Filopodia: The Fingers That Do the Walking

See allHide authors and affiliations

Science's STKE  21 Aug 2007:
Vol. 2007, Issue 400, pp. re5
DOI: 10.1126/stke.4002007re5


Filopodia are actin-based structures composed of parallel bundles of actin filaments and various actin-associated proteins, and they play important roles in cell-cell signaling, guidance toward chemoattractants, and adhesion to the extracellular matrix. Two mechanisms for the formation of filopodia have been suggested, each using different sets of actin-regulating proteins, creating some controversy in the field. New molecules, some of unknown functions, have also been implicated in filopodium formation, suggesting that other possible mechanisms of filopodium formation exist. We discuss established and novel proteins that mediate the formation and dynamics of filopodia, different mechanisms of filopodium formation, and the various functions that distinct filopodia perform.


Both polymerization and movement of the actin cytoskeleton are elaborately regulated by an expanding number of proteins that modulate polymerization, capping, severing, branching, bundling, contraction, and adhesion to the extracellular substrate. The actin-associated proteins present and active in a cell determine the types of filamentous actin (F-actin)–based structures formed, including lamellipodia, filopodia, stress fibers, and podosomes (1). Many of these structures have been the focus of attention of the cell-migration community because of their involvement in leading-edge protrusion, cell body contraction, and adhesion. With such a vast array of proteins that modulate cytoskeletal dynamics, it is not surprising that many actin-based molecular machines exist. One of the best-described structures is the lamellipodium: a thin, densely branched meshwork of F-actin behind the protruding cell edge with fast-growing "barbed" filament ends oriented toward the direction of protrusion and migration (2). The functions and interactions of many of the molecular players necessary for the formation of this structure, such as the Arp2/3 (actin-related protein 2 and 3) complex and cofilin, have been elucidated (35).

The filopodium is another protruding F-actin–based structure composed of parallel bundles of actin filaments that extend in a fingerlike manner beyond the cell edge (6). Similar to the lamellipodium, the barbed ends of the actin filaments within the filopodium are oriented outward (6). Filopodial protrusions and retractions are mediated by a dynamic balance of barbed-end polymerization and F-actin retrograde flow (7). Therefore, regulation of F-actin polymerization at the filopodial tip and F-actin retrograde flow by contraction and adhesion control the formation, maintenance, and dynamics of filopodia. There are a number of established regulators of filopodium formation and dynamics (8), and several distinct mechanisms of filopodium formation have been proposed that involve different molecules, prompting some controversy in the field (911). Several new molecules implicated in filopodium formation and dynamics have been described; however, their functions in filopodium formation are not yet well understood (1113). This review focuses on established and novel molecules involved in filopodium formation, describes how these molecules are implicated in distinct mechanisms of filopodium initiation, and discusses contradictory findings in the field.

Filopodium Function

Filopodia participate in several fundamental physiological processes, such as wound healing, development, and cell signaling. Filopodia function by sensing chemicals in the environment (14, 15), acting as sites of signal transduction, and attaching to the extracellular matrix (ECM). Contained within filopodia are receptors that detect diverse signals in the extracellular milieu (16) and initiate downstream signaling pathways leading to cellular responses. For example, filopodia can initiate the response to guidance cues, such as netrin, in neuronal growth cones, which are motile actin- and microtubule-rich structures at the ends of neurites that guide axons and dendrites to their proper targets (17, 18). Growth cone filopodia orient toward a gradient of guidance cues, an event that precedes the extension, turning, or branching of the growth cone–tipped neurite toward chemoattractants (17, 19, 20). Some studies suggest, however, that filopodia are not required for all types of guidance responses (21, 22), indicating specificity in filopodia function. Dissociated primary neurons on glass cover slips undergo a standard progression of morphologies to reach their mature state with one axon and several dendrites (23). Filopodia appear to be essential for neurite initiation, when the precursors for axons and dendrites first form and extend from the neuronal cell body (24, 25). This suggests that filopodia function to establish neural circuitry during development. Similarly, filopodia are central in the proper alignment and attachment of cells during the dorsal closure of Drosophila embryos (2628). Filopodia help cells to correctly align and adhere to their target partner and close the gap between them, a process known as "adhesion zippering" (28, 29), which also occurs in epithelial sheet fusion in Caenorhabditis elegans (30) and in cultured epithelial cells (31). Interestingly, certain molecules involved with adhesion, such as integrins or cadherins, have long been known to associate with filopodial tips (32) and along filopodial shafts (33). Some filopodial tips even contain activated, yet unligated, β1 integrins (34), which suggests that filopodia sense permissive substrates composed of ligands that allow adhesion, motility, or both. Filopodia act as points of attachment to the substrate and are also thought to produce tension across the ECM (35, 36), which could be used in guidance and migration. These diverse attributes make filopodia important factors in directed cell migration during development and wound healing (2628).

Filopodium Structure and Dynamics

A classic electron microscopy (EM) study carried out more than 30 years ago revealed a dense aggregation at the termini of microvilli (37), which have since been seen by EM in filopodia (10, 33, 38, 39). These protein aggregates suggest the existence of a filopodium tip complex, presumed to either alter filopodial dynamics, or tether F-actin in filopodia to the cortical actin cytoskeleton in the plasma membrane, or both. Although this idea has persisted, the exact molecular composition of this complex is still largely unknown; so, too, is whether different filopodia have distinct tip complexes. Indeed, EM of filopodia from various cellular systems suggests that there are distinct architectures, depending on the cell type. Platinum-replica transmission EM of B16F1 melanoma cells revealed that F-actin within the lamellipodium meshwork becomes bundled and extends along the entire length of a filopodium to its tip (10). In contrast, cryo-electron tomography of vitrified samples showed that the ultrastructure and architecture of the relatively fast-growing filopodia of Dictyostelium cells are quite different from the slower-growing filopodia of mammalian cells (40).

Filopodia in Dictyostelium contain ~10 short filaments that converge into a "terminal cone" within the filopodium tip that are not continuous with the parallel bundles of F-actin within the filopodium shaft. Interestingly, no actin filaments extend the complete length of the filopodia in these cells (40). Whether this terminal cone of F-actin exists in other cell types but has not been observed because of harsher fixation conditions is unclear. The authors suggest that this arrangement of F-actin within the filopodium is best for the force production needed in fast-growing filopodia. Estimating and comparing filopodia protrusion rates from Dictyostelium [~1.0 μm/s (40) or ~0.03 μm/s (41)] with those of B16F1 melanoma tissue culture cells [~0.1 μm/s (10)], neuroblastoma cells [~0.04 to 0.1 μm/s (7)], and primary hippocampal neurons [<0.1 μm/s (42)] indicate that filopodial dynamics differ by orders of magnitude among different cell types. Mathematical models predict that the factors that enhance the protrusion speed include thermal fluctuations of the membrane, the number and spatial arrangement of the filaments in the bundle, and whether the filaments are tethered to the membrane (43). Considering the diversity in the architecture and dynamics of filopodia, it is not surprising that distinct sets of molecules are implicated in filopodium formation. Thus, diverse sets of actin-associated proteins may be associated with filopodial tip complexes of differing structural and dynamic characteristics, and possibly functions that are suited to the needs of particular cell types. It will be interesting to determine those sets of proteins that are associated with particular attributes and functions.

Filopodium Initiation and Maintenance

Mechanisms of filopodium initiation

To date, two mechanisms that use distinct actin-nucleating proteins have been proposed for filopodium initiation (Fig. 1). In the "convergent elongation" model (Fig. 1A) (10), filopodia are initiated by established F-actin regulators, including Arp2/3 complex–mediated nucleation downstream of the signaling of the Rho guanosine triphosphatase (GTPase) Cdc42, Enabled (Ena)/VASP (vasodilator-stimulated phosphoprotein) at filopodial tips, and fascin along filopodia shafts, whereas the opposing model proposes that Dia2 performs most, if not all, of these functions.

Fig. 1.

Mechanisms of filopodium initiation. (A) Convergent elongation model of filopodium initiation involves Arp2/3-mediated F-actin branch formation, the F-actin–bundling activity of fascin, and the activities of Ena/VASP at barbed ends at the filopodial tip as an anticapper, by protecting F-actin from capping proteins and promoting polymerization. These filopodia are anchored into the F-actin dendritic meshwork. (B) De novo filament elongation is likely mediated by an F-actin nucleator and leaky capper, such as Dia2. Dia2 protects the barbed ends of F-actin from capping proteins, promotes bundling, and increases polymerization. F-actin bundling in these filopodia may also be modulated by fascin or other F-actin–bundling proteins, such as filamin, although this is currently unknown. These filopodia may also be anchored to the plasma membrane or cortical F-actin cytoskeleton through such proteins as IRSp53 or LPR1. (C) The reorientation and elongation of peripheral F-actin bundles in neuronal growth cones could also induce filopodium initiation, modulated by several regulators of actin, including Ena/VASP proteins or Dia2 at the barbed ends, and fascin, filamin, or other F-actin crosslinkers along the shafts of filopodia. These filopodia are anchored in the peripheral actin bundles in the growth cone. Additionally, it is unclear whether these three mechanisms are independent and exclusive, or whether multiple mechanisms can be operating within the same cell.

In the convergent elongation model of filopodium initiation, filopodia form from a gradual association of a subset of branched lamellipodial filaments at their barbed ends that are decorated with Ena/VASP proteins. Association of Ena/VASP proteins with the barbed ends of filaments may mark F-actin for subsequent filopodium formation, presumably by clustering barbed ends together, protecting them from capping, and permitting rapid polymerization (10), and possibly bundling filaments (4446). The binding of fascin converts the filaments into bundled filopodia and stabilizes them. After filopodium formation in B16F1 melanoma cells, microtubules extend toward the proximal termini of filopodia. These events correlate with filopodial turning and merging (47). Thus, microtubules may mediate the lateral movement of filopodia, their merging, and thus, their density.

Because many of the players involved in the convergent elongation model have been described in great detail, we will discuss these first. Cdc42 directly interacts with and activates Wiskott-Aldrich syndrome protein (WASP) family proteins, which, in turn, activate the Arp2/3 complex (48, 49). The Arp2/3 complex subsequently binds to the side of an actin filament and nucleates a new filament as a branch from the mother filament. Although it seems counterintuitive that a complex that nucleates filament branches should be involved in the formation of parallel bundles of F-actin, several studies have suggested that filopodia arise from the branched F-actin meshwork of the lamellipodium (10), as opposed to arising from de novo filament nucleation. Evidence for this model has been provided by elegant time-lapse fluorescence microscopy and EM (10); however, genetic data supporting this model are lacking. Interestingly, studies that have suggested that filopodia form in the absence of Arp2/3 (9, 50, 51), and its activator WASP/N-WASP (52, 53), bring into question the role of Arp2/3 and convergent elongation.

The Ena/VASP family of proteins also has a well-established role in the formation and maintenance of filopodia, though the precise nature of EnaVASP function remains controversial (42, 46, 54, 55). These proteins concentrate along the leading edge (56, 57) and in the tips of filopodia (58), both of which are characterized by high actin dynamics. Ena/VASP proteins bind both monomeric actin (G-actin) and F-actin (44, 45, 59). Ena/VASP proteins also bundle F-actin (45, 46, 59, 60) and may cluster barbed ends (54). Studies using mouse embryonic fibroblasts MVD7 cells (61), which are genetically deficient for Mena and VASP and have only trace amounts of the Ena/VASP family member Evl, have revealed new information about this protein family. MVD7 cells show impaired regulation of lamellipodium and filopodium formation, but when exogenously expressed at physiological levels, these functions are restored. Ena/VASP family members stably localize at filopodial tips. Fluorescence recovery after photobleaching (FRAP) studies reveal that these proteins have long residence times at filopodial tips and that this localization is lost in G-actin domain mutants (54), suggesting that the binding of G-actin is necessary for barbed-end association. Ena/VASP proteins can also prevent filament capping by capping protein, while permitting polymerization (59, 62), although this issue remains controversial (46, 63). Although the molecular mechanism of Ena/VASP activity is unclear owing to the use of different experimental conditions in various studies, it is clear that Ena/VASP activity increases the number, length, and growth rates of filopodia in many cell types (42, 55, 64, 65) and the formation of neurites subsequent to filopodium formation in primary neurons (42). Deletion of the F-actin binding motif of the one Ena/VASP ortholog in Dictyostelium blocks filopodium formation. Although this was interpreted as indicative of an exclusive role for bundling in filopodium formation (46), the same motif is required for F-actin binding and the prevention of capping (59). More sophisticated methods will be needed to resolve whether Ena/VASP’s role in filopodium formation is anticapping, F-actin binding, or bundling, or some combination of these. Interestingly, another study has suggested that filopodia form in the absence of Ena/VASP (66) (Fig. 1B), although this needs to be confirmed in a true genetic knockout. In light of this and previously mentioned studies, the convergent elongation model is not sufficient to explain the formation of all filopodia.

The final component of the convergent elongation model is the actin-bundling protein, fascin. Because filopodia are parallel, bundled arrays of F-actin, it is not surprising that fascin, an F-actin–crosslinking protein, is associated with filopodia in many cell types (6769). Although there are many other proteins that crosslink and bundle F-actin, such as fimbrin, filamin, and α-actinin, RNA interference (RNAi) studies in B16F1 mouse melanoma cells have shown that knockdown of fascin inhibits the formation of filopodia and resolvable F-actin bundles along the cell edge, suggesting that fascin is critical for filopodium formation (70). It will be interesting to determine whether other actin-crosslinking proteins form filopodia or whether fascin is always required.

Another mechanism for the formation of filopodia involves the Diaphanous-related formin, Dia2 (50). In vitro Dia2 nucleates linear actin filaments, accelerates actin polymerization (71, 72), slows filament depolymerization (73), protects barbed ends from capping proteins (74, 75), and bundles F-actin (76). Interestingly, all of these actin-modulating activities also occur in the convergent elongation model. Overexpression of Dia2 induces the formation of filopodia in a range of cell types from Dictyostelium to NIH3T3 cells (74, 77, 78). Importantly, knockout studies showed that Dia2 is essential for filopodium formation in Dictyostelium (74). In the proposed mechanism of Dia2-mediated filopodium formation, filopodia arise as a result of de novo filament nucleation, although whether this occurs in vivo is unclear. It will be interesting to determine which functions Dia2 performs during this alternative mechanism of filopodium formation. In addition, in Dictyostelium it appears that both Dia2 and the single VASP ortholog cooperate during filopodium formation (46), which raises the question of whether these two mechanisms of filopodium initiation are independent, or whether Dictyostelium form filopodia by yet another mechanism.

Signaling in filopodium initiation

The small Rho GTPases Rac, Rho, and Cdc42 are established regulators of the actin cytoskeleton. Whereas Rac activity promotes the protrusion of the lamellipodium, Rho is implicated in the formation and contraction of contractile actin structures, and Cdc42 has long been suggested to control the formation and protrusion of the filopodium (7981). It is still unclear whether Cdc42 is necessary for filopodium formation, as is the mechanism by which it acts, because experiments implicating Cdc42 in filopodium formation have mostly involved overexpression of constitutively active or dominant negative mutants of Cdc42 (7981). Filopodia still form in cells lacking Cdc42, as well as cells lacking its effector WASP (52, 53, 82). Other cell types lacking Cdc42 are devoid of filopodia (79). Determining whether Cdc42 activation induces filopodium formation in the absence of WASP activity will be of great interest.

The small Rho GTPase Rif (Rho in filopodia) is also implicated in filopodium formation. Similar to Cdc42, overexpression of Rif induces the formation of filopodia, but these filopodia are long and project from the apical cell surface and the cell periphery (50, 83), whereas Cdc42-induced filopodia are shorter and do not project from the apical cell surface (81, 84). Rif binds to and activates Dia2, and filopodia formed through Rif and Dia2 are independent of Cdc42, WASP, and Arp2/3 activities (50). This further supports the hypothesis that alternative mechanisms, in addition to convergent elongation, exist. Because Cdc42 can also bind to and activate Dia2 (77), it is unclear whether activation of Dia2 by Cdc42 or Rif is the more physiologically relevant mechanism. It will be interesting to determine if filopodium induction by Dia2 is typically associated with activation by Cdc42 or Rif, whether its filopodium-inducing activity is always independent of Arp2/3, and whether fascin is involved in formation or maintanence of these filopodia. It is also unclear whether there are multiple possible mechanisms downstream of these signals that lead to filopodium formation.

Other studies have found additional small Rho GTPases that induce filopodium formation. For example, in fibroblasts, exogenously expressed TC10, another small Rho GTPase, induces the formation of filopodia that are longer than typical Cdc42-induced filopodia, whereas RhoT produces even longer and thicker filopodia (85). However, in porcine aortic endothelial cells, exogenous expression of constitutively active TC10 results in the formation of focal adhesion-like structures and not filopodium-like structures, and Wrch-1 expression induces long and thin filopodia. This study also showed that cells expressing constitutively active RhoD and Rif have extremely long and flexible filopodia, but the same is not true for cells that express RhoT (86). Clearly, more studies on these GTPases are warranted. Although there are multiple Rho GTPases that induce filopodium formation when constitutively active, it is unclear which of these GTPases play a physiological role in filopodium formation, and whether they act upstream of several different mechanisms or converge on similar mechanisms of filopodium formation.

New Players in Filopodium Initiation

Myosin X

Although many of the proteins mentioned above induce the formation of filopodia in relatively well-characterized mechanisms, new players have been discovered that induce filopodium formation by unknown mechanisms. For example, myosin X is an actin-associated protein implicated in filopodium formation and function that localizes to the tips of filopodia. Overexpression of full-length myosin X increases the numbers of both substrate-bound filopodia and dorsal filopodia in various cell types (87). Myosin X moves up and down filopodia shafts, an activity known as intrafilopodial motility (87, 88). Myosin X moves forward by "walking" along actin filaments toward the barbed ends, whereas its rearward movement is due to the retrograde flow of F-actin to which myosin X is tightly bound. Myosin X may therefore shuttle molecules within filopodia. Indeed, it has several interesting binding partners that could affect filopodium formation. Myosin X may transport Ena/VASP proteins to the tips of filopodia (89) to increase F-actin polymerization, clustering, and protection from capping. However, myosin X can induce filopodia independently of Ena/VASP in MVD7 cells (66). Myosin X also binds to β integrins and may transport these adhesive molecules to tips of filopodia, possibly inducing filopodium formation by increasing adhesion to the substrate (90). Myosin X also binds microtubules (91), and therefore may crosslink microtubules to F-actin, one possible mechanism mediating the targeting of microtubules to the proximal ends of filopodia discussed previously (47). It will be useful to determine which of these binding partners are involved in myosin X–mediated filopodium induction.


Another player in the formation of the filopodium is Ror2, a receptor tyrosine kinase activated downstream of Wnt5a, which is expressed in neural crest–derived and mesenchymal cells during embryogenesis (92) and is essential for morphogenesis (93, 94). The C. elegans ortholog, CAM-1, is also involved in cell migration and axonal elongation (95). When expressed ectopically in various cell types, Ror2 stimulates the formation of an increased number of filopodia and colocalizes with F-actin in filopodia (13). This activity is not dependent on the tyrosine kinase activity of Ror2, but rather is dependent on its association with the F-actin–crosslinking protein, filamin A, which is found at the root of a filopodium, and presumably Wnt5a signaling. Indeed, the association of filamin A and Ror2, and possibly the downstream induction of filopodia, are necessary for Wnt5a-dependent cell migration, suggesting that filopodia play an important role in cell migration in some cell types and, further, that filopodia can act downstream of a number of signaling pathways. It will be interesting to determine whether Ror2-induced filopodia are independent of fascin-mediated F-actin bundling, owing to the mandatory interaction of Ror2 with filamin to induce these filopodia. This would suggest that fascin is not the only F-actin–bundling or F-actin–crosslinking protein able to mediate the formation of filopodia.


Another initiator of filopodium formation is the integral membrane protein lipid phosphatase–related protein 1 (LPR1), which induces filopodium formation in an unknown manner but is independent of Cdc42, Rif, Arp2/3, and Ena/VASP (MVD7 cells). LPRs typically either dephosphorylate lipid phosphates or catalyze transphosphatidylation reactions; however, LPR1 is catalytically inactive (11). LPR1 localizes to internal membranes (likely the endoplasmic reticulum and Golgi) and to the plasma membrane, particularly along the lengths of filopodia (11), and its overexpression in both HeLa and Cos-7 cells is associated with a large increase in the number of peripheral and dorsal filopodia. Although the typical filopodial markers mentioned above are absent from LPR1-induced filopodia, myosin X localizes to the tips and fascin is found along the lengths of the filopodia, suggesting that there may be many permutations of molecules present in filopodial tip complexes. LPR1-induced filopodia appear to be more persistent, longer, thinner, and more motile than those induced by Cdc42 and occur on peripheral and dorsal surfaces, further suggesting that different mechanisms of filopodium formation likely produce filopodia with different attributes and functions. Perhaps LPR1 anchors the cytoskeletal architecture within filopodia to the plasma membrane or cortical actin cytoskeleton? Further characterization of the biochemical characteristics of LPR1 will be needed to reveal possible mechanisms of its involvement in filopodium formation, as well as whether LPR1 induction of filopodia requires the presence of myosin X, fascin, or other filopodium-associated proteins.


The insulin receptor tyrosine kinase substrate p53 (IRSp53) is also implicated in the formation of filopodia, although there is some disagreement about its possible mechanisms of action (96, 97). IRSp53 contains an IMD [a lipid binding domain found in both IRSp53 and missing-in-metastasis (MIM)] at its N terminus that has several possible functions including Rac binding, (98), F-actin bundling (99), and lipid binding (12). IRSp53 also contains a central Src-homology 3 (SH3) domain through which it binds WAVE2 (WASP-family protein member 2, and a C-terminal WASP homology domain-2 (WH2) that binds with high affinity to G-actin (100). IRSp53 also binds to Rac and Cdc42 through its N terminus (80). One possible mechanism for the induction of a filopodium by IRSp53 is through downstream activation of Arp2/3 actin-nucleating activity by the interaction of IRSp53 with the WAVE2 complex (96), which may initiate a convergent elongation mechanism of filopodia formation. Alternatively, another study suggested that the IMD domain of IRSp53 induces F-actin bundling in vitro and in tissue culture cells (99). The induction of filopodium formation by Cdc42 in Swiss 3T3 fibroblasts is dependent on the IMD region of IRSp53 (80), and induction is enhanced in the presence of Ena/VASP, possibly by IMD-mediated binding of IRSp53 to Ena/VASP. Cdc42-induced filopodium formation in HeLa cells requires both IRSp53 and Epsin-8 (Eps8). Eps8 is also known to regulate actin dynamics by binding to the barbed ends of F-actin and crosslinking F-actin. The interaction of IRSp53 with Eps8 synergistically increases the F-actin bundling of both proteins in vitro (99). These studies suggest that IRSp53 functions in filopodium formation through its F-actin–bundling activity, which is mediated by the IMD.

However, a recent study questions the relevance of IMD-induced F-actin bundling to filopodium formation and provides an alternative role for this domain. The IMD also interacts with phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2] and is structurally similar to the lipid-binding BAR (Bin/amphiphysin/Rvs) domain of other proteins that typically invaginate the plasma membrane (101). Because the lipid-binding interface of the IMD of IRSp53 and BAR domains display opposite curvatures, it is not surprising that the IMD forms membrane tubes instead of invaginations (12). Indeed, this study showed that IMD-mediated induction of filopodia in U2OS osteosarcoma tissue culture cells does not depend on its F-actin–bundling activity, but rather depends solely on its membrane-curving function (12), suggesting that F-actin bundling may not be the relevant mechanism for IRSp53-induced filopodium formation. Careful measurement by EM of the ultrastructure of filopodia induced by IRSp53 may distinguish between the possible mechanisms, because membrane tubes induced by the IMD have a characteristic width. It will be interesting to clarify which of the different domains and binding partners of IRSp53 are necessary for its induction of filopodium formation, and whether this protein can act through several distinct mechanisms. The interactions between IRSp53 and its binding partners could also link F-actin within filopodia to the plasma membrane and cortical cytoskeleton.

Alternative Mechanisms of Filopodium Formation

To summarize, recent progress in this field has uncovered a multitude of previously unknown proteins associated with filopodium formation and suggested that there may be multiple mechanisms of filopodium formation. Although convergent elongation (Fig. 1A) might be an important pathway for filopodium formation in some cells, it cannot be the sole pathway. Another possible mechanism involves de novo filament nucleation, likely by Dia2 or other formin proteins, and the subsequent polymerization and bundling of filaments (Fig. 1B). A third alternative, particularly in neuronal growth cones, is the formation of filopodia from the reorientation and rapid polymerization of peripheral actin bundles toward the cell edge (Fig. 1C). Neuronal growth cones contain many transverse F-actin bundles (102). It would be interesting to determine whether these bundles can reorient perpendicular to the cell edge and initiate filopodium formation. Another likely possibility, due to the overlapping functions of the many actin-associated proteins, is that different sets of proteins might form filopodia by related mechanisms. Because polymerization and bundling are affected by a plethora of proteins associated with the filopodial tip complex and along the length of the filopodium, these functions may simply depend on which proteins are present and active within a particular cell. Reconstitution of filopodia in vitro with distinct sets of proteins will be enlightening in determining a mandatory set of proteins or protein functions required for filopodium formation.

Why should multiple mechanisms for filopodium formation exist? The expression profile of a particular cell type will determine the mechanisms available for filopodium formation. Similarly, the architecture of a cell may influence the type of mechanism used. For example, cells that are characterized by a highly branched F-actin meshwork mediated by Arp2/3, such as fibroblasts, may use a convergent elongation mechanism for filopodium formation, because most of the F-actin near their cell edges is in a meshwork configuration. Certain cell types lack highly branched actin meshworks, yet still form filopodia (6). For example, hippocampal neuronal growth cones plated on poly-d-lysine are characterized by peripheral F-actin bundles as opposed to densely branched F-actin meshworks. Inhibition of Arp2/3 complex activity in such neurons does not block filopodium elaboration (51, 103). Dorsal root ganglion neurons plated on laminin, however, have Arp2/3 and branched F-actin in the lamellipodial veils of growth cones, and this suggests, in this context, that Arp2/3 may play a role in filopodium formation (104). Other cell types have prominent stress fibers or F-actin bundles instead of F-actin meshworks, and these cells may use an alternative mechanism for filopodium initiation, such as reorientation or extension of preexisting F-actin bundles toward the cell periphery, or both.

Are All Filopodia Created Equal?

One feature that differs among filopodia formed by these different molecular mechanisms is the manner in which the filopodia are anchored into the cellular architecture (Fig. 1). For example, a filopodium formed through convergent elongation is anchored into the lamellipodium F-actin meshwork, but in the case of neuronal growth cones, filopodia are anchored into peripheral actin bundles. It is unknown, however, how filopodia formed from de novo filament nucleation are anchored. In light of studies that showed that membrane proteins such as LPR1 and IRSp53 play roles in the formation of the filopodium, we suggest that these and other similar proteins anchor the filopodium tip or shaft into cortical F-actin and the plasma membrane. The manner in which filopodia are anchored within the cell could affect the downstream functions of filopodia.

How else might filopodia formed by distinct mechanisms differ in their behavior and function? One possibility is that integrins and other cell-adhesion molecules only associate with the tips of filopodia that are formed by particular mechanisms. This might determine which ECM components the filopodia attach to, and which downstream signaling pathways are activated after attachment. Similarly, the structural and signaling proteins within a filopodium might determine which external signals, such as growth factors or guidance cues, the filopodium responds to and how that response is controlled. One could also imagine that the molecules present in the filopodium alter both its biochemical and biophysical properties. For example, the types and numbers of F-actin–bundling or F-actin–crosslinking proteins and the number of actin filaments would greatly affect the persistence length or stiffness of filopodia. These factors may then influence how filopodia respond to external mechanical cues, such as obstacles or shear stress, and how downstream signals are activated.

Future Directions

An important goal for the field is to determine the complete molecular inventory of filopodia and the filopodial tip complexes found in diverse cell types. Similar studies have revealed information about the molecular machinery of pseudopodia (105). Further characterization of the different players and mechanisms of filopodium formation will clarify the independence of different mechanisms, determine whether filopodia formed by different mechanisms have distinct functions, and determine how certain parameters of filopodial behavior, such as length, width, growth rate, and persistence, depend on the mechanism of formation. The involvement of so many molecules in filopodium formation begs the question of whether different mechanisms of filopodium formation evolved independently or were modified over time. Determining a molecular inventory for the many modes of filopodium formation among organisms from Dictyostelium to mammals and inspecting them for commonalities and divergence will likely yield new insight into the mechanisms of filopodium formation. Although all of the molecules involved in filopodium formation have not yet been revealed, it seems highly unlikely that a single mechanism of filopodium formation accounts for the diverse structures, behaviors, and functions of filopodia that occur in a range of cell types and organisms.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
View Abstract

Stay Connected to Science Signaling

Navigate This Article