|Home | About | Journals | Submit | Contact Us | Français|
The swinging crossbridge hypothesis states that energy from ATP hydrolysis is transduced to mechanical movement of the myosin head while bound to actin. The light chain-binding region of myosin is thought to act as a lever arm that amplifies movements near the catalytic site. This model has been challenged by findings that myosin VI takes larger steps along actin filaments than early interpretations of its structure seem to allow. We now know that myosin VI does indeed operate by an unusual ~ 180° lever arm swing and achieves its large step size using special structural features in its tail domain.
Molecular motors have key roles in virtually all cell biology processes. The precise, dynamic organization of cells and tissues depends crucially on these marvellous molecular machines. It is no wonder then that practically every cell has nearly 100 different molecular motors to carry out specific pivotal tasks. And, given the unique demands on the biochemical and biophysical properties of molecular motors to carry out their functions, they have given us a great deal of information about how enzymes work in general.
One such family of molecular motors is the myosin family. Myosins are ATPases that use the chemical energy derived from ATP hydrolysis to produce mechanical work. In 1969, H.E. Huxley1 proposed the swinging crossbridge hypothesis to explain how this energy transduction might occur. According to current versions of this model, following ATP hydrolysis the myosin head domain (the crossbridge between the thick filament of myosin molecules and the thin actin filaments) undergoes a conformational transition into a pre-stroke state. On rebinding to actin and releasing phosphate, the ADP-bound myosin head undergoes a transition from a weak to a strong actin-binding state, which is accompanied by a reverse conformational change to a post-stroke state, resulting in a sliding motion at the actin–myosin interface. The myosin remains tightly bound to actin until ADP is released, at which point ATP rapidly binds and causes release of the myosin from actin. The fraction of the ATPase cycle time that a myosin spends strongly bound to an actin filament is known as the duty ratio. The strongly bound state time determines the maximum velocity of relative movement of myosin along an actin filament. The head cannot move forwards any faster than it can let go.
Crystal structures of several myosins2–5 reveal that they are all composed of a catalytic head that binds to actin and nucleotides, with a converter domain distal to the actin-binding site (FIG. 1). The first myosin crystal structure identified a light chain-binding domain, which extends out from the myosin catalytic domain like a lever arm6. It has been assumed that for all myosins this light chain-binding region serves as a lever arm to amplify movements of the converter domain, which transitions between pre-stroke and post-stroke configurations (FIG. 1; light chains not shown). This transition is proposed to provide a mechanical stroke in a particular direction along the actin filament, which is a polar structure with barbed (also known as plus) and pointed (also known as minus) ends. Thus, the swinging crossbridge hypothesis was renamed the swinging lever arm hypothesis. What follows the light chain-binding domain varies between different myosins. Some myosins have a coiled-coil domain in this region, which is responsible for dimerization. Many have a cargo-binding domain at their carboxy terminus, which associates with specific cargo-binding proteins.
There are ~ 40 different myosin genes in higher eukaryotes7, and each myosin carries out its own special functions in vivo. Muscle myosin (of the myosin II class), for example, is highly specialized to provide appropriate forces and velocities during muscle contraction8. A primary function of non-muscle myosin II is to drive cytokinesis9. Myosin I, myosin VI and myosin VII are involved in the function of the inner ear, and mutations in them result in deafness10. Myosin V and myosin VI are involved in membrane trafficking and other aspects of cell organization. How all the different myosins are coordinated to carry out their various specialized tasks is an emerging field of research.
A particularly interesting member of the myosin family is myosin VI. Hasson et al.11 identified mammalian myosin VI in a kidney proximal tubule cell line, in which it is localized to an apical actin-rich structure in epithelial cells called the brush border. The Drosophila melanogaster equivalent of myosin VI, 95F, had been implicated in the transport of cytosplasmic organelles12,13. Myosin VI is involved in various functions involving multiple cellular organelles (FIG. 2). For example, myosin VI localizes at the base of stereociliary bundles of hair cells (specialized filipodia) of the inner ear and is essential for their structural integrity14. Myosin VI at the cell periphery also has a role in border cell migration during development15,16 and in cancer metastasis17. It is also involved in the transfer of endocytic vesicles from clathrin-coated pits to endosomes, along the cortical filamentous actin network. This includes the trafficking of rhodamine-labelled transferrin18,19, cystic fibrosis transmembrane conductance regulator (CFTR)20, lemur tyrosine kinase 2 (LMTK2; also known as BREK)21,22 and α5β1 integrin23. Furthermore, myosin VI is recruited to the Golgi complex by optineurin24, where it is essential for the maintenance of normal Golgi morphology25. How myosin VI carries out these diverse functions is unclear, although we do know some of the interacting proteins. These include disabled homologue 2 (DAB2)26,27, GIPC1 (REF. 28), synapse-associated protein 97 (SAP97; also known as DLG1)29 and optineurin24, which have been found to target myosin VI to different cellular compartments.
Similar to other myosins, myosin VI comprises a catalytic head with a converter domain, followed by light chain-binding domains (FIG. 2). The catalytic head is similar to that of myosin II and myosin V, with the exception of two inserts. One insert is near the nucleotide-binding pocket, which changes myosin VI’s nucleotide kinetics relative to other myosins4, and the second is an insert of 53 amino acids between the converter domain of the catalytic head and a canonical calmodulin-binding IQ motif11,30. This sequence is known as the unique insert (FIG. 3), and its C-terminal segment binds a single calmodulin31. The tail sequence beyond the calmodulin-binding region consists of the proximal tail, the medial tail, which is an unusual single α-helix (see below), and the cargo-binding domain4,11,32 (FIG. 3).
When the primary sequence of myosin VI was first analysed, a major portion of its tail region seemed to be a coiled coil, similar to the tails of myosin II and myosin V. Because this motif is found in proteins that dimerize in vivo, it was assumed that this motor was a dimer. Nearly all in vitro studies have used a myosin VI construct truncated at Arg992 in the tail domain (near the end of the predicted coiled-coil region), followed by a strong coiled coil — the leucine zipper GCN4 motif — to ensure that the dimer does not dissociate at the low concentrations used for single molecule assays33. The ~ 300 residues at the C terminus that constitute the cargo-binding domain are absent in this myosin VI construct. We refer to this construct as an artificial dimer, as it was meant to be a substitute for a presumed native dimeric molecule involving interactions of the cargo-binding domain with its appropriate cargo. This artificial dimer has been useful to probe details of myosin VI behaviour; however, it is still unclear whether myosin VI functions as a dimer, a monomer or both in vivo.
Myosin VI moves along an actin filament in the opposite direction to all other myosins that have been characterized30. Myosin I34, myosin II35, myosin V36, myosin X37 and myosin XI38 are directed towards the barbed ends of actin filaments. A truncated form of myosin IX was reported to move towards the pointed ends of actin filaments39, but the native myosin IX molecule was subsequently shown to be a barbed end-directed motor40. Myosin VI moves towards the pointed ends of actin filaments, and this ability enables it to carry out specific functions in vivo. For example, the brush border of epithelial cells is polarized with the pointed ends of the actin filaments directed away from the plasma membrane41. This is consistent with roles for myosin VI in endocytosis42 and myosin V in exocytosis43. Similarly, actin bundles in stereocilia are polarized such that myosin VI moves towards their base, thereby increasing membrane tension, which aids in stabilizing and maintaining appropriate tension in the stereocilia.
In addition to having reverse directionality, the artificial dimer of myosin VI has a larger step length on actin filaments than would be predicted from the early interpretations of the structure of the native protein, on the basis of current models of myosin movement44,45. These observations presented a serious challenge to the swinging lever arm hypothesis. In this Review, we discuss the results from in vitro motility assays, single molecule analysis, X-ray crystallography and other biophysical approaches, which proved pivotal in elucidating the unexpected, unique features of myosin VI with respect to its movement on actin filaments.
Huxley’s swinging crossbridge model described above, now known as the swinging lever arm hypothesis, has been the favourite model used to explain how muscles contract and how non-muscle myosins function. However, since Huxley put the model forward in 1969 (REF. 1) there have been fascinating oscillations in its acceptance.
Huxley’s swinging crossbridge model of myosin movement on actin fell out of favour in the 1970s owing to a lack of support on several fronts. All biophysical studies that involved placing probes on a reactive cysteine in the myosin head failed to reveal conformational changes in the head domain (the crossbridge) as it transitioned from putative pre-stroke to post-stroke states46,47. Furthermore, the step size of myosin for each ATP used was frequently measured at > 100 nm48, too large to be accounted for by a lever arm swing, which was unlikely to produce more than ~ 10 nm. These results led to the hypothesis that actin–myosin interactions are loosely coupled to ATP hydrolysis48. Furthermore, the movement of myosin on actin was proposed to be a result of Brownian motion of the myosin head together with prefer ential binding to a site on the actin filament in one direction (for example, towards the barbed end)49. This model of biased thermal diffusion is still prevalent in the current literature49. Indeed, it has been argued recently that biased thermal diffusion can explain the behaviour of processive myosin VI dimers50, with intramolecular strain contributing to the directional bias during myosin stepping.
In the 1980s and 1990s, several developments paved the way for our current understanding of how myosins work. In 1984, single fluorescently labelled actin filaments were observed for the first time51. Subsequently, purified myosin II molecules bound to a microscope slide were shown to move fluorescently labelled purified actin filaments in the presence of ATP, at velocities comparable with those seen during muscle contraction52. Indeed, the myosin head domain was shown to be all that is needed for this in vitro motility53. This in vitro motility assay enabled the examination of small numbers of molecules interacting with an actin filament, under conditions in which many of the complexities of an intact muscle fibre were eliminated.
Given that the myosin head is the motor domain, the crystal structure constrains the lever arm stroke size to ~ 10 nm. However, step size estimates varied from ~ 10 nm54–56 to > 100 nm48,57,58. These estimates were inferred from observations of multiple myosin molecules interacting with one or more actin filaments. It was therefore clear that a single myosin molecule going through a single cycle of interaction with actin had to be watched to directly measure its step size. Direct evidence for the predicted ~ 10 nm stroke for myosin II came from a single molecule optical trap study59. In the presence of ATP, myosin II was found to bind and almost instantaneously (< 1 ms) displace an actin filament by ~ 10 nm. This displacement in the optical trap is consistent with a swing of the end of the myosin lever arm of ~ 70°, resulting in a stroke size of ~ 10 nm. Subsequently, rigorous statistical methods, which form the current standard for the analysis of optical trapping data, confirmed a small step size for myosin60.
These single molecule studies were consistent with later X-ray crystallography studies61, in which a presumed pre-stroke crystal structure of myosin II bound to an ATP analogue was obtained. The lever was in a ‘cocked’ conformation, ~ 70° offset from the post-stroke structure6, which was consistent with an ~ 10 nm displacement (FIG. 1). Since then, crystal structures of several myosins in different nucleotide analogue states have been obtained, which have provided clues to how the myosin head moves to transduce the energy derived from ATP hydrolysis to the motion of the converter (FIG. 1), and consequently of the myosin lever arm. The inferred lever arm stroke from the crystal structures was confirmed by dynamic measurements using fluorescence resonance energy transfer62,63.
Subsequently, the amplification of the stroke through the swing of the lever arm was also seen for myosin V64,65. This myosin is dimeric and has a high duty ratio, which renders it processive (that is, a single motor can move many steps along an actin filament without detaching)36,66. Myosin V has a large step size of ~ 36 nm; presumably a product of its long lever arm of six light chains, which extend beyond each catalytic domain, compared with muscle myosin, which binds only two light chains. Myosin V is arguably the best-studied molecular motor, and research on myosin V has provided strong support for the swinging lever arm model of myosin movement67,68.
Despite strong support from extensive studies using myosin II and myosin V, the swinging lever arm hypothesis initially seemed to be inconsistent with the movement of myosin VI. Studies45,44 found that artificial dimers of myosin VI are processive and have a large step size of ~ 36 nm. This step size was surprising because myosin VI has the same number of bound light chains as myosin II (FIG. 2), which could not support such a large step size on the basis of the ~ 70° lever arm swing predicted from studies of myosin II and myosin V. So how does myosin VI achieve such a large step size?
The first clues to how myosin VI steps along actin came from image reconstructions of cryo-electron micrographs with the myosin VI head domain bound to actin30. These showed that the light chain-binding region points in a direction opposite to that of myosin II in the presumed post-stroke state30. The crystal structure of the myosin VI head in the presumed post-stroke state4 showed that the N-terminus of the unique insert wraps around the converter and forms an integral part of it (FIG. 4). This interaction redirects the myosin VI light chain-binding region in a direction opposite to that of myosin II and myosin V. This mechanical redirection was suggested to be the basis of the reversed direction of movement along actin, compared with that of other myosins4.
To understand the role of the unique insert in myosin VI directionality, myosin VI constructs were created that were truncated after the IQ calmodulin-binding domain, after the unique insert calmodulin-binding domain, near the middle of the unique insert and directly after the converter69. Using in vitro motility and optical trap assays, the constructs truncated after the unique insert calmodulin-binding domain were found to move towards the pointed end of actin, whereas the construct truncated directly after the converter moved towards the barbed end of actin. This is consistent with the myosin head converter domain moving towards the barbed end of actin, as seen for myosin II and myosin V, with the end of the calmodulin-binding domain swinging around towards the pointed end. Thus, the shift in direction of myosin VI movement is clearly brought about by the unique insert domain. Others came to the same conclusion by showing that a myosin VI molecule truncated after the converter and fused to a myosin V lever arm moves towards the barbed end of actin filaments70. Importantly, the change in myosin VI stroke size as a function of the length of its lever arm was found to be consistent with a large (~ 180°) lever arm rotation from the pre-stroke to the post-stroke state69 (FIG. 4). This finding was supported by the crystal structures of the myosin VI head in the pre-stroke and post-stroke states4,5 (FIG. 4). Crystal structures revealed an unexpected, radical change in the arrangement of the α-helices in the converter domain in the myosin VI pre-stroke state5 (FIG. 4). This new tertiary structure adopted by the converter explained the ability of this motor to swing its lever arm through this large angle. Other single molecule studies further confirmed this ~ 180° rotation71,72. The ~ 180° lever arm rotation seen for myosin VI results in a larger stroke than would have been predicted by the ~ 70° stroke that is characteristic of other myosins (FIG. 5), and these observations, together with the studies on myosin II and myosin V, provide overwhelming evidence in support of Huxley’s swinging crossbridge hypothesis.
A fundamental issue, however, still remains. As already described, the step size of artificial myosin VI dimers has been measured at ~ 36 nm44,73–75. By contrast, the stroke size of a single-headed construct of myosin VI truncated shortly after the calmodulin-binding domain (the canonical lever arm) is only ~ 12 nm76, even with an ~ 180° swing. It must be emphasized that the step size of a processive dimeric motor can be longer than the stroke size of its lever arm, as an additional diffusive search by Brownian motion of the leading head of the dimer can result in its binding to an appropriately oriented actin monomer that is many nanometers in front65,77. However, the proposed tail structure of myosin VI could not support the additional 24 nm required to achieve a step size of 36 nm following an ~ 12 nm stroke, as described below.
The IQ calmodulin-binding domain in myosin VI is followed by an ~ 70 amino acid sequence referred to as the proximal tail32,76. The proximal tail is followed by another ~ 70 amino acid sequence, termed the medial tail32,78, which is strongly predicted by the algorithm PAIRCOILS79 to be a coiled-coil domain76. A medial tail as a coiled coil would ensure that myosin VI is dimerized under certain cellular conditions. If the medial tail were a coiled coil, then the proximal tail would have to be an extensible, flexible element to allow the leading head to diffuse forwards the additional ~ 24 nm after the 12 nm stroke45. Indeed, disrupting the proximal tail by substituting a GCN4 leucine zipper coiled-coil motif into it abrogated the movement of myosin VI on actin filaments76. These findings supported a model in which the proximal tail is a flexible linker element that helps to span the ~ 24 nm gap in the myosin VI step.
However, using small-angle X-ray scattering analyses, Spink et al.32 suggested that the proximal tail in solution is a globular, folded domain of 2–3 nm in diameter and not an extended, flexible structure. The proximal tail was then shown to be a globular three-helix bundle by X-ray crystallography80. Thus, the proximal tail seemed most likely to be a structural element of myosin VI that might extend the lever arm, rather than acting as a flexible element to facilitate a large diffusive step.
So which tail structure could allow a dimeric myosin VI to straddle ~ 36 nm along an actin filament? When the strong coiled coil-forming GCN4 leucine zipper motif was inserted into the proximal region of the medial tail (presumed to be a coiled coil already), the myosin VI was unable to efficiently walk processively along an actin filament32. One explanation for this was that the medial tail is not a coiled coil at all. Consistent with this, a segment of the tail domain of myosin X, which was also predicted to be a coiled coil by the algorithm PAIRCOILS79, was shown to be an isolated single α-helix in solution81. The authors pointed out that the medial tail of myosin VI has a similar, highly charged sequence and, therefore, may also be a single α-helix. Subsequent studies showed that this is indeed the case78. The isolated medial tail of myosin VI was found to have a high (> 90%) α-helical content78, and measurements of its molecular weight in solution showed that it is monomeric in solution at concentrations as high as 40 µM32. Furthermore, when two copies of the medial tail were held in close proximity in solution through the formation of a disulphide bond between cysteines at the C-terminus of each monomer, no changes were seen in the melting profile in circular dichroism assays, in contrast to the sharp transition seen for canonical coiled coils32. Thus, the PAIRCOILS algorithm incorrectly predicts the medial tail to be a coiled coil because it assumes that a high density of charged residues, which make coiled coils soluble, is indicative of this structural motif. Indeed, for the most part, a heptad repeat pattern that is characteristic of coiled coils is not apparent in the medial tail. Low resolution reconstructions of the medial tail derived from small-angle X-ray scattering measurements clearly show an elongated structure that is consistent with the medial tail forming an isolated ~ 10 nm-long, single α-helix in solution32. Thus, it was suggested32 that the simplest model of how myosin VI spans 36 nm along actin is that the two medial tail domains do not dimerize but each adds to the lever arm length; indeed, a monomeric construct of myosin VI that includes the medial tail shows a stroke size of 30 nm in the laser trap78.
Although this model is satisfying in its apparent simplicity, Mukherjea et al.80 presented data that relate back to a flexible proximal tail76 and a dimerized medial tail. They proposed that even though the medial tail can be a single α-helix under some conditions, it has a weak dimerization site in its proximal region. They argued that dimerization only occurs when monomers of myosin VI are brought in close proximity, for example when concentrated on an actin filament82. The dissociation rate of these dimers, once formed, is fairly slow even after being released from actin. The authors showed by crystallography that the proximal tail domain is indeed a three-helix bundle and provided evidence that it unfolds into a flexible element, maintaining three independently standing single α-helices. These findings only apply to myosin VI when dimerized. In this model, it is this flexible prox imal tail domain that allows the dimeric myosin VI to span the 36 nm along actin.
One important difference between these two models is that the Spink et al. model proposes that myosin VI has a relatively rigid structure throughout its length, which allows it to step forwards with a normal step size even under backwards load. By contrast, myosin VI in the Mukherjea et al. model has a flexible proximal tail, and a marked reduction in step size would be predicted under backwards load. Laser trap studies with up to 2 pN backwards load have not detected an appreciable change in step size83. Although further experiments are needed to establish which of these models is correct, or indeed whether some combination of the two models is operating, it is clear that one or more of the tail domains beyond the calmodulin-binding domain is responsible for allowing the artificial dimer to span 36 nm along actin.
More studies are confirming that the medial tail can exist as a stable single α-helix in solution78,84. This is an unexpected finding, as protein α-helices are not generally considered to be stable in solution in the absence of the tertiary interactions found in folded proteins85. The sequence of the medial tail is homologous to two previously reported isolated, single, stable α-helices: a segment of the muscle protein caldes mon86 and, as already mentioned, a region in the tail domain of myosin X81. The common feature of these α-helices is a repeating sequence of four negatively charged amino acids (glutamic acid (E)) followed by four positively charged amino acids (a combination of arginines (R) and lysines (K)), which we have termed the ER/K motif. Our studies78 and bioinformatics analyses84,87 have reported that the ER/K motif is present in various proteins from a range of species84.
The stability of the medial tail ER/K motif-containing, single α-helix is a result of dynamic side chain charge–charge interactions between the E and R or K residues along the α-helical backbone78 (FIG. 6a). We have measured the rigidity of this ER/K α-helix and have shown that it has a persistence length of ~ 15 nm88. This large persistence length is consistent with the medial tail in myosin VI, which contributes substantially to the stroke size of myosin VI (FIG. 6b), even in the presence of a large bending force in a dual beam optical trap assay78, which would be expected to greatly reduce the step size of a more flexible element. Indeed, a monomeric construct of myosin VI that includes the medial tail gives a stroke size of ~ 30 nm78. It should be noted that the native (full length) myosin VI molecule has a stroke size of ~ 18 nm89. The reduced stroke size of the full length molecule is probably due to its folded state, with minimal stroke contribution from the medial tail32,89.
We propose that the ER/K α-helix is a tensegrity structure that is stabilized by a balance between two sets of forces, namely those between the oppositely charged E and R/K side chains (tensile force) and the hydrogen bonds in the α-helical backbone (compressive force) (FIG. 6c). Tensegrity is a common design principle in nature90 and sustains a range of biological structures, including the skeleton and the overall structure of the cellular cytoskeleton91–93. The stability of isolated α-helices is limited by interactions between polar water molecules and the hydrogen bonds in the backbone of the α-helix. The hydrogen bonds in the α-helical backbone are brittle and can tolerate little extension. Attack by a polar water molecule will destroy this interaction, causing the α-helix to unravel locally. In the ER/K α-helix, however, this is prevented by the long regions of hydrophobic residues on the side chains, which form a protective shell of hydrophobicity immediately surrounding the core of the α-helix78. Furthermore, the more malleable long-range charge–charge interactions between the E and R or K side chains enclose the back-bone in the event of transient breaks in the backbone hydrogen bonds. Maintaining this α-helical conformation facilitates the formation of backbone hydrogen bonds and thereby stabilizes and strengthens this structure.
The history of the swinging lever arm hypothesis shows how difficult it is to understand how any biological system really works. The myosin motor is one of the most studied enzymes and has seen several reversals in thinking about how it transforms the chemical energy of ATP hydrolysis into mechanical movement. The most recent challenge to the swinging lever arm hypothesis, the most popular model, was derived from the behaviour of myosin VI in single molecule experiments44,45. What was a great mystery a few years ago44,45 can now be understood by a combination of a 180° swing of the myosin VI lever arm and one or another highly unusual tail domain that behaves in unexpected ways. Thus, the myosin that seemed as if it might disprove the hypothesis has corroborated it.
In translating our understanding of myosin VI derived from in vitro experiments to its function in vivo, three pivotal questions need to be answered. First, how do mechanical forces in the molecule and mechanical forces that arise from interactions with other elements in vivo influence myosin VI function? Second, does myosin VI act as a dimer, a monomer or both in vivo? Finally, how do multiple myosins that are constrained to the same membrane compartment coordinate their movements in vivo to carry out their functions inside a cell?
An important concept to emerge from research on molecular motors is tension sensing to coordinate stepping, which is thought to be important for in vivo function. It is interesting that both myosin V and myosin VI give an ~ 25–30 nm stroke when the step size or stride of the dimeric molecules is ~ 36 nm. Why is this? This stroke can bring the trailing free head forwards, most of the way to its preferred actin binding site, but, importantly, not all the way there. Thermal energy would then be captured in a diffusion process, locking in intramolecular tension (arising between the two heads of the dimer) when the free head binds ~ 36 nm in front of the attached head. It is thought that the intramolecular tension between the two heads in the dimer when both heads are attached to actin allows them to have different nucleotide kinetics. In the case of myosin V, backwards force applied to the leading head reduces the ADP release rate67,68. Several kinetic measurements have been carried out to characterize the unique biochemical properties of the myosin VI motor domain as part of a monomer or the artificial dimer33,94–96. For the myosin VI artificial dimer, it has been proposed that it is the rate of ATP binding to the leading head that is reduced, rather than the rate of ADP release95–97.
Similar to intramolecular tension sensing, inter-molecular tension sensing can regulate motor activity. This type of regulation has been proposed for myosin VI, which is thought to act both as a transporter (for example, in the movement of endocytic vesicles) and as a structural anchor (to maintain the spatial organization of the Golgi complex)83. The motor is thought to be able to switch between these two functions by sensing external forces.
Much needs to be done to understand the roles of the different tail domains of myosin VI in transmitting loads through them to the catalytic site, thereby changing the behaviour of the motor. One approach is to replace parts of the tail with artificial protein structural domains, such as α-actinin repeats98,99 or flexible linkers98. For example, myosin VI with α-actinin repeats replacing both the IQ domain and the proximal tail, and with a flexible linker between the α-actinin repeats and the medial tail, shows surprisingly robust processivity98. The effect of mechanical load on the behaviours of this construct, however, has yet to be examined.
Myosins that form native dimers and have a high duty ratio, such as myosin V, are processive in vitro and are highly likely to be dimers and to be processive in vivo36. Myosin VI, however, has been shown to be monomeric89 when isolated in its native, full-length form. The myosin VI monomer seems to be a folded structure32, suggesting that the cargo-binding domain interacts with the head domain, possibly with the lever arm itself (FIG. 7). It is possible that some unfolding of the proximal tail is necessary for the molecule to assume this folded configuration. As noted above, most studies on myosin VI stepping in vitro have forced dimerization by introducing a GCN4 leucine zipper coiled-coil motif at the end of the medial tail, ~ 290 residues from the C-terminus of the native protein33. If the medial tail is not a coiled coil, the effects are of dimerizing the motor artificially at this position and whether this motor ever functions as a dimer in vivo become open questions. Indeed, recent results show that four or more myosin VI monomers working together can transport nanoparticles long distances on a native cellular array of actin filaments100 (FIG. 7). The number and oligomeric state of myosin VI molecules present on a given sub-cellular membrane compartment are not understood and need to be determined. If myosin VI functions as a monomer in vivo, the medial tail may be simply an extension of the canonical light chain-bound lever arm, and this would allow ~ 30 nm stroke sizes as the molecule moves forwards. If myosin VI were to function only as a monomer, then issues about the mechanisms by which dimers might function are not relevant.
What is missing and desperately needed at this point is a thorough biochemical and structural analysis of the interactions of myosin VI and its cargo-binding proteins, and a complete list of myosin VI binding partners101. Only when we understand the structures of entire native complexes of myosin VI bound to its cargoes can we properly begin to focus on how mechanical forces in the molecule and mechanical forces that arise from interactions with other elements in vivo influence myosin VI function, particularly how multiple myosins constrained to the same membrane compartment co ordinate their movements in vivo to carry out their functions inside the cell.
In summary, much needs to be done to determine with certainty the oligomeric state of myosin VI as it functions in vivo. There is considerable circumstantial evidence supporting the idea that it can function as a dimer32,80,82,102–105. Indeed, GIPC and optineurin (which interact with myosin VI) are thought to be dimeric themselves and may therefore induce the dimerization of myosin VI101. However, if dimers are created by precise positioning of two monomers on a dimeric cargo103, these dimers will clearly be structurally different from the experimental artificial dimer created by inserting a GCN4 motif near the end of the medial tail domain. The most important next step is to biochemically reconstitute the native myosin VI with its range of cargoes, allowing complete biochemical and biophysical characterization of such complexes. The feasibility of this approach has recently been shown105.
Understanding myosin VI function has been an important research focus over the past decade, fuelled by two unique characteristics of this myosin. The first is its ability to move along actin filaments in the opposite direction to all other myosins and the second is its ability to take large ~ 36 nm steps with what seemed at first to be a short lever arm. This represented a challenge to the swinging lever arm hypothesis. Both of these issues have now been largely resolved, and the swinging lever arm hypothesis has gained strong support.
The next big challenge for the field of molecular motors lies in translating detailed in vitro biochemical and biophysical characterizations into an understanding of motor function in vivo. Our knowledge of molecular motor function inside cells has been inferred primarily from the proteins or membrane compartments they associate with or from loss-of-function experiments in vivo. For instance, null myosin VI mutants lose the integrity of stereocilia, suggesting that myosin VI helps to generate membrane tension that stabilizes these structures. An important next step in bridging the gap between single molecule and other in vitro studies and in vivo function of motors such as myosin VI will be the thorough biochemical and structural characterization of the complexes made by these motors and their cargo-binding domains. Also needed is the development of new assays and techniques that facilitate the study of their coordinated function. Of particular interest will be understanding why the structure of myosin VI differs so much from other myosins and what purposes this design serves in vivo.
J.A.S is supported by grant GM33289 from the National Institutes of Health. S.S. is supported by an American Cancer Society postdoctoral fellowship.
Competing interests statement
The authors declare no competing financial interests.
Protein Data Bank: http://www.rcsb.org/pdb
James A. Spudich’s homepage:
ALL LINKS ARE ACTIVE IN THE ONLINE PDF