Dissection of Two Hallmarks of the Open Promoter Complex by Mutation in an RNA Polymerase Core Subunit*

Deletion of 10 evolutionarily conserved amino acids from the β subunit of Escherichia coli RNA polymerase leads to a mutant enzyme that is unable to efficiently hold onto DNA. Open promoter complexes formed by the mutant enzyme are in rapid equilibrium with closed complexes and, unlike the wild-type complexes, are highly sensitive to the DNA competitor heparin (Martin, E., Sagitov, V., Burova, E., Nikiforov, V., and Goldfarb, A. (1992)J. Biol. Chem. 267, 20175–20180). Here we show that despite this instability, the mutant enzyme forms partially open complexes at temperatures as low as 0 °C when the wild-type complex is fully closed. Thus, the two hallmarks of the open promoter complex, the stability toward a challenge with DNA competitors and the sensitivity toward low temperature, can be uncoupled by mutation and may be independent in the wild-type complex. We use the high resolution structure of Thermus aquaticus RNA polymerase core to build a functional model of promoter complex formation that accounts for the observed defects of the E. coli RNA polymerase mutants.

The formation of the transcription-competent binary promoter complex by Escherichia coli RNA polymerase (RNAP) 1 holoenzyme is a complex, multi-step process. Kinetic analysis using quantitative abortive initiation assays demonstrates that when RNAP first recognizes the promoter it forms the relatively unstable, "closed" complex (reviewed in Ref. 1). The closed complex is fully reversible: it easily dissociates back to free RNAP and DNA and is therefore highly sensitive to DNA competitors such as heparin, which bind free RNAP and prevent rebinding to promoter DNA. If RNAP commits itself to the template, the closed complex undergoes a series of poorly characterized conformational changes that lead to the formation of the transcription-competent "open" complex. Kinetic analysis at standard (37°C) temperature indicates the existence of at least one kinetically significant, short-lived intermediate between the closed and the open complex (2). On most promoters the open complex formed at standard temperature is essentially irreversible and is fully resistant to DNA competitors.
Direct structural analysis of closed and intermediate complexes at standard temperature is complicated since they are short-lived. In contrast, the open complex, which is the dominant species at 37°C on most promoters, can be easily studied. DNase I footprinting of open complexes at several promoters shows extensive DNA protection from Ϫ55 to ϩ20 relative to the transcription start point at ϩ1. On most promoters, a hypersensitive DNase I site exists in the otherwise well protected area at about position Ϫ25 (3). This hypersensitive site has been interpreted as the site of sharp bending in the DNA (3). Indeed, since the amount of DNA protected in the open complex exceeds the largest linear dimension of RNAP holoenzyme, DNA in the complex must be bent or curved (2,4). Probing with single strand-specific reagents such as KMnO 4 demonstrates that pyrimidines between Ϫ12 to ϩ2 are accessible in the open complex but not in the closed complex. Thus, a fragment of promoter DNA in the open complex is either severely distorted or melted (the so-called transcription bubble).
On most promoters the open complex is temperature-sensitive and is rapidly and cooperatively inactivated when the reaction temperature is lowered to about 20°C or below (5). Footprinting analysis reveals several distinct binary promoter complexes at lower temperatures (6 -9). All these additional complexes are characterized by the absence of DNA reactivity to single strand-specific reagents. In addition, the region of DNA protected by RNAP from DNase I attack is reduced at lower temperatures, mostly in the region downstream of transcription start point, and the degree of protection also varies. The dominant binary complex at 0°C shows an alternating pattern of protection and deprotection between the positions Ϫ55 to Ϫ5, suggesting that RNAP interacts with only one side of the DNA molecule in this complex (7)(8)(9). The 0°C binary complex is highly sensitive to DNA competitors, whereas some of the complexes obtained at higher, 10 -15°C, temperatures are relatively resistant.
The mechanistic significance of binary complexes present at lower temperatures is not known. The limited number of careful kinetic and thermodynamic studies on RNAP-promoter interactions on P R and lac UV5 promoters suggests that binary complexes trapped at lower temperatures may correspond to transient kinetic intermediates of the open complex formation mechanism at 37°C. Thus, the 0°C complex may correspond to kinetically defined, heparin-sensitive closed complex, whereas other trapped complexes can correspond to various stages of isomerization from fully closed to fully open conformation. However, recent detailed analysis revealed that DNase I footprints of the 0°C complex on the phage prmup-1 ⌬265 promoter had both similarities to and differences from true closed complex formed at higher temperatures (10) and, thus, may be off the true promoter opening pathway.
Analysis of mutant RNAPs defective in promoter complex formation provides an alternative and complementary way to trap reaction intermediates and potentially can lead to a better understanding of structure/function relationships in transcription initiation. As expected, mutational analysis indicates that the specificity subunit contributes to the initial promoter recognition, initiation of DNA melting, and possibly promoter complex stability (11)(12)(13). However, mutations in the core enzyme that affect promoter complex stability and structure have also been isolated. RNAP carrying an engineered 10-amino acid deletion in the evolutionarily conserved segment C of the ␤ subunit, ⌬RV (14), was unable to efficiently hold onto promoter DNA, resulting in heparin-sensitive open complexes. More recent studies resulted in isolation of additional mutations that destabilize RNAP-promoter complexes in vitro (15). One point mutation, 4348, occurred close to ⌬RV, underscoring the importance of ␤ segment C in open complex stability. Another engineered mutation, ⌬ (186 -433), removed most of the ␤-dispensable region I, which is adjacent to segment C (Fig.  1, Ref. 16), and resulted in RNAP that formed open promoter complexes that were more resistant to low temperature inactivation than the wild-type open complexes. The mutant complexes remained sensitive to KMnO 4 attack even at Ϫ20°C, whereas the wild-type complexes became unreactive below 15°C, as expected (16). Interestingly, the destabilizing ⌬RV mutation, and the stabilizing ⌬(186 -433) are immediately adjacent to each other in the ␤ subunit primary sequence. Here we reinvestigate promoter complex formation by RNAP carrying the ⌬RV mutation. We confirm that ⌬RV decreases open promoter complex resistance to heparin. Surprisingly, ⌬RV also increases open promoter complex ability to withstand low temperature and, thus, resembles ⌬(186 -433). Thus, the two hallmarks of the open promoter complex, stability toward DNA competitors and sensitivity toward temperature downshift, can be uncoupled by mutation.

MATERIALS AND METHODS
Genetic Engineering-The starting plasmid for genetic manipulations was pMKSe2 containing the wild-type rpoB gene under control of lac promoter (17). The ⌬RV mutation was constructed by treating pMKSe2 with EcoRV (introduces two in-frame cuts in rpoB that are 30 base pairs apart (14)) and religating the plasmid. To construct the ⍀H 6 mutation, two complementary oligonucleotides, 5Ј-p-AGGATCCAT-CACCACCATCACCAC-3Ј (DN1) and 5Ј-p-GTGGTGATGGTGGTGAT-GGATCCT-3Ј (DN2), were used. Both oligonucleotides contain BamHI recognition site (underlined). DN1 and DN2 were annealed to each other and ligated to EcoRV-digested pMKSe2 plasmid. Ligase was then inactivated by heating, and reactions were treated with a high concentration of BamHI (does not cut pMKSe2). BamHI was inactivated by phenol extraction, and the plasmid was religated and transformed in E. coli cells. The DN1/DN2 linker when inserted in the right orientation should introduce eight amino acids, Arg-Ile-His 6 , in place of ␤ amino acids removed by EcoRV. Insertion in the opposite orientation introduces several stop codons in the rpoB reading frame, so that no fullsized ␤Ј can be produced. Recombinants containing the inserts in the right orientation were selected by SDS-polyacrylamide gel electrophoresis, and the insertion was verified by DNA sequencing. The function of mutant rpoB genes was tested in vivo using AJ 2005 strain as described (18).
RNAP Purification-The portion of rpoB harboring deletions was recloned from the original pMKSe2 background into pET15b-␤. In this plasmid, a complete rpoB is cloned between the NdeI and BamHI sites of pET15b (Novagen). The ␤ subunit expressed by pET15b-␤ or its derivatives contains an engineered N-terminal hexahistidine tag. pET15b-␤ and pET15b-␤ ⌬RV were transformed into XL1-blue E. coli cells, and transformants were grown in 4 liters of LB broth without the addition of isopropyl-1-thio-␤-D-galactopyranoside until late log phase. The cells were collected, and RNAP was purified by a combination of the standard purification procedure and metal ion affinity chromatography (19), concentrated by filtration through a C-100 concentrator (Amicon) to ϳ1 mg/ml, and stored in 50% glycerol storage buffer at Ϫ20°C. RNAP from cells carrying 4337 and 4348 mutations was purified using standard procedure (4). RNAP ⍀H6 was prepared by in vitro reconstitution exactly as described (20).
Bacillus subtilis RNAP was purified from a derivative of JH642 strain that contains hexahistidine tag genetically fused to the 3Ј end of the rpoC gene, coding for RNAP ␤Ј (kindly provided by Dr. C. P. Moran, Jr.). Cells were grown at 33°C in LB supplemented with spectinomycin (50 g/ml) to A 600 ϳ 0.8. Cells were collected by centrifugation, resuspended in ice-cold grinding buffer (10 mM Tris-HCl, pH 7.9, 0.5 M NaCl, 5% glycerol, 1 mM ␤-mercaptoethanol, 2.5 mM imidazole), and disrupted by sonication. The lysate was clarified by centrifugation (12,000 ϫ g, 15 min) and applied on a chelating Hi-Trap column (Amersham Pharmacia Biotech) loaded with Ni 2ϩ . The column was washed with grinding buffer containing 20 mM imidazole and then with the buffer containing 200 mM imidazole. The 200 mM imidazole fraction, which contained RNAP, was diluted 3-fold with TGE buffer (10 mM Tris-HCl, 5% glycerol, 0.1 mM EDTA) and applied on a 1-ml heparin Hi-Trap column (Amersham Pharmacia Biotech). After washing the column with TGE buffer containing 300 mM NaCl, RNAP was eluted with TGE containing 600 mM NaCl. The purified enzyme was concentrated by dialysis against storage buffer (20 mM Tris-HCl, pH 8.0, 300 mM NaCl, 5 mM ␤-mercaptoethanol, 50% glycerol) to ϳ1 mg/ml and stored at Ϫ20°C.
Footprinting Reactions-The 106-base pair EcoRI DNA fragment containing the T7 A2 promoter (Ϫ84 to ϩ32) was prepared as described (16). The fragment was 32 P-end-labeled by filling-in EcoRI sticky ends with Klenow enzyme in the presence of [␣-32 P]dATP. The fragment was then treated with AccI (cuts at position Ϫ70) to obtain top strandlabeled fragment or HincII (cuts at position ϩ22) to obtain bottom strand-labeled fragment. Promoter complexes were formed in 20-l reactions containing 0.4 pmol of wt or mutant RNAP, 0.2 pmol of 32 P-end-labeled DNA fragment, 40 mM Tris-HCl, pH 7.9, 40 mM KCl, and 10 mM MgCl 2 . Reactions were preincubated for 15 min at 37°C. DNase footprinting reaction was initiated by the addition of DNase I (2 g/ml DNase I, Worthington). The reaction proceeded for 10 s at 37°C and was terminated by the addition of EDTA to 15 mM followed by phenol extraction and ethanol precipitation. For KMnO 4 probing, promoter complexes were treated with KMnO 4 (1 mM) for 15 s at 37°C. Reactions were terminated by the addition of ␤-mercaptoethanol to 300 mM followed by phenol extraction, ethanol precipitation, and 10% piperidine treatment. Where indicated, reaction mixtures were challenged by the addition of 100 g/ml heparin immediately before footprinting.
To footprint promoter complexes formed at low temperatures, RNAP and 32 P-end-labeled T7 A2 promoter-containing fragments were combined on ice and incubated for 15 min on ice, and the footprinting reaction was performed. The same conditions were used for KMnO 4 probing at low temperature, since control experiments demonstrated that KMnO 4 modification was complete after 15 s. Products of footprinting reactions were analyzed by urea-polyacrylamide gel electrophoresis (7 M urea, 6% polyacrylamide) followed by autoradiography.

RESULTS
Genetic Context of the Conserved Segment C in the ␤ Subunit-Segment C is one of the nine co-linear segments of high sequence conservation found in all ␤ subunit homologs. On the N-proximal side of segment C is dispensable region I (21), a long region of primary sequence (E. coli residues 149 -433) that is present in proteobacteria but is partially absent from homologs from other eubacteria, archaea, and eukaryotes (Fig. 1). Sequence C-terminal to C is not particularly conserved and is followed by the Rif region (Ref. 17 and residues 505-574, Fig. 1).
Mutations at or close to segment C used in this study are shown on Fig. 1. ⌬RV removes amino acids 436 -445 at the left side of segment C (14). The region affected by the deletion contains several highly conserved amino acids and has sequence similarity with barnase-type bacterial RNases. ⌬(186 -433) is a long deletion in dispensable region I and is immediately adjacent to ⌬RV (21). 4348, selected for the ability to confer prototrophy on relA spoT strains occurred close to ⌬RV, changing conserved Arg 454 to His (15). ⍀H 6 is a new mutation. It was constructed by site-specific mutagenesis and introduces eight unnatural amino acids, Arg-Ile-His 6 , in place of segment C amino acids removed by ⌬RV (see "Materials and Methods").
The in vivo phenotype of ⍀H 6 was tested in the E. coli AJ2005 tester strain (18). AJ2005 cells have an amber mutation in the chromosomal copy of rpoB that is suppressed by the supU suppresser carried by an F-factor. The F-factor also carries the lacZ gene, and AJ2005 cells are therefore Lac ϩ . When a plasmid bearing a functional rpoB allele is introduced into AJ2005, the F-factor can be lost, and the plasmid-borne rpoB gene becomes the only source of the ␤ subunit in the cell. As a result, such cells, become Lac Ϫ , and they can be easily distinguished from parental cells using indicator media. AJ2005 cells transformed with the rpoB expression plasmid pMKSe2 bearing rpoB ϩ or ⍀H 6 segregated Lac Ϫ colonies (data not shown). In contrast, no Lac Ϫ colonies was observed when AJ2005 was transformed with pMKSe2 bearing ⌬RV (data not shown, Ref. 14). This result suggests that the insertion functionally compensates for defects caused by ⌬RV and that the total length of this part of segment C is more important for function than the sequence per se. The Open Promoter Complex Formed by RNAP ⌬RV Is Shortened in the Downstream Region-To better understand the marked differences in promoter complex stability caused by the two adjacent mutations in the ␤ subunit, ⌬RV and ⌬(186 -433), promoter complexes formed by RNAP ⌬RV , RNAP ⌬(186 -433) , and RNAP WT were studied by DNase I footprinting (Fig. 2). RNAP ⍀H6 was also included in this analysis.
The DNase I footprint of the RNAP WT on the A2 promoter of bacteriophage T7 was typical for open complexes on 70 promoters. Upstream of the transcription start site, protection started at around position Ϫ40, followed by a region of hypersensitivity at around position Ϫ25. The Ϫ10 promoter region was completely protected from DNase I attack, and this protection extended through the transcription start site to about position ϩ20 (Fig. 2A, lane 5). As described previously (16), RNAP ⌬(186 -433) produced an identical footprint upstream of the transcription start site; however, position ϩ1 was only partially protected, and there was no protection beyond position ϩ5 (Fig. 2A, lane 4). The RNAP ⌬RV footprint was also shortened in the downstream direction and, thus, was qualitatively similar to RNAP ⌬(186 -433) footprint. Unlike RNAP ⌬(186 -433) , RNAP ⌬RV failed to protect the promoter DNA fully, suggesting that significant portion of the enzyme dissociated from the promoter during the 15-s treatment with DNase (note that in this experiment we used 10 times as much RNAP ⌬RV as other enzymes). This result agrees well with the previous findings obtained by filter binding and quantitative abortive initiation assays (14). Kinetic analysis indicated that even at very high RNAP ⌬RV concentrations the steady-state level of mutant promoter complexes was 30% that of the wild type (14). In contrast, RNAP ⍀H6 footprint was indistinguishable from the wildtype footprint; it was fully extended; and protection of DNA was complete (lane 3), in agreement with the results of the in vivo phenotypic analysis (above).
The DNA strand separation in open complexes formed by the wild-type and mutant RNAPs was investigated using KMnO 4 probing. The wild-type and the ⍀H 6 complexes appeared the same, with thymines at Ϫ15, Ϫ12, Ϫ11, Ϫ9, Ϫ7, Ϫ5, Ϫ4, and Ϫ3, as well as cytosine at Ϫ14 being modified by KMnO 4 (Fig.  2B, lanes 3 and 5). As described elsewhere (16), the singlestranded region did not extend as far downstream in complexes formed by RNAP ⌬(186 -433) ; only thymines at positions Ϫ15, Ϫ12, Ϫ11, and Ϫ9 and cytosine at Ϫ14 were modified by KMnO 4 (Fig. 2B, lane 4). As expected, probing of RNAP ⌬RV complex revealed similar shortening of the transcription bubble (Fig. 2B, lane 2). RNAP ⌬RV Promoter Complexes Remain Open at 0°C-The results of the footprinting experiment suggest that in RNAP ⌬RV -T7 A2 promoter complexes, protein-DNA contacts in the downstream region are lost, and DNA melting does not extend beyond the Ϫ5 position relative to transcription start. RNAP ⌬(186 -433) also forms such shortened complexes, which are heparin-resistant and remain sensitive to permanganate at temperatures as low as Ϫ20°C (16). In contrast, the wild-type complexes close at around 15°C on this promoter (16). Since at 37°C RNAP ⌬RV and RNAP ⌬(186 -433) promoter looked similar, we tested if RNAP ⌬RV complexes also remain sensitive to permanganate at lower temperatures. Fig. 3 demonstrates that this was indeed the case. In this experiment, RNAP and radioactively labeled promoter fragment were combined at either 37 or 0°C and incubated for 15 min to allow the complex formation, and KMnO 4 probing was performed. As can be seen, there was no modification of thymines in the wild-type complex at 0°C (lane 3), indicating that the complex was closed, as expected. In contrast, in the ⌬RV complex thymines in the upstream portion of the transcription bubble remained sensitive to permanganate attack even at low temperature (lane 5). Assuming 100% opening at 37°C, more than 50% of the mutant complex was open at low temperature, similar to the value previously obtained for RNAP ⌬(186 -433) (16).

␤ Segment C Point Mutation That Destabilizes Open Complex Does Not Result in Temperature Resistance-Recently, an rpoB
4348 point mutation that changes evolutionarily conserved segment C Arg 454 to His and that renders RNAP-promoter complexes sensitive to heparin attack was described (15). To find out whether unstable open complexes are generally more resistant to low temperature, we purified the RNAP carrying the 4348 mutation, formed promoter complexes at 37 and 0°C, and probed them with KMnO 4 . The Arg 3 His substitution caused by the 4348 mutation is only 10 amino acids away of the ⌬RV deletion (Fig. 1), and we therefore expected that it destabilizes the complex through a similar mechanism. This expectation was not fulfilled. As can be seen, at 37°C the mutant enzyme produced a fully open "wild-type" complex, which was highly sensitive to heparin attack (Fig. 4A, lane 6). When probing was performed at 0°C, no sensitive thymines were observed, indicating that the mutant enzyme was unable to melt promoter DNA at these conditions (Fig. 4A, lane 7). Additional experiments established that RNAP ⍀H6 complexes were indistinguishable from the wild-type complexes, i.e. they were heparin-resistant at high temperature and closed at low temperature (data not shown). It was previously suggested that heparin may actively displace RNAP from promoter complex instead of simply acting as a trap for free RNAP. The experiment presented on Fig. 4 was repeated with poly[dA-dT] as a competitor, and the same result was obtained (data not shown).
Promoter Complexes Formed by RNAP from B. subtilis Are Shortened in the Downstream Region-In Gram-positive bacteria, the region on the N-terminal side of segment C contains insertions and deletions relative to the E. coli sequence (Fig. 1). It is conceivable that compared with the E. coli enzyme, insertions/deletions in these organisms change the position of segment C with respect to other RNAP functional sites. In agreement with this idea it was reported that RNAP from B. subtilis, which contains a long, 90-amino acid insertion relative to E. coli close to segment C (Fig. 1), as well several deletions closer to the N-terminal segment B, forms open promoter complexes that are highly sensitive to heparin (25). Moreover, DNase I footprinting established that promoter complexes formed by B. subtilis enzyme were shortened in the downstream region as compared with E. coli RNAP complexes (25). We purified B. subtilis RNAP and investigated the T7 A2 promoter complexes formed by this enzyme using KMnO 4 footprinting. Purified B. subtilis RNAP recognized the T7 A2 promoter with low efficiency, and high amounts of the enzyme were necessary to observe promoter complex formation. In agreement with the published data, B. subtilis complexes were sensitive to heparin (Fig. 4B, lane 5). Importantly, the pattern of KMnO 4 sensitivity in the transcription bubble in 37°C complexes formed by B. subtilis RNAP was shortened in the downstream direction as compared with the E. coli RNAP complexes and was highly similar or identical to RNAP ⌬(186 -433) and RNAP ⌬RV complexes (Fig. 4B, right). However, at 0°C the KMnO 4 sensitivity in B. subtilis complex disappeared (Fig. 4B, lane 6). We are currently performing deletion mutagenesis in cloned B. subtilis rpoB gene to see if deletions in the B. subtilis ␤ subunit ␤-dispensable region I can result in temperature resistance.

DISCUSSION
Open promoter complexes formed by E. coli RNAP at the physiological temperature of 37°C are highly resistant to polyanion heparin, which acts as a DNA competitor. When the reaction temperature is lowered to about 15°C, heparin-sensitive promoter complexes are formed (23). Thus, it appeared that heparin resistance of promoter complexes and the ability to partially melt promoter DNA are related to each other. One would expect then that mutations that render open complexes heparin-sensitive will also make them more sensitive to lower temperatures. Obviously, RNAP with a short deletion in the ␤-conserved segment C studied here, RNAP ⌬RV , does not meet this expectation; at 37°C the mutant complexes are as sensitive to heparin as closed complexes (14), yet they remain partially open at 0°C, when the wild-type complexes are fully closed. Thus, the two hallmarks of transcription-competent open promoter complex, resistance to inactivation by heparin and sensitivity to low temperature, can be uncoupled. Consequently, RNAP structural elements that underlie the localized melting of promoter DNA and ability to resist heparin challenge must also be distinct.
A seminal paper by Zhang et al. (26) reveals the structure of core RNAP from thermophilic bacterium Thermus aquaticus (Taq) at 3.3 Å resolution. The Taq molecule has a characteristic crab claw-like shape, with a deep, ϳ27-Å-wide channel separating the jaws of the claw. The catalytic Mg 2ϩ ion is located deep in the middle of the channel that likely binds the template DNA. The area of the 27-Å channel that is to the left of catalytic Mg 2ϩ in the representation of Fig. 5B is thought to contact the downstream, double-stranded DNA; the area to the right makes contacts with transcription bubble, DNA-RNA heteroduplex, and upstream DNA. The upper RNAP jaw is composed mostly of ␤ and is bilobate; the downstream lobe is composed entirely of ␤-dispensable region I and portions of conserved segment C; the second, upstream lobe is composed of the remainder of conserved segment C and segment D. The region of Taq ␤ that corresponds to E. coli positions 186 -433 is shown in orange in Fig. 5; Taq amino acids corresponding to those removed by ⌬RV are shown in red. As can be seen, ⌬(186 -433) entirely removes the downstream lobe, whereas ⌬RV removes amino acids at the outside (downstream) edge of this lobe.
We propose that the bilobed structure of the top RNAP jaw holds the key to the understanding of biochemical phenotypes of the mutant enzymes studied here. We envision that both lobes make interactions with promoter DNA and that these interactions occur in a concerted, interdependent manner. The upstream lobe cooperates with the subunit (binds to the coiled-coil structure of ␤Ј located immediately below (Ref. 27, Fig. 5C)) and maintains the initial strand separation at the upstream edge of the transcription bubble. We propose that these upstream protein-DNA interactions are temperature-insensitive and that locking of the upstream jaw also engages, via an allosteric mechanism, the downstream lobe, which results in extended footprint and propagation of melting in the downstream direction. The allosteric mechanism that controls the movement of both lobes is organized such that only two conformations are stable; both lobes are unlocked (i.e. closed complex) or both lobes are locked (open complex). We propose that the locking of the downstream lobe is temperature-sensitive. The ⌬(186 -433) mutation removes the downstream lobe; ⌬RV leaves it intact but destroys the allosteric mechanism. As a result, promoter complexes formed by mutant enzymes are held together by upstream lobe-promoter DNA interactions; they are shortened in the downstream direction, are partially melted, and are temperature-resistant. In the wild-type E. coli RNAP complex or in the ⌬(186 -433) complex the upstream interactions are both temperature-resistant and heparin-resistant. However, mutations such as R454H (shown in cyan and space-fill on Fig. 5), which occurred in the cleft between the two lobes, can induce heparin sensitivity. E. coli RNAP with lesions at ␤ positions 531, 532, and 563 form heparin-sensitive complexes (28). In Taq, the corresponding positions (shown in magenta and space-fill on Fig. 5) are also located between the two lobes, behind the amino acid corresponding to E. coli Arg 454 . Mutational changes in these positions probably weaken DNA contacts made by the upstream lobe and, thus, make the complex sensitive to competitor. The ⌬RV deletion occurred immediately adjacent to the linker, which connects the two ␤ lobes (shown in cyan and ribbons on Fig. 5). Thus, ⌬RV probably affects the DNA interactions of both lobes, which explains its heparin sensitivity and temperature resistance.
In the presence of NTPs the ⌬RV and ⌬(186 -433) complexes appear indistinguishable from the wild-type enzyme complexes but different from the extended wild-type complex formed in the absence of NTP (14,16). Thus, other RNAP segments, most likely in ␤Ј, must participate in these additional interactions with downstream DNA.
Our "two stroke" model of promoter opening is supported by data of Guo and Gralla (24), who studied the interaction of RNAP holoenzyme with artificial DNA templates containing structures mimicking the upstream end of the transcription bubble. Consistent with our model, RNAP complexes formed with these structures are heparin-resistant and temperatureinsensitive. Mustaev and co-workers (29) used extensive protein-DNA cross-linking to build a more detailed model of elongation complex using structural information from Taq RNAP core enzyme. In agreement with our model, the authors found that large scale movements of ␤ lobes toward the DNA are necessary to achieve tight complementary fit between the DNA and the roof of the main RNAP channel.
Curiously, no partially melted complexes were detected when wild-type E. coli RNAP complexes trapped at lower temperatures were analyzed (5-10). Thus, the results obtained with mutant enzymes may complement the existing data ob-tained by temperature trapping. Real time structural, kinetic, and thermodynamic analyses of promoter complex formation by the wild-type RNAP as a function of conditions will be needed to characterize true intermediates of promoter opening and to compare them to complexes obtained using mutant enzymes and temperature trapping.