Regulating Filopodial Dynamics Through Actin-Depolymerizing Factor

Home , Filopodia, MYO1C

Anatomical Science International (2004) 79, 173–183 SpecialBlackwell Publishing, Ltd. Review Based on a Presentation made at the 16th International Congress of the IFAA Regulating ﬁlopodial dynamics through actin-depolymerizing factor/coﬁlin Joseph Fass,1 Scott Gehler,2 Patrick Sarmiere,1 Paul Letourneau2 and James R. Bamburg1 1 Department of Biochemistry and Molecular Biology, Colorado State University, Fort Collins, Colorado and 2 Department of Neuroscience, University of Minnesota, Minneapolis, Minnesota, USA

Abstract The regulation of filopodial dynamics by neurotrophins and other guidance cues plays an integral role in growth cone pathfinding. Filopodia are F-actin-based structures that explore the local environment, generate forces and play a role in growth cone translocation. Here, we review recent research showing that the actin-depolymerizing factor (ADF)/cofilin family of proteins mediates changes in the length and number of growth cone filopodia in response to brain-derived neurotrophic factor (BDNF). Although inhibition of myosin contractility also causes filopodial elongation, the elongation in response to BDNF does not occur through a myosin-dependent pathway. Active ADF/cofilin increases the rate of cycling between the monomer and polymer pools and is critical for the BDNF-induced changes. Thus, we discuss potential mechanisms by which ADF/cofilin may affect filopodial initiation and length change via its effects on F-actin dynamics in light of past research on actin and myosin function in growth cones. Key words: actin-depolymerizing factor, cofilin, filopodia, growth cone.

Introduction processes; growth cone morphology and the behavior of its actin and microtubule cytoskeletal structures Neuronal growth cones rely on dynamic filopodia is modified downstream of guidance molecules to explore their surroundings, as well as for force (Gallo et al., 1997; Gallo & Letourneau, 2004; Zhou generation related to motility (Heidemann et al., & Cohan, 2004) and axonal collateral branching 1990). Filopodia are generally short (0.5 to several (following extensive filopodial sprouting) can result µm) membranous protrusions containing a core of from contact with beads coated with nerve growth colinear, bundled F-actin filaments, but also contain factor (NGF; Gallo & Letourneau, 1998). In addition, den- myosin and components of focal contacts (Bridgman dritic filopodia are precursors to dendritic spines, & Dailey, 1989; Steketee & Tosney, 2002). The F-actin post-synaptic structures whose diverse and dynamic in filopodial bundles is cross-linked with fascin (Cohan morphology is thought to underlie learning and mem- et al., 2001) and is very stable, having a half-life ory (Sala, 2002). Neurotrophins have been implicated more than 10-fold longer (approximately 25 min) in modulating the formation and morphology of both than lamellipodial F-actin (Mallavarapu & Mitchison, dendritic filopodia and spines (Shimada et al., 1998; 1999). Matsutani & Yamamoto, 2004). Furthermore, a point Regulation of filopodia in neurons is important mutation that impairs secretion of the neurotrophin for both growth cone guidance and neurite branch brain-derived neurotrophic factor (BDNF) is corre- formation, both of which determine the ultimate lated with hippocampal functional impairment in arborization and connectivity in the nervous system humans and animals and plasticity impairment in (Goodman & Shatz, 1993; Gallo & Letourneau, 1999). hippocampal cell culture models (Egan et al., 2003; Contact with guidance cues directs both of these Hariri et al., 2003), providing a potential link between filopodial regulation and adult cognitive function. Growth cone filopodia explore their local environ- Correspondence: James R. Bamburg, Department of ment via extension, waving and retraction. When Biochemistry & Molecular Biology, 1870 Campus Delivery, Colorado State University, Fort Collins, CO 80523-1870, USA. adhered at their tips, they can exert a pulling force Email: [email protected] (Heidemann et al., 1990; Smith, 1994; Suter et al., Presented at the 16th International Congress of the IFAA, held 1998; Suter & Forscher, 2000) that can either move at the International Conference Hall, Kyoto, 22–27 August 2004. attached objects (such as beads or other neurite Received 3 September 2004; accepted 13 September 2004. shafts) rearwards towards the growth cone or pull 174 J. Fass et al. material from the growth cone central region antero- of F-actin, active AC proteins can participate in gradely, resulting in engorgement of the filopodium enhancing the rate of cycling, or turnover, between and advance of the growth cone along the path of G- and F-actin in many systems (Carlier et al., 1997; the filopodium (Goldberg & Burmeister, 1986; Smith, Lappalainen & Drubin, 1997; Rosenblatt et al., 1997; 1994; Letourneau, 1996). Engorgement of a filopo- Theriot, 1997), including within neuronal growth cones dium into a neurite is a proposed mechanism for (Meberg & Bamburg, 2000). interactions with guidepost cells during grasshopper How might AC proteins be involved specifically in pioneer axon extension (Sabry et al., 1991) and has filopodial regulation? Recently, Gehler et al. (2004a) been observed during interactions between Aplysia demonstrated that BDNF signals in retinal ganglion growth cones and restrained (immobilized) Aplysia growth cones by inhibiting Rho GTPase. Working cell adhesion molecule (apCAM)-coated beads downstream of Rho, Gehler et al. have found that (Suter et al., 1998), as well as between chick sym- BDNF effects on filopodia are mediated through pathetic neuron growth cones and other cells or modulation of AC protein activity and, although myosin beads coated with polyornithine or laminin (Smith, II-dependent alterations in filopodia length can occur 1994). This filopodial behavior seems analogous to in these growth cones, BDNF does not work through the entire growth cone, in which there is an inverse modulating myosin II activity (Gehler et al., 2004b). relationship between retrograde flow and growth These experiments will be reviewed first and we will cone advance (Lin & Forscher, 1995). This relation- then address their potential implications for how AC ship has led to a model for forward motility whereby proteins may function in filopodial initiation, extension a cellular process will either be retracted at some and retraction. time after protrusion or adhere to some stable substrate, allowing the same forces that would have AC regulation of filopodia caused retraction to result in forward movement of more proximal cellular components. Generation of Gehler et al. previously demonstrated that BDNF pulling force is dependent upon motor proteins (for binding to the p75 neurotrophin receptor (p75NTR) a review, see Brown & Bridgman, 2004), presumably enhances filopodial length by decreasing the activity isoforms of myosin attached to actin filaments with of RhoA (Gehler et al., 2004a). Their results showed opposite orientations (Wylie et al., 1998; Bridgman that several neurotrophins (BDNF, nerve growth et al., 2001; Wylie & Chantler, 2001; Bridgman, factor (NGF), and neurotrophin (NT)-3) cause increases 2002; Brown & Bridgman, 2003b). However, it is as in filopodial length in chick retinal, dorsal root yet unclear which specific isoforms of myosin are ganglion (DRG) and ciliary growth cones (Gehler involved directly in filopodial retraction or, for that et al., 2004a). Furthermore, these length changes were matter, engorgement. mimicked and pre-empted by treatment with either Nevertheless, filopodial behavior is not simply of two p75-specific antibodies: CHEX and AB1554. the result of motor-based rearrangement of static Neurotrophin-induced filopodial length changes were actin structures. F-actin throughout the growth cone not observed in growth cones of p75–/– mouse retinal is constantly turning over, via polymerization at the or DRG neurons, yet these growth cones had longer leading edge, retrograde flow of F-actin and depo- filopodia, similar to neurotrophin-treated p75+/+ growth lymerization within the transitional zone (Forscher & cones, implicating unoccupied p75NTR in downregu- Smith, 1988; Lin & Forscher, 1995; Welnhofer et al., lation of filopodial length. Finally, both p75–/– and 1997; Schaefer et al., 2002). One of the major agents BDNF-treated neurons displayed reduced levels of responsible for this turnover is the actin-depolymerizing RhoA activity. More recent experiments have dis- factor (ADF)/cofilin (AC) family of proteins (reviewed sected the signaling downstream of Rho, beginning by Sarmiere & Bamburg, 2004). The AC proteins with the rho-associated coiled-coil kinase (ROCK), bind preferentially to F-actin on which the associated also called Rho kinase (ROK) or Rho-associated ATP has been hydrolyzed to ADP (Carlier et al., 1997) kinase (Gehler et al., 2004b). and promote both severing of filaments (Bamburg Rho-associated kinase is well characterized as et al., 1999; Ichetovkin et al., 2000, 2002) and depo- an upregulator of myosin II-dependent contractility lymerization from the pointed end. Once in the mon- in non-muscle cells (e.g. Hirose et al., 1998) that acts omer state, profilin induces a rapid exchange of through either direct activation of myosin light chain ATP for ADP on the G-actin and, because AC proteins kinase or inhibition of the phosphatase that removes preferentially bind ADP–actin monomers, they are the phosphate from the light chain (Amano et al., free to recycle to F-actin to repeat their dynamizing 1996; Kimura et al., 1996). In addition, ROCK regu- effects (Pollard et al., 2000). Profilin-bound ATP–G-actin lates AC activity via the activation of specific kinases, readily participates in barbed-end polymerization; namely LIMK1 and 2 (named for the proteins lin-11, thus, rather than simply promoting depolymerization islet-1 and Mec-3, which contain similar LIM domains) ADF/cofilin and filopodial dynamics 175

Figure 1. Schematic representation of the regulatory interactions between several actin regulatory proteins and the actin polymerization/ depolymerization cycle. p21-activated kinase 4 (PAK4) activates (phosphorylates) LIM kinase 1 (LIMK1) and inactivates (phosphorylates) slingshot-1L (SSH-1L) on a site other than the C-terminal tail domain (J. Soosairajah et al., unpubl. obs., 2004). Slingshot inactivates (dephosphorylates) LIMK1 and activates (dephosphorylates) actin-depolymerizing factor (ADF)/cofilin (AC), resulting in a net increase in AC activity and actin filament turnover. The other players in this pathway are F-actin and 14-3-3ζ. F-Actin is necessary for slingshot activity; the interaction of SSH-1L, which has been phosphorylated in its C-terminal tail domain by an unidentified kinase, and F-actin is inhibited by 14-3-3γ and 14-3-3ζ (Nagata-Ohashi et al., 2004; J. Soosairajah et al., unpubl. obs., 2004). Stimulatory interactions (arrows) and inhibitory interactions (blocked lines) are shown. Active AC increases F-actin turnover rate by binding preferentially to ADP–actin subunits (white) as opposed to ADP·Pi-actin (gray) and ATP-actin subunits (black). Bound AC increases F-actin severing and pointed end depolymerization. Free ADP–G-actin monomers can exchange their bound nucleotide (enhanced by profilin) and engage in polymerization (predominantly at the barbed end). and testicular kinase (TESK) 1 and 2, which phos- included the AC-binding scaffolding protein 14-3-3ζ phorylate and inactivate AC proteins at serine 3 and several mutants of Xenopus AC (XAC): a non- (Arber et al., 1998; Yang et al., 1998; Toshima et al., severing mutant XAC KK95,96QQ (KKQQ), a non- 2001a, 2001b). The dephosphorylation (activation) phosphorylatable and, therefore, constitutively active of AC is regulated both by general phosphatases XAC(S3A) (XACA3) and a pseudophosphorylated (Meberg et al., 1998; Samstag & Nebl, 2003) and and, therefore, less active XAC(S3E) (XACE3), which more specific phosphatases, such as those of the acts in a dominant negative manner in polarization slingshot family, of which six isoforms have been of cultured fibroblasts (Dawe et al., 2003). identified in mammalian cells (Niwa et al., 2002; The BDNF-induced increase in growth cone Endo et al., 2003; Ohta et al., 2003; Nagata-Ohashi filopodial length (Gehler et al., 2004a, 2004b) could et al., 2004). Although the mechanisms regulating be mimicked by ROCK inhibition or introduction of the slingshot phosphatases are not all known, one of XACA3 and blocked by introduction of XACE3 or these isoforms, namely slingshot-1L (SSH-1L), requires the non-severing XAC KKQQ mutants or 14-3-3ζ. No binding to F-actin for activity (Ohta et al., 2003; further increases in filopodial length were observed Nagata-Ohashi et al., 2004) and is also phosphor- when BDNF was used to treat cells pretreated with ylated and inactivated by p21-activated kinase 4 a ROCK inhibitor or preloaded with XACA3. Immu- (PAK4; J. Soosairajah et al., unpubl. obs., 2004) and nostaining specific for inactive (phosphorylated) AC other as yet unidentified kinases. In addition, the revealed similar reductions throughout the growth dephosphorylation of AC is regulated by its binding cone after BDNF or ROCK inhibitor treatment; these to members of the 14-3-3 family of phosphoprotein- reductions were blocked in cells loaded with 14-3- binding partners (Gohla & Bokoch, 2002). Both 3ζ, suggesting 14-3-3ζ prevented the filopodial LIMK1 and slingshot also bind 14-3-3 family mem- length increase through its ability to inhibit activation bers and because the consequences of these of endogenous AC. Finally, slingshot immunostaining interactions are not well understood, the effects of was punctate throughout the growth cone and, in the Rho signaling pathway on AC activity may be quite some cases, was observed along and at the tips of complex. However, a summary of the known regula- filopodia, although its involvement in BDNF signaling tory pathways is given in Fig. 1. was not assessed directly. Gehler et al. (2004b) treated cultured retinal Gehler et al. also assessed ROCK and XAC ganglion neurons with a number of inhibitors or loaded effects on filopodial number in the same way, but the the neurons with purified proteins using the Chariot™ results were somewhat more ambiguous (Gehler reagent (ActiveMotif, Carlsbad, CA, USA). Inhibitors et al., 2004b). Inhibition of ROCK and loading with included the ROCK inhibitors Y-27632, HA-1077 and XACA3 mostly recapitulated BDNF-induced increases H-1152 and blebbistatin, a small molecule inhibitor in filopodial number, but further increases were of myosin II (Straight et al., 2003). Proteins loaded observed when these treatments were combined with 176 J. Fass et al.

BDNF. Similarly, XACE3 loading did not completely increase the number of free barbed ends, thus allow- block the effects of BDNF treatment on filopodial ing a burst of actin polymerization that drives lamel- number, whereas loading of XAC KKQQ or 14-3-3ζ lipodial protrusion and can thereby determine the did. These inconsistencies may be the result of direction of lamellipodia-driven migration (Ichetovkin unequal protein loading, partial activity of XACE3 or et al., 2002; DesMarais et al., 2004; Ghosh et al., the presence of endogenous AC proteins. However, 2004). The finding of Gehler et al. (2004b) that the taken together, these results suggest that whereas non-severing KKQQ mutant blocked BDNF-induced BDNF controls filopodial length solely through AC changes as well as the XACE3 mutant suggests that regulation, the effects of BDNF on filopodial number severing is also important in filopodial length and are only partially AC dependent and may involve an number regulation and, on a qualitative basis, BDNF AC-independent pathway. treatment was seen to induce an increase in lamel- Given the downregulation of Rho and ROCK by lipodial area (S. Gehler & P. Letourneau, unpubl. BDNF, another pathway through which filopodial obs., 2004). However, it is not known whether BDNF elongation and numbers could be affected is through induced bulk polymerization or depolymerization in modulation of myosin II activity. To investigate the the experiments of Gehler et al. (2004b). role of myosin II, Gehler et al. (2004b) used bleb- Whereas Gehler et al. (2004b) applied BDNF in bistatin, an inhibitor of non-muscle myosin II, both solution, it seems likely that growth cones would alone and in concert with the various treatments encounter more localized sources of neurotrophins already described. Treatment of growth cones with during development. However, even in the case of blebbistatin resulted in concentration-dependent homogeneous activation, Gehler et al. did not exam- increases in filopodial length and number but, over- ine whether AC proteins are differentially activated all, the morphological responses differed from those or targeted to different compartments of the growth seen following BDNF treatment. Time-lapse imaging cone or F-actin populations. Phosphorylated AC revealed that 20 µmol/L blebbistatin increased staining appears fairly homogeneous throughout the filopodial extension rates, while gradually reducing growth cone (once thickness is accounted for; see retraction rates. Paradoxically, blebbistatin also Fig. 2A) and does not disappear preferentially from reduced the fraction of time filopodia spent extend- any particular region after exposure to BDNF. How- ing or retracting. In addition, blebbistatin treatment ever, simultaneous tracking of both total AC proteins caused lamellipodial retraction, as well as filopodial and their activation state is required in order to detachment, bending and buckling. In contrast, ascertain whether localized activation of AC proteins BDNF treatment did not result in lamellipodial retrac- is occurring in response to even homogeneous tion and filopodia remained substrate bound. Treat- exposure to BDNF. In addition, there may be other ment with BDNF resulted in enhanced filopodial mechanisms localizing the effects of AC in the extension rates, and no change in either retraction growth cone, such as protection of some fraction of rates or in the fraction of time filopodia spent extend- the F-actin population by tropomyosin (DesMarais et ing or retracting. al., 2002). Further evidence that the effects of BDNF on Nevertheless, the work of Gehler et al. (2004b) filopodial length are not mediated via myosin II came strongly suggests that BDNF regulates filopodial length from experiments combining blebbistatin treatment solely through regulation of AC protein activation, with BDNF, ROCK inhibitors and XAC mutants whereas filopodial number may involve additional (Gehler et al., 2004b). Blebbistatin (20 µmol/L) caused signaling pathways. However, it seems unlikely additive increases in filopodial length when used (although not implausible) that length and number of in concert with BDNF, ROCK inhibitor or XACA3 filopodia are regulated directly (i.e. that there is a loading, whereas XACE3 loading did not block direct feedback pathway from some sensor of length blebbistatin-induced increases in filopodial length. or number to a mechanism for controlling length or Blebbistatin did not cause changes in filopodial number number). Rather, we hypothesize that the number at concentrations lower than 50 µmol/L, at which and length of filopodia present at any time depends additivity with other treatments was assessed. Thus, on several mechanisms: initiation, extension and results from Gehler et al. (2004b) suggest that BDNF retraction. Thus, we ask which of the processes acts to regulate growth cone filopodia independently causing filopodia to arise or change in length may of myosin II contractility. be affected by AC regulation. One major question arising from the results described above is: what direct effect is increased Initiation AC activity having on growth cone F-actin? Recent work by Condeelis and colleagues has shown that Data from melanoma cells suggest membrane-bound activated cofilin and Arp2/3 act in concert to enabled/vasodilator-stimulated phosphoprotein (Ena/ ADF/cofilin and filopodial dynamics 177

Figure 2. Brain-derived neurotrophic factor (BDNF)-induced reduction of phosphorlyated actin-depolymerizing factor (ADF)/coﬁlin (pAC) levels in retinal growth cones. Treatment with BDNF for up to 30 min resulted in a progressive reduction in the intensity of pAC staining (A,C,E,G). Growth cones were counterstained with Alexa Fluor 568-phalloidin (B,D,F,H).

VASP) family proteins accumulate in a specific area precede the formation of localized Ena/VASP patches of the leading edge membrane, where they compete or form as a consequence. with capping proteins for barbed ends of actin Regardless, there are no data concerning the filaments in the lamellar meshwork and, thus, allow dependence of either of these hypothesized localized elongation and consolidation (through mechanisms on the rate of actin turnover, which bundling of the elongating filaments; e.g. by fascin) is increased in growth cones expressing XAC(wt) of the affected filaments, a process called conver- (Meberg & Bamburg, 2000) and is presumably gent elongation (Svitkina et al., 2003). Svitkina et al. increased as a result of the AC activation following (2003) termed the initial structures resulting from BDNF treatment and ROCK inhibition and decreased this process ‘Λ-precursors’ due to the arrowhead- following loading with XACE3 or KKQQ mutants. In shaped concentration of local actin filaments the case of the convergent elongation hypothesis, (Fig. 3). The Λ-precursors have not yet been one may imagine that more rapid turnover may allow observed in growth cones and growth cones differ more frequent convergence of peripheral actin from migrating cells in important aspects, such as filaments to form Λ-precursor structures that either the distribution of Arp2/3, which is enriched in the meet membrane-associated concentrations of initia- peripheral ruffling lamella of fibroblasts but, in growth tion factors, such as Ena/VASP, or bring together cones, is enriched in the central domain (Strasser et barbed-end associated proteins that then further the al., 2004). However, there is evidence to suggest that initiation process. the convergent elongation model may extend to neurite growth cones, where Ena/VASP is located Extension at the tips of filopodia and is essential for filopodial formation in response to Netrin-1 (Lebrand et al., 2004). Once nascent filopodia have grown beyond the Another proposed mechanism for growth cone Λ-precursor or bud stage, what determines their filopodial initiation involves actin filament nucleation ultimate length or, rather, the length distribution of all from growth cone structures called focal rings the filopodia on a growth cone at a certain time (Steketee et al., 2001). These structures are found point? G-Actin adds to the tips of filopodia (Okabe at the bases of filopodia and may be associated with & Hirokawa, 1991), where bundled F-actin filaments defining the site of the initial nub, which develops are thought to cause membrane protrusion by an into a filopodium. Convergent elongation and focal elastic Brownian ratchet mechanism (Mogilner & ring-dependent filopodial initiation are not mutually Oster, 2003; Upadhyaya & van Oudenaarden, 2004). exclusive mechanisms, because focal rings could By observing photobleached or photoactivated marks 178 J. Fass et al.

Figure 3. Platinum replica electron microscopy (top) of the leading edge of a mouse melanoma B16F1 cell showing a filopodium (arrow) and a Λ-precursor (arrowhead). Some filaments converging to the vertex of the Λ-precursor are highlighted in cyan. Fluorescence images of the same cell expressing a green fluorescent protein (GFP)-fascin chimera (green) costained with rhodamine-phalloidin (red) are shown at the bottom. The GFP-fascin is accumulated in the mature filopodium, but not in the Λ- precursor. A boxed region from a lower magnification view (lower left) is enlarged in the lower right panel; the electron micrograph (top) depicts a region corresponding to the boxed area in the light microscopic image at the lower right. Bar, 250 nm. (Images courtesy of T. M. Svitkina, University of Pennsylvania.) on filopodial actin bundles, Mallavarapu and Mitchison The BDNF-induced filopodial length increases (1999) revealed two distinct, seemingly independent reported by Gehler et al. (2004b) could have been mechanisms of control over filopodial length versus due to either a reduction in retraction rate or an time. They determined that both retraction rate (indi- increase in polymerization. One possibility is that cated by retrograde motion of the fiduciary marks) increased AC activity may lead to an increased G- and elongation rate (indicated by an increase in the actin concentration, but this is not necessarily the relative distance between a fiduciary mark and the case. However, DesMarais et al. (2004) found that end of the filopodium) governed changes in the posi- by blocking either Arp2/3 or cofilin function they tion of the filopodium tip (i.e. total filopodial length, could prevent an increase in free barbed ends and assuming lack of lamellipodial advance or retraction). concomitant lamellipodial protrusion in rat adenocar- Their data also suggested that there was a stronger cinoma cells. Conceivably, activation of cofilin (and/ correlation between length change and elongation or ADF) leads to a greater prevalence of severing (polymerization) rate than between length changes and then minus-end depolymerization, whereas acti- and retraction rates. However, these studies were vation of both cofilin and Arp2/3 leads, more often, performed in untreated growth cones; drug or growth to severing and then protection of the free pointed factors could, potentially, shift dominance between ends, limiting minus-end depolymerization and these two processes. preserving free barbed ends. The former would cause ADF/cofilin and filopodial dynamics 179 an overall shift from F-actin to G-actin, whereas the surface adhesions and F-actin within the growth latter would cause the reverse. Thus, the BDNF- cone or neurite shaft. Filopodia can also form surface mediated increase in AC activity alone may have led adhesions and it appears that these adhesions to increased turnover as well as somewhat elevated serve different functions (including regulation of G-actin levels, which then increased the rate of lamellipodial veil advance between adjacent filopo- polymerization at filopodial tips. dia) according to their locations along a filopodium Alternatively, AC activity could ‘chop up’ the actin (Steketee & Tosney, 2002). The idea that filopodia meshwork surrounding the base of a filopodium. specifically, and the growth cone in general, can Presumably it is myosin-generated tension produced either retrogradely pull in protrusions (filopodia or between this actin meshwork and the filaments of lamellipodial) or grab hold of substrate adhesions a filopodial actin bundle that results in retrograde and generate tension leading to growth cone advance motion of that filopodium (Jay, 2000) and a highly is known as the clutch hypothesis (Mitchison & severed actin meshwork with less structural integrity Kirschner, 1988; Jay, 2000), but the locations and may serve as a poorer base for tension generation. dynamics of ‘clutching’ (i.e. the force distribution over Less tension leads to less retrograde flow and, time within growth cones) are not well characterized. thus, with even a constant rate of polymerization at Many classes of non-muscle myosins are found filopodial tips, filopodial length would increase. If this in neuronal growth cones and their roles in neurite is the mechanism underlying increased filopodial outgrowth and growth cone motility have been length, it is evidently not entirely effective; as shown reviewed recently (Brown & Bridgman, 2003a; Brown in Fig. 2, there is no marked decrease in F-actin & Bridgman, 2004). Briefly, although myosin II has density evident after BDNF treatment, during the same been the best studied of they myosin classes, time that filopodial length is increasing. Thus, it is there is still controversy over the roles of its specific reasonable that inhibition of myosin II via blebbistatin isoforms (Brown & Bridgman, 2003a). Myosin IIb would further decrease the rate of retrograde ‘reeling appears be responsible for retrograde flow of F-actin in’ of filopodia, further increasing length. in growth cone lamellipodia and myosin IIa may function redundantly. In addition, myosin IIa may have Retraction a role in stabilizing nascent focal contacts (Wylie & Chantler, 2001). Myosin 1B (MYO1B) and D are As mentioned above, we must interpret the results present in neurons, but have not been well studied, of Gehler et al. (2004b) on filopodial retraction and whereas there is some controversial evidence that extension with caution; these may be the result of a MYO1C (and not myosin II) is responsible for retro- combination of changes in polymerization rate at the grade flow (Diefenbach et al., 2002). Although some tip and retrograde ‘reeling in’ of the filopodial actin evidence initially suggested that myosin V resists bundle with respect to the substrate (and growth filopodial retraction (Wang et al., 1996), other studies cone). However, although BDNF does signal through have refuted this and, instead, implicated myosin Va ROCK (Gehler et al., 2004b), which is known to affect in organelle transport (Evans et al., 1997); myosin myosin activity, there seems to be compensatory Vb and c are not well studied in the nervous system. signaling to limit the effects of BDNF to AC proteins. Myosin VI is also found in growth cones, but its physiol- We are led to this conclusion by the finding of Gehler ogical function has not been well studied. et al., (2004b) that enhancing AC activity (which Blebbistatin, a small molecule inhibitor of myosin should not affect myosin activation and, thus, II (Cheung et al., 2002) that works by indirectly inter- retrograde F-actin flow) was sufficient by itself to fering with actin binding (Kovacs et al., 2004), was reproduce the filopodial changes caused by BDNF recently assessed against non-muscle myosins and signaling. found to inhibit human non-muscle myosin II but not The balance of forces and relative motions of human myosins Ib, Va and X (Straight et al., 2003). growth cone cytoskeletal structures during different Thus, myosin VI cannot be ruled out as a possible behaviors is poorly understood. Certainly, tension mediator of blebbistatin-induced filopodial changes generation within growth cones is important, because observed by Gehler et al. (2004b). The observation treatment with the broader-acting and less-specific that blebbistatin increased filopodial length implies myosin inhibitor 2,3-butanedione monoxime (BDM) a reduction of retrograde flow of filopodial F-actin retards retrograde flow and leads to increased bundles, by inhibition of myosin IIa or b. Meanwhile, outgrowth (Lin et al., 1996). Retrograde flow occurs Gehler et al. (2004b) reported that filopodia detach- in the actin meshwork of lamellipodia (Forscher & ment and ‘bending and buckling’ increased in Smith, 1988; Welnhofer et al., 1997) and filopodia response to blebbistatin treatment. Detachment may (Mallavarapu & Mitchison, 1999) and must, ultimately, have been a result of loss of focal contact stabilization result from motor-based tension generation between via myosin IIa inhibition, whereas the bending and 180 J. Fass et al. buckling behavior suggests continued activity of Alternatively, a recent study found that growth some myosin class within the filopodia (perhaps cone turning on a uniform substrate was significantly myosin VI). Alternatively, this behavior may have sim- correlated with lamellipodial size and not with filopo- ply been an effect of random variation of myosin II dial length or number (Wang et al., 2003). Further- inhibition. more, preferential laser inactivation on one side of the growth cone of myosin Ic, but not myosin V, Differential regulation of AC proteins in interfered with turning behavior. Although this study turning behavior suggests a dominant role for both motor activation and lamellipodial area in growth cone turning, the Could differential AC regulation across the growth turning was not a result of guidance cue signaling. cone be a mechanism of achieving directional guid- Thus, it remains to be seen whether turning depends on ance? It has been observed previously that filopodial asymmetric motor protein activation or lamellipodial extension and retraction rates vary across growth protrusion during physiologically relevant guidance cones (Bray & Chapman, 1985; Mallavarapu & behavior. Mitchison, 1999) and that calcium signaling can result from contacts experienced by individual Conclusions filopodia (Gomez et al., 2001). The AC proteins can be regulated downstream of intracellular calcium Filopodia are dynamic structures and similar distri- (Meberg, 2000). In addition, individual filopodia can butions of lengths and numbers of filopodia could respond to contact with guidance cues (Smith, 1994; result from different regulatory processes. In order to Gallo & Letourneau, 2004) and filopodial sprouting more fully understand the regulation of filopodial has been shown to be necessary for turning in behavior, lifetime analysis of individual filopodia (i.e. response to a guidance cue (Zheng et al., 1996). ‘birth’ and ‘death’ times, length histories, locations) This suggests a potential link between AC activity, should be used to characterize the responses to filopodia and turning and, indeed, preliminary data different guidance cues. suggest that AC activity is increased concomitantly The results of Gehler et al. (2004b) document the with filopodial sprouting on the far side of a growth role of AC in the regulation of the length and number cone turning away from a negative guidance cue of growth cone filopodia; however, the mechanisms (Fig. 4). underlying this control remain unclear. Activation of AC may increase filopodial length either by increasing the G-actin pool, thus enhancing polymerization at filopodial tips, or by modifying the actin meshwork within the transitional zone, making motor-based retraction of filopodial F-actin bundles less efficient. Filopodial number may be regulated by AC proteins via enhancement of convergent elongation of lamellipodial F-actin filaments underlying the plasma membrane, thus making filopodial initiation more frequent. However, this mechanism has not yet been demonstrated fully in neurons. Further studies that use pho- tobleaching, photoactivation or speckle microscopy to track F-actin within filopodia and lamellipodia will be helpful in pinning down the mechanism by which AC regulates filopodial length and number. Figure 4. Ratio imaging of total cofilin to phosphorylated actin- depolymerizing factor (ADF)/cofilin (AC) in neuronal growth Acknowledgments cones. The upper panels show a typical growth cone in the absence of any externally applied guidance cue. The white/red We gratefully acknowledge grant support from the colors show the region of highest AC activity (lowest levels of National Institutes of Health (NIH; NS43115, J.N.F.; phosphorylated AC) normalized for every growth cone to a GM35126, NS40371, J.R.B.; HD19950, P.C.L.), the region in the distal neurite shaft where each immunofluorescent Alzheimer’s Association (IIRG-01–2730, J.R.B.), the image is adjusted to the same intensity (to correct for Christopher Reeve Paralysis Foundation (SB-0110–2, differences between antibody affinities). Bottom panels show a growth cone turning away from an aggrecan stripe, which P.D.S.), the National Science Foundation (NSF; contains rhodamine dextran for visualization; there is an IBN0080932, P.C.L.) and the Minnesota Medical accumulation of the active AC away from the repulsive Foundation (P.C.L.). S.G. is a trainee on an NIH Eye guidance cue. (P.D. Sarmiere, unpubl. data., 2004). Institute Training Grant. The authors thank Tatyana ADF/cofilin and filopodial dynamics 181

M. Svitkina, University of Pennsylvania, for kindly Endo M, Ohashi K, Sasaki Y et al. (2003) Control of growth cone providing unpublished data, and Chi Pak, Colorado motility and morphology by LIM kinase and Slingshot via State University, for helpful discussions. phosphorylation and dephosphorylation of cofilin. J Neurosci 23, 2527–37. Evans LL, Hammer J, Bridgman PC (1997) Subcellular localiza- References tion of myosin V in nerve growth cones and outgrowth from Amano M, Ito M, Kimura K et al. (1996) Phosphorylation and dilute-lethal neurons. J Cell Sci 110, 439–49. activation of myosin by Rho-associated kinase (Rho-kinase). Forscher P, Smith SJ (1988) Actions of cytochalasins on the J Biol Chem 271, 20 246–9. organization of actin filaments and microtubules in a neuronal Arber S, Barbayannis FA, Hanser H et al. (1998) Regulation of growth cone. J Cell Biol 107, 1505–16. actin dynamics through phosphorylation of cofilin by LIM- Gallo G, Letourneau PC (1998) Localized sources of neuro- kinase. Nature 393, 805–9. trophins initiate axon collateral sprouting. J Neurosci 18, Bamburg JR, McGough A, Ono S (1999) Putting a new twist on 5403–14. actin: ADF/cofilins modulate actin dynamics. Trends Cell Biol Gallo G, Letourneau PC (1999) Axon guidance: A balance of 9, 364–70. signals sets axons on the right track. Curr Biol 9, R490–2. Bray D, Chapman K (1985) Analysis of microspike movements Gallo G, Letourneau PC (2004) Regulation of growth cone actin on the neuronal growth cone. J Neurosci 5, 3204–13. filaments by guidance cues. J Neurobiol 58, 92–102. Bridgman PC (2002) Growth cones contain myosin II bipolar Gallo G, Lefcort FB, Letourneau PC (1997) The trkA receptor filament arrays. Cell Motil Cytoskeleton 52, 91–6. mediates growth cone turning toward a localized source of Bridgman PC, Dailey ME (1989) The organization of myosin nerve growth factor. J Neurosci 17, 5445–54. and actin in rapid frozen nerve growth cones. J Cell Biol 108, Gehler S, Gallo G, Veien E, Letourneau PC (2004a) p75 95–109. neurotrophin receptor signaling regulates growth cone Bridgman PC, Dave S, Asnes CF, Tullio AN, Adelstein RS (2001) filopodial dynamics through modulating RhoA activity. J Myosin IIB is required for growth cone motility. J Neurosci Neurosci 24, 4363–72. 21, 6159–69. Gehler S, Shaw AE, Sarmiere PD, Bamburg JR, Letourneau PC Brown J, Bridgman PC (2003a) Role of myosin II in axon (2004b) BDNF regulation of growth cone filopodial dynamics outgrowth. J Histochem Cytochem 51, 421–8. is mediated through ADF/cofilin. J Neurosci (in press). Brown ME, Bridgman PC (2003b) Retrograde flow rate is Ghosh M, Song X, Mouneimne G, Sidani M, Lawrence DS, increased in growth cones from myosin IIB knockout mice. Condeelis JS (2004) Cofilin promotes actin polymerization J Cell Sci 116, 1087–94. and defines the direction of cell motility. Science 304, 743–6. Brown ME, Bridgman PC (2004) Myosin function in nervous and Gohla A, Bokoch GM (2002) 14-3-3 regulates actin dynamics by sensory systems. J Neurobiol 58, 118–30. stabilizing phosphorylated cofilin. Curr Biol 12, 1704–10. Carlier MF, Laurent V, Santolini J et al. (1997) Actin depolymer- Goldberg DJ, Burmeister DW (1986) Stages in axon formation: izing factor (ADF/cofilin) enhances the rate of filament Observations of growth of Aplysia axons in culture using turnover: Implication in actin-based motility. J Cell Biol 136, video-enhanced contrast-differential interference contrast 1307–22. microscopy. J Cell Biol 103, 1921–31. Cheung A, Dantzig JA, Hollingworth S et al. (2002) A small- Gomez TM, Robles E, Poo M, Spitzer NC (2001) Filopodial molecule inhibitor of skeletal muscle myosin II. Nat Cell Biol calcium transients promote substrate-dependent growth 4, 83–8. cone turning. Science 291, 1983–7. Cohan CS, Welnhofer EA, Zhao L, Matsumura F, Yamashiro S Goodman CS, Shatz CJ (1993) Developmental mechanisms (2001) Role of the actin bundling protein fascin in growth that generate precise patterns of neuronal connectivity. Cell cone morphogenesis: Localization in filopodia and lamel- 72, 77–98. lipodia. Cell Motil Cytoskeleton 48, 109–20. Hariri AR, Goldberg TE, Mattay VS et al. (2003) Brain-derived Dawe HR, Minamide LS, Bamburg JR, Cramer LP (2003) ADF/ neurotrophic factor val66met polymorphism affects human cofilin controls cell polarity during fibroblast migration. Curr memory-related hippocampal activity and predicts memory Biol 13, 252–7. performance. J Neurosci 23, 6690–4. DesMarais V, Ichetovkin I, Condeelis J, Hitchcock-DeGregori SE Heidemann SR, Lamoureux P, Buxbaum RE (1990) Growth (2002) Spatial regulation of actin dynamics: A tropomyosin- cone behavior and production of traction force. J Cell Biol free, actin-rich compartment at the leading edge. J Cell Sci 111, 1949–57. 115, 4649–60. Hirose M, Ishizaki T, Watanabe N et al. (1998) Molecular DesMarais V, Macaluso F, Condeelis J, Bailly M (2004) Synergistic dissection of the Rho-associated protein kinase (p160ROCK)- interaction between the Arp2/3 complex and cofilin drives regulated neurite remodeling in neuroblastoma N1E-115 stimulated lamellipod extension. J Cell Sci 117, 3499–510. cells. J Cell Biol 141, 1625–36. Diefenbach TJ, Latham VM, Yimlamai D, Liu CA, Herman IM, Ichetovkin I, Han J, Pang KM, Knecht DA, Condeelis JS (2000) Jay DG (2002) Myosin 1c and myosin IIB serve opposing Actin filaments are severed by both native and recombinant roles in lamellipodial dynamics of the neuronal growth cone. dictyostelium cofilin but to different extents. Cell Motil J Cell Biol 158, 1207–17. Cytoskeleton 45, 293–306. Egan MF, Kojima M, Callicott JH et al. (2003) The BDNF Ichetovkin I, Grant W, Condeelis J (2002) Cofilin produces val66met polymorphism affects activity-dependent secretion newly polymerized actin filaments that are preferred for of BDNF and human memory and hippocampal function. dendritic nucleation by the Arp2/3 complex. Curr Biol 12, Cell 112, 257–69. 79–84. 182 J. Fass et al.

Jay DG (2000) The clutch hypothesis revisited: Ascribing the (1997) Xenopus actin depolymerizing factor/cofilin (XAC) is roles of actin-associated proteins in filopodial protrusion in responsible for the turnover of actin filaments in Listeria the nerve growth cone. J Neurobiol 44, 114–25. monocytogenes tails. J Cell Biol 136, 1323–32. Kimura K, Ito M, Amano M et al. (1996) Regulation of myosin Sabry JH, O’Connor TP, Evans L, Toroian-Raymond A, phosphatase by Rho and Rho-associated kinase (Rho-kinase). Kirschner M, Bentley D (1991) Microtubule behavior during Science 273, 245–8. guidance of pioneer neuron growth cones in situ. J Cell Biol Kovacs M, Toth J, Hetenyi C, Malnasi-Csizmadia A, Sellers J 115, 381–95. (2004) Mechanism of blebbistatin inhibition of myosin II. J Sala C (2002) Molecular regulation of dendritic spine shape and Biol Chem 279, 35 557–63. function. Neurosignals 11, 213–23. Lappalainen P, Drubin DG (1997) Cofilin promotes rapid actin Samstag Y, Nebl G (2003) Interaction of cofilin with the serine filament turnover in vivo. Nature 388, 78–82. phosphatases PP1 and PP2A in normal and neoplastic Lebrand C, Dent EW, Strasser GA et al. (2004) Critical role of human T lymphocytes. Adv Enzyme Regul 43, 197–211. Ena/VASP proteins for filopodia formation in neurons and in Sarmiere PD, Bamburg JR (2004) Regulation of the neuronal function downstream of netrin-1. Neuron 42, 37–49. actin cytoskeleton by ADF/cofilin. J Neurobiol 58, 103–17. Letourneau PC (1996) The cytoskeleton in nerve growth cone Schaefer AW, Kabir N, Forscher P (2002) Filopodia and actin motility and axonal pathfinding. Perspect Dev Neurosci 4, arcs guide the assembly and transport of two populations of 111–23. microtubules with unique dynamic parameters in neuronal Lin CH, Forscher P (1995) Growth cone advance is inversely growth cones. J Cell Biol 158, 139–52. proportional to retrograde F-actin flow. Neuron 14, 763–71. Shimada A, Mason CA, Morrison ME (1998) TrkB signaling Lin CH, Espreafico EM, Mooseker MS, Forscher P (1996) modulates spine density and morphology independent of Myosin drives retrograde F-actin flow in neuronal growth dendrite structure in cultured neonatal Purkinje cells. J cones. Neuron 16, 769–82. Neurosci 18, 8559–70. Mallavarapu A, Mitchison T (1999) Regulated actin cytoskeleton Smith CL (1994) Cytoskeletal movements and substrate interac- assembly at filopodium tips controls their extension and tions during initiation of neurite outgrowth by sympathetic retraction. J Cell Biol 146, 1097–106. neurons in vitro. J Neurosci 14, 384–98. Matsutani S, Yamamoto N (2004) Brain-derived neurotrophic Steketee MB, Tosney KW (2002) Three functionally distinct factor induces rapid morphological changes in dendritic adhesions in filopodia: Shaft adhesions control lamellar spines of olfactory bulb granule cells in cultured slices extension. J Neurosci 22, 8071–83. through the modulation of glutamatergic signaling. Neuro- Steketee M, Balazovich K, Tosney KW (2001) Filopodial initiation science 123, 695–702. and a novel filament-organizing center, the focal ring. Mol Meberg PJ (2000) Signal-regulated ADF/cofilin activity and Biol Cell 12, 2378–95. growth cone motility. Mol Neurobiol 21, 97–107. Straight AF, Cheung A, Limouze J et al. (2003) Dissecting Meberg PJ, Bamburg JR (2000) Increase in neurite outgrowth temporal and spatial control of cytokinesis with a myosin II mediated by overexpression of actin depolymerizing factor. inhibitor. Science 299, 1743–7. J Neurosci 20, 2459–69. Strasser GA, Rahim NA, VanderWaal KE, Gertler FB, Lanier LM Meberg PJ, Ono S, Minamide LS, Takahashi M, Bamburg JR (2004) Arp2/3 is a negative regulator of growth cone translo- (1998) Actin depolymerizing factor and cofilin phosphoryla- cation. Neuron 43, 81–94. tion dynamics: Response to signals that regulate neurite Suter DM, Forscher P (2000) Substrate–cytoskeletal coupling as extension. Cell Motil Cytoskeleton 39, 172–90. a mechanism for the regulation of growth cone motility and Mitchison T, Kirschner M (1988) Cytoskeletal dynamics and guidance. J Neurobiol 44, 97–113. nerve growth. Neuron 1, 761–72. Suter DM, Errante LD, Belotserkovsky V, Forscher P (1998) The Mogilner A, Oster G (2003) Force generation by actin polymer- Ig superfamily cell adhesion molecule, apCAM, mediates ization II: The elastic ratchet and tethered filaments. Biophys growth cone steering by substrate-cytoskeletal coupling. J J 84, 1591–605. Cell Biol 141, 227–40. Nagata-Ohashi K, Ohta Y, Goto K et al. (2004) A pathway of Svitkina TM, Bulanova EA, Chaga OY et al. (2003) Mechanism neuregulin-induced activation of cofilin-phosphatase Sling- of filopodia initiation by reorganization of a dendritic network. shot and cofilin in lamellipodia. J Cell Biol 165, 465–71. J Cell Biol 160, 409–21. Niwa R, Nagata-Ohashi K, Takeichi M, Mizuno K, Uemura T Theriot JA (1997) Accelerating on a treadmill: ADF/cofilin (2002) Control of actin reorganization by Slingshot, a family promotes rapid actin filament turnover in the dynamic of phosphatases that dephosphorylate ADF/cofilin. Cell 108, cytoskeleton. J Cell Biol 136, 1165–8. 233–46. Toshima J, Toshima JY, Amano T, Yang N, Narumiya S, Mizuno K Ohta Y, Kousaka K, Nagata-Ohashi K et al. (2003) Differential (2001a) Cofilin phosphorylation by protein kinase testicular activities, subcellular distribution and tissue expression pat- protein kinase 1 and its role in integrin-mediated actin terns of three members of Slingshot family phosphatases reorganization and focal adhesion formation. Mol Biol Cell that dephosphorylate cofilin. Genes Cells 8, 811–24. 12, 1131–45. Okabe S, Hirokawa N (1991) Actin dynamics in growth cones. J Toshima J, Toshima JY, Takeuchi K, Mori R, Mizuno K (2001b) Neurosci 11, 1918–29. Cofilin phosphorylation and actin reorganization activities of Pollard TD, Blanchoin L, Mullins RD (2000) Molecular mecha- testicular protein kinase 2 and its predominant expression in nisms controlling actin filament dynamics in nonmuscle testicular Sertoli cells. J Biol Chem 276, 31 449–58. cells. Annu Rev Biophys Biomol Struct 29, 545–76. Upadhyaya A, van Oudenaarden A (2004) Actin polymerization: Rosenblatt J, Agnew BJ, Abe H, Bamburg JR, Mitchison TJ Forcing flat faces forward. Curr Biol 14, R467–9. ADF/cofilin and filopodial dynamics 183

Wang FS, Wolenski JS, Cheney RE, Mooseker MS, Jay DG Wylie SR, Wu P-J, Patel H, Chantler PD (1998) A conventional (1996) Function of myosin-V in filopodial extension of myosin motor drives neurite outgrowth. Proc Natl Acad Sci neuronal growth cones. Science 273, 660–3. USA 95, 12 967–72. Wang FS, Liu CW, Diefenbach TJ, Jay DG (2003) Modeling the Yang N, Higuchi O, Ohashi K et al. (1998) Cofilin phosphoryla- role of myosin 1c in neuronal growth cone turning. Biophys tion by LIM-kinase 1 and its role in Rac-mediated actin J 85, 3319–28. reorganization. Nature 393, 809–12. Welnhofer EA, Zhao L, Cohan CS (1997) Actin dynamics and Zheng JQ, Wan JJ, Poo MM (1996) Essential role of filopodia in organization during growth cone morphogenesis in Helisoma chemotropic turning of nerve growth cone induced by a neurons. Cell Motil Cytoskeleton 37, 54–71. glutamate gradient. J Neurosci 16, 1140–9. Wylie SR, Chantler PD (2001) Separate but linked functions Zhou FQ, Cohan CS (2004) How actin filaments and microtu- of conventional myosins modulate adhesion and neurite bules steer growth cones to their targets. J Neurobiol 58, outgrowth. Nat Cell Biol 3, 88–92. 84–91.