Metal cation controls phosphate release in the myosin ATPase

Abstract

Myosin is an enzyme that utilizes ATP to produce a conformational change generating a force. The kinetics of the myosin reverse recovery stroke depends on the metal cation complexed with ATP. The reverse recovery stroke is slow for MgATP and fast for MnATP. The metal ion coordinates the γ phosphate of ATP in the myosin active site. It is accepted that the reverse recovery stroke is correlated with the phosphate release; therefore, magnesium “holds” phosphate tighter than manganese. Magnesium and manganese are similar ions in terms of their chemical properties and the shell complexation; hence, we propose to use these ions to study the mechanism of the phosphate release. Analysis of octahedral complexes of magnesium and manganese show that the partial charge of magnesium is higher than that of manganese and the slightly larger size of manganese ion makes its ionic potential smaller. We hypothesize that electrostatics play a role in keeping and releasing the abstracted γ phosphate in the active site, and the stronger electric charge of magnesium ion holds γ phosphate tighter. We used stable myosin–nucleotide analog complex and Raman spectroscopy to examine the effect of the metal cation on the relative position of γ phosphate analog in the active site. We found that in the manganese complex, the γ phosphate analog is 0.01 nm further away from ADP than in the magnesium complex. We conclude that the ionic potential of the metal cation plays a role in the retention of the abstracted phosphate.

Introduction

Myosin is an enzyme that utilizes MgATP to produce a conformational change leading to the generation of force. Upon binding MgATP, myosin changes its conformation from straight (M*) to bent (M**). This conformational change is called the recovery stroke. After the recovery stroke, myosin changes conformation back to M*, which is the reverse recovery stroke (Fig. 1). MgATP binds to myosin in a two‐step process, consisting of rapid equilibrium step K ₁ and a relatively slow isomerization step with the rate k ₊₂. The maximum rate of the ATP‐induced myosin conformational change step reflects the rate of ATP hydrolysis k ₊₃ + k ₋₃. The following phosphate and ADP release steps complete the cycle.1 It is assumed that the role of the magnesium ion is to bridge a nucleotide and a protein, and position a nucleotide within the active site of myosin for subsequent hydrolysis. The active site closes upon ATP binding, holding MgATP and two water molecules inside.2 During ATP hydrolysis, γ phosphate is attacked by a water molecule, this leads to the phosphate abstraction. In the proposed “back door” mechanism, the phosphate release occurs in the direction, opposite to ATP binding, through a narrow tunnel, formed by small rearrangements of neighboring amino acid residues.3 The phosphate release (the rate limiting step of myosin MgATPase) triggers myosin conformational change, from the prepower stroke state M** to the postpower stroke state M*, and as the phosphate release is the rate‐limiting step, the equilibrium between M** and M* structural states is shifted toward M** state.4 Structural states M* and M** are easily distinguished by the intensity of myosin intrinsic fluorescence, M* is the low fluorescence state, M** is the high fluorescence state. Interestingly, when MgATP is replaced with MnATP, the rate of myosin basal ATPase increases and M* − M** equilibrium is shifted toward M* state,5, 6 indicating faster phosphate release. Manganese is similar to magnesium in terms of chemical properties and its inner and outer shell complexation, and has somewhat larger radius (there is 0.01 nm difference). Analysis of octahedral complexes of magnesium and manganese confirms electrostatic nature of interactions in these complexes.7, 8 Magnesium cation usually has stronger partial charge than manganese,7 that makes magnesium a “hard,” difficult to polarize ion. Larger ionic radius and smaller partial charge of manganese cation compared to that of magnesium makes manganese ionic potential weaker (the ionic potential is determined as a ratio of the charge of the ion to its radius, which is a measure of the density of the charge). Smaller charge density of the cation may affect retention of the abstracted phosphate within the active site, and therefore may lead to the decreased lifetime of the S1·ADP·Pi complex.

Figure 1.

Myosin ATPase cycle reaction scheme. M = myosin (M* and M**—myosin states with increased intrinsic fluorescence), T = ATP, D = ADP, P = phosphate.

When myosin is trapped with nucleotide analogs in M* or M** structural states, the position of γ phosphate analog in the active site reflects myosin conformation and stability of the complex (Fig. 2). The longer the distance between the phosphate analog and ADP, the less stable the complex: VO₄ in the S1·MgADP·VO₄ complex (M**) is positioned 0.03 nm further from ADP than BeF₃ in the S1·MgADP·BeF₃ complex (M*).2, 9 The half‐lifetime of the S1·MgADP·BeF₃ complex is almost two times longer than that of the S1·MgADP·VO₄ complex.10 We hypothesize that short lifetime of M** state in myosin–MnATP transient is due to low ionic potential of manganese cation, which does not retain the abstracted phosphate in the active site. This reduced retention of the abstracted phosphate should appear as a longer distance between ADP and the γ phosphate analog in the S1·MnADP·VO₄ complex, compared to the S1·MgADP·VO₄ complex. To examine the hypothesis, we trapped rabbit skeletal myosin S1 in M** state with a stable nucleotide analog MeADP·VO₄ (Me = Mg, Mn), and used Raman spectroscopy.11 We found small, but reproducible red shift of Raman spectrum of the manganese complex, which we interpret as a longer distance between vanadium and ADP. We conclude that electrostatic interaction between metal cation and γ phosphate plays a role in the retention of γ phosphate in the active site, and that metal cation controls the phosphate release rate.

Figure 2.

(A) Metal cation and ATP in the active site of myosin. Metal ion coordinates two residues (rabbit skeletal myosin sequence), two water molecules, β phosphate of ADP and vanadate, mimicking abstracted γ phosphate. (B) Proposed apical shift of vanadate in the S1·MnADP·VO₄ complex. In the vanadate, dashed lines represent apical bonds and solid lines represent equatorial bonds.

Results

Equilibrium of M** and M* states of myosin S1 depends on metal cation

The rate of ATP‐induced myosin conformational change was measured with 0.5 µM S1 (the concentration in the final mixture, here and throughout the text), rapidly mixed with various concentrations of MeATP, Me = Mg, Mn. Both metal cations support myosin ATPase activity, showing increase of M** population upon mixing of myosin S1 with ATP (Fig. 3). The second‐order reaction rate constant of ATP binding to myosin was the same for both cations, K ₁ k ₊₂ = 1.16 ± 0.06 μM⁻¹ s⁻¹ for MgATP and K ₁ k ₊₂ = 1.39 ± 0.12 μM⁻¹ s⁻¹ for MnATP, in excellent agreement with previous reports.1, 2, 3 Maximum rate of myosin conformational change V _max was the same for both cations, V _max = 52.0 ± 3.3 s⁻¹ for MgATP and V _max = 47.2 ± 2.7 s⁻¹ for MnATP; the rate is slower than previously determined1, 2, 3 due to reduced temperature in our experiments (T = 12°C) (Supporting Information, Fig. S1). Figure 3 shows that myosin S1 reaches the steady state M** when mixed with MgATP, reflecting that the reaction rate limiting step is the reverse recovery stroke. In the presence of MnATP myosin reaches M** state, but then M** state depopulates with the rate 1.45 s⁻¹ and reaches an equilibrium at a lower level of myosin intrinsic fluorescence. This indicates that the equilibrium between M** and M* is shifted toward M* state. We found that the rate of M** state depopulation remains constant in our experiments; therefore, the lifetime of M** state does not depend on MnATP concentration (Supporting Information, Fig. S2).

Figure 3.

Transients of myosin–ATP interaction, detected via change of myosin intrinsic fluorescence due to recovery (MgATP and MnATP) and reverse recovery (MnATP) strokes (M* → M** structural transition). Dots, experiment; blue, MgATP; red, MnATP. Lines: fit with exponents, one (MgATP), or two exponents (MnATP). Myosin concentration is 0.5 µM, MgATP, or MnATP concentration is 75 µM. The reaction rate of M** state formation is the same for both MgATP and MnATP. M** state is in steady equilibrium for MgATP, the population of myosin in M** state decreases for MnATP with the rate 1.45 s^{− 1}.

Dependence of Raman spectrum of the S1·ADP·VO₄ complex on metal cation

Raman spectra of S1·ADP·VO₄ complexes with manganese and magnesium are shown in Figure 4. Observed bands are assigned to $V-O$ symmetric and asymmetric stretch modes of the bound vanadate.4 There is a small but statistically significant red shift of the band, corresponding to the symmetric stretch mode when magnesium is replaced with manganese, from 829.4 ± 0.8 to 823.6 ± 0.6 cm⁻¹ accordingly. The band, corresponding to the asymmetric stretch mode does not change its position, within the error of experiment (867.9 ± 1.6 cm⁻¹ (Mg), 866.9 ± 1.3 cm⁻¹ (Mn)). The geometric mean of the observed nonbridging $V-O$ stretches is 855.3 ± 1.1 cm⁻¹ (Mg) and 852.7 ± 0.9 cm⁻¹ (Mn).

Figure 4.

Raman spectra of Mg and Mn complexes. Frequencies of the asymmetric stretch (right peak) are the same in Mg and Mn complexes. Frequencies of the symmetric stretch (left peak) are statistically different.

Discussion

ATP hydrolysis by myosin is a fast and reversible process, which is followed by a slow phosphate release and the reverse recovery stroke. The maximum rate of myosin fluorescence change, V _max, measures the rate of ATP hydrolysis5, 6; according to our results, that rate is not affected by the cation replacement. The rate of phosphate release affects the equilibrium state of myosin post hydrolysis. Therefore, observed destabilization of M** state in myosin MnATPase suggests increased rate of phosphate release. Observed red shift of Raman spectrum of the S1·MnADP·VO₄ complex indicates increase of the length of $V-O$ bond,7 and, according to the modeling,4 increase of the distance between vanadate and ADP in the active site of myosin. This fits the pattern, observed in crystal structures of myosin, trapped with ADP·BeF₃ and ADP·VO₄ nucleotide analogs, which mimic pre‐ and postrecovery stroke states of myosin. Beryllium atom in the S1·MgADP·BeF₃ complex is 0.03 nm closer to β phosphorous of ADP, than vanadium atom in the S1·MgADP·VO₄ complex. We suggest that Mn·ADP·VO₄ traps myosin in the further step post hydrolysis, and the phosphate analog is located further away from ADP (Fig. 2). The length of equatorial $V-O$ bond is calculated according to the empirical expression, which relates $V-O$ bond length to its bond valence4, 8:

$$s(V-O)\approx {(R/1.791)}^{-5.1}$$ (1)

and

$$s\left(V-O\right)={\left[0.2912·\mathrm{l}\mathrm{n}\left(21349/\nu \right)\right]}^{-5.1}$$ (2)

where s(V − O) is the valence bond order in terms of valence units, R is the length of $V-O$ bond in Angstroms, ν is the geometric mean of observed nonbridging stretches, ${\nu =\left[\left({\nu }_{\mathrm{s}}+2{\nu }_{\mathrm{a}}\right)/3\right]}^{1/2}$, ν_s is the symmetric stretch, ν_a is the asymmetric stretch, both in cm⁻¹. We found that valence bond orders are 1.394 and 1.388 for equatorial bonds, and 0.408 and 0.418 for apical bonds for S1·MgADP·VO₄ and S1·MnADP·VO₄ accordingly, in the assumption that the sum of all bonds is the valence of the central vanadium atom, which is equal to 5.8 Determined $R$ = 0.1678 nm and $R\prime$ = 0.1680 nm, where $R$ and $R\prime$ are the lengths of equatorial $V-O$ bonds in magnesium and manganese complexes accordingly, are in excellent agreement with $V-O$ lengths from crystal structure of the S1·MgADP·VO₄ complex, 0.167, 0.167, and 0.164 nm.9 Vanadium and equatorial oxygens form almost planar VO₃ group in the S1·MgADP·VO₄ complex,9 and reported $O-V-O$ angle is θ = 119°.4 Both symmetric and asymmetric stretch bonds of Raman spectrum depend on the equatorial $O-V-O$ bond angle θ as10:

$${\nu }_{\mathrm{s}}^2=\left(\frac{1}{\mu }+2\frac{\mathrm{c}\mathrm{o}\mathrm{s}\theta }{M_{\mathrm{v}}}\right)\left(F_{\mathrm{s}}+2C_{\mathrm{s}\mathrm{s}}\right)$$ (3)

$${\nu }_{\mathrm{a}}^2=\left(\frac{1}{\mu }-\frac{\mathrm{c}\mathrm{o}\mathrm{s}\theta }{M_{\mathrm{v}}}\right)\left(F_{\mathrm{s}}-C_{\mathrm{s}\mathrm{s}}\right)$$ (4)

where μ is the reduced mass of $V-O$ bond, M _v is the mass of vanadium atom, F _s is the $V-O$ stretch force constant, and C _ss is the $O-V-O$ stretch/stretch coupling force constant. Using frequencies of symmetric and asymmetric stretch bonds from our experiment and the angle θ = 119° for the magnesium complex, we have calculated F _s = 541.8 N/m and F _s/C _ss = 10.2, in good agreement with the data of Deng et al.4 Assuming that the ion replacement introduces a small perturbation to the system, we have determined angle θ = 120° for the manganese complex from Eq. (3), using calculated F _s and C _ss. Analysis of Eq. (4) shows that the angle difference of 1° produces smaller frequency shift of asymmetric stretch bond, this explains why such a frequency shift was not observed in our experiments. We estimated the position of vanadium atom from the analysis of isosceles triangles $OVO$ and $OV\prime O$, where $V$ and $V\prime$ are positions of vanadium atom in the S1·MgADP·VO₄ and S1·MnADP·VO₄ complexes accordingly (Fig. 2). The change of the angle $OVO$ from 119° to 120° should be concurrent with the shift of vanadium atom away from ADP in apical direction, making VO₃ group planar. Simple analysis shows that the distance $VV\prime$ = 0.01 nm, determined from ${\left[P^2-R^2\right]}^{1/2}$, where $R$ is the length of $V-O$ bond in the magnesium complex, determined from Eqs. (1) and (2), and $P$ is the projection of $R$ to the plane of the planar VO₃ group. $R$ and $P$ are related as the ratio of sinuses of semi angles $O-V-O$ and $O-V\prime -O$. The length of $V-O$ bond is longer in the manganese complex; therefore, we conclude that the cation replacement leads to the apical shift of vanadium and the elongation of equatorial bonds of VO₃ group. Apparently, these changes in the S1·MnADP·VO₄ complex are due to smaller ionic potential of manganese cation. This weakens the electrostatic interaction between metal cation and vanadate in the S1·MnADP·VO₄ complex, as well as between metal cation and γ phosphate in myosin MnATPase. Considering substantial structural changes of myosin molecule, induced by ATP binding, hydrolysis, and phosphate release, we propose that the role of metal cation is not just in positioning of ATP for successful hydrolysis. Metal cation controls timing of the phosphate release. There are two proposed mechanisms of the phosphate release in myosin, “back door”11 and “trapdoor.”12 In the “back door” mechanism, the abstracted γ phosphate leaves the active site in the direction, opposite to ADP, through a narrow tunnel, formed by a small rearrangement of local residues. The “trapdoor” mechanism proposes the phosphate release as a result of a large‐scale conformational change of the loop switch I, which covers the active site. Our data and crystallographic data for phosphate analogs show that the distance between phosphate and ADP affects timing of the phosphate release, supporting the “back door” mechanism. On the other side, the loop switch I is coordinated by metal cation via residue S244 of the loop. This coordination is important for myosin ATPase activity, Dictyostelium discoideum myosin mutant S237A (corresponds to S244A in rabbit skeletal myosin) exhibits threefold reduced rate of basal ATPase and does not bind mant ADP.13 One can suggest that lower ionic potential of manganese not only reduces interaction between metal cation and γ phosphate, but also affects interaction between metal cation and S244 of the loop switch I. This reduced interaction can affect the time of the closed state of the loop switch I, and therefore modulate γ phosphate release in the “trapdoor” mechanism. Unfortunately, our experiments do not allow us to distinguish unambiguously between two proposed mechanisms of phosphate release by myosin.

Conclusion

A small, but reproducible shift of Raman spectrum of vanadium in the S1·MnADP·VO₄ complex, compared to the S1·MgADP·VO₄, is interpreted as a shift of vanadium atom in apical direction in myosin's active site. As the rate of hydrolysis is not affected by the cation replacement, the shift and observed dramatic destabilization of M** state in myosin MnATPase is likely due to modulation of electrostatic interactions between the metal cation and γ phosphate. We conclude that the metal cation, that coordinates a nucleotide in the myosin active site, plays a role in the regulation of phosphate release in myosin ATPase.

Materials and Methods

Reagents

All chemicals were from Sigma‐Aldrich (Milwaukee, WI), ThermoFisher Scientific (Waltham, MA), and VWR (Radnor, PA).

Protein preparation

Myosin was prepared from rabbit leg and back muscles.14 Chymotryptic S1 was prepared as described,15 and dialyzed into 20 mM MOPS (3‐[N‐morpholino]propanesulfonic acid) pH 7.3, 50 mM KCl buffer. Complexes of S1 with nucleotide analogs were obtained by incubation of S1 with 5 mM MgADP or MnADP (or 5 mM ADP plus 20 mM NaF in case of ADP.AlF₄) for 5 min at 25°C. After that, 5 mM Na₃VO₄ (or 5 mM AlCl₃) were added, and the incubation was continued at 25°C for an additional 20 min. The stock solution of Na₃VO₄ (100 mM) was boiled before use. The complexes were purified and concentrated to 1 mM with spin concentrators (Amicon Ultra 30 kDa, EMD Millipore, Billerica, MA). The experimental buffer contained 20 mM MOPS pH 7.5, 50 mM KCl, 3 mM MgCl₂ or MnCl₂. All reported concentrations are final concentrations.

Raman spectroscopy

Raman spectra of myosin complexes were recorded with Kaiser Optical Systems RXNT‐785 spectrophotometer (Ann Arbor, MI), laser wavelength 752 nm, power 140 mW, immediately after preparation. The spectrum of 100 mM Na₃VO₄ solution was obtained at each experiment to monitor sensitivity of the spectrophotometer. The spectra of S1·MeADP·VO₄ and S1·MeADP·AlF₄ (Me = Mg, Mn) were acquired in the consecutive order and then subtracted from each other to eliminate all protein Raman bands and leave only spectrum of vanadate.4 We used spectrum of S1·ADP·AlF₄ complex as a baseline for spectrum of S1·ADP·VO₄ complex as these complexes have similar structure.9, 16 All experiments were done at T = 21 ± 1°C. To determine position of peaks, Raman spectra were fitted to the sum of three Gaussian lines with Origin 8 (OriginLab Corp, Northampton MA). For statistical significance, we averaged peak positions from multiple experiments with myosin S1 obtained from different preparations (N = 5 for S1·MgADP·VO₄, and N = 3 for S1·MnADP·VO₄).

Transient intrinsic fluorescence of myosin was measured with Bio‐Logic SFM‐300 stopped flow transient fluorimeter (Bio‐Logic Science Instruments SAS, Claix, France), equipped with FC‐15 cuvette. The mixing unit dead time is 2.6 ms. All experiments were done at T = 12°C to reduce the rate of myosin–ATP interaction for reliable detection. Myosin–ATP interaction is too fast for observation at higher temperatures, especially for high ATP concentration. Myosin intrinsic fluorescence was excited by mercury–xenon lamp at 296 nm and detected using a 320 nm cutoff filter. Multiple transients were usually acquired and averaged to improve signal to noise ratio. Eight thousand data points were acquired in each experiment.

Analysis of fluorescence transients

Transients obtained in each experiment were fitted globally with single exponential function S(t) = S _o + A·exp(−k _obs·t) (MgATP) or double exponential function S(t) = S _o + A ₁·exp(−k _obs1·t) + A ₂·exp(−k _obs2·t) (MnATP). S(t) is the observed myosin intrinsic fluorescence at time t, A _i is the signal amplitude, and k _obsi is the observed rate constant. Transients obtained at different MnATP concentrations were fitted globally, and the rate constant k _obs2 was linked in the global fit, reflecting no dependence of M** state decay on [MnATP]. The dependence of observed rates (k _obs and k _obs1 for Mg and Mn accordingly) on the nucleotide concentration was fitted with a hyperbola, v = V _max[ATP]/(K _d +[ATP]), allowing the determination of the maximum rate, V _max (the horizontal asymptote of the hyperbola). To determine the second‐order reaction rate constant, the dependence of the observed rates on the nucleotide concentration was fitted by a straight line at small concentrations of the nucleotide. The rate constant is the slope of the line. All data fits were performed with Origin 8 (OriginLab Corp, Northampton MA).

Supporting Information

Data on reaction rate constants and maximal rate of myosin–ATP reaction, and the illustration of the evolution of myosin intrinsic fluorescence upon MnATP binding and hydrolysis.

Supporting information

Supporting Information