blistery encodes Drosophila tensin protein and interacts with integrin and the JNK signaling pathway during wing development

Tensin is a focal adhesion molecule that binds to actin filaments through its N terminus (Lo et al.,1994a; Lo et al.,1994b; Chuang et al.,1995). In addition, it contains two functional motifs, including a Src homology domain 2 (SH2) and a phosphotyrosine binding (PTB) domain(Davis et al., 1991; Chen et al., 2000; Chen et al., 2002). This conserved domain structure gives significant clues to its possible function,including potential roles in cell signaling.

Tensin is best known as an adaptor protein linking integrin to the actin cytoskeleton. Integrins are a family of the transmembrane receptors that are localized in focal adhesions (Hynes,1992). The extracellular domain of integrin interacts with the extracellular matrix, and its cytoplasmic domain anchors actin filaments to the plasma membrane through the focal adhesion protein complex(Burridge et al., 1988; Hynes, 1992; Lo et al., 1994a; Jockusch et al., 1995). Integrin is also believed to participate in diverse biological events such as cytoskeletal restructuring, cell motility and even cell survival via focal adhesion complexes that include tensin, focal adhesion kinase (FAK), Src kinase and protein kinase C (Burridge et al., 1988; Schwartz et al.,1995; Giancotti and Ruoslahti,1999). Most interestingly, integrin is involved in various cell signaling pathways through its association with focal adhesion proteins. For example, integrin activates ERKMAP kinase by promoting the SH2 domain-mediated association of Grb2 with tyrosine kinases such as FAK and c-Src in focal adhesions (Schlaepfer et al.,1994). The phosphoinositide 3 kinase (PI3K)-dependent signaling pathway is also activated by integrin through FAK-dependent mechanism(Chen and Guan, 1994; Shaw et al., 1997; Reiske et al., 1999). In addition, several reports have demonstrated that JNK is also activated by integrins when cells are attached to the extracellular matrix(Miranti et al., 1998; Oktay et al., 1999).

The role of tensin as an adaptor for integrin and as a required component in focal adhesions suggests the possibility that it may act as a mediator of integrin signaling. In support of this idea, much indirect and direct evidence has been collected. Previously, the SH2 and PTB domains in the C terminus of tensin have been demonstrated to bind tyrosine phosphorylated proteins such as PI3K and p130 CAS (Salgia et al.,1995; Salgia et al.,1996; Auger et al.,1996). In addition, tensin itself is phosphorylated at serine,threonine and tyrosine residues when cells are stimulated by either cell adhesion (Bockholt and Burridge,1993), growth factors (Jiang et al., 1996) or oncogenes, including v-Src and Bcr/Abl(Davis et al., 1991; Salgia et al., 1995). Indeed,a recent study has shown that overexpression of tensin alone can activate JNK in human embryonic kidney 293T cells (Katz et al., 2000). According to this report, the tensin-mediated JNK activities are independent of the activities of small GTP-binding proteins such as Rac and Cdc42, but dependent on the activity of SEK(Katz et al., 2000).

In the present study, we have shown that by, one of the previously reported genes to result in a blistered wing phenotype when disrupted, encodes the Drosophila ortholog of mammalian tensins, and with the by mutants, we were able to characterize the functions of tensin in vivo. The blistered wing phenotype of the by loss-of-function flies was demonstrated to be caused by a defective wing unfolding process after eclosion. Additionally, using a genetic approach and immunohistochemistry, we have shown that tensin functionally interacts with integrin and the JNK signaling pathway. Our results demonstrate the in vivo roles of tensin in development and suggest that tensin might be a transducer of signals from integrin to the JNK signaling pathway.

The Upstream Activation Sequence (UAS) fly lines of basket(bsk; Drosophila JNK), hemipterous (hep; Drosophila MKK7) and rolled^Sem(rl^Sem; a gain-of-function allele of Drosophila ERK) were gifts from Drs M. Mlodzik (EMBL-Heidelberg, Germany), T. Adachi-Yamada (Kobe University, Japan) and D. Bohmann (University of Rochester, USA), respectively. hep¹ fly line was kindly provided by Dr S. Noselli (CNRS, France). The fly lines if³, bsk¹ and by¹were obtained from the Bloomington Stock Center (Bloomington, IN). MS1096-GAL4 driver line was a gift from Dr M. Freeman (MRC Laboratory of Molecular Biology, UK).

The blistery mutant by² is a P-element insertion line. The by² line was isolated accidentally during the generation of UAS lines of a Drosophila EST clone,SD01679. Of many UAS lines generated, this line was distinct from others,showing a blistered wing phenotype and partial sterility for the females. At first, these features made us speculate that they were caused by the inserted transgene, but as other transgenic lines of the EST inserted at other loci did not display the same phenotype, and the imprecise excision lines also showed the same phenotype as their original line, we concluded that the phenotype was the outcome of a gene disruption caused by the inserted P-element. The identity of the disrupted gene was revealed by performing inverse-PCR and sequencing the regions flanking the P-element. The P-element insertion site in this line was 28 bp upstream of the by open reading frame (ORF),which encodes Drosophila tensin protein, and this mutant line was thus named by². P-element excision alleles were generated as Horowitz and Berg performed previously(Horowitz and Berg, 1995),except that TMSΔ2-3 was used as the transposase source and white was a background mutation. The molecular characterization of each mutant line was performed by genomic DNA polymerase chain reactions(PCR). We generated six revertants and two excision alleles of by² from a P-element excision experiment. In by^ex49, one of the imprecise excision alleles, 3.56 kb of genomic region including most of by exon was deleted, and in the other imprecise excision line, by^ex10, 1.1 kb fragment of the original P-element remained in the 5′ upstream of the bylocus (Fig. 3A).

Full-length and partially deleted by genes (ΔN: amino acids 1-450 deleted, ΔC: amino acids 451-720 deleted, ΔPTB: amino acids 567-720 deleted) were each subcloned into separate pUAST vectors. Standard procedures for the egg injection were used for the transformation of Drosophila (Ashburner,1989), using a microinjector model IM300 (Narishige, Japan) and an Axiovert 25 micromanipulator (Carl Zeiss, Germany).

Total RNA, extracted by the easy-Blue^TM system (Intron, Korea),was separated by electrophoresis on denaturing formaldehyde agarose gels in MOPS buffer, transferred onto a nylon membrane, and successively hybridized with nick-translated ³²P-labeled cDNA probe. The probe was made with the PCR product that corresponds to the 3′ region of the by amplified with the primers(5′GGTGTCCGCCGATTCCGTACAATTT3′ for 5′ and 5′GGCCCGCACCGGTGGTAC3′ for 3′). Hybridized probes were visualized by autoradiography.

For RT-PCR, total RNA was isolated from wild-type and the bymutant adult flies using the easy-Blue^TM system (Intron, Korea),and reverse transcribed into cDNA. PCR was performed with the byprimers. One set of PCR primers (5′ATGCGGACGCCGTACGAAGAAAG3′ for 5′ and 5′CACATTCGTATTCGTTGGCC3′ for 3′) was designed to amplify the 5′ portion of by (designated as `a' in Fig. 6A), and the other set(5′ATGCGTCACTTCCTCATCGAG3′ for 5′ and 5′CGAGGAGACCTTGAAGTGCAC3′ for 3′) was prepared to amplify the 3′ region of by (designated as `b' in Fig. 6A).

In situ hybridization experiments were performed with a digoxigenin-labeled RNA probe corresponding to the `b' region of by(Fig. 6A), as previously described (Cho et al.,2001).

Drosophila wing imaginal discs were dissected and fixed in 4%paraformaldehyde in phosphate-buffered saline (PBS). They were then washed in PBST (PBS + 0.1% Tween 20) and blocked in PBST with 3% bovine serum albumin. The samples were incubated overnight at 4°C with either anti-beta PS integrin mouse antibody [1:200 dilution; a gift from Dr D. L. Brower(University of Arizona, Tucson, AZ)] or anti-phosphospecific JNK rabbit antibody (1:200 dilution; Promega, WI). Next, the samples were further incubated for 4 hours at room temperature in either FITC-labeled anti-mouse secondary antibody (1:200 dilution; Sigma, MO) or HRP-conjugated anti-rabbit IgG secondary antibody (1:200 dilution; Molecular Probes, OR), and analyzed using an LSM510 laser confocal microscope (Carl Zeiss, Germany). Alexa Fluor 568 tyramide (Molecular Probes, OR) was used as a substrate for HRP-conjugated anti-rabbit IgG secondary antibody. In order to visualize the actin structure,pupal wing discs were stained with phalloidin-TRITC (Sigma, MO) for 20 minutes. For Acridine Orange (Sigma, MO) staining, third-instar larval wing discs were dissected in PBS and incubated for 5 minutes in 1.6 μM Acridine Orange-PBS solution. The samples were observed by a confocal microscope (Carl Zeiss, Germany).

The sole Drosophila ortholog of mammalian tensins, CG9379 (GenBank accession number: NM_141644, Protein_ID: NP_649901), was identified in the Drosophila genome data bank by BLAST search, and corresponds to the open reading frame deduced from Drosophila EST clone, RH56077(GenBank accession number: AY094941) (Fig. 1A). Drosophila tensin has a predicted molecular weight of 79 kDa, with a much shorter N-terminal region than that of mammalian counterpart. Although the N-terminal sequence of tensin is not conserved, its C-terminal region of about 350 amino acids, which includes the SH2 and PTB domains, exhibits significant homology to the human homologs, with about 40%amino acid identity (Fig. 1B).

To determine the expression stage and transcript complexity of Drosophila tensin, we performed Northern blot analyses at various developmental stages using a probe derived from the Drosophila tensin cDNA (see Materials and Methods). A single transcript of about 3.0 kb was detected throughout all the developmental stages(Fig. 1C).

Next, we decided to investigate the in vivo function of Drosophilatensin using genetic approaches. As described in the Materials and Methods, we found one P-element insertion line, by², containing a P-element in the 5′ flanking sequence of the Drosophila tensin open reading frame (Fig. 1A). To examine whether the P-element insertion in by² hampers the transcription of Drosophila tensin, we performed RT-PCR, which showed highly reduced Drosophila tensin expression level in the whole body (Fig. 3C). In addition to by², six revertants and two imprecise excision alleles of by² were generated as described in the Materials and Methods. These newly obtained mutants were verified by genomic PCR and subsequent sequencing analyses. The expression levels of Drosophilatensin in these lines were also examined by RT-PCR. The transcript levels of Drosophila tensin were completely normal in the revertant lines (data not shown) compared with wild type. By contrast, Drosophila tensin expression was highly reduced or completely missing in the imprecise excision alleles (Fig. 3C).

Homozygous but not heterozygous by² mutant flies showed blisters in their wings (Fig. 2A). To see whether this phenotype is related to Drosophila tensin expression, in situ hybridization was performed in wing imaginal discs. As shown in Fig. 1D, Drosophila tensin expression was dramatically reduced in the by² mutants(Fig. 1D, right panel) compared with the control (Fig. 1D, left panel). Interestingly, the Drosophila tensin transcript was highly enriched in the wing pouch (Fig. 1D, left panel). We confirmed that the blistered wing phenotype was indeed caused by a deficiency in Drosophila tensin expression, by generating transgenic flies specifically overexpressing Drosophilatensin within a homozygotic by² genetic background. Ectopic expression of full-length Drosophila tensin in wing imaginal discs dramatically rescued the wing blister phenotype of by² (Fig. 2B). Moreover, two additional Drosophila tensin loss-of-function alleles, by^ex10(Fig. 2C) and by^ex49 (data not shown) failed to complement the blistered wing phenotype of the by² mutants. However, by^rv8, one of the revertants, fully complemented the by² mutation (Fig. 2D). These results unequivocally demonstrate that the blistered wing phenotype in the Drosophila tensin mutants is due to a defect in Drosophila tensin function.

We searched all the reported mutant lines with a blistered wing phenotype that mapped around the cytogenetic location of Drosophila tensin,85D22, and found one mutant line, blistery¹(by¹) - the loss-of-function allele of the blistery (by) gene. The mutation site of by¹ has been reported to be located around 85D, but specific details have not yet been determined. The by¹mutants display blisters in the wing subterminal region (Glass et al., 1934; Prout et al., 1997; Walsh and Brown, 1998), which is identical to the phenotype of the Drosophila tensin mutants such as by² (Fig. 2A) and by^ex49 (data not shown).

Significantly, the transheterozygotic alleles of by¹and by² failed to complement each other's blistered wing phenotype (Fig. 2F), implying that they are alleles of the same gene. Consistently, by^ex10 and by^ex49 also failed to complement the blistered wing phenotype of by¹ (data not shown).

To further confirm that the Drosophila tensin mutants are allelic to by, we sequenced the Drosophila tensin locus in the by¹ mutants. As expected, we found that by¹ allele contains two missense point mutations that convert tyrosine 62 and threonine 163 to asparagine (Y62N) and arginine(T163R) in tensin, respectively (Fig. 3A). Therefore, we conclude that our Drosophila tensin mutants and by¹ are different alleles of by, and hereafter we refer to Drosophila tensin as by.

The by mutant flies exhibited a blistered wing phenotype with varying degrees of severity, which can be classified into three groups; normal wing, class I wing which contains a small blister (< 1/4 size of the total wing area) within a restricted region, and class II wing with a large blister(> 1/4 size of the total wing area) resulting in a crumpled wing phenotype(Fig. 3B). We determined the severity of the wing blister phenotype by measuring the frequency of each wing class. By comparing the severity in various mutants(Fig. 3B) and their byexpression levels (Fig. 3C), we found that the severity of wing blisters is closely related to the level of the by transcript. For example, a relatively mild phenotype of the by^ex10 mutants can be explained by higher expression of by than the other by mutants.

Besides the blistered wing phenotype, the by mutants exhibited reduced hatching rate of laid eggs (Fig. 3D). The hatching rate of the mutant eggs was reduced to about 60%of wild type, while the survival rate of mutants after hatching was not significantly affected (Fig. 3D). These results demonstrate that tensin plays important roles during Drosophila early development as well. Our unpublished data suggest that the impairment in egg hatching is resulted from defective fertilization, therefore the by mutants are homozygous viable with decreased fertility.

Drosophila wing development after pupariation (AP) consists of two distinct stages: prepupal and pupal wing morphogenesis(Fristrom et al., 1993; Fristrom et al., 1994). And pupal wing morphogenesis is further divided into three stages: separation(11-12 hours AP) of the ventral cell layer from the dorsal layer,re-apposition of the inter-vein cells (21-36 hours AP) and re-separation (60 hours AP) of the two cell layers. Shortly after eclosion, wings expand and unfold by an influx of hemolymph. According to a previous report, PS integrins are required for the attachment of the two wing surfaces during pupal wing re-apposition and for the maintenance of the wing bilayer(Brabant et al., 1996).

To determine the detailed roles of tensin during wing morphogenesis, we examined the pupal wings of the by² flies. As shown in Fig. 4, we were not able to observe any differences in the attachment of two wing surfaces(Fig. 4A-H) and in the integrin localization (Fig. 4I-L)between wild type (Fig. 4, left panels) and by² (Fig. 4, right panels) wings during both prepupal apposition (4-6 hours AP, Fig. 4A-D) and pupal reapposition stages (30-36 hours AP, Fig. 4E-L).

Because the pupal wing development was not disturbed in the by² mutants, we decided to investigate the expansion and unfolding processes of adult wings in the by² mutants. After eclosion, the by² flies displayed folded wings(Fig. 5F) similar to the control (Fig. 5A). Then, a sudden and rapid influx of hemolymph induced the unfolding of folded wings in the by² mutants (Fig. 5G) in the same manner as the control(Fig. 5B). However, as soon as the wings of by² flies unfolded, fluid-filled blisters began to appear at the distal part of the wings(Fig. 5H), and the boundary of the blisters expanded to a certain extent(Fig. 5I). After the fluid dried, the wing blisters were fixed in place(Fig. 5J).

Taken together, although the dorsal and ventral layers of a wing can be brought into close association during apposition and re-apposition processes in the by² flies, the link between them may not be strong enough to resist the hydrostatic pressures during the wing unfolding process.

We examined the functional significance of each domain of tensin in normal wing development. We have generated UAS lines overexpressing either full-length tensin protein or various deletion mutant forms of tensin as shown in Fig. 6A. The ectopic expression of each transgene in the wing imaginal discs by MS1096-GAL4 driver was confirmed by RT-PCR (Fig. 6B). Unlike ΔN (Fig. 6D) and ΔC (Fig. 6E), overexpression of ΔPTB by MS1096-GAL4 driver completely rescued the blistered wing phenotype of by²(Fig. 6C). These data suggest that both the N-terminal region and the SH2 domain of tensin are required for proper attachment of two wing surfaces.

As mammalian tensin is known to participate in the integrin signaling(Zamir and Geiger, 2001), we have examined whether tensin genetically interacts with integrin. As expected,the blistered wing phenotype became more severe in the if³; by²/+ mutants(Fig. 7C) and extremely severe in the double homozygotic mutants for if³ and by² (Fig. 7D), compared with if³ homozygotes(Fig. 7A) or by² heterozygotes (Fig. 7B). In addition, the rate of flies showing blistered wings in the total population greatly increased in the double mutants(Table 1).

To confirm that this enhanced severity of the phenotype in the if³ by² double mutants was not due to an additive effect of two unrelated mutations, we tested the effect of heterozygous by² mutation on other wing blister-exhibiting mutants. Overexpression of rl^Sem using MS1096-GAL4 driver induced extra wing veins and sometimes resulted in a wing blister phenotype in about 21% of the total flies (Fig. 7E, Table 1). However, unlike the integrin mutants, the transgenic flies overexpressing rl^Sem within a heterozygotic by²genetic background did not significantly increase either the severity or the incidence of the blistered wing phenotype of flies overexpressing rl^Sem (Fig. 7F, Table 1). Collectively, these results suggest that tensin functionally interacts with integrin in the wing development of Drosophila in a specific manner.

In mammalian cells, tensin has been implicated in signal transduction related to cell adhesion such as Src, JNK and PI3K(Thomas et al., 1995; Auger et al., 1996; Katz et al., 2000). To examine the role of tensin in the signaling processes related to wing development, we investigated the in vivo interaction between tensin and signaling molecules including rl/Erk, Src, JNK and PI3K. Interestingly, we found that the JNK signaling pathway is tightly correlated with tensin in the wing development(Fig. 8), while other signaling molecules including rl/Erk did not show any interactions with tensin(Fig. 8I,J, and data not shown). Homozygous by² mutants with heterozygotic mutations of the JNK signaling components bsk¹(Fig. 8B, Table 1) or hep¹ (Fig. 8E,F, Table 1) (the loss-of-function mutants for Drosophila JNK and MKK7,respectively) displayed a highly severe blistered wing phenotype compared with either homozygous by²(Fig. 2A, Table 1), heterozygous bsk¹ (Fig. 8A, Table 1) or heterozygous hep¹ (Fig. 8D, Table 1)mutants. Notably, the rate of flies, which showed Class II blistered wings,increased from 46.5% to 70% for these double mutants compared with homozygous by² mutants, and about 15% of these flies had multiple blisters in their wings. Furthermore, the double homozygotic mutants for by² and hep¹ died at pharate adult stage (Fig. 8C). The lethality of these double mutants may be due to an impairment of essential in vivo interactions between tensin and the JNK signaling pathway in Drosophila.

Next, we tested whether overexpressed by also interacts with the components of the JNK signaling pathway. Overexpression of by using MS1096-GAL4 driver turned the adult wings into a convex shape with a smaller overall size (Fig. 8G), and this phenotype became more severe when two copies of the by gene were overexpressed (Fig. 8H). Simultaneous overexpression of bsk(Fig. 8L) or hep(Fig. 8N) with byresulted in a severely curled wing phenotype, which was fully penetrant,whereas overexpression of bsk(Fig. 8K) or hep(Fig. 8M) alone by MS1096-GAL4 driver did not induce any detectable phenotypes in the wing. Collectively,these data suggest that tensin activity is highly related to the JNK signaling pathway during wing development in Drosophila.

To further confirm the genetic interaction between tensin and the JNK pathway, we measured the effect of tensin in JNK activity in vivo. We examined the extent of JNK phosphorylation using anti-phosphospecific JNK antibody in the by overexpression line and the by² mutants. As expected, JNK phosphorylation was dramatically increased in the wing imaginal discs overexpressing by(Fig. 9C) compared with the control (Fig. 9A), which directly demonstrated increased JNK activity by by. On the contrary,JNK phosphorylation in the imaginal discs of the by²mutants was reduced (Fig. 9B)compared with the control (Fig. 9A).

Moreover, the reduced size and the convex wing phenotype observed in the wings overexpressing by (compare Fig. 9E with its control Fig. 9D) can be most easily explained by apoptosis in the wings. As the induction of apoptosis by the JNK signaling is well established (Goberdhan and Wilson, 1998; Adachi-Yamada et al., 1999), we expected that the wing phenotype induced by by overexpression might be due to apoptosis. To confirm the by-induced apoptosis in vivo, we carried out Acridine Orange staining of the relevant wing imaginal discs. As expected, the overexpression of by dramatically increased apoptotic cell death(Fig. 9G) compared with the control (Fig. 9F).

In this study, we have investigated the function of Drosophilatensin ortholog (by) by analyzing its loss-of-function and gain-of-function mutants. By observing the role of tensin in the development of the fly wing and by analyzing its genetic interactions with various signaling components, we were able to gain valuable insights into the in vivo roles of tensin.

The N-terminal region of tensin is responsible for the wing blister phenotype of the by mutants

Our current data demonstrate that by¹ allele contains two missense mutations in the N-terminal region of tensin, and that the expression of by is not affected in the by¹mutants compared with the wild-type control(Fig. 3C). Because the severity of the wing blister phenotype in by¹ was not significantly different from the by-deficient mutants such as by² and by^ex49(Fig. 3B), and the ectopic expression of ΔN form of tensin did not rescue the blistered wing phenotype of by² (Fig. 6), we think that the mutations in the N-terminal region of Drosophila tensin including tyrosine 62 and threonine 163 are responsible for the wing blister phenotype.

However, as Tyr62 and Thr163 are not conserved in other tensins, and there is no information for unmutagenized parental chromosomes of by¹, further studies are required to verify the exact nature of by¹ mutations, and to address the molecular mechanism underlying how mutations of by¹ allele affect the cellular functions of tensin.

During Drosophila wing development, the dorsal and ventral wing epithelia are fused together by highly specified cell-cell adhesions, and defects in this process result in blistered wings. Therefore, the blistered wing phenotype observed in the by mutants indicates that tensin functions in such a cell adhesion process.

In Drosophila, integrin is well known as a central molecule that mediates adhesion between wing layers. Previous studies have shown that the loss of PS integrin function in the wings causes the formation of a fluid-filled blister (Brower and Jaffe,1989; Wilcox et al.,1989; Brabant and Brower,1993; Brown et al.,2000). Moreover, the loss of adaptor proteins such as short stop (Prout et al., 1997)and integrin-linked kinase(Zervas et al., 2001) that mediate attachment of extracellular matrix (ECM)-integrin complexes to cytoskeleton also result in a wing blister phenotype, implying that integrin and its adaptor protein complexes are indispensable in wing layer adhesion. Because tensin is a major component of the ECM-integrin complex, and the wing blister phenotype of an integrin mutant was dramatically enhanced by additional loss of by in this study(Fig. 7 and Table 1) we believe that improperly mediated integrin signaling caused by the loss of tensin results in the wing blister phenotype in the by mutants. This idea is supported by previous mammalian cell studies showing that tensin acts as a molecular linker between actin cytoskeleton and integrin(Zamir and Geiger, 2001), and plays a crucial function when integrin translocates from focal adhesions to fibrillar adhesions (Pankov et al.,2000).

Besides the defects in wing cell adhesion process, we observed another distinct mutant phenotype in the by mutants; they laid rounded eggs due to defective oocyte elongation during oogenesis (K.S.C., unpublished). Our unpublished genetics data suggest the possibility of a functional interaction between tensin and integrin during oogenesis. Interestingly, a similar phenotype has been reported recently in the studies with follicle cell clones lacking PS integrin (Bateman et al.,2001).

Taken together, above findings suggest that tensin and integrins are tightly linked together during most, if not all, of their various functions in Drosophila development.

In this study, tensin was observed to genetically interact with the components of the JNK signaling pathway, and regulate JNK activity during wing development (Figs 8, 9). The supporting evidence for the engagement of tensin in the JNK signaling pathway comes from a recent report that transfected mammalian tensin activates JNK signaling in HEK 293T cells (Katz et al., 2000).

Interestingly, in mammalian cells, JNK is also activated via adaptor proteins p130 CAS and Crk which receive a signal from the FAK/Src tyrosine kinase complex in the cell adhesion sites when cells attach to the ECM(Oktay et al., 1999). As tensin is a possible substrate for FAK(Guan, 1997), and p130 CAS is able to interact with the C terminus of tensin(Salgia et al., 1995; Salgia et al., 1996), it is highly possible that tensin is involved in this signaling cascade and mediates signals from integrin and FAK to the JNK signaling pathway.

In addition, tensin was observed not to genetically interact with other signaling pathways known to interact with integrins such as the ERK-MAPK(Fig. 8) and the PI3K signaling pathways (data not shown). As previously mentioned, integrin signaling is mediated mainly by protein complexes including tensin in focal adhesions. Thus, we think that focal adhesion molecules related to integrin such as tensin are important for directing integrin mediated extracellular signals to a specific signaling pathway. Consequently, we tentatively suggest that at least during Drosophila wing development, tensin has an ability to drive signals from integrin selectively to the JNK signaling pathway. However,further studies are required to confirm this hypothesis and determine the details behind the connection between focal adhesion proteins and related intracellular signaling.

In summary, we have characterized by mutant flies and analyzed the developmental roles of by specifically in wings. We found evidence for the functional interaction between integrin and tensin, and for the modulation of JNK signaling by tensin during Drosophila wing development. These results strongly suggest that tensin is not merely an adaptor protein in focal adhesions, but also an important mediator of signal transduction in Drosophila. Our Drosophila model will be useful in future studies that address the function of tensin as a signaling molecule.

We thank the members of the Chung laboratory and anonymous reviewers for advice and helpful discussions. We also thank Drs. M. Freeman, D. L. Brower,M. Mlodzik, S. Noselli, D. Bohmann and T. Adachi-Yamada for their fly stocks and reagents.