ABSTRACT
Using the Drosophila cell line MLDmBG-1, a monoclonal antibody aBG-1 that can inhibit not only cell clumping but also cell spreading was generated. This antibody immunoprecipitates a complex of molecules consisting of a major 120×103Mr and other components. To characterize the 120×103Mr component, we purified it, generated antibodies to it, and cloned its cDNA. Sequencing of this cDNA suggests that the 120×103Mr molecule is identical to PSβ, a β chain of Drosophila integrins. The other components immunoprecipitated included two a chains of Drosophila integrins, PSlα and PS2α, as revealed using specific antibodies to these molecules. These suggest that aBG-1 recognizes the PSβ associated with PSlα or PS2α. However, immunostaining of embryos and larvae with aBG-1 showed that the staining pattern is similar to that for PS2α but not for PSβ, suggesting that the antibody preferentially recognizes the PSβ associated with particular a chains in situ. We then attempted to characterize the ligands for these integrin complexes, using culture dishes coated with various vertebrate matrix proteins. These cells spread very well on dishes coated with vitronectin and, to a lesser extent, on those with fibronectin. This spreading was partially inhibited by aBG-1, but not by other control antibodies or RGD peptides. The cell attachment to these substrata was not affected by the antibody. The cells also can attach to dishes coated with laminin but without spreading, and this attachment was not inhibited by aBG-1. Furthermore, they do not attach to dishes coated with collagen type I, type IV, and fibrinogen. These results indicate that Drosophila PS integrins can recognize vertebrate vitronectin, and also fibronectin with a weaker affinity, at sites other than RGD sequences, and thus can function in cell-substratum adhesion.
Introduction
Cell adhesion is one of the primary processes through which proper morphogenesis occurs in multicellular organisms (Edelman and Thiery, 1985). Many classes of cell adhesion receptors have been identified and their roles in morphogenesis are being investigated. For such studies, Drosophila is assumed to be a useful model system, because genetic approaches can be applied. In fact, Drosophila homologues of various vertebrate cell adhesion receptors have been identified; these include members of the immunoglobulin superfamily (Harrelson and Goodman, 1988 ; Seeger et al. 1988 ; Bieber et al. 1989) and integrins (Leptin et al. 1987; Bogaert et al. 1987; MacKrell et al. 1988). Also, some extracellular matrix proteins have been detected from this animal, such as laminin and collagen type IV (see review Fessler and Fessler, 1989). Indirect evidence also suggests the presence of fibronectin in this species (Gratecos et al. 1988). However, not all the cell adhesion molecules identified in vertebrates have so far been found in Drosophila. For example, calcium-dependent cell adhesion molecules, cadherins, have not been identified, although some cell surface proteins with cadherin-like sequences have been detected (T. Mahoney and C. Goodman, personal communication).
One of the best characterized cell adhesion receptors in Drosophila is an integrin. In vertebrates, this molecular family includes the major receptors for various extracellular matrix components. Integrins are heterodimers consisting of noncovalently associated α and β subunits, and the combinations of these subunits generate diverse specificities for extracellular matrix proteins. For example, α5β integrin is a receptor for fibronectin while α6β1 integrin is for laminin (see review Hynes, 1987; Ruoslahti, 1988; Akiyama et al. 1990). In Drosophila, two classes of α asubunits, termed PS1α and one class of β subunit, PSβ, are known. These molecules were originally identified by their staining pattern in the wing imaginai disc (Wilcox et al. 1981; Brower et al. 1984), and later they turned out to be members of the integrin family (Leptin et al. 1987; Bogaert et al. 1987; MacKrell et al. 1988). Their functions in morphogenesis have been analyzed utilizing genetic methods (Wilcox et al. 1989; Leptin et al. 1989; Brower and Jaffe, 1989; Zusman et al. 1990), and evidence has been accumulated for their important morphogenetic roles. However, their ligands still remain to be identified.
In searching for novel adhesion molecules, immunological approaches have been successful. In this strategy, antibodies that inhibit cell adhesion are isolated and the antigens are identified as candidates for adhesion molecules. In Drosophila, such approaches have been attempted in a few systems (Gratecos et al. 1990), but they have not been successful. This failure is partly due to the fact that many of the established Drosophila cell lines lack adhesive properties; without appropriate cell lines, it is difficult to analyze cell adhesion mechanisms in accurate ways. Recently, however, several Drosophila cell lines that express both cell-cell and cell-substratum adhesiveness have been isolated, and one of them is MLDmBG-1 (K. Ui and T. Miyake, in preparation).
In the present study, we took advantage of the adhesive properties of the above cell line to identify adhesion molecules. Cells of the MLDmBG-1 line form aggregates in suspension culture, and also attach to and spread on culture substrata if these are coated with appropriate extracellular matrices. We generated a monoclonal antibody (mAb) that inhibits the clumping of MLDmBG-1 cells, and found that it recognizes a molecular complex of PS integrins and inhibits the spreading of these cells on vitronectin-coated dishes. These results demonstrate for the first time that the Drosophila integrins can function as receptors for cell-substratum adhesion recognizing vitronectin.
Materials and methods
Cells and cultures
Drosophila cell line MLDmBG-1 (K. Ui and T. Miyake, in preparation) was used for all the experiments described here. Cells were cultured with M3(BF) medium (Cross and Sang, 1978), which was slightly modified (Ui et al. 1987) and supplemented with 10% heat-inactivated fetal calf serum (FCS) and 10 μgml-1 insulin (M3(BF)-FS), in Nunclon cell culture bottles. Cells were incubated at 25°C under air.
Antibodies
Rat mAbs aBG-1 and 9C7, and a rabbit antiserum against the 120×103 Mr protein were generated as described below. Rat IgG used for control experiments was purchased from Chemicon International. mAbs PS2hc/l (Bogaert et al. 1987) and CF2C7 (Brower et al. 1984) to PS2α chain of integrin, and a rabbit antiserum to PSlα chain (Leptin et al. 1989) were kindly provided by Drs Maria Leptin and Michael Wilcox.
Rat mAbs were purified by using a Sepharose 4B column conjugated with anti-rat IgG (Zymet). Antibody solutions were loaded onto the column, and the antibodies bound were eluted and precipitated with 50% ammonium sulfate. The precipitates were dissolved in C & G’s saline (55 HIM NaCl, 40mM KC1, 15 mM MgSO4, 5mM CaCl2, 10 HIM Tricine, 20 mM glucose and 50mM sucrose, pH6.9) and dialyzed against the same saline. The purity of mAbs was confirmed by electrophoresis, followed by staining of the gels with Coomassie Brilliant Blue.
Assay of cell clumping
Cells were washed with calcium- and magnesium-free (CMF-) C & G’s saline and incubated with 0.1% trypsin (Difco) in CMF-C & G’s saline at room temperature for 5 min. After adding M3(BF)-FS medium, cells were collected by pipetting, and then centrifuged. The pellets were suspended in M3(BF)-FS medium, and then seeded into wells of a non-adherent 96-well plate (Flow Laboratory, 76-232-05). After two days, cells formed small aggregates loosely attaching to the dish. Culture medium was then partly removed leaving 60-70 pl per well, and subsequently the cell aggregates were detached from the dish and suspended into medium by gentle pipetting. The plate was then placed on a gyratory shaker (Model G-2, New Brunswick Scientific) rotating at 100 revs min-1 and incubated at 25°C. During this incubation, cell aggregates adhered to each other to form larger aggregates. The degree of clumping of aggregates was examined, assuming that it reflects cell aggregating activity.
Cell spreading assay
Human vitronectin (Takara), bovine fibronectin (Sigma), bovine collagen type I (Cell Matrix), bovine collagen type IV (Cell Matrix), mouse laminin (Iwaki Glass), and bovine fibrinogen (Organon Tekinka) were used as substrata for cell attachment and spreading. Each protein was dissolved in C & G’s saline in a concentration of 50μgml-1, except that collagens were solubilized with 0.02 M acetic acid. To coat 96-well dishes (Falcon, 3072) with these proteins, each well was incubated with an appropriate amount of the protein solution overnight at 4°C. Then, dishes were washed with C & G’s saline, and incubated with 1 % BSA at room temperature for one hour. After washing three times with C & G’s saline, followed by washing twice with M3(BF) medium without serum, cells were seeded as described below.
The cell spreading assay was performed by the method of Akiyama et al. (1986) with modifications to suit insect cells. Cells grown under semiconfluent conditions were incubated with 0.1 % trypsin (Sigma, T-8003) in C & G’s saline at room temperature. After adding M3(BF)-FS medium, cells were harvested by pipetting. Cells were then washed once with M3(BF) medium without serum and suspended in the same medium at a density of 5×104 ml-1. To this cell suspension, various affinity purified antibodies to be tested were added and the cell suspensions were incubated for 10min at 4°C. Then, these cell suspensions were placed into the wells coated with matrix proteins and then incubated for 3-5 h at 25°C. Cells were then fixed with 10 % glutaraldehyde for 5 min and washed. After taking photographs, the ratio of spread cells to total cells was measured. At least 200 cells were counted for each well, and the mean of the results obtained from duplicate or triplicate experiments was presented.
GRGDSP and GRGESP peptides (Iwaki Glass) were added to the cultures at 200μgml-1. A control experiment showed that this concentration of GRGDSP, but not GRGESP, significantly inhibited the spreading of NIH3T3 cells.
Generation of mAbs
mAb aBG-1 was generated as follows. Donryu rats were immunized by several i.p. injections of living MLDmBG-1 cells. Three days after the last boost, cell fusion was performed according to the method of Kôhler and Milstein (1975). Hybridomas were then screened for the production of antibodies capable of inhibiting cell aggregation.
To generate mAb which could recognize the 120×103 Mr molecule, Donryu rats were immunized by i.p. injection of this protein which was partially immunoaffinity-purified (see below). Antibodies were screened for their ability to recognize the 120×103Mr band on immunoblots. Sera collected from the immunized rats were used as polyclonal antisera against the 120×103 Mr protein.
Partial purification of the 120×103Mr protein
MLDmBG-1 cells were solubilized with 1 % Nonidet P-40 (NP40) in 50mM Tris-buffered saline pH7.6 (TBS) containing phenylmethylsulfonylfiuoride, pepstatin, antipain, leupeptin, and p-toluenesulfonyl-L-arginine methyl ester hydrochloride. After centrifugation, the supernatant was collected, and applied to a precolumn of Sepharose CL-4B (Pharmacia) to remove materials that bind nonspecifically to Sepharose. The flow through was then loaded onto an immunoaffinity Sepharose 4B column conjugated with aBG-1 mAb. After thorough washing with TBS, the antigens were eluted with 100mM glycine-HCl (pH2.5) and then with 50mM triethylamine (pH 11.5). The eluted samples were immediately neutralized with 1M Tris-buffered saline (pH8.0). The eluted fractions containing the 120×103Mr protein was subjected to immunization of rats for generation of mAbs. All the above procedures were done at 4°C.
Immunoprecipitation
For radiolabeling cellular components, cells were incubated in a methionine-free M3(BF) medium supplemented with 1 % heat-inactivated fetal calf serum and 100μCiml-1 of 35S-labeled E. coli hydrolysate (Tran 35S-label, ICN Biochemicals) overnight. The labeled cells were lysed with 1 % NP40 in TBS (one ml 25 cm-2 of dish). After centrifugation, 50^1 of 8% BSA and 60μl 5 M NaCl were added to 1ml of the supernatant. This was preabsorbed by incubation with 300μl of Sepharose 4B beads conjugated with anti-rat IgG. After centrifugation, a half volume of the culture supernatant of aBG-1 hybridoma was added, and incubated for one hour. To this solution, 300 μl of rat IgG-conjugated Sepharose 4B beads were added and incubated for one hour. The immune complexes were washed several times with a washing buffer (0.5% deoxycholate, 0.1% SDS, 1% Triton X-100, 50mM Tris and 150 mM NaCl, pH7.5), and then extracted with 500 μl of SDS electrophoresis sample buffer. All the above procedures were done at 4°C, except for 35S labeling. Prior to electrophoresis, the samples were reduced by boiling in the presence of 5 % β-mercaptoethanol for 3 min unless otherwise specified.
Immunohistochemistry
Cells in culture were fixed with 3.5% paraformaldehyde in 10 mM Hepes-buffered balanced saline (pH 7.5) for 30 min at 4°C. After rinsing with 50mM TBS, the fixed cells were extracted with methanol at —20°C for 10 min, and rinsed again with TBS. Cells were then treated with TBS containing 2.5% skim milk (blocking solution) for 30 min, and subsequently incubated with mAbs for 60 min at room temperature. After washing with TBS, the samples were incubated with a biotinylated second antibody (Amersham) in blocking solution for 60 min, followed by washing again with TBS and finally incubating with fluoresceine-labeled streptavidin (Amersham) in blocking solution for 45 min. After washing, the samples were mounted and examined. The whole mount staining of Drosophila embryos and imaginai discs basically followed the method of Bodmer and Jan (1987).
Immunoblot analysis
Samples were separated by SDS-PAGE using 7.5 % polyacrylamide gels. After electrophoresis, proteins were transferred to nitrocellulose sheets. The sheets were incubated with the blocking solution, mAbs, biotinylated second antibodies and finally alkaline phosphatase-conjugated streptavidine (Amersham). The signals were visualized with nitro blue tetrazohum (NBT) and bromochloroindolyl phosphate (BCIP) in 100 mM Tris-HCl (pH 9.5) containing 100 mM NaCl and 5 HIM MgCl2.
cDNA cloning
A Âgtll cDNA library from 6-12 h Drosophila embryos was kindly provided by Dr Kai Zinn. The library was screened with the antiserum raised against the partially purified aBG-1 antigens as described by Huynh et al. (1985). Approximately 1.5×105 recombinants were screened, and 28 clones were found to react with the antiserum. Then, epitope selection was carried out to detect clones encoding the 120×103Mr protein. Filters of the cloned Âgtll phages were incubated with the antiserum and washed with TBS. After rinsing once with distilled water, antibodies bound to each of the phages were eluted with 100mM glycine-HCl (pH2.5), and the solutions were neutralized with 1 M Tris. Antibodies bound to one clone recognized the 120X103Mr protein in immunoblots of the lysate of MLDmBG-1 cells, and this was subjected to sequencing.
Results
Generation of mAb aBG-1
The MLDmBG-1 cell line derives from the central nervous system of larvae and is a mixture of cells with heterogeneous morphology, including epithelial and fibroblastic types. These cells firmly adhere to culture dishes in the presence of fetal calf serum (FCS) and many of them spread on the substratum (Fig. 1A). However, if these cells are cultured on non-adherent culture dishes, they form aggregates that are loosely attached to the substratum (Fig. 1B). When these aggregates are detached from the plate by pipetting and incubated under gyration, they clump with each other and form larger aggregates within one hour (Fig. 1C).
The above system was utilized to generate monoclonal antibodies (mAbs) that inhibit cell aggregation. From lymphocytes of a rat immunized with intact MLDmBG-1 cells, we obtained one hybridoma clone producing antibodies that can inhibit the aggregate clumping (Fig. 1D) and this mAb was designated as aBG-1. When aBG-1 was added to the monolayer cultures of MLDmBG-1 cells, it inhibited their spreading (Fig. 2). Time-lapse recording of microscopic images of the cells showed that cell movement is arrested within minutes after the addition of aBG-1 and then cells become rounded (data not shown). This effect of the antibody was concentration-dependent,
Characterization of aBG-1 antigens
Although aBG-1 did not react with any bands in immunoblotting, it can immunoprecipitate several components from a detergent extract of 35S-labeled cells. They consist of a major 120×103Mr band and other 145×103Mr, 130×103Mr and 97 ×103Mr bands (Fig. 4A). Since aBG-1 affects cell-substratum adhesion, we tested whether these antigens are related to integrin, by using antibodies against two classes of integrin a chain, PSlα and PS2α. These antibodies stained bands on blots of materials immunoprecipitated with aBG-1. A rabbit antiserum to PS1α recognized a major 130×103Mr and other minor bands (Fig. 4B, lane 1). mAb PS2hc/l to PS2 α recognized a 145×103Mr doublet and other smaller bands (Fig. 4B, lane 2), of which the doublet is assumed to be the native form of PS2α and other bands are probably degradation products of the PS2α, as inferred from previous observations (Bogaert et al. 1987). The 130 ×103 Mr and 145×103Mr bands were detected at the corresponding positions on the autoradiograms of the electrophoresis of 35S-labeled materials immunoprecipitated with aBG-1 (compare Fig. 4A and 4B). These results indicate that aBG-1 immunoprecipitates PSlα and PS2α chains of integrin as components of its antigen complex. However, the major 120×103Mr component was not recognized by these antibodies.
To characterize the 120×103Mr band, we attempted to generate mAbs which specifically recognize this molecule. The antigens to aBG-1 were purified from a lysate of MLDmBG-1 cells using an affinity column conjugated with this mAb. With these materials, rats were immunized, and mAbs were screened for the activity of reacting with the 120×103Mr band on immunoblots. One mAb specifically recognized the 120×103Mr band on the blots of immunoprecipitates with aBG-1 (Fig. 4B, lane 3), and this was designated as 9C7. When this antibody was used for immunoprecipitation, it precipitated antigens whose electrophoretic pattern is similar to that of aBG-1 (Fig. 4C, lane 2). This antibody did not inhibit either cell adhesion or spreading when added to cell cultures (Fig. 3).
For further characterization of the 120×103Mr molecule, we attempted to clone its cDNAs. Using a polyclonal antiserum against the affinity purified aBG-1 antigens, a Agtll cDNA library constructed from embryonic mRNAs was screened and a cDNA clone encoding the 120×103Mr protein was isolated. Partial sequencing of this cDNA revealed that its sequence is identical to that of the myospheroid cDNA from position 1264 to 1548 (MacKrell et al. 1988) which encodes for the PS integrin β chain (PSβ). We also found that the apparent molecular mass of the 120×103Mr component changed if the electrophoretic sample was not reduced; it migrated to a position of 97×103Mr (data not shown). This is consistent with the known properties of the ft chains of integrin (Wilcox et al. 1984; Knudsen et al. 1985).
Taken together, it is most likely that aBG-1 recognizes a complex of PSμ associated with PSIα or PS2α chain of integrin. The 97×103MT band detected in immunoprecipitates with aBG-1 (Fig. 4A) remains to be identified, but this could be a degradation product of PSμ.
Tissue distribution of aBG-1 and 9C7 antigens
The expression pattern of aBG-1 antigens in MLDmBG-1 cells, imaginai discs and whole embryos was examined by immunohistochemical staining. In MLDmBG-1 cells, the antigens were distributed on their entire surface, occasionally at the boundaries between cells, but the intensity of staining varied (Fig. 5A). In imaginai discs of 3rd instar larvae, the staining pattern was similar to that for PS2α, reported previously (Brower et al. 1985; Bogaert et al. 1987). For example, the signals were stronger on the ventral half than the dorsal half in the wing discs (Fig. 5B). In the eye discs, the regions behind the morphogenetic furrow were stained (data not shown).
Tissue localization of aBG-1 antigens in whole embryos was also similar to that of PS2α. aBG-1 intensely reacted with the áttachment sites of somatic and visceral muscles as antibodies to PS2α do (Fig. 5C). In addition, aBG-1 stained the apical surface of the salivary gland (Fig. 5C, insert) where PS2α is not detected (Fig. 5F, insert). We examined whether aBG-1 antigens are expressed in embryos of ifk27e, which is known to lack PS2α (Wilcox et al. 1989), and found that this animal showed a negative reaction in aBG-1 staining at muscle attachment sites but positive at the salivary gland (Fig. 5E, insert).
The staining of the wing imaginai discs with mAb 9C7 gave a pattern different from that with aBG-1. The entire wing imaginai disc was stained (Fig. 5D), although its staining pattern at the muscle attachment sites was similar to that of PS2α (Fig. 5F). These staining patterns are similar to those for PSβ (Brower et al. 1984; Leptin et al. 1989), and consistent with the finding that 9C7 recognizes PSβ.
Identification of ligands for PS integrin in vitro
We examined to which extracellular matrix proteins of vertebrate origins MLDmBG-1 cells can attach. Almost all seeded cells adhered to dishes coated with vitronectin, fibronectin and laminin. The degree of their spreading, however, varied with the coated substances. Vitronectin most effectively supported the spreading of these cells; about 40–60% of the cells spread on vitronectin-coated dishes (Fig. 6A) and this spreading was inhibited by pretreatment of the coated dishes with anti-vitronectin (data not shown). On fibronectin, however, 40% or less of the attached cells spread (Fig. 6C). On laminin, no spreading occurred (Fig. 6E). These cells did not attach to dishes coated with fibrinogen, or collagen types I or IV.
In the presence of aBG-1, cell spreading on vitronectin and fibronectin was partially inhibited (Fig. 6B, D), and this effect was clearer on the vitronectin than fibronectin because the original cell spreading occurred more extensively on the former. The antibody did not affect the attachment of the cells to the substrata (data not shown). 9C7 had no effect on the spreading of cells on these substrata. We also examined the effect of GRGDSP and GRGESP peptides, but they had no effect on either cell adhesion nor spreading, although the former peptide could induce rounding up of vertebrate fibroblasts, such as NIH3T3, when added to their monolayer cultures (data not shown). These results are quantitatively summarized in Fig. 7.
Discussion
Monoclonal antibodies that can inhibit cell adhesion are useful tools for identifying cell adhesion molecules. Many classes of cell adhesion molecules have been identified in this way in various species. The MLDmBG-1 cells used in the present study have made such approaches possible in the Drosophila system, since they express adhesiveness not only to other cells but also to substrata.
We obtained one mAb, aBG-1, that inhibits the spreading of MLDmBG-1 cells and clumping of their aggregates. Identification of the targets for this antibody showed that they are a complex of PS integrins. The present results, thus, provide the first evidence that the Drosophila integrins can function as receptors for cell-substratum adhesion as their vertebrate homologues do. The inhibitory effect of aBG-1 on cell aggregate clumping implies that PS integrins may also effect cell-cell interactions. Indeed, aBG-1 antigens were occasionally detected at the contact sites between cells. The PS integrins localization at cell-cell contacts was originally suggested in tendon cell-muscle and ventral-dorsal wing epithelium attachment sites (Leptin et al. 1989; Wilcox et al. 1989) and also reported in primary cultures of Drosophila embryonic cells (Volk et al. 1990). Furthermore, integrins have recently been implicated in cell-cell adhesion in some vertebrate cells (Carter et al. 1990; Laijava et al. 1990). However, it cannot be excluded that the cell aggregate clumping was mediated by the interaction of aBG-1 antigens with extracellular matrix proteins secreted by the cells. It also must be pointed out that aBG-1 did not inhibit other types of cell aggregation, such as the aggregate formation of cells that spontaneously occurs on nonadherent culture dishes. The question whether PS integrins are involved in pure cell-cell adhesion, therefore, remains to be answered.
The conclusion that aBG-1 recognizes PS integrins was drawn from the observation that this antibody immunoprecipitates a complex of PS integrins. A question then arises which molecule in this complex has the epitope for the antibody. Because of the two reasons, PSβ chain is most likely the target for aBG-1. First, mAb 9C7 directed to PSfβ gave the same immunoprecipitation pattern as aBG-1. Second, aBG-1 immunoprecipitated two different a-chains simultaneously. This can occur only when PSβ chain is the target to the antibody; this subunit can form a heterodimer with either PSlα or PS2α subunits.
Curiously, however, the immunostaining pattern in embryos and larvae of aBG-1 was more like that of PS2α than that of PSβ. For example, aBG-1 stains the ventral half of the wing imaginai disc as the antibodies to PS2α do (Brower et al. 1984). In contrast, 9C7 stains the entire wing imaginai disc as the other antibodies to PSβ do (Brower et al. 1984). These suggest that aBG-1 fails to recognize the PSβ associated with PSIα-under immunocytochemical conditions.
The complicated nature of the aBG-1 epitope may be related to the ability of this antibody to inhibit cell adhesion. Integrins function as cell adhesion receptors only when their α and β chains form heterodimers. Therefore, the active sites for cell adhesion on these molecules should be generated in the form of a molecular complex of the two subunits. Possibly, aBG-1 recognizes a structure on the PSβ that is available only when associated with PS2α. Because aBG-1 can stain the salivary gland in a PS2α-deficient mutant ifk27e, this hypothetical epitope may be formed also on PSβ associated with other classes of a chain. It is, however, equally possible that the association with PSlα might mask the aBG-1 epitope on PSβ in situ.
The present study indicates that vertebrate vitronectin can function as a ligand for Drosophila PS integrins, as aBG-1 inhibited the cell spreading on this protein. Vertebrate fibronectin was also capable of promoting cell spreading but to a lesser extent; it probably has a weaker affinity for PS integrin than vitronectin. The spreading of MLDmBG-1 cells occurs only on these matrix proteins as far as has been tested. Volk et al. (1990) showed that myoblasts derived from Drosophila embryos attach to dishes coated with Drosophila laminin and spread well on these substrata, while MLDmBG-1 cells did not spread on vertebrate laminincoated dishes. The attachment to Drosophila laminin, however, seems to be PS integrin-independent, because cells derived from myospheroid embryos equally attached to this matrix protein. Then, we have to ask what the authentic ligands for Drosophila PS integrins are. This question remains to be answered, mainly because we cannot readily test Drosophila extracellular matrix proteins, although some matrix proteins have been identified in this species (Montell and Goodman, 1988, 1989; Fessler and Fessler, 1989). The present finding would encourage one to identify vitronectin-like molecules in Drosophila.
It is well known that RGD peptides are strong inhibitors for cell spreading on both vitronectin and fibronectin in the vertebrate system. However, we found that these peptides did not inhibit cell spreading in our Drosophila system. This result may not be surprising since some members of the integrin family do not recognize the RGD sequences (see for example Dufour et al. 1988; Springer, 1990), and some integrins might recognize not only the RGD sequence but also other sites on a ligand molecule (Humphries et al. 1986; Obara et al. 1988). From this point of view, Drosophila PS integrins must recognize some other sites than RGD on the vertebrate vitronectin and fibronectin.
The present results also suggest that other members of the integrin family or other types of receptors for extracellular matrix proteins are expressed in MLDmBG-1 cells. The inhibition of cell spreading with aBG-1 was not complete; about 15 % of the cells seeded on vitronectin still spread even at a saturating concentration of this antibody. In addition, this antibody did not inhibit the attachment of the cells. This cell line seems to be useful for the identification of such molecules.
ACKNOWLEDGEMENTS
We are indebted to Drs M. Leptin and M. Wilcox for antibodies to PS antigens and ifk27e fly, and their critical comments on the manuscript, and Dr K. Zinn for the cDNA library. We also thank Dr K. Yamada for their valuable suggestions. This work was supported by a Grant-in-Aid for Priority Area No. 02221101 from the Ministry of Education, Science and Culture of Japan. S.H. is a recipient of a Fellowship of the Japan Society for the Promotion of Science for Japanese Junior Scientists.