Cellular motors (kinesin, dynein and myosin) are ubiquitous. A major task in cell biology is to determine how they function in cells. Here we focus on myosin 10, an intrafilopodial motor, and show how imaging green fluorescent protein fused to myosin 10 or its tail domains can help us understand the function of this myosin.
A major question in cell biology is how substances (e.g. proteins, RNA and vesicles) are moved around inside cells. Their movement cannot rely on simple diffusion, as it would result in random movement, is not easily regulatable and would not allow a build up of high local concentrations. Therefore molecular motors are essential, as they result in highly directional movement, their activity can be regulated, and they provide highly efficient targeted movement.
Typically, a mammalian cell expresses over 20 different myosins, as well as kinesins and dynein. Many of these motors are involved in trafficking cargoes to the required destination, and the different motors co-operate. For example, the ubiquitous kinesin heavy chain, kif5b, interacts with myosin 5 in neurones , myosin 5 interacts with dynein  and there is also likely to be a connection between kinesin and dynein. It is likely that a single vesicle is coated with a number of different types of motor, each of which take a turn in transportation dependent on where they are in the cell and whether they are active.
Myosin 10 is one of the least well-understood molecular motors. It has a motor domain, a neck that consists of 3 ‘IQ’ motifs, to which calmodulin, and a myosin 10-specific light chain  binds (Figure 1). Following the IQ motifs, there is a region of putative coiled-coil that is approx. 125 amino acids long, three PEST sequences, which are cleaved by calpain  and a tail that consists of three types of domain : PH (pleckstrin homology), MyTH4 (myosin tail homology 4) and FERM (band 4.1, ezrin, radixin, moesin) (Figure 1).
The three tandem PH domains (PH1, 2 and 3), totalling 295 residues, are likely to target myosin 10 to its required intracellular location. PH domains bind to phosphatidylinositol phospholipids, rare lipids found in the plasma membrane and intracellular membranes . This binding can regulate protein activity, and/or target the protein to its required intracellular location. Analysis of the sequence of the β1–β2 loop of PH domains, which is indicative of their ability to bind to PtdIns(3,4,5)P3 , shows that PH2 is likely to bind to PtdIns(3,4,5)P3 (Figure 1). This has been demonstrated experimentally [6,7], and we have found that PH123 binds to PtdIns(3,4,5)P3 and PtdIns(3,5)P2 with approximately equal affinity, and to PtdIns(4,5)P2 with approx. 10-fold lower affinity in vitro (D. Tacon and M. Peckham, unpublished work) by a phospholipid overlay assay . PH1 is unlikely to bind to PtdIns(3,4,5)P3, as it lacks the charged residues (arginine or lysine) in the β1–β2 loop, important for interaction with the 3′ phosphate . Although PH3 does have these charged residues, the first residue in the loop, which is normally an amino acid with a short side chain (glycine, alanine, serine or proline) is replaced by the bulkier valine. Thus PH2 is likely to be responsible for the observed binding to PtdIns(3,4,5)P3, and could target myosin 10 to the membrane in response to agonists.
Less is known about the function of the MyTH4 and FERM domains. Intriguingly, the MyTH4 domain has also been found in a kinesin-like calmodulin-binding protein, where it was shown to bind to microtubules [9,10]. While the FERM domain binds to PtdIns(4,5)P2 in ezrin , the FERM domain of myosin 10 lacks the conserved lysines required for this interaction. However, it was recently shown that the FERM domain binds to β-integrins . The full-length tail domain binds to Mena/VASP (mammalian enabled/vasodilator stimulated phosphoprotein) , and it is also likely to bind to a dynein light chain . This might suggest connectivity between myosin 10 and dynein, as previously suggested for myosin 5.
Cellular function of myosin 10
Imaging the dynamic behaviour of myosins fused to GFP (green fluorescent protein) inside living cells can be informative, but this depends on the myosin. Intriguingly, myosin 10 is one of the few myosins that has been observed to move inside living cells , where it was shown to move bidirectionally inside filopodia. As the barbed ends of actin filaments face the tip of the filopodia, myosin 10 will only move actively towards the tip as, like most of the other myosins, it is a ‘barbed-end’ directed motor. Therefore its rearward movement in filopodia is either linked to rearward flow of actin filaments, or driven by a ‘backwards’ motor. The speed of movement towards the tips of filopodia measured in vivo  was similar to that measured in an in vitro motility assay with purified myosin 10: 0.3 μm/s .
Myosin 10 may transport components of the filopodial tip complex, such as Mena or β-integrin , and as such it could either work as part of a team, or work singly. Myosin 5 is a transporter that works singly. It is a ‘high duty ratio’ motor (reviewed in ), processive  and dimeric. However, myosin 10 is unlikely to be a ‘high duty ratio’ motor, is not processive  and we think it is likely to be a monomer (Figure 1). It has been predicted that myosin 10 is dimeric because it contains a region of putative coiled-coil (Figure 1). In fact, it is hard to find a good hydrophobic seam in the putative coiled-coil region, and there is a high proportion of charged residues in this region, similar to that found in myosin 6 (Figure 1). The homologous region of myosin 6 was also predicted to be coiled-coil, and hence myosin 6 was assumed to be a dimer, but in fact the evidence indicates that the full-length myosin 6 is monomeric . Therefore we think it is likely that myosin 10 will also turn out to be a monomer, and the role of the predicted coiled-coil sequence is unclear. This means that, unlike myosin 5, myosin 10 is likely to work as part of a team.
Imaging myosin 10
We have investigated the dynamic behaviour of GFP fused to full-length myosin 10, myosin 10 with a mutation in the motor domain, and GFP fused to the PH domains and to the MyTH4 domain in live cells (Figure 2). Confocal images for each of these constructs shows that expression of full-length myosin 10 GFP in COS-7 cells induces the formation of numerous long filopodia (Figure 2A), as reported previously . In contrast, expression of myosin 10 with a mutation in a highly conserved residue in the motor domain (Ile451→Phe) that is likely to affect motor activity does not induce filopodia formation. However, the mutant GFP-myosin 10 is still recruited into membrane ruffles, and it is also found in intracellular punctae (Figure 2B). This demonstrates that the motor domain must be active for full-length myosin 10–GFP to induce filopodia formation. The PH domains fused to GFP are similarly recruited into membrane ruffles and intracellular punctae, particularly in the perinuclear region (Figure 2C). In contrast, the MyTH4 domain fused to GFP is only found in intracellular punctae, particularly in the perinuclear region, and it is not recruited into membrane ruffles (Figure 2D). Its staining pattern is dependent on an intact microtubular cytoskeleton, as treatment of cells with nocadazole to depolymerize the microtubules disturbs its staining pattern (D. Tacon and M. Peckham, unpublished work), and this staining pattern is restored when nocadazole is washed out. The similarity in the localization of the PH domains and the mutant myosin 10 may suggest that the PH domains are responsible for targeting this myosin to the plasma membrane and intracellular punctae.
Confocal images of GFP–myosin 10 and tail domains
Time-lapse epifluorescence microscopy of COS-7 cells expressing the PH domains fused to GFP shows that these domains are recruited into membrane ruffles, and into newly formed macropinosomes (Figure 2E and Figure 3). We have found that the macropinosomes remain labelled by PH123–GFP for over 20 min as the macropinosomes mature, shrink and move inwards to the centre of the cell. PH123-labelled vesicles in the centre of the cell can be co-labelled with TRITC (tetramethylrhodamine β-isothiocyanate)-dextran, a marker for macropinosomes (D. Tacon and M. Peckham, unpublished work). It is important to note that the inwards movement of the labelled macropinosomes is unlikely to arise from the action of myosin 10, as this is a barbed-end-directed motor which will only move towards the plasma membrane. This myosin could be involved in forming the filopodia and membrane ruffles that are associated with macropinosome formation. However, the inward movement of the mature macropinosomes into the perinuclear region is more likely to be along microtubule tracks, and is likely to be driven by another motor, such as dynein. The association between dynein light chains and myosin 10 may be important for this.
Possible recruitment of myosin 10 to forming macropinosomes
The association of the PH domains with macropinosomes is not surprising, as we know that they will bind to PtdIns(3,4,5)P3 and this lipid has been implicated in macropinocytosis . However, the long-lived association of PH123–GFP with macropinosomes is more surprising. In contrast, PH domains fused to GFP from either PKB/Akt (protein kinase B), which binds to PtdIns(3,4)P2 and PtdIns(3,4,5)P3, or PLCγ (phospholipase Cγ), which binds to PtdIns(4,5)P2, do not label mature macropinosomes, although they are found in membrane ruffles and early macropinosomes (D. Tacon and M. Peckham, unpublished work). Either PH123–GFP binds to all three lipids [PtdIns(3,4,5)P3, PtdIns(4,5)P2 and PtdIns(3,5)P2] in vivo as well as in vitro, and this results in their prolonged association with macropinosomes, or they bind to PtdIns(3,4,5)P3 with much higher affinity than that for Akt-PH.
We have also shown that the PH domains have a high affinity for the plasma membrane by using TIRFM (total internal reflection fluorescence microscopy; ) to image single molecules of PH123 fused to GFP at the plasma membrane. We found that the PH domains bound to the plasma membrane with an apparent on-rate of 0.03 μM−1·μm−2·s−1 and a detachment rate constant of 0.05 s−1. The average time that the PH domains remained bound at the plasma membrane was approx. 20 s. Interestingly, by tracking the single molecules in the TIRFM images, we found that the membrane-bound PH domains had a very limited mobility when bound to the plasma membrane in the mouse myoblasts that we used for this study. More recently, we have found that the PH domains are more mobile in other cell types (G.I. Mashanov, M. Peckham and J.E. Molloy, unpublished work), and this may be related to the organization of the cortical actin cytoskeleton.
In summary, although we still have a lot to learn about the structure and function of myosin 10, imaging either myosin 10 or its tail domains fused to GFP is proving informative.
Lipids, Rafts and Traffic: A Focus Topic at BioScience2004, held at SECC Glasgow, U.K., 18–22 July 2004. Edited by G. Banting (Bristol, U.K.), N. Bulleid (Manchester, U.K.), C. Connolly (Dundee, U.K.), S. High (Manchester, U.K.) and K. Okkenhaug (Babraham Institute, Cambridge, U.K.)
We thank Richard Cheney (Duke University, Durham, NC, U.S.A.) for providing us with the full-length non-mutant isoform of myosin 10 GFP, Doreen Cantrell (University of Dundee, U.K.) for the Akt-PH domain, Vas Ponnabalam (University of Leeds, U.K.) for the PLCγ-PH domain, and Dario Alessi (University of Dundee, U.K.) for help with the PLO assay. This work was funded by BBSRC (the Biotechnology and Biological Sciences Research Council) and the Wellcome Trust.