The Escherichia coli RNA Polymerase·Anti-ς70 AsiA Complex Utilizes α-Carboxyl-terminal Domain Upstream Promoter Contacts to Transcribe from a −10/−35 Promoter*

During infection of Escherichia coli, the phage T4 early protein AsiA inhibits open complex formation by the RNA polymerase holoenzyme Eς70 at −10/−35 bacterial promoters through binding to region 4.2 of the ς70subunit. We used the −10/−35 lacUV5 promoter to study the properties of the Eς70·AsiA complex in the presence of the glutamate anion. Under these experimental conditions, inhibition by AsiA was significantly decreased. KMnO4 probing showed that the observed residual transcriptional activity was due to the slow transformation of the ternary complex Eς70·AsiA·lacUV5 into an open complex. In agreement with this observation, affinity of the enzyme for the promoter was 10-fold lower in the ternary complex than in the binary complex Eς70·lacUV5. A tau plot analysis of abortive transcription reactions showed that AsiA binding to Eς70 resulted in a 120-fold decrease in the second-order on-rate constant of the reaction of Eς70with lacUV5 and a 55-fold decrease in the rate constant of the isomerization step leading to the open complex. This ternary complex still responded to activation by the cAMP·catabolite activator protein complex. We show that compensatory Eς70/promoter upstream contacts involving the C-terminal domains of α subunits in Eς70 become essential for the binding of Eς70·AsiA to the lacUV5promoter.

In Escherichia coli RNA polymerase, the core enzyme E (subunit composition ␣ 2 ␤␤Ј) associates with the subunit to form the holoenzyme E, the species able to initiate transcription at specific promoter sites on bacterial or phage genomes. The nature and properties of the subunit present in a holoenzyme at a given promoter, together with the ability of the ␣-carboxyl-terminal domain (␣-CTD) 1 to recognize an upstream element of the promoter, determine whether and how often transcription will start at this site (1). These principles are nicely illustrated by the regulation of transcription during the development of phage T4 in E. coli. Here, all the phage genes are transcribed by the host RNA polymerase, the structure and molecular properties of which are modified by phagecoded proteins, resulting in the sequential utilization of early, middle, and late promoters (2,3). Immediately after infection, T4 early promoters, with their bacterium-like Ϫ10 and Ϫ35 recognition sequences, are transcribed by E 70 , the host holoenzyme harboring 70 , the major host factor. T4 middle promoters contain a Ϫ10 element closely matching the Ϫ10 consensus sequence for 70 , but the Ϫ35 element is replaced by a so-called MotA box, a Ϫ30 binding site for the phage-coded middle transcriptional activator MotA (4,5). In addition to E 70 and MotA, middle mode transcription also requires the presence of another phage early protein, the anti-factor AsiA (6). Upon binding the MotA box, MotA protein activates recognition of middle promoters by a holoenzyme E 70 ⅐AsiA complex (7,8). Here, AsiA appears to act as a molecular device that switches 70 from the early to the middle class of T4 promoters. Transcription at late promoters is closely coupled with T4 DNA replication and requires a novel T4-encoded subunit, gp55 (9).
Because it is a coactivator of transcription from T4 middle promoters and, simultaneously, a transcription inhibitor of bacterial or T4 early promoters (3), AsiA might also be able to cause the rapid arrest of transcription from early promoters, concomitant with the start of MotA-dependent middle transcription. Assigning this function is a long-standing and unresolved question in phage T4 biology (10). It has been recently shown, however, that this transcription shutoff also occurs in the absence of AsiA (11), although the same study confirmed that transcription of E. coli genes is rapidly and strongly inhibited in vivo when AsiA is overproduced.
In vitro transcription studies have helped to outline the mechanism of inhibition by AsiA. This 10-kDa protein binds to region 4.2 of 70 . In the E 70 ⅐AsiA complex, this binding blocks the normal interaction between 70 and the Ϫ35 upstream promoter element (12)(13)(14). Although AsiA strongly inhibits open promoter complex formation and transcription by E 70 from a Ϫ10/Ϫ35 promoter like lacUV5 or the T4 early promoter P15.0, the holoenzyme is resistant to AsiA inhibition at promoters that, like galP1, lack a Ϫ35 consensus motif and contain an "extended Ϫ10" motif (12,13,15,16). These observations led to a simple model in which the interactions of domain 4.2 in 70 with the Ϫ35 promoter element or with AsiA are mutually exclusive. This model explains the effect of AsiA on E 70 at a Ϫ10/Ϫ35 promoter under classical assay conditions (13). However, at lacUV5, these experimental conditions preclude a detailed analysis of the repression mechanism. In this study, we have looked for conditions allowing the maintenance of residual transcriptional activity. For this purpose, we selected a salt that enhances the interactions between the RNA polymerase partners. Replacing the chloride anion by glutamate increases the affinity of RNA polymerase for its promoters (17,18). Such a change is also likely to improve the interaction between protein partners (19). We expected, and indeed observed, that in the presence of the Glu Ϫ anion, E 70 could still form a kinetically competent complex in the presence of AsiA. This allowed us to quantify free energy changes occurring in the overall reaction and to identify specific compensatory interactions between the holoenzyme and the promoter that permit RNA polymerase to overcome the otherwise strong inhibition brought about by AsiA.

EXPERIMENTAL PROCEDURES
Materials-High Pure spin columns and the Pwo polymerase were from Roche Molecular Biochemicals (Mannheim, Germany). T4 polynucleotide kinase was purchased from New England Biolabs Inc., and DNase I was from Worthington. Magnetic DynaBeads were from Dynal, Inc., and the 8 -25% polyacrylamide PhastGels were from Amersham Pharmacia Biotech.
Plasmids and DNA Fragments-A 207-bp lacUV5 DNA fragment (20) was inserted at the EcoRI site of plasmid pJCD0 (21) upstream of the rrnB T1 and T2 terminators to yield plasmid pGO1. A 665-bp lacUV5 fragment comprising the two terminators was isolated by polymerase chain reaction with template pGO1 and the following primers: 5Ј-CGCCAGGGTTTTCCCAGTCACGAC and 5Ј-GGATTTGTCCTACT-CAGGAG. A 220-bp lacUV5 fragment was isolated by polymerase chain reaction using plasmid pBR-lacUV5 as a template (20) together with primers A1 (5Ј-GGCGTATCACGAGGCCCTTTCG) and B1 (5Ј-GCTG-GCACGACAGGTTTCCCGA). Primer A1 was end-labeled with T4 polynucleotide kinase and [␥-32 P]ATP (3000 Ci/mmol) or used with unlabeled primer B1 in a polymerase chain reaction to prepare the 220-bp lacUV5 fragment labeled at the 5Ј-end on the non-template strand. This fragment was used in the gel shift assay and the DNase I footprinting experiments. Primer A1 biotinylated at the 5Ј-end was used in a polymerase chain reaction to obtain the biotinylated 220-bp lacUV5 fragment.
Purified Proteins and Standard Reaction Conditions-E. coli RNA polymerase holoenzyme was purified according to Burgess and Jendrisak (22) as modified by Marschall et al. (21). AsiA protein from phage T4 was purified as previously described (16). Core RNA polymerase was prepared according to Lederer et al. (23), and 70 was obtained using the overproducing strain M5219/pMRG8 and the published purification procedure (24). Catabolite activator protein (CAP) was prepared as described by Ghosaini et al. (25), and the ⌬␣-235 RNA polymerase was a gift from Dr. Evelyne Richet. The experiments described below were performed in buffer A containing 40 mM Hepes (pH 8.0), 10 mM MgCl 2 , 500 g/ml bovine serum albumin, 1 mM dithiothreitol, and either 100 mM KCl or the indicated KGlu concentration.
Isolation on Magnetic Beads of the Ternary RNA Polymerase⅐AsiA⅐lacUV5 Complex-The biotinylated lacUV5 DNA fragment was prepared as described above. After purification, the biotinylated fragment was immobilized on Dynal streptavidin magnetic beads (12). RNA polymerase holoenzyme (40 pmol) and an 8-fold molar excess of AsiA were incubated in a 30-l reaction for 30 min at 37°C in buffer A containing 100 mM KGlu and 0.125% (v/v) Tween instead of bovine serum albumin. The binding reaction was then added to the immobilized lacUV5 fragment, and the mixture was incubated for 60 min at room temperature. The beads were collected by centrifugation, and the supernatant containing the unbound proteins was withdrawn (see Fig. 2, lane 2). The beads were then washed twice with KGlu buffer, and the bound proteins were eluted after a 1-h incubation at room temperature in 1% SDS (see Fig. 2, lane 3). The samples and controls were analyzed by electrophoresis under denaturing conditions on an 8-25% (w/v) polyacrylamide gradient PhastGel.
Single Round Transcription Assays-Prior to assembly for transcription reactions, RNA polymerase, AsiA, and the lacUV5 template were separately diluted on ice in buffer A containing either 100 mM KCl or the indicated KGlu concentration (100 -400 mM). Runoff transcription reactions were carried out at 37°C under the following conditions: 30 nM RNA polymerase holoenzyme, 2 nM lacUV5 template, the indicated molar excess of AsiA over the polymerase concentration, 100 M ATP, 100 M CTP, 100 M GTP, 10 M UTP, 0.5 Ci of [␣-32 P]UTP, and 250 g/ml heparin. RNA polymerase and AsiA were first mixed at 37°C for 15 min. Template was added to allow open complex formation for 30 min, followed by addition of the nucleotides and heparin. Elongation reactions lasted 10 min and were stopped by mixing equal volumes of reaction mixtures with 20 mM EDTA in xylene cyanol-containing 95% formamide. Following heating at 65°C, the samples were electrophoresed on a 7% polyacrylamide sequencing gel, and the transcripts were quantified with a PhosphorImager.
Abortive Transcription-Abortive transcription reactions (26,27) were carried out at 37°C in buffer A containing the indicated KGlu concentrations and the final concentrations of the following compo-nents: 500 M ApA, 50 M UTP, 2 nM lacUV5 DNA fragment, and 0.5 Ci of [␣-32 P]UTP. The lag assay in Fig. 5A was performed with 30 nM RNA polymerase (final concentration), previously incubated for 15 min either with buffer or with an 8-fold molar excess of AsiA. The experiments with reconstituted RNA polymerases were performed using a fixed time assay at a 100 nM final concentration of the different enzymes. Tau plot analysis (26,28) was carried out to measure the average time obs required for open complex formation at lacUV5 with and without AsiA previously bound to the holoenzyme. For each reaction, the amount of UTP incorporated was plotted versus time. The two parameters obs (minutes) and final steady-state velocity V (mole of product ApApUpU/mol of promoter/min) were determined using a Kaleidagraph program that performed a least-squares fit of the data to the following equation: Y ϭ V⅐t Ϫ V⅐ obs (1 Ϫ exp(t/ obs )), where Y is the amount of product, and t is time in minutes (28).
Gel Retardation Assays-Stock solutions of AsiA and RNA polymerase were prepared on ice in buffer A with 100 mM KCl or KGlu. When present, AsiA was in 6-fold molar excess relative to the highest polymerase concentration used. Incubations were at 37°C. RNA polymerase⅐AsiA complexes were first formed during 15 min, and the enzymes were then incubated for 80 -90 min with the radioactively labeled lacUV5 fragment at 0.02-0.065 nM. After addition of heparin (55 g/ml), the samples were loaded onto 5% native polyacrylamide gels prepared in Tris borate/EDTA buffer and electrophoresed at 120 V at room temperature.
DNase I Footprinting-Complexes with the labeled lacUV5 promoter (at a 3 nM final concentration) were formed during 50 min at 37°C in buffer A containing 200 mM KGlu or 100 mM KCl, using a 6-fold molar excess of AsiA or the cAMP⅐CAP complex as indicated in Fig. 7, with purified RNA polymerase or reconstituted ⌬␣-235 RNA polymerase (each at a 100 nM final concentration). Complexes were then treated with DNase I (at a 80 ng/ml final concentration) for 30 s (or 15 s in the absence of RNA polymerase). Protected bands were identified on the pattern afforded by the migration of the same fragment treated for the G ϩ A sequencing reaction (29).

Single Round Transcription Analysis in KGlu Buffers-
Transcription inhibition by AsiA of E 70 , the RNA polymerase holoenzyme, has been conveniently analyzed with well defined DNA templates (13,15,16). Here, we performed single round transcriptions with a DNA linear fragment containing the lacUV5 promoter upstream of a strong transcriptional terminator and generating a 178-nucleotide transcript. This construct was used to assess the effect of increasing AsiA concentrations (relative to holoenzyme) on runoff transcription reactions when the assays were performed in KGlu buffers of increasing ionic strength as compared with KCl, after a 30-min incubation time with the promoter in each case. In the absence of AsiA, the effect of the KGlu buffers (100 -400 mM KGlu) was a 25-30% decrease in transcriptional activity relative to that measured in 100 mM KCl, an effect that was ionic strengthindependent. Fig. 1 shows that the nature and concentration of the buffer both have a marked effect on the extent of transcription inhibition by AsiA. Strong inhibition was observed in 100 mM KCl and more so in 400 mM KGlu. In contrast, at all other KGlu concentrations (100 -300 mM), transcription was significantly less inhibited, and the corresponding residual activities reached a 75-80% plateau as the AsiA concentration was raised. We therefore utilized the 100 -300 mM KGlu concentration range to investigate in greater detail the behavior of the enzyme⅐inhibitor complex, the E 70 ⅐AsiA entity, relative to the lacUV5 promoter.
Existence of a Ternary Complex Formed at the lacUV5 Promoter by E 70 ⅐AsiA-The results described above prompted us to check whether AsiA was present in a stable ternary complex formed with E 70 and the promoter in the presence of 100 -300 mM KGlu. For this experiment, we used a biotinylated lacUV5 DNA fragment immobilized on streptavidin-agarose magnetic beads. The proteins found to be bound to this DNA fragment were recovered and analyzed on an SDS-polyacrylamide gel (12). AsiA (in an 8-fold molar excess over enzyme) was first added to the holoenzyme in 100 mM KGlu (Fig. 2, lane 1), and the mixture was incubated with the lacUV5 fragment to allow open complex formation. Following this treatment, the holoenzyme was eluted from the beads (Fig. 2, lane 3). Based on the relative Coomassie Blue staining intensities, this eluted fraction was found to contain AsiA in a stoichiometric ratio relative to the subunit present in this holoenzyme preparation (Fig. 2,  lane 5). As a control, the same experiment was performed in the presence of 100 mM KCl. In this case, in contrast to the result observed in KGlu, previous incubation of E 70 with an 8-fold molar excess of AsiA led to the elution from the immobilized promoter of an extremely low quantity of E 70 , containing no detectable amount of AsiA (data not shown). Furthermore, in DNase footprinting experiments performed in the presence of 100 mM KCl with E 70 ⅐AsiA on the lacUV5 fragment, the only species that could be detected was a residual binary complex, E 70 ⅐lacUV5 (see Fig. 7B below). Therefore, in the presence of KGlu, AsiA appeared to take part in a stable ternary complex at the lacUV5 promoter.
KGlu Increases the Affinity of the Holoenzyme⅐AsiA Complex for the lacUV5 Promoter-Given the possibility of forming in KGlu buffers a ternary complex between RNA polymerase, AsiA, and lacUV5, we used gel shift assays to determine the effect of AsiA on holoenzyme affinity for the lacUV5 promoter. We measured the formation of heparin-resistant complexes in KGlu and KCl buffers in the absence and presence of AsiA. Fig.  3A shows a gel retardation assay of complexes formed in 100 mM KGlu, and Fig. 3B shows an analogous experiment performed in the presence of 100 mM KCl. In either buffer, a clear retarded complex was observed at all holoenzyme concentrations. After quantification by phosphorimaging, the gel shift data were fitted to the equation of a rectangular hyperbola using a nonlinear regression program. The results can be summarized as follows. In KGlu buffer, even in the presence of AsiA, a fractional saturation value close to 1 was observed at high holoenzyme concentrations. In 100 mM KGlu, an apparent dissociation constant (K D ) of 0.35 Ϯ 0.07 nM was found for E 70 /promoter binding. This value was Ͼ10-fold higher when AsiA was previously bound to the holoenzyme (K D ϭ 4.7 Ϯ 1.6 nM). In 100 mM KCl and in the absence of AsiA, the binary complex displayed a K D of 1 Ϯ 0.1 nM, indicative of weaker binding of the enzyme to the promoter. In the same buffer and in the presence of 600 nM AsiA, the formation of a retarded species was greatly affected as the enzyme concentration increased. An apparent K D of 100 Ϯ 10 nM was calculated for the enzyme from the gel shift assay. Taken together, these results indicate a qualitative but clear-cut trend: shifting from glutamate to chloride destabilizes the binary complex by a factor of 2.8 in the absence of AsiA. In its presence, the RNA polymerase affinity for the promoter is even more affected because the apparent affinity drops by a factor of 22-25 under the conditions tested. These figures illustrate the destabilizing character of the Cl Ϫ buffers often used in vitro to study systems involving DNA/protein interactions.
KMnO 4 Reactivity of the E 70 ⅐AsiA⅐lacUV5 Complex in Potassium Glutamate-Potassium permanganate has been used to probe exposed pyrimidines in single-stranded DNA (30). KMnO 4 footprinting experiments were performed to confirm that the complex visualized in the gel retardation experiments in the presence of AsiA was an open complex. We first compared the KMnO 4 reactivities of the complexes formed in KGlu with and without AsiA. KMnO 4 sensitivity was observed in the binary complex E 70 ⅐lacUV5 at positions ϩ2, ϩ1, Ϫ8, Ϫ9, and Ϫ11 on the template strand of lacUV5 (data not shown), characteristic of the open complex formed by E 70 at this promoter (30). When AsiA was previously bound to E 70 , the same positions were found to react with KMnO 4 , although more slowly. The time course of establishment of this process was monitored. At zero time, E 70 ⅐AsiA was incubated in 100 mM KGlu with the labeled promoter fragment (50 nM E 70 , 300 nM AsiA, and 2 nM DNA), and the change in KMnO 4 reactivity was measured as a function of time. The results were quantified by comparison with the footprinting signal obtained with E 70 after allowing 60 min for maximal open complex formation. The kinetics observed with the ternary complex fit a single exponential (time constant close to 26 min), and the value measured at 60 min was 85% of the control (Fig. 4). This kinetic profile clearly differs from the fast opening of the promoter in the binary complex (Fig. 4, see the control at 10 min). In the presence of KGlu, the species E 70 ⅐AsiA is therefore able to form an open complex at the lacUV5 promoter, with a time course that is nevertheless significantly slower than for the uninhibited holoenzyme.
Kinetic Analysis of Open Complex Formation by the Holoenzyme⅐AsiA Species-The process studied above by KMnO4 reactivity could also be monitored using an abortive initiation assay that probes only the kinetically competent species. By this method, we analyzed the kinetics of binding of the holoenzyme to the promoter without and with AsiA previously bound to the enzyme. When the reaction was initiated by addition of E 70 ⅐AsiA instead of E 70 alone (40 nM in each case), we systematically observed an important increase in the latency time required to reach the steady-state rate of oligonucleotide synthesis. This increase was monitored at several KGlu concentrations (Fig. 5A). No lag could be easily measured with the holoenzyme alone. When AsiA was previously bound to the holoenzyme, a marked and roughly constant latency time was observed in the presence of 100 -300 mM KGlu (Fig. 5A). We chose then 200 mM as a convenient KGlu concentration and performed a tau plot analysis to determine the [E 70 ] dependence of the observed lag time obs without and with an 8-fold excess of AsiA. At excess RNA polymerase over promoter concentration, the tau plot analysis is based upon a simple twostep model (Equation 1) for open complex formation (31, 32), where R is E 70 , K B is the equilibrium binding constant of R to the promoter P, and k f is the isomerization first-order rate constant (31). Utilizing Equation 1 in the present case is thus an attempt to analyze the effect of binding AsiA to R in terms of this two-step model. Fig. 5B shows the considerable effect induced by AsiA upon the holoenzyme concentration dependence of the lag time , and Table I reports the kinetic parameters derived from these measurements. A strong kinetic penalty is brought about by the addition of AsiA. The overall second-order association constant K B k f is decreased by ϳ120fold. In terms of the two-step model of Equation 1, this effect is mainly expressed at the isomerization step of the pathway. Relative to the situation encountered with the free enzyme R, there is a 55-fold decrease in k f in the presence of excess AsiA. Despite a very large uncertainty in the values of K B (a ratio of 2 is observed between the average K B constants), we can safely conclude that, under these experimental conditions, AsiA is not a competitive inhibitor of RNA polymerase binding to the lacUV5 promoter.
Transcription by E 70 from the galP1 promoter or from a promoter sequence lacking a Ϫ35 hexamer has been shown to be essentially insensitive to inhibition by AsiA (12,13,15). In agreement with this, we did not observe the marked lags described above with lacUV5 when we used a consensus galP1 promoter that lacks a Ϫ35 consensus sequence (data not shown). Thus, the increased lags at lacUV5 specifically reflect an imperfect interaction between E 70 ⅐AsiA and this promoter with its functional Ϫ35 hexamer. We therefore looked for indications of modified interactions between the polymerase and the upstream regions of the promoter that could account for the observed transcriptional activity of E 70 ⅐AsiA at this Ϫ10/Ϫ35 promoter.
The Ternary Complex E 70 ⅐AsiA⅐lacUV5 Is Susceptible to Activation by CAP-At the lacUV5 promoter, the cAMP⅐CAP complex is known to bind a site centered at Ϫ61. 5 and to activate open complex formation (33). We have analyzed the effect of the cAMP⅐CAP complex on the kinetics of open complex formation by E 70 ⅐AsiA. Fig. 6 shows that when cAMP⅐CAP was incubated with the promoter before addition of 30 nM E 70 and with an 8-fold molar excess of AsiA, the observed lag time was reduced from 52 to 18 min; and interestingly, the two kinetics reached the same final steady-state rate after the lag period. The cAMP⅐CAP complex is therefore able to activate the holoenzyme entity E 70 ⅐AsiA. We suspected that this effect could be mediated through the formation of a quaternary E 70 ⅐AsiA⅐lacUV5⅐CAP complex (see below).
Role of Upstream Contacts in the Properties of the Ternary Complex-We used reconstituted holoenzymes to compare the behavior of normal E 70 and of E 70 (⌬␣), a holoenzyme containing ␣ subunits with deletions in their CTDs (34,35). Using the abortive initiation assay, we first measured, in the presence of 180 mM KGlu, the extent of inhibition caused by AsiA (6-fold molar excess) for three different holoenzymes, each at a 100 nM final concentration: native E 70 , reconstituted E 70 , and reconstituted E 70 (⌬␣). Each holoenzyme was incubated for 30 min with lacUV5 prior to the assay. The native holoenzyme was 30% inhibited. Reconstituted E 70 was 60% inhibited, whereas reconstituted E 70 (⌬␣) was almost totally inhibited (97%). An ␣-CTD deletion was therefore able to almost totally suppress the partial insensitivity to AsiA conferred to E 70 by the interactions enhanced by the presence of glutamate. Thus, promoter upstream contacts involving the ␣-CTD appear to be essential to counteract the inhibitory effect of AsiA bound to 70 . Upstream contacts between the promoter and the C-terminal domain of an ␣ subunit of E 70 are known to be further stabilized by the cAMP⅐CAP complex bound upstream (36).
To test this possibility, we used the reconstituted holoen- zymes E 70 and E 70 (⌬␣) in DNase I footprinting experiments to probe the structure of these complexes with and without AsiA. The ternary complex formed by E 70 ⅐AsiA showed an extended protection pattern (Fig. 7A, lane 3) with, however, notable differences compared with that produced by E 70 (lane 4). The most visible difference was the strong hypersensitive bands around position Ϫ35 in the E 70 ⅐AsiA pattern. This footprint also showed an increased protection around positions Ϫ57 to Ϫ62 relative to that of E 70 (Fig. 7A, compare lanes 3  and 4). Under similar conditions and in the presence of 100 mM KCl (Fig. 7B), the RNA polymerase footprint on the DNA fragment was barely detectable, and we were unable to observe the presence of the hyperreactive band that we considered as the signature of the ternary open complex (Fig. 7B, lane 5).
In KGlu, we looked further for the presence of AsiA in the footprint when the cAMP⅐CAP complex was used to activate the system. As in the pattern afforded by the ternary complex, bands at position Ϫ35 were strongly visible in presence of E 70 ⅐AsiA⅐lacUV5 and CAP, strengthening the hypothesis that a quaternary complex forms in the presence of cAMP⅐CAP (Fig.  7A, lane 11). Also, addition of AsiA to a preformed E 70 ⅐lacUV5 complex did not yield the pattern characteristically perturbed by AsiA (Fig. 7A, lanes 5 and 13), confirming that the inhibitor cannot bind to 70 once the holoenzyme has already formed an open complex (13). Finally, when it was preincubated with AsiA, E 70 (⌬␣) failed to form any stable complex, and the lacUV5 fragment (5Ј-labeled on the non-template strand) yielded a pattern similar to that obtained with naked DNA (Fig. 7A, compare lanes 2 and 6). These results confirmed the conclusion that, in the presence of the glutamate anion, AsiA is part of a stable and functional open complex formed by E 70 at the lacUV5 promoter. Moreover, this particular open complex is susceptible to activation by cAMP⅐CAP, as demonstrated by the DNase I protection pattern showing evidence for the presence of the quaternary complex E 70 ⅐AsiA⅐lacUV5⅐CAP.

DISCUSSION
In this study, we describe the behavior of the E 70 ⅐AsiA complex in the presence of KCl and KGlu, two salts widely used in transcription studies. We have studied the mechanism of inhibition exerted by AsiA when it is bound to E 70 acting upon a ؊10/Ϫ35 promoter in greater detail (13,15,16). By runoff transcription analysis on the lacUV5 promoter, we show that the species E 70 ⅐AsiA is much less inhibited in KGlu than in KCl (Fig. 1). These observations suggested the existence of a transcriptionally active ternary complex at lacUV5, a typical bacterial promoter with Ϫ10 and Ϫ35 recognition sequences. Active ternary complexes containing AsiA have been previously observed at several promoters that do not require a Ϫ35 motif for open complex formation (12,13,15). A number of convergent observations support the hypothesis of a ternary complex at a ؊10/Ϫ35 promoter. First, we showed that AsiA was present in a stoichiometric amount when the E 70 ⅐AsiA entity was first incubated with lacUV5 in KGlu (Fig. 2), but not in KCl. Thus, we ruled out the possibility that, in the presence of KGlu, a spurious loss of AsiA could explain the restoration of E 70 activity. Furthermore, in Glu Ϫ buffer, only the extended DNase I footprint formed by E 70 ⅐AsiA at lacUV5 possessed a distinctive AsiA "signature" in the Ϫ35 region (Fig. 7A, lanes 3 and  11). In contrast, with AsiA bound to the holoenzyme in KCl, we observed only a weak protection against DNase I, which we interpreted as due to residual amounts of the binary E 70 ⅐lacUV5 complex (Fig. 7B, lane 5). A third piece of evidence comes from an investigation of the structure of the ternary complex by laser UV photoreactivity. 2 Laser UV photo-footprinting of a RNA polymerase⅐promoter complex allows precise probing of changes in the local structure of DNA (37). When we formed an open complex at lacUV5 with E 70 in 200 mM KGlu, thymine dimer formation at positions Ϫ34 and Ϫ33 was suppressed, probably due to contacts between 70 and the Ϫ35 region (37). In contrast, no such effect was found when AsiA was previously bound to E 70 . However, in this case, we observed signals in the Ϫ10 region, indicative of contacts that are normally present in open complexes (38). This again strongly suggests that even when access to open complex formation via the Ϫ35 region is prevented by AsiA, RNA polymerase can nevertheless form a transcriptionally competent species possessing the downstream structural characteristics of a normal open complex. The presence of a significant fraction of binary complex would have resulted in a corresponding decrease in the Ϫ34 dimer signal. This was not observed. We conclude that the effects we observed in the presence of glutamate belong to a real and probably unique ternary complex referred to as E 70 ⅐AsiA⅐lacUV5.
AsiA inhibits open complex formation by binding directly to the conserved region 4.2 of the 70 subunit. At a ؊10/Ϫ35 promoter, this binding interferes with the interactions between 70 region 4.2 and the Ϫ35 motif (12)(13)(14). Hindrance of these 2 G. Orsini and M. Buckle, unpublished results.  crucial interactions with concurrent inhibition of open complex formation has been recently described in the case of the gene 2 protein of bacteriophage T7 (39). Here, however, the inhibitor gp2 binds to the ␤Ј subunit of E 70 and thus indirectly disrupts the Ϫ35/ 70 region 4.2 interaction (39). Both inhibitors T4 AsiA and T7 gp2 are believed to operate in a mutually exclusive mode: either the inhibitor is bound to the enzyme with ensuing inhibition, or the polymerase is bound to the promoter without inhibition. Moreover, in the latter case, the inhibitor can no longer be bound to the enzyme (13,39).
By analyzing the mode of action of AsiA in the presence of the glutamate anion, we show that this mutual exclusion can be by-passed. As for many other DNA/protein systems in which Cl Ϫ has been replaced by Glu Ϫ , our results illustrate the general observation that, in the presence of Glu Ϫ , there is a substantial increase in the affinity of the proteins for their binding sites on DNA (17,40). The apparent K D values determined by gel shift assay (see above and Fig. 3) are totally in line with previous reports on the "glutamate effect" (40). Using the abortive initiation assay and a simple two-state model (Equation 1) for open complex formation, we showed by tau plot analysis that the rate constant k f of the isomerization step was decreased 55-fold when AsiA was present in the complex (Table  I). Inhibition by AsiA was therefore strongly maintained, as shown by the global 120-fold effect on the second-order association constant K B k f . In a parallel study, we used the abortive initiation assay to assess the stability of the ternary complex by heparin challenge (data not shown). The binary complex (without AsiA) formed in 200 mM KGlu was absolutely stable for Ͼ1 day. The ternary complex was much less stable and showed a first-order decay with a half-life of ϳ18 h. This value is considerably larger that the half-life associated with the conversion of the closed to the open complex. The final ternary open complex formed in KGlu is therefore clearly more stable than the closed intermediate.
The transcriptional activity of the ternary complex strongly implied the existence of a compensatory mechanism. It is well documented that the ␣ subunit of RNA polymerase can also participate in promoter recognition through specific interactions between ␣-CTD and upstream regions of the Ϫ35 hexamer called "UP elements" (41)(42)(43). In glutamate, we show here that the contribution to the binding energy brought about by ␣-CTD binding to lacUV5 is sufficient to partially overcome inhibition by AsiA and to allow the formation of a ternary E 70 ⅐AsiA⅐lacUV5 active complex. Using the reconstituted holoenzyme E 70 (⌬␣), we show that the deletion of the ␣-CTD totally prevents the formation of the ternary complex. This complex displays RNA polymerase/promoter upstream contacts as revealed by a characteristic DNase I footprinting pattern that depends on ␣-CTD binding upstream of position Ϫ40 (Fig.  7, lanes 3 and 11). A similar differential pattern was previously observed in the absence of AsiA at promoters where the UP elements are essential (34,43). Also, ␣-CTD binding can be facilitated by the binding of the cAMP⅐CAP complex at a site centered at Ϫ61.5. Interactions between ␣-CTD and CAP allow one of the two ␣-CTDs to anchor more tightly in the minor groove around position Ϫ43, upstream of the Ϫ35 hexamer (44). Interestingly, this interaction occurs regardless of whether AsiA is bound to the adjacent 70 subunit, and it allows CAP to activate the ternary complex E 70 ⅐AsiA⅐lacUV5, thereby forming a quaternary transcriptionally competent complex (Fig. 6).
The compensatory mechanism documented here at lacUV5 has also been observed, almost unmodified, on lacP S , a parent and weaker promoter in which the Ϫ10 region is altered, but not the sequences participating in upstream contacts. Conversely, when, at lacUV5, the UP element is reinforced by insertion of a proper canonical sequence, the inhibition due to AsiA was found to be already attenuated in the KCl-containing buffer. 3 It is therefore likely that a better anchoring of the RNA polymerase⅐AsiA complex can conceivably take place on the UP element at any promoter, but that a clear balance favoring this repositioning might or might not require the presence of potassium glutamate depending on the relative strengths of the interactions at the Ϫ35 hexamer and at the UP element.
At this point, we will briefly consider the transcription properties of the E 70 ⅐AsiA complex relative to phage T4 early and MotA-dependent middle promoters. AsiA inhibits complex formation at the T4 early promoter P15.0, possessing both a Ϫ35 recognition element and an extended Ϫ10 motif (16). When assayed in KCl buffer, transcription from the T4 middle promoters P uvsX , PrIIB1, PrIIB2, and P1 is strictly dependent on the presence of both activators: MotA bound to the Ϫ30 middle promoter sequence, and AsiA bound to 70 (8). Remarkably, when assayed in KGlu, E 70 is able to transcribe correctly from P uvsX (5,7). This basal transcription by E 70 at a middle promoter is not activated by MotA (45). Furthermore, in KGlu, addition of AsiA to E 70 inhibits both open complex formation at a P uvsX DNA fragment and transcription from this middle promoter (7). Taken together, these observations emphasize the point that AsiA acting alone behaves as a repressor. It inhibits transcription from a MotA-dependent promoter in a manner reminiscent of its action upon a Ϫ10/Ϫ35 promoter. It is therefore likely that most of the RNA polymerase/promoter contacts at Ϫ35 (and at Ϫ30 as well) are perturbed by a local conformational change affecting 70 when it binds AsiA. This perturbation is, in turn, relieved by the new contacts and therefore the new specificity conferred by the addition of MotA.
In this view, this study demonstrates that promoter upstream contacts mediated by ␣-CTD of E 70 are able to reduce the strong structural and kinetic block brought about by AsiA. As a repressor, AsiA changes radically the potential use of the enzyme modules involved in the formation of contacts with the upstream region of the promoter. We propose that in vivo, this drastic change also occurs. Here, however, the ␣-CTDs are first ADP-ribosylated at the onset of T4 infection (46). This modification plays a major role in regulating promoter utilization since it irreversibly blocks the use of the rescued pathway documented here. The efficiency of the E 70 ⅐AsiA complex will now crucially depend on the binding of MotA to those promot-ers containing the Ϫ30 middle promoter motif (47,48) and on the establishment of the proper contacts between the 70 subunit and MotA (49).