Vertebrate Unconventional Myosins

Myosinsarealargefamilyofstructurallydiversemechanoen-zymesthatbindtoF-actinandhydrolyzeATPtoproducemechan-icalforce.Phylogeneticanalysisofthemyosinmotordomainshasidentified 11 distinct classes of myosin, seven of which are ex-pressedinvertebrates(forreviewseeRef.1).Thesesevenmyosinclassesincludeconventionalmyosin,ormyosin-II,andsixmorerecentlyidentifiedunconventionalmyosinclasses,myosins-I,-V,-VI,-VII,-IX,and-X(seeFig.1).AttheNterminus,eachverte-brateunconventionalmyosincontainsaconservedmotordomainthatincludesboththeATPbindingandATP-sensitiveactinbind-ingsites(forreviewofthestructureofthemyosinmotordomainsee Ref. 2). Following the motor is a light chain (LC)

genetic studies that firmly establish a function for a vertebrate myosin-I. Subclass 1, Amoeboid Type Myosins-I: Human Myosin-IC, Chicken BB Myosin-IB, Rat Myr3-It has been more than two decades since the initial purification of myosins-I of this subclass from Acanthamoeba (8), and the wealth of information with regard to the biochemical properties, functional domain structure, and subcellular localization of the multiple myosins-I expressed in Acanthamoeba has been reviewed in detail elsewhere (for references see Refs. 1, 9, and 10). Several subclass I myosins have been identified in vertebrates (see Table I). Based on primary structure, as well as functional domain mapping studies of Acanthamoeba myosins-I, the tails of these myosins have been divided into three tail homology (TH1-3) subdomains. The TH-1 domain, positioned immediately after the neck domain is rich in basic residues and may effect binding to acidic phospholipids. A similar basic domain comprises the full tail of the other three known myosin-I subclasses. The TH-2 domain is rich in proline and also either glycine, as in human myosin-IC ( Fig. 1), or glycine/alanine and glycine/ glutamate, as in Dictyostelium myosins-I. For Acanthamoeba myosins-I, it has been shown that this domain exhibits ATP-insensitive actin binding, thus allowing this myosin to cross-link actin filaments. This critical feature has not been established for vertebrate subclass I myosins. The TH-3 domain consists of a Src homology-3 (SH3) domain. In Acanthamoeba, it has been shown that the head domain of this class of myosin-I must be phosphorylated to be active by a specific myosin-I heavy chain kinase at a serine or threonine residue in a conserved portion of the motor domain (reviewed in Ref. 10). None of the known vertebrate myosins-I contain either a serine or threonine at this site (the TEDS site (11)).
The expression profile of human myosin-IC (12) and its rat ortholog, Myr3 (13), indicate that this myosin-I is present in a wide range of tissues and cell types. Localization studies have been restricted thus far to cultured normal rat kidney cells, and in these cells, Myr3 exhibits a punctate distribution in the cytoplasm and is concentrated in F-actin-rich, elongate structures at sites of cell-cell contact that are distinct from adhesion plaques (13). Interestingly, Concanavalin A treatment of these cells increases the number of these Myr3-rich structures at sites of cell contact. Although no subclass I myosin-I has been purified from a vertebrate source, analysis of the ATP-dependent actin binding of Myr3 present in tissue extracts suggests that the TH-3 domain of Myr3 may modulate the interaction of this myosin-I with actin.
Subclass 2, BB Myosin-I and Myr1/Myosin-I␣-The founding member of this myosin-I subclass is chicken brush border (BB) myosin-I (Fig. 1, Table I). BB myosin-I is expressed in the intestinal epithelia of the large and small intestine, where it comprises the spirally configured bridges that tether the microvillar actin core to the plasma membrane (reviewed in Ref. 14). In contrast, to BB myosin-I, the other members of this subclass, myosin-I␣ and Myr1 (Table I), have been shown to be expressed in a wide range of tissues. Both BB myosin-I and myosin-I␣/Myr1 contain multiple neck-associated CaM LC. Myr1 has been purified from liver (15) and, like BB myosin-I (for review of biochemical studies of BB myosin-I see Refs. 14, 16, and 17), Myr1 exhibits a relatively slow (30 -50 nm/s) in vitro motility that is inhibited by calcium.
Since BB myosin-I is among the best characterized vertebrate unconventional myosins, it serves as an excellent test case to determine if in vitro expression systems can be used for detailed structural and functional analysis of these myosins. Thus, it is noteworthy that expressed chicken BB myosin-I, with properties comparable with tissue-purified protein, has been obtained from transfected COS cells (18) and most recently using the baculovirus expression system in Sf9 cells (19).
A fundamental question to address for these and all other unconventional myosins is what role the tail domain plays in subcellular localization. In the case of BB myosin-I, Footer and Bretscher (20) observed targeting of BB myosin-I to dorsal microvilli of fibroblasts (i.e. not an enterocyte) injected with this myosin. This surprising result suggests that the head domain, through its interaction with the microvillar core bundle, may drive assembly of BB myosin-I rather than its interaction with a specific docking protein complex in the microvillar membrane.
The importance of the head domain in subcellular localization has also been addressed for Myr1, as well as the subclass 3 myosin-I, Myr2. Rü ppert et al. (21) compared the localization of fulllength versus tail domains of Myr1 and -2 in normal rat kidney cells transfected with appropriate epitope-tagged constructs. In these cells, the endogenous Myr1 associates with the plasma membrane, including membrane ruffles, while Myr2 exhibits a punctate pattern throughout the cytoplasm (21). In transfected cells, the fulllength, epitope-tagged protein co-localized with the endogenous myosins. Overexpressed epitope-tagged myosin tail domains did not affect the location of the native protein and exhibited much higher levels of cytosolic expression suggesting that the head domain is necessary for proper subcellular targeting. Lewis and Bridgman (22) have analyzed the location of myosin-I␣ within cultured rat superior cervical ganglion cells at the light and electron microscopic level of resolution. Myosin-I␣ was observed to have a punctate distribution within the superior cervical ganglion with an enrichment at the cell cortex. Interestingly, these punctae were found to not be membranous organelles but rather reflected an association with regions of plasma membrane adjacent to the actin-rich cortex (22). These results force us to rethink the assumption that a punctate staining implicates an association with vesicles.
Subclass 3, Myosin-I␤/Myr2-Myosin-I␤ was first purified from bovine brain and adrenal gland, and the initial biochemical studies indicating that it contains multiple CaM LM and exhibits enzymatic properties similar to that of BB myosin-I have been reviewed elsewhere (1,17). cDNAs encoding presumed orthologous heavy chains have been cloned from frog and rat (termed Myr2; Table I) and, like members of subclass 2, myosin-I␤ is expressed in a wide range of tissues. Localization studies using different antibodies to this myosin-I have given conflicting results even in the same cell type. Myosin-I␤ has been reported to be associated with the cell periphery, including membrane ruffles (23), as well as with the cytokinetic furrow (24). This is in contrast to Ruppert et al. (21) who reported that this myosin was associated with discrete puncta throughout the cytoplasm. As noted by these authors, this discrepancy may well be the result of cross-reactivity of the original myosin-I␤ antibodies with other myosins-I (such as myosin-I␣).
The most significant finding with respect to determining a function for myosin-I␤ has been the demonstration that this myosin (or related immunogen) is expressed in the hair cells of the frog sacculus, where it is restricted in location to the tips of the hair bundles (25). As recently reviewed (26), myosin-I␤ has been proposed, based on this localization, to be the actin-based adaptation motor, which gates the tip link-associated cation channels that open in response to mechanical stimulation of the hair cell bundles. As discussed in detail elsewhere (26,27), a critical property of the myosin serving as the adaptation motor is that, upon influx of Ca 2ϩ due to the mechanical opening of the transduction channel, the mechanochemical activity of the motor should be inhibited. Consistent with this model are the recent studies of Zhu et al. (28), who have conducted an extensive analysis of bovine myosin-I␤ made in Sf9 cells using the baculovirus expression system. In contrast to the initial enzymatic studies using tissue-purified protein, these authors found that expressed myosin-I␤ is a much faster motor, with rates that are comparable with some conventional myosins and myosin-V (500 nm/s versus 60 nm/s for BB myosin-I). Most importantly, the in vitro motility of this myosin is inhibited by Ca 2ϩ even in the presence of saturating CaM, a property consistent with its potential role as the tip link motor.
Subclass 4, Myr4/Myosin-I␥-The only vertebrate member of this subclass to be fully cloned thus far is rat Myr4 (Table I), although homologs of this subclass have been identified in both Drosophila and Caenorhabditis elegans (1,29). Like other vertebrate myosins-I the light chains of Myr4 are CaM (7), although the two IQ motifs that comprise the neck domain exhibit differing Ca 2ϩ sensitivities for CaM binding, based on studies using bacterially expressed protein (7). These studies also revealed a non-IQ, Ca 2ϩ requiring a calmodulin binding site within the tail domain.

Myosin-V
Myosins-V, unlike myosins-I, are two-headed as they self-associate via an ␣-helical coiled-coil portion of the tail (Fig. 1). All myosins-V contain six IQ motifs and have been shown to be associated with multiple CaM LCs. Purified myosin-V is also associated with two as yet unidentified LC distinct from calmodulin. The tail domain of myosin-V is of unknown function but may serve to target this myosin to organelles. Interestingly, sequences similar to the tail of myosin-V were recently uncovered in the non-myosin proteins human AF-6 and its homolog Drosophila Canoe (30), suggesting that the tail of myosin-V contains a conserved functional domain as is seen for many of the other unconventional myosin classes (Fig. 1).
Genetically, myosins-V have been implicated in membrane movements. Mice with mutations in myosin-V exhibit a lightened coat coloring or dilute phenotype. This is due to a defect in transport of the melanosomes, or pigment granules, to the dendrites of melanocytes, which results in inefficient phagocytosis of the pigment by the keratinocytes of the hair shaft (summarized in Ref. 31). To date, actin-based movement of melanosomes has not been observed, but supporting a role for myosin-V in melanosome movements, preliminary work (32) has shown that myosin-V can be immunoprecipitated from isolated melanosomes from normal animals but not from dilute mice.
Other alleles of dilute, termed dilute-lethal, in addition to the lightened coat coloring also produce an abnormal postnatal neurological disorder characterized by convulsions and death by 3 weeks of age (31). No obvious defect in neuroanatomy of these mice has been observed, and the basis for the neurological abnormalities is not understood. Because of the abundant expression of myosin-V within neurons, but not within other cells of the nervous system, it has been postulated that, by analogy to the melanocyte, myosin-V is also necessary for organelle movement in neurons.
Recent work indicates that, indeed, there is a potent actin-based motility, which occurs within vertebrate neurons as well. Using video-enhanced DIC microscopy, Evans and Bridgman (33) ob- served movement of particles within the growth cones of neurons cultured in low doses of nocodazole and cytochalasin, inhibitors of microtubule and microfilament assembly, respectively. This treatment allowed for visualization of movements of particles along actin bundles at rates of 0.48 m/s, a rate very similar to what is seen in vitro for chick brain myosin-V. In preliminary work, Evans and colleagues (34) have located myosin-V by immunoelectron microscopy within these growth cones and have found it in association with small (50 -100 nm) organelles, actin, and plasma membrane domains. These results strongly suggest that myosin-V is indeed an organelle motor in brain.
Recent biochemical studies on purified chick brain myosin-V indicate that the ATP-dependent actin binding and ATPase properties of this motor are consistent with that of an organelle motor (35). In striking contrast to conventional myosins and BB myosin-I, the steady state Mg-ATPase of myosin-V is maximally activated at very low F-actin concentrations, with a K ATPase of ϳ1 M actin. It will be critical to determine if these steady state measurements indicate that myosin-V can remain associated with an actin filament for longer periods than conventional myosin during its duty cycle, a critical feature for a motor expected to translocate organelles.

Myosin-VI
Compared with myosins-I and -V much less is known of the biochemistry of myosin-VI. Sequence analysis of myosin-VI revealed that the 145-kDa heavy chain has a 200-aa coiled-coil domain in its tail, which could allow for dimerization (Fig. 1). In agreement with the presence of a single IQ motif in the neck domain, coimmunoprecipitation studies showed that myosin-VI was associated with CaM (36). The neck domain of myosin-VI is unusual, however, as there is a 50-aa linker between the end of the motor domain and IQ motif (Fig. 1). It is unknown if this linker domain serves as a binding site for a novel LC. The tail domain of myosin-VI is unique, and it is remarkably conserved among myosin-VI homologs (36). Using extracts from rat kidney, myosin-VI was shown to bind F-actin in an ATP-sensitive fashion (36). Myosin-VI motility has not been observed yet, however, perhaps due to regulation by phosphorylation of the heavy chain at a conserved threonine residue (TEDS site) similar to what is seen for Acanthamoeba myosins-I (11,36).
Myosin-VI is ubiquitously expressed, but recent genetic studies suggest that its most critical function is in the maintenance of the cells of the inner ear. Avraham and colleagues (37) were interested in identifying the mouse recessive deafness gene, Snell's waltzer (sv). sv mice are profoundly deaf due to a complete degeneration of the neuroepithelium of the cochlea and exhibit head tossing and circling phenotypes characteristic of vestibular dysfuncion. Positional cloning of sv identified it as the gene for mouse myosin-VI, encoding a polypeptide 91% identical at the amino acid level to porcine myosin-VI (37). Myosin-VI protein was located within the neuroepithelium of the cochlea and was found to be expressed by the inner and outer hair cells of the organ of Corti (37). Within these cells it was seen to be enriched at the cuticular plate, an actin-rich structure at the bases of the stereocilia (37). This location is reminiscent of previous immunolocation studies, which showed that myosin-VI was enriched within the subapical terminal web region of brush border containing polarized epithelial cells (36,38). These locations have suggested a role for myosin-VI is membrane movements, endocytosis, or membrane recycling events within polarized cells.

Myosin-VII
Myosin-VIIa was recently identified as the gene defective in mouse shaker-1, a recessive deafness mutant (39) and in human Usher syndrome type 1B, a recessive disease characterized by congenital deafness, vestibular dysfunction, and retinitis pigmentosa (40). Since these genetic studies, the complete cDNA of human myosin-VIIa has been characterized. 3 As diagrammed in Fig. 1, myosin-VIIa has an N-terminal motor, five IQ motifs in its neck domain, and a short coiled-coil domain at the proximal portion of the tail. The remainder of the tail is comprised of two large ϳ460-amino acid repeats. Each repeat contains two domains, which have been called the MyTH4 domain and the talin-homology domain. 3 The MyTH4 domain, for myosin tail homology 4, is shared by myosin-VIIa and Acanthamoeba high molecular weight myosin-IV and is of unknown function. The talin-homology domain, however, is a conserved sequence shared by members of the band 4.1 superfamily and has been shown for both talin and band 4.1 to bind to acidic phospholipids (reviewed in Ref. 42). These sequence homologies suggest that the tail of myosin-VIIa targets the motor to the plasma membrane, as is seen for ezrin, moesin, and other band 4.1 members.
Immunolocation studies have focused on the expression of myosin-VIIa within tissues affected in Usher syndrome. In the cochlea, myosin-VIIa was expressed exclusively by the sensory hair cells of the organ of Corti. Within the inner and outer hair cells, myosin-VIIa expressed within the stereocilia, as well as within the cuticular plate and throughout the cell body (43). This is distinct from the location of myosin-VI, which was not located to the stereocilia, and from myosin-I␤, the putative adaptation motor, which was specifically located to the stereocilia tips. Since myosin-VIIa is expressed along the entire length of the stereocilia and is potentially a membrane binding protein, myosin-VIIa may serve a structural role, as is seen for brush border myosin-I, tethering the polarized actin filaments of the stereocilia to the plasma membrane.
In the rat retina, myosin-VIIa was expressed exclusively by the pigmented epithelial cells (43). This result suggests that the reti-  nitis pigmentosa observed in Usher patients is due to a defect in this cell layer. This is quite unique, as all previously identified retinitis pigmentosa genes are defects within the phototransduction machinery (44). Within the retinal pigmented epithelial (RPE) cells, myosin-VIIa was enriched at the apical plasma membrane (43). On the apical surface of RPE cells are villi, actin-rich structures involved in phagocytosis of spent photoreceptor outer segments. This phagocytosis is essential for photoreceptor viability, as defects in this process have been shown to lead to retinal degeneration in the RCS rat (45). In these animals, cellular debris accumulates in the intercellular space between the outer segments and the RPE layer, leading to photoreceptor cell death. Perhaps myosin-VIIa serves an essential role in this phagocytic process.

Myosins-IX and -X, Links to Cell Signaling Pathways
Myosins-IX motors are unusual in that they have an N-terminal extension, which is distinct from all other myosins as well as an insert of 150 aa in the actin binding loop (reviewed in Ref. 1), suggesting that the class IX myosins may have unusual mechanochemical properties. Also of interest is a striking tail domain, which contains a chimerin-like apposition of a zinc-binding motif and a GTPase-activating protein (GAP) domain (Fig. 1). This sequence homology suggests that myosins-IX may serve as GAP proteins for the Rho/Rac family of Ras-like GTP-binding proteins. As reviewed recently (46), assays performed using bacterially expressed fusion proteins containing domains from Myr5, a rat myosin-IX, have shown that the tail does indeed bind zinc, but, unlike chimerin, the Myr5 GAP domain preferentially activated the GTPase activity of Rho, and only marginally activated Rac1 (47). Rac and Rho have both been shown to be critical in the membrane reorganizations involved in the induction of membrane ruffling and the assembly of focal adhesions in fibroblasts. Myosins-IX may therefore form a direct link between the actin cytoskeleton and Rho-dependent signaling pathway (reviewed in Ref. 48).
Recently a human class IX cDNA was identified, myosin-IXb (49). Myosin-IXb is highly homologous to Myr5, except at the Cterminal 100-aa tip of the tail suggesting that myosins-IX are alternatively spliced. Expression of myosin-IXb was characterized in a myelocytic cell line HL-60, where it was shown that myosin-IXb exhibited an increase in expression level and a redistribution in location upon differentiation of these cells from promyelocytic to macrophage states (49). This suggests a role for myosin-IXB in the G-protein-coupled actin-based changes in cell shape observed upon differentiation in myelocytic lineages.
Another unconventional myosin with a likely association with G-protein pathways is myosin-X, the most recently identified vertebrate myosin class. First identified in frog (50), myosin-X has now been completely cloned from cow 4 and is shown schematically in Fig. 1. The tail of myosin-X contains a short portion of coiled-coil as well as three pleckstrin homology (PH) domains. PH domains have been found in a large number of proteins that interact with both the membrane cytoskeleton and signal transduction machinery (reviewed in Ref. 51). PH domains have been shown to bind to heterotrimeric GTP binding proteins as well as to phosphatidylinositol 1,4,5-bisphosphate (41), so, like myosins-IX, myosin-X has the potential to play a role linking the signal transduction pathway to the actin cytoskeleton.

Conclusions
Clearly, actin-based motilities are critical for the proper function and development of such diverse cell types as sensory hair cells, melanocytes, retinal epithelia, and neurons. In addition to their presumed functions in motile processes, the studies described here highlight the potential roles of unconventional myosins in signal transduction and in direct mechanochemical regulation of membrane channels.