|Home | About | Journals | Submit | Contact Us | Français|
PilA, the major pilin subunit of Pseudomonas aeruginosa type IV pili (T4P), is a principal structural component. PilA has a conserved C-terminal disulfide-bonded loop (DSL) that has been implicated as the pilus adhesinotope. Structural studies have suggested that DSL is involved in intersubunit interactions within the pilus fiber. PilA mutants with single-residue substitutions, insertions, or deletions in the DSL were tested for pilin stability, pilus assembly, and T4P function. Mutation of either Cys residue of the DSL resulted in pilins that were unable to assemble into fibers. Ala replacements of the intervening residues had a range of effects on assembly or function, as measured by changes in surface pilus expression and twitching motility. Modification of the C-terminal P-X-X-C type II beta-turn motif, which is one of the few highly conserved features in pilins across various species, caused profound defects in assembly and twitching motility. Expression of pilins with suspected assembly defects in a pilA pilT double mutant unable to retract T4P allowed us to verify which subunits were physically unable to assemble. Use of two different PilA antibodies showed that the DSL may be an immunodominant epitope in intact pili compared with pilin monomers. Sequence diversity of the type IVa pilins likely reflects an evolutionary compromise between retention of function and antigenic variation. The consequences of DSL sequence changes should be evaluated in the intact protein since it is technically feasible to generate DSL-mimetic peptides with mutations that will not appear in the natural repertoire due to their deleterious effects on assembly.
The gram-negative opportunistic pathogen Pseudomonas aeruginosa uses polar type IV pili (T4P) to attach to various materials, to move across surfaces via twitching motility, and to initiate host colonization and biofilm formation. T4P are widely distributed among bacteria and have been most extensively studied in Neisseria spp., Escherichia coli, Vibrio cholerae, and P. aeruginosa (8, 16, 42). T4P are divided into two major groups, type IVa and type IVb pili (T4aP and T4bP, respectively); there are several differences that distinguish these subfamilies (reviewed in reference 16). Most P. aeruginosa strains express T4aP composed of one of five different variants of the 15- to 17-kDa PilA protein (37).
The crystal structures of N-terminally truncated or full-length forms of PilA from P. aeruginosa strains PAK and K122-4 have been solved (17, 18, 28, 34), as has the structure of the type IVa pilin from Neisseria gonorrhoeae MS11, called PilE (45). The pilins have a ladle-like structure, with a long, hydrophobic, kinked N-terminal alpha helix joined to a C-terminal domain of antiparallel beta-sheet architecture, terminating in a characteristic disulfide-bonded loop (DSL; also called the D-region). In a recent report describing the cryo-electron microscopy-derived ultrastructure of an assembled type IV pilus from N. gonorrhoeae, Craig and colleagues confirmed the predictions of earlier models that the N-terminal alpha helices of the subunits form the hydrophobic core of the fiber, with the hydrophilic C-terminal beta sheet and loop domains forming its outer surface (17).
P. aeruginosa T4P mediate attachment to, and twitching motility on, an astonishing array of living and nonliving surfaces, from stainless steel and plastic to living cells (15, 20, 22, 25, 27, 44), contributing to the ability of this organism to cause opportunistic infections in a wide range of hosts. Twitching motility involves cycles of pilus extension, adherence, and subsequent pilus retraction that pulls the cell body forward (51). For twitching to occur, the pilus must adhere with sufficient strength that retraction of the pilus will result in translocation of the cell, overcoming the combination of surface tension and other cell surface adhesins that hold the cell body in place.
Most bacterial pili, such as the types 1 and P pili of uropathogenic E. coli, are composed of separate structural (FimA and PapA) and adhesive (FimH and PapG) subunits, with the adhesive subunit present only at the tip of the pilus fiber (7, 32). P. aeruginosa T4P are unusual in this respect, in that the PilA subunit has been reported to act as both the main structural component and the tip adhesin (39, 50). The C-terminal DSL of the PilA subunit has been shown to mediate attachment of piliated P. aeruginosa to host cells and to abiotic surfaces such as stainless steel (25, 39, 50). This subdomain of PilA was shown by immunogold labeling studies to be exposed only at the pilus tip, suggesting that it is otherwise masked by adjacent subunits in the assembled pilus (39). These data are consistent with recent ultrastructural studies of N. gonorrhoeae T4P, which suggest that the C termini of the pilins are involved in intersubunit contacts throughout the length of the pilus fiber (17).
To address the roles of specific residues within the DSL in host cell attachment, Wong and colleagues synthesized peptides corresponding to C-terminal residues 128 to 144 of the pilins from strains PAK and KB7, as well as analogues thereof containing Ala substitutions at each position (57). The peptides were oxidized to allow disulfide bond formation and used in a competition assay, measuring their ability to block binding of biotinylated PAK pili to buccal epithelial cells. Their study confirmed earlier observations that the Cys residues involved in disulfide bond formation contributed significantly to adhesin function and implicated a number of other residues in binding. However, a single adhesinotope common to both peptides could not be defined since they have only partial sequence identity. Conserved residues contributing to conformational elements, particularly type I and type II beta turns, were found to be important while a conserved hydrophobic residue (F137 in the PAK pilin) was not crucial for binding (57).
As a prelude to studies examining the effects of sequence variation within the key DSL region on the adhesive capacity of the pilin subunit, we investigated the effects of PilA mutations on its multiple functions, including participation in protein secretion via the structurally related type II secretion (T2S) system in P. aeruginosa. A previous study (41) reported that PilA could form heterodimers with XcpT, the major pseudopilin of the Xcp T2S, and that PilA mutants were defective in T2S of proteases. In this work, the pilA gene from the laboratory strain PAO1 was mutagenized to generate single-residue variants of PilA that were expressed from an l-arabinose-inducible promoter in a pilA mutant background. This approach permitted the simultaneous interrogation of the effects of the mutations on pilin stability, assembly, and function in terms of twitching motility and pilus-specific bacteriophage susceptibility, as well as potential dominant-negative effects in the wild type upon induction. Here, we show that it is possible to identify single-residue variants of PilA that are affected in each step of pilus assembly and function.
The bacterial strains and genetic constructs used for this study are listed in Table Table1.1. Bacteria were maintained as glycerol stocks at −80°C and grown routinely on Luria-Bertani agar plates supplemented where indicated with l-arabinose at 0.2%, which was previously determined to restore wild-type levels of twitching upon complementation of the pilA mutant with its cognate gene (5) and antibiotics. Antibiotic concentrations for E. coli or P. aeruginosa were 15 or 30 mg/liter gentamicin, 100 mg/liter ampicillin or 300 mg/liter carbenicillin, and 15 or 50 mg/liter tetracycline. For elastase secretion assays (49), tryptic soy agar was supplemented with 0.2% l-arabinose and 0.05% (wt/vol) Congo red-elastin (Sigma) and cooled to 45°C with stirring until just before the plates were poured to ensure an even distribution of insoluble elastin throughout the agar.
Two anti-PilA polyclonal antisera were used in this work. The first was an anti-PilA antibody raised against sheared surface pili (anti-PilA.WP, where WP is whole pili) and affinity purified against denatured PAO1 PilA protein as described previously (36). The second antiserum was raised against an N-terminally truncated fragment of PAO1 PilA expressed in E. coli (anti-PilA.CT, where CT is C terminus).
D-TOPO-pET151-ΔN-PAO1-PilA was produced using standard recombinant DNA techniques. Briefly, primers designed to remove the first 34 amino acids of the prepilin, including the 6-residue signal sequence and the next 28 hydrophobic residues (corresponding to the first 102 bases) of the PAO1 pilA sequence (forward, 5′-CAC CCA GTA TCA GAA CTA TGT TGC GCG T-3′; reverse, 5′-TTA GTT ATC ACA ACC TTT CGG AGT GAA CAT CG-3′) were used to amplify the gene from PAO1 chromosomal DNA. PCR products were purified using a QIAquick gel extraction kit (Qiagen) and ligated into the D-TOPO-pET151 vector, which encodes a tobacco etch virus (TEV) protease-cleavable N-terminal hexahistidine tag (Invitrogen). Positive clones were sequenced (ACGT Corp.), and verified constructs were introduced into E. coli Origami (DE3) cells (Novagen) to permit formation of the C-terminal disulfide bond in the normally reducing environment of the cytoplasm. A 5-ml overnight culture grown at 37°C with shaking (225 rpm) was used to inoculate 1 liter of LB broth supplemented with 100 mg/liter ampicillin, and the culture was grown at 37°C with 225 rpm shaking to an optical density at 600 nm of 0.6. The temperature was then reduced to 16°C, and protein expression was induced by adding isopropyl β-d-thiogalactopyranoside at a final concentration of 1.0 mM.
Cells were grown for 18 h and harvested by centrifugation (3,200 × g at 4°C for 15 min). The pellet was resuspended in 20 ml of lysis buffer (20 mM Tris, pH 8.0, 500 mM KCl, 10% glycerol, 0.1% lauryl dimethylamine N-oxide) and lysed by three passes through a French press at 1,200 lb/in2. The lysate was then centrifuged (48,258 × g at 4°C for 45 min) to remove cellular debris, and the protein was bound to a charged 5-ml Ni Sepharose High Performance column (GE Healthcare Canada) preequilibrated with 25 ml of Ni buffer A (20 mM Tris, pH 8.0, 500 mM KCl, 10 mM imidazole). Step washes were performed with 25 mM, 40 mM, and 55 mM imidazole, followed by elution of the protein with Ni B buffer containing 300 mM imidazole. The eluted protein was dialyzed overnight at 4°C into TEV digestion buffer (20 mM Tris, pH 8.0, 150 mM KCl). The dialyzed protein was incubated with TEV protease at a final concentration of 1 μg/ml for 2 h at room temperature. A second Ni purification was performed as above to separate the cleaved protein from the His6 tag. The cleaved protein was then concentrated to 5 mg/ml, dialyzed in phosphate-buffered saline (PBS), and submitted to Cedarlane Laboratories (Burlington, ON, Canada) for polyclonal antibody production.
PCR with mutagenic primers was used to generate pilA alleles encoding alterations in the amino acid sequence of the PAO1 PilA DSL region. A single upstream primer (Table (Table2)2) was used (except where indicated below) in conjunction with specific downstream primers encoding the change of interest (Table (Table2)2) to amplify pilA using PAO1 chromosomal DNA as a template. The PCR products were purified using Qiaprep columns (Qiagen), digested with restriction enzymes, and ligated into similarly digested pBADGr (38), a modified version of the l-arabinose-inducible vector pMLBAD (40), wherein the dhfrII gene encoding trimethoprim resistance was disrupted with the aacC1 gene from pUCGm (47). To generate the D133N, D133E, P132M, and P132R mutants, site-directed mutagenesis (Quikchange; Stratagene) was used as per the manufacturer's instructions to introduce the desired mutations using the mutagenic primers listed in Table Table22 using the wild-type pilA gene in pBADGr as a template. All constructs were verified by restriction enzyme digestion and DNA sequence analysis. Verified constructs were introduced into the PAO1 NP strain (where NP indicates no pili) (Table (Table1)1) by electroporation and selected on LB plates containing 30 mg/liter gentamicin. Where indicated in Table Table1,1, the same constructs were introduced into PAO1 NP-pilT, a pilA pilT double mutant generated as described previously (53) by disruption of the pilT gene in the NP background with a Flp recombinase target (FRT)-flanked gentamicin resistance cassette, followed by Flp recombinase-mediated excision of the resistance marker (29).
Twitching motility was measured as described previously (24, 48), and the resulting zones of twitching motility were visualized by carefully removing the agar and staining the bacteria adhering to the polystyrene petri plate with 1% crystal violet for 10 min at room temperature, followed by a brief rinse with tap water to remove unbound dye. ImageJ software (NIH) was used to measure and calculate average areas of the resulting twitching zones for quantitative comparative data. Assays were performed at least three times, with a minimum of six replicates per assay.
Surface proteins (flagella and pili) were isolated using the methods of Castric (14) with modifications. Bacteria were streaked in a grid pattern on LB agar plates containing 30 mg/liter gentamicin and 0.2% l-arabinose and incubated overnight at 37°C. Two plates per sample were used. The bacteria were gently scraped from the agar surface using a sterile coverslip and resuspended in 2 ml of sterile PBS (pH 7.4) per plate, and surface proteins were sheared by vigorous vortexing for 30 s. The suspension was transferred to two 1.5-ml microcentrifuge tubes and centrifuged for 5 min at maximum speed to pellet the cells, which were retained for whole-cell lysate Western immunoblotting (below). The supernatant was transferred to a new tube and centrifuged for an additional 25 min at maximum speed at room temperature. To precipitate the sheared proteins, 1/10 volume each of 5 M NaCl and 30% polyethylene glycol (molecular weight range of 8,000) was mixed with the supernatant, and the samples were incubated on ice for 60 min. Samples were centrifuged at maximum speed in a microcentrifuge for 25 min at 4°C. After the supernatant was discarded, the resulting pellets were resuspended in 2× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading dye (125 mM Tris, pH 6.8, 2% [wt/vol] 2-mercaptoethanol, 20% [vol/vol] glycerol, 0.001% [wt/vol] bromophenol blue, 4% [wt/vol] SDS), boiled for 5 min, and resolved on a 15% one-dimensional SDS-PAGE minigel with a prestained Benchmark Protein Ladder (Invitrogen). The proteins were visualized using Coomassie blue dye.
After lysates were vortexed to remove surface proteins (above), the harvested cell pellet was resuspended in sterile PBS to a final optical density of 0.6 at 600 nm. A 200-μl aliquot of the cell suspension was transferred to a 1.5-ml microcentrifuge tube, and the cells were harvested by centrifugation at maximum speed for 5 min. The supernatant was removed, and the cell pellet was resuspended in 150 μl of 1× SDS-PAGE sample buffer and boiled for 5 min, and 8 μl per sample was separated on a 15% SDS-PAGE minigel as described above. After separation, the proteins were transferred to nitrocellulose for Western immunoblot analysis with a rabbit polyclonal antibody raised against and affinity purified with PilA. The blot was blocked with 5% skim milk in PBS overnight at 4°C, incubated with a 1/5,000 dilution of primary antibody for 60 min at room temperature, washed with PBS, and incubated with a 1/5,000 dilution of goat anti-rabbit alkaline phosphatase-conjugated secondary antibody for 60 min at room temperature. After a washing step, the blot was developed with BCIP/NBT (5-bromo-4-chloro-3-indolylphosphate/nitroblue tetrazolium) (Sigma) as per the manufacturer's instructions.
Secretion of elastase was monitored by growing the strains of interest on LB agar containing 0.5% (wt/vol) Congo red-elastin conjugate (Sigma), 0.2% arabinose, and 30 mg/liter gentamicin and by measuring the resulting zones of elastin degradation. xcpY and pilD mutants defective in T2S were used as negative controls (46, 52).
The DSL regions of the five type IVa pilin alleles in P. aeruginosa (37), as well as those from a number of other T4aP-expressing bacterial species were manually aligned from the distal Cys residue (Table (Table3).3). While there is substantial sequence heterogeneity in this region, both in the number and identity of residues present, there are specific residues which are highly conserved: the two Cys residues forming the disulfide bond, a Pro at position −3 relative to the distal Cys, and a hydrophobic residue at position −5 (Table (Table3).3). The C-terminal Pro-X-X-Cys motif forms a type II beta turn (a ~180° turn with a hydrogen bond between the Pro and Cys residues), previously shown to be important for binding and to be the target of antibodies that block attachment (10, 12, 28). The biochemical nature of the X residues is not well conserved since small, bulky, polar, nonpolar, and positively and negatively charged amino acids are all represented, depending on the strain. The type IV pilins from the Haemophilus-Actinobacillus-Pasteurella group (HAP group) of organisms have an additional residue in this region, Pro-X-X-X-Cys (Table (Table3),3), in addition to other unusual features, suggesting that they may represent a third class of type IV pili, which we denote T4cP.
Mutagenic PCR primers were used to create variants of PilA with Ala substitutions at each position through the DSL, as well as a deletion of all residues between the two Cys residues (ΔDSL) and partial replacement of the DSL with the V5 epitope tag (V5DSL) (Fig. (Fig.1).1). The ability of each PilA variant to complement twitching motility of the PAO1 pilA mutant was measured using an agar subsurface assay. The C128A and the C141A mutants, as well as the ΔDSL and V5DSL mutants, were unable to complement twitching motility (Fig. (Fig.2A).2A). The remaining mutant pilins were able to complement twitching motility to various extents. Analysis of susceptibility to the pilus-specific bacteriophage PO4 showed that strains able to twitch were killed by the phage while strains unable to twitch were phage resistant (data not shown). We examined the wild type, NP mutant, and complemented mutants for changes in elastase secretion using both skim milk and Congo red-elastin-containing medium, but no significant differences could be detected as both the wild type and NP mutant had similar levels of elastolytic activity (data not shown).
Lack of twitching motility can arise from loss of pilus retraction, reduced surface assembly of pili, and/or pilin instability. To distinguish between these possibilities, sheared surface proteins from each mutant were examined by SDS-PAGE. No surface pili could be recovered from any of the nontwitching mutants (Fig. (Fig.2B).2B). Examination of strains expressing other PilA variants revealed a decrease of 10 to 80% in the amount of recoverable surface pili relative to the strain expressing unmodified PilA, which did not correlate directly to changes in the size of the resulting twitching motility zones (Fig. (Fig.2B).2B). No gross differences in the morphology of mutant pili compared with those of the wild type could be detected by transmission electron microscopy (data not shown).
To determine if the loss of recoverable surface pili was related to pilin instability, whole-cell lysates of each strain were examined by Western blotting using two different anti-PilA antibodies, one raised against whole purified pili (anti-PilA.WP) and the other against a C-terminal fragment of PilA expressed and purified from E. coli (anti-PilA.CT). Figure Figure2C2C shows that using anti-PilA.WP, the levels of pilin in the nontwitching mutants C128A, C141A, and V5DSL appeared reduced relative to the wild-type level, while the ΔDSL mutant pilin was undetectable. In contrast, probing the same samples with anti-PilA.CT (Fig. (Fig.2D)2D) showed that the C128A, C141A, ΔDSL, and V5DSL proteins were present in whole-cell lysates although only C141A was present at wild-type levels. Therefore, the lack of surface pili in these strains is related to their inability to assemble into surface-exposed fibers rather than lack of pilin expression or complete degradation of unstable pilins. The remaining mutants, regardless of their levels of motility, had intracellular pilin levels similar to the wild-type level (Fig. 2C and D). Attempts to visualize the V5DSL pilin using monoclonal antibodies directed to the V5 epitope were not successful.
The reduced levels of surface pili on the D133A, F136A, and P138A mutants, which had nearly wild-type levels of pilins in whole-cell lysates (Fig. (Fig.2),2), suggested potential defects in pilus assembly. To further explore this phenotype, D133N (loss of negative charge) and D133E (retention of charge but a longer side chain) mutants of PilA were generated. The D133N mutant had twitching defects similar to those seen for D133A while motility of the D133E mutant was similar to that of the strain complemented with wild-type PilA (data not shown). Analysis of sheared surface preparations from these strains showed that only the D133E mutant had levels of recoverable surface pili similar to those of the positive control (Fig. (Fig.3A).3A). Western blot analyses of whole-cell lysates showed that the mutants had levels of intracellular pilins similar to the level of the positive control (Fig. (Fig.3B).3B). Therefore, a negative charge at residue 133 is important for assembly of PAO1 pilins into recoverable surface pili.
The C-terminal type II beta turn implicated in pilus-mediated adherence was observed both in X-ray crystal structures of the P. aeruginosa PAK and K122-4 pilins and in nuclear magnetic resonance (NMR) structures of synthetic PAK and KB7 DSL-mimetic peptides (6, 10, 18, 28), suggesting that it is a native structural feature of this subdomain. Our alignment of type IVa pilins from various species show that this motif is one of the few highly conserved aspects of the DSL region (Table (Table3).3). Although previous studies using peptides corresponding to the DSL region implicated a number of residues in binding to surfaces (25, 57), its potential role in pilus assembly was not tested as this question can only be addressed in the context of a full-length pilin. In addition to the type II beta-turn mutants P138A, K139A, G140A, and C141A examined above, we generated P138M (large, hydrophobic side chain) and P138R (large, positively charged side chain) variants or deleted/inserted single amino acids to generate ΔK139, ΔG140, and PAKGC (wherein an Ala residue was inserted after P138, mimicking the sequence of the HAP group T4cP pilins) (Table (Table3)3) and tested the effect of these changes on pilus assembly. All of these mutant pilins were significantly or completely defective in their ability to complement twitching motility of the PAO1 pilA mutant in an agar subsurface twitching assay (data not shown) due to a >10-fold decrease in recoverable surface pili (Fig. (Fig.4A),4A), with no pili recoverable for the PAKGC variant. However, all of the mutant pilins, including the PAKGC variant, were present at wild-type levels in whole-cell pools when they were probed with the PilA.CT antibody (Fig. (Fig.4B4B).
Loss of recoverable surface pili can be caused by altering the balance between pilus assembly (extension) and pilus disassembly (retraction) events. If the rate of pilus extension is decreased while the rate of pilus retraction is unaffected, the net result is a decrease in surface piliation. By inactivating the pilT gene encoding the retraction ATPase, it is possible to distinguish between mutations that abrogate pilin assembly and those that result in slower rates of assembly (1, 5, 43, 54-56). Therefore, a pilA pilT double mutant of PAO1 (NP-pilT) was generated and complemented with pilin variants that resulted in surface piliation defects in the NP background (C128A, C141A, ΔDSL, V5DSL, PAKGC, ΔK139, ΔG140, D133A, D133N, P138A, P138M, and P138R).
Expression of the PAKGC, ΔK139, and ΔG140 pilins in the pilA pilT double mutant revealed that the ΔK139 pilins were able to assemble into recoverable surface pili but to a much more limited extent than the wild-type PilA protein. However, deletion of the G140 residue or insertion of Ala into the type II beta turn resulted in pilins that were unable to assemble into surface fibers (Fig. (Fig.5).5). Similarly, complementation of the NP-pilT strain with C128A, C141A, ΔDSL, or V5DSL pilins did not result in recovery of surface pili, showing that these variants are not able to assemble. Variants showing assembly defects in the NP mutant (D133A, D133N, P138A, P138M, and P138R) continued to do so in the double-mutant background. However, there was an increase in recovery of P138A and P138M pili relative to the levels obtained when these variants were expressed in the NP background, but presence of the positively charged Arg residue (P138R) had deleterious effects on pilus assembly.
To determine whether mutant pilins demonstrating assembly defects could have dominant-negative effects on pilus assembly in the wild type, the D133A, D133N, P138A, P138M, P138R, ΔK139, ΔG140, and PAKGC constructs were introduced into the PAO1 parent strain, with the wild-type PilA protein acting as a positive control. Twitching motility of the recombinant strains was compared after growth on LB agar without arabinose supplementation (uninduced) or on LB agar with 0.2% arabinose (induced). Upon arabinose induction, twitching motility was unaffected in the strain expressing wild-type PilA (P ≤ 0.08) but significantly reduced in the strains expressing mutant pilins (Fig. (Fig.6).6). In each case, the reduction in twitching motility on 0.2% arabinose plates was correlated with decreased levels of recoverable surface pili (data not shown).
From this study, we conclude that single-residue changes in the C-terminal DSL of PilA can impact twitching motility, not necessarily through changes in adhesive capacity (although the S130A mutant, which has near wild-type piliation but significantly reduced motility, may represent a less adhesive variant) but through effects on pilus assembly. Although the structure of the PAO1 pilin is not yet available, we examined two published structures of the closely related (73% similarity over 144 residues) strain PAK pilin that were solved independently by separate laboratories from crystals belonging to different space groups (18, 28). One structure contains the full-length protein (Protein Data Bank [PDB] 1OQW) while the other has an N-terminal truncation of the hydrophobic alpha helix (PDB 1DZO). Despite these differences, the structures are superimposable, with a Cα root mean square deviation of <1 Å (Fig. (Fig.7A).7A). Inspection of crystal packing features in the full-length versus truncated structures revealed that the DSL is not constrained by crystal contacts in either case but, instead, that its tertiary structure is dictated both by conformational elements (including the type II beta turn near its C terminus) and interactions with the remainder of the protein (Fig. (Fig.7B).7B). In contrast, the NMR solution structure (PDB 1NIM) of a disulfide-bonded peptide corresponding to the PAK pilin DSL (11) could not be superimposed on the equivalent domain in the crystal structure even though the architecture of its type II beta turn is maintained by intramolecular hydrogen bonding (Fig. (Fig.7C).7C). Because the DSL has a substantially different structure in the protein than in the peptide, studies using peptides should be interpreted with caution. Furthermore, it is important to assess the effects of DSL sequence alterations on stability, assembly, and functions of the whole protein since it is possible to synthesize DSL peptides unlikely to appear in the natural repertoire due to their deleterious effects on those key properties.
Mutation of either of the Cys residues forming the disulfide bond of the DSL, deletion of the intervening residues, or insertion of several residues (V5DSL) resulted in a decrease in PilA detection when a polyclonal antibody generated against purified pili was used, but the mutant proteins were readily detected by an antibody raised against the C terminus of PilA purified from E. coli (Fig. 2C and D). The fact that the ΔDSL variant lacking all of the residues between the two Cys residues is not recognized at all by the PilA.WP antibody even though, based on reactivity with the PilA.CT antibody, it is present at levels similar to those of some of the other detectable variants (i.e., C128A) suggests to us that disruption of DSL structure does not cause dramatic defects in stability in most cases but, instead, results in loss of epitopes that are immunodominant in the context of whole pili. In contrast, antisera raised against individual subunits or subunit fragments recognize epitopes both within and outside of the DSL and are therefore better reporters of protein levels. Loss of antibody reactivity due to changes in key epitopes could be misinterpreted as defects in protein stability. Differences in the form of antigen used to prepare antipilin antibodies may explain why some investigators previously reported defects in pilin “stability” when mutations of Cys to Ser were introduced or when the periplasmic disulfide-bond isomerase DsbA was inactivated (21, 26), while others did not observe this phenomenon. It is also possible that the pilins of some species are more susceptible to degradation in the absence of stable disulfide bond formation.
Although large perturbations in DSL sequence (deletion of all intervening residues in ΔDSL or their partial replacement with the V5 epitope) resulted in loss of motility, loss of surface piliation, and loss of protein detection by pilus-specific antibodies, the stable expression of these variants was demonstrated by their detection by pilin-specific antibodies. Other studies showed that transposon insertion into pilA at a position corresponding to residue 125, resulting in the loss of the entire DSL (in P. aeruginosa strain PA103), or replacement of the last nine residues, including the second Cys, in the PAK pilin resulted in attenuated function (23, 33). The PA103 mutants were not tested to determine if surface pili were present although they were reported to be phage susceptible, a hallmark of surface-exposed, retractable pili (33). The mutant PAK pilin was reported to form surface pili which were less able to promote attachment to A549 pneumocytes than wild-type pili (23) though in our hands loss of either Cys residue precluded pilin assembly, even in a retraction-deficient background. A study of Neisseria pilin variation also reported that generation of a ΔDSL-type mutation in PilE, the equivalent of PilA, caused loss of surface piliation and incompetence for DNA transformation (30), while a more recent investigation (19) indicated that accumulation of full-length assembly-deficient monomers was not seen. The latter observation was made on the basis of Western blot analyses, which as demonstrated here depend on the ability of the antibody used to recognize wild-type and mutant pilins with similar efficiencies in order to accurately report pilin levels.
Several of the other PilA point mutations generated in this study had effects on pilus assembly. The D133 residue is conserved in both PAO1 and PAK pilins (D134 in PAK). The requirement for a negatively charged residue at this position for assembly was confirmed by demonstrating that D133E, but not D133N, pilins conferred wild-type levels of twitching motility and that expression of the D133A and D133N pilins in the NP-pilT double mutant could not restore wild-type levels of surface pili. The side chain of D133 projects from the bottom of the DSL (Fig. (Fig.1),1), where it could potentially form a salt bridge with a conserved K110 residue at the bottom of the adjacent loop in the same subunit, thereby constraining the DSL. A second positively charged residue in the same area, K114, is also conserved in both PAK and PAO1 pilins, and it is possible that together K110 and K114 could cause unfavorable conformational changes in the absence of a stabilizing negative charge from D133. Alternatively, D133 could interact with positively charged residues in adjacent subunits during assembly to stabilize intersubunit interactions. We note that F136 similarly projects from the bottom of the DSL, where it is adjacent to the N-terminal hydrophobic alpha helix. The conservation of both the N-terminal helix and the hydrophobic residue among pilins from a wide variety of bacteria (Table (Table3)3) suggests that it is important for conformation of the DSL. Altering this residue to Ala caused a decrease in pilus assembly (Fig. (Fig.2B),2B), suggesting that the longer side chain at this position is important, a notion supported by the presence of Leu at this position in some species. Interestingly, in the HAP pilins although there is an additional residue inserted in the type II beta turn, the −5 position of the conserved Phe residue relative to the distal Cys is preserved (Table (Table33).
Alteration of the highly conserved C-terminal type II beta turn had marked effects on assembly. This structural feature was the focus in previous studies of pilus-mediated adherence since it is present in peptide mimics reported to bind to abiotic surfaces and host cells and because it forms part of the epitope for antibodies able to block pilus-mediated attachment (9, 12, 13, 25, 57). It is possible that alteration of this region affects either the self-interaction of PilA subunits or their interactions with other components of the T4P system that are important for initiation or control of assembly, such as the minor pilins (2, 4). This latter hypothesis is supported by the similarities between the reported phenotypes of minor pilin mutants and those of many of the assembly-defective PilA mutants presented here: both are deficient in surface pili but have wild-type levels of pilins in whole-cell lysates and in some cases can be forced to express surface-exposed pili when retraction is blocked by mutation of pilT (55).
We noted that the P138M and P138R mutants have similar reductions in surface piliation in a pilT-replete background, but piliation of the P138M variant is increased compared with P138R in the pilT-deficient strain. We interpret the data to mean that P138M pilins may assemble more slowly than the wild-type proteins, and therefore in a pilT-replete background, the balance between extension and retraction is skewed, with the net result that the pili are retracted. In a pilT-deficient strain, it is possible to trap P138M pili on the surface although the total amount of pili is reduced compared to the positive control. In contrast, the P138R mutants appear to have more serious assembly defects that reduce the total amount of surface piliation even in a pilT-deficient background. Removing or adding a single residue to the type II beta turn was generally deleterious although the ΔK139 pilins could assemble to some extent, as seen in the NP-pilT double-mutant background (Fig. (Fig.5).5). This finding suggests that the T4cP pilins expressed by the HAP group of bacteria (Table (Table3)3) would not function in Pseudomonas if acquired by horizontal gene transfer.
Although the PilA protein was previously reported by Lu and colleagues (41) to participate in the T2S of proteases in P. aeruginosa, we observed no significant differences in the levels of elastase secretion between the wild type and NP mutant in elastin degradation assays. It is possible that strain-specific differences are responsible for the discrepancy because Lu et al. used the PAK strain of P. aeruginosa for their work while we used PAO1.
When considered on the basis of their overall sequence identity, the type IVa pilins are a heterogeneous group of proteins with only a few highly conserved features, including a characteristic leader sequence that is recognized and cleaved by the prepilin peptidase PilD, the hydrophobic N-terminal alpha helix that forms the core of the pilus fiber, the beta sheet region, and the conserved elements of the C-terminal DSL as described here. The low sequence identity among the pilins, especially in the C termini, likely reflects their exposure on the cell surface and selection for new variants that permit adaptation to new niches or allow escape of immune detection. However, the observed variety in primary sequence also suggests an impressive ability to tolerate mutations in specific domains without loss of function(s), a hypothesis supported here in that most changes did not abrogate motility. Investigation of the limits of pilin sequence flexibility and its relation to protein structure will aid in our understanding of the minimal structural requirements for successful assembly of functional pili and shed light on the capacity of more conserved components of the T4P assembly systems to accommodate pilins of novel sequence. We noted previously (5) that other than the pilin itself, the least conserved components of the T4aP system in P. aeruginosa are the minor pilins, which are required for pilus assembly (3, 4, 55), possibly through interactions with PilA. Therefore, at least some of the observed sequence variation among pilins may reflect the results of coevolution of pilins and minor pilins to permit the continued function of T4P.
We thank Katy Howard for assistance in construction of some of the mutants for this study and Sonali Fonscca for testing the effect of arabinose titration on twitching motility.
This work was funded by grants to L.L.B. from the Canadian Institutes of Health Research (MOP49577 and MOP86639). L.L.B. is the recipient of a CIHR New Investigator award.
Published ahead of print on 28 August 2009.