contraction was demonstrated long before the fine structure of the myofibril became known. In about 1929, ATP was recognized as the energy source for muscle contraction, but it was not until 10 years later that Engelhardt and Ljubimowa showed that isolated myosin preparations catalyzed the hydrolysis of ATP.138 Szent-Györgi139,140 showed that a combination
of the two proteins actin (discovered by F. Straub141)
and myosin was required for Mg2+-stimulated ATP
hydrolysis (ATPase activity). He called this combina- tion actomyosin.
Under the electron microscope the myosin heads can sometimes be seen to be attached to the nearby thin actin filaments as crossbridges. When skeletal muscle is relaxed (not activated by a nerve impulse), the crossbridges are not attached, and the muscle can be stretched readily. The thin filaments are free to move past the thick filaments, and the muscle has some of the properties of a weak rubber band. How- ever, when the muscle is activated and under tension, the crossbridges form more frequently. When ATP is exhausted (e.g., after death) muscle enters the state of rigor in which the crossbridges can be seen by electron
microscopy to be almost all attached to thin filaments, accounting for the complete immobility of muscle in rigor (Figs. 19-12, 14).134
In rigor the crossbridges are almost all firmly attached to the thin actin filaments, making an approx- imately 45° angle to the actin filaments.142– 144 However,
the addition of ATP causes their instantaneous release and the relaxation of the muscle fiber. In contrast, activation by a nerve impulse, with associated release of calcium ions (Section B,4), causes the thin filaments to slide between the thick filaments with shortening of the muscle. An activated muscle shortens if a low tension is applied to the muscle, but at a higher tension it maintains a constant length. Because the maximum tension developed is proportional to the length of overlap between the thick and thin filaments, it was natural to identify the individual crossbridges as the active centers for generation of the force needed for contraction.
The rowing hypothesis. H. E. Huxley145,146 and
A. F. Huxley and R. M. Simmons147 independently
proposed that during contraction the myosin heads attach themslves to the thin actin filaments. The hy- drolysis of ATP is then coupled to the generation of a tension that causes the thick and thin filaments to be pulled past each other. The heads then release them- selves and become attached at new locations on the actin filament. Repetition of this process leads to the sliding motion of the filaments (Fig. 19-14). The evi- dence in favor of this “rowing” or “swinging bridge” hypothesis was initially based largely on electron microscopy. For example, contracting muscle was frozen rapidly and fixed for microscopy in the frozen state.148 Relaxed muscle shows no attached cross-
bridges, but contracting muscle has many. However,
Figure 19-14 A model for the coupling of ATP hydrolysis to force production in muscle based on proposals of H. E. Huxley,
and A. F. Huxley and Simmons. The power stroke is depicted here as a rotation of the crossbridge from a 90° to a 45° configu- ration. Four representative stages are shown: (1) the rigor complex, (3) the dissociated myosin ATP complex, (4) the actomyo- sin ADP pre-power stroke state in which the actin–myosin band has reformed but with a different actin subunit, which may be distant from that in (1), and (6) the actomyosin ADP post-power stroke state. Force production and contraction result from crossbridges passing cyclically through the steps depicted from left to right. Numbering of the stages corresponds approxi-
mately to that in Fig. 19-18. After H. Huxley.146
Myosin Actin 3 S1 1 ATP S2 4 ADP + Pi ADP 6
their appearance was distinct from that seen in rigor. The model was also supported by indirect physical methods.
An impressive demonstration that myosin heads do move along the actin filaments was provided by Sheetz and Spudich, who found that myosin-coated fluorescent beads ~0.7µm in diameter will move along actin filaments from cells of the alga Nitella in an ATP- dependent fashion at velocities similar to those required in muscle.149 The myosin heads literally glide along
the thick cables of parallel actin filaments present in these algae.
Why two heads? The actin filament is a two-start helix, and it is natural to ask whether the two myosin heads bind to just one or simultaneously to both of the actin strands. Most evidence supports a 1:1 interaction of a single head with just one strand of actin. However, the other actin strand may associate with heads from a different thick filament. Another question concerns the role of the pairs of myosin heads. Could the two heads bind sequentially to the actin and exert their pull in a fixed sequence? In the reconstruction of the actomyosin complex in rigor (Fig. 19-12B) two different images are seen for the crossbridges. This suggests the existence of two different conformations for the attached myosin heads. Similar images for smooth muscle heavy meromyosin in its inactive (resting) dephosphorylated state (see p. 1116) show the two heads in very different orientations with one binding to the other of the pair and blocking its movement.121b
Perhaps one head is tightly bound at the end of the power stroke while the other is at a different stage of the catalytic cycle. Nevertheless, single-headed myosin from Acanthamoeba will propell organelles along actin filaments,150 and actin filaments will slide across a
Figure 19-15 Ribbon representation of chicken skeletal myosin subfragment-1 showing the major domains and tryptic
fragments. Prepared with the program MolScript. From Rayment.157
surface coated with single-headed myosin formed by controlled proteolysis.151 The additional interactions
seen in rigor may be peculiar to that state.
Structure of the myosin heads. Myosin and myosin fragments can be isolated in large quantities, but they have been difficult to crystallize. However, Rayment and coworkers purified S1 heads cleaved from chicken myosin by papain and subjected them to reductive methylation (using a dimethylamine–borane complex; see also Eq. 3-34). With most of the lysine side chain amino groups converted to dimethylamine groups, high-quality crystals were obtained, and a structure was determined by X-ray diffraction.152
Since that time various forms of both modified and unmodified myosin heads from several species have been studied by X-ray crystallography.153–160 Especially
clear results were obtained with unmodified myosin from the ameba Dictyostelium discoideum. The head structure, shown in Fig. 19-11, includes a 95-kDa piece of the heavy chain and both light chains. A clearer picture of the neck region containing the light chains was provided by the structure of the “regulatory domain” of scallop myosin.161 Unlike mammalian or
avian myosins, molluscan myosins are regulated by binding of Ca2+ to a site in the essential light chain,
but the structures are similar to those in Figs. 19-10 and 19-15.
Cleavage of the ~850-residue S1 heads with trypsin yields mainly three large fragments that correspond to structural domains of the intact protein as shown in Fig. 19-15. They are known as the 25-kDa (N-terminal), 50-kDa, and 20-kDa fragments, and for myosin from
D. discoideum correspond to residues 1 to 204, 216 to
626, and 647 to 843, respectively. The ATP-binding site is in a deep cleft between the 20-kDa and 50-kDa
Upper domain of 50 kDa region Lower domain of 50 kDa region 50 kDa cleft Nucleotide pocket
C-terminal 20 kDa region
Essential light chain Regulatory light chain Attachment to myosin rod N-Terminal 25 kDa region
Figure 19-16 (A) The nucleotide binding site of myosin
with MgADP
·
BeFx bound in a conformation thought tomimic that of ATP prior to hydrolysis. The β-sheet strands are contributed by both the 25-kDa and 50-kDa domains. The P-loop lies between T178 and E187. The conserved N233 to G240 loop, which also contributes important ATP- binding residues, comes from the 50-kDa region. (B) Stereo- scopic view of the γ-phospho group binding pocket with the
bound MgADP
·
VO4(vanadate) complex. The coordinatedMg2+ and associated water molecules are seen clearly.
Courtesy of Ivan Rayment.157
regions. Figure 19-16 illustrates the binding of an ATP analog, the beryllium fluoride complex of MgADP, in the active site. As can be seen, the ATP binds to loops at the C termini of the β strands of the 8-stranded β sheet from the 25-kDa domain. The conserved P-loop (Chapter 12, E), which lies between T178 and E187, curls around the α and β phospho groups, and has the sequence G(179)ESGAGKT. A second conserved loop N(233)SNSSR-G(240) from the 50-kDa domain contrib- utes to the binding of ATP.
The actin-binding region of the myosin head is formed largely by the 50-kDa segment, which is split by a deep cleft into two separate domains (Fig. 19-15), both of which are thought to participate in binding to actin. A surface loop (loop 1) near the ATP-binding site at the junction of the 25- and 50-kDa regions affects the kinetic properties of myosin, probably by influencing product release. A second loop (loop 2, residues 626 – 647) at the junction of the 50- and 20-kDa regions interact with actin. Loop 2 contains a GKK
sequence whose positive charges may interact with negative charges in the N-terminal part of actin.162– 164
The C-terminal fragment of myosin contains a globular domain that interacts with both the 20-kDa and 50-kDa regions and contains an α-helical neck that connects to the helix of the coiled-coil myosin rod. This helical region is surrounded by the two myosin light chains (Fig. 19-15).157 A pair of reactive thiol
groups (from C697 and C707) in the globular domain are near the active site. Crosslinking of these cysteines by an – S – S – bridge has been utilized to trap nucleotide analogs in the active site.165
How does actin bind? The actin monomer con- sists of four subdomains, 1, 2, 3, and 4 numbered from the N terminus (Fig. 7-10). The negatively charged N-terminal region of actin contains the sequence
A B D E 4 24 25 D E D D. K185 T186 457 S237 3 1 4 G182 W3 2 S236 E459 R238 K185 T186 457 S237 3 1 4 W3 2 G182 E459 S236 R238 Mg++ Mg++
It may interact with loop 1 of myosin, which contains five lysines. However, to form a strong interaction with the myosin head a conformational change must occur in the myosin. A change may also occur in actin. Modeling suggests that a large nonpolar contact region involves actin residues A144, I341, I345, L349, and F352 and myosin residues P529, M530, I535, M541, F542, and P543. A conformational change in actin, which might involve largely the highly conserved actin subdomain 2, may also be required for tight interaction.142,166– 168
Kinesins and other molecular motors. Before considering further how the myosin motor may work, we should look briefly at the kinesins, a different group of motor molecules,168a which transport various
cellular materials along microtubule “rails.” They also participate in organization of the mitotic spindle and other microtubule-dependent activities.168a,b,c See
Section C,2 for further discussion. More than 90 mem- bers of the family have been identified. Kinesin heads have much shorter necks than do the myosin heads. A myosin head is made up of ~ 850 residues, but the motor domain of a kinesin contains only ~ 345. Like
Figure 19-17 Ribbon drawing of human kinesin with
bound Mg
·
ADP. From Gulick et al.174 Courtesy of IvanRayment and Andy Gulick.
myosin, the 950- to 980-residue kinesins have a long coiled-coil C- terminal region that forms a “neck” of ~ 50 residues, a “stalk” of ~ 190 and ~ 330 residue seg- ments with a Pro / Gly-rich hinge between them, and an ~ 45 residue “tail.”169–171
Crystal structures are known for motor domains of human kinesin172 and of a kinesin from rat brain.169,173
The structures of one of six yeast kinesins,174 a protein
called Kar3, and also of a Drosophila motor molecule designated Ncd have also been determined.175 The last
was identified through study of a Drosophila mutant called non-claret disjunctional (Ncd). The motor domains of various members of the kinesin family show ~ 40% sequence identity and very close structural identity (Fig. 19-17).174 Although the sequences are
different from those of the myosin heads or of G proteins, the folding pattern in the core structures is similar in all cases. An 8-stranded β sheet is flanked by three α helices on each side and a P-loop crosses over the ATP- binding site as in Fig. 19-16. Further similarity is found in the active site structures, which, for a monomeric kinesin KIF1A,174a have been determined both with
bound ADP and with a nonhydrolyzable analog of ATP.174b,174c Although there is little similarity in amino
acid sequences the structures in the catalytic core are clearly related to each other, to those of dimeric kinesins,174d to those of myosins, and to those of the
GTP-hydrolyzing G proteins.
A puzzling discovery was that the motor domain of kinesin, which binds primarily to the β subunits of tubulin (Fig. 7-34) and moves toward the fast growing
plus end of the microtubule,176 is located at the N ter-
minus of the kinesin molecule, just as is myosin. How- ever, the Ncd and Kar3 motor domains are at the C- terminal ends of their peptide chains and move their “cargos” toward the minus ends of microtubules.174
Nevertheless, the structures of all the kinesin heads are conserved as are the basic chemical mechanisms. The differences in directional preference are determined by a short length of peptide chain between the motor domain and the neck, which allows quite different geo- metric arrangements when bound to microtubules.173,177
Like Ncd, myosin VI motor domains also move “back- wards” toward the pointed (minus) ends of actin filaments.178–179a
Other major differences between kinesins and myo- sin II heads involve kinetics180,181 and processivity.173
Dimeric kinesin is a processive molecule. It moves rapidly along microtubules in 8-nm steps but remains attached.182,182a Myosins V and VI are also proces-
sive183–183e but myosin II is not. It binds, pulls on actin,
and then releases it. The many myosin heads interact- ing with each actin filament accomplish muscle con- traction with a high velocity in spite of the short time of attachment. Ncd and Kar3 are also nonprocessive and slower than the plus end-oriented kinesins.184
The ATPase cycles of actomyosin and of the kinesins. The properties of the protein assemblies found in muscle have been described in elegant details, but the most important question has not been fully answered. How can the muscle machinery use the Gibbs energy of hydrolysis of ATP to do mechanical work? Some insight has been obtained by studying the ATPase activity of isolated myosin heads (S1) alone or together with actin. Results of numerous studies of ATP binding, hydrolysis, and release of products using fast reaction techniques185–191 and cryoenzymology191a
are summarized in Fig. 19-18. In resting muscle the myosin heads swing freely in the ~ 20-nm space be- tween the thick and thin filaments. However, in acti- vated muscle some heads are bound tightly to actin as if in rigor (complex A
·
M in Fig. 19-18). When ATP is added MgATP binds into the active site of the myosin (Fig. 19-18, step a) inducing a conformational change to form A·
M*·
ATP in which the bond between actin and myosin is weakened greatly, while that between myosin and ATP is strengthened. The complex disso- ciates (step b) to give free actin and (M*·
ATP), which accumulates at – 15°C. However, at higher tempera- tures the bound ATP is hydrolyzed rapidly (step c) to a form M**·
ADP·
Pi in which the ATP has been cleaved to ADP + Pi but in which the split products remain bound at the active site.116,192,192a,b All of these reac-tions are reversible. That is, the split products can recombine to form ATP. This fact suggests that most of the Gibbs energy of hydrolysis of the ATP must be stored, possibly through a conformational change in the myosin head or through tighter bonding to ATP. As long as calcium ions are absent, there is only a slow release of the bound ADP and Pi and replacement with fresh ATP takes place. Thus, myosin alone shows a very weak ATPase activity.
On the other hand, in activated muscle the head with the split ATP products will bind to actin (step d), probably at a new position. The crossbridges that form appear to be attached almost at right angles to the thin filaments. In step e, Pi is released following a conformational alteration that is thought to open a “back door” to allow escape of the phosphate ion.193
In the final two steps ( f and g) the stored energy in the myosin head (or in the actin) is used to bring about another conformational change that alters the angle of attachment of myosin head to the thin filament from ~ 90° to ~ 45°. At least some indication of such a change can be observed directly by electron microscopy.144
Such a change in angle is sufficient to cause the actin filament to move ~ 10 nm with respect to the thick filaments to complete the movement cycle (Fig. 19-18), if the head is hinged at the correct place. However, the existence of at least four different conformational states suggests a more complex sequence.193a,193b
Examination of the three-dimensional structures avail- able also suggest a complex sequence of alterations in
structure and geometry. X-ray crystallographic struc- tures of myosin heads, in states thought to correspond to states 1 and 3 of Figs. 19-14 and 19-18, are also in agreement on an ~10 (5 – 12) nm movement of the lever arm.194,195 Six states of the actomyosin complex are
depicted in Fig. 19-18, but a complete kinetic analysis requires at least eight and possibly 12 states.196,197
Observing single molecules. A major advance in the study of molecular motors has been the develop- ment of ways to observe and study single macromole- cules. The methods make use of optical traps (optical “tweezers”) that can hold a very small (~ 1 µm diameter) polystyrene or silica bead near the waist of a laser beam focused through a microscope objective.198– 202
In one experimental design an F-actin filament is stretched between two beads, held in a pair of optical traps. The filament is pulled taut and lowered onto a stationary silica bead to which a few myosin HMM fragments have been attached (Fig. 19-19). If ATP is present, short transient movements along the filament are detected by observation of displacements of one of the beads when the actin filament contacts HMM heads. An average lateral displacement of 11 nm was observed. Each HMM head exerted a force of 3 – 4 pN, a value consistent with expectations for the swinging bridge model.200 From the duration of a single displace-
ment (≤ 7 ms) and an estimated kcat for ATP hydrolysis
of 10 s–1, the fraction of time that the head was attached
during one catalytic cycle of the head was therefore only 0.07. This ratio, which is called the duty ratio, is low for actomyosin. However, many myosin heads bind to each actin filament in a muscle. Each head exerts its pull for a short time. but the actin is never totally unattached.203 Similar measurements with
smooth muscle revealed similar displacements but with a 10-fold slower sliding velocity and a 4-fold increase in the duty ratio. This may perhaps account for an observed 3-fold increase in force as compared with skeletal muscle.204,204a
Other single molecule techniques involve direct observation of motor molecules or of S1 myosin frag- ments tagged with highly fluorescent labels.205,206
All measurements of single molecule movement are subject to many errors. Brownian motion of the beads makes measurements difficult.207 Not all results are in
agreement, and some are difficult to understand.207a
Most investigators agree that there is a step size of ~4 –10 nm. Kitamura et al. found 5.3 nm as the aver- age.206 However, they also reported the puzzling obser-
vation that some single S1 molecules moved 11– 30 nm in two to five rapid successive steps during the time of hydrolysis of a single molecule at ATP. They suggested that some of the energy of ATP hydrolysis may be stored in S1 or in the actin filament and be released in multiple steps. Veigel et al.208 observed that a