Selective Perturbation of the Myosin Recovery Stroke by Point Mutations at the Base of the Lever Arm Affects ATP Hydrolysis and Phosphate Release*

After ATP binding the myosin head undergoes a large structural rearrangement called the recovery stroke. This transition brings catalytic residues into place to enable ATP hydrolysis, and at the same time it causes a swing of the myosin lever arm into a primed state, which is a prerequisite for the power stroke. By introducing point mutations into a subdomain interface at the base of the myosin lever arm at positions Lys84 and Arg704, we caused modulatory changes in the equilibrium constant of the recovery stroke, which we could accurately resolve using the fluorescence signal of single tryptophan Dictyostelium myosin II constructs. Our results shed light on a novel role of the recovery stroke: fine-tuning of this reversible equilibrium influences the functional properties of myosin through controlling the effective rates of ATP hydrolysis and phosphate release.

Various steps of the myosin mechanochemical cycle are linked to large conformational changes of the motor domain, which contains the actin and ATP binding sites as well as the converter region that forms the base of the extended lever arm domain. The converter/lever arm module is thought to amplify the structural changes occurring at the ATPase active site to produce a large working stroke (1,2).
Upon interacting with ATP, the motor domain undergoes a crystallographically identified large structural rearrangement before hydrolysis takes place. During this transition, the movement of the switch-2 loop of the active site toward the ␥-phosphate of ATP brings catalytically important residues to their active positions. This open-closed transition of switch-2 is cou-pled to a large rotation of the lever arm (from down to a primed up state), and thus the conformational rearrangement has been termed the recovery stroke, which constitutes the priming of the myosin head in an actin-detached state. Rebinding of myosin to actin in the post-recovery (up) conformation is a prelude to the power stroke and force generation (3).
Kinetic studies on a Dictyostelium myosin II motor domain construct containing a single ATP-sensitive tryptophan sensor (W501ϩ located in the relay-converter module) have revealed the correspondence between identified structural states and ATPase intermediates (4,5). In the model shown in Reaction 1, ATP binding to apo-myosin (M apo ) 3 (K 1 k 2 ) is followed by the recovery stroke (K 3a ), which is reversible and rapid compared with the subsequent hydrolysis step (K 3b ). Following hydrolysis, the reversal of the recovery stroke (K 4 ) is thought to occur before the actual release of P i (k 5 ). Reaction 1 and the current study deals with the myosin ATPase in the absence of actin, where lever arm motions are uncoupled from the performance of external work. ADP release (K 6 K 7 ) occurs practically as a reversal of the ATP binding process. (Reaction 1 ignores possible protein structural and/or dynamic differences between the M down ⅐ATP, M down ⅐ADP, and M down ⅐ADP⅐P i states.) REACTION 1 The above mechanism gives rise to the hypothesis that a shift in the equilibrium constant of the recovery stroke (K 3a ) will have an effect on the apparent rate constant of ATP hydrolysis (because app k H ϭ k 3b K 3a /(1ϩK 3a ) ϩ k Ϫ3b ). Furthermore, because P i release is thought to occur in the switch-2 open (down) conformation, the reversal of the recovery stroke (K 4 ) may influence the steady-state ATPase rate, which is controlled by the effective rate of P i release. Thus, besides its role in ATP hydrolysis and priming the myosin head, the recovery stroke may be important in fine-tuning the steady-state distribution of myosin molecules in up and down lever arm orientations and, in turn, the strong and weak actin binding states during the contractile cycle.
To test this hypothesis, we designed Dictyostelium myosin II motor domain mutants to perturb the recovery stroke selectively. This step is accompanied by a large rotation of the converter region that causes disruption of its interface to the N-terminal subdomain of the myosin head (Fig. 1, A and B). One good candidate for a directed change in the interaction pattern of this interface is the Lys 84 (N-terminal subdomain)-Arg 704 (converter) residue pair. In the down (pre-recovery) conformation the two side chains, unusually, run almost parallel to each other and their positive charges are separated by only ϳ0.5 nm (Fig. 1A). In the up (post-recovery) structure this complex is disrupted (Fig. 1B), implying that the introduction of site-directed substitutions in these positions will possibly affect the recovery stroke equilibrium constant. We constructed and characterized the K84M point mutant in which the positive charge of Lys 84 was removed and the side chain replaced by a roughly isosteric one and R704E in which we intended to convert the originally repulsive Lys 84 -Arg 704 interaction into a salt bridge. We introduced these point mutations into the W501ϩ single tryptophan construct, which has been shown to enable very sensitive resolution of the recovery stroke, while retaining essentially identical kinetic properties to the wild-type enzyme (4,5). In later sections we will refer to the W501ϩ control as "wild-type" for simplicity.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-All constructs used in this study are derivatives of the M761 Dictyostelium myosin II motor domain (6). Construction of the plasmid for the W501ϩ motor domain containing the point mutations W36F, W432F, and W584F was described earlier (4). The W501ϩ construct was further mutagenized to yield the K84M and R704E mutants as described previously (7). The pDXA-3H expression vector was used for constitutive expression of the motor domains (8). Dictyostelium AX3 cells were transformed and cultured and the recombinant proteins were prepared as described previously (4). Protein concentrations were determined using Bradford reagent (Sigma). Purity of the preparations (Ͼ95%) was checked by 9% SDS-PAGE.
Nucleotides and Nucleotide Analog Complexes-ATP (special quality, vanadate-free) was from Roche Applied Science. Other nucleotides were purchased from Sigma. The Kinetic and Spectroscopic Measurements-All measurements were carried out in a buffer comprising 20 mM TES, pH 7.5, 40 mM NaCl, and 2 mM MgCl 2 as described previously (4,5,9). Reaction profiles were analyzed by fitting to exponential functions using Origin v7.5 (Microcal Software).
Molecular Dynamics Simulations-Protein Data Bank structure 1FMW and its single point mutants were used as inputs in the modeling studies. Energy minimization and subsequent 5-ns molecular dynamics equilibration were performed with the GROMACS program package (10). 17800 explicit water molecules surrounded the proteins during calculations.

RESULTS
Structural Integrity of the Mutants-Molecular dynamics simulations of the wild-type and mutant proteins in a water box showed that the amino acid substitutions did not cause significant structural changes even in the vicinity of the targeted side chains and that the structures were stable over a 5-ns time scale ( Fig. 1, C-E). The charge separation between Lys 84 and Glu 704 in the R704E mutant was rather similar to that between Lys 84 and Arg 704 of the wild-type enzyme (0.5 nm). Also, the side chain of Met 84 in the K84M mutant adopted a very similar conformation to the corresponding Lys 84 of the wild-type construct and had a similar proximity to Arg 704 .
Steady-state ATPase Activities-The basal (actin-free) steady-state ATPase activity of K84M was ϳ1.5 times, and that of R704E was ϳ3 times, higher than that of the wild-type construct (Table 1). In addition to the NADH-linked assay, ATPase activities were confirmed by multiple turnover tryptophan fluorescence (as described in Ref. (4), data not shown).
Tryptophan Fluorescence Changes on Nucleotide Interaction-In our earlier studies, W501 showed a small quench on ADP binding (transition from apo to down conformation), whereas ATP binding caused a large fluorescence increase as the up state became populated (4, 5, 11) (Reaction 1). Fig. 2 shows tryptophan fluorescence emission spectra of the studied constructs in the absence of nucleotide, and in ADP and ATP. The extent of the quench caused by ADP was similar in all constructs, whereas the fluorescence increase on ATP addition was markedly smaller in R704E than in the other two constructs. The blue shifts characteristic of both ADP and ATP binding to apo-W501ϩ were not affected by the mutations (Fig. 2).
We measured the temperature dependence of the tryptophan fluorescence emission of the studied constructs (Fig. 3, A-C). The slope of the temperature dependence of the wildtype construct in the absence of nucleotide, in ADP, and ADP⅐AlF 4 (an ADP⅐P i analog) was similar to that of free tryptophan, implying that the motor domain adopts three distinct conformations with different fluorescence intensities in these cases (apo in the absence of nucleotide, down in ADP, and up in ADP⅐AlF 4 ) (Fig. 3A) (5). The temperature dependence was clearly different in ATP and in the ATP analogs ADP⅐BeF x and AMPPNP. The intermediate intensities and markedly different slopes in these cases have been inter-preted by means of the formation of a temperature-dependent reversible equilibrium of the down and up states, i.e. the recovery stroke (Fig. 3A) (5).
In the apo, ADP (down), and ADP⅐AlF 4 (up) states, the K84M and R704E constructs showed similar profiles to those of the wild type, indicating that the gross structures of these states are unchanged by the mutations (Fig. 3, A-C). For ADP⅐BeF x and AMPPNP, the equilibrium constant of the recovery stroke (K 3a ) could be calculated from fluorescence intensities as where F x , F down , and F up are fluorescence intensities in the presence of the given nucleotide, ADP (down state), and ADP⅐AlF 4 (up state), respectively. The equilibrium parameters showed that the K84M mutation, and even more so the R704E mutation, caused marked shifts in favor of the down conformation (Table 1). In ATP the situation is more complex because the apparent equilibrium constant of the recovery stroke ( app K 3a ) is also affected by the hydrolysis step ( app K 3a ϭ K 3a (1 ϩ K 3b ), Reaction 1), but a shift toward the down state in R704E is evident even in this case (Fig. 3, A-C).
Although the mutations shifted the equilibrium constant of the recovery stroke (K 3a ) several times, these have only small perturbations in the overall energetic profile of this step. The small ⌬G o associated with the recovery stroke is composed of a large enthalpy increase (⌬H o ϭ 50 -100 kJ/mol) and a compensating entropic contribution (Fig. 3, D and E, Table 1) in all mutants and the wild type.

Recovery stroke
where F down and F up are the fluorescence intensities in the down (ADP-bound) and up (ADP⅐AlF 4 -bound) states, respectively, and F x is the intensity of the given nucleotide analog complex (Fig. 3, A-C). d Calculated from the van't Hoff plots of the recovery stroke (ln K 3a ϭ Ϫ⌬H 0 3a /RT ϩ ⌬S 0 3a /R; Fig. 3, D and E). e Calculated from the hyperbolic dependence of the k obs of ADP binding transients on ͓ADP͔ (k obs ϭ k Ϫ6 ͓ADP͔/(K 7 ϩ͓ADP͔) ϩ k 6 ; Fig. 6A). f From ADP chasing experiments (Fig. 6B). g Equilibrium dissociation constant of ADP binding (K eq ϭ K 6 K 7 ).

Fine Tuning of the Myosin Recovery Stroke
Kinetics of ATP Interaction-On mixing the W501ϩ motor domain with ATP in the stopped-flow, the two-step ATP binding process (K 1 k 2 in Reaction 1) is followed by the more rapid recovery stroke (K 3a ) and the subsequent hydrolysis step (K 3b ) (4,5). The kinetics of the initial phase of the tryptophan fluorescence transients is dictated by nucleotide binding (K 1 k 2 ), but the amplitude of this phase will be set by the equilibrium constant of the oncoming rapid recovery stroke (K 3a ). Because the W501 fluorescence intensity in the down state is lower and in the up state is higher than that in the apo state, the initial phase can be either a burst (fluorescence increase) or a quench (fluorescence drop), depending on the value of K 3a . After this initial phase, there is a subsequent fluorescence enhancement indicating that the hydrolysis step (K 3b Ͼ Ͼ 1) pulls the reaction over to result in the predominance of the high fluorescence M up ⅐ADP⅐P i state (Reaction 1). The observed rate constant (k obs ) of this second phase (i.e. the apparent rate constant of ATP hydrolysis, app k H ) will be a composite of the recovery stroke equilibrium constant and the rate constants of the hydrolysis step ( app k H ϭ k 3b K 3a /(K 3a ϩ 1) ϩ k Ϫ3b ; Reaction 1) (4, 5).
The dependence of the k obs of the initial phase on ATP concentration was quasilinear in all mutants, showing signs of saturation above 1 mM ATP (K 1 Յ 10 Ϫ3 M Ϫ1 , data not shown). The plots had slopes (K 1 k 2 ) ϳ 0.5 M Ϫ1 s Ϫ1 at 5°C, and ϳ1 M Ϫ1 s Ϫ1 at 20°C, consistent with earlier wild-type data and showing little effect of the mutations on ATP binding kinetics (Table 1) (4). However, a more pronounced initial quench (or a lack of an initial burst) together with a slower second phase (smaller app k H ) in the mutants compared with the wild-type demonstrated a marked reduction in the recovery stroke equilibrium constant (K 3a ) caused by the mutations (Fig. 4, A and B). Fig. 4C shows the k obs of the second phase at 20°C plotted against ATP concentration. A marked reduction in the maxi-  , ADP⅐BeF x (circles), and AMPPNP (triangles). Excitation was at 297 nm (1-nm bandwidth), and the emitted light was selected through a 320-nm-long pass filter. Data were normalized to the intensity of the apo state at 20°C. D and E, Van't Hoff plots of the equilibrium constant of the recovery stroke (K 3a ) of the wild-type (squares), K84M (circles), and R704E (triangles) constructs in ADP⅐BeF x (D) and AMPPNP (E) calculated as described in Table 1.

Fine Tuning of the Myosin Recovery Stroke
mal k obs was observed in K84M and even more in R704E. This tendency held throughout the temperature range studied (5-20°C, Fig. 4D). Analysis of the amplitude data of the stopped-flow transients also highlighted the correspondence between the recovery stroke equilibrium constant (K 3a ) and the apparent rate constant of the hydrolysis step ( app k H ). The fluorescence levels after the initial phase were lower in K84M and even more in R704E than in the wild-type (indicating a smaller K 3a ), whereas in the second phase ATP hydrolysis (K 3b ) pulled the reaction to the high fluorescence M up ⅐ADP⅐P i state in all cases (Fig.  4E, Reaction 1).
Kinetics of the Recovery Stroke-Because the recovery stroke is much faster than both the preceding and subsequent steps, its kinetics can be investigated only by its selective perturbation using relaxation kinetic techniques (5,12). Following a rapid pressurejump, the motor domain-nucleotide analog (ADP⅐BeF x and AMPPNP) complexes showed conformational relaxation of the reversible and pressure-sensitive recovery stroke (Fig. 5A). The k obs of these tryptophan fluorescence transients is the sum of the forward and reverse rate constants (k obs ϭ k 3a ϩ k Ϫ3a ). k 3a and k Ϫ3a in the ADP⅐BeF x and AMPPNP complexes of the motor domain could thus be determined from the k obs of the pressure-jump records and the K 3a equilibrium constants determined in equilibrium fluorescence measurements (Fig. 3, Table 1). Fig. 5B shows that the shift toward the down state (a reduction in K 3a ) caused by the mutations is brought about predominantly by a reduction in k 3a , whereas the reverse rate   Table 2.

Fine Tuning of the Myosin Recovery Stroke
constant (k Ϫ3a ) remains largely unaffected. Furthermore, in the Arrhenius plots of these rate constants the slopes of ln k 3a or ln k Ϫ3a versus 1/T were very similar in all constructs, whereas the k 3a values of the mutants were markedly offset from those in the wild-type (Fig. 5, C and D). This means that the mutations caused a reduction in the pre-exponential term of the Arrhenius equation, but the exponential term remained largely unchanged (k ϭ A exp(E # /RT) where A is the pre-exponential term and E # is the activation energy; Table 2).
Kinetics of ADP Binding and Release-On mixing with excess ADP in the stopped-flow, all constructs showed a single exponential quench in tryptophan fluorescence (traces not shown). Analysis of the hyperbolic dependence of k obs on ADP concentration (k obs ϭ k Ϫ6 [ADP]/(K 7 ϩ [ADP]) ϩ k 6 ) showed that the initial low affinity binding step (K 7 ) was not affected by the mutations, whereas the rate constant of the subsequent isomerization (k Ϫ6 ) was slightly reduced (Fig.  6A, Table 1, Reaction 1).
The rate constants of ADP dissociation (k 6 ) were determined by chasing experiments in which the motor domain⅐ADP complexes were mixed with high concentrations of ATP in the stopped-flow (Fig. 6B). The mutations had no profound effect on either k 6 or the overall ADP binding equilibrium constant (K eq ϭ K 7 K 6 ; Table 2).

Selective Perturbation of the Recovery Stroke by Subdomain
Interactions-We aimed to selectively perturb the myosin recovery stroke by converting the repulsive Lys 84 -Arg 704 interaction into a weak attraction in the K84M mutant and a salt bridge in the R704E construct (Fig. 1). These mutations are located at the N-terminal subdomainconverter interface, which is far from the nucleotide binding pocket. As judged from the kinetics of nucleotide binding to and release from the mutants as well as molecular dynamic simulations, the substitutions did not affect the overall structure of the nucleotide binding site (Fig. 1). However, the mutations had a distinct effect on the recovery stroke that involves marked conformational rearrangements involving distant parts of the molecule. Thus, the studied mutations modulated surface charges sufficiently to have an effect on recovery stroke kinetics, but they did not override the structures of the start and end points of this transition. In all constructs ADP induced the same pre-recovery (switch-2 open or lever arm down) state and ADP⅐AlF 4 induced the post-recovery (switch-2 closed or lever arm up) state, as judged by W501 emission intensity (Fig. 3, A-C). Other attractive forces across the subdomain interface in the wild type must dominate over the repulsive force between Lys 84 and Arg 704 in the down state in order to bring them to within 0.5 nm. The wild-type Lys 84 -Arg 704 repulsive interaction exists only in the down state, and we show that this state becomes more favorable in the order wild-type 3 K84M 3 R704E.
It should be noted that the R704E mutation would also disrupt a salt bridge to Glu 755 within the same subdomain; however, this appears to have little consequence on the overall properties of this construct. In particular, in the K84M mutant where the Arg 704 -Glu 755 salt bridge was intact, the kinetic and thermodynamic effects on the recovery stroke were perturbed in the same direction as in the R704E construct.
The Recovery Stroke Is Coupled to ATP Hydrolysis and P i Release-By characterizing the transient fluorescence profiles of the interaction of the mutants with ATP and their steady-state ATPase activity, we demonstrate that even a moderate decrease in the equilibrium constant of the rapid and reversible recovery stroke causes significant slowing of ATP hydrolysis, which takes place after the recovery stroke (Fig. 4, Table 1, Reaction 1). Conversely, P i release, which is the rate-limiting step of the basal ATPase cycle and takes place after the reversal of the recovery stroke (K 4 ), is accelerated by the same mutations (Table 1, Reaction 1). This finding is consistent with earlier work in which trinitropheny-  Table 1. B, kinetics of ADP dissociation as measured by mixing 2 M motor domain plus 50 M ADP with 1 mM ATP in the stopped-flow. Fitted rate constants (k 6 ) are listed in Table 1. The ADP dissociation traces were not clear single exponentials, but the fitted k 6 values were consistent with K eq calculated from equilibrium fluorescence titrations (data not shown).  Table I). In some cases, pressure jump relaxations contained minor slow phases of uncertain origin. b Apparent activation energies. Differences between E # values of the wild-type and mutant constructs were within experimental error. S.E. of all measurements with a given nucleotide are shown. c Pre-exponential factor in the wild-type construct. d Expressed as A mutant /A wild-type . e Not detectable. Differences between wild-type and mutants were within experimental error.

Fine Tuning of the Myosin Recovery Stroke
lation of Lys 84 caused a marked increase in the basal steadystate ATPase activity of Dictyostelium myosin II because the pre-recovery state became predominant during the enzymatic cycle (13).
Mechanism of the Recovery Stroke-We show that the studied mutations selectively decreased the forward rate constant of the recovery stroke (k 3a ), whereas they did not have a remarkable effect on the reverse rate constant (k Ϫ3a ) (Fig. 5B, Table 2). The structural basis of the asymmetric effect is the disruption of the subdomain interface containing residues 84 and 704 upon the recovery stroke that moves these residues ϳ3.8 nm apart from each other (Fig. 1, A and B) (1). Because only a short-range interaction can be formed between these side chains, the mutations do not affect the reverse rate constant (k Ϫ3b ).
It is surprising that, although the mutations targeted a specific side chain interaction that could be expected to have an enthalpic character, the pre-exponential factor changed rather than the apparent activation energy of the recovery stroke (Fig.  5, C and D, Table 2). The mutations investigated in the present study affect short-range interactions in the pre-recovery conformation, and therefore they limit the entry into the pathway leading to the conformational change and possibly reshape the diffusion landscape in the initial part of the pathway. The recent simulations of the reaction coordinate of the myosin recovery stroke using Conjugate Peak Refinement provide some indications as to the nature of this pathway (14). Protein conformational changes that gate enzyme activity and protein folding are likely to share these characteristics.
Role of the Recovery Stroke Equilibrium in the Myosin Mechanism-By affecting the effective rate of ATP hydrolysis and P i release, the equilibrium constant of the recovery stroke may influence the distribution of steady-state enzymatic intermediates during the functioning of myosins. Given the widely varying actin affinities of these intermediates, this effect may also influence the duty ratio, i.e. the fractional occupancy of strongly actin-bound myosin states during steady-state ATPase cycling. The duty ratio is a key functional parameter for the effective functioning of different motors. (For instance, ensemble motors such as muscle myosin II exhibit a low duty ratio, whereas single-molecule stepping by myosin V requires a high duty ratio.) Indeed, the ATP hydrolysis step is quite reversible in most myosins examined: the apparent equilibrium constant of the hydrolysis of enzyme-bound ATP ( app K H ) is between 0.3 and 9 for various myosins (15)(16)(17)(18)(19)(20). Because the chemical step is coupled to the recovery stroke (5,12), the recovery stroke equi-librium constant (K 3a ) is likely to be a key modulator of app K H in most cases (in case of quasi-irreversible ATP binding, app K H ϭ K 3a K 3b /(1 ϩ K 3a ϩ K 3a K 3b ); Reaction 1). app K H , in turn, can significantly influence the duty ratio by limiting the rate of entry into the strong binding actomyosin complex. Interestingly, Arg 704 is highly conserved among at least eight myosin classes (including myosins I, II, and V) whereas Lys 84 is only conserved within class II myosins (19). Thus, the residues mutated in the current study may even play a role in the functional diversity of myosins, which hypothesis needs further investigation.