The Pseudomonas aeruginosa type IV pilin receptor binding domain functions as an adhesin for both biotic and abiotic surfaces

Initial colonization of stainless steel is dependent upon type IV pili in P. aeruginosa

While the involvement of type IV pili in biofilm formation on abiotic surfaces has been well documented (O’Toole and Kolter, 1998; Klausen et al., 2003), the molecular basis for that involvement has not been firmly established. Therefore, we investigated the ability of P. aeruginosa wild-type strains PAK, PAO, K122-4 and KB7, which display considerable differences in their pilin sequences, to bind to stainless steel (Fig. 1A and B). P. aeruginosa pilins are characterized by a highly conserved N-terminal α-helix and a semi-conserved C-terminal disulphide loop region but display minimal sequence similarity through the bulk of the protein (Fig. 1A). However, structural studies indicate that P. aeruginosa pilins are strikingly similar (Fig. 1B) (Hazes et al., 2000; Craig et al., 2003; Audette et al., 2004). P. aeruginosa strains PAK, PAO, K122-4 and KB7 were observed to rapidly bind to stainless steel surfaces (Fig. 2A, D, G and J), in agreement with previous results (Stanley, 1983; VanHaecke et al., 1990). Strains PAO and PAK bound more significantly to stainless steel than either K122-4 or KB7 (Fig. 2A, D, G and J). P. aeruginosa strain MS591 (Starnbach and Lory, 1992), a fliC^– non-flagellated mutant of PAK, was observed to bind at considerably reduced levels compared with the parental PAK strain (Fig. 2A and P). P. aeruginosa strain PAKNP, a non-piliated pilA deficient mutant of PAK (Saiman et al., 1990), did not bind to steel surfaces (Fig. 2M), suggesting that P. aeruginosa adherence to stainless steel may be mediated by type IV pili. To further investigate the initial colonization of stainless steel, the binding kinetics of PAK cells and pili to steel was examined in a quantitative manner employing viable biotinylated cells and purified biotinylated PAK pili. PAK cells (3, 4) and purified PAK pili (Fig. 3B) were observed to bind to steel surfaces in a saturable, concentration-dependent manner while PAKNP biotinylated cells (3, 4) did not bind appreciably to the steel surface. Biotinylation had no effect on the ability of the purified pili to bind stainless steel as identical binding kinetics were observed with native pili when anti-PAK pili antibodies were utilized to quantify binding (data not shown). Furthermore, addition of low concentrations of purified PAK pili competitively inhibited the binding of biotinylated PAK cells to steel (Fig. 3C). As the length of PAK pili is not known, the molarity of the pili cannot be determined (van Schaik et al., 2005).

We established that P. aeruginosa stained with acridine orange were readily visualized by epifluorescence microscopy following binding to stainless steel (Fig. 1D). P. aeruginosa cells were visualized as fluorescent orange rods bound to areas of the steel that fluoresced green either through non-specific interaction of the fluorochrome with grain boundary regions or due to non-specific interaction of the fluorochrome with organic material that interacted with the grain boundaries in the steel. Green fluorescent material primarily associated with grain boundaries in the steel was observed in both strains PAK and PAKNP (Fig. 1D and E). In confirmation of these results, Pseudomonas sp. has recently been shown to colonize preferentially with grain boundaries (Sreekumari et al., 2001).

Effect of flagella and type IV pili mutations on binding to steel

Previous studies using P. aeruginosa FilC^– and PilB^– strains have established the importance of flagella and type IV pili during the initiation and development of biofilms on abiotic surfaces in static cultures (O’Toole and Kolter, 1998). To confirm our qualitative microscopic examination, a quantitative analysis of adherence to steel was performed. The binding of PAK-BΩ (PilB^– mutant), a strain that does not assemble pili but does express PilA (Koga et al., 1993), to steel was compromised relative to wild-type and equivalent to that observed for PAKNP and PAKMS591 (Fig. 4). ELISA evidence indicates that PAKMS591 has less surface-exposed pili than does the PAK wild-type strain (data not shown). Strain PAK-DΩ (pilD-mutant) which lacks the prepilin peptidase and therefore does not express surface-exposed PilA (Koga et al., 1993) or functional type IV pili, bound roughly equivalent to mutant strains (PAKNP, PAKMS591 and PAK-BΩ) (Fig. 4). Although the binding curves differ slightly between mutant strains, all pili-deficient strains (PAKNP, PAK-BΩ and PAK-DΩ), and PAKMS591, bound significantly less than wild-type (Fig. 4). These results indicate that any mutation which abolishes the production of functional pili also reduces the ability to bind stainless steel.

Antibody inhibition studies

Addition of rabbit polyclonal anti-PAK pili antibodies (Lee et al., 1989) but not rabbit pre-immune serum strongly inhibited the binding of biotinylated PAK cells and pili (Fig. 5) to steel in a dose-dependent manner. These data indicate that the type IV pili mediates binding. Murine monoclonal antibody PK99H, which recognizes PAK PilA residues 134–140 exposed at the tip of the pilus (Lee et al., 1994; Wong et al., 1992), significantly inhibited the binding of PAK cells and pili to steel (Fig. 5). The inhibition of PAK binding to stainless steel by PK99H suggests that the C-terminal disulphide loop region of pilin, which contains an epithelial cell binding domain, may also function in mediating attachment to steel surfaces.

Synthetic peptide inhibition of binding to steel

Antibody inhibition assays suggested that the cellular receptor binding domain of the pilus may also mediate binding to stainless steel. Therefore, competitive binding assays were utilized to test the ability of the C-terminal receptor binding domain to inhibit adherence of P. aeruginosa to steel. Previous studies demonstrated that the native C-terminal receptor binding domain (PAK128–144ox) mediates binding to GalNAc-β-D-Gal containing glycoconjugates (Sheth et al., 1994). The PAK pilin receptor binding domain (PAK128–144ox) inhibited the binding of both PAK wild-type cells and PAK pili to steel surfaces with apparent K_i's of ∼4 nM and ∼0.2 nM respectively (Fig. 6A and C). PAK(128–144)ox also inhibited the binding of PAKMS591 (compare Fig. 2P with Fig. 2Q) but had no effect on the binding of PAKNP (compare Fig. 2M and N). The peptide PAK(134–140), which constitutes a portion of the receptor binding domain, and has been demonstrated to bind with low affinity to respiratory epithelial cells (Yu et al., 1996), did not inhibit binding of PAK wild-type cells or PAK pili to steel surfaces, even at the exceptionally high peptide concentration of 100 µg ml⁻¹ (Fig. 6B and D), nor did it inhibit the binding of strains PAO, K122-4 or KB7, in contrast to peptide PAK(128–144)ox (Fig. 2). In addition, neither PAK(22–52), a peptide derived from the N-terminal α-helix (residues 1–58) which should be buried in the native pilus fibre, nor PAK(117–125), a peptide consisting of a portion of β-strands 3 and 4 which models of the pilus fibre suggest will be displayed on the fibre surface (Hazes et al., 2000) (see Table 1 for peptide sequences and Fig. 1B), had any effect, even at high concentrations, on the binding of PAK wild-type cells or PAK pili to steel surfaces (Fig. 6B and D), indicating that these regions do not participate in pilus-mediated binding.

To determine whether the ability to interact with steel surfaces was a general attribute of the C-terminal receptor binding domain or a specific property of the PAK receptor binding domain, the ability of the PAO receptor binding domain to inhibit binding was examined. The synthetic PAO receptor binding domain, PAO(128–144)ox was observed to inhibit pilus-mediated binding to stainless steel in a similar manner to the native PAK peptide (Fig. 6E). To further confirm the specific nature of the receptor binding domain's interaction with steel surfaces, two additional control peptides, a scrambled PAO receptor binding domain PAO(128-144)ox_Scrambled and a linear variant of that sequence where the two cysteine residues have been replaced by alanine residues to eliminate the disulphide bridge, PAO(128–144)C129A/C142A_Scrambled, were utilized to assess the relative importance of sequence versus amino acid composition. Neither scrambled sequence was able to inhibit binding of PAK wild-type cells or PAK pili to steel surfaces, even at very high peptide concentrations (Fig. 6B and D). As a further control, peptides obtained through the trypsinization of bovine serum albumin (BSA) were utilized to confirm that the inhibition of binding was sequence-specific and not a common property of peptides. No inhibition of adherence was observed for either PAK whole cells (data not shown) or PAK pili even at high peptide concentrations (Fig. 6G). A PAO receptor binding domain mutant, PAO(128–144)oxK130I which has high affinity for human buccal epithelial cells (BECs) (Fig. 7C) was unable to inhibit P. aeruginosa whole cells or pili adherence to steel (Fig. 6). To further determine whether the receptor binding domain inhibited binding to steel by a competitive mechanism or by interacting with P. aeruginosa cells or pili, the ability of PAK(128–144)ox to bind to stainless steel was determined using the monoclonal antibody PK99H as a probe of peptide binding to steel (PK99H has been demonstrated to bind to both PAK(128–144)ox and PAK(134–140) when these peptides are bound to a cell surface receptor) (Irvin et al., 1989; Yu et al., 1996). PAK(128–144)ox bound with high affinity to stainless steel while PAK(134–140) bound only marginally to the steel surface at very high concentrations (Fig. 7A and B).

As P. aeruginosa strains vary considerably in their ability to bind to steel surfaces (Fig. 2A, D, G and J), we sought to determine whether the PAK pilin receptor binding domain, PAK(128–144)ox, could inhibit the binding of other P. aeruginosa strains. Utilizing microscopy, we found that at very low concentrations (51 nM), PAK(128–144)ox substantially inhibits the binding of strains PAO, K122-4 and KB7 (compare Fig. 2D, G and J with Fig. 2E, H and K) while very high concentrations (100 µg ml⁻¹) of PAK(134–140) have a minimal effect on binding to steel (compare Fig. 2A, D, G and J with Fig. 2C, F, I and L). The pilin receptor binding domain sequences of strains PAK, PAO, K122-4 and KB7 vary substantially (Fig. 1A and B, and Table 1), but all these receptor binding domains display a conserved antigenic epitope and compete for epithelial cell surface receptors (Sheth et al., 1995).

Binding to other abiotic surfaces

As type IV pili have been implicated in biofilm formation on polystyrene and polyvinylchloride surfaces, we sought to determine if the C-terminal receptor binding domain may function to mediate attachment to a variety of abiotic surfaces. PAK whole cells and pili were found to bind in a concentration-dependent and saturable manner to both polyvinylchloride and polystyrene plates (Fig. 8A–D). The murine monoclonal antibody PK99H significantly inhibited binding to both polyvinyl chloride and polystyrene surfaces (Fig. 8E and F). These data indicate that type IV pili mediate binding to these plastic surfaces which may be dependant on the C-terminal receptor binding domain.

Bacterial strains, DNA and antibody sources

The P. aeruginosa strains used in this study were PAK, PAK 2Pfs (Bradley and Pitt, 1974), PAK-BΩ, a 2 kBΩ fragment containing a transcriptional terminator from pHP45 was inserted into the pilB gene, PAK-DΩ, the same transcriptional terminator was inserted into the pilD gene (Koga et al., 1993), PAKMS591, the gentamycin cassette from a pPC110 was inserted into the fliC gene (Starnbach and Lory, 1992), PAKNP the tetracycline cassette from pB322 was inserted into the pilA gene (Saiman et al., 1990), K122-4, a clinical isolate from a cystic fibrosis patient in Toronto which possesses both pili and flagella (Pasloske et al., 1988) and KB7, an isolate containing both pili and flagella (Wong et al., 1995). Several of these strains were generously provided by Dr Jessica Boyd (NRC Institute for Marine Biosciences, Halifax, Nova Scotia). The strain PAK 2Pfs, a multipiliated retraction-deficient strain, was used for the purification of pili only and not used in experimental conditions. The phenotypes of the P. aeruginosa strains with respect to expression of pili was experimentally verified by Western blotting with anti-PAK pilus-specific antisera and by direct ELISA with whole cells and heat inactivated whole cells (to determine the presence of surface-exposed pili), by the sensitivity to type IV pilus-specific phage, and by monitoring the twitching motility of the strains. P. aeruginosa was routinely grown at 37°C in Luria–Bertani broth (LB) or LB supplemented with 50 µg ml⁻¹ tetracycline (Sigma) for PAKNP, 100 µg ml⁻¹ of gentamycin for PAKMS591, or 50 µg ml⁻¹ of streptomycin for strains PAK-BΩ and PAK-DΩ. The polyclonal antibodies generated against the PAK pili and associated pre-immune antisera used in this study have been reported previously (Paranchych et al., 1979).

Biotinylation of P. aeruginosa cells and purified pili

Biotinylation of bacteria was performed as previously described by Yu et al. (1996) with the following modifications. Harvested cells were suspended in 5 ml of phosphate buffered saline (PBS) (pH 6.8) with 75 µl of 20 mg ml⁻¹ biotinamidocaproate N-α-hydroxysuccinimidyl ester dissolved in dimethylsulfoxide and incubated at 22°C with agitation (200 rpm) in a water bath shaker for 1 h. Cells were harvested by centrifugation (10 000 g for 10 min at 4°C) and washed four times before resuspension in 1.0 ml PBS, pH 6.8. Viable counts were performed before and after biotinylation. PAK pili were purified from PAK 2Pfs as described previously (Paranchych et al., 1979). The purity and integrity of the pili were assessed by 15% SDS-PAGE (Sambrook et al., 1989) and electron microscopy (Fig. 1C and F). The procedure used for the biotinylation of the purified pili has been previously described (Yu et al., 1996). The ability of the biotinylated pili to bind to asialo-GM₁ and GM₁ was determined as previously described (Lee et al., 1994) and the binding specificity for asialo-GM₁ was confirmed to establish the functional binding activity of the pili following biotinylation.

Stainless steel binding assay

Grade 304 stainless steel 2B finish plates (20 gauge – 1 mm thick and 7.6 × 11.5 cm) were washed in 95% ethanol for 10 min, and rinsed with distilled water. Immediately before the binding studies, coupons were washed with 20 ml of acetone for 1 min with gentle agitation and rinsed with distilled water. Coupons were then assembled into a Schleicher and Schuell MinifoldTM System (Mandel Scientific). Biotinylated viable PAK cells or purified PAK pili (biotinylated or unbiotinylated) were added (100 µl per well in replicates of six) to the stainless steel manifold and incubated at 37°C for 1 h with gentle agitation. The manifold was subsequently washed five times with 250 µl per well Buffer A (PBS pH 7.4 containing 0.05% BSA). Binding was assessed using either streptavidin-horseradish peroxidase (HRP) or polyclonal PAK antibodies and secondary goat-anti-rabbit HRP (Bio-Rad). Substrate buffer (0.01 M sodium citrate buffer pH 4.2 containing 1 mM 2,2′-Azino-bis-(3-ethylbenzthiazoline-6-sulfonic acid) diammonium salt (ABTS) (Sigma) and 0.03% (v/v) hydrogen peroxide) was added (125 µl per well) and the manifolds were incubated at room temperature (RT) for 10 min with shaking at 150 rpm. The absorbance was determined at 405 nm using a Multiskan Plus version 2.01 plate reader following transfer of the reaction solution to 96 well flat-bottomed micro titer plates (Corning).

Buccal epithelial cell assay

Buccal epithelial cell assay was completed as described by McEachran and Irvin (1985), with the following modifications. Corning 24 well tissue culture treated ELISA plates were incubated with 500 µl of a 1 µg ml⁻¹ poly L-lysine solution at 75°C overnight before washing for 15 min with PBS three times. Gluteraldehyde (25%, 125 µl per well) was incubated for 1 h at 37°C. Wells were washed as previously stated. BECs, harvested from 10 healthy volunteers, were filtered through a fine (70 µl) nylon mesh (Spectrum) and added to the prepared ELISA plate for overnight fixation at 37°C. Binding studies were performed as noted above.

Antibody inhibition studies

Biotinylated viable PAK whole cells or biotinylated purified PAK pili were mixed with 50 µl of a 10⁻² dilution of pilus-specific antibody (note that all antibodies were initially set to the same titer via an ELISA employing purified pili as an antigen) or pre-immune rabbit sera in PBS buffer, pH 7.4, with a vortex mixer and incubated for 1 h at 37°C. The cell or pili mixture was utilized for binding assays as previously described (Yu et al., 1996). Concentrations of biotinylated PAK cells or biotinylated PAK pili ranged from 0 to 3.0 × 10¹⁶ cfu ml⁻¹ and 0 to 1.5 µg ml⁻¹ respectively. The steel surface was washed five times with Buffer A, incubated with 100 µl of either a rabbit anti-IgG HRP, for the polyclonal antibody or mouse anti-IgG HRP for the monoclonal antibodies. After an 1 h incubation at 37°C, the steel surface was washed as described, and ABTS substrate solution (125 µl per well) was added for 15 min. The two pilus-specific antibodies used in this study, a polyclonal anti-PAK pili antibody (Lee et al., 1989) and monoclonal antibody PK99H (Doig et al., 1990), have been previously described. Rabbit pre-immune serum, which had previously been determined to be free of anti-P. aeruginosa antibodies by ELISA, was utilized as a control.

Peptide synthesis and competitive peptide inhibition assays

The peptides described in Table 1 were synthesized as the N-α-acetylated free carboxyl form, except for PAK(134–140) which was synthesized as the N-α-acetylated and C-terminal amide form, by solid-phase peptide synthesis and purified by reversed-phase high-performance liquid chromatography (HPLC) as previously reported (Wong et al., 1992; Wong et al., 1995). Peptides containing two cysteine residues were air-oxidized to generate the disulphide bridged form of the peptide with disulphide formation being experimentally confirmed (Campbell et al., 1995). Synthetic peptides PAK(128–144)ox, PAK(117–125), PAK(134–140), PAO(128–144)ox, PAK(128–144)oxK130I, PAO(128–144)ox_Scrambled and PAO(128–144)C129A/C142A_Scrambled were dissolved in Buffer A and incubated with either 10¹⁵ cfu ml⁻¹ biotinylated viable PAKwt cells or 0.75 µg ml⁻¹ of biotinylated purified PAK pili such that the final peptide concentration ranged from 51 nM to 51 µM. The samples were then utilized directly in a steel surface binding assay as described above.

Trypsinized BSA peptides

Bovine serum albumin (BSA) (Biotech grade Fisher) was dissolved in PBS pH 7.4, and heat denatured by boiling in a water bath for 1 h. Trypsin (50 µl of a 1 mg ml⁻¹ solution) was then added to a 10 mg ml⁻¹ heat denatured BSA solution and incubated at RT overnight with gentle agitation. Trypsin was heat inactivated via boiling water bath for 1 h before use in competitive peptide inhibition assays as listed above.

Direct binding of peptides to steel

To confirm that the pilin receptor binding domain was directly interacting with the steel surface rather than indirectly inhibiting cell or pilus binding to steel, direct binding of the synthetic receptor binding domain was assessed. The binding of the synthetic peptides to steel was determined by a modified immunoassay that has been previously described (Yu et al., 1996). Synthetic peptides PAK(128–144)ox and PAK(134–140) were prepared in Buffer A (0–51 µM). The peptides were then added directly to wells (100 µl per well in replicates of six) formed on the steel surface and incubated for 1 h at 37°C without agitation. Following five washes with Buffer A (250 µl per well), a 1:5000 dilution of PK99H was added to each well (100 µl per well) and incubated for 1 h, and washed as described. Secondary antibody Goat anti Rabbit IgG HRP (100 µl per well of a 1:3000 dilution) was added and again incubated for 1 h. ABTS substrate solution was added (125 µl per well) and allowed to incubate for 25 min before the absorbance at 405 nm was determined as described.

Acridine orange staining and microscopy

Stainless steel plates, prepared and utilized in bacterial binding studies as described above, were incubated in 1 mM Acridine Orange stain for 1 min, and thoroughly rinsed with distilled water. Coupons were visualized using a Leitz laborlux K microscope equipped with a MSP4 camera, and 40× Neoflour lens with epifluorescent illumination. Micrographs were recorded with Kodak Colormax 35 mm film, processed and digitally scanned immediately after film processing.