NANO LETTERS
Bacterial Motility and Clustering Guided by Microcontact Printing
2009 Vol. 9, No. 12 4553-4557
Claudia Holz,† Dirk Opitz,† Jan Mehlich,‡ Bart Jan Ravoo,‡ and Berenike Maier*,† Biology Department, Westfa¨lische Wilhelms UniVersita¨t, Schlossplatz 5, 48149 Mu¨nster, Germany, and Organic Chemistry Institute, WWU Mu¨nster, Corrensstrasse 40, 48149 Mu¨nster, Germany Received September 23, 2009; Revised Manuscript Received October 30, 2009
ABSTRACT Type IV pili are bacterial nanomotors that mediate two opposing behaviors on surfaces, spreading and clustering. Here we show that the velocity of motile Neisseria gonorrhoeae depends quantitatively on the fluidity of the phospholipid membrane surface. Using microcontact printing, we confined the surface motility to nonfluid islands within a fluid lipid membrane. On an array of islands, the transition from spreading to clustering was analyzed in real time and at the single cell level, showing that it was triggered by the number of bacteria (7.5 ( 0.3) for small islands and by the surface density (56 ( 2%) when the size of the island exceeded 25 µm2.
Bacterial type IV pili are the strongest linear molecular motors characterized to date. Individual pili generate forces that exceed 100 pN1,2 and pilus bundles generate forces in the range of 1 nN.3 These polymeric cell appendages are several micrometers long and 6 nm thick4 and may be assembled in vitro.5 In vivo, their length is dynamic, that is, pili polymerize from pilin subunits stored in the inner membrane and depolymerization supported by the ATPase PilT generates force on the anchoring point of the pilus at a surface. Pilus dynamics has various biological functions including surface adhesion, microcolony formation, surface motility, host cell signaling, biofilm formation, and horizontal gene transfer.6 Twitching motility (Figure 1a,b) is a form of surface motility employed by a multitude of pathogenic bacteria but also by soil-dwellers.7 Motility is driven by a cycle of pilus elongation, surface adhesion, and retraction whereby retraction of pili mediates movement of the bacterial cell body8 (Figure 1b). During infection of epithelial cells, the human pathogens Neisseria gonorrhoeae and Neisseria menigiditis form round dome-shape three-dimensional clusters called microcolonies.9,10 Mutants that do not generate pili are nonmotile and unable to cluster. Mutants that generate pili but are impaired in pilus retraction (depolymerization) are nonmotile and generate large irregular bacterial clusters that do not resemble round microcolonies.9,11 Thus pilus motors are required for two opposing cellular behaviors; they mediate bacterial motility and spreading on the one hand and clustering and microcolony formation on the other hand. * To whom correspondence should be addressed. E-mail: maierb@ uni-muenster.de. Phone: +49 251 8323920. Fax: +49 251 8324723. † Biology Department, Westfa¨lische Wilhelms Universita¨t. ‡ Organic Chemistry Institute, WWU Mu¨nster. 10.1021/nl903153c CCC: $40.75 Published on Web 11/05/2009
2009 American Chemical Society
Figure 1. Twitching motility of Neisseria gonorrhoeae. (a) Time lapse with growing track of an individual bacterium moving on a glass coverslip for 1 min. Scale bar: 5 µm. (b) Model of surface motility (twitching motility) mediated by type IV pilus retraction (left: side view, right: top view). (c) Velocity histogram of motile fraction of bacteria on BSA-treated surface for which velocity was averaged over time points spacing 1 s (218 individual bacteria, 140 min total tracking time).
Here, we address the question whether spreading or clustering can be controlled by the physicochemical properties of the adherent surface. During infection of human epithelial cells, N. gonorrhoeae are associated with rigid cellular structures reminiscent of micrometer-sized islands which include F-actin and lipid rafts.10,12 In these islands, various cellular receptors12 accumulate and it is unclear whether specific receptors or other
components of the host cell are necessary to induce microcolony formation. Therefore, we decided to establish a simple model system consisting of a surface that supports active pilus-mediated twitching motility and a second surface that does not support active motility. Various micropatterning techniques have been used previously to generate bacterial microarrays,13 in particular printing of bacteria,14 microcontact printing of adhesive molecules15,16 and other patterning techniques.17,18 However, since bacteria are bound irreversibly to a surface in all of these methods, they are not useful to study the dynamics of motile bacteria at surfaces. Therefore we explored surfaces with different charge and fluidity for their ability to support bacterial twitching motility. N. gonorrhoeae (MS11 or N400) were grown on agar plates and resuspended in cell culture medium ensuring that initial clusters formed on the agar plates were resolved (Supporting Information). After injection to microscope glass slides, they attached to the surface and a fraction of bacteria immediately began to crawl (Figure 1a, Supporting Information, movie 1). We quantified crawling (twitching motility) of individual bacteria by tracking the position of the bacterium for 1-2 min at a temporal resolution of 100 ms (Figure 1a). The velocity of individual bacteria was calculated in time intervals of 1 s. Since motile bacteria were clearly distinguishable from nonmotile bacteria (Supporting Information, Figure 1), we only analyzed motile bacteria in the following. We found that on a glass surface approximately 15% of bacteria were motile and the velocity distribution of motile bacteria was broad with an average velocity of V ) 1.43 ( 0.06 µm/s. On BSA-treated surfaces around 80% of the bacteria were motile and the average velocity was V ) 1.31 ( 0.02 µm/s (Figure 1c). Phospholipids are abundant in epithelial cell membranes and carry net negative surface charge. Therefore we generated solid-supported phospholipid membranes on glass by vesicle fusion (Figure 2a, Supporting Information). Piliated as well as nonpiliated bacteria attached well and within seconds to all examined solid supported membranes composed of neutral or negatively charged lipids as well as to compositions with different fluidities, demonstrating that no specific receptor was required for adhesion of bacteria to phospholipid membranes. To characterize membrane fluidity, we determined their diffusion coefficients D by fluorescence recovery after photobleaching (FRAP) experiments (Supporting Information). We then analyzed the ability of bacteria to move actively on supported lipid membranes consisting of different compositions of phospholipids. On neutral fluid DOPC (1,2-dioleoyl-sn-glycero-3-phosphatidylcholine) membranes (D ) 1.31 ( 0.78 µm2/s), the average velocity was V ) 0.25 ( 0.03 µm/s (Figure 2b, Supporting Information, Figure 2 and Table 1). For comparison, we analyzed the velocity in a nonmotile bacterial strain that did not generate the PilT motor protein (pilT-strain) and found no significant difference in velocity (Figure 2c), indicating that movement on DOPC was passive. Fluid charged membranes were prepared by vesicle fusion from 90% neutral DOPC and 10% anionic DOPA (1,2-pioleoyl-sn-glycero-3-phosphate) in MES (D ) 0.64 ( 0.49 µm2/s). The average velocity of twitching 4554
Figure 2. Velocity depends on fluidity and charge of solid supported membranes. (a) Sketch of bacterium adhered to supported membrane. (b) Typical 1 min tracks on nonfluid DOPA:DOPC (red) and fluid DOPC (blue). (c) Bacterial velocity V versus diffusion constant D of the membrane. Left, neutral lipids; right, 10% negatively charged lipids; red, wt; black, pilT- (pilus motor not expressed). The respective membrane compositions are listed in Supporting Information, Table 1. (d) Sketch and tracks of bacteria on azolecithin membrane supported by polycarbonate filters. Pores in the polycarbonate filters were visible as black circles. On pores membrane fluidity was high (D > 5 µm2/s) and on solid support fluidity was low (D < 10-3 µm2/s). Scale bars 5 µm.
motility increased to V ) 0.57 ( 0.03 µm/s (Figure 2b,c). Thus the surface charge affected the ability of bacteria to move actively at surfaces but the velocity was significantly lower than on solid surfaces including glass and BSA-treated surfaces. Pili act as grappling hooks that mediate bacterial motion as they generate force between the anchoring point at the surface and the cell body (Figure 1b). We suspected that as pili bound to lipids within a fluid membrane, force was generated by pilus retraction but not transmitted into movement since the pilus slid within the membrane. To test this hypothesis, we decreased the fluidity of the supported membranes while maintaining the surface charge. On neutral gel-like DPPC (1,2-dipalmitoyl-sn-glycero-3-phosphatidylcholine) membranes (D ) 0.06 ( 0.03 µm2/s), the average velocity was V ) 0.25 ( 0.05 µm/s (Figure 2c). On gel-like negatively charged membranes containing 90% DPPC and Nano Lett., Vol. 9, No. 12, 2009
10% negatively charged DPPG (1,2-dipalmitoyl-sn-glycero3-[phospho-rac-(1-glycerol)]) (D ) 0.04 ( 0.03 µm2/s), we found that the velocity increased to V ) 1.10 ( 0.05 µm/s (Figure 2c), that is, the velocity was higher than on fluid membranes but lower than on glass surfaces. Finally, we prepared nonfluid supported membranes (D < 10-4 µm2/s) by vesicle fusion with 90% neutral DOPC and 10% anionic DOPA in ddH2O. Since diffusion was undetectable with FRAP, we assessed the homogeneity of the membrane by FRAP and AFM (Supporting Information, Figure 3) and found that the membrane did not contain stationary defects. To verify that the fluidity of negatively charged nonfluid membranes was independent of the fluorescent probe, we used BODIPY-FL-DHPE and 18:1 NBD PE and found no significant change (data not shown). We therefore assume that strong interaction between the lipids and the surface is responsible for the absence of fluidity. On this surface, more than 80% of the bacteria twitched actively with an average velocity of V ) 1.40 ( 0.07 µm/s. Thus our results clearly show that pilus retraction supports active twitching motility on negatively charged lipid membranes and that the velocity is quantitatively determined by membrane fluidity. To investigate whether surface fluidity could guide bacterial movement, we generated supported membranes with varying fluidity. To this end, we coated polycarbonate filters with a pore diameter of 5 µm with azolecithin membranes (Figure 2d). We found that bacteria moved actively on nonfluid polycarbonate supported membranes (D < 10-3 µm2/ s), but they never entered the fluid membrane regions that were not supported by polycarbonate with D > 5 µm2/s (Figure 2d, Supporting Information movie 2). This experiment demonstrates that bacterial surface motility can be guided by membrane fluidity. We suggest that bacteria do not enter regions of high surface fluidity, since pili cannot exert force on the cell body due to slippage of pilus-bound lipids within the membrane. We therefore asked whether twitching motility could be confined to well-defined islands by microcontact printing. To this end, we adapted a method described by Kung et al.19 We printed BSA-stripes, dots, or triangles with an oxidized PDMS stamp to a glass coverslip and filled the gaps with DOPC membrane by vesicle fusion. Fluorescence labeling of the membrane showed that the BSA stripes were not coated with membrane (Figure 3). Fluidity of the membrane was verified by FRAP. Subsequently, we injected bacteria and analyzed their motility. We found that bacterial movement was confined to the BSA islands and bacteria did not enter the fluid membrane (Figure 3) (Supporting Information, movie 3). This experiment demonstrates that twitching motility can be confined to different geometric patterns through micropatterning without the requirement of threedimensional confinement. At low density of bacteria at the surface, bacteria occasionally collided and usually separated within seconds, indicating that short contacts cannot induce stable pilus-pilus or pilus-cell interactions. When we increased the number of bacteria per island, bacteria collided frequently and Nano Lett., Vol. 9, No. 12, 2009
Figure 3. Confinement of twitching motility to nonfluid microislands. BSA stripes, circles, or triangles were printed onto glass cover slides and the surrounding surface was filled with fluid DOPC membrane. BSA islands are visible as black shapes in the overlay of the brightfield image with the fluorescence image of the membrane containing NBD-C12 HPC. The traces show typical 2 min tracks of individual bacteria over 2 min. Scale bars: 5 µm.
initiated the formation of three-dimensional clusters on BSAcoated islands but not on the DOPC-coated area between the islands (Figure 4a). Initiation of microcolony formation was defined as the time point when multiple bacteria moved out of the focal plane of the microscope and the cluster formation did not resolve over a period of at least 30 min. To verify that microcontact printing of BSA did not by itself induce cluster formation we analyzed the behavior of bacteria on BSA islands on clean glass and found that cluster formation was not supported in the absence of DOPC (data not shown). Within one sample, bacteria were motile on islands with low number of bacteria, whereas on adjacent islands with high number of bacteria three-dimensional clusters were initiated (Figure 4b, Supporting Information, movie 4). It is conceivable that cluster formation is induced by irreversible adhesion of bacteria to BSA. On the other hand, the fact that bacterial movement from the fluid membrane on a BSA-island was likely but reverse movement was unlikely could lead to accumulation of bacteria on the islands. In the first case, we would expect that bacterial colonies would assume the shape of the BSA-coated islands. We therefore generated triangular BSA-islands and found that clusters did not adopt the shape of the triangle but bacteria preferentially formed round and dome-shaped cluster reminiscent of microcolonies that form during infection of epithelial cells (Figure 4c). To assess whether microcolony formation was induced by by a critical number Nc of bacteria confined to an island with area A, or by a critical surface density dc ) Nc·Abacterium/A of bacteria, we analyzed the initiation of microcolonies on islands with different diameters. We found that on islands with an area of 8-25 µm2 the transition to clustering was triggered by a constant number of Nc ) 7.5 ( 0.3 bacteria (Figure 4d). We note that smaller islands displayed a surface density exceeding 100% (Figure 4e) and were therefore disregarded. At 25 µm2, the surface density reached dc ) 56%. As the size of the island was further increased, the 4555
Figure 4. Transition from motility to microcolony formation. (a) Typical microcolony on BSA island and sketch of a bacterial microcolony. (b) Microcolony formation on BSA islands surrounded by DOPC. Left to right: short time scale dynamics. Top to bottom: long time scale dynamics of microcolony formation. (c) Microcolonies on BSA triangles surrounded by DOPC. (d) Minimum number of bacteria Nc required for initiation of microcolony formation as a function of island area A. (e) Minimum surface density of bacteria dc required for initiation of microcolony formation as a function of island area A. The lines are the best linear fits with slope 0. Images are overlay between brightfield (grayscale) and fluorescence of DOPC membrane with NBD-C12-HPC (yellow). Scale bars: 5 µm.
critical surface density dc was constant indicating that dc ) 56 ( 2% triggered the transition to clustering (Figure 4e). In conclusion, we demonstrated that microcontact printing in combination with vesicle fusion is useful for guiding bacterial surface motility and for studying the formation of microcolonies in real time and at the single cell level. Intriguingly, patterned surfaces did not merely act as traps that bind bacteria, but supported formation of dynamic N. gonorrhoeae microcolonies whereby the morphology is dome-shaped irrespective of the shape of the island. We suggest that reduced surface fluidity is likely to support microcolony formation on mammalian host cells through formation of lipid rafts10 and stiffening of the cytoskeleton.9 In contrast to other approaches, our patterning technique used physicochemical surface properties that did not interfere with bacterial motility and dynamics of microcolonies. Thus it should be useful for studying not only formation of microcolonies but also differentiation within bacterial clusters. This is particularly important for understanding the formation of bacterial biofilms which increase antibiotic resistance during bacterial infections. For example, the ability of bacteria to twitch on surfaces directly affects biofilm development with the opportunistic pathogen P. aeruginosa20 and most likely 4556
also with plant pathogens.21 We envision that different proteins and lipids may be printed to the surface that differentially influence bacterial gene expression or simulate matrices of bacterial biofilms. Acknowledgment. We would like to thank M. Koomey and A. Friedrich for supplying us with bacterial strains, C. Fregonese for help with mask design, S. Mu¨ller, A. Gru¨tzner, and W. Linke for support with AFM, and A. Ho¨ne and H.J. Galla for thorough discussions. This work was supported by the DFG through GRK 1409 and SFB 629. Supporting Information Available: Bacterial culture, preparation of solid supported membranes, coating of polycarbonate filter with azolecithin membrane, microcontact printing, FRAP, AFM, image analysis, supplementary figures, supplementary table, and supplementary movies. This material is available free of charge via the Internet at http:// pubs.acs.org. References (1) Maier, B.; Potter, L.; So, M.; Long, C. D.; Seifert, H. S.; Sheetz, M. P. Proc. Natl. Acad. Sci. U.S.A. 2002, 99 (25), 16012–7. (2) Clausen, M.; Jakovljevic, V.; Sogaard-Andersen, L.; Maier, B. J. Bacteriol. 2009, 191 (14), 4633–8. Nano Lett., Vol. 9, No. 12, 2009
(3) Biais, N.; Ladoux, B.; Higashi, D.; So, M.; Sheetz, M. PLoS Biol. 2008, 6 (4), e87. (4) Craig, L.; Volkmann, N.; Arvai, A. S.; Pique, M. E.; Yeager, M.; Egelman, E. H.; Tainer, J. A. Mol. Cell 2006, 23 (5), 651–62. (5) Audette, G. F.; E.J., v. S.; Hazes, B.; Irvin, R. T. Nano Lett. 2004, 4 (10), 1897–1902. (6) Allemand, J. F.; Maier, B. FEMS Microbiol. ReV. 2009, 33 (3), 593– 610. (7) Jarrell, K. F.; McBride, M. J. Nat. ReV. Microbiol. 2008, 6 (6), 466– 76. (8) Skerker, J. M.; Berg, H. C. Proc. Natl. Acad. Sci. U.S.A. 2001, 98 (12), 6901–4. (9) Higashi, D. L.; Lee, S. W.; Snyder, A.; Weyand, N. J.; Bakke, A.; So, M. Infect. Immun. 2007, 75 (10), 4743–53. (10) Mikaty, G.; Soyer, M.; Mairey, E.; Henry, N.; Dyer, D.; Forest, K. T.; Morand, P.; Guadagnini, S.; Prevost, M. C.; Nassif, X.; Dumenil, G. PLoS Pathog. 2009, 5 (2), e1000314. (11) Wolfgang, M.; Lauer, P.; Park, H. S.; Brossay, L.; Hebert, J.; Koomey, M. Mol. Microbiol. 1998, 29 (1), 321–30. (12) Merz, A. J.; Enns, C. A.; So, M. Mol. Microbiol. 1999, 32 (6), 1316– 32.
Nano Lett., Vol. 9, No. 12, 2009
(13) Weibel, D. B.; Diluzio, W. R.; Whitesides, G. M. Nat. ReV. Microbiol. 2007, 5 (3), 209–18. (14) Xu, L.; Robert, L.; Ouyang, Q.; Taddei, F.; Chen, Y.; Lindner, A. B.; Baigl, D. Nano Lett. 2007, 7 (7), 2068–72. (15) Rozhok, S.; Shen, C. K.; Littler, P. L.; Fan, Z.; Liu, C.; Mirkin, C. A.; Holz, R. C. Small 2005, 1 (4), 445–51. (16) Rowan, B.; Wheeler, M. A.; Crooks, R. M. Langmuir 2002, 18, 9914– 17. (17) Suo, Z.; Avci, R.; Yang, X.; Pascual, D. W. Langmuir 2008, 24 (8), 4161–7. (18) Rozhok, S.; Fan, Z.; Nyamjav, D.; Liu, C.; Mirkin, C. A.; Holz, R. C. Langmuir 2006, 22 (26), 11251–4. (19) Kung, L. A.; Kam, L.; Hovis, J. S.; Boxer, S. G. Langmuir 2000, 16 (17), 6773–76. (20) Singh, P. K.; Parsek, M. R.; Greenberg, E. P.; Welsh, M. J. Nature 2002, 417 (6888), 552–5. (21) Li, Y.; Hao, G.; Galvani, C. D.; Meng, Y.; De La Fuente, L.; Hoch, H. C.; Burr, T. J. Microbiology 2007, 153 (3), 719–26.
NL903153C
4557