Walking to work: roles for class V myosins as cargo transporters

Vertebrate myosin Va is structurally and kinetically designed to be able to move processively along actin filaments as a single molecule and is thus well suited to be a cargo transporter. The key features are the presence of two motor domains and kinetics with a high duty ratio. Not all class V myosins are structurally or kinetically similar to vertebrate myosin Va. Some do not have high duty ratios and are not two headed. These class V myosins must adopt other strategies to move cargo, such as having multiple motors bound to the cargo. Budding yeast and plants build actin cytoskeletons that are suitable for long range class V myosin-dependent cargo transport. Indeed, the yeast class V myosins Myo2 and Myo4 drive most, if not all, organelle transport in this organism. The organization of actin within the cortex of vertebrate cells is largely anisotropic, so the class V myosin-dependent transport of organelles in these cells, which in most cases will follow the long-range transport of the organelle on microtubules to the cell periphery, is likely to exhibit minimal directional persistence and be very local. Two recent and compelling examples of class V myosin-dependent organelle transport have been identified within the dendritic spines of neurons, one involving recycling endosomes, and the other involving the endoplasmic reticulum. The 'acid test' as to whether a class V myosin actually moves an organelle in vivo is to show that the organelle moves more slowly when the cell expresses a 'slower' version of the class V myosin. Vertebrate myosin Va is structurally and kinetically designed to be able to move processively along actin filaments as a single molecule and is thus well suited to be a cargo transporter. The key features are the presence of two motor domains and kinetics with a high duty ratio. Not all class V myosins are structurally or kinetically similar to vertebrate myosin Va. Some do not have high duty ratios and are not two headed. These class V myosins must adopt other strategies to move cargo, such as having multiple motors bound to the cargo. Budding yeast and plants build actin cytoskeletons that are suitable for long range class V myosin-dependent cargo transport. Indeed, the yeast class V myosins Myo2 and Myo4 drive most, if not all, organelle transport in this organism. The organization of actin within the cortex of vertebrate cells is largely anisotropic, so the class V myosin-dependent transport of organelles in these cells, which in most cases will follow the long-range transport of the organelle on microtubules to the cell periphery, is likely to exhibit minimal directional persistence and be very local. Two recent and compelling examples of class V myosin-dependent organelle transport have been identified within the dendritic spines of neurons, one involving recycling endosomes, and the other involving the endoplasmic reticulum. The 'acid test' as to whether a class V myosin actually moves an organelle in vivo is to show that the organelle moves more slowly when the cell expresses a 'slower' version of the class V myosin. Cells use molecular motors to position and segregate organelles. Recent studies show that class V myosins function as actin-based cargo transporters in yeast, moving the vacuole, peroxisomes and secretory vesicles. There is also increasing evidence in vertebrate cells that class V myosins can serve as short-range, point-to-point organelle transporters rather than just tethering organelles to actin. Cells use molecular motors, such as myosins, to move, position and segregate their organelles. Class V myosins possess biochemical and structural properties that should make them ideal actin-based cargo transporters. Indeed, studies show that class V myosins function as cargo transporters in yeast, moving a range of organelles, such as the vacuole, peroxisomes and secretory vesicles. There is also increasing evidence in vertebrate cells that class V myosins not only tether organelles to actin but also can serve as short-range, point-to-point organelle transporters, usually following long-range, microtubule-dependent organelle transport.

All cells, in their efforts to move and position their organ elles, use molecular motors. Three classes of molecular motors -dyneins, kinesins and myosinshave evolved to transport organelles and other cargoe s inside cells. The myosin superfamily members, which are separated into at least 35 classes 1 and which move on actin filaments, participate in a plethora of functions in cells, including cytokinesis, cell adhesion, endo cytosis, exocytosis, movement of mRNA, movement of pigment granules, and cell motility. This diversit y of function requires the various myosins to have differ en t kinetic properties and structural adaptations. For example, some myosins are adapted to transport cargo, whereas others may function as loadbearin g tension sensors.
In general, myosins are composed of three domains 2 (FIG. 1). The first is the motor domain or head, which is usually located at the amino terminus and which binds ATP and actin. The motor domain is mechanically linked to a second domain consisting of an extended αhelical 'neck' containing a variable number of IQ motifs, which bind light chains of calmodulin or mem bers of the calmodulin family. The neck domain is sometimes referred to as the lever arm because it moves as a rigid body to generate the power stroke in response to small ATPdependent changes in the conformation of the motor domain [3][4][5][6] . The tail domain contains the most classspecific variations and can include coiled coil motifs for dimerization, as well as domains that accommodate the different functions of distinct myosi n classes, including binding to membranes or to cargo receptors mediating specific interactions with vesicular or nonvesicular cargo 7 .
Class V myosins have been implicated in both organelle transport and dynamic organelle tetherin g in severa l cellular systems. These myosins are found in almost all eukaryotic genomes sequenced to date 1 . Humans have 38 myosin genes, including three class V myosin genes (encoding myosin Va, myosin Vb and myosin Vc) 8 , whereas the yeast Saccharomyces cerevisia e has only five myosin genes, two of which encode class V myosins (MYO2 and MYO4) 8 . The class XI myosins found in plants are almost certainly derived from an ancient class V myosinlike molecule 1 , implyin g that organisms which predate the division of plants and animals possessed this type of myosin.
In this Review, we describe the design features of class V myosins that should allow them to serve as cargo transporters in vivo. We then examine the evidence that these myosins actually transport organelles inside cells and define what we believe is the 'acid test' to prove unequivocally that any particular class V myosin might indeed drive organelle translocation in vivo.

Myosin Va properties and adaptations
The transportation of cargo within cells is a demanding process. Actin filament tracks exist as single filaments, branched networks or bundles of different geometries. In many cases, actin filaments are in close proximity with the two other cytoskeletal filaments, microtubules and inter mediate filaments, forming meshes with small pore sizes. In the turbulent world of the cytoplasm, where the motor is constantly bombarded by water molecules and macro molecules, instantaneous detachment of the myosin from actin would be associated with rapid Brownian diffusion away from the track and an end to directed movement.
Vertebrate myosin Va was the first myosin shown in vitro to be capable of moving along actin fila ments as a single molecule without detaching. This behaviou r, which is known as processive movement, is likely to be a requirement in many cases for efficient cargo transport in vivo 9,10 . As it has been extremely well characterized in vitro, we describe here the array of evolutionary adaptations in both its structure and kinetics that should make myosin Va an efficient cargo transporte r in vivo. The 'neck' domain consists of an α-helical segment of the heavy chain containing six tandem IQ motifs, which each bind a calmodulin molecule. The tail domain consists of a coiled-coilforming domain with periodic interruptions, which dimerizes the heavy chains, and two globular tail domains (GTDs), which bind cargo. b | Kinetic cycle for myosin Va. The cycle alternates between states with a high affinity for actin (myosin Va with no nucleotide bound and myosin Va with ADP bound) and states with a low affinity for actin (myosin Va bound to ATP or to ADP and inorganic phosphate (P i )). ADP release from actomyosin-ADP is rate limiting 31 . c | Pathway of the processive movement of myosin Va along actin. Myosin Va probably dwells in a state with ADP bound to both heads (step 1) 36 and the two heads exert intramolecular strain 32-36 on each other, so that ADP is first released from the trailing head (step 2). This head subsequently binds ATP and rapidly dissociates from actin (step 3). The attached head undergoes a power stroke, positioning the new leading head (bound to ADP and P i ) to find a forward binding site via a thermally driven search 15 (step 4). Upon binding to actin (step 5), the leading head rapidly releases P i and establishes a strong binding conformation 31 , which brings myosin Va to the same state as in step 1 but translated forward by 36 nm. d | Regulation of myosin Va. In the absence of Ca 2+ and cargo, myosin Va adopts a bent conformation that is enzymatically and mechanically inactive (middle), in which the motor domains interact with the cargo binding domains [37][38][39]138,139 . In the presence of Ca 2+ (left), myosin Va unfolds into a conformation that has high ATPase activity but cannot move on actin because the neck is mechanically weakened by dissociation of calmodulins 41 . On cargo binding (right), myosin Va becomes extended and is both mechanically and enzymatically active. This active state has not been observed by electron microscopy but has been inferred from enzymatic studies and in vitro motility studies 44,45 .

Optical trapping
A technique that uses focused laser light in a light microscope to capture and manipulate objects with dielectric constants different from water. Sophisticated traps can be used to measure the mechanical properties of single molecules of motor proteins, such as the class V myosins.

Super-resolution light microscopy
Techniques that use Gaussian fits to the point spread function of light emitted from a single fluorophore to determine its position to within a few nanometres, which is significantly smaller than the diffraction limit.
Total internal reflection fluorescence (TIRF). TIRF microscopy provides improved signal to noise ratios at the interface between media with differing refractive indices (such as coverslip-water), where fluorescence is excited by an evanescent field of light created when the incident light is totally internally reflected. It is ideal for visualizing the movement of single motor proteins.
High-speed atomic force microscope (High-speed AFM). A modified AFM that can acquire images at a high rate from samples in aqueous solution, allowing the dynamic imaging of single molecules as they perform their tasks.

Myosin II
Also called conventional muscle myosin. The first myosin type to be discovered and the most conspicuous of the myosin superfamily. It is responsible for skeletal muscle contraction in muscle cells.

Duty ratio
The fractional time that a myosin spends in a state of high affinity for actin during an ATP hydrolysis cycle.

Structural adaptations allow processive movement.
Myosin Va has coiledcoil forming sequences in its tail domain, which dimerize to form a twoheaded motor 11 , allowing the molecule to 'walk' along actin filaments by alternating the positions of the leading and trailing heads 12,13 (FIG. 1). The two heads must be able to simul taneously bind to two actin monomers within a filament, each of which has a helical pitch with crossover points every 36 nm 14 . If the two heads of the motor can span this distance, then the myosin can effectively transport its cargo linearly along an actin filament, rather than fol lowing the actin's helical pitch. Following the pitch would result in the myosin spiralling around the filament, potentially causing steric clashes between the cargo and the cell membrane (FIG. 1c). Myosin Va accomplishes this feat by having an elongated neck region containing six IQ motifs, each of which binds a calmodulin or a cal modulin family member (FIG. 1a). The 36 nm step size has been measured in vitro in singlemolecule experiments involving optical trapping 9,15,16 and in super-resolution light microscopic observations of moving myosin Va molecules using total internal reflection fluorescence (TIRF) micro scopy 17,18 . In addition, electron microscopic observation of myosin Va molecules associated with actin in the presenc e of ATP show that the two heads are bound to actin monomers spaced 36 nm apart 14 . Moreover, these features were recently confirmed using a novel, highspeed atomic force microscope (AFM) to image myosin Va movin g along an actin filament 19 .
The tail domain of myosin Va is significantly more elastic than that of the conventional skeletal muscle myosi n (myosin II), probably owing to the multiple inter ruptions of its coiledcoil domain by loops of various sizes 20 . This property is interesting, as modelling predicts that an elastic tail would allow the motor to continue moving forward taking regularly spaced steps, albeit more slowly, while the cargo is being pulled through the viscous cytoplasm 21 .
Myosins must have a means of binding their cargo; in the case of myosin Va, this is mediated by a pair of globular tail domains (GTDs) located at the carboxyl term inus, following the coiledcoil region. However, other parts of the molecule, such as flexible loops within the coiled coil domain, also have roles in cargo binding 22 . Often, cargo binding is mediated by the interaction of the GTD with an adaptor protein, which in turn is bound to an organellespecific RABfamily GTPase, thereby linking the myosin to the membranous cargo [22][23][24][25][26] . Furthermore, in cases when a given class V myosin moves multiple cargoes, there has to be means to select the correct cargo. In the case of myosin Va, different alternatively spliced isoforms found in melanocytes and brain cells bind different cargo 22,27,28 .
Finally, myosin Va not only must be capable of navi gating the complex array of actin filaments within cells but also must cooperate with microtubule motors, as most cargoes, at least in animal cells, are transported on both microtubules and actin filaments. Efficient switching between actin filament tracks has been shown in vitro 29,30 . This was found to be mediated by the long neck of myosin Va, which allows it to move from one actin filament to another when the filaments cross or when the actin filament contains a side branch gener ated by the actin nucleator actinrelated protein 2/3 (Arp2/3) 30 . Furthermore, in vitro experiments, in which beads coated with both dynein and myosin V were applied to surfaces containing meshes of actin filaments and microtubules, indicated that track switching can occur efficiently 29 .
Kinetic adaptations allow processive movement. The kinetic cycle of ATP hydrolysis by myosin Va, like that of most myosins, alternates between states with a high affinity for actin (myosin Va alone and myosin Va-ADP) and states with a low affinity (myosin Va-ATP and myosin Va-ADP-inorganic phosphate (P i )) (FIG. 1b). In this cycle, the ratelimiting step is the dissociation of ADP from an actomyosin-ADP complex, whereas ATP hydrolysis and P i release are fast 31 . This means that myosin Va spends most (~70%) of its kinetic cycle bound strongly to actin (that is, it has a high duty ratio, which is required for processive movement). Furthermore, the kinetic cycles of the two heads of myosin Va are strain dependent, such that ADP release from the attached leading head is strongly suppressed as long as the trail ing head is still attached to actin [32][33][34][35][36] (FIG. 1c). The leading head and its neck also cannot easily enter the post powerstroke state as long as the trailing head remains attached 14 . These events ensure that the trailing head loses its ADP, binds a new ATP and detaches from actin, allowing the leading head to undergo its power stroke. This propels the former trailing head forwards to search for a new binding position on actin, whereas the former leading head remains attached.

Regulation of myosin Va.
Given that myosin Va is capabl e of fast processive movement as a single molecule in vitro, it is imperative that its activity is tightly regu lated in cells when it is not engaged in binding to cargo. Under ionic conditions that are similar to those found in most cells (that is, with a submicromolar concentratio n of free Ca 2+ ), myosin Va adopts a specific triangula r, folded conformation, in which the GTDs bind to the motor (head) domains 37,38 (FIG. 1d). This folded confor mation sediments at 14S in the analytical ultracentrifuge (and is thus frequently referred to as 14S), as opposed to 11S for the extended molecule 39 . The ATPase activity of the folded molecule is suppressed and the molecule does not bind strongly to actin 40,41 .
Ca 2+ ions shift the equilibrium to the extended, 11S conformation and activate the actinactivated ATPase activity of myosin Va in vitro. However, Ca 2+ binding also results in the dissociation of one or more calmod ulins from the neck region, resulting in a very flexible neck that cannot undergo mechanical work. Molecules under these conditions do not move on actin filaments unless very high concentrations (~10 μM) of calmoduli n are present 39,42,43 . Thus, Ca 2+ may not be the physio logical activator of myosin Va. Instead, it has been shown that binding to melanophilin, a cargo receptor, activates myosi n Va's ATPase activity, presumably by trapping the molecule in its extended, active conformation, and

Melanosome
The pigment-producing organelle found in pigment-producing cells, such as melanocytes.
restores the myosin's ability to move along actin filaments in vitro in the absence of Ca 2+ (REFS 44,45). Melanophilin, in turn, binds in a GTPdependent manne r to RAB27A, which is attached to the limiting membrane of the melano some. This form of cargodependent control of myosin Va's mechanochemistry serves to avoid energy wastage, prevent cargofree myosin Va molecules from piling up at actin filament plus ends and promote the recycling of the myosin by diffusion.

Properties of myosin V paralogues
The structural and kinetic properties of mammalian myosin Vb are similar to those of myosin Va, and, based on the high duty ratio of myosin Vb, it is pre dicted to be a processive motor at the singlemolecule level 46 . By contrast, although mammalian myosin Vc and Drosophila melanogaster myosin V are similar in struc ture to myosi n Va, kinetic studies demonstrate that ADP dissociation is not rate limiting for these myosins and that they have a lower duty ratio, which would result in nonprocessive interactions with actin at the single molecule level [47][48][49] . However, as is seen with myosin Va, the kinetics of these motors may be affected by intra molecular strain 15,[32][33][34][35] , so there may be conditions in which they can move processively as single molecules. If not, it is possible that clusters of these myosins might transport cargo efficiently. Moreover, some class V myosins may be processive at the singlemolecule level only under specific ionic conditions 50 .
The situation is more complicated in S. cerevisiae. Although the structure of Myo2 is similar to that of mammalian class V myosin, it is only weakly processive as a single molecule 51 , although this can be enhanced using artificial cargo (such as beads) that contain both Myo2 and Smy1, a kinesin family protein that inter acts weakly and electrostatically with actin 52 . By con trast, Myo4 is different in structure from mammalian class V myosin 51 : the tail of one Myo4 molecule forms a heterodimeric coiledcoil with another yeast protein, Swi5dependent HO expression 3 (She3), to form a singl eheaded motor 53 . The monomeric Myo4 motor does not move processively along actin in vitro as a single molecule, but if it is artificially dimerized or aggregated, or if more than one molecule is bound to a polymer bead, processive movements are possible 51,53 .
Given that class V myosins in most organisms seem wellsuited to supporting organelle transport, how does one go about proving that their favourite class V myosin does indeed serve as a pointtopoint organelle trans porter in vivo? BOX 1 and Supplementary information S1 (table) lists the types of evidence that can be used to sup port a claim of class V myosindependent organelle trans location, as well as the evidence currently available for various class V myosin-organelle pairs. The types of evi dence include demonstrating co localization between the myosin and the organelle using class V myosinspecific antibodies and green fluorescent protei n (GFP)tagged class V myosin chimaeras, as well as demonstrating defects in the movement and/ or localization of the organelle when the function of the myosin is abrogated by gene deletion, RNA interference (RNAi)mediated knockdown, or overexpression of a dominantnegative class V myosin construct. The acid test of class V myosin dependent organelle transport is the rescue of mutant cells with mechanochemically compromised versions of the class V myosin in question (such as stepsize mutants or enzymatically impaired mutants) and the demonstra tion of corresponding reductions in the velocity of its cognate organelle in vivo. In our opinion, this crucial test is the stateoftheart approach to verifying motor protein function; however, it has so far been done in only a few instances. In the following sections, we describe these and other examples of apparent class V myosindependent organelle transport in more detail.

Cargo transport in yeast
The manner in which cells accomplish motordependen t organelle movement and positioning is constrained by the organization of the two types of tracks that they can use: microtubules and filamentous actin (Factin). The principal elements of this constraint involve the three dimensional arrangement of these polymers inside the cell, and the fact that individual motors can move (at least, robustly) only in one direction on the polymer (for class V myosins, towards the barbed end of the actin filament). The budding yeast S. cerevisiae has solved this overarching design issue by building actin struc tures that are highly polarized (all barbed ends point in the same direction) and spatially arranged to drive organelle transport 54 . Specifically, this yeast supports most, if not all, organelle transport using actin cables that are generated by the nucleator formin, that run from the mother cell to the bud tip and that are oriented uni formly with their barbed ends pointing towards the bud tip. In this way, the yeast's two class V myosins (Myo2 and Myo4) can support, in an efficient way and in the complete absence of microtubuledependent organelle transport, the host of membrane movements that drive bud growth and organelle inheritance 55,56 (BOX 1; FIGS 2,3; see Supplementary information S1 (table)). Below, we highlight some examples of cargo transport in yeast that support this general conclusion.
Myo2 and secretory vesicle transport. The transport of Golgiderived, Sec4positive secretory vesicles drives the vectorial growth of the bud. In what was the first un equivocal demonstration that a class V myosin actually moves an organelle in vivo, Schott and colleagues 57 showed that complementing the yeast Myo2null mutant with versions of Myo2 possessing progressively shorter lever arms (which should result in the myosin taking progressively smaller steps) resulted in a progressive reduction in the velocity of secretory vesicle transport into the bud (from ~2.5 μm s -1 to ~0.2 μm s -1 ). Somewhat surprisingly, the receptor for Myo2 on these vesicles has until recently evaded identification, despite being the only organelle cargo for this myosin that is required for viability. However, several recent papers [58][59][60][61][62] have provided important new insights into the organization and regulation of this receptor (FIG. 2). Overall, Myo2 is recruited to these vesicles through its sequential interac tion with the functional RAB GTPase pair Ypt31-Ypt32

Box 1 | Evidence for class V myosin-dependent organelle transport in vivo
To demonstrate that a class V myosin transports an organelle in vivo, one should first show colocalization of the endogenous class V myosin with its cognate organelle using specific antibodies, followed by colocalization using a green fluorescent protein (GFP)-tagged version of the myosin and markers for its cognate organelle in another colour to confirm their interaction and follow their dynamics. Second, one should identify abnormalities in the movement and/or distribution of the organelle in cells lacking the function of this class V myosin, which can be generated by gene knockout, RNA interference (RNAi) or the overexpression of a dominant-negative class V myosin construct. Such functional studies require extensive quantification and, in some cases, the application of stress to the mutant cell to demonstrate more clearly the requirement for the myosin. Third, one should complement the defect in mutant cells by reintroducing the myosin gene (or expressing an RNAi-immune version) to prove that the defect in organelle motility and/or distribution is due to the abrogation of the function of this myosin. Complementation also opens the door to defining what domains and/or properties of the myosin are required for organelle targeting and transport. Finally, the 'acid test' of class V myosin-dependent organelle motility is the rescue of mutant cells with mechanochemically compromised versions of the myosin in question (for example, step-size mutants or catalytic mutants) and the demonstration of corresponding reductions in the velocity of its cognate organelle in vivo. Other lines of evidence include the identification of the organelle receptor for the myosin, the identification of mechanisms that regulate the myosin-receptor interaction, and the reconstitution of the myosin-dependent organelle motility in vitro.
The table in this box is available in full Online (see Supplementary information S1 (table)). We discuss only those myosin-organelle and myosin-cargo pairs for which we think there is sufficient evidence to conclude that the myosin actually moves the organelle or non-vesicular cargo in vivo. Note that the wide variations in the speeds reported for yeast class V myosin Myo2-dependent organelle transport could be due, in part, to variations in the loads presented by the different organelles, together with variations in the numbers of Myo2 molecules recruited. Also note that the results of the acid test could be ambiguous for those class V myosin-dependent organelle movements that are extremely local or 'diffusive' in nature.

Secretory vesicles-yeast Myo2
Vectorial transport of the secretory vesicles that drives the growth of the bud

RAB switch
The sequential interaction of an effector protein with two different RAB GTPases.

Exocyst
An eight-member protein complex that is required for vesicle docking and polarized exocytosis and is conserved from yeast to mammals.

Myristoylation
A protein post-translational modification in which the 14-carbon saturated fatty acid myristic acid is covalently attached to an amino-terminal Gly residue.

Palmitoylation
A protein post-translational modification in which the 16-carbon saturated fatty acid palmitic acid is covalently attached to a Cys residue. and Sec4. Myo2 also interacts with the Golgienriched lipid phosphatidylinositol4phosphate (PtdIns4P), and the decrease in the content of this lipid as the vesicle matures regulates both the RAB switch and the recruit ment of Sec15, a lateacting Sec4 effector protein that is important for the final steps in vesicle docking and fusion at the exocyst.
In addition to moving postGolgi secretory vesicles, Myo2 supports the inheritance of the Golgi apparatus by transporting lateGolgi cisternae into the bud at a rate of ~0.22 μm s -1 (REF. 63). Recently, the RAB GTPase Ypt11, which was known from previous work to interact with Myo2's GTD, was also shown to bind Ret2, a subunit of the Golgiassociated coatomer complex that is involved in generating cytosolic coatomer complex I (COPI)coated vesicles for retrograde transport from the Golgi to the endoplasmic reticulum (ER) 64 . These interactions should allow Ypt11 to link Myo2 to lateGolgi elements for their Myo2dependent, vectorial transport into the bud.
Myo2 and vacuole inheritance. The highly polarized mechanism of cell division in S. cerevisiae, with its large mother cell and tiny emerging bud (as opposed to the two evenly sized daughter cells created by median fission in most animal cells), places a high demand on the machinery driving organelle partitioning and inheritance. Myo2 and Myo4 have important roles in the delivery of most, if not all, organelles into the bud. Importantly, although these myosins support organelle inheritance, as evidence by slower growth when the Myo2 or Myo4dependent transport of a partic ular organelle or nonvesicular cargo is specifi cally abrogated, these roles are not absolutely required for viability.
Vacuole inheritance in S. cerevisiae is supported by the Myo2dependent transport of portions of the vacu ole into the emerging bud 65 . During transport, Myo2 accumulates at the leading edge of the segregation struc ture, a fingerlike projection of vacuolar membranes destined for the daughter cell that Myo2 pulls into the bud along the actin cable at a speed of ~0.15 μm s -1 . The vacuolespecific receptor for Myo2 is composed of the vacuolespecific adaptor protei n vacuol e related 17 (Vac17) and the myristoylated and palmitoylate d vacuole membrane protein Vac8 (REFS 66,67) (FIG. 3). Vac17 bridges the indirect inter action between Myo2 and Vac8 by binding simultaneously and directly to the GTD of Myo2 and Vac8. As in all eukaryotic cells, secretory vesicles in yeast form by budding at the trans face of the Golgi apparatus (the trans-Golgi network (TGN)). Budding and subsequent Myo2-dependent vesicle transport involves the sequential interaction of Myo2 with the functional RAB GTPase pair Ypt31-Ypt32 and Sec4, in a process that is coordinated by the Golgi membrane-enriched lipid phosphatidylinositol-4-phosphate (PtdIns4P) 54,[58][59][60]140,141 . Myo2 is initially recruited to forming secretory vesicles by its combined interactions with Ypt31-Ypt32·GTP and PtdIns4P (the latter via an unidentified bridging protein, X). Sec2, the guanine nucleotide exchange factor (GEF) for Sec4, is also recruited to the TGN through interactions with PtdIns4P and Ypt31-YPT32·GTP. This creates a RAB GEF cascade in which the recruitment of Sec2 serves to activate Sec4, which in turn is required for the terminal transport of vesicles to sites of bud growth. The late-acting Sec4 effector protein Sec15, which is required for the final steps in vesicle docking and fusion at the exocyst, also interacts with Sec2, and its interaction competes with that of Ypt31-Ypt32. Importantly, the binding of PtdIns4P to Sec2 inhibits the Sec2-Sec15 interaction, thereby favouring the recruitment of Sec2 by Ypt31-Ypt32 as secretory vesicles initially bud from the TGN. Later, as secretory vesicles mature and approach the site of secretion, PtdIns4P levels in the vesicle membrane decline. This change in membrane lipid composition allows Sec15 to competitively replace Ypt31-Ypt32 on Sec2. This exchange, together with the presence within Myo2's globular tail domain (GTD) of a binding site for Sec4-GTP that is distinct from its binding site for Ypt31-Ypt32, results over time in the replacement of Ypt31-Ypt32 by Sec4 as the Myo2 receptor. The increased vesicle levels of active Sec4 and Sec15 that result from this PtdIns4P-orchestrated handoff serve to drive the final stage of vectorial vesicle transport and to prepare the vesicles for exocyst-dependent tethering, docking and fusion at the bud tip. Vac17 also serves to coordinate vacuole inheritance with the cell cycle in two ways 68 . First, its synthesis and accumulation early in the cell cycle probably facili tates the initiation of vacuole movement. Second, it is phosphorylated by cyclindependent kinase 1 (Cdk1). This modification strengthens the interaction of Vac17 with Myo2 and promotes vacuole inheritance, thereby servin g to coordinate vacuole movement with the cell cycle. Indeed, this may be a common cell cycle control mechanism in yeast, as the peroxisome protein inherit ance of peroxisomes 2 (Inp2; see below) also contains Cdk1 phosphorylation sites.
Finally, Vac17 seems to control the ultimate position of the vacuole in the bud through its cell cycleregulated degradation 67 . Specifically, when the vacuole reaches the bud, Vac17 is specifically degraded by cleavage at its PEST site, thereby releasing Myo2 from the vacuole. This serves to terminate further vacuole movement, thereby depositing the vacuole in the correct location in the bud, and to release Myo2, allowing it to carry out other functions.
Myo2 and peroxisome inheritance. Myo2 supports the inheritance of peroxisomes by transporting them on actin cables into the bud at ~0.45 μm s -1 (REF. 69). Myo2 attaches to peroxisomes by a direct interaction with Inp2, an inte gral peroxisomal membrane protein 69 . Although Inp2 does not have homologues in other species, the peroxi some protein peroxisomal biogenesis 3 (Pex3), which is widely expressed in eukaryotes, plays a part in peroxisome inheritance in the fungus Yarrowia lipolytica 70 . This obser vation suggests that class V myosins may be involved in peroxisome inheritance in vertebrate cells as well.
In S. cerevisiae, the regulation of peroxisome inher itance seems to be focused largely on Inp2, which is subjec t to both cell cycle and positional cues 71 . First, like those of Vac17, Inp2 protein levels rise and fall during the cell cycle, being highest during the period when most peroxisomes are transported into the bud. This suggests that peroxisome inheritance is regulated, at least in part, by the availability of Inp2 for Myo2. Interestingly, the presence of Inp2 in the peroxisome membrane is only readily apparent on peroxisomes that have been trans ported into the bud. The underlying mechanism of this positiondependent asymmetry in receptor distribu tion is thought to be a negative feedback mechanism, in which peroxisomes that have been delivered to the bud relay a signal back to the mother cell that triggers the degradation of Inp2 in the mother cell 71,72 .
Myo4 and mRNA transport. Yeast Myo4 is responsibl e for the transport of asymmetric synthesis of HO 1 (ASH1) mRNA 73,74 , as well as a host of other mRNAs 75 , to the bud tip. In vivo imaging of fluorescently tagged ASH1 mRNA shows that Myo4 supports its continuous transport from mother to bud at a rate of ~0.3 μm s -1 (REF. 76). The fact that the singleheaded Myo4-She3 molecule does not exhibit processive motility raises the question of how this motor generates processive move ment of mRNA in vivo 51 . The answer to this conundrum is in the numbers 77 . Specifically, through the tetrameri zation of the molecule She2, which bridges the indirect interaction between Myo4 and the mRNA by simultane ously binding to a localization element in the mRNA and to She3 (REFS 78,79), multiple copies of Myo4-She3 are recruited to each localization element (FIG. 3). The infer ence, then, is that by recruiting multiple, nonprocessive Myo4 motors in such close proximity, the major hurdl e to processive movement that is inherent in being a mono meric motor can be overcome to allow for the con tinuous movement of the mRNA that is observed in vivo. The fact that mouse myosin Va facilitates the localization of an mRNAbinding protein and its bound mRNA to the dendritic spines of hippocampal neurons 80 , and that D. melanogaster myosin V contributes to the posterior accumulation of Oskar mRNA in embryos 81 , suggests a more general connection between class V myosins and mRNA transport and/or localization.

Myo4 and ER inheritance.
Myo4 seems to have a major role in ER inheritance into the bud. Indeed, Myo4 null cells exhibit a marked decrease in the delivery of cortical ER tubules (which normally move at a rate Multitasking by the yeast class V myosins is coordinated both temporally and spatially, primarily at the level of these receptors, through several mechanisms, including cell cycle-controlled receptor synthesis and proteolysis, and receptor phosphorylation. Myo2 binds multiple receptors through largely distinct regions on the surface of its globular tail domain (GTD). The determination of the structure of Myo2's GTD 142 (which can also be used to model the structure of the GTD in vertebrate class V myosins) has given valuable insights into the molecular details of class V myosin-cargo interactions. The vacuole-specific receptor for Myo2 comprises the vacuole-specific adaptor protein vacuole-related 17 (Vac17) and the myristoylated and palmitoylated vacuole membrane protein Vac8. During vacuole transport, Myo2 accumulates at the leading tip of the segregation structure, a finger-like projection of vacuolar membranes destined for the daughter cell, which Myo2 pulls into the bud along the actin cables 55 . Myo2 attaches to peroxisomes by a direct interaction with inheritance of peroxisomes 2 (Inp2), an integral peroxisomal membrane protein 56 . The cargo adaptor Swi5-dependent HO expression 2 (She2) not only links the Myo4-She3 complex to mRNAs but also serves to cluster Myo4-She3 complexes together in close proximity by virtue of its ability to self-associate. Although individual Myo4-She3 complexes are not processive, their She2-driven aggregation allows processive mRNA transport by increasing the probability, at any one moment during transport, of at least one complex being attached to an actin filament 52,53,77 . of ~0.013 μm s -1 ) into the bud 82 . However, the defect in ER inheritance in Myo4null cells, although signifi cant, is not complete. Indeed, a secondary, less efficient mechanism of ER inheritance seems to exist, in which ER tubules attach to the prebud site and are passively pulled into the growing bud 83 .
Efficient ER inheritance also requires the Myo4interacting protein She3, which is also involved in the Myo4dependent movement of mRNA. A large fraction of She3 cofractionates with ER membranes, suggesting that it may play a part in the attachment of Myo4 to the ER. Different domains of She3 are involved in ER inheritance and mRNA transport, and these two Myo4dependent processes are independent of each other. Interestingly, there is evidence that messenger ribonucleoprotein (RNP) particles migrate with ER tubules into the emerging bud 84 , and that this requires Myo4 and She3.

Cargo transport in D. discoideum
The contractile vacuole complex in Dictyostelium discoideum is a tubulovesicular osmoregulatory organelle that exhibits extensive motility along the actinrich cortex. What is most striking is the conversion, following water discharge, of collapsed bladder membranes into radiating cortical tubules, which move along the cortex at a rate of ~0.5 μm s -1 . It has now been shown that the D. discoideum class V myosin MyoJ powers these movements 85 . Strong support for this conclusion was provided by the complementation of MyoJnull cells with MyoJ contain ing a shorter neck, which resulted in a significant reduc tion in the speed of the radiating tubules. Moreover, as with class V myosins in vertebrate cells (see below), MyoJ was shown to cooperate with bidirectional, microtubule dependent contractile vacuole membrane transport to properly distribute the contractile vacuole complex.
Cargo transport in D. melanogaster When the fly eye is exposed to bright light, pigment granules located deep in the photoreceptor cell cyto plasm move rapidly to the cytoplasmic face of the photo sensitive membrane organelle, the rhabdomere, thereby creating a functional pupil. Granule migration occurs along the rhabdomere terminal web (RTW), a polarized array of actin filaments that emanate from the base of the rhabdomere, extend deep into the photoreceptor cell cytoplasm and are oriented uniformly, with their barbed ends pointing towards the rhabdomere. The fact that myosin V localizes to these pigment granules, and that granule translocation is abrogated in a strong myosi n V lossoffunction mutant, suggests that myosin V is responsible for moving the granules 86 .
Pigment granule migration is triggered by the entry of Ca 2+ into the cytoplasm downstream from lightactivated Transient receptor potential (TRP) Ca 2+ channels, and the increase in intracellular Ca 2+ seems to directly activate myosin V to drive granule migration. Myosin V appears to function, therefore, as a sensory adaptation motor to drive the lightinduced movement of pigment granules. Interestingly, a RABlike protein known as Lightoid, which also associates with pigment granules and is required for granule movement, physically interacts with myosin V and is required for the association of myosin V with the granules, indicating that it serves as an essential component of the granule receptor for myosi n V (REF. 86).
A second important part played by myosin V in photoreceptor cells involves the transport of postGolgi secretory vesicles containing rhodopsin and other car goes that fuel the rapid expansion of the sensory mem brane 87 . Specifically, the movement of these vesicles out the RTW to the developing rhabdomere membrane requires myosin V, RAB11 and the RAB11 effector pro tein, RAB11interacting protein (RIP11). Myosin V and RIP11 bind independently to RAB11 to create a ternary complex, which drives secretory vesicle movement. Given that D. melanogaster myosin V is not processive 48 , pigment granules and secretory vesicles presumably recruit multiple myosin V molecules to allow processive movement of the organelles.

Evidence of cargo transport in vertebrates
Fly photoreceptor cells, with their RTW, and actincentric organisms, like budding yeast and plants cells, use actin tracks of uniform polarity to support longrange, class V myosindependent organelle transport. By contrast, most cells, including vertebrate cells of all types (from minimally polarizedlike liver hepatocytes to highly polarizedlike neurons), seem to use micro tubules and microtubulebased motors to drive the bulk of long range organelle transport. Importantly, the organiza tion of the microtubule cytoskeleton in these cells, in which most microtubule minus ends are anchored at the centro some in the cell centre and most plus ends are found at the cell periphery, provides uniformly polarized, relatively straight, uninterrupted tracks that are ideal for driving organelle movements over much of the cell's threedimensiona l space. Actinbased motors, such as the class V myosins, come into play at the end of the line to support track switching and the capture and/or local movement of organelles in the actinrich periphery 88,89 . Below, we highlight three examples, one in melanocytes and two in neurons, in which a class V myosin cooperates with microtubuledependent motors to drive pointto point organelle transport in this second step.

Class V myosin and the transport of melanosomes.
As with myosin V in fly photoreceptor cells, class V myosins have essential roles in the positioning of melanosomes within pigmentproducing cells in mammals, fish and amphibians 90 . Myosin Va is recruited to the surface of melanosomes in mouse melanocytes by a receptor com plex made up of the RAB GTPase RAB27A and one of its effector proteins, melanophilin 22,24,25,91 . Together, they form a tripartite complex, which connects the melano some to the cortical actin cytoskeleton follow ing the organelle's longrange, microtubuledependent transport to the distal end of the melanocyte's den drites. Numerous sites at which this receptor complex is regulate d have been identified .
The study of mouse melanocytes lacking myosi n Va (known as dilute melanocytes;

Recycling endosome
A generic, centrally located membrane compartment that receives endocytosed membrane receptors and recycles them back to the plasma membrane.

Long-term potentiation
(LTP). A form of synaptic plasticity that is thought to underlie memory formation, in which synapse use leads to long-term strengthening of the synapse.

AMPA receptors
The major excitatory ionotropic Glu receptors found in neurons. Their name comes from their ability to be activated by the artificial Glu analogue AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid).
melanosome transport and distribution 92 (FIG. 4b). This model stressed a role for myosin Va in tethering melano somes in the actinrich periphery following their transport there by longrange, micro tubuledependent transport. This was based on evidence showing that myosi n Va dependent melanosome capture in the periphery seemed to be sufficient to drive melano some positioning in cooperation with bidirectional, microtubulebased melanosome transport 92 . This was in contrast to the widely held and incorrect belief at the time that myosin Va drives the longrange trans port of melanosomes. Moreover, crucial evidence for myosi n Va dependent melanosome movement was lack ing. However, a class of melanosome movements (with a speed of ~0.15 μm s -1 ) that were myosin Va dependent (present in wildtype but not dilute melanocytes) and microtubuleindependent, and that exhibited little direc tional persistence, were identified 92 . These movements could be due either to myosin Va moving melanosomes on actin or to myosin holding the organelles on actin and the actin moving (see also REF. 93).
Significantly stronger evidence that class V myosins move melanosomes has been obtained from studies of fish and frog pigment cells (known as melanophores; FIG. 4c). Specifically, the ~0.04 μm s -1 melanosome move ments attributed to class V myosin within these cells can, over time, drive the longrange movement of the organ elle, can often exhibit significant directional persistence and can show a significant decrease in run length when actin dynamics are suppressed by drugs that stabilize Factin [94][95][96][97] . Moreover, highresolution tracking of mela nosome movements in vivo identified translocation events that occur in 35nmlong steps, a size which corresponds to the step size of class V myosin 98 . Perhaps the 'nail in the coffin' would be to complement class V myosinnull melanocytes or melanophores with versions of the myosin that walk more slowly and show a corresponding decrease in the speed of melanosome movement in vivo.

Myosin Vb transports endosomes in hippocampal neurons.
Previous studies have shown that vertebrate myosin Vb binds independently to RAB11A, a resident RAB GTPase in the recycling endosome, and to one of its effec tor proteins, RAB11 familyinteracting protein 2 (FIP2), through its GTD 99,100 . Although the exact transport event supported by myosin Vb in this pathway remains unclear, its importance has been demonstrated for many cargoes by the overexpression of a dominantnegative myosin Vb tail construct, which markedly inhibits recycling [101][102][103] . Defining the precise role of myosin Vb should be facili tated by the recent demonstration that microvillus inclu sion disease, a rare human disease characterized by the lack of apical microvilli on intestinal epithelial cells, is caused by lossoffunction mutations in myosin Vb 104 .
Recycling endosomes also exist at the base of dendritic spines in hippocampal neurons, and these vesicles move into the spine when the neuron is stimulated strongly, that is, with a stimulus that can increase intraspinal Ca 2+ levels and induce long-term potentiation (LTP). Importantly, exo cytosis from these relocated recycling endosomes, which occurs locally within the spine and involves an unidenti fied transport carrier, appears to serve two critical func tions 105 . First, it seems to serve as a major source of AMPA receptors for insertion into the postsynaptic membrane to create LTP. Second, it appears to provide the membrane lipids and other molecules needed to drive the activity dependent growth in spine size and other structural remodelling that accompany LTP induction 105 .
A recent study has now made a strong case for the idea that the stimulusdependent translocation of recy cling endosome membranes into spines is driven by myosi n Vb 106 (FIG. 5a). Moreover, evidence was pre sented that the spike in intraspinal Ca 2+ levels which occurs following the strong stimulation of spine NMDA (Nmethyldaspartate) receptors serves to activate myosin Vb by pushing it from its folded, quiescent 14S conformation to its extended, active 11S conformation. This Ca 2+ dependent unfolding of myosin Vb exposes the RAB11A and FIP2binding sites in its cargobinding GTD, allowing the myosin to attach to recycling endo some membranes and transport them into the spine along actin tracks. Myosin Vb appears to act, therefore, as a Ca 2+

Box 2 | The melansome receptor for myosin Va
The organization of the melanosome receptor for myosin Va 22,25,128 . RAB27A, when in its GTP-bound, active state and attached to the melanosome membrane via geranylgeranyl groups, binds its effector protein melanophilin, which then recruits myosin Va (see the figure). The labelled sites in the figure highlight the following: (a) the regulation of the nucleotide state of RAB27A by a RAB27A-specific guanine nucleotide exchange factor (RAB3-specific GDP-GTP exchange protein (RAB3GEP)) 129 and GTPase-activating protein (EBP50-PDX interactor of 64 kDa (EPI64)) 130 ; (b) melanophilin's conserved helixzinc-finger-helix SLP homology domain (SHD), which confers a specific interaction with RAB27A 131 ; (c) melanophilin's low-affinity binding site for the globular tail domain (GTD) of myosin Va 132 ; (d) exon F, an alternatively spliced, 27-residue exon that is present in the melanocyte-spliced isoform of myosin Va and binds melanophilin with high affinity 22 ; (e) a conserved site within the GTD of frog class V myosin that is phosphorylated during mitosis by Ca 2+ -and calmodulin-dependent kinase II (CaMKII), uncoupling the myosin from pigment granules 133 ; (f) the ability of melanophilin to drive the unfolding of myosin Va from its closed, quiescent, 14S conformation to its extended, active 11S conformation 44,45 ; (g) a binding site for filamentous actin (F-actin) within the carboxy-terminal domain of melanophilin which may increase the processivity of myosin Va when it is bound to melanophilin 134 ; (h) two SKIP motifs that mediate melanophilin's interaction with the core microtubule plus end-tracking protein end-binding 1 (EB1), through which melanophilin associates transiently with the plus end of growing microtubules 135,136 ; and (i) a short stretch of putative coiled-coil in melanophilin that is required for its function 137 . sensor to translate increases in spine Ca 2+ into postsynap tic membrane transport. However, it is unclear how this Ca 2+ dependent activation step can avoid the dissocia tion of neck calmodulins and the concomitant decrease in mechanochemical integrity of the myosin that is seen in vitro 107 .

Nature Reviews | Molecular Cell Biology
Of note, a second study, which used RNAi experiments and the expression of a dominantnegative tail construct (which may not be isoform specific), argued that it is in fact myosin Va that mediates the movement of recycling endosomes into spines, although its attachment to recy cling endosome membranes seems to be driven by a direct interaction with the C terminus of the AMPA receptor Glu receptor 1 (GLUR1) and with RAB11 (REF. 108). These findings are surprising, as pre vious studie s have only identified a role for myosin Vb in recycling endosome dynamics, and hippocampal neurons from myosin Va null mice exhibit normal postsynaptic GLUR distribution, excitatory synaptic transmission and both shortterm and longterm potentiation 109,110 .

Myosin Va transports the ER in Purkinje neurons.
The extension of the ER into the dendritic spines of cere bellar Purkinje neurons, which is required for synaptic plas ticity, does not occur when myosin Va is missing 111,112 , indicating a requirement for myosin Va in this process. Indeed, a recent study 113 has now provided clear evidence that myosin Va acts as a pointtopoint organelle trans porter to pull ER tubules into these spines at a maximum velocity of ~0.45 μm s -1 (FIG. 5b). Myosin Va was shown to concentrate at the tip of the ER tubule as it moves into the spine, and rescue of dilute mutant Purkinje neurons with slowwalking versions of myosi n Va results in cor responding decreases in the velocity of ER movement into spines. In keeping with the paradigm discussed above, this shortrange, myosi n Va dependen t move ment of ER into spines occurs downstream of long range, microtubuledependent transport of ER out neuronal dendrites. The myosin Va mediated transport of ER into spines was also shown to be required for the rise in spinal Ca 2+ levels downstream of metabotrophic GLUR1 (mGLUR1) activation, which leads to long-term depression 113 (LTD), the major form of synaptic plasticity exhibited by Purkinje neurons (FIG. 5b). Interestingly, a role for class V myosins in ER transport may be evo lutionarily ancient, given the role of yeast Myo4 in ER movement 82 and of class XI myosins, the plant versions of myosin V, in powering the rapid (60 μm s -1 ) transport of ER networks that drives cytoplasmi c streaming via viscous drag 114,115 .   Figure 5 | Type V myosins transport membrane compartments into dendritic spines to support synaptic plasticity. a | Myosin Vb transports recycling endosomes into the dendritic spines of hippocampal neurons following stimulation 106 . These recycling endosomes serve as a major source of AMPA receptors (AMPARs) for insertion into the postsynaptic membrane to support long-term potentiation (LTP). Importantly, the rise in spinal Ca 2+ levels downstream of strong NMDA (N-methyl-d-aspartate) receptor (NMDAR) activation is argued to open up the myosin, allowing it to bind RAB11 and RAB11 family-interacting protein 2 (FIP2) and thereby attach to recycling endosomes. b | Myosin Va translocates endoplasmic reticulum (ER) tubules into the dendritic spines of cerebellar Purkinje neurons 113 . In this system, the myosin appears to be unfolded and activated by interaction with its membrane cargo rather than by raised levels of cytosolic Ca 2+ . Note that the receptor in the ER membrane for myosin Va has not been identified. Importantly, the spike in spinal Ca 2+ levels downstream of strong signals that activate metabotrophic Glu receptor 1 (mGLUR1) requires the myosin Va-dependent transport of ER into the spine, as this Ca 2+ comes from that stored in spine ER. The inositol-1,4,5-trisphospha te (Ins(1,4,5)P 3 ) generated downstream of mGLUR1 activation through the G protein G q and phospholipase C (PLC) normally binds to the Ins(1,4,5)P 3 receptor IP 3 R1 in the spine ER membrane, causing Ca 2+ to flow out of the ER and into the spine head. In the absence of myosin Va and spine ER (as in the dilute mutant), the spike in spinal Ca 2+ levels, and the subsequent protein kinase C (PKC)-driven phosphorylation of AMPARs that stimulates their endocytosis, resulting in a reduction in their levels at the postsynaptic membrane (the basis for LTD), does not occur 112,113 . For both of the class V myosin-dependent membrane transport steps depicted in this figure to occur with reasonable efficiency, the organization of filamentous actin (F-actin) in spines must be such that it can support outward movement (that is, with the majority of filaments being oriented with their barbed ends pointing towards the postsynaptic density) [144][145][146] . Please see REF. 107 for an in-depth discussion of the Ca 2+ -dependent regulation of class V myosins in vitro and in vivo.

Anisotropic
A term referring to a complete lack of uniformity in orientation.

Filopodia
Thin, dynamic, cellular extensions containing actin filaments aligned in parallel with their barbed ends pointing towards the tip. They are often found in growth cones and at the leading edge of migrating cells.

Challenges in vertebrate cells.
The key to obtaining clear examples of class V myosindependent organelle move ment in animal cells, and presumably the reason why such movements are more often than not very hard to discern, almost certainly revolves around the organi zation of Factin in the cortex. In those cases in which class V myosindependent organelle transport has been identified in a definitive way, such as the movements of ER and recycling endosomes into dendritic spines, the actin tracks must be sufficiently aligned and polarized over the relatively short distances travelled to support the efficient, directed transport that is seen. Typically, however, the organization of Factin in the periphery of cells is highly anisotropic, although as it approaches the membrane the actin should become increasingly oriented with its barbed end facing 'out' , owing to the nature of the machinery that drives actin assembly at the plasma membrane 116 (note that filopodia, the only place in animal cells where highly polarized, barbed endout actin filaments exist, rarely contain organelles). Anisotropic actin organization does not allow organelles to move persistently in any one direction, making it diffi cult to demonstrate class V myosindependent organelle movement. Examples of this situation may include the class V myosindependent capture of melanosomes 92 and secretory vesicles 117,118 in the cortex. However, class V myosins can readily navigate the Arp2/3generated actin side branches that probably permeate much of the cortex, and they can also hop from one actin filament onto a crossing filament in anisotropic actin networks 30 . Moreover, such in vitro behaviours have now been witnessed in living cells using quantum dotlabelled myosin Va molecules 119,120 . Although the paths of individual myosin Va molecules look random in most cases, and their meansquared dis placements on timescales longer than 1 second resemble random diffusion, on short timescales (<0.2 seconds), clear evidence of directed transport is seen. These movements, which occur with the characteristic step size (36 nm) and speed (~0.5 μm s -1 ) of myosin Va, almost certainly represent very short periods of pro cessive movement within the highly anisotropic Factin network in the cell cortex 119 . This suggests that future studies of class V myosindependent organelle move ment should focus on observing very short periods of persistent movement in vivo.
Why does the cell not just build cortical actin tracks that are better suited to allow class V myosins to move organelles to the plasma membrane? Perhaps the cost of doing this, in terms of the other major functions sup ported by cortical actin (such as cell locomotion, cortical integrity and endocytosis), is too high. Moreover, driving relatively random walks on anisotropic actin networks that exhibit even modest barbedendout bias should allow class V myosins to eventually deliver vesicle s to the membrane. Alternatively, the organization of cortical actin in cells within tissues may be organized for more effective class V myosindependent vesicle deliverythat is, with filaments more highly aligned and polarize d towards the plasma membrane 121 .

Conclusions
Class V myosins clearly drive organelle transloca tion in plants and yeast. However, neither of the yeast class V myosins is processive (Myo4 is not even dimeric), so this organism uses 'tricks' like the cargodriven clus tering of Myo4 to allow robust transport. Conversely, although vertebrate class V myosins are highly processive, concrete evidence that they move membranes over appre ciable distances inside cells has been slower to appear. Indeed, debate continues regarding the extent to which class V myosins in animal cells serve as pointtopoint organelle transporters or dynamic organelle tethers 88,[122][123][124][125][126][127] , and several reviews have commented on the relative pau city of data supporting the transport role 88,123,125 . As dis cussed above, this may be due in large part to the fact that, in animal cells, the actin tracks for class V myosins are not usually organized in a way that supports persistent move ment in any one direction. Therefore, efforts to prove the existence of class V myosindependent organelle trans port in vertebrate cells should focus on identifying short range movements, such as those identified recently in the cell cortex using quantum dotlabelled myosi n Va 119 and in the dendritic spines of neurons 106,113 .