Current address: Hoffmann-La Roche, Nutley, New Jersey, United States of America.
Current address: Department of Biochemistry and Molecular Biology, University of South Alabama, Mobile, Alabama, United States of America.
Conceived and designed the experiments: AN LLL DLB MPJ. Performed the experiments: AN LLL DLB MPJ. Analyzed the data: AN LLL DLB MPJ. Wrote the paper: AN LLL DLB MPJ.
I have read the journal's policy and have the following conflicts: MPJ is a member of the scientific advisory board of Schrodinger LLC.
Actin filament assembly by the actin-related protein (Arp) 2/3 complex is necessary to build many cellular structures, including lamellipodia at the leading edge of motile cells and phagocytic cups, and to move endosomes and intracellular pathogens. The crucial role of the Arp2/3 complex in cellular processes requires precise spatiotemporal regulation of its activity. While binding of nucleation-promoting factors (NPFs) has long been considered essential to Arp2/3 complex activity, we recently showed that phosphorylation of the Arp2 subunit is also necessary for Arp2/3 complex activation. Using molecular dynamics simulations and biochemical assays with recombinant Arp2/3 complex, we now show how phosphorylation of Arp2 induces conformational changes permitting activation. The simulations suggest that phosphorylation causes reorientation of Arp2 relative to Arp3 by destabilizing a network of salt-bridge interactions at the interface of the Arp2, Arp3, and ARPC4 subunits. Simulations also suggest a gain-of-function ARPC4 mutant that we show experimentally to have substantial activity in the absence of NPFs. We propose a model in which a network of auto-inhibitory salt-bridge interactions holds the Arp2 subunit in an inactive orientation. These auto-inhibitory interactions are destabilized upon phosphorylation of Arp2, allowing Arp2 to reorient to an activation-competent state.
The Arp2/3 complex consists of seven associated protein subunits including Arp2 and Arp3 that play a central role in the formation of actin filaments. Filament formation by the Arp2/3 complex drives important cell processes such as cell movement and endocytosis. The function of the Arp2/3 complex is highly regulated, and improper regulation of its activity has been linked to cancer metastasis. One level of regulation is post-translational phosphorylation, in which a −2 charged phosphate group is added to the uncharged amino acids threonine 237 and 238 of Arp2. We use molecular dynamics simulations and biochemical studies to show that Arp2 phosphorylation results in large structural changes of the Arp2/3 complex consistent with low-resolution structural studies. The simulations suggest phosphorylation allows the complex to reorient to an activation competent state by destabilizing interactions that hold Arp2 in an inactive position. Further simulations suggested that mutation of the Arp2/3 complex could allow complex activation, and we verified this gain-of-function mutation biochemically. We propose a model for Arp2/3 complex activation in which phosphorylation destabilizes the inactive state of the complex, allowing structural changes that are permissive for activation by nucleation-promoting factors and binding to the mother filament.
Spatial and temporal control of the assembly and disassembly of actin filaments is crucial for a number of distinct cell processes, including endocytosis and cell migration
Direct regulation of the nucleating activity of the Arp2/3 complex leads to a second level of control over actin filament assembly. The Arp2/3 complex is composed of seven subunits: the actin-related proteins Arp2 and Arp3, and ARPC1–5. While binding of the mother filament contributes significantly to activation
The structures of the
An additional, more recently identified requirement for activating nucleation by the Arp2/3 complex is threonine or tyrosine phosphorylation of the Arp2 subunit
Computational studies have the potential to elucidate aspects of Arp2/3 complex function and regulation. While the impact of computation in this regard has been limited thus far due to the large system size, molecular dynamics simulations have been used to examine dynamics of the ATP binding cleft in Arp2 and Arp3
Here, we use unbiased molecular dynamics simulations to determine how phosphorylation at Arp2 T237 and T238 may change the structure of the Arp2/3 complex and permit activation by NPFs. We find large conformational changes in the Arp2/3 complex upon phosphorylation, including the reorientation of Arp2 relative to Arp3, toward the short-pitch dimer orientation. Our simulations suggest a mechanism by which a complex network of positively and negatively charged amino acids at the Arp2/Arp3/ARPC4 interface holds the complex in an inactive configuration, and phosphorylation disrupts these auto-inhibitory interactions. To test this prediction, we designed, based on further computational simulations, mutations of the Arp2/3 complex that we predicted would disrupt the auto-inhibitory interactions. Biochemical assays reveal that this mutant, R105/106A ARPC4, does in fact show nucleation activity even in the absence of NPFs.
We previously reported that phosphorylation of Arp2 T237/238 or Y202 is necessary for activation of the Arp2/3 complex in the presence of NPF
The rate of assembly of pyrene-labeled actin into filaments, an index of Arp2/3 complex nucleation activity, was similar with WT rArp2/3 (0.193 nM filament ends) and with Arp2/3 complex purified from bovine thymus (0.209 nM filament ends at concentrations of 5 nM Arp2/3 complex with 4 µM actin) (
In the absence of NPF, the rate of actin filament assembly and the concentration of filament ends with mutant rArp2/3 complex containing T237/238A-Y202A Arp2 (t1/2 = 500 s, 0.260 nM) were similar to native and WT complexes (
To nucleate a new actin filament, the Arp2/3 complex binds to filament pointed ends, which reflects its capping activity. We used actin seeds capped at the barbed end with gelsolin to measure pointed end capping by WT and mutant Arp2 rArp2/3 complex. Gelsolin-capped actin filaments elongated from their pointed ends in the presence of 4 µm actin (
We previously hypothesized that phosphorylation may induce a conformational change that allows activation by mother filament and NPF
Due to the large size of the Arp2/3 complex, the simulations required substantial computational resources on a supercomputer. We performed simulations only with Arp2 T237 and T238 individually phosphorylated, as well as the unphosphorylated ‘control’ simulation. For each of these systems, we ran duplicate 30 ns molecular dynamics simulations to control for simulation dependence on the initial conditions and stochastic fluctuations. Observing large conformational changes using molecular dynamics simulations is difficult because of the gap between experimental and computationally feasible timescales, and we do not expect to see the full range of structural change in these simulations. Nonetheless, large conformational changes were observed for Arp2/3 when phosphorylated on either T237 or T238 of Arp2. Backbone root mean square deviations (RMSDs) following global alignment of simulation snapshots to the starting model revealed modest conformational changes of 3–4 Å RMSD for the unphosphorylated simulations compared with larger conformational changes of 4–8 Å for the phosphorylated simulations (
The conformational changes induced by phosphorylation were dominated by changes in the orientation of Arp2, ARPC1, and ARPC3 relative to other subunits (
(a) Arp3 and Arp2 subunits from snapshots from the last ns of the unphosphorylated (Arp3-gray; Arp2-blue) and phosphorylated T237 Arp2 (tan, red) wild-type simulations are shown following alignment of Cα atoms of subdomains 1 and 2 of Arp3. These and all other structural figures were produced using the molecular graphics program UCSF Chimera
Analyzing the Cα RMSD of each residue of Arp2 after alignment of Arp3 subdomains 1 and 2 reveals that the largest changes in Cα position of Arp2 occur in the C-terminal tail as well as in subdomains 1 and 2, which are disordered in most unphosphorylated Arp2/3 complex crystal structures (
The results of the molecular dynamics simulations are consistent with the hypothesis that phosphorylation induces conformational changes that contribute to adopting a nucleation-competent form
Stereoimages of the salt-bridge network between Arp2 (pink) and ARPC4 (green) in simulations of wild-type (a) unphosphorylated or (b) phosphorylated Arp2 T237 Arp2/3 complex.
By contrast, these interactions were dramatically rearranged with phosphorylation of either T237 or T238 (
Because T237 and T238 are near the interfaces with several other subunits, perturbations to the electrostatic network induced by phosphorylation can cause large conformational changes. We hypothesized that mutating key residues that interact with pT237 and pT238 would abolish the ability of phosphorylation to induce these conformational changes, and hence activation of the nucleation activity. This hypothesis is based on studies of phosphorylation-mediated activation in systems such as protein kinases, in which attractive interactions with arginine residues that interact with phosphorylated residues drive conformational changes key to phospho-activation
Contrary to our expectations, simulations of the unphosphorylated Arp2/3 complex with the R105A ARPC4 mutation produced a similar, but somewhat smaller, structural change to that induced by phosphorylation at T237 or T238 Arp2 (
Stereoimages of changes in Arp2-Arp3 orientation in R105A ARPC4 mutant Arp2/3 complex from snapshots from the last ns of the unphosphorylated wild-type(Arp3-gray; Arp2-blue) and the (a) unphosphorylated R105A ARPC4 mutant complex (pink, orange) or (b) phosphorylated T237 Arp2 ARPC4 R105A ARPC4 mutant complex (cyan, magenta) simulations are shown following alignment of Cα atoms of subdomains 1 and 2 of Arp3. The other subunits of the complex from the snapshot of the unphosphorylated simulation, colored as in
(a) Distribution of root-mean-square deviations (RMSD) of Cα atoms of Arp2 to Arp2 atoms in the active short-pitch dimer orientation (B. Nolen, personal communication) after alignment of subdomains 1 and 2 of Arp3 over the last 20 ns of simulations. (b) Distribution of number of contacts between Arp3 and Arp2 heavy atoms over the last 20 ns of simulations. Coloring is as follows: U(black) – unphosphorylated; R105A(orange) – unphosphorylated R105A ARPC4 mutant; and pT237/R105A(magenta) – Arp2 T237 phosphorylated R105A ARPC4 mutant. Green line represents the corresponding RMSD (31.2 Å) and number of contacts (35) for the unphosphorylated starting model as in
To test predictions from our molecular dynamics simulations on the role of arginines in ARPC4, we generated rArp2/3 complex with R105 and R106 of ARPC4 mutated to alanine (R105/106A ARPC4). While we did not simulate structural changes associated with the R106A ARPC4 mutation, the simulations indicated that R106 ARPC4 upon T238 phosphorylation played a role analogous to that of R105 ARPC4, forming a stable interaction with the phosphate group (
(a) Recombinant Arp2/3 complex WT (rWT) and R105/106A ARPC4 expressed and purified from sf21 cells. (b) Pyrene actin assembly assays comparing nucleation activity of rWT and R105/106A ARPC4 rArp2/3 complex. (c) Pointed end capping assay with actin alone (black), rWT (red) and R105/106A ARPC4 rArp2/3 (green), and R105/106A ARPC4 rArp2/3 pretreated with Antarctic phosphatase (AP-R105/106A ARPC4) (purple).
Pointed end capping by the R105/106A ARPC4 rArp2/3 complex appeared similar to WT rArp2/3. Using gelsolin-capped actin seeds, the rate of pointed end elongation was not significantly different between mutant and wild-type Arp2/3 complex, and treating with AP only slightly attenuated capping activity of the mutant (
Our simulations revealed large structural differences between the phosphorylated or mutant complexes and the unphosphorylated wild-type complex. These structural changes included the movement of Arp2 toward the active short-pitch dimer orientation relative to Arp3. However, the conformational changes observed in these short molecular dynamics simulations are much smaller than those assumed to occur upon full activation, and even with longer simulations we would not expect the phosphorylated complex to adopt the putative active conformation, since phosphorylation is necessary but not sufficient for activation. Rather, we propose a model in which phosphorylation destabilizes the inactive state, leading to conformational changes that relieve the auto-inhibition and thus are permissive for full activation by NPF binding.
It is impossible to say whether the actual structural differences in response to phosphorylation or mutation are realized at the end of these simulations, but this is very unlikely. Each of the duplicate simulations of the same state of the complex show differences, even the dual simulations of the unphosphorylated state (see, for example, the two peaks in the Arp2 RMSD to the active orientation (
Besides the changes in the Arp2-Arp3 orientation, large changes in the orientation of ARPC1 and ARPC3 relative to Arp3 were observed upon phosphorylation (
The findings here and previously
In all cases in which phosphorylation is required for functional activation, the unphosphorylated state can be considered auto-inhibited. A common mechanism for converting from the auto-inhibited to the activated state is one in which phosphorylation induces attractive interactions between the phosphorylated residue and other residues that are required for activating structural changes. This is seen in Ser/Thr protein kinases, and also in more traditionally auto-inhibited systems such as the phosphorylation of the tail of Tyr protein kinases (reviewed in
In contrast, mutation of arginine residues in the Arp2/3 complex enhances activation by phosphorylation (
Based on the activation model laid forth by Dalhaimer and Pollard
Due to the large computational expense of the simulations (∼500,000 cpu-hours during the course of this study), we were unable to perform all of the potentially informative simulations. For example, we have not investigated phosphorylation of Y202 on Arp2, which has been proposed as the site of tyrosine phosphorylation and is located close to T237/238 Arp2. The proximity of Y202 to the salt-bridge network around T238 leads us to speculate that its phosphorylation would exert its effects by a similar mechanism, but this hypothesis remains to be examined. We have also not investigated dual phosphorylation of both T237 and T238 Arp2. Nonetheless, despite these limitations and the short timescale probed by molecular dynamics, the simulations suggested a structural mechanism for phosphoregulation of Arp2/3 and predicted a gain-of-function mutation, which was confirmed experimentally. As such, this study provides an example of how computational simulations can be used to create testable models of regulatory phosphorylation, which is valuable when it is difficult to obtain direct, atomic-resolution structural information, as is often the case. Here, we have provided a new model for Arp2/3 regulation in which a network of electrostatic interactions helps to hold the complex in the inactive state, and this auto-inhibition must be relieved by phosphorylation to permit activation.
Plasmids encoding Arp2/3 complex subunits were obtained from M. Welch (UC Berkeley) and were generated as described
Recombinant Arp2/3 complex was expressed and purified as described
Pyrene actin polymerization assays were performed with 4 µM monomeric actin containing 5% pyrene-labeled actin in KMEI (50 mM KCl, 1 mM MgCl2, 1 mM EGTA, and 10 mM imidazole, pH 7), 2.5 to 50 nM Arp2/3 complex, and 500 nM N-WASP VCA domain. Measurements were made with an RF-5301PC spectrophotometer (Shimadzu) at 1 s intervals. Growing filament ends were calculated by determining the rate of actin assembly at 80% of polymerization and using the relationship
Binding constants of Arp2/3 complex for NPFs were determined by using GST-NWASP VCA covalently coupled to Activated CH-sepharose 4B (GE Healthcare, Piscataway, NJ). GST-NPF-coupled beads were added to mock-treated or Antarctic phosphatase-treated Arp2/3 complex and incubated at room temperature for 30 min. NPF-coupled beads were spun at 700× g for 5 min, the supernatant removed and beads resuspended in SDS-PAGE sample buffer. Coomasie-stained gels were scanned and quantified using a LabWorks imaging system and LabWorks Software (UVI, CA). The data were plotted and fitted using GraphPad Prism software (GraphPad Software, Inc., San Diego, CA). Binding constants for Arp2/3 complex for actin filaments were determined by actin co-sedimentation as described
Systems were prepared for molecular dynamics simulations starting from the crystal structure of the
The full system was then energy minimized using DESMOND
Each system was then simulated for 30 ns using the Martyna-Tobias-Klein integrator
Coordinates of the Cα atoms from the last 20 ns of each unphosphorylated, Arp2 pThr237, and Arp2 pThr238 simulation were collected into a single trajectory on which Principal Component Analysis
A contact between the Arp2 and Arp3 subunits was defined as the number of heavy atoms in Arp3 that were within 3.5 Å of any heavy atom in Arp2. The number of contacts between the Arp2 and Arp3 subunits was calculated for every 10th frame (100 ps) of each simulation. The results for the duplicate simulations of each wild-type or mutant complex were then pooled and compared.
(TIFF)
(TIFF)
(TIFF)
(TIFF)
(TIFF)
(TIFF)
(TIFF)
(TIFF)
(DOC)
We thank Drs. Michael Gonzales and Byoung-Do Kim of the TACC supercomputer center and Dr. Justin Gullingsrud of D.E. Shaw Research for assistance with the simulations. We also thank the Matthew Welch laboratory (University of California, Berkeley) for help in preparing recombinant Arp2/3 complex, and Torsten Wittmann, Bradley Webb, and André Schönichen for helpful comments on the manuscript. Finally, we thank the reviewers of this manuscript, whose comments substantially improved this work.