Structure of the Yeast Hst2 Protein Deacetylase in Ternary Complex with 2′-O-Acetyl ADP Ribose and Histone Peptide

Structure, Vol. 11, 1403–1411, November, 2003, 2003 Elsevier Science Ltd. All rights reserved.

DOI 10.1016/j.str.2003.09.016

Structure of the Yeast Hst2 Protein Deacetylase in Ternary Complex with 2ⴕ-O-Acetyl ADP Ribose and Histone Peptide Kehao Zhao1, Xiaomei Chai,1 and Ronen Marmorstein1,2,* 1 The Wistar Institute 2 The Department of Chemistry University of Pennsylvania Philadelphia, Pennsylvania 19104

Summary Sir2 proteins are NADⴙ-dependant protein deactylases that have been implicated in playing roles in gene silencing, DNA repair, genome stability, longevity, metabolism, and cell physiology. To define the mechanism of Sir2 activity, we report the 1.5 A˚ crystal structure of the yeast Hst2 (yHst2) Sir2 protein in ternary complex with 2ⴕ-O-acetyl ADP ribose and an acetylated histone H4 peptide. The structure captures both ligands meeting within an enclosed tunnel between the small and large domains of the catalytic protein core and permits the assignment of a detailed catalytic mechanism for the Sir2 proteins that is consistent with solution and enzymatic studies. Comparison of the ternary complex with the yHst2/NADⴙ complex, also reported here, and nascent yHst2 structure also reveals that NADⴙ binding accompanies intramolecular loop rearrangement for more stable NADⴙ and acetyllysine binding, and that acetyl-lysine peptide binding induces a trimer-monomer protein transition involving nonconserved Sir2 residues. Introduction The yeast Sir2 protein was first identified as a factor required for all forms of transcriptional silencing, including the silencing of silent mating loci, telomeres, and rDNA, through its promotion of higher order chromatin structure (reviewed in Guarente, 1999; Loo and Rine, 1995; Sherman and Pillus, 1997). The biochemical basis for this activity was elucidated when Guarente and coworkers reported that the Sir2 protein from yeast and mouse harbored NAD⫹-dependant histone deacetylase activity (Imai et al., 2000). The observation that the deactylase activity of Sir2 required NAD⫹ set it apart from other deactylase enzymes that catalyze a similar hydrolysis reaction in the absence of a cosubstrate (reviewed in Marmorstein, 2001) and fueled studies linking NAD⫹ hydrolysis by Sir2 to metabolism and aging (reviewed in Guarente, 2000; Tissenbaum and Guarente, 2001). Subsequent studies of Sir2 homologs ranging from bacteria to human demonstrated that this family of enzymes also deactylate nonhistone substrates in vivo, including the archaeal chromatin-associated protein Alba (Bell et al., 2002), the p53 tumor suppressor protein (Luo et al., 2001; Vaziri et al., 2001), ␣-tubulin (North et al., 2003), and bacterial acetyl-CoA synthetase (Starai et al., 2002). *Correspondence: [email protected]

Together, these studies suggest that Sir2 proteins may have broad cellular functions ranging from the regulation of gene expression to cellular metabolism. The mechanism for Sir2 activity has been extensively studied by several laboratories at both the structural and enzymatic levels. The structures of the conserved catalytic domain of Sir2 homologs from archaea (Avalos et al., 2002; Chang et al., 2002; Min et al., 2001) and human (Finnin et al., 2001) reveal an elongated structure containing a large Rossmann fold domain characteristic of NAD⫹ binding proteins and a small domain containing a structural zinc ion. A series of loops traverse between the large and small domain forming a pronounced extended cleft between the two protein domains. Binary NAD⫹ and acetyl-lysine containing complexes bound to two different archaeal Sir2 homologs reveal that the cosubstrates are bound in adjacent clefts between the small and large domains (Avalos et al., 2002; Min et al., 2001). Although several disparate enzymatic mechanisms for the reaction have been proposed (Avalos et al., 2002; Chang et al., 2002; Jackson and Denu, 2002; Min et al., 2001; Sauve et al., 2001; Tanner et al., 2000), the available data are most consistent with initial formation of a 2⬘-O-acetyl ADP ribose product that proceeds through an ␣-1⬘-acetyl ADP ribose intermediate, in addition to formation of nicotinamide and lysine products (Sauve et al., 2001). S. cerevisiae Hst2 (homologue of Sir two-2, yHst2) is the most universally conserved member of the Sir2 homologs, is cytoplasmically localized, and can modulate nucleolar and telomeric silencing when overexpresed in yeast (Perrod et al., 2001). Although the physiological target for yHst2 has not yet been established, yHst2 accounts for the majority of detectable NAD⫹-dependant deactylase activity in yeast cell extracts (Smith et al., 2000), and yHst2 has been shown to exhibits greater activity than ySir2 for histone substrates in vitro (Landry et al., 2000). We have previously reported on the structure of intact, nascent yHst2, revealing a conserved catalytic domain and nonconserved C- and N-terminal domains that autoinhibit NAD⫹ and acetyl-lysine binding, respectively (Zhao et al., 2003). Attempts to obtain substrate complexes with intact yHst2 were unsuccessful, leading us to prepare a C-terminal deletion construct missing the NAD⫹ autoinhibitory domain for cocrystallization with substrates. Using this yHst2 protein construct, we have crystallized and determined the 1.5 A˚ resolution structure of yHst2 in ternary complex with NAD⫹ and an 11 residue histone H4 peptide centered around lysine-16, the preferred substrate for yeast Sir2 (Imai et al., 2000) (Table 1). This structure allows us to define a conserved mechanism for catalysis by Sir2 proteins. We also report the structure of the binary yHst2/NAD⫹ complex (Table 1) for comparison with the ternary complex and nascent yHst2 to define the precise structural rearrangements that accompany binding of substrates and catalysis by the Sir2 proteins.

Structure 1404

Table 1. Data Collection and Refinement Statistics Data Set

Hst2/NAD⫹

Hst2/NAD⫹/Histone H4

Space group Cell parameters

C2 a ⫽ 146.16 A˚, b ⫽ 109.68 A˚, c ⫽ 85.83 A˚, ␤ ⫽ 104.9⬚ 30–2.7 35492 98.8 (99.9) 4.1 15.0 (10.9) 6.6 (20.7)

R3 a ⫽ b ⫽ 98.42 A˚, c ⫽ 77.25 A˚

18.6 (26.5) 24.7 (33.5)

18.3 (21.8) 20.9 (23.1)

6986/49.2

2270 84/22.4

Resolution (A˚) Unique reflection Completeness (%)a Multiplicity I/Ioa Rmergeb (%) Refinement statistics Rworkingc (%) Rfreed (%) Number of atoms/B factors (A˚2) Protein H4 histone ADP-ribose AADPR Water Zn,Cl ions Rms deviations Bond length (A˚) Bond angle (⬚)

30–1.5 43331 96.9 (93.9) 6.1 31.5 (4.2) 4.9 (17.2)

108/36.8 162/40.1 3,4/59.5,42.4

39/10.6 303/30.2 1,0/21.7

0.0066 1.35

0.0047 1.22

a

Values in parentheses are from the highest resolution shell. Rmerge ⫽ ⌺|I ⫺ ⬍I⬎|/⌺⬍I⬎. c Rworking ⫽ ⌺||Fo| ⫺ |Fc||/⌺|Fo|. d Rfree ⫽ ⌺T||Fo| |Fc||/⌺bT|Fo| (where T is a test data set of percentage [10% for Hst2/NAD and 5% for Hst2/NAD/H4] of the total reflections randomly chose and set aside in prior to refinement). b

Results and Discussion Overall Structure of the yHst2/NADⴙ/Histone H4 Complex The ternary yHst2/NAD⫹/histone H4 complex shows that the two substrates enter the protein through opposite sides of a cleft between the small and large domains of the catalytic core and that the functional groups of both the protein and substrates are buried within a protein tunnel that harbors the region of highest conservation within the Sir2 proteins (Figure 1). The high-resolution structure reveals that we have not cocrystallized a true Michaelis-Menten complex, but instead unambiguously shows both an acetyl-lysine substrate and a 2⬘-O-ADP ribose reaction product, consistent with the reaction product identified previously (Jackson and Denu, 2002; Sauve et al., 2001) (Figures 2A and 2B). We presume that the ternary complex between protein, acetyl-lysine, and NAD⫹ substrates initially formed, producing the 2⬘-O-ADP ribose reaction product, which remained bound while the deacetylated lysine substrate dissociated and was replaced with a new acetyl-lysine substrate producing the ternary complex that we observe in the crystals. We propose that this structure represents a product-inhibition complex consistent with our solution studies demonstrating inhibition of the enzyme at high NAD⫹ concentrations (data not shown). In the ternary complex, the carbonyl group of acetyllysine is 3.1 A˚ away from the 1⬘-carbon of the ribose ring, and His 135 is 2.8 A˚ away from the 3⬘ oxygen of the ribose ring. Histidine 135 is strictly conserved within the Sir2 proteins, and its mutation to alanine has been shown to result in background levels of deacetylase activity, but detectable levels of nicotinamide exchange

activity (Min et al., 2001). In addition, Asn 116, another strictly conserved residue that shows nearly background levels of catalysis when mutated to alanine (Finnin et al., 2001), forms a 2.7 A˚ H bond to the most wellordered water molecule in the ternary complex. This water molecule is located 2.8 A˚ away from the 3⬘-oxygens of the ribose ring, suggesting that it may play an important catalytic role. The potential significance of this water molecule is further supported by its presence in each of the three yHst2 structures reported by us (this work; and Zhao et al., 2003), as well as its presence in each of the Sir homolog crystal structures reported by others (Avalos et al., 2002; Chang et al., 2002; Min et al., 2001), with SIRT2 offering the one exception (Finnin et al., 2001). Mechanism of Catalysis Taking together the reaction products identified by Schramm and coworkers (Sauve et al., 2001), the mutational data described above, and the high-resolution ternary complex reported here, we propose the catalytic mechanism shown in Figure 2C. In this mechanism, the carbonyl oxygen of acetyl-lysine has the appropriate distance and geometry to carry out nucleophilic attack of the 1⬘ carbon of the ribose ring nucleating the displacement of the nicotinamide ring, with inversion of stereochemistry at the 1⬘ position and formation of an O-alkyl amidate intermediate. The imidazole nitrogen of His 135 has the appropriate distance and geometry to then function as a general base to directly deprotonate the 3⬘ hydroxyl of the ribose ring. We then propose that proton migration from the 2⬘ to 3⬘ oxygen of the ribose ring results in formation of a cyclic acyldioxalane involving the 1⬘ and 2⬘ oxygens of the ribose ring. We then

Structure of Ternary Yeast Hst2 Complex 1405

Figure 1. Overall Structure of the Ternary yHst2/2⬘-O-Acetyl ADP Ribose/Histone H4 Complex (A) Schematic structure of the ternary complex with the protein shown in blue, with different shades distinguishing the small and large domains of the catalytic core, and the connecting loops colored in purple. The 2⬘-O-acetyl ADP ribose is colored in yellow and the histone H4 peptide is colored in green. (B) A surface representation of the ternary complex highlighting the regions of strict (blue) and conservative (cyan) conservation within the Sir2 family. In the blow-up view on the right, the ␤1-␣2-loop and part of the ␣2 helix are rendered as a loop, and Phe 184 and Val 228 were deleted so that the interior of the catalytic “tunnel” could be presented. (C) Sequence and secondary structure assignment of yHst2 using the same color-coding as in (A). Residues with strict and conservative conservation within the Sir2 proteins are indicated in dark and light shading, respectively. Residues that contact NAD⫹ and the histone H4 peptide substrates in the ternary complex are indicated with closed triangles and circles, respectively.

propose that the “structurally conserved” water molecule that is held in place by, and possibly also activated by, Asn 116 carries out nucleophilic attack of the cyclic acyldioxalane, resulting in the collapse of the cyclic intermediate to the 2⬘-O-ADP ribose and lysine reaction products. Modeling of the cyclic acyldioxalane suggests that this water molecule would be about 4.5 A˚ away, suggesting that some structural rearrangement of the protein or NAD⫹ molecule would have to occur to facilitate this step of the reaction. His 135 is also in position to mediate the reprotonation of the lysine reaction product during the collapse of the cyclic intermediate. Another residue that is strictly conserved among the Sir2 pro-

teins and whose mutation significantly effects catalytic function is Ser 36 (Min et al., 2001). Previous studies have speculated that this residue might mediate formation of an ADP-ribose intermediate (Min et al., 2001) or be involved in general protein stability (Chang et al., 2002). While our structure is incompatible with the formation of a Ser 36 ADP-ribose intermediate, it is compatible with a role for Ser 36 in protein stability, as it makes direct and water-mediated contacts to residues within the large domain of the catalytic core. Specifically, Ser 36 is within hydrogen bonding distance of the N␦ of Asn 116 (3.7 A˚) and the O␦ of Gln 118 (2.8 A˚), another strictly conserved and mutationally conserved residue within

Structure 1406

Figure 2. The yHst2 Active Site (A) Stereo view of the yHst2 active site highlighting the acetyl-lysine substrate (green) and the 2⬘-O-ADP-ribose product (yellow) as well as the key catalytic residues (cyan) and a catalytic water molecule (red). (B) Simulated-annealing omit electron density map of the active site contoured at 3.5 ␴. (C) Proposed catalytic mechanism for yHst2.

the Sir proteins (Guarente, 2000). A well ordered water molecule also links these three residues via hydrogen bonds. This hydrogen bond network may also serve to position Asn 116 appropriately for presenting the catalytic water molecule for the deacetylation reaction. Comparison with Binary yHst2/NADⴙ and Nascent yHst2 The binary yHst2/NAD⫹ complex reveals unambiguous density for all of the substrate except for the nicotinamide group. This is consistent with the Af1-Sir2/NAD⫹ structure, which also does not contain interpretable electron density for the nicotinamide group (Min et al., 2001), and we presume that in our structure this group is either highly mobile or prematurely hydrolyzed. A superposition of binary yHst2/ NAD⫹ with the ternary yHst2 complex reveals a high degree of superposition with an rms deviation over all atoms of 1.65 A˚2 (Figure 3A).

Surprisingly, the unacetylated NAD⫹ molecule in the binary complex also superimposes almost perfectly with the 2⬘-O-ADP ribose reaction product of the ternary complex with an rms deviation over all atoms of 0.29 A˚2. This observation suggests that NAD⫹ does not undergo significant structural rearrangement upon acetyl-lysine binding and catalysis, in contrast with some previous proposals that called for a rearrangement of NAD⫹ for catalysis (Avalos et al., 2002; Min et al., 2001). The NAD⫹ is primarily contacted by the ␤1-␣2-loop connecting the large and small domain of the catalytic core on one side and the ␤7-␣9 loop of the large domain on the opposite side (Figure 3B). Virtually all of the residues that contact the NAD⫹ substrate of the binary complex and the 2⬘-O-ADP ribose product of the ternary complex are conserved within the Sir2 family, and strikingly, three of these highly conserved residues (Val 182, Phe 184, and Val 228) also make critical interactions for acetyl-lysine

Structure of Ternary Yeast Hst2 Complex 1407

Figure 3. The yHst2-2⬘-O-Acetyl ADP Ribose Interface (A) Overlay of the ternary yHst2/2⬘-O-acetyl ADP ribose/histone H4 complex (cyan) with the binary yHst2/NAD⫹ complex (gray) and the nascent yHst2 protein (red). (B) Stereo view of yHst2-2⬘-O-acetyl ADP ribose interactions within the ternary complex. Hydrogen bonds are indicated with a dashed line. Residues that mediate van der Waals interactions are also shown. (C) Summary of yHst2-2⬘-O-acetyl ADP ribose interactions. Hydrogen bonds are indicated with a dashed line, and van der Waals interactions are indicated with a half-moon symbol. The residues highlighted in cyan and red highlight interactions with NAD⫹ that are conserved and nonconserved, respectively, with the protein-NAD⫹ interactions observed in the Af1-Sir2/NAD⫹ structure.

recognition, possibly serving to help direct NAD⫹ and acetyl-lysine hydrolysis to the same active site (Figures 3 and 4). A superposition of nascent yHst2 (Zhao et al., 2003) with the binary and ternary complexes reveals that the most structurally dynamic region of yHst2 is the ␤1-␣2 loop that is disordered in the nascent structure but becomes ordered upon NAD⫹ binding and mediates nearly half of the NAD⫹ contacts. Ordering of this loop upon NAD⫹ binding is consistent with the defined trace of the

corresponding loop of the Af1-Sir2/NAD⫹ complexes (Chang et al., 2002; Min et al., 2001) and the more poorly defined or disordered conformation of the corresponding loop in the reported Sir2 structures without bound NAD⫹ (Avalos et al., 2002; Finnin et al., 1999). The rearrangement of the ␤1-␣2 loop of yHst2 also appears to be correlated with a rigid-body rotation of the small domain relative to the large domain of the catalytic core, which in turn appears to reposition the ␤6-␣8 loop for more optimal acetyl-lysine interactions (Figure 3A). In-

Structure 1408

Figure 4. The yHst2-Histone H4 Interface (A) Stereo view of yHst2-histone H4 interactions within the ternary complex. Hydrogen bonds are indicated with a dashed line. Residues that mediate van der Waals interactions are also shown. (B) Summary of yHst2-histone H4 interactions. Hydrogen bonds are indicated with a dashed line, and van der Waals interactions are indicated with a half-moon symbol. For clarity, histone H4 side chains that do no participate in direct protein-peptide interactions are not shown. The residues highlighted in cyan and red highlight interactions with acetly-lysine peptide substrate that are conserved and nonconserved, respectively, with the protein-peptide interactions observed in the Af2-Sir2/p53 peptide structure. (C) The p53 peptide (purple) from the Af2-Sir2/p53 peptide structure and the “pseudosubstrate” from the nascent yHst2 structure (yellow) are overlayed with the histone H4 peptide (green) onto a surface representation of yHst2 from the ternary complex. Protein residues that make conserved interactions between the three substrates are indicated in blue, and protein residues that mediate variable interactions are indicated in red.

Structure of Ternary Yeast Hst2 Complex 1409

deed, residues in the ␤6-␣8 loop (Val 182, Phe 184, Gly 185, Glu 186, and Leu 188) are highly conserved within the Sir 2 proteins (Figure 1C) and mediate nearly half of the histone H4 contacts in the ternary complex (Figure 4B). The observed flexibility of the small domain of the catalytic core is consistent with the structural divergence of this domain between the structures of the differentially liganded Sir2 homologs (Avalos et al., 2002; Chang et al., 2002; Finnin et al., 1999; Min et al., 2001). Taken together, NAD⫹ binding to the Sir2 proteins appears to play both a catalytic and structural role in acetyl-lysine catalysis and binding. Acetyl-Lysine Binding Specificity by yHst2 The histone H4 peptide of the ternary complex is bound in a bent conformation adjacent to the NAD⫹ molecule. The histone substrate is flanked on one side by a short loop connecting the ␤6-strand of the small domain to the ␣8 helix of the large domain, and on the opposite side by the ␤7-␣9 loop of the large domain, the same region that flanks the NAD⫹ molecule (Figures 4A and 4B). Acetyl-Lys 16 makes the most extensive interactions with yHst2. Specifically, the aliphatic arm of this residue makes extensive van der Waals interactions with residues Val 182, Phe 184, and Leu 188 of the ␤6-␣8loop and residues Val 228 and Pro 230 of the ␤7-␣9 loop. Phe 184 also interacts with the acetyl methyl group of acetyl-lysine 16, and the acetyl carbonyl group H bonds to the 1⬘ OH of the ADP ribose ring. The backbone amino group of acetyl-lysine 16 also makes ␤ sheet interactions with the backbone of residues 185 and 186. Each of the residues that contact the acetyl-lysine are highly conserved among the Sir2 proteins, suggesting that the observed interactions are a conserved feature of the Sir2 proteins. Consistent with this proposal, the structure of Af2-Sir2 bound to a acetyl-lysine p53 peptide shows nearly identical interactions in this region (Avalos et al., 2002) (Figure 4B). Ten of the 11 histone H4 residues have well ordered main chain density, and 8 of the 10 residues also have well defined side chain density. Despite this, outside of acetyl-lysine 16, the yHst2-histone H4 peptide interactions are very limited and largely restricted to backbone interactions involving residues that are not conserved among the Sir2 proteins (Figure 4B). This relatively sparse set of protein-peptide interactions probably reflects the fact that histone H4 may not be the true in vivo substrate for yHst2 and/or that substrate specificity determinants may involve other regions of the Sir2 proteins and/or regions of the substrate that are distal to the acetyl-lysine site. Consistent with the latter argument, a superposition of the ternary complex reported here with the Af2-Sir2/p53 peptide complex and yHst2 bound to a “pseudo-substrate” within the autoinhibited homotrimeric form (Zhao et al., 2003) reveals that each of the acetyl-lysine substrates (or substrate mimics in the case of the autoinhibited yHst2) diverge greatly in path and

region of contact on the protein directly outside of the acetyl-lysine and the two flanking residues of the substrate (Figure 4C). Substrate-Dependant Oligomerization of yHst2 We have previously reported that intact nascent yHst2 forms a symmetrical homotrimer both in a crystal lattice and in solution (Zhao et al., 2003). This trimer is held together, in part, by the insertion of the N-terminal strand of one yHst2 protomer into the acetyl-lysine binding site of an adjacent protomer of the trimer, resulting in autoinhibition of the unliganded protein. Consistent with this finding, the binary yHst2/ NAD⫹ complex also forms a similar homotrimer, despite the different crystallization conditions and crystal parameters with the nascent protein (Figure 4D). In contrast, the ternary yHst2/ NAD⫹/ histone H4 complex does not show a similar trimer in the crystal lattice (Figure 4D). Instead, in the ternary complex, the N-terminal strand is disordered and the histone takes its place in the acetyl-lysine binding site. Based on this data, we propose that yHst2 undergoes a trimer-monomer transition upon acetly-lysine substrate binding, possibly setting apart the substrate-specific activity of the yHst2 member of the Sir2 proteins. Conclusions Taken together, the structural studies presented here on the yHst2 member of the Sir2 proteins, in combination with available solution and mutational studies, provide a detailed catalytic mechanism for NAD⫹-dependant histone deactylation, and a mode for NAD⫹ and acetyllysine binding that is likely to be general for this class of proteins. Given that eukaryotic organisms often have multiple Sir2 proteins, with presumably distinct biological functions, what may be the distinguishing features of these proteins? A comparison of the differently liganded yHst2 proteins reveals that the protein undergoes a trimer-monomer transition upon acetyl-lysine substrate binding. The lack of sequence conservation with the Sir2 proteins in the regions of yHst2 that mediate trimer formation suggests that this may be a unique property of yHst2. Also likely to be unique for different Sir2 proteins is substrate specificity. A comparison of different Sir2 proteins bound to acetyl-lysine or an acetyl-lysine mimic shows little commonality in protein-substrate interaction outside of the two residues that flank acetyllysine. This observation suggests that substrate specificity of different Sir2 proteins may be derived from regions outside of the immediate acetyl-lysine target site and possibly also outside of the catalytic Sir2 domain. Perhaps, a specific Sir2/protein substrate interface may share similarities to that seen at the protein-protein interface of the nascent yHst2 homotrimer (Zhao et al., 2003; and Figure 4D). Interestingly, the apparently poor degree of substrate specificity observed with Sir2 peptide substrates is also consistent with what has been observed for histone methyltransferase (Xiao et al., 2003) and his-

(D) Backbone overlay of yHst2/NAD⫹ (gray) and nascent yHst2 (cyan) homotrimers with the yHst2/2⬘-O-acetyl ADP ribose/histone H4 monomer (red). The ADP-ribose is highlighted in yellow, the histone H4 peptide is highlighted in green, and the C-terminal domain of nascent yHst2 is highlighted in purple.

Structure 1410

tone acetyltransferase proteins (Rojas et al., 1999). These ternary complexes also reveal a poor degree of substrate-specific interactions with peptides centered on their respective reactive lysine substrates. Uncovering the mode of substrate specificity by the Sir2 proteins is one of the major hurdles before us in understanding the mechanism of action of these unique deacetylase enzymes. Experimental Procedures The 64 residue C-terminal deletion construct of yHst2 was purified essentially as described previously (Zhao et al., 2003). The recombinant protein was produced in BL21/DE3 E. coli cells. The 6⫻His tagged protein was purified by Ni-NTA (Qiagen) and Superdex-75 gel filtration chromatography. The protein was concentrated to 20–30 mg/ml and stored at ⫺70⬚C prior for crystallization in 20 mM Tris-HCl (pH 8.5), 100 mM NaCl, and 10 mM DTT. Crystals of Hst2/ NAD⫹ and Hst2/NAD⫹/histone H4 peptide were grown at room temperature using the hanging drop vapor diffusion method. The Hst2/ NAD⫹ crystals were obtained by mixing 8 mg/ml protein, with a reservoir solution containing 1.2–1.4 M (NH4)2SO4, 50 mM HEPES (pH 7.0), and 10 mM MgSO4, and equilibrating over 0.5 ml of reservoir solution. The Hst2/NAD⫹/histone H4 crystals were grown using 10 mg/ml protein and reservoir solution consisting of 0.2 M NH4(OAc), 100 mM Na-citrate (pH 5.6), and 25% PEG 4000 . Before data collection, Hst2/NAD⫹ crystals were flash frozen in a reservoir solution supplemented with 15% glycerol, and Hst2/NAD⫹/histone H4 crystals were cryoprotected with a five-step gradual transfer into reservoir solution containing 25% glycerol. The data for Hst2/NAD⫹ and Hst2/NAD⫹/histone H4 crystals were collected on the F1 beamline at CHESS. All data were processed with DENZO and scaled with SCALEPACK (Gewirth et al., 1993). The structures of both complexes were solved with the program AMoRe (Navaza, 1994) using residues 1–294 from the nascent yHst2 structure (Zhao et al., 2003) as a search model. Structures were refined by simulated annealing and torsion angle dynamics in CNS (Brunger et al., 1998), with iterative manual adjustments of the models using the program O (Jones, 1978). At the later stages of refinement, individual atomic B factors were adjusted, and solvent molecules were modeled into the electron density map. The final models were checked for errors with composite-simulated annealing omit maps, and a final round of refinement resulted in models with excellent refinement statistics and geometry. The final model for yHst2/ NAD⫹ includes three molecules of yHst2 in the asymmetric unit. The proteins from both structures contain a disordered region from residues 207 to 214 that could not be modeled, and residues 1–4 of the ternary complex also could not be modeled, presumably due to disorder. Both structures also have excellent density for the ADP-ribose portion of NAD⫹, and the ternary complex also shows interpretable density for 10 of the 11 residues of the histone H4 peptide (12-KGGAKAcRHRKIL-22). The peptide side-chains K21, I21, and L22 could not be completely modeled, presumably due to disorder. Acknowledgments We thank A. Poux, W. Ho, D.M. Fitzgerald, D. McCafferty, and D. Christianson for useful discussions, and the staff at CHESS for assistance with beam line F1. This work was supported by NIH grants to R.M. and by a grant from the Commonwealth Universal Research Enhancement Program, Pennsylvania Department of Health, awarded to the Wistar Institute. Received: July 9, 2003 Accepted: July 14, 2003 Published: November 4, 2003 References Avalos, J.L., Celic, I., Muhammad, S., Cosgrove, M.S., Boeke, J.D., and Wolberger, C. (2002). Structure of a Sir2 enzyme bound to an acetylated p53 peptide. Mol. Cell 10, 523–535.

Bell, S.D., Botting, C.H., Wardleworth, B.N., Jackson, S.P., and White, M.F. (2002). The interaction of Alba, a conserved archaeal chromatin protein, with Sir2 and its regulation by acetylation. Science 296, 148–151. Brunger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., et al. (1998). Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. D54, 905–921. Chang, J.H., Kim, H.C., Hwang, K.Y., Lee, J.W., Jackson, S.P., Bell, S.D., and Cho, Y. (2002). Structural basis for the NAD-dependent deacetylase mechanism of Sir2. J. Biol. Chem. 277, 34489–34498. Finnin, M.S., Donigian, J.R., Cohen, A., Richon, V.M., Rifkind, R.A., Marks, P.A., Breslow, R., and Pavletich, N.P. (1999). Structures of a histone deacetylase homologue bound to the TSA and SAHA inhibitors. Nature 401, 188–193. Finnin, M.S., Donigian, J.R., and Pavletich, N.P. (2001). Structure of the histone deacetylase SIRT2. Nat. Struct. Biol. 8, 621–625. Gewirth, D., Otwinowski, Z., and Minor, W. (1993). The HKL Version 1.0 Manual (New Haven, CT: Yale University Press). Guarente, L. (1999). Diverse and dynamic functions of the Sir silencing complex. Nat. Genet. 23, 281–285. Guarente, L. (2000). Sir2 links chromatin silencing, metabolism, and aging. Genes Dev. 14, 1021–1026. Imai, S., Armstrong, C.M., Kaeberlein, M., and Guarente, L. (2000). Transcriptional silencing and longevity protein Sir2 is an NADdependent histone deacetylase. Nature 403, 795–800. Jackson, M.D., and Denu, J.M. (2002). Structural identification of 2⬘- and 3⬘-O-acetyl-ADP-ribose as novel metabolites derived from the Sir2 family of beta-NAD⫹-dependent histone/protein deacetylases. J. Biol. Chem. 277, 18535–18544. Jones, T.A. (1978). A graphics model building and refinement system for macromolecules. J. Appl. Crystallogr. 11, 268–272. Landry, J., Sutton, A., Tafrov, S.T., Heller, R.C., Stebbins, J., Pillus, L., and Sternglanz, R. (2000). The silencing protein SIR2 and its homologs are NAD-dependent protein deacetylases. Proc. Natl. Acad. Sci. USA 97, 5807–5811. Loo, S., and Rine, J. (1995). Silencing and heritable domains of gene expression. Annu. Rev. Cell Dev. Biol. 11, 519–548. Luo, J., Nikolaev, A.Y., Imai, S., Chen, D., Su, F., Shiloh, A., Guarente, L., and Gu, W. (2001). Negative control of p53 by Sir2alpha promotes cell survival under stress. Cell 107, 137–148. Marmorstein, R. (2001). Structure of histone deacetylases: insights into substrate recognition and catalysis. Structure 9, 1127–1133. Min, J., Landry, J., Sternglanz, R., and Xu, R.M. (2001). Crystal structure of a SIR2 homolog-NAD complex. Cell 105, 269–279. Navaza, J. (1994). AMoRe: an automated package for molecular replacement. Acta Crystallogr. A50, 157–163. North, B.J., Marshall, B.L., Borra, M.T., Denu, J.M., and Verdin, E. (2003). The human Sir2 ortholog, SIRT2, is a NAD⫹-dependent tubulin deacetylase. Mol. Cell 11, 437–444. Perrod, S., Cockell, M.M., Laroche, T., Renauld, H., Ducrest, A.L., Bonnard, C., and Gasser, S.M. (2001). A cytosolic NAD-dependent deacetylase, Hst2p, can modulate nucleolar and telomeric silencing in yeast. EMBO J. 20, 197–209. Rojas, J.R., Trievel, R.C., Zhou, J., Mo, Y., Li, X., Berger, S.L., Allis, D., and Marmorstein, R. (1999). Structure of the Tetrahymena GCN5 bound to coenzyme-A and a histone H3 peptide. Nature 401, 93–98. Sauve, A.A., Celic, I., Avalos, J., Deng, H., Boeke, J.D., and Schramm, V.L. (2001). Chemistry of gene silencing: the mechanism of NAD⫹dependent deacetylation reactions. Biochemistry 40, 15456–15463. Sherman, J.M., and Pillus, L. (1997). An uncertain silence. Trends Genet. 13, 308–313. Smith, J.S., Brachmann, C.B., Celic, I., Kenna, M.A., Muhammad, S., Starai, V.J., Avalos, J.L., Escalante-Semerena, J.C., Grubmeyer, C., Wolberger, C., and Boeke, J.D. (2000). A phylogenetically conserved NAD⫹-dependent protein deacetylase activity in the Sir2 protein family. Proc. Natl. Acad. Sci. USA 97, 6658–6663.

Structure of Ternary Yeast Hst2 Complex 1411

Starai, V.J., Celic, I., Cole, R.N., Boeke, J.D., and Escalante-Semerena, J.C. (2002). Sir2-dependent activation of acetyl-CoA synthetase by deacetylation of active lysine. Science 298, 2390–2392. Tanner, K.G., Landry, J., Sternglanz, R., and Denu, J.M. (2000). Silent information regulator 2 family of NAD-dependent histone/protein deacetylases generates a unique product, 1-O-acetyl-ADP-ribose. Proc. Natl. Acad. Sci. USA 97, 14178–14182. Tissenbaum, H.A., and Guarente, L. (2001). Increased dosage of a sir-2 gene extends lifespan in Caenorhabditis elegans. Nature 410, 227–230. Vaziri, H., Dessain, S.K., Ng Eaton, E., Imai, S.I., Frye, R.A., Pandita, T.K., Guarente, L., and Weinberg, R.A. (2001). hSIR2(SIRT1) functions as an NAD-dependent p53 deacetylase. Cell 107, 149–159. Xiao, B., Jing, C., Wilson, J.R., Walker, P.A., Vasisht, N., Kelly, G., Howell, S., Taylor, I.A., Blackburn, G.M., and Gamblin, S.J. (2003). Structure and catalytic mechansim of human histone methyltransferase SET7/9. Nature 421, 652–656. Zhao, K., Chai, X., Clements, A., and Marmorstein, R. (2003). Structure and autoregulation of a full-length yeast homologue of Sir2. Nat. Struct. Biol., 10, 864–871. Accession Numbers Coordinates of the Hst2/NAD⫹ and Hst2/NAD⫹/histone H4 complexes have been deposited to the Rutgers Collaborative Structural Bioinfomatics database under accession numbers 1Q17 and 1Q1A, respectively.