Binding of the RNA Polymerase I Transcription Complex to Its Promoter Can Modify Positioning of Downstream Nucleosomes Assembled in Vitro*

We have studied the reconstitution of chromatin-like structures in vitro, using purified RNA polymerase I transcription complexes and histone octamers. The plasmid construct used in these studies is a pUC8 de-rivative in which we have inserted an RNA polymerase I core promoter region of Acanthamoeba castellanii upstream of four repeats of the 5 S rDNA nucleosome positioning sequence (208 base pairs) from Lytechinus variegatus. When histone octamers were reconstituted onto the naked DNA template, the expected nucleosome positioning previously observed using tandem repeats of the same 208-base pair fragment was not obtained (as assayed by restriction enzyme digestion mapping of the inserted region of the plasmid). We show that the location of the RNA polymerase I core promoter region with regard to the tandemly repeated 208-base pair positioning sequence is a major determinant in the positioning of the histone octamers. Reconstituting first with the stalled transcription complex excluded octamers from the promoter region and recovered the expected nucleosome positioning downstream on the four repeats of the 5 S positioning sequence. The observed competition between histone octamers and the transcription complex for the promoter region suggests a great similarity with what has been reported from in vitro studies of RNA polymerase I1 and I11 transcription systems. We may be looking at a mechanism of

We have studied the reconstitution of chromatin-like structures in vitro, using purified RNA polymerase I transcription complexes and histone octamers. The plasmid construct used in these studies is a pUC8 derivative in which we have inserted an RNA polymerase I core promoter region of Acanthamoeba castellanii upstream of four repeats of the 5 S rDNA nucleosome positioning sequence (208 base pairs) from Lytechinus variegatus. When histone octamers were reconstituted onto the naked DNA template, the expected nucleosome positioning previously observed using tandem repeats of the same 208-base pair fragment was not obtained (as assayed by restriction enzyme digestion mapping of the inserted region of the plasmid). We show that the location of the RNA polymerase I core promoter region with regard to the tandemly repeated 208-base pair positioning sequence is a major determinant in the positioning of the histone octamers. Reconstituting first with the stalled transcription complex excluded octamers from the promoter region and recovered the expected nucleosome positioning downstream on the four repeats of the 5 S positioning sequence. The observed competition between histone octamers and the transcription complex for the promoter region suggests a great similarity with what has been reported from in vitro studies of RNA polymerase I1 and I11 transcription systems. We may be looking at a mechanism of regulation of transcription for the RNA polymerase I.
A major problem in the field of eukaryotic transcription concerns the role and behavior of nucleosomes occupying the transcribed regions (see van Holde et al. (1992) for a recent review). Most of the attempts to study this behavior have utilized linear templates (Lorch et al., 1987(Lorch et al., , 1988Losa and Brown, 1987;Izban and Luse, 1991, for example). One published study has employed a circular template, in order to investigate effects of DNA supercoiling (Pfaffle et al., 1990); a second recent study used tandemly repeated 5 S genes inserted in a closed circular plasmid (O'Neill et al., 1992). *This research was supported by National Institutes of Health Grants GM-12296 and GM-22580 and by National Institute of Environmental Health Sciences Grant 5 PO-1 ES 04766. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
However, both of these works utilized prokaryotic promoter and polymerases.
In order to be able to study in vitro transcription using a eukaryotic polymerase, we have constructed a circular template for the RNA polymerase I transcription complex from Acanthamoeba castellanii. The plasmid has been called pPol I 208-4 and contains an RNA polymerase I promoter region immediately upstream of four repeats of the 5 S rDNA nucleosome positioning sequence from Lytechinus uariegatus (Simpson and Stafford, 1983). This repeated sequence has been used extensively in our laboratory and elsewhere in studies on nucleosome positioning (Simpson and Stafford, 1983;Simpson, 1986;Hansen et al., 1989;Dong et al., 1990;Pennings et al., 1991; and see Thoma (1992) for a review).
In order to keep the promoter site free of a nucleosome, which would interfere with initiation, we have reconstituted the RNA polymerase I plus its transcription factors on the plasmid before the nucleosome structure is formed. The step dialysis method commonly used to deposit histone octamers onto their target sequences cannot be applied to this system. The ionic strength is a critical parameter for the stability of the polymerase and its factors and has to be kept close to the physiological 150 mM NaC1. When the reconstitution is carried out using the step dialysis method, the system is stepwise dialyzed from 2.2 M NaCl down to the required salt concentration; the high salt concentration encountered would dissociate the transcription complex. Therefore, the reconstitution of chromatin-like structure was carried out using polyglutamic acid (PGA)' as carrier for the deposition of the nucleosomes onto the DNA. The polyglutamic acid method has been studied by Retief (Retief et al., 1984) and has been shown to give consistent results in reconstituting the DNA and histones into a chromatin-like structure.
Before beginning transcription studies, we felt it important to first determine whether normal nucleosome positioning was obtained after this kind of reconstitution, particularly in the repeated region downstream from the promoter. The positioning was investigated by restriction enzyme digestions. Surprisingly, no defined positioning was observed when the circular template was reconstituted with histone octamers by this technique. However, we observed a rescue of positioning on the repeated 5 S genes when the reconstitution was performed in the presence of transcription factors TIF-IB and aUBF plus the RNA polymerase I stalled at position +8. No The abbreviations used are: PGA, polyglutamic acid; rDNA, ribosomal DNA; TIF-IB, transcription and initiation factor (also called SL-1); aUBF, upstream binding factor (also called SF-1); bp, base pairk); PCR, polymerase chain reaction; Pol I, RNA polymerase I. rescue was observed when the transcription factors were provided in the absence of the RNA polymerase I.
We further compared the reconstitution of templates that had been linearized by cleavage at different sites uersu covalently closed circular plasmid, on the assumption that the integrity of specific regions might be necessary for the nucleation of positioning. Indeed, we find that the polymerase I promoter contains a strong positioning sequence that competes with the 5 S rDNA signals and leads to randomization of nucleosome positioning.
The complete system has been demonstrated to be transcriptionally active. This activity will be the subject of a subsequent paper.

MATERIALS AND METHODS
Construction of the pPol I208-4 Plasmid-A 94-bp fragment (-75 to +19) from the RNA polymerase I promoter sequence from A.
castellanii was amplified by PCR. The template used for the amplification was the plasmid vector pEBHlO (Kowning et al., 1985) harboring a 200-bp sequence from the promoter region. The primers were designed to contain a PstI restriction site on the 5' terminus of the promoter and to incorporate a XbaI restriction site on the 3' terminus. A 4-bp extension at the end of the primer sequence was included to assure a satisfactory digestion of the PCR product with restriction endonucleases. After gel purification, the PCR product was cloned into pUC19 after restriction with PstI and XbaI and sequenced.
The 5 S ribosomal sea urchin sequence was obtained from the plasmid pAT153, PCR-amplified and sequenced, generating a 259-bp fragment. The primers were also designed to contain an XbaI restriction site at the 3' terminus and a PstI restriction site at the 5' terminus. The purified product was digested with PstI and XbaI digestion and cloned into pUC19. The polymerase I promoter region was then ligated to the 208-5 S sequence.
The PstIIPstI insert was PCR-amplified, and the monomer was excised with AuaI, which cuts once on each repeat and at the 3' end of the construct, opening up an AuaI insertion site.
Individual AuaI fragments were polymerized by ligation and cloned into the AuaI site. The asymmetry of the AuaI site allows only headto-tail ligation, forcing the orientation of the monomeric fragments.
Series of plasmids were prepared containing the RNA polymerase I promoter region upstream of up to 35 repeats of 208-5 S. These plasmids were called pPol I 208-n (where n is the number of repeats).
Preparation of Histone Octamers-Histone octamers were obtained from purified nucleosome monomers isolated from chicken erythrocytes according to method of Yager et al. (1989). Nuclei isolated from White Leghorn rooster blood were digested for 5 min with 14 units of micrococcal nuclease (Worthington Biochemical) per mg of DNA. The long chromatin was spun down at 6900 X g for 20 min, and the pellet was resuspended in 10 mM Tris-HC1,0.25 mM EDTA, and 0.35 M NaCI, pH 8.0. The removal of histone H1/H5 was accomplished by incubating the chromatin with 30 pg/ml carboxymethyl-Sephadex for 3 h at 4 "C, followed by centrifugation at 7700 X g for 30 min and dialysis of the supernatant against TE (10 mM Tris, 1 mM EDTA). A 4-min micrococcal nuclease digestion of the stripped long chromatin with 5 units of micrococcal nuclease per yl of DNA reduced the long chromatin to monomers, which were then concentrated in an Amicon The concentrated nucleosome monomer solution was made 2.2 M in NaCl and 0.1 M in potassium phosphate at pH 6.7 and chromatographed on a hydroxylapatite column equilibrated with the same buffer (Simon and Felsenfeld, 1979). The collected fractions were electrophoresed to check the histone content and stoichiometry. The concentration was determined from measurements of A2m (Stein, 1979).
Purification of the Transcription Factors and RNA Polymerase I-RNA polymerase I was purified by a modification of the method of Iida and Paule (1992). Instead of a whole cell extract, a 1.6-3.0 M ammonium sulfate fraction from a nuclear extract of A . castellunii (Zwick et al., 1991) was used as starting material. This was dialyzed to 100 mM KC1 in buffer A (20 mM Tris-HC1, pH 7.9, 0.2 mM EDTA, 1 mM dithiothreitol, 20% glycerol, 0.1 mM phenylmethanesulfonyl fluoride and loaded onto a 11 X 1.5-cm Bio-Rex 70 column in place of a phosphocellulose column, and the step-eluted fraction between 450 and 650 mM KC1 was collected. The DE52 column was step-XM-50. eluted (75-250 mM fractions) and the heparin-Sepharose column was likewise step-eluted (300-500 mM KCl) instead of running gradients. The TIF-IB/aUBF fraction utilized the 0.5-1.6 M ammonium sulfate fraction from the nuclear extract and was chromatographed through 14 X 1.5-cm DEAE fast flow (Pharmacia LKB Biotechnology Inc.) by loading it at 75 mM KC1 in buffer A and, after a wash in the same buffer, eluting with a linear gradient of KC1 in buffer A from 75-500 mM. The TIF-IB/aUBF-containing fractions (at approximately 300 mM KCl) were pooled, diluted to 150 mM KCl, and chromatographed through a 9.5 X 0.9-cm Bio-Rex 70 column using a KC1 gradient in buffer A from 150 to 900 mM. The fractions containing TIF-IB and aUBF, eluted at approximately 430 mM KCl, were dialyzed to 100 mM KC1 in buffer A and stored at -70 "C.
Reconstitution of Octamers onto the Plasmid-The pPol I 208-4 was reconstituted with histone octamers according to the method of Retief et al. (1984), which was modified according to the requirements of our system. PGA (Miles Laboratories) was used as a carrier for the deposition of the histones onto the circular template. The ionic strength of the medium was kept at 150 mM of monovalent cations in order to prevent displacement of the transcription factors and to keep the conditions optimal for transcription by RNA polymerase I.
Reconstitution was carried out using histone octamers purified from chicken erythrocytes as described above. Twenty-five yg of plasmid were first treated with 0.6 units of topoisomerase I (GIBCO-BRL) per pg of DNA for 90 min at 37 "C. The 2.2 M NaCl concentration of the octamer solution was reduced to 150 mM NaCl by diluting in TE; the octamers were then incubated at room temperature for 60 min in the presence of a 10 mg/ml solution of polyglutamic acid a t a ratio of PGAhistone of 2 1 (w/w). The relaxed plasmid was then added to the mixture. The final DNA concentration was 0.05 mg/ml. Input ratios of histone to DNA from 0.6 to 2.3 (g of histone/g of DNA) were tested in order to optimize the conditions of reconstitution. The 500-pl reaction mixture was reconstituted a t 37 "C, overnight, under constant shaking to avoid aggregation and precipitation. The reaction mixture was spun down on a IEC centra" centrifuge at top speed for 5 min to verify that no material had aggregated. The reconstitute was run on a 0.8% agarose gel in 0.5 X e-buffer (e-buffer contains 20 mM Tris HCI, 0.5 mM EDTA, and 15 mM NaOAc at pH 8.0) to monitor the formation of nucleoprotein complex (data not shown).
To generate a system with the potential for transcriptional activity, the appropriate amount of partially purified transcription initiation factor TIF-IB, upstream binding factor (aUBF), and RNA polymerase I were incubated for 15 min a t 25 "C, in 500 pl of final volume, in the presence of 12.5 pg of pPol I 208-4 and 0.5 mM each of ATP and GTP, before reconstitution with histone. The transcription complex will bind to the promoter region, start to transcribe, and stop at position +8 because of lack of CTP needed a t +8, making the complex more stable and less likely to fall off the DNA template. The system was then reconstituted with histone octamers plus PGA at an input ratio of 2.05 histone to DNA and 2 to 1 PGA to histone, according to the protocol previously defined. The same protocol was used for reconstituting with histone octamers, TIF-IB and aUBF, in the absence of polymerase I. Sedimentation Velocity Analysis-The plasmid and histone complexes were submitted to sedimentation velocity analysis on a Beckman model E analytical ultracentrifuge to verify the homogeneity of the system and to monitor the efficiency of the reconstitution. The runs were performed utilizing 12-mm double sector cells in a fourhole AN-F rotor. The temperature was kept constant within 0.1 "C. The solutions used for the sedimentation velocity studies had an AZGS = 0.8-1.0. The rotor speed, in different runs, was between 18,000 and 22,000 rpm. The scans were analyzed by the method of van Holde and Weischet (1978) using the Ultrascan ultracentrifuge data collection and analysis program. All data were corrected to standard conditions.
Micrococcal Nuclease Digestion-Micrococcal nuclease digestions of reconstitutes were performed in 100-pl volumes, using 5 pg of chromatin, a t 0, 25, and 50 units of micrococcal nuclease per pg of DNA. The reaction mixtures were incubated for 30 s a t 37 "C and then stopped by making the reaction 40 mM EGTA. The products were treated with 20 pl of 10 mg/ml proteinase K for 1 h a t 37 "C and then phenol/chloroform-extracted and ethanol-precipitated. The final products were resuspended in 10 pl of H20. The material was run in a 1.5% agarose gel in 1 X Tris/borate/EDTA for 3 h and 30 min a t 5 V/cm, ethidium bromide-stained, and photographed.
Restriction Digestion-To attempt to map the positions of the nucleosomes on the insert region, cleavages with PstI, XbaI, and EcoRI were performed. All digestions were performed under the same low Mg2' buffer conditions, whether naked or reconstituted, circular or linear DNA was used. Amounts of 0.5-1 pg of the different DNA templates were digested for 60 min a t 37 "C with EcoRI, PstI, AuaI, or XbaI a t 10 unitslpg of DNA. The buffer used for EcoRI digestion was 50 mM Tris-HCI, pH 8.0, 2.5 mM MgCI,, and 50 mM NaCI. The buffer used for PstI, AvaI, and XbaI was 50 mM Tris-HCI, pH 8.0, 2.5 mM MgC12, and 100 mM NaC1. The fragments were electrophoresed in a 1% agarose gel in 0.5 X e-buffer. After the restriction digestion, half of the reconstituted DNA was phenol-extracted or loaded directly in the gel with 0.5% SDS loading dye to strip the DNA of proteins.
To linearize the plasmid with XbaI or SspI, before attempting reconstitution, the restriction digestions were done under the conditions indicated by the manufacturers of the enzymes. The digestions of pPol I 208-4 for the binding competition assay using PuuII and XbaI were also done according to the manufacturer's reaction conditions. When comparing XbaI-linearized to SspI-linearized and reconstituted pPol I 208-4, the gels were scanned using a Zeineh scanning densitometer SL-504-XL. Peak heights were measured and normalized so that the total amount of DNA per lane is 100%. The results were plotted to compare the efficiency of cutting in both cases.

RESULTS
Reconstitution of Nucleosomes onto the pPol I 208-4-The plasmid pPol I 208-4 was designed to contain an rRNA core promoter region directly upstream of four repeats of a sequence containing the 5 S rDNA from L. uariegatus (Fig. 1). This latter sequence is known for its ability to define the binding position of histone octamers on a linear DNA template in uitro (Simpson and Stafford, 1983;Simpson et al., 1985;Simpson, 1986). Our initial studies were aimed a t testing the efficiency of the reconstitution, which was carried out a t 150 mM NaCl using the polyglutamic acid method. After overnight incubation a t 37 "C, the reconstituted plasmids were submitted to sedimentation velocity analysis. The integral distribution of s~~,~, obtained for reconstituted material at input ratios from 0.6 to 2.0 g of histonelg of DNA, showed an underreconstitution demonstrated by the presence of heterogeneous material with sedimentation coefficients ranging from about 19 S (corresponding to the supercoiled naked DNA) up to about 85 S (see Fig. 2). When the input ratio was increased up to 2.2, the template appeared to be overreconstituted, exhibiting the presence of heterogeneous material with a n S value over 80 S and up to 120 S (data not shown). The optimal input ratio was found to be R = 2.05, a t which ratio the distribution of the S values obtained covers a narrow range (between 73 and 76 S), as shown in Fig. 2. It should be noted that input ratios are almost certainly higher than the stoichiometry of the complex because histones are lost on surfaces when working with such small volumes.
To determine the average spacing between the nucleosomes, a mild micrococcal nuclease digestion was performed. The digestion products were run on a 1.5% agarose in 0.5 x ebuffer (Fig. 3). The expected ladder pattern was obtained, but the spacing was found to range between 123 and 159 bp, suggesting close packing. For the region containing the 5 S gene repeats, the spacing is expected to be about 200 base pairs. Although our results indicate close packing of the overall plasmid under these conditions, these data cannot describe the positioning of the nucleosomes on the insert region. Nucleosomes Are Incorrectly Positioned on the 5 S RNA Genes When the Plasmid Is Reconstituted in the Absence of the pol I Transcription Complex-To investigate nucleosome positioning after reconstitution of the template, the reconstituted material was digested with several restriction endonucleases, and the results were compared with the patterns obtained with naked DNA. The rationale of this experiment is to monitor the availability of the restriction sites. If the nucleosomes are positioned as described by Dong et al. (1990) and Meersseman et al. (1991), the EcoRI, PstI, and XbaI sites should be fully available for cutting, on the reconstituted plasmid as well as on the naked template (see Figs. 1 and 4). The digestion patterns showed that the expected cutting was not observed when the reconstitution had been carried out on this circular template using the polyglutamate method. For example, on a template with properly positioned nucleosomes, the PstI and AuaI sites and the most upstream EcoRI site of each 5 S gene should be available for restriction. The extra bands in lanes 5 (EcoRI digestion), 8 (PstI digestion), and 11 (XbaI digestion) indicate that only partial digestion of the plasmid had occurred, due to the obstruction of some of the sites by nucleosomes. Lanes 6,9, and 12 show the same pattern as lanes 5, 8, and 11; the only difference is that the bands are shifted downward after proteinase K treatment. This band  (Dong et al., 1990).
shift confirms the presence of nucleosome structures on the DNA templates. Lanes 8 and 9 showed about 50% of complete digestion, suggesting that either only one of the two PstI sites is accessible or both are accessible 50% of the time. The XbaI site seemed to be more open, but still displayed some protection (lunes 11 and 12). Neither does the pattern obtained with the EcoRI-digested R = 2.05 pPol I 208-4 match the naked plasmid digestion pattern (lunes 4, 5, and 6). The partial protection of the EcoRI sites again indicates a mispositioning of the histone octamers. Although it is possible to see some partial digestion for some of the enzymes used in this study, the overall significance of the patterns is to demonstrate that the positioning of the histone octamers onto the DNA does not match what was expected from studies made on linear arrays of tandem repeats of 5 S genes reconstituted by salt gradient dialysis (Dong et al., 1990;Meersseman et al., 1991).
There are several possible explanations for such results.
1) The topological constraints of a circular plasmid might have a major effect in determining the positioning or displacement of histone octamers (see Freeman and Garrard (1992) for review). However (see below), simple linearization of the plasmid does not, in itself, assure correct positioning.
2) It is also possible that some feature of the reconstitution protocol (PGA or low ionic strength) could be interfering with proper positioning, as has been shown on short linear DNA templates (Pennings et al., 1989). However, a step dialysis reconstitution was attempted on pPol I 208-4 at an input ratio of R = 2.05. This resulted in restriction digestion patterns very similar to those observed when PGA was used to reconstitute (data not shown). This result indicates that it is not the method of reconstitution but some other feature of the plasmid that produces the irregular positioning.
3) Finally, the plasmid sequence or the RNA polymerase I promoter region might contain regions with high affinity for histone octamers, which would in turn influence nucleosome positioning in the adjacent 5 S gene region.

Incorrect Positioning in the Repeat Region Results from the
Proximity of the RNA Polymerase I Promoter-To assess the relative importance of DNA topology uersus the effect of the proximity of the RNA polymerase I promoter region, the plasmid was linearized in two different ways prior to reconstitution; cleavage was by either XbaI or SspI. The XbaIlinearized plasmid does not contain the polymerase I promoter region upstream of the stretch of 5 S genes; rather, it is moved to a far downstream position. Thus, any possible interference Binding competition assay monitored by gel shift assay. The three fragments obtained from the double digestion with PuuII and XbaI were reconstituted in the presence of increasing ratios of histone to DNA. After reconstitution, the DNA was electrophoresed in a 3.5% acrylamide gel to monitor band shifts due to the binding of histone octamer(s) onto the DNA templates. from that region should disappear. On the other hand the SspI-linearized plasmid will still contain the promoter in its normal position upstream of the 5 S genes and therefore will give information about the effect of that sequence on the mispositioning effect. After reconstitution, the complexes were run on an high performance liquid chromatography C8 column to assure the absence of free DNA before submitting the reconstituted sample to digestion. The samples showed no free DNA (data not shown).
The efficiency with which different restriction endonucleases cut linearized pPol I 208-4 before and after reconstitution was determined by comparing the amount of digested products obtained from naked DNA and reconstituted DNA (see Fig.  5). It was found that reconstituted XbaI-digested plasmid treated with EcoRI displays 90% of the efficiency of cutting at the EcoRI restriction sites observed in the case of naked pPol I 208-4. On the other hand, when the SspI-linearized plasmid was digested, the relative amount of the 208-bp fragment produced drops to about 50%, showing more protection of the EcoRI sites and therefore reflecting a less accurate positioning (Fig. 5, cf. lanes 4 and 5).
A similar analysis was performed on the SspI-and XbaIlinearized plasmids utilizing PstI. The relative efficiency of cutting was again higher in the case of the XbaI-treated plasmid (75%), as compared with the 45% obtained for the SspI-treated pPol I 208-4.
In short, in every case, the cutting was found to be more efficient when the plasmid was linearized with XbaI. These results demonstrated that positioning was less regular when the SspI-linearized DNA was provided as a template for the reconstitution than when the plasmid had been linearized with XbaI. Thus, the proximity of the promoter region to the tandem repeat region inhibits proper reconstitution in the latter. A possible explanation is that the binding of one nucleosome on the promoter region may be changing the phasing of histone octamers on the adjacent tandem-repeat region (see Fig. 6). The RNA Polymerase I Promoter Region Competes Strongly with Other Sequences for Histone Octamers-The above results imply that sequences from the Pol I promoter region might have a higher affinity for histone octamers than do the tandemly repeated 5 S gene sequences. This was investigated directly by allowing three regions of the plasmid to compete for histone octamers under conditions in which histones were limiting. The pPol I 208-4 was cut with PuuII and XbaI, generating three fragments: 1) a linear fragment (199 bp) containing the Pol I promoter region, 2) a 1080-bp fragment containing four copies of the 208-bp positioning sequence, and 3) a fragment containing 2320 base pairs of the pUC8 sequence (see Fig. 7). A mixture of these three DNA fragments was used for the competition studies. Reconstitution was via our usual PGA technique; however, the histone:DNA ratio was varied from 0.6 to 2.05 in order to assay competition. The material obtained after overnight reconstitution was analyzed by band shift on a 3.5% acrylamide gel. This analysis showed that as the histone:DNA ratio is increased, the 199-bp fragment containing the RNA polymerase I promoter region plus 104 bp competes efficiently for the binding of histone octamers in a titration experiment with the 1080-bp fragment containing four copies of the positioning sequences or with the entire 2320-bp pUC8 fragment. This argues that the pol I promoter region has a nucleosome binding affinity in the same range of magnitude as do four copies of the 5 S gene DNA.
These results may also explain why nucleosomes reconstituted on DNAs containing this promoter sequence upstream from the repeat were not correctly positioned on the repeats of the 5 S gene (Figs. 4 and 5). It seems likely that the tight binding of histone octamers to the promoter region disrupts the regular nucleosome phasing across the region that contains the four repeated 5 S genes. One factor that may be important in the high affinity of the Pol I promoter for histone octamers is DNA bending. It was recently shown that the bent DNA of trypanosome kinetoplast minicircles bound nucleosomes 6-7-fold more tightly than bulk sequences. Especially significant for our studies was the observation that the location of a bend affected the position of neighboring octamers (Trifonov, 1980;Shrader and Crothers, 1989;Constanzo et al., 1990). Recently, intrinsically bent DNA has been found near the promoter of the transcription initiation site of the Physarum rDNA (Schroth et al., 1992). We analyzed the A c a n t h m o e b a Pol I promoter region used in these experiments by computer modeling in the manner of Schroth et al. (1992) and detected a 35" bend centered about 23 bp from the positioning sequence and +8 bp from the transcription start site. If, as has been observed for the positioning sequence itself, a favored nucleosome position puts this bend at the dyad axis, this would overlap the 5 S rDNA sequence as shown in Fig. 6 (top). This could then disturb subsequent positions in the repeat region. A Stalled Transcription Complex Restores Correct Nucleosome Positioning on the 5 S rRNA Genes-If a nucleosome bound to the promoter region causes changes in positioning of adjacent nucleosomes, what will be the effect of the binding of the transcription complex? To test for effects of transcription factors TIF-IB and aUBF and RNA polymerase I on the nucleosome positioning, we first assembled these proteins onto pPol I 208-4 and then reconstituted with nucleosomes and probed restriction site availability. The plasmid was first incubated in the presence of the two transcription factors, TIF-IB and aUBF, and the RNA polymerase I. The transcription complex was then initiated by the addition of ATP and GTP and stalled at position +8 by starving it for UTP and CTP. Once the transcription complex was engaged, reconstitution was carried out. The reconstituted plasmid was then digested with XbaI, PstI, AuaI, or EcoRI; each preparation was then phenol-extracted. The digestion products were electrophoresed in a 1% agarose gel in 0.5 X e-buffer next to similar digests of naked pPol I 208-4 (Fig. 8).
The restriction endonuclease sites in these constructs exhibit availability consistent with correct or nearly correct nucleosome positioning. All digestions went to completion with 5 units of restriction enzyme/pg of DNA, whereas some dimers and trimers were visible in the AuaI and EcoRI digestions digested with only 1 unit of enzyme/pg of DNA.
We conclude that positioning on the 5 S rDNA downstream from the promoter was rescued by the addition of the transcription factors plus the RNA polymerase I. On the other hand, the presence of the transcription factors TIF-IB and aUBF, in the absence of polymerase I, did not rescue the positioning (data not shown). A possible reason for this is shown in Fig. 6 (bottom). The polymerase may prevent deposition on the promoter and yet not interfere with adjacent nucleosomes. Why the factors themselves do not rescue is entirely unclear; at this point we cannot exclude the possibility that a nucleosome can displace the factors but not the factors plus the polymerase.

DISCUSSION
We have shown that the expected positioning of reconstituted nucleosomes on a tandemly repeated array of 5 S genes is not seen when the array is placed adjacent to the Acanthamoeba polymerase I core promoter on a circular plasmid. When the template was linearized by restriction endonuclease cutting before the reconstitution, the subsequent position pattern depended upon where the cut had been made. Retention of the promoter sequence upstream from the 5 S gene repeats resulted in incorrect positioning, whereas more regular positioning was observed if the region was moved away from the 5 S gene repeat. This argues that the promoter region somehow interferes with positioning. Reconstitution competition assays showed the unexpected result that the Pol I promoter has an affinity for nucleosomes comparable with that of the 208-bp positioning sequences. This may be explained by modeling studies, which predict a bent sequence in the promoter region. Such a sequence might strongly bind a nucleosome that would overlap the first 5 S gene repeat and might then interfere with further positioning by the 5 S RNA repeats. The recovery of the positioning, upon prior formation of a stalled transcription complex, suggests that the presence of the complex prevents deposition of a nucleosome at this site. This event would then prevent interference with positioning of an octamer on the first 5 S gene and allow the subsequent nucleosomes to adopt the expected positions (see Fig. 6).
The positioning on the pUC 8 portion of the plasmid was not examined, but the micrococcal ladder indicates a close packing. The input ratio of 2.05 histone/DNA appears high, and while it may not correspond to the actual stoichiometry of the complex, it may also imply a close packing of the nucleosomes onto most of the plasmid. The fact that the promoter region contains a site of high affinity for nucleosomes may have wider significance. Indeed, this may be related to the proposed mechanism of regulation of transcription involving the binding of nucleosomes onto the promoter regions of RNA polymerases (Wasylyk and Chambon, 1979;Morse, 1989;Almouzni et al., 1990; see Grunstein (1990) for a review).