The protrusion of two distinct actin-containing organelles, lamellipodia and filopodia, is thought to be regulated by two parallel pathways: from Rac1 through Scar/WAVEs to lamellipodia, and from Cdc42 through N-WASP to filopodia. We tested this hypothesis in Drosophila, which contains a single gene for each WASP subfamilies, SCAR and WASp. We performed targeted depletion of SCAR or WASp by dsRNA-mediated interference in two Drosophila cultured cell lines expressing lamellipodial and filopodial protrusion. Knockdown was verified by laser capture microdissection and RT-PCR, as well as western blotting. Morphometrical, kinetic and electron microscopy analyses of the SCAR-depleted phenotype in both cell types revealed strong inhibition of lamellipodial formation and cell spreading, as expected. More importantly, filopodia formation was also strongly inhibited, which is not consistent with the parallel pathway hypothesis. By contrast, depletion of WASp did not produce any significant phenotype, except for a slight inhibition of spreading, showing that both lamellipodia and filopodia in Drosophila cells are regulated predominantly by SCAR. We propose a new, cascade pathway model of filopodia regulation in which SCAR signals to lamellipodia and then filopodia arise from lamellipodia in response to additional signal(s).
In eukaryotic cells, controlled actin polymerization underlies cell motility, contractility and maintenance of cell polarity. Protrusion of the leading edge represents the first locomotory response of a cell to external signals and occurs within two organelles: flat broad lamellipodia and thin long filopodia (Small et al., 2002). Lamellipodia and filopodia have a strikingly different organization of the actin polymerization machinery and they are controlled by different signaling pathways (Hall, 1998; Svitkina and Borisy, 1999a). In lamellipodia, short actin filaments form an extensively branched network (Svitkina et al., 1997; Svitkina and Borisy, 1999b). The critical element of network assembly is dendritic nucleation of actin filaments mediated by the Arp2/3 complex (Mullins et al., 1998). Newly nucleated filaments generate the pushing force that drives lamellipodial protrusion. Capping of growing filaments at their barbed ends after some period of elongation is required to maintain a steady-state of filament number (Borisy and Svitkina, 2000). In filopodia, long parallel actin filaments are tightly cross-linked to form a stiff bundle (Wood and Martin, 2002). Protrusion of filopodia occurs by elongation at barbed ends facing the membrane (Mallavarapu and Mitchison, 1999). The cross-linking protein fascin is considered to be the major bundler in filopodia (Kureishy et al., 2002), which cross-links filaments and makes them efficient at pushing. Ena/VASP (vasodilator-stimulated phosphoprotein) family proteins allow for continuous filament elongation by binding to growing barbed ends and protecting them from capping (Bear et al., 2002).
Small GTPases of the Rho family are key regulators of the actin cytoskeleton (Hall, 1998). Among them, Cdc42 and Rac1 are known to control filopodial protrusion and to regulate lamellipodia, respectively. The signaling pathways leading from these GTPases to actin polymerization are not completely established. However, critical roles belong to proteins of the WASP family, which activate the intrinsically inactive Arp2/3 complex (Machesky and Insall, 1998; Higgs and Pollard, 2001). At this key point, signaling meets the protrusive machinery. The mammalian genome contains at least five members of the WASP† family organized in two subfamilies: one containing the closely related N-WASP and WASP members and the other containing WAVE 1, 2 and 3, also known as Scar from its founding Dictyostelium orthologue (Bear et al., 1998). N-WASP (Rohatgi et al., 2000) and WASP (Higgs and Pollard, 2000) are activated by Cdc42, whereas Scar/WAVEs become activated downstream of Rac1 (Miki et al., 2000; Eden et al., 2002). Consistent with the specifics of their upstream GTPases, the downstream cellular effects of WASPs and Scar/WAVEs are thought to lead to the formation of filopodia and lamellipodia, respectively (Takenawa and Miki, 2001). Therefore, the current paradigm for actin-based protrusion is that Cdc42 activates N-WASP, which induces filopodia, whereas Rac1 activates WAVEs, which induce lamellipodia. This parallel pathway model predicts that in the absence of WAVEs, a cell would be able to form filopodia, but not lamellipodia, whereas removal of WASPs would have the converse effect.
Recent data, however, have challenged the idea of strict specialization of different WASP family members with respect to the type of induced protrusions. For example, cells derived from N-WASP knockout mice successfully formed filopodia (Snapper et al., 2001; Lommel et al., 2001). Also, some WAVEs were found to localize to filopodia, not only to lamellipodia (Nozumi et al., 2003). We have recently shown in mouse melanoma cells that filopodia are formed by reorganization of the lamellipodial dendritic network (Svitkina et al., 2003), suggesting that lamellipodia represent an important intermediate in filopodial formation. Finally, in vitro data do not support a model in which a lamellipodial or filopodial mode of actin organization is determined by specific activators of the Arp2/3 complex (Vignjevic et al., 2003). Taken together, these considerations suggest that signaling pathways leading to the formation of lamellipodia and filopodia may not be entirely parallel and independent.
In this study, we have tested the predictions of the parallel pathway model and have investigated the contribution of individual Arp2/3 activators to lamellipodial and filopodial protrusion in vivo. Because of the complexity of the mammalian genome, we addressed these issues in Drosophila, which has a single gene for each WASP subfamily, SCAR and WASp (Ben-Yaacov et al., 2001; Zallen et al., 2002). The roles of SCAR and WASp in lamellipodial and filopodial protrusion were analyzed using Drosophila cultured cell lines in which we performed targeted depletion of SCAR or WASp by the RNAi technique. Also, we characterized Drosophila culture cells at the electron microscopic level and showed that they are suitable for cell motility studies.
Our results suggest an alternative to the conventional parallel pathway hypothesis. We propose a new model, which we call the cascade pathway model. In the cascade mechanism, activation of any WASP family protein leading to stimulation of the Arp2/3 complex may induce formation of the dendritic network, which subsequently may be reorganized into filopodial bundles if additional signals are received.
Materials and Methods
Cell culture and RNAi
Drosophila permanent cell lines, S2R+ and Dm-BG2-c2 (referred to as BG2 throughout the paper), and culture conditions have been described previously (Ui et al., 1994; Yanagawa et al., 1998). Cells were split every 4-5 days. RNAi was performed according to the procedure of Clemens (Clemens et al., 2000). The following sequences were used for amplification of ∼500 bp fragments for the target proteins. SCAR (CG4636, FBan0004636): 5′-CAGTCGATGA GGACGCACTA-3′; 5′-GTCTGTTCAG CTTCTTGCCC-3′. WASp (CG1520, FBan0001520): 5′-TCATTAAATC GCTGCGTGAG-3′; 5′-AGAGCAACA ATGTCCTGGCT-3′.
WASp rabbit polyclonal antibodies were raised against a peptide NNAKDKKRKV TKADISRPTN, corresponding to Dm WASp (protein id is NP 788755.1 in NCBI) amino acids 220-239. GP12 SCAR antibody was a generous gift from Jen Zallen (Princeton University, Princeton, NJ). 5G2 antibody to Enabled was from Hybridoma Bank, Iowa. Arp3 antibody was a gift from William Theurkauf (University of Massachusetts, Worchester, MA). Monoclonal α-tubulin (clone B-5-1-2) antibody, polyclonal anti-actin A5060 and rhodamine-labeled secondary antibody were from Sigma. HRP-labeled goat anti-mouse and goat anti-rabbit antibodies were from Jackson or KPL. AlexaFluor 488-phalloidin was from Molecular Probes.
Light microscopy was performed using an inverted microscope (Nikon Eclipse) equipped with a Plan 100×, 1.3 NA or a 40×, 1.3 NA objective and a back-thinned, cooled CCD camera (model CH250; Roper Scientific) driven by MetaMorph® imaging software (Universal Imaging Corp.). For morphometrical analyses, cell projected area (A) and perimeter (P) were determined using automated functions in MetaMorph® software. Ratio P/A1/2 served as an index of shape irregularity. Data from experiments were analyzed by using a Student's t-test. Presented values are expressed as mean ± s.d. The number of cells counted is indicated in each figure legend. Statistical significance was accepted at P<0.05.
For fluorescence microscopy, different conditions were used depending on the particular antibody. For SCAR-specific staining cells were fixed with 4% formaldehyde in cytoskeleton buffer (10 mM PIPES buffer pH 6.1, 138 mM KCl, 3 mM MgCl2, 2 mM EGTA, 320 mM sucrose) for 20 minutes, washed with the same buffer and then with PBS, and permeabilized with 0.1% Triton X-100 in PBS for 5 minutes. For Arp3-specific staining, cells were extracted for 3-5 minutes with 1% NP-40 (Pierce) in PEM buffer (100 mM PIPES, pH 6.9, 1 mM MgCl2 and 1 mM EGTA), containing 4% polyethylene glycol 40,000 (SERVA), protease inhibitors cocktail (Sigma-Aldrich) and 2 μM phalloidin (Sigma-Aldrich) followed by fixation with 0.2% glutaraldehyde in 0.1 M Na-cacodylate (pH 7.3) and quenching with 1 mg/ml NaBH4. For Ena staining, fixation with 0.25% glutaraldehyde, 0.5% NP-40 and 10 μm phalloidin in PEM buffer (pH 6.9) was followed by a 20 minute quenching with 1 mg/ml NaBH4. For all protocols, after blocking for 1 hour in 1% glycine in PBS with 1% Triton X-100 or 5% BSA in PBS, cells were incubated with primary antibody to SCAR (GP12, 1:50), Ena (5G2, 1 μg/ml) or Arp3 (1:25) for 1 hour followed by rhodamine-labeled secondary antibody (1:100) for 40 minutes. For phalloidin staining, AlexaFluor 488-labeled phalloidin was added to the secondary antibody. FITC and Cy3 filter sets were used for imaging of AlexaFluor 488-phalloidin or rhodamine-labeled antibody, respectively. To measure levels of F-actin, cells were fixed with 0.2% glutaraldehyde without extraction and stained with AlexaFluor 488-phalloidin. Fluorescence was integrated over individual cells using MetaMorph® software and expressed as a number in arbitrary units per cell.
For platinum replica electron microscopy, cells on glass coverslips were extracted as described for Arp3 staining above, then washed twice with PEM, fixed with 2% glutaraldehyde and processed for electron microscopy as described previously (Svitkina and Borisy, 1998).
Laser Capture Microdissection (LCM) and RT-PCR
LCM technology (Arcturus) was used to capture cells with the desired morphology from a heterogeneous population. A CapSure™ cap bearing a transparent polymer film was positioned over the microdissection target area. Low-energy laser pulse causes the film to melt, extend and adhere to individual target cells. Lifting the cap separates targeted cells from unselected population. Cells were grown on microscope slides, fixed with 70% ethanol, dehydrated according to the Arcturus instructions and captured onto CapSure™ caps using the following settings: laser power = 40 mW, duration = 1 ms. 100 cells were captured for each sample. PicoPure RNA isolation kit (Arcturus) was used to isolate and purify total RNA. Regular conditions and a SuperScript First-Strand Synthesis kit (Invitrogen) were used for RT-PCR. Primers for PCR reaction were as follows: SCAR forward 1020-1039: ATACACGTCC GCCACGTCCT; SCAR reverse 1561-1580: TTGCCCGTCA TACCGATGCT. r49 forward 42-61: GCGCACCAAG CACTTCATCC; r49 reverse 362-381: GCGACCGTTG GGGTTGGTGA. Sequence used for rp49 is: NM 170461, [gi:24651275] in NCBI database.
Immunoblotting and densitometry analysis
Regular conditions were used for SDS-PAGE and immunoblotting. Antibodies to SCAR and to WASp were used in 1:1000 dilutions overnight. Actin-specific antibody was diluted 1:250. Densitometry analysis was performed using NIH Image software. Representative blots obtained in at least three independent experiments with similar results are presented.
Drosophila cell lines as a model system for motility
Because Drosophila cells were not previously used as a model system for in vitro motility, we first characterized the protrusive behavior and structural organization of lamellipodia and filopodia in two cultured cell lines, S2R+ and Dm-BG2-c2 (hereafter referred to as BG2). The S2R+ cell line is a derivative of the S2 line expressing a wingless pathway receptor (Yanagawa et al., 1998). These cells are thought to be derived from embryonic hemocytes, although the exact origin is unknown. BG2 cells are thought to be cells of neuronal origin (Ui et al., 1994). Both types of cells are able to attach and spread on the substratum.
Although we found the populations of S2R+ cells to be rather heterogeneous, the dominant morphology was large flat cells with broad and thin circumferential lamellipodia, which had a rather smooth outline (Fig. 1A). Kinetic analysis of cell motility in culture revealed that lamellipodia of these cells displayed constant protrusive-withdrawal activity (Fig. 1B), but cell translocation was not very efficient, presumably because of lack of cell polarity. Staining with fluorescently labeled phalloidin suggested that lamellipodia of S2R+ cells were filled with a dense actin network (Fig. 1C), which constituted the major part of the actin cytoskeleton in these cells. Platinum replica electron microscopy (EM) confirmed this impression. It showed that lamellipodia were filled with a dense highly branched actin network consisting of relatively short filaments (Fig. 1D). Filopodial-like bundles were rare in S2R+ cells, and the small number observed consisted of a few relatively long loosely bundled filaments (data not shown).
BG2 cells were rather small in size and had polar morphology, in that flat and thin lamellar regions were not present all around the cell perimeter (Fig. 2A). Kinetically, BG2 cells displayed efficient locomotion over the substratum. At the leading edge active lamellipodia and filopodia were observed, with filopodia being particularly prominent (Fig. 2B). Phalloidin staining (Fig. 2C) revealed bright actin fluorescence in filopodia and in numerous stress-fibers in the cell interior; weaker diffuse fluorescence was present in lamellipodia. EM analysis showed that BG2 lamellipodia contained a branched actin network, whereas the abundant filopodia contained classical tight bundles of long actin filaments (Fig. 2D).
Thus, we showed that both cell lines display actin-based protrusive activity similar to that described in other animal cells. We found that S2R+ cells use a predominantly lamellipodial type of protrusion, whereas BG2 cells are rich in filopodia. This makes these two cell lines good models in which to study the roles of SCAR and WASp in lamellipodial and filopodial protrusion. We first present results on SCAR knockdown in S2R+ and BG2 cells and then on WASp knockdown.
SCAR knockdown in S2R+ cells
S2R+ cells cultured in the presence of SCAR-specific dsRNA (dsSCAR) developed a distinct phenotype which became especially obvious by day 3-4. The phenotype was characterized by decreased spreading and formation of a highly irregular shape (Fig. 3A). To test whether SCAR was specifically depleted by the knockdown procedure, we analyzed the levels of SCAR protein and mRNA after RNAi treatment for the entire cell population (by western blotting) and for individual cells (by RT-PCR and immunostaining).
In immunoblots, Drosophila SCAR-specific antibody recognized a ∼78 kDa protein in extracts prepared from Drosophila S2R+ cells (Fig. 3B, lane 0). After SCAR RNAi, band staining gradually decreased over time (Fig. 3B, lanes 1-4), reaching >95% depletion by day 4, as determined by densitometry. WASp levels were not affected (Fig. 3B). The reduction of SCAR protein was reversible, as the protein level was restored when SCAR-depleted cells were re-plated and cultured for an additional 4-5 days in the absence of dsRNA (Fig. 3B, lane 4+5 WO). Concurrently, cell shape returned to normal (not shown).
Although the penetration of the SCAR RNAi phenotype was high, the development of an abnormal morphology was asynchronous, leading to a certain heterogeneity in the morphological phenotype at any given time point, especially at early stages. To verify that the cells with a profound morphological defect lack SCAR, we analyzed SCAR mRNA levels in these cells. For this purpose we used laser capture microdissection technique (LCM, Arcturus), which allows cells with the desired morphology to be captured on a special cap and separated from a heterogeneous population. Target cells were subjected to RNA extraction and RT-PCR. We used the LCM technique to capture 100 control cells and 100 cells subjected to SCAR RNAi for 4 days. Well-spread cells with prominent lamellipodia were chosen from the control population, and cells showing pronounced phenotype with numerous spiky protrusions similar to that shown in Fig. 3A were selected from the dsSCAR-treated population. Total RNA was isolated from these two subpopulations and subjected to RT-PCR. From densitometry analysis of samples after 30 PCR cycles, we estimated that SCAR mRNA levels were reduced by >95% in dsSCAR-treated cells (Fig. 3C). By contrast, RT-PCR of the same samples for ribosomal protein rp49 mRNA did not show any significant difference. These results confirm that SCAR mRNA is specifically depleted in cells exhibiting the characteristic SCAR RNAi phenotype.
Immunofluorescence staining for SCAR protein was used as an additional assay to determine the levels of SCAR in cells with a specific phenotype (Fig. 3D). In control cells, SCAR was enriched at the leading edge of lamellipodia. After dsSCAR treatment, specific staining in peripheral protrusions was abolished, consistent with the expected depletion of SCAR protein. In all samples, a nonspecific background staining was observed in the center of cells because of the low affinity of the available antibody.
SCAR depletion inhibits lamellipodia formation in S2R+ cells
Phenotype development during SCAR depletion was monitored over a period of 7 days. During this time, cells became more compact and developed a highly irregular shape, displaying long narrow processes (Fig. 3A and Fig. 4B,C). Although the pace and extent of phenotype development varied among cells in the population, the penetration of the phenotype was essentially 100%.
To quantify the reduction in cell spreading, we compared the projected area of control and SCAR-depleted cells (Fig. 4A, left). After 4 days of SCAR RNAi, the cell projected area decreased by an average of >75%. The irregularity of cell shape was evaluated using a dimension-less index – namely, the ratio of perimeter (P) over the square root of projected area (A) (Fig. 4A, right). For a circle, this index has a minimal value equal to 3.56 [P/A1/2=2πR/(πR2)1/2=2π1/2=3.56]. The value of the index grows with the irregularity of cell shape. Control S2R+ cells had an average P/A1/2 value close to 4, indicating that their normal cell shape intrinsically was highly regular and circular. Depletion of SCAR led to a fourfold increase in the value of P/A1/2 index, reflecting the development of a highly irregular cell shape.
The decreased cell spreading and the irregular shape suggested an impaired protrusive behavior in SCAR-depleted cells. To test how the SCAR phenotype arose, we performed a kinetic analysis of leading edge activity by time-lapse phase contrast microscopy at different levels of SCAR depletion – namely, after 3, 4 and 5 days of SCAR RNAi. At 3 days of treatment, cells still were able to protrude lamellipodia, although their activity was more erratic compared with untreated cells, and lamellipodia protruded with discrete protuberances instead of a smooth regular outline (Fig. 4B). At 4-5 days treatment, cells were unable to form lamellipodia; instead, they protruded long narrow processes, which expressed dynamics at their tips, as well as along their length (Fig. 4C). Over time, such discontinuous protrusion, in combination with retraction elsewhere along the cell perimeter, led to the formation of long processes, which although resembled filopodia superficially, were both curved and flaccid. With continued depletion, cells were not able to form any protrusions and detached from the substratum.
The inefficient protrusive activity of SCAR-depleted cells suggested an impairment in actin cytoskeleton assembly. Therefore, we analyzed actin organization by phalloidin staining (Fig. 3D) and EM (Fig. 4D,E). After 4 days of dsSCAR treatment, the level of phalloidin staining at the cell periphery was markedly reduced (Fig. 3D). Average F-actin content per cell, as estimated by quantification of phalloidin fluorescence, dropped approximately eightfold in SCAR-depleted cells (from 50.4±35.2, n=20 to 5.8±3.5 arbitrary units (AU), n=34), which is consistent with the idea that the primary pathway of actin polymerization is through SCAR-activated, Arp2/3-mediated nucleation. Levels of total actin as shown by western blotting also decreased but only by 1.6-3-fold (n=3) (blot not shown). Because G-actin has been shown to provide autoinhibitory regulation of actin expression (Lyubimova et al., 1997), the lower total actin levels are probably a secondary effect of SCAR depletion because of net actin depolymerization and transient increase of G-actin levels.
By EM, residual lamellipodia of SCAR-depleted cells contained very sparse filament networks that retained dendritic features (Fig. 4D). In the narrow processes, actin filaments were also organized into a branched network of rather short actin filaments (Fig. 4E), but not into parallel bundles as is characteristic of filopodia. Occasionally, microtubules penetrated into some processes from the cell interior, possibly because of low density of actin network there. These structural data indicate that the processes, although superficially looking like filopodia, may represent residual protrusive activity by the dendritic nucleation mechanism.
To test this idea, we examined the distribution of lamellipodial and filopodial markers in the peripheral narrow processes (Fig. 5). The lamellipodial marker Arp3 was enriched in the lamellipodia of control S2R+ cells (Fig. 5A) and was also detected at much lower levels in processes of SCAR-depleted cells (Fig. 5B); this was consistent with the EM data, which showed sparse dendritic network in these structures. We used Ena as a filopodial marker because its mammalian homologues Mena (Lanier et al., 1999) and VASP (Rottner et al., 1999) are highly enriched at filopodial tips and are present – although less concentrated – at the lamellipodial edge (Rottner et al., 1999). In Drosophila cells, Ena formed a thin punctate line at the very leading edge of lamellipodia, but could not be detected at the tips of processes in SCAR-depleted cells (Fig. 5D). This result strongly indicates that narrow processes of SCAR-depleted cells are not filopodia, because normal filopodia in Drosophila BG2 cells display bright Ena staining at their tips (not shown).
Thus, we found that depletion of SCAR in S2R+ cells resulted in progressive inhibition of cell spreading, lamellipodial protrusion and assembly of the actin network.
SCAR depletion inhibits lamellipodia and filopodia formation in BG2 cells
Immunoblotting data showed that SCAR levels in BG2 cell extracts were reduced by 90-95% following RNAi, whereas WASp was unaffected (Fig. 6A). The morphological phenotype of SCAR-depleted BG2 cells resembled that of S2R+ cells, although it appeared less dramatic (Fig. 6B). Cell spreading was decreased by ∼45%, and the P/A1/2 index increased by ∼40% (Fig. 6C). The increase in the irregularity of shape reflected the formation of long narrow processes.
Fluorescence microscopy using phalloidin showed that actin filament levels were markedly reduced in dsSCAR-treated cells, although some stress fibers in the cell interior remained (Fig. 6B). EM analysis revealed that neither normal lamellipodia nor filopodia were present in SCAR-depleted cells (Fig. 6D). The cell processes usually contained distal patches of a sparse dendritic network, presumably representing residual lamellipodial-like protrusions, which were associated with more central actin filament bundles.
These bundles probably corresponded to stress fibers, but they also might represent modified filopodial bundles and thus point to sustained filopodia formation in SCAR-depleted cells.
To evaluate this latter possibility, we analyzed de novo filopodia formation in SCAR-depleted BG2 cells. Untreated BG2 cells and cells subjected to RNAi for 4 days were replated and allowed to spread for 30-120 minutes. Untreated cells expressed intense filopodial formation, whereas dsSCAR-treated cells formed very few filopodia during spreading (Fig. 6E). To quantify the rate of filopodia formation, we counted the number of newly formed filopodia in 2 minute time-lapse sequences of individual cells and found that de novo filopodia formation was reduced approximately eightfold in SCAR-depleted BG2 cells (Fig. 6F). Thus, depletion of SCAR in BG2 cells inhibited the formation of both lamellipodia and filopodia, which led to reduced leading edge protrusion and cell spreading.
WASp knockdown in S2R+ and BG2 cells
As a tool for this study, we generated a polyclonal antibody against a peptide of WASp (see Materials and Methods). The affinity purified antibody recognized a protein with a molecular weight of ∼65 kDa in extracts from Drosophila S2R+ or BG2 cells (Fig. 7A). Four to eight days following the treatment of cells with WASp-specific dsRNA (dsWASp), the intensity of the immunoreactive band was reduced to 25-30% of the control level (Fig. 7A). However, this band remained unaltered in cells treated for 4 days with dsSCAR (Fig. 3B and Fig. 6A). From these results, we conclude that our antibody specifically recognizes Drosophila WASp and that the RNAi treatment was effective in eliminating ∼70-75% of WASp protein from both S2R+ and BG2 cells. Unfortunately, the antibody was not efficient for immunostaining so we were not able to determine WASp localization in cells.
In contrast to results with SCAR depletion, we were unable to identify a phenotype following WASp depletion in both cell lines tested. By live-cell imaging and phalloidin staining, WASp-depleted cells were essentially similar to respective nontreated controls (Fig. 7B,C; compare with Fig. 1A and Fig. 2A). One difference was a decrease in the projected cell area, by ∼45% in S2R+ and ∼30% in BG2 cells (Fig. 7D), although the P/A1/2 index, and therefore cell shape, did not change. The EM analysis also did not reveal significant structural changes in WASp-depleted cells, except slightly longer filaments and a little more `translucent' actin meshwork in S2R+ cells (Fig. 7E). In addition, filopodial formation was unaffected in spread BG2 cells (Fig. 7F) or in the filopodia re-formation assay (Fig. 6F). We conclude that WASp makes only a minor contribution to lamellipodial and filopodial protrusion in both S2R+ and BG2 cells.
In this study we investigated the relative contributions of two key regulators, SCAR and WASp, in actin-dependent protrusion in order to distinguish between two models of regulation of filopodia and lamellipodia: the conventional model, which predicts parallel regulatory pathways from N-WASP to filopodia and from Scar/WAVEs to lamellipodia (Takenawa and Miki, 2001), and an alternative model suggesting cascade regulation, in which WASP and/or Scar/WAVEs initiate lamellipodia, which are subsequently transformed into filopodia by additional signals (Fig. 8A,B). Our results do not support the parallel pathway model, but are consistent with lamellipodia functioning as a necessary preliminary step in the formation of filopodia.
Drosophila cells as a model system for investigating roles of WASP family proteins in protrusion
The Drosophila genome contains a single orthologue for each subfamily of WASP family proteins, WASp and SCAR (Ben-Yaacov et al., 2001; Zallen et al., 2002). By analogy with mammalian cells, one might assume that in Drosophila WASp and SCAR will regulate filopodial and lamellipodial protrusion, respectively. During development, Drosophila WASp and SCAR perform distinct, nonoverlapping functions upstream of the Arp2/3 complex (Ben-Yaacov et al., 2001; Hudson and Cooley, 2002; Tal et al., 2002; Zallen et al., 2002), but it remains unclear whether SCAR and WASp are specialized at the subcellular level with respect to the mode of actin polymerization they induce.
We chose two adherent Drosophila cell lines as potential model systems for in vitro motility to analyze the protrusive behavior of these cells. When plated onto glass or plastic, both cell types spread to form protrusions which structurally and kinetically are similar to those observed in other motile cells. As in mammalian cells, protrusions were filled with actin filaments organized into dendritic networks in lamellipodia or into parallel bundles in filopodia. On the basis of these data, we conclude that Drosophila cell lines display conventional actin-based protrusive behavior and can be used for genetic analysis, which may give insight into mechanisms of cell motility in general. The availability of cell lines that express either lamellipodia (S2R+) or lamellipodia and filopodia (BG2) allowed for the possibility of determining the pathways for these two distinctive actin protrusions.
SCAR is the major regulator of lamellipodia in S2R+ cells
We found that depletion of SCAR from S2R+ cells correlated with the development of a strong, penetrant and highly reproducible phenotype, which reflected the declining ability of cells to form lamellipodia and spread over the substratum. Although some heterogeneity inevitably accompanied the development of the phenotype (probably due to uneven dsRNA uptake), the progression clearly occurred from the appearance of slight irregularity of the lamellipodial outline to the formation of long narrow processes to virtual failure of a cell to spread.
The formation of linear processes reminiscent of filopodia during SCAR depletion could have resulted from the undepleted activity of WASp leading to the induction of filopodia, as is predicted by the model of two parallel signaling pathways. However, a detailed analysis of these aberrant processes showed that they are more like extremely narrowed residual lamellipodia than filopodia. The curved shape, uneven thickness and kinetic behavior of these processes were distinct from those seen during normal filopodial protrusion. Structurally, the processes of SCAR-depleted cells did not contain bundles of long parallel filaments that are characteristic of filopodia. Instead, they were filled with a sparse dendritic network like that seen in lamellipodia. The weak presence of the lamelipodial marker Arp3 and absence of the filopodial marker Ena in these processes also suggests that they are not filopodia. Thus, the depletion of SCAR in S2R+ cells did not induce filopodia, but produced narrow aberrant processes, probably by a combination of residual SCAR activity and retraction elsewhere along the cell perimeter. Our findings in Drosophila cells are in agreement with the previous work in Dictyostelium (Bear, 1998), where SCAR-deficient cells manifested abnormal cell size and morphology, and instead of normal broad lamellipodia formed `pseudopodia-like extensions' depleted of F-actin.
The formation of narrow lamellipodia following depletion of SCAR provides an additional insight into the function of SCAR. Scar/WAVEs in mammalian (Hahne et al., 2001; Nozumi et al., 2003) and Drosophila cells (this study) are localized to the extreme edge of protruding lamellipodia. Possibly, SCAR acts cooperatively in the process of protrusion. That is, protrusion at the leading edge may be more probable at sites of a pre-existing protrusion. A consequence of SCAR depletion would be that protrusion would become progressively restricted to smaller domains.
In contrast to the striking effect of SCAR depletion, WASp depletion did not generate a phenotype in S2R+ cells aside from a minor, albeit statistically significant, reduction in the cell projected area. One possible explanation for the lack of a significant phenotype might be the incomplete knockdown of WASp, as we were able to achieve only ∼75% depletion of WASp compared with >95% depletion of SCAR. However, considering that we used several diverse assays for phenotypic analysis, we believe that if WASp made a significant contribution to protrusive behavior, a phenotype would have been detectable.
Taken together, our results using S2R+ cells indicate that SCAR is the major regulator of protrusive activity in these cells. It is necessary for the assembly of the actin dendritic network, formation of lamellipodia and cell spreading onto the substratum. By contrast, WASp makes little, if any, contribution to actin-based protrusion in these cells.
SCAR is the major regulator of lamellipodia and filopodia in BG2 cells
The lack of filopodia formation in S2R+ cells after SCAR inhibition may be explained by an intrinsic deficiency of the filopodial machinery in these cells. Therefore, to test the role of SCAR in filopodial formation we switched to BG2 cells, which, in addition to lamellipodia, are very rich in filopodia, similar to other cells of neuronal origin.
The parallel signaling hypothesis predicts that depletion of the lamellipodial regulator, SCAR, will not affect filopodia, which, according to the model, would be induced by WASp. By contrast, the cascade model predicts that SCAR depletion will inhibit not only lamellipodia, but also filopodia. Our results support the latter prediction. We confirmed that SCAR in BG2 cells, like in S2R+ cells, regulates lamellipodia. Indeed, depletion of SCAR in BG2 cells inhibited lamellipodia, leading to decreased spreading and altered shape similar to results obtained with S2R+ cells. By EM criteria, the lamellipodial dendritic network in BG2 cells was significantly reduced; only a few small patches of the dendritic network remained at some cell edges.
The conclusive evidence allowing us to distinguish between two alternative models of regulation was expected from effects of SCAR depletion on filopodia formation. Our results clearly show that SCAR RNAi in BG2 cells inhibits not only lamellipodia, but also filopodia. Although light microscopic evaluation of static cultures per se gave an impression of filopodia sustaining in SCAR-depleted BG2 cells, more detailed analyses, namely, EM structural assay and a quantitative kinetic analysis of filopodial initiation during cell spreading, revealed that filopodia formation, in fact, was strongly inhibited. These results allowed us to rule out the parallel pathway model and validate the cascade pathway model for actin-based protrusions in Drosophila cell lines.
In contrast to SCAR, depletion of WASp in BG2 cells did not result in any changes in cell morphology, kinetics of protrusion or cytoskeletal architecture except for a slight inhibition of spreading, similar to what we observed in S2R+ cells. BG2 cells after WASp depletion retained not only lamellipodia, but also numerous perfectly looking filopodia indistinguishable from those in control cells. These results show that WASp does not play a distinctive role for filopodia formation, thus further corroborating the priority of the cascade pathway model over the parallel pathway model.
Regulation of lamellipodia and filopodia
The requirement of SCAR activity in Drosophila cells for both lamellipodial and filopodial protrusion is fully consistent with the idea that filopodial initiation in motile cells occurs downstream of lamellipodial protrusion, as suggested by our previous studies (Svitkina et al., 2003; Vignjevic et al., 2003). Therefore, we propose a model stipulating a two-step cascade regulation of filopodial protrusion in Drosophila cells (Fig. 8C). SCAR activates the Arp2/3 complex which then induces assembly of the dendritic network in lamellipodia. The lamellipodium by itself represents an important part of the protrusive machinery, but we suggest that it also provides a foundation for subsequent reorganization into filopodial bundles. Such reorganization would be dependent on additional as-yet-unspecified signals. Indirect evidence in support of the cascade regulation model comes from in vivo studies of Drosophila. In mutations removing all three Drosophila isoforms of Rac, both lamellipodia and filopodia were abolished during dorsal closure (Hakeda-Suzuki et al., 2002). Such a result would not be predicted if Rac signaled only to lamellipodia and a parallel pathway existed for filopodia.
Cascade signaling may also underlie filopodia formation in mammalian cells, even though they have a more complex complement of WASP family proteins. Because all WASP family members are potent activators of the Arp2/3 complex, any one of them may potentially contribute to the first step of the cascade, the formation of the dendritic network. However, the unimpaired filopodial formation in N-WASP knockout cells (Snapper et al., 2001; Lommel et al., 2001) suggests that the contribution of N-WASP to formation of the dendritic network is unlikely to be essential. The expression of mammalian WASP is limited to cells of hematopoietic origin. Interestingly, WASP-deficient platelets from Wiscott-Aldrich syndrome patients were able to form lamellipodia and filopodia (Falet et al., 2002), suggesting that WASP does not play an essential role in the protrusive activity of crawling cells in the hematopoietic lineage. On the basis of these considerations, we predict that the Scar/WAVE subfamily may play an essential role for filopodial formation in mammalian cells similar to that observed in Drosophila cells.
The functional roles of WASP subfamily proteins remain to be clarified. Our data indicate that WASp in Drosophila cells is dispensable for protrusive motility – both lamellipodial and filopodial. Then, what is WASp doing? As potent activators of the Arp2/3 complex, WASP subfamily members are likely to play a role in the formation of some dendritic actin arrays. The Arp2/3 complex and other proteins of the actin machinery are involved in processes other than lamellipodial protrusion, including endocytosis (Schafer, 2002; Qualmann et al., 2002), formation of certain junctions, such as podosomes (Linder et al., 2000), and establishment of immunological synapses (Thrasher, 2002). It is possible that the regulation of specialized dendritic arrays relies on WASP subfamily members.
In general, our results and interpretations of the roles of SCAR and WASp at the cellular level are consistent with conclusions drawn from studies on Drosophila development in vivo: SCAR mutations resulted in more severe phenotypes and affected many structures, whereas mutations in WASp had very little phenotype (Ben-Yaacov et al., 2001; Zallen et al., 2002; Hudson and Cooley, 2002; Tal et al., 2002). SCAR, rather than WASp, was the major mediator of Arp2/3 function during Drosophila oogenesis and in the development of the central nervous system (Zallen et al., 2002). SCAR and Arp2/3 were essential for proper spatial distribution of nuclei during cortical divisions, as well as for metaphase furrow formation and ring canal formation. Overall actin levels were lower in SCAR and Arp2/3 mutant cells. By contrast, WASp function seemed to be restricted to specific processes in development, such as the proper execution of asymmetrical cell divisions in neural lineages (Ben-Yaacov et al., 2001) or in bristle formation in ommatidia (Zallen et al., 2002). Nevertheless, WASp performed its role via the Arp2/3 complex (Tal et al., 2002). Although WASp did not seem to contribute to protrusive activity in the Drosophila cell lines tested in our experiments, consideration of mutant developmental phenotypes suggests that WASp can activate the Arp2/3 complex in certain cellular contexts and induce downsteam events that are dependent on dendritic nucleation of actin filaments.
If WASP subfamily members are not essential for filopodia formation, then what are the signaling molecules for the second step of our postulated cascade pathway? We have suggested recently that a key element driving network reorganization into filopodial bundles is a multiprotein complex at filopodial tips (Svitkina et al., 2003). We propose that the tip complex is a site at which the second step of the cascade signaling is directed. In favor of this hypothesis, IRSp53, an effector of Cdc42, induces filopodia by recruiting Mena (Krugmann et al., 2001), a protein which localizes to the filopodial tips (Lanier et al., 1999) and promotes the elongation of actin filaments by protecting them from capping (Bear et al., 2002). Also, a formin family protein Drf3 has recently been shown to be a downstream effector of Cdc42, localize to filopodial tips and play a role in filopodia formation (Peng et al., 2003). The IRSp53 or Drf3 pathway or an as-yet-unidentified pathway, rather than N-WASP, may be responsible for Cdc42 signaling to filopodia. Irrespective of the specific molecules involved in the signaling pathway, our structural-kinetic analyses suggest that they share the property of targeting the filopodial tip complex to drive filopodial formation.
We thank John Peloquin for preparing antibodies to WASp, and Dr Jen Zallen and Dr William Theurkauf for gifts of the SCAR and the Arp3 antibody. The 5G2 monoclonal antibody to Enabled developed by Dr Corey Goodman was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. Supported by grant NIH GM 62431 to G.G.B.
↵† WASP/WASPs indicate mammalian proteins, whereas WASp is used for the Drosophila protein.
Note added in proof
While this paper was in review, two relevant studies consistent with our findings on the role of SCAR were published. Rogers and co-workers recognized SCAR as a major regulator of lamella formation in Drosophila S2 cells in a screen of proteins implicated in actin function, and gave insights on regulation of the SCAR complex (Rogers et al., 2003). Kunda et al., focused on the role of Sra-1, kette, and Abi in stability, localization, and function of the SCAR complex downstream of GTPase signaling (Kunda et al., 2003).
- Accepted October 3, 2003.
- © The Company of Biologists Limited 2004