Dcas Supports Cell Polarization and Cell-Cell Adhesion Complexes in Development

Mammalian Cas proteins regulate cell migration, division and survival, and are often deregulated in cancer. However, the presence of four paralogous Cas family members in mammals (BCAR1/p130Cas, EFS/Sin1, NEDD9/HEF1/Cas-L, and CASS4/HEPL) has limited their analysis in development. We deleted the single Drosophila Cas gene, Dcas, to probe the developmental function of Dcas. Loss of Dcas had limited effect on embryonal development. However, we found that Dcas is an important modulator of the severity of the developmental phenotypes of mutations affecting integrins (If and mew) and their downstream effectors Fak56D or Src42A. Strikingly, embryonic lethal Fak56D-Dcas double mutant embryos had extensive cell polarity defects, including mislocalization and reduced expression of E-cadherin. Further genetic analysis established that loss of Dcas modified the embryonal lethal phenotypes of embryos with mutations in E-cadherin (Shg) or its signaling partners p120- and β-catenin (Arm). These results support an important role for Cas proteins in cell-cell adhesion signaling in development.

Overexpression of Cas proteins contributes to the development of human cancer (reviewed in [3,16]). BCAR1 is required for Srcdependent cellular transformation of murine fibroblasts [17], and conserves with NEDD9 the ability to enhance the production of matrix metalloproteases [18], enhancing tumor cell invasion of the extracellular matrix (ECM) [19], promoting mammary tumorigenesis and lung metastasis in MMTV-HER2 and other mouse models of cancer [19,20]. NEDD9 has been defined as a component of an intracellular signaling switch that is important for epithelial-mesenchymal transition (EMT), based on activation of its downstream effector Rac [21]. TGFb promotes EMT during tumor cell invasion through the ECM, and tissue remodeling in development. In mammals, TGFb regulates both transcription and proteasomal degradation of NEDD9 [22,23]; conversely, BCAR1 and NEDD9 reciprocally bind and regulate the activity of a subset of TGFb effectors [24,25,26]. BCAR1 overexpression may predict aggressive estrogen receptor-negative cancers [27,28]. Overexpression of NEDD9 supports oncogenic signaling in malignancies of the hematopoietic system [13,[29][30][31][32], and has been linked to increased cellular invasive behavior in breast and colorectal cancer cell lines [18,33], squamous cell carcinomas of the head and neck [34], and enhanced metastatic potential in glioblastomas [35], melanomas [12], and some lung cancers [36], and to cell migration and EMT induced by chemical carcinogens [37]. Conversely, a null NEDD9 genotype significantly increases the latency of tumor incidence in the MMTV-PyVmT mammary cancer model [38].
While studies of the Cas group have emphasized important roles in cancer and other pathogenic conditions, little is known of their roles in normal development. Knockout of BCAR1 in mice leads to an embryonal lethal phenotype at day 11.5-12.5, associated with marked systemic congestion and growth retardation, and disordering of actin-based structures in the heart [17]. In contrast, knockout of NEDD9 results in viable, fertile animals, with minor defects in immune system maturation [39]. The presence of 4 paralogous family members with overlapping expression profiles [7], together with the difficulty of performing detailed phenotypic analysis in early embryonal development, have made it difficult to establish the required functions of Cas proteins in mammalian development.
By contrast, there is only a single Cas family protein in Drosophila, Dcas (CG1212). Dcas is highly expressed in the embryonic nervous system at stage 16 [40], as well as in the ventral ectoderm and ventral nerve cord primordial at earlier developmental stages (stages 9-12 [41]). The importance of DCas in Drosophila has been unclear. One recent study used an existing allele with a P-element insertion in an intron within the Dcas coding region, and a deficiency mutation overlapping Dcas and 5 adjacent genes, to establish a modifier role for Dcas in axonal fasciculation and axon guidance [40], but did not address the question of any potential early embryonal phenotypes. Although the protein is highly conserved with mammalian family members (68% with NEDD9 and 70% with BCAR1 [40]), null mutations in Drosophila orthologs of some of the most important mammalian interactors of Cas proteins, such as FAK (Fak56) [42] produce limited phenotypes. In the present study, we have used a FRTexcision-based strategy to delete the Dcas locus. Upon identification of an embryonal lethal phenotype affecting 10% of maternalzygotic null embryos, we subsequently extensively probed the genetic interactions of Dcas relevant to cell migration and EMT. This work indicated evolutionary conservation of core Cas family signaling involving FAK, Src, and integrins. Combination of mutations in Dcas and Fak56 perturbed localization of polarity markers, including particularly E-cadherin (Shotgun, Shg), implying that DCas might also interact with the E-cadherinassociated cell junctional proteins. Subsequent experiments directly testing this idea identified novel and potent genetic interactions between Dcas and the cell-cell adhesion proteins Shotgun, Armadillo and p120-catenin, influencing cell polarity. These findings inform the understanding of Cas protein action both in development and in cancer progression.

Generation and characterization of a Dcas null allele
To study Dcas function in Drosophila development, we used a modification of FRT-excision technology [43]. A FRT-containing P-element upstream of the Dcas gene was provided by a P-element located within 50 bp of the start of the Dcas open reading frame (ORF). A downstream transposon was provided by a Pbac located between the end of the Dcas coding sequences and the assigned start codon of the CG7049 ORF. Using this technique, we generated a precise excision of the complete Dcas ORF on chromosome 3 (Fig. 1A). The resulting allele, which we call Dcas 1 , contains a deletion spanning the first through final coding exons of the Dcas gene, but retains the Dcas promoter region and flanking genes, as confirmed by extensive quantitative PCR using probes directed against the DNA of Dcas and flanking genes (results not shown). Homozygous Dcas 1 /Dcas 1 mutants produce fertile progeny and can be maintained as a stable Dcas null strain.
To exclude the possiblity of secondary mutations contributing to any observed Dcas deletion-associated phenotypes, we used a number of discrete approaches to separately test the Dcas 1 strain. First, we crossed Dcas 1 /Dcas 1 stock to a stock containing the small Df(3L)Exel6083 deletion, which removes Dcas as well as Pk61c, CG6845, and CG7049 (Fig. 1A), to allow analysis of the phenotypes of Dcas 1 /Dcas 1 versus Dcas 1 /Df(3L)Exel6083 flies. Second, we crossed the Dcas 1 /Dcas 1 stock with a previously described Dcas P1 hypomorphic allele [40] which has a GAL4containing P-element inserted in the Dcas promoter, resulting in limited Dcas transcript levels, then analyzed Dcas 1 /Dcas P1 flies. Third, in Dcas P1 /Dcas P1 flies, we also introduced the DCas expressing a GAL4-activated UAS promoter fusion, a UAS-GFP-Dcas transgenic allele (as described in [40]), and assessed the Dcas P1 , UAS-GFP-Dcas/Dcas P1 , UAS-GFP-Dcas and Dcas P1 , UAS-GFP-Dcas/Dcas 1 phenotypes. Fourth, we assessed the mRNA expression of Dcas and flanking genes in the Dcas 1 /Dcas 1 and other mutant backgrounds.
Analysis of cDNA prepared from Dcas 1 /Dcas 1 stocks indicated complete absence of Dcas transcript. Dcas P1 /Dcas P1 mutants had significantly reduced but still detectable levels of the Dcas transcript ( Figure 1D). While the Dcas 1 /Dcas 1 stock had somewhat elevated expression of the adjacent CG7049 locus (which is predicted to encode a protein, but has no described phenotype or known function), the Dcas 1 /Df(3L)Exel6083 and Dcas P1 /Dcas P1 stocks did not: indeed, gene expression of CG7049 was diminished in Df(3L)Exel6083/+ stock. Together, these expression results argue against variation in CG7049 expression as contributing to the observed GBR/DC phenotype ( Figure 1D), and suggest the minimal phenotype observed with the DCasP 1 allele reflect the fact that this strain reduces (to 43% of wt) but does not eliminate DCas mRNA expression.
Synthetic lethality of Dcas 1 with FAK56 and Src42A mutations, and modifier interactions between Dcas and integrins The best defined signaling partners of Cas proteins in mammals are components of the integrin signaling network. For instance, in mammals, interactions of the Cas proteins with FAK, Src, and integrins are critical for cell migration [1,16]. We hypothesized that the weak Dcas 1 phenotypes might be exacerbated by additional targeting of the Drosophila orthologs of these genes. The Drosophila FAK ortholog, Fak56D, is not essential for viability or fertility and a null mutation, Fak56 CG1 , has no gross phenotypes associated with cell migration [45], although homozygous mutations in Fak56D have been associated with morphogenesis of the optic stalk in second and third instar larvae [42]. Drosophila have two Src-related genes, Src42A and Src64, which have redundant function in GBR and DC (with double mutants having phenotypes similar to those seen in 10% of Dcas 1 /Dcas 1 mutants, [46]), and other developmental processes [47]. Homozygous null alleles in Src42A have a high frequency of death before hatching or as first instar larvae, although some adult escapers of the hypomorphic allele Src42A JP45 have mild dorsal cleft phenotypes [46].
We first analyzed the genetic interactions of Dcas 1 with Fak56 CG1 . We created a double-balanced stock which carried both Dcas 1 and FAK56D CG1 mutations. Dcas 1 /Dcas 1 in combination with either heterozygous or homozygous Fak56 CG1 yielded no viable adult offspring (Table 1, Figure 2A). The Fak56 CG1 / Fak56 CG1 ; Dcas 1 /+ genotype also significantly reduced the viability of adults. Analysis of the Fak56 CG1 /Fak56 CG1 ; Dcas 1 /Dcas 1 lethal phenotype indicated that most (95%) of the embryos did not hatch. The few escapers survived to pupal stages, but did not emerge. We then crossed double mutants to a stock with a green compound balancer CyO-TM3-GFP, which constitutively expresses GFP from the Hsp70 promoter during all developmental stages, and selected Fak56 CG1 , Dcas 1 /CyO-TM3-GFP embryos, to separate double homo-and heterozygotes for analysis of the cuticles of double homozygotes (Figs 2A). In contrast to embryos with either the Dcas 1 or Fak56 CG1 homozygotes, only 5% of Fak56 CG1 /Fak56 CG1 ;Dcas 1 /Dcas 1 mutants produced cuticles (Fig. 2B), and almost all observed cuticles were marked by dorsal and/or ventral holes, indicating dorsal closure defects. Additionally, Fak56 CG1 /Fak56 CG1 ; Dcas 1 /Dcas 1 cuticles had fused or missing (not shown) denticle belts, phenotypes never observed in cuticles of Fak56 CG1 homozygotes alone.
Src42A k10108 is a mild hypomorphic allele of Src42A: Src42A k10108 /Src42A k10108 homozygous embryos hatch, but die as first instar larvae from defects in tail morphology, head involution, and tracheal necrosis [48]. However, we observed synthetic lethality in Dcas 1 /Dcas 1 ; Src42A k10108 /CyO adult flies ( Table 1,  Figure 2A, C) and Dcas 1 /Dcas 1 ; Src42A k10108 /Src42A k10108 embryos, as fewer than 46% of these double mutant embryos hatched (data not shown). Cuticles were assessed by a strategy similar to that described for DCas 1 6Fak56D CG1 mutants. 90% of the cuticles of Src42A k10108 /Src42A k10108 ; Dcas 1 /Dcas 1 embryos had holes in or absence of the head cuticle, and 10% of Dcas 1 /Dcas 1 ; Src42A k10108 / Src42A k10108 cuticles had additional GBR defects. Similar phenotypes were obtained crossing Dcas 1 /Dcas 1 to two additional Src42A mutant strains including the mild allele Src42A miri and the strong allele Src42A E1 (Table 1 and not shown). We confirmed the synthetic lethal interactions between DCas and Src42A and Fak56D, by in each case also assessing the phenotype of the Src and Fak alleles with Dcas 1 /Df(3L)Exel6083, in order to exclude the influence of potential secondary mutations in the Dcas 1 /Dcas 1 stock. Fak56 CG1 /CyO; Dcas 1 / Df(3L)Exel6083 adults did not emerge, indicating complete lethality. Src42A k10108 /CyO; Dcas 1 /Df(3L)Exel6083 and Src42A E1 / CyO; Dcas 1 /Df(3L)Exel6083 were semi-viable, emerging at approximately 50% of the rate of phenotypically normal adult siblings from the same cross (results not shown). Interestingly, the same cross to a double balanced Dcas P1 allele did not result in the same substantial decrease in the numbers of Src42A E1 /CyO; Dcas 1 /Dcas P1 progeny, implying that moderate expression of Dcas is sufficient to support the survival of Src42A E1 mutants.
Interestingly, the low percentage of Src42A k10108 /+; Dcas 1 / Dcas 1 and Src42A E1 /+; Dcas 1 /Dcas 1 adult escapers manifested wing blistering defects similar to those seen with mutants in integrin subunits [49] ( Figure 3A). These data implied that simultaneous reduction in Dcas and Src42A function combined to impact an important integrin-dependent effector pathway. Based on these results, we also assessed whether Dcas interacted genetically with Drosophila orthologs of integrin a (mew EY09631 and mew G0429 ; If B2 and If 3 ) and b (mys 1 ) subunits, which are upstream activators of SRC and FAK. Double balanced stocks of integrin mutants and Dcas 1 were crossed to make double heterozygous stocks of each mutant in combination with Dcas, and analyzed for adult viability and visible phenotypes ( Table 1).  K10108 and Dcas 1 in Cross 2) to allow analysis of the viability of resulting progeny. Each row in the graph represents percentage of viable progeny of indicated genotype. Total number of expected progeny was calculated from the number of phenotypically viable double balanced adult heterozygotes Fak56 CG1 /CyO; Dcas 1 /TM6B, Ubx, y+ in Cross 1 and Src K10108 /CyO; Dcas 1 /TM6B, Ubx, y+ in Cross 2, which we considered 100% viable. The percentage of viability for the remaining progeny was calculated in agreement with Mendel's law of independent assortment for two alleles. B. Cuticle preparations of stage 16 Fak56D CG1 /Fak56D CG1 and Fak56D CG1 /Fak56D CG1 ;Dcas 1 /Dcas 1 embryos transitioning to 1 st instar larvae, viewed laterally, ventral side to right. Arrows indicate holes in head and dorsal cuticle, * indicates missing and/or fused denticle belts. C Cuticle preparations of stage 16 Src42A k10108 /Src42A k10108 and Src42A k10108 /Src42A k10108 ;Dcas 1 /Dcas 1 embryos. Arrows indicates holes in head; * indicates GBR defect associated with incomplete DC. doi:10.1371/journal.pone.0012369.g002 Loss of Dcas in If mutants only moderately affected viability of adult flies: 27% of If 3 /If 3 ; Dcas 1 /Dcas 1 and 56% of If B2 /+; Dcas 1 /Dcas 1 emerge as adults ( Figure 3B). However, the characteristic wing blistering defects of the If integrin a mutants [49] were significantly exacerbated in all viable progeny of double homozygous If 3 /If 3 ; Dcas 1 /Dcas 1 (Table 1, Figure 3B). Further, loss of a single or both copies of Dcas in combination with mew EY09631 (a viable weak hypomorphic allele) caused a dramatic reduction in the viability of adult mew EY09631 homozygotes (Table 1). Moreover, 2% of mew EY09631 /+; Dcas 1 / Dcas 1 flies had wing blisters and smaller, more rounded wings ( Figure 3B). These interactions were enhanced using a lethal allele of mew (mew G0429 ), with viability of mew G0429 /+; Dcas 1 / Dcas 1 significantly reduced (Table 1). Finally, loss of Dcas significantly lowered the percentage of viable Dcas 1 /Dcas 1 ; mys 1 / + adults (Table 1), although no wing phenotypes were observed (not shown).

Dcas interacts with Fak56D to influence cell polarity and cytoskeleton
Based on the defined biology of mammalian Cas proteins (reviewed in [50]), the defects seen with Dcas and Fak56 mutant flies may reflect defects in cellular morphology (e.g., attachment and polarization) that inhibit appropriate migration during development. To begin to explore these mechanisms, we assessed the localization of markers of apical and basolateral polarity in flies with mutations in Dcas and Fak56. We compared localization of a set of polarity markers in embryos undergoing DC in mutants and wild type homozygotes and heterozygotes. Localization of polarity markers in heterozygous mutants was in all cases comparable to wild type (not shown). (Crumbs, CRB1) and aPKC localize to the subapical region/marginal zone in wild type embryos. For Crb, this localization was diminished in the 10% of phenotypically affected Dcas 1 /Dcas 1 mutant embryos, while in Dcas 1 /Dcas 1 ; Fak56 CG1 /Fak56 CG1 Dcas 1 /Dcas 1 embryos, Crb staining was generally reduced and diffuse in the cytoplasm (Fig 4A). aPKC staining was abnormally punctate specifically in the 10% of Dcas 1 / Dcas 1 embryos which had discernible GBR defects. Staining intensity of aPKC was both generally reduced and more cytoplasmically diffuse in Dcas 1 /Dcas 1 ;Fak56 CG1 /Fak56 CG1 embryos ( Fig 4A).
The septate junction markers Fas3 (fasciclin 3) and Dlg (Discs large), and an adherens junction marker, Shg (shotgun, Ecadherin), localize to the basolateral cell surface of epithelial cells. Fas3 expression and localization were unaffected in Dcas 1 / Dcas 1 ; Fak56 CG1 /Fak56 CG1 embryos beginning DC, although staining suggested a multi-layering of cells that was also indicated by the nuclear staining pattern obtained with the DNA label DRAQ5. Dlg staining patterns become more punctate and apical in Dcas 1 /Dcas 1 ; Fak56 CG1 /Fak56 CG1 embryos. Interestingly, Shg staining was markedly altered (Figs 4A-C) in embryos lacking Dcas, Fak56, or both. Shg staining in double mutant embryos was more cytoplasmic and diffuse compared to same stage embryos of other genotypes, with particular reduction of intense staining in the apical and lateral compartments. Detailed analysis of the intracellular distribution of Shg (Figs 4B, C) revealed significant reduction in the lateral punctate Shg staining in Dcas 1 /Dcas 1 ; Fak56 CG1 /Fak56 CG1 embryos. Although immunofluorescence analysis indicated Shg expression was maintained overall in Dcas null embryos, the 10% of embryos with GBR/DC defects had evidence of mislocalized Shg, with greater accumulation in a disorganized pool of Shg at the lateral and basal cell surface.
Dcas and Fak56 negatively regulate shg/E-cadherin protein localization in Drosophila embryos The loss of Shg from the adherens junctional complex might reflect defects in localization of the protein, or reduced Shg expression. To discriminate these possibilities, we analyzed extracts made from Drosophila embryos (stage 13-16), 2 nd -3 rd instar larvae, and adults (Fig. 5A). Quantitive Western blot analysis indicated that Dcas 1 /Dcas 1 embryos or larvae contain 2-fold higher levels of E-cadherin/Shg compared to WT, although no differences were seen in adult flies (Fig. 5A, graph). We next compared Ecadherin/Shg expression in Dcas 1 /Dcas 1 , Fak56 CG1 /Fak56 CG1 or Dcas 1 /Dcas 1 ;Fak56 CG1 /Fak56 CG1 embryos. A Fak56 CG1 /Fak56 CG1 genotype elevated E-cadherin protein levels to the same extent as Dcas 1 / Dcas 1 , while the double mutant had 3.1-fold more protein relative to wild type levels (Fig. 5B). E-cadherin transcription levels were not affected in Dcas and/or FAK mutants. The simplest interpretation of these results is that Shg does not effectively localize to lateral junctions in the absence of Dcas and Fak56. This defect initiates compensatory signals that modestly upregulate Shg at the level of translational control or protein stability, but this Shg remains trapped in the basal cytoplasmic cellular compartment.
Interestingly, Crb has been reported to support Shg localization to adherens junctions [51]. The fact that the localization of Crb, Dlg, and Shg, but not Fas3, was strongly affected in the assessed double mutants indicated that Dcas and Fak56 did not ubiquitously affect localization controls. Further, in Drosophila, the Src orthologs genetically and functionally interact with shg, and phosphorylate the Shg partner protein armadillo (arm, b-catenin) promoting its degradation [47,52]. As our data indicated Dcas both interacted with FAK56D and Src42A and resulted in mislocalized shg, these results together suggested close interactions between Dcas and Shg function.
shg is important for morphogenesis of the head and ventral epithelium [53]. Although homozygous shg 2 embryos complete DC due to abundant maternal contribution, some shg 2 /shg 2 embryos may have small irregularities of the leading edge [53]. However, no shg 2 /shg 2 embryos progress to 1 st instar larvae because of moderate (49%, Fig. 6B, panels i versus ii) to severe (42% and 9%, Fig. 6B, panels iii and iv) defects in the embryonic head and ventral cuticle. By contrast, none of shg 2 /shg 2 ; Dcas 1 /Dcas 1 embryos hatched; rather, shg 2 /shg 2 ; Dcas 1 /Dcas 1 double homozygotes arrested in late embryogenesis. Although most of the embryos formed at least partial cuticles, absence of Dcas significantly enhanced shg 2 /shg 2 cuticle defects, with all shg 2 /shg 2 ; Dcas 1 /Dcas 1 embryos showing severe defects that essentially eliminated head and ventral cuticle, while incomplete DC was reflected by the presence of holes in the dorsal cuticle (Fig. 6B, panel v).
To confirm specificity of the genetic interaction between Dcas and Shg, we subsequently crossed Dcas 1 with two other alleles of Shg, shg E17B and shg K03401 . Both alleles are embryonal lethal at embryonal stage 16, with embryos failing to produce head and ventral cuticle [53,55]. Double mutants shg E17B /shg E17B ; Dcas 1 / Dcas 1 and shg K03401 /shg K03401 ; Dcas 1 /Dcas 1 were lethal or semilethal and had similarly enhanced cuticle defects, as observed with shg 2 /shg 2 ; Dcas 1 /Dcas 1 , indicating the genetic interaction observed was not allele-specific.
We extended our analysis to two additional alleles of arm (arm 3 and arm 8 ). Like arm 2 , these alleles express a truncated form of Armadillo [58,59], due to either amino acid replacements resulting in a stop codon (arm 2 and arm 3 ) or p-element insertion (arm 8 ) within the arm repeats. arm 3 is embryonic lethal typically at stage 16, and is characterized by segment polarity and DC defects, while arm 8 undergoes normal DC, but has strong segment polarity defects and dies in pupae. Dcas 1 genetically interacts with both arm 3 and arm 8 , as only 40% of the expected adult progeny with the genotypes arm 3 /FM7i-GFP, B; Dcas 1 /Dcas 1 and arm 8 /FM7i-GFP, B; Dcas 1 / Dcas 1 can be recovered ( Table 2).
The p120-catenin homozygotes are viable and fertile; p120ctn 308 mutation has been reported to induce a delayed, but complete DC and subtle irregularities of the leading edge in the majority of mutant embryos [60]. A p120ctn 308 /p120ctn 308 ; Dcas 1 /Dcas 1 genotype caused misalignment of segments and fused denticle belts (Fig. 7D, right panel, arrow), phenotypes not observed in p120ctn 308 /p120ctn 308 mutants (Fig. 7D). Double mutant embryos p120ctn 308 /p120ctn 308 ; Dcas 1 /Dcas 1 successfully complete embryogenesis and form larvae, but produce few pupae and no adult flies ( Figure 7B and Table 2). The later point of lethality may indicate a less direct interaction than that between Dcas, shg, and arm.

Specificity of Dcas genetic interactions
To rule out the possibility of secondary hits accumulated during double balancing and multiple crosses influencing phenotypes, we chose strongest shg and arm alleles (shg 2 , shg E17B and arm 2 ) double   indicating sensitivity of shg mutants to even slight loss of Dcas expression. Finally, we also explored other potential Dcas genetic interactions suggested from studies of the mammalian Cas paralogs NEDD9 and BCAR1. In mammals, NEDD9 interacts with Aurora-A kinase to regulate cell cycle [61], while BCAR1 interacts with the adaptor protein NCK in growth factor signaling [62]. Combination of Dcas 1 with mutants in the Drosophila ortholog of Aurora-A (the amorphic alleles Aur 1 or Aur 87Ac-3 ) or of Nck (dreadlocks, the amorphic dock 04723 allele) resulted in no synthetic lethality (Table S1). These negative results suggest a more ancestral and specific relationship of Dcas with the other genes yielding positive phenotypes.

Discussion
This work identifies a strong interaction between the Dcas, and integrin pathway genes, including integrins and their effector kinases Fak56D and Src42A, during early embryonal development in Drosophila. The synthetic lethal phenotypes found in double mutants of Dcas and Src or FAK56D were marked by defects in dorsal closure and in some cases by the appearance of anterior cuticle holes that suggested head involution defects. These defects were commonly accompanied by abnormalities in epithelial function, including failure to appropriately localize shg/E-cadherin to cell junctions, and reduced shg expression. Our data are compatible with the idea that either Fak56D or Dcas is sufficient to support shg/E-cadherin localization and cell polarization during morphogenetic movements in Drosophila embryos, but the absence of both cannot be sustained.
Building from these observations, we established a novel synthetic lethal relationship between DCas, shg, and arm. As with crosses to alleles of Fak56D and Src42A, the point of lethality was at the time of dorsal closure, at embryonal stages 15-16, and associated with defective cuticle formation. One way to integrate these observations is to hypothesize that the DCas, Fak56D, and Shg protein products are normally in dynamic balance, with Dcas regulating Shg cycling. The fact that Crb and Dlg1, a mammalian homolog of Dlg, have been reported to support Shg localization to adherens junctions [51,63], suggests that Dcas/Fak56/Src42A specifically interact to support this cell polarity/cell junctional control system. In this context, it is suggestive that the Crb family protein CRB3 has been described as part of a complex including CRB3, Pals1, and PatJ that becomes tightly associated with Src kinase during reorganization of cell polarity [64]. In the absence of DCas and Fak56D, Shg cannot localize properly; the moderately elevated levels of Shg proteins found in these embryos most likely arises as part of a cellular compensatory mechanism in response to decreased functional Shg signaling complexes. In further indirect support of the idea that this is a specific Dcas action, the fact that genetic interactions were not observed between Dcas 1 and Aur or Dock indicates that Dcas does not promiscuously interact with other genetic lesions to reduce viability.
A previous study demonstrated a role for Dcas in axonal guidance in the development of the nervous system of adult flies [40]. That work analyzed the hypomorphic Dcas mutant allele Dcas P1 , and the small deficiency Df(3L)Exel6083, including Dcas and five adjacent genes, which we have also used in this study. The earlier study focused exclusively on analyzing the contribution of Dcas to axonal guidance in late (stage 16/17) embryos: in that analysis, Dcas functioned similarly to integrins, and genetically interacted with integrins (if, mew, and mys) in regulating axon guidance and axonemal defasciculation. In this context, it is intriguing that the mammalian Cas family NEDD9 gene is abundant during neuronal development, has been proposed as a candidate locus for oral cleft defects in humans based on its chromosomal location near the OFC-1 locus [65], and has recently been implicated in control of neural migration and neuronal cell fate [66,67]. Together these findings raise the possibility that this specific Dcas paralog has a specific role in human neuronal migration and morphogenesis of the head. As with our data using the new Dcas 1 allele, homozygous deletion of Dcas in conjunction with integrins had moderate effect on viability of adult flies, although our work for the first time demonstrates an interaction between Dcas and if and mew, and also between Dcas and Src, in regulation of wing development.
Generation of the first null allele of Dcas provides a useful new tool to study the role of this protein in Drosophila development. This work illuminates the evolutionary conservation of Dcas function within the integrin and receptor tyrosine kinase network, including FAK, Src, and integrins genes. The finding that a low percentage of embryos with mutant Dcas and all embryos with double mutations in Dcas and Fak56D, have perturbed localization of polarity markers, including Shg, indicates a novel function for Cas family in regulation of cell polarity. To date, the evidence directly connecting Cas proteins to a known mechanism for control of cell polarity is sparse. Although NEDD9 was in fact discovered in a functional genomics screen for cell cycle and polarity modifiers in budding yeast (leading to its designation as HEF1, Human Enhancer of Filamentation 1) [5], the mechanism involved was not established, and given the great evolutionary distance involved, may not be relevant to a role in metazoans. Both BCAR1 and NEDD9 interact physically with proteins that influence cell polarity controls during pseudopod extension and other actin polarization processes: these include the GTP exchange factor AND-34 [68] and Rac1 [69].
Our data in the present study indicating genetic interactions with cell-cell junction regulatory proteins Shg, Arm and p120catenin may have considerable significance in the sphere of cancer research, as it implies that overexpression of Cas proteins may promote cancer progression by influencing the polarized movement of cells and influencing lateral attachments. The fact that one report has indicated interactions between BCAR1 and nephrocystins at cell-cell junctions in polarized epithelial cells [70] implies that a potentially direct interaction of Cas proteins in these structures is conserved through mammals. However, given the known interactions of Cas proteins with FAK and SRC at focal adhesions, another possibility is that Cas may additionally or alternatively impact Shg function through indirect signaling emanating from these structures. Notably, Bui et al. recently reported that NEDD9 overexpression induced by dioxin caused downregulation of E-cadherin [37], and it will be of great interest to study the consequences of overexpressing Dcas on Drosophila development. Consequences for loss of NEDD9 expression on Ecadherin expression or localization are not yet known. Resolving these questions will provide intriguing directions for future studies.

Construction of CG1212/Dcas knockout and viability calculations
To generate a Dcas null allele, yw hs-flp; FRT82neurIF63/Tm3, Sb' females were crossed to FRT-containing PBac{WH}f00059 (BDSC, Bloomington, IN) males to produce yw hs-flp/Y; PBac{WH}f00059/Tm3, Sb' males, which were backcrossed to yw hs-flp; FRT82neurIF63/Tm3, Sb'. Next, yw hs-flp; PBac{WH}f00059/Tm3, Sb' females were crossed with w, P5-HA-2428 males (Szeged Drosophila Stock Centre, Hungary). Male progeny with the genotype yw hs-flp/Y; PBac{WH}f00059/ P5-HA-2428 were backcrossed to yw hs-flp; PBac{WH}f00059/ Tm3, Sb' to obtain yw hs-flp; PBac{WH}f00059/P{RS5}5-HA-2428 flies. We then initiated double stranded breaks and isolated knockouts as described in [71]. Endpoints of excision were defined by the P-element P{RS5}5-HA-2428, positioned within 50bp from the Dcas start codon and containing an FRT site in the same orientation as the piggyBac Pbac{WH}f00059 transposon, located in between the Dcas ORF and the adjacent downstream gene CG7049 (Fig. 1A). Flippase-activated excision produced 20 potential mutant stocks, which were then analyzed by quantitative RT-PCR with seven sets of primers spanning the Dcas coding region, and in flanking genes (Fig. 1A). Correct endpoints of the Dcas deletion were initially confirmed using qRT-PCR to analyze DNA from adult flies, using primers directed at the promoter, and first and last coding exons of Dcas, as well as flanking upstream and downstream chromosomal sequences, to confirm that the Dcas gene was not detectable in Dcas 1 /Dcas 1 mutant stocks, although readily detected in WT flies. Dcas transcript levels were also measured in knockout flies using qRT-PCR to analyze at least 3 independent samples of RNA prepared using the RNeasy kit (Qiagen) from adult flies and larvae. Dcas 1 null flies were serially backcrossed to 3 rd chromosome balancer stocks (TM3,TM6 and TM2) to exclude additional recombination-associated mutations on other chromosomes before further characterization.

Preparation of embryos and immunohistochemistry
For analysis of localization of polarity markers, embryos were prepared as described by [72], with minor modifications. Briefly, embryos collected off apple-agar plates were washed in 50% bleach for 2 minutes, then rinsed twice in PBS and gently shaken on a platform for 40 minutes in 1:1 8% PFA/heptane mixture containing 2 units/ml of phalloidin (Invitrogen, Carlsbad, CA). Embryos accumulating at the interface between PFA and heptane were collected, applied to double stick tape, then de-vitellinated with a fine glass needle. Rehydrated embryos were blocked with 10% BSA/PBS for an hour, washed in 1% BSA/PBS, and incubated with a primary antibody for 2 hours, followed by two 5minute washes in PBS, and incubation with secondary antibody for an hour. Embryos were visualized with Leica TCS SL and Nikon C1 confocal microscopes, and images analyzed using Metamorph and EZ-C1 freeViewer software.

Western blot analysis
Whole embryo, larval and fly lysates were prepared by homogenization in 36 Laemmli sample buffer containing 10% SDS, and then boiling for 5 minutes. Samples were separated by 10% Bis-Tris NuPage PAGE (Invitrogen, Carlsbad, CA). Western blots were performed using standard protocols.

Cuticle preparation
Embryos were collected for 26 hours after the parents were removed from the apple-agar plates, dechorionated with 50% bleach, devitellinized in 1:1 heptane/methanol for 15 minutes, washed 26 with methanol, 26 with lactic acid, transferred on a slide into a drop of Hoyer's medium, and photographed using a phase contrast microscope (Nikon Eclipse TE-2000-U). The stocks carrying mutations in a gene of interest and Dcas were rebalanced over a green compound balancer to separate GFP-negative double homozygotes and GFP-positive double heterozygotes. Homozygotes were then collected for cuticle preparations. Where applicable, GFP-negative homozygotes and GFP-positive heterozygotes were separated using fluorescent dissecting microscope, prior to bleaching.

Wing preparation
Wings were detached and mounted into Hoyer's medium. Images were taken at 46 magnification on Leica TCS SL microscope.

Embryo Hatch Rate Analysis
Mutant embryos were collected from indicated crosses. For hatch rate determination, embryos were collected on apple-agar plates for 4 hr, than the parents were removed and more than 500 embryos per genotype were counted after 2 days' incubation.

Real Time PCR analysis
Stage 13-15 embryos were bleached and rinsed with distilled water. 30 mg of embryos were used to isolate mRNA with the RNAeasy kit (Qiagen, Valencia, CA) and overall mRNA quality assessed using an Agilent Bioanalyzer (Agilent, Santa Clara, CA). The concentration of Shg mRNA in was quantified via real-time PCR with the Smartcycler detection system (Cepheid, Sunnyvale, CA) using SYBR green I (Molecular Probe, Eugene, OR) in three independent experiments. The primers used in the PCR reaction were as follows: Shotgun, forward primer 59-GCGCTACGAC-GAATCCATG-39 and reverse primer 59-AGATAATACCC-GACTCCTTGTCAATC-39; and as a normalization control, the housekeeping gene RpII140, forward primer 59-CGCACGTG-GAAGTTGGTAAT-39 and reverse primer 59-ACAATCA-GAGTCCGCGTA ACAC-39.

Supporting Information
Table S1 Additional genetic interactions of Dcas 1 . For data shown, the parental crosses for Dcas 1 and dock 04723 were performed as described in Materials and Methods. Alleles of Aurora kinase (aur 1 and aur 87Ac-3 ) were first recombined to position both mutations to the same chromosome with Dcas 1 , balanced over TM3, Ser balancer to establish a double heterozygous stock (i.e. aur 1 , Dcas 1 /TM3). Double heterozygotes were crossed with Dcas 1 /Dcas 1 to produce aur 1 , Dcas 1 /Dcas 1 , aur 87Ac-3 , and Dcas 1 /Dcas 1 which were then crossed back to double heterozygotes. The viable progeny of indicated genotypes was collected and compared to phenotypically normal double heterozygous siblings, (i.e. aur 1 , Dcas 1 /TM3) in each of 3 independent experiments. Found at: doi:10.1371/journal.pone.0012369.s001 (0.03 MB DOC)