Genetic characterization of the Drosophila homologue of coronin

We report cloning and characterization of coro, which codes for the Drosophila homologue of the F-actin binding protein coronin. Viable alleles of coro produce a variety of phenotypes in leg, wing and eye development, which are similar to the phenotypes observed as a result of mutations in genes associated with the actin cytoskeleton and/or membrane trafficking. Homozygous lethal mutations in coro results in the disruption of the actin cytoskeleton in wing imaginal discs. Formation of both basolateral septate junctions and apical adherens junctions are also adversely affected in epithelial cells. Both viable and lethal alleles of coro show genetic interactions with syntaxin1A, a gene required for membrane trafficking. They also show enhanced response to over-expression of Decapentaplegic (Dpp) and its receptor Thick vein. Tracing of Dpp morphogen using a Dpp::GFP fusion construct suggested defects in the endocytic pathway, which resulted in uniform distribution of Dpp along the AP axis rather than a gradient from the AP boundary. Our results provide a genetic link between endocytosis/exocytosis events involving F actin-coated vesicles and the establishment of morphogen gradient.


Introduction
The actin cytoskeleton and associated proteins play many important roles both during development and during normal function of a differentiated cell. They are required for cell shape control, cortical stiffness/tension, spindle rotation, cytokinesis, cell and nuclear migration, vesicle and organelle transport, transport of morphogens and other cell fate determinants. Reorganization of the cytoskeleton, in particular actin polymerization, is the driving force for epithelial cell-cell adhesion (Vasioukhin et al., 2000;Vasioukhin and Fuchs, 2001). Very often microtubule and actin filament networks cooperate functionally during cellular processes such as transport of vesicles and organelles (Goode et al., 2000). Understanding the biochemical mechanisms that control the organization of actin is thus important and has implications for health and disease.
Amongst the many actin-associated proteins that mediate various functions of the actin cytoskeleton, the coronin family of proteins is highly conserved: from Dictyostelium to human. They are known to promote actin polymerization, cytokinesis and all actin-mediated processes (de Hostos et al., 1993) (for a review, see de Hostos, 1999). Analyses of coronin-GFP (green fluorescent protein) fusions and knockout mutants show that coronin participates in the remodeling of the cortical actin cytoskeleton that is responsible for phagocytosis and macropinocytosis (Maniak et al., 1995;Hacker et al., 1997). Likewise, in mammalian neutrophils, a coronin-like protein is associated with the phagocytic apparatus (Grogan et al., 1997). In addition to their association with proteins such as vinculin at focal contacts, they are also part of the actin cross-bridge between the actin cytoskeleton and the plasma membrane (Nakamura et al., 1999). Establishment of anterior-posterior (AP) polarity in the Caenorhabditis elegans embryo depends on POD-1, a coronin-like protein. Loss of POD-1 from early embryos results in a loss of physical and molecular asymmetries along the AP axis (Rappleye et al., 1999). It also causes the formation of large vesicles of endocytic origin, membrane protrusions, unstable cell divisions, a defective eggshell and deposition of extracellular material (Rappleye et al., 1999). Human coronin protein (TACO) is involved in the survival of mycobacteria and Helicobacter pylori in phagosomes (Ferrari et al., 1999;Zheng and Jones, 2003). TACO is actively retained on the outer surface of the phagosomes by living mycobacteria and thereby prevents their delivery to lysosomes, allowing mycobacteria and Helicobacter pylori to survive within macrophages. It has been further shown that cholesterol mediates the phagosomal association of TACO (Gatfield and Pieters, 2000). By entering host cells at cholesterol-rich domains of the plasma membrane, mycobacteria may ensure their subsequent intracellular survival in TACO-coated phagosomes.
We report cloning and characterization of coro, which codes for the Drosophila homologue of the F-actin binding protein coronin. Viable alleles of coro produce a variety of phenotypes in leg, wing and eye development, which are similar to the phenotypes observed as a result of mutations in genes associated with actin-cytoskeleton and/or membrane trafficking. Homozygous lethal mutations in coro results in the disruption of the actin cytoskeleton in wing imaginal discs. Formation of both basolateral septate junctions and apical adherens junctions are adversely affected in the epithelial cells of these strong alleles. Both viable and lethal alleles of coro show genetic interactions with syntaxin1A (syx1A), a gene We report cloning and characterization of coro, which codes for the Drosophila homologue of the F-actin binding protein coronin. Viable alleles of coro produce a variety of phenotypes in leg, wing and eye development, which are similar to the phenotypes observed as a result of mutations in genes associated with the actin cytoskeleton and/or membrane trafficking. Homozygous lethal mutations in coro results in the disruption of the actin cytoskeleton in wing imaginal discs. Formation of both basolateral septate junctions and apical adherens junctions are also adversely affected in epithelial cells. Both viable and lethal alleles of coro show genetic interactions with syntaxin1A, a gene required for membrane trafficking. They also show enhanced response to over-expression of Decapentaplegic (Dpp) and its receptor Thick vein. Tracing of Dpp morphogen using a Dpp::GFP fusion construct suggested defects in the endocytic pathway, which resulted in uniform distribution of Dpp along the AP axis rather than a gradient from the AP boundary. Our results provide a genetic link between endocytosis/exocytosis events involving F actin-coated vesicles and the establishment of morphogen gradient. required for membrane trafficking. They also show enhanced response to Decapentaplegic (Dpp) and its receptor Thickvein (Tkv) over-expression. Tracing of Dpp morphogen using a Dpp::GFP fusion construct suggested uniform distribution of Dpp along the AP axis rather than a gradient from the AP boundary. In addition, over-expression of Dpp restored the normal actin cytoskeleton, which supports earlier finding on the role of Dpp in cytoskeletal regulation. These results on cloning and characterization of coro, thus, provide a new tool to employ the power of Drosophila genetics to understand the role of the actin cytoskeleton in various developmental processes.

Materials and Methods
Standard genetic techniques were used for balancing excision mutations, making recombinant chromosomes and combinations of different mutations and/or markers. 409-GAL4 was identified in the lab in a screen for enhancer-traps to identify targets of homeotic gene function. Polytene chromosome in situ hybridization, plasmid rescue and sequencing of the flanking genomic DNA were carried out as per the standard protocols. Excision mutations were generated by crossing 409-GAL4 to a genetic source of transposase. UAS lines used in this study were, UAS-lacZ (Brand and Perrimon, 1993), UAS-GFP (Lee and Luo, 2001), UAS-Dpp::GFP (Teleman and Cohen, 2000), UASactivated Thickvein (UAS-Tkv*) (Nellen et al., 1996) and dominant negative form of Syntaxin1A SKP and LSS,unpublished). omb-lacZ is described by Tsai et al. (Tsai et al., 1997) and dpp-GAL4 by Morimura et al. (Morimura et al., 1996). UAS-DNsyx1A is an EP insertion within the transcribed region (800 bp from the transcription start site) of syx1A. It was isolated in an EP screen using Ubx-GAL4 driver (Pallavi and Shashidhara, 2003) to identify suppressors and enhancers of Ubx (S.K.P. and L.S.S., unpublished). RT-PCR analysis confirmed that this EP line expresses a truncated transcript. In all tissues, this EP insertion induced phenotypes that were identical to the phenotypes reported for loss of function alleles of syx1A (Shulze and Bellen, 1996), thus suggesting that it expresses a dominant negative form of Syx1A.

Histology
RNA in situ hybridization was done essentially as described by Tautz and Pfeifle (Tautz and Pfeifle, 1989) using cDNA clone LD32977 as the probe. X-gal staining and immunohistochemical stainings were essentially as described previously (Ghysen and O'Kane, 1989;Patel et al., 1989). Visualization of Dpp::GFP was as described by Teleman and Cohen (Teleman and Cohen, 2000). FITC-labeled actin-phalloidin was purchased from Molecular Probes, USA. The primary antibodies used were, anti-Arm (Riggleman et al., 1990) anti-Ci (Motzny and Holmgren, 1995), anti β-galactosidase (Sigma, St Louis, MO), anti-Dll (Vachon et al., 1992), anti-En (Patel et al., 1989), anti-FASIII (Patel et al., 1987), anti-Sal (Kuhnlein et al., 1994) and anti-Wg (Brook and Cohen, 1996). Anti-Arm, Anti-En, Anti-FasIII and anti-Wg antibodies were obtained from the Development Studies Hybridoma Bank, University of Iowa, USA. Confocal microscopy was carried out on a Meridian Ultima or on Zeiss LSM510/Meta microscope. Trypsin-mediated dissociation of imaginal disc cells was done according to the method of Neufeld et al., (Neufeld et al., 1998). The adult appendages were processed for microscopy as described previously (Shashidhara et al., 1999). For scanning electron microscopy (SEM), adult heads were air dried and coated with gold. SEM was carried out in an Hitachi S520 machine.

Cloning of coro
In a GAL4-enhancer trap screen for segmentally modulated expression patterns, we identified a GAL4 insertion (409-GAL4), which showed segment-specific expression patterns in the embryonic CNS (V.B. and L.S.S., unpublished). The Pelement insertion was mapped to polytene location 42C-D by in situ hybridization (data not shown). We cloned the genomic region adjacent to the site of the P-element insertion by plasmid rescue and sequenced the same. The 409-GAL4 enhancer-trap insertion was mapped to the 5′ (at bp 56) untranslated region of CG9446 a predicted gene within AE003790 with considerable homology to coronin family of proteins (hereafter referred to as Coro). At the genomic level, the insertion was at position 82185 bp of AE003790, the genomic contig from the 42C8 region of the polytene chromosome (as per the BDGP Release 1.0; http://www.fruitfly.org; Fig. 1A).
By northern blot analysis, we identified a single species of transcript of size ~2.5 kb (Fig. 1B). Amongst the many cDNA clones from BDGP EST collection, LD32977 is 2610 bp long. We completely sequenced this cDNA clone and it has been submitted to GenBank (accession no. AF467271). The transcription unit is 10443 bp long (as per the LD32977), consisting of seven exons and six introns (Fig. 1A). As per the BDGP Release 3.1, CG9446 transcript RA is 2941 bp long and the transcription unit is 10768 bp long. CG9446-RA is longer than LD32977 (and many other ESTs such as AT08250, RE67756, LD28262, RE30002, LD15561, RE13103, etc) by 70 bp at the 5′ end and by 260 bp at the 3′ end. The predicted translation product is 528 amino acid long (Fig. 1C), which is the same as the predicted protein of CG9446. The protein product has 2 WD repeats, which forms a β-propeller-like structure, believed to be important for coronin function (de Hostos, 1999). At the C-terminal end it has a α-helical coiled coil domain (Fig. 1D), a dimerization domain, which is implicated in its localization to cell periphery (Asano et al., 2001).
It shows considerable identity at the protein sequence level to a number of coronin and coronin-like proteins from different organisms across the entire eukaryotic spectrumfrom yeast to human (http://flybase.bio.indiana.edu/.bin/ fbidq.html, FBgn0033109). Its closest homologues are coronins from mosquito (74% identity across 376 residues) and zebrafish (55% identity across 265 residues), followed by human, Xenopus, mouse and rat (around 54% identity across 265 residues), in that order.

Coro is expressed in all cells
The 409-GAL4 enhancer-trap driver showed a tissue-specific expression pattern. In imaginal discs its expression is restricted to only the peripodial membrane ( Fig. 2A). More than 100 ESTs have been identified for coro from a number of different cDNA libraries (such as LD, CK, AT, RE, RH, GM and SD libraries for embryonic, larval discs, adult head etc) from both BDGP and RIEKEN projects (http://www.fruitfly.org). This suggests that it is expressed in all tissues, probably at very high levels. Indeed, RNA in situ hybridization using the cDNA clone LD32977 suggested that coro is expressed in all cell types including differentiating embryonic CNS and differentiating photoreceptor neurons in eye discs ( Fig. 2B-D,G-H). However, it is expressed at very low levels in the larval CNS (Fig. 2G). Recent report on the large-scale microarray analyses to determine gene expression during the life cycle of Drosophila homologue of coronin Drosophila, suggests that Coro is expressed at very high levels at 0 hours AEL, suggesting a maternal contribution (Arbeitman et al., 2002). Its expression gradually comes down during embryogenesis, reaches lowest levels in the larval stages and rises again significantly in pupal stages.
Generation of mutations in coro P-element insertion in the 409-GAL4 enhancer-trap line is within the 5′ UTR of coro. However, the strain is viable and did not show any abnormal phenotype. We generated deletion mutations within the Drosophila coronin gene by imprecise excision of the P-element. We generated six homozygous lethal and four partially viable mutations. All the ten alleles belong to a single complementation group. Amongst the six lethal excisions, five were early pupal lethal (coro ex2 , coro ex3 , coro ex4 , coro ex6 and coro ex7 ) and one (coro ex8 ) was lethal at pharate adult stages. The viable alleles (coro ex9 , coro ex10 , coro ex11 and coro ex12 ) were lethal over the stronger alleles. All the alleles showed similar phenotypes in varying degrees of severity in homozygous and/or trans-allelic (transhomozygous) combinations.
We used two representative alleles in all our experiments; coro ex6 and coro ex11 . For these two alleles we did extensive polymorphism studies to identify the molecular nature of the mutation. First, we performed RT-PCR analyses to identify the nature of mutations. Since Coro is contributed maternally, we examined Coro transcripts in the third instar wing discs of homozygous coro ex6 and coro ex11 larvae. In coro ex6 , we detected a translocation of 231 bp from CG16791 of the third chromosome (cytology: 93C7) in the third exon of coro (Fig.  3A). Since we also noticed normal RT-PCR product, the translocated region may get spliced out, albeit inefficiently. We also designed primers to PCR-amplify genomic DNA flanking the site of the P-element insertion. In total 39 pairs of primers were designed to amplify 18 kb of genomic DNA as 500-800 bp long fragments from wild-type, homozygous coro ex6 and coro ex11 larvae. In addition to the translocation mentioned above, coro ex6 showed a polymorphism at 1250 bp upstream of the transcription start site (Fig. 3A). Sequencing of this region indicated a 100 bp deletion. According to BDGP release 3.1, this region corresponds to the 3′ end of the predicted gene CG9447. Since no cDNA or EST clone is known for CG9447 (http://www.fruitfly.org), we attempted to examine if there is indeed a gene in this region. We designed two pairs of primers to amplify predicted exon1 alone and both exon1 and exon2. We did not see any amplification at the RT-PCR level (using RNA isolated from both embryos and larvae), whereas these primers amplified products of expected sizes from the genomic DNA.
coro ex11 showed GAL4 activity (expression patterns were identical to that of original 409-GAL4), suggesting incomplete excision of the P-element. Similar GAL4 activity was noticed in two more alleles, coro ex8 and coro ex10 . Further PCR analyses indicated that they all have internal deletions of the P-element (thereby, removing w + activity). Northern blot analyses of RNA isolated from coro ex11 embryos showed two species of transcripts, one of normal size and the other longer. The longer transcript hybridizes to both LD32977 and GAL4 cDNA, suggesting a read-through transcript initiated at the P-element promoter (data not shown). Such read-through transcription initiated at the P-element promoter has been reported for vg 2a33 allele of vestigial (Hodgetts and O'Keefe, 2001). We did not detect any other molecular lesion in the entire gene for coro ex11 .
Since all the excision alleles belonged to one complementation group, all show identical phenotypes (although with varying degree of severity) and, at least four alleles (coro ex6 , coro ex8 , coro ex10 and coro ex11 ) show molecular lesion in coro; the observed phenotypes are due to mutations in the coro locus. Furthermore, analyses of wing disc mRNA suggest that both the alleles selected are regulatory in nature.

Mutations in coro affect adult morphogenesis
Homozygous coro ex11 is lethal at early to late pupal stages. Viability is 10%, although all homozygous female flies are sterile. The ovary in these flies is very small and devoid of any oocytes (data not shown). Homozygous coro ex11 flies exhibited a variety of leg, wing and eye phenotypes. Legs of coro ex11 flies appeared to be short and thick (Fig. 3C). We observed duplication of sex combs (Fig. 3D,E), an indication of ventralization of legs. However, we did not observe any altered phenotype at the levels of Wingless (Wg), Dpp and Distal-less (Dll) expression patterns (data not shown). The trans-alleleic combination coro ex6 /coro ex11 pharate adults often showed total Journal of Cell Science 117 (10)  loss of one or more legs, which was more pronounced in segments T2 and T3 (data not shown). All coro ex11 adult flies showed loss of anterior and/or posterior wing margin (Fig.  3G,H). They also showed thickening of veins and Delta vein phenotypes at the distal most regions (vein 3 and 4; Fig. 3J). Very often coro ex11 flies showed a small-wing phenotype (Fig.  3K). They exhibited rough eye phenotypes (Fig. 4B,F) and often, total loss of eyes (Fig. 4C). They also showed lack of appropriate adhesion properties, which was reflected in loss of proper organization of ommatidia in the adult eye. We saw ommatidia loosely attached to each other, often protruding ( Fig. 4D) and sometimes fused together (Fig. 4G).

coro mutants show disruption in actin cytoskeleton networks
The general appearance of all imaginal discs in coro ex6 , coro ex11 and coro ex6 /coro ex11 suggested that cells are loosely attached to each other. Trypsin (4.5 mg/ml) digestion of wing discs led to dissociation of coro ex6 and coro ex11 disc cells within 30-45 minutes, whereas wild-type wing discs took as long as 4 hours for complete dissociation of disc cells (Neufeld et al., 1998). coro ex11 wing discs stained with FITC-labeled actinphalloidin showed marginal defects in the cytoskeleton (Fig.  5B). However, actin-phalloidin staining of coro ex6 imaginal discs indicated that the actin-cytoskeleton is severely disrupted (Fig. 5C). The normal regular actin network was absent and we observed brightly stained spots, an indication that actin filaments are retracted and have formed actin-rich foci. One would predict that such severe disruption of the actin cytoskeleton in coro ex6 discs would be lethal to cells. Indeed, coro ex6 animals died at early pupal stages. Similar defects in the actin cytoskeleton were observed for coro ex6 /coro ex11 wing discs (data not shown).
In both invertebrates and vertebrates, disruption in the actin cytoskeleton leads to loss of cell adhesion properties, particularly loss of adherens junctions (reviewed by Hall, 1998). For example, mutations in the small GTPase Rho, which regulates the actin cytoskeleton, affect DE-cadherin-based cellcell adhesion in Drosophila embryos (Bloor and Kiehart, 2002). We therefore examined the effect of coro mutations on the expression of various markers of different spatial domains of epithelial cells. Along the apicobasal axis of epithelial cells, FITC-labeled actin-phalloidin, Armadillo (Arm) and Drosophila E-cadherin are localized at the apical end. Particularly, the latter two proteins label the site of the adherens junction, whereas antibodies to Discs large (Dlg), Fasciclin III (FasIII) and Coracle (Cor) label the more basal septate junctions (Woods, 1997). We observed moderate (in coro ex11 wing discs) to severe (in coro ex6 and coro ex6 /coro ex11 wing discs) reduction in actin-phalloidin staining in the apical surface of epithelial cells (Fig. 5B,C). In both coro ex6 and coro ex6 /coro ex11 wing discs, Arm and FasIII were also reduced and very often show diffuse expression (data shown only for coro ex6 ; Fig. 5E). Consistent with loss of actin-phalloidin staining, reduction in Arm levels was more severe than FASIII in both coro ex11 (Fig. 5G) and coro ex6 (Fig. 5H) discs. This by itself may not suggest loss of apicobasal polarity. Nevertheless, disruption in cell morphology is evident along the apicobasal axis, which is probably due to defective actin cytoskeleton.
Defective compartmental boundaries in coro mutant wing discs Homozygous coro ex11 wing discs are marginally smaller than wild-type discs. However, in as many as 30% (n=35) of coro ex11 discs the anterior (A) and posterior (P) compartments are not properly joined along the AP boundary. They show a deep cleft along the AP boundary (Fig. 6B). Since animals showing such severe wing disc phenotypes were early pupal lethal, we could not ascertain corresponding adult wing phenotypes. Homozygous coro ex6 wing discs do not show any such phenotypes in the AP or DV axes, probably because of severe defects in the actin cytoskeleton leading to cell death. Indeed, wing discs in coro ex6 and a trans-alleleic combination coro ex6 /coro ex11 are very small and often show total reduction of wing pouch area (Fig. 6C,D). A small number of coro ex6 /coro ex11 wing discs did show a deep cleft along either AP or DV boundary (Fig. 6D).
Reduced viability of peripodial cells of the wing disc (caused by the suppression of Dpp signaling) causes similar deep clefts along the AP and DV boundaries, more frequently in the AP boundary (Gibson et al., 2002). As mentioned above, 409-GAL4 is expressed only in the peripodial membrane ( Fig.  2A). Since mutation in the coro ex11 strain is the result of incomplete excision of this P-element, we examined the viability and morphology of peripodial cells in coro ex11 wing discs. DAPI-stained wing discs did not reveal any effect on the viability of peripodial cells (data not shown). We stained the wing discs with Arm antibody, which outlines all cells. Large peripodial cells were examined using a laser scanning confocal microscope. The observed cleft along the AP boundary of coro ex11 wing discs was restricted to the disc proper and the peripodial membrane did not show any defect. Individual peripodial cells also appeared normal (Fig. 6F). Thus, formation of the cleft along the AP boundary may not be related to the reported dependence of disc growth on the viability of peripodial cells.
The observed cleft was always along the AP boundary. Either it is caused by enhanced cell-sorting between A and P cells or by defective AP boundary formation. Alternatively, the observed cleft could be the result of irregular Dpp signaling from the boundary, thus causing disproportionate growth between A and P compartments. The AP boundary is specified by the activity of Hedgehog (Hh), which diffuses from the P compartment to the A compartment to activate Dpp expression. In the presumptive AP boundary, Hh activates Cubitus interruptus (Ci) by stabilizing its full-length isoform, which in turn activates Dpp expression (reviewed by Aza-Blanc and Kornberg, 1999;Ingham and McMohan, 2001). Dpp signaling from the AP boundary is important for growth along the entire AP axis.
We examined possible patterning defects in coro ex11 wing imaginal discs at the levels of Engrailed (En), Ci and Dpp staining. We did not see any obvious defects in En (data not shown), Ci and Dpp expression patterns (Fig. 7B,D) in wing imaginal discs. This suggests that specification of the AP boundary is normal. In addition, expression patterns of DV markers such as Wg (Fig. 7F), Cut (Ct) and Dll in coro ex11 wing discs (data shown) were also normal.
Spalt (Sal) and Optomotor blind (Omb) are the two wellstudied genes that respond to morphogenic activity of the Dpp gradient. In coro ex11 wing imaginal discs, we observed narrowing of the Sal expression domain in both anterior and posterior compartments; more severely in the latter compartment (Fig. 7H,I). Omb expression was also down regulated in the posterior compartment, while it was marginally Journal of Cell Science 117 (10) , coro ex11 /coro ex11 (B) and coro ex6 /coro ex6 (C) wing discs. A′ and A′′ show disc in A at higher magnifications. Similarly B′ and C′ are higher magnification images of discs shown in B and C, respectively. Note moderate to severe disruption of fine network of the actin cytoskeleton in B and C, respectively. coro ex6 /coro ex6 discs show a large number of darkly stained actin foci. Actin-phalloidin is normally localized to the apical surface of epithelial cells, which is reflected in higher intensity of staining at the DV boundary and at the presumptive hinge regions (A). In coro ex6 /coro ex6 discs, such intense staining of actin-phalloidin is lost (C) suggesting loss of actin-phalloidin from the apical surface. (D,E) Wild-type (D) and coro ex6 /coro ex6 (E) wing discs stained for Arm (green; D,E) and FASIII (red; D′,E′). D′′ and E′′ are the merged images of D and D′ and E and E′, respectively. Note that in coro ex6 /coro ex6 discs levels of both Arm and FASIII are very low and the staining is more diffused. (F-H) X-Z sections of wild-type (F), coro ex11 /coro ex11 (G) and coro ex6 /coro ex6 (H) wing discs stained for Arm (green) and FASIII (red). Note that normally Arm is apical and FASIII is basolateral. In wild-type wing discs both are expressed at very high levels. Both coro ex11 /coro ex11 and coro ex6 /coro ex6 show significant reduction in Arm and marginal reduction in FASIII staining indicating a possible defect in apicobasal polarity. In addition, cells are loosely arranged in corodiscs compared to wild-type disc, wherein cells are more compact. Normally coro ex6 /coro ex6 discs show severe loss of FASIII and Arm staining (as shown in E) and general loss of cell and disc morphology (as shown in C). To ascertain specific effects of loss of coro on apicobasal polarity of wing discs, we have chosen those coro ex6 wing discs, which show milder phenotype. affected in the anterior compartment (Fig. 7K). These observations suggest down regulation of Dpp signaling to cells away from the AP boundary.

Enhanced response to Dpp over-expression in coro mutants
In flies, Dpp is transported from the source to recipient cells through a cycle of endocytosis and exocytosis of Dppcontaining vesicles (Gonzalez-Gaitan and Jackle, 1999; Teleman and Cohen, 2000;Entchev et al., 2000;Arquier et al., 2001). Furthermore, it is proposed that the balance between recycling and degradation of the ligand in the endocytic pathway determines the gradient of Dpp morphogen (Entchev et al., 2000). In Dictyostelium, coronin is required to regulate trafficking and fusion of actin-coated vesicles (Rauchenberger et al., 1997;Gotthardt et al., 2002). We therefore examined the dynamics of the Dpp gradient in coro mutant discs using a biologically active Dpp::GFP fusion protein (Teleman and Cohen, 2000). We over-expressed Dpp::GFP in the AP boundary (using a dpp-GAL4 driver) in wild-type and in coro mutant backgrounds. Surprisingly, coro mutant (both heterozygous and homozygous) discs over-expressing Dpp::GFP were much larger than wild-type discs overexpressing Dpp::GFP (Fig. 8B). Since coro ex11 shows residual GAL4 activity, the phenotype in coro ex11 and coro ex6 /coro ex11 wing discs could be due to over-expression of Dpp::GFP in the peripodial cells. However, identical phenotypes were observed in the coro ex6 background, which does not have any GAL4 activity of its own. Thus, the observed over-growth phenotype is caused by an enhanced response of coro mutant cells to Dpp::GFP over-expressed in the AP boundary.
In the wild-type background, Dpp::GFP forms an unstable gradient that spreads rapidly in the disc (Teleman and Cohen, 2000). Bright punctate spots of over-expressed Dpp::GFP are visible only in cells close to the AP boundary, the source of the protein (Teleman and Cohen, 2000;Entchev et al., 2000) (Fig. 8C). In addition, both brightness and number of spots per unit area decrease with the distance from the source, suggesting the formation of a gradient of Dpp::GFP. In coro ex6 (C) coro ex6 /coro ex6 wing disc. Note that the disc is very small and no specific defect in the AP or DV boundary is seen. (D) coro ex6 /coro ex11 wing disc. Note both small-disc phenotype as in coro ex6 /coro ex6 discs and separation of dorsal and ventral compartments along the DV boundary. (E,E′′) Optical sections of anti-Arm-stained wild-type wing disc to show that the epithelium is contiguous in all focal planes. (F,F′′) Optical sections of anti-Arm-stained coro ex11 /coro ex11 wing disc to show that peripodial membrane is contiguous and large squamous epithelial cells of the peripodial membrane are normal. Also in the disc proper, the columnar epithelium is contiguous, although mainly at the basal end of the cells. Images in E,E′′ and F,F′′ were generated by taking nearly 60 optical sections each of 0.5 µm thickness, which were then used to reconstruct disc images at three equal focal planes. wing discs is normal, although they show defective AP boundary (arrows). (C,D) Expression pattern of Dpp in wildtype (C) and coro ex11 /coro ex11 (D) wing discs. Dpp is visualized with the help of dpp-GAL4/UAS-GFP. In coro ex11 /coro ex11 wing disc Dpp expression is normal despite separation of anterior and posterior compartments along the AP boundary (arrows). GFP seen in cells outside the AP boundary is mainly because of residual GAL4 activity of coro ex11 . (E,F) Wg expression patterns of wild-type (E) and coro ex11 /coro ex11 (F) wing discs. Wg expression along the DV boundary is normal in the coro ex11 wing disc. (G-I) Wild-type (G), coro ex11 /coro ex11 (H) and coro ex6 /coro ex6 (I) wing discs stained for Sal expression. Note reduced levels of Sal in coro ex11 and coro ex6 discs, particularly in the posterior compartment. Here too we have chosen those coro ex6 wing discs that show milder phenotype, to examine the effect on Dpp signaling. (J-K) Wild-type (J) and coro ex11 /coro ex11 (K) wing discs stained for Omb expression. Note that Omb expression in the posterior compartment is nearly absent. and coro ex11 (both heterozygous and homozygous) and coro ex6 /coro ex11 wing discs we observed bright spots of Dpp::GFP in all cells of the disc, even in cells farthest from the AP boundary (Fig. 8D,E). Dpp::GFP pattern appeared more uniform, suggesting that it has reached an equilibrium rather than a gradient with slopes on either side of the AP boundary. In addition, the size and number of bright spots of Dpp::GFP were higher in coro mutant discs than in wild-type discs (Fig.  8G). Since bright spots of Dpp::GFP represent a proteaseresistant form of Dpp enclosed within endocytic vesicles (Teleman and Cohen, 2000;Entchev et al., 2000), it is possible that Dpp turnover is slowed down in coro mutant discs. Wing discs double stained for Arm showed that Dpp::GFP bright spots are not internalized in corowing discs (Fig. 8G,I). They remain on the cell membrane, mostly on the inside surface of the cells, suggesting defective membrane trafficking in corowing discs.
Interestingly, coro ex6 /coro ex6 and coro ex6 /coro ex11 discs overexpressing Dpp showed a partial rescue of the actin cytoskeleton and apicobasal polarity as visualized with actinphalloidin, Arm and FasIII staining (Fig. 8J,K). This is consistent with the role of Dpp in maintaining the actin cytoskeleton Ricos et al., 1999;Martín-Blanco et al., 2000). Dcdc42 and JNK and Dpp pathways interact to maintain the normal actin cytoskeleton during dorsal closure in embryogenesis Ricos et al., 1999) and thorax closure during metamorphosis (Agnes et al., 1999;Martín-Blanco et al., 2000). Thus, these results suggest that severe loss of Dpp signaling could be the reason for defective active cytoskeleton associated with coro ex6 discs. This correlates well with the fact that coro ex6 animals die at late larval or early pupal stages (normally Dpp expression in wing discs is activated in late second/early third instar stages).

Over-expression of activated thickvein mimics coro phenotypes
Dpp is transported in endocytic vesicles, which would eventually be fused with lysosomes and degraded (Entchev et al., 2000). If Dpp turnover is slowed down in corowing discs because of impaired membrane trafficking and/or fusion with the lysosomes, it may lead to enhancement in Dpp signaling. If the amount of Dpp protein is a limiting factor, enhanced response would be seen only in the cells closer to the AP boundary, the source of the signal. The observed cleft along the AP boundary in homozygous coro ex11 wing discs could therefore be due to enhanced Dpp signaling in the AP boundary. To examine this, we over-expressed activated the receptor of Dpp, Thickvein (Tkv*) using a dpp-GAL4 driver in wild-type and in coro mutant backgrounds. Interestingly, over-expression of Tkv* in the AP boundary of wild-type wing discs caused mild cleft formation along the boundary (Fig. 9A). Over-expression of Tkv* in coro ex6 and coro ex11 heterozygous backgrounds caused severe cleft phenotype with 85% penetrance (n=35) along the AP Journal of Cell Science 117 (10) Fig. 8. Enhanced response to Dpp over-expression in coro mutants. (A,B) Low magnification images of dpp;GAL4/UAS-Dpp::GFP (A) and coro ex11 /coro ex11 ; dpp;GAL4/UAS-Dpp::GFP (B) wing discs. Note differences in the size of the two discs. coro ex11 and coro ex6 heterozygous and coro ex6 /coro ex6 and coro ex6 /coro ex11 flies too showed enhanced growth response to Dpp-overexpression (data not shown). (C-E) Dpp::GFP-expressing wild-type (C), coro ex11 /coro ex11 (D) and coro ex6 /coro ex6 (E) wing discs stained for both GFP (green) and Arm (red). In the wild-type background, bright punctate spots of over-expressed Dpp::GFP are visible only in cells close to the AP boundary, the source of the morphogen. In coro mutant discs, bright spots of Dpp::GFP are visible in all cells of the disc, even in cells farthest from the AP boundary. (F,G) Dpp::GFPexpressing wild-type (F) and coro ex11 /coro ex11 (G) wing discs shown at higher magnification. Note that in the coro ex11 disc, bright spots of Dpp::GFP are larger, more numerous and mostly associated with the plasma membrane. (H,I) Intensity profile of wing discs shown in C and D, respectively. Posterior compartment cells immediately adjacent and parallel to the AP boundary were used for intensity profiling in both the discs. Note large number of Dpp::GFP peaks (green) outside Arm peaks (red) in wild-type background. However, nearly all DPP::GFP peaks coincide with Arm peaks in coro ex11 /coro ex11 background suggesting that Dpp containing-vesicles are on the membrane. (J) coro ex6 /coro ex6 ; dpp;GAL4/UAS-Dpp::GFP wing disc stained with FITC-conjugated actin-phalloidin. Note considerable rescue of actin cytoskeleton. Compare this disc with that of coro ex6 /coro ex6 shown in Fig. 5C. (K) coro ex6 /coro ex6 ; dpp;GAL4/UAS-Dpp::GFP wing disc stained with Arm (green) and FASIII (red). Note that the levels of both Arm and FASIII are now comparable to the wild-type (shown in Fig. 5D).
boundary of the wing disc (Fig. 9B,C). Enhancement of the coro phenotype was more pronounced in adult wing blades. Overexpression of Tkv* in the wild-type background caused defects in anterior cross-vein development; otherwise the wing blade is normal (Fig. 9D). Being recessive, coro heterozygous flies do not show any phenotype in any tissue, including wing blades (data not shown). However, over-expression of Tkv* in the coro heterozygous background caused a dramatic deep cleft along the AP boundary of wing blades (Fig. 9E). These results, therefore, suggest that cleft formation is indeed due to localized enhancement of Dpp signaling.

Loss of syntaxin enhances coro phenotypes
The above-mentioned results on accumulation of Dppcontaining vesicles in corodiscs and enhanced Dpp signaling in the AP boundary suggest a role for Coro in vesicle transport. Independent evidence on this line came from studies on genetic interactions between coro and syx1A. The adult wing phenotype (notching) and eye phenotypes (reduction in the number of ommatidia, fusion of ommatidia and protrusion of head cuticle into the region of eye proper) and rudimentary ovaries observed in coro ex11 are remarkably similar to the phenotypes reported for syx1A alleles (Shulze and Bellen, 1996). Syx1A interacts with other members of the SNARE (soluble NSF attachment protein receptor) complex in membrane trafficking and in particular, it is required for membrane fusion. F-actin, coronin and vinculin are all part of transport vesicles (Gotthardt et al., 2002). Interestingly, loss of POD-1 (homologue of coronin) in C. elegans causes accumulation of large vesicles of endocytic origin. We therefore, examined if syx genetically interacts with coro.
We used a dominant negative form of Syx1A (DN-Syx1A; see Materials and Methods) for localized removal of its function in wild-type and coro backgrounds. Over-expression of DN-Syx1A using dpp-GAL4 driver in the wild-type background does not cause any significant phenotype in wing discs (Fig. 9F). Its expression in both coro ex11 and coro ex6 homozygous background enhanced the lethality associated with these alleles, suggesting genetic interactions between the two genes. In the coro ex6 homozygous background, all the animals died at early third instar larval stage. Over-expression of DN-Syx1A in coro ex11 homozygous background also caused higher incidences (78%; n=17) of cleft formation along the AP boundary of the wing disc (Fig. 9G), suggesting that coro phenotype is enhanced by the loss of syx1A.

Sequestration of Dpp in coro mutant background
Although over-expression of Tkv* causes cell-autonomous enhancement of Dpp signaling, it sequesters Dpp, leading to loss of signaling in the neighboring cells. For example, overexpression of Tkv* in the AP boundary causes reduction in the Sal domain and limits its expression to the AP boundary (Lecuit and Cohen, 1998) (Fig. 9H). Similar down regulation of Sal is observed with coro alleles suggesting that in corodiscs Dpp is not transported to the distant cells, causing its sequestration within the AP boundary. Over-expression of Tkv* in the corobackground resulted in severe loss of Sal expression in both anterior and posterior cells distant to the AP boundary (Fig. 9I,J), which suggests stronger sequestration of Dpp.
To conclude, adult phenotypes, cytoskeletal defect in wing discs, a cleft along the AP boundary, enhanced response to Dpp signaling and accumulation of large Dpp-containing vesicles observed in coro mutants are all linked and the main cause appears to be the disruption of F-actin-coated vesicle trafficking and/or its fusion with the membrane. Fig. 9. Over-expression of activated Tkv mimics coro phenotypes. (A-C) dpp;GAL4/UAS-Tkv* (A), coro ex11 /+; dpp;GAL4/UAS-Tkv* (B) and coro ex6 /+; dpp;GAL4/UAS-Tkv* (C) wing discs stained for Arm to outline the cells. Over-expression of Tkv* in the AP boundary causes cleft formation similar to that in corodiscs (Fig. 6B), although the phenotype is milder. Overexpression of Tkv* in coro ex6 and coro ex11 heterozygous backgrounds, however, results in the formation of a deep cleft, which is normally seen in coro ex11 homozygous background (Fig. 6B). Over-expression of Tkv* in coro ex6 and coro ex11 homozygous backgrounds caused very high levels of lethality in early larval stages. (D,E) Adult wing blades of dpp;GAL4/UAS-Tkv* (D) and coro ex11 /+; dpp;GAL4/UAS-Tkv* (E) flies. Note severe notching of the wing blade in E. coro ex6 /+; dpp;GAL4/UAS-Tkv* also showed similar phenotypes. (F,G) dpp-GAL4/UAS-DNSyx1A (F) and coro ex11 /coro ex11 ; dpp-GAL4/UAS-DNSyx1A (G) wing discs stained for Arm to outline the cells. Over-expression of DN-Syx1A in the AP boundary does not cause any significant phenotype. The severity and penetrance of cleft phenotype is enhanced when DN-Syx1A is expressed in coro ex11 background. G and G′ show two different levels of optical sections. coro ex6 over-expressing DN-Syx1A were lethal at early-third instar larval stages and therefore we were unable to examine third instar wing discs. (H,I) dpp;GAL4/UAS-Tkv* (H), coro ex11 /+; dpp;GAL4/UAS-Tkv* (I) and coro ex6 /+; dpp;GAL4/UAS-Tkv* (J) wing discs stained for Sal. Over-expression of Tkv* causes down regulation of Sal in distant cells, which is more pronounced in the posterior cells (Lecuit and Cohen, 1998). Sal levels are severely reduced in both anterior and posterior cells when Tkv* is over-expressed in coro heterozygous backgrounds.

Discussion
We have reported here the cloning and genetic characterization of the Drosophila homologue of coronin, a F-actin binding protein. The sequence analyses suggest that coro shares most features of coronin and coronin-like proteins. It has 2 WD repeats and an α-helical coiled coil domain at the C-terminal end. In the coro mutant background, we observed a number of phenotypes, such as disruption in actin networks and change in cell adhesion properties that are reported in the literature for loss of coronin function. Considering the very high sequence homology and similarities in loss-of-function phenotypes between coro and a number of other coronin genes from a range of species from Dictyostelium to human, suggest that coro is a functional homologue of coronin.
Our RNA in situ data suggests that Coro is expressed at high levels in all cells. This is further supported by the expression profile of coro during the life cycle of Drosophila as determined by Arbeitman et al. (Arbeitman et al., 2002). They carried out microarray-based gene expression studies of nearly one-third of all Drosophila genes, one of which is coro. Coro transcripts are seen at very high levels in 0-5-hour embryosreflecting maternal contribution. Its expression is drastically reduced in 15-20-hour embryos and maintains a steady state level until pupal stages. A slight increase in transcript level has been observed in pupal stages. Very high levels of maternal contribution of coro transcript is reflected in the observations that all coro alleles are either early or late pupal lethal. Interestingly, Enabled (another actin binding protein) and Costa (a microtubule binding protein) show the highest correlation of expression to Coro (Arbeitman et al., 2002). Developmental stages at which these genes show higher levels of expression may be stages in which actin and microtubule structures play significant roles.
We have generated a number of coro alleles, both viable and lethal. Viable coro mutants displayed prominent wing and eye phenotypes. In addition, all homozygous females of viable alleles are sterile with severe defects in ovary development. These phenotypes are very similar to those reported for syx1A alleles (Shulze and Bellen, 1996). Furthermore, we have observed that down regulation of Syx1A in the AP boundary is sufficient to enhance the cleft phenotype observed in corodiscs. Syx1A is part of the SNARE complex and mediates membrane trafficking and fusion. Coronin protein (in C. elegans and Dictyostelium) is also required for vesicle transport (Rauchenberger et al., 1997;Rappleye et al., 1999). Although further biochemical investigations are required for confirmation, our genetic experiments suggest that Drosophila Coronin protein functions with Syx1A to mediate trafficking and fusion of F actin-coated vesicles.
Over-expression of Dpp::GFP (a fusion protein, which otherwise functions as normal Dpp) in the corobackground (either heterozygous or homozygous) caused a dramatic overgrowth phenotype compared to its over-expression in wild-type wing discs. coro mutant wing discs over-expressing Dpp::GFP in the AP boundary showed large number of bright spots of GFP all along the AP axis suggesting more of Dpp in endocytic vesicles. Rapid turnover of Dpp is necessary to establish the morphogen gradient (Gonzalez-Gaitan and Jackle, 1999;Teleman and Cohen, 2000;Entchev et al., 2000). In corowing discs, either Dpp is not degraded or its degradation is slowed down, causing uniformly high levels of Dpp all along the AP axis and thereby causing the over-growth phenotype.
Wing discs heterozygous for coro mutations that were otherwise normal, showed high incidences of cleft formation when activated Tkv, the Dpp receptor in wing discs, was overexpressed in the AP boundary. This suggests that cleft formation is due to localized enhancement in Dpp signaling in the AP boundary. However, levels of Dpp targets in the cells distant to the AP boundary are down regulated in coro mutants, suggesting that Dpp is not transported to the distant cells. Enhanced Dpp signaling in the AP boundary, therefore, is attributed to its sequestration. The observed cleft along the AP boundary may reflect a fold caused by the overgrowth of anterior cells at the boundary and/or undergrowth of posterior cells.
However, over-expression of Dpp::GFP along the AP boundary (using dpp-GAL4 driver) caused enhanced growth all along the AP axis of corowing discs. Furthermore, we observed uniformly high levels of Dpp::GFP all over the coro mutant wing disc rather than a gradient, with high levels only in cells closer to the source of the signal. This appears to be contrary to the above-mentioned results, which suggested down regulation of Dpp signaling in cells away from the AP boundary. However, this may simply reflect the relative amounts of Dpp present in two genetic backgrounds. In corowing discs, the normal amount of Dpp is made in the AP boundary. If all of this Dpp is endocytosed, but not recycled because of defective exocytosis (and also not degraded since vesicles are not fused with the lysosomes for degradation) it may lead to sequestration of Dpp in the cells within and closer to the AP boundary. If the amount of Dpp protein was not a limiting factor (for example, when Dpp is over-expressed), there would be accumulation of Dpp in all the cells and as a consequence enhanced growth in the entire wing disc. The product of the Drosophila tumor suppressor gene lethal (2) giant larvae (l(2)gl) is required for the normal secretion of Dpp through exocytosis (Arquier et al., 2001). Similar to coro, l(2)gl is also required in the cytoskeleton for cell shape and polarity determination. In l(2)gl mutants too, slow turnover of Dpp could be the reason for the observed paradoxical overgrowth phenotype. Finally, Dpp::GFP spots/speckles were always associated with the plasma membrane in corowing discs, whereas they were mostly inside the cells in wild-type wing discs. This suggests that Coro is required either for exocytosis (which recycles Dpp-containing vesicles) or for the fusion of endocytic vesicles with the lysosomes for protein degradation. Work (at cellular and biochemical levels) is in progress to determine if coronin is needed for vesicle recycling or for its degradation or both.