Histone Acetylation Alters the Capacity of the Hl Histones to Condense Transcriptionally Active/Competent Chromatin*

The relationship between histone acetylation and the capacity of Hl histones to cause the 0.15 M NaCl- induced aggregation/precipitation of transcriptionally active/competent gene chromatin fragments was inves- tigated. Previous studies have shown that transcrip- tionally active/competent, but not repressed, gene chromatin polynucleosomes, which were isolated from chicken erythrocytes, remained soluble in 0.15 M NaCl after being reconstituted with Hl histones. This result suggested that some component of the active/competent gene nucleosome altered the capacity of the Hl histones to condense the chromatin fiber. Recently, Hebbes et al. (Hebbes, T. R., A. W., and Crane-Robinson, EMBO J. demonstrated di- rectly that active, but not repressed, gene chromatin of chicken erythroid cells contain high levels of acetylated histones. Here, we show that the solubility of active/competent gene chromatin fragments in 0.15 M NaCl is dependent on

Here, we show that the solubility of active/competent gene chromatin fragments in 0.15 M NaCl is dependent on the level of acetylated histone species, with induction of hyperacetylation increasing the solubility of this gene chromatin. Also, we show that lowering the levels of the acetylated histone forms reduces the ability of the active/competent gene chromatin fragments to resist exogenously added Hl-histone-induced 0.15 M NaCl aggregation/precipitation. These results suggest that histone acetylation alters the capacity of the Hl histones to form compact higher order chromatin structures such that active/competent gene chromatin is maintained in a less folded state than the bulk of chromatin.
The Hl (or linker) histones, which play a key role in the folding of chromatin, are general repressors of gene expression (Weintraub, 1985). Mature chicken erythrocyte nuclei contain linker histones Hl and H5, an extreme variant of Hl, in a stoichiometric ratio of 0.4 of Hl and 0.9 of H5 per nucleosome (Bates and Thomas, 1981). At the final stages of erythropoiesis, histone H5 accumulates in erythroid nuclei. Although there is more than enough Hl histones to cover the entire genome of immature cells, this high level of linker histones does not prevent the transcription of erythroid-specific histone H5 and P-globin genes (Affolter et al., 1987).
Transcriptionally active/competent chromatin is thought to be in an unfolded conformation (as reflected, for example, by an increased sensitivity to DNase I enriched chromatin fractions contain increased levels of modified and variant histone species. Nucleosomes containing such histones may vary somewhat in structure from the bulk and thereby affect nucleosome-dependent transitions to higher order structures. Differences in nucleosomal structure that result in differences in higher-order structure may be of primary importance in determining the role of chromatin structure in the regulation of transcription. One of the most universally observed features of active gene-enriched chromatin is the presence of hyperacetylated nucleosomal histones (Reeves, 1984;Vidali et al., 1988;Loidl, 1988). Hyperacetylation of the amino-terminal region of the nucleosomal histones (H2A, H2B, H3, and H4) may be linked to the chromatin solubility in the presence of Mg2+ and destabilization of chromatin higher order structures Chalkley, 1981, 1982;Allan et al., 1982;Annunziato et al., 1988). Although not all data make a clear distinction, acetylation also seems to affect the structure of the nucleosome (a good discussion of this point is in Oliva et al., 1987). Numerous reports have suggested that histone acetylation functions in the maintenance or control of the transcriptional and/or replicative capacity of chromatin regions (see Loidl, 1988, andVidali et al., 1988 for discussions).
In immature chicken erythrocytes, approximately 3.7% of modifiable histone lysine sites are undergoing acetylation and deacetylation Nelson, 1986,1988, a and. These sites are rapidly acetylated (half-life of 12 min) and deacetylated (Zhang and Nelson, 1988, a and b). The histones undergoing dynamic acetylation are associated with active/ competent DNA Alonso et al., 1987;Zhang and Nelson, 1988a). Moreover, Hebbes et al. (1988) have demonstrated directly that active, but not repressed, genes of chicken erythoid cells are associated with significant amounts of acetylated histones.
We have demonstrated that the 0.15 M NaCl-soluble polynucleosomes from mature or immature chicken erythrocyte nuclei are highly enriched in transcriptionally active (e.g. histone H5 and /3-globin) and competent (e.g. t-globin) genes. These salt-soluble polynucleosomes are enriched in acetylated species of histones H2B, HPA.Z, and H4, poly-and monoubiquitinated species of H2A and H2B, and histone variants H3.3 and H2A.Z. Moreover, these active/competent geneenriched chromatin fragments are complexed with linker histones Hl and H5 (Ridsdale and Davie, 1987;Nickel et al., 1989;Delcuve and Davie, 1989). Reconstitution experiments revealed that active/competent gene-enriched chromatin fragments are much more resistant than repressed gene chromatin fragments to exogenously added linker-histone-induced precipitation in 0.15 M NaCl (Ridsdale et al., 1988). These observations suggest that some feature of the active/competent gene chromatin fragments prevents the linker histones from folding the fiber into a compact higher order structure.

Histone Acetylation and HI Histone Function
In this report, we investigated whether histone acetylation influences the capacity of Hl histones to aggregate/precipitate chromatin fragments in 0.15 M NaCI. Our results show that the degree of solubility of the active/competent gene chromatin fragments in 0.15 M NaCl is correlated with the level of acetylated histone species. Furthermore, the results suggest that the level of histone acetylation is the major determinant of the resistance of active/competent gene chromatin fragments to Hl/HS-induced salt precipitation. These results suggest that histone acetylation alters the capacity of linker histones to form higher order chromatin structures such that transcriptionally active/competent gene chromatin is maintained in a less folded state than the bulk of chromatin.  (Ferenz and Nelson, 1985). Blood was collected and washed of the buffy coat in ice cold 75 mM NaCl, 25 mM EDTA. For the incubation in the presence or absence of sodium butyrate, red blood cells were divided into two portions and washed once more in Swim's S-77 medium, pH 7.2, either with, or without, sodium butyrate added to 10 mM. The cells were incubated with gentle agitation for 60 min at 37 "C in the same media at a density of approximately 25 ml of cells (volume when packed by low speed centrifugation) in 500-ml final volume. After the incubation, the cells were collected by centrifugation, frozen, and stored at -70 "C. Nuclei were isolated and digested with micrococcal nuclease as previously described, with all digestions being for 30 min unless otherwise specified (Ridsdale and Davie, 1987 Thomas (1980). Autoradiograms of hybridized slot blots were scanned with a densitometer, and the data were analyzed as described (Ridsdale et al.. 1988: Delcuve and et al., 1978). In the presence of sodium butyrate, nucleosomal histones complexed to active/ competent genes become hyperacetylated, while in the absence of sodium butyrate, these histones have reduced levels of acetylated species (Alonso et al., 1987;Zhang and Nelson, 1988a). Chromatin isolated from these cells was fractionated. The distribution of DNA among chromatin fractions Pz, SE, Piso, and &So was 25.9 _+ 6.4, 74.4 f 5.9, 66.3 f 6.0, and 7.8 f 0.8 (-butyrate; n = 3) and 20.2 f 2.6, 79.9 +-2.6, 70.5 f 3.0, and 9.5 f 0.1 (+butyrate; n = 3), respectively. Thus, incubation of the cells with or without butyrate did not influence the fractionation of the bulk of the chromatin fragments. The partitioning of active and competent DNA with fraction Px, which is not shown in this report, was not affected by these incubations (Delcuve and Davie, 1989). Fig. 1 shows the protein content of fractions SE and S150 isolated from cells incubated in the absence or presence of butyrate. Acetylation levels can be easily judged by examining the different acetylated forms of H4. The levels of acetylated histone species do not change dramatically in the SE fractions which contain the bulk of erythrocyte chromatin. This is consistent with the results of Zhang and Nelson (1986)  Slier (+butyrate) has greater amounts of the acetylated species of histones H4, H2B, and H3 than fraction S,," (-butyrate) (see Fig. 1). The amount of ubiquitinated histones in the saltsoluble chromatin fractions is similar. The chromatin fragments of fraction &,() were size-resolved by gel exclusion chromatography. Fig. 2 demonstrates that incubation of cells in the absence of butyrate results in a major decline in the 260 nm absorbing material in the polynucleosome fractions.
DNA fragments, which were isolated from chromatin fractions SE, S,w,, and PIdO, were electrophoretically separated on a 1% agarose gel. Fig. 3 (DNA) shows that although the DNA fragment sizes of chromatin fractions SE (+-or -butyrate) are similar, there is a striking difference in the size distribution of the DNA fragments present in fractions S1 i0 (compare +-and -butyrate).
Chromatin fraction Slso (+butyrate) clearly contains a greater abundance of longer DNA fragments than chromatin fraction Slso (-butyrate), as indicated by gel filtration (Fig. 2). Southern blot analysis of the DNA shows that incubation of the cells in the absence or presence of butyrate results in different distributions of the competent DNA (t-globin and vimentin) among the salt-soluble and salt-insoluble chromatin fractions (Fig. 3). The salt-soluble chromatin fraction from butyrate-treated cells is enriched (approximately 3.6-fold) in competent (t-globin and vimentin), but not repressed (vitellogenin), DNA sequences (for determination of whether a DNA sequence is enriched or depleted see "Materials and Methods").
In contrast, the salt-soluble chromatin fragments from cells incubated in the absence of butyrate are depleted in competent DNA sequences (approximately 0.6-fold). In addition, there is an increase in the content of competent DNA found in the aggregation-prone, salt-insoluble chromatin fraction Plso as a result of cells being incubated in the absence of butyrate. The partitioning of the repressed DNA (vitellogenin) among the salt-soluble and -insoluble chromatin fractions is not affected by incubation in the presence or absence of butyrate, and repressed DNA is salt-soluble primarily as mononucleosomes in 150 mM NaCl. The percentage of DNA sequences in the salt-soluble and -insoluble chromatin fractions was quantified ( Table I). The percentage of active/competent DNA sequences in fraction SlsU (+butyrate) is greater than that in fraction SlsO (-butyrate). The enrichment of competent and active DNA in the salt-soluble chromatin fraction decreased 6.0 + 0.7-fold (n = 4; combined average for t-globin and vimentin) and 3.0 + 0.7fold (n = 3; combined average for P-globin and histone H5), respectively, when cells were incubated in the absence of butyrate.
Effect of Histone Acetylation on the 0.15 M NaCl Solubility of Active/Competent Gene Chromatin Fragments Reconstituted with Linker Histones-In a previous report, Hl/H5stripped chromatin fragments of mature chicken erythrocytes were reconstituted with varying 1inker:nucleosomal histone ratios (Ridsdale et al., 1988). Active/competent gene chromatin fragments were more resistant than bulk chromatin fragments to added linker histone-induced NaCl precipitability. The salt-soluble, active/competent gene-enriched chromatin fragments, which contained histones Hl and H5, were enriched in acetylated and ubiquitinated histone species. These observations suggested that some component(s) of the active/competent gene nucleosome altered the capacity of the Hl histones to condense the active/competent gene chromatin fiber in 0.15 M NaCl.
In this study, we determined whether modulating the level of acetylated histone species would alter the ability of active/ competent gene chromatin fragments to resist exogenously added linker histone-induced NaCl precipitation. Histone Hl/ H5-stripped EDTA-soluble chromatin fragments, which were isolated from immature erythroid cells incubated in the presence or absence of butyrate, were reconstituted with varying amounts of linker histones. The linker histone-reconstituted chromatin fragments were fractionated into 0.15 M NaClsoluble and -insoluble fractions. As the amount of reconstituted linker histones approached native chromatin levels (linker histone density = l), there was a decrease in the percentage of chromatin fragments that were salt-soluble, a decrease in the concentration of salt-soluble polynucleosome fragments, and an increase in the level of mononucleosomes (see Ridsdale et al., 1988). Fig. 4  The histones, which were isolated from the salt-soluble chromatin fragments, were electrophoretically resolved on AUT 15% polyacrylamide gels. Fig. 5 shows that as the amount of linker histones added increases to native levels (linker histone density = l), there is an increase in the content of hyperacetylated histone H4 and histone H2B species associated with the salt-soluble chromatin fragments of +butyrate-treated cells. This increase in the level of acetylated histone species is not observed for salt-soluble chromatin fragments of -butyrate-treated cells. Note that the salt-soluble chromatin fractions contain linker histones Hl and H5. We have previously demonstrated that these linker histones are associated with the salt-soluble polynucleosomes (Ridsdale et al., 1988). Fig. 6 shows the solubility of reconstituted chromatin as a function of the amount of linker histone added. Butyrate incubation does not significantly alter the solubility of bulk chromatin at any given amount of added linker histone. The sigmoidal shape of the curve defined by this relation suggests a cooperative interaction among linker histones in giving rise to salt-precipitable chromatin structures. Fig. 7 shows a quantitative assessment of the amount of competent t-globin and repressed vitellogenin gene chromatin soluble in 0.15 M NaCl as a function of the amount of linker histone added. The same relationship is shown for total chromatin.
Note that the total chromatin and vitellogenin curves are distinctly sigmoidal in shape. The t-globin gene chromatin fragments isolated from cells incubated in the presence or absence of butyrate show markedly different degrees of solubility as a function of linker histone density. c-Globin gene chromatin fragments from butyrate-incubated cells remain completely soluble until the amount of linker histones added is equivalent to the levels of these histones found in native chromatin fragments. Further increases in the amount of linker histones reconstituted onto the chromatin fragments result in a decline in the solubility of the globin gene chromatin fragments. However, t-globin chromatin fragments, which were isolated from cells incubated in Hl/H5-stripped chromatin i f fragments of fraction SE, which were isolated from cells incubated in the presence (+) and absence (-) of butyrate, *,auI,,,+&&"H"2'," --I-. The results are expressed as the percentage of 0.2, n = 4; 0.4, n = 4; 0.5, n = 10; 0.7, n = 10; 0.8, n = 11; 0.9, n = 13; gene chromatin fragments in fraction SE that are soluble in 0.15 M 1.0, n = 10; 1.1, n = 4. NaCl.
the absence of butyrate, lost most of their ability to resist linker histone-induced NaCl precipitation. Thus, when about 11% of the bulk chromatin fragments are soluble in 0.15 M NaCl, t-globin gene chromatin fragments from cells incubated in the presence or absence of butyrate are 100% and 18% soluble, respectively. This represents a 5.6-fold difference in the solubility of competent gene chromatin fragments. Active gene chromatin fragments, which are considerably shorter than competent gene chromatin fragments (Delcuve and Davie, 1989), also had this difference in salt solubility, with the ratio of histone H5 gene chromatin solubility (+butyrate:-butyrate) at a nominal l-fold linker histone density being 2.1 f 0.2 (n = 4). (The solubility of bulk chromatin fragments at this linker histone density was 11.3 f 1.5 for chromatin isolated from butyrate-treated cells and 11.0 + 1.0 .z 20 -for chromatin isolated from cells incubated in the absence of butyrate (n = 4).) Within each experiment with reconstituted chromatin (six separate experiments), active/competent gene chromatin was always more enriched in the NaCl-soluble chromatin fragments from cells incubated in the presence of butyrate than those from cells incubated in its absence.

DISCUSSION
Linker histones are known to act in the salt-dependent formation of higher order chromatin structures which are observable by electron microscopy (Thoma et al., 1979). The formation of aggregation-prone, NaCl-precipitable structures shows a similar histone dependency (this study; Allan et al., 1981). The results of Widom's (1986)  These observations suggest that the Hl histones associated with highly acetylated nucleosomes are not able to condense the active/competent gene chromatin fiber.
of the reconstituted chromatin did not diminish this constraint. However, yeast minichromosomes, which contain highly acetylated histones (Davie et al., 1981), do allow thermal untwisting of the DNA (Morse et al., 1987). The liberation of DNA from the nucleosome and/or change in nucleosome shape as a consequence of acetylation may alter the path of the DNA enterinn and leaving the nucleosome which in turn Nelson and colleagues have shown that altering the level of acetylated histones has a profound effect on the solubility of active/competent gene chromatin fragments in buffers containing 3 mM MgC12, with increased levels of acetylated histones resulting in enhanced solubility (Alonso et al., 1987). It should be noted that Hl/H5-stripped chromatin fragments are insoluble in 3 mM Me, but they are soluble in 0.15 M NaCl (Ausio et al., 1986). To be precipitated in 0.15 M NaCl, chromatin fragments must be complexed with Hl histones. Removal of the basic amino-terminal "tails" of the nucleosomal histones, which contain the sites of acetylation, does not prevent the Hl histones from binding to the chromatin fiber (Allan et al., 1982). But removal of the tails does interfere with the capacity of the linker histones to condense the chromatin fiber (Allan et al., 1982). Thus, acetylation of lysyl residues located within the amino-terminal basic domains of the nucleosomal histones may have the same effect on altering Hl histone action as does the removal of this basic domain.
The EDTA-soluble competent gene chromatin fragments isolated from micrococcal nuclease-digested immature erythrocyte nuclei are considerably longer than the active gene chromatin fragments (Delcuve and Davie, 1989). This disparity in chromatin fragment sizes may account for the more noticeable transition of competent (6-fold) uersus active (2-3-fold) gene chromatin to an aggregation-prone state (native or linker histone reconstituted) as a consequence of deacetylation. In studies of exchange of linker histones between chromatin fragments, Thomas and Rees (1983) demonstrated that at an ionic strength of 0.75 M, histone H5 of short chromatin fragments preferred to associate with long chromatin fragments that had formed higher order structures. Short active gene chromatin fragments may be more susceptible than long competent gene chromatin fragments to losing their linker histones. The loss of linker histones and the reduced ability to form higher order structures would tend to decrease the salt-induced aggregation and precipitation of active gene chromatin fragments. Recently, Norton et al. (1989) reported that histone acetylation reduces the amount of negative DNA supercoils constrained by the nucleosome.
Removal of the nucleosomal histone tails also leads to the loss of DNA normally constrained by the nucleosome (Allan et al., 1982). Conversely, Cantor (1985, 1986) demonstrated that plasmid DNA reconstituted with acid-extracted histones was constrained from thermal untwisting, and that trypsin treatment may alter the interaction between Hl histones and nucleosomal/linker DNA (Allan et al., 1980). These alterations in DNA path and linker histone-nucleosome interaction may prevent (or alter) the formation of compact higher order structures.
In conclusion, our observations suggest that, in immature chicken erythrocytes, dynamic acetylation of nucleosomal histones complexed with transcriptionally active/competent DNA prevents histone Hl and H5 interactions that allow for normal condensation of the chromatin fiber. Chan et al. (1988) presented evidence indicating that histone acetyltransferase is preferentially associated with the active/competent gene chromatin domains. This would ensure that the transcriptionally active gene chromatin is maintained in a less folded state than the bulk of chromatin.