SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Retraction of type IV pili
  5. How do they do it?
  6. A sticky problem; how do T4P mediate secure attachment?
  7. Getting to the bottom of it; the pilus motor
  8. A look at the hole problem: crossing the cell envelope
  9. Acknowledgements
  10. References

Twitching motility is a unique form of bacterial propulsion on solid surfaces associated with cycles of extension, tethering and retraction of type IV pili (T4P). Although investigations over the last two decades in a number of species have identified the majority of the genes involved in this process, we are still learning how these pili are assembled and the mechanics by which bacteria use T4P to drag themselves from one place to another. Among the puzzles that remain to be solved is the mechanism by which hydrolysis of ATP is coupled to pilus assembly and disassembly, and how the cell envelope structure is modified to accommodate the passage of the pilus through the periplasm. Unravelling of these and other enigmas in the T4P system will not only teach us more about these important colonization and virulence factors, but also help us to understand related processes such as type II secretion, which relies on a set of proteins homologous to those in the T4P system, and bacterial conjugation, involving retractable pili belonging to the F-like subgroup of the type IV secretion family. This review focuses on recent discoveries relating to the assembly and function of T4P in generation of twitching motility.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Retraction of type IV pili
  5. How do they do it?
  6. A sticky problem; how do T4P mediate secure attachment?
  7. Getting to the bottom of it; the pilus motor
  8. A look at the hole problem: crossing the cell envelope
  9. Acknowledgements
  10. References

Type IV pili (T4P) are adhesive appendages synthesized by a wide variety of Gram-negative bacteria ranging from pathogens, where they contribute to virulence, to environmental species such as Myxococcus, Synechocystis, Aquifex and Shewanella, reviewed recently in Mattick (2002) and Craig et al. (2004). T4P mediate adherence to host cells (Saiman et al., 1990; Whitchurch and Mattick, 1994), twitching motility across solid surfaces (Saiman et al., 1990; Semmler et al., 1999; Mattick, 2002), bacteriophage adsorption (Bradley, 1974), DNA uptake (Mattick, 2002) and biofilm initiation and development (O’Toole and Kolter, 1998; Chiang and Burrows, 2003). T4P have been most extensively characterized in Neisseria and Pseudomonas, but studies on related systems in bacteria like enteropathogenic Escherichia coli (bundle-forming pili, Bfp) and Vibrio cholerae (toxin-co-regulated pili, Tcp) are adding to our knowledge base. This review will highlight the current state of knowledge regarding assembly of T4P through the complex Gram-negative envelope and the mechanism by which the forces necessary to transport bacteria towards or across surfaces may be generated.

Retraction of type IV pili

  1. Top of page
  2. Summary
  3. Introduction
  4. Retraction of type IV pili
  5. How do they do it?
  6. A sticky problem; how do T4P mediate secure attachment?
  7. Getting to the bottom of it; the pilus motor
  8. A look at the hole problem: crossing the cell envelope
  9. Acknowledgements
  10. References

A fascinating feature that distinguishes T4P from many other types of bacterial fimbriae is their ability to be retracted through the cell wall while the pilus tip remains firmly adhered to surfaces of various types, allowing the pili to act as fishing rods or grappling hooks for translocation of the cell body. While conjugative pili belonging to the F-like subfamily of type IV secretion (T4S) systems have also been observed to undergo retraction in the process of mating pair formation (Novotny and Fives-Taylor, 1974; Harris and Silverman, 2004), that process has not been investigated to the same extent as T4P retraction. Early electron microscopy work by Bradley demonstrated that the pili of Pseudomonas aeruginosa could be retracted into the cell, a process that was impaired by binding of the pili by the bulky capsids of pilus-specific bacteriophages or anti-pilin antibodies (Bradley, 1972a,b). Studies in Myxococcus xanthus showed that wild-type cells tethered by their T4P to a solid surface exhibited a jiggling motion, bringing the cell body into juxtaposition with the surface over time (Sun et al., 2000). In contrast, mutants that were piliated but lacked social motility (the M. xanthus version of twitching) became tethered but were stationary and did not approach the surface. Skerker and Berg (2001) showed retraction of P. aeruginosa T4P in real time by filming fluorescently labelled whole cells, capturing images of the pili being rapidly withdrawn into stationary cells or of tethered cells jerking toward the point of pilus attachment. Elegant microscopy studies of Neisseria cells undergoing T4P retraction used laser tweezers to counteract the retraction force (Merz et al., 2000; Maier et al., 2002). These studies not only confirmed that T4P can shorten, bringing the cell body closer to the distal attachment point of the pilus, but also measured the forces generated during this process. Pilus retraction was estimated to generate forces of up to 140 pN, distinguishing the pilus motor as one of the strongest molecular machines reported to date (Maier et al., 2002). Remarkably, retracting forces can be generated by single pili, as Skerker and Berg's study showed that the extended pili on any particular cell retracted independently of one another. This phenomenon, coupled with the flexibility of the pili, affords the bacteria the opportunity to twitch along a number of vectors tangential to the long axis of the cell, depending on where the pilus tip becomes tethered. Tethering of multiple pili will increase the likelihood that the cell body will be brought into close contact with a surface. Twitching motility enables bacteria to move across surfaces at approximately 1 µm s−1, which is approximately the same as the rate of pilus retraction measured during the laser tweezer experiments. Skerker and Berg (2001) estimated P. aeruginosa pilus retraction rates of approximately 0.5 µm s−1, in the same range as those observed with Neisseria.

Twitching motility plays a number of roles in biofilm formation, from promoting initial adherence to various substrata to promoting detachment of cells to maintain biofilm morphology (O’Toole and Kolter, 1998; Chiang and Burrows, 2003; Klausen et al., 2003a). An interesting study by Klausen et al. (2003b) showed that twitching motility affects spatial differentiation within a biofilm. They showed that in mixed biofilms, wild-type cells use twitching motility to clamber on top of their isogenic non-piliated sessile counterparts to form the ‘cap’ component of mushroom-like macrocolonies. Twitching motility also affects virulence of pathogens expressing T4P. Comolli et al. (1999) showed that piliated but non-twitching mutants of P. aeruginosa were capable of colonizing the lungs of mice at levels comparable to the wild type, but were impaired in their ability to disseminate to other organs. P. aeruginosa mutants lacking various T4P genes were also found to be non-cytotoxic, likely due to an inability to engage their contact-dependent type III secretion (T3S) systems without T4P retraction to bring the bacteria into close apposition to host cell membranes (Kang et al., 1997). Studies in intrepid human volunteers showed that enteropathogeneic E. coli (EPEC) expressing T4P but lacking twitching motility lost the ability to disperse from autoaggregates and were 200-fold less virulent, causing significantly less diarrhoea than the wild type (Bieber et al., 1998). Neisseria T4P mediate binding to a number of cell types, eliciting Ca++ fluxes and cortical plaque formation due to signals generated from the application of tensile stress on eukaryotic membranes during pilus retraction (Merz et al., 1999; Merz and So, 2000).

How do they do it?

  1. Top of page
  2. Summary
  3. Introduction
  4. Retraction of type IV pili
  5. How do they do it?
  6. A sticky problem; how do T4P mediate secure attachment?
  7. Getting to the bottom of it; the pilus motor
  8. A look at the hole problem: crossing the cell envelope
  9. Acknowledgements
  10. References

The prevalent model of pilus extension and retraction suggests that extremely rapid (1500 subunits s−1), ATP-dependent association/dissociation of pilin subunits at the base results in pilus assembly/disassembly. For twitching motility to occur, the tip of the pilus must be firmly adhered to a surface in order to provide an anchor against which the pilus motor can exert force. Although significant attention has been focused recently on the motor proteins required for the assembly/disassembly events at the cell proximal end of the pilus fibre (below), there has been relatively little study of the non-specific adhesive events occurring at the distal pilus tip that are essential for twitching to occur. At the proximal end, the base of the pilus must be affixed within the cell envelope in such a way that the application of tensile forces will not readily result in detachment of the pilus from the cell, but that will simultaneously allow longitudinal movement of the pilus either into or out of the cell wall during retraction and extension events respectively. Unlike flagella, which have a well-characterized basal ring and hook structure, no obvious base structure for T4P has been identified. The following sections will consider each of these issues.

A sticky problem; how do T4P mediate secure attachment?

  1. Top of page
  2. Summary
  3. Introduction
  4. Retraction of type IV pili
  5. How do they do it?
  6. A sticky problem; how do T4P mediate secure attachment?
  7. Getting to the bottom of it; the pilus motor
  8. A look at the hole problem: crossing the cell envelope
  9. Acknowledgements
  10. References

Due to the important role of T4P as virulence-related adhesins, there has been a historic focus on identification of T4P proteins or protein subdomains involved in interaction of T4P with host cell receptors. In contrast, there is a dearth of studies aimed at elucidating the structural basis for the ability of T4P to mediate non-receptor-mediated binding, a crucial event that enables twitching motility to occur. T4P can promote binding to, and twitching motility upon, a wide variety of materials including tissue, glass, plastics and metals, key steps in the formation of biofilms. The attachment of the pilus to a surface must be sufficiently robust to withstand the force of pilus retraction in order that the cell body may move before detachment of the pilus tip. If the cell is suspended in a liquid medium, the shear forces on the cell body are probably low. In contrast, when the cell body is in contact with a solid or semi-solid material, holding forces including liquid surface tension and those generated by other cell surface-associated adhesins must be overcome in order for the cell body to translocate. The dynamic interplay between the pilus-mediated retraction force and sudden loss of surface contacts during twitching motility on surfaces is likely to be responsible in part for its often-reported ‘jerky’ nature (Semmler et al., 1999).

The major structural subunit, and one of the best-studied components of the T4P system, is a small (c. 15–23 kDa) pilin protein called PilA in Pseudomonas and PilE in Neisseria. Crystallographic investigations of type IV pilins from Neisseria, Pseudomonas and V. cholerae have shown that the subunits consist of a highly conserved extended N-terminal hydrophobic alpha-helix region followed by a globular C-terminal domain containing beta-strands (Parge et al., 1995; Hazes et al., 2000; Craig et al., 2003). The protein terminates in a disulphide-bonded loop (DSL) the size of which varies among pilin alleles (Kus et al., 2004). The subunits are predicted to assemble such that the conserved hydrophobic N-terminal alpha-helices form the core of the pilus fibre, while the C-terminal domains decorate the surface (Fig. 1). The extended N-terminal helix region likely serves to retain subunits in the inner membrane for processing by the peptidase/N-methylase PilD and, in some strains, post-translational modification with glycans (Castric, 1995; Stimson et al., 1995), alpha-glycerophosphate (Stimson et al., 1996), phosphocholine, phosphoethanolamine (Hegge et al., 2004) and/or phosphate (Forest et al., 1999). The effect of these decorations upon twitching motility has not been widely investigated. Upon pilus assembly, the N-terminal helices are positioned in the interior of the fibre, stabilizing the structure so it can withstand the substantial forces produced by twitching motility. Upon fibre retraction, the pilins are thought to dissociate at the pilus base, ‘melting’ back into the cytoplasmic membrane where they could form a recyclable subunit pool to be used in a new round of pilus extension. The involvement of minor pilins (i.e. PilE, PilV, PilW, PilX, FimT and FimU of P. aeruginosa) in pilus assembly has been demonstrated by mutational analyses (Russell and Darzins, 1994; Alm and Mattick, 1995; 1996; Alm et al., 1996; Winther-Larsen et al., 2005), but defining the specific functions of these proteins requires further analyses. It is extremely difficult to detect these proteins in intact pili because they are present in vanishingly small quantities, but without them, surface piliation and its associated functions are lost. They may play roles in priming of pilus extension or prevention of pilus retraction; in control of pilus length; or in pilus-specific functions including adherence, transformation competence or motility.

image

Figure 1. Model of the type IV pilus of Pseudomonas aeruginosa PAK. The disulphide-bonded loop (DSL) region implicated in adherence is shown in red and the remainder of the protein in green. The N-terminal alpha-helices of each subunit pack into the core of the fibre, and the C-terminal DSL is partly exposed on the fibre surface. Illustration kindly provided by L. Craig.

Download figure to PowerPoint

While the structure of the pilin subunit has been defined through crystallographic studies, the final form of the assembled fibre has been more difficult to nail down. Although it is widely accepted that the pilins assemble in a helical fashion, the number of subunits per turn of the helix, the number of independent starting points and the handedness of the helix are features that have not been experimentally determined (Craig et al., 2004). Early models suggested that the fibre was a left-handed helix containing approximately five subunits per turn (Parge et al., 1995; Hazes et al., 2000), while more recent calculations have pointed to four subunits per turn, with three independent starts (Craig et al., 2003). In contrast, the pseudopilus formed by overexpression of the major pseudopilin PulG is predicted to have a similar number of subunits per turn (17 monomers over four turns) but to be a one-start helix (Kohler et al., 2004). Confirmation of these models requires high-resolution X-ray diffraction studies, since accurate estimates of the pitch of the helix and retraction rates would be useful to calculate rates of pilin assembly and disassembly. Current models show that the conserved N-terminal helices pack into the core of the fibre with the globular C-terminal domains decorating its surface. The critical DSL appears to be at least partly buried along the fibre but exposed at its tip (Fig. 1). If there are three independent starts in the pilus fibre as the model suggests, up to three DSL subdomains could be displayed at the pilus tip, increasing the valency and strength of attachment.

Both specific and non-specific attachment events mediated by T4P occur via the pilus tip. Many bacterial pili have separate major structural and minor adhesive subunits, with the latter typically positioned at the distal end of the pilus to promote adherence. In some cases, the adhesive subunit has been shown to be involved in both receptor-specific and non-specific adherence. Non-retractile E. coli type I fimbriae, which bind specifically to mannose-containing receptors on human cells, are also important for non-specific attachment events that initiate biofilm formation. E. coli type I fimbriae-dependent biofilm formation can be inhibited in a dose-dependent fashion upon addition of alpha-mannoside, suggesting that the FimH adhesin is responsible for both receptor-specific and non-specific attachment (Pratt and Kolter, 1998; Schembri and Klemm, 2001). In the case of T4P, the presence of a tip-associated adhesin is still somewhat controversial. In P. aeruginosa, the C-terminal DSL of the pilin subunit was shown by monoclonal antibody binding studies to be exposed only at the tip of the pilus and to mediate attachment to host cell receptors, suggesting PilA itself acts as both a structural subunit and an adhesin (Lee et al., 1994). The DSL structure is highly conserved among type IV pilins of all species, although the size of the loop (from 12 to 31 amino acids in P. aeruginosa) and its sequence are highly variable (Kus et al., 2004). It is not yet clear whether the DSL region of type IV pilins is also responsible for the non-specific tethering events that are necessary for twitching motility to occur. In the process of creating a pilA knockout of P. aeruginosa, Farinha et al. (1994) inserted a chloramphenicol resistance cassette into the pilA structural gene. Serendipitously, they generated a mutant expressing pilin in which the last nine amino acids of the protein, including the second Cys residue, were replaced by 11 amino acids encoded on the antisense strand of the marker gene. This mutant pilin was shown to assemble into surface pili that were less able to promote attachment to A549 pneumocytes compared with wild-type pili. Unfortunately, the ability of the mutant pili to mediate twitching motility was not tested. However, a transposon insertion into the 3′ end of the pilA gene of P. aeruginosa PA103, corresponding to truncation of the protein at amino acid 125 and therefore loss of the entire DSL, was reported to result in delayed twitching motility, reduced adherence to MDCK cells and an 80% reduction in cytotoxicity compared with the wild type (Kang et al., 1997). In our hands, site-directed mutation of either of the two Cys residues of the DSL, or deletion of the amino acids between them, results in a dramatic reduction in cellular pilin levels with a concomitant loss of surface pili, precluding analysis of twitching motility (M. Habash, K. Howard and L.L. Burrows, unpublished data). A similar decrease in pilin stability was noted when either of two Cys residues within bundlin, the type IV pilin subunit of EPEC, was mutated to Ser (Zhang and Donnenberg, 1996).

Studies of T4P expression in strains lacking the periplasmic disulphide-bond isomerase DsbA, required for formation of the DSL via generation of the bond between the Cys residues, have demonstrated a range of effects. A dsbA mutant of P. aeruginosa lacked twitching motility and showed reduced expression of the pilA gene; the effect of the dsbA mutation on pilin protein levels was not determined (Ha et al., 2003). EPEC strains lacking DsbA showed normal levels of bundlin gene transcription but a marked reduction in protein stability (Zhang and Donnenberg, 1996). In V. cholerae, inactivation of the DsbA homologue, TcpG, resulted in expression of T4P that appeared morphologically normal but were non-functional in a number of in vitro and in vivo assays (Peek and Taylor, 1992). Similarly, inactivation of two of the three DsbA homologues in Neisseria meningitidis resulted in wild-type levels of pilins both in whole-cell lysates and on the cell surface, but drastic reductions in T4P-dependent DNA transformation competence and alterations in the pilin isoelectric point (Tinsley et al., 2004). Therefore, the formation of the DSL appears to be important for pilin function, and in some bacteria loss of this bond results in unstable pilins that are rapidly degraded and may trigger negative feedback on pilin gene expression. Because of the potential for pleotropic effects in dsbA mutants, targeted studies of PilA structure and its relationship to function are preferable.

While PilA appears with play a dual role as both a structural protein and an adhesin in P. aeruginosa, studies of T4P-mediated adherence in Neisseria have implicated additional proteins such as PilC (corresponding to PilY1 in Pseudomonas) in pilus-mediated adherence. This functional assignment was based in part on the observations that PilC is both outer membrane- and pilus-associated, and that isolated PilC binds to human cells, blocking attachment of piliated gonococci (Nassif et al., 1997; Scheuerpflug et al., 1999; Merz and So, 2000). More recent data have emerged to show that PilC is also involved in antagonism of pilus retraction (below), and for that reason, influences twitching motility in Neisseria (Morand et al., 2004). Its specific influence on non-receptor-mediated adherence of T4P has not been reported, but since Neisseria is an obligate pathogen, the role of PilC in adherence is likely specific to host interactions.

Studies comparing T4P with type II secretion (T2S) show striking similarities between the systems (for a recent review, see Peabody et al., 2003). Both make use of pilin-like proteins (called pseudopilins in the T2S system) that assemble to form a fibre-like structure after processing by a common signal peptidase/N-methylase (PilD); both use large multimeric outer membrane channels (GspD/XcpQ/PilQ) to permit passage to the cell's exterior; and both are associated with ATPases that are required to promote assembly of the structural subunits. One of the curious differences between pilins and pseudopilins is (with a few exceptions; Pugsley et al., 2001) the lack of a C-terminal DSL in the latter. Whether this structural difference reflects differences in the function of these proteins is not yet clear. There have been scattered reports of cross-over between the T4P and T2S systems. For example, recent studies showed that the T4P of V. cholerae are involved in secretion of a soluble colonization factor, independent of the native T2S system (Kirn et al., 2003), and that T2S is reduced in a pilA mutant of P. aeruginosa (Lu et al., 1997). At least two T2S systems can be artificially manipulated to form adhesive pseudopili that enhance bacterial adherence to surfaces and biofilm formation (Durand et al., 2003; Vignon et al., 2003), although this phenomenon is not thought to be physiologically relevant (Kohler et al., 2004).

Getting to the bottom of it; the pilus motor

  1. Top of page
  2. Summary
  3. Introduction
  4. Retraction of type IV pili
  5. How do they do it?
  6. A sticky problem; how do T4P mediate secure attachment?
  7. Getting to the bottom of it; the pilus motor
  8. A look at the hole problem: crossing the cell envelope
  9. Acknowledgements
  10. References

What powers retraction of the pilus to generate twitching motility? Genetic studies of P. aeruginosa and Neisseria established that predicted T4P ATPases PilB (PilF in Neisseria), PilT and PilU were necessary for twitching motility, and that the non-twitching mutants either were bald, lacking pili altogether (PilB), or were hyperpiliated (PilT and PilU) (Nunn et al., 1990; Whitchurch et al., 1991; Lauer et al., 1993). From this work, it was concluded that PilB was involved in polymerization of the subunits to form the pilus and that PilT and PilU were involved in depolymerization. These three proteins belong to the AAA+ (ATPases associated with various activities) superfamily of oligomeric molecular ratchets which typically hydrolyse ATP to power mechanical folding/unfolding or polymerization/depolymerization events (Patel and Latterich, 1998; Ogura and Wilkinson, 2001). PilT and PilB are widely conserved among T4P-expressing bacteria, but PilU (sometimes called PilT2) is less broadly distributed (Chiang et al., 2005). It is possible that PilU arose from a gene duplication event, as it has strong homology to PilT (67% similarity in P. aeruginosa) and the pilT and pilU genes are contiguous in the chromosomes of both P. aeruginosa and Neisseria.

PilB and PilT are thought to play antagonistic roles, with PilB promoting pilin association and fibre formation and PilT promoting pilus retraction via pilin disassembly. The role of the third member of the trio, PilU, is less clear. Inactivation of either PilT or PilU in P. aeruginosa results in a similar hyperpiliated, non-twitching phenotype (Whitchurch and Mattick, 1994), while in Neisseria inactivation of pilT, but not pilU, causes loss of twitching motility (Park et al., 2002). Despite apparently normal levels of piliation, Neisseria pilU mutants have abnormal phenotypes including a spreading colony morphology, loss of pilus-associated autoagglutination and increased binding to host cells. Despite the similarity in phenotype of P. aeruginosa mutants lacking PilT or PilU, the proteins appear to have quite distinct functions. Neither protein can compensate for loss of the other, even when overexpressed in trans. Mutants lacking PilT are resistant to pilus-specific bacteriophages, which exploit pilus retraction to bring them in contact with the bacterial cell surface for injection of their genetic material (Bradley, 1974). In contrast, PilU mutants surprisingly remain susceptible to such phage despite an apparent inability to retract their T4P (Whitchurch and Mattick, 1994). In the presence of a shear force, P. aeruginosa pilT mutants form exaggerated biofilms, but pilU mutants are unable to form biofilms, a phenotype that suggests decreased pilus strength or stability in pilU strains (Chiang and Burrows, 2003). Interestingly, inactivation of the PilQ outer membrane secretin in pilT mutants of Neisseria results in the lethal phenotype of unopposed pilus extension within the periplasm (Fig. 2). In contrast, pilQ pilU mutants do not express ingrown pili (Wolfgang et al., 2000), suggesting PilU is not directly involved in pilus retraction.

image

Figure 2. Phenotype of a pilT pilQ mutant of Neisseria gonorrhoeae. If pilus extension is forced by inactivation of PilT in a strain lacking the outer membrane secretin PilQ, the pili are extended in the periplasm, causing outer membrane bulges and extrusions. Photo kindly provided by M. Koomey and reprinted with permission from EMBO J19:6408–6418, copyright 2000, Macmillan Publishers.

Download figure to PowerPoint

Using fluorescent chimeras, we showed recently that these motor-associated proteins have disparate localization patterns in P. aeruginosa, with PilT and PilB localizing to both poles and to the division septum of dividing cells, while PilU is found only at the piliated pole (Chiang et al., 2005). While localization of PilU to the piliated pole was not surprising, given its role in twitching motility, finding PilB and PilT at the non-piliated pole was intriguing – what are they doing there? One possible explanation comes from observations of twitching bacteria. Although P. aeruginosa is typically piliated at the flagellated (old) pole, twitching cells have been observed to rapidly reverse direction when moving across a surface (Semmler et al., 1999). Similar observations have been made in M. xanthus (Sun et al., 2000). It is possible that PilB and PilT are part of a twitching motor subassembly, which, upon complexing with additional proteins such as PilU, is made functional.

Through generation of chimeric proteins, the differential localization of the highly homologous PilT and PilU proteins was found to be directed by the N-terminal region, which may specify interactions with as yet unidentified partners. PilB requires the inner membrane protein PilC (corresponding to PilG in Neisseria) for polar localization in P. aeruginosa, but no protein partner for either PilT or PilU has yet been identified (Chiang et al., 2005). Protein–protein interaction studies of components of the type IV bundle-forming pili system in E. coli revealed similar interactions between BfpD (PilB) and a cytoplasmic loop of the membrane protein BfpE (PilC in P. aeruginosa) but also demonstrated interaction of BfpF (PilT) with a separate region of BfpE (Crowther et al., 2004). Additional interactions of BfpD with the N-terminus of membrane-spanning protein BfpC (PilN in P. aeruginosa) were identified, suggesting a functional interaction between the cytoplasmic ATPases and specific membrane proteins (whose localization within the cell has not yet been determined) is required for pilus function. On the topic of protein interactions, a major deficiency in the field is knowledge of how the PilB and PilT motor proteins interact with pilins to promote their association or dissociation. At present it is unclear if there is direct physical contact between the ATPases and the pilins, or whether intermediary proteins such as the membrane proteins listed above and/or the minor pilins are also involved.

Recent studies of pilus retraction in Neisseria revealed that the adhesin protein PilC (corresponding to PilY1 in P. aeruginosa) plays an important role in control of pilus retraction. While pilC mutants of Neisseria gonorrhoeae appeared unable to assemble pili, this deficit could be overcome by introducing a pilT mutation into the pilC mutant background (Wolfgang et al., 1998). Further investigation of this phenomenon showed that PilC, which is localized to the outer membrane, acts as a ‘clip’ on the pilus to counteract its retraction by PilT. In the absence of PilC, the pili are rapidly retracted into the cell and the cells appear bald (Wolfgang et al., 1998; Morand et al., 2004). This is an important observation that emphasizes the need to distinguish between mutations that simply abrogate pilus assembly altogether, versus those that skew the balance of extension and retraction to give the appearance of non-piliation. The role of the corresponding protein PilY1 protein of P. aeruginosa has not been investigated, but both PilY1 and the Neisseria PilC protein are encoded in the same operon as the minor pilins.

Recent work in the Koomey laboratory (Winther-Larsen et al., 2005) showed that inactivation of the minor pilins in N. gonorrhoeae results in a phenotype similar to that of a pilC mutant; near-complete loss of piliation in the wild-type background, which could be suppressed in a pilT background. Their results suggest that the role of the minor pilins, like PilC, is to tip the scale in favour of pilus extension, and that in the absence of these proteins, retraction is more likely to occur. This work also demonstrated that the minor pilins do not play strictly structural roles, as levels of piliation in minor pilin mutant-pilT suppressor backgrounds are similar to those seen when all of the minor pilins are present (Winther-Larsen et al., 2005).

As noted above, studies comparing T4P with T2S have noted striking similarities between the systems. One interesting dichotomy is the lack of PilT and PilU homologues in the T2S system. Homologues of PilB are readily identified, and current hypotheses about the mechanism of secretion include the possibility of a piston- or sewing machine needle-like action of the pseudopilus to push folded substrates through the outer membrane secretin pore (Hobbs and Mattick, 1993; Pugsley, 1993; Nunn, 1999). However, such a mechanism would require that the pseudopilus, like the T4P, was alternately extended and retracted. It is possible that the PilB-like T2S ATPases provide both extension and retraction functions, and that the T4P protein PilB has lost the ability to promote retraction over time due to its partnership with PilT. The colocalization of these two proteins at both poles of the cell in P. aeruginosa supports their cooperative roles. The structure of a typical T2S ATPase from V. cholerae, EpsE, was recently solved (Robien et al., 2003), and that of PilT from Aquifex aeolicus is approaching resolution (Forest et al., 2004). Determination of the structure of a PilB homologue from the T4P system will be very useful for comparison, and may shed light on the potential differences in function between these related proteins.

While it is clear that the motor proteins are involved in pilus extension and retraction, and that ATP hydrolysis is necessary for these events (Turner et al., 1993), the actual mechanism by which pilin polymerization and depolymerization occurs remains obscure. Mattick (2002) has suggested, based on the helical structure of the pilus fibre and the observation that cells with tethered pili do not rotate during retraction, that the hexameric motor protein complexes may revolve around the pilus during the process of pilin polymerization and depolymerization, in a manner analogous to that described for the F1/F0 ATPase. This is an interesting concept that remains to be tested experimentally.

The general picture emerging is that there is a balance between pilus retraction by PilT and pilus extension by PilB. Perturbation of components of the system, such as the minor pilins or the PilC (PilY1) protein, tilts the system towards pilus retraction. In this scenario, PilT acts in a quality control role, retracting any pili that are missing one or more of the necessary components (Winther-Larsen et al., 2005). In contrast, in the absence of the retraction-promoting factor PilU, the ability of PilT to effect retraction appears to be reduced and the system is skewed towards pilus extension. The concept of PilU as an agonist of PilT-mediated retraction is supported by the phenotype of Neisseria pilU mutants, which show defects in host cell contact-induced dispersal, requiring PilT activity (Park et al., 2002). However, Neisseria pilU mutants lack the exaggerated autoaggregation phenotype that is exhibited by pilT mutants, suggesting PilU does not merely support PilT function.

In the case of P. aeruginosa, pilT mutants are completely incapable of pilus retraction, but pilU mutants may retain a residual level of retraction such that they are still susceptible to pilus-specific phage despite their inability to twitch (Whitchurch and Mattick, 1994; Wolfgang et al., 2000). Even within individual cells, high local levels of PilU (which appears unipolar in P. aeruginosa) may favour pilus retraction at the high-PilU pole, and pilus extension at the opposite pole where PilU levels are low (Fig. 3). While our fluorescent fusion studies showed that PilU appeared to be localized to the flagellar (and therefore likely the piliated) pole of P. aeruginosa, these findings must be confirmed using complementary high-resolution techniques, preferably in live cells undergoing twitching motility. By analogy to other polar proteins such as the cell division Min proteins (Raskin and de Boer, 1999) or the Caulobacter differentiation protein DivK (Matroule et al., 2004), it is possible that PilU concentrations at a particular pole, and thus the balance of pilus extension versus retraction, could change rapidly by diffusion, allowing the bacteria to move by twitching in appropriate directions. High-resolution real-time microvideography of cells expressing fluorescent PilU fusions will be necessary to test this hypothesis. The basis for hyperpiliation at a single pole in pilU mutants is unknown, but suggests that other components likely contribute to the identification of the active pole.

image

Figure 3. Model of PilU function in Pseudomonas aeruginosa.

Download figure to PowerPoint

A look at the hole problem: crossing the cell envelope

  1. Top of page
  2. Summary
  3. Introduction
  4. Retraction of type IV pili
  5. How do they do it?
  6. A sticky problem; how do T4P mediate secure attachment?
  7. Getting to the bottom of it; the pilus motor
  8. A look at the hole problem: crossing the cell envelope
  9. Acknowledgements
  10. References

In considering the processes required for twitching motility to occur, an interesting dilemma arises. How can the pilus be simultaneously fixed within the cell envelope to the extent that detachment is prevented upon application of a pulling force, yet free to move through the envelope along its long axis to enable retraction to take place? A related problem involves the requirement for a sufficiently large conduit to allow passage of the pilus (estimated to be approximately 6 nm in diameter) (Craig et al., 2004) through the periplasm, including the peptidoglycan component, without compromising integrity of the cell wall. These issues have received relatively little attention despite increasingly intensive study of T4P and T2S over the last two decades.

Peptidoglycan (PG) forms a continuous mesh that completely surrounds the cytoplasmic membrane, providing mechanical protection against turgor-induced stresses and determining cell shape (reviewed in Holtje, 1998). As maintenance of an intact PG skeleton is essential for survival, enzymatic remodelling of the cell wall to allow growth and division is a highly regulated process. The PG mesh size of approximately 4–5 nm forms a permeability barrier to the passage of proteins larger than approximately 50 kDa (Demchick and Koch, 1996); therefore, controlled localized rearrangements in PG structure are necessary to allow macromolecular structures involved in adherence, motility or secretion to traverse the cell wall (Koraimann, 2003). Conjugative plasmids that encode T4P (i.e. E. coli Bfp or the thin pili of plasmid R64) often include genes for transglycosylase proteins, enzymes that cleave the beta-1,4 glycosidic bonds in the PG backbone. However, mutants lacking these proteins (i.e. BfpH or PilT of R64) appear to have wild-type levels of piliation, suggesting that their main function is not controlled remodelling of the PG to allow passage of the T4P. Instead, the T4P transglycosylase mutants show reduced ability to transfer DNA, suggesting that the proteins generate a PG conduit for this macromolecule in recipient cells.

Most bacteria possess multiple homologues of transglycosylase and transpeptidase proteins with nominally similar activities, complicating assignment of specific functions to individual proteins. Often these proteins can be inactivated without apparent phenotypical consequences (Denome et al., 1999), a result that is sometimes ascribed to functional redundancy. However, recent studies have begun to show that subtle differences in function are detectable under appropriate conditions (Ghosh and Young, 2003; Meberg et al., 2004). By screening P. aeruginosa mutants lacking specific autolysins, we have identified several that lack both twitching motility and surface piliation (C.V. Gallant, C. Daniels, J. Leung and L.L. Burrows, in preparation), suggesting that these chromosomally encoded proteins may be necessary for remodelling PG to allow surface expression of T4P. Other intriguing links between genes encoding PG biosynthetic proteins and those encoding T4P were noted over a decade ago (Hobbs and Mattick, 1993). For example, in both P. aeruginosa and N. gonorrhoeae, the gene encoding the bifunctional transglycosylase–transpeptidase PBP1a (ponA) is immediately upstream of the pilM-Q gene cluster (see below). Similarly, the P. aeruginosa homologue of the peptidoglycan-linked lipoprotein ComL, required for T4P-mediated transformation competence in many species (Fussenegger et al., 1996), is encoded immediately upstream of the pilRS two-component regulatory system and the fimT-pilE cluster encoding the minor pilins (Stover et al., 2000).

In addition to its barrier function, PG could act as a scaffold on which components of the macromolecular structures are assembled. Such a scaffold may provide the bracing function required to apply force to the pilus during retraction. Analysis of non-twitching mutants of P. aeruginosa generated by random transposon mutagenesis led to the identification of PilM and FimV, proteins potentially involved in T4P scaffold–cell wall interactions. PilM, which is the first gene in a cluster encoding PilM through PilQ, has homology to actin-like proteins MreB and FtsA (Martin et al., 1995), and intriguingly, MreB was recently shown to be involved in bipolar positioning of proteins (Soufo and Graumann, 2003). The other proteins in the pilM gene cluster could potentially form a polar cell envelope-spanning scaffold since PilN and PilO are predicted to be inner membrane proteins, PilP is a lipoprotein localized to the outer membrane (Drake et al., 1997) and PilQ is the dodecameric outer membrane secretion pore through which the pilus is extruded (Drake and Koomey, 1995; Martin et al., 1995; Bitter et al., 1998).

A transposon mutagenesis screen for non-twitching mutants led to the identification of FimV, a putative PG-binding protein required for twitching motility (Semmler et al., 2000). While the function of FimV is not clear, it has been implicated in PG remodelling as its overexpression resulted in formation of filamentous cells. This phenotype implicates FimV as a component involved in the process of creating a PG-spanning passage for the pilus, or in coupling the pilus scaffold to the cell wall. We have preliminary evidence suggesting that modulation of PG structure through inactivation of specific penicillin-binding proteins involved in PG remodelling can also have a negative effect on twitching motility without impairing growth of the cells (C.V. Gallant, C. Daniels, J. Leung and L.L. Burrows, in preparation), suggesting that a large number of proteins may participate in opening holes in a controlled fashion to allow pilus passage.

In summary, work over the past two decades has defined a large number of proteins involved in the regulation, synthesis, assembly and function of T4P. Clever studies have confirmed early predictions that the T4P are indeed retractable organelles, allowing future investigators to focus on the mechanics of this process, including the function of the motor proteins involved. Application of structural biology techniques has revealed the structure of the basic pilin subunit and at least one member of the motor protein family, EpsE. Such studies allow us to ask more specific questions about the function of individual motifs within these proteins, and to propose mechanisms of assembly. Protein–protein interaction studies have begun to reveal the nature of the assembly complex but technical challenges remain, as many of the key components are associated with the inner or outer membranes. Further study of these and related aspects of pilus biogenesis have the potential to explain mechanisms underlying the multiple functions of these fascinating bacterial organelles.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Retraction of type IV pili
  5. How do they do it?
  6. A sticky problem; how do T4P mediate secure attachment?
  7. Getting to the bottom of it; the pilus motor
  8. A look at the hole problem: crossing the cell envelope
  9. Acknowledgements
  10. References

I thank the members of my lab for their hard work, interesting ideas and creativity; Lisa Craig for supplying the illustration for Fig. 1; and Mike Koomey for providing the photo for Fig. 2. Work on T4P in my laboratory is supported in part by funding from the Canadian Institutes of Health Research, the Natural Sciences and Engineering Research Council of Canada, the Canadian Cystic Fibrosis Foundation and a Premier's Research Excellence Award from the Government of Ontario.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Retraction of type IV pili
  5. How do they do it?
  6. A sticky problem; how do T4P mediate secure attachment?
  7. Getting to the bottom of it; the pilus motor
  8. A look at the hole problem: crossing the cell envelope
  9. Acknowledgements
  10. References