Load-dependent modulation of non-muscle myosin-2A function by tropomyosin 4.2

Tropomyosin isoforms play an important role in the organisation of cytoplasmic actomyosin complexes in regard to function and cellular localisation. In particular, Tpm4.2 is upregulated in rapidly migrating cells and responsible for the specific recruitment of the cytoplasmic class-2 myosin NM-2A to actin filaments during the formation of stress fibres. Here, we investigate how the decoration of F-actin with Tpm4.2 affects the motor properties of NM-2A under conditions of low and high load. In the absence of external forces, decoration of actin filaments with Tpm4.2 does not affect the gated release of ADP from NM-2A and the transition from strong to weak actin-binding states. In the presence of resisting loads, our results reveal a marked increase in the mechanosensitive gating between the leading and trailing myosin head. Thereby, the processive behaviour of NM-2A is enhanced in the presence of resisting loads. The load- and Tpm4.2-induced changes in the functional behaviour of NM-2A are in good agreement with the role of this myosin in the context of stress fibres and the maintenance of cellular tension.

The tropomyosin (Tpm) family consists of more than 40 isoforms generated by alternative splicing from four genes. The great majority of Tpm isoforms are classified as cytoskeletal isoforms 1 . They associate with actin filaments made from cytoskeletal β -or γ -actin and are responsible for the formation and segregation of actin filament populations that display distinct localization patterns, turnover dynamics, and cellular functions (for reviews see 2,3 ). Compositional variation of cytoplasmic actin-Tpm complexes appears to provide a mechanism for the selective recruitment of myosin isoforms [4][5][6] . Studies in yeast show that decoration of actin with cytoplasmic Tpm isoforms is required for myosin-5 to show processive behaviour 7,8 . The low molecular weight, cytoplasmic isoform Tpm4.2 is produced in a wide range of human cells and tissues. Depending on the cell type, Tpm4.2 appears to play a role in supporting different motile and contractile events. Tpm4.2 is prominent in growth cones of developing neurons, plays an essential role in the formation of a cap over the inner face of the sealing zone in osteoclasts, stabilizes actin-based structures associated with podosomes, and appears to contribute to the formation of focal adhesions and stress fibres in fibroblasts [9][10][11] . Tpm4.2 production is upregulated as contractile smooth muscle cells dedifferentiate into non-contractile, migrating cell types and some types of cancer cells. In the case of infiltrating ductal breast carcinomas, the aggressiveness of the tumour has been directly linked to Tpm4 expression 12 . The myofibroblastic cells in some inflammatory myofibroblastic tumours contain chromosomal rearrangements involving TPM4 and ALK, which encodes a receptor tyrosine-kinase. While ALK expression is normally restricted to neural tissues, the fusion protein is overproduced by the myofibroblastic spindle cells that make up this tumour, which can occur in many different parts of the body 13 .
Mammals produce three distinct filament-forming non-muscle myosin-2 isoforms (NM-2A, NM-2B, NM-2C). The corresponding myosin heavy chains are encoded by three different genes (MYH9, MYH10, MYH14). All three NM-2 isoforms require activation by phosphorylation of their regulatory light chain 14 . However, even in the activated form, they display much slower ATP turnover and motile activity than muscle myosin-2. Activated NM-2 heavy chain dimers assemble into bipolar myosin bundles via their coiled-coil tails. Bipolar NM-2 bundles are responsible for the contraction of antiparallel actin filaments that are found in cytoskeletal structures such as stress fibres and the contractile rings formed during cytokinesis (for reviews

Results
Tpm4.2 facilitates the transition of NM-2A to the strongly actin-binding myosin-ADP state. We produced active, soluble NM-2A motor constructs using the Sf9/baculovirus system. HMM-and S1-like myosin heavy chain fragments were co-produced with human native essential and regulatory light chains MYL6 and MYL12b. Typical yields were 0.8 mg NM-2A-HMM and 9 mg NM-2A-S1 from 2 · 10 9 cells. The S1 construct is constitutively active, while the HMM construct requires activation by phosphorylation of MYL12b 15,27 . Tpm4.2 was produced in E. coli with yields of approximately 1.3 mg of homogeneous protein per litre of culture medium ( Supplementary Fig. S1).
To elucidate how the decoration of actin filaments with Tpm4.2 influences their interaction with NM-2A, we measured the actin-activated ATPase activity of NM-2A-HMM during steady-state at varying concentrations of bare and Tpm4.2-decorated F-actin (Fig. 1a). Using bare F-actin, the maximum turnover rate k cat of NM-2A-HMM was 0.22 ± 0.03 s −1 . In the presence of 10 μ M Tpm4.2, k cat increased 1.5-fold to 0.34 ± 0.02 s −1 . The apparent dissociation equilibrium constant for F-actin binding in the presence of ATP (K actin ), which is equal to the actin concentration at the half-maximum rate, was unchanged by the presence of Tpm4.2 (see Table 1). Similar results were obtained for NM-2A-S1 (see Supplementary Figure S2 and Table 1). The activating effect on the NM-2A-HMM ATPase rate reached a plateau value at a concentration greater than 10 μ M Tpm4.2 (Fig. 1b). It appears likely that saturation is reached at lower concentrations with the N-terminally acetylated form of Tpm4.2 28 .
The rate-limiting step in the ATPase cycle of NM-2A was shown to be the transition from the myosin-ADP-P i state to the strongly actin-bound myosin-ADP state 18 . We wanted to examine whether the increase of the maximum ATPase rate is caused by the acceleration of this step. To facilitate the interpretation of the signals in the stopped-flow experiments, we used single-headed NM-2A-S1. Using the sequential mixing mode of a stopped-flow system, NM-2A was mixed with ATP, aged for 5 s in order to form a myosin-ADP-P i complex, before mixing with bare or Tpm4.2-decorated F-actin. Phosphate release was followed by the fluorescence increase of the phosphate sensor MDCC-PBP (see methods). As shown in Fig. 1c, under the experimental conditions, the dissociation rate constant for the release of phosphate from NM-2A-S1 is more than 2-fold faster with Tpm4.2-decorated actin filaments (0.179 ± 0.012 s −1 ) compared to bare filaments (0.084 ± 0.003 s −1 ). This finding supports the notion that the decoration of F-actin with Tpm4.2 accelerates the transition of the NM-2A motor to the strongly actin-bound states. ADP release was shown to limit the detachment rate of NM-2A from actin at saturated ATP concentrations 18 . Therefore, a change in the rate of ADP release has a major impact on the fraction of time NM-2A spends in strongly actin-bound states. We measured the ADP release rate from NM-2A-S1 on bare and Tpm4.2-decorated actin filaments by rapidly mixing an actin-NM-2A-S1-mant-ADP complex with a large excess of ATP (4 mM). Under these conditions, the rate of ATP-binding is limited by the dissociation of mant-ADP, which can be followed by the drop in mant-fluorescence that is associated with the displacement of mant-ADP from the nucleotide-binding pocket (see Supplementary Figure S3a and b). In the presence of Tpm4.2, the dissociation rate constant of ADP release from actin-NM-2A-S1 was reduced by approximately 13%, as shown in Fig. 1d (no Tpm: 6.05 ± 0.25 s −1 ; Tpm4.2: 5.29 ± 0.06 s −1 ; p < 0.01).
The intramolecular strain between the two heads of NM-2A-HMM is known to influence their ADP release rates 23 . The leading head that experiences a resisting intramolecular strain from the trailing head was shown to release ADP at a slower rate than the single-headed S1. The trailing head's ADP release is stimulated by assisting strain. We determined the individual ADP release rates for the two HMM heads on bare and Tpm4.2-decorated actin filaments according to the protocol described by Kovacs et al. 23 . A preformed complex of actin-NM-2A-HMM with mant-ADP was rapidly mixed with a large excess of ADP (2 mM), which caused a drop in mant fluorescence upon mant-ADP displacement (see Supplementary Figure S3c and d). On bare actin filaments, the rate of ADP release from the leading head of HMM is 12.7-fold slower (0.48 ± 0.07 s −1 ; Fig. 1d) than the rate Scientific RepoRts | 6:20554 | DOI: 10.1038/srep20554 observed for the S1 construct. The trailing head releases ADP 1.5-fold faster (9.1 ± 2.2 s −1 ; Fig. 1d) than the S1 construct on bare actin filaments. Thereby, we confirm that our NM-2A-HMM construct exhibits strain-sensitive ADP release. A similar extent of gating was observed for the HMM construct in the presence of Tpm4.2 (0.50 ± 0.07 s −1 ; 7.5 ± 2.0 s −1 ; Fig. 1d).
Tpm4.2 increases the number of actin landing events and the maximum filament sliding velocity on an NM-2A-coated surface. We used landing assays to investigate the effect of Tpm4.2 on the interaction between NM-2A and actin filaments 29 . NM-2A-HMM was immobilised on the surface of a nitrocellulose-coated coverslip. In a TIRF microscope, we observed the transient interactions of fluorescently labelled actin filaments with the myosin-decorated surface. In the presence of 4 mM Mg 2+ -ATP, the landing rate of bare actin filaments dropped rapidly, when the NM-2A density on the surface was step-wise reduced. In contrast, the number of productive interactions with the surface dropped much slower when the actin filaments were decorated with Tpm4.2 (Fig. 2a). The data shown in Fig. 2a can be described with the model devised by Hancock and Howard 29 , where R n is the landing rate, Z is a parameter that determines the collision rate of filaments with the surface, ρ is the myosin density, ρ 0 corresponds to the surface area over which myosins interact with an actin filament, and n can be interpreted as the minimal number of myosin molecules required for a productive landing event. The Tpm4.2-induced change of n in the landing assay from 43 ± 26 to 3.4 ± 1.5 shows that NM-2A is shifted towards strongly actin-attached states by Tpm4.2. Next, we examined the influence of Tpm4.2 on the speed of translocation of actin filaments in an in vitro motility assay. We decorated coverslips with varying densities of NM-2A-HMM and recorded the movement of bare and Tpm4.2-decorated actin filaments following the addition of assay buffer containing 4 mM Mg 2+ -ATP. In the case of bare actin filaments the mean sliding velocity reached a plateau value of 203 ± 7 nm/s at a myosin density of approximately 10,000 molecules per μ m 2 ( Fig. 2b; dashed blue line). The speed of Tpm4.2-decorated filaments continued to increase up to twice the surface density of NM-2A-HMM molecules, reaching a plateau value of 287 ± 11 nm/s ( Fig. 2b; dashed red line). A widely accepted model by Uyeda et al. 30,31 suggests that at high myosin densities the sliding velocity in the in vitro motility assay is limited by the detachment rate of myosin from actin, which in the case of NM-2A is determined by the rate of ADP release from the complex 18 . In the absence and presence of Tpm4.2, our stopped-flow experiments show no significant change in the rate of ADP release (Table 1). Therefore, the 1.4-fold Tpm4.2-mediated increase in velocity cannot be explained by a change  in the rate of ADP release. It appears more likely that the Tpm4.2-mediated increase in velocity results from the improved coordination of the power-strokes of myosin heads interacting with an actin filament.
NM-2A takes processive steps along actin filaments. The rate of ADP release from the leading head of NM-2A-HMM is close to the overall cycle rate (see above), which suggests that the myosin is likely to take a step before both heads dissociate. NM-2A-HMM should therefore be able to perform short processive runs on actin filaments. We used a three-bead optical trap setup to observe single myosin molecules interacting with single actin filaments (Fig. 3a). Interactions between actin and NM-2A-HMM were detected by a reduction of the noise amplitude of the bead-actin-dumbbell position signal. Processive movement was detected by analysing the signal of the mean position along the filament axis of the dumbbell. In fact, we observed processive runs with NM-2A-HMM at 2 mM ATP on bare actin (Fig. 3b). While the majority of runs ended after a few steps, we frequently recorded runs in which the myosin moved longer distances up to a constant resisting force (stall force) of 3.4 ± 0.9 pN (number of events used to determine the stall force: n = 24). After reaching the stall force, the NM-2A-HMM construct underwent successive back-and forward stepping as illustrated in Fig. 3b. Figure 3c illustrates the step size distribution of 1204 observed steps. Forward and backward step sizes appear as distinct populations. The apparent mean forward step size of NM-2A-HMM is 5.0 ± 1.6 nm. The apparent mean backward step size is 4.7 ± 1.9 nm. These values provide an estimate of the minimal step size, since the movement of the myosin is partially buffered by the compliance in the bead-actin link. We compared the measured step sizes at low displacement with those at high displacements, where the bead-actin link is strongly stretched and compliance is negligible (see Supplementary Figure S6). Considering this compliance effect, the mean values above underestimate the step sizes by approximately 10-20%. Thus, the measured step sizes are in good agreement with the distance of 5.5 nm between the neighbouring actin subunits in a strand 32,33 . Decoration of actin filaments with Tpm4.2 enhances the processivity of NM-2A in the presence of resisting loads. On bare actin filaments, the stepping rate of NM-2A-HMM remains nearly constant during processive runs until the myosin reaches the stall plateau (Fig. 3b). In contrast, we observed a marked reduction in the stepping rate for myosin approaching the stall position on Tpm4.2-decorated actin filaments. As the examples shown in Fig. 4 illustrate, the wait-time between steps increased markedly at higher loads. Detailed comparison of the wait-times between steps recorded for runs on bare actin filaments and on Tpm4.2-decorated filaments ( Fig. 5a and b, respectively) reveals that Tpm4.2 reduces the stepping rate of NM-2A when the myosin works under load. We applied a simple exponential model to describe the data shown in panels a and b of Fig. 5  (Fig. 5c).
Here, k step,0 is the stepping rate at zero load, F is the resisting force, and δ is a distance parameter that is a measure of force-sensitivity of the stepping rate [34][35][36][37]  . The stepping rate at zero load is similar to the ADP-release rates that we determined for the NM-2A-S1 and the trailing head of NM-2A-HMM in our stopped-flow experiments. This result suggests that the stepping rate is limited by ADP release. We did not determine the lifetimes of strongly-bound states that follow ADP release. However, others have shown that ATP-induced dissociation is an order of magnitude faster than ADP release for NM-2A at saturating ATP concentrations 18 . Therefore, it is unlikely that the rate of ATP-induced dissociation contributes to the stepping rate in our experimental setup. The distance parameter δ is increased 2-fold by the presence of Tpm4.2 (no Tpm: δ = 1.68 (+ 0.59/− 0.35) nm; Tpm4.2: δ = 3.42 (+ 0.59/− 0.44) nm). Based on these results, we conclude that Tpm4.2-decoration of actin filaments raises the energy barrier and changes the position of the force-sensitive transition state for taking a step, when the motor works against resisting loads. Extended dwell-times of the myosin under high loads are expected to support its processive behaviour. Therefore, we determined the probability of NM-2A-HMM taking a processive step as a function of the resisting force (Fig. 5d). Both, in the case of bare and Tpm4.2-decorated actin, we observed a steep rise in the probability of processive stepping with increasing loads. On bare actin filaments, the maximum probability for taking a processive step (0.77 ± 0.04) is reached at a resisting force of 1.38 ± 0.04 pN. In the presence of Tpm4.2, NM-2A reaches its maximum processivity (stepping probability of 0.77 ± 0.02) already at a resisting force of 0.93 ± 0.04 pN. The combination of both a slower stepping rate and increased processivity on Tpm4.2-decorated actin extends the attachment time for NM-2A working against an external force. As shown in Fig. 5e, both the mean and maximum time that NM-2A-HMM spent in the stall position until dissociation were prolonged 5-fold by Tpm4.2. Strikingly, we observed attachment times of more than 300 s on Tpm4.2-decorated filaments.

Discussion
We aimed to elucidate how the decoration of actin filaments with Tpm4.2 affects the motor properties of NM-2A.
A cell-based study has previously demonstrated that Tpm4.2 binding to actin in emerging stress fibres recruits NM-2A to these filaments 11 . The cryo-EM structure of a Tpm-actin-myosin complex revealed stereospecific interactions between tropomyosin and myosin 38 . In the case of Tpm4.2-decorated actin filaments, the stereospecific contact area with NM-2A is predicted to be enlarged by approximately 300 Å 38 . Our kinetic analysis suggests that enhanced recruitment of NM-2A is achieved in part by accelerating the rate-limiting transition to the strongly actin-bound myosin-ADP state (Fig. 6a). Tighter binding and increased coupling between actin binding and release of the hydrolysis products can act like a selectivity filter for a specific myosin isoform. The avidity effect that results from NM-2A assembling into bipolar filaments may further enhance selective binding, leading to a situation were Tpm4.2-decorated actin filaments promote the specific recruitment of bipolar NM-2A filaments. The kinetic properties of single-headed NM-2A-S1 were described to be compatible with a low duty-ratio motor 18 . However, experiments performed by the same group with NM-2A-HMM did already reveal a strain-sensitive ADP release step and gating between the two heads 23 . We confirm these results, as we find that ADP is released from NM-2A-HMM at two clearly distinct rates. The slow rate that is associated with ADP release from the leading head 23 is slightly faster (0.48 ± 0.07 s −1 ) than the overall ATPase rate (0.22 ± 0.03 s −1 ), which is limited by re-binding of a detached head in the myosin-ADP-P i state. These kinetic data predict that NM-2A should be able to take several processive steps under load-free conditions. Our optical trap experiments  Figure S5). Gaussian fits give a mean forward step size of 5.0 ± 1.6 nm and a mean backward step size of 4.7 ± 1.9 nm (n = 1204 steps; mixed from runs on bare and Tpm4.2-decorated filaments). The distance between neighbouring actin subunits corresponds to 5.5 nm 32,33 .
Scientific RepoRts | 6:20554 | DOI: 10.1038/srep20554 demonstrate that the extent of gating between the heads of NM-2A-HMM is indeed sufficient for the myosin to move in a processive fashion. As expected, most processive runs are short, caused by the frequent simultaneous dissociation of both heads. Different from myosins that are involved in the long-distance transport of cargo, such as myosin-5 and myosin-6, single molecules of NM-2A do not perform long distance runs. In their physiological environment multiple motors of NM-2A are assembled as bundles. Thus, sustained attachment to an actin filament is achieved by simultaneous binding of several motors of a bundle. Similar to NM-2B 36 , NM-2A takes short steps of 5.5 nm making it less suitable for long-distance transport, as this requires the motor to rotate around the actin filament axis. In contrast, a motor like myosin-5 takes 36 nm steps along the pseudo-helical actin filament repeat 39 , which allows the myosin to stay continuously on the same side of the filament. The fact that the processive stepping of NM-2A leads to a rotation of the interacting actin filament has interesting implications for stress fibre contractility. Katoh et al. have indeed provided evidence that actin filaments in stress fibres are rotated during contraction. The authors suggest that a rotation of actin filaments induces coiling of the stress fibres and thereby enables the fibres to shorten more extensively 40 .
The changes in the stepping rate observed in our optical trap experiments indicate that the detachment of NM-2A from F-actin is delayed in the presence of resisting forces. In the absence of Tpm4.2, the characteristic distance parameter δ (1.68 nm), which is a measure of load-sensitivity for the detachment rate, is low. The δ -value determined for NM-2A is similar to the δ -values reported for myosins that are involved in tension maintenance and work under high load conditions, such as smooth muscle myosin-2 (δ = 1.3 nm) 37 and NM-2B (δ = 0.99 nm) 36 . In comparison, a δ -value of 4.3 nm was reported for a long-distance transporter like myosin-5 41 . The load-sensitivity of the detachment rate is amplified for NM-2A on Tpm4.2-decorated actin filaments with δ increasing 2-fold. Consequently, the detachment of NM-2A heads is strongly delayed under high loads. As a result, NM-2A will stay anchored to Tpm4.2-decorated filaments and move in a processive manner under high-load conditions.
In conclusion, our findings support a role of NM-2A as a motor that is involved in tension maintenance and contractility of cellular actin networks. The properties of Tpm4.2 are ideally suited to mediate the assembly of contractile actomyosin structures. We propose that Tpm4.2-decoration of actin filaments can act in part as a selectivity filter promoting the specific recruitment of NM-2A (Fig. 6b). The results from our in vitro motility assay provide evidence for the ability of Tpm4.2 to synchronise the power-stroke of NM-2A heads that are attached to the same filament. Once NM-2A is incorporated into an F-actin network and load is applied, the improved synchronisation of NM-2A molecules and their ability to step in a processive fashion will enhance the function of bipolar bundles and stabilise the contractile network (Fig. 6c). Studying the functional consequences of Tpm4.2-decoration of actin filaments for bipolar bundles of NM-2A in more detail is an interesting task for future experiments.  (MLCK). NM-2A-S1 was used without prior phosphorylation. The steady-state ATPase rates were determined in an NADH-coupled assay 42 (2) was fitted to the data in (a) (blue curve) and (b) (red curve). The curves represent best fits for the parameters k step,0 and δ . The coloured areas indicate 95% confidence limits based on the errors of the best-fit parameters (see text). A sliding mean filter (20 points window size) was applied to the data in (a and b) and plotted as blue and red points, respectively. (d) NM-2A-HMM shows load-enhanced processivity. Tpm4.2decoration of the actin filament reduces the resisting force that is required to reach maximum processivity. The probabilities for a processive step were determined by relating the frequency of myosin taking a follow-on step (omitting initial steps) to the frequency of dissociation events. The frequencies were grouped into force bins (n = 19-182 frequencies per bin). Each bin centre is represented by a point in the diagram. The intersection of the two trend lines indicates the force at which the myosin processivity reaches its maximum.  The ADP-state complex of the S1 construct was formed by mixing 2 μ M bare or Tpm4.2-decorated actin filaments, 2 μ M NM-2A-S1 and 2 μ M mant-ADP in MOPS Experimental Buffer (20 mM MOPS pH 7.0, 50 mM KCl, 5 mM MgCl 2 ) containing 0-20 μ M Tpm4.2. Displacement of mant-ADP was triggered by rapid mixing with an equal volume MOPS Experimental Buffer containing 4 mM ATP. The drop in mant-ADP fluorescence (365 nm excitation, 389 nm long-pass emission filter) upon dissociation from the myosin was recorded for 2 s. The fluorescence transients were best-described by fitting a single exponential decay function to the data.

Steady
The ADP-state complex of the HMM construct was formed by mixing 2 μ M bare or Tpm4.2-decorated actin filaments, 0.4 μ M NM-2A-HMM and 10 μ M mant-ADP in MOPS Experimental Buffer containing 0 or 10 μ M Tpm4.2. Displacement of mant-ADP was triggered by rapid mixing with an equal volume MOPS Experimental Buffer containing 2 mM ADP. The drop in mant-ADP fluorescence upon dissociation from the myosin was recorded for 20-50 s. The fluorescence transients were best-described by fitting a double exponential decay function to the data.
Phosphate release kinetics. Phosphate release kinetics were measured as described earlier 18,43  Inactive myosins that could interfere with the assay were removed by spin-down with 5 μ M actin in the presence of 2 mM ATP at 100,000 g in a table-top ultracentrifuge (Beckman Optima Max) at 4 °C. The supernatant was diluted in Assay Buffer to myosin concentrations of 25-500 μ g/ml. Myosin densities on the nitrocellulose-surface were calculated assuming that all myosins in solution would completely bind to the surface. According to this assumption, a stock of 500 μ g/mL NM-2A-HMM that is infused into the flow cell provides a density of 93,750 molecules per μ m 2 . To reduce deviations in the myosin concentrations between the bare actin and Tpm4.2-actin cases as much as possible, the same myosin dilutions from the same myosin stock were used for the respective myosin densities. One chamber volume of the myosin dilution was infused into the flow cell. It was incubated for 2 min to allow the myosin to bind to the nitrocellulose-coated glass surface. The flow cell was then flushed with 3 volumes of BSA/AB (0.5 mg/ml bovine serum albumin in Assay Buffer) and incubated for 2 min to block the surface. Next, 10 nM TRITC phalloidin-labelled actin filaments in BSA/AB with 0 or 10 μ M Tpm4.2 were infused and allowed to bind to the myosin surface for 2 min. Excess actin was washed out with 2 volumes of BSA/AB containing 0 or 10 μ M Tpm4.2. Finally, BSA/AB containing 4 mM ATP, 0 or 10 μ M Tpm4.2 and an anti-bleach system consisting of 5 mg/ml glucose, 0.1 mg/ml glucose oxidase and 0.02 mg/ml catalase was infused. Filament gliding over the myosin surface at 30 °C was observed on an Olympus IX70 Fluorescence Microscope with a 60 × Plan Apo 1.49 NA oil immersion TIRF objective and an additional 1.5-fold magnification. Filaments were excited with light of 550 nm wavelength in epi-fluorescence mode. Fluorescence light was filtered using a 580 nm band pass dichroic mirror and detected with an Orca Flash 4.0 CMOS camera (Hamamatsu Photonics). Filament movement was tracked using DiaTrack 3 45 .
Landing assay. Decorated actin filaments were formed by pre-incubation for 30 min with 10 μ M Tpm4.2 at room temperature. To assure a comparable filament length distribution, bare and Tpm4.2-decorated filaments were made from the same actin stock and treated similarly. NM-2A-HMM was prepared and myosin densities were calculated as described for the in vitro motility assay. First, the myosin was allowed to bind to the nitrocellulose for 5 min. Then, the surface was blocked with 3 volumes of BSA/AB for 2 min. Finally, 2 volumes of BSA/ AB containing 4 mM ATP, 20 nM TRITC phalloidin-labelled actin filaments, 0 or 10 μ M Tpm4.2 and anti-bleach solution containing 5 mg/ml glucose, 0.1 mg/ml glucose oxidase and 0.02 mg/ml catalase were infused. Filament landing rates were observed at 30 °C using the Olympus Fluorescence Microscope in TIRF mode and a 565 nm laser for excitation. Landing rates were counted manually from 30 s movies at a frame rate of 2 per second. Only filaments that touched the surface and were bound for at least two frames (1 second) were counted as an event. Filament lengths were measured using the ImageJ plugin Jfilament 46 . The average filament length was 4.1 ± 2.1 μ m (mean ± SD) for bare filaments and 3.7 ± 2.3 μ m for Tpm-4.2-decorated filaments. The difference in the landing rates is therefore not a consequence of differences in the filament length. Optical trap assay. Optical trap experiments were carried out as described previously ( 32,47 for setup).
Flow cells were made as described for the in vitro motility assay, but the cover slips were coated with a suspension of 1.5 μ m glass beads in 0.05% nitrocellulose in amyl acetate. Actin was coupled to polystyrene beads via biotin-neutravidin according to Ishijima et al. 48 . To visualise these biotin-coated 0.8 μ M polystyrene beads (Kisker Biotech, Steinfurt, Germany) in the same fluorescence channel as F-actin, we treated them with a 1:100 mixture of Texas-Red-labelled streptavidin and unlabelled neutravidin. For Tpm4.2-decoration, TRITC phalloidin-labelled actin filaments were pre-incubated with 10 μ M Tpm4.2 at room temperature for at least 15 min. NM-2A-HMM was phosphorylated as described for the in vitro motility assay, diluted to 0.5 μ g/ml in Assay Buffer and infused into the flow cell. The myosin was allowed to bind to the surface for 1 min. Next, a blocking step with BSA/ AB was carried out for 1 min. Finally, a mix of the streptavidin-coated polystyrene beads and 0.6 nM TRITC phalloidin-labelled biotinylated actin filaments in Assay Buffer containing 20 mM DTT, 1 μ M phalloidin, 2 mM ATP, 0 or 10 μ M Tpm4.2, 2 mg/ml glucose, 1 mg/ml glucose oxidase and 0.2 mg/ml catalase was added. The chamber was sealed with vacuum grease. The optical trap experiments were carried out on a custom-build system as described previously 47 . The trap stiffness was set to 0.06-0.08 pN/nm for both traps. The actin dumbbell was stretched to an initial bead-actin link stiffness of approximately 0.1 pN/nm. The link stiffness and its change during processive runs were considered when calculating resisting forces on the myosin. Motor activity was monitored by recording the position of the trapped beads using two 4-quadrant detectors at a sampling rate of 10 kHz. The difference in the noise amplitude between a myosin-bound and unbound actin dumbbell was increased by applying a low-amplitude (80 nm peak to peak), high-frequency oscillation (1 kHz sinusoid) to one of the traps 37 . The oscillation did not affect processive runs of the myosin. Data records were evaluated using MatLab routines. The traces of processive runs were tracked manually to find the transition between steps.
To ensure that observed processive events were the result of single NM-2A-HMM interactions, the glass bead surface was sparsely coated with the myosin construct. We assayed a flow cell with less than one activity on 15 platforms and still observed the same processive runs. For major data acquisition we worked in a regime where 51 out of 122 glass beads displayed ATP-sensitive interactions with the actin dumbbell. According to these statistics, the probability that an observed activity is due to multiple motors is approximately 29%. However, we observed processive movement with 44 of the 51 activities. The binomial probability that all 44 of these 51 activities are due to multiple motors is 2.7 × 10 −17 . This calculation does not take into account positional restraints on the bead surface that may limit the motors' ability to interact with the same actin filament and to adopt a proper orientation to support processive movement. Considering these criteria, the probability that the events are caused by multiple motors is even lower. In addition, when applying a sliding mean filter to the data, individual steps of the myosin could be resolved (Figs 3b and 4a) and this clear staircase activity is unlikely to be the result of more than one motor.
Events that were caused by more than one motor could be easily recognised and sorted out as they usually produced multiples of the reported stall force and more than one stall plateau (see Supplementary Figure S7).