Reconstitution of Nucleosome Positioning, Remodeling, Histone Acetylation, and Transcriptional Activation on the PHO5 Promoter*

The PHO5 gene promoter is an important model for the study of gene regulation in the context of chromatin. Upon PHO5 activation the chromatin structure is reconfigured, but the mechanism of this transition remains unclear. Using templates reconstituted into chromatin with purified recombinant yeast core histones, we have investigated the mechanism of chromatin structure reconfiguration on the PHO5 promoter, a prerequisite for transcriptional activation. Footprinting analyses show that intrinsic properties of the promoter DNA are sufficient for translational nucleosome positioning, which approximates that seen in vivo . We have found that both Pho4p and Pho2p can bind their cognate sites on chro-matin-assembled templates without the aid of histone-modifying or nucleosome-remodeling factors. However, nucleosome remodeling by these transcriptional activators requires an ATP-dependent activity in a yeast nuclear extract fraction. Finally, transcriptional activation on chromatin templates requires acetyl-CoA in addition to these other activities and cofactors. The addition of acetyl-CoA results in significant core histone acetylation. These findings indicate that transcriptional activation requires Pho4p, Pho2p, nucleosome remodeling, and nucleosome acetylation.

Chromatin functions to compact and organize DNA in the nucleus of eukaryotic cells in a manner that allows regulated access to genes for transcription and DNA replication. The role of nucleosomes in transcriptional regulation has become a major area of study. From in vitro studies, it is clear that nucleosomes can repress transcription by RNA polymerase II (1)(2)(3). Further confirmation that nucleosomes play a prominent role in gene regulation came from studies showing that histone H4 depletion in yeast cells results in nucleosome loss and transcription derepression of several RNA polymerase II-transcribed genes (4,5). Barring artificial loss, nucleosomes must be reconfigured prior to transcriptional activation of these genes. Recently, many yeast activities that remodel chromatin have been identified, including the SWI/SNF, INO80, ISW1, ISW2, and RSC complexes (reviewed in Ref. 6). All of these activities contain a DNA-or chromatin-dependent ATPase subunit required for remodeling. In addition, many transcriptional activators recruit histone acetyltransferase activities, which in yeast include the ADA, Spt-Ada-Gcn5 acetyltransferase (SAGA), NuA3, and NuA4 complexes (7,8).
The chromatin structure of the yeast PHO5 promoter regulates RNA polymerase II transcription of the PHO5 gene, which encodes the major, secreted acid phosphatase in yeast (9). Under repressive conditions (adequate phosphate) the PHO5 promoter is bound in an array of positioned nucleosomes (10). Activation of PHO5 through phosphate starvation is accompanied by a loss or reconfiguration of four nucleosomes from the promoter region (11).
Activation is initiated through a signal transduction pathway that ultimately results in dephosphorylation of the transcription factor Pho4p, allowing its transport into the nucleus where it can bind to two upstream activation sequences (UAS) 1 on the PHO5 promoter, UASp1 and UASp2 (12)(13)(14)(15). Pho2p, a second transcription factor involved in PHO5 activation, binds cooperatively with Pho4p in vitro at both UASs (16). Pho4p and Pho2p must physically interact in vivo for transcription activation to occur (17). This interaction can occur regardless of phosphate concentration but does require the presence of DNA (18). Both transcription factors are required for full chromatin remodeling and activation of the PHO5 promoter (19). However, Pho4p is the primary trigger for activation, because Pho2p appears to be constitutively expressed and active. In addition, overexpression of PHO4 in a pho2 null strain is sufficient for the full chromatin transition and partial transcriptional activation (15).
Chromatin remodeling on the PHO5 promoter occurs in the absence of replication and does not require transcription (20,21). These findings suggest that chromatin remodeling and activation of transcription can occur independently. Gcn5p, the catalytic subunit of ADA and Spt-Ada-Gcn5 acetyltransferase HAT complexes is required for full PHO5 derepression in a ⌬pho80 strain or a ⌬rpd3 strain and full activation in the absence of one of the UASs (22,23). Furthermore, the rates of PHO5 chromatin remodeling and transcriptional activation are significantly delayed (2-to 3-fold) in a ⌬gcn5 strain (24). Interestingly, this delay appears to be specific for chromatin structure rather than for activator (Pho4p). Esa1p, the catalytic subunit of NuA4, is indicated to be involved in maintaining the H4 acetylation state on the PHO5 promoter (23). SWI/SNF is not required for activation of PHO5, because the PHO5 chromatin transition is unaffected and acid phosphatase levels achieve 70% of wild type levels under activating conditions in strains carrying an SNF2 disruption (25). A requirement for RSC has not been tested, and ISW1 and ISW2 do not seem to have a role in PHO5 regulation (26). However, INO80 has been reported to be required for full PHO5 activation (27,28). Although the trans-acting factors and cis-acting sequences required for chromatin structure modulation and transcription activation of PHO5 have been extensively studied, the detailed mechanisms of this process are not yet understood.
Using minichromosome templates isolated from yeast cells, remodeling on PHO5 promoter was shown to require Pho4p, Pho2p, ATP, and fractionated nuclear extract, further elucidating the mechanism of remodeling (29). However, to study the fine details of chromatin structure modulation, a fully defined chromatin template was required. Here we describe the reconstitution of chromatin templates with purified, recombinant yeast core histones possessing many important aspects of the repressed PHO5 promoter. Based on footprinting results, we conclude that sequence-dependent intrinsic properties of the DNA can produce translational positioning of nucleosomes on the PHO5 promoter that approximates that seen in vivo. Consistent with nucleosome positioning, reconstitution of nucleosomes on the PHO5 promoter strongly repressed transcription. Both Pho4p and Pho2p can bind their cognate sites on chromatin-assembled templates without the aid of remodeling factors. However, nucleosome remodeling by these transcriptional activators requires an ATP-dependent activity in a yeast nuclear extract fraction. Finally, transcriptional activation on chromatin templates requires acetyl-CoA in addition to these other activities and cofactors. These findings indicate that transcriptional activation requires Pho4p, Pho2p, nucleosome remodeling, and nucleosome acetylation. Furthermore, DNA binding, nucleosome remodeling, and transcriptional activation are separable steps, in concordance with the ability to separate nucleosome remodeling and transcriptional activation in vivo (15,20,21). Finally, our results suggest that Pho4p and Pho2p binding to UASp2 in nucleosome Ϫ2 occurs prior to nucleosome remodeling or acetylation, which then function subsequent to activator binding.

EXPERIMENTAL PROCEDURES
Plasmids-The pPHO5-G-less plasmid was produced by subcloning the 526-bp BamHI to ApoI fragment (Ϫ542 to Ϫ16) of pMH313 (5) and a DNA fragment containing no guanine bases in the RNA-like strand into pUC19. In this construct, three guanines were changed to cytosines in the RNA-like strand at positions Ϫ24, Ϫ22, and Ϫ17 relative to the ATG. The plasmid pMH313-G-less was produced by inserting a 100-bp DNA fragment containing no guanine residues in the RNA-like strand into the ApoI and BamHI sites. The same three guanines were changed to cytosines between the RNA start sites and the ATG.
Nucleosomal Template Reconstitution-Recombinant yeast core histones, purified as described previously (30), were provided by the Luger laboratory. Yeast core histones and recombinant yeast Nap1p were combined at a total core histones to Nap1p ratio of 1:8 (w/w). Acetylated bovine serum albumin (New England BioLabs) was added to 100 g/ml, Igepal (Sigma) was added to 0.02%, and the mixture was dialyzed for 16 h at 4°C against 10 mM Tris-HCl, pH 8.0, 0.5 mM EDTA, 120 mM NaCl, 0.2 mM PMSF, 0.5 mM DTT, and 0.02% Igepal. Following dialysis, glycerol was added to 5%, and the complex was stored at 4°C. For chromatin reconstitution, the core histones and Nap1p were preincubated at 37°C for 15 min. Then the appropriate amount of core histones and Nap1p was combined with DNA under the same conditions. Reconstitution was allowed to proceed at 30°C for 45 min, and reconstituted chromatin templates were stored at 4°C.
Nucleosome Footprinting by Micrococcal Nuclease Digestion of Reconstituted Chromatin and Multiple-round Primer Extension-Reconstituted chromatin (3 g of DNA) was digested at 37°C in 200 l at 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 50 mM NaCl, and 4 mM CaCl 2 . Micro-coccal nuclease (Worthington) was added to 1.0 unit/ml and digested for 2 and 4 min. Free DNA (3 g) was incubated under the same conditions except that micrococcal nuclease was added to 0.35 unit/ml and digested for 1 and 2 min. Aliquots of 100 l were transferred to tubes containing 12 l of 0.5 M EDTA, 5 l of 2.5 mg/ml proteinase K (Sigma) and incubated at 37°C for 10 min. The DNA fragments were extracted, precipitated, washed, and dried. The DNA was resuspended in H 2 O, and the concentration was determined by measuring absorbance (A 260 ). 75 nmol of a 32 P-end-labeled primer was added to 250 ng of DNA. The sequence of the primers used were 5Ј-CCACGTGTGAGTGCCAAG-3Ј (mn2), 5Ј-ATGAGGAAAGGAGAGTAGGGTGGTATA-3Ј (ml58), and 5Ј-GAATTGTCGAAATGAAACG-3Ј (mn5). The primers bind where indicated in Fig. 1A. Multiple-round primer extension reactions were carried out in 50-l reactions containing 0.1 mM dNTPs, 1.5 mM MgCl 2 , 10 mM HEPES-KOH, pH 8.4, and 50 mM KCl. Five cycles of primer extension were done. Extension products were extracted, precipitated, resuspended in formamide loading buffer, and resolved on a 6% sequencing gel. The gel was dried and exposed to a PhosphorImager screen (Amersham Biosciences).
One-dimensional Topological Analysis-To analyze the degree of reconstitution, 5 g of DNA was relaxed with 10 units of topoisomerase I (MBI Fermentas) in 50 mM Tris-HCl, pH 7.5, 50 mM KCl, 10 mM MgCl 2 , 1 mM DTT, 0.5 mM EDTA, and 30 g/ml BSA for 40 min at 37°C in a 100-l reaction. Following initial relaxation, 10 units more of topoisomerase I was added. Reconstitution reactions contained 500 ng of relaxed DNA, 3 mM MgCl 2 , and octamer⅐Nap1p complex in reconstitution buffer in a 19-l reaction volume. Reconstitution was allowed to proceed for 2 h at 30°C. The reaction was stopped by the addition of 100 l of STOP (20 mM EDTA, 0.1% SDS, 200 mM NaCl, and 0.25 mg/ml glycogen) and 12.5 g of proteinase K and incubated at 37°C for 20 min. The DNA was extracted and precipitated as before. The DNA was divided and run on two 1% TBE (90 mM Tris, 90 mM boric acid, 2 mM EDTA)-agarose gels, one of which contained 1.8 g/ml chloroquine, and visualized with ethidium bromide.
Micrococcal Nuclease Digestion of Reconstituted Templates-Two micrograms of DNA, either naked or reconstituted chromatin (chromatin: DNA, w/w ratio of 1:1), was digested in a 100-l reaction volume in reconstitution buffer and 5 mM CaCl 2 . Naked DNA was digested with 0.002, 0.004, 0.008, 0.016, and 0.024 unit of MNase, and chromatin was digested with 0.005, 0.02, 0.05, 0.10, and 0.20 unit of MNase at 37°C for 10 min. DNA was purified as before and resolved on a 10-cm 1% agarose-TBE gel. The gel was stained with ethidium bromide and digitally scanned.
Pho4p Purification-Purification of Pho4p (expression vector T7-PHO4, a gift from the O'Shea laboratory) was adapted from Kaffman et al. (31) and is as follows. Pho4p was expressed in BL21(DE3) Escherichia coli with 1 mM isopropyl-1-thio-␤-D-galactopyranoside at an A 600 of 0.7 for 3 h at 30°C in 1 liter of LB broth containing 100 g/ml ampicillin. The cells were washed once in 30 ml of RB0.1 buffer (20 mM Tris-OAc, pH 7.9, 1 mM EDTA, 10% glycerol, 1 mM DTT, 1 mM PMSF, 10 mM ␤-mercaptoethanol, and 100 mM potassium acetate), resuspended in 5 ml of RB0.1, and sonicated, and the debris was pelleted at 10,000 rpm for 20 min at 4°C in a Sorvall SS-34 rotor. The lysate was loaded at 0.7 ml/min onto a 10-ml DEAE-Sepharose FF (Amersham Biosciences) column equilibrated in RB0.1 column, and the column was washed with RB0.1. Proteins were eluted with RB1.0 (same as RB0.1, but with 1 M potassium acetate), and the peak Pho4p protein fractions were determined by SDS-PAGE and Coomassie Blue staining. The peak fractions were pooled and dialyzed into RB0.3 for 6 h at 4°C then loaded at 0.4 ml/min onto a 5-ml SP-Sepharose FF (Amersham Biosciences) column equilibrated in RB0.3. The column was washed with RB0.3, then the proteins were eluted with RB1.0. Peak fractions were determined as before. The Pho4p SP-Sepharose pool was dialyzed against RB0.1 then loaded onto a Mono-Q column equilibrated in the same buffer. The column was washed with RB0.1, and proteins were eluted with an RB0.1 to RB1.0 linear gradient at 0.5 ml/min over 20 column volumes. Peak fractions were determined as described above, and protein concentrations were quantitated using the Coomassie Plus Protein Assay kit (Pierce) and stored in liquid nitrogen.
Pho2p Purification-Purification of Pho2p was adapted from Brazas and Stillman (32). The Pho2-His expression vector (M2025, a gift from David Stillman) was transformed into BL21(pLysS) E. coli. 250 ml of LB was supplemented with 34 g/ml chloramphenicol and 100 g/ml ampicillin. Pho2-His was induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside at an A 600 of 0.8 for 3 h at 30°C. Cells were washed once in His binding buffer (20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 5 mM imidazole, and 1 mM PMSF). Cells were pelleted and resuspended in 10 ml of His binding buffer. Lysozyme was added to 1 mg/ml, and the cells were incubated on ice for 15 min then sonicated. Igepal (Sigma) was added to 0.1%, and the lysate was clarified by centrifugation at 12,000 ϫ g for 30 min at 4°C. The lysate was applied to a 1-ml nickel-nitrilotriacetic acid-agarose column (Qiagen) equilibrated in His binding buffer. The column was washed with His binding buffer followed by His binding buffer plus 60 mM imidazole. Pho2-His was eluted with a linear 60 to 600 mM imidazole gradient in His binding buffer. The peak fractions were determined by Coomassie-stained SDS-PAGE gel and dialyzed for 4 h at 4°C against storage buffer (20 mM HEPES-KOH, pH 7.9, 10% glycerol, 0.5 mM EDTA, 0.5 mM DTT, 100 mM potassium acetate, 1 mM PMSF, 10 mM ␤-mercaptoethanol, and 1 mM benzamidine). Protein concentrations were determined using the BCA Protein Assay kit (Pierce).
Electrophoretic Mobility Shift Assays-For the competition EMSA both wild type and mutant probes were 24-bp double-stranded oligomers. The EMSA gel showing Pho4p and Pho2p ternary complex formation was run with the indicated amounts of Pho4p and Pho2p, ( Pho4p and Pho2p Footprinting by Deoxyribonuclease I Digestion and Multiround Primer Extension-To verify reconstitution, chromatin templates were digested with MNase and analyzed on a 1.2% TBEagarose gel for production of a ϳ170-bp ladder. Naked or reconstituted chromatin DNA (500 ng), 100 ng of poly(dI-dC), purified Pho4p and Pho2p (amounts indicated in Fig. 4 legend), 1% polyethylene glycol 3350, 50 mM HEPES-KOH, pH 7.6, 100 mM potassium glutamate, 5 mM EDTA, 10 mM magnesium acetate, 2.5 mM DTT, and 10% glycerol in a 20-l reaction volume were incubated at 30°C for 15 min. DNase I (Worthington) was added (2 units/ml for naked DNA, 16 units/ml for chromatin) and DNA was digested at 30°C for 2 min. The reaction was stopped, and the DNA was purified as described above (one-dimensional topological analysis). Twenty cycles of primer extension were conducted in 50 l of 50 mM KCl; 10 mM Hepes-KOH, pH 8.4; 0.75 mM MgCl 2 ; 0.05 mM dATP, dTTP, dGTP, and dCTP; 125 ng of DNA; 1 unit of Taq polymerase; and 0.1 pmol of [␥-32 P]ATP-end-labeled primer (5Ј-ATTTGGCATGTGCGATCTCTT-3Ј). DNA products were extracted and precipitated. Products were resuspended in TBE-formamide dye and electrophoresed on a 5% polyacrylamide-urea gel next to a dideoxy sequence generated by using the same primer and template. The gel was then dried and visualized using a PhosphorImager screen.
In Vitro Chromatin Remodeling-One microgram of reconstituted chromatin was digested with micrococcal nuclease (MNase, Worthington) in a 100-l reaction volume following a 30-min incubation at 30°C with various components as indicated in the Fig. 5 legend. Reactions contained 1 g of DNA, 6 mM CaCl 2 , proteins, and ATP as described in the figure legend, and were brought to the final volume with reconstitution buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 150 mM NaCl, 100 g/ml BSA, 0.02% Igepal). Naked DNA was digested with 0.004, 0.016, or 0.032 unit of MNase and chromatin was digested with 0.05, 0.25, and 0.75 unit of micrococcal nuclease. Reactions were stopped by the addition of 100 l of STOP (20 mM EDTA, 200 mM NaCl, 0.1% SDS, and 250 g/ml glycogen) and 5 l of 2.5 mg/ml proteinase K, incubated at 37°C for 20 min, extracted with phenol/chloroform/isoamyl alcohol, extracted with chloroform/isoamyl alcohol, and precipitated with ethanol. Two hundred nanograms of DNA was run on a 20-cm, 1.5% agarose gel in 1ϫ TBE, Southern blotted (Turboblotter, Schleicher and Schuell), and probed with either a 169-bp MfeI to BstEII fragment representing nucleosome Ϫ2 or a 110-bp ApoI to BamHI fragment representing nucleosome ϩ1 (Fig. 1A). Before reprobing, blots were stripped by boiling in 0.1% SDS and cooling to room temperature.
Histone Acetyl Transferase Assays-Two micrograms of core histones, free or assembled into chromatin, were incubated with yeast WCE from YS27 cells at 30°C for 60 min. The reaction mixture (100 l) contained 50 mM Tris-HCl, pH 8.0, 5% glycerol, 0.1 mM EDTA, 50 mM KCl, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, 10 mM sodium butyrate, 0.2 Ci of [ 14 C]acetyl-CoA, and 4.5 M acetyl-CoA. After incubation, the proteins were resolved by SDS-PAGE (15% acrylamide; 30:0.8, acrylamide:bis). The gels were Coomassie Blue-stained, dried, and exposed to a PhosphorImager screen. Alternatively, the incubations contained 0.5 g of core histones in chromatin and only unlabeled acetyl-CoA. After SDS-PAGE, the proteins were transferred to a polyvinylidene difluoride membrane, and the acetylation state was determined by Western blot using anti-unacetylated H4 antibody (Upstate Biotechnology). The same blotted membrane was Western-blotted again using anti-acetylated H4 antibody (Upstate Biotechnology).

Primer Extension Footprinting Analysis of Nucleosome
Positioning on the PHO5 Promoter-Nucleosomes are positioned on the PHO5 promoter in vivo. To determine chromatin structure on the reconstituted PHO5 promoter, we used micrococcal nuclease digestion in conjunction with multiround primer extension analysis to footprint the nucleosomes (34,35). The primers used for this analysis are shown in Fig. 1A.
Nucleosome Ϫ1-We observed cleavage protection over the TATA-box and RNA start sites (Fig. 1, B and C). The protection occurred between approximately ϩ7 and Ϫ123 bp, relative to the ATG (translational start codon) and corresponds to nucleosome Ϫ1. At the downstream end of nucleosome Ϫ1, strong protection ends at ϩ7, with slight protection seen at ϩ10 (Fig.  1B). No cleavage protection or enhancement is seen from ϩ19 to ϩ55. However, there is some ambiguity as to where protection ends due to lack of cleavage between ϩ10 and ϩ19 on both free DNA and chromatin samples. The upstream edge of nucleosome Ϫ1 was inferred from the protection at Ϫ123 (Fig.  1C). From Ϫ129 through Ϫ150 there are several sites of equal or stronger cleavage, defining the linker region between nucleosome Ϫ1 and Ϫ2. The lack of cleavage between Ϫ123 and Ϫ129 on both free DNA and the chromatin precludes defining the downstream edge of nucleosome Ϫ1 with greater precision. Thus, we place nucleosome Ϫ1 between ϩ7 (Ϯ5) bp to Ϫ123 (Ϯ9) bp.
The clear footprint observed with primer ml58 and primer mn5 indicates a strong translational setting for nucleosome Ϫ1. In addition, the length of DNA associated with nucleosome Ϫ1 protection is very close to the 145 bp normally associated with a core particle. This translational setting places both the initiation region and the TATA element within nucleosome Ϫ1, ϳ30 bp from each edge. Note the cleavage protection over 5 nt of the TATA element, indicating a lack of accessibility for TATA-binding protein (TBP) binding.
A comparison of nuclease cleavage patterns on chromatin and free DNA templates in the region from ϩ19 to ϩ55 and Ϫ150 to Ϫ129 indicates a complete lack of protection. This pattern is consistent with these regions comprising stretches of linker DNA ϳ30 bp long between nucleosomes Ϫ1 and ϩ1 and nucleosomes Ϫ1 and Ϫ2.
Nucleosome Ϫ2-Protection from micrococcal nuclease cleavage is seen from Ϫ150 (Ϯ10 bp) through Ϫ311 (Ϯ6 bp), corresponding to nucleosome Ϫ2 (Fig. 1, C and D). The nuclease protection is nearly complete (Fig. 1C, near the top), which is unusual for a nucleosome. Placing the downstream edge of this nucleosome at around Ϫ150 is based upon the clear protection through this site (Fig. 1C). Setting the upstream edge at around Ϫ311 is based on the protection from Ϫ289 to Ϫ311 and the lack of protection seen from Ϫ317 to Ϫ370 (see Fig. 1D). The slightly greater than 145-bp area of protection generally associated with a nucleosome core particle suggests multiple translational or rotational positions. As seen in vivo, nucleosome Ϫ2 contains UASp2 (the Pho4p binding site) near the nucleosome dyad and a Pho2p binding site closer to the upstream edge.
Enlarged Linker-An important feature of the repressed PHO5 promoter chromatin structure is the nucleosome-free region between nucleosomes Ϫ2 and Ϫ3, which includes UASp1 (the distal Pho4p binding site). Nuclease protection beginning from approximately Ϫ370 indicates that nucleosome Ϫ3 is positioned upstream of UASp1 (Fig. 1D). In addition, we detect a region lacking protection between approximately Ϫ366 and Ϫ317 or around 50 bp in length. This linker is larger than the ϳ30-bp length of the other linkers, corresponding to the situation in vivo (11). Previously, this enlarged linker (referred to as HS2) was mapped to a region from approximately Ϫ410 to Ϫ315 in vivo by indirect end-labeling and restriction endonuclease cleavage analysis (11). The micrococcal nuclease protection suggests that nucleosome Ϫ3 is located farther downstream on the reconstituted templates than in vivo. However, what is most important is that an enlarged linker is formed that leaves the Pho4p binding site in UASp1 accessible.

The PHO5 Promoter Is Strongly Repressed by Nucleosome
Formation in Vitro-We have reconstituted plasmids with a PHO5 promoter construct containing 526 bp of the promoter upstream of a G-less cassette (pPHO5-Gless). Nucleosomes were formed at the ratios (w/w) of core histones to DNA indicated in Fig. 2. At the histone to DNA ratio of 1.0, nucleosomes were formed at physiological density as determined by topological assay ( Fig. 2A). Micrococcal nuclease analysis verified that complete nucleosomes were formed with uniform spacing of ϳ160 to 170 bp (Fig. 2B). Transcription from the chromatin templates was strongly repressed at core histone to DNA ratios of 0.43 and 0.57 and essentially completely repressed at a ratio of 1.0 (Fig. 2C).
Pho4p and Pho2p Bind to Both Free and Reconstituted Chromatin DNA-Pho4p and Pho2p are both required for the chromatin transition and transcriptional activation of PHO5 upon phosphate depletion (15). Therefore, it was necessary to purify both proteins to study the mechanism of chromatin structure transition and transcriptional activation in vitro (Fig. 3A). Both proteins bind a 106-bp UASp1 DNA probe (TthIII to MfeI, Fig.   FIG. 2. Strong repression of PHO5 transcription by reconstituted nucleosomes. Plasmid templates, containing both UASs and the core promoter were reconstituted. A, one-dimensional topological analysis. Plasmid DNA was reconstituted at the core histone to DNA ratio (w/w) indicated at the top, purified, and resolved on 1% agarose gels in the absence (ϪChl) or presence (ϩChl) of chloroquine. Lanes S and R are supercoiled and relaxed markers, respectively. In the right margin, the positions of relaxed (Ir) and supercoiled (Is) closed circular DNA are indicated. B, micrococcal nuclease digestion analysis was used to verify that complete nucleosomes with uniform spacing were formed. Lane 1 contains undigested DNA. Free DNA (lanes 2-4) and reconstituted chromatin (lanes 5-7) were digested with increasing concentrations of nuclease, and the products were resolved on an agarose gel. The digestion products corresponding to a mono-, di-, and trinucleosome are indicated as such in the right margin. Lanes M, 100-bp DNA size markers. C, free DNA (lane 1) and chromatin templates reconstituted at the core histone to DNA ratio (w/w) indicated at the top (lanes 2-9) were transcribed using yeast whole cell extract. The PHO5 transcripts are indicated at the left of the figure. These results were found to be reproducible through several repetitions of this experiment. 1A) independently and form a ternary complex when combined (Fig. 3B) (16). Because Pho2p, a homeodomain protein, does not have a well-defined binding site, we tested the binding specificity of our purified Pho2p to UASp2. An unlabeled wild type probe, identical to the labeled probe (MfeI to BstEII, Fig. 1A), successfully competed for Pho2p binding (lanes 1-8, Fig. 3C). However, a probe in which the order of the base pairs in the AT-rich region required for Pho2p binding had been randomized did not compete for Pho2p binding (lanes 9 -14, Fig. 3C).
Pho4p and Pho2p protect both naked DNA and chromatin from deoxyribonuclease I (DNase I) cleavage (Fig. 4, A-C). Pho4p protected UASp1 and UASp2 on free DNA and chromatin (vertical lines, Fig. 4A; horizontal lines in Fig. 4, B and C). Even though Pho2p can bind DNA in the absence of Pho4p in EMSAs (lanes 5 and 6, Fig. 3B), the Pho2p protection (asterisks in Fig. 4, A-C) was stronger with Pho4p present on both free DNA and chromatin. Protection is observed on chromatin DNA at the same Pho4p and Pho2p concentrations that produced protection on free DNA. However, protection on chromatin is more dependent on both proteins being present than protection on free DNA (particularly apparent in Fig. 4, B and C). Pho2p protected the binding site near UASp2 more strongly on chromatin than on free DNA. Pho4p protected a larger region on UASp1 on chromatin than seen on free DNA but produced none of the adjacent heightened nuclease sensitivity. Increasing the concentration of the activator proteins past that shown did not result in more complete protection. These higher concentrations caused reproducible, but nonspecific, alterations in the cleavage pattern, presumably due to aggregation of proteins on the DNA (results not shown).

A 0.3 M SP Nuclear Extract Fraction Remodels the PHO5 Chromatin Structure in a Pho2p-, Pho4p-, and ATP-dependent
Manner-Chromatin templates were incubated with micrococcal nuclease, and the DNA fragments were resolved by agarose gel electrophoresis and detection through Southern blot analysis. The chromatin structure over the promoter region was determined using the MfeI to BstEII probe encompassing nucleosome Ϫ2 (Fig. 1A). Pho4 and Pho2p do not remodel chromatin on their own (lanes 10, 11, and 12, Fig. 5A, and the corresponding tracing in Fig. 5B). A nuclear extract fraction from wild type yeast cells was shown to remodel the PHO5 chromatin structure of isolated minichromosomes (29). On our defined chromatin templates, we found that nucleosome remodeling was mediated by this nuclear extract fraction and that it, too, was dependent on the presence of Pho4p, Pho2p, and ATP (compare lanes 7-9 with lanes 13-15, Fig. 5A, and the corresponding tracings in Fig. 5B). To ascertain the range of the remodeling that occurs in the transcribed region, these blots were stripped and hybridized with the ApoI to BamHI (110 bp) probe (Fig. 1A) complementary to DNA in nucleosome ϩ1. Much less remodeling occurs over the transcribed region (compare lanes 13-15 with lanes 16 -18, Fig. 5A, and the corresponding tracings in Fig. 5B).
Acetyl-CoA Is Required for Activation of Transcription-The NE fraction containing chromatin remodeling activity does not support transcription, so a WCE from a pho4⌬ was used. The addition of Pho4p, Pho2p, and the 0.3 M SP NE fraction, sufficient for chromatin remodeling, to transcription reactions containing reconstituted chromatin did not result in transcriptional activation (results not shown). It is important to note that the formation of nucleosomes does not render the DNA irreversibly transcriptionally incompetent, because transcription from the reconstituted template can be restored by proteinase K digestion and organic extraction of the DNA (data not shown). To test whether histone acetylation was required in addition to nucleosome remodeling, acetyl-CoA was added to the transcription reactions. Addition of Pho4p activates transcription approximately 3-to 4-fold on free DNA (Fig. 6, compare lanes 2-6 with lanes 7-11). Acetyl-CoA had no stimulatory effect and even had a slight inhibitory effect on transcription from free DNA templates at higher levels ( Fig. 6, lanes 1-11). Interestingly, on chromatin the addition of acetyl-CoA was able to fulfill the additional requirement needed to relieve transcriptional repression and to allow transcriptional activation from the chromatin templates (Fig. 6, lanes 13-23). Pho4p activated transcription further on chromatin (lanes 19 -23), but the stimulatory effect of acetyl-CoA was not dependent on Pho4p (lanes 14 -18). Pho4p was not able to activate transcription from chromatin without acetyl-CoA (data not shown).
Addition of Acetyl-CoA to the Yeast Whole Cell Extract Results in Core Histone Acetylation on Chromatin Templates-Acetyl-CoA stimulates transcription from chromatin, but not from free DNA, suggesting that this cofactor is functioning through histone acetylation. To directly determine that the core histones are acetylated in our transcription reactions, we added [ 14 C]acetyl-CoA. The primary acetyl group acceptors were histones H2A and H4, which represents an acetylation pattern similar to that of NuA4 (Fig. 7A). Even in the presence of WCE, no other protein was detectably labeled. In addition, the level and pattern of the HAT activity is independent of ATP-dependent remodeling activity and Pho4p activation (right-hand two lanes, Fig. 7A). Western blot analysis of core histone acetylation by the activity in our whole cell transcription extract was conducted, as well (Fig. 7B). The results demonstrate that histone H4 is unacetylated in the absence of acetyl-CoA. Interestingly, the H4 acetylation appears to be quantitative, because the histone H4 is no longer recognized by antibody directed against the unacetylated H4 tail. NuA4 has been shown to fully acetylate histone H4 (forms tetra-acetylated H4) by transferring an acetyl group to the lysines at positions 5, 8, 12, and 16 (36). These antibodies were raised against an H4 amino-terminal tail peptide, so it is conceivable that they would not recognize the fully acetylated histone H4 tail.

Reconstitution of the Repressed State-The
Hörz and Bergman laboratories independently identified an array of at least four translationally positioned nucleosomes on the repressed PHO5 promoter (11,37). Although extensive mutant searches have been conducted, to date no mutation has been found that disrupts nucleosome positioning. Therefore, we hypothesized that nucleosomes are positioned primarily through intrinsic  13-15 and lanes 16 -18). A, Southern blot analysis probing for nucleosomal protection over the UASp2 (nucleosome Ϫ2) and transcribed (nucleosome ϩ1) regions. Following micrococcal nuclease digestion, purification, and resolution on an agarose gel, the DNA was blotted to a membrane and probed with a fragment that binds the nucleosome Ϫ2 region (lanes 1-15) or the nucleosome ϩ1 region (lanes 16 -18). A ladder indicates no remodeling. A smear indicates remodeled chromatin. B, graphical representation of the distribution of counts in the lanes containing the most digested samples (lanes 3, 6, 9, 12, 15, and  18). A series of density peaks indicates no remodeling. A lack of peaks (a broad smear) indicates remodeled chromatin.
properties of the promoter DNA sequence. To test this hypothesis, we determined the micrococcal nuclease cleavage pattern on chromatin and free DNA templates to define the nucleosome positions. We observed positioning of a nucleosome over the TATA-box and RNA start sites (nucleosome Ϫ1) and another nucleosome over the downstream Pho2p binding site and UASp2 (nucleosome Ϫ2). Moreover, we found that an enlarged linker region between nucleosomes Ϫ2 and Ϫ3 containing UASp1 is established on our reconstituted templates. Thus, the nucleosome positioning on our reconstituted chromatin templates is consistent with that seen in vivo.
The reconstituted chromatin templates were formed using purified yeast core histones and Nap1p, so no other DNAbinding proteins are present. Hence, the nucleosome positioning on our reconstituted templates is directed primarily by sequence-dependent intrinsic properties of the DNA. From the results of a series of deletion experiments, Fascher et al. (21) concluded that intrinsic properties of the promoter DNA make "an essential contribution to the chromatin organization at the PHO5 promoter." Therefore, the mechanism of nucleosome positioning in vivo is likely to be much the same as that driving the positioning on our reconstituted templates. Nucleosomes reconstituted on tandem repeats of the sea urchin 5 S rRNA gene formed arrays of positioned nucleosomes (38). In addition, a 200-bp fragment from the Drosophila Adh promoter will translationally position a single nucleosome correctly in vitro (39). However, the PHO5 promoter and the murine mammary tumor virus long terminal repeat promoter are the only singlecopy gene promoters thus far determined to contain an array of nucleosomes that are translationally positioned primarily through histone-DNA interactions (40).
There is one minor but interesting difference between the nucleosome positioning of the reconstituted and in vivo chromatin. Although the endpoints of the enlarged linker have not been mapped to the base pair in vivo, the evidence available suggests that nucleosome Ϫ3 is positioned farther upstream in vivo than in vitro (11). In vivo, the enlarged linker was mapped to approximately Ϫ410 through Ϫ340 (70 bp) by restriction endonuclease accessibility and low resolution indirect end labeling. A closer analysis of a higher resolution indirect endlabeling experiment (11) suggests boundaries of Ϫ393 to Ϫ313 (70 -80 bp). The enlarged linker on our reconstituted templates is formed from approximately Ϫ366 to Ϫ317 (50 -60 bp). Therefore, the downstream end point of the enlarged linker on our reconstituted templates corresponds well with that seen in vivo. However, nucleosome Ϫ3 produces clear protection through Ϫ370. Therefore, the location of nucleosome Ϫ3 appears to be shifted 20 -30 bp downstream in vitro. Similarly, on partially purified minichromosomes, nucleosome Ϫ3 is located farther downstream, suggesting that the positioning factor is being lost during preparation (29). This positioning factor is absent on our reconstituted templates, as well.  13-17 versus lanes 2-6). The addition of Pho4p (0.38 g) stimulated transcription from both free DNA and chromatin (lanes 7 and 18). The -fold activation is represented as transcript levels relative to those obtained from free DNA or chromatin templates in the absence of both acetyl-CoA and Pho4p (lanes 2 and 13). These results were found to be reproducible through several repetitions of this experiment. Pho4p is exported from the nucleus and is not bound to UASp1 under repressing conditions (12,31). In addition, the PHO5 chromatin structure, as analyzed by low resolution indirect end labeling in pho2 and in pho4 cells, appears to be the same as that of wild type cells (15). Finally, deletion of both the Pho2p and Pho4p binding sites at UASp1 does not significantly affect the PHO5 promoter chromatin structure (21). However, a 20-or 30-bp downstream shift of nucleosome Ϫ3 would not have been resolved in these studies. Therefore, although Pho4p is unlikely to be involved in positioning nucleosome Ϫ3 farther upstream, such a role for Pho2p or another, unidentified protein cannot be excluded at this time.
We have succeeded in reconstituting many aspects of the in vivo chromatin structure on the repressed PHO5 promoter. Nucleosome formation has long been known to repress transcription in vitro (42,43). However, the strong repression of transcription from the PHO5 promoter on our chromatin templates at sub-saturating densities of nucleosomes suggests that the PHO5 core promoter has a high affinity for nucleosome formation, positioning a nucleosome over the core promoter. Primer extension footprint analysis indicates this is the case.
Reconstitution of Transcriptional Activation-We began to study the mechanism of PHO5 activation by adding back various activities and cofactors to determine their effects on the transcription level from our well-characterized chromatin template. We have found that recombinant Pho4p and Pho2p can bind to both sets of PHO5 promoter regulatory elements (UASps) on free DNA and assembled into chromatin. This finding reiterates the importance of regulation of Pho4p function through phosphorylation and nuclear localization (12,31). Differences in nuclease protection patterns indicate that these transcription factors bind their cognate sites and interact with each other differently on free DNA versus chromatin. For example, Pho4p produces enhanced cleavages downstream of the UASps in free DNA but protects these regions in chromatin. In addition, although Pho2p cleavage protection at both binding sites is enhanced by Pho4p, the effect at the site upstream of UASp2 is much more apparent on chromatin than on free DNA. These findings are consistent with an important role for Pho4p-Pho2p interactions in PHO5 activation and indicate that chromatin structure participates in these interactions.
In our system, Pho4p is sufficient for transcription activation on naked DNA, presumably through increased recruitment of the general transcriptional factors and RNA polymerase II. Pho4p physically interacts with TFIIB, TFIIE␤, and the TATAbinding protein (TBP) (18). However, Pho4p alone could not counteract nucleosome repression.
We have shown that Pho4p and Pho2p can bind to the promoter in the presence of nucleosomes. Their ability to bind both in the enlarged linker and on nucleosome Ϫ2 suggests that their binding is not ordered but, rather, that they can occupy both UASps simultaneously.
The presence of Mg 2ϩ in the transcription buffer might promote compaction of the templates (44). However, although compaction of the DNA may occur thereby preventing core promoter access by the general transcription machinery, it does not prevent all access to the promoter, because the transcription factors Pho4p and Pho2p have access to the DNA at the same Mg 2ϩ concentrations used for in vitro transcription experiments (Fig. 4).
Pho4p and Pho2p binding is not sufficient for nucleosome remodeling. In the presence of a nuclear extract fraction and ATP, these transcriptional activators can remodel nucleosomes Ϫ2. Whether Pho4p and Pho2p function in remodeling through recruitment or nucleosome restructuring or sliding remains to be determined. It is clear that remodeling is not dependent on histone acetylation, because it occurs without the addition of acetyl-CoA. Finally, transcriptional activation is still not observed (data not shown). Thus, remodeling is not sufficient, and there is another barrier to transcription.
Core histone tail acetylation, which primarily affects chromatin compaction, is a potential additional requirement for transcriptional activation (45). To test this idea, we added acetyl-CoA to the reactions and found that nucleosomal repression was relieved. Furthermore, we determined the core histones, particularly H4, were highly acetylated under these conditions (Fig. 7). The level of HAT activity in our WCE is quite high and was not dependent on recruitment. Consistent with this result, the histone acetylation state on the PHO5 promoter in vivo has not been found to change significantly between the repressed and activated states (23). Because transcription is dependent on the presence of acetyl-CoA and only the core histones are detectably acetylated, core histone acetylation is likely to be a prerequisite for transcription from the chromatin-assembled PHO5 promoter. A requirement for acetylation of other substrates after transcriptional initiation cannot be ruled out. However, the lack of acetyl-CoA stimulation of transcription from free DNA templates argues against this idea.
The core histone acetylation specificity in the yeast WCE resembles that of the NuA4 complex (46). This acetylation pattern suggests that this complex is the most abundant or active HAT activity in the extract and may be primarily responsible for counteracting chromatin repression of basal and activated transcription in the presence of acetyl-CoA. Supporting this idea, we find that addition of NuA4, but none of the other three complexes, stimulated transcription further (results not shown).
Although we have not shown directly that transcriptional activation requires chromatin remodeling, the simplest interpretation of our results is that maximal activation from the reconstituted PHO5 promoter requires Pho4p, Pho2p, remodeling activity, and histone acetylation. Our results suggest that the role of histone acetylation and nucleosome remodeling lie downstream of Pho4p and Pho2p binding. Possibilities include allowing recruitment of the basal transcriptional machinery, initiation, and elongation. Current experimentation in our laboratory is designed to differentiate between these possibilities.
The events in PHO5 regulation are separable. The nucleosome positioning and the transcriptionally repressed state can be reproduced with core histones and PHO5 promoter DNA alone. Pho4p and Pho2p binding can occur without remodeling. Remodeling appears to be dependent on binding of these factors but is not dependent on histone acetylation. In fact, we have tested our yeast nuclear extract fraction (NE(S0.3)) and found it to be devoid of HAT activity (data not shown). Finally, acetylation is not dependent on either Pho4p binding or ATP-dependent remodeling. The ability to separate these events greatly facilitates their mechanistic dissection. The system we have designed to study chromatin remodeling is unique in that it uses only purified components for reconstitution. This allows the study of PHO5 transcriptional regulation in a well-defined environment. In addition, our experimental system facilitates combined transcription, nucleosome remodeling, and histone acetylation studies, giving a more complete picture of the processes of PHO5 transcriptional activation.