Myosin VI is the only minus-end actin motor and it is coupled to various cellular processes ranging from endocytosis to transcription. This multi-potent nature is achieved through alternative isoform splicing and interactions with a network of binding partners. There is a complex interplay between isoforms and binding partners to regulate myosin VI. Here, we have compared the regulation of two myosin VI splice isoforms by two different binding partners. By combining biochemical and single-molecule approaches, we propose that myosin VI regulation follows a generic mechanism, independently of the spliced isoform and the binding partner involved. We describe how myosin VI adopts an autoinhibited backfolded state which is released by binding partners. This unfolding activates the motor, enhances actin binding and can subsequently trigger dimerization. We have further expanded our study by using single-molecule imaging to investigate the impact of binding partners upon myosin VI molecular organization and dynamics.
Myosins are actin-based molecular motors which perform vital roles in numerous of cellular processes . Myosin VI (MVI) is associated with several cellular functions, ranging from endocytosis to transcription [2–9]. MVI is unique, in that it is the only member of the myosin family with the ability to move towards the minus end of actin filaments  and its functional diversity relies on its association with various binding partners [11,12].
MVI consists of a motor domain, followed by a neck region consisting of a unique insert, which confers the reverse directionality, and an IQ domain (Figure 1A). Both of these domains bind calmodulin. The N-terminal tail domain contains three structural domains: a three-helix-bundle (amino acids 835–916) , a single-alpha-helix (amino acids 942–978), followed by a short coiled-coil . The C-terminal tail domain consists of the globular cargo binding domain (CBD 1060-end). In addition, two regions within the tail can be alternatively spliced resulting in a 31-residue insertion (large-insert, LI) proximal to the CBD, and/or an nine-residue insertion within the CBD (small-insert, SI). This leads to four splice isoforms, the non-insert (NI), SI, LI and LI + SI , each with distinct intracellular distributions and functions [15,16]. For example, the MVI-NI isoform is able to enter the nucleus, whereas the MVI-LI is confined to the cell periphery .
Backfolding of the myosin VI tail.
The CBD domain enables interactions with several binding partners which control the intracellular localization and function of MVI . This diverse list of partners includes disabled-2 (Dab2) and nuclear dot protein 52 (NDP52). These partners predominantly bind to one of the two established motifs within the CBD of MVI, the WWY and RRL, respectively [8,17,18]. However, interactions occur over wider regions. Binding partner selectivity is driven by isoform splicing, whereby the MVI-LI encodes an alpha helix which occludes the RRL motif . This prevents partners, such as NDP52, from interacting with the protein, and therefore interactions for this isoform can be driven by the WWY motif. In contrast, in the MVI-NI isoform, both motifs are available for binding. However, the RRL site can display higher affinity for partners over the WWY motif, in order to select for those interactions . We have previously revealed that MVI-NI can adopt a back-folded conformation, in which the CBD is brought into close proximity to the motor domain [4,20,21]. NDP52 then interacts through the RRL binding motif, which leads to unfolding and subsequent dimerization of MVI through an internal dimerization site.
Several questions remain unanswered: It is unknown if the structurally distinct MVI-LI isoform is regulated in the same manner, or whether the WWY site can trigger the same structural rearrangements in MVI. Do binding partners enable unfolding of MVI or stabilize an open conformation? How do micromolar protein affinities regulate MVI in the cell? Lastly, how does the release of backfolding regulate motor activity?
To this end, we have now expanded our studies to assess whether this mechanism applies generally to MVI, independent of partner and isoform preference. We have also further explored the molecular basis for this mechanism, and investigated the cellular organization and dynamics with respect to binding partners. Overall, we present a detailed generic model governing the activation of MVI from an inactive back-folded state to an active motor capable of actin binding and dimerization.
Myosin VI back-folding is independent of isoform
We have previously shown that the MVI-NI isoform is back-folded in vitro and in cells . To address whether back-folding is a generic feature of MVI, independent of the isoform, we investigated the conformation of MVI-LI. We utilized a previously employed FRET-based assay [4,6,22] by titrating Alexa555-CBD against FITC-MVI814–1091 (containing the LI) to assess whether there is an interaction between the N- and C-terminal tail domains, as would occur in a back-folded state. A significant concentration-dependent change in FRET was measured, indicating that the two domains are in close proximity (Figure 1B). The same was observed for the MVI-NI isoform MVI814–1060 (Figure 1C), consistent with previous results . The calculated equilibrium dissociation constants (Kd) corresponding to these data were 5.98 (±0.58) µM and 4.8 (±0.61) µM for MVI-LI- and MVI-NI, respectively. These are relatively low micromolar affinities, suggesting the interactions are likely to be dynamic.
Binding partners regulate back-folded myosin VI
We have previously shown  that unfolding of the MVI-NI tail is directly driven by NDP52 following its interaction with the RRL motif and not through Calcium–Calmodulin interactions, as proposed by a different study , and we have previously addressed this discrepancy . To address whether regulation of backfolding by binding partners is a generic mechanism, we followed a similar approach for both MVI-LI and MVI-NI isoforms, but now focusing upon the WWY partner of MVI, Dab2. Due to the instability of recombinant full-length Dab2, we used a recombinant C-terminal truncation of the protein (residues 649–770), which contains the MVI binding site , as performed previously . This truncation of Dab2 will be referred to as tDab2 throughout the manuscript. To assess whether Dab2 can regulate back-folding, the FRET assay between the CBD and the MVI814–1091 (LI) or the MVI814–1060 (NI) tail was repeated following pre-incubation of the CBD with an excess of tDab2. As with NDP52, tDab2 sequestered the CBD, preventing the interaction between the two domains (Figure 1B,C), suggesting that Dab2 is also able to disrupt the intramolecular back-folding in both isoforms. However, tDab2 was not as efficient as NDP52, which is consistent with its weaker affinity for MVI .
To further confirm the direct effect of binding partners upon MVI unfolding, we used the FRET-based MVI tail conformation reporter, which we have previously developed . Our reporter was based on a GFP-RFP FRET pair with the MVITAIL(NI) or MVITAIL(LI), placed in the middle (Figure 1D). As shown by the fluorescent spectra, a high FRET population was observed with both of these reporters, supporting the conclusion that the tails of both isoforms have the ability to back-fold (Figure 1E,F). The addition of 5 µM NDP52 dimers to the MVITAIL(NI) FRET reporter, resulted in loss of the high FRET population. Similarly, 5 µM tDab2 was able to deplete the FRET population of the MVITAIL(LI). However, 20 µM of tDab2 were required to induce an equivalent effect on MVITAIL(NI), given its low affinity for this tail . Also as expected, NDP52 induced little, if any, loss of the high FRET population of the MVITAIL(LI) reporter, given that the RRL site is masked by the LI .
Altogether, these data demonstrate that the intramolecular backfolding of MVI is not isoform specific, but rather an intrinsic feature of the protein. Moreover, binding partners interacting at either motif can regulate this backfolding by triggering the unfolding step.
Binding partners associate to back-folded myosin VI and trigger unfolding
Stopped-flow transient kinetics were then employed to further explore the role of binding partner interactions during the unfolding process. In particular, we endeavoured to determine whether binding partners first interact with the backfolded MVI triggering its unfolding, or whether they bind to spontaneously unfolded MVI stabilizing the conformation. Given that both MVI isoforms showed the same response for either binding partner tested, further experiments only focused upon the NI isoform and NDP52.
First, FITC-MVITAIL(NI) and AF555-NDP52 were used as a FRET pair to report upon the interaction. 1 µM FITC-MVITAIL(NI) was mixed with excess AF555-NDP52 under pseudo-first order conditions (Figure 2A). The fluorescence traces were characterized by two phases: an increase in FRET signal, followed by a partial decrease (Figure 2B). The first phase was fitted to a signal exponential function (Figure 2C) and the observed rate constant was found to be dependent on the concentration of NDP52 (Figure 2E). We extracted an association rate constant of 1.65 µM−1 s−1 and dissociation rate constant of 3.2 s−1, giving a Kd of 1.94 µM. This is consistent with the Equilibrium Dissociation constant previously derived from titrations . The second phase was also fitted to a single exponential function (Figure 2D), however, the derived rate constants (average 2.2 s−1) were independent of NDP52 concentration (Figure 2E). This indicates that the second phase corresponds to a first order process, such as a conformation change. We therefore propose that these biphasic traces directly report upon a two-step process: first, binding of the partner onto backfolded MVI, corresponding to the initial increase in FRET, and second, the subsequent unfolding of the myosin, corresponding to the consequent decrease in FRET due to a greater distance between the donor-acceptor dyes (Figure 2F). This suggests that direct binding of the unfolded MVI is unlikely to occur.
NDP52 association kinetics with the myosin VI tail.
To further understand this partner-induced conformation change, we used fluorescently labelled calmodulin bound to the MVITAIL(NI) as an environmental reporter. This approach would enable the probe to report upon either the partner binding step or the MVI unfolding step, or both processes (Figure 2F,G). Cy3B-calmodulin was pre-mixed with MVITAIL(NI) in a 2 : 1 molar excess. 1 µM Cy3B-calmodulin MVITAIL(NI) was then mixed against an excess of non-fluorescent NDP52, under pseudo-first order conditions. A decrease in fluorescence was observed for all concentrations tested (Figure 2H), which does not arise solely due to Cy3B-calmodulin reacting to NDP52 (Supplementary Figure S1). Interestingly, when the curves were fitted to single exponentials, the derived rate constants (1.95 s−1) were independent of NDP52 concentration (Figure 2I) and were identical with the rate of constants of the second exponential phase observed in the AF555-NDP52 experiments (Figure 2E). We therefore propose that the Cy3B-calmodulin probe reports upon unfolding. This is not unanticipated because unfolding would lead to the largest local environmental change for the dye, as the partner itself binds to the CBD not the neck region. These observations are consistent with the partners not binding to a spontaneously unfolded MVI but rather to its backfolded conformation, which then triggers the unfolding of the protein (Figure 2F).
Myosin VI dimerization is an intrinsic property
Unfolding of the NI isoform subsequently exposes dimerization sites, leading to protein oligomerization , similar to the LI isoform with binding partners . There is a debate as to whether MVI dimerizes intrinsically or through a binding partner mediator [4,13,23–25]. We have previously reported that a tail region ahead of the CBD (Figure 3A) of MVI can dimerize independently of binding partners, but we also suggested that this region is blocked until interactions with binding partners occur .
Dimerization of myosin VI.
NDP52 is a dimeric protein  (Figure 3B) and therefore capable of dimerizing MVI with one CBD bound to each monomer. However, tDab2, which also has been previously shown to be able to dimerize MVI , is monomeric (Figure 3B). This finding reinforces our previous conclusion that dimerization is an intrinsic property of MVI. This allowed us to estimate that the complex would consist of one binding partner (monomer or dimer) per MVI molecule. For example, we would expect 4 NDP52 molecules (or 2 NDP52 dimers) in complex with 2 MVI proteins.
To explore this hypothesis, we performed ATPase measurements on two MVI constructs, namely MVI1–814 and MVI1–1060, with the latter lacking the CBD and therefore having the proposed dimerization region exposed (Figure 3C). The ATPase rates for MVI1–814 (kcat 4.4 s−1) were similar to full-length MVI (kcat 4.1 s−1), while MVI1–1060 displayed a lower ATPase rate (kcat 2.79 s−1), as expected for a dimeric protein. This occurs due to molecular gating, whereby the ATPase activity of the individual motors is coordinated so that only a single motor turns over ATP at any given moment . This rate was also similar to those measured for full-length MVI in the presence of dimeric NDP52 (kcat 2.11 s−1) and monomeric tDab2 (kcat 2.55 s−1). Overall, these ATPase measurements show that dimerization is a feature of MVI and not due to binding partner crosslinking. In addition, size-exclusion chromatography strongly supported the formation of MVI1–1060 dimers. Full-length MVI eluted as a single peak ∼13 ml, whereas MVI1–814 eluted at 15.6 ml (Figure 3D), consistent with this construct being smaller in size. However, MVI1–1060 eluted earlier than both full length MVI and MVI1–814 constructs, suggesting it has adopted a distinct structure which supports its dimeric nature observed in the ATPase measurements. Experiments were also performed with MVI and NDP52, however, the proteins separated during the chromatography, so a complex could not be resolved. Finally, size-exclusion chromatography also showed that MVI814–1060 elutes in two equal peaks, indicating monomer and dimer species. In contrast, MVITAIL(NI) and CBD elute as largely single species (Figure 3E), further supporting our finding that dimerization occurs within amino acids 814–1060. Taken together, the ATPase and size-exclusion chromatography data further support the presence of a dimerization site between amino acids 814–1060, consistent with our previous FRET experiments [4,6].
Lastly, in order to follow the binding partner induced dimerization in real-time, we performed a FRET-based stopped flow assay. 1 µM FITC-MVITAIL(NI) and 1 µM AF555-MVITAIL(NI) were pre-mixed, before being mixed in the stopped-flow with an excess of NDP52, under pseudo-first order conditions (Figure 4A). The fluorescence trace revealed a single exponential increase in FRET signal, with an observed rate constant of 2.45 s−1 (Figure 4A,B). This was independent of NDP52 concentration (Figure 4B), but importantly, it was similar to the unfolding kinetics of the MVITAIL(NI) (Figure 2E,H. Altogether, we propose that dimerization is an intrinsic property of MVI and is a rapid process that occurs once unfolding has been triggered by the interaction with the binding partners, exposing the otherwise masked dimerization site (Figure 4C).
Binding partner driven dimerization of Myosin VI.
Clustering of myosin VI generates high local densities within the cell
We have explored the molecular mechanism underpinning MVI unfolding and dimerization. However, the affinities determined here and previously  are in the low micromolar range. Whilst it is possible that these interactions are part of large multi-valent complexes which enhances the overall complex affinities, the biochemical constants defined here suggest that binding partner interactions and subsequent dimerization would be rare events. We therefore focused upon investigating the in cellulo spatial organization of MVI and NDP52 using super resolution imaging — Stochastic Optical Reconstruction Microscopy (STORM) to investigate cellular interactions .
Widefield imaging against endogenous NDP52 and MVI showed that both proteins are distributed throughout HeLa cells (Figure 5A). STORM imaging of both proteins resolved NDP52 and MVI clusters (Figure 5B). To determine whether this distribution is indeed clustered or random (Figure 5C), we performed cluster analysis using the linearized form of Ripley's K function  L(r)-r, where r is the radius. A plot of L(r)-r versus r gives a value of zero for a random distribution (blue spots), but deviates from zero, due to molecular clustering (Figure 5C). This analysis showed that both NDP52 and MVI assemble into clusters, rather than being randomly distributed. To further understand this clustering behaviour, we used the Clus-DoC software , which allows to quantify the spatial distribution of a protein by generating cluster maps (Figure 5D). In this way, we calculated that, in the cytoplasm, 65% and 70% of MVI and NDP52 molecules, respectively, are in clusters (Figure 5E). Overexpression of Halo-MVI reveals an increase in the mean number percentage of MVI molecules in clusters (Figure 5E), which suggests clustering is concentration dependent. Interestingly, mutation of the RRL binding sites leads to a 50% reduction in the percentage of molecules in clusters (Figure 5E), which links binding partner interactions to molecular clustering. Lastly, treatment with small molecule inhibitor, TIP , which leads to a decrease in actin-activated ATPase activity, results in a significant decrease in the percentage of molecules in a cluster (Figure 5E). Moreover, TIP leads to a significant decrease in the total number of clusters and number of molecules per cluster (Supplementary Figure S3), suggesting that molecular clustering requires MVI ATPase activity.
Cellular clustering of myosin VI and NDP52.
To explore the interaction between MVI and NDP52 within the cell, we performed colocalization analysis of the clusters (Figure 6). Colocalization can be represented by transforming the STORM images (Figure 6A) into colocalization heat maps (Figure 6B), which assigns a different colour to the clusters of each protein depending on the level of colocalization between them. TIP has a clear impact on the fraction of MVI which is colocalized with NDP52, resulting in an almost two-third reduction (Figure 6C). Exploring the cluster data further revealed that, for each protein, the colocalized clusters represent ∼20% of all clusters (colocalized and non-colocalized) (Figure 6D). This is not a surprise given that both proteins interact with various other partners and are also involved in distinct pathways. However, the MVI colocalized clusters are 3-fold larger than the non-colocalized ones, and contain 15-fold more molecules. Treatment with TIP reduces the colocalized cluster area by a third and reduces the number of molecules per cluster by 50%. For NDP52, the colocalized clusters are 2-fold larger than the non-colocalized ones, and contain almost 7-fold more molecules.
Cluster analysis and colocalization of myosin VI and NDP52.
Taken together, the cluster analysis suggests that molecular clustering of MVI does occur within the cell and is promoted by binding partners, while requiring motor activity. As clustered MVI does not represent freely diffusing molecules, the high local density generated by clustering would enhance the impact of its biochemical properties. Within the confined and denser environment of the clusters, the low-affinity interactions of MVI with its partners are more likely to occur. In this way, clustering would ensure that the subsequent unfolding and dimerization of MVI is readily facilitated and implemented, when needed.
Interactions with binding partners regulate the cellular dynamics of myosin VI
Following the single-molecule localization experiments, we performed live-cell 3D single-molecule tracking to observe the impact of binding partners on the dynamics of MVI, using an aberration-corrected multi-focal microscope (acMFM) system . This technique allows the simultaneous acquisition of nine focal planes covering 4 µm in the z axis, with a 20 × 20 µm field of view. We visualized the NI isoform of MVI in HeLa using an N-terminal HaloTag fusion  labelled with JF549  and then performed single-particle tracking.
The trajectories of MVI revealed different populations of molecules, some undergoing confined motion, some random diffusion and others directed movement (Figure 7A). The diffusion constants (D) extracted from these trajectories had a mean value of 0.3 µm2 s−1, under normal conditions, with the large majority of molecules exhibiting slow diffusion (Figure 7B). To better understand the diffusion properties of MVI, we further split these diffusion constants into three groups: a) static/slow moving molecules with D below 0.1 µm2 s−1, b) molecules with D between 0.1 and 2 µm2 s−1 and c) fast moving molecules with D above 2 µm2 s−1 (Figure 7C). We found that MVI molecules are almost equally split between the first two categories, indicating that MVI mainly exists in a static/slow and medium diffusive state. In addition, we assessed the type of particle motion exhibited by MVI by calculating the anomalous diffusion alpha value (Figure 7D). Values below 1 are indicative of confined motion, values above 1 are suggestive of directed motion, whereas values of 1 occur for random diffusion. The alpha value for MVI under normal conditions had a mean value of 0.69, therefore suggesting it mainly undergoes confined movement. This confined motion was further represented using a roseplot (Figure 7E) to depict the angular change in direction within the trajectories. Angles around 180 degrees were enriched and represented trajectories where molecules reversed direction, as would more likely occur for a confined molecule.
Live cell single-molecule dynamics of myosin VI.
To assess the role of binding partners in the cellular dynamics of MVI, we tracked the movement of two MVI mutants in which Dab2 and NDP52 binding was disrupted, namely MVIWWY/WLY and MVIRRL/AAA, respectively. MVIWWY/WLY did not show any significant change in its diffusion properties compared with wild type MVI (Figure 7B,C). This is not surprising since, as shown previously, the NI isoform displays selectivity for RRL binding partners, rather than the lower affinity WWY-mediated interactions . Conversely, MVIRRL/AAA led to a significant shift in the distribution of diffusion constants, with a 30% increase in the mean diffusion constant to a value of 0.39 µm2 s−1 (Figure 7B). This can also be seen by the decrease in the static/slow moving population (<0.1 µm2 s−1) and the increase in the medium and highly diffusive pools (0.1–2 µm2 s−1 and >2 µm2 s−1) (Figure 7C). This was also matched by the significant change in the type of anomalous diffusion exhibited by the MVI mutant molecules, with an increase in the mean anomalous diffusion alpha value to 0.79 (Figure 7D). As evidenced by the dramatic change in the shape of the distribution, disruption of MVI interaction with its RRL binding partners caused a significant shift from confined motion towards random diffusion.
As expected from the biochemical parameters, these data suggest that there is an impact of the binding partners upon the activity of MVI, which influences its cellular dynamics. This was further supported by the effect of transiently overexpressed GFP-tagged CBD into cells stably expressing wild type Halo-MVI. Over-expression of this construct is known to have a dominant negative effect by displacing wild type MVI from binding partner and lipid-based interactions . Indeed, the presence of the CBD led to a considerable increase in the diffusion constant of MVI to a mean value of 0.43 µm2 s−1 (Figure 7B), an increase in the population of highly diffusive species (Figure 7C), and a shift of the alpha value to mean of 0.85 (Figure 7D). Once again, this is consistent with the STORM measurements which revealed less clustering of MVI RRL mutant. To assess the contribution of actin to the pool of static/slow diffusing molecules, the dynamics of MVI were observed following treatment with an actin polymerization inhibitor, Latrunculin B. Not surprisingly, there was a significant increase in MVI diffusion (Figure 7B,C), due to the loss of actin-based interactions.
Overall, our data demonstrate that the interaction of MVI with its binding partners plays an important role in regulating the cellular dynamics of the protein. The increase in random diffusion observed following disruption of these interactions could reflect a loss of interaction with cargo and anchoring sites. It could also reflect loss of direct interactions with actin, given that MVI would then be in a back-folded inactive state.
Binding partner interactions enhance actin binding
To directly test how binding partners impact actin binding, we performed actin isolation assays with recombinant proteins (Figure 8A,B). As expected, MVI is isolated by F-actin. The amount of MVI bound to actin is 50% where binding is dependent upon actin concentration and ATP (Supplementary Figure S4). The addition of an excess of NDP52 led to above 80% of MVI being bound to actin. As expected, the RRL–AAA mutant did not show the enhanced binding in the presence of NDP52. Whilst these experiments are in alignment with our assays above, this finding contradicts the claims in , where actin pelleting assays were performed in cell lysates. Based on the findings from Arden et al., we would expect to find these mutant proteins decorating actin filaments. To this end, we transfected HeLa cells with eGFP–WT, RRL–AAA and WWY–WLY MVI (Figure 8C). Some colocalization with actin filaments are observed but there are no obvious differences between mutants and wild type MVI. We therefore conclude we are not enhancing actin binding above wild type levels with the mutant proteins.
Binding partners enhance actin binding.
This study has provided novel insights into the mechanisms underlying the regulation of MVI and its function within a cell. We have revealed that binding partners directly regulate the structural transitions of MVI from an auto-inhibited backfolded monomer to an active motor, which is linked to actin and cargo (Figure 8A). Our data suggest that this mechanism is generic, irrespective of MVI isoform or choice of binding partners. Importantly, this model does not assume that binding partners must be dimeric. Moreover, pre-steady-state kinetics has allowed us to define that binding partners first bind to the backfolded MVI and subsequently trigger unfolding, rather than binding to a spontaneously unfolded MVI and stabilizing that state. This is an important regulatory step which enables binding partners to turn myosin VI on and off.
We have shown that once MVI is unfolded, it has the intrinsic ability to dimerize, as it has been reported before [4,6,13,24,25,35]. Here, we have further clarified these models by showing that MVI dimerizes internally and that binding partner dimerization is not the driving factor. We therefore propose that the complex stoichiometry exists as one active binding partner (monomer or dimer) per MVI molecule (Figure 9A). This stoichiometry has been reported from structural studies [25,36]. Greater differences in oligomerization status may occur between binding partners where monomers or dimers can be favoured. Moreover, larger oligomers, while maintaining 1 : 1 stoichiometry, could also be produced .
Model describing the activation of myosin VI by binding partners.
Rapid mixing kinetics allowed us to probe the individual mechanistic steps and therefore propose that dimerization can occur rapidly following unfolding of MVI, with unfolding being the rate-limiting step once binding partner interaction occurs.
MVI exists as four alternative spliced isoforms and interacts with a wide range of binding partners. In our study, we assessed the generic nature of MVI unfolding and dimerization, using examples of two MVI isoforms, which represent the two largest structural variants, and two binding partners, which bind at alternative motifs. Although we cannot exclude potential differences in behaviour in other isoforms and with other MVI partners.
Our biochemical analysis has furthered our understanding of how MVI interacts with its binding partners and highlighted the impact of these interactions upon MVI activity. We have previously shown that this relies on the interplay between isoform splicing and the differential affinities between the RRL and WWY binding sites motifs . Here, we have provided insights into the order of events occurring during partner binding, unfolding and dimerization. However, by nature, all these biochemical studies occur under an environment allowing free molecular diffusion, which is strikingly different from the crowded cellular environment. Our high-resolution imaging data shed light on how the biochemical features of MVI measured in vitro could translate into the complex cellular context. They revealed that over 60% of MVI molecules are clustered within the mammalian cell. The high local densities within these clusters enhance the probability of interactions with its binding partners, which have been biochemically shown to be in the micromolar range. Moreover, our data have shown that association with binding partners further increases these clusters in size and molecular density. Although this was tested here only for the partner NDP52, it is very likely to apply for all binding partners. The reason is that, within such clusters, multivalent interactions of MVI to cargo through various binding partners would lead to stable complexes capable of long-range processive movement. Enzymatic clustering has been reported for many cellular processes [37,38], where is it hypothesized to increase local concentrations to enhance efficiency. Therefore, given the multifunctional and complex nature of MVI intracellular activity, it would not be surprising that this myosin follows a mechanism, which allows stability, efficiency and high level of regulation.
In addition to what was demonstrated biochemically, the impact of binding partners upon MVI is also evident from the live cell single-molecule tracking data. Here, we have shown how mutations, or truncations that lead to the disruption of interactions between MVI and binding partners have a profound effect on the dynamics of the protein inside the cell, leading to loss of the static and slow-moving molecules, which display restrictive diffusion, for more rapidly diffusing molecules exhibiting random motion. To account for such a behavioural switch, we propose that the direct association with the actin cytoskeleton and cellular cargo is decreased when interactions with binding partners are disrupted. In this manner, a backfolded MVI would freely diffuse around the cell and then interact with the cytoplasm when activated by a binding partner on a cargo.
It was previously reported that disruption of MVI — binding partner/cargo interactions leads to an increase in actin binding using a cell-based pull-down assays . However, while imaging was performed with other mutations within their study to show enhanced actin binding, those targeting the RRL and WWY sites were not observed. We performed imaging of these mutations and could not find them decorating actin filaments, as would have been expected based on the study. Instead, the results are consistent with the data presented here, and with our recombinant actin binding assays, where we find that binding partners enhance actin associations.
These observations allowed us to further expand our activation model of MVI. We propose that the back-folded MVI, being inactive, can randomly diffuse in the cell. In this state, MVI can be sequestered by binding partners to cellular cargo or organelles, and get subsequently activated (Figure 9B). Diffusion of MVI in its inactive state is important because it allows sampling of various regions of the cell before recruitment by its partners to function in specific biological processes. Interestingly, we would expect to observe a decrease in ATPase activity if the protein is not bound to actin. However, these changes have not been observed through biochemical analysis, possibly due to the marked differences between the chemical environment of biochemical assays and the cellular environment. Also, within the complex cellular environment, other processes, such as further regulation of the motor by calcium–calmodulin or changes in the phosphorylation status of MVI upon perturbation of binding partner interactions, could lead to inhibition of the ATPase activity, and these would not be observed in vitro. These factors remain to be determined.
In summary, our multidisciplinary study provides new insights into the regulation of MVI by binding partners in the cell. The regulation is complex and our knowledge is not complete. It remains unknown as to how monomeric or dimeric states of the protein are selected within the cell, which may be the role of the binding partners to fulfil. Interestingly, due to the internal dimerization of MVI, we suggest that monomeric activities would arise from binding partners activity halting dimerization. Given the established impact of MVI in several diseases, including cancer [19,39–42] and deafness [34,43], defining the mechanistic details of its regulation and function is critical for understanding the impact of mutations, truncations, or altered expression of MVI or partners during disease. Overall, these insights provide new avenues for exploring how the activity of this multi-functional motor protein is regulated within the cell and how these processes may be perturbed during disease.
A list of constructs and PCR primers are provided in Supplementary Tables S1 and S2, respectively. Constructs generated in this work are described below: RRL/AAA and WWY/WLY mutations were made by site-directed mutagenesis using standard Quick-Change site-directed mutagenesis protocol with pLV-Tet0-Halo MVI as the template. All plasmids were verified by DNA sequencing. Proteins were expressed and purified, as descried in the Supplementary Methods.
Protein expression and purification in Escherichia coli
Recombinant constructs were expressed in E.coli BL21 DE3 cells (Invitrogen) in Luria Bertani media. Proteins were purified by affinity chromatography (HisTrap FF, GE Healthcare). The purest fractions were desalted through a PD10 column (GE Healthcare) to remove imidazole before treatment with TEV protease for 4 h at 25°C. The samples were then passed through a second HisTrap column. The cleaved protein was further purified through a Superdex 200 16/600 column (GE Healthcare).
Protein expression using Baculovirus system
Full-length MVI NI, MVI1–814, MVI1–1060, MVIRRL/AAA and Xenopus calmodulin were expressed in Sf9 and Sf21 (Spodoptera frugiperda) insect cells using the Bac-to-Bac® Baculovirus Expression System (Invitrogen). Sf9 cells were cultured in Sf900 media (Gibco). Recombinant bacmids were generated following the manufacturer's instructions and were transfected into adherent Sf9 cells to generate the P1 viral stock. Sf9 cells were infected in suspension at 27°C and 100 rpm with 1 in 50 dilution of P1 and P2 viral stocks to yield P2 and P3 stocks, respectively. Finally, expression of recombinant proteins was set up by infecting sf21 cells with the P3 viral stock in Spodopan media (PAN Biotech). To ensure correct folding of the MVI constructs, cells were simultaneously infected with P3 viral stock of the MVI constructs together with calmodulin at a 0.75 ratio The cells were harvested after 3 days by centrifugation for 15 min at 700×g and at 4°C and resuspended in ice cold myosin extraction buffer (90 mM KH2PO4, 60 mM K2HPO4, 300 mM KCl, pH 6.8), supplemented with Proteoloc protease inhibitor cocktail (Expedeon) and 100 µM PMSF, before proceeding to protein purification. Prior to sonication, an additional 5 mg recombinant calmodulin was added together with 2 mM DTT. After sonication, 5 mM ATP and 10 mM MgCl2 were added and the solution was rotated at 4°C for 30 min before centrifugation (20 000g, 4°C, 30 min). Then, the cell lysate was subjected to the purification. Proteins were purified by affinity chromatography (HisTrap FF, GE Healthcare). The purest fractions were further purified through a Superdex 200 16/600 column (GE Healthcare).
Proteins were transferred into 50 mM Na-phosphate (pH 6.5) using a PD10 desalting column. Samples were then incubated with a 5-fold excess of dye for 4 h, rotating at 4°C. Excess dye was removed using a PD10 desalting column pre-equilibrated with 50 mM Na-Phosphate, 150 mM NaCl and 1 mM DTT. Labelling efficiency was calculated based on the absorbance at 280 nm and the absorbance maximum of the dye. Typical efficiency was 90%, whereby the less than complete labelling was taken as an indicator for a single dye per protein. This was tested for isolated preparations in mass spectroscopy, which revealed both an unlabelled and single labelled population.
Size-exclusion chromatography and multi-angle light scattering
An amount of 100 µl samples of 2 mg/ml purified protein, was applied to a Superdex 200 (30 × 1 cm) analytical column (GE Healthcare) equilibrated in 150 mM NaCl, 50 mM Tris.HCl (pH 7.5) and 1 mM DTT and controlled using Waters 626 HPLC at room temperature. Eluted proteins were analysed with Viscotek SEC-MALS 9 and Viscotek RI detector VE3580 (Malvern Panalytical). Molecular mass was determined using OmniSEC software.
Constructs were incubated at the specified concentrations in reaction buffer (150 mM NaCl, 50 mM Tris.HCl (pH 7.5), 1 mM MgCl2 and 1 mM DTT), for 10 min at RT. 5 µM F-actin (mixture 20% Biotinylated actin (Cytoskeleton Inc.)) was added to the mixture and incubated at RT for 10 min. Streptavidin Dynabeads (M-280) (Invitrogen) were washed three times according to the manufacturer's instructions and then finally in reaction buffer. An amount of 0.2 mg/ml beads were added to the samples. The samples were incubated for 5 min at RT before magnetic isolation. The isolated sample was re-suspended in an equal volume to the supernatant. The samples were analyzed by SDS–PAGE.
Steady-state ATPase activity of MVI
Ca2+-actin monomers were converted to Mg2+-actin with 0.2 mM EGTA and 50 µM MgCl2 before polymerizing by dialysis into 20 mM Tris–HCl (pH7.5), 20 mM imidazole (pH 7.4), 25 mM NaCl and 1 mM DTT. A 1.1 molar equivalent of phalloidin (Sigma) was used to stabilize actin filaments, as previously described .
All reactions were performed at 25°C in a buffer containing 50 mM Tris·HCl (pH 7.5), 150 mM sodium chloride and 1 mM DTT in a final volume of 100 µl. Measurements were performed using a ClarioStar Plate Reader (BMG Labtech). Intensity measurements were performed at the following wavelengths: FITC (ex. 490 nm), Alexa Fluor 555 (ex. 555 nm). FITC to Alexa Fluor 555 FRET measurements were performed using the following wavelengths ex. 470 nm and em. 575 nm.
Stopped flow measurements
A HiTech SF61DX2 apparatus (TgK Scientific Ltd, Bradford-on-Avon, U.K.) with a mercury-xenon light source and HiTech Kinetic Studio 2 software was used [45,46]. For FRET experiments, excitation was at 495 nm with emission through a 570 nm cut-off filter (Schott Glass). For Cy3B, excitation was at 550 nm with emission through a 570 nm cut-filter (Schott Glass). In all experiments, the quoted concentrations are those in the mixing chamber, except when stated. All experiments were performed at 25°C in 50 mM Tris–HCl (pH 7.5), 150 mM NaCl, 1 mM DTT and 3 mM MgCl2. The dead time of the stopped-flow instrument was ∼2 ms: during this initial time no change in fluorescence can be observed.
Analysis of kinetic data
Cell culture and transfection
HeLa (ECACC 93021013) cells were cultured at 37°C and 5% CO2, in Gibco MEM Alpha medium with GlutaMAX (no nucleosides), supplemented with 10% Fetal Bovine Serum (Gibco), 100 units/ml penicillin and 100 µg/ml streptomycin (Gibco). For the transient expression of MVI mutants, HeLa cells grown on glass coverslips were transfected using Lipofectamine 2000 (Invitrogen), following the manufacturer's instructions. Depending on the construct, 24 h–72 h after transfection, cells were subjected to further analysis. To inhibit actin polymerization, cells were treated with 1 µM Latrunculin B (Sigma) for 1 h at 37°C.
HeLa cells were fixed for 15 min at room temperature in 4% (w/v) paraformaldehyde (PFA) in PBS and residual PFA was quenched for 15 min with 50 mM ammonium chloride in PBS. All subsequent steps were performed at room temperature. Cells were permeabilized and simultaneously blocked for 15 min with 0.1% (v/v) Triton X-100 and 2% (w/v) BSA in PBS. Cells were then immuno-stained against the endogenous proteins by 1 h incubation with the indicated primary and subsequently the appropriate fluorophore-conjugated secondary antibody (details below), both diluted in 2% (w/v) BSA in PBS. The following antibodies were used at the indicated dilutions: Rabbit anti-myosin VI (1 : 200, Atlas-Sigma HPA0354863), Mouse anti-NDP52 (1 : 250 Abcam ab124372), Donkey anti-mouse Alexa Fluor 488-conjugated (1 : 250, Abcam Ab181289), Donkey anti-rabbit Alexa Fluor 647-conjugated (1 : 250, Abcam Ab181347). Coverslips were mounted on microscope slides with Mowiol (10% (w/v) Mowiol 4-88, 25% (w/v) glycerol, 0.2 M Tris–HCl, pH 8.5), supplemented with 2.5% (w/v) of the anti-fading reagent DABCO (Sigma).
Glass coverslips were cleaned by incubating them for 3 h, in etch solution, made of 5 : 1 : 1 ratio of H2O: H2O2 (50 wt. % in H2O, stabilized, Fisher Scientific) : NH4OH (ACS reagent, 28–30% NH3 basis, Sigma), placed in a 70°C water bath. Cleaned coverslips were repeatedly washed in filtered water and then ethanol, dried and used for cell seeding.
Cells were seeded on pre-cleaned No 1.5, 25 mm round glass coverslips, placed in six-well cell culture dishes. Cells were fixed in pre-warmed 4% (w/v) PFA in PBS and residual PFA was quenched for 15 min with 50 mM ammonium chloride in PBS. Immunofluorescence (IF) was performed in filtered sterilized PBS. Cells were permeabilized and simultaneously blocked for 30 min with 3% (w/v) BSA in PBS supplemented with 0.1% (v/v) Triton X-100. Permeabilized cells were incubated for 1 h with the primary antibody and subsequently the appropriate fluorophore-conjugated secondary antibody, at the desired dilution in 3% (w/v) BSA, 0.1% (v/v) Triton X-100 in PBS. The antibody dilutions used were the same as for the normal IF protocol (see above). Following incubation with both primary and secondary antibodies, cells were washed three times, for 10 min per wash, with 0.2% (w/v) BSA, 0.05% (v/v) Triton X-100 in PBS or TBS. Cells were further washed in PBS and fixed for a second time with pre-warmed 4% (w/v) PFA in PBS for 10 min. Cells were washed in PBS and stored at 4°C, in the dark, in 0.02% NaN3 in PBS, before proceeding to STORM imaging.
Before imaging, coverslips were assembled into the Attofluor® cell chambers (Invitrogen). Imaging was performed in freshly made STORM buffer consisting of 10% (w/v) glucose, 10 mM NaCl, 50 mM Tris — pH 8.0, supplemented with 0.1% (v/v) 2-mercaptoethanol and 0.1% (v/v) pre-made GLOX solution which was stored at 4°C for up to a week (5.6% (w/v) glucose oxidase and 3.4 mg/ml catalase in 50 mM NaCl, 10 mM Tris — pH 8.0). All chemicals were purchased from Sigma.
Imaging was undertaken using the Zeiss Elyra PS.1 system. Illumination was from a HR Diode 642 nm (150 mW) and HR Diode 488 nm (100 mW) lasers where power density on the sample was 7–14 kW/cm2 and 7–12 kW/cm2, respectively.
Imaging was performed under highly inclined and laminated optical (HILO) illumination to reduce the background fluorescence with a 100× NA 1.46 oil immersion objective lens (Zeiss alpha Plan-Apochromat) with a BP 420–480/BP495–550/LP 650 filter. The final image was projected on an Andor iXon EMCCD camera with 25 ms exposure for 20000 frames.
Image processing was performed using the Zeiss Zen software. Where required, two channel images were aligned following a calibration using a calibration using pre-mounted MultiSpec bead sample (Carl Zeiss, 2076–515). The channel alignment was then performed in the Zeiss Zen software using the Affine method to account for lateral, tilting and stretching between the channels. The calibration was performed during each day of measurements.
The images were then processed through our STORM analysis pipeline using the Zen software. Single molecule detection and localization was performed using a nine pixel mask with a signal to noise ratio of six in the ‘Peak finder' settings while applying the ‘Account for overlap’ function. This function allows multi-object fitting to localize molecules within a dense environment. Molecules were then localized by fitting to a 2D Gaussian.
The render was then subjected to model-based cross-correlation drift correction. Typical localization precision was 20 nm for Alexa-Fluor 647 and 30 nm for Alexa-Fluor 488. The final render was then generated at 10 nm/pixel and displayed in Gauss mode where each localization is presented as a 2D gaussian with a standard deviation based on its precision. The localization table was exported as a txt for import in to Clus-DoC.
The single molecule positions were exported from Zeiss black version and imported into the Clus-DoC analysis software  (https://github.com/PRNicovich/ClusDoC). Cytoplasmic areas were selected as ROIs for cluster analysis. First, the Ripley K function was completed on each channel identifying the r max. The r max was then assigned for DBSCAN if one channel was being analysed or Clus-Doc if two channel colcalisation was being analysed. The clustering size was set to a minimum of five molecules, with smoothing set at 7 nm and epsilon set at the mean localization precision for the dye. All other analyses parameters remained at default settings. Data concerning each cluster was exported and graphed using Plots of Data.
Multi-focal imaging and particle tracking analysis
Cells stably or transiently expressing Halo-tag constructs were labelled for 15 min with 10 nM HaloTag-JF549 ligand, in cell culture medium at 37°C, 5% CO2. Cells were washed for three times with warm cell culture medium and then incubated for further 30 min at 37°C, 5% CO2. Cells were then washed three times in pre-warmed FluoroBrite DMEM imaging medium (ThermoFisher Scientific), before proceeding to imaging.
Single molecule imaging was performed using an aberration-corrected multifocal microscope (acMFM), as described by Abrahamsson et al. . Briefly, samples were imaged using 561 nm laser excitation, with typical irradiance of 4–6 kW/cm2 at the back aperture of a Nikon 100× 1.4 NA objective. Images were relayed through a custom optical system appended to the detection path of a Nikon Ti microscope with focus stabilization. The acMFM detection path includes a diffractive multifocal grating in a conjugate pupil plane, a chromatic correction grating to reverse the effects of spectral dispersion, and a nine-faceted prism, followed by a final imaging lens.
The acMFM produces nine simultaneous, separated images, each representing successive focal planes in the sample, with ca. 20 µm field of view and nominal axial separation of ca. 400 nm between them. The nine-image array is digitized via an electron multiplying charge coupled device (EMCCD) camera (iXon Du897, Andor) at up to 32 ms temporal resolution, with typical durations of 30 s.
3D + t images of single molecules were reconstructed via a calibration procedure, implemented in Matlab (MathWorks), that calculates and accounts for (1) the inter-plane spacing, (2) affine transformation to correctly align each focal plane in the xy plane with respect to each other, and (3) slight variations in detection efficiency in each plane, typically less than ±5–15% from the mean.
Reconstructed data were then subject to pre-processing, including background subtraction, mild deconvolution (3–5 Richardson-Lucy iterations), and/or Gaussian de-noising prior to 3D particle tracking using the MOSAIC software suite . Parameters were set where maximum particle displacement was 400 nm and a minimum of 10 frames was required. Tracks were reconstructed, and diffusion constants were extracted via MSD analysis  using custom Matlab software assuming an anomalous diffusion model.
Unless stated, data fitting and plotting was performed using Plots of data  and Grafit Version 5 (Erithacus Software Ltd). Cartoons were generated using the BioRender software.
The data supporting the findings of this study are available from the corresponding author on request.
The authors declare that there are no competing interests associated with the manuscript.
We thank the UKRI-MRC (MR/M020606/1) and UKRI-STFC (19130001) for funding. Aberration-corrected multi-focal microscopy was performed in collaboration with the Advanced Imaging Center at Janelia Research Campus, a facility jointly supported by the Howard Hughes Medical Institute and the Gordon and Betty Moore Foundation.
CRediT Author Contribution
Chris Toseland: Conceptualization, Resources, Data curation, Supervision, Funding acquisition, Investigation, Writing — original draft, Project administration. Alia dos Santos: Conceptualization, Investigation. Natalia Fili: Investigation. Yukti Hari-Gupta: Investigation. Rosemarie E. Gough: Investigation. Lin Wang: Resources. Marisa Martin-Fernandez: Resources. Jesse Arron: Resources. Eric Wait: Resources. Teng Leong Chew: Resources.
We also thank Darren Griffin (University of Kent) and Mike Geeves (University of Kent) for sharing of equipment and reagents, and Satya Khuon (Janelia Research Campus) for assisting with cell culture. The JF549 dye was kindly provided by Luke Lavis (Janelia Research Campus).
These authors contributed equally to this work.
Current address: School of Life Sciences, University of Lincoln, Lincoln LN6 7DL, U.K.
Current address: MRC LMCB, University College London, Gower Street, London WC1E 6BT, U.K.