Conserved Interactions with Cytoskeletal but Not Signaling ...

Developmental Biology 243, 260–271 (2002)doi:10.1006/dbio.2002.0571, available online at http://www.idealibrary.com on

Conserved Interactions with Cytoskeletal but NotSignaling Elements Are an Essential Aspectof Drosophila Wasp Function

Tamar Tal, Dalit Vaizel-Ohayon, and Eyal D. Schejter1

Department of Molecular Genetics, Weizmann Institute of Science, Rehovot 76100, Israel

Wiskott–Aldrich Syndrome proteins (WASp) serve as important regulators of cytoskeletal organization and function. Thesemodular proteins, which are well-conserved among eukaryotic species, act to promote actin filament assembly in responseto cues from various signal transduction pathways. Genetic analysis has revealed a requirement for the single Drosophilahomolog, Wasp (Wsp), in cell-fate decisions governing specific neuronal lineages. We have used this unique developmentalcontext to assess the contributions of established signaling and cytoskeletal partners of WASp. We present biochemical andgenetic evidence that, as expected, Drosophila Wsp performs its developmental role via the Arp2/3 complex, indicatingconservation of the cytoskeletal aspect of Wsp function in vivo. In contrast, we find that association with the key signalingmolecules CDC42 and PIP2 is not an essential requirement, implying that activation of Wsp function in vivo depends onadditional or alternative signaling pathways. © 2002 Elsevier Science (USA)

Key Words: WASp; Drosophila; Arp2/3 complex; CDC42; PIP2; sensory organs.

INTRODUCTION

Members of the evolutionarily conserved Wiskott–Aldrich Syndrome protein (WASp) family serve as keymediating elements between signal transduction pathwaysand functional reorganization of the actin-based cytoskel-eton (for recent reviews, see Mullins, 2000; Higgs andPollard, 2001; Millard and Machesky, 2001). The amino-terminal two-thirds of these modular proteins function asregulatory regions, being composed of a series of subunitsthat can bind and respond to various signaling molecules,including small rho-like GTPases, tyrosine kinase signalingelements, and phospholipids (Fig. 1). According to a gener-ally accepted model (Miki et al., 1998; Kim et al., 2000),association of signaling molecules with the regulatory re-gions relieves an autoinhibitory conformation, allowing aC-terminal domain (termed WA) to engage elements of thecytoskeletal machinery. In particular, the WA domain in-teracts with monomeric (G) actin and with Arp2/3, anevolutionarily conserved complex of seven protein subunits(Machesky et al., 1994; Welch et al., 1997). Upon activation

1 To whom correspondence should be addressed. Fax: 972-8-
9344108. E-mail: [email protected].
260

by WASp and WASp-related proteins, the Arp2/3 complexnucleates de novo formation of microfilaments, whicharrange in extensive branched arrays (Machesky et al., 1999;Yarar et al., 1999). Arp2/3-based polymerization of actin istranslated into forces that can reshape the cell surface,contribute to cell motility, and provide a mechanistic basisfor propulsion of vesicles and microbial cells within thecytoplasm (reviewed in Svitkina and Borisy, 1999; Pollard etal., 2000; Pantaloni et al., 2001).

We have previously reported on the isolation and charac-terization of mutant alleles of Wasp (Wsp), which encodesthe single WASp family homolog in the fruitfly Drosophila(Ben-Yaacov et al., 2001). Analysis of the Wsp mutantphenotype has provided a developmental context for WASpfunction. Our studies have identified a surprisingly specificdevelopmental role for Wsp in the establishment of lineageand cell fate, primarily during elaboration of the peripheralnervous system. In particular, Wsp is required for properexecution of distinct asymmetric divisions that govern theDrosophila sensory organ lineage (for review, see Lu et al.,2000). In Wsp mutants, sensory organ cells generally as-sume a neuronal fate, at the expense of other cell types.Both the nature of the Wsp mutant phenotypes and ob-
served genetic interactions suggest the Notch signaling
0012-1606/02 $35.00© 2002 Elsevier Science (USA)

All rights reserved.

pathway as a molecular context for Wsp function (Ben-Yaacov et al., 2001).

The wide variety of interactions described for WASpproteins leaves open the issue of the specific molecularpathways utilized by these important elements in vivo.Our identification of a defined developmental capacityfor a WASp family homolog enables us to address issuesof in vivo function, by assessing which of the structuraldomains of this protein, and which of its potentialmolecular partners, are required to carry out its develop-mental role. We report here biochemical evidence that,according to expectation, Wsp, the protein encoded bythe Wsp locus, associates with well-established signalingand cytoskeletal partners of WASp-family proteins. Fur-thermore, we present genetic evidence that Wsp operatesin vivo through its primary cytoskeletal effector, theArp2/3 complex. Surprisingly, we find that association ofWsp with the established activators, CDC42 and PIP2, isnot an essential aspect of in vivo Wsp function, implyingalternative modes of activation.

MATERIALS AND METHODS

Biochemical Methods

Wsp constructs were prepared by PCR amplification with appro-priate primers using a full-length Wsp cDNA as template, andsubcloned into RSET vectors (Invitrogen). All constructs weresequenced to ensure fidelity of the amplification and subcloning.Bead immobilized 6xHis-Wsp fusion proteins were prepared byincubation of bacterial cell lysates with Ni-NTA agarose beads(Qiagen). Beads prepared in this manner were incubated with eitherG-actin (Cytoskeleton, binding assay performed as in Rohatgi et al.,1999), bovine brain extract (kindly provided by O. Reiner), or

lysates of 293T cells transfected with CDC42 expression con-structs, and subjected to Western immunoblot analysis usingantibodies to actin (Sigma), human Arp2/3 complex components(kindly provided by L. Machesky), or CDC42 (Transduction Labo-ratories), respectively. For the vesicle cosedimentation assay (Kav-ran et al., 1998), Wsp proteins were eluted from the beads, pre-cleared at 100,000g, incubated for 40 min at room temperature withlipid vesicles containing 48% phosphatidylserine, 48% phosphati-dylcholine, and 4% PI or PIP2, and centrifuged at 100,000g. Pelletsresulting from this procedure underwent Western blot analysisusing anti-6xHis antibodies (Sigma).

Genetic Methods

Rescue constructs were prepared by subcloning of the amplifiedWsp regions into the pUAST vector (Brand and Perrimon, 1993),followed by germline transformation using standard methods (Sprad-ling, 1986). Multiple transgenic lines were established and at leasttwo independent insertions of each construct were used separatelyin the phenotypic rescue experiments. Expression of the transgenicWsp constructs was ascertained following crosses of the variouslines to the engrailed-GAL4 driver line (see Flybase (http://flybase.bio.indiana.edu/) for details concerning all fly stocks de-scribed) and detection of an engrailed striped expression pattern inprogeny embryos following immunostaining with anti-Wsp anti-bodies (Ben-Yaacov et al., 2001). Phenotypic rescue of Wsp1/Df(3R)3450 flies was attempted following introduction of both theUAS-Wsp constructs and the ubiquitous drivers armadillo-GAL4and T80-GAL4 into this mutant background. Similar expression ofall UAS-Wsp constructs described here in a wild-type backgroundhas no noticeable effect on viability or on the pattern and develop-ment of sensory organs. Mosaic head clones were obtained iney-FLP; Arpc1Q25sd FRT40A/l(2)cl-L31 FRT40A and ey-FLP; FRT42Dsktl�20/FRT42D l(2)cl-R111 flies.

FIG. 1. Structure of Wsp, the Drosophila WASp homolog. The domain structure of Drosophila Wsp fully resembles its mammalianhomologs (Ben-Yaacov et al., 2001; Higgs and Pollard, 2001). Residue numbers at the boundaries of major domains are indicated. Keymolecules shown to interact with mammalian WASp are shown above the Wsp schematic. Boxed domains include: an N-terminal regionhomologous to the EVH1 domain found in VASP/MENA proteins (also known as WH1), which binds the proline-rich WASp InteractingProtein (WIP; Ramesh et al., 1997); the regulatory region, which contains separate binding sites for PIP2 (B, basic region), GTP-boundCDC42 (GBD, GTPase binding domain) and a proline-rich (PR) region that binds various SH3-domain proteins and profilin; and theC-terminal WA effector domain (also known as WH2), which contains binding sites for monomeric actin (V, verprolin homologous) and theArp2/3 complex (C, central region; and A, acidic tail).

261Cytoskeletal and Signaling Aspects of Wasp

© 2002 Elsevier Science (USA). All rights reserved.

Tissue Preparation, Staining, and Examination byMicroscopy

Adult cuticles were prepared by warming to 50°C for 10 min in10% NaOH, to aid in removal of soft tissue, mounted in Hoyer’smedium, examined, and photographed with a Zeiss Axioplanmicroscope. Pupal heads (60 h old) were dissected and stained asdescribed (Ben-Yaacov et al., 2001). Primary antibodies used in-cluded an anti-ElaV mouse monoclonal (Developmental StudiesHybridoma Bank, used at a 1:10 dilution), and rat anti-Su(H) (Ghoet al., 1996; used at a 1:1000 dilution). Fluorescent images werecollected by using a Bio-Rad MRC-1024 confocal system, using anArgon/Krypton mixed gas laser, and mounted on a Zeiss Axiovertmicroscope. Adult heads were prepared for scanning electronmicroscopy (using a JEOL JSM-6400 microscope) by dehydration inan ethanol series, critical-point drying, and sputter coating with agold film. All images were prepared for publication by using AdobePhotoshop.

RESULTS

Conserved Associations of Wsp with EstablishedCytoskeletal Partners

We have employed a combination of biochemical andgenetic experimental approaches, based on a series of Wspconstructs, in which key functional domains have beenmanipulated or omitted. These constructs were studied in

two major ways: (1) Bacterially expressed 6xHis-tagged-fusion versions were tested for their ability to interact witheffector molecules in solution. (2) Utilizing the GAL4-UASsystem for regulated in vivo expression of Drosophila genes(Brand and Perrimon, 1993), we assessed the capacity oftransgenic copies of the Wsp constructs to rescue thedevelopmental defects characteristic of a Wsp mutant back-ground.

We first sought to determine whether Drosophila Wspfunction involves the same cytoskeletal machinery throughwhich WASp operate in other systems. Pull-down assaysusing 6xHis-Wsp proteins constructs (Fig. 2A) immobilizedon Ni-NTA agarose beads demonstrated that full-lengthWsp associates with pure monomeric (G) actin (Fig. 2B).This association could be repeated by using the C-terminalWA domain of Wsp, while the regulatory N-terminal half ofWsp failed to interact with G-actin. These observations areconsistent with the mapping of the G-actin binding regionof mammalian WASp to the WA domain, and in particular,to the verprolin-homologous sequences within this region(Machesky and Insall, 1998; Higgs et al., 1999; Rohatgi etal., 1999).

Using a similar experimental approach, we determinedthat the WA domain of Wsp is capable of associating withthe endogenous Arp2/3 protein complex (Fig. 2C). Mamma-lian cell-extract material pulled down by bead-immobilized

FIG. 2. Wsp interacts with cytoskeletal elements via the WA domain. (A) Structure of the 6xHis-tagged Wsp constructs used in G-actinand Arp2/3 pull-down assays. FL, full length Wsp; N, N-terminal half of Wsp (residues 1–267); WA, C-terminal effector domain of Wsp(residues 427–526). (B) G-actin binds to the WA domain of Wsp. Equal amounts of bacterially expressed 6xHis-tagged Wsp constructs wereimmobilized on Ni-NTA agarose beads and incubated with pure monomeric actin. Following low-speed centrifugation, G-actin bound tothe bead pellets (P) or remaining in the supernatants (S) was assayed by immunoblotting with anti-actin antibodies. (C) The Arp2/3 complexassociates with the WA domain of Wsp. Bead-immobilized Wsp constructs were incubated with a bovine brain extract, and material boundto the beads was subjected to Western blot analysis using antibodies specific to three components of the Arp2/3 complex, Arp2, Arp3, andp34/ARPC2 (Machesky and Insall, 1998). Asterisks mark the appropriately sized Arp complex-specific bands (anti-p34 consistentlyrecognized two bands in our preparations).

262 Tal, Vaizel-Ohayon, and Schejter


Wsp WA contained at least three components of the Arp2/3complex (Arp2, Arp3, and ARPC2/p34), as determined byantibodies specific to these proteins. The anti-ARPC2 anti-bodies identified an appropriately sized band in materialobtained by Wsp WA pull-down from Drosophila sources aswell (data not shown). Arp2/3 complex subunits did notinteract with the N-terminal half of Wsp in this assay.

Taken together, these binding experiments suggest thatDrosophila Wsp interacts with the universal WASp-familycytoskeletal effectors, G-actin and the Arp2/3 complex, andthat this association is carried out via the conservedC-terminal WA domain.

Wsp Functions in Vivo via its Association with theArp2/3 Complex

A genetic approach was employed, in parallel, to deter-mine the in vivo requirement for the cytoskeleton-interacting domain of Wsp. Zygotic mutations in Wspresult in pharate adult lethality, in which the single clearmorphological abnormality is a pronounced lack of mech-

anosensory bristles, the external manifestations of sensoryorgans, on most regions of the adult cuticle (Ben-Yaacov etal., 2001). Ubiquitous expression of a transgenic constructbased on the full-length Wsp cDNA (FL; Fig. 3A) fullyrestores viability and normal sensory organ development inflies bearing loss-of-function mutations in both copies ofthe Wsp gene (Fig. 3B). However, a similar construct, whichproduces a protein lacking only the C-terminal 30 residues(�CA), completely fails to rescue the Wsp bristle-loss phe-notype (Fig. 3B), as well as late-pupal lethality. Since themissing residues correspond to most of CA, the Arp2/3-binding portion of the WA domain, this finding implies thatassociation with Arp2/3 is essential for normal Wsp func-tion. A construct lacking only the (primarily acidic) ex-treme C-terminal 15 residues (�A), while unable to rescueWsp lethality, partially restores the bristle-loss phenotype(Fig. 3B). This finding suggests that a significant degree offunctional association with the Arp2/3 complex is medi-ated via the central (C) region of the cytoskeleton-interacting domain, which is retained in the �A construct.Interestingly, a transgenic construct expressing the full Wsp

FIG. 3. The WA domain is required but not sufficient for Wsp in vivo function. (A) Structure of the UAS-Wsp constructs used in theArp2/3-related phenotypic rescue assays. FL, full length Wsp; �CA, a construct lacking most of CA, the Arp2/3 binding portion of the WAdomain (residues 497–526); �A, a construct lacking the acidic residue stretch at the end of the WA domain (residues 512–526); WA, isolatedC-terminal effector domain of Wsp (residues 427–526). (B) Photographs show bristle arrangement on magnified portions of abdominalsegments of adult flies. Comparison of wild-type (WT) and Wsp1/Df(3R)3450 (Wsp) adult cuticles demonstrates the Wsp bristle-lossphenotype. Rescue of Wsp1/Df(3R)3450 flies harboring a UAS-FL transgene (FL) and driven by the ubiquitous armadillo-GAL4 driver isreadily observed. Similar expression of UAS-�CA (�CA) and UAS-WA (WA) in a Wsp1/Df(3R)3450 mutant background has no noticeableeffect, while using UAS-�A (�A) provides for significant restoration of the bristle pattern. Scale bar, 100 �m.



WA domain alone completely fails to rescue aspects of themutant phenotype (Fig. 3B), demonstrating that this keydomain is necessary but not sufficient for Wsp in vivofunction.

A complementary genetic experiment was designed todetermine whether Arp2/3 function is indeed required forthe Wsp-mediated process of sensory organ development(Fig. 4). Toward this end, we made use of strong loss-of-function alleles of Arpc1, the Drosophila homolog of thegene encoding the ARPC1/p41 subunit of the Arp2/3 com-plex (Hudson and Cooley, 2002). Unlike Wsp, zygotic mu-tants for Arpc1 do not survive to adult stages. We thereforemade use of the eyeless-FLP-FRT system (Newsome et al.,2000) to generate mosaic Arpc1 heterozygous flies, in whichhead capsule structures and cuticle are derived from largehomozygous mutant clones induced in the eye imaginaldisc. Compared with wild type (Fig. 4A), Arpc1 mosaicheads exhibit a pronounced loss of external sensory organstructures, which include both the bristle-shafts and thesocket structures from which they emanate (Fig. 4E). Thissmooth cuticle phenotype is a highly characteristic featureof Wsp mutant animals (Fig. 4C). To establish the develop-mental basis for the Arpc1 bristle-loss phenotype, we dis-sected and immunostained pupal mosaic heads using twoinformative sensory organ markers: Suppressor-of-Hairless[Su(H)], which specifically accumulates in socket cells (Ghoet al., 1996), and the neuronal-specific nuclear antigen ElaV(Robinow and White, 1991). In wild-type flies, each sensoryorgan lineage gives rise to five distinct cell types, includinga single neuron and a single socket cell (Hartenstein andPosakony, 1989; Gho et al., 1999; Reddy and Rodrigues,1999; Fig. 4B). In Arpc1 mosaics (Fig. 4F), a clear preponder-ance of ElaV-positive neurons is observed, coupled with amarked decrease in the number of cells expressing thesocket cell marker Su(H), highly similar to the cell-fatetransformation characteristic of Wsp mutants (Fig. 4D).

An abnormal, Wsp-like distribution of cell fates, in whichexcess neurons develop at the expense of other cell types, istherefore observed within the sensory organ lineages ofArpc1 mosaics. This finding significantly supports ourassertion that the functional requirement for DrosophilaWsp during cell-fate specification is mediated via theArp2/3 complex, and is thus likely to involve reorganiza-tion of the actin-based cytoskeleton.

The Conserved Association between Wsp andCDC42 Is Not Essential for Wsp Function

The small GTPase CDC42 is considered a primary acti-vator of mammalian WASp (Miki et al., 1998; Rohatgi et al.,1999; reviewed in Carlier et al., 1999). The GTPase bindingdomain (GBD) of WASp, which encompasses the CDC42binding site and flanking sequences, is well conserved inthe Drosophila homolog (Ben-Yaacov et al., 2001; Fig. 5B),and we had previously shown that Wsp interacts withactivated (GTP-bound) rho-like GTPases, with a strong

preference for CDC42 (Ben-Yaacov et al., 2001). We con-firmed and extended this observation using pull-down as-says in which various Wsp constructs (Fig. 5A) were reactedwith either wild-type CDC42 or a mutant form (G12V,referred to as CDC42act), which mimics the active, GTP-bound state (Qiu et al., 1997). Full-length Wsp stronglyinteracts in this assay with CDC42act, but not with wild-type CDC42 (Fig. 5C). The interaction can be repeated usingonly the N-terminal half of Wsp, but not with thecytoskeleton-interacting WA domain.

To confirm that the Wsp-CDC42act interaction maps tothe GBD, two additional 6xHis-Wsp constructs were gener-ated (Figs. 5A and 5B). The H242D variant, in which acrucial histidine within the GBD is replaced with anaspartate residue, was fashioned after a similar mutantwhich disrupts the interaction of mammalian WASp withactivated CDC42, and which has been extensively used instudies of CDC42 regulation of mammalian WASp (Miki etal., 1998; Rohatgi et al., 1999). The second construct(�CRIB), harbors a deletion of the 14-residue CRIB domain,which constitutes the core sequence of the GBD in smallGTPase binding proteins (Burbelo et al., 1995; Pirone et al.,2001). The association between Wsp and CDC42act observedin the pull-down assay is abolished by either disruption ofthe GBD (Fig. 5C). These observations demonstrate that thenature of the physical association between activatedCDC42 and WASp is well conserved by the Drosophilahomolog.

Having established that Wsp interacts with CDC42 in amanner indistinguishable from its mammalian counter-parts, we sought to determine the functional significance ofthe Wsp–CDC42act association. Toward this end, we gener-ated transgenic fly lines harboring Wsp constructs in whichthe GBD was mutated as above, and thus should lack theability to bind CDC42act (Fig. 6A). Surprisingly, introduc-tion and expression of these constructs in a Wsp mutantbackground fully rescues the pupal lethality and bristle-lossphenotypes of Wsp mutant flies (Fig. 6B), similar to therescue achieved when using a wild-type construct. Thisresult suggests that association, and presumably, activationof Wsp by CDC42 is not essential for its function duringDrosophila development. Unfortunately, we cannot di-rectly assess the consequences of reducing CDC42-basedsignaling in this system, since large adult cuticle clones ofstrong Drosophila CDC42 alleles cannot be obtained(Genova et al., 2000).

PIP2-Based Activation Is Not Essential for WspFunction

The ability of Wsp to perform its in vivo roles in theapparent absence of an association with CDC42act impliesthe presence of an alternative activating element. Recentstudies have identified phosphatidylinositol 4,5-bis-phosphate (PIP2) as an important phospholipid activator ofmammalian WASp, acting in this capacity on its own or in



concert with CDC42 (Higgs and Pollard, 2000; Prehoda etal., 2000; Rohatgi et al., 2000). We therefore examined thefeasibility of PIP2 as an independent or parallel activator toCDC42 in the Drosophila system.

Using a vesicle cosedimentation assay, we first deter-mined that the N-terminal regulatory half of Wsp preferen-tially associates with PIP2 in comparison to PIP (Fig. 7B), aprofile demonstrated for mammalian WASp. The interac-tion of PIP2 with mammalian WASp is achieved through ashort stretch of basic residues, found just upstream of theGBD (Prehoda et al., 2000; Rohatgi et al., 2000). A similar,albeit shorter, domain is found at the same position in

Drosophila Wsp (Fig. 5B). The preferential association ofthe Wsp N-terminal half with PIP2 is abolished uponremoval of this basic sequence (Fig. 7B), demonstratingconservation of the phosphoinositide binding site in thefly homolog.

The functional significance of the PIP2-binding sitewas assessed following generation of transgenic fliesharboring a Wsp construct that differed from the full-length form by removal of the basic region (Fig. 8A). Thisconstruct was capable of fully rescuing the pupal lethal-ity and sensory organ defects of Wsp mutant flies (Fig.8B), suggesting that association of Wsp with the PIP2

FIG. 4. The Arp2/3 complex subunit Arpc1 is required, like Wsp, for cell-fate specification in sensory organs. Scanning electronmicrographs (SEMs) of adult heads (left column) and confocal micrographs of stained preparations of 60-h-old pupal heads (right column)of wild-type (A, B) and Wsp1/Df(3R)3450 flies (C, D), and of mosaic clones in (E) and (F), ey-FLP; Arpc1Q25sd FRT40A/l(2)cl-L31 FRT40A flies.SEMs reveal a pronounced loss of head and interommatidial (eye) bristles in both Wsp and Arpc1 mosaics. Pupal heads were stained for theneuronal nuclear marker Elav (green) and the socket-cell cytoplasmic marker Su(H) (red). In wild-type pupae (B), pairs of Elav andSu(H)-positive cells mark the position of sensory organs (arrowheads) which will give rise to bristles. In both Wsp mutants (D) and Arpc1mosaics (F), an excess number of neurons is specified (arrows), while socket cell number is strongly reduced. A small, variable number ofproperly differentiated sensory organs is observed in both Wsp and Arpc1 mutants (see also Ben-Yaacov et al., 2001). The positions of theocelli light-sensing organs (oc) are indicated for orientation. Scale bar, 100 �m.



activator is not essential for its in vivo function. Further-more, the possibility that association with either PIP2 orCDC42act alone is sufficient for Wsp function is negatedby the full phenotypic rescue achieved using a modifiedconstruct, �B-H242D (Figs. 8A and 8B). This constructnot only lacks the basic region, but also harbors theH242D point mutation, and therefore should lack theability to bind either activator.

We sought to complement the phenotypic rescue resultsby compromising PIP2-based signaling activity, and assess-ing the subsequent effects on adult sensory organ develop-

ment. Toward this end, we examined the phenotypic con-sequences of eyeless-FLP induced mosaics of a strong loss-of-function allele of skittles (sktl), sktl�20 (Hassan et al.,1998), which encodes one of three isoforms of DrosophilaPIP5 kinase (Morrison et al., 2000), a key enzyme in themajor biosynthetic pathway for PIP2. Although disruptionof sktl function has wide-ranging effects on Drosophiladevelopment (Hassan et al., 1998), we find that the bristlepattern in large sktl�20 head clones is mildly affected, if at all(Fig. 8C). This observation indicates that reduction ofPIP2-based signaling does not interfere with sensory organ

FIG. 5. Wsp interacts with activated CDC42 via the GBD. (A) Structure of the 6xHis-tagged Wsp constructs used in CDC42 pull-downassays. FL, N, and WA, as in Fig. 2. H242D, full-length Wsp variant in which histidine242 (within the GBD) is replaced by an aspartateresidue. �CRIB, full-length Wsp variant in which the core CRIB domain of the GBD (residues 234–247) is deleted. (B) Sequence alignmentof Drosophila Wsp (residues 212–256) and bovine N-WASp (residues 181–225; Miki et al., 1996), demonstrating the close sequencesimilarity within the CDC42 and PIP2 binding sites of the regulatory region. The 14-residue CRIB domain deleted in the �CRIB constructsis boxed, and an arrow indicates the critical histidine242 residue. The boxed six-residue sequence designated �B encompasses the stretch ofbasic residues crucial for binding of PIP2 (see Fig. 7). (C) Binding of CDC42 to Wsp requires an intact GBD. Equal amounts ofbead-immobilized Wsp constructs were incubated with lysates of 293T cells expressing wild-type human CDC42 (left) or the constitutivelyactive variant CDC42G12V (CDC42act), which mimics the GTP-bound state (right). CDC42 pelleting with the centrifuged beads (P) orremaining in solution (S) was assayed by immunoblotting with anti-CDC42 antibodies.



development and is consistent with the apparent absence ofa requirement for PIP2 in activating Wsp. Taken together,the observations described here imply that association withthe major established in vitro activators of WASp, PIP2 andCDC42act, is not essential for the developmental rolescarried out by the fly homolog, Wsp.

DISCUSSION

Conservation of the Cytoskeletal Aspect of WspFunction in Vivo

The results reported above provide a detailed assessmentof the molecular pathway through which a WASp familyhomolog, Drosophila Wsp, performs essential in vivo func-tions. Wsp is specifically required during cell-fate decisionsin neural lineages (Ben-Yaacov et al., 2001), and the evi-dence presented here suggests that this unanticipated func-tion involves established cytoskeletal partners of WASp,and in particular, the Arp2/3 protein complex. Our bindingstudies demonstrate a capacity for Wsp to directly associatewith monomeric actin via WA, the C-terminal cytoskeleton-interacting domain present in all WASp and WASp-related

proteins (Higgs and Pollard, 2001). In parallel, the WA domainof Wsp is shown to interact with components of the Arp2/3complex, the primary downstream target of signal transduc-tion pathways operating through WASp family proteins (Ma-chesky and Insall, 1999). The in vivo significance of theseassociations, which are characteristic of WASp elements ingeneral, is demonstrated by a dual genetic approach. The final30 residues at the C-terminal end of the WA domain of Wspprove necessary for rescue of Wsp mutant phenotypes, whilemutations in the Arp2/3 complex subunit Arpc1 lead tocell-fate transformations and neuronal excess during sensoryorgan development, a distinct, Wsp-like phenotype. Takentogether with the binding studies, these genetic observationsimply that engagement of the cytoskeletal machinery via theC-terminal WA domain is an essential aspect of Wsp functionin vivo.

Several additional inferences can be drawn from thereported results, regarding the mechanism by which thecytoskeleton-interacting domain of Wsp operates. We notethat significant function is retained after removal of theextreme C-terminal 15 residues, corresponding to the A(acidic) portion of the WA domain. This observation sug-gests that the remaining WA sequences, comprising the

FIG. 6. Mutations in the Wsp GBD do not interfere with in vivo function. (A) Structure of the UAS-Wsp constructs used in theCDC42-related phenotypic rescue assays. FL, full length Wsp; H242D, full-length Wsp variant in which histidine242 (within the GBDdomain) is replaced by an aspartate residue; �CRIB, full-length Wsp variant in which the entire CRIB domain (residues 234–247) is deleted.(B) Comparison of wild-type (WT) and Wsp1/Df(3R)3450 (Wsp) adult cuticles demonstrates the Wsp bristle-loss phenotype. Restoration ofthe wild-type pattern in a Wsp1/Df(3R)3450 mutant background is achieved by ubiquitous (armadillo-GAL4 driven) expression of theUAS-FL transgene (FL). Similar expression of the UAS-H242D (H242D) and UAS-�CRIB (�CRIB) transgenic constructs, which are incapableof binding CDC42, completely rescues the Wsp mutant phenotype. Scale bar, 100 �m



so-called central (C) domain, contribute significantly to thefunctional interaction with Arp2/3, and is in good keepingwith a recent study highlighting the importance of the Cdomain in WASp-based activation of the actin-nucleatingactivity of Arp2/3 (Hufner et al., 2001). A second notewor-thy aspect is the inability of the full Wsp WA domain torescue Wsp mutant phenotypes on its own, even thoughbiochemical studies have repeatedly demonstrated consti-tutive Arp2/3 activation by the isolated WA domains ofvarious WASp and WASp-related proteins (Machesky et al.,1999; Rohatgi et al., 1999; Winter et al., 1999). Thisobservation implies that N-terminal regions absent fromthe truncated protein play important in vivo roles, beyondtheir established capacity to relieve a self-inhibitory con-formation. These may include proper localization of theactivated protein to specific cellular sites of function, as hasbeen proposed elsewhere (Moreau et al., 2000).

The cellular mechanism by which Wsp influences lineagedecisions during Drosophila development remains un-known. The data presented here strongly suggest that thefunctional requirement for Wsp is mediated via the Arp2/3

complex, thereby implying that Wsp-dependent cell-fatespecification involves reorganization of the actin-basedcytoskeleton. It remains to be seen just how this intriguingconnection between the cytoskeletal machinery and a keydevelopmental mechanism is carried out.

The Signaling Aspect of Wsp Function

In contrast to the demonstration of a functional con-nection in vivo between Wsp and the established cy-toskeletal partners of WASp proteins in general, our datasuggest that association with the major established acti-vators of WASp, the small GTPase CDC42 and thephosphoinositide PIP2, is not essential for the develop-mental roles carried out by the Drosophila WASp ho-molog. Characterization of the prototype WASp as aCDC42-binding protein is a longstanding observation(Aspenstrom et al., 1996; Kolluri et al., 1996; Symons etal., 1996), and a functional connection between CDC42activation of WASp elements and reorganization of theactin cytoskeleton via Arp2/3 has been firmly established(Miki et al., 1998; Rohatgi et al., 1999). Association withPIP2 (Miki et al., 1996) has gained recent prominence asan alternative activating mechanism of WASp, whileoptimal activation is achieved by the combined action ofboth signaling molecules (Higgs and Pollard, 2000; Pre-hoda et al., 2000; Rohatgi et al., 2000). We show here thatthe Drosophila Wsp protein interacts with both CDC42(in its activated state) and PIP2, and that this associationmaps to the well-conserved domains identified and char-acterized as the CDC42 and PIP2 binding sites in WASp.Elimination of these sites, however, does not interferewith the ability of Wsp transgenic constructs to rescueWsp mutant phenotypes, suggesting that associationwith these elements, either separately or in combination,is not an essential aspect of Wsp function during Dro-sophila development.

Several issues are raised by these unexpected observa-tions and warrant further discussion. One matter is thebasis for evolutionary conservation of the activator bindingsites, despite the absence of an essential developmentalrole. This situation is not without precedence (see, e.g., Suand Kiehart, 2001), and may indicate that the conservedsites function in a relatively subtle context, which thephenotypic studies have failed to identify. A second, cardi-nal issue is the implications these findings have on ourunderstanding of the Wsp molecular pathway. In particular,the possibility that elements other than CDC42 and PIP2contribute significantly to Wsp activation must be consid-ered. Elements of tyrosine-kinase signaling pathways con-stitute possible alternative candidates, as several such mol-ecules have been shown to associate with and activateWASp (Banin et al., 1996; She et al., 1997; Rohatgi et al.,2001). This contribution could act in concert with thefunctions performed by CDC42 and PIP2, but may wellprovide the primary activating signal in this particular in

FIG. 7. Wsp interacts with PIP2 via the basic domain of theregulatory region. (A) Structure of the 6xHis-tagged Wsp constructsused in vesicle cosedimentation assays. N, N-terminal half of Wsp(residues 1–267); N�B, similar construct lacking a six-residuestretch within the basic region (see Fig. 5B). (B) Purified Wspconstructs were incubated with lipid vesicles containing either PIPor PIP2, followed by cosedimentation at 100,000g. Pellets wereanalyzed by Western blots for the presence of the Wsp fusionproteins using anti-6xHis antibodies. (Left) The intact N constructcosediments with PIP2-containing vesicles, while its associationwith PIP vesicles is at background (-) levels. (Right) The N�Bconstruct, lacking the short basic region, does not exhibit prefer-ential affinity towards PIP2 vesicles, as measured by this assay.



vivo setting. Our observations thus suggest caution indrawing inferences from in vitro studies, and underscorethe need for further work, with particular emphasis ongenetic screens designed to identify additional, physiologi-cal activators of the Wsp pathway.

ACKNOWLEDGMENTS

We thank various colleagues for providing crucial reagents,experimental advice, and assistance during the course of thesestudies. We thank Andy Hudson and Lynn Cooley for the Arpc1flies and for stimulating discussions regarding the functions of theArp2/3 complex in Drosophila, Bassem Hassan and Hugo Bellen forthe sktl�20 flies, Laura Machesky and Francois Schweisguth forArp2/3-complex and Su(H) antibodies, respectively, Orly Reiner forthe bovine brain extract, Misha Shtutman and Sasha Bershadsky forinstruction and help with the CDC42 assays, Krishnan Venkatara-man and Tony Futerman for instruction and help with the lipid

vesicle cosedimentation assay, and Dr. Eugenia Klein for instruc-tion and skillful operation of the scanning electron microscope.This research was supported by research grants from the IsraelScience Foundation and the Minerva Foundation.

REFERENCES

Aspenstrom, P., Lindberg, U., and Hall, A. (1996). Two GTPases,Cdc42 and Rac, bind directly to a protein implicated in theimmunodeficiency disorder Wiskott–Aldrich syndrome. Curr.Biol. 6, 70–75.

Banin, S., Truong, O., Katz, D. R., Waterfield, M. D., Brickell, P. M.,and Gout, I. (1996). Wiskott–Aldrich syndrome protein (WASp) isa binding partner for c-Src family protein-tyrosine kinase. Curr.Biol. 6, 981–988.

Ben-Yaacov, S., Le Borgne, R., Abramson, I., Schweisguth, F., andSchejter, E. D. (2001). Wasp, the Drosophila Wiskott–AldrichSyndrome gene homologue, is required for cell fate decisionsmediated by Notch signaling. J. Cell Biol. 152, 1–13.

FIG. 8. Association of Wsp with PIP2 is not required for in vivo function. (A) Structure of the UAS-Wsp constructs used in the PIP2-relatedphenotypic rescue assays. �B, a full-length UAS-Wsp transgene from which the short basic region was deleted; �B-H242D, a similarconstruct bearing the H242D point mutation in addition to removal of the basic residues. (B) Restoration of the wild-type bristle patternin a Wsp mutant background (Wsp1/Df(3R)3450) is achieved by ubiquitous (armadillo-GAL4 driven) expression of either the UAS-�B (�B)or UAS-�B-H242D (�B-H242D) constructs. (C) Reduction of PIP2 by genetic means does not affect sensory organ development. The headbristle pattern of a sktl�20 mosaic fly (sktl; ey-FLP; FRT42D sktl�20/FRT42D l(2)cl-R111) is similar to wild type (WT). The strong effect oneye morphology in the sktl mosaic head attests to the generation of large head clones. Scale bars, 100 �m.



Brand, A., and Perrimon, N. (1993). Targeted gene expression as ameans of altering cell fates and generating dominant phenotypes.Development 118, 401–415.

Burbelo, P. D., Drechsel, D., and Hall, A. (1995). A conservedbinding motif defines numerous candidate target proteins forboth Cdc42 and Rac GTPases. J. Biol. Chem. 270, 29071–29074.

Carlier, M. F., Ducruix, A., and Pantaloni, D. (1999). Signalling toactin: The Cdc42-N-WASP-Arp2/3 connection. Chem. Biol. 6,R235–R240.

Genova, J. L., Jong, S., Camp, J. T., and Fehon, R. G. (2000).Functional analysis of Cdc42 in actin filament assembly, epithe-lial morphogenesis, and cell signaling during Drosophila devel-opment. Dev. Biol. 221, 181–194.

Gho, M., Lecourtois, M., Geraud, G., Posakony, J. W., and Schweis-guth, F. (1996). Subcellular localization of Suppressor of Hairlessin Drosophila sense organ cells during Notch signalling. Devel-opment 122, 1673–1682.

Gho, M., Bellaiche Y., and Schweisguth, F. (1999). Revisiting theDrosophila microchaete lineage: A novel intrinsically asymmet-ric cell division generates a glial cell. Development 126, 3573–3584.

Hartenstein, V., and Posakony, J. W. (1989). Development of adultsensilla on the wing and notum of Drosophila melanogaster.Development 107, 389–405.

Hassan, B. A., Prokopenko, S. N., Breuer, S., Zhang, B., Paululat, A.,and Bellen, H. J. (1998). skittles, a Drosophila phosphoinositol4-phosphate 5-kinase, is required for cell viability, germlinedevelopment and bristle morphology, but not for neurotransmit-ter release. Genetics 150, 1527–1537.

Higgs, H. N., Blanchoin, L., and Pollard, T. D. (1999). Influence ofthe C terminus of Wiskott–Aldrich syndrome protein (WASp)and the Arp2/3 complex on actin polymerization. Biochemistry38, 15212–15222.

Higgs, H. N., and Pollard T. D. (2000). Activation by Cdc42 andPIP2 of Wiskott–Aldrich Syndrome Protein (WASp) stimulatesactin nucleation by Arp2/3 complex. J. Cell Biol. 150, 1311–1320.

Higgs, H. N., and Pollard, T. D. (2001). Regulation of actin filamentnetwork formation through Arp2/3 complex: Activation by adiverse array of proteins. Annu. Rev. Biochem. 70, 649–676.

Hudson, A. M., and Cooley, L. (2002). A subset of dynamic actinrearrangements in Drosophila requires the Arp2/3 complex. J.Cell Biol. 156, in press.

Hufner, K., Higgs, H. N., Pollard, T. D., Jacobi, C., Aepfelbacher,M., and Linder, S. (2001). The VC region of Wiskott–Aldrichsyndrome protein induces Arp2/3 complex-dependent actinnucleation. J. Biol. Chem. 276, 35761–35767.

Kavran, J. M., Klein, D. E., Marco Falasca, A. L., Isakoff, S. J.,Skolnik, E. Y., and Lemmon, M. A. (1998). Specificity andpromiscuity in phosphoinositide binding by pleckstrin homologydomains. J. Biol. Chem. 273, 30497–30508.

Kim, A. S., Kakalis, L. T., Abdul-Manan N., Liu, G. A., and Rosen,M. K. (2000). Autoinhibition and activation mechanisms of theWiskott–Aldrich syndrome protein. Nature 404, 151–158.

Kolluri, R., Tolias, K. F., Carpenter, C. L., Rosen, F. S., andKirchhausen, T. (1996). Direct interaction of the Wiskott–Aldrich syndrome protein with the GTPase Cdc42. Proc. Natl.Acad. Sci. USA 93, 5615–5618.

Lu, B., Jan, L., and Jan, Y. N. (2000). Control of cell divisions in thenervous system: Symmetry and asymmetry. Annu. Rev. Neuro-sci. 23, 531–556.

Machesky, L. M., Atkinson, S. J., Ampe, C., Vandekerckhove, J.,and Pollard, T. D. (1994). Purification of a cortical complexcontaining two unconventional actins from Acanthamoeba byaffinity chromatography on profilin-agarose. J. Cell Biol. 127,107–115.

Machesky, L. M., and Insall, R. H. (1998). Scar1 and the relatedWiskott–Aldrich syndrome protein, WASP, regulate the actincytoskeleton through the Arp2/3 complex. Curr. Biol. 8, 1347–1356.

Machesky, L. M., and Insall, R. H. (1999). Signaling to actindynamics. J. Cell Biol. 146, 267–272.

Machesky, L. M., Mullins, R. D., Higgs, H. N., Kaiser, D. A.,Blanchoin, L., May, R. C., Hall, M. E., and Pollard, T. D. (1999).Scar, a WASp-related protein, activates nucleation of actin fila-ments by the Arp2/3 complex. Proc. Natl. Acad. Sci. USA 96,3739–3744.

Miki, H., Miura, K., and Takenawa, T. (1996). N-WASP, a novelactin-depolymerizing protein, regulates the cortical cytoskeletalrearrangement in a PIP2-dependent manner downstream of ty-rosine kinases. EMBO J. 15, 5326–5335.

Miki, H., Sasaki, T., Takai, Y., and Takenawa, T. (1998). Inductionof filopodium formation by a WASP-related actin-depolymerizing protein N-WASP. Nature 391, 93–96.

Millard, T. H., and Machesky, L. M. (2001). The Wiskott–Aldrichsyndrome protein (WASP) family. Trends Biochem. Sci. 26,198–199.

Moreau, V., Frischknecht, F., Reckmann, I., Vincentelli, R., Rabut,G., Stewart, D., and Way, M. (2000). A complex of N-WASP andWIP integrates signalling cascades that lead to actin polymeriza-tion. Nat. Cell Biol. 2, 441–448.

Morrison, D. K., Murakami, M. S., and Cleghon, V. (2000). Proteinkinases and phosphatases in the Drosophila genome. J. Cell Biol.150, F57–F62.

Mullins, R. D. (2000). How WASP-family. proteins and the Arp2/3complex convert intracellular signals into cytoskeletal struc-tures. Curr. Opin. Cell Biol. 12, 91–96.

Newsome, T. P., Asling, B., and Dickson, B. J. (2000). Analysis ofDrosophila photo-receptor axon guidance in eye-specific mosa-ics. Development 127, 851–860.

Pantaloni, D., Le Clainche, C., and Carlier, M. F. (2001). Mecha-nism of actin-based motility. Science 292, 1502–1506.

Pirone, D. M., Carter, D. E., and Burbelo, P. D. (2001). Evolutionaryexpansion of CRIB-containing Cdc42 effector proteins. TrendsGenet. 17, 370–373.

Pollard, T. D., Blanchoin, L., and Mullins, R. D. (2000). Molecularmechanisms controlling actin filament dynamics in nonmusclecells. Annu. Rev. Biophys. Biomol. Struct. 29, 545–576.

Prehoda, K. E., Scott, J. A., Mullins, R. D., and Lim, W. A. (2000).Integration of multiple signals through cooperative regulation ofthe N-WASP-Arp2/3 complex. Science 290, 801–806.

Qiu, R. G., Abo, A., McCormick, F., and Symons, M. (1997). Cdc42regulates anchorage-independent growth and is necessary for Rastransformation. Mol. Cell. Biol. 17, 3449–3458.

Ramesh, N., Anton, I. M., Hartwig, J. H., and Geha, R. S. (1997).WIP, a protein associated with Wiskott–Aldrich syndrome pro-tein, induces actin polymerization and redistribution in lym-phoid cells. Proc. Natl. Acad. Sci. USA 94, 14671–14676.

Reddy, G. V., and Rodrigues V. (1999). A glial cell arises from anadditional division within the mechanosensory lineage duringdevelopment of the microchaete on the Drosophila notum.Development 126, 4617–4622.



Robinow, S., and White, K. (1991). Characterization and spatialdistribution of the ELAV protein during Drosophila melano-gaster development. J. Neurobiol. 22, 443–461.

Rohatgi, R., Ma, L., Miki, H., Lopez, M., Kirchhausen, T., Tak-enawa, T., and Kirschner, M. W. (1999). The interaction betweenN-WASP and the Arp2/3 complex links Cdc42-dependent signalsto actin assembly. Cell 97, 221–231.

Rohatgi, R., Ho, H.-y.H., and Kirschner, M. W. (2000). Mechanismof N-WASP activation by Cdc42 and phosphatidylinositol-4,5-bisphosphate. J. Cell Biol. 150, 1299–1309.

Rohatgi, R., Nollau, P., Ho, H. Y., Kirschner, M. W., and Mayer, B. J.(2001). Nck and phosphatidylinositol 4,5-bisphosphate synergis-tically activate actin polymerization through the N-WASP-Arp2/3 pathway. J. Biol. Chem. .276, 26448–26452.

She, H., Rockow, S., Tang, J., Nishimura, R., Skolnik, E. Y., Chen,M., Margolis, B., and Li, W. (1997). Wiskott–Aldrich syndromeprotein is associated with the adapter protein Grb2 and epider-mal growth receptor in living cells. Mol. Biol. Cell 8, 1709–1721.

Spradling, A. C. (1986). P element-mediated transformation. In“Drosophila: A Practical Approach” (D. B. Roberts, Ed.), pp.175–197. IRL Press, Oxford, UK.

Su, Z,., and Kiehart, D. P. (2001). Protein kinase C phosphorylatesnonmuscle myosin-II heavy chain from Drosophila but regula-

tion of myosin function by this enzyme is not required forviability in flies. Biochemistry 40, 3606–3614.

Svitkina, T. M., and Borisy, G. G. (1999). Progress in protrusion:The tell-tale scar. Trends Biochem. Sci. 24, 432–436.

Symons, M., Derry, J. M. J., Karlak, B., Jiang, S., Lemahieu, V.,McCormick, F., Francke, U., and Abo, A. (1996). Wiskott–Aldrichsyndrome protein, a novel effector for the GTPase Cdc42hs, isimplicated in actin polymerization. Cell 84, 723–734.

Welch, M. D., DePace, A. H., Verma, S., Iwamatsu, A., andMitchison, T. J. (1997). The human Arp2/3 complex is composedof evolutionarily conserved subunits and is localized to cellularregions of dynamic actin filament assembly. J. Cell Biol. 138,375–384.

Winter, D., Lechler, T., and Li, R. (1999). Activation of the yeastArp2/3 complex by Bee1p, a WASP-family protein. Curr. Biol. 9,501–504.

Yarar, D., To, W., Abo, A., and Welch, M. D. (1999). The Wiskott–Aldrich syndrome protein directs actin-based motility by stimu-lating actin nucleation with the Arp2/3 complex. Curr. Biol. 9,555–558.

Received for publication September 26, 2001Accepted December 3, 2001

Published online February 11, 2002



Conserved Interactions with Cytoskeletal but Not Signaling ...

Documents

Transcript of Conserved Interactions with Cytoskeletal but Not Signaling ...