We demonstrate using a dinucleosome template that acetylation of the core histones enhances transcription by RNA polymerase III. This effect is not dependent on an increased mobility of the core histone octamer with respect to DNA sequence. When linker histone is subsequently bound, we find both a reduction in nucleosome mobility and a repression of transcription. These effects of linker histone binding are independent of core histone acetylation, indicating that core histone acetylation does not prevent linker histone binding and the concomitant transcriptional repression. These studies are complemented by the use of a Xenopus egg extract competent both for chromatin assembly on replicating DNA and for RNA polymerase III transcription. Incorporation of acetylated histones and lack of linker histones together facilitate transcription by >10‐fold in this system; however, they have little independent effect on transcription. Thus core histone acetylation significantly facilitates transcription, but this effect is inhibited by the assembly of linker histones into chromatin.
Histones and the chromatin structures they assemble have an active role in the transcription process (Felsenfeld, 1992; Grunstein et al., 1992; Wolffe, 1995). Nucleosomes have been shown in vitro to both repress (Lorch et al., 1987) and activate transcription (Schild et al., 1993). In vivo experiments also indicate that chromatin components have a dual functional role in both activating and repressing gene activity (Kayne et al., 1988; Durrin et al., 1991; Fisher‐Adams and Grunstein, 1995; Shen and Gorovsky, 1996). The exact role of chromatin structure in regulating any particular promoter in a natural context remains to be completely defined (Svaren and Horz, 1995, 1996).
In vitro experiments with model chromatin templates have proven useful in defining how the core histones and linker histones might restrict trans‐acting factor access to nucleosomal DNA (Hayes and Wolffe, 1992a; Owen‐Hughes and Workman, 1994). Certain trans‐acting factors such as the glucocorticoid and thyroid hormone receptors bind efficiently to nucleosomal DNA (Perlmann and Wrange, 1988; Truss et al., 1995; Wong et al., 1995); others such as TFIIIA, Gal4 or TBP do not (Hayes and Wolffe, 1992a, b; Imbalzano et al., 1994; Vettese‐Dadey et al., 1994; Godde et al., 1995). Transcription factor association with nucleosomal DNA can be facilitated by removal of histone H1 (Juan et al., 1994; Ura et al., 1995), removal of core histones H2A/H2B (Hayes and Wolffe, 1992b; Chen et al., 1994), proteolysis of the N‐termini of the core histones (Lee et al., 1993; Vettese‐Dadey et al., 1994; Godde et al., 1995) or acetylation of the N‐termini of the core histones (Lee et al., 1993; Vettese‐Dadey et al., 1996). Of these various modifications to nucleosomal structure, acetylation is currently of most interest since the identification of transcription factors that can act as histone acetyltransferases (Brownell et al., 1996; Mizzen et al., 1996; Ogryzko et al., 1996; Yang et al., 1996).
High levels of histone acetylation are correlated with gene activity (Hebbes et al., 1988, 1992, 1994; Tazi and Bird, 1990; Bartsch et al., 1996) and reduced levels with gene silencing (Braunstein et al., 1993). The patterns of acetylation on individual histones are complex (Turner and O‘Neill, 1995), and exceptions exist to the general association of histone acetylation with transcription (O'Neill and Turner, 1995). Nevertheless, the fact that transcription factors can direct acetylation of the histones suggests that modification of the nucleosome may have a causal role in facilitating transcription. Newly synthesized histones H3 and H4 are acetylated (Ruiz‐Carrillo et al., 1975) and deacetylated shortly after their incorporation into the nascent chromatin assembled immediately following DNA replication (Jackson et al., 1976). Newly synthesized H4 is invariably diacetylated at lysines 5 and 12 (Sobel et al., 1995). It is in this form that H4 associates with chromatin assembly factor 1 (CAF1) (Verreault et al., 1996). It has been proposed that the incorporation of acetylated histones into nascent chromatin might facilitate transcription factor access to DNA immediately following replication (Lee et al., 1993; Verreault et al., 1996).
We have approached the role of histone acetylation in the transcription of chromatin using a model dinucleosomal template reconstituted using purified histones (Ura et al., 1995, 1996). This model template enables the characterization of both chromatin structure and transcriptional activity. In earlier work, we found that physiologically spaced histone octamers are mobile with respect to DNA sequence and repress transcription to 15‐30% of the level of naked DNA (Ura et al., 1995). Addition of histone H1 at a level of one molecule per nucleosome restricts nucleosome mobility and further represses transcription (Ura et al., 1995). Here we report that acetylation of the core histones significantly relieves transcriptional repression without influencing the mobility of histone octamers. Histone acetylation does not influence the repression of transcription by histone H1. These results are extended using a distinct experimental approach in which the acetylated histones are assembled into chromatin in a replication‐coupled in vitro system (Almouzni and Méchali, 1988a, b; Almouzni et al., 1991). We also find in this system that histone acetylation and linker histone deficiency significantly augment transcription.
Histone acetylation facilitates transcription of dinucleosomal templates containing only histone octamers
We reconstituted a 424 bp template containing two Xenopus somatic 5S RNA genes with histone octamers (Ura et al., 1995). We made use of the strong nucleosome positioning signals in the 5S RNA gene to separate nucleosomes (Simpson, 1991) and the capacity to have very efficient in vitro transcription of these genes as short linear DNA fragments in vitro (Wolffe et al., 1986). Histone octamers were reconstituted onto radiolabeled DNA fragments by exchange from HeLa cell chromatin enriched in acetylated core histones or from chromatin with minimal levels of histone acetylation (see Figure 1C, Ura et al., 1994). The reconstituted chromatin was fractionated on a sucrose gradient and each fraction was analyzed by nucleoprotein gel electrophoresis. Free DNA was resolved from mono‐, di‐ and trinucleosomal complexes (data not shown, Ura et al., 1995). Transcription of naked DNA, mono‐ and dinucleosome templates reconstituted with histones having minimal levels of acetylation, in extracts of Xenopus oocyte nuclei under efficient conditions (Birkenmeier et al., 1978; Wolffe et al., 1986), reveals a progressive reduction in transcriptional activity concomitant with the increase in the number of histone octamers bound (Figure 1A). Our results assay three predominant types of transcripts (Wolffe et al., 1986; Hansen and Wolffe, 1992, data not shown): there are transcripts that initiate at the start site of the 5S rRNA gene and terminate at the end of the gene (experimental 5S rRNA transcript), other transcripts initiate at the start site of the 5S rRNA gene and fail to terminate at the end of the gene continuing to the end of the DNA fragment (read‐through transcripts), and transcripts that initiate at the end of the DNA fragment and continue the entire length of the fragment (end‐initiated transcript). These transcripts provide information concerning both the access of the transcriptional machinery to the regulatory elements of the 5S rRNA gene (experimental 5S rRNA transcript) and the processivity of RNA polymerase III through chromatin templates (read‐through transcripts and end‐initiated transcripts). In all of our experiments, we find that all of the transcripts change in abundance to very similar extents; this is indicative of a general change in the access of the transcriptional machinery to DNA in response to changes in chromatin composition. We also include a naked internal control plasmid template which encodes a maxi 5S rRNA gene in our transcription reactions (maxi‐gene internal control, Bogenhagen et al., 1982); this template demonstrates that the repressive effects of chromatin are specific to the nucleosomal template. Finally, we can assay the recovery of nucleic acid from the transcriptional reaction from the abundance of the radiolabeled DNA (DNA) that serves as the template for transcription. Quantitation of the transcriptional repression due to reconstitution of dinucleosomes with minimally acetylated core histones (using a Phosphoimager, data not shown) indicates an 85% reduction in transcriptional activity (Figure 1A, compare lanes 1 and 3).
Our next experiments examined the differences between the transcriptional activity of mononucleosomes and dinucleosomes reconstituted with core histones having different levels of acetylation (Figure 1B). We compared transcriptional activity in reactions to which no histone deacetylase inhibitor was added (Figure 1B, lanes 1‐4), or to which the deacetylase inhibitors sodium butyrate (lanes 5‐8) or trichostatin A (lanes 11‐12, Yoshida et al., 1995) were added. In each instance the results were very similar. Histone acetylation was without significant consequence for transcription of mononucleosomal templates; however, high levels of histone acetylation facilitate the transcription of dinucleosomal templates 5‐ to 8‐fold (Figure 1B brackets, compare lanes 3 and 4, lanes 7 and 8, and lanes 11 and 12). Thus acetylation of core histones facilitates the transcription of dinucleosomal templates containing only histone octamers.
The histones on the dinucleosomal templates purified on the sucrose gradients are shown in Figure 1C. The octamers enriched in acetylated core histones contain histone H4 with a distribution of acetylated isoforms containing predominantly two, three and four acetylated lysines (lane 1). Use of antibodies against acetylated lysine 16 (H4. Ac 16, Turner et al., 1992) indicates reactivity with all these H4 isoforms, as anticipated from earlier work (S.Dimitrov, data not shown; Turner and O'Neill, 1995). In contrast, the dinucleosomal templates assembled using histones containing minimal levels of histone acetylation contained core histones in which no significant levels of acetylated H4 isoforms were present (Figure 1C, lane 2) and that did not react with antibodies against acetylated Lys16 (S.Dimitrov, data not shown). We examined whether histone acetylation was maintained through the time course of the experiment (Figure 1D). The relative abundance of acetylated isoforms of histone H4 is maintained during the transcription reaction in the presence of butyrate (compare lanes 1 and 2) or trichostatin A (compare lanes 1 and 3). In the absence of histone deacetylase inhibitors there is a partial loss of acetylated isoforms which decline from an average of three to that of two acetylated lysines per histone H4 (data not shown). The core histones also remain intact during the transcription reaction (Figure 1D, lanes 4 and 5). Having established that the histones and their acetylation state were maintained under the conditions of transcription, we next wished to determine what structural properties of dinucleosomal template might change in response to acetylation and contribute to facilitating transcription.
The influence of histone acetylation on nucleosome mobility and position
Nucleosome mobility has been suggested to have a major role in allowing access of the transcriptional machinery to regulatory DNA (Meersseman et al., 1992; Becker, 1994; Ura et al., 1995; Varga‐Weisz et al., 1995). We examined the influence of histone acetylation on the mobility of histone octamers relative to DNA sequence. To assess nucleosome mobility, we made use of electrophoresis on non‐denaturing polyacrylamide gels (Meersseman et al., 1992; Pennings et al., 1994). Resolution of nucleoprotein complexes on polyacrylamide gels is sensitive to the conformation of the histone‐DNA complex (Meersseman et al., 1992). Thus a single histone octamer associated with a 424 bp DNA fragment can be resolved into multiple complexes on a single dimension of electrophoresis dependent on the translational position of the histone octamer along the DNA fragment. Evidence for the mobility of a histone octamer comes from carrying out a second dimension of electrophoresis (Meersseman et al., 1992). If the octamer changes position during a 1 h incubation at 4 or 37°C prior to the second dimension, then this will be detected by the appearance of a nucleoprotein complex that migrates at a position away from a simple diagonal. Nucleosome mobility is temperature dependent: using a 424 bp DNA fragment reconstituted with histone octamers containing high levels of acetylation (Figure 2A) or minimal levels of acetylation (Figure 2B), more nucleoprotein complexes migrate off the diagonal at 37°C than at 4°C. Acetylated histone octamers do not show more mobility than control octamers. Two‐dimensional analysis of dinucleosomal templates reconstituted with two histone octamers again reveals comparable levels of nucleosome mobility with acetylated (Figure 2C) and control octamers (Figure 2D). We conclude that histone acetylation does not have a major influence on the mobility of histone octamers with respect to DNA sequence.
We next made use of an independent nuclease mapping methodology (Meersseman et al., 1991) to investigate nucleosome mobility through the associated variation in nucleosome position. Micrococcal nuclease digestion of acetylated and control nucleosomal complexes results in the accumulation of core particle size DNA (core, 146 bp) as a kinetic intermediate in digestion (Figure 3). Both acetylated (lanes 1‐4) and control mononucleosome reconstitutes (lanes 6‐9) show comparable digestion kinetics, as do acetylated (lanes 11‐14) and control dinucleosomes (lanes 15‐18). These digestions were followed by restriction endonuclease digestion of the core particle sized DNA. This allows the mapping of the boundaries of the histone octamer relative to the 5S rRNA gene sequences on the dinucleosome template. A single histone octamer adopts a preferred location with boundaries at −68 and +79 relative to the transcription start site (+1) of either 5S rRNA gene independent of acetylation status (Figure 4A, lanes 1‐11, dots show the 5′ and 3′ boundaries of the histone octamer, Figure 4B). When two histone octamers are present, this preference is less apparent (Figure 4A, lanes 12‐22). We find that the 5′ histone octamer adopts at least five distinct translational positions independent of histone acetylation (Figure 4A, compare lanes 13‐16 with 18‐21, dots show the 5′ boundaries of the octamer). These multiple translational positions indicate nucleosome mobility (Meersseman et al., 1991, 1992). A unique restriction site (Fnu4HI, Figure 4B, marked by an asterisk) allows discrimination between the boundaries of DNA contact made by the 5′ and 3′ histone octamers (Figure 4A, lanes 6, 11, 16 and 21). The 3′ histone octamer has fewer positions and, therefore, appears less mobile than the 5′ octamer (Ura et al., 1995). There is also a fraction of DNA that remains resistant to cleavage by EcoRV (Figure 4A, lanes 4, 9, 14 and 19). Cleavage by HgiAI in addition to EcoRV (lanes 5, 10, 15 and 20) indicates that most of this core particle DNA is present within a population of nucleosomes that have the 5′ end of the DNA fragment as their 5′ boundary. The proportion of nucleosomes in this fraction is also unchanged by histone acetylation. Thus we conclude that alterations in nucleosome positioning or the stability with which nucleosomes maintain their position with respect to DNA sequence do not occur in response to acetylation of the core histones.
Linker histone binding restricts nucleosome mobility and represses transcription independently of core histone acetylation
Acetylation of the core histones correlates with a deficiency of linker histones in newly assembled chromatin (Perry and Annunziato, 1989, 1991). It has also been suggested that core histone acetylation interferes with the association of linker histones with mononucleosomes (Juan et al., 1994). In contrast to these results, we have found that the chromatin of Xenopus embryonic nuclei in which core histones are maintained in a hyperacetylated state incorporate normal stoichiometries of histone H1 (Dimitrov et al., 1993). Moreover, mononucleosomes assembled using acetylated core histones and the Xenopus 5S rRNA gene also incorporate histone H5 without impediment (Ura et al., 1994). However, since dinucleosomes offer the opportunity to examine histone H1‐dependent transcriptional repression (Ura et al., 1995, 1996), and since histone H1 binds with different affinities to mononucleosome compared with dinucleosome substrates (Hayes et al., 1996; Nightingale et al., 1996a,b; Ura et al., 1996), we next explored whether linker histones would stabilize acetylated histone octamers comparably to controls and whether transcription would be repressed with similar efficiency.
We find that histone H1 binds with equivalent efficiency to dinucleosomal templates reconstituted with acetylated (Figure 5A, lanes 1‐6) or control histone octamers (Figure 5A, lanes 7‐12). These nucleoprotein complexes also aggregate with equivalent efficiency (Figure 5A, lanes 6 and 12). Control experiments indicated that chromatosome (166 bp) intermediates in micrococcal nuclease digestion accumulated in the presence of histone H1 independently of core histone acetylation state (data not shown; Ura et al., 1994). We next examined the consequences of incorporating histone H1 for transcription of the dinucleosomal templates.
In earlier work, we had established that the incorporation of histone H1 into a dinucleosomal template would restrict transcription of the 5S rRNA genes (Ura et al., 1995). We repeated these experiments using dinucleosomes containing acetylated or control core histones, incorporating trichostatin into the reaction mixture to impede deacetylation of the core histones (see Figure 1). We find that histone H1 effectively represses transcription of dinucleosomal templates to comparable levels, independently of histone acetylation state (Figure 5B, compare lanes 2 and 3 with 5 and 6). The repression of transcription dependent on histone H1 has been attributed to the restriction in the mobility of histone octamers relative to the 5S rRNA gene (Ura et al., 1995). We next examined the mobility of histone octamers on the dinucleosomal template reconstituted with acetylated or control core histones in the presence or absence of histone H1. Histone H1 restricted nucleosome mobility at 37°C independently of core histone acetylation (data not shown). We conclude that the correlation between the restriction in nucleosome mobility and transcriptional repression is maintained and that histone H1 retains the capacity to repress transcription from acetylated chromatin.
In earlier work, we and others had observed that histone acetylation had no influence on the rotational positioning of DNA on the surface of the histone octamer (Norton et al., 1990; Bauer et al., 1994). We had also found that incorporation of linker histones into mono‐ or dinucleosomal templates had no influence on the rotational positioning of DNA on the surface of the histone octamer(s) (Hayes et al., 1993; Ura et al., 1995). Since the emphasis of the experiments reported here is on constraints on the mobility of the histone octamer with respect to DNA sequence, we wished to establish whether the DNA might adopt different rotational positions in the dinucleosome dependent on histone acetylation and/or histone H1. We find that there is no change in the rotational positioning of DNA with respect to the histone surface dependent on either acetylation or linker histones, as detected by hydroxyl radical cleavage (data not shown).
Diacetylation of histone H4 and linker histone deficiency together facilitate the assembly of transcriptionally competent chromatin during replication
Our experiments using model dinucleosomal templates demonstrate that acetylation of the core histones can facilitate transcription in the absence of histone H1. These studies make use of an experimental strategy requiring the prior assembly of chromatin of defined composition (Figure 1C) before addition to a transcription extract enriched in transcription factors for RNA polymerase III and deficient in chromatin assembly itself (Birkenmeier et al., 1978; Wolffe et al., 1986, 1987). Thus the facilitation of transcription attributed to acetylation of the core histones occurs after assembly of the histone octamer is complete. This might reflect the consequences of the potentially targeted acetylation of the core histones conferred by transcription factors (Brownell et al., 1996; Mizzen et al., 1996; Ogryzko et al., 1996; Yang et al., 1996). However, the consequences for transcription might be very different when acetylated core histones are incorporated into nascent chromatin following DNA replication (Ruiz‐Carrillo et al., 1975; Verreault et al., 1996). To investigate this issue, we made use of a physiological chromatin assembly system. Xenopus egg extracts will assemble chromatin in a replication‐coupled reaction during second strand synthesis on single‐stranded DNA templates (Almouzni and Méchali, 1988a, b; Almouzni et al., 1990a). This system has proven useful in determining the role of various core histones in directing transcriptional repression during the staged assembly of chromatin (Almouzni et al., 1990b, 1991). Histone H4 is stored in a diacetylated form in Xenopus eggs (Figure 6A, panel 1, Shimamura and Worcel, 1988; Dimitrov et al., 1994). In agreement with predictions (Sobel et al., 1995), the use of antibodies specific for acetylated H4 isoforms indicates that lysines 5 and 12 are acetylated in this storage form (S.Dimitrov, data not shown; Turner and Fellows, 1989). Following incorporation into chromatin during embryogenesis, histone H4 is progressively deacetylated (Figure 6A, panels 3 and 4). This deacetylation can be inhibited with sodium butyrate (Figure 6A, panel 2). The Xenopus egg extract does not contain histone H1, but does contain an oocyte‐specific linker histone variant, histone B4 (Dimitrov et al., 1994; Nightingale et al., 1996a). Immunoblotting (Figure 6B) indicates that maintenance of histone diacetylation does not influence the incorporation of histone B4 into chromatin. Thus we can explore the role of histone H4 diacetylation and linker histone B4 in the repression of 5S rRNA gene transcription.
We depleted histone B4 from Xenopus egg extract using established protocols (Figure 6C, Dasso et al., 1994; Dimitrov et al., 1994) and examined the assembly of nucleosomes in the presence or absence of sodium butyrate (Figure 6D). We find that nucleosomes are assembled with comparable efficiency (data not shown) and comparable resistance to micrococcal nuclease independently of sodium butyrate (Figure 6D, compare lanes 1‐6 with 7‐14). There is a slight increase in nuclease sensitivity following depletion of histone B4 (Figure 6D, compare lanes 1‐3 and 7‐10 with 4‐6 and 11‐14). Our next experiments examined the consequences for the transcription of two class III genes: satellite I DNA (Wolffe, 1989) and a 5S rRNA gene, that are assembled into chromatin under these various conditions on a replicating template. As a control, we made use of double‐stranded DNA templates that are assembled into chromatin inefficiently and that remain transcriptionally competent under these reaction conditions (Almouzni et al., 1990b, 1991). We find that the depletion of histone B4 from the extract leads to a modest 2‐fold stimulation of transcription (Figure 6E, compare lanes 3 and 4), as does the addition of sodium butyrate (Figure 6E, compare lanes 3 and 5). However, when the extract is both depleted of histone B4 and maintained in the presence of sodium butyrate, there is a >10‐fold augmentation of transcription (Figure 6E, compare lanes 3 and 6). Transcription levels become similar to those of a comparable mass of double‐stranded DNA that is only inefficiently assembled into chromatin in this system (Figure 6E, compare lanes 1 and 6). Thus linker histone deficiency and the maintenance of histone acetylation states also contribute to facilitating transcription from chromatin assembled under these conditions (see also Figures 1 and 5).
The major conclusion from this work is that acetylation of the core histones facilitates transcription of chromatin templates in the absence of linker histones (Figures 1, 5 and 6). Incorporation of linker histones into chromatin represses transcription independently of core histone acetylation (Figures 5 and 6). We have also excluded two molecular mechanisms by which acetylation of the core histones might facilitate transcription. Neither nucleosome mobility nor nucleosome positioning change as a consequence of histone acetylation (Figures 2,3,4). We suggest that local changes in the stability with which the N‐terminal tails of the core histones contact DNA in the nucleosome dependent on their acetylation (Hong et al., 1993) will have the major influence on transcription within these model templates.
Histone acetylation facilitates transcription
Histone acetylation has long been correlated with transcriptional activity (Allfrey et al., 1964; Hebbes et al., 1988; reviewed by Turner, 1991, 1993). An important and unresolved issue has been whether the acetylation of core histones could have a causal influence on transcription, or whether the accumulation of acetylated histones in chromatin was a consequence of transcriptional activity. Three lines of experimental work now suggest that histone acetylation has a causal role in facilitating the transcription process. (i) Histone acetylation can occur in genes that will become transcriptionally competent, but are yet to be transcribed (Hebbes et al., 1988, 1992, 1994). (ii) Histone acetylation, in particular that of H4 (Vettese‐Dadey et al., 1996), facilitates transcription factor access to nucleosomal DNA (Lee et al., 1993; Vettese‐Dadey et al., 1996) and the transcription of nucleosomal templates in vitro (Figures 1, 5 and 6). (iii) An increasing number of transcriptional activators have been characterized that have the capacity to acetylate the core histones (Brownell et al., 1996; Mizzen et al., 1996; Ogryzko et al., 1996; Yang et al., 1996). Although it is probable that these transcriptional activators have other functions that facilitate transcription (Barlev et al., 1995), modification of the histones is established as functioning to increase the accessibility of nucleosomal DNA to the transcriptional machinery and provides a viable regulatory step for the transcription process (Brownell et al., 1996; Wolffe and Pruss, 1996).
In this work, we have shown that the assembly of histone octamers onto a dinucleosomal template represses transcription, but that acetylation of the core histones can substantially relieve this repression (Figures 1, 5 and 6). How might histone acetylation facilitate transcription? Nucleosome mobility has been proposed as a molecular mechanism that permits the remodeling of repressive chromatin structures to allow gene activation (Ura et al., 1995; Varga‐Weisz et al., 1995). Through a related mechanism, it is possible that histone acetylation might alter the exact distribution of histone octamers with respect to gene sequence such that key regulatory elements become accessible to the transcriptional machinery (Ura et al., 1995). Our experimental results (Figures 2, 3 and 4) indicate that acetylation of the core histones does not influence nucleosome mobility, and is without effect on the translational or rotational positioning of DNA with respect to the histone octamer. Earlier work has shown that mononucleosomes containing acetylated core histones and the 5S rRNA gene become more accessible to the gene‐specific transcription factor TFIIIA (Lee et al., 1993). Although TFIIIA could gain access to the 5S rRNA gene, the DNA templates used in these experiments were too short to be transcribed efficiently (Wolffe et al., 1986). However, the increase in access of a transcription factor to mononucleosomal DNA is consistent with acetylation of the core histones facilitating transcription without a dependence on moving the histone octamer to alternate transcriptional positions, since on a short DNA template (186 bp, Wolffe et al., 1986) there are only limited opportunities for the histone octamer to move to alternative translational positions. Acetylation of the H4 N‐terminal tail renders the tail much more mobile in a nucleosomal context and greatly reduces its affinity for DNA in isolation (Cary et al., 1982; Hong et al., 1993). However, the stimulation of TFIIIA binding to the 5S nucleosome by histone acetylation is not dependent on the precise positioning of the histone octamer with respect to DNA sequence (Lee et al., 1993), nor is the stimulation of Gal4 binding to nucleosomal DNA by H4 acetylation dependent on nucleosome position (Vettese‐Dadey et al., 1996). Thus it is possible that acetylation might exert a general destabilizing influence on histone‐DNA interactions in the nucleosome.
Levels of histone acetylation such as those used in our experiments lead to a reduction in linking number change per 5S nucleosome in vitro (Norton et al., 1989, 1990). This reduction in linker number occurs without changing the helical periodicity of DNA on the surface of the histone octamer (Bauer et al., 1994). This suggests that the writhe of DNA in the nucleosome changes when the histones are acetylated, consistent with an allosteric change in nucleosome conformation. Others have suggested that the integrity of histone‐histone interactions within the nucleosome is reduced following acetylation of the N‐terminal tails in vivo (Chen and Allfrey, 1987; Chen et al., 1990; Oliva et al., 1990). However, rigorous biophysical studies indicate that histone hyperacetylation alone does not destabilize nucleosomes substantially in vitro (Ausio and van Holde, 1986; Garcia‐Ramirez et al., 1995). We suggest that the core histone tails have a major role in restricting transcription and that their acetylation can release this restriction through a local transition in nucleosome structure without alterations in the DNA sequences bound by individual histone octamers. Acetylation of the histone tails may be a necessary step in a staged process that requires additional histone modifications or events to destabilize the nucleosome further in vivo.
Linker histones repress transcription of nucleosomal templates independently of histone acetylation
Acetylated core histones are generally associated with transcriptionally active chromatin (Allegra et al., 1987; Hebbes et al., 1988), which is also relatively deficient in histone H1 (Kamakaka and Thomas, 1990; Tazi and Bird, 1990; Bresnick et al., 1992). Since the stability of chromatin higher order structures is dependent on both the core histone tails (Allan et al., 1982; Annunziato and Seale, 1983; Garcia‐Ramirez et al., 1992) and the linker histone H1 (Thomas, 1984), it has been suggested that histone acetylation and histone H1 deficiency might act in concert to destabilize any higher order structure. Such an effect would be augmented further if core histone acetylation itself destabilizes the association of linker histone with chromatin (Ridsdale et al., 1990; Perry and Annunziato, 1991; Juan et al., 1994). However, in earlier work using mononucleosomes containing the 5S rRNA gene, we have found that histone H5 associates with nucleosomal DNA independently of histone acetylation status (Ura et al., 1994). Here we extend these observations to demonstrate that histone H1 binds efficiently to dinucleosomes reconstituted with either acetylated or control core histones (Figure 5A) and represses transcription (Figure 5B). Our results indicate that histone acetylation in isolation does not severely restrict the incorporation of linker histones in chromatin. It is possible that other proteins that recognize acetylated core histones such as the small subunit of CAF1 (p48, Verreault et al., 1996), might be stably incorporated into nascent acetylated chromatin such that histone H1 is excluded. In our model dinucleosome, incorporation of histone H1 restricts the translational mobility of both acetylated and control nucleosomes, without influencing the rotational positioning of DNA on the surface of the histones (data not shown). Restriction of the mobility of histone octamers by histone H1 can account for the observed repression of transcription (Ura et al., 1995).
The repression of transcription mediated by histone H1 on acetylated templates implies that removal of H1 will still be a necessary component of any transcriptional activation process in which targeted core histone acetylation is involved.
A role for histone acetylation during chromatin assembly in facilitating transcription
Our experiments with acetylated histones and model dinucleosome templates provide some insight into the consequences of the acetylation of pre‐formed nucleosome structures for transcription. Such regulated acetylation might be directed by targeted components of the transcriptional machinery (Brownell et al., 1996; Mizzen et al., 1996; Ogryzko et al., 1996; Yang et al., 1996). However, a predominant form of histone acetylation during S phase of the cell cycle is that associated with the assembly of nascent chromatin (Ruiz‐Carrillo et al., 1975; Parthun et al., 1996; Verreault et al., 1996). In particular, histone H4 is diacetylated on lysines 5 and 12 (Sobel et al., 1995) prior to nucleosome assembly and then deacetylated as chromatin matures (Jackson et al., 1976; Shimamura and Worcel, 1989). Inhibition of deacetylation using sodium butyrate (Figure 6A) augments transcription >10‐fold in the absence of linker histones (Figure 6E). The significance of the capacity of histone acetylation to potentiate the transcriptional competence of chromatin under linker histone‐deficient conditions is that it provides a potential mechanism for the programing of genes during replication‐coupled chromatin assembly (Lee et al., 1993; Verreault et al., 1996). Since nucleosome assembly is staged in vivo with the initial deposition into stable nucleoprotein complexes of histones H3/H4, followed by H2A/H2B and finally linker histones, such as B4 and H1 (Worcel et al., 1978), intermediate chromatin structures enriched in acetylated histones but deficient in linker histones will be relatively accessible to components of the transcriptional machinery. This would provide a window of opportunity to pre‐assemble preset chromatin configurations competent for subsequent transcription (Becker, 1994; Wallrath et al., 1994).
Materials and methods
Plasmids and DNA fragments
The construction of the pX5S 197‐2 plasmid containing two tandem repeats of the 5S RNA gene has been described (Ura et al., 1995). A 424 bp XbaI‐XhoI fragment derived from plasmid pX5S 197‐2 was isolated from non‐denaturing acrylamide gels for nucleosome reconstitution after end‐labeling at the XbaI site with T7 polynucleotide kinase or Klenow fragment (New England Biolabs). This fragment was reconstituted into nucleosomes. Single‐stranded DNA from bacteriophage M13 mp18 and class III gene derivatives was prepared as previously described (Almouzni and Méchali, 1988a, b; Almouzni et al., 1990a, b).
Chromatin was purified from HeLa cells according to Lee et al. (1993). Nuclei isolated from cells treated with or without 10 mM sodium butyrate for 24 h were sonicated extensively, producing chromatin lengths of 10‐30 nucleosomes. Purified chromatin was treated with Dowex 50 W‐X2 in 0.2 M NaCl, 0.05 M potassium phosphate (pH 6.8) to remove linker histone H1 (Thoma et al., 1979). Nucleosome cores were reconstituted onto radioalabeled DNA fragments by the histone exchange method (Tatchell and van Holde, 1977). A 15‐fold template mass excess of chromatin was mixed with radiolabeled DNA in tubes followed by slow adjustment of the NaCl concentration (to 1 M). Tubes were incubated at 37°C for 15 min. Samples were transferred to a dialysis bag (with a molecular size limit of 6‐8 kDa) and dialyzed against 1.0 M NaCl, 10 mM Tris‐HCl (pH 7.5), 1 mM EDTA, 0.1 mM 2‐mercaptoethanol overnight. In this case, almost all products were dinucleosome cores and no naked DNA fragment or mononucleosome cores were detectable by nucleoprotein agarose electrophoresis.
After reconstitution, the oligonucleosome cores were loaded on 5‐20% sucrose gradients containing 10 mM Tris‐HCl (pH 7.5) 1 mM EDTA and 0.1 mM phenylmethylsulfonyl fluoride (PMSF), and then centrifuged for 16 h at 35 000 r.p.m. at 4°C in a Beckman SW41 rotor. Fractions were collected and analyzed by nucleoprotein agarose (0.7%) gel in 0.5× TBE (1× TBE is 90 mM Tris base/90 mM boric acid/2.5 mM EDTA). Fractions containing mono‐, di‐ or trinucleosomes were pooled separately, concentrated to ∼2.5 μg/ml using microcon‐30 (amicon), and dialyzed against 10 mM Tris‐HCl (pH 7.5), 0.1 mM EDTA, 1 mM 2‐mercaptoethanol overnight at 4°C. Samples were stored on ice until use. Chicken erythrocyte oligonucleosomes (chromatin lengths of 1‐30 nucleosomes) were prepared after removal of linker histones (Lee et al., 1993), and used as a control for isolation of the native dinucleosome complex.
Analysis of histone modification
Histones were resolved both on SDS‐18% polyacrylamide gel (Laemmli, 1970) and on Triton acid‐urea‐15% polyacrylamide gels (Zweidler, 1978). Triton acid‐urea gels were modified from the published protocol to optimize resolution (19:1 15% acrylamide:bisacrylamide, 6.3 M urea, 7 mM Triton X‐100, 5% acetic acid). Gels were pre‐run for 2‐3 h with a top buffer layer of 6.3 M urea, 7 mM Triton X‐100 containing 0.5 M 2‐aminoethanethiol‐HCl (Aldrich) as a scavenger. Electrophoresis was carried out in 5% acetic acid with an applied constant current of 3 mA/cm in a buffer‐cooled electrophoresis tank for ∼5 h. Analytic gels subsequently were fixed and stained with Coomassie blue. Proteins from Triton acid‐urea gels were electrotransferred onto nitrocellulose and stained with Ponceau S (Sigma) to evaluate transfer efficiency. Western blotting and immunochemical staining with polyclonal antibodies for the acetylated isoform of histone H4 (Turner and Fellows, 1989; Turner et al., 1992) was as previously described (Dimitrov et al., 1994; O'Neill and Turner, 1995). Two‐dimensional electrophoresis was performed as described (Russanova et al., 1980, 1989). The proteins were first separated in a 15% polyacrylamide slab gel containing 7 M urea and 5% acetic acid (Panyim and Chalkley, 1969). The strip with the separated proteins was then cut out from the gel and placed on the top of a second gel, which was made of a 2‐3 cm 5% stacking gel and a 12‐15 cm separating gel, containing 0.4% Triton X‐100 and 6 M urea (West and Bonner, 1980). The gels were stained either with 0.1% Coomassie brilliant blue R‐250 (Bio‐Rad Laboratories, Cambridge, MA) or with silver nitrate as described (Wray et al., 1981).
H1 binding experiments
A total of 100 ng (DNA content) of reconstituted nucleosome cores (0.6 pmol) were incubated with various amounts of histone H1 in 10 μl of binding buffer [10 mM Tris‐HCl (pH 8.0), 50 mM NaCl, 0.1 mM EDTA, 5% (v/v) glycerol] at room temperature for 15‐30 min (Hayes and Wolffe, 1993). Samples were located directly onto running 0.7% agarose gel in 0.5× TBE. After electrophoresis, the gels were dried and autoradiographed.
Two‐dimensional gel experiments
Two‐dimensional gel experiments to show the redistribution of nucleosome cores were performed following the procedure of Meersseman et al. (1992) with slight modifications. Reconstitutes with or without histone H1 were loaded onto non‐denaturing 4% polyacrylamide (29:1 acrylamide: bisacrylamide) gels at 4°C in 0.5× TBE. The gels were run at a maximum of 10 V/cm. Each lane was cut in half lengthwise. One half of each lane was left at 4°C, and the other was sealed and immersed at 37°C for 1 h. The gel strips were then arranged on top of a second non‐denaturing gel in the cold, and the second dimension was electrophoresed at 4°C under the same conditions as the first dimension.
Hydroxyl radical footprinting of dinucleosomes
Reconstitutes with or without histone H1 were treated with hydroxyl radicals prior to resolving nucleoprotein complexes on preparative 0.7% agarose gels (Hayes and Wolffe, 1992b). Samples contained labeled dinucleosome (60 ng of DNA) and chicken erythrocyte core particles (∼1 μg) and were incubated with or without 200 ng of histone H1 (one molecule/nucleosome core) as described above. After electrophoresis, bound or unbound H1 dinucleosome complexes were excised from the gel. DNA from these complexes was isolated and analyzed by denaturing polyacrylamide (6%) gel electrophoresis. Specific DNA markers were produced by Maxam and Gilbert cleavage at G residues.
Micrococcal nuclease mapping of dinucleosomes
Dinucleosomes (80 ng of DNA) in the absence or presence of 16 ng of histone H1 (molar ratio of histone to DNA = 1) were digested with 0.075‐0.6 U of micrococcal nuclease (Pharmacia) for 5 min at 22°C. Incubation with H1 was as described above. Ca2+ was adjusted to 0.5 mM concomitantly with addition of micrococcal nuclease. Digestions were terminated with addition of EDTA (5 mM), SDS (0.25%, w/v) and proteinase K (Gibco BRL) (1 mg/ml). The DNA was recovered and 5′‐end‐labeled with [γ‐32P]ATP and T4 polynucleotide kinase, and the end‐labeled DNA fragments were separated by electrophoresis in non‐denaturing 6% polyacrylamide gels. DNA fragments of nucleosome core and chromatosome products were recovered and digested with restriction endonucleases to determine micrococcal nuclease cleavage sites (Hayes and Wolffe, 1993).
Transcription reactions for dinucleosomes
Mono‐, di‐ or trinucleosome complexes previously resolved and separated by sucrose gradient centrifugation or naked DNA were used as templates for transcription in an extract from Xenopus oocyte nuclei. Oocyte nuclear extract was prepared as described previously (Birkenmeier et al., 1978). Transcription reaction conditions were as follows: 10 ng of radiolabeled template were added to a 10 μl reaction mixture containing 5 μl of nuclear extract in J buffer [10 mM HEPES (pH 7.4), 50 mM DCl, 7 mM MgCl2, 2.5 mM dithiothreitol (DTT), 0.25 U/μl RNasin (Gibco/BRL) and 0.1 mM EDTA] and pre‐incubated for 20 min before addition of exogenous triphosphates to 250 μM ATP, CTP and GTP, 50 μM UTP with 2.5 μCi of added [α‐32P]UTP. The reaction temperature was 22°C. Labeling was continued for 40 min after pre‐incubation. Radiolabeled transcripts were extracted with phenol, precipitated with ethanol and analyzed by electrophoresis in a 6% denaturing polyacrylamide gel. The level of 5S RNA transcription was quantitated with a Molecular Dynamics PhosphorImager. The radiolabeled 5S DNA template served as an internal control for recovery.
Egg collection and preparation of egg extract
Xenopus laevis frogs were purchased from Xenopus I, Ann Arbor, Michigan. Unfertilized eggs were collected from X.laevis frogs by injection of chorionic gonadotropin (Pregnyl; Organon). Eggs were collected in modified high‐salt Barth saline, dejellied with 2% cysteine in 1/10 Barth saline (pH 7.9), rinsed and sorted under a dissecting microscope to remove all damaged or abnormal eggs. The egg extract was prepared at 4°C as described (Almouzni and Méchali, 1988a, b). The eggs were washed in extraction medium [20 mM HEPES (pH 7.5), 70 mM potassium chloride, 1 mM DTT, 5% glycerol] and packed in centrifuge tubes. Excess medium was removed, leaving only the interstitial buffer between the cells. The cells were broken by direct centrifugation (12 000 g for 30 min), and the supernatant was collected. The supernatant was then recentrifuged at 150 000 g for 60 min in a Beckman 50 Ti rotor. The extract was stored in small aliquots at 80°C. Extracts were depleted of histone B4 as previously described (Dimitrov et al., 1994).
DNA synthesis, chromatin assembly and transcription reactions in egg extract
Unless otherwise specified, our standard reaction mixtures contained 10‐20 μg of DNA per ml of egg extract supplemented with exogenous 3 mM ATP and 5 mM MgCl2. We added 10‐15 μCi of [α‐32P]dATP per ml to monitor the DNA synthesis. Unless otherwise specified, chromatin assembly reactions were digested by micrococcal nuclease after addition of 3 mM CaCl2. Aliquots were taken during digestion, made up to 30 mM EDTA and 0.5% SDS, and treated as described previously (Almouzni et al., 1991), either for gel electrophoresis or for counting of acid‐insoluble material. When subjected to electrophoresis, the samples were deproteinized by phenol extraction after proteinase I (500 μg/ml) treatment. Each transcription reaction contained duplex DNA at a final concentration of 10‐20 μg/ml, 500 μM ATP, 500 μM UTP, 500 μM CTP, 100 μM GTP, 10 μCi of [α‐32P]GTP (3000 Ci/mmol) and 5 U of human placental RNase inhibitor (Bethesda Research Laboratories). Reactions were incubated for the times indicated. The samples were then processed for sequencing gel analysis as previously described (Almouzni et al., 1991).
We thank Dr Bryan Turner (University of Birmingham, UK) for the kind gift of antibodies specific for histone H4 isoforms. We thank Ms Thuy Vo for manuscript preparation.
- Copyright © 1997 European Molecular Biology Organization