The family of unconventional myosins is ever growing and the functions attributed to them seem to expand in parallel. These actin-based motor proteins have been implicated in processes as seemingly diverse as endocytosis and exocytosis, the transport of organelles, in spermatogenesis and in neurosensory functions such as hearing and sight. A common myosin function may underlie them all — the regulation of intracellular membrane traffic.
It is now clear that much of intracellular trafficking is powered by motor proteins: the myosins, kinesins and dyneins. This current view largely comes from greatly improved imaging techniques and the development of in vitro motility assays. Microtubule-based motility can be readily reconstituted and visualised in vitro, facilitating the initial isolation and characterisation of organelles and their associated motors. Actin-based organelle motility, in contrast, has proved more difficult to study and, as a consequence, has been underappreciated. However, the situation has changed recently and the motors of the myosin superfamily are now acknowledged to play important roles in many aspects of membrane traffic within the cell.
Myosins were actually the first motor proteins suggested to play a role in organelle movement, based on the elegant early analyses of cytoplasmic streaming in Characean algae (reviewed in ). It was not until 1992, when George Langford and colleagues developed an in vitro system that maintained the integrity of actin networks in extruded squid axoplasm, that actin-based transport of organelles was observed directly in vitro. Vesicles and organelles were seen to move on extruded F-actin networks in the absence of microtubules, but the movement ceased when inhibitors of actin polymerisation were added . The implication from these experiments was that members of the myosin superfamily, the only characterised actin-based motor proteins, must be driving the motility.
The myosin superfamily members share a common N-terminal motor domain which generates movement along actin filaments in an ATP-dependent process. The motor domain is connected to a tail by a flexible neck region. The myosins are subdivided into an ever-increasing number of classes (18 at the last count) according to phylogenetic analyses of the motor domains and are named with roman numerals (i.e. myosin III, IV, etc.), roughly according to their order of discovery. Class II myosins are often termed the ‘conventional myosins’, the best-studied example of which is skeletal muscle myosin — the standard by which other myosins are judged. All other myosins (i.e. those that are not class II myosins) are therefore called ‘unconventional myosins’. Generally, the tail domain structure is conserved within each class but differs between classes . In light of the high degree of conservation between the motor domains, the myosin tail is thought to determine the specificity of the protein, both for interactions such as dimerisation and for the type of cargo it will carry. In contrast to the motor domain, which has been extensively studied in both conventional and several of the unconventional myosins, the functions of the individual tail domains are less well understood .
The roles attributed to myosins are bewildering in their diversity. Family members have been demonstrated to participate in ameoboid cell motility, pinocytosis and phagocytosis, vesicle and organelle transport, the maintenance of cellular microstructures, hearing, sight and spermatogenesis to name but some ( Fig. 1). This review will concentrate on the functions of the unconventional myosins in membrane trafficking processes, in particular of classes I, V, VI and VII.
Vesicular Transport and Myosins
Even the simplest of eukaryotes express multiple myosins. The budding yeast, Saccharomyces cerevisiae, for example has five myosin genes: one conventional class II myosin and two each of classes I and V. Work in this system has provided perhaps the most compelling evidence that the class V myosins are involved in transport of vesicle. The first myosin V to be identified (the product of the MYO2 gene), is also the only yeast myosin essential for viability . A temperature sensitive allele of MYO2 (myo2-66) causes cells to arrest in an enlarged, unbudded state at the restrictive temperature, with an accumulation of secretory vesicles and disorganisation of the actin cytoskeleton . The vesicles do not reach their usual destination at the bud site, suggesting that Myo2p directs the polarised transport of vesicles along actin cables and into the bud. Through a series of genetic interactions with sec mutants, the accumulated vesicles were determined to be post-Golgi vesicles, placing the functions of Myo2p at a post-Golgi stage of the exocytic pathway. It is of interest to note that myo2-66 interacts largely with genes that encode proteins of the exocyst, a large, multi-protein complex that plays a role in the docking of secretory vesicles at the plasma membrane [5,6]. However, many proteins are secreted normally in these cells, so Myo2p is required for the secretion of only a subset of vesicles .
Myo2p has been localised to sites of polarised growth — the presumptive bud site, the neck of the bud during cytokinesis and at the tips of mating projections — and consistent with these locations, has been implicated in several other aspects of polarised growth in S. cerevisiae. These include directed secretion of chitin and transport of exocytic vesicles [4,5]. The partitioning of the vacuole (the yeast equivalent of a lysosome) between the mother cell and bud involves a process of polarisation along F-actin cables within the mother cell followed by transport of the vacuole to the presumptive bud site that can be visualised by the presence of a ring of F-actin . With some mutant alleles of myo2 (and also of actin) the movement of the vacuole towards the bud fails . One of the vacuolar inheritance mutants, myo2-2, the result of a single amino acid change in the C-terminus, causes a loss of Myo2p localisation from polarised sites of active growth as well as from the vacuole itself. Interestingly, in contrast to the myo2-66 allele, the myo2-2 allele is viable and does not exhibit any obvious accumulation of vesicles . The myo2-2 allele also does not exhibit any synthetic lethality with the exocyst genes (e.g. sec2, sec9, sec15) as the myo2-66 allele does . These observations suggest that this region of the Myo2p tail is important for Myo2p localisation and transport of the vacuole into the growing bud but not for polarised secretion of other materials. One obvious and attractive conclusion, therefore, is that Myo2p directly carries the vacuole along the actin cables towards the bud. However, given the myo2-2 phenotype, it is also possible the Myo2p plays a role in anchoring actin filaments at the site of organelle targeting and that loss of the appropriate actin organisation results in failed vacuole movement. Directed transport of post-Golgi vesicles and the vacuole to the bud is likely to be controlled by distinct domains within the tail of Myo2p.
Myo4p, the other myosin V in S. cerevisiae, has no apparent membrane trafficking function, however, it is required for mRNA transport into the bud [10,11]. Deletion of MYO4 prevents the accumulation of ASH1 mRNA in the daughter cell. ASH1 encodes an HO transcriptional repressor, so Myo4p is required for the restriction of mating type switching to the mother cell only. Myo4p is localised similarly to Myo2p (at the emerging bud tip) and may play a similar role in intracellular trafficking, with different determinants (presumably in the tail region) that dictate cargo selection. Molecular genetic analysis of these two myosins in yeast should yield key information about their targeting specificity.
Myosin V is also thought to play a role in organelle trafficking in neurons. Isolated organelles from both squid giant axons and brain contain tightly associated myosin V [12,13]. These organelles are capable of moving along actin filaments and, interestingly, this movement can be inhibited by the addition of function blocking antibodies. The myosin V-containing squid giant axon organelles have been shown to be derived from the endoplasmic reticulum  while the brain vesicles are a mixture of synaptic vesicles and an as yet unidentified population of organelles [12,14]. Thus, it is likely that this class of myosin plays a role in various aspects of intracellular trafficking.
Of Mice and Men (and Fish!)
Whilst yeast has proved to be an excellent system for analysing myosin V function in vesicle transport, it is not the only genetic model for myosin V. Two mammalian genetic disorders have been mapped to the myosin V locus: a rare human condition known as Griscelli Syndrome which results in loss of pigmentation of the skin and immunological and neurological disorders  and the dilute lethal mouse mutation, which causes a dilution of the coat colour, from black to grey, brown to beige, etc. and also neurological defects . In both cases, the pigmentation problems are caused by an abnormal distribution of the melanosomes — the delivery vehicles which transport melanin, synthesised in the melanocyte, to the keratinocytes. Under normal circumstances, melanosomes are found concentrated in the dendrites and dendritic tips of the melanosome. The keratinocytes then engulf the dendritic melanosomes by a phagocytosis-like process and the hair becomes pigmented. In Griscelli sufferers and in dilutelethal mice the melanosomes become concentrated in the centre of the cell and not in the dendrites [15,17]. Myosin V has been localised to mammalian melanosomes [17–19] and the loss of its function is believed to result in a failure of melanosomes to be transported to their correct intracellular location.
An extensive analysis of the patterns of melanosome movement within the melanocytes of wild-type and dilute mice has provided a potential explanation for the altered distribution in myosin V mutant cells [17,20] and has also highlighted the close co-operation that must exist between the microtubule-based and actin-based motility systems. Melanosomes (and their fish equivalents described below) have both been shown to move rapidly and bidirectionally on microtubules (see [20,21] and references therein). These movements occur over large distances in both the wild-type and dilute melanocytes (in some cases movement along a linear trajectory was seen to exceed 15 μm) and carry the melanosomes from the centre of the cell to the periphery. It is at the periphery that differences in motility were observed. In wild-type cells the melanosomes exhibit short range, slow movements in the actin-rich periphery, but these movements are generally absent in the dilute cells. It appears that microtubule motility is used to move the melanosomes rapidly outwards from the centre of the cell, but once there the microtubule motors hand over the responsibility for their cargo to myosin V which shifts the melanosome onto the actin network to retain it at the periphery. In light of these findings, it would be unsurprising if myosin V was found to be part of a complex with microtubule motors on the surface of the melanosomes (see below). Two possible roles for myosin V in contributing to the peripheral distribution of melanosomes have been postulated. Myosin V may move the particles along the randomly oriented actin filaments in the periphery to effect the random dispersal of the pigment granules  or play a role in ‘capturing’ the melanosomes at the actin-rich periphery  or may play both roles.
Supporting evidence for melanosome transport by myosin V comes from slightly different, but related cell systems — fish and amphibian melanophores. In these cells the distribution of the melanosomes is altered rapidly to change the colour of the organism in response to external stimuli. Recent work has shown definitively that melanosomes can and do move on actin networks, in addition to movement on microtubules [21,22]. Disrupting the microtubule networks with nocodozole does not abolish melanosome motility, but the additional loss of F-actin filaments does. Purified Xenopus melanosomes contain myosin V and are capable of moving on actin filaments extruded from Nitella. It will be interesting to see if this movement can be blocked by antibodies to myosin V, as demonstrated for ER vesicles moving on actin filaments from squid axoplasm  and if any other myosins are associated with these pigment granules.
Biochemistry of Myosin V as a Vesicle Motor
One problem when thinking about any motor protein as a vesicle transporter is the biochemical properties of its interaction with the appropriate filament system. The myosins that have been best characterised biochemically belong to classes I and II. Both are termed ‘short-duty’ motors. They bind to actin for only a short period during their duty-cycle so, in contrast to the highly processive microtubule motor kinesin, this type of myosin is likely to diffuse away from actin filaments when not attached [23,24]. For myosin II, this may not be a problem since it is organised in a tightly packed array (i.e. the thick filament) that allows for numerous brief interactions with actin, resulting in productive movement along actin. The single-headed myosin I may solve the problem by forming an as yet unidentified assembly of numerous myosins on either a membrane or on actin itself through one of the domains in its tail [23,25]. However, it is difficult to envisage how a myosin could transport vesicles over long distances in areas of few actin filaments without losing contact with those filaments if it were not present in a high local concentration. The movement of melanosomes on actin filaments described above was seen over considerable distances within melanocytes with a persistence of direction and was not limited to short, random movements that might be expected of a non-processive motor detaching regularly from the filaments .
Consistent with its predicted role as a vesicle motor, myosin V behaves differently from either myosins I or II. Firstly, unlike myosin I or II, a significant amount of purified myosin V binds to actin in the presence of 10 mM ATP. This suggests that myosin V can stay attached to actin throughout the duty-cycle . Secondly, its ATPase activity is activated at a lower actin concentration than for myosin II, a useful property for operating in areas of less concentrated actin . Recent direct observations using filament-sliding assays at high resolution revealed that the movement of actin filaments over tethered myosin V molecules is smooth and continuous — indicative of a processive motor. Furthermore, the average number of ATPase cycles that a myosin V molecule was able to complete without detaching from the actin filament was estimated to be 40–50 . The step size (the distance moved upon the actin filament in one cycle) of myosin V is large, which allows a substantial distance to be covered in fewer ATP-cycles. The step size for myosin V was estimated to be 30–38 nm, in comparison to 4–15 nm for myosin II and 8 nm for kinesin [27,28]. Myosin V, therefore, would appear to have the necessary biochemical properties to be the vesicle transport motor predicted from the available genetic data. It remains to be determined if a single head of myosin V is itself processive or if the double-headed nature of myosin V plays an integral role in maintaining the motor's attachment to the actin filament for long periods. Ongoing structural and biophysical analyses will no doubt resolve this question.
Myosin I is for Eating, Drinking and Digesting
Unlike myosin V, myosin I is not an obvious candidate for a long-duty motor. The molecules have only a single motor domain, have been shown to be monomeric in solution and do not have the same high affinity for actin when in the ATP-bound state as myosin V [23,29]. However, all myosin I class members share a positively charged membrane-binding domain and have been shown to bind to membranes with high affinity . Some possess a second ATP-independent actin-binding domain that may allow the cross-linking of actin filaments. The most likely role for the class I myosins is to move membranes along actin or actin filaments against each other.
Myosin Is have been implicated in several membrane traffic processes  ( Fig. 1). Study of myosin I function in lower eukaryotes, especially the amoebae Acanthamoeba and Dictyostelium, in yeast and in Aspergillus nidulans, has identified roles in pinocytosis, phagocytosis and endocytosis. Dictyostelium expresses six myosin I genes, none of which are essential . Where examined, it has been shown that most of these are localised to actin-rich cortical regions of the cell, such as at the leading edge of an extending pseudopod or in phagocytic cups . Cell lines in which a single myosin I gene has been targeted show no obvious alterations of the actin cytoskeleton, or of membrane trafficking processes, but loss of two or more myosin Is does have an effect. Double mutants show a defect in the pinocytosis of fluid-phase markers [31,32]. Mild overexpression of myosin I produces a similar pinocytic defect to the double knockout, the likely explanation being an increase in cortical tension [33,34]. The myosin I proteins also seem to play a role in the regulation of secretion. Lysosomal enzymes are secreted at increased rates in two of the single mutants and in a double myosin I mutant . These defects are likely to arise from an inability to efficiently manipulate their actin-rich cortex in the absence of myosin I.
The phenotypes of the yeast and Aspergillus myosin I mutants correlate well with those originally described for the Dictyostelium myosin I mutants. S. cerevisiae expresses two myosin Is, encoded by the MYO3 and MYO5 genes . These myosins are localised to actin-rich cortical structures within yeast . As with Dictyostelium, the single mutants are relatively normal (although a defect in internalisation of α-factor has been reported for myo5;). A double deletion mutant has a severely defective actin cytoskeleton and defects in receptor-mediated and fluid-phase endocytosis [36,37]. Myosin I also plays a role in membrane trafficking in Aspergillus. Expression of a constitutively activated form of the Aspergillus myosin I, MYOA, causes an activation of endocytosis that, in turn, leads to an accumulation of intracellular membranes . In contrast, the AspergillusmyoA deletion mutant shows no defect in endocytosis, but rather has a defect in polarised growth and secretion . It seems likely that myosin I plays multiple roles in membrane trafficking processes in all of these organisms. It remains to be determined if the fungal class I myosins participate in organising the actin cytoskeleton for the polarised movement of vesicles or if they are directly associated with secretory and endocytic vesicles and power their movement along actin.
A sub-class of myosin I proteins are expressed solely in metazoans. The founding member is the brush border myosin I found in the microvilli of intestinal epithelia. The members of this class lack two of the domains found in the ‘classical’ myosin Is described above — the second actin-binding site and the SH3 domain. The membrane binding domain is present, however, in all members . Whilst it was thought that these myosins play a structural role by crosslinking actin filaments to the plasma membrane in the brush border, there is growing evidence they may also be regulators of vesicle trafficking from the Golgi to the apical domains of polarised cells, as well as in late endosomal-lysosomal trafficking. Firstly, myosin I was localised to the ‘outer’ or cytoplasmic surface of Golgi-derived apical transport vesicles isolated from cells of intestinal crypts in the process of assembling the brush border . Further investigation revealed that the microtubule motor dynein is also present on these vesicles, suggesting a model in which Golgi membranes are transported on microtubules to the actin-rich cortex of the cells where myosin I takes over and delivers materials to the base of the microvillus .
Myosin Iα, a brush border-type myosin I has been localised by immunofluorescence and immunoEM to the plasma membrane, but also to endosomes and lysosomes . Expression of the full-length myosin 1α tail in mammalian cells reveals that the tail can direct the localisation of this myosin to the actin-rich cortex adjacent to the plasma membrane and to intracellular vesicles [42,43]. Overexpression of the tail inhibits delivery of fluid phase markers from endosomes to lysosomes in a hepatoma cell line and appeared to compete with the endogenous myosin Iα for membrane binding sites . However, no obvious abnormalities were reported when a similar experiment was performed in an NRK cell line . Overexpression of full-length myosin Iα also affected the distribution of endocytic compartments in a hepatoma cell line . While it is unclear exactly how inhibition is occurring in this situation or what is being inhibited, the accumulated data are consistent with a role for myosin I in endosomal-lysosomal trafficking pathway ( Fig. 1).
Are Myosins VI and VII Vesicle Transporters?
The mouse myosin VI gene is mutated in the classic mouse mutant, Snell's waltzer, which is deaf due to structural defects of the inner ear . Similarly, mutations in myosin VIIa cause deafness in shaker-1 mice  and combined deafness and blindness in humans (Ushers’ Syndrome; ). In neither case is there an obvious link between the pathology and a function for myosins, but recent data for both classes of myosin suggests that intracellular transport may be affected in each mutant.
Drosophila myosin VI (called 95F) has been shown to be responsible for the transport of cytoplasmic vesicles in the syncytial blastoderm, based on inhibitory antibody injection studies ( Fig. 2). These vesicles surround the nuclei of the syncytium during interphase but are transported outwards during nuclear division to the regions of the pseudo-cleavage furrow where the membrane transiently invaginates between the spindles of neighbouring nuclei. There, they are thought to contribute to this pseudo-cytokinesis, perhaps by providing membrane to the invaginating furrow . 95F myosin also plays a role in the intracellular transport of particles through the ring canals of Drosophila oocytes, consistent with a general role for myosin VI in organelle trafficking .
Interestingly, loss of testis-specific expression of 95F causes a spermatogenesis defect and consequential male sterility (K. Miller, personal communication). This phenotype is also seen in a mutant of hum-3, a myosin VI in C. elegans (J. Kelleher, M. Titus, unpublished). In both cases, spermatids do not form correctly and the males are sterile. Again, in both cases the reasons behind the defect appear to be the same in spite of the significant differences in the process of spermatogenesis in the two organisms. Delivery of membrane-bound components appears to be aberrant, consistent with myosin VI acting as an organelle trafficking motor.
Myosin VI has been localised to the Golgi and also to membrane ruffles in mammalian cell lines . The suggestion is that myosin VI may be directing the polarised transport of vesicles to the leading edge. In bullfrog inner ear cells, myosin VI is present in the cuticular plate, where it may have a structural function and also in the pericuticular necklace, a region containing a large number of vesicles that are thought to be transport vesicles moving from the Golgi to the apical surface . In both cases, the suggestion is that myosin VI may be directing the transport of Golgi vesicles to polarised areas of the cell.
EM studies of the retinas of the shaker-1 mice show altered distribution of melanosomes (a recurring theme?) . The melanosomes do not extend into the outer periphery of the retinal pigment epithelial (RPE) cells, suggesting that myosin VIIa may contribute to transporting them to this region of the cell. Alternatively, myosin VIIa may play a role in organising the actin at the apical end of these cells and loss of this organisation might impair proper melanosome distribution. The photoreceptors of shaker-1 mice also show a slight alteration in opsin distribution. Opsin molecules are synthesised at a huge rate in the inner section of the inner segment of the photoreceptor and transported to the outer segment ( Fig. 2), but in the shaker-1 mice the transport is less efficient, with some opsin molecules remaining in the connecting cilia . Myosin VIIa has been localised to the connecting cilium of the photoreceptor where it may play a function in transporting the opsin, either directly or indirectly. Having noted these two retinal phenotypes, it is unclear why the shaker-1 mice do not suffer the retinitis pigmentosa that humans with Ushers’ Syndrome do. It has been hypothesised  that the lack of obvious retinal dysfunction in the shaker-1 mice may be due to the significantly shorter lifespan of the mouse. The mutant retinas may need to be significantly older or stressed in some way in order for retinitis pigmentosa to become apparent or there may be other unconventional myosins expressed in the mouse retina that can compensate for the loss of myosin VIIa.
Another role for myosin VIIa in endocytosis has been proposed based on the analysis of inner and outer cochlear hair cells in the shaker-1 mutant  and of a Dictyostelium myosin VIIa null mutant . The inner and outer cochlear hair cells of mice internalise aminoglycosides (such as gentamicin) by an as yet undefined endocytic pathway. Analysis of shaker-1 alleles reveals that aminoglycoside uptake in the hair cells is significantly inhibited . Interestingly, there is no obvious perturbation of receptor-mediated endocytosis in the hair cells, based on electron microscopic observation of clathrin coated pits on the apical surface of these cells. These results suggest that loss of myosin VIIa may result in the inhibition of an actin-based endocytic pathway or may act in a later step in endocytosis, perhaps by playing a role in the efficient trafficking of vesicles along the endosomal-lysosomal pathway. A Dictyostelium mutant that lacks myosin VIIa also has an endocytic defect. These mutants have significantly reduced phagocytic activity . These results are consistent with the speculation that the observed retinitis pigmentosa in Usher's patients may be due to the loss of phagocytic activity of the RPE  ( Fig. 2). While the basis of the defect remains to be determined, it is likely to be due either to a failure in the engulfment step or in the transport of the nascent phagosome along the degradation pathway, as has been suggested for mouse myosin VIIa.
And Conventional Myosin?
Whilst most is known about the functions of non-muscle myosin II in processes such as cytokinesis, recent reports suggest that it may also play a role in Golgi transport. A 200 kD protein found on transport vesicles of the TGN was identified as myosin II and found to be recruited to TGN vesicles in a GTP-dependent, brefeldin A-sensitive manner [56,57]. It has been reported that vesicles containing haemagglutinin protein (an apical marker) were able to bud normally from the TGN following p200 immunodepletion but those containing vesicular stomatitis virus G protein (a basolateral marker) were somewhat inhibited . Purified myosin II added back to the system overcame the inhibition. Myosin II appears to be recruited to TGN vesicles via its motor domain , indicating that it is not bound directly to the Golgi membrane and could simply be associated based on the presence of an actin-cytoskeleton on the Golgi. Hence, the specificity for only those vesicles containing apical markers may be based on the composition or organisation of the associated actin cytoskeleton. However, a second report suggested myosin II was not required in a similar experimental set-up . The brefeldin A-dependent redistribution of myosin II and available in vitro evidence are consistent with myosin II being associated with a subclass of Golgi membranes but whether or not it plays an important role in the budding of basolateral vesicles remains unclear. If myosin II is not directly associated with vesicles via its tail then it may accelerate vesicle budding by pulling on opposing actin filaments associated with Golgi membranes, resulting in the severing of a vesicle. Alternatively, it may participate in reorganising the local actin cytoskeleton in a manner that facilitates vesicle transport away from the Golgi. These intriguing yet conflicting observations suggest that further investigations are necessary to clarify the role of conventional myosin in Golgi transport.
It would seem likely that a key determinant in the targeting of vesicles to their destinations will be the ability of the vesicle to interact with the correct cytoskeletal components, a process regulated by its associated motor proteins. It also seems likely that the different families of motor proteins co-operate closely in the transport of vesicles and organelles within the cell, enabling the cargo to be switched efficiently between microtubule and actin networks. Evidence to support such a model is growing. Firstly, organelle switching between microtubule and actin networks has been directly observed for ER-derived vesicles in squid axoplasm, mitochondria in neurons and melanosomes in mouse melanocytes [2,20,59]. Moreover, the importance of this co-operation for successful targeting of cargo within the cell is seen by the mis-localisation of the melanosomes of dilute mice caused by their apparent failure to switch from microtubules onto actin filaments [17,20]. It is also observed in dilute neurites where synaptic vesicles accumulated in regions rich in tyrosinated tubulin (i.e. near microtubule ends) .
The switching of cargo between microtubules and actin appears to be controlled by motors acting in concert. Myosin V and a kinesin have been found to co-localise by immunoEM on ER vesicles in squid axons , myosin I and a dynein family member are found together on Golgi-derived transport vesicles  and myosin V, dynein and kinesin II have been found associated with Xenopus melanosomes [21,62]. Furthermore, two interactions between myosins and microtubule binding proteins have been shown biochemically: mouse myosin V directly interacts with the ubiquitously expressed kinesin, KhcU  and the Drosophila 95F protein (myosin VI) interacts with a microtubule-binding protein of the cytoplasmic linker protein (CLIP) family . A scenario in which cargo (the melanosome, for example) is transported along microtubules by a motor protein complex containing the appropriate microtubule motor, a CLIP and a myosin is possible. Upon reaching its destination (the plus end of the microtubule at the periphery, in the case of the melanosome), the vesicle may be tethered to this microtubule end via the CLIP  until responsibility for vesicle transport can be handed over to the associated myosin ( Fig. 1). Once the myosin binds to a local actin filament, the cargo is transferred to the actin network. How the complex of motor proteins is regulated to allow such an event is an intriguing question. The molecular basis for this switching could either occur through mechanical signals or differential activation of motors. The ability to isolate vesicles carrying the native complement of motors should enable investigators to directly test these models.
The roles of the motors that use the actin filaments to carry cargo and produce forces have come more into focus as the contribution of the actin cytoskeleton to membrane targeting have received wider recognition. When considering just those functions uncovered by loss-of-function genetics (engineered or natural), myosins of several classes have been shown to be involved in pinocytosis and exocytosis in yeast and Dictyostelium, melanosome transport or localisation in humans and mice and directed vesicle transport in mammalian cells and the syncitial Drosophila embryo. The myosins do not appear to act alone in intracellular trafficking. Vesicles exploit both actin and microtubule filaments to achieve intracellular transport and targeting. The challenge for the next phase of research is firstly to define the minimal or optimal motor complement for a given class of transport vesicles — do all vesicles have only one type of myosin, one type of kinesin and one type of cytoplasmic dynein or do they have multiple types of each motor type associated with them? If they do have a specific subset of motors, what dictates which motor is associated with a given vesicle or membrane? Finally, once all of the players are identified it is imperative to know how their functions are integrated. If they directly bind to each other, as is likely for myosin V and KhcU, then do they directly respond to mechanical signals when tension is placed on one motor by the action of the other pulling along its filament system? How are these motors regulated by intracellular signals? Are there zones of activation depending on whether or not the vesicle must traffic along one filament track or another? Taking on the challenge of answering these questions is certain to stimulate a great deal of future work and will contribute to understanding the molecular basis of intracellular transport.
The authors would like to thank Drs Joe Kelleher, Mary Porter, Jeff Baker and Tom Hays for many helpful comments on the manuscript. RIT is supported by a postdoctoral fellowship from the Human Frontiers Science Program and MAT is supported by grants from the National Institutes of Health, National Science Foundation and an Established Investigator Award from the American Heart Association.