Actin: Simulations and Animations

Actin filaments are dynamic polymers whose ATP-driven assembly in the cell cytoplasm drives shape changes, cell locomotion and chemotactic migration. Actin filaments also participate in muscle contraction. The structure of the filament is not known at atomic resolution, but several models were produced in the laboratory of Ken Holmes (MPI for medical research, Heidelberg, Germany) by refinement against X-ray fiber diffraction data:

Fig. 1 (click to enlarge): Lorenz model of F-actin. A single G-actin monomer with inter-actin contact surfaces is shown on the right, the entire F-actin on the left. This Figure, created with the graphics program VMD, is actually quite popular and has been downloaded by numerous researchers and printed in two textbooks. Please see copyright notice below.

1. Actin's "Back Door"

The mechanical properties of the filaments can be altered by actin-binding agents such as the toxin phalloidin from the mushroom "Amanita phalloides". Phalloidin binding to actin has been shown to delay the release of inorganic phosphate after ATP hydrolysis (Dancker & Hess, Biochim. Biophys. Acta, 1990, 1035:197). Conversely, no effect of inorganic phosphate on the phalloidin binding kinetics has been observed (De La Cruz & Pollard, Biochemistry, 1994, 33:14378). Our research goal was to resolve this seemingly paradoxical situation by identifying the dissociation pathway of the phosphate.

The simulation of monomeric actin required the development of a proper force field for ATP and ADP nucleotides (parameters available by download). The major discovery was the diffusion of water molecules into the buried nucleotide binding site along two distinct pathways (Fig. 2). Of special interest is the "back door" diffusion pathway which we believe to be relevant to the dissociation of the phosphate after hydrolysis. Our hypothesis for phosphate release would explain the seemingly paradoxical kinetics of actin-phalloidin interaction. Phalloidin, whose position in the F-actin filament structure is known, would in our model block the exit of the back door pathway and, thus, delay the release of the phosphate.

Fig. 2 (click to enlarge): Diffusion of water molecules into the nucleotide binding site of actin [visualized with the program VMD (side view)].

ATP and the Mg++ ion are represented as white spheres in the center of the protein, which is rendered as a transparent ribbon. The color of the traces of the water oxygens codes for different simulation trajectories. The Figure reveals two water diffusion pathways with the "back door" pathway on the right.

2. Actin Phosphate Release - The Movie

In order to substantiate our back door hypothesis of actin phosphate dissociation we studied further the dissociation of the nucleotide and the role of the proposed back door pathway for actin's enzymatic activity. To reveal the microscopic processes underlying the release of the inorganic phosphate, we explored possible release pathways in steered molecular dynamics simulations by measuring adhesion forces when pulling the substrate out of its binding pocket. The resulting structural prediction of the dissociation mechanism has important implications also for other putative back door enzymes such as the ATP-driven motor protein myosin (Yount et al., Biophys J., 1995, 68:44s).

Animations (total length 4 minutes):

160x120 RealVideo 9 (modem speed) [ part 1 | part 2 | part 3 | part 4 | part 5 ] (616 KB)
320x240 RealVideo 9 (T1/LAN speed) [ part 1 | part 2 | part 3 | part 4 | part 5 ] (10 MB)
640x480 Quicktime 6 [ part 1 | part 2 | part 3 | part 4 | part 5 ] (29 MB)

Based on these studies, we proposed that the imidazole moiety of histidine 73, which is methylated in most actins, interacts with the dissociating phosphate and that electrostatic stabilization retards its release from actin following ATP hydrolysis. To investigate the effect of H73 further, mutagenesis experiments (substitution with R,K,A,E,Q) were performed in the laboratory of Peter Rubenstein. The observed trends in these experiments are consistent with results obtained by simulated annealing simulations of the mutant actins. Two publications summarizing the latest experimental and simulation results were published in JCB (see below).

3. Actin Filament Severing by Cofilin

Fig. 3 (click to enlarge): Cover figure of the Oct. 9, 1998, issue of J. Molec. Biol. Shown on the left are the nonbonded interaction energies arising in the complex of cofilin (top) with actin (bottom). Shown on the right are the contact surfaces in the complex, colored by atom type.

An understanding of the actin-depolymerizing function attributed to members of the ADF/cofilin/destrin superfamily requires a structural model of these proteins in complex with actin. As a step toward defining actin-cofilin interactions, the complex of yeast cofilin with monomeric actin was predicted by molecular dynamics simulation in collaboration with Jay X. Tang and Paul A. Janmey. The predicted complex exhibits strong interactions between the N-termini of actin and cofilin, mediated by a salt bridge of cofilin Arg3 with actin Asp1. The forming of this salt bridge could be prevented by the phosphorylation of cofilin Ser4 which is believed to inhibit cofilin depolymerization activity. Recent mutagenesis studies, crosslinking experiments and peptide binding studies are consistent with the predicted model of the actin-cofilin complex. The structural homology between cofilin and gelsolin segment-1 binding to actin was confirmed experimentally by two types of competitive binding assays.

Fig. 4 (click to enlarge): Filament severing by steric exclusion. When aligned to the actin filament in the monomer binding mode, cofilin exhibits steric clashes with actin's subdomain 2 (black). The model is consistent with a conformational transition from the filamentous to the monomeric cofilin binding mode that weakens the interactions of adjacent actin monomers in the filament.

More recently, we modelled the atomic structure of the cofilin-F-actin complex based on electron microscopy data from the laboratory of Edward H. Egelman:

Fig. 5 (click to enlarge): Cover figure of the April 2, 2001 issue of Journal of Cell Biology. The image shows the atomic model (left) and an electron microscopy map (right) of doubly ADF-decorated actin filaments.

4. Cross-Links and Actin Filament Structure

We refined the model of F-actin further by modeling of cysteine cross-linking constraints on the structure. The cross-links are introduced experimentally in the laboratory of Emil Reisler and provide stringent constraints for the arrangement of actin subunits and their contact interface. The DNase I binding loop (residues 38-52), the hydrophobic plug (residues 262-274), and the C-terminus region are among the structural elements of monomeric (G-) actin that were proposed to form the intermonomer interface in F-actin. Cysteines were introduced into yeast actin either at residue 41 or 265. These mutations allowed for specific cross-linking of F-actin between C41 and C265, C265 and C374, and C41 and C265. The calculated cross-linked structures showed a better fit to the Holmes than to the refined Lorenz model of F-actin. It is predicted on the basis of such calculations that image reconstruction of electron micrographs of disulfide cross-linked C41-C374 F-actin should provide a conclusive test of these two similar models of F-actin structure.

Fig. 6 (click to enlarge): Multi-resolution rendering of conformational changes induced by the 41-374 disulfide bridge. (a) The Holmes model of F-actin after refinement. Individual monomers are shown as colored backbone traces, Ca2+-ADP as black spheres. The 41-374 cross- bridges are shown in red. (b) Low-resolution map of the original Holmes model, shown as an isocontour at half-maximal density. The spatial resolution of the map (full width half-maximum of the Gaussian kernel) is 13.6A (c) Low-resolution map of the refined model shown as a wire-mesh, otherwise rendered as in (b). In addition, the difference density map relative to (b) is visualized by a red isocontour at -10% of the maximum density value. Positive difference densities are below the +10% density threshold. (d) The refined Lorenz model with the cross-linked sites (red). (e) Low-resolution map of the original Lorenz model, shown as an isocontour at half-maximal density. (f) Low-resolution map of the Lorenz model in the presence of the 41-374 cross-bridge shown as a wire-mesh. The difference density map relative to (e) is visualized by isocontours at -10% (red) and + 10% (blue) of the maximum density value.

References

Xiaoyi Yao, Vinh Nguyen, Willy Wriggers, and Peter Rubenstein. Regulation of Yeast Actin Behavior by Interaction of Charged Residues Across the Interdomain Cleft. J. Biol. Chem., 2002, Vol. 25, pp. 22875-22882. [Abstract] [Article] [Predicted Structures]
Vitold E. Galkin, Albina Orlova, Natalya Lukoyanova, Willy Wriggers, and Edward H. Egelman. ADF Stabilizes an Existing State of F-actin and Can Change the Tilt of F-actin Subunits. J. Cell Biol., 2001, Vol. 153, pp. 75-86 [Abstract] [Article] [Predicted Structure]
Eldar Kim, Willy Wriggers, Martin Phillips, Kevin Kokabi, Peter Rubenstein, and Emil Reisler. Cross-Linking Constraints on F-actin Structure. J. Molec. Biol., 2000, Vol. 299, pp. 421-429. [Abstract] [Article] [Predicted Structures]
Xiaoyi Yao, Stephanie Grade, Willy Wriggers, and Peter Rubenstein. His73, Often Methylated, Is an Important Structural Determinant for Actin - A Mutagenic Analysis of H73 of Yeast Actin. J. Biol. Chem., 1999, Vol. 274, pp. 37443-37449. [Abstract] [Article] [Predicted Structures]
Willy Wriggers and Klaus Schulten. Investigating a Back Door Mechanism of Actin Phosphate Release by Steered Molecular Dynamics. Proteins: Structure, Function, and Genetics, 1999, Vol. 35, pp. 262-273. [Abstract] [Article]
Sergei Izrailev, Sergey Stepaniants, Barry Isralewitz, Dorina Kosztin, Hui Lu, Ferenc Molnar, Willy Wriggers, and Klaus Schulten. Steered Molecular Dynamics. Chapter in "Computational Molecular Dynamics: Challenges, Methods, Ideas", Springer Verlag, 1999 [Abstract] [Article - 25MB]
Willy Wriggers, Jay X. Tang, Toshifumi Azuma, Peter Marks, and Paul A. Janmey. Cofilin and Gelsolin Segment-1: Molecular Dynamics Simulation and Biochemical Analysis Predict a Similar Actin Binding Mode. J. Molecular Biology, 1998, Vol. 282, pp. 921-932. [Abstract] [Article] [Predicted Structure]
Willy Wriggers and Klaus Schulten. Stability and Dynamics of G-actin: Back Door Water Diffusion and Behavior of a Subdomain 3/4 Loop. Biophysical Journal 1997, Vol. 73, pp. 624-639. [Abstract] [Article - 27MB] [Nucleotide Parameters]
Thomas E. Angelini, Hongjun Liang, Willy Wriggers and Gerard Wong. Like-charge attraction between polyelectrolytes induced by counterion charge density waves. PNAS, 2003, Vol. 100, pp. 8634-8637. [Abstract] [Article] [Software]