The Assembly of Regularly Spaced Nucleosomes in the Xenopus Oocyte S-150 Extract Is Accompanied by Deacetylation of Histone H4*

Histone proteins, which were assembled into chromatin using the Xenopus oocyte 5-150 extract, were analyzed on acid-urea gels and Triton-acid-urea gels to determine their state of modification. We find that histone H4, which is present in a diacetylated form in the oocyte S-150, gradually loses its acetate groups as the DNA is packaged into chromatin. Thus, this process parallels the one observed in vivo during chromatin formation in growing eucaryotic cells. Histone H4 deacetylation in the oocyte S-150 is a DNA-dependent reaction. This reaction is blocked when butyrate (an inhibitor of histone deacetylase) is added at the onset of the chromatin assembly process. When butyrate is added at the end of the assembly process, no de novo acetylation of the nucleosomal histone H4 is observed. Chromatin playing may contain either deacetylated when assembled under when Both chromatin identical upon digestion with The potential applications of this system toward the study of the naturally occurring diacetylated histone H4 are discussed.

contain either deacetylated histone H4 when assembled under standard conditions or diacetylated H4 when assembled in the presence of butyrate. Both types of chromatin display identical structures upon digestion with nucleases. The potential applications of this system toward the study of the naturally occurring diacetylated histone H4 are discussed.
In order to gain a mechanistic understanding of the role that chromatin plays in the regulation of eucaryotic gene expression, a great deal of effort has been devoted in a number of laboratories to the development of i n uitro chromatin assembly systems. Several methods to reconstitute nucleosomes onto DNA using purified components have been reported. These procedures use either dialysis from high salt (Oudet et al., 1975) or negatively charged carrier molecules which can transiently bind histones and promote their transfer to DNA (Laskey et al., 1978;Stein et al., 1979;Earnshaw et al., 1980;Nelson et al., 1981;Bonne-Andrea et al., 1984). Methods involving the slow reassociation of preformed histone H3-H4 tetramers and H2A-H2B dimers onto DNA (Ruiz-Carrillo et al., 1979) or the use of stoichiometric quantities of DNA topoisomerase I (Germond et al., 1979) have also been described. These procedures result in the formation of supercoiled DNA containing nucleosomes, but they fail to assemble regularly spaced nucleosomes, which is the hallmark of chromatin i n uiuo. So far, assembly of a regular nucleosomal array on random sequence DNA i n uitro has only been achieved using crude extracts derived from Drosophila embryos (Nelson et al., 1979), Xenopus eggs (Laskey et al., 1977), and Xenopus oocytes (Glikin et al., 1984;Shimamura et al., 1988).
Much evidence indicates that competition between transcription factors and nucleosomes for binding to regulatory DNA sequences play a key role in the establishment of the active or inactive state of eucaryotic genes (Bogenhagen et al., 1982;Gottesfeld and Bloomer, 1982;Knezetic and Luse, 1986;Matsui, 1987;Workman and Roeder, 1987;Knezetic et al., 1988;Workman et al., 1988). Thus, it would be useful for the study of gene commitment to develop i n uitro systems which reflect the i n vivo situation not only in the final chromatin structure produced, but also in the process by which chromatin is assembled.
During DNA replication, newly assembled nucleosomes are more easily displaced from the DNA than are mature nucleosomes (Schlaeger and Knippers, 1979;Jackson et al., 1981). It has also been observed that the histone H4 that initially binds to the DNA as the chromatin replicates i n uiuo is dimodified. This dimodified histone H4, which carries either two acetyl groups (Jackson et al., 1976;Cousens and Alberts, 1982;Chambers and Ramsay Shaw, 1984;Allis et al., 1985), or one acetyl plus one phosphoryl group (Ruiz-Carrillo et al., 1975), is gradually demodified as the chromatin matures i n uiuo. Acetylation has been associated with nucleosome destabilization (Christensen and Dixon, 1982;Loidl and Grobner, 1987), and it is possible that the acetylation of newly assembled nucleosomes may contribute to their weakened binding to the DNA. It follows, then, that any transcription factors attempting to form active complexes on the nascent chromatin in growing somatic cells would be competing and interacting with modified nucleosomes, not with conventional nucleosomes.
T o study the possible role of acetylation during chromatin assembly, we have analyzed an in vitro system derived from Xenopus oocytes (Glikin et al., 1984;Shimamura et al., 1988) for qualitative changes in the state of the nucleosomes that occur as a function of chromatin assembly. Proteins that are assembled into chromatin on a circular DNA template (which is a simple model system that is topologically equivalent to the looped DNA domains found in genomic DNA in uiuo; Benyajati and Worcel, 1976;Paulson and Laemmli, 1977;Cockerill and Garrard, 1986) can be conveniently analyzed in this system. We document here the changes in histone H4 modification that occur during chromatin formation in uitro, and we investigate the effects of blocking histone demodification on the final chromatin structure attained.

MATERIALS AND METHODS
Mature female Xenopus laeois frogs (10.5 cm) were obtained from Nasco. Recently, Nasco has initiated a program whereby the frogs are starved during the summer. These starved frogs produce poor extracts, so we recommend that some frogs be set aside for regular feeding during these months.
The plasmid DNA pXbsF2Ol used in these experiments carries the 240-base pair Xenopus borealis somatic 5 S ribosomal RNA gene inserted into the HindIII-BamHI site of pUC 9 (Razvi et a/., 1983). This is the standard DNA plasmid used in our laboratory for chromatin assembly studies. The chromatin assembly reaction does not require specific DNA sequences and works as readily on all circular DNA plasmids tested. Sodium butyrate and DNase I were obtained from Sigma.
Chromatin Assembly Reaction-The oocyte S-150 extract was prepared as described (Shimamura et al., 1988) except that the levels of collagenase type I1 (Sigma) used to disperse the oocytes was reduced to 0.075% to minimize the effects of nonspecific proteases.
The chromatin assembly reactions (Shimamura et a/., 1988) were performed with the modifications described (Rodriguez-Campos et al., 1989). except that fresh additional phenylmethylsulfonyl fluoride (10 pg/ml), leupeptin (2 pg/ml), and pepstatin (2 pg/ml) were present in all reactions. When chromatin was assembled in the presence of sodium butyrate, an equal concentration of NaCl was added to the reactions performed without butyrate to control for any salt effects. When butyrate levels higher than 10 mM were utilized, the concentration of creatine phosphate (Boehringer Mannheim), which is supplied as a disodium salt, was decreased in order to avoid inhibition of the chromatin assembly reaction by salt. Histone H1 was isolated from X. laeois erythrocytes and added at the beginning of the chromatin assembly reactions as described (Rodriguez-Campos et al., 1989).
Isolation of the Minichromosomes-The minichromosomes were isolated through sucrose gradients essentially as described (Shimamura et al., 1988) except that the gradient buffers and the sucrose cushion used in the final pelleting step contained 10 mM butyrate and 5 mM @-glycerophosphate, and the concentration of NaCl was decreased to 80 mM to accommodate the additional salt. Addition of butyrate preserves the modified state of the histones by inhibiting any further deacetylation during the long isolation procedure.
Nuclease Digestion-The chromatin was digested with micrococcal nuclease as described (Shimamura et a/., 1988). The nucleosome repeat length was determined as follows. The size of the larger oligonucleosomes present a t early digestion times was determined using molecular weight markers. The length, in base pairs, of each oligonucleosome was then divided by the number of nucleosomes it contained. This approach minimizes the effect of exonucleolytic trimming by micrococcal nuclease (Noll and Kornberg, 1977).
Digestion with DNase I (0.45 unit/pl reaction) was performed in the presence of an additional 4 mM MgCI2 and 0.2 mM CaC12, and the samples were processed as described for micrococcal nuclease digestions.
Protein Analysis-The histone proteins were analyzed on Tritonacid-urea gels (0.75 mm thick) essentially as described (Shimamura et a/., 1988) except that a stacking gel (Spiker, 1980) was poured after the resolving gel had been prerun and scavenged with cysteamine. The stacking gel consisted of 7.5% acrylamide (with an acrylamide to bis ratio of 300.8), 6 M urea, 0.37% Triton X-100, and 0.375 M potassium acetate, pH 4.0. Photochemical polymerization of the stacking gel was accomplished with 0.0004% riboflavin and 1% N,N, N',N'-tetramethylethylenediamine. Acid-urea gel electrophoresis was performed as described above with the omission of Triton X-100 from the stacking and resolving gels. The gels were stained with silver nitrate after prestaining with Amido black (Mold et al., 1983).
To analyze the unfractionated chromatin assembly reactions on the acid-urea gels, an equal volume of 2-fold sample dye solution (0.4 M HCI, 0.5 mg/ml protamine sulfate, 8 M urea, and 0.02% pyronin Y) was added to the reactions, which were then loaded directly onto the gel.

Diacetylated Histone H4 Is Demodified as Chromatin Is
Assembled-Chromatin assembly in the S-150 extract, like the process occurring in vivo (Levy and Jakob, 1978;Worcel et al., 1978;Schlaeger and Knippers, 1979;Annunziato and Seale, 1982), involves a gradual change from a nascent chro-matin structure into mature chromatin containing regularly spaced nucleosomes (Glikin et al., 1984;Ruberti and Worcel, 1986;Shimamura et al., 1988). Although plasmid DNA added to the S-150 appears to be supercoiled on a one-dimensional gel after a 1-h incubation at 27 "C, digestion with micrococcal nuclease does not yet produce a distinct nucleosomal ladder. Chromatin assembly is not complete until 6 h of incubation a t 27 "C, at which time analysis on a two-dimensional gel reveals the accumulation of additional negative supercoils with a tighter distribution of DNA topoisomers, and digestion with micrococcal nuclease produces a distinct and regularly spaced ladder (see also Shimamura et dl., 1988).
T o analyze the state of the histone proteins over the course of chromatin assembly, minichromosomes were purified by sucrose gradient centrifugation after various times of assembly a t 27 "C, and the proteins bound to the DNA were visualized on an acid-urea gel. As shown in Fig  FIG. 1. Chromatin assembly is accompanied by deactylation of histone H4. a, relaxed 5 S DNA was incubated in the S-150 at histone H4 was well resolved from the other proteins in the extract. The acid-urea gel shows that histone H4 exists in the diacetylated form in the extract. Identification of the modified groups as acetates was made based on the following criteria: 1) incorporation of radioactive acetate into these bands following the addition of ['HH]acetyl-CoA with or without exogenously added histones to the extract; 2) no change in migration of histone H4 following treatment with bacterial alkaline phosphatase or calf intestinal phosphatase; and 3) failure to incorporate radioactive phosphate into histone H4 following incubation of the extract with [y-"P]ATP. Our findings are consistent with the report of Woodland (1979) that histone H4 in the oocyte nucleus is present in a diacetylated form.
The Modified State of Histones H2A, H2B, and H3 Does Not Change during Chromatin Assembly-Histones H2A and H3 are not well resolved on the acid-urea gel, but they are resolved on a Triton-acid-urea gel. As shown in Fig. 2, three histone H2A variants are present in the Xenopus oocyte extract (see also Dilworth et al., 1987;Shimamura et al., 1988). Only one of the variants, histone H2A2, displays the same mobility as the typical somatic histone H2A on Triton-acidurea gels and on SDS gels (see Fig. 2 in this paper and Shimamura et al., 1988).
No changes in the modification states of histones HZA, H2B, and H3 were apparent during chromatin formation in vitro, while the demodification of histone H4 was again observed (Fig. 2). Histone H3 is present as multiple bands on the Triton-acid-urea gel. The nature of these modified histone H3 species is not clear, and several histone H3 variants may be present (Dilworth et al., 1987). The question of the nature of these modified species is not addressed further here because no apparent change in histone H3 occurred as a function of chromatin assembly.
These results parallel the process observed in vivo in that no change in modification of histones H2A or H2B has been detected during chromatin replication in vivo. Whether or not histone H3 undergoes changes in modification during chromatin maturation appears to vary with the in vivo system Histone H4 is the only core histone to undergo changes in modification during chromatin assembly. Chromatin was assemhled for the indicated times at 2'i "C, and the minichromosomes were isolated through sucrose gradients. The proteins bound to the DNA were loaded onto Triton-acid-urea gels and visualized with silver. The various modified species of histone H4 are indicated. The intensities of the two unidentified bands migrating between histones H2R and H4 vary from gel to gel, and they are not present in acid-urea gels (Fig. 1). Lane 4 , Xenopus erythrocyte histones.

Deacetylation of Histone H4 Is Coupled to the Assembly of a Compact Chromatin
Structure-To rule out the possibility that the deacetylation reaction was not coupled to chromatin assembly but was merely a consequence of the prolonged incubation of the extract at 27 "C, chromatin was assembled in the presence of different levels of DNA, and the total, unfractionated S-150 proteins were analyzed on an acid-urea gel (as in the last lane in Fig. 1). As shown in the first lane of Fig. 3a, when the S-150 was incubated under the conditions used for chromatin assembly in the absence of DNA and then loaded directly onto an acid-urea gel, no histone H4 deacetylated occurred. As a positive control (see adjacent lanes in the same figure), the deacetylation of histone H4 is clearly apparent with this approach when DNA is added to reactions performed in parallel. Thus, the deacetylation reaction is coupled to chromatin assembly.
The efficiency of chromatin assembly in this system is greatly affected by the concentration of magnesium present in the reaction. At 27 "C, the assembly of regularly spaced nucleosomes is optimal at levels of 1-2 mM MgC12; at 10 mM MgC12, the DNA is only partially supercoiled and the nucleosomal ladder is disrupted (Rodriguez-Campos et al., 1989). Similarly, the deacetylation reaction occurred efficiently at low levels of MgC12 but was inhibited at high levels (compare the intensities of the H4aco band among the different lanes of Fig. 36). Thus, the inhibition of the deacetylation reaction by high levels of MgC12 parallels the inhibitory effect of MgC12 on the chromatin assembly reaction.
The optimal levels of MgCI2 for chromatin assembly at 37 "C is shifted from 1 mM M&12 to 4-5 mM MgC12 (Rodriguez-Campos et al., 1989). The deacetylation reaction at 37 "C again paralleled the formation of a periodic chromatin structure: H4 deacetylation was inefficient at both lower (0-1 nM) and higher (10 mM) levels of MgC12 and was optimal at 5 mM MgC12 (data not shown).
Examination of the total histone H4 present in the unfractionated extract following chromatin assembly reveals that much of the H4 remained diacetylated (Fig. 3). At the lower levels of input DNA (4-8 ng DNA/pl S-150, which is well below the capacity of the extract to supercoil DNA), the histone H4 which had been assembled onto the minimicrosome was deacetylated in 6 h (Fig. 1); therefore, the diacety- lated histone H4 remaining in the extract was not bound to the DNA. Increasing the levels of input DNA, however, failed to deacetylate correspondingly greater proportions of histone H4 (Fig. 3a). T o distinguish whether the deacetylation reaction proceeded inefficiently a t high levels of DNA, or only a subset of H4 molecules were binding to the DNA and becoming fully deacetylated, the proteins bound to the minichromosomes a t high levels of input DNA were purified through sucrose gradients and visualized on an acid-urea gel (Fig. 4b).
The data indicate that histone H4 did bind to the DNA but failed to undergo efficient deacetylation. Note that in a parallel control reaction performed with the standard concentration of DNA used for chromatin assembly (5 ng of DNA/pl S-150), histone H4 again became deacetylated.
At these high DNA input levels, the DNA did not become fully supercoiled in the S-150 (Fig. 4a; compare with the supercoiling obtained at the standard DNA concentration used for chromatin assembly in Figs. la and 6 4 . Digestion with micrococcal nuclease produced a diffuse distribution of DNA fragments which were rapidly degraded to small subnucleosomal lengths (Fig. 4c; compare with the micrococcal nuclease ladder produced at the standard DNA concentration used for chromatin assembly in Fig. 6b). A particle the size of a monomer accumulated at longer times of digestion. These patterns indicate that while some nucleosomes had been formed, they were not regularly spaced and much of the DNA remained free. Under these conditions of DNA excess, failure to generate a compact chromatin structure is accompanied by inefficient deacetylation.
Butyrate Blocks the DNA-dependent Histone H4 Deucetylation in the Oocyte S-150-Butyrate is an inhibitor of histone deacetylases Candido et al., 1978;Vidali et al., Cousens et al., 1979). When butyrate was added at the onset of the chromatin assembly reaction, histone H4 failed to undergo deacetylation (Fig. 5a, lanes 9 and 10). As a positive control for the deacetylation reaction, parallel reactions were performed in the absence of butyrate; again, monoacetylated and deacetylated histone H4 could be readily detected (Fig.  513, lanes 3 and 4; see also Fig. 3a). Analysis of the gradientpurified minichromosomes revealed that the diacetylated histone H4 was bound to the DNA (Fig. 5b, lane 3). The other core histones were also assembled onto the DNA, but no apparent changes in their modification were seen.
The addition of 5 mM butyrate for 30 min after chromatin had been formed, however, did not lead to any significant reacetylation of the demodified nucleosomal histone H4 (Fig.  5a, lanes 5-8). Raising the concentration of butyrate to levels as high as 50 mM for up to 1 h after chromatin had been assembled yielded the same results. No hyperacetylation of histone H4 in the presence of butyrate could be detected even a t early times of chromatin assembly (data not shown).

Butyrate Inhibition of Histone H4 Deacetylation Does Not Affect the Formation of Regularly Spaced Nucleosomes or the
Binding of Histone HI-The structure of minichromosomes containing fully diacetylated histone H4 was next analyzed and compared to that of control minichromosomes assembled in the absence of butyrate. The rate and extent of DNA supercoiling was the same in the presence and absence of butyrate (Fig. 6a). Micrococcal nuclease digestion of the minichromosomes containing diacetylated histone H4 revealed regularly spaced nucleosomes with a periodicity identical to the one seen in control minichromosomes (Fig. 6b, left two  panels). Upon mild digestion with micrococcal nuclease, 17 nucleosomes with a periodicity of 170 base pairs per nucleosome, could be counted on this 2.9-kilobase pair DNA plasmid, both in the presence and absence of butyrate. Furthermore, the diacetylated and deacetylated minichromosomes were digested at the same rate by micrococcal nuclease. These findings were not surprising since studies in viuo have also reported no change in the micrococcal nuclease digestion pattern of chromatin following the addition of butyrate (Mathis et al., 1978;Nelson et al., 1978;Simpson, 1978;Annuziato and Seale, 1983).
Studies in vivo have reported that chromatin becomes more sensitive to digestion with DNase I following treatment with butyrate (Mathis et al., 1978;Nelson et al., 1978;Simpson, 1978;Vidali et al., 1978;Annunziato and Seale, 1983). The diacetylated minichromosomes assembled in the presence of butyrate in uitro, however, exhibited the same sensitivity to DNase I as the deacetylated minichromosomes (Fig. 6c, left  two panels). Since  histone H1, and this, in turn, could render the chromatin more sensitive to digestion by DNase I.
To test for a possible effect of butyrate on the structure of histone HZ-containing chromatin, we assembled minichromosomes in the presence of exogenously added histone H1. This protein is not found in the minichromosomes (Shimamura et al., 1988) unless exogenous H1 is added to the extract (Rodriguez-Campos et al., 1989). By varying the amounts of H1 added to the extract as well as the times at which H1 is added to the assembly reaction, the nucleosomal repeat length of the assembled minichromosomes can be increased from 170-180 base pairs to as much as 220 base pairs (Rodriguez-Campos et al., 1989).
To determine whether the binding of histone H1 was affected by the acetylation state of histone H4, histone H1 was added to the oocyte extract, and the chromatin assembly relaxed 5 S DNA was assembled into chromatin for the indicated times a t 27 "C in the presence or absence of 50 mM butyrate. The deproteinized DNA was analyzed on a 1% agarose gel. b, relaxed 5 S DNA was assembled into chromatin for 6 h a t 27 "C in the presence or absence of 10 mM butyrate. Histone H1 (at a ratio of 1.5 molecules of H1 to 190 base pairs DNA) was added at the beginning of the assembly reactions in the two right-hand panels. The reactions were digested with micrococcal nuclease (0.1 unit/pl reaction) for 0, 1, 2, 4, and 8 min. The DNA was analyzed on a 1.5% agarose gel. c, relaxed 5 S DNA was assembled into chromatin as described in b. DNase I (0.45 unit/pl reaction) was added for 0, 1, 2, 4, and 8 min, and the reaction products were analyzed on a 1.5% agarose gel. A 123-base pair ladder (Bethesda Research Laboratories, Inc.) is used as molecular weight markers flanking each nuclease digestion panel. The size of the nucleosome oligomers ( n ) was determined at the earliest time of digestion with nuclease to minimize effects of exonucleolytic trimming (Noll and Kornberg, 1977).
reaction was performed in the presence and absence of butyrate. After assembly was complete, the chromatin structure was examined by nuclease digestions as before. Analysis of the proteins in acid-urea gels confirmed that histone H4 deacetylation still occurred when histone H1 was assembled into the minichromosomes (data not shown). As shown in Fig. 6b, histone H1 induced a longer nucleosomal repeat length. In this case, the nucleosomal repeat was changed from 170 to 200 base pairs by histone H1, as expected (Rodriguez-Campos et al., 1989). Note that at early digestion times, the pentanucleosome-size DNA fragments migrate with the 861base pair marker (7-mer) in the absence of H1 (Fig. 6b, left panels), whereas they migrate above the 984-base pair marker (8-mer) when H1 is present (Fig. 66, right panels).
The binding of histone H1 to the minichromosomes assembled i n vitro can also be ascertained by digestion with DNase I. This enzyme produces a diffuse pattern of DNA fragments when it digests histone H1-containing chromatin. On the other hand, in the absence of histone H1, DNase I preferentially cleaves the exposed linker DNA to produce a periodic pattern resembling the one generated by micrococcal nuclease (Worcel et al., 1983). DNase I digestions of minichromosomes generate a distinct nucleosomal ladder when histone H1 is absent (Fig. 6c, leftpanels; see also Ruberti and Worcel, 1986), but when histone H1 is present, a more diffuse pattern of DNA fragments is produced (Fig. 6c, right panels).
The patterns produced by nuclease digestions and the susceptibility to the nucleases were the same in the presence or absence of butyrate in all cases examined (Fig. 6, b and c). We therefore conclude that the persistence of the diacetylated state of histone H4 does not have a major effect on the structure of these minichromosomes. We also conclude that deacetylation of histone H4 is not required for the binding of histone H1 to chromatin.

DISCUSSION
We report that chromatin assembly in the Xenopus oocyte S-150 is accompanied by the deacetylation of histone H4. The deacetylation reaction fails to occur in the absence of DNA. Conditions which disrupt the chromatin assembly reaction also inhibit the deacetylation reaction. The deacetylation reaction, however, is not required for the assembly of regularly spaced nucleosomes.
Relationship to Chromatin Assembly in Viuo-The observed qualitative change in histone H4, which goes from a diacetylated to an unmodified form, is the same as that observed during chromatin assembly i n vivo in systems as diverse as mammalian hepatoma cells (Jackson et al., 1976), sea urchin (Chambers andRamsay Shaw, 1984), and Tetrahymena (Allis et al., 1985). We have not detected changes in modification of the other core histones. Data obtained in vivo also report no changes in histones H2A and H2B. The i n vivo data on histone H3 is variable. No change in histone H3 was observed in duck erythroid cells (Ruiz-Carrillo et al., 1975), although some deacetylation was noted in mammalian hepatoma cells (Jackson et al., 1976) and in Tetrahymena (Allis et al., 1985).
Histone H4 Acetylation and Nuclease Sensitiuity-The chromatin assembled in vitro by the S-150 in the presence of butyrate contains diacetylated histone H4 and exhibits a regularly spaced nucleosomal ladder with micrococcal nuclease. No change in DNase I sensitivity is observed in this system following assembly in the presence of butyrate. Addition of histone H1, which leads to an increase in the nucleosomal repeat length (Rodriguez-Campos et al., 1989), also fails to elicit any difference in DNase I sensitivity following treatment with butyrate in this system. Inhibition of the deacetylation reaction by butyrate during chromatin assembly i n vivo similarly does not affect the formation of a regularly spaced ladder with micrococcal nuclease, but it does prevent the subsequent resistance to DNase I which normally develops as the chromatin matures (Annunziato and Seale, 1983). Many other studies in vivo have also reported that butyrate causes no change in the micrococcal nuclease ladder but leads to a marked increase in the susceptibility to digestion by DNase I (Mathis et al., 1978;Nelson et al., 1978;Simpson, 1978;Vidali et al., 1978). One explanation for the different results obtained with DNase I in vivo versus i n uitro may be that most of the in vivo studies with butyrate induced hyperacetylation of histone H4 to the triacetylated and tetraacetylated species (Mathis et al., 1978;Nelson et al., 1978;Simpson, 1978;Vidali et al., 1978), while the oocyte S-150 contains only the diacetylated form of histone H4. Annunziato and Seale (1983) sought to minimize the formation of such hyperacetylated species by using very short times of exposure to butyrate; the increased sensitivity of acetylated chromatin to DNase I in this case was proposed to result from changes in higher order structure. Hyperacetylation of the histones i n vivo may affect the condensation of chromatin into the 30-nm fiber (Annunziato et al., 1988). The inability of the small DNA plasmid used in these studies to form any significant higher order structures would preclude the detection of any effects of butyrate on the folding of the chromatin fiber.
Regulation of Histone H4 Deacetylation-The mechanisms regulating the deacetylation reaction which accompanies chromatin assembly remain an open question. The addition of butyrate to the deacetylated nucleosomes fails to reacetylate histone H4. While it is possible that the endogenous acetylase in the S-150 might have lost activity over the time of incubation required for chromatin assembly, it is also possible that the activity of histone acetylases may be regulated by the chromatin structure (Cousens et al., 1979;Covault and Chalkley, 1980;Cousens and Alberts, 1982). Thus, free histone H4 may be more accessible to the acetylase, while nucleosomal histone H4 may be sequestered away from the enzyme. A cytoplasmic acetylase which recognizes free H4 but not nucleosomal H4 has been described (Garcea and Alberts, 1980;Sures and Gallwitz, 1980;Wiegand and Brutlag, 1981). Another possibility is that the nucleosome undergoes a conformational change following the deacetylation of histone H4 such that it is no longer recognized by the acetylase enzyme. A change in nucleosomal conformation resulting from the acetylation of histone H4 has been reported (Muller et al., 1982). Experiments using the isolated acetylase and deacetylase enzymes to study chromatin assembly i n vitro may help answer these questions.
Possible Role of the Diacetylated Histone H4 in Nascent Chromatin-We find that the deacetylation reaction of the DNA-bound histone H4, which accompanies chromatin assembly in the S-150, is not necessary for the formation of a regularly spaced nucleosomal ladder.
The interesting alternative question, whether the initial diacetylated state of histone H4 is required for proper chromatin assembly, cannot yet be addressed as there is no satisfactory procedure for deacetylating the histones present in the S-150 prior to chromatin assembly. The initial diacetylated state of histone H4 may be important for proper chromatin formation, possibly by transiently destabilizing the nascent nucleosome during the process of nucleoprotein assembly.
The oocyte S-150 is notable in that it allows exogenously added histone H1 to increase the nucleosomal repeat length on any circular DNA template, independent of its sequence. Previous attempts to assemble nucleosomes with histone H1 using dialysis from 2 M NaCl (Fulmer and Fasman, 1979) or H3-H4 tetramers and H2A-H2B dimers (Ruiz-Carrillo et al., 1979) have failed to establish these longer physiological spacings. The assembly of nucleosomes with H1 or H5 onto chicken erythrocyte DNA using polyglutamic acid led to native spacing at the level of dinucleosomes (Kunzler and Stein, 1983); however, physiological spacing beyond the level of the dimer could only be achieved on a poly[d

Nucleosomes on poly[d(A-T)]-poly[d(A-T)]
DNA are apparently destabilized by polyglutamic acid, and, under these conditions, they can slide apart in the presence of histones H1 or H5. The acetylation of histone H4 may exert a similar effect during chromatin assembly in the S-150 by destabilizing the nucleosomes and thereby allowing histone H1 to establish the longer nucleosomal periodicities found in uiuo (Rodriguez-Campos et al., 1989).
The modification state of histone H4 may facilitate the binding of transcription factors to DNA by destabilizing the nucleosomes. The observation that acetylated histone H4 in S-phase chromatin from Physarum polycephalun is displaced by lower concentrations of protamines than the unmodified histone H4 supports such a view (Loidl and Grobner, 1987). Curiously, this preferential release of acetylated histone H4 by protamines is not seen when histone H4 is artificially acetylated with sodium butyrate (Loidl and Grobner, 1987). Thus, the role of acetylation may be more complex than the nonspecific neutralization of highly positive charged residues. A model has been proposed (Loidl, 1988) whereby specific combinations of sites of acetylation may serve as signals for different processes which require the destabilization of nucleosomes. In support of such a model, sites of acetylation have been reported to differ between transcriptionally active chromatin and replicating chromatin (Chicoine et al., 1986).
The nascent chromatin that forms in vitro when DNA is added to the S-150 carries the naturally occurring diacetylated form of histone H4. Further studies on the nascent chromatin may illuminate both the mechanism of nucleosome maturation and the special structural and functional properties of this chromatin.