Disruption of the Dictyostelium myoi (myo7) gene encoding DdMyo7 results in a significant defect in filopodia formation that is rescued by expression of GFP-DdMyo7 (Figure 1A, Figure 1—figure supplements 1A and 2A; Tuxworth et al., 2001; Petersen et al., 2016). Most strikingly localized to filopodia tips, DdMyo7 is also in the cytosol and localized to the leading edge of cells (Figure 1B, Figure 1—figure supplement 1A). In the course of characterizing functionally important regions of DdMyo7, it was observed that the tail domain (aa 809 - end) is localized all around the cell periphery in contrast to the full-length myosin which is often restricted at one edge of the cell cortex, (Figure 1B; Petersen et al., 2016; Arthur et al., 2019). This was unexpected as the myosin tail region is largely regarded as playing a key, even determining, role in targeting myosins. DdMyo7-mCherry and GFP-DdMyo7 tail were co-expressed in Dictyostelium cells (Figure 1B, Figure 1—figure supplement 1B) and a line scan through the cell showed that while both are present at a region of the cell that is extending outwards and producing filopodia (i.e. the leading edge), the tail is also strongly enriched in the cell rear (Figure 1C, line from Figure 1B). The extent of co-localization of the full-length myosin with the tail domain was assessed using cytofluorograms. This method quantifies co-localization of two proteins by comparing the intensity of the two fluorescent signals on a pixel-wise basis for a field of cells with r representing the correlation coefficient (Bolte and Cordelières, 2006). Quantification revealed less correlation between DdMyo7 and the tail (r = 0.63) than is seen for the two full-length myosins (r = 0.86) (Figure 1D). Due to the transient polarity of vegetative cells, a second measure of distribution of DdMyo7 on the cortex was carried out. To quantify the localized enrichment of DdMyo7 without any bias for manually identifying the leading edge, the intensity all around the cortex of cells was measured using a high throughput automated image analysis FIJI macro, Seven (Petersen et al., 2016; Figure 1E). Actin is typically enriched at the leading edge with high intensity at one edge and lower intensity elsewhere (Figure 1F,G). The standard deviation of intensities values around the periphery was calculated for each cell (as exemplified in Figure 1H). This analysis shows that actin (RFP-Lifeact; Brzeska et al., 2014) and GFP-DdMyo7 are localized asymmetrically with a high standard deviation of cortical intensity (Figure 1F,G,I). In contrast, the tail appears uniformly localized around the periphery (Figure 1F,G) and has a lower average SD for the cortical intensity around the cell periphery (Figure 1I). Interestingly, a tail-less forced dimer of the motor (motor-FD; Figure 1A) that is capable of inducing filopodial protrusions (Arthur et al., 2019) is also localized asymmetrically with a high standard deviation of intensity (Figure 1F,I). Together, these data establish that restricted localization of full-length DdMyo7 is dependent on its motor domain.

The apparent asymmetrical distribution of DdMyo7 to the leading edge (Figure 1F) suggested that the actin at the cortex has a role in the recruitment and localization of DdMyo7. DdMyo7 and actin appear to colocalize at the leading edge in cells expressing both GFP-DdMyo7 and an actin marker (Lifeact) (Figure 2A). Linescans show that indeed there is strong correlation of Lifeact and DdMyo7 fluorescence around the cell periphery (Figure 2B, yellow line from 2A). Cytofluorogram analysis also revealed strong correlation of the DdMyo7 and actin fluorescence intensities (Figure 2C). A line scan taken perpendicular to an extending leading edge showed a steady accumulation of DdMyo7 intensity at the same time as an increase in actin intensity (Figure 2D,E). DdMyo7 and actin normalized intensities during leading edge extension were measured for 10 independent cells and a spline fit shows a robust correlation with actin intensity (Figure 2F).

The dependence of DdMyo7 localization on actin polymerization at the cortex was tested by treating cells with actin modulating drugs. CytochalasinA (cytoA) binds to the fast-growing (barbed) end of actin filaments and blocks incorporation of actin monomers, capping and stabilizing filaments (Cooper, 1987). LatrunculinA (latA) sequesters monomers and prevents actin filament growth, and CK-666 blocks the Arp2/3 complex and actin branching (Coué et al., 1987; Nolen et al., 2009). Cells were incubated with each of these drugs and the impact on DdMyo7 cortical targeting was assessed by measuring the intensity ratio of DdMyo7 in a 0.8 μm band around the perimeter versus the cytoplasm (see Figure 1E). Control cells have a cortex:cytoplasm ratio of about ~1.2, indicating an overall 20% enrichment of DdMyo7 on the cell cortex (Figure 2G,H; Arthur et al., 2019). Treatment of cells with either cytoA or latA reduced the ratio to ~1 indicating a total loss of cortical localization of DdMyo7 (Figure 2H). CK-666 also significantly reduced DdMyo7 cortical recruitment (Figure 2G,H). Consistent with the reduction or loss of polymerized actin and cortical DdMyo7, filopodia formation was significantly reduced with these actin modulating drugs (Table 1). Cells were also treated with jasplakinolide (jasp) that promotes monomer nucleation and stabilizes ADP-Pi actin filaments (Merino et al., 2018). Jasp treatment had the opposite effect, resulting in increased recruitment of DdMyo7 to the cortex and increased filopodia formation (Figure 2G,H; Table 1). The addition of PI3Kinase inhibitors did not disrupt DdMyo7 cortical targeting or filopodia formation, but did disrupt the localization of a PIP3 reporter (GFP-CRAC, Parent et al., 1998; Figure 2—figure supplement 1A–C, Table 1). Microtubules have been implicated in regulating filopodia in melanoma cells, modulating their localization, density and merging (Schober et al., 2007). Treatment of Dictyostelium cells with the microtubule depolymerizing compound nocodazole disrupts cytosolic microtubules (visualized by cells expressing GFP-tubulin Neujahr et al., 1998, Figure 2—figure supplement 1E). Nocodazole reduced DdMyo7 cortical targeting and partially inhibited filopodia formation, particularly at higher concentrations (50 µM) (Figure 2—figure supplement 1E,G,I; Table 1). Interestingly, cortical targeting of the DdMyo7 tail fragment is not reduced in cells treated with nocodazole (Figure 2—figure supplement 1F,H) in spite of the presence of two MF domains with micromolar affinity for microtubules in the DdMyo7 tail (Planelles-Herrero et al., 2016). Additionally, DdMyo7 is not strongly localized to microtubules (Figure 2—figure supplement 1D,E). Together, the data suggest that microtubule-actin crosstalk may impact the dynamics of the cortical actin network and thus indirectly control the cortical recruitment of full-length DdMyo7.

Filopodia initiation and extension is driven by regulators of actin polymerization such as VASP and formin (Mattila and Lappalainen, 2008; Figure 3A,B). The actin bundler/polymerase DdVASP accumulates at the leading edge of cells and is important for filopodia formation (Han et al., 2002; Breitsprecher et al., 2008). Interestingly, the Dictyostelium vasp null mutant phenocopies the myo7 null mutant - it lacks filopodia, has reduced adhesion and smaller cell size (Han et al., 2002; Tuxworth et al., 2001). This is of particular note as mammalian Myo10 and VASP are observed to co-transport in filopodia and co-immunoprecipitate, suggesting a role for Myo10 in the transport of VASP to filopodia tips to promote filopodia growth (Tokuo and Ikebe, 2004; Lin et al., 2013). There is currently no evidence that DdMyo7 and Dictyostelium VASP (DdVASP) interact with each other and co-immunoprecipitation experiments in Dictyostelium did not detect any interaction (Figure 3—figure supplement 1A). Surprisingly, in spite of this lack of interaction, DdMyo7 fails to target efficiently to the cortex of vasp- cells (Figure 3C,D). In contrast, DdVASP localizes to the cortex regardless of the presence of DdMyo7 (Figure 3C,D).

Formins are localized to the leading edge of migrating cells and play an important role in filopodia formation (Schirenbeck et al., 2005; Pellegrin and Mellor, 2005; Yang et al., 2007). Formins promote actin polymerization by incorporating actin monomers at the barbed end of actin filaments and elongate parallel actin filaments (Breitsprecher and Goode, 2013; Mellor, 2010). The Dictyostelium diaphanous related formin dDia2 makes significant contributions to filopodia formation and is required for normal filopodia length (Figure 3A,B; Schirenbeck et al., 2005). Formin activity is not required for cortical recruitment of DdMyo7 as it is found to localize normally to the cortex of dDia2 null (Figure 3C,D). These data reveal that DdVASP or its actin polymerization activity is critical for DdMyo7 recruitment and suggests that this bundler/polymerase acts upstream of DdMyo7.

The finding that DdVASP is critical for DdMyo7 cortical targeting raises the question of how it is acting to recruit this myosin. One simple explanation is that DdMyo7 is targeting to regions of the cell cortex where dynamic actin polymerization occurs. This hypothesis was tested by examining if robust recruitment of DdMyo7 could occur in the absence of DdVASP when actin polymerization was stimulated using different manipulations. Treatment of Dictyostelium with latrunculinA treatment followed by washout produces robust actin waves (Gerisch et al., 2004). These waves are made of a dense Arp2/3 branched actin meshwork and class I myosins (MyoB, MyoC) (Jasnin et al., 2019; Brzeska et al., 2020). Control or vasp null cells expressing both RFP-LimEΔcoil (a marker for F-actin) and GFP-DdMyo7 were induced to form waves that are readily apparent as broad circles of actin that emanate outwards toward the cell periphery. Robust actin waves marked by RFP-LimEΔcoil were observed but DdMyo7 was not observed in the waves in either cell line (n = 6 per genotype; Figure 4A, Figure 4—figure supplement 1A,B). Next, vasp-cells were treated with jasplakinolide to stimulate actin polymerization (Figure 4—figure supplement 1C). No increase in cortical targeting of DdMyo7 was observed in vasp nulls treated with jasplakinolide when compared to untreated cells (Figure 4B,E). Blocking capping protein by overexpressing the capping protein inhibitor V-1 stimulates filopodia formation in Dictyostelium (Figure 4—figure supplement 1D, Jung et al., 2016). V-1 also stimulates cortical actin network formation in vasp null Dictyostelium (Figure 4C). However, no increase in targeting was observed in vasp nulls with induced overexpression of V-1 (Figure 4B,D), nor was filopodia formation restored (Figure 4F).

Activation of formins can also result in robust polymerization of actin filaments that are organized into linear bundles (Breitsprecher and Goode, 2013). These polymerases are tightly controlled by autoinhibition with the conserved DAD domain binding to the N-terminal DID domain, blocking actin nucleation activity. Mutation of conserved basic residues in the C-terminal DAD domain creates a constitutively activated (CA) formin (Wallar et al., 2006). A pair of charged residues are present in the dDia2 DAD domain and these were mutated - R1035A, R1036A - to create a CA formin (see alignment in Figure 4—figure supplement 1E). Either dDia2 or dDia2-CA overexpressing vasp null cells exhibit restored and increased DdMyo7 cortical targeting (Figure 4B,E). Overexpression of dDia2-CA also promoted a modest rescue of filopodia formation (Figure 4F; Table 2). Linescans of cells stained with phalloidin show that while vasp nulls have less cortical F-actin than wild type cells, expression of dDia2-CA in vasp nulls resulted in increased cortical F-actin (Figure 4D). In summary, stimulation of actin polymerization by inducing actin waves, treating cells with jasplakinolide, or blocking capping protein is not sufficient to promote DdMyo7 cortical recruitment in vasp null cells. In contrast, expression of activated formin is sufficient for cortical recruitment in vasp null cells. This strongly suggests that DdMyo7 cortical targeting specifically requires the activity of actin regulators that generate parallel actin filaments.

The properties of VASP critical for DdMyo7 recruitment were investigated using separation of function mutations. VASP accelerates the rate of actin polymerization, bundles actin filaments and blocks capping protein from binding to the barbed ends of actin filaments (Breitsprecher et al., 2008; Breitsprecher et al., 2011; Hansen and Mullins, 2010). VASP is a constitutive tetramer that can bundle actin by binding to the sides of filaments (see Figure 5A; (Bachmann et al., 1999; Breitsprecher et al., 2008; Brühmann et al., 2017). DdVASP F-actin binding is mediated by a region within the EVH2 domain (aa 264–289) and tetramerization by a region at the C-terminus (aa 341–375) (Schirenbeck et al., 2006; Breitsprecher et al., 2008). Two different mutants were used to test if the VASP bundling activity is required for DdMyo7 recruitment (Figure 5D). A monomeric VASP was created by deleting the tetramerization domain (∆tet; ∆341–375)) and an F-actin binding mutant was generated in which conserved lysines were mutated to glutamate (FAB K-E; KR275, 276EE + K278E + K280E; based on Hansen and Mullins, 2010). The FAB K-E mutations are predicted to slow but not eliminate actin polymerization (Breitsprecher et al., 2008; Schirenbeck et al., 2006; Applewhite et al., 2007; Hansen and Mullins, 2010). Co-expression of either DdVASP-∆tet or DdVASP-FAB K-E with DdMyo7 in vasp null cells cannot fully restore DdMyo7 cortical recruitment (Figure 5B,E). The monomeric VASP-∆tet mutant partially rescues filopodia formation while the FAB K-E mutant does not (Figure 5C). The residual filopodia forming activity of the VASP-∆tet monomer could be attributed to its remaining anti-capping activity that is lost in the F-actin binding mutant (Breitsprecher et al., 2008; Hansen and Mullins, 2010). Both mutants are predicted to lack any bundling activity and show decreased recruitment of DdMyo7 to the cortex. These data indicate that bundling linear F-actin at the membrane could promote DdMyo7 targeting.

DdMyo7 is enriched at the leading edge of cells, in contrast to the tail alone that localizes all around the cortex (Figure 1F,I). This difference suggests that head-tail autoinhibition restricts DdMyo7 localization. DdVASP is required for cortical localization but the apparent lack of interaction between DdMyo7 and DdVASP suggests an indirect role. Furthermore, the GFP-DdMyo7 tail domain in vasp null cells targets to the cortex efficiently (Figure 6A–C). This confirms that direct DdMyo7 tail - DdVASP binding is not required for localization to the cortex. Together these results suggested that VASP mediated actin networks release autoinhibition of DdMyo7. If so, then loss of autoinhibition regulation of DdMyo7 should eliminate the requirement of VASP for DdMyo7 to target to the cortex. To test this, two highly conserved charged residues at the extreme C-terminus of DdMyo7 (K2333, K2336) essential for head-tail autoinhibition were mutated (Yang et al., 2009; Petersen et al., 2016). Overexpression of this constitutively active mutant (DdMyo7-KKAA) stimulates filopodia formation and increases cortical localization in wild-type cells (Table 2, Petersen et al., 2016; Arthur et al., 2019). DdMyo7-KKAA was expressed in vasp null cells where it targets to the cortex (Figure 6A–B). The cortical targeting of DdMyo7-KKAA in vasp nulls is not as robust as seen in control cells (Figure 6C), and it is localized more uniformly around the cortex in vasp nulls (less enriched in the leading edge or pseudopod) compared to controls (Figure 6E). These observations are consistent with VASP-mediated actin assembly having a role in local activation and recruitment of DdMyo7 to the cortex.

The finding that loss of head-tail autoinhibition restores cortical recruitment of DdMyo7 raised the question of whether an autoactivated motor is sufficient to rescue the filopodia formation defect seen in the vasp null cells. Expression of DdMyo7-KKAA in vasp nulls did not rescue their filopodia formation defect (Figure 6D), indicating that a specific DdVASP activity is required for filopodia initiation. The requirement for both cortical DdVASP and DdMyo7 for filopodia formation was further tested by targeting of the myosin to the cortex or membrane by fusing a prenylation sequence to its C-terminus (DdMyo7-CAAX, adapted from Weeks et al., 1987). DdMyo7-CAAX was robustly localized to the cortex in myo7 null, wild type and vasp null cells (Figure 6A,B; Table 2). Expression of DdMyo7-CAAX in wildtype or myo7 nulls cells significantly stimulated filopodia formation (Figure 6A,D, Table 2). However, filopodia are not formed when DdMyo7-CAAX was expressed in vasp null cells (Figure 6A,D). These results establish that targeting DdMyo7 to the membrane alone is not sufficient for filopodia initiation and that DdVASP must also be present.

The data suggest that activation of autoinhibited DdMyo7 is promoted at the front of the cell in regions rich in newly polymerized F-actin. Actin binding by myosin involves conformational changes in the myosin head (Houdusse and Sweeney, 2016) and could destabilize the interactions between the motor head and the tail. To test this, two DdMyo7 mutants were designed, each with mutations in highly conserved connectors of the motor domain. Their functions are established by studies performed on the highly conserved motor domain of DdMyo2 (Sasaki et al., 2003; Friedman et al., 1998). The non-hydrolyzer mutant (E386V) binds ATP but cannot hydrolyze it, and thus stays in weak actin-binding state, while the uncoupler mutant (I426A) can undergo conformational changes allowing strong interactions with F-actin but these conformational changes are not transmitted to the lever arm to produce force (illustrated in Figure 7A).

The non-hydrolyzer (E386V) has a mutated glutamate in Switch II that is required to assist hydrolysis (Figure 7—figure supplement 1A, Friedman et al., 1998). The non-hydrolyzer failed to efficiently cosediment with the actin cytoskeleton upon centrifugation (Figure 7D, Figure 7—figure supplement 1B) indicating that it was indeed a weak actin binding mutant. The DdMyo7-E386V mutant does not target efficiently to the cell periphery (Figure 7B,E) or rescue filopodia formation (Figure 7F; Table 2). The lack of cortical localization by the non-hydrolyzer suggests that the tail remains bound to the motor, favoring an autoinhibited conformation, and is not free to bind to the cortex (see Figure 7A). To test this, the autoinhibition point mutations (KKAA) were introduced into the non-hydrolyzer to disrupt the head/tail interface. Blocking autoinhibition (KKAA) in the non-hydrolysis mutant (DdMyo7-E386A + KKAA) strikingly restores cortical targeting (Figure 7B,E; Table 2) but not filopodia formation (Figure 7F; Table 2).

A second motor mutation, I426A, resides in the relay helix and disrupts the interface with the converter that is critical to direct the swing of the lever arm during force generation. This mutant uncouples ATP hydrolysis from force generation (Figure 7—figure supplement 1A, Sasaki et al., 2003). Thus, the uncoupler (DdMyo7-I426A) undergoes actin-activated ATP hydrolysis, likely due to conformational changes directed from binding to actin, but it cannot exert force (illustrated in Figure 7A). DdMyo7-I426A cortical targeting is similar to what is seen for wild-type DdMyo7 (Figure 7B,E). This indicates that actin binding destabilizes the autoinhibited form in this mutant. DdMyo7-I426A fails to efficiently rescue the filopodia formation defect of myo7 null cells despite its targeting to the cortex (Figure 7F and Table 2), establishing that force generation by DdMyo7 is essential for filopodia initiation.

The uncoupler mutant motor is predicted to have normal actin binding and indeed its cortical asymmetry is similar to wild-type DdMyo7 (Figure 7B,C). In contrast, the motor of the non-hydrolyzer DdMyo7-E386A + KKAA mutant has weak actin binding so interaction with the cortex would be directed by the tail. As predicted, the DdMyo7-E386A + KKAA mutant is localized more uniformly around the cortex (low cortical SD, Figure 7B,C). These observations provide support for the model that while the tail aids overall cortical localization, motor-actin binding refines leading edge targeting of DdMyo7.

Disruption of the dynamic actin network at the cortex that occurs by loss of vasp (Figure 4D) or by treatment with anti-actin drugs such as cytoA or latA results in loss of cortical localization of DdMyo7 (Figure 2G,H). Dictyostelium VASP is a potent actin polymerase, its activity speeds actin elongation, bundles filaments and blocks capping proteins from binding to the growing ends of actin filaments (Breitsprecher et al., 2008). Both its actin polymerization and bundling activities could be important for DdMyo7 recruitment (Figure 5). In the absence of DdVASP, increased cortical actin generated by the formin dDia2 can also recruit and activate DdMyo7 (Figure 4B-E). However, recruitment is not restored by several other means of increasing actin polymerization, including generating actin waves, polymerizing actin with jasplakinolide or blocking capping protein by overexpression of V-1. Thus, DdVASP does not simply recruit DdMyo7 by generating new actin polymers, but rather it is the nature of the network that DdVASP builds, likely growing linear actin filaments that are bundled together in parallel, that is critical for activation of DdMyo7.

The actin cortex of vasp nulls is less dense than found for wild-type cells as revealed by phalloidin staining, suggesting that a robust leading edge actin network collapses without VASP (Figure 4D). Interestingly, evidence for VASP-dependent changes in the cortical actin network has recently been shown in B16 melanoma cells (Damiano-Guercio et al., 2020). Their cortical actin network is normally dense and with many filaments oriented perpendicular to the membrane but in the absence of VASP, the network is more disperse and actin filaments are oriented in shallower angles with respect to the membrane (Damiano-Guercio et al., 2020). Thus, the formation of a VASP-mediated actin network at the cortex is likely to be important in this context as well as in other cell types.

The motor domain of DdMyo7 and not the tail localizes this myosin to the dynamic leading edge of the cell (Figure 1). The tail alone is uniformly localized all around the cortex suggesting that it only provides general cortical targeting while the motor-FD is enriched at the front of the cell (Figure 1). Cortical targeting requires DdVASP or activated formin activity, suggesting an enrichment of parallel actin filaments in the cortex is needed for normal DdMyo7 activation (Figure 3D, Figure 4D). Mutation of DdMyo7 to lock the motor in a weak actin-binding state, even in the absence of autoinhibition such as in the case of the double non-hydrolyzer/autoinhibition mutant (E386A; KKAA), does not promote leading edge targeting. In contrast, a mutant that can bind actin but uncouples force generation from ATP hydrolysis (I426A) is localized more similarly to wild-type DdMyo7 (Figure 7A–C). Together these observations reveal that a functional motor that can hydrolyze ATP and bind actin is necessary for targeted, cortical recruitment of DdMyo7 to a DdVASP generated actin network, predominantly at the leading edge of the cell (Figure 7).

The motor is typically sequestered by interaction with the tail in the autoinhibited state, a widely used mechanism for controlling myosin activity and reducing unnecessary energy expenditure (Heissler and Sellers, 2016). DdMyo7 head/tail autoinhibition is mediated by charged residues in the FERM domain, similar to fly myosin 7a, and this regulates both cortical targeting and filopodia number (Petersen et al., 2016; Arthur et al., 2019; Yang et al., 2009). Preventing autoinhibition (KKAA) results in enhanced targeting to actin and, in the presence of DdVASP, filopodia formation (Petersen et al., 2016; Arthur et al., 2019; Figure 7B,C,E,F). It also bypasses the loss of cortical targeting observed in the absence of DdVASP, as the tail is now free to bind to any target sites in the cortex (Figure 6). The autoinhibited off state occurs when the equilibrium is biased toward the closed conformation. In this state, the motor has reduced affinity for actin and the partner binding sites in the tail are occluded. The results here lead to the following appealing model for how the motor specifies targeting of DdMyo7 to the leading edge (Figure 8). The compact, autoinhibited DdMyo7 freely diffuses throughout the cell. High local concentrations of actin in the cortical region facilitate interactions with actin, which destabilizes the inactive myosin, shifting the conformational equilibrium toward the open state. This makes the tail available to interact with partners and dimerize, and allows an active motor domain to engage with the actin network. If the motor does not dimerize, then DdMyo7 likely returns to a closed, off state, which diffuses and can be recycled back into the cytoplasm. It is tempting to speculate that the closely apposed actin filament bundles generated by VASP (or dDia2) favor stabilizing the open state because of the high local concentration of actin. This would promote dimerization, either because the tails could interact with each other more readily or because a dimerization partner is also present in the network. Once dimerized, this likely processive motor can productively engage with the cortex (e.g. generating force to align and/or bundle filaments) in collaboration with VASP to promote filopodia extension. Thus, motor domain mediated binding to a specific actin network is a key means of restricting localization of DdMyo7 during filopodia formation.

The mechanism of autoinhibition via motor-tail stabilizing interactions has been conserved throughout the evolution of myosins (Umeki et al., 2011; Sakai et al., 2011; Yang et al., 2009; Petersen et al., 2016; Weck et al., 2017; Heissler and Sellers, 2016). Mechanisms to overcome autoinhibition likely developed to restrict the recruitment and activation of myosins in time and space, allowing cells to fine-tune their activity. Here it is shown that DdMyo7 autoinhibition is relieved by the cortical actin network generated by DdVASP. While this differs from the PIP3-mediated recruitment of metazoan Myo10, once activated these two evolutionarily distant filopodia myosins use similar mechanisms to generate filopodia. They both dimerize upon recruitment to the actin-rich cortex (Arthur et al., 2019; Lu et al., 2012) where they likely contribute to the reorganization of the Arp2/3 branched actin network to orient actin filaments perpendicular to the membrane (Svitkina et al., 2003; Tokuo et al., 2007; Ropars et al., 2016; Arthur et al., 2019). Interestingly, both DdMyo7 and Myo10 work in cooperation with VASP, although they appear to do so in different ways. In the case of Myo10, it is seen to co-transport with VASP along the length of filopodia during extension and co-immunoprecipitate with VASP, although it is not known if the two proteins interact directly in vivo (Tokuo and Ikebe, 2004; Kerber et al., 2009; Lin et al., 2013). In contrast, the activity of DdVASP generates an actin-rich leading edge that recruits and activates DdMyo7 and there is no evidence that the two proteins interact with each other at present (Figures 1, 3 and 6, and Figure 3—figure supplement 1A).

Dictyostelium filopodia have actin filaments that are loosely bundled (Medalia et al., 2007). They lack an obvious fascin homologue, a bundling protein localized to filopodia in many cell types. However, there is a calcium-sensitive bundler ABP-34 with three actin binding sites, similar to fascin (Fechheimer, 1987; Lim et al., 1999). ABP-34 is selectively localized to filopodia and has been suggested to play a role in filopodia formation in Dictyostelium but its exact function remains unclear (Fechheimer, 1987; Rivero et al., 1996). In the absence of a clear filopodia bundling protein, it is possible that DdMyo7 dimers contribute to actin bundling.

Dictyostelium are highly motile cells with dynamic, short lived filopodia. Given the early origins of Amoebozoa, a simple mechanism of coupling motor activation to actin dynamics at the pseudopod suggests that amoeboid filopodia formation is driven by a minimal regulatory circuit. It is tempting to speculate that VASP-dependent recruitment of an MF myosin represents an early form of cooperation between these two proteins. Perhaps as filopodia played wider roles in development and migration as multicellularity evolved, a signaling-based mechanism of MF myosin recruitment emerged with PIP3 binding to the three PH domain motif of Myo10 that replaced the first MF domain in present in other MF myosins. Thus, once Myo10 activation became dependent on PIP3, its motor activity was then used to promote VASP transport and filopodia extension. The evolution of the functional relationship between VASP and MF myosin would be interesting to explore in organisms such as Drosophila that make filopodia but lack Myo10 (Drosophila instead have a Myo22 with two MF domains like DdMyo7) and in the earliest organisms that have Myo7, Myo22, and Myo10 such as the filasterean Capsaspora and choanoflagellate Salpingoeca (Kollmar and Mühlhausen, 2017). It is possible that VASP plays a role in the recruitment of Myo10 to initiation sites or its activation in some systems, but this remains to be determined. The development of genetic tools in several evolutionarily significant organisms such as Capsaspora and Salpingoeca (Parra-Acero et al., 2018; Booth et al., 2018; Booth and King, 2020), unicellular organisms at the onset of multicellularity, should now allow for the study of the evolution of the VASP-filopodial MF myosin relationship in the targeting and activation of filopodial myosins.

Dictyostelium control/wild-type (AX2 or AX3), myo7 null (HTD17-1) (Tuxworth et al., 2001), vasp null (Han et al., 2002), and dDia2 null (Schirenbeck et al., 2005) cells were cultured in HL5 media (Formedium). Transgenic lines were generated by methods similar to those described (Gaudet et al., 2007; Paschke et al., 2018). Briefly, axenic lines grown in HL5 cells were harvested, washed twice with ice cold H50 (20 mM HEPES, pH 7.0, 50 mM KCl, 10 mM NaCl, 1 mM MgSO₄, 5 mM NaHCO₃, 1 mM NaH₂PO₄ and flash spun at 8,000–10,000 X g in a microcentrifuge tube until the rotor reached speed). Cells were resuspended at 5e7 cells/mL and 100 µL of cells was combined with 10 or 15 µg DNA in a 0.1 cm gap cuvette. Cells were electroporated by pulsing twice, 5 s apart with a Bio-Rad Gene-Pulser set to 0.85 kV, 25 µF, and 200 Ω. Cells were recovered 10 min on ice and plated in a 10 cm dish for 24 hr before moving to selection media, either 10 µg/mL G418, 35 µg/mL HygromycinB or both.

Expression of the fusion proteins or null backgrounds was verified by western blotting using either anti-Myo7 (UMN87, Tuxworth et al., 2005), anti-GFP (Biolegend - B34), or anti-VASP (Breitsprecher et al., 2008) with anti-MyoB used as a loading control (Novak et al., 1995; Figure 1—figure supplement 2). Bands were detected using fluorescent secondary antibodies (LiCor) imaged with the LiCor Odyssey.

The GFP-DdMyo7 expression plasmid was created by fusing gfp to the 5' end of the myoi gene (Titus, 1999) (dictyBase DDB: G0274455; Dictybase.org [Fey et al., 2013]) then cloning it into the pDXA backbone with the actin-15 promotor and a NeoR cassette as described (Tuxworth et al., 2001). Plasmids used for the expression of the full length tail (aa 809 - end) (Tuxworth et al., 2001), the KKAA autoinhibition mutant (K2333A/K2336A) (Petersen et al., 2016), motor forced dimer (aa 1–1020 followed by the mouse Myo5A coiled coil region and a GCN4 leucine zipper) Arthur et al., 2019 have been described previously. An expression clone for the full-length GFP-DdMyo7 with a C-terminal prenylation site (CAAX) was generated using Q5 mutagenesis (New England Biolabs) to add codons encoding the CTLL* prenylation motif from Dictyostelium RasG (UNIPROT: P15064) to the 3' end of the myoi gene. A DdMyo7-mCherry expression plasmid was generated by first TA cloning a PCR product (myo42 to myi185 +2) encompassing the 3' region of the myoi gene (encoding aa 475 - end) minus the stop codon using StrataClone (Agilent) (pDTi289 +2). This fragment was then cloned into pDMCCherry, a modified pDM358 (Veltman et al., 2009) with the mCherry gene inserted for C-terminal fusions, generating pDTi299. The 5' end of myoi was then inserted by restriction cloning to create pDTi340. A motor mutant that cannot hydrolyze MgATP, the non-hydrolyzer E386V was designed based on a characterized Dictyostelium Myo2 mutant (Friedman et al., 1998). The combined non-hydrolyzer + KKAA was made by standard ligation cloning to introduce the motor domain sequence from the non-hydrolyzer mutant into KKAA full-length expression plasmid by restriction enzyme digest with BsiWI and BstEII. The uncoupler mutant (I426A) based on a characterized Dictyostelium Myo2 mutant (Sasaki et al., 2003) was cloned by Q5 mutagenesis. Fluorescent protein fusions of DdMyo7 were made using a combination of Q5 mutagenesis, Gibson assembly and restriction enzyme cloning. A DdMyo7-Scarlet I expression plasmid, pDTi517, was generated by first cloning a full-length myoi gene that has a BglII site at the 5' end, lacks its internal BglII site and also its 3' stop codon, pDTi515 +2, using a combination of Q5 mutagenesis, Gibson assembly and restriction enzyme cloning. The base Scarlet I-pDM304 expression plasmid for C-terminal fusions was generated by restriction enzyme cloning a codon-optimized synthesized Scarlet I gene (GenScript) into the extrachromosomal expression plasmid pDM304 (Veltman et al., 2009). The myoi gene was then cloned into mScarlet I-pDM304. Restriction cloning was used to create wild type and non-hydrolyzer mNeon DdMyo7 expression plasmids (pDTi516 and pDTi527, respectively). First, the base mNeon-pDM304 expression plasmid for N-terminal fusions was generated by restriction enzyme cloning a codon-optimized synthesized mNeon gene (GenScript) into the extrachromosomal expression plasmid pDM304 (Veltman et al., 2009). Then a full-length myoi gene that has a BglII site at the 5' end and lacks its internal Bgl II site was generated and cloned into either pDM448 (Veltman et al., 2009) for GFP-DdMyo7 expression (pDTi492) or mNeon-pDM304 for mNeon-DdMyo7 expression (pDTi516). Restriction cloning was used to exchange the 5' region of the gene carrying the E386V mutation into the wild type pDTi516 expression plasmid, creating pDTi527. The dDia2-CA mutant was created by first PCR-amplifying the forH gene (using forH4L and forH8 oligos) from GFP-dDia2 (Schirenbeck et al., 2005) then TA cloning the product using StrataClone (Agilent) to generate dDia2-SC. Mutations to generate a double R1035A, R1036A mutation were introduced dDia2-SC by Q5 mutagenesis (dDia2-CA SC). The wild type or mutant genes were restriction cloned into the extrachromosomal expression plasmid pDM449 (Veltman et al., 2009) to generate mRFPmars-dDia2-CA. An inducible V-1 expression plasmid was created by first PCR-amplifying the mtpn gene with V-1 F and V-1 R oligos (dictyBase:DDB_G0268038) using Ax2 genomic DNA then TA cloning the product using StrataClone (Agilent) to generateV-1 SC. The V-1 insert was then restriction cloned into the extrachromosomal expression plasmid pDM334 (Veltman et al., 2009) to generate GFP-inducible V-1. The sequence of all PCR generated clones was confirmed by Sanger sequencing (GeneWiz and University of Minnesota Genomics Center).

The GFP-VASP expression plasmid was a gift from Dr. Richard Firtel (UCSD) (Han et al., 2002). The VASP tetramer and FAB, Δtet mutants were not fused to GFP to avoid any steric hindrance with the fluorescent protein. The full-length VASP cDNA (dictyBase:DDB_G0289541) was cloned into the pDM344 shuttle vector (Veltman et al., 2009) and the NgoM-IV fragment from this plasmid was ligated into pDM358-mApple that has the mApple gene (Shaner et al., 2008) cloned in between the act6 promoter and act15 terminator of pDM358 (Veltman et al., 2009). VASP-∆tet was created by introducing a SmaI site and stop codon into the vasp gene, altering the coding sequence from 334 PSLSAPL to 334 PSLSAPG* using Q5 mutagenesis. The F-actin-binding mutant (FAB K-E) was based on mutating previously identified critical F-actin binding residues (K275, R276, K278, and K280); (Schirenbeck et al., 2006) to glutamic acid (Hansen and Mullins, 2010) by Q5 mutagenesis. The sequence of all PCR generated clones was confirmed by Sanger sequencing (University of Minnesota Genomics Center). Oligonucleotides used are in the key resource table.

Statistics were calculated in Prism 8 (GraphPad). One-way ANOVA analysis with post hoc Tukey test or Dunnett’s multiple comparison to wild-type control was used to compare groups; Student’s t test was used when only comparing two datasets. Statistical tests were calculated on full datasets, experimental means are shown on graphs to demonstrate experimental variability. Data points deemed definite outliers (0.1%) by Rout method were excluded for cortex:cytoplasm ratio. Error bars are SEM, unless noted. Significance differences are in comparison to control (DdMyo7), unless noted. Tables have filopodia number as average number of filopodia in cells with at least one filopod. SEM is standard error of the mean. Capitalized ‘N’ indicates number of experiments lowercase n is number of cells. Full statistical analysis is included in the source data tables accompanying each figure.