Bacterial polymerase and yeast polymerase II use similar mechanisms for transcription through nucleosomes.

We have previously shown that nucleosomes act as a strong barrier to yeast RNA polymerase II (Pol II) in vitro and that transcription through the nucleosome results in the loss of an H2A/H2B dimer. Here, we demonstrate that Escherichia coli RNA polymerase (RNAP), which never encounters chromatin in vivo, behaves similarly to Pol II in all aspects of transcription through the nucleosome in vitro. The nucleosome-specific pausing pattern of RNAP is comparable with that of Pol II. At physiological ionic strength or lower, the nucleosome blocks RNAP progression along the template, but this barrier can be relieved at higher ionic strength. Transcription through the nucleosome by RNAP results in the loss of an H2A/H2B dimer, and the histones that remain in the hexasome retain their original positions on the DNA. The results were similar for elongation complexes that were assembled from components (oligonucleotides and RNAP) and elongation complexes obtained by initiation from the promoter. The data suggest that eukaryotic Pol II and E. coli RNAP utilize very similar mechanisms for transcription through the nucleosome. Thus, bacterial RNAP can be used as a suitable model system to study general aspects of chromatin transcription by Pol II. Furthermore, the data argue that the general elongation properties of polymerases may determine the mechanism used for transcription through the nucleosome.

In eukaryotes, the DNA within the nucleus is packaged by histones into nucleosomes. In each nucleosome core, 146 bp of DNA are wrapped in about 1.7 superhelical coils around an octamer containing two molecules each of histones H2A, H2B, H3, and H4. The nucleosome core has a tripartite structure, with the H3/H4 tetramer organizing the central ϳ90 bp of nucleosomal DNA and two H2A/H2B dimers bound to the ends of the nucleosomal DNA (1).
Nucleosomes are present in vivo even when genes are actively transcribed by RNA polymerase II (Pol II), 1 indicating that even if nucleosomal structure is disrupted during transcription, recovery occurs almost immediately after passage of the enzyme (see Refs. 2 and 3 for reviews). Analysis of a Pol II encounter with a mononuclesome resulted in the discovery that the nucleosome acts as a potent barrier to the core Pol II enzyme in vitro, which cannot be relieved even when transcription is conducted in the presence of general elongation factors (4,5). More recently, it has been shown that transcription through the nucleosome by Pol II results in the loss of an H2A/H2B dimer without changing the position of the histones on the DNA (6).
The bacteriophage SP6 and T7 RNA polymerases have been extensively used as model systems for the in vitro analysis of transcription through the nucleosome (7). The properties of nucleosomal transcription by the multisubunit Pol II and the single-subunit phage SP6 RNAP are dramatically different. The nucleosome is a relatively weak barrier to transcription by bacteriophage RNAP (8 -10), and during transcription, the complete histone octamer is transferred from in front of to behind the enzyme (11)(12)(13)(14)(15). A nucleosome transfer mechanism is also probably used during DNA replication (16 -18) and nucleosome mobilization by ATP-dependent remodeling enzymes (see Ref. 19 for a review). Remarkably, the multisubunit eukaryotic RNA Pol III transcribes nucleosomal templates in a similar manner to SP6 RNAP (15). Thus, different RNAPs use different mechanisms to overcome nucleosomes in vitro (6).
It is surprising that the bacteriophage RNAP is able to transcribe a nucleosomal template more efficiently than Pol II, since of all the enzymes analyzed in vitro, only Pol II is likely to actually encounter nucleosomes in vivo. Indeed, bacteriophage RNAPs never encounter chromatin in vivo; Pol III transcribes only short genes covered with initiation factors that may exclude nucleosomes (reviewed in Ref. 20). The difference between Pol II, on one hand, and bacteriophage RNAPs and Pol III, on the other hand, could be a special evolutionary acquisition developed by Pol II to broaden factor-dependent regulation of genes in vivo. Indeed, this view is supported by the discovery of several transcription factors that facilitate Pol II transcription through chromatin in vitro and in vivo (see Ref. 21 for review). Alternatively, the low efficiency of transcription through nucleosomes by Pol II in vitro as compared with Pol III and phage RNAPs can be due to the differences in the properties of elongation complexes (ECs) formed by these RNAPs. To distinguish between these two possibilities, Pol II was compared with Escherichia coli RNAP, which does not transcribe chromatin templates in vivo but otherwise has similar transcription elongation properties to Pol II.
Pol II and E. coli RNAP belong to the class of RNAPs that are responsible for synthesis of messenger RNA in prokaryotes and eukaryotes. Although subunit composition varies in these enzymes, they all have similar protein architecture and a com-mon catalytic mechanism (22). The amino acid composition of the two largest catalytic subunits is very similar among all members of the family (23). General elongation properties of E. coli RNAP and Pol II ECs on DNA are very similar (24,25). At the same time, E. coli RNAP is a prokaryotic enzyme that never encounters nucleosomes in vivo. Here we test whether the enzyme from E. coli resembles SP6 RNAP and Pol III or uses the Pol II-type mechanism to transcribe through nucleosomes.
To address this question, transcription of nucleosomal templates by E. coli RNAP was analyzed. The data obtained using E. coli ECs clearly indicate that this enzyme utilizes the Pol II-type mechanism. All distinct features of the mechanism, including higher nucleosomal barrier to transcription, lack of nucleosome translocation, and displacement of an H2A/2B dimer, were recapitulated with E. coli RNAP. These data suggest that the general elongation properties shared by all members of the Pol II family may determine the mechanism of transcription though the nucleosome and the fate of the histone octamer.
EC Assembly and Ligation to the 204-bp DNA or Nucleosomal Template-Hexahistidine-tagged Pol II and E. coli RNAP were purified according to published protocols (25,26). The assembly and ligation were done as described (6,25,27), with the following modifications; 3 pmol of core RNAP or Pol II was incubated with 13-26 pmol of duplex (the 9-nt RNA primer (RNA9) annealed to the 50-nt template DNA strand (TDS50)), and the fully complementary nontemplate DNA strand (nontemplate DNA strand 59) (1.3-2.6 nmol) was incorporated as described (25,27). The Ni 2ϩ -NTA-agarose (50 l of 50% suspension; Qiagen, Chatsworth, CA) was washed three times with TB40, pretreated with 0.5 mg/ml acetylated BSA (Sigma) for 10 min, and washed again two times with 1 ml of TB40. EC9 was immobilized on the resin by constant shaking for 15 min at room temperature, washed three times with 1 ml of TB40, incubated for 10 min in 1 ml of TB1000, and washed twice with 1 ml TB40. For analyzing the templates after transcription, ECs were assembled with TDS50 that was labeled at the 5Ј-end using T4 polynucleotide kinase (New England Biolabs, Beverly, MA) and [␥-32 P]ATP (7000 Ci/mmol, ICN Biomedicals, Inc., Irvine, CA). Where RNA was analyzed, TDS50 was phosphorylated at the 5Ј-end using T4 polynucleotide kinase and cold ATP. In some cases, RNA was analyzed by assembling ECs with RNA9 that was 5Ј-end-labeled using T4 polynucleotide kinase. Washed EC9 was incubated in the presence of 100 -200 ng of TspRI-cut 204-bp template, 100 M ATP, 1% polyethylene glycol 8000, 0.5 mg/ml acetylated BSA, and 50 units T4 DNA ligase (New England Biolabs, Beverly, MA) at 12°C for 1-2 h. The ligated EC was washed three times with 1 ml of TB300, incubated in TB300 for 10 min, and washed three times with 1 ml of TB40.
Promoter Initiation and Ligation to the 204-bp DNA or Nucleosomal Template-A 110-bp DNA fragment containing the T7A1 promoter was used for promoter initiation (26). The sequence of this template is identical to the 50-bp DNA used for EC assembly from Ϫ10 to the TspRI site; the only difference is an extra 60 bp upstream (Ϫ70 to Ϫ11; Fig.  1B). This DNA fragment was prepared by annealing two long overlapping oligonucleotides and filling in the ends with Klenow (New England Biolabs, Beverly, MA). The 133-bp DNA fragment (5Ј-GGATCCAGAT-CCCGAAAATTTATCAAAAAGAGTATTGACTTAAAGTCTAACCTAT-AGGATACTTACAGCCATCGAGAGGGACACGGCGAAAAGCCAACC-CAAGCGAC ACCGGCACTGGGCACACACAGGAAAC-3Ј) containing the TspRI site was PCR-amplified and digested with TspRI, and the resulting 110-bp fragment was purified in an 8% polyacrylamide gel (19:1) containing 1ϫ TAE and 4 M urea. If the templates were to be analyzed after transcription, the promoter fragment was 5Ј-end-labeled by incubation with [␥-32 P]ATP and T4 polynucleotide kinase. To form open complexes, a 9-l reaction containing 1 pmol (0.5 g) of hexahistidine-tagged holo-RNAP enzyme (26) and 0.7 pmol of 110-bp T7A1 promoter fragment in TB40 was incubated at 37°C for 5-10 min. To initiate transcription and form EC11, 1 l of 10ϫ start mix (200 M ApUpC (Oligos, Etc. Inc., Wilsonville, OR), 200 M ATP, and 100 M GTP) was added, and the reaction was incubated at 37°C for 5 min. 20 l of Ni 2ϩ -NTA-agarose was treated with acetylated BSA, and EC11 was immobilized as described for the assembled ECs. The ECs were washed six times with 0.25 ml of TB40. Washed EC11 was ligated to a 204-bp fragment as described above, washed three times with 0.5 ml of TB300, incubated in TB300 for 10 min, and washed three times with 0.5 ml of TB40.
Transcription and Analysis of RNA-EC45 was formed by incubating assembled EC9 or promoter-initiated EC11 in the presence of 10 M ATP, CTP, and GTP and 0.5 mg/ml acetylated BSA for 10 min at room temperature. EC45 was washed six times with TB40 (the volume of each wash was 1 ml for assembled ECs and 0.5 ml for promoterinitiated ECs). EC48 was formed by incubating EC45 in the presence of 0.4 M [␣-32 P]GTP (3000 Ci/mmol; PerkinElmer Life Sciences) and 5 M cold UTP for 5 min at room temperature. The ECs were washed six times with TB40. The samples were supplemented with 0.5 mg/ml acetylated BSA, the KCl concentration was adjusted as indicated under "Results" and in the figure legends, and transcription was performed at room temperature for the indicated times in the presence of 200 M NTPs (or 250 M NTPs for Fig. 2). The reaction was stopped with an equal volume of loading buffer containing 10 M urea, 25 mM EDTA, and 0.25% each of xylene cyanol and bromphenol blue. The sample was boiled and loaded on an 8% (19:1) denaturing polyacrylamide gel. Quantitation was performed using a Cyclone Storage Phosphor System (Packard Instrument Co.).
Analysis of Transcribed Templates-EC45 was formed by incubating assembled EC9 or promoter-initiated EC11 in the presence of 100 M ATP, CTP, and GTP and 0.5 mg/ml acetylated BSA for 10 min at room temperature. Reactions with promoter-initiated ECs also contained 20 g/ml rifampicin (Sigma) during transcript elongation. Supernatant containing RNA and DNA released from a fraction of the unstable ECs with 15-45-nt RNAs (data not shown; results are similar to those obtained with Pol II (6)) was collected and used as a control (nontranscribed) template. The pellet was washed three times with TB300, incubated in TB300 for 20 min at room temperature, and washed three times with TB300 (each wash was 1 ml for assembled ECs and 0.5 ml for promoter-initiated ECs). The samples were divided into two aliquots and incubated in the presence of either 200 M ATP, CTP, and GTP (mock transcription) or 200 M all four NTPs (transcription) for 5-15 min at room temperature. Supernatant was collected from each reaction. For transcription in solution, EC45 was eluted with 100 mM imidazole in TB300 containing 0.5 mg/ml acetylated BSA, and the sample was diluted 2-fold prior to transcription. Analysis of templates by restriction enzyme sensitivity assays and native PAGE were performed as described (6,8). Mobility controls with the hexasome or nucleosome ligated to the 50-bp DNA duplex used for assembly or the 110-bp T7A1 fragment were prepared as described (6). Quantitation was performed as described above.

Assembly of E. coli RNAP ECs and Ligation to Defined Nucleosomal
Templates-To determine the mechanism of transcription through the nucleosome used by E. coli RNAP and to compare it with the Pol II-type mechanism, transcription was performed with both enzymes under very similar conditions using promoter-initiated and assembled ECs (6) (Fig. 1A). Briefly, ECs were assembled by annealing a 50-nt template DNA strand (TDS50) to a 9-nt RNA oligonucleotide (RNA9) and allowing the core polymerase to bind (27). The transcription bubble was formed by incorporation of the 59-nt nontemplate DNA strand into the complex. This created a long 3Ј-overhang (TspRI site) to be used for ligation. Finally, the completed EC9 (with the numerical index referring to the length of the RNA in the EC) was immobilized on Ni 2ϩ -NTA-agarose beads through the hexahistidine tag at the carboxyl terminus of the ␤Ј subunit of RNAP (26) or at the amino terminus of the Rpb3 subunit of Pol II (24). Alternatively, promoter-initiated EC11 was formed by incubation of RNAP and 110-bp T7A1 promoter-containing DNA fragment in the presence of ApUpC, ATP, and GTP. EC9 and EC11 were ligated to the 204-bp TspRI-cut DNA or mononucleosomes with defined positions (see Fig. 1, A and B) (6). The polymerase was walked to a point beyond the ligation junction (ϩ45), and the ECs were washed. Thus, polymerase from any ECs that were not ligated would run off the template, the unligated templates would be washed away, and only ligated templates would remain in the reaction. Transcription was resumed by the addition of all four NTPs.
The 204-bp template is only large enough to accommodate a single nucleosome per DNA molecule. Despite the fact that this template contained a nucleosome positioning sequence from the Xenopus 5 S RNA gene (6,28), reconstitution resulted in a mixture of mononucleosomes with different positions along the DNA (Fig. 1B). Nucleosome positioning was previously determined by restriction enzyme mapping and micrococcal nuclease mapping (6). Two regions of the template were preferred locations for nucleosome formation (N1 and N2; N2 is positioned on the 5 S sequence). Each of these regions had two local heterogeneous positions (a and b), giving rise to four different nucleosome positions (N1a, N1b, N2a, and N2b). The N1 and N2 nucleosomes are present in an ϳ1:1 ratio. The amounts of N1a and N1b are about equal, whereas N2b is present at about 2.5 times the amount of N2a.
The Nucleosome-specific Pausing Patterns Are Similar for E. coli RNAP and Pol II at Physiological Ionic Strength-To estimate the level of similarity in transcription through the nucleosome by E. coli RNAP and Pol II, elongation was per-formed with both enzymes under very similar conditions. ECs were assembled with 5Ј-labeled 9-nt RNA oligonucleotide and ligated to the 204-bp nucleosomal template. The resulting template is shown in Fig. 1B. RNAP was walked to the ϩ45position, the ECs were washed, and transcription was resumed by the addition of all four NTPs for the indicated time points at physiological ionic strength (150 mM KCl). The RNA resulting from transcription through the nucleosome by each enzyme was run side by side on a denaturing polyacrylamide gel (Fig.  2). The patterns of nucleosome-induced pausing and the slow elongation rate within the nucleosome were very similar for the bacterial and eukaryotic polymerases. Transcription of the naked DNA was much more efficient for both enzymes (data not shown). Thus, the bacterial polymerase, which never encounters eukaryotic chromatin in vivo, showed a very similar rate of elongation and intensity of the nucleosomal barrier to Pol II.
The Nucleosome Is a Strong but Reversible Barrier to E. coli RNAP-The region of nucleosome-specific pausing by E. coli RNAP and the strength of the barrier were determined by comparing transcription of the histone-free DNA and nucleosomal templates at different concentrations of salt using assembled ECs (Fig. 3A). RNA in EC9 was extended to 45 nt by incubation with ATP, CTP, and GTP (EC45 was formed), and the complexes were washed. The RNA was labeled during the FIG. 1. Experimental system. A, the experimental approach. Pol II and RNAP EC9 were assembled on 50-bp DNA, or promoter-initiated EC11 was formed on 110-bp DNA. The ECs were immobilized on Ni 2ϩ -NTA agarose beads, ligated to the 204-bp DNA or nucleosomal template, and washed. The transcribed sequences for both templates were the same. The movement of RNAP is shown only for the assembled ECs. RNAP was walked to the ϩ45-position or in some cases the ϩ48 position using [␣-32 P]NTPs to label the RNA. Alternatively, the RNA was labeled at the 5Ј-end prior to EC assembly. The ECs were washed after each manipulation, and transcription was resumed by the addition of all four NTPs. B, nucleosome positioning on the ligated 254-bp template used for assembly (EC9, top) and the 314-bp template used for promoter initiation (EC11, bottom). The 110-bp TspRI-cut fragment of DNA containing the T7A1 promoter was the exact same sequence of DNA used for EC assembly (Ϫ10 to ϩ244), except that there was an extra 60 bp of DNA upstream of the promoter (Ϫ70 to Ϫ11). The numerical labeling on the template represents the positions along the DNA relative to the transcription start site (ϩ1). Locations of the TspRI, MspI, EcoRI, and EcoRV restriction sites are indicated. The 254-bp template was labeled at the ligation site (TspRI site; star), and the 314-bp template was labeled at the promoter-proximal end (star). The nucleosome positions, N1a, N1b, N2a, and N2b are shown as ovals (a is solid, b is dashed) below the template. formation of EC48 (by adding cold UTP and [␣-32 P]GTP; see the sequence in Fig. 3A), and the ECs were washed again. Transcription was resumed from EC48 with the addition of all four NTPs at different concentrations of KCl. Transcription of histone-free DNA was very efficient and was completed within 5 min at all of the salt concentrations tested (Fig. 3A, lanes 1-6). However, at 150 mM KCl and below, the nucleosome arrested most of the RNAP in the region from ϩ48 (near the N1 nucleosomal border) to ϩ150 (lanes 8 and 9). The minor amount of run-off transcript that was formed under these conditions (Ͻ10% of the total radioactivity of the lane) most likely resulted from transcription of the small amount of free DNA present in the nucleosome preparation (Ͻ15%). In 300 mM KCl, ϳ25-30% of the RNAP could pass through the nucleosome (lane 10). This moderate ionic strength is enough to weaken some of the histone-DNA interactions, but the nucleosomal structure, EC integrity, and fate of the nucleosome during transcription are still maintained (6,29,30). When the ionic strength is raised to 1 M KCl, the nucleosome is destroyed (6, 30), but the EC remains stable (27). Under these conditions, RNAP was able to complete transcription of the template with the same efficiency as that on the naked DNA (compare lanes 5 and 11). Moreover, when transcription was initially conducted at 40 mM KCl, and then the salt concentration was raised to 1 M, more than half of the RNAP that was originally stopped in the nucleosome was able to complete transcription at the higher salt concentration (lane 12). Thus, for the majority of RNAP that could not transcribe through the nucleosome, the ECs were intact, and the arrest was reversible. These properties of nucleosomal transcription, including the pattern of pausing and the strength of the barrier, were strikingly similar to what was observed for Pol II under very similar conditions (6).
Time courses of transcription on the free DNA template and through the nucleosome were conducted at 300 mM KCl so that a substantial fraction of the RNAP (ϳ25-30%) could complete transcription of the nucleosomal template (Fig. 3B). Transcription of the free DNA template was nearly complete after 30 s (lane 3). In contrast, the same time point for the nucleosomal template only had a small amount of run-off transcript produced (probably from transcription of the free DNA contaminating the nucleosome preparation; lane 9). The majority of the RNAP was arrested in the nucleosome. Even at a salt concentration 2 times higher than physiological ionic strength, the rate of transcription through the nucleosome was ϳ10 times slower than that of the DNA template. Again, this result was remarkably similar to what was seen for transcription through the same nucleosomes by Pol II (6). Moreover, some of the pauses and arrests coincide with the intrinsic pause sites on the naked DNA as they do with Pol II (6,31).
Nucleosomes Form Similar Barriers to Assembled and Promoter-initiated E. coli ECs-The authenticity of assembled ECs of E. coli and yeast polymerases has previously been verified (24,25,27). Assembled ECs of E. coli RNAP were found to be indistinguishable from promoter-initiated ECs in terms of EC stability, footprint size, and rate of RNA polymerization (27). However, it had not been determined whether the mechanisms of transcription through the nucleosome were similar for promoter-initiated and assembled ECs. In our previous work with Pol II (6), we were unable to address this issue because promoterdependent Pol II in vitro systems are extremely inefficient in terms of initiation (32). We solved this problem using E. coli RNAP, which supports much more efficient promoter initiation than Pol II and requires no additional transcription factors.
As illustrated in Fig. 1B, a 110-bp TspRI-cut DNA fragment containing the A1 promoter of bacteriophage T7 was used for this purpose (26). This DNA fragment is identical in sequence to the 50-bp DNA used for EC assembly from the region Ϫ10 to the TspRI (ligation) site, but it contains an extra 60 bp of DNA upstream of the template used for EC assembly (Ϫ70 to Ϫ11; Fig. 1B). Thus, the transcribed sequence is identical for both templates. Promoter initiation was achieved as follows: (i) the 110-bp promoter fragment and E. coli holo-RNAP were incubated at 37°C to form initiation complexes; (ii) transcription was initiated by adding ApUpC, ATP, and GTP and incubating at 37°C to stall Pol II before incorporating CTP in the ϩ12position and allow the formation of EC11; and (iii) the EC was immobilized on Ni 2ϩ -NTA-agarose beads and washed. The rest of the experimental procedure is identical to that used for transcription with assembled ECs (Fig. 1A).
The strength and pattern of nucleosome-specific pausing by promoter-initiated E. coli RNAP was compared with that of assembled ECs. This was done by analyzing the efficiency of transcription on the DNA and nucleosomal templates at different KCl concentrations (compare Fig. 4A with Fig. 3A). RNAP was walked from EC11 to EC45, the RNA was labeled during the formation of EC48, and transcription was resumed from EC48 as described for the experiment in Fig. 3A. As for the FIG. 2. The nucleosome-specific pausing of E. coli RNAP is similar to that of Pol II. Time courses of transcription on the nucleosomal template at 150 mM KCl are compared side-by-side for the two enzymes. The RNA was labeled at the 5Ј-end (marked by an asterisk in the sequence below the gel) prior to EC assembly, and the polymerases were walked to form EC45 and washed. Transcription was resumed at 150 mM KCl by adding all four NTPs for the indicated time periods. The transcribed sequence is shown below the gel. The arrows at the left indicate the 45-nt starting material (lanes 1 and 7), the 64-nt readthrough product, and the 244-nt run-off transcript. Nucleosome positions are indicated on the right. assembled ECs, promoter-initiated RNAP was able to complete transcription of the DNA template at each of the salt concentrations analyzed (lanes 1-6). The nucleosome was again found to be a strong barrier. Full-length transcript formation was prevented at physiological ionic strength or below, and RNAP paused in the region from ϩ48 (near the N1 nucleosomal border) to ϩ150 (lanes 7-9). The strength of the barrier and the nucleosome-specific pausing pattern were indistinguishable between promoter-initiated E. coli RNAP and assembled ECs.
Time courses of transcription on the DNA and nucleosomal templates at 300 mM KCl were also analyzed for promoterinitiated RNAP (Fig. 4B). Again the results were very similar to what was seen for assembled ECs (compare with Fig. 3B). As expected, histone-free DNA transcription was almost complete after 30 s (lane 3), and transcription of the nucleosome was ϳ10 times slower (lanes 7-12). This similarity was further confirmed by the quantitative analysis of the data (Fig. 4C) 5A; see Fig. 1B). EC9 was incubated with ATP, CTP, and GTP to form EC45, and the templates contained in the supernatant were analyzed in a native gel. As was the case for Pol II, a fraction of the ECs dissociated during this walking step, and therefore nontranscribed templates were released into the supernatant (Fig. 5A, lane 1; see Ref. 6). Prior to transcription, the N1 nucleosomes are resistant to cleavage with EcoRI, and N1a is sensitive to cleavage with EcoRV (lanes 2  and 3). N1b is only slightly sensitive to EcoRV, because this restriction enzyme site is right on the nucleosomal border. The N2 nucleosomes are resistant to cleavage by EcoRV and sensitive to cleavage by EcoRI. The nucleosomes do not reposition during ligation to the ECs (Fig. 1B) (6). Next, EC45 was  lanes 1 and 7), and the sample in lane 14 was incubated with NTPs at 300 mM KCl. Chase with NTPs resulted in the appearance of two bands on the gel that should not be attributed to DNA-specific or nucleosome-specific pauses, because they appear without ligation (lanes 13 and 14). B, time courses of transcription on the DNA and nucleosomal templates at 300 mM KCl. The arrows at the left indicate the 48-nt starting material (lanes 1 and 7) and the 244-nt run-off transcript. Nucleosome positions are indicated on the right along with RNA lengths estimated from denatured DNA markers (MspI digest of pBR322). A shaded bar represents the region of nucleosome-specific pausing.
washed so that any templates that were released by the polymerase and not transcribed to completion were removed from the analysis. A second incubation of the ECs with ATP, CTP, and GTP demonstrated that only a small amount of ECs fell apart after the addition of the subset of NTPs (Fig. 5A,  lane 4). After the addition of all four NTPs, primarily fully transcribed templates were released into the supernatant (Fig. 5A, lane 5). This was also confirmed by labeled transcript analysis, where the primary RNA product released into solution at this step was full-length run-off (not shown; see Ref. 6). Transcription through the nucleosome resulted in the appearance of a faster migrating band (lane 5, Hex.) as compared with the mobility of the original nucleosomes (N1 and N2, lane 1). This band had the same mobility in a native gel as the reconstituted hexasome control (lane 6, see "Experimental Procedures"), suggesting that an H2A/H2B dimer was lost during transcription. The addition of histones H2A and H2B back to the transcribed templates led to the restoration of nucleosomes with mobilities similar to N1 and N2, indicating that the surviving histones remained bound at their original positions on the DNA (Fig. 5B, lanes 1 and 2). As expected, some of the transcribed templates were sensitive to EcoRI and resistant to EcoRV, whereas the others were sensitive to EcoRV and resistant to EcoRI (data not shown; see Fig. 1B). Thus, transcription though the nucleosome by E. coli RNAP led to the loss of an H2A/H2B dimer and the formation of a hexasome. This is the same mechanism that was seen for Pol II transcription through the nucleosome (6).
Our assay for determining the fate of the nucleosome after transcription relies, in part, on the mobility of the template during native gel electrophoresis. Mobilities of nucleosomes in native gel depend on their positions on the DNA (33). Note that on the 254-bp template, the N1 and N2 nucleosomes are re-solved in a native gel, but the local heterogeneous positions of the nucleosomes are indistinguishable. N1a is not resolved from N1b, and N2a is barely resolved from N2b (Fig. 5A, lane  1). However, on the longer template used for promoter initiation, the different positions of the nucleosomes (N1a, N1b, N2a, and N2b) are more distinguishable during native gel electrophoresis (compare Fig. 5A, lane 1, with Fig. 6A, lane 1). To confirm that the increased resolution of the nucleosomes on the longer template was not the result of new positions being created from nucleosome sliding along the DNA, restriction enzyme mapping for the 314-bp template used for promoter initiation was conducted (Fig. 6A). The complete digestion with MspI confirms that the nucleosomes did not slide onto the ligated promoter fragment (lane 2). Digestion with EcoRI and EcoRV demonstrates that the same fraction of nucleosomes that were sensitive/resistant to these enzymes on the templates with assembled ECs exists for the promoter-initiated templates (compare Fig. 6A, lanes 3 and 4, with Fig. 5A, lanes 2 and 3; see also Fig. 1B). Thus, as for the shorter template, ligation to the longer promoter fragment does not induce nucleosome repositioning.
The fate of the nucleosome during transcription by promoterinitiated and assembled E. coli ECs were compared (Fig. 6B).
To allow for only a single round of transcription, rifampicin was added to the transcription reaction to block reinitiation (34). EC11 was walked to EC45. Any nontranscribed templates released from ECs that fell apart as a result of transcription of the promoter-proximal region (lane 1) were washed away. Note that the fraction of unstable ECs is much less for promoter initiation than for EC assembly (compare Fig. 6B, lane 1, with  Fig. 5A, lane 1). The mock transcription control (Fig. 6B, lane 2) illustrates that any EC dissociation is at a background level after washing. Finally, with the addition of all four NTPs, the completely transcribed templates are released (lane 3). The mobility of the transcribed nucleosome is faster than the original nucleosomes (lane 4) and comparable with that of the reconstituted hexasome control (lane 5). Furthermore, DNA controls do not give rise to a band with hexasome mobility in the native gel (data not shown). Thus, similar mechanisms of transcription through the nucleosome are utilized by promoterinitiated and assembled ECs, each resulting in the formation of hexasome. This is the same mechanism that was seen for Pol II. Moreover, as for transcription with Pol II (6), transcription in solution using EC45 eluted with imidazole gives the same result (data not shown), indicating that the formation of hexasome occurs both during transcription in solution and while using the immobilized system. DISCUSSION Here, we report that RNA polymerase from E. coli recapitulates all of the properties of nucleosomal transcription by yeast Pol II in vitro (6). The nucleosome imposes a very strong barrier for bacterial RNAP that is partially reduced at elevated salt concentration. The pausing starts within the first 20 bp of the nucleosomal DNA and spreads almost across its entire length. The nucleosome-induced pausing of E. coli RNAP is very similar to that of Pol II in terms of strengths of the pauses and pausing patterns. Moreover, the passage of E. coli RNAP causes quantitative removal of one H2A/H2B dimer from the histone octamer, without changing the position of the remaining histones on the DNA. These results are highly similar to what was obtained with Pol II under identical conditions on the same nucleosomal template (6).
Our previous data with Pol II relied entirely on the ability to assemble Pol II ECs in the absence of initiation factors and promoter sequence (6). It has not been verified if the elongation properties of the assembled ECs on the nucleosomal template are the same as those of the "native" promoter-initiated ECs. As we show here, the properties of transcription through the nucleosome by promoter-initiated E. coli RNAP and assembled ECs are indistinguishable (compare Figs. 3 and 4 as well as Figs. 5 and 6). This similarity further establishes EC assembly as a valid and credible technique for studying transcription through the nucleosome in vitro.
The similarity of the mechanisms of transcription through the nucleosome by E. coli RNAP and yeast Pol II is most likely explained by the high structural and functional similarity of these enzymes. Both polymerases have very similar structure, and the amino acid composition of the active center is similar across almost the entire length of the template/transcript-binding site (23,35,36). In fact, the amino acid composition of these sites is almost 90% identical in the E. coli and yeast enzymes (23,35).
Reflecting a close similarity in protein design, the nucleic acid architecture is also very similar in bacterial and yeast Pol II ECs. These ECs protect ϳ30 -35 bp of the double-stranded DNA and contain a 10 -12-nt melted segment that forms the transcription bubble (25,36). In the bubble, 8 -9 nt of the 3Ј-proximal RNA hybridize with the template DNA strand to form an RNA:DNA hybrid, which represents a major stability determinant in both ECs (25,27,37). Two DNA strands come together 1-2 nt downstream and 2-3 nt upstream from the FIG. 5. Transcription through the nucleosome by assembled E. coli RNAP ECs results in the loss of an H2A/H2B dimer. Markers are an MspI digest of pBR322. A, the transcribed template had the same mobility in a native gel as the reconstituted hexasome. Labeled 254-bp templates were analyzed in a native gel before transcription (EC9 ϩ ACG; templates released during transcription of the first 45 bp of the template) after mock transcription (washed EC45 ϩ ACG; templates released after NTP addition), and after transcription (washed EC45 ϩ NTPs; templates released due to complete transcription) and before and after restriction enzyme mapping as described above the gel. The hexasome mobility control (Hex) was obtained by ligating 204 bp of reconstituted hexasome to labeled TDS50-nontemplate DNA strand 59 duplex (see Ref. 6). The mobilities of the nucleosomes (N1a/b (not resolved), N2a, and N2b), hexasomes (not resolved by position) (6), and free DNA are indicated. B, the addition of histones H2A and H2B back to the transcribed templates restores the original nucleosomes. Fully transcribed templates (EC45 ϩ NTPs) were incubated with H2A/H2B in TB300 and analyzed by native PAGE. The mobilities of the nucleosomes, hexasomes, and free DNA are indicated.
RNA:DNA hybrid. Both polymerases are believed to use the same "ratchet" mechanism for translocation along the DNA (36,38). In addition to forward movement, both enzymes are capable of backward motion without degrading the nascent RNA. This reaction occurs by backward sliding of the enzyme along the template with extrusion of the 3Ј-end of the RNA from the active center and results in transcription arrest (24,39). Notably, it has been demonstrated that neither Pol II nor E. coli RNAP dissociate from the template upon the encounter with the nucleosome, indicating that the DNA-bound histones may induce transcription arrest of these polymerases. It is known that Pol II and E. coli RNAP ECs respond similarly to arrest sites and various roadblocks (40 -42).
The high sensitivity of Pol II and E. coli RNAP ECs to various roadblocks, including nucleosomes, could be explained by the arrest caused by backtracking over a large distance (10 nt or more; see Refs. 24, 43, and 44 and references therein). In contrast, an exceptionally high elongation rate of the bacteriophage RNAPs (45), taken together with a very limited (not more than 2-3 nt) lateral movement of the polymerase in the EC (46), suggests that the ECs are likely to be resistant to arrest induced by the roadblocks. Indeed, SP6 RNAP can efficiently overcome the nucleosomal barrier to transcription (6,8,29). Similarly, yeast Pol III ECs are also much less sensitive to various roadblocks (47,48), including nucleosomes (15). Thus, the differences in catalytic properties or propensity to backtrack of the various RNAPs may account for the differences in the efficiency of transcription through the nucleosome by these enzymes. The fact that E. coli RNAP encounters the same block during transcription through the nucleosome as Pol II may be indicative that it is also regulated to some degree by DNA packaging. E. coli RNAP does not encounter nucleosomes in vivo; nonetheless, the bacterial DNA is highly compacted by basic proteins (49,50) that can possibly form a barrier to transcribing RNAP.
E. coli RNAP represents one of the most simply organized members of its family, which makes it particularly valuable for the study of the mechanism of nucleosomal transcription. The catalytic core of the bacterial enzyme, capable of normal elongation and termination, is made of two large ␤ and ␤Ј subunits and a dimer of ␣ subunits, which correspond to RPB2, RPB1, and a heterodimer of two RPB3/RPB5 subunits in the yeast Pol II, respectively (23,36). There is a vast pool of already known mutants of E. coli RNAP with altered elongation properties, and a well established genetic system makes it easy to obtain and characterize novel mutants (51). Moreover, the biochemical system allows for reconstitution of E. coli RNAP from either separately purified subunits or even separate domains within the subunits, which extends the list of manipulations possible with mutant RNAPs in vitro (52). These advantages of the model system based on E. coli RNAP may promote identification of the subunits or structural domains responsible for specific properties of transcription of nucleosomal templates by bacterial RNAP and Pol II.
Acknowledgment-We are grateful to Jodi Becker for purification of E. coli RNAP.
FIG. 6. Transcription through the nucleosome by promoter-initiated E. coli RNAP results in the loss of an H2A/H2B dimer. Markers are an MspI digest of pBR322. A, restriction enzyme mapping of the 314-bp nucleosomal template. 314-bp nucleosomes were obtained by ligating 204-bp nucleosomes to 5Ј-end-labeled 110-bp T7A1 promoter fragment, and the templates were analyzed by native PAGE before and after restriction enzyme digestion. The mobilities of the nucleosomes (N1a, N1b, N2a, and N2b), hexasomes, and free DNA are indicated. Positions of the restriction enzyme sites along the DNA are shown in Fig. 1B. B, labeled 314-bp templates were analyzed in a native gel before transcription (EC9 ϩ ACG; templates released during transcription of the first 45 bp of the template), after mock transcription (washed EC45 ϩ ACG; templates released due to NTP addition), and after transcription (washed EC45 ϩ NTPs; templates released due to run-off transcription) and before and after restriction enzyme mapping as described above the gel. The nucleosome (Nucl.) and hexasome (Hex.) mobility controls were created by ligating 204-bp reconstituted nucleosome and hexasome to a 5Ј-end-labeled 110-bp T7A1 promoter fragment. The mobilities of the nucleosomes (N1a, N1b, N2a, and N2b), hexasomes (not resolved by position), and free DNA are indicated.