Lowery+axon+growth+2009 PDF Review

REVIEWS The trip of the tip: understanding the growth cone machinery Laura Anne Lowery and David Van Vactor Abstract | The central component in the road tr...

REVIEWS The trip of the tip: understanding the growth cone machinery Laura Anne Lowery and David Van Vactor Abstract | The central component in the road trip of axon guidance is the growth cone, a dynamic structure that is located at the tip of the growing axon. During its journey, the growth cone comprises both ‘vehicle’ and ‘navigator’. Whereas the ‘vehicle’ maintains growth cone movement and contains the cytoskeletal structural elements of its framework, a motor to move forward and a mechanism to provide traction on the ‘road’, the ‘navigator’ aspect guides this system with spatial bias to translate environmental signals into directional movement. The understanding of the functions and regulation of the vehicle and navigator provides new insights into the cell biology of growth cone guidance. Chemotropic cue During the development of the nervous system, each but rather to the specific receptors that are engaged An external chemical cue, often neuron extends an axon through a complex and chang­ in the growth cone and the internal signalling milieu of found in a gradient, that leads ing environment to reach its final destination. At the the growth cone. In particular, the ‘navigator’ function to a directional growth in tip of each axon is the growth cone (BOX 1). The highly of the growth cone comprises the intracellular signalling response. dynamic behaviour of the growth cone and its respon­ elements that determine how environmental directions Protrusion siveness to multiple sources of spatial information lead to a given guidance response4. The stage of growth cone allows it to find its target with an impressive level of Despite notable advances over decades of research, progression in which there is accuracy. The growth cone ‘vehicle’ cannot move for­ our current understanding of how the growth cone extension of filopodia and ward without a ‘road’ to travel along. For growth cones, achieves its impressive road trip is far from complete. lamellipodia-like veils. this road comprises adhesive molecules that are either Here, we examine the basic cell biological features of Engorgement presented on a neighbouring cell surface (for example, growth cone guidance, focusing on the cytoskeletal The stage of growth cone transmembrane cell adhesion molecules (CAMs)1) or mechanisms that the growth cone uses as its vehicle to progression in which assembled into a dense extracellular matrix (ECM; move forward, as well as elements of the navigation sys­ microtubules further invade into the growth cone, fixing the for example, laminin and fibronectin2) (FIG. 1). These tem that convert spatial bias into steering by translating new axonal growth direction. molecules provide defined surfaces to which growth environmental guidance cues into localized cytoskeletal cone receptors can adhere, but they also activate intra­ remodelling. Although changes in membrane dynamics, Consolidation cellular signalling pathways that are used by the growth including the regulation of endocytosis and exocytosis, The stage of growth cone cone guidance machinery. Additionally, anti­adhesive, also have crucial roles in growth cone migration and are progression in which actin filaments at the neck of the surface­bound molecules (such as slits and ephrins3,4) likely to be targets of guidance cue signalling 11,12, this growth cone depolymerize and can prohibit the advance of the growth cone and thus topic is beyond the scope of this Review. We conclude the membrane shrinks to form provide ‘guard rails’ that determine roadway bounda­ by highlighting some of the key unsolved questions in a cylindrical axon shaft around ries. Finally, diffusible chemotropic cues represent the growth cone dynamics, and propose explanations for the bundle of microtubules. ‘road signs’ that present further steering instructions how recent technological advances would allow future to the travelling growth cone (FIG. 1). These include a investigations to further knowledge in these areas. range of molecules, including factors that were initially Harvard Medical School, identified explicitly in axon guidance assays3,4, as well as The growth cone vehicle 240 Longwood Avenue, Boston, Massachusetts morphogens5, secreted transcription factors6,7, neuro­ The growth cone engages its cytoskeleton to drive for­ 02115, USA. trophic factors8,9 and neurotransmitters10. Whereas ward and turn, continuously progressing through three Correspondence to D.V.V. it was originally thought that some cues always func­ stages of advance that are influenced by environmental e-mail: tion as attractive ‘go’ signals (for example, netrins) and factors: protrusion, engorgement and consolidation13,14 [email protected] doi:10.1038/nrm2679 others as repulsive ‘stop’ signals (for example, ephrins), (BOX 2). For the growth cone to navigate according to Published online it is now clear that the response to attraction or repul­ spatial landmarks, the motility machinery that drives 17 April 2009 sion is not due to the intrinsic property of the cue, forward movement must have the capacity to be biased 332 | MAy 2009 | VoluME 10 www.nature.com/reviews/molcellbio © 2009 Macmillan Publishers Limited. All rights reserved  REVIEWS seen substantial improvements in our molecular under­ Box 1 | The structure of the growth cone standing of F­actin retrograde flow and how it relates to growth cone motility and protrusion. It has been convincingly demonstrated that F­actin ret­ F-actin bundle rograde flow is driven both by contractility of the motor F-actin network protein myosin II, which seems to be tethered through protein–protein interactions in the transition (T) zone F-actin arc (the region between the peripheral (P) and central (C) Axon shaft domains of the growth cone), and the ‘push’ from F­actin Filopodium polymerization in the P domain (the region of the growth cone that includes filopodia and lamellipodia-like veils)17. Myosin II­driven compression across the T zone cir­ cumference causes buckling of the F-actin bundles (FIG. 2a), Stable which might be enhanced by pushing from leading edge microtubule actin polymerization17. This leads to bundle severing Lamellipodia-like veil C domain near the proximal ends17 and probably involves actin T zone filament­severing proteins of the actin­depolymerizing P domain factor (ADF)/cofilin family18. A recent paper suggests Dynamic that myosin II might also actively depolymerize actin fila­ microtubule ments19. After severing, the actin fragments are recycled into individual actin subunits and are available for trans­ The structure of the growth cone is fundamental to its function. The leading edge port to the periphery for further actin polymerization at Nature Reviews | Molecular Cell byBiology consists of dynamic, finger-like filopodia that explore the road ahead, separated the leading edge20 (FIG. 2a). sheets of membrane between the filopodia called lamellipodia-like veils (see the figure). The cytoskeletal elements in the growth cone underlie its shape, and the growth cone Engaging the clutch and forming traction to push ahead. can be separated into three domains based on cytoskeletal distribution14. The peripheral How does the growth cone use the actin engine to move (P) domain contains long, bundled actin filaments (F-actin bundles), which form the forward? Mitchison and Kirschner first proposed the filopodia, as well as mesh-like branched F-actin networks, which give structure to lamellipodia-like veils. Additionally, individual dynamic ‘pioneer’ microtubules (MTs) ‘clutch’ hypothesis21, also called the substrate–cytoskeletal explore this region, usually along F-actin bundles. The central (C) domain encloses stable, coupling model22, which links growth cone protru­ bundled MTs that enter the growth cone from the axon shaft, in addition to numerous sion to actin dynamics16,23. They suggested that growth organelles, vesicles and central actin bundles. Finally, the transition (T) zone sits at the cone receptor binding to an adhesive substrate leads interface between the P and C domains, where actomyosin contractile structures (termed to the formation of a complex that acts like a mole­ actin arcs) lie perpendicular to F-actin bundles and form a hemicircumferential ring33. cular clutch, mechanically coupling the receptors and The dynamics of these cytoskeletal components determine growth cone shape and F­actin flow, thus anchoring F­actin to prevent retro­ movement on its journey during development. grade flow and driving actin­based forward protrusion of the growth cone on the adhesive substrate (FIG. 2a). Indeed, growth cone–substrate adhesions have long spatially and to achieve accurate steering. In fact, been shown to be important for growth cone migra­ Actin treadmilling the steering and drivetrain are intimately connected tion 24 and, in fact, the generation of traction also The process by which the at a physical level. Therefore, if we are to fully grasp requires myosin II25. continual addition of actin subunits at the barbed end how guidance occurs, it is essential to understand the Filopodia, in particular, are considered to be guid­ of an actin polymer and underlying cytoskeletal mechanisms that propel ance sensors at the front line of the growth cone and disassembly of the polymer the vehicle forward and have the potential to be affected might have a major role in establishing growth cone– at the pointed end ensures asymmetrically. substrate adhesive contacts during environmental explo­ that the polymer stays of ration26. Studies show that filopodia function as points constant length, but individual subunits move along. Turning on the engine: F‑actin retrograde flow. Growth of attachment to the substrate and produce tension that cone motility and protrusion of the leading edge mem­ is used for growth cone progression27,28. Whereas earlier Filopodium brane depend on the dynamic properties of actin (BOX 3). studies that blocked filopodia formation using general A thin, transient actin Although actin might not be the only engine that powers F­actin inhibitors showed abnormal growth cone steer­ protrusion that extends from the cell surface and is formed axon elongation per se (axons that lack actin polymer­ ing 29,30, a recent study that specifically targeted filopodial by the elongation of bundled ization can still move forward, albeit with abnormal F­actin suggests that filopodia are dispensible for accu­ actin filaments in its core. growth cone morphology and substratum selectivity)15, rate growth cone guidance but are indeed required for actin is a central part of the mechanism that controls normal growth cone motility 31, supporting their role in Lamellipodia-like veil growth cone exploration. A combination of filamen­ forming adhesive contacts. A thin, sheet-like extension of cytoplasm between filopodia tous (F)­actin treadmilling and F­actin retrograde flow Accumulating evidence in recent years supports the that is formed by branched (the continuous movement of F­actin from the leading clutch model, in particular in vitro live growth cone actin networks. edge towards the centre of the growth cone) provide imaging experiments that use APCAM, a neural CAM the ‘motor’ that keeps the growth cone engine idling (NCAM) orthologue in Aplysia californica 32, a model F-actin bundle (FIG. 2a) and available to drive movement in response system with large growth cones that allow the high Long actin filaments that are crosslinked together in parallel, to directional cues16. Following increased technological resolution imaging of their cytoskeletal dynamics33. forming the core of filopodia. advances in live cell imaging, the past few years have Following APCAM­mediated growth cone–substrate NATuRE REVIEWS | Molecular cell Biology VoluME 10 | MAy 2009 | 333 © 2009 Macmillan Publishers Limited. All rights reserved REVIEWS to provide a molecular basis for similar clutch linkages Adhesive substrate-bound cues Repellent substrate-bound cues ‘The roadway’ ‘The roadway guard rails’ between actin and other CAMs and ECM receptors CAMs (Igs, cadherins and LRR) Slits and ephrins in growth cones. In non­neuronal systems, the focal ECM (laminin and fibronectin) Chondroitin sulphate proteoglycans adhesion proteins talin and vinculin provide a proto­ typic molecular clutch that is mediated by the integrin ECM receptors36. More recently discovered examples of growth cone­specific clutch machinery include caten­ ins, which mechanically couple N­cadherin receptors and F­actin flow in rat neurons37, and the novel protein shootin 1, which mediates linkage between l1CAM receptors and F­actin flow 38. A complete molecular understanding of the clutch mechanism will allow the transition from in vitro findings to embryonic sys­ tems and will provide a framework for understanding the overall logic that governs forward progression of the growth cone in vivo. MTs help steer the vehicle. Although actin structures Diffusible chemotropic cues are rapidly remodelled in response to guidance cues, ‘The road signs’ Classic guidance molecules actin is not the only component of the vehicle, as growth (netrins and semaphorins) cones cannot move forward without MT function 39. Morphogens and growth factors Whereas the mechanisms that control MTs have gener­ (Wnt, SHH, BMP and BDNF) Neurotransmitters ally received less attention than those of actin, recent Secreted transcription factors studies have confirmed early seminal experiments40 by showing that MTs have major roles in the process of Figure 1 | Directions for the trip. The growth cone encounters many different types Nature Reviews | Molecular Cell Biology growth cone steering in two complementary ways: indi­ of cues in its environmental terrain. It travels on a ‘road’ that is made up of adhesive molecules that are either presented directly on a neighbouring cell surface (for vidual P domain MTs act as guidance sensors, whereas example, transmembrane cell adhesion molecules (CAMs)1) or assembled into a dense bulk C domain MTs steer growth cone advance41. and complex extracellular matrix (ECM; for example, laminin and fibronectin2). First, before growth cone protrusion (BOX 2), a popu­ Additionally, anti-adhesive surface-bound molecules (such as slits, ephrins and lation of individual MTs actively explore the P domain42 chondroitin sulphate proteoglycans) can prohibit growth cone advance and thus (FIG. 2b) using their property of dynamic instability (BOX 3). provide the ‘guard rails’ that determine the road boundaries. Finally, diffusible Because the introduction of a localized adhesive cue leads chemotropic cues are the ‘road signs’ that present further steering instructions to the to an increase in the number of exploratory MTs that growth cone and include various diffusible chemotropic molecules (such as netrins interact with the adhesion site35, it has been proposed and semaphorins3,4), as well as morphogens (such as Wnt, sonic hedgehog (SHH) and that these MTs might act as guidance sensors (FIG. 2b). bone morphogenetic protein (BMP))5 and growth or neurotrophic factors (such as This might occur either by carrying signals involved in brain-derived neurotrophic factor (BDNF))5, secreted transcription factors6–9 and neurotransmitters10. Whereas it was originally thought that some cues function as steering to and/or from the cortical membrane, or by attractive ‘go’ signals (for example, netrins) and others as repulsive ‘stop’ signals (for acting as a scaffold for the localized recruitment of key example, ephrins), it is now clear that the response of attraction or repulsion is not due signalling components needed for navigation43. For to the intrinsic property of the particular cue, but rather to the specific growth cone example, dynamic MTs are required for the localized receptors that are engaged and the internal signalling of the growth cone. Green and accumulation of active Src family kinase signalling at red circles are interpreted as attractive and repulsive cues, respectively. sites of adhesion, which is necessary for the turning response of a growth cone to an adhesive substrate43, and MTs probably bring other signalling molecules as well adhesion, dramatic local reorganization of actin occurs. (for example, Rho family GTPase regulators). Increased levels of localized actin assemble at the site The second major role of MTs in steering occurs dur­ of adhesion, followed by regional slowing of retrograde ing engorgement (BOX 2), after initial actin remodelling in flow and growth cone protrusion16,34,35. Subsequently, response to cues. At this point, stable, bundled C domain F­actin bundles disappear between the adhesion site MTs move into the area of new growth, as consolida­ F-actin arc and the C domain (the region of the growth cone in tion of a new region of axon shaft occurs behind them, An actomyosin contractile which the microtubules (MTs) from the axon shaft enter thereby fixing the axonal direction16,44. Further support­ structure that is perpendicular the growth cone), and F-actin arcs reorientate from the ing the instructive role of MTs in growth cone steering, to bundles of actin, forming a C domain towards the adhesion site, creating a corridor the inhibition of MT dynamics prevents growth cone hemicircumferential ring in the between the two regions. Actomyosin­driven tension turning in response to guidance cues, whereas localized transition zone. builds up between the actin assembly at the adhesion MT stabilization induces turning 44. Dynamic instability site and the actin arcs with their associated MTs, and the The state used to describe growth cone undergoes engorgement as the C domain MT interactions with actin. The role of MTs during microtubule polymer moves forward35 (BOX 2). growth cone steering clearly requires the participation dynamics, in which microtubule polymers cycle through Whereas many studies that investigated the clutch of, and interaction with, actin45,46. Recent live­imaging periods of growth, shrinkage hypothesis focused on the cell adhesion molecule studies show that the function of actin dynamics might and occasional pausing. APCAM in A. californica, recent studies have begun be to provide spatio­temporal guidance to MTs to steer 334 | MAy 2009 | VoluME 10 www.nature.com/reviews/molcellbio © 2009 Macmillan Publishers Limited. All rights reserved  REVIEWS Box 2 | Stages of axon outgrowth As dynamic MTs preferentially explore the growth cone periphery, they usually follow the trajectories a Encounter substrate c Engorgement of F­actin bundles42, which are thought to guide MT advance into the P domain41,45. A recent study, however, showed that these F­actin bundles are not required for MT advance; indeed, they inhibit MT penetration into the P domain when the MTs are coupled to F­actin bundle­ specific retrograde flow 47 (by MT–actin crosslinking proteins). As MT coupling to F­actin retrograde flow directly affects the ability of MTs to explore the P domain, C domain moves forward it seems likely that regulation of MT–actin coupling and b Protrusion d Consolidation uncoupling (the release of MTs from F­actin retrograde flow) would have an effect on MT dynamics. This pre­ New axon diction was shown in a more recent study that examined shaft forms P domain MTs35. Not only does increased MT–actin uncoupling allow dynamic MTs to explore growth cone sides more frequently than central regions under uniform conditions (which might account for increased sensitivity to guidance cues that are not on the current axon out­ Filopodia and growth path), but it also allows an increased number of lamellipodia-like veils Microtubules C domain F-actin T zone MTs to explore sites of APCAM­mediated adhesion35. It extend forward Substrate P domain is possible that repellent guidance cues induce an oppo­ site response in the actin–MT coupling state, thus leading A traditional description of the axon outgrowth process separates it into three stages: to an opposite effect on the cytoskeletal machinery. Nature Reviews | Molecular Cell Biology protrusion, engorgement and consolidation13,14. These occur upon encountering Interestingly, the absence of F­actin bundles does attractive, adhesive substrates. This sequence during growth cone progression provides a framework for understanding detailed molecular mechanisms, and we assume that some not lead to the inappropriate advance of C domain of the same mechanistic events are used in response to diffusible chemotropic cues. MTs47, suggesting that although F­actin bundles regu­ The distal end of the growth cone contacts an adhesive substrate (see the figure, late exploratory P domain MTs during protrusion, stable part a). The binding of growth cone receptors activates intracellular signalling C domain MT movement into the growth cone during cascades and begins the formation of a molecular ‘clutch’ that links the substrate to the engorgement might be regulated by another mechanism, actin cytoskeleton. During protrusion, this clutch strengthens, resulting in regional namely the actin network and actin arcs. Indeed, another attenuation of filamentous (F)-actin retrograde flow. This anchors the actin with study showed this to be the case. Disruption of actin arcs respect to the substrate so that, as F-actin polymerization continues in front of the results in the failure of MT consolidation during axon clutch site, the lamellipodia-like veils and filopodia of the peripheral (P) domain move outgrowth, leading to an abnormally broad C domain49. forward to extend the leading edge (see the figure, part b) (see ReF. 124 for a During engorgement, as the C domain advances towards discussion of the molecular ratchet model for membrane protrusion). Engorgement occurs after actin clears from the corridor between the adhesion and the central (C) an adhesion site, actin arcs on the sides of the C domain domain, perhaps as F-actin behind the clutch is severed and removed (see the figure, become more prominent (BOX 2) and mechanical connec­ part c). F-actin arcs reorientate from the C domain towards the site of new tivity between extracellular adhesion, actin arcs and the growth16,34,35, followed by the invasion of C domain microtubules (MTs) into this region, C domain is apparent49. Thus, actin arcs normally form a which are guided by transition (T) zone actin arcs and C domain actin bundles. Finally, barrier around the C domain that regulates MT advance consolidation of the recently advanced C domain occurs as the proximal part of the by capturing MTs on the sides of the growth cone and growth cone compacts at the growth cone neck to form a new segment of axon shaft transporting them into the C domain. In a separate study, (see the figure, part d). The myosin II-containing actin arcs compress the MTs into the it was shown that myosin II, which mediates the contrac­ newly localized C domain (followed by MT-associated protein stabilization). Retraction tion of antiparallel actin filaments found in actin arcs, has of the filopodia away from the area of new growth occurs as F-actin protrusive activity an important role in actively transporting MTs from the is suppressed in these regions (also promoted by myosin II activity52), further promoting axon shaft consolidation. These three continuous and overlapping stages occur during sides into the C domain, compressing them into bun­ the formation of nascent axons, and also when new growth cones form from an axon dles and perhaps securing them50 until they are stably shaft during axon branching14,125. crosslinked by MT­associated proteins (MAPs) in the growth cone neck51. Myosin II in the growth cone neck has also been shown to suppress F­actin protrusion to allow axon shaft consolidation52, and this function might the growth cone in the right direction. In particular, contribute to its function during growth cone turning 53. actin has a pivotal role in determining MT localiza­ Therefore, distinct classes of actin structures seem to tion in the growth cone, acting as both a barrier to regulate different populations of MTs. Whereas F­actin premature MT invasion and as a guide to MTs dur­ bundles can inhibit the protrusive activities of P domain ing their advance45,47. Furthermore, local perturbation MTs when they are coupled together, F­actin arcs regu­ of actin structures leads to the redistribution of MTs late the engorgement and consolidation activities of Actin network and a change in the direction of growth48. Here, we C domain MTs (FIG. 2c). Therefore, an obvious question Actin filaments that are crosslinked in a branched discuss two important interactions between MTs and is: how are MT–actin interactions regulated in a spatio­ pattern, forming the structure actin: between P domain MTs and F­actin bundles, temporal manner in response to guidance cues? This topic of lamellipodia-like veils. and between C domain MTs and actin arcs (FIG. 2c). is addressed below. NATuRE REVIEWS | Molecular cell Biology VoluME 10 | MAy 2009 | 335 © 2009 Macmillan Publishers Limited. All rights reserved REVIEWS Box 3 | Cytoskeletal dynamics Actin filaments are polarized polymers that are composed Actin filaments of actin monomers. Their formation, stability and Minus Plus destruction are carefully regulated at every stage in the (pointed) (barbed) growth cone. Actin monomers can be added to either end end end (see the figure), but changes in the equilibria of Facing T zone Facing leading edge polymerization dynamics depend on whether ATP or ADP Transport for recycling is associated with actin. In the growth cone, ATP–actin is Microtubules usually added to the plus (or barbed) end that points Minus Plus towards the cell membrane, ATP hydrolyses to form end end ADP–actin, and ADP–actin disassembles at the minus Facing T zone Facing leading edge (or pointed) end that faces the transition (T) zone. Monomer-binding proteins then transport actin back to the leading edge to support further growth. Other actin-binding proteins include nucleation factors that create new actin plus ends for growth, capping proteins Nature that Reviews block growth | Molecular or Cell Biology disassembly, antagonists of capping proteins, actin filament-severing proteins and filament-stabilizing proteins, such as those that assemble F-actin into higher-order structures (for example, bundles and networks) and those that anchor F-actin to specific regions of the membrane (reviewed in ReF. 126). Microtubules (MTs) are polarized structures that are composed of α- and β-tubulin dimers and are assembled into linear arrays. A linear array of alternating α- and β-tubulin subunits form a protofilament, 11–15 of which form the wall of the MT. GTP–tubulin dimers are added to the plus end, and GDP–tubulin dimers dissociate from the minus end following GTP hydrolysis (see the figure). In growth cones, MT plus ends, which face outwards towards the periphery, exhibit dynamic instability — they cycle through periods of growth, shrinkage and occasional pausing127. Numerous proteins bind to MTs: some proteins stabilize MTs (for example, MT-associated protein 1B (MAP1B)111), some act as MT motors (for example, dynein and kinesin128) and others are part of a family called plus-end tracking proteins (+TIPs), which have been implicated in dynamic control of plus end MTs and linking MTs with actin- or membrane-associated structures (for example, end-binding proteins (EBs), adenomatous polyposis coli (APC) and CLASP (cytoplasmic linker protein-associated protein97,98)). The growth cone as a navigator (for example, fibroblasts, neutrophils and Dictyostelium Thus far, we have described how changes in the cytoskel­ discoideum60,61). Because substantial differences exist in etal machinery drive the forward progression of the molecular content between cell types, we must be careful growth cone vehicle. However, growth cone pathfinding not to assume that the systems act identically. Nonetheless, obviously does not consist solely of moving forward; it is numerous studies do suggest certain parallels in cytoskel­ a dynamic process in which the growth cone progresses, etal signalling between neuronal and non­neuronal pauses, turns and retracts as it navigates through the cells60,61. Additionally, common cell biological mecha­ embryonic landscape and encounters various directions nisms underlie growth cone guidance and regulation of for its trip. Spatial bias in a given direction can occur other aspects of axon biology, such as axon initiation and through either positive cues that increase protrusion modelling of secondary axonal branches62. Thus, studies (towards the side of new growth) or negative cues that on cytoskeletal signalling in other systems might provide decrease protrusion (occurring on the side away from insights into the mechanisms of growth cone guidance new growth). For spatial discontinuities in the environ­ (and vice versa). ment to drive growth cone steering and, in particular, to accurately interpret numerous cues simultaneously, Rho GTPases behind the wheel. Rho GTPases, which the growth cone requires a navigation system that can include RhoA, RAC1 and CDC42, are signalling nodes translate multiple environmental directions and inte­ that couple upstream directional cues and downstream grate separate signalling pathways to locally modulate cytoskeletal rearrangements to either enhance actin the dynamics of the cytoskeletal machinery. The overall polymerization for protrusion or promote disassembly logic that governs this process is still emerging. There and actomyosin contraction for retraction58,59 (FIG. 3). If is a vast literature that describes specific aspects of the regulation of Rho GTPase activity is to convey guidance growth cone navigation system, but many studies focus information, its upstream regulators must be activated in on individual pathways that are engaged by particular a spatially specific manner, internally reflecting the extra­ cues or receptors. Although there are numerous signal cellular environment. upstream regulators include the transduction molecules that convey guidance informa­ proteins that activate Rho GTPases (guanine nucleotide­ tion, including kinases54,55, phosphatases56 and calcium exchange factors (GEFs)) and those that inactivate them ions57, our most comprehensive understanding is of the (GTPase­activating proteins (GAPs))58,63 (FIG. 3; a further, Rho family of GTPases, a class of molecules that control newly implicated method of Rho GTPase regulation is cytoskeletal dynamics downstream of nearly all guidance the local translation downstream of guidance signalling 64, signalling receptors58,59 (FIG. 3). which is briefly discussed in BOX 4). Many Rho GTPase Although increasing numbers of studies have analysed regulators have been studied in non­neuronal systems, the functions of Rho GTPases and their cytoskeletal effec­ but our understanding of their specific functions in the tors in the growth cone, much of our understanding of growth cone is not as advanced63. Growth cone guid­ these molecules still comes from non­neuronal systems ance receptors can contain their own GTPase regulatory 336 | MAy 2009 | VoluME 10 www.nature.com/reviews/molcellbio © 2009 Macmillan Publishers Limited. All rights reserved  REVIEWS b MTs as guidance sensors Exploratory MTs recruited in response to guidance cues Growth cone b a MTs explore filopdia MTs act as scaffolding for c guidance cue signalling a Idling growth cone engine F-actin c Interactions between actin and MTs retrograde flow Protrusion Actin severing and Arcs constrain and depolymerization guide C domain MTs Actin treadmilling Bundles inhibit and guide F-actin polymerization P domain MTs at leading edge Myosin II contractility F-actin retrograde Protrusion Engaging the clutch and pushing ahead flow slows increases Substrate coupled F-actin polymerization to actin; clutch at leading edge engaged continues MTs F-actin Substrate C domain T zone P domain Figure 2 | The growth cone ‘vehicle’. Boxed regions of the growth cone are shown in subsequent Nature Reviewspanels. a | Together, | Molecular Cell Biology filamentous (F)-actin treadmilling (in which F-actin is polymerized at the leading edge and severed at the transition (T) zone, with the subunits recycled back to the leading edge) and F-actin retrograde flow (the continuous movement of F-actin from the leading edge towards the centre of the growth cone) keep the growth cone engine idling. When retrograde flow and polymerization forces are balanced, no protrusion occurs. When filopodia encounters an adhesive substrate, growth cone receptors bind to the substrate and are coupled to F-actin through ‘clutch’ proteins. This engages the clutch, anchoring F-actin with respect to the substrate and attenuating F-actin retrograde flow. Further F-actin polymerization pushes the membrane forward, which results in growth cone protrusion. b | Peripheral (P) domain microtubules (MTs) explore filopodia along F-actin bundles and might act as guidance sensors. As a filopodium encounters a guidance cue, exploratory MTs might act as scaffolding for further signalling, and additional MTs are recruited to the region. c | Actin has a role in determining MT localization in the growth cone. Actin arcs constrain and guide central (C) domain MTs (single arrows), and F-actin bundles inhibit and guide P domain MTs (double arrows). domains, such as the Plexin family of receptors for the guidance cue. For example, ephrin A4 can lead to RhoA semaphorins65, but many other receptors transduce activation, Rac protein inactivation or Rac activation, their activation state through separate regulators that depending on which receptors and GEFs or GAPs are are recruited by cytoplasmic domains. For example, the engaged58, and over 70 GEFs and 80 GAPs have been guidance cue ephrin B3 binding to the EphA4 recep­ described in mammals70. Many of them regulate several tor activates the regulatory RacGAP α­chimerin (also different Rho GTPases, and a particular GTPase might known as N­chimaerin) to inhibit growth cone exten­ be regulated by numerous GEFs and GAPs that all reside sion66; EphA receptors can trigger the activation of the in the same cell. How can this complex network of inter­ RhoGEF ephexin to activate RhoA but inhibit CDC42 actions be functionally explained in the growth cone? and RAC1 to induce growth cone collapse67; and the A recent profiling study of the proteome in neuroblas- guidance cues slit and netrin can both signal through toma cells suggests that Rho GTPase spatial localization the RacGEF and RhoGEF TRIo to regulate growth cone and activation might be the answer, which is a popular dynamics68,69. idea that is supported by previous studies and specula­ Neuroblastoma A potentially confusing feature of Rho GTPase sig­ tion71. This particular profiling study found that specific A tumour derived from nalling is that the possible signalling network combina­ GEFs and GAPs are differentially localized between the primitive ganglion cells that can partially differentiate into tions of GEFs, GAPs and GTPases are numerous and cell body and the axonal process72; of the 14 GEFs that cells that have the appearance complex. Multiple GTPases (that have antagonistic are expressed in neuroblastoma cells, 11 are enriched of immature neurons. functions) can be activated in response to the same in axonal processes compared with the cell body, as NATuRE REVIEWS | Molecular cell Biology VoluME 10 | MAy 2009 | 337 © 2009 Macmillan Publishers Limited. All rights reserved REVIEWS Guidance cues Netrin Slit led to increased filopodia without cell spreading, and Ephrin UNC5/DCC Robo Semaphorin and receptors Eph Plexin/L1 loss of TRIo, the GEF of RhoA and RAC1, led to long but unstable filopodia. Thus, even though RAC1 could be targeted by seven different GEFs and three different GAPs, and CDC42 could be targeted by four GEFs and Ephexin TRIO α-Chimerin SRGAP p190RhoGAP five GAPs, all of which are located in the same cell, each upstream regulator is probably required for distinct cell­ GEF GAP ular functions. These data suggest that the same GTPase Rho GTPase might control various aspects of growth cone cytoskeletal dynamics, such as F­actin assembly, disassembly and P RhoA RAC1 CDC42 Integration retrograde flow, by receiving different upstream inputs GAPs of GEFs and GAPs in time and space. Rho Rho Following activation, how do distinct Rho GTPases GTP GDP mediate downstream growth cone responses to affect (Active) (Inactive) Coordination growth cone steering? Intriguingly, activation of the same GEFs GTPase can lead to opposite responses of the growth cone: for example, whereas RhoA activation leads to GTP GDP ROCK Cytoskeletal growth cone retraction (by promoting myosin II contrac­ effectors tile activity)73, it can also be required for axon outgrowth74 MLCK LIMK (by inhibiting the actin filament­severing protein family ADF/cofilin)75. Again, as with the upstream regulators, an explanation for this discrepancy is that Rho GTPases Myosin II Cofilin Formin ENA/VASP Arp2/3 have different functions in the growth cone depend­ ing on their localization and, specifically, depending Actomyosin F-actin F-actin Effect on which downstream effector molecules are activated. contraction disassembly polymerization Numerous Rho GTPase effectors have been identi­ Figure 3 | The growth cone as a ‘navigator’. Rho family GTPases act as key navigation fied76,77, but only a few (such as Rho kinase (RoCK)) signalling nodes to integrate upstream directionalNature Reviews cues and | Molecular coordinate Cell Biology downstream have been well studied in the growth cone (see ReF. 59 cytoskeletal rearrangements. The activation of receptors by guidance cues leads to the for a review). Following activation by Rho GTPases, activation of Rho GTPase regulators. These include guanine nucleotide-exchange factors these effectors either directly or indirectly regulate (GEFs) and GTPase-activating proteins (GAPs), which activate and inactivate Rho GTPases, respectively. Rho GTPases integrate the responses of upstream pathways and numerous downstream targets to modify the cyto­ coordinate downstream effects by modifying the function of cytoskeletal effectors. skeleton to direct the growth cone vehicle in a spatially Activation or inactivation of cytoskeletal effectors leads to responses such as actomyosin biased manner. contraction, filamentous (F)-actin disassembly or F-actin polymerization. The resulting growth cone turning response depends on the localization of the guidance signalling Control of actin dynamics at the leading edge. Rho inside the growth cone. Only some of the known examples of guidance cues and GTPase cytoskeletal effectors are known to regulate receptors, GEFs, GAPs and cytoskeletal effectors that are downstream of Rho GTPases all of the aspects of the actin cycle that affect growth are shown in the figure. Arrows do not necessarily denote direct interaction. The boxed cone steering, including F­actin assembly at the periph­ inset shows the Rho GTPase activation–inactivation cycle, in which GAPs lead to the ery, F­actin retrograde flow towards the C domain, and hydrolysis of GTP to GDP, whereas GEFs catalyse the exchange of GDP for GTP. Arp2/3, disassembly and recycling of actin at the T zone. actin-related protein2/3; ENA/VASP, enabled/vasolidator-stimulated phosphoprotein; LIMK, LIM domain kinase; MLCK, myosin light chain kinase; ROCK, Rho kinase; SRGAP, Actin polymerization at the leading edge must occur slit–robo GAP; UNC5, uncoordinated protein 5. for the engine to run. This process is controlled by multi­ ple regulators, including the actin nucleators, actin­ related protein (Arp)2/3 complex and the formins, and are 6 of the 7 neuroblastoma GAPs. The authors propose the F­actin polymerization factors enabled/vasolidator­ that spatial compartmentalization of Rho GTPase regu­ stimulated phosphoprotein (ENA/VASP). The Arp2/3 lators might allow the same GTPase to be regulated by complex is a major effector of RAC1 and CDC42 and distinct GEFs or GAPs in different locations throughout is thought to control the nucleation of F­actin polymer­ the growth cone. Moreover, time­lapse microscopy after ization and F­actin branching by binding to existing individual knockdown of these GEFs or GAPs showed F­actin78. Several studies have shown that Arp2/3 is distinguishable axonal phenotypes. Whereas depletion required for guidance79,80, but whether this complex of several regulators (including the GAP of RAC1 and functions in a similar way in neuronal and non­neuronal CDC42 — ARHGAP30 — and dedicator of cytokinesis 4 cells has been questioned79. However, it was recently (DoCK4), the GEF of RAC1) led to an increase in axon shown that Arp2/3 is present in the branched F­actin extension on an ECM substrate but normal cytoskeletal networks of growth cones and does affect their protru­ structure, silencing of others led to changes in axon sion dynamics81,82. Inhibition of Arp2/3 in neurons blocks extension along with distinct and obvious perturbations protrusion of both lamellipodia­like veils and filopodia in the actin cytoskeleton72. For example, knockdown of and also increases RhoA activity 81, but future studies the GAP of CDC42, slit–robo GAP 2 (SRGAP2), resulted will be needed to determine its full role in growth cone in increased filopodia and cell spreading, knockdown motility. Downstream of Rho GTPase signalling, formins of breakpoint cluster region (BCR), the GAP of RAC1, nucleate and then remain continuously associated with 338 | MAy 2009 | VoluME 10 www.nature.com/reviews/molcellbio © 2009 Macmillan Publishers Limited. All rights reserved  REVIEWS Box 4 | Regulation of localized protein translation and degradation in the growth cone Localized translation in response to guidance cues has emerged as an important mechanism that mediates cytoskeletal dynamics, including RhoA signalling, during growth cone steering events129–131. In particular, attractive cues, such as netrins and brain-derived neurotrophic factor (BDNF), induce the asymmetric localization of mRNA and translation of cytoskeletal components, such as β-actin, on the side of the growth cone in which new filamentous (F)-actin polymerization (and thus growth cone extension) occurs132,133. Furthermore, repulsive cues, such as slit, induce the asymmetric translation of proteins that break down the cytoskeleton, such as actin-depolymerizing factor (ADF)/cofilin (a family of actin filament-severing proteins that disassemble F-actin filaments)134 as well as β-thymosin (an actin mono- mer-sequestering protein that inhibits actin polymerization)135. The microtubule (MT)–F-actin crosslinking protein short stop, which is required for growth cone steering, also binds directly to a newly discovered translation inhibitor, krasavietz (also known as extra bases)136. Furthermore, the plus-end tracking protein (+TIP) family member adenomatous polyposis coli (APC), which also binds to MT plus ends and has a role in steering, has recently been shown to be required for RNA localization to cell protrusions in non-neuronal migrating cells137 and, thus, it might function in a similar way in growth cones. In addition to cytoskeletal elements, local translation of the signalling molecule RhoA GTPase is required for the collapsing response of Xenopus laevis growth cones to the repulsive cue semaphorin 3A64, suggesting that localized translation is a common theme for many molecules involved in growth cone steering. Finally, localized protein degradation might also have a role in the regulation of growth cone dynamics. This has been shown to be true for RhoA at the leading edge in migrating fibroblasts138, and seems also to be true in neuronal cells, during their outgrowth139,140. elongating F­actin barbed ends83. Formin is required for Not only are they directly downstream of RoCK91, growth cone filopodia formation, and the Drosophila they can interact with specific growth cone receptors melanogaster formin DAAM was recently shown to (such as l1CAM)92, suggesting that they might have a act together with Rac GTPases and ENA during axonal role in crosstalk between different signalling pathways. growth regulation84. The novel actin nucleator, cordon­ Activated ERM proteins are asymmetrically localized in bleu, is also highly enriched in rat brain and might have response to guidance cues in the growth cone and pro­ a role during growth cone guidance85. mote growth cone motility in response to the guidance ENA/VASP proteins are a family of proteins that cue semaphorin 3A93, but how they function to remodel enhance F­actin elongation by several methods, includ­ actin dynamics has not been clarified. ing binding to F­actin barbed ends at the leading edge These examples are just a few of the many effectors to antagonize capping proteins (which inhibit F­actin of actin dynamics that are downstream of Rho GTPase elongation), a process that is also observed for the form­ signalling. Additionally, there are other important actin ins, and also by recruiting actin subunit complexes to the effectors that are not clearly linked to the Rho GTPases. P domain for further polymerization86. Although ENA/ The complete picture of how guidance cues affect all VASP proteins are mainly thought to be regulated by aspects of actin dynamics on a global growth cone scale protein kinases downstream of guidance cue signalling is still emerging. (such as the cyclic nucleotide­activated kinases, including protein kinase A (PKA)86), there are genetic interactions Steering and MT–F-actin interactions between ENA and TRIo in D. melanogaster 87, and the As with actin, there are many aspects of MT dynamics ENA regulator Ableson tyrosine kinase functions through that can be controlled, including nucleation, polymeri­ Rac and Rho GTPases in both neuronal and non­neuronal zation, stabilization and translocation along F­actin, all cells88,89, suggesting that there is crosstalk between Rho of which have roles in steering the growth cone vehicle GTPase signalling and ENA/VASP proteins. and which are coordinated by cytoskeletal effectors Additionally, cytoskeletal effectors that are down­ downstream of guidance signalling. However, although stream of RhoA GTPase have crucial roles in regulating numerous studies have elucidated the functions of a F­actin retrograde flow and disassembly of actin in the multitude of actin­binding proteins, fewer studies have T zone. RoCK, one of the most widely studied down­ analysed the detailed functions of MAPs. Whereas Rho stream effectors of RhoA, has multiple phosphoryl­ GTPase signalling has a central role in regulating actin ation targets that are implicated in actin dynamics in dynamics, this is not as evident for MTs, although there the growth cone, including myosin light chain kinase are several examples of MT­specific RhoGEFs in non­ (MlCK), lIM domain kinase (lIMK) and ezrin– neuronal cells, such as RHoGEF2 in D. melanogaster 94 radixin–moesin (ERM) proteins. Phosphorylation of and XlFC in Xenopus laevis 95, and RAC1 can regulate MlCK induces myosin II activity and promotes its MT dynamics in non­neuronal cells96. However, this association to F­actin, leading to actomyosin contrac­ connection has not yet been clearly shown in growth tion and driving F­actin retrograde flow 59,90. Active cones. Based on recent studies, it is clear that the ability lIMK inactivates the actin filament­severing protein of MTs to explore the growth cone periphery and enter ADF/cofilin by phosphorylation, thereby stabilizing into filopodia (an important early step that is necessary actin filaments and promoting the forward progres­ for proper steering in response to environmental cues) is sion of the vehicle75. Finally, ERM proteins are another highly dependent on how MTs interact with the F­actin group of actin­binding proteins that have an impor­ bundles and network, and thus Rho signalling at least tant but unclear role in growth cone actin dynamics. indirectly affects MT dynamics. NATuRE REVIEWS | Molecular cell Biology VoluME 10 | MAy 2009 | 339 © 2009 Macmillan Publishers Limited. All rights reserved REVIEWS A central focus of studies on MT dynamics in the example, the dynein regulator nudel binds to and inhib­ growth cone revolves around the control of MT–actin its the Rho GTPase regulator CDC42GAP in migrating interactions; specifically, the coupling and uncoupling of mouse cells104), suggesting that Rho GTPases control MTs and F­actin retrograde flow. For example, positive lIS1 and dynein­dependent uncoupling downstream of guidance cues, such as the adhesive molecule APCAM, guidance cues. Interestingly, a recent study showed that increase the frequency of MT–actin uncoupling, leading to the motor protein kinesin 5, working antagonistically increased MT exploration at sites of cue binding 35. By con­ with dynein, has a key role in controlling MT extension trast, repellent cues might increase coupling between MTs into the P domain during growth cone turning and is and actin, and thus reduce MT exploration and increase specifically phosphorylated on the side opposite the MT ‘looping’, because the growing plus ends of MTs are invasion of MTs before turning 105, further demonstrat­ transported back towards the C domain by their linkage to ing that motor proteins are likely targets of guidance cue F­actin retrograde flow, thereby forming MT loops in the signalling. growth cone. Thus, one function of the growth cone navi­ Similar to lIS1, APC is another +TIP that promotes gation system is to locally control MT–actin interactions MT–actin uncoupling in the growth cone and also in response to asymmetric guidance cues. The detailed directly interacts with IQGAP1 (ReF. 106). APC binds to signalling pathways by which this occurs are still unclear, a subset of MTs in the growth cone and its binding loca­ but it is obvious that components of the navigation system tion indicates the future growth direction of the axon. can control MT dynamics by directly or indirectly modu­ local enhancement of APC association with MTs leads lating the activity of MAPs, and, in particular, those that to growth cone steering towards that axis107, possibly function as MT–actin crosslinking proteins. by preventing MT binding to F­actin. This speculation is supported by a recent study that shows that APC loss Revealing the roles of MAPs in growth cone dynamics. from MT plus ends leads to increased MT looping and one group of MAPs that has recently been shown to the prevention of growth cone progression108. APC loss have a crucial role in growth cone dynamics is the family occurs in response to the morphogen WNT3A, which of plus­end tracking proteins (+TIPs; including end­ acts through the intracellular Wnt signalling pathway binding proteins (EBs) and adenomatous polyposis member dishevelled 1 and can induce axonal remodel­ coli (APC)), which bind specifically to the plus ends of ling in mouse dorsal root ganglia neurons. In this case, the MTs97,98. These proteins were originally implicated in Wnt signalling pathway might be regulating APC–MT affecting the stabilization of MT plus ends, but they also association by modifying the APC phosphorylation state. mediate MT crosslinking to actin, either in a positive or When APC is lost from the MT plus ends, this might allow negative way. Interestingly, the +TIP EB1 (also known other MT–actin crosslinkers to bind to the plus ends and as MAPRE1) is required for RHoGEF2 association with couple MTs to F­actin retrograde flow, leading to MT MT plus ends in non­neuronal cells94, which suggests the looping. For example, the +TIP member ClASP (cyto­ intriguing possibility that navigator signalling molecules, plasmic linker protein­associated protein), promotes MT such as Rho GTPase regulators, might actually harness looping in the growth cone when overexpressed109 and can MT dynamics to move themselves into areas of guidance link MT ends to actin in non­neuronal cells110, although cue signalling 99. This could be one explanation for how its actin­binding activity in the growth cone has not been exploratory MTs are required early for steering. examined. If this function holds true in growth cones, Evidence suggests that at least two +TIP members, ClASP might be a +TIP member that drives MT–actin lissencephaly 1 (lIS1; also known as PAFAH1B1) and coupling and MT looping behaviour downstream of envi­ APC, promote MT–actin uncoupling in the growth ronmental ‘stop’ signals. Interestingly, glycogen synthase cone. In particular, lIS1 cooperates with the MT motor kinase 3β (GSK3β), which is downstream of a number protein dynein to allow the uncoupling of dynamic MTs of guidance cues, including Wnt, netrin and semaphorin from actin retrograde flow in chick and rat neurons signalling 111, can regulate whether ClASP binds to the plated on a laminin substrate100,101. Inhibition of dynein MT plus ends or along the entire MT, and this change in and lIS1 prevents MT advance into the periphery and localization might regulate its function112. reduces the ability of MTs to resist F­actin retrograde MAP1B, which is considered to be a scaffold protein flow 101. Consequently, the growth cone cannot accurately that stabilizes MTs but can also bind actin, also regulates steer along a laminin border 100. This is further confirmed growth cone steering and motility of mouse neurons by by the blockage of retrograde flow using the myosin II controlling MT–actin dynamics113. In particular, the role inhibitor blebbistatin, which leads to the recovery of MT of MAP1B in growth cones might be to couple MTs and extension into the P domain in dynein­depleted growth F­actin in response to guidance cues, as MAP1B function cones and demonstrates that actin retrograde flow can is required for the MT looping that occurs in response prevent MT exploration100. Whereas the pathways by to treatment with lysophosphatidic acid (lPA; a phos­ which guidance cues lead to lIS1 and dynein activity pholipid derivative that triggers axonal process retrac­ in the growth cone have not been elucidated, lIS1 can tion and growth cone collapse)113. As MAP1B function specifically interact with the RAC1 and CDC42 regu­ can be modulated by phosphorylation111, this suggests a lator IQGAP1 in migrating mouse neurons102, and lIS1 mechanism by which the ability of MAP1B to couple MT deficiency is associated with dysregulation of CDC42, and actin depends on upstream guidance cue signalling. RAC1 and RhoA103. Additionally, connections exist For example, MAP1B is a phosphorylation substrate of between dynein and Rho GTPases in other systems (for GSK3β114, as is ClASP. 340 | MAy 2009 | VoluME 10 www.nature.com/reviews/molcellbio © 2009 Macmillan Publishers Limited. All rights reserved  REVIEWS Whereas coupling MTs to F­actin retrograde flow Whereas much work has investigated how indi­ prevents MT extension to the periphery in these cases, vidual cues are interpreted, the next target will be to interactions between MTs and F­actin can also promote understand how the growth cone interprets multiple MT extension. For example, a recent study showed that overlapping gradients of cues. Previously, it has been the F­actin­associated protein drebrin binds directly difficult to examine complex effects on the growth to the +TIP protein EB3 in embryonic rat growth cones, cone cytoskeletal machinery, in part owing to techni­ an interaction that occurs specifically in the proximal cal limitations of experimental assays. Recent work region of filopodia when EB3­bound MTs enter and align that uses microfluidic devices has allowed the produc­ alongside drebrin­bound F­actin bundles115. Strikingly, tion of stable, precisely controlled gradients in various loss of drebrin function prevents MT extension into combinations of both diffusible and substrate­bound filopodia, suggesting that the drebrin–EB3 interaction is factors116,117. These types of combined cues can more important for allowing MT exploration of filopodia115. faithfully recapitulate the complexity of the in vivo These recent studies that examine the growth cone environment and, thus, future experiments that use functions of MT–actin regulatory proteins are begin­ this method, in combination with high­resolution cyto­ ning to uncover the complete and detailed mechanisms skeletal imaging, hold considerable potential for refin­ by which environmental cues are translated to changes ing our understanding of the growth cone guidance in the cytoskeleton, and, especially, they seek to answer mechanism. the intriguing question of how MTs interact function­ In the past decade, there have been great advances in ally with the actin network during growth cone steering. high­resolution imaging methods that can be used for However, there is still much to understand of the role of analysing growth cone cytoskeletal dynamics, includ­ MTs and MT–actin interactions in the growth cone, and ing fluorescent speckle microscopy 118 and total internal this is a promising area of growth cone guidance research reflection fluorescence microscopy 119. Combinations of for the future. these microscopy methods with new cross­correlation speckle­tracking algorithms for quantitatively measur­ Conclusions and future perspectives ing the details of cytoskeletal dynamics have provided like an experienced driver, the growth cone navigation further insights into the roles of actin and MTs dur­ system can integrate multiple environmental variables, ing normal vehicular function of the growth cone36,120, including road conditions, stop lights and street signs, including the finding that exploratory MTs might have to direct the vehicle and reliably reach its destination. a role in early signalling from guidance cues, whereas Navigation system signalling through Rho GTPases and actin dynamics guide and control more stable MTs to fix downstream cytoskeletal effectors is superimposed on the the direction of new growth. The same types of quantita­ cytoskeletal mechanisms of the vehicle, including F­actin tive methodology will need to be applied to understand retrograde flow and F­actin guidance of MTs, to introduce protein–protein interactions and catalytic activations spatial bias for steering the growth cone in the right direc­ of signalling molecules, through advanced fluores­ tion. understanding how multiple cytoskeletal effectors cent sensors and tags121–123. This is an exciting time for work in concert to achieve this, in combination with the studying growth cone dynamics, with new techniques elucidation of additional signalling pathways that mediate and microscopy tools that are now providing further growth cone navigation, will give an integrated picture for opportunities to explore the outstanding questions the logic of growth cone dynamics over time. of the growth cone vehicle and its navigation. 1. Maness, P. F. & Schachner, M. Neural recognition 10. Mattson, M. P., Dou, P. & Kater, S. B. Outgrowth- neuronal growth cones. Nature Cell Biol. 8, 215–226 molecules of the immunoglobulin superfamily: regulating actions of glutamate in isolated (2006). signaling transducers of axon guidance and hippocampal pyramidal neurons. J. Neurosci. 8, Provides strong evidence that growth cone F‑actin neuronal migration. Nature Neurosci. 10, 19–26 2087–2100 (1988). retrograde flow is driven both by contractility of (2007). 11. Tojima, T. et al. Attractive axon guidance involves the motor protein myosin II in the T zone, and 2. Evans, A. R. et al. Laminin and fibronectin modulate asymmetric membrane transport and exocytosis in the F‑actin bundle treadmilling in the P‑domain. inner ear spiral ganglion neurite outgrowth in an growth cone. Nature Neurosci. 10, 58–66 (2007). Compression across the T zone circumference that in vitro alternate choice assay. Dev. Neurobiol. 67, 12. Bonanomi, D. et al. Identification of a is driven by myosin II leads to buckling and 1721–1730 (2007). developmentally regulated pathway of membrane severing of F‑actin bundles near the proximal 3. Dickson, B. J. Molecular mechanisms of axon retrieval in neuronal growth cones. J. Cell Sci. 121, ends. guidance. Science 298, 1959–1964 (2002). 3757–3769 (2008). 18. Sarmiere, P. D. & Bamburg, J. R. Regulation of the 4. Chilton, J. K. Molecular mechanisms of axon guidance. 13. Goldberg, D. J. & Burmeister, D. W. Stages in axon neuronal actin cytoskeleton by ADF/cofilin. Dev. Biol. 292, 13–24 (2006). formation: observations of growth of Aplysia axons in J. Neurobiol. 58, 103–117 (2004). 5. Zou, Y. & Lyuksyutova, A. I. Morphogens as conserved culture using video-enhanced contrast-differential 19. Haviv, L., Gillo, D., Backouche, F. & Bernheim- axon guidance cues. Curr. Opin. Neurobiol. 17, 22–28 interference contrast microscopy. J. Cell Biol. 103, Groswasser, A. A cytoskeletal demolition worker: (2007). 1921–1931 (1986). myosin II acts as an actin depolymerization agent. 6. Brunet, I. et al. The transcription factor Engrailed-2 14. Dent, E. W. & Gertler, F. B. Cytoskeletal dynamics and J. Mol. Biol. 375, 325–330 (2008). guides retinal axons. Nature 438, 94–98 (2005). transport in growth cone motility and axon guidance. 20. Zicha, D. et al. Rapid actin transport during cell 7. Butler, S. J. & Tear, G. Getting axons onto the right Neuron 40, 209–227 (2003). protrusion. Science 300, 142–145 (2003). path: the role of transcription factors in axon 15. Marsh, L. & Letourneau, P. C. Growth of neurites 21. Mitchison, T. & Kirschner, M. Cytoskeletal dynamics guidance. Development 134, 439–448 (2007). without filopodial or lamellipodial activity in the and nerve growth. Neuron 1, 761–772 (1988). 8. Gundersen, R. W. & Barrett, J. N. Neuronal presence of cytochalasin B. J. Cell Biol. 99, 22. Lin, C. H., Thompson, C. A. & Forscher, P. chemotaxis: chick dorsal-root axons turn toward high 2041–2047 (1984). Cytoskeletal reorganization underlying growth cone concentrations of nerve growth factor. Science 206, 16. Suter, D. M. & Forscher, P. Substrate–cytoskeletal motility. Curr. Opin. Neurobiol. 4, 640–647 1079–1080 (1979). coupling as a mechanism for the regulation of growth (1994). 9. Sanford, S. D., Gatlin, J. C., Hokfelt, T. & cone motility and guidance. J. Neurobiol. 44, 97–113 23. Jay, D. G. The clutch hypothesis revisited: ascribing Pfenninger, K. H. Growth cone responses to growth (2000). the roles of actin-associated proteins in filopodial and chemotropic factors. Eur. J. Neurosci. 28, 17. Medeiros, N. A., Burnette, D. T. & Forscher, P. protrusion in the nerve growth cone. J. Neurobiol. 44, 268–278 (2008). Myosin II functions in actin-bundle turnover in 114–125 (2000). NATuRE REVIEWS | Molecular cell Biology VoluME 10 | MAy 2009 | 341 © 2009 Macmillan Publishers Limited. All rights reserved REVIEWS 24. Letourneau, P. C. Cell–substratum adhesion of neurite 45. Zhou, F. Q. & Cohan, C. S. How actin filaments and 69. Bateman, J. & Van Vactor, D. The Trio family of growth cones, and its role in neurite elongation. Exp. microtubules steer growth cones to their targets. guanine-nucleotide-exchange factors: regulators of Cell Res. 124, 127–138 (1979). J. Neurobiol. 58, 84–91 (2004). axon guidance. J. Cell Sci. 114, 1973–1980 (2001). 25. Bridgman, P. C., Dave, S., Asnes, C. F., Tullio, A. N. & 46. Rodriguez, O. C. et al. Conserved microtubule–actin 70. Heasman, S. J. & Ridley, A. J. Mammalian Rho Adelstein, R. S. Myosin IIB is required for growth cone interactions in cell movement and morphogenesis. GTPases: new insights into their functions from in vivo motility. J. Neurosci. 21, 6159–6169 (2001). Nature Cell Biol. 5, 599–609 (2003). studies. Nature Rev. Mol. Cell Biol. 9, 690–701 26. Matti