- Email: [email protected]
The three-dimensional structure of a molecular motor
REVIEWS
TIBS 19 - MARCH 1994
MYOSIN IS AN FaNZYMEthat catalyses the hydrolysis of adenosine triphosphate (ATP) and, in the presence of actin, has the ability to convert the energy of hydrolysis into directed movement During muscle contraction, this enzymatic process causes an array of thick filaments, composed mostly of myosin, to actively slide past an array of thin filaments, containing primarily actin Lz. This 'sliding filament model' for muscle contraction has been known for many years but has proved difficult to understand on a molecular level, in part because the tertiary structure of myosin was unknown. The stumbling block in determining the three-dimensional structure has been the difficulty in obtaining X-ray quality crystals of the protein. This has been all the more frustrating because myosin is an exceedingly abundant molecule: for example, in skeletal muscle it comprises nearly 50% of the total cellular protein. One reason for the difficulty in growing crystals of myosin can be ascribed to the highly asymmetrical shape of the molecule, which consists of two globular heads attached to a long tail Gig. 1). it has a molecular weight of approximately 500 kDa and consists of six polypeptide chains: two heavy chains (total molecular mass = 400kDa) and two sets of light chains, with each lightchain species having a molecular mass of approximately 20 kDa. Each globular head is composed of approximately 850 amino acid residues contributed by one of the respective hea~ chains and two of the light chains. The remaining portions of the two hea~ chains assemble to form an extended coiled coil that is nearly 1500 A in length. The globular heads contain the nucleotidebinding sites and the actin-binding regions, whereas the long rod-like portion of myosin forms the backbone of the thick filament. The myosin heads can be readily cleaved from the rest of the molecule by mild proteolysis to yield the so-called 'subfragments-r, which are very soluble. These subfragments, first identified in 1962 ~ef. 3), contain all the necessary elements to generate movement of actin during ATP hydrolysis 4. Since the myosin heads can be prepared in very large quantities 5, they seemed ideal candidates for pursuing an X-ray crystallographic analysis.
The three-dimensional structure of a molecular motor Ivan Raymentand Hazel M. Holden Myosin is one of only three proteins known to convert chemica! energy into mechanical work. Although the chemical, kinetic and physiological characteristics of this protein have been studied extensively, it has been difficult to define its molecular basis of movement. With the recent X-ray structural determination of the myosin head, however, it is now possible to put forward a hypothesis on how myosin might function as a molecular motor.
Unfortunately, they proved exceedingly difficult to crystallize by conventional crystallization techniques. After many years of unsuccessful crystallization in various laboratories, this problem was overcome by the unusual approach of mild reductive methylation of lysine residues 6. By this method, two methyl groups were successively attached to nearly all of the lysine residues contained in the myosin head through the steps shown in Box 1. Given this unique strategy for crystallizing the myosin heads, it was necessary to demonstrate that chemical modification introduced no major structural or enzymatic changes in the molecule. Consequently, an X-ray study of chemically modified hen.egg-white lysozyme was undertaken and demonstrated that, although the crystallization conditions were radically different, there were very few changes in the
three-dimensional structure of the protein 7 (Fig. 2). Furthermore, previous biochemical studies had already demonstrated that hen-egg-white lysozy~ne retained most of its enzymatic activity following reductive methylation 8. The kinetic properties of methylated myosin subfragment-1 were also investigated 9 and the results indicated that, although the kinetic parameters for the hydrolysis of ATP had changed, the protein was still enzymatically active. Taken together, these investigations demonstrated that major structural changes in the myosin head were not introduced by reductive al~lation. The thre~lmensional structure
The myosin head is highly asymmetrical .(Fig. 3). it has a length of over 165 A and is approximately 65 A wide and 40,~ deep at its thickest end. it is
~
~
Heavy chain
\ELC
R L C ~
<-- 950 A -> Rgure 1
I. Rayment and H. M. Holden a=~ at the
Department of Biochemistryand the Institute for EnzymeResearch, 1710 University Avenue, Madison, WI 53705, USA.
© 1994,ElsevierScienceLtd 0968-0004/94/$07.00
A schematic drawingof the myosin molecule (not to scale) showingthe organizationof the six polypeptidechains that form this macromolecularassembly. Eachmyosin head is composed of 840 amino acids of the heavychain and a regulatory(RLC) and essential (ELC) class of light chain. 129
REVIEWS
TIBS 19 - MARCH 1994
components necessary to generate movement. The fact that movement is possible with just the head region has The reaction involves the initial formation of an adduct or Schiff's base between an amino been recently established by expression group of the lysine sidechain and formaldehyde, which is then reduced with sodium boroof a truncated myosin in Dictyostelium hydride or a borane complex to form a secondary or tertiary amine. discoideum lacking the light-chainreduction binding regio~, which can move actin in R-NH2 + CH20 [ R-N=CH2 + H20 =' R-NH-CH 3 (1) an in vitro motility assay ~7. l-he motor segment of the myosin head exhibits a complex arrangement of secondary OH3 structural elements centered around a I reduction R-NH-CH3 + CH20 [ R-N-CH20H ~ RN-(CH3)2 (2) large, seven-stranded, mostly parallel ~sheet. This motif is formed by comThis reaction proceeds rapidly to form the dimethyl-lysine product since the pKa of the ponents from all three tryptic fragmonomethyl-lysine residue is lower than that of lysine itself. ments. The nucleotide-binding pocket is located at the end of the central clear that the secondary structure is two additional modes of interaction of ~strand of this sheet and is abutted on dominated by many long (z-helices. The this family of proteins with their target one side by the 25 kDa segment and on the other by the 50 kDa segment of the longest of these extends for 85 ,~ from peptides 14,Is. The heavy chain forms the thick heavy chain. The location of this the thick end of the molecule towards the carboxyl terminus of the heavy- portion of the head. In Fig. 3, the heavy nucleotide-binding pocket was deterchain fragment. This helix is enveloped chain is displayed in green, red and mined by the positions of amino acid by both the essential and the regulatory blue to delineate the amino-terminal, residues originally identified from affinity light chains, indicating that a nmjor role central and carboxy-terminal fragments photolabeling studies to be active-site of the light chains is the stabilization of that extend over residues Asp4-Glu204, residues ~s, and by the position of the this (z-helix. This supports earlier Gly216-Tyr626 and Gln647-Lys843, re- consensus sequence for phosphateobservations by electron microscopy, spectively. These segments are separ- binding loops previously observed in which showeO ~hat the molecule col- ated by disordered loops in the X-ray adenylate kinase and Ras ~9-21.As shown lapses when the regulatory light chain structure and were previously identified in Fig. 4, the pocket is approximately is removed 1°. In addition, these light by mild tryptic cleavage of the myosin 13 ,~ deep and 13 A wide and is empty, chains might serve to amplify the head as the 25 kDa, 50 kDa and 20 kDa except for a sulfate ion which results effects of conformational changes associ- fragments, respectively 16. While it is because the crystals were obtained ated with the binding and release of convenient to discuss the arrangement from ammonium sulfate. Since the crysnucleotide at the thick portion of the of the tertiary structural elements in tals were grown in the absence of head. The light c~ains have consider- the myosin head with respect to these nucleotide, the subfragment-1 structure able sequence and structural similarity three fragments, it is important to keep described here most probably repto both calmodulin and troponin C~-13. in mind that they are not structural resents the open conformation for the Since the structures of these light domains as originally believed. active site. chains are observed in their natural setThe thick portion of the myosin head The myosin head is also characterting, i.e. associated with the myosin contains both the nucleotide- and actin- ized by several other prominent clefts heavy chain, they serve to illustrate binding sites and thus has all of the and grooves that define structural domains ill the subfragment. The most obvious cleft splits the 50 kDa region into an upper and a lower domain. This narrow cleft runs from directly under the nucleotide-binding pocket towards the end of the molecule. These two parts of the molecule in the X-ray structure are not in Van der Waals contact with one another. It has been suggested that this cleft plays a central role in the functioning of myosin as a molecular motor, as discussed below 22. Before any model for muscle contraction can be proposed, however, the kinetic behavior of the protein and the structure of C the actomyosin complex must be considered. Box 1. Reduotivemethylotion of myosinto I~UC¢ a crystallLmbleprotein
Figure 2 Stereo comparison of the (~-carbon positions for native (red) and reductively methylated (blue) hen-egg-white lysozyme. The structure of the methylated protein was determined to 1.8 A resolution. X-raycoordinates for the unmodified protein were kindly provided by Kei)h Wilson and Brian W. Matthews (University of Oregon). This and all ribbon drawings were prepared with the software package MOLSCRIPT43. "
130
Kinetic behavior of myosin The temporal relationship between
ATP hydrolysis and the generation of movement has been established through an extensive series of kinetic
REVIEWS
TIBS 19 - MARCH 1994
studies 23-25. The two most significant features of this process are: (i) that the hydrolysis of ATP occurs when myosin is either dissociated or only weakly bound to actin 26.27, and (2) that the power stroke occurs during product release. The cnntractile cycle can be summarized in the following manner. Starting in the absence of ATP, myosin is tightly bound to actin. Upon the addition of nucleotide, myosin is rapidly released from actin and, thereafter, ATP is hydroly~ed to form a metastable myosin-products complex. Rebinding of myosin to actin catalyses the release of the y-phosphate and ADP. Studies conducted on muscle fibers have shown that force is generated during the release of the hydrolysis products 28. viiuv1,uiv
v i Lll~ l i ~ i V i i l ~ U O i l i b V i i | ~ i ; A
The actomyosin complex is a very large macromolecular assembly that has been studied extensively by electron microscopy, with the result that the tightly bound form of the complex has been well characterized 29,3°. Additionally, the th='ee-dimetlsional X-ray structure of monomeric actin has been determined, and a model for the actin filament has been proposed 3~,32.A preliminary atomic model for the actomyosin complex has been achieved by combining these two sources of information with the structure of myosin subfragment-I (Ref. 22). Even though the resolution of the image reconstruction was approximately 30A, it was possible to assemble the model with an accuracy of 5--8A because only a few positional parameters had to be determined. As a consequence, the actomyosin model provides a reasonable estimate of the relationship between myosin and actin that is presumed to represent the end of the contractile cycle, it demonstrates clearly that the actin-binding site on myosin includes components from both the upper and lower domains of the 50 kDa heavychain fragment. Furthermore, it suggests that myosin binds to actin in a highly stereospecific manner. One of the key features of this model for the actomyosin complex is a close contact between actin and the lower domain of the 50 kDa region of the structure, indicating that a conformational change might be induced in myosin when it binds to actin. This conformational change might result in closure of the long narrow cleft that divides the thick portion of the myosin head (Fig. 4), thereby allowing for corn-
Rgute 3 Ribbon representation of the myosin subfragmeoL-1 head. The green, red and blue segmerits represent the 25, 50 and 20 kDa segments of heavy chain, and the yel;ow and magenta stretches correspond to the essential and regulatory light chain~, respectiveiy.
munication between the actin- and nucleotide-binding sites, which are separated by about 35 A.
cause the movement of actin relative to myosin. The model must also conform to fundamental chemical principles of protein structure and function. These requirements can be readily met if it is assumed that the actin- and nucleotidebinding properties of myosin are mediated through a series of conformational changes between structural domains. A summary of the structural hypothesis for the contractile cycle is shown in Fig. 5. The starting point for the contractile cycle is taken as the point where
Molecular modelfor contraction The basic requirements for a molecular model of the contractile cycle are that there should be a change in the affinity o[ myosin for actin when ATP binds in the active site; that hydrolysis should only occur when myosin is dissociated or weakly bound to actin; and that rebinding of myosin to actin should catalyse the release of products and
ATP
,,,e
I
,,¢3
T
sile
58O
ACTIN
'Collision
Towa~s Z-disc
I
Colhslon
Towards Z-disc
Rgure 4 Stereo view of the interaction of myosin with actin as viewed approximatelyalong the thinfilament axis.
131
REVIEWS
TIBS 19 - MARCH 1994
molecule is thought to be in a conformationally bent state that is primed and ready for release of inorganic phosJ phate and ,~d)P. ATPpocket j [ ~ [ [ ATP binds, j[ p Rebinding of myosin to actin is also closes, cleftopens 4 ~ r Myosin hydro,ysis ~ believed to be a multistep process. The ~ i ~ dissociates occurs W'oJW initial model-building studies indicate fromactin that actin and myosin interact wRh each other in a stereospecific manner. This suggests that reformation of the tight complex between actin and myosin, after hydrolysis but before release of products, will only occur when the myosin head adopts a stereospecific orientation relative to actin. It Myosinstartsto y-phosphate, must be emphasized that myosin is rebindto actin initiationof ~ Lessof attached to the thick filament via a powerstroke fairly rigid pair of intertwined m-helices, ~1~' binding ~ binding such that its degrees of freedom relative to actin are limited. As can be seen schematically in Fig. 5d, the myosin head must rotate ~ y diffusion) into a new position one or more actin subFigure 5 units along the thin filament before it Schematic representation of the contractile cycle. This figure incorporates the various can adopt the appropriate stereostructural features of the myosin head and their proposed involvemunt in the contractile specific orientation. process. Actin is represented by spheres. In the near-axial third of the myosin head, the Once this rotation has occurred, narrow cleft that splits the 50 kDa segment of the myosin-heavy-chain sequence into two rebinding results in release of the domains is, for simplicity, represented as a vertical gap perpendicular to the filament axis. y-phosphate by closure of the narrow In the three-dimensional model, this cleft lies at an angle of ~30 ° to the filament axis such cleft Gig. 5e). In this model, the presthat the opening and closing of the cleft would not be evident from this particular orientation. The representation of the nucleotide-bound state and its associated conformational ence of the y-phosphate in the active changes relative to the X-ray structure of myosin is conceptual in nature. site of myosin and tight binding of myosin to actin are mutually exclusive. myosin is tightly bound to acthi in the Kinetic and spectroscopic measure- The loss of the y-phosphate triggers the absence of nucleotide (Fig. 5a). in this ments suggest that nucleotide binding start of the power stroke and allows the state, it i~ assumed that the narrow is a nmltistep process 3'~. In the first myosin molecule to reverse the conforcleft splitting the upper and lower stage, it is hypothesized that only the ct-, mational changes induced by the binddomains of the 50 kDa region is closed. [~- and y-phosphates, and perhaps part ing of the adenine portion of the The first atep in the contractile cycle of the ribose moiety of the nucleotide, nucleotide (Fig. 5e). As a result, the thus requires a reduction of the binding bind to subfragment-1 in tl,e 'P-loop' at active-site pocket reopens and the affinity of myosin for actin when ATP the base of the active-site pocket. It is myosin head returns to its rigor state binds. The myosin.head structure not envisaged that the initial stage of (Fig. 50, while concurrently moving the suggests that this might be ac- nucleotide binding induces any major carboxy-terminal segment of the complished by the opening of the nar- change in the curvature or overall myosin molecule relative to actin. After row cleft when the y-phosphate of ATP shape of the protein. The major shape this process, ADP is released, wherebinds, which would serve to disrupt the change is proposed to occur during the upon ATP can rapidly rebind and actin-binding surface on myosin (Fig. second stage of ATP binding, when the restart the contractile cycle. This analy5b). The potential binding site for the active-site pocket closes around the sis assumes that the conformation of y-phosphate is located close to the base of the nucleotide to produce a myosin observed in the rigor state repconfluence of the upper and lower conformationally bent myosin head resents the end of the force-producing domains of the 50 kDa region (l:ig. 4). (Fig. 5c). Thus, in the proposed model, contractile cycle. It should be noted This model also predicts that the weak the major conformational change in the that this mechanism requires the binding interaction of myosin with head occurs when the head is dis- turnover of one ATP molecule per poweractin, which was deduced from kinetic sociated from actin. This two-stage pro- stroke and is totally inconsistent with studies': and occurs even when cess is necessary to avoid tlJe pfob!,~m recent suggestions that myosin is nucleotide is bound to myosin, can be of the molecule reversing its power capable of executing many poweraccounted for by the retention of an stroke while it is tightly bound to actin strokes per hydrc'ysis event34. interaction between the flexible loop at in a stereospecific manner, which would the junction of tile 50 kDa and 20 kDa result in no net movement of myosin Evidencefor the model tryptic fragments ot myosin with actin, relative to actin. It is further proposed One of the more frustrating aspects This latter interaction would then be that hydrolysis of ATP only occurs of the study of myosin over the past nonstereospecific in nature and is pre- when the active-site pocket is in the forty years has been the difficulty in dicted to help keep myosin close to closed conformation and results in a detecting any major conformational actin during the hydrolysis step. metastable state. At this stage, the changes in the subfragment-1 head dur-
©
O
132
O"
G
REVIEWS
TIBS 19 - MARCH 1994
ing the contractile cycle. It is only recently that very careful, low-angle scattering measurements have provided evidence for a change in the shape of the molecule upon nucleotide binding3~. Earlier chemical crosslinking studies, however, clearly suggested that significant movement in the head must occur, since the distance between two reactive cysteine residues (Cys697 and CysT07) changed dramatically from 19 A to 2 A when nucleotide was bound in the active-site pocket. Furthermore, these studies demonstrated that it was possible to trap ADP in the active site by crosslinking Cys697 and Cys707 (Ref. 36). In the X-ray structure, determined in the absence of nucleotide, these reactive sulfhydryl groups are separated by an ¢x-helixand point to opposite sides of the molecule (Fig. 6). Evidence for closure of the active-site pocket was further provided by photolabeling studies in which amino acid residues, now known to lie on opposite sides of the active-site pocket and separated by about 13 A, were chemically labeled by nucleotide analogs ~s. It is thus instructive to examine the subfragment-1 model to determine if there is a pathway by which the activesite pocket can be closed through a series of inter-domain movements. It is difficult to justify rearrangement within a domain, since this would be energetically unfavorable. The active-site pocket is abutted on one side by the aminoterminal segment of the heavy chain (Fig. 3). This region provides three of the seven [~-strands of the major tertiary structural motif within the catalytic part of the myosi=~head, and also forms a small domain that constitutes one of the walls of the pocket, it is unlikely that the [~-sheet bends when the active-site pocket closes around the nucleotide. However, rotation of this small domain around the end of the sheet might be an energetically feasible movement. Additionally, this small domain is characterized by a substantial protein-protein interface with the essential light chain (Fig. 3), such that any movement of this domain would be transduced to the light-chain and hence to the rest of the light-chain-binding region. When myosin is oriented onto the actin filament, the active-site pocket faces away from the filament and opens towards the center of the sarcomere or thick filament. Closure of the active-site pocket would cause the carboxyterminal end of the heavy chain to rotate upwards relative to the actin-
i
TowardsM-line
,..~
"..t'."
..-,,,,~
,,:~
. ~.: ~',,;:.~
:..~:,..
'~'"
~'~."
I
• •
~....z.,,.o
t
°" w
: z . ".,
(a)
!
Rgure 6 (a) A ribbon and space-filling representation of the actin-myosin ;nteraction and (b) a close-up stereo view of the reactive cysteine pocket in the same orientation. The arrows indicate putative domain movements necessary to close the active site.
binding site. This is exactly the direction of movement of the tail necessary for priming the molecule for the power stroke. These movements are indicated in Fig. 6a.
There is one inherent problem with the proposed domain movement and that is due to the thickness of the myosin molecule. The base of the activesite pocket is located approximately
t33
REVIEWS half way through the thickness of the head. Thus, closure of the nucleotide pocket by a domain rotation must be associated with a rearrangement below the active site, since the protein cannot be compressed or stretched without a large one,eric penalty, it is thus noteworthy that this region below the active site contains the reactive cysteine residues discussed earlier. These residues represent a dividing line between those domains identified to bind to actin and those that might be associated with closing the active-site pocket and hence moving the carboxyl terminus of the heavy chain. Figure 6b shows that there is a deep cleft associated with one of the cysteine residues, and this is at the start of the long a-helix that constitutes the backbone of the light-chain-binding region. At this point, the long a-helix has little contact with the rest of the heavy chain. If it is true that the conformational change associated with closure of the active-site pocket is transduced to the light chains via contact with the essential light chain, then closure of this pocket would allow the helix to rotate relative to the actin-binding site without introducing any distortions within the domains themselves. These types of movement are consistent with those ~bserved in other enzymatic reactions su~.h as the active-site closure in hexokinase o,,
Magnitude of the powerstroke The crystal structure of the myosin head predicts a minimum step size or power stroke of 60 A based on simple closure of the active-site pocket. However, the physical dimensions of the protein place an upper limit on the size of the step size of approximately 200 A, assuming that the molecule operates through a 90° power stroke. These lengths are consistent with earlier estimates deduced from the tension measurements of muscle fibers after a rapid change in their length38. Likewise, an analysis of the kinetic properties of rapidly contracting muscle fibers yields an estimate of about 100A for the power stroke39. By contrast, much larger values have been suggested based on the careful study of rapid length changes in muscle fibers induced by the photolysis of caged substrates, and from irt vitro motility studies 34A0. These and other studies have given rise to the hypothesis that myosin can undergo several or many power strokes per ATP hydrolysed4L It is difficult to reconcile the multiple-powerstroke hypothesis 134
TIBS 19 - MARCH1994 973-976 3 Muller. H. and Perry, S. Vo(1962) Biochem..I. 85, 431-439 4 Toyoshima, Y. Y. et aL (1987) Nature 328, 536-539 5 Margossian S. S. and Lowey, S. (1982) Methods Enzymo/. 85, 93-116 6 Rayment, I. et al. (1993) Science 261, 50-58 7 Rypniewski, W. R., Holdeu, H. M. and Rayment, I. (1993) Biochemistry 32, 9851-9858 8 Chen, J., Dong, D. E. and Andrade,J. D. (1982) J. Colloid Interface Sci. 89, 577-580 9 White, H. W. and Rayment, I. (1993) Biochemistry 32, 9859--9865 10 Ricker, F. F., Wallimann, T. and Vibert, P. (1983) J. Mol. Biol. 169, 723-741 11 Herzberg, O. and James, M. N. G. (1985) Nature 313, 653-659 12 Babu, Y. S., Bugg, C. E. and Cook, W. J. (1988) J. Mol. Biol. 204, 191-204 13 Collins, J. H. (1991) J. Musc. Res. Cell Motil. 12, 3-25 Concludingremarks 14 Ikura, M. etal. (1992) Science 256, 632-638 There are many questions that still 15 Meador, W. E,, Means, A. R. and Quiocho, F. A. (1992) Science 257, 1251-1255 remain to be answered concerning the 16 Cooke, R. (1986) CRC Crit. Roy. Biochem. 2!, structural and enzymatic basis of 53-118 myosin-based motility. The most obvi- 17 Itakura, S. etal. Bibchem. Biophys. Res. ous are 'what type~ of molecular conCommun. (in press) formations are exhibited by the myosin 18 Yount, R. G., Cremo, C. R., Grammer, J. C. and Kerwin, B. A. (1992) Philos. Trans. R. Soc. head upon nucleotide a~d actin bindLond. Ser. B 336, 55-61 ing?' and 'how does the st~bfragment-1 19 Walker, J. E., Saraste, M., Runswick, M. J. and Gay, N. (1982) EMBO J. 1, 945-951 function within the context of a muscle fiber?' Additional questions arise from 20 Pai, E. F. et al. (1990) EMBOJ. 9, 2351-2359 the manner in which the ATPase ac- 21 Muller, C. W. and Schulz, G. E. (1992) J. MoL tivity is regulated in myosin from differBioL 224, 159-177 ent biological sources. Skeletal-muscle 22 Rayment, I. etal. (1993) Science 261, 58-65 myosin is only one member of a vast 23 Taylor,E. W. (1979) CRC Crit. Rev. Biochem. 6, 103-164 group of motor molecules that utilize 24 Hibberd, M. G. and Trentham, D. R. (1986) the myosin-heavy.chain motif 42. There Annu. Rev. Biophys. BIochern. 15, 119-161 is still much to be learned about the cellu- 25 Geeves, M. A. (1991) Biochern. J. 274, 1-14 R, W, and Taylor, E. W, (1971) lar function of these proteins, Fortu- 26 Lymn, Biochemistry 10, 4617-4624 nately, the structure of the myosin head 27 Stein, L. A., Schwartz, R. P,, Chock, P. B. and now provides a framework upon which Elsenberg, E. (1979) Biochemistry 18, 3895-3909 to dissect these problems through the M. G., Dantzlg, J. A., Trentham, combination of molecular biology, bio- 28 Hibberd, D, R. and Goldman, Y. E. (1985) Science physics and biochemistry. Indeed, this 228, 1317-1319 is an especially exciting time for the 29 Milllgan, R. A, and Flicker, P. F. (1987) J. Cell Biol. 105, 29-39 study of directed movement which is, 30 Milligan, R. A., Whitaker, M. and Safer, D. after all, one of the essential character(1990) Nature 348, 217-221 istics of living organisms. 31 Kabsch, W. et aL (1990) Nature 347, 37-44 32 Holmes, K. C., Popp, 0,, Gebhard, W. and Kabsch, W. (1990) Nature 347, 44-49 Acknowledgements 33 Taylor, E, W, (1979) CRC CriL Rev. Biochem. The study of muscle contraction has 6, 103-164 generated a vast literature over the past 34 Yan3gida,T., Harada, Y. and Ishijima, A. (1993) "l'rends Biochem. Sci. 18, 319-32a forty years and this information pro35 Wakabayashi, K. et aL (1992) Science 258, vides the foundation for the hypotheses 443-447 described in this review. Unfortunately, 36 Wells, J. A. and Yount, R. G. (1979) Proc. Natl Acad. Sci. USA 76, 4966-4970 through the confines of this short article, it is impossible to give adequate 37 Anderson, C. M., Zucker, F. H. and Steit2, T. A. (1979) Science 204, 375-380 and due credit to the n~any outstanding 38 Huxley, A. F. and Simmons, R. M. (1971) Nature researchers in the muscle field. This 233, 533-538 work was supported in part by funds 39 Pate, E., White, H. and Cooke, R. (1993) Proc. Natl Acad. Sci. USA 90, 2451-2455 from the NIH and the American Heart 40 Higuchi, H. and Goldman, Y. E. (1991) Nature Association. 352, 352-354 41 Lombardi, V., Piazzesi, G. and Linari, M. (1992) Nature 355, 638-641 References 42 Goodson, H. V. and Spudich, J. A. (1993) Proc. 1 Huxle~ A.E and hiudergerke, R.(1954)Nature Natl Acad. Sci. USA 90, 659-663 173,971-973 43 Kraulis, P. J. (1991) J. Appl. Crystallogr. 24, 2 Huxley, H. and Hanson, J.(1954)Na~re 173, 946-950
with the structure of myosin. If it is true that the source of movement is a series of distinct conformational changes induced by the sequential binding of nucleotide to myosin followed by hydrolysis and rebinding to actin, and that each of these steps leads to a decrease in free energy, then it is thermodynamically impossible for a single hydrolysis reaction to give rise to multiple power-generating steps. Resolution of this conflict between physiological measurements and the three-dimensional structure of myosin should lead to a greater insight into the biological mechanism of muscle contraction.
TIBS 19 - MARCH 1994
MYOSIN IS AN FaNZYMEthat catalyses the hydrolysis of adenosine triphosphate (ATP) and, in the presence of actin, has the ability to convert the energy of hydrolysis into directed movement During muscle contraction, this enzymatic process causes an array of thick filaments, composed mostly of myosin, to actively slide past an array of thin filaments, containing primarily actin Lz. This 'sliding filament model' for muscle contraction has been known for many years but has proved difficult to understand on a molecular level, in part because the tertiary structure of myosin was unknown. The stumbling block in determining the three-dimensional structure has been the difficulty in obtaining X-ray quality crystals of the protein. This has been all the more frustrating because myosin is an exceedingly abundant molecule: for example, in skeletal muscle it comprises nearly 50% of the total cellular protein. One reason for the difficulty in growing crystals of myosin can be ascribed to the highly asymmetrical shape of the molecule, which consists of two globular heads attached to a long tail Gig. 1). it has a molecular weight of approximately 500 kDa and consists of six polypeptide chains: two heavy chains (total molecular mass = 400kDa) and two sets of light chains, with each lightchain species having a molecular mass of approximately 20 kDa. Each globular head is composed of approximately 850 amino acid residues contributed by one of the respective hea~ chains and two of the light chains. The remaining portions of the two hea~ chains assemble to form an extended coiled coil that is nearly 1500 A in length. The globular heads contain the nucleotidebinding sites and the actin-binding regions, whereas the long rod-like portion of myosin forms the backbone of the thick filament. The myosin heads can be readily cleaved from the rest of the molecule by mild proteolysis to yield the so-called 'subfragments-r, which are very soluble. These subfragments, first identified in 1962 ~ef. 3), contain all the necessary elements to generate movement of actin during ATP hydrolysis 4. Since the myosin heads can be prepared in very large quantities 5, they seemed ideal candidates for pursuing an X-ray crystallographic analysis.
The three-dimensional structure of a molecular motor Ivan Raymentand Hazel M. Holden Myosin is one of only three proteins known to convert chemica! energy into mechanical work. Although the chemical, kinetic and physiological characteristics of this protein have been studied extensively, it has been difficult to define its molecular basis of movement. With the recent X-ray structural determination of the myosin head, however, it is now possible to put forward a hypothesis on how myosin might function as a molecular motor.
Unfortunately, they proved exceedingly difficult to crystallize by conventional crystallization techniques. After many years of unsuccessful crystallization in various laboratories, this problem was overcome by the unusual approach of mild reductive methylation of lysine residues 6. By this method, two methyl groups were successively attached to nearly all of the lysine residues contained in the myosin head through the steps shown in Box 1. Given this unique strategy for crystallizing the myosin heads, it was necessary to demonstrate that chemical modification introduced no major structural or enzymatic changes in the molecule. Consequently, an X-ray study of chemically modified hen.egg-white lysozyme was undertaken and demonstrated that, although the crystallization conditions were radically different, there were very few changes in the
three-dimensional structure of the protein 7 (Fig. 2). Furthermore, previous biochemical studies had already demonstrated that hen-egg-white lysozy~ne retained most of its enzymatic activity following reductive methylation 8. The kinetic properties of methylated myosin subfragment-1 were also investigated 9 and the results indicated that, although the kinetic parameters for the hydrolysis of ATP had changed, the protein was still enzymatically active. Taken together, these investigations demonstrated that major structural changes in the myosin head were not introduced by reductive al~lation. The thre~lmensional structure
The myosin head is highly asymmetrical .(Fig. 3). it has a length of over 165 A and is approximately 65 A wide and 40,~ deep at its thickest end. it is
~
~
Heavy chain
\ELC
R L C ~
<-- 950 A -> Rgure 1
I. Rayment and H. M. Holden a=~ at the
Department of Biochemistryand the Institute for EnzymeResearch, 1710 University Avenue, Madison, WI 53705, USA.
© 1994,ElsevierScienceLtd 0968-0004/94/$07.00
A schematic drawingof the myosin molecule (not to scale) showingthe organizationof the six polypeptidechains that form this macromolecularassembly. Eachmyosin head is composed of 840 amino acids of the heavychain and a regulatory(RLC) and essential (ELC) class of light chain. 129
REVIEWS
TIBS 19 - MARCH 1994
components necessary to generate movement. The fact that movement is possible with just the head region has The reaction involves the initial formation of an adduct or Schiff's base between an amino been recently established by expression group of the lysine sidechain and formaldehyde, which is then reduced with sodium boroof a truncated myosin in Dictyostelium hydride or a borane complex to form a secondary or tertiary amine. discoideum lacking the light-chainreduction binding regio~, which can move actin in R-NH2 + CH20 [ R-N=CH2 + H20 =' R-NH-CH 3 (1) an in vitro motility assay ~7. l-he motor segment of the myosin head exhibits a complex arrangement of secondary OH3 structural elements centered around a I reduction R-NH-CH3 + CH20 [ R-N-CH20H ~ RN-(CH3)2 (2) large, seven-stranded, mostly parallel ~sheet. This motif is formed by comThis reaction proceeds rapidly to form the dimethyl-lysine product since the pKa of the ponents from all three tryptic fragmonomethyl-lysine residue is lower than that of lysine itself. ments. The nucleotide-binding pocket is located at the end of the central clear that the secondary structure is two additional modes of interaction of ~strand of this sheet and is abutted on dominated by many long (z-helices. The this family of proteins with their target one side by the 25 kDa segment and on the other by the 50 kDa segment of the longest of these extends for 85 ,~ from peptides 14,Is. The heavy chain forms the thick heavy chain. The location of this the thick end of the molecule towards the carboxyl terminus of the heavy- portion of the head. In Fig. 3, the heavy nucleotide-binding pocket was deterchain fragment. This helix is enveloped chain is displayed in green, red and mined by the positions of amino acid by both the essential and the regulatory blue to delineate the amino-terminal, residues originally identified from affinity light chains, indicating that a nmjor role central and carboxy-terminal fragments photolabeling studies to be active-site of the light chains is the stabilization of that extend over residues Asp4-Glu204, residues ~s, and by the position of the this (z-helix. This supports earlier Gly216-Tyr626 and Gln647-Lys843, re- consensus sequence for phosphateobservations by electron microscopy, spectively. These segments are separ- binding loops previously observed in which showeO ~hat the molecule col- ated by disordered loops in the X-ray adenylate kinase and Ras ~9-21.As shown lapses when the regulatory light chain structure and were previously identified in Fig. 4, the pocket is approximately is removed 1°. In addition, these light by mild tryptic cleavage of the myosin 13 ,~ deep and 13 A wide and is empty, chains might serve to amplify the head as the 25 kDa, 50 kDa and 20 kDa except for a sulfate ion which results effects of conformational changes associ- fragments, respectively 16. While it is because the crystals were obtained ated with the binding and release of convenient to discuss the arrangement from ammonium sulfate. Since the crysnucleotide at the thick portion of the of the tertiary structural elements in tals were grown in the absence of head. The light c~ains have consider- the myosin head with respect to these nucleotide, the subfragment-1 structure able sequence and structural similarity three fragments, it is important to keep described here most probably repto both calmodulin and troponin C~-13. in mind that they are not structural resents the open conformation for the Since the structures of these light domains as originally believed. active site. chains are observed in their natural setThe thick portion of the myosin head The myosin head is also characterting, i.e. associated with the myosin contains both the nucleotide- and actin- ized by several other prominent clefts heavy chain, they serve to illustrate binding sites and thus has all of the and grooves that define structural domains ill the subfragment. The most obvious cleft splits the 50 kDa region into an upper and a lower domain. This narrow cleft runs from directly under the nucleotide-binding pocket towards the end of the molecule. These two parts of the molecule in the X-ray structure are not in Van der Waals contact with one another. It has been suggested that this cleft plays a central role in the functioning of myosin as a molecular motor, as discussed below 22. Before any model for muscle contraction can be proposed, however, the kinetic behavior of the protein and the structure of C the actomyosin complex must be considered. Box 1. Reduotivemethylotion of myosinto I~UC¢ a crystallLmbleprotein
Figure 2 Stereo comparison of the (~-carbon positions for native (red) and reductively methylated (blue) hen-egg-white lysozyme. The structure of the methylated protein was determined to 1.8 A resolution. X-raycoordinates for the unmodified protein were kindly provided by Kei)h Wilson and Brian W. Matthews (University of Oregon). This and all ribbon drawings were prepared with the software package MOLSCRIPT43. "
130
Kinetic behavior of myosin The temporal relationship between
ATP hydrolysis and the generation of movement has been established through an extensive series of kinetic
REVIEWS
TIBS 19 - MARCH 1994
studies 23-25. The two most significant features of this process are: (i) that the hydrolysis of ATP occurs when myosin is either dissociated or only weakly bound to actin 26.27, and (2) that the power stroke occurs during product release. The cnntractile cycle can be summarized in the following manner. Starting in the absence of ATP, myosin is tightly bound to actin. Upon the addition of nucleotide, myosin is rapidly released from actin and, thereafter, ATP is hydroly~ed to form a metastable myosin-products complex. Rebinding of myosin to actin catalyses the release of the y-phosphate and ADP. Studies conducted on muscle fibers have shown that force is generated during the release of the hydrolysis products 28. viiuv1,uiv
v i Lll~ l i ~ i V i i l ~ U O i l i b V i i | ~ i ; A
The actomyosin complex is a very large macromolecular assembly that has been studied extensively by electron microscopy, with the result that the tightly bound form of the complex has been well characterized 29,3°. Additionally, the th='ee-dimetlsional X-ray structure of monomeric actin has been determined, and a model for the actin filament has been proposed 3~,32.A preliminary atomic model for the actomyosin complex has been achieved by combining these two sources of information with the structure of myosin subfragment-I (Ref. 22). Even though the resolution of the image reconstruction was approximately 30A, it was possible to assemble the model with an accuracy of 5--8A because only a few positional parameters had to be determined. As a consequence, the actomyosin model provides a reasonable estimate of the relationship between myosin and actin that is presumed to represent the end of the contractile cycle, it demonstrates clearly that the actin-binding site on myosin includes components from both the upper and lower domains of the 50 kDa heavychain fragment. Furthermore, it suggests that myosin binds to actin in a highly stereospecific manner. One of the key features of this model for the actomyosin complex is a close contact between actin and the lower domain of the 50 kDa region of the structure, indicating that a conformational change might be induced in myosin when it binds to actin. This conformational change might result in closure of the long narrow cleft that divides the thick portion of the myosin head (Fig. 4), thereby allowing for corn-
Rgute 3 Ribbon representation of the myosin subfragmeoL-1 head. The green, red and blue segmerits represent the 25, 50 and 20 kDa segments of heavy chain, and the yel;ow and magenta stretches correspond to the essential and regulatory light chain~, respectiveiy.
munication between the actin- and nucleotide-binding sites, which are separated by about 35 A.
cause the movement of actin relative to myosin. The model must also conform to fundamental chemical principles of protein structure and function. These requirements can be readily met if it is assumed that the actin- and nucleotidebinding properties of myosin are mediated through a series of conformational changes between structural domains. A summary of the structural hypothesis for the contractile cycle is shown in Fig. 5. The starting point for the contractile cycle is taken as the point where
Molecular modelfor contraction The basic requirements for a molecular model of the contractile cycle are that there should be a change in the affinity o[ myosin for actin when ATP binds in the active site; that hydrolysis should only occur when myosin is dissociated or weakly bound to actin; and that rebinding of myosin to actin should catalyse the release of products and
ATP
,,,e
I
,,¢3
T
sile
58O
ACTIN
'Collision
Towa~s Z-disc
I
Colhslon
Towards Z-disc
Rgure 4 Stereo view of the interaction of myosin with actin as viewed approximatelyalong the thinfilament axis.
131
REVIEWS
TIBS 19 - MARCH 1994
molecule is thought to be in a conformationally bent state that is primed and ready for release of inorganic phosJ phate and ,~d)P. ATPpocket j [ ~ [ [ ATP binds, j[ p Rebinding of myosin to actin is also closes, cleftopens 4 ~ r Myosin hydro,ysis ~ believed to be a multistep process. The ~ i ~ dissociates occurs W'oJW initial model-building studies indicate fromactin that actin and myosin interact wRh each other in a stereospecific manner. This suggests that reformation of the tight complex between actin and myosin, after hydrolysis but before release of products, will only occur when the myosin head adopts a stereospecific orientation relative to actin. It Myosinstartsto y-phosphate, must be emphasized that myosin is rebindto actin initiationof ~ Lessof attached to the thick filament via a powerstroke fairly rigid pair of intertwined m-helices, ~1~' binding ~ binding such that its degrees of freedom relative to actin are limited. As can be seen schematically in Fig. 5d, the myosin head must rotate ~ y diffusion) into a new position one or more actin subFigure 5 units along the thin filament before it Schematic representation of the contractile cycle. This figure incorporates the various can adopt the appropriate stereostructural features of the myosin head and their proposed involvemunt in the contractile specific orientation. process. Actin is represented by spheres. In the near-axial third of the myosin head, the Once this rotation has occurred, narrow cleft that splits the 50 kDa segment of the myosin-heavy-chain sequence into two rebinding results in release of the domains is, for simplicity, represented as a vertical gap perpendicular to the filament axis. y-phosphate by closure of the narrow In the three-dimensional model, this cleft lies at an angle of ~30 ° to the filament axis such cleft Gig. 5e). In this model, the presthat the opening and closing of the cleft would not be evident from this particular orientation. The representation of the nucleotide-bound state and its associated conformational ence of the y-phosphate in the active changes relative to the X-ray structure of myosin is conceptual in nature. site of myosin and tight binding of myosin to actin are mutually exclusive. myosin is tightly bound to acthi in the Kinetic and spectroscopic measure- The loss of the y-phosphate triggers the absence of nucleotide (Fig. 5a). in this ments suggest that nucleotide binding start of the power stroke and allows the state, it i~ assumed that the narrow is a nmltistep process 3'~. In the first myosin molecule to reverse the conforcleft splitting the upper and lower stage, it is hypothesized that only the ct-, mational changes induced by the binddomains of the 50 kDa region is closed. [~- and y-phosphates, and perhaps part ing of the adenine portion of the The first atep in the contractile cycle of the ribose moiety of the nucleotide, nucleotide (Fig. 5e). As a result, the thus requires a reduction of the binding bind to subfragment-1 in tl,e 'P-loop' at active-site pocket reopens and the affinity of myosin for actin when ATP the base of the active-site pocket. It is myosin head returns to its rigor state binds. The myosin.head structure not envisaged that the initial stage of (Fig. 50, while concurrently moving the suggests that this might be ac- nucleotide binding induces any major carboxy-terminal segment of the complished by the opening of the nar- change in the curvature or overall myosin molecule relative to actin. After row cleft when the y-phosphate of ATP shape of the protein. The major shape this process, ADP is released, wherebinds, which would serve to disrupt the change is proposed to occur during the upon ATP can rapidly rebind and actin-binding surface on myosin (Fig. second stage of ATP binding, when the restart the contractile cycle. This analy5b). The potential binding site for the active-site pocket closes around the sis assumes that the conformation of y-phosphate is located close to the base of the nucleotide to produce a myosin observed in the rigor state repconfluence of the upper and lower conformationally bent myosin head resents the end of the force-producing domains of the 50 kDa region (l:ig. 4). (Fig. 5c). Thus, in the proposed model, contractile cycle. It should be noted This model also predicts that the weak the major conformational change in the that this mechanism requires the binding interaction of myosin with head occurs when the head is dis- turnover of one ATP molecule per poweractin, which was deduced from kinetic sociated from actin. This two-stage pro- stroke and is totally inconsistent with studies': and occurs even when cess is necessary to avoid tlJe pfob!,~m recent suggestions that myosin is nucleotide is bound to myosin, can be of the molecule reversing its power capable of executing many poweraccounted for by the retention of an stroke while it is tightly bound to actin strokes per hydrc'ysis event34. interaction between the flexible loop at in a stereospecific manner, which would the junction of tile 50 kDa and 20 kDa result in no net movement of myosin Evidencefor the model tryptic fragments ot myosin with actin, relative to actin. It is further proposed One of the more frustrating aspects This latter interaction would then be that hydrolysis of ATP only occurs of the study of myosin over the past nonstereospecific in nature and is pre- when the active-site pocket is in the forty years has been the difficulty in dicted to help keep myosin close to closed conformation and results in a detecting any major conformational actin during the hydrolysis step. metastable state. At this stage, the changes in the subfragment-1 head dur-
©
O
132
O"
G
REVIEWS
TIBS 19 - MARCH 1994
ing the contractile cycle. It is only recently that very careful, low-angle scattering measurements have provided evidence for a change in the shape of the molecule upon nucleotide binding3~. Earlier chemical crosslinking studies, however, clearly suggested that significant movement in the head must occur, since the distance between two reactive cysteine residues (Cys697 and CysT07) changed dramatically from 19 A to 2 A when nucleotide was bound in the active-site pocket. Furthermore, these studies demonstrated that it was possible to trap ADP in the active site by crosslinking Cys697 and Cys707 (Ref. 36). In the X-ray structure, determined in the absence of nucleotide, these reactive sulfhydryl groups are separated by an ¢x-helixand point to opposite sides of the molecule (Fig. 6). Evidence for closure of the active-site pocket was further provided by photolabeling studies in which amino acid residues, now known to lie on opposite sides of the active-site pocket and separated by about 13 A, were chemically labeled by nucleotide analogs ~s. It is thus instructive to examine the subfragment-1 model to determine if there is a pathway by which the activesite pocket can be closed through a series of inter-domain movements. It is difficult to justify rearrangement within a domain, since this would be energetically unfavorable. The active-site pocket is abutted on one side by the aminoterminal segment of the heavy chain (Fig. 3). This region provides three of the seven [~-strands of the major tertiary structural motif within the catalytic part of the myosi=~head, and also forms a small domain that constitutes one of the walls of the pocket, it is unlikely that the [~-sheet bends when the active-site pocket closes around the nucleotide. However, rotation of this small domain around the end of the sheet might be an energetically feasible movement. Additionally, this small domain is characterized by a substantial protein-protein interface with the essential light chain (Fig. 3), such that any movement of this domain would be transduced to the light-chain and hence to the rest of the light-chain-binding region. When myosin is oriented onto the actin filament, the active-site pocket faces away from the filament and opens towards the center of the sarcomere or thick filament. Closure of the active-site pocket would cause the carboxyterminal end of the heavy chain to rotate upwards relative to the actin-
i
TowardsM-line
,..~
"..t'."
..-,,,,~
,,:~
. ~.: ~',,;:.~
:..~:,..
'~'"
~'~."
I
• •
~....z.,,.o
t
°" w
: z . ".,
(a)
!
Rgure 6 (a) A ribbon and space-filling representation of the actin-myosin ;nteraction and (b) a close-up stereo view of the reactive cysteine pocket in the same orientation. The arrows indicate putative domain movements necessary to close the active site.
binding site. This is exactly the direction of movement of the tail necessary for priming the molecule for the power stroke. These movements are indicated in Fig. 6a.
There is one inherent problem with the proposed domain movement and that is due to the thickness of the myosin molecule. The base of the activesite pocket is located approximately
t33
REVIEWS half way through the thickness of the head. Thus, closure of the nucleotide pocket by a domain rotation must be associated with a rearrangement below the active site, since the protein cannot be compressed or stretched without a large one,eric penalty, it is thus noteworthy that this region below the active site contains the reactive cysteine residues discussed earlier. These residues represent a dividing line between those domains identified to bind to actin and those that might be associated with closing the active-site pocket and hence moving the carboxyl terminus of the heavy chain. Figure 6b shows that there is a deep cleft associated with one of the cysteine residues, and this is at the start of the long a-helix that constitutes the backbone of the light-chain-binding region. At this point, the long a-helix has little contact with the rest of the heavy chain. If it is true that the conformational change associated with closure of the active-site pocket is transduced to the light chains via contact with the essential light chain, then closure of this pocket would allow the helix to rotate relative to the actin-binding site without introducing any distortions within the domains themselves. These types of movement are consistent with those ~bserved in other enzymatic reactions su~.h as the active-site closure in hexokinase o,,
Magnitude of the powerstroke The crystal structure of the myosin head predicts a minimum step size or power stroke of 60 A based on simple closure of the active-site pocket. However, the physical dimensions of the protein place an upper limit on the size of the step size of approximately 200 A, assuming that the molecule operates through a 90° power stroke. These lengths are consistent with earlier estimates deduced from the tension measurements of muscle fibers after a rapid change in their length38. Likewise, an analysis of the kinetic properties of rapidly contracting muscle fibers yields an estimate of about 100A for the power stroke39. By contrast, much larger values have been suggested based on the careful study of rapid length changes in muscle fibers induced by the photolysis of caged substrates, and from irt vitro motility studies 34A0. These and other studies have given rise to the hypothesis that myosin can undergo several or many power strokes per ATP hydrolysed4L It is difficult to reconcile the multiple-powerstroke hypothesis 134
TIBS 19 - MARCH1994 973-976 3 Muller. H. and Perry, S. Vo(1962) Biochem..I. 85, 431-439 4 Toyoshima, Y. Y. et aL (1987) Nature 328, 536-539 5 Margossian S. S. and Lowey, S. (1982) Methods Enzymo/. 85, 93-116 6 Rayment, I. et al. (1993) Science 261, 50-58 7 Rypniewski, W. R., Holdeu, H. M. and Rayment, I. (1993) Biochemistry 32, 9851-9858 8 Chen, J., Dong, D. E. and Andrade,J. D. (1982) J. Colloid Interface Sci. 89, 577-580 9 White, H. W. and Rayment, I. (1993) Biochemistry 32, 9859--9865 10 Ricker, F. F., Wallimann, T. and Vibert, P. (1983) J. Mol. Biol. 169, 723-741 11 Herzberg, O. and James, M. N. G. (1985) Nature 313, 653-659 12 Babu, Y. S., Bugg, C. E. and Cook, W. J. (1988) J. Mol. Biol. 204, 191-204 13 Collins, J. H. (1991) J. Musc. Res. Cell Motil. 12, 3-25 Concludingremarks 14 Ikura, M. etal. (1992) Science 256, 632-638 There are many questions that still 15 Meador, W. E,, Means, A. R. and Quiocho, F. A. (1992) Science 257, 1251-1255 remain to be answered concerning the 16 Cooke, R. (1986) CRC Crit. Roy. Biochem. 2!, structural and enzymatic basis of 53-118 myosin-based motility. The most obvi- 17 Itakura, S. etal. Bibchem. Biophys. Res. ous are 'what type~ of molecular conCommun. (in press) formations are exhibited by the myosin 18 Yount, R. G., Cremo, C. R., Grammer, J. C. and Kerwin, B. A. (1992) Philos. Trans. R. Soc. head upon nucleotide a~d actin bindLond. Ser. B 336, 55-61 ing?' and 'how does the st~bfragment-1 19 Walker, J. E., Saraste, M., Runswick, M. J. and Gay, N. (1982) EMBO J. 1, 945-951 function within the context of a muscle fiber?' Additional questions arise from 20 Pai, E. F. et al. (1990) EMBOJ. 9, 2351-2359 the manner in which the ATPase ac- 21 Muller, C. W. and Schulz, G. E. (1992) J. MoL tivity is regulated in myosin from differBioL 224, 159-177 ent biological sources. Skeletal-muscle 22 Rayment, I. etal. (1993) Science 261, 58-65 myosin is only one member of a vast 23 Taylor,E. W. (1979) CRC Crit. Rev. Biochem. 6, 103-164 group of motor molecules that utilize 24 Hibberd, M. G. and Trentham, D. R. (1986) the myosin-heavy.chain motif 42. There Annu. Rev. Biophys. BIochern. 15, 119-161 is still much to be learned about the cellu- 25 Geeves, M. A. (1991) Biochern. J. 274, 1-14 R, W, and Taylor, E. W, (1971) lar function of these proteins, Fortu- 26 Lymn, Biochemistry 10, 4617-4624 nately, the structure of the myosin head 27 Stein, L. A., Schwartz, R. P,, Chock, P. B. and now provides a framework upon which Elsenberg, E. (1979) Biochemistry 18, 3895-3909 to dissect these problems through the M. G., Dantzlg, J. A., Trentham, combination of molecular biology, bio- 28 Hibberd, D, R. and Goldman, Y. E. (1985) Science physics and biochemistry. Indeed, this 228, 1317-1319 is an especially exciting time for the 29 Milllgan, R. A, and Flicker, P. F. (1987) J. Cell Biol. 105, 29-39 study of directed movement which is, 30 Milligan, R. A., Whitaker, M. and Safer, D. after all, one of the essential character(1990) Nature 348, 217-221 istics of living organisms. 31 Kabsch, W. et aL (1990) Nature 347, 37-44 32 Holmes, K. C., Popp, 0,, Gebhard, W. and Kabsch, W. (1990) Nature 347, 44-49 Acknowledgements 33 Taylor, E, W, (1979) CRC CriL Rev. Biochem. The study of muscle contraction has 6, 103-164 generated a vast literature over the past 34 Yan3gida,T., Harada, Y. and Ishijima, A. (1993) "l'rends Biochem. Sci. 18, 319-32a forty years and this information pro35 Wakabayashi, K. et aL (1992) Science 258, vides the foundation for the hypotheses 443-447 described in this review. Unfortunately, 36 Wells, J. A. and Yount, R. G. (1979) Proc. Natl Acad. Sci. USA 76, 4966-4970 through the confines of this short article, it is impossible to give adequate 37 Anderson, C. M., Zucker, F. H. and Steit2, T. A. (1979) Science 204, 375-380 and due credit to the n~any outstanding 38 Huxley, A. F. and Simmons, R. M. (1971) Nature researchers in the muscle field. This 233, 533-538 work was supported in part by funds 39 Pate, E., White, H. and Cooke, R. (1993) Proc. Natl Acad. Sci. USA 90, 2451-2455 from the NIH and the American Heart 40 Higuchi, H. and Goldman, Y. E. (1991) Nature Association. 352, 352-354 41 Lombardi, V., Piazzesi, G. and Linari, M. (1992) Nature 355, 638-641 References 42 Goodson, H. V. and Spudich, J. A. (1993) Proc. 1 Huxle~ A.E and hiudergerke, R.(1954)Nature Natl Acad. Sci. USA 90, 659-663 173,971-973 43 Kraulis, P. J. (1991) J. Appl. Crystallogr. 24, 2 Huxley, H. and Hanson, J.(1954)Na~re 173, 946-950
with the structure of myosin. If it is true that the source of movement is a series of distinct conformational changes induced by the sequential binding of nucleotide to myosin followed by hydrolysis and rebinding to actin, and that each of these steps leads to a decrease in free energy, then it is thermodynamically impossible for a single hydrolysis reaction to give rise to multiple power-generating steps. Resolution of this conflict between physiological measurements and the three-dimensional structure of myosin should lead to a greater insight into the biological mechanism of muscle contraction.