Catalytic activation of histone acetyltransferase Rtt109 by a histone chaperone

Abstract

Most histone acetyltransferases (HATs) function as multisubunit complexes in which accessory proteins regulate substrate specificity and catalytic efficiency. Rtt109 is a particularly interesting example of a HAT whose specificity and catalytic activity require association with either of two histone chaperones, Vps75 or Asf1. Here, we utilize biochemical, structural, and genetic analyses to provide the detailed molecular mechanism for activation of a HAT (Rtt109) by its activating subunit Vps75. The rate-determining step of the activated complex is the transfer of the acetyl group from acetyl CoA to the acceptor lysine residue. Vps75 stimulates catalysis (> 250-fold), not by contributing a catalytic base, but by stabilizing the catalytically active conformation of Rtt109. To provide structural insight into the functional complex, we produced a molecular model of Rtt109-Vps75 based on X-ray diffraction of crystals of the complex. This model reveals distinct negative electrostatic surfaces on an Rtt109 molecule that interface with complementary electropositive ends of a symmetrical Vps75 dimer. Rtt109 variants with interface point substitutions lack the ability to be fully activated by Vps75, and one such variant displayed impaired Vps75-dependent histone acetylation functions in yeast, yet these variants showed no adverse effect on Asf1-dependent Rtt109 activities in vitro and in vivo. Finally, we provide evidence for a molecular model in which a 1∶2 complex of Rtt109-Vps75 acetylates a heterodimer of H3-H4. The activation mechanism of Rtt109-Vps75 provides a valuable framework for understanding the molecular regulation of HATs within multisubunit complexes.

In eukaryotes, histone acetylation regulates nucleosome assembly, chromatin folding, transcription, and DNA repair (1). Histone acetyltransferases, or HATs, transfer the acetyl group of acetyl CoA onto the ε-amino group of lysine residues. HATs are almost exclusively found within larger multisubunit complexes with accessory proteins that modulate enzymatic activity and direct substrate specificity (2–4). For example, the yeast HATs Esa1 (KAT5) and Gcn5 (KAT2) alone are ineffective catalysts toward nucleosomal histones; both enzymes require association with other protein subunits for efficient acetyl transfer on nucleosomal substrates (5–7). Despite the prevalent reports of multisubunit HAT complexes, the molecular mechanisms by which accessory proteins regulate the acetyltransferases are largely unknown.

HAT complexes formed by the acetyltransferase Rtt109 (KAT11) are remarkable. Distinct histone chaperones (Vps75 and Asf1) help direct Rtt109 substrate selection for different biological processes, and each stimulates the acetyltransferase activity of Rtt109 (8–11). In Candida albicans, Rtt109 is required for pathogenesis and, thus, could provide a unique target for antifungal therapies (12). Notably, Rtt109 lacks sequence homology to previously characterized HATs (11, 13–15), although Rtt109’s structure has revealed similarity to the mammalian HAT p300 (KAT3B) (16).

Rtt109 is responsible for acetylating multiple lysine residues on nonnucleosomal histone H3 substrates (11, 17). Rtt109 acetylates lysine 56 on the histone H3 core domain (H3K56), a mark that occurs globally on newly synthesized histones in Saccharomyces cerevisiae and Schizosaccharomyces pombe and is required for genome stability (13, 15, 18–21). H3K56ac was recently detected in humans, where it has been shown to be prominent in multiple cancers and is enriched at genes that are key regulators of stem cell pluripotency (22–24). H3K56ac is absent in yeast cells lacking Asf1 (asf1Δ mutants), indicating an essential role for Asf1 in the Rtt109-dependent acetylation of H3K56 (11, 13, 25). The Nap1 family histone chaperone Vps75 copurifies with Rtt109, but yeast cells lacking Vps75 (vps75Δ mutants) display normal levels of H3K56ac. Instead, vps75Δ cells display a reduction in acetylation of H3K9 and H3K23 (8, 9). Additionally, the Rtt109-Vps75 complex contributes to H3K27ac, an overlapping function with the HAT Gcn5 (26). In vitro, both Asf1 and Vps75 stimulate the histone acetyltransferase activity of Rtt109 (8, 10, 11, 27, 28). However, a biochemical and structural understanding for how these discrete histone chaperones stimulate catalysis and direct distinct cellular functions is lacking.

Here we utilize biochemical, structural, and genetic analyses to explore the functional Rtt109-Vps75 complex and the molecular mechanism of catalytic activation. We demonstrate that catalytic activation of Rtt109 by Vps75 is achieved via enhanced acetyl transfer that occurs due to stabilization of the active enzyme conformation. Additionally, we provide a model of the Rtt109-Vps75 complex derived from X-ray crystallographic studies and define critical interacting surfaces in vitro and in vivo that are required for specific Vps75 activation of Rtt109. Finally, we present evidence supporting a molecular model in which a 1∶2 complex of Rtt109-Vps75 acetylates histone H3-H4 heterodimers. This report describes a functional HAT-chaperone complex at the biochemical, genetic, and structural levels. These results provide important clues toward our general understanding for the roles of accessory proteins in multisubunit HAT complexes.

Results

Mechanism of Rtt109 Activation by Vps75.

Rtt109 alone is an inefficient acetyltransferase (k_cat = 2.3 ± 0.7 × 10^-3 s^-1 for histone H3); however, when Rtt109 is purified in complex with Vps75, its enzymatic activity increases ∼100-fold (8). Rtt109-Vps75 complex can be separated by strong hydrophobic interaction chromatography or expressed individually and recombined to produce fully activated complex, demonstrating the reversibility of binding (29). We hypothesized that the rate enhancement of the Rtt109-Vps75 complex could be due either to an increase in the rate of acetyl transfer (k_cat effect) or to the ability of the histone chaperone to affect substrate binding affinity (K_m effect). To determine the activation mechanism of Vps75 on Rtt109, we performed steady-state kinetic analyses with increasing H3 concentrations and varied levels (0.2×–8×) of chaperone relative to enzyme (Fig. 1A). Each resulting dataset was fitted to the Michaelis–Menten equation. The series of curves revealed a substantial increase in k_cat and k_cat/K_m values as the ratio of Vps75 to Rtt109 increased, whereas no general trend for K_m values was observed (Fig. 1A). Similar trends were observed when H3-H4 was employed as substrate (Table S1). These results indicated that the mechanism of Rtt109 activation by Vps75 involves enhanced catalytic turnover (k_cat) but not increased binding affinity for histone. Next, we explored the mechanism by which Vps75 enhances the rate of catalysis by Rtt109. To first ascertain the step of catalysis affected by Vps75, the rate-limiting step for Rtt109 alone was determined. Previously, we identified the chemical step of acetyl transfer as rate limiting for the Rtt109-Vps75 complex (29). A change in the rate-limiting step in the absence of Vps75 would suggest that the chaperone activates Rtt109 by enhancing the rate of substrate binding or product release. To determine the rate-limiting step for Rtt109 alone, pre-steady-state kinetic analysis was performed and compared to the results obtained with the complex (Fig. S1A) (29). For Rtt109-Vps75, the amount of product formed over time produced a linear curve yielding a rate of 0.53 ± 0.02 s^-1. For Rtt109 alone, the amount of acetylated product measured over time (0–480 s) yielded a linear curve, indicating that the rate-limiting step for both unactivated Rtt109 and Rtt109-Vps75 is the chemical step of acetyl transfer. As observed for the complex, a lag or a burst phase was undetectable for Rtt109 alone. In addition, the linear curves yielded rates (2.30 ± 0.03 × 10^-3 s^-1) that are consistent with steady-state k_cat values (2.3 ± 0.7 × 10^-3 s^-1); thus, the stimulated activity reflects enhanced acetyl transfer and not a change in the rate of substrate binding or product release.

Fig. 1.

Enzymatic characterization of Rtt109 activation by Vps75. (A) H3 saturation curves were performed to determine the K_m and k_cat for Rtt109 with varying amounts of chaperone. Data from two trials performed in duplicate are shown. Kinetic constants obtained from the saturation curves are listed below with standard error. (B) Arrhenius plots of Rtt109 and Rtt109-Vps75. The natural logarithm of k_cat values for Rtt109 alone (circles) and Rtt109-Vps75 (triangles) are plotted against the reciprocal of absolute temperature. Data were fitted to the natural logarithm of the Arrhenius equation, and activation energy values with standard error were calculated using the slope of the lines.

We then investigated the possibility that Vps75 stimulates the enzymatic activity of Rtt109 by contributing an ionizable catalytic residue to the active site. We have previously shown that the k_cat pH profile of Rtt109-Vps75 exhibits a single critical ionization (pK_a = 8.5), which likely corresponds to the substrate lysine that must be unprotonated for catalysis (29). To determine if Vps75 contributes a general base to the active site of Rtt109, we performed a k_cat pH profile analysis of Rtt109 alone (Fig. S1B). The unactivated enzyme yielded a similar pK_a(8.1 ± 0.1) to that determined for the activated complex (pK_a = 8.5) (29), indicating that Vps75 does not enhance acetyl transfer by contributing an active-site general base or by lowering the observed pK_a.

We envisioned two scenarios for the activation of Rtt109 by Vps75. First, Rtt109 could exist in an equilibrium between active and inactive conformers, where only a small fraction of Rtt109 is capable of catalyzing the acetyl transfer. In this case, Vps75 could enhance the rate of acetyl transfer by stabilizing the active conformation of Rtt109. Second, formation of the Rtt109-Vps75 complex could enhance catalysis by lowering the activation energy of acetyl transfer; thus, the activated complex and Rtt109 alone represent two distinct transition states with different energies of activation. To distinguish between these possibilities, we determined the effect of temperature on the k_cat of Rtt109 and Rtt109-Vps75 using S. cerevisiae H3-H4 tetramers as the substrate. The k_cat values were determined at various temperatures, and the data were fitted using the natural logarithm of the Arrhenius equation ln(k) = (-E_a/R)(1/T) + ln(A) (Fig. 1B). The energy of activation, E_a, was calculated from the resulting slope of the plot, which is equal to -E_a/R, where R is the ideal gas constant, 8.314472 J K^-1 mol^-1. No significant difference was observed between the activation energy of Rtt109 alone (E_a = 6.0 ± 0.4 × 10⁴ J mol^-1) and Rtt109-Vps75 (E_a = 6.1 ± 0.4 × 10⁴ J mol^-1). These data are consistent with the pH rate analysis and argue against Vps75 contributing to transition state stabilization. Thus, Rtt109 alone and the Rtt109-Vps75 complex utilize the same transition state, supporting a model in which Vps75 enhances catalytic turnover via stabilization of the active conformation of Rtt109.

To provide further evidence for this model, we asked if the activation of Rtt109 at various chaperone concentrations could be attributed to the fraction of bound complex. We calculated the apparent dissociation constant () of Rtt109-Vps75 using the kinetic data of Fig. 1A. The concentration of Vps75 dimer was plotted against the fraction of catalytically active complex (Fig. S2A). The calculated value of 13 ± 3 nM is consistent with the K_d determined using an equilibrium binding method (10 ± 2 nM) (29). Because the magnitude of activation correlates with the fraction of bound complex, this result provides further evidence that catalytic activation is attributed to Vps75 binding and stabilization of the active Rtt109 conformation.

Structural Insights into the Rtt109-Vps75 Complex.

To provide a structural understanding of this unique HAT-chaperone complex, the complex of Rtt109 and Vps75 was crystallized for diffraction studies. Crystals of a His-tagged fragment of Rtt109 (residues 1–405) bound to a fragment of Vps75 (residues 9–223) were obtained after coexpression in Escherichia coli and purification through three chromatographic steps. The resulting crystallized complex diffracted to a 4.25-Å resolution (Table S2). Molecular replacement was performed using the previously determined Rtt109 (16) and Vps75 (8) structures as search models. The asymmetric unit contained two Rtt109 monomers and four Vps75 monomers, which were arranged into apparent trimeric arrangements with two Vps75 monomers capping a single Rtt109 monomer (Fig. 2A). Interestingly, the Vps75 monomers pack together by interaction of the long helices that mediate similar homodimeric Vps75 interactions previously observed (8, 10), although this arrangement was not a constraint used in the molecular replacement. The limited resolution of the diffraction data did not permit stable refinement of the Rtt109-Vps75 model; however, this structural model provided an excellent platform for directed biochemical experiments that probe the interfaces between subunits. Using native PAGE, Rtt109 was titrated with increasing concentrations of Vps75 and complex formation was assessed (Fig. S3). Consistent with the formation of a trimeric complex, 2 M equivalents of Vps75 to Rtt109 produced maximal levels of complex. These data are also in agreement with the kinetic data (Fig. 1 and Fig. S2A) as 2M equivalents of Vps75 to Rtt109 resulted in near maximal activation.

Fig. 2.

Structure of the Rtt109-Vps75 complex. (A) 2F_o-F_c difference map of the Rtt109-Vps75 complex. Rtt109 is shown in green and the Vps75 homodimer is shown in blue. (B) Electrostatic maps of Vps75 dimer (Top) and Rtt109 (Bottom) illustrate the charge–charge interaction surfaces seen in the crystal structure. The active-site channel is positioned directly in the center of the substrate cavity formed by the complex. Structural figures and electrostatic maps were made using MacPyMOL.

To investigate the importance of the two electrostatic interfaces observed between subunits within the Rtt109-Vps75 complex (Fig. 2B), two sets of amino acid substitutions in Rtt109 were generated and functionally assessed for their ability to affect Vps75-dependent histone acetylation. Within the two α5 helices of the Vps75 dimer, residues R173 and K177 are predicted to interact with Rtt109 through E374 and E378 on helix α9 and E299, E300, and D301 on helix α6 (Fig. 3). Thus, we made three sets of substitutions within Rtt109 at these residues [E374A/E378A (A2), E374K/E378K (K2), and E299K/E300K/D301K (K3)] (Fig. 3). To verify that the point substitutions did not affect the overall structural integrity of Rtt109, steady-state kinetics of the unactivated enzyme variants were performed (Fig. 3). All variants showed similar activity toward H3 in comparison to wild-type enzyme, suggesting that these substitutions do not grossly affect the structure of Rtt109 in the absence of Vps75. We then assessed the ability of Vps75 to activate the Rtt109 variants by determining the initial rates of acetylation at increasing chaperone concentrations (1–25× relative to Rtt109) under saturating substrate conditions (Fig. 3). The data show a > 10-fold reduction in activation for the charge reversal variants at both interfaces, whereas the neutralization variant had a less dramatic effect. If the loss in activation results from a decrease in binding affinity between Rtt109 and Vps75, then the k_cat for wild type and the variants should be equivalent but require more Vps75 to reach the k_cat for K2 and K3; however, the k_cat for these variants is >10-fold lower than that of wild-type Rtt109. Equilibrium binding measurements yielded K_d values for K2 and K3 of 12 ± 1 and 16 ± 3 nM, respectively (Fig. S2); thus, both variants are able to bind to Vps75 with similar affinity as wild-type Rtt109 [K_d = 10 ± 2 nM (29)]. We next asked if the point substitutions in Rtt109 caused a change in the K_m value for H3. Steady-state kinetic analyses of Rtt109-Vps75 variants K2 and K3 revealed no significant change in H3 K_m (7.5 ± 1 and 4.6 ± 1 μM, respectively, Fig. S4), relative to wild-type Rtt109 (6.5 ± 2 μM, Fig. 1A). These results suggest that the K2 and K3 substitutions do not significantly alter substrate binding or the overall affinity between Rtt109 and Vps75. Native PAGE was performed to provide additional evidence of the variants to form complexes with Vps75 (Fig. S5). Although the Rtt109 K2 mutant generated Rtt109-Vps75 complexes that were indistinguishable from that of wild type, interestingly, Rtt109 K3-Vps75 generates a slightly faster-running complex, suggesting that the hydrodynamic properties are different from that observed with wild-type Rtt109 and the K2 variant (Fig. S5B). However, quantitative titrations revealed that the K3-Vps75 complex forms in a 1∶2 M ratio (Fig. S5C), just as that observed with the wild-type complex (Fig. S3 A and B). Collectively, these observations indicate that perturbation of the electrostatic interfaces yields Rtt109-Vps75 complexes that fail to attain full catalytic activation, but are otherwise similar. To demonstrate that the impairments in catalytic activation are specific to Rtt109-Vps75 contact surfaces (Fig. 3), we investigated the ability of A2, K2, and K3 variants to be activated by Asf1 using H3-H4 as substrate (Fig. S6A). The results indicate that the Rtt109 point substitutions (A2, K2, and K3) caused no significant change in Asf1-dependent acetylation compared to wild-type Rtt109.

Fig. 3.

Variants of Rtt109 at the proposed Rtt109-Vps75 complex interface. Point substitutions were made at acidic residues of Rtt109 (green) proposed to interact with basic patches of Vps75 (blue overlaid with a vacuum electrostatics surface generated in MacPyMOL). E374 and E378 were replaced with either Lys or Ala, whereas E299, E300, and D301 were changed to Lys. Initial rates of acetylation by Rtt109 variants alone or with increasing [Vps75] were measured to determine k_cat values for each Rtt109 variant. Kinetic constants obtained from three trials of samples performed in duplicate are listed with error being one standard deviation for unactivated values and the standard error for Rtt109-Vps75 complex values.

To provide in vivo support for the functional importance and chaperone specificity of the electrostatic interaction sites revealed in our biochemical and structural studies, we investigated the Asf1- and Vps75-dependent cellular functions of these Rtt109 variants. Yeast strains carrying the mutant rtt109-K2 and -K3 alleles were created and grown on either synthetic media [synthetic complete media lacking leucine (SC-Leu)] as a positive control for cell growth or on rich media [yeast extract/peptone/dextrose (YPD)] containing camptothecin (CPT, 8 μM) to measure genotoxic stress resistance (Fig. S6B and Fig. 4A). CPT covalently traps topoisomerase I leading to double-stranded DNA breaks (30), making Rtt109-Asf1 (and not Rtt109-Vps75) stimulated histone deposition essential for survival (31, 32). Both Rtt109 variants as well as wild-type Rtt109 were able to restore CPT resistance to the rtt109Δ strain, suggesting that the acetylation of H3K56 by Rtt109-Asf1 is unaffected by the K2 or K3 substitutions.

Fig. 4.

Variants of Rtt109 do not affect Rtt109-Asf1 activity in vivo. (A) Wild-type Rtt109 and the Rtt109 variants are able to rescue genome stability of the rtt109Δ strain. Yeast strains carrying the indicated RTT109 alleles were plated on rich media (YPD) containing 8 μM camptothecin (CPT) to indicate resistance to genotoxic stress. Plates were grown at 30 °C for 3 d prior to photography. (B) Histone H3 K9, K23, K27, and K56 are the primary sites of acetylation on H3-H4 tetramers for Rtt109-Vps75, as determined by quantitative MS/MS analysis. (C) H3K56ac is not affected by the point substitutions in Rtt109; however, H3K9ac and H3K27ac are reduced in the rtt109Δgcn5Δ strain with the mutant Rtt109 allele for K3 substitutions. Whole-cell alkaline lysis extracts of the indicated strains were analyzed by immunoblotting, with the indicated antiacetyl histone antibodies and reprobed with anti-H3 antibody as the loading control.

To assess the in vivo effect of Rtt109 variants on Vps75-dependent H3 tail acetylation, we examined the acetylation of H3 by immunoblotting. We previously demonstrated that Rtt109-Vps75 readily acetylates residues K9, K14, and K23 in vitro, and K56 on free H3 alone and residues K9 and K23 in cells (8). Here, we examined the primary sites of Rtt109-Vps75-catalyzed acetylation of reconstituted H3-H4 tetramers. Using quantitative MS/MS analysis, the acetylation sites were determined after 30 min of reaction, which represents non-steady-state conditions (Fig. 4B). Acetylation of H3 on K9 (100%), K23 (82%), K27 (31%), and K56 (100%) was clearly observed. Under steady-state conditions, we compared the site specificity of the wild type and the K3-Vps75 complex toward free histone H3. Consistent with previous observations (8, 27), the Rtt109-Vps75 complex displays a marked preference for H3K9. Although the rate was > 10 times slower (Fig. 3), the K3-Vps75 complex displays a similar preference for H3K9 (Fig. S6C), suggesting that the K3 variant does not alter the enzyme specificity. Once we verified the sites of acetylation on the H3-H4 substrate, we used antibodies against H3K9ac, K27ac, and K56ac to investigate the Vps75-dependent activity of Rtt109 variants in vivo. Whole-cell extracts of rtt109Δ and rtt109Δ gcn5Δ strains carrying the plasmid-borne RTT109 alleles were probed for acetylation at K56, K9, and K27 of histone H3. The latter strain was used to investigate acetylation at H3K9 and H3K27, as both Rtt109 and Gcn5 acetylate these sites (8, 26, 27). In agreement with the genome stability data (Fig. 4A), both K2 and K3 variants supported normal levels of H3K56 acetylation (Fig. 4C and Fig. S6 D and E). Interestingly, although normal levels of H3K9 and H3K27 acetylation were observed in cells expressing wild-type Rtt109 and the variant Rtt109-K2, the variant Rtt109-K3 generated reduced levels of K9 and K27 acetylation in the rtt109Δgcn5Δ strain. These findings were supported by immunoblotting titrations of each variant using the double knockout strain (Fig. S6 D and E). Thus, we have developed variants of Rtt109 that selectively perturb the Vps75-dependent acetylation of H3 in vitro and in vivo. Additionally, the results demonstrate that the Rtt109-Vps75 interfaces revealed with our crystal model are unique to the functional interaction between Rtt109 and Vps75, and that activation of Rtt109 by Asf1 is mediated through a distinct mechanism.

A Model for H3-H4 Dimer Acetylation by Rtt109-Vps75.

With a structural and biochemical model of the Rtt109-Vps75 complex in hand, we next investigated the relevant histone substrates that would effectively fit into the cavity formed between Rtt109-Vps75. From our structural model, the available space for substrate binding is approximately 27 × 32 Å, which would allow for the binding of an H3-H4 dimer but not a tetramer. We therefore modeled the histone H3-H4 dimer (33) into the Rtt109-Vps75 structure by positioning Lys56 from an H3-H4 dimer into the Rtt109 active site and minimizing structural clashes between the models (Fig. 5). This position is meant to illustrate a possible arrangement of the Rtt109-Vps75-H3-H4 complex and to demonstrate the capacity of the Rtt109-Vps75 complex to accommodate the histone dimer. However, this model does not exclude the potential for structural conformations that could take place upon substrate binding, which would allow for alternative arrangements of histones.

Fig. 5.

H3-H4 dimer is a substrate for Rtt109-Vps75. Model of the Rtt109-Vps75-H3-H4 complex. K56 from the histone dimer was manually positioned in the Rtt109 active site, and structural clashes were minimized between the models. Kinetic analysis with yH3-H4 tetramer or yH3(A110E)-H4 dimer. Kinetic constants obtained from substrate saturation curves of three trials of samples performed in duplicate are listed with standard error.

Histone H3-H4 complexes exist in an equilibrium between heterotetrameric and heterodimeric forms (34, 35); thus from previous kinetic analyses, it was unclear if the H3-H4 dimer is an efficient substrate for Rtt109-Vps75. The ability of Rtt109-Vps75 to acetylate a dimeric form of H3-H4 was assessed kinetically. We compared the steady-state acetylation rates of reconstituted histone tetramers versus a variant histone construct, H3(A110E)-H4, which has been demonstrated to exist in only dimeric form in solution (34). Substrate saturation curves were fitted to the Michaelis–Menten equation, revealing k_cat values of 0.41 ± 0.02 s^-1 and 0.39 ± 0.04 s^-1 and K_m values of 1.4 ± 0.4 μM and 2.4 ± 0.7 μM for H3-H4 and H3(A110E)-H4, respectively (Fig. 5). Thus, both substrates display similar efficiency within the error of the experiment. The results indicate that the H3-H4 mutant dimer is an efficient substrate and that Rtt109-Vps75 does not require an obligate H3-H4 tetramer for acetylation. Collectively, the biochemical evidence along with our structural model provides compelling evidence that the activated, high-affinity Rtt109-Vps75 trimer represents a functionally relevant complex that can acetylate an H3-H4 dimer.

Discussion

Rtt109-Vps75 and Rtt109-Asf1 complexes are associated with discrete biological functions in vivo and catalyze different histone acetylation reactions. However, the mechanisms by which Vps75 and Asf1 activate Rtt109 and direct lysine specificity are unknown. Here we investigated the mechanism of activation of Rtt109 by Vps75 using biochemical, structural, and genetic approaches. We demonstrated that Rtt109 can be activated > 250-fold by Vps75 via an increase in the rate of acetyl transfer, which occurs through the stabilization of the catalytically active form of Rtt109. To gain insight into the specific interactions of Rtt109 and Vps75, we have produced a molecular replacement solution of the Rtt109-Vps75 complex. The subunit arrangement was comprised of a 1∶2 complex of Rtt109-Vps75 that suggested two important electrostatic contact sites exist between the enzyme and chaperone. We validated the two distinct interfaces biochemically and genetically using point substitutions. These structure-based variants disrupted Vps75-dependent activity and function but did not adversely affect the Asf1-dependent functions of Rtt109 in vitro and in vivo. Additionally, the structure revealed a charged cavity capable of acetylating a heterodimer of H3-H4. By demonstrating that the H3-H4 dimer is an efficient substrate for Rtt109-Vps75, we provided kinetic evidence for a model involving a functional Rtt109-Vps75-H3-H4 complex (Fig. 5).

Several lines of biochemical evidence suggest that interaction surfaces other than those available in our structural model contribute to the formation of the Rtt109-Vps75 complex. However, the limited resolution of the structure impedes our ability to model potential interfaces that are not present in the previously determined Rtt109 and Vps75 structures, which were used as search models. We demonstrated that although the two electrostatic contact sites illustrated in the structural model are important for catalytic activation, they contribute very little to overall binding affinity (Fig. 3 and Fig. S2B). Though not observed in any solved structures to date, previous studies have implicated important binding regions within residues 130–179 of Rtt109 (16, 36) and the α8 helix (residues 209–222) of Vps75 (28). Consistent with this idea, the Rtt109-Vps75 complex is stable to high salt and requires the use of strong hydrophobic chromatography to disrupt the native complex (29). Based on these data, we propose that the high-affinity interaction between Rtt109 and Vps75 is mediated primarily through these hydrophobic interfaces, likely involving the unstructured loop of Rtt109 (residues 130–179). The electrostatic contact surfaces, which we investigated in the present study, are critical to permit transition to the fully active conformer of Rtt109 and allow for the optimal positioning of acetyl CoA and the attacking lysine from histone substrates.

This study represents the biochemical, structural, and genetic analysis of a HAT-regulatory subunit complex. The activation mechanism of Rtt109-Vps75 provides a valuable framework for understanding the molecular regulation of other HATs within multisubunit complexes. For example, it appears likely that the Esa1 and Tip60 HATs experience a similar regulatory mechanism (5, 37). Interaction of Esa1 with two subunits (Epl1 and Yng2), stimulates the k_cat of Esa1 by ∼100-fold, without alteration of the K_m for histone substrates (5). Stabilization of the active HAT conformation might be one critical method of minimizing spurious activity, allowing the cell to coordinate the timing and specificity of acetylation by linking transferase activity to the amount or type of accessory proteins.

Methods

Reagents.

Reagents were purchased from Sigma-Aldrich, Fisher Scientific, or RPI unless otherwise noted. [³H]-acetyl CoA (2–25 Ci/mmol) was obtained from Morevek, acrylamide/bisacrylamide from Bio-Rad, site-directed mutagenesis kits from Stratagene, anti-H3K9ac and K27ac from Upstate, and anti-H3 from Abcam.

Expression and Purification of Proteins.

Recombinant untagged Rtt109, His₆-tagged Rtt109, His₆-tagged Vps75, His₆-tagged Asf1, and Xenopus laevis and S. cerevisiae histones were expressed in either Rosetta 2 (DE3) pLysS (Novagen) or BL21-CodonPlus (DE3)-RIPL (Stratagene) competent cells and purified as described previously (8, 29, 33). Rtt109 and Vps75 concentrations were determined as previously described (29). Histone concentrations were determined using extinction coefficients of ε = 4,040 for X. laevis H3 and ε = 2,560 and 5,120 M^-1 cm^-1 for S. cerevisiae H3 and H4, respectively, which were derived from the method described by Gill and von Hippel (38). S. cerevisiae H3-H4 oligomers and H3(A110E)-H4 dimers were reconstituted as previously described (39). Unincorporated H3 and H4 monomers were removed by purification over a HighLoad 16/60 Superdex 200 prep grade column (GE Healthcare). Protein concentrations were determined using an extinction coefficient of ε = 7,680 M^-1 cm^-1.

HAT Assays.

Reactions were analyzed by filter binding and scintillation counting of products (40). Unless otherwise noted, reaction buffers were comprised of 50 mM Tris (pH 7.5 at 25 °C) and 1 mM DTT and reaction contained 40 μM [³H]-acetyl CoA (0.2–0.4 Ci/mmol). The details of specific experiments are described in SI Text.

Structure Determination of 6xHisRtt109(1-405)-Vps75(9-223).

The details of expression, purification, crystallization, X-ray diffraction data collection, and structure determination are described in SI Methods. Briefly, Rtt109(1-405) and Vps75(9-223) were coexpressed in E. coli using a pET-derived vector containing a 6xHis epitope tag and a pET3a vector, respectively. Recombinant protein was purified using three chromatographic steps, and crystals were generated in a hanging drop vapor diffusion crystallization experiment. Diffraction data were indexed and scaled using HKL2000 (41). Molecular replacement was carried out using Phaser (42) with the previously determined Rtt109 structure (16) and a monomer Vps75 structure (8) as search models.

Mass Spectrometry Quantitation Assay.

Enzymatic acetylation sites were determined using chemical acetylation combined with MS/MS peptide sequencing as described previously (43). The details of specific experiments are described in SI Text.

Yeast Assays.

The details of yeast strains are described in SI Text. For immunoblotting experiments, alkaline whole-cell extracts were prepared from 2 mL of cells grown to an OD₆₀₀ of 0.8 in YPD or Leu media (44). The final volume of the extracts was 30 μL, from which 2–10 μL was loaded onto 17% SDS-PAGE for Western blot analysis.