View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Structure, Vol. 9, 1127–1133, December, 2001, 2001 Elsevier Science Ltd. All rights reserved. PII S0969-2126(01)00690-6 Structure of Histone Deacetylases: Review Insights into Substrate Recognition and Catalysis

Ronen Marmorstein sequence and functional properties of the HDAC super- The Wistar Institute and the Department of Chemistry family of and discuss in more detail structural University of Pennsylvania information that has been recently derived for them. I Philadelphia, Pennsylvania 19104 will also discuss how a correlation of the structural and functional studies leads to a framework for understand- ing the substrate specific binding and catalysis by these Summary enzymes. Those who are interested in a more detailed discussion of HDAC function are referred to several ex- Histone deacetylases catalyze the removal of the ace- cellent and recent reviews [9, 11, 12, 15]. tyl moiety from acetyl-lysine within histones to pro- mote repression and silencing. These enzymes The Superfamily fall into distinct families based on primary sequence To date, there have been over 45 HDACs identified within homology and functional properties in vivo. Recent yeast, Drosophila, maize, chicken, mouse, and human structural studies of histone deacetylases and their (Table 1), and there have also been homologous proteins homologs from bacteria have provided important in- identified in bacteria. Together, the eukaryotic proteins sights into the mode of substrate recognition and ca- fall into at least three distinct classes based on se- talysis by these enzymes. quence homology. Members of the class I subgroup have a high degree of to yeast Rpd3 (yRpd3) and are referred to as Rpd3-like. In addition to Introduction yRpd3, members of this family include hHDAC1-3 and It has been nearly 40 years since Allfey and coworkers yHos3, with yHos3 being the most divergent (Table 1). have proposed that the status of the histone The sequence homology extends over about 300 resi- proteins that make up chromatin is correlated with the dues with particularly striking homology within an inter- transcriptional competence of a given gene, whereby residue stretch (Figure 1). For example, within 70ف nal containing hypoacetylated and hyperacetylated the 300 residue region of conservation, yRPD3 and histones are transcriptionally repressed and activated, hHDAC1 have 80% similarity and 66% identity, and respectively [1]. Despite, the time that has elapsed since within the more homologous 70 residue stretch, these the initial studies of Allfrey, the proteins that modulate protein have 99% similarity and 91% identity. Many histone acetylation have only recently been identified, members of class I HDACs have been shown to be and structural information has only become available sensitive to inhibition by a family of small molecule com- over the last 2 years. A major breakthrough in under- pounds that have homology to trichostatin (TSA) such standing the mechanism of histone acetylation came as suberoylanilide hydroxamic acid (SAHA) and trapoxin with the cloning of a histone acetyltransferase (HAT) (TPX) (reviewed in [16]). Moreover, members of the class from Tetrahymena [2] that had strong sequence homol- I HDACs have been shown to be tightly associated with ogy to the Gcn5 transcriptional mediator from yeast several other protein subunits, such as Sin3 and N-CoR, [3, 4]. Subsequently, several other previously identified to mediate histone deacetylation and transcriptional transcriptional coactivators, such as CBP/p300 [5, 6], corepression in vivo (reviewed in [17, 18]). TAFII250 [7], and SRC-1 [8], were shown to have HAT The founding member of the class II HDACs is yeast activity, reinforcing the connections between gene acti- Hda1 (yHda1) [19, 20]; this family is referred to as Hda1- vation and histone acetylation. like and includes members such as hHDAC4-8 and Almost in parallel to the discovery of HAT enzymes yHOS1-2 (Table 1). Although members of this family of came the identification of several histone deacetylase HDACs share sequence homology within a roughly 320 (HDAC) enzymes [9] and their correlation with transcrip- residue enzymatic domain characteristic of class I tional repression and gene silencing. First, in 1996, HDACs (Figure 1), members of the class II HDACs are Schreiber and coworkers reported on the identification characterized by additional structural and functional of a mammalian histone deacetylase related features (reviewed in [15]). The class II proteins have to the yeast transcriptional regulator, Rpd3, that was more extensive N-terminal sequences than the class I isolated by virtue of its affinity to the known histone proteins. Like the class I HDACs, members of the class deacetylase inhibitor, trapoxin [10]. Shortly thereafter, II subgroup function in the context of other protein sub- other HDACs were identified, including several that had units in vivo and are sensitive to inhibition by TSA and poor sequence homology to Rpd3 (reviewed in [11, 12]). its homologs. However, the class II proteins appear to Interest in HDACs was further heightened by their corre- interact with a different set of proteins in vivo than the lation with several cancers, including promyelocytic leu- class I proteins. In addition, it appears that the class II kemia [13, 14]. These findings underscored the key gene proteins are actively maintained in the cytoplasm and regulatory role played by HDACs and how their misregu- are imported to the nucleus when required [21–25], sug- lation can lead to disease. gesting that their regulation may be more complex than In this report, I will highlight what is known about the the predominantly nuclear class I proteins. Human HDAC6 is an unusual member of the class II proteins Correspondence: [email protected] since it has two catalytic deacetylase domains [26]. Structure 1128

Table 1. Classification of Histone Deacetylases Class Members Properties I. yRdp3-like y, ma RPD3; y, h, m, d, c, ch HDAC1 -TSA-sensitive y, h, m, d, c, ch HDAC2; -Tightly associated with regulatory proteins y, h, m, d, c, ch HDAC3; yHOS3 -N-terminal deacetylase domain

II. yHda1-like yHDA1; hHDAC4; hHDAC5; -TSA-sensitive hHDAC6 h, ch HDAC7; -Tightly associated with regulatory proteins h, ch HDAC8; mHDA1; -C-terminal deacetylase domain m, d HDA2; yHOS1; yHOS2 -Shuttles between nucleus and cytoplasm

III. ySir2-like Ia - ySIR2, hSIRT1, yHST1 -NAD-dependant Ib - yHST2, hSIRT2, hSIRT3 -Different members are nuclear or cytoplasmic Ic - yHST3, yHST4 II - hSIRT4 III - hSIRT5 IV - hSIRT6, hSIRT7

Abbreviations: y ϭ yeast, h ϭ human, m ϭ mouse, d ϭ drosophila,cϭ C. elegans,chϭ chicken, ma ϭ maize. Only Sir2-like homologs from yeast (S. cerevisiae) and humans are shown here. A more complete list of homologs can be found in Frye, 2000 (Biochem. Biophys. Res. Commun., 273, 793–798).

Yeast Sir2 (ySir2) is the founding member of the class quence homology to ySir2, suggesting that histones may III HDACs, and this family of proteins shows significant not be the only targets of the eukaryotic Sir2 proteins. sequence and functional divergence from class I and II Sequence homology among the Sir2 proteins is re- subgroups (Table 1 and Figure 2). The Sir2 family has stricted to a roughly 270 residue domain that has been further been phylogenetically classified into four distinct shown to be sufficient for catalytic activity (Figure 2). subclasses (I-IV) [27]. A functional distinction of the Sir2- Within this 270 residue region, there are four patches like proteins is that their deacetylase activity is NAD of particularly high sequence conservation that are dis- dependent [28–30]. The Sir2-like proteins are extensive tributed throughout the domain. Interestingly, the intact and broadly conserved from yeast to man [27]. Within S. bacterial homologs are not much larger than the 270 cerevisiae and human there are five and seven members, residue enzymatic domain, while the eukaryotic homo- respectively (Table 1). Surprisingly, bacteria (that are logs usually contain more extended N- and C- terminal free of histones) also contain proteins that have se- sequences, suggesting an additional level of regulation

Figure 1. Sequence Alignment of the Class I (yRpd3-like) and Class II (yHda1-like) Histone Deacetylases Sequences are aligned against the histone deacetylase-like protein from hyperthermophi- lic bacterium Aquifex aeolicus (aHDLP). Typical class I members, human HDAC1 (hHDAC1) and yeast Rpd3 (yRpd3), are compared with the most distantly related class I member, yeast Hos3 (yHos3), and a typical class II member, human HDAC6 (hHDAC6). Identical residues in at least three out of the five pro- teins are indicated with black shading, and conserved residues are indicated with gray shading. The secondary structural elements of aHDLP are identified above the aligned sequences where ␣ helices are represented by rectangles and ␤ strands are represented by bold arrows. Asterisks above the aHDLP sequence identifies residues that contact the trichostatin (TSA) inhibitor in the aHDLP/TSA complex, and gray bars below the sequence alignment represents loop regions of aHDLP that line the pronounced protein pocket and that mediate contacts to the TSA inhibitor. Review 1129

Figure 2. Sequence Alignment of the Class III (ySir2-like) Histone Deacetylases Sequences are aligned against human SIRT2 and a homolog of the Class III-like histone deacetylases from the bacterium Archaeo- globus fulgidus, Af1. At least one member from each of the phylogenetically defined subclasses of the ySir2 family (I–IV) is aligned. The secondary structural elements for SIRT2 and Af1 are compared above the sequence alignment. Asterisks above the sequence alignment identify residues that contact the NAD ligand in the Af1/NAD complex. Bar des- ignations below the sequence alignment indi- cate regions of Af1 and SIRT2 that comprise part of the large domain (gray), small domain (hatched), and extended connection loop segments (open).

of these proteins in . Among eukaryotes, dif- and C-terminal ends of the ␤ strands. The longest loop ferent members of the family also have different localiza- segments are at the C-terminal end of the ␤ strands tion. For example, within yeast, Sir2 and Hst1 are pre- and, together with many of the helices that don’t flank dominantly nuclear, while Hst2 (and its human homolog the ␤ sheet, form a pronounced pocket along one face SIRT2) is cytoplasmic [31]. of the structure (Figure 3b). The TSA and SAHA HDAC To date, there has been only one structure of a bona inhibitors occupy this pocket in the HDLP/inhibitor com- fide histone deacetylase enzyme reported: the class III plexes, with the aromatic end of the inhibitors pointing human HDAC, SIRT2 [32]. However, the structure of away from the protein and the extended carbon chain the Sir2 homolog from the bacterium Archaeoglobus of the inhibitor extending deep into an internal protein fulgidus, Af1, has been determined in complex with NAD cavity within the pocket region. A bound Zn2ϩ ion is also [33]. The structure of a protein from the hyperthermophi- located near the end of the inhibitor that is located deep lic bacterium Aquifex aeolicus with sequence homology within the internal cavity of the protein. All of the inhibitor to the class I and II HDACs (termed HDLP) has also and Zn2ϩ ion interactions are mediated by the loop re- been determined alone and in complex with the TSA gions that line the pocket and cavity region of HDLP. and SAHA inhibitors [34]. Together, these structures provide a framework for understanding the structural Functional Implications for the yRpd3- and mechanistic basis for HDAC activity. and yHda1-like HDACs A mapping of residues that are highly conserved across Overall Structure of HDLP the class I and II HDACs onto the HDLP structure shows Over the 320 residue HDAC homology region of the class that the highest degree of sequence conservation maps I and II HDACs, HDLP shares about 35% and 27% se- to the pronounced protein pocket that is surrounded by quence identity with human HDAC1 and HDAC6, respec- the large surface loops at one end of the protein (Figure tively. The HDLP structure belongs to the open ␣/␤ class 3b). Consistent with the importance of this site for ace- of folds and shows a compact architecture containing tyl-lysine binding is the observation that the TSA and an 8 stranded parallel ␤ sheet at the center of the struc- SAHA HDAC inhibitors bind within this region. A detailed ture with 13 helices surrounding the ␤ sheet (Figure 3a). view of this protein region reveals that its chemical prop- Four helices pack against each side of the ␤ sheet, erties are ideally suited for acetyl-lysine binding (Figure characteristic of the open ␣/␤ class of folds, and the 3c). Specifically, the bottom of the cavity contains four remaining helices are largely clustered toward the histidine residues that are strictly conserved within the C-terminal end of the ␤ strands. A series of loops that class I and II families (residues 131, 132, 170, and 171). connect the secondary structural elements are largely These histidine residues may play a role in binding the localized to opposite sides of the molecule, at the N- negatively charged tip of acetyl lysine and/or (as de- Structure 1130

Figure 3. Structure of the aHDLP/TSA Complex (a) Schematic structure of the aHDLP/TSA complex. Helices are colored in blue, ␤ strands are colored in green, and loops are colored in aqua. Loop segments that contact TSA are colored in brown; the TSA inhibitor is colored in red, and the bound Zn2ϩ ion is colored in purple. (b) Surface representation of aHDLP with bound TSA (red). Highly conserved regions within the class I and II family of HDACs are highlighted in yellow. (c) The putative acetyl-lysine binding and catalytic site is highlighted. The bound TSA molecule has been removed from this image for clarity. Basic residues that are highly or strictly conserved within the class I/II HDACs and that are proposed to play a direct role in catalysis are identified in red; hydrophobic residues that are proposed to play a role in contacting the aliphatic region of the acetyl-lysine residue are identified in pink. scribed below) a role in catalysis. In addition, there are structure contains a bound Zn2ϩ ion instead of Mn2ϩ, a number of highly conserved aspartic acid residues (166, the geometry is remarkably similar to a mono-Mn2ϩ argi- 168 and 173) and a strictly conserved cysteine residue nase. In HDLP, the Zn2ϩ ion is bound to two aspartic (142) that also may play a role in catalysis. In the HDLP/ acids, one histidine and a water molecule. In addition, inhibitor structures, several of the histidine and aspartic a Zn2ϩ ion hydrogen bond with the TSA inhibitor appears acid residues contact either the bound Zn2ϩ ion or the to mimic the interaction with the third aspartic acid resi- inhibitor. Further out from the cavity, there are also some due in . It is possible that the bound water mole- highly conserved hydrophobic residues (Pro22, Phe141, cule is activated by a combination of contributions from Phe198, Leu265, and Tyr297) that may also facilitate the activated Zn2ϩ ion (like in arginase) and the His-Asp association with the aliphatic arm of acetyl-lysine. charge relay pair (like in serine proteases). In the HDLP/ Although the detailed enzymatic mechanism must TSA structure Tyr297 and His132 also hydrogen bond await further analysis, there are some tantalizing correla- to the carbonyl and amine nitrogen, respectively, of TSA. tions with other better characterized enzymes. Specifi- The high-degree sequence conservation of these resi- cally, near the bottom of the pronounced protein cavity dues within class I and II HDACs also argues for their of HDLP, the histidine and aspartic acid residues form role in catalysis, possibly in a later step. The role played two hydrogen bonded Asp-His pairs (Asp166-His131 by the residues mentioned above in catalysis is also and Asp173-His132) that are characteristic of the charge supported by various mutagenesis studies [32, 37, 38]. relay systems found at the active sites of serine prote- ases that are used in these enzymes to polarize and Overall Structure of the ySir2-like HDACs increase the basicity of the N⑀ histidine nitrogens [35]. The core catalytic region of SIRT2 forms an elongated In the serine proteases, the polarized histidine residues structure containing two globular domains that are con- serve to increase the nucleophilicity of a bound water nected by four extended strands which traverse be- molecule for participation in catalysis. Interestingly, in tween the two domains four times [32] (Figure 4a). The the HDLP structure, His131 is bound to a water molecule large domain of SIRT2 contains a central six-stranded which is also bound to the Zn2ϩ ion. It is possible that parallel ␤ sheet with three helices flanking each face of the water molecule may mediate nucleophilic attack of the sheet, forming a Rossmann fold that is characteristic the carbonyl carbon of the acetyl-lysine substrate. of many NAD/NADP binding enzymes [39]. The small The chemical environment proximal to the Zn2ϩ ion domain of SIRT2 contains four helices and a two- also shows similarities with metalloproteinases. The stranded antiparallel ␤ sheet. Within the small domain, most striking homology appears to be with the diman- a structural Zn2ϩ ion is liganded by four cysteine residues ganese metalloenzyme arginase [36]. In arginase, each that are highly conserved in the Sir2 family. The overall Mn2ϩ ion is hydrogen bonded to one histidine and three fold around the Zn2ϩ ion has a topology that is similar aspartic acid ligands, and a hydroxide ion bridges the to a RING finger [40]. The large and small domains do Mn2ϩ ions and one of the Mn2ϩ-bound aspartic acid not make direct interactions, but have fixed orientation ligands. This arrangement appears to polarize the hy- to each other by virtue of interactions that are mediated droxide ion for nucleophilic attack of the quanidinium by the extended connection loops. The region between carbon of the arginine substrate. Interestingly, arginase the small and large domains forms a pronounced groove reconstituted with only one Mn2ϩ ion still contains signifi- that runs roughly perpendicular to the long axis of the cant, albeit less, catalytic activity. Although the HDLP molecule. Review 1131

Figure 4. Structure of the ySir2-like Histone Deacetylases (a) Schematic structure of SIRT2. The small domain is colored in green, the large domain is colored in blue, the extended connecting loop- rich region is colored in orange, and the bound Zn2ϩ ion is colored in purple. (b) Structure of the Af1/NAD complex using the same color-coding as in (a). The bound NAD molecule in shown in red. (c) Surface representation of Af1 with bound NAD (red). Highly conserved residues within the class III (ySir2-like) family of HDACs are highlighted in yellow. The protein sites proposed to mediate acetyl-lysine and NAD hydrolysis are labeled B and C, respectively. (d) The putative catalytic sites for acetyl-lysine (B site) and NAD hydrolysis (C site) within Af1 are highlighted. The bound NAD molecule is identified in green. Basic residues that are highly or strictly conserved within the ySir2-like HDACs and that are proposed to play a direct role in catalysis are identified in red, and hydrophobic residues that are proposed to play a role in contacting the aliphatic region of the acetyl- lysine residue are identified in pink.

The structure of the Afl Sir2 homolog bound to NAD strictly conserved among the Sir2 family of HDACs are shows the same overall shape and domain architecture mutationally sensitive for NAD-nicotinamide exchange as the catalytic core of SIRT2 (Figure 4b), but the overall and deacetylase activity [33]. Biochemical studies with alignment is surprisingly poor, despite the 26% se- Sir2 family members have revealed that hydrolysis of quence identity between the two proteins within the NAD forms nicotinamide and ADP-ribose and that this HDAC domain. The corresponding large domains super- hydrolysis is tightly coupled to the deacetylation of ace- impose considerably better than the small domains, with tyl-lysine [42, 43]. Moreover, it has been suggested that a root-mean-square deviation of 1.6 A˚ for 86 of the 145 this coupled reaction generates a novel product, 1-O- residues of the large domain. The small domain of Af1 acetyl-ADP ribose, in which the acetyl group from ace- contains an antiparallel three-stranded ␤ sheet and a tyl-lysine is transferred to the nicotinamide position of helix-turn-helix module. The structural zinc module NAD. The structure of the Af1/NAD complex suggests within the small domain of Af1 forms a zinc ribbon struc- that acetyl-lysine and NAD hydrolysis occur at two dis- ture initially characterized in TfIIB [41]. The NAD mole- tinct locations within the region of high conservation, cule in Af1 is bound along a face of the large domain and these positions have been referred to as sites B and partly within the groove formed at the interface and C, respectively [33]. between the small and large domains and the connect- Site B is located on the extreme edge of the conserved ing loops. The NAD molecule adopts an extended con- surface closest to the nicotinamide side of the NAD formation that is commonly found in other Rossman fold/ molecule and contains an accessible and deep channel NAD complexes [39]; however, the NAD is bound in an that has been proposed to be a for the inverted orientation relative to most other NAD-linked de- acetyl-lysine substrate. This site has a chemical environ- hydrogenases. The nicotinamide group is not visible in ment that is consistent with a putative role in acetyl- the electron density map and is tentatively modeled lysine binding (Figure 4d). Specifically, at the bottom of based on the position of the attached ribose ring. this channel is a histidine residue that is strictly con- served within the Sir2 family (His116 in Af1 and His187 in Functional Implications for ySir2-like HDACs SIRT2) that is likely to play an important role in catalysis. A mapping of residues that are highly conserved across Correlating with the importance of this histidine residue the Sir2 family of HDACs onto the Af1/NAD structure is its high sensitivity to mutation. In addition, the region shows that the highest degree of sequence conservation proximal to the histidine residue and toward the outside extends across the extended loop regions connecting of the channel contains conserved hydrophobic resi- the small and large subdomains (Figure 4c). In addition, dues that may serve to complement the aliphatic chain NAD is bound near the center of this region of conserva- of acetyl-lysine. Among these hydrophobic residues, tion. Together, these observations implicate the con- Leu163 and Pro192 (Af1 numbering) are strictly con- served and extended loop regions connecting the small served; leucines 115, 168, and 196 (Af1 numbering) are and large domains of Af1 as the site of acetyl-lysine also highly conserved. Interestingly, this “hydrophobic substrate binding and catalysis. This hypothesis is sup- border” is considerably wider in SIRT2 than it is in Af1, ported by mutational studies across the conserved sur- and it may be that the width of this region is modulated face which reveal that several charged residues that are by NAD binding. This may indeed be a mechanism to Structure 1132

assure that acetyl-lysine does not bind in the absence small domain is important for deacetylase activity [44], of NAD. and it is tempting to propose that this region may be Site C also contains a pronounced protein channel particularly important for specificity for the acetyl-lysine and is located toward the center of the patch of Sir2 containing substrate. The sequence and structural di- sequence conservation and is inaccessible due to steric vergence within the small domain further suggests that hindrance of the bound NAD molecule. Because of the the small domain may play a modulatory role for sub- inaccessibility of this site and its relative proximity to strate specificity within the Sir2 family. The absence of the nicotinamide group, it has been proposed to be a corresponding small domain in the class I/II HDACs the nicotinamide cleavage site [33]. This proposal is suggests that other proteins may functionally substitute consistent with the chemical environment of this site. for this domain in vivo. This is consistent with the obser- Specifically, at the base of this channel are several po- vation that class I/II HDACs associate with other proteins tential catalytic residues that are strictly conserved in vivo for catalytic activity. Although the detailed cata- within the Sir2 family (Ser24, His80, Asn99, and Asp101 lytic mechanism of any HDAC is still unknown, it is al- in Af1). There is also an ordered water molecule that is ready clear that the mechanisms of the class I/II and hydrogen bonded to each of these residues. In addition, Sir2 families diverge significantly. Although both types of His104, which is either a histidine of glutamate in other deacetylases require a for activity, deacetylase Sir2 homologs, is also located in this channel. The equiv- activity by the Sir2 family of HDACs is NAD dependent, alent residues to Ser24 and Asn99 of Af1 in ySir2 and whereas the class I/II family of proteins appears to de- yHst2 have also been shown to be sensitive to mutation pend on the presence of a catalytic Zn2ϩ ion. [30, 33], further supporting the importance of the C site for NAD catalysis. Intriguingly, the presence of a serine, histidine, and aspartic acid in the C site is reminiscent Remaining Questions and Perspectives of the serine proteases; however, the geometrical con- Taken together, the structure of the histone deacetylase figuration of these residues in the Af1/NAD structure is SIRT2 and the bacterial histone deacetylase homologs different. Paradoxically, the 1 position of the ribose ring Af1 and HDLP have provided a structural framework for that is attached to the nicotinamide group that is pro- understanding the mechanism of acetyl-lysine binding posed to undergo cleavage is located too far from these and catalysis by HDACs. We are now poised to address putative residues to allow for catalysis. This important mechanistic issues. Paramount among these suggests that NAD cleavage would require structural issues are the mechanistic details underlying substrate rearrangement of either the protein, the NAD molecule, binding, catalysis, and target specificity. As in the case or both. Such repositioning may also be facilitated by of histone acetyltransferases, it may be that some acetyl-lysine binding, which would serve to couple ace- HDACs deacetylate nonhistone factors or tyl-lysine and NAD hydrolysis. even nontranscription factor proteins. Indeed, the bac- terial HDAC homolog must deacetylate nonhistone pro- teins in vivo since bacteria do not contain histones. Comparison of HDAC Structures Another important question is why many HDAC proteins Based on their similar enzymatic activities, one would require modulatory subunits for in vivo activity? What expect the proteins within the HDAC superfamily to is the role of these subunits and, in particular, do they share some structural features, despite the sequence have a role in acetyl-lysine substrate specificity or pro- divergence between the class I/II and Sir2 members. A moter selection specificity? Addressing these and re- comparison of the available structural information re- lated questions will require significantly more biochemi- veals that this is indeed the case. The large domain of cal and structural analysis. From a structural standpoint Af1/SIRT2 has structural similarity to HDLP in that each we will require complexes with substrate and reaction domain has a central parallel ␤ sheet network that is intermediates. Detailed biochemical and enzymology flanked on opposite sides by helix-rich segments, al- studies addressing catalytic mechanism and substrate though in HDLP both the ␤ sheet and helical regions preference will also be required. Furthermore, to get at are somewhat more extensive. Moreover, sequence the role of HDACs in repression, it will be necessary to conservation among the respective subfamilies sug- characterize both biochemically and structurally physio- gests that in both cases, substrate binding and catalysis logically relevant HDAC complexes. The involvement of is mediated by a common face of this domain, proximal HDACs in human diseases such as leukemias increase to the C-terminal tip of the ␤ sheet, which in both cases the motivation to get at the mechanism of action of is rich in loop segments that appear to be critical for these fascinating enzymes that regulate transcription function. In addition, analysis of the loop segments re- and possibly also other biological processes. veals that, in each case, they form protein cavities that are implicated for acetyl-lysine binding by virtue of their References suitable chemical environment and mutational sensitivity. Despite these notable similarities, there are striking 1. Allfrey, V.G., Faulkner, R., and Mirsky, A.E. (1964). Acetylation structural differences between the class I/II and Sir2 and methylation of histones and their possible role in the regula- HDACs. Most notable is the small globular domain of tion of RNA synthesis. Proc. Natl. Acad. Sci. USA 51, 786–794. the Sir2 family, which is absent in the class I/II HDACs. 2. Brownell, J.E., et al., and Allis, C.D. (1996). Tetrahymena histone acetyltransferase A: A homolog of yeast Gcn5p linking histone In Af1/SIRT2, this region is positioned along one side acetylation to gene activation. Cell 84, 843–851. of the cleft harboring the region of conservation within 3. Georgakopoulos, T., and Thireos, G. (1992). Two distinct yeast the Sir2 family. Mutagenesis studies suggest that the transcriptional activators require the function of the GCN5 pro- Review 1133

tein to promote normal levels of transcription. EMBO J. 11, 28. Landry, J., et al., and Sternglanz, R. (2000). The silencing protein 4145–4152. SIR2 and its homologs are NAD-dependent protein deacety- 4. Berger, S.L., et al., and Guarente, L. (1992). Genetic isolation lases. Proc. Natl. Acad. Sci. USA 97, 5807–5811. of ADA2: a potential transcriptional adaptor required for function 29. Frye, R.A. (1999). Characterization of five human cDNAs with of certain acidic activation domains. Cell 70, 251–265. homology to the yeast SIR2 gene: Sir2-like proteins () 5. Shikama, N., Lyon, J., and LaThangue, N.B. (1997). The p300/ metabolize NAD and may have protein ADP-ribosyltransferase CBP family: Integrating signals with transcription factors and activity. Biochem. Biophys. Res. Commun. 260, 273–279. chromatin. Trends Cell Biol. 7, 230–236. 30. Imai, S., Armstrong, C.M., Kaeberlein, M., and Guarente, L. 6. Bannister, A.J., and Kouzarides, T. (1996). The CBP co-activator (2000). Transcriptional silencing and longevity protein Sir2 is an is a histone acetyltransferase. Nature 384, 641–643. NAD-dependent histone deacetylase. Nature 403, 795–800. 7. Mizzen, C.A., et al., and Allis, C.D. (1996). The TAF(II)250 subunit 31. Perrod, S., et al., and Gasser, S.M. (2001). A cytosolic NAD- of TFIID has histone acetyltransferase activity. Cell 87, 1261– dependent deacetylase, Hst2p, can modulate nucleolar and 1270. telomeric silencing in yeast. EMBO J. 20, 197–209. 8. Spencer, T.E., et al., and Omalley, B.W. (1997). Steroid receptor 32. Finnin, M.S., Donigian, J.R., and Pavletich, N.P. (2001). Structure -1 is a histone acetyltransferase. Nature 389, of the histone deacetylase SIRT2. Nat. Struct. Biol. 8, 621–625. 194–198. 33. Min, J., Landry, J., Sternglanz, R., and Xu, R.M. (2001). Crystal 9. Cress, W.D., and Seto, E. (2000). Histone deacetylases, tran- structure of a SIR2 homolog-NAD complex. Cell 105, 269–279. scriptional control, and cancer. J. Cell. Physiol. 184, 1–16. 34. Finnin, M.S., et al., and Pavletich, N.P. (1999). Structures of a 10. Taunton, J., Hassig, C.A., and Schreiber, S.L. (1996). A mamma- histone deacetylase homologue bound to the TSA and SAHA lian histone deacetylase related to the yeast transcription regu- inhibitors. Nature 401, 188–193. lator Rpd3p. Science 272, 408–411. 35. Fersht, A.R., and Sperling, J. (1973). The charge relay system in 11. Gray, S.G., and Ekstrom, T.J. (2001). The human histone deace- chymotrypsin and chymotrypsinogen. J. Mol. Biol. 74, 137–149. tylase family. Exp. Cell Res. 262, 75–83. 36. Kanyo, Z.F., Scolnick, L.R., Ash, D.E., and Christianson, D.W. 12. Graessle, S., Loidl, P., and Brosch, G. (2001). Histone acetyla- (1996). Structure of a unique binuclear manganese cluster in tion: plants and fungi as model systems for the investigation of arginase. Nature 383, 554–557. histone deacetylases. Cell. Mol. Life Sci. 58, 704–720. 37. Hassig, C.A., et al., and Schreiber, S.L. (1998). A role for histone 13. Grignani, F., et al., and Pelicci, P.G. (1998). Fusion proteins of deacetylase activity in HDAC1-mediated transcriptional repres- the retinoic acid receptor-alpha recruit histone deacetylase in sion. Proc. Natl. Acad. Sci. USA 95, 3519–3524. promyelocytic leukaemia. Nature 391, 815–818. 38. Kadosh, D., and Struhl, K. (1998). Histone deacetylase activity 14. Lin, R.J., Nagy, L., Inoue, S., Shao, W.L., Miller, W.H., and Evans, of Rpd3 is important for transcriptional repression in vivo. Genes R.M. (1998). Role of the histone deacetylase complex in acute Dev. 12, 797–805. promyelocytic leukaemia. Nature 391, 811–814. 39. Bellamacina, C.R. (1996). The nicotinamide dinucleotide binding 15. Wade, P.A. (2001). Transcriptional control at regulatory check- motif: a comparison of nucleotide binding proteins. FASEB J. points by histone deacetylases: molecular connections be- 10, 1257–1269. tween cancer and chromatin. Hum. Mol. Genet. 10, 693–698. 40. Borden, K.L. (2000). RING domains: master builders of molecu- 16. Weidle, U.H., and Grossmann, A. (2000). Inhibition of histone lar scaffolds? J. Mol. Biol. 295, 1103–1112. deacetylases: a new strategy to target epigenetic modifications 41. Zhu, W., Zeng, Q., Colangelo, C.M., Lewis, M., Summers, M.F., for anticancer treatment. Anticancer Res. 20, 1471–1485. and Scott, R.A. (1996). The N-terminal domain of TFIIB from 17. Ng, H.H., and Bird, A. (2000). Histone deacetylases: silencers Pyrococcus furiosus forms a zinc ribbon. Nat. Struct. Biol. 3, for hire. Trends Biochem. Sci. 25, 121–126. 122–124. 18. Knoepfler, P.S., and Eisenman, R.N. (1999). Sin meets NuRD 42. Tanner, K.G., Landry, J., Sternglanz, R., and Denu, J.M. (2000). and other tails of repression. Cell 99, 447–450. Silent information regulator 2 family of NAD- dependent histone/ 19. Carmen, A.A., Rundlett, S.E., and Grunstein, M. (1996). HDA1 protein deacetylases generates a unique product, 1-O-acetyl- and HDA3 are components of a yeast histone deacetylase (HDA) ADP-ribose. Proc. Natl. Acad. Sci. USA 97, 14178–14182. complex. J. Biol. Chem. 271, 15837–15844. 43. Tanny, J.C., and Moazed, D. (2001). Coupling of histone deace- 20. Rundlett, S.E., Carmen, A.A., Kobayashi, R., Bavykin, S., Turner, tylation to NAD breakdown by the yeast silencing protein Sir2: B.M., and Grunstein, M. (1996). HDA1 and RPD3 are members Evidence for acetyl transfer from substrate to an NAD break- of distinct yeast histone deacetylase complexes that regulate down product. Proc. Natl. Acad. Sci. USA 98, 415–420. silencing and transcription. Proc. Natl. Acad. Sci. USA 93, 44. Sherman, J.M., Stone, E.M., Freeman-Cook, L.L., Brachmann, 14503–14508. C.B., Boeke, J.D., and Pillus, L. (1999). The conserved core of 21. Grozinger, C.M., and Schreiber, S.L. (2000). Regulation of his- a human SIR2 homologue functions in yeast silencing. Mol. Biol. tone deacetylase 4 and 5 and transcriptional activity by 14-3- Cell 10, 3045–3059. 3-dependent cellular localization. Proc. Natl. Acad. Sci. USA 97, 7835–7840. 22. Zhou, X., Richon, V.M., Wang, A.H., Yang, X.J., Rifkind, R.A., and Marks, P.A. (2000). Histone deacetylase 4 associates with extracellular signal-regulated kinases 1 and 2, and its cellular localization is regulated by oncogenic Ras. Proc. Natl. Acad. Sci. USA 97, 14329–14333. 23. McKinsey, T.A., Zhang, C.L., Lu, J., and Olson, E.N. (2000). Signal-dependent nuclear export of a histone deacetylase regu- lates muscle differentiation. Nature 408, 106–111. 24. Verdel, A., et al., and Khochbin, S. (2000). Active maintenance of mHDA2/mHDAC6 histone-deacetylase in the cytoplasm. Curr. Biol. 10, 747–749. 25. Wang, A.H., et al., and Yang, X.J. (1999). HDAC4, a human histone deacetylase related to yeast HDA1, is a transcriptional . Mol. Cell. Biol. 19, 7816–7827. 26. Grozinger, C.M., Hassig, C.A., and Schreiber, S.L. (1999). Three proteins define a class of human histone deacetylases related to yeast Hda1p. Proc. Natl. Acad. Sci. USA 96, 4868–4873. 27. Frye, R.A. (2000). Phylogenetic classification of prokaryotic and eukaryotic Sir2-like proteins. Biochem. Biophys. Res. Commun. 273, 793–798.