The kinetic mechanism of Myo1e (human myosin-IC).

Myo1e is the widely expressed subclass-1 member of the myosin-I family. We performed a kinetic analysis of a truncated myo1e that consists of the motor and the single IQ motif with a bound calmodulin. We determined the rates and equilibrium constants for the key steps in the ATPase cycle. The maximum actin activated ATPase rate (V(max)) and the actin concentration at half-maximum of V(max) (K(ATPase)) of myo1e are similar to those of the native protein. The K(ATPase) is low (approximately 1 microm), however the affinity of myo1e for actin in the presence of ATP is very weak. A weak actin affinity and a rapid rate of phosphate release result in a pathway under in vitro assay conditions in which phosphate is released while myo1e is dissociated from actin. Actin activation of the ATPase activity and the low K(ATPase) are the result of actin activation of ADP release. We propose that myo1e is tuned to function in regions of high concentrations of cross-linked actin filaments. Additionally, we found that ADP release from actomyo1e is > 10-fold faster than other vertebrate myosin-I isoforms. We propose that subclass-1 myosin-Is are tuned for rapid sliding, whereas subclass-2 isoforms are tuned for tension maintenance or stress sensing.

Myosin-I Expression and Purification-The cDNA for human myosin-IC (accession NM 004998), kindly provided by W. M. Bement (University of Wisconsin), was truncated at Glu 720 , generating a construct containing the motor domain and the only IQ motif (referred to as myo1e IQ throughout the paper). A FLAG peptide sequence was inserted at the C terminus and subcloned into the baculovirus transfer vector pVL1392 (Invitrogen). Recombinant baculovirus was generated using standard procedures.
Myo1e IQ with bound calmodulin was purified from Sf9 cells that were coinfected with virus containing recombinant myo1e IQ and CaM (Fig.  1). Four liters of log phase cells (2 ϫ 10 6 cells/ml) were infected and incubated at 27°C for 60 h with shaking. Cells were harvested by centrifugation; suspended in lysis buffer (10 mM Tris, pH 7.0, 200 mM NaCl, 2 mM MgCl 2 , 5 mM DTT, 1 mM phenylmethylsulfonyl fluoride, 0.01 mg/ml aprotinin, 0.01 mg/ml leupeptin), 2 mM MgATP, and 0.05% Igepal at 4°C; and homogenized with five strokes in a Dounce homogenizer. Cell extract was centrifuged at 100,000 ϫ g for 1 h. The supernatant was loaded onto anti-FLAG antibody columns (Sigma). Columns were washed with 5 column volumes of lysis buffer ϩ 2 mM MgATP and 5 column volumes of lysis buffer. Myo1e IQ was eluted with 10 mM Tris, pH 8.0, 100 mM NaCl, 1 mM DTT, 5 M CaM, 0.2 mg/ml FLAG peptide (Sigma), 0.01 mg/ml aprotinin, 0.01 mg/ml leupeptin. Eluted protein was loaded directly on to an 8-ml Mono Q column (Amersham Biosciences) equilibrated in column buffer (10 mM Tris, pH 8.0, 25 mM KCl, 1 mM DTT) and eluted with a linear 25 mM-1 M KCl gradient. The Mono Q column separated myo1e IQ from FLAG peptide, ADP, ATP, and free CaM. Fractions containing myo1e IQ were dialyzed versus storage buffer (10 mM Tris, pH 7.5, 100 mM KCl, 1 mM EGTA, 1 mM DTT, 50% glycerol), which concentrated the protein and allowed for storage at Ϫ20°C. Quantitative densitometry showed that myo1e IQ was Ͼ95% pure, and the CaM:myo1e IQ ratio was 1:1. Approximately 2 mg of pure myo1e protein was obtained from 4 liters of cells.
Steady-state ATPase Activity and Sedimentation Assays-Steadystate ATPase activities were measured in KMg50 or KMg0 buffer at 25°C using the NADH-coupled assay as described (27). Steady-state binding of 0.2 M myo1e IQ to 0 -50 M actin was measured in KMg50 and KMg0 at 25°C in the presence of 2 mM MgATP by ultracentrifugation assays (350,000 ϫ g for 20 min). The fraction of actin-bound myosin was determined by assaying the NH 4 /EDTA ATPase activity of the supernatant (29).
Stopped Flow, Quenched Flow, and Kinetic Modeling-Transient kinetic measurements were made at 25°C with an Applied Photophysics (Surrey, UK) SX.18MV stopped flow having a 1.2-ms dead time. Tryptophan fluorescence ( ex ϭ 295 nm) was measured using a 320 nm WG long pass emission filter (Oriel). A 400 nm long pass filter (Oriel) was used to monitor pyrene ( ex ϭ 365 nm), and mantADP and man-tATP ( ex ϭ 295 nm) fluorescence. Usually three to five transients were averaged before nonlinear least square fitting. The time courses presented in the figures show the average of one to four individual traces. Transients were fitted to exponential functions using the software supplied with the stopped flow. Transient P i release was measured using the coupled assay system containing the fluorescently labeled mutant of the phosphate-binding protein (P i BP (30,31)) with the stopped flow in sequential mixing mode using an excitation wavelength of 425 nm and a 440 nm long pass filter. The dead time of the instrument in this configuration was ϳ2 ms. Quenched flow measurements were performed with a KinTek (State College, PA) RQF-3 apparatus. Errors reported are standard errors in the fits.
Kinetic modeling and simulations were performed using Scheme 1 of the actomyosin ATPase, where A is actin and M is myosin. Computer simulations were performed with KSIM (Neil C. Miller) or Berkeley-Madonna (Robert I. Macey and George F. Oster; University of California, Berkeley).

Steady-state ATPase and Actin Binding Parameters-The
ATPase activities of subclass-1 myosin-I isoforms from all lower eukaryotes are activated by phosphorylation of serine or threonine at the TEDS site in an actin binding, surface loop by proteins that have homology to p21-activated kinases (20). Although myo1e has serines immediately adjacent to the TEDS site, it is not predicted to be regulated by heavy chain phosphorylation (20). We were unable to phosphorylate myo1e IQ with Pak3 or Acanthamoeba myosin-I heavy chain kinase (not shown), thus confirming the TEDS rule (20).
The steady-state ATPase activity of myo1e IQ is activated ϳ2-fold by actin filaments. Steady-state ( Fig. 2) and single turnover measurements (not shown) gave rates of ϳ0.6 s Ϫ1 in the absence of actin (Table I) in KMg50. The actin concentration at half-maximum of the steady-state ATPase rate (K ATPase ) is 1.2 Ϯ 0.36 M, and the maximum ATPase rate (V max ) at saturating actin is 1.2 Ϯ 0.03 s Ϫ1 . The triphasic ATPase activity observed with native myo1e was not observed (19).
Steady-state actin binding measurements show that myo1e IQ is dissociated predominantly from 0 -50 M actin in KMg0 (Fig. 2) and KMg50 (not shown) in the presence of 2 mM MgATP. Therefore, in the presence of ATP, the predominant steady-state myo1e IQ intermediates have a very weak affinity for actin with Quantitative densitometry shows that the CaM:heavy chain stoichiometry was 1:1. The predicted molecular mass of the heavy chain from the amino acid sequence is 83 kDa.

AM
Strongly bound Weakly bound Strongly bound Kinetic Mechanism of Myo1e an equilibrium dissociation constant (K d ) greater than 50 M. Myo1e IQ Binding to Actin Filaments-An ϳ80% fluorescence quenching upon strong binding of myo1e IQ to pyrene-actin allowed us to monitor the association of myo1e IQ with actin in the absence (Reaction 1) and presence (Reaction 2) of MgADP. Data were modeled as where A* represents the unquenched fluorescent state of pyrene-actin. Time courses of myo1e IQ and myo1e IQ -ADP binding to pyrene-actin follow single exponentials rates that depend linearly on the actin concentration (Fig. 3A). The apparent second order rate constant for myo1e IQ binding to actin obtained from the slope is k Ϫ7 ϭ 9.0 Ϯ 0.7 M Ϫ1 s Ϫ1 (Table II). The apparent second order rate constant for myo1e IQ -ADP binding to actin is k Ϫ11 ϭ 1.2 M Ϫ1 s Ϫ1 (Table II).
The rate of dissociation of myo1e IQ and myo1e I -ADP from pyrene-actin was measured by competition with 60-fold excess of unlabeled actin (Fig. 3B). Time courses fit single exponentials with rates k ϩ7 ϭ 0.056 s Ϫ1 and k ϩ11 ϭ 0.065 s Ϫ1 (Table II). Dissociation equilibrium constants (K 7 and K 11 ) calculated from the dissociation and association rates (k ϩ7 /k Ϫ7 and k ϩ11 / k Ϫ11 ) are K 7 ϭ 6.2 nM and K 11 ϭ 55 nM (Table II).
MantATP Binding to Myo1e IQ and Actomyo1e IQ -We used the enhancement in fluorescence of the ATP analog mantATP to measure the rate of nucleotide binding. Unlike skeletal muscle myosin-II (32), an intrinsic tryptophan fluorescence change upon MgATP binding was not detected. The transient increase The actin concentration dependence of the steady-state turnover of 50 nM myo1e IQ in KMg0 (f) and KMg50 (q) was measured using the NADH-coupled assay. The data points represent acquisitions from three different preparations, and the solid lines are best fits to rectangular hyperbolas. The fraction of myo1e IQ bound to actin in KMg0 (ࡗ) in 2 mM ATP was measured by cosedimentation.
a Maximum steady-state ATPase rate in the presence of saturating actin filaments.
b Actin filament concentration at half maximum activation of steadystate ATPase.
c Steady-state ATP turnover rate in the absence of actin filaments. d Uncertainties represent standard errors in the best fits of the data.
in mantATP fluorescence was best fit to two exponentials (data not shown). The two exponentials are not the result of the presence of multiple mantATP isomers (33) because a two exponential fluorescence increase was also detected when experiments were performed with dmantATP, which is a singleisomer preparation of mantATP. The amplitude of the fast exponential phase is 70 -80% of the signal, and its rate is linearly related to the dmantATP concentration ( Fig. 4A). We assumed a two-step ATP binding reaction, where K 1 is a rapid equilibrium and k ϩ2 is a rate-limiting isomerization to the high fluorescence state of dmantATP (ATP*). The apparent second order rate constant for dmant-ATP binding, given by the slope of the plot of the observed rate versus nucleotide concentration, is K 1 k 2 ϭ 11 Ϯ 0.4 M Ϫ1 s Ϫ1 (Fig. 4A). Fits of the slow exponential phase yielded an apparent second order rate constant 0.65 M Ϫ1 s Ϫ1 (Fig. 4A, inset). It was proposed for skeletal muscle subfragment-1 that this slow rate reports the conformational change that accompanies ATP hydrolysis (k 3 , Scheme 1 (33)). A two-exponential fluorescence increase was also observed in the presence of actin. The amplitude of the fast exponential phase is 40 -60% of the signal, and its rate is linearly related to the nucleotide concentration with an apparent second order (Fig. 4B). The linear fit of the rates does not pass through the origin. The apparent second order rate constant determined from the slow exponential phase is 0.15 M Ϫ1 s Ϫ1 (Fig. 4B, inset).
ATP-induced Population of the Weakly Bound States-Pyrene-actin fluorescence was used to monitor the ATP-induced population of the weakly bound states (13,34). Mixing ATP with pyrene-actomyo1e IQ resulted in an increase in fluorescence (Fig. 5A, upper trace in inset). Fluorescence transients were best fit to single exponentials at all ATP concentrations examined, and the rate of the fluorescence increase was hyperbolically related to the ATP concentration (Fig. 5A). The mechanism of ATP-induced fluorescence enhancement was modeled as shown in Reaction 4, where K 1 Ј is a rapid equilibrium, k 2 Ј is a rate-limiting isomerization to the high fluorescence A*M⅐ATP state, and k ϩ9 is the dissociation rate. The maximum rate is k 2 Ј ϭ 440 Ϯ 9.0 s Ϫ1 and K 1 Ј ϭ 397 Ϯ 23 M (Table II). The association rate constant for MgATP binding to actomyo1e IQ obtained from the initial slope 5B) with a y intercept of zero.
Light scattering measurements were used to monitor the rate of actomyo1e IQ dissociation by MgATP (Fig. 5A). This measurement is different from the pyrene-actin measurements in that pyrene-actin fluorescence reports the population of the FIG. 4. dmantATP binding to myo1e IQ and actomyo1e IQ . A, rate of dmantATP binding to 0.5 M myo1e IQ and B, dmantATP binding to 0.5 M actomyo1e IQ as a function of nucleotide concentration. The observed rates (k obs ) were obtained by fitting the stopped flow fluorescence data at each nucleotide concentration to the sum of two exponentials. Apparent second order association rate constants were determined from the slopes of the linear fits of the fast phase of the two exponential fits (solid lines). The insets show the nucleotide dependence of the slow phase of the two exponential fits. attached and detached M⅐ATP states (Scheme 1). Mixing MgATP with actomyo1e IQ resulted in a rapid decrease in light scattering (Fig. 5A, lower trace in inset). The time courses follow single exponentials with rates that depend hyperbolically on the MgATP concentration (Fig. 5A). Fits to Reaction 4 yielded rate constants (K 1 Ј ϭ 334 Ϯ 46 M; k ϩ2 Ј ϭ 439 Ϯ 23 s Ϫ1 ) nearly identical to those obtained with pyrene-actin (Table II). Therefore, the maximum rate of dissociation (k ϩ9 ) upon ATP binding is limited by k ϩ2 Ј (Reaction 4).
ATP Hydrolysis-The rate of ATP hydrolysis in the absence of actin (k ϩ3 ϩ k Ϫ3 ; Scheme 1) was measured directly by quenched flow (Fig. 6). There is a rapid initial burst of ATP hydrolysis which fits a single exponential with a rate of 108 Ϯ 14 s Ϫ1 at 50 M MgATP. The rate of ATP binding under experimental conditions is expected to be Ͼ500 s Ϫ1 (Table II), so this measurement is not limited by the rate of nucleotide association. The presence of a phosphate burst indicates that the rate-limiting step occurs after ATP hydrolysis. The amplitude of the phosphate burst is 0.55 P i /myosin (Fig. 6). If we assume that B ϭ K 3 /(1 ϩ K 3 ) and K 3 ϭ (k ϩ3 /k Ϫ3 ), then K 3 ϭ 1.2, k ϩ3 ϭ 59 s Ϫ1 , and k Ϫ3 ϭ 49 s Ϫ1 (Table II).
Phosphate Release-Fluorescently labeled P i BP was used to measure directly the rate of phosphate release (k ϩ4 and k ϩ4 Ј) in sequential mix, single turnover, stopped flow experiments ( Fig.  7A (30)). Apyrase-treated myo1e IQ (2 M) was mixed with 1 M ATP, aged for 400 ms to allow for ATP binding, and mixed with 0 -40 M actin (Fig. 7B). P i BP was included with the myo1e IQ and the actin to prevent transients resulting from phosphate released during the age time or phosphate contamination in the actin.
The time course of phosphate release in the absence of actin follows a single exponential with a rate of 1.5 s Ϫ1 in KMg0 and KMg50 (Fig. 7A). There is no lag in the phosphate release transients (Fig. 7A, inset), indicating that phosphate release precedes ADP release. Unlike previously characterized myosins (28,30), actin did not increase the rate of phosphate release (Fig 7B). This result is likely caused by the weak affinity of the M⅐ADP⅐P i state for actin (see "Discussion").
ADP Association and Dissociation-The fluorescence increase of mantADP was used to determine the rate constants for ADP association to myo1e IQ and actomyo1e IQ (Fig. 8). In the absence of actin, time courses of fluorescence fit two exponential rates with approximately equal amplitudes. The fast phase was hyperbolically related to the nucleotide concentration (Fig.  8A) and modeled as a two-step binding reaction, where * indicates high mantADP fluorescence. The maximum rate is k Ϫ5 ϭ 237 Ϯ 13 s Ϫ1 and K 6 ϭ 6.0 Ϯ 0.9 M (Table II). The rate of the slow exponential phase shows only a slight nucleotide dependence and plateaus at ϳ11 s Ϫ1 at Ͼ10 M mantADP (Fig. 8A, inset). The origin of this second exponential is not known, but it may report the population of a second M⅐ADP state (see "Discussion").
In the presence of actin, time courses of fluorescence were best fit to single exponential rates that were linearly related to the nucleotide concentration (Fig. 9A). The signal to noise ratio of the fluorescence signal limited the range of testable mant-ADP concentrations, so we were not able to measure the maximum rate of mantADP binding (k Ϫ5 Ј). Therefore, the apparent second order rate constant for mantADP binding was obtained by a linear fit to the data assuming the two-step binding model of Reaction 5 (k Ϫ5 Ј/K 6 Ј ϭ 13 Ϯ 1.8 M Ϫ1 s Ϫ1 ; Fig. 9A). The non-zero y intercept reveals a dissociation rate constant of k ϩ5 Ј ϭ 93 Ϯ 5 s Ϫ1 .
The rates of mantADP dissociation from myo1e IQ (Fig. 8B) and actomyo1e IQ (Fig. 9A, inset) were measured by competition with 250-fold excess of MgATP. Fluorescence time courses of mantADP dissociation fit two-exponential rates of 6.3 and 1.7 s Ϫ1 in the absence of actin and a single exponential rate of k ϩ5 Ј ϭ 104 Ϯ 4 s Ϫ1 in the presence of actin (Table II). The rate of MgADP dissociation from actomyo1e IQ is similar to that reported by the y intercept in the mantADP association experiment (Fig. 9A).
The affinity of MgADP for actomyo1e IQ was determined by competition experiments in which the ternary pyrene- actomyo1e IQ -ADP complex was mixed with MgATP (Fig. 9B). Fluorescence transients fit single exponential rates that had a hyperbolic dependence on the MgADP concentration, suggesting that AM⅐D and AM states are in rapid equilibrium. Therefore, the dependence of k obs on MgADP concentration is as follows, where k obs is the observed dissociation rate, k 0 is the dissociation rate in the absence of ADP (K 1 Јk 2 Ј[ATP]; Table II), and K d is the dissociation equilibrium constant for ADP (K 5 ЈK 6 Ј). Nonlinear least square fits to the data yielded a dissociation equilibrium constant of K 5 ЈK 6 Ј ϭ 3.9 Ϯ 0.29 M (Table II), which is similar to the calculated value (K 5 ЈK 6 Ј ϭ 8 M) using association and dissociation rate of mantADP.

DISCUSSION
Overview of the Myo1e IQ ATPase-Myo1e IQ has a high basal ATPase rate (v o ; Table I), yet all measured rate constant are Ͼ2-fold faster than v 0 (Table II), so we cannot assign a single step as rate-limiting. The slower turnover number is the result of two sequential steps (phosphate release and ADP release) with nearly equivalent rates. Kinetic modeling of the ATPase reaction using our determined rate constants (Table II) yields an ATPase rate (v 0 ϭ 0.65 s Ϫ1 in KMg50), which is similar to the experimentally determined value (v 0 ϭ 0.58 Ϯ 0.04 s Ϫ1 ; Table I). The predominant steady-state intermediates in the absence of actin are M⅐ADP⅐P i and M⅐ADP (Scheme 1).
Overview of the Actomyo1e IQ ATPase-Myo1e IQ has a low K ATPase and has a ϳ2-fold actin activation of its ATPase rate (Table I), which is similar to that found for the native molecule. Actin (0 -40 M) does not increase the rate of P i release in KMg0 or KMg50 (Fig. 7B). Therefore, actin activation of P i release is not the mechanism for actin activation of the ATPase rate, as it is for other myosins. Rather, the increased ATPase rate is the result of actin activation of ADP release. The predominant pathway under actin concentrations used in our assays is one in which (a) ATP binding, actin dissociation, and ATP hydrolysis are fast and not rate-limiting; (b) P i is released while myosin is detached from actin; (c) M⅐ADP binds to actin; and (d) ADP is released from actin-bound myosin (Fig. 10A,  outlined pathway). The rate-limiting step at saturating actin (V max ) is defined by the rate of P i release (k ϩ4 ), because at saturating actin ADP release is ϳ100 s Ϫ1 (Table II). The K ATPase value is related to the affinity of the M⅐ADP state for actin. Because binding measurements indicate that myo1e IQ binds weakly to actin in the presence of ATP (Fig. 2), the predominant steady-state intermediate is an actin-detached M⅐ADP⅐P i state.
ATP Binding and Actomyo1e IQ Dissociation-The binding of ATP to myo1e IQ and actomyo1e IQ , as determined by dmantATP fluorescence, is fast. The rate of binding at physiological nucleotide concentrations is Ͼ100-fold faster than the rate-limiting step(s). Therefore, the nucleotide-free state of myo1e IQ is not significantly populated. The rate of population of the weak binding states as measured by pyrene-actin fluorescence and light scattering is also fast, and ATP-induced dissociation from actin is limited only by the rate of ATP binding.
The rate of ATP binding to actomyo1e IQ as measured by mant fluorescence is ϳ4-fold faster than the rate measured by pyrene-actin fluorescence or light scattering. Additionally, the linear fit of the rates of dmantATP binding to actomyo1e IQ does not pass through the origin (Fig. 4B) but gives a y intercept of 25 s Ϫ1 . The value of the y intercept likely reports a reverse isomerization rate (k Ϫ2 Ј) of myo1e IQ . However, because the rate of ATP-induced dissociation of actomyo1e IQ is Ͼ400 s Ϫ1 (Fig. 5) and essentially irreversible under the low actin concentration conditions of the experiment, a relatively slow reverse isomerization of the ATP binding step should not be observable. Therefore, the faster rate of mantATP binding and the observed reversal rate (k Ϫ2 Ј) are likely the result of an artifact because of the fluorescent modification of the 3Ј position of ATP. Although the dmantATP experiments probably do not report the true rate constants for ATP binding in the presence of actin, the experiments do reveal differences in the structural isomerization required for tight ATP binding in the presence and absence of actin. Additionally, because similar mant artifacts are not seen with other characterized myosins (e.g. 14), the experiments show that there are significant structural differences at the nucleotide binding sites of myosin-I isoforms.
Phosphate Release-The v 0 and V max rates and the extent of actin activation of myo1e IQ (Table I) are similar to the values measured for native myo1e purified from rat liver (19), so the biochemical rate and equilibrium constants determined for Sf9-expressed myo1e IQ are likely identical to the native molecule. Therefore, the fast rate of P i release (k ϩ4 ) is not an artifact of the expression system or of the expression of a truncated protein.
The rate of phosphate release from all previously characterized myosins is increased by actin binding (28,30,35). However, we did not detect actin activation of phosphate release from myo1e IQ . We propose that the lack of actin activation is caused by the low affinity of the M⅐ADP⅐P i state for actin. Other myosins bind actin in the presence of ATP with equilibrium dissociation constants Ͻ50 M at low ionic strength conditions (36,37), whereas myo1e IQ clearly binds with a dissociation constant Ͼ 50 M (Fig. 2). The ionic component of actin binding has been shown to be mediated by positive charges in surface loop-2 of myosin (36). Loop-2 of myo1e is shorter than those found in other characterized myosins (Fig. 10B) and does not contain the positively charged amino acids in the region that has been shown to be required for actin binding (Fig. 10B (36)). However, loop-2 of myo1e contains the KK region (Fig. 10B,  underlined) shown to be crucial for actin activation of phosphate release (37). Therefore, it is possible that actin activates phosphate release from myo1e IQ , but we are not able to achieve high enough actin concentrations to populate the weakly bound AM⅐ADP⅐P i state. Weak actin binding may be a kinetic adaptation required for cellular function (see below).
It is possible that phosphate release by myo1e IQ is uncoupled from actin binding, so the very high basal rate of P i release (k ϩ4 ) is also the rate when myo1e IQ is attached to actin (k ϩ4 Ј). Why myo1e would evolve a mechanism that seemingly wastes ATP is not clear; however, other mechanisms for the suppression of the basal ATPase rate may exist. For example, the tail domain (which is not present in myo1e IQ ) is thought to regulate the ATPase activity of myo1e allosterically (19). Further experiments are required to determine whether the myo1e tail regulates phosphate release directly.
ADP Binding and Release-Time courses of mantADP binding to myo1e IQ are best fit to two exponentials. The rate of the fast phase is hyperbolically related to the nucleotide concentration as expected for a two-step binding reaction (Reaction 6). The rate of the slow exponential does not show a significant nucleotide dependence and possibly reports the population of a second higher fluorescence state M⅐ADP state, where * indicates the fluorescence state of myosin, k Ϫ5a is the maximum rate of the fast phase, and k Ϫ5b is the maximum rate of the slow fluorescence change. It has been shown previously that myosins exists in more than one ADP state (14,38), so it is not surprising that we are able to detect multiple conformations of myo1e IQ . Dissociation of mantADP in the absence of actin also shows two exponentials that are likely related to these two states. However, a more detailed analysis of ADP binding is required to characterize these states better.
The similarity of the rates of P i release and ADP release in the absence of actin raises the possibility that the products are released simultaneously (Table II). However, if ADP and P i were released together, steady-state measurements of v 0 would be Ͼ2-fold faster (see above).
Time courses of mantADP binding to actomyo1e IQ fit single exponentials that are linearly related to the nucleotide concentration (Fig. 9A). A linear fit to the data does not pass through the origin but reveals a dissociation rate (k ϩ5 Ј) of ϳ100 s Ϫ1 . This dissociation rate is nearly identical to the dissociation rate determined directly by displacing myo1e IQ -bound mantADP with unlabeled ATP (Fig. 9A, inset). The dissociation equilibrium constant (K 5 ЈK 6 Ј) calculated by dividing the dissociation rate (k ϩ5 Ј) by the association rate (k Ϫ5 Ј/K 6 Ј) obtained from the mantADP experiments is within a factor of 2 of that determined from the pyrene-actin fluorescence experiments. Therefore, it is unlikely that the mantADP experiments are reporting artifactually high rates as shown for dmantATP binding to actin (see above).
Comparison with Other Myosin-I-Myo1e is the first vertebrate subclass-1 myosin-I to be characterized. Like the subclass-1 isoforms from Acanthamoeba (13), myo1e has relatively large rate constants that are similar to those of skeletal muscle myosin-II (Table II). Most notably, the rates of ADP release from all subclass-1 isoforms are Ͼ10-fold faster than subclass-2 isoforms, i.e. the lifetimes of the AM⅐ADP states are Ͼ10 longer for subclass-2 isoforms (13)(14)(15)(16). The rate of ADP release limits sliding velocity (39), so we propose that subclass-1 isoforms are better tuned for fast motility, whereas subclass-2 isoforms are better tuned for maintenance of force (14,15).
A difference among the amoeba and vertebrate subclass-1 isoforms is that Acanthamoeba myosin-IA and -IB have low basal ATPase rates (Ͻ0.1 s Ϫ1 ) and a large actin activation of the steady-state ATPase (Ͼ 50-fold (40)), whereas myo1e has an high basal ATPase rate and a low actin activation (Table I). Like myo1e, vertebrate subclass-2 isoforms also have high basal ATPase rates and relatively low actin activations (41). This might be an adaptation to the cellular environment of vertebrate cells, or it might be related to the regulation of myosin-I in these cells (8). A second difference between amoeba subclass-1 isoforms and myo1e is, like subclass-2 isoforms, myo1e has a very high ADP affinity (K 5 ЈK 6 Ј Ͻ 10 M). Therefore, it is possible that ADP release from both vertebrate subclasses is highly force-dependent (14,15).
Relevance of Myo1e Kinetics to in Vivo Function-Like all other characterized myosin-I isoforms (13)(14)(15)(16), myo1e is a low duty ratio motor, so under unloaded conditions it is predominantly weakly bound or detached from actin filaments. Therefore, for myole to support motility, a high effective duty ratio must be created by bringing together locally high concentrations of myosin and actin. Such a mechanism is consistent with its observed cellular localization in regions of high F-actin concentration (17,18).
The force-generating power stroke of myosin is thought to accompany phosphate release (42). Therefore, an ATPase pathway in which phosphate is released while myosin is detached from the actin filament (Fig. 10A, outlined pathway) would not generate force. It is possible that the power stroke occurs at a different step in the myo1e ATPase pathway (e.g. ADP release); however, this possibility is unlikely given the sequence and mechanistic conservation among myosin isoforms. We propose that the low actin affinity of myo1e in the presence of ATP is an adaptation to the high actin concentration environment in which myo1e functions (17,18). Native myo1e cross-links actin filaments via an ATP-insensitive actin binding site in its tail, creating densely packed and cross-linked bundles (19). If the weak binding states (M⅐ATP and M⅐ADP⅐P i ) of myo1e had a high actin affinity, then ATP hydrolysis and the recovery stroke would occur while attached to actin (30). Therefore, an associated hydrolysis pathway would result in the reversal of myosin's power stroke. To ensure that myo1e is detached from actin during the recovery stroke, a weak affinity must be maintained in the pre-force-generating states. We therefore predict that under the high actin concentrations of the cell, the M⅐ADP⅐P i state will bind and release phosphate while attached to actin (Fig. 10A, gray pathway) as proposed for other myosins (43). Further experiments are required to confirm this predic-tion and to understand better the unique kinetic adaptations of all vertebrate myosin-I isoforms.