Intracellular movement is a fundamental property of all cell types. Many organelles and molecules are actively transported throughout the cytoplasm by molecular motors, such as the dimeric type V myosins. These possess a long neck, which contains an IQ motif, that allow it to make 36-nm steps along the actin polymer. Live cell imaging of the fission yeast type V myosin Myo52 reveals that the protein moves rapidly throughout the cytoplasm. Here, we describe analysis of this movement and have established that Myo52 moves long distances on actin filaments in an ATP-dependent manner at ∼0.5 μm/second. Myo51 and the microtubule cytoskeleton have no discernable role in modulating Myo52 movements, whereas rigour mutations in Myo52 abrogated its movement. We go on to show that, although dimerisation is required for Myo52 movement, deleting its neck has no discernable affect on Myo52 function or velocity in vivo.
Myosin V proteins are dimeric-actin-associated motor proteins involved in diverse cell biological processes (Mehta et al., 1999). They possess four distinct domains: the motor domain, the neck region that normally contains six IQ motifs to which light chains bind, the coiled-coil region to facilitate dimerisation and the globular tail to which cargoes bind. Although type V myosins have been characterised biochemically, the regions within the motor domain that give it its distinctive motor activity are not yet well understood. Recent studies have brought into question fundamental properties of myosin Vs, including the role played by the neck region in determining the step-size of myosin Vs when `walking' along actin cables. Step size is normally 36 nm, which coincides with the helical repeat of actin (Mehta et al., 1999). Research using myosin V head constructs (lacking neck and tail domains) has suggested that step size is independent of the presence of IQ domains (Tanaka et al., 2002; Watanabe et al., 2004). This is in contrast to studies using full-length myosin V, which indicate a direct relationship between neck-length, step size and sliding-filament velocities, with myosin V possessing six IQs moving significantly faster (0.36±0.08 μm/second) than a myosin V with a shorter, two IQ-motif-containing neck (0.29±0.07 μm/second) (Purcell et al., 2002; Sakamoto et al., 2003; Sakamoto et al., 2005; Schott et al., 2002). In addition, it is also unclear at present whether myosin Vs have a conserved motor function between organisms, because studies to date suggest the budding yeast and Drosophila myosin Vs move non-processively along actin (Reck-Peterson et al., 2001; Toth et al., 2005).
Myo51 and Myo52 are the fission yeast type myosin Vs, each with distinctive functions. Myo51 primarily functions during the meiotic lifecycle (A. Doyle, R. M. -G., S. B. and D. P. M., unpublished), whereas Myo52 has multiple functions during the vegetative cycle. Myo51 localises to the actin ring during the mitotic cell cycle, whereas Myo52 localises to foci that move throughout the cytoplasm, concentrating at regions of cell growth (Motegi et al., 2001; Win et al., 2001). Here, we characterise Myo52 movements and movements of novel Myo52 mutants – including rigour mutants; of a Myo52 mutant that lacks its coiled-coil domain and is therefore unable to dimerise; and of a mutant lacking its entire neck domain. The latter moved and functioned normally within the cell, indicating the neck domain is not required for myosin V motor function in vivo, and that properties other than the neck size determines the cellular velocity of the protein.
Results and Discussion
Myo52 in vivo movements
The fission yeast, Schizosaccharomyces pombe, is amenable to live cell imaging techniques allowing in vivo analysis of movements of one of its class V myosin, Myo52. Using strains in which the chromosomal copy of the myo52 gene had been fused to cDNA encoding GFP at either the N-terminal (Myo52-nGFP) or the C-terminal (Myo52-cGFP) end, we analysed over 1000 individual in vivo movements (supplementary material Movie 1) of Myo52, which revealed at least two distinct types of Myo52 movement. One was a brief, slow (∼0.25 μm/second) and erratic movement with Myo52 moving a short distance before either stopping or moving in a different direction (focal movement had no discrete direction over three subsequent time points, with movements seeming random; Fig. 1a, yellow and green arrows). The other was of a higher velocity and with an easily discernable overall direction towards either a cell pole or the cell equator, with foci moving on a discernable path for at least three subsequent frames (often up to 20 frames; Fig. 1a, red, white and blue arrowheads). Kymograph analyses highlight the distinctions between the rapid and slow classes of movement (Fig. 1g). The position of the GFP fluorophore on Myo52 did not affect the velocity or duration of movements (Myo52-nGFP: 0.502±0.09 μm/second; Myo52-cGFP: 0.511±0.07 μm/second). We observed no bias in rapid Myo52 movements towards any particular cellular locus, or any significant differences in the velocity of cytoplasmic Myo52 movements (e.g. any correlation to the cell cycle, differential brightness of Myo52 foci, Myo52 expression levels or cellular location). Bidirectional movements were often seen (Fig. 1a, white arrow; Fig. 1g,h, asterisks), probably brought about by the complex actin filament network in S. pombe.
Myo52 movement and microtubules
Studies from diverse organisms show a functional relationship between Myosin V proteins and microtubules (Cao et al., 2004; Huang et al., 1999; Motegi et al., 2001; Wu et al., 2005). To determine whether rapid long-distance mobility of Myo52 relies upon the movement of kinesin on microtubules, its movements were monitored in cells treated with the microtubule-depolymerizing drug carbendazim. Addition of 25 μg/ml carbendazim led to a rapid disappearance of microtubules, seen in nmt81gfp-atb2 cells (Fig. 1c, arrows). Microtubule depolymerisation had no affect upon Myo52 movement or its ability to switch direction (Fig. 1h; supplementary material Movies 2 and 3). Furthermore Myo52 movement was not affected in strains that lack kinesins with overlapping function – such as the two Kar3-like kinesins (pkl1Δklp2Δ cells) – or both central-motor-centromere-associated kinesins (klp5Δ klp6Δ cells), and in strains lacking any single kinesin alone. Consistent with these findings Myo52 foci were never seen associated with microtubules when visualised simultaneously in the same cells (supplementary material Fig. S1a). Simultaneous observation of Myo52 and microtubules in this HcRed1-myo52 gfp-atb2 strain also allowed us to follow recruitment of Myo52 to the cell equator during the metaphase-anaphase transition, when microtubules are incorporated into the short (2.1±0.4 μm) mitotic spindle at the cell equator. Myo52 then moves from the cell tips, extending 6.4±0.3 μm from the cell equator, which indicates that microtubules cannot be involved in the medial recruitment of Myo52. Inhibition of this medial recruitment by microtubule depolymerisation (Motegi et al., 2001) is therefore likely to be due to the action of either a checkpoint mechanism or regulating molecules recruited to the cell equator by microtubules.
Myo52 movement is dependent upon actin filaments
We went on to establish whether Myo52 movement is dependent upon actin. When compared with cortical actin patches, actin filaments have a higher susceptibility to the F-actin depolymerising affects of the drug latrunculin A and allows the opportunity to examine Myo52 in the absence of specific components of the actin cytoskeleton. Actin structures can be monitored simultaneously by following movements of the actin patch component Crn1 (Pelham and Chang, 2001) (supplementary material Movie 4). Treatment with 1 μM latrunculin depolymerised actin filaments as determined by actin patches losing their polarised localisation pattern and ceasing to move (Fig. 1e, white arrow). Latrunculin also abolished rapid long-distance Myo52 movements (Fig. 1i; supplementary material Movie 5) but did not affect slow short-range movements. The addition of 20 μM latrunculin resulted in the depolymerisation of all F-actin structures, as observed by the loss of a discrete GFP-Crn1 signal (Fig. 1f – white arrow). Similarly, all measurable Myo52 movements were abolished and foci disappeared within a few minutes (Fig. 1f; supplementary material Movie 6).
We investigated the role of different actin-containing structures in Myo52 movements by using conditional mutants. Inactivation of the S. pombe tropomyosin homologue Cdc8 destabilises actin filaments, whereas inactivation of profillin (Cdc3) renders cells incapable of forming F-actin structures (Balasubramanian et al., 1992; Balasubramanian et al., 1994). These conditional mutants confirmed our findings in that long distance movements were abolished in cdc8-110 (Fig. 2a) and cdc3-124 (Fig. 2b) cells at the restrictive temperature. In addition, Myo52 was only seen to move at the cell equator in cells that lack the formin For3 and that only possess CAR-associated actin filaments (Fig. 2c, supplementary material Movie 7) (Feierbach and Chang, 2001). Together, these data indicate that Myo52 moves upon actin filaments to yielding rapid long-distance movements.
Myo52 moves actively upon actin filaments
Having established that rapid Myo52 movements are dependent upon the presence of actin cables, we asked whether Myo52 movement arises from an attachment to polymerizing and depolymerising actin cables, and therefore represents a passive mode of transport. We compared velocities of the actin patch component Crn1, which gives a direct readout of actin-cable dynamics, with those of directional Myo52 movement. The velocity of 1375 Myo52-foci movements fitted a Gaussian distribution with an average velocity of 0.51 μm/second (Fig. 2d), with no significant difference between the distribution of Myo52 movements to poles or equator. As Myo52 velocity does not change at different positions within the cell, it suggests that mechanisms regulating Myo52-driven movement operate uniformly throughout the cell. A population of 247 individual, directed Crn1 foci movements fitted a narrow Gaussian distribution, with a mean velocity of 0.28±0.04 μm/second. Student's t-test analysis indicated that the two populations differ significantly (99.99% confidence), and show that Myo52 moves almost twice as fast as actin filaments grow and shrink, indicating it is not moving passively on growing actin polymers.
As Myo52 movement is unaffected by the absence of Myo51 (Table 1), its movement is likely to be a result of its own ATPase activity and should, therefore, be dependent upon ATP. To investigate this further, cells were treated with the oxidative-phosphorylation uncoupler carbonyl cyanide p-trifluoromethoxyphenyl-hydrazone (FCCP), to rapidly reduce levels of cellular ATP (Cheng et al., 2003). At concentrations of FCCP that reduced ATP levels more than 1000 times (determined by luciferase assay), Myo52 foci ceased to move throughout the cytoplasm (Fig. 2f), indicating that the movement of Myo52 upon actin filaments is an active process.
Introducing mutations within Myo52 can be used to modulate its interaction with actin, which in turn might affect its cellular velocity. Mutating residues within the conserved relay loop responsible for communicating movements between the active site and the actin-binding site (Tsiavaliaris et al., 2002) or mutating the conserved glycine residue within the reactive thiol region (Batra et al., 1999; Uyeda et al., 2002) can change the affinity of myosin for actin. We therefore created strains that constitutively express Myo52-nGFP and contain comparable mutations within its motor domain. None of these mutants fully complemented the myo52Δ phenotype, and resulted in either a reduction in Myo52 velocity (Y509G & G689A) or abolition of Myo52 movements (F490A and G689V) (Fig. 2g,h). These data demonstrate that the movement of Myo52 foci is the result of their own Myo52 motor activity, which is necessary for Myo52 function in the cell.
Movement and velocity of Myo52 are not dependent upon the neck region
Having established that Myo52 moves actively on actin cables, we made use of this in vivo assay to examine the role individual domains play in facilitating its movement (Fig. 3a). Using the yeast two-hybrid assay, we examined whether (1) the short coiled-coil domain of Myo52 (107 residues) can dimerise – because this domain is significantly longer in myosin V proteins of Drosophila and vertebrates (500 residues) and, (2) whether Myo52 associates with calmodulin, a conserved IQ-motif-binding protein. The ability of the Myo52 tail (neck, coiled-coil and globular-tail domain; Fig. 3a) to interact with itself suggests that Myo52 can indeed dimerise. Its high affinity to the fission yeast calmodulin Cam1 suggests that calmodulin can bind to Myo52 (Fig. 3b).
Constructs that express Myo52-nGFP but lack the entire Myo52 neck region (GFP-Myo52ΔIQ, lacking residues 800-940) or the coiled-coil domain (GFP-Myo52ΔCC, lacking residues 940-1060) were created and integrated into the genome of myo52Δ cells. Both truncations resulted in the expected increase of mobility in SDS-PAGE experiments (Fig. 3c). Myo52ΔCC was seen to localise only weakly to the cell equator and only partially complemented the growth defects associated with myo52Δ cells (Fig. 3d). It never localised to motile foci within the cell, indicating that dimerisation is required for Myo52 function and in vivo movement. By contrast, the cellular distribution of Myo52ΔIQ was indistinguishable from that of the full-length protein (Fig. 3e). Kymographs of timelapse movies of GFP-Myo52ΔIQ cells demonstrated that the protein was capable of making slow, rapid, and bidirectional movements (Fig. 3f, supplementary material Movie 8). The Myo52ΔIQ truncation was able to complement the growth and morphology defects associated with myo52Δ cells (Fig. 3e). Myo52ΔIQ did not associate with the IQ-motif-binding calmodulin light chains and is unlikely to be able to make the predicted 36-nm steps along the actin helix repeat. However, these data demonstrate that deleting all six IQ domains has no effect on in vivo movement of Myo52 and suggest the neck region is not required for it to function within the cell. Analysis of 233 rapid, directed movements of Myo52ΔIQ revealed that the velocities (mean velocity 0.522±0.14 μm/second) fitted a tight Gaussian curve (Fig. 3g). The mean velocity of Myo52ΔIQ does not differ significantly from that obtained for the full-length protein (0.502±0.09 μm/second), which suggests that deletion of the IQ motifs does not have an adverse affect on the motor properties of the protein. The ability of Myo52 to dispense with its neck for normal in vivo function highlights differences in the evolutionary and functional significance IQ motifs of the neck regions play in myosin V proteins of different organisms; moreover, this phenomenon has not been found in any vertebrate myosin V described to date, showing that the cellular velocity of Myo52 is determined by properties other than neck length. Multiple Myo52 dimers may associate into arrays capable of working together with coiled-coils acting as lever arms, indicating that Myo52 might be capable of functioning non-processively under these conditions. It is unlikely that the coiled-coil region of Myo52 replaces the IQ neck region as a lever arm in Myo52ΔIQ, because its shorter length (97 residues compared with 139 residues) would result in the protein having a reduced cellular velocity. In addition, the lever would be unable to operate in a processive manner, because there would be only one lever for two heads. Similarly, the short length of the coiled-coil domain indicates that the region is not used as a flexible linker – like the coiled-coil within myosin VI – to allow stretching between actin sites. Even if the coiled-coil domain were able to unwind completely it would not be capable of reaching the same distance as the neck.
Although removal of the IQ domains had no effect on the mean velocity of Myo52, the distribution of Myo52 and Myo52ΔIQ foci velocities differ significantly, and indicate that the IQ domain might regulate Myo52 function – potentially through binding to light chains or by allowing the globular tail domain to associate with the motor domains, thus inhibiting motor function (Liu et al., 2006; Thirumurugan et al., 2006).
Materials and Methods
Yeast cell culture and strains
5×107 cells were harvested from mid-log phase cultures of S. pombe cells, washed in ice-cold STOP buffer (Moreno et al., 1991; Simanis and Nurse, 1986), and frozen in liquid nitrogen prior to storage at –80°C. Protein was extracted by resuspending cells in EXPY buffer [HE buffer (Simanis and Nurse, 1986) supplemented with 1 μg/ml AEBSF, 1 μg/ml antipain, 1 μg/ml aprotinin, 10 μM benzamidine, 1 μg/ml chymostatin, 1 μg/ml E64, 1 μg/ml leupeptin, 1 μg/ml pepstatin, 5 μM phenanthroline, and 1 mM PMSF] and disrupting them with 500 μl of acid-washed glass beads in a ThermoSavant Fastprep 120 (MP Biochemicals, Irvine, USA). Samples were then boiled with SDS-PAGE loading buffer. Western blotting detection was carried out using alkaline phosphatase reagents. Anti-GFP antibodies (generous gift from W. Gullick, University of Kent) were used at a 1:100 dilution.
Live cells were grown, visualised and images were processed as described previously (Grallert et al., 2006). Cells were visualised on a Deltavision Spectris system (Applied Precision Instruments), based around an Olympus IX71 microscope. Enclosed in an opaque environmental chamber, the system uses a Zeiss Alpha-Plan-Fluar ×100 1.45NA objective lens and a Sedat ET-filter set (Chroma) which is located on external filter wheels. Illumination is via a Sutter 300W xenon light source, which is delivered to the microscope through a liquid light guide. All of the equipment is isolated through an anti-vibration table and all components that create vibration are isolated from the microscope. The hardware is controlled by a Win95 PC, whereas the capture software is on a Linux Redhat PC; the software used to control image capture is softworx (API).
Cells were cultured and imaged in minimal media at 25°C (unless stated otherwise) as described above. Once the type of movement had been determined (by following its movement using a maximum projection timelapse of the z series stacks), the time elapsed and distance moved (3D) between the frames and the moved dot were determined manually using Imaris image analysis or Metamorph particle-tracking software (Molecular Devices, Downington, PA). The tracking continued until either no detectable movement was observed or until the dot moved out of the field of view. This was repeated for several 100, independent Myo52 movements and used to describe a population distribution of Myo52 movements.
We thank P. Nurse and F. Chang for strains, and I. Hagan and M. Geeves for their support and comments on the manuscript. This work was supported by the BBSRC (a David Phillips Fellowship to D.P.M.) and Cancer Research UK (CRUK).