Region 3.2 of the σ Subunit Contributes to the Binding of the 3′-Initiating Nucleotide in the RNA Polymerase Active Center and Facilitates Promoter Clearance during Initiation*

Region 3.2 of the RNA polymerase σ subunit forms a loop that protrudes toward RNA polymerase active center and partially blocks RNA exit channel. To provide some insights into the functional role of this region, we studied a deletion variant of the Escherichia coli σ70 subunit that lacked amino acids 513–519 corresponding to the tip of the loop. The deletion had multiple effects on transcription initiation including: (i) a significant decrease in the amount of short abortive RNAs synthesized during initiation, (ii) defects in promoter escape, (iii) loss of the contacts between the σ subunit and the nascent RNA during initiation and, finally, (iv) dramatic increase in the Km value for the 3′-initiating nucleotide. At the same time, the mutation did not impair promoter opening and the binding of the 5′-initiating purine nucleotide. In summary, our data demonstrate an important role of σ region 3.2 in the binding of initiating substrates in RNA polymerase active center and in the process of promoter clearance.

The subunit of bacterial RNA polymerase (RNAP) 2 plays the key role in promoter-specific initiation of RNA synthesis. The subunit is involved in multiple processes during initiation including initial promoter recognition, DNA melting, abortive synthesis, and promoter clearance (1). Recent progress in structural studies of RNAP together with the bulk of biochemical data allowed the creation of an integral picture of transcription initiation and a proposal of functional roles for individual domains of the subunit. In RNAP holoenzyme, the subunit occupies the upstream part of the main RNAP channel. Two DNA binding domains of , involved in the recognition of the Ϫ10 and Ϫ35 promoter elements, are formed by conserved regions 2 and 4 and interact with a coiled-coil element of the ␤Ј subunit and the flap domain of the ␤ subunit, respectively (2)(3)(4)(5). The promoter recognition domains of are connected by a flexible linker formed by conserved region 3.2 (amino acids 498 -526 in Escherichia coli numbering). A hairpin-like loop from region 3.2 protrudes toward the active center of RNAP and occupies a part of the RNA exit channel (Fig. 1). This led to several predictions about the functional role of this region. First, the region 3.2 loop was proposed to directly participate in the binding of the 5Ј-initiating nucleotide in the i-site of the RNAP active center (6,7). In support of this, E. coli 70 was shown to cross-link to a 5Ј-initiating ATP analogue in a segment between amino acids 508 and 561 containing regions 3.2 and 4.1 (8). However, no experiments were done demonstrating the functional role for this region in substrate binding. Second, region 3.2 was proposed to play an important role in the process of promoter escape (3,6,7,9). As this region blocks the path for RNA exit (Fig. 1), the elongating RNA transcript must either displace it from the RNA exit channel thus promoting dissociation or dissociate itself from the complex and be released as an abortive product. Thus, direct competition between the elongating RNA transcript and region 3.2 may be one of the major causes for abortive initiation. In support of this, deletion of the C-terminal part of including region 3.2 was shown to lead to a significant decrease in the amount of abortive products synthesized during initiation from an extended Ϫ10 promoter (6). However, the deletion also removed conserved region 4 that led to multiple defects during initiation and complicated interpretation of the experimental data. The role of region 3.2 in promoter escape has not been directly studied. Intriguingly, a finger-like domain of the eukaryotic general transcription factor TFIIB was found to occupy a similar position at the RNA exit channel of RNAP II and was proposed to clash with the growing RNA during initiation (10). This indicates that prokaryotic and eukaryotic RNAPs may share a common mechanism of abortive synthesis and promoter escape (11,12).
In this work, we show that, in agreement with structural predictions, displacement of region 3.2 by nascent RNA is required for efficient promoter escape. At the same time, we demonstrate that region 3.2 is not involved in the binding of the 5Ј-initiating nucleotide at the i-site but instead contributes to the binding of the 3Ј-substrate at the iϩ1 site of RNAP active center.

EXPERIMENTAL PROCEDURES
Protein Purification-E. coli core RNAP bearing a hexahistidine tag at the C terminus of the ␤Ј subunit was purified as described (13,14). The mutant rpoD gene encoding the ⌬513-519 70 subunit was obtained by standard PCR mutagenesis methods. Both wild type and mutant rpoD genes were cloned between the NdeI and EcoRI sites of the pET28 vector and overexpressed in the E. coli BL21(DE3) strain. The resulting proteins contained a six-histidine tag at the N terminus. Bacterial pellet from 1 liter of cell culture was resuspended in 25 ml of lysis buffer (20 mM Tris-HCl, pH 7.9, 500 mM KCl, 0.1% Tween-20, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 5% glycerol) and sonicated. Supernatant obtained after centrifugation was loaded onto a 5-ml chelating HiTrap column (GE Healthcare) charged with Ni 2ϩ and equilibrated with the same buffer. The column was washed with buffer containing 35 mM imidazole, the subunit was eluted with 200 mM imidazole and precipitated with ammonium sulfate. The pellet was dissolved in 15 ml of buffer containing 40 mM Tris-HCl, pH 8.3, 1 mM EDTA, 0.1 mM dithiothreitol, 5% glycerol, and DEAE ion-exchange chromatography was performed as described (13).
Transcription in Vitro-DNA fragments bearing the sequences of either wild type or consensus T7A1 promoters (from Ϫ85 to ϩ53 nucleotides relative to the starting point of transcription) were obtained by PCR from synthetic oligonucleotide templates. Holoenzyme RNAP was prepared by incubating the core polymerase (100 nM) and either the wild type or the mutant 70 subunit (500 nM) in transcription buffer (40 mM Tris-HCl, pH 7.9, 40 mM KCl, 10 mM MgCl 2 ) for 5 min at 37°C. The DNA template was added (30 nM), and the samples were incubated for 5 min at 37°C. Nucleotide substrates were added (25 M concentration of each NTP with addition of [␣-32 P]GTP, 3000 Ci/mmol, PerkinElmer Life Sciences) and transcription was proceeded for 5 min. The CpA primer was present at 50 M, when indicated. RNA products were separated on 20% denaturing polyacrylamide gel and analyzed by phosphorimaging. Apparent K m values for the initiating substrates were measured on the wild type T7A1 promoter; reactions contained 100 nM core RNAP, 500 nM subunit, and 10 nM DNA in 10 l of the transcription buffer. One of the two initiating nucleotides (ATP or UTP) was taken at 1 mM (with the addition of the corresponding ␣-32 P-labeled nucleotide), while the concentration of the other was varied from 1 M to 6 mM. The samples were incubated for 1 min at 37°C, and the reaction was stopped with 10 l of solution containing 8 M urea and 20 mM EDTA. Dinucleotide product was separated from mononucleotides on 30% polyacrylamide gel and quantified with PhosphorImager (Amersham Biosciences). The data were fit to the hyperbolic equation rate at indefinite concentration of NTP) using GraFit software (Erithacus Software).
KMnO 4 Footprinting-The promoter DNA fragment used in the footprinting experiments contained a EcoRI site at position Ϫ81 relative to the starting point of transcription. DNA was treated with EcoRI and labeled at 3Ј-end of the template promoter strand with Klenow enzyme (Amersham Biosciences) and [␣-32 P]dATP. Holoenzymes containing either wild type or mutant subunits were incubated with the labeled promoter fragment (10 nM) in 10 l of the transcription buffer for 10 min at 37°C. 1 l of 20 mM KMnO 4 was added, and the reaction was stopped after 15 s by addition of 5 l of stop-solution containing 1 M ␤-mercaptoethanol and 1 M sodium acetate, pH 4.8. DNA was processed as described (15) and analyzed on 10% denaturing polyacrylamide gel.
Cross-linking Experiments-Cross-linking of RNAP subunits with the initiating ATP analogue in the open promoter complex was performed as described (8). The structure of the ATP analogue is shown on Fig. 4A. Prior to cross-linking, the aldehyde group of the reagent was reduced with 10 mM NaBH 4 for 10 min at 25°C. Open complexes formed by either wild type or mutant RNAP on the wild type T7A1 promoter were immobilized on nickelnitrilotriacetic acid-agarose (Qiagen) through a hexahistidine tag present at the C terminus of RNAP ␤Ј subunit and incubated with the ATP analogue (at 1 mM final concentration) for 1 h at 37°C. The uncross-linked reagent was removed by extensive washing with transcription buffer, and the samples were incubated with 100 M [␣-32 P]UTP for another 0.5 h at 37°C. Cross-linking products were separated on 8% SDS-polyacrylamide gel and analyzed by phosphorimaging.

RESULTS AND DISCUSSION
Role of Region 3.2 in NTP Binding-To clarify the role of region 3.2 during transcription initiation, we generated a deletion variant of the 70 subunit lacking amino acids 513-519, which correspond to the tip of the loop formed by this region (amino acid sequence . . . 512-GDDEDSHLG-520 . . . , the deleted residues are bold underlined) (Fig. 1). The activity of RNAP containing the mutant subunit was studied in an in vitro transcription test. Two variants of a T7A1 promoter were used as templates. The first one was the wild type T7A1 promoter; the second (T7A1cons) was based on the sequence of T7A1 but contained two nucleotide changes in the Ϫ10 element bringing it to the consensus (TATAAT instead of GATACT in the wild type promoter). Transcription was performed either in the absence or in the presence of a dinucleotide primer (CpA) that was complementary to the Ϫ1 and ϩ1 nucleotides of the template promoter strand. The reaction was supplemented with [␣-32 P]GTP that allowed to label RNA products longer than 4 nucleotides (see the initial transcribed sequence of T7A1 on Fig. 2A). RNAP containing the wild type 70 subunit was active on both promoters independently of the presence of the primer dinucleotide ( Fig. 2A, lanes 1-4). In contrast, RNAP containing the mutant subunit was essentially inactive in the absence of the initiating primer ( Fig. 2A, lanes 5 and 6, and Fig. 2C). KMnO 4 footprinting demonstrated that the efficiency of promoter melting and the size of transcription bubble in the open promoter complex were the same for wild type and mutant RNAPs (Fig. 2B). Thus, inability of the mutant polymerase to initiate transcription could not be explained by abnormalities in promoter melting. The activity of the mutant RNAP was greatly stimulated in the presence of the initiating dinucleotide CpA (Fig. 2A, lanes 7 and 8, and Fig. 2C). This suggested that the defect of the mutation may be in the binding of initiating substrates and the first phosphodiester bond formation.
To test this hypothesis, we determined apparent K m values for the first (ATP, bound at the i-site) and the second (UTP, bound at the iϩ1-site) initiating NTPs for wild type and mutant RNAPs in a reaction of dinucleotide synthesis on the T7A1 promoter (Fig. 3). We found that the deletion of amino acids 513-519 from region 3.2 did not impair the binding of the 5Ј-initiating nucleotide (ATP). In fact, the mutant had even lower K m value for ATP than the wild type polymerase (450 Ϯ 200 M versus 920 Ϯ 300 M for wild type RNAP) (Fig. 3). This observation is consistent with the hypothesis that the RNA polymerase initiation site, specific for purine nucleotides, is present in the core RNAP and neither the DNA template nor the subunit is required for its formation (16). At the same time, the mutation had dramatic effect on K m for the second substrate (UTP) which was increased by about two orders of magnitude (250 Ϯ 70 M versus 3.3 Ϯ 0.6 M for wild type RNAP) (Fig. 3). From this, we conclude that inability of the mutant RNAP to start RNA synthesis from mononucleotides can be attributed to the defect in the binding of 3Ј-initiating NTP in the RNAP active center.
To test whether region 3.2 could be involved in direct contacts with initiating nucleotides, we performed a cross-linking experiment using a ␥-phosphate-modified initiating ATP analogue (Fig. 4A) (8). The alkylating group of the reagent has a broad specificity and can react with any nucleophilic amino acid side chain (17). Prior to the cross-linking, the aldehyde group of the analogue was reduced with NaBH 4 that resulted in a significant increase of the reactivity of the alkylating group (17). Previously, this reagent was shown to cross-link between amino acids 508 and 561 of the 70 subunit in the initiating promoter complex (8). We repeated the cross-linking experiment with RNAP containing the ⌬513-519 subunit. The experiment was done in two steps. First, open complexes formed by the wild type and mutant RNAPs on the T7A1 promoter were immobilized on nickel-nitrilotriacetic acid-agarose, and incubated with the ATP analogue to allow its covalent attachment to RNAP subunits. Second, the uncross-linked reagent was removed by extensive washing and the samples were incubated with [␣-P 32 ]UTP. This resulted in the attachment of the radioactive label specifically to those ATP molecules that were cross-linked in the vicinity of the RNAP active center. In accordance with published data, in the case of wild type RNAP the cross-links were formed with the and ␤/␤Ј subunits (Fig. 4B, lane 2). In contrast, only the large subunit(s) was labeled in the case of the mutant RNAP (with the efficiency of about 40% as compared with the wild type polymerase) (Fig. 4B, lane 3). Core RNAP did not form any cross-links confirming that the reaction was specific for the open promoter complex (Fig. 4B, lane 1). The absence of cross-linking in the case of the mutant subunit strongly suggested that in the wild type the cross-link was localized between amino acids 513 and 519. Thus, in the open promoter complex the region 3.2 loop is located in a close proximity to the RNAP active site (the distance between and the 5Ј-carbon atom of the initiating ATP should be less than 18 Å, which is the length of the cross-linking arm in an extended conformation, Fig. 4A). This region of may therefore directly participate in substrate binding during initiation. It should be noted, however, that in the structure of the Thermus thermophilus holoenzyme RNAP this region is located too far from the active site to contact initiating NTPs (the closest distance between and initiating nucleotides in the model of the open promoter complex is more than 12 Å, Fig. 1) (3, 18). Thus, during initiation of RNA synthesis, the region 3.2 loop may adopt a more extended conformation to allow direct contacts with initiating substrates. Alternatively, might contribute to the binding of the 3Ј-initiating substrate by an allosteric mechanism. Indeed, it was proposed that region 3.2 may affect the structure of the RNAP active center by changing the conformation of the template DNA strand (19). The experimental validation of this hypothesis is a subject of further studies.

Role of Region 3.2 in Abortive Initiation and Promoter
Escape-Transcription on the mutant T7A1cons promoter resulted in the synthesis of a large number of short RNA transcripts up to 16 nucleotides in length (Fig. 2A, lane  2). Control experiments demonstrated that these RNAs were not associated with transcribing RNAP and therefore represented real abortive products (data not shown). In this property, the T7A1cons promoter is similar to the semisynthetic T5 N25 antiDSR promoter (20). However, in the case of T5 N25 antiDSR the appearance of long abortive products was associated with unfavorable initially transcribed sequence, while in the case of T7A1cons it is most likely a result of strong interactions between RNAP and the core promoter region. T7A1cons was therefore used as a model promoter to compare the processes of abortive synthesis and promoter escape by the wild type and mutant RNAPs.
The reaction was performed in the presence of the CpA primer, i.e. at conditions when both RNAPs were active. The efficiency of the full-length RNA synthesis by wild type RNAP on T7A1cons was about 2-fold lower than on the wild type promoter indicative of the problems in promoter clearance ( Fig. 2A,  lanes 3 and 4, and Fig. 2C). The defect in promoter clearance became more pronounced in the case of mutant RNAP that synthesized four times less of the full-length transcript on the consensus promoter as compared with the wild type template (Fig. 2A, lanes 7 and 8, and Fig. 2C). Analysis of the abortive products synthesized on the T7A1cons promoter in the presence of CpA  Ϫ13 and ϩ3 is shown below the gel, the Ϫ10 promoter element is boxed. The modified thymine nucleotides in the melted promoter region are indicated on the right of the gel and below the promoter sequence. C, the efficiency of RO synthesis by RNAPs containing the wild type and mutant s on T7A1 and T7A1cons promoters (quantitation of data from A). The activity for each reaction is shown in percent of the activity of the wild type RNAP on the wild type T7A1 in the absence of the primer. D, the efficiency of abortive synthesis on the T7A1cons promoter by RNAPs containing the wild type and mutant 70 subunits. For each RNA product, the ratio of the radioactivity in the corresponding band on the gel to the radioactivity in the RO band is shown.
revealed that the mutant RNAP produced much lower amounts of short 5-7-nt transcripts relative to the wild type polymerase (Fig. 2, A and D). At the same type, the ratio of longer abortive transcripts (8 -16 nt) to the full-length RNA was severalfold higher for the mutant enzyme than for wild type RNAP (Fig. 2D). This indicated that although removal of region 3.2 stabilized short RNAs in the RNAP active center, it resulted in the increase of the amount of longer transcripts and led to a general defect in promoter escape. Our data are in agreement with the results of Hernandez et al. (21) who demonstrated that mutations in positions 504 and 506 of 70 region 3.2 decreased the number of short abortive RNAs (4 -8 nt) but stimulated synthesis of longer (9 -10 nt) abortive transcripts. In general, these results suggest that direct competition between region 3.2 and growing RNA during initiation facilitates promoter clearance, probably by weakening the contacts of the subunit with core RNAP or/and the non-template DNA strand. This is consistent with the conclusions made from structural consideration (6,7).
The increase in the amount of long abortive transcripts in the case of the mutant subunit is likely explained by steric clashing of growing RNA with the C-terminal part of the subunit further upstream of the RNAP active site (see Fig. 1). Remarkably, the longest abortive transcript that was synthesized in large quantities by both wild type and mutant RNAPs had the length of 16 nucleotides. Longer transcripts were present in much lower quantities and their synthesis did not differ between the two polymerases ( Fig. 2 and data not shown). These data are in agreement with the recently proposed model that region 4 of 70 bound to the ␤-flap domain presents a barrier to the RNA elongation past 16 nucleotides (3,22). The RNA-mediated displacement of region 4 from the ␤-flap results in the formation of a stable elongation complex and stopping of abortive synthesis. As the length of long abortive products does not differ for the wild type and mutant RNAPs, this final step of promoter clearance is apparently unaffected by the mutation.
It is interesting to compare our results with the results of Zenkin et al. (23), who studied initiation of primer RNA synthesis on the single stranded DNA replication origin of bacteriophage M13. On that template, the core enzyme alone was able to synthesize the first phosphodiester bond from two initiating NTPs but was unable to elongate RNA beyond the second position in the absence of the 70 subunit or in the case of mutant lacking amino acids 507-519 from region 3.2 (23). Although core RNAP possessed certain defects in the binding of the 3Ј-substrate (whose apparent K m was increased ϳ10 times in comparison with the holoenzyme), 3 these defects could not explain its inability to synthesize the full-length RNA as the core was unable to escape into elongation even in the presence of high concentrations of a dinucleotide primer. It was therefore proposed that the main role of the subunit was to stabilize short abortive transcripts in the RNAP active center and allow them to be elongated (23). Our results demonstrate that, during initiation on the dou-ble stranded T7A1 promoter, region 3.2 of is indispensable for the binding of the 3Ј-initiating nucleotide and the first phosphodiester bond formation but is not absolutely required for the subsequent cycles of nucleotide addition. This demonstrates that RNAP can use different mechanisms for initiation of RNA synthesis on different kinds of templates.