Recruitment of sigma54-RNA polymerase to the Pu promoter of Pseudomonas putida through integration host factor-mediated positioning switch of alpha subunit carboxyl-terminal domain on an UP-like element.

The interactions between the sigma54-containing RNA polymerase (sigma54-RNAP) and the region of the Pseudomonas putida Pu promoter spanning from the enhancer to the binding site for the integration host factor (IHF) were analyzed both by DNase I and hydroxyl radical footprinting. A short Pu region centered at position -104 was found to be involved in the interaction with sigma54-RNAP, both in the absence and in the presence of IHF protein. Deletion or scrambling of the -104 region strongly reduced promoter affinity in vitro and promoter activity in vivo, respectively. The reduction in promoter affinity coincided with the loss of IHF-mediated recruitment of the sigma54-RNAP in vitro. The experiments with oriented-alpha sigma54-RNAP derivatives containing bound chemical nuclease revealed interchangeable positioning of only one of the two alpha subunit carboxyl-terminal domains (alphaCTDs) both at the -104 region and in the surroundings of position -78. The addition of IHF resulted in perfect position symmetry of the two alphaCTDs. These results indicate that, in the absence of IHF, the sigma54-RNAP asymmetrically uses only one alphaCTD subunit to establish productive contacts with upstream sequences of the Pu promoter. In the presence of IHF-induced curvature, the closer proximity of the upstream DNA to the body of the sigma54-RNAP can allow the other alphaCTD to be engaged in and thus favor closed complex formation.

Bacterial promoters are modular DNA regions able to establish productive interactions both with subunits of RNA polymerase holoenzyme (RNAP 1 ; subunit composition: ␣ 2 ␤␤Ј) and regulatory proteins (1)(2)(3)(4). The core promoter elements are signature tags for factor selectivity (4). The major sigma factor 70 generally directs RNAP to interact with the core DNA elements Ϫ10 and Ϫ35 (hexamers with consensus 5Ј-TATAAT-3Ј and 5Ј-TTGACA-3Ј, respectively) (5). The core promoter elements can be both overlapped and flanked by protein-bound DNA sites involved in the fine modulation of promoter activity (2,4). In the last decade, a considerable amount of attention has been given to a A ϩ T-rich promoter sequence, the UP element, located upstream of the core promoter region and consisting of two distinct subsites, each of which, by itself, can be bound by the carboxyl-terminal domain of the RNAP ␣ subunit (␣CTD) (3, 6 -10). ␣CTD recognizes and interacts with the backbone structure in the minor groove of the UP element. The A ϩ T-rich sequence of the UP element are needed to provide the optimum width of minor groove for interaction with ␣CTD (7,11). Several lines of evidence showed that the role of the UP elements is to stimulate transcription in an activator protein-independent manner and to a different extent (from 1.5 to ϳ90-fold) depending on the similarity with the consensus UP element sequence (3,9). It is currently believed that the transcription stimulation by an UP element has to be traced mainly to the cooperation of the sigma factor and the ␣ subunit in RNAP binding to the promoter (1,12). Thus, the presence of an UP element in a promoter plays the major role of increasing the initial equilibrium constant of closed complex formation between RNAP and promoter DNA (13,9). However, influences of ␣-UP element interaction on later steps in transcription initiation were also reported (13,14). The location of the UP element with respect to the transcription start site can influence the degree of transcription stimulation (15). In the Escherichia coli rrnB P1 promoter, the UP element is located in a region spanning form the Ϫ40 and Ϫ60 positions and is able to increase transcription from 30-to 70-fold (6,13). The artificial upstream re-location of the rrnB P1 UP element by a single turn of DNA helix decreases but does not prevent transcription stimulation, while further displacements abolish UP elementdependent transcription (15). The ability of ␣CTD to contact DNA and/or activator molecules at different locations upstream of the core promoter (8,(15)(16)(17)(18)(19)(20)(21) has been attributed to the flexibility of the linker connecting ␣CTD to the ␣ amino-terminal domain (␣NTD) (8,22) assembled in the body of RNAP. This linker flexibility also accounts for the ability of the two copies of ␣CTD to function interchangeably with respect to the subsite recognition within the UP element (10). Sufficient length of the linker between ␣CTD and ␣NTD is also needed for UP element-dependent transcription activation. The linker is flexible but structured to a certain extent to facilitate the positioning of the ␣CTD to a proper location for interaction with the UP element (23,15).
UP-like elements were also found in promoters recognized by alternative sigma factors, such as the D -dependent flagellin promoter of Bacillus subtilis (24) and the 54 -dependent Pu promoter of Pseudomonas putida ( Fig. 1) (25,26). The latter drives transcription of TOL plasmid upper operon for the degradation of toluene (27) and shows the typical modular structure of the 54 -dependent promoters (for review, see Refs. 28 and 29) that consists of: (i) the Ϫ12/Ϫ24 region (consensus: TGGCAC N5 TTGCa/t located between positions Ϫ11 and Ϫ26) (30) recognized by 54 and considered the functional analogue of Ϫ10/Ϫ35 core promoter bound by 70 (31), (ii) DNA enhancer sequences (known as upstream activating sequences or UAS) targets for the activators of the 54 -RNAP, usually located at Ͼ100 bp from the transcription start site, and (iii) an intervening sequence between UAS and Ϫ12/Ϫ24 motif that may contain a target site of the IHF (32), which, by its ability to bind and bend DNA sequences, assists the looping out required to bring closer together the 54 -activator prebound at UAS and 54 -RNAP assembled in a closed complex with Ϫ12/Ϫ24 DNA region. The productive contact between 54 -activator and 54 -RNAP closed complex triggers promoter opening (open complex) and eventually transcription initiation (33,34).
Within this typical modular scheme for 54 -dependent promoters, the P. putida Pu promoter presents unique features. In fact, our previous results showed the additional IHF role of stimulating the otherwise limiting step of closed complex formation between 54 -RNAP and Pu DNA (26,35). We also showed that the recognition of Pu promoter by 54 -RNAP involves not only the Ϫ12/Ϫ24 region but also a functional equivalent of an UP element located in the intervening region, upstream to the IHF binding site. Furthermore, our data strongly suggested that the Pu UP element could play a key role in the IHF-mediated stimulation of closed complex formation by 54 -RNAP. In this work, we closely inspected the interactions, both in the presence and absence of IHF, between 54 -RNAP and the Pu intervening region located upstream the IHF binding site. The data strongly support the notion of a non-canonical arrangement of the stimulating DNA sequences functioning as UP element. 54 -RNAP upstream interactions concentrate on two sites located in the surroundings of positions Ϫ104 and Ϫ78, respectively, thus being distant about 25 bp. In the absence of IHF and probably due to asymmetrical positioning of the upstream DNA, the two sites can be contacted interchangeably only by one ␣CTDs of the ␣ 2 ␤␤Ј 54 complex constituting the 54 -RNAP. On the contrary, the bending by IHF apparently introduces symmetry to the nucleoprotein complex allowing the other ␣CTD to interact with the two sites. Thus, the IHFmediated stimulation of closed complex would result from curvature-dependent increased probability of wide range upstream interactions by 54 -RNAP through the ␣CTDs.

MATERIALS AND METHODS
Bacterial Strains, Plasmids, and General Procedures-Plasmid pEZ9 (25) contains the entire Pu promoter sequence as a 312-bp EcoRI-BamHI insert in pUC18 spanning positions Ϫ208 to ϩ93. P. putida strains KT2442 hom.fg/xylRS and its derivative HFPu (Pu::lacZ, xylR ϩ ) have been described previously (36,37). KT2442PuXhoCla (PuXhoCla::lacZ, xylR ϩ ), KT2442PuScra1 (PuScra1::lacZ, xylR ϩ ), and KT2442PuScra2 (PuScra2::lacZ, xylR ϩ ) carrying mutant Pu::lacZ fusions in the same location of the chromosome as HF Pu were obtained as follows. The Pu version, cloned in pUC-PuClaI-79, derived from pEZ9 and bearing a ClaI site engineered within positions Ϫ79 and Ϫ84 (26), was subjected to site-directed mutagenesis by the QuikChange TM site-directed mutagenesis kit (Stratagene) to engineer a XhoI site within nucleotides Ϫ121 to Ϫ126. This procedure generated the plasmid pUC-PuClaXho. The replacement of the 47-bp XhoI-ClaI fragment of pUC-PuClaXho for synthetic XhoI-ClaI fragments harboring scram-bled sequences from nucleotides Ϫ105 to Ϫ120 and from Ϫ95 to Ϫ120 gave rise to plasmids pUC-PuScra1 and pUC-PuScra2, respectively. The Pu versions present in pUC-PuClaXho, pUC-PuScra1, and pUC-PuScra2, respectively, were rescued as 312-bp EcoRI-BamH fragments, fused to lacZ by cloning in the corresponding sites of pBK16 vector (36) and recombined with the homology fragment inserted in the chromosome of KT2442 hom.fg/xylRS as described previously (36). All cloned inserts and DNA fragments were verified before use by automated DNA sequencing. Recombinant DNA manipulations were carried out according to published protocols (38).
For gel retardation assays, the PCR-amplified fragments described above were end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase. Radioactive nucleotides not incorporated in DNA were removed by centrifuging briefly in small Sephadex G-25 columns. Binding reactions were performed in a total volume of 25 l of transcription buffer containing 35 mM Tris acetate, 70 mM KAc, 5 mM MgAc 2 , 20 mM NH 4 Ac, 2 mM CaCl 2 , 1 mM dithiothreitol, 3% glycerol, and 40 g/ml of poly[d(I-C)]. Labeled fragments, added to the buffer at a final concentration of 5 nM, were incubated with 100 nM IHF, 60 nM core RNAP, and a 3-fold molar excess of 54 factor for 25 min at 30°C. The entire reaction volume was loaded onto non-denaturing 4% polyacrylamide gels (acrylamide:bis ratio 80:1) in 0.5ϫ TBE buffer (45 mM Tris borate, pH 8.3, 0.1 mM EDTA, 5 mM MgCl 2 ), electrophoresed at 12 mM at 4°C for 6 h, and dried. Bands were visualized and quantified by Typhoon 8600 variable mode imager (Amersham Biosciences) upon storage phosphor autoradiography. DNA footprinting assays were performed in a total volume of 50 l and with similar concentrations of end-labeled fragments and proteins used in the gel mobility-shift assays. For DNase I footprinting, after preincubation of end-labeled Pu DNA and proteins in transcription buffer for 25 min at 30°C, 3 ng of DNase I were added to each sample and further incubated for 3.5 min. Reactions were stopped by addition of 25 l of STOP buffer containing 0.1 M EDTA, pH 8, 0.8% SDS, 1.6 M NH 4 Ac, and 300 g/ml sonicated salmon sperm DNA. Nucleic acids were precipitated with 175 l of ethanol, lyophilized and directly resuspended in denaturing loading buffer (7 M urea, 0.025% bromphenol blue, and 0.025% xylene cyanol in 20 mM Tris, pH 8) prior to loading on a 7% DNA sequencing gel. A ϩ G Maxam and Gilbert reactions (41) were carried out with the same fragments and loaded onto the gels along with the footprinting samples. For hydroxyl radical footprinting, after preincubation of end-labeled Pu DNA and proteins for 25 min at 30°C in hydroxyl radical buffer containing 25 mM HEPES, 70 mM KAc, 5 mM MgAc 2 , 19 mM NH 4 Ac, 0.7 mM DTT, 1% glycerol, and 40 g/ml of poly[d(I-C)], 3 l each of [Fe(EDTA)] 2Ϫ (125 mM (NH 4 ) 2 Fe(SO 4 ) 2 ⅐6H 2 O, 250 mM EDTA), 28 mM ascorbate, and 0.84% hydrogen peroxide were added to the samples and then incubated for a further 5 min. Reactions were stopped by addition of 15 l of 0.1 M thiourea and 25 l of STOP buffer (as above), respectively. Nucleic acids were precipitated with 200 l of ethanol and treated for separation and visualization as described above for DNase I footprinting. For DNA footprinting in the presence of (Fe⅐BABE)-RNAPs, after preincubation of end-labeled Pu DNA and proteins for 25 min at 30°C in hydroxyl radical buffer, 1 l each of 100 mM ascorbate and 0.6% hydrogen peroxide were added to the samples and then incubated for a further 15 min. Reactions were stopped by addition of 10 l of 0.1 M thiourea and 25 l of STOP buffer (as above), respectively. Nucleic acids were pre-cipitated with 200 l of ethanol and treated for separation and visualization as described above for DNase I footprinting.

-RNAP Per Se Can Contact Sequence Elements Located
Upstream the IHF Binding Site of Pu Promoter-The affinity of 54 -RNAP for Pu promoter DNA lacking the sequence region located upstream the position Ϫ79 is strongly reduced ( Fig. 1) (26). These data suggested that 54 -RNAP could utilize per se additional DNA affinity elements located far upstream of the Ϫ12/Ϫ24 region involved in the interaction with 54 . Since ␣CTD-deleted derivatives of 54 -RNAP showed reduced promoter affinity (26), we also suggested that ␣CTD could be directly involved in the recognition of such upstream DNA, reminiscent, in this case, of the UP elements of 70 promoters. Furthermore, an ␣CTD/UP-like interaction also seemed to be involved in the mechanism behind the stimulation of closed complex formation between Pu DNA and 54 -RNAP on IHF binding and bending, which we referred to as IHF-mediated recruitment of 54 -RNAP to the Pu promoter (26). To map the contact sites of 54 -RNAP upstream Ϫ79 position and to reveal any possible influence of IHF-induced bending on these upstream contacts, we carried out either DNase I and hydroxyl radical footprinting assays on end-labeled DNA fragments bearing the entire Pu mixed with increasing amounts of 54 -RNAP, both in the presence and in the absence of sub-saturating concentrations of purified IHF. As shown in Fig. 2A (lanes  1-3), the footprint of the 54 -RNAP in the region spanning approximately from Ϫ83 to Ϫ108 consisted in a generalized hypersensitivity to DNase I. The addition of IHF ( Fig. 2A, lanes  4 and 5) caused a further slight increase of upstream DNA reactivity to DNase I, with the exception of positions Ϫ90 and Ϫ91, appearing to be more protected from DNase I cleavage than when in the presence of 54 -RNAP alone. Similar conditions of labeled Pu DNA/protein ratio were employed to per-form hydroxyl radical footprinting assays (Fig. 2B). The inspection of the Pu DNA sequence upstream of the IHF binding site revealed both protected and hypersensitive sites in a region spanning from Ϫ95 to Ϫ109. In particular, as shown in Fig. 2B  (lanes 1-4), the hydroxyl radical footprint of 54 -RNAP consisted of three hypersensitive positions, Ϫ103 to Ϫ105, which seemed to be flanked by short protected regions. Apparently, the addition of IHF protein (Fig. 2B, lanes 5-7) caused minimal changes to the pattern of hydroxyl radical cutting. Examined together, these results strongly indicated that, even when unassisted, 54 -RNAP can establish contacts to DNA sites located far upstream of the Ϫ12/Ϫ24 region. In the case of the hydroxyl radical footprint, the DNA region involved in interactions with 54 -RNAP appeared to be limited to a short sequence in the surroundings of position Ϫ104, which we named UP-like Pu I . In our previous work (26), to explain the mechanism behind the IHF-mediated recruitment of 54 -RNAP, we suggested that the distance between the Ϫ12/Ϫ24 site and the UP element(s) might disfavor either the formation or the maintenance of simultaneous binding by 54 -RNAP through 54 and ␣CTD, respectively. Thus, the key recruiting action of IHF-induced bending would have consisted of increasing 54 -RNAP affinity for Pu DNA by bringing the Ϫ12/Ϫ24 site and the UP element(s) into a closer proximity. Apparently, from the footprinting analysis presented in Fig. 2, the ability of 54 -RNAP to establish contacts with the Pu region located upstream of the IHF binding site seemed to be enhanced to a limited extent only by the addition of IHF. However, further analysis with free radical-delivering 54 -RNAPs (see below) showed more clearly that IHF-induced bending can cause increased occupancy by ␣CTD of the region upstream to the IHF binding site.
The us to identify an upstream DNA element, UP-like Pu I , which is contacted by 54 -RNAP in the closed complex with the Pu promoter. To evaluate the role of the contacts with UP-like Pu I in determining the affinity of 54 -RNAP for the Pu promoter, we ran gel retardation assays on the nucleoprotein closed complexes formed by 54 -RNAP with DNA fragments bearing either the Pu sequence up to position Ϫ126 (Pu-126; Fig. 1) supposed to include the whole UP-like Pu I site or progressively shorter Pu sequences extending up to Ϫ114, Ϫ105, and Ϫ85, respectively (Pu-114, Pu-105, Pu-85; Fig. 1). Side-by-side comparison of the amounts of complex assembled by 54 -RNAP with either Pu-126 or Pu-114, respectively, showed no significant difference (data not shown). On the contrary, the amounts of complex formed by 54 -RNAP with Pu-105 were strongly reduced with respect to Pu-114 (Fig. 3, lanes 5 and 6). A more extended deletion up to position Ϫ85 (Pu-85) (Fig. 3, lane 4) did not decrease further the amounts of nucleoprotein complex with 54 -RNAP.
To test the requirement of UP-like Pu I integrity for the IHFmediated recruitment of 54 -RNAP, we added a sub-saturating concentration of purified IHF protein to the mixtures of 54 -RNAP with Pu-85, Pu-105, and Pu-114, respectively. As shown in Fig. 3 (lanes 4 -9), while the binding of 54 -RNAP to Pu-114 could be stimulated by IHF as shown previously (26), IHF failed to enhance closed complex formation with Pu-105 and Pu-85, respectively. Thus, these results strongly indicated that the interactions established by 54 -RNAP with UP-like Pu I site in the closed complex are instrumental in determining promoter affinity. In addition, as the integrity of UP-like Pu I is also required to observe IHF-mediated enhancement of 54 -RNAP recruitment, we speculated that the promoter architecture imposed by IHF binding contributes to the interactions between 54 -RNAP and the UP-like Pu I site.

The Scrambling of UP-like Pu I DNA Region Affects Pu Performance in Vivo-In view of the previous results in vitro, the disruption of the integrity of UP-like Pu
I site was expected to affect Pu activity. To address this issue, we aimed to monitor in vivo the consequences of altering the sequence spanning the UP-like Pu I on Pu expression pattern. To this end, we introduced progressive sequence scrambling into the DNA region from Ϫ120 to Ϫ93 sites (Fig. 1) by replacement with synthetic double-stranded oligomers for the wt DNA sequence located between the XhoI and ClaI sites that were opportunely engineered within positions Ϫ126 and Ϫ121, and Ϫ84 and Ϫ79, respectively, in the Pu variant, PuXhoCla (Fig. 1). From this procedure, we obtained two Pu derivatives, PuScra1 and PuS-cra2 (Fig. 1), that, along with their parental PuXhoCla, were fused to lacZ and recombined into the chromosome of P. putida KT2442 hom.fg/xylRS as described previously (36). As shown in Fig. 4A, the comparison of accumulation of ␤-galactosidase upon toluene induction of HFPu (Pu::lacZ, xylR ϩ ) and KT2442PuXhoCla (PuXhoCla::lacZ, xylR ϩ ), respectively, revealed that the engineering of the XhoI and ClaI sites described above did not affect Pu performance. On the contrary, as shown in Fig. 4B, the comparison of ␤-galactosidase accumulation upon toluene induction of KT2442PuXhoCla, KT2442PuScra1 (PuScra1::lacZ, xylR ϩ ) and KT2442PuScra2 (PuScra2::lacZ, xylR ϩ ), respectively, revealed that the promoter activity of PuScra2 was severely impaired. Since the sequence scrambling did not introduce either phase alteration of key regulatory sites (UAS, IHF box and Ϫ12/Ϫ24 sites) or a predictable drastic variation of promoter DNA curvature (42), we inferred that the reduction of promoter activity of PuScra2 was caused by the destruction of upstream contacts between 54 -RNAP and the UP-like Pu I site. Monitoring the Positioning of the Two ␣CTDs of 54 -RNAP along Upstream Sequences of the Pu Promoter in the Absence and in the Presence of IHF-In the ␣ 2 ␤␤Ј core RNAP complex, the two identical ␣ subunits can be distinguished by their arrangement with respect to ␤ and ␤Ј subunits. In fact, one ␣, ␣ I , interacts with the ␤, whereas the other, ␣ II , interacts with ␤Ј (43,40,17). To provide a more precise definition of the positioning of the ␣CTD of ␣ I and ␣ II (␣CTD I and ␣CTD II , respectively) along the Pu DNA region upstream the IHF site, we set out to exploit the UP DNA cleavage capability caused by free radicals originated from (p-bromoacetamidobenzyl)-EDTA⅐Fe (Fe⅐BABE) attached to the UP contact surface of one or both ␣ subunits assembled in the RNAP holoenzyme complex (40,17). We prepared free radical-releasing 54 -RNAP by adding a saturating amount of 54 both to oriented-␣ and non-oriented (Fe⅐BABE)-labeled RNAP core complexes: ␣ I (Fe)/␣ II and ␣ I / ␣ II (Fe), in which the (Fe⅐BABE) moiety is bound to ␣CTD I and ␣CTD II , respectively, and ␣ I (Fe)/␣ II (Fe), in which the (Fe⅐BABE) moiety is bound to both ␣CTDs. Such (Fe⅐BABE)labeled 54 -RNAPs were incubated with end-labeled Pu promoter DNA, and the pattern of Pu DNA fragmentation was analyzed in sequencing gels as in the footprinting experiments presented in Fig. 2 (A and B). To test any differential influence of IHF on upstream DNA occupancy by ␣CTD I or ␣CTD II , a set of reactions was also incubated in the presence of IHF. As shown in Fig. 5, in the absence of IHF, both ␣ I (Fe)/␣ II (Fe) and ␣ I (Fe)/␣ II (lanes 2 and 3)

FIG. 4. Involvement of UP-like Pu
I region in Pu promoter activity in vivo. P. putida KT2442 derivatives bearing the Pu-lacZ, PuXho-Cla-lacZ, PuScra1-lacZ, and PuScra2-lacZ transcriptional fusions recombined into the same site of the chromosome, respectively, were tested for the performance of expression of the lacZ reporter gene during the growth at 30°C in LB medium (38). Each strain was grown until cultures had an absorbance of 0.5 at 600 nm. Toluene was then administered and the incubation continued for the subsequent 3.5 h. Accumulation of ␤-galactosidase along the time and growth curves of each strain after toluene addition are shown. A, comparison of the performances of Pu-lacZ and PuXhoCla-lacZ in the presence and in the absence of induction with toluene, respectively. B, comparison of the performances of PuXhoCla-lacZ, PuScra1-lacZ, and PuScra2-lacZ, respectively, upon toluene induction. a fragmentation pattern that was very similar to the hydroxyl radical hypersensitivity profile at positions Ϫ103, Ϫ104, and Ϫ105 displayed by the end-labeled Pu in the presence of unlabeled 54 -RNAP (Fig. 2B, lanes 1-4). Furthermore, both ␣ I (Fe)/ ␣ II (Fe) and ␣ I (Fe)/␣ II could produce another discrete pattern of fragmentation closer to the IHF binding site, involving posi-tions spanning from Ϫ74 to Ϫ79. We named this second ␣CTDinteracting region, which did not result so evident in the previous footprinting experiments (Fig. 2, A and B), UP-like Pu II . Unlike ␣ I (Fe)/␣ II (Fe) and ␣ I (Fe)/␣ II , in the absence of IHF the other oriented-␣ 54 -RNAP, ␣ I /␣ II (Fe), could not produce any significant fragmentation of the end-labeled Pu DNA (Fig. 5,  lane 4). The addition of IHF did not substantially modify the fragmentation pattern by ␣ I (Fe)/␣ II (Fe) and ␣ I (Fe)/␣ II . On the contrary, the presence of IHF increased the fragmentation by ␣ I /␣ II (Fe). In fact, under these conditions, ␣ I /␣ II (Fe) was also able to generate a fragmentation pattern (Fig. 5, lane 8) identical to that of ␣ I (Fe)/␣ II (Fe) and ␣ I (Fe)/␣ II . Taken together, these results clearly revealed that ␣CTD interaction with the Pu upstream region can occur at two unusually distant sites, UP-like Pu I and UP-like Pu II , respectively. Both sites can be bound interchangeably by ␣CTD I , both in the absence and in the presence of IHF. Unlike ␣CTD I , and probably due to an asymmetrical arrangement of the DNA upstream the IHF site, ␣CTD II can efficiently contact both UP-like Pu I and UP-like Pu II only in the presence of IHF protein. These results strongly suggest that when the Pu promoter DNA conformation is structured by the binding and bending of IHF, the two ␣CTD can distribute simultaneously and interchangeably among UPlike Pu I and UP-like Pu II .

DISCUSSION
The activity of the Pu promoter of P. putida is strongly influenced by DNA architecture. Remarkably, the bending activity of IHF strongly augments the probability of interaction between the activator XylR and 54 -RNAP (44) and also positively influences the docking of 54 -RNAP on the promoter (26). Our previous studies showed the involvement of ␣CTD in the interaction with a Pu region reminiscent of an UP element located upstream to the IHF binding site (26). Furthermore, our results suggested that the upstream interactions of 54 -RNAP by ␣CTD could play an active role in its IHF-mediated recruitment. However, the relationship between topology of the promoter DNA and topography of such upstream interactions required further clarification. In fact, this might be at the basis of the IHF-mediated enhancement mechanism of closed complex formation. In this study, we investigated the nature and the role of the upstream contacts that 54 -RNAP is able to establish with the Pu promoter both in the presence and in the absence of IHF protein. The footprinting experiments presented in Fig. 2 sustained more clearly than our previous works (26) the notion that 54 -RNAP per se is able to establish contacts with DNA sequences located upstream to the IHF binding site. In addition, these experiments allowed us to identify a discrete upstream DNA site surrounding position Ϫ104, UPlike Pu I (Fig. 2B), engaged in the interactions with 54 -RNAP. Then, we addressed the issue of the functional significance of the 54 -RNAP/UP-like Pu I contacts. At least two lines of evidence indicated that the 54 -RNAP contacts with UP-like Pu I are functional interactions participating in Pu promoter activity. First, alterations of the Pu sequence spanning from Ϫ114 to Ϫ85 positions (Pu-105 and Pu-85, respectively) affect promoter affinity for 54 -RNAP per se (Fig. 3, lanes 4 -6). Second, the rearrangement of the sequence surrounding UP-like Pu I in Pu-Scra2 derivative severely affects Pu activity in vivo. The experiment presented in Fig. 3 also indicated that UP-like Pu I participates in the IHF-mediated stimulation of closed complex formation by 54 -RNAP even though its occupancy by 54 -RNAP did not seem to change significantly upon the addition of IHF (Fig. 2B). In our previous work (26), to explain the IHFmediated recruitment of 54 -RNAP, we considered the possibility that IHF binding and bending could strengthen the upstream contacts established by ␣CTD with Pu DNA. We figured The location of the IHF binding site, some coordinates along the promoter sequence, and the regions corresponding to the UP-like Pu I and UP-like Pu II sites are indicated to the sides, using the Maxam and Gilbert A ϩ G reaction as a reference. Nucleotides, which become hypersensitive to the cleavage with hydroxyl radicals, are indicated with closed arrows (see Fig. 1). The concentration of IHF used was 100 nM. RNAPs contained 60 nM, core enzyme mixed, in each case, with a 3-fold molar excess of purified 54 . that this could be accomplished by one or the combination of the following two mechanisms: (i) IHF-induced curvature centered at Ϫ68 would bring into closer proximity the Ϫ12/Ϫ24 region and an UP element located upstream of the IHF binding site, thus favoring the simultaneous interaction of 54 -RNAP with both sites, and (ii) the IHF binding would locally distort the double helix of DNA strengthening the ␣CTD/UP interaction. However, the idea that IHF could reinforce upstream contacts by ␣CTD contrasted with the evidence presented in Fig. 2 that IHF addition did not substantially alter the pattern of the upstream footprint of the 54 -RNAP. Furthermore, this had to be reconciled with the fact that the integrity of UPlike Pu I was required both to determine promoter affinity and stimulate IHF-mediated closed complex formation. The experiments with (Fe⅐BABE)-labeled 54 -RNAPs showed that the hypothesis of the IHF-mediated strengthening of the ␣CTD upstream contacts was correct. However, the possible scenario (Fig. 6) can be more complex than previously thought (Ref. 26 and see above). First, ␣CTD can bind to UP-like Pu I and also to another site, UP-like Pu II , located downstream. In the absence of IHF (Fig. 6A), only one ␣CTD, ␣CTD I (i.e. ␣CTD of ␣ that associates with ␤), can bind interchangeably to both UP-like Pu I and UP-like Pu II . These asymmetrical upstream ␣CTD interactions determine a first level of promoter affinity. The fact that in the absence of IHF only ␣CTD I can interact efficiently with upstream DNA could be explained by: (i) asymmetries in the ␣ 2 ␤␤Ј 54 complex, (ii) the axis of the upstream DNA is not in line with the body of 54 -RNAP, and (iii) a combination of point i and ii. However, the binding and bending of IHF makes the ␣CTD interactions more symmetrical (Fig. 6B). In fact, in this case, the other ␣CTD, ␣CTD II (i.e. ␣CTD of ␣ that associates with ␤Ј), can also bind interchangeably to both UP-like Pu I and UP-like Pu II . These symmetrical ␣CTD interactions determine a higher second level of promoter affinity which would underlie the IHF-mediated recruitment of 54 -RNAP to the Pu promoter. The topological shift from asymmetrical to symmetrical ␣CTD interactions can be attributed to the IHF-induced bending that brings the Ϫ12/Ϫ24 site and the upstream DNA into closer proximity. In fact, this would favor the 54 -RNAP contacts with the upstream DNA also through the ␣CTD II domain. Despite this IHF-mediated topological switch, the flexibility of the long unstructured interdomain linker connecting ␣CTD to the rest of ␣ (8, 22) might also account for the ability of 54 -RNAP to contact interchangeably the UP-like Pu I and UPlike Pu II sites located at a distance from the Ϫ12/Ϫ24 core promoter region.
Both UP-like Pu I and UP-like Pu II sites may resemble the UP element subsites, each of which constitutes a binding site for ␣CTD (10). The certain assignment of UP-like Pu I and UPlike Pu II to one of the two classes, distal-or proximal-type, of UP subsite (10) cannot be deduced from this study. Moreover, it remains to be clarified whether UP-like Pu I and UP-like Pu II are arranged adjacently as in the UP element consensus sequence (10) or are separated by turns of DNA helix. Hypothetically, the A-tracts at positions Ϫ102 to Ϫ105 and Ϫ77 to Ϫ82 within UP-like Pu I and UP-like Pu II (Fig. 1), respectively, could constitute the core subsites (7,10,45). In view of this, we suggested that the distance between the core A-tracts is not consistent with a side-by-side arrangement as for the UP subsites in the canonical UP element (10), and there would be turns of helix between UP-like Pu subsites. Despite the distance between UPlike Pu I and UP-like Pu II , it would still be possible that the binding of one ␣CTD to one UP-like Pu subsite cooperatively assists the second copy of ␣CTD to bind to the other UP-like Pu subsite. This could be accomplished by a combination of protein-protein interaction between the two ␣CTDs and local DNA flexibilization. The hypersensitivity to DNase I of the region upstream to the IHF binding site ( Fig. 2A), both in the absence and in the presence of IHF, could trace the DNA bending of the UP-like Pu arising from the cooperative binding of the two ␣CTDs. We suggest that even in the absence of IHF, the binding of ␣CTD I to one UP-like Pu subsite may recruit an otherwise distant ␣CTD II to the other UP-like Pu subsite, and this would originate bending of the UP-like Pu . The promoter bending introduced by IHF that renders the UP-like Pu region more accessible to ␣CTD II (see above) may result in mutual and stronger cooperative binding of two the ␣CTDs. The slight increase in the general DNase hypersensitivity and DNase I protection of positions Ϫ90 and Ϫ91 in the presence of IHF ( Fig. 2A, lanes 3  and 4) might account for the strengthening of the cooperative binding of ␣CTD to the UP-like Pu region. Since no evidence for the occupation of other sites different to UP-like Pu I and UP-like Pu II resulted from the assay with ␣ I (Fe)/␣ II (Fe) 54 -RNAP, we also suggest that the cooperative occupancy of the upstream DNA region involves only UP-like Pu I and UP-like Pu II subsites.
In summary, the evidence presented in this work strongly supports the notion of IHF-mediated topological switch that governs the occupancy of a promoter by RNAP through the curvature-mediated modulation of ␣CTD interactions with the upstream promoter region. It was shown previously that, by protein-protein interaction, transcription factors can direct the ␣CTD positioning on the upstream promoter region (17,19). In UP-like Pu II (II) subsites, thus enhancing the closed complex formation of the holoenzyme with its target sequences at Ϫ12/Ϫ24. A, in the absence of IHF protein, the 54 -RNAP complex interacting with Ϫ12/Ϫ24 motif through the factor 54 can establish upstream contacts mainly with ␣CTD I which collocates interchangeably on either I or II site. The interactions of ␣CTD II with the upstream DNA are probably hindered by an unfavorable distance of both I and II sites from the body of the holoenzyme. Even though the DNA/␣CTD II interactions are disfavored, the binding of ␣CTD I to the upstream sites may cooperatively assist the docking of ␣CTD II (not represented). B, the promoter geometry created by the IHF binding can make the distance of both I and II sites from the body of the holoenzyme more favorable for ␣CTD II . This supposedly strengthens both the ␣CTD II interactions with I and II sites alone and the cooperative binding of the two ␣CTDs. The full positioning of the two ␣CTDs on the upstream DNA in the presence of IHF-induced bending would enhance closed complex formation. a novel way, the positioning of ␣CTD on the Pu promoter would be directed predominantly by the DNA architecture.