|Home | About | Journals | Submit | Contact Us | Français|
During secretory events, kinesin transports cargo along microtubules, then shifts control to myosin V for delivery on actin filaments to the cell membrane . When kinesin and myosin V are present on the same cargo, kinesin interacts electrostatically with actin to enhance myosin V-based transport in vitro . The relevance of this observation within the cell was questioned. In budding yeast, overexpression of a kinesin-family protein (Smy1p) suppressed a transport defect in a strain with a mutant class V myosin (Myo2p) . We postulate that this is a cellular manifestation of the in vitro observation. We demonstrate that Smy1p binds electrostatically to actin bundles. While a single Myo2p cannot transport cargo along actin bundles, addition of Smy1p causes the complex to undergo long-range, continuous movement. We propose that the kinesin-family protein acts as a tether that prevents cargo dissociation from actin, allowing the myosin to take many steps before dissociating. We demonstrate that both the tether and the motor reside on moving secretory vesicles in yeast cells, a necessary feature for this mechanism to apply in vivo. The presence of both kinesin and myosin on the same cargo may be a general mechanism to enhance cellular transport in yeast and higher organisms.
Polarized transport in budding yeast is carried out by Myo2p, a class V myosin that transports organelles along actin cables to the growing bud . A temperature-sensitive mutant strain, myo2-66, accumulates secretory vesicles at the restrictive temperature [5, 6]. Myo2p expressed by this strain has a charge reversal point mutation in the actin-binding interface , resulting in defects in actin binding and in vitro motility (Table S1) . Overexpression of the kinesin-family protein Smy1p partially suppresses the transport defect in the myo2-66 strain. SMY1 is not essential in wild-type cells, but is essential in the myo2-66 strain. Smy1p and Myo2p both localize to regions of polarized growth, and Smy1p overexpression enhances Myo2p localization in wild-type cells [3, 7]. Smy1p does not use an alternative microtubule pathway to transport secretory vesicles. Overexpression of Smy1p still compensated for the mutant Myo2p when microtubules were depolymerized, or when Smy1p was mutated to abolish potential motor activity . The molecular mechanism by which a kinesin-related protein compensates for a defective myosin is unknown.
To characterize the interaction of purified Smy1p with Myo2p and actin, we expressed Smy1p and Myo2p, each with a C-terminal biotin tag for binding to streptavidin-coated quantum dots, in Sf9 cells (Fig. S1). Full-length Myo2p has a long lever arm with six IQ motifs, followed by an alpha-helical coiled-coil region that dimerizes the molecule, and a globular tail that binds cargo . Myo2p was inferred to be non-processive, based on the concentration dependence of the rate of actin filament movement in an ensemble in vitro motility assay (Fig. S2) [8, 11]. A non-processive motor remains bound to its track for a small part of its ATPase cycle, and cannot step continuously along the track as a single molecule. This contrasts with the processive motors kinesin-1 and vertebrate myosin Va, which can move continuously as single molecules for a micron or more along their respective tracks [12, 13].
The molecular properties of Smy1p have never been characterized. The N-terminal motor domain of Smy1p has significant sequence similarity to the motor domains of the kinesin superfamily. The ATP binding region conforms to the consensus sequence for the P-loop (GX4GKT), but is atypical in containing a proline residue (GPSFSGKS). A region of predicted alpha-helical coiled-coil which dimerizes the molecule follows . The C-terminal domain is divergent from kinesin .
Total internal reflection fluorescence (TIRF) microscopy was used to determine if Smy1p interacts with microtubules (Fig. 1a). Streptavidin-coated quantum dots (Qdots) were mixed with an excess of biotinated Smy1p, resulting in an estimated ~4–6 Smy1p molecules per Qdot. Smy1p-Qdots bound to but did not move unidirectionally along microtubules. Instead, Smy1p executed a one-dimensional (1D) random walk along the microtubule (Fig. 1b, Supplemental Movie S1). The mean squared displacement versus time was linear, indicating diffusive behavior (Fig. 1c, diffusion constant = 0.11 ± 0.07 µm2/sec, mean ± S.D.). This behavior was identical in absence of nucleotide, and thus is not driven by motor activity. If Smy1p is an active but non-processive motor, multiple Smy1p molecules will be required to move microtubules. This was not the case, because an ensemble of Smy1p molecules attached to a coverslip was able to bind microtubules, but not move them (data in Fig. 2 establishes that Smy1p binds to coverslips). Conventional kinesin-1 under these conditions moved microtubules at a speed of 0.49 ± 0.03 µm/sec. FPLC analysis of bound nucleotide demonstrated that Smy1p binds but does not hydrolyze MgATP. Smy1p is thus not an active microtubule-based motor.
The ability of Smy1p to interact with actin was assessed by TIRF microscopy (Fig. 1d). The actin cables that serve as a track for Myo2p in budding yeast were mimicked by using fascin to form bundles of parallel actin filaments spaced ~9 nm apart . Smy1p-coated Qdots bound to actin bundles (Fig. 1e, Movie S2). Approximately half of the bound Qdots diffused along the actin bundle and half were stationary. Diffusing Qdots were constrained to an ~1 µm length of actin, in contrast to the >5 µm that were explored when Smy1p diffused on a microtubule. Qdots, or Qdots coated with biotinated bovine serum albumin, did not decorate actin bundles.
The number of Smy1p-saturated Qdots bound per µm of actin bundle decreases as the ionic strength increases (Fig. 1f), indicating that the binding is electrostatic. Fewer interactions and a steeper ionic strength dependence were observed when Qdots with a single bound Smy1p were added to actin bundles. Multiple Smy1p molecules likely enhance interactions with actin filaments in the bundle, thus stabilizing the binding of the complex.
This study was motivated by an in vivo observation, which implies that Smy1p interacts with actin cables strongly enough to suppress the transport defect in the myo2-66 strain. How is this reconciled with the observation that the interaction between Smy1p and F-actin is reduced in vitro near physiological ionic strength? When the assay was altered to increase the number of Smy1p molecules interacting with a single actin filament (Fig. 2a), the interaction became less ionic-strength dependent and persisted at salt concentrations as high as 0.5 M K-Acetate (Fig. 2b). This assay more closely resembles cellular conditions in that the 50–100 nm diameter secretory vesicles [6, 16] potentially allow many Smy1p molecules to interact with actin cables. Moreover, macromolecular crowding and confinement effects in the cytoplasm can greatly increase association rates and equilibrium constants .
The non-processive nature of Myo2p was confirmed by the lack of processive runs on actin bundles when a single Myo2p motor was attached to a Qdot (Movie S3). To ensure at most one motor per Qdot, Myo2p was mixed with Qdots at a ratio of 1:10, such that ~10% of the Qdots have a single motor attached, ~90% have no motor, and <1% have 2 or more Myo2p motors attached .
Our key finding is that Smy1p and actin interact strongly enough to enhance the motility of a single Myo2p molecule. Excess Smy1p was added to Qdots bound to a single Myo2p motor, so that ~10% of the Qdots have one Myo2p and multiple Smy1p molecules attached, and ~90% of the Qdots have multiple Smy1p but no Myo2p (see Methods). When this mixture was added to actin bundles, long, unidirectional, processive runs were observed (Fig. 3, Movie S4). The average speed of these runs (1.4 ± 1.0 µm/s) is similar to the ensemble motility of Myo2p (Fig. S2) . Fig. 3c shows a histogram of all runs that display unidirectional movement >1 µm. Of these runs, 49% moved to the end of the actin bundle before dissociating, thus the histogram underestimates the actual run lengths. Myo2p by itself is non-processive, but adding Smy1p to the Qdot creates a processive complex. 7% of the Qdots which bound to actin moved processively, consistent with the calculated percentage of Qdots with a single Myo2p attached (~10%). The remaining actin-bound Qdots were either stationary (39%) or underwent short-range (<1 µm) diffusion (54%).
Our interpretation of these data is that Smy1p acts as an electrostatic tether. After Myo2p undergoes a powerstroke and dissociates from actin, the electrostatic interaction between Smy1p and actin tethers the Qdot “cargo” to the actin bundle until Myo2p can re-bind and undergo another power stroke. In this way, cargo containing the non-processive Myo2p and Smy1p can take many steps along the actin bundle. This idea does not require a direct interaction between Myo2p and Smy1p, but does necessitate that the motor and the tether are both bound to the same cargo. In further support of the idea that Smy1p can act at a distance as a tether, a different non-processive motor (a single-headed mouse myosin Va) was also able to move a Smy1p-coated Qdot processively (Fig. S3). Smy1p can exert its effect in the absence of a direct interaction with Myo2p.
For Smy1p to act as a tether in vivo, it must bind to Myo2p or be bound to cargo transported by Myo2p. Evidence for a direct interaction between Smy1p and Myo2p is inconclusive. An interaction between Myo2p and Smy1p was inferred from yeast two-hybrid analysis, but could not be confirmed by co-immunoprecipitation or by co-affinity purification . We have likewise been unable to demonstrate a direct interaction (unpublished data and Fig. S2). These results suggest that Smy1p exerts its effect by being present on the same cargo.
It is well-established that Myo2p moves secretory vesicles [5, 6]. Myo2p and Smy1p both localize to regions of polarized growth, but the localization of both proteins is disrupted in mutant strains in which the late stages of secretion are blocked , implying that Smy1p resides on secretory vesicles. Here we show that Smy1p is present on secretory vesicles in living cells. Smy1p-mCherry and Sec4p-GFP, a marker for secretory vesicles , co-localize to the bud tip in small-budded cells and to the mother-bud neck in large-budded cells (Fig. S4), a distribution consistent with previous reports [7, 20]. By focusing above or below the focal plane containing the bud tip or neck localization and using deconvolution software to remove out-of-focus light, small particles representing either single secretory vesicles or small clusters of vesicles  were resolved (Fig. 4a). Most particles moved, but some were stationary. Smy1p-mCherry and Sec4-GFP co-localize to these stationary particles. We could not determine if both proteins co-localize on the moving particles due to their high speed of movement.
Moving secretory vesicles were compared in strains containing either Smy1p-GFP or Sec4p-GFP. The distribution of Smy1p-GFP and Sec4p-GFP was identical, consistent with the dual-labeled strain. Moving particles were observed in both strains (Fig. 4b–c, Movie S5–Movie S6). In both cases, particles moved across the mother cell to the mother-bud neck, and then to the bud tip, similar to previously reported results for Sec4p-GFP . The average particle movement speed within the mother cell was 1.60 ± 0.45 µm/s (N = 38, mean ± S.D.) for Sec4-GFP, and 1.59 ± 0.41 µm/s (N = 28) for Smy1p-GFP, values identical within experimental error (p = 0.87). These results place Smy1p on secretory vesicles, in close physical proximity to Myo2p and the actin cable track along which it moves, consistent with a role as an electrostatic tether.
Our results likely explain the mechanism by which overexpression of Smy1p compensates for the defective transport by a mutant Myo2p in the myo2-66 strain . Does Smy1p have other roles in the cell? Smy1p diffuses on microtubules in vitro, but it is not known if this is biologically relevant. In cases where kinesin shows diffusive behavior on microtubules, the advantages of this behavior relate directly to the motor’s function. The depolymerizing kinesin-13 MCAK uses a one-dimensional diffusive search to rapidly target microtubule ends where it binds and performs its cellular function of microtubule depolymerization . Smy1p differs from other diffusive kinesins in that it lacks motor activity, and it is possible that its sole function is to tether secretory vesicles to actin cables.
In higher eukaryotes, both kinesin and myosin V must be present on the same cargo for transport to the cell periphery . Conventional kinesin-1 enhanced the processive run length of vertebrate myosin Va when both proteins were present on the same cargo in vitro , but the cellular relevance of these observations was questioned because these events became less pronounced near physiologic ionic strength. Here we showed that when many Smy1p tethers can interact with actin, the interaction with actin persists until well beyond physiological ionic strength. Moreover, the observation that overexpression of the kinesin-family protein Smy1p compensates for a defective myosin V in living yeast suggests that these interactions are significant in vivo.
Myo2p is one of several non-processive myosin V motors that have been characterized, including human myosin Vc, Drosophila myosin V, and yeast Myo4p [8, 23–25]. One strategy a non-processive motor can use to transport cargo continuously is to have multiple motors engaged with the track at all times. Alternatively, a passive electrostatic tether can be used to create a processive complex, which has the advantage of not consuming ATP. The use of electrostatic tethers to enhance transport is widespread . Examples exist where the tether is a binding-partner, a domain within the motor, or even other motors. Dynactin increases the processivity of dynein [27–29]. An ATP-independent microtubule binding site in the tail of Drosophila Ncd (kinesin-14A) serves as an electrostatic tether that allows continuous movement along microtubules, with a unidirectional and a diffusive component. Processive motion is enhanced on bundled microtubules, similar to the observation here with actin bundles . A large insertion in loop 2 of the single-headed myosin IXb is thought to function as an actin-based tether, allowing the motor to move processively [31, 32]. We propose that the use of electrostatic tethers to enhance processivity may be a common mechanism to ensure robust transport under cellular conditions for both actin and microtubule-based motors, from yeast to mammals.
Actin bundles were formed by mixing 4 µM rhodamine-labeled actin filaments with 4 µM fascin in 25 mM imidazole (pH 7.4) plus 25 mM KCl, and incubating on ice for at least 24 hours. Smy1p was clarified for 20 minutes at 400,000 × g, then mixed with streptavidin-coated 655 nm Qdots (Invitrogen) at a ratio of 20 Smy1p per Qdot, and incubated on ice for at least 15 minutes. Based on biotin binding assays and the ~20 nm size of the Qdot, we estimate an average of 4–6 Smy1p molecules bind per Qdot. Flow cells were prepared by introducing the following solutions into the flow cell: 0.3 mg/ml N-ethylmaleimide-modified (NEM) skeletal muscle myosin (5 minute incubation), 5X rinse of 1 mg/ml BSA (2 minutes), microtubules or actin bundles (2–5 minutes), 5X rinse of Motility Buffer [25 mM imidazole, pH 7.4, 4 mM MgCl2, 1 mM EGTA, 50 mM DTT, 1 mg/mL BSA, and an oxygen scavenging system with varying salt and nucleotide concentrations]. Finally, the Smy1p/Qdot mixture was diluted to 0.05 – 0.25 nM in Motility Buffer and added to the flow cell. TIRF microscopy was performed at 24°C. N-ethylmaleimide-modified skeletal muscle myosin forms a strong and ATP-insensitive bond with actin, is commonly used to attach actin filaments to a coverslip , and also bound microtubules reasonably well. As a control, NEM myosin did not bind either Myo2p- or Smy1p-coated Qdots.
Myo2p-ΔGT (0.2 µM) was mixed with 0.4 µM F-actin and 2 mM MgATP in 300 mM KCl, and centrifuged for 20 minutes at 400,000 × g to remove any myosin that was unable to dissociate from actin in the presence of ATP. The supernatant was mixed with streptavidin-coated Qdots at a ratio of one Myo2p per ten Qdots. Flow cells were prepared as described above. The Myo2p-Qdot mixture was diluted to 0.025 – 0.5 nM in Motility buffer with 25 mM KCl, 2 mM MgATP, 6 µM Mlc1p, and 6 µM yeast calmodulin, and added to the flow cell.
To determine the effect of Smy1p on Myo2p, a similar procedure was employed. Myo2p-ΔGT was mixed with Qdots in a 1–10 ratio and incubated on ice for 15 minutes. Smy1p was clarified and then added at a ratio of 20 Smy1p per Qdot.
Cells were mounted directly onto a glass slide in Complete Synthetic Medium and sealed with Valap. Particles were tracked by hand using ImageJ. Speeds were measured as displacement of the particle image over time. Particles tended to slow down near the mother-bud neck, so those trajectories were not included in the analysis. Speeds are probably underestimated because we did not take into account movement perpendicular to the focal plane . To test if speeds of Sec4p-GFP and Smy1p-GFP particles were equal, a Student’s T-Test (two-tailed, two-sample unequal variance) was performed in Microsoft Excel.
Details on constructs, protein expression and purification, microscopes and experimental setups, data analysis, particle tracking, nucleotide hydrolysis, and yeast strains can be found in Supplemental Data.
Movie S1 Smy1p diffuses on microtubules. A Smy1p-coated Qdot (red) undergoes a 1-dimensional random walk on a microtubule (green). Field of view is 7.5 × 7.1 µm. Data collected at 12 frames per second and played back at 24 frames per second.
Movie S2 Smy1p-coated Qdots bind to and undergo short-range diffusion on actin bundles. Field of view is 9.1 × 6.5 µm. Data collected at 30 frames per second and played back at 15 frames per second.
Movie S3 Single Myo2p motors are not processive. Streptavidin Qdots bound to at most one Myo2p motor were added to actin bundles. No processive runs were observed. Field of view is 11.3 × 8.4 µm. Data collected at 12 frames per second and played back at 24 frames per second.
Movie S4 A single Myo2p can transport a Qdot long distances when Smy1p is also present on the same Qdot “cargo”. Streptavidin Qdots bound to at most one Myo2p motor, along with multiple Smy1p molecules, were added to actin bundles. Long processive runs, attributed to Qdots with one Myo2p and multiple Smy1p molecules, were observed. More frequently, Qdots bound to actin bundles were stationary or underwent local diffusion. These were attributed to Qdots with multiple Smy1p molecules bound but no Myo2p. Field of view is 15.0 × 3.3 µm. Data collected and played back at 30 frames per second.
Movie S5 Moving Smy1p-GFP particles. Field of view is 7.2 × 8.8 µm. Data collected and played back at 20 frames per second.
Movie S6 Moving Sec4p-GFP particles. Field of view is 10.3 × 7.2 µm. Data collected and played back at 10 frames per second.
The authors thank Matt Lord for providing yeast strains, technical advice, and use of equipment. We thank Jackie Vogel for providing strains and technical advice, Anthony Bretscher for providing strains and plasmids, Raviteja Devalla for imaging yeast cells, Brittany Weldon for ensemble motility experiments, Susan Lowey for critical reading of the manuscript, and Todd Clason, Guy Kennedy, and David Warshaw for technical advice and use of equipment. Microscopy was supported by NIH Grant Number 2 P20 RR016435-06 from the COBRE Program of the National Center for Research Resources. This work was supported by funds from the National Institutes of Health (GM078097 to KMT).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.