Developmental territories created by mutual antagonism between Wingless and Decapentaplegic

The generation of at least three axes of asymmetry in appendage primordia is an essential element of patterning (see Postlethwait, 1978; Cohen, 1993). The A/P axes in Drosophila appendages are established by the A/P compartments (strict lineage restrictions) (Garcia-Bellido et al., 1973, 1976; Garcia- Bellido, 1975; Steiner, 1976; Wieschaus and Gehring, 1976; Lawrence and Morata, 1977) which are defined by non-over- lapping expression of cubitus interruptus(ci) in the anterior compartment (Eaton and Kornberg, 1990; Dominguez et al., 1996) and engrailed (en) in the posterior compartment (Kornberg et al., 1985; Sanicola et al., 1995; Zecca et al., 1995). The apical basal polarity of the epithelium may provide another axis. After the establishment of A/P asymmetry, the wing disc is further subdivided into dorsal and ventral com- partments (Bryant, 1970; Garcia-Bellido et al., 1976; Blair, 1993) by the activity of the apterous (ap) gene (Blair, 1993; Diaz-Benjumea and Cohen, 1993; Lawrence and Morata, 1993; Williams et al., 1993; Blair et al., 1994) but it has not been possible to demonstrate a D/V lineage restriction in the leg (Steiner, 1976). Is it possible that the establishment of D/V asymmetry in the leg utilizes a mechanism different from a lineage restriction?

In third instar leg discs, wingless (wg) is expressed in a ventral anterior wedge (Baker, 1988b; Couso et al., 1993) while decapentaplegic (dpp) is expressed strongly in an anterior stripe that abuts the A/P compartment boundary, and weakly in a ventral domain that overlaps wg expression (Masucci et al., 1990; Raftery et al., 1991). Several observa- tions suggest that WG and DPP can interact in particular tissues (Jackson and Hoffmann, 1994; Kaphingst and Kunes, 1994; Staehling-Hampton and Hoffmann, 1994; Tabata and Kornberg, 1994; Ma and Moses, 1995; Treisman and Rubin, 1995; Penton and Hoffmann, 1996; Wiersdorff et al., 1996; Yu et al., 1996). We have compared the effect of WG and DPP signaling on each other’s transcription and on cell fate speci- fication in discs. We find that WG signaling inhibits dpp expression, and DPP signaling inhibits wg expression, in leg and eye/antennal discs but not in dorsal discs (i.e. wing and haltere). This mutual repression provides a mechanism for maintaining separate regions of gene expression in developing leg discs. The D/V restriction of wg and dpp divides the anterior compartment into dorsal and ventral territories, thus creating a D/V axis of asymmetry that is necessary for gener- ating a chiral appendage. We define a territory as a group of cells that are under the domineering influence of a particular morphogen. Neighboring territories are not separated by lineage restriction nor are cells within a territory or their descendants ‘determined’ (Lawrence and Struhl, 1996) to remain as part of that territory. Descendants of cells from one territory that, by growth displacement or injury, find them- selves closer to the source of another morphogen will become part of the territory defined by the second morphogen. In the leg disc, cells primarily within the range of influence of DPP will acquire dorsal/dorsolateral positional values, while cells that lie within the range of influence of WG will acquire ventral/ventrolateral positional values regardless of their lineage. Cells in the zone between the two territories will receive varying levels of both signals. The mutual repression will serve to maintain separate territories of morphogen expression.

A y w sn stock was used as a control for the temperature upshifts, the mRNA in situ hybridizations and the immunostaining reactions because this allowed controls and mutants to be stained in the same tubes and then separated by color of mouthparts, thus, permitting comparisons of the levels of staining.

Three alleles of punt were tested as homozygotes and in heteroal- lelic combinations with each other for temperature sensitivity. All het- eroallelic combinations of punt^135-22, punt^Δ61 and puntP1 exhibited temperature sensitivity (Letsou et al., 1995). The punt^135-22 allele was a kind gift from A. Letsou, University of Utah. The punt mutant stocks were balanced with a TM6 balancer chromosome that carries the dominant marker Tubby (Tb). Heteroallelic punt larvae were identi- fied by their non-Tb phenotype. The temperature sensitive wg allele, wgI^L114 (Nüsslein-Volhard et al., 1984), was balanced with the compound balancer TSTL which has a translocation between the CyO and TM6B,Tb balancers. Homozygous mutant larvae were identified by their non-Tb phenotype.

dpp^blk/dpp^blk; dpp^blk>Gal4 [39B2]/TM6, Tb flies were mated to flies homozygous for one of the following transgenes: UAS>dpp- myc4, UAS>dpp-myc7 (kind gifts from R. Nichols and W. M. Gelbart, Harvard University) and UAS>wg (a kind gift from I. Livne-Bar and

H. Kraus, University of Toronto) to generate larvae expressing either dpp or wg along the A/P boundary. Larvae that received the dpp^blk>Gal4 driver were identified by their non-Tb phenotype. The dpp^blk>Gal4; UAS>dpp-myc4 and dpp^blk>Gal4; UAS>dpp-myc7 combinations gave similar results and are referred to as dpp^blk>dpp in this manuscript. Clones of tkv were generated by heat shocking y w P[hs-FLP1]; tkvIIB09 P[FRT40A]/P[y+25A]P[FRT40A] larvae.

wg and dpp expression were monitored by whole-mount in situ hybridization using digoxigenin-labeled antisense RNA probes. Plasmids used as templates for probes were wg651 (a kind gift from B. Cohen), a 3 kb wg cDNA, dppE55, a 4 kb dpp cDNA (Padgett et al., 1987), and hhC11, a 2.3 kb cDNA (Lee et al., 1992), all in blue- script. The probes were prepared according to the protocol accompa- nying the digoxigenin RNA labeling mix (Boehringer Mannheim 1277 073). Unincorporated nucleotides were removed by LiCl pre- cipitation and the RNA from 1 μg of template was resuspended in 100 μl DEPC water; 20 μl of each probe were hydrolyzed, LiCl precipi- tated, resuspended in 100 μl hybridization solution and diluted 1:50μl for the hybridization reaction.

The prehybridization procedure and hybridization conditions used are based on the protocol of Tautz and Pfeifle (1989) with the following modifications. Late third instar larvae and white prepupae were dissected in PBS (pH7), stored for less than 20 minutes on ice, and fixed for 20 minutes in 4% formaldehyde in PBT (PBS + 0.1% Tween 20). Discs were stored at −20°C in ethanol, progressively rehy- drated, and digested with 5 μg/ml proteinase K in PBT for 5 minutes. After rinsing, the discs were fixed in 4% formaldehyde in PBT for another 20 minutes. The hybridization was carried out at 55°C for 2 days and the discs were washed for 3 days at 55°C after hybridization. After one 20-minute 50:50 PBT:hyb solution rinse and five 25-minute PBT rinses, discs were incubated in polyclonal sheep anti-digoxigenin Fab fragments conjugated to alkaline phosphatase (Boehringer Mannheim 1093 274) diluted 1/2000 in PBT overnight at 4°C. The AP color reaction was developed according to the protocol accompa- nying the antibody.

Discs were fixed as above and incubated overnight at 4°C with anti- EN monoclonal antibody diluted 1:1 with PBT (PBS + 0.1% Triton X-100) + 3% BSA. An AP-conjugated secondary antibody (Jackson Immunological Laboratory) was used at a1/1000 dilution and the AP color reaction was developed as described for in situs.

The DPP receptor is a heterodimer composed of two distantly related transmembrane serine/threonine kinases called type I and type II receptors (reviewed by Massaguè, 1996). The punt gene encodes a type II receptor that is essential for reception of the DPP signal (Letsou et al., 1995; Ruberte et al., 1995). We used temperature sensitive punt mutants to determine the consequences of loss of DPP signaling on wg expression. Mutant larvae were transferred from 18°C to the restrictive temperature, 25°C, for different lengths of time during larval development (measured retrospectively from the time of puparium formation) and stained for wg mRNA.

Loss of punt function results in the ectopic activation of wg expression along the dorsal A/P boundary of the leg disc (Fig. 1). By 23 hours after upshifting puntP1/punt135-22 larvae, wg expression has expanded into the dorsal region (Fig. 1B), and it continues to expand until reaching a maximum after about 40 hours at the restrictive temperature (Fig. 1C). Similar results are also observed in the eye-antennal disc. The antennal portion of the eye-antennal disc is analogous to a leg disc, but inverted in the D/V and A/P axes (Struhl, 1981). Thus wg, which is expressed in the ventral region of the leg disc, is expressed in the dorsal region of the wild-type antennal disc (Baker, 1988a) (Fig. 1D). wg is also expressed around the periphery of the eye portion of the disc (Baker, 1988a; Ma and Moses, 1995) (Fig. 1D). When temperature sensitive punt mutants are upshifted to 25°C, wg expression expands into the ventral domain of the antenna to form a continuous stripe along the A/P boundary from dorsal to ventral (Fig. 1E). In addition, wg expression expands from its normal location at the periphery of the eye anlage into the morphogenetic furrow (Fig. 1E). In wing discs, the pattern of wg expression and the morphology of the discs is normal in pntts animals even with a 70-hour upshift. In legs, the maximal ectopic wg expression is seen after only 40 hours at restrictive temperature (Fig. 1C).

In summary, when DPP signaling is blocked by transferring temperature-sensitive punt mutants from 18°C to 25°C, ectopic wg expression appears in regions of the leg and eye-antennal discs that coincide with regions of dpp expression in the wild type.

The previous experiments show that loss of DPP signaling causes ectopic wg expression. To perform the reciprocal test, namely whether increased DPP signaling represses wg expression, we expressed increased levels of dpp in its normal domain and monitored wg expression. In leg discs, dpp is strongly expressed in a dorsal stripe that abuts the A/P boundary and is weakly expressed in the ventral region of the disc, where its expression overlaps the wg expression domain (Masucci et al., 1990; Raftery et al., 1991). A dpp^blink>Gal4 transgene (Staehling-Hampton et al., 1994) was used to drive expression of a UAS>dpp transgene. In the leg disc, this com- bination drives dpp expression in a two wedges that abut the anterior side of the A/P boundary (Fig. 2C). The increase in dpp expression in the ventral region of the leg disc causes a reduction of wg expression (Fig. 2B). The normal anterior ventral wedge of wg expression (Fig. 2A) is lost and wg expression is restricted to a narrow D/V stripe across the tip of the tarsus (Fig. 2B). This region of wg expression coincides with the only region along the A/P boundary that lacks dpp expression in this genetic background (Fig. 2C and Fig. 6A). Similar reductions in wg expression were also observed by expressing a ligand-independent, constitutively active type I DPP receptor, thick veins (tkv), using a dpp^blk>Gal4 driver in leg discs.

Unlike the punt temperature-shift experiments described previously, where the transcription response of wg occurs within 23 hours, expression of dpp driven by dpp^blk>Gal4 is continuous during the approximately 4.5 days of disc devel- opment. This provides time for regulative growth to occur in response to ectopic dpp expression. This pattern regulation produces leg discs that are wider in the anterior-posterior axis. To better understand the basis of the morphological changes, we monitored hedgehog (hh) expression to demarcate posterior from anterior compartments. The wide discs exhibit an expansion of the posterior compartment of the disc as evidenced by expanded hh expression (Fig. 2D and Fig. 6A). Thus, continuous elevated dpp expression in the ventral region of the leg disc suppresses wg expression and causes expansion of the posterior compartment in this region.

Expression of either dpp or ligand independent tkv in wing discs using the dpp^blk>Gal4 driver produced extremely wide wing discs but the pattern of wg expression, even in the face of extensive ectopic growth, was essentially normal. Since expression driven by the dpp^blk>Gal4 driver intersects the bands of wg expressesion only along the A/P boundary, we used the A9>Gal4 driver to express both activated tkv and dpp in a broad band that overlaps the band of wg expression along the wing margin. Again, no reduction in wg expression was observed, indicating that the mutual antagonism between WG and DPP does not operate in the wing.

The loss of wg expression in the dpp^blk>dpp discs and the expanded wg expression in the punt mutants demonstrates that DPP signaling restricts the wg expression domain in the leg disc.

To examine the effect of WG signaling on dpp expression, we used a temperature-sensitive allele of wg. Larvae, mutant for wgIL114 (Nüsslein-Volhard et al., 1984), were raised at 18°C, transferred to 25°C for 24 hours and 40 hours BPF and the expression of dpp in discs of white prepupae was examined. As controls, y w sn larvae were upshifted on the same schedule and stained in the same vials. Within 24 hours, loss of wg function causes dpp in the ventral region of the leg to increase and expand (Fig. 3B). In the eye-antennal disc, dpp expression also expands into the dorsal region of the antenna within 40 hours at 25°C (Fig. 3E). The changes in dpp expression caused by loss of wg function provide an explanation for the pheno- types seen in wgts pharate adults, which exhibit a duplication of dorsal in place of ventral structures in the legs (Couso et al., 1993).

If WG signaling represses dpp expression, then dpp expression should also expand when transduction of the wg signal is blocked, such as in dsh mutants (Klingensmith et al., 1994; Theisen et al., 1994). Maternal and zygotic loss of dsh causes embryonic lethality with a cuticular phenotype indis- tinguishable from that of wg mutants (Perrimon and Mahowald, 1987). However, when dsh is maternally supplied, homozygous mutant larvae survive to early third instar with very small leg discs but reasonably developed eye-antennal discs. In the antennal discs of these mutants, dpp (which is normally confined to the ventral region (Masucci et al., 1990), Fig. 3C) is ectopically expressed in the dorsal region of the antenna (Fig. 3D). In wing discs, the pattern of dpp expression and the disc morphology is normal in wg^ts animals upshifted for 49 hours, whereas a 24-hour upshift is sufficient to achieve maximal ectopic dpp expression in the leg discs (Fig. 3B). Thus, loss of WG signaling, causes dpp expression to expand in leg and eye-antennal discs but not in wing discs.

Blocking WG signaling causes an expansion of dpp expression. To test whether ectopic WG signaling inhibits dpp expression, we used the dpp^blk>Gal4 (Staehling-Hampton et al., 1994) driver to activate a UAS>wg transgene in a stripe along the A/P boundary of the leg disc (Fig. 4B). The ectopic expression of wg in the dorsal region of the disc suppresses expression of dpp to the point that the level of expression appears similar to or lower than the weak ventral expression of dpp seen in normal discs (compare Fig. 4A with 4C). Note also that the weak ventral expression of dpp which coexists with wg in normal discs is unaffected by ectopic expression of wg (Fig. 4C). Thus, ectopic expression of wg reduces dpp expression in leg discs. These discs become long and narrow in contrast to the short wide shape of discs with increased dpp signaling (Figs 4B-D compared with 2B-D and 6A). The expansion of en expression reveals that the dorsal posterior compartment is enlarged in these leg discs (Figs 4D and 6A). Wing discs with dppblk driving wg, exhibit extensive pattern regulation but strong expression of dpp in a D/V stripe remains, indicating that ectopic WG does not suppress dpp in wing discs.

To determine whether alterations in gene expression have pre- dictable developmental consequences, we examined pharate adults from upshifted puntts animals and adults with tkv clones for altered cuticular patterns. puntts pharate adults produced by late first/early second-instar upshifts have legs in which dorsal structures are absent and ventral structures are duplicated (Fig. 5A,B). The duplication of ventral leg elements is consistent with ectopic wg expression since wg is normally expressed ventrally (Baker, 1988b; Couso et al., 1993) and specifies ven- trolateral fates when ectopically expressed in the leg disc (Struhl and Basler, 1993; Wilder and Perrimon, 1995). Inter- estingly, the ectopic dorsal expression of wg in puntts animals is correlated with specification of extreme ventral fates in the dorsal region. For example, on the distal tibia of the second leg shown in Fig. 5B there are two apical bristles which are indi- cators of the ventral-most cell fate in leg discs. In addition, puntts pharate adults exhibit reduced or missing eyes (Fig. 5C) and duplicated antennae (Fig. 5D). The loss of eye tissue is consistent with the ectopic expression of wg in the eye field antagonizing DPP and consequently restricting the morpho- genetic furrow (Ma and Moses, 1995; Treisman and Rubin, 1995). The duplicated antennae branch from the ventral side as would be expected from regulatory growth if wg were ectopically expressed on this side.

The tkv gene encodes a type I receptor that is essential for reception of the DPP signal (Brummel et al., 1994; Nellen et al., 1994; Penton et al., 1994). We induced tkv clones in second instar larvae. In the tarsus, dorsal tkv clones convert dorsal cells to ventral or ventrolateral cell fates. These clones often produce outgrowths that include normal neighbors, some of which have also been respecified to ventral cell fates (not shown). This is consistent with DPP specifying dorsal cell fates and its loss converting cells to ventral identity followed by an intercalary response as expected when ventral cells confront dorsal cells (French et al., 1976). Unexpectedly, tkv clones in the ventral tarsus produce excess ventral-most structures (e.g. the peg-like bristles characteristic of wg-expressing cells on the tarsus, Fig. 5E). These clones often initiate outgrowths consisting only of mutant cells (Fig. 5E). This observation demonstrates that DPP signaling also plays a role in patterning the ventral territory of leg discs. Tarsal tkv clones also cause polarity disruptions in neighboring normal cells (e.g. reversed bristle-bract vectors; reversed bristles, Fig. 5E,F) that show a positional bias (Fig. 5G) with all ventral tkv clones associated with polarity defects (9 of 9) while less than half of the dorsal clones (4 of 10) show polarity disruptions. The polarity effects are not solely the result of extensive regulative growth since they occur even when the patterning response is negligible (Fig. 5F). Clones of tkv in the tarsal region, where WG and DPP signaling normally overlap, contain many marked cells while clones in the tibia, femur and more proximal segments or clones induced in first instar contain only one or two marked cells, suggesting that growth of cells in the tarsus is less sensitive to the loss of DPP signaling than other regions of the disc. The cuticle effects of puntts and tkv clones are consistent with the ectopic expression of wg affecting cell fate and with WG and DPP signals being antagonistic both in terms of domain of gene expression and in terms of effects on cell fate.

We find that the D/V asymmetry of wg and dpp expression in the anterior compartment of leg and antennal discs is main- tained by inhibition of wg expression by DPP signaling, and inhibition of dpp expression by WG signaling (Fig. 6B). Blocking response to DPP signaling in puntts mutants leads to expansion of wg expression, but only along the A/P boundary (Fig. 1C,E). Conversely, ventrally boosting DPP signaling inhibits wg expression in the leg (Fig. 2B). Similarly, blocking WG signaling with temperature sensitive wg mutants or dsh mutants, leads to expanded dpp expression, but again only along the A/P boundary (Fig. 3B,D), while ectopic expression of wg dorsally reduces dpp expression (Fig. 4C).

The restriction of wg and dpp expression to the region along the A/P boundary is governed by negative inputs. Repression by EN excludes wg and dpp from the posterior compartment (Sanicola et al., 1995). HH protein diffusing anteriorly across the A/P boundary allows wg and dpp to be expressed in domains that abut the A/P compartment boundary (Fig. 6B) (Basler and Struhl, 1994; Tabata and Kornberg, 1994). The positive effect of HH may be mediated by it antagonizing a repressive activity of Patched (PTC) or protein kinase A on dpp and/or wg expression in the anterior compartment (Phillips et al., 1990; Basler and Struhl, 1994; Capdevila et al., 1994; Jiang and Struhl, 1995; Johnson et al., 1995; Li et al., 1995).

The leg disc is specified in the embryo at a point were ventral wg and dorsal dpp-expressing cells abut (reviewed by Williams and Carroll, 1993). During the first and second instar, wg and dpp expression become largely overlapping and separate again during the third laval instar (Masucci et al., 1990; Couso et al., 1993). Our data suggest that the changes in the domains of dpp and wg expression during the third laval instar is due at least in part to a mutual repression that operates throughout devel- opment. The temperature shift experiments demonstrate that the mutual repression operates until late in larval development while the clonal analysis and dsh mutants suggest it operates early. It is unclear whether the initial D/V bias of dpp and wg seen in the embryo is partially maintained through first and second instar or whether other mechanisms are responsible for reinitiating the restriction of wg and dpp to the ventral and dorsal regions respectively. Nevertheless, once separate domains of expression are established, the mutually repressive interactions between WG and DPP signaling can then maintain the expression of these genes in different regions of the leg disc as growth and patterning proceed. This mutual repression can also provide a mechanism to regenerate patterning domains in the event of injury to the disc.

Based largely on studies of the wing imaginal disc, a model of pattern formation involving stepwise delineation of compart- ments has been proposed (Garcia-Bellido, 1975; Lawrence and Struhl, 1996). Compartments are defined by strict lineage restriction which once defined, do not change. Each compart- ment acquires a genetic address that is defined by the expression of selector genes (e.g. en, ci, ap) which once turned on or off become fixed in the founder cells and their descen- dants (Lawrence and Struhl, 1996). The demarcation of anterior and posterior compartments (Blair, 1993), follows directly from embryonic segmentation which generates adjacent stripes of ci and en expressing cells (Kornberg et al., 1985; Eaton and Kornberg, 1990) that are preserved as lineage restrictions (compartments) in both larval segments and discs (Garcia-Bellido et al., 1973; Garcia-Bellido, 1975; Steiner, 1976; Wieschaus and Gehring, 1976; Lawrence and Morata, 1977; Szabad et al., 1979; Lawrence and Struhl, 1982). While it has been possible to demonstrate a D/V compartment restric- tion in wing discs (Bryant, 1970; Garcia-Bellido et al., 1976; Blair, 1993; Diaz-Benjumea and Cohen, 1993; Lawrence and Morata, 1993; Williams et al., 1993; Blair et al., 1994), demon- stration of D/V compartments or proximal/distal compartments in other discs has remained elusive (Steiner, 1976). As an alter- native, we use the term ‘territory’ to describe a region of cells that are under the dominating influence of a particular morphogen. For example, the ventral cells in the leg disc that are responding to the predominant influence of WG, even though they are integrating high WG and low DPP input, con- stitute a ventral territory (Fig. 6B). Territories differ from com- partments in that they are not defined by lineage or by sharp boundaries of irreversibly committed selector gene expression (e.g. en, ci, ap). If, as a result of growth displacement, the descendants of territory-founding cells find themselves closer to a different territory-defining morphogen, they will acquire the properties of cells in that territory regardless of lineage origin. Thus, territory borders are less sharp than compartment boundaries and they are more dynamic. We propose that the anterior compartments of the leg and antennal discs are divided into dorsal and ventral territories by the mutual antagonism between WG and DPP signaling (Fig. 6B).

The effect of puntts on wg expression in the eye is consis-tent with the observation that WG inhibits the ability of DPP to propagate the morphogenetic furrow (Ma and Moses, 1995; Treisman and Rubin, 1995). The reduced eyes seen in puntts animals (Fig. 5C) suggest that the ectopic expression of wg that extends from the periphery into the eye field (Fig. 1E) restricts the domain of DPP influence in specifying eye tissue. Loss of eye tissue accompanied by ectopic wg expression is also seen in Mad clones which provide a downstream block to DPP signaling (Wiersdorff et al., 1996). Thus, the regulatory inter- actions between WG and DPP are similar in the eye and leg discs.

We find that the regulatory interactions between WG and DPP seen in leg and eye/antennal discs do not hold in dorsal discs (wing and haltere). In wing discs, the pattern of wg expression and the morphology of the discs is normal in pntts animals even with an extented temperature upshift of 30 hours past the time needed for maximal ectopic wg expression in legs (i.e. 40 hours, Fig. 1C). Similarly, the pattern of dpp expression and the morphology is normal in wgts animals upshifted for twice the time (49 hours) that is sufficient (24 hours) to achieve maximal ectopic dpp expression in leg discs (Fig. 3B). The dif- ference may lie in the different developmental histories of legs and wings. During arthropod evolution legs appeared first followed later by wings (Kukalova-Peck, 1978; Birket-Smith, 1984; Williams and Carroll, 1993; Williams et al., 1994). In Drosophila, a colony of cells migrates away from the leg disc anlage to become the wing disc anlage (Cohen et al., 1993) after the A/P compartments have been established (Steiner, 1976; Wieschaus and Gehring, 1976; Lawrence and Morata, 1977). Once separated, the wing and leg discs take distinct developmental paths. In wing discs, wg expression is absent during first instar but reappears in the second instar in a ventral patch of expression which overlaps dpp (Couso et al., 1993). A D/V compartment boundary is established by a series of genes whose functions in D/V compartment specification are unique to wing and haltere (e.g. vg, ap; Blair, 1993; Diaz- Benjumea and Cohen, 1993; Williams et al., 1993; Blair et al., 1994; Williams et al., 1994). Concurrently, the expression of wg and dpp change rapidly so that by mid third instar, dpp and hh expression are uniform along the D/V axis (Masucci et al., 1990; Lee et al., 1992; Tabata et al., 1992) while bands of wg expression run perpendicular to the stripe of dpp expression in several locations (Baker, 1988b; Couso et al., 1993). No candidate genes that might specify D/V compartments in the legs have emerged, suggesting different developmental strate- gies may be operative in wings and legs. The reason for expression of ectopic wg in clones of punt or tkv that fall near the distal crossover point of dpp and wg expression in the wing blade (Penton and Hoffmann, 1996) while temperature shift experiments do not alter wg expression, may be due to clones eliciting an intercalary regenerative response due to sharp posi- tional discontinuities while temperature shifts cause a general depression of the signaling response. These considerations and the failure of either WG or DPP to affect the other’s expression in wing discs, either when signaling is compromised or enhanced, support the view that the molecular basis for gener- ation of D/V asymmetry in legs and wings may be different.

In the gut, WG and DPP signaling positively affect each others expression in parasegments 7 and 8 (Staehling-Hampton and Hoffmann, 1994; Yu et al., 1996) and it has been suggested that WG promotes dpp expression in the germ band retracting embryo (Jackson and Hoffmann, 1994) providing two other examples where interactions between WG and DPP differ among tissues.

The patterning responses of adult cuticle elements to manipu- lations of wg or dpp are consistent with the hypothesis that wg specifies ventral positional values and dpp specifies dorsal positional values. Changes in shape and size of discs suggest that pattern regulation also occurs in response to the ectopic expression of either wg or dpp (Fig. 6A). Since dpp signaling inhibits wg expression, strong dpp expression along the A/P compartment boundary should replace the anterior ventral territory with the anterior dorsal one. The Polar Coordinate (French et al., 1976; Bryant et al., 1981) and Boundary models (Meinhardt, 1983) would predict a discontinuity in positional values at the ventral A/P compartment boundary followed by intercalation leading to an expansion of the posterior compart- ment. In situ hybridization to hh in discs with dppblk>dpp confirms that the posterior compartment has expanded, resulting in a pear shaped disc (Fig. 6A). In contrast, ectopic wg expression dorsally causes a mirror image duplication of the anterior ventral territory while repressing formation of the anterior dorsal territory producing an expansion of posterior and suppression of anterior values leading to a long thin disc (Fig. 6A). The pattern of en expression in dppblk>wg discs confirms the expansion of the posterior compartment (Fig. 4D).

Ectopic expression of wg in the dorsal region of the leg, even at high levels, failed to specify the extreme ventral fates that normally arise from the ventral wg-expressing cells (Struhl and Basler, 1993; Wilder and Perrimon, 1995). Thus, it is surpris- ing that ectopic expression of wg seen in puntts mutants is able to specify extreme ventral cell fates (Fig. 5A,B). The key dif- ference between these experiments may be that, in puntts animals, ectopic wg is accompanied by blocking DPP signaling by puntts (Letsou et al., 1995; Ruberte et al., 1995). This suggests that the potential of an ectopically expressed morphogen to affect cell fate may be fully realized only when competing antagonistic signals are removed (e.g. DPP in this case). We conclude that WG and DPP are mutually antago- nistic at the level of cell fate specification as well as tran- scription.

Previously, it was unclear whether the weak dpp expression, which overlaps wg in the ventral region of leg discs (Fig. 3A), had a functional role in patterning of the leg. The fact that ventral tkv clones in tarsi overproduce the ventral peg-like bristles (Fig. 5E) suggests that the weak dpp expression is func- tioning to antagonize and modulate WG signaling during normal patterning. The genetic insertion of these ‘hyper wg- like’ cells by tkv clones accounts for fatter legs and the out- growths that do not incorporate neighboring cells. The fact that ventral tkv clones also cause polarity disruptions could be explained if excess WG signaling reduced the amount of dsh available for establishing polarity. The effect of loss of tkv in the ventral region of leg discs where WG signaling is operative, suggests that cells may integrate input from both signaling pathways to determine cell fate. Such integration of competing signals could provide a mechanism for specification of inter- mediate cell fates.

In the leg, both punt and tkv clones located dorsally cause bifurcations (Penton and Hoffmann, 1996; our observations). However, ventral punt clones are reported to cause no abnor- malities (Penton and Hoffmann, 1996) while we see that ventral tkv clones cause excess ventral cells and bifurcations. One interpretation is that punt alleles may not be nulls. An alternative explanation is that ventral DPP signaling requires tkv but not punt. A recently isolated second BMP type II receptor that is expressed in all discs might provide an alter- native to signaling through punt in the ventral leg disc (Marques and O’Connor, unpublished observations).

A number of synergistic interactions between growth factors have been described (e.g. Kimelman et al., 1992; Rothbacher et al., 1995; Watabe et al., 1995) but the results reported here add antagonistic interactions to the repertoire of regulatory mechanisms available during patterning. Negative interactions may play an important role in integrating multiple positional cues during the specification of cell fate. The mutually negative effect of WG and DPP on each other’s expression and the antagonistic influence of each on cell fate choice adds a new dimension to the role these factors play in patterning of discs. It also raises the possibility of such negative feedback loops playing a general role during patterning in other systems such as vertebrate limb specification.

This work was supported by an NIH Research Program Project PO1 HD27173 to J. L. M.(Prog. Director, P.J.Bryant). H. T. was supported in part by a PHS training grant #5T32 GM07311-17. M. B. O. is supported by a NIH #GM00599 and GM 47462 and T. E. H. by a CRCC grant from the University of California. The authors gratefully acknowledge the resources of the National Drosophila Stock Center in Bloomington, IN and appreciate the stocks received from I. Livne- Bar, H. Kraus, A. Letsou, R.Nichols, W.M.Gelbart, and D. Brower.

A recent communication (Jiang, J. and Struhl, G. Cell86, 401; 1996) examines the mutually antagonistic effects of WG and DPP at the level of protein accumulation reaching similar conclusions to our studies examining mRNA levels. Thus, the conclusion that regulation of transcript levels is the major mechanism of regulation is supported.