Dynamics of Myo1c (Myosin-Iβ) Lipid Binding and Dissociation

Myosin-I is the single-headed member of the myosin superfamily that associates with lipid membranes. Biochemical experiments have shown that myosin-I membrane binding is the result of electrostatic interactions between the basic tail domain and acidic phospholipids. To better understand the dynamics of myosin-I membrane association, we measured the rates of association and dissociation of a recombinant myo1c tail domain (which includes three IQ domains and bound calmodulins) to and from large unilamellar vesicles using fluorescence resonance energy transfer. The apparent second-order rate constant for lipid-tail association in the absence of calcium is fast with nearly every lipid-tail collision resulting in binding. The rate of binding is decreased in the presence of calcium. Time courses of myo1c-tail dissociation are best fit by two exponential rates: a fast component that has a rate that depends on the ratio of acidic phospholipid to myo1c-tail (phosphatidylserine (PS)/tail) and a slow component that predominates at high PS/tail ratios. The dissociation rate of the slow component is slower than the myo1c ATPase rate, suggesting that myo1c is able to stay associated with the lipid membrane during multiple catalytic cycles of the motor. Calcium significantly increases the lifetimes of the membrane-bound state, resulting in dissociation rates ≪ 0.001 s−1.

Myosin-I is the single-headed member of the myosin superfamily that associates with lipid membranes. Biochemical experiments have shown that myosin-I membrane binding is the result of electrostatic interactions between the basic tail domain and acidic phospholipids. To better understand the dynamics of myosin-I membrane association, we measured the rates of association and dissociation of a recombinant myo1c tail domain (which includes three IQ domains and bound calmodulins) to and from large unilamellar vesicles using fluorescence resonance energy transfer. The apparent second-order rate constant for lipid-tail association in the absence of calcium is fast with nearly every lipid-tail collision resulting in binding. The rate of binding is decreased in the presence of calcium. Time courses of myo1c-tail dissociation are best fit by two exponential rates: a fast component that has a rate that depends on the ratio of acidic phospholipid to myo1c-tail (phosphatidylserine (PS)/tail) and a slow component that predominates at high PS/tail ratios. The dissociation rate of the slow component is slower than the myo1c ATPase rate, suggesting that myo1c is able to stay associated with the lipid membrane during multiple catalytic cycles of the motor. Calcium significantly increases the lifetimes of the membrane-bound state, resulting in dissociation rates < < 0.001 s ؊1 .
Myosin-I is the widely expressed single-headed member of the myosin superfamily. Myosin-I isoforms play roles in membrane dynamics, cell structure, and mechanical signal transduction (1). Subcellular fractionations of vertebrate cells indicate that most myosin-I is associated with the membrane and cytoskeleton fractions with only a small soluble pool (2,3). Myosin-I is dynamically localized to cell membranes (4 -6), suggesting regulated membrane association. However, very little is known about the control of myosin-I localization and membrane association, though it has been proposed that calcium-calmodulin and phosphorylation play roles in modulating membrane attachment (7).
Electrostatic interactions play an important role in the binding of myosin-I to membranes. Positively charged tail domains of myosin-I bind to acidic phospholipids without any apparent specificity toward the acidic head groups (8 -10). For example, myo1a (brush border myosin-I) binds tightly to vesicles com-posed of phosphatidylserine and phosphatidylglycerol but binds very weakly to phosphatidylcholine (8). Although myosin-I tail binding proteins have been identified in lower eukaryotes (11)(12)(13)(14)(15), none have been identified in vertebrates. Additionally, it is not known if these proteins play a role in anchoring myosin-I to the membrane.
Myo1c (also known as myr2 and myosin-I␤) is widely expressed in vertebrates, is enriched in the perinuclear regions and dynamic cell margins, and is found concentrated in the tips of stereocilia of inner-ear hair cells (for a review, see Ref. 1). Chromophore-assisted laser inactivation (16) and chemical inhibition (17) studies have identified roles for myo1c in lamellipodial structure and mechanical signal transduction. To better understand the dynamics and regulation of myo1c membrane association, we determined the lipid association and dissociation rate constants using a recombinant protein construct containing the three IQ domains and the C-terminal tail region of myo1c (myo1c-tail). Membrane association was detected by fluorescence resonance energy transfer from myo1c-tail tryptophans to dansyl-labeled phosphatidylserine (18,19).

MATERIALS AND METHODS
Reagents, Buffers, and Lipid Preparation-All experiments were performed in HNa100 (10 mM Hepes, pH 7.0, 100 mM NaCl, 1 mM EGTA, 1 mM DTT). 1 Calcium concentrations were adjusted by adding CaCl 2 to HNa100 and are reported as free calcium. All kinetic experiments were performed with 2 M excess calmodulin.
Large unilamellar vesicles (LUVs) were prepared by extrusion. Lipid components were mixed in the desired ratios in chloroform and dried completely under a stream of nitrogen. Lipids were resuspended in HNa100 to a total concentration of 2 mM and were subjected to five cycles of freeze-thaw. Lipids were then passed through 100-nm filters (11 times) using a mini-extruder (Avanti Polar Lipids). LUVs were discarded after 1 day. PS percentages reported throughout the text are the mole percentages of total PS (PS ϩ dPS) with PC making up the remaining lipid. All fluorescent LUVs (dLUVs) contain 5% dPS. Lipid concentrations are given as total lipid (PC ϩ PS ϩ dPS).
Protein Expression and Purification-We cloned and expressed myo1c-tail (amino acids 690 -1028), which consists of three IQ motifs, the tail domain, and an N-terminal HIS 6 tag for purification. The cDNA for myo1c was kindly provided by D. P. Corey (Massachusetts General Hospital), and the appropriate bases were amplified with the following primers: 5Ј-cgcggatcctatgcatcaccatcaccatcacggcgccacagaggactccct-3Ј and 5Ј-cggggtacctcaccgagaattcagccgtg-3Ј. The amplified fragment was cloned into baculovirus transfer vector pBlueBac4.5 (Invitrogen). Recombinant baculovirus was generated using standard procedures and screened by plaque assay.
Myo1c-tail with bound calmodulin was purified from Sf9 cells that were co-infected with virus containing recombinant myo1c-tail and calmodulin. Cells were suspended in 25 mM Tris, pH 7.5, 20 mM imid-* This work was supported by National Institutes of Health Grant GM57247. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  azole, 300 mM NaCl, 0.5% igepal, 0.5 mM EGTA, 1 mM 2-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 0.01 mg/ml aprotinin, and 0.01 mg/ml leupeptin 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 sonicated three times for 15 s, incubated with 10 g/ml RNase A and 5 g/ml DNase I on ice for 15 min, and loaded on to nickel-nitrilotriacetic acid (Qiagen) column. The column was washed with 5 column volumes of the same buffer without igepal. Myo1c-tail was eluted with 5 ml of 125 mM imidazole, 300 mM NaCl, 25 mM Tris, pH 8.0, 0.5 mM EGTA, 1 mM 2-mercaptoethanol, 0.01 mg/ml aprotinin, 0.01 mg/ml leupeptin, and 10 m calmodulin (Fig. 1). The elution was diluted in 15 ml of 10 mM Hepes, pH 7.0, 200 mM NaCl, 1 mM EGTA, and 1 mM 2-mercaptoethanol and loaded on to MonoS column (Amersham Biosciences) equilibrated in 10 mM Hepes, 50 mM NaCl, 1 mM EGTA, and 1 mM DTT. Myo1c-tail was separated and eluted with a linear 50 mM-1 M NaCl gradient. Myo1c-tail with bound calmodulin elutes at 0.5 M NaCl. Fractions containing myo1c-tail were combined and dialyzed versus 1 mM Tris, pH 8.0, 50 mM KCl, 1 mM EGTA, and 1 mM DTT and loaded on to MonoQ column (Amersham Biosciences) equilibrated in 10 mM Tris, pH 8.0, 25 mM KCl, and 1 mM DTT (Fig. 1). Myo1c-tail was eluted with a linear 25 mM-1 M KCl gradient (Fig. 1A). Myo1c-tail with bound calmodulin elutes at 0.43 M KCl. Fractions containing myo1c-tail were combined and concentrated to one-fifth volume using Microsep concentrators (Pall Filtron). The concentrated protein was dialyzed versus HNa100. Myo1c-tail concentrations were determined using the Coomassie Plus assay (Pierce). Quantitative scanning of protein gels labeled with SYPRO Red (Molecular Probes) indicates that approximately three calmodulins are bound per tail (not shown). A typical yield of myo1c-tail with bound calmodulin from 4 liters of cells is ϳ1 mg.
Detection of a Calcium-induced Change in Calmodulin Affinity-Myo1c-tail with bound calmodulin (1 nmol) was added to 50 l of nickel beads (Qiagen) and equilibrated in HNa100 with 1 mM ␤-mercaptoethanol instead of 1 mM DTT. Myo1c-tail-bound beads (0.4 M protein) were sedimented in HNa100 with 100 M 80% PS LUVs in the absence or presence of 250 M free calcium by spinning at 3000 ϫ g for 4 min. No free calmodulin was added to the buffer. The supernatant was removed, and the beads were suspended to original volume in HNa100. Proteins in the supernatant and pellets were resolved on a 15% SDS-polyacrylamide gel and stained with SYPRO Red. Protein bands were scanned and visualized with a PhosphorImager (Molecular Dynamics). The protein bands shown in Fig. 1B are from the same gel scan and are shown at identical exposure and identical gray-level scale.
Stopped-Flow Kinetics and Analysis-Transient kinetic measurements were made at 25 Ϯ 0.1°C with an Applied Photophysics (Surrey, UK) SX.18MV stopped-flow fluorometer. Tryptophan was excited with a monochromator set at 285 nm (2.3-nm slit width). A 440-nm long-pass filter (Oriel) was used to detect dansyl fluorescence. Experimental transients were fitted with software supplied with the instrument. All stated concentrations are those after mixing (i.e. the final reaction concentrations). The mixing ratio of the stopped-flow apparatus was 1:1, thus the concentrations in the stopped-flow syringes before mixing are two times those reported.
Association transients were obtained by mixing myo1c-tails with dLUVs. Two to eight association transients were averaged before nonlinear least-squares fitting. Dissociation transients were obtained by mixing pre-equilibrated myo1c-tail-dLUVs complexes with a large molar excess of unlabeled 80% PS LUVs. When mixed, myo1c-tails that dissociate from dLUVs have a much higher probability of rebinding to unlabeled LUVs resulting in decreased dLUVs fluorescence. Reported dissociation rates are the averages of four to eight transients from three different protein preparations.

RESULTS
Lipid Association-A fluorescence increase upon binding of myo1c-tail to dPS-containing LUVs (dLUVs) allowed us to monitor the rate of association of myo1c-tail with lipid. Fluorescence time courses of myo1c-tail binding to dLUVs in the absence ( Fig. 2A) and presence of 250 M free calcium (Fig. 2B) follow single exponential rates that depend linearly on the total lipid concentration (Fig. 2C). We did not detect lag phases, deviations from single exponential time courses, or fluorescence increases at longer time scales. However, a mixing artifact at high lipid concentrations (Ն100 M) limited our ability to analyze the first 50 ms of the transient (Fig. 2, A and B).
The apparent second-order rate constant for the binding of myo1c-tail to 40% PS dLUVs (0.16 Ϯ 0.02 M Ϫ1 s Ϫ1 ) was the same as 60% PS dLUVs (0.18 Ϯ 0.01 M Ϫ1 s Ϫ1 ). However, the apparent rate of association was significantly reduced for both 40% PS (0.020 Ϯ 0.01 M Ϫ1 s Ϫ1 ) and 60% PS (0.049 Ϯ 0.02 M Ϫ1 s Ϫ1 ) dLUVs in the presence of 250 M free calcium. Calcium alone does not affect the fluorescence of dLUVs.
Myo1c-Tail Lipid Dissociation-The rate of myo1c-tail dissociation from dLUVs was measured by mixing myo1c-tail-dLUVs with excess unlabeled 80% PS LUVs (see "Material and Methods"). After mixing, the starting fluorescence (F ϩtail ) decreased to the fluorescence level of dLUVs in the absence of tail (F Ϫtail ), indicating complete dissociation of tail from dLUVs (Fig. 3A). The signal-to-noise ratios of the dissociation curves ( Fig. 3A) were much better than the association curves ( Fig. 2) because the longer time scales allowed for longer signal integration times. Fluorescence time-courses of dissociation were best fit by a two-exponential rate function (Fig. 3A). The fast (k fast ) and slow (k slow ) rates were dependent on the mole fraction of total PS in the dLUVs at a single myo1c-tail concentration (Fig. 3B). The relative amplitude of the fast component (A fast ) decreased from ϳ70% of the total signal change at 20% PS to Ͻ40% of the total signal change at 80% PS.
The dissociation rate was examined as a function of myo1ctail concentration at a single 20% PS lipid concentration (Fig.  4A). At the lowest myo1c-tail concentrations tested (15-25 nM), fluorescence transients were adequately fit with a single-exponential rate function. At protein concentrations Ͼ25 nM, a two-exponential fit was required. The rate of k fast increased significantly between 25 and 100 nM myo1c-tail and did not increase at Ͼ100 nM myo1c-tail. The rate of k slow showed a small dependence on the myo1c-tail concentration (Fig. 4A).
The amplitude of the slow component (A slow ) reached its maximum level at ϳ100 nM myo1c-tail then slightly decreased, and A fast reached its maximum level at ϳ200 nM myo1c-tail. The total amplitude of the fluorescence change (A slow ϩ A fast ) reached its maximum at 200 nM myo1c-tail, indicating that the tail binding sites on the dLUVs were saturated at this concentration (Fig. 5).
Dissociation and the PS/Tail Ratio-To better understand the relationship between the dissociation rate and the PS content of LUVs, we measured dissociation as a function of myo1ctail concentration (15-375 nM) from 20%, 40%, and 60% PS dLUVs (Fig. 6). The rates of the fast and slow components are plotted as a function of moles of PS/moles of myo1c-tail (Fig. 6). The dependence of the rate on the PS/tail ratio is nearly identical for dLUVs of the different PS percentages. Therefore, the dependence of the dissociation rate of the fast component on the percentage PS at a single myo1c-tail concentration (Fig.  3B) is due to the variation in the PS/tail ratio, and it is not a function of the percentage of PS or PC in a vesicle.

Myo1c-Tail Lipid Dissociation in the Presence of Calcium-
The myo1c-tail dissociation rate from 20% PS dLUVs was examined as a function of myo1c-tail concentration in the presence of 250 M free calcium. As shown previously for myo1c and other myosin-I isoforms (reviewed in Ref. 1), calcium weakens the affinity of calmodulin for myo1c-tails. This is seen by the partial dissociation of calmodulin from myo1c-tails bound to beads in a pull-down experiment in the presence of 250 M calcium (Fig. 1B).
The dissociation transients in the presence of calcium were best fit to two-exponential rates at all myo1c-tail concentrations. k fast and k slow (Fig. 4C) and their associated amplitudes, A fast and A slow (Fig. 4D), showed a much smaller dependence on the myo1c-tail concentration than in the absence of calcium. The starting fluorescence levels of the dissociation transients in the presence of calcium (F ϩtail ) were the same as those obtained in the absence of calcium (Fig. 3A), suggesting equivalent binding of myo1c-tail to dLUVs. However, the total amplitude change (A fast ϩ A slow ) of the transient after mixing with unlabeled LUVs in the presence of calcium was Ͼ2-fold less than the expected signal change (F ϩtail -F Ϫtail ) at almost every myo1c-tail concentration tested (Fig. 5), suggesting that Ͼ60% of myo1c-tail does not dissociate from the dLUVs under experimental conditions. A further amplitude change was not observed when the acquisition time was increased to 1000 s. A further decrease in the fluorescence signal was also not observed when the plus-calcium dissociation reaction was mixed with 10 mM EGTA to chelate the free calcium, thus indicating that the effect of calcium on myo1c-tail dissociation is not reversible on the time scale of the experiments (not shown). Similar calcium effects on the rates and amplitudes were observed with 40% PS dLUVs (data not shown).

DISCUSSION
Association Rates-Association transients follow a single exponential at all concentrations tested. The rate constants for myo1c-tail binding to 40% PS and 60% PS LUVs in the absence of calcium were not statistically different, suggesting that the association rate is not dependent on the PS percentages that we tested. If we assume that the molecular weight of an LUV is ϳ1.2 ϫ 10 8 (20), then the apparent second-order rate constant for binding in terms of LUV concentration is ϳ2 ϫ 10 10 M Ϫ1 s Ϫ1 for 40% PS and 60% PS. This rate is on the same order as the collisional association rate (1-3 ϫ 10 10 M Ϫ1 s Ϫ1 ) as estimated by the Smoluchowski equation (20,21). Therefore, nearly every tail-vesicle interaction in the absence of calcium results in binding.
Dissociation Rates-In most cases, dissociation transients are not described by single exponential functions (Fig. 3A), so dissociation can not be modeled as a single-step process limited by one rate constant. We propose a simple model in which the dissociation rate depends on the number of PS molecules bound to each tail: the myo1c-tail is in equilibrium between two states that differ in the number of tail-bound PS molecules; and the relative populations of the two states in the absence of calcium are determined in part by the PS/tail ratio.
At high PS/tail ratios, myo1c-tails do not compete with each other for PS, and a single slowly dissociating population predominates. We can estimate the effective number of PS molecules bound to the slowly dissociating population by examining the myo1c-dependent changes in k fast and k slow . Fig. 6 shows that the rate of the fast component approaches the rate of the slow component at ϳ100 PS/tail. If 50% of the lipid is inaccessible due to its location within the inner leaflet of the lipid bilayer, then the effective number of PS molecules bound to each slowly dissociating tail is ϳ50.
At low PS/tail ratios, myo1c-tails compete with each other for PS, resulting in a fraction of tails with Ͻ50 bound PS molecules. Fewer electrostatic interactions between PS and tail result in an increased rate of dissociation. The fast component reaches its maximum rate at ϳ25 PS/tail (Fig. 6), suggesting that the minimum number of PS molecules bound to a tail is ϳ13. The rate of the slow component also increases at high myo1c-tail concentrations (Fig. 4A), which might be due to fewer PS molecules bound to the tail or to steric effects due to saturation of lipid with protein.
At the lowest PC/tail ratios in the absence of calcium, the slow component represents 20 -30% of the total dissociation amplitude for dLUVs containing 20% PS (Fig 4B), 40% PS (not shown), and 60% PS (not shown). Therefore, there are factors other than the PS/tail ratio that define the relative population of the fast and slow dissociating components (22). These may include cooperative association of PS with membrane-bound tails or stable conformational states of the protein that do not bind less than 50 PS molecules.
Calcium Regulation-The rate of lipid association is decreased ϳ3-fold in the presence of 250 M free calcium. The apparent second-order rate constants expressed in terms of dLUVs concentration (rather than total lipid) are 3 ϫ 10 9 M Ϫ1 s Ϫ1 for 40% PS dLUVs and 8 ϫ 10 9 for 60% PS dLUVs. Therefore, even in the presence of very high calcium concentrations, the rate of association of myo1c-tail with lipids is very fast. We do not know why the rate constant for lipid binding is decreased in the presence of calcium. However, possibilities include competition between calcium and myo1ctail for PS and alteration of the myo1c-tail conformation by calcium binding.
As in the absence of calcium, dissociation transients are best fit to a two-exponential rate function in the presence of 250 M free calcium. k fast and k slow do not show any dependence on the myo1c-tail concentration, and the total fluorescence amplitudes of the transients (A fast ϩ A slow ) are Ͼ2-fold smaller than in the absence of calcium (Fig. 5). The starting fluorescence levels (F ϩtail ) of the dissociation transients in the presence and absence of calcium are the same, so it is unlikely that the decreased amplitudes are due to fewer lipid-bound myo1c-tails. Rather, a large fraction of the myo1c-tails remain lipid-bound and do not dissociate on the time scale of our measurements. Qualitative lipid sedimentation experiments also suggest that calcium does not decrease myo1c-tail binding to LUVs (not shown). Experiments with myo1a (brush border myosin-I) suggest that calcium increases the number of myosin-I molecules bound to lipid vesicles (7). However, the similar fluorescence levels (F ϩtail ) in the presence and absence of calcium in the present study suggest that this is not the case for myo1c (Fig.  3). Highly stable protein-lipid interactions mediated by electrostatic interactions are not unique to myo1c. For example, protein kinase C has been shown to dissociate from PS-containing vesicles with a half-time Ͼ24 h (23).
It has been proposed for myo1a that calcium dissociates myosin-bound calmodulin unmasking positive charges that can bind acidic phospholipids (7). Because calcium weakens the affinity of calmodulin for myo1c and myo1c-tail ( Fig. 1B; see Ref. 1 for review), we suggest that myo1c-tail follows a similar mechanism, resulting in a highly stable protein-lipid interaction. Such a mechanism might explain the data of Cyr et al. (24), which shows that the IQ motifs play a role in directing subcellular localization of myo1c. However, it remains to be determined if a calmodulin-dissociated state of myo1c is physiological. It is also possible that calcium mediates an electrostatic interaction between a cluster of negative charges on the tail and PS as seen for other membrane-binding proteins (25). In either case, the calcium effect is not reversible under our experimental conditions and time scales, so it is possible that other regulatory elements play a role in mediating membrane attachment/detachment in the presence of calcium.
Physiological Relevance of the Dissociation Rates-The steady-state actin-activated ATPase rate of myo1c under calcium-free conditions similar to the present experiments is 0.5 s Ϫ1 (26), which is 10-fold faster than k slow (0.02-0.06 s Ϫ1 ). At high PS/tail ratios, most myo1c motors should be able to undergo several catalytic cycles before dissociating from the membrane. Therefore, an "anchoring" protein is not necessary to keep the motor associated with the membrane under these conditions. Remarkably, k slow is nearly identical to the fluorescence-recovery-after-photobleach rate of green fluorescent protein-myo1a and green fluorescent protein-myo1a-tail as measured in the brush border of kidney epithelial cells (5). Therefore, if membrane-associated tail-binding proteins are present in brush border microvilli, we predict that they do not increase the lifetime of membrane association.
The magnitude of k fast (0.2-1.0 s Ϫ1 ) is on the same order as the myo1c ATPase rate in the absence of calcium. So under conditions where the fast component predominates (i.e. low PS/tail ratios), the probability that myo1c will perform a power stroke while bound to a membrane is greatly decreased. The acidic phospholipid content in the membrane is regulated in both concentration and in spatial distribution (27). Therefore, the membrane dissociation rate of cellular myosin-I might be coupled directly to the regulation of lipid charge and acidic phospholipid density.
The motor domain of myosin-I interacts preferentially with tropomyosin-free actin filaments (4,28), so tropomyosin and spatially regulated actin polymerization play important roles in specifying the subcellular localization of myosin-I (4). In the absence of a specific myosin-I tail receptor, as in ruffling membranes (2, 4), we predict that myosin-I will associate with acidic phospholipids adjacent to tropomyosin-free actin. Assuming a high acidic lipid/tail ratio, these membrane-linked motors can perform multiple power strokes as they move toward the barbed end of the actin filament. Because myosin-I membrane interactions are relatively nonspecific, this model predicts that multiple myosin-I isoforms will concentrate on the same cellular membranes. Overlapping localization of vertebrate myosin-I isoforms has been shown (1, 2). However, it has also been shown that myosin-I isoforms have distinct non-overlapping localizations in certain subcellular regions (2), thus it is likely that myosin-I tails contain information for specifying localization to specific cellular regions (2,4). Further experiments are required to better understand the mechanisms and regulation of myosin-I targeting.