Differential Regulation of Twitching Motility and Elastase Production by Vfr in Pseudomonas aeruginosa

Bacterial strains, plasmids, and media.

E. coli strain DH5α was used in all genetic manipulations and in the preparation of DNA sequencing templates, and E. coli S17-1 was used as the donor strain in bacterial conjugation (45). Details of bacterial strains, plasmid construction, and the primers used in this study are described in Table 1. P. aeruginosa competent cells and transformations were prepared by MgCl₂ treatment as described previously (29). E. coli and P. aeruginosa liquid cultures were maintained in Luria-Bertani (LB) broth (41), and solid medium (LBA) was prepared by adding 1 to 1.5% Select agar (Gibco-BRL). Pseudomonas Isolation Agar (PIA; Difco) was used for pyocyanin quantitation from plate cultures. Light microscopy was performed with nutrient medium (4 g of Tryptone liter⁻¹, 2 g of yeast extract liter⁻¹, 2 g of NaCl liter⁻¹) solidified with 8 g of GelGro (ICN) liter⁻¹ for greater optical clarity. The following antibiotic concentrations were used for selection of E. coli: 12.5 μg of tetracycline ml⁻¹ for plasmid selection, 40 μg of tetracycline ml⁻¹ for cosmid selection, 100 μg of ampicillin ml⁻¹, and 50 μg of kanamycin ml⁻¹. The concentrations of antibiotics used for selection of P. aeruginosa were 500 μg of carbenicillin ml⁻¹, 250 μg of chloramphenicol ml⁻¹, 20 μg of rifampin ml⁻¹, and 200 μg of tetracycline ml⁻¹.

DNA manipulations.

DNA manipulations, plasmid DNA extractions, Southern hybridization, and radiolabeling of probe DNA were performed by standard procedures (41). The enzymes for DNA manipulation were supplied by Roche and New England Biolabs.

Construction and screening of PAO1293 transposon mutant library.

P. aeruginosa strain PAO1293, a chloramphenicol-resistant derivative of completely sequenced strain PAO1, was mutagenized with the 2.2-kb mini-Tn 5-tet (hereafter abbreviated mTn 5-Tc) transposable element. A library containing 92,260 tetracycline-resistant clones was characterized by the following four criteria: (i) estimation of the frequency of exconjugant mutants, (ii) genomic profiling of 100 Tc^r clones selected via PCR amplification of an internal 398-nt Tc gene product, (iii) Southern hybridization, and (iv) sequencing of mTn 5-Tc insertion endpoints for 30 clones. On the basis of these criteria and biostatistical analysis as a binomial probability, the library was estimated to cover the 6.3-Mb chromosome at 14.6 genome equivalents.

In order to facilitate mass screening for mutants deficient in twitching motility, aliquots of the original library were diluted and plated on LBA containing tetracycline and ∼12,000 individual colonies were randomly picked into 96-well microtiter plates that were stored at −70°C after incubation. A 96-prong twitching stab tool was constructed with stainless steel prongs in a Perspex base to allow rapid screening of the entire gridded library for loss of twitching motility. After overnight incubation, mutants with absent or reduced twitching motility were selected and confirmatory twitching stabs were performed manually (see below).

The sites of mTn 5-Tc insertion from mutants with absent or reduced twitching motility were determined by generating and sequencing marker rescue clones. Briefly, P. aeruginosa mTn 5-Tc mutant chromosomal DNA was digested with PstI or XhoI and then shotgun cloned into pUCPSK and used to transform E. coli DH5α cells that were subsequently plated on LBA containing ampicillin and tetracycline. Plasmid DNA was extracted from ampicillin- and tetracycline-resistant colonies, and the site of transposon insertion was identified by sequence analysis as described below. Mutant 118C8 was found to contain an mTn 5-Tc insertion within vfr.

Sequencing and sequence analysis.

Automated DNA sequencing was performed by the Australian Genome Research Facility (University of Queensland, Brisbane, Queensland, Australia) by using big-dye terminator chemistries and Taq cycle sequencing kits from Perkin-Elmer Applied Biosystems and analysis on an ABI PRISM 377 DNA sequencer (Perkin-Elmer Applied Biosystems). A primer (Tn5.I; 5′-GCGGCCAGATCTGATCAAGAG-3′) complementary to a region near the I terminus of the mini-Tn5 cassette was used to obtain the sequence adjacent to the point of transposon insertion in marker rescue clones. Primer Tn5.I was manufactured by Pacific Oligos, Lismore, New South Wales, Australia. BLAST (5) searches of the P. aeruginosa PAO1 unfinished genome sequence at the National Center for Biotechnology Information (Bethesda, Md.) identified the position of transposon insertion, and the DNA sequence from that region was obtained via the Institute for Molecular Bioscience P. aeruginosa interactive database (http://www.bit.uq.edu.au/pseudomonas) (13). Further BLAST searches were performed by using the nonredundant nucleotide and protein databases at the National Center for Biotechnology Information. Where quoted, protein identities were calculated from a CLUSTALW pairwise alignment with an open gap penalty of 10 and a gap extension penalty of 0.1 with a blosum 30 similarity matrix.

Construction of isogenic mutants.

vfr allelic-exchange mutants were constructed by using the sucrose selection system described previously (4, 43). To construct PAK vfr::mTn 5-Tc, the vfr::mTn 5-Tc allele was subcloned from the marker rescue clone of mutant 118C8 into the vector pOK12. To construct PAO1 vfr::Tc, the 1.3-kb XhoI fragment from pUCP vfrA was subcloned into pOK12 and the tetracycline cassette from pSM-TET was inserted as a blunt fragment into the BclI site (5′ extensions were removed with mung bean nuclease) within vfr. The resulting clones were then digested with SpeI, and the disrupted genes were inserted into the suicide vector pRIC380 (Table 1). This vector carries the genes sacBR, which promote sensitivity to sucrose, and oriT, which enables conjugal transfer. The constructs were then used to transform E. coli donor strain S17-1 in preparation for mating with P. aeruginosa. Following conjugation, the transconjugants were plated on 5% sucrose medium containing tetracycline to select for colonies in which the plasmid had been excised while leaving the homologously recombined mutated gene in the chromosome. Mutants were confirmed by Southern blot analysis and examined with the subsurface twitching assay.

Twitching motility assay.

Twitching motility was assayed as described previously (3). Briefly, the P. aeruginosa strain to be tested was stab inoculated through a 1% agar plate and, after overnight growth at 37°C, the zone of twitching motility between the agar and petri dish interface was visualized by staining with Coomassie brilliant blue R250.

Western blotting.

Whole cells from LBA plates were resuspended to an optical density at 600 nm of 1.0 in 50 mM sodium carbonate buffer, pH 9.6. One-milliliter samples were centrifuged, and the cell pellet was resuspended in 100 μl of sample buffer (60 mM Tris-HCl [pH 6.8], 2% sodium dodecyl sulfate, 10% glycerol, 5% β-mercaptoethanol, 0.001% bromophenol blue). To reduce viscosity, the samples were sheared by 20 passages through a 27-gauge needle and heated to 100°C for 5 min. Samples were then displayed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis with a 15% polyacrylamide gel and a 5% stacking gel (27) and transferred electrophoretically to Hybond-C nitrocellulose (Amersham) with a Tris-glycine buffer system (49). Proteins were detected with anti-PilA serum (1:5,000), followed by goat anti-rabbit immunoglobulin G conjugated to alkaline phosphatase (1:5,000; Boehringer Mannheim).

ELISAs.

Enzyme-linked immunosorbent assays (ELISAs) (15) to determine the amount of surface pili were performed as follows. Cells were resuspended in 50 mM sodium carbonate buffer, pH 9.6, at an optical density at 600 nm of 1.0, and 200 μl was loaded into wells of a 96-well ELISA plate. After overnight incubation at 4°C, the wells were washed with phosphate-buffered saline containing 0.1% Tween 20, blocked with 3% bovine serum albumin for 1 h, and then exposed to an anti-PilA antibody at a starting dilution of 1:500 for 2 h at 37°C. After removal of antisera, the wells were again washed with phosphate-buffered saline containing 0.1% Tween 20, and then goat anti-rabbit immunoglobulin G conjugated with alkaline phosphatase was added (1:5,000) and the mixture was incubated for 2 h at 37°C. Detection was done with 20 mg of p-nitrophenyl phosphate (Sigma) ml⁻¹ in 1 M Tris buffer (pH 8.0), and the tray was read at 405 nm with an ELISA reader (Bio-Rad).

Light microscopy.

Light microscopy was performed as described previously (44). Briefly, sterile microscope slides were submerged in molten GelGro medium to obtain a thin layer of medium coating the slide. The slides were allowed to set in a horizontal position and air dried briefly prior to use. The slides were then inoculated with a small loopful of bacteria taken from an overnight plate culture. A sterile glass coverslip was placed over the point of inoculation, and the slide was transferred to a large petri dish containing a moist tissue and sealed with Nescofilm (Bando Chemical Industries) to maintain humid conditions. Incubation times ranged from 2 to 6 h at 37°C. Slide cultures were examined with a Zeiss Axioskop 50 microscope with Nomarski facilities at magnifications of ×200 to ×400.

Three-dimensional protein homology modeling.

With the SWISS-MODEL server (http://www.expasy.ch/swissmod/SWISS-MODEL.html) and the Swiss Protein Data Bank (PDB) viewer (http://www.expasy.ch/spdbv/) (19), three-dimensional homology models of Vfr and VfrΔEQERS were automatically constructed with the cAMP-CRP structure as a template (PDB entry PDB 1D 1G6N) (35). In the Vfr model, residues forming the cyclic nucleotide monophosphate (cNMP)-binding pocket all have main-chain dihedral angles that are in allowed regions of the Ramachandran plot. However, the additional residues in Vfr between β strands 6 and 7 make it impossible to predict the exact conformation of this nonconserved loop with the CRP structure alone. Therefore, the sequence between β strands 6 and 7 in Vfr (LFEKEGSE) was modeled with the loop-building function of the Swiss PDB viewer and the nonconserved loop database. Multiple loop conformations with favorable torsion (main-chain dihedral angles with the allowed regions of the Ramachandran plot) and energetics were identified when β strands 6 and 7 were constrained at L77 and E86, respectively. Therefore, the conformation predicted for Vfr β strands 6 and 7 is theoretically possible despite the difference in the intervening loop. The same procedure was used to optimize the model of VfrΔEQERS.

Elastase assay.

Assay of the elastolytic activity of P. aeruginosa spent culture supernatant was done essentially as previously described (33). Briefly, bacterial cultures were incubated with shaking at 37°C for 16 h in LB broth. After centrifugation of the culture, 500 μl of supernatant (diluted 1:10) was added to 10 mg of Elastin Congo Red (Sigma) in glass tubes and the mixture was incubated at 37°C for 16 h. After centrifugation of the slurry, the supernatant was diluted 1:10 and the absorbance at 495 nm was measured with a spectrophotometer zeroed on a control Elastin Congo Red sample incubated with LB broth alone. The relative elastolytic activity of each strain is represented by the absorbance at 495 nm of triplicate assays after adjustment for dilution.

Assay for pyocyanin production.

For plate assays, LBA or PIA plates were seeded with 100 μl of overnight broth culture and spread plated to achieve confluent growth prior to incubation for 16 h at 37°C. Plate cultures were shaken gently after the addition of 5 ml of H₂O to the plate surface. After 30 min, the slurry was removed by pipette and the cells were removed by centrifugation. Pyocyanin was extracted from culture supernatants and measured essentially as described by Essar et al. (16). Briefly, 500 liters of culture supernatant was extracted three times with 0.333 liter of diethyl ether and the pyocyanin was extracted from the pooled ether fractions into 0.2 M HCl as a red pigment that was subsequently measured spectrophotometrically at 520 nm.

Vfr is required for twitching motility.

During a screen of a P. aeruginosa PAO1293 mTn 5-Tc transposon library, we identified a mutant (118C8) that exhibited a large (∼70%) reduction in twitching motility levels, as measured by the standard subsurface agar assay. The growth rate of 118C8 was not significantly less than that of the PAO1293 parent (data not shown), indicating that the loss of twitching motility was not simply due to a growth defect. In order to characterize the site of transposon insertion in 118C8, we rescued a PstI chromosomal fragment containing the mTn 5-Tc marker. Sequencing outward from the transposon with primer Tn5.I indicated that P. aeruginosa 118C8 contained a mTn 5-Tc insertion within vfr (Fig. 1A).

A cosmid encompassing the vfr gene was isolated from the Holloway P. aeruginosa PAO1 cosmid library (22, 39), and a 4-kb PstI fragment containing vfr was subcloned into pUCPSK. From this intermediary plasmid, a 1.3-kb XhoI fragment containing vfr was subcloned into pUCPSK in both orientations relative to the lac promoter (Fig. 1B). The defective twitching motility of 118C8 was restored to wild-type levels by complementation with wild-type vfr in either orientation (data not shown), indicating that the mTn 5-Tc transposon insertion in vfr, and not a secondary-site mutation, was the cause of the twitching defect.

To allow comparison with the other twitching-defective mutant strains identified previously in this laboratory (21), the majority of which were constructed with a P. aeruginosa strain K (PAK) parent, we also reconstructed the vfr knockout mutation in PAK by transferring the mTn 5-Tc insertion by allelic exchange. As observed with 118C8, the PAK vfr::mTn 5-Tc mutant has a severe defect in twitching motility characterized by a small, misshaped twitching zone by stab assay (Fig. 2). Although the twitching defect in PAK vfr::mTn 5-Tc is not as severe as that observed in PAK pilA::Tc (which lacks type IV pili and is devoid of any twitching motility), the strain is clearly distinguishable from the wild type. Light microscopy revealed that the twitching edge of the PAK vfr::mTn 5-Tc mutant was not undifferentiated, like that of PAK pilA::Tc, but showed large rafts of cells moving outward, albeit slowly, from the colony (data not shown). However, these rafts were unable to break away from the colony edge and there was no subsequent formation of the intricate lattice-like network of cells characteristic of wild-type twitching motility (44).

The mTn 5-Tc insertion was very close to the vfr stop codon (17 nt upstream), raising the possibility that the partial-twitching phenotype may be due to a truncation of Vfr rather than a true null mutation. To address this concern, we created an independent isogenic knockout mutant by inserting a tetracycline resistance cassette centrally within vfr in P. aeruginosa PAO1 to create PAO1 vfr::Tc (Fig. 1A). PAO1 vfr::Tc similarly exhibited a severe defect in twitching motility that could be complemented by pUCPvfrA and pUCPvfrF (data not shown), confirming that loss of vfr causes the observed twitching defect. On the other hand, the vfr mutation in strain PAK vfr::mTn 5-Tc could only be complemented by pUCP vfrA (vfr cloned in the opposite orientation with respect to the lac promoter), whereas twitching motility was actually inhibited by pUCP vfrF in wild-type PAK (presumably due to overexpression of vfr in this clone), indicating that there are differences between strains PAK and PAO in the balance of the regulatory circuitry connecting Vfr and twitching motility.

vfr-null mutants are defective in type IV pilus assembly.

To further characterize the defect in vfr-null mutants, we undertook ELISA and Western analyses. ELISA of whole cells with antisera against the major structural subunit PilA indicated that vfr mutants have very low levels of surface pili (Fig. 3A). Western blots of sheared surface pili with anti-PilA sera similarly demonstrated that the vfr mutant has a small but significant amount of surface piliation, which may explain its ability to twitch slowly (Fig. 3B). Western blots of whole-cell lysates showed that PAK vfr::mTn 5-Tc produces wild-type levels of cellular PilA (Fig. 3B), indicating that the vfr mutation does not cause a defect in pilA transcription. Therefore, the twitching defect in vfr-null mutants is most likely caused by an inability to properly coregulate the assembly or function of surface pili, rather than by a defect in the production of pilA. Given the high homology of Vfr to CRP and its previously described function as a transcriptional regulator, it is most likely that Vfr acts as a transcriptional regulator of a gene(s) encoding functional elements of type IV pili or signal transduction pathways that control twitching motility. However, the target(s) that is affected by Vfr in the control of twitching motility has yet to be determined.

Vfr appears to be capable of activation by both cAMP and cGMP.

A sequence comparison of Vfr and E. coli CRP (Fig. 4) shows that Vfr shares with CRP five of the six residues in the cAMP-binding domain involved in the strongest noncovalent interactions with cAMP (54). The cAMP-binding domains of Vfr and CRP belong to the larger family of cNMP-binding domains (Prosite accession no. PDOC00691 [http://ca.expasy.org/prosite/]), along with other CRP homologs, cAMP- and cGMP-dependent protein kinases, and vertebrate cyclic nucleotide gated channels. All of the well-characterized cNMP-binding domains (including CRP) can bind both cAMP and cGMP, but the degree of activation varies, depending on the bound nucleotide. According to various mutational studies, CRP residues T127 and S128 participate in intersubunit communication and determine nucleotide selectivity, i.e., T127 of one subunit, cAMP, and S128 of the other subunit of the CRP dimer are connected by noncovalent bonds (54). While CRP is not normally activated by cGMP, a threonine substitution at S128 will permit activation of CRP S128T by both cAMP and cGMP (28). Vfr contains an identical threonine substitution at the position equivalent to CRP S128, which suggests that Vfr may also be activated by both cAMP and cGMP. In addition, although Vfr can substitute for CRP in E. coli (in a cAMP-dependent fashion), CRP cannot reciprocally complement a vfr mutation in P. aeruginosa, suggesting that CRP is unresponsive to another signal recognized by Vfr and/or that CRP cannot recognize some Vfr-specific promoters (56). Moreover, it has been shown that Crp/Vfr and Vfr/Crp chimeras are able to activate expression of lasR but not toxA or regA (S. E. H. West, L. J. Runyen-Janecky, and L. A. Smith, Pseudomonas ’99: Biotechnology and Pathogenesis, abstr. 150, 1999). Taken together, this suggests that Vfr is activated by both cyclic nucleotides to allow differential regulation of different regulons.

VfrΔEQERS appears to be capable of binding cAMP.

In the accompanying report, we show that P. aeruginosa lasI quorum-sensing mutants (PAK lasI-RT) spontaneously develop defects in twitching motility via a precise 15-nt deletion in vfr (6). Significantly, the deleted residues (EQERS) form one-half of an imperfect tandem repeat that lies in the region corresponding to the cNMP-binding pocket of vfr, suggesting that further study of VfrΔEQERS may provide insight into Vfr signal transduction. To better understand the effect of this deletion on Vfr function, we constructed a homology-based model of Vfr and VfrΔEQERS with the backbone of the known CRP-cAMP three-dimensional structure (35) (Fig. 5). This analysis predicts that the residues encoded by EQERS in Vfr correspond to the motif between β strands 6 and 7 of CRP, which forms part of the cAMP-binding domain. In Vfr, there are three additional residues in the turn between β strands 6 and 7 but, despite this difference, we predict that Vfr should still be capable of binding cAMP (Fig. 5).

It is clear from the simple alignment (Fig. 4) that the deletion of EQERS from Vfr results in the net loss of two residues corresponding to CRP R82 and S83, both of which strongly interact with the cyclic phosphate moiety of cAMP (54). Intriguingly, our model of VfrΔEQERS suggests that this mutant protein is still capable of binding a cAMP molecule (Fig. 5). The retention of binding potential is bought about by the compensatory repositioning of EGS in place of the deleted ERS residues (Fig. 5). The potential for Vfr to adapt to this deletion is dependent on both the longer chain length of Vfr relative to CRP between β strands 6 and 7 and the similarity of the imperfect tandem repeats (i.e., EKEGS/EQERS). However, it should be noted from the model that the substitution of a glycine for the essential arginine residue removes the ability to form some noncovalent bonds with the cyclic phosphate moiety of a cNMP molecule. Therefore, it is likely that VfrΔEQERS has less affinity for cNMP molecules, which may affect the allostery imparted by cNMP binding.

VfrΔEQERS retains the ability to control elastase and pyocyanin production.

In light of the potential for VfrΔEQERS to bind cAMP, we were interested in whether this mutant protein retains any functionality. By transforming PAO1 vfr::Tc with cloned PCR amplimers of vfr from wild-type PAK and PAK lasI-RT, we could measure the ability of Vfr and VfrΔEQERS, respectively, to rescue the phenotype of the vfr-null mutant. As expected, VfrΔEQERS could not complement the defective twitching motility phenotype of PAO1 vfr::Tc (Table 2). Vfr has also been reported to control the expression of lasR, which is required for expression of a range of quorum-sensing-related factors, including elastase (1). Consistent with a lack of lasR expression, PAO1 vfr::Tc (Table 2) and PAK vfr::mTn 5-Tc (data not shown) are deficient in elastolytic activity. The PAO1 vfr::Tc elastase deficiency can be complemented by introducing wild-type vfr on a plasmid. Moreover, VfrΔEQERS was also able to restore wild-type levels of elastase production in PAO1 vfr::Tc (Table 2). Therefore, although the loss of EQERS drastically affected the ability of Vfr to control twitching motility, it had no effect on the ability of Vfr to control elastase production.

We were interested in whether defects in the production of any other Vfr-controlled virulence factors could be similarly complemented by VfrΔEQERS. We had observed that when cultured in LB broth, PAO1 vfr::Tc produces greatly reduced quantities of pyocyanin compared to those produced by the wild-type strain (data not shown). Interestingly, we found that PAO1 vfr::Tc, when it was cultured on PIA plates, produced a greater quantity of pyocyanin than the wild type, whereas the opposite was true when LBA was used (Table 2). As observed with elastase production, wild-type levels of pyocyanin production were restored to PAO1 vfr::Tc when it was complemented with either Vfr or VfrΔEQERS, on both LBA and PIA (Table 2).

These results provide strong support for the idea that Vfr possesses the novel ability to differentially activate (or repress) discrete regulons in response to two different signals (probably cAMP and cGMP) recognized in the same cNMP-binding domain (see Discussion). Such a model suggests that the partial loss of function observed in VfrΔEQERS is due to an inability to bind and/or be activated by either cAMP or cGMP while retaining the ability to respond to the other.