|Home | About | Journals | Submit | Contact Us | Français|
Yeast Rtt109 promotes nucleosome assembly and genome stability by acetylating K9, K27 and K56 of histone H3 through interaction with either of two distinct histone chaperones, Vps75 or Asf1. We report the crystal structure of an Rtt109-AcCoA/Vps75 complex revealing an elongated Vps75 homodimer bound to two globular Rtt109 molecules to form a symmetrical holoenzyme with a ~12 Å diameter central hole. Vps75 and Rtt109 residues that mediate complex formation in the crystals are also important for Rtt109-Vps75 interaction and H3K9/K27 acetylation both in vitro and in yeast cells. The same Rtt109 residues do not participate in Asf1-mediated Rtt109 acetylation in vitro or H3K56 acetylation in yeast cells, demonstrating that Asf1 and Vps75 dictate Rtt109 substrate specificity through distinct mechanisms. These studies also suggest that Vps75 binding stimulates Rtt109 catalytic activity by appropriately presenting the H3–H4 substrate within the central cavity of the holoenzyme to promote H3K9/K27 acetylation of new histones prior to deposition.
Histone acetyltransferase (HAT) enzymes modify histone lysine residues to modulate various DNA-templated processes including replication, transcription, and DNA repair. Rtt109 is a fungal-specific HAT that acetylates lysine 56 on newly synthesized histone H3 (H3K56) during S-phase to mediate nucleosome assembly during DNA replication and DNA repair. Rtt109 is also important for cell survival following treatment with a number of genotoxic agents (Collins et al., 2007; Driscoll et al., 2007; Han et al., 2007a; Tsubota et al., 2007). Together with Gcn5, Rtt109 was more recently shown to contribute to the acetylation of histone H3K9 (Fillingham et al., 2008) and H3K27 (Burgess et al., 2010). Rtt109 harbors very low acetyltransferase activity on its own (Driscoll et al., 2007; Tsubota et al., 2007), but its activity is stimulated by association with either of the histone chaperone proteins Asf1 or Vps75 (Albaugh et al., 2010; Berndsen et al., 2008; Han et al., 2007b, c; Tsubota et al., 2007). In vivo, Rtt109-mediated acetylation of H3K56 requires Asf1, an evolutionarily conserved histone chaperone that binds to H3–H4 dimers (English et al., 2006), although Asf1 does not appear to form a stable complex with Rtt109 in vitro (Driscoll et al., 2007; Han et al., 2007b; Tsubota et al., 2007). In addition, both the Asf1 and Vps75 histone chaperones contribute to H3K9 acetylation in vivo (Adkins et al., 2007; Berndsen et al., 2008; Fillingham et al., 2008). Vps75 is a member of the NAP1 histone chaperone family that binds to both (H3–H4)2 tetramers and H2A–H2B dimers (Park et al., 2008; Selth and Svejstrup, 2007). The Rtt109/Vps75 complex can also acetylate H3K56 in vitro (Berndsen et al., 2008; Lin and Yuan, 2008; Tsubota et al., 2007), but a vps75 null mutant shows no decrease in H3K56 acetylation in vivo (Tsubota et al., 2007; Han et al., 2007c). Recently, H3K56 acetylation has been observed to overlap strongly with the binding of key pluripotency regulators at active and inactive promoters in human embryonic stem cells (Xie et al., 2009), but there have been conflicting reports about whether this modification is mediated by the GCN5 (Tjeertes et al., 2009) or p300/CBP HAT in association with ASF1 (Das et al., 2009).
Our group and others have reported on the X-ray crystal structure of Rtt109 (Lin and Yuan, 2008; Stavropoulos et al., 2008; Tang et al., 2008a), leading to the unexpected observation that Rtt109 is structurally related to p300/CBP, despite the lack of significant sequence conservation. We and others have also reported that Vps75 adopts a dimeric head-phone-like structure (Berndsen et al., 2008; Park et al., 2008; Tang et al., 2008b) that is distinct from the monomeric β-sandwich fold of Asf1 (Daganzo, 2003; English et al., 2006). To obtain direct molecular insights into how Rtt109 activity is modulated by the binding of histone chaperones, we now report on the X-ray crystal structure of an Rtt109-AcCoA/Vps75 complex. We also present structure-based mutagenesis, combined with biochemical and, enzymological data, and studies in yeast cells to derive molecular insights into the mechanism by which histone chaperones enhance and mediate lysine-specific histone acetylation by Rtt109.
We prepared the Rtt109/Vps75 complex by coexpressing the full-length Rtt109 protein with the core domain of Vps75 (residues 1–232) in bacteria and purifying the tightly associated complex to homogeneity using a combination of affinity, ion exchange and size exclusion chromatography. The protein complex was mixed with acetyl coenzyme A (AcCoA) and crystals were obtained in spacegroup P21212 (native crystal) that diffracted to 3.2 Å resolution. Soaking of these crystals with a 14-amino acid peptide centered around H3K9 (peptide soaked crystal) produced crystals that diffracted to a higher resolution of 2.8 Å, but with the peptide bound in a non-physiologically relevant manner, but otherwise essentially identical to the native Rtt109/Vps75 complex (Figure S1 and see Supplemental Experimental Procedures for more details). Because of its higher resolution, the peptide-soaked Rtt109/Vps75 complex is used for further analysis as described below. Within the crystal lattice, each asymmetric unit contains one molecule each of Rtt109 bound to AcCoA and Vps75, related by a crystallographic 2-fold symmetry axis to form the functional heterotetramer with two bound AcCoA molecules. The structure was refined to 2.8 Å resolution with refinement statistics of Rwork=0.234 (Rfree=0.298) and geometrical parameters of RMSDbond length=0.011 Å and RMSDbond angle=1.352° (Table 1).
The Rtt109-AcCoA/Vps75 complex reveals that the obligate Vps75 homodimeric headphone-like structure, which contains an elongated helical dimerization domain and globular earmuff domains at each end of the dimerization domain, is bound on opposite ends by two globular Rtt109 molecules to form a symmetrical ring with a hole of ~12 Å diameter (Figure 1). The two Rtt109 subunits make extensive sequence-specific contacts with the Vps75 dimer burying a solvent-accessible surface (SAS) of 3725 Å2 for each Rtt109-Vps75 pair, while the two Rtt109 subunits make more modest and non-specific interactions with each other burying 1736 Å2 of SAS. Each Rtt109 subunit contacts the Vps75 dimer in two distinct regions. The Rtt109-α8-α9 helices pack against the α2-α5 helices of the Vps75 earmuff domain (burying 1982 Å2 of SAS) and the Rtt109 130–179 segment reaches around the Vps75 earmuff domain to engage the Vps75 dimerization helices (burying 1743 Å2 of SAS). The less extensive Rtt109-Rtt109 interface primarily involves non-specific contacts between the α7-β7 loops of both molecules.
The Rtt109-α8-α9 helices make predominantly hydrogen-bond contacts to the α2-α5 helices of the Vps75 earmuff domain (Figure 2A). Specifically, lysines 356 and 363 of Rtt109-α8 hydrogen bond to Gln64 of Vps75-α2 and Asn70 of Vps75-α3, respectively; and Tyr364 of Rtt109-α8 makes van der Waals contact to Ala74 of Vps75-α4. Glu374, Glu378, and Arg390 of Rtt109-α9 make salt-bridges to Arg173 and Lys177 of Vps75-α5, and Arg73 and Asp81 of Vps75-α4, respectively. Asn382 of Rtt109-α9 also makes hydrogen bond to the backbone NH of Ala74, and Leu389 of Rtt109-α9 makes van der Waals contacts to Phe77 and the aliphatic region of Lys78 of Vps75-α4.
The 130–179 segment of Rtt109 is not conserved in all species of yeast (Figure S2) and is either absent (deleted from the crystallized protein) or disordered in previously reported Rtt109 structures (Lin and Yuan, 2008; Tang et al., 2008a). However, in the Rtt109-AcCoA/Vps75 complex, this segment contains a well-ordered central helix from residues 144–155 (called α3i as it forms in between the previously defined α3 and α4 helices) and surrounding loop regions that are partially ordered (Figure 2B). The Rtt109-α3i helix contacts both Vps75 subunits through a combination of hydrogen bonding and van der Waals interactions. Arg154 of Rtt109-α3i forms a salt bridge with Asp222 of Vps75-α8, a residue at the base of the Vps75 earmuff domain that also forms two main-chain H-bonds with Ile138 of Rtt109. In addition, Arg149 of Rtt109-α3i forms a salt-bridge with Glu23 of Vps75-α1′ from the opposite subunit. Other residues of Rtt109-α3i also contact the α8 and α1′ helices of the Vps75 earmuff and dimerization domains, respectively, through mostly van der Waals interactions (Figure 2B). Specifically, Ala145, Leu148, Ile150, Leu151, and Ala152 of Rtt109-α3i and Ile138 within the preceding loop form a hydrophobic network with a cluster of Vps75-α8 (Val213 and Tyr216) and α1′ (Phe15, Leu16, and Ala19) residues.
Except for the 130–179 segment of the Rtt109-CoA/Vps75 complex, the uncomplexed Rtt109 and Vps75 proteins superimpose well with no significant structural changes and RMS deviations for all ordered Cα atoms of 0.75 Å and 1.14 Å, respectively. Strikingly, even the Rtt109 active site residues Tyr199 and Trp222, the bound AcCoA and the acetylated Lys290 superimpose well (Figure 2C). This observation suggests that Vps75-mediated stimulation of Rtt109 HAT activity does not involve an alteration of Rtt109 conformation or its active site upon Vps75 binding. Vps75 binding to Rtt109 strongly stimulates the kcat (about 100-fold) and has little effect on the Km for H3 substrates (Berndsen et al., 2008). Based on this and the structures of the holoenzyme and free Rtt109, we propose that Vps75 stimulates Rtt109 HAT activity by productively positioning specific lysine residues of histone H3 within the Rtt109 active site.
To determine the functional significance of the Rtt109-Vps75 contacts that are observed in the crystal structure of the complex, we carried out structure-based mutagenesis of Rtt109 followed by Rtt109-Vps75 interaction and enzymatic assays using recombinant (H3–H4)2 tetramer as a substrate.
Given that in the Rtt109-AcCoA/Vps75 complex, one Rtt109 molecule contacts both Vps75 subunits, we first asked whether Vps75 dimerization is required for its ability to stimulate Rtt109 HAT activity. For this purpose, we prepared the Vps75-(C21E,V25S,I28E,V32E) mutant previously shown to produce monomeric Vps75 (Berndsen et al., 2008) and found that it is strongly defective in binding to Rtt109 (Figure 3A) as well as in stimulating Rtt109 HAT activity (Figure 3B). This result confirms that Vps75 dimer formation is required for Rtt109 interaction and optimal stimulation of Rtt109 HAT activity.
As Rtt109 interacts with Vps75 through two distinct surfaces (Figures 2 and S2A), we also asked whether both surfaces are required for Rtt109 function in vitro. We have previously reported mutagenesis data implicating two distinct Vps75 surfaces for Rtt109 interaction (Tang et al., 2008b). Specifically, we showed that the Vps75-(E218K,D222K) mutant abolishes Rtt109 interaction, while the distal Vps75-(R173E,K177E) mutant and Vps75-Δ(167–178) deletion greatly compromise Rtt109 interaction. This data is consistent with the Rtt109-AcCoa/Vps75 structure reported here. Vps75-Asp222 and –Glu218 interact with the 130–179 segment of Rtt109 (Figure 2B), whereas Vps75-Arg173 and -Lys177 interact with the α8-α9 region of Rtt109. As shown in Figure 3A, another mutation, Vps75-(R73D,A74D), that disrupts the interface with Rtt109-α8-α9, completely abolishes interaction with Rtt109. Vps75 mutations that reduce its interaction with Rtt109 in pull-down assays also show a diminished ability to stimulate Rtt109 HAT activity (Figure 3B). In particular, the Rtt109 binding defective mutants, Vps75-(R73D, A74D), -(R173E,K177E), and -(E218K,D222K) each show a reduced ability to stimulate Rtt109 HAT activity relative to the wild-type Vps75 protein.
Mutation of Rtt109 residues that contact Vps75 also show a defect in pull-down assays and Vps75-mediated stimulation of Rtt109 HAT activity. As shown in Figure 3C, removal of the 130–179 segment of Rtt109 that contacts the Vps75 dimerization domain, or substitution mutations of Vps75-interacting residues in this segment, Rtt109-(L148D) or Rtt109-(I150D,L151D), abolish Vps75 binding. Similarly, combined substitution mutations in the Rtt109 α8-α9 residues that mediate contacts with the Vps75 earmuff domain, Rtt109-(E378R,N382R) and Rtt109-(R355E,K356E), also diminish Vps75 binding. Correlating with the reduced Rtt109-Vps75 interaction, each of these Rtt109 mutants show reduced Vps75-stimulated HAT activity relative to wild-type Rtt109 (Figure 3D). Taken together, these studies indicate that the 130–179 segment and α8-α9 region of Rtt109 (Figure S2B) are both required for optimal Vps75 interaction and HAT activity stimulation.
Comparison of free and Vps75-bound Rtt109 structures suggests that the Rtt109 130–179 segment is only ordered in the presence of Vps75 (Lin and Yuan, 2008; Tang et al., 2008a). To confirm that this is also true in solution, we probed the proteolytic sensitivity of Rtt109 as a function of Vps75 binding. To carry out these studies, we prepared three recombinant Rtt109 proteins: wild-type Rtt109, Rtt109-Δ(130–179) in which residues 130–179 are removed and Rtt109-(L148D), a mutant defective in Vps75 binding. As shown in Figure 3E, in the absence of Vps75, Rtt109-wt and Rtt109-(L148D) are readily cleaved by trypsin, whereas the Rtt109-Δ(130–179) mutant is significantly more protease resistant. In the presence of Vps75, however, Rtt109-wt is protected from proteolysis. In contrast, the Rtt109-(L148D) mutant which cannot bind to Vps75 but retains the 130–179 segment, remains sensitive to trypsin even in the presence of Vps75. Taken together, these studies reveal that, in the absence of Vps75, Rtt109 is highly susceptible to trypsin cleavage, and this appears to be nucleated by the 130–179 segment of Rtt109. Interestingly, a recent report has shown that the in vivo stability of Rtt109 is compromised in cells lacking Vps75 (Fillingham et al., 2008). Based on the structural and biochemical results reported here, we hypothesize that the reduced stability of Rtt109 in the absence of Vps75 in vivo correlates with the exposure of the 130–179 segment, which renders Rtt109 susceptible to degradation. Interestingly, Rtt109 orthologs from several yeast species do not contain the 130–179 region (Figure S2B), suggesting that these species may compensate for the absence of the 130–179 segment by employing other surfaces of Rtt109 to stabilize its interaction with Vps75. Alternatively, these species may not use Vps75 stimulate the HAT activity of Rtt109.
Given that Rtt109 HAT activity can be stimulated by both the Vps75 and Asf1 histone chaperones, we asked whether the regions of Rtt109 that mediate Vps75 interaction and stimulation of HAT activity are also employed for Asf1-mediated Rtt109 stimulation. To determine whether the stimulation of HAT activity by Asf1 requires the 130–179 segment of Rtt109, we assayed the acetyltransferase activity of Rtt109-wt and two Rtt109 mutants, Rtt109-Δ(130–179) and Rtt109-(L148D), in the presence or absence of Asf1. As shown in Figure 3D, Asf1 enhances the HAT activity of the Rtt109-Δ(130–179) and Rtt109-(L148D) mutants as effectively if not better than the wild-type protein, demonstrating that Asf1 does not employ the 130–179 segment of Rtt109 to stimulate its HAT activity.
This is in striking contrast to Vps75, which poorly increases the HAT activity of Rtt109-Δ(130–179) and Rtt109-(L148D) (Figure 3D). In addition, Rtt109 α8-α9 mutations that significantly cripple stimulation of Rtt109 HAT activity by Vps75 do not show a defect in Asf1-stimulated HAT activity (Figure 3D). These include Rtt109-(R355E,K356E), Rtt109-(E374R), Rtt109-(E378R) and Rtt109-(E378R,N382R). Taken together, these results suggest that Vps75 and Asf1 stimulate Rtt109 HAT activity through distinct mechanisms.
To further probe the lysine substrate specificity of the Rtt109/Vps75 complex, we prepared histone H3 mutants that contain either K9R/K27R (K56 is available for acetylation), K27R/K56R (K9 is available for acetylation) or K9R/K56R (K27 is available for acetylation) substitutions in the context of the (H3–H4)2 tetramer and used them as substrates for Rtt109/Vps75. Kinetic analysis revealed that the wild-type Rtt109/Vps75 complex exhibits comparable catalytic efficiency towards the wild-type and H3-K27R/K56R mutant histone substrates but significant defects towards any of the histone substrates harboring a K9R mutation, showing a decrease in kcat of about 10-fold and elevated Km values (Figure 3F and S3A). This data is consistent with quantitative MS and Edman sequencing analyses of the reaction products obtained with wild-type (H3–H4)2, which showed that the Rtt109/Vps75 complex preferentially acetylates H3K9 over H3K56 and other acetylation sites in the N-terminal tail of H3 (Figure S3D).
In our survey of Rtt109 mutants that exhibit reduced HAT activity in the presence of histone chaperones, we were intrigued by an Rtt109-(R292E) mutant that did not affect Rtt109/Vps75 complex formation (Figure 3C) and only had a modest effect on Vps75-stimulated H3 acetylation, but showed more dramatic defects in Asf1-stimulated acetylation (Figure 3D). To further probe this mutation, we carried out more detailed kinetics on the Rtt109/Asf1 and Rtt109-(R292E)/Asf1 complexes using the wild-type (H3–H4)2 substrate. We found that, in contrast to wild-type Rtt109/Vps75 and Rtt109-(R292E)/Vps75 that acetylate the (H3–H4)2 substrate with comparable activity (< 2-fold decrease in kcat/Km) and similar activity profiles for each of the substrates bearing H3 mutants (Figures 3F and S3B), the Rtt109-(R292E)/Asf1 mutant shows about a 45-fold decrease in kcat/Km for the (H3–H4)2 substrate relative to wild-type Rtt109/Asf1 (Figures 3F and S3C). This observation demonstrates that residue Rtt109-R292 plays a more important role in Asf1- over Vps75-mediated Rtt109 acetylation of histone H3.
The observation that the acetyl group of AcCoA and the active site residues of Rtt109 are facing towards the interior of the ring structure implies that the histone substrate is bound in the interior of the ring-shaped Rtt109-AcCoA/Vps75 complex. This hypothesis is supported by a mapping of sequence conservation onto the surfaces of both Rtt109 and Vps75, which reveals that the most highly conserved surfaces of Rtt109 and Vps75 are brought together in the interior of the ring (Figures 4A and 4B). In contrast, the exterior surface of the complex shows relatively poor sequence conservation (Figure 4C).
We also identified several conserved and solvent exposed Vps75 and Rtt109 residues within the ring that result in reduced HAT activity when mutated (Figures 3B, 3D and and4D).4D). These mutants included Rtt109-(R194E), Rtt109-(E368R) and Rtt109-(EEYD368RRYR), as well as Vps75-(E206K,E207K). To determine the histone acetylation defects of these mutants more accurately, we determined their steady-state kinetic parameters in the presence of a large excess (presumed saturating) of Ac-CoA (Figures 4E and S4). This analysis revealed that the Vps75-(E206K,E207K) mutant predominantly has a histone tetramer Km defect (greater than10-fold), while the other mutants have defects in kcat of between 4 and 70-fold. While the Km defect of the Vps75-(E206K,E207K) mutant likely reflects a defect in histone binding, the kcat defect of the other mutants could reflect a destabilization of the transition state of the acetylation reaction. Given that the mutated residues are not near the active site of the enzyme and are not involved in Rtt109/Vps75 interaction, we propose that they mediate correct positioning of the histone H3 substrate in the active site for optimal acetylation. This proposal is consistent with our structural observation that Vps75 binding to Rtt109 does not significantly alter the Rtt109 active site (Figure 2C), as well as a previous report showing that Vps75 binding stimulates Rtt109 activity by elevating the kcat, rather than by reducing the Km for the H3 substrate (Berndsen et al., 2008). Taken together, these data further support the role of the interior of the ring-shaped complex in binding the histone substrate in a manner that is productive for H3 acetylation.
It has been reported that a truncation of the C-terminal acidic tail of Vps75 (residues 224–264) impairs its ability to activate Rtt109 without affecting Vps75 binding to Rtt109 (Park et al., 2008). Since we found Vps75 residues as C-terminal as D225 involved in Rtt109 interaction, we asked how the longer fragment (residues 1–232) used in our structure determination may impact Rtt109 HAT activity. To this end, we compared the Rtt109 activation capabilities of Vps75-wt, Vps75-(1–232) and Vps75-(1–223). We found that Vps75-(1–232) and Vps75-(1–223) activate Rtt109 to similar levels, and with an ~4-fold reduced rate relative to full-length Vps75 (Figure 3B). These results suggest that Vps75 residues C-terminal to Leu232 are required for optimal activation of Rtt109. Since the last ordered residue of Vps75 in the structure of the complex, Glu226, points towards the interior of the ring, we propose that the Vps75 C-terminal acidic tail may also be in position to associate with histones bound within the ring-shaped enzyme complex.
In vitro, the Rtt109/Vps75 holoenzyme cannot acetylate nucleosomes (Han et al., 2007c; Tsubota et al., 2007a), suggesting that it may exclusively act on newly synthesized histones prior to their deposition onto DNA. Among the five acetylatable lysine residues in the N-terminal tail of H3, lysines K9 and K27 were found to be the most extensively acetylated following a short pulse of [3H]-lysine in S. cerevisiae (Kuo et al., 1996). Furthermore, in contrast to H3K56 acetylation, which is entirely dependent upon Rtt109 and Asf1 (Driscoll et al., 2007; Drogaris et al., 2008; Han et al., 2007a; Tsubota et al., 2007a), both Gcn5 and Rtt109-Vps75 contribute to H3K9 and H3K27 acetylation in vivo (Berndsen et al., 2008; Burgess et al., 2010; Fillingham et al., 2008).
In order to characterize the effects of the structure-based Rtt109 mutations in vivo, we first validated our assays by monitoring H3K9/K27 and H3K56 acetylation in strains where selected HAT genes were deleted. Based on mass spectrometry (MS), H3K9 and H3K27 acetylation is abundant in vivo, with approximately 20% of H3 molecules acetylated at K9 and K27 in wild-type cells (Figure 5B–C). As judged by MS, the levels of H3K9 and H3K27 acetylation were considerably reduced, but not completely abolished in gcn5 single mutants, whereas H3K56 acetylation was as abundant as in wild-type cells (Figure 5A–C). In contrast, essentially no H3K9 or H3K27 acetylation remained in gcn5Δ vps75Δ or gcn5Δ rtt109Δ double mutants and, consistent with previous reports, the rtt109Δ mutation essentially abolished H3K56 acetylation (Figure 5A–C) (Driscoll et al., 2007; Han et al., 2007b). As judged by MS, a vps75 null mutation did not significantly cripple acetylation of H3K14, H3K18 or H3K23, whereas H3K18 was essentially abolished in gcn5 single mutants (data not shown). These results demonstrate that, in vivo, the Rtt109/Vps75 enzyme functionally overlap with Gcn5 to acetylate H3K9 and H3K27, whereas Rtt109 and Asf1 cooperate to promote H3K56 acetylation.
Importantly, when introduced into either gcn5Δ rtt109Δ or gcn5Δ vps75Δ double mutants, low copy plasmids expressing wild-type Rtt109-Flag or Vps75-Flag, respectively, from their natural promoters, restored H3K9, K27 and K56 acetylation (Figures 5A–H). This provided assays to monitor the effects of structure-based Rtt109 mutations on histone acetylation. The structure-based Rtt109 mutants that we tested in vivo showed defects in H3K9 and K27 acetylation, with the Rtt109-(L148D) and Rtt109-(E378R,N382R) mutants being the most defective (Figures 5E–G and data not shown). Therefore, we focused our in vivo studies on these two Rtt109 mutants. Based on MS and immunoblotting, the Rtt109-(L148D) and Rtt109-(E378R,N382R) were defective in H3K9/K27 acetylation, but not H3K56 acetylation in vivo (Figure 5D–G). As judged by MS and immunoblotting, the Rtt109-(L148D) mutation abolished its interaction with Vps75 while the Rtt109-(E378R,N382R) mutant showed residual binding to Vps75 in vivo (Figure 6A–B). Consistent with the greater sensitivity of Rtt109-L148D and Rtt109-E378R,N382R to mutations, these residues are located within the Rtt109 (130–179) segment and α8-α9 helix region of Rtt109, respectively, which mediate contacts to the two distinct binding surfaces on Vps75 (Figure 2). In the absence of Vps75, Rtt109 has been reported to be less stable than in wild-type cells (Fillingham et al., 2008). In agreement with its strong defect in interaction with Vps75, the Rtt109-(L148D) mutation (and to a lesser extent the Rtt109-(E378R,N382R) mutation) destabilized the protein to an extent similar to that observed in a vps75 null mutant even though each of the mutant proteins are expressed at similar steady-state levels (Figure 6D).
We also determined the effects of Vps75 mutations on histone acetylation and complex formation in vivo. In keeping with the structure of the holoenzyme (Figure 2), the Vps75-(R173E,K177E) and Vps75-(E218K,D222K) mutations reduced the interaction of Vps75 mutants with Rtt109 both in vitro (Figure 3A) and in vivo (Figure 6C). In contrast, mutation of residues E206 and E207, which line the central cavity of the Rtt109/Vps75 holoenzyme and are not involved in interaction with Rtt109 (Figure 3A and and4D),4D), only has a mild effect on the interaction of Rtt109 and Vps75 in vivo (Figure 6C). However, compared with WT Vps75, all three Vps75 mutations reduced H3K9/K27 acetylation without perturbing H3K56 acetylation in vivo (Figure 5H). These results taken together, demonstrate that mutations that decrease Rtt109/Vps75 interaction and H3K9 but not H3K56 acetyltransferase activity in vitro correlates with similar activities in vivo.
In vivo, mutations that cripple the acetylation of new histones or interfere with replication-coupled nucleosome assembly confer sensitivity to genotoxic agents and defects in cell proliferation and heterochromatin-mediated gene silencing (Burgess et al., 2010; Driscoll et al., 2007; Han et al., 2007a; Han et al., 2007b; Li et al., 2008; Tsubota et al., 2007). This is indeed what we observed in rtt109 single and rtt109 gcn5 double mutants (Figures S5 and S6). In rtt109Δ gcn5Δ double mutants, H3K9/K27 in the N-terminal tail and H3K56 acetylation are essentially abolished (Figure 5A–C). However, none of our structure-based mutations that selectively disrupt the structure and/or activity of the Rtt109/Vps75 holoenzyme resulted in proliferation, genotoxic agent sensitivity or heterochromatin-mediated silencing (Figures S5A, S5B and S6C). This was true even when structure-based Rtt109/Vps75 mutations were combined with other mutations that should reduce the efficiency of the replication-coupled nucleosome assembly pathway, such as null mutations in gcn5 (which reduces N-terminal acetylation of new H3 molecules (Burgess et al., 2010), hat1 (an enzyme that acetylates new H4 molecules at lysines 5 and 12) (Kelly et al., 2000; Qin and Parthun, 2002) or rtt106 (a chaperone that binds to new H3/H4 molecules) (Huang et al., 2007) (Figures S5B and S6A–C). These results suggested that the absence of the Rtt109/Vps75 enzyme is not sufficient to confer a physiologically significant defect in replication-coupled nucleosome assembly. To further confirm this hypothesis, we affinity-purified Chromatin Assembly Factor 1 (CAF-1) and determined by MS the amounts of histones bound to CAF-1. CAF-1 is the prototypical replication-coupled nucleosome assembly factor (Kaufman et al., 1997) and cells where H3K56 acetylation is impaired contain low amounts of H3/H4 bound to CAF-1 (Kaufman et al., 1997; Li et al., 2008). Consistent with the fact that they do not confer any striking phenotype, the Rtt109-(L148D) and Rtt109-(E378R,N382R) mutants did not show any striking decrease in the amounts of H3/H4 bound to CAF-1, even in cells lacking Gcn5 (Figure 5I). Taken together, these studies show that the H3K9/K27 acetylation activity of the Rtt109/Vps75 complex is not sufficient to confer phenotypes associated with defects in nucleosome assembly.
The Rtt109 histone acetyltransferase promotes nucleosome assembly and genome stability by acetylating K9, K27 and K56 on new non-nucleosomal histone H3 molecules through its interactions with either of two distinct histone chaperones, Vps75 or Asf1. H3K9 acetylation, in particular, is evolutionarily conserved from yeast to human cells and is the most prominent site of acetylation in the N-terminal tail of new H3 molecules in S. cerevisiae (Adkins et al., 2007; Kuo et al., 1996). Despite the importance of the acetylation of new histones, the mechanism and the molecular basis of chaperone-mediated histone lysine acetylation specificity have gone largely unexplored.
To explore the mechanism by which histone chaperones promote Rtt109 HAT activity and lysine specificity, we determined the X-ray crystal structure of an Rtt109-AcCoA/Vps75 complex that revealed a 2-fold symmetrical heterotetrameric ring containing an interior cavity of ~12 Å diameter. Biochemical, enzymatic and in vivo studies further demonstrated that Rtt109-Vps75 contacts observed in the crystals are important for optimal H3K9/K27 but not H3K56 acetylation both in vitro and in yeast cells. A comparison of the Rtt109-AcCoA/Vps75 complex with the free Rtt109 and Vps75 proteins also showed that Vps75 binding to Rtt109 does not alter the Rtt109 active site, suggesting that Vps75 binding stimulates Rtt109 catalytic activity by appropriately presenting histone H3 for acetylation.
We also used information derived from the structure of the Rtt109-AcCoA/Vps75 complex to demonstrate that Asf1 and Vps75 stimulate the HAT activity of Rtt109 via different mechanisms. Along with other groups, we previously reported the identification of residues generally important for Rtt109-mediated H3 acetylation, including Rtt109 residues Asp89, Tyr199, Trp222 and Asp287 (Han et al., 2007b; Tang et al., 2008a; Tsubota et al., 2007). In this study, we extended these findings to residues that play more dedicated roles in Vps75-mediated activation of Rtt109 for H3K9 acetylation. We showed that the 130–179 segment of Rtt109 and other residues that contribute to interaction surfaces between Rtt109 and Vps75 play a particularly important role in Vps75-dependent histone H3K9 acetylation both in vitro and in yeast cells. We also showed that Rtt109-R292 plays a more important role in Asf1-mediated histone acetylation.
The structure of the Rtt109-AcCoA/Vps75 complex, together with results from in vitro enzymological assays and analysis of acetylation in yeast cells, also indicate that histone substrate binding occurs within the interior surface of the ring-shaped Rtt109/Vps75 complex. Three lines of evidence support this model. First, the acetyl groups of the two Rtt109-bound AcCoA cofactors in the complex point into the interior of the ring. Second, the internal surface of the ring shows a much higher degree of sequence conservation than the exterior. Third, we have identified several Rtt109 and Vps75 substitution mutations within the interior of the ring that reduce histone H3 acetylation in vitro and H3K9/K27 but not H3K56 acetylation in vivo. The 2:2 stoichiometry and 2-fold symmetry of the complex are also consistent with the observation that Vps75 preferentially binds (H3–H4)2 heterotetramers (Selth and Svejstrup, 2007) and strongly suggests that both H3 molecules of the heterotetramer are acetylated for histone deposition. In this way, the Vps75 histone chaperone serves as a cofactor for Rtt109-mediated acetylation by both appropriately positioning histone H3K9 for acetylation in the Rtt109 active site and ensuring homogenous histone H3 acetylation before H3–H4 deposition into nascent chromatin.
The structure of the Rtt109-AcCoA/Vps75 complex also enabled us to design Rtt109 and Vps75 separation-of-function mutants. These mutations leave H3K56 acetylation unaffected, but completely abolish the H3K9/K27 acetylation that remains in gcn5Δ cells (Figures 5 and and6).6). Significantly, mutations that selectively perturb the Rtt109/Vps75 enzyme do not exacerbate the phenotypes of gcn5Δ cells nor do they exhibit phenotypes that are commonly observed in mutants where replication-coupled nucleosome assembly is defective. This is likely because of the overlapping substrate specificity of Gcn5, Rtt109/Asf1 and Rtt109/Vps75 for H3K9/K27 and suggests that the Rtt109/Vps75 complex contributes to, but is not essential for replication-coupled nucleosome assembly.
It is currently unclear why several distinct HATs (Hat1, Gcn5, Rtt109/Vps75, Rtt109/Asf1 and possibly others) contribute to the acetylation of new H3/H4 molecules. By analogy with isoenzymes, some of these HATs may not be functionally redundant under certain growth or cellular stress conditions. However, new tools will be necessary to determine whether this is the case. Thus far, studies aimed at assigning functions to the acetylation of newly synthesized histones have been mainly performed by creating yeast strains where H3 and H4 carry several lysine-to-arginine mutations in the N-terminal tails. This experimental strategy has been helpful, but suffers from two limitations. First, these mutations cripple the acetylation of both newly synthesized and pre-existing histones and global disruption of histone acetylation interferes with transcription. Second, phenotypes that result from histone lysine-to-arginine substitutions might be due to mutations of the lysines, rather than the absence of acetylation. Rtt109/Vps75 is an enzyme that exclusively acetylates non-nucleosomal histones at two specific residues in the N-terminal tails of new H3 molecules. Therefore, the structure-based Rtt109 mutants described here, which are specifically defective in H3K9/H3K27, but not H3K56 acetylation, may provide invaluable tools to investigate the elusive function of the acetylation of the N-terminal tails of newly synthesized histones.
Individual Rtt109 and Vps75 proteins were engineered as GST fusions as previously described (Tang et al., 2008a; Tang et al., 2008b), with the GST moiety removed by TEV protease as necessary. Quick-change site-directed mutagenesis (Stratagene) was employed to introduce selected protein mutations. Recombinant human 6His-ASF1a protein (Tang et al., 2006) and yeast (H3–H4)2 heterotetramer (Luger et al., 1999) were prepared as previously described. Yeast 6His-Asf1N(1–154) was cloned and prepared similarly to human ASF1a. To produce complexes by co-expression, we subcloned DNAs encoding full-length Vps75 (or a 1–232 fragment as used in crystallization) and full-length Rtt109 proteins into the MCS1 and MCS2 sites, respectively, of a modified 6His-TEV-pCDF-Duet1 vector. Protein complexes were expressed in bacteria and purified to homogeneity through a combination of Ni-affinity, TEV protease cleavage, MonoQ anion exchange and Superdex S200 gel filtration chromatography.
Rtt109-Vps75 pull-down assays were carried out with wild-type or mutant GST-Rtt109 proteins bound to glutathione resin and untagged wild-type or mutant forms of Vps75. Following binding reactions, proteins retained on the resin were resolved by SDS-PAGE. For enzymatic assays, we adapted the radioactive HAT assay as previously described (Lau et al., 2000; Thompson et al., 2001) using yeast (H3–H4)2 tetramer as a substrate. Proteolysis studies of Rtt109 protein constructs were carried out with trypsin protease in the absence or presence of stoichiometric amounts of full length Vps75 and terminated by boiling the samples for resolution on SDS-PAGE.
Native crystals of the Rtt109-AcCoA/Vps75-(1–232) complex were obtained using hanging-drop vapor-diffusion from a reservoir solution containing 10.0% (v/v) PEG8000, 8% (v/v) ethylene glycol and 100 mM Hepes pH7.5 buffer. Peptide-soaked Rtt109-AcCoA/Vps75-(1–232) crystals were prepared by soaking native crystals with 1 mM CoA and 1 mM of a H3K9 14-amino acid peptide. The structure of the peptide soaked Rtt109-AcCoA/Vps75-(1–232) complex was determined to 2.8 Å resolution using molecular replacement with Rtt109-Δ (130–179) (PDB ID 3D35) (Tang et al., 2008a) and a Vps75-(1–232) monomer (PDB ID 3DM7) (Tang et al., 2008b) as search models. The models were adjusted and refined to include the 130–179 segment of Rtt109, AcCoA, and residues 11–14 of the H3K9 peptide [(NH2)A-R-T-K-Q-T-A-R-K-S-T-G-G-K (CONH2)] (Figure S1 and Table 1). Given that the peptide does not make extensive protein contact and is located far from the active site, we infer that this peptide-binding mode is not biologically relevant. The final model, with the addition of 54 water molecules, was checked for errors using composite simulated annealing omit maps (Table 1). The structure of the native Rtt109-AcCoA/Vps75-(1–232) crystals was determined using the refined peptide-soaked Rtt109-AcCoA/Vps75-(1–232) model and refined to 3.2 Å resolution (Table 1). Comparison of the two crystal forms reveals that the presence of the H3K9 peptide does not alter the overall Rtt109-AcCoa/Vps75 complex structure.
Yeast strains and plasmids to express Rtt109, as well as protocols to monitor H3K9 and K56 acetylation by immunoblotting and mass spectrometry, are described in detail in the Table S1 and the Supplemental Methods.
Vps75 homodimer interacts with two Rtt109 monomers to form a 2:2 ring-like complex
Vps75-mediated Rtt109 acetylation of H3K9 requires stable Rtt109-Vps75 interaction
Vps75 and Asf1 histone cheperones mediate Rtt109 acetylation by distinct mechanism
H3/H4 binds to the central cavity of the Rtt109-Vps75 ring-like complex
We thank X. Liu for initial contributions to this project and Dr. Zhiguo Zhang for the gift of an antibody specific for histone H3K27 acetylation. This work was supported by NIH grants to R.M. (R01 GM060293, P01 AG031862) and P.A.C (R01 GM062437), and by funds from CIHR and NSERC to A.V. and P.T. IRIC’s infrastructure is supported by the Canadian Center of Excellence in Commercialization and Research, the Canadian Foundation for Innovation, and the Fonds de la Recherche en Santé du Québec. Part of this research was conducted at GM/CA-CAT beamline 23ID-B at the Advanced Photon Source, Argonne National Laboratory. Use of the APS was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. W-31-109-ENG-38.
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.
Accession codes. Protein Data Bank: Atomic coordinates and structure factors of the Rtt109-AcCoA/Vps75 complexes have been deposited with accession codes XXXX (native crystal) and YYYY (peptide-soaked crystal), respectively.