On the Role of the Escherichia coli RNA Polymerase σ70 Region 4.2 and α-Subunit C-terminal Domains in Promoter Complex Formation on the Extended –10 galP1 Promoter

Bacterial promoters of the extended –10 class contain a single consensus element, and the DNA sequence upstream of this element is not critical for promoter activity. Open promoter complexes can be formed on an extended –10 Escherichia coli galP1 promoter at temperatures as low as 6 °C, when complexes on most promoters are closed. Here, we studied the contribution of upstream contacts to promoter complex formation using galP1 and its derivatives lacking the extended –10 motif and/or containing the –35 promoter consensus element. A panel of E. coli RNA polymerase holoenzymes containing two, one, or no α-subunit C-terminal domains (αCTD) and either wild-type σ70 subunit or σ70 lacking region 4.2 was assembled and tested for promoter complex formation. At 37 °C, αCTD and σ70 region 4.2 were individually dispensable for promoter complex formation on galP1 derivatives with extended –10 motif. However, no promoter complexes formed when both αCTD and σ70 region 4.2 were absent. Thus, in the context of an extended –10 promoter, αCTD and σ70 region 4.2 interactions with upstream DNA can functionally substitute for each other. In contrast, at low temperature, αCTD and σ70 region 4.2 interactions with upstream DNA were found to be functionally distinct, for σ70 region 4.2 but not αCTD was required for open promoter complex formation on galP1 derivatives with extended –10 motif. We propose a model involving σ70 region 4.2 interaction with the β flap domain that explains these observations.

Most Escherichia coli promoters are characterized by the presence of two 6-bp sequence elements centered ϳ10 and 35 nucleotides upstream of the transcription initiation point. These promoters are referred to as Ϫ10/Ϫ35 class promoters. Interaction of the RNA polymerase (RNAP) 1 70 subunit with Ϫ10 and Ϫ35 promoter elements is responsible for promoter recognition and transcription initiation. 70 conserved region 4.2 recognizes the Ϫ35 promoter element, while 70 conserved region 2.4 recognizes the Ϫ10 promoter element (reviewed by Gross et al.,Ref. 1; see also Ref. 2). Multiple alignments of promoter sequences permit the derivation of consensus sequences for the Ϫ10 and Ϫ35 promoter elements and show that most promoters deviate from the consensus (3). Promoter elements of strong promoters tend to deviate from consensus less than promoter elements of weak promoters. Thus, assuming that promoter elements with consensus sequence are preferred binding sites for 70 regions 2.4 and 4.2, the strength of regions 2.4 and 4.2 interaction with their respective promoter elements determines the efficiency of promoter complex formation.
For most promoters, RNAP 70 regions 2.4 and 4.2 interactions with their target promoter elements are sufficient for promoter complex formation. On some promoters, the presence of RNAP ␣-subunit C-terminal domains (␣CTDs) greatly increases transcription initiation beyond the basal level achieved through 70 -promoter element interactions (4). On these promoters, ␣CTDs make sequence-specific interactions with an A-rich promoter element (the "UP-element") located upstream of the Ϫ35 promoter element (reviewed by Gourse et al.,Ref. 5).
In the absence of a UP-element, ␣CTD non-specifically interacts with upstream DNA and the stimulatory effect of these interactions is less significant.
There exists a minor class of promoters that lack recognizable Ϫ35 promoter elements and whose Ϫ10 elements are extended with an upstream dinucleotide TG. Genetic data show that specific interaction between an additional region of 70 , conserved region 2.5, and the TG motif is required for promoter complex formation on promoters of this class (6). Evidently, this additional contact is strong enough to make promoter complex formation on extended Ϫ10 promoters independent of 70 region 4.2 and Ϫ35 promoter element interaction (7).
In order for template-directed RNA synthesis to occur, promoter DNA has to become locally melted (opened). In the catalytically competent open promoter complex, the melting extends from Ϫ12 to ϩ3 positions and thus includes the entire Ϫ10 promoter element. Promoter opening is temperature-dependent, and promoter complexes formed on Ϫ10/Ϫ35 type promoters "close" below 15°C (8). In contrast, promoter complexes on the extended Ϫ10 galP1 promoter remain open at temperatures as low as 5°C (9 -12). The reason for this unusual behavior is not completely understood. For example, while it is clear that the extended Ϫ10 motif contributes to promoter opening at low temperature (9), it is not sufficient, since open promoter complexes on some extended Ϫ10 promoters are sensitive to low temperature (12,13).
Previous work attempted to compare promoter complexes formed on the Ϫ10/Ϫ35 class and the extended Ϫ10 class promoters (9,12,14). However, due to technical constrains, relatively large fragments of promoter DNA were altered. As a result, interpretation of some of the published data is complicated, since promoter complex formation can be affected by sequences outside of promoter consensus elements. In this work, we used several derivatives of the galP1 promoter obtained by site-specific mutagenesis and a set of RNAP mutants that lacked ␣CTD and/or 70 region 4.2 to study promoter complexes formation in the absence of ␣CTD-UP element interactions, 70 region 4.2-Ϫ35 promoter consensus element interactions and 70 region 2.5-extended Ϫ10 motif interactions. Our results indicate that on the extended Ϫ10 galP1 promoter, ␣CTD-DNA interactions and 70 region 4.2-DNA interactions are functionally equivalent, and at least one of these interactions is necessary for promoter complex formation at physiological temperature of 37°C. In contrast, galP1 promoter complex formation at low temperature of 6°C is strictly dependent on the presence of 70 region 4.2.
To purify C-terminally truncated 70 -(1-565), E. coli XL-1Blue cells were transformed with the pCYB2_ 1-565 plasmid. Transformants were grown in 1 liter of LB with ampicillin (100 g/ml) at room temperature to an OD 600 of 0.6 -0.8, and expression was induced by the addition of IPTG to 1 mM. After 6 h, cells were harvested by centrifugation and resuspended in buffer H containing 20 mM HEPES, pH 8.0, 500 mM NaCl, 0.1 mM EDTA. Cells were lysed by sonication; the lysate was clarified by low speed centrifugation and loaded onto a 1-ml chitin column equilibrated in buffer H. The column was washed with 15 ml of buffer H and then quickly flushed with 3 column volumes of freshly prepared buffer H containing 30 mM dithiothreitol. The column outlet was sealed, and the column was left overnight at 4°C. Pure 70 -(1-565) was eluted with 3 column volumes of buffer H without dithiothreitol, dialyzed against buffer H, and stored at Ϫ20°C in the presence of 50% glycerol.
RNAP Core Enzymes-The wild-type RNAP core enzyme and the mutant lacking both ␣CTDs were prepared by in vitro reconstitution as described (18,20). The core and holoenzyme fractions were separated on a 1-ml Protein-Pak Q 8 HR column (Waters) attached to a Waters 650 FPLC. RNAP was loaded on the column in TGE buffer (10 mM Tris-HCl, pH 7.9, 1 mM EDTA, 5% glycerol) and eluted using a linear 60-ml gradient of NaCl from 0.23 to 0.4 M. Fractions containing RNAP core and holoenzymes were pooled separately, concentrated to 1-2 mg/ml and stored at Ϫ20°C in the presence of 50% glycerol. Heterodimeric RNAP core enzyme containing one wild-type ␣-subunit and one ␣-subunit lacking CTD was purified from XL1-Blue strain transformed with the plasmid pREII-NH␣45A (1-235) according to the procedure of Niu and co-workers (21). The core and holoenzyme fractions were separated on the Protein-Pak Q8 HR column (Waters) and stored as described above.
AsiA-E. coli BL21 (DE3) cells containing the pLysE plasmid (Novagen) were transformed with pET21AsiA. Transformants were grown in 2 liters of LB with ampicillin (100 g/ml) and chloramphenicol (25 g/ml) at room temperature to OD 600 of 0.6 -0.8 and expression was induced by the addition of IPTG to 1 mM. After 6 h, cells were harvested by centrifugation and resuspended in buffer C containing 20 mM Tris-HCl, pH 7.9, 5% glycerol, 500 mM NaCl. Cells were lysed by sonication and after a low speed centrifugation, cell extract was loaded onto a 5-ml chelating Sepharose column (Amersham Biosciences) loaded with Ni 2ϩ and attached to an FPLC. The column was washed with buffer C containing 20 mM imidazole, and AsiA was eluted with 100 mM imidazole in the buffer. The eluate was diluted 5-fold with buffer D (50 mM Tris-HCl, pH 7.9, 5% glycerol, 50 mM NaCl, 1.65 M (NH 4 ) 2 SO 4 , 1 mM ␤-mercaptoethanol) and loaded onto a 8 ml phenyl-Toyopearl column equilibrated with the same buffer. The column was washed and eluted with an (NH 4 ) 2 SO 4 gradient from 1.3 to 0 M. Homogeneous AsiA HIS protein was eluted from the column at 20 -0 mM (NH 4 ) 2 SO 4 , concentrated to 1 mg/ml concentration and stored at Ϫ20°C in the presence of 50% glycerol.

In Vitro Transcription
Abortive transcription initiation reactions contained, in 10 l of transcription buffer (50 mM Tris-HCI, pH 7.9, 10 mM MgCl 2 , 40 mM KCl), 20 nM RNAP core enzyme and 320 nM 70 or 640 nM 70 -(1-565). Reactions were incubated for 10 min at 37°C, followed by the addition of 10 nM promoter fragments, 0.1 mM initiating dinucleotide CpA, and 3 Ci (3000 Ci/mmol) of [␣-32 P]UTP. Reactions proceeded for 10 min at 37°C and were terminated by the addition of an equal volume of loading buffer containing 9 M urea. Reaction products were resolved by electrophoresis in denaturing (8 M urea) 20% (19:1) polyacrylamide gel, visualized by autoradiography, and quantified using the Molecular Dynamics PhosphorImager.

Determination of Open Complex Dissociation Kinetics
Open complex lifetime was measured on linear promoter-containing DNA fragments using an abortive transcription assay. Reactions contained, in 40 l of buffer (40 mM Tris-HCl, pH 7.9, 10 mM MgCl 2 , 175 mM NaCl, 2 mM ␤Ϫmercaptoethanol), 80 nM wild-type and mutant RNAP core enzymes and 80 nM 70 or 160 nM 565 . Reactions were incubated for 10 min at 30°C, followed by the addition of 10 nM promoter fragments and additional incubation for 15 min at 30°C. A 10-l reaction aliquot was transferred to tubes containing 0.1 mM dinucleotide CpA, 10 M UTP, and 3 Ci (3000 Ci/mmol) of [␣-32 P]UTP. Abortive transcription initiation was allowed to proceed for 10 min and was terminated by the addition of an equal volume of loading buffer containing 9 M urea. The remainder of promoter complex reaction was supplemented with 100 g/ml heparin and incubation at 30°C was continued. 10-l reaction aliquots were withdrawn immediately after the heparin addition and 30 and 90 min after the heparin addition and assayed for abortive transcription as described above. Reaction products were resolved, visualized, and quantified as described above.

DNase I Footprinting
The 147-bp DNA fragments (Ϫ96 to ϩ51) harboring the galP1 promoter and its derivatives were amplified by PCR from appropriate plasmids using universal T7 promoter and M13 reverse primers. PCR fragments were digested with HindIII and labeled at nontemplate strand by filling the HindIII sticky end with Klenow enzyme in the presence of [␣-32 P]dCTP. Promoter complexes were formed in 15 l of transcription buffer containing 200 nM RNAP core enzymes and a 4-fold excess of 70 or 8-fold excess of 70 -(1-565). Reactions were incubated for 10 min at 37°C, followed, when necessary, by the addition of 1 M of AsiA and further 10 min incubation at the same temperature. Reactions were next supplemented with 10 nM 32 P-endlabeled promoter fragments and incubated for 20 min at assay temperature. DNase I footprinting reactions were initiated by the addition of 0.1 or 1 unit of DNase I (Worthington) for 37 and 6°C reactions, respectively. The reactions proceeded for 30 s at the assay temperature and were terminated by addition of 85 l stop-mixture (20 mM EDTA, 10 g of denatured calf thymus DNA, and water) followed by phenol extraction and ethanol precipitation. The DNA samples were dissolved in 8 l of formamide-loading dye and analyzed using 7% polyacrylamide/8 M urea sequencing gels.

KMnO 4 Probing
Reactions were set up as described above for DNase I footprinting. Promoter complexes were treated with 1 mM KMnO 4 for 15 s at the assay temperature. Reactions were terminated by the addition of ␤-mercaptoethanol to 300 mM, followed by phenol extraction, ethanol precipitation, and 20 min treatment with 10% piperidine at 95°C. Reaction products were analyzed using 7% polyacrylamide urea gels.

Gel Retardation Assay
The reactions contained, in 20 l of transcription buffer, 10 nM RNAP wild-type and mutant core enzymes combined with appropriate amounts of 70 or 70 -(1-565). Reactions were incubated for 15 min at 37°C, transferred to 6°C, and incubated for additional 10 min, followed by the addition of 10 nM 32 P-endlabeled galP1 promoter fragment. After further 30 min of incubation at 6°C, reactions were combined with 4 l of loading buffer (transcription buffer containing 50% glycerol and 0.05% bromphenol blue). When necessary, heparin was added to the final concentration 50 g/ml, and reactions were immediately loaded on 5% (29:1) polyacrylamide Tris borate/EDTA gel. The gel was run in a cold room and reactions products were revealed by autoradiography.

The galP1 Promoter Derivatives Used in This Work-A galP1
promoter derivative previously constructed and characterized by Burns et al. (12) was used as a starting point of this work. The promoter differs from the natural galP1 promoter by three point mutations. Two substitutions at positions Ϫ9 and Ϫ8 relative to the start point of transcription create a consensus extended Ϫ10 promoter element, TGcTATAAT, instead of the wild-type sequence TGcTATggT. The third substitution introduces a T at position Ϫ19 and destroys the overlapping galP2 promoter, thus simplifying the analysis of galP1 promoter complexes. We used site-directed mutagenesis to construct "secondgeneration" derivatives of the optimized galP1 promoter constructed by Burns et al. (Fig. 1A). The first derivative, galP1-35, contains the consensus Ϫ35 promoter element, TTGACA, incorporated 18 base pairs upstream of T at position Ϫ12, the first base of the Ϫ10 consensus element. The second derivative, galP1-TG, was constructed by substituting the TG motif of the extended Ϫ10 promoter consensus element for AC. The resultant promoter was named galP1-TG. The final construct, galP1-35-TG, is a derivative of galP1-35 and also has the TG extended Ϫ10 motif substituted with AC. Note that our nomenclature differs from that adopted by Kamali-Moghaddam and Geiduschek in the accompanying article (28).
The ability of promoter fragments containing galP1 and its derivatives to serve as templates for the synthesis of CpApU from the CpA primer and the UTP substrate by E. coli RNAP 70 holoenzyme was investigated (Fig. 1B). DNA fragments containing galP1, galP1-35 as well as galP1-35-TG supported high and comparable levels of CpApU synthesis. In contrast, little transcription was detected when RNAP was combined with the galP1-TG fragment (less than 5% compared with other promoters used here). These result indicate the following. (i) galP1 has no functional Ϫ35 promoter consensus element. (ii) The Ϫ10 consensus promoter element alone, in the absence of the TG motif, is insufficient for transcription initiation, and (iii) the consensus Ϫ35 element sequence introduced in galP1-35-TG promoter is functional.
Promoter Complex Formation by RNAP Mutants Lacking ␣CTD and/or 70 Region 4.2 on galP1 Derivatives-We wished to determine the contribution of RNAP domains capable of interaction with upstream promoter sequences, the ␣CTD and the 70 subunit region 4.2, to promoter complex formation on galP1 and its derivatives. To this end, 5 mutant RNAP holoenzymes that lack one or both ␣CTDs and/or 70 region 4.2 and wild-type 70 RNAP holoenzyme were prepared. RNAP holoenzymes were reconstituted in vitro by combining RNAP core containing two copies of the wild-type ␣ subunit, RNAP core that has one ␣ and another copy truncated at amino acid 235, and RNAP core that contains two copies of truncated ␣ with either wild-type 70 , or with a 70 mutant that is truncated at amino acid position 565, 70 -(1-565), and thus lacks region 4.2 (16,22). Native PAGE analysis showed that reconstituted RNAP holoenzymes contained less than 5% RNAP core (data not shown). The ability of mutant holoenzymes and wild-type RNAP control to form promoter complexes at 37°C with DNA fragments containing galP1 and its active derivatives was investigated by DNase I footprinting. The footprinting results obtained with DNA fragments that contained functional promoters galP1, galP1-35, and galP1-35-TG are presented in Fig.  2, A-C, respectively. On all three promoters, the wild-type holoenzyme protected DNA from ϩ20 to Ϫ23 (Fig. 2, A-C, lane  2). DNase I hypersensitive sites centered at positions Ϫ25, Ϫ36, Ϫ44, Ϫ57, and Ϫ67 were also observed in complexes formed on all three promoters with wild-type RNAP. The DNA between the hypersensitive sites was protected (Fig. 2, A-C,  lane 2). This periodic pattern of protection and hypersensitivity has been attributed to wrapping of upstream DNA around RNAP (Attey et al.,Ref. 11). The hypersensitive site at position Ϫ25, was very prominent in the galP1-35 and galP1-35-TG complexes, and was less pronounced in galP1 complexes. DNase I hypersensitivity at about Ϫ25 is a common feature of promoter complexes formed by bacterial RNAP (3); it may result from DNA bending that arises when RNAP establishes simultaneous contacts with the Ϫ35 and Ϫ10 promoter consensus elements (2). If this interpretation is correct, weaker hypersensitivity in the galP1 complexes is expected, since 70 region 4.2 does not make specific contacts with DNA on galP1.
On all three promoters, complexes formed by 70 holoenzymes lacking one of the ␣CTDs were very similar to the corresponding wild-type RNAP complexes, though protection of a DNase I-sensitive band at Ϫ52 was decreased somewhat (Fig.  2, A-C, compare lanes 2 and 3). In complexes formed by 70 holoenzyme lacking both ␣CTDs, protection at Ϫ52 and Ϫ48 was further decreased (Fig. 2, A-C, compare lane 4 with lanes  3 and 2), suggesting that in the wild-type RNAP complexes, protection of AT-rich sequence between positions Ϫ52 and Ϫ48 is due to ␣CTD binding, in agreement with earlier results (11).
No footprint was observed when 70 -(1-565) holoenzymes were combined with galP1-35-TG (Fig. 2C, lanes 5-7). This result is expected, since the interaction between 70 region 4.2 and the Ϫ35 promoter consensus element is essential for pro-moter complex formation on the Ϫ10/Ϫ35 class promoters. In contrast, footprints were readily observed when 70 -(1-565) holoenzymes reconstituted from wild-type core enzyme or from core enzyme lacking one ␣CTD were combined with galP1 and galP1-35, also as expected (Fig. 2, A and B, lanes 5 and 6). These footprints are distinct from the corresponding 70 holoenzyme footprints in that position Ϫ37, which is fully protected in the presence of 70 RNAP complexes, is hypersensitive in the presence of 70 -(1-565) RNAP, presumably because specific interactions between 70 region 4.2 and the Ϫ35 promoter element (galP1-35) or nonspecific interactions between 70 region 4.2 and promoter DNA (galP1) are lacking. In complexes formed by 70 -(1-565) RNAP holoenzymes containing both ␣CTDs, protection at around Ϫ40/Ϫ50 is present (Fig. 2, A  and B, lane 5), suggesting that ␣CTD is able to interact with its binding site independently of 70 region 4.2. Hypersensitivity at Ϫ67 and at Ϫ57 is absent in 70 -(1-565) complexes. The reasons for this are unclear, since 70 region 4.2 is only expected to interact with promoter DNA ϳ35 base pairs upstream of the transcription start point (1).
Surprisingly, RNAP holoenzyme lacking both ␣CTDs and 70 region 4.2 produced no footprints on extended Ϫ10 galP1 or galP1-35 promoters (Fig. 2, A and B, lane 7). Since the removal of these RNAP domains individually had no effect on promoter complex formation, we conclude that on galP1 and galP1-35, ␣CTD, and 70 region 4.2 interactions with upstream promoter DNA can substitute for each other. However, the removal of both of these interactions prevents promoter complex formation even in the context of a consensus extended Ϫ10 promoter.
Stability of Promoter Complexes-The dissociation kinetics of promoter complexes formed on galP1 and its two active derivatives, galP1-35 and galP1-35-TG, was investigated. Promoter complexes were formed, challenged with DNA competitor heparin, and the amount of complexes that survived heparin chal-

FIG. 2. Promoter complex formation on galP1 promoter derivatives using mutant RNAP holoenzymes lacking ␣CTD and/or 70 region 4.2.
Promoter complexes were formed at 37°C using the indicated RNAP holoenzymes on DNA fragments containing indicated functional galP1 promoter derivatives. Promoter fragments were 32 P-endlabeled at the top (non-template strand). Promoter complexes were footprinted with DNase I, reaction products were separated by electrophoresis in a 7% sequencing gel and revealed by autoradiography. lenge was determined by withdrawing reaction aliquots at various times after heparin addition and supplementing them with the CpA primer and radioactive UTP substrate to allow the production of radioactively labeled abortive transcript CpApU. Reaction products were separated by denaturing PAGE and the decline of CpApU synthesis over time was monitored. The results are presented in Fig. 3A. As can be seen, in the case of the wild-type RNAP promoter complexes on galP1-35, no decline in abortive synthesis was observed even after 90-min incubation in the presence of heparin. Complexes on galP1-35-TG were the least stable, and decayed with a half-life of ϳ30 min. Complexes formed on galP1 had intermediate stability and decayed with a half-life of ϳ90 min. The result is consistent with the idea that interactions between 70 region 2.5 and the TG motif contribute to promoter complex stability. The results also suggest that specific interactions between 70 region 4.2 and the Ϫ35 promoter   FIG. 3. Heparin-induced dissociation of RNAP-promoter complexes formed on galP1 promoter derivatives. A, promoter complexes were formed on indicated promoter fragments using wild-type 70 RNAP holoenzyme. Complexes were challenged with heparin (ϩhep). At the indicated times after heparin addition, reaction aliquots were withdrawn and combined with substrate mixture containing CpA and [␣-32 P]UTP. Reactions were allowed to proceed for 10 min, reaction product, CpApU, was separated by denaturing PAGE and quantified. Each experiment was repeated at least three times. Mean values are presented. On the right, an autoradiograph of a representative gel is shown. B, experiment was performed as in A using mutant RNAP holoenzymes. consensus element increase promoter complexes ability to withstand heparin challenge (see, however, below). We next measured the dissociation kinetics of promoter complexes formed by RNAP mutants (Fig. 3B). As can be seen, 70 holoenzymes on galP1, galP1-35 and galP1-35-TG dissociated with the same kinetics irrespective of the presence of ␣CTD(s). Similarly, the dissociation kinetics of promoter complexes formed by 565 holoenzymes on galP1 and galP1-35 did not depend on the presence of ␣CTD. Thus, ␣CTD interactions with upstream galP1 promoter DNA do not contribute to promoter complex ability to withstand heparin challenge, in agreement with data obtained on other promoters (23).
Curiously, comparison of 70 -(1-565) and 70 RNAP complexes on galP1 (or galP1-35) revealed no differences in stability. Furthermore, 70 -(1-565) RNAP-galP1-35 complexes were more stable than 70 -(1-565) RNAP-galP1 complexes. It therefore follows that increased stability of galP1-35 complexes compared with galP1 complexes is not caused by 70 region 4.2 interactions with the Ϫ35 promoter consensus element. We conclude that 70 region 4.2 interactions with galP1 promoter DNA do not contribute to promoter complex stability as measured by heparin challenge assay. Increased stability of galP1-35 complexes compared with galP1 complexes must be due to intrinsic properties of galP1-35 DNA and/or due to galP1-35 DNA interactions with RNAP domains other than ␣CTD or 70 region 4.2.
Promoter Complex Formation at Low Temperature-Promoter complexes formed on galP1 at temperatures as low as 6°C are open, as judged by KMnO 4 probing (9). This is an unusual behavior since promoter complexes formed on most promoters are insensitive to KMnO 4 at this temperature, and are therefore closed. KMnO 4 probing of promoter complexes formed by the wild-type RNAP at 6°C revealed the presence of KMnO 4 -sensitive bands in galP1-35, but not in galP1-35-TG complexes (Fig. 4A, compare lanes 2 and 6). Complexes formed on galP1 were also open at 6°C (data not shown). We conclude, in agreement with earlier data (12), that the TG motif is essential for low temperature opening.
To determine whether RNAP contacts with upstream DNA contribute to open complex formation at low temperature, we performed KMnO 4 probing of galP1 complexes formed by different 70 and 70 -(1-565) RNAP holoenzymes at 37 and 6°C. Since RNAP holoenzymes containing one ␣CTD behaved indistinguishably from the corresponding holoenzymes containing two ␣CTDs (data not shown), we only present data obtained with holoenzymes that have or lack both ␣CTDs. As expected from DNase I footprinting results, all RNAP holoenzymes, with the exception of the mutant lacking both ␣CTDs and 70 region 4.2, formed open complexes at 37°C (Fig. 4B, compare lanes 2,  3, and 6 with lane 7). 70 holoenzymes formed open promoter complexes at low temperature irrespective of the presence of ␣CTD (Fig. 4B, compare lanes 4 and 5). 70 -(1-565) holoenzyme did not form open complexes at low temperature irrespective of the presence of ␣CTD (Fig. 4B, lanes 8 and 9). Thus, at low temperature, upstream DNA binding by 70 region 4.2 but not by ␣CTD is necessary for open complex formation on galP1.
The inability of RNAP holoenzyme lacking 70 region 4.2 to form open complexes at 6°C can be due to its inability to perform localized DNA melting, which can become rate-limiting at lower temperatures. Alternatively, the lack of open complex formation at lower temperatures can be caused by the inability of the mutant enzyme to bind to promoter DNA. To determine which of these possibilities is realized, we first footprinted galP1 complexes formed at 6°C with DNase I (Fig. 4C).
The results indicate that only 70 holoenzymes protected DNA from DNase I digestion at low temperature. The overall pattern FIG. 4. Promoter complex formation and promoter opening at low temperature. A, promoter complexes were formed at indicated temperatures using the indicated 32 P-endlabeled (top-strand) promoter-containing fragments and wild-type 70 RNAP holoenzyme. Complexes were probed with KMnO 4 , reaction products were separated on a sequencing gel and revealed by autoradiography. B and C, the indicated RNAP holoenzymes were used to form complexes with 32 P-endlabeled (top strand) galP1 promoter containing fragment at 6 or 37°C (B) or 6°C (C), probed with KMnO 4 (B) or footprinted with DNase I (C), and reaction products were analyzed as above. D, promoter complexes were formed at 6°C using the indicated RNAP holoenzymes with 32 P-endlabeled galP1 promoter containing fragment. Reaction products were separated, with or without heparin challenge, on a native 5% polyacrylamide gel. The gel was run at 6°C. Reaction products were revealed by autoradiography.
of protection was very similar to the 37°C complexes protection pattern (compare Fig. 4C, lanes 2 and 3, with Fig. 2A, lanes 2  and 4). In contrast, the pattern of DNase I digestion of promoter DNA in the presence of 70 -(1-565) RNAP holoenzymes was similar to the pattern obtained with naked DNA.
The results presented in Fig. 4C suggest that the low temperature melting defect exhibited by 70 -(1-565) RNAP holoenzymes may simply be a consequence of their inability to bind to promoter DNA at these conditions. An alternative explanation would be that closed promoter complexes formed by enzymes that lack 70 region 4.2 are so short-lived that they are not detected during the 15-s DNase I footprinting reaction. We therefore attempted to follow closed complex formation using gel retardation assay, hoping that the caging effect during gel electrophoresis may stabilize complexes that were not detected by footprinting (Fig. 4D). RNAP holoenzymes were combined with radioactively labeled galP1 promoter fragment at 6°C, complexes were allowed to form, and reactions were separated by electrophoresis in a native polyacrylamide gel kept at 6°C. To monitor open complex formation, reactions were treated with heparin prior to loading on the gel. Reactions that were subjected to electrophoresis without heparin treatment were expected to reveal the presence of both closed and open promoter complexes, as well as complexes formed due to endbinding and other non-promoter complexes. To decrease the amounts of non-promoter complexes, reactions were set up in the presence of excess promoter DNA over RNAP holoenzymes. Electrophoretic separation of reaction products containing 70 holoenzymes revealed, in the absence of heparin challenge, free promoter DNA and two shifted bands (Fig. 4D, lanes 2 and 3). Both shifted bands disappeared upon addition of heparin and a single band with intermediate mobility appeared. This heparin-resistant band must correspond to open promoter complexes detected by KMnO 4 probing and DNase I footprinting (Fig. 4, A-C). Bands seen in the absence of heparin must also correspond to open complexes, because addition of heparin did not result in significant increase in free DNA present in reactions (Fig. 4D, compare lanes 2 and 3 to lanes 3 and 5, correspondingly). Apparently, the addition of heparin changes the complex mobility, either because heparin interacts with the complex, or because some RNAP-DNA contacts are broken in the presence of heparin.
As expected from KMnO 4 probing and DNase I footprinting data, little or no heparin-resistant complexes was observed when 70 -(1-565) RNAP holoenzymes were used for promoter complex formation (Fig. 4D, lanes 9 and 10). A different situation was observed in the absence of heparin. The 70 -(1-565) RNAP holoenzyme that lacked both ␣CTDs was unable to shift promoter DNA (Fig. 4D, lane 8), while the enzyme that had wild-type ␣ produced two bands with the same mobility as those observed in 70 holoenzyme lanes (Fig. 4D, lane 7). We take these results as evidence that 70 -(1-565) holoenzyme reconstituted with wild-type RNAP core fails to form an open promoter complex at low temperature because it fails to isomerize to open promoter complex. One the other hand, the double mutant holoenzyme that lacks 70 region 4.2 and ␣CTDs fails to bind promoter DNA.
Effect of T4 AsiA on Promoter Complex Formation-Bacteriophage T4 anti-AsiA binds to two sites in 70 , in regions 4.1 and 4.2. AsiA binding to 70 region 4.2 prevents the recognition of the Ϫ10/Ϫ35 class promoters by interfering with 70 region 4.2 interaction with the Ϫ35 promoter element (19,24). AsiA does not prevent promoter complex formation on extended Ϫ10 promoters because region 4.2 interactions with upstream DNA are not critical on these promoters (19,24). Interaction of AsiA with region 4.1 (in the context of 70 -(1-565) holoenzyme) stim-ulates transcription from bacteriophage T4 extended Ϫ10 class middle promoters through an undefined mechanism (22). We studied promoter complexes formed by various RNAP holoenzymes in the presence of AsiA. Promoter complexes were formed in the presence of 5-fold excess of AsiA over 70 present in the reaction. The amounts were sufficient to completely block promoter complex formation on galP1-35-TG by wild-type RNAP holoenzyme (data not shown). The DNase I footprinting experiment presented in Fig. 5A was conducted on the galP1-35 promoter at 37°C. An identical result was obtained on galP1 (data not shown). As can be seen, promoter complexes formed by 70 RNAP holoenzymes in the presence of AsiA differed from complexes formed in its absence primarily by increased accessibility of DNA positions Ϫ34 and Ϫ35 to DNase I attack (compare lanes 2 and 3, and lanes 4 and 5). In addition, AsiA abolished DNase I hypersensitivity at Ϫ67 and Ϫ57, consistent with the results in Fig. 2 that show that hypersensitivity of these sites requires full-length 70 . The addition of AsiA to 70 RNAP lacking both ␣CTDs diminished, but did not completely prevent promoter complex formation (Fig. 5A, compare lanes 6  and 7). Since no promoter complexes were formed when both ␣CTDs and 70 region 4.2 were missing (Figs. 2, A and B and 4B), the result implies that 70 region 4.2 is able to interact with DNA even in the presence of bound AsiA. Alternatively, AsiA itself may interact with DNA (25).
Promoter complexes formed by ␣CTD-containing 70 -(1-565) RNAP holoenzymes in the presence of AsiA differed from complexes formed in its absence primarily by decreased accessibility of DNA position Ϫ37 and increased accessibility of DNA positions Ϫ34 and Ϫ35 (Fig. 5B). As a result, ␣CTDcontaining 70 holoenzymes and 70 -(1-565) holoenzymes produced very similar footprints in the presence of AsiA. Addition of AsiA did not allow the 70 -(1-565) holoenzyme that lacked both ␣CTDs to form a promoter complex (Fig. 5B,  compare lanes 6 and 7).
Our last experiment investigated the effect of AsiA on low temperature promoter opening by 70 -(1-565) holoenzyme. As shown above (Fig. 5B), the binding defect of RNAP that lacked 70 region 4.2 and ␣CTDs could not be overcome by the addition of AsiA at 37°C. Likewise, AsiA had no effect on promoter binding by this mutant enzyme at 6°C (data not shown). Since RNAP containing the wild-type ␣ but lacking 70 region 4.2 appeared to bind promoter DNA normally at 6°C, we asked if AsiA could stimulate open complex formation by this enzyme. The addition of AsiA stimulated promoter opening (Fig. 5C,  compare lanes 4 and 5) and thus allowed to partially overcome the defect caused by the absence of 70 region 4.2.

DISCUSSION
Several conclusions about the galP1 promoter complex formation by E. coli RNAP 70 holoenzymes can be drawn from the experiments presented herewith. First, at physiological temperature, the ␣CTD upstream promoter DNA interactions and 70 region 4.2 upstream promoter DNA interactions are functionally interchangeable on galP1 derivatives containing extended Ϫ10 motif. At least one of these interactions is necessary for efficient open promoter complex formation. Mechanistically, each of these interactions may serve to recruit RNAP holoenzyme to promoter and allow 70 region 2.5 to engage the extended Ϫ10 element. The presence of these upstream interactions has no effect of the dissociation kinetics of preformed promoter complex as measured by heparin challenge experiments.
␣CTD upstream promoter DNA interactions and 70 region 4.2 Ϫ35 promoter consensus element interactions are functionally distinct on a galP1 promoter derivative lacking the extended Ϫ10 TG motif: 70 region 4.2 Ϫ35 promoter consensus element interaction is required for open complex formation on this promoter, while ␣CTD upstream promoter DNA interaction is dispensable. It is possible that this difference is simply a consequence of different strengths of ␣CTD and 70 region 4.2 interactions with their targets. Alternatively, 70 region 4.2 Ϫ35 promoter element interaction but not ␣CTD upstream promoter DNA interaction may allow 70 region 2.4 to productively engage the Ϫ10 promoter element and initiate promoter melting in the absence of specific interactions between region 2.5 and the TG motif. The differential ability of ␣CTD and 70 region 4.2 DNA interactions to allow to engage the Ϫ10 promoter element could be due to the fact that ␣CTDs are attached to RNAP through unstructured and highly flexible tethers, while 70 region 4.2 is connected to RNAP through an extensive protein-protein interaction with the ␤ flap domain (2,26,27). The ␤ flap has only limited mobility and may therefore allow "signal transduction" from the Ϫ35 promoter element to the Ϫ10 promoter element.
Analysis of promoter complex formation at low temperature of 6°C demonstrates that even on an extended Ϫ10 promoter 70 region 4.2 and ␣CTD play different roles in promoter complex formation. Our results show that even though the extended Ϫ10 TG motif is necessary for low temperature galP1 open complex formation, it is not sufficient, and 70 region 4.2 (and presumably its interaction with upstream DNA) is also required. Both sequence-specific and nonspecific 70 region 4.2 interactions are sufficient for low temperature open complex formation in the context of extended Ϫ10 promoter since low temperature complexes form on galP1-35 and galP1. The latter promoter lacks a functional Ϫ35 promoter element as evidenced by the inactivity of the galP1-TG promoter (Fig. 1). In contrast to strict requirement for 70 region 4.2 for promoter complex formation at low temperature, ␣CTDs are not re-quired, in agreement with accompanying work of Kamali-Moghaddam and Geiduschek (28). In this respect, low temperature complex formation on extended Ϫ10 galP1 derivatives is similar to the situation observed on a Ϫ10/Ϫ35 derivative at 37°C.
Since ␣CTD does not substitute for 70 region 4.2 at 6°C, 70 region 4.2 somehow contribute to low temperature DNA opening. The gel retardation results appear to support change for this, since an enzyme reconstituted from wild-type RNAP core and 70 lacking region 4.2 forms heparin-sensitive closed complexes but fails to isomerize to heparin-resistant complexes. Since RNAP holoenzyme reconstituted from wild-type core and 70 lacking region 4.2 readily forms open complexes at physiological temperature, there must exist a cold-sensitive step in promoter complex formation that is overcome in the presence of 70 region 4.2. The movement of the ␤ subunit flap domain could be a potential candidate for such a step. We propose that in the context of 70 -(1-565) holoenzyme, the ␤ flap can adopt several (at least two) conformations relative to the rest of RNAP core molecule. One of these conformations is not productive and does not allow promoter complex formation on extended Ϫ10 promoters. This conformation is infrequent at 37°C, but becomes predominant at low temperature. In the context of the 70 RNAP holoenzyme, the ␤ flap interaction with 70 region 4.2 interaction stabilizes an alternative active conformation of the ␤ flap, at both the low and the high temperatures. According to this view, AsiA may stimulate low temperature melting by 70 -(1-565) holoenzyme not because AsiA interacts with promoter DNA but because AsiA helps stabilize the active conformation of the ␤ flap at low temperature. Indeed, specific binding of AsiA to the ␤ flap has recently been reported (29). FIG. 5. Effect of T4 AsiA on promoter complex formation. Promoter complexes on the 32 P-endlabeled (top-strand) galP1-35 promotercontaining DNA fragment were formed at 37°C using the indicated enzymes in the presence or in the absence of T4 AsiA. Complexes were treated with DNase I and analyzed as above. A, effect of AsiA on complexes formed by holoenzymes containing 70 . B, effect of AsiA on complexes formed by holoenzymes containing 70 -(1-565). C, promoter complexes were formed at 6°C using the indicated enzymes in the presence or in the absence of T4 AsiA. Complexes were probed with KMnO 4 and analyzed as above.