Metal Switch-controlled Myosin II from Dictyostelium discoideum Supports Closure of Nucleotide Pocket during ATP Binding Coupled to Detachment from Actin Filaments*

Background: Myosins convert the energy of ATP hydrolysis into force production. Results: Substitution of the metal-interacting serine in switch-1 with cysteine renders the motor sensitive to manganese. Conclusion: This technology provides a reversible structural linkage between the nucleotide pocket and actin-binding region in the myosin motor domain. Significance: Understanding the ATPase mechanism requires a description of allosteric communication in molecular motors. G-proteins, kinesins, and myosins are hydrolases that utilize a common protein fold and divalent metal cofactor (typically Mg2+) to coordinate purine nucleotide hydrolysis. The nucleoside triphosphorylase activities of these enzymes are activated through allosteric communication between the nucleotide-binding site and the activator/effector/polymer interface to convert the free energy of nucleotide hydrolysis into molecular switching (G-proteins) or force generation (kinesins and myosin). We have investigated the ATPase mechanisms of wild-type and the S237C mutant of non-muscle myosin II motor from Dictyostelium discoideum. The S237C substitution occurs in the conserved metal-interacting switch-1, and we show that this substitution modulates the actomyosin interaction based on the divalent metal present in solution. Surprisingly, S237C shows rapid basal steady-state Mg2+- or Mn2+-ATPase kinetics, but upon binding actin, its MgATPase is inhibited. This actin inhibition is relieved by Mn2+, providing a direct and experimentally reversible linkage of switch-1 and the actin-binding cleft through the swapping of divalent metals in the reaction. Using pyrenyl-labeled F-actin, we demonstrate that acto·S237C undergoes slow and weak MgATP binding, which limits the rate of steady-state catalysis. Mn2+ rescues this effect to near wild-type activity. 2′(3′)-O-(N-Methylanthraniloyl)-ADP release experiments show the need for switch-1 interaction with the metal cofactor for tight ADP binding. Our results are consistent with strong reciprocal coupling of nucleoside triphosphate and F-actin binding and provide additional evidence for the allosteric communication pathway between the nucleotide-binding site and the filament-binding region.

Purine nucleotide-binding proteins constitute a large fraction of the hydrolase enzymes (EC 3.6) discovered in biological systems. G-proteins, kinesins, and myosins are hydrolases that utilize a common fold consisting of a central ␤-sheet surrounded by ␣-helices. Specifically, there are three primary structural elements in the active site that are used to bind the nucleotide and metal ion (typically divalent magnesium, Mg 2ϩ ) and coordinate nucleotide hydrolysis (1). These conserved active site elements are called the phosphate-binding loop (P-loop 2 or Walker A motif, consensus sequence GXXXXGK(S/ T)), switch-1 (NXXSSR in kinesins and myosins, T in G-proteins), and switch-2 (Walker B motif; DXXG) (2)(3)(4)(5). The P-loop interacts with the phosphates of the nucleotide and the Mg 2ϩ cofactor. The switch-1 and switch-2 motifs function to sense and respond to the presence or absence of the ␥-phosphate of the nucleotide, which is accomplished through direct interaction with the nucleotide as well as Mg 2ϩ coordination. Four of the six ligands that satisfy the octahedral geometry for metal ion coordination are typically the hydroxyl oxygen from the side chain of the conserved P-loop serine/threonine, an oxygen from the ␤-phosphate, and two distinct water molecules (6). The remaining two ligands vary depending on the particular hydrolase and the nucleotide state (e.g. ATP-bound versus ADP-bound). Although there has been a wealth of research directed to understand the protein conformational changes needed for nucleotide catalysis, we have a poor understanding of the detailed mechanistic biochemistry that functions to lower the transition state free energy and achieve the accelerated rate of hydrolysis.
Myosins and kinesins are related cytoskeletal molecular motor proteins that convert the free energy of ATP hydrolysis into force generation along their filament track (filamentous actin for myosins and microtubules for kinesins). They contain a similar central core of structural elements that are thought to facilitate conformational changes in the motor domain during the ATP hydrolysis cycle (7)(8)(9). The movement of switch-1 and switch-2 is linked to the transmission of structural changes at the filament-binding interface to modulate the affinity of the motor for its track as well as to trigger displacement of the rod-like lever arm in myosins or the neck linker in kinesins. Despite the structural similarities between myosins and kinesins, the kinetic and thermodynamic mechanisms are quite different. For example, upon ATP binding to the actomyosin complex, myosin⅐ATP rapidly detaches from the filament before ATP hydrolysis (10), whereas ATP binding to the microtubule⅐kinesin complex does not typically dissociate the motor from the filament, rather it remains tightly bound and undergoes hydrolysis on the filament before detaching (11). Nevertheless, the rate-limiting reaction in the ATP hydrolysis cycle for both motors is stimulated in the presence of their respective filaments. The structural details for this filamentpromoted stimulation of ATPase activity remain largely unspecified.
Actomyosin force generation occurs through cyclic interactions of filamentous actin (F-actin) and myosin cross-bridges that are mechanochemically linked to the myosin ATP hydrolysis cycle (12). The genetically truncated and recombinantly expressed non-muscle myosin II motor (called S1dC) from Dictyostelium discoideum has been a good model system for studying the actomyosin mechanochemical cycle (13). It has long been known that ATP binding to myosin results in closure of the nucleotide pocket (13,14). Based on x-ray crystallographic and cryo-electron microscopy data, myosin switch-1 has been hypothesized to critically function in the transmission of structural information from the nucleotide-binding site to the actinbinding regions (15). The rigor-like (nucleotide-free) structure of myosin (16 -18) and the post-rigor (ATP-bound) structure (19,20) show significant conformational differences in the switch-1 loop and the actin-binding cleft. Conibear et al. (21) utilized pyrene excimer fluorescence to demonstrate reciprocal opening of the actin-binding cleft upon closure of the nucleotide-binding site during ATP binding. Using engineered tryptophan substitutions in the switch-1 region of D. discoideum myosin II, Kintses et al. (22) demonstrated that switch-1 existed in a dynamic equilibrium between two states (open and closed). Recent electron paramagnetic resonance spectroscopic studies (23,24) revealed the thermodynamics of nucleotide-binding cleft closure coupled to actin filament binding. Studies using molecular dynamics simulation methodologies have corroborated the reciprocal coupling of switch-1 closing in the ATPbound state and actin-binding cleft opening to weaken filament affinity (25,26). Using a combination of nonhydrolyzable nucleotide analogues and myosin mutations that block ATP hydrolysis, Yengo et al. (27) provide kinetic evidence for two distinct myosin⅐ATP states that have different actin binding affinities, consistent with this allosteric communication pathway. Collectively, these studies provide a strong argument for a mechanical link between the nucleotide pocket conformation and a structural change in the actin-binding cleft.
In a recent study (28), we described a strategy to control the enzymatic activity and force-generating ability of molecular motors from the kinesin superfamily using an engineered metal switch technology. This technology centers on the substitution of the conserved metal-interacting switch-1 serine (NXXSSR) with cysteine (NXXSCR), thus allowing us to take advantage of the different affinities of Mg 2ϩ and Mn 2ϩ for serine (-OH) and cysteine (-SH) amino acids. This substitution resulted in diminished catalytic activity in kinesins, because the serine interaction with Mg 2ϩ was critical for catalysis. By substituting Mg 2ϩ with Mn 2ϩ , which interacts more strongly with the thiol of cysteine, we observed a restoration of metal interaction with this residue, thus rescuing ATP catalysis. This technology has provided a powerful and direct probe for catalytically important interactions between the metal, nucleotide, and protein.
In this work, we engineered the homologous switch-1 substitution into D. discoideum myosin II to determine rescue of the switch-1 interaction with the metal during the myosin ATPase cycle. Utilizing enzyme kinetics and thermodynamic methodologies, we show that this substitution modulates the actomyosin interaction based on the divalent metal present in solution.

EXPERIMENTAL PROCEDURES
Experimental Conditions-Experiments were conducted in ATPase buffer (20 mM HEPES, pH 7.2, using sodium hydroxide, 5 mM magnesium chloride, 95 mM potassium chloride, 150 mM sucrose, and 1 mM dithiothreitol (DTT)) at 298 K. Where relevant, magnesium chloride was substituted with manganese chloride or calcium chloride. Rabbit skeletal muscle actin was purified from acetone powder (29) and labeled with pyrenyliodoacetamide (30). Both unlabeled and labeled G-actin were purified using Superdex 75 (GE Healthcare) gel filtration chromatography (31) and stored in 2 mM Tris-HCl, pH 8.0, 0.5 mM DTT, 0.2 mM ATP, 0.1 mM calcium chloride, and 0.01% (w/v) sodium azide under constant dialysis at 277 K. Pyrene-labeled F-actin was prepared as described (32). Unless otherwise noted, final concentrations of the reactants are reported.
Construction, Expression, and Purification of M761(WT) and S237C-Escherichia coli strain Mach1 (Invitrogen) was used for all cloning reactions. D. discoideum S1dC myosin II (DdMhcA) consisting of Pro-3-Arg-761 followed by a FLAG purification tag (C terminus encoding 761 RGDYKDDDDK) was cloned into a modified pDXA-3C expression vector that codes for the native myosin II Met-1-Asp-2 by incorporation of a BamHI site (33). The S237C substitution was introduced using the QuikChange site-directed mutagenesis kit (Stratagene) with primers 5Ј-CCACCCGTAACAACAATTCATGTCGTTTC-GGTAAATTCATTG-3Ј and 5Ј-CAATGAATTTACCGAAA-CGACATGAATTGTTGTTACGGGTGG-3Ј, and the resulting DNA mutation was confirmed by sequencing. D. discoideum AX3-ORF ϩ cells were transfected with each expression vector as described (34), and constitutive recombinant protein expression was performed by growing cells in 3 liters of AX medium (Formedium) at 294 K at 120 rpm to 10 7 cells/ml. The M761(WT) myosin motor domain (hereafter referred to as WT) was purified as before (34 -36). Specifically, cells (ϳ15 g) were lysed for 15 min at 277 K in 60 ml of 50 mM Tris-HCl, pH 8.0, 2 mM EDTA, 0.2 mM EGTA, 1 mM DTT, 5 mM benzamidine, 10 M phenylmethylsulfonyl fluoride (PMSF), 1 tablet of Complete protease inhibitor mixture (Roche Applied Science), 2% (v/v) Triton X-100, 30 g/ml RNase A, and 100 units of alkaline phosphatase. Cell lysis was followed by centrifugation at 186,000 ϫ g for 45 min and homogenization of the cytoskeletal pellet with 100 ml of extraction buffer (50 mM HEPES, pH 7.3, using sodium hydroxide, 100 mM sodium chloride, 30 mM potassium acetate, 10 mM magnesium acetate, 5 mM benzamidine, 10 M PMSF, and 1 tablet of Complete protease inhibitor mixture (Roche Applied Science)). The centrifugation and homogenization procedure was repeated in 20 ml of the extraction buffer with 20 mM ATP to release the myosin motor domain from actin. The final centrifugation step (370,000 ϫ g for 45 min) produced a myosin-containing supernatant that was loaded onto an anti-FLAG M1-agarose affinity resin (Sigma) equilibrated with wash buffer (50 mM Tris-HCl, pH 8.0, 150 mM sodium chloride, and 1 mM EDTA). The protein was eluted from the column using elution buffer (wash buffer plus 0.1 mg/ml FLAG peptide (Sigma)); the pooled fractions were adjusted to 5 mM EDTA to chelate any trace divalent cations, and the protein was incubated at 277 K for 15 min before buffer exchange using a HiPrep desalting column (GE Healthcare) into final buffer (20 mM HEPES, pH 7.3, with sodium hydroxide, 95 mM potassium acetate, 0.1 mM EDTA, 150 mM sucrose, and 1 mM DTT). The protein was concentrated to 4 -5 mg/ml, frozen in small aliquots using liquid nitrogen, and stored at 193 K. For the M761(S237C) protein (hereafter referred to as S237C), manganese chloride at 10 mM was added to the extraction buffer plus ATP for promoting actomyosin dissociation. Protein concentrations were determined using Bradford reagent (Sigma).
Actomyosin Cosedimentation Assays-The actomyosin cosedimentation experiments were performed by mixing myosin with phalloidin-stabilized F-actin for 10 min before the addition of either 10 mM MgATP or MnATP. These complexes were immediately centrifuged at 100,000 ϫ g for 30 min. Gel samples were prepared for the supernatant and pellet fractions for each reaction, and proteins were analyzed by SDS-PAGE. Densitometry was performed using ImageJ software (37).
Steady-state ATPase Measurements-The basal and actinstimulated ATPase activities of WT and S237C were measured by the NADH coupled assay as described (38). Briefly, myosin or actomyosin was diluted to twice the final concentration in ATPase buffer minus Mg 2ϩ , Mn 2ϩ , or Ca 2ϩ (named Mx 2ϩ ). ATP was also diluted to 2ϫ final concentration in ATPase buffer with 2ϫ final Mx 2ϩ concentration and 2ϫ NADH mixture (final concentrations as follows: 0.5 mM phosphoenolpyruvate, 0.4 mM NADH, 100 units/ml pyruvate kinase (Sigma), 20 units/ml lactate dehydrogenase (Sigma)). To initiate the reaction, equal volumes of myosin or actomyosin and Mx 2ϩ -ATP plus mixture were thoroughly mixed by pipetting, and 150 l of each reaction was transferred to a 96-well microplate for absorbance reading at 340 nm for 30 -240 min using a BioTek Epoch microplate spectrophotometer. A standard curve from 0 to 800 M NADH was used to convert A 340 to ADP product concentration.
ATPase activity of WT and S237C was monitored as a function of F-actin concentration using the NADH assay. The observed rates of ATP hydrolysis were normalized to 1 M enzymatic sites and plotted as a function of F-actin concentration. For conditions where F-actin stimulation was observed, the data were fit to modified Michaelis-Menten Equation 1, where E o is the myosin concentration; k cat is the maximum ATPase rate at infinite F-actin; k basal is the ATPase rate at zero F-actin, and K 0.5, actin is the F-actin concentration required to provide one-half the maximal velocity. For conditions where F-actin inhibition of myosin ATPase was observed, the data were fit to hyperbolic inhibition Equation 2, where k basal is the velocity at zero F-actin concentration; A o is the amplitude of inhibition defined as the rate at infinite F-actin minus the basal velocity, and K i, actin is the inhibition constant for F-actin. Michaelis-Menten steady-state kinetics were measured by reacting myosin and actomyosin complexes with increasing ATP concentrations, and the observed rates of ATP hydrolysis were normalized to 1 M enzymatic sites and plotted as a function of ATP concentration. Each data set was fit to hyperbolic Michaelis-Menten Equation 3, Presteady-state Kinetic Experiments-Stopped-flow measurements were performed at 298 K using an SF-2004 (Dartmouth College) or SF-300X (Indiana University) stoppedflow apparatus (KinTek Corp.) equipped with a xenon arc lamp (Hamamatsu). In experiments to measure ATP-promoted actomyosin dissociation, pyrenyl-F-actin filaments (1:5 labeled/unlabeled) were equilibrated with 2-fold excess myosin for 30 min, followed by rapid mixing in the stoppedflow with Mx 2ϩ ATP. Pyrene fluorescence was monitored over time, ex, max ϭ 365 nm, em, max ϭ 407 nm (400-nm long pass filter). For experiments at increasing ATP concentration, the observed exponential rate of ATP-promoted actomyosin dissociation was plotted against ATP concentration and fit to the following hyperbola based on Scheme 1, where k obs is the observed rate of the exponential change of fluorescence intensity; k ϩ2 is the rate constant for the ATP-dependent isomerization (Scheme 1), and K d, ATP is the dissociation constant for weak ATP binding forming the collision complex AM⅐ATP.
Inorganic phosphate (P i ) product release from myosin was determined using the MDCC-PBP coupled assay, as described previously (11, 39 -41). Specifically, WT and S237C were equilibrated with MDCC-PBP and P i mop, followed by rapid mixing SCHEME 1

Metal Switch for Non-muscle Myosin II
with Mx 2ϩ -ATP plus P i mop. The P i mop removes the 1-2 M contaminating P i present in the buffers, and the concentrations of the P i mop reagents (0.05 units/ml purine nucleotide phosphorylase, 75 M 7-methylguanosine) were determined to effectively prevent competition with the MDCC-PBP for P i during the reaction. The experimental design assumes that, after ATP hydrolysis, P i product will be released from the active site of myosin and bind rapidly and tightly to the MDCC-PBP, thus triggering the fluorescent enhancement of the MDCC-PBP⅐P i complex, ex, max ϭ 425 nm, em, max ϭ 464 nm (450-nm long pass filter (39)). To convert the observed change in fluorescence intensity into units of P i concentration, a P i standard curve was used. Each nonlinear transient in Fig. 5 was fit to the following burst Equation 5, where A o corresponds to the concentration myosin⅐ADP⅐P i complexes that proceed forward toward P i release; k b is the rate of the exponential burst phase, and k ss is the rate of the linear phase (M P i s Ϫ1 ) defining subsequent ATP turnovers.
Kinetics of the interaction of mant-ADP with myosin was measured by equilibrating a myosin⅐mant-ADP (1:1) complex in excess Mx 2ϩ followed by rapid mixing with a high concentration of the complementary Mx 2ϩ and ATP to chase the mant-ADP from the active site as described (44). The mant fluorescence was monitored over time, ex, max ϭ 356 nm and em, max ϭ 448 nm (400-nm long pass filter).
Data Analysis-DYNAFIT software (BioKin Ltd., Pullman, WA) was used to model the P i product release kinetics for 2 M WT at 500 M MnATP to the mechanism proposed in Scheme 2 (47). The kinetic constants for ATP binding (K ϩ1 k ϩ2 ) and ATP hydrolysis (k ϩ3 and k Ϫ3 ) were fixed in value based on previous studies with WT under similar experimental conditions (Table 3) (48,49). Because P i binding to MDCC-PBP was both rapid and tight, P i release from WT was modeled as an irreversible step. ADP release was also modeled as irreversible because only 0.2% of the ATP was hydrolyzed during the time course, and thus ADP product rebinding to WT would be improbable. The forward rate constants for P i release (k ϩ4 ) and ADP release (k ϩ5 ) were allowed to float in the DYNAFIT regression analysis.

RESULTS
Metal Dependence of S237C Interaction with F-actin-We began our study by engineering a serine (residue 237) to cysteine substitution in switch-1 (WT, SSR, and S237C, SCR) of the motor domain of D. discoideum S1dC myosin II. This resi-due has been shown to coordinate the divalent Mg 2ϩ at the active site of myosin II when adenosine nucleotide is bound (19, 20, 36, 50 -56). We hypothesized the S237C motor would display divalent cation sensitivity due to the differential affinities of Mg 2ϩ and Mn 2ϩ for serine (-OH) and cysteine (-SH) amino acids, similar to the switch-1 serine-to-cysteine substitution in related kinesin motors (28). Substitution of this active site serine with cysteine is expected to diminish catalytic activity if the serine interaction with Mg 2ϩ is critical for catalysis. Substituting Mg 2ϩ with Mn 2ϩ , which interacts more strongly with cysteine, should therefore restore the metal interaction with this residue, thus rescuing catalysis. WT and S237C proteins were expressed and purified from D. discoideum using cytoskeleton affinity and ATP release (see under see "Experimental Procedures" and Refs. 34 -36) followed by anti-FLAG affinity chromatography. The WT enzyme preparation occurred as expected, but at the ATP release step, S237C was found to predominantly sediment with the F-actin cytoskeleton in the presence of Mg 2ϩ (data not shown). However, including Mn 2ϩ in the extraction buffer promoted actomyosin dissociation and S237C partitioned with the soluble fraction.
Actomyosin cosedimentation experiments were performed to assess the fraction of WT and S237C that bound to phalloidin-stabilized F-actin under different metal conditions (Fig. 1). In the absence of added nucleotide, greater than 99% of myosin bound and partitioned with F-actin, which, under these experimental conditions, was consistent with a very low K d,actin that has been previously reported (48). For WT, a comparable fraction of myosin detached from F-actin in the presence of excess MgATP or MnATP (76 and 67%, respectively), which represented a weakening of actin binding by ϳ100-fold over nucleotide-free conditions. However, S237C retained relatively tight binding to F-actin (16% detached) in the presence of MgATP, yet MnATP weakens the acto⅐S237C interaction (61% detached) similar to WT. Therefore, the S237C substitution appears to modulate the actomyosin interaction based on the divalent metal present in solution.

Steady-state ATPase Kinetics Versus F-actin Concentration-
ATPase activities of WT and S237C were monitored in the presence of Mg 2ϩ , Mn 2ϩ , and Ca 2ϩ with increasing concentrations of F-actin using the NADH-coupled assay (Fig. 2). WT displayed typical basal and actin-stimulated steady-state MgATPase kinetics (basal, k cat ϭ 0.05 s Ϫ1 site Ϫ1 ; F-actin, k cat ϭ 2.0 s Ϫ1 site Ϫ1 ; see Fig. 2A and Table 1) with a weak F-actin binding interaction (estimated K 0.5, actin ϭ 192 M) as observed previously for D. discoideum S1dC myosin II (13). The MnATPase kinetics for WT showed much higher basal activity (k cat ϭ 0.84 s Ϫ1 site Ϫ1 ; Table 1) that was stimulated 2-fold by F-actin (k cat ϭ 1.6 s Ϫ1 site Ϫ1 ; Fig. 2B; Table 1). WT binding to F-actin in MnATP was tightened by greater than an order of magnitude (K 0.5, actin ϭ 17 M) compared with MgATP conditions. WT demonstrated an enhanced basal CaATPase rate (k cat ϭ 0.46 s Ϫ1 site Ϫ1 ; Fig. 2C and Table 1) compared with MgATP, yet F-actin inhibited its CaATPase (K i,actin ϭ 2.0 M) due to the weakening of CaATP affinity for WT (see below) .
Surprisingly, S237C displayed rapid basal MgATPase (k cat ϭ 0.77 s Ϫ1 site Ϫ1 ) that was inhibited (rather than activated) in the presence of F-actin ( Fig. 2D and Table 1), which yielded an apparent K i, actin at 0.25 M. Based on this apparent K i, actin , we would expect 84% of the myosin to partition with 1.5 M F-actin in excess MgATP under these conditions, which was consistent with our cosedimentation experiments (Fig. 1). In Mn 2ϩ , S237C displayed rapid ATPase (k cat ϭ 0.65 s Ϫ1 site Ϫ1 ; Fig. 2E and Table 1) that was stimulated by F-actin (k cat ϭ 1.0 s Ϫ1 site Ϫ1 ) and showed similar actin binding affinity (K 0.5, actin ϭ 17  (Table 1). M) compared with WT. The basal CaATPase of S237C was also very rapid (k cat ϭ 0.48 s Ϫ1 site Ϫ1 ), comparable with WT, but demonstrated strong actin-promoted inhibition (k cat ϭ 0.03 s Ϫ1 site Ϫ1 and K i, actin ϭ 4 nM; Fig. 2F and Table 1). Together, these results support strong evidence for divalent cation regulation of the actomyosin interaction for both WT and S237C motors.
ATP Concentration Dependence of Myosin and Actomyosin ATPase-To determine the relative affinity of WT and S237C for ATP in Mg 2ϩ , Mn 2ϩ , and Ca 2ϩ in the absence and presence of F-actin, the ATPase kinetics were monitored as a function of ATP concentration (Fig. 3). WT MgATPase kinetics in the absence and presence of F-actin (Fig. 3A) were comparable with those of a similar D. discoideum S1dC myosin II construct (57). Notably, the WT MgATPase enzymatic efficiency (k cat / K m, ATP ) was significantly higher with F-actin (basal ϭ 0.22 M Ϫ1 s Ϫ1 versus F-actin ϭ 5.3 M Ϫ1 s Ϫ1 ; Table 1). Although F-actin stimulated WT MnATPase (Fig. 3B), the enzymatic efficiency was similar (basal ϭ 0.11 M Ϫ1 s Ϫ1 versus F-actin ϭ 0.14 M Ϫ1 s Ϫ1 ). In Ca 2ϩ (Fig. 3C), F-actin promoted a dramatic weakening of ATP binding for WT (K m,ATP ϭ 500 M), thus leading to a marked reduction (ϳ300-fold lower) in enzymatic efficiency. Therefore, the presence of different excess divalent cations in solution regulates the WT⅐ATP and acto⅐WT interactions.
S237C MgATPase kinetics in the absence and presence of F-actin showed significantly different ATP binding affinities (K m, ATP ϭ 4.7 M versus 7800 M, respectively; see Fig. 3D) leading to Ͼ1000-fold reduction in enzymatic efficiency. These data suggest that the slow rate-limiting reaction in the S237C MgATPase cycle has changed compared with WT, and the populated intermediate in the cycle is a tightly actin-bound state. In Mn 2ϩ (Fig. 3E), the ATP binding affinity to S237C in the presence of F-actin and the F-actin binding affinity (Fig. 2E) were restored to nearly WT levels ( Table 1). In Ca 2ϩ , S237C showed very weak ATP binding and slow CaATPase activity (Fig. 3F), coupled to very tight F-actin association (Fig. 2F). Together, these results support the strong reciprocal coupling of ATP and actin binding affinities, which support the model of allosteric communication between switch-1 closure via S237C interaction with the metal and weakening of the actin-binding interface.
ATP-promoted Dissociation of the Actomyosin Complex-WT and S237C were complexed with pyrene-labeled F-actin (2:1) to ensure complete saturation of the F-actin lattice. This actomyosin complex was rapidly mixed with ATP and different divalent cations in a stopped-flow instrument, and pyrene fluorescence was monitored over time (Fig. 4). The experimental design assumed that myosin binding to pyrenyl-F-actin resulted in the quenching of pyrene fluorescence, but upon  (Table 1). myosin detachment, the quenching was diminished (10,48,58,59). In Fig. 4A, transients for acto⅐WT showed a rapid exponential increase in pyrene fluorescence upon reacting with either MgATP (k obs ϭ 137 s Ϫ1 ) or MnATP (k obs ϭ 127 s Ϫ1 ) but a much slower exponential increase with CaATP (k obs ϭ 9 s Ϫ1 ) on the 0.1-s time scale. Acto⅐S237C did not show any observable change in fluorescence with either MgATP or CaATP on the same time scale, yet for MnATP, S237C dissociated rapidly (k obs ϭ 49 s Ϫ1 ) albeit at a slower rate compared with WT (Fig.  4A). If the pyrene fluorescence was monitored for 10 s (Fig. 4B), S237C still did not dissociate from F-actin with CaATP but did show slow detachment with MgATP (k obs ϭ 0.11 s Ϫ1 ). The observed rate of S237C detachment from pyrenyl-F-actin increased as a function of ATP concentration (Fig. 4, C and D). When the observed rate was plotted against ATP concentration and the data were fit to Equation 4, the k ϩ2 was 0.58 s Ϫ1 and K d,ATP was 620 M, which was consistent with the weak K m,ATP derived from our ATP concentration dependence of the acto⅐S237C ATPase activity (Fig. 3D). P i Product Release Kinetics-The kinetics of P i product release from WT and S237C were monitored using the MDCC-PBP-coupled assay in a stopped-flow instrument (39). For WT, slow linear kinetics were observed in Mg 2ϩ (Fig. 5A), and the observed rates of P i release were comparable with the rates measured in MgATP steady-state experiments ( Figs. 2A and  3A). In Mn 2ϩ , P i release was more rapid and displayed burst kinetics (i.e. single exponential phase followed by a linear phase; Fig. 5, A, C, and D). The observed exponential burst rate (k b ) for WT was 2.1 s Ϫ1 , whereas the slow steady-state rate (k ss ) was 1.0 s Ϫ1 site Ϫ1 , which correlated well with the k cat values determined from steady-state assays (Fig. 2B; Table 1). For S237C in MgATP and MnATP (Fig. 5B), burst kinetics were observed, with more rapid P i release during the exponential burst phase (k b ϭ 92 and 178 s Ϫ1 , respectively).
Using intrinsic tryptophan fluorescence (Fig. 5E) and myosin tryptophan to mant-ATP FRET (Fig. 5F), the kinetics of ATP binding were measured for WT and S237C in Mg 2ϩ and Mn 2ϩ ( Table 2). The observed rates of ATP binding using tryptophanto-mant FRET were similar; yet we measured 1.5-fold faster kinetics for WT ϩ Mn 2ϩ compared with WT ϩ Mg 2ϩ and 1.4-fold slower kinetics for S237C ϩ Mn 2ϩ compared with S237C ϩ Mg 2ϩ using intrinsic tryptophan fluorescence enhancement ( Table 2). The amplitudes of the S237C transients were consistently lower compared with WT transients, although the explanation for this observation remains elusive. Overall, these results demonstrate that both WT and S237C are capable of rapidly binding Mg-or Mn-ATP, and the rate of this step does not limit the overall steady-state ATPase cycle under our experimental conditions. The kinetics of P i release for WT reacted with MnATP were analyzed by a four-step model (Scheme 2) using DYNAFIT (47). Given our kinetics of MnATP binding to WT are similar (based on FRET kinetics; Table 2) or are 1.5-fold faster (based on tryptophan fluorescence; Table 2), we modeled the kinetics of ATP binding as reported previously (Table 3) (48). Assuming the kinetics of ATP hydrolysis were unchanged in the presence of Mn 2ϩ , the modeled rates of P i release and ADP release were 0.78 and 1.78 s Ϫ1 , respectively ( Table 3). The similar magnitude of these modeled intrinsic rate constants was consistent with the k cat value approximately half the average rate constant for these two steps (60). In other words, both P i release and ADP release steps contribute to the slow steady-state catalytic rate constant of WT in the presence of MnATP. ADP Product Release Kinetics-The kinetics of ADP product release from myosin were monitored using mant-ADP fluorescence as described (44). The experiment assumed that when mant-ADP was bound to the active site of myosin, its fluorescence intensity was greater than when it was free in solution. For WT, similar exponential decreases in mant fluorescence were observed in Mg 2ϩ and Mn 2ϩ yielding similar ADP release rates and amplitudes (Mg 2ϩ , k obs ϭ 1.24 s Ϫ1 and A o ϭ 1.44; Mn 2ϩ , k obs ϭ 1.98 s Ϫ1 and A o ϭ 1.37; Fig. 6A). It is worth noting that the observed rate of mant-ADP release in Mn 2ϩ (k obs ϭ 1.98 s Ϫ1 ) was similar to the DYNAFIT modeled rate constant for ADP release (k ϩ5 ϭ 1.78 s Ϫ1 ) using P i release kinetics. S237C showed weak ADP binding in Mg 2ϩ as demonstrated by the very small amplitude of the representative transient (A o ϭ 0.06; Fig. 6A). Modeling these mant-ADP binding kinetics using a two-step rapid equilibrium binding mechanism (60) supports the increase of the off-rate for the myosin⅐ADP collision complex as responsible for the ϳ24-fold reduction of amplitude. Yet Mn 2ϩ seemed to rescue ADP binding, although release occurred at a slower rate (k obs ϭ 0.9 s Ϫ1 ) and near WT amplitude (A o ϭ 1.11). In Ca 2ϩ (Fig. 6B), both WT and S237C showed rapid ADP release (k obs ϭ 31 and 66 s Ϫ1 , respectively), yet S237C had ϳ25% of the amplitude compared with WT, suggesting CaADP bound more weakly to S237C compared with WT.  Table 2). F, WT and S237C myosin tryptophan to mant-ATP FRET in Mg 2ϩ and Mn 2ϩ . Each transient in E and F was fit to a single exponential function.

DISCUSSION
The kinetic and thermodynamic parameters that govern the minimal basal and F-actin-stimulated ATPase mechanism of WT and S237C myosin II have been determined in the presence of different divalent cations (Mg 2ϩ , Mn 2ϩ , and Ca 2ϩ ). We have provided evidence for an allosteric communication pathway between the nucleotide-binding site and the filament-binding region. When the switch-1 serine (Ser-237 in D. discoideum S1dC) was substituted with cysteine, its interaction with Mg 2ϩ was presumed to be weakened as a result of sulfur being a poorer ligand for the hard Mg 2ϩ ion. We observed very slow ATP binding kinetics coupled to actomyosin dissociation for S237C, suggesting that the mutant motor is still capable of closing the nucleotide pocket to trigger the conformational change at the actin-binding cleft, albeit at a markedly reduced rate. However, in the presence of Mn 2ϩ , the cysteine mutant appears to form a complete and functional active site with Mn 2ϩ , but not Mg 2ϩ as its cofactor, leading to rapid detachment of the S237C motor from the filament. We attribute this phenomenon to the sulfur of the S237C interacting more strongly with this softer Mn 2ϩ ion. Overall, this study demonstrates key differences between different molecular motor superfamilies (myosin versus kinesin (28)) when comparing details of nucleotide pocket movements and how these conformational changes are mapped onto their respective ATPase cycles.

Similarities and Differences of Mn 2ϩ Versus Mg 2ϩ and Serine
Versus Cysteine-Thio-substitution and metal ion rescue experiments were developed to study the metal-catalyzed reactions of ribozymes (61) and DNA transposases (62)(63)(64). Mg 2ϩ and Mn 2ϩ have the same octahedral coordination geometry and a similar covalent radius (1.41 Å versus 1.39 Å, respectively). Mg 2ϩ is a relatively hard metal ion (low polarizability and high charge density) that participates in nucleotide catalysis through coordination of oxygens from hydroxyls, carboxylates, and/or phosphates. In previous metal rescue experiments, oxygen was substituted with sulfur, which does not bind as tightly to Mg 2ϩ but has a higher affinity for softer transition metal ions with lower charge density and somewhat greater polarizability, such as Mn 2ϩ or Zn 2ϩ (65). Therefore, in previous studies, the addition of Mn 2ϩ resulted in the catalytic rescue of the sulfur-substituted macromolecule, which provided evidence for metal-binding sites that are required for the catalytic activity of different ribozymes (66 -68). Mutating a serine residue to a cysteine is often thought to be a very conservative amino acid substitution in proteins due to only one atom being different (-OH in serine versus -SH in cysteine). However, the serine oxygen has an atomic radius that is 0.4 Å smaller than the sulfur in cysteine, with modestly different bond lengths and angles (69 -71). These structural differences can lead to dramatic changes in ligand binding affinity, both due to steric effects as well as a change in the detailed hydrogen bonding geometry (72). However, in this study, the interaction between serine or cysteine and the divalent metal cation involves a direct bond between the metal and oxygen (from serine) or sulfur (from cysteine) with no intervening hydrogen. Therefore, although steric effects could certainly alter the structure of the myosin-active site for S237C with bound Mn 2ϩ , the effects are not expected to be as drastic as those when cysteine is acting as a hydrogen bond donor; future structural studies will provide insight into the magnitude of these effects.
Mn 2ϩ Stimulation of WT ATPase-The mechanism of Mn 2ϩ -ATPase for skeletal muscle heavy meromyosin (HMM) was investigated in previous studies (73)(74)(75). These studies revealed that the kinetic parameters of HMM satisfy a modified Lymn-Taylor model (12) wherein the rate of P i product release (or preceding conformational change necessary for productive lever arm movement), which limited the overall ATPase cycle, was accelerated in the presence of Mn 2ϩ . Our data on D. discoideum S1dC correlated well with these previous studies on HMM. Although our global analysis using DYNAFIT was limited to a single ATP concentration, the rate constants for P i and ADP product release (k ϩ4 ϭ 0.71 s Ϫ1 and k ϩ5 ϭ 1.78 s Ϫ1 , respectively; Scheme 2) were similar to the intrinsic rates for HMM (k ϩ4 ϭ 0.7 s Ϫ1 and k ϩ5 ϭ 1.2 s Ϫ1 (74,75)). Our mant-ADP release results (Fig. 6) corroborate the intrinsic rate constant for ADP release with an observed rate constant of 1.98 s Ϫ1 . We suggest that our steady-state acceleration in the presence of Mn 2ϩ demonstrates a similar effect as Ca 2ϩ activation of S1dC (13). The catalytic constant (k cat ) includes two relatively slow steps in the basal ATPase cycle (in the absence of F-actin) leading to an overall k cat approximately half the magnitude of the intrinsic constants (60).

Structural Link Demonstrated for Communication between the Nucleotide Pocket and Actin-binding Cleft Using Metal
Switch Technology-Despite the acceleration of the Mn 2ϩ -ATPase cycle, the overall mechanochemical cycle of WT S1dC in the presence of F-actin remained the same. We observed strong reciprocal coupling of nucleotide and F-actin binding such that upon rapid ATP binding, the myosin⅐MnATP complex rapidly detached from the filament (Fig. 4). For S237C in the absence of F-actin, the binding of MgATP was relatively unaffected, but in the presence of F-actin, ATP binding was weak as evidenced by steady-state ATPase (Fig. 3) and ATPpromoted actomyosin dissociation (Fig. 4) experiments. Our cosedimentation data (Fig. 1) demonstrated that S237C remains tightly bound to F-actin in MgATP, consistent with the reciprocal binding mechanism for myosin (i.e. weak ATP binding coupled to strong F-actin binding and vice versa). We propose the S237C switch-1 closes very slowly when bound to F-actin in the presence of Mg 2ϩ , and thus, the structural linkage between the nucleotide pocket and actin-binding cleft opening is perturbed. However, upon reversibly restoring the interaction between S237C switch-1 and the Mn 2ϩ metal, the structural communication pathway is established, and the enzyme can bind MnATP tightly and thus weaken its affinity for F-actin. This represents evidence for a direct and experimentally reversible linkage (i.e. swapping of metals in the reaction) of switch-1 and actin-binding cleft through conformational switching. This finding generally agrees with previous studies of different stages of force production in myosin (76 -79) and supports the previously proposed model (16,80) that ATP binding induces closure of switch-1, resulting in release of myosin from actin.
Rapid Basal ATPase for S237C in MgATP Supports Dynamic Switch-1 in the Absence of Actin-Unexpectedly, we observed ϳ17-fold enhancement of basal S237C MgATPase and similar CaATPase compared with WT ( Fig. 2; Table 1). Shimada et al. (57) characterized the ATPase activity of D. discoideum S1dC mutant S237A and found ϳ3-fold reduction in basal MgATPase as well as ϳ10-fold reduction in actin-stimulated and CaATPase. There are a variety of methodologies that support at least two conformational states of switch-1 when different nucleotides are bound at the active site, and the equilibrium distribution of these states is modulated through actin binding (22,23,81,82). Our results suggest that S237C switch-1 must be able to rapidly close in the MgATP state to establish the proper conformation necessary to lower the transition state free energy of hydrolysis. However, upon binding to actin, allosteric communication from the actin-binding interface promotes the opening of switch-1 and significantly weakens the interaction of Mg 2ϩ by S237C such that substrate binding becomes ratelimiting in the ATPase cycle. It is thought that upon reaching the rigor state (nucleotide-free myosin bound to actin) the central ␤-sheet undergoes a conformational change through twisting of the sheet (15,16,18,26,56,83). This favorable or low energy state of the central ␤-sheet must be reversed through favorable interactions between the residues in the active site and MgATP. Our results support the model that, in the presence of Mg 2ϩ , S237C tips the conformational equilibrium toward the open switch-1 state, thus abolishing rapid untwisting of the central ␤-sheet to drive actin detachment. It remains to be determined whether this slow closure of switch-1 S237C upon binding MgATP is capable of generating force during the ATPase cycle. Although the M761 fragments behave normally in all the assays presented here, it has not been found to be capable of actin movement in motility assays (48). Thus, the S237C mutation will need to be tested in future studies using a longer fragment of D. discoideum myosin II or a M761 fusion with one or two ␣-actinin repeats, which have been shown to display robust actin motility (45,48).
Switch-1 Interaction with Metal Is Important for the Mechanism of Product Release in Myosin-Serine 237 has been shown to coordinate the divalent Mg 2ϩ at the active site of D. discoideum S1dC myosin II when adenosine nucleotide is bound (19, 20, 36, 50 -56). Although most myosin crystal structures in the MgADP state show switch-1 closed and coordinating the Mg 2ϩ , several kinetic and thermodynamic studies demonstrate that the switch-1 conformation is dynamic in the MgADP state with the closed conformation being thermodynamically favorable (22,24,82,84,85). However, kinesin structures in the MgADP state show switch-1 predominantly open, although conformational dynamics are observed (86 -88). When comparing the analogous switch-1 serine-to-cysteine substitution in kinesin (28) and myosin, there is a dramatic difference in mant-ADP release behavior in these motors. For S233C in kinesin-5 (human Eg5/KSP), there was little difference in ADP binding affinity or release kinetics in the absence or presence of microtubules (28). However, in this study, we observe a significant difference in mant-ADP binding affinity for S237C in Mg 2ϩ as demonstrated by the ϳ24-fold reduction in amplitude of the S237C transient under these conditions (Fig. 6). Despite this reduction of amplitude, the observed rate constant was comparable with WT in Mg 2ϩ , suggesting that switch-1 dynamics are not limiting the rate constant for release, but the loss of interaction between switch-1 S237C and Mg 2ϩ increases the off-rate for the myosin⅐ADP collision complex. However, in the presence of Mn 2ϩ , the switch-1 interaction is established with the metal, and the affinity for ADP was rescued to near WT level.
Metal Switch Technological Applications in Other ATPases-There are vast arrays of hydrolases that utilize a threonine or serine residue to bind a divalent magnesium cofactor in the nucleotide-binding site. As this technology has been demonstrated for kinesins (28) and now myosin, it will be interesting to speculate on the application of this technology for other P-loop-containing ATPases.