Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2007 Oct 2.
Published in final edited form as: Mol Cell. 2007 May 11;26(3):449–457. doi: 10.1016/j.molcel.2007.04.017

Insights into the influence of nucleotides on actin family proteins from seven new structures of Arp2/3 complex

Brad J Nolen 1, Thomas D Pollard 1
PMCID: PMC1997283  NIHMSID: NIHMS25453  PMID: 17499050

Summary

ATP is required for nucleation of actin filament branches by Arp2/3 complex, but the influence of ATP binding and hydrolysis are poorly understood. We determined crystal structures of bovine Arp2/3 complex cocrystalized with various bound adenine nucleotides and cations. Nucleotide binding favors closure of the nucleotide binding cleft of Arp3, but no large scale conformational changes in the complex. Thus, ATP binding does not directly activate Arp2/3 complex, but is part of a network of interactions that contribute to nucleation. We compared nucleotide-induced conformational changes of residues lining the cleft in Arp3 and actin structures to construct a movie depicting the proposed ATPase cycle for the actin family. Chemical crosslinking stabilized subdomain 1 of Arp2, revealing new electron density for 69 residues in this subdomain. Steric clashes with Arp3 appear to be responsible for intrinsic disorder of subdomains 1 and 2 of Arp2 in inactive Arp2/3 complex.

Introduction

Arp2/3 complex is an assembly of seven proteins that nucleates actin filament branches from the sides of pre-existing actin “mother” filaments (Pollard, 2007). The complex is intrinsically inactive, and activation requires interactions with ATP, an actin monomer, a mother filament and one of a diverse array of nucleation promoting factors (NPFs), which include WASp/Scar family proteins, yeast type I myosins, Listeria monocytogenes Act A protein, cortactin, Abp1p, and Pan1p (D’Agostino and Goode, 2005; Higgs and Pollard, 2001). The complex contains two actin-related proteins, Arp2 and Arp3, for which it is named. The crystal structure of inactive bovine Arp2/3 complex suggested that a rearrangement of Arp2 and Arp3 is required for them to initiate a daughter filament (Robinson et al., 2001).

Like actin, Arp2 and Arp3 bind adenosine nucleotides, with the adenine ring and the ribose filling a notch at the interface of subdomains 3 and 4 and the phosphates clamped by two highly conserved loops from opposite sides of the large central cleft (Nolen et al., 2004). After incorporation into a filament, actin rapidly hydrolyzes its bound ATP (Blanchoin and Pollard, 2002) and then slowly releases the γ-phosphate (Melki et al., 1996). The biochemical differences between ADP- and ATP-actin subunits provide a timing mechanism which allows the cell to distinguish between old and new actin filaments.

During branch formation Arp2/3 complex hydrolyzes ATP bound to Arp2 but not to Arp3. Some experiments suggest that hydrolysis is necessary for nucleation (Dayel et al., 2001; Dayel and Mullins, 2004; Le Clainche et al., 2001), while others point to a role in regulating branch disassembly (Le Clainche et al., 2003; Martin et al., 2006). Mutations that inhibit ATP binding to Arp3 reduce nucleation by Saccharomyces cerevisiae Arp2/3 complex (Martin et al., 2005) and decrease the effect of Mg-ATP on fluorescence resonance energy transfer (FRET) between labeled subunits of human Arp2/3 complex, suggesting that ATP binding causes a conformational change in Arp3 important for activation (Goley et al., 2004). However, soaking crystals of nucleotide-free bovine Arp2/3 complex with ATP caused a relatively minor structural change in which the cleft of Arp3 closed by 1.5 Å (Table 1 and (Nolen et al., 2004)).

Table 1.

Data collection and refinement statistics

Data Collection 2P9K 2P9I 2P9U 2P9S 2P9L 2P9P 2P9N
Beamline 19ID 19ID 19ID X29A X29A X29A X29A
Ligand ATP/Ca2+ ADP/Ca2+ ANP/Ca2+ ATP/Mg2+ none ADP/Ca2+ ADP/Ca2+
Growth Condition A A A A A A B
Crosslink yes yes no no no no no
Resolution (Å) 30.0–2.59 30.0–2.43 30.0–2.75 30.0–2.68 30.0–2.65 30.0–2.9 30.0–2.85
Space group P212121 P212121 P212121 P212121 P212121 P212121 P212121
Cell constants a = 110.62 a = 109.88 a = 111.12 a = 111.02 a = 111.06 a = 110.97 a = 111.13
b = 128.05 b = 127.80 b = 129.35 b = 128.69 b = 128.06 b = 128.93 b = 129.27
c = 198.29 c = 197.74 c = 203.80 c = 201.40 c = 204.17 c = 202.29 c = 203.86
Mosaicity (°) 0.2 0.16 0.15 0.41 0.35 0.74 0.47
Measured reflections 964956 990924 627423 1591990 1131076 1110806 1121069
Ubique reflections 88027 105244 76183 80893 81688 57341 68523
Mean I/σ 19.3/4.0 21.0/2.4 18.4/2.7 22.3/2.6 12.7/2.0 13.4/2.4 11.6/2.4
Rsym (%)a 10.1/41.3 6.3/54.2 8.0/41.1 10.4/54.4 11.1/65.1 12.1/62.1 11.9/54.9
Completeness (%) 99.5/97.5 98.4/96.4 98.9/92.4 99.4/96.1 95.5/89.9 88.8/88.6 98.2/99.0
Refinement
Modeled atoms 14366 14312 13719 13612 13504 13196 13545
Water molecules 326 335 251 233 123 0 0
Average B-factor (Å 2)
 main chain 35.2 47.6 45.3 52.7 54.2 59.8 51.5
 side chain 35.6 48.7 45.6 53.1 54.5 61.1 52.1
 nucelotide (Arp3) 27.5 37.0 48.8 45.1 - 59.4 72.7
 nucelotide (Arp2) 32.6 40.9 61.8 65.4 - 67.6 63.0
 water 37.4 40.5 39.7 44.7 40.0 - -
RMS from ideal
 bond lengths 0.007 0.006 0.007 0.007 0.007 0.008 0.008
 bond angles 1.34 1.23 1.37 1.34 1.31 1.42 1.38
Ramachandran plot
 most favored 1358 1376 1306 1300 1295 1255 1265
 additionally allowed 175 171 180 177 186 185 221
 generously allowed 16 7 10 16 15 15 11
 disallowed 0 0 0 0 0 0 0
Rfree 26.0 26.1 26.3 26.2 26.6 28.8 27.2
Rworkb 21.9 21.7 21.7 22.2 22.3 23.4 23.2
Open in a new tab
a

Rsym = Σ|Ih − <Ih>|/Σ Ih, where <Ih> is the average intensity over symmetrically equivalent reflections.

b

Rwork =Σ|Fo − Fc||/Σ Fo, where Fo is an observed amplitude and Fc a calculated amplitude; Rfree is the same statistic calculated over a subset (5%) of the data that has not been used for refinement.

In this study we determined structures of bovine Arp2/3 complex co-crystallized with Ca2+ or Mg2+, ATP, ADP or the ATP analog AMP-PNP. We also tested the effect of crosslinking crystals with glutaraldehyde. These structures have clarified how nucleotides affect the conformation of Arp2/3 complex and allowed us to assemble a movie of the ATPase cycle of the actin family of proteins.

Results and Discussion

Co-crystals of bovine Arp2/3 complex with bound ATP- or ADP-Ca2+ grew consistently in a new condition (precipitant A: 7.5 % PEG 3350, 50 mM HEPES pH 7.0, 100 mM KSCN, and 10% sucrose) with the same space group and nearly identical unit cell dimensions as crystals of Arp2/3 complex without bound nucleotides (Table 1). We used the original apo-Arp2/3 complex structure as a starting point for rigid body refinement to solve each of the co-crystal structures. The Arp3 nucleotide binding cleft adopted a range of conformations that were accommodated by compensatory variation of the longest unit cell axis (c) (Table 1) with only minor differences in crystal contacts.

Nucleotide binding and cleft closure in Arp3: implications for Arp2/3 complex function

Nucleotide binding to Arp2/3 complex before or after crystallization influences the closure of the nucleotide cleft of Arp3 and conformations of residues lining the cleft but does not cause major rearrangements of the complex. The cleft of Arp3 closes by a rigid body movement of subdomains 3 and 4 of Arp3 and ARPC3 (p21) relative to the rest of the complex (Figure 1a). We classified each Arp3 structure as open, intermediate or closed based on the distance between Gly15 in the P1 loop and Asp172 in the P2 loop on opposite sides of the nucleotide binding cleft (Table 2, Figure 1b).

Figure 1.

Figure 1

Open in a new tab

Nucleotide binding causes changes in the cleft of Arp3. (A) Superposition of Cα traces of apo-Arp2/3 complex (1K8K) and the ADP-cocrystallized, crosslinked Arp2/3 complex (2P9I). Structures were superposed by aligning subdomains 1 and 2 of Arp3. ARPC1, ARPC2, ARPC4, ARPC5 and Arp2 overlay well (both complexes gray). A rigid body motion of subdomains 3 and 4 of Arp3 and ARPC3 (cyan in ADP-complex, red in apo-complex) closes the cleft of structure 2P9I. Additional residues built for subdomain 1 in Arp2 of the crosslinked ADP cocrystals (2P9I) are also in cyan. (B) Stereo figure showing overlay of the nucleotide binding cleft of Arp3 in the crosslinked ADP cocrystal (2P9I, cyan) and the crosslinked ATP cocrystal. (2P9K, yellow). ADP is magenta and ATP is purple. Labeled residues mark key features: Thr14 for the P1 loop; Val174 for the P2 loop; and His80 for the sensor loop. Two distances define the width of the cleft: B1 (Thr14 Cα to Gly173 Cα; atoms shown as orange spheres) and B2 (Gly15 Cα to Asp172 Cα; atoms shown as brown spheres). Distance B2 was used to categorize clefts in structures of Arp2, Arp3 and actin as open, closed or intermediate (Table 2). (C) Stereo figure showing an overlay of the nucleotide binding cleft of Arp3 in the crosslinked ADP co-crystal (2P9I, cyan protein, magenta ADP) and the previously published ADP-soaked structure (1U2V, orange protein, yellow ADP). (D) Summary of conformational changes in the ATP binding cleft during the ATPase cycle of actin family proteins. The dotted lines show hydrogen bonds between the γ-phosphate and the loops when the cleft is fully closed. Residue numbering is for bovine Arp3. Conformational changes observed in Arp3 and actin are numbered in red. Wavy red lines connect structural features that show correlated changes in one or more actin or Arp3 structures. (1) Two positions of a valine (Val174 in Arp3) in the P2 loop observed in Arp3 and actin structures suggest this valine may be involved in sensing the nucleotide binding state. (2) A rotomer flip in Ser14 (Thr14 in Arp3) and a slight inward collapse (cyan arrowhead) of the P1 loop occurs in the ADP-actin structures (1J6Z and 2HF4) and in a structure of Arp3 with bound ADP (2P9I). This movement is accompanied by a flip of the backbone carbonyl of a residue in the sensor loop in both Arp3 and actin. (3) Rigid body motions of subdomains 1 and 2 relative to subdomains 3 and 4 result in opening or closing of the nucleotide cleft. These structural changes have been observed in actin by comparing the single open structure (1HLU) to each of the other ADP and ATP-containing structures, all of which are closed. Open and closed conformations have also been observed in Arp3, where the nucleotide state is correlated to the degree of opening of the cleft.

Table 2.

Measurement of cleft distances in Arps and actin. Structures with an asterisk are presented here, others are previously published. Columns indicate for each Arp2/3 complex structure the identity of bound ligand (ligand), precipitation agent (ppt; A: see text, B: 6% PEG 8000, 100 mM HEPES pH 7.5, 100 mM KSCN), ligand incorporation method (prep), and whether (y) or not (n) the crystal was crosslinked with glutaraldehyde (crsk). B1 is the distance between the Cα of Thr14 in the P1 loop and Cα of Gly173 in the P2 loop on opposite sides of the nucleotide binding cleft in Arp3. B2 is the distance between the Cα of Gly15 in the P1 loop and Cα of Asp172 in the P2 loop. Distance B2 measures the width of the nucleotide binding cleft, as indicated in the column labeled “cleft” as follows: B2 ≤ 6.0 Å closed, 6.0 Å < B2 > 7.0 Å intermediate, B2 ≥ 7.0 Å open). The conformations of the carbonyl of Glu79 (sensor) and Val174 (V159/174) are indicated. The table includes actin crystal structures mentioned in the text for reference.

structure ligand ppt prep crsk B1 B2 cleft sensor V159/174
Arp3
1K8K APO B n 8.5 7.7 open up up
2P9L* APO A n 7.9 7.1 open up up
1U2V ADP/Ca B soak n 8.8 8.0 open up up
2P9N* ADP/Ca B co n 8.2 7.6 open up up
2P9U* ANP/Ca A co n 7.6 7.0 open up up
2P9P* ADP/Ca A co n 6.5 6.2 intermediate up up
1TYQ ATP/Ca B soak n 6.8 6.2 intermediate up up
2P9I* ADP/Ca A co y 5.9 5.5 closed down down
2P9K* ATP/Ca A co y 6.0 5.7 closed up up
2P9S* ATP/Mg A co n 5.7 5.6 closed up up
Arp2
2P9I* ADP/Ca A co y 5.5 5.2 closed down
2P9K* ATP/Ca A co y 6.1 6.7 intermediate down
Actin
1NWK ATP/Ca 5.3 5.5 closed up down
1ATN ATP/Ca 5.3 5.5 closed up down
1J6Z ADP/Ca 5.2 5.6 closed down down
1D4X ATP/Mg 5.4 5.1 closed up up
1MA9 ATP/Mg 5.7 5.3 closed down down
1HLU ATP/Ca 8.3 8.2 open up down
Open in a new tab

ATP binding causes closure of Arp3 in all conditions we tested. Cleft closure may contribute to branch formation by favoring interactions of Arp2/3 complex with a nucleation promoting factor (Dayel et al., 2001), a mother filament or the first subunit in the daughter filament. Together these cooperative interactions rearrange the Arps like two consecutive subunits in an actin filament, as observed in reconstructions of branch junctions from electron tomograms (Rouiller et al., unpublished data). However, ATP binding alone does not activate Arp2/3 complex and the conformational change associated with ATP binding is much smaller in crystals than anticipated by FRET measurements (Goley et al., 2004).

Closure of the nucleotide cleft of Arp3 with bound ADP varies depending on the conditions (Table 2, Figure 1b,c). The cleft was completely closed when Arp2/3 complex was cocrystallized with ADP in precipitant A and then crosslinked (2P9I) but was open in crystals soaked in ADP or cocrystallized with ADP in precipitant B (1U2V, 2P9N). In both open structures, ADP adopts an unusual bent conformation with the α-phosphate flipped up and both phosphates stretched across the open cleft to hydrogen bond to the P1 loop (Figure 1c). The Arp3 cleft adopts an intermediate conformation in the ADP/precipitant A cocrystals (2P9P), with the phosphates of ADP in the usual extended conformation. The observation of ADP-Arp3 with both open and closed clefts indicates that a low energy barrier exists between these states and suggests that in solution the open conformation may be more populated when ADP is bound to Arp3 than when ATP is bound.

The nucleotide binding cleft of Arp3 is much more open in crystals grown with the non-hydrolyzable ATP-analogue AMP-PNP than with ATP (Table 2). Therefore, the failure of AMP-PNP to support nucleation (Dayel et al., 2001) may be due to the conformations of Arp3 and/or Arp2 rather than a requirement for Arp2 to hydrolyze ATP. This is consistent with experiments showing that hydrolysis of ATP bound to Arp2 is a consequence of branch formation (Dayel and Mullins, 2004) rather than a requirement (Martin et al., 2006). Dissociation of γ-phosphate from Arp2 may influence the stability of branches (Martin et al., 2006) by opening the cleft of Arp2, as observed for Arp3-ADP.

Influence of the γ-phosphate on the conformation of the sensor and DNase loops

The loop between β6 and αB of actin (residues 69 to 78 in actin and 78 to 85 in Arp3) is called the sensor loop because it is postulated to detect and relay conformational differences between ADP- and ATP-actin from the nucleotide cleft to other parts of the molecule (Graceffa and Dominguez, 2003; Otterbein et al., 2001; Rould et al., 2006). In the closed ADP-Arp3 structure (2P9I), the absence of the γ-phosphate allows the P1 loop to collapse inward slightly and the sensor loop to adopt the “flipped” conformation observed in two ADP-actin structures (Figure 1b) (Otterbein et al., 2001; Rould et al., 2006). This change is not observed in the open or intermediate ADP structures or in the ATP-bound structures of Arp3 (Figure 1b,c, Table 2).

The flip of the sensor loop was suggested to cause the DNase binding loop of actin to fold into an α-helix (Otterbein et al., 2001). In Arp3, the change in the sensor loop does not influence the conformation of the “DNase loop”, which is disordered in all ten Arp2/3 complex structures solved to date. The distinct packing arrangements of Arp2/3 complex versus rhodamine-actin in crystals may contribute to this difference. In the ADP-rhodamine actin structure the DNase helix is packed against a symmetry related molecule. Flipping of the sensor loop upon loss of the γ-phosphate changes the relative position of the symmetry-related molecule, providing a favorable interaction surface for the DNase helix (Graceffa and Dominguez, 2003; Otterbein et al., 2001; Rould et al., 2006).

The conformation of Val174 of Arp3 is not tightly coupled with the presence of the γ-phosphate

A valine in the P2 loop (Val159 in actin, Val174 in Arp3) on the subdomain 3 side of the nucleotide cleft is postulated to couple γ-phosphate release to opening the cleft of actin (Belmont et al., 1999). We observed two distinct conformations for this residue in Arp3. In the “down” conformation, the backbone amide is oriented to form a hydrogen bond with the γ-phosphate (Figure 1b,c) and in the “up” conformation it is not. The position is not correlated with the identity of the bound nucleotide in our crystal structures; Val174 is in the up conformation in all but the closed ADP-Arp3 structure (2P9I); i.e., it is up in three of four ADP structures, all three ATP structures, the AMP-PMP structure and the two nucleotide-free structures. In actin, the up position is found only in the C. elegans ATP-actin-gelsolin crystal (Vorobiev et al., 2003) (1D4X).

Reconstruction of the actin family ATPase cycle from crystal structures

Our ten structures of Arp2/3 complex together with the available actin structures reveal a spectrum of conformations that could influence the biochemical properties of these proteins (Figure 1d). We assembled the available crystallographic snapshots of the nucleotide-binding cleft of actin, Arp2 and Arp3 to postulate a sequence of conformational changes associated with the ATPase cycle of actin family proteins. The accompanying movies morph this sequence of structures from one to the next by linear interpolation and adiabatic mapping carried out using a modified CNS (Brunger et al., 1998) script downloaded from the Gerstein lab morphing server (Krebs and Gerstein, 2000). The motions of actin in the movie represent one possible path between the individual crystallographic snapshots and are not empirically based. However, they provide a useful tool for visualizing the documented structural changes as they might occur during the ATPase cycle.

First, ATP binds to the open cleft of an actin monomer to occupy the groove between subdomains 3 and 4 (Nolen et al., 2004). The cleft must be open to avoid steric clashes with the protein as the nucleotide diffuses into the groove. The fact that nucleotide binds to subdomains 3 and 4 of Arp2 even though subdomains 1 and 2 are disordered shows that interactions of the nucleotide with subdomains 3 and 4 provides much of the binding energy.

Second, the cleft closes around ATP. The phosphates are clamped into place by hydrogen bonds to the P1 and P2 loops (for details see the movie legend in supplemental materials). The networks of interactions of ATP with actin and Arp3 are identical, although in some cases the hydrogen bond between the γ-phosphate and Val159 (Val173 in Arp3) is missing, as described below.

Third, the ATP is hydrolyzed. Crystallographic (Vorobiev et al., 2003) and mutational analysis (Martin et al., 2006) suggests that a conserved histidine (His161 in actin) may play a role in activating and/or positioning a nucleophilic water located near the γ-phosphate. In the majority of ATP-actin crystal structures, His161, the proposed nucleophilic water, and the γ-phosphate are not aligned for catalysis. This may account for the very low rate of ATP hydrolysis by monomers (0.000007 s−1) (Rould et al., 2006), but the fact that incorporation of actin into a filament increases the rate of hydrolysis >40,000-fold to 0.3 s−1 (Blanchoin and Pollard, 2002) raises the possibility of a substantial rearrangement of the active site and a catalytic mechanism quite different from that suggested by available structures. A high resolution structure of filamentous actin will be required to understand the details of the mechanism.

Fourth, the free γ-phosphate moves away from the β-phosphate, but maintains the hydrogen bond to the backbone amide of Val159 in the P2 loop as this residue flips up. No crystal structure of actin with bound ADP and inorganic phosphate is available, so we modeled this step using the ADP-Pi intermediate of another member of the family of hexokinase fold proteins, Hsc70 (Wilbanks and McKay, 1995). The flip of Val159 does not cause changes outside the active site in the movie, but must in reality, since the rate constants differ substantially for dissociation and association of ADP-Pi-actin and ATP-actin at both ends of filaments (Fujiwara et al., 2007).

Fifth, γ-phosphate dissociates from the protein. Steered molecular dynamics simulations of actin with ADP and inorganic phosphate suggested a “back door” release mechanism, in which phosphate escapes through a small solvent channel underneath the sensor loop and away from the solvent-exposed face of actin in a filament (Wriggers and Schulten, 1999). Since this pathway is available without opening the cleft, it is consistent with the observation that slow ATP hydrolysis and phosphate release occur in some crystals of actin without perturbing the crystal (Hertzog et al., 2004; Kabsch et al., 1990). Wriggers’ simulation suggests that separation from the calcium is the rate-limiting step in phosphate release. The high energy barrier for this separation presumably accounts for the low rate constants for phosphate dissociation from (0.003 s−1) and association with (2 M−1s−1) ADP-actin filaments.

Sixth, in the absence of the γ-phosphate the P1 loop collapses slightly and the backbone of the sensor loop flips, as seen in ADP-Arp3 and ADP-actin structures. It is unclear how this change affects the biochemical properties of actin. While a structural connection between the sensor loop and the DNase binding loop has not been established, limited proteolysis experiments show that the nucleotide state influences the conformation of the DNase binding loop (Strzelecka-Golaszewska et al., 1993). The DNase binding loop can adopt multiple conformations in both ADP- and ATP-bound states, so the dominant factor influencing its fold may be interactions with other proteins (Graceffa and Dominguez, 2003; Kabsch et al., 1990; Otterbein et al., 2001; Rould et al., 2006). These structural transitions, regardless of how they are connected to the active site, are probably important in the context of the filament.

It is not known if the closed ADP conformation is the most favored conformation in filaments and/or monomers in solution. The ADP-actin crystal structures are closed, suggesting that the closed state is thermodynamically favored (Otterbein et al., 2001; Rould et al., 2006). The cleft of ADP-actin the cleft remained closed in molecular dynamics (MD) simulations (Chu and Voth, 2005; Dalhaimer P, Nolen BJ and Pollard TD, unpublished data). Furthermore, subunits in muscle actin filaments do not exchange nucleotide (Pollard et al., 1992), which probably requires opening of the cleft. On the other hand, the cleft is open in reconstructions of electron micrographs of filaments of ADP-bound yeast actin (Belmont et al., 1999). Differences in the cleft or other parts of the molecule must explain how the nucleotide bound to monomers and filaments determines their affinities for a number of proteins (see below) (Dominguez, 2004).

Seventh, the cleft opens and ADP adopts the bent conformation seen in ADP-loaded Arp3 in the 1U2V structure. Dissociation of the γ-phosphate probably favors this step, because bridging interactions between the phosphate and the P1 and P2 loops are lost. Steered MD simulations show that less force is required to open the cleft in ADP-actin than ATP-actin (Dalhaimer P, Nolen BJ and Pollard TD, unpublished data). As the cleft opens, the hydrophobic groove on the barbed end of actin between subdomains 1 and 3 closes. This region is a hotspot for actin binding proteins (Dominguez, 2004). Differences in its width caused by opening or closing of the nucleotide cleft could contribute to the higher affinity of thymosinβ4, profilin and WASp-family members for ATP-actin monomers (Hertzog et al., 2004) on one hand, and the higher affinity of ADF/cofilin and gelsolin for ADP-actin monomers on the other hand (Blanchoin and Pollard, 1998; Blanchoin and Pollard, 1999; Laham et al., 1993).

Eighth, ADP dissociates with its bound divalent. Separation of the P1 and P2 loops, opening of the cleft and the bent conformation of ADP facilitate release by providing the adenine ring, ribose, and phosphates with an unobstructed exit path.

We hope that our hypothesis for the ATPase cycle of actin monomers embodied in the movie will stimulate further thought and experimentation. Viewers should note that the movie has several limitations.

First, the physiologically relevant pathway involves actin monomers that bind ATP and dissociate ADP and polymerized actin subunits that hydrolyze ATP and dissociate the γ-phosphate, but not ADP. At the present time little can be said about the details of the conformations of polymerized actin where interactions with neighboring subunits are certain to be influential.

Second, each state along the pathway is shown as a static structure, whereas in solution each chemical state is actually an ensemble of rapidly interconverting structures. The conformations captured in crystals are a sample of the population of a particular chemical state, but these crystallographic samples may not be the dominant conformation in solution. This is illustrated by inconsistencies between chemical states and conformations in crystal structures of both actin and Arps. For example, the backbone amide of Val159 can hydrogen bond to the γ-phosphate when the side chain is in the down position, so this conformation might be favored with bound ATP. The up conformation of the side chain, if observed, might be more likely with bound ADP. However, Val174 was flipped up in all ATP-bound Arp3 structures and one ATP-actin structure (1D4X). Only the crosslinked ADP-Arp2/3 complex structure had the Val174 of Arp3 in the down conformation. Similar inconsistencies can be seen in the conformations of the sensor loop in Arp3 and actin structures. It is likely that the alternative conformations have similar energies and are readily interconverted. Additional crystal structures and molecular dynamics simulations should clarify how the various conformations contribute to the ATPase cycle.

Third, the movie shows equal time intervals between each state, whereas some steps are fast (ATP binding), and some are very slow, such as ATP hydrolysis by actin monomers and phosphate dissociation from actin filaments. While the crystal structures of actin and Arp2/3 complex have allowed us to propose a sequence for the ATPase cycle, other biophysical methods are needed to relate the kinetic and thermodynamic parameters for ATP binding (De La Cruz and Pollard, 1996), ATP hydrolysis (Blanchoin and Pollard, 2002), γ-phosphate release (Melki et al., 1996) and ADP dissociation (Selden et al., 1999) to the occupancy of each chemical intermediate along this pathway in a population of actin or Arp molecules in solution.

The architecture of the inactive complex may prevent ATP hydrolysis by Arp2

Treatment of Arp2/3 complex-nucleotide co-crystals (2P9I and 2P9K) with the non-specific crosslinker glutaraldehyde dramatically increased the positive electron density in initial Fo-Fc maps in the regions expected to contain subdomains 1 and 2 of Arp2 (Figure 2a,b, S1). The new density allowed us to model an additional 85 residues, including 69 residues in subdomain 1. Crosslinking captures the P1 and P2 loops in a closed conformation (Table 2), which must be occupied at least transiently in crystals. However, much of subdomains 1 and 2 of Arp2 remain disordered, and the newly modeled regions have relatively high B-factors. This suggests a high energy barrier for stable closure of subdomains 1 and 2 around the nucleotide. The structural basis for this energy barrier is apparent in a model of Arp2/3 complex built by superposing subdomains 1 and 2 of rabbit skeletal muscle actin (1ATN) onto the ordered regions of subdomain 1 of Arp2 in the crosslinked structures (Figure 2c). In this model, side chains from subdomain 2 of Arp2 clash with side chains on both the αI/αJ loop and the αK/β15 insert of Arp3. This αK/β15 insert is in Arp3 has been conserved for a billion years (Beltzner and Pollard, 2004) and may provide a steric bumper which prevents the cleft of Arp2 from closing, hence inhibiting spontaneous activation and hydrolysis of bound ATP until the complex is activated.

Figure 2.

Figure 2

Open in a new tab

Crosslinking bovine Arp2/3 complex crystals with glutaraldehyde increases order in subdomain 1 of Arp2. (A) Final 3σ Fo-Fc electron density map and Cα trace of modeled regions of Arp2 in uncrosslinked bovine Arp2/3 complex-ATP-Mg2+ cocrystals (2P9S). The map shows little density for subdomains 1 and 2. (B) Final 3σ Fo-Fc omit map and Cσ trace of modeled regions of Arp2 in bovine Arp2/3 complex-ATP-Ca2+ cocrystals treated with glutaraldehyde (2P9K). The newly modeled regions were not included in the map calculation. (C) Steric hindrance with Arp3 may prevent Arp2 from closing in the inactive complex. Cα traces show Arp3 (cyan) and Arp2 (blue) from the structure of crosslinked ADP-Arp2/3 complex (2P9I) with the addition of a model of four disordered residues at the end of the αK/β15 loop of Arp3 (orange). Actin (red) is overlaid onto Arp2 to show potential clashes (red arrows) of subdomain 2 with the αI/αJ loop and the αK/β15 loop of Arp3. The yellow Cα trace (highlighted with arrow) shows how αK is connected to β15 in actin. The αK/β15 insert in Arp3 makes the αK helix three turns longer in Arp3 than actin.

Experimental Methods

Arp2/3 complex was purified from bovine calf thymus (Robinson et al., 2001). Crystals were grown using the hanging drop vapor diffusion method at 4° Celsius using either precipitant A (7–9% PEG 3350, 50 mM HEPES pH 7.0, 100 mM KSCN, and 10% sucrose) or precipitant B (7–9% PEG 8000, 50 mM HEPES pH 7.5 and 100 mM KSCN) and 25–40 μM Arp2/3 complex. For co-crystallization, Arp2/3 complex stocks were mixed with 10 mM ATP, ADP or AMP-PNP and 10 mM CaCl2 or MgCl2 to bring the nucleotide and divalent concentration to 0.5 mM prior to mixing with precipitant. Crystals grown in precipitant A were soaked in 18% PEG 3350, 20 mM HEPES pH 7.0, and 24% sucrose plus nucleotide and divalent for one hour before flash freezing. Crystals grown in precipitant B were soaked in 18 % PEG 8000, 20 mM HEPES pH 7.5, and 20% glycerol plus nucleotide and divalent for one hour before flash freezing. Crystals 2P9U, 2P9K and 2P9I were grown in the presence of 25 μM human WASp construct B-GBD-CA (residues A223-Q310-GGSGGS-Q461-D502, a gift from the laboratory of Michael Rosen). Peptides corresponding to the C and A domains of N-WASp were present at 500 μM each in the cryo-soak for crystal 2P9U. No density was observed for activators in these structures, and a 2.8 Å resolution data set confirmed that activator peptides had no effect on conformation of Arp2/3 complex in the 2P9U crystal (data not shown). Crosslinked crystals 2P9K and 2P9I were prepared by suspending crystallization drops over a microbridge containing 50% glutaraldehyde for one hour, then transferring to cryo-buffer and flash freezing.

Data were collected at 19ID (ALS) or X29A (NSLS) (Table 1) and processed with HKL2000 (Otwinowski and Minor, 1997). The original apo-enzyme Arp2/3 complex structure (1K8K) was used as a starting model for rigid body refinement and subsequent rounds of minimization in CNS (Brunger et al., 1998). Models were rebuilt using COOT and O (Emsley and Cowtan, 2004; Jones et al., 1991). Water molecules were added at the end of the refinement using a σ cutoff of 2.5.

Supplementary Material

Movie Legends
Movie 1
Download video file (5.2MB, mpg)
Movie 2
Download video file (7MB, mpg)

Acknowledgments

We thank Hongli Chen for assistance with preparation of bovine Arp2/3 complex. We thank Howard Robinson and Wuxian Shi at NSLS X29A beam line and Younchang Kim, Andrzej Joachimiak, and Rongguang Zhang at APS beamline 19ID for assistance with data collection. Use of the National Synchrotron Light Source, Brookhaven National Laboratory, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-98CH10886. Use of the Advanced Photon Source at Argonne National Laboratory was supported by the U.S. Department of Energy, Office of Biological and Environmental Research under contract DE-AC02-06CH11357. BJN was supported by National Institutes of Health under Ruth L. Kirschstein National Research Service Award (F32GM074374). This work was supported by NIH research grant PO1GM066311 to TDP.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Belmont LD, Orlova A, Drubin DG, Egelman EH. A change in actin conformation associated with filament instability after Pi release. Proc Natl Acad Sci U S A. 1999;96:29–34. doi: 10.1073/pnas.96.1.29. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Beltzner CC, Pollard TD. Identification of functionally important residues of Arp2/3 complex by analysis of homology models from diverse species. J Mol Biol. 2004;336:551–565. doi: 10.1016/j.jmb.2003.12.017. [DOI] [PubMed] [Google Scholar]
  3. Blanchoin L, Pollard TD. Interaction of actin monomers with Acanthamoeba actophorin (ADF/cofilin) and profilin. J Biol Chem. 1998;273:25106–25111. doi: 10.1074/jbc.273.39.25106. [DOI] [PubMed] [Google Scholar]
  4. Blanchoin L, Pollard TD. Mechanism of interaction of Acanthamoeba actophorin (ADF/Cofilin) with actin filaments. J Biol Chem. 1999;274:15538–15546. doi: 10.1074/jbc.274.22.15538. [DOI] [PubMed] [Google Scholar]
  5. Blanchoin L, Pollard TD. Hydrolysis of ATP by polymerized actin depends on the bound divalent cation but not profilin. Biochemistry. 2002;41:597–602. doi: 10.1021/bi011214b. [DOI] [PubMed] [Google Scholar]
  6. Brunger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve RW, Jiang JS, Kuszewski J, Nilges M, Pannu NS, et al. Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallogr D Biol Crystallogr. 1998;54(Pt 5):905–921. doi: 10.1107/s0907444998003254. [DOI] [PubMed] [Google Scholar]
  7. Chu JW, Voth GA. Allostery of actin filaments: molecular dynamics simulations and coarse-grained analysis. Proc Natl Acad Sci U S A. 2005;102:13111–13116. doi: 10.1073/pnas.0503732102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. D’Agostino JL, Goode BL. Dissection of Arp2/3 complex actin nucleation mechanism and distinct roles for its nucleation-promoting factors in Saccharomyces cerevisiae. Genetics. 2005;171:35–47. doi: 10.1534/genetics.105.040634. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Dayel MJ, Holleran EA, Mullins RD. Arp2/3 complex requires hydrolyzable ATP for nucleation of new actin filaments. Proc Natl Acad Sci U S A. 2001;98:14871–14876. doi: 10.1073/pnas.261419298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dayel MJ, Mullins RD. Activation of Arp2/3 Complex: Addition of the First Subunit of the New Filament by a WASP Protein Triggers Rapid ATP Hydrolysis on Arp2. PLoS Biol. 2004;2:E91. doi: 10.1371/journal.pbio.0020091. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. De La Cruz EM, Pollard TD. Kinetics and thermodynamics of phalloidin binding to actin filaments from three divergent species. Biochemistry. 1996;35:14054–14061. doi: 10.1021/bi961047t. [DOI] [PubMed] [Google Scholar]
  12. Dominguez R. Actin-binding proteins--a unifying hypothesis. Trends Biochem Sci. 2004;29:572–578. doi: 10.1016/j.tibs.2004.09.004. [DOI] [PubMed] [Google Scholar]
  13. Emsley P, Cowtan K. Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr. 2004;60:2126–2132. doi: 10.1107/S0907444904019158. [DOI] [PubMed] [Google Scholar]
  14. Fujiwara I, Vavylonis D, Pollard TD. Polymerization kinetics of ADP- and ADP-Pi-actin determined by fluorescence microscopy. Proc Nat Acad Sci USA. 2007 doi: 10.1073/pnas.0702510104. In press. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Goley ED, Rodenbusch SE, Martin AC, Welch MD. Critical conformational changes in the Arp2/3 complex are induced by nucleotide and nucleation promoting factor. Mol Cell. 2004;16:269–279. doi: 10.1016/j.molcel.2004.09.018. [DOI] [PubMed] [Google Scholar]
  16. Graceffa P, Dominguez R. Crystal structure of monomeric actin in the ATP state. Structural basis of nucleotide-dependent actin dynamics. J Biol Chem. 2003;278:34172–34180. doi: 10.1074/jbc.M303689200. [DOI] [PubMed] [Google Scholar]
  17. Hertzog M, van Heijenoort C, Didry D, Gaudier M, Coutant J, Gigant B, Didelot G, Preat T, Knossow M, Guittet E, Carlier MF. The beta-thymosin/WH2 domain; structural basis for the switch from inhibition to promotion of actin assembly. Cell. 2004;117:611–623. doi: 10.1016/s0092-8674(04)00403-9. [DOI] [PubMed] [Google Scholar]
  18. Higgs HN, Pollard TD. Regulation of actin filament network formation through ARP2/3 complex: activation by a diverse array of proteins. Annu Rev Biochem. 2001;70:649–676. doi: 10.1146/annurev.biochem.70.1.649. [DOI] [PubMed] [Google Scholar]
  19. Jones TA, Zou JY, Cowan SW, Kjeldgaard Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr A. 1991;47(Pt 2):110–119. doi: 10.1107/s0108767390010224. [DOI] [PubMed] [Google Scholar]
  20. Kabsch W, Mannherz HG, Suck D, Pai EF, Holmes KC. Atomic structure of the actin:DNase I complex. Nature. 1990;347:37–44. doi: 10.1038/347037a0. [DOI] [PubMed] [Google Scholar]
  21. Krebs WG, Gerstein M. The morph server: a standardized system for analyzing and visualizing macromolecular motions in a database framework. Nucleic Acids Res. 2000;28:1665–1675. doi: 10.1093/nar/28.8.1665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Laham LE, Lamb JA, Allen PG, Janmey PA. Selective binding of gelsolin to actin monomers containing ADP. J Biol Chem. 1993;268:14202–14207. [PubMed] [Google Scholar]
  23. Le Clainche C, Didry D, Carlier MF, Pantaloni D. Activation of Arp2/3 complex by Wiskott-Aldrich Syndrome protein is linked to enhanced binding of ATP to Arp2. J Biol Chem. 2001;276:46689–46692. doi: 10.1074/jbc.C100476200. [DOI] [PubMed] [Google Scholar]
  24. Le Clainche C, Pantaloni D, Carlier MF. ATP hydrolysis on actin-related protein 2/3 complex causes debranching of dendritic actin arrays. Proc Natl Acad Sci U S A. 2003;100:6337–6342. doi: 10.1073/pnas.1130513100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Martin AC, Welch MD, Drubin DG. Arp2/3 ATP hydrolysis-catalysed branch dissociation is critical for endocytic force generation. Nat Cell Biol. 2006;8:826–833. doi: 10.1038/ncb1443. [DOI] [PubMed] [Google Scholar]
  26. Martin AC, Xu XP, Rouiller I, Kaksonen M, Sun Y, Belmont L, Volkmann N, Hanein D, Welch M, Drubin DG. Effects of Arp2 and Arp3 nucleotide-binding pocket mutations on Arp2/3 complex function. J Cell Biol. 2005;168:315–328. doi: 10.1083/jcb.200408177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Melki R, Fievez S, Carlier MF. Continuous monitoring of Pi release following nucleotide hydrolysis in actin or tubulin assembly using 2-amino-6-mercapto-7-methylpurine ribonucleoside and purine-nucleoside phosphorylase as an enzyme-linked assay. Biochemistry. 1996;35:12038–12045. doi: 10.1021/bi961325o. [DOI] [PubMed] [Google Scholar]
  28. Nolen BJ, Littlefield RS, Pollard TD. Crystal structures of actin-related protein 2/3 complex with bound ATP or ADP. Proc Natl Acad Sci U S A. 2004;101:15627–15632. doi: 10.1073/pnas.0407149101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Otterbein LR, Graceffa P, Dominguez R. The crystal structure of uncomplexed actin in the ADP state. Science. 2001;293:708–711. doi: 10.1126/science.1059700. [DOI] [PubMed] [Google Scholar]
  30. Otwinowski Z, Minor W. Processing of X-ray diffraction data collected in oscillation mode. Methods in Enzymology. 1997;276:307–326. doi: 10.1016/S0076-6879(97)76066-X. [DOI] [PubMed] [Google Scholar]
  31. Pollard TD. Regulation of Actin Filament Assembly by Arp2/3 Complex and Formins. Annu Rev Biophys Biomol Struct. 2007;36 doi: 10.1146/annurev.biophys.35.040405.101936. [DOI] [PubMed] [Google Scholar]
  32. Pollard TD, Goldberg I, Schwarz WH. Nucleotide exchange, structure, and mechanical properties of filaments assembled from ATP-actin and ADP-actin. J Biol Chem. 1992;267:20339–20345. [PubMed] [Google Scholar]
  33. Robinson RC, Turbedsky K, Kaiser DA, Marchand JB, Higgs HN, Choe S, Pollard TD. Crystal structure of Arp2/3 complex. Science. 2001;294:1679–1684. doi: 10.1126/science.1066333. [DOI] [PubMed] [Google Scholar]
  34. Rould MA, Wan Q, Joel PB, Lowey S, Trybus KM. Crystal structures of expressed non-polymerizable monomeric actin in the ADP and ATP states. J Biol Chem. 2006;281:31909–31919. doi: 10.1074/jbc.M601973200. [DOI] [PubMed] [Google Scholar]
  35. Selden LA, Kinosian HJ, Estes JE, Gershman LC. Impact of profilin on actin-bound nucleotide exchange and actin polymerization dynamics. Biochemistry. 1999;38:2769–2778. doi: 10.1021/bi981543c. [DOI] [PubMed] [Google Scholar]
  36. Strzelecka-Golaszewska H, Moraczewska J, Khaitlina SY, Mossakowska M. Localization of the tightly bound divalent-cation-dependent and nucleotide-dependent conformation changes in G-actin using limited proteolytic digestion. Eur J Biochem. 1993;211:731–742. doi: 10.1111/j.1432-1033.1993.tb17603.x. [DOI] [PubMed] [Google Scholar]
  37. Vorobiev S, Strokopytov B, Drubin DG, Frieden C, Ono S, Condeelis J, Rubenstein PA, Almo SC. The structure of nonvertebrate actin: implications for the ATP hydrolytic mechanism. Proc Natl Acad Sci U S A. 2003;100:5760–5765. doi: 10.1073/pnas.0832273100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Wilbanks SM, McKay DB. How potassium affects the activity of the molecular chaperone Hsc70. II. Potassium binds specifically in the ATPase active site. J Biol Chem. 1995;270:2251–2257. doi: 10.1074/jbc.270.5.2251. [DOI] [PubMed] [Google Scholar]
  39. Wriggers W, Schulten K. Investigating a back door mechanism of actin phosphate release by steered molecular dynamics. Proteins. 1999;35:262–273. [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Movie Legends
NIHMS25453-supplement-01.doc (36KB, doc)
Movie 1
Download video file (5.2MB, mpg)
Movie 2
Download video file (7MB, mpg)

RESOURCES