Kinetic analyses of a truncated mammalian myosin I suggest a novel isomerization event preceding nucleotide binding.

MI(1IQ) is a complex of calmodulin and an epitope-tagged 85-kDa fragment representing the amino-terminal catalytic motor domain and the first of 6 calmodulin-binding IQ domains of the mammalian myosin I gene, rat myr-1 (130-kDa myosin I or MI(130)). We have determined the transient kinetic parameters that dictate the ATP hydrolysis cycle of mammalian myosin I by examining the properties of MI(1IQ). Transient kinetics reveal that the affinity of MI(1IQ) for actin is 12 nm. The ATP-induced dissociation of actin-MI(1IQ) is biphasic. The fast phase is dependent upon [ATP], whereas the slow phase is not; both phases show a Ca(2+) sensitivity. The fast phase is eliminated by the addition of ADP, 10 micrometer being required for half-saturation of the effect in the presence of Ca(2+) and 3 micrometer ADP in the absence of Ca(2+). The slow phase shares the same rate constant as ADP release (8 and 3 s(-)(1) in the presence and absence of Ca(2+), respectively), but cannot be eliminated by decreasing [ADP]. We interpret these results to suggest that actin-myosin I exists in two forms in equilibrium, one of which is unable to bind nucleotide. These results also indicate that the absence of the COOH-terminal 5 calmodulin binding domains of myr-1 do not influence the kinetic properties of MI(130) and that the Ca(2+) sensitivity of the kinetics are in all likelihood due to Ca(2+) binding to the first IQ domain.

Class I myosins in mammals are mechanochemical molecules with an amino-terminal motor domain containing an ATP and actin-binding region, a neck region with one or more socalled IQ domains to which calmodulin binds, and a carboxylterminal tail region (1). One member of the class I myosins, MYR-1, is ubiquitously expressed in mammalian cells. myr-1 contains up to 6 IQ domains; alternate splice forms containing 4, 5, or 6 IQ domains exist (2). The 130-kDa myosin I isolated from rat liver is a myr-1 gene product (3,4). Although quantitation of calmodulin has indicated that this preparation contains 6 mol of calmodulin/130-kDa myosin I heavy chain (4), isoforms corresponding to the 5 IQ and 4 IQ variants are also expressed in liver (2) and may be present.
The 130-kDa myosin I translocates actin filaments slowly and in a Ca 2ϩ -sensitive manner (5). At 10 M free Ca 2ϩ and above, motility is inhibited. This decrease in motility can be reversed by the addition of exogenous calmodulin, indicating that a calcium-induced dissociation of calmodulin might be responsible for the decrease in motility. Laser trap analyses have indicated that the 130-kDa myosin I translocates actin in a two-step process (6). The authors proposed that the two mechanical steps are coupled to P i and ADP release, respectively. Transient kinetic analyses have indicated that the mechanical step coupled to ADP release is unlikely to contribute to force generation or to motility, but could be a system for providing a strain-sensitive ADP release mechanism. This, together with the slow ATP-induced dissociation of actin-myosin I, suggests that this myosin is best suited for maintenance of tension (7).
We have coexpressed in baculovirus calmodulin and a fragment representing the first 728 amino acids of MYR-1, which codes for the amino-terminal motor domain and 1 IQ domain; we refer to this complex as MI 1IQ 1 (8). MI 1IQ translocates actin filaments in vitro. Unlike the parent molecule, the rate of actin translocation is not affected by the Ca 2ϩ concentration over the range of pCa 4 -7 and the rates of movement of MI 130 and MI 1IQ are comparable.
The availability of a homogeneous preparation from the baculovirus expression system has allowed us to explore in detail the kinetics of this construct representing the motor domain and the first IQ domain. Our results indicate that the truncated myosin I possesses kinetic properties indistinguishable from the parent molecule. Thus, the 5 deleted IQ domains and their associated calmodulins play no role in defining the unloaded properties of the myosin I head. Furthermore, our results indicate that, when bound to actin, myosin I exists in two conformations in equilibrium and that ATP can bind to only one of the two conformations. The equilibrium between the two forms is sensitive to both Ca 2ϩ and ADP concentrations and we propose that the conformational change between the two forms may correlate with (i) the double step observed in laser trap studies with this molecule (6) and (ii) the ADP-dependent conformational change identified by electron microscopy for a closely related myosin I, brush border myosin I (9, 10).

EXPERIMENTAL PROCEDURES
Proteins-MI 1IQ was expressed in baculovirus and purified from insect lysates as described in the accompanying article (8). MI 130 was prepared from rat liver as described previously (3). Rabbit skeletal muscle actin was prepared according to (11) and, in some cases, labeled with pyrene at Cys-374 according to Ref. 12.
Transient Enzyme Kinetics-All kinetic experiments were performed at 19.8°C in 20 mM MOPS, 100 mM KCl, 5 mM MgCl 2 , 1 mM dithiothre-itol, and 1 mM EGTA or 1 mM EGTA and 1.1 mM CaCl 2 at pH 7.0 except for the Ca 2ϩ dependence of the ADP release rate, which used 2 mM EGTA and 2 mM CaEGTA mixed in the appropriate proportions to give the required pCa (13,14). All measurements were performed with a Hi-Tech Scientific SF-61 single mixing stopped-flow system using a 100-watt xenon/mercury lamp and a monochromator for wavelength selection. Pyrene fluorescence was excited at 365 nm and emission detected after passing through a KV 389-nm cut-off. Tryptophan fluorescence was excited at 297 nm and observed through a WG 320 filter. The stated concentrations of reactants are those after mixing in the stopped-flow observation cell. Stopped-flow data were fitted to exponentials by a non-linear least-squares curve fit using software provided by Hi-Tech.
The analysis of the titration of MI binding to actin was performed as described by Kurzawa and Geeves (15), where the amplitude of the observed transient is assumed to be proportional to the concentration of actin-myosin complex present before addition of ATP. The data were fitted to the physically significant root of the following quadratic equation.
␣ is the fraction of actin with myosin bound, [M] is the total concentration of myosin added, [A] 0 is the concentration of actin and K d is the dissociation constant of myosin for actin.
Data Interpretation-As shown in Scheme 1, part A, we interpret the kinetics of the interaction of myosin I or its expressed truncated form with nucleotide (T, ATP; D, ADP) in terms of the model described by Bagshaw et al. (16) for conventional myosin (M) where k ϩi , k Ϫi are the forward and reverse rate constants, respectively, and K i (ϭ k ϩi /k Ϫi ) represents the equilibrium constant of the ith reaction. Normal characters are used to indicate reactions involving myosin only, and bold characters refer to reactions in which actin is also present.
ATP binds rapidly to myosin in a two-step reaction before ATP is reversibly hydrolyzed on the protein. This results in a conformational change, which limits phosphate release and the faster ADP release. The ATP-induced dissociation of actin-MI 130 or actin-MI 1IQ and the inhibition of the reaction by ADP have also been interpreted in terms of the models developed for conventional myosin by Millar and Geeves (17) and Siemankowski and White (18). As shown in Scheme 1, part B, ATP binds rapidly and reversibly to actin-myosin and is followed by a rate-limiting isomerization (k ؉2 ) of the complex, which leads to rapid dissociation of actin. ADP competes with ATP for the nucleotide binding site. The dissociation constants of actin for myosin, actin for myosin-ADP, and ADP for actin-myosin are K A , K DA , and K AD , respectively.

RESULTS
The transient kinetics of the interaction of MI 1IQ with actin and ATP were examined and compared with our previous measurements on the native protein, MI 130 (7). In all cases the results were very similar. ATP required for half saturation of k obs (see Fig. 1C). By analogy with other actomyosin systems, the maximal rate and the ATP concentration at half the maximal rate were assigned to k ؉2 , and 1/K 1 , respectively, the rate constant for an isomerization of the actin-myosin complex, which limits the dissociation of actin and the affinity of ATP for the actin-myosin complex (Table I).
Repeating the measurement in the absence of Ca 2ϩ gave similar results, except that the slow component of the transient comprised 50% of the total amplitude with k obs of 3.4 s Ϫ1 and the best fit to the [ATP] dependence of the k obs of the fast component gave k ؉2 ϭ 36 s Ϫ1 and 1/K 1 ϭ 1.2 mM for an apparent second order rate constant (ϪCa 2ϩ ) of 3 ϫ 10 4 M Ϫ1 s Ϫ1 (Fig. 1C).
We had previously observed for MI 130 that the fast phase of the transient was eliminated by incubating the proteins with 20 M ADP before initiating the reaction by mixing with ATP. A similar result was observed here for the expressed myosin I fragment ( Fig. 2A). In this case we were able to titrate the fast phase of the transient, and a plot of the amplitude against [ADP] in the presence and absence of Ca 2ϩ is shown in Fig. 2B. A fit of the amplitude to a hyperbola gave a best fit of 10 and 3 M, respectively, in the presence and absence of Ca 2ϩ , and this is assigned to the affinity of ADP for the A.MI 1IQ complex (K AD ). The k obs of the slow phase remained almost constant over the range of [ADP] used (8.5 s Ϫ1 in Ca 2ϩ ; 2.45 s Ϫ1 , ϪCa 2ϩ ) and was assigned to k ؊AD , the rate of ADP dissociation from A.MI 1IQ .ADP. Since K AD ϭ k ؊AD /k ؉AD , it is possible to calculate k ؉AD , the apparent second order rate constant of ADP binding to A⅐MI 1IQ . This gave values of 0.8 ϫ 10 6 M Ϫ1 s Ϫ1 (ϩCa 2ϩ ) and 1.1 ϫ 10 6 M Ϫ1 s Ϫ1 (ϪCa 2ϩ ), i.e. the apparent second order rate constant of ADP binding is Ca 2ϩ independent and 15-40 fold faster than the apparent second order rate constant of ATP binding to A⅐MI 1IQ .
To verify that the rate of ADP binding is faster than ATP binding, we examined the rates of competitive binding of ADP and ATP to A⅐MI 1IQ . A⅐MI 1IQ at 30 nM was mixed in the stopped flow apparatus with 1 mM ATP and 0 -40 M ADP. The amplitude of the fast phase decreased, whereas k obs increased with increasing [ADP] (data not shown). This demonstrates that ADP effectively competes with ATP even at less than one-tenth of the concentration. If the rate of ADP dissociation from the A⅐M⅐D complex is much less than the rate at which ADP and ATP bind to A⅐M (i.e.
At a fixed ATP concentration, k obs increased from 23 to 42 s Ϫ1 at 20 M ADP and further increased to 73 s Ϫ1 at 40 M. The data over this limited range are therefore compatible with a value of k ؉AD of 0.95-1.25 ϫ 10 6 M Ϫ1 s Ϫ1 in agreement with the estimate above.
The slow phase of the ATP-induced dissociation of A⅐MI 1IQ has a k obs that is very similar to k ؊AD , the rate constant of ADP dissociation from the A⅐MI 1IQ ⅐D complex even in the absence of added ADP. It is possible, therefore, that the slow phase rep-

FIG. 2. Influence of ADP on the ATP-induced dissociation of actin-MI 130 .
A, addition of 2.5 mM ATP to 25 nM pyrene-actin-MI 130 in the absence of ADP resulted in a rapid increase in fluorescence best described by a double exponential (k obs ϭ 43.8 and 8.5 s Ϫ1 and amplitudes of 18.8 and 9.4%, respectively). In the presence of 20 M ADP, the change in fluorescence can be described by a single exponential with a k obs ϭ 10.3 s Ϫ1 and an amplitude of 19.8%; however, the data can be equally described by a double exponential with k obs ϭ 37.7 and 8.5 s Ϫ1 and amplitudes of 3.8% and 17.3%, respectively. B, titration of the amplitude of the fast phase against [ADP] added to the protein before mixing with ATP in the presence and absence of Ca 2ϩ . The data were fitted to a hyperbola in each case, and the apparent affinity (K AD ) for ADP is 10 M in the presence of Ca 2ϩ and 3 M in its absence.
Expressed Truncated Mammalian Myosin I: Kinetics resents a fraction of A⅐MI 1IQ that is isolated with ADP bound. Extensive treatment of the protein with apyrase did not eliminate the slow phase of the reaction. In contrast, if the protein was treated with 20 M ADP such that the fast phase was eliminated, then treatment with apyrase restored the fast phase but only to the same extent as in the original measurement. Thus, apyrase treatment does eliminate the protein bound ADP effectively. We therefore conclude that the slow phase is not caused by the presence of ADP bound to the protein.
Another possibility is that ATP could be the source of contaminating ADP. ATP normally contains about 1% ADP. If ADP binds to the protein faster than ATP (as shown above), then the contaminant ADP could bind a fraction of the protein to produce the slow phase. The true substrate for myosin is MgATP, and the product is MgADP. Since ATP binds Mg 2ϩ more tightly than ADP, reducing the free Mg 2ϩ concentration to a minimum should reduce the contaminant MgADP concentration; however, under limiting Mg 2ϩ concentrations, the slow phase remained constant (data not shown). Furthermore, addition of up to 5% ADP into the ATP had no effect on the amplitude of the slow phase (see above). These results indicate that the slow phase is not due to ADP contamination in the ATP.
The effect of Ca 2ϩ on the rate of ADP release from MI 130 was measured by determining in buffers containing fixed amounts of Ca 2ϩ , the rate of the ATP-induced dissociation of pyrenelabeled actin-MI in the presence of saturating amounts of ADP (Fig. 3). The k obs was plotted as a function of [Ca 2ϩ ], and the line represents the best fit to the Hill equation with midpoint of 6.8 Ϯ 0.6 M Ca 2ϩ and k obs in the presence and absence of Ca 2ϩ of 8.0 and 1.5 s Ϫ1 , respectively. The slope of the graph that defines the Hill coefficient was poorly defined as 3 Ϯ 1.5. These data indicate that binding of Ca 2ϩ to the myosin I complex is positively cooperative.
The binding of ATP to MI 1IQ was followed using intrinsic protein fluorescence and gave results that were indistinguishable from those of MI 130 (data not shown), indicating a second order rate constant for ATP binding of 0.1 ϫ 10 6 M Ϫ1 s Ϫ1 that was independent of Ca 2ϩ . There was no evidence of a slow component of the reaction; however, the signal to noise ratio was very poor in these measurements, and the possibility of a slower component cannot be eliminated.
The affinity of MI 1IQ for actin was measured by monitoring the amplitude of the ATP-induced dissociation of pyrene-labeled A⅐MI 1IQ as a function of [MI 1IQ ] (15). Using 30 nM pA and 400 M ATP, the amplitude of the reaction was measured for MI 1IQ from 10 to 150 nM (Fig. 4). Analysis of the data gave a value of K A of 12 nM, in good agreement with the value obtained for the native protein by a less direct method (7). DISCUSSION Our results demonstrate that MI 1IQ is essentially indistinguishable from MI 130 in terms of its interaction with actin and nucleotide, indicating that the presence of the 5 additional IQ domains and associated calmodulins in the parent molecule has no effect on these properties (Table I). The truncated myosin I shows the same actin-activated Mg 2ϩ -ATPase activity as the parent molecule (8) and the ATP, ADP, and actin binding to A⅐MI 1IQ are identical to the parent. Since the actin and nucleotide binding to MI 1IQ are unchanged, the coupling between actin and nucleotide binding is also expected to remain unaffected by the missing IQ domains and their associated calmodulins. In addition, the Ca 2ϩ sensitivity of all of the above properties is identical to that of the parent molecule, MI 130 . These results demonstrate that the Ca 2ϩ sensitivity of these properties is not a function of the missing calmodulins and must result from Ca 2ϩ binding to the first calmodulin or from a novel independent Ca 2ϩ binding site in the motor domain.
All of the events in the acto-MI ATPase cycle that we have measured are significantly faster than the turnover rate and are therefore not rate-limiting in the ATPase reaction. It is therefore likely that P i release is rate-limiting, as has been observed for other myosins, and in this case the P i release must be regulated by Ca 2ϩ , as previously observed for scallop muscle myosin II (19). A biphasic fluorescent transient was observed upon introduction of ATP to complexes of pyrene-labeled actin with both MI 130 and MI 1IQ . The amplitudes and rates of the two phases were similar for the two proteins, and in both cases the rate constant of the slow phase was very similar to the rate constant limiting ADP release from A⅐MI. To address the possibility that the slow component was due to ADP present either in association with the expressed myosin I or as a trace contaminant in the ATP, several experiments designed to reduce ADP levels were performed. In all cases, the slow phase remained unaltered.
For the native protein we considered the alternative possi-bility that the slow phase in the ATP dissociation reaction was due to the presence of a fraction of damaged myosin I, e.g. missing one or more of the calmodulins, or that a contaminant myosin I could be present and therefore responsible for the biphasic nature of the ADP-induced dissociation. It seems unlikely, however, that the same contaminant would be present in both myosin I preparations (i.e. the native protein isolated from liver and the expressed fragment from insect cells) since they derive from different cell types and involve different purification schemes. We are therefore forced to conclude that this represents an intrinsic property of both the expressed protein and the parent MI 130 . The simplest explanation of the biphasic nature of the ATPinduced dissociation is that the protein exists in two forms: one to which ATP can bind readily (at an apparent second order rate constant of 5.3 ϫ 10 4 M Ϫ1 s Ϫ1 in the presence of Ca 2ϩ ) and the other, which cannot bind ATP without first isomerizing. One possible model is shown in Fig. 5, where the nucleotide pocket must open before nucleotide can bind. (In this model, we have assumed a direct link between the opening of the nucleotide pocket and an ADP-induced structural change, i.e. "wagging" of the myosin neck; see below.) The rate constant for the opening of the pocket is very similar to the net rate constant of ADP dissociation suggesting a similar process limits ADP re-FIG. 5. Proposed model depicting the isomerization of the nucleotide-binding pocket that must occur before ATP or ADP can bind or ADP can be released. The conformational change is represented as a swing of the converter domain of the myosin head with respect to the actin binding domain, which is coupled to the accessibility of the nucleotide binding pocket. The data in Figs. 1 and 2 allow assignment of all of the rate and equilibrium constants. Analysis of the amplitudes of the ATP-induced dissociation reaction in the presence of Ca 2ϩ shows a 60:30 ratio of the two forms. This ratio defines the equilibrium constant between the two forms of A⅐M, K ␣ , with a value of ϳ2.5. Since k ؉␣ ϭ 8 s Ϫ1 (the slow phase of ATP binding) and K ␣ ϭ k ؉␣ /k ؊␣ , then k ؊␣ ϭ 3.2 s Ϫ1 . In the absence of Ca 2ϩ , the amplitudes of the two phases are similar, consistent with the equilibrium lying closer to the closed form of A⅐M with K ␣ ϭ 1-2 and since k ؉␣ ϭ 3.4 s Ϫ1 , k ؊␣ is unchanged by Ca 2ϩ at 3-4 s Ϫ1 . Since (i) the rate constant of ADP dissociation from A⅐M⅐D (k ؉␣D ) and the rate constant of the isomerization of A⅐M (k ؉␣ ) are similar and (ii) they are both reduced 2-3-fold on removal of Ca 2ϩ , it suggests that the two events are closely related and may represent the same isomerization of the A⅐M complex as shown. The displacement of ADP from A⅐M occurs in a single phase and suggests little occupancy of A⅐MЈ⅐D; therefore, in both the presence and absence of Ca 2ϩ , K ␣D Յ 0.1. The affinity of ADP for the complex, K AD , is defined by K ADP ⅐K ␣D and has a value of 10 M in the presence of Ca 2ϩ and 3 M in the absence of Ca 2ϩ . Although K ADP and K ␣D are not defined individually, K ␣D is k ؉␣D /k ؊␣D and therefore K AD ϭ K ADP ⅐k ؉␣D /k ؊␣D or, after rearranging, k ؊␣D /K ADP ϭ k ؉␣D /K AD . Since k ؉␣D has a value of 8 s Ϫ1 (ϩCa 2ϩ ) and 3.4 s Ϫ1 (ϪCa 2ϩ ), the apparent ADP on rate, k ؊␣D /K ADP , has a value of 0.8 ϫ 10 Ϫ6 M Ϫ1 s Ϫ1 in the presence of Ca 2ϩ or 1.1 ϫ 10 Ϫ6 M Ϫ1 in the absence of Ca 2ϩ . This is in good agreement with the directly measured values and is independent of calcium. In this interpretation the rate of the conformational change giving access to the site is Ca 2ϩ -dependent but independent of ADP bound to the pocket, whereas reversal of the conformational change is ADP-dependent and Ca 2ϩ -independent. lease (k ؉␣ ϭ k ؉␣D ϭ 8 s Ϫ1 , Fig. 5). From the data presented in Figs. 1 and 2, we can define the rate and equilibrium constants of each of the transitions shown in Fig. 5. The assignment of the constants is described in the legend to Fig. 5, and the values are listed in Table II.
The evidence points toward a significant proportion (ϳ30%) of the conformation with a closed nucleotide pocket being present in the absence of nucleotide. By increasing the rate constant of pocket closing, ADP causes almost all of the myosin I to be in the closed pocket form. In contrast, the presence of Ca 2ϩ increases the proportion of the open pocket form by increasing the rate of the opening of the pocket in both the presence and absence of ADP. In terms of nucleotide binding, Ca 2ϩ lowers the affinity of A⅐MI for ADP by increasing the net rate constant of ADP release (k ؉␣D ). Ca 2ϩ also increases the rate of ATPinduced dissociation of actin (k ؉2 , Scheme I and Table I) from A⅐MI, but does not affect the affinity of ATP for A⅐MI (K 1 ). Thus, Ca 2ϩ can stimulate the detachment of actin from A⅐MI⅐D by increasing the rate of ADP release and the rate of the subsequent dissociation by ATP. Note, however, that the effects of Ca 2ϩ are small (in all cases, values do not differ by more than a factor of 3) and therefore probably do not represent an on/off switch but rather a modulator of activity.
At physiological nucleotide concentrations of 1 mM ATP and 10 -50 M ADP, the net rates of nucleotide binding to A⅐MI will be 18 s Ϫ1 for ATP and 10 -50 s Ϫ1 for ADP in the absence of Ca 2ϩ . Thus, the A⅐MI that has released ADP is as likely to rebind ADP as to bind ATP. The low efficiency for binding ATP and detaching from actin supports our proposal that myosin I is designed for tension maintenance not motility (7). The presence of Ca 2ϩ increases the net rate of ATP binding (K 1 k ؉2 ) but does not alter the rate of ADP rebinding k ؊␣D /K ADP .
The possibility that myosin I exists in two conformational states, as shown by the kinetic evidence presented here, is of great interest since an isomerization of A⅐MI⅐D can be responsible for the second displacement seen in laser trap assays for MI 130 , brush border myosin I (6), a close relative to MI 130 , and smooth muscle myosin S1 (smS1) (20). It is also required for the structural changes or "tail wagging" seen in three-dimensional reconstructions from electron micrographs for some actomyosins including brush border myosin I and smS1 (9,21). If our assignment is correct, then the identified isomerization provides a direct link between the accessibility of nucleotide to its binding pocket and the mechanical transient and lends strong support for the role of nucleotide release in providing a strainsensing mechanism (22). We previously argued that a MI crossbridge bearing the typical isometric tension could have ADP release reduced up to 100-fold. In the current model, the strain would act directly on the isomerization shown in Fig. 5. Any load or strain on the cross-bridge would inhibit the swing of the tail against the load. The effect would be to reduce k ؉␣D and hence K ␣ .
The Ca 2ϩ sensitivity of the A⅐MI isomerization also establishes a link between the binding of Ca 2ϩ (presumably to the remaining calmodulin), and the biochemical, structural, and mechanical events discussed above. If the protein isomerization is a mechanical sensor that slows down the rates of both ADP release and ATP binding in the presence of strain, then the effect of Ca 2ϩ may be far more dramatic for a head bearing strain. Ca 2ϩ could act on the strained head by binding to calmodulin to alter the elasticity of the calmodulin-IQ complex and so reduce the strain leading to acceleration of the isomerization, ADP release, and cross-bridge detachment. The other 5 calmodulin-IQ domains could also contribute to the Ca 2ϩ -assisted release of strain if they are elastically distorted in a strained cross-bridge. All that is required is for the Ca 2ϩinduced change in calmodulin conformation to alter the rest length of each calmodulin-IQ domain such that the strain on a A⅐MI⅐D head is modulated. It has previously been noted that Ca 2ϩ binding reduces the affinity of calmodulin for the IQ domain but probably not enough to cause calmodulin dissociation at cellular concentrations of calmodulin.
A remaining question about the A⅐MI conformation is its relation to conformational changes in MI alone and to similar structural changes in other myosins. To date we have no evidence for two conformations of MI in the absence of actin and the rate of ATP binding is reasonably fast (0.1 ϫ 10 6 M Ϫ1 s Ϫ1 ); however, the fluorescent signal changes that monitor ATP binding are very small and the presence of a second component cannot be eliminated. smS1 shows a similar mechanical and structural change on binding ADP to A⅐M, yet a close examination of earlier data (22) shows no evidence of a second phase in the ATP-induced dissociation of A⅐smS1. There are two possible explanations for the lack of a second component. Either K ␣ . is much larger (Ͼ10), such that in the absence of ADP the closed conformation is not significantly occupied, or the rate of pocket opening is very fast (k ؉␣ Ͼ 200 s Ϫ1 ), such that the opening can only be observed at very high ATP concentrations. The first possibility is compatible with the smS1 data, where electron micrographs indicate an ADP-induced swing of the myosin neck. The second possibility is formally equivalent to a strain-dependent transition state for nucleotide binding and release proposed for skeletal muscle myosin (23). The proposed model may therefore reflect a property of myosins in general. In this respect it is noteworthy that the neck of the related molecular motor, kinesin, has recently been proposed to adopt various nucleotide-dependent conformations when attached to its microtubule track (24). The interpretation of our data and of the recent kinesin data allows for a strain-sensitive ADP to ATP exchange mechanism. In the kinesin mechanism, this could provide gating of the interaction between the two heads of kinesin. In the case of myosin I, it provides a mechanism for a single-headed molecule to maintain tension with low ATP turnover. If this myosin I clusters on a membrane or vesicle as has been proposed for Acanthamoeba myosin I (25), then a similar gating mechanism could also apply.