Type IV pilus (T4P) dynamics is important for various bacterial functions including host cell interaction, surface motility, and horizontal gene transfer. T4P retract rapidly by depolymerization, generating large mechanical force. The gene that encodes the pilus retraction ATPase PilT has multiple paralogues, whose number varies between different bacterial species, but their role in regulating physical parameters of T4P dynamics remains unclear. Here, we address this question in the human pathogen Neisseria gonorrhoeae, which possesses two pilT paralogues, namely pilT2 and pilU. We show that the speed of twitching motility is strongly reduced in a pilT2 deletion mutant, while directional persistence time and sensitivity of speed to oxygen are unaffected. Using laser tweezers, we found that the speed of single T4P retraction was reduced by a factor of ≈ 2 in a pilT2 deletion strain, whereas pilU deletion showed a minor effect. The maximum force and the probability for switching from retraction to elongation under application of high force were not significantly affected. We conclude that the physical parameters of T4P are fine-tuned through PilT2.
Type IV pili (T4P) are polymeric cell appendages that are generated by a large variety of bacterial species. They are involved in initial adhesion to surfaces, surface motility, and transformation. Furthermore, they determine the morphology of microcolonies (Higashi et al., 2009) and biofilms (Lappann et al., 2006).
In contrast to other bacterial pili, the length of T4P is highly dynamic (Merz et al., 2000). They can elongate by polymerization of the pilin subunits (PilE) and retract by depolymerization. Cycles of T4P elongation, adhesion at surfaces, and subsequent retraction power surface motility, also termed twitching motility (Skerker and Berg, 2001). The distribution of pili along the cell body and the mechanisms of co-ordination of multiple pili are different for different species (Bulyha et al., 2009; Gibiansky et al., 2010; Holz et al., 2010; Jin et al., 2011). The ATPase PilF (Freitag et al., 1995) is essential for elongation. PilT, another ATPase, acts as an antagonist of PilF and supports retraction of T4P (Wolfgang et al., 1998). In the human pathogen Neisseria gonorrhoeae, PilT is essential for T4P retraction, whereas in Myxococcus xanthus T4P retraction occurs in a pilT deletion strain, albeit at a strongly reduced frequency (Clausen et al., 2009a).
During T4P retraction, individual T4P motors generate high force exceeding 100 pN (Maier et al., 2002; Clausen et al., 2009a). The concentration of PilT determines the probability for directional switching from T4P retraction to elongation (Maier et al., 2004; Clausen et al., 2009b). With increasing force the probability for elongation increases (Clausen et al., 2009b). The T4P motor of both N. gonorrhoeae and M. xanthus can work at two distinct speeds, namely at either ≈ 2 μm s−1 or ≈ 1 μm s−1 (Clausen et al., 2009a,b). In N. gonorrhoeae speed switching can be triggered through oxygen depletion (Kurre and Maier, 2012). Speed switching occurs within seconds and is independent of gene expression. So far, no genetic determinant of T4P retraction speed has been identified to our knowledge.
Different bacterial species have different numbers of PilT paralogues (Table S1). We performed a blast search for paralogues in 16 representative bacterial species and found that their number ranges from 0 (e.g. Francisella tularensis, Streptococcus sanguinis) to 4 in M. xanthus. Their sequence similarity with their respective PilT proteins is in the range of 60–70%. There is evidence that their phenotype is not conserved between different bacterial species. PilU shows high amino acid identity with PilT and both genes reside in the same operon. While pilU is essential for twitching in Pseudomonas aeruginosa (Whitchurch and Mattick, 1994), it is dispensable for T4P retraction (Park et al., 2002; Maier et al., 2004) and twitching motility in N. gonorrhoeae (Park et al., 2002; Maier et al., 2004) and N. meningiditis (Brown et al., 2010), but its deletion causes a strong increase in host cell adherence (Park et al., 2002). The Neisseria species (including N. gonorrhoeae, N. meningidits, N. lactamica) have a second paralogue of pilT, named pilT2. pilT2 is located within an operon together with pilZ and shows high amino acid similarity (61%) with pilT without apparent clustering apart from the Walker A and Walker B regions. Brown et al. found that deletion of pilT2 in N. meningiditis does not abolish twitching motility, although the level of piliation is enhanced by a factor of 2 and the formation of microcolonies is accelerated (Brown et al., 2010). PilT2 interacts with PilT, PilU, PilF, and with PilG (Georgiadou et al., 2012). Therefore it has been proposed to act as a hub with all the other traffic ATPases. Still, it is unclear how the pilT paralogues affect the dynamics and the probability for switching between different states of the T4P system.
Here, we quantified the effect of the pilT paralogues pilT2 and pilU on the physical properties of type IV pili in N. gonorrhoeae. We show that the speed of single T4P retraction and of twitching motility is strongly enhanced through PilT2, while force generation is unaffected. PilU plays a minor role in the regulation of T4P dynamics. Thus, we demonstrate that in addition to physical inputs, the speed of T4P retraction can be tuned by a genetically encoded factor.
PilT2 enhances the speed of twitching motility
To investigate the effect of the pilT-like gene pilT2 on dynamics and force generation by type IV pili, we generated a pilT2 deletion strain (ΔpilT2). By using an antibody against PilT2, Western blot analysis showed detectable levels of PilT2 in both wt and a pilT deletion strain, indicating that PilT was not essential for the stability of PilT2 (Fig. S1).
We quantified T4P-mediated motility (twitching motility) of individual cells. Cells were inoculated at low density onto BSA-coated glass coverslides and sealed. Upon contact with the glass surface, gonococci immediately started moving (Fig. 1a). The position of individual bacteria was tracked for 1–2 min at a temporal resolution of 100 ms (Fig. 1a) and the speed of individual bacteria was calculated in time intervals of 1 s. The speed distribution was centred around v = (1.58 ± 0.02) μm s−1 for wild type gonococci (Fig. 1b). We found that the pilT2-deletion mutant the average speed was lower with v = (1.04 ± 0.02) μm s−1. Complementation of pilT2 in the non-essential iga locus restored the speed close to wild type level.
The mean squared displacement (MSD) ⟨δ2(τ)⟩ is a measure for how far the bacterium moves away from its starting point within a given period of time. is the bacterial displacement between two time points. In other words, the higher the MSD, the more efficient is bacterial movement in exploring a surface. The two important parameters are the characteristic speed vc and the correlation time τc which describes how long the bacterium moves in one direction. We fitted our data to the model of a correlated random walk with an offset A (Holz et al., 2010) (Fig. 1c).
The correlation time obtained by the MSD (0.1 s < τ < 15 s) was nearly identical for all strains with τc = (2.3 ± 0.5) s for wt, τc = (2.3 ± 0.6) s for ΔpilT2 and τc = (2.2 ± 0.5) s for ΔpilT2 pilT2compl (Fig. 1d). As expected PilT2 deficiency led to a decreased characteristic speed of vc = (1.15 ± 0.08) μm s−1 in comparison to wt with vc = (1.60 ± 0.10) μm s−1 and ΔpilT2 pilT2compl with vc = (1.49 ± 0.10) μm s−1. Thus, in absence of PilT2 the correlation length lc = vc· τc decreased to lc = (2.6 ± 0.7) μm, whereas in wt lc = (3.7 ± 0.8) μm and in ΔpilT2 pilT2compl lc = (3.3 ± 0.8) μm. Due to the fact that the characteristic length of a single pilus is only lp = (0.9 ± 0.1) μm (Holz et al., 2010), movement over the correlation length must be still performed by multiple pili. These results indicate that PilT2 enhances twitching speed but does not alter characteristics of correlated persistent movement in N. gonorrhoeae.
The speed transition upon oxygen depletion is independent of PilT2
At this point, we tested whether PilT2 was a candidate for a molecular switch inducing global speed switching after oxygen depletion (Kurre and Maier, 2012). To address this question, we performed twitching motility assays in sealed chambers. Oxygen-dependent global switching was induced by bacterial oxygen consumption or actively by adding an oxygen scavenger system based on 50 nM protocatechuate-3,4-deoxygenase (PCD, Sigma-Aldrich) and 2.5 mM protocatechuic acid (PCA, Sigma-Aldrich) (Kurre and Maier, 2012). The PilT2 deficient strain ΔpilT2 showed oxygen-dependent speed switching with the same characteristic behaviour as found in wt (Fig. 2a). We found a low speed mode of vl = (0.46 ± 0.04) μm s−1 which is comparable with the low speed mode of wt with vl = (0.51 ± 0.02) μm s−1 (Fig. 2b). These experiments show that speed switching upon oxygen depletion occurs independently of pilT2 expression.
PilT2 enhances single pilus retraction speed
To assess whether reduced speed of twitching motility was caused by reduced T4P retraction speed, we characterized the dynamics of individual T4P using a previously established laser tweezers assay (Clausen et al., 2009b). We set the external force to F = 8 pN, i.e. the T4P experienced a pulling force of 8 pN. Oxygen-dependent global switching was induced by bacterial oxygen consumption or actively via oxygen scavenger (50 nM PCD/2.5 mM PCA) treatment.
We found that PilT2 deficiency results in a twofold decreased retraction speed in both speed modes (Fig. 3). As control, complementation of pilT2 led to complete recovery to wild type level. Single pilus retraction speed was not affected in a strain which had an additional copy of the pilT2 gene under the control of the pilE promoter in a non-essential locus (Fig. 3). Furthermore, we investigated the effect of the second pilT paralogue, pilU. Here, PilU deficiency led only to a slightly increased low speed mode, indicating that PilU plays only a minor role in fine-tuning pilus retraction speed. For both the high speed mode and the low speed mode, the retraction speed of the double mutant ΔpilT2 ΔpilU was significantly reduced as compared with the wt. The speed was slightly higher than that of the ΔpilT2 mutant (Table 1), but the difference is small. We conclude that deletion of pilT2 reduces the speed of single pilus retraction.
Table 1. Comparison of high speed mode vh and low speed mode vl
PilT overproduction does not complement the function of PilT2
PilT and PilT2 have a high sequence similarity. We therefore assessed whether PilT overexpression was able to complement PilT2 in its function of supporting fast T4P retraction. To this end, we constructed strain ΔpilT2 pilToe that has a chromosomal copy of pilT under the control of the pilE promoter, generating fivefold increased levels of PilT protein as compared with wt background (Clausen et al., 2009b). Twitching motility assays revealed no difference in speed between the PilT2 deficient strain ΔpilT2 and the PilT overexpressing strain ΔpilT2 pilToe (Fig. 4a), indicating that the function of enhancing speed was specific to PilT2. Using laser tweezers, we found that single pilus retraction speed of ΔpilT2 pilToe increased slightly as compared with ΔpilT2 but it did not restore wt level (Fig. 4b).
Force–dependence of T4P retraction in a pilT2 deletion strain
Using laser tweezers in force clamp mode, we investigated the effect of external force on the speed of T4P retraction (Fig. 5). Increasing the force from 8 pN to 60 pN caused a strong reduction in the high speed mode (aerobic) for both ΔpilT2 and ΔpilT2 pilToe. The effect was less distinct for the low speed mode (anaerobic). At 60 pN the speed was independent of oxygen, i.e. the high and low speed modes were indistinguishable.
In addition to pilus retraction speed, we sought to determine the stall force of the pilus machinery in the ΔpilT2 mutant. We therefore performed single pilus retraction assays in position clamp mode. Please note that in position clamp mode the retraction speed obtained at low loads (15 pN < F < 50 pN) is underestimated in the initial phase of pilus retraction. Stalling events were determined by screening all retraction events for retraction periods with speeds |v| < 150 nm s−1 for at least 20 ms. The stalling force of a single event was calculated by the average force of the longest period fulfilling this criterion. Events containing only corresponding retraction pauses were not considered. We measured the distribution of stall forces for wt and found that the stall force was considerably lower under anaerobic conditions with F = (133 ± 5) pN than under aerobic conditions with F = (206 ± 6) pN (Fig. 6). Unfortunately, force clamp data of ΔpilT2 and ΔpilT2 pilToe revealed the high speed mode merged with the low speed mode for F > 30 pN (Fig. 5). Thus, for pilT2 mutants detection of oxygen-dependent global switching was not successful. Consequently, we analysed stalling events without considering oxygen availability. For ΔpilT2 we determined 57 stalling events of in total 1248 retraction events. We found an average stall force F = (184 ± 7) pN which is close to wt level under aerobic conditions (Fig. 6). Note that the stall force histogram of ΔpilT2 exhibits a pronounced tail at lower forces. It is very likely that stalling events before and after oxygen depletion were collected, suggesting a bimodal distribution. We therefore interpret the distribution of ΔpilT2 as an overlay of aerobic and anaerobic conditions, indicating that PilT2 does not influence the stalling force.
Furthermore, the probability for retraction, elongation and periods of pausing as well as the duration of pausing was analysed for wt and ΔpilT2. To calculate, for example, the probability of retraction, we summed up the time of all retraction periods divided by the total time of all events. In comparison to wt (Fig. 7a), PilT2 deficiency does not influence elongation probability even at high external load (Fig. 7c). The frequency of retraction pauses was increased with respect to wt (Fig. 7b and d). We conclude that PilT2 is mainly responsible for accelerating pilus retraction and does not affect force generation by the pilus machinery.
The role of pilT paralogues in fine-tuning T4P dynamics
Neisseria gonorrhoeae possesses two paralogues of the pilus retraction motor PilT designated as PilT2 and PilU. Here, we have demonstrated that PilT2 accelerates twitching motility without affecting persistent movement. The decrease of twitching speed due to PilT2 deficiency correlates with a decrease of single pilus retraction speed. As we have significantly improved the accuracy of speed measurement since our previous report (Maier et al., 2004), we have re-evaluated the effect of PilU on the speed of pilus retraction. PilU deficiency only slightly enhanced pilus retraction speed in a pilT2 deletion background and therefore we conclude that it does not play a major role in regulating the speed of T4P retraction.
We assessed whether PilT2 or PilU showed phenotypes other than fine-tuning of T4P retraction speed. We found that deficiency of PilT2 did not affect the stall force or the elongation probability of T4P dynamics. Maier et al. demonstrated already that deficiency of PilU did not influence the stall force (Maier et al., 2004). Therefore, we conclude that high-force generation can be performed by PilT alone. In this context it is noteworthy that the stalling force was significantly higher than the stalling force obtained in previous experiments (Maier et al., 2002; Opitz et al., 2009). In previous experiments, we did not distinguish between aerobic and anaerobic conditions. Here, we found that under anaerobic conditions the stalling force was by a factor of ≈ 2 higher than under anaerobic conditions. Most likely, our previous experiments were performed mostly under anaerobic conditions. We verified in these experiments that the stalling events were generated by single pili, as the histogram of stalling forces was identical in mutants that produced a lower number of pili per cell (Maier et al., 2002). Here we most likely measured the stalling force of individual pili as the histograms show only one mode. Nevertheless, we note that our data are likely to be biased by the fact that T4P retraction is slower in ΔpilT2. As the connection between the pilus and the bead holds only for several hundreds of milliseconds and pausing/elongation was detected, the probability of reaching very high forces is likely to be decreased in ΔpilT2. Taken together, we conclude that PilT2 is mainly responsible for modulating the retraction speed.
Potential molecular mechanisms of tuning T4P dynamics
In N. gonorrhoeae the PilT subunit is essential for initializing retraction. Thus, it is very likely that PilT2 and PilU modulate the function of PilT. We envision two mechanisms, namely physical interaction between PilT and its paralogues, and control of pilin subunit integration. We will discuss both mechanisms in the following.
It is most tempting to speculate that PilT forms heterohexamers with its paralogues. It is very likely that ATP hydrolysis is cooperative (Satyshur et al., 2007; Misic et al., 2010), similar to other ring-like ATPases (Smith et al., 2011). Inserting a paralogue of PilT into the ring would then alter the cooperativity and change the speed of T4P retraction. Supporting this hypothesis, Georgiadou et al. recently found that PilT2 interacts with itself, PilT, PilU and PilF (Georgiadou et al., 2012). It is also conceivable that the PilT paralogues interact with PilT directly or indirectly without inserting into the ring-like structure and tune the speed of the pilus motor.
In addition to the major pilin subunit PilE, minor pilin subunits exist and are likely to be inserted into the pilus as they alter its phenotype (Aas et al., 2002; Winther-Larsen et al., 2005). Interestingly, deletion of several of the minor pilins leads to a piliation defect that is restored by simultaneously deleting pilT in gonococci (Winther-Larsen et al., 2005) and meningococci (Carbonnelle et al., 2006). It is therefore conceivable that the PilT paralogues control the composition of the pilus in terms of minor pilins. Their presence may alter the speed of T4P retraction.
Putative functional consequences of tuning T4P retraction speed
So far the biological relevance of fine-tuning T4P dynamics by PilT paralogues is not known. One can speculate that N. gonorrhoeae downregulates PilT2 and hence retraction speed upon a so-far unknown stimulus to become more sessile and promote biofilm formation. In agreement with this hypothesis, Brown et al. found that PilT2 deficiency in N. meningitidis accelerates cell clustering in vitro (Brown et al., 2010). Their pilT2 deletion mutant had a twofold increase in piliation with respect to wt, suggesting that faster microcolony formation of the pilT2 deletion mutant was caused by increased piliation.
Type IV pilus dynamics is described by a set of physical parameters including the speed of retraction, the retraction frequency, and the probability of directional reversal of the motor. These parameters can be fine-tuned by various inputs including force, oxygen concentration, and concentration of PilT retraction proteins. Here, we describe a genetic factor, namely PilT2, enhancing the speed of T4P retraction. This growing number of control parameters of pilus dynamics prompts us to suggest that bacterial behaviour including the transition between motile and sessile state can respond to different environmental inputs at various time scales.
Bacterial strains and plasmids
Gonococcal strains used in this study are described in Table 1. They were grown as described earlier (Opitz et al., 2009). Antibiotics were used for selection of gonococcal transformants at following concentrations: erythromycin, 8 μg ml−1 and kanamycin, 5 μg ml−1. Escherichia coli strain HB101 was used for plasmid propagation. It was grown on Luria–Bertani agar with 50 μg ml−1 kanamycin. Isolation and purification of plasmid DNA were performed using QIAGEN columns according to the manufacturer's specifications (QIAGEN, Düsseldorf, Germany).
Construction of pilT2 mutant strains
The oligonucleotides used as primers are listed in Table S2. The pilT2 knockouts were constructed by substituting the pilT2 gene in frame with either a kanamycin or an erythromycin resistance cassette in MS11 (wt). This was performed by transformation with PCR fusions containing the resistance cassette and flanking regions of pilT2 (pilT2::nptll, pilT2::erm). The primers N89 with N90 were designed for the upstream region and N93 with N94 for the downstream region of pilT2. The kanamycin resistance gene was amplified from the vector pUP6 (Wolfgang et al., 1998) using the primers N91 with N92. The erythromycin resistance gene was amplified from the vector p2/16/1 (Wolfgang et al., 2000) using the primers N106 with N107. The PCR product pilT2::nptll was used to generate the strain ΔpilT2. For strain ΔpilT2 pilT2compl, pilT2 including its promotor sequence was amplified from gDNA by the primers N99 and N100 and inserted into the unique SacI restriction site of the vector p2/16/1. The latter enables the insertion of pilT2 at an ectopic location (iga). The final construct was transformed into MS11, generating ΔpilT2.
To construct the pilT2 overexpressing strain pilT2oe, the pilT2 gene was fused to the strong pilE promotor and cloned into p2/16/1.
The pilU–pilT2 double knockout mutant GUT2 was generated by transformation of gDNA of ΔpilT2 into GU2.
For the strain ΔpilT2 pilToe, the PCR product pilT2::erm was transformed into a pilT overexpressing strain. For generating the pilT overexpressing strain (Clausen et al., 2009b), we transformed gDNA from an iga::PpilEpilT(kan) ΔrecA(tet) strain (generous gift from Michael Koomey) into MS11 and selected for kanamycin resistance, resulting in iga::PpilEpilT(kan).
Using a pilT inducible strain (Lee et al., 2005), the strain pilTind pilT2oe was constructed by transformation with gDNA of pilT2oe.
The transformations were carried out with either 500 ng of DNA (either PCR fragment or plasmid) or with 5 μg (gDNA) and 500 μl of gonococcal cells (∼ 5 × 108 cells ml−1) in GC broth (19) with 7 mM MgCl2 and 1× Isovitalex (BD Biosciences) at 37°C, 5% CO2, and under constant shaking with 250 r.p.m. After 30 min the cells were diluted 1:10 in GC broth with 7 mM MgCl2 and 1× Isovitalex and then incubated for 4 h before plating at different dilutions onto selective agar plates.
Twitching motility assays
Twitching motility assays of N. gonorrhoeae were performed on BSA (Sigma-Aldrich) coated glass in a commercial microscope equipped with a heated thermal insulation box equilibrated at 37°C [Details can be found elsewhere (Kurre and Maier, 2012)]. Bacterial motility was monitored via standard video microscopy (10 frames per second) and subsequent cell tracking in MATLAB R2009b (MathWorks). Tracking is based on the algorithms of J.C. Crocker and D. Grier originally written in IDL and transferred to MATLAB code by D. Blair and E. Dufresne (Crocker and Grier, 1996) (Download: http://physics.georgetown.edu/matlab/). Velocities vi(t) of single bacterial tracks were calculated by vi(t) = (r(t + Δt) − r(t))/Δt, where Δt = 1 s and r(t) is the position vector at time t.
Single pilus retraction assays
Single pilus retraction events were measured with optical tweezers in position clamp mode and force clamp mode (Clausen et al., 2009b). In short, in position clamp mode, the T4P pulls the bead out of the centre of the laser trap and the optical restoring force increases linearly with the deflection. This mode is useful for determining the stall force, but less convenient for determining speed since the elastic properties of the T4P are non-linear. In force clamp, a software-based feedback clamps the deflection of the bead from the centre of the trap at a constant position, thus the force acting on the T4P is constant. Bacteria were fixed to polystyrene spin-coated glass slides. Carboxylated polystyrene microspheres (Molecular Probes, Eugene, Oregon, USA) with 2 μm in diameter were added to cell suspension before sealing the chamber. Pilus retraction events were analysed in MATLAB as previously described (Kurre and Maier, 2012).
We thank Vladimir Pelicic for directing our attention to PilT2, Michael Koomey for providing us with strains, Thorsten Volkmann and Gerda Scheidgen for experimental support, and Ute Höcker for access to her imaging system. This project was supported by the Deutsche Forschungsgemeinschaft through MA3898.