Genetic Interaction Between Integrins and moleskin, a Gene Encoding a Drosophila Homolog of Importin-7

Corresponding author: Danny L. Brower, Life Sciences South Bldg., 1007 E. Lowell St., University of Arizona, Tucson, AZ 85721., dbrower{at}u.arizona.edu (E-mail)

Communicating editor: T. CLINE

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The Drosophila PS1 and PS2 integrins are required to maintain the connection between the dorsal and ventral wing epithelia. If PS subunits are inappropriately expressed during early pupariation, the epithelia separate, causing a wing blister. Two lines of evidence indicate that this apparent loss-of-function phenotype is not a dominant negative effect, but is due to inappropriate expression of functional integrins: wing blisters are not generated efficiently by misexpression of loss-of-function PS2 subunits with mutations that inhibit ligand binding, and gain-of-function, hyperactivated mutant PS2 proteins cause blistering at expression levels well below those required by wild-type proteins. A genetic screen for dominant suppressors of wing blisters generated null alleles of a gene named moleskin, which encodes the protein DIM-7. DIM-7, a Drosophila homolog of vertebrate importin-7, has recently been shown to bind the SHP-2 tyrosine phosphatase homolog Corkscrew and to be important in the nuclear translocation of activated D-ERK. Consistent with this latter finding, homozygous mutant clones of moleskin fail to grow in the wing. Genetic tests suggest that the moleskin suppression of wing blisters is not directly related to inhibition of D-ERK nuclear import. These data are discussed with respect to the possible regulation of integrin function by cytoplasmic ERK.

INTEGRINS are a strongly conserved family of cell surface receptors (HYNES 1992 ), and genes for - and ß-subunits of integrin heterodimers have been found in the most primitive metazoans. Most integrins bind components of the extracellular matrix (ECM), although some integrins in vertebrates recognize other cell surface proteins. Typically, integrins make strong connections between the ECM and the actin cytoskeleton (YAMADA and MIYAMOTO 1995 ; DEDHAR and HANNIGAN 1996 ). Integrins are also signaling proteins, and integrin ligand binding can have a multitude of effects in regulating cellular events (DEDHAR and HANNIGAN 1996 ; HOWE et al. 1998 ). Moreover, cells can often regulate the function of their integrins in what is referred to as "inside out" signaling (FERNANDEZ et al. 1998 ; HUGHES and PFAFF 1998 ).

Drosophila genetics has been instrumental in the identification and analysis of an extraordinary number of genes encoding proteins important for developmental and cell biological processes. The genetic study of integrin function in Drosophila has included a combination of classical forward and reverse genetics approaches (GOTWALS et al. 1994 ; BROWN et al. 2000 ). The gene encoding the ßPS subunit (myospheroid) was originally identified by mutation and analyzed extensively by genetics before it was discovered that it encoded an integrin subunit (WRIGHT 1960 ; NEWMAN and WRIGHT 1981 ; MACKRELL et al. 1988 ). The PS1, PS2, and PS3 proteins were all identified biochemically as integrins, and gene localization subsequently was used to identify the corresponding genes, mew (WEHRLI et al. 1993 ; BROWER et al. 1995 ), inflated (BOGAERT et al. 1987 ; WILCOX et al. 1989 ), and scab (STARK et al. 1997 ), respectively (see also GROTEWIEL et al. 1998 , who came upon the PS3 gene independently in a forward genetic screen). The PS4, PS5, and ß genes have yet to be extensively analyzed genetically.

As in vertebrates, studies of integrin function in Drosophila are moving toward analyses of components that work in conjunction with the ß-heterodimers. The elucidation of the fly genome makes it relatively straightforward to generate and study mutations in proteins previously associated with integrins from other systems. Also, forward genetic screens can identify novel cellular components involved in integrin function. The PS1 and PS2 integrins are required to maintain the connection between dorsal and ventral wing surfaces (reviewed by BROWN et al. 2000 ), and this phenotype has been used to devise relatively efficient strategies for identification of integrin-related genes that are likely to be recessive lethals when mutated. For example, the generation of homozygous clones of mutant cells has been used in screens to find integrin and related mutants that cause the wing surfaces to separate (BROWER et al. 1995 ; PROUT et al. 1997 ; WALSH and BROWN 1998 ).

Another tactic for identification of interacting components is to look for dominant enhancers or suppressors of weak integrin phenotypes (e.g., WILCOX 1990 ). Screens for suppressors of a phenotype, where a defect must be repaired by the mutation, typically are more reliable than enhancer screens in identifying components that are functionally associated with the initial mutation. Since most mutations are due to loss of gene function, suppressor screens often begin with a gain-of-function phenotype to be suppressed.

During the late larval and early pupal development of the Drosophila wing, the PS1 and PS2 integrins show a predominantly dorsal and ventral, respectively, restriction in their expression (BROWER et al. 1985 ). Soon after pupariation, the wing pouch evaginates and folds along the nascent wing margin, bringing together the basal surfaces of the dorsal and ventral cells. After 10–12 hr, the dorsal and ventral epithelia separate as the wing epithelium expands. Approximately 10–12 hr later, the two sides reappose and remain attached until the adult fly ecloses from the pupal case (WADDINGTON 1941 ; FRISTROM et al. 1993 ).

If integrin -subunits are inappropriately expressed in the developing wing, the epithelia do not reappose and wing blisters result (BRABANT et al. 1996 ; N. BROWN, personal communication). It is not clear which is more important for blister formation, expression of an -subunit on the wrong surface or the unusually high level of expression typically required (see also BROWN et al. 2000 ). What is known is that the critical time for this phenotype is during the initial apposition of the dorsal and ventral epithelia; high level or spatially incorrect integrin expression later is without effect (BRABANT et al. 1996 ). It appears that if some necessary event does not take place at the initial apposition, reapposition is prevented, even if integrin expression is returned to normal. Whether the defect results from interference with a specific intracellular signal or simply from the disruption of dorsoventral connections necessary for reapposition is unclear.

Early gene dosage experiments suggested that the wing blistering due to misappropriate integrin expression results from a gain-of-function effect (BRABANT et al. 1998 ), although this had not been demonstrated convincingly. If so, the phenotype can be a useful starting point to look for suppressor mutations that might function downstream of the integrin misexpression. Here we show that the wing blistering is indeed a gain-of-function phenotype and describe a screen for suppressors that has identified mutations in a gene involved in the nuclear import of the phosphorylated form of the mitogen-activated protein (MAP) kinase, D-ERK.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly husbandry and stocks:
For all crosses, animals were grown on the food described by CONDIE and BROWER 1989 . For crosses in which phenotypes were to be assayed, all flies from a vial were scored to guard against potential developmental rate variations.

Misexpression of PS2 in the wing was generated in most cases using the GAL4 UAS system (BRAND et al. 1994 ). The GAL4 enhancer traps 684 (wing pouch) and 337 (most tissues) are described by MANSEAU et al. 1997 ; these P elements are inserts into the third chromosome and are marked with w⁺. Blistermaker is a third chromosome (homozygous viable) containing both the 684 enhancer trap and a P element with UAS-PS2m8 (this particular insertion is designated m8K6), also marked with w⁺.

The tubulin1-PS2m8 flies were generated by first creating an inserted P element (P[tub1-promoter FRT y⁺ FRT PS2m8] on chromosome 3) with a cassette for making PS2-expressing clones using the FRT FLP system (STRUHL and BASLER 1993 ). The y⁺ sequence was recombined out by expression of FLPase in the germ line, and the tub-PS2-containing chromosome was balanced over TM3, Ser. Other stocks used in crosses to assay suppression of blistering or vein formation are: y csw^eOP w sev^d2 f car/FM7 (from Mike Simon); y Draf^C110 sn/Binsc and y w spl sn Dsor^r2/Binsc (from Yasuyoshi Nishida); y w; HS-rho^27B/TM3, Sb and y w; HS-rho^30A/TM3, Sb (from Ethan Bier); en-GAL4 enhancer trap (from Ruth Palmer); nw^D pu² Egfr^E1 Pin^Yt/SM1 (from the Bloomington Stock Center); Draf^HM7; rl^Sem/CyO; and rl¹. The UAS-msk chromosome is described in LORENZEN et al. 2001 .

Mutant PS2 experiments:
The PS2-LOF (222-224 YWQ>AWA) and PS2-GOF (deletion of the cytoplasmic CGFFN) mutations were made by PCR mutagenesis, confirmed by sequencing, and inserted into pUASPS2m8 or pUASPS2C for fly transformation (BRABANT et al. 1996 ) or into pHSPS2m8 or pHSPS2C for transformation into S2 cells (BUNCH and BROWER 1992 ; ZAVORTINK et al. 1993 ). Numerous independent chromosomal insertions of the mutant PS2 subunits were generated by embryo injection. All fly transformants were of the "m8" isoform of PS2 (BROWN et al. 1989 ). LOF inserts were designated H and K on chromosome 2 and B, L, and P on chromosome 3. GOF inserts that showed some blistering activity were G on chromosome 2 and A, C, and O on chromosome 3. The wild-type cDNA insert was m8Z4 on the X chromosome. S2 cell transformants were made as described (BUNCH and BROWER 1992 ). Both PS2m8 and PS2C isoforms were generated; the data shown in Fig 1 derive from the "c" isoform.

View larger version (16K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Cell spreading mediated by PS2 mutants. S2 cells transformed with the indicated PS2 and wild-type ßPS subunit genes were allowed to spread on various concentrations of a fragment of the ECM protein Tiggrin. Only the loss-of-function mutant does not support spreading in this assay. Flow cytometry of these cells indicates that the wild-type and LOF mutants are expressed on the cell surface at similar levels, while the GOF mutant is expressed at 5- to 10-fold lower levels.

Spreading of transformed S2 cells was performed as previously described (JANNUZI et al. 2002 ). Briefly, cells were cleared with protease, heat shocked, allowed to spread for 3–4 hr on a recombinant fragment of the matrix ligand Tiggrin, and scored by direct observation.

To score wing blistering and examine expression in wing discs, animals bearing inserts of mutant or wild-type UAS-PS2 genes were crossed to animals bearing the 337 enhancer trap. For most experiments, including those for which discs were stained, homozygous stocks of all chromosomes were crossed, so that the animals to be examined were heterozygous for both the enhancer trap and the UAS-PS2 transgene. For experiments to score interactions with myospheroid mutants, we made stocks with either a LOF (insert B) or a GOF (insert C) UAS-PS2 recombined onto a chromosome with the 337 enhancer trap, balanced over TM3, Sb. These stocks were then crossed to mys/FM7c females (myospheroid alleles M2, G4, and G1; see JANNUZI et al. 2002 ) to score blistering and viability. The 337 UAS-PS2 mutant animals displayed reduced viability; to minimize this, eggs were generally laid and progeny were allowed to develop for 2–3 days (through embryogenesis) at 22°, and then raised to the temperature at which blistering was to be scored. Immunostaining of integrins in wing imaginal discs was performed as described (BROWER et al. 1984 ) using the monoclonal antibodies DK.1A4 (PS1) and CF.2C7 (PS2).

Suppressors of Blistermaker:
Oregon-R (wild-type) males were mutagenized with EMS (GRIGLIATTI 1986 ) and crossed to w; Blistermaker/Blistermaker virgin females at 28°. Progeny with two wild-type wings were kept; retesting and mapping were initiated by crossing these flies to w; Blistermaker Sb/TM3, Ser. Although subsequent mapping and balancing should have theoretically isolated suppressors on any of the three large chromosomes, the screen turned out to be strongly biased toward the third chromosome. This was likely because identification of other suppressors relied on suppression of Blistermaker/TM3 animals during mapping, and TM3 turned out to be a general enhancer of blistering. Chromosomes with suppressing activity all contained recessive lethal mutations, and these were tested against one another for complementation. Only the moleskin complementation group contained multiple alleles, and the moleskin recessive lethals were mapped by recombination to a region of chromosome 3L. The locus was fine mapped using a set of deficiencies; moleskin alleles failed to complement Df(3L)pbl-X1 and Df(3L)66C-G28 as well as a smaller deficiency that was generated in an attempt to hop a nearby P element into the locus. The locus was finally identified molecularly by sequencing candidate genes within the deficiencies from the moleskin mutant chromosomes.

DNA sequencing:
Genomic DNA from moleskin alleles balanced over TM3 was prepared using a QIAGEN (Valencia, CA) QIAamp tissue kit. Using the Drosophila genome sequence, PCR primers were designed to amplify the potential coding exons. The products of amplification were prepared using QIAGEN's QiaQuick PCR purification kit and sequenced directly by the University of Arizona LMSE Automated DNA Sequencing Service. All mutants were confirmed by sequencing of both strands.

Generation of moleskin clones:
Males of the genotype y w; en-GAL4 47m1UAS-DIM-7/+; mwh msk⁵ P[w⁺]70C P[FRT]80B/+ were crossed to y w hsFLP; P[y⁺ FRT]80B/TM3, Sb at 22°, and the progeny were given 60-min heat shocks (37°) at various times during larval development to induce recombination at the FRT sites. Wings were mounted in Euparol and clone sizes were scored using the multiple wing hairs marker.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Blistermaker results from a gain of function:
The wing blister phenotype caused by inappropriate integrin subunit expression resembles the loss-of-function phenotype, as evidenced by viable integrin mutants or wing clones homozygous for null integrin alleles. This might suggest that the Blistermaker chromosome (containing an PS2 gene driven by the wing pouch enhancer trap, 684; BRABANT et al. 1996 ; MANSEAU et al. 1997 ) causes a dominant negative situation. For example, since both PS1 and PS2 integrins share a common ßPS subunit, flooding dorsal wing cells with PS2 subunits might reduce dorsal PS1ßPS dimer expression below a critical level. Alternatively, the extra PS2 integrins might directly lead to wing blistering through some gain-of-function event, such as the activation of an unknown regulatory pathway or an inappropriate adhesion. Gene dosage studies tended to support the gain-of-function proposal (BRABANT et al. 1998 ; D. L. BROWER, unpublished data). For example, reducing ßPS expression via heterozygosity for null mutations in the ßPS-encoding myospheroid gene does not have the enhancing effects that would be predicted by the dominant negative scenarios, and increasing ßPS expression with transgenes does not suppress the effects of PS2 overexpression.

We sought a more direct demonstration that Blistermaker does indeed result from a gain of function. We transformed flies with one of two mutant PS2 genes, under the control of the GAL4 UAS (BRAND et al. 1994 ). In one gene (PS2 loss of function or PS2-LOF), residues 222–224 (YWQ) of the extracellular domain are changed to AWA. This alteration is expected to inhibit extracellular ligand binding (e.g., IRIE et al. 1995 ), and indeed we find that in Drosophila S2 cells transfected with PS2-LOF, cell spreading is severely inhibited relative to wild type (Fig 1). The other mutant is a deletion of the cytoplasmic, membrane proximal CGFFN sequence (residues 1366–1370 for PS2C), which is expected to lead to activation of integrin heterodimers (e.g., O'TOOLE et al. 1994 ). This expectation is supported by observations of S2 cells transfected with this mutant gene (PS2-GOF). These cells spread very efficiently on PS2 ligands, even at very low levels of integrin expression (Fig 1).

To drive mutant integrin expression in developing flies, we used a GAL4 enhancer trap (337; MANSEAU et al. 1997 ) that is expressed fairly ubiquitously, so that relative expression levels could be discerned easily in the dorsal proximal wing disc (outside of the wing pouch), where PS2 expression is normally close to zero. Numerous chromosomal inserts of each gene were generated and tested at various temperatures (wing blistering due to Blistermaker is greater at higher temperatures). It should be noted that both sets of mutant animals often show significant lethality, especially at 28°, and for the experiments at higher temperatures the flies are allowed to transit embryogenesis at 22° to increase viability.

Examination of PS2 expression in wing imaginal discs from larvae bearing the PS2-LOF mutant shows that it can be expressed at high levels (Fig 2B), similar to those seen when wild-type PS2 is driven from the same enhancer trap. However, even with this high level of surface expression, adults have a relatively low frequency of wing blisters. For example, the disc illustrated in Fig 2B is typical of expression levels from three different transgenes, but in the 25° crosses done for disc staining, virtually no blisters were observed in adults expressing these transgenes. Overall, a number of PS2-LOF inserts cause significant (>50%) blistering at 28°, but even for the strongest inserts, the frequency of defects falls abruptly at 25°, typically to <5%. Using the same enhancer trap, a wild-type PS2 insert typically blisters at close to 100% levels at 25°.

View larger version (100K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Expression of mutant integrins on the cell surface in late third instar (just prior to pupariation) wing imaginal discs. (A) Wild-type PS2 is primarily restricted to ventral cells. (B) Using the 337 enhancer trap, a UAS-PS2-LOF (deficient for ligand binding) transformant (line "L") is expressed throughout the wing at high levels. Adult wings from these animals rarely blister. (C) A UAS-PS2-GOF (activated) transformant (line "G") is expressed dorsally at low levels. Adult wings from this cross blister at >50% frequency. (D) A UAS-PS2 wild-type transformant (line m8Z4). Dorsal PS2 is expressed at a level below that of the loss-of-function mutant in B, but expression is much greater than that for the activated mutant in C. (To assess ectopic dorsal expression relative to ventral expression, compare the dorso-ventral boundaries at the posterior margins, indicated by the brackets.) About 2% of the wings from this cross are blistered, more than for the loss-of-function cross in B but less than for the gain-of-function cross in C. (E and F) High magnification of dorsal folds from PS2-GOF discs, stained with antibody against PS2 (E) or PS1 (F). The activated PS2-containing integrins are clustered in basal plaques on each cell, but the PS1-containing heterodimers are not.

The activated PS2-GOF integrins are expressed at low levels in imaginal discs (Fig 2C), just as they are in S2 cells in culture (not shown). This is consistent with findings from integrins in situ bearing similar mutations (MARTIN-BERMUDO et al. 1998 ). The PS2-GOF-containing integrins are typically clustered in a plaque on the basal surface of each imaginal disc cell (Fig 2E); this is similar to the plaques of wild-type integrins observed during pupal stages by BRABANT et al. 1996 . The dorsal PS1 integrins in the same cells are not clustered by the activated PS2 dimers (Fig 2F). The mutant subunits are quite capable of generating wing blisters, even though expressed at levels that, for wild-type PS2, would never make blisters. For example, wings from adults grown (at 25°) in the same vials as the larval disc shown in Fig 2C had a blister frequency of 65%. By contrast, Fig 2D shows a disc expressing wild-type PS2m8 driven by the same enhancer trap (at 18°); although ectopic PS2 expression is much greater than that for the activated mutants, adult flies from this cross have a blister frequency of 2%. (At 25°, the same wild-type-expressing cross shows dorsal PS2 expression similar to that seen for the PS2-LOF in Fig 2B and a blister frequency of close to 100%.) Overall, four of eight PS2-GOF inserts cause blisters as heterozygotes, although others begin to do so when homozygous. The PS2-GOF inserts display a temperature sensitivity similar to that of wild-type Blistermaker, with the penetrance of blistering at 25° typically being at least 50% or more of the 28° frequency.

In summary, nonfunctional PS2 subunits blister wings poorly, and activated subunits blister wings more efficiently than do wild type; these experiments demonstrate that the Blistermaker phenotype results from a gain of integrin function.

Finally, some additional observations indicate not only that the PS2-LOF subunits are less effective at making wing blisters than are the wild-type or activated proteins, but also that the loss-of-function mutants may have their effects through a different mechanism as well. The inability of reduced ßPS expression to enhance the penetrance of the wild-type Blistermaker was one of the reasons for originally thinking that this was a gain-of-function phenotype (BRABANT et al. 1998 ). By contrast, heterozygosity for mutations in myospheroid (ßPS) strongly enhances the wing blistering of PS2-LOF-expressing animals (Fig 3). Additional evidence that the PS2-LOF behaves as a dominant negative comes from its synthetic lethality with myospheroid (ßPS) null mutations. That is, myospheroid mutants are typically recessive, and heterozygotes are completely viable and wild type. However, if myospheroid heterozygotes also express PS2-LOF subunits, they are killed by high temperatures (28°). Even if embryogenesis (which is the most sensitive stage) is allowed to proceed at low temperature (22°), a postembryonic shift to 28° reduced adult viability to 2% in one experiment (which included data from three different myospheroid alleles).

View larger version (27K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Wing blistering of PS2-LOF but not PS2-GOF mutants is enhanced by reducing the dosage of ßPS (myospheroid). Males with UAS-PS2 transgenes and the 337-GAL4 enhancer trap were crossed at 22° to females heterozygous for a strong myospheroid mutation over the FM7 balancer chromosome, and progeny went through late larval and pupal development at 25° or 28°. Heterozygosity for myospheroid increased wing blisters induced by the loss-of-function PS2; for the gain-of-function mutant, blistering was reduced to zero in myospheroid heterozygotes. The asterisks above the LOF 28° bar indicate that this value is based on only two wings, due to the greatly reduced viability of myospheroid heterozygotes that express PS2-LOF; however, examinations of pharate adults trapped in pupal cases is consistent with the notion that this genotype results in completely nonapposed wings. For this experiment, PS2-LOF line "B" and PS2-GOF line "C" were used.

Second-site suppressors of Blistermaker:
Since the Blistermaker phenotype results from an inappropriate integrin-related process, one might hope to suppress the phenotype by reducing the activity of related functions. We had found previously that the Blistermaker phenotype could be suppressed by heterozygosity for various mutant chromosomes (e.g., BRABANT et al. 1998 ). Unfortunately, further analyses indicated that the phenotype was very sensitive to genetic background; that is, chromosomes from different wild-type strains could also have large effects on the penetrance of wing blistering. When testing for interactions by crossing Blistermaker to a stock with a particular mutant chromosome, the control and experimental classes of progeny will be genetically different at almost 20% (for the X chromosome) to 40% (for chromosomes 2 and 3) of the genome. We found that for most mutant chromosomes tested by crossing to different stocks, it was not possible to map all of the suppressing activity to a single mutant locus on the chromosome.

To circumvent the genetic background problem as well as to find potential unanticipated suppressing loci, we performed a genetic screen for suppressors of Blistermaker. The screen, illustrated in Fig 4, asks for mutations that will suppress Blistermaker in a dominant manner; the mutated chromosomes may or may not be recessive lethals. Because the beginning strains are isogenic, the genetic background is uniform and any changes in activity should result from mutations created by the EMS. Complementation between the different suppressing chromosomes was examined, using recessive lethality as an assay, and we found that one complementation group on the third chromosome was represented five times (although two of these alleles subsequently proved to be duplicates). Because this locus was identified as a suppressor of blistering, we named the corresponding gene moleskin (msk).

View larger version (34K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Screen for suppressors of Blistermaker.

Before proceeding further, it was important to show that moleskin-mediated suppression is due to an effect independent of integrin expression. For example, we had found in earlier work (before the screen was undertaken) that some of our Blistermaker-suppressing chromosomes acted by reducing expression through the 684-GAL4 enhancer trap. We examined integrin expression directly in moleskin/Blistermaker animals and detected no obvious differences relative to Blistermaker heterozygotes alone (not shown). We also looked in detail at ß-galactosidase expression in wing discs from animals bearing the 684 enhancer trap and a UAS-lacZ insert, with and without moleskin mutations (Fig 5). Finally, we asked if moleskin heterozygosity could suppress blistering in flies in which integrin expression is driven by a completely different set of regulators. These test animals contain an PS2 transgene driven directly by a tubulin promoter, with no enhancer trap or GAL4 intermediate. As shown in Fig 6, wing blistering is suppressed by moleskin regardless of the mode of expression. Chromosomes known to suppress the Blistermaker chromosome via reduction of expression from the enhancer trap are completely ineffective in suppressing the tubulin-PS2 animals.

View larger version (53K): [in this window] [in a new window] [Download PPT slide]   Figure 5. ß-Galactosidase expression in the wing pouch of late third instar imaginal discs, resulting from a UAS-lacZ transgene and the 684-GAL4 enhancer trap. (A) Wild type. (B) Heterozygote for the DrafC110 chromosome, which suppresses Blistermaker apparently by reducing expression from the 684-GAL4 system. (C) Heterozygote for msk5; there is no clear reduction in ß-galactosidase expression.

View larger version (25K): [in this window] [in a new window] [Download PPT slide]   Figure 6. moleskin suppression is independent of the GAL4-UAS expression system. Wing blistering suppression is similar for the msk4 and csweOP heterozygotes whether integrin expression is driven by the 684 enhancer trap and a UAS-PS2 gene (Blistermaker) or directly from a transgene with a tubulin promoter. The Dsorr2 and DrafC110 chromosomes suppress Blistermaker completely (0% wings blistered), but have little effect on the tubulin-regulated blisters. For all crosses, blistering in the controls with the relevant balancer chromosomes (TM3 for moleskin, FM7 for corkscrew, and Binsc for Dsor and Draf) is 83–100%.

moleskin encodes Drosophila importin-7:
To determine the molecular nature of the moleskin gene, the suppressing activity was genetically mapped to a region on the left arm of the third chromosome. Using the recessive lethality and a series of deficiencies, the gene was further localized to a small region within polytene bands 66B8–10. Molecular determination of deficiency breakpoints and comparison to the Drosophila genome defined a set of potential open reading frames, and one was selected for further analysis on the basis of the finding that its encoded protein, Drosophila importin-7 (DIM-7), was found to bind to the cytoplasmic tyrosine phosphatase Corkscrew in a two-hybrid screen (LORENZEN et al. 2001 ). The "Corkscrew connection" seemed potentially relevant since we had earlier found that a chromosome containing an antimorphic allele of corkscrew (csw^eOP) was a dominant suppressor of Blistermaker, although this finding has not been verified to be independent of other genetic background effects. Since the identification of moleskin as DIM-7, we have found that the csw^eOP mutant chromosome also suppresses blistering from the tubulin-PS2 insert (Fig 6). Dominant Blistermaker suppression by corkscrew appears to be dependent on the dominant negative properties of the csw^eOP mutation (ALLARD et al. 1996 ), since two chromosomes containing other strong alleles of corkscrew fail to suppress at similar levels (not shown). Sequencing of the moleskin chromosomes revealed that four of the five alleles contain mutations expected to truncate the encoded DIM-7 protein (Table 1); all of these alleles terminate translation in the first one-third of the predicted coding sequence of 1049 amino acids. These alleles have a similar lethal recessive phenotype (death of late embryos or early larvae, with a normal-looking cuticle) and, considering that the msk⁵ allele contains a stop in the second codon, we believe that this is likely to be a protein null allele. The animals most likely survive embryogenesis as a result of a significant maternal contribution of wild-type activity (LORENZEN et al. 2001 ), which we have not been able to eliminate genetically (see below). We found no lesions in the coding region of the msk¹ chromosome. This allele has a weaker lethal phenotype (most homozygotes die as pupae), and it therefore probably results from a regulatory mutation. The DIM-7 protein is a member of the ß-importin family of nuclear import proteins (LORENZEN et al. 2001 ). It is a close homolog of human importin-7 (also known as Ran binding protein 7, or RanBP7), being 53% identical in sequence, with the homology extending throughout the sequences. Recently, we showed that DIM-7 was important for the nuclear import of the activated MAP kinase D-ERK in response to signaling from receptor tyrosine kinases (LORENZEN et al. 2001 ).

moleskin function is required for growth:
To examine the phenotype of cells lacking wild-type DIM-7 in the wing, we generated clones of cells homozygous for the msk⁵ allele via somatic recombination in heterozygous animals (XU and RUBIN 1993 ). Induction of clones (identified by the cell marker multiple wing hairs) at various times during development fails to yield moleskin mutant clones >4–8 cells. Developmentally early induction often yields no clones at all, suggesting that the small clones depend on perdurance of wild-type gene product from the heterozygous clone precursor cell.

Since the entire left arm of the mutant chromosome 3 is made homozygous by this procedure, we wanted to make certain that the lack of clone growth was due to the moleskin mutation, and not to some other lesion. To do this, we made msk⁵ clones in wings in which wild-type DIM-7 was expressed in the posterior compartment, under the control of an engrailed-GAL4 enhancer trap. In these wings, posterior msk⁵ homozygous clones grow to typical sizes of >100 cells, while anterior clones of >8 cells are not observed (Fig 7). Thus, DIM-7 function is required for growth of cells in the wing epithelium. This requirement for cell growth is likely to be fairly general, as we also failed to generate moleskin mutant clones in the female germ line, in an attempt to produce embryos missing the strong maternal component of DIM-7.

View larger version (102K): [in this window] [in a new window] [Download PPT slide]   Figure 7. Wing with homozygous clones of msk5 cells (outlines), induced during larval development. The posterior (lower) wing expresses wild-type DIM-7 protein, under the control of an engrailed enhancer trap. The wild-type DIM-7 rescues the growth inhibition phenotype of the posterior clones, which can grow to hundreds of cells. Anterior clones are typically eight or fewer cells.

In the course of the above experiments, we noticed that abnormal expression of DIM-7 affects the patterning of the adult wings. With many GAL4 enhancer traps to drive UAS-DIM-7 we found that the animals are killed, but the engrailed-GAL4 trap yields a variety of wing abnormalities, in part depending on temperature (the GAL4 system often expresses at greater levels at higher temperatures). Comparing a series of wings of varying severity, we find that the two cross-veins move closer with increasing DIM-7, until they line up into one large vein (Fig 8). Also, in a small number of wings, blisters are formed.

View larger version (91K): [in this window] [in a new window] [Download PPT slide]   Figure 8. Wild-type wing (A) and wings expressing DIM-7 under the control of an engrailed-GAL4 enhancer trap (B and C), which is specific for the posterior wing. High levels of DIM-7 can cause the cross-veins to move closer (as in Fig 7) or to actually fuse into one long cross-vein (B). In some cases, blisters can result (C).

Blistermaker suppression and D-ERK nuclear import:
Integrins have been shown in numerous systems to interact with growth factor receptors in regulating ERK activity (GIANCOTTI and RUOSLAHTI 1999 ; ASSOIAN and SCHWARTZ 2001 ), and so it seemed possible that Blistermaker suppression by moleskin mutants could indicate an integrin activation of the nuclear import of activated D-ERK. While D-EGFR (Drosophila epidermal growth factor receptor) and D-ERK are important throughout the wing epithelium for normal growth, increased function of each is also necessary in a complex series of steps to induce and support the differentiation of veins in the wing (STURTEVANT and BIER 1995 ; GUICHARD et al. 1999 ; MARTIN-BLANCO et al. 1999 ). We examined the effects of heterozygosity for moleskin mutations in various mutants that affect vein formation. In each case we assayed the effects of the moleskin-containing chromosome relative to the control TM3 balancer chromosome. These are the same moleskin/TM3 stocks that have been used for Blistermaker crosses, where moleskin does suppress relative to TM3. Thus, if Blistermaker suppression is via D-ERK import, we might expect to see similar relative effects on vein formation.

The rhomboid gene encodes a protein that is believed to be involved in processing an extracellular activator of D-EGFR (KLAMBT 2000 ), and rhomboid activity is specifically upregulated in the vein-forming regions of the wing (STURTEVANT and BIER 1995 ; GUICHARD et al. 1999 ). We used two lines of flies that contain an inserted rhomboid transgene under the control of a heat-shock promoter (HS-rhomboid). Both inserts tested express low levels of rhomboid throughout the wing, resulting in ectopic veins (HS-rhomboid^27B) or at higher expression levels, wing blisters (HS-rhomboid^30A), without the need for inducing heat shocks. As shown in Fig 9, the msk⁴ chromosome does not suppress wing blisters generated by HS-rhomboid^30A nor does it suppress the extra veins of HS-rhomboid^27B. Indeed, the phenotypes are typically stronger in the moleskin-containing animals than in those bearing the TM3 balancer chromosome. There also is no obvious suppression by the antimorphic csw^eOP chromosome or by two chromosomes that suppress Blistermaker by repressing the 684 enhancer trap (not shown).

View larger version (64K): [in this window] [in a new window] [Download PPT slide]   Figure 9. Wings from cross of msk4/TM3 to flies with HS-rhomboid inserts that are expressed constitutively at relatively low (HS-rhomboid27B; A and B) or high (HS-rhomboid30A; C and D) levels. The inappropriate expression of rhomboid induces extra vein material (B and C) or, at high levels, wing blisters (D). In both cases, the phenotype is more severe in the animals with the msk4 chromosome than in those with the TM3 balancer; this is in contrast to the relative effects of these chromosomes on the Blistermaker phenotype.

Wing vein formation requires a complex scenario of D-ERK regulation in space and time (MARTIN-BLANCO et al. 1999 ), and so we examined the ability of moleskin chromosomes to suppress other activating mutants. Ellipse (Egfr^E1) is a gain-of-function mutation of the gene encoding D-EGFR (BAKER and RUBIN 1989 ), and Sevenmaker (rl^Sem) is a dominant gain-of-function mutation in the gene (rolled) encoding D-ERK (BRUNNER et al. 1994 ). Both alleles tested here cause ectopic vein formation. As for the HS-rhomboid phenotypes above, neither the Ellipse nor the Sevenmaker wing phenotype is suppressed by the msk⁴ or msk⁵ chromosomes (not shown).

We also examined the effects of our Blistermaker suppressor stocks on a viable hypomorphic mutation in the D-ERK-encoding gene, rolled¹ (rl¹). Reduced D-ERK function in homozygous rl¹ flies often results in gaps in wing vein 4. The msk⁴ and msk⁵ chromosomes showed no clear enhancement of the rl¹ wing gaps (Fig 10); in fact, relative to the TM3 balancer, these chromosomes enhanced the phenotype.

View larger version (87K): [in this window] [in a new window] [Download PPT slide]   Figure 10. Wing from rolled1; msk4/+ animal, showing breaks in vein 4. This phenotype is not significantly enhanced by msk relative to the TM3 control chromosome (see also Table 2).

In summary, we have crossed moleskin mutant stocks to a variety of mutations that alter the D-EGFR D-ERK signaling pathway and gene expression. When comparing the effects of the same moleskin and control TM3 chromosomes in each case, we see no positive correlation between the ability of a moleskin mutant chromosome to suppress integrin-induced wing blistering and its ability to suppress (in the case of gain-of-function mutations) or enhance (for loss-of-function mutations) phenotypes dependent on D-ERK-regulated gene expression (Table 2).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Blistermaker is a gain of function:
The Blistermaker phenotype is similar to the wing blistering that results from a loss of integrin function, either in integrin weak alleles or in clones of strong mutants (BROWN et al. 2000 ). However, the results reported here support the general idea of an active role for integrins in blistering. Especially striking is the fact that, although the ligand-binding-deficient subunits behave as enhancers of other loss-of-function phenotypes [for example, expression of PS2-LOF can make myospheroid (ßPS) heterozygotes lethal], they are poor inducers of wing blistering on their own. Moreover, the wing blistering that PS2-LOF does induce is enhanced by myospheroid heterozygosity, suggesting that it results from a different mechanism from that of Blistermaker. The complementary finding that hyperactivated PS2 subunits can create wing blisters when expressed at relatively low levels, which have no significant effect on endogenous integrin expression, demonstrates convincingly that Blistermaker is indeed a gain-of-function phenotype.

Blistermaker and D-ERK-dependent gene expression:
The Blistermaker phenotype is suppressed when expression of the ß-importin DIM-7 is reduced in heterozygous moleskin mutant flies. The ß-importins comprise a large family, responsible for the nuclear import of a wide variety of proteins (GORLICH and KUTAY 1999 ). The vertebrate homolog of DIM-7, importin-7, has been shown to translocate ribosomal proteins and histones (JAKEL and GORLICH 1998 ; JAKEL et al. 1999 ). Most interesting, though, is the demonstrated function of DIM-7 in the nuclear import of the activated MAP kinase D-ERK (LORENZEN et al. 2001 ), especially in light of the extensive literature connecting integrins and ERK signaling (GIANCOTTI and RUOSLAHTI 1999 ; ASSOIAN and SCHWARTZ 2001 ). Although the direct functional connection between DIM-7 and D-ERK has been demonstrated primarily in embryos, the inability of homozygous moleskin mutant clones to grow in the wing epithelium is consistent with a requirement downstream of growth factor receptor signaling in this tissue as well (e.g., DIAZ-BENJUMEA and GARCIA-BELLIDO 1990 ; DIAZ-BENJUMEA and HAFEN 1994 ).

However, it seems unlikely that moleskin mutants dominantly suppress Blistermaker by reducing expression of genes that depend on the nuclear translocation of activated D-ERK. We say this because wing vein formation, which depends on a series of specific growth factor-initiated D-ERK signals (STURTEVANT and BIER 1995 ; GUICHARD et al. 1999 ; MARTIN-BLANCO et al. 1999 ), and Blistermaker display different relative sensitivities to the moleskin and TM3 chromosomes.

At first glance, it might appear paradoxical that moleskin chromosomes do not seem to suppress (and even enhance) events known to require nuclear import of activated D-ERK or, conversely, that moleskin chromosomes do not seem to enhance the effects of rolled (D-ERK) loss-of-function alleles. Indeed, we previously reported data that appear to contradict the current findings (LORENZEN et al. 2001 ). However, it must be remembered that in each cross we are comparing a chromosome with a moleskin allele to another chromosome, in this case a third chromosome balancer, TM3. Thus, 40% of the genomes are different between the experimental and control classes, and comparisons with other crosses (in which the control is a first chromosome balancer) suggest that our TM3, Sb chromosome is something of a suppressor of the activated D-ERK phenotypes. So, in each case here we are assaying the effects of the moleskin-containing chromosome relative to a chromosome that also may be a suppressor of vein formation; we are not asking if moleskin is a suppressor or enhancer in an absolute sense. What is important is that these same stocks are those that have been used for Blistermaker crosses, in which moleskin does suppress relative to TM3. Thus, if Blistermaker suppression is via D-ERK import, we would expect to see similar relative effects on vein formation, and we do not.

Although DIM-7 immunoprecipitates activated D-ERK (LORENZEN et al. 2001 ), we have no direct evidence that moleskin mutants mediate Blistermaker suppression through D-ERK. There almost certainly are other cargos for DIM-7, and we have no data concerning the sensitivities of these potential nuclear transport or other events to moleskin dosage. However, for simplicity, we will focus the following discussion of mechanistic possibilities on ERK, since we know that it can both associate with DIM-7 and regulate integrin function. Formally, the arguments presented could be applicable to other, currently unknown regulators as well.

Does moleskin suppress Blistermaker by increasing cytoplasmic D-ERK?
H-Ras can be a suppressor of integrin activation, and data suggest that it may act via activated ERK in a transcription-independent manner (HUGHES et al. 1997 ). These findings lead to possible models for Blistermaker suppression by moleskin heterozygosity, via the negative regulatory activity on integrins of activated cytoplasmic D-ERK. For example, reducing DIM-7 by 50% in a moleskin heterozygote would be expected to reduce nuclear import of activated D-ERK, although since DIM-7 is not necessarily the major rate-limiting step in D-ERK signaling, the reduction in activated D-ERK in the nucleus would likely be <50%. But because the amount of activated D-ERK that transits to the nucleus is typically much greater than that which travels to the cell periphery, a modest reduction in nuclear import can result in a relatively large increase in cytoplasmic D-ERK activity. Thus, compared to their effects on gene expression, moleskin heterozygotes might have a greater capacity to affect integrin activation directly, suppressing the effects of Blistermaker.

Two observations might seem to argue against the notion that DIM-7 and D-ERK regulate integrin function in the wing, but in fact they do not. The first is that moleskin chromosomes appear to be able to suppress blistering from the PS2-GOF transgenes (our unpublished data), which should not easily be regulated by cytoplasmic events. This suppression is less dramatic than that for Blistermaker, and its interpretation is subject to some of the genetic background difficulties described earlier. Most importantly, however, it must be remembered that blistering may depend on some amount of active PS1-containing integrins in the same dorsal cells, and these will still be susceptible to D-ERK-mediated regulation.

A second potential difficulty is that this model predicts that activated D-ERK generally should be a Blistermaker suppressor. In apparent contradiction, the Sevenmaker mutation of the D-ERK-encoding gene rolled, which leads to elevated levels of phosphorylated D-ERK, is an enhancer of Blistermaker, not a suppressor. However, the Sevenmaker mutation is in a docking domain of D-ERK that alters the ability of the protein to interact with downstream effectors as well as with D-ERK-regulating kinases and phosphatases (TANOUE et al. 2000 ). Thus, although Sevenmaker enhances some phenotypes that require D-ERK-activated gene expression, it is difficult to predict its effects on a particular event a priori.

A function for cortical DIM-7?
The ß-importin family of proteins is principally linked with nuclear import of protein cargos. However, recently other functions have been associated with members of the importin superfamily. For example, importin-ß, in some cases with importin-, functions in vertebrates to sequester microtubule polymerization factors early in mitosis (GRUSS et al. 2001 ; NACHURY et al. 2001 ; WIESE et al. 2001 ). Mitotic microtubule formation can be triggered by the release of the polymerization regulators by RanGTP, just as RanGTP binding to importin-ß leads to release of cargos inside the nucleus.

DIM-7 protein can be detected immunologically at the cell cortex, both in early Drosophila embryos (LORENZEN et al. 2001 ) and in S2 cells in culture (our unpublished results). It thus seems reasonable to consider a more direct connection between the peripheral DIM-7 and integrin regulation. Additionally, it appears that a mutation in corkscrew, the Drosophila SHP-2 homolog (PERKINS et al. 1992 , PERKINS et al. 1996 ), can also suppress Blistermaker and that Corkscrew protein binds directly to DIM-7 (LORENZEN et al. 2001 ). Although Corkscrew has been implicated primarily in signaling events downstream of receptor tyrosine kinases (PERKINS et al. 1992 , PERKINS et al. 1996 ; ALLARD et al. 1996 ; HERBST et al. 1996 ; CLEGHON et al. 1998 ; JOHNSON HAMLET and PERKINS 2001 ), vertebrate SHP-2 has been implicated in signaling via a host of growth factor receptors, cytokines, hormones, and antigens (reviewed by FENG 1999 ). Most relevant to our discussion, SHP-2, often in association with the membrane glycoproteins PECAM-1 or SHPS-1, has been shown to be involved in many integrin-dependent signaling events and also to be important in regulating integrin-mediated cell adhesion, spreading, or migration (JACKSON et al. 1997 ; SAGAWA et al. 1997 ; TSUDA et al. 1998 ; YU et al. 1998 ; DEMALI et al. 1999 ; MANES et al. 1999 ; OH et al. 1999 ; INAGAKI et al. 2000 ; SCHOENWAELDER et al. 2000 ; LACALLE et al. 2002 ). While SHP-2 is a cytoplasmic tyrosine phosphatase, some experiments suggest that it can serve as a scaffolding protein at or near the plasma membrane. For example, a Corkscrew protein mutated in the phosphatase domain retains significant wild-type activity in situ, and this activity is increased if the protein is targeted to the plasma membrane (ALLARD et al. 1998 ).

It is likely therefore that cell surface receptors mediate a localized Corkscrew/SHP-2 activation of cortical DIM-7. This active DIM-7, in combination with associated factors such as D-ERK, could then function more directly in integrin regulation. A more direct connection between DIM-7 and integrin function is also consistent with the fact that moleskin mutations were especially common among the suppressors isolated in the screen. A key question for future work, therefore, will be defining the subcellular location at which DIM-7 functions with respect to integrin-related phenotypes.

Integrin regulation of nuclear import:
Recently, evidence has begun to appear that integrin engagement with the ECM can regulate nuclear import of regulatory molecules. For example, BIANCHI et al. 2000 found an association between Lß2 and the c-Jun coactivator JAB1 and suggest that this connection regulates the nuclear localization of JAB1. More directly relevant to our results, APLIN et al. 2001 reported that ERK nuclear translocation in fibroblasts is dependent on an integrin-mediated event, also involving the actin cytoskeleton. Also, HIRSCH et al. 2002 found that primary mouse embryo fibroblasts with a ß1 integrin cytoplasmic mutant show reduced nuclear translocation of phosphorylated ERK. Regardless of the importance of nuclear transport in Blistermaker suppression, our genetic data indicate a functional connection between integrins and a specific importin-ß that can transport activated ERK and suggest another potential molecular mechanism whereby integrin and growth factor signals can be integrated by the cell.

	FOOTNOTES

¹ Present address: Department of Plant Health, Torrey Mesa Research Institute, 3115 Merryfield Row, San Diego, CA 92121.
² Present address: Division of Biology 0349, UC San Diego, 9500 Gilman Dr., La Jolla, CA 92093.

	ACKNOWLEDGMENTS

We thank Dominic Jezierski, Augustine Lau, and Jeff Rosenberger for help with the screens and other genetics of suppressors; Barb Carolus for help with FACS analyses; and Brian Coullahan and the group in the LMSE sequencing facility. S.E.B. especially wishes to thank Marc Brabant, Tracy Futch, Brian James, and Leona Mukai for instruction and patience. Many people have been generous with fly stocks and reagents over the course of these experiments, including Konrad Basler, Ethan Bier, Suzanne Eaton, Mike Forte, Yasuyoshi Nishida, Ruth Palmer, Norbert Perrimon, Mike Simon, Gary Struhl, John Thomas, and the Bloomington Stock Center. Financial support includes National Institutes of Health (NIH) grant no. 5F32GM19149 (S.E.B.); NIH grant no. 5F32GM17901 and the American Cancer Society, Massachusetts (J.A.L.); University of Arizona UBRP (S.W.M.); NIH grant no. HL48728 (M.H.G.); National Science Foundation IBN9723509 and NIH grant no. R01GM61707 (L.A.P); and NIH grant no. R01GM42474 (D.L.B.).

Manuscript received February 7, 2002; Accepted for publication June 25, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALLARD, J. D., H. C. CHANG, R. HERBST, H. MCNEILL, and M. A. SIMON, 1996  The SH2-containing tyrosine phosphatase Corkscrew is required during signaling by Sevenless, Ras1 and Raf. Development 122:1137-1146.[Abstract]

ALLARD, J. D., R. HERBST, P. M. CARROLL, and M. A. SIMON, 1998  Mutational analysis of the SRC homology 2 domain protein-tyrosine phosphatase Corkscrew. J. Biol. Chem. 273:13129-13135.[Abstract/Free Full Text]

APLIN, A. E., S. A. STEWART, R. K. ASSOIAN, and R. L. JULIANO, 2001  Integrin-mediated adhesion regulates ERK nuclear translocation and phosphorylation of Elk-1. J. Cell Biol. 153:273-282.[Abstract/Free Full Text]

ASSOIAN, R. K. and M. A. SCHWARTZ, 2001  Coordinate signaling by integrins and receptor tyrosine kinases in the regulation of G1 phase cell-cycle progression. Curr. Opin. Genet. Dev. 11:48-53.[Medline]

BAKER, N. E. and G. M. RUBIN, 1989  Effect on eye development of dominant mutations in Drosophila homologue of the EGF receptor. Nature 340:150-153.[Medline]

BIANCHI, E., S. DENTI, A. GRANATA, G. BOSSI, and J. GEGINAT et al., 2000  Integrin LFA-1 interacts with the transcriptional co-activator JAB1 to modulate AP-1 activity. Nature 404:617-621.[Medline]

BOGAERT, T., N. BROWN, and M. WILCOX, 1987  The Drosophila PS2 antigen is an invertebrate integrin that, like the fibronectin receptor, becomes localized to muscle attachments. Cell 51:929-940.[Medline]

BRABANT, M., C. D. FRISTROM, T. A. BUNCH, and D. L. BROWER, 1996  Distinct spatial and temporal functions for PS integrins during Drosophila wing morphogenesis. Development 122:3307-3317.[Abstract]

BRABANT, M. C., D. FRISTROM, T. A. BUNCH, S. E. BAKER, and D. L. BROWER, 1998  The PS integrins are required for a regulatory event during Drosophila wing morphogenesis. Ann. NY Acad. Sci. 857:99-109.[Abstract/Free Full Text]

BRAND, A., A. MANOUKIAN and N. PERRIMON, 1994 Ectopic expression in Drosophila, pp. 635–654 in Drosophila melanogaster: Practical Uses in Cell and Molecular Biology, Methods in Cell Biology, Vol. 44, edited by L. GOLDSTEIN and E. FYRBERG. Academic Press, San Diego.

BROWER, D. L., M. WILCOX, M. PIOVANT, R. J. SMITH, and L. A. REGER, 1984  Related cell-surface antigens expressed with positional specificity in Drosophila imaginal discs. Proc. Natl. Acad. Sci. USA 81:7485-7489.[Abstract/Free Full Text]

BROWER, D. L., M. PIOVANT, and L. A. REGER, 1985  Developmental analysis of Drosophila position-specific antigens. Dev. Biol. 108:120-130.[Medline]

BROWER, D. L., T. A. BUNCH, L. MUKAI, T. E. ADAMSON, and M. WEHRLI et al., 1995  Nonequivalent requirements for PS1 and PS2 integrin at cell attachments in Drosophila: genetic analysis of the _PS1 integrin subunit. Development 121:1311-1320.[Abstract]

BROWN, N. H., D. L. KING, M. WILCOX, and F. C. KAFATOS, 1989  Developmentally regulated alternative splicing of Drosophila integrin PS2 transcripts. Cell 59:185-195.[Medline]

BROWN, N. H., S. T. GREGORY, and M. D. MARTIN-BERMUDO, 2000  Integrins as mediators of morphogenesis in Drosophila.. Dev. Biol. 223:1-16.[Medline]

BRUNNER, D., N. OELLERS, J. SZABAD, W. H. BIGGS, III, and S. L. ZIPURSKY et al., 1994  A gain-of-function mutation in Drosophila MAP kinase activates multiple receptor tyrosine kinase signaling pathways. Cell 76:875-888.[Medline]

BUNCH, T. A. and D. L. BROWER, 1992  Drosophila PS2 integrin mediates RGD dependent cell-matrix interactions. Development 116:239-247.[Abstract]

CLEGHON, V., P. FELDMANN, C. GHIGLIONE, T. D. COPELAND, and N. PERRIMON et al., 1998  Opposing actions of CSW and RasGAP modulate the strength of Torso RTK signaling in the Drosophila terminal pathway. Mol. Cell 2:719-727.[Medline]

CONDIE, J. M. and D. L. BROWER, 1989  Allelic interactions at the engrailed locus of Drosophila: engrailed protein expression in imaginal discs. Dev. Biol. 135:31-42.[Medline]

DEDHAR, S. and G. E. HANNIGAN, 1996  Integrin cytoplasmic interactions and bidirectional transmembrane signalling. Curr. Opin. Cell Biol. 8:657-669.[Medline]

DEMALI, K. A., E. BALCIUNAITE, and A. KAZLAUSKAS, 1999  Integrins enhance platelet-derived growth factor (PDGF)-dependent responses by altering the signal relay enzymes that are recruited to the PDGF ß receptor. J. Biol. Chem. 274:19551-19558.[Abstract/Free Full Text]

DIAZ-BENJUMEA, F. J. and A. GARCIA-BELLIDO, 1990  Behaviour of cells mutant for an EGF receptor homologue of Drosophila in genetic mosaics. Proc. R. Soc. Lond. Ser. B. Biol. Sci. 242:36-44.[Medline]

DIAZ-BENJUMEA, F. J. and E. HAFEN, 1994  The sevenless signalling cassette mediates Drosophila EGF receptor function during epidermal development. Development 120:569-578.[Abstract]

FENG, G. S., 1999  Shp-2 tyrosine phosphatase: signaling one cell or many. Exp. Cell Res. 253:47-54.[Medline]

FERNANDEZ, C., K. CLARK, L. BURROWS, N. R. SCHOFIELD, and M. J. HUMPHRIES, 1998  Regulation of the extracellular ligand binding activity of integrins. Front Biosci. 3:D684-D700.

FRISTROM, D., M. WILCOX, and J. FRISTROM, 1993  The distribution of PS i