ExsA Recruits RNA Polymerase to an Extended −10 Promoter by Contacting Region 4.2 of Sigma-70

ABSTRACT ExsA is a member of the AraC family of transcriptional activators and is required for expression of the Pseudomonas aeruginosa type III secretion system (T3SS). ExsA-dependent promoters consist of two binding sites for monomeric ExsA located approximately 50 bp upstream of the transcription start sites. Binding to both sites is required for recruitment of σ70-RNA polymerase (RNAP) to the promoter. ExsA-dependent promoters also contain putative −35 hexamers that closely match the σ70 consensus but are atypically spaced 21 or 22 bp from the −10 hexamer. Because several nucleotides located within the putative −35 region are required for ExsA binding, it is unclear whether the putative −35 region makes an additional contribution to transcription initiation. In the present study we demonstrate that the putative −35 hexamer is dispensable for ExsA-independent transcription from the P exsC promoter and that deletion of σ70 region 4.2, which contacts the −35 hexamer, has no effect on ExsA-independent transcription from P exsC . Region 4.2 of σ70, however, is required for ExsA-dependent activation of the P exsC and P exsD promoters. Genetic data suggest that ExsA directly contacts region 4.2 of σ70, and several amino acids were found to contribute to the interaction. In vitro transcription assays demonstrate that an extended −10 element located in the P exsC promoter is important for overall promoter activity. Our collective data suggest a model in which ExsA compensates for the lack of a −35 hexamer by interacting with region 4.2 of σ70 to recruit RNAP to the promoter.

Pseudomonas aeruginosa is an opportunistic human pathogen that causes a variety of acute and chronic infections in immunocompromised individuals (52,53). A primary determinant of P. aeruginosa virulence is a type III secretion system (T3SS) (24,70). The T3SS consists of a macromolecular conduit through which effector toxins are translocated into eukaryotic host cells (70). The translocated toxins promote tissue destruction and evasion of the host immune response (3,55,69). Mutants lacking a functional T3SS have reduced virulence in both tissue culture and animal infection models (2,33).
The central regulator of T3SS gene expression is ExsA (25,67,68). ExsA directly binds to 10 different promoters and activates transcription of the core genes required for assembly and function of the T3SS (12,64). ExsA belongs to the family of AraC/XylS transcriptional regulators. AraC family members generally consist of an amino-terminal ligand interaction domain and two carboxy-terminal helix-turn-helix DNA-binding motifs (14). AraC regulators are often classified by the type of ligand that regulates their activity. ExsA belongs to a small subset of AraC regulators that respond to protein ligands and control T3SS gene expression (50). Representatives of this subfamily are also found in Vibrio parahaemolyticus (ExsA), Shigella flexneri (MxiE), and Salmonella enterica (InvF) (17,25,42,71). ExsA-dependent transcription in P. aeruginosa is antagonized by ExsD, which functions as an antiactivator by inhibiting the DNA-binding activity and self-association properties of ExsA (13,43,61). Similarly, the ExsA homolog in Vibrio parahaemolyticus is required for T3SS1 gene expression, and the ExsD homolog is thought to antagonize ExsA activity (71). Transcriptional activation by S. flexneri MxiE is also antagonized by a protein ligand (OspD1), but the inhibitory mechanism has not been established (48). In contrast, transcriptional activation by the S. enterica regulator InvF is positively regulated by the SicA coactivator through a direct binding interaction (18). S. flexneri IpgC, which copurifies with MxiE, may also function as a coactivator (49). In summary, modulation of activator function by protein ligands can occur in a positive or negative fashion and may be a common theme for AraC family members that regulate T3SS gene expression.
ExsA-dependent promoters in P. aeruginosa consist of two adjacent binding sites for monomeric ExsA. Binding site 1 completely overlaps a putative 70 -RNA polymerase (RNAP) Ϫ35 recognition hexamer, while binding site 2 extends upstream and includes an adenine-rich region (12). ExsA binds via a monomer assembly pathway in which ExsA bound to site 1 recruits a second ExsA monomer to binding site 2 (12,14). Like most AraC family members, ExsA is dependent on 70 for transcriptional activation (64), and ExsA-dependent promoters contain apparent 70 -RNAP hexamers that closely resemble the P. aeruginosa consensus sequences (TTGACA and TATAAT for the Ϫ35 and Ϫ10 sites, respectively) (12,34). The placement of the Ϫ10 hexamers and transcription start sites has been established for several ExsA-dependent promoters by potassium permanganate footprinting experiments and 5Ј rapid amplification of cDNA ends (RACE) mapping, respectively (64,67). These experiments indicated that 70 -dependent transcription originates from the same start sites in the presence and absence of ExsA (64). The apparent Ϫ35 and Ϫ10 hexamers of ExsA-dependent promoters are spaced 4 to 5 bp farther apart than the 17 bp typical of most 70 -dependent promoters. Increased spacing has not been reported for any AraC family regulators but is seen for Spo0A, a transcriptional activator of the sporulation regulon in Bacillus subtilis (45). Spo0A activates promoters with extended spacing (20 to 22 bp) between near-consensus A -RNAP (the 70 homolog) Ϫ35 and Ϫ10 hexamers (8,39,57). The current model suggests that Spo0A activates transcription by repositioning RNAP prebound to the Ϫ35 site such that A region 2 can interact with the Ϫ10 hexamer to initiate open complex formation (39). Despite the apparent similarity to the extended spacing of Spo0A-dependent promoters, genetic and biochemical experiments suggest an entirely different mechanism for ExsA-dependent activation. A kinetic analysis of abortive transcript production from the P exsC and P exsD promoters reveals that the primary function of ExsA is to recruit 70 -RNAP to promoter DNA (64). Additionally, ExsA-dependent promoters in which the spacing between the Ϫ35 and Ϫ10 hexamers has been reduced to 16 or 17 bp are severely reduced in expression. These data suggest that ExsA, unlike Spo0A, does not function by compensating for increased promoter spacing and raises the question as to whether the Ϫ35-like elements of ExsA-dependent promoters represent actual 70 -RNAP contact points.
Transcriptional activators typically promote transcription through specific contacts with the ␣ and/or 70 subunit of RNAP (41). The specific RNAP contacts made by these proteins are thought to depend largely on the location of the activator promoter-binding site relative to the Ϫ35 hexamer. Class I promoters usually contain an activator DNA-binding site located Ն20 bp upstream of the Ϫ35 hexamer (22). The available data suggest that activation of a class I promoter is typically mediated by specific contacts between the activator protein and the carboxy-terminal domain of the RNAP ␣ subunit (␣-CTD) (22). In contrast, class II promoters usually contain an activator DNA-binding site positioned in close proximity to or overlapping the Ϫ35 hexamer. Activation of a class II promoter is thought to occur by contacts with the 70 subunit or both the 70 and ␣ subunits of RNAP (51,56). Activation by AraC family members often involves interactions with region 4.2 of 70 RNAP. This region contains a DNA-binding domain that recognizes the Ϫ35 hexamer. The carboxy-terminal end of region 4.2 also interacts with a diverse group of class II transcriptional activators (20). For example the AraC family regulators RhaR and RhaS, which are involved in the metabolism of rhamnose, contact several amino acids in region 4.2 of 70 , and this interaction is required for transcriptional activation (7,66).
In this study we characterized the interaction between ExsA and the 70 subunit. Our data indicate that ExsA functions as a class II transcriptional activator at the P exsC and P exsD promoters, does not require the ␣ subunit of RNAP, and instead contacts several amino acids in region 4.2 of 70 . We also provide evidence that the Ϫ35-like element of the P exsC promoter is not an authentic RNAP recognition hexamer for ExsA-independent or -dependent transcription and demonstrate that ExsA-independent transcription at the P exsC promoter requires an extended Ϫ10 promoter element.

MATERIALS AND METHODS
Bacterial strains and culture conditions. The bacterial strains and plasmids used in this study are summarized in Table 1. Escherichia coli strains were maintained on LB agar plates containing the following antibiotics/chemicals as necessary: gentamicin (15 g/ml), ampicillin (50 or 100 g/ml), tetracycline (10 g/ml), kanamycin (50 g/ml), and indole-3-acrylic acid (0.5 mM). P. aeruginosa strains were maintained on Vogel-Bonner minimal medium (65) with antibiotics as indicated (carbenicillin [300 g/ml] and tetracycline [50 g/ml]). For the LuxR experiments bacteria were grown in the presence of 200 nM N-(3-oxo-hexanoyl)-L homoserine lactone (Sigma, St. Louis, MO). To assay for ExsA-dependent gene expression in the presence of mutant and wild-type RNAP subunits, E. coli strains from LB agar plates grown overnight were inoculated into 10 ml of LB to an optical density at 600 nm (OD 600 ) of 0.1 and grown with vigorous aeration at 30°C until the OD 600 reached 0.6. ␤-Galactosidase assays were performed as previously described (19), and the reported values are the averages from at least three independent experiments.
Plasmid construction and promoter mutagenesis. Plasmid construction is summarized in Table 2, and the primers used are listed in Table 3.
The P exsC-lacZ , P exsD-lacZ , and P exoT-lacZ transcriptional reporters were generated by PCR amplification of the promoters and cloning into the KpnI/EcoRI (P exsC /P exsD ) or SalI/EcoRI (P exoT ) sites of the integration plasmid pAH125 (28). The P luxI-lacZ translational fusion reporter was generated by cloning the AatII/EcoRI restriction fragment from plasmid pluxI-lacZ (63) into plasmid pAH125. The resulting plasmids were integrated at the attachment site of E. coli strains GS162 and/or GA2071 by an electroporation method as described previously (28).
The carboxy-terminal hexahistidine-tagged ␣ subunit expression vector pET24-rpoA His was created by PCR amplifying the rpoA gene from P. aeruginosa strain PA103 lacking its native stop codon by using NdeI-NotI-containing primers and cloning the resulting fragment into pET-24a (Novagen). The P. aeruginosa ␤ and ␤Ј subunit expression vectors (pET24-rpoB and pET24-rpoC, respectively) were created by PCR amplification of rpoB or rpoC from P. aeruginosa strain PA103 by using NdeI-NotI containing primers and cloning the resulting fragment into pET-24a. The carboxy-terminal hexahistidine-tagged 70 expression vector (pET23-rpoD His ) was created by PCR amplification of the rpoD gene lacking its native stop codon from P. aeruginosa strain PA103 using primers incorporating NdeI-HindIII restriction sites and cloning the resulting fragment into pET-23b (Novagen). Point mutations in rpoD were introduced by QuikChange site-directed mutagenesis (Stratagene).
Purification of P. aeruginosa RNAP core enzyme, 70 and holoenzyme reconstitution. ExsA was purified as previously described under native conditions as an amino-terminal decahistidine-tagged fusion protein (12). Individual P. aeruginosa RNAP subunits were purified as described previously (60) with modifications. E. coli Tuner(DE3) carrying pET24-rpoA His was grown at 37°C in 50 ml of Luria broth containing 50 g/ml kanamycin to an OD 600 of 0.7, at which time IPTG (isopropyl-␤-D-thiogalactopyranoside) (1 mM) was added and the culture was incubated for an additional 3 h at 37°C. Bacteria were harvested by centrifugation and suspended in 4 ml of buffer A (20 mM Tris-HCl [pH 7.9], 500 mM NaCl, and 5 mM imidazole). Cells were lysed via sonication on ice, and unbroken cells were removed by centrifugation (15 min, 16,000 ϫ g, 4°C). Solid ammonium sulfate (60% of saturation) was added, and samples were allowed to precipitate for 15 min at 4°C with agitation. The precipitate was collected by centrifugation (20 min, 16,000 ϫ g, 4°C) and resuspended in 10 ml of buffer B (20 mM Tris-HCl [pH 7.9], 6 M guanidine HCl, and 500 mM NaCl) containing 5 mM imidazole. Prior to Ni 2ϩ affinity chromatography (see below), the material was subjected to ultracentrifugation (30 min, 100,000 ϫ g, 4°C) to remove particulates.
E. coli Tuner(DE3) carrying pET23-rpoD His was grown at 37°C in 200 ml LB containing 200 g/ml ampicillin to an OD 600 of 0.5, at which time IPTG (1 mM) was added and the culture was incubated for an additional 3 h at 37°C. Bacteria were harvested by centrifugation and suspended in 5 ml buffer B containing 5

3598
VAKULSKAS ET AL. J. BACTERIOL. mM imidazole. Cells were lysed via sonication on ice, and unbroken cells were removed by centrifugation (15 min, 38,000 ϫ g, 4°C). The ␣ and subunits were denatured and solubilized with guanidine as described above and purified under denaturing conditions by Ni 2ϩ affinity chromatography. Lysates were applied to a 1-ml HisTrap column (GE Healthcare) previously equilibrated with buffer B containing 5 mM imidazole, washed with 10 ml buffer B containing 30 mM imidazole, and developed with a 10-ml linear imidazole gradient (30 to 500 mM) in buffer B. The elution peaks were established by SDS-PAGE. The purified ␣ subunit was stored on ice for immediate use in core RNAP reconstitution. Purified 70 was dialyzed overnight against buffer E (50 mM Tris-HCl [pH 7.9], 200 mM KCl, 10 mM MgCl 2 , 10 M ZnCl 2 , 1 mM EDTA, 5 mM 2-mercaptoethanol, and 20% [vol/vol] glycerol) at 4°C, subjected to ultracentrifugation (30 min, 100,000 ϫ g, 4°C), and stored in 50% glycerol at Ϫ20°C.
The ␤ and ␤Ј RNAP subunits were purified from E. coli inclusion bodies. E. coli Tuner(DE3) carrying either pET24-rpoB or pET24-rpoC was grown at 37°C in 1 liter of LB containing 50 g/ml kanamycin to an OD 600 of 0.5, at which time IPTG (1 mM) was added and the culture was incubated for an additional 3 h at 37°C. Bacteria were harvested by centrifugation and suspended in 16 ml of buffer C (40 mM Tris-HCl [pH 7.9], 300 mM KCl, 10 mM EDTA, 1 mM dithiothreitol [DTT], and 1ϫ protease inhibitor cocktail [Roche]) containing 0.2 mg/ml lysozyme and 0.2% (wt/vol) sodium deoxycholate. The bacteria were incubated on ice for 20 min and lysed by sonication. Inclusion bodies were collected by centrifugation (30 min, 38,000 ϫ g, 4°C) and washed with 16 ml buffer C containing 0.2% n-octyl-␤-D-glucoside. Inclusion bodies were sonicated and centrifuged as described above, followed by a final wash with 16 ml buffer C. Washed inclusion bodies were solubilized in 2 ml of buffer D (50 mM Tris-HCl [pH 7.9], 6 M guanidine-HCl, 10 mM MgCl 2 , 10 M ZnCl 2 , 1 mM EDTA, 10 mM DTT, and 10% [vol/vol] glycerol) and incubated at 25°C for 10 min. The resulting material was subjected to ultracentrifugation (30 min, 100,000 ϫ g, 4°C), and the soluble fraction was stored on ice for immediate use in core enzyme reconstitution.
RNAP core enzyme was reconstituted by mixing 0.3 mg purified ␣ subunit, 1.5 mg purified ␤ subunit, and 3 mg ␤Ј subunit in buffer D (2 ml) and dialyzing twice against 500 ml of buffer E at 4°C with constant stirring. The resulting material was subjected to ultracentrifugation (30 min, 100,000 ϫ g, 4°C), and the soluble  In vitro transcription assays. Supercoiled transcription templates containing the P exsC and P exsD promoters were described previously (14,64). The pOM90-P exsC template was generated by PCR amplifying the P exsC promoter (nucleotides [nt] Ϫ207 to ϩ192 relative to the transcriptional start site) and cloning as an EcoRI fragment into pOM90 (54). The resulting template contains a fusion of the P exsC promoter to the rpoC transcriptional terminator (rpoC ter ) on pOM90 and directs synthesis of a 261-nt transcript. The pOM90-P trc180 and pOM90-P trc250 templates were generated by PCR amplifying the P trc promoter (nucleotides Ϫ61 to ϩ109/179 relative to the transcriptional start site) from pTRCHIS-b (Invitrogen) and cloning as an EcoRI fragment into pOM90. The pOM90-P trc180 and pOM90-P trc250 templates fuse the P trc promoter to rpoC ter , resulting in 180and 250-base transcripts, respectively. Finally, the pOM90-P RE# template was generated by annealing complementary oligonucleotides, and the resulting BamHI-EcoRI fragment was cloned into pOM90. The pOM90-P RE# template fuses the synthetic P RE# promoter to the rrnB T1 terminator and results in a 135-base transcript.
Single-round transcription assays (20-l final volume) were performed by incubating ExsA His (35 nM) with transcription templates (2 nM) at 25°C in 1ϫ transcription buffer (40 mM Tris-HCl [pH 7.5], 50 mM KCl, 10 mM MgCl 2 , 1 mM DTT, 0.1% Tween 20, and 0.5 mg/ml BSA) containing the initiating nucleotides ATP and GTP (0.75 mM). After 10 min, 25 nM reconstituted P. aeruginosa RNAP holoenzyme was added, and open complexes were allowed to form for 1 min at 25°C in the presence of ExsA His or for 20 min at 25°C in the absence of ExsA His . Elongation was allowed to proceed by the addition of the remaining nucleotides (0.25 mM ATP/GTP/CTP, 0.75 mM UTP, and 2.5 Ci [␣-32 P]CTP) in 1ϫ transcription buffer containing heparin (final concentration, 50 g/ml). Reactions were stopped after 5 min at 25°C by the addition of 20 l stop buffer (98% formamide, 20 mM EDTA, 0.05% bromophenol blue, and 0.05% xylene cyanol). Samples were heated at 95°C for 5 min and immediately incubated on ice before electrophoresis on 5% denaturing polyacrylamide gels.

RESULTS
The carboxy-terminal domain of the RNAP ␣ subunit is not required for ExsA-dependent transcriptional activation. Since ExsA activates transcription primarily through recruitment of RNAP (64) and many transcriptional activators that recruit do so by contacting the carboxy-terminal domain of the RNAP ␣ subunit (␣-CTD), we tested the hypothesis that ExsA uses a similar mechanism. Previous studies have shown that ExsA activates transcription in vitro to similar levels using RNAP from either P. aeruginosa or E. coli (64). To demonstrate that ExsA can activate transcription from the P exsC , P exsD , and P exoT promoters in E. coli, ExsA was expressed from a plasmid (p2UY21-exsA) under the transcriptional control of a constitutive ␣-CTD-independent promoter. ExsA-dependent transcription was measured from transcriptional reporters consisting of ExsA-dependent promoters (P exsC , P exsD , and P exoT ) fused to lacZ and integrated at the E. coli phage attachment site. Significant ExsA-dependent activation of all three promoters was observed relative to a control plasmid (Fig. 1A), demonstrating that ExsA is sufficient to activate transcription from P exsC , P exsD , and P exoT , as was previously shown for P exsC in E. coli (61).
To determine the role of the ␣-CTD, we used an established E. coli assay in which the native ␣ subunit (␣-wt) or ␣ lacking the C-terminal 239 amino acids (␣-⌬CTD) was expressed from a plasmid such that its cellular concentration exceeded that of native ␣ subunit expressed from the chromosome. This approach was necessary because deletion of the ␣ subunit CTD is lethal in E. coli (37). ExsA-dependent transcription following overexpression of ␣-⌬CTD was plotted as a percentage of the activation observed with overexpressed ␣-wt. As a control, we also measured LuxR-dependent activation of a P luxI-lacZ tran-scriptional fusion (1). LuxR is an activator known to require the ␣-CTD (59). ExsA-dependent activation of the P exsC-lacZ , P exsD-lacZ , and P exoT-lacZ reporters in the presence of ␣-⌬CTD was Ն100% of that seen with ␣-wt, indicating that ExsA does not require the ␣-CTD for transcriptional activation (Fig. 1B). Curiously, activation from the P exsC promoter in the presence of ␣-⌬CTD was 125% of the wild-type value, suggesting that the ␣-CTD might have an inhibitory function at this promoter.
In contrast, activation of the P luxI-lacZ reporter was reduced to ϳ33% of the wild-type value in the presence of ␣-⌬CTD. to test for activator-specific defects in gene expression (40). Those experiments were performed in E. coli strain GA2071 where expression of chromosomal rpoD is tightly repressed. To measure ExsA activity, the P exsC-lacZ and P exsD-lacZ transcrip-  1. The RNAP ␣-CTD is not required for ExsA-dependent activation of transcription. (A) E. coli strain GS162 carrying the indicated transcriptional reporters (P exsC-lacZ , P exsD-lacZ , or P exoT-lacZ ) was transformed with a vector control (pJN105) or a constitutive ExsA expression plasmid (p2UY21, labeled pExsA in the figure). The resulting strains were grown in LB to an OD 600 of 0.6 and assayed for ␤-galactosidase activity (reported in Miller units). (B) E. coli GS162 carrying a P luxI-lacZ reporter and a LuxR expression plasmid (p2UY21-luxR) and the reporter strains from panel A were transformed with a plasmid expressing the native ␣ or ␣⌬CTD subunit. The resulting strains were grown in LB to an OD 600 of 0.6 and assayed for ␤-galactosidase activity. The reporter activities obtained in cells expressing ␣⌬CTD were normalized to the same strain expressing native (WT) ␣ and reported as the percentage of native activity. The results represent the averages for three independent experiments, and error bars represent the standard errors of the means.

VOL. 192, 2010
ExsA CONTACTS SIGMA-70 REGION 4.2 3601 tional reporters were introduced at the phage attachment site of E. coli strain GA2071, and the resulting strains were transformed with a plasmid expressing wild-type RpoD or the RpoD point mutants. Given the tendency for reversion of rpoD point mutants to the native sequence (40), the integrity of each expression plasmid was verified by nucleotide sequence analysis after introduction into strain GA2071. ExsA was constitutively expressed from plasmid p2UY21. Since the RpoD mutants in this library do not affect activator-independent transcription, ExsA expression levels were similar in each of the tested strains ( Fig. 2A) (40). ExsA-dependent expression from the P exsC and P exsD reporters in the presence of RpoD mutants was plotted relative to that in the presence of wildtype RpoD ( Fig. 2B and C). The most drastic effect on ExsAdependent activation of the P exsC-lacZ reporter resulted with the K593A, R596A, and R599A substitutions, which exhibited 50%, 29%, and 25% of the activity seen with native RpoD, respectively (Fig. 2B). Similarly, P exsD-lacZ reporter activity was also impaired by the K593A, R596A, and R599A substitutions to 42%, 67%, and 36% of the native RpoD levels (Fig. 2C). The effects observed from the single amino acid substitutions were modest (2-to 3-fold) and likely reflect the fact that each of the individual positions represents only a portion of the ExsA-70 interaction site. We predicted that ExsA-dependent transcription might result from synergistic interactions with each of the three amino acid positions. This proved to be true, as the activity of the P exsC and P exsD reporters in the presence of a triple RpoD mutant (K593A, R596A, R599A) was only 15% of the wild-type activity in both cases ( Fig. 2B and C). We did note that strain GA2071 expressing the RpoD triple mutant exhibited a 2-fold growth defect yet had ExsA levels equivalent to those of GA2071 expressing native RpoD ( Fig. 2A).
ExsA-dependent transcription in vitro is dependent on P. aeruginosa 70

region 4.2.
To further characterize the role of 70 region 4.2, the mutations from E. coli rpoD (K593A, R596A, and R599A) that affect ExsA-dependent transcription in vivo were introduced into P. aeruginosa rpoD. Native and mutant forms of P. aeruginosa RpoD were expressed in E. coli and purified under denaturing conditions by Ni 2ϩ affinity chromatography. Core RNAP was generated by expressing the P. aeruginosa ␣, ␤, and ␤Ј subunits in E. coli, purifying the individual purified components (Fig. 3A), and reconstituting -saturated RNAP holoenzyme with either native RpoD, RpoD-K593A, RpoD-R596A, RpoD-R599A, or the triple RpoD mutant. RNAP holoenzyme activity was normalized between the different RpoD-reconstituted polymerases by comparing the production of single-round in vitro transcription products from the P trc promoter. Transcription from the P trc promoter was not affected by the K593A, K596A, or K599A mutation in region 4.2 of 70 (40).
Reconstituted RNAP holoenzymes were then assayed for ExsA-dependent transcription in vitro using supercoiled plasmid templates containing the P exsC and P exsD promoters fused to the rpoC ter terminator. Each of the templates generates a 261-nucleotide, terminated transcript. As expected, terminated transcripts were not observed with core RNAP alone. We initially tested the individual 70 mutants (K597A, R600A, and R603A) for ExsA-dependent activation of the P exsC or P exsD promoters but found that none had an activation defect greater than 50% of native RpoD (data not shown). This result was not surprising given that a similar observation was made when testing the individual 70 mutants for activation in E. coli ( Fig.  2B and C). In contrast, the triple RpoD mutant produced far less exsC and exsD transcripts than did native RpoD ( Fig. 3B and C). These combined data indicate that 70 region 4.2 is required for ExsA-dependent activation of the P exsC and P exsD promoters both in vivo and in vitro.
The near-consensus ؊35 sequence at the P exsC promoter is not required for ExsA-independent transcription. We previously demonstrated that the P exsC promoter has low basal activity in the absence of ExsA (64). To determine whether the putative Ϫ35 sequence is required for ExsA-independent promoter activity, we generated P exsC transcription templates containing point mutations at each of the Ϫ35 nucleotide positions. Each of the nucleotide substitutions, with the exception of G41T, was divergent from the 70 consensus (Fig. 4A). The mutant promoters were assayed for ExsA-independent transcript levels and compared to the native P exsC promoter and to a negative control containing a single point mutation (T8G) in the established Ϫ10 Pribnow box (Fig. 4A). To account for subtle differences in template concentration and purity, the P exsC transcripts were normalized to a constitutive transcript generated from a promoter located on the plasmid backbone (64). Whereas the negative control (T8G) lacking a functional Ϫ10 hexamer exhibited a 50-fold decrease in transcription compared to P exsC (Fig. 4), the remaining point mutants had little (less than 2-fold) or no effect on ExsA-independent transcription ( Fig. 4B and C). These data indicate that the putative Ϫ35 hexamer is not important for ExsA-independent transcription at the P exsC promoter. Although a near-consensus, but improperly spaced, Ϫ35 sequence is present at the P exsC promoter, it is possible that a weak, unrecognizable Ϫ35 hexamer with a poor match to the 70 consensus is present and optimally spaced (16/17 bp) from the Ϫ10 hexamer. Potential Ϫ35 hexamers spaced at either 16 or 17 bp would have the sequence AAAGCG or AAAAGC, respectively (matches to consensus are underlined). To test this hypothesis, we constructed a single point mutation in the P exsC promoter (A33G) such that the potential Ϫ35 hexamer spaced 16 bp (AAGGCG) from the Ϫ10 hexamer more closely resembles the Ϫ35 consensus sequence and the potential Ϫ35 hexamer spaced at 17 bp (AAAGGC) would be a weaker match to the consensus. The A33G mutation had no significant effect (Ͻ2-fold) compared to native P exsC . These combined data suggest that the putative Ϫ35 sequence is not important for ExsA-independent transcription.
The P exsC promoter sequence located immediately upstream of the Ϫ10 box resembles an extended Ϫ10 promoter (Fig.   FIG. 3. ExsA-dependent in vitro transcription is dependent on re- ExsA His (35 nM) was incubated with 2 nM supercoiled P exsC or P exsD promoter template (pOM90-P exsC or pOM90-P exsD ) at 25°C in the presence of rATP and rGTP. After 10 min, P. aeruginosa core RNAP, 70 -RNAP, or 70 (K597A/R596A/R599A)-RNAP was added (25 nM each; the activity of -saturated enzymes was normalized with P trc ), and the reaction mixture was incubated for 1 min at 25°C. Heparin and substrate nucleotides (including 2.5 Ci [␣-32 P]CTP) were immediately added, and the reaction mixture was incubated for 5 min at 25°C. Reactions were terminated, and the resulting products were electrophoresed on a 5% denaturing polyacrylamide-urea gel and subjected to phosphorimaging. The ExsA-dependent terminated transcripts (261 nt) from the P exsC or P exsD promoter and the runoff transcripts (250 or 180 nt) from the P trc promoter are indicated. 4A). Extended Ϫ10 promoters contain the sequence TGxTA TAAT and can function in either the presence or absence of a Ϫ35 hexamer (4,44). To determine whether the P exsC promoter contains an extended Ϫ10 element, we mutated the consensus TG sequence to AC (here referred to as P exsC-TG ). As expected, the mutant P exsC-TG promoter had a significant reduction in ExsA-independent transcription (5-fold) compared to the native P exsC promoter ( Fig. 4B and C). These combined data suggest that the P exsC promoter lacks a Ϫ35 hexamer and that an extended Ϫ10 element may provide basal promoter activity. The extended ؊10 element is important for ExsA-independent and -dependent P exsC promoter activity. Since ExsA-independent activity of the P exsC promoter requires an apparent extended Ϫ10 sequence, we asked whether ExsA-dependent activation had a similar requirement using in vitro transcription assays. P exsC-TG promoter activity was reduced 3-fold in the presence of ExsA, demonstrating that the extended Ϫ10 element affects P exsC to similar extents in the presence and absence of ExsA ( Fig. 5A and B). In contrast, the T8G mutation ablates both ExsA-dependent and -independent promoter activity. Note that ExsA-independent transcripts were not observed under these conditions due to the short RNAP incubation time (1 min) required to detect ExsA-dependent open complex formation in the linear range ( Fig. 5A and data not shown). To rule out the trivial explanation that the DNAbinding activity of ExsA is affected by the TG mutation, we employed electrophoretic mobility shift assays (EMSAs) and found no significant difference in the binding affinity of ExsA for the P exsC-TG and native P exsC promoters or in formation of shift complexes 1 and 2 (Fig. 5C).
Region 4.2 of 70 is required for ExsA-dependent but not ExsA-independent transcription. Region 4.2 of 70 recognizes the Ϫ35 hexamer and is essential for recognition of most bacterial promoters (15). Region 4.2 is also a common target for AraC family transcriptional activators. We have provided evidence that ExsA interacts with this region and that the putative Ϫ35 sequence is not a determinant for RNAP recruitment at the P exsC promoter. Based on these data, we hypothesized that the P exsC extended Ϫ10 element compensates for the lack of a functional Ϫ35 hexamer. To test this idea, we employed in vitro transcription assays utilizing RNAP holoenzyme reconstituted with 70 lacking the carboxy-terminal 43 amino acids, 70⌬4.2 , which encompasses region 4.2. Whereas deletion of region 4.2 renders promoters that are dependent upon Ϫ35 hexamers nonfunctional, the same deletion has little effect on transcription initiation and elongation from extended Ϫ10 promoters (38). The following promoters were used as controls for this experiment: (i) P trc , which contains a strong Ϫ35 hexamer and requires region 4.2 of 70 , and (ii) P RE# , a synthetic promoter which lacks a Ϫ35 hexamer and does not require region 4.2 but is dependent upon an extended Ϫ10 element (10, 38) (Fig.  6A). Although 70⌬4.2 has slightly reduced affinity for core RNAP enzyme (38), holoenzyme reconstituted with 70⌬4.2 (here referred to as RNAP-⌬4.2 ) and native RNAP holoenzyme generated similar levels of transcript from the P RE# promoter (Fig. 6B). In contrast, RNAP-70⌬4.2 generated significantly less transcript from the P trc promoter than did RNAP- 70 (Fig. 6B). Consistent with our hypothesis that the P exsC promoter lacks a functional Ϫ35 hexamer, RNAP-70⌬4.2 and RNAP-70 generated similar levels of P exsC transcript in the absence of ExsA (Fig. 6B). In addition, the P exsC-TG and P exsCT8G mutants were essentially devoid of RNAP-70⌬4.2dependent activity. Finally, we tested whether ExsA-dependent transcripts were produced from the P exsC promoter using RNAP-70⌬4.2 . Although ExsA-dependent transcription was drastically reduced with RNAP-70⌬4.2 , a detectable transcript was made when reactions were allowed to proceed for 1 min for open complex formation. These same conditions do not support the detection of transcription in the absence of ExsA using native RNAP-70 holoenzyme (Fig. 5A). It is unclear whether the weak ExsA-dependent transcription in the absence of region 4.2 represents additional contacts between ExsA and 70 outside region 4.2 or additional contacts between ExsA and other RNAP subunits.

DISCUSSION
In the present study we find that recruitment of RNAP by ExsA does not require the CTD of the RNAP ␣ subunit, a common target for AraC family regulators. Although these studies were performed with E. coli we believe the findings would be identical for P. aeruginosa. Data supporting this claim include the following: (i) ExsA activates transcription from T3SS promoters in vitro to similar extents with RNAP (normalized for specific activity using an ␣-CTD-independent promoter) from either E. coli or P. aeruginosa (64). (ii) the carboxy-terminal 90 amino acids of the ␣ subunits from E. coli and P. aeruginosa share 86% identity, and (iii) heterologous activators known to require the ␣-CTD, including LuxR from Vibrio fischeri (used in this study), can efficiently activate E. coli RNAP (59). For these reasons, we believe that the involvement of the ␣-CTD in ExsA-dependent activation would have been detected in our experiments.
Interestingly, ExsA-dependent transcription from the P exsC promoter was slightly elevated (125%) following expression of ␣-⌬CTD compared to the full-length ␣ subunit (Fig. 1B). A possible explanation for this finding is that the ␣-CTD may bind the P exsC promoter and antagonize ExsA function. In this scenario, the ␣-CTD-P exsC promoter interaction might sterically hinder the DNA-binding activity of ExsA or its ability to contact RNA polymerase. We did not test whether ExsA interacts with the amino-terminal domain (NTD), since Egan et al. have shown that an extremely diverse group of AraC family members do not require this domain for transcriptional activation (23).
Using a plasmid-based mutant rpoD expression library, we found that ExsA requires the K593, R596, and R599 amino acids of 70 for full activation of the P exsC and P exsD promoters (Fig. 2). These specific residues are some of the most frequently observed contact points for AraC family members and unrelated transcriptional regulators (20,40). In fact, an alignment of ExsA with the AraC family members RhaS and MelR reveals a conserved aspartate residue known to interact with R599 of 70 (27,66). Whether this aspartate or other conserved positions are important for the interaction of ExsA with 70 will be the subject of future studies. Although ExsA-dependent activation defects of greater than 2-fold were not routinely observed with a single point mutation in rpoD, expression of the chromosomal rpoD gene is only suppressed in these experiments, and leaky expression of rpoD may result in higher levels of ExsA-dependent activation that would bias the data toward transcriptional activation defects smaller than those observed. Furthermore, the literature suggests that RNAP-activator interaction regions most likely consist of several amino acid contacts (40). Consistent with this, we find that the 70 triple mutant (K593A, R596A, R599A) showed a cumulative 6-fold effect on ExsA-dependent transcription in vitro and in vivo ( Fig. 2 and 3). It is possible that ExsA may interact with amino acids in region 4.2 that we did not test, other regions in 70 , and/or different RNAP subunits. The 16 amino acids in the mutant rpoD expression library were selected because these positions are reported to have little effect on activator-independent transcription (40). Some amino acids in region 4.2 were omitted from this library because alanine substitution resulted in unstable protein or because they are required for interaction with the Ϫ35 hexamer (40). It is therefore possible that other amino acids are also important for the interaction with ExsA. Finally, the finding that a 70 derivative lacking region 4.2 ( ⌬4.2 ) is still capable of weak ExsA-dependent activation supports the hypothesis that ExsA interacts with several regions of 70 and/or multiple RNAP subunits (Fig. 6B).
Although the mechanism of transcription activation is known for only a small number of AraC family activators, most activate transcription by facilitating both closed and open complex formation (11). Activators that facilitate both closed and open complex formation do so by contacting the ␣-CTD and 70 region 4.2, respectively (11). It is therefore curious that ExsA requires region 4.2 of 70 and functions primarily to recruit RNAP. A possible explanation for this finding is that the ExsA-70 region 4.2 interaction affects the rate of isomerization to an open complex. This explanation seems unlikely, however, as disruption of the ExsA-70 region 4.2 interaction results in at least a 5-fold defect in activation, whereas ExsA is known to only marginally affect (2-fold) the rate of isomerization to an open complex. We believe a more likely explanation is that ExsA interaction with 70 region 4.2 results primarily in the recruitment of RNAP. This is in contrast to the reported activity of the well-characterized cI protein of phage lambda, which increases the isomerization rate at the P RM promoter by contacting 70 region 4.2 (21,30). The cI example is somewhat paradoxical, since it has been well established that in the absence of a transcriptional activator, 70 region 4.2 normally interacts directly with DNA at the Ϫ35 position to facilitate the initial binding of RNAP to the promoter (15). In fact, the observation that ExsA recruits RNAP through contacts with 70 region 4.2 seems to better support the known function of region 4.2. We believe the most likely explanation for these  70 and 70⌬4.2 reconstituted RNAP holoenzymes normalized for specific activity using the P RE# extended Ϫ10 promoter (lanes 3 and 4). Reactions were performed as described in the legend to Fig. 3, and open complexes were allowed to form for 1 min (lanes 1 to 4, 9, and 10) or 20 min (lanes 5 to 8) as indicated. VOL. 192, 2010 ExsA CONTACTS SIGMA-70 REGION 4.2 3605 discrepancies is that protein-protein interactions with 70 region 4.2 can affect both closed and open complex formation. In support of this claim, a single point mutation (R596H) in 70 region 4.2 changes the mechanism of cI activation to an enhancement of closed complex formation while having almost no effect on the rate of isomerization to an open complex (21). This finding may indicate that the specific contacts between transcriptional activators and 70 region 4.2 do not determine whether closed or open complex formation is enhanced. In fact, Dove et al. have suggested that the promoter sequence and location of the activator-binding site may play the most important part in determining the mechanism of transcriptional activation by an activator (21). Further studies analyzing the structure of activator-RNAP complexes are needed to address this curious observation. We provide evidence that the putative Ϫ35 hexamer in the P exsC promoter is not sufficient for ExsA-independent expression. This is consistent with a previous study demonstrating that the Ϫ35 hexamer from P exsD , although a close match to the 70 consensus, is also not used as an RNAP recognition site (12,64). To further characterize the role of the P exsC Ϫ35 region, we generated point mutations at every position in the Ϫ35 site, and the resulting mutations had no significant effect (Ͻ2-fold) on ExsA-independent transcription, while control mutations in the Ϫ10 hexamer resulted in undetectable levels of transcript (Fig. 4). An explanation for this result is that an authentic Ϫ35 hexamer is located at a more favorable position (16 or 17 bp relative to the Ϫ10 sequence) but has few matches to the consensus sequence. We tested this hypothesis by creating a single point mutation in P exsC (A33G), which should significantly increase or decrease ExsA-independent activation if the Ϫ35 hexamer is positioned 16 or 17 bp from the Ϫ10 hexamer, respectively (46). No significant effect was observed with this mutant, suggesting that a Ϫ35 hexamer is not required for ExsA-independent transcription at the P exsC promoter. Unfortunately, we were unable to assess the role of the Ϫ35 hexamer with respect to ExsA-dependent transcription, since mutations in the Ϫ35 region are known to disrupt ExsA binding to site 1 (12).
Consistent with the hypothesis that a Ϫ35 hexamer is not required for ExsA-independent transcription from the P exsC promoter, we identified a putative extended Ϫ10 element (38). A point mutation within this element resulted in a significant reduction in both ExsA-dependent and ExsA-independent transcription ( Fig. 4 and 5). EMSA experiments demonstrated that the extended Ϫ10 mutation had no effect on ExsA binding to the promoter (Fig. 5C). These data indicate that the P exsC promoter contains an extended Ϫ10 promoter that might partially compensate for the lack of a functional Ϫ35 hexamer. Since exsA expression is autoregulated through the P exsC promoter, it is tempting to speculate that the extended Ϫ10 element is important in maintaining a basal level of the exsCEBA transcript. The fact that the extended Ϫ10 element is required for maximal P exsC promoter activity, however, prevented us from directly testing this hypothesis. Nevertheless, 5Ј RACE promoter-mapping experiments demonstrate that exsCEBA transcript is detectable in an exsA mutant (64), suggesting that P exsC exhibits some level of basal activity. We propose a model in which ExsA recruits RNA polymerase to an extended Ϫ10 promoter (P exsC ) by contacting 70 region 4.2. Interestingly, the residual transcription from P exsC seen with 70⌬4.2 -RNAP was shown to be ExsA dependent (compare Fig. 5A and 6B), further suggesting that an additional region of 70 or perhaps an RNAP subunit other than ␣ and 70 may be involved in ExsAdependent transcriptional activation, as has been suggested for other AraC regulators (6,23,32,35).