Kinetic signatures of myosin-5B, the motor involved in microvillus inclusion disease

Myosin-5B is a ubiquitous molecular motor that transports cargo vesicles of the endomembrane system in intracellular recycling pathways. Myosin-5B malfunction causes the congenital enteropathy microvillus inclusion disease, underlining its importance in cellular homeostasis. Here we describe the interaction of myosin-5B with F-actin, nucleotides, and the pyrazolopyrimidine compound myoVin-1. We show that single-headed myosin-5B is an intermediate duty ratio motor with a kinetic ATPase cycle that is rate-limited by the release of phosphate. The presence of a second head generates strain and gating in the myosin-5B dimer that alters the kinetic signature by reducing the actin-activated ADP release rate to become rate-limiting. This kinetic transition into a high-duty ratio motor is a prerequisite for the proposed transport function of myosin-5B in cellular recycling pathways. Moreover, we show that the small molecule compound myoVin-1 inhibits the enzymatic and functional activity of myosin-5B in vitro. Partial inhibition of the actin-activated steady-state ATPase activity and sliding velocity suggests that caution should be used when probing the effect of myoVin-1 on myosin-5–dependent transport processes in cells.

Double-headed class 5 myosins are evolutionary ancient, prototypic actin-based molecular motors implicated in the spatiotemporal segregation and the transport of organelles across phylogeny (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). A prerequisite for efficient transport is that a single motor can either take multiple discrete steps on an actin filament without detaching, a property also referred to as processivity, or that the motor works in an ensemble with other myosins-5 to efficiently move a cargo. These differences in physiological capacities and limitations are directly linked to the mechanoenzymatic signatures of class 5 myosins (11,12).
Mammals express three myosin-5 genes (MYO5A, MYO5B, and MYO5C) in almost all cells (6,13). Compared with the well studied paralog myosin-5A that exhibits single molecule processivity and the nonprocessive myosin-5C, little is known about the kinetic features of myosin-5B (14 -18). Myosin-5B is best known for the transport of Rab GTPase positive vesicles of the endomembrane system along the F-actin cytoskeleton to the plasma membrane (4, 19 -21). The long-range vesicle transport system is a prerequisite for diverse physiological processes including axon development, postsynaptic plasticity, cell polarization, and the prevention of polyspermy (19,(22)(23)(24)(25). The mutation-induced loss of function of MYO5B is manifested in debilitating pathologies including microvillus inclusion disease that results in death in neonates and children (20,26,27). As reviewed recently, the majority of the Ͼ40 disease-associated mutations in MYO5B cluster in the coding region for the motor and the tail domain of the myosin heavy chain (26). The motor domain harbors a prototypic nucleotide-binding pocket and a binding region for F-actin (Fig. 1A). The tail domain contains a coiled-coil region that promotes the dimerization of two myosin heavy chains and terminates in a distal globular tail domain that constitutes the binding site for Rab GTPase family members and other adaptor proteins (4,13,25,28). Both domains are separated by the neck that contains six IQ motifs, predicted binding regions for calmodulin (CaM) 3 or calmodulin-like light chains ( Fig. 1A) (29,30). The distribution of mutations in the myosin heavy chain suggests that either (i) the nucleotide-dependent interaction between the motor domain and the actin filament may be altered or (ii) the interaction between the molecular motor and its cargo vesicle may be altered. Both scenarios are expected to result in the aberrant vesicle trafficking patterns in patients with microvillus inclusion disease (31,32).
To understand the mechanoenzymatic basis of the actinbased transport powered by myosin-5B in normalcy, we studied the interaction between the motor, nucleotides, and F-actin ( Fig. 1B) with ensemble solution kinetics and functional assays in vitro. Our experiments show that single-headed myosin-5B has an intermediate duty ratio and that a second head is required to establish a high-duty ratio, a kinetic prerequisite for single molecule processivity. The data also allows for the comparative analysis of kinetic signatures in the class 5 myosome. Moreover, we show that the small molecule myoVin-1 inhibits the enzymatic and functional activity of myosin-5B in vitro. The observed effects resemble the previously described effect on myosin-5A and suggest that myoVin-1 may inhibit all class 5 myosins.

Design of expression constructs for kinetic experiments
We recombinantly overproduced and purified the singleheaded M5B S1 that contains all six IQ motifs and the doubleheaded M5B HMM (Fig. 1A) in complex with the light chain CaM in milligram quantities in the baculovirus/Sf9 insect cell system. The proteins allowed us to directly compare the mechanoenzymatic properties of single-headed and double-headed myosin-5B constructs and their interaction with nucleotides, F-actin, and the small molecule myoVin-1.

Enzymatic activity of M5B S1 under steady-state conditions
We first determined the enzymatic activity of M5B S1 with an NADH-coupled assay under steady-state conditions. As shown in Fig. 1C, F-actin potently activates the basal ATPase activity of M5B S1 from a k basal of 0.07 Ϯ 0.01 s Ϫ1 more than 2 orders of magnitude. Fitting a Michaelis-Menten equation k obs ϭ k cat ͓F-actin͔ K app ϩ ͓F-actin͔ (Eq. 1) to the data results in a k cat of 10.0 Ϯ 0.1 s Ϫ1 . The K app is 12.6 Ϯ 0.4 M, the catalytic efficiency k cat /K app ϳ0.8 M Ϫ1 s Ϫ1 . The steady-state kinetic parameters of M5B S1 are compiled in Table  1 and compared with previously studied class 5 myosins. Figure 1. Domain organization, kinetic scheme, and steady-state ATPase activity of M5B S1 . A, domain organization of myosin-5B and expression constructs used in this work. The myosin motor domain in the myosin heavy chain is shown in gray. The six IQ motifs in the myosin neck domain are shown in orange, and the tail domain is in blue. The tail domain mediates the dimerization of M5B FL and M5B HMM by means of a coiled-coil and interacts with binding partners. B, simplified kinetic scheme of the myosin and actomyosin ATPase cycle. The events of ATP binding, ATP hydrolysis, and phosphate release are shown for myosin in the actin detached (top row) and attached (bottom row) states. The main flux through the pathway is highlighted in orange, and the weak and strong actin-binding states are indicated. The notation distinguishes between the kinetic constants in the presence and absence of F-actin by using regular versus bold type; subscripts A and D refer to F-actin (K A ) and ADP (K D ), respectively. Dissociation equilibrium constants were calculated as K x ϭ k Ϫx /k ϩx . M, myosin; A, actin. C, steady-state ATPase activity of M5B S1 . Increasing concentrations of F-actin activate the steady-state ATPase activity to a k cat of 10.0 Ϯ 0.1 s Ϫ1 with a K app of 12.6 Ϯ 0.4 M.

Kinetics of myosin-5B ATP binding to myosin
To determine the rate-limiting step of the M5B S1 kinetic cycle, we measured the individual parameters that describe the events of ATP binding, ATP hydrolysis and product release in the F-actin attached and detached states (Fig. 1B) in a stoppedflow spectrophotometer. As shown in the inset of Fig. 2A, ATP binding to M5B S1 causes a transient increase in the intrinsic tryptophan fluorescence that is best described by a single exponential fit. The observed rate constants k obs , obtained by varying the ATP concentration, follow a hyperbolic dependence according to and approach a maximum k max value of 85.4 Ϯ 1.5 s Ϫ1 , representing the ATP hydrolysis rate k ϩ3 ϩ k Ϫ3 . Half-saturation (1/K 0.5 ) is observed at an ATP concentration of 225.1 Ϯ 14.7 M ( Fig. 2A). At low ATP concentrations, the dependence of the observed rate constant on ATP is linear. The slope of the straight line yields the apparent second-order rate constant for ATP binding, K 1 k ϩ2 ϭ 0.39 Ϯ 0.01 M Ϫ1 ⅐s Ϫ1 (Fig. 2B). An identical value of 0.4 Ϯ 0.01 M Ϫ1 s Ϫ1 is obtained after binding of the nucleotide analog mantATP to M5B S1 under pseudo-first order conditions. Fig. 2B shows the binding data of both nucleotides up to a concentration of 25 M. All transient-state kinetic parameters of M5B S1 are compiled in Table 2 and compared with previously studied single-headed myosins-5.

ATP binding to actomyosin
The time-dependent change in light scattering signal was used to assay the interaction between actomyosin and ATP. Mixing of 0.3 M actoM5B S1 with excess ATP results in a single exponential decrease in light scattering signal, as shown in the inset of Fig. 2C. The observed rate constant k obs values follow a hyperbolic dependence as a function of ATP that is best described by the following equation. 4 The fit converges toward a maximum k ؉2 value of 373.5 Ϯ 18.7 s Ϫ1 . Half-saturation (1/K 0.5 ϭ 1/K 1 ) is observed at 745.6 Ϯ 120.7 M ATP (Fig. 2C). At low [ATP] up to 50 M, the data set is best fit with a linear function, representing the apparent sec-ond-order binding rate constant K 1 k ؉2 ϭ 0.28 Ϯ 0.01 M Ϫ1 s Ϫ1 (Fig. 2D).

ADP binding and release from myosin
Binding kinetics of the fluorescent nucleotide analog mantADP to M5B S1 were determined in a stopped-flow spectrophotometer. Mixing of 0.5 M M5B S1 under pseudo-first order conditions with mantADP results in a single exponential fluorescence increase (Fig. 3A, inset). A linear fit to the data set according to results in a second-order binding rate constant k ϩD of 3.05 Ϯ 0.16 M Ϫ1 s Ϫ1 (Fig. 3A). Extrapolation of the fit to [mantADP] ϭ 0 determines the ADP release rate constant k ϪD ϭ 14.3 Ϯ 1.61 s Ϫ1 . This value is an order of magnitude higher than for previously characterized class 5A myosins ( Table 2). As an independent measure, we directly determined k ϪD in chasing experiments in which mantADP is displaced from the myosin⅐ mantADP complex with excess ATP. Independently, we measured the ADP release rate constant from the myosin⅐ADP complex when chased with excess ATP. The latter experimental design uses the intrinsic tryptophan fluorescence of myosin as readout. Both experimental setups result in almost identical values for k ϪD of 18.0 Ϯ 0.29 s Ϫ1 (mantADP) and 18.9 Ϯ 0.09 s Ϫ1 (ADP) that are in good agreement to the determined value from the binding data (k ϪD ϭ 14.3 Ϯ 1.61 s Ϫ1 ) ( Fig. 3B and Table 2). From the ratio of the ADP release and binding rate constants (K D ϭ k ϪD /k ϩD ), the dissociation equilibrium constant of ADP was calculated to K D ϭ 4.7-6.2 M.

ADP binding to actomyosin
ADP binding to the actomyosin complex was probed by mixing mantADP under pseudo-first order conditions with actoM5B S1 . The associated time-dependent fluorescence increase is best described by a single exponential fit (Fig. 3C). The plot of the observed rate constants k obs versus [mantADP] follows a linear trend that is best described by the following equation.

ADP release from actomyosin
The ADP release rate constant k ؊AD was directly determined by chasing the actomyosin⅐ADP complex with excess ATP. 4 The notation used throughout this article distinguishes between the kinetic constants in the presence and absence of F-actin by using regular (K1) versus bold (K1) type; subscripts A and D refer to F-actin (K A ) and ADP (K D ), respectively.

Kinetics of myosin-5B
Two different experimental strategies were employed. First, mantADP was chased from actomyosin. Second, the change in light scattering upon the ATP-induced dissociation of the actomyosin⅐ADP complex was detected. In this indirect experimental setup, the ADP release limits the dissociation of the actomyosin complex. The former results in an ADP release rate k ؊AD of 16.69 Ϯ 0.11 s Ϫ1 , and the latter results in a k ؊AD of 17.13 Ϯ 0.14 s Ϫ1 (Fig. 3D). These independent measurements suggest that the fast ADP release rate is a kinetic signature of actoM5B S1 . The kinetic coupling ratio, the efficiency of F-actin to activate the ADP release, k ؊AD /k ϪD ϭ ϳ1-1.5, is low. Accordingly, the dissociation equilibrium constant K AD ϭ k ؊AD /k ؉AD ϭ 6.1-8.1 M is similar to the respective constant in the absence of F-actin (K D ϭ 4.7-6.2 M; Table 2).

Interaction with F-actin
The binding kinetics of M5B S1 to F-actin in the presence and absence of excess ADP was probed by light scattering. The transients are best described with single exponential fits as exemplarily shown in the inset of Fig. 4A. The obtained k obs increase linearly within the F-actin concentration range tested. Linear fit of the data to k obs ϭ k ؉DA ͓F-actin͔ ϩ k ؊DA (Eq. 6) results in the second-order binding rate constants k ؉A of 8.39 Ϯ 0.16 M Ϫ1 s Ϫ1 and k ؉DA of 5.98 Ϯ 0.19 M Ϫ1 s Ϫ1 in the absence and presence of ADP, respectively (Fig. 4A). The dissociation rate constants k ؊A and k ؊DA are close to 0 and do not allow for a firm determination of the respective parameters (Fig. 4A). M5B S1 does not significantly quench pyrene fluorescence upon binding to pyrene-labeled F-actin. This observation precludes the use of this fluorescence label to directly determine the F-actin affinities in the presence and absence of ADP and the F-actin release rate constants, respectively. It is of note that the inability of M5B S1 to quench pyrene fluorescence is not an indication that the actomyosin complex is not formed, because a light scattering signal is observed under identical assay conditions (Fig. 4A, inset). Concomitantly, we find that the actoM5B S1 complex is formed under identical buffer conditions in high-speed cosedimentation assays (Fig. 5B).

Duty ratio of M5B S1
The steady-state and transient kinetic characterization of single-headed M5B S1 allowed us to calculation the duty ratio, the time of its kinetic cycle M5B S1 spends strongly bound to actin (Fig. 1B). Values of ϳ0.46 -0.6 were obtained with the following equation
The duty ratio values obtained for M5B S1 are lower than for single-headed constructs of other myosin-5 paralogs and unconventional myosins associated with cellular transport function (35)(36)(37). A duty ratio Ͼ0.5/head is a theoretical cutoff value and a minimum requirement for processivity, but duty ratios Ͼ0.95 are required for efficient single molecule processivity of a double-headed myosin (11,35,38). Calculation of the duty ratio of a myosin-5B dimer (39, 40) based on duty ratio M5B HMM ϭ 1 Ϫ ͑1 Ϫ duty ratio M5B S1 ͒ 2 (Eq. 8) neglecting any steric factors, gating, and strain between the two motor domains results in values of 0.71-0.84.

Steady-state kinetic properties and key transient-state kinetic parameters of M5B HMM
The steady-and transient-state kinetic parameters of M5B S1 measured in the absence of load indicate that the motor is unlikely to function as an efficient cargo transporter inside cells (Tables 1 and 2). This finding conflicts with the described physiological function of myosin-5B as transporter of membrane vesicles along the F-actin cytoskeleton (4, 19 -21). We therefore tested whether a second head, as it is found in the myosin-5B dimer inside the cell, would change the enzymatic signatures and render the molecule a high-duty ratio motor. First, we measured the kinetic properties of M5B HMM under steady-state conditions. F-actin activates the basal ATPase activity k basal from 0.12 Ϯ 0.02 s Ϫ1 ϳ65-fold to a maximum k cat value of 7.91 Ϯ 0.15 s Ϫ1 per head (Fig. 5A), a 20% reduction in k cat when compared with M5B S1 (Tables 2 and 3). The K app is with 3.12 Ϯ 0.26 M 60% lower than the respective parameter obtained with M5B S1 , the catalytic efficiency k cat /K app with ϳ2.5 elevated (Tables 1 and 3).
The difference in k cat must be reflected in a transient kinetic parameter that rate-limits the actomyosin ATPase cycle and led us to test whether the presence of a second head would influence the ADP release rate from M5B HMM . As shown in Fig. 5B, the ADP release rate from M5B HMM in the presence of a 320fold molar excess ADP (k ϪD ϭ 19.53 Ϯ 0.25 s Ϫ1 ; Fig. 5B) is identical to the parameter measured for single-headed M5B S1 (k ϪD ϭ 18.9 Ϯ 0.09 s Ϫ1 ) in the absence of F-actin ( Fig. 3D and Table 2). Moreover, the fluorescence transient is best described by a single exponential, indicating that the presence of a second head per se does not affect the kinetics in the absence of F-actin. When bound to F-actin, the ADP release rate from M5B HMM and the transient change in light scattering signal is double exponential with a fast rate k ؊AD,HMM,fast of 7.24 s Ϫ1 and a slow phase k ؊AD,HMM,slow of 1.27 s Ϫ1 (Fig. 5C). The fast rate is in good agreement with the titrated k cat of 7.91 Ϯ 0.15 s Ϫ1 obtained in the steady-state ATPase assay ( Fig. 5A and Table 3) and probably represents the ADP release rate from the positively strained trail head, and the slow rate probably represents the ADP release rate from the negatively strained lead head (18). Taken together, the steady-state and transient-state Table 2 Transient state kinetic parameters of mouse M5B S1 in comparison with previously characterized class-5 myosins Numbering of the kinetic constants refers to Fig. 1B. ND, not determined. NA, not experimentally accessible.

Kinetics of myosin-5B
kinetic experiments indicate that the actin-activated ADP ratelimits the M5B HMM but not the M5B S1 kinetic cycle. Accordingly, the duty ratio of the lead head is high and approaches unity based on the calculation according to Equation 7 with the measured rates k ؊AD,HMM,fast ϭ 7.24 Ϯ 0.1 s Ϫ1 and k cat ϭ 7.91 Ϯ 0.15 s Ϫ1 .

Chemical inhibition of M5B S1 with myoVin-1
Our detailed kinetic characterization of M5B S1 and M5B HMM allowed us to probe the effect of the small molecule inhibitor myoVin-1 on its mechanoenzymatic properties. The pyrazolopyrimidine compound (Fig. 6A) was previously identified as an uncompetitive inhibitor for myosins-5A in vitro and has been used in cell biological studies to probe for myosin-5 motor function (41)(42)(43)(44)(45). To address whether the small molecule has selectivity for myosin-5A, we measured its effect in the actinactivated steady-state ATPase activity of M5B S1 . The addition of myoVin-1 to M5B S1 leads to a partial inhibition of the steadystate ATPase activity at a fixed F-actin concentration of 30 M. The K I , obtained after fitting the data, normalized to the ATPase activity in the absence of inhibitor, to a dose-response function is 23.12 Ϯ 2.07 M.
To address whether the partial reduction in the steady-state ATPase activity is caused by a decrease in k cat or K app , we measured the ATPase activity of M5B S1 as a function of [F-actin] at increasing myoVin-1 concentrations up to 30 M. Higher . Transient kinetic interaction between M5 S1 and actoM5 S1 with ADP and ADP analogs. A, the observed rate constants upon rapidly mixing M5B S1 and mantADP follow a linear dependence in the concentration range tested. The straight line describes the second-order rate constant for ADP-binding k ϩD to 3.05 Ϯ 0.16 M Ϫ1 s Ϫ1 . From the y intercept, the ADP release rate k ϪD was determined to be 14.3 Ϯ 1.61 s Ϫ1 . Inset, representative fluorescence increase after mixing 0.5 M M5B S1 and 12.5 M mantADP in a stopped-flow spectrophotometer. Single exponential fit to the transient results in a k obs of 48.65 s Ϫ1 . B, ADP release kinetics from M5B S1 . Shown are the normalized data of the time-dependent increase in the intrinsic fluorescence signal after mixing 0.5 M M5B S1 in the presence of a 25 M ADP with 1.5 mM ATP. The transient was fit to a single exponential, yielding the ADP release rate constant k ϪD ϭ 18.9 s Ϫ1 . Similar, rapidly mixing 0.5 M M5B S1 in the presence of a 50-fold molar excess of ADP (5 M mantADP and 20 M ADP) with 1.5 mM ATP resulted in a k ϪD of 18.0 s Ϫ1 . C, mantADP dependence of the observed rate constant k obs after mixing with actoM5B S1 under pseudo-first order conditions. The solid line describes the second-order binding rate constant k ؉AD ϭ 2.71 Ϯ 0.11 M Ϫ1 s Ϫ1 and the ADP release rate constant k ؊AD ϭ 21.9 Ϯ 0.9 s Ϫ1 . Inset, time-dependent fluorescence increase after mixing 0.5 M actoM5B S1 with 7.

Kinetics of myosin-5B
inhibitor concentrations resulted in precipitation in the presence of high F-actin concentrations. MyoVin-1 inhibits the actin-activated ATPase activity of M5B S1 in a dose-dependent manner by reducing k cat (Fig. 6, B and C). The almost linear dependence of the ATPase activity on [F-actin] at high concentrations of myoVin-1 additionally suggests that the compound causes an increase in K app and k cat (Fig. 6, B and C).
Next, we tested whether myoVin-1 would inhibit the speed of actin filament gliding over a M5B HMM -decorated surface in the in vitro motility assay. M5B HMM smoothly moves actin filaments with a speed of 354 Ϯ 119 nm s Ϫ1 . The presence of 500 M myoVin-1 reduces the gliding speed to 201 Ϯ 97 nm s Ϫ1 without completely inhibiting the motion, and many actin filaments move less regularly with periods of pauses (Fig. 6C). This observation is in line with the increased K app observed in steady-state kinetic assays and suggests that the actin-activated ADP release is reduced in the presence of the compound (Fig. 6, B and C).
To investigate in more detail the transient kinetic transitions that are modulated by myoVin-1, we performed single-turnover measurements. By rapidly mixing M5B S1 with substoichiometric concentrations of mantATP, the molecular events of nucleotide binding, hydrolysis, and the release of the hydrolysis products can be monitored by the fluorescence increase, plateau, and decrease of the mant moiety. As shown in Fig. 6D, myoVin-1 predominantly reduces the release of mantADP in a concentration-dependent manner.
To explore how myoVin-1 modulates M5B S1 motor activity at the structural level, we performed in silico blind docking for unbiased mapping of the binding site and binding patterns in a homology model of the human myosin-5B motor domain with Mg⅐ADP⅐VO 4 in the active site. The nucleotide analog was included to account for the noncompetitive mode of action of myoVin-1. Of the top 10 ranked binding conformations, 7 ligands clustered near and around the relay helix, the thiolreactive region (SH1-SH2), three of the seven-stranded ␤-sheets of the transducer, and the converter region of the myosin motor domain (Fig. 6, E and F) at a distance of 17 Å from the active site. The close-up view of energetically most favorable binding pose (binding energy ϭ Ϫ10.5 kcal/mol) of myoVin-1 shows that the ligand interacts closely with interfaces formed by the seven-stranded ␤-sheets of the transducer on one side and hydrophobic residues including Trp-697 from the converter and the hydrophobic methylene group of Arg-152. The pyrazolopyrimidine group of myoVin-1 is further stabilized by Gln-149 of the ␣-helix that connects the sevenstranded ␤-sheet transducer (Fig. 6F). The predicted binding site in great distance to the active site supports previous reports that identified myoVin-1 as an uncompetitive myosin inhibitor (41).

Discussion
We chose to determine the kinetic and regulatory features of mammalian myosin-5B with single-headed and double-headed constructs that contain the entire neck domain with its six IQ motifs (Fig. 1A). The M5B S1 construct is therefore considerably longer than the myosin-5 constructs commonly studied (16, 35, 46 -48). We chose this approach because recent studies indicate that the light chains bound to the myosin neck region are (i) not passive components of the holoenzyme, (ii) can directly interact with the myosin motor domain and each other in some cases, and (iii) may therefore directly influence myosin motor function (29,30). Moreover, the length of the neck domain, which defines the number of IQ motifs in the myosin heavy chain, determines the kinetic and mechanical activity of the protein when artificially altered (49 -51). The M5B S1 construct with its six IQ motifs allows a direct comparison between with the single-headed and double-headed myosin constructs and the contribution of a second head on myosin motor function and intramolecular gating. Moreover, our detailed kinetic characterization allows for the comparative analysis of kinetic signatures in the class 5 myosome, the description of the interaction of myosin-5B with the small molecule inhibitor myoVin-1, and the prediction of the impact of disease-causing mutations on myosin-5B motor function. . Transient kinetic interaction between M5B S1 and F-actin in the presence and absence of ADP. A, increasing concentrations of F-actin were mixed with M5B S1 in a stopped-flow spectrophotometer in the presence or absence of saturating ADP. Linear fits to the data sets describe the second-order binding rate constants k ؉A and k ؉DA to 8.39 Ϯ 0.16 and 5.98 Ϯ 0.19 M Ϫ1 s Ϫ1 , respectively. Inset, time-dependent change in light scattering after rapidly mixing 0.5 M M5B S1 with 4 M F-actin in the presence of 80 M ADP. Single exponential fit to the data set results in a k obs of 13.6 1 s Ϫ1 . B, a cosedimentation assay shows that an actomyosin complex is formed in SF buffer. Sample 1, 2 M M5B S1 (control); sample 2, 10 M F-actin (control); sample 3, 2 M M5B S1 and 10 M F-actin. Abbreviations S and P refer to supernatant and pellet, respectively.

Steady-state and transient kinetic signatures of M5B S1
With thorough transient and steady-state kinetic assays, we show that myosin-5B is a bona fide actin-based molecular motor with a high actin-activated ATPase activity under steady-state conditions. The transient kinetic signatures of M5B S1 include low thermodynamic (K AD /K D ϭ ϳ1-1.7) and kinetic coupling ratios (k ؊AD /k ϪD ϭ ϳ1-1.5), a high k ؉AD / K 1 k ؉2 ϭ ϳ10 ratio, and an intermediate duty ratio of ϳ0.46 -0.6 ( Table 2). Other characteristics of the M5B S1 ATPase cycle are a low k cat /k ؉2 ratio of 0.027 that favors the dissociation of the actomyosin complex and a catalytic efficiency of k cat /K app ϳ0.8 M Ϫ1 s Ϫ1 that indicates a low number of ATP turnovers per F-actin encounter (48). The kinetic signatures of M5B S1 indicate that the molecule has an intermediate duty ratio, in between the parameters determined for the nonprocessive M5C S1 and the processive M5A S1 (Tables 1 and 2).
Our results disagree to some extent with an initial report on the kinetic properties of human myosin-5B (52). Although this study did not include a detailed description of the myosin kinetic cycle in the actin-detached states and the interaction of M5B S1 with F-actin, we find that the presented second-order binding rate constants for ATP and the Michaelis-Menten parameter of the steady-state ATPase activity agree with the parameters determined in this study (52). Discrepancies were found in the nucleotide release rates that are slightly reduced when compared with our work and result in a higher duty ratio for a single-headed M5B S1 construct (52). In summary, the kinetic characteristics of M5B S1 presented in this study suggest that myosin-5B has a limited processive ability and is likely to work in an ensemble of motors to efficiently transport vesicles along the cellular F-actin cytoskeleton.

The second head in M5B HMM establishes the kinetic prerequisite for single molecule processivity
Single molecule processivity is a feature of higher eukaryotic myosins-5A (17,18). This behavior is in striking contrast to mammalian myosin-5C, Drosophila myosin-5, and the Saccharomyces myosins-5 Myo2p and Myo4p that have mechanoenzymatic properties that do not support single molecule processivity in biochemical in vitro assays (16,47,53). To either power or maximize the efficiency of myosin-5-based transport functions in the three-dimensional actin cytoskeleton of a cell, the motors cluster on the surface of cargo vesicles in densities up to Ͼ100 molecules to establish processivity (1,54). This number may be crucial to maintain trafficking function, especially to compensate for either a reduced number of myosins-5 resulting in low surface densities or motors with compromised mechanoenzymatic activity as they are found in some disease states (32,55). This makes the information regarding whether the myosin-5B dimer possesses the kinetic prerequisites for single molecule processivity essential to understand its function in normalcy and pathology.

Kinetics of myosin-5B
To address the question of whether mammalian myosin-5B has the kinetic prerequisites to establish single molecule processivity, we explored the kinetic properties of double-headed M5B HMM . The presence of a second head induces gating in the dimer that reduces the actin-activated ADP release rate k ؊AD of ϳ65% and the k cat of ϳ20% when compared with M5B S1 under identical assay conditions (Tables 1-3). Our transient and steady-state kinetic ensemble measurements further indicate that the ADP release is most likely not the rate-limiting step in the M5B S1 kinetic cycle but the rate-limiting step in the M5B HMM kinetic cycle. This observation highlights that a second head in M5B HMM influences the kinetics of the first head, as has been shown for the gating between the two heads of other unconventional myosins associated with transport function (18, 56 -59).
As reported previously, myosins that move processively, either as dimers or as filaments, typically have a calculated duty ratio Ͼ0.95 (35,38). Based on our duty ratio calculations on M5B HMM that neglect every form of communication and steric constraints between the two heads, we speculate that the myosin-5B dimer is weakly processive and that a higher duty ratio and hence robust processivity may be achieved in the presence of actin-binding proteins or when stain is induced by a vesicle load. Both regulatory effects on the mechanical processivity have been described for some myosins-5, including nonprocessive myosins-5, and will be addressed for myosin-5B in future studies (60 -63).

Potential impact of disease-associated mutations in the myosin-5B motor domain
14 mutations in MYO5B result in amino acid changes in the motor domain at the protein level (26). To explore the possible effect of these mutations on myosin motor function, we predicted the functional impact of the mutations, mapped the mutations in a homology model of myosin-5B, and linked the structural analysis to our kinetic data (64).
The functional impact score of the mutations is shown in Fig.  7A, the respective location in the myosin motor domain in Fig.  7B. Of the 14 mutations, 2 have low, 2 have medium, and 10 have high functional impact scores (Fig. 7A), thereby ranking their predicted impact on myosin motor function (64). The mutations are distributed throughout the entire myosin motor domain including the active site (G168R/R219H) and the actinbinding region (C514R/R531W/P619L) (Fig. 7B). The other mutations are located outside the functional structural ele-ments in allosteric regions of the motor domain: mutations V108G, G316R, R401H, N456S, and R656C were previously attributed to play crucial roles in allosteric communication pathways in the myosin motor domain, whereas mutations G143R, G435(R/E), and R656C are predicted to induce protein instability and misfolding (26).
Among the mutations with a medium or high functional impact score are G168R and R219H in the conserved P-loop and switch-1 loop in the myosin active site. Both structural elements play crucial roles in the coordination of the nucleotide throughout the myosin kinetic cycle (Fig. 1B). Switch-1 mutations at equivalent positions to the highly conserved Arg-219 have previously been described to reduce the actin-activated ATPase activity and motile properties of class 5 and 2 myosins by interfering with the hydrolysis of ATP (65)(66)(67). The resulting low-duty ratio motors are expected to have a reduced processivity when compared with the wild-type protein. This property may be of physiological significance in the myosin-5B R219H mutant that, based on our kinetic characterization of the wild-type protein, is not expected to have robust single molecule processivity. A reduction in duty ratio and processivity may be reflected in the aberrant vesicle trafficking and segregation patterns in patients with microvillus inclusion disease.
Similarly, the replacement of a negatively charged with a bulky residue in the myosin-5B mutant R531W is expected to weaken the interaction with the actin filament that prominently involves electrostatic interactions between positively charges residues in the myosin motor domain and the negatively charged surface of the actin filament. A reduced F-actin interaction is expected to weaken the affinity of the motor domain for its track in the presence of ATP, thereby limiting the degree of processivity (59).

Why does myosin-5B binding not quench the fluorescence of pyrene-labeled F-actin?
Apart from a kinetic description of single-and doubleheaded myosin-5B constructs, this study revealed that the motor does not significantly quench pyrene fluorescence in transient-kinetic in vitro assays. Site-specific labeling of actin on Cys-374 is a powerful method to measure binding of most myosins to F-actin (68 -71). Cys-374 is located on the surface of the actin subunit of a filament in the vicinity of the myosinbinding site. The fluorescence of pyrene-actin is only quenched when myosin binds in the strong binding state (AM or AM⅐ADP; Fig. 1B), but not in the weak binding state (AM⅐ATP Table 3 Steady

-state and key transient-state kinetic parameters of mouse M5B HMM in comparison with previously characterized class-5 myosins
Numbering of the transient-state kinetic constants refers to Fig. 1B

Kinetics of myosin-5B
and AM⅐ADP⅐P i ; Fig. 1B) (34). The light scattering signal in contrast can be used as readout for the binding of myosin to actin in both the strong and the weak states (34). Our observation that the interaction between M5B S1 and pyrene-actin does not quench fluorescence but induces a change in light scattering signal further supports the results from the transient kinetic analysis that indicate that the motor binds F-actin weakly. However, actin sedimentation assays in the same buffer indicate that the actomyosin complex can form under the experimental conditions and that the lack of pyrene signal is caused by different interactions at the actomyosin interface when compared with other myosins-5. A similar result has been obtained for other myosins, including Drosophila indirect flight muscle myosin-2, rat myosin-9B, and also a truncated, single-headed myosin-5A construct (48,72,73).

Chemical inhibition with myoVin-1
The pyrazolopyrimidine compound myoVin-1 was identified as an uncompetitive inhibitor for myosin-5A based on privileged chemical scaffold design. The small molecule did not inhibit skeletal muscle myosin-2, nonmuscle myosin-2, and myosin-6 in vitro, suggesting selectivity for class 5 myosins (41). In this work, we show that myoVin-1 inhibits M5B S1 (K I ϭ 23.12 Ϯ 2.07 M) with an almost identical K I as a single-headed myosin-5A construct (K I ϭ 24 Ϯ 4 M) (41). Similar to the effects reported for myosin-5A, we find that myoVin-1 partially inhibits the actin-activated ATPase activity of M5B S1 (41). Partial inhibition is also observed in the in vitro motility assay, suggesting that a decrease in the actin-activated ADP release is the kinetic step in the actomyosin mechanoenzymatic cycle that is altered in the presence of the small molecule compound.
MyoVin-1 has been used to inhibit the myosin-5B-dependent trafficking of podocalyxin to apical membrane insertion sites and myosin-5-dependent intersynaptic vesicle exchange in live cells (43,44). Vesicle velocity and directionality are slightly reduced in the presence of myoVin-1, whereas the frequency of vesicle pausing increases (43). These quantitative in vivo results agree with the observed partial inhibition of the myosin ATPase activity and functional activity and the repetitive motion of actin filaments in the in vitro motility assay in this study (Fig. 5, A-D). Intriguingly, myosin-5 moves vesicles with a speed of 320 Ϯ 80 nm s Ϫ1 in vivo that drops to 280 Ϯ 60 nm s Ϫ1 in the presence of 30 M myoVin-1 (43). This effect is very similar to the speeds measured in the in vitro motility assay (354 Ϯ 119 nm s Ϫ1 (no myoVin-1); 201 Ϯ 97 nm s Ϫ1 (500 M myoVin-1) (Fig. 6C). It is also of note that myoVin-1 inhibits the activity of skeletal muscle myosin-2 in prothrombin activation, a finding that contrasts the above mentioned selectivity for class 5 myosins (74).
Based on our blind docking results, we predict the inhibitor binding pocket to be located near the converter region in the myosin motor domain (Fig. 6, E and F). The predicted binding pocket is remote from the active site, underlining the uncompetitive mode of inhibition of myoVin-1. Interestingly, the same pocket was recently identified in structural studies as the binding site for the myosin-2 inhibitor CK-571, suggesting that the converter/SH1-SH2/relay interface is a promising site to target myosin motor function with small molecules (75).
In summary, the inhibition of the kinetic and functional activity of myosin-5B constructs in biochemical in vitro assays at high inhibitor concentrations, together with the results of previous in vitro and in vivo studies, suggest that myoVin-1 is not a potent inhibitor of class 5 myosins. The lack of selectivity for a myosin-5 paralogs, the almost identical K I values for myosins-5A and -5B in vitro, and the partial inhibition of the sliding velocity and steady-state ATPase activity of potentially all three mammalian myosins-5 in a cell suggest that caution should be used when interpreting the results from cell biological experiments in which myoVin-1 is used.

Construction of expression constructs, recombinant protein production, and preparation
cDNA constructs encoding for murine M5B S1 (amino acids 1-906, molecular mass of 107 kDa) and M5B HMM (amino acids

Kinetics of myosin-5B
1-1095, molecular mass of 127 kDa) were inserted in a modified pFastBac1 encoding for a C-terminal Flag tag with standard cloning techniques. Rat CaM cDNA (molecular mass of 17 kDa) that is at the protein level 100% identical with murine CaM was expressed from pFastBac1. Recombinant baculovirus generation and gene expression were performed as recommended by the manufacturer (Thermo Fisher Scientific). For protein production, Sf9 insect cells were infected with recombinant baculoviruses encoding M5B S1 /M5B HMM and CaM, and the protein complexes were purified as described via Flag capture (76). M5B HMM was further purified to electrophoretic homogeneity by size exclusion chromatography on a HiLoad Superdex 16/600 pg (GE Healthcare Life Sciences). Actin was prepared from rabbit skeletal muscle acetone powder (Pel-Freez Biologicals) and labeled with N- (1-prenyl)iodoacetamide (pyrene) as described (69,77).

Steady-state and transient state kinetic assays
Steady-state kinetic assays were performed in a Cary 60 spectrophotometer (Agilent Technologies) at a temperature of 25°C in buffer containing 10 mM MOPS, pH 7.0, 2 mM ATP, 50 mM NaCl, 2 mM MgCl 2 , 0.15 mM EGTA, 40 units/ml L-lactic dehydrogenase, 200 units/ml pyruvate kinase, 200 M NADH, and 1 mM phosphoenolpyruvate at a protein concentration of Ͻ50 nM M5B S1 or M5B HMM as described (76). Transient kinetic assays were carried out on a Sf-61 DX2 stopped-flow (Hi-Tech Scientific) equipped with a 75-watt mercury-xenon arc lamp at a temperature of 20°C in SF buffer containing 25 mM MOPS, pH 7.0, 100 mM KCl, 5 mM MgCl 2 , and 0.1 mM EGTA unless stated otherwise. Concentrations throughout the text represent post-mixing concentrations. Light scattering was excited at a wavelength of 320 nm, and emission was measured at a fixed angle of 90°. The intrinsic tryptophan fluoresce was excited at a wavelength of 297 nm, and the emission was recorded after passage through a WG320 filter. The fluorescence of mant-nucleotides was either directly excited at a wavelength of 365 nm or indirectly after energy transfer from tryptophan residues, and the emission was detected after passage through a LP390 filter.
The small molecule myoVin-1 (Calbiochem) was diluted in DMSO to a concentration of 50 mg/ml (92.83 mM) and stored at Ϫ20°C. Kinetic measurements in the presence of myoVin-1 were carried out at a constant concentration of 2% DMSO. Reaction mixtures were incubated for 10 min at room temperature protected from light prior to the assay.

Cosedimentation assay
Proteins (2 M M5B S1 , 10 M F-actin) were incubated in SF buffer for 10 min at room temperature. The samples were centrifuged (30 min, 100,000 ϫ g, 4°C), and supernatant and pellet fractions were separated on a 4 -12% Bis-Tris gel (Thermo Scientific). The gel was stained with PageBlue (Thermo Scientific), destained with water, and documented on an Odyssey scanner (LI-COR Biosciences).

In silico homology modeling and docking
The human myosin-5B motor domain (amino acids 1-767) was modeled using the human myosin-5C motor domain (Protein Data Bank code 4ZG4) pre-powerstroke state structure as a template. Both proteins share 68% sequence identity and 81% homology. The model was built using Modeler 9.18 (78). For the myoVin-1-binding site analysis, the human myosin-5B motor domain model bound to Mg⅐ADP⅐VO 4 was used for docking studies. AutodockTools was used to prepare the ligands and to assign Gasteiger charges (79). Blind docking was performed using autodock Vina by defining a cubic grid size of 79 Å. The exhaustiveness level was set at 50, and default options were used for the remaining parameters. The top 10 best docked ligand poses were selected based on the free energy predictions (Ϫ10.5 kcal/mol to Ϫ8.5 kcal/mol).