Head-to-tail regulation is critical for the in vivo function of myosin V

The spatial organization of eukaryotic cells depends on a regulated cytoskeleton and motor proteins that use it as tracks for cargo transport and delivery. The major motor proteins implicated in cargo transport are the kinesins and dyneins that use microtubule tracks and unconventional myosins, most notably myosin Vs, which use actin filaments. The delivery cycle of transporting motors must be precisely regulated both spatially and temporally to allow for proper cargo capture and dissociation. Here we examine the in vivo consequences of defects in myosin V motor regulation.

Myosin Vs are highly conserved motor proteins that transport specific cargoes in fungal, plant (called myosin XI), and animal cells (Hammer and Sellers, 2012). In mammalian cells, myosin V motors have been implicated in transporting melanosomes (Wu et al., 1998), recycling endosomes (Lapierre et al., 2001), and the endoplasmic reticulum (Wagner et al., 2011). To perform these functions, all myosin V motors consist of two heavy chains with N-terminal motors dimerizing along a coiled-coil stalk, six associated light chains decorating each force-transducing lever arm, and two cargo-binding globular tail domains at the C terminus (Hammer and Sellers, 2012).

Biochemical and structural data has suggested that mammalian myosin Vs undergo an autoinhibitory interaction between the head domain and globular tail domain. The existence of the folded inhibited state of mammalian myosin V is supported by sedimentation analysis (Krementsov et al., 2004; Li et al., 2004; Wang et al., 2004), cryoelectron microscopy reconstruction (Liu et al., 2006; Thirumurugan et al., 2006), and ATPase activity assays (Li et al., 2008). Activation of the motor occurs after cargo receptor binding for MyoVa in vitro (Li et al., 2005; Sckolnick et al., 2013), as is the case for yeast myosin V in vivo (Donovan and Bretscher, 2012). Notably, the residues predicted to form the ionic bond between domains have been identified (Li et al., 2008; Nascimento et al., 2013). These include mouse MyoVa residues D121/D134/D136 in the head domain and K1706/K1779 in the tail domain. However, despite a decade of work characterizing this interaction biochemically, to our knowledge no one has defined how misregulation of myosin V affects motor functions such as cargo capture and recycling in vivo.

Budding yeast is an especially amenable system in which to probe the function of myosin V motors because of the absence of microtubule-based transport systems (Pruyne et al., 2004). In addition, spatially segregated areas of cargo capture in the mother cell and delivery to the daughter cell enable easy phenotypic analysis. The budding yeast myosin V motor Myo2p is responsible for transporting a wide array of organelles and secretory membranes to the daughter cell (Bretscher, 2003). Myo2p also retains the overall domain structure of mammalian myosin V motors, though its coiled-coil rod domain lacks flexible loops and is much shorter than mammalian myosin V.

Here we show that Myo2p is regulated by a head-to-tail interaction and that the length of the lever arm is critical for this interaction. We then document the in vivo consequences of motor misregulation. Defects in this regulatory mechanism lead to a host of cellular problems, including impaired cargo capture and delayed motor recycling.

We first performed alignment and structural modeling to determine Myo2p’s potential to undergo autoinhibition. The five residues predicted to form the ionic bond between the head and tail domains show strong conservation between yeast Myo2p and vertebrate myosin V motors (Fig. 1 A). Although the exact residues may change, the nature of the charges remains the same. These basic tail residues mapped to a patch on the face opposite from the secretory vesicle binding region, whereas the acidic residues in the head were clustered on a modeled Myo2p head in a manner consistent with MyoVa structural studies (Fig. 1 B; Coureux et al., 2003).

As previous attempts to detect a Myo2p head–tail interaction have been unsuccessful (Pashkova et al., 2006), we investigated this interaction by taking a charge complementation approach. The chromosomal MYO2 gene was mutated such that the basic residues in the interaction were changed to glutamic acid (E) and acidic residues were changed to arginine (R). Mutating the three acidic residues in the head of Myo2p generated the myo2-3DR allele (so named because all three residues in mouse MyoVa are aspartic acid), whereas mutating the two basic residues in the tail generated the myo2-REKE allele. These two alleles were also combined into the myo2-3DR-REKE allele, which effectively swaps the charged patches on the head and tail domains. All mutants expressed similarly to wild type (Fig. S1 A).

The myo2-3DR and myo2-REKE strains grew much slower than wild-type MYO2 cells (Fig. 1 C). Note that the strain containing myo2-REKE grew slightly slower than myo2-3DR; this might be related to the finding that the R1402/K1473 basic patch is also the binding site for exocyst complex component Sec15p, thereby disturbing this nonessential interaction (Fig. S1 B; Jin et al., 2011). To our surprise, the charge-reversal myo2-3DR-REKE allele grew nearly as well as wild type, showing that charge complementation between the head and tail is sufficient to return Myo2p to wild-type–like growth. These suppression data, in conjunction with previously published MyoV in vitro work, strongly imply that Myo2p undergoes a head–tail autoinhibitory interaction via these residues (Li et al., 2008).

To determine the effect of the mutant motors in vivo, we first examined their distribution in the cell. We tagged the chromosomal gene encoding Myo2p with a single copy of GFP at its C terminus in wild-type and mutant strains. These tagged strains were functional across a broad range of temperatures and showed no difference in growth compared with the untagged alleles. As expected, Myo2-GFP is polarized toward sites of growth in the daughter cell (Fig. 2 A). However, the bulk of the Myo2-3DR-GFP and Myo2-REKE-GFP motors hyperpolarize toward the bud tip; whole-cell summed projections of medium-budded cells show that 75% of all motors are present in the bud compared with 37% for wild type (Fig. 2 B). Strikingly, when charge reversal mutations in the head and tail were combined in the Myo2-3DR-REKE-GFP strain, the motor phenotype was similar to wild type, with 30% of all motors present in the bud (Fig. 2 B). Therefore, an interaction between the head and tail domains is necessary and sufficient to regulate myosin V in vivo to achieve a normal distribution.

The two internal loops present in mammalian myosin V coiled-coil stalks have been shown in vitro to be required for autoinhibition, likely because of the flexibility gained to bring the head and tail domains together (Li et al., 2006). Myo2p lacks these internal loops, though the length of the lever arm and coiled coil domains are nearly identical (~21 and 24 nm, respectively). We therefore explored the effect of disturbing this geometry by subtracting or adding light-chain binding IQ motifs, thereby shortening or lengthening its lever arm (Schott et al., 2002). Motors with 2, 4, or 8 IQ motifs on the lever arm showed significantly more motors in the bud compared with wild-type 6IQ motors (Fig. 2 C). Thus, the absolute length of the lever arm is important for proper motor regulation in vivo, with mismatched lengths resulting in more motors in the active state.

Next, we explored in detail how the motor delivery cycle is perturbed in the Myo2p charge mutants. A myosin V motor must be deactivated for efficient recycling from sites of delivery, and this occurs when it releases its cargo (Donovan and Bretscher, 2012). To determine if the head–tail interaction is necessary to deactivate the motor and allow normal recycling, we performed fluorescence loss in photobleaching (FLIP) experiments. Motors in the mother cell were bleached every 6 s, whereas the intensity of motor fluorescence in the bud was monitored every 2 s. Wild-type motors followed known recycling kinetics, with the vast majority of motors leaving the bud after ~30 s (Fig. 2, D and E; Donovan and Bretscher, 2012). Myo2-3DR-GFP or Myo2-REKE-GFP motors resided in the bud for much longer, with half-times of loss being ~2–3 min. This suggests that motors unable to form the head–tail interaction are hyperactive and unable to recycle. Strikingly, in the charge complementation Myo2-3DR-REKE-GFP allele, motors recycled at rates matching wild-type motors. Thus, the head–tail autoinhibitory interaction is required for proper recycling dynamics of myosin V.

What effect does misregulation of the autoinhibitory interaction have on cargo transport? Myo2p has two known essential cargoes: mitochondria, where Mmr1p and the Rab Ypt11p serve as alternative receptors, and secretory vesicles, where the Rab Sec4p and an unknown component dependent on the lipid PI4P serve as receptors (Itoh et al., 2002, 2004; Santiago-Tirado et al., 2011; Jin et al., 2011; Chernyakov et al., 2013). Importantly, these cargoes bind to different locations on the tail (Pashkova et al., 2006; Eves et al., 2012); Mmr1p binds to subdomain I, whereas the Rabs Sec4p/Ypt11p bind to subdomain II (and on the face opposite Mmr1p binding).

Wild-type cells show very robust polarization of the secretory vesicle marker GFP-Sec4 (Fig. 3, A and B). In contrast, in both charge mutant alleles there is a large buildup of GFP-Sec4–marked vesicles in the mother cell and a large decrease in the fraction of cells with polarized GFP-Sec4 (Fig. 3 B). This does not appear to be a defect in Sec15p binding, an interaction also using the R1402/K1473 basic patch (Jin et al., 2011), because the result is phenocopied in the myo2-3DR strain. Because both Myo2-3DR and Myo2-REKE motors are constitutively active and accumulate at the bud tip, it is likely that the defect in vesicle transport is caused by a scarcity of available motors in the mother cell. In the charge reversal myo2-3DR-REKE strain, GFP-Sec4 polarization is significantly restored, though not to quite to the same degree as seen in wild-type cells. Exocyst component Sec15-GFP was also polarized in the myo2-3DR-REKE allele (Fig. S1 C), suggesting that defects in polarized transport related to this site may not be completely caused by a compromised Sec15p–Myo2p interaction as has been previously reported (Jin et al., 2011), but instead caused by motor misregulation.

To explore if the failure of the unregulated mutants to transport secretory vesicles affected exocytosis of secretory vesicle cargo, we examined the internal pool of the β-endoglucanase Bgl2p, a marker for post-Golgi secretory vesicles (Harsay and Bretscher, 1995). As a benchmark, the conditional exocyst mutant sec6-4 accumulates substantial Bgl2p after a 1-h shift to the restrictive temperature (Fig. 3 C). 13% of the total Bgl2p population is present inside wild-type cells, which increases >2.5-fold in each Myo2p charge mutant (Fig. 3 C). The charge reversal myo2-3DR-REKE mutant also exhibited significantly increased Bgl2p accumulation; this could be a result of slight defects in vesicle fusion because the Sec15p interaction through R1402/K1473 is compromised (Fig. S1 B).

We next turned to see how mitochondria are distributed in the cell. In wild-type cells, mitochondria appear as a tubulated network in both the mother and bud as visualized with a mito-RFP probe (Fig. 3 D). In either of the myo2-3DR or myo2-REKE strains, mitochondria adopted a less tubular and aggregated appearance toward the back of the mother cell. This phenotype is also observed in cells that have an impaired ability to transport mitochondria into the bud (Chernyakov et al., 2013; Klecker et al., 2013). When both mutations were combined, there was a significant decrease in the fraction of cells with this phenotype and a majority of cells had a normal mitochondrial distribution and appearance (Fig. 3 E).

We also examined the orientation of the mitotic spindle, a process in which the receptor Kar9p binds to subdomain II of the Myo2p tail to link the motor to the end of a microtubule via Bim1p (Yin et al., 2000; Eves et al., 2012). Single-charge myo2 mutants had significantly fewer cells displaying oriented spindles, as scored by the spindle axis passing directly through the bud neck. Complementing charges in the myo2-3DR-REKE mutant fully restored orientation to wild-type levels (Fig. 3, F and G). Thus, misregulation of autoinhibition compromises the ability of Myo2 to orient the mitotic spindle.

The similar deleterious effects of the charge mutants and their suppression in the charge reversal mutant, for secretory vesicle polarization, mitochondrial distribution, and spindle orientation, are important because the cargo receptors bind to different regions of the Myo2p tail. Thus, it is unlikely that such large-scale cellular changes are caused by defects in binding cargo, but are instead caused by motor misregulation. Therefore, motor misregulation in vivo leads to severe cargo transport defects.

The data presented here indicate that a head-to-tail inhibitory interaction regulates yeast Myo2p as it does mammalian Myosin Va. More importantly, our results provide the first direct evidence that misregulation of a myosin V motor in vivo leads to wide-ranging cellular problems, including reduced abilities to capture and transport cargo as well as recycle after cargo delivery. Because the bulk of misregulated motors are hyperpolarized to sites of growth and kinetically delayed in recycling, the pool of motors available to allow continuous cargo transport is insufficient. Thus, myosin V motors have evolved and conserved the ability to remain firmly inactive until cargo is available, transport it in a processive manner, and finally coordinate delivery with motor deactivation.

Cells were grown and selected for using standard molecular biology techniques. Yeast transformations were performed using lithium acetate–based methods (Gietz et al., 1995). In brief, log phase 5-ml cultures were transformed for 45 min at 42°C with the plasmid/PCR product in a mixture containing ~40% wt/vol PEG-3350 (Sigma-Aldrich), 0.1 M lithium acetate (Sigma-Aldrich), and 0.28 mg/ml SS carrier DNA (Sigma-Aldrich). Dilution assays were spotted on appropriate media by 1:10 serial dilutions and incubated at the indicated temperature.

Protein sequences of myosin V motors were obtained from the Uniprot database (accession nos. Q9Y4I1, Q99104, Q02440, K4JEU1, and P19524) and aligned with JalView software (Waterhouse et al., 2009) using the TCoffeeWS algorithm. The location of the basic patch on the Myo2p tail was found using the Myo2p tail crystal structure (PDB accession no. 2F6H; Pashkova et al., 2006). A structural model of the Myo2p head domain was constructed by threading the primary sequence (residues 1–785) of the Myo2p head domain around a chicken MyoVa head structure (PDB accession no. 1OE9; Coureux et al., 2003) using the SWISS-MODEL server (University of Basel, Basel, Switzerland).

To generate the myo2-3DR mutant, constructs were created to mutate residues D123/E135/E137 to arginine and integrate them into the endogenous MYO2 locus. The region −500 to +750 (containing the promoter and the first 750 nucleotides of MYO2) was amplified from genomic DNA of a wild-type BY4741 strain with XhoI and BamHI sites at the end of the forward and reverse primers, respectively, using the Expand High Fidelity PCR kit (Roche). After overnight digestion, this was then ligated into a similarly cut pRS303 vector. A second homology region containing nucleotides −968 to −500 was amplified and ligated between the BamHI and NotI sites of the vector. To mutate residues in the head proposed to be involved in the interaction to arginine, the QuikChange II XL kit (Agilent Technologies) was used with primers designed to introduce the appropriate mutations. Sequencing was performed on all amplified products as well as after site-directed mutagenesis to ensure fidelity. After digestion of the unique BamHI site in the vector, the vectors containing the mutations were transformed into cells and selected for using SD-His media. Sequencing genomic DNA of the resulting colonies ensured the mutations were present at the myo2 locus.

A similar strategy was used to mutate the R1402/K1473 residues in the tail to glutamic acid, using the pRS303-Myo2tail-3′UTR integrating vector as a template (Schott et al., 1999). This base vector contains nucleotides 3272–4725 (those after an internal SpeI restriction site in the tail domain) along with a 197-nucleotide fragment of the 3′UTR; digesting with SpeI will linearize the vector and allow for integration of any introduced mutations into the endogenous MYO2 gene. To generate the myo2-3DR-REKE strain, this plasmid with associated REKE mutations was swapped into the pRS305 vector (LEU2 marker) to allow for its selection when transformed into myo2-3DR::HIS3. Wild-type and mutant expression levels were performed via glass-bead disruption methods and probed with a Myo2p tail polyclonal antibody (Schott et al., 1999; generated in rabbit against the Myo2p tail domain).

Motors were observed by appending GFP to the C terminus using a pRS306-cctail-GFP construct as previously described (Donovan and Bretscher, 2012). This construct contains the cctail region of MYO2 (nucleotides 2776–4725) fused to one copy of GFP at the C terminus, followed by the MYO2 terminator. When digested with the internal unique HindIII site, the vector integrates into the endogenous MYO2 locus. The appropriate mutations from the myo2-REKE allele were transferred into this construct before integration to ensure the mutations were not selected against. Sequencing of genomic DNA was performed to ensure the mutations remained.

Live cell imaging of log-phase cells was performed at room temperature (21°C) on a CSU-X spinning disc confocal microscope system (Intelligent Imaging Innovations) using an inverted microscope (DMI600B; Leica), 100×/1.46 NA objective, and either the CoolSnap HQ2 camera (Photometrics) or QuantEM EMCCD camera (Photometrics). Strains were imaged on a 2% agarose pad with synthetic complete media. For longer time-lapse experiments with photobleaching, a 35-mm dish pretreated with 0.05 mg/ml concanavalin A (EY Laboratories) to adhere the cells was used. Micrographs were acquired in Slidebook (Intelligent Imaging Innovations), which was also used to create maximum image projections for single images.

The percentage of motors in the bud was calculated in ImageJ using the integrated density of summed projections of the mother cell and bud, after subtracting out the background; only cells with medium-sized buds (2-µm diameter) were used in the analysis. FLIP experiments were performed such that a 3-µm z-stack of the cell (which included all of the medium-sized buds) was imaged every 2 s, with mother-cell bleach events occurring every 6 s. Bleaching was performed using the Vector control system (Intelligent Imaging Innovations) with 1-ms dwell times and laser power set to 60%. Analysis of FLIP photobleaching was also performed in Slidebook software, where the motor intensity in the bud of summed movie projections over time was calculated (after subtracting background and using an adjacent unbleached control cell for normalization).

Cargo phenotypes in charge mutants were observed and scored by transforming the appropriate cargo marker. Secretory vesicles were imaged using the endogenously integrated GFP-Sec4 marker (Donovan and Bretscher, 2012), which contains 400 nucleotides of the SEC4 promoter, GFP, SEC4, and the SEC4 terminator. Cells were scored as having polarized or not polarized secretory vesicles. Mitochondria were observed by transforming in the pRS306-mitoRFPff-ura3d integrating vector (Chernyakov et al., 2013) and scored as either having aggregated mitochondria or normal mitochondria in the mother cell. Spindle orientation was observed using a GFP-Tub1 construct (pAFS125-GFP-Tub1, containing the HIS3 promoter, GFP, and TUB1 locus and terminator) integrated into the TUB1 locus (Yin et al., 2000). Cells were scored as either oriented (with the spindle axis passing through the bud neck) or not oriented. Sec15-GFP::kanMX6 polarization was detected by C-terminal integration of a PCR cassette into the chromosomal SEC15 locus (Longtine et al., 1998).

To probe for secretion defects, the fraction of internal β-endoglucanase Bgl2p was determined by separating internal and external fractions. Cells were grown in 5 ml of yeast extract peptone-dextrose at 26°C to log phase. A control strain, sec6-4, was transferred to 37°C for 1 h. Cells were then killed in killing buffer (10 mM NaF/10 mM NaN₃ in 10 mM Tris, pH 7.4) and washed on ice in buffer B (1.2 M sorbitol, 25 mM K-P_i, and 25 mM BME, pH 7.4) containing killing buffer. Cells were spheroplasted to remove the cell wall with 100 µl of 25 µM Zymolyase 100T in buffer B at 37°C for 1 h. After a low speed spin at 500 g, the supernatant containing the external fraction was gently harvested from the pellet containing the internal fraction. The pellet was resuspended in an equal volume of 10 mM MOPS buffer as the supernatant, and then both fractions were boiled after addition of Laemmli sample buffer. Fractions of all strains were run out on SDS-PAGE gels, transferred by semi-dry methods, and probed using a rabbit-derived anti-Bgl2p antibody serum. The internal fraction was found by dividing the internal fraction by the total amount of Bgl2p in both fractions.

Student’s t test and one-way analysis of variance were used to determine statistical significance between measured values. A 95% confidence interval (P < 0.05) was accepted as significant.

Fig. S1 provides data showing charge mutant strains are expressed similarly to wild type and affect Sec15p binding and localization as expected. A list of strains used in this study is also included in Table S1.

We thank L. Xu for initial construction of the head domain vector and other members of the Bretscher Laboratory for their comments and suggestions during the course of this work.

This work was funded by Public Health Service grant GM39066 to A. Bretscher.

The authors declare no competing financial interests.