Motor domain constructs

PCR products amplified from human heart cDNA were cloned into the MCS of pShuttle-CMV modified to seamlessly add a 6xHistidine tag in frame with the C-terminus of S1, where an AflII site was silently coded into the conserved S1 C-terminal amino acid sequence LKSA. Translation is terminated after this sequence in S1 such that it includes Met1-Ala843. The respective α protein terminates at the same conserved amino acid, but due to sequence differences is slightly longer, including Met1-Ala845. Unmodified pShuttle-CMV was used in cloning β short-S1 (β-sS1) where no tag was added and translation terminated after the ELC-binding IQ domain at Arg808. The human ventricular ELC, MYL3, was cloned into unmodified pShuttle-CMV and an N-terminal 6xHistidine tag was added by sequential elongating PCR steps.

Adenovirus production

Adenovirus production was performed using the AdEasy kit (Qbiogene) with modifications [¹³]. Briefly, after cloning each construct into pShuttle-CMV, the shuttle vector was linearized with PmeI and homologously recombined with pAdEasy in bacteria. Successfully recombined plasmids were linearized with PacI and transfected into HEK293 cells stably expressing the E1 protein to complement pAdEasy for replication competence. Infected cell lysates were used to infect increasing numbers of cells, then virus isolated from the lysates by sequential step and linear CsCl gradients. Purified virus was stored at −20°C in 100 mM Tris pH 7.5, 250 mM NaCl, 1 mM MgCl₂, 1 mg/ml BSA, 50% glycerol.

Cell culture and protein expression

Murine C₂C₁₂ myoblasts (ATCC) were cultured in Dulbecco’s modified Eagle’s Medium (DMEM) (Gibco) with 10% fetal bovine serum (FBS) (Hyclone). Cells were differentiated into myotubes at ~100% confluence in 2% horse serum media (Gibco-BRL). Forty-eight hours later, myotubes were infected with 1 × 10⁶–1 × 10⁸ plaque-forming units of adenovirus per 100 mm plate. Approximately 4 days after infection, the cells were collected in 250 μl per plate of lysis buffer (20 mM Imidazole pH 7.0, 100 mM NaCl, 50 mM Tris, 3 mM ATP, 0.5% Tween-20, 1 mM DTT, 1× Complete, EDTA-free protease inhibitor cocktail [Roche]) and lysed using a Dounce homogenizer. Lysates were clarified by centrifugation in a Beckman 50 Ti rotor at 30 kRPM for 30 min. Clarified lysates were run through 1-ml HisTrap HP columns (GE Healthcare) by FPLC using an AKTA Purifier (GE Healthcare). Single peak fractions were isolated by 300 mM Imidazole step gradient elution. After dialysis into 25 mM Imidazole pH 7.0, 10 mM KCl, 4 mM MgCl₂, 1 mM DTT, eluates were further purified by FPLC through 1-ml HiTrapQ HP columns (GE Healthcare). Single peak fractions were isolated at ~300 mM NaCl by a 20-column volume linear gradient elution. Purified proteins were stored at 4°C on ice and dialyzed into the appropriate reaction buffers before use.

Proteins derived from muscles

Cardiac myosin was prepared from a heart of an adult mouse (CD1) according to Margossian and Lowey [²²]. S1 was prepared from myosin by digestion with chymotrypsin as described by Weeds and Taylor [²³]. The mouse α-S1 was not stable for more than few days on ice and was therefore stored in the presence of an EDTA-free protease inhibitor cocktail (Roche), and used without extensive purification within 48 h of digestion by chymotrypsin. Actin was prepared from rabbit muscle as described by Spudich and Watt [²⁴] and labeled with pyrene iodoacetamide as described by Criddle et al. [²⁵].

Transient kinetics

Quenched-flow experiments were performed using a Hi-Tech RQF-63. All measurements were made using a buffer containing 20 mM MOPS pH 7.0, 100 mM KCl, 5 mM MgCl₂. S1 was mixed with ten-fold excess ATP and incubated at 20°C for various time points (10–500 ms), then quenched by 1:1 addition of 6.25% (w/v) trichloroacetic acid. After neutralization with NaOH and a clarification spin at 3,000 × g for 5 min, ADP and ATP were separated for each time point by HPLC using a Hypersil ODS (3 μm) column by isocratic flow, and quantified by integration of peak areas. For each time point, the ratio of ADP over total nucleotide concentration was used to calculate the hydrolysis rate constant for S1.

Stopped-flow experiments were performed essentially as described previously [²⁶] using a HiTech Scientific SF-61DX2 stopped flow system. All measurements were made at 20°C (unless indicated otherwise) in 20 mM MOPS pH 7.0, 100 mM KCl, 5 mM MgCl₂, 1 mM DTT. Fluorescence transients were measured using intrinsic S1 tryptophan fluorescence (excitation at 295 nm, emission through a WG320 filter) or pyrene-labeled actin (excitation 365, emission through a KV389 filter) [²⁶]. Fluorescence transients were fitted to single or double exponential equations:

orand the fitted values of k_obs and the amplitudes were used to deduce rate and equilibrium constants of the events in the cross-bridge cycle as defined in Fig. 1. k_obs values were plotted against the varied nucleotide or S1 concentrations and fitted to the appropriate equation. Concentration in the text and figures refer to the concentration after mixing 1:1 unless otherwise stated.

The outline of the ATPase cycle is shown in Fig. 1. In each case, the steps in the cycle are known to be more complex. For example, ATP binding, ADP release, and P_i release each consist of at least two steps: a diffusion-limited fast equilibrium step and a protein isomerization or conformational change. Similarly, the ATP hydrolysis step consists of at least two steps: the closure of switch 2 and the associated lever arm swing, which is known as the recovery stroke, occurs before ATP cleavage into ADP and P_i. However, the simpler outline of Fig. 1 is sufficient for most of the data presented here except as outlined below in Scheme 1.

In Scheme 1, T and D represent ATP and ADP and where actin·S1 exists in two conformations (A·M and A·M′) in equilibrium. A·M′ is unable to bind nucleotide and must isomerize to A·M before ATP can bind. A similar pair of conformations exist in the presence of bound ADP, where A·M′·D must isomerize to A·M·D before ADP can dissociate.

For Figs. 3b, 4d, and 6b, the dependence of k_obs on ATP concentration is defined by:

where k_max is the maximum value of k_obs and K_0.5 is the nucleotide concentration required to give k_obs = k_max/2.

For the ATP dissociation of A·M in Fig. 4a, d [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K_{\rm max} = K^{\prime}_{+2} $$\end{document}], and [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K_{\rm max} = K^{\prime}_{1}$$\end{document}]K_0.5 = 1/K′. For Fig. 6b the ATP binding to S1, the ATP binding is considered to be irreversible and the data is not expected to conform to a hyperbola. In this case the fit is used to estimate the value of k_max, which can be either k₊₂ or k₊₃ + k₋₃ and the initial slope of the plot defines K₁k₊₂.

For Fig. 4a the competitive inhibition of ATP induced dissociation by ADP can be described by [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ K_{{{\rm{obs}}}} = K^{\prime}_{1} K^{\prime}_{{ + 2}} [ - {\rm{ATP]/(1 + }}K^{\prime}_{1} [{\rm{ATP] + [ADP]/}}K^{\prime}_{5}.$$\end{document}]

If [ATP] is low, i.e., [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{1}[\rm ATP] \ll 1 $$\end{document}] then the above equation simplifies to: [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ K_{{{\rm{obs}}}} = K^{\prime}_{1} K^{\prime}_{{ + 2}} [ {\rm{ATP]/(1 + [ADP]/}}K^{\prime}_{5}) $$\end{document}] and this is rewritten as: [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K_{{{\rm{obs}}}} /K_{{\rm{o}}} = K_{{{\rm{obs}}}} K^{\prime}_{1} K^{\prime}_{{ + 2}} [ - {\rm{ATP] = 1/(1 + [ADP]/}}K^{\prime}_{5} ) $$\end{document}] where k_o = the value of k_obs in the absence of ADP and at a fixed low concentration of ATP [¹¹]. This assumes that ADP is in rapid equilibrium with A·M on the time scale of the measurement.

For Fig. 5 the analysis of the titration of S1 binding to actin was performed as described by Kurzawa and Geeves [²⁷]. Using a fixed concentration of pyrene-actin and increasing S1 concentrations, the concentration of the actin·S1 complex can be estimated from the amplitude of the observed fluorescence transient when ATP is added to dissociate the complex. The amplitude dependence on [S1] data was then fitted to the physically significant root of the following quadratic equation.

α is the fraction of actin with myosin bound, [M] is the total concentration of S1 added, [A]₀ is the concentration of actin and K_D is the dissociation constant of S1 for actin (i.e., K_A or K_DA).

For Fig. 6d, the two amplitudes of the ADP displacement reaction are proportional to the concentration of the S1 present as free S1 or S1·ADP. The amplitudes depend on the total ADP concentration as defined by:

and

In all cases, the figures refer to individual experimental measurements whereas Table 1 gives the mean values of the fitted constant for 2–3 separate measurements.

Recombinant human α- and β-S1 proteins

Using a previously described muscle cell expression system [¹³], we have produced recombinant human α- and β-S1 motors. S1 constructs encode binding domains for both ELC and RLC. A schematic diagram of each expression construct is shown in Fig. 2a. Adenoviruses encoding human α- and β-S1 were used to infect C₂C₁₂ myotubes and the protein was purified using the His-tag on the C-terminus of the motor. α-S1 and β-S1 co-purify with endogenous mouse ELC and RLC from the C₂C₁₂ muscle cells which were previously identified by mass spectrometry of the purified proteins as MLC1F, MLC2F, MLC3F, and MLC1A [¹³]. Because of a potential concern about using mouse light chains on a human motor, we also prepared a shorter β subfragment (β-sS1) corresponding to chymotryptic S1 [²³] co-expressed with the human ventricular ELC isoform MYL3 and purified it using a 6xHistidine tag on the N-terminus of the ELC (Fig. 2a). Figure 2b shows a Coomassie-stained gel of the purified proteins.

ATP-induced dissociation of actin·S1 is biphasic for human α- and β-S1

The steps in the cross-bridge cycle are shown in Fig. 1. The maximum shortening velocity of a muscle fiber is thought to be limited by how fast the myosin cross-bridge detaches from actin. This is limited either by the rate of ADP release from the cross-bridge (Step 5′) or the subsequent ATP-binding step (Step 1′ and 2′) [¹⁰, ¹¹, ²⁸]. The rate constants for these steps are a characteristic of the myosin isoform expressed in a muscle fiber [¹⁰]. For fast skeletal muscle myosins such as MyHC IIx from rabbit psoas, ADP release is very fast and detachment may be limited by how fast ATP binds to the cross-bridge [¹⁰]. For a slow muscle myosin, such as β-myosin, ADP dissociation is slow and limits how fast ATP can bind and thereby induce detachment [²⁶, ²⁹]. We measured the ATP-induced dissociation of actin·S1 for both human cardiac isoforms using stopped-flow methods as shown in Fig. 3. After rapid mixing of 0.5 mM ATP with 0.1 μM pyrene-actin·S1, for both cardiac S1 isoforms the pyrene fluorescence transient was best described by a two exponential fit with a large fast phase and smaller slow phase (Fig. 3a). For the fast phase, the observed rate constants for the two motors differed: k_obs = 658 s⁻¹, amp = 21% for α and k_obs = 436 s⁻¹, amp = 31% for β whereas for the slow phase a similar k_obs was found for both isoforms, k_obs = 61 s⁻¹, amp = 2.2% for α and 64 s⁻¹, amp = 5% for β. For both isoforms the k_obs of the fast phase showed a hyperbolic dependence on ATP concentration and saturates at high [ATP] with k_max = 1,667 and 1,432 s⁻¹ for α and β respectively (see Fig. 3b) where k_max corresponds to [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{+2} $$\end{document}] in Scheme 1. The values of K_0.5 (= 1/[\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{1} $$\end{document}]K′₁) were 769 μM for α and 1,075 μM for β-S1. The slow phase is virtually independent of ATP concentration and k_obs (= k_+α) = 40–60 s⁻¹ for both α- and β-S1. Table 1 contains the mean values for at least three measurements. The apparent second-order rate constant of ATP-binding ([\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{1}K^{\prime}_{+2} $$\end{document}]) is about two-fold faster for α-S1 compared to β-S1. This is mainly caused by a two-fold tighter affinity of actin·α-S1 for ATP (1/[\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{1}$$\end{document}]) compared to β-S1 with little difference on [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{+2}$$\end{document}]k′₊₂ (Table 1). The slow phase measured here has been seen previously for non-muscle and slow striated muscle myosin isoforms [²⁶, ³⁰]. The relative amplitude of the slow phase (~10% of the total amplitude for both α- and β-S1) defines the fraction of the actin·S1 complex that has a closed nucleotide pocket (equilibrium constant K_α, Scheme 1) that must isomerize (with rate constant k_+α) before ATP can bind [²⁶]. The closed-pocket form of actin·S1 is not illustrated in Fig. 1 because it is not on the primary cross-bridge cycle.

ADP release from actin·S1 is >ten-fold faster for α- than β-S1

We measured the ADP-affinity ([\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{5}$$\end{document}]) and ADP release rate constant (k′₊₅) for the two cardiac isoforms in order to establish whether ADP release could be rate-limiting for the cross-bridge detachment and thus the maximum shortening velocity of cardiac muscle fibers. The equilibrium constant for ADP binding to α and β actin·S1 can be estimated by measuring the ADP inhibition of the observed rate constant of ATP-induced dissociation of actin·S1 as described above. The results are depicted in Fig. 4a and typical transients are shown in supplementary Fig. S1A. Figure 4a shows that the k_obs of the fast phase of the ATP induced dissociation reaction decreased as the ADP concentration increased. The data in Fig. 4a clearly show that actin·β-S1 binds ADP much more tightly than actin·α-S1. Analysis of the data shows that K′₅ = 127 μM for α-S1 and 23 μM for β-S1. This tight affinity for ADP is similar to that reported for β-S1 isolated from slow muscle [²⁶] but α-S1 behaves more like fast skeletal muscle myosin which typically has a much lower affinity for ADP (see mean values in Table 1) [¹⁰].

Actin·β-S1 has a tight affinity for ADP and therefore a stable actin·S1·ADP complex can be formed in vitro for β-S1. Hence, saturating actin·β-S1 with ADP and then competing ADP off with a large excess of ATP allows the rate constants for ADP release to be measured. 0.5 μM actin·β-S1 was pre-incubated with 100 μM ADP and then rapidly mixed with a large excess of ATP (4 mM ATP). The resulting pyrene fluorescence transient can be described by a two exponential fit (Fig. 4b shows a one and a two exponential fit of the data). The fast component with k_obs = 84 s⁻¹ (amp +38%) defines the rate constant of ADP release ([\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{5}$$\end{document}]) whereas the slower phase defines the rate constant of nucleotide pocket opening to allow ADP release k_+αD 15.9 s⁻¹ (amp +6%). For β-S1, the rate constant for ADP release (mean value [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{5}$$\end{document}]  = 92 ± 8 s⁻¹) is very similar to ADP release from actin·β-S1 isolated from bovine masseter or bovine cardiac muscle [²⁶, ²⁸] as is the rate constant of nucleotide pocket opening (k_+αD = 15 ± 1 s⁻¹). The relative amplitude of the two phases defines the fraction of the actin·S1·ADP complex in the open and closed pocket forms (Scheme 1); thus the equilibrium constant, K_αD is 0.38/0.06 = 6.3 and a mean value of 7 ± 1 was observed over a series of three preparations. The temperature dependence of the observed fast and slow rates (Fig. 4c) allows one to determine the activation energy values E_a = 89.8 kJ (fast phase) and 38 kJ (slow phase). The ratio of amplitudes of the two phases, A_fast:A_slow had a small dependence on the temperature with A_f:A_s = 6.4 at 15°C, reducing to 3.5 at 40°C. At 37°C the rate constant of ADP release, [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{5}$$\end{document}]  = 575 s⁻¹ and the isomerization rate constant k_+αD = 25 s⁻¹. The same measurement cannot be made for α-S1 as the affinity of S1 for ADP and actin in the ternary actin·α-S1·ADP complex is too weak at accessible protein concentrations (see below).

The low affinity of ADP for actin·α-S1 predicts that the ADP release rate ([\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{5}$$\end{document}]) for α-S1 will be faster than the maximal rate of ATP induced actin·S1 dissociation ([\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{+2}$$\end{document}]) and therefore too fast to be measured by ADP displacement with excess ATP. This was shown to be true in Fig. 4d where the experiment of Fig. 3b was repeated in the presence of 125 μM ADP (see supplementary Fig. S1B for sample transients). Here we used a lower temperature (12°C) to improve the precision of the maximum value of k_obs that can be measured. At 20°C the maximum k_obs (1,500 s⁻¹) is near the limit of the resolution of the instrumentation. A plot of the k_obs versus [ATP] is shown in Fig. 4d and the maximum value for k_obs saturates at approximately the same value in the presence and absence of ADP (754 ± 50 and 837 ± 84 s⁻¹). Thus ADP release does not limit the maximum k_obs for ATP binding to actin·α-S1·ADP. ADP binding is therefore a rapid equilibrium step and [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{5}$$\end{document}] is much greater than [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{+2}$$\end{document}] (>750 s⁻¹ at 12°C or >1,500 s⁻¹ at 20°C). As noted above, this behavior of α-S1 is similar to that of fast skeletal muscle myosin [¹⁰] and not of any previously documented slow or cardiac isoform. The two cardiac myosin isoforms are therefore predicted to have very different lifetimes of the force-holding states (actin·S1·ADP and actin·S1) and these differences predict a faster shortening velocity for muscle fibers expressing α-MyHC than those expressing only β-MyHC; consistent with measurements of cardiac shortening velocity [³¹].

β-S1 binds actin five- to ten-fold tighter than α-S1

The affinity of actin for S1 in the rigor, actin·S1 complex (K_A), and in the actin·S1·ADP complex (K_DA) gives an indication of the strength of the actin–myosin bond that holds the force that myosin generates. It also yields the value for the thermodynamic coupling between actin and nucleotide binding (K_AD/K_D = the extent to which actin decreases the affinity of S1 for ADP and conversely K_DA/K_A = the extent to which ADP decreases the affinity of S1 for actin); a key characteristic of the mechanical activity of a myosin isoform [³⁰]. The thermodynamic coupling (K_AD/K_D) between actin and ADP binding to S1 can be validated by measuring the affinity of S1 for actin in the presence and absence of ADP (K_DA/K_A). Phalloidin-stabilized pyrene-actin (15 nM) incubated with various concentrations of α-S1 or β-S1 was mixed with ATP and the amplitude of the dissociation reaction was used to estimate the fraction of actin bound to S1. Figure 5a shows a series of transients observed on ATP induced dissociation of β-S1 from pyrene-labeled actin·β-S1 at a fixed 30 nM pyrene-actin with increasing S1 concentration. Since the ATP concentration was fixed, the k_obs values did not change but the amplitude increased as the S1 concentration increased until the actin was saturated. A plot of observed amplitude versus [S1] (given in Fig. 5b) shows that β-S1 has a tight actin binding (K_A = 8 nM). When the experiment was repeated in the presence of saturating ADP the affinity was reduced ~20-fold (K_DA = 190 nM). The same measurement was repeated for α-S1 and Fig. 5c shows a transient observed in the presence and absence of 1 mM ADP. This shows a much smaller amplitude in the presence of ADP. The plot of amplitude of the observed transient versus [α-S1] shown in Fig. 5d gives the affinities of α-S1 for actin as K_A = 44 nM and K_da = 2.4 μM.

The average actin-affinity measured for β-S1 (8 ± 2 nM) is similar to the affinity of bovine β-S1 (7 nM) whereas the affinity for α-S1 (37 ± 11 nM) is similar to rabbit fast muscle S1 (33 nM) (Table 1) [²⁶]. In the presence of saturating ADP concentrations affinities were reduced 50-fold for α-S1 (1.84 μM) and 24-fold for β-S1 (Table 1). The weak affinity of S1 for actin is again a feature common to α-S1 and fast muscle myosin isoforms and distinct from slow muscle-derived β-S1.

ATP hydrolysis is ten-fold faster for α-S1 than β-S1

The isometric force that a muscle can generate is proportional to the number of attached myosin cross-bridges in the sarcomere. While ADP release and ATP binding limit the exit from the attached states, entry into the attached states is influenced by other steps in the cross-bridge cycle, such as myosin access to actin-binding sites (at the high local concentration of actin in the sarcomere this is largely limited by the geometry of the thick and thin filaments and sliding velocity [³²]), the release of P_i that commits myosin to strongly bind actin, and the hydrolysis step which defines the time of the myosin recovery stroke (Step 3) before it can again bind to actin [³²].

Myosin S1 contains tryptophan residues that have intrinsic fluorescence and can be used to detect both the ATP-binding and the ATP hydrolysis steps (Fig. 1 Steps 1–3) in the absence of actin. Upon mixing 500 μM ATP with 0.2 μM S1 the tryptophan fluorescence signal of showed a single phase for α-S1, whereas for β-S1 the fluorescence signal showed two phases, a fast phase similar to α-S1 and an additional slow phase (Fig. 6a). The fast phase had a k_obs of 151 s⁻¹ (amp 5.1%) for α-S1 and 124 s⁻¹ (amp 8.4%) for β-S1. The fast phases were hyperbolically dependent upon ATP concentration as shown in Fig. 6b and the data were fitted to Eq. 3, as defined in the legend.

The slow phase observed for β-S1 was small and not always easy to observe (see single exponential fit in Fig. 6a) but was consistently observed when the S1 was treated with apyrase to remove any contaminate ADP. A similar slow phase was previously seen in other β-S1 preparations from Bovine masseter and rabbit soleus muscles [²⁶, ²⁹]. It was also clearly present in β-sS1 preparation (see supplementary Fig. 2E). The slow phase was independent of ATP concentration above 100 μM with k_obs = 19 s⁻¹ and an amplitude of 0.8%.

For β-S1 derived from slow skeletal muscle a similar slow phase of the fluorescence change has been assigned to the hydrolysis step (Step 3; k_obs = k₊₃ + k₋₃) [²⁶]. To ascertain if this was also true for human β-S1 we used quenched-flow methods to directly measure the rapid burst of ATP hydrolysis. The mouse C₂C₁₂ cell expressed protein is limited in both quantities and the S1 concentrations that can be achieved. We therefore could not complete a full quenched-flow study but simply attempted to distinguish between the hydrolysis step coinciding with either the fast or slow phase of the fluorescence transient. The maximum β-S1 available after mixing was 3.5 μM, and this was mixed with a ten-fold excess of ATP and incubated at 20°C before quenching the reaction in acid and analyzing the amount of ADP generated. A rapid initial burst of ADP production was observed and at 350 ms with a burst size of 0.59 ADP/S1. Three time points were chosen at 16, 33, and 68 ms to give the best chance of distinguishing the two possibilities. The results gave the ADP produced as 16, 43, and 61% of the burst amplitude. A single exponential fit to the data suggests a k_obs value of ~15 s⁻¹. Using the values given in Table 1 the expected k_obs for the fast phase is K₁k₊₂[ATP] = 1.5 × 10⁶ × 35 × 10⁻⁶ = 52.5 s⁻¹. This predicts the ADP generated at the three time points of 57, 82 and 95%. The ADP burst rate is thus a third that of the predicted fast phase fluorescence. The quality and quantity of the data does not warrant a detailed fitting of the data to a model. However a simple M → M·ATP → M·ADP·P_i model where the first step is the irreversible ATP binding at K₁k₊₂[ATP] = 52.5 s⁻¹ and the second step is the maximum observed value of k_obs for the slow phase of 17 s⁻¹ (Table 1) predicts a burst with a half time of ~66 ms and thus a k_obs of 10.5 s⁻¹. A much closer fit to the 15 s⁻¹ observed than the predicted fast phase fluorescence. The data are therefore compatible with the hydrolysis step being slower than the fast fluorescence transient (assigned to ATP binding). The similarity of the data presented here together with that reported for tissue-purified β-S1 suggests that the hydrolysis step does correspond to the slow fluorescence change but given the error on each measurement the assignment is provisional.

Quenched-flow measurements with α-S1 at a single time point of 13 ms indicated 33% of the burst amplitude, which is much higher than observed for the β-S1. This is compatible with a k_obs of 85 s⁻¹ and a predicted value for the fast phase of 100 s⁻¹. There was no observable slow phase fluorescence for the α-S1 and therefore we assume the maximum observed rate constant for the fast fluorescence transient corresponds to the hydrolysis step. Further evidence for this assignment of the single phase for α-S1 is that the amplitude of the fluorescence change decreased from 10% at low ATP (<100 μM) to ~6% at higher ATP (>300 μM). This could indicate the presence of a fast fluorescence change associated with ATP binding which becomes lost in the dead time of the system at high ATP concentrations. For α-S1, these characteristics are again similar to the characteristics of fast skeletal muscle myosin [³³, ³⁴]. Thus α- and β-S1 differ greatly in the speed with which the myosin can complete the hydrolysis step/recovery stroke before it can once again bind to actin.

There is only a very small change in tryptophan fluorescence when ADP binds to cardiac S1. Therefore, the larger fluorescence change which occurs on ATP binding can be used to measure the displacement of ADP from S1. Therefore, in order to measure the ADP affinity (K₅) and ADP off-rate constant (k₊₅), ADP was displaced by addition of a large excess of ATP [²⁶]. Figure 6c shows the fluorescence transients observed on displacing ADP (0–1.6 μM) from 0.2 μM β-S1. The transients show a fast and a slow phase; the fast phases represent ATP-binding to free S1 and the amplitude decreases as the ADP concentration increases. The slow phase is the displacement of ADP from S1·ADP by ATP and the amplitude increases with increasing ADP concentration. The ratio of the two amplitudes therefore represents [S1]_free/[S1·ADP]. Fitting the amplitude dependence on [ADP] to Eq. 5A and 5B (Fig. 6d) gives an estimate of the ADP-affinity K₅ = 0.5–0.8 μM for β-S1. The k_obs of the slow phase was independent of ADP concentration and ATP >100 μM and therefore represents the ADP off-rate constant k₊₅ = 0.8 s⁻¹. The ADP affinity was weaker for α-S1 (K₅ = 2.8 μM) with a faster ADP release rate k₊₅ = 2.7 s⁻¹.

The humanized ELC form, β-sS1, is kinetically similar to β-S1

We repeated all of the measurements described above for the “humanized” β-sS1 construct. In almost all cases, the data show little difference between the two β-constructs (see example data in Supplementary Fig. S2A–F summarized in Table 1 and Fig. 7). The constants, K₁k₊₂ and k₊₂ for S1 alone do not differ between the two constructs. All other constants differ by no more than two-fold with the exception of k_+αD which is seven-fold faster for β-S1 than for β-sS1. Thus, as seen for other myosin constructs, in the absence of load the light chains have little influence on the biochemical kinetic properties of S1.

Mouse cardiac α-S1 is similar to the recombinant human α-S1

In order to compare recombinant protein with myosin purified from cardiac tissue, we isolated myosin from mouse heart, which almost exclusively expresses the α myosin isoform [³⁵]. Myosin was isolated from a single mouse heart and was digested with chymotrypsin as described above. The S1 thus generated was active for only a few days, therefore the preparation from heart to S1 was completed in 24 h and used without extensive purification. The digested myosin was simply spun down to remove insoluble material. All measurements were completed within 48 h of digestion of the myosin. Figure 2c shows a Coomassie-stained gel of the purified α-myosin and α-S1. Almost all of the measurements reported above were repeated on mouse cardiac α-S1. Examples of data are shown in Supplementary Fig. S3 and the data are summarized in Table 1 and Fig. 7. These show that the properties of tissue-purified mouse α-S1 and the human α-S1 differ by no more than two-fold for any of the parameters measured, with the exception of 1/[\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{1}$$\end{document}]. However, mouse α-S1 is quite distinct from β-S1.

Implications for cardiac muscle fibers

Differences in transient kinetic properties between the α and β isoforms are more striking than might have been expected from the two- to three-fold differences in shortening velocity, in vitro motility and ATPase activity reported in the literature for the α and β isoforms from other species [⁵, ¹², ³⁵, ⁴⁵–⁴⁸]. Single molecule mechanical studies of tissue isolated α- and β-myosins have shown that the step size and unitary force do not change. However, the lifetimes of the attached states do differ two-fold [⁶]. Using a higher time resolution laser trap Capitanio et al. [⁴⁹] reported that the actin–myosin mechanical event can be resolved into two components, the first limited by ADP release and the second limited by the concentration of ATP used. For full-length β-myosin isolated from rat soleus muscle, the ADP limited step was ~42 s⁻¹ compared to the 400 s⁻¹ for a fast type myosin. For slow muscle myosins (the β isoform) the ADP release step measured in solution ([\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{+5}$$\end{document}]) correlates with the event that limits the maximum shortening velocity [²⁶, ²⁹]. For fast muscle myosins, the 400 s⁻¹ ADP-linked event in the laser trap is slower than the value of [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{\alpha {+5}}$$\end{document}] observed in solution for ADP release. It is therefore thought to be an isomerization that is not accessible by simply adding ADP to actin·S1 and can only be accessed following ATP hydrolysis and P_i release. In several non-muscle myosins the ADP release event is load dependent and slows down by 5- to 100-fold [⁵⁰, ⁵¹] when the myosin is bearing a load equivalent to isometric force. This marked difference in the ADP release event between fast and slow muscle myosins is compatible with our β-S1 data which shows a >ten-fold slower ADP release for β-S1 than either α or that reported for fast skeletal muscle myosins [¹⁰, ²⁶]. The ATP limited steps observed in the optical trap are more similar for fast and slow muscle myosins as we also observed for α- and β-S1. No fast laser trap measurements have been made for any α-myosin.

Like the single molecule measurements and our solution data, mechanical measurements on human cardiac myofibrils and cardiomyocytes show that the differences between the myosins may be larger than expected from ATPase and motility data alone. Studies of single human atrial (predominantly α-myosin) and ventricular (predominantly β-myosin) myofibrils show that while the isometric tension varies by 10–20% (108–125 mN mm⁻²) the rate of tension development after a period of rapid shortening (known as the rate constant k_tr) is five-fold faster for atrial than for ventricle samples [⁵²]. Similarly a study of the myosin isoform content of human atrial cardiomyocytes from patients with atrial fibrillation or dilation have shown that the β-myosin content increases from 10% in normal hearts to up to 60% [⁵³]. Again, while the isometric force generation was not correlated with β-myosin content, the rate of force development (k_tr) decreased from 10 s⁻¹ for cardiomyocytes with 10% β-myosin to 1 s⁻¹ for 60% β-myosin [⁵³]. A direct correlation of k_tr with solution data is not currently possible but Piroddi et al. [⁵²] did establish that the differences in k_tr were not attributable to changes in calcium binding or release from the thin filaments. Currently k_tr is thought to be limited by cycling cross-bridges as they change from unloaded to loaded isometric conditions. This slowing down of loaded contractions is believed to be linked to load-limited ADP release.

As discussed above, the rate of ADP release from actin·S1 does appear to be the event that limits the shortening velocity in slow muscle and our low value of the ADP release in β-S1 is consistent with this view. For fast muscle myosin ADP release as measured in solution is very fast and too fast to limit the shortening velocity. Our results with α-S1 are consistent with this view. The data in Fig. 4c show the value of the ADP release rate constant ([\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{+5}$$\end{document}]) over the temperature range from 10 to 40°C. The value of [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K^{\prime}_{+5}$$\end{document}] increases almost seven-fold between 20 and 37°C (from 84 to 575 s⁻¹) and therefore predicts a similar increase in shortening velocity over this temperature range.

Using the rate constants in Table 1 extrapolated to physiological ATP concentration (5 mM) and assuming unlimited actin concentration (the situation in a rapidly contracting muscle fiber), we can estimate the time that the two isoforms spend strongly bound to actin [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(1/K^{\prime}_{{ + 5}} + 1/(K^{\prime}_{1} K^{\prime}_{{ + 2}} [{\rm{ATP}}])/(1 + K^{\prime}_{1} [{\rm{ATP]))}}$$\end{document}] and the time free from actin (1/(k₊₃ + k₋₃)) at 20°C. By these calculations α-S1 spends ~0.7 ms strongly bound to actin and ~6.0 ms unbound while β-S1 spends ~12.9 ms strongly bound and ~67.0 ms unbound. In this modeled contractile cycle the two motors have a >ten-fold difference in cycle time: 6.7 (0.7 + 6.0) ms for α-S1 and 79.9 (67 + 12.9) ms for β-S1. Thus, although α and β spend very different lengths of time in these bound and unbound states they have a similar duty ratio (bound/bound + unbound) which remains low at ~10% for α-S1 and ~16% for β-S1. This is compatible with observations that the unitary force per myosin does not change and that the isometric force is not greatly altered. Thus, the number of myosin heads attached is similar and the force per head is also the same. The consequences of a similar duty ratio but a longer lifetime of the attached force-holding states has implications for the rate of activation and relaxation of the thin filament. If strongly bound myosin heads are a significant component of the thin filament activation process then actin filaments with an attached β-myosin will be much slower to relax than with an α-myosin. Similarly, the faster cycling α-myosin could activate the thin filament faster than β-myosin with an equivalent level of calcium present. These cooperative effects are unlikely to be linearly related to the fraction of α and β present in a given tissue. In several mechanical assays, the presence of a small amount of one isoform has a dominant effect on the overall phenotype observed and this is particularly true of power output and the rate of relaxation [¹⁸].

The roles of cardiac myosin isoforms have been explored in cell and transgenic studies. For example, persistent expression of α-MyHC in hearts of rabbits (which normally express β-MyHC) that have tachycardia-induced failure or myocardial infarction provides cardioprotection [⁵⁴, ⁵⁵]. Recently, gene transfer of human α- and β-MyHCs into adult cardiac myocytes was performed, achieving the transfer of human α into failing human and rabbit cardiac myocytes and human β into rat cardiac myocytes. The investigators found that human α-MyHC representing 21% of total MyHC 24 h after transduction or 30% of total after 48 h increased the degree of contractility as well as the kinetics of contraction and relaxation without affecting Ca²⁺ transients [⁵⁶]. Transfer of human β-MyHC into the predominantly α-MyHC rat myocytes where human β-MyHC represented 18 or 38% of the total MyHC decreased the amount of contraction by 42 and 57%, respectively [⁵⁷]. These effects were also Ca²⁺ independent. Together, these results support the notion that the myosin composition of the human heart influences cardiac function. Understanding how the two myosins work together will be an important next step.

The transient assays used here have the great advantage that they are not sensitive to the presence of contaminant myosin activity, but this makes them unsuitable for examining the potential dominant role of isoform mixture. Full-length constructs such as those described above can be used in mechanical studies these will be sensitive to the isoform mixture present and will be the subject of future work. Work with full-length human α-MyHC in adult cardiac myocytes has demonstrated dominant effects of small amounts of α-MyHC on contractility and relaxation in failing human cardiac myocytes [⁵⁶]. Similarly, the study of the activation and relaxation process in solution with α- and β-myosins (and in mixtures) will be a major study that has not been addressed to date. The impact of these kinetic constants on muscle contraction lies in their contribution to defining contractile processes. These findings indicate >ten-fold differences between the cardiac isoforms α and β in the kinetic constants implicated in controlling fiber shortening velocities and cycle times, as well as subtler differences in constants influencing other contractile events. The elucidation of these values is a step towards understanding the role of myosin isoform composition in the human heart in health and disease.

Click here for additional data file (18_2012_927_MOESM1_ESM.doc)

Conflict of interest

The authors declare that they have no conflict of interest.

Human α- and β-S1 constructs with mouse light chains or with human MYL3 (β-sS1) are compared to rabbit fast skeletal S1 and bovine masseter slow skeletal myosin S1 from muscle. The data presented here represent the mean values of 2–3 individual measurements. Measurements were performed at 20°C and 100 mM KCl, pH 7.0 unless indicated otherwise

nd not determined

^aThis study

^bFrom Ritchie et al. [⁵⁸] unless otherwise stated

^cFrom Bloemink et al. [²⁶]

^dFrom Miller and Geeves [⁵⁹]

^eFrom Nyitrai et al. [¹⁰]. This work gives values of k₊₂ (= 740 s⁻¹) and 1/K′₁ (= 520 μM) at 12°C. The value of 1/[\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ k_{1}^{\prime} $$\end{document}] is independent of temperature so the value of [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ k_{+2}^{\prime} $$\end{document}] at 20°C can be estimated from the [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ k_{1}^{\prime} $$\end{document}][\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ k_{+2}^{\prime} $$\end{document}] value at 20°C given in the table and 1/[\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ k_{1}^{\prime} $$\end{document}], i.e., [\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ k_{+2}^{\prime} $$\end{document}] = 2.4 × 520 = 1,248 s⁻¹

No detectable slow component observed