Kinetic and Spectroscopic Evidence for Three Actomyosin:ADP States in Smooth Muscle*

Smooth muscle myosin II undergoes an additional movement of the regulatory domain with ADP release that is not seen with fast skeletal muscle myosin II. In this study, we have examined the interactions of smooth muscle myosin subfragment 1 with ADP to see if this additional movement corresponds to an identifiable state change. These studies indicate that for this myosin:ADP, both the catalytic site and the actin-binding site can each assume one of two conformations. Relatively loose coupling between these two binding sites leads to three discrete actin-associated ADP states. Following an initial, weakly bound state, binding of myosin:ADP to actin shifts the equilibrium toward a mixture of two states that each bind actin strongly but differ in the conformation of their catalytic sites. By contrast, fast myosins, including Dictyosteliummyosin II, have reciprocal coupling between the actin- and ADP-binding sites, so that either actin or nucleotide, but not both, can be tightly bound. This uncoupling, which generates a second strongly bound actomyosin ADP state in smooth muscle, would prolong the fraction of the ATPase cycle time that this actomyosin spends in a force-generating conformation and may be central to explaining the physiologic differences between this and other myosins.

Although all myosins share basic structural and enzymatic features, they differ widely in their physiologic behavior. Comparing myosin II and myosin V provides an illustrative example (1,2). Myosin V is designed to work as a single, processive motor, whereas myosin II is designed to work in a large, filamentous ensemble. Even within the myosin II family there are major differences in shortening velocity, economy of force production, and in vitro motility between fast myosins, such as those from vertebrate skeletal muscle, and slow myosins, including vertebrate smooth muscle and non-muscle myosin II (3). Because all myosins appear to use the same general kinetic scheme for their ATPase cycles, differences in their properties must be reflected in corresponding differences in the rate and equilibrium constants that describe the individual kinetic schemes.
During the course of its ATPase cycle, myosin alternates between conformations that bind actin weakly and those that bind actin strongly (4,5). The strong binding conformations generate force, and the lifetime of the strong binding state limits the rate of actin-myosin filament sliding. One of the major differences between myosin isoforms is in the duty ratio, which is defined as the fraction of the total cycle time that a myosin isoform spends in a strong binding conformation (6,7). Such differences have been described between fast skeletal and smooth muscle myosin II and include an increased duty ratio, as well as slower rates of hydrolysis (8) and ADP release (9) for smooth muscle myosin II.
An increase in the duty ratio of smooth muscle myosin II implies a longer lifetime of the strong binding state, which under physiologic conditions means a longer lifetime of ADPbound state(s) (4). This could be accomplished either by adding an additional, unique actomyosin:ADP state for smooth muscle myosin or by slowing the ADP release step. Information relevant to this issue has been obtained from structural studies of gizzard S1. Whittaker et al. (10) produced three-dimensional density maps of nucleotide-free and ADP-bound chicken gizzard acto-S1, generated by cryo-electron microscopy and helical image reconstruction. These maps demonstrated that ADP release induces an additional 23 o tilt of the regulatory domain. Similar conclusions were also reached from studies utilizing EPR spectroscopy and x-ray fiber diffraction (11,12), as well as from studies with brush border myosin I (13). Interestingly, this movement is not seen in the corresponding skeletal muscle isoform (10). These findings have been interpreted to mean that ADP release contributes to the power stroke in a subset of myosin II isoforms, as well as in some of the unconventional myosins (13,14), perhaps through the presence of an additional structural transition not present in skeletal muscle myosin II.
However, several other lines of evidence call into question the significance of this extra tilt. Cremo and Geeves (15) examined the equilibrium binding and kinetic properties of smooth muscle S1:ADP 1 with actin. They concluded that the tight affinity of ADP for acto-S1 (K d ϭ 5 M) would make ADP release energetically unfavorable under physiologic conditions and therefore could not provide free energy to drive rotation of the regulatory domain. Furthermore, equilibrium binding data were used to argue that the thermodynamic coupling between ADP and actin binding to S1 was appreciably lower in smooth muscle than in skeletal muscle myosin, which suggested a strain-dependent mechanism would be required to drive ADP release. Dantzig et al. (16) examined the effects of MgADP on rigor tension in gizzard smooth muscle and found that, contrary to expectation, its addition did not reduce tension. They postulated that the additional movement of the regulatory domain may contribute to the latch state, by slowing release of ADP from attached heads.
The presence of an additional tilt in the mechanism of smooth muscle myosin II implies that there is an additional actomyosin ADP intermediate in the pathway of nucleotide dissociation. If this is so, the rate of transition through this intermediate might contribute to economy of force production by smooth and non-muscle myosin II isoforms by controlling the rate of ADP release. In order to evaluate this, we have examined the interactions of ADP with smooth muscle S1 and acto-S1 with transient fluorescent and kinetic methodologies. We have also examined myosin II from Dictyostelium, which has been the subject of extensive structural and biochemical characterization (17)(18)(19). Although we find evidence for only one myosin:ADP state for myosin II from Dictyostelium, corresponding studies on smooth muscle myosin II clearly demonstrate the presence of two such states, both of which are on the pathway to ADP release. The relevance of these results to models of strain-dependent release will be discussed.

EXPERIMENTAL PROCEDURES
Reagents-The N-methylanthraniloyl derivative of 2Јdeoxy-ADP was synthesized as described (20). N-1-pyrenyl iodoacetamide was obtained from Molecular Probes (Portland, OR). Protease inhibitors and chemicals used for buffers were obtained from Sigma. Sephacryl S-300-HR and prepoured Sephadex G-25 columns were obtained from Amersham Pharmacia Biotech.
Proteins-Dephosphorylated smooth muscle myosin was obtained from chicken gizzards (21). Actin was prepared from rabbit acetone powder as described (22). S-1 was prepared by digestion of myosin with Staphylococcus aureus V8 protease and column purified on Sephacryl S-300-HR (23).
Recombinant Proteins-All myosin heavy chain constructs were subcloned into the baculovirus transfer vector, pVL 1393 (Invitrogen). Generation of the recombinant baculoviruses coding for the myosin heavy chain constructs followed published techniques (24). Expression of these constructs with the smooth regulatory and essential (LC 17a ) light chains and subsequent purification of the recombinant myosins followed published procedures (25).
To generate the recombinant Dictyostelium myosin S1, the cDNA for Dictyostelium myosin II was truncated at the codon corresponding to lysine 833 to create an S1-like fragment. A Flag peptide sequence (35) followed by a stop codon was appended. The S1-like construct and the Dictyostelium cDNAs encoding for the regulatory and essential light chains were subcloned into baculovirus transfer vectors (pVL 1392/93, Invitrogen). Expression of the truncated heavy chain with the regulatory and essential light chains and subsequent purification of the recombinant Dictyostelium myosin followed published procedures (30).
Fluorescence Methodologies-Steady state fluorescence measure-ments were made on an SLM/Aminco 8000C fluorescence spectrophotometer, with sample holder thermostated at 20°C. Fluorescence lifetimes were measured using a picosecond laser as excitation source, as described (27) in samples that had been equilibrated in 20 mM KCl, 25 mM Hepes, 1 mM MgCl 2 , 1 mM dithiothreitol, pH 7.50. Kinetic Methodologies-Measurements of the rates of mant nucleotide and pyrenyl actin binding to and release from S1 were made using a Hi-Tech Scientific stopped-flow spectrometer equipped with a 150watt xenon lamp. The instrument dead time was determined to be 1.4 ms. For studies with mant nucleotide, fluorescence was excited by energy transfer from protein tryptophan, using a 295 nm input from a monochrometer, and a 400 nm cut filter was used to measure the fluorescence emission at a right angle from the incident beam. Data were fit to a sum of exponentials. For studies using pyrene actin fluorescence, the excitation wavelength was 365 nm, and a 405 nm interference filter was used to collect the fluorescence emission (Omega Optical).
Protein Purification and Concentration Assays-Proteins were run on SDS-polyacrylamide electrophoresis as described (28). Protein concentrations were determined colorimetrically (Bio-Rad Protein Assay).
Electron Microscopy and Image Analysis-Recombinant S1, derived from Dictyostelium and smooth myosin II, at 2-5 mg/ml in 10 mM imidazole, pH 7.0, 50 mM KCl, 2 mM MgCl 2 , 0.5 mM EDTA, 0.1 mM EGTA, 0.5 mM dithiothreitol was used to decorate rabbit skeletal muscle actin filaments adsorbed to holey carbon support films on EM grids. For the nucleotide-free condition, apyrase was added to the solution (final concentration ϳ10 IU/ml) on the grid just prior to blotting and freezing in liquid ethane slush. For the ADP condition, the myosin buffer was supplemented with 5 mM MgADP. A Gatan cold stage, operating at Ϫ180°C, was used to examine the grids in a Philips CM200 FEG electron microscope. Images of filaments spanning holes in the support film were recorded at 40,000ϫ at 1.1-2.0 m under focus. Selected images of well ordered filaments were digitized with spot and step sizes equivalent to 4.97 Å at the specimen. Image analysis, averaging, and calculation of three-dimensional maps were carried out essentially as described previously (29). The three-dimensional maps were visualized with Volvis.

Transient Fluorescence Measurements of Myosin:2ЈdmD-
We had previously shown that the fluorescence of recombinant smooth muscle S1 complexed with the fluorescent ADP analogue 2ЈdmD could be characterized by two lifetimes, suggesting that smooth S1:ADP consists of an equilibrium mixture of two states (30). In order to examine this more closely, we prepared proteolytic gizzard S1, and we measured the time-dependent fluorescence decays of complexes with 2ЈdmD as a function of temperature. The large concentrations of protein required for these measurements precluded our performing this experiment with recombinant S1. At all temperatures, the decay data could be adequately fitted to the sum of two exponentials, yielding two lifetimes. At 20°C, the lifetimes of these components were 7.95 Ϯ 0.30 and 0.81 Ϯ 0.10 ns, with fractional amplitudes of 0.72 and 0.28, respectively (Table I). Neither of these lifetimes represents unbound nucleotide, since the lifetime of unbound mant-ADP is 3.8 Ϯ 0.2 ns (27), and at the concentration of S1 used in this study (110 M), less than 1% of TABLE I Fluorescence lifetimes and equilibrium constants for complexes of S1 and acto-S1 with 2ЈdmD The conditions were 20 mM KCl, 25 mM Hepes, 1 mM MgCl 2 , 1 mM dithiothreitol, pH 7.5.

Sample
Source where 1 and 2 are the two lifetimes with two associated fractional amplitudes ␣ 1 and ␣ 2 (31). The two fractional intensities provide a measure of the mole fractions of the two components. Thus, the ratio of f 2 /f 1 is an equilibrium constant describing the equilibrium between these two S1:ADP states. We measured the fluorescence decay parameters of proteolytic S1:2ЈdmD as a function of temperature, and Fig. 1 shows a van't Hoff plot depicting ln(f 1 /f 2 ) versus the reciprocal of temperature, in degrees Kelvin. The component with longer lifetime is favored at all temperatures. The data fit a two-state equilibrium with ⌬H 0 ϭ 10.6 kcal/mol and ⌬S 0 ϭ 29.3 cal/ (degrees⅐mol). At 20°C, f 2 /f 1 0.045, and at 10°C, f 2 /f 1 ϭ 0.011 (Table I). We have previously reported that the fluorescence decay of 2ЈdmD complexed with recombinant S1 likewise consists of two components, with lifetimes of 8.73 Ϯ 0.2 ns and 1.8 Ϯ 0.1 ns and fractional components of 0.70 and 0.30, respectively (30). This would correspond to f 2 /f 1 ϭ 0.075 at 20°C (Table I).
Similar studies using recombinant S1 from Dictyostelium myosin II, a fast myosin, also revealed two components. However, one of these had a lifetime of 3.82 Ϯ 0.20 ns, identical to free nucleotide, and its fractional emission of 0.30 was nearly identical to the fraction of free nucleotide determined by equilibrium dialysis (0.27). The other, corresponding to bound nucleotide, had a lifetime of 8.31 Ϯ 0.30 ns (Table I). We therefore conclude that the short lifetime component seen with smooth muscle myosin represents an additional myosin ADP state not seen in fast myosin II isoforms.
We next repeated these measurements for a complex of 12.4 M proteolytically prepared smooth muscle S1, 12.4 M 2ЈdmD, and 25 M phalloidin-stabilized actin. The lower concentration of S1 was used to reduce the turbidity of the acto-S1 mixture. Given the dissociation constant of S1:ADP for actin in the steady state (15), essentially all of the S1 should remain bound. We measured the concentration of 2ЈdmD not bound to the acto-S1 complex by sedimenting the complex in an Airfuge and measuring the absorbance of the supernatant at 356 nm. This yielded an unbound nucleotide concentration of 6.8 M, corresponding to an unbound fraction of 0.48 and a dissociation constant of 5.7 M. The fluorescence decay of this complex at 20°C could be described by three exponential terms with lifetimes of 7.97 Ϯ 0.3, 3.97 Ϯ 0.2, and 0.28 Ϯ 0.1 ns and fractional intensities of 0.17, 0.60, and 0.23, respectively (Table I). The fractional intensity of 0.60 is similar to the fraction of unbound nucleotide, and the lifetime for this component is likewise very close to that for free nucleotide. Furthermore, the fluorescence decay of a mixture of 12 M 2ЈdmD and 20 M phalloidinstabilized actin yielded a single lifetime of 3.87 Ϯ 0.2 ns, and we therefore conclude that the component of the acto:S1 complex with lifetime of 3.97 ns is free nucleotide. The other two lifetime components, representing two acto:S1:2ЈdmD states, are related by an equilibrium constant f 2 /f 1 ϭ 1.32. Repeating this experiment at 10°C again produced three components, with lifetimes of 6.98 Ϯ 0.3, 4.23 Ϯ 0.2, and 0.30 Ϯ 0.1 ns and fractional intensities of 0.25, 0.61, and 0.14 ( Table I). This corresponds to an equilibrium constant of 0.56 between the two acto:S1:2ЈdmD states. These results thus demonstrate that actin binding shifts the equilibrium distribution of the two S1:ADP states toward the shorter lifetime component by a factor of 30 -50.
Repeating these experiments with a complex of 6.0 M recombinant smooth S1 ϩ 6.0 M 2ЈdmD, and 40.0 M phalloidinstabilized actin also revealed the presence of three components in the fluorescence decay. At 20°C, the lifetimes were 7.82 Ϯ 0.3, 3.70 Ϯ 0.2, and 0.72 Ϯ 0.2 ns with fractional intensities of 0.23, 0.63, and 0.14, respectively. The corresponding lifetimes and fractional intensities at 10°C were 7.58 Ϯ 0.3, 3.99 Ϯ 0.2, and 0.67 Ϯ 0.2 ns and 0.25, 0.70, and 0.05. The component with lifetime of 3.70 Ϯ 0.2 ns, with unbound fraction at 20°C of 0.63, is considered free nucleotide, since its fractional intensity is in good agreement with direct measurements of unbound nucleotide that were obtained by sedimenting the complex in an Airfuge. This revealed a free nucleotide concentration of 4.1 M, corresponding to an unbound fraction of 0.68 and dissociation constant of 8.3 M. Measurement of actin-bound S1:ADP was performed with a sedimentation assay using 3 H-labeled S1, as described (8), and revealed an unbound fraction of Ͻ0.07. Thus, lifetime studies of recombinant S1:2ЈdmD:actin also reveal evidence of two states that are related by an equilibrium constant of 0.20 at 10°C and 0.61 at 20°C (Table I).
These results demonstrate the presence of two S1:ADP and two acto-S1:ADP states in smooth muscle, as defined by a probe in the nucleotide-binding site. In order to determine how these states relate to actin binding affinity, we next performed a series of transient state kinetic measurements with pyrenelabeled actin.
Kinetic Studies of Recombinant S1:ADP Binding to Pyrene Actin-The formation of a strongly bound actomyosin state can be monitored by using a pyrenyl probe on actin, since strong binding of S1 to actin quenches the pyrene fluorescence (32). Several investigators have utilized this to demonstrate that binding of ADP to acto:S1 has no effect on the pyrene fluorescence for both smooth muscle myosin II and brush border myosin I (14,15). We found similar results with recombinant S1 bound to phalloidin-stabilized pyrene-labeled actin (data not shown), and we conclude from this that the two acto:S1: ADP states detected from fluorescence lifetime studies are both in a strong actin-binding conformation. It does not automatically follow, however, that the predominant S1:ADP state detected from lifetime studies retains this strong binding conformation. This is supported by the results depicted in Fig. 2. In this experiment, a 10-fold molar excess of pyrene-labeled actin was mixed with recombinant S1 ϩ 2 mM ADP at 20°C in the stopped flow. The quenching of the pyrenyl probe followed a single exponential process following a lag phase (Fig. 2, inset), as has been described with skeletal S1 (32). The rate of the quenching varied hyperbolically with actin concentration, with an initial slope defining an apparent second order rate constant of 3.4 M Ϫ1 s Ϫ1 . Fitting to a hyperbolic dependence on actin concentration yielded an apparent binding constant of 2.4 M and a maximum rate of 6.9 s Ϫ1 (Fig. 2). The kinetics of this reaction are consistent with the following Reaction 1, originally described for skeletal muscle S1 (32), where A* is pyrene actin with unquenched fluorescence; A is quenched pyrene actin; D is ADP; and M is myosin. This implies that, as in the case of skeletal muscle S1, the initial M⅐D complex is in a weak actin binding conformation. Formation of the M⅐D⅐A* complex, associated with the lag phase, is then followed by an isomerization to strong binding complex (M⅐D⅐A) whose fluorescence is quenched.
We next examined the kinetics of ADP dissociation from acto:S1 by mixing pyrene actin ϩ recombinant S1 ϩ ADP in the stopped flow with 4 mM ATP. At 20°C, the resulting fluorescence enhancement occurred as a single exponential process whose rate constant showed a small concentration dependence, and over an ADP concentration range of 20 -200 M was between 61 and 77 s Ϫ1 (data not shown). Repeating the experiment at 10°C produced a fluorescence transient that was best fit by two exponential terms. The ratio of the amplitudes of these two phases remained constant at 0.20 -0.25 in favor of the slower phase, over an ADP concentration range of 30 -300 M. The rate constant of the faster process demonstrated an inverse relationship with [ADP], whereas that for the second phase showed little dependence on [ADP] and was in the range of 6 -10 s Ϫ1 (Fig. 3A). The dependence of the rate of the faster phase on [ADP] could be described by Equation 2: where obs is the observed rate constant; K ATP is the apparent binding constant of ATP to the acto-S1 complex; k d is the rate constant of acto-S1 dissociation produced by mixing with ATP at concentration [ATP]; and K ADP is the dissociation constant for ADP binding to the acto-S1 complex. These values are summarized in Table II. A similar result had been reported for brush border myosin I at room temperature (14). We had previously reported that loop 1, a surface-exposed segment of random coil bridging the catalytic site and connecting the 25-and 50-kDa domains, modulates the rate of ADP release and that its deletion reduced the rate of ADP release by nearly 10-fold (30). We repeated the above experiments with a mutant of smooth muscle S1 in which this loop had been deleted (⌬25/50 S1), and we found that at 20°C, it produced a biphasic fluorescence transient with roughly equal amplitudes. Furthermore, the rates for the two phases demonstrated an ADP concentration dependence very similar to wild type S1 at 10°C (Fig. 3B). The rate of the faster phase of the transient for ⌬25/50 S1 could likewise be fit to Equation 2, and results are summarized in Table II. A double exponential decay would be expected if release of ADP occurs subsequent to an isomerization reaction, as has been described for brush border myosin I (14). For smooth muscle S1, this isomerization can be detected by either reducing the temperature or by eliminating loop 1, both of which appear to slow this isomerization. The fluorescence lifetime data discussed above indicate that both acto-S1:ADP states are populated at 20 o C. Thus, the presence of only a single phase in the fluorescence transient for native S1 at 20°C implies that at FIG. 2. Apparent rate constant for the binding of recombinant S1:ADP to pyrene-labeled actin, as a function of actin concentration. Conditions as in Fig. 1, 20°C. Actin:S1 concentration ratio Ն10:1. Rate constants were determined by fitting to a single exponential decay following a lag, and the solid line is fitted to a two-step reaction in which formation of a non-quenched weakly bound complex is followed by formation of a strongly bound complex, as described for skeletal muscle S1 (32). Fitting to a hyperbolic dependence on actin concentration yielded an apparent binding constant of 2.4 M and a maximum rate of 6.9 s Ϫ1 . Inset, fluorescence transient produced by mixing 1 M S1:ADP with 10 M actin, demonstrating the lag and subsequent fluorescence decay. Smooth curve is a fit to a single exponential decay following a lag phase. this temperature, these two states equilibrate at least as rapidly as the rate of ADP release. In order to evaluate this further, we next examined the kinetics of 2ЈdmD binding to and release from recombinant acto-S1.
Kinetic Studies of Binding and Release of 2ЈdmD from Acto-S1-Mixing of recombinant S1:mant-ADP with actin ϩ ATP at 20°C is associated with a single exponential fluorescence decay with maximum rate greater than 30 s Ϫ1 (30). In order to slow the rate of interconversion of the two S1:ADP states, we repeated these experiments at 10°C. At this temperature, re-lease of 2ЈdmD from S1 was associated with a two phases in the fluorescence transient with rates of 4.5 and 1.2 s Ϫ1 and relative amplitudes of 0.08 and 0.92, respectively. Mixing S1:2ЈdmD with actin ϩ ATP at 10°C also produced a biphasic transient, and the rate constants of both phases were best fit to a hyperbolic dependence on actin concentration (Fig. 4). The maximum rates of the two phases were 30.4 and 3.0 s Ϫ1 , with dissociation constants of 2.5 and 23.8 M, respectively (Table III). The amplitude of the faster phase constituted 4 -6% of the total amplitude. These results are consistent with those using pyrene actin at this temperature (above) and suggest that at 10°C two acto-S1:ADP states are in a slow equilibrium relative to the rate of ADP dissociation. It also suggests that there is a roughly 10-fold difference in affinities for these two S1:ADP states with actin.
We next examined the kinetics of 2ЈdmD release from a preformed acto-S1 complex at 10 and 20°C. In this experiment 3 M recombinant S1 ϩ10 M phalloidin-stabilized actin ϩ 30 M 2ЈdmD were mixed in the stopped flow with 2 mM ATP. Given our measured value for the dissociation constant under steady state conditions, essentially all of the S1 should remain bound to actin. The resulting fluorescence transient, monitored by energy transfer, demonstrated a single exponential process at both temperatures. At 10°C, the rate constant was 26.2 Ϯ 3.7 s Ϫ1 , nearly identical to the faster rate produced by mixing S1:2ЈdmD with actin. At 20°C, the rate was 96.4 Ϯ 10.7 s Ϫ1 . This experiment was repeated with 20 mM phosphate added to the acto-S1:2ЈdmD-containing syringe, and the rate at 10°C, 25.2 Ϯ 4.6 s Ϫ1 , was essentially the same as in the absence of phosphate.
Binding of 2ЈdmD to a rigor acto-S1 complex was examined at 10°C in the stopped flow by mixing varying concentrations of 2ЈdmD with 3 M recombinant S1 ϩ 12 M actin ϩ 1.0 unit/ml apyrase, the latter to ensure the acto:S1 complex is  Table III. Conditions are as in Fig. 1.  3. Rates of the two phases of the fluorescence transient produced by mixing pyrene actin ؉ S1 ؉ ADP with 4 mM MgATP, plotted as a function of [ADP], for wild type S1 at 10°C (A) or ⌬25/50 S1 at 20°C (B). A, the rate constant of the faster phase (closed squares) could be fit to Equation 2, yielding values for K ATP , k d , and K ADP summarized in Table II, whereas the slower phase (open squares) showed little ADP concentration dependence and remained in the range of 6 -10 s Ϫ1 . B, the rate constant for the faster phase (closed squares) could be fit to Equation 2, and values of K ATP , k d , and K ADP are summarized in Table II. The rate of the slower phase remained relatively constant at 5-7 s Ϫ1 . Conditions are as in Fig. 1

TABLE II
Kinetic parameters for the release of ADP from recombinant S1:pyrene actin and ⌬25/50 S1:pyrene actin nucleotide-free. The rate of the fluorescence enhancement produced by excitation at 295 nm was monophasic, and a plot of rate versus nucleotide concentration was linear in the 0 -15 M range (Fig. 5). This defines an apparent second order rate constant of 2.9 M Ϫ1 s Ϫ1 , a dissociation rate constant of 12.1 s Ϫ1 , and a dissociation constant of 4.2 M. As noted above, the fluorescence decay of complexes of 2ЈdmD with S1 from Dictyostelium myosin II is monoexponential, indicating the presence of only one S1:ADP state. This is consistent as well with the kinetics of 2ЈdmD release, measured by mixing Dictyostelium myosin II:2ЈdmD with an excess of actin ϩ 4 mM ATP in the stopped flow. Release of nucleotide at both 10 and 20°C occurred in a single phase, with rate constants demonstrating a linear dependence on actin concentration (Fig. 6).
Cryo-EM Maps of Dictyostelium Myosin II-To test further the assertion that the second strongly bound ADP state in smooth muscle is linked to the movement observed in cryo-electron microscopy, we calculated three-dimensional maps of actin decorated with recombinant Dictyostelium myosin II, with and without MgADP present. The resulting maps are shown in Fig. 7 and are compared with previously generated maps for smooth muscle myosin II (10). Consistent with the kinetic data, there is no movement of the light chain domain of Dictyostelium myosin II upon ADP release. DISCUSSION Previous studies of the interaction of smooth muscle S1 with ADP and actin have been interpreted in terms of a kinetic scheme that invokes the existence of one S1:ADP state (15). However, several results in the literature suggest that the properties of smooth muscle myosin:ADP can only be explained by assuming the existence of several such states. First, there is an apparent discrepancy between the binding constant of smooth muscle S1:ADP for actin, measured by equilibrium techniques (15), and that determined from transient kinetic measurements (30). This could be most readily explained by the presence of an additional S1:ADP state, not seen in skeletal FIG. 5. Apparent rate constant for binding of 2deoxy mant-ADP to nucleotide-free acto:S1, as a function of nucleotide concentration, 10°C. Conditions as in Fig. 1. Binding was examined by mixing varying concentrations of 2ЈdmD with 3 M recombinant S1 ϩ 12 M actin and was monitored by energy transfer from an S1 tryptophan to the mant fluorophore. The linear slope of the plot of rate versus nucleotide concentration defines an apparent second order rate constant of 2.9 M Ϫ1 s Ϫ1 , a dissociation rate constant of 12.1 s Ϫ1 , and a dissociation constant of 4.2 M. This compares to a calculated value of 6.1 M at 10°C from results with pyrene actin (Fig. 3A and Table II

FIG. 7. Comparison of ADP (blue) and nucleotide free (red) three-dimensional wire-frame maps of Dictyostelium myosin II (A) and smooth muscle myosin II (B) attached to F-actin.
All maps are at the same scale. Note that there is no significant difference between the two Dictyostelium maps in A. However, as previously published (10), the light chain region (lever arm) undergoes a rotation of approximately 20 -25°for smooth muscle myosin II upon ADP release. The label motor is adjacent to the myosin II motor domain, into which has been inserted the ␣-carbon backbone (yellow) derived from the crystal structure (17). ELC denotes the essential light chain and RLC the regulatory light chain of myosin II, which together constitute the effective "lever arm." Note that the RLC density is weak, especially in A, which may be indicative of light chain mobility. or other fast myosins, and could also explain the additional movement of the regulatory domain seen with ADP release (10,11). In this study, we have examined this issue by investigating the interaction of S1:ADP with actin using both spectroscopic and kinetic probes.
Fluorescence lifetime studies of complexes of 2ЈdmD with S1 demonstrate the existence of two S1:ADP states. When bound to actin, the fluorescence decay of S1:2ЈdmD also demonstrates the presence of two states, and the finding that adding ADP to pyrene actin:S1 does not alter the fluorescence of the pyrenyl probe indicates that both of these states bind actin strongly. Nevertheless, the hyperbolic dependence of the rate of binding of S1:ADP to pyrene actin (Fig. 2) suggests that when dissociated from actin, the vast majority of S1:ADP is in a conformation that binds actin weakly (32,33). Previous studies, using spectroscopic probes of skeletal muscle S1:ADP, were also interpreted to suggest the existence of two S1:ADP states (31). Our results from this study, in conjunction with these previous studies, therefore suggest a model for smooth muscle myosin interactions with ADP and with actin as shown in Reaction 2, where A is actin, M is myosin, D is ADP, K represents an equilibrium constant, k is a rate constant, and where K 1 ϭ k 1 /k -1 ϭ M S D W /M S D S , etc. The species highlighted in boldface would be the predominant ones at equilibrium, based on the calculated values of the various equilibrium constants (see below). In Reaction 2, each myosin and actomyosin state can be characterized by the conformations of the nucleotide and actinbinding sites. We propose that these two sites can each assume one of two conformations, "strong" binding and "weak" binding. The state of the actin-binding site (S indicates strong, and W indicates weak) is indicated by the subscript next to the symbol for myosin, while the state of the nucleotide-binding site is indicated by the subscript next to the symbol for ADP. Tight coupling between the conformations of the nucleotide and actin-binding sites would produce two actomyosin states, such as is seen in skeletal muscle, whereas looser or no coupling could produce up to four such states. This is reminiscent of the model of Cremo and Geeves (15), in which strain-dependent nucleotide release was proposed to drive ADP dissociation from a system that had relatively low thermodynamic coupling. In such a loosely coupled system, the affinity of a myosin species for actin would solely be determined by the state of its actin-binding site. Hence, K b ϭ K c and therefore K 1 ϭ K 1 Ј. As noted above, the lack of a fluorescence change when nucleotidefree pyrene actin:S1 is mixed with ADP indicates that K 0 Ј Ͼ Ͼ 1 and AM W D S is not populated in the steady state. This is also consistent with the finding that added phosphate did not affect the kinetics of 2ЈdmD dissociation from acto:S1. Hence, the equilibrium constant measured from the fluorescence decay of 2ЈdmD bound to acto-S1 (Table I) is a measure of K 1 and K 1 Ј. The fluorescence lifetime-determined equilibrium constant, f 2 /f 1 (Table I), is a measure of M S D W /(M S D S ϩ M W D S ). Hence, the value of this equilibrium constant can be used to solve for K 0 and K 1 . These are 0.14 and 0.61, respectively, for recombinant S1:ADP at 20°C and 0.03 and 1.3 for proteolytic S1:ADP at the same temperature. Thus, for smooth muscle myosin: ADP, the predominant state would be a weak actin-binding, strong nucleotide-binding state (M W D S ) that would be equiva-lent to that proposed by Sleep and Hutton (33) for skeletal muscle myosin. However, a small fraction would be in a weak nucleotide-binding, strong actin-binding conformation (M S D W ). If K 1 is a slow equilibrium at 10°C relative to the binding reactions with actin (K a and K b ), then mixing S1:dmD with actin at this temperature could produce a biphasic fluorescence transient (Fig. 4). Binding to actin would shift the equilibrium almost completely over to the two strong actin-binding states (e.g. K 0 Ј Ͼ Ͼ 1), one whose ADP affinity is high (M S D S ) and one with lower ADP affinity (M S D W ).
While this study did not directly measure K 0 Ј, a lower limit of 10 appears reasonable, given the sensitivity of the pyrenyl probe on actin to detect the AM W D S state. The quenching of pyrene fluorescence produced by binding of S1:ADP to pyrene actin would therefore be expected to proceed in a two-step process as shown in Reaction 3, The apparent binding constant for the hyperbolic dependence of rate on actin concentration provides a measure of K a of 0.42 M Ϫ1 at 20°C (Fig. 2). Given that K 0 ϭ 0.14 and that K b ϭ K a ⅐K 0 Ј/K 0 , this would predict K b ϭ 33 nM, which is remarkably close to the value measured by Cremo and Geeves (24 nM,Ref. 15).
Two rate processes were seen in the release of 2ЈdmD from S1 at 10°C, whereas we had previously reported only one process was seen at 20°C (30). In neither case was an appreciable lag phase seen. These findings could be explained by assuming that K 0 is a rapid equilibrium relative to the subsequent steps and by assigning the MD S state to that with the longer lifetime component. The resulting fluorescence transient for 2ЈdmD dissociation would be predicted to fit a double exponential decay. The apparent rate constants, 1,2 are described by Equations 3 and 4 (9), The solution of the rate equations must reasonably agree with the calculated value of K 1 ϭ k 1 /k -1 , as well as explain the observed rates. These requirements can be satisfied for recombinant S1 by assigning values of k 1 ϭ 0.3 s Ϫ1 , k -1 ϭ 0.8 s Ϫ1 , and k 2 ϭ 4 s Ϫ1 at 10°C; and k 1 ϭ 3 s Ϫ1 , k -1 ϭ 2 s Ϫ1 , and k 2 ϭ 4 s Ϫ1 . At the lower temperature, this would produce a biphasic transient with rates of approximately 5 and 0.6 s Ϫ1 . Two rate processes would theoretically also be observed at 20°C, although the rate constants would be approximately 2.5 and 6 s Ϫ1 and could appear as a single exponential process. These results thus suggest a strong temperature dependence on k 1 , which is consistent as well with the effect of temperature on K 1 ( Fig. 1 and Table I). This implies that this transition represents a concerted reaction, such as would be seen with a large conformational change (see below). Solution of the rate equations for 2ЈdmD release from actomyosin takes the same form as that for myosin, although each of the rate constants now represents an averaged rate constant whose value depends hyperbolically on actin concentration, as described previously for skeletal muscle S1 (9). Since K 0 Ј Ͼ Ͼ 1, the AM W D S state would not be populated and would not con-tribute to the fluorescence transient. At 10°C, a biphasic fluorescence transient was observed, and at 20°C, fitting of the transient could be accomplished with a single exponential function. This could be explained by setting k 1 Ј ϭ 3 s Ϫ1 , k -1 Ј ϭ 9 s Ϫ1 , and k 2 Ј ϭ 15 s Ϫ1 at 10°C. This would predict rates of 1.5 and 25.2 s Ϫ1 , which are very close to the extrapolated maximum rates at this temperature (Table III). It would also predict that the rate of 2ЈdmD dissociation from acto-S1 at 10°C determined by extrapolation (12.1 s Ϫ1 , Fig. 5) would differ from that measured by mixing acto-S1:2ЈdmD with ATP (26.2 s Ϫ1 ) since the former is largely determined by the value of k 2 . At 20°C, a single phase would be expected if 1 and 2 differed by less than a factor of 3, and this could be accomplished by setting k 1 Ј ϭ 30 s Ϫ1 , k -1 Ј ϭ 15 s Ϫ1 , and k 2 Ј ϭ 45 s Ϫ1 . Both kinetic and fluorescence lifetime data suggest that k 1 Ј has a large temperature dependence, which would be expected if this transition were associated with the tilting of the regulatory domain.
This model is also consistent with results using pyrene actin. The lack of fluorescence change after mixing pyrenyl actin:S1 with ADP implies that binding of myosin:ADP to actin shifts the equilibrium over to the M S state. If k 1 Ј Ϸ k 2 Ј at 20°C, then the model would predict a single phase for mixing pyrenyl actin:S1:ADP with ATP, with rate approximating k 2 Ј. By contrast, at 10°C a biphasic release transient would be expected if k 1 Ј Ͻ Ͻ k 2 Ј, as predicted from the model. The basic conclusion that release of ADP from acto-S1 occurs in two steps remains valid even if the 4 mM ATP used in this experiment (Fig. 3) were to contain a small contaminating amount of ADP. Although the presence of such a contaminant would influence the calculated values of K ADP (Equation 2), it would not influence the biphasic nature of the transients or its significance.
The model also explains the role of loop 1. Deleting this loop has three effects as follows. It enhances the affinity of S1 for ADP. It enhances the affinity of S1:ADP for actin. It slows actin-activated ADP release (30). All of these effects could be explained by arguing that deleting loop 1 increases K 0 and decreases K 1 which would have the combined effect of favoring formation of M S D S . Evidence for the former is provided by our previous study (30), which showed that ⌬25/50S1:dmD consists of a predominance of the shorter lifetime state and which implies that deletion of the loop increases K 0 . Evidence for the latter is provided in the demonstration that deletion of the loop slows mant-ADP release from S1 (30) as well as ADP release from acto-S1 (present study, Fig. 3B). These effects imply that deletion of loop 1 also reduces K 1 by reducing k 1 .
What is the physiologic significance of a three-step transition for ADP release from smooth muscle myosin? The temperature dependence of the K 1 Ј transition suggests that it is a concerted reaction, linked to a large scale structural change. However, the free energy change for this step would be nearly zero. This apparent paradox could be explained if it is assumed that an additional movement of a lever arm, which would occur in this step, is coupled to a change in the state of the catalytic site. This could also provide a mechanism for strain-dependent release of nucleotide. In this scheme, the reduction in ADP affinity and subsequent nucleotide release would require a rotation of the lever arm, with the free energy released by one transition compensated by free energy expenditure by the other.
The kinetics of fast skeletal myosin II and Dictyostelium myosin II both indicate that a second strongly bound ADP state does not exist in those myosins. Consistent with this is the lack of movement upon ADP dissociation seen in the cryo-electron microscopy maps of Dictyostelium myosin II (Fig. 7) and the previously published EPR data comparing skeletal and smooth muscle myosin (11). Thus, it would appear that that the large scale structural change inferred by the temperature depend-ence of the isomerization between the two strong smooth actomyosin⅐ADP states is the structural change observed in movement of the light chain domain upon ADP release from smooth muscle myosin. It is interesting that the ADP/rigor position of the Dictyostelium lever arm is similar to the rigor position of the smooth lever arm. On the other hand, previously published fast skeletal EM maps and EPR data (10,11) reveal that the ADP/rigor lever arm position of skeletal myosin is similar to the ADP position of smooth. This could reflect either differences in the size of the power stroke among these myosins or differences in the connection between the light chain helix and the converter domain of the myosin motor. In particular the position of the essential light chain density of Dictyostelium myosin II appears somewhat different with respect to the motor domain than that of smooth muscle myosin II (Fig. 7).
It is ironic that neither the myosin for which there is the most functional data (fast skeletal myosin II) nor the myosin for which there is the most structural data (Dictyostelium myosin II) demonstrate an additional transition associated with ADP release from actomyosin. It is likely that the coupling that produces the additional movement has been lost to allow for greater velocities of filament sliding. However, this appears to be in conflict with data from skeletal muscle fibers undergoing isometric contraction, conditions under which ADP release is greatly slowed, yet the myosin remains attached strongly enough to support force transduction. The resolution of this paradox likely results from a strain-induced uncoupling of strong actin and ADP binding. We propose that there is sufficient compliance within the myosin molecule to allow a strong actin interface to form without completion of the lever arm swing. Indeed, in contracting muscle there is evidence for lever arm positions that are in between the unstrained positions revealed by the crystal structures (17,36,37). If ADP release cannot occur rapidly until the lever arm approaches the unstrained position, then this would provide a means to have both strong actin binding and strong ADP binding during isometric contraction. However, in the absence of such a load, as occurs in solution, the lever arm would reach its unstrained position coincident with strong actin binding, and this would give rise to the reciprocal relationship between actin and ADP binding that we observe for fast skeletal and Dictyostelium myosin II.
By contrast, smooth muscle myosin II, brush border myosin I, and myosin VI all display large scale movements associated with ADP release (10,13,34). This movement indicates that there is loose coupling between actin and ADP binding and implies the existence of a state that binds both actin and ADP strongly in the absence of imposed strain. The critical difference between these myosins on the one hand and fast skeletal and Dictyostelium myosin II on the other is that the former will have a long lived ADP state in the absence of load. This feature may be critical for a number of unconventional myosins in order to generate a long duty cycle in the absence of load. For smooth muscle myosin, this structural transition could generate long lived, force-producing cross-bridges that would allow the sustained force generation seen in this tissue. It is reasonable to assume that other members of the myosin family, such as myosin V, will also display such movements. Indeed, in the case of myosin V, DeLaCruz et al. (2) have argued that the kinetics of nucleotide binding by this myosin could be best explained by invoking the presence of an additional, strongly bound actomyosin ADP state. In all cases, the structural coupling between an ADP-occupied nucleotide pocket and the myosin lever arm will generate strain-dependent ADP release that could be exploited for a range of physiologic purposes.