ATP-dependent Nucleosome Remodeling and Histone Hyperacetylation Synergistically Facilitate Transcription of Chromatin*

Drosophila nucleosome remodeling factor (NURF) is an ISWI-containing protein complex that facilitates nucleosome mobility and transcriptional activation in an ATP-dependent manner. Numerous studies have implicated histone acetylation in transcriptional activation. We investigated the relative contributions of these two chromatin modifications to transcription in vitro of a chromatinized adenovirus E4 minimal promoter that contains binding sites for the GAL4-VP16 activator. We found that NURF could remodel chromatin and stimulate transcription irrespective of the acetylation status of histones. In contrast, hyperacetylation of histones in the absence of NURF was unable to stimulate transcription, suggesting that NURF-dependent chromatin remodeling is an obligatory step in E4 promoter activation. When chromatin templates were first hyperacetylated and then incubated with NURF, significantly greater transcription stimulation was observed. The results suggest that changes in chromatin induced by acetylation of histones and the mobilization of nucleosomes by NURF combine synergistically CL4B gel filtration, a (40 chromatin incubated with 0.2 pmol of GAL4-VP16, m of NURF (2.5 ng/ m l glycerol gradient fraction), m M ATP 30 min to allow chromatin remodeling. incubated nuclear followed the addition of NTPs for transcription. transcription purified components. extension analysis of transcription reactions was performed as except for the use of purified Drosophila general transcription factors ( GTFs and polymerase II ( Pol -fold activation correlate

The compaction of the eukaryotic genome in nucleosomes and the higher order folding of nucleosome arrays present barriers to regulatory proteins and the multisubunit enzymes that process genetic information (reviewed in Refs. [1][2][3][4][5]. In the nucleosome core particle, winding of 147 bp 1 of DNA over an octamer of histones creates severe distortion of the DNA helix and obscures roughly half of the helix surface (reviewed in Refs. 6 and 7). To counteract the constraints imposed by chromatin architecture, cells employ several distinct mechanisms. Strategies to destabilize chromatin include the use of homopolymeric stretches of DNA that resist bending; architectural, high mobility group-type proteins that unfold nucleo-some arrays; histone-modifying enzymes that covalently alter specific residues of the histone tails; and ATP-dependent chromatin remodeling complexes that facilitate nucleosome mobility (reviewed in Refs. 8 -15). In addition, the passage of an elongating RNA polymerase is also facilitated by distinct proteins that alter chromatin structure (16).
ATP-utilizing chromatin remodeling complexes can be classified into two main groups, containing either the SWI2/SNF2 or the related ISWI ATPases and their close relatives (reviewed in Refs. 17 and 18). SWI2/SNF2-containing complexes are large assemblies in the megadalton size range and composed of 11-15 distinct polypeptides. ISWI-containing complexes are smaller and are composed of 2-5 subunits. Both types of chromatin remodelers use the free energy of ATP hydrolysis to increase nucleosome mobility by changing nucleosome conformation (19 -22). A large body of evidence implicates the SWI2/ SNF2-containing complexes in transcription regulation. The swi2/snf2 gene was originally identified genetically as a transcriptional regulator in yeast (reviewed in Refs. 23 and 24). Recent genetic studies show that Drosophila iswi is required for engrailed and Ultrabithorax expression in vivo (25). In addition, SWI2/SNF2-and ISWI-containing complexes isolated from yeast, flies, and mammals can assist the transcriptional activation of model chromatin templates in vitro (26 -32).
Although the separate contributions of histone hyperacetylation and ATP-driven chromatin remodeling to transcription are well known, the interrelationships between these two major types of chromatin modifications and their relative contributions to the activation process are only beginning to be explored. Genetic studies indicate that the yeast SWI2/SNF2 and GCN5 genes perform independent but overlapping functions during transcriptional activation (55)(56)(57)(58)(59)(60)(61). To date, however, no biochemical studies have examined the mechanistic relationship between histone hyperacetylation and ATP-de-pendent chromatin remodeling. To address this question, we analyzed the relative contributions of Drosophila nucleosome remodeling factor (NURF) and histone acetylation to the activation of a model chromatin template in vitro.
Drosophila NURF was identified as a four-subunit, ISWIcontaining complex that has been implicated in transcriptional activation of chromatin (62)(63)(64)(65)(66)(67). Other ISWI-containing complexes related to NURF have subsequently been characterized from Drosophila (ACF (68) and CHRAC (21)) and human cells (RSF (30) and WCRF/hACF (69,70)). In a previous report, we demonstrated that purified NURF enables activation of a chromatin template by GAL4-HSF at an early step in the process of transcription (27). Here, we demonstrate synergism between the hyperacetylation of histones and nucleosome remodeling by NURF. Histone hyperacetylation alone did not stimulate transcription from our model chromatin template. However, in combination, the addition of NURF to hyperacetylated chromatin leads to synergistic activation of transcription. We suggest that a hierarchy exists in which NURF-dependent chromatin remodeling is an obligatory step during promoter activation. By the use of single and multiple round transcription experiments we demonstrate that the two types of chromatin modification stimulate transcription of chromatin both at the preinitiation and reinitiation stages.

EXPERIMENTAL PROCEDURES
Expression and Purification of Recombinant p300 and Human P/CAF Complexes-FLAG epitope-tagged human p300 was expressed in Sf9 cells after infection with recombinant baculovirus, and protein was purified as described elsewhere (71). Human P/CAF complex was purified from nuclear extracts of a HeLa cell line expressing FLAG epitope-tagged P/CAF as described (72).
Purification of NURF-NURF was purified from nuclear extracts of 0-12 h Drosophila embryos up to the glycerol gradient step as described (64).
Purification of Preassembled Chromatin by Sucrose Gradient Sedimentation-Sucrose gradient purification of preassembled chromatin was performed essentially as described (79). Briefly, 2.8 ml of preassembled chromatin mixture was applied onto a Beckman SW41 tube and centrifuged in a Beckman SW41 rotor at 26,000 rpm for 16 h at 4°C. The sucrose gradient buffer was 10 mM HEPES-KOH (pH 7.6), 1 mM EDTA, 35 mM NaCl containing from 30 to 50% (w/v) sucrose. 1.0-ml gradient fractions were collected, and DNA concentration was determined after 1% agarose gel electrophoresis of an aliquot (40 l) by ethidium bromide staining.
Histone Hyperacetylation of Chromatin by p300 and P/CAF Complex-The sucrose gradient fraction was directly subjected to treatment of the histone acetyltransferases. Typically, 100 l (1.4 g of DNA equivalent) of the gradient fraction was incubated at 26°C for 30 min with 1 mM dithiothreitol, 1 mM sodium butylate, 1 mM acetyl-CoA (lithium salt; Amersham Pharmacia Biotech), and ϳ5 pmol of p300 or P/CAF complex. Hyperacetylated chromatin was further purified and subjected to chromatin remodeling and transcription. For conventional purification of hyperacetylated chromatin, Sepharose CL4B (Amersham Pharmacia Biotech) was used in a SizeSep-400 spin column (Amersham Pharmacia Biotech) preequilibrated with elution buffer (10 mM HEPES-KOH (pH 7.6), 0.5 mM EGTA, 5 mM MgCl, 10% (v/v) glycerol, 50 mM KCl, 10 mM ␤-glycerophosphate, 1 mM dithiothreitol). Bovine serum albumin (Roche Molecular Biochemicals) was added to chromatin fractions after purification to a final concentration of 0.5 mg/ml. DNA content was estimated by agarose gel electrophoresis and ethidium bromide staining.
Histone Acetyltransferase Assay-For analysis of histone acetylation by SDS-PAGE and fluorography, 100 l (1.4 g of DNA equivalent) of the sucrose gradient purified chromatin was processed as described above, except that 1 nmol of [ 3 H]acetyl-CoA was introduced in place of 1 mM acetyl-CoA (lithium salt). The reaction was quenched by the addition of SDS-PAGE sample buffer and analyzed on 15% SDS-PAGE. The gels were then stained by Coomassie Brilliant Blue and exposed for fluorography. For Triton-acid-urea gel analysis, samples were processed as described above, using 1 mM acetyl-CoA (lithium salt). The acetylated chromatin was then precipitated with trichloroacetic acid and analyzed on a TAU gel, essentially as described (82).
In Vitro Transcription of Chromatin Template-In vitro transcription and primer extension analysis were performed as described previously (27,83,84). For the purified transcription system, typically 40 ng of chromatin (40 l of spin column-purified chromatin) was preincubated in 80 l (final volume) containing 25 ng of TFIIA, 15 ng of TFIIB, ϳ120 ng of immunopurified TFIID, 50 ng of TFIIE, 50 ng of TFIIF, ϳ40 ng of purified TFIIH, 10 ng of TFIIS, ϳ30 ng of RNA polymerase II, 20 mM HEPES-KOH (pH 7.6), 5 mM MgCl, 40 mM KCl, 2.6% (v/v) polyethylene glycol 8000 (final concentration), 3.75 mM (NH 4 ) 2 SO 4 , 1 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride for 30 min at 26°C to form preinitiation complexes and then transcribed for 10 min at 26°C with the addition of ribonucleotide triphosphates (0.55 mM final concentration). Transcripts were detected by primer extension using a 32 P-labeled AdE4 primer corresponding to positions 72-99 of the transcribed strand of pWEGIE-0. cDNA products were analyzed on a 6% denaturing polyacrylamide gel. Quantitation of transcripts was performed on a Fuji Bio-Image Analyzer.
For transcription in a crude system, soluble nuclear fraction (nuclear extract) was prepared from 0 -12-h Drosophila embryos as described (85,86). Reactions were performed as described above, except that 70 g of the soluble nuclear fraction was used instead of purified general transcription factors and RNA polymerase II.
For the heparin challenge protocol, reactions were performed essentially as described for transcription with the purified system. A synthetic RNA polymerase II pause at ϩ5 site was created by the addition of 0.55 mM ATP, 0.55 mM CTP, and 0.55 mM UTP. Following incubation at 26°C for 5 min, 400 ng of heparin and 0.55 mM GTP were added to block reinitiation of RNA polymerase II, and the reactions were incubated for a further 30 min at 30°C (87,88). Transcribed RNAs were analyzed by primer extension as above.
Micrococcal Nuclease and Restriction Enzyme Digestion of Chromatin-Micrococcal nuclease (MNase) digestion analysis and sequential Southern blot hybridization were performed as described (62,80,89). A restriction enzyme accessibility assay was performed essentially as described (90). Typically, 50 l (ϳ50 ng of DNA equivalent) of spin column-purified chromatin, 0.5 l of NURF fraction, ATP (0.5 mM final concentration), and 0.5 pmol of GAL4 derivative (GAL4-VP16) were mixed and adjusted to 65 l with elution buffer. The reaction was incubated at 26°C for 30 min and subjected to micrococcal nuclease or restriction enzyme digestion. For MNase digestion, the reaction mixture was directly treated with enzyme and analyzed as described (27). For restriction enzyme digestion, the reaction mixture was incubated with either 0.5 unit of BamHI (digested at position Ϫ46) or 0.5 unit of HaeII (digested at position ϩ98), at 26°C for 30 min. Digested DNA samples were deproteinized, precipitated with ethanol, and redigested with PstI and ClaI. Samples were loaded on a 1.3% agarose gel and analyzed by Southern blot hybridization as for the MNase assay.

In Vitro Transcription of Hyperacetylated Chromatin-To
analyze the relative contributions of histone acetylation and ATP-dependent chromatin remodeling to transcription, we used a 9-kilobase pair cosmid (pWEGIE-0) containing five tandemly repeated GAL4 binding sites upstream of the adenovirus E4 core promoter (80). The experimental procedure for reconstitution and acetylation of pWEGIE-0 chromatin is outlined in Fig. 1A. After assembly with the Drosophila embryo S-190 extract and purified Drosophila core histones, chromatin is purified by sucrose gradient sedimentation. This procedure removes the bulk of nonhistone proteins, leaving histones as the predominant proteins in the chromatin preparation, as judged by SDS-PAGE and silver staining (79). The purified chromatin is then modified with a HAT and acetyl-CoA, followed by repurification on a Sepharose CL4B spin column, to preclude acetylation of remodeling proteins and transcription factors that are introduced subsequently. Chromatin is next incubated with saturating amounts of the transcription activa- After histone hyperacetylation by p300 and further purification by CL4B gel filtration, a portion (40 ng) of the chromatin was incubated with 0.2 pmol of GAL4-VP16, 0.2 l of NURF (2.5 ng/l glycerol gradient fraction), and 0.5 mM ATP for 30 min to allow chromatin remodeling. The sample was then incubated with nuclear extract, followed by the addition of NTPs for transcription. D, in vitro transcription using purified components. Primer extension analysis of transcription reactions was performed as above, except for the use of purified Drosophila general transcription factors (GTFs) and RNA polymerase II (Pol II). In C and D, the -fold activation is the average of three independent experiments. Hence, the numerical values do not fully correlate with visual inspection of the presented autoradiograph. tor GAL4-VP16 and saturating amounts of purified NURF and ATP to allow the mobilization of nucleosomes. This is followed by the assembly of preinitiation complexes with either a Drosophila soluble nuclear fraction that has little ATP-dependent chromatin remodeling activity (27,80) or a purified transcription system consisting of native and recombinant Drosophila general transcription factors and RNA polymerase II. Transcription is then initiated upon the addition of ribonucleotide triphosphates (NTPs), and the RNA products after 10 min of transcription are analyzed by primer extension. We note that Drosophila embryo extracts contain histone acetyltransferases and deacetylases that can modify histones during chromatin assembly (45,47,82). However, we and others (47,82) have found the chromatin product after in vitro assembly yields histones whose final levels of acetylation are extremely low, thus providing an appropriate substrate for histone hyperacetylation.
NURF and Histone Hyperacetylation Synergistically Facilitate Transcription of Chromatin-We utilized recombinant human p300 as a potent HAT enzyme to hyperacetylate nucleosomal histones (71,91); p300 preferentially acetylates nucleosome core histones in vitro at the same sites that are acetylated in vivo (92). We confirmed that histone hyperacetylation by p300 occurs in our reactions by the incorporation of [H 3 ]acetyl-CoA into all four core histones (Fig. 1B, middle  panel). In addition, Triton-acid-urea gel electrophoresis showed that p300 quantitatively modifies, in an acetyl CoA-dependent fashion, the bulk of histone H4 in reconstituted chromatin, leading to a conversion from a predominantly unacetylated form (for purposes of discussion, we refer to this state as "unacetylated") to the mono-, di-, and triacetylated states (Fig. 1B,  bottom panel). The sequential operations of histone hyperacetylation and ATP-dependent chromatin remodeling by NURF resulted in very high level activation of the E4 promoter in a crude transcription system (139-fold; Fig. 1C, lane 16). (In concurrence with previous studies in vivo and in vitro, initiation from the E4 promoter occurs at A ϩ1 and T Ϫ6 positions (103,104); unless otherwise noted, we use the sum of the transcription signals from both positions as a measure of -fold activation. In addition, the numerical values given are averages of three experiments.) This 139-fold level of activation was ϳ5-fold higher than the activation conferred by NURF in the absence of histone hyperacetylation (30-fold; Fig. 1C, lane 4). The ϳ5-fold increase was dependent on both p300 and acetyl CoA, since it was not observed upon the omission of either reagent (Fig. 1C,  lanes 8 and 12). Given that p300 is known to harbor other activities (51,71,91), the requirement for acetyl-CoA suggests that it is the HAT activity of p300 that is responsible for the further increase of transcription. As previously noted (e.g. Refs. 27 and 80), transcription of chromatin was also highly dependent on the presence of the VP16 activation domain (data not shown; Fig. 1C).
We next investigated whether histone hyperacetylation could substitute for the ATP-dependent nucleosome remodeling step by omitting NURF from the reaction protocol. Interestingly, there was little or no activation of the E4 promoter in the absence of remodeling by NURF, despite bulk histone hyperacetylation by p300 (1.1-fold; Fig. 1C, lane 14). To confirm the above transcription results, which were obtained with the Drosophila soluble nuclear fraction, we prepared a purified Drosophila polymerase II transcription system using bacterially expressed and native components (see "Experimental Procedures") and performed the same set of experiments. With the purified transcription system, we were able to demonstrate again the synergy between histone hyperacetylation and nucleosome remodeling by NURF. Transcription of chromatin with the purified polymerase II system showed that activation was increased further by ϳ5-fold when histone hyperacetylation by p300 is included in the protocol (202-fold; Fig. 1D, lane  16, versus 33-, 42-, and 37-fold; lanes 4, 8, and 12). Moreover, little or no activation was observed when NURF was omitted from the protocol (1.1-fold; Fig. 1D, lane 14). Thus, core histone hyperacetylation can operate synergistically with NURF but is unable to replace it functionally in activating transcription of chromatin, suggesting that ATP-dependent nucleosome remodeling is an obligatory event for the GAL4-E4 promoter.
We performed additional experiments that underscore a role for histone hyperacetylation in facilitating transcription of chromatin. Introduction of desulfocoenzyme A (DSA), an acetyltransferase inhibitor, along with p300 and acetyl CoA blocked the synergistic activation of chromatin (35-fold; Fig. 2A, lane 22), while synergy was retained when desulfocoenzyme A was introduced after chromatin was hyperacetylated (163-fold; Fig. 2A, lane 18). We also deliberately reintroduced acetyl-CoA at a late stage of the protocol (after the removal of p300), to assess whether synergistic activation might be caused by undetected HAT activities that could be contaminating the NURF preparation or are dominant in the transcription extract. As shown in Fig. 2B, continuous exposure of the chromatin template to acetyl-CoA during ATP-dependent remodeling, preinitiation complex formation, and transcription did not further increase activation (32-fold (Fig. 2B, lane 2) versus 30-, 29-, and 28-fold (lanes 6, 10, and 14)). Taken together, the results indicate that hyperacetylation of bulk histones can operate synergistically with NURF in GAL4-VP16-mediated transcription of chromatin.
Hyperacetylation of Core Histones Fails to Mobilize Nucleosomes-To examine how histone hyperacetylation might assist in transcription, we analyzed the structure of the acetylated chromatin by digestion with micrococcal nuclease (MNase), which cleaves nucleosome linker DNAs. After histone hyperacetylation with p300, chromatin was repurified and incubated with GAL4-VP16, digested with MNase, and processed for Southern blot hybridization with an E4 promoter-specific oligonucleotide. As previously reported, MNase digestion of chromatin reconstituted with the Drosophila extract produces a "ladder" of DNA fragments with a repeat length of ϳ180 bp, corresponding to periodically spaced nucleosomes (Ref. 27; Fig.  3A, panel 1; two digestion points per panel). Previous studies (27,28) also indicated that, in the absence of NURF, a saturating amount of GAL4-VP16 is able to bind to the five cognate sites (spanning ϳ100 bp) without disturbing periodic nucleosome spacing at the E4 promoter. We observed similar results (Fig. 3A, panel 2; data not shown). (GAL4-VP16 binding to linker DNA protects it from MNase digestion and increases the yield of fragments corresponding to dinucleosomes).
Upon remodeling by NURF, the mobilization of nucleosomes away from GAL4 binding sites to slightly distal locations produces a smeared ladder of DNA fragments in the MNase digestion assay (Ref. 27; Fig. 3A, panel 4). Importantly, in the absence of remodeling by NURF, the hyperacetylated chromatin retained periodic nucleosome spacing at the E4 promoter, as shown by the integrity of the DNA ladder (Fig. 3A, panel 14). Hence, histone hyperacetylation by p300 could not substitute for NURF to mobilize nucleosomes over GAL4 binding sites. Furthermore, the combination of histone hyperacetylation by p300 and chromatin remodeling by NURF did not lead to an increase in the kinetics of MNase digestion (Fig. 3A, panels 4,  8, and 12 versus panel 16).
As an alternative probe of chromatin accessibility, we used restriction endonucleases instead of MNase. Chromatin was treated with BamHI (cleavage position Ϫ46, within the region of DNase I hypersensitivity induced by GAL4-VP16 and NURF; Ref. 27) or with HaeII (cleavage position ϩ98, in the downstream nucleosome). In the absence of remodeling by NURF, Southern blot analysis showed that BamHI digestion of hyperacetylated chromatin is only slightly increased over unacetylated controls (18% versus 21, 20, and 22% digestion) (Fig. 3B,   FIG. 2. Histone acetyltransferase activity of p300 2, 6, and 10). Similarly, HaeII digestion was not substantially affected by histone hyperacetylation in the absence of remodeling by NURF (1.0% versus 1.5, 1.5, and 1.1% digestion) (Fig. 3C, lane 14 versus lanes 2, 6, and 10). In contrast, remodeling by NURF caused a significant increase of BamHI digestion from ϳ20 to ϳ60%. This level of digestion was found for both unacetylated chromatin (Fig. 3B, lanes 2, 6, and  10 versus lanes 4, 8, and 12) and hyperacetylated templates (Fig. 3B, lane 14 versus lane 16), confirming that histone hyperacetylation did not alter nucleosome structure appreciably. Similarly, remodeling by NURF increased HaeII digestion from 1.0 to 1.5% to 14 -19% (Fig. 3C, lanes 2, 6, 10, and 14 versus  lanes 4, 8, 12, and 16), independent of the acetylation state of chromatin. Overall, the results indicate that the structural changes induced by histone hyperacetylation that increase transcription occur beyond the level of the nucleosome core particle and most likely involve changes in higher order chromatin folding. (A study reporting increased accessibility to restriction nuclease upon histone hyperacetylation may not be inconsistent with our results, because the restriction enzyme reaction contained ATP and HeLa nuclear extract, a potential source of NURF-like nucleosome remodeling activities; see Ref. 46).
NURF and Histone H3 Hyperacetylation Facilitate Transcription of Chromatin-Given that p300 potently acetylates all four core histones, we used PCAF, a HAT with more restricted substrate specificity, to analyze the effects of histone hyperacetylation (PCAF acetylates histone H3 at lysine 14 preferentially; see Ref. 92; Fig. 4B). We found that histone H3 hyperacetylation by the purified PCAF complex leads to a 4.5-fold increase of transcription over the level detected in its absence (166-fold (Fig. 4C, lane 16) versus 37-, 33-, and 36-fold (lanes 4, 8, and 12)). Consistent with the observations for p300, histone H3 hyperacetylation by PCAF could not substitute for NURF in remodeling chromatin for activation by GAL4-VP16 (Fig. 4C, lane 14). Moreover, MNase and restriction enzyme digestion of histone H3 hyperacetylated chromatin showed a lack of nucleosome mobilization at the E4 promoter when NURF was omitted from the protocol (Fig. 5A, panel 14 versus  panel 16; Fig. 5, B and C (lane 14 versus lane 16)). These findings underscore the obligatory requirement for nucleosome mobilization by NURF for transcription of our model chromatin template. The results also suggest that hyperacetylation of one histone (H3) may be sufficient to create synergy with NURF in facilitating activation by GAL4-VP16.
NURF and Histone Hyperacetylation Facilitate Initiation and Reinitiation of Transcription-Our transcription reactions usually proceed for only 10 min. In this way, the total number of rounds of transcription is restricted. Therefore, our assay predominantly should reveal effects on the preinitiation or initiation stages of transcription. Nonetheless, the observed activation of the E4 promoter could also include a contribution from transcription reinitiation during the 10-min interval. Accordingly, we modified the protocol to allow only a single round of transcription on the chromatin template (Fig. 6A). After hyperacetylation by p300 or PCAF, remodeling by NURF, and assembly of preinitiation complexes, initiation is allowed with only three rNTPs (ATP, CTP, and UTP), creating an artificial pause at ϩ6 (the location of the first G in the transcript). GTP and heparin are then introduced to permit read-through of the first round transcript and to block reinitiation of transcription, respectively (87,88,93) (Fig. 6A).
The single round transcription experiments revealed that transcription from the Ϫ6-position of the E4 promoter was stimulated ϳ20-fold when chromatin was treated with NURF (Fig. 6B, lanes 10, 12, and 14). There were no single round transcripts when NURF was omitted from the protocol (data FIG. 3. Histone hyperacetylation by p300 does not remodel  nucleosomes. A, nucleosome organization of Gal4-E4 promoter after histone hyperacetylation by p300. An aliquot of chromatin used for in vitro transcription after the remodeling step was analyzed by MNase digestion (two digestion points) and Southern blot hybridization, using a radiolabeled oligonucleotide probe specific for the E4 promoter region (Ϫ57 to Ϫ25 bp). B and C, restriction enzyme accessibility. An aliquot of chromatin after the remodeling step was digested with BamHI (B) or HaeII (C) and analyzed by Southern blot hybridization using a oligonucleotide probe corresponding to positions 72-99 of the E4 promoter. The uncut PstI/ClaI fragment and the BamHI-or HaeII-cut fragments are indicated by arrowheads. The percentage of digestion is given by the average value from three sets of experiments. D, locations of restriction enzyme cleavage sites on the GAL4-E4 promoter are given in the diagram. not shown). When histone hyperacetylation by p300 was combined with the action of NURF, we observed an additional ϳ1.9-fold increase in the single-round transcript from the Ϫ6 start site (38-fold (Fig. 6B, lane 14) versus 21-fold (lane 16)). Hence, as anticipated, NURF, and histone hyperacetylation combined with NURF, stimulate transcription at the preinitiation or initiation stage. Interestingly, single-round transcription occurred only from the Ϫ6-position of the E4 promoter. This unexpected finding raises the possibility that subsequent rounds of transcription starting from ϩ1 may be dependent on subtle repositioning of promoter-proximal nucleosome(s), caused by passage of the first polymerase.
When multiple rounds of transcription were allowed (without heparin), NURF in the absence of histone hyperacetylation conferred the typical ϳ36-fold activation, or ϳ1.8 rounds of transcription in 10 min from the Ϫ6-position (Fig. 6B, lanes 2,  4, and 6). Multiple round transcription of chromatin hyperacetylated by p300 and remodeled by NURF gave ϳ183-fold activation, or ϳ4.8 rounds of transcription in 10 min, a ϳ2.7fold increase in reinitiation attributable to histone hyperacetylation (Fig. 6A, lane 8 versus lane 6). (For calculations of the rounds of transcription, we have not included transcripts initiating at ϩ1, although reinitiation from this position is also increased when histone hyperacetylation is combined with remodeling by NURF). Our results indicate that both types of chromatin modification can stimulate reinitiation of transcrip-tion. We have obtained similar findings showing a ϳ1.8-fold increase of single round transcription and a ϳ2.8-fold increase in the rounds of transcription when the combined effects of NURF and histone H3 hyperacetylation by PCAF are compared with the effects of NURF alone (Fig. 6C). DISCUSSION An increasing number of in vitro experiments utilizing chromatin templates provide compelling evidence that noncovalent and covalent modifications of chromatin can directly participate in the process of transcriptional control by sequence-specific activators. To date, however, studies show that ATP-dependent chromatin remodeling or histone hyperacetylation can each function to facilitate transcription. But there is little or no information on how these two major types of chromatin modification might be mechanistically integrated in the transcription process. In this report, we have begun to address this issue by comparing the relative contributions of histone hyperacetylation and ATPdependent chromatin remodeling to transcription of a model chromatin template in vitro. We found evidence of synergism between histone hyperacetylation and nucleosome remodeling by NURF in allowing GAL4-VP16-dependent activation of the adenovirus E4 minimal promoter. In combination, the two modifications caused dramatic stimulation of preinitiation or initiation as well as reinitiation of transcription.
To demonstrate the effects of histone hyperacetylation on  Fig. 1. C, in vitro transcription using a purified transcription system. Primer extension analysis was performed as in Fig. 1, with the exception of using human P/CAF complex instead of human p300.
GAL4-VP16-dependent transcription of chromatin, we used two characterized histone acetyltransferases, p300 and PCAF, as enzymatic tools. In addition to its potent HAT activity on all four nucleosomal histones (71,92), p300 has also been shown to acetylate sequence-specific transcription factors such as p53 (94) and to function as a multifunctional transcriptional coactivator (e.g. Refs. 51 and 95). Therefore, to deploy only the HAT activity of p300, it was necessary first to hyperacetylate chromatin with p300 and then to remove it by gel filtration prior to introduction of NURF, GAL4-VP16, and the transcription ma-chinery. (This technical requirement constrained the order of addition of the chromatin modifiers in our reaction protocol, which should not be taken necessarily as the order of chromatin modifications as they occur in vivo).
The combination of both chromatin modifications led to a very high level (Ͼ100-fold) of transcriptional activation, with a close correlation between increased activation and bulk hyperacetylation of histones. Such a state of histone modification on the reconstituted 9-kilobase pair cosmid chromatin may be compared with the "global" status of histone hyperacetylation FIG. 5. Histone hyperacetylation by P/CAF complex does not remodel nucleosomes. A, nucleosome organization of Gal4-E4 promoter after histone hyperacetylation by P/CAF complex. An aliquot of chromatin used for in vitro transcription (1.0 h) was analyzed by MNase digestion (two digestion points) and processed as in Fig. 3. B and C, restriction enzyme accessibility. An aliquot of chromatin used for in vitro transcription (1.0 h) was digested with BamHI (B) or HaeII (C) and analyzed as in Fig. 3. for an activated locus in vivo, e.g. as found at the chick ␤-globin gene cluster (96). Interestingly, preferential hyperacetylation of histone H3 on cosmid chromatin by PCAF, a more specific HAT enzyme, gave a level of synergistic activation similar to that provided by p300, indicating that general modification of a specific histone could suffice to assist the transcription process. Histone hyperacetylation can also be limited to several or more nucleosomes surrounding promoter regions (59,60,97,98). This aspect of histone hyperacetylation is not addressed by our study, but it was recently shown that GAL4-VP16 targeting of p300 can cause histone H3 and H4 hyperacetylation within proximal nucleosomes (54, 102) and transcriptional activation of chromatin in vitro (54).
The requirement for the VP16 activation domain of GAL4-FIG. 6. Histone hyperacetylation and NURF facilitate transcription preinitiation and reinitiation. A, experimental protocol. B, primer extension analysis showing multiple round (left) and single round transcription (right) after histone hyperacetylation by p300. 5 min after initiation with ATP, CTP, and UTP, heparin was introduced into the reaction, followed by GTP; transcription was allowed for 10 min. Transcripts originating from two major start sites at Ϫ6 and ϩ1 on the E4 promoter are indicated. For multiple round transcription, reactions were performed as in single round transcription, except for the omission of heparin. C, primer extension analysis showing multiple round and single round transcription after histone hyperacetylation by P/CAF complex. The reaction was performed as above, except for using P/CAF complex instead of p300. -Fold activation in B and C is the average of three independent experiments. VP16 in revealing the effects of chromatin modifications on transcription suggests that histone hyperacetylation and ATPdependent nucleosome remodeling help to stimulate recruitment of general transcription factors and the RNA polymerase II machinery by the sequence-specific activator. It is of interest that, despite the ability to synergize with NURF, histone hyperacetylation alone was unable to stimulate recruitment of the transcription apparatus. This suggests that ATP-dependent nucleosome repositioning is obligatory to activate the E4 promoter and that the two modes of chromatin modification enable GAL4-VP16-dependent transcription using distinct mechanisms. It should be noted that the obligatory requirement for NURF in this study is not inconsistent with studies of other promoters (e.g. human immunodeficiency virus-1, hsp26, and AdML promoters) demonstrating significant effects of histone hyperacetylation on transcription, in the apparent absence of remodeling by NURF (45-47, 51, 52, 54). In those studies, transcription of hyperacetylated chromatin templates was likely to have occurred in the presence of NURF-like activities present in the high salt nuclear extracts employed as a source of the transcriptional machinery. It is also possible that the E4 promoter may possess a specific requirement for nucleosome remodeling by NURF.
How might histone hyperacetylation facilitate GAL4-VP16dependent recruitment of the transcription apparatus in the context of ATP-dependent nucleosome mobility? While it is possible that hyperacetylation of chromatin could help recruitment by extending the range or efficiency of nucleosome remodeling by NURF, we did not observe significant enhancement of nucleosome mobility under saturating NURF conditions, as measured by MNase and restriction enzyme digestions. Moreover, histone hyperacetylation of chromatin did not promote the binding of the GAL4 activator, shown in a previous study (99), since effects on activator binding were bypassed in our experiments by the use of saturating amounts of GAL4-VP16. Rather, we favor two nonexclusive mechanisms by which chromatin modification could facilitate recruitment of the transcription apparatus. It is possible that histone hyperacetylation could decondense the higher order folding of nucleosome arrays in a manner not readily detectable by probing MNase or restriction endonucleases. Such a view is consistent with studies in which histone hyperacetylation was found to unfold arrays of 5 S nucleosomes and dramatically enhance transcription by RNA polymerase III (100). Alternatively, histone H4 hyperacetylation could directly enhance recruitment of the general transcription factor TFIID to promoter chromatin through increased affinity for the bromodomains of the TAF II 250 subunit (101).
Thus, the recruitment of the transcription apparatus by GAL-4VP16 could be assisted by at least three distinct, chromatin-based processes: 1) the repositioning or sliding of nucleosomes to distal locations, mediated by the GAL4 DNA binding domain and catalyzed by NURF, 2) the unfolding of higher order structure caused by histone hyperacetylation, and 3) the increased binding of TFIID to hyperacetylated nucleosomes. For the GAL4-UAS-minimal E4 promoter, the disposition of nucleosomes over promoter sequences may render nucleosome repositioning obligatory for activation, despite higher order unfolding, thus creating the observed regulatory hierarchy. While such a scenario may be envisaged for some gene promoters, it is likely that other gene promoters may possess intrinsic mechanical properties and a series of DNA cis-regulatory elements that dictate a different hierarchical order (and a different set) of chromatin modifications. Our present findings should provide a useful foundation for future studies to eluci-date the multistep process of activating natural promoters in the context of chromatin.