| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Division of Molecular Microbiology, Biozentrum, Klingelbergstrasse 50–70, CH-4056 Basel, Switzerland
2 Department of Plant Sciences, University of Oxford, South Parks Road, Oxford OX1 3RB, UK
3 Centre for Ecology and Hydrology, Mansfield Road, Oxford OX1 3SR, UK
4 School of Biological Sciences, University of Auckland, Private Bag 92019, Auckland, New Zealand
Correspondence
P. B. Rainey
p.rainey{at}auckland.ac.nz
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Lists of mutant wspR constructs produced by PSM, mutant wspR constructs produced by SOE-PCR, and oligonucleotides used in the study, BLAST alignment data for WspR and PleD, and a figure showing attachment to glass for SM 
wspR versus WS wspR, are available as supplementary data with the online version of this paper.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
; D'Argenio & Miller, 2004; Römling et al., 2005). Although the biological pathways through which c-di-GMP regulates persistence, cell aggregation and the switch to the commensal lifestyle are currently poorly understood, the proteins responsible for c-di-GMP synthesis and degradation have been determined and, at least partially, characterized. Bacterial diguanylate cyclase (DGC) activity is found in GGDEF domain-containing proteins (Paul et al., 2004; Ryjenkov et al., 2005; Hickman et al., 2005; Kulesekara et al., 2006), whilst EAL domain-containing proteins have been shown to have c-di-GMP-specific phosphodiesterase (PDE) activity (Bobrov et al., 2005; Christen et al., 2005; Schmidt et al., 2005; Kulesekara et al., 2006).
Despite the publication of much biological data concerning GGDEF domains, biochemical demonstration of DGC activity (Paul et al., 2004; Ryjenkov et al., 2005), and the solution of the crystal structure of the GGDEF response regulator PleD (Chan et al., 2004), detailed structure–function analyses of GGDEF proteins are relatively scarce. Christen et al. (2006) have analysed the site of c-di-GMP product inhibition (the ‘I site’) for the Caulobacter crescentus enzymes PleD and DgcA. Additionally, site-specific mutagenesis studies have been carried out, focusing on single residues of interest (Kirillina et al., 2004; Garcia et al., 2004; Goymer et al., 2006). A comprehensive structure–function analysis stands to contribute insights into various issues, including the role of the GGDEF motif, the mode of activation of individual GGDEF proteins, and potential alternative mechanisms of downstream signal transduction.
The GGDEF domain was first recognized in PleD, a response regulator responsible for mediating the swarmer-to-stalked cell transition in C. crescentus (Hecht & Newton, 1995), and has since been identified in large numbers in many bacterial species (Galperin et al., 1999, 2001; Galperin, 2005). GGDEF domains are often associated with receiver, PAS and GAF sensor domains, and are frequently paired with EAL domains (Galperin et al., 2001).
Numerous studies in diverse species have linked GGDEF domain proteins to processes involved in biofilm formation and aggregative behaviour (reviewed by Römling, 2005; Römling et al., 2005). The effects of GGDEF proteins on exopolysaccharide levels have been probed via genetic studies in numerous species, including Pseudomonas putida (Gjermansen et al., 2005; Ude et al., 2006), Pseudomonas fluorescens SBW25 (Spiers et al., 2002, 2003; Goymer et al., 2006; Ude et al., 2006), Salmonella typhimurium (Simm et al., 2004; Garcia et al., 2004; Simm et al., 2005), Thermotoga maritima (Johnson et al., 2005) and Vibrio cholerae (Bomchil et al., 2003; Rashid et al., 2003; Kovacikova et al., 2005).
In addition, GGDEF-containing proteins have been repeatedly implicated in the regulation of cell motility, attachment and virulence (reviewed by Jenal, 2004; D'Argenio & Miller, 2004; Römling, 2005). For example, PleD controls the onset of motility during the C. crescentus cell cycle (Aldridge et al., 2003; Paul et al., 2004). In Pseudomonas aeruginosa, autoaggregation is controlled by the GGDEF response regulator WspR (D'Argenio et al., 2002; Hickman et al., 2005), and twitching motility by FimX, which contains GGDEF and EAL domains (Huang et al., 2003; Kazmierczak et al., 2006). Additional examples of the effects of GGDEF domains on cell attachment and motility include ScrC regulation of attachment factors in Vibrio parahaemolyticus (Boles & McCarter, 2002), control of curli fimbriae by AdrA in S. typhimurium (Simm et al., 2004), and RocS effects on V. cholerae motility (Rashid et al., 2003). A recent survey of all GGDEF-containing genes in P. aeruginosa PA14 has shown that, in several cases, overexpression or inactivation has profound effects on virulence and cell attachment to surfaces (Kulesekara et al., 2006).
The GGDEF response regulator WspR was initially discovered as a consequence of investigations into the genetic causes of adaptive radiation in experimental populations of P. fluorescens SBW25 (Rainey & Travisano, 1998; Spiers et al., 2002). One class of derived genotype colonizes the air–liquid interface of static broth microcosms; these genotypes produce distinctive wrinkled colonies on agar plates and are termed wrinkly spreaders (WS). Transposon mutagenesis has been used to identify genes that determine WS phenotype, focusing initially on a single WS genotype, the large-spreading wrinkly spreader (LSWS) genotype. Amongst other genes, the wrinkly spreader phenotype (wsp) operon has been identified (Spiers et al., 2002; Goymer et al., 2006), which encodes a chemosensory system with homology to the chemotaxis pathway of Escherichia coli (Bantinaki et al., 2007).
WspR is the final gene product and primary output component of the Wsp pathway, and is activated by a currently unknown signal processed by the rest of the Wsp complex (Spiers et al., 2002). A wspR : : mini-Tn5 WS mutant displays a smooth colony morphology (Spiers et al., 2002), and does not produce cellulose or attach to surfaces (Spiers et al., 2003). Expression of wspR in trans stimulates attachment and exopolysaccharide synthesis in various species (D'Argenio et al., 2002; Aldridge et al., 2003; Goymer et al., 2006; Ude et al., 2006). The evolutionary cause of LSWS is a single point mutation in the gene encoding the WspF methylesterase, which results in overactivation of the WspE kinase and constitutive activation of the primary output component of the Wsp pathway, WspR (Bantinaki et al., 2007). Parallel analysis in P. aeruginosa PAO1 has linked the emergence of an aggregative, wrinkled colony morphology to disruption of wspF (D'Argenio et al., 2002; Hickman et al., 2005). The wsp operon is conserved in both SBW25 and PAO1, and the two systems are thought to function in a similar way.
P. fluorescens SBW25 WspR is a 333 residue protein comprising an N-terminal response-regulator receiver domain and a C-terminal GGDEF domain, separated by a linker region (Spiers et al., 2002, 2003). Recently, a combination of random and site-specific mutagenesis has been used to outline basic physical characteristics of WspR (Goymer et al., 2006). Asp 67 has been established as the site of WspR phosphorylation, and WspR activation has been shown to be dependent upon phosphorylation. In addition, constitutively active [e.g. WspR 19 (R129C); D'Argenio et al., 2002; Aldridge et al., 2003] and dominant-negative [e.g. WspR 9 (G296R)] wspR alleles have been identified. A model has been proposed to explain the dominant-negative effect seen upon the production of certain WspR variants. In this model, N-terminal receiver domains without functional C termini competitively inhibit the interaction of chromosomally expressed WspR with the Wsp kinase machinery, and hence reduce the downstream signal (Goymer et al., 2006).
The work described here expands on the findings of Goymer et al. (2006), employing two distinct mutagenesis strategies to probe in detail the structure–function relationship of WspR. In the first of these, short in-frame stretches of DNA were introduced at positions throughout wspR. In the second, the amino acid residues of the RYGGEEF motif were substituted via site-specific mutagenesis. Following mutagenesis, phenotypic assays were applied to the libraries of wspR mutant alleles. Morphological analysis and assays for surface attachment and cellulose production were carried out on SBW25 strains expressing the wspR insert and substitution libraries in trans, allowing for qualitative, quantitative and in silico analysis of WspR. These data, combined with biochemical analyses of cell extracts and purified protein, were used to revise and expand upon models for the structure–function relationship of WspR. More broadly, these models were applied to our understanding of both response regulators and GGDEF domain-containing proteins.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Bacteria were grown in either King's medium B (KB) or Luria–Bertani (LB) medium (King et al., 1954; Miller, 1972) at 28 °C (P. fluorescens) or 37 °C (E. coli) with shaking. KB microcosms contained 6 ml KB medium in 35 ml universal glass vials and were used for attachment and growth assays. Kanamycin was used at a final concentration of 50 µg ml–1, streptomycin at 100 µg ml–1 and tetracycline at 12.5 µg ml–1. Oligonucleotide primers used in this work are listed in Supplementary Table S3.
|
). pME6010-wspR was constructed by ligation of wspR, excised as a BamHI/EcoRI fragment from pWspR12 (Aldridge et al., 2003) between the BglII and EcoRI sites of pME6010 (Heeb et al., 2000), destroying the BglII site in the process. pET42b-wspR was constructed by ligation of the relevant wspR PCR fragment (amplified with primers WspRPurFor and WspRPurRev from pME6010-wspR), between the XhoI and NdeI sites of pET42b (Novagen).
Overexpression and purification of His6-WspR.
Overnight cell cultures of E. coli BL21-(DE3) containing pET14b-wspR and pET42b-wspR were used to inoculate LB medium plus ampicillin, to initial OD600 0.1. Cell cultures were incubated for 150 min at 37 °C with shaking, before protein expression was induced with IPTG (Sigma) solution at a final concentration of 100 µM for pET14b-wspR, and 1 mM for pET42b-wspR. Cell cultures were then incubated for an additional 150 min at 30 °C. Cells were harvested by centrifugation and resuspended in running buffer (25 mM Tris/HCl, pH 8.0, 250 mM NaCl, 1 % 
-mercaptoethanol), before lysis by sonication and French press. Following lysis, the samples were centrifuged (30 000 g, 15 min, 4 °C) and His6-WspR was purified from the supernatant via nitrilotriacetic acid (NTA)-nickel chromatography, according to the manufacturer's protocol (Qiagen). Both His6-WspR samples were eluted in the 200 mM imidazole fraction, and N-terminal His6-WspR was verified by MS. The protein concentration of samples was determined as described elsewhere (Bradford, 1976) (N-terminal His6-WspR 1.6 mg ml–1, and C-terminal His6-WspR 5.0 mg ml–1).
Western blotting.
Overnight cell cultures were lysed by incubation (95 °C, 5 min) with 1 volume SDS loading buffer (10 mM Tris/HCl, pH 6.8, 0.005 % bromophenol blue, 10 % glycerol, v/v, 2 % SDS). Proteins from the resulting cell extracts were separated on pre-poured 12 % Tris/HCl gels (Bio-Rad) and blotted onto PVDF membranes (Millipore). After overnight incubation in blocking solution (15 mM Tris/HCl, pH 7.0, 10 mM KCl, 10 mM NaCl, 0.01 % Tween-20, 10 % glycerol, v/v, 5 % milk powder), WspR was detected with a 1 : 2500 dilution of WspR-specific polyclonal antiserum (Harlan Seralabs) and a 1 : 3000 dilution of anti-rabbit secondary antibody (Invitrogen). Bound antibodies were visualized with ECL chemiluminescent detection reagent (Amersham Biosciences) and photographic film.
Assay for DGC activity.
DGC activity was assayed as described previously (Paul et al., 2004). Overnight cultures of strain SBW25 were grown, cells were harvested, resuspended in running buffer, and partially lysed by gentle sonication. Assays were run in 50 µl (final volume) running buffer containing 10 µl cell extract or purified WspR, and started by addition of 100 µM (final concentration) GTP (18.5 kBq [
-32P]GTP) (Amersham Biosciences). Retardation factor (Rf) values were recorded as the ratio of spot migration distance to the distance migrated by the solvent front.
PCR.
A standard reaction contained 5 µl 2 mM dNTP mix, 2.5 µl 10x polymerase buffer, 1 U Pfu turbo polymerase (Stratagene), 0.4 pmol each primer, and 10–20 ng template DNA, made up to 25 µl with deionized water. Following an initial template denaturation step of 3 min at 94 °C, amplification was performed by 25 cycles of denaturation for 30 s at 94 °C, annealing for 30 s at the appropriate temperature (56–62 °C, depending on the reaction), and strand extension for 1 min 30 s at 72 °C. A final extension step for 5 min at 72 °C was performed before samples were kept at 4 °C.
Pentapeptide scanning mutagenesis (PSM).
PSM was used to insert 15 bp sections of DNA into pME6010-wspR, as described by Hallet et al. (1997), except that a morphological screen was used for transposon insertions in wspR. P. fluorescens WS-4 (WS wspR : : mini-Tn5) was transformed with pooled pME6010-wspR : : Tn4430 DNA, and transformant colonies exhibiting a smooth morphology were selected. These colonies contained constructs unable to complement WS-4; hence they contained transposon insertions in wspR. Following mutagenesis, the site of insertion in each case was verified by sequencing.
Strand overlap extension PCR (SOE-PCR).
SOE-PCR (Ho et al., 1989) was used during the systematic mutagenesis of wspR. DNA on either side of the point of mutation was amplified by a standard PCR with primers SOEFor/Rev 1–25 that complemented at the site for joining (and containing the altered nucleotide sequence). PCR products were purified and 1 µl of each used as template in the subsequent SOE-PCR. The SOE-PCR was carried out with primers SOEFOR and SOEREV in a final volume of 25 µl. Following an initial template denaturation step of 3 min 30 s at 94 °C, strand extension was carried out in two steps: primer annealing for 30 s at 56 °C, followed by extension for 2 min at 72 °C. DNA was then amplified under the conditions described for the standard PCR protocol, with annealing at 56 °C for 30 s. SOE-PCR fragments were confirmed by sequencing, and ligated between the XhoI and EcoRI sites of pME6010.
Degenerate PCR-based mutagenesis.
In order to efficiently mutagenize the GGEEF motif of wspR, an upstream restriction site was required. This allowed mutagenesis to proceed via PCR with degenerate primers, and reduced the size of the inserts produced from 1 kb to 
300 bp (the region of wspR DNA between the introduced site and the downstream EcoRI site). A SacI restriction site (GAGCTC, encoding E-L) was introduced at positions 339/240 of wspR, via SOE-PCR with primers SacIFor and SacIRev. PCR was carried out using the degenerate primers DEG1–14 with the primer SOEREV, and pME6010-wspR-SacI as a template. PCR products were ligated as a pool between the SacI and EcoRI sites of pME6010-wspR-SacI. Smooth (SM) cells were transformed with the pooled constructs, and colony morphology on LB agar was determined. Individual colonies were selected on the basis of morphology, and the substituted codon in each case was verified by sequencing.
DNA sequencing.
A 50–100 ng aliquot of plasmid DNA was mixed with 1 pmol primer SeqFor or SeqRev and 4 µl Big-Dye ready reaction mix version 3.0 (Applied Biosystems), in a final volume of 10 µl. Samples were thermocycled for 25 cycles of 10 s at 96 °C, 5 s at 50 °C, and 4 min at 60 °C, before cooling to 4 °C. Following the reaction, samples were ethanol-precipitated and submitted for sequence analysis on an ABI 3100 sequencer.
Measurement of overnight culture cell density.
Assays for attachment and cellulose production described below were inoculated with 24 h cell cultures. Accordingly, the effects of WspR variants on SM culture growth were measured. KB microcosms inoculated with 60 µl aliquots of overnight cultures, and with the lids loosely taped in place, were incubated at 28 °C with shaking. All cultures displayed similar optical densities (OD600) after 24 h growth. A Spectronic-20 spectrometer (Genesys) was used throughout.
Attachment assay.
Bacterial attachment was determined quantitatively using crystal violet (CV), as described by Spiers et al. (2003). KB microcosms were inoculated with 60 µl aliquots of overnight cultures and incubated statically at 28 °C for 48 h. Vials were emptied and washed vigorously with deionized water. One millilitre of 0.05 % (w/v) CV (Sigma) was added and mixed for 2 min. The vials were then emptied and washed with deionized water. CV was eluted in 5 ml ethanol with vigorous shaking for 1 h, and A570 determined.
Congo Red (CR) binding assay.
Cellulose expression was measured using a CR binding assay adapted from that described by Spiers et al. (2003). Ten-microlitre drops of overnight cultures were grown on 25 ml KB agar plates for 24 h at 28 °C. Colonies were resuspended in 1 ml 0.005 % (w/v) CR (Sigma) and incubated for 2 h at 37 °C. Colony material was then pelleted by centrifugation. The amount of CR remaining in the supernatant was determined by measurement of A490 and comparison with appropriate CR standards. CR binding was expressed as a fraction of the CR bound by the control strain.
Homology modelling.
BLAST (Altschul et al., 1997) was used to align the primary sequence of WspR and the monomeric structure of C. crescentus PleD (Chan et al., 2004; PDB code 1w25B). Twenty-two residues were deleted from the sequence of WspR and gaps were added where necessary to give the optimum alignment (Supplementary Fig. S1), which was then used as a template for SWISS-MODEL analysis (Schwede et al., 2003). The resulting WspR homology model was visualized and manipulated using DeepView/Swiss-PdbViewer version 3.7 (Guex & Peitsch, 1997). The reliability of the homology model was assessed by comparison of the predicted secondary structures of WspR and PleD with the known secondary structure of PleD (Chan et al., 2004), using PROFsec from the Predict Protein database (Rost et al., 2004).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
(c) also corresponded to WspR and was likely to be a degradation fragment of WspR.
|
). Approximately equal amounts of WspR were detected in SM and WS cell lysates, supporting the hypothesis that the WS wrinkled phenotype does not arise primarily as the result of WspR overexpression, in agreement with previous observations (Spiers et al., 2002).
Given the sequence homology (74 % identity) between P. fluorescens SBW25 and P. aeruginosa PAO1 WspR (recently shown to function as a DGC by Hickman et al., 2005) and the fact that WspR R129C complements PleD activity in C. crescentus (Aldridge et al., 2003), it was reasonable to assume that WspR is a DGC, catalysing the formation of c-di-GMP from two molecules of GTP. Nevertheless, it was important to independently verify DGC activity for SBW25 WspR. N- and C-terminal His6-WspR were assayed for DGC activity, as described by Paul et al. (2004). Radiolabelled GTP was incubated with purified His6-WspR, and samples were removed at regular intervals. When the nucleotide samples were separated by TLC, a spot corresponding to c-di-GMP was generated over time by C-terminal His6-WspR, determined by comparison with a c-di-GMP control sample (Christen et al., 2005) (Fig. 1a). No DGC activity was detected for N-terminal His6-WspR (data not shown), possibly as a result of disruption of the active state by the His6-tag, similar to that seen for N-terminal His6-PleD (R. Paul and U. Jenal, personal communication).
To investigate the effects of the WS background on WspR DGC activity, cell extracts from overnight cultures of SM, WS and isogenic strains lacking WspR (SM wspR and WS wspR), and containing pME6010 and pME6010-wspR as indicated, were assayed for DGC activity (Fig. 1b) in the same manner as above. A slow-running spot, thought to correspond to c-di-GMP, on the basis of comparison with previous studies (Paul et al., 2004), was generated in WS cell lysates expressing wspR both from pME6010 and the chromosome. No c-di-GMP generation was seen for the other samples tested. Hence, the WS cell extract activated WspR DGC activity, whereas the SM cell extract did not.
Mutagenesis of WspR
Nineteen wspR mutant alleles were produced by PSM (Hallet et al., 1997) (see Supplementary Table S1). The 5 aa insert resulting from PSM differed among mutants, depending on the specific site of Tn4430 insertion into wspR. However, the disruption to protein tertiary structure caused by the insertion of 5 aa into the protein was considered likely to outweigh any effects caused by variation in the amino acid sequence of the insert, and this latter difference was discounted (Hayes & Hallet, 2000).
To complement the PSM variants and fully mutagenize wspR, SOE-PCR (Ho et al., 1989) was used to introduce 25 GVPTK insertions at specific positions throughout wspR (Supplementary Table S2). The resulting library (44 variants) contained approximately one insert for every 7–8 aa of WspR, ensuring a high probability of disrupting most elements of secondary structure present in the protein.
To determine whether soluble protein was expressed from each pME6010-wspR mutant allele, all alleles were expressed separately in SM wspR, to allow WspR produced from pME6010 to be detected by Western blotting. Forty wspR mutant alleles produced detectable amounts of protein, but no WspR was detected for variants with inserts at residue positions 99, 182, 246 and 264 (Fig. 2), suggesting that either an undetectably small amount of WspR was present in each case, or that these variants produced highly unstable protein that was rapidly degraded during sample preparation. However, when expressed in WS, these four variants ameliorated the wrinkled colony morphology (see below). Given clear phenotypic effects, these alleles were included in subsequent analyses. No correlation was seen between protein levels and the phenotypic effects of expression, suggesting that the amount of protein present was not a major factor in determining the WspR activity state.
|
). WspR is activated by the WspE kinase (Bantinaki et al., 2007), therefore, development of WS morphology in strains lacking the wsp operon (e.g. SM wspA–wspF) indicates the presence of constitutively active WspR. In cases where a WS morphology develops in the presence of the Wsp machinery (e.g. in SM), but not in a wsp deletion background, the protein is signal-dependent and behaves like wild-type WspR. Inactive proteins have no effect in any SBW25 genotype, whilst dominant-negative WspR variants produce a smooth morphology in the WS genotype. To investigate whether the presence of chromosomal wspR has any effect on the behaviour of trans-expressed wspR variants, the morphological analysis was expanded to include several wspR deletion strains (e.g. SM wspA–wspR and SM/WS wspR).
To fully distinguish between all possible activity states, the wspR insert alleles were expressed in several SBW25 backgrounds: SM, WS, SM wspA–wspF, SM wspA–wspR, SM wspR and WS wspR, and their effects on colony morphology were determined. In general, changes in colony morphology conformed to predictions, and allowed us to assign one of the four activity states described previously to every tested WspR variant (Table 2, Fig. 3). Morphological analysis provided no evidence for interactions between WspR variants and chromosomal WspR; little morphological difference was seen upon WspR production in SM versus SM wspR, or SM wspA–wspF versus SM wspA–wspR.
|
|
) produced a wrinkled morphology in WS wspR, but had no effect in other tested strains. These variants appeared to have some residual activity; enough to recover the wrinkled morphology in the WS genotype, but not enough to induce a wrinkled morphology in SM. These variants indicate that less WspR activity is required to substitute for deleted chromosomal WspR than is required to induce a wrinkled morphology in an SM genotype. For simplicity, these strains were classed as inactive for the purpose of further analysis.
The emerging model of WspR function was refined through analysis of the effects of WspR variants on different aspects of biofilm formation. WspR is known to be essential for the enhanced levels of attachment seen in WS (Spiers et al., 2003). Therefore, the effect of wspR mutant allele expression on attachment in SBW25 strains was measured using a quantitative CV assay (Fig. 4). To distinguish between the effects of constitutively active and signal-dependent variants on attachment, the assay was repeated in SM and SM wspA–wspR backgrounds.
|
To determine whether the close correlation between attachment and colony morphology also applied to cellulose production, the effects of WspR variants on cellulose production in SM were determined quantitatively using a CR binding assay, as described by Spiers et al. (2003) (Fig. 5). [In addition, the presence or absence of cellulose in each sample was determined qualitatively by staining with calcofluor (CF; Sigma). No major discrepancies emerged between cellulose seen upon CF staining, and the results of the CR assay (data not shown)]. In general, increased cellulose production relative to SM pME6010 was seen for those strains displaying WS colony morphology in SM (Table 2). High levels of cellulose production (relative to the SM pME6010-wspR control) were observed for several N-terminal and linker region variants, whilst C-terminal variants did not affect cellulose production. The CR assay results were then compared with the results for attachment in the SM genotype. Several variants were identified that strongly upregulated cell attachment, but did not effect a corresponding large increase in cellulose production (e.g. WspR6). These results were consistent with several hypotheses, discussed in more detail below.
|
) made possible the use of homology modelling software to produce a putative 3D structure for WspR (Fig. 6). The PleD structure contains two response-regulator receiver domains and a C-terminal GGDEF domain, which has a fold similar to that of the adenylyl cyclase catalytic domain (Zhang et al., 1997), consistent with the earlier predictions of Pei & Grishin (2001). WspR shares significant sequence homology with PleD across the second response-regulator receiver (R2) and GGDEF (DGC) domains, and the PleD and WspR DGC domains are functionally interchangeable in C. crescentus (Aldridge et al., 2003). Therefore, the PleD crystal structure represented the best candidate upon which to base the model. The program SWISS-MODEL (Schwede et al., 2003) was used to produce a homology model for WspR (Fig. 6). This model was then used to refine the structure–function models of WspR. The insertion positions of the N-terminal and linker region WspR variants were mapped onto the WspR homology model (Fig. 6), and the relationship between the position of each insert and its effect on function was analysed.
|
). Substitution variants were produced as a random pME6010-wspR-SacI plasmid pool, in which all residue substitutions (except tryptophan) were possible throughout RYGGEEF. This pool was used to transform SM, and WspR substitution variants were recovered and checked. In total, 76 different WspR substitution variants were isolated. Seventy-five of these were from colonies displaying smooth colony morphology (i.e. those in which the WspR variant could not induce the WS phenotype) (Table 3). One additional variant, WspRE250D, produced a wrinkled colony phenotype in SM, suggesting that the protein was active in this case.
|
|
). |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
), and the WS phenotype was shown to stimulate WspR DGC activity in cell extracts, in proportion to the level of wspR gene expression. This provided a definite mechanism for WspR function, and suggests that the WS phenotype may arise (at least in part) as the result of increased levels of WspR activation and concomitant c-di-GMP production.
Secondly, to probe the mechanisms of WspR activation and regulation, systematic mutagenesis was used to disrupt WspR at positions throughout its length. The effects of the resulting WspR insert library on SBW25 colony morphology, attachment and cellulose production were analysed, and used to determine the activity state of the various WspR variants. Several distinct observations were made of the relationship between WspR activity and insert position. Firstly, inserts in the C terminus (residues 164–333) of WspR always eliminated enzymic activity. Whilst several insertions in the N terminus also eliminated enzymic function, the same sensitivity to mutation was not seen in the domain as a whole. Secondly, a number of constitutively active WspR variants were found with insertions in the N terminus (residues 1–130), and insertions in the linker region (residues 131–163) produced constitutively active enzymes in every case (Fig. 3). To our knowledge, WspR is the only GGDEF protein for which a comprehensive structure–function analysis of this kind has been undertaken.
Detailed structure–function analysis provides significant insight into the means of protein regulation, and in this case allowed the construction of a model for WspR activation (Fig. 8). In this model, the C-terminal effector domain is inhibited in the inactive state by the N-terminal domain. Regulation through effector-domain inhibition occurs in a number of response regulators, including E. coli NarL (Baikalov et al., 1996; Eldridge et al., 2002) and S. typhimurium CheB (Djordjevic et al., 1998). Phosphorylation (Fig. 8a), or mutation of the N-terminal domain (Fig. 8b), leads to the release of inhibition. This model expands on the observations of Goymer et al. (2006), who have noted that mutations in or near the linker region of WspR mimick the phosphorylated state of the protein. Biochemical (Ryjenkov et al., 2005) and structural (Chan et al., 2004) evidence suggests that DGCs function as dimers (although no evidence for WspR dimerization was found in this study). If WspR functions as a dimer, disruption of the inactive form of the protein might lead to exposure of the dimer interface, and hence activation (Fig. 8b). These data do not rule out the possibility of an alternative activation mechanism, whereby the phosphorylated N-terminal domain stabilizes the DGC-activated conformation. However, it is unlikely that the precise interactions required by such a mechanism would be reproduced by every activating insertion tested, strongly suggesting that the mechanism of WspR activation proceeds via release of effector-domain repression.
|
for suppression of the WS wrinkled morphology by the isolated WspR N terminus. Alternatively, inactive heterodimers, formed between inactive WspR variants and chromosomal wild-type WspR may interrupt Wsp signalling at the level of WspR. Neither model of inhibition can be ruled out at this stage.
When the positions of the 5 aa inserts were plotted onto the N terminus and linker region of the homology model (Fig. 6), they clustered according to their effects on WspR activity, suggesting that each group affected WspR function via a common mechanism. N-terminus insertions producing constitutively active proteins were found at the interface between the N- and C-terminal domains, and in adjoining loops and helices. These insertions presumably activated WspR via disruption of the interface between the two domains, a mechanism common to many response-regulator activation mechanisms (Eldridge et al., 2002; Park et al., 2002; Muchova et al., 2004). A comparison of the domain interaction surfaces of four full-length, two-domain response-regulator crystal structures (Robinson et al., 2003) suggests that the activation mechanisms of the different proteins, NarL (Baikalov et al., 1996), CheB (Djordjevic et al., 1998), DrrD (Buckler et al., 2002) and DrrB (Robinson et al., 2003), vary with the extent of their interdomain interfaces. For example, comparison of the DrrB and DrrD crystal structures suggests that different levels of intramolecular communication between the N- and C-terminal domains lead to significant differences in effector-domain regulation in each case (Robinson et al., 2003).
Several activating insertions were clustered in the -helical section of the linker region (Fig. 6). Mutational analysis of the linker region between the receiver and effector domains of OmpR by Mattison et al. (2002) has shown that the composition of the linker region affects the interplay of the two domains, and hence plays a role in transducing the receiver signal to the effector. In addition, the importance of interdomain linkers in the determination of the response-regulator activation mechanism has been shown using OmpR–PhoB hybrid proteins (Walthers et al., 2003). The linker region appears to play an important role in regulation of WspR activation. However, studies with NarL have assigned little functional significance to the linker region (Eldridge et al., 2002), suggesting that linker-mediated regulation is not a universal feature of response regulators.
Insertions producing inactive and dominant-negative WspR variants were spread over the surface of the model, furthest from the domain interface, surrounding the predicted phosphorylation site (Fig. 6). N-terminal dominant-negative insertions may have disrupted protein folding without affecting the domain interface or promoting dimerization, in the same manner as was seen for the activating insertions. Alternatively, these insertions could have affected WspR interaction with the Wsp chemosensory kinase.
For the purposes of this study, two morphological states (SM and WS), and four activity states for WspR were described. These distinctions are a necessary simplification of the situation in vivo that allows us to propose models for WspR function. Both in this study and elsewhere (Römling et al., 2005), it is evident that the phenotypic consequences of changes in c-di-GMP levels are highly complex and rely on a number of interrelated factors. For example, the wrinkled phenotype caused by overexpression of wild-type wspR in the SM background was accompanied by little increase in cellulose production or attachment compared to the LSWS phenotype, yet both were classified here as WS. (In the first case, most WspR in the cell was thought to be in the inactive form. In the second, less WspR was present but a far larger proportion of this was activated). The physiological relevance of overexpression phenotypes should not be overstated. Nonetheless, within the context of isolated, controlled experiments, overexpression is a valid tool for the analysis of GGDEF domain proteins.
Quantitative analysis of the effects of WspR variants showed that generation of the wrinkled phenotype in SM strains was associated with increases in both cell attachment and cellulose production (Figs 3, 4 and 5). An important exception, however, was the discovery of several WspR variants whose expression produced little effect on cellulose production, but greatly stimulated cell attachment, relative to the other WspR variants tested (e.g. WspR6 and WspR120). In addition, greatly decreased levels of cellulose production in WS were observed upon expression of the 20 dominant-negative wspR substitution mutant alleles. However, cell attachment to glass in WS strains expressing these wspR mutant alleles was not repressed, relative to that seen with the wild-type control (Fig. 7).
There are several possible explanations for these data. The first is that WspR regulates attachment and cellulose production via distinct and separable mechanisms, possibly regulating one function through DGC activity and another via protein–protein interactions. Andrade et al. (2006) have suggested that some GGDEF proteins may function in this way. Secondly, if wspR functions solely via DGC activity, and if upregulation of attachment requires lower levels of c-di-GMP than does cellulose production, then WspR variants that upregulate attachment only (e.g. WspR6) presumably do not produce enough c-di-GMP to trigger cellulose production. Likewise, inhibition of DGC activity by dominant-negative WspR substitution variants (e.g. WspRR246L) would reduce c-di-GMP levels to the point at which cellulose production, but not attachment, is repressed. Finally, the Wsp system may modulate different processes through two or more response regulators, including WspR. Repression of DGC activity (and hence cellulose production) in WS by dominant-negative WspR variants would have no effect on attachment in this scenario. If a second Wsp-controlled response regulator existed, we would expect to see high levels of attachment for WS wspR relative to that for SM wspR. In fact, attachment levels in this strain are comparable to those of SM wspR (Supplementary Fig. S2), strongly arguing against this model. Biochemical analysis of DGC levels for the different classes of WspR variants is required to confirm which explanation of the data is correct. Alternatively, the discovery of a hypothetical WspR variant that upregulates cellulose production whilst having no effect on attachment would provide strong support for the first model.
For several wspR insert variants (e.g. SOEwspR130), higher levels of attachment were observed upon expression in SM wspA–wspR than in SM (Fig. 4). This may be related to the loss of pellicle structure in SM wspA–wspR compared to SM, allowing more cells to attach at the meniscus than would be possible for strains producing a structured biofilm. Similarly, repression of cellulose production, and hence loss of pellicle structure, may explain why several dominant-negative WspR substitution variants (e.g. SUBwspRR246F) produced increased WS attachment levels compared to those of wild-type WspR (Fig. 7). It is probable that biofilm formation in SBW25 consists of numerous interconnected processes (Spiers et al., 2003), and care must be taken when interpreting the phenotypic data.
Finally, the RYGGEEF motif of WspR was investigated using site-specific mutagenesis, and was shown to be highly sensitive to mutation. Substitution of residues 246–252 (RYGGEEF) led to a loss of protein function in almost every case tested. Of the 76 wspR substitution alleles produced, only one produced an active protein (wrinkled morphology in SM). As WspR functions as a DGC, it is reasonable to suggest that the level of functional conservation seen for the WspR RYGGEEF motif may also apply to other DGCs; that this motif (GGD/EEF) is absolutely required for DGC activity. [The GGD/EEF motif is not always required for domain function, however, as functional proteins containing GGDEF domains with degenerate active site motifs have been identified (Christen et al., 2005; Kazmierczak et al., 2006)]. Mutagenesis of single residues in RYGGD/EEF has eliminated protein function in numerous species (Paul et al., 2004; Simm et al., 2004; Kirillina et al., 2004; Garcia et al., 2004; Goymer et al., 2006), and the crystal structure of PleD places the RYGGEEF motif close to the ribose group and alpha-phosphate of bound c-di-GMP, suggesting a critical role in nucleotide binding or catalysis (Chan et al., 2004).
The single active WspR variant, WspRE250D, contained the motif RYGGDEF, rather than RYGGEEF. An analysis of sequence conservation among GGDEF domains (Galperin et al., 2001) has indicated that aspartate and glutamate are found with similar frequency at position 250. It is possible that the conformational constraints on the first glutamate in the RYGGEEF motif are not so pronounced as for the second, allowing aspartate or glutamate to be present. Interestingly, Tyr 247 was absolutely conserved in this study, despite the residue at this position showing substantial variation across different GGDEF domains (Galperin et al., 2001). When three WspR substitution variants were chosen to represent each position of the RYGGEEF motif and analysed in more depth, almost every variant tested produced a dominant-negative effect on WS morphology (Table 3). The only exception was WspRY247F, which produced an inactive protein. The structural similarity between tyrosine and phenylalanine suggests that an undefined protein misfolding event, avoided in the case of a structurally conserved substitution such as that in WspRY247F, was responsible for the observed dominant-negative effect of the tested residue substitutions on WspR activity.
This work sheds light on several previously unanswered aspects of the WspR structure–function relationship. WspR functions as a DGC, whose enzyme activity is stimulated in the WS genotype compared with the (wild-type) SM genotype. Upon phosphorylation, or disruption of the linker region or the interdomain interface, activation proceeds via release of effector-domain repression by the receiver domain. Quantitative analysis of biofilm formation has raised questions concerning the downstream regulatory mechanisms of WspR. Finally, the RYGGEEF motif of WspR is highly conserved functionally, with almost every substitution tested abolishing function in a dominant-negative fashion. As the time and effort required to produce high-quality structural models of proteins continues to decrease, this study underlines the continued utility of systematic mutagenesis to structure–function analyses.
|
ACKNOWLEDGEMENTS |
|---|
Edited by: J. Anné
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Altschul, S. F., Madden, T. L., Schaffer, A. A., Zhang, J., Zhang, Z., Miller, W. & Lipman, D. J. (1997). Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25, 3389–3402.
Andrade, M. O., Alegria, M. C., Guzzo, C. R., Docena, C., Rosa, M. C., Ramos, C. H. & Farah, C. S. (2006). The HD-GYP domain of RpfG mediates a direct linkage between the Rpf quorum-sensing pathway and a subset of diguanylate cyclase proteins in the phytopathogen Xanthomonas axonopodis pv citri. Mol Microbiol 62, 537–551.[CrossRef][Medline]
Baikalov, I., Schroder, I., Kaczor-Grzeskowaik, M., Grzeskowaik, K., Gunsalus, R. P. & Dickerson, R. E. (1996). Structure of the Escherichia coli response regulator NarL. Biochemistry 35, 11053–11061.[CrossRef][Medline]
Bantinaki, E., Kassen, R., Knight, C. G., Robinson, Z., Spiers, A. J. & Rainey, P. B. (2007). Adaptive divergence in experimental populations of Pseudomonas fluorescens. III. Mutational origins of wrinkly spreader diversity. Genetics (in press).
Bobrov, A. G., Kirillina, O. & Perry, R. D. (2005). The phosphodiesterase activity of the HmsP EAL domain is required for negative regulation of biofilm formation in Yersinia pestis. FEMS Microbiol Lett 247, 123–130.[CrossRef][Medline]
Boles, B. R. & McCarter, L. L. (2002). Vibrio parahaemolyticus scrABC, a novel operon affecting swarming and capsular polysaccharide regulation. J Bacteriol 184, 5946–5954.
Bomchil, N., Watnick, P. & Kolter, R. (2003). Identification and characterization of a Vibrio cholerae gene, mbaA, involved in maintenance of biofilm architecture. J Bacteriol 185, 1384–139
