Bacterial Two-Hybrid Analysis of Interactions between Region 4 of the ς⁷⁰ Subunit of RNA Polymerase and the Transcriptional Regulators Rsd from Escherichia coli and AlgQ from Pseudomonas aeruginosa

Abstract

A number of transcriptional regulators mediate their effects through direct contact with the ς⁷⁰ subunit of Escherichia coli RNA polymerase (RNAP). In particular, several regulators have been shown to contact a C-terminal portion of ς⁷⁰ that harbors conserved region 4. This region of ς contains a putative helix-turn-helix DNA-binding motif that contacts the −35 element of ς⁷⁰-dependent promoters directly. Here we report the use of a recently developed bacterial two-hybrid system to study the interaction between the putative anti-ς factor Rsd and the ς⁷⁰ subunit of E. coli RNAP. Using this system, we found that Rsd can interact with an 86-amino-acid C-terminal fragment of ς⁷⁰ and also that amino acid substitution R596H, within region 4 of ς⁷⁰, weakens this interaction. We demonstrated the specificity of this effect by showing that substitution R596H does not weaken the interaction between ς and two other regulators shown previously to contact region 4 of ς⁷⁰. We also demonstrated that AlgQ, a homolog of Rsd that positively regulates virulence gene expression in Pseudomonas aeruginosa, can contact the C-terminal region of the ς⁷⁰ subunit of RNAP from this organism. We found that amino acid substitution R600H in ς⁷⁰ from P. aeruginosa, corresponding to the R596H substitution in E. coli ς⁷⁰, specifically weakens the interaction between AlgQ and ς⁷⁰. Taken together, our findings suggest that Rsd and AlgQ contact similar surfaces of RNAP present in region 4 of ς⁷⁰ and probably regulate gene expression through this contact.

Sigma factors are subunits of bacterial RNA polymerase (RNAP) that direct the holoenzymes that contain them to promoters of a specific class (20). In Escherichia coli there are seven different species of ς factors, and ς⁷⁰ is the principal ς factor (27). The presence of different types of ς factors within a cell with distinct DNA sequence binding specificities provides a mechanism for coordinate regulation of genes that are controlled by promoters of the same class. Competition between different ς factors for the available RNAP core enzyme in part determines which genes are transcribed within a cell at any given time (27). This competition can be influenced by anti-ς factors, which are regulatory proteins that bind ς factors and often prevent their association with the RNAP core enzyme (19, 26). Anti-ς factors ultimately inhibit transcription from the class of promoters recognized by the ς factors that they sequester.

The ς⁷⁰ subunit of RNAP participates in a number of protein-protein interactions, including interactions with other subunits of the polymerase complex (43, 52) and interactions with transcriptional regulators (18, 23). The regulators that interact with ς⁷⁰ often contact a region of ς that contains a putative helix-turn-helix DNA-binding motif responsible for contacting the −35 elements of ς⁷⁰-dependent promoters (18, 23, 37). This DNA-binding region of ς is conserved in members of the ς⁷⁰ family of proteins and is called region 4 (36).

Recently, Jishage and Ishihama identified a protein in E. coli that was preferentially made by cells during the stationary phase of growth and was associated with the ς⁷⁰ subunit of RNAP in stationary-phase extracts (28). The protein was named Rsd (which stands for regulator of sigma D) since it was found to associate specifically with ς⁷⁰ (but not with several alternative ς factors) and was shown to be capable of inhibiting ς⁷⁰-dependent transcription from certain promoters in vitro (28). The binding site for Rsd on ς⁷⁰ was mapped to a C-terminal tryptic fragment encompassing conserved region 4 (28). On the basis of these observations and because the synthesis of Rsd coincides with the general shutdown in ς⁷⁰-dependent transcription that occurs as cells enter the stationary phase of growth, Jishage and Ishihama suggested that Rsd might be an anti-ς factor (28). Subsequent work has shown that consistent with this idea, Rsd may facilitate the replacement of ς⁷⁰ by the stationary-phase-specific ς factor ς³⁸ in functional RNAP holoenzyme complexes as cells go from the exponential phase to the stationary phase of growth (29).

The sequence of putative anti-ς factor Rsd is similar to the sequence of a regulator of alginate production in Pseudomonas aeruginosa called AlgQ (or AlgR2) (28). Alginate is an important virulence factor that imparts the characteristic mucoid phenotype to P. aeruginosa isolated from the lungs of cystic fibrosis patients (17). P. aeruginosa isolates from other sources are typically nonmucoid and do not exhibit activated expression of genes involved in alginate production (10). However, the production of alginate is believed to promote survival of P. aeruginosa in the special environment of the lungs of cystic fibrosis patients, contributing to resistance to both immune responses and antibiotics (10). AlgQ was originally identified as a positive transcriptional regulator of the key alginate biosynthetic gene algD (8, 32), which is expressed at high levels in mucoid cells. The regulation of the algD gene is complex and involves two different ς factors; transcription initiates from two superimposed promoters, one of which is recognized by RNAP containing ς^E (AlgU/AlgT) and the other of which is recognized by RNAP containing ς⁵⁴ (RpoN) (3, 9, 11, 21, 38, 49, 59). Furthermore, at least one DNA-binding protein, AlgR (AlgR1), is known to bind to specific sites upstream of the algD promoter and activate transcription (31, 41, 42). The mechanism by which AlgQ positively regulates transcription of the algD gene is not known.

We were interested in testing the idea that like Rsd, AlgQ interacts with region 4 of the ς⁷⁰ subunit of RNAP. In this study we tested this idea explicitly by using a bacterial two-hybrid system. We found that both Rsd and AlgQ can interact with a ς⁷⁰ moiety encompassing region 4 in vivo, and we identified an amino acid substitution in region 4 that specifically weakens the interaction of Rsd and AlgQ with the ς moiety.

MATERIALS AND METHODS

Plasmids and strains.

E. coli XL1-blue (Stratagene) was used as the recipient strain for all plasmid constructions. E. coli KS1 harbors on its chromosome the lac promoter derivative placO_R2-62 driving expression of a linked lacZ reporter gene and has been described previously (14). E. coli SF1 harbors an F′ episome containing the lac promoter derivative plac O_R2-55/Cons −35 driving expression of a linked lacZ reporter gene and has also been described previously (13).

Plasmids pACλcI, pACΔcI, pBRα, pBRα-ς³⁸, pBRα-ς⁷⁰, and pBRα-ς⁷⁰ (R596H) have all been described previously (13, 14). Plasmid pACλcI-Rsd encodes λcI (residues 1 to 236) fused to Rsd (residues 1 to 158) via three alanine residues. pACλcI-Rsd was made by cloning the appropriate NotI-BamHI-digested PCR product into NotI-BstYI-digested pACλcI32 (24); expression of the cI-rsd fusion gene was therefore under the control of the lacUV5 promoter. Plasmid pACλcI-AlgQ encodes λcI (residues 1 to 236) fused to AlgQ (residues 1 to 160) via three alanine residues, and it was made in a manner similar to the manner in which pACλcI-Rsd was made. Expression of the cI-algQ fusion gene on pACλcI-AlgQ is also under the control of the lacUV5 promoter. Plasmid pACλcI-AsiA encodes λcI (residues 1 to 236) fused to AsiA from bacteriophage T4 (residues 1 to 90) via three alanine residues, and it was made in a manner similar to the manner in which pACλcI-Rsd was made. Plasmid pACΔ-35λcI-AsiA contains the cI-asiA fusion gene from plasmid pACλcI-AsiA under the control of a lacUV5 promoter variant in which the −35 element of the promoter has been deleted. Plasmid pACΔ-35λcI-AsiA, therefore, expresses less of the λcI-AsiA fusion protein than plasmid pACλcI-AsiA expresses under identical conditions. pACΔ-35λcI-AsiA was made by cloning the appropriate HindIII-BstYI fragment from pACλcI-AsiA into plasmid pA3B2 (57) cut with both HindIII and BstYI. Plasmid pBRα-ς⁷⁰PA encodes residues 1 to 248 of the α subunit of E. coli RNAP fused to residues 532 to 617 of the ς⁷⁰ subunit of P. aeruginosa RNAP. The hybrid α-ς⁷⁰PA gene was made by performing PCR and was cloned into HindIII-BamHI-digested pBRα. Expression of the chimeric gene on pBRα-ς⁷⁰PA is, therefore, under the control of tandem lpp and lacUV5 promoters. pBRα-ς⁷⁰PA (R600H) is a derivative of pBRα-ς⁷⁰PA in which the R600H substitution in the ς moiety of the chimera was introduced by PCR.

The lac promoter derivative placCOP-93+O_L2-62 was made by the PCR and contains two λ operators; a near-consensus λ operator (TACCACCGGCGGTGATA) and O_L2 (CAACACCGCCAGAGATA) are centered 93 and 62 bp, respectively, upstream of the transcriptional start site of the lac core promoter. Plasmid pFW11-COP-93+O_L2-62 was constructed by cloning an EcoRI-HindIII-cut PCR product containing placCOP-93+O_L2-62 into pFW11 (56) cut with EcoRI and HindIII. Plasmid pFW11-COP-93+O_L2-62 was then transformed into strain CSH100, and the promoter-lacZ fusion was recombined onto an F′ episome and mated into strain FW102 (56) to create reporter strain F′93+62. The PCR-amplified regions of all plasmids were sequenced to confirm that no errors had been introduced as a result of the PCR process.

Experimental procedures.

Cells were grown in LB supplemented with kanamycin (50 μg/ml), chloramphenicol (25 μg/ml), carbenicillin (50 μg/ml), and isopropyl-β-d-thiogalactoside (IPTG) at the concentration indicated. Cells were permeabilized with sodium dodecyl sulfate-CHCl₃ and assayed for β-galactosidase activity essentially as described previously (39). Assays were performed at least three times in duplicate on separate occasions, and representative data sets are shown below. The values are averages based on one experiment; duplicate measurements differed by less than 10%.

RESULTS

λcI-Rsd activates transcription from a test promoter in the presence of a chimeric α-subunit harboring region 4 of E. coli ς⁷⁰.

We sought to detect an interaction between Rsd and region 4 of ς⁷⁰ in vivo by using a bacterial two-hybrid system that we had recently developed (12, 14). This bacterial two-hybrid system is based on the finding that any sufficiently strong interaction between a protein bound upstream of a suitable test promoter and a component of RNAP can activate transcription in E. coli (12, 14). Thus, two proteins that interact with one another can mediate transcriptional activation in E. coli provided that one protein is fused to a DNA-binding protein and the other is fused to a component of RNAP (12, 14).

Our strategy for detecting an interaction between Rsd and region 4 of ς⁷⁰ involved the use of two chimeric proteins, one comprising Rsd fused to the repressor of bacteriophage λ (λcI) and the other comprising a modified form of the α-subunit of RNAP in which the C-terminal domain (CTD) of α has been replaced by a C-terminal fragment of ς⁷⁰. We reasoned that RNAP containing the resulting α-ς⁷⁰ chimera would display a target for Rsd that could be contacted by a DNA-bound λcI-Rsd dimer (Fig. 1A). Having fused the entire Rsd protein (residues 1 to 158) to the C terminus of λcI, we placed the gene encoding this chimeric protein on a plasmid vector downstream of the IPTG-inducible lacUV5 promoter, thus creating plasmid pACλcI-Rsd. We used plasmid pBRα-ς⁷⁰ as a source of the α-ς⁷⁰ chimera. This plasmid encodes a chimera in which residues 528 to 613 of E. coli ς⁷⁰ are fused to residues 1 to 248 of the α-subunit of RNAP (13). We introduced plasmids pACλcI-Rsd and pBRα-ς⁷⁰ into E. coli KS1 (14), which harbors on its chromosome the lac promoter derivative placO_R2-62 (bearing a single λ operator centered 62 bp upstream of the transcriptional start site) linked to a lacZ reporter gene. We then tested the ability of the λcI-Rsd chimera to activate transcription from the placO_R2-62 test promoter in the presence of the α-ς⁷⁰ chimera. Figure 1B shows that λcI-Rsd activated transcription from the test promoter up to ∼24-fold in cells containing the α-ς⁷⁰ chimera compared to control cells containing only wild-type α. An additional control revealed that λcI (lacking the fused Rsd moiety) did not activate transcription from the test promoter in the presence of the α-ς⁷⁰ chimera (Fig. 1B). We also found that λcI-Rsd did not activate transcription from the test promoter in the presence of an α-ς³⁸ chimera (comprising residues 1 to 248 of α fused to residues 243 to 330 of ς³⁸) encoded by plasmid pBRα-ς³⁸ (13; data not shown).

FIG. 1.

Transcriptional activation by λcI-Rsd in the presence of the α-ς⁷⁰ chimera. (A) Replacement of the RNAP α-CTD by a C-terminal fragment of E. coli ς⁷⁰ (residues 528 to 613) permits interaction with the Rsd moiety of a λcI-Rsd chimera bound to DNA. The diagram depicts the test promoter placO_R2-62, which bears the λ operator O_R2 centered 62 bp upstream from the transcriptional start site of the lac core promoter. In reporter strain KS1 the placO_R2-62 test promoter is located on the chromosome and drives expression of a linked lacZ gene. The N-terminal domain of α is designated αNTD. (B) Effect of λcI-Rsd on transcription in vivo from placO_R2-62 in the presence of the α-ς⁷⁰ or α-ς⁷⁰(R596H) chimera. KS1 cells harboring compatible plasmids expressing the indicated proteins were grown in the presence of different concentrations of IPTG and assayed for β-galactosidase activity.

Substitution R596H in the ς moiety of the α-ς⁷⁰ chimera weakens the interaction between ς and Rsd

Our ability to link the protein-protein interaction between Rsd and its target on ς⁷⁰ to transcriptional activation provided us with a useful genetic tool for dissecting this interaction. We were particularly interested in identifying mutant forms of ς⁷⁰ that were specifically defective in the ability to interact with Rsd. We therefore introduced a variety of amino acid substitutions into the ς moiety of the α-ς⁷⁰ chimera and tested the effects of these substitutions on the ability of the λcI-Rsd fusion protein to mediate transcriptional activation from the test promoter. Figure 1B shows that substitution R596H in the ς moiety of the α-ς⁷⁰ chimera strongly reduced the magnitude of λcI-Rsd-dependent activation (to a factor of ∼5). Substitution R596A in the ς moiety of the α-ς⁷⁰ chimera had a nearly identical effect on the magnitude of the activation (data not shown). In contrast, substitutions L573A, E591A, E591Q, H600A, and H600R did not decrease the magnitude of λcI-Rsd-dependent activation (data not shown).

Substitution R596H in the ς moiety of the α-ς⁷⁰ chimera does not compromise the ability of a superactivating variant of λcI to stimulate transcription from an appropriate test promoter.

To determine whether the R596H substitution affects the interaction of ς⁷⁰ with Rsd specifically, we tested its effect on the ability of another protein to interact with the ς moiety of the chimera. We showed recently that the λcI protein (a transcriptional activator as well as a repressor) can interact specifically with the ς moiety of the α-ς⁷⁰ chimera and stabilize its binding to a promoter −35 element (13). The in vivo assay which we designed to detect the interaction between λcI and region 4 of ς⁷⁰ is shown in Fig. 2A. In this experimental setup a DNA-bound λcI dimer activates transcription from test promoter placO_R2-55/Cons-35 by stabilizing the binding of the ς moiety of the α-ς⁷⁰ chimera to the ectopic −35 element present upstream of the core promoter elements (13) (Fig. 2A). Transcriptional activation from this test promoter is dependent not only on the protein-protein interaction between λcI and the tethered ς moiety but also on the protein-DNA interaction between the tethered ς moiety and the ectopic −35 element (13).

FIG. 2.

Transcriptional activation by λcI superactivator in the presence of the α-ς⁷⁰ chimera. (A) λcISa109 stabilizes the binding of region 4 of ς⁷⁰ to an ectopic −35 element. The diagram depicts the test promoter placO_R2-55/Cons-35, which bears an ectopic −35 element and the λ operator O_R2 centered 45.5 and 55 bp, respectively, upstream from the transcriptional start site of the lac core promoter. In reporter strain SF1 the placO_R2-55/Cons-35 test promoter is located on an F′ episome and drives expression of a linked lacZ gene. (B) Effect of substitution R596H in the ς moiety of the α-ς⁷⁰ chimera on the ability of λcISa109 to activate transcription in vivo from placO_R2-55/Cons-35. SF1 cells harboring compatible plasmids expressing the indicated proteins were grown in the presence of different concentrations of IPTG and assayed for β-galactosidase activity.

We tested whether the R596H substitution in the α-ς⁷⁰ chimera has an effect on the ability of a superactiviating variant of λcI to activate transcription from test promoter placO_R2-55/Cons-35 (Fig. 2A). Reporter strain SF1 carries promoter placO_R2-55/Cons-35 fused to the lacZ gene in single copy on an F′ episome (13). We assayed the ability of λcI superactivator 109 (λcISa109) (5, 13) to activate transcription from this reporter in the presence of the α-ς⁷⁰ chimera with or without the R596H substitution in the ς moiety. λcISa109 activated transcription a maximum of ∼5.5-fold from the test promoter in the presence of the α-ς⁷⁰ chimera and a maximum of ∼7.5-fold in the presence of the chimera harboring the R596H substitution (Fig. 2B). (Although the R596H substitution has previously been shown to inhibit the ability of wild-type λcI to activate transcription from P_RM [35], we have observed that this substitution does not inhibit the ability of λcISa109 to activate transcription from P_RM [see below].) We concluded that substitution R596H in the ς moiety of the α-ς⁷⁰ chimera does not result in a general defect in the ability of the ς moiety to interact with other proteins.

We also tested the effect of the R596A substitution in the ς moiety of the α-ς⁷⁰ chimera on the ability of λcISa109 to activate transcription from the same test promoter. Unlike the R596H substitution, the R596A substitution significantly reduced the magnitude of the activation by λcISa109 (data not shown). Using another assay, however, we were able to determine that this reduction in activation was not due to a general defect caused by the R596A substitution (see below).

λcI-AlgQ activates transcription from a test promoter in the presence of a chimeric α-subunit harboring region 4 of P. aeruginosa ς⁷⁰.

Jishage and Ishihama (28) noted that Rsd exhibits 31% identity with the alginate regulatory protein AlgQ (also known as AlgR2) from P. aeruginosa. AlgQ was originally identified as a positive regulator of alginate production in P. aeruginosa (8, 32) but has subsequently been shown to regulate several other gene products in this organism (see below).

Very little is known about how AlgQ mediates its effects on gene expression. In order to test the hypothesis that AlgQ, like Rsd, can interact with region 4 of ς⁷⁰, we made a cI-algQ fusion gene analogous to the cI-rsd fusion gene. We also made a fusion gene encoding a protein analogous to the α-ς⁷⁰ chimera that contained region 4 of ς⁷⁰ from P. aeruginosa. We fused the entire AlgQ protein (residues 1 to 160) to the C terminus of λcI. We placed the gene encoding this chimeric protein on a plasmid vector downstream of the IPTG-inducible lacUV5 promoter, creating plasmid pACλcI-AlgQ. We then fused residues 532 to 617 of P. aeruginosa ς⁷⁰ (equivalent to residues 528 to 613 of E. coli ς⁷⁰) to residues 1 to 248 of α. The resulting chimera was called α-ς⁷⁰PA. The hybrid α-ς⁷⁰PA gene encoding this chimera was placed on a plasmid vector downstream of tandemly arranged lpp and lacUV5 promoters, creating plasmid pBRα-ς⁷⁰PA. We introduced plasmids pACλcI-AlgQ and pBRα-ς⁷⁰PA into E. coli KS1. We then tested the ability of the λcI-AlgQ chimera to activate transcription from placO_R2-62 in the presence of the α-ς⁷⁰PA chimera (Fig. 3A). Figure 3B shows that λcI-AlgQ activated transcription from the test promoter a maximum of ∼17-fold in cells containing the α-ς⁷⁰PA chimera compared to control cells containing only wild-type α. An additional control revealed that λcI without the fused AlgQ moiety did not activate transcription from the test promoter in the presence of the α-ς⁷⁰PA chimera (Fig. 3B).

FIG. 3.

Transcriptional activation by λcI-AlgQ in the presence of the α-ς⁷⁰PA chimera. (A) Replacement of the RNAP α-CTD by a C-terminal fragment of P. aeruginosa ς⁷⁰ (residues 532 to 617) permits interaction with the AlgQ moiety of a λcI-AlgQ chimera bound to DNA. The diagram depicts the test promoter placO_R2-62, which bears the λ operator O_R2 centered 62 bp upstream from the transcriptional start site of the lac core promoter. In reporter strain KS1 the placO_R2-62 test promoter is located on the chromosome and drives expression of a linked lacZ gene. (B) Effect of λcI-AlgQ on transcription in vivo from placO_R2-62 in the presence of the α-ς⁷⁰PA or α-ς⁷⁰PA(R600H) chimera. KS1 cells harboring compatible plasmids expressing the indicated proteins were grown in the presence of different concentrations of IPTG and assayed for β-galactosidase activity.

Substitution R600H in the ς moiety of the α-ς⁷⁰PA chimera weakens the interaction between ς and AlgQ.

The ς⁷⁰ subunits from E. coli and P. aeruginosa are very similar to one another, exhibiting ∼83% identity over the length of the ς fragments (86 amino acids) that we used in our experiments (36). We wanted to test whether substitution R600H in ς⁷⁰ from P. aeruginosa, which corresponds to the R596H substitution in E. coli ς⁷⁰, had any effect on the ability of λcI-AlgQ to activate transcription in the presence of the α-ς⁷⁰PA chimera. To do this, we made a version of the α-ς⁷⁰PA chimera harboring substitution R600H [α-ς⁷⁰PA(R600H)] and assayed the ability of λcI-AlgQ to activate transcription in KS1 cells expressing this chimera. Figure 3B shows that λcI-AlgQ activated transcription from the reporter gene a maximum of ∼5-fold in the presence of the α-ς⁷⁰PA(R600H) chimera, compared to a maximum of ∼17-fold with the chimera derived from the wild-type form of P. aeruginosa ς⁷⁰.

Substitution R600H in the ς moiety of the α-ς⁷⁰PA chimera does not compromise the ability of a superactivating variant of λcI to stimulate transcription from an appropriate test promoter.

In order to assess whether the effect of the R600H substitution was specific for the interaction between ς⁷⁰PA and AlgQ, we first tested whether λcISa109 could interact with the ς moiety of the α-ς⁷⁰PA chimera. We found that λcISa109 stimulated transcription from test promoter placO_R2-55/Cons-35 up to ∼3.5-fold specifically in the presence of the α-ς⁷⁰PA chimera (Fig. 4). Furthermore, introduction of the R600H substitution into the ς moiety of the α-ς⁷⁰PA chimera did not abrogate the stimulatory effect of λcISa109 (instead it resulted in a modest increase in the observed activation), suggesting that the effect of the R600H substitution is specific for the λcI-AlgQ chimera (Fig. 4B).

FIG. 4.

Transcriptional activation by λcI superactivator in the presence of the α-ς⁷⁰PA chimera. (A) λcISa109 stabilizes the binding of region 4 of ς⁷⁰ from P. aeruginosa to an ectopic −35 element. The diagram depicts the test promoter placO_R2-55/Cons-35, which bears an ectopic −35 element and the λ operator O_R2 centered 45.5 and 55 bp, respectively, upstream from the transcriptional start site of the lac core promoter. In reporter strain SF1 the placO_R2-55/Cons-35 test promoter is located on an F′ episome and drives expression of a linked lacZ gene. (B) Effect of substitution R600H in the ς moiety of the α-ς⁷⁰PA chimera on the ability of λcISa109 to activate transcription in vivo from placO_R2-55/Cons-35. SF1 cells harboring compatible plasmids expressing the indicated proteins were grown in the presence of different concentrations of IPTG and assayed for β-galactosidase activity.

The E. coli protein Rsd can interact with region 4 of ς⁷⁰ from P. aeruginosa, and the P. aeruginosa protein AlgQ can interact with region 4 of ς⁷⁰ from E. coli.

Given the high degree of similarity between the regions of ς⁷⁰ from E. coli and P. aeruginosa used in our experiments, we thought that each regulator might be able to contact region 4 of ς⁷⁰ from either E. coli or P. aeruginosa. We explicitly tested whether Rsd could interact with region 4 of ς⁷⁰ from P. aeruginosa and also whether AlgQ could interact with region 4 of ς⁷⁰ from E. coli. Figure 5A shows that the λcI-Rsd chimera activated transcription from the test promoter in KS1 cells a maximum of ∼52-fold in the presence of the α-ς⁷⁰PA chimera, compared to ∼24-fold in the presence of the E. coli α-ς⁷⁰ chimera. We also found that introduction of substitution R600H into the α-ς⁷⁰PA chimera reduced the ability of λcI-Rsd to stimulate transcription from the test promoter to a factor of ∼13 (Fig. 5A).

FIG. 5.

Rsd can interact with region 4 of ς⁷⁰ from P. aeruginosa, and AlgQ can interact with region 4 of ς⁷⁰ from E. coli. (A) Effect of λcI-Rsd on transcription in vivo from placO_R2-62 in the presence of the α-ς⁷⁰PA, α-ς⁷⁰PA(R600H), or α-ς⁷⁰ chimera. KS1 cells harboring compatible plasmids expressing the indicated proteins were grown in the presence of different concentrations of IPTG and assayed for β-galactosidase activity. (B) Effect of λcI-AlgQ on transcription in vivo from placO_R2-62 in the presence of the α-ς⁷⁰, α-ς⁷⁰(R596H), or α-ς⁷⁰PA chimera. KS1 cells harboring compatible plasmids expressing the indicated proteins were grown in the presence of different concentrations of IPTG and assayed for β-galactosidase activity.

Figure 5B shows that the λcI-AlgQ chimera activated transcription from the test promoter in KS1 cells a maximum of ∼7-fold in the presence of the α-ς⁷⁰ chimera, compared to ∼15-fold in the presence of the P. aeruginosa α-ς⁷⁰PA chimera. Introduction of substitution R596H into the ς moiety of the E. coli α-ς⁷⁰ chimera reduced the ability of λcI-AlgQ to activate transcription from the test promoter to a factor of ∼2 (Fig. 5B).

λcI-AsiA activates transcription from a test promoter in the presence of either α-ς chimera.

We took advantage of another protein known to interact with region 4 of E. coli ς⁷⁰ to test further the specificity of the effect of the R596H substitution on the interaction of ς⁷⁰ with either Rsd or AlgQ. The AsiA protein encoded by bacteriophage T4 is an anti-ς factor that has been shown to interact specifically and very strongly with region 4 of E. coli ς⁷⁰ (1, 7, 22, 40, 50, 51). AsiA is thought to work by interacting with ς⁷⁰ that is free in solution rather than with ς⁷⁰ that is already complexed with the RNAP core enzyme (22). The interaction between AsiA and E. coli ς⁷⁰ inhibits the activity of RNAP holoenzyme (containing the AsiA-ς⁷⁰ complex) by preventing region 4 of ς from contacting the −35 element of ς⁷⁰-dependent promoters (7, 51).

We therefore sought to detect the interaction between AsiA and region 4 of ς⁷⁰ by fusing AsiA to λcI and measuring the ability of the resulting λcI-AsiA chimera to activate transcription from placO_R2-62 in the presence of either the α-ς⁷⁰ chimera or the α-ς⁷⁰PA chimera. We fused the entire AsiA protein (residues 1 to 90) to the C terminus of λcI. We placed the gene encoding this chimeric protein on a plasmid vector downstream of the IPTG-inducible lacUV5 promoter, creating plasmid pACλcI-AsiA. Initial experiments demonstrated that the λcI-AsiA chimera was extremely toxic to cells at the levels provided by the expression vector pACλcI-AsiA (data not shown). For subsequent experiments we constructed a different expression vector, pACΔ-35λcI-AsiA, in which the chimeric cI-asiA gene was under control of an IPTG-inducible variant of the lacUV5 promoter that lacked the normal −35 element and therefore directed the synthesis of smaller amounts of the fusion protein. Since pACΔ-35λcI-AsiA made insufficient levels of λcI-AsiA to saturate the λ operator present on the placO_R2-62 promoter construct, we also constructed a new test promoter bearing two relatively strong λ operators in the upstream region (Fig. 6A; also see Materials and Methods). Figure 6B shows that the λcI-AsiA chimera activated transcription from the modified test promoter in the presence of the α-ς⁷⁰ chimera derived from ς⁷⁰ of either E. coli or P. aeruginosa (similar findings with the α-ς⁷⁰ chimera from E. coli have been obtained by S. Pande and D. Hinton [submitted for publication]). The data also show that substitution R596H in the E. coli ς moiety or the corresponding R600H substitution in the P. aeruginosa ς moiety had only a modest effect on the ability of the λcI-AsiA chimera to activate transcription from placCOP-93+O_L2-62. Similarly, the R596A substitution in the E. coli ς moiety did not reduce the stimulatory effect of the λcI-AsiA chimera (data not shown). Since these substitutions do not appear to cause nonspecific defects in the abilities of the ς moieties to interact with other proteins, we suggest that residue R596 of E. coli ς⁷⁰ is at or near the contact surface for Rsd and the equivalent residue of P. aeruginosa ς⁷⁰, R600, is at or near the contact surface for AlgQ (see Discussion). Furthermore, the finding that substitution R596A in E. coli ς⁷⁰ results in a strong defect in the ability of ς to interact with Rsd suggests that the arginine side chain may make an energetically significant contact with Rsd.

FIG. 6.

Transcriptional activation by λcI-AsiA in the presence of the α-ς chimeras. (A) Schematic diagram of test promoter placCOP-93+O_L2-62 used to determine the effects of substitutions in the α-ς chimeras on transcriptional activation by λcI-AsiA. The test promoter placCOP-93+O_L2-62 bears both a consensus λ operator and O_L2 centered 93 and 62 bp, respectively, upstream from the transcriptional start site of the lac core promoter. In reporter strain F′93+62 the placCOP-93+O_L2-62 test promoter is located on an F′ episome and drives expression of a linked lacZ gene. (B) Effect of λcI-AsiA on transcription in vivo from placCOP-93+O_L2-62 in the presence of the α-ς chimeras. F′93+62 cells harboring compatible plasmids expressing the indicated proteins were grown in the presence of different concentrations of IPTG and assayed for β-galactosidase activity. ΔcI refers to the fact that no λcI was made by control plasmid pACΔcI.

DISCUSSION

Rsd and AlgQ can interact with region 4 of ς⁷⁰ in vivo.

We found that the E. coli Rsd protein and the P. aeruginosa AlgQ protein can interact in vivo with a C-terminal fragment of the ς⁷⁰ subunit of E. coli and P. aeruginosa RNAP, respectively. This fragment of ς⁷⁰ encompasses conserved region 4, which contains a DNA-binding domain that mediates recognition of the −35 element of ς⁷⁰-dependent promoters (18). Furthermore, each protein can also interact with the corresponding ς⁷⁰ fragment from the heterologous organism. We also identified a single amino acid substitution (R596H and the corresponding substitution R600H in E. coli and P. aeruginosa ς⁷⁰, respectively) that inhibits the binding of both Rsd and AlgQ to either ς⁷⁰ fragment. The interchangeability of the E. coli and P. aeruginosa ς⁷⁰ fragments in our experiments suggests that regulators from P. aeruginosa or E. coli that interact with region 4 of ς⁷⁰ might in general be expected to function in either organism. In fact, a previous in vitro study showed that the λcI protein (which contacts region 4) can activate transcription from the λ promoter P_RM by the P. aeruginosa RNAP (16).

Analysis of the interaction between Rsd and region 4 of ς⁷⁰.

Our bacterial two-hybrid results for Rsd and region 4 of ς⁷⁰ are consistent with the biochemical data of Jishage and Ishihama (28). These authors found that Rsd could interact with ς⁷⁰ in vitro but not with ς³⁸ (or several other alternative ς factors), and they showed that Rsd could interact with a tryptic fragment of ς⁷⁰ composed of residues ∼500 to 613. We found that Rsd can interact in vivo with an 86-amino-acid C-terminal fragment of ς⁷⁰ that contains region 4 but cannot interact with the equivalent 88-amino-acid C-terminal fragment of ς³⁸. Jishage et al. (30) have recently reported that alanine substitutions at two positions flanking residue 596 (L595 and L598) disrupt the association of Rsd with ς⁷⁰ in vitro. However, in contrast with our results, they reported that substitution R596A in ς⁷⁰ did not prevent association of Rsd with ς⁷⁰ (in a glutathione S-transferase pulldown assay). To better compare the results of Jishage et al. with our results, we introduced substitutions L595A and L598A into the ς moiety of the α-ς⁷⁰ chimera and tested their effects on the interaction with Rsd in vivo. We found that both the L595A substitution and the L598A substitution strongly inhibited the interactions with Rsd and that their effects were more severe than that of the R596A substitution (data not shown). We suggest, therefore, that the in vivo assay may be more sensitive than the in vitro assay, allowing us to detect the less severe effect of the R596A substitution. Moreover, a structural model of region 4 of ς⁷⁰ based on the crystal structure of the NarL protein (see below) suggests that the side chain of residue 595, at least, is likely to be partially buried. Substitutions at position 595 and possibly also at position 598 may affect the interaction with Rsd indirectly; consistent with this possibility, we found that substitutions L595A and L598A both completely eliminated the ability of λcISa109 to activate transcription from our artificial test promoter in the presence of the α-ς⁷⁰ chimera (data not shown).

Amino acid substitution R596H has been isolated previously. It was first identified based on its effect on expression of the arabinose operon (25, 53). Expression of the ara genes is positively controlled by two regulators, the global regulator cyclic AMP receptor protein (CRP) and the operon-specific regulator AraC (15). CRP is active only when it is complexed with the small molecule effector cyclic AMP, and consequently mutant strains that are not able to synthesize cyclic AMP (cya mutants) cannot induce expression of CRP-dependent operons, including the arabinose operon. A mutation in the ς⁷⁰ (rpoD) gene specifying the R596H substitution was isolated as a suppressor that restored expression of the arabinose operon in a cya mutant background (25). It has been suggested that the R596H substitution enhances the ability of AraC to interact productively with RNAP (25, 37). This same amino acid substitution was subsequently isolated based on its ability to suppress the effect of a λcI positive control mutation at the λ promoter P_RM. Suppressor mutations in the rpoD gene were sought that would reverse the activation defect of a λcI mutant bearing substitution D38N in its activating region, and a single mutation specifying the R596H change was obtained (35). Although the R596H substitution in ς⁷⁰ reduces the ability of wild-type λcI to activate transcription from P_RM, we have shown that this substitution actually enhances the ability of λcISa109 (which bears a non-wild-type residue at position 38) to stimulate transcription from P_RM (unpublished data). Evidently, certain amino acid-amino acid combinations at position 38 of λcI and position 596 of ς⁷⁰ permit efficient activation, while others do not.

Taken together, the data obtained in previous studies and data obtained in this study suggest that R596 of ς⁷⁰ is exposed on the surface of the polypeptide such that it can be contacted by interacting proteins. This suggestion is supported by the results of structural modeling. Region 4 of ς⁷⁰ contains a putative helix-turn-helix motif, and the structure of this DNA-binding domain has been modeled based on the three-dimensional crystal structures of two related helix-turn-helix proteins, the E. coli NarL protein and the bacteriophage 434 Cro protein (4, 37). In both structural models, residue 596 is solvent exposed and accessible when the protein domain is bound to DNA.

AlgQ and regulation of gene expression in P. aeruginosa.

Our findings with AlgQ may be relevant to understanding the mechanism by which it regulates gene expression in P. aeruginosa. The finding that AlgQ, like Rsd, can interact with a C-terminal fragment of the ς⁷⁰ subunit of RNAP provides strong support for the proposal that it is a functional homolog of Rsd. This conclusion is reinforced by our finding that the interactions of Rsd and AlgQ with region 4 of ς⁷⁰ are both weakened by the same substitution in ς⁷⁰.

The algQ (algR2) gene was originally identified as a regulatory gene implicated in activation of the algD promoter in experimentally derived mucoid strains of P. aeruginosa (8, 32). In particular, a putative algQ mutation was found to eliminate transcription from the algD promoter in a laboratory strain of P. aeruginosa (8). Furthermore, overexpression of algQ was shown to reverse the nonmucoid phenotype of spontaneous Alg⁻ mutants derived from mucoid cystic fibrosis isolates of P. aeruginosa (32). Interestingly, expression of algQ was also shown to mediate strong activation of the algD promoter in E. coli cells grown under high-osmolarity conditions (32). Activation of the algD promoter has been shown to occur by at least two different mechanisms involving one of two alternative sigma factors, ς^E and ς⁵⁴ (3, 9, 11, 21, 38, 44, 49, 59). In particular, mutations that activate ς^E (encoded by the algU gene) lead to increased expression of algD (reviewed in reference 9), but algD expression can also be activated by a ς⁵⁴-dependent pathway (3). Our results support the idea that the effect of AlgQ on algD expression is likely to be indirect since there is no evidence that the ς⁷⁰ form of RNAP can recognize the algD promoter. Possibly AlgQ functions as an anti-ς factor, increasing the amount of RNAP core that is available to bind the relevant alternative ς factor (either ς^E or ς⁵⁴), thereby increasing the occupancy of the algD promoter. It is interesting that algD gene expression is negatively controlled by an anti-ς factor (the product of the mucA gene), which is specific for ς^E (44, 48, 58). Thus, most mucoid cystic fibrosis isolates of P. aeruginosa have been found to bear mutations in the mucA gene, which result in constitutive alginate production (2).

Our finding that AlgQ can bind to ς⁷⁰ from E. coli as well as to ς⁷⁰ from P. aeruginosa could be relevant to its ability to activate the algD promoter in E. coli (32). Nevertheless, it is possible that the role of AlgQ in activation of the algD promoter is unrelated to its ability to bind to ς⁷⁰ in either P. aeruginosa or E. coli. Although AlgQ was originally reported to have a kinase activity (45), the subsequent finding that it regulates production of a kinase (nucleoside diphosphate kinase) which has a similar molecular weight suggests that AlgQ is not itself a kinase (47).

The regulatory effects of AlgQ are not limited to alginate production. For example, AlgQ regulates production of a variety of secretable virulence factors, up-regulating a neuraminidase and a siderophore and down-regulating extracellular proteases and a rhamnolipid biosurfactant (6, 46, 54). AlgQ also regulates production of Ndk (see above) and succinyl coenzyme A synthetase, an enzyme of the tricarboxylic acid cycle that forms a complex with Ndk in P. aeruginosa (33, 46, 47). Finally, characterization of an algQ null mutant revealed a dramatic loss of viability in the stationary phase of growth, as well as reductions in the intracellular concentrations of GTP, ppGpp, and inorganic polyphosphate (34). Although the molecular basis for these regulatory effects has not been defined, the pleiotropic nature of AlgQ-dependent phenotypes and our results support the suggestion that AlgQ is a global regulator of transcription in P. aeruginosa. In addition to its similarity to Rsd, AlgQ exhibits 58% identity with PfrA, a positive regulator of siderophore biosynthetic genes in Pseudomonas putida (54). Interestingly, both PfrA and a putative member of the extracytoplasmic function family of alternative ς factors (PfrI) are required for transcriptional activation of siderophore biosynthetic genes under iron limitation conditions in P. putida (54, 55).

Two-hybrid assay for the interaction of transcriptional regulators with region 4 of ς⁷⁰.

The two-hybrid system that we used to study the interactions of Rsd and AlgQ with region 4 of ς⁷⁰ should facilitate studies of other regulators that interact with this region of ς⁷⁰ from E. coli, P. aeruginosa, or other bacteria. Use of the α-ς⁷⁰ chimeras, in particular, could facilitate genetic analysis of these interactions by providing a convenient vehicle for mutagenesis of region 4 of ς⁷⁰. Whereas isolation and analysis of rpoD mutations are complicated by the fact that ς⁷⁰ is an essential protein that exerts global effects on cellular transcription, our α-ς chimeras exert their effects at specifically designed test promoters. Moreover, mutant chimeras can be assayed to determine their abilities to interact with a number of different regulators so that the specificities of their effects can be assessed.