2s Biophysical Journal Volume 68 April 1995 2s-7s The Actomyosin Interaction and Its Control by Tropomyosin

Kenneth C. Holmes Max Planck Institute for Medical Research, 69120 Heidelberg, Germany

An atomic model for actomyosin has been proposed the strongly bound state is created in a strained conformation (Rayment et al., 1993b) in which the and nucleotide from the weakly bound form. Because the strongly bound binding sites of S1 are linked by a cleft that may close on state has at equilibrium a different gross orientation to the binding to F-actin. The initial actomyosin contact probably actin filament (450) but is tethered by its attachment to the involves only part ofthe interface (a postulated stereospecific filament to remain in the initial 900 position, in the weak interaction); the full rigor complex develops on closure new conformational state it is strained. Release of strain en- of the cleft. These states appear similar in properties to the ergy brings about the rowing-like sweep. A and R states described by Geeves (1992). Central to this kind ofcycle is the idea that the cross-bridge By analysis of the low-angle x-ray diffraction from rabbit must first bind in an unstrained conformation (weak) and muscle at nonoverlap with and without Ca21 (Poole et al., then undergo an isomerization to the strained (strong) form. 1994) the positions of tropomyosin can be derived. Com- Hill (1974) emphasized that the work done in sliding the bining these with the atomic model of actomyosin shows that filament, often referred to as the power stroke, should not with tropomyosin in the "off' state (no Ca21) the weak ste- occur between two states but within a state so that the stored reospecific binding would be inhibited. In the presence of energy can be smoothly given up into the muscle filament Ca21 the weak stereospecific interaction would be allowed. lattice. The process is modulated by the status of the nucle- However, the development of the full rigor interface would ne- otide binding site, in particular by the presence or absence cessitate a further small movement ofthe tropomyosin to permit of the y-phosphate. This is the of the contraction pro- the cleft to shut. These three states of the thin filament may be cess. A key question that we wish to address is a possible tentatively equated with the states "blocked," "closed," and structural basis for the weak-to-strong isomerization. "open" identified by McKillop and Geeves (1993). THE ISOMERIZATION BETWEEN A AND THE CROSS-BRIDGE CYCLE R STATES A widely accepted theory to explain the mutual sliding The initial attachment is weak in the sense that the bound of actin and myosin filaments is the cross-bridge theory of cross-bridge is in rapid equilibrium with the dissociated (see H. E. Huxley, 1969; A. F. Huxley, form. However, it should be stereospecific binding; other- 1974). This theory suggests that the sliding process is driven wise it would not be possible to provide a structural basis for by the myosin cross-bridges, which extend from the myosin the ensuing isomerization to the strong binding state. The filament and cyclically interact with the actin filament by a stereospecific weak binding to actin is required to enable rowing motion of the myosin head or part of the head as ATP (catalyze) the transition to the strong binding state. This se- is hydrolyzed. The Lymn-Taylor cycle for the hydrolysis of quence provides the necessary vectoriality to the process ATP by actomyosin (Lymn and Taylor, 1971) may easily be (Jencks, 1980). Without this property the cross-bridge would incorporated into the cross-bridge theory. This resulting bind unstrained in the strong state and nothing would happen. cycle has been elaborated by Eisenberg and Green (1980) to This scheme is not a unique paradigm for achieving a strained incorporate the fact that release from actin is not an obliga- bound cross-bridge. However, the two-step binding of S1 to tory step in the hydrolysis ofATP. In their cross-bridge cycle actin is in accordance with the biochemical data (Taylor, each sweep of an oarlike cross-bridge cycle is not linked 1991; Geeves, 1992) and provides a good starting model. rigorously to a biochemical event but the myosin cross- Geeves calls the two states the A and R states. The following bridge alternates between a weak binding conformation and quote from Geeves' paper gives the essential properties of a strong binding conformation, which differ in the structural this isomerization: way in which they interact with actin. In the muscle lattice "Studies of the interaction of actin and myosin S1 in solution have shown that the association reaction takes place in at least two steps. Initially the association is relatively weak to form a complex called the A state which can isomerise into the R state. The rate and equilibrium constants for the isomerisation have been measured and are shown to depend upon the nucle- Address reprint requests to Dr. Kenneth C. Max Holmes, Planck Institute otide bound to the Sl ATPase site; with ATP bound the A state is preferred for Medical Research, Jahnstrasse 29, 69120 Heidelberg, Germany. Tel.: but as ATP is hydrolyzed and the products are sequentially released then the 49-9221-486270; Fax: 49-6221-486437; E-mail: holmes@mpimf- complex gradually shifts to the R state. An extensive series of heidelberg.mpg.de. experiments ... have shown the isomerisation ... to be closely linked with the key events C) 1995 by the Biophysical Society in the ATP-driven contraction cycle: the conformational change can be 0006-3495/95/04/02s/06 $2.00 monitored by probes on either actin or the nucleotide; the isomerisation can Holmes The Actomyosin Interaction and Its Control by Tropomyosin 3s be perturbed by an increase in hydrostatic pressure, the actin-induced ac- separates the upper and lower 50K fragments. Examination celeration of the product release from myosin is coupled to the A to R of the acto-myosin interface suggests that there are poten- isomerisation; tropomyosin may control actin and myosin interaction by three of interaction with actin: an ionic controlling the isomerisation step and finally pressure perturbations of con- tially types initially tracting muscle fibres shows there to be a close coupling between the isom- interaction involving the disordered 50-20K loop; secondly, erisation of acto-Sl and the force generating event of muscle contraction" a stereospecific interaction involving hydrophobic residues (Geeves, 1992, p. 63). that we associate with the weak A state; and thirdly, a strengthening of this interaction by the recruitment of more THE LOW IONIC STRENGTH INTERACTION MAY hydrophobic interactions allowed by the coming together of BE A PRE-WEAK STATE the 50K upper and lower fragments, which could produce the strong rigor interaction. Although the weak state we discuss here is conceptually The segment between residues 626 and 647 (50/20Kjunc- similar to the weak state discussed by Eisenberg and Green tion), which is disordered in the three-dimensional structure, (1980) it may not be the same as the weak states found at low contains five lysines and nine glycines (depending on the ionic strength (Brenner et al., 1982) since it has been shown, species). However, these lysine residues are protected from e.g., by Pollard et al. (1993) that at low ionic strength the proteolysis in the presence of actin, which suggests that while cross-bridges take up all possible orientations with respect to they are flexible in solution they are physically protected by the actin filament and appear therefore not to be stereospe- actin in the rigor complex (Mornet et al., 1979, 1981) or only cifically bound to actin. The interactions are presumably adopt a distinct conformation when bound in the actomyosin electrostatic in nature and appear to be incompetent to make interface. From their location the residues between 626 and the transition to the strong state (for discussion, see Wray 647 may interact with the negatively charged residues 1-4 on et al. (1988)). Such electrostatic weak states should be actin. It is expected that this interaction will be predomi- viewed rather as the cross-bridge making short-lived colli- nately ionic (five lysines in the loop and four carboxylic acid sion complexes with actin. Since the N-terminus of actin, groups near the N-terminus of actin) and therefore would be which is somewhat disordered, is highly negatively charged sensitive to ionic strength. The main contact of myosin with and the 50-20K myosin loop (see below), which is unstruc- actin involves the heavy chain segment from 525 to 552. This tured, is highly positively charged one could easily foresee consists of a helix that extends from 516 to 542, a loop from an interaction between these components ofactin and myosin 543 to 546, and a second helix from 547 to 558. The first leading to nonspecific complexes. helix, of which only part is in the actin binding site, contains a prominent bulge at residue 529. These two helices run at NAMING OF PARTS an angle of 10° to each other and are located at the end of the lower 50 K domain. These secondary structural elements Myosin S1 can be split by trypsin into three fragments are in close proximity to residues 341-354 and 144-146 of (Mornet et al., 1979) (25K, 50K, 20K from N- to C-termi- actin. This interface has a considerable hydrophobic com- nus), which were thought to be domains. Rayment et al. ponent. In addition residues 552 and 558 of myosin are close (1993a) show that these "domains" do not reflect structural enough to make contact with residues 40-42 in the actin divisions but rather that the trypsin-sensitive sequences lie in subunit below that, which forms the major interaction. Resi- loops that are disordered and are therefore available for pro- dues 647 to 655 of the 20K fragment are also located in the teolysis. However, the terminology (50K, etc.) is now well actin-myosin interface. established and will be used in what follows. Rayment et al. In addition to the disordered loop at the 50/20K junction, (1993a) show further that the 50K fragment is split into two an ordered loop between 405 and 415 extends toward the by a long cleft that runs from the nucleotide binding site to actin filament. If the cleft between the upper and lower 50K the actin binding site. The two parts are referred to as the fragments should close, this loop would be well placed to upper and lower 50K fragments. The upper 50K fragment, interact with actin, probably around the pair of exposed pro- which is structurally contiguous with the C-terminal part of line residues (332-333) on actin subdomain 3. The impor- the 25K fragment, contains a seven-stranded (3 sheet and tance of this loop in the function of the has been includes the nucleotide binding site. The lower 50K fragment implicated from genetic studies of familial hypertrophic car- consists of long a-helices and actually incorporates the first diomyopathy, which showed that in one case analyzed the ahelix of the 20K fragment. In the following reference will disease arises from a point mutation of residue Arg 405 to be made to the chicken sequence (Maita et al., 1991). Gln in cardiac myosin (referred to as the chicken sequence) (Geisterfer-Lowrance et al., 1990). Moreover, this mutation SUMMARY OF THE RESULTS OF has been shown to alter the rate of actin-activated ATPase RAYMENT ET AL. (1993) (Sweeney et al., 1994), which would be expected if this loop were an essential part of the rigor interaction. The actomyosin interaction is built from components from The cleft between the upper and lower 50K fragments is both the upper and lower 50K fragments on myosin as well the apparent line of communication between the actin and as the first helix of the 20K region in the sequence. These nucleotide binding sites; a closing of this cleft would clearly interactions are found on both sides of the narrow cleft that affect the nucleotide binding site. The open state would be 4s Biophysical Journal Volume 68 April 1995 favored by the presence of a y-phosphate group, whereas loop is highly variable between species, the C-terminal end model building indicates that the closed state would have dif- of the loop, which model building implicates in the actin- ficulty accommodating the -y-phosphate. On the other hand, the myosin interaction, is rather well conserved. Other evidence open state would bind weakly to actin because the actin binding for the involvement of the 626-647 loop in the formation of site is split. It is therefore rather natural to equate the open and the rigor (R) interaction is the protection from hydrolysis in closed states with the A and R states of Geeves, and it is con- rigor (Mornet et al., 1979). Moreover, the A-R transition is venient to use Geeves' nomenclature. One should not forget, pressure sensitive (Geeves, 1992) i.e., is associated with a large however, that the structural evidence for the closed state is still volume change. The formation offourto five ion pairs associated rather preliminary (see Rayment et al. (1993b)). with the folding of the 626-647 loop against the surface of actin would be a ready candidate for such a volume change. Ifformation ofthe A state led to the folding ofthe 626-647 A MECHANISM FOR CONTROLLING THE loop and this in turn led to the closing of the 50K cleft, then WEAK/STRONG TRANSITION one would have an explanation of the sequential control of The structure of myosin at the interface suggests a mecha- the A-to-R transition. nism for the transition from A-to-R state (stereospecific weak to strong). As noted, the A interface probably comprises parts of the lower 50K fragment (i.e., myosin residues 525-552 EFFECT OF TROPOMYOSIN and 547-558). This (steric) interaction brings together the By analysis of the low-angle x-ray diffraction from rabbit positively charged 626-647 flexible loop and the negatively muscle fibers at nonoverlap with and without Ca>2 (Poole charged actin N-terminus which, by charge neutralization et al., 1994) the position of tropomyosin in the "off' state (no and reduction of the loop entropy, may create a favorable Ca>2) and in the "on" (Ca2+ present) state can be derived. environment for the folding of the loop. Also the myosin These results agree rather well with the electron microscopic phe646 comes close to phe2l on actin, which may provide examination of Limulus thin filaments in the two states a hydrophobic component for the interaction. Furthermore, (Lehman et al., 1994). Furthermore, Lorenz et al. (1994) have the 627-646 loop actually straddles the cleft separating the analyzed the x-ray fiber diffraction patterns of synthetic upper and lower domains. The formation of a structure in- actin-tropomyosin filaments in the absense of to volving the flexible loop and the N-terminus of actin would show that the position taken up by tropomyosin on the sur- be expected to shorten the loop and could thereby exert ten- face of actin is very close to the position given by regulated sion across the cleft, thereby urging the cleft to close. The filaments with Ca>. Model building the actomyosin com- importance of the loop has been demonstrated by transplant- plex (Rayment et al., 1993b) into cryo-electron microscope ing the rabbit 50K-20K loop sequence into Dictyostelium reconstructions of Si-decorated thin filaments (Milligan and myosin and showing that the actin-activated ATPase rate is Flicker, 1987; Milligan et al., 1990) shows that the position determined by the sequence of the loop (J. Spudich, personal of tropomyosin in the rigor complex is similar to the position communication). Moreover, although the sequence in the given by Ca2+ on thin filaments but appears to be about 5 A

FIGURE 1 (a), (b), and (c) each show three neighboring actin monomers in a thin filament viewed at right angles to the filament axis (figure prepared with the program GRASP; A. Nicholls and B. Honig, Columbia University, New York). The middle monomer is shown as a surface representation, the outer two as worms passing through the C. atoms of each residue. Also shown are C. worms representing tropomyosin in three states: (a) "off" state, no Ca2"; (b) "on" state, Ca2+ present but no myosin interaction; (c) rigor complex. (a) and (b) are deduced from low-angle x-ray scattering studies ofthin filaments at nonoverlap (Poole et al., 1994). (c) is deduced from model building into an electron microscope reconstruction of Sl-decorated thin filaments (Milligan and Flicker, 1987; Milligan et al., 1990). Holmes The Actomyosin Interaction and Its Control by Tropomyosin 5s a b c FIGURE 2 A diagram looking down the axis of the thin filament to- Unov | U143V ward the z line showing the relative positions of tropomyosin and myosin Actin Actin Actin Si produced by combining the results T + -+ given in Fig. 1 with the atomic model T / for decorated actin (Rayment et al., 4. ,4, 9 1993b). (a) "Off' state; the tropomyo- \ \ sin inhibits the stereospecific weak p binding. (b) "On" state; note the steric hindrance between the 50K upper \lyosin S1 Myosin S1 fragment and tropomyosin. (c) Rigor Myosin S \ complex; the tropomyosin lies outside P the myosin binding site. further away from the myosin binding site. These three po- Jencks, W. P. 1980. The utilisation of binding energy in coupled vectorial sitions are shown in Fig. 1. processes. Adv. Enzymol. Relat. Areas Mol. Biol. 51:75-106. Lehman, W., R. Craig, and P. Vibert. 1994. Ca2"-Induced tropomyosin Furthermore, model building shows that in the "off' state movement in Limulus thin filaments revealed by three-dimensional re- the myosin would straddle the tropomyosin on binding to construction. Nature. 368:65-67. actin so that the cleft could not close (Fig. 2). Therefore even Lorenz, M., D. Popp, K V. Poole, G. Rosenbaum, and K C. Holmes. 1994. An if myosin did manage to find its binding site on actin in this atomic model of the unregulated thin filament obtained by x-ray diffraction situation the tropomyosin would act as a gag to prevent the from orientated actin-tropomyosin gels. Biophys. J. 68:347s. Lymn, R. W., and E. W. Taylor. 1971. Mechanism ofadenosine triphosphate cleft from closing, i.e., the A-to-R transition would be com- hydrolysis of actomyosin. Biochemistry. 10:4617-4624. pletely inhibited. In the presence of Ca2' a partial interaction Maita, T., E. Yajima, S. Nagata, T. Miyanishi, S. Nakayama, and G. would be permitted, but there would still be steric clashes Matsuda. 1991. The primary structure of heavy chain: IV with the 405-415 loop of myosin. Development of the full sequence ofthe rod, and the complete 1938-residue sequence of the heavy rigor interface would be inhibited. However, the myosin no chain. J. Biochem. (Japan). 110:75-87. McKillop, D. F. A., and M. A. Geeves. 1993. Regulation of the interaction longer straddles the tropomyosin. A further small movement between actin and myosin subfragment 1: evidence for three states of the of the tropomyosin would be necessary to allow the cleft to shut thin filament. Biophys. J. 65:693-701. and permit the full rigor interaction. Fig. 2 shows these situations Milligan, R. A., and P. F. Flicker. 1987. Structural relationships of actin, diagrammatically. The three structural states ofthe thin filament myosin, and tropomyosin revealed by cryo-electron microscopy. J. Cell. we describe correspond rather nicely with the properties ofthree Biol. 105:29-39. states ofthe thin filament and iden- Milligan, R. A., M. Whittaker, and D. Safer. 1990. Molecular structure of "blocked," "closed," "open" F-actin and location of surface sites. Nature. tified by and Geeves binding 348:217-21. McKillop (1993). Momet, D., R. Bertrand, P. Pantel, E. Audemard, and R. Kassab 1981. Structure of the acto-myosin interface. Nature (Lond.). 292:301-306. I am very grateful to Ivan Rayment for making the coordinates of myosin Mornet, D., P. Pantel, E. Audemard, and R. Kassab. 1979. The limited Si available and to Ron for the data tryptic cleavage of chymotriptic S1: an approach to the characterisation Milligan kindly providing from his of the actin site in myosin heads. Biochem. Biophys. Res. Commun. 89: reconstructions. My warmest thanks are due to my colleagues Kate Poole 925-932. and Michael Lorenz for making their results available and for their valuable criticism. Pollard, T. D., D. Bhandari, P. Maupin, D. Wachstock, A. Weeds, and H. Zol. 1993. Direct visualization by electron microscopy of the weakly- bound intermediates in the actomyosin ATPase cycle. Biophys. J. 64: 454-471. REFERENCES Poole, K. V., K. C. Holmes, I. Rayment, and M. Lorenz. 1994. Control of Brenner, B., M. Schoenberg, J. M. Chalovich, L. E. Greene, and E. Eisen- the actomyosin interaction. Biophys. J. 68:348s. berg. 1982. Evidence for cross-bridge attachment in relaxed muscle at low Rayment, I., H. M. Holden, M. Whittaker, C. B. Yohn, M. Lorenz, K. C. ionic strength. Proc. Natl. Acad. Sci. USA. 79:7288-7291. Holmes, and R. A. Milligan. 1993b. Structure of the actomyosin complex Eisenberg, E., and L. E. Green. 1980. The relation of muscle biochemistry and its implications for muscle contraction. Science. 261:58-65. to muscle physiology. Annu. Rev. Physiol. 42:293-309. Rayment, I., W. R. Rypniewski, K. Schmidt-Base, R. Smith, D. R. Tom- Geeves, M. A. 1992. The actomyosin ATPase: a two-state system. Philos. chick, M. M. Benning, D. A. Winkelmann, G. Wesenberg, and H. M. Trans R. Soc. Lond. Ser. B. Bio. Sci. 336:63-70. Holden. 1993a. The three-dimensional structure of a molecular motor, Geisterfer-Lowrance, A. A. T., S. Kass, G. Tanigawa, H.-P. Vosberg, W. myosin subfragment-1. Science. 261:50-58. McKenna, C. E. Seidman, and J. G. Seidman. 1990. A molecular basis Sweeney, H. L., A. Straceski, L. Leinwand, B. Tikunov, and L. Faust. 1994. for familial hypertrophic cardiomyopathy: a ,B-cardiac myosin heavy Heterologous expression and characterization of a cardiac myosin mutant chain missense mutation. Cell. 62:999-1006. that causes hypertrophic cardiomyopathy. J. Bio. Chem. 269:1603-1605. Hill, T. L. 1974. Theoretical formulation for the Taylor, E. W. 1991. Kinetic studies on the association and dissociation of of contraction in striated muscle. Part 1. Prog. Biophys. Mol. Biol. myosin subfragment 1 and actin. J. Biol. Chem. 266:294-302. 28:267-340. Wray, J., R. S. Goody, and K. C. Holmes 1988. Towards a molecular mecha- Huxley, A. F. 1974. Muscular contraction. J. Gen. Physiol. 243:1-43. nism for the cross bridge cycle. Adv. Exp. Med. Biol. 226:49-59. (Mo- Huxley, H. E. 1969. The mechanism of muscular contraction. Science. 164: lecular Mechanism of Muscle Contraction: Hakone Symposium 1986. H. 1356-1366. Sugi and G. H. Pollach, editors.)