Stable DNA Opening within Open Promoter Complexes Is Mediated by the RNA Polymerase (cid:1) (cid:1) -Jaw Domain *

DNA opening for transcription-competent open promoter complex (OC) formation by the bacterial RNA polymerase (RNAP) reliesuponacomplexnetworkofinteractionsbetweenthestructur-ally conserved and flexible modules of the catalytic (cid:1) and (cid:1) (cid:1) -sub-units, RNAP-associated (cid:2) -subunit, and the DNA. Here, we show that one such module, the (cid:1) (cid:1) -jaw, functions to stabilize the OC. In OCs formed by the major (cid:2) 70 -RNAP, the stabilizing role of the (cid:1) (cid:1) -jaw is not restricted to any particular melted DNA segment. In contrast, in OCs formed by the major variant (cid:2) 54 -RNAP, the (cid:1) (cid:1) -jaw and a conserved (cid:2) 54 regulatory domain co-operate to stabilize the melted DNA segment immediately upstream of the transcription start site. Clearly, regulated communication between the mobile modules of the RNAP and the functional domain(s) of the (cid:2) subunit is required for stable DNA opening.

DNA opening for transcription-competent open promoter complex (OC) formation by the bacterial RNA polymerase (RNAP) relies upon a complex network of interactions between the structurally conserved and flexible modules of the catalytic ␤ and ␤-subunits, RNAP-associated -subunit, and the DNA. Here, we show that one such module, the ␤-jaw, functions to stabilize the OC. In OCs formed by the major 70 -RNAP, the stabilizing role of the ␤-jaw is not restricted to any particular melted DNA segment. In contrast, in OCs formed by the major variant 54 -RNAP, the ␤-jaw and a conserved 54 regulatory domain co-operate to stabilize the melted DNA segment immediately upstream of the transcription start site. Clearly, regulated communication between the mobile modules of the RNAP and the functional domain(s) of the subunit is required for stable DNA opening.
The DNA-dependent RNA polymerase (RNAP) 3 is central to all steps of transcription, the most regulated stage of gene expression. The core enzymes of multisubunit RNAPs from bacteria, archaea, and eukaryotes display striking structural similarity to each other. For promoter-specific and regulated transcription to occur, the RNAP core needs to associate with auxiliary transcription initiation factors. In bacteria, RNAP core (subunit composition: ␣ 2 ␤␤Ј, E) associates with a sigma () subunit to form an RNAP holoenzyme (subunit composition: ␣ 2 ␤␤Ј, E), which is capable of promoter-specific transcription initiation. The major type of bacterial RNAP holoenzyme contains a subunit belonging to the 70 -class. The major variant bacterial RNAP holoenzyme contains an unrelated subunit, 54 , which is the sole representative of the 54 -class. The processes of transcription initiation by E 70 and E 54 are mechanistically distinct (1,2). Specific transcription initiation requires that both types of Es bind to their respective promoters to form a closed complex. In the case of holoenzymes containing 70 -class subunits, the closed complex usually spontaneously isomerizes to form a transcription-competent open complex, in which the DNA strands are separated and placed in the catalytic cleft of the RNAP. In contrast, RNAP holoenzyme containing 54 relies on specialized activator proteins that belong to the AAA (ATPases associated with various cellular activities) family to convert the closed complex to the open complex. The AAA activator binds to enhancer-like DNA sequences located upstream of E 54 promoters and upon ATP hydrolysis promotes open complex formation (2). A class of mutant 54 exists called activator-bypass 54 , which do not require activation to form open complexes in vitro (2).
In both types of bacterial RNAPs, transcription-competent open complex formation is accompanied by several conformational changes in the subunit and structurally conserved mobile modules of the core RNAP (␤-lobes, ␤-flap, ␤Ј-clamp, and the ␤Ј-jaw) (3)(4)(5)(6). These conformational changes trigger DNA strand separation and/or accommodate the separated strands of the promoter DNA within the catalytic cleft of the RNAP. In a model of the transcription elongation complex formed by the Escherichia coli RNAP, the double-stranded DNA downstream of the catalytic center is positioned in a trough formed by the ␤Ј-jaw and the ␤Ј-clamp modules (7). The E. coli RNAP containing a deletion in the ␤Ј-jaw (⌬1149 -1190; Fig. 1A) forms unstable E 70 open complexes, although the property is evident only on some 70 -dependent promoters (8). The ␤Ј-jaw is required for several other transcription-related activities including transcriptional pausing, replication of ColEI-type plasmids, and bacteriophage M13 minus-strand DNA synthesis (8,9). The ␤Ј-jaw is also bound by the T7 bacteriophage gp2 protein, a strong inhibitor of 70 -dependent transcription from most promoters (10,11).
The bacterial ␤Ј-jaw is the structural homologue of the RPB1 jawlobe of the eukaryotic RNAPII and its role in transcription has not been clearly determined in either RNAPs. In the bacterial RNAP, the role of the ␤Ј-jaw in binding, closed complex formation, and in steps leading to open complex formation have not been established. How the deletion of the ␤Ј-jaw affects open complex stability is unknown. Here, we investigate the role of the ␤Ј-jaw in open complex formation by three forms of the bacterial RNAP: E 54 , activator-bypass E 54 , and E 70 . The results of this study provide insights into how DNA opening and open complex formation by the two mechanistically distinct types of the bacterial RNAP might occur; and provides one example of how regulated communication between the mobile modules of the RNAP and a functional domain of the subunit is required to establish the transcription proficient open complex.

EXPERIMENTAL PROCEDURES
Proteins and Plasmids-Wild-type E. coli RNAP was purchased from Epicenter Technology (Madison, WI). E. coli RNAP deleted for ␤Ј residues 1149 -1190 was purified as described (12). To ascertain that equal quantities of wild-type and mutant core RNAP are used in the reactions, samples of each RNAP preparation were checked on a SDS-PAGE gel. Wild-type and mutant 54 and PspF-(1-275) were purified as described (13). The activation-defective mutant PspF-(1-275) harbored the T86A mutation and was purified and characterized as described (14). 54 containing a carboxyl-terminal RRASV tag (to label with [ 32 P]ATP using heart-muscle kinase) was used to generate 32 P-labeled 54 . Briefly, 2 g of 54 -RRASV was labeled by using 50 Ci (5,000 Ci/mmol) of [␥-32 P]ATP and 0.5 unit of heart-muscle kinase for 10 min at 37°C in 20 mM Tris, pH 7.5, 100 mM NaCl, 12 mM MgCl 2 , 4 mM dithiothreitol in a final volume of 10 l. Aliquots of labeled protein were stored at Ϫ70°C and used only once after thawing. The plasmids for the in vitro transcription assays were constructed by cloning of PCR fragments containing the Sinorhizobium meliloti nifH, E. coli glnHp2, pspA, and pspG promoter sequence into pTE103 (13).
Native Gel Mobility Assay-All binding reactions (10 l) were conducted in STA buffer (25 mM Tris acetate, pH 8.0, 8 mM Mg-acetate, 10 mM KCl, and 3.5% (w/v) PEG 6000) at 37°C and are described in Ref. 13. Mutant and wild-type E 54 was reconstituted using 100 nM core RNAP and 1 M 54 (unless otherwise stated in the figure legends). Reactions containing [ 32 P] 54 also contained 5 g/l of ␣-lactoalbumin. The promoter probes used in the native gel mobility assay (at a final concentration of 10 nM) were made as described (13) and either 5Ј-labeled with 32 P or 5Ј-tagged with fluorescein. PspF-(1-275) was present (where indicated) at a final concentration of 5 M. The activator interaction assays were conducted as described (15). All reactions were separated on a 4.5 (w/v) native gel that was run for 80 min at 60 V. The gels were analyzed, and the complexes were quantified using a FLA-5000 PhosphorImager Fluorescent Image Analyzer. All native gel mobility assays were repeated at least twice, and the values shown in the figures represent an average of these replicates with an error range of Ϯ5%.
In Vitro Transcription Assays-Single and multiple round transcription reactions from plasmid-based templates (10 nM final concentra-tion) were conducted exactly as described (13). Where indicated, reactions were challenged with 100 g/ml of heparin for between 1 and 120 min (exact time specified in the figures and legends). Transcription elongation was initiated by adding a mixture containing 1 mM ATP, CTP, GTP, 0.05 mM UTP, and 3 Ci of [␣-32 P]UTP. The reactions were separated on a 6% denaturing gel and analyzed and quantified as above. For abortive transcription reactions, open complexes were formed on plasmid-based or linear templates and, where indicated, challenged with heparin, exactly as described above. For the abortive synthesis of a 4-nucleotide transcript (UGGG) from the S. meliloti nifH promoter, a mixture containing 0.5 mM initiating dinucleotide UpG, 4 Ci of [␣-32 P]GTP, and 0.5 M GTP was added for 20 min. For the abortive synthesis of a 3-nucleotide transcript (AAU) from the lacUV5 promoter, a mixture containing 0.5 mM ApA, 4 Ci of [␣-32 P]UTP, and 0.5 M UTP was added for 20 min. The reactions were separated on a 20% denaturing gel and quantified as above. All transcription assays were repeated at least three times, and the values shown in the figures represent an average of these replicates with an error range of Ϯ5-8%.
DNase I and KMnO 4 Footprinting Assays-DNase I and KMnO 4 footprinting assays on closed and open complexes (formed as described above) were conducted as described (16).

RESULTS
Removing ␤Ј Residues 1149 -1190 Affects the Ability of E 54 to Form Closed Promoter Complexes-A semi-quantitative native gel assay was used to test the ability of ⌬1149 -1190 E to bind radioactively labeled 54 , [ 32 P] 54 . A fixed amount of [ 32 P] 54 was combined with increasing amounts of RNAP core, and reaction products were separated by native PAGE and revealed by autoradiography ( Fig. 2A). The free [ 32 P] 54 migrates as a diffused, high mobility band ( Fig. 2A, lane 1), whereas E[ 32 P] 54 migrates as a sharp, low mobility band ( Fig. 2A, lane 4). At [ 32 P] 54 :E ratio of 1:1, ⌬1149 -1190 E formed ϳ50% less holoenzyme than the wild-type RNAP core ( Fig. 2A, lanes 4 and 10). A Fig. 2A, lane 12). Thus, it appears that removing ␤Ј residues 1149 -1190 leads to a moderate defect in 54 binding to the core RNAP. Based on the similar mobilities of the mutant and wild-type E 54 complexes on native gels (Fig. 2, A-C), we infer that deletion of ␤Ј residues 1149 -1190 has no strong effect on the overall conformation of E 54 . In all the experiments described below, we used saturating amounts of 54 over the mutant and wild-type RNAP core to ensure that all core RNAP is converted into the holoenzyme.
The ability of ⌬1149 -1190 E 54 to form closed promoter complexes was tested using the well characterized S. meliloti nifH promoter (18). A fixed amount of 32 P-labeled linear promoter probe was incubated with increasing amounts of E 54 and the reaction products were separated by native PAGE. As evident from Fig. 2B, compared with WT E 54 , ⌬1149 -1190 E 54 was ϳ2-fold less active in closed complex formation. Because specific promoter DNA contacts in the closed complex appear to be made exclusively by 54 (17,19), we suggest that removal of ␤Ј residues 1149 -1190 affects, directly or indirectly, DNA-interacting domains of 54 within E 54 thus explaining the reduced promoter binding by ⌬1149 -1190 E 54 . This view is supported by functional data that showed that deletion of the amino-terminal, promoter DNA-proximal region I of 54 changes the conformation of the ␤Ј-jaw (11).
We incubated the ⌬1149 -1190 E 54 closed complex with an AAA activator in the presence of an ATP hydrolysis transition state analogue, ADP-AlF x (15), to determine whether the ⌬1149 -1190 E 54 closed com- plex is able to interact with the activator. For experimental simplicity, a DNA-binding mutant of the E. coli AAA activator, phage shock protein F (PspF), was used. The mutant, PspF-(1-275), is able to interact with and activate the E 54 closed complex from solution (14). The molar ratio of E 54 to DNA was adjusted so that approximately equal amounts of E 54 closed complexes were formed in reactions containing either the mutant or wild-type RNAP (Fig. 2C, lanes 2 and 3). As can be seen, in the presence of ADP-AlF x , PspF-(1-275) binds equally well to both the mutant and the wild-type closed complexes (Fig. 2C, lanes 4 and 5). We conclude that removal of ␤Ј residues 1149 -1190 does not grossly change the ability of the mutant closed complex to interact with the AAA activator.
⌬1149 -1190 E 54 Fails to Form Stable Open Complexes-The ability of ⌬1149 -1190 E 54 to form open complexes and to initiate transcript synthesis from the S. meliloti nifH promoter was tested using an abortive transcription assay. The sequence of the nifH promoter around the transcription start site is TGGG (from Ϫ1 to ϩ3, non-template strand).
In the presence of the UpG dinucleotide, radioactive GTP, and dATP (the latter is needed for the ATPase activity of the AAA activator), E 54 reiteratively synthesizes the tetranucleotide UGGG. The UGGG transcript (and therefore the transcriptionally proficient open complex) is formed only when PspF-(1-275) and dATP are added to reactions containing the E 54 closed complex (Fig. 3A, lane 6). When the polyanion heparin, a DNA competitor, is added to reactions after open complex formation, but prior to the addition of transcription substrates, the abortive transcription assay can be used to measure the stability of open complexes by determining the amount of transcript synthesized after incubation with heparin. With WT E 54 , the amount of transcript synthesized gradually decreases with time, reflecting dissociation of the open complex in the presence of heparin (Fig. 3B). When the experiment is repeated with ⌬1149 -1190 E 54 , the amount of synthesized transcripts decreases much more rapidly (Fig. 3B). For example, after a 20-min incubation with heparin, ϳ12-fold fewer transcripts were synthesized by the mutant than by the wild-type RNAP (Fig. 3B, lane 6). Control assays established that the decrease in the amount of transcripts synthesized in reactions containing ⌬1149 -1190 E 54 was not because of inactivation of the mutant RNAP during the assay (data not shown).
The instability of the ⌬1149 -1190 E 54 open complex upon addition of heparin may simply reflect a more rapid (compared with reactions containing WT E 54 ) dissociation of the mutant core RNAP from 54 . To prove that this is not the case, we measured the mutant and wild-type open complex stabilities in the absence of heparin ( Transcription assays were repeated under conditions that allowed a single-round synthesis of a full-length transcript from the nifH promoter. As shown in Fig. 3D, and consistent with the results of the abortive transcription assays (Fig. 3B), ⌬1149 -1190 E 54 produced ϳ9-fold less full-length transcript than WT E 54 after a 20-min incubation with heparin (Fig. 3D, lanes 1 and 5). In the absence of heparin, a condition when multiple rounds of transcription initiation can occur, ⌬1149 -1190 E 54 produced ϳ1.5-fold less full-length transcripts than WT E 54 (Fig. 3D,  lanes 9 and 13). A similar trend was observed on three other 54 -dependent promoters (E. coli glnHp2, pspA, and pspG) (Fig. 3D). The instability of ⌬1149 -1190 E 54 open complexes was most pronounced in reactions containing the pspG promoter (Fig. 3D, lanes 4, 8, 12, and 16). Overall, we conclude that ⌬1149 -1190 E 54 open complexes, once formed, are clearly much less stable than wild-type open complexes.  4 and 6). Further analysis of mutant and wild-type closed and open complexes by a range of other footprinting reagents, such as S1 nuclease (which also reports DNA opening) and ortho-copper phenanthroline (which reports a distortion near the start site-proximal consensus promoter element characteristic of E 54 closed complexes; (21)) also did not reveal any detectable differences (data not shown).
We used photoactive S. meliloti nifH promoter probes to further characterize mutant and wild-type open complexes. The photoactive probes were made by derivatizing a single, specifically placed phosphorothioate with p-azidophenacyl bromide. Upon UV irradiation, the aryl azide group of p-azidophenacyl bromide reacts with protein moieties that are within ϳ12 Å from its attachment site on the DNA, resulting in the formation of covalently cross-linked protein-DNA complexes (17,22). Because the p-azidophenacyl bromide-modified DNA strand also contains a [ 32 P]phosphate at its 5Ј end, protein-DNA cross-links can be easily detected after fractionation of reaction products by SDS-PAGE. Derivatization of several positions of the nifH promoter (both in the template and non-template strand; Fig. 4C) led to good yields of protein-DNA cross-links within the open complex. 4 Based on the mobilities of non-cross-linked ␤, ␤Ј, and subunits on a SDS-PAGE gel, the faster migrating band was identified as 54 and the slower migrating band(s) were identified as cross-linked ␤/␤Ј (17). 4 Comparisons of mutant and wild-type open complex patterns revealed no differences either in the number or intensity of cross-linked species. Overall, the results from DNase I and KMnO 4 footprinting and the photocross-linking experiments suggest that the inherent instability of ⌬1149 -1190 E 54 open complexes is not because of a lack of a direct specific interaction between ␤Ј 4 P. C. Burrows, S. R. Wigneshweraraj, and M. Buck, manuscript in preparation.  Pre-opening Promoter DNA Stabilizes ⌬1149 -1190 E 54 Open Complexes-Because ⌬1149 -1190 E 54 forms open complexes as efficiently as WT E 54 (Fig. 4B), the instability of ⌬1149 -1190 E 54 open complexes could be because of the inability of the ⌬1149 -1190 E 54 to maintain DNA strand separation within the open complex. To test this idea, we performed abortive transcription assays using a heteroduplex variant of the S. meliloti nifH promoter as a template. Such heteroduplex promoter probes are faithful templates for studying the transcription initiation process by the bacterial RNAP (23). The S. meliloti nifH heteroduplex promoter probe used here contains a non-complementary segment between positions Ϫ10 and Ϫ1 and thus mimics the conformation of promoter DNA within the open complex (24,25). In the presence of the activator, WT E 54 forms heparin-resistant complexes on the Ϫ10 to Ϫ1 heteroduplex promoter probe, and thus likely adopts a conformation that is similar to that formed on homoduplex templates after the normal DNA opening process is completed (24,25). In contrast to results with homoduplex promoter DNA (both on supercoiled plasmid and a linear DNA fragment; Fig. 3B and data not shown), the ⌬1149 -1190 mutant and WT E 54 produced equal amounts of abortive transcripts from the heteroduplex promoter probe, even after preincubation with heparin for more than 1 h (Fig. 5). Thus, destabilization of promoter complexes caused by deletion of the ␤Ј-jaw is suppressed by pre-opening of promoter DNA (compare Figs. 3B and 5). It therefore appears that ␤Ј residues 1149 -1190 contribute to stable maintenance of DNA in the "opened" conformation once DNA opening per se has been completed.
The melting of promoter DNA during open complex formation is a multistep process (26,27). To determine which step en route to the transcription-competent open complex requires ␤Ј residues 1149 -1190, we performed abortive transcription assays using nifH promoter probes containing non-complementary segments between positions Ϫ10 and Ϫ6, Ϫ5 and Ϫ1, and Ϫ3 and Ϫ1. As shown in Fig. 6, A and B, ⌬1149 -1190 E 54 and WT E 54 synthesized approximately equal amounts of abortive transcripts from all three heteroduplex promoter probes in the absence of heparin. However, when complexes were incubated with heparin for 5 min after activation but prior to the addition of transcription substrates, ⌬1149 -1190 E 54 synthesized ϳ4 times less abortive transcripts from the Ϫ10 to Ϫ6 heteroduplex probe than did WT E 54 . In contrast, the amounts of abortive transcripts synthesized by ⌬1149 -1190 E 54 from promoter probes containing heteroduplexes from Ϫ5 to Ϫ1 and from Ϫ3 to Ϫ1 were comparable with the amounts synthesized by WT E 54 even in the presence of heparin (Fig. 6, A and B).
We next used promoters containing non-complementary segments between promoter positions Ϫ3 and Ϫ2, Ϫ2 and Ϫ1, and at Ϫ1 as templates in the abortive transcription assays, to further localize promoter positions that need to be pre-opened to allow heparin-resistant open complex formation by ⌬1149 -1190 E 54 . As shown in Fig. 6C, pre-opening of any promoter position between Ϫ3 and Ϫ1 allows   Fig. 6D, in the absence of heparin, the wild-type and the mutant E 70 synthesized approximately equal amounts of abortive transcripts from the homoduplex lacUV5 promoter probe. However, when the open complexes were incubated with heparin for 5 min, ⌬1149 -1190 E 70 synthesized ϳ5 times less abortive transcripts than WT E 70 . In contrast, ⌬1149 -1190 E 54 synthesized only ϳ1.5 times less abortive transcripts than WT E 70 when the Ϫ1 heteroduplex lacUV5 promoter probe was used as the template. When the incubation time with heparin was increased to 20 min, ⌬1149 -1190 E 70 synthesized ϳ25 times less abortive transcripts than WT E 70 from the homoduplex lacUV5 probe (Fig. 6E). However, when Ϫ1 heteroduplex probe was used as the template, ⌬1149 -1190 E 70 synthesized ϳ1.7 times less abortive transcripts than WT E 70 . Additional assays with different heteroduplex variants of lacUV5 revealed that pre-opening of any position downstream of the Ϫ10 promoter consensus element resulted in increased stability of ⌬1149 -1190 E 70 open complexes (data not shown). Thus, pre-opening of promoter DNA has a strong positive effect on the stability of open complexes formed by the RNAP containing the ␤Ј⌬1149 -1190 mutation in the context of the 54 and 70 subunits; for E 54 the effect is localized to positions Ϫ3 to Ϫ1. 54 Region I and ␤Ј Residues 1149 -1190 Cooperate during Stable Open Complex Formation-54 lacking the amino-terminal regulatory Region I ( 54 ⌬RI ) or carrying substitutions in Region I (e.g. 54 Ala24 -26 ) enables E 54 to form heparin-resistant complexes on a Ϫ10 to Ϫ1 heteroduplex promoter in the absence of the activator (24). In contrast, wild-type E 54 complexes require activation to become heparin-resistant (24). We reconstituted RNAP holoenzymes from ⌬1149 -1190 E and 54 Region I mutants to investigate if there is a functional link between 54 Region I and ␤Ј residues 1149 -1190. We used native gel analysis (as in Fig. 2A) to ascertain that the 54 mutants bound the ⌬1149 -1190 E just as well as the wild-type 54 (data not shown). As shown on Fig. 7A, neither E 54 ⌬RI nor E 54 Ala24 -26 form heparin-resistant complexes on the het-eroduplex promoter in the absence of ␤Ј residues 1149 -1190 (compare lanes 3, 4, 9, and 10). However, in the presence of GTP (which allows the synthesis of an RNA trimer from the S. meliloti nifH promoter), heparin-resistant ⌬1149 -1190 E 54 ⌬RI and ⌬1149 -1190 E 54 Ala24 -26 complexes were detected (Fig. 7A, compare lanes 6, 7, 12, and 13). Presumably, in reactions containing GTP, the initiation of transcription prevents the decay of unstable open complexes formed by the ⌬1149 -1190 E 54 ⌬RI and ⌬1149 -1190 E 54 Ala24 -26 . Control reactions with dGTP, which is not a substrate for RNA synthesis, did not yield any heparin-resistant heteroduplex complexes with either ⌬1149 -1190 E 54 ⌬RI or ⌬1149 -1190 E 54 Ala24 -26 (data not shown).
Because 54 Region I constitutes the major AAA activator interaction site (15), the AAA activator does not greatly improve heparin resistance of E 54 ⌬RI on the Ϫ10 to Ϫ1 heteroduplex probe (24). In contrast, heparin resistance of E 54 Ala24 -26 complexes on the Ϫ10 to Ϫ1 heteroduplex probe increases in the presence of the activator (28). As can be seen in Fig. 7B, heparin resistance of complexes between the double mutant ⌬1149 -1190 E 54 Ala24 -26 and the Ϫ10 to Ϫ1 heteroduplex template was not markedly increased under activating conditions. This result suggest that pre-opening the DNA does not restore mutant open complex stability, if the integrity of Region I is compromised. Thus, the result strongly suggests that a functional link exists between 54 Region I and ␤Ј residues 1149 -1190 during open complex formation by E 54 .
␤Ј Residues 1149 -1190 Are Required for Activator "Bypass" Transcription by E 54 -AAA activator-independent transcription by E 54 reconstituted with activator-bypass 54 mutants (e.g. 54 harboring the F318A or R336A substitutions) is efficiently inhibited by bacteriophage T7 gp2 protein, which binds to the ␤Ј-jaw (11). We wished to determine whether the E 54 F318A and E 54 R336A mutants could transcribe in an activator-independent fashion in the absence of the ␤Ј-jaw. As shown in Fig. 7C, ⌬1149 -1190 E 54 F318A does not transcribe from the S. meliloti nifH promoter (lanes 3 and 4). Identical results were obtained with ⌬1149 -1190 E 54 R336A (data not shown). Control assays showed that both ⌬1149 -1190 E 54 F318A and ⌬1149 -1190 E 54 R336A formed closed complexes on the S. meliloti nifH promoter (data not shown). Thus, transcription by E 54 F318A and E 54 R336A fails when the ␤Ј jaw is absent or if its function is compromised by the binding of gp2.

DISCUSSION
The principal finding of this study is the demonstration that during transcription by E 54 , E. coli RNAP ␤Ј residues 1149 -1190, which constitute a large part of the ␤Ј-jaw module, help stabilize the segment of the melted DNA immediately upstream of the transcription start site in the open promoter complex. Thus, the ␤Ј-jaw appears to have a related role in stabilizing the open promoter complex during transcription initiation by E 54 (this work) and E 70 (Ref. 8 and this work). Analysis of conformational flexibility of bacterial RNAP has revealed that the ␤Ј-subunit pincer has a tendency to pivot into and away from the main DNA binding channel of the RNAP (29). The ␤Ј-jaw and the ␤Ј-clamp are integral mobile modules of the ␤Ј subunit (Fig. 1B). In the closed complex, the ␤Ј-clamp restricts the access of the DNA to the catalytic site (6). However, upon open complex formation, the ␤Ј-clamp moves, enabling DNA access to the catalytic site. We envisage that the movement of the ␤Ј-clamp could be allosterically relayed to the ␤Ј-jaw, which in turn helps to stabilize the open complex and perhaps helps propagate DNA opening further downstream. Indeed, models of the open complex and transcription elongation complexes place the DNA downstream of the catalytic center within a trough formed by the ␤Ј-jaw and the ␤Ј-clamp (7,8). The function of the ␤Ј-jaw appears to be akin to that of its structural homologue, the RPB1 jaw lobe of the eukaryotic RNAP II. The RPB1 jaw lobe undergoes a conformational change and moves toward DNA upon interaction with the transcription factor TFIIB (30).
The results also highlight the point that regulated communication between the mobile modules of the RNAP and the functional domain(s) of an RNAP-associated factor is required for stable DNA opening. In the 54 -dependent system, the ␤Ј-jaw appears to function cooperatively with the conserved amino-terminal Region I domain, a key regulatory domain of 54 (see later). This can explain why ⌬1149 -1190 E 54 has reduced activity for closed complex formation. Because DNA contacts within the closed complex are exclusively made by 54 (17,19), 4 we propose that deletion of ␤Ј residues 1149 -1190 leads to repositioning of 54 domains, notably the Region I domain, which is located close to the start site-proximal consensus promoter DNA element within the closed complex (17,16,31). Region I of 54 is a major target of the AAA activator. Proximitybased footprinting studies have placed Region I of 54 at the upstream edge of the RNAP (Fig. 1A) (32). The ␤Ј-jaw is located toward the downstream edge of the RNAP (Fig. 1A). Therefore, we envisage that cooperation between Region I and the ␤Ј-jaw during open complex formation occurs indirectly, via a conformational signaling pathway within E 54 . In support of this view, we have previously shown that removal of Region I or AAA activator-triggered movements in Region I alter the conformation of the ␤Ј-jaw (11). Our new results suggest that one role for Region I-triggered conformational changes of the ␤Ј-jaw is to confer stability to the open complex. Because 54 Region I is also required for stabilizing E 54 interactions on melted DNA (24), it appears that 54 Region I and the ␤Ј-jaw domain cooperatively contribute to the stability of E 54 open complexes, consistent with the view that downstream DNA within open complexes lies within the trough formed by the ␤Ј-jaw and the ␤Ј-clamp.
Stable opening of DNA immediately upstream of the transcription start site appears to depend on the integrity of the ␤Ј-jaw. However, ⌬1149 -1190 E 54 is otherwise normal for DNA opening and transcription.
Because pre-melting the DNA near the transcription start site markedly increased ⌬1149 -1190 E 54 open complex stability, it appears that the ␤Ј-jaw, rather than making any direct interactions with DNA, contributes to interactions between 54 and the RNAP that confer stability and heparin resistance to open complexes. The ␤Ј-jaw seems to influence interactions with a small segment of promoter DNA (between positions Ϫ3 and Ϫ1) that are required for open complex stability. No single part of this short DNA segment seems to be more important than any other part, suggesting that properties of ⌬1149 -1190 E 54 are not because of a specific defect in an isolated amino acid-DNA interaction. Rather, a general property of DNA melting or DNA distortion at or close to the ϩ1 position seems to be important. The evidence that mutant and wildtype promoter complexes are indistinguishable by several footprinting methods and in photocross-linking experiments supports this view.
DNA strand separation during open complex formation is energetically unfavorable because the complementary DNA strands must be physically kept separate under solution conditions that would otherwise greatly favor base pairing. In the case of E 70 , several conserved domains of 70 cooperate with core RNAP subunits to establish favorable protein-DNA interactions required for stable maintenance of the open complex (33); the ␤Ј-jaw seems to play a prominent role in this process (8). We have now shown that in the case of E 54 , the cooperation, particularly near the transcription start site, between the highly conserved Region I domain of 54 and the ␤Ј-jaw also helps to establish interactions favoring stable maintenance of open complexes. Mutations in 54 Region I and ␤Ј-jaw appear to abolish these interactions and impose a substantial energetic penalty on open complex stability that can only be compensated for by pre-opening of the promoter DNA. This also seems to be true for E 70 , because pre-opening the promoter markedly improves the stability of ⌬1149 -1190 E 70 open complexes formed on lacUV5.
Interestingly, the stability of E 70 open complexes on the lacUV5 promoter appears to be restored to wild-type levels when the promoter is pre-opened at any position between the transcription start site proximal RNAP binding site (Ϫ10 region) and the transcription start site. In contrast, stability of E 54 open complexes on the S. meliloti nifH promoter is restored to wild-type levels only when the promoter is preopened near the transcription start site. Based on the observation that 54 can open (in the presence of the AAA activator and a hydrolysable nucleotide) the S. meliloti nifH promoter up to position Ϫ8 in the absence of the core RNAP subunits (18,34), we envisage that initial promoter opening during 54 and 70 -dependent transcription initiation requires different elements of core RNAP. The results presented here argue that during 54 -dependent DNA opening, the core RNAP subunits contribute to later stages of the DNA opening process. This view is further supported by the observation in which core RNAP subunits do not become cross-linked to the promoter DNA segment between Ϫ10 and Ϫ1 within an intermediate E 54 promoter complex that forms en route to the bona fide open complex (17). In contrast, ␤ and ␤Ј subunits only cross-link to DNA when the open complex has formed at 54 -dependent promoters. 4 It seems that the T7 bacteriophage encoded transcription inhibitor gp2, which binds to the ␤Ј-jaw domain, shuts down host transcription by compromising the ␤Ј-jaw function. It has been shown that T7 gp2 inhibits host transcription by preventing open complex formation on some promoters (10). Our results with activator bypass 54 mutants suggest that one other way by which gp2 could inhibit host transcription is by negatively influencing the ability of RNAP to make or capture opened DNA by interfering with ␤Ј-jaw activity.