Mutational Analysis of an Extracytoplasmic-Function Sigma Factor To Investigate Its Interactions with RNA Polymerase and DNA

The extracytoplasmic-function (ECF) family of sigma factorscomprises a large group of proteins required for synthesis ofa wide variety of extracytoplasmic products by bacteria. Residuesimportant for core RNA polymerase (RNAP) binding, DNA melting,and promoter recognition have been identified in conserved regions2 and 4.2 of primary sigma factors. Seventeen residues in region2 and eight residues in region 4.2 of an ECF sigma factor, PvdSfrom Pseudomonas aeruginosa, were selected for alanine-scanningmutagenesis on the basis of sequence alignments with other sigmafactors. Fourteen of the mutations in region 2 had a significanteffect on protein function in an in vivo assay. Four proteinswith alterations in regions 2.1 and 2.2 were purified as His-taggedfusions, and all showed a reduced affinity for core RNAP invitro, consistent with a role in core binding. Region 2.3 and2.4 mutant proteins retained the ability to bind core RNAP,but four mutants had reduced or no ability to cause core RNApolymerase to bind promoter DNA in a band-shift assay, identifyingresidues important for DNA binding. All mutations in region4.2 reduced the activity of PvdS in vivo. Two of the region4.2 mutant proteins were purified, and each showed a reducedability to cause core RNA polymerase to bind to promoter DNA.The results show that some residues in PvdS have functions equivalentto those of corresponding residues in primary sigma factors;however, they also show that several residues not shared withprimary sigma factors contribute to protein function.

INTRODUCTION

Top

Introduction

Bacterial core RNA polymerase (RNAP; ${alpha}$ ₂ßß')requires a fifth subunit, the sigma factor, in order to initiatespecific transcription from double-stranded templates. CoreRNAP-sigma factor complexes bind to gene promoters in a sequence-specificmanner (reviewed in references 18, 25, and 46). Sigma factorproteins of the ${sigma}$ ⁷⁰ family bind at specific DNA sequences thatare usually centered at about the –35 and –10 regionof the promoter. Multiple sigma factor proteins can be presentin a single bacterium, and each has an important role in controllinggene expression, with different sigma factors recognizing differentpromoter sequences (25, 66). Primary sigma factors, such as ${sigma}$ ⁷⁰ in Escherichia coli, are essential proteins that are responsiblefor the majority of RNA synthesis in exponentially growing cells.Alternative sigma factors are nonessential proteins requiredunder certain circumstances and, except for ${sigma}$ ⁵⁴ and related proteins,have various degrees of sequence similarity to ${sigma}$ ⁷⁰. The extracytoplasmic-function(ECF) family is a large group of sigma proteins that regulatethe production of various extracytoplasmic products in a varietyof bacteria (reviewed in references 20 and 40). Although theyhave conserved sequence features present in ${sigma}$ ⁷⁰-type proteins(34), ECF sigma factor proteins are much smaller than most othersigma factors (typically 20 to 25 kDa) and therefore may havesignificant differences in their interactions with DNA, coreRNAP, or other proteins.

Sigma factors have been divided into regions based on sequencealignments (33, 34, 66), with region 2 sequences showing thehighest level of similarity (Fig. 1A). Residues in region 2are required for binding to core RNAP (regions 2.1 and 2.2),recognition of the –10 sequence element (region 2.4),and melting of promoter DNA (region 2.3). The evidence for thisfirst came from mutagenesis studies with ${sigma}$ ⁷⁰ and ${sigma}$ ³² ( ${sigma}$ ^H) fromE. coli and ${sigma}$ ^A and ${sigma}$ ^E from Bacillus subtilis (1, 27, 52, 53, 56)and is strongly supported by the crystal structure of ${sigma}$ ⁷⁰ inholoenzyme complexes (43, 59). ECF proteins have significantsimilarities with other ${sigma}$ ⁷⁰-related proteins in region 2, butthere are also significant differences, and it is necessaryto introduce a gap in order to align aromatic residues of region2.3 (34); these residues are associated with open complex formation(13, 28).

View larger version (69K):
[in this window]
[in a new window]
FIG. 1. Targeted mutagenesis of region 2 from PvdS. (A) Sequence alignment of region 2 from sigma factor proteins. Sigma factor proteins are divided into regions 1 to 4 based on blocks of sequence similarity (66), and four subregions within region 2 (2.1 to 2.4) are indicated. A sequence alignment of this region from various sigma factors was generated using CLUSTAL_X (58) and manually adjusted to incorporate structure-based information as described previously (8). The alignment was displayed by MacBoxShade (http://www.isrec.isb-sib.ch/ftp-server/boxshade/MacBoxshade). Similar alignments have been presented previously (33, 34). The functions of the different sigma factors are described in references 20 and 66. Residues of PvdS mutated in this study are indicated by asterisks. P. syringae, Pseudomonas syringae; M. xanthus, Myxococcus xanthus; P. fluorescens, Pseudomonas fluorescens; P. putida, Pseudomonas putida. (B) The effects of mutations in region 2 on PvdS activity in vivo. The activity of PvdS was determined using a lacZ reporter gene assay as described in Materials and Methods, with values expressed in Miller units and standard errors shown in brackets. WT, wild type.

A helix-turn-helix (HTH) DNA-binding motif in region 4.2 ofthese sigma factors contains residues that make sequence-specificcontacts with DNA bases in the –35 hexamer sequence (7,42). The HTH is a very common motif found in many DNA-bindingproteins, and several crystal structures that provide informationon how it interacts with DNA are available (reviewed in reference65). The first helix acts to position the second helix, whichprotrudes from the surface of the protein to make discriminatorysequence-specific contacts with the DNA nucleotides. Mutationsmade to residues of the ${sigma}$ ⁷⁰ HTH and the ${sigma}$ ^E HTH alter the promotersequences recognized by these sigma factors (17, 29, 54, 57),a fact that emphasizes the role of the HTH in promoter recognition.

PvdS is an ECF protein required for the production of pyoverdine,exotoxin, and PrpL protease, three virulence factors secretedby the pathogen Pseudomonas aeruginosa (11, 41, 45, 62). PvdSis a true sigma factor, binding with core RNAP to pyoverdinesynthesis gene promoters and triggering transcription of thesegenes (32, 63). PvdS binds with core RNAP to promoters containinga DNA sequence motif, TAAAT, spaced 16 bases from a second sequencemotif, CGT (60, 64). By analogy with ${sigma}$ ⁷⁰, regions 4.2 and 2.4of PvdS are most likely to be responsible for the recognitionof these motifs. The activity of PvdS is controlled posttranslationallyby an anti-sigma factor, FpvR, that can bind to PvdS, inhibitingits activity (31, 48).

The research described here aimed to identify residues withinregion 2 and region 4.2 of PvdS that are involved in interactionswith core RNAP and with DNA. Each targeted residue was mutatedto an alanine, and the mutations were examined for their effecton PvdS function in vivo. Selected proteins were also purifiedand used in in vitro assays to determine whether they couldbind core RNAP or DNA and direct transcription.

MATERIALS AND METHODS

Top

Materials and Methods

Bacterial strains and constructs. Plasmids were maintained in E. coli strain MC1061 (9). Wild-typeand mutant pvdS genes were in the vector pProEX (GIBCO BRL).E. coli was grown in LB broth (49) at 37°C with aeration.Ampicillin and chloramphenicol were added to final concentrationsof 100 µg/ml and 30 µg/ml, respectively.

Site-directed mutagenesis. The GeneEditor site-directed mutagenesis system (Promega) wasused following the protocol provided by the manufacturer, withpProEX::pvdS (63) as the template. Mutagenic oligonucleotides(synthesized by GIBCO BRL) were designed for each mutation toreplace the target residue codon with one encoding alanine.Each mutagenic oligonucleotide was 33 bases in length and had15 bases identical to wild-type pvdS sequence on either sideof the mutated codon. Oligonucleotides were designed to incorporatea change in a restriction site at the site of the mutation.DNA carrying potential mutations was screened by restrictiondigestion, and mutant genes were sequenced (Centre for GeneResearch, University of Otago) to confirm the presence of theintended mutations and the absence of other mutations.

ß-Galactosidase assays. E. coli MC1061 cells containing the reporter plasmid pMP190::pvdEwere transformed with pProEX plasmids containing wild-type ormutant pvdS genes. Plasmid pMP190::pvdE has a PvdS-dependentpromoter, pvdE, cloned upstream of a lacZ reporter gene so thatlacZ expression and consequent ß-galactosidase productionare dependent on the activity of PvdS (11, 31, 64). Cultureswere grown to an optical density at 600 nm of 0.3, and the expressionof pvdS was induced by the addition of IPTG (isopropyl-ß-D-thiogalactopyranoside)to a final concentration of 1 mM. Aliquots (100 µl) wereremoved after 1 h and assayed for ß-galactosidaseby use of the Miller assay (39).

Purification of PvdS. Wild-type and mutant PvdS proteins were purified as describedpreviously (63). Briefly, the pvdS and mutant pvdS genes werepresent in the pProEX (GIBCO BRL) vector that results in expressionof proteins fused to six-histidine tags at their N termini.E. coli MC1061 cells expressing His-tagged PvdS proteins werecollected and lysed by sonication, and PvdS was purified underdenaturing conditions on a Ni-nitrilotriacetic column (QIAGEN).Purified proteins were renatured by dialysis and stored in TGEDbuffer (50% glycerol, 50 mM Tris-HCl, pH 7.9, 0.01% [vol/vol]Triton X-100, 0.1 mM EDTA, 50 mM NaCl, and 0.1 mM dithiothreitol).Protein concentrations were determined by the Bradford proteinassay method (Bio-Rad).

Quantification of PvdS. Cells expressing wild-type and mutant PvdS were collected andlysed by boiling in phosphate-buffered saline (0.15 M NaCl,0.012 M NaH₂PO₄, 0.036 M Na₂HPO₄ [pH 7.2], and 0.1% sodium dodecylsulfate [SDS]), and the protein concentrations were determinedby using the Bradford assay (6). Two dilutions were made, andequal amounts of protein (approximately 0.42 and 0.18 ng) fromdifferent cultures were loaded onto a nitrocellulose membraneusing a slot blot apparatus (Schleicher and Schull Minifold-II).PvdS protein was detected using an anti-PvdS monoclonal antibodyas described previously (63), except that bound secondary antibodywas detected by use of West Pico chemiluminescent substrate(Pierce). Bands were then quantified using Molecular Analystsoftware (Bio-Rad), and values were expressed as percentagesof the value for the wild type.

Core-binding assays. The protocol used was based on the method of Bowers and Dombroski(5). Core RNAP (4 pmol; Epicenter) was incubated with 4 pmolof histidine-tagged PvdS in protein dilution buffer (PDB; 10mM Tris-Cl [pH 7.5], 1 mM ß-mercaptoethanol, 0.3 MNaCl, 10 mM MgCl₂, 0.1 mM EDTA, 0.05% Tween 20, 250 µg/mlbovine serum albumin, and 5% glycerol) in a final volume of30 µl for 15 min at 37°C. Each mixture was incubatedwith antibodies (1 µl) against core RNA polymerase for1 h on ice, and then core-sigma complexes were purified usingprotein A-agarose (Sigma). After they were washed three timeswith PDB to remove unbound sigma factor, samples were heatedin 6x SDS loading buffer, and the supernatant was loaded ontoa 10% SDS-polyacrylamide gel. Following electrophoresis, proteinswere electroblotted onto a membrane and subjected to Westernanalysis with anti-PvdS antibodies. Bands were quantified withMolecular Analyst (Bio-Rad) software.

Gel shift assays. Gel shift assays were carried out as described previously (63).Briefly, PvdS proteins (8 pmol) were incubated with core RNAP(3.4 pmol) and digoxigenin-labeled promoter probe DNA (pvdE/F)at 37°C for 15 min and subsequently for 5 min at room temperature.Reaction mixtures were loaded onto a 5% native polyacrylamidegel and subjected to electrophoresis at 4°C. DNA probeswere transferred to an N⁺ membrane (Roche Molecular Biochemicals)and detected using anti-digoxigenin antibodies following themanufacturer's protocol (Roche Molecular Biochemicals).

RESULTS

Top

Results

Mutational analysis of regions 2.1 and 2.2 of PvdS. Studies with other sigma factors have indicated that these regionscontain residues required for binding to core RNAP. Residuesto be mutated were selected on the basis of their alignmentwith ${sigma}$ ⁷⁰ family members and ECF sigma proteins (Fig. 1) and includedpositions where (i) the same or similar residues are presentin primary, alternative, and ECF sigmas (L25, R31, Q44, F47,R49); (ii) the same or similar residues are present in ECF sigmasbut not in primary sigmas (E40, D41, D45) and may confer class-specificfunctions; or (iii) the residue in PvdS is different from thoseof other sigmas and may confer PvdS-specific functions or maynot be critical for function (N21, R22, R36, S52). Residueslikely to be important for maintaining the tertiary structureof PvdS (identified by comparison to the ${sigma}$ ⁷⁰ crystal structure)were avoided. Targeted residues were individually mutated toalanine, and the mutated genes were expressed in E. coli asHis-tagged fusions from the pProEX expression vector. QuantitativeWestern analysis showed that all the PvdS derivatives were presentin cells in amounts comparable to that for wild-type protein(data not shown), indicating that the mutations did not significantlydisrupt protein folding and stability.

The plasmid constructs were then transformed into E. coli containingthe reporter plasmid pMP190::pvdE. This has a PvdS-dependentpromoter, pvdE, upstream of a lacZ reporter gene such that theexpression of lacZ in E. coli is dependent upon the activityof PvdS (11, 31, 64). Mutations of eight residues (R22, L25,E40, D41, Q44, D45, F47, and R49) resulted in a level of activityless than 35% of that of the wild type, and mutations N21A andR31A reduced activity to about 60% of that of the wild type(Fig. 1B). L25, Q44, F47, and R49 are conserved throughout sigmaproteins, and residues at these positions are implicated inthe binding of core RNAP by ${sigma}$ ⁷⁰ (42, 43, 52, 59). D41 and toa lesser extent E40 and D45 are highly conserved in the ECFfamily (Fig. 1A), and residues at the equivalent positions in ${sigma}$ ⁷⁰ have been shown to be involved in binding to core RNAP (52).Residue N21 is not conserved and R31 is semiconserved, withsimilar residues being present in many but not all sigma factors(Fig. 1A), and the data indicate that these residues are importantbut not critical for PvdS function. Mutations R36A and S52A,where the residues in PvdS differ from those in other sigmas,had little or no effect on protein activity. Mutation R22A hada major effect on the activity of PvdS, even though this isnot a conserved residue.

Four mutant proteins, namely, R22A, D41A, F47A, and R49A, wereselected for in vitro study. These proteins were purified asHis-tagged fusions, and their abilities to bind to core RNAPwere determined by use of an immunoaffinity assay (Fig. 2).All four proteins had significant reductions in their abilitiesto bind to core RNAP relative to that of wild-type PvdS, althoughthey did retain some core-binding activity. These data suggestthat these residues are specifically involved in the recognitionof core enzyme by PvdS. It is likely that a reduced affinityof PvdS for core RNAP has a more severe effect on activity invivo due to the presence of other sigma factors that competefor a limiting pool of core RNAP (35) and are absent in thein vitro studies. The relative effects of the different mutationson core binding did not completely correlate with the relativeeffects of the mutations in the reporter gene assay (Fig. 1),suggesting that the mutations may have other effects in additionto affecting core binding.

View larger version (46K):
[in this window]
[in a new window]
FIG. 2. Affinities of mutant PvdS proteins for core RNAP. PvdS proteins carrying mutations in region 2.1/2.2 were incubated with equimolar amounts of core RNAP, and core-PvdS complexes were purified by immunoaffinity. PvdS coprecipitating with core RNAP was detected by Western analysis, and representative results are shown (upper panel). Each assay was repeated at least two times, and the amounts of immunoprecipitated PvdS as percentages of the wild-type (WT) amount are shown (lower panel).

Mutational analysis of regions 2.3 and 2.4 of PvdS. In ${sigma}$ ⁷⁰, regions 2.3 and 2.4 form part of a continuous ${alpha}$ -helix(36, 43, 59) and are involved in DNA melting and recognitionof the –10 sequence motif, respectively. Five residues(Y66, Q69, N73, D77, and R80) that are likely to lie on theoutward face of a corresponding ${alpha}$ -helix in PvdS and thereforewould be positioned to interact with promoter DNA were selectedfor mutagenesis. Y66 corresponds to a conserved aromatic residuepresent at this position in all ${sigma}$ ⁷⁰-type proteins (Fig. 1) whichhas been previously implicated in DNA melting in ${sigma}$ ⁷⁰/ ${sigma}$ ^A proteins(28, 47). Residues N73, D77, and R80 are conserved in otherECF factors thought to recognize similar –10 region sequenceelements, whereas different residues are present in sigma factorsthat recognize different promoters, so that these residues arecandidates for DNA recognition (Fig. 3A). Each residue was mutatedto an alanine, and quantitative Western analysis showed thatall the mutants were expressed in amounts similar to that seenfor the wild type (data not shown). Mutant proteins were assayedfor activity in vivo (Fig. 3B). All mutations except for Q69Aessentially abolished PvdS activity, indicating that these residuesare essential for PvdS function. The mutant proteins with negligibleactivity were purified and tested for the ability to bind tocore RNAP and DNA. All four bound to core enzyme to the sameextent as wild-type PvdS (Fig. 4A), indicating that the mutationshad not affected interactions with core RNAP. Mutant proteinswere also tested for the ability to bind with core RNAP to promoterDNA in a gel shift assay (Fig. 4B). Y66A, D77A, and R80A proteinsdid not give rise to detectable DNA-protein complexes, indicatingthat these mutations affected the interaction with promoterDNA or the stability of the DNA-protein complexes. N73A gaverise to DNA-protein complexes, although in amounts consistentlylower than that found in wild-type protein.

View larger version (39K):
[in this window]
[in a new window]
FIG. 3. Mutation of region 2.4 from PvdS. (A) A comparison of region 2.4 sequences with the proposed –10 sequences for some ECF sigma factor proteins. The sequences of region 2.4 of the following ECF sigma factor proteins along with their proposed –10 recognition sequences are shown: CarQ (38), ${sigma}$ ^W (21), ${sigma}$ ^X (22), HrpL (67), AlgU (16), and ${sigma}$ ^C (55). Residues of PvdS mutated in this study are indicated by asterisks, and boxes denote conserved residues that are required for PvdS function. (B) The effects of mutations in region 2.4 on PvdS activity in vivo as assayed by ß-galactosidase reporter assays. WT, wild type.

View larger version (35K):
[in this window]
[in a new window]
FIG. 4. Activities of region 2.3 and 2.4 mutants in vitro. (A) Purified mutant proteins were tested for their ability to bind to core RNAP in a core-binding immunoaffinity assay (see Materials and Methods). PvdS coprecipitating with core RNAP was detected by Western analysis, and representative results are shown (upper panel). Each assay was repeated at least two times, and the amounts of immunoprecipitated PvdS as percentages of the wild-type amount are shown (lower panel). (B) Gel shift assay with region 2.3 and 2.4 mutant proteins. The gel shift assay was carried out with PvdS proteins, core RNAP, and a PvdS-dependent promoter fragment as described in Materials and Methods. The positions of unbound DNA and DNA-protein complexes are indicated.

Mutational analysis of region 4.2 of PvdS. Alignment of the sequence of PvdS with those of other sigmafactors allowed the identification of the HTH DNA-binding motifin region 4.2. This region of PvdS is shown aligned with otherHTH regions in Fig. 5A. The predicted HTH of PvdS was determinedto have a 90% chance of having an HTH structure by use of theDodd-Egan matrix method (14). A model for the PvdS HTH is shownin Fig. 5B. In this model, conserved hydrophobic residues inthe second helix (residues V15, I19, and L23) face away fromthe DNA and form the interior of the HTH, whereas more-hydrophilicresidues are on the external face of this helix and are positionedto interact with DNA. This model is consistent with HTH motifsfor which a three-dimensional structure has been determined(reviewed in reference 65).

View larger version (52K):
[in this window]
[in a new window]
FIG. 5. Targeted mutagenesis of region 4.2 of PvdS. (A) Alignment of HTH motifs. Amino acid sequences of predicted HTH motifs from sigma factor proteins PvdS (11, 41), FecI (44), ${sigma}$ ⁵⁴ (10), and ${sigma}$ ^E (56) are aligned with the HTH motifs from the ${sigma}$ ⁷⁰ (43, 59) and the DNA-binding proteins ${lambda}$ repressor (4), ${lambda}$ cro (2), and CAP (catabolite activator protein) (51), for which crystal structures have been determined, as well as with the predicted HTH motif from TyrR (24). The numbers at the ends of the amino acid sequences indicate the positions of the residues within each protein. Residues of PvdS that were mutated in this study are circled. Underlined residues have been shown from crystal data to interact with DNA, and residues in light gray boxes have been shown by mutagenesis to be important for function. The light gray circled residue in ${sigma}$ ⁵⁴ has been shown not to be important for function. (B) A model of the HTH from PvdS. This model was constructed by comparison with HTHs from other DNA-binding proteins whose crystal structures are known (panel A). Residues that may interact with DNA are circled and were mutated in this study. (C) The effects of mutations in region 4.2 on PvdS activity in vivo as assayed by ß-galactosidase reporter assays. See the legend to Fig. 1B for further details.

Eight residues that had the potential to interact with DNA (Fig.5B) were altered to alanine residues. Amounts of six of themutant proteins were similar to, or slightly greater than, thatfor the wild type (Fig. 6A). The strain expressing M161A containedabout 50% less PvdS protein than the wild type, and the strainexpressing P155A contained about 30% less. These mutations mayhave affected protein stability with a consequent degradationof misfolded PvdS.

View larger version (84K):
[in this window]
[in a new window]
FIG. 6. Quantification and gel shift analysis of region 4.2 mutant proteins. (A) Effects of mutations on amounts of PvdS. Quantification of PvdS was carried out as described in Materials and Methods. Each strain was assayed at least three times, and PvdS protein was detected using an anti-PvdS monoclonal antibody; representative results are shown. Bands were then quantified, and the amounts of mutant PvdS relative to that of wild-type (WT) PvdS are shown to the right, with standard deviations shown in parentheses. (B) Gel shift assays with PvdS (F160A) and PvdS (R163A). PvdS proteins were purified and assayed for their ability to bind DNA as described in Materials and Methods. The positions of DNA-protein complexes are indicated by arrows. Neither core RNAP in the absence of PvdS nor PvdS without core RNAP form DNA-protein complexes in this assay (63). (C) Core-binding immunoaffinity assay. Wild-type PvdS or mutant proteins (F160A and R163A) were incubated with core enzyme, and complexes were purified from solution with antibodies against core enzyme as described in Materials and Methods. Results are averages of three independent assays.

The effect of each mutation on PvdS function in vivo was assessedwith a lacZ reporter assay. All of the mutations reduced thelevel of PvdS function below that of the wild type (Fig. 5C),indicating that each mutated residue contributed to PvdS activity,although all of the mutated proteins retained significant activity.Mutations in this part of PvdS are likely to affect the bindingof PvdS-RNAP to promoters. To test this, two mutant proteins(F160A and R163A) were purified and their activities assessedin vitro in order to analyze the effects of the mutations. Ingel shift assays (Fig. 6B), both mutant proteins had a reducedability, relative to wild-type PvdS, to bind to promoter DNAwith core RNAP. Both mutants bound core RNAP to the same extentas wild-type protein (Fig. 6C), indicating that these proteinswere folded correctly and had retained the ability to interactwith core RNAP. The mutations therefore specifically affectedthe ability of the PvdS-core RNAP complex to interact with DNA.

DISCUSSION

Top

Discussion

Alanine-scanning mutagenesis was carried out on 25 amino acidresidues in PvdS. The targeted residues were expected to besurface exposed and available to interact with core RNAP orDNA rather than to form part of the hydrophobic core of PvdS.Most of the mutations had little or no effect on the amountof PvdS in the bacteria (Fig. 6A; also data not shown), andproteins with mutations in regions 2.3/2.4 and 4.2 retainedthe ability to bind core RNA polymerase. In contrast, a mutation,V158R, at a residue predicted to be required for correct foldingof PvdS, and also mutations causing the insertion of short (four-residue)peptides in PvdS, resulted in no detectable PvdS protein (M.J. Wilson, T. T. Caradoc-Davies, and I. L. Lamont, unpublishedobservations), most likely because PvdS that is not correctlyfolded is rapidly degraded. It is therefore very likely that,with possible exceptions in the cases of P155A and M161A, theeffects of the mutations reflect the roles of the correspondingamino acid residues in the activity of PvdS and are not simplydue to misfolding of the protein.

Only 3 of the 17 mutations in region 2 had little or no effecton PvdS function (over 85% of wild-type activity in vivo) (Fig.1B and 3B), and all of these were at nonconserved positions.For non-ECF sigma factors, region 2.1/2.2 is involved in interactionswith core enzyme (27, 52, 53), and our in vitro data (Fig. 2)indicate that the same is true for PvdS. The large number ofresidues where mutation resulted in reduced or no function indicatesa complex interaction between the two proteins involving thecooperative participation of a large number of amino acid residues.In this respect, PvdS is similar to ${sigma}$ ⁷⁰, for which mutationaland structural studies (43, 52, 59) revealed a very extensivesigma factor-core enzyme interface. This is notable becauseof the much smaller size of PvdS and its low level of overallsequence similarity with ${sigma}$ ⁷⁰. As might be expected, mutationsat residues that are conserved or highly conserved among sigmafactors (L25, R31, Q44, F47, and R49) greatly reduced transcriptionfrom a PvdS-dependent promoter. PvdS F47A and R49A had impairedcore binding, and residues corresponding to L25 and Q44 areimportant for core binding in ${sigma}$ ^H and ${sigma}$ ⁷⁰, so it is very likelythat these residues all play a role in core RNAP interactionsof PvdS.

E40, D41, and D45 are conserved amino acids within the ECF family,but different residues are present at corresponding positionsin primary sigma factors (Fig. 1). Mutations at each of theseresidues had major effects on the activity of PvdS (Fig. 1A).The D41A protein had reduced affinity for core enzyme in vitro(Fig. 2), and the mutations lie in region 2.2, for which theprimary function is core binding, so it is very likely thatthese residues contribute to core binding for PvdS. Mutationsat a residue of ${sigma}$ ⁷⁰ that corresponds to E40 of PvdS, L402, affectedthe affinity of ${sigma}$ ⁷⁰ for core enzyme, showing that this positionat least is important for core binding in different sigma factors,even though the residues that are present are quite different.It may be that the exact nature of the residues at these positionscontributes to determining the relative affinities of differentsigma factors for core enzyme; it has been shown that the affinitiesof different sigmas for core can vary over at least a 16-foldrange (35). Residue R22 was very important for PvdS function(Fig. 1B) and core RNAP binding (Fig. 2), even though quitedifferent residues are present at corresponding positions inother sigma factors. The corresponding residue in ${sigma}$ ⁷⁰ contributesto core binding, with an L384A mutation reducing the affinityof ${sigma}$ ⁷⁰ for core enzyme (52). This indicates that at this positionalso, residues in different sigmas contribute to core binding,even though the exact natures of the residues can be quite different.Overall, it is clear that an extended region of PvdS spanningat least residues R22 to R49 in the linear sequence is involvedin binding core RNA polymerase and that critical residues includeones that are common to different sigma factors and ones thatare sigma factor specific and may determine the relative affinitiesof different sigmas for core RNA polymerase.

There is good evidence that region 2.3 in ${sigma}$ ⁷⁰ is involved inopen complex formation and that this involves aromatic residuesthat contribute to open complex formation by forming stackinginteractions between nucleotide bases to stabilize DNA thathas melted into the single-stranded form (13, 15, 28, 47). Residuesin region 2.3 also participate in the binding of holoenzymeto double-stranded DNA (15). ECF sigma factors are quite differentfrom ${sigma}$ ⁷⁰ in region 2.3, with relatively few aromatic residues,and it is necessary to introduce a gap in this region in orderto align the different sequences (Fig. 1). Only one conservedaromatic residue, Y66, is present in region 2.3 of PvdS (Fig.1). The Y66A mutant had wild-type affinity for core enzyme (Fig.4A) but had no activity in vivo and showed no activity in thegel shift assay (Fig. 4B). This indicates that Y66A directlyaffects DNA binding by the core-PvdS complex, although it isalso possible that the mutation affects DNA melting and thatthis is required in order to obtain DNA-protein complexes inthe band shift assay.

Region 2.4 is involved in the recognition of –10 promotermotifs by ${sigma}$ ⁷⁰, ${sigma}$ ^H, and ${sigma}$ ^E with the N-terminal part of this regionnearest to the transcription start site (12, 15, 26, 61). Thethree mutations in this region (at N73, D77, and R80) greatlyaffected the activity of PvdS in vivo (Fig. 3B) and in bandshift assays (Fig. 4B) but did not affect binding of PvdS tocore enzyme (Fig. 4A). These data are consistent with mutationsin this part of PvdS affecting the recognition of target promoters.The proposed –10 sequence recognition motifs of variousECF sigma factors, along with the region 2.4 sequence, are shownin Fig. 3A. There is a striking correlation between the residuesplaced to interact with DNA at the conserved positions in region2.4 and the sequences of the –10 motifs. All of the proteinsexcept AlgU/RpoE have an R residue at the position correspondingto R80 in PvdS, and all of them except AlgU/RpoE recognize aC · G base pair at the first position in the –10motif. Similarly, the proteins that have a D residue at theposition corresponding to D77 in PvdS are those for which aG · C base pair is present at the second base in the–10 sequence, and those with an N at this position havea C · G base pair in the second position of the –10.For four of the five proteins with an N at position 73, thereis a T · A base pair in the third –10 position,with the only exception being HrpL. Furthermore, studies with ${sigma}$ ^E and ${sigma}$ ^H from B. subtilis (12, 56) indicate that the residuescorresponding to N73 and D77 in PvdS make discriminating –10region contacts for these sigma factors. These findings leadto the hypothesis that the three mutated residues in region2.4 of PvdS are involved in recognition of the –10 promotersequence element. It should be noted, however, that ${sigma}$ ^C, whichhas an H at the position corresponding to N73, also recognizesa T · A base pair in the third –10 position.

All of the mutations in region 4.2 affected PvdS function, althoughall of the mutant proteins retained significant activity invivo (Fig. 5C). This indicates that at least eight residuesin this region of PvdS contribute to protein function, mostlikely through interactions with the –35 region of thepromoter, with no single residue of those examined being criticalto DNA binding. Residues at positions equivalent to T156 andT159 in the HTH are involved in sequence-specific DNA-base interactionsof ${sigma}$ ⁷⁰ (17, 29) as well as in other DNA-binding proteins (2-4,24, 37, 51). The residues corresponding to positions 156, 159,163, and 164 can all form hydrogen bonds with bases in DNA andmake sequence-specific DNA contacts in many other DNA-bindingproteins (37), so they are good candidates for involvement inpromoter recognition. Histidine can carry a positive chargeand is present in many DNA-binding proteins, making contactswith the DNA phosphate backbone and stabilizing the DNA binding(37), and this may be the role of H168.

The likely roles of P155, M161, and F160 are less clear, asthese amino acids do not commonly make sequence-specific contactswith DNA. Mutations P155A and M161A resulted in reduced amountsof PvdS, and this may be at least part of the reason for thelower activities of these proteins in vivo. The F160A mutationhad the most detrimental effect on PvdS activity of all themutations in region 4.2 (Fig. 5C). Phenylalanine is a hydrophobicamino acid, and it is possible that it forms part of the hydrophobicface of the helix, interacting with other parts of PvdS. However,the F160A mutation did not affect amounts of PvdS (Fig. 6A)and purified protein was still able to bind to core RNAP (Fig.6C), indicating that the F160A mutation did not affect proteinstability or folding and therefore is most likely to be at theprotein surface, as predicted by analogy with other HTH proteindomains (Fig. 5B). The phenylalanine side chain cannot formH bonds that could contribute to sequence-specific interactionsbetween proteins and DNA and is not often found in HTH motifs.The crystal structure of the Arc repressor showed that a phenylalanineresidue contacts the DNA backbone, and this residue is importantfor DNA binding and contributes to binding specificity (50).It may be that F160 makes contacts with the sugar-phosphatebackbone at pyoverdine promoters and that these contacts areimportant in enabling PvdS-RNA polymerase to bind to DNA.

Overall, these results suggest that at least three residuesin region 2.4 and several residues in region 4.2 of PvdS collectivelycontribute to promoter recognition. While this work was in preparation,a study using similar approaches with the non-ECF sigma factor ${sigma}$ ³² from E. coli was published (30). Four mutations in region4.2 of ${sigma}$ ³² reduced ${sigma}$ ³²-dependent gene expression to less than30% of that of the wild type. This was in contrast to resultsfor PvdS, in which only one mutation had such a marked effect(Fig. 6). Conversely, mutations in regions 2.3 and 2.4 of ${sigma}$ ³²did not have as marked an effect on gene expression as mutationsin these regions of PvdS (Fig. 2). These differences may reflectdifferences in the molecular recognitions of promoter DNA bythe different sigma factors or differences in the initiationsof transcription following promoter recognition. Determinationof the structure of PvdS in complex with DNA and with core enzymewill be required to obtain a detailed picture of the interactionsbetween PvdS and these binding partners. In addition, the activityof PvdS is posttranslationally regulated by an anti-sigma factor,FpvR (31). Anti-sigma factors bind to their cognate sigmas toprevent their interaction with RNA polymerase core enzyme (19,23), and FpvR has recently been shown to interact with PvdSin vivo (48). The mutants generated in this study, in conjunctionwith studies on protein structure, will be very useful in exploringFpvR-PvdS interactions.

ACKNOWLEDGMENTS

We thank Michael Chamberlin and Malay Ray for supplying anti-coreRNAP antibodies and Michael Vasil for anti-PvdS antibodies.

This project was funded in part by a grant from the New ZealandLottery Health Board.

FOOTNOTES

* Corresponding author. Mailing address: Department of Biochemistry, University of Otago, P.O. Box 56, Dunedin, New Zealand. Phone: 64 3 479 7869. Fax: 64 3 479 7866. E-mail: iain.lamont{at}stonebow.otago.ac.nz.

REFERENCES

Top