Electrostatic interactions in the SH1-SH2 helix of human cardiac myosin modulate the time of strong actomyosin binding

Abstract

Two single mutations, R694N and E45Q, were introduced in the beta isoform of human cardiac myosin to remove permanent salt bridges E45:R694 and E98:R694 in the SH1-SH2 helix of the myosin head. Beta isoform-specific bridges E45:R694 and E98:R694 were discovered in the molecular dynamics simulations of the alpha and beta myosin isoforms. Alpha and beta isoforms exhibit different kinetics, ADP dissociates slower from actomyosin containing beta myosin isoform, therefore, beta myosin stays strongly bound to actin longer. We hypothesize that the electrostatic interactions in the SH1-SH2 helix modulate the affinity of ADP to actomyosin, and therefore, the time of the strong actomyosin binding. Wild type and the mutants of the myosin head construct (1–843 amino acid residues) were expressed in differentiated C₂C₁₂ cells, and the duration of the strongly bound state of actomyosin was characterized using transient kinetics spectrophotometry. All myosin constructs exhibited a fast rate of ATP binding to actomyosin and a slow rate of ADP dissociation, showing that ADP release limits the time of the strongly bound state of actomyosin. The mutant R694N showed a faster rate of ADP release from actomyosin, compared to the wild type and the E45Q mutant, thus indicating that electrostatic interactions within the SH1-SH2 helix region of human cardiac myosin modulate ADP release and thus, the duration of the strongly bound state of actomyosin.

Introduction

Myosin II is a nanoscale motor that transduces the chemical energy of ATP to generate force during muscle contraction and perform transport functions in non-muscle cells. ATP binds actomyosin and initiates its dissociation, and ATP hydrolysis drives myosin conformational change, or the recovery stroke, that primes myosin for subsequent strong binding to actin to produce the power stroke. The rate of ATP binding to actomyosin and the rate of ADP release determine the duration of the strong actomyosin binding, which is directly related to the average force produced by muscle [1]. General organization of the motor domain is conserved among myosins [2], the myosin fold provides a structural frame for the molecular motor function. Myosins have different functions, and their kinetic properties vary dramatically. Kinetics of the actomyosin cycle depends on myosin sequence, and therefore, on inter-residue interactions, which vary in different myosins. There are amino acid residues critical for myosin function, such as I499 in D. discoideum myosin, which couples the relay helix and the converter [3]. Mutation I499A inhibits the converter swing and abolishes myosin motility [3]. There are residues, less critical for myosin function, and mutagenesis of such residues modulates myosin kinetics (for example G691A in D. discoideum myosin [4]). Inter-residue interactions can assist or resist the transition between structural states, increasing or decreasing the rate of the myosin conformational change. Therefore, inter-residue interactions may regulate myosin function. Interatomic forces of physical origin, electrostatics and dispersion, maintain functional interactions between myosin structural elements. Since the dispersion force is a short-range force (r ⁻⁶ dependence), and the electrostatic force between charged residues is a long-range force (r ⁻² dependence), we hypothesize that electrostatic interactions between charged residues of myosin structural elements within the myosin head is the major determinant of myosin kinetics. In this work we focus on the SH1-SH2 helix region, the part of the myosin head, responsible for significant conformational change during the myosin ATPase cycle. Two helices of the force generation region, the relay helix and the SH1 helix, connect the converter domain with the rest of the myosin head (Figure 1). The relay helix extends from the myosin active site to the converter domain. During the power stroke, the bent relay helix unbends and rotates the converter domain 60° around the SH1 helix [5].

Figure 1.

The SH1-SH2 helix of the myosin head. Left panel, alpha isoform, right panel – beta isoform. Spheres – charged residues, sticks – permanent salt bridges. The bridge E700:R703 exists in both isoforms, the bridge E98:R703 in beta isoform switches to the bridge E98:R708 in alpha isoform. Bridges E98:R694 and E45:R694 exist only in the beta isoform.SH1 helix unwinds in beta isoform in the MD simulations [6].

Alpha (MYH6) and beta (MYH7) isoforms of human cardiac myosin share 93% sequence identity of the head domain but are different kinetically. ADP release from actomyosin is faster in the alpha isoform, resulting in the shorter duration of the strongly bound state of actomyosin. Usually, the sequence alignment is used to pick a site for a mutation to examine if the mutation site affects the molecule’s kinetic properties. We chose a different approach. Previously, based on the available crystal structure of the beta isoform (4DB1.pdb, UniProtKB P12883), we built a model of the alpha isoform and ran a molecular dynamics simulation for the myosin constructs of both isoforms [6]. We analyzed obtained trajectories in terms of electrostatic interactions within the myosin head. We found significant differences in the distribution of electrostatic interactions within the SH1-SH2 helix of the alpha and beta isoforms (Figure 1). In the beta isoform, we found a well-structured electrostatic network of permanent salt bridges, staying put during the whole trajectory. In the alpha isoform, our simulation showed a smaller number of permanent salt bridges, and therefore, higher flexibility of the construct.

The SH1-SH2 helix is well conserved among myosins and 100% conserved in the alpha and beta isoforms of human cardiac myosin. There are five charged residues in the SH2 and SH1 helices, R694, E700, R703, R706, K707 (MYH7 sequence). Four of these residues form permanent salt bridges in the beta isoform (Figure 1). The effect of mutagenesis of the residues E700, R703, R706 was studied in different myosins before. R706H is the known deafness mutation in nonmuscle myosin IIA (MYH9) [7], myosin with this mutation shows motility impairment [8], but undisturbed ATPase activity, indicating that this mutation uncouples the ATPase activity and force production. R703C results in decreased ATPase activity and reduced actin mobility in the in vitro motility assay [9]. E700K mutation [10] results in slow myosin ATPase activity and slow ATP-induced actomyosin dissociation [11]. According to our computer simulations, E700 and R703 form a permanent salt bridge, apparently essential for SH1 helix structural integrity during the myosin ATPase cycle. We observed this salt bridge in both isoforms. R703 and R706 form a permanent salt bridge with E98 in the beta and alpha isoforms accordingly. Highly conserved E98 is located at the N terminus of the helix C [12], and the mutation of the next residue, H97K (MYH7 sequence), introducing positive charge near the negatively charged E98, stops myosin ATPase activity and abolishes actin motility in the in vitro motility assay [9]. Myosin mutant H97K cannot bind actin strongly, presumably due to the disturbed release of the products of hydrolysis. The appearance of the positively charged residue next to the negative charge of E98, involved in the electrostatic network, can affect the network, and we suspect that this is the case for the H97K mutant. One can conclude that the electrostatics of the SH1 helix is important for proper myosin functioning since the SH1 helix is the pivoting helix in myosin power stroke conformational change. It is interesting to mention that disturbed electrostatics not only uncouples myosin ATPase activity and force production (R706) but affects myosin interaction with nucleotide (E700, R703). Newly developed myosin activator omecamtiv mecarbil binds myosin at the SH1 helix [13]. According to our computer simulations, R694 forms two permanent salt bridges with E98 and E45 only in the beta isoform, these salt bridges are not present in the alpha isoform. We hypothesized that removal of one salt bridge (using E45Q mutant) or both salt bridges (using R694N mutant) can modulate the kinetics of the nucleotide binding and retention and will affect the kinetics of the strongly bound state of actomyosin. We found previously that the mutation R694N accelerates the release of ADP from actomyosin [6]. In the current paper, we report on the kinetics of myosin mutant E45Q and compare obtained kinetic rates with the data for the R694N mutant. The goal of the work was to determine if the salt bridge R694:E45 or R694:E98 plays a role in the regulation of ADP dissociation from actomyosin. In our previous analysis, we assumed the rapid equilibrium between actomyosin, ADP, and ATP in the reaction when ATP-induced actomyosin dissociation is inhibited by ADP. We defined the rapid equilibrium as an equilibrium establishing on the timescale of actomyosin dissociation. Our current analysis shows that this is not the case, the true rate of ADP dissociation is modulated by the probability of actomyosin binding either ATP or ADP. Although the general conclusion of our previous paper [6] remains the same, in the current paper we revise the reaction rates of ADP dissociation from actomyosin. We compare obtained rates of ATP and ADP binding actomyosin with the calculated rate of the diffusion-controlled reaction and conclude that myosin’s active site may be partially closed. We hypothesize that the difference between the rates of ATP and ADP binding actomyosin may reflect the overall negative charge of the active site. Obtained rate constants allow us to analyze the range of the equilibrium rate constants for the two-state actomyosin·ADP complex, proposed before [14–16]. Even though our data do not provide any evidence of existence of two states of the actomyosin·ADP complex, the analysis of the obtained kinetic rates suggest that mutations E45Q and R694N shift equilibrium of actomyosin·ADP population toward just one actomyosin·ADP state.

Materials and Methods

Reagents

N-(1-pyrene)iodoacetamide (pyrene) was from Life Technologies Corporation (Grand Island, NY), phalloidin, ATP, and ADP were from Sigma-Aldrich (Milwaukee, WI). All other chemicals were from ThermoFisher Scientific (Waltham, MA) and VWR (Radnor, PA).

Protein preparation

The β isoform construct of human cardiac myosin motor domain contains 1–843 residues and FLAG affinity tag at the C-terminus. Adenoviruses encoded with the wild type and R694N myosin mutant were purchased from Vector Biolabs (Malvern, PA), amplified using HEK293 cells (ATCC CRL-1573), and purified using CsCl gradient centrifugation. Recombinant human cardiac myosin was expressed in C₂C₁₂ (ATCC CRL-1722) mouse myoblast cells. C₂C₁₂ cells were grown to a 95% confluence on 15 cm diameter plates and infected with the optimum dosage of virus determined by a viral-titration assay. Cells were allowed to differentiate post-infection and collected seven days post-infection to extract and purify myosin. Collected cells were washed and lysed in the presence of a millimolar concentration of ATP. The cell lysate was incubated with anti-FLAG magnetic beads (Sigma-Aldrich, Milwaukee, WI). Beads were washed and myosin was eluted from the beads by 3x FLAG peptide (ApexBio, Houston, TX). Myosin purity was assessed by Coomassie-stained SDS-polyacrylamide gels and protein concentration was determined by measuring the absorbance at 280 nm using the extinction coefficient ε_280nm = 93,170 M⁻¹cm⁻¹, calculated using the ProtParam tool of ExPASy web server.

Actin was prepared from rabbit leg and back muscles [17]. F-actin was labeled with pyrene iodoacetamide (Life Technologies Corporation, Grand Island, NY) with the molar ratio 6:1, label:actin. After labeling, actin was cleaned from the excess of label, re-polymerized, stabilized with phalloidin at the molar ratio of 1:1, and dialyzed for two days at T=4°C against the experimental buffer. Concentration of unlabeled G-actin was determined spectrophotometrically assuming the extinction coefficient ε_290nm = 0.63 (mg/ml)⁻¹cm⁻¹ [18]. Concentration of labeled G-actin and labeling efficiency were determined spectroscopically using the following expressions: [G-actin]=(A_290nm–(A_344nm·0.127))/26,600 M⁻¹ and [pyrene]=A_344nm /22,000 M⁻¹ [19]. Pyrene labeling efficiency of actin was usually about 70%. The experimental buffer contained 20 mM MOPS (3-[N-morpholino]propanesulfonic acid) pH 7.3, 50 mM KCl, 3mM MgCl₂ total concentration. Since log₁₀K_A for MgATP is 4.29 [20], where K_A is the association constant, 3mM MgCl₂ chelated all ATP used in our experiments, since used ATP concentration was 0.9 mM or less. We do not expect any measurable effect from the KATP complex since the association constant for such a complex is three orders of magnitude smaller than the constant for MgATP [21]. All reported concentrations are final concentrations.

Acquisition of fluorescent transients

In the ATP-induced actomyosin dissociation experiment, usually 0.5 μM actomyosin was rapidly mixed with ATP solution of variable concentrations. In the ADP inhibition of the ATP-induced actomyosin dissociation experiment, 0.5 μM actomyosin was rapidly mixed with the premixed ATP and ADP solution. The concentration of ATP in solution was 0.6 mM or 0.9 mM and the concentration of ADP varied from 20 μM to 200 μM. Transient fluorescence of pyrene-labeled actin was measured with a Bio-Logic SFM-300 stopped flow transient fluorimeter (Bio-Logic Science Instruments SAS, Claix, France), equipped with an FC-15 cuvette. The pyrene fluorescence was excited at 365 nm and detected using a 420 nm cutoff filter. Multiple transients were acquired and averaged to improve the signal to noise ratio. 8000 points were acquired in each experiment. All experiments were performed at T=20° C. Transient experiments were performed with myosin constructs from at least three independent preparations.

Analysis of fluorescence transients

The transients obtained in each experiment were fitted by the single exponential function S(t) = S_o+A·exp(−k_obs·(t-t₀)), or the double exponential function S(t) = S_o+A₁·exp(−k_obs1·(t-t₀)) + A₂·exp(−k_obs2·(t-t₀)). S(t) is the observed signal at the time t, A_i is the signal amplitude, t₀ is the time before the flow stops, and k_obsi is the observed rate constant. Transients, obtained for the same actomyosin preparation at different concentrations of the nucleotide were fitted together, assuming the known value of t₀, measured in a separate experiment, and the constant value of S₀, which depends on the concentration and labeling efficiency of pyrene-labeled actin in the actomyosin preparation. In the case of the double exponential global fit, we kept amplitudes of the transients constrained, A₁+A₂=const, to account for the conservation of mass. Transients of the ATP induced actomyosin dissociation were fitted with the single exponential function to determine the rate constant k_obs. The dependence of the observed rates on the ATP concentration was fitted by a hyperbola, v = V_max·[ATP]/(K_app +[ATP]), allowing the determination of the maximum rate, V_max (the horizontal asymptote). In the case of the ATP-induced actomyosin dissociation, the rate constant k_+2T is the V_max and the equilibrium constant of the collision complex formation K_1T is 1/K_app. To determine the bimolecular rate (K_1Tk_+2T, Scheme 1), the dependence of the observed rates on the ATP concentration was fitted by a straight line at small concentrations of ATP. All data fits with an exponential function, hyperbola, and polynomial were performed with Origin 8 (OriginLab Corp, Northampton MA). The statistical significance of results was tested with ANOVA integrated into Origin 8 software. A significance level of P < 0.05 was used for all analyses. We also fitted the acquired transients to the time-dependent numerical solution of differential equations corresponding to Scheme 1 and Scheme 2. The equations are shown in the Supporting Information. The numerical solution was obtained using the built-in symbol NDSolve in Wolfram Mathematica. The solution was fitted to an acquired transient using the built-in symbol NMinimize by minimizing the residual sum of squares. The scripts are shown in the Supporting Information. The goodness of the fit was assessed with the Pearson χ² test, χ² was 98%–99% in all fits. Representative fits of the experimental data are shown in Figures 6 and 7. The signal intensity of transients was expressed in the units of actin concentration based on the assumption that all actomyosin is dissociated at the end of the reaction. The assumption is supported by our experimental observation that all transients of the same protein preparation have the same final amplitude of pyrene fluorescence, indicating complete dissociation of actomyosin at the end of the reaction. Transients obtained for the same actomyosin preparation were fitted globally. Data of at least three biological replicates were fitted, the obtained rate constants were averaged, and mean values and standard deviations are reported.

Scheme 1.

ATP-induced actomyosin dissociation. A = pyrene-labeled actin, M = myosin, T = ATP. A* = actin with unquenched pyrene fluorescence.

Scheme 2.

ATP-induced actomyosin dissociation, competitive inhibition with ADP.

Figure 6.

Typical transients in the ATP-induced actomyosin dissociation experiment. Actomyosin (0.5 μM) is rapidly mixed with ATP (900 μM, upper trace, black dots, or 60 μM, lower trace, blue dots). The transients are fitted to the numerical solution of differential equations (Eq. S1), red curves. All transients obtained for the same myosin preparation are fitted globally with the same set of kinetic constants. A, WT myosin, B, R694N myosin mutant, C, E45Q myosin mutant.

Figure 7.

Typical transients in the experiment when actomyosin (0.5 μM) is rapidly mixed with the premixed ATP and ADP. [ATP] = 900 μM (WT and R694N) and 600 μM (E45Q). [ADP] = 0 μM, upper trace, black dots, and [ADP] = 100 μM, lower trace, blue dots. The transients are fitted to the numerical solution of differential equations (Eq. S2), red curves. All transients obtained for the same myosin preparation are fitted globally with the same set of kinetic constants. A, WT myosin, B, R694N myosin mutant, C, E45Q myosin mutant.

Results

Design and preparation of myosin S1 constructs.

We choose residues E45 and R694 for mutagenesis because they participate in the isoform-specific electrostatic interactions within the SH1-SH2 helix of the myosin head. The roles of other charged residues in the region were studied previously [7–11]. We prepared constructs of the wild type and two mutants, E45Q and R694N, of the human cardiac myosin motor domain (1–843 amino acid residues long) in the beta isoform background. All constructs have the FLAG tag at the C-terminus. We used differentiated C₂C₁₂ cells to express constructs and FLAG tag affinity chromatography to purify expressed constructs. The purity of the expressed constructs was confirmed with SDS PAGE (Figure 2).

Figure 2.

SDS-PAGE of the purified recombinant myosin head. 98 kDa human cardiac S1 co-purifies with murine ELC and RLC.

ATP-induced actomyosin dissociation.

We monitor ATP-induced actomyosin dissociation using fluorescence of pyrene labeled actin. In the absence of ATP, actin strongly binds myosin, and pyrene fluorescence is quenched. Pyrene fluorescence increases upon actomyosin dissociation and formation of the weakly bound actomyosin complex [22, 23]. Therefore, pyrene fluorescence reports the lifetime of the strongly bound state of actomyosin. ATP binds actomyosin in a two-step process (Scheme 1). The first step is the rapid equilibrium when strongly bound actomyosin forms a collision complex with ATP, and k_−1T > k_+2T. Pyrene fluorescence does not change during this step. Upon isomerization of the actomyosin·ATP collision complex, ATP binding results in irreversible dissociation of actomyosin. In our experiments, the formation of the actomyosin complex (prepared by mixing of expressed myosin constructs with the pyrene-labeled rabbit skeletal actin) led to decreased pyrene fluorescence. This decrease was similar for all myosin constructs, indicating strong binding of actin and expressed myosin constructs. When prepared actomyosin is rapidly mixed with ATP, the time course of pyrene fluorescence follows single exponential kinetics. The final amplitudes of all transients, obtained for the same preparation of actomyosin, were the same for all used concentrations of ATP, showing complete dissociation of actomyosin complex and confirming that the ATP-induced dissociation is irreversible. Figure 3 shows a fluorescence transient observed for the WT myosin at 20°C when 0.5 μM actomyosin is mixed with 900 μM ATP in the stopped-flow fluorimeter (all reported concentrations are in the final mixture, here and throughout the text). The observed rate constants for the WT myosin and the mutants depend on ATP concentration hyperbolically (Figure 4), the reaction rate constant k_+2T and the association constant of the collision complex K_1T can be determined from the fit to a hyperbola. The second-order association rate constant (K_1Tk_+2T) is determined at small concentrations of ATP (Figure 5) when the dependence of the reaction rate on ATP is linear [24]. We determined all kinetic constants, shown in Scheme 1 in the fit of obtained transients to the numerical solution of differential equations (Eq. S1) (Table 1). We calculated constants K_1T and K_1Tk_+2T to compare with the constants obtained in the fit of the transients to a single-exponential function, and the following fit to a hyperbola and a straight line. We found a good agreement of the kinetic constants, obtained by two fitting methods. For the WT myosin, K_1T was 4.5±1.4 mM⁻¹ and 11.2 ± 2.4 mM⁻¹, obtained in the fit to a hyperbola and in the fit to the numerical solution of differential equations, respectively. For the R694N mutant, the equilibrium constant K_1T was 7.8±2.6 mM⁻¹ and 6.8 ± 1.8 mM⁻¹, respectively. For the E45Q mutant, the equilibrium constant K_1T was 7.3±1.9 mM⁻¹ and 12.3 ± 2.8 mM⁻¹, respectively. The rate k_+2T for the WT myosin was determined as 491.5±74.1 s⁻¹ in the fit to a hyperbola and 406.2 ± 14.1 s⁻¹ in the fit to the solution of differential equations. For the R694N mutant, the rate constants were 338.8±47.8 s⁻¹ and 449.1 ± 55.5 s⁻¹, respectively. For the E45Q mutant, the rate constants were 618.5±75.3 s⁻¹ and 619.6 ± 52.3 s⁻¹, respectively. The second-order association rate constant K_1Tk_+2T for the WT myosin was determined as 2.12±0.21 μM⁻¹s⁻¹ from the linear fit of the reaction rates at small ATP concentrations, and as 4.56 ± 0.98 μM⁻¹s⁻¹ from the fit to the solution of differential equations respectively. For the R694N mutant, the rate constants were 1.85±0.03 μM⁻¹s⁻¹ and 3.03 ± 0.88 μM⁻¹s⁻¹ respectively. For the E45Q mutant, the rate constants were 3.26±0.31 μM⁻¹s⁻¹ and 7.61 ± 1.86 μM⁻¹s⁻¹, respectively. Smaller values of the equilibrium constant K_1T for the WT and the E45Q mutant likely reflect a slight deviation from the assumption of the rapid equilibrium condition [24]. Smaller values of the rate constant K_1Tk_+2T obtained from the linear fit of the reaction rates at small ATP concentrations likely reflect an excessive ATP concentration, which ideally should be [ATP] < 0.01·K_app [24]. The rate constant k_−2T is below 10 s⁻¹ for all studied myosin constructs, as anticipated [25]. Representative fits for all studied myosin constructs are shown in Figure 6.

Figure 3.

Typical transients in the experiment of the ATP-induced actomyosin dissociation with and without ADP. Actomyosin (0.5 μM) rapidly mixed with ATP (upper trace, black), or the mixture of ATP and ADP (lower trace, blue). [ATP] = 900 μM in both cases. [ADP] = 200 μM when present in the mixture. Final pyrene fluorescence is the same, indicating complete actomyosin dissociation regardless of the concentration of ADP used. Fitting curves (red) are single exponential function (upper trace, the actomyosin + ATP experiment) and double exponential function (lower trace, the actomyosin + (ATP+ADP) experiment). Deadtime, measured in a separate experiment, constrains the fit (all fits intercept at the mixing time in the bottom left corner). The vertical dashed line shows the time of the flow stop and the beginning of the fit. Meaningful kinetic traces lay on the right side of the dashed line. On the left, there are flow artifacts.

Figure 4.

Rate of ATP-induced actomyosin dissociation. Black circles, WT, N=3, red squares, R694N mutant, data from [6], N=3 (the point at [ATP]=600 μM, N=1), blue triangles, E45Q mutant, N=3. Reaction rates fitted with a hyperbola, k_+2T = 491.5±74.1 s⁻¹, 338.8±47.8 s⁻¹, 618.5±75.3 s⁻¹ for WT, R694N, and E45Q respectively. Data points are mean ± SD. N is the number of biological replicates.

Figure 5.

ATP-induced actomyosin dissociation. Observed reaction rates at low [ATP] are fitted by a straight line, the second-order reaction rate constant is determined from the slope of the line. Black circles, WT, N=3, and red squares R694N mutant, N=3, data from [6]. Blue triangles, E45Q mutant, N=3. K_1TK_+2T = 2.12±0.21 μM⁻¹s⁻¹, 1.85±0.03 μM⁻¹s⁻¹, 3.27±0.31 μM⁻¹s⁻¹ for WT, R694N, and E45Q respectively. Data points are mean ± SD. N is the number of biological replicates.

Table 1.

Actomyosin kinetic rate constants, obtained in the fit of the transients to the numerical solution of differential equations Eq. S1 and Eq. S2, mean ± SD. Data are averages of three independent protein preparations.

	WT(a)	R694N(a)	E45Q
Actomyosin dissociation
k+1T, μM−1s−1	7.4 ± 1.4	8.9 ± 2.1	11.2 ± 1.4
k−1T, s−1	661.0 ± 72.9	1312.5 ± 160.6	915.1 ± 178.6
k+2T, s−1	406.2 ± 14.1	449.1 ± 55.5	619.6 ± 52.3
k−2T, s−1	3.6 ± 2.5	0.008 ± 0.009	1.8 ± 0.3
K1Tk+2T, μM−1s−1	4.6 ± 1.0	3.0 ± 0.9	7.6 ± 1.9
Actomyosin collision complex formation
K1T, mM−1	11.2 ± 2.4	6.8 ± 1.8	12.3 ± 2.8
ADP binding to and release from actomyosin
k+1D, μM−1s−1	26.6 ± 8.1	94.7 ± 52.1	82.6 ± 4.3
k−1D, s−1	199.6 ± 51.8	776.1 ± 285.3	301.3 ± 74.3
ADP dissociation from actomyosin
K1D, μM	7.5 ± 3.0	8.2 ± 5.4	3.6 ± 0.9

^(a)Raw data from [6], refitted with the Eq. S1 and Eq. S2.

ADP dissociation from actomyosin.

ADP has a high affinity to the WT human cardiac actomyosin. The equilibrium dissociation constant of ADP is in the micromolar range [6, 26]. To measure the rate of ADP dissociation from actomyosin we rapidly mix pyrene-labeled actomyosin with premixed ADP and ATP. In our experiments, we kept ATP concentration constant (near saturation, but not saturated, 600 μM or 900 μM) and vary ADP concentration from 20 μM to 200 μM. Upon mixing, actomyosin can bind either ATP or ADP and form either an actomyosin·ATP or actomyosin·ADP complex (Scheme 2). ATP binding results in actomyosin dissociation and increased pyrene fluorescence. ADP binding does not dissociate actomyosin and therefore does not change the intensity of pyrene fluorescence. If actomyosin, actomyosin·ATP, and actomyosin·ADP complexes are in rapid equilibrium, compared to the rate of ATP induced actomyosin dissociation, we should observe a single-exponential increase of pyrene fluorescence, corresponding to the competitive inhibition of the ATP-induced actomyosin dissociation. At the saturated concentration of ATP, there should be no ADP dependence on the rate of actomyosin dissociation [24]. In our experiments with the constructs of human cardiac myosin, we usually observe the double-exponential kinetics of ATP-induced actomyosin dissociation in the presence of ADP (Figure 3). Our observation of the double-exponential kinetics suggests that there is no rapid equilibrium in the mixture of actomyosin, ATP, and ADP on a timescale of actomyosin dissociation. ADP readily binds actomyosin, and the rate of ADP dissociation from actomyosin is slower or comparable with the rate of ATP-induced actomyosin dissociation. The amplitude of the fast component decreases with the increase of ADP concentration in the mixture (Figure S1). This decrease in the amplitude shows that the probability of ATP binding decreases. The range of ADP concentrations in this experiment was suggested by the resolution of kinetic components of a transient. At small concentrations of ADP ([ADP] < 20 μM) we observed a single-exponential transient with the rate constant corresponding to the rate of ATP-induced actomyosin dissociation. This observation shows that the probability of ATP binding is higher due to a significantly higher concentration of ATP. At large concentrations of ADP ([ADP] > 200 μM) we observed a single-exponential transient with the rate, slower than the rate of ATP-induced actomyosin dissociation. This observation shows that at high concentration ADP preferentially binds actomyosin instead of ATP, and the rate of ATP-induced actomyosin dissociation is modulated by the rate of ADP binding, and, most likely, by several consecutive ADP bindings. The rate of the fast component of the observed double-exponential kinetics does not depend on ADP concentration. We conclude that there is no fast exchange between ADP and actomyosin·ADP complex in the beta isoform actomyosin, wild type, and mutants. The rate of the slow component of double-exponential transient depends on the concentration of ADP and reflects ADP binding to actomyosin, followed by ADP dissociation and subsequent ATP binding, resulting in actomyosin dissociation. To determine the rate of ADP binding to and ADP dissociation from actomyosin we fit the acquired transients globally to the numerical solution of differential equations, corresponding to Scheme 2 (Eq. S2). In the fit, we used the rate constants, previously determined for the ATP-induced actomyosin dissociation reaction, and varied only the rates k_+1D and k_−1D. The fit shows that ADP dissociates faster from actomyosin with mutant R694N than from actomyosin with WT and E45Q constructs, k_−1D = 199.6 ± 51.8 s⁻¹ for the WT myosin, and 776.1 ± 285.3 s⁻¹ and 301.3 ± 74.3 s⁻¹ for the mutants R694N and E45Q respectively. ADP binds actomyosin slower in the case of WT myosin than the mutants, k_+1D is 26.6 ± 8.1 μM⁻¹s⁻¹, 94.7 ± 52.1 μM⁻¹s⁻¹, and 82.6 ± 4.3 μM⁻¹s⁻¹ for the WT, R694N, and E45Q myosin constructs respectively. As expected, the equilibrium constant of ADP dissociation is in the micromolar range for all myosin constructs, in a good agreement with the previous reports [26, 27]. Representative fits for all studied myosin constructs are shown in Figure 7.

Discussion

In this study, we experimentally verified our hypothesis that electrostatic interactions in the SH1-SH2 helix of the myosin head modulate the rate of ADP dissociation from actomyosin. A single mutation, R694N, disrupts electrostatic interaction between SH2 helix and the helix C of the myosin head, which exists only in the slow beta isoform of human cardiac myosin. The mutant R694N shows a faster rate of ADP dissociation from actomyosin, and therefore, shorter duration of the strongly bound actomyosin state, t_s. The timing of the strongly bound state depends on how fast actomyosin exits the state, t_s = 1/k_+2T + 1/k_−1D, where k_+2T is the rate of ATP-induced actomyosin dissociation, and k_−1D is the rate of ADP dissociation from actomyosin. For the WT beta isoform human cardiac myosin t_s = 7.5±1.3 ms, for the R694N mutant t_s = 3.5±0.5 ms, 47% shorter. Mutation E45Q, disrupting the isoform-specific salt bridge R694:E45, which is responsible for the electrostatic interactions between SH2 helix and SH3-like domain of the myosin head, also leads to the shorter timing of the strongly bound state of actomyosin, t_s = 4.9±0.8 ms. Structural details of such regulation of myosin kinetics are yet to be determined. If in the WT and the E45Q constructs the rate k_−1D of ADP dissociation from actomyosin is slower than the rate k_+2T of ATP-induced actomyosin dissociation, and therefore, the time of ADP release governs the time of the strongly bound state of actomyosin. In the R694N construct, the rate k_−1D of ADP release from actomyosin statistically equal to the rate k_+2T of ATP-induced actomyosin dissociation, then the time of the strongly bound state is determined by two of these processes.

The diffusion-controlled rate of ATP and actomyosin association is k_diff = 255μM⁻¹s⁻¹, this rate is calculated using the expression k_diff = 4·D_ATP·a [28], where a is the radius of myosin active site (we assume it equals to 0.4 nm, the radius of ATP), and D_ATP is the diffusion coefficient of ATP, D_ATP = (2.65±0.09)·10⁻¹⁰ m² ATP s⁻¹ at T=20°C [29]. Assumption of the larger diameter of the active site makes the rate of association faster, therefore, the value k_diff = 255 μM⁻¹s⁻¹ is the minimal value for the rate of the diffusion-controlled reaction of ATP and actomyosin. The expression k_diff = 4·D_ATP·a follows from the consideration of the diffusion-controlled bimolecular reaction between the large macromolecule with a single ligand-binding site and a small uniformly reactive ligand [28, 30]. The expression is accurate when (a) the diffusion of actomyosin is much slower than the diffusion of ATP, (b) the active site is a small part of the surface of actomyosin, and (c) the diameter of ATP is much smaller than the diameter of the actomyosin complex.

The hyperbolic dependence of ATP-induced actomyosin dissociation indicates that ATP binds actomyosin in two steps, first forming the collision actomyosin·ATP complex, which is in rapid equilibrium with ATP and actomyosin, and then binding actomyosin practically irreversibly, causing actomyosin dissociation. Both mutants and the WT myosin construct exhibit that hyperbolic dependence. Our fits show that the rate of the collision complex formation k_+1T is slower than the rate of the diffusion-controlled bimolecular reaction for both mutants and the WT actomyosin. This can be interpreted as if the active site is not open all the time and is in equilibrium between the open and closed states. It is interesting to mention that our fits give the rate of the collision complex dissociation k_−1T of the same order of magnitude that the rate of the ATP-induced actomyosin dissociation k_+2T. Knowing the rate of the collision complex dissociation, one can determine the half-lifetime of the collision complex as t_1/2 = ln(2)/k_−1T. t_1/2 = 1.0 ± 0.1 ms for the WT myosin, and 0.5 ± 0.06 ms and 0.7 ± 0.1 ms for R694N and E45Q mutants respectively.

Assuming millimolar concentration of ATP in muscle [31], the decrease of the equilibrium constant K_1T from about 11 mM⁻¹ in the case of the WT myosin to 7 mM⁻¹ in the case of the R694N construct virtually does not change the population of the actomyosin·ATP complex.

Our fits show that ADP binds actomyosin faster than ATP (Table 1), possibly reflecting the difference in charge of these molecules. At pH 7 and higher, both MgADP and MgATP are ionized [32], and the total charge of MgADP and MgATP is (−1e), and (−2e) accordingly. This charge dependence of the kinetics of nucleotide binding may indicate an overall negative charge of the myosin active site, and therefore electrostatic repulsion when nucleotide binds actomyosin.

When we mix actomyosin with the mixture of ATP and ADP, the actomyosin dissociation follows double-exponential kinetics. This indicates that there is no rapid equilibrium between actomyosin and the actomyosin·ADP complex on the timescale of ATP binding actomyosin. The fast exponential component of the reaction corresponds to ATP-induced actomyosin dissociation, and the slow component corresponds to ADP binding to actomyosin and therefore delayed ATP-induced actomyosin dissociation. There is no [ADP] dependence of the fast component of actomyosin dissociation, this also confirms the absence of rapid equilibrium between actomyosin and actomyosin·ADP complex. To determine true rates of ADP binding to and dissociation from actomyosin, we fit acquired transients to the numerical solution of differential equations, corresponding to Scheme 2 (Eq. S2). The fits show that the rates of ADP binding to actomyosin are slower than the rate of the diffusion-controlled reaction. For the WT myosin construct, the rate of ADP dissociation from actomyosin k_−1D is two times slower than the rate of irreversible binding of ATP and actomyosin k_+2T. The rate of ADP dissociation k_−1D for the mutant R694N is statistically faster than for the WT construct, confirming our hypothesis that electrostatic interactions within the SH1-SH2 helix modulate ADP release from actomyosin. Observed rates are close to the rates reported for the cardiac myosin S1 from different species [33–36] and close to the kinetic difference between the alpha and beta isoforms of cardiac myosin [34].

Since ADP is similar to ATP in size and structure, we may suggest that ADP binds and releases from actomyosin in two steps (Scheme 3), similar to ATP. The hypothesis of the several sequential actomyosin·ADP states in the actomyosin cycle was discussed before [14, 15, 37–39]. Our data do not provide support for this hypothesis, but the kinetic rates, obtained in our experiments, allow us to conclude on how mutations R694N and E45Q change the population of these consecutive states and the equilibrium constant of these states, K_2D. Our analysis uses determined rates k_+1D and k_−1D, the rate of the diffusion-controlled reaction of nucleotide binding to actomyosin k_diff, and condition k_−1D ≤ k_−+2T due to the absence of rapid equilibrium between actomyosin, actomyosin·ADP, and actomyosin·ATP in reaction. Assuming the two-step mechanism, shown in Scheme 3, and assigning the superscripts ⁽¹⁾ and ⁽²⁾ to distinguish between the rates of one-step and two-step reaction, the rate k⁽²⁾_+1D should be smaller than the rate k_diff and larger than the rate k⁽¹⁾_+1D, or k⁽¹⁾_+1D ≤ k⁽²⁾_+1D ≤ k_diff. The rate k⁽²⁾_−1D should be larger than the rate k⁽¹⁾_−1D, but smaller than the rate k⁽¹⁾_+1T to satisfy the condition of absence of rapid equilibrium, observed in the experiment, or k⁽¹⁾_−1D ≤ k⁽²⁾_−1D ≤ k⁽¹⁾_+2T. The total equilibrium constant of the two-step reaction should be the same as the equilibrium constant of the one-step reaction, K⁽²⁾_2D·K⁽²⁾_1D = K⁽¹⁾_1D, where K⁽²⁾_2D is the equilibrium constant of the complex (AMD)₂ formation from the complex (AMD)₁ (Scheme 3), and K⁽²⁾_1D is the equilibrium constant of the complex (AMD)₁ formation in the reaction when ADP binds actomyosin. Then

$$K_{2D}^{(2)}{K}_{1D}^{(2)}=\left(\frac{k_{+2D}^{(2)}}{k_{-2D}^{(2)}}\right)\left(\frac{k_{+1D}^{(2)}}{k_{-1D}^{(2)}}\right)=\frac{k_{+1D}^{(1)}}{k_{-1D}^{(1)}}=K_{1D}^{(1)}.$$

Scheme 3.

ATP-induced actomyosin dissociation, inhibition with ADP. Two sequential states of the actomyosin ADP complex considered. Superscript⁽²⁾ indicates rate constants of the two-step reaction mechanism.

The population of the intermediate state (AMD)₁ is determined by the dissociation constant 1/K⁽²⁾_2D, reflecting dissociation of (AMD)₂ into (AMD)₁. The equilibrium constant 1/K⁽²⁾_1D, reflects dissociation of (AMD)₁ into actomyosin and ADP. Using considered range of rate constants k⁽²⁾_+1D and k⁽²⁾_−1D and known k⁽¹⁾_+1D and k⁽¹⁾_−1D the value of equilibrium constant 1/K⁽²⁾_2D is in the range 0.49±0.13 – 9.58±2.92 for the WT myosin, 0.49±0.13 – 3.09±0.16 for the mutant E45Q, and 1.73±0.67 – 4.65±3.13 for the R694N mutant. Corresponding range of the equilibrium constant 1/K⁽²⁾_1D is 15.27±4.68 μM – 0.78±0.20 μM for the WT myosin, 4.74±2.67 μM – 1.76±0.22 μM for the mutant R694N, and 7.50±0.74 μM – 1.18±0.29 μM for the mutant E45Q. Since the equilibrium constant 1/K⁽²⁾_2D reflects formation of the (AMD)₁ complex from the (AMD)₂ complex and the constant 1/K⁽²⁾_1D reflects dissociation of ADP from the complex (AMD)₁ in two-step reaction, it is clear that in the case of the WT myosin, the complex (AMD)₁ is more populated, than the complex (AMD)₁ of both mutants.

Conclusion.

Our experiments support the hypothesis that the electrostatic interactions in the SH1-SH2 helix within the head of human cardiac myosin modulate the rate of ADP release from actomyosin, and thus, the time of the strongly bound state of actomyosin. Obtained rates of ATP and ADP binding to actomyosin suggest that the active site can be partially closed, limiting the rate of nucleotide binding. The difference in the rates of ATP and ADP binding to actomyosin may indicate that the overall charge of the nucleotide binding site is negative.