HDA1 and HDA3 Are Components of a Yeast Histone Deacetylase (HDA) Complex*

Histone acetylation is maintained through the action of histone acetyltransferases and deacetylases and has been correlated with increased gene activity. To inves- tigate the functional role of these enzymes in the regulation of transcription, we have purified from Saccharo- myces cerevisiae two histone deacetylase activities, HDA and HDB, with molecular masses of (cid:59) 350 and 600 kDa, respectively. In vitro , the HDA activity deacetylates all four core histones, has a preference for histone H3, and is strongly inhibited by trichostatin A (a specific inhib- itor of histone deacetylases). HDB is considerably less sensitive to trichostatin A. We report the extensive pu- rification of the HDA activity and the identification of peptides (p75, p73, p72, and p71) whose presence corre- lates with deacetylase activity on native polyacrylamide gels. An antibody to p75 immunoprecipitates peptides with molecular masses similar to those in the 350-kDa complex. Additionally, antibodies to p75 and p71 specif- ically precipitate histone deacetylase activity and co-immunoprecipitate each other. Gene disruptions of p75 ( HDA1 ) or p71 ( HDA3 ) cause the loss of the 350-kDa (but not the 600-kDa) activity from our chromatography pro- files. These data argue strongly that HDA1 and HDA3 are subunits of the HDA complex, which is structurally distinct from the second, HDB complex. Nucleosomes The dynamic

Nucleosomes can have both positive and negative effects on transcription (1)(2)(3)(4)(5)(6). The dynamic effects of nucleosomes on transcription may be due, in part, to the acetylation of lysine residues present in the hydrophilic amino-terminal portions (tails) of the core histones (7). This acetylated state is maintained through the competing activities of histone acetyltransferases and deacetylases. Histone acetyltransferases transfer the acetate from acetyl-CoA to neutralize the positively charged ⑀-amino group of an unacetylated lysine residue. In contrast, deacetylation is achieved through the hydrolysis of the acetyl moiety, restoring a positive charge. Numerous studies have demonstrated the colocalization of nucleosomes containing hyperacetylated histones with active or potentially active genomic domains. Conversely, histone hypoacetylation has been correlated with heterochromatic regions that are predominantly silent (7)(8)(9)(10).
Despite these correlations, it is still not known whether the acetylated state of histones has a causal effect on histone deposition or gene activity. To understand the physiological effects of histone acetylation and deacetylation, it will be necessary to identify the protein subunits of the appropriate enzymes and to obtain loss-of-function mutations in their genes, whose effects on gene activity may then be monitored in vivo. Partial purification and characterization of histone acetyltransferases have led to the conclusion that there are several different enzymes within the cell with different substrate specificities (18 -20). Recently, a cytoplasmic yeast histone acetyltransferase subunit with specificity for H4 Lys-12 in vitro has been found (21). In addition, an active acetyltransferase (55 kDa) subunit from Tetrahymena macronuclei with a homologue in yeast (GCN5) has also been identified (22,23).
In contrast to histone acetyltransferases, less is known about deacetylases. Partial purification studies have detected activities in yeast (24,25) and higher eukaryotes (26 -29). Sanchez Del Pino et al. (25) reported the partial purification from yeast whole cell extracts of two forms of deacetylase that were able to deacetylate histones. Other researchers have demonstrated that histone deacetylase activity is tightly associated with the nuclear matrix (29 -31). A deacetylase activity is also associated with yeast nuclei (24). Sodium butyrate inhibits histone deacetylation in yeast spheroplasts (24) and higher eukaryotes (32). However, sodium butyrate is neither a very potent inhibitor (requiring millimolar concentrations) of the deacetylase nor specific in its actions (33,34). In contrast, trichostatin A (TSA) 1 is a very specific and potent inhibitor (demonstrating inhibition in the nanomolar range) of histone deacetylases (25,28).
In this report, we describe the identification of two distinct nuclear associated histone deacetylases: HDA, a 350-kDa activity that is highly sensitive to TSA; and HDB, a 600-kDa activity that is less sensitive. The HDB activity is likely the same as that reported previously (25). We have purified HDA extensively and have found that it copurifies chromatographically with at least four peptides (p75, p73, p72, and p71). Our chromatography, immunoprecipitation, and gene disruption data argue that we have identified at least two of these peptides (p75 coded for by HDA1 and p71 coded for by HDA3) as members of the histone deacetylase (HDA) complex and that the HDA complex is structurally distinct from the larger histone deacetylase HDB complex.

Isolation of [ 3 H]Acetyllysine-labeled HeLa Histones
6 liters of HeLa cells were grown to a density of 5 ϫ 10 5 cells/ml in Dulbecco's modified Eagle's medium (9418, Irvine Scientific, Santa Ana, CA) supplemented with 10% newborn calf serum, 1% fetal calf serum, 0.1 g/liter streptomycin, and 0.06 g/liter penicillin. The cells were centrifuged (1500 ϫ g) and resuspended in 120 ml of phosphate-buffered saline (140 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , and 1.8 mM KH 2 PO 4 ), pH 7.4, and the solution was brought to 100 g/ml cycloheximide, 10 mM sodium butyrate, and 0.2 mCi/ml [ 3 H]acetic acid and incubated for 1 h at 37°C with gentle stirring. Cells were chilled on ice, centrifuged (500 ϫ g, 5 min), washed three times in 50 ml of phosphatebuffered saline and 10 mM sodium butyrate, and then lysed in 40 ml of NIB buffer (1% Nonidet P-40 in IB buffer (10 mM Tris-HCl, pH 7.4, 2 mM MgCl 2 , 3 mM CaCl 2 , 10 mM sodium butyrate, and 1 mM phenylmethylsulfonyl fluoride)). Nuclei were collected (500 ϫ g) and washed twice in 40 ml of NIB buffer followed by one wash with NIB buffer supplemented with 100 mM NaCl. One additional wash was performed in 40 ml of 100 mM NaCl and IB buffer, and then the nuclei were high salt-extracted twice in 40 ml of 400 mM NaCl and IB buffer followed by centrifugation. The nuclear pellet was extracted twice in 10 pellet volumes of 0.2 M H 2 SO 4 for 90 min on ice and centrifuged (30,000 ϫ g, 25 min). The supernatants were pooled and dialyzed extensively at 4°C against 100 mM acetic acid. The extracted histones were lyophilized and resuspended in water to a concentration of 4 mg/ml. Typically, a specific activity of 2.5 ϫ 10 6 cpm/mg of protein was obtained.

Isolation and Analysis of Individual Histones
Isolation and analysis of individual histones were performed by reverse-phase HPLC, and quantitation of the histones was done by the method of Marvin et al. (11). Triton-acid-urea-18% polyacrylamide gels were performed according to the method of Zweidler (35). Gels were prepared for fluorography using Amersham Amplify.

Deacetylase Assay
Column fractions were assayed by incubating 20 g of [ 3 H]acetyllabeled HeLa histones and 1 ϫ HDA assay buffer (75 mM Tris-HCl, pH 7.0, 275 mM NaCl, 2.0 mM 2-mercaptoethanol, and 0.1 mM EDTA) with 10 -50 l (or as noted) of each fraction for 30 -60 min at 30°C in a total volume of 200 l. Assays were performed in duplicate or as indicated, and counts/minute released represent the average of two independent assays. The reactions were stopped, and released acetate was extracted and assayed according to the procedures of Hendzel et al. (29).

Chromatography for Extensive Purification
The dialyzed material was filtered on a 0.8-m filter, chromatographed on a DEAE-Sepharose FF column (50 mm ϫ 5.0 cm), washed with 150 mM NH 4 Cl and Io buffer (5.0 mM 2-mercaptoethanol, 0.1 mM EDTA, 10% glycerol, and 15 mM Tris-HCl, pH 7.8) until A 280 nm Ͻ 0.05, and step-eluted at 300 mM NH 4 Cl (in Io buffer). The eluted pool was dialyzed against DBII buffer (125 mM NaCl in Io buffer), chromatographed on an S-Sepharose FF column (25 mm ϫ 10 cm), washed with 125 mM NaCl (in Io buffer) until A 280 nm Ͻ 0.05, and then step-eluted at 300 mM NaCl (in Io buffer). This activity was dialyzed in DBII buffer, filtered on a 0.2-m filter, loaded onto a Mono Q HR 10/10 column, and eluted on a gradient from 230 mM NaCl to 390 mM NaCl (in Io buffer) over 130 ml (see Fig. 1A). Peak activity fractions from 290 to 315 mM NaCl from the Mono Q column were pooled and dialyzed into DBIII buffer (100 mM sodium phosphate, pH 6.8, 10% glycerol, and 5.0 mM 2-mercaptoethanol). This material was chromatographed over a Bio-Rad Macro-Prep ceramic hydroxyapatite column (type I, 20 m, 10 mm ϫ 12 cm) and eluted over a 90-ml linear gradient from 100 to 400 mM sodium phosphate, pH 6.8, 10% glycerol, and 5.0 mM 2-mercaptoethanol (see Fig. 1B). The peak of activity from 265 to 333 mM sodium phosphate was pooled and concentrated by ultrafiltration to 0.5 ml at 2°C. After centrifugation (14,000 ϫ g), the sample was eluted at 350 mM NaCl (in Io buffer) on a Superdex 200 prep-grade column (1.0 ϫ 46 cm) (see Fig. 1C). The Superdex HDA peak was dialyzed against DBII buffer and loaded and chromatographed on a Mono S HR 5/5 column over a 20-ml linear gradient from 125 to 400 mM NaCl (see Fig. 1D). All chromatography steps were carried out at 4°C on a Pharmacia FPLC apparatus. Protein was quantitated using the Pierce micro BCA protein assay reagent.
The peak enzyme fractions were dialyzed against loading buffer (25 mM Tris-HCl, pH 7.0, 100 mM NaCl, 2.0 mM EDTA, 5.0 mM 2-mercap-toethanol, and 10% glycerol) and concentrated by spin ultrafiltration, and glycerol was added to 50% (v/v). The enzyme was overlaid with running buffer, and the tube gel was electrophoresed (anode to cathode) at 2.0 mA/tube for 16 h at 4°C.
The tube gel was removed, cut into 3.3-mm sections, and homogenized. One-fourth of each homogenate was used for deacetylase assays, and the remainder was retained for SDS-PAGE analysis. Concentrated Mono S fractions (equivalent to one-tenth of each fraction) or one-fourth of the homogenate from the upper 5 mm of the tube gel was loaded on the gel. SDS-PAGE was performed on an 8% acrylamide gel (30:0.8 acrylamide/bisacrylamide) and visualized using Silver Stain Plus (Bio-Rad).

Immunoprecipitation Using an Antibody Directed against HDA Subunits
A yeast nuclear extract was prepared from 400 g of cells and then partially purified via chromatography on DEAE-Sepharose FF, Mono Q HR 10/10, and Superdex 200 columns as described above. The pool from the Superdex column was diluted from 7.5 to 15 ml in 350 mM NaCl (in Io buffer supplemented with 1% Nonidet P-40), and 0.5 ml of the extract was used for each immunoprecipitation. Polyclonal rabbit antibodies directed against HDA1 (residues 1-226)-GST (GST gene fusion in plasmid pGEX-2T (Pharmacia)) and HDA3 (residues 170 -336)-GST fusion Escherichia coli proteins were obtained by standard procedures, and the serum was purified over a G-Sepharose FF column. The antibodies were titrated at various concentrations and allowed to incubate overnight at 0°C. Approximately 4.0 mg of washed and equilibrated protein A-Sepharose CL-4B beads were added to each immunoprecipitation and allowed to incubate for 1 h with gentle rocking. The beads were then centrifuged (500 ϫ g) and washed three times in 1.0 ml of the Superdex 200 column buffer. Half of the pellet was used for enzyme assays, and the rest was retained for SDS-PAGE. Western blotting of the SDSpolyacrylamide gel was visualized by using the Amersham ECL detection system according to the manufacturer's instructions after electrotransfer to Immobilon-P (Millipore Corp., Bedford, MA).

Enzyme Properties of HDA
Histone Specificity and K m -The HDA complex can deacetylate all four core histones, but has a preference for H3 in vitro. Using purified H3 as the substrate the K m values were determined to be 2.3 and 10.5 nM for a mixture of total histones. These are extremely low values for a K m and suggest that this enzyme has an exceptionally strong substrate affinity.
pH Optimum-The pH optimum for the enzyme was ϳ7.0 (at 250 mM NaCl with 0.25 pH unit steps), with approximately one-third of the activity remaining at pH 5 or 8. Bio-Rad Rotophor analysis determined that the pI of the protein was ϳ7.1.
Ionic Strength Optimum-The optimum NaCl concentration for the enzyme was found to be ϳ275 mM (at pH 7 with 50 mM steps). The enzyme was found to be stable between 50 and 500 mM NaCl, with total activity restored upon dialysis or dilution with additional NaCl.
Metal Ions-Enzyme activity is not enhanced by any metal ion cofactors; however, many metal ions inhibit the deacetylase, including Zn 2ϩ , Fe 3ϩ , Cu 2ϩ , Co 2ϩ , and Cr 2ϩ . Zn 2ϩ was a particularly potent inhibitor, reducing the activity of the enzyme by Ͼ95% at 0.5 mM. Although not as extensively purified, the activity described earlier did not require metal ion cofactors and was similarly inhibited by Zn 2ϩ and Cu 2ϩ (24).
Sodium Butyrate/Sodium Acetate-Sodium acetate and sodium butyrate were found to be mild inhibitors of HDA. Sodium acetate completely eliminated activity at 25 mM and was found to be a superior inhibitor compared with sodium butyrate (which reduced activity by 70% at this concentration). This result suggests that acetate, which is released by the deacetylation of histones, may acts as a feedback inhibitor of the enzyme. N 8 -Acetylspermidine-No activity was detected for this substrate even upon prolonged incubations.
Polypeptides Containing Chemically Acetylated Lysine-Poly-L-lysine or poly(Lys,Ala) (1:1), when chemically acetylated, reduced deacetylase activity to Ͻ1% when present in 20-fold excess (w/w for the competitor/histones) ( Table I). The unacetylated forms of these polyamino acids were 30-fold less effective at reducing the measured activity at the same concentrations. Acetylated forms of histone H3 and H4 tails reduced measured enzyme activity to Ͻ1% when present in 20-fold excess, similar to the acetylated polyamino acids. However, we found that the unacetylated forms of these histone tail peptides had the ability to inhibit the enzyme independent of acetylation. Therefore, acetylated polypeptides are strong competitors for enzyme activity, but the enzyme may also recognize other sequences within the histone tails.
N ⑀ -Acetyllysine and N ␣ -Acetyllysine-Monoacetylated forms of lysine have no effect on enzyme activity at 2O-fold excess (w/w) (a concentration that eliminates activity using acetylated peptides) ( Table I). This suggests that the enzyme requires more than a single acetylated lysine for recognition.

Purification of Histone Deacetylases HDA and HDB-Yeast
histones are neither efficiently labeled nor easily purified in large quantities. Therefore, in vivo radiolabeled HeLa histones were used to follow the release of [ 3 H]acetate to assay the activity of the histone deacetylase enzyme, as described under "Materials and Methods." To obtain the deacetylase, highly purified yeast nuclei were prepared as described (see "Materials and Methods"), and nuclear proteins were extracted using 450 mM NaCl. The unusually high ionic strength of the low salt wash (175 mM NaCl) helped to remove many contaminating proteins that were loosely associated with the nuclei. This nuclear extract was precipitated at 50% ammonium sulfate saturation. After dialysis, the preparation was step-eluted in series from DEAE-Sepharose and S-Sepharose columns. The S-Sepharose column resolved the deacetylase activity into two forms: the 300 mM eluting HDA activity, which we have extensively purified in this paper, and the higher salt eluting HDB activity (see Fig. 2). To further fractionate the HDA activity, we employed a Mono Q HR 10/10 column (Fig. 1A). This strong anion-exchange column yielded a single sharp peak of enzyme activity between 290 and 320 mM NaCl. This activity was further fractionated on a Macro-Prep hydroxyapatite column into a single peak of activity that eluted between 265 and 333 mM sodium phosphate (Fig. 1B). The HDA activity was sizefractionated as a 350-kDa peak on a Superdex 200 column (Fig.  1C). To minimize aggregation, gel filtration chromatography was performed at 350 mM NaCl. Detergents and these higher salt conditions did not alter the molecular mass of the HDA activity, but ionic strengths above 400 mM NaCl did tend to reduce overall HDA activity even with subsequent dialysis. In contrast, the HDB activity was retained even in the 1.0 M NaCl step elution and could be directly rechromatographed, revealing a 600-kDa complex on Superdex 200 chromatography (Fig.  2). Fractions from the gel filtration column containing peak HDA activity were chromatographed using a Mono S HR 5/5 column, yielding a single broad peak of enzyme activity (Fig.  1D) over a range of 190 -290 mM NaCl. Mono S chromatography offers only a slight purification of the deacetylase as measured, but has the ability to further separate the HDA complex into a broad peak of activity whose peptides may be gel-purified (see below). In summary, we have purified a 350-kDa histone deacetylase activity to near homogeneity, which we term HDA, and have shown that it is distinct from a second, 600-kDa activity, termed HDB.
As described under "Materials and Methods," we have characterized certain other features of the HDA complex. It has an exceptionally high substrate affinity for histones (K m ϭ 2.3 nM for histone H3 and 10.5 nM for total core histones), a pH optimum of ϳ7.0, and an NaCl optimum of 275 mM and is strongly inhibited by Zn 2ϩ and by trichostatin A and less so by sodium butyrate. Additionally, the HDA enzyme does not deacetylate acetylspermidine, but is specifically inhibited by polypeptides containing chemically acetylated lysine and both acetylated and unacetylated histone H3 and H4 N termini. This latter result suggests that it may also interact with the regions around the acetylated lysine residues at the histone N termini.
The enzyme activity present in crude lysates or nuclear fractions prior to the ammonium sulfate precipitation is much lower than after the ammonium sulfate cut, suggesting the presence of an inhibitor in these preparations (Table II). Due to this inhibition, the -fold purification from the initial crude lysate cannot be accurately determined. Chromatography on both DEAE-Sepharose and S-Sepharose increases the amount of activity present overall, suggesting the removal of inhibitors. The amount of activity after the S-Sepharose column chromatography is especially enhanced when one considers that approximately one-third of the starting activity is diverted into the HDB 1.0 M NaCl salt step. We have not characterized the inhibition, but it may be caused by unlabeled substrates within the crude preparations, allosteric regulators present in the whole cell lysate, or possible enzyme trapping within nuclear structures since histone deactylases are tightly associated with the nuclear matrix (29,30). (25,28,36). We have also found that TSA causes histone hyperacetylation in yeast spheroplasts, but we have had to use concentrations (10.0 M) 5-fold higher than those used previously (Ref. 25 and data not shown). The presence of 10.0 M TSA increases the overall levels of acetate incorporated by 3-fold compared with the wild-type control. Additionally, we observed on Triton-acid-urea gels proportionally more diacetylated H4; increases in acetylated forms of H2B, H3, and H2A were also evident (data not shown).

HDA Histone Deacetylase Activity Is Strongly Inhibited by Trichostatin A-TSA is a potent and specific inhibitor of both mammalian and yeast histone deacetylase activities in vitro
TSA was also found to be a potent noncompetitive inhibitor of the HDA activity in vitro as shown by a Lineweaver-Burk plot (Fig. 3). This plot demonstrates that the K m for the enzyme is unaffected, but the rate of reaction is reduced with increasing concentrations of TSA. Replotting this data on a Dixon plot (Fig. 3, inset) determined that TSA has a K i of 9.7 nM for HDA. While our calculated value for K i is exceptionally low for an inhibitor, its value is ϳ3-fold higher than the value reported by Yoshida et al. (28) for the mammalian deacetylase. It is interesting to note that the K i for TSA is very similar to the K m for total histones that we calculated at 10.5 nM, suggesting that the HDA affinity for both TSA and histones is similar. In contrast to HDA, the HDB activity was found to be less sensitive to TSA, with a K i at least 10-fold greater. We also found that the peak HDA activity (Mono S fraction 12) was inhibited by 80% at 10 nM TSA at 50 -100 g/ml histone, whereas the HDB activity (Mono S fraction 25) was inhibited by Ͻ20% (Fig.   FIG. 1. Column chromatography of HDA histone deacetylase. A, Mono Q HR 10/10 chromatography fractions were assayed for deacetylase activity as described under "Materials and Methods." 25 microliters of each 7.5-ml fraction were assayed for activity; total counts/minutes released for a 30-min assay are indicated (E). B, fractions 7-9 were pooled, dialyzed, and loaded onto a hydroxyapatite column (10 mm ϫ 12 cm). 25 microliters of each 4.0-ml fraction were assayed as described above, and fractions 14 -18 were pooled and dialyzed. C, shown is the Superdex 200 gel filtration chromatography (10 mm ϫ 46 cm) profile with activity for 10 l of each 1.0-ml fraction. Peak activity fractions 6 -8 were pooled and dialyzed. Protein molecular mass standards (in kilodaltons) are indicated by arrows. D, shown is the Mono S HR 5/5 chromatography profile and activity measured from 25 l of each 1.0-ml fraction. HDA Activity Co-chromatographs with Four Peptides (p75, p73, p72, and p71)-Four peptides with relative molecular masses of 75, 73, 72, and 71 kDa co-migrate with enzyme activity in the Mono S fractions on SDS-PAGE (Fig. 5). Sequence analysis of gel-isolated proteins has confirmed that p75, p73, and p71 are distinct from each other. 2 We have not yet confirmed that p73 and p72 have different primary structures. It is interesting to note that the band representing the 73-kDa peptide is less obvious in the earlier Mono S fractions (Fig. 5,  fraction 10). While the decrease in the p73 band intensity does not destroy enzyme activity, we did notice that loss of p73 in fraction 10 is inversely proportional to the appearance of a smaller 61-kDa peptide. Whether the 61-kDa species is a degradation product of p73 has not yet been further investigated.
p75, p73, p72, and p71 Are Associated with an Active Complex under Native Conditions of Electrophoresis-To determine if p75, p73, p72, and p71 are present in a single active complex, native gel electrophoresis was performed on peak enzyme fractions from the Mono S column as described under "Materials and Methods." The peak enzyme activity was found within one 3.3-mm slice of the native tube gel (slice 4), migrating slightly more than 1.0 cm into the 10-cm tube gel. No other active complex was found within the remainder of this tube or in an identical tube run under reverse polarity. SDS-PAGE analysis of proteins in this portion of the tube gel (Fig. 5, right-most lane) reveals the presence almost solely of proteins with the migration of p75, p73, p72, and p71, suggesting that the active complex contains these four peptides.
Antibodies Directed against p75 and p71 Immunoprecipitate an Active Deacetylase Complex-To confirm the importance of the purified peptides in constituting the deacetylase complex, immunoprecipitation was performed using antibodies directed against p75 and p71. These were raised against E. coli translation products of their encoded genes, HDA1 and HDA3, respectively, which were cloned using the sequenced peptides of p75 and p71 to design probes for the genes. 2 While immunoprecipitation from a large quantity of crude nuclear extract can immunoprecipitate an active complex, this contains not only the very rare HDA1 (p75) and HDA3 (p71) proteins, but many additional peptides. This is presumably due to the presence of abundant cross-reactive proteins in the crude mixture, as revealed by silver staining (data not shown). Therefore, a partially purified extract of higher activity and purity was used as described under "Materials and Methods." Immunoprecipitation with the anti-HDA1 antibody (Fig. 6A) from partially purified extracts (lane 5; input) demonstrates the presence of peptides (lane 3) having similar electrophoretic (SDS-PAGE) migration compared with p75, p73, p72, and p71 from highly purified Mono S fractions rechromatographed on Mono Q (lanes 1 and 2 represent Mono S fractions 12-14 and 9 -11, respectively). This contrasts with bands precipitated by the preimmune serum (lane 4). Note that the immunoprecipitated material (lane 3) contains relatively less of the p73 species as compared with fractions 12-14 (lane 1). This may be due to the lability of p73 as described above. We also observed a faint additional band present between the p75 and p73 proteins that is present in lanes 1-3. This band is found in fractions across FIG. 4. Sensitivity to TSA differs in fractions containing HDA versus HDB histone deacetylase activities. Inhibition at 100 g/ml histone substrate was calculated as a ratio between enzyme reactions containing 10 nM TSA versus no TSA. Enzyme reactions were performed in triplicate for a fixed time point and represent an average percent inhibition. 10-Microliter Superdex 200 fractions from the wild-type extract as indicated in Fig. 7 were incubated over various histone substrate concentrations with and without TSA. The ratio of inhibition is indicated for fraction 5 (enriched for HDB) or fraction 7 (enriched for HDA). One-half of the pool of activity from Superdex fractions 5-7 was dialyzed and further chromatographed over a Mono S HR 5/5 column as described under "Materials and Methods." Mono S fractions 12 (HDA) and 25 (HDB) were dialyzed into Io buffer plus 125 mM NaCl and assayed for TSA inhibition as described above. The sensitivity to TSA for the HDA-containing fraction 12 and HDB-containing fraction 25 is as indicated.  chromatography profiles in other preparations and may represent a modified form of p75 as we have obtained seemingly identical proteolytic peptide analysis of the two proteins (data not shown).
To assay for the specificity of the antibodies and to determine if p75 and p71 co-immunoprecipitate, the precipitates were assayed by Western blotting (Fig. 6B). As shown in Fig. 6B, the HDA1 antibody detects a 75-kDa band (lane 1), while the HDA3 antibody detects only the 71-kDa species (lane 10), showing little if any cross-reaction between the antibodies. Lanes 3 and 8 contain immunoprecipitations using the HDA1 antibody. It is evident that the HDA1 antibody immunoprecipitates both p75 (lane 3) and p71 (lane 8). Conversely, the HDA3 antibody precipitates both p71 (lane 6) and p75 (lane 5). We observed that the HDA1 antibody is more effective at immunoprecipitating both proteins, although we observed only slightly less activity present in the anti-HDA3 immunoprecipitate. We have also found that Western blot analysis using these antibodies demonstrates staining across HDA activity profiles in our chromatographic analyses, but not against fractions containing only purified HDB activities (data not shown).
In addition, as shown in Table III, antibodies directed against both HDA1 (p75) and HDA3 (p71) recovered a significant amount of activity in the pellets (28% for anti-HDA1 and 21% for anti-HDA3). Both antibodies for HDA1 and HDA3 also immunodepleted a significant portion of the total activity (80% for anti-HDA1 and 67% for anti-HDA3), whereas incubation with preimmune serum or an antiserum directed against an unrelated protein (enolase 1) led to no immunodepletion. The residual activity either may be cross-contaminating HDB or may be due to incomplete immunoprecipitation. These data FIG . 5. Proteins p75, p73, p72, and p71 copurify with histone deacetylase activity. Fractions 9 -14 containing enzyme activity from a final Mono S HR 5/5 column were electrophoresed on 8% SDS-polyacrylamide gel and stained with silver as shown. In addition, an aliquot of the pooled and concentrated fractions (12)(13)(14) was electrophoresed on a native polyacrylamide gel as described under "Materials and Methods," and a 3-mm slice (slice 4) containing peak histone deacetylase activity was homogenized to obtain protein and re-electrophoresed under denaturing conditions (NATIVE GEL lane) along with the Mono S fractions described. Molecular mass standards are indicated to the left at 200, 97.3, 69, 61, and 46 kDa. It is evident that the native activity contains mainly the p75, p73, p72, and p71 peptides shown. The band at 61 kDa is a putative degradation product of p73. The activity of each Mono S fraction is indicated by the bar graph directly below the silverstained fractions; the assay was performed as described for Fig. 1 and under "Materials and Methods." FIG. 6. Immunoprecipitation with an antibody directed against HDA1 (p75) indicates the presence of a polyprotein complex containing p75, p73, p72, and p71. A, proteins with the molecular masses of those in the HDA complex co-immunoprecipitate with antibody to HDA1 (p75). HDA enzyme fractions were electrophoresed on 8% SDS-polyacrylamide gel stained with silver. Lane 1, peak HDA activity aliquot from Mono S column fractions 12-14 pooled and rechromatographed over a Mono Q HR 5/5 column as described for Fig. 5; lane 2, identical to lane 1, but contains Mono S fractions 9 -11; lane 3, immunoprecipitation from a partially purified fraction (see "Materials and Methods") using antibody directed against HDA1; lane 4, identical immunoprecipitation done with preimmune serum; lane 5, total input protein used in immunoprecipitations. 1% of the total immunoprecipitation was loaded in lanes 3 and 4. 0.2% of the total input protein was loaded in lane 5. Proteins that both immunoprecipitate and are present in the highly purified Mono Q fractions (lanes 1 and 2) are indicated to the left of the chromatograph (p75, p73, p72, and p71). Note that there is also a faint band between p75 and p73 that is found in lanes 1-3 (please see "Results" for further description). The heavy band just above the 46-kDa marker at ϳ55 kDa in lanes 3 and 4 is the IgG heavy chain from the antibodies used in the reactions. B, Western blot analysis of fractions immunoprecipitated using the HDA1 (p75) or HDA3 (p71) antibody shows the presence of both proteins. Lanes 1 and 10, input protein; lanes 2 and 9, HDA1 preimmune serum immunoprecipitation; lanes 3 and 8, HDA1 antibody immunoprecipitation; lanes 4 and 7, HDA3 preimmune serum immunoprecipitation; lanes 5 and 6, HDA3 antibody immunoprecipitation. The aliquots of input protein, HDA1 preimmune serum immunoprecipitation, and HDA1 antibody immunoprecipitation shown in A were used in B. The gel was electrotransferred to an Immobilon-P membrane, and the membrane was cut into two sections, immunoblotted using either anti-HDA1 (lanes 1-5) or anti-HDA3 (lanes 6 -10) antibody, and visualized as described under "Materials and Methods." Note that both HDA1 and HDA3 are detected in immunoprecipitations with anti-HDA1 (lanes 3 and 8). Similarly, both HDA1 and HDA3 are detected in immunoprecipitations with anti-HDA3 (lanes 5 and 6).
argue strongly that HDA1 (p75) and HDA3 (p71) are members of the same active histone deacetylase HDA complex.
HDA Activity Is Selectively Disrupted by Deletions in the Genes Coding for Either p75 or p71-If the HDA and HDB activities result from the assembly of different protein subunits, one would predict that deletions of subunits in the HDA complex would disrupt the HDA activity, but not the HDB activity. A Superdex 200 profile (on a preparation that retains both HDA and HDB activities) (Fig. 7A) separated two deacetylase activities, revealing a higher molecular mass complex (600-kDa HDB activity; fraction 5) and a smaller HDA complex (fraction 7). Deletions in either the p75 (HDA1) or p71 (HDA3) genes 2 disrupt the smaller HDA activity, yet leave the larger HDB complex intact (Fig. 7). This is especially evident when the peak Superdex fractions 5-7 were pooled and rechromatographed on a Mono S column. This illustrates the elimination of the HDA peak and the retention of the HDB peak in either hda1 or hda3 mutant strains (Fig. 7B). These data demonstrate that HDA1 (p75) and HDA3 (p71) components are required for HDA deacetylase activity on these chromatography columns, but not for HDB enzyme activity. DISCUSSION We have characterized two histone deacetylase activities, HDA (350 kDa) and HDB (600 kDa), from yeast. Several lines of evidence suggest that the activities are functionally distinct from each other and that we have identified at least two peptides (p75 and p71) that are associated with HDA enzyme activity. The HDA activity is highly sensitive to the specific histone deacetylase inhibitor TSA. The HDB activity is only modestly sensitive over a wide range of substrate concentrations. Estimates of the abundance of the HDA deacetylase complex from our purifications suggest that the complex is not very abundant and is represented by Ͻ100 copies/cell. However, the K m for this enzyme is exceptionally low (10.5 nM) for a heterogeneous acetylated histone substrate in vitro, suggesting very high substrate specificity.
Purification of the HDA activity demonstrates four main bands on SDS-PAGE representing peptides p75, p73, p72, and p71 that correlate with activity. On native gels, a slice containing the HDA activity contains almost exclusively peptides with the molecular masses of p75, p73, p72, and p71 when re-electrophoresed by SDS-PAGE. Antibodies to p75 co-immunoprecipitate a complex that contains peptides with molecular masses similar to those in the HDA peak. Western blot analysis confirms that antibodies to p75 co-immunoprecipitate p71 and vice versa. Finally, disruptions of the genes coding for either p75 or p71 cause the loss of the HDA activity from our chromatography columns. These disruptions do not cause the loss of the HDB activity. These data argue strongly that p75 (HDA1) and p71 (HDA3) are components of HDA histone deacetylase activity. We do not yet know genetically whether p73 and p72 are distinct from each other or whether they are required for HDA enzyme activity.
Previously, the partial purification of both a high mass (500 kDa) and a low mass (150 kDa) deacetylase activity has been reported (25). This work showed that the high mass activity was less sensitive to TSA than the low mass activity. Additionally, the high mass form was lost at higher ionic strengths as the low mass form increased. We did not find a 150-kDa activity within our crude nuclear preparations. However, either it may have been removed during the preparation, or our isolation conditions were not conducive to generating the smaller product. It is likely that the activity reported previously as high mass is the same as HDB. Why two histone deacetylase activities exist with different sensitivities to TSA is unclear. The HDA activity is specific for histones in vitro since it will not deacetylate acetylspermidine, and it is strongly inhibited by polypeptides containing chemically acetylated lysine or by histone H3 and H4 N-terminal peptides, whether or not they are acetylated (see "Materials and Methods"). We do not yet know the specificity of the HDB activity.
Most of the HDA activity is directed toward histone H3 in vitro. However, this result must be interpreted with caution regarding in vivo specificity. The HDA activity is capable of active histone deacetylase complex The percent activity recovered in each immunoprecipitated pellet was calculated from parallel assays of a fraction of the immunoprecipitated activity versus a similar fraction of the starting material. The assays were performed as described under "Materials and Methods." Antibody Activity in pellet a (%) Anti-HDA1 (p75) 28 Anti-HDA3 (p71) 21 Preimmune (HDA1) 0 Preimmune (HDA3) 0 Anti-enolase 1 0 a Percent activity as compared with the supernatant of a sample with the same treatment, but no antibody added.
FIG. 7. Mutations in hda1 or hda3 cause the loss of HDA histone deacetylase activity from chromatography profiles. A, gel filtration column. Partially purified extracts from either wild-type (WT) or mutant strains were run on a Superdex 200 column (10 mm ϫ 46 cm) as described under "Materials and Methods." B, Mono S column. Superdex fractions 5-7 from each strain above were pooled, dialyzed, and rechromatographed on a Mono S HR 5/5 column as described under "Materials and Methods." 10 microliters of each 1.0-ml fraction were assayed in duplicate, and total counts/minute released were averaged for each fraction as indicated. Ⅺ, wild-type strain; ࡗ, hda1⌬; f, hda3⌬. deacetylating all four histones, and we have no evidence as yet for the presence of other histone deacetylases that have specificity for histones other than H3 in vitro. It is certainly possible that the apparent specificity for H3 reflects the in vitro assay used. HeLa cells grown in high concentrations of butyrate possess H3 acetylated at Lys-27. However, this is only in the penta-acetylated form, which is also the least abundant even in the presence of the histone deacetylase inhibitor sodium butyrate, composing Ͻ1% of the H3 in HeLa cells (11). It is possible that the turnover rate of H3 Lys-27 acetylation is more rapid than that of other lysines in H3 in the presence of butyrate, resulting in disproportional labeling of this site. The yeast deacetylase may then reverse this reaction (of Lys-27 or other high turnover sites), rapidly resulting in the apparent specificity for H3. In vivo, we find that when the yeast deacetylase activity is inhibited, the acetylation state of all the histones increases. 2 While this report is the first to identify the individual subunits of a histone deacetylase, numerous questions still need to be addressed. As of yet, we do not know the function of the individual subunits of the HDA complex and whether either HDA1 or HDA3 is the catalytic subunit. Moreover, histone deacetylation may require the targeting of the complex to specific chromosomal domains. It is possible that one of the subunits identified has this role. While the HDB complex is distinct in its sensitivity to TSA and its response to HDA-specific gene disruptions, we do not know whether it is functionally distinct, histone-specific, and/or redundant in vivo. We have also not eliminated the possibility that these two complexes share a common catalytic subunit that is not removed by gene disruption to HDA1 or HDA3. The differential inhibition by TSA (between HDA and HDB) may then be due to the modulation of its sensitivity due to the altered associations within the two complexes. Finally, we do not yet know whether these two activities are in fact important for gene regulation in yeast. By using both enzyme purification and genetic analysis, it is our hope that these questions will soon be answered.