JBC Papers in Press. Published on June 18, 2007 as Manuscript R

Transcription

1 JBC Papers in Press. Published on June 18, 2007 as Manuscript R The latest version is at CONTROL OF ACTIN ASSEMBLY DYNAMICS IN CELL MOTILITY Marie-France Carlier and Dominique Pantaloni Cytoskeleton dynamics and motility group, Laboratoire d Enzymologie et Biochimie Structurale CNRS, Gif-sur-Yvette, France Address correspondence to : Marie-France Carlier, LEBS, CNRS, 1 avenue de la Terrasse, Gif-sur-Yvette, France. Phone : ; carlier@lebs.cnrs-gif.fr Motile processes at the origin of cell migration, cell division, synaptic plasticity and endocytosis, but also morphogenetic processes that define anterior-posterior and ventral-dorsal polarity of the embryo, or left-right asymmetry, are driven by spatially and temporally controlled assembly of actin filaments (1-4). Cell biological studies and reconstituted motility assays indicate that polarized filament barbed end growth, at specific sites at the membrane, generates the forces responsible for lamellipodia and filopodia extension, as well as tensile forces that strengthen focal adhesions. Fluorescence speckle microscopy of actin and of regulatory proteins in motile cells show that morphologically and dynamically distinct networks of actin filaments initiated at the plasma membrane coexist in the cell and define cellular compartments (5, 6). How are these different turnover rates coordinated to achieve directional movement in response to extracellular cues? How can regulatory proteins recognize distinct actin networks? What are the relationships between the force produced by filament barbed end growth that deforms the membrane and the molecular mechanism of the protein machineries that link barbed ends to the membrane? New insight in these issues emerge from recent cellular, physical and biochemical approaches. Control of filament barbed end growth. In living cells, actin filaments (F-actin) are assembled at a steady state and turn over via pointed end depolymerization balanced by barbed end polymerization, a process called treadmilling. The treadmilling of pure muscle actin is very slow (of the order of 0.1 subunit/s, meaning that a 3 µm long filament is renewed in 2 hours). Treadmilling is accelerated by two orders of magnitude in motile processes like lamellipodia or filopodia, due to the activity of ADF/cofilins (7). In other structures like microvilli, treadmilling is slowed down, by spatially regulated barbed end capping of e.g. actin bundles in the stereocilia of the inner ear (8). Treadmilling thus appears as a universal mechanism that drives actin dynamics in eukaryotes. The stationary concentration of polymerizable ATP-G-actin is determined by the overall flux of pointed end depolymerization and by the availability and reactivity of barbed ends. Both are finely regulated. Polymerizable monomeric ATP-actin consists of free ATP-G-actin and complexes of ATP-G-actin with profilin or with functional homologs of profilin like WH2-domain proteins of the actobindin family (9-12). These proteins make a complex with G-actin that participates in barbed end assembly specifically. Profilin and its functional homologs thus play a crucial role in barbed end growth and motility. Most often, ATP-G-actin is in rapid equilibrium with these proteins (called collectively Wi) while filament assembly is an actual process. Each G-actin-Wi complex associates with free barbed ends with its own rate constant k B +i, and displays its own intrinsic barbed end critical concentration C B Ci. The steady-state concentrations of ATP- G-actin, [T] SS, and of the assembly competent complexes, [TWi] SS, depend on the flux of pointed end depolymerization. Proteins of the ADF/cofilin family, acting in synergy with Aip1, enhance pointed end depolymerization in a signal-responsive fashion (13). By increasing the steady-state concentration of ATP-G-actin, they contribute to increase the rates of barbed end nucleation and growth. The specific localization of ADF in the lamellipodium has been recently explained by the fact that the ADF-activating Slingshot phosphatase (14, 15) becomes active by binding the filaments that are initiated upon lamellipodium formation (16). The thermodynamic (C B Ci ) and kinetic (k B +i, k B -i ) parameters are affected by proteins that associate with barbed ends and may either block barbed end assembly and depolymerization, like capping proteins, slow down assembly and/or disassembly, or enhance barbed end growth like formins. Hence barbed ends are distributed in a variety of dynamically distinct classes. The values of [T] SS and [TWi] SS that are established at a given time in the cell depend on the number of 1 Copyright 2007 by The American Society for Biochemistry and Molecular Biology, Inc.

2 filament pointed ends and on the number of barbed ends in each class (Figure 1). The kinetic parameters of capping protein interaction with barbed ends vary greatly from one protein to the other. Capping Protein ab dissociates so slowly from barbed ends that this process could be limiting in motility. The uncapping activity of CARMIL (17, 18) may therefore be important in the regulation of barbed end growth. The capping activity may also be integrated in a modular protein, like Eps8, which spatially and temporally restricts its function, in response to local signals (19). Finally a protein known to sequester ADP-actin, twinfilin, has also been identified as a capping protein. One of the consequences of the combined activities of twinfilin is revealed in an integrated motility assay in which formin and twinfilin are operating simultaneously. The sequestration of ADP-actin has a buffering effect on the level of barbed end capping by twinfilin. In turn this buffering effect maintains a low remanent activity of formin (20). These results should foster experiments aiming at testing potential genetic interaction between twinfilin and formin. Conversely, the discovery of genetic interactions between two actin regulatory proteins should foster the design of reconstituted assays combining both proteins, to discover new emerging properties. Most capping proteins are soluble, however proteins that nucleate barbed ends or that regulate barbed end growth rate often operate at the plasma membrane in a signal-responsive fashion. The best example is provided by formins. The auto-inhibited fold of formins is relieved in two consecutive steps by the binding of Rho GTPase (21), thus targetting activated formin at the membrane. The function of formins in actin filament dynamics is tightly regulated by profilin, a cofactor which is required for the activity of formins in vivo. In the absence of profilin, formins nucleate filaments and cap barbed ends more or less tightly. Cdc12 is a strong capper (22), while mdia1 and Bni1 are «leaky» cappers (23). Formins use the property of profilin to couple ATP hydrolysis to filament assembly, to catalyze rapid processive barbed end assembly, enhancing the rate constant for barbed end growth by one order of magnitude (28). Thus rapid formin-based motile processes may occur concomitantly with other slower actin-based motile processes mediated by free barbed end growth, at a given steady-state concentration of ATP-G-actin. As a result, agents that lower the steady-state concentration of G-actin by stabilizing filaments (e.g. tropomyosin) are expected to further depress the formation of the branched filament array while still allowing formin-based motile processes. In conclusion, what we know from the biochemistry of actin and its regulators lets us anticipate that the assembly dynamics at individual barbed ends can vary to large extents in different regions of the cell. A more complex situation may occur if the diffusion of ATP-Gactin is low enough for significant gradients of G- actin to build up in regions of the cytoplasm. Control of the G-actin/F-actin ratio : Sequestering proteins In addition to polymerizable G-actin, cells contain a pool of non-polymerizable G- actin, which consists of G-actin in complex with sequestering proteins (S), like ß-thymosins (24). ß-Thymosin-actin complexes (TS) do not participate in filament assembly at either end and are in rapid equilibrium with free ATP-G-actin. Hence the amount of sequestered actin at steady state is determined, at any time, by the concentration of free ATP-G-actin, [T] SS : [TS] SS = [S 0 ].[T] SS /([T] SS + K S ) (1) where [S 0 ] is the total concentration of sequestering agent, and K S is the equilibrium dissociation constant for the TS complex. K S is higher than [T] SS, hence according to equation (1), changes in [T] SS generate large changes in the pool of sequestered actin, reflected by changes in the amount of F-actin. G-actin sequesterers do not affect the velocity of motile processes, since the TS complex does not feed barbed end growth. However, they do affect the force produced, because they determine the amount of F-actin that builds up when the creation of new filaments in response to signals causes the relaxation of the steady state of actin assembly to a lower value of [T] SS. Note that this relaxation process takes place without any massive production of free ATP-Gactin. Measuring the concentrations of polymerizable and non-polymerizable G-actin in vivo upon stimulus-induced changes in the motile state of the cell is a challenging task that will require refined imaging procedures that have not been developed yet. Insight into the mechanism of actin-based motility from reconstitution assays. Forces produced by the growth of actin filaments against a membrane, in response to 2

3 signalling (25) generate lamellipodial and filopodial protrusions, internalization of endocytic vesicles or propulsion of endosomes. Actin polymerization against the surface of intracellular pathogens Listeria and Shigella causes their propulsion, which made these bacteria the first natural functionalized particles mimicking the leading edge of migrating cells. The reconstitution of actin-based propulsion of a particle functionalized with N-WASP (26, 27) or with formin (28) in a solution of F-actin maintained at steady state with ATP as an energy source validated the principle of treadmilling as the fuel of movement in cells and delineated the minimum number of required proteins (ADF and profilin for formin, and additionally Arp2/3 and capping proteins for N-WASP). The same proteins were found essential for lamellipodium extension in vivo (29). Reconstituted motility assays also enabled controlled measurements of the force produced by actin polymerization (30). WASP family proteins use Arp2/3 complex to branch filaments at the leading edge, while formins may contribute to lamellipodium or filopodium or nerve growth cone extension, or even supplement the absence of Arp2/3 (31, 32). How does a lamellipodium start extending? Arp2/3 generates new filaments only by branching. When a N-WASP-coated bead or vesicle is placed in a reconstituted motility solution of barbed end-capped filaments, Arp2/3, ADF and profilin, the filaments are never seen to adsorb onto N-WASP-Arp2/3 bound to the surface. The unpolarized actin meshwork that forms at the bead surface actually starts from G- actin. The treadmilling model accounts for this fact as follows. The motility medium maintains a steady-state concentration of ATP-G-actin well above the critical concentration of barbed ends, favoring nucleation. Nuclei that have free barbed ends either abort by binding a barbed end capper, or branch at the bead surface upon transient interaction with the N-WASP-actin-Arp2/3 complex. Branching thus transiently protects from barbed end capping. It is likely that the same mechanism accounts for the initiation of the branched lamellipodial array. Similarly, formins use the dimer pre-nuclei present at steady state to initiate processive barbed end growth of new filaments at the leading edge. This process is more likely than the capture of barbed ends of existing filaments, which cannot diffuse rapidly in this confined cellular environment. How is actin assembly coordinated in the dendritic array and formin-induced bundles? How are the two arrays recognized by their specific regulators or bundling proteins? Reconstituted motility assays show that optimized actin-based movement is not obtained with the exact same medium composition for the WASP/Arp2/3 and formin systems. Propulsion of a WASP-functionalized particle by formation of a branched filament arrray relies on the balance between the sustained generation of new filaments at the particle surface by WASP- Arp2/3-induced filament branching and the disappearance of growing filaments by barbed end capping. This balance maintains a stationary number of growing filaments. The branched filament array thus contains an equal number of incorporated molecules of Arp2/3 and capping protein. Increasing the concentration of capping protein in the medium leads to an increase in branching density of the meshwork (27). In contrast, capping proteins, which compete with formin for barbed end binding, are negative regulators of the propulsion of formincoated beads (28). In conclusion, in vitro, capping proteins favor the formation of a highly densely branched array but inhibit formin-based processes. Remarkably, the same phenotype is observed in vivo (33): depletion of capping protein abolishes lamellipodia and favors formininduced filopodia (34). In conclusion, the same physical-chemical principles govern the dynamics of individual machineries and organize the crosstalk between them. The fact that two populations of filaments with different dynamics may be regulated by different proteins is amazing. The possibility cannot be discarded that the machineries that control barbed end growth also control the structure of the filament over some distance (35, 36). The actin filament has a versatile structure and subtle structural changes induced by barbed end-binding regulators may favor the binding of e.g. distinctive bundling proteins, thus further differentiating the function of actin arrays. This might explain why bundling proteins like fascin (37) or espin (38) are found associated with bundles of filaments initiated by different machineries. Attachments between the membrane and actin filaments during protrusion : from molecular processes to macroscopic behavior. The directionality of cell protrusions is determined and maintained by dynamic links between the membrane and the actin filaments. Different mechanisms were proposed to link the 3

4 membrane to growing filaments. The forcevelocity relationships predicted by the models derived from these different mechanisms are different. Protrusive force produced by the growth of a branched filament array has been most extensively sudied. Activated Arp2/3 complex was proposed to associate with the sides of actin filaments in the lamellipodium to initiate lateral branches. Insertional polymerization would occur in the small interval made available by the fluctuations of the membrane between the barbed ends and the lipid bilayer, and filaments would never attach to the membrane (39). This model is unable to reasonably account for directional protrusion. The perpetuation of the polarized dendritic pattern during sustained lamellipodium extension requires some kind of cyclic either transient or permanent - connection between the membrane and the filaments as they grow (40). Biomimetic assays of propulsion of N-WASP-functionalized microspheres or vesicles demonstrated that the actin tail was attached to the particle surface (30, 41, 42), suggesting that similar bonds exist between the filaments and the membrane during protrusion. Lateral mobility of lipids appears strongly hampered at the leading edge by association of filament barbed ends with membrane components (43). What kind of link then could generate filament attachment, yet allow barbed end growth? Are the attachments permanent during filament growth, or transient? Are the attachments linked to the filament branching activity of N-WASP or are they independent of branching? These issues, which are central to physical models of protrusion, are no doubt to be solved only by a thorough analysis of the biochemical steps in the reactions of filament branching and elongation, using solution polymerization kinetics, structural analysis of the complexes involved and microscopy assays of individual filaments initiated at a surface. Progress is still expected from all methods. Polymerization assays in bulk solution so far provided conflicting results regarding the molecular mechanism of filament branching. Some data favored side branching of filaments by Arp2/3 complex «activated» by N-WASP (44). Other data showed that branching occurred at and required free barbed ends (45-47). In the branching «activated» Arp2/3 complex, the WH2 domain of N-WASP binds one molecule of G-actin and the CA domain binds one molecule of Arp2/3. The WH2-actin subcomplex, like profilin-actin, can associate productively with barbed ends (48). Therefore N-WASP is an immobilized enzyme, either at the membrane or at the surface of Shigella or of beads and Arp2/3 complex, G-actin and the filament its subtrates (49). It was proposed that N-WASP-actin-Arp2/3 complex associated with a filament via productive association of WH2-actin to the barbed end, CA-Arp2/3 initiating a lateral branch. The rate of detachment of the branch junction from N-WASP plays a determinant role in propulsion. In motility assays, a pulse of fluorescent Arp2/3 led to initial labeling of N- WASP at the bead surface, followed by diffusional penetration of fluorescent Arp2/3 from the bead surface into the actin tail, at a rate identical to filament growth (27). A model of «tethered ratchet» for actin-based motility was derived, in which filaments are transiently attached to the bead surface (50, 51). Recent data show that the complex of the isolated WH2 domain with ATP-G-actin binds barbed ends and mediates the attachment of filaments at the beadtail interface (52). The attachments appear dynamic when N-WASP, containing both WH2 and CA moieties, is able to branch filaments, while they remain quasi-permanent when only the WH2-actin moiety associates with barbed ends. It is possible that ATP hydrolysis causes dissociation of WH2 from the barbed end (53). The Arp2/3 complex may also play a role in the transient attachment. In support to this view, the segregation of N-WASP at the rear of propelling giant liposomes appears to depend on Arp2/3 complex (Delatour et al., submitted). Structural organization of the subunits of Arp2/3 complex at the branch junction is expected to provide mechanistic insight into the branching process. Conflicting orientations of the complex axis have been proposed (54-56), and no information exists on the position of the WH2- bound actin incorporated in the branch. Biases may occur when computing differences in projection maps of branched filaments containing Arp2/3 associated or not with tags. In summary, the complete kinetic analysis of the elementary reactions involved in filament branching, and the characterization of transient complexes in the cycle of filament attachment-branching are important future trends of research. Formins do maintain a permanent attachment to the filament barbed ends during processive growth, and truly behave as endtracking stepping motors (57). The molecular mechanism of processivity of formins and the 4

5 role of ATP hydrolysis associated with profilinactin processive assembly are not fully understood. Non-invasive methods have not always been used to show evidence for processivity, and forces that maintain growing filament ends in the vicinity of the immobilized formin may bias the conclusions. Some data indicate that filaments can be assembled processively from ADP-actin as well as from ATP-actin, and that profilin is not required (58-60). Other data (28) indicate that formin uses a major property of profilin, which is to couple ATP hydrolysis to filament growth. The crystal structure of the formin-actin complex (61) shows that formin caps barbed ends and a structural change implying the weakening of a formin-actin contact is required to allow processive assembly. Recent data support the structural model. As long as ATP is not hydrolyzed following association of profilin-actin to barbed ends, formin-bound profilin actually caps barbed ends (62). If ATP hydrolysis is involved in processivity of formin, the forces developed by processive filament assembly may be much greater than the force of the order of a piconewton which is produced by a freely fluctuating filament end (63). This measurement has not been performed yet. Biochemical and structural analysis of the transient complexes that are formed at the forminbound barbed end is needed to understand how molecular reactions control macroscopic behavior. So far formins and WASP/WAVE- Arp2/3 are the only known machineries that control barbed end nucleation and growth in motile processes. Other actin-binding proteins involved in morphogenetic or developmental processes are emerging. The WH2 domain protein Spire/Eg6 (64-66) plays a role in the definition of anterior-posterior and dorso-ventral axes in early embryogenesis. Spire nucleates actin filaments in vitro (65). The phenotypes of mutants of Spire and of the formin Cappuccino are similar and are mimicked by profilin depletion, suggesting that some functional interplay links these three proteins to optimize barbed end growth in processes directing oocyte polarity. Concluding remarks Self-assembly of actin filaments is at the heart of morphogenesis and movement in eukaryotic cells. The in vivo analysis of the dynamics of distinct actin arrays, the discovery of protein machineries that couple actin filaments to membrane dynamics and signalling, the development of reconstituted motility assays, the use of nanophysics methods to measure forces developed by actin, have largely contributed to the expansion of the actin field to boundary disciplines. It may not be a dream to imagine that using reconstituted systems of increasing complexity, the coordinated motility of an artificial cell will eventually be mimicked. The intrinsic properties of actin itself and their potential use by regulators are constantly being unveiled in the light of cellular phenomena. Yet, in many instances the analysis of molecular events using classical biochemical and structural approaches remains the limiting factor in future progress. REFERENCES 1. Pollard TD, Borisy GG. (2003) Cell 112, Rafelski SM, Theriot JA. (2004) Annu Rev Biochem. 73, Dormann D, Weijer CJ. (2006) Curr Opin Genet Dev. 16, Kaksonen M, Toret CP, Drubin DG. (2006) Nat Rev Mol Cell Biol. 7, Ponti A, Machacek M, Gupton SL, Waterman-Storer CM, Danuser G. (2004) Science 305, Iwasa JH, Mullins RD. (2007). Curr Biol. 17, Carlier MF, Ressad F, Pantaloni D. (1999) J Biol Chem. 274, Lin HW, Schneider ME, Kachar B. (2005) Curr Opin Cell Biol. 17, Paunola E, Mattila PK, Lappalainen P. (2002) FEBS Lett. 513, Boquet I, Boujemaa R, Carlier MF, Preat T. (2000) Cell 102, Hertzog M, Yarmola EG, Didry D, Bubb MR, Carlier MF. (2002) J Biol Chem. 277, Hertzog M, van Heijenoort C, Didry D, Gaudier M, Coutant J, Gigant B, Didelot G, Preat T, Knossow M, Guittet E, Carlier MF (2004) Cell 117, Ono S, Mohri K, Ono K. (2004) J Biol Chem. 279, Niwa R, Nagata-Ohashi K, Takeichi M, Mizuno K, Uemura T. (2002) Cell 108,

6 15. Sarmiere PD, Bamburg JR. (2004) J Neurobiol. 58, Nagata-Ohashi K, Ohta Y, Goto K, Chiba S, Mori R, Nishita M, Ohashi K, Kousaka K, Iwamatsu A, Niwa R, Uemura T, Mizuno K. (2004) J Cell Biol. 165., Yang C, Pring M, Wear MA, Huang M, Cooper JA, Svitkina TM, Zigmond SH. (2005) Cell 119, Uruno T, Remmert K, Hammer JA 3rd. (2006) J Biol Chem. 281, Disanza A, Carlier MF, Stradal TE, Didry D, Frittoli E, Confalonieri S, Croce A, Wehland J, Di Fiore PP, Scita G. (2004) Nat Cell Biol. 6, Helfer E, Nevalainen EM, Naumanen P, Romero S, Didry D, Pantaloni D, Lappalainen P, Carlier MF. (2006) EMBO J. 25, Rose R, Weyand M, Lammers M, Ishizaki T, Ahmadian MR, Wittinghofer A. (2005) Nature 435, Kovar DR, Kuhn JR, Tichy AL, Pollard TD. (2003) J Cell Biol. 161, Li F, Higgs HN. (2003) Curr Biol. 13, Safer D, Nachmias VT. (1994) Bioessays 16, Ridley AJ. (2006) Trends Cell Biol.16, Loisel TP, Boujemaa R, Pantaloni D, Carlier MF. (1999) Nature 401, Wiesner S, Helfer E, Didry D, Ducouret G, Lafuma F, Carlier MF, Pantaloni D. (2003) J Cell Biol.160, Romero S, Le Clainche C, Didry D, Egile C, Pantaloni D, Carlier MF. (2004) Cell 119, Rogers SL, Wiedemann U, Stuurman N, Vale RD. (2003) J Cell Biol.162, Marcy Y, Prost J, Carlier MF, Sykes C. (2004) Proc Natl Acad Sci U S A 101, Gupton SL, Anderson KL, Kole TP, Fischer RS, Ponti A, Hitchcock-DeGregori SE, Danuser G, Fowler VM, Wirtz D, Hanein D, Waterman-Storer CM. (2005) J Cell Biol. 168, Nardo A, Cicchetti G, Falet H, Hartwig JH, Stossel TP, Kwiatkowski DJ. (2005) Proc Natl Acad Sci U S A 102, Mejillano MR, Kojima S, Applewhite DA, Gertler FB, Svitkina TM, Borisy GG. (2004) Cell 118, Shirenbeck A, Bretschneider T, Arasada R, Schleicher M, Faix J. (2005) Nat Cell Biol. 7, Orlova A, Prochniewicz E, Egelman EH. (1995) J Mol Biol. 245, Papp G, Bugyi B, Ujfalusi Z, Barko S, Hild G, Somogyi B, Nyitrai M. (2006) Biophys J. 91, Vignjevic D, Kojima S, Aratyn Y, Danciu O, Svitkina T, Borisy GG. (2006) J Cell Biol.174, Sekerkova G, Zheng L, Loomis PA, Mugnaini E, Bartles JR. (2006) Cell Mol Life Sci. 63, Pollard TD (2007) Annu Rev Biophys Biomol Struct. 36, Takenawa T, Suetsugu S. (2007) Nat Rev Mol Cell Biol. 8, Gerbal F, Laurent V, Ott A, Carlier MF, Chaikin P, Prost J. (2000). Eur Biophys J. 29, Trichet L, Campas O, Sykes C and Plastino J (2007) Biophys. J. 92, Weisswange I, Bretschneider T, Anderson KI.(2005) J Cell Sci. 118, Amann KJ, Pollard TD. (2001) Nat Cell Biol. 3, Pantaloni D, Boujemaa R, Didry D, Gounon P, Carlier MF. (2000) Nat Cell Biol. 2, Boujemaa-Paterski R, Gouin E, Hansen G, Samarin S, Le Clainche C, Didry D, Dehoux P, Cossart P, Kocks C, Carlier MF, Pantaloni D. (2001) Biochemistry 40, Falet H, Hoffmeister KM, Neujahr R, Italiano JE Jr, Stossel TP, Southwick FS, Hartwig JH. (2002) Proc Natl Acad Sci U S A. 99, Egile C, Loisel TP, Laurent V, Li R, Pantaloni D, Sansonetti PJ, Carlier MF.(1999) J Cell Biol. 146, Pantaloni D, Le Clainche C, Carlier MF. Mechanism of actin-based motility. (2001) Science 292, Mogilner A, Oster G. (2003) Biophys J. 84, Mogilner A. (2006) Curr Opin Cell Biol. 18, Co C, Wong DT, Gierke S, Chang V, Taunton J. (2007) Cell 128,

7 53. Le Clainche C, Pantaloni D, Carlier MF. (2003) Proc Natl Acad Sci U S A 100, Beltzner CC, Pollard TD (2004) J Mol Biol 336, Aguda AH, Burtnick LD, Robinson RC (2005) EMBO Rep 6, Egile C, Rouiller I, Xu XP, Volkmann N, Li R, and Hanein, D (2005) PLoS Biol 3, Dickinson RB, Caro L, Purich DL.(2004) Biophys J. 87, Kovar DR, Pollard TD. (2004) Proc Natl Acad Sci U S A 101, Kovar DR, Harris ES, Mahaffy R, Higgs HN, Pollard TD. (2006) Cell 124, Vavylonis D, Kovar DR, O'Shaughnessy B, Pollard TD. (2006) Mol Cell. 21, Otomo T, Tomchick DR, Otomo C, Panchal SC, Machius M, Rosen MK.(2005) Nature 433, Romero S, Didry D, Larquet E, Boisset N, Pantaloni D, Carlier MF. (2007) J Biol Chem 282., Footer MJ, Kerssemakers JW, Theriot JA, Dogterom M. (2007) Proc Natl Acad Sci U S A 104, Wellington A, Emmons S, James B, Calley J, Grover M, Tolias P, Manseau L. (1999) Development 126, Quinlan ME, Heuser JE, Kerkhoff E, Mullins RD. (2005) Nature 433, Le Goff C, Laurent V, Le Bon K, Tanguy G, Couturier A, Le Goff X, Le Guellec R.(2006) Biol Cell. 98, FOOTNOTES We acknowledge financial support from HFSPO (grant RGP C), the ANR (Physique et Chimie du Vivant), the Ligue Nationale contre le Cancer (équipe labellisée) and the european STREP (BIOMICS). FIGURE LEGEND Figure 1. Coordinated treadmilling of distinct arrays of actin filaments in cell motility This scheme emphasizes that an overall steady state of actin assembly allows different populations of filaments to coexist with distinct dynamics and barbed end growth rates. All filaments deplymerize from their pointed ends. ADF/cofilin binding to a fraction of actin filaments enhances pointed end depolymerization and feeds a pool of ADF-ADP-G-actin that feeds the pool of ATP-G-actin. ATP-Gactin is in rapid equilibrium with thymosin ß4 and with profilin (the «profilin-actin» pool actually encompasses the pool of all complexes of actin with proteins that are functional homologs of profilin). Barbed end growth is fed by ATP-G-actin and profilin-actin in various arrays : branched filaments initiated by immobilized N-WASP-Arp2/3 and non-branched filaments processively assembled by formins. Note that a gradient of ADP-Pi-F-actin (red) to ADP-F-actin (blue) exists on filaments that assemble from G-actin and profilin-actin, while ATP hydrolysis is coupled to formin-catalyzed processive assembly of filaments from profilin-actin. Transient (N-WASP-Arp2/3-mediated) or permanent (formin-mediated) attachments of barbed ends occur at the membrane during barbed end growth. The rate of barbed end growth is controlled by proteins (weak or strong cappers, formin, ) bound to the barbed end in rapid equilibrium. 7

8 8