A novel labeling strategy reveals that myosin Va and myosin Vb bind the same dendritically polarized vesicle population

Abstract

Neurons are specialized cells with a polarized geometry and several distinct sub-domains that require specific complements of proteins. Delivery of transmembrane proteins requires vesicle transport, which is mediated by molecular motor proteins. The myosin V family of motor proteins mediates transport to the barbed end of actin filaments, and little is known about the vesicles bound by myosin V in neurons. We developed a novel strategy to visualize myosin V-labeled vesicles in cultured hippocampal neurons and systematically characterized the vesicle populations labeled by myosin Va and Vb. We find that both myosins bind vesicles that are polarized to the somatodendritic domain where they undergo bidirectional long-range transport. A series of two-color imaging experiments showed that myosin V specifically colocalized with two different vesicle populations: vesicles labeled with the transferrin receptor and vesicles labeled by low-density lipoprotein receptor. Finally, coexpression with Kinesin-3 family members found that myosin V binds vesicles concurrently with KIF13A or KIF13B, supporting the hypothesis that coregulation of kinesins and myosin V on vesicles is likely to play an important role in neuronal vesicle transport. We anticipate that this new assay will be applicable in a broad range of cell types to determine the function of myosin V motor proteins.

1 |. INTRODUCTION

Members of the Myosin V family are the only dimeric actin-based motor proteins that mediate processive movement to the barbed end of actin filaments.^1,2 Myosin V functions as a homodimer with two domains, an N-terminal motor domain that generates locomotive force along actin filaments and a C-terminal tail domain that interacts with vesicles.³ Three myosin V proteins are encoded in the mammalian genome, two of which—myosin Va and myosin Vb—are expressed in the nervous system.^4–7 Myosin V mediates short-range transport and fine-tunes the localization of vesicles. Neuronal myosin Vs interact with a variety of cargoes and in a range of cellular pathways.^8,9 For example, myosin V mediates neurotransmission by tethering synaptic vesicles in the active zone of presynaptic sites^10,11 and maintains postsynaptic sites.^12–14 Myosin Va is also responsible for localizing ER into spines to facilitate synaptic plasticity¹⁵ and maintaining the spine apparatus through interactions with synaptopodin.¹⁶ One important function of the myosin V family is to bring receptors to the plasma membrane at local dendritic sites, including spines.^5,17,18

In addition to these roles, myosin Va has also been proposed to function in maintaining the polarized transport of dendrite-selective vesicles.^19,20 Dendrite-selective vesicles move in the somatodendritic domain but do not enter the axon beyond the axon initial segment.^21,22 One leading model posits that myosin Va specifically binds dendrite-selective vesicles, but not axonal vesicles.¹⁹ In this model, myosin Va acts in the proximal axon by arresting vesicle transport in an actin-dense region, allowing other motors such as dynein to retrieve vesicles and return them to the soma. Long-term expression of dominant-negative myosin Va—but not myosin Vb—caused the mislocalization of transferrin receptor (TfR), the metabotropic glutamate receptor mGluR1, and the potassium channel KV4.2 into the axon of cultured cortical neurons.¹⁹ Most studies addressing the role of myosin V in the transport of neuronal vesicles relied on analysis of the steady-state localization of cargo molecules following long-term expression of dominant-negative myosin V constructs or RNAi. While often useful, the inherent connectivity of the endomembrane system makes this approach susceptible to indirect effects. Disruptions in one transport step could impact the transport machinery that is involved in other downstream transport steps. There are no systematic studies analyzing the transport behavior of vesicle-bound myosin V in neurons, probably because imaging vesicle-bound myosin V is challenging. While the expression of fluorescently tagged myosin V sometimes allows visualization of vesicular structures, this approach generally yields inconsistent labeling; in most cells myosin V expression results in diffuse cytoplasmic fluorescence that obscures any vesicle-bound pool.

Here we describe a novel strategy that consistently produces myosin V vesicle labeling by expressing only the vesicle-binding tail region of myosin V. We used this labeling strategy to systematically analyze myosin V-labeled vesicles in cultured hippocampal neurons. We found that myosin Va and Vb both bind the same dendrite-selective vesicles. A series of two-color imaging experiments showed that these myosin V-labeled vesicles specifically colocalize with two different dendritically polarized vesicle populations: vesicles labeled with transferrin receptor and, to a lesser degree, vesicles labeled by low-density lipoprotein receptor (LDLR). Finally, myosin V cotransported with the Kinesin-3 family members KIF13A and KIF13B in dendrites, which supports the hypothesis that coregulation of kinesins and myosin V on vesicles is likely to play an important role in neuronal vesicle transport.

2 |. RESULTS

2.1 |. Visualizing vesicle-bound myosin Vs with fluorescently tagged myosin V tails

An intuitive approach to characterize the vesicle populations myosin Vs interact with is to express the full-length motor protein with a fluorescent tag. While this approach has been used on occasion,^15,17 in practice, it yields inconsistent results, and only a small fraction of expressing cells produce vesicle labeling. To illustrate this, we expressed full-length GFP-tagged myosin Va in cultured hippocampal neurons (Figure 1A). Transfected cells exhibited minimal vesicle labeling. Any vesicle-bound myosin V was not visible because of the bright cytoplasmic signal—the construct behaved much like a soluble fill. High magnification images show little fluorescence in the axon and diffuse fluorescence in dendrites. Kymographs from the axon and a dendrite reveal a near absence of transport events. On a kymograph, anterograde transport events are indicated by lines with positive slopes and retrograde transport events by lines with negative slopes. The few visible transport events in dendrites are short and barely above the background. This approach lacks the consistency needed to systematically evaluate the localization and behavior of myosin V-labeled vesicles.

FIGURE 1.

Myosin V tail labeling strategy enhances vesicle labeling. Representative images of 6 to 8 DIV hippocampal neurons expressing full-length myosin Va-GFP (A), GFP-myosin Va tail (B), or GFP-myosin Vb tail (C), for 5 to 8 hours at comparable levels. Yellow lines indicate the axons and dendrites from which high magnification views and kymographs were generated. For clarity, transport events from kymographs were redrawn as white lines. Kymograph lines with a positive slope indicate anterograde transport and lines with a negative slope indicate retrograde transport. This convention is followed for all figures. Full-length myosin Va-GFP labeled very few vesicles. Both tail constructs resulted in the visualization of vesicle-bound myosins. Labeled vesicles were polarized to the dendrites, where they underwent bidirectional long-range transport. The graphs show the quantification of run-length (D), velocity (E), and the number of transport events per 100 µm (F) for dendritic vesicles labeled with myosin Va or myosin Vb tails. There were no statistically significant differences between the measured transport parameters of myosin Va and Vb tail-labeled vesicles (P > 0.1). Error bars indicate the SEM. Myosin Va: 14 cells, 531 events; myosin Vb: 20 cells, 594 events. Scale bars: 10 µm

In an attempt to improve vesicle labeling, we decided to test an approach we previously developed to visualize vesicle-bound kinesins.²³ Similar to kinesins, most cytoplasmic myosin V molecules are in an autoinhibited conformation in which the globular tail binds to the motor domain.^24–26 In this autoinhibited state, myosin V dimers no longer bind to vesicular cargoes.²⁷ When imaging full-length fluorescently tagged myosin Vs, the bright diffuse signal from cytoplasmic myosin V molecules obscures any labeled vesicles. For kinesins, expressing only the vesicle-binding domain rather than the full-length protein dramatically improves vesicle labeling in neurons.²³ We reasoned that because myosin V is structurally similar to kinesin, a comparable approach may enhance myosin V vesicle labeling. To test this, we expressed a truncated myosin Va tail domain tagged with an N-terminal fluorophore (Figure 1B and Video S1). Expressing the tail domain dramatically improved vesicle labeling. Myosin Va-labeled vesicles were confined to the soma and dendrites, with little labeling in the axon beyond the axon initial segment. Kymographs show that vesicles underwent long-range bidirectional transport in dendrites. We employed the same strategy for the other neuronal myosin V family member, myosin Vb (Figure 1C). Myosin Vb tail expression also resulted in improved vesicle labeling, although the cytoplasmic background was higher than for myosin Va. This may indicate that there are fewer overall binding sites for myosin Vb on vesicles than for myosin Va. Another possibility is that the difference in labeling between myosin Va and Vb is because of the differences in construct design. The myosin Vb tail construct we used contains less of the coiled-coil domain which may make it less likely to dimerize. If dimerization is required for vesicle binding, the shorter construct length of the myosin Vb tail could explain its less efficient vesicle labeling. Despite the higher background, myosin Vb tail clearly labeled vesicles. These vesicles were distributed throughout the somatodendritic region and excluded from the axon. Kymographs show that myosin Vb-labeled vesicles underwent long-range bidirectional transport in dendrites. Together, these results show that the expression of the myosin V tail domain significantly improves vesicle labeling.

Consistent vesicle labeling in combination with live-cell imaging allowed us to measure the transport parameters of myosin V-labeled vesicles. Vesicle transport parameters, such as run-length and velocity, differ between different vesicle populations, giving most vesicles a specific transport “signature.”^28–31 Long-range transport (excursion lengths greater than ~3–5 µm) is microtubule-based and mediated by kinesins and dynein.^32,33 Correspondingly, vesicles labeled by different kinesins also exhibit signature transport parameters.²³ Therefore, even though myosin V itself does not mediate long-range transport, analysis of transport parameters provides insight into the identity of the kinesin motors that mediate its movement. We analyzed the transport parameters of dendritic vesicles labeled with myosin Va or Vb to determine if they bound vesicles with different transport behaviors (Figure 1D,E). Kymograph analysis revealed that myosin Va-labeled vesicles exhibited an average run length of 6.3 ± 0.3 µm and an average velocity of 2.4 ± 0.07 µm/s. Myosin Vb-labeled vesicles exhibited an average run length of 6.8 ± 0.2 µm and an average velocity of 2.5 ± 0.09 µm/s. Because about 50% of dendritic microtubules are oriented with the plus end toward the cell body and 50% away from the cell body,^34,35 we combined anterograde and retrograde transport events for analysis. There were no statistically significant differences between the vesicle populations labeled by the two myosins (run-length: P = 0.15; velocity: P = 0.82). The only observed difference was in the number of events, with myosin Vb labeling more vesicles than myosin Va (Figure 1F). This is the first analysis of myosin V-labeled vesicle behavior in neurons. The transport parameters we obtained are similar to those of vesicles that contain dendritic receptors.²⁸

Because the transport parameters of myosin Va- and myosin Vb-labeled vesicles were indistinguishable and both myosins labeled vesicles that were restricted to the somatodendritic domain, we used two-color live-imaging to determine whether the two myosins bound two different vesicle populations with similar transport parameters, or the same vesicle population (Figure 2A and Video S2). High magnification images show that most vesicles were labeled by both myosin Va and Vb. High colocalization was corroborated by a Pearson correlation coefficient of 0.54 in dendrites (n = 23 cells, SD = 0.11). This is substantially higher than the Pearson correlation coefficient for labels that do not colocalize, which is typically around 0.3 because of the correlation of low-intensity areas in which both fluorescent proteins are absent. Overlap of two proteins in stationary images can be coincidental, whereas spurious cotransport is extremely unlikely. Such cotransport of two proteins—simultaneous movement in the same direction and at the same velocity—is perhaps the strongest evidence that two proteins bind the same vesicles. Therefore, we used kymograph analysis to determine cotransport of myosin Va- and myosin Vb-labeled vesicles. We found that 45% of moving myosin Va-labeled vesicles cotransported with myosin Vb and 43% of moving myosin Vb-labeled vesicles cotransported with myosin Va (Figure 2B). These results show conclusively that myosin Va and Vb interact to a large degree with the same vesicle populations. While previous neuronal studies have focused on either myosin Va or Vb and attributed their findings to one of the two motors, this is the first analysis directly comparing the vesicle-bound populations of the neuronal myosin Vs. These findings raise the possibility that there may be more redundancy between myosin V isoforms than previously thought.

FIGURE 2.

Myosin Va and Vb tails label overlapping vesicle populations. A, A representative image of a 6 DIV hippocampal neuron expressing GFP-myosin Vb tail and HaloTag-myosin Va tail, visualized with JF549 dye. Yellow lines indicate the axon and dendrite from which high magnification views and kymographs were generated. Kymographs show that myosin Va and myosin Vb tails predominantly label the same vesicles. For clarity, transport events were redrawn (green: myosin Vb only, magenta: myosin Va only, white: cotransport). B, The table shows the quantification of cotransport of myosin Va and myosin Vb vesicles. n = 23 cells; myosin Va events: 938; myosin Vb events: 1058. Scale bar: 10 µm

2.2 |. Myosin V specifically binds dendrite-selective vesicle populations

To identify the cargo proteins in myosin V-bound vesicles, we coexpressed GFP-myosin Va tail with a series of Halo-tagged vesicle markers (Figure 3 and Video S3). Because myosin Va and Vb bind the same vesicles and myosin Va produced better labeling, we focused on myosin Va. First, we coexpressed GFP-tagged myosin Va tail with HaloTag-transferrin receptor (TfR) that was visualized with JF549.³⁶ TfR is a canonical marker for dendrite-selective vesicles^21,28,29 and both myosin V isoforms have been proposed to interact with TfR vesicles.^17,19,20 Both proteins labeled vesicles in the somatodendritic region and were excluded from the axon (Figure 3A). A high magnification image from a dendrite shows substantial colocalization between myosin Va and TfR. The Pearson correlation coefficient derived from dendrites corroborated this with a value of 0.49 (n = 21 cells, SD = 0.09). This was significantly different from all other cargoes (P < 0.01). Kymographs show that in dendrites many myosin V-labeled vesicles underwent cotransport with TfR. Kymograph analysis found that 41% of myosin Va-labeled vesicles cotransported with TfR (Figure 3E). These results show that myosin V interacts with dendrite-selective TfR vesicles. To confirm that the presence of myosin Va tail on moving TfR vesicles did not alter the transport behavior, we measured the velocities of moving TfR vesicles that were cotransporting with myosin Va tail and TfR vesicles without myosin Va (Figure 3F). As expected, both populations moved at comparable velocities. Vesicles labeled with TfR only moved at 1.9 ± 0.1 µm/s and vesicles labeled by TfR and myosin V tail at 1.8 ± 0.1 µm/s.

FIGURE 3.

Myosin Va specifically binds dendrite-selective vesicles. Representative images of 6 to 7 DIV hippocampal neurons expressing GFP-myosin Va tail with HaloTag-transferrin receptor (TfR) (A), HaloTag-low-density lipoprotein receptor (LDLR) (B), brain-derived neurotrophic factor (BDNF)-HaloTag (C), or neuron-glia cell adhesion molecule (NgCAM)-HaloTag (D). HaloTags were visualized with JF549 dye. Yellow lines indicate the axons and dendrites from which high magnification views and kymographs were generated. For clarity, transport events were redrawn (green: myosin Va only, magenta: cargo label only, white: cotransport). Myosin Va exhibited the most cotransport with TfR and less cotransport with LDLR, proteins that label two different dendrite-selective vesicle populations. Myosin Va did not overlap with BDNF or NgCAM. Graphs show quantification of overlap between myosin Va and four neuronal vesicle markers. E, Cotransport in dendrites was determined by kymograph analysis. F, Velocity measurements of TfR vesicles with and without bound myosin Va. There was no statistical difference between the two populations (P > 0.4). Error bars indicate the SEM. TfR: 21 cells, 1086 events; LDLR: 21 cells, 1539 events; BDNF: 21 cells, 3421 events; NgCAM: 21 cells, 970 events. Scale bars: 10 µm

We next asked whether myosin V interacts with a second dendrite-selective vesicle population, labeled with the low-density lipoprotein receptor (LDLR).^37,38 LDLR labels a population of vesicles that is different from TfR-labeled vesicles, but is also polarized to the somatodendritic region.³⁷ We coexpressed GFP-myosin Va and HaloTag-LDLR (Figure 3B). High magnification images of a dendrite show that some myosin Va vesicles colocalized with LDLR (Figure 3B), but substantially fewer than with TfR (Figure 3A). Quantitative analysis of colocalization (Pearson correlation coefficient = 0.39, n = 21 cells, SD = 0.13) and kymograph analysis of cotransport corroborate this observation (Figure 3E). This shows that myosin V binds to two distinct vesicle populations that are polarized to the somatodendritic domain.

To confirm the specificity of this interaction, we determined the colocalization and cotransport of myosin Va with two other important neuronal vesicle populations: those labeled with brain-derived neurotrophic factor (BDNF) or neuron-glia cell adhesion molecule (NgCAM).

BDNF labels secretory granules that move bidirectionally in the axon and dendrites.^39,40 NgCAM labels vesicles that move bidirectionally in dendrites but preferentially enter the axon where they undergo mostly anterograde transport.^28,41 Neither of these vesicle populations exhibited significant colocalization with myosin Va (Pearson correlation coefficients: BDNF = 0.32, n = 21 cells, SD = 0.09; NgCAM = 0.27, n = 21 cells, SD = 0.11). Both conditions were significantly different from LDLR (p < 0.05) but not from each other (P > 0.1). Neither BDNF nor NgCAM cotransported with myosin Va (Figure 3C–E). These data show that myosin V specifically interacts with dendrite-selective vesicles, primarily those labeled by TfR.

2.3 |. Myosin Va binds recycling endosomes

One myosin V isoform, myosin Vb, functions in the recycling of AMPA receptors and TfR at the dendritic plasma membrane.¹⁷ Because we found that myosin Va and Vb label the same vesicles, we asked whether myosin Va also binds to vesicles that mediate the endocytic recycling of transferrin. TfR resides in at least two different vesicle populations: Golgi-derived vesicles and vesicles in the endocytic/recycling pathway where resident TfR functions as an iron transporter and facilitates the endocytosis and release of transferrin. To determine whether myosin Va binds vesicles in the recycling pathway, we transfected neurons with Halo-myosin Va tail (visualized with JF646³⁶) and TfR-GFP and exposed these cells to fluorescent transferrin (Tf555) in the culture medium (Figure 4A). Fluorescent transferrin was endocytosed and labeled endocytic and recycling vesicles.¹⁷ As expected, all three labels were polarized to the somatodendritic region of the cell (Figure 4B). High magnification images show considerable overlap between TfR, transferrin, and myosin Va in dendrites. Kymograph analysis found that myosin Va cotransported with TfR and Tf (Figure 4C). This indicates that myosin Va binds the endocytic/recycling vesicles that are known to interact with myosin Vb, suggesting that myosin Va and Vb may both perform the same function of bringing recycling endosomes to the plasma membrane.

FIGURE 4.

Myosin Va interacts with recycling endosomes. A, Representative image of a 7 DIV hippocampal neuron that has endocytosed Tf555 and is expressing HaloTag-myosin Va tail and GFP-TfR. The HaloTag was visualized with JF646 dye. B, Yellow lines indicate the axon and dendrite from which high magnification views and kymographs were generated. For clarity, transport events present in all three channels were redrawn as white lines. C, There was no significant difference between myosin Va cotransport with TfR or Tf (P > 0.3). Error bars indicate the SEM. Twenty cells, myosin Va: 1426 events; TfR: 1180 events; Tf: 1239 events. Scale bar: 10 µm

2.4 |. Myosin V-labeled vesicles colocalize with kinesin-3 family members KIF13A and KIF13B

Myosin V-labeled vesicles undergo long-range transport, which is mediated by microtubule-based motors. We utilized the tail-labeling strategy to identify the kinesins that bind moving myosin Va-labeled vesicles.²³ We had previously found that dendrite-selective vesicles are moved by members of the Kinesin-3 family.^37,42 KIF13A may be a kinesin that is specialized for dendritic transport, as it binds exclusively to dendrite-selective vesicles in hippocampal neurons.²³ The closely related KIF13B also interacts with dendrite-selective TfR vesicles,³⁷ but also mediates the transport of axonal cargoes.^23,43,44 KIF1A, another Kinesin-3 family member, interacts with dendrite-selective vesicles that carry LDLR,³⁷ but most prominently mediates the transport of secretory granules^23,39 and presynaptic vesicles.^30,45,46 To determine which of these candidate kinesins move the motile fraction of myosin V-labeled vesicles, we coexpressed each with fluorescently tagged myosin Va tail.

Figure 5A shows an image of a neuron expressing GFP-myosin Va tail and HaloTag-KIF13A tail, visualized with JF549. Both labeled vesicles that were polarized to the somatodendritic region, with little labeling in the axon. High magnification images show that many of these vesicles overlapped (Figure 5A). Kymograph analysis of cotransport revealed that 27% of moving KIF13A-labeled vesicles cotransported with myosin Va (Figure 5D). This number likely under-estimates true cotransport, because KIF13A is a relatively poor vesicle labeler. We next coexpressed the GFP-myosin Va tail with HaloTag-KIF13B tail (Figure 5B,D). KIF13B labeled vesicles in the axon and dendrites. High magnification images and kymographs show that a few dendritic vesicles exhibited colabeling and cotransport (Figure 5B). Finally, we expressed HaloTag-myosin Va tail with KIF1A-GFP (Figure 5C). KIF1A labeled vesicles in axons and dendrites and exhibited no overlap with myosin Va (Figure 5C,D). These results show that two Kinesin-3 family members, KIF13A and KIF13B, cotransport with myosin Va, suggesting that coregulation of these motors is required for proper transport of these vesicles.

FIGURE 5.

Myosin Va binds to vesicles labeled by KIF13A and KIF13B. Representative images of 7 to 8 DIV hippocampal neurons expressing GFP-myosin Va tail and HaloTag-KIF13A tail (A), GFP-myosin Va tail and HaloTag-KIF13B tail (B), or HaloTag-myosin Va and KIF1A-GFP (C). HaloTag was visualized with JF549 dye. Yellow lines indicate the axons and dendrites from which high magnification views and kymographs were generated. For clarity, transport events were redrawn (green: GFP, magenta: HaloTag, white: cotransport). D, Myosin Va cotransported with KIF13A and KIF13B but not with KIF1A. Error bars indicate the SEM. KIF13A: 17 cells, myosin Va events: 1303, KIF13A events: 304; KIF13B: 23 cells, myosin Va events: 1133, KIF13B events: 1230; KIF1A: 21 cells, myosin Va events: 2146, KIF1A events: 965. Scale bars: 10 µm

3 |. DISCUSSION

Here we developed a novel tail labeling strategy to systematically analyze the neuronal localization of myosin V family members. With this approach, we quantitatively analyzed the transport parameters of vesicles bound by myosin Va and Vb. Surprisingly, both myosin Vs bound the same vesicle population. Myosin V-labeled vesicles were present throughout the somatodendritic domain, where a subset underwent bidirectional long-range transport, and was excluded from the distal axon. Two-color imaging with a series of canonical vesicle markers identified these as recycling endosomes. Colabeling with kinesins found that myosin V and kinesins concurrently bind stationary and moving vesicles. This indicates that actin- and microtubule-based motor activity is likely mediated by a form of on-vesicle regulation.⁴⁷ The tail labeling strategy described will be useful in future studies—including in nonneuronal cell types—to determine the function of myosin Vs and their cargo vesicles.

3.1 |. Visualizing vesicle-bound myosin V

Our results show that expression of fluorescently labeled tail constructs results in robust labeling of vesicle-bound myosin V. This strategy is a considerable improvement compared with overexpressing full-length myosin V. Expressing fluorescent myosin V tails is an unbiased approach and can in principle identify all vesicle populations that interact with a particular myosin V, as long as there are enough binding sites available. This labeling approach allowed us to quantify the localization and transport parameters of vesicles labeled by both neuronal myosin V family members. Two-color live-cell imaging experiments determined the vesicle population myosin V binds in hippocampal neurons. Consistent vesicle labeling with myosin V allowed us to systematically test several vesicle populations using the same culture conditions and imaging parameters. The results show that vesicle labeling is highly specific, enhancing confidence in the assay.

As with any assay, there are some limitations. Although tail labeling greatly enhances vesicle labeling, the imaging is still technically challenging. Even in neurons with successful labeling, the signal of labeled vesicles was still dim. Consequently, an absence of vesicle labeling needs to be interpreted with caution. On vesicles, myosin V family members are thought to act as ensembles of 10 to 20 molecules,^48–50 which is detectable by modern sCMOS or EMCCD cameras. However, vesicles with fewer copies of myosin V may exist and could be missed even when using tail labeling.

Another concern with these experiments is the possibility of a dominant-negative effect. The tail could saturate the binding sites on vesicles and prevent endogenous myosin V molecules from binding to vesicles. In fact, myosin V tail constructs have been used as dominant-negative constructs.^5,18,19,51 Those studies differ from our experiments in several important ways. First, in studies that sought to cause a dominant-negative effect, myosin tail constructs were expressed at high levels and often with expression times above 24 hours. In our experiments, we recorded as early as possible, typically 3 to 6 hours after transfection. Only cells with relatively low expression had successful tail labeling. Cells with higher expression levels—and thus more likely to exhibit dominant-negative effects—were excluded from our analysis simply because the fluorescent signal was largely soluble. Second, a series of publications from Arnold and colleagues found that expression of myosin Va dominant constructs caused TfR vesicles to enter the axon and lose their polarized distribution.^19,20 We did not see any change in the polarity or any other transport characteristic of TfR vesicles following low levels of expression of myosin Va tail in our experiments. Third, it is worth noting that most myosin V dominant-negative experiments use steady-state localization of marker proteins as an indirect readout of alterations of vesicle transport instead of imaging transport directly, as disruptions in one transport step could lead to changes in the transport machinery that is required for other downstream transport steps. We conclude that while vigilance is important, it is unlikely that dominant-negative effects are disrupting normal vesicle transport when imaging tail constructs.

Finally, our experiments cannot determine whether myosin V that is bound to a particular vesicle is active or not. A subset of labeled carriers undergoes long-range microtubule-based transport, suggesting that myosin V can bind vesicles in an inactive state.

3.2 |. The role of myosin V in neuronal vesicle transport

Many pathways for myosin V have been identified, but relatively little is known about the overall role of myosin V in neuronal vesicle transport. A surprising outcome from this work is that both myosin V isoforms bound the same vesicles. This observation appears to challenge the dogma that myosin Va and Vb are each specialized to function in different sets of neuronal pathways. The C-terminal domains, ~400 residues that mediate vesicle binding³ have nearly 70% sequence identity between Va and Vb in humans. While our data do not rule out specific specialized functions, most of both myosin V isoforms appear to bind the same carriers, suggesting that they may fulfill the same function on these vesicles.

Our results definitively show that both myosin V isoforms bind preferentially to vesicles labeled by transferrin receptor, in accordance with previous findings.^17,52,53 Myosin V binds to these vesicles and targets them to the plasma membrane in response to calcium release.¹⁷ Our data show that myosin V binds both stationary and moving vesicles. This is consistent with a model in which myosin V binds vesicles in an inactive state and that on-vesicle activation in response to an external signal causes engagement with actin and myosin V-mediated translocation to the plasma membrane. While calcium is one regulator of myosin V activity, other signals likely exist.

Finally, these experiments show directly that myosin V and kinesins simultaneously bind to vesicles. Understanding the coregulation of motor ensembles on vesicles is a long-standing challenge.^47,54,55 While much attention has been given to the coregulation of kinesin and dynein, myosin V must factor into the transport problem, especially in neurons. Previously gained information about myosin V and kinesin coregulation from other model systems, such as Xenopus laevis melanophores^56–58 and Drosophila melanogaster oocytes,^59–61 may be applicable to neuronal motor regulation and inform future experiments. At least in some instances, handoff between kinesins and myosin V can be regulated by the relative abundance of microtubule and actin substrates.^61,62 Artificial recruitment of constitutively active myosin V motor domains to moving vesicles rapidly arrest movement, indicating that activity of vesicle-bound motors must be finely tuned.^63,64 Future progress will require the development of systems in which endogenous motor regulation can be probed in living cells.

4 |. MATERIALS AND METHODS

4.1 |. Cell culture

Primary rat hippocampal neurons were cultured as described previously.^65–67 In brief, hippocampi were dissected from E18 rats, trypsinized, dissociated, and plated on poly-L-lysine-treated 18 mm glass coverslips. Cultures were grown in minimal essential medium with N2 supplements and maintained at 37°C in an incubator with a 5% CO2 atmosphere. DNA constructs were transfected into stage 4 hippocampal neurons (6–10 days in culture) with Lipofectamine 2000 (Thermo Fisher) and allowed to express for 3 to 18 hours before imaging.

4.2 |. DNA constructs

All constructs used for multi-color imaging had a beta-actin promoter. Details of constructs are described in Table 1. The original GFP-myosin V constructs were gifts from Don Arnold¹⁹ and Jose Esteban,¹⁸ respectively, and were subcloned into a vector containing a chicken beta-actin promoter.

TABLE 1.

Expression constructs

Construct	Construct design	Accession number	Source
GFP-myosin Va tail	GFP-GGGSS-myosin Va1006−1828	NM_022178.1	19
GFP-myosin Vb tail (CMV promoter)	GFP-SGLRSREAQ-myosin Vb1436−1846	NM_017083.1	18
GFP-myosin Vb tail (chicken beta actin promoter)	GFP-SGLRSREAQ-myosin Vb1436−1846	NM_017083.1	This paper
HaloTag-myosin Va tail	HaloTag-LYG-3myc-YKGGGSS-myosin Va1006−1828	NM_022178.1	This paper
TfR-HaloTag	TfR-GDPPVATMASLEPTTEDLYFQSDND-HaloTag	M11507	This paper
TfR-GFP	TfR-GDPPVAT-GFP	M11507	41
HaloTag-LDLR	HaloTag-GGGEF-LDLR	BC014514.1	This paper
BDNF-HaloTag	BDNF-GGGPRGGGSGGGSGGG-HaloTag	NM_001270630.1	This paper
NgCAM-HaloTag	NgCAM-KLGAPRPTMASLEPTTEDLYFQSND-HaloTag	Z75013	This paper
HaloTag-KIF13A tail	HaloTag-YSDLELKLRILQSTVPRAR-KIF13A361−1749	NM_010617.2	This paper
HaloTag-KIF13B tail	HaloTag-LYGGGSGGGSGGG-KIF13B442−1843	NM_001081177.1	This paper
KIF1A-GFP	KIF1A-GGGSGGGSGGPRT-GFP	NM_001294149.1	This paper

4.3 |. Imaging

For live imaging, cells were maintained in Hibernate E medium without phenol red (Brain-Bits). Images and movies were acquired using a microscope with an Andor Dragonfly built on a Ti2 (Nikon) with a CFI Apo 60× 1.49 objective (Nikon) and two sCMOS cameras (Zyla 4.2+, Andor). The imaging stage, microscope objectives, and cell samples were kept at 37°C in a warmed enclosure (full lexan incubation ensemble; OkoLab). During imaging, the z-axis movement was controlled by the Perfect Focus system on the Ti-E microscope (Nikon). Live movies were recorded for 30 seconds at two frames per second.

Axons were identified with anti-neurofascin antibody (NeuroMab, Cat #: 75–027) conjugated to CF405 (Mix-n-Stain CF405S Antibody Labeling Kit; Biotum, Cat# 92231) in the imaging medium. Cells expressing constructs with HaloTag were treated with 50 nM Janelia Fluor 549 dye³⁶ for 10 minutes and washed with the conditioned medium for 10 minutes before imaging. Three-color imaging: Tf555 at 6 µg/mL was maintained in the cellular medium before imaging. HaloTag-myosin Va tail was visualized with 50 nM Janelia Farm 646. Each 500 ms frame includes three pictures: HaloTag-myosin Va and TfR-GFP were imaged simultaneously and Tf555 subsequently.

4.4 |. Image analysis

MetaMorph software (Molecular Devices) was used to generate kymographs from movies. Transport events were traced on the kymographs and the coordinates were exported to Microsoft Excel for analysis. All the kymograph analyses were done in dendrites. A single-blinded analyst identified transport events on all kymographs and determined overlapping events to maintain consistency. The analyst only scored clear and “indisputable” transport events, avoiding those that were borderline, and any transport events <3 µm were excluded from the analysis. Each continuous line with a constant slope was scored as a single transport event and its velocity, run length, and other parameters were calculated. A single vesicle could undergo multiple transport events if there was a distinct pause between each event. To determine vesicle overlap between two channels, the kymograph lines for each channel were drawn independently. The number of overlapping lines for each channel was counted by overlaying the kymograph lines for both channels on a single kymograph. For each cell, the transport events and overlapping lines were combined from each dendrite to yield a single overlap percentage for each cell. The overlap percentages for all the cells in each condition were averaged to yield a single overlap percentage represented on the graph. Between 14 and 23 cells were evaluated for each condition, including cells from at least two independent cultures.

Pearson correlation coefficient data were generated from the same dendritic regions as were used as in the kymograph analysis. After a background subtraction, a region was drawn around a dendrite on the first frame of the movie and the coloc2 Fiji plugin was used to generate the correlation coefficient. For each cell, this process was repeated for all analyzed dendrites. The Pearson correlation coefficient values for all the analyzed dendrites were averaged for each cell. P-values for data sets with equal variance were determined by two-tailed Student’s t tests. P-values for data sets with unequal variance (Figures 3E and 5D) were determined by Welch’s t test.