Structural modeling and molecular dynamics simulation of the actin filament

Actin is a major structural protein of the eukaryotic cytoskeleton and enables cell motility. Here, we present a model of the actin filament (F-actin) that not only incorporates the global structure of the recently published model by Oda et al. but also conserves internal stereochemistry. A comparison is made using molecular dynamics simulation of the model with other recent F-actin models. A number of structural determents such as the protomer propeller angle, the number of hydrogen bonds, and the structural variation among the protomers are analyzed. The MD comparison is found to reflect the evolution in quality of actin models over the last 6 years. In addition, simulations of the model are carried out in states with both ADP or ATP bound and local hydrogen-bonding differences characterized.     

Coarse-grained molecular simulations of large biomolecules

Recently, we saw a dramatic increase in the number of researches that rely on coarse-grained (CG) simulations for large biomolecules. Here, first, we briefly describe recently developed and used CG models for proteins and nucleic acids. Balance between structure-based and physico-chemical terms is a key issue. We also discuss the multiscale algorithms used to derive CG parameters. Next, we comment on the dynamics used in CG simulations with an emphasis on the importance of hydrodynamic interactions. We then discuss the pros and cons of CG simulations. Finally, we overview recent exciting applications of CG simulations. Publicly available tools and software for CG simulations are also summarized.     

Coarse-grained modeling of conformational transitions underlying the processive stepping of myosin V dimer along filamentous actin

To explore the structural basis of processive stepping of myosin V along filamentous actin, we have performed comprehensive modeling of its key conformational states and transitions with an unprecedented residue level of details. We have built structural models for a myosin V monomer complexed with filamentous actin at four biochemical states [adenosine diphosphate (ATP)-, adenosine diphosphate (ADP)-phosphate-, ADP-bound or nucleotide-free]. Then we have modeled a myosin V dimer (consisting of lead and rear head) at various two-head-bound states with nearly straight lever arms rotated by intramolecular strain. Next, we have performed transition pathway modeling to determine the most favorable sequence of transitions (namely, phosphate release at the lead head followed by ADP release at the rear head, while ADP release at the lead head is inhibited), which underlie the kinetic coordination between the two heads. Finally, we have used transition pathway modeling to reveal the order of structural changes during three key biochemical transitions (phosphate release at the lead head, ADP release and ATP binding at the rear head), which shed lights on the strain-dependence of the allosterically coupled motions at various stages of myosin V's work cycle. Our modeling results are in agreement with and offer structural insights to many results of kinetic, single-molecule and structural studies of myosin V.     

Under physiological conditions actin disassembles slowly from the nonpreferred end of an actin filament

Incubation of the isolated acrosomal bundles of Limulus sperm with skeletal muscle actin results in assembly of actin onto both ends of the bundles. Because of the taper of these cross-linked bundles of actin filaments, one can distinguish directly the preferred end for assembly from the nonpreferred end. Loss of growth with time from the nonpreferred end was directly assessed by electron microscopy and found to be dependent upon salt concentration. Under physiological conditions (100 mM KCl, 1 mM MgCl2) and excess ATP (0.5 mM), depolymerization of the newly assembled actin filaments at the nonpreferred end over an 8-h period was 0.024 micron/h. Thus, even after 8 h, 63% of the bundles retained significant growth on their nonpreferred ends, the average length being 0.21 micron +/- 0.04. However, in the presence of 1.2 mM CaCl2, disassembly of actin monomers from the nonpreferred end increased substantially. By 8 h, only 7% of the bundles retained any actin growth on the nonpreferred ends, and the depolymerization rate off the nonpreferred end was 0.087 micron/h. From these results we conclude that, in the absence of other cellular factors, disassembly of actin subunits from actin filaments (subunit exchange) is too slow to influence most of the motile events that occur in cells. We discuss how this relates to treadmilling.     

Dynamic polymorphism of single actin molecules in the actin filament

Actin filament dynamics are critical in cell motility. The structure of actin filament changes spontaneously and can also be regulated by actin-binding proteins, allowing actin to readily function in response to external stimuli. The interaction with the motor protein myosin changes the dynamic nature of actin filaments. However, the molecular bases for the dynamic processes of actin filaments are not well understood. Here, we observed the dynamics of rabbit skeletal-muscle actin conformation by monitoring individual molecules in the actin filaments using single-molecule fluorescence resonance energy transfer (FRET) imaging with total internal reflection fluorescence microscopy (TIRFM). The time trajectories of FRET show that actin switches between low- and high-FRET efficiency states on a timescale of seconds. If actin filaments are chemically cross-linked, a state that inhibits myosin motility, the equilibrium shifts to the low-FRET conformation, whereas when the actin filament is interacting with myosin, the high-FRET conformation is favored. This dynamic equilibrium suggests that actin can switch between active and inactive conformations in response to external signals.     

Coarse-grained MD simulations of membrane protein-bilayer self-assembly

Complete determination of a membrane protein structure requires knowledge of the protein position within the lipid bilayer. As the number of determined structures of membrane proteins increases so does the need for computational methods which predict their position in the lipid bilayer. Here we present a coarse-grained molecular dynamics approach to lipid bilayer self-assembly around membrane proteins. We demonstrate that this method can be used to predict accurately the protein position in the bilayer for membrane proteins with a range of different sizes and architectures.     

Mechanism of actin filament nucleation

We used computational methods to analyze the mechanism of actin filament nucleation. We assumed a pathway where monomers form dimers, trimers, and tetramers that then elongate to form filaments but also considered other pathways. We aimed to identify the rate constants for these reactions that best fit experimental measurements of polymerization time courses. The analysis showed that the formation of dimers and trimers is unfavorable because the association reactions are orders of magnitude slower than estimated in previous work rather than because of rapid dissociation of dimers and trimers. The 95% confidence intervals calculated for the four rate constants spanned no more than one order of magnitude. Slow nucleation reactions are consistent with published high-resolution structures of actin filaments and molecular dynamics simulations of filament ends. One explanation for slow dimer formation, which we support with computational analysis, is that actin monomers are in a conformational equilibrium with a dominant conformation that cannot participate in the nucleation steps.     

Coarse-grained simulations of the membrane-active antimicrobial Peptide maculatin 1.1

Maculatin 1.1 (M1.1) is a membrane-active antimicrobial peptide (AMP) from an Australian tree frog that forms a kinked amphipathic α-helix in the presence of a lipid bilayer or bilayer-mimetic environment. To help elucidate its mechanism of membrane-lytic activity, we performed a total of ∼8 μs of coarse-grained molecular dynamics (CG-MD) simulations of M1.1 in the presence of zwitterionic phospholipid membranes. Several systems were simulated in which the peptide/lipid ratio was varied. At a low peptide/lipid ratio, M1.1 adopted a kinked, membrane-interfacial location, consistent with experiment. At higher peptide/lipid ratios, we observed spontaneous, cooperative membrane insertion of M1.1 peptide aggregates. The minimum size for formation of a transmembrane (TM) aggregate was just four peptides. The absence of a simple and well-defined central channel, along with the exclusion of lipid headgroups from the aggregates, suggests that a pore-like model is an unlikely explanation for the mechanism of membrane lysis by M1.1. We also performed an extended 1.25 μs simulation of the permeabilization of a complete liposome by multiple peptides. Consistent with the simpler bilayer simulations, formation of monomeric interfacial peptides and TM peptide clusters was observed. In contrast, major structural changes were observed in the vesicle membrane, implicating induced membrane curvature in the mechanism of active antimicrobial peptide lysis. This contrasted with the behavior of the nonpore-forming model peptide WALP23, which inserted into the vesicle to form extended clusters of TM α-helices with relatively little perturbation of bilayer properties.     

Structure and dynamics of the actin filament

The structures of filamentous Mg-ATP-actin (F actin) in the presence and absence of KCl have been mapped with hydroxyl radicals (*OH) generated by synchrotron X-ray radiolysis. Proteolysis and mass spectrometry (MS) analysis revealed 52 reactive side-chain sites from 27 distinct peptides within actin. The reactivities of these probe sites with *OH in the F-actin states are compared with those of Mg-ATP-G-actin (monomers) analyzed previously [Guan, J.-Q. et al. (2003) Biochemistry 42, 11992-12000]. Filament-dependent protection within subdomains 2, 3, and 4 and at the C terminus is consistent with longitudinal contacts of monomers within the filament helical structure as predicted by the Holmes model. In the absence of KCl, the extent of filament-dependent protection rarely reached 3-fold, consistent with a highly dynamic filament characterized by relatively weak interactions between actin protomers. However, in the presence of KCl, the extents of protection are significantly increased, consistent with a well-ordered, more tightly packed filament structure. Filament-dependent enhancements of reactivity not predicted by the Holmes model are seen for a peptide that overlaps the "hydrophobic plug" (H-plug) region and for a peptide that forms contacts with the polyphosphate moiety of the bound nucleotide. Overall, these data are both consistent with and complementary to a recent deuterium-exchange MS study of filamentous actin [Chik, J. K., and Schriemer, D.C. (2003) J. Mol. Biol. 334, 373-385], which also did not detect any burial of the H plug upon formation of filaments.     

Coarse-grained modeling of mucus barrier properties

We designed a simple coarse-grained model of the glycocalyx layer, or adhesive mucus layer (AML), covered by mucus gel (luminal mucus layer) using a polymer lattice model and stochastic sampling (replica exchange Monte Carlo) for canonical ensemble simulations. We assumed that mucin MUC16 is responsible for the structural properties of the AML. Other mucins that are much smaller in size and less relevant for layer structure formation were not included. We further assumed that the system was in quasi-equilibrium. For systems with surface coverage and concentrations of model mucins mimicking physiological conditions, we determined the equilibrium distribution of inert nanoparticles within the mucus layers using an efficient replica exchange Monte Carlo sampling procedure. The results show that the two mucus layers penetrate each other only marginally, and the bilayer imposes a strong barrier for nanoparticles, with the AML layer playing a crucial role in the mucus barrier.     

Characterization of actin filament severing by actophorin from Acanthamoeba castellanii

Actophorin is an abundant 15-kD actinbinding protein from Acanthamoeba that is thought to form a nonpolymerizable complex with actin monomers and also to reduce the viscosity of polymerized actin by severing filaments (Cooper et al., 1986. J. Biol. Chem. 261:477-485). Homologous proteins have been identified in sea urchin, chicken, and mammalian tissues. Chemical crosslinking produces a 1:1 covalent complex of actin and actophorin. Actophorin and profilin compete for crosslinking to actin monomers. The influence of actophorin on the steady-state actin polymer concentration gave a Kd of 0.2 microM for the complex of actophorin with actin monomers. Several new lines of evidence, including assays for actin filament ends by elongation rate and depolymerization rate, show that actophorin severs actin filaments both at steady state and during spontaneous polymerization. This is confirmed by direct observation in the light microscope and by showing that the effects of actophorin on the low shear viscosity of polymerized actin cannot be explained by monomer sequestration. The severing activity of actophorin is strongly inhibited by stoichiometric concentrations of phalloidin or millimolar concentrations of inorganic phosphate.     

Models of the actin monomer and filament from fluorescence resonance-energy transfer.

We have developed algorithms for combining fluorescence resonance-energy transfer (FRET) efficiency measurements into structural models which predict the relative positions of the chemical groups used in FRET. We used these algorithms to construct models of the actin monomer and filament derived solely from FRET measurements based on seven distinct loci. We found a mirror-image pair of monomer models which best fit the FRET data. One of these models agrees well with the atomic-resolution crystal structure recently published by Kabsch et al. in Heidelberg [Kabsch, W., Mannherz, H. G., Suck, D., Pai, E. F. & Holmes, K. C. (1990) Nature 347, 37-44]. The root-mean-square deviation between this FRET model and the crystal structure was about 0.9 nm. Other macromolecular models assembled from FRET measurements are likely to have a similar resolution. The largest discrepancy was for the Cys10 locus which deviated 1.44 nm from the crystal position. We discuss the limitations of the FRET method that may have contributed to this discrepancy, and conclude that the Cys10 FRET data have probably located Cys10 incorrectly in the FRET monomer model. Using the FRET monomer models, we found three orientations in the filament which best fit the intermonomer FRET data. These orientations differ substantially from the atomic-resolution filament model proposed by the Heidelberg group [Holmes, K., Popp, D., Gebhard, W. & Kabsch, W. (1990) Nature 347, 44-49], largely because of the discrepancies in the Cys10 data. These data should probably be excluded from the analysis; however, this would leave too few measurements to assemble a filament model. In the near future, we hope to obtain additional FRET measurements to other actin loci so that the filament modelling can be done without the Cys10 data.     

Model of formin-associated actin filament elongation

Formin FH2 domains associate processively with actin-filament barbed ends and modify their rate of growth. We modeled how the elongation rate depends on the concentrations of profilin and actin for four different formins. We assume that (1) FH2 domains are in rapid equilibrium among conformations that block or allow actin addition and that (2) profilin-actin is transferred rapidly to the barbed end from multiple profilin binding sites in formin FH1 domains. In agreement with previous experiments discussed below, we find an optimal profilin concentration with a maximal elongation rate that can exceed the rate of actin alone. High profilin concentrations suppress elongation, largely because free profilin displaces profilin-actin from FH1. The model supports a common polymerization mechanism for the four formin FH1FH2 constructs with differences attributed to varying parameter values. The mechanism does not require ATP hydrolysis by polymerized actin, but we cannot exclude that formins accelerate hydrolysis.     

Non-muscle actin filament elongation from complexes of profilin with nucleotide-free actin and divalent cation-free ATP-actin

Using vertebrate cytoplasmic actin consisting of a mixture of beta and gamma isoforms, we previously characterized profilin and nucleotide binding to monomeric actin (Kinosian, H. J., et al. (2000) Biochemistry 39, 13176-13188) and F-actin barbed end elongation from profilin-actin (PA) (Kinosian, H. J., et al. (2002) Biochemistry 41, 6734-6743). Our initial calculations indicated that elongation of F-actin from PA was more energetically favorable than elongation of F-actin from monomeric actin; therefore, the overall actin elongation reaction scheme described by these two linked reactions appeared to be thermodynamically unbalanced. However, we hypothesized that the profilin-induced weakening of MgATP binding by actin reduces the negative free energy change for the formation of profilin-MgATP-actin from MgATP-actin. When this was taken into account, the overall reaction scheme was calculated to be thermodynamically balanced. In our present work, we test this hypothesis by measuring actin filament barbed end elongation of nucleotide-free actin (NF-A) and nucleotide-free profilin-actin (NF-PA). We find that the free energy change for elongation of F-actin by NF-PA is equal to that for elongation of F-actin from NF-A, indicating energetic balance of the linked reactions. In the absence of actin-bound divalent cation, profilin has very little effect on ATP binding to actin; analysis of elongation experiments with divalent cation-free ATP-actin and profilin yielded an approximately energetically balanced reaction scheme. Thus, the data in this present report support our earlier hypothesis.     

Uncoupling actin filament fragmentation by cofilin from increased subunit turnover

The actin depolymerizing factor (ADF)/cofilin family of proteins interact with actin monomers and filaments in a pH-sensitive manner. When ADF/cofilin binds F-actin it induces a change in the helical twist and fragmentation; it also accelerates the dissociation of subunits from the pointed ends of filaments, thereby increasing treadmilling or depolymerization. Using site-directed mutagenesis we characterized the two actin-binding sites on human cofilin. One target site was chosen because we previously showed that the villin head piece competes with ADF for binding to F-actin. Limited sequence homology between ADF/cofilin and the part of the villin headpiece essential for actin binding suggested an actin-binding site on cofilin involving a structural loop at the opposite end of the molecule to the alpha-helix already implicated in actin binding. Binding through the alpha-helix is primarily to monomeric actin, whereas the loop region is specifically involved in filament association. We have characterized the actin binding properties of each site independently of the other. Mutation of a single lysine residue in the loop region abolishes binding to filaments, but not to monomers. Using the mutation analogous to the phosphorylated form of cofilin (S3D), we show that filament binding is inhibited at physiological ionic strength but not under low salt conditions. At low ionic strength, this mutant induces both the twist change and fragmentation characteristic of wild-type cofilin, but does not activate subunit dissociation. The results suggest a two-site binding to filaments, initiated by association through the loop site, followed by interaction with the adjacent subunit through the "helix" site at the opposite end of the molecule. Together, these interactions induce twist and fragmentation of filaments, but the twist change itself is not responsible for the enhanced rate of actin subunit release from filaments.     

The interaction of cofilin with the actin filament

Cofilin is a key actin-binding protein that is critical for controlling the assembly of actin within the cell. Here, we present the results of molecular docking and dynamics studies using a muscle actin filament and human cofilin I. Guided by extensive mutagenesis results and other biophysical and structural studies, we arrive at a model for cofilin bound to the actin filament. This predicted structure agrees very well with electron microscopy results for cofilin-decorated filaments, provides molecular insight into how the known F- and G-actin sites on cofilin interact with the filament, and also suggests new interaction sites that may play a role in cofilin binding. The resulting atomic-scale model also helps us understand the molecular function and regulation of cofilin and provides testable data for future experimental and simulation work.