Actin filament dynamics in the actomyosin-VI complex is regulated allosterically by calcium-calmodulin light chain

Abstract

The contractile and enzymatic activities of myosin VI are regulated by calcium binding to associated calmodulin light chains. We have used transient phosphorescence anisotropy (TPA) to monitor the microsecond rotational dynamics of erythrosin iodoacetamide-labeled actin with strongly-bound myosin VI (MVI) and to evaluate the effect of MVI-bound calmodulin (CaM) light chain on actin filament dynamics. MVI binding lowers the amplitude but accelerates actin filament microsecond dynamics in a Ca²⁺ - and CaM-dependent manner, as indicated from an increase in the final anisotropy and a decrease in the correlation time of TPA decays. MVI with bound apo-CaM or Ca²⁺ - CaM weakly affects actin filament microsecond dynamics, relative to other myosins (e.g. muscle myosin II and myosin Va). CaM dissociation from bound MVI damps filament rotational dynamics (i.e. increases the torsional rigidity), such that the perturbation is comparable to that induced by other characterized myosins. Analysis of individual actin filament shape fluctuations imaged by fluorescence microscopy reveals a correlated effect on filament bending mechanics. These data support a model in which Ca²⁺ - dependent CaM binding to the IQ domain of MVI is linked to an allosteric reorganization of the actin-binding site(s), which alters the structural dynamics and the mechanical rigidity of actin filaments. Such modulation of filament dynamics may contribute to the Ca²⁺ – and CaM–dependent regulation of myosin VI motility and ATP utilization.

Introduction

Interaction of actin and myosin is required for force generation and contractility in muscle and non-muscle cells. A large body of studies on the molecular mechanism of motility approaches the problem by focusing primarily on one of the two crucial issues of this mechanism: the structure of the actomyosin interface and nucleotide-induced structural changes at the light chain-binding domain of myosin. However, since biochemical studies indicate functional interdependence (i.e. coupling) of the actin and light chain binding domains of several myosins, the two issues are linked and these interrelations must be elucidated for a complete molecular understanding of myosin-based motility.

The essential light chain isoform bound to skeletal muscle myosin strongly influences actin-activated ATPase¹, in vitro motility sliding velocity of actin filaments², and the ability of the S1 motor domain to accelerate actin polymerization³. Actin-activated ATPase of smooth muscle myosin is regulated by phosphorylation-induced structural transitions in the N-terminal region of bound light chains⁴. The activity of molluscan striated muscle myosin is regulated by Ca²⁺ binding to the regulatory light chain⁵.

Light chain-regulation of actin-myosin interactions requires long-range allosteric communication between the actin- and light chain- binding regions of myosin. Crystal structures of muscle and non-muscle myosins indicate that small movements within the myosin motor domain can be transmitted through the converter domain to the light chain binding region^{6; 7}. Functional interdependence between the motor properties of myosin and the light chain-binding domain is particularly clear in myosin VI (herein referred to as MVI) - the only known myosin family member that moves toward the pointed end of actin filaments rather than the barbed end⁷. MVI achieves reverse directionality by rotating its lever arm in the direction opposite to that of other myosins. Such rotation is probably enabled by the presence of an unique insert between the converter and the calmodulin (CaM) light chain binding (IQ) domain, which influences coupling between the light chain binding domain orientation and the nucleotide and actin binding sites^{7; 8}.

The insert and IQ domains of MVI each bind a single CaM molecule. It has been suggested that the insert-bound CaM principally plays a structural role and the IQ-bound CaM may be involved in Ca²⁺ -regulation MVI function⁹, namely slowing actin motility, actin-activated ATPase, and ADP release^{10; 11}. Early studies using gel densitometry showed that binding of CaM to MVI HMM dimers is not affected by Ca²⁺ up to ~100 µM¹⁰. Mass spectrometry confirms that CaM binding to the insert and to the IQ region is Ca²⁺ –independent and that CaM binds with a higher affinity (K_d ~ 30 nM) to the insert domain⁹. Furthermore, structural analysis by cryo-electron microscopy shows that Ca²⁺ weakens CaM binding to the IQ domain, but not to the insert¹². Since CaM binds the insert region of MVI independent of calcium with a high affinity⁹, while Ca²⁺ weakens the affinity of CaM for the IQ domain of MVI¹², it is possible that CaM binding and/or a CaM-linked conformation of the IQ domain modulates actin filament stiffening.

There is increasing evidence that actin filaments adopt multiple conformational states and that the equilibria among these states are modulated by interaction with regulatory proteins, including myosin contractile proteins^{13; 14}. Numerous studies have shown that changes in the structural state(s) of actin modulate filament sliding and actomyosin ATPase^{15; 16}, and that the effect of myosin on actin structure depends on the structural states of myosin^{17; 18; 19}. The impact of Ca²⁺-CaM regulation on the interaction between MVI and actin in terms of actin structural properties is less clear.

We initiated this study to determine whether Ca²⁺-CaM regulation of MVI involves allosteric modulation of the actin-binding regions that modulates the structural state fluctuations of actin. We measured the actin filament microsecond dynamics using transient phosphorescence anisotropy (TPA) and the actin filament flexural rigidity from images of thermally-driven filament shape conformations acquired using fluorescence microscopy. Previously, we demonstrated that the structural dynamics of actin is an important determinant of functional properties of the actomyosin complex^{17; 18; 20}. The present results with MVI reveal that the actomyosin interface and resulting changes in filament dynamics can be allosterically regulated by ligand-linked changes in the light chain-binding domain of myosin.

Results & Discussion

A. Effects of MVI on actin TPA decays

MVI binding affects the TPA decays of actin filaments in a Ca²⁺ - and CaM-dependent manner (Fig. 1). In the presence of free calcium (200 µM) and no added CaM, MVI substantially increases the final anisotropy, corresponding to a reduction in the amplitude of filament microsecond rotational dynamics (i.e. torsional stiffening; Fig. 1, red), as observed with other myosin isoforms¹⁹. However, the effects of bound MVI are substantially diminished in the absence of calcium (i.e. with bound apo-CaM; Fig. 1, blue) or upon addition of 30 µM Ca²⁺-CaM (Fig. 1, green). Equilibrium binding assays confirm MVI binds actin stoichiometrically under these conditions (Fig. 2), indicating that the observed effects reflect Ca²⁺ and CaM-dependent structural changes in actomyosin VI. Calcium and/or CaM alone have negligible effects on the TPA decays of bare actin (data not shown).

Figure 1. Effects of Ca²⁺ and CaM on the TPA decays of actomyosin VI.

Conditions: 200 µM Ca²⁺ (red); 200 µM Ca²⁺ and 30 µM CaM (green); 1 mM EGTA (blue). Bare actin (grey) is shown for comparison. Smooth lines represent fits to the sum of two exponential terms (Eq. 1). Both panels show the same data over different time scales.

Figure 2. Equilibrium binding of myosin VI and actin filaments.

Binding of myosin VI to actin in the presence (blue) and absence (red) of 200 µM Ca²⁺ measured from pyrene actin fluorescence quenching. The solid line is the best fit to a quadratic binding expression using the binding affinities³⁶ and total reactant concentrations^{37; 38}, yielding binding stoichiometries of 0.98±0.07 and 0.98±0.04 in the presence and absence of Ca²⁺, respectively.

B. Ca²⁺ and CaM-dependence of MVI effects on actin torsional dynamics

We evaluated the CaM-dependence of actin TPA decays, specifically the final anisotropy and the average correlation time (Eq. 1, Eq. 2, Fig. 3), with sub-stoichiometric MVI (myosin VI binding density = 0.25) to determine if torsional stiffening of actin filaments by MVI is regulated by CaM. The final anisotropy and the average correlation time of actin depend hyperbolically (Eq. 3) on the [CaM] when Ca²⁺ is saturating (^{9; 11}) (Fig. 3), yielding an apparent binding affinity (K_d) of 0.6 µM (final anisotropy) - 1.5 µM (average correlation time) for Ca²⁺ -CaM binding to the IQ domain of MVI. Ca²⁺ -[CaM]-dependent changes in the individual component anisotropy amplitudes (r₁ and r₂) and correlation times (ϕ₁ and ϕ₂) yield essentially identical results as the final anisotropy and average correlation time (Fig. 4; K_d = 0.5 – 1.2 µM). Thus, Ca²⁺ -CaM similarly affects both slow (r₁, ϕ₁) and fast (r₂, ϕ₂) components of the anisotropy decays. We conclude, in agreement with cryo electron microscopy studies¹², that Ca²⁺ lowers the CaM binding affinity of MVI and exogenous CaM is needed to saturate the IQ domain of MVI; binding to the insert IQ peptide is much tighter (K_d ~ 30 nM) and independent of calcium⁹.

Figure 3. Effects of exogenous CaM on TPA decays of actomyosin VI.

Conditions: 200 µM Ca²⁺ or 1 mM EGTA. The myosin binding density is 0.25. The solid lines through the data represent the best fits to rectangular hyperbolae (Eq. 3). Uncertainty bars represent one standard deviation from the mean.

Figure 4. Effects of CaM on the individual TPA decay component correlation times and anisotropy amplitudes.

Conditions: 200 µM Ca²⁺. The myosin binding density is 0.25. The solid lines through the data represent the best fits to rectangular hyperbolae (Eq. 3). Uncertainty bars represent one standard deviation from the mean.

Inclusion of exogenous CaM in the absence of calcium (1 mM EGTA present) has no detectable effect on the TPA decays (Fig. 3). The simplest explanation of this result is that MVI is saturated with CaM from the initial purification and this endogenous CaM binds MVI with high affinity (K_d <130 nM) in the absence of Ca²⁺. Thus, three distinct states that differentially affect actin rotational dynamics exist and their distribution is dictated by the CaM and Ca²⁺ concentrations.

C. Allostery of MVI effects on actin torsional dynamics in the presence and absence of calcium and CaM

Experiments thus far (Fig. 1 to Fig. 4) evaluate the effects of CaM at a single bound MVI concentration. We have previously shown that the perturbation of actin filament microsecond torsional dynamics by various myosin isoforms displays non-nearest neighbor (i.e. long-range) cooperative interactions, such that an individual bound myosin influences the dynamics of ~10 unoccupied filament subunits¹⁹. To determine whether and how the extent of filament saturation with MVI affects the regulatory effects of CaM, actin was titrated with MVI in the presence of calcium without and with CaM added at saturating (30 µM) concentration as well as in the absence of calcium, and the MVI-induced changes in the final anisotropy and average correlation time were fitted to a one-dimensional lattice model (Eq. 4, Fig. 5). In the presence of Ca²⁺ but absence of CaM, the increase in final anisotropy is highly cooperative and a single bound MIV affects the dynamics of 8.5±2.7 filament subunits, similar to that observed for myosin Va¹⁹. A comparable cooperative unit (N = 8.2±3.7) is observed in the reduction of the average correlation time.

Figure 5. Effects of myosin VI on the final anisotropy and average correlation time 〈 ϕ 〉 of actin filaments.

Conditions: 200 µM Ca²⁺ (red); 200 µM Ca²⁺ and 30 µM CaM (green); 1 mM EGTA (blue). Myosin binding densities were calculated from the equilibrium binding affinities and total concentrations. Molar ratios ranged from 0 to 1.2 myosin per actin. The solid lines represent the best fit of the data to the expression for binding to a linear one-dimensional lattice with cooperative non-nearest neighbor interactions (Eq. 4). Uncertainty bars represent one standard deviation from the mean.

Either addition of CaM in the presence of calcium (Ca²⁺ -CaM) or removal of calcium by EGTA (apo-CaM), two conditions favoring CaM-bound at the MVI IQ domain, has substantial effects on actin dynamics. Cooperative changes in final anisotropy for Ca²⁺ -CaM and apo-CaM are less pronounced (N = 4.6±0.1 and N = 6.1±1, respectively), with a modest reduction in the magnitude of change in final anisotropy. The effects on the correlation time are non-cooperative (Fig. 5). These data indicate that three MVI states with distinct actin filament interactions and cooperativity (Table 1) exist in a reversible Ca²⁺ – and CaM– linked equilibrium (Table 1). MVI with bound apo-CaM or Ca²⁺ - CaM weakly affects actin filament microsecond torsional dynamics compared to other myosins (e.g. muscle myosin II and myosin Va¹⁹). CaM dissociation from bound MVI stiffens filaments (Fig. 6), to an extent comparable to other myosins¹⁹. Ca²⁺ – and CaM–dependent modulation of acto-MIV filament dynamics may contribute to regulation of myosin VI motility and ATP utilization^{10; 11}.

Table 1.

Summary of actomyosin VI filament interactions and dynamics.

Conditions	CaM occupancy	Actin Cooperativity	Final anisotropy	Rotational dynamics
High Ca2+, Low CaM	IQ domain: unoccupied Insert: bound CaM	Induces cooperative (N ~8 subunits) actin structural changes	High	Cooperative acceleration
Ca2+-free (EGTA)	IQ domain: bound apo-CaM Insert: bound CaM	Weakly/Non- cooperative	Intermediate	Non-cooperative acceleration
High Ca2+, High CaM	IQ domain: bound Ca2+-CaM Insert: bound CaM	Weakly/Non- cooperative	Intermediate	Weak acceleration

Figure 6. Bending flexibility of actin filaments modulated by myosin VI.

The best fits of the average angular correlation of actin filaments to the 2D persistence length function (Eq. 7): bare (black) and fully myosin VI-decorated with 200 uM Ca²⁺ (red), 200 µM Ca²⁺ and 30 µM CaM (green) or 1 mM EGTA (blue) yield persistence lengths of 8.2±0.2, 21.4±0.7 and 12.8±0.2 µm, respectively.

D. Bending persistence length of actomyosin VI filaments modulated by Ca²⁺ and CaM

We determined the actin and filament bending persistence length, L_p, by fitting the 2D average cosine correlation function, 〈C(s)〉, to the average cosine of correlated tangential angles (θ) along segment lengths of actin filaments^{21; 22}. MVI increases the actin filament L_p from 8.8 ± 0.8 µm to 15 ± 1 µm (in the presence of 1 mM EGTA; Fig. 6). Adding 200 µM Ca²⁺ to actomyosin VI dissociates myosin-bound CaM and stiffens filaments, as indicated by a higher L_p of 23 ± 2 µm (Fig. 6). When additional CaM is supplemented to favor Ca²⁺ -CaM binding, actomyosin VI filaments have an L_p of 14 ± 1 µm (in the presence of 200 µM Ca²⁺ and 30 µM CaM; Fig. 6), which is comparable to actomyosin MVI in the presence of 1mM EGTA. These measurements indicate that myosin VI-, Ca2⁺ - and CaM-dependent changes in actin filament flexural rigidity correlate with effects in torsional dynamics assayed by TPA.

E. Origin of CaM-dependent actomyosin VI filament dynamics

CaM-regulated dynamics of actomyosin VI (Fig. 3; Fig. 5) does not result from CaM binding to actin (e.g. bound CaM interaction with actin’s C-terminus, as observed for the essential light chain of muscle S1²³) since the light chain binding domain of actin-bound MVI points away from the actin filament, while that of muscle myosin II and MV tilt towards the filament^{7; 8}. It is more likely that CaM-regulated dynamics of MVI-bound actin occurs via allosteric structural changes at the actin binding regions of MVI that originate from the light chain binding IQ domain. Reorganization of the actomyosin interface presumably compromises the energy and/or interface area between filament subunits, which influences the filament compliance²⁴.

We observe a remarkable agreement between MVI-induced changes in the microsecond torsional rigidity and the millisecond flexural rigidity of actin filaments (Fig. 7). Correlated, but opposite, changes in the twisting and bending motions of actin filaments are induced by cofilin - bound cofilin lowers the torsional²⁵ and flexural rigidity of actin filaments²². These observations are consistent with a coupling between actin filament twisting and bending motions²⁶. The non-nearest neighbor effects in filament dynamics and elasticity could influence the stepping behavior of dimeric myosin VI and interaction with filament regulatory proteins. Actin subdomains 1 and 2 are regions likely to be affected by MVI binding as it results in reorganization of the actomyosin interface²⁷. We expect MVI – induced changes in subdomain 1 to be subtle since the phosphorescence lifetime of ErIA-actin (156.8±5.4 µs) – a sensitive indicator of accessibility of the probe environment - is not significantly affected by MVI either in the absence of Ca²⁺/CaM (148.4±3.6 µs) or the presence of 30 µM CaM (137.1 µs). Alteration of the DNase loop within actin subdomain 2, has been implicated in determination of actin mechanical properties²⁸, as well as the dynamics and distribution of structural states within actin filaments, particularly upon interaction with a variety of actin-binding proteins^{29; 13; 25; 30; 22; 31}. The increased flexibility and dynamics of cofilin-decorated actin filaments were interpreted in terms of possible disruption of stabilizing intersubunit contacts, particularly contacts involving subdomain 2 induced by rearrangement of the DNase loop^{22; 31}. It is possible that MVI stiffens actin filaments by stabilizing the same DNase loop-mediated intersubunit contacts that modulate the dynamics and rigidity of actin filaments. MVI could potentially favor formation of a subset of the various actin filament subunit solution conformations^{13; 14}. Differences in the TPA of MVI as compared to other myosins (e.g. final anisotropy, correlation time, cooperativity¹⁹) may reflect different distributions of the various filament thermal conformers that are populated with bound myosin.

Figure 7. CaM-dependent microsecond dynamics and millisecond flexural rigidity of actomyosin VI filaments.

The amplitude of the microsecond dynamics calculated from the average anisotropy, r, from 400 to 500 µs (blue) and bending L_p from fits of the average angular correlation (red) of actin filaments upon the addition of saturating concentrations of myosin VI alone and with 200 uM Ca²⁺ and/or 30 µM CaM. Error bars indicate the SEM.

F. Implications for biological functions of MVI

The effects of Ca²⁺ -CaM on the dynamics of actomyosin VI could potentially have functional implications. MVI stiffens actin filaments and possibly more resistant to fragmentation, similar to bound MV¹⁹, dystrophin and utrophin³². MVI is present in many non-muscle cells, and Ca²⁺ associated modulation of actomyosin VI stiffness may be one of the structural determinants of the mechanical properties of actin bundles such as in hair cells in the inner ear³³. Studies on the dynamics of hair bundles indicate that the bundle stiffness increases with Ca²⁺ concentration, interpreted as Ca²⁺ - modulation of actin filament stiffness³⁴. A significant increase in the hair bundle stiffness is observed at ~200 µM Ca²⁺ (Ref.³⁴), which agrees with the increased actin filament stiffness upon the addition of MVI and 200 µM Ca²⁺ observed here. Thus, Ca²⁺ regulation of CaM binding to the IQ domain of MVI (i.e. by Ca²⁺ fluxes in the inner ear) likely enhance the “grip” of MVI on actin filaments and the mechanical stiffness of actin-based cytoskeletal structures.

Material and Methods

Protein preparations

Actin was prepared from rabbit skeletal muscle and labeled with erythrosin iodoacetamide (ErIA, AnaSpec)¹⁷ pyrene-iodoacetamide (Invitrogen) at Cys 374 or Alexa 488 succinimidyl ester (Molecular Probes, Eugene, OR) and gel-filtered^{22; 30}. A labeling efficiency with ErIA of 0.83±0.13 (mean±SD, n=11) was determined from the absorbance at 538 nm using extinction coefficient 83000³⁵ and ~0.8 for Alexa 488. Freshly prepared ErIA labeled actin was immediately stabilized against label-induced destabilization by adding a molar equivalent of phalloidin. Actin-bound Ca²⁺ was exchanged for Mg²⁺ with 200 µM EGTA and 80 µM MgCl₂ immediately before polymerization. Single-headed porcine myosin VI (T406A mutant; truncated at Gly840) with bound CaM light chain^{36; 37; 38} was purified from Sf9 cells by Flag affinity chromatography. Purity was >98% for all preparations. CaM was purified from E. coli using calcium-dependent hydrophobic interaction chromatography on phenyl sepharose 4B³⁹.

TPA experiments

Phalloidin-stabilized ErIA-F-actin was diluted in KMg50 buffer (50 mM KCl, 2 mM MgCl₂, 10 mM imidazole pH 7.0 with 0.2 mM CaCl₂ or 1 mM EGTA) to 0.5 µM, and 0.05 – 0.6 µM MVI was added to form the actomyosin VI complexes, as indicated in the text. To remove residual ATP and ADP from actin, all samples prior to measurement were incubated for 20 min with 0.5 units/ml apyrase. To prolong the excited-state lifetime and prevent photo-bleaching of the dye, oxygen was removed from the sample by 5 min incubation with glucose oxidase (55 µg/ml), catalase (36 µg/ml), and glucose (45 µg/ml)¹⁷.

TPA data analysis: model-independent fit to the sum of exponentials

Time-resolved phosphorescence anisotropy was measured at 25°C as described previously¹⁹, by recording 30 cycles of 1000 laser pulses (500 in each orientation of the polarizer). The initial anisotropy r₀, rotational correlation times ϕ₁ and ϕ₂ and corresponding amplitudes r₁ and r₂ were determined by fitting the anisotropy to the sum of two exponential terms and a constant r_∞ as described previously²⁰:

$$r(\mathrm{t})=r_1\text{exp}(-\mathrm{t}/{\mathrm{\varphi }}_1)+r_2\text{exp}(-\mathrm{t}/{\mathrm{\varphi }}_2)+r_{\infty }$$ Eq. 1

The time course of TPA decay was fitted in the 10 – 500 µs time range. The fit quality indicated residuals < 2% of the signal. Extending the fit to the full scale of decay (1 ms) increased the residuals due to lower signal to noise ratio at long times, but yielded r_∞ values within 5% of the calculated average r(t) in the 400–500 µs time range, since the decays are nearly completed within this time scale. The calculated average value of r(t) in the 400 – 500 µs time range has been shown previously to provide the most sensitive and precise measurement of actin’s microsecond rotational dynamics²⁰ and therefore was defined as the final anisotropy, determining the amplitude of microsecond time scale motions in actin. The component lifetimes and amplitudes of the two exponential fit were used to calculate a single weighted average correlation time (〈ϕ〉):

$$\langle \mathrm{\varphi }\rangle =({\mathrm{\varphi }}_1{r}_1+{\mathrm{\varphi }}_2{r}_2)/(r_1+r_2)$$ Eq. 2

The apparent affinity K_d of CaM to MVI was determined by fitting the effects of CaM on the final anisotropy and average correlation time (y) to a rectangular hyperbola with offset using Origin.8 program:

$$y=y_{\mathit{0}}+(y_{\mathit{\text{max}}}[\text{Ca}\mathrm{M}])/({\mathrm{K}}_{\mathrm{d}}+[\text{Ca}\mathrm{M}])$$ Eq. 3

where y₀ = y in the absence of CaM.

The effects of bound MVI on the observed final anisotropy (r_obs) of actin was analyzed in terms of the linear one-dimensional lattice model with non-nearest neighbor interactions^{20; 40}, in which binding to an individual actin filament subunit allosterically affects the dynamics of a filament segment containing N protomers according to:

$$y={\mathrm{y}}_{\mathit{\text{max}}}-(y_{\mathit{\text{max}}}-y_{\mathit{\text{action}}}){(1-\mathrm{v})}^N$$ Eq. 4

where y_actin and y_max are the limiting values of final anisotropy and average correlation time <ϕ> at 0 and infinite concentrations of MVI, respectively, and ν is the MVI binding density (i.e. bound [MVI]/[actin]). The unconstrained parameters in the least-squares fit were y_max and N.

The phosphorescence intensity (unpolarized) was calculated as I(t) = (I_v(t)+2*GI_h(t))/3, and fitted to

$$I(\mathrm{t})=a_1\text{exp}(-\mathrm{t}/{\mathrm{\tau }}_1)+a_2\text{exp}(-\mathrm{t}/{\mathrm{\tau }}_2)+a_3\text{exp}(-\mathrm{t}/{\mathrm{\tau }}_3)$$ Eq. 5

The amplitudes (a_i) and the triplet excited-state lifetimes (τⁱ) were used to calculate a single weighted average lifetime, 〈τ〉:

$$\langle \mathrm{\tau }\rangle =(a_1{\mathrm{\tau }}_1+a_2{\mathrm{\tau }}_2+a_3{\mathrm{\tau }}_3)/(a_1+a_2+a_3)$$ Eq. 6

Equilibrium binding measurements

Myosin VI binding to actin filaments was measured from the [myosin] dependence of pyrene actin fluorescence quenching³⁷. MVI was mixed with 500 nM phalloidin-stabilized pyrene actin and equilibrated at 25 °C (±0.1 °C) for 40–60 min in the presence of the indicated [CaM] and [Ca²⁺]. Steady-state fluorescence intensities were measured using a Photon Technologies International (New Brunswick, NJ) Alphascan fluorescence spectrometer.

Determination of filament flexural rigidity

Cation exchanged Alexa 488-labeled actin was polymerized in KMg50 buffer and equilibrated with and without saturating concentrations of myosin VI. Samples were diluted with KMg50 buffer supplemented with 15 mM dextrose, 100 mM DTT, 0.1 mg mL⁻¹ glucose oxidase, and 20 µg mL⁻¹ catalase and either 200 uM Ca²⁺, 200 µM Ca²⁺ and 30 µM CaM or 1 mM EGTA to a final actin concentration of 50 nM. Analysis of filaments undergoing thermal fluctuations and those adsorbed to poly-L-lysine-treated slides yielded comparable results²¹. Images of individual filaments were acquired using a Nikon Eclipse TE300 microscope equipped with a Coolsnap HQ cooled CCD camera (Roper Scientific, Tucson, AZ) and µManager, processed using imageJ (http://imagej.nih.gov/ij/) and analyzed with code written in Matlab (The Mathworks, Natick, MA) as described²². The bending persistence length (L_p) was determined by fitting the average of >100 angular (θ) cosine correlation measurements (C(s)) of a segment length, s, corrected for measurement variance, to the following two-dimensional correlation function:

$$\langle 𝒞(s)\rangle =\langle \text{cos}[\theta (s)-\theta (0)]\rangle ={\mathrm{e}}^{\frac{s}{2{\mathrm{L}}_{\mathrm{p}}}}$$ Eq. 7

Statistical analysis of data

Each result is reported as mean ± SEM, unless indicated otherwise.