The actin/actin interactions involving the N-terminus of the DNase-I-binding loop are crucial for stabilization of the actin filament

Correlation between polymerizability and conformation in scallop beta-like actin and rabbit skeletal muscle alpha-actin

In order to investigate the structural basis for functional differences among actin isoforms, we have compared the polymerization properties and conformations of scallop adductor muscle beta-like actin and rabbit skeletal muscle alpha-actin. Polymerization of scallop Ca(2+)-actin was slower than that of skeletal muscle Ca(2+)-actin. Cleavage of the actin polypeptide chain between Gly-42 and Val-43 with Escherichia coli protease ECP 32 impaired the polymerization of scallop Mg(2+)-actin to a greater extent than skeletal muscle Mg(2+)-actin. When monomeric scallop and skeletal muscle Ca(2+)-actins were subjected to limited proteolysis with trypsin, subtilisin, or ECP 32, no differences in the conformation of actin subdomain 2 were detected. At the same time, local differences in the conformations of scallop and skeletal muscle actin subdomains 1 were revealed as intrinsic fluorescence differences. Replacement of tightly bound Ca(2+) with Mg(2+) resulted in more extensive proteolysis of segment 61-69 of scallop actin than in the case of skeletal muscle actin. Furthermore, segment 61-69 was more accessible to proteolysis with subtilisin in polymerized scallop Ca(2+)-actin than in polymerized skeletal muscle Ca(2+)-actin, indicating that, in the polymeric form, the nucleotide-containing cleft is in a more open conformation in beta-like scallop actin than in skeletal muscle alpha-actin. We suggest that this difference between scallop and skeletal muscle actins is due to a less efficient shift of scallop actin subdomain 2 to the position it has in the polymer. The possible consequences of amino acid substitutions in actin subdomain 1 in the allosteric regulation of the actin cleft, and hence in the different stabilities of polymers formed by different actins, are discussed.

Divalent cation-, nucleotide-, and polymerization-dependent changes in the conformation of subdomain 2 of actin

Conformational changes in subdomain 2 of actin were investigated using fluorescence probes dansyl cadaverine (DC) or dansyl ethylenediamine (DED) covalently attached to Gln41. Examination of changes in the fluorescence emission spectra as a function of time during Ca2+/Mg2+ and ATP/ADP exchange at the high-affinity site for divalent cation-nucleotide complex in G-actin confirmed a profound influence of the type of nucleotide but failed to detect a significant cation-dependent difference in the environment of Gln41. No significant difference between Ca- and Mg-actin was also seen in the magnitude of the fluorescence changes resulting from the polymerization of these two actin forms. Evidence is presented that earlier reported cation-dependent differences in the conformation of the loop 38-52 may be related to time-dependent changes in the conformation of subdomain 2 in DED- or DC-labeled G-actin, accelerated by substitution of Mg2+ for Ca2+ in CaATP-G-actin and, in particular, by conversion of MgATP- into MgADP-G-actin. These spontaneous changes are associated with a denaturation-driven release of the bound nucleotide that is promoted by two effects of DED or DC labeling: lowered affinity of actin for nucleotide and acceleration of ATP hydrolysis on MgATP-G-actin that converts it into a less stable MgADP form. Evidence is presented that the changes in the environment of Gln41 accompanying actin polymerization result in part from the release of Pi after the hydrolysis of ATP on the polymer. A similarity of this change to that accompanying replacement of the bound ATP with ADP in G-actin is discussed.

Intrastrand cross-linked actin between Gln-41 and Cys-374. I. Mapping of sites cross-linked in F-actin by N-(4-azido-2-nitrophenyl) putrescine

A new heterobifunctional photo-cross-linking reagent, N-(4-azido-2-nitrophenyl)-putrescine (ANP), was synthesized and covalently bound to Gln-41 of rabbit skeletal muscle actin by a bacterial transglutaminase-mediated reaction. Up to 1.0 mol of the reagent was incorporated per mole of G-actin; at least 90% of it was bound to Gln-41 while a minor fraction (about 8%) was attached to Gln-59. The labeled G-actin was polymerized, and the resulting F-actin was intermolecularly cross-linked by irradiation with UV light. The labeled and cross-linked peptides were isolated from either a complete or limited tryptic digest of cross-linked actin. In the limited digest the tryptic cleavage was restricted to arginine by succinylation of the lysyl residues. N-terminal sequencing and mass spectrometry indicated that the cross-linked peptides contained residues 40-50 (or 40-62 in the arginine limited digest) and residues 373-375, and that the actual cross-linking took place between Gln-41 and Cys-374. This latter finding was also supported by the inhibition of Cys-374 labeling with a fluorescent probe in the cross-linked actin. The dynamic length of ANP, between 11.1 and 12.5 A, constrains to that range the distance between the gamma-carboxyl group of Gln-41 in one monomer and the sulfur atom of Cys-374 in an adjacent monomer. This is consistent with the distances between these two residues on adjacent monomers of the same strand in the long-pitch helix in the structural models of F-actin [Holmes, K. C., Popp, D., Gebhard, W., and Kabsch, W. (1990) Nature 347, 44-49 and Lorenz, M., Popp, D., and Holmes, K. C. (1993) J. Mol. Biol. 234, 826-836]. The effect of cross-linking on the function of actin is described in the companion papers.

Intrastrand cross-linked actin between Gln-41 and Cys-374. III. Inhibition of motion and force generation with myosin

Structural and functional properties of intrastrand, ANP (N-(4-azido-2-nitrophenyl)-putrescine) cross-linked actin filaments, between Gln-41 and Cys-374 on adjacent monomers, were examined for several preparations of such actin. Extensively cross-linked F-actin (with 12% un-cross-linked monomers) lost at 60 degrees C the ability to activate myosin ATPase at a 100-fold slower rate and unfolded in CD melting experiments at a temperature higher by 11 degrees C than the un-cross-linked actin. Electron microscopy and image reconstruction of these filaments did not reveal any gross changes in F-actin structure but showed a change in the orientation of subdomain 2 and a decrease in interstrand connectivity. Rigor and weak (in the presence of ATP) myosin subfragment (S1) binding and acto-S1 ATPase did not show major changes upon 50% and 90% ANP cross-linking of F-actin; the Kd and Km values were little affected by the cross-linking, and the Vmax decreased by 50% for the extensively cross-linked actin. The cross-linking of actin (50%) decreased the mean speed and the number of sliding filaments in the in vitro motility assays by approximately 35% while the relative force, as measured by using external load in these assays, was inhibited by approximately 25%. The mean speed of actin filaments decreased with the increase in their cross-linking and approached 0 for the 90% cross-linked actin. Also examined were actin filaments reassembled from cross-linked and purified ANP cross-linked dimers, trimers, and oligomers. All of these filaments had the same acto-S1 ATPase and rigor S1 binding properties but different behavior in the in vitro motility assays. Filaments made of cross-linked dimers moved at approximately 50% of the speed of the un-cross-linked actin. The movement of filaments made of cross-linked trimers was inhibited more severely, and the oligomer-made filaments did not move at all. These results show the uncoupling between force generation and other events in actomyosin interactions and emphasize the role of actin filament structure and dynamics in the contractile process.

Transglutaminase-induced cross-linking between subdomain 2 of G-actin and the 636-642 lysine-rich loop of myosin subfragment 1

G-actin was covalently cross-linked with S1 in a bacterial transglutaminase-catalyzed reaction. The cross-linking sites were identified with the help of fluorescent probes and limited proteolysis as the Gln-41 on the DNase I binding loop of subdomain 2 in G-actin and a lysine-rich loop (residues 636-642) on the S1 heavy chain. The same lysine-rich loop was cross-linked to another region of G-actin in a former study (Combeau, C., D. Didry, and M-F. Carlier. 1992. J. Biol. Chem. 267:14038-14046). This indicates the existence of more than one G-actin-S1 complex. In contrast to G-actin, no cross-linking was induced between F-actin and S1 by the transglutaminase reaction. This shows that in F-actin the inner part of the DNase I binding loop, where Gln-41 is located, is not accessible for S1. The cross-linked G-actin-S1 polymerized upon addition of 2 mM MgCl2 as indicated by electron microscopy and sedimentation experiments. The filaments obtained from the polymerization of cross-linked actin and S1 were much shorter than the control actin filaments. The ATPase activity of the cross-linked S1 was not activated by actin, whereas the K+ (EDTA)-activated ATPase activity of S1 was unaffected by the cross-linking. The cross-linking between G-actin and S1 was not influenced by the exchange of the tightly bound calcium to magnesium; however, it was inhibited by the exchange of the actin-bound ATP to ADP. This finding supports the view that the structure of the DNase binding loop in ADP-G-actin is somewhere between the structures of ATP-G-actin and F-actin.

Analysis of long-range structural effects induced by DNase-I interaction with actin monomeric form or complexed to CapZ

Two fundamental properties of monomeric actin were examined in this study, ie its interaction with DNase-I, and the inhibition of endonuclease activity consecutive to the association of the two molecules. In particular, the topological independence between catalytic site of DNase-I and interface with actin, structural changes in actin monomer and the absence of conformational changes in DNase-I were described. We demonstrated a loss of flexibility of antigenic structures in actin subdomain I (ie epitopes 18-28 and 95-105) as well as modification in the exposure of Cys10 and Cys374 after DNase-I binding. Furthermore, the conformational changes induced by DNase-I into the actin molecule weakened the interaction of CapZ to its binding site located in the C-terminal region of actin monomer. These structural changes were time-dependent. When actin was cleaved in the DNase-I binding loop (sequence 38-52) at position 42 by E coli A2 strain protease, a tight DNase-I binding to split actin and the conformational changes were still observed, whereas the DNase-I inhibition activity was completely abolished. Finally, when we substitute Ca2+ by Mg2+ (ATP-Mg2+ monomeric actin) which induces a tighter conformation of actin and partially restores the inhibitory ability of split actin, long-range conformational effects of DNase-I are prevented and the ternary complex DNase-I-actin-CapZ is obtained.

Conformational changes in actin induced by its interaction with gelsolin

Actin cleaved by the protease from Escherichia coli A2 strain between Gly42 and Val43 (ECP-actin) is no longer polymerizable when it contains Ca2+ as a tightly bound cation, but polymerizes when Mg2+ is bound. We have investigated the interactions of gelsolin with this actin with regard to conformational changes in the actin molecule induced by the binding of gelsolin. ECP-(Ca)actin interacts with gelsolin in a manner similar to that in which it reacts with intact actin, and forms a stoichiometric 2:1 complex. Despite the nonpolymerizability of ECP-(Ca)actin, this complex can act as a nucleus for the polymerization of intact actin, thus indicating that upon interaction with gelsolin, ECP-(Ca)actin undergoes a conformational change that enables its interaction with another actin monomer. By gel filtration and fluorometry it was shown that the binding of at least one of the ECP-cleaved actins to gelsolin is considerably weaker than of intact actin, suggesting that conformational changes in subdomain 2 of actin monomer may directly or allosterically affect actin-gelsolin interactions. On the other hand, interaction with gelsolin changes the conformation of actin within the DNase I-binding loop, as indicated by inhibition of limited proteolysis of actin by ECP and subtilisin. Cross-linking experiments with gelsolin-nucleated actin filaments using N,N-phenylene-bismaleimide (which cross-links adjacent actin monomers between Cys374 and Lys191) reveal that gelsolin causes a significant increase in the yield of the 115-kDa cross-linking product, confirming the evidence that gelsolin stabilizes or changes the conformation of the C-terminal region of the actin molecule, and these changes are propagated from the capped end along the filament. These results allow us to conclude that nucleation of actin polymerization by gelsolin is promoted by conformational changes within subdomain 2 and at the C-terminus of the actin monomer.

Polymerization and in vitro motility properties of yeast actin: a comparison with rabbit skeletal alpha-actin

Actin purified from the yeast (Saccharomyces cerevisae) was polymerized faster than rabbit skeletal alpha-actin by MgCl2. The two actins polymerized at similar rates in the presence of CaCl2. Yeast actin, up to 25 microM, was not polymerized by KCl (100-300 mM); the monovalent salt also inhibited the MgCl2-induced polymerization of actin. The local structure of the subdomain-2 region in yeast actin filaments was probed by subtilisin and trypsin digestions. Loop 38-52 appeared more flexible and accessible to subtilisin in yeast than in rabbit actin. In contrast, tryptic digestions at Lys-61 and -68 occurred at the same rate for yeast and alpha-actin filaments. Modification of yeast actin by a sulfhydryl reagent CPM [7-(diethylamino)-3-(4'-maleimidophenyl)-4-methylcoumain] was specific to the Cys-374 residue; no labeling of a yeast actin mutant containing an alanine substitution for cysteine 374 was observed. The rates of Cys-374 labeling by CPM were similar for yeast and muscle actin, suggesting a similar environment for the C terminus in both polymers. In the in vitro motility assays, yeast actin required higher concentrations of heavy meromyosin (HMM) for its sliding than did the rabbit actin. At saturating concentrations of HMM, the sliding velocities of both actins were the same (3.0 microns/s). Relative forces generated by HMM with yeast and muscle actin were assessed by monitoring their in vitro motility in the presence of NEM-HMM load. The sliding of yeast actin was stopped at a level of external load (molar ratio NEM-HMM/HMM = 0.25) lower than that of muscle actin (NEM-HMM/HMM = 0.43), suggesting lower force production with yeast actin. These results are discussed in terms of the myosin cross-bridge cycle and actomyosin interactions.

Intermolecular coupling between loop 38-52 and the C-terminus in actin filaments

The recently reported structural connectivity in F-actin between the DNase I binding loop on actin (residues 38-52) and the C-terminus region was investigated by fluorescence and proteolytic digestion methods. The binding of copper to Cys-374 on F- but not G-actin quenched the fluorescence of dansyl ethylenediamine (DED) attached to Gin-41 by more than 50%. The blocking of copper binding to DED-actin by N-ethylmaleimide labeling of Cys-374 on actin abolished the fluorescence quenching. The quenching of DED-actin fluorescence was restored in copolymers (1:9) of N-ethylmaleimide-DED-actin with unlabeled actin. The quenching of DED-actin fluorescence by copper was also abolished in copolymers (1:4) of DED-actin and N-ethylmaleimide-actin. These results show intermolecular coupling between loop 38-52 and the C-terminus in F-actin. Consistent with this, the rate of subtilisin cleavage of actin at loop 38-52 was increased by the bound copper by more than 10-fold in F-actin but not in G-actin. Neither acto-myosin subfragment-1 (S1) ATPase activity nor the tryptic digestion of G-actin and F-actin at the Lys-61 and Lys-69 sites were affected by the bound copper. These observations suggest that copper binding to Cys-374 does not induce extensive changes in actin structure and that the perturbation of loop 38-52 environment results from changes in the intermolecular contacts in F-actin.

Purification and characterization of the proteinase ECP 32 from Escherichia coli A2 strain

The proteinase previously described as an unidentified component of E. coli A2 extracts which hydrolyses actin at a new cleavage site (Khaitlina et al. (1991) FEBS Lett. 279, 49) was isolated and further characterized. A chromatographic method of proteinase purification was developed by which a purity of more than 80% was attained. The enzyme was identified as a single, 32 kDa polypeptide (ECP 32) by SDS-PAGE and non-denaturing electrophoresis as well as by ion-exchange chromatography and gel filtration. The N-terminal sequence of ECP 32 was determined to be: AKTSSAGVVIRDIFL. The activity of ECP 32 is inhibited by o-phenanthroline, EDTA, EGTA and zincone. The EDTA-inactivated enzyme can be reactivated by cobalt, nickel and zinc ions. Based on these properties ECP 32 was classified as a metalloproteinase (EC 3.4.24). Limited proteolysis of skeletal muscle actin between Gly-42 and Val-43 was observed at enzyme substrate mass ratios of 1:25 to 1:3000. Two more sites between Ala-29 and Val-30, and between Ser-33 and Ile-34 were cleaved by ECP 32 in heat- or EDTA-inactivated actin. Besides actin, only histones and DNA-binding protein HU were found to be substrates of the proteinase, confirming its high substrate specificity. Its molecular mass, N-terminal sequence and enzymatic properties distinguish ECP 32 from any known metalloproteinases of E. coli, and we therefore conclude that it is a new enzyme.

Structural changes in subdomain 2 of G-actin observed by fluorescence spectroscopy

The influence of DNase I binding to Ca-ATP-G-actin and of Ca2+/Mg2+ and ATP/ADP exchange on the conformation of G-actin were investigated by measuring the fluorescence of dansyl cadaverine (DC) conjugated to Gln41 in subdomain 2 of the protein. Fluorescence resonance energy transfer (FRET) between this probe and N-[4-(dimethylamino)-3,5-dinitrophenyl]maleimide (DDPM) attached to Cys374 in subdomain 1 was also measured. Contrary to an earlier report [dos Remedios, Kiessling and Hambly (1994) in Synchrotron Radiation in the Biosciences (Chance, B., Deisenhofer, J., Ebashi, S., Goodhead, D. T., Helliwell, J. R., Huxley, H. E., Iizuka, T., Kirz, J., Mitsui, T., Rubenstein, E. et al., eds.), pp. 418-425, Oxford University Press, Oxford], the distance between these probes did not change significantly when DNase I was bound to actin. A small but reproducible increase in the quantum yield and a blue shift of the DC fluorescence maximum were observed when bound Ca2+ was replaced by Mg2+. A large increase (about 70%) in the quantum yield and an approx. 12 nm blue shift of the emission spectrum occurred when ATP in Mg-G-actin was replaced by ADP. These changes were not accompanied by any significant change in the FRET distance between the dansyl donor and DDPM acceptor probes. A substantial change in the fluorescence of DC-actin was observed after proteolytic removal of the last three residues of actin, in accordance with earlier evidence suggesting that there is a conformational coupling between subdomain 2 and the C-terminal segment in subdomain 1 of actin. The results are discussed in relation to recently published data obtained with another fluorescent probe and to earlier observations based on limited cleavage using proteolytic enzymes.

Conformational changes in subdomain I of actin induced by proteolytic cleavage within the DNase I-binding loop: energy transfer from tryptophan to AEDANS

Alteration of the actin polypeptide chain within the DNase I-binding loop by cleavage with E. coli A2 protease or subtilisin was shown to increase the efficiency of energy transfer from tryptophan residues to AEDANS attached to Cys-374. Analysis of structural and fluorescence data suggested that only two of four actin tryptophan residues, namely, Trp-340 and/or Trp-356, can be energy transfer donors. It was also found that labelling with AEDANS induces perturbations in the environment of the tryptophan residues, these perturbations being smaller in the cleaved actin. These changes are consistent with a shift of the C-terminal segment of actin monomer upon cleavage and confirm the existence of high conformational coupling between subdomains 1 and 2 of actin monomer. We also suggest that tryptophan residues 340 and/or 356 are located in the focus of this coupling.

Conformational changes in subdomain 2 of G-actin: fluorescence probing by dansyl ethylenediamine attached to Gln-41

Gln-41 on G-actin was specifically labeled with a fluorescent probe, dansyl ethylenediamine (DED), via transglutaminase reaction to explore the conformational changes in subdomain 2 of actin. Replacement of Ca2+ with Mg2+ and ATP with ADP on G-actin produced large changes in the emission properties of DED. These substitutions resulted in blue shifts in the wavelength of maximum emission and increases in DED fluorescence. Excitation of labeled actin at 295 nm revealed energy transfer from tryptophans to DED. Structure considerations and Cu2+ quenching experiments suggested that Trp-79 and/or Trp-86 serves as energy donors to DED. Energy transfer from these residues to DED on Gln-41 increased with the replacement of Ca2+ with Mg2+ and ATP with ADP. Polymerization of Mg-G-actin with MgCl2 resulted in much smaller changes in DED fluorescence than divalent cation substitution. This suggests that the conformation of loop 38-52 on actin is primed for the polymerization reaction by the substitution of Ca2+ with Mg2+ on G-actin.

Actin and the actomyosin interface: a review

This review deals with the structure of the actin monomer, its assembly into filaments and the loci on F-actin involved in binding myosin. Two distinctly different arrangements of monomers have been suggested for actin filaments. One model proposed by Holmes et al. is well developed. It places the so-called 'large' domain close to the filament axis and the so-called 'small' domain out near the surface of the filament. A second, less-well developed, model proposed by Schutt et al. locates the 'small' domain close to the filament axis and they rotate the monomer so that 'bottom' of the 'large' domain is at the highest radius. We analyze the available evidence for the models of F-actin derived from X-ray diffraction, reconstructions from electron micrographs, fluorescence resonance energy transfer spectroscopy, chemical cross-linking, antibody probes, limited proteolysis, site-directed and natural mutations, nuclear magnetic resonance spectroscopy and other techniques. The result is an actin-centered view of the loci on actin which are probably involved in its interaction with the myosin 'head'. From these multiple contacts we speculate on the sequence of steps between the initial weak-binding state of S-1 to the actin filament through to the stable strong-binding state seen in the absence of free Mg-ATP, i.e., the rigor state.

Deletion of amino acids from the carboxy-terminal end of actin

A series of deletions was made from the C-terminal end of actin by inserting termination codons into a full length cDNA of human alpha-skeletal muscle actin. These included deletions of 2, 3, 10, 20, 30, and 40 amino acids. The cDNA clones were transcribed and the resulting mRNAs were translated in vitro using 35S-labeled methionine. The 35S-labeled actin and actin mutants were then tested for the ability to coassemble with carrier actin, bind DNAse I, bind myosin S-1, bind a 27 kDa proteolytic fragment of alpha-actinin, and incorporate into myofibrils in vitro. Removal of the C-terminal two or three amino acids did not grossly alter the properties of actin tested. Deletion of an additional 7 amino acids (10 amino acids total) significantly decreased coassembly, binding to DNAse I, and incorporation into myofibrils, but did not dramatically reduce binding to myosin S-1 or the 27 kDa fragment of alpha-actinin. Deletion of 20 or more amino acids virtually abolished all normal actin function tested. By examining the structure of actin, we propose that the effect of removing residues 356-365 is due to the important role Trp356 plays in maintaining hydrophobic bonds between three non-contiguous segments of actin. We also suggest that removal of residues 366-372 adversely affected the structure or orientation of the DNAse I binding loop and that this change can account for defects in actin binding to DNAse I, coassembly with wild type actin, and incorporation into myofibrils.

Sequence 18-29 on actin: antibody and spectroscopic probing of conformational changes

Experimental evidence for the involvement of the 18-29 site within actin subdomain-1 in the actomyosin weak binding interface includes the inhibition of actomyosin ATPase activity by specific peptide antibodies [Adams, S., & Reisler, E. (1993) Biochemistry 32, 5051-5056] and by the Dictyostelium actin mutant D24H/D25H [Johara, M., et al. (1993) Proc. Natl. Acad. Sci. U.S.A. 90, 2127-2131]. In this work, the effect of the 18-29 peptide antibodies on the polymerization and conformation of actin has been characterized. Binding of antibody to the 18-29 site strongly inhibited the MgCl2-induced polymerization of G-actin, had a much weaker impact on the CaCl2 polymerization of actin, and showed very little effect on the NaCl polymerization of G-actin. These observations were linked to the binding of the 18-29 antibody to the different forms of actin. In sedimentation assays, the (18-29) IgG bound more strongly to Mg-F- and Mg-G-actins than to Ca-F- and Ca-G-actins, respectively. The binding of IgG to F-actin decreased sharply with an increase in ionic strength. Antibody binding to the 18-29 site induced conformational changes within the nucleotide cleft, both slowing the rate of nucleotide exchange and increasing the fluorescence intensity of actin-bound epsilon ATP. The increased fluorescence of a dansyl probe attached to Gln-41 and a pyrene probe attached to Cys-374 demonstrated that antibody binding also caused local perturbations in the DNase I loop of subdomain-2 and at the C-terminus of actin. These results are discussed in terms of actin plasticity and its implications for actomyosin interactions.