letters to nature 17,18 19. White, J. H. et al. Heterodimerization is required for the formation of a functional GABAB receptor. diameter and up to several micrometres in length . Type IV pilus Nature 396, 679–682 (1998). biosynthesis occurs through the type II protein translocation path- 20. Kaupmann, K. et al. GABAB-receptor subtypes assemble into functional heteromeric complexes. Nature 396, 683–687 (1998). way and requires several accessory proteins in addition to the pilin 1 21. Liu, F. et al. Direct protein–protein coupling enables cross-talk between dopamine D5 and subunit . The PilT protein is dispensable for Tfp biosynthesis but is g-aminobutyric acid A receptors. Nature 493, 274–280 (2000). essential for both Tfp-mediated motility3,19,20 and for the DNA 22. Dixon, B. S., Sharma, R. V., Dickerson, T. & Fortune, J. Bradykinin and angiotensin II: activation of 13,20 protein kinase C in arterial smooth muscle. Am. J. Physiol. 266, 406–420 (1994). uptake step of genetic transformation . PilT belongs to a highly 23. Tsuchida, S. et al. Potent antihypertrophic effect of the bradykinin B2 receptor system on the renal conserved family of presumed ATPases, which partition to the inner vasculature. Kidney Int. 56, 509–516 (1999). membrane and cytosol and are thought to energize type II and IV 24. Bascands, J. L. et al. Bradykinin-induced in vitro contraction of rat mesangial cells via a B2 receptor protein translocation systems1,13,19,21. Two proteins of this family type. Am. J. Physiol. 267, F871–F878 (1994). 25. Zhuo, J. et al. Localization and interactions of vasoactive peptide receptors in renomedullary form hexameric rings strikingly similar to the rings formed by many 22 interstitial cells of the kidney. Kidney Int. 67, S22–S28. (1998). AAA-type ATP-dependent chaperones and proteases . Electron- a a bg 26. Quitterer, U. & Lohse, M. J. Crosstalk between G i- and G q-coupled receptors is mediated by G microscope studies of pilT mutants led to the hypothesis that Tfp exchange. Proc. Natl Acad. Sci. USA 96, 10626–10631 (1999). 3,15 27. Quitterer, U., Zaki, E. & AbdAlla, S. Investigation of the extracellular accessibility of the connecting promote cellular motility by retracting , perhaps through PilT- 1,4,19 loop between membrane domains I and II of the bradykinin B2 receptor. J. Biol. Chem. 274, 14773– mediated filament disassembly . Indirect evidence further sug- 14778 (1999). gests that other bacterial surface filaments might retract, including 28. Mastro, R. & Hall, M. Protein delipidation and precipitation by tri-n-butylphosphate, acetone, and the conjugative F pili of Escherichia coli23 and a type III export methanol treatment for isoelectric focusing and two-dimensional gel electrophoresis. Anal. Biochem. 24 273, 313–315 (1999). filament of Salmonella enteriditis associated with cell contact . Supplementary information is available on Nature’s World-Wide Web site However, filament retraction has been neither observed directly (http://nature.com) or as paper copy from the London editorial office of Nature. nor proven to power motility in any prokaryotic system. N. gonorrhoeae, the causative agent of gonorrhoea, provides an Acknowledgements excellent model for studies of twitching motility. It lacks rotary We thank B. Nu¨rnberg, for anti-Ga-common antibodies, and J. Heukeshoven for helpful flagella and components for type III export, and it produces only advice on high-resolution electrophoresis of hydrophobic proteins. This work was one known fimbrial structure, the Tfp. When we suspended Tfp- supported in part by the Deutsche Forschungsgemeinschaft. producing N. gonorrhoeae cells at low density in liquid medium, Correspondence and requests for materials should be addressed to U.Q. they attached to and crawled over the surface of a glass coverslip (e-mail: [email protected]). (Fig. 1). Under optimal conditions (see Methods), over half of the bacteria on the coverslip were motile. Figure 1b depicts the path of a crawling diplococcus (a joined pair of cells is the neisserial func- tional unit). Although most movements were short and directional changes occurred frequently, many directed movements of 2–5 mm ................................................................. − were observed. Cells crawled at ,1 mms 1 (Fig. 1c, d). This motility Pilus retraction powers bacterial was not due to passive diffusion. Motile cells consistently crawled twitching motility a Alexey J. Merz*†, Magdalene So* & Michael P. Sheetz†‡ *Department of Molecular Microbiology and Immunology, µ Oregon Health Sciences University, Portland, Oregon 97201-3098, USA 5 m ‡ Department of Cell Biology, Duke University Medical School, Durham, North Carolina 27705, USA .................................. ......................... ......................... ......................... ......................... ........ Twitching and social gliding motility allow many Gram negative bacteria to crawl along surfaces, and are implicated in a wide b range of biological functions1. Type IV pili (Tfp) are required for twitching and social gliding, but the mechanism by which these 1–4 filaments promote motility has remained enigmatic . Here we 5 µm 5 use laser tweezers to show that Tfp forcefully retract. Neisseria c gonorrhoeae cells that produce Tfp actively crawl on a glass 3 ) surface and form adherent microcolonies. When laser tweezers –1 2 are used to place and hold cells near a microcolony, retractile m s 1 µ Speed forces pull the cells toward the microcolony. In quantitative ( 0 experiments, the Tfp of immobilized bacteria bind to latex 0 30 60 90 120 beads and retract, pulling beads from the tweezers at forces that Time (s) can exceed 80 pN. Episodes of retraction terminate with release or d breakage of the Tfp tether. Both motility and retraction mediated − by Tfp occur at about 1 mms 1 and require protein synthesis and function of the PilT protein. Our experiments establish that Tfp filaments retract, generate substantial force and directly mediate 0 0.5 1 1.5 2 2.5 3 –1 cell movement. Speed (µm s ) Type IV pili are implicated in motility1–3,6, biofilm formation7, virulence8–11 and all three modes of prokaryotic horizontal genetic Figure 1 Piliated N. gonorrhoeae cells crawl on an inert surface. a, Cells crawling on a transfer (transformation12,13, conjugation14 and transduction15,16). glass coverslip. This is the first frame of the sequence analysed in b–d, and the tracked The Tfp fibre, a helical polymer of the pilin protein, is 6 nm in diplococcus is circled. Note that most cells are present as diplococci. b, Tracking of the diplococcus indicated in a during an interval of 140 s. Small circles show the position of † Present addresses: Department of Biochemistry, Dartmouth Medical School, Hanover, New Hamp- the tracked diplococcus at intervals of 67 ms. Large circle indicates the start point. c, Plot shire 03755-3844, USA (A.J.M.); Department of Biological Sciences, Columbia University, New York, New York 10027, USA (M.P.S). of velocity against time for the same track. d, Histogram of the velocities shown in c. 98 © 2000 Macmillan Magazines Ltd NATURE | VOL 407 | 7 SEPTEMBER 2000 | www.nature.com letters to nature out of a laser-tweezers trap strong enough to firmly arrest sus- When anti-pilus 1-mm beads were placed within 1–3 mmof pended cells or 1-mm latex beads (20 pN at 100 nm displacement). immobilized cells, the beads became dynamically tethered and Cells crawled at temperatures from 20 to 42 8C. Motility ceased were repeatedly pulled towards the immobilized cells (Fig. 3b). − ,10 min after the addition of chloramphenicol or tetracycline and The mean speed of retraction was 1.17 6 0.49 mms 1 (mean 6 s.d; n resumed upon drug washout, indicating a requirement for protein = 713), and was independent of the distance traversed by the beads synthesis (Table 1). Similarly, bacteria were non-motile and unable (Fig. 3c). Retraction events were separated by intervals of 1–20 s. to grow in defined medium lacking L-glutamine and pyruvate, and One-micrometre beads lacking the anti-pilin monoclonal antibody became motile upon addition of these nutrients. As expected, non- did not bind to pili as frequently as anti-pilin beads; but when they piliated pilE and pilF mutants were completely non-motile. Three bound, they also were pulled toward immobilized bacteria. Type IV piliated pilT mutants also failed to undergo large movements pili therefore exert retractile force through both nonspecific and (. 1 mm) and never crawled out of the laser trap (Table 1). specific binding interactions, consistent with our finding that Tfp N. gonorrhoeae cells thus can crawl over surfaces at rates of facilitate bacterial motility on inert substrates (Fig. 1). Retraction − ,1 mms1 by an active process requiring protein synthesis, Tfp ceased ,10 min after the addition of chloramphenicol or tetra- biogenesis and PilT3,19,20. cycline and resumed after drug washout, again indicating a require- Tfp-producing N. gonorrhoeae cells not only crawl but aggregate ment for protein synthesis. into microcolonies that contain twitching or writhing cells. Cells The restoring force of the laser tweezers increases with radial within 1–5 mm of a microcolony often move into the colony, displacement from the trap centre, and can be calibrated for whereas cells within a microcolony move out of the colony2,6–8.To homogeneous particles using laminar flow (see Methods). Figure determine whether retractile forces would pull dispersed cells 3c shows a trace of displacement against time, with force shown on together, we used laser tweezers to position isolated cells 1–5 mm the displacement axis. Isolated, immobilized diplococci were often (1–2 pilus lengths) away from microcolonies attached to a coverslip able to exert forces of 80 pN or more on trapped 1-mm beads. By (Fig. 2). The trapped cells were repeatedly pulled from the laser trap comparison, ,20 pN is the force needed to extract an integral toward the microcolonies, directly showing that there are retractile membrane protein from a lipid bilayer25, and ,30 pN of tensile forces between cells (Fig.