Mechanochemical tuning of -I by the PNAS PLUS N-terminal region

Michael J. Greenberg, Tianming Lin, Henry Shuman, and E. Michael Ostap1

Pennsylvania Muscle Institute and the Department of Physiology, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA 19104

Edited by James A. Spudich, Stanford University School of Medicine, Stanford, CA, and approved May 13, 2015 (received for review April 3, 2015) are molecular motors that generate force to power a wide only do these myosins have different responses to force, but also array of motile cellular functions. Myosins have the inherent ability they have different biochemical transitions that are affected by to change their ATPase kinetics and force-generating properties force (i.e., the rate of ADP release for Myo1b and the rate of ATP- when they encounter mechanical loads; however, little is known induced actomyosin detachment for Myo1c). about the structural elements in myosin responsible for force The high-resolution crystal structures of nucleotide-free sensing. Recent structural and biophysical studies have shown that Myo1b (i.e., rigor-like state) (4) and ADP.vanadate-bound Myo1c myosin-I isoforms, Myosin-Ib (Myo1b) and Myosin-Ic (Myo1c), have (i.e., pre–power-stroke state) (8) have recently been determined. similar unloaded kinetics and sequences but substantially different Despite being in different conformational states, these structures responses to forces that resist their working strokes. Myo1b has the show that Myo1b and Myo1c have a high degree of structural properties of a tension-sensing anchor, slowing its -detachment homology to each other and to other myosins (4, 16). A prominent kinetics by two orders of magnitude with just 1 pN of resisting feature in the Myo1b structure is the positioning of the N-terminal force, whereas Myo1c has the properties of a slow transporter, region (NTR), which is in a conformation that has not been ob- generating power without slowing under 1-pN loads that would served in other myosin structures (Fig. 1A). In myosin-II, -V, and stall Myo1b. To examine the structural elements that lead to differ- -VI the NTR includes an SH3-like domain that lies to the side of ences in force sensing, we used single-molecule and ensemble ki- the motor domain (17–19). In contrast, the Myo1b NTR sits in a netic techniques to show that the myosin-I N-terminal region (NTR) hydrophobic pocket between the motor and the lever arm helix plays a critical role in tuning myosin-I mechanochemistry. We found (LAH) and interacts with the first calmodulin light chain. As such, that replacing the Myo1c NTR with the Myo1b NTR changes the it is in a position that might enable it to communicate the position identity of the primary force-sensitive transition of Myo1c, resulting of the LAH to the nucleotide-binding site (4). Although the motor in sensitivity to forces of <2 pN. Additionally, we found that the – domain sequences of myosin-I isoforms are highly conserved, in- NTR plays an important role in stabilizing the post power-stroke cluding the residues that create the hydrophobic pocket in which conformation. These results identify the NTR as an important struc- the NTR sits, the NTR sequences are not conserved (Fig. 1B). We tural element in myosin force sensing and suggest a mechanism for hypothesized that the NTR plays a role in establishing kinetic generating diversity of function among myosin isoforms. diversity among myosins (4). To investigate whether the NTR of Myo1c plays a role in mechanochemistry | optical tweezers | mechanosensing | mechanosensing, we expressed Myo1c constructs with the native transient kinetics | single molecule NTR present, with the NTR deleted, and with the Myo1b NTR in place of the native sequence and measured their kinetic and me- yosin motors use the energy from ATP hydrolysis to power chanical properties in the presence and absence of force. We find Ma wide array of cellular processes including muscle con- that the NTR of Myo1c plays important roles in tuning Myo1c’s traction, cell migration, membrane trafficking, cell division, and kinetics in the presence and absence of load and in stabilizing the intracellular transport (for review, see ref. 1). To optimally

function in such diverse processes, different myosin isoforms BIOPHYSICS AND Significance

have evolved distinct kinetic and mechanical properties to meet COMPUTATIONAL BIOLOGY their physiological demands (for review, see ref. 2), including differing abilities to adapt their ATPase kinetics and power Myosin molecular motors generate forces in the cell and act as outputs in response to mechanical loads (3). Although many mechanosensors, adjusting their power outputs in response to studies have elucidated structural elements important for force mechanical loads. Little is known about the structural elements generation in myosin motors, not much is known about the re- involved in myosin mechanosensing. Our results identify the gions important for tuning a myosin’s ability to modulate power N-terminal region (NTR) of the myosin-I as having an output in response to mechanical loads. important role in tuning mechanochemistry. Appending the An opportunity to study these structural elements has emerged NTR from a highly tension-sensitive myosin (Myo1b) onto a less with the structural and mechanochemical characterization of two tension-sensitive motor (Myo1c) changes the identity of the myosin-I family members with similar sequences, Myosin-Ib primary force-sensitive transition of Myo1c, making it sensitive < (Myo1b) (4–7) and Myosin-Ic (Myo1c) (8–10). These motors have to forces 2 pN. Moreover, we show that the NTR stabilizes the – similar unloaded ATPase kinetics, where they are both low duty- post power-stroke conformation. These results identify the ratio motors (i.e., they spend a majority of their biochemical cycles NTR as an important structural element in myosin force sensing detached from actin) with motility rates that are limited by the rate and suggest a mechanism for generating diversity of function among myosin isoforms. of ADP release (7, 10–14). Although these motors have similar sequences and unloaded ATPase kinetics, they have very different Author contributions: M.J.G., H.S., and E.M.O. designed research; M.J.G. and T.L. per- mechanical outputs under load (for review, see ref. 15). Myo1b is formed research; M.J.G., T.L., and H.S. contributed new reagents/analytic tools; M.J.G., extraordinarily sensitive to small loads, where forces of 1 pN slow T.L., H.S., and E.M.O. analyzed data; and M.J.G., H.S., and E.M.O. wrote the paper. its motility rate more than 50-fold. As such, Myo1b has the The authors declare no conflict of interest. expected properties of a tension-sensitive anchor (5, 6). In contrast, This article is a PNAS Direct Submission. 1 pN of force does not appreciably slow the rate of Myo1c motility, 1To whom correspondence should be addressed. Email: [email protected]. enabling the motor to generate power over a range of forces. Thus, This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. Myo1c has the expected properties of a slow transporter (10). Not 1073/pnas.1506633112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1506633112 PNAS | Published online June 8, 2015 | E3337–E3344 Downloaded by guest on September 25, 2021 Scheme 1), where the rate of ADP release limits the Myo1c A motility rate at saturating [ATP] (10). We used stopped-flow transient kinetic techniques to determine whether these steps were altered by deletion or swapping of the NTR (Table 1; see Materials and Methods for details). The rate constant for ATP- induced dissociation was measured by mixing pyrene-labeled actomyosin with varying ATP concentrations (Fig. 2). Similar to Δ Myo1c (10), fluorescence transients obtained with Myo1c N Δ > and Myo1c N- b are best fitted by the sum of two exponential functions. The fast component has a hyperbolic dependence on the [ATP] (Fig. 2A) and was modeled as shown in Scheme 1 (20). Δ Δ > Both Myo1c N and Myo1c N- b have substantially accelerated L11 −1 maximal rates of ATP binding (k+2′ = 160 ± 4.6 s and 160 ± L10 − Δ Δ > 2.8 s 1 for Myo1c N and Myo1c N- b, respectively) compared with −1 Myo1c (k+2′ = 18 ± 0.99 s ; P < 0.001). The rate of the slow M14 ΔN −1 phase measured for both Myo1c (kslow = 56 ± 3.2 s )and ΔN->b −1 Myo1c (kslow = 54 ± 2.0 s )is>10-foldfasterthanthat −1 found for Myo1c (kslow = 4.0 ± 0.034 s ; P < 0.001) (Fig. 2B). In Myo1c, the slow phase is due to an isomerization of actin-bound myosin from a nucleotide-free state that is not capable of binding ATP to a state than can bind ATP (12, 21); however, it does not B 10 20 30 appear that the slow phase observed for the NTR mutants cor- Myo1b MAKKEVKSSLLDNMIGVGDTVLLEPLN-EETF ID relates with the same transition (see below). Myo1c ---MESALTARD-RVGVQDFVLLENFTSEAAFIE The rate of ADP release was measured by preincubating NTR 10 20 30 pyrene-labeled actomyosin with saturating [ADP] and then ob- C serving the increase in fluorescence upon ATP-induced acto- 110 767 myosin dissociation (Fig. 2D). The rate of ADP release from k ′ = ± −1 MESALTARDRVGV Myo1c Motor IQ IQ IQ AviTag Myo1c ( +5 3.9 0.060 s ) was reported previously (10). For Myo1c ΔN ΔN Myo1c , the fluorescence transient is best described by the Myo1c MVGV Myo1c Motor IQ IQ IQ AviTag sum of two exponential functions where the slower phase (0.93 ± −1 ΔN->b Myo1c Motor IQ IQ IQ AviTag 0.010 s ) makes up 79% of the amplitude and the faster phase Myo1c MAKKEVKSSLLDNMVGV − Δ > (4.2 ± 0.18 s 1) makes up 21% of the amplitude. For Myo1c N- b,the 114 fluorescence transient is best fitted by a single exponential k ′ = ± −1 Fig. 1. (A, Inset) The structure of apo-Myo1b (PDB ID code 4L79) showing function ( +5 0.68 0.033 s ). Removal of the Myo1c NTR the NTR (green) sandwiched between the motor domain, the LAH, and the causes a twofold weakening in the ADP affinity (K5 = 0.42 ± bound calmodulin (red) (4). The myosin is colored gray and residues that 0.015 μM; P < 0.001; Fig. 2E) compared with native Myo1c (K5 = create the hydrophobic pocket are colored in yellow. (A) Close-up of the 0.22 ± 0.050 μM), which is reversed when the Myo1b NTR is hydrophobic pocket showing the NTR residues (L10, L11, and M14) that appended (K5 = 0.16 ± 0.062 μM; P = 0.1). reach into the pocket. (B) Alignment of the initial sequences of Myo1b and The rates of ADP release (k+5′) and the slow phase of ATP- Myo1c. The green box encloses the regions we define as the NTR. Conserved k residues are highlighted in blue. (C) Sequence maps of myosin constructs. induced actomyosin dissociation ( +α; Scheme 1) are similar to each other in Myo1c (10, 12, 14) and other myosins (21, 22), leading to the suggestion that these rates are reporting a transition between post–power-stroke conformation. Importantly, appending the Myo1b similar structural states (21, 22). Interestingly, these rates are not ΔN ΔN->b NTRtoMyo1ccausesthemyosintobehavemorelikeMyo1b; correlatedineitherMyo1c or Myo1c , suggesting that the slow specifically, it causes the rate of actoMyo1c detachment to become phase of ATP-induced dissociation and ADP release are distinct sensitive to forces <1 pN and it changes the identity of the primary transitions in the ATPase pathway and that changes to the NTR force-sensitive transition that limits actin detachment. These results cause uncoupling of these transitions. clearly identify the NTR as an important structural element that Unloaded Actin Filament Gliding Assays. In vitro motility assays in helps define myosin-I’s force-sensing and power-generating prop- which fluorescently labeled actin filaments are propelled by surface- erties. Moreover, these results suggest that mechnochemical tuning bound myosin molecules were conducted at 37 °C (Fig. 3A). For by the NTR is an underappreciated mechanism for generating Myo1c, removal of the NTR causes a reduction in the sliding ve- functional diversity within the myosin family of motors. locity from 83 ± 5.9 nm/s to 60 ± 4.6 nm/s (P < 0.001). Appending Results the Myo1b NTR causes a further reduction in sliding velocity to 32 ± 4.9 nm/s (P < 0.001). These results are consistent with the rate We probed the role of the NTR in tuning Myo1c mechanochem- istry by examining the biochemical, mechanical, and motile prop- erties of three truncated, recombinant Myo1c constructs (Fig. 1C): a protein with the native NTR sequence (Myo1c), a construct AM ΔN lacking the NTR (Myo1c ), and a Myo1c with the Myo1b NTR k-α k+α ΔN->b k ’ k ’k’ (Myo1c ). All constructs included the motor domain, the LAH K1’ +2 +4 +5 AM’ AM(ATP) AM.ATP AM.ADP.Pi AM.ADP + Pi AM’ consisting of the three calmodulin-bound IQ motifs, a FLAG-tag k-5’ for purification, and an AviTag for site-specific biotinylation. K8 K9 k+3 Biochemical Analysis of actoMyo1c Detachment Kinetics. The actin A + M.ATP A+M.ADP.Pi k-3 detachment rate of cycling Myo1c is kinetically limited by the sequential steps of ADP release (k+5′) and ATP binding (k+2′; Scheme 1.

E3338 | www.pnas.org/cgi/doi/10.1073/pnas.1506633112 Greenberg et al. Downloaded by guest on September 25, 2021 − − PNAS PLUS Table 1. Rate and equilibrium constants for key steps of the rates of the fast (2.7 ± 0.12 s 1) and slow (0.73 ± 0.026 s 1) phases actomyo1c ATPase are within twofold of the rates of the fast and slow phases of − − Δ Δ > 1 1 Parameter Myo1c Myo1c N Myo1c N- b ADP release measured in the stopped flow (4.2 s and 0.93 s , respectively). The relative amplitudes of the fast (37%) and slow ATP binding phases (63%) for actomyosin detachment are also within 20% of ′ μ ± † ± ± 1/K1 , M12031 450 48 320 24 the relative amplitudes of the fast (21%) and slow (79%) phases ′ −1 ± † ± ± ΔN->b k+2 ,s 18 0.99 160 4.6 160 2.8 of ADP release measured in the stopped flow. Myo1c ′ ′ μ −1· −1 ± † ± ± K1 k+2 , M s 0.15 0.038 0.35 0.039 0.49 0.038 attachment durations are best fitted by the sum of two expo- −1 † k+α or k+slow,s 4.0 ± 0.034 56 ± 3.2 54 ± 2.0 − † nential functions, where the predominant phase (72% of the am- 1 ± ± ± −1 k−α or k−slow,s *121.2 32 4.7 58 13 plitude) has a rate (0.35 ± 0.049 s ) that is approximately twofold ± † ± ± Afast:Aslow 0.33 0.034 1.8 0.25 0.94 0.21 slower than expected from ADP release measurements. The slower Δ > ADP release rate may be due to the force sensitivity of actoMyo1c N- b disso- ′ −1 ± † ± ± − k+5 (fast), s 3.9 0.06 4.2 0.18 (21%) 0.68 0.033 ciation (see below). The fast phase (9.8 ± 0.29 s 1) does not cor- ′ −1 ± k+5 (slow), s 0.93 0.010 (79%) relate with any measured biochemical transition and may represent ′ μ ± † ± ± K5 , M 0.22 0.05 0.42 0.015 0.16 0.062 detachment of actomyosin through a noncanonical pathway. KMg25 buffer: 60 mM Mops (pH 7.0), 25 mM KCl, 1 mM EGTA, 1 mM DTT,

1 mM MgCl2,20°C. Mechanics of the Myo1c Working Stroke. Myo1c was shown to have *Calculated. a two-substep working stroke, with an initial displacement (5.8 ± † Values are from ref. 10. 0.03 nm, state 1) that correlates with the transition to strong binding and phosphate release followed by a second displacement (2.0 ± 0.03 nm, state 2) that correlates with the rate of ADP release of ADP release limiting Myo1c sliding velocity, because the rate of (Fig. 3D) (10). To examine whether deleting or changing the > ΔN > ΔN->b D ADP release for Myo1c Myo1c Myo1c (Fig. 2 ). NTR affects the size and/or kinetics of working-stroke sub- Δ Δ > steps, we ensemble averaged Myo1c N and Myo1c N- b unitary Measurement of actoMyo1c Attachment Durations Using Optical Δ displacements acquired using the optical trap (25, 26). Averages were Trapping Techniques. Mechanical interactions of Myo1c N and ΔN->b obtained by aligning interactions upon their initial actin attachment Myo1c with actin were measured using an assay in which (time forward) and upon actin detachment (time reverse; Fig. 3D). Δ Δ > single actin filaments were suspended between two optically Myo1c N and Myo1c N- b show considerable force fluctua- trapped beads and lowered onto pedestal beads sparsely coated tions during actomyosin interactions, unlike Myo1c (Fig. 3C). with myosin (23, 24). Mechanical interactions between actin and Strikingly, neither time-forward nor time-reversed ensemble av- Δ Δ > myosin were identified, cumulative distributions of attachment erages of the Myo1c N and Myo1c N- b working strokes show durations were constructed, and single- or double-exponential exponential rises in force associated with mechanical substeps; functions were fitted to the data as justified by statistical testing rather, the force appears to fluctuate about a mean value (Fig. (Table 2 and Fig. 3 B and C)(Materials and Methods). 3D). Therefore, a clearly resolved second mechanical substep is Δ The distribution of Myo1c attachment durations was previously not observed in the ensemble averages of either Myo1c N or Δ > reported to be best fitted by a single-exponential function with a Myo1c N- b, likely due to the absence of a second substep, the rate that is consistent with the rate of ADP release measured bio- introduction of compliance when the native Myo1c NTR is al- Δ chemically (Fig. 3B) (10). The distribution of Myo1c N attachment tered, and/or fluctuations between conformational states that are durations is best fitted by the sum of two exponential functions. The not tightly linked to biochemical states (27).

A 150 B 50 C 2.5 BIOPHYSICS AND ) ) 40 2.0 -1 -1 slow 100 COMPUTATIONAL BIOLOGY Myo1c 30 1.5 / A Myo1cΔN 20 fast 1.0 50 Δ N->b A Rate (s Myo1c Rate (s 10 0.5 0 0 012345 012345 012345 D [ATP] (mM) E [ATP] (mM) [ATP] (mM) 1.0 1.0 Myo1c Δ 0.8 0.8 Myo1c N 0.6 0.6 0.4 0.4 Myo1cΔN->b 0.2 0.2 Fluorescence (A.U.) 0 Fraction Slow Phase 0 01234 02 4 68 10 Time (s) [ADP] (μM)

Fig. 2. Stopped-flow transient kinetic measurements. (A–C) The rate of ATP-induced actomyosin dissociation was measured by rapidly mixing pyrene-acto- myosin with varying [ATP] and then observing the increase in pyrene fluorescence as myosin dissociated from the actin. Data were best fitted by the sum of two exponential functions. Each point is the average of at least five measurements. (A) The rate of the fast phase of the fluorescence transient as a function of [ATP].

A hyperbola was fitted to the data to yield the equilibrium constant for ATP binding (K1′) and the maximal rate of ATP-induced dissociation (k+2′). Values for these rates can be found in Table 1. The maximal rate of ATP binding is faster in the deletion and swap constructs. (B) The rate of the slow phase of the

fluorescence transient as a function of [ATP]. A hyperbola was fitted to the data to yield k+α or k+slow.(C) The ratio of the fast (Afast) to slow (Aslow) amplitudes of the fluorescence transients as a function of [ATP]. (D) Fluorescence transients showing rate of ADP release from actomyosin. A total of 10 μM Mg.ADP equilibrated with 0.5 μM pyrene-actomyosin was rapidly mixed with 5 mM Mg.ATP (all concentrations after mixing) and the increase in fluorescence was measured as actomyosin dissociated. Note that apparent changes in the variance are due to changes in the sampling rate during acquisition to look for fast phases in the transient. (E) The affinity of actomyosin for ADP. Actomyosin was preincubated with ADP and the fractional amplitude of the slow phase of ATP- induced actomyosin dissociation was measured as a function of [ADP]. A quadratic function was fitted to the data to obtain the ADP affinity (Table 1).

Greenberg et al. PNAS | Published online June 8, 2015 | E3339 Downloaded by guest on September 25, 2021 A B 120 1.0 100 0.8 80 0.6 60 Myo1c ΔN 40 0.4 Myo1c 0.2 ΔN->b Speed (nm/s) 20 Myo1c

0 Δ Δ Cumulative Fraction 0 Myo1c Myo1c N Myo1c N->b 042816 0 C Time (s) Myo1c

Myo1cΔN * * ** * *

Δ Myo1c N->b 25 nm ***** 5 s D

Myo1c Myo1cΔN ΔN->b State 2 Myo1c 4 nm State 1

1 s 0.2 s

Fig. 3. (A) The rate of actin gliding by myosin was measured using the in vitro motility assay. Error bars show the SD (n = 50–75 filaments). (B) Optical trapping techniques were used to measure single-molecule interactions between actin and myosin at 50 μM ATP. Cumulative distributions of attachment durations were generated and were fitted by either single- or double-exponential functions as justified by an F-test. Distributions were normalized to account for the dead time. The rates obtained from the fitting are reported in Table 2. (C) Sample data traces showing interactions between actin and myosin in the absence of positional feedback. Individual binding interactions are denoted by black lines. Boxes show expanded interactions with blue asterisks marking force reversals. (D) Ensemble averages of the myosin working stroke were constructed by averaging individual binding interactions as previously described (5). The curvy black lines mark a break in the time of the time-reverse averages, 1 s before actomyosin detachment. Values obtained from fitting of the averages are reported in Table 2. The cartoon shows a model for a generic myosin with a two-substep power stroke. The swap and deletion constructs do not have an observable second substep in their working strokes.

Force Sensitivity of Actomyosin Detachment. To examine whether attachment durations over a range of loads in each experiment. the NTR plays a role in tuning the loaded mechanochemistry of Fig. 4A shows representative data traces of actomyosin in- Myo1c, a feedback system was used to maintain the actin near its teractions collected using the isometric optical clamp. In the isometric position while myosin undergoes its working stroke (23, presence of forces that resist the power stroke, Myo1c binding 28). The isometric optical clamp allows measurement of actin- interactions are generally short lived and the force remains

Table 2. Mechanical and kinetic parameters determined via optical trapping techniques Parameter Myo1c, n = 369 Myo1cΔN, n = 550 Myo1cΔN->b, n = 230

Actomyosin detachment rate in the absence of force −1 kfast,s 3.5 ± 0.012* 2.7 ± 0.12 (37%) 9.8 ± 0.29 (28%) −1 kslow,s NA 0.73 ± 0.026 (63%) 0.35 ± 0.049 (72%)

Δ Δ > Myo1c, n = 369 Myo1c N, n = 550 Myo1c N- b, n = 230 Working stroke displacements Total, nm 7.8 ± 0.05* 5.9 ± 0.5 4.1 ± 0.3 Substep 1, nm 5.8 ± 0.03* 5.9 ± 0.5 4.1 ± 0.3 Substep 2, nm 2.0 ± 0.03* NA NA

Δ Δ > Myo1c, n = 670 Myo1c N, n = 153 Myo1c N- b, n = 316 Force sensitivity −1 kf,s 29 (+9/−6)* NA 1.2 (+0.42/−0.27)

ddet, nm 5.2 (+0.5/−3.6)* NA 5.3 (+1.3/−0.96) −1 ki,s 5.6 (+1.6/−0.8)* 0.51 (+0.24/−0.15) 0.059 (+0.027/−0.020)

NA, not applicable. *Values are from ref. 10.

E3340 | www.pnas.org/cgi/doi/10.1073/pnas.1506633112 Greenberg et al. Downloaded by guest on September 25, 2021 A is best modeled as two sequential transitions, one force in- PNAS PLUS dependent and one force dependent (model 3; Fig. 4B and Table Myo1c 2). The rate of the force-independent transition is consistent with Δ k ′ Myo1c N the rate of ADP release ( +5 ) measured via solution kinetics, whereas the rate of the force-sensitive transition is consistent Δ Myo1c N->b 4 pN with the transition that limits ATP-induced dissociation at sat- urating [ATP] (k+2′; Scheme 1) (10). In contrast, the rate of 10 s Δ B actoMyo1c N detachment as a function of force is best described 100 ) by a kinetic model in which there is a single, force-independent -1 transition that limits detachment (model 1; Fig. 4B and Table 2). Δ > Myo1c The distribution of actoMyo1c N- b attachment durations as a Δ Myo1c N function of force is best fitted by a model in which there are two 10 Δ Myo1c N->b parallel pathways for detachment, one force dependent and one force independent (model 4; Scheme 2). The pathway that < 1 dominates dissociation at forces 2 pN is force sensitive, and the transition that dominates at higher loads is force insensitive. The −1 rate of the force-sensitive transition [kf = 1.2 (+0.42/−0.28) s ] 0.1 is within twofold of the biochemical rate of ADP release with an effective distance to the transition state (ddet) of 5.3 (+1.3/−0.96) nm. The rate of the force-insensitive transition (ki) is 0.059 + − −1 Mean Detachment Rate (s ( 0.027/ 0.020) s , which likely represents the dissociation of 0.01 -1 0123 45 myosin.ADP from actin (5). Force (pN) Traces obtained from single myosins contained interactions with and without force reversals. We analyzed the interactions without Fig. 4. Optical trapping experiments using the isometric optical clamp to force reversals independently of interactions with force reversals to apply a load to the myosin. (A) Representative data traces showing in- determine whether dwelling in force-reversed states is responsible teractions between actin and myosin. Individual interactions identified by for the force dependence of actin detachment (Fig. 5A). MLE analysis of the force covariance of the optically trapped beads are marked by Δ > fitting of model 4 in Scheme 2 to the data shows that the dis- black bars. The binding interactions for Myo1c N- b are longer lived than the tance to the transition state (ddet) for interactions lacking [4.7 interactions for Myo1c and frequent force reversals are observed during single (+1.1/−0.98) nm] and possessing [4.5 (+2.2/−1.1) nm] force re- attachment events. (B) Mean detachment rate of actomyosin measured in the versals is within 15% of the distance to the transition state for all presence of force. A positive force is defined as a force that resists the power + − stroke. Each point is the average of 20 binned binding interactions. Different of the binding interactions [5.3 ( 1.3/ 0.96) nm]. Therefore, the kinetic models were fitted to the unbinned data using maximum-likelihood force reversals are not responsible for the observed force de- estimation (MLE), and the best model was selected based on statistical testing pendence of actomyosin detachment. Moreover, the rates of as described in Materials and Methods. The thick line shows the best fit de- detachment through the force-independent pathway (ki)bothin − termined by MLE fitting and the thinner lines show the 95% confidence in- the absence [0.076 (+0.072/−0.056) s 1] and in the presence [0.040 − tervals. The force-sensing behaviors of the various constructs are different. (+0.028/−0.020) s 1] of force reversals are within twofold of each The detachment rate of actoMyo1c is independent of forces <1 pN, whereas Δ > other. The primary difference between interactions containing thedetachmentrateofactoMyo1cN- b is sensitive to these forces. The and lacking force reversals is the rate of detachment through transition that limits detachment at 0 pN force (i.e., the rate of ADP release) is − Δ Δ > 1 N N- b the force-sensitive pathway [kf = 2.0 (+0.82/−0.52) s and 0.65 force insensitive for Myo1c and Myo1c and force sensitive for Myo1c . − (+0.46/−0.17) s 1, respectively]. These data show that appending the Myo1b NTR to Myo1c introduces force sensitivity to the ADP relatively constant throughout the binding interactions. Attach- release transition, a behavior that is observed in Myo1b (5, 6), but BIOPHYSICS AND ΔN ment durations of actoMyo1c interactions are slightly longer not in Myo1c (10). COMPUTATIONAL BIOLOGY and appear to be force independent. However, in some binding Δ > interactions, the force dramatically fluctuates during the in- Myo1c N- b Force Reversals. The number of force reversals per teraction in a behavior we termed force reversals. Force reversals actomyosin binding interaction was calculated as a function of Δ are not due to dissociation and rebinding of actoMyo1c N, be- the average force during the interaction and then normalized cause the force covariance of the optically trapped beads remains based on the average length of a binding interaction at that force low during the force reversals. The force reversals do not appear (Fig. 5B). Force does not change the time-normalized probability to represent transitions between states 1 and 2 revealed in the Myo1c ensemble averages (Fig. 3D). Rather, we propose that they represent a destabilization of the post–power-stroke state, evidenced by the fact that the force exerted by the myosin will Model Detachment rate occasionally drop to 0 pN (i.e., the baseline when actomyosin is ki = kk 1. Attached Detached det i dissociated) during a force reversal (Fig. 4A). The durations of Δ > N- b • dF actoMyo1c attachments appear to increase with force, k (F) kFk *)( exp(−= det ) f det f *Tk suggesting that the actomyosin detachment rate is more force 2. Attached Detached B sensitive than the other two constructs. The force-reversal be- Δ > N- b 1 1 1 havior is also more pronounced in Myo1c interactions. ki kf (F) = + Attached (1) Attached (2) Detached Fk )( • dF k Actomyosin attachment durations were measured as a func- 3. det k *exp(− det ) i f *Tk tion of the average force on the myosin during the binding in- B teraction, and maximum-likelihood estimation (MLE) was used kf (F) 4. Attached Detached • dF kFk *)( exp(−= det ) + k to fit four different kinetic models to the data (Scheme 2). The ki det f *Tk i model that best fitted the data was selected based on statistical B Detached testing (Materials and Methods and Table S1). As demonstrated previously (10), the force dependence of actoMyo1c detachment Scheme 2.

Greenberg et al. PNAS | Published online June 8, 2015 | E3341 Downloaded by guest on September 25, 2021 )

A -1 B 10 6 Myo1c Δ 5 Myo1c N ΔN->b 1 4 Myo1c 3 0.1 2 No Force Reversals 1 Force Reversals 0.01 0

-1 01234 Force reversals per event / -1 01 234 Mean Detachment Rate (s Force (pN) mean attachment duration (s) Force (pN) ) C D -1 10

1

k+r kf(F) Low force state High force state Detached 0.1 ki k-r ki Experimental Data Theoretical Curve 0.01 Detached Detached -1 01 234 Mean Detachment Rate (s Force (pN)

Fig. 5. Analysis of Myo1cΔN->b force reversals. (A) The rate of actoMyo1cΔN->b detachment as a function of force determined by MLE fitting for events both with (pink) and without (purple) force reversals. Each point is the average of 15 binned binding interactions. Interactions both with and without reversals are similarly sensitive to forces <1pN.(B) The number of force reversals per actomyosin binding interaction (including interactions without force reversals) was measured, binned based on the force on the myosin immediately before the force reversal, and then normalized based on the mean attachment duration at that force. (C) Simple kinetic model used for calculating the theoretical detachment rate as a function of force (Materials and Methods). Actin-attached myosin transits Δ > between a high-force state and a low-force (i.e., force-reversed) state. (D) The theoretical detachment rate for actoMyo1c N- b as a function of force based on the kinetic scheme in C is shown in black. The green triangles are experimental data showing the binned average of 10 points and the green lines show the best fit determined by MLE (thick line) and the 95% confidence intervals (thin lines). It is important to note that the black line is a theoretical curve, not fitted to the data.

of undergoing a force reversal, demonstrating that the increased fitting of the relationship between the mean attachment duration number of force reversals observed at larger forces is only due to and the force. The theoretical rate of detachment (Fig. 5D) the binding interactions becoming longer with force. based on the kinetic scheme in Fig. 5C agrees well with the To better understand the relationship between the rate of measured rate. This modeling indicates that the primary source actomyosin dissociation and the force reversal behavior, kinetic of the force sensitivity of actomyosin detachment stems from modeling was used (Fig. 5C; details in SI Materials and Methods). force-induced slowing of exit from the high-force state. The modeling assumed that during a binding interaction, the myosin can transit between high- and low-force states (i.e., the Discussion force-reversed state) where the rates of entry (k−r; Fig. 5B) and The NTR of nucleotide-free Myo1b was found to be in a con- exit (k+r; Fig. S1) from the low-force state are force independent. formation well positioned to communicate the position of the The myosin can detach from the actin via either the force- LAH to the motor domain (4). This finding led us to the hy- dependent (kf(F)) or force-independent (ki) pathways (Figs. 4B pothesis that the NTR may be important for myosin mechano- and 5A). The rates of these key transitions were fixed based on sensing. In the recently published crystal structure of Myo1c (8), mechanical experiments. k+r was fixed based on the cumulative the NTR was not resolved, possibly due to its conformational − distribution of time spent in the reversed state (Fig. S1; 1.3 s 1). flexibility or changes in its position based on the nucleotide- The equilibrium constant between the high- and low-force states binding state of the myosin; however, the hydrophobic cluster of in the absence of force was calculated by measuring the fraction residues important for positioning the Myo1b NTR in the nu- of time spent in each substate (0.95) and this value permitted the cleotide-free state (V17, V21, L22, L23, Y47, S50, F77, Y78, and −1 calculation of rate of entry into the low-force state (k−r = 1.2 s ). P82 in the motor and F634 and F694 in the LAH) is also present The force-dependent rate of actomyosin dissociation was fixed in Myo1c (V13, V17, L18, L19, Y44, P47, F74, Y75, and P79 in based on the force dependence of actomyosin detachment in the the motor and F634 and F689 in the LAH). Despite these simi- −1 absence of force reversals (Fig. 5A; kf = 2.0 s and ddet = 4.7 nm). larities, the residues important for anchoring the Myo1b NTR The force-independent rate of detachment was fixed based on the (L10, L11, and M15) in this pocket and for interacting with cal- rate of the force-independent pathway from the MLE fitting (ki = modulin (K7) are not conserved in Myo1c (Fig. 1 A and B), so we − 0.059 s 1). Moreover, the data show that the high-force state is expected isoform-specific sequence differences to confer unique formed immediately after actomyosin association because there is mechanical properties (4). Despite not knowing the conformation not an ∼1-s delay before the development of force at the initiation of the NTR in the Myo1c constructs, our results demonstrate of a binding interaction (Fig. 4A). clearly that the NTR of Myo1c serves an important role in de- A simple kinetic scheme can be constructed to describe both fining the mechanochemical properties of the motor. Deletion of Δ > the force sensitivity of actoMyo1c N- b detachment and the force the Myo1c NTR or appending the Myo1b NTR to Myo1c leads to reversals (Fig. 5C). Using statistical kinetics (details in SI Ma- changes in the unloaded kinetics, speed of unloaded sliding, terials and Methods and Fig. S2), it is possible to derive the mechanics of the working stroke, and, strikingly, the ability of theoretical overall actomyosin detachment rate as a function of myosin to adjust its kinetics in response to load. force for this kinetic scheme (Fig. 5D). It is important to note that this curve is derived using statistical kinetic modeling based The NTR Tunes Myosin-I Mechanochemistry. Although the actomy- Δ on the independently measured rate constants and not MLE osin detachment rates and sliding velocities of Myo1c, Myo1c N,

E3342 | www.pnas.org/cgi/doi/10.1073/pnas.1506633112 Greenberg et al. Downloaded by guest on September 25, 2021 Δ > PNAS PLUS and Myo1c N- b appear to be limited by the rate of ADP release whether it plays a role as an allosteric regulator of the post– in the absence of force, the three different constructs examined power-stroke state. One possibility is that force causes the dis- have distinct responses to load. Myo1c’s actin-detachment rate is sociation and reassociation of the bound calmodulin in these insensitive to forces <1.5 pN, but it shows force sensitivity at constructs, leading to transient drops in force; however, additional Δ resisting loads >2 pN. Myo1c N’s actin-detachment rate is in- structural information is required to better understand these in- Δ > sensitive to forces <5 pN, and Myo1c N- b’s rate is sensitive to teractions. The frequency of force reversals for Myo1c is in- Δ Δ > forces <2 pN (Fig. 4B). creased in both Myo1c N and Myo1c N- b. Similar force reversals The actin-detachment rate of Myo1c slows at forces >2pN were also observed in a Myo1b construct lacking the NTR (4) and due to force-induced slowing of the rate of ATP-induced disso- native myosin-V (31), which does not have an NTR that sits be- ciation (10). This same transition might also be force sensitive tween the motor and the LAH. This behavior suggests a more Δ for Myo1c N, but we are unable to resolve it because the rate of general role of the NTR in stabilizing the post–power-stroke state. ATP-induced actomyosin detachment is substantially increased It is tempting to speculate that the myosin-V NTR is positioned to by the removal of the NTR (Fig. 2A and Table 1). allow reversals of the power stroke, enabling the myosin to walk For both Myo1b (4) and Myo1c in the absence of load, removal large distances past obstacles in the dense actin meshwork. Con- of the NTR slows the rate of ADP release while accelerating the sistent with this idea, the frequency of undergoing a force reversal maximal rate of ATP-induced dissociation. This indicates that in in myosin-V increases with load (31), enabling the myosin to dy- the absence of load, removing the native NTR increases the ac- namically regulate the probability of back stepping (31, 32). In tivation energy barrier for ADP release while decreasing the contrast, by stabilizing the post–power-stroke conformation in barrier for ATP-induced dissociation. Appending the NTR of myosin-I over a range of forces, the NTR may prevent transient Myo1b to Myo1c is not able to rescue these changes. Although drops in force that would cause actomyosin dissociation during Δ Δ > Myo1c N and Myo1c N- b have similar unloaded kinetics, a major force-induced long-lived interactions. difference in their energy landscapes becomes apparent when the myosins are placed under load. Specifically, whereas the distance Relationship to Other Myosins. The NTR sequence is highly di- Δ to the transition state of ADP release for Myo1c and Myo1c N is vergent between different myosin-I isoforms, suggesting that it Δ > <0.5 nm, it is 4.5 nm for Myo1c N- b. Thus, the transition state for plays a role in generating functional diversity among family Δ > ADP release from actoMyo1c N- b is more similar to the products members. Moreover, it has been shown that the NTR of Myo1c than to the reactants (29), which is consistent with what we found is alternatively spliced, yielding isoforms with different sub- previously for Myo1b (5–7). The NTR thus regulates both the cellular localization (33–35). This alternative splicing might also height and position of the transition state. It is likely that the NTR play a role in establishing diversity of function among these serves as an allosteric regulator of the active-site conformation different isoforms by mechanochemically tuning these myosins. through its close positioning with the transducer and loop-helix- Most myosin isoforms (including myosin-II, -V, and -VI) have loop motif of the motor domain (4). The myosin-II N terminus is SH3-like domains at their NTRs. In myosin-V and -VI, the NTR also positioned near these structural elements and its removal does not interact with either the LAH or a conserved loop-helix- leads to slowing of the rate of ADP release (30), suggesting that loop (LHL) motif in the motor domain (18, 19) whereas in myosin- the NTR may serve a similar role in some other myosin isoforms. II, it interacts with the LHL domain but not the LAH (17). The Although the ADP release transitions of both Myo1b and NTRs in these myosins likely also play roles in allosterically tuning Δ > Myo1c N- b are force sensitive, the magnitude of their force myosin’s mechanochemistry. Consistent with this notion, deletion sensitivities (as defined by the distance to the transition state, of the Dictyostelium myosin-II NTR leads to significant changes in ΔN->b ddet) is different. The distance to the transition state for Myo1c the unloaded kinetics and actomyosin motility (30) and alternative [4.5 (−1.1/+2.2) nm] is ∼2.5-fold smaller than for Myo1b splicing of the Drosophila muscle myosin-II NTR changes the [12 (−0.3/+1.6) nm] (5). This means that whereas 2 pN of force myosin’s power output, sliding velocity, and ATPase rates (36, 37). will slow the motility rate of Myo1b by 340-fold, it will slow the Future studies are necessary to determine whether the role of the Δ > motility rate of Myo1c N- b by only 8.9-fold. Nevertheless, the NTR in tuning mechanochemistry is a universal feature of myosins. data clearly demonstrate that the NTR affects both the regime of BIOPHYSICS AND forces over which the myosin is sensitive and the kinetic transition Materials and Methods COMPUTATIONAL BIOLOGY that is affected by force. Myo1c constructs were expressed in Sf9 cells and purified as previously de- scribed (10). These constructs were examined using stopped-flow transient The Native NTR Stabilizes the Post–power-Stroke Conformation. The kinetic techniques, optical trapping techniques, and in vitro motility assays appearance of force reversals when the native NTR is removed as previously described (10). Details of experimental conditions used for the or swapped indicates that the NTR plays a role in energetically optical trapping, in vitro motility, and transient kinetic experiments are stabilizing a post–power-stroke state by reducing the probability described in SI Materials and Methods. Data were analyzed using custom of entering the force-reversed state. Not all binding interactions software written in Matlab and Labview. Details of the data analysis and kinetic modeling are described in SI Materials and Methods. from the same molecule contain force reversals, suggesting that additional interactions likely contribute to the stabilization of the – ACKNOWLEDGMENTS. The authors acknowledge Michael Woody for help- post power-stroke state. From our data, it is not clear whether the ful discussions on MLE fitting and acknowledge grants from the National NTR plays a direct role in mechanically stabilizing this state or Institutes of Health (R01GM057247 to E.M.O. and K99HL123623 to M.J.G.).

1. Hartman MA, Spudich JA (2012) The myosin superfamily at a glance. J Cell Sci 125(Pt 7. Lewis JH, Greenberg MJ, Laakso JM, Shuman H, Ostap EM (2012) Calcium regulation 7):1627–1632. of myosin-I tension sensing. Biophys J 102(12):2799–2807. 2. De La Cruz EM, Ostap EM (2004) Relating biochemistry and function in the myosin 8. Münnich S, Taft MH, Manstein DJ (2014) Crystal structure of human myosin 1c—the superfamily. Curr Opin Cell Biol 16(1):61–67. motor in GLUT4 exocytosis: Implications for Ca2+ regulation and 14-3-3 binding. J Mol 3. Bloemink MJ, Geeves MA (2011) Shaking the myosin family tree: Biochemical kinetics Biol 426(10):2070–2081. defines four types of myosin motor. Semin Cell Dev Biol 22(9):961–967. 9. Batters C, et al. (2004) Myo1c is designed for the adaptation response in the inner ear. 4. Shuman H, et al. (2014) A vertebrate myosin-I structure reveals unique insights into EMBO J 23(7):1433–1440. myosin mechanochemical tuning. Proc Natl Acad Sci USA 111(6):2116–2121. 10. Greenberg MJ, Lin T, Goldman YE, Shuman H, Ostap EM (2012) Myosin IC generates 5. Laakso JM, Lewis JH, Shuman H, Ostap EM (2008) Myosin I can act as a molecular force power over a range of loads via a new tension-sensing mechanism. Proc Natl Acad Sci sensor. Science 321(5885):133–136. USA 109(37):E2433–E2440. 6. Laakso JM, Lewis JH, Shuman H, Ostap EM (2010) Control of myosin-I force sensing by 11. Lewis JH, Lin T, Hokanson DE, Ostap EM (2006) Temperature dependence of nucleotide alternative splicing. Proc Natl Acad Sci USA 107(2):698–702. association and kinetic characterization of myo1b. Biochemistry 45(38):11589–11597.

Greenberg et al. PNAS | Published online June 8, 2015 | E3343 Downloaded by guest on September 25, 2021 12. Lin T, Greenberg MJ, Moore JR, Ostap EM (2011) A hearing loss-associated 28. Takagi Y, Homsher EE, Goldman YE, Shuman H (2006) Force generation in single con- mutation (R156W) decreases the myosin duty ratio and force sensitivity. Biochemistry ventional actomyosin complexes under high dynamic load. Biophys J 90(4):1295–1307. 50(11):1831–1838. 29. Howard J (2001) Mechanics of Motor and the (Sinauer Asso- 13. Coluccio LM, Geeves MA (1999) Transient kinetic analysis of the 130-kDa myosin I ciates, Sunderland, MA). (MYR-1 product) from rat liver. A myosin I designed for maintenance of tension? 30. Fujita-Becker S, et al. (2006) Functional characterization of the N-terminal region of J Biol Chem 274(31):21575–21580. myosin-2. J Biol Chem 281(47):36102–36109. 14. Adamek N, Coluccio LM, Geeves MA (2008) Calcium sensitivity of the cross-bridge 31. Sellers JR, Veigel C (2010) Direct observation of the myosin-Va power stroke and its cycle of Myo1c, the adaptation motor in the inner ear. Proc Natl Acad Sci USA 105(15): reversal. Nat Struct Mol Biol 17(5):590–595. 5710–5715. 32. Kad NM, Trybus KM, Warshaw DM (2008) Load and Pi control flux through the 15. Greenberg MJ, Ostap EM (2013) Regulation and control of myosin-I by the motor and branched kinetic cycle of myosin V. J Biol Chem 283(25):17477–17484. – light chain-binding domains. Trends Cell Biol 23(2):81 89. 33. Nowak G, et al. (1997) Evidence for the presence of myosin I in the nucleus. J Biol 16. Kollmar M, Dürrwang U, Kliche W, Manstein DJ, Kull FJ (2002) Crystal structure of the Chem 272(27):17176–17181. – motor domain of a class-I myosin. EMBO J 21(11):2517 2525. 34. Pestic-Dragovich L, et al. (2000) A myosin I isoform in the nucleus. Science 290(5490): 17. Rayment I, et al. (1993) Three-dimensional structure of myosin subfragment-1: A 337–341. – molecular motor. Science 261(5117):50 58. 35. Ihnatovych I, Migocka-Patrzalek M, Dukh M, Hofmann WA (2012) Identification and 18. Coureux PD, et al. (2003) A structural state of the myosin V motor without bound characterization of a novel myosin Ic isoform that localizes to the nucleus. Cyto- nucleotide. Nature 425(6956):419–423. skeleton 69(8):555–565. 19. Ménétrey J, et al. (2005) The structure of the myosin VI motor reveals the mechanism 36. Swank DM, et al. (2003) Variable N-terminal regions of muscle myosin heavy chain of directionality reversal. Nature 435(7043):779–785. modulate ATPase rate and actin sliding velocity. J Biol Chem 278(19):17475–17482. 20. De La Cruz EM, Ostap EM (2009) Kinetic and equilibrium analysis of the myosin AT- 37. Swank DM, Kronert WA, Bernstein SI, Maughan DW (2004) Alternative N-terminal Pase. Methods Enzymol 455:157–192. regions of Drosophila myosin heavy chain tune muscle kinetics for optimal power 21. Geeves MA, Perreault-Micale C, Coluccio LM (2000) Kinetic analyses of a truncated output. Biophys J 87(3):1805–1814. mammalian myosin I suggest a novel isomerization event preceding nucleotide 38. Spudich JA, Watt S (1971) The regulation of rabbit skeletal muscle contraction. binding. J Biol Chem 275(28):21624–21630. 22. Nyitrai M, Geeves MA (2004) Adenosine diphosphate and strain sensitivity in myosin I. Biochemical studies of the interaction of the - complex with – motors. Philos Trans R Soc Lond B Biol Sci 359(1452):1867–1877. actin and the proteolytic fragments of myosin. J Biol Chem 246(15):4866 4871. 23. Takagi Y, Shuman H, Goldman YE (2004) Coupling between phosphate release and 39. Putkey JA, Slaughter GR, Means AR (1985) Bacterial expression and characterization force generation in muscle actomyosin. Philos Trans R Soc Lond B Biol Sci 359(1452): of proteins derived from the chicken calmodulin cDNA and a calmodulin processed 1913–1920. gene. J Biol Chem 260(8):4704–4712. 24. Finer JT, Simmons RM, Spudich JA (1994) Single myosin molecule mechanics: Pico- 40. Meijering E, Dzyubachyk O, Smal I (2012) Methods for cell and particle tracking. newton forces and nanometre steps. Nature 368(6467):113–119. Methods Enzymol 504:183–200. 25. Chen C, et al. (2012) Kinetic schemes for post-synchronized single molecule dynamics. 41. Press WH (1992) Numerical Recipes in C: The Art of Scientific Computing (Cambridge Biophys J 102(6):L23–L25. Univ Press, Cambridge, UK), 2nd Ed. 26. Veigel C, et al. (1999) The motor protein myosin-I produces its working stroke in two 42. Wilks SS (1938) The large-sample distribution of the likelihood ratio for testing steps. Nature 398(6727):530–533. composite hypotheses. Ann Math Stat 9(1):60–62. 27. Nesmelov YE, et al. (2011) Structural kinetics of myosin by transient time-resolved 43. Moffitt JR, Chemla YR, Bustamante C (2010) Methods in statistical kinetics. Methods FRET. Proc Natl Acad Sci USA 108(5):1891–1896. Enzymol 475:221–257.

E3344 | www.pnas.org/cgi/doi/10.1073/pnas.1506633112 Greenberg et al. Downloaded by guest on September 25, 2021