Wrinkly spreader (WS) genotypes evolve repeatedly in model Pseudomonas populations undergoing adaptive radiation. Previous work identified genes contributing to the evolutionary success of WS. Here we scrutinize the GGDEF response regulator protein WspR and show that it is both necessary and sufficient for WS. Activation of WspR occurs by phosphorylation and different levels of activation generate phenotypic differences among WS genotypes. Five alleles of wspR, each encoding a protein with a single amino acid substitution, were generated by mutagenesis. Two alleles are constitutively active and cause the ancestral genotype to develop a WS phenotype; the phenotypic effects are allele specific and independent of phosphorylation. Three alleles contain changes in the GGDEF domain and when overexpressed in WS cause reversion to the ancestral phenotype. Ability to mimic this effect by overexpression of a liberated N-terminal domain shows that in WS, regulatory components upstream of WspR are overactive. To connect changes at the nucleotide level with fitness, the effects of variant alleles were examined in both structured and unstructured environments: alleles had adaptive and deleterious effects with trade-offs evident across environments. Despite the proclivity of mutations within wspR to generate WS, sequence analysis of wspR from 53 independently obtained WS showed no evidence of sequence change in this gene.
THE diversity of life poses a myriad of problems for biologists; foremost among these are the causes of ecological and morphological diversity. Concepts that stem from Darwin's theory of evolution by natural selection have produced an ecological theory of diversification (Darwin 1890; Lack 1947; Dobzhansky 1951; Simpson 1953; Schluter 2000), but a corresponding genetic theory remains underdeveloped. Necessary for progress is an understanding of how genotype, phenotype, and fitness are connected through evolutionary time (Orr 2005): in short, a genetical theory of evolutionary development (West-Eberhard 2003).
Our work has focused on the ecological and genetic causes of diversity in simple laboratory populations of Pseudomonas fluorescens (strain SBW25). When propagated in spatially structured microcosms, populations of P. fluorescens rapidly diversify, giving rise to an extensive range of niche specialist genotypes (Rainey and Travisano 1998). One common class of niche specialist colonizes the air–liquid interface of static broth microcosms and produces colonies with a characteristic wrinkled morphology on agar plates; these mutants are referred to as “wrinkly spreaders” (WS). WS morphs (of which there are many morphologically distinct types) arise by spontaneous mutation from the ancestral genotype and show a significant negative frequency-dependent fitness advantage. Their ability to colonize the air–liquid interface stems from the formation of a self-supporting mat that is a product of the combined (cooperative) activities of the individual cells (Rainey and Rainey 2003).
To define the genetic and phenotypic causes of WS success a top-down analysis of a representative WS genotype, the “large spreading wrinkly spreader” (LSWS), was initiated. The LSWS genotype was subjected to a suppressor analysis, and mutants defective in expression of the wrinkled colony morphology and compromised in their ability to colonize the air–liquid interface were identified (Spiers et al. 2002; Gehrig 2005).
Among the genes contributing to the LSWS phenotype is a set of 10 open reading frames, collectively termed wss, which form a single operon and together produce an acetylated cellulose polymer (Spiers et al. 2003). In the ancestral genotype the polymer is barely detectable; however, in the LSWS it is overproduced and is the primary phenotypic target of selection (Spiers et al. 2002). Sequence and other analyses of the wss operon from LSWS ruled out the possibility that a mutation within this set of genes was the cause of LSWS evolution (Spiers et al. 2002). Nevertheless, given that polymer production is tightly regulated in the ancestral genotype, but constitutively overproduced in the LSWS genotype, it seemed likely that the ultimate mutational causes of LSWS might reside in regulators of cellulose production.
Regulation of cellulose is complex and not fully understood, but pioneering studies in Acetobacter xylinus (now Gluconacetobacter xylinus) showed that bis-(3′-5′)-cyclic-dimeric-guanosine monophosphate (c-di-GMP) is an allosteric activator of cellulose biosynthetic enzymes (Ross et al. 1987). Subsequent work implicated proteins containing a conserved Gly–Gly–Asp/Glu–Glu–Phe motif—a motif that defines the ubiquitous GGDEF domain family of proteins—as diguanylate cyclases (DGC) responsible for the biosynthesis of c-di-GMP (Ross et al. 1987; Tal et al. 1998; Galperin et al. 2001; Pei and Grishin 2001). Recent work has confirmed these predictions (Paul et al. 2004; Hickman et al. 2005; Ryjenkov et al. 2005).
Among the genes that determine the wrinkled spreading phenotype of LSWS is the GGDEF response regulator WspR. WspR is the seventh gene of a complex chemosensory signal transduction pathway that controls extracellular polysaccharide production and biofilm formation in P. aeruginosa (cellulose is not encoded by the P. aeruginosa genome) (D'Argenio et al. 2002; Hickman et al. 2005). In the LSWS genotype of P. fluorescens polar mutations at any point within the wsp operon abolish the biosynthesis of cellulose (Spiers et al. 2002), the attachment factor (Spiers et al. 2003), the wrinkled colony morphology of LSWS, and the ability to colonize the air–liquid interface (Spiers et al. 2002).
Several features suggest the importance of wspR in evolution and expression of the WS phenotype: first, when inactivated in LSWS all traces of the WS phenotype are abolished (Spiers et al. 2002); second, the GGDEF domain is firmly implicated as a DGC [the GGDEF domain of WspR from P. fluorescens is interchangeable with the GGDEF domain of PleD from Caulobacter crescentus (Aldridge et al. 2003), and PleD has proven DGC activity (Paul et al. 2004)]; and third, the receiver domain is CheY-like and contains all conserved residues diagnostic of CheY-like response regulators (Stock et al. 2000), including Asp67, the predicted site of phosphorylation. This led to predictions as to the means by which WspR might be activated, namely, by phosphorylation at Asp67. Indeed, mutations are known within the N-terminal domains of response regulators that cause constitutive activation of the output domain (Hoch and Silhavy 1995; Chamnongpol and Groisman 2000; Da Re et al. 2002; Aldridge et al. 2003; Siam and Marczynski 2003; Smith et al. 2004). We therefore considered it plausible that mutations might arise within WspR to cause WS—that such mutations might cause constitutive activation of the GGDEF domain, overproduction of c-di-GMP, and concomitant overproduction of acetylated cellulose polymer.
Here we report the results of a study in which we scrutinize wspR because of its role in the evolution of WS. We manipulate wspR in vitro to gain insight into the function of the protein and then introduce allelic variants into both ancestral and derived (WS) genotypes so as to explore the morphological and fitness consequences of in vitro manipulations. Our data show that wspR is a proximate (but not ultimate) cause of the WS phenotype.
MATERIALS AND METHODS
Bacterial strains, growth conditions, and manipulation:
The ancestral (wild-type) “smooth” (SM) strain is P. fluorescens SBW25 and was isolated from the leaf of a sugar beet plant grown at the University Farm, Wytham, Oxford, in 1989 (Rainey and Bailey 1996). The ancestral strain was placed at −80° immediately after isolation to minimize lab adaptation. The niche-specialist wrinkly spreader (LSWS) genotype (PR1200) was derived from the ancestral genotype following 3 days of selection in a spatially structured microcosm (Spiers et al. 2002). PR1252 (SBW25ΔpanB) is a pantothenate-requiring auxotroph of SM and has a clean deletion of the entire panB gene (Rainey 1999); it is phenotypically SM. The mutation is neutral in a pantothenate-replete environment (Spiers et al. 2002). PR3521 (WSΔpanB) was derived from SBW25ΔpanB following 3 days of selection in a spatially structured microcosm; it is phenotypically WS and its fitness is equivalent to LSWS.
P. fluorescens strains were cultured at 28° in Luria–Bertani (LB) (Miller 1972), Pseudomonas agar F (PAF; Difco), or King's medium B (KB) (King et al. 1954). Escherichia coli DH5α, XL1-Blue MRF′ (Stratagene), and S17-1λpir (Simon et al. 1983) were cultured at 37° in LB. Antibiotics were used at the following concentrations: ampicillin, 100 μg ml−1; kanamycin, 75 μg ml−1 (50 μg ml−1 for E. coli); and tetracycline, 25 μg ml−1. Pantothenate was added to media at a final concentration of 0.24% (w/v) where required. CFC (Oxoid) was added to media to select for P. fluorescens strains following conjugation. Congo Red (CR) was used at a final concentration of 0.001% (w/v) in KB or LB agar without NaCl (NaCl causes CR to precipitate). Calcofluor (Sigma Fluorescent Brightener 28) was used in LB at 1–10 μm and cells or the polymer incubated with the stain for 30 min before examination by fluorescent microscopy.
Plasmid DNA was introduced into E. coli and P. fluorescens by transformation (electroporation) or conjugation using standard procedures. The helper plasmid pRK2013 (Figurski and Helinski 1979) was used to facilitate transfer of plasmids between E. coli and P. fluorescens.
Molecular biology techniques:
Plasmid DNA was isolated from E. coli using QIAprep plasmid kits (QIAGEN). Standard agarose gel electrophoresis and other recombinant DNA techniques were performed according to Sambrook et al. (1989). DNA fragments were recovered from agarose using a QIAEX II gel extraction kit (QIAGEN). pSK+ (pBluescriptII SK+; Stratagene) and pVSP61 (Loper and Lindow 1994) were used for cloning of wspR alleles. The neomycin phosphotransferase gene plus promoter were derived from plasmid pCPP2988 (Alfano et al. 1996). Allelic exchange was performed using the suicide reporter plasmid pUIC3 (Rainey 1999). DNA sequencing was carried out on an ABI310 (Perkin-Elmer) automated sequencer. Oligonucleotide primers used for sequencing and construction of wspR alleles are available upon request.
Construction of deletion mutants:
Deletion mutants were constructed as described previously (Rainey 1999). First, ∼500 bp of DNA flanking either side of the gene (or genes) to be deleted were spliced together by PCR (Horton et al. 1989). Oligonucleotide primers were designed on the basis of the available SBW25 genome sequence (Sanger Centre). The spliced DNA was cloned into pBluescript, from where it was sequenced to check for errors. The spliced fragment was then excised as a BglII–SpeI fragment and ligated into the suicide vector pUIC-3 (Rainey 1999).
To introduce the spliced DNA [the DNA flanking the gene(s) to be deleted] into the SBW25 chromosome a two-step allelic replacement strategy was used (Rainey 1999). pUIC-3 containing the DNA to be exchanged into the genome was mobilized into P. fluorescens by conjugation. Integration into the chromosome by a single homologous recombination event was selected by plating on LB agar containing tetracycline and X-gal. To complete the allelic exchange, single, blue-colored, tetracycline-resistant colonies were propagated overnight in LB broth without antibiotic selection. The cells were then harvested and plated on LB agar containing X-gal and white colonies screened for loss of tetracycline resistance. White colonies were further checked by PCR to identify deletion mutants.
Mutagenesis of wspR:
The wspR gene was amplified via a standard polymerase chain reaction using nonproofreading (Taq) polymerase. PCR products were cloned into pVSP61 between BamHI and EcoRI and transformed into E. coli DH5α. Site-directed mutants were constructed using specifically designed oligonucleotides. Mutagenesis of wspR was achieved by PCR, but with the reaction mixture containing a fivefold greater concentration of Mg2+ than recommended by the manufacturer.
CR binding assay:
Cellulose expression was measured using a CR binding assay as described by Spiers et al. (2003). A total of 10 μl drops of overnight cultures were grown on KB agar plates for 24 hr at 28°. Colonies were resuspended in 0.005% (w/v) CR and incubated for 2 hr at 37°. Cells were then removed by centrifugation. The amount of CR remaining in the supernatant was determined by measurement of A490 and comparison with appropriate CR standards to obtain milligrams of CR. CR binding was expressed as milligrams of CR .
Calcofluor staining of cellulose:
Cellulose production was additionally measured using a Calcofluor binding assay adapted from Spiers et al. (2003). A total of 5 μl drops of overnight cultures were grown on 25-ml KB agar plates for 24 hr at 28°. Colonies were scraped from plates and then transferred into 1 μm Fluorescent Brightener 28 (Calcofluor, Sigma) solution. Samples were incubated at 28° for 30 min, and stained cell material was viewed under UV illumination.
Fitness of genotypes:
Competitive fitness of SM and LSWS genotypes carrying alleles of wspR was determined by directed competition between each strain and panB deletion derivatives of both the ancestral genotype (PR1252) and WS (PR3521) in spatially structured and unstructured microcosms (Rainey and Travisano 1998). Relative fitness was calculated as the ratio of the Malthusian parameters of the two strains being compared (Lenski et al. 1991). Cultures were founded with 105 cells of each competitor and the ratio of competing genotypes was determined after 24 hr growth on vitamin-free KB agar supplemented with 4.8 × 10−6% pantothenic acid; on this medium the pantothenate-marked strain is readily distinguished by its greatly reduced size. Antibiotic selection was not maintained during competitions; preliminary experiments showed the rate of segregational loss in the absence of antibiotics was below the level of detection. All cultures were whirlimixed for 60 sec before dilution plating to maximally disperse clumped cells.
In silico analysis of WspR:
In prokaryotes signal transduction occurs via highly conserved His–Asp phospho-relay systems (Hoch and Silhavy 1995). At their simplest, His–Asp relays compose a sensor protein and a response regulator: the sensor protein detects an environmental stimulus, autophosphorylates at a conserved histidine residue, and then transfers the phosphoryl group to a conserved aspartate residue on the input domain of the response regulator. The phosphorylation state of the response regulator in turn controls its output.
The WspR protein (AAL71852) is a response regulator and output component of the Wsp chemosensory (signal transduction) operon (AY074937). WspR is 333 amino acids in length and is composed of two clearly defined domains separated by a linker (Figure 1). Residues 1–131 compose a response regulator receiver domain with homology to CheY from E. coli. Included within this domain are all conserved residues diagnostic of CheY-like response regulators, including Asp67, the predicted site of phosphorylation, and Asp22, Asp23, Ser97, and Lys117, a set of conserved residues that form a cluster adjacent to the active site (Stock et al. 2000). The C-terminal output domain extends from residues 164 through 333 and contains the conserved GGDEF domain (plus associated residues) indicative of diguanylate cyclase (DGC) function (Pei and Grishin 2001; Galperin 2004; Paul et al. 2004; Romling et al. 2005). Note that WspR from P. fluorescens SBW25 has a glutamine residue at the third position of the conserved motif instead of an aspartate residue rendering its conserved motif GGEEF.
The role of wspR in development of the LSWS phenotype:
Previous analysis of WS-4, a mutant of LSWS defective in expression of the wrinkled colony morphology and unable to colonize the air–liquid interface of a static broth microcosm (Spiers et al. 2002, 2003), showed that the cause of the mutant phenotype is a transposon insertion in wspR (Spiers et al. 2002). To verify the central role of this gene in the LSWS phenotype, the complete wspR open reading frame was deleted from the genome of LSWS by a two-step allelic exchange strategy. The resulting wspR deletion mutant (LSWS ΔwspR) was ancestral-like in appearance and indistinguishable from WS-4; it was unable to colonize the air–liquid interface of static broth microcosms and unable to produce cellulose, as evidenced by yellow (not orange) colony color on agar plates containing CR and absence of fluorescent material following staining with Calcofluor. All traits characteristic of LSWS were restored when a cloned copy of wspR (expressed from the kanamycin promoter in pVSP61, pVSP61-wspR) was introduced into LSWS ΔwspR; wspR is thus necessary and sufficient to generate the WS morphology in the LSWS genotype.
WspR exerts its effects on acetylated cellulose production at a post-transcriptional level:
The similarity between WspR and known DGCs suggests that WspR exerts its primary effect on cellulose production at a post-transcriptional level. To obtain evidence of post-transcriptional activity the amount of CR-binding material produced by three different strains was compared. CR binds (1–4) β-d-glucopyranosyl units, basic or neutral extracellular polysaccharides, and some proteins and can be readily assayed (Spiers et al. 2003). The ancestral SM genotype produced 0.158 ± 0.015 mg (errors are 95% confidence intervals) of CR-binding material per OD600, whereas a strain lacking wspR (SM ΔwspR) produced significantly less (0.134 ± 0.013). This reduction is consistent with the known regulatory role of WspR in cellulose biosynthesis (Spiers et al. 2003). If the reduction is due to decreased wss transcription, i.e., WspR is a transcriptional regulator, then the defect should be readily reversed by increasing the level of wss transcription. [We have previously shown that the ancestral strain (with intact wspR) produces excess CR-binding material when wss transcription is placed under control of the nptII promoter (Spiers et al. 2002).] Conversely, if WspR exerts its effects at a post-transcriptional level (for example, by providing an essential cofactor for cellulose biosynthetic enzymes) then increasing wss transcription in a strain lacking wspR should not affect the production of cellulose. To distinguish between these two possibilities a constitutive nptII promoter was introduced into the chromosome of the wspR deletion strain and placed immediately upstream of (and in place of) the wss promoter (see Spiers et al. 2002). The amount of CR-binding material produced by this strain was 0.132 ± 0.017 mg of CR-binding material per OD600 and not different from the amount of CR-binding material produced by the wspR deletion strain. One-way ANOVA showed a significant difference among the three strains (F[2,12] = 6.788, P < 0.01) and comparisons with the ancestral genotype using Dunnett's test showed that both wspR mutant strains produced significantly less CR-binding material than the ancestral strain (P < 0.05). This result is consistent with predictions, namely, that WspR exerts its primary effect at a post-transcriptional level.
Overexpression of WspR in the ancestral SM genotype causes development of the WS phenotype:
Given the strong similarity between WspR and other response regulators we considered the possibility that the primary effect of the causal LSWS mutation might be activation of WspR with concomitant effects on production of the allosteric activator of cellulose synthase (c-di-GMP). If so, then conversion of the ancestral SM genotype to WS might be achieved by increasing the dosage of WspR. [Increasing the dosage of a response regulator can mimic the effect of activation by phosphorylation (Hoch and Silhavy 1995).] Accordingly, pVSP61-wspR was introduced into the ancestral SM genotype; the resulting transconjugant colonies displayed a wrinkled colony morphology, stained dark orange on media containing CR (Figure 2), fluoresced brightly after staining with Calcofluor, and colonized the air–liquid interface of static broth microcosms. When pVSP61-wspR was introduced into the LSWS genotype it led to the development of smaller, more tightly compacted WS colonies (Figure 2).
The conserved aspartate (Asp67) of WspR is required for activity:
Drawing upon similarities between WspR and both CheY and PleD (Aldridge et al. 2003), it seemed likely that activity of WspR would be dependent upon phosphorylation. To test this hypothesis a variant allele of wspR was produced in which nucleotide 199 was changed from G to A, resulting in replacement of the conserved aspartate at residue 67 with asparagine. This replacement was predicted to abolish the ability of WspR to be phosphorylated while causing minimum structural alteration. The phosphorylation-deficient wspR (wspR D67N) was cloned into pVSP61 (pVSP61-wspR D67N) and expressed in the ancestral SM genotype. [An antibody raised against WspR confirmed that WspR D67N was expressed (J. G. Malone, R. Williams, A. J. Spiers and P. B. Rainey, unpublished data)]. The colony morphology remained smooth and the strain was unable to colonize the air–liquid interface. Assays for cellulose were negative.
Allelic variants of WspR:
Having successfully altered the predicted site of phosphorylation and observed a phenotypic effect we sought additional allelic variants of wspR. To increase the likelihood of obtaining variant alleles, wspR was amplified by PCR in the presence of excess Mg2+. The resulting products were ligated into pVSP61, transformed into E. coli, and then transferred by conjugation into both ancestral SM and derived LSWS genotypes of P. fluorescens. While most pVSP61-wspR clones caused the ancestral SM genotype to convert to WS, two pVSP61-wspR clones were identified that caused development of WS-type colonies. In both instances the colonies were smaller in diameter and more reticulate than those generated by overexpression of pVSP61-wspR. In addition to the effects on colony morphology, these pVSP61-wspR clones caused the cells to overproduce cellulose and conferred the ability to colonize the air–liquid interface of static broth microcosms. A second set of pVSP61-wspR clones was identified that had no effect on the phenotype of the ancestral SM genotype, but when introduced into LSWS they caused the wrinkled morphology to revert to SM. In each case these LSWS (pVSP61-wspR) genotypes no longer produced orange-staining colonies on CR plates and were unable to colonize the air–liquid interface of static broth microcosms.
DNA sequence analysis of five pVSP61-wspR alleles, whose effects on SM and LSWS are shown in Figure 2, revealed, in each case, a single nucleotide change (transition) resulting in a single amino acid alteration to the WspR sequence. The specific changes and their phenotypic effects are recorded in Table 1. The two variants with dominant effects (WspR R129C and WspR D159G) harbor mutations in or immediately adjacent to the linker region, whereas alleles with dominant negative effects (WspR D206G, WspR F252S, and WspR G296R) carry mutations in the GGEEF output domain, including one (WspR F252S) with a mutation in the GGEEF motif itself.
Dominant wspR alleles do not require phosphorylation:
When overexpressed in trans in the ancestral genotype, the two alleles with mutations in the linker region caused development of tightly compacted WS colonies similar to that seen when wild-type wspR was introduced into—and overexpressed in—LSWS. This suggested that these mutations might activate WspR to a high level; indeed, mutations within response regulators that either prolong the lifetime of the activated (phosphorylated) state or mimic key features of the phosphorylated state are well known (Smith et al. 2004). Such mutations can have a number of specific effects; they can (1) increase the rate of autophosphorylation, (2) make the protein more resistant to dephosphorylation, or (3) mimic the conformational change ordinarily induced by phosphorylation. Activation in the first two scenarios requires the transfer of phosphoryl groups to the response regulator, whereas in the third scenario activation is likely to be phosphorylation independent.
To gain insight into mechanism, WspR R129C (as representative of the two dominant alleles) was subject to further manipulation. First, the D67N mutation, which removes the phosphorylation site, was created in the R129C (dominant) WspR background and the gene encoding this protein was introduced into the ancestral SM genotype. Upon overexpression WspR D67N R129C caused SM to develop the WS morphology and was indistinguishable from the effects of WspR R129C. This indicates that the R129C mutation causes phosphorylation-independent activation of the output domain.
According to accepted models of response regulator function, mutations in the input domain should be overridden by mutations in the C-terminal output domain; i.e., a mutation in the output domain should be epistatic to a mutation in the input domain (Hoch and Silhavy 1995). To see whether WspR conforms to this model a mutation was made in the output domain of WspR R129C; specifically, the sequence of WspR R129C was altered at nucleotide 667 (the T was changed to a C) to produce WspR R129C L226S (the leucine residue is highly conserved across all GGDEF proteins). When the gene encoding this protein was overexpressed in the ancestral SM genotype, colony morphology was unaffected; however, when overexpressed in LSWS the phenotype reverted to SM (i.e., it had a dominant-negative effect) demonstrating that the C-terminal mutation is epistatic to the R129C mutation. Interestingly, this result indicates that WspR R129C remains capable of phosphorylation.
Dominant-negative wspR alleles competitively inhibit phosphorylation of the wild-type protein:
Mutations in the output domain of WspR are likely to inactivate its enzymatic activity; this is especially so in the case of WspR F252S, which has a mutation in the highly conserved GGEEF motif (Kirillina et al. 2004). At the same time, the input domain and its capacity to accept phosphoryl groups is unlikely to be affected. As a consequence, when overexpressed in trans these alleles should have no effect on the phenotype of the ancestral SM genotype. However, if, in LSWS, a component that donates phosphoryl groups to WspR (e.g., a kinase) is overactive, then overexpression of a variant of WspR with a defective output domain is likely to “soak up” (quench) excess signal and lead to competitive inhibition of wild-type WspR, and thus a reversal of phenotype to SM.
If the dominant-negative alleles cause the LSWS phenotype to revert to SM by quenching then it should be possible to reproduce this effect by overexpressing solely the N-terminal domain of WspR; moreover, the dominant-negative effect should be dependent upon the phosphorylation site being intact. To this end a truncated variant of WspR was generated consisting of amino acid residues 1–141 (WspR Nterm) and cloned into pVSP61. When introduced into the ancestral SM genotype this truncated type had no observable phenotypic effect; however, when introduced into LSWS, pVSP61-WspR Nterm had a dominant-negative effect identical to that observed with the C-terminal mutants. Next, the D67N mutation was introduced into WspR Nterm; when expressed in LSWS no phenotypic effect was observed, consistent with the prediction that the effect of C-terminal mutants is to repress the activity of the chromosomal copy of WspR by competitive inhibition.
Given that the liberated receiver domain of WspR was functional (able to compete for activated phosphate) we considered the possibility that a liberated C-terminal domain might also be active. Such a domain would, if active, jam the WS phenotype into the on state. The C-terminal domain (starting at residue 158) was cloned into pVSP61 and introduced into both ancestral SM and LSWS genotypes, but no phenotypic effects were observed.
The Wsp pathway is required for activation of wild-type WspR but not of the dominant (constitutive) variants:
The contribution of upstream components of the Wsp signal transduction pathway on the effects of each of the wspR alleles was considered. Each of the alleles was introduced into SM ΔwspR and SM ΔwspABCDEFR. Alleles with C-terminal mutations (and dominant-negative effects) had no effect on phenotype. Both the wild-type allele and the two dominant alleles caused SM ΔwspR to convert to WS in a manner identical to the effect caused by the introduction of these alleles into the ancestral SM genotype. The two dominant alleles also caused SM ΔwspABCDEFR to convert to WS; however, when the wild-type copy of wspR was introduced into SM ΔwspABCDEFR there was no change in phenotype. This indicates that wild-type WspR is activated by the Wsp pathway and that the activity of WspR R129C and WspR D159G is kinase independent.
Fitness effects of wspR alleles:
The results described thus far show that manipulation of wspR (and allelic variants thereof) in both ancestral and derived WS genotypes generates phenotypes of likely evolutionary significance. This adds further weight to the notion that wspR plays a central role in the evolution of WS; however, as phenotype alone can be misleading, direct insight into the fitness consequences of these manipulations was sought. To this end the effects of the six different wspR alleles (three dominant-negative alleles and three dominant alleles, including the wild-type allele) on fitness of both the ancestral SM and derived LSWS genotypes were determined. In each case the genotype containing one of the test alleles was competed (in separate experiments) against both the ancestral SM genotype and a single WS genotype. Populations were founded with equal numbers of competing strains and competitions (over 24 hr) were performed in both spatially structured and unstructured environments. A neutral pantothenate marker was used to distinguish competing types and plasmid pVSP61 (without wspR) was used as a control. The results are shown in Figure 3.
For the purpose of statistical analysis the data presented in Figure 3 were analyzed in four sets (each representing a separate set of fitness assays) as follows: (1) fitness effects of alleles in SM relative to SM ΔpanB (a neutral marker) in structured and unstructured environments, (2) fitness effects of alleles in SM relative to WS ΔpanB in structured and unstructured environments, (3) fitness effects of alleles in LSWS relative to SM ΔpanB in structured and unstructured environments, and (4) fitness effects of alleles in LSWS relative to WS ΔpanB in structured and unstructured environments. In each case a two-way ANOVA with allele as a fixed effect revealed a significant interaction between allele and environment: (F[6,28] = 37.60, P < 0.0001), (F[6,23] = 12.75, P < 0.0001), (F[6,28] = 5.83, P = 0.0005) and (F[6,28] = 8.78, P < 0.0001), respectively. Subsequent analysis involved a series of linear contrasts (broken down by environment) based on comparisons between the two allele classes [dominate negative (WspR D206G, WspR F252S, WspR G296R) and dominant (WspR, WspR R129C, WspR D159G)]; in addition, each allele class was contrasted with the control (plasmid pVSP61). The results of these contrasts are shown in Table 2. In all contrasts significant differences were observed between the effects of the two different allele classes.
The fitness effects of wspR alleles in the ancestral genotype were largely consistent with a priori predictions on the basis of the visible effects of alleles on colony morphology: the dominant-negative alleles had no fitness effects (Figure 3, A and C), whereas the dominant alleles did. Of interest is the fact that the effects of the dominant alleles were allele specific, such that WspR and WspR D159G increased fitness in the structured environment, whereas WspR R129C caused a decrease in fitness (Figure 3A). The increased fitness conferred by WspR and WspR D159G in the structured environment appeared to be traded off against diminished fitness in the unstructured environment.
When introduced into the LSWS genotype wspR alleles generally resulted in a decrease in fitness relative to controls in both structured and unstructured environments (Figure 3, B and D). Linear contrasts among allele classes (broken down by environment; Table 2) showed highly significant effects for all comparisons except for those between the dominant negative alleles and the control (in unstructured microcosms). This is consistent with the observed effects of dominant-negative alleles, which is to convert LSWS to the ancestral-like phenotype.
No allelic variation in wspR from independent WS genotypes:
Ability to generate the WS phenotype from the ancestral SM genotype by increasing the level of expression of wspR implicates wspR as a central component in the evolution of WS. Moreover, the fact that different alleles of wspR (when overexpressed) are capable of generating WS genotypes that differ in both form (Figure 2) and fitness (Figure 3)—in a manner indicative of the differences that arise spontaneously during the course of adaptive radiation of the ancestral SM genotype—implicates wspR as a locus within which mutations might spontaneously arise to cause the evolution of WS.
A test of this appealing possibility was undertaken: the nucleotide sequence of wspR was obtained from 53 independent WS genotypes by direct DNA sequencing of PCR-amplified products. No difference in the sequence of wspR was detected.
In previous work we identified, via a suppressor analysis, numerous genes that contribute toward expression of the traits and characters that define the WS phenotype (Spiers et al. 2002). Understanding why these genes matter and the nature of their contribution remains an important goal; equally, there is interest in the mutational causes of the WS phenotype and the molecular basis of morphological diversity among WS genotypes. Our earlier work focused on structural components of the WS phenotype: the acetylated cellulose polymer (Spiers et al. 2002), attachment factors (Spiers et al. 2003), and more recently the contribution of lipopolysaccharide (Spiers and Rainey 2005). Having understood the contribution of each of these, attention has shifted to regulatory components required for expression of the WS phenotype.
Certain features of the WspR protein, in particular the N-terminal response regulator domain and C-terminal GGDEF output domain, suggested that it might play a central role in evolution and expression of the WS phenotype. Accordingly, wspR was subject to a series of in vitro manipulations and the effects of the variant forms were examined in both ancestral and derived genotypes. One outcome of these manipulations was demonstration of a relationship between wspR and the WS phenotype—and importantly, one where phenotypic variation among WS can be accounted for by changes in the level of activity of a key regulator.
Overexpression of wspR in the ancestral SM genotype was, in and of itself, sufficient to bring about development of the WS phenotype with concomitant effects on cellulose production and niche specialization; WspR is thus a proximate cause of cellulose overproduction (and more generally the WS phenotype). Ordinarily, signal transduction pathways switch between active and inactive states in a sensitive and reversible way. Overexpression of WspR shows that it is possible to tip the balance of the pathway toward a constitutively active state, as can happen when response regulators are overexpressed in the absence of their cognate kinase (Powell and Kado 1990; Hoch and Silhavy 1995; Aguilar et al. 2001). Suspecting that overactivation reflected enhanced WspR phosphorylation, the predicted site of phosphorylation (Asp67) was changed to asparagine, an amino acid that is almost identical to aspartate from a structural perspective. Failure of WspR D67N to convert SM to WS when overexpressed in the ancestral SM genotype strongly supports the conjecture that the cause of WspR activation is due to a chemical modification of Asp67.
Evidence that different levels of WspR activity are a cause of phenotypic differences among WS genotypes comes from several angles. First, overexpression of WspR in the ancestral SM and LSWS genotypes generated distinctly different WS phenotypes. Given that WspR is likely to be constitutively active in LSWS (a view supported by the effects of the dominant negative alleles of wspR; see below), further increases in the dosage of WspR are likely to further increase the level of active WspR. In this case the increased WspR dosage in LSWS caused development of a tighter, more compacted WS form with a significantly reduced fitness in spatially structured microcosms relative to WS ΔpanB. Second, when overexpressed in the ancestral genotype, the constitutively active (phosphorylation-independent) wspR alleles resulted in two phenotypically distinct WS types, with significantly different fitness relative to the ancestral SM genotype (Figures 2 and 3). Given that the only difference between SM WspR R129C and SM WspR D159G is a single amino acid in the WspR sequence, the differences in phenotypes and fitnesses must reflect different activity levels of the respective WspR proteins. Third, the ability of the dominant negative alleles to completely revert the WS phenotype of LSWS to the smooth ancestral morphology by competitive inhibition of the wild-type chromosomal copy shows the radical impact of eliminating WspR activity.
The immediate consequences of WspR phosphorylation have not been examined here, but given that the output domain of WspR can substitute for the GGDEF output domain of PleD from C. crescentus (Aldridge et al. 2003) combined with the proven DGC activity of PleD (Paul et al. 2004) [and WspR from P. aeruginosa (Hickman et al. 2005)] it is highly likely that the immediate effect of phosphorylation is activation of the output domain and concomitant biosynthesis of c-di-GMP. Indeed, recent studies have confirmed the DGC activity of WspR (J. G. Malone and P. B. Rainey, unpublished data). Given that c-di-GMP is a known allosteric activator of cellulose biosynthetic enzymes (Ross et al. 1987) and that cellulose is a critical determinant of the WS phenotype (Spiers et al. 2002), the scenario whereby activation of WspR leads to overactivation of the enzymes encoded by the wss operon—and related adhesive factors—is a highly plausible explanation for the WS phenotype.
In terms of understanding the mutational cause(s) of the LSWS phenotype, the C-terminal (dominant-negative) wspR alleles proved significant. Overactivation of WspR by phosphorylation at Asp67 is necessary to generate LSWS, but the ultimate cause of overactivation must lie upstream of WspR and most likely in components of the Wsp signal transduction pathway of which WspR is a part. Data from analysis of the phosphate-quenching effects of the C-terminal mutant alleles (WspR D206G, WspR F252S, and WspR G296R) strongly support this conjecture; that these alleles could competitively inhibit the wild-type chromosomal copy of WspR means that the flow of active phosphate to WspR must be elevated in the LSWS genotype, most likely as a consequence of a mutation that causes overactivation of the cognate kinase WspE. Indeed, analyses reported in a forthcoming article show that the mutational cause of LSWS is a single nucleotide change in wspF (P. B. Rainey, E. Bantinaki, C. G. Knight, R. Kassen, Z. Robinson and A. J. Spiers, unpublished observations), a methyl esterase that controls the activity of the WspE kinase (D'Argenio et al. 2002; Hickman et al. 2005).
Clearly the mutation causing LSWS does not—and cannot—reside within wspR. Nevertheless, it seemed reasonable to expect that in at least some WS genotypes the causal mutation would reside in wspR; sequencing, though, suggests otherwise. One possible explanation may be that in single copy in the chromosome, alleles such as wspR C385T (which encodes WspR R129C) are incapable of having the effects that are evident when overexpressed on a medium copy plasmid under the control of the nptII protomer, but arguing against this is the fact that the wsp operon is actively transcribed in both SM and LSWS genotypes (Bantinaki 2002). An alternate explanation might be the rarity of gain-of-function mutations of the kind identified in this study relative to loss-of-function mutations elsewhere in the pathway leading to WspR activation, for example, in the methyl esterase wspF.
Analysis of the fitness consequences of different wspR alleles (when expressed in both the ancestral and derived genotypes) provided some insights into the evolutionary significance of wspR. In general the fitness effects were consistent with a priori expectations; that is, the dominant alleles (which convert SM to WS) increase fitness of the ancestral genotype in spatially structured microcosms, whereas the dominant-negative alleles (which convert LSWS to SM) reduce fitness of LSWS in the spatially structured environment. However, there was one notable exception: whereas wild-type WspR and WspR D159G increased the fitness of the ancestral genotype relative to both the ancestral genotype and WS ΔpanB, in spatially structured microcosms, WspR R129C, despite causing SM to convert to WS (Figure 2), caused a significant reduction in fitness (relative to both SM and WS). The causes of this are not known, but it is possible that either WspR R129C causes too great an elevation in the level of production of adhesive factors or it has additional pleiotropic effects within the cell. It is interesting to note that all three dominant alleles caused a reduction in the fitness of LSWS. It might be argued that because LSWS evolved de novo from the ancestral genotype in an identical microcosm environment as used here it resides on or near an adaptive peak. As a consequence further manipulations may well decrease fitness, as was observed here.
As anticipated given their morphological effects, the alleles with C-terminal mutations had minimal impact on fitness of the ancestral type, but they significantly (and with equal effect) decreased the fitness of LSWS in the structured environment. Interestingly the effect was greater than the expected reduction in fitness to ancestral levels indicating that the C-terminal alleles have additional negative fitness consequences over and above those caused by inactivation of cellulose and associated attachment factors. This is reminiscent of previous findings that indicate that complete eradication of cellulose biosynthetic capacity in the ancestral genotype is costly (Spiers et al. 2002).
Finally, although our work was not motivated by a specific aim to understand the function of GGDEF domain proteins, the mutant alleles and manipulations reported here have proved useful. For a start, discovery that wspR can be constitutively activated by single nucleotide changes opens the door to a detailed molecular-level understanding of the causes of DGC activation. This is particularly valuable in light of the recent crystal structure of PleD (Chan et al. 2004) and given the fact that WspR is one of the simplest GGDEF proteins known (and best understood on account of the CheY-like receiver domain) with a readily assayable phenotype. Indeed, the discovery of WspR R129C and the ability of the R129C mutation to constitutively activate the output domain of PleD motivated an ultimately successful search for constitutively active PleD alleles (Aldridge et al. 2003). WspR R129C also provided evidence that in P. aeruginosa WspR is the proximate cause of the autoaggregation phenotype (D'Argenio et al. 2002).
The mechanisms by which the output domains of response regulator proteins are activated are poorly understood. Several mechanisms have been proposed and these include relief of active site obstruction (CheB) (Djordjevic and Stock 1998; Lee et al. 2001); dimerization, as in PleD (Chan et al. 2004), DrrB, and DrrD (Robinson et al. 2003); or both, as is the case for NarL (Baikalov et al. 1996; Maris et al. 2002). The precise conformational effects of the two activating mutations in WspR are not known, but given the similarity between WspR and PleD it is likely that WspR is activated by dimerization. A recent, intensive and fine-scale mutational analysis of WspR strongly supports the view that the active conformation may, in addition to dimerization, involve relief of active site obstruction (J. G. Malone, R. Williams, A. J. Spiers and P. B. Rainey, unpublished data).
Explaining even the simplest cases of adaptive evolution is a daunting task (Bull et al. 1997; Treves et al. 1998; Wichman et al. 1999; Cooper et al. 2001; Cooper et al. 2003; Crozat et al. 2005). Much emphasis is placed on understanding the mutational causes of new phenotypes, but knowledge of mutations alone goes only a small way toward explaining the origins and adaptive significance of new phenotypes. A central challenge is to link DNA sequence change to phenotype and to fitness via the incorporation of development. Our top-down analysis of experimental P. fluorescens populations has revealed much about the development of the WS phenotype. In common with a number of other studies (e.g., Doebley et al. 1997; Treves et al. 1998; de Rosa et al. 1999), the origin of the new WS phenotype owes much to alterations in regulation. What is striking in the case of the WS is not just that subtle changes in regulation can have profound effects, but that different levels of regulator activity can account for phenotypic variation among WS genotypes. This suggests that the genetic architecture of WspR regulation, and perhaps more generally the c-di-GMP network, might be evolutionarily significant, possibly with a capacity for phenotypic exploration or evolvability (Gerhart and Kirschner 1997). We anticipate that future studies on components of the wsp operon and the genetic architecture underlying the c-di-GMP network will provide significant insight.
We thank Chris Knight for valuable comments on the manuscript, Allen Rodrigo and Chris Knight for advice on statistical analysis, and Zena Robinson and Julie Stansfield for technical assistance. This work was supported by the Biotechnology and Biological Sciences Research Council (UK).
- Received January 14, 2006.
- Accepted March 20, 2006.
- Copyright © 2006 by the Genetics Society of America