Role and structural mechanism of WASP-triggeredconformational changes in branched actin filamentnucleation by Arp2/3 complexMax Rodnick-Smitha,b, Qing Luana, Su-Ling Liua, and Brad J. Nolena,b,1

aInstitute of Molecular Biology, University of Oregon, Eugene, OR 97403; and bDepartment of Chemistry and Biochemistry, University of Oregon, Eugene,OR 97403

Edited by Thomas D. Pollard, Yale University, New Haven, CT, and approved May 18, 2016 (received for review September 8, 2015)

The Arp2/3 (Actin-related proteins 2/3) complex is activated by WASP(Wiskott–Aldrich syndrome protein) family proteins to nucleatebranched actin filaments that are important for cellular motility.WASP recruits actin monomers to the complex and stimulates move-ment of Arp2 and Arp3 into a “short-pitch” conformation that mimicsthe arrangement of actin subunits within filaments. The relative con-tribution of these functions in Arp2/3 complex activation and themechanism by which WASP stimulates the conformational changehave been unknown. We purified budding yeast Arp2/3 complexheld in or near the short-pitch conformation by an engineered cova-lent cross-link to determine if the WASP-induced conformationalchange is sufficient for activity. Remarkably, cross-linked Arp2/3 com-plex bypasses the need forWASP in activation and is more active thanWASP-activated Arp2/3 complex. These data indicate that stimulationof the short-pitch conformation is the critical activating function ofWASP and that monomer delivery is not a fundamental requirementfor nucleation but is a specific requirement for WASP-mediatedactivation. During activation,WASP limits nucleation rates by releasingslowly from nascent branches. The cross-linked complex is inhibited byWASP’s CA region, even though CA potently stimulates cross-linking,suggesting that slow WASP detachment masks the activating poten-tial of the short-pitch conformational switch. We use structure-basedmutations and WASP–Arp fusion chimeras to determine how WASPstimulates movement toward the short-pitch conformation. Our dataindicate that WASP displaces the autoinhibitory Arp3 C-terminal tailfrom a hydrophobic groove at Arp3′s barbed end to destabilize theinactive state, providing a mechanism by which WASP stimulates theshort-pitch conformation and activates Arp2/3 complex.

actin | Arp2/3 | WASP

Proper regulation of the actin cytoskeleton is critical for cells toorchestrate processes such as division, differentiation, motility,

and endocytosis. Assembly of new actin filaments is limited by aslow nucleation step, and cells rely on three classes of actin fila-ment nucleators—tandemWH2 domain proteins, formins, and theArp2/3 (Actin-related proteins 2/3) complex—to control preciselywhen and where actin filament networks assemble (1, 2). TheArp2/3 complex is unique among actin filament nucleators in thatit typically nucleates branched (as opposed to linear) actin fila-ments (3–5). Branched actin filaments are important for assem-bling the dendritic networks required for lamellipodial protrusionand endocytosis (6, 7).The Arp2/3 complex is tightly regulated, and multiple factors,

including preformed actin filaments, actin monomers, ATP, and aclass of proteins called “nucleation-promoting factors” (NPFs) aretypically required for activation (5–7). In some species, phosphor-ylation of the Arp2/3 complex is also required for activation (8).WASP (Wiskott–Aldrich syndrome protein) family proteins are thebest understood NPFs and share a conserved polypeptide sequencecalled “VCA” (verprolin homology, central, acidic), which consti-tutes the minimal region sufficient for Arp2/3 complex activation.WASP proteins interact with actin monomers through their Vand C regions and bind Arp2/3 complex through the C and A re-

gions (9, 10). How WASP proteins activate the Arp2/3 complex isunknown, but available data suggest they play multiple roles in thebranching nucleation reaction. For instance, several lines of evi-dence indicate that the CA segment of WASP proteins stimulates aconformational change in the complex important for activation(11–15). In addition, WASP-VCA recruits actin monomers to thecomplex through its V region. Kinetic simulations of the weakNPF-independent activity of the Arp2/3 complex suggested actinmonomers must bind the complex to form a nucleus, leading to theidea that WASP-recruited monomers could “jump start” nucleation(16, 17). The observation that the V region is required for WASP-mediated activation of the complex supported this idea and indi-cated that direct monomer tethering not only might acceleratenucleation but also might be a general requirement for the as-sembly of the nucleus (9, 10, 12). In addition to its activating in-fluence, WASP-VCA also can have an inhibitory influence on thekinetics of branching nucleation. Specifically, recent single-mole-cule total internal reflection fluorescence (TIRF) experimentsshowed that VCA must be released from nascent branch junctionsbefore nucleation and that slow WASP-VCA release limits nu-cleation rates (18, 19). Efforts to understand the relative contri-butions of each of these roles of WASP on activation have beenhampered by a lack of methods that can disentangle these func-tions or directly probe the connection between conformationand activity.Two of the seven protein subunits in the Arp2/3 complex, Arp2

and Arp3, are homologous to actin. High-resolution crystal struc-tures of inactive Arp2/3 complex show Arp2 and Arp3 interacting

Significance

Assembly of actin filaments is tightly regulated to orchestratebasic cellular processes such as motility, division, and differenti-ation. To control when and where actin filaments assemble, cellsrely on actin filament nucleators including the Arp2/3 (Actin-related proteins 2/3) complex, a seven-subunit protein assemblythat nucleates branched actin filaments to create dendritic actinnetworks. Activity of the Arp2/3 complex is controlled by WASP(Wiskott–Aldrich syndrome protein) proteins, which bind directlyto it to activate nucleation. To understand how WASP proteinsactivate the complex, we used chemical cross-linking to engineeran Arp2/3 complex that is locked into an “on” state withoutWASP. In addition, we used biochemical experiments to de-termine howWASP proteins stimulate the on state. These resultshave important implications for understanding how cells controlthe actin cytoskeleton.

Author contributions: M.R.-S. and B.J.N. designed research; M.R.-S., Q.L., S.-L.L., and B.J.N.performed research; M.R.-S. and B.J.N. analyzed data; and M.R.-S. and B.J.N. wrote thepaper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: bnolen@uoregon.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1517798113/-/DCSupplemental.

E3834–E3843 | PNAS | Published online June 20, 2016 www.pnas.org/cgi/doi/10.1073/pnas.1517798113

Dow

nloa

ded

by g

uest

on

Nov

embe

r 24

, 202

0

end-to-end in a “splayed” conformation (20–23). Upon activation,Arp2 is hypothesized to undergo a rigid body movement that movesit ∼25 Å into position next to Arp3, forming an Arp2–Arp3 “short-pitch” dimer that structurally mimics two consecutive actin subunitswithin a filament (20, 24). Low-resolution EM structures of Arp2/3complex bound to the WASP proteins N-WASP (neural Wiskott–Aldrich syndrome protein) or WAVE1 (WASP-family verprolin-homologous protein-1) support this hypothesis, because they showArp2 and Arp3 in a filament-like arrangement (15). In addition,recent experiments demonstrated that N-WASP stimulates site-specific cross-linking of engineered cysteines on Arp2 and Arp3 thatreact only when the complex adopts the short-pitch conformation(11). These studies suggest a general function of WASP pro-teins may be to stimulate the short-pitch conformation, providing anexplanation for how they could activate the complex. However,whether this function is conserved among WASP proteins andwhether adoption of the short-pitch conformation is sufficient foractivation is unknown. Importantly, because WASP must recruitactin monomers to the complex for potent activation (9, 12), it ispossible that the short-pitch Arp2–Arp3 dimer provides only theinitial core of a nucleus to which one or more WASP-tethered actinmonomers must be added for nucleation (7). Therefore, whetherstimulation of the short-pitch conformation without direct mono-mer tethering could activate the complex is unknown.Understanding WASP-mediated activation requires knowledge

of both the role of the short-pitch conformation and the mecha-nism by which WASP proteins could stimulate this conformationalchange. Despite numerous studies, how theWASP-CA region bindsthe Arp2/3 complex is still unclear. Biophysical and biochemicalexperiments, including analytical ultracentrifugation, isothermal ti-tration calorimetry (ITC), and cross-linking indicate that the Arp2/3complex binds twoWASP-CAmolecules (25–27). Cross-linking andlabel-transfer experiments suggest one CA binding site is on Arp3and that the second spans Arp2 and ARPC1 (Actin-related protein2/3 complex, subunit 1) (10, 25, 28–30). NMR line-broadening ex-periments and mutational analyses suggest that the C region formsan amphipathic helix when it binds to the complex (31). Togetherwith homology modeling and recent low-resolution crystal struc-tures, these data suggest that the C helix binds the barbed-endgroove of each Arp just as the V region helix binds the barbed-endgroove of actin (Fig. 1A) (25–27, 32). Less is known about inter-actions with A, but a short segment of electron density betweensubdomains 3 and 4 in Arp3 in a cocrystal structure was interpretedas the conserved tryptophan from the A region of one boundWASP (27). The second A region is thought to bind the ARPC1subunit, because this subunit in isolation can bind to VCA in amode dependent on the conserved tryptophan (33). These data,together with distance measurements reported in a recent FRETstudy (26), have led to an approximate model for CA binding ateach site (Fig. 1A). This model is consistent with the majority, butnot all, of the published data (15) and provides an importantstarting point for understanding the molecular basis for WASP-mediated conformational changes. Surprisingly, few studies haveused mutational dissection of the complex to investigate WASPbinding, and little high-resolution structural informational is avail-able (27, 34), so the model remains tentative. Furthermore, the lackof biochemical dissection leaves open the important question ofhow the engagement of WASP could stimulate the conformationalrearrangement of the complex thought to be required for activation.Here we use a newly developed engineered cysteine cross-linking

assay to investigate the role of the short-pitch conformation inArp2/3 complex activation (11). Cross-linking the engineered cys-teines to hold Arp3 and Arp2 in or near the short-pitch confor-mation bypasses the need for WASP in activation. Remarkably, thecomplex shows NPF-independent hyperactivity when cross-linkedin this conformation. This result indicates that the switch to theshort-pitch conformation is a critical activation step and that directactin monomer recruitment by WASP-VCA is not required forpotent activity. Our cross-linking experiments indicate that stimu-lation of the short-pitch conformation is a conserved feature ofWASP proteins and that the CA region is sufficient for this func-

tion. However, our data also suggest that the activating effect of theWASP-mediated conformational change may be masked by theslow release of WASP from nascent branch junctions. Usingstructure-based mutational analysis and WASP-Arp fusion chi-meras, we determined one mechanism by which WASP proteinscan stimulate movement of the Arp2 and Arp3 subunits into theshort-pitch conformation. We show that WASP competes forbinding to the barbed-end groove of Arp3 with the Arp3 C-terminaltail, a structural element that forms an allosteric switch responsiblefor autoinhibition of the complex. These data support a model inwhich WASP binds the Arp3 barbed-end groove, displacing theArp3 C-terminal tail to relieve autoinhibition and stimulate theshort-pitch conformation.

ResultsStimulation of the Short-Pitch Conformation Is a Conserved Featureof WASP Family NPFs. Several studies show that WASP familyproteins cause conformational changes in the Arp2/3 complex andthat these changes are likely important for activation. For example,binding of WASP-VCA to an Arp2/3 complex tagged with fluo-rescent proteins on ARPC1 and ARPC3 caused a conformationalchange detected by increased FRET (12). Mutations that blocked the

Fig. 1. Hypothesized CA binding mode and schematic of the short-pitch cross-linking assay. (A) Hypothetical model showing proposed binding sites for theWASP family CA region on the Arp2/3 complex. The Arp2/3 complex surfacerepresentation is based on the structure of the GMFγ (Glia maturation factorgamma)-bound Arp2/3 complex (PDB ID code 4JD2) (23). (B) Ribbon diagrams ofArp2 and Arp3 in the splayed (Left) and short-pitch (Right) conformationsshowing the design of the engineered cross-linking assay. The structure of theGMFγ-bound Arp2/3 complex (PDB ID code 4JD2) was used to make both panels.In the right panel, the actin filament structure of Oda et al. (50) was used tomove Arp2 into the short-pitch conformation. The engineered cysteines Arp3(L155C) and Arp2(R198C) are highlighted in cyan. The black arrow shows the di-rection of the ∼25-Å movement required to position Arp2 in the short-pitchconformation. The structure of the chemical cross-linker (BMOE) used in the short-pitch cross-linking assay is indicated. The cross-linking distance between engi-neered cysteines is 32.5 Å in the splayed conformation and ranges from∼8–13 Å indifferent models of the short-pitch conformation. (Also see SI Appendix, Fig. S1).

Rodnick-Smith et al. PNAS | Published online June 20, 2016 | E3835

BIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 24

, 202

0

conformational change decreased the activity of the complex (12).Similarly, EM reconstructions of negatively stained Arp2/3 complexesshowed the WASP protein Las17 caused structural rearrangementsinterpreted as repositioning of Arp2 (13, 14). Electron densi-ties from higher-resolution EM reconstructions (∼2.0 nm) ofN-WASP– and WAVE1-bound complexes showed a short-pitcharrangement of Arp2 and Arp3 that could template nucleation(15). We reasoned that if stimulation of the short-pitch confor-mation is important for activation, it should be a conservedfunction of WASP family proteins. To test this hypothesis, we tookadvantage of a recently developed cross-linking assay in whichengineered cysteines on budding yeast Arp2 and Arp3 cross-linkonly when the complex is in or very near the short-pitch confor-mation (Fig. 1B) (11). Treatment of the dual-cysteine engineeredArp2/3 complex with the 8-Å cross-linker BMOE (bis-maleimido-ethane) produces a high-molecular-weight band that is cross-reactive with both Arp2 and Arp3 antibodies (Fig. 2A). This banddoes not form in either of the single-cysteine complexes (Fig. 2A).Furthermore, the engineered cysteines do not significantly influencethe activity of the complex, demonstrating that the cross-linkingassay can be used to probe for the conformational switch (11). Wetested the influence of canonical WASP family members frommammals and budding yeast on cross-linking of the dual-cysteineengineered complex. Each NPF was added at multiple concentra-tions to reach saturation and remove the influence of binding-siteoccupancy on the conformation. VCA segments from N-WASP,WAVE1, WASP, and Las17 each potently increased cross-linking(Fig. 2 B and C and SI Appendix, Table S1). In 1-min cross-linkingreactions, nearly saturating concentrations of each WASP familyprotein produced ∼23–32% cross-linked complex, a 7- to 10-foldincrease over reactions without an NPF. Multiple lines of evidenceindicate that the influence of these NPFs results not from a changein the chemical reactivity of individual engineered cysteine residuesbut instead from increased population of the short-pitch confor-mation. First, although WASP binding could influence the pKa of

the cysteine thiol groups and formation of the thiolate, the highconcentration of negative charges of the WASP A regions wouldbe more likely to decrease than increase cysteine reactivity (35).Second, a chemically conservative point mutation in N-WASPpotently reduces stimulation of the short-pitch cross-linking evenat saturation (see below). This residue was previously de-termined by FRET to influence WASP-mediated conformationalchanges in the complex (12).To understand the structure of the BMOE cross-linked complex

better, we used morphing software to move Arp2 from its positionin an inactive (splayed) crystal structure [Protein Data Bank (PDB)ID code 4JD2] to its short-pitch position in the low-resolution EMstructure of the Arp2/3 complex at a branch junction (24). At eachpoint in the trajectory we calculated the solvent-accessible surfacecross-linking distance between the engineered cysteines (36). Giventhe cross-linker link of ∼8 Å for BMOE, this analysis suggests thatthe cross-linked complex is either in or very near the short-pitchconformation (SI Appendix, Fig. S1). Therefore, for simplicity, wewill refer to the covalently trapped conformation as the short-pitchconformation in the sections that follow. We discuss the possiblelimitations of this interpretation in Discussion. Given these con-siderations, we conclude from the data above that diverse WASPproteins stimulate the movement of Arp2 and Arp3 into the short-pitch “filament-like” arrangement. Importantly, N-WASP-CAalone stimulated this conformation as potently as N-WASP-VCA,as is consistent with binding assays showing that CA but not V in-teracts with Arp2/3 complex (Fig. 2 B and C) (10).

CA Binding to Both Arp2 and Arp3 Is Important for Stimulation of theShort-Pitch Conformation. Our data indicate that diverse WASPproteins stimulate the short-pitch conformation, suggesting that thisstructural state is important for activation. To understand betterthe role of the short-pitch conformation and how it is promoted,we attempted to create a “locked” short-pitch complex by fusingWASP-CA segments onto the C terminus of either Arp2 or Arp3

Fig. 2. Stimulation of the short-pitch conformation is a conserved feature of WASP family NPFs. (A) Single- or two-color Western blots of short-pitch cross-linkingreactions containing 1 μMwild-type (wt), dual-cysteine (Cys-Arp), or single-cysteine (Arp2-Cys or Arp3-Cys) Saccharomyces cerevisiae (Sc) Arp2/3 complexes and 25 μMBMOE reacted for 0 or 5 min. (B) Single-color Western blot of 1-min cross-linking reactions containing 1 μM dual cysteine (Cys-Arp), 25 μM BMOE, and 30 μM WASPfamily protein, as indicated. (C) Quantification of reactions similar to those described in B, except that multiple concentrations were tested. Concentrations ofWASP-CA used (expressed in micromolar) are shown in parentheses. Error bars show SE from three reactions.

E3836 | www.pnas.org/cgi/doi/10.1073/pnas.1517798113 Rodnick-Smith et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 24

, 202

0

in the budding yeast Arp2/3 complex. We designed these fusionsbased on the binding mode previously proposed for WASP-CA oneach Arp (Fig. 1A) (25–27). Fusions were made in the context ofthe engineered cysteine mutations to allow us to test for formationof the short-pitch conformation. We reasoned that this design po-tentially could create a constitutively activated complex and pro-vides a method to dissect the importance of CA engagement ateach binding site. Furthermore, the fused WASP segments lack Vregions, conferring two additional advantages. First, the fused CAcannot recruit monomers to the complex, potentially allowing us toisolate the role of CA-stimulated conformational changes in acti-vation. Second, the V region is proposed to block the elongation ofnascent branch junctions, necessitating WASP release (32, 37). Bynot including the V region, we aimed to eliminate the requirementfor WASP release in the nucleation reaction. We show belowthat the fused CA segments engage the complex properly. Un-expectedly, we found that although CA fusion to either Arp2 orArp3 stimulated the short-pitch conformation, the effect wasmodest (Fig. 3 A–C and SI Appendix, Table S2). Both CA fusionsfailed to cross-link to the levels observed for the wild-type complexsaturated with free N-WASP-CA. These observations suggest thatthe binding of WASP-CA to both Arp2 and Arp3 is required tostimulate the conformational switch maximally. To challenge thishypothesis, we added free CA to each of the fusion complexes.Free WASP-CA added to the Arp3-CA fusion complex increasedcross-linking to a level nearly identical to that observed with wild-type complex with saturating CA (23 ± 2% versus 26 ± 2%),demonstrating that binding to both sites is required for potentshort-pitch stimulation (Fig. 3 A–C). This result also shows thatCA fused to Arp3 is nearly fully functional in its ability to stim-ulate the short-pitch conformation, because the addition of freeCA (to saturate the unoccupied Arp2 site) increases cross-linkingefficiency to nearly the same level as CA-saturated wild-type com-plex. The addition of CA to the Arp2-CA fusion complex yieldedless short-pitch cross-linking than in CA-saturated wild-type com-plex (19 ± 2% versus 26 ± 2%), suggesting that the Arp2-CA fusionis only partially functional in stimulating the short-pitch conforma-

tion, potentially because of a minor defect in the way the fused CAengages Arp2. We also attempted to create a dual CA-fusioncomplex, but sporulation of diploid yeast strains containing bothArp3-CA and Arp2-CA chimeras never yielded a haploid straincontaining both chimeras in the absence of the wild-type Arps, in-dicating that the dual CA fusion complex does not support viability.

CA Release from Arp3 Is Likely Important for Nucleation. We nextasked how fusion of CA to the complex influenced its nucleationactivity. Specifically, we wondered if the modest increased pop-ulation of the short-pitch conformation in each CA fusion wouldresult in increased nucleation activity of these complexes. Asexpected, the Arp2-CA fusion complex showed increased NPF-independent activity in pyrene actin polymerization assays com-pared with the wild-type complex (Fig. 3D). In contrast, CA fusedto Arp3 potently inhibited the complex, even though it stimu-lated the short-pitch conformation (Fig. 3D). Previous experi-ments showed that WASP must be released from the nascentbranch junction before nucleation (19), so we hypothesized thatthe decreased activity of the Arp3-CA fusion complex results fromslow (or no) CA release. To test this hypothesis, we measured theinfluence of free CA on the activity of wild-type and Arp2-CAfusion complexes. The addition of free CA blocked the activity ofboth these complexes in pyrene actin polymerization assays (Fig.3E). Therefore, these data are consistent with a model in whichslow CA release stalls branching nucleation. This result was un-anticipated, because previous structural modeling predicted that V(and not CA) sterically blocks actin monomers from associatingwith nascent branches to initiate elongation, necessitating WASPrelease (32, 37). Our data suggest that CA alone can block elon-gation. However, it is important to note that we currently cannoteliminate the possibility that the fusion of CA to Arp3 may inhibitthe complex for reasons unrelated to CA release. For instance,one study reported that CA binding to one of the two sites couldinhibit the complex from binding to preformed filaments, so fusedCA could block activation by inhibiting filament binding (27).

Fig. 3. CA binding to both Arp2 and Arp3 is importantfor stimulating the short-pitch conformation change.(A) Anti-Arp3 Western blots of short-pitch cross-linkingassays containing 1 μM wild-type or CA-fusion ScArp2/3complexes (with dual-cysteine mutations) and 25 μMBMOE reacted for 1 min in the presence or absence of20 μM N-WASP-CA. (B) Quantification of reactions rununder the conditions described in A. Error bars show SEof three reactions. (C) Cartoon summarizing the short-pitch cross-linking results for wild-type and CA fusioncomplexes with or without N-WASP-CA. The averagepercentage of short-pitch cross-linked product formedunder each condition is indicated. (D) Maximumpolymerization rate vs. concentration of wild-type orArp-CA fusion complexes in 3 μM 15% pyrene actinpolymerization assays. (E) Time courses of 3 μM 15%pyrene actin polymerization assays containing 100 nMwild-type or Arp2-CA fusion complex with or without20 μM N-WASP-CA. AU, arbitrary units.

Rodnick-Smith et al. PNAS | Published online June 20, 2016 | E3837

BIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 24

, 202

0

Locking the Arp2/3 Complex into the Short-Pitch ConformationCauses Potent NPF-Independent Activity. We created the CA fu-sion complexes to isolate the influence of the short-pitch confor-mation on the branching nucleation reaction. However, the CAfusions had complicated effects on the reaction, both stimulatingthe short-pitch conformation and presumably decreasing nucleationby preventing CA release. Therefore, we sought a different strategyto probe the role of the short-pitch conformation. Our cross-linkingexperiments showed that, even without WASP, a small fraction ofthe complex adopts the short-pitch conformation (Fig. 2 A and C),consistent with the weak NPF-independent activity of the buddingyeast complex (13, 38). Therefore, we reasoned that cross-linkingcould be used to trap a fraction of the complex in the short-pitchconformation to test its influence on nucleation activity directly. Wecross-linked the dual-cysteine complex with BMOE, quenched thereaction at various time points with DTT, and tested the product inpyrene actin polymerization assays without NPFs (Fig. 4). Re-markably, cross-linking potently activated the NPF-independentactivity of the complex, demonstrating that adoption of the short-pitch conformation is a critical activation step (Fig. 4). These resultswere unexpected because, as noted above, direct recruitment ofactin monomers by WASP is required for WASP-mediated acti-vation of the Arp2/3 complex (9). Our data demonstrate that, evenin the absence of VCA-mediated monomer tethering, the complexis activated once it is locked in the short-pitch conformation.

The Complex Locked into the Short-Pitch Conformation Is Hyperactivated.Our data suggest the dominant activating influence of WASP onthe kinetics of branching nucleation stems from its influence on theconformation of the Arp2/3 complex. However, WASP-mediatedactivation of the Arp2/3 complex involves additional steps that alsocould influence branching rates, including direct monomer re-cruitment and WASP release from the branch junctions. To un-derstand better the relative contribution of each of these steps tothe reaction kinetics, we directly compared the NPF-independentactivity of the cross-linked complex and the WASP-activated non–cross-linked complex. To do so, we first separated the cross-linkedcomplex from the non–cross-linked complex using an affinity col-umn charged with GST-tagged Dip1, a recently described Arp2/3complex activator which binds preferentially to the cross-linkedcomplex (SI Appendix, Fig. S2) (5). This protocol yielded 98% purecross-linked complex (Fig. 5A). In pyrene actin polymerization as-says, 10 nM purified cross-linked complex produced a maximumpolymerization rate of 40.6 ± 2.2 nM/s with a time to half-maximal

polymerization (t1/2) of 33.3 ± 5.8 s (Fig. 5 B–D). By comparison,activation of the non–cross-linked complex by the optimal con-centration of N-WASP-VCA produced a slower maximum poly-merization rate (28.5 ± 1.0 nM/s) and an increased t1/2 (93.3 ± 5.8 s)(Fig. 5 B–D). These observations led to the surprising conclusionthat the cross-linked complex is more active than the N-WASP–activated non–cross-linked complex. In some contexts, dimerizingWASP increases NPF activity (25, 37, 39), so we also comparedoptimal activation of the non–cross-linked complex by GST-Las17-VCA and leucine zipper (LZ)-dimerized N-WASP-VCA to theactivity of the cross-linked complex. Neither of these NPFs activatedthe non–cross-linked complex as potently as cross-linking (Fig. 5 Band C). The maximum polymerization rate in reactions containingArp2/3 complex with either GST-Las17-VCA or LZ-N-WASP-VCAwas approximately fourfold less than in reactions with the sameconcentration of short-pitch cross-linked complex. These datademonstrate that the cross-linked complex is hyperactivatedcompared with the WASP-activated Arp2/3 complex. One ex-planation for this phenomenon is that the slow WASP-releasestep is bypassed in the WASP-independent activity of the short-pitch cross-linked Arp2/3 complex. To challenge this model, wetested the influence of free N-WASP-CA on the activity of thecross-linked complex. As predicted, CA inhibited nucleation,supporting a model in which locking the complex into the short-pitch conformation hyperactivates it by allowing it to bypass theslow WASP release step (Fig. 5 E and F). We note that otherattributes of the cross-linked complex also may contribute to itshyperactivity. For instance, the WASP-bound Arp2/3 complexmay not fully populate the short-pitch state, leading to decreasednucleation rates compared with the cross-linked complex. Inaddition, if the cross-linked complex requires preformed fila-ments for nucleation, CA could inhibit it by decreasing actinfilament binding (27). However, CA did not block the cross-linked complex from binding filaments in copelleting assays, ar-guing against this mechanism (SI Appendix, Fig. S3). Finally,inhibition of spontaneous nucleation by WASP also may de-crease the maximal polymerization rate even at optimal WASPconcentrations, leading to greater activity of the short-pitchcross-linked complex (16).

Residues in the C Region Influence the Ability of WASP to Stimulatethe Short-Pitch Conformational Change.How CA binding stimulatesthe movement of Arp2 into the short-pitch position is unknown. Asa first step in dissecting the mechanism, we used the cross-linking

Fig. 4. Locking the Arp2/3 complex into the short-pitch conformation bypasses the requirement for NPFsin activation. (A) Western blot of SDS/PAGE gel of cross-linking reactions containing 1 μMdual-cysteine ScArp2/3complex (Cys-Arp) and 25 μM BMOE. Reactions werequenched with 1 mM DTT at the times indicated.(B) Quantification of data in A. Error bars show SEsfrom three separate reactions. (C) Time course of pyreneactin polymerization for assays containing 3 μM 15%pyrene-labeled actin and 10 nM dual-cysteine complexcross-linked for the times indicated. (D) Fold activationof the dual-cysteine complex calculated by dividing themaximum polymerization rate at a given cross-linkingtime by the maximum polymerization rate of the non–cross-linked complex. Data were taken from C. (E) Thenumber of barbed ends created plotted versus time forreactions in C.

E3838 | www.pnas.org/cgi/doi/10.1073/pnas.1517798113 Rodnick-Smith et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 24

, 202

0

assay to test the role of specific WASP-CA residues in stimulatingthe conformational change. The C consensus sequence is charac-terized by several conserved hydrophobic residues (Fig. 6A) (31).NMR line-broadening experiments and mutational analyses sug-gest these hydrophobic residues occupy one face of an amphipathichelix that forms when WASP binds the Arp2/3 complex (31). Im-mediately C-terminal to the proposed amphipathic helix is a con-served arginine, R477 in WASP. The R477K (R478K in N-WASP)mutation is found in patients with Wiskott–Aldrich syndrome andreduces WASP-mediated activation of the Arp2/3 complex (Fig. 6B)

(9, 40). This mutation decreased cross-linking by ∼2.5-fold atsaturation compared with wild-type N-WASP, demonstrating theimportance of this residue in stimulating the short-pitch switch(Fig. 6C and SI Appendix, Fig. S4 and Table S3). Previous datashowed that another mutation within the C region, WASP L470A(L471A in N-WASP), decreased activity in polymerization assaysand blocked a conformational change measured by FRET, sug-gesting that this residue also may be important for stimulating theshort-pitch conformation (12, 31). Consistent with this hypothesis,the L471A mutation decreased short-pitch cross-linking by ap-proximately threefold compared with wild-type N-WASP-VCA(Fig. 6C). These observations support a model in which contactswith the C region are critical for stimulating the short-pitch con-formation and identify two key residues for stimulating this switch.The A region of WASP family proteins is characterized by a

conserved tryptophan surrounded by several acidic residues (Fig. 6A).The A region confers most of the Arp2/3 complex binding affinity toWASP proteins (9). The acidic residues in A play an important rolein branching nucleation kinetics. Specifically, deletion of acidic resi-dues in N-WASP-VCA increases the affinity of N-WASP for thecomplex and decreases branching nucleation rates (19), whereasaddition of acidic residues to WAVE1 increases nucleation rates(30). To test the influence of the acidic residues in stimulating theshort-pitch switch, we created a (monomeric) N-WASP constructwith five acidic resides deleted from A. This construct stimulatedcross-linking to the same extent as the wild-type N-WASP-VCA (Fig.6C and SI Appendix, Fig. S4 and Table S3). Therefore, decreasedactivation of the complex by the N-WASP(ΔDEDED) mutant can-not be explained by a failure to induce the short-pitch conformationand instead is likely caused by the slow release of the N-WASP(ΔDEDED) from branch junctions (19).

CA–Fusion Complexes Suggest the N-WASP C Region Binds to theBarbed-End Grooves of Arp2 and Arp3. Our data demonstrate thatCA regions from diverse WASP proteins stimulate the short-pitchconformation and that residues in the C region are critical forstimulating the switch. However, understanding how C stimulatesthis conformation requires knowledge of how C interacts with thecomplex. Based on biochemical and biophysical studies describedabove, WASP-C has been proposed to interact with the barbed-endgrooves of Arp2 and Arp3 (25–27, 32, 34). The C termini of Arp2and Arp3 terminate between subdomains 1 and 3, filling the backportion of the barbed-end grooves in a position that either overlapsor sits immediately adjacent to the predicted binding sites for C(Fig. 7) (20). This arrangement provides a fortuitous opportunity totest the proposed binding mode. As described above, we createdchimeric subunits in which the CA region of N-WASP is fused tothe C terminus of either Arp2 or Arp3 (Fig. 7 A–C). Based on theproposed binding mode, the tip of the Arp3 C terminus stericallyoverlaps the WASP C helix, so we truncated Arp3 by seven resi-dues, removing the tip of the C-terminal tail and connecting thebase of the tail to the N terminus of the N-WASP CA region with ashort linker (GSG). We reasoned that fused CA would engageArp3 properly only if the front of the barbed-end grove is thecorrect binding site for C, as predicted by the model. A similarstrategy was used for Arp2, except that the Arp2 C terminus isshorter, so the binding model predicted that truncation of onlythree residues from the C terminus would allow the C helix to foldinto its predicted position in the groove (Fig. 7C).To determine if fused CA segments are properly engaged, we

asked if they could compete with VCA for binding to the complex.To do so, we first reacted the bifunctional cross-linker benzophe-none-4-maleimide (B4M) with N-WASP-VCA at an engineeredcysteine residue (T464C) immediately N-terminal to the C region(Fig. 7D). We used UV radiation to cross-link this adduct,N-WASP-VCA-T464C-B4M, to the wild-type complex, producingVCA cross-linked adducts of both Arp2 and Arp3, as expectedbased on previously published cross-linking studies (Fig. 7 E and F)(28). Unlabeled VCA competes with N-WASP-VCA-T464C-B4Mfor cross-linking to the wild-type complex at the Arp2 and Arp3sites, demonstrating the reaction is specific. Importantly, the

Fig. 5. The short-pitch cross-linked complex is more active than the WASP-activated non–cross-linked complex. (A, Left) Coomassie-stained SDS/PAGE gelshowing the pre–cross-linked dual-cysteine complex, the post–cross-linking re-action, and the purified cross-linked complex after passing through a MonoQcolumn and a GST-Dip1 affinity column. (Right) Anti-Arp3 Western blot of pu-rified cross-linked and non–cross-linked samples. (B) Time courses of 3 μM 15%pyrene actin polymerization for reactions containing 10 nM dual cross-linkedcysteine ScArp2/3 complex (“xlinked Arp2/3”) or 10 nM non–cross-linked complexwith either no NPF (“dual Cys Arp2/3”), 250 nM N-WASP-VCA, or 50 nM GST-Las17-VCA, as indicated. Multiple reaction time courses are shown for the fastestreactions. (C) Maximum polymerization rate versus NPF concentration for reac-tions containing 10 nM non–cross-linked complex with N-WASP-VCA, GST-Las17-VCA, or LZ-dimerized N-WASP-VCA in conditions identical to those in B. Theaveragemaximum polymerization rate for purified cross-linked complex withoutNPFs is shown as a red circle. Error bars indicate the SD (n = 3). (D) Comparison ofmaximum polymerization and time to t1/2 polymerization for 10 nM cross-linkedcomplex versus 10 nM non–cross-linked complex with 250 nM N-WASP-VCA.Error bars indicate the SD (n = 3). (E) Time course of polymerization of a reactioncontaining 3 μM15% pyrene-labeled actin, 2.5 nM purified cross-linked complex,and 0–40 μM N-WASP-CA, as indicated. (F) Plot of the maximum polymerizationrate versus N-WASP-CA concentration for reactions in E.

Rodnick-Smith et al. PNAS | Published online June 20, 2016 | E3839

BIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 24

, 202

0

Arp3-CA fusion completely blocked N-WASP-VCA-T464C-B4Mfrom cross-linking to Arp3 (Fig. 7 E–G). This result suggests thatCA properly engages its site on Arp3 in the fusion chimera,providing support for a model in which the C region binds theArp3 barbed-end groove. Fusion of CA to Arp3 did not influenceN-WASP-VCA-T464C-B4M cross-linking to Arp2, indicatingthat the Arp3-CA fusion specifically blocks binding of CA to theArp3 site. We repeated the B4M cross-linking experiments totest whether the Arp2-CA fusion engages the Arp2 CA-bindingsite. Under our reaction conditions, 67% of Arp2 from wild-typecomplex cross-linked to N-WASP-VCA-T464C-B4M, whereasonly 4.3% of the Arp2 was cross-linked in the Arp2-CA chimera(Fig. 7 E–G). This potent reduction in cross-linking demonstratesthat CA fused to the C terminus of Arp2 blocks the Arp2 CA-binding site. Together these experiments support the proposedbinding modes, in which the C region ofWASP binds the barbed-endgroove of either Arp2 or Arp3. The proposed binding modes also areconsistent with recent low-resolution X-ray crystal structures ofArp2/3 complex crystals grown in the presence of WASP-CA, whichshow putative electron density for WASP-C bound to the barbedends of Arp2 and Arp3 (34).

WASP-C Likely Displaces the C Terminus from the Barbed-End Grooveof Arp3 to Relieve Autoinhibition and Stimulate the Short-PitchConformation. How could WASP-CA stimulate the short-pitchconformation? We analyzed crystal structures of inactive Arp2/3complex in the context of the CA-binding model to identify inter-actions that could destabilize intersubunit contacts important forholding the complex in the splayed conformation. We also analyzedlow-resolution EM models of the Arp2/3 complex at a branchjunction to identify WASP-CA interactions that could stabilize theshort-pitch conformation (24). These analyses, in conjunction withresults from a recent study (41), led us to hypothesize that inter-actions of WASP-C with the Arp3 barbed-end groove are criticalfor stimulating the short-pitch conformation. We recently foundthat the Arp3 C terminus is an autoregulatory switch important forholding the complex inactive in the absence of NPFs. The Arp3C-terminal tail engages the Arp3 barbed-end groove and occupiesa critical position at the splayed Arp2–Arp3 interface (Fig. 8 A and

B). The base of the Arp3 C-terminal tail directly contacts Arp2 tostabilize the splayed conformation, whereas the most C-terminalresidues in the Arp3 tail (the “tip”) do not contact Arp2 directly butinstead fill the front portion of the barbed-end groove (20–22).Deletion of the entire Arp3 C-terminal tail in either budding orfission yeast Arp2/3 complexes stimulates the short-pitch confor-mation and increases NPF-independent activity (41). Furthermore,mutations that disrupt interactions of the tip of the tail with thebarbed-end groove also stimulate the short-pitch conformation, sup-porting a model in which release of the tip from the barbed-endgroove allosterically relieves autoinhibition by destabilizing thesplayed Arp2–Arp3 interface (41) (Fig. 8A). Importantly, our CA-fusiondata, along with recently published low-resolution crystal structures(34), suggest that WASP-C engages the barbed-end groove of Arp3,where it would overlap with the tip of the C-terminal tail (Figs. 7 and8 A and B). Therefore, we hypothesized that WASP could activatethe complex by displacing the Arp3 C-terminal tail from the Arp3barbed-end groove. To test this hypothesis, we first asked if the Cterminus and WASP-CA compete for binding to the complex. Weused fluorescence anisotropy to measure the affinity of rhodamine-labeled WASP proteins for the budding yeast and fission yeastArp3ΔC complexes. Using a single-site binding model (SI Appendix,SI Materials and Methods and ref. 27), we found the affinity ofrhodamine-labeled WASP proteins increased four- to sixfold for theArp3ΔC mutants compared with wild-type complexes (Fig. 8C andSI Appendix, Fig. S5). To eliminate the influence of the rhodaminelabel, we ran competition assays to measure binding of unlabeledWASP proteins. These experiments showed that the Arp3ΔCcomplexes bind to WASP proteins 4- to 13-fold more tightly thanwild-type complex (Fig. 8C). Although the fluorescence anisotropyassay does not allow us to distinguish between binding to Arp2 or toArp3, we took advantage of the observation that truncation of theArp3 C terminus causes a previously unreactive endogenous cysteinein Arp3, C426, to react in the BMOE cross-linking assay (Fig. 8D).Arp3 C426 lines the barbed-end groove (Fig. 8B), and deletion ofthe C terminus allows this residue to cross-link with Arp2 to form anon–short-pitch cross-link that migrates below the short-pitch cross-linked Arp3–Arp2 adduct (Fig. 8D and SI Appendix, Fig. S6). Wereasoned that if CA binds to the barbed-end groove of Arp3, it coulddirectly block the reactivity of C426. Consistent with this prediction,the addition of CA to the Arp3ΔC complex decreased cross-linkingbetween Arp2 and Arp3 C426 (Fig. 8D). To provide additionalevidence that WASP-C binds the barbed end of Arp3, we asked ifWASP-C could be directly cross-linked to C426 in Arp3. To thisend, we designed a VCA construct [N-WASP-VCA(V468C)] with acysteine residue immediately N-terminal to C (SI Appendix, Fig. S7).N-WASP-VCA(V468C) formed a disulfide cross-link with C426 inthe Arp3ΔC complex in the presence of CuSO4 (Fig. 8E). The cross-linking was abolished by the addition of cysteine-free N-WASP-CA,indicating that the cross-linking occurs between bound N-WASP-VCA(V468C) and Arp3. These data support a model in whichWASP-C competes directly with the Arp3 C terminus to displace itfrom the barbed-end groove, revealing a structural mechanism bywhich WASP binding can stimulate the short-pitch conformation.However, we note that in the absence of high-resolution structuresof Arp3 with bound WASP-CA, we cannot eliminate the possibilitythat competition is not direct but instead involves an allosteric linkbetween the WASP-CA–binding site and the Arp3 C terminus.

DiscussionStimulating the Short-Pitch Conformation Is a Critical NPF Function.Previous models for activation of the Arp2/3 complex ascribed twomajor functions to WASP: (i) stimulation of an activating confor-mational change in the complex (11–13, 15) and (ii) direct re-cruitment of actin monomers to assemble a nucleus consisting ofArp2, Arp3, and one or more actin monomers (9, 10, 12). Therelative contributions of each of these functions was unknown. Herewe show that locking the Arp2/3 complex into the short-pitchconformation potently activates it without NPFs, demonstrating thatdirect monomer tethering is not required to assemble the nucleus,as previously anticipated. Therefore, whether direct actin monomer

Fig. 6. Residues in the C region are important for stimulating the short-pitchconformational change. (A) Sequence alignment of WASP family CA regions.Conserved hydrophobic (green boxes) and basic (red boxes) residues are in-dicated, and acidic residues in the A region are highlighted in cyan. N-WASPpoint mutations and deletion constructs investigated in this study are indicated.Bt, Bos taurus; Hs Homo sapiens; Sc, S. cerevisiae. (B) Maximum polymerizationrate versus NPF concentration for polymerization reactions containing 3 μM15% pyrene actin, 10 nM dual-cysteine Arp2/3 complex, and the indicatedconcentration of each NPF. (C) Results of 1-min short-pitch cross-linking reac-tions containing 1 μM dual-cysteine Arp2/3 complex, 25 μM BMOE, and theindicated concentrations (in micromolar) of mutant or wild-type N-WASP-VCA.Error bars show the SE from three reactions.

E3840 | www.pnas.org/cgi/doi/10.1073/pnas.1517798113 Rodnick-Smith et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 24

, 202

0

tethering is required for potent activation by any NPF likely de-pends on the identity and mechanism of the NPF and is not afundamental requirement of the branching nucleation reaction.Consistent with this hypothesis, WISH/DIP/SPIN90 (WDS) familyproteins can potently activate the Arp2/3 complex but lack knownactin monomer-binding domains (5). Although little is known aboutthis class of NPFs, we previously used the cross-linking assay toshow that WDS protein Dip1 stimulates formation of the short-pitch conformation, and our data here suggest this stimulation issufficient for Dip1-mediated activation of the complex (5). Like theWDS family proteins, type II NPFs such as cortactin and fissionyeast myosin I interact with the Arp2/3 complex and actin filamentsbut not with actin monomers (37, 42–44). Unlike WDS proteins,type II NPFs are generally weak activators on their own (7). Recentexperiments indicate that, at least in the case of cortactin, this weakactivity is caused by an inability to deliver monomers directly to theArp2/3 complex (42). Type II NPFs typically interact with actinmonomer-binding proteins that allow them to deliver monomers tothe complex indirectly and potently increase their ability to activate(7, 45, 46). Therefore, as with WASP, a requirement for actinmonomer recruitment may be programmed into the activationmechanism of type II NPFs. An important future direction will be todetermine why actin monomer recruitment is required for potentactivation by some NPFs but not by others and how this re-quirement is structurally encoded into the activation mechanism.Here we propose a model in which WASP-CA stimulates the

short-pitch conformation by displacing the Arp3 C-terminal tailfrom the barbed-end groove, thereby triggering destabilization of

the splayed conformation. Our data indicate that this mechanismmust work coordinately with other activation triggers. For example,in a recent study, we showed that ATP bound to the Arp3 nucle-otide cleft allosterically disrupts interactions between the Arp3 tailand the barbed-end groove, providing another mechanism to de-stabilize the splayed conformation (41). Structural features otherthan the Arp3 C-terminal tail also must be involved in triggeringactivation. For example, here we show that CA engagement withArp2 is also important for stimulation of the short-pitch confor-mation. In addition, phosphorylation of Arp2 near its interface withARPC4/ARPC2 is required for activation of the complex in somespecies and is thought to destabilize the splayed conformation (8,47). Understanding how the Arp2/3 complex structurally integrateseach of these activating responses remains an important openquestion in the field.

The Role of WASP, Actin Monomers, and Actin Filaments in Activation.Because actin monomers, WASP, and actin filaments are requiredto activate the complex in most contexts (4, 5, 9, 10), it has gen-erally been assumed that cooperation between each of these fac-tors is required to shift the conformation into the short-pitchconformation, thereby stimulating activation. Our data led us toreevaluate the relationship between activity and conformation.Specifically, the observation that WASP-VCA (or CA) alone canstimulate cross-linking between short-pitch engineered cysteinesmakes it unclear why WASP-recruited monomers and actin fila-ments also are required for activation. Because the cross-linkingassay conformationally traps the complex, one possibility is that

Fig. 7. WASP-CA fused to the C terminus of Arp2 or Arp3 blocks the binding of soluble WASP-VCA to the CA-fusion subunit. (A) Sequence alignment of the Cterminus of Arp2, Arp3, and actin, showing the design of Arp CA-fusion constructs. Residues conserved in Arp2 and Arp3 (cyan boxes), Arp2 only (pink boxes), orArp3 only (green boxes) are indicated. Residues in Arp3 that mark the tip (blue line) and base (red line) of the Arp3 C-terminal tail are indicated. The majority ofresidues from the fused CA segments [indicated by “(CA)”] were omitted from the alignment. (B) Hypothetical model of the Arp3-CA fusion structure. The fusedCA sequence is shown in magenta, and the linker is in green. The model was constructed from the crystal structure of the Arp2/3 complex with bound ATP (PDB IDcode 2P9K) (22), with the WASP-C region modeled into position based on a WASP-V–bound actin structure (PDB ID code 2A3Z) (32). The A region was placedbetween subdomains 3 and 4 based on the structure (PDB ID code 3RSE) (27). (C) Hypothetical model of the Arp2-CA fusion structure. The model was constructedfrom the crystal structure of the GMF-bound Arp2/3 complex (PDB ID code 4JD2) (23), withWASP-C modeled into position as described above. (D) Schematic of theB4M cross-linking assay to probe WASP-CA binding at each site on the complex. (E) Anti-Arp3 Western blot of B4M UV cross-linking reactions containing 1.5 μMdual-cysteine (“wt”), Arp2-CA fusion, or Arp3-CA fusion complexes with 200 μM ATP and with no NPF, 4 μM N-WASP-VCA(T464C)-B4M, or 100 μM LZ-dimerizedVCA (LZ-VCA), as indicated. All three complexes harbor the dual-cysteine engineered mutations. (F) Reactions from E analyzed byWestern blotting with anti-Arp2antibody. (G) Quantification of B4M cross-linking reactions as described in E and F. Error bars are SEs from three reactions.

Rodnick-Smith et al. PNAS | Published online June 20, 2016 | E3841

BIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 24

, 202

0

WASP-bound Arp2/3 complex may not populate the short-pitchstate significantly, but by cross-linking we amplify a functionallyinsignificant effect of WASP on the concentration of short-pitchcomplex. Previous experiments indeed demonstrate that WASPbinding is not sufficient to populate the short-pitch conformationfully, because WASP-recruited actin monomers potently increasethe amount of short-pitch cross-linking over reactions with WASPalone (11). However, comparisons of the relative amount of com-plex cross-linked under different conditions, although yielding noinformation about absolute equilibrium populations, indicate thebias toward the short-pitch conformation in the presence of WASPalone should be functionally significant. Without WASP, only ∼3%of the Arp2/3 complex was cross-linked after 1 min of cross-linking,suggesting that the complex adopts the short-pitch conformationonly weakly in the absence of NPF. However, this amount isfunctionally significant, because the ScArp2/3 complex showsNPF-independent activity in pyrene actin polymerization assays.Therefore, the ∼7- to 10-fold increase in cross-linking stimulated byWASP-CA (or VCA) proteins without actin probably reports on

functionally significant increases in the short-pitch conforma-tion. Several previously published biochemical and biophysicalexperiments agree with this interpretation, because they indi-cate that WASP alone causes a significant shift in the confor-mational equilibrium of the complex, presumably toward anactivated state (12–15).A second explanation for the insufficiency of WASP-CA in ac-

tivation (despite its influence on the conformation of the complex)is that WASP-CA binding stimulates an intermediate conformationclose to but distinct from the short-pitch arrangement. Our datacannot eliminate this possibility, because BMOE cross-linking maynot necessarily constrain the complex strictly to the short-pitchstate, as noted above. From an intermediate WASP-CA–stimulatedconformation, WASP-recruited actin monomers and filamentscould trigger population of the “fully” short-pitch state. However,recent EM reconstructions of negatively stained NPF-bound Arp2/3complexes argue against intermediate WASP-bound conformations,because the particles segregated into only two different conforma-tional classes (15). One class corresponded to the inactive splayedconformation seen in crystal structures, and the second (NPF-bound) class produced 3D density consistent with a filament-like (short-pitch) arrangement of Arp2 and Arp3. This study isin contrast with slightly lower-resolution reconstructions of NPF-bound Arp2/3 complexes that indicated a third conformational classcorresponding to an intermediate state (13). Thus, the precisestructures of WASP-bound complexes and their role in activationremain an important open question.Finally, another possibility is that actin filaments (and potentially

WASP-recruited monomers) could stimulate activating steps dis-tinct from the adoption of short-pitch conformation. In addition tothe movement of Arp2 and Arp3 into the filament-like arrange-ment, the EM branch-junction model describes intramolecularrearrangements within the Arp subunits that may be important foractivation and could be stimulated by actin filaments or monomers.This hypothesis is attractive, given recent experiments suggestingthat WASP must be released from nascent branch junctions fornucleation to occur (19, 37). Conformational changes initiated byfilaments but distinct from the short-pitch switch could help accel-erate WASP release, potentially explaining the requirement for fil-aments in activation. We anticipate that in addition to the engineeredshort-pitch dual-cysteine Arp2/3 complex, additional biophysical toolsand high-resolution structures will be required to dissect the con-formational pathway to activation, including the molecular basis forWASP release. We note that our data provide two observations thatmay be important for understanding the release mechanism. First,although previous studies used single-molecule TIRF to showWASPrelease triggers nucleation, these studies used only dimeric WASPproteins (19). Experiments here indicate slow detachment of mo-nomeric WASP proteins also might limit branching rates. Second,previous structural models suggested that the WASP V region blocksmonomers from associating with nascent branch junctions (32, 37),explaining the requirement for WASP release. Here we show thatCA alone stalls nucleation, demonstrating that the structural featuresthat program WASP release into the nucleation mechanism aremore complicated than previously anticipated.

Role of WASP Engagement at Each Binding Site. We show that en-gagement of WASP-CA with Arp3 is critical for stimulating theshort-pitch conformation. However, experiments with the CA fu-sion complexes revealed that engagement of WASP-CA at theArp2–ARPC1 site is also important for this conformationalchange. WASP engaged at this site likely uses a mechanism otherthan C-terminal tail displacement, because, unlike Arp3, the Arp2C terminus does not make contacts that stabilize the splayed(inactive) conformation (20–23). ITC binding measurements showa 10-fold difference in the binding affinity at each of the twoWASP-CA–binding sites (26, 27), and it will be important to de-termine how the asymmetry in binding affinities influences thereaction kinetics. We note that our studies indicate that Arp3harbors the high-affinity site, because the fluorescence anisotropybinding assay measures binding to the high-affinity site (27), and

Fig. 8. WASP competes with the Arp3 C terminus for binding the barbed-endgroove of Arp3. (A) Model for the relief of autoinhibition of the Arp2/3complex by WASP-CA binding. B, base of the Arp3 C-terminal tail; NBC,nucleotide-binding cleft; T, tip of the Arp3 C-terminal tail. Subdomains of Arp3and Arp3 are numbered 1–4. (B) Hypothetical model of WASP-CA (magenta)bound to Arp3 showing a clash between WASP-C and the tip of the Arp3C-terminal tail (yellow). Cysteine 426, a residue that becomes reactive when theC terminus is deleted, is shown as green spheres. The Arp3 ribbon diagram isbased on PDB ID code 1K8K, and WASP-C is modeled into barbed-end grooveof Arp3 based on the WASP-V–bound actin structure (PDB ID code 2A3Z) (32).(C) Tabulated results from fluorescence anisotropy binding assays. (D) Two-color (anti-Arp3/anti-Arp2, Upper) or one-color (anti-Arp3, Lower) Westernblots of cross-linking reactions containing 1.0 μM wild-type or Arp3ΔC orArp3ΔC(C426A) fusion complexes (each with dual-cysteine mutations) with orwithout 20 μM N-WASP-CA showing short-pitch (sp) or non–short-pitch (nonsp) products. (E) Anti-Arp3 Western blot of VCA cross-linking reactions con-taining 0.5 μM wild-type, Arp3ΔC, or Arp3ΔC(C426A) S. cerevisiae Arp2/3complex, 2 μM N-WASP-VCA(V468C), 50 μM CuSO4,10 mM imidazole (pH 7.0),50 mM KCl, 50 μM EGTA, 1 mMMgCl2, and 200 μMATP with or without 50 μMN-WASP-CA. Note that the Arp2/3 complexes in this assay lack the engineeredcysteines for short-pitch cross-linking.

E3842 | www.pnas.org/cgi/doi/10.1073/pnas.1517798113 Rodnick-Smith et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 24

, 202

0

affinities measured by this method increased when competitionwith the Arp3 C-terminal tail was eliminated. This conclusion is inagreement with other studies indicating that the high-affinity site ison Arp3 (48). However, some experiments point to the Arp2-ARPC1 site as the high-affinity site (26, 27, 33). For instance, arecent ITC study directly demonstrated that WASP delivers actinmonomers to the Arp2 site first, suggesting that, at least in thepresence of monomers, the Arp2 site is the higher-affinity site(49). Although the complexity of this issue has precluded a clearconsensus, we speculate that the preferred binding site may de-pend on whether WASP is bound to actin monomers.

Materials and MethodsPlease see SI Appendix for a complete description of the methods. This de-scription includes construction of budding yeast and fission yeast plasmids andstrains harboring Arp2/3 complex mutations; protein purification and labeling;pyrene actin polymerization assays; BMOE, B4M, and CuSO4 cross-linking as-says; actin filament copelleting; and fluorescence anisotropy binding assays.

ACKNOWLEDGMENTS. We thank Tom Stevens (University of Oregon) and RongLi (Stowers Institute) for sharing plasmid and yeast strains, Ken Prehoda for com-ments on the manuscript, and Will Storck for assistance with purification of theN-WASP-VCA mutants. This work was supported by National Institutes of HealthGrant R01-GM092917 (to B.J.N.) and the Pew Biomedical Scholar Program (B.J.N.).

1. Campellone KG, Welch MD (2010) A nucleator arms race: Cellular control of actinassembly. Nat Rev Mol Cell Biol 11(4):237–251.

2. Skau CT, Waterman CM (2015) Specification of Architecture and function of actinstructures by actin nucleation factors. Annu Rev Biophys 44:285–310.

3. Amann KJ, Pollard TD (2001) The Arp2/3 complex nucleates actin filament branchesfrom the sides of pre-existing filaments. Nat Cell Biol 3(3):306–310.

4. Achard V, et al. (2010) A “primer”-based mechanism underlies branched actin fila-ment network formation and motility. Curr Biol 20(5):423–428.

5. Wagner AR, Luan Q, Liu SL, Nolen BJ (2013) Dip1 defines a class of Arp2/3 complexactivators that function without preformed actin filaments. Curr Biol 23(20):1990–1998.

6. Rotty JD, Wu C, Bear JE (2013) New insights into the regulation and cellular functionsof the ARP2/3 complex. Nat Rev Mol Cell Biol 14(1):7–12.

7. Goley ED, Welch MD (2006) The ARP2/3 complex: An actin nucleator comes of age.Nat Rev Mol Cell Biol 7(10):713–726.

8. LeClaire LL, 3rd, Baumgartner M, Iwasa JH, Mullins RD, Barber DL (2008) Phosphorylationof the Arp2/3 complex is necessary to nucleate actin filaments. J Cell Biol 182(4):647–654.

9. Marchand JB, Kaiser DA, Pollard TD, Higgs HN (2001) Interaction of WASP/Scar pro-teins with actin and vertebrate Arp2/3 complex. Nat Cell Biol 3(1):76–82.

10. Kelly AE, Kranitz H, Dötsch V, Mullins RD (2006) Actin binding to the central domainof WASP/Scar proteins plays a critical role in the activation of the Arp2/3 complex.J Biol Chem 281(15):10589–10597.

11. Hetrick B, Han MS, Helgeson LA, Nolen BJ (2013) Small molecules CK-666 and CK-869inhibit actin-related protein 2/3 complex by blocking an activating conformationalchange. Chem Biol 20(5):701–712.

12. Goley ED, Rodenbusch SE, Martin AC, Welch MD (2004) Critical conformational changesin the Arp2/3 complex are induced by nucleotide and nucleation promoting factor. MolCell 16(2):269–279.

13. Rodal AA, et al. (2005) Conformational changes in the Arp2/3 complex leading toactin nucleation. Nat Struct Mol Biol 12(1):26–31.

14. Martin AC, et al. (2005) Effects of Arp2 and Arp3 nucleotide-binding pocket muta-tions on Arp2/3 complex function. J Cell Biol 168(2):315–328.

15. Xu XP, et al. (2012) Three-dimensional reconstructions of Arp2/3 complex with boundnucleation promoting factors. EMBO J 31(1):236–247.

16. Higgs HN, Blanchoin L, Pollard TD (1999) Influence of the C terminus of Wiskott-Aldrichsyndrome protein (WASp) and the Arp2/3 complex on actin polymerization. Biochemistry38(46):15212–15222.

17. Mullins RD, Heuser JA, Pollard TD (1998) The interaction of Arp2/3 complex with actin:Nucleation, high affinity pointed end capping, and formation of branching networksof filaments. Proc Natl Acad Sci USA 95(11):6181–6186.

18. Martin AC, Welch MD, Drubin DG (2006) Arp2/3 ATP hydrolysis-catalysed branchdissociation is critical for endocytic force generation. Nat Cell Biol 8(8):826–833.

19. Smith BA, et al. (2013) Three-color single molecule imaging shows WASP detachmentfrom Arp2/3 complex triggers actin filament branch formation. eLife 2:e01008.

20. Robinson RC, et al. (2001) Crystal structure of Arp2/3 complex. Science 294(5547):1679–1684.21. Nolen BJ, Littlefield RS, Pollard TD (2004) Crystal structures of actin-related protein

2/3 complex with bound ATP or ADP. Proc Natl Acad Sci USA 101(44):15627–15632.22. Nolen BJ, Pollard TD (2007) Insights into the influence of nucleotides on actin family

proteins from seven structures of Arp2/3 complex. Mol Cell 26(3):449–457.23. Luan Q, Nolen BJ (2013) Structural basis for regulation of Arp2/3 complex by GMF. Nat

Struct Mol Biol 20(9):1062–1068.24. Rouiller I, et al. (2008) The structural basis of actin filament branching by the Arp2/3

complex. J Cell Biol 180(5):887–895.25. Padrick SB, Doolittle LK, Brautigam CA, King DS, Rosen MK (2011) Arp2/3 complex is

bound and activated by two WASP proteins. Proc Natl Acad Sci USA 108(33):E472–E479.26. Boczkowska M, Rebowski G, Kast DJ, Dominguez R (2014) Structural analysis of

the transitional state of Arp2/3 complex activation by two actin-bound WCAs. NatCommun 5:3308.

27. Ti SC, Jurgenson CT, Nolen BJ, Pollard TD (2011) Structural and biochemical charac-terization of two binding sites for nucleation-promoting factor WASp-VCA on Arp2/3complex. Proc Natl Acad Sci USA 108(33):E463–E471.

28. Liu SL, Needham KM, May JR, Nolen BJ (2011) Mechanism of a concentration-dependent switch between activation and inhibition of Arp2/3 complex by coronin.J Biol Chem 286(19):17039–17046.

29. Kreishman-Deitrick M, et al. (2005) NMR analyses of the activation of the Arp2/3 com-plex by neuronal Wiskott-Aldrich syndrome protein. Biochemistry 44(46):15247–15256.

30. Zalevsky J, Grigorova I, Mullins RD (2001) Activation of the Arp2/3 complex by theListeria acta protein. Acta binds two actin monomers and three subunits of the Arp2/3complex. J Biol Chem 276(5):3468–3475.

31. Panchal SC, Kaiser DA, Torres E, Pollard TD, Rosen MK (2003) A conserved amphi-pathic helix in WASP/Scar proteins is essential for activation of Arp2/3 complex. NatStruct Biol 10(8):591–598.

32. Chereau D, et al. (2005) Actin-bound structures of Wiskott-Aldrich syndrome protein(WASP)-homology domain 2 and the implications for filament assembly. Proc NatlAcad Sci USA 102(46):16644–16649.

33. Pan F, Egile C, Lipkin T, Li R (2004) ARPC1/Arc40 mediates the interaction of the actin-related protein 2 and 3 complex with Wiskott-Aldrich syndrome protein family acti-vators. J Biol Chem 279(52):54629–54636.

34. Jurgenson CT, Pollard TD (2015) Crystals of the Arp2/3 complex in two new spacegroups with structural information about actin-related protein 2 and potential WASPbinding sites. Acta Crystallogr F Struct Biol Commun 71(Pt 9):1161–1168.

35. Britto PJ, Knipling L, Wolff J (2002) The local electrostatic environment determinescysteine reactivity of tubulin. J Biol Chem 277(32):29018–29027.

36. Kahraman A, Malmström L, Aebersold R (2011) Xwalk: Computing and visualizingdistances in cross-linking experiments. Bioinformatics 27(15):2163–2164.

37. Helgeson LA, Nolen BJ (2013) Mechanism of synergistic activation of Arp2/3 complexby cortactin and N-WASP. eLife 2:e00884.

38. Wen KK, Rubenstein PA (2005) Acceleration of yeast actin polymerization by yeast Arp2/3complex does not require an Arp2/3-activating protein. J Biol Chem 280(25):24168–24174.

39. Yamaguchi H, et al. (2000) Two tandem verprolin homology domains are necessaryfor a strong activation of Arp2/3 complex-induced actin polymerization and inductionof microspike formation by N-WASP. Proc Natl Acad Sci USA 97(23):12631–12636.

40. Derry JM, et al. (1995) WASP gene mutations in Wiskott-Aldrich syndrome andX-linked thrombocytopenia. Hum Mol Genet 4(7):1127–1135.

41. Rodnick-Smith M, Liu S-L, Balzer CJ, Luan Q, Nolen BJ (2016) Identification of an ATP-controlled allosteric switch that controls actin filament nucleation by Arp2/3 complex.Nat Commun, in press.

42. Helgeson LA, Prendergast JG, Wagner AR, Rodnick-Smith M, Nolen BJ (2014) Inter-actions with actin monomers, actin filaments, and Arp2/3 complex define the roles ofWASP family proteins and cortactin in coordinately regulating branched actin net-works. J Biol Chem 289(42):28856–28869.

43. Uruno T, et al. (2001) Activation of Arp2/3 complex-mediated actin polymerization bycortactin. Nat Cell Biol 3(3):259–266.

44. Huang C, et al. (1997) Down-regulation of the filamentous actin cross-linking activity ofcortactin by Src-mediated tyrosine phosphorylation. J Biol Chem 272(21):13911–13915.

45. Sirotkin V, Beltzner CC, Marchand JB, Pollard TD (2005) Interactions of WASp, myosin-I,and verprolin with Arp2/3 complex during actin patch assembly in fission yeast. J CellBiol 170(4):637–648.

46. Kinley AW, et al. (2003) Cortactin interacts with WIP in regulating Arp2/3 activationand membrane protrusion. Curr Biol 13(5):384–393.

47. Narayanan A, LeClaire LL, 3rd, Barber DL, Jacobson MP (2011) Phosphorylation of theArp2 subunit relieves auto-inhibitory interactions for Arp2/3 complex activation. PLOSComput Biol 7(11):e1002226.

48. Nolen BJ, Pollard TD (2008) Structure and biochemical properties of fission yeast Arp2/3complex lacking the Arp2 subunit. J Biol Chem 283(39):26490–26498.

49. Boczkowska M, Rebowski G, Dominguez R (2013) Glia maturation factor (GMF) in-teracts with Arp2/3 complex in a nucleotide state-dependent manner. J Biol Chem288(36):25683–25688.

50. Oda T, Iwasa M, Aihara T, Maéda Y, Narita A (2009) The nature of the globular- tofibrous-actin transition. Nature 457(1):441–446.

Rodnick-Smith et al. PNAS | Published online June 20, 2016 | E3843

BIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 24

, 202

0