Histone Deacetylase Is a Component of the Internal Nuclear Matrix*

In chicken immature erythrocytes, approximately 4% of the modifiable histone lysine sites participate in active acetylation. There are two categories of actively acetylated histone H4. Although both are acetylated at the same rate (tllz = 12 min), one is acetylated to the tetraacetylated form and is rapidly deacetylated (class l), and the other is acetylated to mono- and diacety- lated forms and is slowly deacetylated (class 2). We show that the chromatin distribution of the class 1 labeled tetraacetylated H4 species paralleled that of the transcriptionally active DNA sequences. For ex- ample, the chromatin fragments of the insoluble nuclear material contained 76% of the active DNA and 74% of the labeled tetraacetylated H4. Class 2 labeled acetylated H4 species were found in repressed chro- matin and were enriched in active/competent gene-enriched chromatin fragments. The majority of the histone deacetylase activity (75-80%) was located with the insoluble residual nuclear material. Further, approximately 40-50% of the enzyme activity was associated with nuclear matrices prepared by two methods using high salt and intermediatelhigh salt ex- traction. Histone deacetylase was solubilized by extracting the nuclear matrices with high salt and 2- mercaptoethanol, a procedure that generates nuclear pore-lamina complexes. These results demonstrate that histone deacetylase is a component of the internal nuclear matrix.

acetylated histones in chromatin preparations. In chicken immature erythrocytes approximately 4% of the modifiable histone lysine sites participate in active acetylation and deacetylation (Zhang and Nelson, 1986). In these cells there is only one rate of acetylation which has a tllP of approximately 12 min (Zhang and Nelson, 1988a;Hendzel and Davie, 1991). However, there are two categories of metabolically active histone acetylation. One type of acetylated histone species becomes hyperacetylated (e.g. the tetraacetylated form of histone H4) in the presence of sodium butyrate, a histone deacetylase inhibitor. Upon removal of the inhibitor, the hyperacetylated histone species are rapidly deacetylated ( tIr2 = 5 min; Zhang and Nelson, 1988b). We refer to this type of metabolically active acetylation as dynamic, class 1 acetylation. Another population of the metabolically active acetylated histone species only achieve low levels of acetylation (e.g. mono-and diacetylated forms of histone H4) in the presence of sodium butyrate. In the absence of butyrate these histones are slowly deacetylated. We refer to this type of acetylation as class 2 acetylation.
Transcriptionally active gene chromatin has a soluble and/ or insoluble nature; that is, active genes are located in two types of chromatin fragments: those that are soluble in 0.15 M NaCl and/or 2 mM MgClz and those that co-fractionate with the low salt-insoluble residual nuclear material (Rose and Garrard, 1984;Cohen and Sheffrey, 1985;Stratling et al., 1986;Einck et al., 1986;Stratling, 1987;Delcuve and Davie, 1989). Interestingly, the partitioning of inducible genes with the low salt-insoluble residual nuclear material was related to their transcriptional activity (Stratling et al., 1986;Einck et al., 1986).
In chicken immature erythrocytes, transcriptionally active DNA sequences are located in chromatin fragments that are soluble in 0.15 M NaCl and in chromatin fragments that are associated with the low salt-insoluble residual nuclear material (Delcuve and Davie, 1989). The latter chromatin fraction was found to be enriched in histones H3 and H4 which were undergoing methylation and in newly synthesized histones, both of which paralleled active gene content Davie, 1989,1990). The association of the class 1, dynamically acetylated histones with the 0.15 M NaCl or 2 mM MgCLsoluble, active gene chromatin fragments has been demonstrated (Zhang and Nelson, 1988a;Ridsdale et al., 1990). However, whether the histones associated with the active gene chromatin fragments bound to the low salt-insoluble residual nuclear material partake in dynamic acetylation remains to be determined. Furthermore, little is known about the chromatin distribution of the class 2 acetylated histones.
Dynamic acetylation is a consequence of the activities of two histone-modifying enzymes, the histone acetyltransferase(s) and the histone deacetylase(s). Because dynamic acetylation is a nonrandom phenomenon restricted to specific regions of the eucaryotic genome, to aid in the understanding Histone Deacetylase(s) and Histone Acetylation in Chromatin 21937 of this process it is necessary that the nuclear distribution of these two enzymes be determined. In chicken erythrocytes the histone acetyltransferase(s1 is highly enriched in the transcriptionally active/competent gene-enriched, 0.15 M NaC1-soluble polynucleosomes (Chan et al., 1988). In contrast, little is known about the distribution of histone deacetylase(s) within the nucleus. Hay and Candido (1983a, 198313) have reported the presence in HeLa cells of histone deacetylase activity in high molecular weight material after mild micrococcal nuclease digestion. This fraction resembles the nuclear scaffold in that it is partially dissociated with 2-mercaptoethanol or neocuproine (Hay and Candido, 1983a). Mold and McCarty (1987) reported that histone deacetylase is associated with a unique class of mononucleosomes which can be resolved by nondenaturing polyacrylamide gel electrophoresis.
In this study we determined the chromatin distribution of the metabolically active class 1 and class 2 acetylated histone H4 species and the histone deacetylase activity in chicken immature and mature erythrocytes. We provide evidence that dynamic class 1 histone H4 acetylation was limited to the transcriptionally active chromatin regions. Dynamically acetylated H4 histone species were associated with the active gene-enriched, salt-soluble chromatin fragments and the active chromatin fragments bound to the residual nuclear material. Class 2 acetylated histone H4 species were found in repressed, active and competent chromatin, with a preference for the latter two chromatin regions. The majority of the histone deacetylase activity was located with the residual nuclear material. We demonstrate that the histone deacetylase is a component of the internal nuclear matrix.

Isolation and Incubation of Chicken Erythrocytes
Chicken mature and immature erythrocytes were isolated from adult White Leghorn chickens as described (Hendzel and Davie, 1989). Cells were washed once with an isotonic buffer (130 mM NaC1, 6.2 mM KCI, 7.5 mM MgC12, 10 mM Hepes,' pH 7.2, and once in Swim's S-77 medium (Sigma), pH 7.2). Cells resuspended at onethird volume packed erythrocytes and two-thirds volume medium were preincubated for 30 min with 20 pM cycloheximide. Cells were then labeled with ["Hlacetic acid, sodium salt (27 Ci/mmol; ICN Radiochemicals) at a final concentration of 0.1 mCi/ml. Labeling proceeded for either 15 or 120 min as indicated. Cells were collected by centrifugation and washed with Swim's S-77 medium containing 0.1 mM sodium acetate and 10 mM sodium butyrate. Subsequently, cells resuspended in the same medium with sodium acetate and sodium butyrate were incubated a further 60 min essentially as described by Zhang and Nelson (1988a).

Nuclei Isolation, Digestion, and Chromatin Fractionation
Nuclei were digested and the chromatin fractionated as described previously (Delcuve and Davie, 1989) except that the micrococcal nuclease incubations (Pharmacia LKB Biotechnology Inc.) were for 20 min. Under these digestion conditions 2.0-2.4% of the nuclear DNA was rendered acid soluble. The 0.15 M NaCI-soluble chromatin lragments of fraction SI,,, were size fractionated on either a 2.8 X 90cm Bio-Gel A-5m column at a flow rate of 25 ml/h or a 2.8 X 20-cm Rio-Gel A-5m column at a flow rate of 1.2 ml/min. Protein concent.ration of the column fractions was determined by the Bio-Rad protein microassay. Fraction PE was extracted with 0.6 M NaCI. The suspension was then kept on ice for 10 min and then centrifuged at 12,000 X g for 10 min to obtain a soluble (So.,;) and an insoluble fraction (P,, ,J.

Preparation of Nuclear Matrices
High Salt Nuclear Matrix Isolation-NaC1 matrices from chicken erythrocytes were prepared according to the procedure of Cockerill I The abbreviation used is: Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.
IntermediatelHigh Salt Nuclear Matrix Isolation-Nuclear matrices from chicken erythrocytes were prepared as described by Roberge et al. (1988). Nuclei were resuspended a t a concentration of 40 Ar,,r/ml in RSB, 0.25 M sucrose, pH 7.5, and digested for 1 h at room temperature with 200 pg/ml DNase I. Nuclei were collected by centrifugation and the supernatant (fraction S ) , which contained very low histone deacetylase activity, was discarded. The nuclear pellet was resuspended in 0.4 M KCI, 0.2 mM MgCl,, 1 mM phenylmethylsulfonyl fluoride, 0.25 M sucrose, and 10 mM Tris-HCI, p H 7.4. After a 15-min incubation on ice the matrices were collected by centrifugation, and both the pellet (fraction 0.4 M KC1 P) and supernatant (fraction 0.4 M KC1 S) were saved. The pellet was resuspended in the same buffer with 2.0 M KC1 and treated as before. The supernatant (fraction 2.0 M KC1 S) and pellet (fraction 2.0 M KC1 P) fractions were dialyzed against TEN, pH 8.0. Often a precipitate would form after dialysis, and this was resuspended in the supernatant, and the samples were assayed as suspensions.
Nuclear Pore-Lamina Complex Isolation-Nuclear pore-lamina complexes were prepared from chicken erythrocytes as described by Kaufmann et al. (1983). Nuclei were resuspended at a concentration of 40 A,Bo/ml in RSB, 0.25 M sucrose, pH 7.5, and digested for 1 h on ice with 500 Ng/ml DNase I and 500 pg/ml RNase A (boiled). The nuclear pellet was saved after centrifugation. The supernatant (fraction S ) , which contained very low histone deacetylase activity, was discarded. The nuclear pellet was resuspended in a low salt buffer (10 mM Tris-HCI, p H 7.4, and 0.2 mM MgSO,). High salt buffer (10 mM Tris-HCI, pH 7.4, 0.2 mM MgSO,, 2 M NaCI) was then added to a final NaCl concentration of 1.6 M. 2-Mercaptoethanol was added subsequently with gentle agitation to a final concentration of 1% (v/ v), and the suspension was incubated on ice for 15 min. The nuclear pore-lamina complexes (pellet) and supernatant were collected by centrifugation. The pellet was resuspended in the same manner in the absence of 2-mercaptoethanol. After a 15-min incubation on ice, the pellet (fraction 1.6 M NaCl P) and supernatant were collected.
The supernatants from the salt-extracted nuclear pellets were pooled, yielding fraction 1.6 M NaCl S. The pellet was resuspended in low salt buffer, and all fractions were dialyzed overnight against TEN, pH 8.0.

Assay for Deacet-ylase Activity
Chromatin fractions were brought to equivalent AZ6!, concentrations (typically 4 absorbance units in a final volume of 300 pl) in TEN buffer. For the suspension of nuclear matrices or nuclear envelopes and the accompanying supernatant fractions, a portion (300 pl) of the fraction was assayed. Approximately 100 pg of acidsoluble histones isolated from ["Hlacetate-labeled erythrocytes was added to each reaction. Cells were then incubated for 60 min at 37 "C. Nonenzymatic release of label was typically below 100 dpm. Incubations were typically done in triplicate. Incubation was terminated by adding acetic acid/HCl to a final concentration of 0.12 N/0.72 N, and 2 volumes of ethyl acetate were added to the reaction mix. The samples were centrifuged for 2 min in a microcentrifuge. Half of the added volume of ethyl acetate was then counted in a scintillation counter to determine the amount of liberated label. The enzyme activity of fraction PE, which was the chromatin fraction with the highest level of activity, was linear up to 16 A,,,, units in a 1-h incubation, and the enzyme activity was linear with time up to 2 h at 4 units of fraction PE.

Preparation and Analysis of Samples
DNA was isolated and quantitated as described by Delcuve and Davie (1989). Histones were isolated from the various chromatin fractions by extraction with 0.4 N HzSO, as described (Nickel et al., 1987). Polyacrylamide gel electrophoresis was performed as described (Nickel et al., 1987). Fluorography was performed as described bv Hendzel and Davie (1989).
Histone Deacetylase(s) and Histone Acetylation in Chromatin

RESULTS
Dynamically Acetylated Histones Are Preferentially Associated with Transcriptionally Actioe Gene Chromatin-To label the dynamically acetylated (class 1) histones, chicken immature erythrocytes were pulse labeled with ["Hlacetate for 15 min followed by a 60-min chase in Swim's S-77 medium containing 10 mM sodium butyrate. Fig. lR, lane T, shows that during this 60-min chase period a portion of the metabolically active histone H4 population became hyperacetylated, with approximately 33.7% of the labeled histone H4 being tetraacetylated. In addition to histone H4, the acetylated species of histones H2A, H2A.Z, H3.2, H3.3, and H2R were also labeled (Fig. lR, lane T). We observed that of the two histone H3 variants, H3.2 and H3.3, the latter variant preferentially participated in dynamic acetylation. Densitometric analysis of electrophoretic patterns of histones resolved on long AUT (acetic acid/6.7 M urea, 0.375% (w/v) Triton X-lOO)-polyacrylamide gels (not shown) demonstrated that histone variant H3.3 was labeled to a specific activity 3.2-fold greater than t.hat of histone H3.2, the major histone H3 variant in chicken erythrocytes.
We have described previously a chromatin fractionation protocol that separates chromatin fragments differing in their content of transcriptionally active, competent, and repressed DNA sequences ( Fig. 2; Table I; Delcuve and Davie, 1989). Transcriptionally active DNA sequences ( e . g &globin, histone H5, and histone H2A.F) were enriched in 0.15 M NaCI- Ten pg of acidsoluhle protein from each chromatin fract ion was electrophoretically resolved on 15% polyacrylamide AUT gels. P o n d A shows the Coomassie 13lue-stained gel pattern, and p o n d H shows the accompanying lluorogram. The acetylated species of histone H4 are denoted numerically 0 , I, 2, 3 , and 4 , representing the un-, mono-, di-, tri-, and tetraacetylated species, respectively. u denotes the uhiquitinated histone species.

TARIX 1 Chromatin distrihution of thr lrlraacrl,vlalrd hislonv H 4 spwirs in chichrn irnmnturr rpthrocytcs
Histones from the chromatin fractions (see Fig. 2) isolated from cells pulse laheled for 15 min or 2 h were electrophoresed on 1.5"; AUT-polyacrylamide gels and suhjected to fluorography. The flrrorogram was analyzed hy densitometry. and the rel~ltive proportion o f laheled acetylated histone H4 to total laheled histone H4 was determined. The percentages of active (histone H5 antl ,f-glnhin) and competent (6-glohin) IINA sequences in the chromatin frartions were determined from data presented in 1)elruve and Ihvie (1989 (e.g e-globin, c-myc, thymidine kinase), which are no longer expressed in immature or mature erythroc.ytes of adult birds, retain a DNase I-sensitive structure. Fig. 1 demonstrates that relative to total histones, the levels of the labeled acetylated histones were elevated in the 0.1 5 M NaCI-soluble chromatin fragments (fractions SI , , , FI, FII, FIII. and which contained salt-soluble mononucleosomes) and fraction P,.:. The 0.15 M NaCI-insoluble chromatin fragments (fraction PI7,,J had low amounts of the labeled histones, with mono-and diacetylated histone H4 forms being labeled. Labeled tetraacetylated histone H4 was not detectahle in this fraction. The highest concentration of labeled histones was present in the 0.15 M NaCI-soluble oligonucleosomes (Fll antl FII1). Note that the content of labeled histones H2A.Z and ubiquitinated H2R in these fractions was significant. Labeled histone H2A.Z, hut not ubiquitinated H2H, was detected in fraction PI.:.
T o ascertain the extent to which the histones in each chromatin fraction were participating in dynamic class 1 acetylation, we determined the percentage of the labeled histone H4 forms of each fraction which were tetraacetylated (Fig. 3). The small amount of H4 of fraction P I , , , , which was metabolically active was of the class 2 acetylation t-ype as indicated by the label being localized in the mono-and diacetylated species (Fig. 3, Pl:,,,15). I t should he noted that there are specific patterns of site usage for acetylation which differ in their rates of turnover (Pesis and Matthews, 1986;Chicoine et al., 1986;Turner, 1989;Thorne rt al., 1990). Thus, the distribution of label among the various acetylated species of H4 shown in Fig. 3 represents averages of kinetically different isoforms. The histone H4 of the highly competent geneenriched, 0.15 M NaCI-soluble polynucleosomes (F115) was a combination of the class 1 and class 2 acetylation types, with 28.4% of the labeled H4 being tetraacetylated. Histone H4 of the 0.15 M NaCI-soluble oligonucleosomes (Flll15) and of the T o determine the chromatin distribution of the dynamically acetylated class 1 histones, the percentage of labeled tetraacetylated histone H4 located in each chromatin fraction was ascertained. Table I demonstrates that the distribution of the labeled tetraacetylated H4 species (class 1 acetylation) paralleled that of active DNA among the chromatin fractions. A similar match in the partitioning of competent DNA was not as obvious. The chromatin fragments associated with the residual nuclear material (fraction PE) contained the majority of the active DNA and labeled tetraacetylated histone H4 species. The remainder of the labeled tetraacetylated histone H4 was located with the active and competent gene-enriched 0.15 M NaC1-soluble chromatin fragments (fraction Size fractionation of these fragments demonstrated that the 0.15 M NaC1-soluble oligonucleosomes (FII and Fill) had a greater proportion of the labeled tetraacetylated H4 than the 0.15 M NaCl-soluble polynucleosomes, illustrating the parallel partitioning of class 1 acetylated histones and active DNA. Similar results were obtained when we repeated these experiments with mature erythrocytes.
Metabolically Actioe Class I Acetylated Histone Species in Fraction Pf: Co-purify with Actioe Gene Chromatin-The chromatin fragments associated with the residual nuclear material contained competent and repressed DNA as well as an enrichment in active DNA. Fig. 4A shows that the salt elution characteristics of bulk DNA and active DNA sequences from the PE fraction differ. Bulk chromatin was eluted more readily than the transcriptionally active histone H5 gene chromatin. We used this differential extractability of bulk uersus active gene chromatin to obtain a greater level of enrichment of active gene chromatin. Fig. 4C shows that after the treatment of the fraction PE with 0.6 M NaCl, the amount of labeled histones associated with the insoluble chromatin (fraction Po.6) was elevated. A striking 51.1% of the labeled H4 of fraction Po.6 was tetraacetylated. Moreover, the percentage of total labeled tetraacetylated H4 (dynamic, class 1 acetylation type) in fraction Po.6 correlated with the enrichment in active gene chromatin fragments ( Table I).
Chromatin Distribution of Class 2 Acetylated Histone Spe- cies-To label the class 2 acetylated histones, chicken immature erythrocytes were pulse labeled with ["HI acetate for 2 h followed by a 60-min chase in Swim's S-77 medium containing 10 mM sodium butyrate. Incubation of cells in the absence of sodium butyrate resulted in the loss of the 0.15 M NaCl solubility of the active and competent gene polynucleosomes. Incubation of the cells with sodium butyrate was necessary to restore this solubility (Ridsdale et af., 1990). The slow rate of deacetylation of the class 2 acetylated histones and the rapid deacetylation of the class 1 acetylated histones result in the preferential labeling of the class 2 acetylated histones during this 2-h labeling period. Fig. 3 shows that the predominating labeled H4 species of unfractionated chromatin (T2h) were the mono-and diacetylated forms. For all of the chromatin fractions except the 0.15 M NaCI-soluble oligonucleosomes (fraction F1112h), the labeled mono-or diacetylated H4 species were a t a higher level than the labeled tetraacetylated H4 (Fig. 3).
The content of the labeled histones of each chromatin fraction was similar to those shown in Fig. 1, with 0.15 M NaC1-soluble oligonucleosomes (fractions FII and Fill) having the greatest concentration of labeled histones and the 0.15 M NaC1-insoluble chromatin fragments having the lowest levels (not shown). The chromatin distribution of the labeled monoand diacetylated histones (class 2 acetylated histone forms) was different from that of class 1 acetylated (tetraacetylated H4) histone (Table I). Although the 0.15 M NaC1-soluble polyand oligonucleosomes (fractions FI, FII, and F,II) were enriched in class 2 acetylated histones (5.6-, 7.5-and 4.6-fold, respectively), the 0.15 M NaC1-insoluble chromatin fragments contained a substantial amount of the labeled mono-and diacetylated histone H4 species (Table I). Further, these labeled histone forms were only slightly enriched (1.3-fold) in the chromatin fragments complexed to the insoluble nuclear material (fraction PE). These observations suggest that class 2 acetylated histones were associated with repressed, competent and active gene chromatin fragments, with a preference toward competent gene chromatin fragments. Obviously, the chromatin distribution of the class 2 acetylated histones did

Histone Deacetylasefs) and Histone Acetylation in Chromatin
not parallel that of the active DNA.

Distribution of Histone Deacetylase Activity in Chicken Immature and Mature Erythrocyte Chromatin-
The analysis of the immature and mature erythroid chromatin distribution of the dynamically acetylated histones demonstrated that the majority of the rapidly deacetylated, class 1 histone H4 was localized with the chromatin fragments complexed to the insoluble nuclear material. This observation suggested that the majority of the histone deacetylase activity would also be localized in this fraction. Table I1 shows the activity of the histone deacetylase of the chromatin fractions. Fraction PE was found to possess the majority of the enzyme activity. Of three fractionations, the proportion of the total histone deacetylase activity located in the EDTA-insoluble residual nuclear material (fraction PE) was highly reproducible, with this fraction containing 77.6-78.9% of the enzyme activity. Zhang and Nelson (1988b) reported that the deacetylation rate of labeled histone H4 of mature erythrocytes was lower than that of immature cells. Consistent with this observation, we found that the histone deacetylase activity of mature erythroid nuclei was approximately 35% that of the enzyme activity of immature erythroid nuclei. However, the distribution of the histone deacetylase activity among the mature erythroid chromatin fractions was similar (Table 11).
Whereas treatment of the residual nuclear material (fraction PE) with 2 M NaCl extracted approximately 90% of the associated chromatin (see Fig. 4), the histone deacetylase activity was not solubilized, suggesting that the enzyme was not bound to chromatin. Also, the enzyme activity remained associated with the residual nuclear material that was treated with 10 mM 2-mercaptoethanol, indicating that the enzyme was not retained by the residual nuclear material (nuclear matrix) solely via disulfide bonds.

Histone Deacetylase Is a Component of the Nuclear Matrix-
The observation that histone deacetylase activity was primarily located in fraction P E suggested that the enzyme may be a component of the nuclear matrix. Nuclear matrices of immature and mature erythrocytes were prepared by extensively digesting nuclei with DNase I and subsequently extracting the residual nuclear material with 2 M NaCl (Fig. 5, high salt  matrix). Table I1 shows that 50-57% of the histone deacetylase activity was associated with the nuclear matrix of mature and immature erythrocytes.
There is concern that the direct high salt extraction of Chromatin distribution of histone deacetylnse actiuity in chicken immature erythrocytes Four AZM units of the various chromatin fractions (see Fig. 2) were assayed for deacetylase activity by incubation with approximately 100 pg of substrate for 1 h at 37 "C. The released label in ethyl acetate was determined by scintillation counting, and the distribution of enzyme was determined by multiplying the dpm released by the percentage of total in the chromatin fraction and then dividing b y the dpm value for fraction T. For nuclear matrix preparations (see Fig. 5, high salt matrix). the total enzyme activity in the pooled soluble fractions and the matrix was determined, and the distribution among the fractions was the percentage of the combined activities.
Each value represents the mean f S.E. from three measurements.

TABLE I11
Histone deacetylnse actiuity is a component of the nuclear matrix Equivalent proportions of the various fractions from chicken immature erythrocytes were assayed for deacetylase activity. Fig. 5 shows the procedures used to isolate the intermediate/high salt matrix, which yields fractions 0.4 M KC1 S, 2.0 MKCI s, and 2.0 M KC1 P (matrix), or nuclear pore-lamina complexes, which yield fraction 1.6 M NaCl S and 1.6 M NaCl P (nuclear pore-lamina complex). The total enzyme activity in the various fractions was determined, and enzyme distribution was quantitated by determining the relative activity of each fraction to the combined total. Each value represents the mean f S.E. from three measurements.  Fig. 5). Thus, we prepared nuclear matrices by the intermediate/high salt matrix isolation procedure. Table I11 shows that 42% of the histone deacetylase remained bound to the intermediate/high salt prepared nuclear matrices of immature erythrocytes. These results provided strong evidence that approximately half of the histone deacetylase activity of mature and immature erythrocytes was complexed to the nuclear matrix.
We attempted to determine the histone deacetylase activity associated with the nuclear scaffold prepared by the lithium diiodosalicylate low ionic strength procedure (Mirkovitch et al., 1984). However, we found that lithium diiodosalicylate irreversibly denatured the enzyme, and we were unable to detect enzyme activity in either soluble or scaffold fractions (data not shown).
To determine whether the histone deacetylase was a component of the internal nuclear matrix or of the nuclear porelamina fraction, nuclear pore-lamina complexes were isolated according to the protocol of Kaufmann et al. (1983) (see Fig. 5, nuclear pore-lamina complex). Treating nuclear matrices with 1% 2-mercaptoethanol, 1.6 M NaCl solubilized the internal matrix, leaving the nuclear pore-lamina complexes. This procedure was recently used to solubilize and purify a protein (ARBP) which is responsible for binding DNA to the nuclear matrix (von Kries et al., 1991). Table I11 shows that approximately 95% of the histone deacetylase activity was found in the 1% 2-mercaptoethanol, 1.6 M NaCl extract. The nuclear pore-lamina complex contained less than 5% of the histone deacetylase activity. These results demonstrated that like Histone deacetylase is associated with the 150 mM NaC1-soluble poly-and oligonucleosomes. Approximately 3 mg of fraction Sls0 in a volume of 1.0 ml was applied to a Bio-Gel A-5m (2.8 X 20-cm) column. The column was run at a flow rate of 1.2 ml/ min in 10 mM Tris-HC1, pH 8.0, 1 mM EDTA, and 150 mM NaCI, and 1-ml fractions were collected. A portion (290 pl) of the column fraction was then assayed for deacetylase activity as described under "Materials and Methods." ARBP the histone deacetylase is a specific component of the internal nuclear matrix. Fig. 6 shows the elution of enzyme activity uersus the elution of the 0.15 M NaC1soluble chromatin fragments from a Bio-Gel A-5m column. Four peaks of enzyme activity were detected. Peaks 1 and 2 were associated with the poly-and oligonucleosomes, respectively. The third peak of enzyme activity eluted with the mononucleosomes, and the fourth peak eluted from the column in a position consistent with the elution of free enzyme.' Approximately 28, 15, and 57% of the histone deacetylase activity in fraction slso was present in the 0.15 M NaClsoluble poly-and oligonucleosomes (peaks 1 and 2), mononucleosomes (peak 3), and free enzyme (peak 4), respectively. The specific enzyme activity of the oligonucleosomes (fraction 30; 14.7 dpm/pg of protein) was greater than that of the polynucleosomes (fraction 18; 12.2 dpm/pg of protein) or mononucleosomes (fraction 40; 10.7 dpmlpg of protein).

DISCUSSION
Chicken immature erythrocytes contain two categories of metabolically active acetylated histones. Both groups of histones are acetylated at the same rapid rate, but they differ in both the extent of acetylation and the rate of deacetylation. We provide evidence that the dynamically, class 1 acetylated histones, which attain high acetylation levels and are rapidly deacetylated, are complexed principally to active DNA. Parallel chromatin distributions of the active DNA sequences and labeled tetraacetylated histone H4 species were observed, with the 0.15 M NaC1-soluble oligonucleosomes and the chromatin fragments bound to the insoluble nuclear material containing the majority of the active DNA and labeled tetraacetylated histone H4. The strong bias of the dynamically acetylated class 1 histones to active gene-enriched chromatin regions was not apparent with the class 2 acetylated histones. These metabolically active acetylated histones, which are acetylated to low levels and are slowly deacetylated, were located in repressed, competent and active chromatin regions. However, it is important to note that the class 2 acetylated histones were more abundant in active and competent geneenriched chromatin fractions. The enrichment of the class 2 acetylated histones in the 0.15 M NaC1-soluble poly-and M. J. Hendzel and J. R. Davie, unpublished results. oligonucleosomes was 8.0-fold greater than the 0.15 M NaClinsoluble, repressed gene chromatin fraction. This implies that the majority of the histones associated with the 0.15 M NaC1-insoluble chromatin fragments do not participate in metabolically active acetylation, and they remain at the acetylation levels observed on Coomassie Blue-stained AUT gels.
The spectrum of labeled acetylated histones of the active gene-enriched chromatin fractions, the 0.15 M NaC1-soluble oligonucleosomes, and the insoluble chromatin associated with the residual nuclear material were similar, but some differences were noted. Labeled histones H2A, H2A.Z, H3.2, H3.3, HZB, and H4 were present in both chromatin fractions. Labeled ubiquitinated H2B was found only in the salt-soluble oligonucleosomes. Although the 0.15 M NaC1-soluble polynucleosomes contained HBA.Z, it was labeled to a much lower specific activity than the H2A.Z of the 0.15 M NaC1-soluble oligonucleosomes. It is of interest to note that although the amount of histone H3.2 is considerably higher than that of histone H3.3, the level of labeled histone H3.3 was similar to that of H3.2. Waterborg (1990) reported a similar observation that the alfalfa minor histone H3 variant H3.2 was also preferentially acetylated.
The observation that rapid acetyl group turnover occurs primarily with histones of active gene chromatin (Boffa et al., 1990;Ip et al., 1988;Nelson, 1988a, 1988b;this study) predicts that the modifying enzymes, histone acetyltransferase and histone deacetylase, would also be localized in the active gene chromatin domains. Chan et al. (1988) demonstrated that in chicken erythrocytes the histone acetyltransferase was associated with the nuclease-sensitive regions of the active gene-enriched, 0.15 M NaC1-soluble poly/ oligonucleosomes. Our results show that the histone deacetylase was also found in the active gene-enriched oligonucleosomes as well as active gene-enriched insoluble chromatin fragments associated with the residual nuclear material. These observations are in agreement with other reports showing that the deacetylase of HeLa cells was located in a high molecular weight fraction with some of the characteristics of the nuclear scaffold (Hay andCandido, 1983a, 1983b) and with a subpopulation of mononucleosomes (Mold and Mc-Carty, 1987).
The nuclear matrix is operationally defined as the residual nuclear structure that remains after the high salt extraction of nuclease-digested nuclei. It may represent an underlying structural framework that is the putative site of many nuclear metabolic processes (Cook, 1988;Verheijen et al., 1988). The histone deacetylase is a component of this internal nuclear matrix. The following observations support this conclusion: the enzyme activity was associated with the low salt-insoluble residual nuclear material after a micrococcal nuclease digest, the high salt nuclear matrix, and the intermediate salt/high salt nuclear matrix. We found that more than 40% of the histone deacetylase activity was located in the intermediate salt/high salt nuclear matrix. This result is in marked contrast to the < 10% of the RNA polymerase I and I1 which remained associated with the intermediate salt/high salt nuclear matrix (Roberge et al., 1988) and argues that the histone deacetylase is associated with such a nuclear structure. Histone deacetylase was solubilized by extracting the nuclear matrices with high salt and 2-mercaptoethanol, a procedure that generates empty shells of nuclear pore-lamina complexes. This result demonstrates that as with the matrix/scaffold attachment region-binding protein (von Kries et al., 1991), histone deacetylase is a component of the internal nuclear matrix.
The dual locations of the histone deacetylase with active gene chromatin regions and the internal nuclear matrix raises

Histone Deacetylase(s) and Histone Acetylation in Chromatin
intriguing questions as to the enzyme's function. In addition to its enzymatic role in modulating the levels of acetylated histones complexed to transcriptionally active DNA, the histone deacetylase may also have a structural role similar to that of topoisomerase I1 (Adachi et al., 1991). It is conceivable that histone deacetylase complexed to transcriptionally active chromatin may mediate the dynamic interaction of active gene chromatin with the internal nuclear matrix.