Article Information

• PDF (196 kb)
• Full Text with Large Figures
• Cited by in Scopus (6)

PubMed

• Articles by T.J. Minehardt
• Articles by E. Pate

Related Articles

• Thymosin-β4 Changes the Conformation and Dynamics of Actin M...

  Thymosin-β4 Changes the Conformation and Dynamics of Actin Monomers Biophysical Journal, Volume 78, Issue 5, 1 May 2000, Pages 2516-2527 Enrique M. De La Cruz, E.Michael Ostap, Rodney A. Brundage, K.S. Reddy, H. Lee Sweeney and Daniel Safer Abstract Thymosin- (T) binds actin monomers stoichiometrically and maintains the bulk of the actin monomer pool in metazoan cells. T binding quenches the fluorescence of -iodoacetyl--(5-sulfo-1-naphthyl)ethylenediamine (AEDANS) conjugated to Cys of actin monomers. The of the actin-T complex depends on the cation and nucleotide bound to actin but is not affected by the AEDANS probe. The different stabilities are determined primarily by the rates of dissociation. At 25°C, the free energy of T binding MgATP-actin is primarily enthalpic in origin but entropic for CaATP-actin. Binding is coupled to the dissociation of bound water molecules, which is greater for CaATP-actin than MgATP-actin monomers. Proteolysis of MgATP-actin, but not CaATP-actin, at Gly on subdomain 2 is >12 times faster when T is bound. The C terminus of T contacts actin near this cleavage site, at His. By tritium exchange, T slows the exchange rate of approximately eight rapidly exchanging amide protons on actin. We conclude that T changes the conformation and structural dynamics (“breathing”) of actin monomers. The conformational change may reflect the unique ability of T to sequester actin monomers and inhibit nucleotide exchange. Abstract | Full Text | PDF (256 kb)

• Insights into the Influence of Nucleotides on Actin Family P...

  Insights into the Influence of Nucleotides on Actin Family Proteins from Seven Structures of Arp2/3 Complex Molecular Cell, Volume 26, Issue 3, 11 May 2007, Pages 449-457 Brad J. Nolen and Thomas D. Pollard 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 cocrystallized 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. Summary | Full Text | PDF (882 kb)

• The β-Thymosin/WH2 Domain

  The β-Thymosin/WH2 Domain Cell, Volume 117, Issue 5, 28 May 2004, Pages 611-623 Maud Hertzog, Carine van Heijenoort, Dominique Didry, Martin Gaudier, Jérôme Coutant, Benoı̂t Gigant, Gérard Didelot, Thomas Préat, Marcel Knossow, Eric Guittet and Marie-France Carlier Summary The widespread β-thymosin/WH2 actin binding domain has versatile regulatory properties in actin dynamics and motility. β-thymosins (isolated WH2 domain) maintain monomeric actin in a “sequestered” nonpolymerizable form. In contrast, when repeated in tandem or inserted in modular proteins, the β-thymosin/WH2 domain promotes actin assembly at filament barbed ends, like profilin. The structural basis for these opposite functions is addressed using ciboulot, a three β-thymosin repeat protein. Only the first repeat binds actin and possesses the function of ciboulot. The region that shows the strongest interaction with actin is an amphipathic N-terminal α helix, present in all β-thymosin/WH2 domains, which recognizes the ATP bound actin structure and uses the shear motion of actin linked to ATP hydrolysis to control polymerization. Crystallographic (H, N), NMR, and mutagenetic data reveal that the weaker interaction of the C-terminal region of β-thymosin/WH2 domain with actin accounts for the switch in function from inhibition to promotion of actin assembly. Summary | Full Text | PDF (1172 kb)

• …more

Copyright © 2006 The Biophysical Society. All rights reserved.
Biophysical Journal, Volume 90, Issue 7, 2445-2449, 1 April 2006

doi:10.1529/biophysj.105.072900

Muscle and Contractility

The Open Nucleotide Pocket of the Profilin/Actin X-Ray Structure Is Unstable and Closes in the Absence of Profilin

T.J. Minehardt^*, 2, P.A. Kollman^*, 1, R. Cooke^†, ‡ and E. Pate^§, ,  

^* Department of Pharmaceutical Chemistry, University of California, San Francisco, California
^† Department of Biochemistry and Biophysics, University of California, San Francisco, California
^‡ Cardiovascular Research Institute, University of California, San Francisco, California
^§ Department of Mathematics, Washington State University, Pullman, Washington

Address reprint requests to E. Pate, Dept. of Mathematics, Washington State University, Pullman, WA 99164. Tel.: 509-335-3151; Fax: 509-335-1188.

¹ This article is dedicated to deceased colleague P. A. Kollman.

² T. J. Minehardt’s present address is Lone Star Biotechnologies, Austin, TX 78736.

The cycling of actin between the monomeric G-actin species and filamentous F-actin is a dynamic process, with the equilibrium between the two species dependent upon immediate cellular needs. The transition serves as an integral part of the function of eukaryotic cells, with polymerized F-actin both providing a structural role in cells and functioning as an obligatory component of myosin-based motility 1. The initial x-ray structure of G-actin showed it to be a globular protein comprised of four contiguous subdomains (Fig. 1) 2. There is a nucleotide-binding site between the lobes of the protein formed by subdomains 1 and 2, and by subdomains 3 and 4, with the nucleotide interacting with all four subdomains. Although modulated by actin-associated proteins, ATP at the nucleotide site of G-actin favors polymerization at the barbed-end of the actin filament, whereas the posthydrolysis, ADP-bound state favors depolymerization at the pointed end (reviewed in Pollard et al. 3 and references therein). The requirement for nucleotide exchange in a polymerization-depolymerization cycle has been taken to imply that an opening and closing of the nucleotide-binding site is associated with movements of the subdomains, and that these nucleotide-dependent domain movements are involved in the G-/F-actin transition. Thus, there has been intense interest in attempting to understand the relationship between subdomain movements in actin, the opening and closing of the nucleotide site, and the bound nucleotide.

Display large version of this figure

Figure 1

Open-nucleotide-pocket x-ray structure of actin•CaATP without profilin (cyan, Protein Data Bank 1HLU) and the structure after 1000ps of MD simulation (tan) are shown. Structures have been superimposed via a least-squares distance minimization (Cα coordinates) of subdomains 3 and 4. The four subdomains of actin are labeled. ATP is shown in purple. The nucleotide-binding site has closed in the tan structure due to a 4.5-Å displacement of subdomain 1 (a significant component is out of the plane of the page) and a 12° rotation of subdomain 2, both movements relative to subdomains 3 and 4. The Cα-Cα distance across the ATP triphosphate-binding domain decreases by 2.6Å. R335 is shown in magenta. A Protein Data Bank format file (Word) of the structure of actin•CaATP with profilin removed after 1000ps of MD simulation is available at http://www.math.wsu.edu/faculty/pate/actin_1000ps.doc.

The most compelling, and frequently quoted, structural evidence for an open conformation of the actin nucleotide site has come from the x-ray crystallography structure of the open nucleotide binding site structure of the profilin:actin•CaATP complex. This open nucleotide site conformation (Protein Data Bank 1HLU 4), coupled with previous observations that profilin promotes nucleotide exchange 5,6,7,8, has led to wide postulation that this structure is the open structure of actin for nucleotide exchange (reviewed in Schuler 9 and Sablin et al. 10, and references therein). Others have challenged this assignment 11,12.

Difficulties arise with interpretation. This is due to the fact that x-ray structures of actin have only been possible when the polymerization of G-actin has been inhibited either by complexing G-actin with ancillary proteins such as DNAase, profilin, gelsolin, vitamin D-binding protein, etc. 2,4,13,14,15,16, by complexing it with drugs 17, or via covalent modification of actin by the binding of organic molecules 11,12. Ambiguities thus arise as to whether the properties of the x-ray structures of the modified proteins represent true properties of actin or are simply the result of the modifications necessary for crystallization. All current actin x-ray structures have a closed nucleotide-binding site, with the single exception of the profilin:actin•CaATP structure 4 discussed above.

Molecular dynamics (MD) simulations can provide a valuable tool for helping to sort out the ambiguities present in the x-ray database associated with protein modification. For MD simulation, the ancillary bound proteins or covalent modifications can be easily removed. The thermodynamic properties of the remaining native protein in the absence of crystal packing forces can then be examined in a modeled aqueous environment under physiological conditions. Of equal importance, parallel MD simulations with retention of the ancillary protein modification can further provide control comparisons with the x-ray crystallography results. We use this approach to demonstrate that the open nucleotide-site actin structure is unstable in the absence of profilin, and closes to a structure with a nucleotide site resembling that in the closed profilin:actin•SrATP x-ray structure 13.

Molecular dynamics simulations were performed using the Amber 7 18 suite of codes and the Cornell force field 19. The structures of profilin:actin•CaATP (Protein Data Bank 1HLU, ref. 4) and profilin:actin•SrATP (Protein Data Bank 2BTF 13) were used as the starting points for simulation. For simulations in the absence of profilin, the coordinates of profilin were manually edited out of the PDB file. For simulations of nucleotide-free actin states, ATP and the metal ion were also edited out of the file. To substitute between SrATP and CaATP at the active site, ATP was retained in the PDB file and the metal ion alone was changed. ADP was simulated at the nucleotide site by deleting the γ-phosphate moiety from ATP in the PDB file. Charges for ATP and ADP were determined by first performing a single-point energy calculation at the Hartree-Fock level of theory using a 6-31G* basis set to obtain electrostatic charges. These are then fit to the molecules using the restrained electrostatic potential procedure 20. Van der Waals parameters for Ca²⁺ and Sr²⁺ were taken from Aqvist 21. The tLeaP module of Amber 7 was used to add hydrogens to the x-ray structures for simulation. Histidines were protonated as appropriate for their local environment. Na⁺ counterions were then added to produce overall charge neutrality. A box of TIP3P water molecules 22 was used to solvate the system, with a 10-Å border between the edge of the box and the closest point on the protein surface. The total size of all simulations was >78,000 atoms.

The entire protein-water system was energy-minimized using 500 steps of the steepest descent algorithm and a subsequent 500 steps of the conjugate gradient algorithm. The minimized system was then gradually heated and maintained at 300K using the Berendsen algorithm 23 for maintenance of temperature via coupling to an external bath. The particle-mesh Ewald procedure 24,25,26 with a nonbonded cutoff of 8Å was used to handle electrostatic interactions. MD simulations employed the SHAKE algorithm 27 and a 2-fs time step. More limited simulations employing a 0.1-fs time step yielded similar results. Unless otherwise stated, all simulations were for 1000ps. Periodic boundary conditions were employed at constant pressure. Simulation results were visualized using the Midas 28 and Chimera 29 molecular graphics suites of codes.

We performed MD simulations based upon the open nucleotide site profilin:actin•CaATP x-ray structure 4 to investigate the implications of protein modification. The results of these simulations were compared with simulation of the corresponding profilin:actin•SrATP structure 13 from the same laboratory, but containing a closed nucleotide-binding site. Two simulations were initially done. In the first, the profilin:actin•CaATP structure with an open nucleotide-binding site was taken directly from the Protein Data Bank and simulated in a full box of explicit waters. The MD simulation showed virtually no change in the structure. The open nucleotide-binding site remained open. The root mean-square (RMS) displacement (Cα atoms) between the profilin:actin•CaATP x-ray structure and the stable structure obtained after 1000ps of MD simulation was <0.2Å. Thus, the MD simulation and x-ray crystallography agree that the open-nucleotide-site, profilin:actin•CaATP x-ray structure is stable under modeled physiological conditions. For the second simulation, profilin was now deleted from the x-ray structure of the profilin:actin•CaATP complex. The open nucleotide site of the remaining, isolated G-actin x-ray structure now rapidly closed down during the MD simulation. Profilin binds across both subdomains 1 and 3 of actin 4,13, making it structurally plausible that profilin could be stabilizing an otherwise globally strained conformation (see Fig. 1). Indeed, the major portion of the conformational change observed occurred within the first 200ps of the MD simulation, demonstrating that the profilin-free actin complex was indeed quite unstable. The RMS displacement of the MD-simulated structure as a function of time from the initial x-ray structure is shown in Fig. 2.

Display large version of this figure

Figure 2

Root mean-square displacement of the MD-simulated actin•ATP structure in the absence of profilin as a function of time. The nonzero intercept (●) represents the displacement that occurred during the energy minimization process after removal of profilin. A new equilibrium is reached after ∼500ps.

More important, the open nucleotide site evolved into a closed structure topologically similar to that previously described in the related x-ray structure of the profilin:actin•SrATP complex from the same lab containing a closed nucleotide site 13. Comparison of the initial open actin structure and the MD-simulated closed actin structure obtained after 1000ps of simulation showed a clamshell closing of the nucleotide site resulting from movements of subdomains 1 and 2 relative to subdomains 3 and 4 (Fig. 1). The transition from the open to the closed conformation could be modeled by changes in backbone ϕ-angles of −5° at R335 and of −10°, 12°, −16°, and −13° at A138, G146, R147, and T149, respectively, in the open-nucleotide-site structure. The important point in comparing the MD simulation results with previous crystallographic analysis is that R335 and the spanning α-helix and adjacent loop between subdomains 1 and 3 containing amino acids 138–149 (Fig. 1) are the identical structural elements that have previously been identified as being involved in the closing of the nucleotide pocket in the actin x-ray structures 4,11,13,30. Further evidence for changes in the vicinity of R335 and the spanning α-helix comes from the observations that the average magnitudes of the differences between ϕ/ψ-angles (average Δϕ and Δψ) in the open x-ray structure and the MD-simulated structure in the region P333–Y337 are 21.7°/25.2° and 11.8°/11.4° for the region A138–T149. The MD simulation thus reproduces previous crystallographic observations. The net effect is a rotation and translation of subdomains 1 and 2 relative to subdomains 3 and 4, as shown in Fig. 1. An examination of the RMS displacement of the MD-simulated, closed-nucleotide-site actin structure as a function of time demonstrated that a new equilibrium had been reached at 1000ps (Fig. 2).

The MD simulation with profilin removed yields a 2.6-Å closing of the triphosphate-binding pocket (distance measured as S14↔G158, Cα↔Cα). This is virtually identical to the 2.7-Å closing observed between the open and closed x-ray structures of the profilin:actin•ATP x-ray complexes 4,13. Fig. 3 shows in greater detail the correspondence between the nucleotide-binding domains. As is evident in Fig. 3, the hydrogen bonding patterns are virtually identical. Hydrogen atoms cannot be resolved in x-ray structures, but are present in the MD simulation, which must be taken into consideration when comparing hydrogen bond lengths. The only difference between the interactions of protein with the nucleotide is that a water molecule is trapped in the closing of the nucleotide-binding site in the MD simulation, mediating hydrogen-bonding interactions between the backbone amide groups of G15 and M16 with the β-phosphate of ATP. These are instead direct hydrogen bonds in the closed profilin:actin•SrATP x-ray structure (red hydrogen bonding patterns, Fig. 3). The water appeared at the nucleotide site during the initial solvation of the x-ray structure in a box of explicit water molecules, and remained stable for the duration of the simulation. After MD simulation, H73 also now forms a hydrogen bond across the cleft to the backbone oxygen of G158 as seen in the closed x-ray structure. The hydrophobic interaction of the adenine ring with the ethyl groups on the side chain of E214 is also preserved (detail not shown in Fig. 3). The crucial point that must be considered in the evaluation of the results of an MD simulation is the degree to which the atomic-level force field employed in the simulation accurately captures the true forces involved. The existing x-ray structure of the closed-pocket, actin•nucleotide structure 13 serves as the gold standard for this comparison. As shown above, our results indicate excellent agreement between the closed structure to which the MD simulation evolves and an actual x-ray structure.

Display large version of this figure

Figure 3

(A) Hydrogen-bonding pattern of the closed nucleotide site of profilin:actin•SrATP x-ray structure is shown. (B) The hydrogen-bonding pattern after 1000ps MD simulation of the profilin:actin•CaATP x-ray structure with profilin removed is shown after the nucleotide site had closed. Adenine is omitted from the structures of ATP to enhance detail. (C) Distances in the open profilin:actin•CaATP x-ray structure are shown for comparison. Hydrogen atoms are not resolved in the x-ray structures. They are present in the MD simulation. In A and B, the two hydrogen-bonding patterns are virtually identical. The only difference is that a water molecule is trapped during the closing of the nucleotide pocket in the MD simulation (right) and now mediates the hydrogen bonding to the β-phosphate moiety of ATP. Green dashed lines show hydrogen bonds conserved between the two closed structures (A and B). Red dashed lines show the water-mediated hydrogen bond. Color scheme: cyan, hydrogen; gray, carbon; red, oxygen; and blue, phosphorus.

An additional advantage of MD simulation is that we could perform the following control simulations to provide additional confidence in the conclusions. 1), MD simulations of the closed nucleotide site profilin:actin•SrATP structure remained closed whether profilin was present or absent. 2), MD simulations of the structure resulting from the substitution of SrATP for CaATP, or with the elimination of CaATP (apo structure) in the profilin:actin•CaATP x-ray structure, while retaining profilin, remained open. 3), MD simulation of the structure resulting from the substitution of SrATP for CaATP in the actin•CaATP structure (profilin removed for the simulation) showed a closing of the nucleotide site. 4), MD simulation of the profilin:actin•CaATP structure with profilin removed and CaATP replaced with CaADP showed a closing of the nucleotide site. 5), MD simulation of the profilin:actin•CaATP structure with both profilin and CaATP removed (open apo structure) continued to show a closing of the nucleotide site. This latter simulation implied that the domain movements we were seeing were the result of instabilities generated in the fundamental protein structure resulting from the interaction with profilin, and not simply the result of charge effects at the nucleotide site. In summary, all combinations of closed x-ray structures examined remained closed during MD simulation. All combinations of open profilin:actin•nucleotide complexes remained open. In the absence of profilin, all open actin structures closed down. These control simulations provide further confidence in the generality of our conclusions. The open structure in the absence of profilin evolves into a structure homologous to other closed actin structures. We also note that in all simulations, with and without profilin, the DNAase binding loop (D-loop) remained a loop.

The dynamic polymerization/depolymerization cycle of actin requires nucleotide hydrolysis and exchange of ADP for ATP. Thus, considerable effort has been directed at identifying nucleotide-site conformational changes and their relationship to the global conformational changes associated with polymerization. Studies using proteolysis, fluorescence, and binding as the reporter signal 31,32,33,34,35,36,37,38 have all been consistent with the presence of nucleotide-site conformational changes in actin. Recent control experimental investigations taking advantage of a known x-ray actin structure, however, have suggested that these localized reporter signals may not always signal a nucleotide cleft opening 39. Electron microscopy reconstructions of actin filaments have implied that the actin cleft is more open in the ADP-bound state compared to the ATP-bound state 10,40. However, these EM studies were done with yeast actin, which may show significant differences from vertebrate skeletal muscle actin. The presence of open nucleotide sites in the x-ray structures of proteins such as nucleotide-free and glucose-free hexokinase 41, nucleotide-free Arp2, Arp3 42, and ParM 43, all of which share the homologous “actin fold” 44, has further argued for an open state in actin 11. It is important to emphasize that our results do not preclude an open state in actin. However, they do indicate that a stable atomic level structure is not in hand, and conclusions based on the present open structure must be evaluated with this in mind.

In summary, our results support previous suggestions that profilin facilitates nucleotide exchange in actin 5,6,7,8. However, the MD simulations imply that profilin does so by stabilizing an otherwise thermodynamically unstable, open-nucleotide-pocket state of actin. The MD simulations imply that the open-nucleotide-site actin structure state is not on the in vivo nucleotide-exchange pathway.