Filopodia are dynamic structures that are primarily composed of F-actin bundles and whose initiation and elongation are precisely regulated by the rate of actin filament assembly, convergence and cross-linking.
The actin treadmilling mechanism of elongation is essential in protrusion  and any change in the frequency of initiation, or the balance of extension versus retraction of actin filaments, can result in the gain or loss of filopodia. Filopodial actin filaments are unbranched , and observations of filopodial formation have revealed that actin assembles at the filopodial tips, moves backwards, and dissipates at the rear . A complete model for how a filopodium is formed has recently been reviewed .
Current research into these models suggests that both remain plausible mechanisms for actin filament nucleation and filopodia initiation in vivo, however only the “convergent elongation model” is supported with direct experimental evidence. Much of this evidence is based on the effect of modulating activators or regulators of the Arp2/3 complex, although some findings are directly attributed to the expression and function of the Arp2/3 complex itself. One such study suggested that filopodia containing factin-1 mediated actin bundles formed from existing lamellipodia, and were important in serving as a template for the formation of new lamelliopodia .
It has been shown, for example, that Arp2/3 knockdown in cultured neurons  as well as loss-of-function mutations in C. elegens  and cultured Drosophila neurons , leads to a disruption in filopodia initiation and subsequently a decrease in the number of filopodia. Modulation of nucleation-promoting factors (NPFs) that act directly on Arp2/3 have also been shown to influence filopodia initiation. In one example, targeted depletion of SCAR by RNA interference reportedly inhibited both lamellipodia and filopodia formation in Drosophila where as depletion of N-WASP did not . It has also been reported that although N-WASP was not essential for filopodia formation, its activation lead to an increased number of filopodia, again implicating the Arp2/3 complex in filopodia initiation. This was observed in COS-7 cells co-expressing N-WASP with a Myc-tagged Cdc42 mutant (an activator of N-WASP) where ‘extremely long’ microspikes were described in half the anti-Myc positive cells . Similarly, expression of a mutant form of WASP (Y291E), which mimics phosphorylated Tyr291, was shown to induce filopodia in macrophages more effectively than wildtype WASP . The same study also reported that filopodia formation was induced in macrophages as a result of WASP over-expression, and noted that WASP co-localized with F-actin at the filopodia base . More recently, Robo4 was shown to induce filopodia formation in endothelial cells. This role was attributed to its activation of NPFs such as WASP, which in turn activates the Arp2/3 complex. It was also proposed that Robo4 may act as a molecular scaffold for the recruitment of of proteins that mediate actin nucleation . Additional experimental evidence supporting the “convergent elongation model” has been summarized extensively .
Once nucleation has taken place, actin filaments begin to extend. This process is primarily facilitated by members of the formin family of proteins, however numerous other proteins also play important roles. In the ‘convergent elongation model’ filament barbed ends are locally associated with each other and must be protected from capping in order for extension to occur . This protection is provided by the activity of proteins such as the Ena/VASP family of proteins which is delivered to the filopodia tip by myosin X . Here, Ena/VASP enhances filament polymerization, and promotes F-actin bundling, thereby stimulating filopodial protrusion .
Actin filaments in filopodia are unbranched , indicating filaments elongate primarily through the activity of formins. Following the rapid polymeriztaion facilitated by formins, a steady state mechanism that is common to many actin based structures will maintain filament length. This is known as actin treadmilling  and is described in greater detail in the functional module; actin filament polymerization.
As actin filaments extend, force is exerted on the cellular membrane, leading to filopodia protrusion. For efficient protrusion, membrane curvature rigidity must be overcome. This is aided by actin filament cross-linking, which gives the structure the strength required to push against the compressive force of the plasma membrane . In nerve growth cones, more than 15 parallel filaments may be bundled together by crosslinking . Bundle stiffness increases with the number of bundled filaments and so contributes to the overall length of the filopodium . Filament bundling results from crosslinking proteins, many of which co-localize at the base of filopodia and work in concert to produce crosslinked filaments . Examples include α-actinin and fascin. The filamin family of proteins are also crucial actin crosslinking and scaffolding proteins and bind to both actin and a number of signaling molecules, including Rho GTPases. Crosslinking also increases the ATPase activity of myosins and increases the tension on filaments .
As well as mediating cargo transport along actin filament bundles, recent studies have implicated myosin-X as integral to the initiation of filopodia and the elongation of long filopodia. This has been attributed to a mechanism of myosin-X that promotes actin filament bundling, similar to cross-linking.
Studying the localization and motility of single myosin-X molecules using TIRF microscopy, Watanabe et al, hypothesized that binding of cargo, and the preceding dimerization of myosin-X monomers at the cell periphery is important for filopodia initiation as it promotes actin filament bundling . This was proposed to occur in a similar manner to regular crosslinking. Previous reports had indicated that even without the cargo-binding FERM domain, lateral movement of myosin-X along the leading edge of lamellipodia promoted actin reorganization and through its motor activity, filopodia initiation. In this case the length of the head and neck domain of myosin-X was proposed to be important for initiation. In the study by Watanbe et al a rapid increase in the rate of recruitment and assembly of myosin-X at filopodia initiation sites moments before protrusion commenced was observed and again this was shown to be independent of the presence of the FERM domain.
After removing the FERM domain from myosin-X (using a FERM domain-truncated construct called M10-ΔFERM) the protein was still able to walk along actin filament bundles and continued to localize at both the leading edge and filopodia tip. Significant changes to the length and stability of the growing filopodia were however noted. Not only were filopodia significantly shorter and more unstable, as previously reported , but the phased-extension mechanism of elongation observed when complete myosin-X was present, did not occur.
Although it was noted that without the FERM domain the transport of essential cargo to the tips of the filopodia would be insufficient to permit continued elongation, it was also proposed that through FERM-β-integrin interactions at the tips of filopodia (when filopodia are attached to substrates via focal adhesion sites), myosin-X may also possess adhesive qualities. In this hypothesis, as filopodia enter the retraction phase and actin filaments move back towards the cell body by retrograde flow, myosin-X remains at the tips of filopodia together with the adhesion complex. Upon shrinkage of the tips, the protein will rebind to actin filament bundles and allow a new phase of extension to begin from adhesion sites. This mechanism of phased-extension is supported by observations that without the FERM domain myosin-X diffuses back into the cell body during the retraction phase whilst intact myosin-X remains at the tips .
The extension rate of a filopodium will differ depending on the cell type (table 1). In each case however, this rate is controlled by the availability of G-actin-ATP, associated structural components and the energetics of membrane bending. The growth of long filopodia (>10 μm in length) requires the rapid transport of key materials towards the growing end  and this process is facilitated by the myosin motor proteins such as myosin-X or myosin V using an ATP-dependent ‘walking’ mechanism. Although the extension of filopodia is often described in a highly ordered manner, and does rely on the defined movements of Myosin-X for component delivery to the filopodia tip, the contribution of random diffusion of components must also be considered. Stochastic simulation models were recently presented that describe such phenomena, where ‘molecular noise’ may influence concentration gradients of G-actin as well as efficacy of the machinery responsible for filopodia growth.
One such study indicates that although the spatial distribution of static Myosin-X is universally consistent, and not altered by organelle length, the concentration of walking motors may vary. “Traffic jams” of myosin-X may occur for example at the base of the filopodia. Although logically this will impede progression of the proteins and their cargo down the filament, it was calculated that following a build up of G-actin at the blockage site, a concentration gradient is generated that enables its diffusion down the filopodia, (so long as the G-actin is not sequestered by the blocked motors) which subsequently sustains filopodia extension . Similarly, the constant association and dissociation of capping protein at the barbed ends of actin filaments has been shown, also using stochastic simulations, to influence the dynamics of filopodia extension. In this case amplification of these regulatory proteins from initially low concentrations may trigger the fast retrograde flow of actin and induce repeated extension-retraction cycles that occur on a micro-meter scale. Compared to actin-only models, these dynamic cycles enable the growth of substantially longer filopodia .
Filopodia are motile structures, being able to extend, retract and move laterally as they sense their environment. Lateral movement is particularly important in allowing the structure to sense stimuli prior to its adhesion with another cell or substrate.
Lateral movement may be a consequence of filopodial protrusions being tilted with respect to the retrograde flow of actin . This movement is inhibited once filopodia adhere to cells or substrates , however it is important to note that newly emerged filopodia seldom adhere to the substratum at their tips and are instead more likely to adhere at their bases .
Lateral movement and merging of filopodia may also stimulate further elongation of filopodia, suggesting that this may be an alternative method for promoting filopodial maturation and growth cone advancement on less adherent substrates .
Typically filopodia are quite dynamic and are constantly growing or retracting. Thus, periods of stasis are often short-lived and even adhesion of the filopodial tip to a substrate will not last long before the cell pulls on the site and recruits additional components or retracts, leaving behind a thin tube of membrane. There are several events that may promote stasis in retracting filopodia.
After a ligand binds to a receptor on the filopodium, lateral connections between the receptor and the actin bundle(s) prevent retrograde actin movement relative to the ligand. This inhibition converts the previous retrograde movement of the retracting filaments into tension on the bundles. This tension might influence polymerization and stability of actin filaments and/or myosin activity.
Retracting filopodia become static when receptor binding leads to filament uncapping and rapid polymerization at the barbed end. The retraction force is converted to retrograde flux and the filament length remains constant.
If retraction occurs by barbed end disassembly, capping of unstable filaments by an activated receptor will block retraction and induce stasis.
A diverse array of cellular responses can result when a filopodium makes contact with a ligand or substrate. These responses are dependent on the coupling of membrane-bound proteins to the backward (retrograde) flow of actin that drives filopodia elongation and motility. This contributes to larger processes such as cell pulling, which is needed for cell migration in wound healing, neurite growth , filopodia collapse and momentary stasis. These responses are influenced by the strength of the adhesion and it is known that contact differences between substrates or cell types influence the number of protruding filopodia .
Three distinct types of adhesions can be identified within filopodia. Adhesion components may be localized to the tips of filopodia or be actively transported down the bundled actin filaments by myosin proteins. Each adhesion may function independently or work in concert to produce the overall guidance response:
Filopodia on apposed cells interact directly through their tips  and/or ‘slide’ past each other (interdigitate) in order for adhesions to form (through cadherin-cadherin contact) between the tip of one filopodia and the cell membrane at the base of the adjacent filopodium . Tip adhesions are likely to contain proteins that are found in nascent focal complexes . Depending on the type of substrate, and strength of the adhesion, the resulting response may differ. In filopodia extending from neuronal growth cones, signals originating from tip adhesions can result in the formation of shaft adhesions, veil advance, and ultimately, growth cone movement .
A single filopodium can have both non-adherent and adherent regions along the shaft. Shaft adhesions develop de novo along the filopodium and do not represent former tip adhesions . Growth cone veils advance easily on non-adherent regions and cease movement as they encounter stable shaft adhesions; lateral mobility, veil advance and the merging of filopodia are also regulated by shaft adhesions .
Basal adhesions play a specific role in filopodia initiation and are found in ~98% of all filopodia, where they anchor the filopodial base that usually remains immobile despite considerable flexibility in the shaft . These are stable adhesions that contain a focal ring structure believed to convert tension forces into filopodia formation. Basal adhesions are stable and are formed before the filopodium emerges. This has been observed in growth cones where they remain in place as the growth cone advances .
Increasing the force on an adhesion, either by an external source or by increasing cellular contractility, strengthens and enlarges the adhesion . Similarly, in basal adhesions of filopodia in neuronal growth cones, the size and stability of nascent adhesion increases in a maturation process that is reminiscent of the focal adhesions (FAs) found in non-neuronal cells . Though smaller than FAs, these adhesions share a number of signal components, pathways, and proteins leading to adhesion site formation and maturation .The structure of the growth cone adhesion site varies with the substrate  and adhesion directs growth cone navigation and movement (reviewed in ).
Although a reterograde motion of actin filaments is intrinsic in the formation of filopodia, the forces generated by actin treadmilling are too weak to facilitate the “pulling” mechanism required for rigidity sensing and other mechanosensing processes. This characteristic of filopodia is instead produced by the activity of myosin motor proteins such as Myosin II . This process, as described in the Functional Module: ‘Microfilament Motor Activity’, is similar to that observed in the actin filament networks of the lamellipodia and lamella. Disassembly of actin bundles at the filopodial tip combined with the tension of the lipid bilayer also theoretically support the forces needed for pulling .
Filopodia pulling is important in a number of processes additional to mechanosensing. One example is in the immunological response to pathogens where immune cells such as macrophages will pull a pathogen toward the cell body for active uptake and processing . In another example the filopodia of mature osteoclasts bind to extracellular substrates at their tips and transport the particles rearward by retraction for resorption to the cell body, leaving the cell body adhered to the substrate . It has also been shown that matrix coated beads are pulled rearward by an active process rather than by diffusion after the beads bind to integrins .
In each case the mechanism behind pulling remains the same. While the powerstroke of Myosin II essentially drives the movement of filaments, the pulling process occurs over three distinct stages, each defined by varying rates of retraction. These can be described using the process by which pathogens are pulled towards macrophage cell bodies as an example. Here, immediately following filopodial adhesion to a pathogen, a slow motion retraction of the filopodia commences. This is followed by a phase of rapid movement where the pathogen is pulled towards the cell. A final phase of slow retraction occurs, resulting in the pathogen being fixed to the cell surface. This final phase of retraction is also characterized by a thickening of the filopodial base which is necessary to build-up large retraction forces for internalization . Filopodial attachment to a surface (in addition to the ligand) produces counteracting adhesion forces which influence the retraction speed. In general, the rearward movement or ‘step size’ of a retracting filopodia decreases as the filopodia encounters a greater counteracting force . Importantly, the mechanism may also be hijacked by a pathogen to ensure an efficient infection. This has been described for the Murine Leukemia Virus, amongst others . In this case the virus binds to surface receptors at the tips of the filopodia and essentially rides the retracting actin filaments to entry points in the cell where it is internalized.
Binding of filopodia to certain ligands or substratum may hinder filament assembly, thereby leading to changes that promote retraction, collapse or growth cone turning . For example, substrate contacts with a repulsive signal on one side of a filopodium causes growth cone turning, while contact across the entire filopodial tip circumference causes complete filopodium collapse . Normal retrograde flow of material continues under both circumstances and collapse is independent of the maintenance of growth cone protrusive activity . Resorption and collapse are related processes and likely involve the same core proteins.
Rapid collapse produces a large number of filopodial strands tightly connected to the substrate by long tethers. F-actin bundles  and monomeric actin  disappear from collapsing filopodia without a compensatory rise in F-actin at the growth cone center; this indicates a net loss of actin rather than a rearward translocation. Furthermore, active nucleation and protrusion of filopodia is still found in discrete areas of collapsing growth cones, which argues against sequestration or modification of actin as the mechanism responsible for the loss of F-actin during the collapse . It has also been shown that filopodial retraction involves periodic helical and rotational motion of the actin shaft, together with the retrograde flow of actin, and these dynamics were together responsible for filopodial shortening during retraction 
A number of factors regulate collapse and retraction. For example, capping proteins promote filopodial retraction by shielding the barbed end of filaments from further assembly and elongation . Inhibition of F-actin polymerization and protrusion during collapse are mediated by RhoA kinase activity . Collapse may result from the exposure of a “naive” growth cone to a high concentration of a repellent followed by an overactive response . The repulsive component appears to shut down the growth program and is therefore dominant over the growth-stimulating effects of adhesion molecules. In addition, the repellent also interferes with mechanisms that would normally result in filopodial retraction .
Growth cone collapse is a complex phenomenon involving numerous signal pathways including Rho-GTPases , ADF , and various kinases . A model for filopodia collapse in growth cones was created using the guidance signal, semaphorin IIIA (SemaIIIA; collapsin-1). SemaIIIA causes termination of protrusive activity and growth cone collapse  through decreased phosphorylation of the ezrin–radixin–moesin (ERM) family of F-actin binding proteins . Phosphorylation of ERM proteins activates the F-actin binding domain and regulates filopodia assembly/protrusion by linking filopodial membranes with F-actin (reviewed in ). Inactivation of the phosphoinositide 3-kinase (PI3K) signal pathway by SemaIIIA may also be linked to reduced ERM protein activity and growth cone collapse .