Pseudomonas aeruginosa Type IV Pilus Expression in Neisseria gonorrhoeae: Effects of Pilin Subunit Composition on Function and Organelle Dynamics

Abstract

Type IV pili (TFP) play central roles in the expression of many phenotypes including motility, multicellular behavior, sensitivity to bacteriophages, natural genetic transformation, and adherence. In Neisseria gonorrhoeae, these properties require ancillary proteins that act in conjunction with TFP expression and influence organelle dynamics. Here, the intrinsic contributions of the pilin protein itself to TFP dynamics and associated phenotypes were examined by expressing the Pseudomonas aeruginosa PilA^PAK pilin subunit in N. gonorrhoeae. We show here that, although PilA^PAK pilin can be readily assembled into TFP in this background, steady-state levels of purifiable fibers are dramatically reduced relative those of endogenous pili. This defect is due to aberrant TFP dynamics as it is suppressed in the absence of the PilT pilus retraction ATPase. Functionally, PilA^PAK pilin complements gonococcal adherence for human epithelial cells but only in a pilT background, and this property remains dependent on the coexpression of both the PilC adhesin and the PilV pilin-like protein. Since P. aeruginosa pilin only moderately supports neisserial sequence-specific transformation despite its assembly proficiency, these results together suggest that PilA^PAK pilin functions suboptimally in this environment. This appears to be due to diminished compatibility with resident proteins essential for TFP function and dynamics. Despite this, PilA^PAK pili support retractile force generation in this background equivalent to that reported for endogenous pili. Furthermore, PilA^PAK pili are both necessary and sufficient for bacteriophage PO4 binding, although the strain remains phage resistant. Together, these findings have significant implications for TFP biology in both N. gonorrhoeae and P. aeruginosa.

Type IV pili (TFP) are proteinaceous surface structures found ubiquitously in gram-negative species of medical, environmental, and ecological importance. TFP are defined by their shared structural, biochemical, and morphological features and a highly conserved biogenesis pathway (26). In line with these conserved features, trans-species complementation of various TFP biogenesis mutants has been achieved in a number of cases. Trans-species complementation was first documented when the pilin subunit protein from Dichelobacter nodosus was expressed as TFP in Pseudomonas aeruginosa (11). D. nodosus TFP purified from this heterologous source engenders immunity to ovine foot rot and recombinant D. nodosus TFP remains in use today as a vaccine (12). This same methodology was subsequently exploited with success in a number of studies involving the expression of pilins from Moraxella bovis (3), Neisseria gonorrhoeae (20), Escherichia coli (33), and Pseudomonas syringae (29) as TFP in P. aeruginosa. Similar gene swapping and complementation studies of the related type II secretion systems also revealed relaxed specificity with heterologous expression of pilin-like proteins of the PulG family resulting in the formation of pseudopilus structures analogous to TFP (10, 36).

Strains expressing heterologous pilin subunits provide unique opportunities to examine relationships between TFP, accessory factors, and associated phenotypes. The latter include colonization of biotic and abiotic surfaces, motility across solid surfaces, multicellularity, and susceptibility to bacteriophage infection, as well as horizontal gene transfer via transformation and conjugation. However, these phenotypes are not universally seen in TFP-expressing species. This may reflect constraints associated with structural features unique to particular pilin subunits or the requirement for ancillary factors that act in concert with an intact TFP biogenesis pathway to impart function. TFP pilin subunits from PAK and PAO strains of P. aeruginosa pilin have been shown to complement pilin-null mutants of Pseudomonas stutzeri in DNA uptake required for transformation (16), as does the P. aeruginosa PAK pilin subunit when expressed in N. gonorrhoeae (2). It is important to note that competence for natural transformation has not been documented in P. aeruginosa strains. Since DNA uptake in both these backgrounds requires an intact TFP biogenesis pathway and a pilin subunit capable of being assembled, it is presumed that heterologous pili are expressed in these cases, but this has yet to be demonstrated. Heterospecific complementation of type II secretion defects likewise suggest that PulG type pseudopilins function by virtue of their abilities to oligomerize into pseudopilus polymers (10, 36). In both the TFP-associated competence and secretion systems, functionality is not attributable to the unique structural features of individual pilins but rather correlates best with the assembly proficiencies of the subunits.

Numerous bacteriophages utilize pili as a primary receptor to infect cells. Infection here involves binding of phage particles to the sides or tip of the pilus and retraction of the pilus, which brings the phage into contact with a second, cell surface-associated receptor. Phage components governing the molecular recognition of pili have been well defined in a number of systems. For example, minor coat proteins of the g3p family act as pilus receptors in many filamentous phages (8, 18, 19). The structural features and components of pili that participate in these interactions have yet to be defined in any system. With regard to TFP-bacteriophage systems, strain P. aeruginosa PAK and the pilus-dependent bacteriophage PO4 have been utilized extensively. In these studies, the expression of a remarkably diverse group of exogenous TFP pilin subunits in P. aeruginosa PAK backgrounds (lacking endogenous pilin-pilA) has been reported to restore twitching motility (indicative of pilus retraction) and productive PO4 infection (29, 37). Similarly, three distinct pilin genes were reported to restore PO4 sensitivity in a pilA strain of P. stutzeri (16). Based on the reported relaxed specificity of PO4 phage vis-á-vis subunit composition, some have surmised that a “minor” pilus associated protein might actually be serving as the primary receptor (29). Alternatively, PO4 phage might recognize a degenerate motif widely distributed in TFP polymers or pilin subunits. To further cloud this issue, one study reported that P. aeruginosa strain PAO is readily susceptible to PO4 (5), while another concluded that strain PAO only became PO4 sensitive when complemented by the PAK pilin gene (37). The molecular contribution of TFP to phage binding and recognition in P. aeruginosa, or for that matter in any TFP-expressing species, thus remain a matter of controversy.

Another issue of interest in TFP biology relates to the mechanisms by which pili promote adherence to mammalian tissue. TFP-mediated adherence of N. gonorrhoeae to human epithelial cells requires the coordinated expression of PilC, which has intrinsic adhesin properties, and six pilin-like proteins that appear to impact on adherence by influencing PilC activity or trafficking (39, 40, 42). Each of these proteins copurify with TFP, and all but the PilV pilin-like protein impact on TFP dynamics by promoting extension and/or polymerization events in the presence of the PilT pilus retraction ATPase. In contrast, TFP-mediated epithelial cell adherence exhibited by P. aeruginosa PAK is thought to require a receptor-binding domain located within residues 128 to 144 of the C-terminal region of the PilA pilin subunit itself (21). This receptor-binding domain is only exposed at the tip of the pilus (23). To date, no study has directly examined adherence phenotypes related to heterologous expression of PilA as TFP.

In an effort to delineate the potential contributions of pilin subunit structure and chemistry to TFP-associated phenotypes, we undertook a systematic assessment of phenotypes imparted by the expression of P. aeruginosa PilA^PAK pilin in N. gonorrhoeae.

MATERIALS AND METHODS

Bacterial strains and plasmids.

Gonococcal strains used in the present study are described in Table 1. Antibiotics were used at concentrations previously described for selection of gonococcal transformants (40). E. coli HB101 was used for plasmid propagation. Isolation and purification of plasmid DNA were performed using QIAGEN columns according to the manufacturer's specifications (QIAGEN, Chatsworth, CA). The plasmid pPilE::cat (17) was used to inactivate the wild-type pilE locus, creating the strain KS45. The pilT_ind pilC2_off (phase-off) mutant (KS56) was isolated by a procedure described earlier (42).

TABLE 1.

Bacterial strains considered in this study

Strain	Parental strain	Relevant genotypea	Source or reference
VD300	MS11		22
N400	VD300	recA6(tetM)	34
N401	N400	recA6(kan)	41
MW4	N401	pilTindc	41
MW7	N401	pilTindpilC1::ermC pilC2::cat	41
MW24	N401	pilEind	41
GE101	VD300	pilEind	43
GE107	GE101	pilEindiga::pilA	2
GE200	N400	iga::pilE	27
GE45	MW4	iga::pilE pilTind	1
KS45b	N400	pilE::cat	This study
KS46	GE200	iga::pilE pilE::cat	This study
KS47	GE45	iga::pilE pilE::cat pilTind	This study
KS48	N400	iga::pilA	This study
KS49	MW4	iga::pilA pilTind	This study
KS50	MW24	pilEindiga::pilA	This study
KS51	KS50	pilEindiga::pilA pilT::cat	This study
KS52	KS48	pilE::cat iga::pilA	This study
KS53	KS49	pilE::cat iga::pilA pilTind	This study
GV6	MW4	pilTindpilVfs	39
KS54	GV6	pilTindpilVfs iga::pilA	This study
KS55	KS54	pilTindpilVfs iga::pilA pilE::cat	This study
KS56	MW4	pilTindpilC2offd	This study
KS57	KS56	pilTindpilC2off iga::pilA	This study
KS58	KS57	pilTindpilC2off iga::pilA pilE::cat	This study

^aThe genotypes are indicated here with antibiotic resistance markers and other relevant genotypes, although they are mentioned only as simple mutant alleles throughout the text.

^bConstruction of the plasmid used to transform N400 to make strain KS45 was described previously (17).

^cpilT_ind is an IPTG-inducible allele of pilT.

^dpilC2_off is a phase-off variant of the pilC2 allele (42).

Characterization of twitching motility.

Twitching motility was assessed by observation of cells at the periphery of colonies using a Stereozoom 7 (Bausch and Lomb) stereomicroscope as well as by the slide culture method, in which cells are inoculated onto GC agar slices on microscope slides, covered with a coverslip, and visualized under a Zeiss phase microscope using a ×40 objective lens (41).

TFP retraction assay.

For retraction experiments, 3-μm silica beads (Polysciences) were coated with poly-l-lysine and adsorbed to glass coverslides by centrifugation. Then, 2-μm carboxylated latex beads (Polysciences) were added without further treatment to a suspension of gonococci, which were then mounted on a microscope slide and sealed. The optical tweezers system, including calibration methods and the determination of the velocity-versus-force curve, has been described previously (25).

Epithelial cell bacterial adherence.

Primary cultures of human corneal epithelial cells were established (35), and adherence assays of gonococcal strains to the human corneal epithelial cells were performed as described previously (39).

Pilus purification.

Pili were purified by the ammonium sulfate procedure as previously described (41) except that the bacterial cells were collected from two heavily streaked petri dishes and suspended in 1 ml of 0.15 M ethanolamine (pH 10.5) and the centrifugation time lengths were reduced to 5 min.

SDS-PAGE, immunoblotting, and staining.

Procedures for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), immunoblotting, and Coomassie blue staining have been described previously (15), and silver staining SDS-PAGE analyses were performed according to standard techniques. PilA, PilE, PilC, and PilV were detected by immunoblotting of pilus preparations and whole-cell lysates using specific rabbit polyclonal antibodies and alkaline phosphatase-coupled goat anti-rabbit antibodies (Tago). The PilA-specific sera was a gift from E. Gotschlich (Rockefeller University), and the PilC-specific sera was a gift from A.-B. Jonsson (Uppsala University). The PilV- and PilE-specific sera were described previously (1, 39).

Immunofluorescence microscopy.

Immunofluorescence microscopy was performed as described previously (40), except that the gonococcal strains were grown up to an optical density at 550 nm of 0.2 before they were fixed on poly-l-lysine-coated glass coverslips. The gonococcal pili were labeled using rabbit serum raised against purified TFP from strain N400 (9), and the PAK pili were labeled with PilA-specific antiserum. The images shown correspond to fields that are representative of the overall observations.

Immunogold and transmission electron microscopy.

Sample grids were prepared by touching carbon-coated Formvar copper grids to individual bacterial colonies grown on GC agar (18 h, 37°C, 5% CO₂) and fixed with 0.5% glutaraldehyde in phosphate-buffered saline (PBS; pH 7.4) for 5 min. Grids were washed three times on drops of PBS and subsequently negatively stained with 0.5% ammonium molybdate in water for 5 min before being viewed in a Philips CM100 transmission electron microscope. For immunogold labeling the grids with fixed bacteria were first blocked with 0.5% newborn calf serum and then incubated with rabbit anti-pilin PilA-specific antibody (dilution, 1:500) for 10 min at room temperature. After four washes on drops of PBS, the grids were incubated with gold-conjugated protein A (5 nm). After three rinses on drops of PBS, followed by three rinses on drops of water, the grids were stained 4 min with uranyl acetate. The same procedure was repeated using rabbit antisera specific for PilE: lots 2-66 and 904 (dilution 1:100), lot T36 (1:10), and lot T40 (1:20). The T36 and T40 antibodies have been reported to specifically recognize epitopes exposed at the tips of the PilE pilus structure (14). PilE 904 antibodies were made against purified N. gonorrhoeae pilin subunit from the strain N400.

For the double immunogold labeling the grids with N. gonorrhoeae were incubated with PilA-specific antisera (dilution 1:400 for 30 min), rinsed three times for 5 min each time on drops of PBS before incubation with gold-conjugated Protein A (10 nm gold for 30 min). The grids were washed three times for 5 min each time on drops of PBS and then blocked by incubation on free protein A for 15 min and rinsed three times for 5 min each time on drops of PBS before being incubated with the rabbit anti-PilE specific antibodies (dilution 1:400) for 30 min at room temperature. This was followed by three washes of 5 min each with PBS and incubation with gold-conjugated protein A (5-nm gold for 30 min) before rinsing three times for 5 min each time on drops of PBS, followed by two 5-min washes on drops of distilled water. Finally, the grids were stained for 10 min with 2% uranyl acetate. The procedure was repeated by changing the PilE- and PilA-specific antiserum incubation steps. As an experimental control to exclude the possibility of cross-reactivity, the incubation step with the second primary antibody was replaced with incubation on drops of PBS.

Interactions of phage PO4 with TFP.

Standard methods were used for phage PO4 preparation and phage titration (5, 6). Electron microscopy was used to study PO4 virions absorbed to N. gonorrhoeae cells by a method derived from (4). N. gonorrhoeae cells grown overnight on GC plates (37°C, 5% CO₂) were resuspended in GC broth to 2 × 10⁸ cells/ml and mixed with an equal volume of phage PO4 at 6 × 10⁹ PFU/ml. After gentle shaking at 37°C for 10 min, bacteria were subjected to immunogold labeling and negatively stained for electron microscopy.

Sample preparation for intact protein MS.

Isolated PilA protein was rinsed and washed by a methanol-chloroform precipitation procedure described previously (38). Briefly, 100 μl of the aqueous PilA solution at 2 to 3 mg/ml protein was diluted 1:3 (vol/vol) with methanol and mixed briefly. Both 100 μl of CHCl₃ and 200 μl of water were added consecutively, and this was followed each time by a mixing step. Phase separation was achieved by centrifugation (4,000 × g, for 8 min), yielding precipitated PilA at the interface. The upper methanol-water phase was removed, and 400 μl of methanol was added. After mixing, the pellet was recovered by centrifugation (13,000 × g, 8 min). The pellet was dried for 5 min in the inverted tube before the sample was dissolved in 50 μl of water, 70% formic acid, and acetonitrile (3:1:3 [vol/vol/vol]). Samples were subjected immediately to mass spectrometric (MS) analyses or frozen at −80°C.

Infusional MS analysis of intact protein.

All data were acquired on a quadrupole time-of-flight mass spectrometer (Q-Tof Micro; Micromass, Manchester, United Kingdom) equipped with the standard z-spray electrospray ionization (ESI) source. Sample solutions were infused into the ESI source at a flow rate of 5 μl/min by using a syringe pump (SP 100i; Cole-Parmer Instrument Company, Vernon Hills, IL). The source block temperature was maintained at 80°C. Nitrogen was used as both desolvation and nebulizing gas with flow rates of 300 and 20 liters/h, respectively. MS analyses were performed in the electrospray positive mode with the following parameter settings (parameter names as used in the MassLynx NT software, version 3.5): capillary voltage, 3,000 V; sample cone voltage, 25 V; extraction cone voltage, 4.3 V; ion energy, 3 V; and collision energy, 10 eV. The MS resolution was typically 4,000. The MS survey was obtained in a mass range from 150 to 1,700 m/z. Mass calibration in a mass range of 100 to 2200 m/z was performed using the ES tune mix solution from Agilent (Agilent, Palo Alto, CA). The MS spectra were analyzed by using the MassLynx software (version 3.5). For deconvolution, spectra were processed with the MaxEnt1 program of the MassLynx software.

RESULTS

P. aeruginosa TFP expression in N. gonorrhoeae.

We previously demonstrated the ability of PilA^PAK pilin to partially complement competence for natural transformation in a N. gonorrhoeae strain defective in expression of endogenous PilE (2). PilA^PAK pilin was expressed in this background as a pilE translational fusion such that, after processing by prepilin peptidase, the mature polypeptide was entirely P. aeruginosa derived. This approach also ensured levels of expression equivalent to that of endogenous pilin. When examined in whole-cell lysates, the levels of heterologous pilin were indistinguishable from those of endogenous PilE (Fig. 1). In contrast, the levels of PilA^PAK pilin in purifiable TFP were dramatically reduced relative to that of PilE pilin. This effect was not related to an inhibitory effect of PilA^PAK pilin on the integrity of the biogenesis machinery since the levels of purifiable PilE remained unaltered in a strain coexpressing PilA^PAK. Furthermore, the levels of PilA^PAK pilin in the shear-recoverable fraction were nearly equivalent to those of PilE when tested in a mutant background lacking the PilT retraction ATPase. Thus, the failure to recover high levels of PilA^PAK pili here is related to distorted TFP dynamics associated with organelle retraction rather than an intrinsic inability of PilA^PAK pilin to polymerize into TFP in an N. gonorrhoeae background.

FIG. 1.

Characterization of PilE and PilA expression in N. gonorrhoeae using whole cells lysates and purified pili. (A, D, and E) Silver-stained (A and E) or Coomassie blue-stained (D) SDS-PAGE gels loaded with either whole-cell lysates (A) or purified pili (E and D). (B, C, F, and G) Immunoblotting of whole-cell lysates (B and C) or purified pili (F and G) by using anti-PilE-specific antibodies (B and F) or anti-PilA-specific antibodies (C and G). Lanes: 1, N401 (wild type); 2, KS48 (iga::pilA); 3, KS49 (iga::pilA pilT); 4, MW24 (pilE); 5, KS50 (pilE iga::pilA); 6, KS51 (pilE iga::pilA pilT). Note that the polyclonal antibodies to PilA cross-react with PilE due to a shared epitope encompassed within the highly conserved amino termini of the proteins (28).

Immunofluorescence microscopy provided further insights into heterospecific PilA^PAK pilus expression (Fig. 2). In contrast to the bundles of short, lateral aggregates seen for endogenous TFP, pili in the strain solely expressing PilA^PAK pilin were detected primarily as long individual filaments, together with a lower abundance of short fibers. Moreover, the levels of PilA^PAK pilus antigen seen by this technique were similar to those seen for endogenous pilus antigen. Coexpression of endogenous PilE in this background resulted in a dramatic increase in the amount of PilA^PAK pilus antigen detected. Thus, PilE stimulates the levels of PilA^PAK antigen seen using this technique. In both of these instances, there was a clear discrepancy between the levels of PilA^PAK pili seen by purification yield versus those seen in the immunofluorescence assay. The expression of PilA^PAK pilin alone in a pilT background led to a strong increase in the levels of immunoreactive pili.

FIG. 2.

TFP characterization in gonococcal strains using immunofluorescence. Gonococcal cells (green) and TFP (red) are detected using indirect immunofluorescence. Anti-PilE specific antibodies (2-66) were used in the two leftmost columns, while anti-PilA specific antibodies were used for the two rightmost columns. Wild-type (wt; N400), KS45 (pilE), KS48 (iga::pilA), KS49 (iga::pilA pilT), KS52 (iga::pilA pilE), and KS53 (iga::pilA pilE pilT) strains are all shown as labeled.

Evidence for a unique tip structure on P. aeruginosa TFP in N. gonorrhoeae.

Immunoelectron microscopy confirmed the specificities of the PilE- and PilA-derived sera seen in the immunofluorescence assays (Fig. 3). In addition, differential labeling using sequential treatment with coupled 10- and 5-nm gold particles showed two serologically distinct pilus populations on each cell. There was no evidence for polymer heterogeneity vis-á-vis PilE and PilA^PAK. Nonetheless, a significant proportion of PilA^PAK pili (ca. 50% on average) stood out due to the absence of immunolabeling at their free ends (Fig. 4). In instances where it was possible to orient and track filaments back to cells, it was clear that such ends corresponded to the distal pilus end. These nonreactive segments were additionally noteworthy due to their relatively constant contour length. In addition, nonreactive tip structures were strongly associated with concurrent PilE expression since they were undetectable in its absence. A simple explanation for this phenomenon might then be that the tip structures were comprised of PilE. Efforts to test this were equivocal due to the low valency with which PilE antibodies reacted with homologous pili. We also investigated this by using antibodies previously published to bind specifically to the tip of PilE pili (14), but no binding was detected (results not shown). Regardless of the basis for these findings, this is to our knowledge the first direct evidence in any TFP system for a distinct, filament tip structure.

FIG. 3.

Piliation of wild-type and mutant gonococcal strains analyzed by immunogold labeling and transmission electron microscopy. (A and B) Negatively stained and 5-nm-immunogold-labeled TFP using anti-PilA specific antibodies. (C and D) Negatively stained and sequentially immunogold-labeled TFP fibers using anti-PilE specific antibodies (10-nm gold particles), followed by anti-PilA-specific antibodies (5-nm gold particles). (E and F) Negatively stained and sequentially immunogold labeled TFP using anti-PilA-specific antibodies (10-nm gold particles), followed by anti-PilE-specific antibodies (5-nm gold particles). The strain used was KS49 (iga::pilA pilT).

FIG. 4.

Evidence for a unique tip structure on P. aeruginosa TFP in N. gonorrhoeae. Negative staining and immunogold transmission electron microscopy using anti-PilA specific antibodies (5-nm gold particles). The lower left panel was only labeled with anti-PilA-specific antibodies and not with gold particles. Bar, 200 nm. Strains KS48 (iga::pilA), KS49 (iga::pilA pilT), KS52 (iga::pilA pilE), and KS53 (iga::pilA pilE pilT) are shown as marked.

Characterization of TFP retraction-associated events in N. gonorrhoeae expressing P. aeruginosa PilA^PAK.

As dynamic polymers, TFP undergo rounds of extension and retraction modeled as pilin subunit polymerization and depolymerization (43). To assess whether PilA^PAK pili were capable of undergoing retraction in N. gonorrhoeae, strains were examined microscopically for the expression of twitching motility by examining cell movement at the periphery of colonies, as well as by the slide culture method. Zones of “crawling” cells moving in jerky fashion relative to one another were readily detectable in the PilA^PAK pilin background, and this movement was abolished in a pilT background. These phenotypes in the PilA^PAK pilin background were indistinguishable from those seen for wild-type control strain (data not shown).

We next characterized the kinetics and force generation of PilA^PAK pilus retraction using laser tweezers (Fig. 5). The method has been described previously (25). In short, single cells of N. gonorrhoeae were immobilized at the surface of a microscope cover slide. Subsequently, a 2-μm latex bead was trapped in the laser tweezers and placed in close vicinity of the bacterium. When a pilus bound to the bead and retracted, the displacement of the bead was used as a measure of pilus displacement as a function of time as a function of the optical restoring force acting on the bead. It was found that pili retracted at a frequency of 0.16 ± 0.03 retractions/s, i.e., on average, a pilus would bind to the bead and retract once in 6 ± 1 s. The retraction velocity was constant at v = 1,600 × 100 nm/s at forces below 40 pN. At forces of >40 pN the velocity decreased and the velocity was 270 ± 30 nm/s at 100 pN, i.e., the velocity decreased by a factor of ∼6 from 0 to 100 pN. Force-dependent elongation of pili was not observed with this strain.

FIG. 5.

TFP-associated force generation for strains expressing PilA^PAK pili. Velocity-versus-force relationship for pilus retraction for N. gonorrhoeae expressing PilA pili KS52 (pilE iga::pilA).

Characterization of human epithelial cell adherence mediated by P. aeruginosa TFP in N. gonorrhoeae.

Current models invoke that human epithelial adherence mediated by PAK pili requires a pilin subunit receptor-binding domain localized to residues 128 to144 (21). To directly test this hypothesis, we first examined the interaction of the strain solely expressing PilA^PAK pilin and found that it did not adhere and was indistinguishable from a mutant devoid of pilin subunit expression (Fig. 6). Since this finding might be due to the low steady-state levels of pili expressed in this background, its pilT derivative was examined and found to adhere well with an average of more than 100 bacteria/epithelial cell being observed. We then examined the influence of PilC, the N. gonorrhoeae TFP-associated epithelial cell adhesin, and the PilV pilin-like protein on this phenotype. Although PilC is also required for high, steady-state TFP levels, this role in organelle expression is dispensable in a pilT background. As shown in Fig. 7, the levels of purifiable PilA^PAK TFP were undiminished in the pilC pilT and pilT pilV backgrounds. Nonetheless, epithelial adherence was significantly diminished in these backgrounds (i.e., to the level seen for a pilin-null mutant) (Fig. 6). Despite the presence of the PilA^PAK subunit, epithelial cell adherence proficiency was dependent on PilC and PilV. Moreover, like the situation in the homologous PilE-expressing background (39), PilC was critical to the ability of PilV to copurify with TFP, whereas the levels of copurifying PilC were only moderately influenced by PilV (Fig. 7). In addition, the levels of copurifying PilV in the PilA^PAK background never achieved those seen in the corresponding PilE background (Fig. 7, lanes 2 versus 8).

FIG. 6.

Adherence of wild-type and mutant gonococcal strains to human corneal epithelial cells. (A) Adherent cells are stained with crystal violet. All panels are shown at the same level of magnification. Adherence was also quantitated by determining the average numbers of adherent bacteria per cell (see Table S1 in the supplemental material). (B) Indirect immunofluorescence on gonococcal TFP using rabbit antibodies specific for PilA, followed by Alexa red (594 nm)-conjugated goat anti-rabbit immunoglobulin G antibodies (red). Gonococci were detected using fluorescence-labeled monoclonal antibodies (green). Wild-type (wt; N400), KS45 (pilE), KS48 (iga::pilA), KS49 (iga::pilA pilT), KS52 (iga::pilA pilE), KS53 (iga::pilA pilE pilT), KS58 (iga::pilA pilC pilE pilT), and KS55 (iga::pilA pilE pilT pilV) strains are shown as marked.

FIG. 7.

Quantitative analysis of PilC and PilV in gonococcal strains. (A, B, D, and E) Immunoblot of purified pili (A and B) or whole-cell lysates (D and E) using rabbit antibodies specific for PilC (A and D) or PilV (B and E). (C) Coomassie blue-stained SDS-PAGE gel showing relative amounts of PilE and PilA in purified pili. Lanes: 1, KS46 (iga::pilE pilE); 2, KS53 (iga::pilA pilE pilT); 3, KS58 (pilT pilC2_off iga::pilA pilE); 4, KS56 (pilT pilC2_off); 5, MW7 (pilT pilC1 pilC2); 6, KS55 (pilT pilV iga::pilA pilE); 7, GV6 (pilT pilV); 8 KS47 (iga::pilE pilE pilT).

P. aeruginosa PilA^PAK pilin in its assembled form is both necessary and sufficient for bacteriophage PO4 binding.

The capacity of heterologous TFP to function as primary bacteriophage receptors has primarily been assessed by productive infection, as seen by plaque formation or growth inhibition. Since N. gonorrhoeae strains expressing PAK pili showed no discernible phenotypes after exposure to high-titer PO4 phage stocks (data not shown), TFP-phage interactions were assessed directly by transmission electron microscopy using a strain expressing solely exogenous pilin. Using this approach, PAK pili were clearly decorated with PO4 phage in a manner indistinguishable from that reported for strain P. aeruginosa PAK (Fig. 8) (5). Specifically, phage particles were distributed tail-first along PAK pili filaments, and these interactions appeared to be mediated by a direct interaction between pili and phage tail fibers. To further confirm the specificity of these events, strains coexpressing endogenous and PAK pili were first exposed to phage and then immunolabeled with PAK-specific antibodies. As shown in Fig. 8E, PO4 phage were exclusively associated with PAK pili. We conclude that PilA^PAK pilin in its assembled form is both necessary and sufficient for PO4 recognition. In addition, the resistance of N. gonorrhoeae strains expressing PAK pili to productive PO4 infection must result from a block subsequent to phage binding its primary receptor.

FIG. 8.

PilA^PAK pilin is necessary and sufficient for bacteriophage PO4 binding to pili in N. gonorrhoeae. The bacteriophage PO4 binds tail first along PilA pili. Negative staining (A, B, and C) and immunogold transmission electron microscopy using anti-PilA specific antibodies (5-nm gold particles) (D and E) is depicted. Bar, 200 nm. The strain used was KS49 (iga::pilA pilT). PO4 does not bind to PilE pili (seen as unlabeled bundles in panel E).

Intact mass analysis of P. aeruginosa PAK pilin.

Endogenous PilE pilin undergoes unique posttranslational modifications, including glycosylation and the covalent addition of the phosphoforms phosphethanolamine and phosphocholine (17). To assess the status of PilA^PAK pilin in N. gonorrhoeae, ESI MS of the intact protein was performed. After deconvolution of the data signals, this revealed one well-defined peak at m/z 14,993 consistent with the mass values predicted from PilD proteolytically processed but otherwise unmodified protein (see Fig. S1 in the supplemental material). This finding was further corroborated by analysis of intact PilA^PAK pilin derived from P. aeruginosa, which yielded an identical value. Thus, PilA^PAK pilin does not undergo covalent posttranslational modifications in either background. Therefore, the phenotypes seen for PAK-expressing N. gonorrhoeae strains are not influenced by additional or alternative modifications in subunit structure and chemistry.

DISCUSSION

Prior studies of TFP functions utilizing expression of heterologous pilin subunits have focused on a limited set of phenotypes comprised of twitching motility, phage sensitivity, and competence for natural transformation. Our findings with N. gonorrhoeae expressing P. aeruginosa PilA^PAK pilin provide new insight into the contributions of subunit constitution to TFP function. Most notably, PilA^PAK complemented a pilE mutant in adherence for human primary corneal epithelial cells, and this property required the PilC adhesin and the PilV pilin-like protein. Thus, the pilin subunit itself is not the main determinant of adherence. Rather, in its polymeric form, it appears to act in a generic fashion so as to support the display of the PilC adhesin. In line with this, PilC and PilV were recoverable in the sheared TFP fraction from the PilA^PAK background. Moreover, as with endogenous pilin (39), TFP-associated PilC levels were moderately reduced in a pilV background and PilV levels were reduced in the pilC background. Together with the PilC copurification data, immunolabeled electron miscroscopy localization studies suggest that PilC is directly associated with TFP (31). For such a model to be true here, PilC would have to associate directly or indirectly with the heterologous TFP structure. However, we cannot rule out that PilA-derived TFP or the process of TFP assembly and extrusion in itself supports PilC function independent of its direct association with TFP. The data also reveal signs that adherence proficiency imparted by heterologous TFP is not equivalent to that seen in the wild-type background. Notably, adherence imparted by PilA^PAK was only detected in a pilT mutant. This failure to detect epithelial cell binding in the wild-type background could be ascribed to low, steady-state TFP levels. Even when this defect was suppressed in the pilT background, the levels of adherence never reached those seen in the wild-type strain. This may in part be attributed to the nonautoagglutinating phenotype exhibited by the complemented strain as bacterial-bacterial aggregation contributes to overall adherence proficiency (27).

The requirement for PilC and PilV for PilA^PAK pilin-mediated adherence was surprising given the prevailing dogma that P. aeruginosa TFP-mediated adherence to human epithelial cell requires a receptor-binding domain located within the C-terminal region of pilin (21). Clearly, adherence mediated by such an intrinsic domain is not manifest in the gonococcal background. The in vivo relevance of the PilA binding domain and its putative asialo-GM1 glycosphingolipid receptor (23) has recently been brought into question (13). Assuming that there is in fact an intrinsic PilA receptor-binding domain, we surmise that it is not properly exposed in the gonococcal background.

Our previous work documented the ability of PilA^PAK pilin to complement DNA binding and uptake required for natural transformability (2). However, the frequency of transformation supported by PilA was only 2.5% of that seen in the wild-type background. Many studies have shown that strains expressing very low, steady levels of endogenous TFP (resulting from reduced levels of pilin expression or pilE alleles partially defective in assembly) retain high-level transformability, and such strains express far lower levels of TFP than those seen here for the PilA^PAK pilin (1, 24, 30). The reduced activity of PilA^PAK vis-á-vis transformation seems therefore unrelated to a quantitative defect in TFP expression. Rather, the data imply a degree of functional incompatibility between PilA^PAK itself or PilA^PAK TFP and other components involved in DNA binding and uptake. It is important to note here that overexpression of the pilin-like ComP protein enhanced DNA uptake and transformability over a 100-fold in this background, while no increase in the levels of TFP were seen (2). Thus, PilA^PAK appears defective in supporting ComP function or a ComP-dependent pathway in a manner unrelated to its assembly proficiency.

In addition to the suboptimal human cell adherence and transformation phenotypes associated with PilA^PAK, other findings suggest diminished compatibility between the exogenous pilin and endogenous machinery. Most notably, the levels of purifiable PilA^PAK TFP were remarkably low in a wild-type background but in the absence of the PilT retraction ATPase were nearly equivalent to those of endogenous TFP. This PilT-mediated effect was specific for exogenous TFP since the levels of endogenous TFP expressed simultaneously were not similarly diminished. Thus, this phenomenon involved a specific rather than a general perturbation of TFP dynamics in the cell. Steady-state TFP levels are partly dictated by the relative frequencies of extension and/or polymerization event initiation versus the frequencies with which extension or polymerization events are terminated by retraction or disassembly (mediated by PilT) (40). It appears then that PilA^PAK pilus assembly sites are disproportionately susceptible to attack by the PilT-associated retraction machinery. This phenotype is reminiscent of that seen in N. gonorrhoeae null mutants lacking so-called effectors of pilus homeostasis (40, 42, 43). These factors act by promoting extension or polymerization events in the presence of PilT and include PilC and all of the five pilin-like proteins encoded within the pilH-L locus. Therefore, we propose that PilA^PAK may interact suboptimally with these resident effector proteins. Like PilC, the PilH, -I, -J, -K, and -L proteins are each required for both human cell adherence and transformation competence (40). Therefore, the disparate phenotypes seen in the PilA^PAK background might also reflect a diminished functionality of these components.

Simultaneous expression of endogenous pilin and P. aeruginosa PilA^PAK at identical levels revealed previously unseen interactions. First, PilA pili were demarcated by a tip structure whose presence was PilE dependent. The simplest explanation could be that this tip structure is comprised of components requiring PilE for localization to this site, but further studies are required to assess these possibilities. Whatever its nature, the presence of the tip element is not essential to TFP-associated functions. Second, PilE influenced the length distributions of PilA^PAK pilus filaments seen by immunofluorescence, with a heterogeneous mixture of pilus lengths being seen in its presence and only long, single filaments being seen in its absence. This differential effect was only seen in the presence of PilT, suggesting that the action of PilE was in some way influenced by pilus retraction. In contrast, PilE had no effect on the levels of purifiable PilA^PAK pili, which were profoundly reduced relative to those of endogenous TFP. These discordant results seen for PilA^PAK TFP levels in wild-type backgrounds using the purification versus the direct immunodetection methods are difficult to reconcile. Obviously, the cultivation conditions are clearly different since TFP are purified from strains propagated on agar medium, whereas immunodetection required shorter times of propagation in liquid medium on coverslips. The use of poly-l-lysine-coated coverslips here might be a factor since TFP undergoing extension and retraction might be irreversibly trapped or captured externally so as to prevent retraction or otherwise distort organelle dynamics. By way of example, treatment of P. aeruginosa PAO1 with pilus-specific RNA phage has been demonstrated to stimulate pilus formation by an unknown mechanism (7).

With regard to twitching motility, our results were not unexpected since other studies have documented the ability of diverse pilin subunits to form TFP and support this property in P. aeruginosa. We showed here that replacement of PilE by PilA^PAK does not significantly impact force generation by TFP retraction. Like pilus retraction with endogenous TFP (25), the velocity is constant at forces of <40 pN, and at higher forces the velocity decreases but pili were still able to retract against 100pN. However, the average retraction velocity at low forces was increased compared to the wild type. This observation suggests that the maximum force is likely determined by a molecular motor in or near the membrane and that the structure and/or composition of the pilin subunit may influence the speed of pilus retraction.

Despite the vast number of instances in which bacteriophages are known to initiate infection by adsorption to pili, the pilus component involved has yet to be unambiguously identified in any system. In the present study, we demonstrated that the expression of P. aeruginosa PilA as TFP in N. gonorrhoeae was both necessary and sufficient to engender binding of bacteriophage PO4. It follows then that a specific PO4 receptor is encompassed within PilA or within a PilA oligomer. Nonetheless, other species expressing pilin subunits structurally distinct from PilA^PAK exhibit PO4 sensitivity (in a TFP-dependent fashion), and it has been reported that the expression of any one of diverse set of TFP pilin subunits in a P. aeruginosa PAK pilA background restored PO4 sensitivity (29, 37). It is formally possible then that in some strains either a functional redundancy of PO4 receptors exists or that PO4 exhibits relaxed specificity for TFP organelles in some (but not all) instances. It is perhaps worth noting here that some filamentous phages are able to infect F⁻ E. coli at frequencies of ca. 10⁻⁶, and infectivity can be dramatically increased by treatment that perturbs the outer membrane (32). There is thus clear precedence for the ability to bypass the requirement for the primary pilus receptor, albeit with reduced frequency. Further studies examining the direct interactions between phage and TFP as detailed here and elsewhere (19) should help clarify these issues.

In summary, this systematic analysis of phenotypes imparted by heterologous pilin subunit expression in N. gonorrhoeae provides new insights into the correlations between TFP expression, structure, and associated functions.