Biochemical Society Transactions

Cellular Cytoskeletal Motor Protein

Regulation of myosin 5a and myosin 7a

Verl B. Siththanandan, James R. Sellers

Abstract

The myosin superfamily is diverse in its structure, kinetic mechanisms and cellular function. The enzymatic activities of most myosins are regulated by some means such as Ca2+ ion binding, phosphorylation or binding of other proteins. In the present review, we discuss the structural basis for the regulation of mammalian myosin 5a and Drosophila myosin 7a. We show that, although both myosins have a folded inactive state in which domains in the myosin tail interact with the motor domain, the details of the regulation of these two myosins differ greatly.

  • actin
  • ATPase activity
  • calcium
  • myosin

Introduction

The myosin superfamily throughout phylogeny consists of at least 35 classes of motors with another 145 motors that are orphans and do not fall into any defined class [1]. The human genome contains 39 myosin genes from 12 classes [2]. Within cells, these myosins perform a plethora of tasks such as cargo transport, contraction of muscle, endocytosis, exocytosis, cytokinesis and cell adhesion. Myosins participate in the formation and maintenance of actin bundles such as are found in filopodia, hair cell stereocilia and enterocyte brush border microvilli.

In general, myosins have three distinctive domains. A motor domain, usually located at the N-terminus, contains the binding sites for nucleotide and actin. A neck region follows this and is composed of IQ motifs that bind calmodulin or other calmodulin family member proteins. The number of IQ motifs determines the length of the neck and, since the neck region acts as a rigid lever arm to generate the power stroke, a longer-necked myosin will typically have a larger power stroke. This is followed by the tail region, which is the most diverse domain in myosins. In many myosins, the tail contains coiled-coil-forming amino acids which dimerize to form ‘two-headed’ myosins. Most myosins, however, lack a functional coiled-coil-forming domain and are monomeric with respect to motor domains. Other functional domains such as MyTH4 (myosin tail homology 4)–FERM (4.1/ezrin/radixin/moesin) domains, PH (pleckstrin homology) domains and SH3 (Src homology 3) domains can also be found in the tails of different myosins. In general, within a given myosin class, the tail domain structure is conserved. In most cases, the activity of the myosin is regulated, but the form of this regulation can vary. An exception to this is myosin from mammalian skeletal muscle whose enzymatic activity of is inherently unregulated. Instead, the troponin–tropomyosin complex regulates the interaction of this myosin with actin [3]. In vertebrate smooth and non-muscle myosin class 2 molecules, the regulation is via phosphorylation of the regulatory light chain by myosin light chain kinase [4].

In the present review, we compare the regulation of myosin 5 and myosin 7. In both cases, regulation is brought about by an intramolecular folding event, but the nature of the folding differs in detail.

Regulation of myosin 5a

Mammals have three genes for class 5 myosins [2]. The best studied of these is myosin 5a, which is a processive actin-dependent cargo motor [5]. In melanocytes, it is required for the localization of pigment granules termed melanosomes to the dendritic processes and also moves endoplasmic reticulum into the spines of Purkinje cells [69]. In melanocytes, the long-range movements of pigment granules are via microtubule motors, kinesin and dynein, which are also bound to the melanosomes. Myosin 5a motors engage actin filaments in the tips of the dendrite where the density of actin filaments is high and possibly carry out short-range transport to position the melanosomes near the membrane so that they can be transferred to keratinocytes.

In vitro, myosin 5a is capable of moving processively along actin, taking 50 or more ATP-dependent steps per encounter before dissociating [10]. The step size is 36 nm, which matches the half-helical repeat of the actin filament, essentially allowing myosin 5a to walk ‘straight’ across the top of an actin filament. To accomplish this feat, myosin 5a evolved three structural adaptations that allow it to be an effective cargo motor. It has a long neck region consisting of six IQ motifs that allows it to have a long power stroke capable of generating the 36 nm steps. At the C-terminus of the molecule, there is a cargo-binding GTD (globular tail domain) that is able to bind membranous cargo through receptor molecules. Finally, it has an extensive coiled-coil-forming sequence that allows it to form a two-headed motor and provides a spacer between the mechanical ends of the molecule (the motor and neck regions) and the cargo-binding ends. Two heads are critical for the processive movement, as it allows one motor domain to be attached to actin, while the other is searching for a new forward binding site.

Myosin 5a has also evolved two kinetic features that supports processive movement. The rate-limiting step in its kinetic cycle is the release of ADP from an actomyosin·ADP complex [11]. With bound ADP, myosin 5a has a high affinity for actin. Thus, in the steady state, a myosin 5a motor has a high duty ratio which means that it will spend the majority of its kinetic cycle strongly bound to actin. In addition, there is a strain-dependent gating of the kinetic cycles of the two motor domains such that ADP is unlikely to be released from the leading head as long as the trailing head is still attached to actin [1215]. This feature enhances the processivity of the myosin even further.

The study of the regulation of myosins have been greatly aided by the ability to express full-length myosin and truncations of the molecule in the baculovirus/Sf9 insect cell system. Using this system combined with standard molecular biological techniques, mutants can be made that lack portions of the molecule that may or may not be required for regulation. Regulation can be studied at several levels. The actin-activated MgATPase activity and the sliding actin in vitro motility assay which measures ability of the myosin to move fluorescently labelled actin filaments when the myosin is bound to the surface can be easily measured in most laboratories [16]. A variation on the standard in vitro motility assay involves the use of TIRF (total internal reflection fluorescence) microscopy to measure the movement of single fluorescently labelled myosin molecules along actin that is bound to a coverslip surface [17]. Together, these assays define how the enzymatic and mechanical properties of myosin is regulated.

Full-length myosin 5, either purified from brain or expressed in the baculovirus/Sf9 system, requires micromolar levels of Ca2+ ions for activation of its actin-activated MgATPase activity in vitro [1821]. In the absence of Ca2+, the myosin interacts only weakly with actin and its MgATPase activity is very low, even at high actin concentrations [22]. Truncation of the GTD and a portion of the coiled-coil motifs results in the formation of a two-headed HMM (heavy meromyosin)-like fragment. This HMM molecule has a high actin-activated MgATPase activity regardless of the actin concentration, implicating the tail region in the suppression of activity in the absence of Ca2+ [23]. Ultracentrifugation experiments have shown that, in the absence of Ca2+, full-length myosin 5a sedimented at 14S, but, in the presence of Ca2+, it sedimented at 11S, suggesting that the molecule is more compact in the absence of Ca2+ [19,21]. In contrast, Ca2+ had little effect on the sedimentation of myosin 5a HMM, again implicating the GTD region in the folding mechanism.

Electron microscopy showed that, in the absence of Ca2+, myosin 5a adopts a conformation in which the two motor domains fold back and each contact the globular tail domain (Figure 1). This structure was seen in both negatively stained images of single myosin 5a molecules or in two-dimensional crystals of myosin 5a grown on lipid-covered surfaces [24,25]. Single particle averaging techniques enhance the resolution and show that the region of the motor domain contacted by the tail is distinct from the actin-binding regions and thus the inhibition of actin-activated MgATPase activity must be allosteric. The region of contact appears to be a surface loop containing two negatively charged residues (Asp136 and Asp134) that are highly conserved among class 5 myosins [25]. Subsequent mutation of Asp136 abolished the regulation of the molecule [26]. These authors of this study also identified two arginine residues (Lys1706 and Lys1779) in the GTD as participants in the regulation. The structure of the molecule is tightly coupled to its ability to self-regulate. Shortening the neck region by removing IQ motifs or altering the length of the coiled-coil tail interfere with the regulation [27].

The inhibited complex could be formed in trans by mixing myosin 5a HMM with a purified complex of GTD fused to GST (glutathione transferase) [25,26]. Since GST dimerizes, this would create a dimeric GTD complex that might resemble the structure of the dimeric GTD complex found at the end of myosin 5a. Recall that the actin-activated MgATPase activity of myosin 5a HMM is high even in the absence of Ca2+. When the MgATPase of this molecule was measured in the presence of increasing concentrations of GST–GTD in the presence of actin, the enzymatic activity was suppressed in the absence of free Ca2+. At high GST–GTD concentrations, addition of Ca2+ activated the enzymatic activity of the myosin similar to the way in which Ca2+ activates the actin-activated MgATPase of full-length myosin 5a. Electron microscopy of the myosin 5a HMM–GST–GTD complex reveal triangular-shaped molecules with an appearance virtually identical with that of full-length myosin 5a molecules in the absence of Ca2+ [25].

Figure 1 Regulated conformation of myosin 5a

Upper panels: averaged image classes of myosin 5a molecules. Lower panel: superimposition of a coloured cartoon depicting the major domains of myosin 5a. Data taken from [25].

Ca2+ activation of the regulated myosin 5a molecule is not likely to be the physiological activator. Although Ca2+ activates the MgATPase activity of myosin 5a in the presence of actin, it abolishes the ability of myosin 5a to interact in a mechanically competent manner with actin [18,19]. This was noticed in the first paper describing the effect of Ca2+ on myosin 5a [18]. The authors found that the full-length myosin 5a, when bound to a surface, only moved actin filaments when no Ca2+ was present. This was confirmed in subsequent publications using both full-length myosin and HMM and both geometries of in vitro motility assays [19,28]. The reason for this lies in the fact that Ca2+ results in the dissociation of one or more calmodulins from the neck of myosin 5 [29,30]. Studies have shown that, when an IQ motif in a myosin neck is not occupied by a light chain, the neck is ‘floppy’ and will not transmit mechanical force [21,31]. How then can one explain the fact that full-length myosin 5a bound to a surface will translocate actin filaments even in the absence of Ca2+, where, in solution, its MgATPase activity is inhibited. The answer probably lies in how myosin 5a binds to the coverslip surface. Some of the molecules probably unfold when bound to the surface and are active and able to interact productively with actin. In the single-molecule TIRF motility assay, the frequencies of initiation of movement by full-length myosin 5a and HMM can be compared directly. Under these conditions, HMM has 10–40 times more movement events than does full-length myosin, since the latter molecules are regulated and are not interacting with actin [25,28].

This raises the question of how myosin 5a is regulated in cells. In melanocytes, the answer is clear (Figure 2). The C-terminus of melanophilin binds to the GTD of myosin 5a [79]. The N-terminus of melanophilin binds to Rab27a, which itself in integrally bound to the melanocyte. The interaction between myosin 5a and melanophilin is sufficient to unfold the molecule and relieve the inhibition in vitro. This mode of regulation can be demonstrated in vitro. Addition of melanophilin to full-length myosin 5a in the presence of actin activates its MgATPase activity in the absence of Ca2+ [32]. Similarly, a GFP–Rab27a–melanophilin–myosin 5a complex can be observed moving along actin in the single-molecule TIRF motility assay [33]. Presumably there are similar proteins in neuronal cells that link brain myosin 5a to the endoplasmic reticulum.

Figure 2 Receptor complex for melanosome trafficking

Cartoon of how melanosomes are moved along actin filaments by a complex of myosin 5a, melanophilin and Rab27a. Taken from reference [7] with permission.

Regulation of myosin 7a

There are two genes for myosin 7 in both mammals and Drosophila [2,34]. Mutations in human myosin 7a are associated with deafness at birth followed by retinal degeneration resulting in blindness [35]. In Drosophila, myosin 7a mutations are embryonic- and larval-lethal, but a small percentage of the fruitflies escape this lethality and mature into adults which are deaf and have bristle defects [36,37]. What all of these defects have in common are actin-rich structures. The stereocilia of the hair cells in ears are a tightly packed actin bundle and actin bundles are also found in the Johnston organ of the fruitfly which is the sensory organ for hearing [37]. Drosophila bristles are built from massive actin bundles [38], and the apical region of retinal pigmented epithelial cells in the eye has dense parallel arrays of actin filaments.

Drosophila myosin 7a has a motor domain followed by a neck region with five IQ motifs. There is a short segment that was predicted to form a coiled-coil, but instead is more likely to be an extended single SAH (stable α-helical) [39] domain which serves to extend the neck region. Following this are two MyTH4–FERM domains separated by an SH3 domain. FERM domains have three subdomains and the last subdomain of the myosin 7 FERM domains is highly conserved among class 7 myosins and was termed a MyTH7 domain [40]. Actin activates the MgATPase activity of full-length myosin 7a to a Vmax of approximately 1 s−1, but the amount of actin needed for half-maximal activation (KATPase) is high, at approximately 40 μM [41]. In contrast, an S1 (subfragment-1)-like fragment consisting of the motor domain and first IQ motif is also activated to a Vmax of approximately 1 s−1, but the KATPase is now only approximately 1 μM. This means that, at low actin concentrations (i.e. 5 μM), the activity of the full-length myosin is very low, whereas that of the tail-less fragment is at Vmax. This ‘regulation’ is inherently different from that of myosin 5a described above where the ‘off’ state is largely independent of actin. Here, it is the affinity for actin that is being regulated. Ca2+ ions also regulate the activity of full-length myosin 7a [42], and the regulation may involve the calmodulin bound to the third IQ motif.

Electron microscopy analysis of full-length myosin 7a in the presence of ATP shows that the molecule is tightly folded to the extent that it is difficult to distinguish the tail and the head [42]. In the absence of ATP, the molecule unfolds allowing the identification of the motor domain (Figure 3). These images also show that myosin 7a is single-headed. The C-terminal tail region was systematically truncated and the various fragments assayed for ATPase activity [41]. All of the fragments behaved essentially the same; their MgATPase activity could be activated at low actin concentrations, in contrast with the behaviour of the full-length molecules. Electron microscopy analysis of one of the truncated molecules revealed that it was not able to fold in the presence of ATP, suggesting that the folding was the basis of the inhibited state. The shortest C-terminal truncation assayed only eliminated the last 99 amino acids which comprise the MyTH7 domain of the second FERM domain. This sequence is very homologous among myosin 7 molecules. Mutation of a pair of conserved positively charged amino acids in this region (Arg2140 and Arg2143) eliminated the ability to fold and resulted in the MgATPase activity being activated at low actin concentrations [41].

Figure 3 Regulated conformation of myosin 7a

Left-hand panel: an averaged image of full-length (FL) myosin 7a in the folded state. Right-hand panel: a cartoon depiction of how the molecule might be folded. Images taken from [41] with permission and the model for folding was conceived by Michelle Peckham and Tom Baboolol (University of Leeds).

This does not answer the question of why myosin 7a can be activated at high actin concentrations. The answer may lie in a second low-affinity actin-binding site in the last FERM domain. This domain was expressed and shown to be able to bind weakly to actin with a Kd of approximately 30 μM [41]. This value is similar to the KATPase value obtained by titrating the actin-activated MgATPase. Possibly at high actin concentrations, myosin 7a binds to actin through this low-affinity second actin-binding site which then causes the myosin to unfold, freeing up the motor domains to interact with actin in a productive manner. This could also be the basis for targeting this molecule to actin bundles. It is also possible that myosin 7a can be regulated by binding to membranes or that there are proteins that bind to the tail region and unfold the molecule.

Myosin 7a is purified from Sf9 insect cells as a monomeric protein, but enzymatic characterization of the motor domain fragment demonstrated that it has an even higher duty ratio than that of myosin 5a [43,44]. This strongly suggested that a dimeric myosin 7a would be processive, and it was found that if two myosin 7a motor domains were dimerized by the inclusion of a leucine zipper motif at the C-terminal end of the putative coiled-coil region, the molecule was indeed processive in the single-molecule TIRF motility assay [43]. This raises the question of whether the oligomeric state of myosin 7a might be regulated in cells to produce two functionally distinct types of motor. Such a suggestion has been made with myosin 6, and there is evidence to show that this motor can dimerize if brought into close proximity on actin or in the presence of certain myosin 6-binding proteins [45,46]. There are two MyTH4–FERM domains and an SH3 domain in the tail of myosin 7a, both of which are known to function as binding motifs when present in other proteins. There have been several binding proteins described for mammalian myosin 7a [47], and it is likely that the Drosophila homologue of many of these might bind to Drosophila myosin 7a. It is possible that some of these proteins might potentially dimerize the molecule.

The kinetics of myosin 7a indicate that it is a very slow molecular motor, and, in its monomeric form, it might best function to tether cargo to actin. Some myosins appear to act as molecular force sensors which stall when subjected to higher forces and remain attached to actin with little expenditure of ATP [48]. Given the very slow kinetics of myosin 7a, it is possible that it is such a myosin. Alternatively, myosin 7a might function as an actin-bundling protein via the ATP-sensitive binding site in the motor and the ATP-insensitive site located in the tail or might serve as a link between actin via the motor domain and binding to membranes via one of the FERM domains. Clearly, there are many functional possibilities for myosin 7a.

Conclusions

The enzymatic activity of both myosin 5a and myosin 7a are regulated by an intramolecular folding event in which a domain in the tail contacts the motor domain. However, the details of how this is accomplished are quite different and may be related to the function of the molecule. Myosin 5a is a two-headed motor. The electron microscopy images of the inhibited complex clearly show that each head interacts with a lobe of the GTD. The folding almost completely shuts down the molecule in terms of its activation by actin. A cargo receptor has been identified that binds the tail and activates the enzymatic activity. This cargo receptor is shown to be essential for movement of myosin 5a-dependent cargo in cells. In contrast, the myosin 7a molecule is purified as monomer. It appears to adapt a wide range of shapes when forming the inhibited complex and it is difficult to distinguish what part of the motor is interacting with what part of the tail. The activity of myosin 7 increases as actin concentration is increased and this may be due to a low-affinity actin-binding site in the tail. No cargo receptor molecules have been reported for myosin 7a and it is not clear whether the molecule can ever exist as dimers inside the cell.

Funding

This work was funded by the Intramural Research Division of the National Heart, Lung and Blood Institute.

Acknowledgments

We thank Yi Yang, Kavitha Thirumurugan, Takeshi Sakamoto, Tom Babalool, Matt Walker, John Hammer, Peter Knight and Michelle Peckham for helpful discussions during the collaboration that led to much of the work reviewed in the present paper.

Footnotes

  • Cellular Cytoskeletal Motor Proteins: A Biochemical Society/Wellcome Trust Focused Meeting held at Wellcome Trust Genome Campus, Hinxton, Cambridge, U.K., 30 March–1 April 2011. Organized and Edited by Folma Buss (Cambridge, U.K.) and John Kendrick-Jones (MRC Laboratory of Molecular Biology, Cambridge, U.K.).

Abbreviations: FERM, 4.1/ezrin/radixin/moesin; GST, glutathione transferase; GTD, globular tail domain; HMM, heavy meromyosin; MyTH, myosin tail homology; SH3, Src homology 3; TIRF, total internal reflection fluorescence

References

View Abstract