SCAR is a primary regulator of Arp2/3-dependent morphological events in Drosophila

The spatially regulated polymerization of actin monomers into filaments provides a widely used mechanism underlying cell morphology and motility. During cell migration and axon extension, localized actin filament assembly occurs at the advancing cell surface (Borisy and Svitkina, 2000). Through a related mechanism, actin polymerization propels the movement of intracellular pathogens within host cytoplasm (Dramsi and Cossart, 1998). Actin polymerization is also required in nonmotile cells for cytoplasmic organelle transport, endocytosis, and the organization of dynamic actin structures (Winter et al., 1997, 1999; Qualmann et al., 2000; Pelham and Chang, 2001; Taunton, 2001).

Biochemical studies have provided detailed information about the molecules that influence actin dynamics (Pantaloni et al., 2001). Of particular significance is the Arp2/3 complex that stimulates microfilament nucleation, the rate-limiting step in actin polymerization (Mullins et al., 1998; Welch et al., 1998). The Arp2/3 complex consists of seven protein subunits, including the actin-related Arp2 and Arp3, and is conserved among eukaryotes (Machesky and Gould, 1999; May, 2001). Members of the evolutionarily conserved Wiskott-Aldrich Syndrome protein (WASp)^* and Scar/WAVE family function as strong potentiators of Arp2/3 complex activity (Higgs and Pollard, 2001). Distinct WASp and Scar/WAVE branches of this family have been recognized in diverse organisms, including Dictyostelium, Caenorhabditis elegans, Drosophila, and mammals. WASp and Scar/WAVE proteins share a common domain structure that mediates activation of the Arp2/3 complex in response to multiple signaling pathways. All members of the WASp-Scar/WAVE family possess a common COOH-terminal (WA) domain that stimulates actin polymerization through association with monomeric actin and the Arp2/3 complex, whereas their NH₂-terminal domains are structurally distinct and serve as signal-responsive regulatory regions (Fig. 1). The molecular mechanisms controlling WASp function are well characterized (Fawcett and Pawson, 2000), whereas regulatory aspects of Scar function are now beginning to emerge (Takenawa and Miki, 2001).

The single WASp/Scar protein in budding yeast is required for processes that have been shown to be Arp2/3 dependent (Li, 1997; Naqvi et al., 1998), indicating a functional connection in vivo as well as in vitro. In what cellular contexts does this system operate during development of multicellular organisms? Are the distinct WASp and Scar homologs present in such organisms involved in common or separate Arp2/3-dependent processes? Here, we describe the first mutant alleles of the single Drosophila Scar/WAVE homologue, SCAR, and compare its functions to those of the previously identified WASp homologue, Wasp (Wsp) (Ben-Yaacov et al., 2001). SCAR and Wsp appear to represent the only homologs of their respective subfamilies in the Drosophila genome, providing an opportunity to compare the functional requirements for these two major branches of the WASp/Scar protein family. Furthermore, the in vivo relevance of WASp and Scar to Arp2/3 complex functions during development of a multicellular organism can be assessed using mutant alleles in components of the Drosophila Arp2/3 complex (Hudson and Cooley, 2002). Our results suggest that Wasp and SCAR mediate distinct subsets of Arp2/3-dependent processes during Drosophila development. Although Wasp is required specifically for proper execution of asymmetric cell divisions in neural lineages, SCAR plays a major role in the Arp2/3 complex-dependent regulation of cell morphology.

A search of the sequenced Drosophila genome identified a single Scar/WAVE homologue (SCAR, corresponding to transcription unit CG4636) that maps to cytogenetic band 32C4-5 on the second chromosome. We determined the structure of the SCAR transcript by sequencing three ESTs from the Berkeley Drosophila Genome Project database (Fig. 1). The 2,184 nucleotide SCAR transcript is predicted to encode a 613 amino acid protein possessing the major hallmarks of Scar/WAVE proteins (Fig. 1, B and C).

To examine SCAR protein expression and subcellular localization, we generated a polyclonal antibody to the unique SCAR NH₂-terminal domain (see Materials and methods). We found that SCAR protein is present in early blastoderm embryos and in the embryonic CNS (Fig. 2) consistent with its mRNA expression (unpublished data). In the blastoderm, SCAR protein colocalizes with filamentous actin structures that are dynamically regulated during the cell cycle (Fig. 2, A and C; see below). In the CNS, SCAR protein is specifically localized to axons (Fig. 2, E and F). This pattern of SCAR protein expression in the embryo provided us with an initial indication of the potential sites of SCAR gene activity.

To investigate SCAR function, two mutant alleles of SCAR were identified and characterized. The recessive lethal P-element insertion mutation l(2)k13811 (Spradling et al., 1999) lies within the 5′ UTR of the SCAR transcript, 208 nucleotides upstream of the translation start codon (Fig. 1 A) (Berkeley Drosophila Genome Project; unpublished data). Chromosomes bearing precise excisions of this insertion complemented the lethality of l(2)k13811 and were homozygous viable. In addition, ubiquitous expression of a full-length SCAR cDNA rescued the lethality of the l(2)k13811 insertion. These data demonstrate that the zygotic recessive lethality is due to disruption of the SCAR gene by the l(2)k13811 insertion, and we refer to this allele as SCAR^k13811. Moreover, embryos that are maternally and zygotically mutant for the SCAR^k13811 allele display a strong reduction in staining with the anti-SCAR antibody (Fig. 2, B and D), confirming that this insertion disrupts SCAR expression.

To obtain deletions in the SCAR locus, we generated imprecise excision alleles of the SCAR^k13811 insertion, all of which were homozygous lethal and failed to complement the lethality of SCAR^k13811. The homozygous lethality of the Δ37 excision allele was rescued by ubiquitous expression of the full-length SCAR cDNA, and we refer to this allele as SCAR^Δ37. SCAR^Δ37 was molecularly characterized and removes all SCAR sequences downstream of the insertion site (Fig. 1 A). This excision event also removes portions of the neighboring piwi transcription unit. Since piwi function is restricted to maintenance and proliferation of germline stem cells (Cox et al., 1998, 2000), the SCAR^Δ37 phenotypes described below, in distinct tissues, are likely to represent consequences of disrupting SCAR function alone. Developmental defects were considerably weaker in SCAR^k13811, indicating that the insertion allele retains partial SCAR activity.

In addition to SCAR and Wasp (Ben-Yaacov et al., 2001), the sequenced Drosophila genome contains predicted homologs of the seven members of the Arp2/3 complex (Fyrberg et al., 1994; Goldstein and Gunawardena, 2000). Mutations have been recovered in two Arp2/3 complex components, Arp3 (Rørth, 1996; Berkeley Drosophila Genome Project) and Arpc1 (Hudson and Cooley, 2002). This set of mutations provides an opportunity to analyze the role of Arp2/3-based signaling in different contexts within a multicellular organism and to ascertain the physiological contributions of the SCAR and Wasp activators.

Homozygous mutations in either SCAR allele result in zygotic lethality, which can occur during late embryogenesis, larval, and early pupal stages. However, maternally provided SCAR gene products may compensate for loss of zygotic gene function and mask an essential requirement during embryogenesis. To interfere with the maternal gene contribution, we used FLP-mediated recombination to produce homozygous mutant clones within the germline of heterozygous females (Chou and Perrimon, 1996). Strong disruption of maternal SCAR or Arpc1 in this manner results in developmental arrest during oogenesis (see below). However, germline clones homozygous for the weaker SCAR^k13811 or Arpc1^R337st alleles give rise to fertilizable eggs, enabling study of functional requirements for SCAR and the Arp2/3 complex during embryogenesis. These embryos are designated SCAR^mat and Arpc1^mat, respectively. To compare the roles of SCAR and Wasp, we examined embryos derived from germline clones for the strong loss of function Wsp³ allele (Wsp^mat embryos). The Wsp³ frameshift mutation truncates the protein before the highly conserved WA domain that is required for Arp2/3 activation and is a probable null allele (Ben-Yaacov et al., 2001).

The early blastoderm embryo undergoes 13 nuclear divisions without accompanying cytokinesis, producing a multinucleate syncytium. The majority of nuclei migrate to the surface by cycle 10, where they undergo four synchronous rounds of division before their compartmentalization into individual cells during interphase of cycle 14 (Zalokar and Erk, 1976). Surface nuclei maintain a uniform distribution throughout these final syncytial divisions (Fig. 3, A and B). Examination of the spatial distribution of nuclei revealed a requirement for SCAR and Arpc1, but not Wasp, during these cortical division cycles (Fig. 3, C–H). SCAR^mat and Arpc1^mat mutants exhibited defects in the uniform spacing of interphase nuclei beginning in cycle 11, whereas Wsp^mat mutants displayed wild-type nuclear organization. By cycles 12 and 13, increased defects in nuclear spacing in SCAR and Arpc1 were accompanied by the appearance of abnormal nuclear morphologies, including large or elongate DNA masses that are likely to represent the fusion of adjacent nuclei.

In wild-type syncytial embryos, nuclei are maintained in two separate populations: a uniform layer of surface nuclei and an internal mass of yolk nuclei. In SCAR^mat and Arpc1^mat embryos, displacement of surface nuclei into the interior was first observed at cycle 12 and increased in severity by cycle 14 (Fig. 3, L and N) (96% of cycle 14 SCAR^mat embryos, n = 24; 100% of cycle 14 Arpc1^mat embryos, n = 24). The severity of these defects was strongly correlated with increased division cycles, demonstrating a late onset progressive defect (for SCAR^mat p = 10⁻¹⁷, n = 76 embryos, and r_sp = 0.80; for Arpc1^mat p = 10⁻²¹, n = 69 embryos, and r_sp = 0.87; r_sp is the Spearman rank correlation coefficient). Nuclear displacement was rarely observed in wild-type and Wsp^mat embryos (PFig. 3, J and P) (3% of cycle 14 wild-type embryos, n = 33; 0% of cycle 14 Wsp^mat embryos, n = 32).

The syncytial blastoderm contains well-defined filamentous actin structures that exhibit dynamic cell cycle regulation (Karr and Alberts, 1986). Actin is organized into caps overlying individual nuclei during interphase of cortical cycles 10–14 (Fig. 4, A and C). During mitosis, actin is redistributed into a network of metaphase furrows that separate adjacent spindles (Fig. 5, A and C). Genetic and drug interference studies demonstrate that organization of the actin cytoskeleton is crucial for the uniform arrangement of blastoderm nuclei (Foe et al., 1993; Schejter and Wieschaus, 1993). The nuclear defects in SCAR mutants, and SCAR protein colocalization with filamentous actin (Fig. 2), raise the possibility that SCAR may function in the regulation of actin structures in the blastoderm embryo.

In SCAR^mat embryos, actin caps appeared largely intact during interphase, although some defects were observed. In particular, SCAR^mat actin caps were consistently smaller and less rounded than in wild type (Fig. 4 E, 22/35 embryos). In a subset of embryos, primarily in later syncytial divisions 12 and 13, actin caps appeared less discrete and gaps were observed in regions where nuclei were abnormally clustered (Fig. 4 G, 13/35 embryos). In contrast to the relatively mild defects in interphase caps, metaphase furrows were completely absent in the majority of SCAR^mat embryos undergoing mitosis, and actin accumulated in aberrant structures positioned above rather than between individual spindles (Fig. 5, G, H, and L, 21/28 embryos). A partial and discontinuous metaphase furrow network was observed in a minority of SCAR^mat embryos (Fig. 5, I and J, and Fig. 2 J, 7/28 embryos). These two mitotic phenotypes were mutually exclusive and consistent across the entire embryo surface. These observations suggest an abnormal reorganization of interphase actin as SCAR^mat embryos enter mitosis, indicating that the transition between interphase caps and metaphase furrows requires SCAR function.

Major defects in cortical actin structures were also observed in Arpc1^mat embryos, where interphase actin caps were abnormal and actin appeared to be depleted from the regions above individual nuclei (Fig. 4, I and K, 37/37 embryos). This depletion is apparent most readily in cross-section (Fig. 5 Q). As in SCAR, metaphase furrows failed to form in Arpc1^mat embryos (Fig. 5 M, 12/12 embryos). However, unlike SCAR, metaphase actin exhibited a diffuse localization to the broad region between spindles (Fig. 5 R). These results demonstrate that the Arp2/3 complex component Arpc1 is required for the formation of both interphase actin caps and metaphase actin furrows. The greater severity of the Arpc1 phenotype compared with SCAR could reflect a difference in residual gene activity of these partial loss of function alleles.

SCAR and Arpc1, therefore, provide functions that are critical for proper formation of cortical actin structures. In contrast, actin caps and furrows formed normally in Wsp^mat embryos (27 interphase and 5 metaphase embryos) (PFig. 5, O and P). It is worth emphasizing in this context that both the SCAR and Arpc1 phenotypes result from a partial loss of gene function, whereas early embryogenesis can proceed normally despite complete lack of Wsp gene activity.

In addition to defects in organization of microfilament structures, overall actin levels in Arpc1^mat and SCAR^mat embryos appeared consistently lower than in wild type. To rigorously assess differences in actin levels, we quantitated surface filamentous actin in syncytial embryos using a phalloidin fluorescence assay. We found that both Arpc1^mat and SCAR^mat mutants exhibited significantly reduced levels of surface actin compared with control embryos (Fig. 6). The more severe loss of actin in Arpc1^mat correlates with the greater disruption of actin structures in this mutant. These results indicate that Arpc1 and SCAR are both required for the generation of bulk filamentous actin in the blastoderm and suggest a common basis for the defects in cortical actin structures of Arpc1 and SCAR mutant embryos.

The striking enrichment of SCAR expression in the CNS (Fig. 2, E and F) prompted us to examine CNS morphology in SCAR mutants using the axon-specific BP102 monoclonal antibody (Fujita et al., 1982). In wild-type embryos, CNS axons travel in two longitudinal bundles that flank the midline and two commissural bundles that cross the midline in each segment (Fig. 7 A). Although no apparent defects were observed in homozygous SCAR embryos (Fig. 7, B and I), maternal contribution of SCAR transcript or protein may provide sufficient wild-type SCAR activity to mask a functional requirement in the CNS. Partial maternal SCAR function provided by the weak SCAR^k13811 allele at lower temperature (20–22°C rather than 25°C) is sufficient to produce embryos that develop normally through the blastoderm stages described above, allowing an examination of later CNS development.

Reduction of SCAR function achieved in this manner indeed caused dramatic CNS defects (Fig. 7, C and I). The phenotypes observed in these mutants (designated SCAR^mat/zyg embryos) required disruption of zygotic SCAR function (unpublished data). In SCAR^mat/zyg embryos, frequent breaks occurred in longitudinal and commissural bundles (93% of segments, n = 242). In extreme cases, a severe depletion of all axon bundles was observed (Fig. 7 C, 46% of segments). At a lower frequency, SCAR^mat/zyg embryos exhibited defects in commissure fasciculation and separation (18% of segments), and medial (13%) or lateral (9%) displacement of axons.

Since SCAR is essential for normal CNS axon morphology, we also examined the zygotic effect of mutations in two members of the Arp2/3 complex, Arp3 and Arpc1. Arp3 zygotic mutant embryos exhibited a partially penetrant defect in CNS axon morphology with a range of phenotypes that strongly resemble SCAR^mat/zyg mutants (Fig. 7, E and J). In particular, the majority of Arp3 mutants displayed breaks in the longitudinal and commissural axon bundles (Fig. 7 E) (48% of segments in mutant embryos, n = 115 segments). A subset of Arp3 mutants exhibited defects such as commissure defasciculation or fusion (28% of segments) and medially or laterally displaced axons (3% of segments). Arp3 heterozygotes also exhibited a low penetrance of axon defects (Fig. 7 J). The CNS morphology of zygotic Arpc1 single mutants appeared normal, perhaps due to the continued presence of maternal gene products. However, combining zygotic Arpc1 mutations with an Arp3 heterozygous background produced defects that were significantly more severe than in Arp3 heterozygotes alone (Fig. 7, F and J). These phenotypes demonstrate a similar functional requirement for SCAR and Arp2/3 complex components during CNS development.

The contribution of Wasp function to CNS axon morphology is more difficult to assess, since complete removal of maternal and zygotic Wsp using the strong Wsp³ allele (Wsp^mat/zyg embryos) produces cell fate defects in CNS lineages (Ben-Yaacov et al., 2001). An apparent thickening of commissural bundles suggestive of an increase in neuronal number was observed in a majority of Wsp^mat/zyg embryos (Fig. 7 D). In addition, most Wsp^mat/zyg embryos contained one to two segments with axon bundles collapsed at the midline (Fig. 7 D) (73% of embryos, n = 41). Despite these phenotypes, Wsp^mat/zyg embryos did not exhibit the severe defects in axon morphology present in SCAR and Arp3 mutants. Although removal of zygotic SCAR or Wsp function alone did not disrupt CNS morphology (Fig. 7, B and I), zygotic reduction of SCAR and Wsp together produced significant defects (Fig. 7, G, H, and I) that resemble the strong SCAR^mat/zyg phenotype. Therefore, although loss of Wasp function alone does not cause the significant axon defects produced by loss of SCAR, Wasp can influence axon morphology in situations where SCAR function is compromised.

Although the partial reduction of SCAR function associated with the SCAR^k13811 insertion allele is sufficient for normal egg production, the more severe SCAR^Δ37 excision allele produces small and abnormally shaped eggs indicative of a defect in oogenesis. Drosophila ovaries house a series of egg chambers that each contain 16 cells (the oocyte and a 15-cell nurse cell complex) interconnected by cytoplasmic bridges (ring canals) that arise from incomplete cytokinesis during mitosis (Spradling, 1993). Morphological defects are apparent in SCAR^Δ37 germline clones during the final phases of oogenesis (Fig. 8). In particular, nurse cells become multinucleate, as many of the actin-lined nurse cell membranes are absent (Fig. 8 B). The morphological abnormalities extend to additional structures, including the actin-rich ring canals, which are significantly smaller than in wild-type and aberrantly shaped (Fig. 8, F and H).

The defects observed in SCAR mutant egg chambers closely resemble phenotypes described in mutants for the Arp2/3 complex subunits Arpc1 and Arp3 (Hudson and Cooley, 2002), resulting in late stage deterioration of the nurse cell complex (Fig. 8 C). In marked contrast to the Arpc1, Arp3, and SCAR phenotypes, oogenesis in germline clones for the strong loss of function Wsp³ allele appears wild type. No apparent morphological abnormalities were observed in Wsp³ late stage egg chambers (Fig. 8 D), which can support embryonic development after fertilization (Ben-Yaacov et al., 2001). This phenotypic analysis indicates that SCAR, rather than Wasp, is the major mediator of Arp2/3 function during Drosophila oogenesis, much as was observed in the blastoderm and the embryonic CNS.

The above phenotypic analyses identify several Arp2/3-dependent morphological processes that rely on SCAR but are largely independent of Wasp. We therefore asked whether the reciprocal situation exists, namely, are there Arp2/3-mediated events that rely on Wasp but are independent of SCAR? Wasp provides an essential contribution to cell fate decisions in several neural lineages in the Drosophila embryo and adult (Ben-Yaacov et al., 2001). Furthermore, the Arp2/3 complex component Arpc1 is required for Wasp-dependent cell fate changes during sensory organ development, and association with Arp2/3 is essential for Wasp function in this context (Tal et al., 2002). This requirement provides an opportunity to examine whether developmental events dependent on Wasp also require SCAR function.

In the adult peripheral nervous system, a primary consequence of mutations in Wsp is the excessive differentiation of sensory organ neurons at the expense of nonneuronal cell types, resulting in a marked absence of mechanosensory bristles (Fig. 9, A and B). We used the ey-FLP-FRT system (Newsome et al., 2000) to generate mosaic SCAR and Arpc1 heterozygous flies in which head capsule structures and cuticle are derived from homozygous mutant clones induced in the eye imaginal disc. Arpc1 mosaic heads like, Wsp, display a pronounced bristle loss phenotype (Fig. 9 C), which results from cell fate defects similar to those present in Wsp mutants (Tal et al., 2002). In contrast, the sensory organ pattern in mosaic heads of strong SCAR alleles appears wild type (Fig. 9 D), suggesting that SCAR does not play an essential role in lineage decisions mediated by Wasp and the Arp2/3 complex.

In addition to loss of sensory organ structures, Arpc1 mosaics display abnormalities in eye structure, including a reduction in the overall size of the eye, irregularly shaped ommatidia, and a distinct loss of lens material in most eye facets (Fig. 9, C and G). Mosaics for the SCAR^Δ37 excision allele present a very similar eye phenotype (Fig. 9, D and H), with the exception that interommatidial bristles are largely present. As noted previously (Ben-Yaacov et al., 2001), mutations in Wsp have no discernible effect on eye morphology apart from the bristle loss phenotype (Fig. 9, B and F). This analysis provides a striking example of the distinct requirements for SCAR and Wasp, which mediate separate aspects of Arp2/3 complex function during adult development.

These results present the first genetic analysis of Scar function in a multicellular organism and demonstrate a requirement for Scar and the Arp2/3 complex in regulating the morphology of multiple cell types. The highly similar requirements for Drosophila SCAR and Arp2/3 complex components in regulating cytoplasmic organization in the blastoderm and cell morphology in CNS neurons, egg chambers, and adult eyes suggest that they function in a common pathway in vivo, consistent with their well-established regulatory interaction in vitro. These roles of SCAR and the Arp2/3 complex are largely independent of Wasp function, suggesting that SCAR is the primary regulator of Arp2/3-dependent morphological processes in Drosophila. In contrast, Wasp is specifically required for the Arp2/3-dependent regulation of asymmetric cell divisions (Ben-Yaacov et al., 2001; Tal et al., 2002) a process that is independent of SCAR (this study). These results demonstrate that SCAR and Wasp perform generally nonoverlapping functions during Drosophila development and that the Arp2/3 complex can participate in distinct cell biological events in response to different regulators. Although SCAR and Wasp can account for all characterized Arp2/3 complex functions in Drosophila, recent studies have described Arp2/3 complex regulators outside of the Scar/Wasp family (Jeng and Welch, 2001). Therefore, homologs of such elements (such as Cortactin and Eps15/Pan1p) may also play a role in Arp2/3-dependent processes during Drosophila development.

Here we demonstrate a requirement for SCAR in the regulation of axon morphology in the Drosophila CNS. The striking enrichment of SCAR protein in axons is consistent with a direct role for SCAR in axon development. In particular, the breaks in longitudinal and commissural axon bundles in SCAR mutant embryos may indicate a defect in axon growth. However, these phenotypes could also reflect defects in other aspects of nervous system formation, such as axon guidance, axon initiation, or neuronal differentiation. Morphological characterization of SCAR mutants at single neuron resolution will provide greater insight into the processes that require SCAR function.

The CNS axon defects in SCAR mutant embryos resemble defects caused by simultaneous zygotic disruption of the Abl tyrosine kinase and a diverse set of elements including the Fasciclin I transmembrane protein, Armadillo/β-catenin, Chickadee/profilin, and the Trio Rac/Rho guanine nucleotide exchange factor (Lanier and Gertler, 2000). Interestingly, Scar/WAVE-1 has been shown to associate with the SH3 domain of the Abl tyrosine kinase, suggesting that they may directly interact in vivo (Westphal et al., 2000). The observation that multiple zygotic mutations are required to replicate the SCAR phenotype is consistent with a model where SCAR functions downstream of multiple signaling pathways that converge on regulation of the actin cytoskeleton.

The defects in axon morphology caused by reduction of maternal and zygotic SCAR are similar to those produced by zygotic disruption of Arp3 or simultaneous zygotic disruption of Arp3 and Arpc1 or SCAR and Wasp. These results suggest that SCAR, Wasp, and the Arp2/3 complex may affect a common process in neuronal development involving actin regulation. The contribution of both SCAR and Wasp to axon morphology could be explained by several possible mechanisms. In one model, SCAR and Wasp might regulate a common activity of the Arp2/3 complex, such as in the context of a specific actin structure or in contribution to bulk actin levels. Their functional differences in vivo could be achieved through differences in expression, activation, or subcellular localization. Alternatively, SCAR and Wasp could regulate distinct activities of the Arp2/3 complex, producing different actin structures that participate in diverse cell biological processes such as cell morphology (SCAR) and asymmetric cell division (Wasp). It will be interesting to examine how SCAR and Wasp intersect with regulators and effectors to achieve the specific organization of actin structures in different contexts.

The dramatic reduction in actin levels of Drosophila Arpc1 mutants indicates that the Arp2/3 complex is an essential source of filamentous actin in the blastoderm embryo. This is consistent with experiments in other systems, where the Arp2/3 complex is required for actin polymerization in yeast actin patches (Pelham and Chang, 2001) and cell extracts in response to the Cdc42 GTPase (Ma et al., 1998; Mullins and Pollard, 1999) or the Listeria pathogen (Welch et al., 1997). Our results also suggest that the SCAR regulator mediates this Arp2/3-dependent actin polymerization in the blastoderm. A similar reduction in filamentous actin is observed in Dictyostelium Scar mutants (Bear et al., 1998), and budding yeast Bee1 is required for actin polymerization at actin patch structures in a permeabilized cell assay (Lechler and Li, 1997). Together, these results demonstrate a conserved role for the Arp2/3 complex and WASp/Scar proteins in promoting actin polymerization in vivo as well as in vitro.

In budding and fission yeast, inducible disruption of Arp2/3 complex function first leads to a cessation of actin patch movement followed by their eventual dissolution (Winter et al., 1997; Pelham and Chang, 2001). Therefore, the Arp2/3 complex is required for the motility of actin structures and their formation. Similarly, Dictyostelium Scar mutants exhibit a selective disruption of specific actin structures that cannot easily be explained by an overall reduction of actin. Actin correctly localizes to the cell cortex and extending pseudopods as in wild type; however, leading edge actin fails to coalesce in response to chemoattractant, often leading to the aberrant formation of multiple pseudopods (Bear et al., 1998). These results suggest that Scar is involved in the dynamic organization of actin structures as well as their generation.

An exciting possibility is that Scar and the Arp2/3 complex direct both the configuration and polymerization of actin in the Drosophila blastoderm. SCAR embryos in metaphase contain more than half the actin of wild-type embryos, yet this substantial amount of actin often fails to form even a discontinuous network of metaphase furrows. Instead, actin remains in aberrant surface structures that are not normally found at the surface of mitotic embryos. These observations suggest that SCAR plays a role in actin redistribution, perhaps through a local Arp2/3-dependent polymerization event that triggers a global cell cycle–dependent change in actin organization. This role of SCAR in the Drosophila embryo may be analogous to the reorganization of actin structures that occurs in other contexts, such as during cytokinesis or at the leading edge of migrating cells.

Flies were maintained using standard methods. Wild-type stocks were Oregon R (immunohistochemistry) and oskar¹⁶⁶ (SCAR antibody controls and actin quantitation) (Lehmann and Nusslein-Volhard, 1986). See Flybase (http://flybase.bio.indiana.edu) for details concerning fly stocks. Alleles used were Arp3^EP(3)3640 (Rørth, 1996; Berkeley Drosophila Genome Project), Wsp¹, Wsp³ (Ben-Yaacov et al., 2001), Arpc1^R337st, Arpc1^Q25sd, Arpc1^W108R (Hudson and Cooley, 2002), SCAR^k13811 (Spradling et al., 1999; Berkeley Drosophila Genome Project), and SCAR^Δ37.

Germline clones were generated as described (Chou and Perrimon, 1996) by heat shock of hs-FLP; ovoD FRT40A/SCAR FRT40A larvae, hs-FLP; FRT82B ovoD/FRT82B Wsp larvae, or hs-FLP; ovoD FRT40A/Arpc1 FRT40A larvae. Adult germline clone females were mated to Oregon R males (blastoderm and oogenesis analysis) or to SCAR^k13811/CyO en-lacZ males (SCAR^mat/zyg) or Df(3R)3450/TM6B abdA-lacZ males (Wsp^mat/zyg) (CNS). Mosaic head clones were obtained in ey-FLP; Arpc1^Q25sd FRT40A/l(2)cl-L3¹ FRT40A and ey-FLP; SCAR^Δ37 FRT40A/l(2)cl-L3¹ FRT40A flies. SCAR^k13811 and Wsp³ germline clones (blastoderm), Wsp³ germline clones (CNS), and all zygotic mutants were generated at 25°C. Arpc1^R337st germline clones (blastoderm) and SCAR^k13811 germline clones (CNS) were generated at 20–22°C.

We observed no contribution of zygotic gene activity to the blastoderm defects of embryos derived from SCAR^k13811 and Arpc1^R337st germline clones (unpublished data). Wsp^mat embryos include embryos defective for both maternal and zygotic Wsp function and embryos defective only for maternal Wsp function.

The full-length SCAR cDNA from the SD02991 EST was cloned as an EcoRV-XhoI (blunt) fragment into the pUASP vector (Rørth, 1998). Transgenic flies containing the UAS-SCAR transgene were generated by P-element–mediated transformation and two independent lines used for phenotypic rescue. Rescue to adult viability was obtained in SCAR/SCAR; UAS-SCAR/P{tubP-GAL4} for SCAR^k13811 and SCAR^Δ37 homozygotes.

The SCAR NH₂-terminal region (amino acids 1–237) was cloned into the pRSET vector (Invitrogen) to generate a 6×His-tagged protein, which was purified and injected into guinea pigs (Cocalico Biologicals). SCAR polyclonal antibody was used at a 1:50 dilution to stain formaldehyde-fixed embryos as described (Theurkauf, 1994). SCAR polyclonal antibody was visualized with Alexa 546–conjugated secondary antibody and costained with Alexa 488–conjugated phalloidin (Molecular Probes).

The l(2)k13811 P-element insertion in the 5′ UTR of SCAR was mobilized by introducing transposase on the chromosome CyO Δ2-3. Excisions were identified in F2 progeny derived from single males of the genotype w; SCAR^k13811/CyO Δ2-3. 15 homozygous viable alleles complemented the SCAR^k13811 lethality and are likely to represent precise excision events, verified by sequencing one such chromosome. 11 (presumably imprecise) excision alleles failed to complement the SCAR^k13811 lethality and were homozygous lethal. The Δ37 excision allele complements the lethality of l(2)06225, an insertion within the CG6105 transcription unit 400 bp upstream of SCAR (Fig. 1 A) and fails to complement the sterility of the downstream piwi gene.

SCAR^Δ37 homozygotes were chosen as nonfluorescing larvae from a SCAR^Δ37/CyO, Act-GFP stock. Genomic DNA was prepared from five individual larvae, and PCR amplification for sequencing was performed using primers from the SCAR region.

Embryos were fixed for 20 min in 3.7% formaldehyde/PBS:heptane and devitellinized in heptane:methanol. Embryos were stained with mouse mAb BP102 (1:10 dilution; Developmental Studies Hybridoma Bank) and rabbit anti–β-galactosidase (1:1,500 dilution; Cappel), followed by fluorescence-conjugated Alexa 488 and Alexa 546 secondary antibodies (Molecular Probes), and mounted in Aquapolymount (Polysciences, Inc.). Images were z-series projections obtained on a ZEISS LSM 510 confocal microscope. CyO en-lacZ, TM6B abdA-lacZ, and TM3 Ubx-lacZ balancers were used to genotype embryos. For Arp3 heterozygotes, Arp3/balancer females were mated to WT males. When deficiencies were used, mutant/balancer females were mated to deficiency/balancer males. For double mutant analyses, double mutant females were crossed to deficiency/balancer or SCAR^k13811/balancer males. Statistics were computed using the Primer of Biostatistics program (Stanton Glantz) and Numerical Recipes in C (Press et al., 1992).

Embryos were collected at 20–22°C (Arpc1^mat and oskar control) or 25°C (SCAR^mat and oskar control), fixed for 45 min in 19% formaldehyde/PBS:heptane, and hand devitellinized. Control and mutant embryos were pooled and incubated in a single tube for 2 h with 6.6 nM Alexa 488–conjugated phalloidin (Molecular Probes) and 0.1 μg/ml Hoechst (Sigma-Aldrich). Mean surface fluorescence intensity was measured for two areas of each embryo (the brightest surface 1.5 μm optical slice) and averaged. Images were obtained on a ZEISS LSM 510 confocal microscope using a C-Apochromatic 40×/1.2 NA water immersion objective. Images were generated using identical linear gain settings at zoom 3 and 1,024 pixel² resolution to achieve Nyquist resolution, an optimal sampling rate. Gain settings were determined empirically to allow a range of intensities to be detected with minimal saturation of the higher control signal. Fluorescence intensity was quantitated in ImagePro Plus (Media Cybernetics). Background fluorescence was determined from 15 embryos processed identically but without phalloidin; these embryos displayed a nearly identical nonzero fluorescence, which was subtracted from the measured fluorescence in all images. Standard error of the mean was calculated in Kaleidagraph. In separate experiments to assess actin distribution (Fig. 5), embryos were costained with mouse anti–β-tubulin antibodies (1:500 dilution; Sigma-Aldrich) for precise cell cycle staging. Antibody staining obscured the difference in phalloidin fluorescence; therefore, no antibodies were included in the quantitation experiment.

Ovaries were dissected from 3–5 d-old females and fixed for 15 min in 6% formaldehyde/PBS. Germline clones could first be identified after stage 8 of oogenesis (Spradling, 1993). Egg chamber microfilaments were visualized by staining with rhodamine-phalloidin (1 U/ml, 20 min; Molecular Probes), nuclei with OliGreen (1:5,000, 10 min, after a 1-h treatment with 5 μg/ml RNase; Molecular Probes), and ring canals with monoclonal Hts-RC antibody (clone 7C, 1:10 dilution).

Adult heads underwent critical point drying and sputter coating with a gold film after dehydration in an ethanol series. Scanning EM was performed using a JEOL JSM-6400 microscope.

We are grateful to Joe Goodhouse for assistance with confocal microscopy, Bahram Houchmandzadeh for advice on statistical analysis, and Eugenia Klein for instruction and skillful operation of the scanning electron microscope. We thank Mimi Shirasu-Hiza, Rachel Hoang, Jörg Großhans, Ali Nouri, Cheryl Van Buskirk, and anonymous reviewers for helpful comments on the article. Some stocks were obtained from the Bloomington Drosophila Stock Center.

J.A. Zallen was supported by a Koshland Postdoctoral Fellowship and is a Postdoctoral Fellow of the Damon Runyon-Walter Winchell Foundation Cancer Research Fund. E. Wieschaus is an Investigator of the Howard Hughes Medical Institute. E.D. Schejter is supported by grants from the Israel Science Foundation and the Minerva Foundation.