Three mammalian tropomyosin isoforms have different regulatory effects on nonmuscle myosin-2B and filamentous β-actin in vitro

The metazoan actin cytoskeleton supports a wide range of contractile and transport processes. Recent studies have shown how the dynamic association with specific tropomyosin isoforms generates actin filament populations with distinct functional properties. However, critical details of the associated molecular interactions remain unclear. Here, we report the properties of actomyosin–tropomyosin complexes containing filamentous β-actin, nonmuscle myosin-2B (NM-2B) constructs, and either tropomyosin isoform Tpm1.8cy (b.–.b.d), Tpm1.12br (b.–.b.c), or Tpm3.1cy (b.–.a.d). Our results show the extent to which the association of filamentous β-actin with these different tropomyosin cofilaments affects the actin-mediated activation of NM-2B and the release of the ATP hydrolysis products ADP and phosphate from the active site. Phosphate release gates a transition from weak to strong F-actin–binding states. The release of ADP has the opposite effect. These changes in dominant rate-limiting steps have a direct effect on the duty ratio, the fraction of time that NM-2B spends in strongly F-actin–bound states during ATP turnover. The duty ratio is increased ∼3-fold in the presence of Tpm1.12 and 5-fold for both Tpm1.8 and Tpm3.1. The presence of Tpm1.12 extends the time required per ATP hydrolysis cycle 3.7-fold, whereas it is shortened by 27 and 63% in the presence of Tpm1.8 and Tpm3.1, respectively. The resulting Tpm isoform–specific changes in the frequency, duration, and efficiency of actomyosin interactions establish a molecular basis for the ability of these complexes to support cellular processes with widely divergent demands in regard to force production, capacity to move processively, and speed of movement.

an integral part of most actin filaments in animals. Over 40 Tpm isoforms are produced by alternate promoter selection and splicing of four different genes: TPM1, TPM2, TPM3, and TPM4. Direct differences between the coding regions of TPM1-4 and alternative usage of variant exons 1, 2, 6, and 9 contribute to Tpm isoform diversity (1). The resulting gene products can be grouped in high-molecular-weight and lowmolecular-weight isoforms. High-molecular-weight isoforms are encoded by nine exons and bind seven consecutive actin subunits along the actin long-pitch helix. Low-molecularweight isoforms lack sequences encoded by exon 2 and interact only with six actin subunits. Although most Tpm isoforms show a wider tissue distribution (2), they are frequently referred to according to their dominant localization as striated muscle, smooth muscle, brain, or cytoskeletal isoforms.
In the sarcomere of striated muscles, isoforms Tpm1.1, Tpm1.2, Tpm2.2, and Tpm3.12 form a complex with troponin that regulates cardiac and skeletal muscle myosin-2 motor activity in a Ca 2ϩ -dependent manner (3). In smooth muscle cells, Tpm1.3, Tpm1.4, and Tpm2.1 play a more modulatory role in regulating contraction (4). Cytoskeletal tropomyosins (e.g. Tpm1.7, Tpm1.8, Tpm1.12, Tpm3.1, and Tpm4.2) represent the largest group of Tpm isoforms. They have been shown to critically affect a wide range of actomyosin-dependent processes including endo-and exocytosis events, the formation and dynamics of cell-surface extensions, rigidity sensing, cell migration, cytokinesis, nuclear division, anchorage-dependent growth, and embryogenesis (5)(6)(7). Additionally, changes in the abundance of specific cytoskeletal Tpm isoforms are commonly observed in cells that are undergoing transformation (8). In skeletal and cardiac muscle, the troponin-induced azimuthal translocation of Tpm across the F-actin surface, which controls access of the myosin-binding sites on F-actin, is favored by the low energy cost of the translocation between the A and M states (9 -11) and depends critically on tight control of the sarcomeric Ca 2ϩ concentration as well as the short duration of strong actin-bound myosin states relative to the overall duration of the reaction cycle (12). Based on a fairly complete analysis of the chemomechanical properties for most of the common myosin isoforms (13)(14)(15)(16) and reports showing that cooperative units extend beyond single Tpm dimer boundaries (17,18), it appears unlikely that cytoskeletal and smooth muscle Tpm isoforms play a similar regulatory role as gatekeeper of the myosin-bind-ing sites on actin. Head-to-tail interactions and the formation of a continuous cable contribute critically to the extension of cooperative units and enhance the interaction of all Tpm isoforms with the filamentous actin (F-actin) substrate (19). Continuous Tpm cables bind to F-actin with ϳ1000-fold higher affinity than individual Tpm molecules (20). Structural studies show how the C-terminal coiled-coil region of the Tpm dimer opens over the last 11-15 amino acids, whereas the N-terminal coiled-coil inserts into the resulting cleft (21)(22)(23)(24). In the case of muscle Tpm isoforms, N-acetylation of the N-terminal methionine contributes critically to the formation and stabilization of a stable overlap complex. In contrast, the ability of Tpm isoforms to interact with F-actin appears to be less critically affected by N-terminal acetylation. Thus, the effect of a lack of N-acetylation of the skeletal muscle isoform Tpm1.1 on actin binding is compensated by the replacement of exons 1a and 2b with the N-terminal exon 1b that is present in cytoskeletal isoforms Tpm1.8, Tpm1.12, or Tpm3.1 (25). Moreover, subtle changes near the N terminus appear to have a critical effect on head-to-tail interactions. The ability of muscle Tpm isoforms to polymerize and efficiently bind to F-actin has been reported to be enhanced by the addition of a single glycine residue, an Ala-Ser dipeptide, or a Gly-Ala-Ser tripeptide to their N termini (24,26,27). Exon 9-encoded sequences are another major determinant in regard to head-to-tail interactions. Replacement of the striated muscle specific exon 9a encoded C terminus with exon 9d, which is found in smooth muscle and cytoskeletal Tpm isoforms, allows nonacetylated hybrid Tpm to efficiently bind to F-actin (28).
Nonmuscle myosin-2 (NM-2) isoforms are among the most prominent members of the myosin family that associate with Tpm-enveloped cytoskeletal F-actin (29,30). In mammals there are three genes (MYH9, MYH10, and MYH14) encoding NM-2A, NM-2B, and NM-2C heavy chains, respectively. Socalled NM-2 monomers consist of two heavy chains, each associated with an essential and a regulatory myosin light chain. Phosphorylation of the regulatory light chain is required for the monomers to unfold from an autoinhibitory conformation, acquire motor activity, and gain the ability to form bipolar thick filaments that typically contain 20 -30 monomers (31,32). At least in the case of NM-2A and NM-2B, both activated monomers and bipolar thick filaments appear to contribute to distinct cellular functions (33). Advances in the structural characterization of large protein complexes by electron cryomicroscopy have contributed new and important insights into actin-Tpm-myosin complexes (34). Structures resolved to subnanometer scale show that Tpm forms a 210 Å 2 interaction surface with F-actin and a 300 Å 2 interaction surface with the myosin motor domain (9,35). Based on these results, it is possible to begin to relate known differences in their interactions (4) to the structural features of individual myosin and Tpm isoforms. Cytoskeletal Tpm isoforms specify functionally distinct F-actin populations acting as selectivity filters that modulate allosteric coupling and mediate the isoform-specific recruitment of myosins in the context of stress fibers, transverse arcs, and other actin-based superstructures (37,38). The modulation of myosin motor activity by cytoskeletal Tpm isoforms is not restricted to changes in actin affinity but extends to the kinetics of individual steps in the ATP hydrolysis cycle (39 -42). The nature of the M-Tpm contact areas suggests that myosin can greatly enhance Tpm binding to F-actin (10), as was previously shown for a series of Escherichia coli expressed variants of rat ␣-tropomyosin (25).
In this study, we characterized the kinetic and motor properties of complexes composed of filamentous ␤-actin, NM-2B, and either Tpm1.8cy (b.-.b.d), Tpm1.12br (b.-.b.c), and Tpm3.1cy (b.-.a.d). Tpm3.1 is encoded by TPM3 and contains three regions that differ from the corresponding regions in Tpm1.8 and Tpm1.12. In addition to the changes introduced by the use of TPM3 exon 6a, differences occur in the regions encoded by exons 1 and 9. The regions encoded by exons 1b and 6a contain 5 and 4 charge changes compared with Tpm1.8 and Tpm1.12, respectively. Exon 9d contains 3 charge changes compared with Tpm1.8 and 11 charge changes compared with Tpm1.12. The program Paircoil2 (43) predicts that only the C-terminal ends of Tpm1.8 and Tpm1.12 are affected by differences in their amino acid sequence. In the case of Tpm3.1, Pair-coil2 indicates a slightly reduced likelihood of coiled-coil formation in the region from residues Val 134 to Cys 170 . The organization of the C-terminal end of Tpm3.1 is predicted to more closely resemble Tpm1.8.

NM-2B and Tpm isoforms in neuroblastoma cells
NM-2B is the dominant NM-2 isoform in neuronal cells (44). In differentiated SK-N-BE(2) neuroblastoma cells, NM-2B is widely distributed but somewhat enriched within the cell body and in axonal projections. The abundance and localization of specific Tpm isoforms in neuroblastoma cells can be assessed using antibodies that recognize peptides encoded by exons 1b, 9c, and 9d of TPM1 and exon 9d of the TPM3 (1). Each of these antibodies recognizes multiple Tpm isoforms as summarized in Table 1. Using this approach, Gunning and co-workers (45) revealed the presence of Tpm1.10, Tpm1.11, Tpm1.12, and Tpm4.2 in neuroblastoma cells. Similar to most tumor cells, neuroblastoma cells do also produce Tpm3.1. Our results show that in addition, Tpm1.6, Tpm1.7, Tpm1.8, and Tpm3.2 are produced in these cells (Fig. 1A). Immunofluorescence staining revealed partially overlapping localization patterns for the different Tpm isoforms and strong colocalization with NM-2B (Fig. 1B). Stress fibers that are prominent before the induction of differentiation are difficult to detect at the final stage but can be identified in cells with a neuroepithelial phenotype after 2 days of differentiation (Fig. 1C). Any observed differences between Tpm1.8 and Tpm1.12 are caused by differences in amino acid sequence occurring in the C-terminal region, which is encoded by exons 9d and 9c, respectively (Fig. 1D). Tpm1.8 carries a C-terminal Asn-Asn-Met tripeptide extension. The preceding 23 residues differ in 18 positions, with 15 changes corresponding to non-conservative substitutions. Tpm3.1 is encoded by Tpm3 and differs in 58 and 62 residues from Tpm1.8 and Tpm1.12, respectively. Clusters of non-conservative substitutions are located in the region between residues 155 and 175 and near to the C terminus. In addition, 18 substitu-

Tropomyosin-mediated actin-Tpm-myosin diversification
tions occur within the 50 N-terminal residues, including 5 that lead to charge changes.

Cooperative binding of Tpm and NM-2B to filamentous ␤-actin
We performed cosedimentation assays to determine binding affinities for the interaction between filamentous ␤-actin and individual Tpm isoforms. Our results show that in the absence of NM-2B, Tpm1.12 binds filamentous ␤-actin with much lower affinity than Tpm1.8 and Tpm3.1. The corresponding K 50% values correspond to Ն40 M (Tpm1.12), 1.6 Ϯ 0.2 M (Tpm3.1), and 0.10 Ϯ 0.03 M (Tpm1.8) ( Fig. 2A). Similar results were described in previous reports performed with ␣-actin (25,46). Specifically, Moraczewska et al. (25) showed that chicken skeletal muscle myosin subfragment-1 greatly increases the affinity of Tpm1.12 and other cytoskeletal Tpm variants for F-actin to biologically significant levels. The rate of NM-2B binding to filamentous ␤-actin was assessed by monitoring the transient increase in light scattering upon mixing of the single-headed construct NM-2B0 (47) with either bare or Tpm-enveloped F-actin. In each case, the transients followed a single-exponential function. A secondary plot of the observed rate constants against the concentration of bare or Tpm-enveloped F-actin revealed a linear dependence for each combination of proteins (Fig. 2C). This is compatible with a simple one-step binding process, where k obs ϭ [A] k ϩA ϩ k ϪA (48). The value of the second-order association rate constant k ϩA is given by the slope of the plot. Decoration with Tpm1.12 led to a 20% reduction in k ϩA, whereas an ϳ50% increase in k ϩA was observed with Tpm1.8 and Tpm3.1. The y intercept was used to estimate the first-order dissociation rate constant k ϪA .
The affinity of NM-2B0 for filamentous ␤-actin (K A ) was calculated from the ratio k ϪA /k ϩA. The results are summarized in Table 2.

Modulation of the actin-activated ATPase activity and duty ratio of NM-2B by Tpm isoforms
The actin-activated Mg 2ϩ ATPase activity of myosin light chain kinase (MLCK)-treated NM-2B-HMM was measured with bare-and Tpm-enveloped filamentous ␤-actin. Decoration with Tpm1.8 or Tpm3.1 increases the maximum actinactivated ATPase activity (k cat ) of NM-2B-HMM 1.3-and 1.6fold, respectively, whereas Tpm1.12 reduces k cat 3.6-fold (Fig.  3A). The actin concentration required for half-maximal activation (K app ) is significantly reduced by all three Tpm isoforms ( Table 2). It should be noted that the individual values for k cat and K app are only estimates and must be treated with some caution. Reliable readings were obtained only up to 50 M enveloped F-actin, which is in the order of K app . In contrast, the values shown in Table 2 for the apparent second-order rate constant k cat /K app are well defined by the initial slope of the ATPase activity versus [F-actin] plot. They reflect the behavior of the fully activated complex and are a measure of the coupling efficiency between the actin and nucleotide-binding sites of myosin (49). The coupling efficiency increases 2.9fold in the presence of Tpm1.8 and 2.5-fold in the presence of Tpm3.1 but is 14.3% reduced in the presence of Tpm1.12 ( Fig. 3A and Table 2).
To determine the influence of Tpm isoforms on individual steps of the NM-2B ATP-turnover cycle, we performed a series of stopped-flow experiments. In a first step, we measured the effect of Tpm decoration on the rate of P i release from acto-NM-2B-HMM. Compared with bare F-actin (k obs ϭ 0.078 Ϯ 0.02 s Ϫ1 ), we observed ϳ5.3and 240-fold acceleration of the release of inorganic phosphate (P i ) from the active site of NM-2B in the presence of Tpm1.8 (0.41 Ϯ 0.09 s Ϫ1 ) and Tpm3.1 (18.7 Ϯ 2.2 s Ϫ1 ), respectively. In contrast, Tpm1.12 only marginally affected the P i release rate of the NM-2B construct (0.13 Ϯ 0.03 s Ϫ1 ) ( Fig. 3B and Table 2).
Next, we evaluated the rate of ADP release from acto-NM-2B-HMM using 2Ј-deoxy-3Ј-mant-ADP (dmADP) (50). This approach exploits the decrease of mant-fluorescence that occurs when dmADP is displaced from acto-NM-2B-HMM in the presence of excess ADP. The ensuing transients display biphasic behavior. Sellers and co-workers (51) assigned the fast

Tropomyosin-mediated actin-Tpm-myosin diversification
Tropomyosin-mediated actin-Tpm-myosin diversification   Table 2. The occupancy of ␤-actin with Tpm1.12 in this experiment (preincubation in the absence of NM-2B-HMM) is in the range of ϳ30%; thus the observed reduction of the actin binding rate can be assumed to be more pronounced under cellular conditions. The data are expressed as the means Ϯ S.E. The data shown in A and B were obtained from four independent measurements. The data shown in C are the average of six measurements and were reproduced at least once with a different batch of protein.   3C and Table 2). Both the Tpm-induced acceleration in the rate of phosphate and the slower rate of ADP release contribute to an increase in the duty ratio, the fraction of time that NM-2B spends in strongly F-actin bound states during ATP turnover (52).

Tpm-mediated changes in the motor activity and processivity of NM-2B
To analyze the Tpm-mediated changes of NM-2B motor activity directly, we performed in vitro motility assays. In the standard in vitro motility assay first described by Kron and Spudich (53), robust, unidirectional movement was observed. Compared with the situation with bare F-actin, the average unloaded velocity measured in this assay was 7, 28, and 35% slower in the presence of Tpm1.8, Tpm1.12, and Tpm3.1, respectively (Table 2).
To assess the extent to which Tpm-mediated changes of the duty ratio affect the motor properties of NM-2B, we performed landing assays. Actin filament recruitment was measured at varying NM-2B-HMM surface densities. The data were fitted using equation 1 (see "Experimental procedures"), where n approximates the minimal number of myosin molecules required to bind and propel actin filaments. Values of n close to 1 imply that only one motor molecule is required for processive movement along actin filaments. The measured values for n decrease from 5.0 Ϯ 1.2 for bare-actin to 0.9 Ϯ 0.1, 1.5 Ϯ 0.4, and 1.0 Ϯ 0.2 for Tpm1.8-, Tpm1.12-, and Tpm3.1-enveloped F-actin, respectively. Thus, the duty ratio of NM-2B increases from ϳ20% with bare F-actin to 67% with Tpm1.12 and close to 100% with both Tpm1.8 and Tpm3.1. Compared with the situation with bare F-actin, the presence of each of the Tpm iso-

Tropomyosin-mediated actin-Tpm-myosin diversification
forms led to a large increase in the landing rate at low NM-2B surface densities. In the presence of Tpm1.8 or Tpm3.1, the number of observed landing events was also significantly increased at NM-2B surface densities greater than 1000 molecules/m 2 . In contrast, Tpm1.12 caused a large drop in the landing rate at high NM-2B surface densities (Fig. 4A). Landing events were further analyzed to determine how the myosin density affects the duration of interactions between F-actin and NM-2B. Fig. 4B shows that Tpm3.1 and Tpm1.8 increase the cooperativity of NM-2B binding to F-actin but reduce the duration of the interactions. Tpm1.12-enveloped F-actin has the opposite effect. The cooperativity of NM-2B binding to Tpm1.12-enveloped F-actin is reduced, but once NM-2B has bound Tpm1.12-enveloped F-actin, it remains associated for significantly longer periods.

Tpm decoration affects the run-length of NM-2B on ␤-actin filaments
To directly evaluate the processivity of NM-2B on bare and Tpm containing ␤-actin filaments, we performed single-molecule motility assays using a quantum dot (Qdot)-labeled NM-2B-HMM-construct (Fig. 5A). The NM-2B-HMM construct displays short processive runs on bare actin filaments, with an average run distance of 209 Ϯ 37 nm (n ϭ 108). Tpm1.8-and Tpm3.1-enveloped F-actin supported an average run length of 650 Ϯ 61 nm (n ϭ 110) and 384 Ϯ 42 nm (n ϭ 112), respectively (Fig. 5, B and C). Assuming a step size of ϳ5.5 nm, the longest processive runs recorded with Tpm1.8-enveloped F-actin correspond to nearly 500 consecutive steps. In the presence of Tpm3.1, we observed run lengths consistent with more than 300 consecutive steps. The velocity of the Qdot-labeled NM-2B-HMM construct showed significant increases by ϳ30% in the presence of Tpm1.8 and Tpm3.1 (Fig. 5D). The frequency of productive runs increased 3-fold in the presence of Tpm1.8 and 2-fold for Tpm3.1. No processive runs were observed on Tpm1.12-enveloped filaments (Fig. 5E). Differences in motor activity observed between the standard and single molecule in vitro motility assays reflect changes in the number of interacting myosin heads, drag force, and mode of surface attachment.

Discussion
Cell biological, structural, and biochemical studies lend support to the notion that Tpm isoforms contribute or even play an essential role in specifying the functional properties of cytoskeletal actin filament populations (9,30,41,54). Our results show how different Tpm isoforms that coexist within the same cell type affect myosin motor activity. The fact that cytoskeletal actin filaments are mostly associated with tropomyosin cofilaments stresses the need for a conceptual change in the cytoskeletal actomyosin field. Namely, it requires a shift away from a generic view of the actin filament and toward a greater consideration for the properties of distinct cellular actin filament populations. Our results show the extent to which product release steps and the duty ratio of the myosin motor are affected by the association of F-actin with different Tpm isoforms. Dependent on the composition of the actin-Tpm cofilaments, the same cytoskeletal myosin motor is able to switch function from efficient transporter to tension holder, force sensor, or processive signal transducer.
NM-2B displays slow actin-activated ATP turnover and sliding velocity (51,55). During basal ATP turnover, the release of the hydrolysis product P i is the rate-limiting step of the NM-2B ATPase cycle. Similar to most myosins, P i release is greatly accelerated by F-actin binding. However, unlike most myosins, binding to F-actin has been reported to slow the release of the second hydrolysis product ADP from the active site of NM-2B (51,56). This type of negative coupling between actin binding and ADP release has also been observed with human myosin-7A, but is uncommon for other myosins (57). The force generating interaction of NM-2B with bare F-actin creates a situation where the release rates for both hydrolysis products become similar and contribute to limiting the rate of the overall cycle.
Our results show that all three Tpm isoforms bind NM-2Bdecorated F-actin with affinities that are of physiological significance and affect the balance between P i and ADP release from the active site of NM-2B in a specific manner. The associated changes in dominant rate-limiting steps have a direct effect on the duty ratio, because P i release gates the transition from weak to strong F-actin binding states, and ADP release does the

Tropomyosin-mediated actin-Tpm-myosin diversification
opposite. The duty ratio is increased ϳ3-fold in the presence of Tpm1.12 and 5-fold for both Tpm1.8 and Tpm3.1. In regard to their effect on the time required per ATP turnover, the presence of Tpm1.12 extends the cycle time 3.7-fold, whereas it is shortened by 27 and 63% in the presence of Tpm 1.8 and Tpm3.1, respectively.
Tpm-induced changes in the second-order rate constant for NM-2B binding to F-actin and the first-order rate constant for dissociation from F-actin result in 2.9-and 2.2-fold greater affinity for F-actin in the presence of Tpm1.8 and Tpm1.12. The largest changes were observed in the presence of Tpm3.1, which resulted in a 9-fold greater affinity for F-actin. The com-bined effects of the observed changes in turnover rate, duty ratio, and actin affinity brought about by each of the Tpm isoforms can be probed using in vitro assays that monitor the functional competence of the myosin motor. Previously, Rock and co-workers (58) used a three-bead optical trapping assay to show that NM-2B dimers bind to filamentous ␣-actin and cause multiple displacements before detachment. In contrast, Sellers and co-workers (55) reported that only short bipolar NM-2B filaments move processively on filamentous ␣-actin. This apparent difference in processive behavior may be explained by the force-dependent ADP release kinetics described by both the Sellers and Rock groups (36,69). It is also compatible with the

Tropomyosin-mediated actin-Tpm-myosin diversification
results obtained in the landing assays, where NM-2B moving against zero loads on bare actin appears to require multiple molecules for continuous motion, whereas the increased Stokes drag associated with the attachment of an 18-nm Qdot to the C terminus of NM-2B-HMM appears to facilitate processive movement in our single myosin molecular experiments. A Qdot-mediated increase in Stokes drag with the associated decrease in diffusion rate along DNA has previously been reported for the motion of Qdot-labeled DNA N-glycosylases belonging to the helix-hairpin-helix and Fpg/Nei families (59). As expected from our transient kinetics results, NM-2B performs longer runs in the TIRF single-molecule motility assay on Tpm1.8 -or Tpm3.1-␤-actin cofilaments than on bare filaments. Interactions of NM-2B with Tpm1.12-␤-actin cofilaments lasted typically three times longer but did not result in active translocation of the motor along the actin filament. Our kinetic data do not predict that at NM-2B surface densities above 1500 molecules/m 2 , the number of landing events is lower in the presence of Tpm1.12 than with bare F-actin. In contrast to the two other Tpm isoforms, the association of Tpm1.12 with ␤-actin does not increase the cooperativity of NM-2B-binding to F-actin but rather decreases it. The major effect of Tpm1.12 appears to be that it gears the activity of the NM-2B motor toward tension-bearing rather than contractile functions. NM-2B is less likely to bind Tpm1.12-␤-actin cofilaments in the crowded environment of actin arcs or stress fibers. However, once it binds, it remains strongly attached for longer periods. The opposite is true for Tpm 1.8 and Tpm3.1. Decoration of ␤-actin filaments with one of these Tpm isoforms increases the motile activity of NM-2B, its frequency of productive interactions, and ability to take multiple steps against an external force. The specific changes brought about by association with Tpm 1.8 or Tpm3.1 extend the functional range of NM-2B, within both the contexts of bipolar thick filaments and activated NM-2B monomers (33).
In summary, our work shows the extent and scope of the functional changes that are introduced by the decoration of cytoskeletal actomyosin complexes with different cytoskeletal tropomyosin isoforms. Future challenges include the elucidation of the composition of additional actin-Tpm-myosin complexes, their involvement in specific cellular tasks, and how their function and structural integrity are affected by posttranslational modifications.

Cell lysis and immunoblot analysis
SK-N-BE(2) cells differentiated with 10 M all-trans-retinoic acid for a period of 6 days were lysed in SDS sample buffer at a concentration of 2 ϫ 10 4 cell/l. Lysates were separated on 12% acrylamide gels by SDS-PAGE. Samples were transferred onto a nitrocellulose membrane, blocked for 1 h at room temperature in 5% skim milk powder and labeled overnight with the following Tpm-specific sheep polyclonal antibodies: ␥/9d (Merck Millipore, catalog no. AB5447), ␣/1b (Merck Millipore, catalog no. ABC499), and ␣/9d (Merck Millipore, catalog no. AB5441) (1). A donkey anti-sheep IgG-HRP secondary antibody (Santa Cruz Biotechnology, catalog no. sc-2473) was applied for 1 h at room temperature before HRP activity was developed using the SuperSignal TM West Femto maximum sensitivity substrate (Thermo Fisher, catalog no. 34095). Membranes were imaged with a Bio-Rad ChemiDoc TM MP system using Image La TM software (Bio-Rad).

Cloning, expression, and protein purification
To generate the expression construct for the heavy chain of the double-headed NM-2B-HMM construct with C-terminal octahistidine and Avi tags, the DNA sequence that encodes amino acids 1-1344 was PCR-amplified using human cDNA as a template. The primers used were 5Ј-CGCGCGGCGCGCAT-GGCGCAGAGAACTGGACTCGA-3Ј and 5Ј-CAGCTGGT-CGACGTGGTGATGATGATGATGATGATGCTCCTCTT-CCAGCTGCCGGAT-3Ј containing restriction sites for BssHI and SalI, respectively. The pFast Bac TM vector containing the NM-2B heavy chain sequence and the pFastBac TM Dual vector containing the sequences for MYL6 and MYL12b were cotransformed for protein production in SF9 cells. Vectors and cells were obtained from Thermo Fisher Scientific. The generation of the expression vector encoding the motor domain of human NM-2B0 (residues 1-782), two Dictyostelium discoideum ␣-actinin repeats, and a C-terminal octahistidine tag was described elsewhere (47). Sf9 cells were transfected with the recombinant bacmids using the FuGENE HD transfection reagent (Promega). To purify the free MYL6 and MYL12b light chains, the genes were cloned into the pGEX expression vector (GE Healthcare Life Sciences) for protein production in E. coli.
Expression constructs for rat Tpm1.

Tropomyosin-mediated actin-Tpm-myosin diversification
for expression in E. coli strain Rosetta TM (Merck). We refer to the different tropomyosin isoforms in the text mostly by their short name in accordance with a recently introduced systematic nomenclature (60). Full formal protein names are used when we refer to differences in exon usage between the isoforms. Alternate names and changes in exon usage are listed in Table 1. Moreover, we refer to the unassembled double-stranded coiled-coil protein as dimer. Cytoskeletal Tpm isoforms form predominantly homopolymers of Tpm homodimers. The extensions "br" and "cy" in the formal protein name indicate prominent but not exclusive associations with brain tissue and cytoskeletal structures. Tpm1 or Tpm3 specifies the protein is encoded by TPM1 or TPM3. The fourletter code extension of the formal protein name indicates splice form usage for the variable exons 1, 2, 6, and 9. All three isoforms lack the region encoded by exon 2 (60). The short names Tpm1.8, Tpm1.12, and Tpm3.1 are used throughout the text, unless we refer to the splicing of the four exons that vary in vertebrates.
Purification of chicken skeletal muscle ␣-actin (61) and recombinant human ␤-actin was performed as described previously (62). Pellets of Sf9 cells producing NM-2B0 motor domain or HMM constructs were suspended in lysis buffer (50 mM HEPES, pH 7.5, 200 mM NaCl, 10 mM ␤-mercaptethanol, 4 mM MgCl 2 , 4 mM ATP, 1 mM EGTA, 0.5% Triton X-100, and cOmplete TM protease inhibitor mixture (Merck). The cell suspension was sonicated four times for 1 min at 40% power and 50% duty cycle (Bandelin Sonoplus, Berlin, Germany). The lysate was centrifuged at 100,000 ϫ g for 1 h at 4°C. The supernatant was loaded onto a Ni 2ϩ -nitrilotriacetic acid affinity column (Qiagen) equilibrated with lysis buffer. The column material was washed with 10 column volumes lysis buffer and 10 column volumes wash buffer (25 mM HEPES, pH 7.3, 200 mM NaCl, 0.5 mM EGTA, 3 mm MgCl 2 , 65 mM imidazole). The protein was eluted with 6 column volumes elution buffer (25 mM HEPES, pH 7.3, 200 mM NaCl, 1 mM EGTA, 1 mM EDTA, 1 mM DTT, and 200 mM imidazole). In addition, the free forms of human recombinant ELC (essential myosin light chain) (MYL6) and RLC (regulatory myosin light chain) (MYL12b) constructs were produced in E. coli. The purified light chains were added at 1 M concentration to the elution buffer used for the purification of the HMM construct. Individual fractions were analyzed by SDS-PAGE, concentrated, and dialyzed overnight against dialysis buffer (25 mM HEPES, pH 7.3, 200 mM NaCl, 1 mM EGTA, 1 mM EDTA, 1 mM DTT, and 3% trehalose). Dialyzed protein was applied to a HiLoad Superdex 200 prep grade 16/60 gel filtration column (GE Healthcare Life Sciences) equilibrated with dialysis buffer. The resulting protein fractions were concentrated to a final concentration of ϳ6 mg ml Ϫ1 . The purified protein was supplemented with 7% (w/v) trehalose, flash frozen in liquid nitrogen, and stored at Ϫ80°C.
Tpm production was induced by the addition of 1 mM isopropyl ␤-D-thiogalactopyranoside, followed by incubation with vigorous shaking for 3-4 h at 37°C. In addition to human Tpm3.1, we produced rat isoforms Tpm1.8 and Tpm1.12. Changes between rat and human Tpm1.8 and Tpm1.12 occur at S30T, Q37H, and K184R. In the case of Tpm1.12, a fourth change corresponds to H226Q. All changes occur in the c posi-tion of the repeating heptad repeat pattern (63). Residues in the c position are located at the outside of the coiled-coil and contribute to stability through ␣-helical propensity (64). The program Paircoil2 (43) predicts the parallel coiled-coil fold of Tpm1.8 and Tpm1.12 to be unchanged by these species specific substitutions. The recombinant, tag-free Tpm constructs were purified by ion exchange chromatography as described by Coulton et al. (65) with minor modifications. Purified Tpm was stored in 5 mM potassium phosphate, pH 7.0, 100 mM NaCl, and 5 mM MgCl 2 . Protein concentration was determined using the Bradford assay and protein absorbance readings at 276 nm. Measurements were performed with a UV-2600 spectrophotometer (Shimadzu, Duisburg, Germany). Recombinant myosin light chain kinase was obtained from Sigma-Aldrich.

Binding assays
The affinity of Tpm1.8, Tpm1.12, and Tpm3.1 for filamentous ␤-actin was measured and analyzed using a cosedimentation assay as described by Coulton et al. (65). 10 M ␤-actin was preincubated with varying Tpm concentrations (0 -10 M) for 30 min at room temperature. Assay buffer contained 20 mM MOPS, pH 7.0, 5 mM MgCl 2 , and 100 mM KCl. NM-2B-induced Tpm binding to ␤-actin was monitored following the cosedimentation protocol described by Moraczewska et al. (25). The 6:1 acto-Tpm complex was mixed with increasing concentrations of NM-2B (0 -16 M), incubated for 30 min in assay buffer (20 mM MOPS, pH 7.0, 5 mM MgCl 2 , and 50 mM KCl), and pelleted for 20 min at 100,000 ϫ g. To increase the sensitivity and linearity of the cosedimentation assays, supernatant and pellet fractions were treated with Instant-Bands sample loading buffer (EZBiolab Inc., Carmel, IN) and resolved by SDS-PAGE. Quantification of band intensities was performed using densitometry on a Bio-Rad ChemiDoc TM MP system (Bio-Rad). The images were analyzed with the Fiji release of ImageJ software version 1.49s (59).

Kinetic measurements
NM-2B-HMM was phosphorylated immediately prior to use at 30°C for 30 min. NM-2B-HMM was incubated with myosin light chain kinase at a stoichiometric ratio of 20:1 in a reaction mixture containing 2 M RLC, 20 mM MOPS, pH 7.0, 50 mM KCl, 2 mM MgCl 2 , 1 mM CaCl 2 , 0.15 mM EGTA, 0.2 M calmodulin, 2 mM DTT, and 1 mM ATP. The single-headed NM-2B0 construct does not require activation and was used without prior phosphorylation.
Steady-state ATPase rates were measured using an enzymebased assay, linking ATP hydrolysis to the oxidation of NADH. The change of absorbance was followed over 30 min at 25°C in assay buffer containing 25 mM HEPES, pH 7.4, 5 mM MgCl 2 , and 50 mM KCl. Increasing actin concentrations (0 -50 M) were preincubated with a 3-fold molar (18-fold stoichiometric) excess of Tpm for 30 min at 23°C. Values for the maximum actin-activated ATPase activity (k cat ) and the actin concentration required for half-maximal activation (K app ), and the apparent second-order rate constant for F-actin binding (k cat /K app ) were determined as described previously (57).
Transient kinetic experiments were performed at 20°C in either a HiTech Scientific SF-SF-61 SX or SF-61 DX stopped-Tropomyosin-mediated actin-Tpm-myosin diversification flow system. Both systems are equipped with a 75 W mercuryxenon arc lamp. The assay buffer used contained 20 mM MOPS, pH 7.0, 100 mM KCl, and 5 mM MgCl 2 . In the case of Tpm-enveloped actin filaments, the complex of actin and Tpm was preincubated for at least 30 min at 23°C.
The binding of NM-2B to actin was measured by following the increase in light scattering at 405 nm occurring upon mixing 4 M NM-2B with 0.5-4 M actin or actin-Tpm complex. The P i release assay was performed as described previously (66,67). First, 2 M NM-2B was mixed with 1 M ATP. This was followed by incubation for 15 s in a delay line to allow ATP binding and hydrolysis to occur. In the second mixing step, the release of phosphate was triggered by the addition of 40 M actin or actin-Tpm, where Tpm constructs were added in a 3-fold molar excess over actin. Binding of P i to N-[2-(1-maleimidyl)ethyl]-7-(diethylamino)coumarin-3-carboxamide-labeled phosphate-binding protein was monitored.
The rate of ADP release from the HMM construct was determined as described by Sellers and co-workers (51). The decrease in dmADP fluorescence upon mixing acto-NM-2B-HMM complexes with a large excess of unlabeled ADP in the presence or absence of Tpm was measured. Premix concentrations in the stopped-flow apparatus correspond to 0.4 M myosin heads, 10 M dmADP, 2 M actin, 2 mM ADP, and either 0 or 20 M Tpm construct.
Kinetic Studio software (TgK Scientific, Bradford on Avon, UK) was used for initial data inspection and analysis. Detailed data analysis was performed with OriginPro (Northampton, MA) 9.0G graphing and data analysis software. Goodness-of-fit criteria were evaluated using the coefficient of determination R 2 and 2 tests as implemented in OriginPro 9.0G. Each data point corresponds to the average of 3-6 single measurements and was verified at least once with protein from different preparations. The error bars indicate the standard deviation.

Assays for actin sliding movement
Unloaded actin sliding motility was measured at 30°C as described previously (53). HMM was directly bound to the nitrocellulose-coated surface. Actin filament velocity was determined with the help of the program DiaTrack 3.01 (Semasopht, Switzerland). Data analysis was performed with Origin 9.0G (OriginLab). Goodness-of-fit criteria were evaluated using the coefficient of determination R 2 and 2 tests as implemented in Origin 9.0G.
Landing assays were performed as described previously (68) with the following modifications: NM-2B-HMM molecules were directly immobilized on nitrocellulose-coated coverslips to obtain surface densities between 200 and 8,000 myosin molecules/m 2 (69). The assay was started by the addition of TRITC-phalloidin-labeled actin (20 nM, with and without decoration with Tpm) to the motility buffer containing 4 mM Mg 2ϩ -ATP. Landing events were recorded in TIRF mode using a customized inverted microscope (Olympus IX81) equipped with a 532 nm diode laser (Novalux, Sunnyvale, CA) and fitted with a 60ϫ/1.49 NA oil immersion lens (ApoN, Olympus). The landing rate was measured by counting the number of actin filaments that landed and moved Ն0.3 m in an observation area of ϳ7,100 m 2 . The density of myosin motors on the assay surface (molecules/m 2 ) is plotted against the observed landing rate (mm Ϫ2 s Ϫ1 ). Landing rates were best fit to Equation 1 according to the model by Hancock and Howard (68).
Here, Z is a parameter that incorporates collision of actin with the surface, and 0 corresponds to the surface area over which motors interact with actin. The duty ratio is given by the slope n, which can be obtained from the graph. At n ϭ 1, only one motor is required to propel an actin filament forward.

Determination of processive run length
To investigate the run lengths of NM-2B motors, biotinylated NM-2B-HMM was mixed with a 10 -20-fold molar excess of 655-nm streptavidin-coated Qdots (Invitrogen), which ensures that the majority of motile Qdots are bound to a single NM-2B-HMM. Biotinylation of the C-terminal AviTag of NM-2B-HMM was performed, along with RLC phosphorylation. Following the addition of 50 M biotin, 0.6 M BirA, and 0.005 M myosin light chain kinase to 1 M NM-2B-HMM (final concentrations) in buffer containing 25 mM imidazole, 1 mM CaCl 2 , 0.15 mM EGTA, 0.2 M calmodulin, 1 mM DTT, and 1 mM ATP, the reaction mixture was incubated at 30°C for 30 min. Flow cells made from glass coverslips were prepared by introducing the following solutions into the flow cell: 0.05 mg/ml anti-His antibody (5 min incubation) and 0.2 mg/ml N-ethylmaleimide-modified D. discoideum myosin-2 motor domain construct with artificial lever arm and C-terminal His tag (70). Incubation for 2 min was followed by a wash with buffer containing 1 mg/ml BSA and 1 mM DTT, incubation with rhodamine-phalloidin-labeled ␤-actin filaments with or without Tpm (2-5 min), and a wash with motility assay buffer. Finally, we added 0.1 M NM-2B with 1 or 2 M Qdots in motility buffer containing 4 mM Mg-ATP and an oxygen-scavenging system (36). For experiments with actin-Tpm, a very large excess of Tpm (10 M) was included in the motility buffer to prevent dissociation from F-actin. TIRF fluorescence microscopy was performed at room temperature (22°C) using a Nikon Eclipse Ti spinning disc microscope equipped with a 100ϫ TIRF Apo objective lens (1.49 NA). Qdots and TRITCphalloidin-labeled actin were excited with the 488-and 561-nm laser lines, respectively. Images were acquired simultaneously using a two-camera adaptor and Andor iXon Ultra EMCCD cameras. The pixel resolution was 76.0 nm, and data were collected at 60 frames/min. Qdot movement along actin filaments was tracked manually using ImageJ. For each event, we required Qdot-labeled NM-2B-HMM to move continuously for at least five frames to qualify as a run. Runs that artificially terminated by running off the end of an actin filament and runs shorter than 2 pixels were not included in the runlength analysis.