Myosins-I are widely expressed actin-dependent motors which bear a phospholipid-binding domain. In addition, some members of the family can trigger Arp2/3 complex (actin-related protein 2/3 complex)-dependent actin polymerization. In the early 1990s, the development of powerful genetic tools in protozoa and mammals and discovery of these motors in yeast allowed the demonstration of their roles in membrane traffic along the endocytic and secretory pathways, in vacuole contraction, in cell motility and in mechanosensing. The powerful yeast genetics has contributed towards dissecting in detail the function and regulation of Saccharomyces cerevisiae myosins-I Myo3 and Myo5 in endocytic budding from the plasma membrane. In the present review, we summarize the evidence, dissecting their exact role in membrane budding and the molecular mechanisms controlling their recruitment and biochemical activities at the endocytic sites.
Myosins-I are widely expressed actin-dependent motors that share a conserved structural organization. Besides the N-terminal actin-activated ATPase, all myosins-I bear a TH1 (tail homology 1) domain that binds acidic phospholipids . The motor head and the TH1 domain are separated by an α-helical neck that associates with CaM (calmodulin) or CaM-related light chains, often in a calcium-independent manner . In addition, the long-tailed myosins-I bear a Cext (C-terminal extension), which can trigger Arp2/3 complex (actin-related protein 2/3 complex)-dependent actin polymerization. The Cext includes a TH2 domain, which binds to filamentous actin , and an SH3 (Src homology 3) domain, which associates with proline-rich motifs . The fungal myosins-I also bear an acidic peptide which participates directly in the activation of the Arp2/3 complex [3–8]. The long-tailed protozoal and mammalian myosins-I lack this acidic peptide, but they might indirectly activate the Arp2/3 complex via association with CARMIL (capping protein, Arp2/3 and myosin-I linker) .
Myosins-I were first discovered in protozoa in the early 1970s , from which they were purified and extensively characterized at the biochemical level . The subsequent development of powerful genetic tools in protozoa and mammals and discovery of these motors in yeast during the 1990s allowed the demonstration of their specific roles in membrane traffic along the endocytic and secretory pathways, in protozoal vacuole contraction, in cell motility and in mechanosensing [12,13].
Early evidence indicating a role for the yeast myosins-I in the endocytic uptake
Saccharomyces cerevisiae is one of the genetic model systems that have contributed the most to define the function and regulation of myosins-I in clathrin-mediated endocytosis. The first myosin-I discovered in S. cerevisiae was named Myo3, based on the order of the myosin discovery . Deletion of MYO3 did not cause any apparent phenotype, indicating the existence of other functionally redundant genes . Using a PCR-based screen  and a genetic screen for mutations that cause synthetic growth defects in a myo3Δ background , the second myosin-I, Myo5, was discovered in 1996. Deletion of both myosins-I is synthetically lethal in the yeast background used by Geli and Riezman , and therefore a temperature-sensitive mutant was designed in order to analyse their cellular functions. The myo5-1 allele, encoding an E472K substitution in the actin-binding surface of the motor head, is temperature-sensitive in a myo3Δ background and exhibits a strong defect in the ligand-induced internalization of the G-protein-coupled receptor Ste2 when shifted to restrictive conditions . The uptake defect is specific, since the kinetics of secretion and traffic to the Golgi were unaffected . The function of myosins-I in endocytic uptake nicely correlated with previous studies demonstrating a strong requirement for actin  and a calcium-independent function for CaM  in the process. On the other hand, Goodson et al.  used a genetic background that allows growth of the myo3Δ myo5Δ double knockout. In agreement with a role of myosins-I in endocytosis, these authors demonstrated that the myo3Δ myo5Δ cells are unable to take up the fluid-phase endocytic marker Lucifer Yellow. In addition, they reported a strong defect in actin polarization in the myosin-I mutant and suggested the possibility that the polarization defect was primary to the endocytic defect . Subsequently, a direct role of the myosins-I in endocytic uptake has been convincingly demonstrated by studies showing its recruitment and function at the endocytic sites (see below). However, it is still not clear what the role of these motors in actin polarization might be. Polarization of certain membrane proteins depends on their proper endocytic uptake  and therefore the polarization defect might be secondary to the uptake block.
The yeast myosins-I are transiently recruited to the sites of endocytic uptake
The cellular localization of the yeast myosins-I was first investigated by indirect immunofluorescence [16,20]. Myo5 was shown to be present in a subset of cortical actin-patch-like structures, which were already suspected to correspond to the sites of endocytosis [21,22]. Recruitment of Myo5 to the cortical actin patches is at least partially dependent on the interaction of its SH3 domain with the yeast homologue of human WIP (Wiskott–Aldrich interacting protein) Vrp1 , a protein that bears several G-actin-binding sites .
The use of live-cell microscopy to study the dynamics of endocytic proteins at the yeast plasma membrane definitively unveiled the nature of the cortical actin patches . The group of Drubin first showed that these structures are highly dynamic structures with a lifespan of approximately 15 s and that the nucleation of actin is always preceded by recruitment of components of the endocytic clathrin coat . Massive actin polymerization coincides with the initiation of a 200 nm slow inward movement of the coat into the cytosol, which is converted into a fast inward movement on arrival of the yeast amphiphysins [24,25]. Idrissi et al.  demonstrated, using quantitative immunoelectron microscopy, that the slow inward movement corresponds to the growth of a tubular invagination of diameter 50 nm, which carries a hemispherical clathrin coat at the tip.
The dynamics of Myo5 relative to other endocytic markers was solved by Li and Drubin's laboratories. The lifespan of GFP (green fluorescent protein)-tagged Myo5 at the endocytic sites is approximately 10–15 s and its arrival coincides with the initiation of massive actin polymerization [6,27]. Consistent with a role for Vrp1 in recruiting Myo5 at the endocytic sites, Vrp1 was shown to arrive approximately 10 s before Myo5 .
The use of quantitative immunoelectron microscopy to analyse endocytic uptake further improved the resolution at which we could visualize the process . In 2008, Idrissi et al.  used statistical methods to describe the distribution of gold particles labelling different proteins along endocytic invaginations and demonstrated that the length of the profiles could be used as a parameter to describe their age. This kind of analysis showed that most Myo5 is strongly associated with the base of the invaginations during their elongation (Figure 1B). However, a small pool of Myo5 appeared on the clathrin coat immediately before fission, indicating that the myosins-I could have different functions during membrane invagination and vesicle scission .
Model for the mode of action of myosins-I at the base of endocytic invaginations
Dissecting the myosin-I biochemical activities required for endocytic uptake
A number of groups have given significant insights into the biochemical activities of the yeast myosins-I and their role in endocytosis. The observation that the E472K mutant abolishes endocytic uptake already suggested that the actin-activated ATPase might be essential to the process . In 2006, Sun et al.  demonstrated that purified Myo5 could actually translocate actin filaments at 0.17+0.04 μm/s in gliding assays and that mutations abolishing or slowing down motility (G132R and S357A) also prevent the slow inward movement of the coat.
Besides the ATPase, the myosins-I C-terminal actin, ANPA (actin nucleating promoting activity), is required for endocytic uptake [6,28]. Lechler et al. and Evangelista et al. [4,5] first pointed out the sequence homology of the yeast myosins-I C-terminal acidic peptide with that of the members of the WASP (Wiskott–Aldrich syndrome protein) family, and showed functional redundancy between myosins-I and Las17, the yeast WASP homologue. Interestingly, however, the myosin-I ANPA could not be demonstrated using purified components, suggesting that other co-activators were required. Consistently, it was shortly afterwards demonstrated that myosin-I-induced actin polymerization in the presence of yeast cytosol requires not only a functional Arp2/3 complex  but also the interaction of the myosin SH3 domain with Vrp1 . Lechler et al. , and later on Sun et al. , showed in pyrene–actin polymerization assays that the interaction with Vrp1 was not only necessary but also sufficient for Myo5 to develop a strong ANPA.
The first indication that the myosin-I/Vrp1 ANPA was required for endocytic uptake came from the observations that deletion of the Myo5 Cext or depletion of Vrp1 strongly prevents receptor-mediated internalization [28,29]. Some years later, it was demonstrated that disruption of the myosin-I/Vrp1 ANPA prevents the slow inward movement of the clathrin coat [6,30]. The myosin-I acidic peptide can partially be substituted by that of Las17, but the yeast WASP homologue has a more prominent role earlier, during initiation of actin polymerization [6,30]. Kaksonen et al. [24,25] had previously shown that, during the slow inward movement of the coat, addition of actin monomers probably occurs close to the basal plasma membrane, and they postulated that the ANPA should be located at the base of the invaginations. Consistently, most Myo5 accumulates at the base of the endocytic profiles  (Figure 1).
Besides a role of Myo3 and Myo5 in membrane invagination, a number of observations suggest that they also function to promote vesicle scission. Mutation of a conserved residue in the Myo5 TH1 domain (E901A) causes accumulation of plasma membrane invaginations , and deletion of MYO5 is synthetically lethal with deletion of the yeast amphiphysin RVS167 , which encodes a BAR (Bin/amphiphysin/Rvs)-domain-containing protein that participates in scission . Interestingly, the human homologue of Myo5, Myo1E, interacts with the GTPase dynamin  and follows its very same dynamics at the plasma membrane , supporting a conserved role of myosins-I in vesicle fission.
Regulation of the recruitment and the biochemical activities of the yeast myosins-I in endocytic uptake
One can envision that the recruitment and the activity of myosins-I need to be precisely regulated in time and space to generate productive forces capable of deforming the lipid bilayer and promoting vesicle scission. On the other hand, since myosins-I can induce polymerization of filamentous actin, which in turn will activate its ATPase, it is likely that their biochemical activities are tightly regulated to prevent energy waste.
Most studies on the regulation of the actin-activated ATPase of the yeast myosins-I have focused on the phosphorylation of the TEDS (Thr-Glu-Asp-Ser) site. The actin-activated ATPase of the protozoal and fungal myosins-I is induced by phosphorylation of a conserved serine or threonine residue (the TEDS site), positioned on an actin filament contact loop [33,34]. The protozoal myosin-I TEDS site kinase was identified as a member of the PAK (p21-activated kinase) family [35,36]. PAKs are ubiquitous Cdc42 (cell division cycle 42) and Rac-activated kinases . Numerous results demonstrate that the yeast myosins-I are also targets of the Cdc42-activated PAKs controlling actin assembly and polarization in yeast . However, even though the TEDS site phosphorylation is required for endocytic uptake [6,38], the role of the Cdc42/PAK signalling cascade in endocytosis is far from being understood. Phosphorylation of the TEDS site is essential for ligand-induced Ste2 internalization . However, Cdc42 and the PAKs are dispensable, suggesting that other signalling cascades can control TEDS site phosphorylation . Consistently, it was demonstrated that the yeast SGK (serum and glucocorticoid-induced kinase) homologue Ypk2 interacts with Myo5 and can phosphorylate its TEDS site . Ypk2 and its homologue Ypk1 are downstream of a signalling cascade controlled by sphingolipids and participate in the regulation of actin dynamics, translation initiation, cell wall integrity and endocytosis . Interestingly, the spatial distribution of actin patches roughly follows the polarized distribution of Cdc42 throughout the cell cycle . In contrast, Ste2 is evenly distributed over the cell surface during vegetative growth . Thus it is feasible that the Cdc42/PAK signalling pathway controls the activity of the myosin-I pool required for the compensatory endocytosis linked to polarized secretion, while Ypk1 and Ypk2 control endocytic uptake at other cellular locations. It remains to be determined how transient the TEDS site phosphorylation is and where are the PAKs and Ypks phosphorylating Myo3 and Myo5.
Recent evidence indicates that the myosin-I ANPA is also regulated, albeit that the signalling cascades leading to its activation have not yet been defined . The data provided by Grötsch et al.  indicate that the cytosolic Myo5 is kept in an autoinhibited conformation where the TH1 domain interacts with the Cext and prevents binding of Myo5 with Vrp1 and, as a consequence, Myo5-induced actin polymerization. The autoinhibited conformation is stabilized by binding of CaM to the Myo5 neck . Dissociation of CaM from Myo5, possibly at the endocytic sites, promotes binding to Vrp1 and the onset of massive actin polymerization  (Figure 2). What triggers dissociation of the tight myosin-I–CaM interaction is still uncertain. It is reasonable to hypothesize that myosins-I directly or indirectly recognize a physicochemical change at the endocytic sites, which in turn triggers CaM dissociation and Myo5 activation. Increasing evidence indicates that N-WASP (neuronal WASP)-induced actin polymerization is regulated by the release of an autoinhibitory interaction on association of N-WASP with certain lipids and/or on sensing specific membrane curvatures [41–43]. Both the lipid composition and the membrane curvature change locally during endocytic budding and therefore similar mechanisms might be involved in the recruitment and activation of the myosins-I.
Model for the regulation of the myosin-I ANPA at the endocytic sites
The genetic, the live-cell and the immunoelectron microscopy tools developed to study the endocytic uptake in yeast, together with the established assays to measure the biochemical activities of Myo3 and Myo5, provide a unique experimental system to further dissect the function and regulation of these conserved motors. A number of questions are still open in the field and their definitive answer will require carefully designed in vivo and in vitro experiments that can now be performed in yeast.
As an example, the lipid-binding characteristics of the yeast myosins-I remain to be determined and the role of the TH1 domain in endocytosis is currently unclear. Mutation in a conserved residue in the Myo5 TH1 domain (E901A) impairs endocytic uptake, but whether the effect is due to impaired lipid binding was not established . Even though all myosins-I bear a TH1 domain, the specificity of lipid binding varies depending on the motor studied ( and references therein). At least in some cases, binding to PtdIns(4,5)P2 contributes to targeting the myosins-I to their cellular locations [44–47]. PtdIns(4,5)P2 is actually enriched at endocytic sites . However, the TH1 domain of the yeast myosins-I is not essential for their recruitment to clathrin-coated pits  and, in fact, the TH1 domain rather precludes Cext-mediated recruitment of Myo5 . Alternatively, and similar to N-WASP (see above), the interaction of Myo5 with certain lipids via the TH1 domain could have a signalling role and participate in the release of autoinhibition.
Another almost unexplored field is the regulation of the myosins-I by post-translational modifications other than the phosphorylation of the TEDS site. Over the last few years, the improvement of proteomic techniques has contributed to identify new sites phosphorylated in vivo, which lie within different regions of the myosins-I, including the Head, the TH1 domain and the Cext (Table 1). However, except for the TEDS site, the functional significance of Myo3/Myo5 phosphorylation in these locations is completely unknown.
Phosphorylated residues are followed by an asterisk.
Phosphorylated residues have been localized with >95% certainty, except when they are shown in italic. The position may change in the literature depending on the sequence database used. Residues annotated in this Table came from the Uniprot/KB (Universal Protein Resource Knowledgebase) database.
Residues are assigned to Myo3, but they might also correspond to Myo5 Ser775 and Ser777 respectively.
Identifying the lipid species that physiologically interact with the yeast myosins-I and defining the signalling cascades involved in the post-translational modification of these motors will probably reveal new levels of regulation and/or unveil novel functions of these conserved proteins.
Cellular Cytoskeletal Motor Proteins: A Biochemical Society/Wellcome Trust Focused Meeting held at Wellcome Trust Genome Campus, Hinxton, Cambridge, U.K., 30 March–1 April 2011. Organized and Edited by Folma Buss (Cambridge, U.K.) and John Kendrick-Jones (MRC Laboratory of Molecular Biology, Cambridge, U.K.).
This work was supported by the Ministerio de Ciencia e Innovación [grant numbers BFU2008-03500 and CSD2009-00016 (to M.I.G.)].
These authors contributed equally to this work.