ABSTRACT
Decapentaplegic (Dpp) is a Drosophila member of the Transforming Growth Factor-β (TGF-β)/Bone Morpho-genetic Protein (BMP) superfamily of growth factors. Dpp serves as a classical morphogen, where concentration gradients of this secreted factor control patterning over many cell dimensions. Regulating the level of Dpp signaling is therefore critical to its function during development. One type of molecule proposed to modulate growth factor signaling at the cell surface are integral membrane proteo-glycans. We show here that division abnormally delayed (dally), a Drosophila member of the glypican family of integral membrane proteoglycans is required for normal Dpp signaling during development, affecting cellular responses to this morphogen. Ectopic expression of dally+ can alter the patterning activity of Dpp, suggesting a role for dally+ in modulating Dpp signaling strength. These findings support a role for members of the glypican family in controlling TGF-β/BMP activity in vivo by affecting signaling at the cell surface.
INTRODUCTION
The Transforming Growth Factor-β (TGF-β) class of secreted molecules serve as critical regulators of differentiation and cell division during development. Recently, a Drosophila member of the TGF-β/Bone Morphogenetic Protein (BMP) superfam-ily, Decapentaplegic (Dpp), has been shown to have morphogen activity, directing distinct cellular responses at different extracellular concentrations (Lecuit et al., 1996; Nellen et al., 1996). Gradients of Dpp can affect patterns of gene expression and are critical for establishing the anterior-posterior axis of the wing, as well as the proximal-distal axis of the leg (Campbell et al., 1993; Diaz-Benjumea et al., 1994; Jiang and Struhl, 1996; Penton and Hoffmann, 1996). Regu-lating the level of Dpp signaling is therefore critical to its role in patterning during development. One potential site for regu-lating Dpp signaling is at the cell surface, where Dpp binds to its signaling receptor, a heteromer of both type I and type II serine/threonine kinases encoded by the punt (pnt), saxaphone (sax) and thickvein (tkv) genes (Brummel et al., 1994; Penton et al., 1994; Ruberte et al., 1995).
A class of cell surface molecule implicated in regulating growth factor signaling is the integral membrane proteoglycans (reviewed in David, 1993). Proteoglycans bear long unbranched disaccharide polymers, glycosaminoglycans, attached to serine residues of the core protein. These sugar polymers bind a host of extracellular molecules including many growth factors. Studies using tissue culture systems have shown that cell-associated glycosaminoglycans affect signaling mediated by Fibroblast Growth Factor (FGF) (Olwin and Rapraeger, 1992; Rapraeger et al., 1991), Wingless (Wg) (Reichsman et al., 1996), TGF-β (Lopez-Casillas et al., 1993), Hepatocyte Growth Factor (HGF) (Zioncheck et al., 1995) and Heparin Binding-Epidermal Growth Factor (HB-EGF) (Aviezer and Yayon, 1994). Betaglycan, a molecule identified on the basis of its affinity for TGF-β, is a transmembrane pro-teoglycan that potentiates TGF-β responses in transfected cells by promoting the interaction of TGF-β with its signaling receptors (Lopez-Casillas et al., 1993). Both syndecans and glypicans, two different types of integral membrane proteo-glycans, can affect responses to FGF in tissue culture cells (Mali et al., 1993; Steinfeld et al., 1996). These studies have made it clear that cell surface proteoglycans can affect growth factor signaling but do not address the role of these molecules in vivo, or during development.
The importance of cell surface proteoglycans in vivo was recently demonstrated by the identification of mutants in Drosophila and human glypicans. Glypicans are attached to the outer leaflet of the plasma membrane via a glycosylphos-phatidylinositol linkage and bear glycosaminoglycan chains of the heparan sulfate type (David et al., 1990; Filmus et al., 1995; Karthikeyan et al., 1992; Litwack et al., 1994; Stipp et al., 1994; Watanabe et al., 1995). The Drosophila glypican, division abnormally delayed (dally) is required for the control of cell division in the developing visual system, and morpho-genesis of the eye, wing, antenna and genitalia (Nakato et al., 1995). Mutations in human Glypican-3 are responsible for the Simpson-Golabi-Behmel Syndrome, a disease characterized by prenatal and postnatal overgrowth, a host of morphological abnormalities, and a high incidence of neuroblastomas and Wilm’s tumors (Pilia et al., 1996). While these findings have established the importance of glypicans in morphogenesis and cell division control, the mechanism of their action has not been examined in vivo. We show here that dally affects cellular responses to Dpp and is required for normal Dpp signaling during development. These findings are consistent with a role for glypicans as growth factor co-receptors and suggest that they serve to control the activity of morphogens at the cell surface.
MATERIALS AND METHODS
Genetic interaction experiments
Genetic interactions between dally and dpp were tested using dally alleles, dallyP2, dallyΔP-188 and dallyΔP-527 (Nakato et al., 1995), and dpp alleles dppd5, dppd6, dppd12, dppd14 and dppblk (Spencer et al., 1982; Blackman et al., 1987; Masucci, 1990). dallyP2 and dallyΔP alleles were derived from different mutageneses experiments. Increas-ing the dosage of dpp+ was accomplished using CyO23, a second chromosome balancer bearing a transposed dpp+ gene (Wharton et al., 1993). All flies were reared at 25°C unless stated otherwise.
Monitoring expression of dpp and dpp target genes in imaginal tissues
We examined the expression of dpp in third instar imaginal discs using a dpp-lacZ (BS3.0) reporter that accurately reflects the level of dpp transcription, and all β-galactosidase staining of larval tissues was performed according to previously described methods (Blackman et al., 1991). We monitored the expression of two dpp target genes optomotor blind (omb) and spalt (sal) (Grimm and Pflugfelder, 1996; Kuhnlein et al., 1994; Pflugfelder et al., 1990), using the enhancer trap insert in omb, ombP1 (Sun et al., 1995), and for sal expression either a sal::lacZ line (Kuhnlein et al., 1994), or anti-Spalt (Kuhnlein et al., 1994) and Spalt-R antibodies (Reuter et al., 1996). Antibody stainings were performed at 1:50 dilution according to procedures described (Nakato et al., 1995) and visualized using an MRC 600 scanning confocal microscope.
Ectopic expression of dpp+ and dally+
A UAS-dally+ transgene was made by cloning a dally cDNA into the polylinkers of pUAST (KpnI-XbaI dally fragment) (see Brand and Perrimon, 1993), and a hs-dally+ construct using pCaSpeR-hs (dally cDNA fragment SphI-XbaI into the EcoRI/XbaI sites of the polylinker) (see C. Thummel, and V. Pirrotta, Drosophila Information Newsletter volume 2 for map of pCaSpeR-hs vector). Transgenic flies were created using standard P-element transformation procedures.
A UAS-dpp+ transgenic fly (94B1) together with the A9 GAL4 line (kindly provided by M. O’Connor) were used to direct expression of dpp+ in the wing pouch. Animals were reared at 16°C for these exper-iments. Ectopic expression of dpp+ in the anterior segment of the developing wing margin was accomplished using a dominant allele of hedgehog, Moonrat (hhMrt) (Tabata et al., 1995).
RESULTS
dpp is a genetic enhancer of dally
As a first step in determining if dally could play a role in Dpp signaling, we tested the ability of dpp mutants to affect the severity of the phenotypes found in dally adults. Flies het-erozygous for dpp and dally show phenotypes never observed in animals heterozygous for either dpp or dally alone, and the reduction in the eye observed for dally homozygotes is greatly enhanced by reducing dpp function. Shown in Fig. 1 are eye phenotypes observed in animals heterozygous for dppblk, an eye specific, viable allele, and either heterozygous or homozy-gous for dally. Heterozygosity for several dpp alleles also increased the penetrance of phenotypes found in dally homozy-gotes for eye, antenna and genitalia defects (Fig. 2A). In addition, the severity, or expressivity, of the phenotypes was also increased by reducing dpp function. For example, 31% of the dppd14/+; dallyP2/dallyP2 flies shown in Fig. 2A had severe eye abnormalities (eyes reduced by greater than 50%), while none of the dallyP2/dallyP2 animals had eye defects scored as severe. The wing phenotypes found in dally mutants, incom-plete wing vein V and wing notching, were suppressed by reducing dpp function, however, suggesting that dally is doing something different in the wing disc than it is in other imaginal tissues.
dpp is a dominant enhancer of dally. dally adult phenotypes are enhanced by dpp mutations. Flies heterozygous for both dally and dpp show eye abnormalities never observed in animals heterozygous for dally or dpp alone (top row). Animals homozygous for dally show more severe reductions in the eye when heterozygous for dpp (bottom row). (All heterozygotes bear dallyP2 and dppblk mutations over TM3 or CyO respectively, balancer chromosomes wild type for dally function.) The scale bar represents 100 μm.
dpp is a dominant enhancer of dally. dally adult phenotypes are enhanced by dpp mutations. Flies heterozygous for both dally and dpp show eye abnormalities never observed in animals heterozygous for dally or dpp alone (top row). Animals homozygous for dally show more severe reductions in the eye when heterozygous for dpp (bottom row). (All heterozygotes bear dallyP2 and dppblk mutations over TM3 or CyO respectively, balancer chromosomes wild type for dally function.) The scale bar represents 100 μm.
Tissue-specific genetic interactions between dpp and dally mutants. (A) Adult phenotypes of dallyP2 mutants are affected by dppd14. The penetrance of eye, antenna and genitalia defects are increased by reducing dpp function, while wing phenotypes are suppressed. The graph compares dppd14/+; dallyP2 (n=95) to Sco/+; dallyP2 (n=97) animals. (B) More severe alleles of dpp enhance dally phenotypes to a greater degree. dpp/CyO; dallyΔP-188/TM3 trans-heterozygotes with a more severe disk V allele (dppd12, n=245) show higher penetrance of eye and antenna defects than trans-heterozygotes with a milder disk II allele (dppd5, n=156). Similar effects are seen in dallyP2 homozygotes with other disk V and disk III alleles of dpp (data not shown).
Tissue-specific genetic interactions between dpp and dally mutants. (A) Adult phenotypes of dallyP2 mutants are affected by dppd14. The penetrance of eye, antenna and genitalia defects are increased by reducing dpp function, while wing phenotypes are suppressed. The graph compares dppd14/+; dallyP2 (n=95) to Sco/+; dallyP2 (n=97) animals. (B) More severe alleles of dpp enhance dally phenotypes to a greater degree. dpp/CyO; dallyΔP-188/TM3 trans-heterozygotes with a more severe disk V allele (dppd12, n=245) show higher penetrance of eye and antenna defects than trans-heterozygotes with a milder disk II allele (dppd5, n=156). Similar effects are seen in dallyP2 homozygotes with other disk V and disk III alleles of dpp (data not shown).
To further test the hypothesis that dally affects events directed by dpp, we have examined if more severe dpp mutants alter dally phenotypes to a greater degree. Shown in Fig. 2B are the phenotypes found in dpp, dally heterozygotes, for dppd5, and dppd12, mild and severe alleles of dpp respectively, and another, independently derived allele of dally, dallyΔP-188. As expected the penetrance of dpp/+; dally/+ phenotypes increased with the severity of the dpp allele (Fig. 2B). The expressivity of the phenotypes are likewise affected in more severe versus mild dpp alleles for both dally heterozygotes and homozygotes (data not shown).
dally phenotypes are rescued by increasing the dosage of dpp+
If dally mutations compromise Dpp signaling, dally mutants should be rescued by increasing the expression of dpp+. Increasing the dosage of dpp+ does indeed suppress dally phe-notypes in the eye, antenna and genitalia, the same tissues where we observe enhancement of phenotypes with partial loss-of-function dpp alleles (Fig. 3). Consistent with the genetic interactions between dpp and dally mutants, increasing the dosage of dpp+ has the opposite effect in the wing as it does in the eye, antenna and genitalia; dally wing phenotypes are enhanced in the presence of 3 copies of dpp+. These findings suggest that dally is antagonistic to Dpp signaling in the wing, in contrast to its function in the eye, antenna and genitalia.
An additional copy of dpp+ rescues dally mutants. dallyP2 homozygotes with 3 copies of dpp+ (n=408) exhibited lower penetrance for abnormalities in the eye, antenna and genitalia than their siblings with 2 copies of dpp+ (n=374). dally wing phenotypes however, were enhanced by increasing the dosage of dpp+. To compare dally homozygotes with and without CyO23 (a chromosome bearing a transposed dpp+ gene), CyO23/Sco; dally/TM3, Sb were crossed to dally/TM3, Sb flies. Siblings bearing either Sco, or CyO23 were compared. dally homozygotes from these crosses showed lower penetrance of adult phenotypes than in other experiments, presumably on account of other chromosomes introduced in establishing the Sco/CyO23; dally/TM3, Sb stock. The same effect of CyO23 for eye and genitalia phenotypes was observed with a milder, independently derived dally allele, dallyΔP-188 (data not shown).
An additional copy of dpp+ rescues dally mutants. dallyP2 homozygotes with 3 copies of dpp+ (n=408) exhibited lower penetrance for abnormalities in the eye, antenna and genitalia than their siblings with 2 copies of dpp+ (n=374). dally wing phenotypes however, were enhanced by increasing the dosage of dpp+. To compare dally homozygotes with and without CyO23 (a chromosome bearing a transposed dpp+ gene), CyO23/Sco; dally/TM3, Sb were crossed to dally/TM3, Sb flies. Siblings bearing either Sco, or CyO23 were compared. dally homozygotes from these crosses showed lower penetrance of adult phenotypes than in other experiments, presumably on account of other chromosomes introduced in establishing the Sco/CyO23; dally/TM3, Sb stock. The same effect of CyO23 for eye and genitalia phenotypes was observed with a milder, independently derived dally allele, dallyΔP-188 (data not shown).
dally mutants show reduced expression of dpp target genes
The genetic interaction between dpp and dally suggested that Dpp signaling is compromised in dally mutants. We therefore examined the level of expression of two genes, omb and sal, that are activated by Dpp signaling (Grimm and Pflugfelder, 1996; Kuhnlein et al., 1994). Fig. 4 shows the results of one experiment where we determined that omb expression is severely reduced in the antenna and eye discs of dally mutants. We observe similar reductions in sal expression for these imaginal tissues, as well as in genitalia discs of dally mutants (data not shown).
Expression of the dpp target gene, omb, is reduced in dally mutants. Eye (labelled E)/antenna (labelled A) disc complexes from third instar larvae bearing an omb::lacZ reporter ombP1, and either wild-type or homozygous for dallyΔP-527 (upper two panels). A dpp-lacZ reporter shows mild reductions in expression in dally mutants (compare lower two panels). The scale bar represents 100 μm.
Expression of the dpp target gene, omb, is reduced in dally mutants. Eye (labelled E)/antenna (labelled A) disc complexes from third instar larvae bearing an omb::lacZ reporter ombP1, and either wild-type or homozygous for dallyΔP-527 (upper two panels). A dpp-lacZ reporter shows mild reductions in expression in dally mutants (compare lower two panels). The scale bar represents 100 μm.
The reduction in Dpp signaling found in dally mutants could be accounted for by either affecting the level of Dpp itself, or altering the responses of cells to Dpp. We therefore examined dpp expression in dally mutants using a dpp-lacZ reporter gene. In the eye and antenna discs we observed a very modest reduction in dpp-lacZ expression compared to the severe loss of target gene expression (Fig. 4). These findings suggest that the primary effect of dally mutations is to affect cellular responses to Dpp and not expression of the growth factor itself. In fact Dpp has been shown to activate its own expression in imaginal discs and compromising Dpp reception would therefore be expected to also reduce dpp expression to a degree (Chanut and Heberlein, 1997; Pignoni and Zipurski, 1997).
dally mutants suppress phenotypes resulting from ectopic expression of Dpp in the wing disc
The analysis of dally mutants described above suggested that dally serves a function downstream of dpp, affecting cellular responses to this secreted factor. To test this relationship directly, we ectopically expressed dpp+ in the wing disc and determined if reducing dally function could suppress the over-growth and patterning defects that result from ectopic Dpp signaling. Our first set of experiments made use of hhMrt, a dominant allele that results in ectopic expression of dpp+ in the anterior compartment of the wing disc (Tabata et al., 1995). hhMrt adults show overgrowth and wing vein patterning defects limited to the anterior segment of the wing. These phenotypes are rescued in a graded fashion by reducing dally function (Fig. 5). We observed a partial rescue in dally heterozygotes and a complete suppression of defects in dally homozygotes. A dally enhancer trap insertion shows that dally is expressed along the wing margin, where reductions in its function could rescue the effects of ectopic dpp+ expressed along the anterior segment of the future wing margin in hhMrt mutants (Fig. 5).
dally mutants suppress phenotypes generated by ectopic expression of dpp+ in the wing disc. The top row shows adult wings from wild-type and hhMrt mutants either heterozygous or homozygous for dally (arrow marks a wing vein defect characteristic of dally mutants). The second row of panels shows the expression pattern of dpp-lacZ in third instar larval wing discs of these same genotypes (arrows mark ectopic dpp+ expression). The third row shows the pattern of sal::lacZ in third instar discs, with the exception of dally hhMrt/dally where anti-SalM and anti-SalR antibodies were used to detect the expression of SalM and SalR proteins. Scale bar represents 100 μm. The image in the bottom row shows the expression of a dally enhancer trap insert (dallyP2), in a third instar larval wing disc. The bracket marks a 10-to 14-cell-wide stripe of dally-expressing cells at the future wing margin.
dally mutants suppress phenotypes generated by ectopic expression of dpp+ in the wing disc. The top row shows adult wings from wild-type and hhMrt mutants either heterozygous or homozygous for dally (arrow marks a wing vein defect characteristic of dally mutants). The second row of panels shows the expression pattern of dpp-lacZ in third instar larval wing discs of these same genotypes (arrows mark ectopic dpp+ expression). The third row shows the pattern of sal::lacZ in third instar discs, with the exception of dally hhMrt/dally where anti-SalM and anti-SalR antibodies were used to detect the expression of SalM and SalR proteins. Scale bar represents 100 μm. The image in the bottom row shows the expression of a dally enhancer trap insert (dallyP2), in a third instar larval wing disc. The bracket marks a 10-to 14-cell-wide stripe of dally-expressing cells at the future wing margin.
Consistent with the effects in the adult wing, dally mutants reduce the ectopic activation of dpp target genes in hhMrt wing discs without eliminating the expression of dpp-lacZ (Fig. 5). 32% of the hhMrt, dally/dally discs show ectopic dpp+ expression (Fig. 5, n=197), and 100% of these larvae show complete rescue of the adult wing phenotype (n=823). Thus, there remains significant levels of ectopic dpp+ expression that is without effect when dally function is decreased.
The reductions in dpp-lacZ expression that we observed in dally mutants bearing the hhMrt mutation prompted us to use another system for ectopically expressing dpp+, where autoregulation of dpp would not play a role. In addition, Hh is known to bind heparan sulfate and it was a formal possi-bility that Dally could affect Dpp signaling via its interaction with Hh (Lee et al., 1994). We therefore tested the ability of dally mutants to rescue phenotypes produced by ectopic expression of Dpp derived from a UAS-dpp+ transgene. Con-sistent with our experiments using hhMrt, dally mutations sup-pressed phenotypes produced by ectopic Dpp (Fig. 6), as well as reduce the activation of dpp target genes (data not shown). These findings show that dally affects events downstream of Dpp, altering the responses of cells to this secreted growth factor.
Effects of decreasing or increasing dally function on phenotypes produced by ectopic expression of Dpp from a UAS-dpp+ transgene. For the experiments depicted in the top row, ectopic expression of dpp+ in the wing disc was produced using a GAL4 enhancer trap insert (A9) combined with a UAS-dpp+ transgene. Flies heterozygous for dallyΔP-527 rescue wing margin defects resulting from ectopic dpp+ derived from this heterologous expression system (compare top two panels). The arrowheads mark wing vein defects produced by ectopic dpp+ that are suppressed by reducing dally function (top row). hs-dally+ and UAS-dally+ transgenes convert the wing overgrowth phenotype observed in hhMrt/+ animals to mirror image wing duplications (bottom two rows). The pattern of wing margin bristles show the mirror-image symmetry of the duplicated structures (see for example, the high magnification view of a wing from a hs-dally+ bearing animal, lower right-hand panel).
Effects of decreasing or increasing dally function on phenotypes produced by ectopic expression of Dpp from a UAS-dpp+ transgene. For the experiments depicted in the top row, ectopic expression of dpp+ in the wing disc was produced using a GAL4 enhancer trap insert (A9) combined with a UAS-dpp+ transgene. Flies heterozygous for dallyΔP-527 rescue wing margin defects resulting from ectopic dpp+ derived from this heterologous expression system (compare top two panels). The arrowheads mark wing vein defects produced by ectopic dpp+ that are suppressed by reducing dally function (top row). hs-dally+ and UAS-dally+ transgenes convert the wing overgrowth phenotype observed in hhMrt/+ animals to mirror image wing duplications (bottom two rows). The pattern of wing margin bristles show the mirror-image symmetry of the duplicated structures (see for example, the high magnification view of a wing from a hs-dally+ bearing animal, lower right-hand panel).
Ectopic dally+ potentiates the patterning activity of ectopic Dpp
Dpp can serve as a morphogen, directing different patterns of gene expression at different extra-cellular concentrations (Lecuit et al., 1996; Nellen et al., 1996). The effect of dally on Dpp signaling suggests that Dally serves to modulate the activity of Dpp. We tested this idea by providing ectopic dally+ expression to a level of ectopic dpp+ that produces overgrowth and pattern-ing defects in the wing, namely the phenotypes found in hhMrt mutants. Flies bearing UAS-dally+ or hs-dally+ transgenes potentiate the activity of Dpp, converting overgrowth phenotypes to wing duplications (Fig. 6, we have observed wing duplications with two independent UAS-dally+ insertions on the X and another on the second chromosome). Wing duplications are produced by high levels of ectopic dpp+ expression, and in animals bearing mitotic clones of dpp+-expressing cells(Kojima et al., 1994; Zecca et al., 1996). Modest levels of ectopic Dally are apparently sufficient to have a marked effect on Dpp signaling since we observe wing duplications with several UAS-dally+ inserts in the absence of a GAL4 activator. UAS constructs can have some low level of tran-scription in the absence of the GAL4 activator and this apparently accounts for the effects that we have observed with multiple UAS-dally+ inserts.
DISCUSSION
A role for cell surface proteoglycans in growth factor signaling during development
A great deal of work has implicated cell surface proteoglycans as regulators of growth factor signaling, affecting the assembly or composition of the ligand-signaling receptor complexes. Betaglycan has been shown by cross-linking studies to serve as a co-receptor for TGF-β (Lopez-Casillas et al., 1993), and both syndecans and glypicans affect responses to FGF in tissue culture cells (Mali et al., 1993; Steinfeld et al., 1996). The recent identification of a glypican in Drosophila, dally, has allowed us to determine if a cell surface proteoglycan affects events directed by a secreted growth factor in vivo. Our findings demonstrate that (1) dally is required for normal Dpp signaling in at least some imaginal tissues during development, (2)dally functions downstream of Dpp, affecting cellular responses to this morphogen, (3) dally+ can increase the pat-terning activity of Dpp, suggesting a role for dally+ in modu-lating Dpp signaling strength.
The genetic interaction experiments described here have made use of known hypomorphic dpp and dally alleles. In fact, molecular and genetic characterization of dally mutants suggests that all the alleles that we have obtained thus far are hypomorphic (Nakato et al., 1995). We have isolated over 20 excision-derived alleles of dally and the most severe of these do not remove the initiating AUG codon although the deletions approach to within a few tens of base pairs (H. Nakato, unpub-lished data). These findings suggest that our failure to obtain a null allele is because dally loss-of-function alleles are haplo-lethal. It is therefore possible that the incomplete penetrance that we observe with existing dally alleles in some tissues reflects that some dally functions remains in these animals.
Our findings are entirely consistent with Dally serving as a co-receptor, where Dally binds Dpp and participates in forming a signaling receptor complex. However, other molecular mech-anisms are possible. First, dally could affect a parallel signaling pathway that alters the responses of cells to activation of Dpp signaling. If Dally does affect the Dpp signaling pathway directly, it could influence the distribution or availability of Dpp. It is also possible that Dally and its associated heparan sulfate affect the activity of extracellular enzymes that regulate Dpp. Tolloid, a metalloprotease related to BMP-1, potentiates Dpp activity and could potentially be a target for glycosamino-glycan regulation of its protease activity (Finelli et al., 1995; Shimell et al., 1991). Heparin, a short chain glycosaminogly-can synthesized by mast cells and basophils activates the protease inhibitor Antithrombin III, providing a precedent for a glycosaminoglycan controlling extracellular enzyme activity (reviewed in Hornebeck et al., 1994). Whatever the mechanism, cell surface proteoglycans add another dimension to the regu-lation of growth factor signaling at the cell surface.
Recently, a gene encoding an enzyme required for gly-cosaminoglycan synthesis, UDP glucose dehydrogenase, has been shown to be involved in Wg signaling (Haerry et al., 1997; Häcker et al., 1997; Binari et al., 1997). Mutations in UDP-glucose dehydrogenase are also capable of suppressing phenotypes resulting from ectopic activation of the Dpp signaling pathway in the wing disc (Haerry et al., 1997). These findings make it evident that proteoglycans can affect signaling mediated by both Wg and Dpp. Dally bears heparan sulfate chains (T. Futch, M. Tsuda and S. Selleck, unpublished data), glycosaminoglycans that require UDP-glucose dehydrogenase activity for their synthesis. It is therefore possible that the phe-notypes observed in UDP-glucose dehydrogenase mutants reflect in part, changes in Dally function.
Dally has different functions in different tissues
Our genetic analysis of dpp-dally interactions showed that dally does not serve the same role in all tissues. dally mutant phenotypes are enhanced by decreasing dpp function in the eye, antenna and genitalia, but are suppressed by dpp mutations in the wing. Increasing dpp+ function likewise rescues eye, antenna and genitalia defects, while enhancing wing abnor-malities. These findings show that the interaction between dpp and dally in the wing cannot be accounted for by the trivial explanation that existing dally alleles affect the wing to a lesser degree.
In fact the pattern of dally enhancer trap expression along the third instar disc wing margin is coincident with the domain of cells that show immunoreactive Wg and are known to respond to this secreted protein (Couso et al., 1994; Rulifson et al., 1996). We have recently determined that dally expression in the embryo is in segmentally repeated stripes that coincide with engrailed-expressing cells (C. Golden and E. Siegfried, unpublished observations), cells known to respond to Wg (reviewed in Klingensmith and Nusse, 1994). Perhaps Dally affects Wg-mediated events at the developing wing margin and in the embryonic epidermis, in contrast to its role in promoting Dpp signaling in the eye, antenna and genitalia discs.
It is not clear whether a role for Dally in Wg signaling at the wing margin could account for the genetic interactions that we observe between dally and dpp in this tissue. The relation-ship between Wg and Dpp signaling is complex and cell type specific. In the leg, eye and antenna discs, for example, Wg and Dpp signaling are mutually antagonistic (Jiang and Struhl, 1996; Theisen et al., 1996), whereas in the wing disc Wg is apparently not able to repress dpp expression (Heslip et al., 1997). The wing margin is particularly complex, with Wg repressing its own activation in some cells in order to refine the boundary between wg-expressing and proneural cells (Rulifson et al., 1996). In embryonic midgut, Wg and Dpp actually promote each others expression (Immerglück et al., 1990; Yu et al., 1996). Further study will be required to determine if Dally signaling somehow suppresses the activity of Dpp at the wing margin via Wg.
dally-and dpp-mediated control of cell division
A link between dally and dpp is further established by the cell division defects found in the eye disc of dally mutants and the requirement for Dpp signaling in the normal synchronization of the cell cycle within the morphogenetic furrow (Penton et al., 1997). dally mutants show a delayed entry into M phase from G2 for the first synchronized division in the morpho-genetic furrow (Nakato et el., 1995). Loss of Dpp signaling in clones of cells defective for tkv, sax or shn, disrupts Cyclin B expression in this same division cycle, consistent with a failure to progress into M phase. Thus dally mutants phenocopy cell cycle abnormalities found when Dpp signaling is compromised.
Dpp clearly promotes cell division in Drosophila (Burke and Basler, 1996) and several lines of evidence suggest this regu-lation occurs in G2. First, the analysis of Dpp signaling defective clones in the eye, indicate a G2-M failure (Penton et al., 1997). Second, cells throughout the wing disc, which are known to require Dpp for their division, are found in clusters synchronized in G2 (Milan et al., 1996). These cells are not clonally related and therefore must be receiving some inter-cellular signal to arrest in G2, an arrest that is apparently overcome by Dpp. Dpp clearly affects both cell division and differentiation during development. How are these functions of Dpp related and do cell surface proteoglycans affect the nature of the Dpp signal?
Glypicans as regulators of tissue growth and tumor progression
Recently the human overgrowth syndrome, Simpson-Golabi-Behmel Syndrome (SGBS) was shown to result from loss-of-function mutations in Glypican-3, one of several members of this gene family represented in vertebrates (Pilia et al., 1996). Remarkably, SGBS patients, in addition to abnormalities in somatic growth and morphogenesis, have a high susceptibility to neuroblastomas and Wilm’s tumors of the kidney. Our findings showing a functional interaction between glypicans and the TGF-β superfamily of growth factors raise the possi-bility that SGBS patients are defective for TGF-β signaling. In contrast to Drosophila where Dpp promotes cell division (Burke and Basler, 1996; Penton et al., 1997), in many ver-tebrate cells, TGF-β arrests cell cycle progression (for review see Massague and Polyak, 1995). In addition, both TGF-βs and BMPs are able to induce programmed cell death in vertebrates (Rorke and Jacobberger, 1995; Tsukada et al., 1995; Wang et al., 1995; Zou and Niswander, 1996). Loss of Glypican-3 function in SGBS patients could compromise TGF-β/BMP-mediated cell cycle arrest and apoptosis, leading to overgrowth and tumor progression.
Given the number of growth factors that bind heparan sulfate, and the varied expression of different glypicans in tissues (David et al., 1990; Filmus et al., 1995; Karthikeyan et al., 1992; Litwack et al., 1994; Pilia et al., 1996; Stipp et al., 1994; Watanabe et al., 1995), it is possible that this family of integral membrane proteoglycans serves to regulate growth factor signaling in a wide range of contexts. Growth factors have different effects in different cells and perhaps cell surface proteoglycans provide one mechanism of controlling the nature or strength of the signal generated.
ACKNOWLEDGEMENTS
We thank F. M. Hoffmann, W. Gelbart, T. Tabata, M. O’Connor and S. Carroll for kindly providing dpp mutants, BS3.0, hhMrt bearing flies, the A9 GAL4 line, and α-SalM and α-SalR antibodies. We are indebted to David Bentley for his expertise with Scanning Electron Microscopy; Momoko Fujise for ably assisting with the hhMrt exper-iments, Marc Brabant and Tom Bunch for advice on P-element trans-formation, Huan Vuong for assistance with mapping of UAS-dally+ transformants, and T. Weinert, M. Ramaswami, A. Lander, S. Saunders and anonymous reviewers for careful reading of the manu-script. This work was supported by grants from the March of Dimes and the American Cancer Society (DB-111). S. B. S. is a fellow of the Arthur P. Sloane Foundation.