We have shown that expression of low levels of exogenous myosin-Vb (25–40% of endogenous levels) does not alter the trafficking of transferrin [20]. However, the dynamic tethering hypothesis predicts that exceeding endogenous levels with wild-type exogenous myosin-Vb will alter the balance of forces, reducing the extent and/or rate of retrograde movement from peripheral to perinuclear compartments. To test this prediction, we increased the amount of myosin-Vb associated with those compartments by transiently transfecting HeLa cells with a full-length, wild-type myosin-Vb construct. To allow imaging of live cells, we used a construct with an N-terminal eGFP tag [16]. We compared the distribution of eGFP-tagged myosin-Vb with that of our C-terminal -tagged (V5 and 6x-His) version [20], and observed no significant differences (data not shown). At low levels of eGFP-myosin-Vb expression, we observed only occasional, highly dynamic, colocalization of myosin-Vb and transferrin (arrows, Fig. 1D,E,F; Additional file 1).
Figure 2 and Additional files 2, 3, 4, 5, 6, 7, 8 show transferrin accumulation in peripheral compartments as a function of the overexpression level of eGFP-myosin-Vb, which the dynamic tethering hypothesis predicts will cause the coalescence and caging of peripheral endosomes by actin (Fig. 2A). A coalescence of actin around the enlarged peripheral endosomes is shown by the colocalization of myosin-Vb and actin (Fig. 2B,C).
HeLa cells that endocytosed fluorescent transferrin before and during overexpression of eGFP-myosin-Vb sequestered transferrin in large peripheral compartments decorated with myosin-Vb (Fig. 2D,E,F; Additional file 2), suggesting that fission of the compartments in which myosin-Vb and transferrin normally transiently colocalize (arrow, Fig. 1D,E,F; Additional file 1) was inhibited. By contrast, when transferrin was introduced after overexpression of myosin-Vb, transferrin was not colocalized with myosin-Vb in the enlarged peripheral compartments (arrows, Fig. 2F–K; Additional files 3, 4, 5). In addition, transferrin failed to accumulate in perinuclear compartments. As a negative control, we expressed a truncated myosin-Vb consisting of the head domain and first IQ domain, which had no effect on transferrin localization (data not shown). These data suggest that that overexpression of myosin-Vb prevents transferrin from both entering into and exiting from a normally dynamic, short-lived endocytic compartment.
In isolation, the static images shown in Figure 2 can be fit to the anterograde transport model if overexpression caused rapid transport of transferrin from perinuclear compartments while delaying its passage through cortical actin. However, Additional files 3, 4, 5 show that transferrin is not reaching perinuclear compartments.
As these data suggest that fission of vesicles from peripheral endocytic compartments and/or their transport to perinuclear compartments had been prevented by increased tethering to cortical actin, we examined the distribution of the endocytic markers Rab11a, Rab4, and Rab5. Cotransfections with eGFP-myosin-Vb and the recycling endosome marker mRFP-Rab11a showed virtually complete colocalization at high levels of myosin-Vb expression (Additional file 6). By contrast, little colocalization was observed in cells cotransfected with eGFP-myosin-Vb and the early endosome markers mRFP-Rab4 (Additional file 7) and mRFP-Rab5 (Additional file 8), suggesting that trafficking through early endosomes was not prevented. The videos also show that the enlarged endosomes are relatively static, consistent with increased tethering forces and caging by actin.
In a previous study, we used a chemical-genetic approach to show that induction of tight binding of sensitized myosin-Vb to actin, before addition of transferrin, prevented transferrin from accumulating in perinuclear compartments [20]. Our hypothesis is diagrammed in Fig. 3A, and the effect of inhibition before transferrin uptake, demonstrated previously, is shown in Fig. 3B. If myosin-Vb is required for transport from perinuclear compartments to the plasma membrane, then inducing tight binding of myosin-Vb to actin after transferrin loading should increase transferrin accumulation in perinuclear compartments, just as myosin-Vb tail overexpression does. We therefore transfected HeLa cells with Y119G sensitized mutant (Fig. 3) and wild-type control (not shown) myosin-Vb, loaded them with fluorescent transferrin, and microinjected the specific inhibitor of Y119G myosin-Vb, N6-(2-phenylethyl)-ADP (PE-ADP) [20]. Only cells with a punctate eGFP localization, representing lower expression levels, were chosen for microinjection. When PE-ADP was injected 10 min (data not shown) and 30 min (Fig. 3D,E,F) following the addition of transferrin, we still observed a decrease in fluorescence intensity in the perinuclear region of the transfected and injected cells (Fig. 3D,E,F) as well as rapid movement of transferrin when it did not colocalize with myosin-Vb (Additional file 9). These data, as well as the limited colocalization between transferrin and myosin-Vb, indicate that myosin-Vb activity is not required to transport transferrin from perinuclear compartments to the plasma membrane. These data are much more consistent with the peripheral tethering hypothesis, because the peripheral site of myosin-Vb function has been bypassed by loading with transferrin before induction of tight binding of myosin-Vb to actin.
While the inhibition of the Y119G sensitized mutant myosin-Vb in preloaded cells did not cause transferrin accumulation in perinuclear compartments, the data were not as simple as they were predicted to be by the dynamic tethering hypothesis, as myosin-Vb inhibition retarded the depletion of transferrin from perinuclear compartments relative to control cells (Fig. 3G,H). Upon closer examination, our induction of binding of myosin-Vb to actin had the general effect of halting nearly all motion of myosin-Vb-decorated structures within the cell (Fig. 4A). The motility of eGFP-myosin-Vb before and after microinjection was analyzed using kymographs (Fig. 4B,C for the cells shown in Fig. 4A, Fig. 4D,E for additional negative control cells; also see Additional files 10, 11, 12). Binned measurements of instantaneous particle speeds in the presence and absence of PE-ADP (Fig. 4F, Additional files 10 and 11) show that not only was slower actin-based motility (0.15 – 0.3 μm/s) inhibited, but higher-speed movements of myosin-Vb-decorated particles caused by microtubule-based motors (> 0.7 μm/s) were halted as well. No such inhibition was observed under control conditions, which included cells expressing Y119G myosin-Vb after injection of vehicle plus fluorescent Dextran without PE-ADP (data not shown), as well as cells expressing wild-type myosin-Vb after PE-ADP injection (Fig. 4G, Additional file 12).
Additional file 10: Chemical-genetic inhibition of sensitized mutant (Y119G) eGFP-myosin-Vb by PE-ADP microinjection (cell on left) halts movement of all myosin-Vb-decorated particles, including those being transported via microtubules; same field as Figure 3A. Uninjected control cell is on the right. Frame acquisition rate, 1/sec; frame display rate, 10/sec. (MOV 8 MB)
The arrest of microtubule-based motility of myosin-Vb-decorated particles was unexpected, and we initially suspected that it might have been an artifact of high effective ADP concentration in the form of the microinjected PE-ADP analog. To test the hypothesis that myosin-Vb interacts transiently with actin filaments during microtubule-based transport under normal conditions, we measured the speeds of particles decorated with wild-type eGFP-tagged myosin-Vb before and after the addition of latrunculin A. If myosin-Vb (or other myosins) normally interacts with actin filaments, latrunculin A treatment should increase both mean speed and the proportion of vesicles moving at 0.7–1.0 μm/sec. This prediction was confirmed, as latrunculin treatment nearly doubled the proportion of particles exhibiting rapid movement (Figure 4H), in contrast with results from melanosome transport in fish melanophores [34]. The modification of the dynamic tethering hypothesis to account for these data is diagrammed in Fig. 4I and 4J.
The dynamic tethering hypothesis further predicts that some markers found in peripheral endocytic compartments are likely to be shifted to a more perinuclear distribution by myosin-Vb tail overexpression (Fig. 5A). We tested this prediction for early endosomal antigen-1 (EEA1), which had a dispersed pattern in control cells (Fig. 5B,C,D, arrowhead), while in cells expressing the eGFP/myosin-Vb tail chimera [16], EEA1 was much more concentrated, in an asymmetric pattern primarily on one side of the nucleus (Fig. 5B,C,D, arrows).
Based on the change in distribution of EEA1 coupled with its failure to colocalize with the myosin-Vb tail, we hypothesize that in the presence of the tail, endosomes still are transported to more perinuclear regions of the cytoplasm, but the fission between their domains that normally occurs in peripheral regions occurs in a more perinuclear location. We then confirmed the effect of the myosin-Vb tail on Rab11a redistribution. As observed by Lapierre et al., the dispersed pattern observed in untransfected control cells (Fig. 5E,F,G, arrowheads) was changed to a more perinuclear pattern by overexpression of the eGFP/myosin-Vb tail (arrows).
We next examined Rab8a, which has been shown to interact in vitro with myosin-Vb [19]. We observed a nearly normal distribution of Rab8a despite the overexpression of the myosin-Vb tail (Fig. 5H,I,J). These results are consistent with the differences between Rab11a and Rab8a compartments and pathways observed by Roland et al., and indicate that the affinity of the myosin-Vb tail domain for Rab8a is much lower than its affinity for Rab11a. This result also is consistent with their inability to observe interaction between myosin-Vb and Rab8a in a cellular context.
Since mosaic endosomes have been observed with every possible combination of Rab4, Rab5, and Rab11a [35], we examined Rab4 and Rab5 distribution. Overexpression of the myosin-Vb tail produced a slight alteration in the distribution of Rab4 (Fig. 6A,B,C,D), but no significant effect on Rab5 distribution (Fig. 6F,G,H), which is puzzling given the association between EEA1 and Rab5 [36].
To summarize our model, myosin-Vb is associated with multiple compartments, of which only some are involved in transferrin trafficking. Myosin-Vb primarily tethers a subset of peripheral, Rab11a-positive endocytic compartments to cortical actin, opposing forces from dynein or minus-end-directed kinesins and retaining the compartment in the actin-rich periphery. This is analogous to the mechanism of Velcro™, except that instead of hooks bending, the myosin-Vb heads are going through the ATPase cycle and periodically releasing from actin. In this analogy, overexpression of full-length, wild-type myosin-Vb (Fig. 2) causes greater retention of normally endocytic compartments in the periphery, leading to their coalescence, because the increase in number of myosins outweighs their individual cycling off and back onto actin. These data strongly suggest that while this compartment is normally rare and transient (Fig. 1), all transferrin still must pass through it to reach perinuclear compartments. Both our overexpression of myosin-Vb and overexpression of the myosin-Vb tail create artifacts. In the former case (Fig. 2), caging by actin causes coalescence and blockage of both entry and exit; while in the latter case [16], release from actin causes what we believe to be virtually the same compartment to collapse to a perinuclear location. Chemical-genetic inhibition is analogous to preventing individual Velcro™ hooks from bending, but does not prevent entry and exit into this compartment via the peripheral pathway [20].
Myosin-Va, the founding member of this myosin family, appears to have a similar function, albeit involving different compartments. In melanocytes, we first suggested a peripheral tethering function for myosin-Va based on the mutant phenotype, a perinuclear accumulation of melanosomes [37]. The best-characterized system for melanosome transport has been Xenopus melanophores [4, 5, 38, 39], with similar, but less dynamic, results from murine melanocytes [40]. In both cases, myosin-Va function was hypothesized to provide not only peripheral capture, but transport within the periphery as well. While pauses in microtubule-based movement attributed to myosin-Va have been observed [41, 42], these studies represent the first such observation for myosin-Vb. In general, myosin-Vb appears to perform the same function in the endocytic pathway as myosin-Va performs in exocytic pathways, and our future experiments will test the validity of our generalization. In a technical context, our results suggest that membrane compartments cannot necessarily be reliably identified by their locations within the cytoplasm in cells in which trafficking has been grossly perturbed by manipulation, particularly overexpression, of any relevant component.