R

I

SN

a

AA

KIICFMF

C

1d

Seminars in Cell & Developmental Biology 21 (2010) 350–356

Contents lists available at ScienceDirect

Seminars in Cell & Developmental Biology

journa l homepage: www.e lsev ier .com/ locate /semcdb

eview

-BAR domains, IRSp53 and filopodium formation

ohail Ahmed ∗, Wah Ing Goh, Wenyu Bueural Stem Cell Laboratory, Institute of Medical Biology, 8A Biomedical Grove, #05-37 Immunos, Singapore 138648, Singapore

r t i c l e i n f o

rticle history:vailable online 11 November 2009

eywords:

a b s t r a c t

Filopodia and lamellipodia are dynamic actin-based structures that determine cell shape and migration.Filopodia are thought to sense the environment and direct processes such as axon guidance and neu-rite outgrowth. Cdc42 is a small GTP-binding protein and member of the RhoGTPase family. Cdc42 and

-BARRSp53dc42-actinembrane protrusion

ilopodia

its effector IRSp53 (insulin receptor phosphotyrosine 53 kDa substrate) have been shown to be stronginducers of filopodium formation. IRSp53 consists of an I-BAR (inverse-Bin-Amphiphysin-Rvs) domain, aCdc42-binding domain and an SH3 domain. The I-BAR domain of IRSp53 induces membrane tubulationof vesicles and dynamic membrane protrusions lacking actin in cells. The IRSp53 SH3 domain inter-acts with proteins that regulate actin filament formation e.g. Mena, N-WASP, mDia1 and Eps8. In thisreview we suggest that the mechanism for Cdc42-driven filopodium formation involves coupling I-BAR

domain-induced membrane protrusion with SH3 domain-mediated actin dynamics through IRSp53.

© 2009 Elsevier Ltd. All rights reserved.

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3501.1. Cell morphogenesis and migration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3501.2. Lamellipodia and filopodia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3511.3. Aims of the review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351

2. Filopodia: diversity, form and composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3513. Definition of mammalian filopodia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3514. Small GTPases as regulators of filopodium formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351

4.1. Cdc42 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3514.2. Rif . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352

5. IRSp53 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3525.1. Domain organisation and binding partners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3525.2. IRSp53 protein family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3525.3. Cellular function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352

6. I-BAR domains and membrane protrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3527. Actin polymerisation in filopodium formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3548. A mechanism for filopodium formation: coupling membrane protrusion and actin polymerisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3549. Filopodial dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355

10. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

∗ Corresponding author. Tel.: +65 6407 0165; fax: +65 6464 2048.E-mail address: sohail.ahmed@imb.a-star.edu.sg (S. Ahmed).

084-9521/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.oi:10.1016/j.semcdb.2009.11.008

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356

1. Introduction

1.1. Cell morphogenesis and migration

Cells form the fundamental building blocks of all living mat-ter [1]. Thus understanding the form and function of cells willhelp to reveal the complex biology of tissues, and ultimately wholeorganisms. Cardinal features of cells are their shape or morphology,

evelopmental Biology 21 (2010) 350–356 351

aeact

1

acithc

sfiasswatigdp[

1

cnfiptd

2

aobafftstDtfbattbpte

Table 1Morphological characteristics of mammalian filopodia.

Cell line/cDNA transfection Length (�m) Lifetime (s)

EndogenousN1E115 15 142HeLa 14 131Cos7 10 123B16F1 8 79CHO ND ND

TransfectedCdc42V12, Rac1N17 (fibroblasts) 8.4 157IRSp53 (N1E115) 6.8 187N-WASP (N1E115) 7.4 154Toca-1 (N1E115)a 6.6 128

S. Ahmed et al. / Seminars in Cell & D

nd their ability to migrate. Disease states such as neurodegen-ration and cancer can be linked to defects in cell morphologynd migration. Two cellular compartments, the membrane and theytoskeleton, play pivotal roles in regulating cell shape and migra-ion.

.2. Lamellipodia and filopodia

The actin-based structures at the leading edge – lamellipodiand filopodia – determine cell shape and ability to migrate. Motileells put forward thin, sheet-like protrusive structures at their lead-ng edge as they crawl across the substratum. The region closest tohe leading edge is referred to as the lamellipodium. It is made up ofighly branched dendritic microfilaments assembled by the Arp2/3omplex [2].

Filopodia are membrane-based actin-rich finger-like protru-ions that are highly dynamic. Filopodia extend and retract rapidlyrom the cell surface as the cell explores its environment, seek-ng biological cues. Their movements are not limited to extensionnd retraction in the horizontal plane; filopodia are also able towing laterally, as well as lift up away from and down towards theubstratum in the vertical plane. Ultimately they form adhesionsith the matrix, facilitating lamellae to fill gaps between them

nd move the cell forward. Filopodia are thought to play impor-ant roles in a number of cellular and developmental processes,ncluding (i) neuritogenesis [3,4], (ii) axon guidance in neuronalrowth cones [5–7], (iii) receptor–ligand endocytosis [8,9], (iv)engue virus uptake [10], (v) detection of pathogen targets forhagocytosis [11] and (vi) dorsal closure in Drosophila embryos12].

.3. Aims of the review

Here, we show how studies of the I-BAR and SH3 domain-ontaining protein IRSp53 have begun to reveal a mecha-ism of filopodium formation. Essential to this model forlopodium formation is that the Cdc42–IRSp53 effector com-lex allows the coupling of membrane protrusion (driven byhe I-BAR domain) to actin dynamics (mediated by the SH3omain).

. Filopodia: diversity, form and composition

In mammalian cells, each individual filopodium is made up ofcylindrical plasma membrane extension enclosing a tight bundlef 15–20 linear actin filaments all oriented in parallel, with theirarbed ends distal from the cell body [13]. In addition to actin fil-ments, a number of proteins are associated with filopodia. Theormin Dia2 (Diaphanous 2) nucleates actin filaments and has beenound in both mammalian and Dictyostelium filopodia, includinghe tips of these structures [14,15]. Ena/VASP (enabled/vasodilator-timulated phosphoprotein) proteins are also found at filopodialips. Ena/VASP and Mena (mouse Ena), together with mDia2 (mouseia2), have been proposed to form a ‘tip complex’. Ena/VASP facili-

ates the barbed end growth of actin filaments by protecting themrom capping proteins (reviewed in [11]). Myosin X is a VASP-inding protein that localises to both the tips and shafts of filopodia,

nd appears to transport Ena/VASP and other components to theips, using its motor domain that travels along actin filamentsowards their barbed ends [16]. Along the filopodial shaft, the actin-undling protein fascin crosslinks individual actin filaments as theyolymerise. This gives rise to stiff bundles that are rigid enough noto buckle when they push against the membrane as the filopodiumxtends [11].

Rif (N1E115)b 4.4 155

ND: not detected. Filopodial width 0.6–1.2 �m. Data from [31], except a[53] andb(Goh and Ahmed, unpublished observations).

3. Definition of mammalian filopodia

At the outset of this review it is important to define the structureand dynamics of filopodia so that results from different laboratoriescan be compared and their discrete features investigated. Mam-malian filopodia can be followed in cell culture using time-lapsemicroscopy. Widefield dual channel fluorescence microscopy usingsensitive CCD (charged-coupled device) cameras are an ideal set-upto follow filopodia. Individual frames can be acquired in the rangeof 100 ms each and at a rate of six frames a minute giving a totalof 600 frames over a 10 min time course. Generally speaking, thisformat allows dynamic data of filopodia to be acquired withoutsignificant bleaching or toxicity. We have followed filopodium for-mation in a variety of cell types using GFP-actin to label dynamicactin structures and DIC microscopy to track changes in cell mor-phology. The use of GFP-actin allows observation of structuresin real time. From time-lapse analysis of endogenous structuresin N1E115, HeLa, Cos7 and B16F1 cells, the following features offilopodia were determined: lifetime, length, width and morphol-ogy (Table 1). Mammalian filopodia are rarely longer than 15 �mand have a lifetime of 79–142 s. They have a uniform width of0.6–1.2 �m along their length, and are never branched or taperedin appearance. Filopodia usually emerge from the cell periphery orleading edge individually and never in clusters. In contrast, retrac-tion fibres are non-dynamic, tapered in appearance and are foundin clusters.

4. Small GTPases as regulators of filopodium formation

Several members of the Ras superfamily of small GTPases havebeen linked to filopodium formation, with strongest evidence hav-ing emerged for Cdc42 [17], Rif (Rho in filopodia) [18] and Rab35(Ras-like protein in brain 35) [19]. Apart from these, other smallGTPases implicated in filopodium formation include RalA (Ras-likeA) [20], TC10, Wrch-1 (Wnt-1 responsive Cdc42 homologue-1) andWrch-2 (Wnt-1 responsive Cdc42 homologue-2) [17].

4.1. Cdc42

In 1995, two studies reported that Cdc42 could regulatethe formation of filopodia in mammalian cells. The first studyshowed that bradykinin could activate Cdc42 leading to three dis-tinct morphological effects – filopodium formation, Rac-mediated

lamellipodium formation and inactivation of RhoA [21]. In the otherstudy, Nobes and coworkers demonstrated that Cdc42 inducedfocal complexes associated with filopodia [22]. In a variety of mam-malian cells Cdc42 dominant negative protein inhibits filopodiumformation [21–24]. Cdc42 functions as a molecular switch by reg-

3 evelop

uppat

4

a1atttaicvt

ia(bRCfcfi

5

5

aaas(o

lit[amIft[

5

t(oSamWi

likely to be an in vitro artefact [33,40]. In fact, even at high saltconcentrations, the F-actin bundling activity of the IRSp53 I-BARdomain is poor and at least 83-fold weaker than fascin, a well-known F-actin bundling protein [31].

52 S. Ahmed et al. / Seminars in Cell & D

lating the formation and disassembly of protein complexes at thelasma membrane. It achieves this by alternating between the cyto-lasm, where it is inactive, and the plasma membrane, where it isctive. Cdc42 activation and cellular localisation is controlled byhe RhoGTPase activation cycle (reviewed in [25]).

.2. Rif

Rif is a novel Rho family GTPase that shares only 32–49% aminocid sequence identity with other family members. The N-terminal9 residues of Rif bear no homology to other Rho family GTPases,nd attachment of an epitope tag to this end of the protein appearedo interfere with the full phenotype, hence this region is believedo play a role in cellular localisation [18]. The untagged Rif pro-ein induces filopodia in fibroblast cells; however these structuresppear somewhat different from those induced by Cdc42. The Rif-nduced filopodia are shorter (Fig. 1A), project not only from theell periphery but also the apical surface, and their tips lack theinculin-rich focal complexes found in their Cdc42-induced coun-erparts.

To date, the formin mDia2 is the sole putative effector of Rifdentified thus far. Although mDia2 contains a CRIB domain, Rif isble to bind it even when the conserved histidine within the motifamino acid residue 160) is mutated. Furthermore, this Cdc42-inding defective mutant of mDia2 could, together with activatedif, induces filopodia in fibroblast cells. Sequestration of eitherdc42 or the Arp2/3 complex did not affect the ability of Rif to

orm filopodia, and this leads to the hypothesis that Rif and mDia2onstitute a novel Cdc42- and Arp2/3-independent pathway oflopodium formation [14].

. IRSp53

.1. Domain organisation and binding partners

Also known as BAIAP2 (brain-specific angiogenesis inhibitor-1ssociated protein 2), IRSp53 is composed of three main domains –n N-terminal I-BAR domain, followed by a partial CRIB domain andn SH3 domain. At the extreme C-terminal there exists a PDZ (post-ynaptic density 95, disc large, zonula occludens-1) and/or a WH2WASP homology 2) domain. The isoforms and tissue distributionf IRSp53 have been reviewed by Scita et al. [26].

IRSp53 was found to bind Cdc42 [23], Rac1 [27] and the polypro-ine region of WAVE1 [28] in yeast two-hybrid screens. Thesenteractions have since been mapped to the partial CRIB domain,he I-BAR domain, and the SH3 domain of IRSp53 respectively23,28,29]. Apart from WAVE1, the IRSp53 SH3 domain also inter-cts with other regulators of actin dynamics, namely WAVE2, Mena,Dia1, Dynamin1 [30,31], Eps8 and N-WASP (see [26] for review of

RSp53 and its binding partners). The 14-3-3 protein has also beenound to bind IRSp53, and this interaction may well serve to keephe protein inactive in the cytoplasm before activation by Cdc4232].

.2. IRSp53 protein family

The IRSp53 family of proteins comprises IRTKS (insulin recep-or tyrosine kinase substrate; also known as BAIAP2L1), MIM/ABBAmissing in metastasis/actin-bundling protein with BAIAP2 homol-gy), and FLJ22582 (BAIAP2L2). Of these, only MIM/ABBA lacks an

H3 domain, suggesting that it regulates actin dynamics by a mech-nism that is either different from that employed by other familyembers, or possibly absent. However, MIM/ABBA does contain aH2 actin monomer binding domain [33]. IRSp53 is unique in that

t is the only family member that possesses a Cdc42-binding site.

mental Biology 21 (2010) 350–356

This partial CRIB domain of IRSp53 binds Cdc42 specifically andwith high affinity in the nanomolar range [23].

5.3. Cellular function

In N1E115 mouse neuroblastoma cells, IRSp53 induces neuritesthat are branched and complex. The neurite complexity comprisesextensive filopodium and lamellipodium formation and membraneruffling [23]. Cdc42 interaction with IRSp53 is essential for thesephenotypes as a mutant of IRSp53 that is unable to bind Cdc42cannot induce such morphological changes. This IRSp53 mutantalso does not colocalise with F-actin. Mutation of the four lysineresidues (142, 143, 146 and 147) in the I-BAR domain, which areinvolved in binding to actin, also results in a loss of the IRSp53phenotype [31]. The FP/AA mutation of the IRSp53 SH3 domain pre-vents interaction with binding partners, but this mutant is still ableto induce neurites. Interestingly, the neurites generated by IRSp53-FP/AA lack complexity, with filopodia absent from most of them.These results reveal the importance of the IRSp53 SH3 domain ingenerating filopodia.

6. I-BAR domains and membrane protrusion

The BAR (Bin-Amphiphysin-Rvs) domain is a highly conservedprotein domain involved in remodelling of cellular membranes.BAR domains form dimers with natural curvature and can inducecurvature of membranes that they bind to. There are three distinctfamilies of BAR domain-containing proteins: classical BAR, F-BAR(Fer/CIP4-homology-BAR) and I-BAR. Each BAR domain dimer caninduce distinct degrees of membrane curvature depending on itsshape. For example, the F-BAR domains induce curvature consistentwith the size of endocytic vesicles and they do this by end to endoligomerisation [34–36]. The I-BAR domain was first identified inIRSp53 based on sequence homology [37]. However, the sequencehomology was weak and arguably marginal. Crystal structure anal-ysis of the N-terminal 250 amino acid residues of IRSp53 revealedstrong structural similarity to the BAR domain family; the domainformed a dimer with six helices, with a distinct, rather flat, cigarshaped curvature [38] (Fig. 1B). The I-BAR domain was named IMD(IRSp53/MIM homology domain) because two proteins, IRSp53 andMIM, possessed this domain. For simplicity, in this review we willrefer to the N-terminal sequences of both IRSp53 and MIM as I-BAR domains. A series of charged residues in the IRSp53 I-BARdomain were identified as possible actin binding sites. This raisedthe possibility that the I-BAR domain could bundle F-actin. Indeed,when the I-BAR domain of IRSp53 was expressed in cells, filopodia-like structures containing actin were seen. Furthermore, the I-BARdomain could bundle F-actin in vitro1 [38,39]. In both of these stud-ies, examining the effect of the I-BAR domain on cells revealedthe formation of membrane protrusions containing actin. How-ever, the structures did not resemble filopodia and whether thesewere dynamic structures was not evaluated by Millard et al. [38]or Yamagishi et al. [39] (see Section 3 for definition of mammalianfilopodia). Subsequent analysis of the F-actin bundling activity ofthe I-BAR domain revealed that the original experiments were car-ried out under non-physiological salt concentrations and were thus

1 Protein activities of actin filament regulators (such as actin filament capping,actin filament elongation or F-actin bundling) observed in vitro do not necessarilymean that the same activities will be seen in vivo. It is therefore important to treatin vitro protein activities of actin filament regulators with caution.

S. Ahmed et al. / Seminars in Cell & Developmental Biology 21 (2010) 350–356 353

F am (ts

fipmsttdCwadd(IBr

ig. 1. I-BAR domains and membrane protrusion. Shown below the ribbon diagrchematic of the I-BAR domain shape.

Our work demonstrating that overexpression of IRSp53 inducedlopodia in mammalian cells also found that some of therotrusions generated did not contain actin and were thusembrane-only structures [23]. This initial observation was sub-

tantiated and developed in [29]. The Takenawa group was the firsto report that the I-BAR (termed the RCB or Rac-binding domain inhat study) domain of IRSp53 (and MIM) could bind phospholipids,eform liposomes and generate membrane protrusions in A439 andos7 cells [29]. Low affinity binding of Rac-GTP to the I-BAR domainas also detected and postulated to be essential for I-BAR domain

ctivity. The true functional nature of the IRSp53 and MIM I-BARomain was revealed when Mattila and coworkers showed that this

omain could generate 40–90 nm tubules from lipid vesicles [33]Fig. 1C). Furthermore, the membrane deformation induced by the-BAR domain was found to be opposite to that induced by classicalAR or F-BAR domains, and consistent with membrane protrusionather than membrane invagination. Interestingly, they found no

aken from website: http://www.endocytosis.org/F-BAR proteins/I-BAR.html) is a

evidence for an essential role for Rac in the membrane deformingactivity of the I-BAR domain [33]. Careful examination of the effectof the I-BAR domain of IRSp53 on mammalian cells was done bycoexpressing GFP-tagged I-BAR domain with mRFP- or mCherry-labelled actin and studying the structures formed by dual channeltime-lapse fluorescence microscopy. It was found that the I-BARdomain generates three types of protrusions – static aberrantlyshaped membrane protrusions with aggregates of actin (type 1),static membrane protrusions lacking actin (type 2), and dynamicmembrane protrusions lacking actin (type 3) [31] (Fig. 1A). Thetype 1 structures are similar to the protrusions described in earlierstudies [38,39] (Fig. 1A).

It is now clear that the I-BAR domain functions primar-ily to deform membranes and that this activity is essential forfilopodium formation. Saarikangas and coworkers recently com-pared the relative abilities of the I-BAR domains of various IRSp53family members to deform giant unilamellar vesicles containing

354 S. Ahmed et al. / Seminars in Cell & Developmental Biology 21 (2010) 350–356

ia the

flfC7ti(be

BdcHm3ogFdebABF

7

fitfisbfia

which exists in an autoinhibited state in the cytoplasm, to theplasma membrane. IRSp53 dimers become active on the plasmamembrane, with their I-BAR domains deforming the membranewhile their SH3 domains increase the concentration of actin fil-ament regulators such as Mena (Ena/VASP), N-WASP, mDia12 and

Fig. 2. Actin filament formation v

uorescence-labelled lipids, as well as multilamellar vesicles. Theyound that: (i) the I-BAR domains of IRSp53, MIM, IRTKS, ABBA and. elegans MIM induce tubules of distinct sizes ranging from 30 to0 nm, (ii) ABBA and MIM are more active in deforming membraneshan IRSp53 and IRTKS, and do so by a distinct mechanism – bynserting an N-terminal amphipathic helix into the membrane, andiii) there is dynamic interchange of I-BAR domains on/off mem-ranes with rate of 8 s (MIM/ABBA), 5 s (IRSp53/IRTKS) and 86 s (C.legans BAR) [41].

Do I-BAR domains generate filopodia? It is evident that the I-AR domain when overexpressed in mammalian cells causes aramatic induction of membrane protrusions, which in some casesontain actin [33,38,39] and putative ‘markers’ of filopodia [41].owever, to represent physiologically relevant structures, mam-alian filopodia must possess the features described in Section

. Central amongst these features is the lifetime and morphol-gy of filopodia. A direct side by side comparison of structuresenerated by the I-BAR domain and IRSp53 was recently made.rom this analysis it was concluded that the I-BAR domainoes not generate filopodia [31]. In particular, structures gen-rated by I-BAR had aberrant actin aggregates as opposed toundled microfilaments, and had lifetimes of more than 10 min.

comparison of the features of structures generated by the I-AR domain and IRSp53 in mammalian cells is presented inig. 1A.

. Actin polymerisation in filopodium formation

Lamellipodia have been postulated to be prerequisite forlopodium formation, as filopodia were observed to arise fromhem [42]. This led to the ‘convergent elongation model’ of

lopodium formation, whereby filopodial tip complex proteinsuch as Ena/VASP and Dia2 bring together and protect thearbed ends of selected Arp2/3-nucleated lamellipodial micro-laments from capping proteins. These microfilaments are thusble to continue elongating to form long, unbranched fila-

SH3 domain partners of IRSp53.

ments that are bundled together by fascin to give rise to afilopodium [43,44]. However, filopodia were found to form evenin the absence of lamellipodia, when proteins essential for lamel-lipodium formation such as the Arp2/3 complex were depletedby sequestration [14] or knockdown [45]. Coupled with theobservations that Dia2 localised to filopodial tips, and that Dic-tyostelium Dia2-knockout cells failed to produce filopodia, anArp2/3-independent, formin-driven ‘de novo filament nucleation’model emerged as an alternative. In this model, a formin-containingfilopodial tip complex assembles at a site on the plasma mem-brane, and nucleates unbranched actin filaments that are bundledtogether as they elongate to form a filopodium (reviewed in [46])(Fig. 2).

8. A mechanism for filopodium formation: couplingmembrane protrusion and actin polymerisation

We now integrate the components that have been discussedduring the course of this review – Cdc42, IRSp53, membrane protru-sion (induced by I-BAR domain) and actin polymerisation (drivenby SH3 domain binding partners) – into a model for filopodiumformation. The process starts with the activation of Cdc42 viamembrane-bound receptors and exchange factors. Active Cdc42forms ‘hotspots’ on the plasma membrane for recruitment of pro-teins that are necessary to build filopodia. Cdc42 recruits IRSp53,

2 We have found that mDia1, but not mDia2, induces and synergises with IRSp53in filopodium formation. FRET (fluorescence resonance energy transfer) experi-ments show that IRSp53 and mDia1 interact in filopodia (Lim, Sudhaharan andAhmed, unpublished data).

S. Ahmed et al. / Seminars in Cell & Developmental Biology 21 (2010) 350–356 355

Fig. 3. Mechanism of filopodium formation.

s indi

Emaafi

9

dcdUroifloi(t

Fig. 4. Model for filopodial dynamics. Orange outline

ps8. The I-BAR domain of IRSp53 has an actin-binding site and thisay be used to localise actin filaments to sites where membranes

re being deformed into tubular protrusions. Membrane protrusionnd actin filament formation become coupled through IRSp53 andlopodia emerge from the cell periphery (Fig. 3).

. Filopodial dynamics

Actin assembly at tips and retrograde flow can be used toescribe filopodial dynamics. Mallavarapu and Mitchison usedaged rhodamine-labelled actin and GFP-actin to examine filopo-ial growth and retraction in the growth cones of NG108 cells.sing these probes they measured rates of actin assembly and ret-

ograde flow throughout the lifetimes of filopodia. In the majorityf filopodia examined assembly rates at the tip were dominantn determining filopodial extension or retraction, with retrograde

ow rates remaining relatively constant [47]. Developing thesebservations further we propose that IRSp53, in addition to induc-ng membrane protrusion, serves to localise Mena [27], mDia1 [30]Lim and Ahmed, unpublished observations), and Eps8 [48,49] tohe tip complex. Mena, mDia1 and Eps8 have anti-capping, fila-

cate inactivation. Black outlines indicate dominance.

ment elongation and capping activities, respectively (Fig. 4). Wesuggest that the synergistic or competing activities of these threeproteins could determine the degree of actin polymerisation. Sig-nalling pathways can modify the activities of Mena, mDia1 andEps8 through phosphorylation or changing their cellular localisa-tion. Mena is known to be a target of protein kinases A and G [50].Interestingly, a recent study has revealed that the phosphorylationof Eps8 at MAPK sites influences its capping activity downstream ofBDNF (brain derived neurotrophic factor) [51]. We also know thatmDia1 binds RhoA and this interaction could influence its localisa-tion. Further support for a role for capping proteins in determiningfilopodial dynamics comes from computer simulations of actinpolymerisation. The results suggest that filament capping activityinduces macroscopic instability in filopodial dynamics consistentwith filopodial retraction rates and lifetimes [52].

10. Conclusions

The discovery of the I-BAR domain in IRSp53 and its abil-ity to induce membrane protrusion has begun a new chapterin understanding filopodium formation. The Cdc42–IRSp53 effec-

3 evelop

tdtmdmria

A

bt

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[[

[

[

[

[

56 S. Ahmed et al. / Seminars in Cell & D

or complex initiates filopodium formation. It is likely that I-BARomain oligomerisation is important for membrane protrusion andhe recruitment of SH3 domain partners for actin filament for-

ation. Once formed, filopodia can either become stabilised orisassemble. Filopodium stabilisation is connected with the for-ation of focal complexes and adhesion to the substrate. In future

esearch it will be important to investigate how filopodial dynam-cs, stabilisation and disassembly are regulated and which proteinsnd signalling pathways are involved.

cknowledgements

We would like to thank members of the SA lab for their contri-ution over the years in the IRSp53 project. We would also like tohank A*STAR for supporting this research.

eferences

[1] Wolpert L. Evolution of the cell theory. Philos Trans R Soc Lond B Biol Sci1995;349:227–33.

[2] Chhabra ES, Higgs HN. The many faces of actin: matching assembly factors withcellular structures. Nat Cell Biol 2007;9(10):1110–21.

[3] Dent EW, Kwiatkowski AV, Mebane LM, Philippar U, Barzik M, RubinsonDA, et al. Filopodia are required for cortical neurite initiation. Nat Cell Biol2007;9(12):1347–59.

[4] Kwiatkowski AV, Rubinson DA, Dent EW, Edward van Veen J, Leslie JD, Zhang J,et al. Ena/VASP is required for neuritogenesis in the developing cortex. Neuron2007;56:441–55.

[5] O’Connor TP, Duerr JS, Bentley D. Pioneer growth cone steering decisions medi-ated by single filopodial contacts in situ. J Neurosci 1990;10(12):3935–46.

[6] Chien C-B, Rosenthal DE, Harris WA, Holt CE. Navigational errors made bygrowth cones without filopodia in the embryonic Xenopus brain. Neuron1993;11:237–51.

[7] Zheng JQ, Wan J-J, Poo M-M. Essential role of filopodia in chemotropicturning of nerve growth cone induced by a glutamate gradient. J Neurosci1996;16(3):1140–9.

[8] Lidke DS, Nagy P, Heintzmann R, Arndt-Jovin DJ, Post JN, Grecco HE, et al. Quan-tum dot ligands provide new insights into erbB/HER receptor-mediated signaltransduction. Nat Biotechnol 2004;22:198–203.

[9] Lidke DS, Lidke KA, Rieger B, Jovin TM, Arndt-Jovin DJ. Reaching out for signals:filopodia sense EGF and respond by directed retrograde transport of activatedreceptors. J Cell Biol 2005;170:619–26.

10] Zamudio-Meza H, Castillo AM, Gonzalez-Bonilla C, Meza I. Rac1 and Cdc42GTPases cross-talk regulates formation of filopodia required for Dengue virustype-2 entry into HMEC-1 cells. J Gen Virol 2009;90(12):2902–11.

11] Mattila PK, Lappalainen P. Filopodia: molecular architecture and cellular func-tions. Nat Rev Mol Cell Biol 2008;9(6):446–54.

12] Millard TH, Martin P. Dynamic analysis of filopodial interactions during thezippering phase of Drosophila dorsal closure. Development 2008;135(4):621–6.

13] Le Clainche C, Carlier M-F. Regulation of actin assembly associated with pro-trusion and adhesion in cell migration. Physiol Rev 2008;88:489–513.

14] Pellegrin S, Mellor H. The Rho family GTPase Rif induces filopodia throughmDia2. Curr Biol 2005;15:129–33.

15] Schirenbeck A, Bretschneider T, Arasada R, Schleicher M, Faix J. TheDiaphanous-related formin dDia2 is required for the formation and mainte-nance of filopodia. Nat Cell Biol 2005;7:619–25.

16] Tokuo H, Ikebe M. Myosin X transports Mena/VASP to the tip of filopodia.Biochem Biophys Res Commun 2004;319(1):214–20.

17] Sorokina EM, Chernoff J. Rho-GTPases: new members, new pathways. J CellBiochem 2005;94:225–31.

18] Ellis S, Mellor H. The novel Rho-family GTPase Rif regulates coordinated actin-based membrane rearrangements. Curr Biol 2007;10:1387–90.

19] Zhang J, Fonovic M, Suyama K, Bogyo M, Scott MP. Rab35 controls actin bundlingby recruiting fascin as an effector protein. Science 2009;325(5945):1250–4.

20] Ohta Y, Suzuki N, Nakamura S, Hartwig JH, Stossel TP. The small GTPase RalAtargets filamin to induce filopodia. Proc Natl Acad Sci USA 1999;96(5):2122–8.

21] Kozma R, Ahmed S, Best A, Lim L. The Ras-related protein Cdc42Hs andbradykinin promote formation of peripheral actin microspikes and filopodiain Swiss 3T3 fibroblasts. Mol Cell Biol 1995;15(4):1942–52.

22] Nobes CD, Hall A. Rho, rac, and cdc42 GTPases regulate the assembly of mul-timolecular focal complexes associated with actin stress fibers, lamellipodia,and filopodia. Cell 1995;81(1):53–62.

23] Govind S, Kozma R, Monfries C, Lim L, Ahmed S. Cdc42hs facilitates cytoskeletalreorganization and neurite outgrowth by localizing the 58-Kd insulin receptorsubstrate to filamentous actin. J Cell Biol 2001;152(3):579–94.

[

[

mental Biology 21 (2010) 350–356

24] Allen WE, Jones GE, Pollard JW, Ridley AJ. Rho, Rac and Cdc42 regulate actinorganization and cell adhesion in macrophages. J Cell Sci 1997;110:707–20.

25] Bustelo XR, Sauzeau V, Berenjeno IM. GTP-binding proteins of the Rho/Racfamily: regulation, effectors and functions in vivo. BioEssays 2007;29:356–70.

26] Scita G, Confalonieri S, Lappalainen P, Suetsugu S. IRSp53: crossing the roadof membrane and actin dynamics in the formation of membrane protrusions.Trends Cell Biol 2008;18(2):52–60.

27] Krugmann S, Jordens I, Gevaert K, Driessens M, Vandekerckhove J, Hall A. Cdc42induces filopodia by promoting the formation of an IRSp53:Mena complex. CurrBiol 2001;11(21):1645–55.

28] Miki H, Yamaguchi H, Suetsugu S, Takenawa T. IRSp53 is an essential interme-diate between Rac and WAVE in the regulation of membrane ruffling. Nature2000;408(6813):732–5.

29] Suetsugu S, Murayama K, Sakamoto A, Hanawa-Suetsugu K, Seto A, Oikawa T, etal. The RAC binding domain/IRSp53-MIM homology domain of IRSp53 inducesRAC-dependent membrane deformation. J Biol Chem 2006;281:35347–58.

30] Fujiwara T, Mammoto A, Kim Y, Takai Y. Rho small G-protein-dependent bind-ing of mDia to an Src homology 3 domain-containing IRSp53/BAIAP2. BiochemBiophys Res Commun 2000;271:626–9.

31] Lim KB, Bu W, Goh WI, Koh E, Ong SH, Pawson T, et al. The Cdc42 effector IRSp53generates filopodia by coupling membrane protrusion with actin dynamics. JBiol Chem 2008;283(29):20454–72.

32] Mackie S, Aitken A. Novel brain 14-3-3 interacting proteins involved in neu-rodegenerative disease. FEBS J 2005;272:4202–10.

33] Mattila PK, Pykäläinen A, Saarikangas J, Paavilainen VO, Vihinen H, JokitaloE, et al. Missing-in-metastasis and IRSp53 deform PI(4,5)P2-rich mem-branes by an inverse BAR domain-like mechanism. J Cell Biol 2007;176(7):953–64.

34] Frost A, Perera R, Roux A, Spasov K, Destaing O, Egelman EH, et al. Structuralbasis of membrane invagination by F-BAR domains. Cell 2008;132(5):807–17.

35] Frost A, Unger VM, De Camilli P. The BAR domain superfamily: membrane-molding macromolecules. Cell 2009;137:191–6.

36] Shimada A, Niwa H, Tsujita K, Suetsugu S, Nitta K, Hanawa-Suetsugu K, et al.Curved EFC/F-BAR-domain dimers are joined end to end into a filament formembrane invagination in endocytosis. Cell 2007;129:761–72.

37] Habermann B. The BAR-domain family of proteins: a case of bending and bind-ing? EMBO Rep 2004;5:250–5.

38] Millard TH, Bompard G, Heung MY, Dafforn TR, Scott DJ, Machesky LM, et al.Structural basis of filopodia formation induced by the IRSp53/MIM homologydomain of human IRSp53. EMBO J 2005;24(2):240–50.

39] Yamagishi A, Masuda M, Ohki T, Onishi H, Mochizuki N. A novel actinbundling/filopodium-forming domain conserved in insulin receptor tyro-sine kinase substrate p53 and missing in metastasis protein. J Biol Chem2004;279(15):14929–36.

40] Lee SH, Kerff F, Chereau D, Ferron F, Klug A, Dominguez R. Structural basis for theactin-binding function of missing-in-metastasis. Structure 2007;15(2):145–55.

41] Saarikangas J, Zhao H, Pykäläinen A, Laurinmäki P, Mattila PK, Kinnunen PK,et al. Molecular mechanisms of membrane deformation by I-BAR domain pro-teins. Curr Biol 2009;19(2):95–107.

42] Biyasheva A, Svitkina T, Kunda P, Baum B, Borisy G. Cascade pathway of filopodiaformation downstream of SCAR. J Cell Sci 2004;117(6):837–48.

43] Svitkina TM, Bulanova EA, Chaga OY, Vignjevic DM, Kojima S, Vasiliev JM, et al.Mechanism of filopodia initiation by reorganization of a dendritic network. JCell Biol 2003;160(3):409–21.

44] Mejillano MR, Kojima S, Applewhite DA, Gertler FB, Svitkina TM, Borisy GG.Lamellipodial versus filopodial mode of the actin nanomachinery: pivotal roleof the filament barbed end. Cell 2004;118(3):363–73.

45] Steffen A, Faix J, Resch GP, Linkner J, Wehland J, Small JV, et al. Filopodia forma-tion in the absence of functional WAVE and Arp2/3-complexes. Mol Biol Cell2006;17(6):2581–91.

46] Faix J, Rottner K. The making of filopodia. Curr Op Cell Biol 2006;18:18–25.47] Mallavarapu A, Mitchison T. Regulated actin cytoskeleton assembly at

filopodium tips controls their extension and retraction. J Cell Biol1999;146(5):1097–106.

48] Funato Y, Terabayashi T, Suenaga N, Seiki M, Takenawa T, Miki H. IRSp53/Eps8complex is important for positive regulation of Rac and cancer cell motil-ity/invasiveness. Cancer Res 2004;64:5237–44.

49] Disanza A, Mantoani S, Hertzog M, Gerboth S, Frittoli E, Steffen A, et al. Regula-tion of cell shape by Cdc42 is mediated by the synergic actin-bundling activityof the Eps8-IRSp53 complex. Nat Cell Biol 2006;8:1337–47.

50] Drees F, Gertler FB. Ena/VASP: proteins at the tip of the nervous system. CurrOpin Neurobiol 2008;18:53–9.

51] Menna E, Disanza A, Cagnoli C, Schenk U, Gelsomino G, Frittoli E, et al. Eps8regulates axonal filopodia in hippocampal neurons in response to brain-derived

neurotrophic factor (BDNF). PLoS Biol 2009;7:e1000138.

52] Zhuravlev PI, Papoian GA. Molecular noise of capping protein binding inducesmacroscopic instability in filopodial dynamics. Proc Natl Acad Sci USA2009;106:11570–5.

53] Bu W, Chou AM, Lim KB, Sudhaharan T, Ahmed S. The Toca-1-N-WASP complexlinks filopodial formation to endocytosis. J Biol Chem 2009;284(17):11622–36.