Myosins are a large family of actin-based motors that play fundamental roles in eukaryotic motility (Sellers, 1999). They are implicated in a number of important cell functions, including cytokinesis and organelle transport and trafficking. Myosins are also important in cell polarization (Yin et al., 2000) and signal transduction (Bähler, 2000). Each of the 18 known classes of myosin (Furusawa et al., 2000) is characterized by the presence of a heavy chain with a conserved head domain, generally near the N-terminus, followed by an alpha-helical light-chain-binding region consisting of one or more IQ motifs, and a C-terminal tail. N-terminal extensions are also characteristic of a few classes of myosin.
The functions of each class of myosin have been shown to be distinct. Class I is implicated in endocytic and exocytic membrane traffic, whereas class II, or conventional myosin, is known to be a component of the contractile ring in dividing cells and the sarcomere in muscle cells. Myosin III is localized to the photoreceptor cells in the retina and functions in signal transduction (Montell and Rubin, 1988). Myosin V and XI have been shown to be organelle motors in animals and plants, respectively. Myosin VI has been shown to move toward the minus end of actin filaments; therefore, it is a reverse motor (Wells et al., 1999), and one that is important for development in many multicellular organisms (Mermall and Miller, 1995). The 12 other classes of myosin all possess specific functional properties, although most have not been characterized biochemically.
Traditionally, these distinct myosins are grouped based on phylogenetic analyses of their respective head domains (Goodson and Spudich, 1993; Cheney et al., 1993; Hodge and Cope, 2000). Head sequences are used because myosins are defined by the actin-dependent MgATPase activity of their head domains (Cheney et al., 1993), and because the conserved head sequences allow phylogenetic comparison. Recent studies of the class-specific tails yield results parallel to those derived from head domain analysis. Both approaches segregate the known myosins similarly across all 18 classes (Korn, 2000). This suggests that the head, neck, and tail domains of myosin heavy chains have coevolved and are predicted to be functionally interdependent (Korn, 2000). However, despite efforts to include all of the extant myosin genes into this imposed class structure, there remain a few myosins that evade classification, possibly because they represent novel classes or divergent members of existing classes (Berg et al., 2001).
The protein sequence databases now contain protein sequences for well over 200 myosins from about 65 different species widely distributed across the taxonomic spectrum. Several studies have used compilations of these sequence data to produce phylogenetic trees that organize each myosin in relation to all known myosins. Such phylogenetic trees enable conclusions to be drawn about biological function based on similarities among proteins.
Over the past 8 years there have been a number of important contributions to the overall phylogenetic and evolutionary study of the myosin superfamily. Goodson and Spudich (1993) and Cheney et al. (1993) proposed an expansion of the myosin classifications beyond myosins I and II, based on distance matrix analysis of sequences available at the time. Moreover, this study challenged the then-prevailing notion that myosin heads were freely interchangeable units whose function was uniquely determined by the tails to which they were attached. About 5 years later, in the midst of burgeoning knowledge of myosin diversity, a study by Mermall et al. (1998) integrated the functions of the various myosins into a coherent, interconnected picture. In addition, this study contained a phylogenetic analysis of the then-identified 14 classes of myosin. A year later, Oliver et al. (1999) published a comprehensive review of myosins and their functions in multicellular animals. That work pulled together important recent information, including the discovery of Myosin Tail Homology 4 (MyTH4) and FERM (band 4.1, Ezrin, Radixin, Moesin) domains in myosin, and other relevant biochemical information published by researchers in the field. Hodge and Cope (2000) produced a myosin family tree based on phylogenetic analysis of the myosin head domain, neatly classifying the known myosins. Similarly, Korn (2000) analyzed the head and tail segments of the myosins independently and was able to show that the head and tail domains of the myosins coevolved. Sellers (2000) also published an important review, delving further into the relatedness of the myosin classes. Lastly, Berg et al. (2001) published a review that incorporated all of the latest data and began one of the first analyses of the evolution, structural relationships, and functional similarities of the myosin class.
However, these approaches do not address the important question of the evolution of the myosin superfamily. In order to relate gene family phylogeny to evolutionary events, it is necessary to tie the phylogenetic data to the taxonomic groups of organisms represented in the sequence data matrix (Thornton and DeSalle, 2000). In the present study we performed such an analysis, which to our knowledge has not been done previously in the field of molecular motors. Thus, it is both appropriate and important that an evolutionary study of the myosin superfamily be undertaken to determine when and how the various individual classes of myosin evolved.
Sequences and Accession Numbers
From the databases, 232 myosin heavy-chain amino acid sequences were downloaded and organized into 19 groups representing the 18 classes of myosin (groups 1–18) plus one additional category (group U) for those sequences that could not be readily classified. Most of these sequences came from database entries accessed through hyperlinks available on the myosin homepage (http://www.mrc-lmb.cam.ac.uk/myosin/). A number of other sequences came from accession numbers in works by Berg et al. (2001) and Korn (2000). The remainder of the sequences were located through a number of GenBank searches, using terms such as “myosin” and “myosin heavy chain” (Benson et al., 2000; Wheeler et al., 2000). A complete list of accession numbers for sequences used in this study is included in Table 1.
Table 1. Accession numbers used, grouped by class and organism
Taxonomic Classification of Organisms Used in This Study
The taxonomic classifications of each organism for which sequence data is represented in this study were obtained by using the taxonomy browser available at the NCBI website (Benson et al., 2000; Wheeler et al., 2000). Translation of the taxonomic groups into a branching order yielded a rooted, unresolved tree of the hypothetical evolutionary history of these organisms. The basic branching structure of this tree was then refined by comparison with the “Tree of Life” available online through the University of Arizona (http://phylogeny.arizona.edu/tree/life.html). Branching structure was also refined and edited based on the phylogenetic information available through the University of California Museum of Paleontology website (http://www.ucmp.berkeley.edu/alllife/eukaryotasy.html). The precise basal metazoan branching pattern was further supported by recent findings (Medina et al., 2001). In addition, the relationships between the fungi, plant, nematode, arthropod, and chordate clades were supported by an analysis of 75 nuclear genes by Wang et al. (1999). Vertebrate branching order was supported by two independent studies (Felsenstein, 1985; Venkatesh et al., 2001), and mammalian phylogeny was supported by data from Murphy et al. (2001). Plasmodium phylogeny was based on a published study by Rich et al. (1998).
It is important to note that the deeper (more ancient) branches of this tree are not as reliable as the more recent ones because early events in eukaryotic evolution are uncertain, and there remains considerable debate over the precise divergence pattern of crown eukaryotes (Baldauf and Doolittle, 1997; Philippe and Adoutte, 1998). However, different phylogenies based on data from comparisons of small-subunit rRNA (Woese, 1996), chaperonin 60 and heat shock protein 70 (Roger, 1999), elongation factor 1-α (Baldauf and Doolittle, 1997), and RNA polymerase II (Stiller et al., 1998) agree with the general branching structure of the proposed tree. Three notable exceptions are Acanthamoeba castellanii (Ac), Tetrahymena thermophila (Tt), and Dictyostelium discoideum (Dd). Based on these phylogenetic studies and the aforementioned phylogenetic data available on the Web, the Ac, Tt, and Dd organisms group differently depending on the type of analysis performed. Of the three, the branching location of Dd, post-divergence of plants, seems to be the best-supported phylogenetic hypothesis. The only dissenting study in this case is based on what now appears to be misleading rRNA data (Baldauf and Doolittle, 1997; Abouheif et al., 1998). However, the placement of Ac and Tt is much less reliable, and therefore, conclusions from data associated with these organisms must be taken lightly.
Overall, the constructed phylogenic tree of the 65 organisms used in this study is reasonable and well supported, although debate remains over the deep-level phylogeny of the eukaryotes (Baldauf et al., 2000).
Estimating Evolutionary Relationships of Myosins
The combination of organism phylogeny and classes of myosins available for each organism was used to infer when each class of myosin first appeared. The criterion of “probable appearance” was applied to the complete data set contained in this study. The results are shown in Figure 1. This approach, sometimes referred to as the cladistic approach, has been used by others to infer the phylogenetic relationships of ray-finned fish (Venkatesh et al., 1999), the evolution of vertebrate HOX clusters (Amores et al., 1998), and the historic relationship of SINE genes (Shimamura et al., 1997).
For a probable appearance to be assigned to a class on a given node, three conditions must be met: 1) At least two of the branches from that node must each contain at least one instance of the particular class being considered. 2) The class being considered cannot be found outside of the clade defined by the node under question. This condition assumes the absence of lateral gene transfer, which is a reasonable assumption to make for this family of proteins. 3) If a class is unique to a given leaf (branch terminus) of the tree, then a probable appearance cannot be bound to a given node. Instead, the probable appearance is assigned to the leaf itself. It should be noted that the probable appearance of each class of protein on a given tree is as reliable as the assignment into classes of the proteins themselves. It should also be noted that these criteria assume a significantly higher probability of a gene-loss event than the independent creation of the same protein class in two different species.
The presence of every myosin class was analyzed in the context of a phylogeny of organisms. The resulting tree is shown in Figure 1. This figure shows the emergence of each class of myosin during the “branching history” of these organisms. Using this technique, it is possible to assign an evolutionary order to the appearance of each class in the myosin family.
Myosins With Broad Distribution
Myosins I and II evolved the earliest of all the myosin groups, and are found in amoebae, fungi, worms, flies, and other eukaryotes. Myosin II in particular is widely found. Of the 65 organisms used in this study, 40 were identified as having at least one class II myosin. This broad distribution and early evolutionary appearance of the class II myosins underscores their importance. Similarly, myosin I was found to be an ancient and important member of the myosin superfamily. Like myosin II, it is also present in both lower and higher eukaryotes. However, myosin I is not as broadly distributed as myosin II. Only 14 of the 65 organisms studied have been found to possess a class I myosin. Yet, for a number of organisms, several members of the myosin I family are expressed for a variety of functions. In Dictyostelium, at least six different myosin I proteins cooperatively perform functions as varied as endocytosis (Novak et al., 1995; Jung et al., 1996), morphogenesis (Novak et al., 1995), chemotaxis (Ostap and Pollard, 1996), and secretion (Temesvari et al., 1996). In vertebrates as well, myosin I is involved in a variety of important membrane-based phenomena (Allen and Aderem, 1995; de Oca et al., 1997; Poucell-Hatton et al., 1997).
Myosin XI appeared before the divergence of plants and fungi/metazoa. It is virtually ubiquitous in plants and is also found in Dictyostelium. However, it is absent in both fungi and metazoans. In its stead, myosin V pervades the fungal and metazoan clades, appearing before the divergence of the two. Remarkably, myosins V and XI share similar sequences, structural features (Kinkema and Schiefelbein, 1994), and functions (Berg et al., 2001). This suggests that myosin XI and V represent two distinct members of a broadly defined V/XI class. Members of this larger V/XI class appear in both plants and animals, indicating that it arose before the divergence of these two branches of eukaryotes. Such relationships suggest that the myosin V/XI class is ancient, widespread, and important.
Class VII myosins evolved after the offshoot of plants, but before the actual divergence of fungi and metazoans. This makes class VII the fourth most ancient myosin. It is present in a diverse set of organisms, including slime molds, worms, flies, fish, frogs, and humans, and plays an important role in each one. Sahly et al. (1997) demonstrated that myosin VII is expressed in mice exclusively in epithelial cell lines, coupled with the appearance of cilia and microvilli. Wolfrum et al. (1998) independently tested and confirmed these results by immunoelectron microscopy. In addition, artificial constructs of the MyTH4 domain, present twice in the tail of myosin VII, have been shown to bind to microtubules (Narasimhulu and Reddy, 1998). These results, coupled with the early appearance of the class, suggest that myosin VII plays important roles in animals.
Myosins VI and IX appeared during the early periods of metazoan radiation and are present in a variety of organisms, including worms, fish, and mice. Class VI in particular is found in most of the major metazoan clades. Myosin IX is not quite as widely represented as myosin VI; however, it too is found in many of the major metazoan clades.
Around the time of the appearance of classes VI and IX, there was a tremendous explosion of diversity, termed the “metazoan radiation.” Exon shuffling acquired major significance at this time, and has been implicated in an acceleration of modular protein evolution (Patthy, 1999). This has significant relevance, as the diversification of metazoan lineages also saw an acceleration in the diversification of the myosin superfamily.
Myosins III, XV, and XVIII appeared during the metazoan radiation, prior to the divergence of arthropods and mammals, but after the split of nematodes. Each of these classes is found in both fruit flies and humans. Myosin III is also found in fish. However, myosins III, XV, and XVIII are not well represented among other organisms, unlike the above-mentioned classes, which are well represented. Indeed, this study found only six instances of myosin III, five of myosin XV, and three of class XVIII. Moreover, these myosins were found in a total of five, three, and two species, respectively. Thus, myosins III, XV, and XVIII appear to have evolved much more recently than other groups of myosin.
Class X myosin appeared during the formative phases of vertebrate evolution, prior to the divergence of amphibians, birds, and mammals. Myosin X is found exclusively in frogs, cows, humans, and mice. Thus, it is a relative newcomer on the evolutionary scene.
Myosins With Limited Distribution
According to the current data, myosin XVI appeared very recently, sometime during the evolution of mammals. The class has only been found in two species: Homo sapiens and Mus musculus. Functionally, the class is important for neuronal migration and brain development (Patel et al., 2001). Myosin XVI appears to be an evolutionary infant, and is unique and specific in both its structure and function.
Myosin VIII appeared prior to the radiation of terrestrial plants, but after the divergence of green algae. Members of the myosin VIII class were found in all of the plants analyzed in this study, suggesting that it is an important component of plant physiology. Indeed, myosin VIII appears to be important for the maturation of the cell plate and the reestablishment of cytoplasmic actin cables at the sites of intercellular communication (Reichelt et al., 1999).
Myosin XIII appears to be limited specifically to Acetabularia cliftonii and evolved after the divergence of Chlamydomonas. Likewise, myosin IV appears to be organism-specific. This class likely appeared sometime during the evolution of the Acanthamoebae, post divergence from the rest of the eukaryotic lineages. Myosin XIV is another example of a class limited in scope. Group XIV evolved just prior to apicomplexan radiation and it appears to be the exclusive form of myosin for apicomplexan parasites (Heintzelman and Schwartzman, 2001).
Myosin class XVII evolved somewhere within the diversification of the fungi. The precise appearance of the class is impossible to pinpoint without a broader sampling of fungi. However, based simply on the presence of myosin XVII in the Magnaporthe and Emericella genuses, one would predict that the class appeared sometime prior to the diversification of the euascomycota, a subfamily of the ascomycetes (sac fungi). The class, which consists of a chitin synthase fusion protein, has metabolic activity specific to the fungi, and thus it is reasonable that it should be limited to fungi (Fujiwara et al., 1997).
Myosin XII appeared sometime late during the evolution of Caenorhabditis elegans, currently the only species in which this class is found. Like classes IV and XIII, myosin XII represents a species-specific protein.
Relating the myosin classes in an evolutionary context enables one to predict those classes of myosins that should be present for a given species. The absence of a member of a class where one is predicted to occur is indicative of either “missing” data (genes that have not been found yet) or a gene-loss event (wherein a deleterious mutation removed a gene from an organism's lineage). Table 2 provides a complete list of all of the classes of myosin that are absent, based on the time of appearance indicated on the branches of the tree in Figure 1.
Table 2. Compilation of the myosin classes that are present and those that are absent from each organism in this study*
Myosin(s) absent due to gene loss
Myosin(s) absent due to missing data
The absence of a myosin class may be due to either a gene loss event or missing (incomplete) data.
Denotes that the organism is not well studied.
Denotes that the organism's genome has been sequenced.
(2) Denotes sequence data that appears to be myosin II, but is unconfirmed.
Neither myosin I nor II has been found in plants (Yamamoto et al., 1999) or alveolate protists (Heintzelman and Schwartzman, 2001), despite extensive searches to try to locate instances of these classes. It is thought that the absence of conventional (class II) myosins in plants is tied to their different mode of plant cytokinesis (whereby the two daughter cells are separated not by the action of a contractile ring, as is the case in animals, but rather by the construction of a cell plate (Field et al., 1999)). The apicomplexa also lack myosins II and I. This is understandable in light of the fact that Dictyostelium cells are able to undergo traction-mediated cytofission, despite the absence of myosin II (De Lozanne and Spudich, 1987; Spudich, 1989; Pasternak et al., 1989). Furthermore, on an adhesive surface, these traction-mediated cytofission events are coupled to mitosis (Zang et al., 1997; Neujahr et al., 1997). However, on a non-adhesive hydrophobic surface, Dictyostelium cells do not divide properly (De Lozanne and Spudich, 1987; Knecht and Loomis, 1987; Manstein et al., 1989). These findings suggest that while myosin II is advantageous for cytokinesis, it is not essential. Thus, a loss of the gene for myosin II in the plant lineage may have been accompanied by the transition from fission to cell plate construction.
The loss of class I myosins from plant and alveolate lineages is also explicable. As previously mentioned, myosin I plays a variety of important roles in Dictyostelium and vertebrates. However, there are only two class I myosins each in Drosophila and C. elegans. Drosophila does not even possess an SH3 domain-containing member of the class. This suggests that myosin I may function as a generalist in most metazoan lineages, whereas proliferation of the group in Dictyostelium and in vertebrates allowed for occupation of various functional niches. Thus, loss of a general myosin I in the plant and alveolate lineages is understandable.
Conversely, myosin XI is not found outside of the plant kingdom. Although class XI appears in the lineage of all fungi and metazoans, it has not been found in any animal genomes. Instead, a similar class of myosins (class V) appears to have replaced the class XI in fungi/metazoa.
Fungi also appear to have lost class VII myosins. The Saccharomyces cerevisiae (Sc) genome has been sequenced and well studied. However, no instances of class VII have turned up in the genome. The same is true of Schizosaccharomyces pombe (Spo). Because Sc and Spo represent ancestral branches prior to the appearance of the ascomycetes, it should also hold that other fungi lack class VII myosins as well.
Likewise, platyhelminthes, arthropods, and molluscs all lack instances of class IX myosins. Given that the Drosophila genome is complete and has been inspected for putative myosins (Adams et al., 2000; Goldstein and Gunawardena, 2000), and given that Drosophila has been well studied biochemically, it seems unlikely that myosin IX is found in flies. Because the platyhelminthes, arthropods, and molluscs are all related, it would seem that class IX myosins were lost from these three phyla prior to their divergence from higher eukaryotes.
Absence of Myosin Due to Missing Data (Not Yet Detected, or Unknown)
According to the data in Figure 1, some of the myosins predicted to be present in certain organisms are “missing” from the databases. That is, the classes have not yet been identified in these organisms, despite predictions that they exist. A summary of missing myosins is provided in Table 2. Surprisingly, these assumptions of incomplete information apply to a number of fairly well known organisms.
First, Mus musculus is missing classes III and XVI. Both of these classes are found in the evolutionary history of the species. Thus, these classes should be found in mice unless a gene-loss event eliminated these genes from the organism's history. However, because the closely related Rattus norvegicus has class XVI myosins, it seems likely that myosin XVI should also be found in mice. Similarly, myosin III is found in humans, which are close relatives of mice. This would appear to indicate that mice have this class as well.
Rattus norvegicus is also missing a number of classes of myosin. Many of these classes are found in mice and humans. Therefore, it is expected that these genes will be identified when the rat genome is fully sequenced.
Frogs and zebrafish are also missing a few widespread classes of myosin, including classes I and V (chickens contain both class I and class V). These organisms are fairly well studied, but an exhaustive search for myosin classes has not been conducted. It seems likely that these organisms possess a number of myosin classes for which the data are missing from the current databases.
Both Acetabularia cliftonii and Oryza sativa lack an instance of myosin XI. Each of these species contains the class within its evolutionary history, so each is predicted to possess class XI. This assumption is supported by their close relatedness with other algae and plants, respectively.
In addition, the higher fungi are missing class II and class V myosins. Magnaporthe grisea is also missing class I myosin. The higher fungi are close relatives of yeast and should contain similar genetic information. It is thus predicted that higher fungi should have classes II and V myosin.
Lastly, there are a large number of other organisms for which data are missing. For instance, data are missing for a number of groups of myosin that are predicted to be present in many of the less studied platyhelminthes, molluscs, fish, and mammals. A complete inventory is provided in Table 2.
It is highly unlikely that the myosin classes were a product of 18 separate de novo evolutionary lineages. Instead, the classes of myosin must have shared a common ancestor. The motor head domain is a trait common to each class. The IQ motif is also widespread. Thus, it appears that a myosin ancestor contained a head domain plus one or two IQ motifs.
However, the tails of the various myosins are extremely divergent. Some tails contain domains that are common to other protein families as well. For instance, the SH3 domain found in myosins I, IV, VII, and XV is a common feature of proteins involved in signal transduction related to cytoskeletal organization. The PH domain is found in myosin X, in addition to a number of other proteins involved in signal transduction pathways, including phospholipase C-gamma (Falasca et al., 1998) and protein kinase B (Tanaka et al., 1999). It is exceedingly unlikely that these domains arose de novo in myosin tails. The divergent tail structures and corresponding functions imply dramatic changes in sequence over time. Such changes are not probable via a normal evolutionary scheme based on point mutations and frame shifts. Indeed, research on genome evolution has shown that de novo generation of gene segments coding for protein domains is rare. Most genome novelties are produced by gene duplication, exon shuffling, and retrotransposition of genes (Levinson et al., 1990; Long and Langley, 1993; Chen et al., 1997a, b; Naas et al., 1998; Nurminsky et al., 1998; Long et al., 1999; Brosius, 1999; Moran et al., 1999; Courseaux and Nahon, 2001). The myosin superfamily is no exception to this. Over time, the myosin head domain found itself in different regions of DNA, which led to changes in the structural features, localization, expression, regulation, and function of the gene.
Figure 2 presents a hypothesis of the evolution of the myosin superfamily, based on the principles of evolution that could have led to such diversity. These relationships are not reported as conclusive statements concerning what happened over the course of evolution of the superfamily. Instead, they are presented as an evolutionary hypothesis, supported by phylogeny and structural comparisons.
Inferred Evolutionary Relationships of the 18 Known Classes of Myosin
Myosins I and II both emerged early in the evolutionary history of eukaryotes. Class I appears to have given rise, both directly and indirectly, to most of the myosins that exist today. The class II myosins appear to have given rise only to the myosin V/XI lineage, which consists of only four classes. These evolutionary relationships have been predicted here based on protein structure, function, localization, approximate evolutionary appearance, and a number of phylogenetic comparisons of the head and tail domains of each class.
Myosin II Lineage
Myosin XI is likely to have been derived from myosin II, rather than myosin I. This relationship is not certain, but a number of factors point to class II myosin as the progenitor of myosin XI. Myosin XI and myosin II share coiled-coil regions in their tail domains and similar N-terminal extensions. The myosin II coiled-coil is much more extensive than the corresponding domain in myosin XI. Nonetheless, the two classes share this feature. Furthermore, myosin I lacks an N-terminal extension, thus class II is the better candidate for the myosin XI precursor. In addition, a number of phylogenetic analyses of myosin motor domains group myosin XI more closely with myosin II than with myosin I. Studies by Patel et al. (2001) and Mermall et al. (1998) have shown this relationship. Thus, in preference to myosin I, myosin II appears to be the progenitor of myosin XI.
Myosin V and myosin XI have similar sequences and structural properties. Moreover, their functions are related: both act as organelle motors. The concurrent appearance of myosin V and disappearance of myosin XI in animal lineages indicates that the two are divergent members of the same evolutionary ancestor. That is, an ancient class XI gene was retained in plants, but was altered substantially in animals. This phenomenon is the most probable explanation for the appearance of class V myosin.
Myosin XIII and myosin XI are also clearly related. In a phylogenetic analysis of the core motor domains of myosins performed by Sellers (2000), myosin XIII and XI grouped together in >80% of the bootstrap trials performed. The Hodge and Cope (2000) myosin family tree also groups class XIII closely with class XI. Moreover, the domain architecture of the two classes is fairly similar (see Fig. 2). Lastly, class XI appears earlier in the lineage of Acetabularia than does class XIII. Thus, class XI could easily have served as a precursor to myosin XIII. This is indicated on the tree of myosin evolution presented in Figure 2.
Myosin VIII is also related to myosin XI. Not only do the two classes appear in the same plant lineages, but they are also similar to each other. Domain architecture similarity is the first indication of this. Additionally, phylogenetic analysis reproducibly groups myosin XI and VIII together. Previous studies (Hodge and Cope, 2000; Berg et al., 2001) ascertained the same phylogenetic relationships between the two classes. Thus, it can be concluded myosin XI served as a precursor to myosin VIII.
Myosin I Lineage
Myosin IV and myosin I appear to be related. The trees developed by Mermall et al. (1998) and Hodge and Cope (2000) demonstrate that myosin I links loosely with class IV. Moreover, the presence of an SH3 signaling domain in the tail of class IV indicates that it probably evolved from myosin I, which also contains an SH3 domain at its C-terminus (Horowitz and Hammer, 1990). Recall, however, that the phylogeny/placement of class IV is uncertain. Hence, this particular conclusion must be taken with a grain of salt. Nonetheless, this relationship is partially independent of the branching location of the Acanthamoeba lineage, and thus is important to note.
Myosin VI appears to have evolved from myosin I. Head domain comparisons performed by Patel et al. (2001) show a relationship between the two classes. This is also distantly supported by other analyses (Mermall et al., 1998; Sellers, 2000). Further, all myosin VI molecules have a threonine residue at the TEDS rule site (Bement and Mooseker, 1995), similar to lower eukaryotic myosin I (Sellers, 2000). Although the evidence is inconclusive, it appears that myosin VI may have evolved from myosin I.
Class VII appears to have been derived from myosin I as well. Like myosin I, class VII also contains an SH3 domain. In addition, myosin VII contains two MyTH4 domains and two FERM domains. Neither of these domains is present in myosin I. However, the domain architecture of class IV myosin includes a MyTH4 followed by an SH3 domain. This similarity of domain architecture is hardly coincidental. Rather, class VII appears to share common features with class IV. However, myosins IV and VII are not found within the same evolutionary lineages. Thus, it seems that a common progenitor containing a MyTH4 and SH3 domain architecture yielded both myosin IV and myosin VII. Indeed, it has been shown that the MyTH4 domains of class VII and IV myosins are similar (Sellers, 2000).
Myosin XIV appears to have evolved from myosin I in the alveolate protist lineage. In fact, the disappearance of myosin I from alveolates coincides with the appearance of myosin XIV. Thus, it seems likely that myosin XIV is a divergent form of myosin I. The Hodge and Cope (2000) myosin family tree shows a correlation between the myosin I and XIV head domains. Evolutionarily, the only other myosin available as a precursor to class XIV was class II. However, class II repeatedly groups distantly from class XIV. The relationship between myosin XIV and myosin I appears to hold true, but is fairly inconclusive.
Myosin X appears to have evolved from myosin VII. The similarities of protein domain architecture between the two are striking. Both genes contain a C-terminal MyTH4/FERM domain pair. It is unlikely that this similarity is the product of two separate de novo evolutionary events. Rather, phylogenetic comparison groups the two classes closely. For instance, previously published trees (Hodge and Cope, 2000; Berg et al., 2001) show a clear correlation between classes VII and X. Over the course of evolution of class X myosin, it appears that a region of DNA encoding for pleckstrin-homology domains “jumped” into the myosin VII sequence from which class X came. This phenomenon would explain the homology of the head domains of myosin X and VII, as well as the C-terminal MyTH4 and FERM similarities.
Myosin XV also appears to have evolved from class VII. The domain architecture of myosin XV is extremely close to that of myosin VII. Both genes contain two MyTH4 domains followed by FERM domains. These two MyTH4/FERM domain groups surround an SH3 domain found in both myosin XV and myosin VII. Moreover, the spacing between the two MyTH4 domains is conserved across the myosin XV/VII divide (Oliver et al., 1999). Phylogenetic analysis of the head domains of myosins XV and VII also groups them closely. In a number of phylogenetic trees (e.g., Hodge and Cope, 2000; Berg et al., 2001), myosins XV and VII group together reproducibly.
Myosin XII evolved from the class VII lineage as well. Both proteins contain structural similarities, including two MyTH4 domains that are spaced roughly the same number of amino acids apart. In addition, phylogenetic analysis by Mermall et al. (1998) demonstrated that the head domains of the two are related. Thus, it appears that myosin XII evolved from myosin VII in the C. elegans lineage.
Class IX myosin may also have come from class VII. Though the tail domain structures between the two genes are not strictly related, head domain phylogeny repeatedly groups the two classes together. It seems likely that myosin IX diverged from myosin VII sometime early in metazoan radiation. However, in view of the domain structure of myosin IX, the appearance of the class is not likely to have been due to divergence over time. Rather, a radical change in head domain environment, caused by retroposition or other forces of evolution, must have produced the class IX myosins.
Myosin III may have evolved from class VII. Both proteins are found in the photoreceptor cells. Overlapping expression patterns are indicative of similarities between the two. However, the functions of the two proteins are clearly divergent. Myosin VII is thought to play a role in phagocytosis and the transport of opsin (Hasson et al., 1995; El-Amraoui et al., 1996; Liu et al., 1998; Liu et al., 1999), whereas myosin III links a well-known multiprotein signaling complex to actin filaments (Xu et al., 1998; Wes et al., 1999). In addition, tail domain structure is also divergent. Thus, any evolutionary event that could have produced these two distinct classes of myosin must have been major. Nonetheless, a phylogenetic analysis by Patel et al. (2001) of head domains indicates that myosin III is potentially related to myosin VII.
Myosin XVI is evolutionarily related to myosin III. The two classes group together in the Hodge and Cope (2000) analysis of the myosin head domains, as well as in other analyses by Patel et al. (2001).
Myosin XVII in fungi may be an evolutionary descendent of class VII myosins. The Hodge and Cope (2000) head domain comparisons show a relatedness of the two classes, which is also supported in other studies (Berg et al., 2001; Korn, 2000). However, this relationship is weakly inferred.
Myosin XVIII appears to be related to myosin VII in an evolutionary sense. This inference is supported by other phylogenetic trees (Hodge and Cope, 2000; Berg et al., 2001). It is difficult to say whether or not the appearance of class XVIII was a direct result of some evolutionary change that occurred in a class VII gene. Indeed, class XVIII may have arisen from class III. Either way, myosin VII appears to be an evolutionary ancestor to class XVIII.
General Observations on Domain Structure
It is interesting to note the relative simplicity of the domain structure of the early myosin genes vs. that of the later myosins in the class I lineage. Moreover, there appears to be a great diversification of the myosins during the time of metazoan radiation, coincident with the evolutionary appearance of many new classes of the protein. The appearance of so many new and structurally diverse myosins may thus have allowed for the expansion of the roles played by the various myosins. In fact, increased domain complexity in later myosin classes may also be tied to the more precise functional specificity of the proteins. In contrast, earlier myosins may function more as generalists, carrying out a number of important tasks in a variety of cell types.
It is highly likely that new and novel members of the myosin superfamily will surface over the coming months and years as research continues to expand our knowledge of this gene family. In addition, it is likely that a search for the “missing” myosins noted in Table 2 will turn up previously undiscovered classes and pseudogenes in a host of organisms. Moreover, as future studies refine our ideas of early eukaryotic evolution and as additional myosins are discovered in early eukaryote lineages, it will become possible to assess the validity of the hypotheses presented herein. It may also be possible to garner further insight into the evolution of this unique gene family.
The great diversity represented in the myosin superfamily must have come about through specific evolutionary events. Indeed, class I and II myosins appear to have served as templates for the evolution of the entire superfamily. However, there must have been a common precursor to myosins I and II, and that hypothetical precursor must have arisen based on preexisting gene segments. It is likely that this myosin progenitor was originally related to ancestral kinesin. A few years ago, this notion would have been flatly rejected. Prior to recent studies of the crystal structures of the two proteins, it appeared as though myosin and kinesins were unrelated. However, close overlap of the crystal structures of myosin and kinesin has pointed to a number of short stretches of well-conserved sequences within the core motor domains of the two proteins (Kull et al., 1996; Vale and Fletterick, 1997). This suggests that myosin and kinesin share a common ancestor. In fact, the myosin and kinesin lineages may be linked even further back in evolutionary history to a progenitor of the G-protein superfamily (Kull et al., 1998). Although there has been substantial divergence among the myosin, kinesin, and G-protein superfamilies, a highly conserved core mechanism underlies these proteins (Vale and Milligan, 2000). This conserved core has proven extremely versatile as a foundation for the evolution of a multitude of functionally diverse proteins. The mechanochemical relevance and potency, and the evolutionary diversity and descent of this long lineage of molecular motors is fundamental and is certain to be the focus of studies for years to come.
We thank Esther Warshauer-Baker and Jonathan M. Budzik for critically reading the manuscript. In addition, we thank Jeremiah R. Brown and Mark McPeek for feedback and suggestions. This work was supported by NSF grant MCB9974709 to G.M.L., and a Dartmouth College Presidential Scholars Internship to R.F.T.