6
A closer look at energy transduction in muscle Hirofumi Onishi* and Manuel F. Morales †‡ *Exploratory Research for Advanced Technology ‘‘Actin-Filament Dynamics’’ Project, Japan Science and Technology Agency, c/o RIKEN Harima Institute SPring-8 Center, Kouto, Sayo, Hyogo 679-5148, Japan; and University of the Pacific, San Francisco, CA 94115 Contributed by Manuel F. Morales, June 18, 2007 (sent for review January 1, 2007) Muscular force is the sum of unitary force interactions generated as filaments of myosins move forcibly along parallel filaments of actins, understanding that the free energy required comes from myosin-catalyzed ATP hydrolysis. Using results from conventional biochemistry, our own mutational studies, and diffraction images from others, we attempt, in molecular detail, an account of a unitary interaction, i.e., what happens after a traveling myosin head, bearing an ADP-P i , reaches the next station of an actin filament in its path. We first construct a reasonable model of the myosin head and actin regions that meet to form the ‘‘weakly bound state’’. Separately, we consider Holmes’ model of the rigor state [Holmes, K. C., Angert, I., Kull, F. J., Jahn, W. & Schro ¨ der, R. R. (2003) Nature 425, 423– 427], supplemented with several hereto- fore missing residues, thus realizing the ‘‘strongly bound state.’’ Comparing states suggests how influences initiated at the inter- face travel elsewhere in myosin to discharge various functions, including striking the actins. Overall, state change seems to occur by attachment of a hydrophobic triplet (Trp-546, Phe-547, and Pro-548) of myosin to an actin conduit with a hydrophobic guiding rail (Ile-341, Ile-345, Leu-349, and Phe-352) and the subsequent linear movement of the triplet along the rail. actin ATPase mutation myosin H ere, we attempt a molecular account of the central inter- actions that occur when muscle contracts or sustains ten- sion. Because of a peculiar structural organization of muscle, the contractile apparatus can be thought of as effectively a linear filament of myosin molecules forcibly advancing along a parallel filament of actin molecules, the process being thermodynami- cally paid for by losing free energy of myosin-bound ATP hydrolysis (1). In this sense, contraction is an instance of transducing chemical (free) energy into mechanical work. For our purposes, however, it is simpler to consider our task to be explaining how a traveling myosin head, bearing a partly hydro- lyzed ATP, on reaching interaction distance to an actin pair, fastens to it, first weakly and later strongly, and then, how it, in a remarkable cooperation with the actins, casts off the terminal phosphate of ATP and delivers a mechanical impulse to the actins. The account that follows is put together by using reports from many laboratories. These are of three general kinds, conven- tional biochemical [e.g., Scheme 1, § showing the time- dependence of forming complexes of myosin and actin (2–4)]; mutational [e.g., recent works identifying myosin surface loops engaged in complexing with actin (5, 6)]; and x-ray diffraction [e.g., studies of Rayment (7–10), of Cohen (11), and of their pioneering associates]. Application of diffraction techniques has dramatically improved resolution and has shown that a globular head of myosin has specialized ‘‘organs,’’ like an enzyme pocket in which to conduct ATP hydrolysis, a long, stiff -helix (‘‘relay helix’’) to transmit linear force, a converter to turn it into a rotation, and a lever arm to deliver a mechanical impulse to (at the time attached) actins (Fig. 1). In addition, from fitting crystal structures of actin and the myosin head to the 3D structure reconstructed from cryoEM pictures of actin filaments deco- rated with S1 (an isolated single myosin head), Rayment et al. (7) have suggested that ionic and hydrophobic residues, which are possibly involved in actin binding, are localized at one end of the myosin head, which is far from either the enzyme pocket or the converter (Fig. 1). At the end of the head, the so-called ‘‘actin-binding’’ cleft separates the 50-kDa domain into two parts, named the upper and lower subdomains, both of which are involved in actin binding (7, 12). Because the enzyme pocket was connected to the apex of the cleft, they proposed that actin binding causes the cleft closure, resulting in opening the enzyme pocket and accelerating product releases. However, the precise mechanism is not yet clear, because of limited information on the weakly bound complex, which is initially formed when a traveling myosin head encounters two actins. Recently, many functional tests have been carried out with expressed myosin mutants at sites thought from crystallography to be a patch for binding actins and demonstrated that some residues are mainly engaged in the formation of the weakly bound complex, whereas other residues are mainly engaged in the transition from the weakly- to-strongly bound complex (Table 1). Here, based on knowl- edge from these tests, we construct a reasonable model of the myosin and actin regions that meet to form weak binding. Comparing our model with a rigor complex model (12), we suggest how influences initiated at the actin–myosin interface travel elsewhere in myosin to discharge various functions, in- cluding cleft closure, P i release, and a mechanical impulse. Results and Discussion Modeling of the Weakly Bound Complex of Fibrous Actin (F-Actin) and M.ADP.Pi . A model of the weakly bound state to be constructed must have some properties like those of its predecessor, some properties that draw it to both members of the complex, and some properties that move it toward its successor. We assume that, on entry, the myosin active site continues to be posthydro- lytic and that complex formation is mainly driven by the rela- tively long-range electrostatic attractions between myosin and the two actins. To form such a model of the weakly bound complex, we use the crystal structure of a Dictyostelium myosin motor domain complexed with MgADP.VO 4 , which has been Author contributions: H.O. and M.F.M. designed research; H.O. performed research; H.O. contributed new reagents/analytic tools; H.O. analyzed data; and H.O. and M.F.M. wrote the paper. The authors declare no conflict of interest. Freely available online through the PNAS open access option. Abbreviation: HMM, double-headed myosin fragment. To whom correspondence may be addressed. E-mail: [email protected] or manuel. [email protected]. § A, M, and Pi are actin, a myosin head, and orthophosphate, respectively. Improved resolution indicates that the myosin head interacts with two actins of an actin filament (see below); the 1 : 1 ratio pictured in the scheme should for now be taken as a kinetics approximation. Also to be noted is the distinction between ‘‘A’’ and ‘‘A’’. This simply reminds one that after a myosin completes an interaction ending with AM, it reacts with a new ATP and is thus freed to travel to the next station, where it forms A.ADP.Pi (A and A, of course, are chemically identical but are distinguishable because of their separate locations on the actin filament). This loop has been first reported by using cardiac myosin to participate functionally in the actomyosin interaction (13). Recently a study mutating Glu-365 at the loop 4 of Dictyo- stelium myosin (corresponding to Asp-374 of smooth muscle myosin) into glutamine suggested that this residue has an important role in the actomyosin interaction particu- larly in the weakly bound state [a late abstract Pos-L96 in the 50th Biophysical Society Annual Meeting (February 22, 2006)]. © 2007 by The National Academy of Sciences of the USA 12714 –12719 PNAS July 31, 2007 vol. 104 no. 31 www.pnas.orgcgidoi10.1073pnas.0705525104

A closer look at energy transduction in · PDF fileA closer look at energy transduction in muscle ... and a lever arm to deliver a mechanical impulse to (at ... parts,namedtheupperandlowersubdomains,bothofwhichare

Embed Size (px)

Citation preview

Page 1: A closer look at energy transduction in · PDF fileA closer look at energy transduction in muscle ... and a lever arm to deliver a mechanical impulse to (at ... parts,namedtheupperandlowersubdomains,bothofwhichare

A closer look at energy transduction in muscleHirofumi Onishi*† and Manuel F. Morales†‡

*Exploratory Research for Advanced Technology ‘‘Actin-Filament Dynamics’’ Project, Japan Science and Technology Agency, c/o RIKEN Harima InstituteSPring-8 Center, Kouto, Sayo, Hyogo 679-5148, Japan; and ‡University of the Pacific, San Francisco, CA 94115

Contributed by Manuel F. Morales, June 18, 2007 (sent for review January 1, 2007)

Muscular force is the sum of unitary force interactions generatedas filaments of myosins move forcibly along parallel filaments ofactins, understanding that the free energy required comes frommyosin-catalyzed ATP hydrolysis. Using results from conventionalbiochemistry, our own mutational studies, and diffraction imagesfrom others, we attempt, in molecular detail, an account of aunitary interaction, i.e., what happens after a traveling myosinhead, bearing an ADP-Pi, reaches the next station of an actinfilament in its path. We first construct a reasonable model of themyosin head and actin regions that meet to form the ‘‘weaklybound state’’. Separately, we consider Holmes’ model of the rigorstate [Holmes, K. C., Angert, I., Kull, F. J., Jahn, W. & Schroder, R. R.(2003) Nature 425, 423–427], supplemented with several hereto-fore missing residues, thus realizing the ‘‘strongly bound state.’’Comparing states suggests how influences initiated at the inter-face travel elsewhere in myosin to discharge various functions,including striking the actins. Overall, state change seems to occurby attachment of a hydrophobic triplet (Trp-546, Phe-547, andPro-548) of myosin to an actin conduit with a hydrophobic guidingrail (Ile-341, Ile-345, Leu-349, and Phe-352) and the subsequentlinear movement of the triplet along the rail.

actin � ATPase � mutation � myosin

Here, we attempt a molecular account of the central inter-actions that occur when muscle contracts or sustains ten-

sion. Because of a peculiar structural organization of muscle, thecontractile apparatus can be thought of as effectively a linearfilament of myosin molecules forcibly advancing along a parallelfilament of actin molecules, the process being thermodynami-cally paid for by losing free energy of myosin-bound ATPhydrolysis (1). In this sense, contraction is an instance oftransducing chemical (free) energy into mechanical work. Forour purposes, however, it is simpler to consider our task to beexplaining how a traveling myosin head, bearing a partly hydro-lyzed ATP, on reaching interaction distance to an actin pair,fastens to it, first weakly and later strongly, and then, how it, ina remarkable cooperation with the actins, casts off the terminalphosphate of ATP and delivers a mechanical impulse to theactins.

The account that follows is put together by using reports frommany laboratories. These are of three general kinds, conven-tional biochemical [e.g., Scheme 1,§ showing the time-dependence of forming complexes of myosin and actin (2–4)];mutational [e.g., recent works identifying myosin surface loopsengaged in complexing with actin (5, 6)]; and x-ray diffraction[e.g., studies of Rayment (7–10), of Cohen (11), and of theirpioneering associates]. Application of diffraction techniques hasdramatically improved resolution and has shown that a globularhead of myosin has specialized ‘‘organs,’’ like an enzyme pocketin which to conduct ATP hydrolysis, a long, stiff �-helix (‘‘relayhelix’’) to transmit linear force, a converter to turn it into arotation, and a lever arm to deliver a mechanical impulse to (atthe time attached) actins (Fig. 1). In addition, from fitting crystalstructures of actin and the myosin head to the 3D structurereconstructed from cryoEM pictures of actin filaments deco-rated with S1 (an isolated single myosin head), Rayment et al. (7)have suggested that ionic and hydrophobic residues, which arepossibly involved in actin binding, are localized at one end of the

myosin head, which is far from either the enzyme pocket or theconverter (Fig. 1). At the end of the head, the so-called‘‘actin-binding’’ cleft separates the 50-kDa domain into twoparts, named the upper and lower subdomains, both of which areinvolved in actin binding (7, 12). Because the enzyme pocket wasconnected to the apex of the cleft, they proposed that actinbinding causes the cleft closure, resulting in opening the enzymepocket and accelerating product releases. However, the precisemechanism is not yet clear, because of limited information on theweakly bound complex, which is initially formed when a travelingmyosin head encounters two actins. Recently, many functionaltests have been carried out with expressed myosin mutants atsites thought from crystallography to be a patch for bindingactins and demonstrated that some residues are mainly engagedin the formation of the weakly bound complex, whereas otherresidues are mainly engaged in the transition from the weakly-to-strongly bound complex (Table 1).¶ Here, based on knowl-edge from these tests, we construct a reasonable model of themyosin and actin regions that meet to form weak binding.Comparing our model with a rigor complex model (12), wesuggest how influences initiated at the actin–myosin interfacetravel elsewhere in myosin to discharge various functions, in-cluding cleft closure, Pi release, and a mechanical impulse.

Results and DiscussionModeling of the Weakly Bound Complex of Fibrous Actin (F-Actin) andM.ADP.Pi. A model of the weakly bound state to be constructedmust have some properties like those of its predecessor, someproperties that draw it to both members of the complex, andsome properties that move it toward its successor. We assumethat, on entry, the myosin active site continues to be posthydro-lytic and that complex formation is mainly driven by the rela-tively long-range electrostatic attractions between myosin andthe two actins. To form such a model of the weakly boundcomplex, we use the crystal structure of a Dictyostelium myosinmotor domain complexed with MgADP.VO4

�, which has been

Author contributions: H.O. and M.F.M. designed research; H.O. performed research; H.O.contributed new reagents/analytic tools; H.O. analyzed data; and H.O. and M.F.M. wrotethe paper.

The authors declare no conflict of interest.

Freely available online through the PNAS open access option.

Abbreviation: HMM, double-headed myosin fragment.

†To whom correspondence may be addressed. E-mail: [email protected] or [email protected].

§A, M, and Pi are actin, a myosin head, and orthophosphate, respectively. Improvedresolution indicates that the myosin head interacts with two actins of an actin filament(see below); the 1 : 1 ratio pictured in the scheme should for now be taken as a kineticsapproximation. Also to be noted is the distinction between ‘‘A’’ and ‘‘A’’. This simplyreminds one that after a myosin completes an interaction ending with AM, it reacts witha new ATP and is thus freed to travel to the next station, where it forms A.ADP.Pi (A andA, of course, are chemically identical but are distinguishable because of their separatelocations on the actin filament).

¶This loop has been first reported by using cardiac myosin to participate functionally in theactomyosin interaction (13). Recently a study mutating Glu-365 at the loop 4 of Dictyo-stelium myosin (corresponding to Asp-374 of smooth muscle myosin) into glutaminesuggested that this residue has an important role in the actomyosin interaction particu-larly in the weakly bound state [a late abstract Pos-L96 in the 50th Biophysical SocietyAnnual Meeting (February 22, 2006)].

© 2007 by The National Academy of Sciences of the USA

12714–12719 � PNAS � July 31, 2007 � vol. 104 � no. 31 www.pnas.org�cgi�doi�10.1073�pnas.0705525104

Page 2: A closer look at energy transduction in · PDF fileA closer look at energy transduction in muscle ... and a lever arm to deliver a mechanical impulse to (at ... parts,namedtheupperandlowersubdomains,bothofwhichare

proposed to be an analog of the transient state or the posthy-drolytic state by Smith and Rayment (10), and dock it to anatomic model of F-actin. As shown in a stereoview of Fig. 2, fourpositively charged segments of myosin fit well spatially to fournegatively charged segments of the two actins, respectively;Lys-652� and Lys-653 of myosin fit to Asp-24 and Asp-25 of thefirst actin, Arg-530 of myosin to the negatively charged clusterbetween Asp-1 and Glu-4 (Residues 1, 2, and 3 are missing in thecrystal structure of rabbit skeletal actin) of the first actin,Lys-573 of myosin to Glu-99 and Glu-100 of the second actin, andLys-576 and Lys-578 of myosin to Glu-93 of the second actin.This is consistent with previous studies using a zero-lengthcross-linker that bridges a carboxylic group and lysine (16, 17).In these studies, a myosin head was cross-linked to two distinctactins at its lower 50-kDa subdomain and at its C-terminal20-kDa domain. In our model, Asp-374 at the so-called ‘‘loop 4’’of myosin also fits well to Lys-326 of the first actin. Many recentkinetic studies with actin mutants (18–20) or myosin mutants (5,6, 21, 22) also show that the foregoing residues are involved informing the weakly bound complex.

Interestingly, our model suggests that before cleft closure,both Trp-546 and Phe-547 in the hydrophobic triplet of myosinare already close to Ile-345 and Leu-349 of the first actin,although a surface loop bearing Pro-534 and Pro-535 (‘‘proline-rich loop’’) is not as yet close to either Leu-349 or Phe-352 of thefirst actin. In this early stage, Val-409 and Gly-410 at thecardiomyopathy loop of myosin are also close to Pro-332,Pro-333, and Glu-334 of the first actin, but this loop seems tomove freely to either side on the actin surface. Therefore, wespeculate that the myosin head in this complex does not yetcompletely orient on the actin filament and that this complex caneasily dissociate into actin and M.ADP.Pi.

Involvement of Four Surface Loops in the Transition from the Weakly-to-Strongly Bound State. Functional tests of mutants suggest thatin contrast to the early stage, four surface loops of myosin areinvolved in the later Pi-leaving stage of the actin-activatedATPase reaction (Table 1). One loop bears a hydrophobic tripletof residues, the second is the proline-rich loop, the third is theloop 2, and the fourth is the cardiomyopathy loop. Because thefluorescence of F-actin (whose Cys-374 is labeled with pyrene-iodoacetamide) is quenched by � 80% on complexing withwild-type myosin (23), measurements of pyrene–actin fluores-cence have been often used to detect formation of a rigorcomplex. In an immediately preceding paper (6), we reportedthat triplet-compromised mutants (W546A, F547A, and P548A)had a reduced extent of fluorescence quenching when com-plexed with pyrene-labeled F-actin, whereas the maximum ex-

tents of quenching by double-headed myosin fragments (HMMs)mutated at the C terminus of loop 2 or the cardiomyopathy loopwere identical to that obtained with wild-type HMM. From thisresult and data listed in Table 1, we conclude that, although allthree loops work together in cleft closure and actin activation,the triplet-bearing loop is somehow different from two otherloops in the route by which it communicates with its specifictarget elsewhere in myosin. As described below, we also thinkthat the proline-rich loop contact with the first actin is engagedin closing the cleft.

Comparison Between Two Complex Models. To know what happensat the actin–myosin interface when the weakly bound stateconverts into the strongly bound state, we compare our weaklybound complex model with the acto-S1 rigor complex modelpublished by Holmes et al. (12), by superimposing their actinswith ours. The myosin head has rotated upward (toward the firstactin), suggesting that its binding with the first actin is strength-ened (arrow a in Fig. 3), whereas that with the second actin isweakened (arrow b in Fig. 3). This is consistent with our previousresults with HMMs mutated lysines of interest here into alanines,suggesting that two lysines (Lys-652 and Lys-653) at the Cterminus of loop 2 are involved either in the weakly boundAM.ADP.Pi-forming stage or in the later Pi-releasing stage andthat two lysines (Lys-576 and Lys-578) at the second actin-binding loop are also involved in the former stage but not in thelatter stage (Table 1).

To deduce the role of Lys-652 and Lys-653 more precisely, wemust know their exact locations in the rigor complex. In Holmes’rigor complex model (12); however, the entire loop 2 is missing,so we cannot use it to find the location of the two lysines. Wesupplement his model with 9 aa residues including these lysines,using the crystal structure of Dictyostelium myosin motor domainwith MgADP.VO4

� (10). We do this by superimposing theirconnecting helices and �-sheet strands (Fig. 4A). The distancesof these lysines to Asp-24 or Asp-25 of the first actin (Lys-652�-carbon to Asp-24 �-carbon, 4.8 Å and Lys-653 �-carbon toAsp-25 �-carbon, 6.4 Å) are now closer than those in the weaklybound state (5.4 and 6.5 Å), suggesting that electrostatic inter-actions between these charged residues may have important rolesdifferent from those functioned in the weakly bound complex.Lys-652 and Lys-653 are, in one direction, connected to the third

�Although the amino acid sequence of proteins is different in various organisms (or indifferent major tissues of the same organism), the residues of interest here are highlyconserved. Throughout, we use the sequence numeration appropriate for smooth musclemyosin, as extracted from chicken gizzard (14). We note that Arg-247, Asp-374, Val-409,Gly-410, Asp-412, Val-414, Glu-470, Arg-530, Thr-532–Asn-533–Pro-534–Pro-535, Trp-546–Phe-547–Pro-548, Glu-557, Lys-573, Lys-576–Lys-578, and Lys-652–Lys-653 correspond torabbit skeletal Arg-245, Glu-372, Val-408, Gly-409, Glu-411, Val-413, Glu-468, Lys-528,Met-530, Met-541–Phe-542–Pro-543, Gln-552, Lys-569, Lys-572–Lys-574, and Lys-641–Lys-642, respectively, and to Dictyostelium discoideum Arg-238, Glu-365, Ala-400, Gly-401,Asp-403, Val-405, Glu-459, Arg-520, Gln-521–Pro-522–Pro-523, Val-534–Phe-535–Pro-536,Thr-545, Arg-562, Lys-565, and Lys-622–Lys-623, respectively. We also use the rabbitskeletal sequence numeration for actin (15).

Fig. 1. Specialized ‘‘organs’’ of myosin. The N-terminal 25-kDa (residues2–202), the upper 50-kDa (207–468), the lower 50-kDa (469–651), and theC-terminal 20-kDa (652–791) segments of the myosin motor domain heavychain are colored yellowish green, red, ocher, and dark blue, respectively. Thelever arm is not shown. Locations of actin-binding surface loops: 1, the secondactin-binding loop; 2, the hydrophobic triplet; 3, the proline-rich loop; 4, theC terminus of loop 2; 5, the cardiomyopathy loop; and 6, loop 4. Mutatedresidues in these loops are indicated as balls. The crystal structure of the motordomain of Dictyostelium myosin with MgADP.VO4

� is adapted from ref. 10.MgADP.VO4

� is shown as balls in white (carbon), blue (nitrogen), red (oxygen),green (magnesium), and ocher (phosphorus).

Scheme 1.

Onishi and Morales PNAS � July 31, 2007 � vol. 104 � no. 31 � 12715

BIO

CHEM

ISTR

Y

Page 3: A closer look at energy transduction in · PDF fileA closer look at energy transduction in muscle ... and a lever arm to deliver a mechanical impulse to (at ... parts,namedtheupperandlowersubdomains,bothofwhichare

strand of the seven-stranded �-sheet via a helix named ‘‘HW’’ byCoureux et al. (24); in another direction, they reach a part (asegment from Pro-604 to Val-628) of the upper 50-kDa subdo-main, via a loop called a ‘‘strut’’ by Sutoh (25). Therefore, it isnatural to assume that the contact of the C terminus of loop 2with the first actin (a in Fig. 4A) is very important in positioningthe backbone �-sheet, on which both the upper 50-kDa subdo-main and the N-terminal 25-kDa domain are carried. In the rigorstate, the cardiomyopathy loop, which has moved freely towardeither side in the weakly bound state, is caught by a dead-endsurface or ‘‘ditch’’ (surrounded by residues Ala-26, Pro-27,Ala-29, Val-30, Tyr-337, and Tyr-338) of the first actin (b in Fig.4A). So, this loop can no longer move freely on the actin surface.In fact, the Holmes’ model shows an overlap of masses aroundthe contact of the cardiomyopathy loop of myosin and the firstactin (12). We conclude that the rigor myosin head, particularlythe upper 50-kDa subdomain and the N-terminal 25-kDa do-main, orients at a specific angle to the actin filament by makinguse of the two anchors described above. We also conclude thatthis stereospecific attachment is necessary before both actinactivation and delivery of the mechanical impulse, because themutation of Lys-652 or Lys-653, or of the cardiomyopathy loop(Ile-407, Asp-412, or Val-414), into an alanine significantlyreduces actin-activated ATPase and in vitro motility (6).

A Driving Force for the Movement of the Lower 50-kDa Subdomain. Toclose the actin-binding cleft, the lower 50-kDa subdomain must

move differently from the upper 50-kDa subdomain or theN-terminal 25-kDa domain. What force causes the lower sub-domain to move so? From previous results that mutations of thehydrophobic triplet most remarkably weakened the affinity ofHMM for F-actin (6), we assume that, upon cleft closure, thestrongest contact with the first actin is that of the triplet-bearingloop. Residues between Lys-336 and Phe-352 of actin form asurface �-helix (Fig. 4A), and among them, residues Ile-341,Ile-345, Leu-349, and Phe-352 are exposed on the bindinginterface for myosin (Fig. 4B). These four residues, properlyaligned, form a ‘‘conduit’’ along which the hydrophobic triplet ofmyosin moves, the side chains of the residues form a guiding railthat preserves direction. We also assume that the movementalong the conduit occurs when the cleft closes. In Holmes’ rigormodel, this �-helix is aligned antiparallel to the �-helix of Ile-515to Cys-545 (adjacent to the triplet Trp-546–Phe-547–Pro-548) ofmyosin, and there are two bridges, with mainly hydrophobicinteractions, between the two helices. Trp-546 and Phe-547 ofthe triplet contact the hydrophobic conduit of the first actin, justas in the weakly bound complex (a in Fig. 4B). But in the rigorcomplex, they hold on to a node between Ile-341 and Ile-345**of the conduit but not to the node between Ile-345 and Leu-349.Thus, the attached triplet linearly slides by one turn along theaxis of the �-helix of actin when the complex converts from theweakly into the strongly bound state. From these observations,we think that the approach of the triplet of myosin to the conduitof actin, and its slide along the conduit, triggers several impor-tant movements in the myosin head. These movements close thecleft, resulting in Pi release and a mechanical impulse. Thisexplains why triplet-compromised mutations cause serious dam-age in both actin activation and motility (5, 6).

The proline-rich loop of smooth muscle or Dictyosteliummyosin, composed of residues between Glu-529 and Gly-536,projects from the center of a long �-helix composed of residuesIle-515 to Cys-545 of the myosin (10, 11). Because Holmes’ rigormodel used skeletal muscle S1 as the binding partner of actins,Thr-532–Asn-533–Pro-534–Pro-535 were substituted by only amethionine (12). To compare Holmes’ model with ours, wesupplement his model with six residues in the proline-rich loopusing the same Dictyostelium myosin structure (10), superim-posing the connecting �-helices. Pro-534 and Pro-535 in thisloop, just like Met-530 of skeletal muscle S1, fit well to Leu-349and Phe-352 at the end of the hydrophobic conduit of the firstactin (b in Fig. 4B). This hydrophobic interaction network hasnot yet appeared in the weakly bound complex, so we speculatethat the new formation of these hydrophobic interactions movesthe triplet along the conduit of the first actin, leading finally tostabilizing the actin–myosin interaction. Our view agrees wellwith various proposals (7, 27, 28), namely that hydrophobic

**Yeast actin mutants with alanine replacing Ile-341 or Ile-345 grow normally, althoughthe mutant with alanines or lysines replacing both isoleucines does not survive (26). Sowe suppose that a remaining isoleucine can still support the linear sliding of thehydrophobic triplet of myosin, although the movement may not be so smooth as thatobtained by a pair of isoleucines.

Fig. 2. Stereoview of the weakly bound AM.ADP.Pi complex. The same colorsare used for segments of the myosin heavy chain as in Fig. 1, whereas the firstactin and the second actin are colored cyan and dark green, respectively. Thedirection of the pointed end of the actin filament is upward. Myosin residuesthat are possibly involved in binding for actins (as balls): 1, Lys-652 and Lys-653(blue); 2, Arg-530 (blue); 3, Lys-573 (blue); 4, Lys-576 and Lys-578 (blue), 5,Asp-374 (red), and 6, Trp-546 (purple), Phe-547 (dark blue), and Pro-548 (pink).The corresponding actin residues are also shown as balls. The cardiomyopathyloop is indicated as a red ribbon marked 7. Crystal structures of the myosinmotor domain and monomeric rabbit skeletal actin are adapted from refs. 10and 33, respectively. Side chains of some residues (for example, Asp-374,Lys-652, and Lys-653 for myosin and Glu-4 for actin), important for actinbinding, are missing in these structures.

Table 1. Functional tests of myosins mutated at sites thought to be a patch for actin binding

Sites

Vmax Kapp

Reaction stage to be involved in Refs.Actin-Activated ATPase

Second actin-binding loop No change Increase Weak binding 5Hydrophobic triplet Decrease No change Transition 5, 6Proline-rich loop Decrease No change Transition 5C-terminal end of loop 2 Decrease Increase Weak binding and transition 6Cardiomyopathy loop Decrease No change Transition 6Loop 4 (cardiac) No change Increase Weak binding ¶

12716 � www.pnas.org�cgi�doi�10.1073�pnas.0705525104 Onishi and Morales

Page 4: A closer look at energy transduction in · PDF fileA closer look at energy transduction in muscle ... and a lever arm to deliver a mechanical impulse to (at ... parts,namedtheupperandlowersubdomains,bothofwhichare

interaction networks at the actin–myosin interface expand dur-ing the weak-to-strong conversion. Previously, we reported thatVmax values of smooth muscle mutants at Thr-532–Asn-533–Pro-534–Pro-535 substituted by one methionine or one lysine were �2-fold greater or 1/2 of that of the wild-type, respectively (5). So,we suggest that this proline-rich loop is important for the rate ofcleft closure (and thus the rate of Pi release), because this loophas an important role in forming the second hydrophobicinteraction network with the first actin. In the rigor complex,Arg-530 of myosin, a part of the proline-rich loop, is rather farfrom Glu-4 of the first actin (although they, behind the proline-rich loop, are not seen in Fig. 4B); nevertheless, they mayelectrostatically interact with each other, if the side chain ofArg-530 is twisted, or if missing N-terminal residues (Asp-1,Glu-2, and Asp-3) are added to Glu-4 of the first actin. Becausethis arginine has already made contact with N-terminal nega-tively charged residues of the first actin in the early, weaklybound state (Fig. 2), this f lexible bridge may assist the proline-rich loop to approach the end of the conduit of the attachedactin.

How Do Events Initiated at the Interface Transmit Their Influences tothe Cleft, the Pocket, and the Lever Arm? It is widely believed thatthe mechanical impulse to actins is delivered somewhere in thereaction steps from AM.ADP.Pi to the rigor complex (4).Because we have obtained snapshots of the pre- and postpower-stroke states, we now try to infer what occurs in between the twostates. Our approach (constructing a weakly bound complex andthen comparing it with the rigor complex) per se does not permittime sequencing of events in the interval. However, together withconventional biochemical and mutational studies, we can makea reasonable guess as described in the following. We assume thatan attachment of the triplet of myosin to the conduit of actin andthe subsequent run of the triplet along the conduit initiatevarious paths of influence within the myosin head in threedistinct stages. In the first stage (step 2 of Fig. 5), the approachof the triplet to the conduit releases the myosin head from thesecond actin and rotates it toward the first actin (two curvedarrows colored black in Fig. 5). The C terminus of loop 2 keepscontact with the first actin. The cardiomyopathy loop begins tomove on the actin surface toward the dead end. Then the cleftbegins to close, and the bound myosin head orients to a fixedangle to the axis of the actin filament.

In the second stage (step 3 of Fig. 5), the linear movement of

the triplet along the conduit and the formation of the secondhydrophobic interaction network force the lower 50-kDa sub-domain to rotate (curved arrow colored ocher in Fig. 5). Withinthe lower subdomain, a four-stranded �-sheet (residues Lys-568to Lys-571, Glu-580 to Leu-584, Lys-589 to Ala-594, and Glu-473to Phe-477) behaves as a stiff connector, which is especiallyimportant in transmitting displacements to two other functionalorgans. One is the enzyme pocket, and the other is the converter.The last strand of the four-stranded �-sheet is connected toswitch 2 of the enzyme pocket via a short loop, so the twistedrotation of the �-sheet (curved arrow near b of Fig. 4B) movesswitch 2 away from the �-phosphate of ATP, without deformingother parts of the pocket. These movements break the salt bridgebetween Arg-247 (on switch I) and Glu-470 (on switch II), andopen the so-called ‘‘back door’’ to accelerate Pi release (29). Theforegoing explains why actin facilitates the rate of Pi release frommyosin and also why myosin releases products in a sequence [firstPi and then ADP, even when the myosin head binds F-actin (4)].

The rotational movement of the lower subdomain initiated bythe linear movement of the triplet also causes the rotation of therelay helix, but this small rotation does not fully explain therotation of �70° of the converter around the principal axis ofthe SH1 helix. X-ray crystallographic studies of myosin motordomain complexes with various ATP analogs suggested that

Fig. 3. Comparison between myosin heads in our weakly bound complexmodel (colors are same as those in Fig. 1) and in the rigor complex model (cyan)reported by Holmes et al. (12). Actins of two models are superimposed, andmovements of the myosin head are shown by two curved arrows, a and b. Twoactins are omitted, except for several residues (numbers and colors are thesame as in Fig. 2) that bind myosin. The direction of the pointed end of theactin filament is upward (straight arrow).

Fig. 4. The rigor complex of acto-S1. Structures shown in A and B are derivedfrom the same structure reported by Holmes et al. (12). Segments of themyosin heavy chain and the first actin are colored the same as in Fig. 1 or 2, butthe second actin is not shown. (A) The lower 50-kDa subdomain is omittedfrom the myosin structure to look at two attachment sites: a, Lys-652 andLys-653 (blue balls) of myosin and Asp-24 and Asp-25 (red balls) of actin andb, the cardiomyopathy loop (red ribbon) of myosin and the dead-end surface(colored balls) surrounded by residues 26–30 and 337–338 of actin. Theseven-stranded �-sheet is indicated by ribbons colored yellowish green (strand1, 114–117; strand 2, 120–123; strand 4, 171–176), dark blue (strand 3, 678–684), and red (strand 5, 459–467; strand 6, 249–256; and strand 7, 262–269).The position of Lys-652 and Lys-653 is deduced by supplementing theseresidues from the crystal structure of the Dictyostelium myosin motor domainwith MgADP.VO4

� (10). The light purple ribbon indicates the superimposedDictyostelium HW helix and the third strand of the �-sheet. A helix of residues336–352 of the first actin is shown as a ribbon colored cyan. A big slantedarrow indicates the direction of the pointed end of the actin filament. (B) Onlythe lower 50-kDa subdomain and the C-terminal 20-kDa domain are shown tolook at two specific attachment sites of the lower 50-kDa subdomain for thefirst actin: a, the hydrophobic triplet (purple for Trp-546, dark blue forPhe-547, and pink for Pro-548) of myosin and Ile-341 and Ile-345 (green) ofactin and b, the proline-rich loop (pink for Pro-534 and Pro-535 and red forGlu-557) of myosin and Leu-349 (green) and Phe-352 (dark blue) of actin. Alinear arrangement of Ile-341, Ile-345, Leu-349, and Phe-352 forms a hydro-phobic conduit on actin. The proline-rich loop is supplemented to the helix ofresidues 515–545 from the Dictyostelium crystal structure (10). The orangeribbon indicates the superimposed Dictyostelium 515–545 helix. Three bigarrows indicate a linear movement of the hydrophobic triplet along theconduit, a twisted rotation of a proximal part of the lower 50-kDa subdomain,and a rotation of the converter, respectively. Switch 1 (residues 237–246) andswitch 2 (469–476) in the enzyme pocket are shown as ribbons colored red andyellow, respectively, and Arg-247 and Glu-470 are shown as balls colored blueand red, respectively.

Onishi and Morales PNAS � July 31, 2007 � vol. 104 � no. 31 � 12717

BIO

CHEM

ISTR

Y

Page 5: A closer look at energy transduction in · PDF fileA closer look at energy transduction in muscle ... and a lever arm to deliver a mechanical impulse to (at ... parts,namedtheupperandlowersubdomains,bothofwhichare

opening of switch 2 triggers structural changes of force-transmitting devices between the switch and the converter (9, 11,30). After these changes, the relay helix, which has been bent atits center when switch 2 has been closed, releases from theseven-stranded �-sheet and straightens. Straightening out of therelay helix enhances the converter rotation by additionallytwisting the distal half of the relay helix and the SH1 helix asdescribed in refs. 30 and 31. By this mechanical cascade, thelinear movement of the triplet finally converts into a largerotation of the converter and then of the rigidly attached leverarm (curved arrow colored violet in Fig. 5). By the rotation, thelever arm, lying distal to the actin filament, strikes the actinsupward (toward the pointed end of the filament) so as to advancemyosin toward the next station of the actin filament. Thecardiomyopathy loop is also important in delivering the mechan-ical impulse, as this loop has to be anchored to the first actinduring the power stroke. Functional importance of this loop issupported by Sutoh’s clever experiments showing that mutagenicdeletion of the cardiomyopathy loop abolishes the strong bindingbetween myosin and actin, in actin-activated ATPase and in vitromotility (32).

In the final stage of the power stroke (step 4 of Fig. 5), theupper 50-kDa subdomain largely rotates on an axis passingaround strands 5, 6, and 7 of the seven-stranded �-sheet (curvedarrow colored red in Fig. 5). We think that this is a result ofcomplicated global changes, within both the upper 50-kDasubdomain and the N-terminal 25-kDa domain; it is also initiatedby the linear movement of the triplet. The extreme movement ofthe relay helix finally results in a distortion of the seven-stranded�-sheet bearing the upper subdomain, because this helix isconnected with the first two strands of the �-sheet via hydro-phobic bridges. At the same time, however, the movement of theupper subdomain is restricted at two peripheral points (a and bin Fig. 4A) by anchoring to the first actin. As a result, the uppersubdomain rolls away from the N-terminal 25-kDa domain. This

forces the enzyme pocket to open, and ADP is ready to bereleased from the pocket. When a new ATP enters, the pocketcloses again by forces attracting both sides, and the uppersubdomain is returned to its original orientation. Because thecardiomyopathy loop, which has functioned as an anchor duringthe power stroke, is now detached from the first actin, the myosinhead as a whole dissociates from the actins and travels to the nextstation of the actin filament (step 5 in Fig. 5). During traveling,the conformation of the myosin head returns to the prepower-stroke state, and ATP is hydrolyzed into ADP and Pi. This stepbecomes the restoring action (‘‘reverse stroke’’) necessary for thestart of the next power stroke (30). The myosin head bearingADP-Pi then rebinds to the actin filament, and the cycle startsagain (step 1 in Fig. 5).

In summary, the foregoing findings account for the centralevents of muscle contraction in totally molecular terms, explain-ing the event cycle inferred by Lymn and Taylor (4) from kineticobservations. Although our reasoning has yet to be confirmed bycomputer simulations or precise docking programs, we can nowplausibly visualize the difference between weakly (prepower-stroke) and strongly (postpower-stroke) -bound states of thecontractile system. Comparison between states suggests that inthe transition between two bound states, the actin–myosininterface has a large expansion of hydrophobic interactionnetworks. If this result is interpreted as a large increase inconfigurational entropy of the system, then it can be consideredconsistent with the Van’t Hoff analyses of force development (28).

MethodsDocking Crystal Structures of Myosin to a 3D Model of F-Actin. Modelsof the rigor complex (rigor�complex.pdb), F-actin (3actin.pdb),and the myosin motor domain (motor�domain.pdb), and thecrystal structure of Dictyostelium myosin motor domain withMgADP.VO4

� (1VOM.pdb) were downloaded from the ProteinDatabase. Docking was performed visually by using RasTop ver.2.0.0.0, replacing atoms with balls with a Van der Walls radius.

1. Huxley HE (1969) Science 164:1356–1366.2. Bagshaw CR, Trentham DR (1974) Biochem J 141:331–349.3. Bagshaw CR, Eccleston JF, Eckstein F, Goody RS, Gutfreund H, Trentham

DR (1974) Biochem J 141:351–364.4. Lymn RW, Taylor EW (1971) Biochemistry 10:4617–4624.5. Kojima S, Konishi K, Katoh K, Fujiwara K, Martinez HM, Morales MF, Onishi

H (2001) Biochemistry 40:657–664.6. Onishi H, Mikhailenko SV, Morales MF (2006) Proc Natl Acad Sci USA

103:6136–6141.

7. Rayment I, Holden HM, Whittaker M, Yohn CB, Lorenz M, Holmes KC,Milligan RA (1993) Science 261:58–65.

8. Rayment I, Rypniewski WR, Schmidt-Base K, Smith R, Tomchick DR,Benning MM, Winkelmann DA, Wesenberg G, Holden HM (1993) Science261:50–58.

9. Fisher AJ, Smith CA, Thoden JB, Smith R, Sutoh K, Holden HM, Rayment I(1995) Biochemistry 34:8960–8972.

10. Smith CA, Rayment I (1996) Biochemistry 35:5404–5417.11. Dominguez R, Freyzon Y, Trybus KM, Cohen C (1998) Cell 94:559–571.

Fig. 5. A serial diagram depicting, step 1, the initial contact between a myosin head and two actins (first, cyan and second, dark green) of a filament; step 2,rotating the myosin head toward the first actin that follows the weak-to-strong transition; step 3, the rotation of the lower 50-kDa subdomain that results inPi release and movement of the converter–lever arm system; step 4, rolling of the upper 50-kDa subdomain that accelerates the ADP release; and step 5, therelease that frees the myosin head to move away toward the next station. The N-terminal 25-kDa domain, the upper 50-kDa subdomain, the lower 50-kDasubdomain, and the converter of myosin are colored yellowish green, red, ocher, and violet, respectively. The blue line in the myosin head indicates the C-terminalheavy-chain segment connecting between an actin-binding loop and the converter. The black spiral indicates a long �-helix that constitutes the core of thelever arm.

12718 � www.pnas.org�cgi�doi�10.1073�pnas.0705525104 Onishi and Morales

Page 6: A closer look at energy transduction in · PDF fileA closer look at energy transduction in muscle ... and a lever arm to deliver a mechanical impulse to (at ... parts,namedtheupperandlowersubdomains,bothofwhichare

12. Holmes KC, Angert I, Kull FJ, Jahn W, Schroder RR (2003) Nature 425:423–427.

13. Ajtai K, Garamszegi SP, Watanabe S, Ikebe M, Burghardt TP (2004) J BiolChem 279:23415–23421.

14. Yanagisawa M, Hamada Y, Katsuragawa Y, Imamura M, Mikawa T, MasakiT (1987) J Mol Biol 198:143–157.

15. Collins JH, Elzinga M (1975) J Biol Chem 250:5915–5920.16. Mornet D, Bertrand R, Pantel P, Audemard E, Kassab R (1981) Nature

292:301–306.17. Sutoh K (1983) Biochemistry 22:1579–1585.18. Razzaq A, Schmitz S, Veigel C, Molloy JE, Geeves MA, Sparrow JC (1999)

J Biol Chem 274:28321–28328.19. Johara M, Toyoshima YY, Ishijima A, Kojima H, Yanagida T, Sutoh K (1993)

Proc Natl Acad Sci USA 90:2127–2131.20. Miller CJ, Reisler E (1995) Biochemistry 34:2694–2700.21. Joel PB, Trybus KM, Sweeney HL (2001) J Biol Chem 276:2998–3003.

22. Van Dijk J, Furch M, Lafont C, Manstein DJ, Chaussepied P (1999) Biochem-istry 38:15078–15085.

23. Kouyama T, Mihashi K (1981) Eur J Biochem 114:33–38.24. Coureux PD, Sweeney HL, Houdusse A (2004) EMBO J 23:4527–4537.25. Sasaki N, Ohkura R, Sutoh K (2000) J Biol Chem 275:38705–38709.26. Miller CJ, Doyle TC, Bobkova E, Botstein D, Reisler E (1996) Biochemistry

35:3670–3676.27. Geeves MA, Conibear PB (1995) Biophys J 68:194S–199S.28. Zhao Y, Kawai M (1994) Biophys J 67:1655–1668.29. Rayment I, Smith C, Yount RG (1996) Annu Rev Physiol 58:671–702.30. Geeves MA, Holmes KC (2005) Adv Protein Chem 71:161–193.31. Ohki T, Mikhailenko SV, Morales MF, Onishi H, Mochizuki N (2004)

Biochemistry 43:13707–13714.32. Sasaki N, Asukagawa H, Yasuda R, Hiratsuka T, Sutoh K (1999) J Biol Chem

274:37840–37844.33. Otterbein LR, Graceffa P, Dominguez R (2001) Science 293:708–711.

Onishi and Morales PNAS � July 31, 2007 � vol. 104 � no. 31 � 12719

BIO

CHEM

ISTR

Y