Crystallographic studies of the chicken gizzard G-actin X DNase I complex at 5A resolution

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.

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.

The structural basis for the intrinsic disorder of the actin filament: the "lateral slipping" model

Three-dimensional (3-D) helical reconstructions computed from electron micrographs of negatively stained dispersed F-actin filaments invariably revealed two uninterrupted columns of mass forming the "backbone" of the double-helical filament. The contact between neighboring subunits along the thus defined two long-pitch helical strands was spatially conserved and of high mass density, while the intersubunit contact between them was of lower mass density and varied among reconstructions. In contrast, phalloidinstabilized F-actin filaments displayed higher and spatially more conserved mass density between the two long-pitch helical strands, suggesting that this bicyclic hepta-peptide toxin strengthens the intersubunit contact between the two strands. Consistent with this distinct intersubunit bonding pattern, the two long-pitch helical strands of unstabilized filaments were sometimes observed separated from each other over a distance of two to six subunits, suggesting that the intrastrand intersubunit contact is also physically stronger than the interstrand contact. The resolution of the filament reconstructions, extending to 2.5 nm axially and radially, enabled us to reproducibly "cut out" the F-actin subunit which measured 5.5 nm axially by 6.0 nm tangentially by 3.2 nm radially. The subunit is distinctly polar with a massive "base" pointing towards the "barbed" end of the filament, and a slender "tip" defining its "pointed" end (i.e., relative to the "arrowhead" pattern revealed after stoichiometric decoration of the filaments with myosin subfragment 1). Concavities running approximately parallel to the filament axis both on the inner and outer face of the subunit define a distinct cleft separating the subunit into two domains of similar size: an inner domain confined to radii less than or equal to 2.5-nm forms the uninterrupted backbone of the two long-pitch helical strands, and an outer domain placed at radii of 2-5-nm protrudes radially and thus predominantly contributes to the outer part of the massive base. Quantitative evaluation of successive crossover spacings along individual F-actin filaments revealed the deviations from the mean repeat to be compensatory, i.e., short crossovers frequently followed long ones and vice versa. The variable crossover spacings and diameter of the F-actin filament together with the local unraveling of the two long-pitch helical strands are explained in terms of varying amounts of compensatory "lateral slipping" of the two strands past each other roughly perpendicular to the filament axis. This intrinsic disorder of the actin filament may enable the actin moiety to play a more active role in actin-myosin-based force generation than merely act as a rigid passive cable as has hitherto been assumed.

Molecular structure of F-actin and location of surface binding sites

Comparisons of three-dimensional maps of vertebrate muscle thin filaments obtained by cryo-electron microscopy and image analysis, reveal the molecular structure of F-actin, the location of the C terminus of the monomer and the positions of the binding sites of tropomyosin, the myosin head and the N-terminal portion of the myosin A1 light chain on the filament. These data provide strong constraints for evaluating models built from the atomic structure of the monomer and the subsequent identification of molecular contacts.

Three-dimensional reconstruction of an actin bundle

We present the three-dimensional structure of an actin filament bundle from the sperm of Limulus. The bundle is a motile structure which by changing its twist, converts from a coiled to an extended form. The bundle is composed of actin plus two auxiliary proteins of molecular masses 50 and 60 kD. Fraying the bundle with potassium thiocyanate created three classes of filaments: actin, actin plus the 60-kD protein, and actin plus both the auxiliary proteins. We examined these filaments by transmission electron microscopy and scanning transmission electron microscopy (STEM). Three-dimensional reconstructions from electron micrographs allowed us to visualize the actin subunit and the 60- and 50-kD subunits bound to it. The actin subunit appears to be bilobed with dimensions 70 X 40 X 35 A. The inner lobe of the actin subunit, located at 20 A radius, is a prolate ellipsoid, 50 X 25 A; the outer actin lobe, at 30 A radius, is a 35-A-diam spheroid. Attached to the inner lobe of actin is the 60-kD protein, an oblate spheroid, 55 X 40 A, at 50 A radius. The armlike 50-kD protein, at 55 A radius, links the 60-kD protein on one of actin's twin strands to the outer lobe of the actin subunit on the opposite strand. We speculate that the 60-kD protein may be a bundling protein and that the 50-kD protein may be responsible for the change in twist of the filaments which causes extension of the bundle.

Structural relationships of actin, myosin, and tropomyosin revealed by cryo-electron microscopy

We have calculated three-dimensional maps from images of myosin subfragment-1 (S1)-decorated thin filaments and S1-decorated actin filaments preserved in frozen solution. By averaging many data sets we obtained highly reproducible maps that can be interpreted simply to provide a model for the native structure of decorated filaments. From our results we have made the following conclusions. The bulk of the actin monomer is approximately 65 X 40 X 40 A and is composed of two domains. In the filaments the monomers are strongly connected along the genetic helix with weaker connections following the long pitch helix. The long axis of the monomer lies roughly perpendicular to the filament axis. The myosin head (S1) approaches the actin filament tangentially and binds to a single actin, the major interaction being with the outermost domain of actin. In the map the longest chord of S1 is approximately 130 A. The region of S1 closest to actin is of high density, whereas the part furthest away is poorly defined and may be disordered. By comparing maps from decorated thin filaments with those from decorated actin, we demonstrate that tropomyosin is bound to the inner domain of actin just in front of the myosin binding site at a radius of approximately 40 A. A small change in the azimuthal position of tropomyosin, as has been suggested by others to occur during Ca2+-mediated regulation in vertebrate striated muscle, appears to be insufficient to eclipse totally the major site of interaction between actin and myosin.

Fluorescence resonance energy transfer measurements of distances in actin and myosin. A critical evaluation

The contractile proteins actin and myosin are of considerable biological interest. They are essential for muscle contraction and in eukaryotic cells they play a crucial role in most contractile phenomena. Over the years since the first fluorescence resonance energy transfer (FRET) paper appeared, an extensive body of literature has accumulated on this technique using actin, myosin and the actomyosin complex. These papers are reviewed with several aims in mind: we assess the reliability and consistency of intra- and inter-molecular distances measured between the fluorescent probes attached to specific sites on these proteins; we determine whether the measurements can be assembled into an internally consistent model which can be fitted to the known dimensions of the actomyosin complex; several of the FRET distances are consistent with the available structural data from crystallographic and electron microscopic dimensions; the modelled FRET distances suggest that the assumed value of the orientation factor (k2 = 2/3) is reasonable; we conclude that the model has a predictive value, i.e. it suggests that a small number of the published dimensions may be incorrect and predicts the magnitude of a larger number of measurements which have not yet been reported; and finally (vi) we discuss the contribution of FRET determinations to the current debate on the molecular mechanism of contraction.

Cryo-electron microscopy and three-dimensional reconstruction of actin filaments

Actin filaments have been examined by electron microscopy whilst in a frozen-hydrated state. Filaments embedded in a vitreous water layer are basically similar to those prepared by negative staining and show characteristic helical substructure, where the pitches of the helices are about 70 nm and 6 nm. Variability in spacing between long pitch helix cross-over points has been observed, which is consistent with intrinsic angular disorder between successive filament subunits. Fourier transforms of the most ordered filaments show four strong layer lines that index as the first, fifth, sixth and seventh orders of a 35 nm repeat. A three-dimensional helical reconstruction, calculated to a resolution of about 4 nm, shows the individual subunits to be orientated with their long axes roughly perpendicular to the filament axis. Each subunit is somewhat curved and is resolved into two domains. Most connections between successive subunits appear to be made close to the filament axis. We also report on the performance of the specimen holder (Philips PW 5699) used in this work.

Structure of myosin decorated actin filaments and natural thin filaments

Negatively stained paracrystals of reconstituted thin filaments decorated with myosin subfragment 1 (S1), at high calcium concentrations (greater than or equal to 10(-5) M), exhibit pgg plane group symmetry with component filaments having 28 subunits in 13 turns of the actin genetic helix. Isolated S1 decorated F-actin filaments trapped in a stain film were also observed to form spontaneously paracrystals with pgg plane group symmetry. We conclude that a favourable S1-S1 interaction must exist in order to stabilize these structures. Three-dimensional helical reconstructions, calculated from these paracrystals show S1 to be curved, 12 to 14 nm long and tilted with respect to the helical axis, in broad agreement with previous reconstructions calculated from isolated particles. Reconstructions of S1 and HMM decorated filaments that resolve actin show a principal myosin binding site located on the side of the actin subunit reported by Taylor & Amos [J. molec. Biol. 147, 297-324 (1981)] and a possible small interaction on the opposite side. The appearance, symmetry and helical reconstructions of isolated F-actin filaments decorated with heavy meromyosin (HMM) were similar to those of S1 decorated filaments, except at high radii where extra mass was observed. This probably arose from the connection between the two heads of HMM bound to the same long-pitch strand of actin. In contrast to most studies on thin filaments, which use reconstituted filaments, we present data on natural I-segments of muscle homogenates. Individual filaments exhibited actin helical symmetry which on reconstruction gave a two-domain motif oriented consistently with its long axis approximately perpendicular to the helical axis, but inclined towards the 5.9 nm genetic helix. Our original interpretation of these maps [Seymour & O'Brien, Nature, Lond. 283, 680-2 (1980)] depended upon reconstructions from F-actin paracrystals, which suggested actin was rather symmetrical in shape. New data from two- and three-dimensional crystal studies and reconstructions of actin-tropomyosin filaments show that actin is rather elongated and consists of two domains. These results indicate that actin contributes towards both domains of our I-segment motif and are consistent with the monomer long axis lying approximately perpendicular to the helical axis. Although tropomyosin is not resolved, comparison of the actin-tropomyosin and I-segment reconstructions suggests that tropomyosin is strongly merged with the actin domain at a lower radius from the helical axis and that the domain at higher radius arises solely from actin.

Fluorescence energy transfer between points in G-actin: the nucleotide-binding site, the metal-binding site and Cys-373 residue

Fluorescence energy transfers were studied in order to investigate the spatial relationships between the nucleotide-binding site, the metal-binding site and the Cys-373 residue in the G-actin molecule. When 1-N6-ethenoadenosine-5'-triphosphate (epsilon-ATP) in the nucleotide-binding site and Co2+ or Ni2+ in the metal-binding site were used as fluorescence donor and acceptor, respectively, the fluorescence intensity of epsilon-ATP was perfectly quenched by Ni2+ or Co2+. This indicated that the nucleotide-binding site is very close to the metal-binding site; the distance should be less than 10 A. When N-iodoacetyl-N'-(5-sulfo-1-naphthyl)ethylenediamine (IAEDANS) bound to Cys-373 residue and Co2+ in the metal-binding site were used as a fluorescence donor and an acceptor, respectively, the transfer efficiency was equal to 5 +/- 1%. The corresponding distance was calculated to be 23-32 A, assuming a random orientation factor K2 = 2/3.

F-actin is intermolecularly crosslinked by N,N'-p-phenylenedimaleimide through lysine-191 and cysteine-374

The bifunctional reagent N,N'-p-phenylenedimaleimide (PDM) is being used in an attempt to measure distances between specific side chains in adjacent monomers within F-actin. [14C]PDM was synthesized and was used to crosslink F-actin. Uncrosslinked actin was removed by gel filtration, and, from an arginine-specific tryptic digest of the covalently crosslinked dimers and higher oligomers, one radioactive crosslinked peptide was obtained in high yield. Amino acid composition and sequence analysis indicated that it comprises residues 184-196 of one monomer and 373-375 of an adjacent actin molecule, bridged by PDM through Cys-374 and Lys-191. Thus, these groups are shown to be 1.2-1.4 nm apart in adjacent actin monomers in F-actin. This information may be crucial in establishing the orientation of actin monomers within F-actin.

Electron microscopy and image processing applied to the study of protein structure and protein-protein interactions

We review the application of electron microscopy and image processing at the molecular level to an ever increasing range of biological specimens. Although recent advances have been due in part to development of more sophisticated instrumentation and/or processing algorithms, widespread application of the well-known techniques of image enhancement and structure reconstruction has depended on new strategies of in vitro crystallization and polymerization, some of which are outlined here. We also discuss the use of stoichiometric labeling and/or "cocrystallization" in identifying the different subunits in multisubunit complexes and in studying protein-protein interactions.

Tubular arrays of the actin-DNase I complex induced by gadolinium

We describe the preparation and structural analysis of ordered tubular arrays of the actin-DNase I complex. These structures consist of helically stacked rings; each ring is 73 A thick, has a 240 A outer and a 120 A inner diameter, and has 7-fold rotational symmetry. The actin-DNase I complex forms tubes under conditions in which actin alone aggregates into crystalline sheets-i.e., in the presence of the trivalent cation gadolinium. Moreover, upon addition of an equimolar amount of DNase I, crystalline actin sheets are slowly converted to tubes. The rings making the tubes contain a radial dyad axis that may be identical to the dyad axis of the unit cell of the crystalline actin sheet. Evidence is presented for this identification, which in turn allows tentative assignment of actin- and DNase I-containing regions in three-dimensional reconstructions of the rings. The structural analysis presented here may be useful in aligning available three-dimensional molecular models of actin determined from crystals of the actin-DNase I complex and from crystalline actin sheets with each other and ultimately within the biologically important actin filament.

Towards an alignment of the actin molecule within the actin filament

Electron micrographs of negatively stained actin filament paracrystals and single-layered filament rafts showing different interfilament spacings have been studied and three-dimensional reconstructions have been computed from them. Lateral ordering of the filaments in rafts was lost when interfilament spacings exceeded 8.5 nm, suggesting this distance as an upper limit for the filament diameter. Further, all reconstructions showed the same structural features at the 3 nm resolution level, except that the filaments from ordered single-layered rafts appeared 10-20% wider than those from multi-layered paracrystals. A comparison between electron microscopical and X-ray filament data, and synthetic filaments generated using different tentative molecular models and/or orientations for actin did not allow a single best model to be selected.

A consistent picture of the actin filament related to the orientation of the actin molecule

We show that freeze-dried actin filaments which have been rotary shadowed with a light coat of platinum appear very similar in morphology and width to negatively-stained filaments. The addition of a thicker coat of platinum to such preparations gives the actin filaments a different morphology and width, which are similar to those of the rotary-shadowed, quick-frozen filaments described by Heuser and Kirschner (J. Cell Biol. 1980, 86:212-234). The consistent view of the actin filament presented here, particularly its 7-8-nm width, can be interpreted in terms of the overall orientation of the actin subunit in the actin filament.