document

© 2001 Macmillan Magazines Ltd

brief communications

NATURE CELL BIOLOGY VOL 3 MARCH 2001 http://cellbio.nature.com306

The Arp2/3 complex nucleates actin filament branches from the sides ofpre-existing filaments

Kurt J. Amann and Thomas D. Pollard*Structural Biology Laboratory, Salk Institute for Biological Studies, 10010 North Torrey Pines Road, La Jolla, California 92037, USA

*e-mail: [email protected]

Regulated assembly of actin-filament networks providesthe mechanical force that pushes forward the leadingedge of motile eukaryotic cells1 and intracellular patho-genic bacteria2 and viruses3. When activated by bindingto actin filaments and to the WA domain ofWiskott–Aldrich-syndrome protein (WASP)/Scar proteins,the Arp2/3 complex nucleates new filaments that growfrom their barbed ends4–8. The Arp2/3 complex binds tothe sides9 and pointed ends10,11 of actin filaments, local-izes to distinctive 70° actin-filament branches present inlamellae12, and forms similar branches in vitro6,8,10. Theseobservations have given rise to the dendritic nucleationmodel for actin-network assembly10,13, in which theArp2/3 complex initiates branches on the sides of olderfilaments. Recently, however, an alternative mechanismfor branch formation has been proposed8. In the ‘barbed-end nucleation’ model, the Arp2/3 complex binds to thefree barbed end of a filament and two filaments subse-quently grow from the branch. Here we report the use ofkinetic and microscopic experiments to distinguishbetween these models. Our results indicate that the acti-vated Arp2/3 complex preferentially nucleates filamentbranches directly on the sides of pre-existing filaments.

We used light microscopy to examine filaments in samplesobtained at intervals from bulk polymerization reactions, todistinguish side branching from barbed-end branching. The

morphology of the products reveals much about the mechanism ofbranch formation. Fluorescent phalloidins can be used to stabilizeand visualize the filaments6. Phalloidin also inhibits phosphaterelease after ATP hydrolysis. Thus, filaments stained with phal-loidin during polymerization contain ADP and inorganic phos-phate (ADP–Pi), whereas filaments stained with phalloidin at timepoints after polymerization contain ADP.

We measured the lengths of branches formed during sponta-neous polymerization of monomeric actin in the presence of acti-vated Arp2/3 complex and rhodamine–phalloidin. The barbedends of both daughter6 and mother10 filaments extended from theacute angle of the branch formed by the Arp2/3 complex. Ifbranches arise from Arp2/3 complex bound to the barbed ends offilaments, and if the two branches are initiated simultaneously andelongate at equal rates, they should have similar lengths. On theother hand, if branches emanate from the sides of filaments, thelengths of the two filaments distal to the branch would not be cor-related.

The reaction produced branched structures (Fig. 1a) in whichthe lengths of mother and daughter filaments distal to the branchwere very poorly correlated (Fig. 1b–d). The correlation coeffi-cients of the lengths of mother and daughter filaments produced byeither amoebic (Fig. 1b, d) or bovine (Fig. 1c, e) Arp2/3 complex

were 0.033 (n = 104) and 0.217 (n = 83), respectively. Daughter fil-aments formed by both amoebic and bovine Arp2/3 complex were,on average, half as long as their mother filaments (1.6 µm com-pared with 3.4 µm, n = 183), as measured distal to the branch. Over80% of filaments had one or more branches which varied in lengthrelative to the mother. Some daughter filaments gave rise to further‘generations’ of filaments (Fig. 1a).

To rule out the effects of phalloidin on branching and of anneal-ing of short filaments, we polymerized actin monomers withbovine Arp2/3 complex and the Scar WA domain (Scar-WA),adding rhodamine–phalloidin only after the polymerization reac-tion was ~80% complete. Dilution of the sample into fluorescencebuffer after 2 min minimized filament annealing. The productswere very similar to those assembled in the presence of phalloidin.Of filaments distal to 265 branches, daughter filaments were1.12 ± 0.61 µm in length and mother filaments were 2.55 ± 1.78µm long. The average ratio of daughter- to mother-filament lengthwas 0.37, and the correlation coefficient was only 0.068.

These results indicate that daughter branches may initiate froma point proximal to the barbed end of the mother filament. Thiscontrasts with previous results8, which indicated a strong correla-tion of branch lengths, as measured by electron microscopy of fila-ments polymerized with the Arp2/3 complex and N-WASP.Specimen preparation may account for this difference. Hydratedfilaments imaged by light microscopy are consistently longer thanthose prepared for electron microscopy (reviewed in ref. 14), prob-ably as a result of filament breakage, whereas lengths measured bylight microscopy are similar to those determined in polymerizationassays14.

To distinguish further between the two filament-nucleationhypotheses, we determined the effect of barbed-end capping on theability of filaments to serve as secondary activators of nucleation bythe Arp2/3 complex. Secondary activation by filaments reduces thelag at the outset of polymerization of actin monomers with theArp2/3 complex and the WA domain4–6,8. For example, 10 nMArp2/3 complex activated with Scar-WA accelerated polymeriza-tion of 2 µM actin monomers, but only after a lag of ~300 s were10% of the monomers polymerized (Fig. 2). As little as 10 nMpolymerized actin reduced this lag time to 210 s, whereas 40 nMand 200 nM polymerized actin further reduced the lag to 120 s and50 s, respectively. The barbed-end-nucleation hypothesis requiresthe presence of free barbed ends for this secondary activation of theArp2/3 complex.

We found that short filaments capped on their barbed ends withcapping protein stimulated nucleation by the Arp2/3 complex andthe WA domain just as well as much longer, uncapped filaments(Fig. 2). Although the concentration of free capping protein waslow, we included profilin to inhibit pointed-end elongation of fila-ments nucleated by free capping protein. The reduction in the lagtime depended only on the mass of actin filaments added, and the



NATURE CELL BIOLOGY VOL 3 MARCH 2001 http://cellbio.nature.com 307

observed rate of polymerization (see ref. 5 for the method of calcu-lation) indicated that short, capped filaments generated the sameconcentration of new barbed ends as uncapped long filaments. Thisresult is inconsistent with Arp2/3 complex binding to and nucleat-ing from the barbed ends of filaments.

Similar data have been presented previously8, as the result of anexperiment designed to examine the ability of the Arp2/3 complexactivated by the WA domain to block polymerization at the point-ed ends of filaments capped on their barbed ends with gelsolin.Actin polymerized rapidly in the presence of gelsolin-capped fila-ments and a range of concentrations of the Arp2/3 complex. Thesefindings have been interpreted as reflecting a failure of the Arp2/3complex to cap pointed ends, followed by barbed-end branchingfrom spontaneously formed nuclei. Given the low concentration ofactin monomer and the absence of a significant lag in the time

courses determined in this earlier study, we propse that a betterinterpretation is that gelsolin-capped filaments stimulated nucle-ation of branches that grew in the barbed direction. This vigorousbarbed-end growth would have prevented inhibition of pointed-end growth by the Arp2/3 complex from being observed. We con-firmed both of these interpretations by light microscopy (seebelow).

Our third test of the two hypotheses involved imaging branchesgrown directly from pre-existing, marked filaments with either freeor capped barbed ends (Figs 3, 4). In the first of these experiments,we polymerized ATP–actin with Alexa 488–phalloidin. We used theresulting green-stained mother filaments containing ADP–Pi,together with the Arp2/3 complex and Scar-WA, to stimulate nucle-ation of actin monomers in the presence of rhodamine–phalloidin.New red filaments (~35% of branches) grew as branches on thesides of pre-existing green filaments (Fig. 3). Five per cent of redbranches grew from the sides of red filaments that had grown fromeither the free barbed end or from the side of an original green fila-ment. Fifty per cent of red filaments (n = 100) were not associatedwith a green filament as a result of nucleation in solution by theArp2/3 complex or of spontaneous nucleation favoured by phal-loidin. Sixty per cent of red branches grew from the sides of thesenew red filaments. Per unit length of mother filament, the frequen-cy of branching from red and green filaments was similar — 1branch per 14 µm of green filament (n = 16) and 1 for every 13.5µm of red filament (n = 44). As green filaments were present fromthe outset of the reaction, this indicates a slight bias towards branch-ing from new filaments. Transiently, the new filaments have ATP asthe bound nucleotide and no bound phalloidin, but this is subject torapid change, as the half time for ATP hydrolysis is only 2 s (L.Blanchoin and T.D.P., manuscript in preparation) and the half time

Num

ber

Mean = 0.81 ± 0.91n = 104

Num

ber

Mean = 0.65 ± 0.75n = 79

Mother-filament length (µm)

Amoeba

Daughter/mother length ratio

Amoeba

Mother-filament length (µm)

Bovine

Daughter/mother length ratio

Bovine

Correlationcoefficient:0.033

Correlationcoefficient:0.217

Dau

ghte

r-fil

amen

t len

gth

(µm

)

Dau

ghte

r-fil

amen

t len

gth

(µm

)

0

1

2

3

4

5

6

7

0

5

10

15

20

0

1

2

3

4

5

6

7

0

5

10

15

20

25

30

0 2 4 6 8 10 0 2 4 6 8 10

0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3

b

a

c

d e

Figure 1 Length distribution of actin filament branches mediated by theArp2/3 complex. a, Fluorescence micrograph of a field of products of polymer-ization of 4 µM actin with 20 nM amoeba Arp2/3 complex, 200 nM Scar-WA and 4µM rhodamine–phalloidin (see Methods). Scale bar represents 10 µm. b, c, Scatterplots of the lengths of mother and daughter filaments (distance from branchingpoint to filament tip) of actin branches formed by 4 µM actin, 200 nM Scar-WA and4 µM rhodamine–phalloidin in the presence of 20 nM amoeba (b) or bovine (c)Arp2/3 complex. Diagonal lines represent a 1:1 ratio, corresponding to filamentbranches of equal length. d, e, Histograms of the length ratios of mother anddaughter filaments from experiments in a–c.

Flu

ores

cenc

e (a

rbitr

ary

units

)

Time (s)

7

6

5

4

3

2

0

1

0 100 200 300 400 500 600 700

Figure 2 Effect of capped and uncapped filaments on actin polymerizationin the presence of activated Arp2/3 complex. Conditions were as follows: 2µM actin (5% pyrene-labeled), 4 µM profilin, 300 nM Scar-WA, 10 mM imidazole pH7.0, 50 mM KCl, 1 mM MgCl2, 1 mM EGTA, 0.2 mM ATP and 1 mM dithiothreitol at22 °C. Uncapped actin filaments were prepared by polymerization of 2 µM actinmonomers for 1 h at 22 °C in the same buffer. Capped filaments were prepared bypolymerization under the same conditions in the presence of a 1:30 molar ratio ofamoeba capping protein. Inverted triangles, conditions described above with noaddition; open circles, 10 nM polymerized actin; open diamonds, 40 nM polymer-ized actin; filled circles, 10 nM capped filaments; filled diamonds, 40 nM capped fil-aments; no symbols, 10 nM amoeba Arp2/3 complex; open triangles, 10 nMamoeba Arp2/3 complex plus 10 nM polymerized actin; closed triangles, 10 nMamoeba Arp2/3 complex plus 10 nM capped filaments; open squares, 10 nMamoeba Arp2/3 complex plus 40 nM polymerized actin; closed squares, 10 nMamoeba Arp2/3 complex plus 40 nM capped filaments. Data are from one of twoindependent experiments.




for phalloidin binding is 6 s (ref. 15) under these conditions. Thelengths of polymer distal to branches varied widely between moth-er and daughter filaments. Some daughters and mothers (Fig. 3b, c,e) were of similar lengths, but most daughter filaments were signif-icantly shorter than their mothers (Fig. 3a, d, e).

In a second microscopy experiment, we capped green-labelled,ADP-containing mother filaments on their barbed ends with anexcess of capping protein and used them to stimulate branching byWA–Arp2/3 complex in the presence of rhodamine–phalloidin tolabel new filaments and to stabilize branches by inhibiting phos-phate dissociation (Fig. 4b–j). The products were similar to those inthe experiment with uncapped mother filaments, except for the lackfor red growth on the barbed ends of most green mother filaments,which verified that capping was effective. One or more red branch-es emanated from the sides of capped, phalloidin-stabilized greenfilaments and occasionally from red filaments (Fig. 4g). New red fil-aments were shorter and less variable in length than those withoutcapping protein (with capping protein, 1.5 ± 0.5 µm, n = 24; with-out capping protein, 2.6 ± 1.4 µm, n = 43), as a result of terminationof barbed-end growth by capping protein. As with uncapped fila-ments, many new filaments were not associated with a pre-existinggreen filament (Fig. 4j).

In a third microscopy experiment, we used WA–Arp2/3 complexand green, ADP–actin-containing mother filaments capped on theirbarbed ends with gelsolin to stimulate polymerization of actinmonomers in the presence of rhodamine–phalloidin to label newfilaments. Branches formed on the sides of both green (67%, n = 73)and red (33%) mother filaments (Fig. 4k, l). Gelsolin effectivelyblocked elongation of barbed ends, but in the absence of profilin allmother filaments grew slowly at their free pointed ends. This veri-fies the geometry of branching inferred from earlier studies6,10,11.

Because these polymerization reactions were carried out in thepresence of phalloidin and were allowed to equilibrate before visu-alization, we carried out a fourth microscopy experiment. We incu-bated WA-activated Arp2/3 complex with Alexa 488-labelled fila-ments containing ADP–Pi, that were capped with capping protein.After a short incubation, we added profilin and actin monomers tothe mixture. When the polymerization reached 80% completion webriefly labelled the newly grown filaments with rhodamine–phal-loidin and quickly diluted the sample to minimize annealing. As inthe continuous presence of phalloidin, short red branches formeddirectly on the sides of capped green mother filaments (Fig. 4a).Neither in this experiment, nor in other microscopy experiments inwhich the Arp2/3 complex stimulated branching from capped oruncapped phalloidin-stabilized filaments, did we observe a prefer-ence for branches at what had been the barbed end of the stimula-tory filament, as indicated either by the filament end (capped fila-ments) or by the green/red boundary (uncapped filaments).

In the absence of filament dynamics (as a result of the presenceof phalloidin) and of improbable annealing events in which theArp2/3 complex binds between two filament ends, the Arp2/3 com-plex can only associate with the middle of a mother filament by lat-eral binding. Although our results clearly show that the Arp2/3complex can form branches on the sides of filaments, they do notstrictly rule out the occurrence of branching at barbed ends.However, it is difficult to reconcile side-binding with the proposedmechanism of barbed-end branching8, in which the Arp2/3 com-plex is incorporated into both the mother and daughter filaments,as such an intercalation into a mother filament would require dis-placement of monomers from the filament.

Pantaloni et al.8 have presented several arguments to support thebarbed-end branching mechanism. These authors found thatmother and daughter filaments nucleated by the Arp2/3 complexare of similar lengths by electron microscopy. Using a light-micro-scopic assay that may more accurately represent the native lengthsof filaments, we found no such correlation and in fact founddaughter branches to be significantly shorter than their mothers.Pantaloni and colleagues have presented kinetic modelling of actin

polymerization nucleated by the Arp2/3 complex as support forbarbed-end nucleation. As we explained previously4,5, we agree thatnucleation mediated by the Arp2/3 complex is autocatalytic, withproducts (actin filaments) promoting catalysis of their own pro-duction. However, beyond the fact that some of the proposed reac-tions are purely hypothetical, this modelling in no way differenti-ates between barbed-end and side-binding mechanisms of nucle-ation and branching. Similarly, their fascinating observation thatListeria comet tails branch in ‘fish-bone’ structures in limiting con-centrations of gelsolin is consistent with branching nucleation bythe Arp2/3 complex, but does not differentiate between nucleationat barbed ends or on the sides of filaments. Pantaloni and col-leagues have interpreted other results8 as demonstrating that sec-ondary stimulation of nucleation by filaments depends on the con-centration of filament ends, rather than on the concentration ofpolymerized actin. However, these experiments were complicatedby the fact that the large numbers of free barbed ends introducedinto the experiment rapidly elongated, quickly producing a totalconcentration of polymerized actin that was uncontrolled and dif-ficult to predict accurately.

Neither the experiments presented here nor our interpretationof the experiments of Pantaloni and colleagues supports barbedend branching or the idea of ‘functional antagonism’ between cap-ping protein and the Arp2/3 complex. As described here and else-where6, capping protein and the Arp2/3 complex act synergisticallyto favour the densely branched structures seen in cells. No experi-ment has demonstrated competition between these two proteinsfor barbed ends of actin filaments. Furthermore, we have shownhere that the Arp2/3 complex mediates branching from the sides ofpre-existing actin filaments, rather than from their barbed ends.

Challenges remain in elucidating the molecular steps in theassembly and disassembly of actin-filament networks. Ultimately,real-time observation of nucleation, branch formation and elon-gation, and disassembly of filaments by severing and monomer

Figure 3 Fluorescence micrographs of actin filaments nucleated byuncapped green mother filaments and the Arp2/3 complex. Conditions wereas follows: 4 µM actin monomers and 4 µM rhodamine–phalloidin added to 10 mMimidazole pH 7.0, 50 mM KCl, 1 mM MgCl2, 1 mM EGTA, 0.2 mM ATP, 1 mM dithio-threitol, 300 nM Scar-WA and 1 µM Alexa 488–phalloidin-labelled actin with 25 nMbovine (a, b) or amoebic (c–e) Arp2/3 complex. Samples were incubated for 15min at 22 °C. Scale bar represents 5 µm.



NATURE CELL BIOLOGY VOL 3 MARCH 2001 http://cellbio.nature.com 309

dissociation will provide valuable insights. Furthermore, the rolesof many of the participants require better definition. For example,although the Arp2/3 complex is known to localize to branchpoints12, it is not yet known how long WASP/Scar proteins remainassociated with the Arp2/3 complex at branches after nucleation. Todate, most visualization of Arp2/3 complex branches has beenaided by the presence of phalloidin, either fluorescent phalloidinfor light microscopy or unlabelled phalloidin to allow extraction ofcytoplasm for electron microscopy12. New experimental approach-es are needed to provide a clearer understanding of the molecularmechanisms involved in this complex system.

MethodsProteins.Actin was isolated from rabbit skeletal-muscle acetone powder16 and further purified by gel filtration

on Sephacryl S-300 in Ca-G buffer (2 mM imidazole pH 7.0, 0.2 mM ATP, 0.1 mM CaCl2 and 0.5 mM

dithiothreitol (DTT)). Pyrenyl–actin was produced by labelling Cys374 with pyrene iodoacetamide17,18.

The Arp2/3 complex was purified from Acanthamoeba by poly-L-proline affinity chromatography19,

further purified by S-300 gel-filtration chromatography in complex storage buffer (10 mM imidazole

pH 7.0, 150 mM NaCl, 0.1 mM EGTA, 0.2 mM MgCl2, 0.2 mM ATP, 1 mM DTT and 0.02% NaN3)

and stored at –20 °C in 50% glycerol. Bovine Arp2/3 complex was purified from thymus5. Scar1-WA,

the 65 carboxy-terminal residues of human Scar1 protein4, was produced in bacteria from pMW172

(ref. 20) without induction. Bacterial cell lysates were clarified at 65,000g, gel-filtered on S-300, sub-

jected to chromatography on DEAE cellulose and stored at –20 °C. This resulted in highly purified

Scar1-WA consisting of a single species of relative molecular mass 7,650, as determined by MALDI DE.

Profilin-I (ref. 21) and capping protein22 were purified from Acanthamoeba. Filaments capped with

capping protein were produced by polymerizing 1 µM actin for 1 h at 22 °C in Mg-G buffer contain-

ing 1 × KMEI (50 mM KCl, 1 mM MgCl2, 1 mM EGTA and 10 imidazole pH 7.0) with 10 nM or

33 nM capping protein. Gelsolin–actin seeds were produced by incubating a 1:2 ratio of full-length

recombinant human-plasma gelsolin23 to actin in G buffer containing 200 µM CaCl2 at 22 °C for 2 h

and then incubating overnight at 4 °C (ref. 11). Gelsolin-capped filaments were produced by incubat-

ing 2 µM actin with a 1:1000 ratio of gelsolin to actin seeds for 30 min at room temperature in Mg-G

buffer plus 1 × KMEI. For use in fluorescence microscopy, a molar equivalent of Alexa 488–phalloidin

was added to capped filaments after polymerization and the mixture was incubated for 10 min at 22

°C.

Pyrenyl–actin polymerization kinetics.Immediately before polymerization, Ca–ATP–actin was converted to Mg–ATP–actin by incubation

with one-tenth of a volume of 10 × exchange buffer (10 mM EGTA and 1 mM MgCl2) for 2 min at

22 °C. Proteins were mixed at 22 °C in Mg-G buffer, and one-tenth of a volume of 10 × KMEI

(500 mM KCl, 10 mM MgCl2, 10 mM EGTA and 100 mM imidazole, pH 7.0) was added to initiate

polymerization. Pyrene fluorescence of samples (200 µl) was monitored using a Photon Technologies

International spectrofluorimeter with excitation at 365 nm and emission at 407 nm.

Branching assay.Products of polymerization were examined by fluorescence microscopy11. Proteins other than actin

monomers were mixed in Mg-G buffer containing 1 × KMEI. Actin monomers and one molar equiva-

lent of Alexa 488–phalloidin (Molecular Probes) or rhodamine–phalloidin (Fluka, Steinheim,

Germany) were added and gently mixed. After polymerization for 15 min (unless otherwise specified)

at 22 °C, samples were diluted 500-fold in fresh fluorescence buffer (50 mM KCl, 1 mM MgCl2, 100

mM DTT, 10 mM imidazole pH 7.0, 0.5% methylcellulose, 20 µg ml–1 catalase, 100 µg ml–1 glucose oxi-

dase and 3 mg ml–1 glucose). Samples (1.6 µl) were applied to coverslips (22 × 22 mm) previously

coated either with 0.1% nitrocellulose in amyl acetate or for 10 min with 20 µg ml–1 poly-L-lysine.

Samples were examined by mercury illumination at 100 W on an Olympus IX-70 inverted microscope

using a ×100, 1.35 NA objective lens and rhodamine (ex:HQ545/30X, em:HQ610/75M, dc:565DCLP)

and fluorescein (ex:HQ4880/40, em: HQ535/50, dc:Q505LP) filter sets (Chroma Technology Corp.,

Battleboro, Vermont). Images were recorded using a Hamamatsu C4742-95 digital camera controlled

by Metamorph software (Universal Imaging). Branches were verified as not being merely artifacts of

overlain filaments by visualizing their brownian vibration11. Images were processed with Metamorph

and Adobe Photoshop to generate two-colour images from monochromatic images. Random fields (n

≈ 10) were selected and the lengths of mother and daughter filaments for each branch in the field (n ≈100) were measured manually using Metamorph. The lengths of mother and daughter filaments were

defined as the distance from the branching point to the filament tip. Length measurements were veri-

fied by a blinded observer.

RECEIVED 1 SEPTEMBER 2000; REVISED 13 NOVEMBER 2000; ACCEPTED 24 NOVEMBER 2000;PUBLISHED 16 FEBRUARY 2001.

1. Theriot, J. A. & Mitchison, T. J. Nature 352, 126–131 (1991).

2. Loisel, T. P., Boujemaa, R., Pantaloni, D. & Carlier, M. F. Nature 401, 613–616 (1999).

Figure 4 Fluorescence micrographs of actin filaments nucleated by cappedgreen mother filaments and the Arp2/3 complex. a, Alexa 488-labelled fila-ments (10 µM), containing ADP–Pi, were added to 10 mM imidazole pH 7.0, 50 mMKCl, 1 mM MgCl2, 1 mM EGTA, 0.2 mM ATP and 1 mM dithiothreitol and incubatedwith 200 nM capping protein at room temperature for 1 min, after which 300 nMScar-WA and 40 nM bovine Arp2/3 complex were added. After 1 min the samplewas diluted to 2 µM actin in the same buffer containing 200 nM capping protein.Actin monomers (2 µM) pre-incubated with a twofold excess of profilin were addedfor 1 min and rhodamine–phalloidin was then added to 2 µM for 1 min before dilu-

tion of the sample into fluorescence buffer. b–j, Actin monomers (2 µM) and rho-damine–phalloidin (2 µM) were added to buffer as in a with 20 nM amoeba Arp2/3complex, 200 nM Scar-WA, 200 nM capping protein and 500 nM Alexa 488–phal-loidin-labelled actin filaments previously capped by polymerizing in the presence of1:100 capping protein. k, l, Actin monomers (2 µM) and rhodamine–phalloidin (2µM) were added to buffer as in a containing 10 nM amoeba Arp2/3 complex, 200nM Scar-WA and 500 nM Alexa 488–phalloidin-labelled actin filaments previouslycapped by polymerizing in the presence of 1:1000 gelsolin. Scale bar represents13 µm (a), 10 µm (b–d, g), 5 µm (e, f, h–j) and 2.5 µm (k, l).




3. Frischknecht, F. et al. Nature 401, 926–929 (1999).

4. Machesky, L. M. et al. Proc. Natl Acad. Sci. USA 96, 3739–3744 (1999).

5. Higgs, H. N., Blanchoin, L. & Pollard, T. D. Biochemistry 38, 15212–15222 (1999).

6. Blanchoin, L. et al. Nature 404, 1007–1011 (2000).

7. Marchand, J. B., Kaiser, D. A., Pollard, T. D. & Higgs, H. N. Nature Cell Biol. 3, 76–82 (2001).

8. Pantaloni, D., Boujemaa, R., Didry, D., Gounon, P. & Carlier, M-F. Nature Cell Biol. 2, 385–391 (2000).

9. Mullins, R. D., Stafford, W. F. & Pollard, T. D. J. Cell Biol. 136, 331–343 (1997).

10. Mullins, R. D., Heuser, J. A. & Pollard, T. D. Proc. Natl Acad. Sci. USA 95, 6181–6186 (1998).

11. Blanchoin, L., Pollard, T. D. & Mullins, R. D. Curr. Biol. 10, 1273–1282 (2000).

12. Svitkina, T. M. & Borisy, G. G. J. Cell Biol. 145, 1009–1026 (1999).

13. Pollard, T. D., Blanchoin, L. & Mullins, R. D. Annu. Rev. Biophys. 29, 545–576 (2000).

14. Xu, J., Casella, J. F. & Pollard, T. D. Cell Motil. Cytoskeleton 42, 73–81 (1999).

15. De La Cruz, E. M. & Pollard, T. D. Biochemistry 35, 14054–61 (1996).

16. Spudich, J. A. & Watt, S. J. Biol. Chem. 246, 4866–4871 (1971).

17. Pollard, T. D. J. Cell Biol. 99, 769–777 (1984).

18. Kouyama, T. & Mihashi, K. Eur. J. Biochem. 114, 33–38 (1981).

19. Machesky, L. M., Atkinson, S. J., Ampe, C., Vandekerckhove, J. & Pollard, T. D. J. Cell Biol. 127,

107–115 (1994).

20. Way, M., Gooch, J., Pope, B. & Weeds, A. G. J Cell Biol 109, 593–605 (1989).

21. Kaiser, D. A., Goldschmidt-Clermont, P. J., Levine, B. A. & Pollard, T. D. Cell Motil. Cytoskeleton 14,

251–262 (1989).

22. Cooper, J. A., Blum, J. D. & Pollard, T. D. J. Cell Biol. 99, 217–225 (1984).

23. Yu, F. X., Zhou, D. M. & Yin, H. L. J. Biol. Chem. 266, 19269–75 (1991).

ACKNOWLEDGMENTSThis work was supported by research grants from the National Institutes of Health. We thank D. Kaiserfor amoeba profilin and Scar-WA, H. Higgs for bovine Arp2/3 complex, L. Blanchoin for amoeba cap-ping protein, V. Sirotkin and K. M. Roh for measuring filaments, and members of our laboratory fordiscussions. We also thank D. Pantaloni and M-F. Carlier for discussion of their work on barbed-endnucleation and for providing us with a copy of their Nature Cell Biology paper before publication, andM-F. Carlier in particular for suggesting the mother/daughter nomenclature for branches.Correspondence and requests for materials should be addressed to T.D.P.

document

Documents