wingless inhibits morphogenetic furrow movement in the Drosophila eye disc

The ordered array of ommatidia in the Drosophila compound eye results from a progressive pattern of differentiation in the eye imaginal disc (for review see Thomas and Zipursky, 1994). The eye disc is an epithelial monolayer derived from a primordial group of cells determined during embryogenesis. During the third larval instar, rows of photoreceptor clusters differentiate in succession, starting at the posterior margin and continuing anteriorly. The first morphological marker of differentiation is the morphogenetic furrow, an indentation formed by contraction of the cells in the apical/basal dimension and accompanied by constriction of their apical profiles (Ready et al., 1976). This furrow moves across the disc from posterior to anterior as development proceeds. Immediately anterior to the furrow, cells are arrested in the G₁ phase of the cell cycle (Ready et al., 1976). Within the furrow clusters of cells form in a regularly spaced array (Wolff and Ready, 1991) and, posterior to the furrow, these cells differentiate as photoreceptors in a defined sequence (Tomlinson and Ready, 1987a).

The successive addition of rows of ommatidia has been shown to require expression of hedgehog (hh) in the differentiated photoreceptors (Heberlein et al., 1993; Ma et al., 1993); HH is a secreted protein which is thought to constitute an inductive signal (Lee et al., 1992). Its effects may be mediated by decapentaplegic (dpp), a member of the TGF-β family, which shows hh-dependent expression within the morphogenetic furrow (Heberlein et al., 1993; Ma et al., 1993). Ectopic expression of hh in patches of cells anterior to the furrow is sufficient to induce differentiation which appears to propagate radially (Heberlein et al., 1995). dpp is one of the first molecules to be induced by hh, and hh-independent induction of ectopic dpp expression, caused by loss of protein kinase A function, has identical effects (Pan and Rubin, 1995; Strutt et al., 1995).

Initiation of furrow movement must occur by a different mechanism than its propagation, as hh is not present prior to differentiation. Since dpp is expressed around the edges of the eye disc before initiation (Masucci et al., 1990), it may be involved in this process. The eyes absent (eya; Bonini et al., 1993), sine oculis (so; Cheyette et al., 1994) and dachsund (dac; Mardon et al., 1994) genes are also likely to act in initiation (Thomas and Zipursky, 1994). All of these genes can mutate to eyeless phenotypes, are expressed in a similar early pattern at the margins of the disc and encode nuclear proteins. The eyeless (ey) gene is presumably upstream of all these genes, since it is sufficient to initiate eye differentiation in other imaginal discs (Halder et al., 1995).

The wingless (wg) gene encodes another secreted protein that is required for segmentation of the embryo and development of the imaginal discs (Rijsewik et al., 1987; Baker, 1988a). It is a member of the Wnt gene family (Nusse and Varmus, 1992); vertebrate Wnt genes are involved in axis specification (McMahon and Moon, 1989), determination of the midbrain (Thomas and Capecchi, 1990; McMahon and Bradley, 1990) and limb development (Parr and McMahon, 1995). The WNT proteins appear to bind to the extracellular matrix and to act over a short range (Bradley and Brown, 1990; Papkoff and Schryver, 1990; van den Heuvel et al., 1993). Their receptors have not been characterized, although it has been suggested that Notch might act as a WG receptor (Couso and Martinez-Arias, 1994). Genetic methods have been used to identify a number of other molecules acting in the wg pathway (reviewed by Klingensmith and Nusse, 1994; Siegfried and Perrimon, 1994); these include shaggy/zeste-white3 (sgg/zw3; Bourois et al., 1990; Siegfried et al., 1990), a protein kinase that is negatively regulated by wg (Siegfried et al., 1992).

The functions of wg in imaginal disc development have recently been characterized. In the leg imaginal disc, wg is expressed in a stripe along the ventral half of the anteriorposterior compartment boundary (Baker, 1988b). At the point where it abuts a stripe of dpp along the dorsal half of the A-P boundary, the homeobox genes distalless and aristaless are induced and direct distal outgrowth of the leg (Campbell et al., 1993; Diaz-Benjumea et al., 1994). Ectopic expression of wg promotes ventrolateral cell fates (Struhl and Basler, 1993) and can induce a secondary proximal-distal axis if it is present near the region of dpp expression (Struhl and Basler, 1993; Campbell et al., 1993). Even high levels of wg expression do not allow cells to adopt the ventralmost fates, although these can be induced by the absence of sgg/zw3 (Diaz-Benjumea and Cohen, 1994; Wilder and Perrimon, 1995). These results have cast doubt on the theory that WG acts as a graded morphogen and suggest that additional components may contribute to determining ventral cell fates. In the wing disc, wg has an early role in specifying the ventral compartment and indirectly determining the entire wing (Morata and Lawrence, 1977; Couso et al., 1993; Williams et al., 1993). Later, it acts through the transcription factors achaete and cut to form the margin of the wing (Phillips and Whittle, 1993; Couso et al., 1994). In the eye disc, wg is expressed at the dorsal and ventral margins in the regions that will form head cuticle (Baker, 1988b). We show here that it acts to prevent dpp present in these regions from initiating a wave of photoreceptor development. We also show that ectopic wg can inhibit the propagation of normal photoreceptor development. Thus wg and dpp interact to define the region in which the morphogenetic furrow can initiate.

The wg alleles used were wg^IL114 (Nusslein-Volhard et al., 1984), with secondary lethals removed (Couso et al., 1994), and wg¹ (Sharma and Chopra, 1976). Other alleles used were dpp^d-blk (Masucci et al., 1990), hh¹ (Heberlein et al., 1993) and sgg^D127 (Bourois et al., 1990). Double mutants were made by standard genetic methods. Reporter genes used were the enhancer traps wg^P (Kassis et al., 1992) and P30 (Lee et al., 1992), and the dpp-lacZ construct BS3.0 (Blackman et al., 1991). The Act5c>y⁺>wg lines 411.28 and 411.33 (Struhl and Basler, 1993) were combined with hs-FLP (Xu and Rubin, 1993) to induce flp-out clones expressing wg. Larvae were heat shocked for 1 hour at 38°C in both the first and second instar. Clones of sgg mutant cells were induced by the same heat-shock regimen, using the 18.2πM stock (Xu and Rubin, 1993). Myc expression was induced by a 1 hour heat shock at 38°C and larvae were allowed to recover for 30 minutes at 25°C before dissection. The dpp-GAL4 line used was 40.C6 (Staehling-Hampton et al., 1994) and this was crossed to UAS-wg^ts M7-2-1 (Wilder and Perrimon, 1995).

Flies were prepared for scanning electron microscopy as described by Kimmel et al. (1990). Adult eyes were fixed, embedded and sectioned as described by Tomlinson and Ready (1987b). Eye imaginal discs were stained with antibodies as described by Tomlinson and Ready (1987a), except that the detergent used was 0.2% Triton. Rat anti-Elav monoclonal antibody was diluted 1:1; mouse anti-Glass monoclonal antibody (Ellis et al., 1993) was diluted 1:5. Mouse myc ascites was diluted 1:100. For double labeling with antibody and X-gal, the antibody staining was performed first and followed by a wash in PBS and incubation in X-gal staining buffer.

The spatially restricted pattern of expression of the wingless (wg) gene in the eye disc suggested that wg might function in pattern formation in the eye. wg RNA is expressed at the dorsal and ventral edges of the eye disc, in regions fated to become head cuticle (Baker, 1988b; Ma and Moses, 1995; Fig. 3E, 6E). We have used a temperature-sensitive wg mutation, wg^IL114 (Nusslein-Volhard et al., 1984), to characterize the role of wg at this stage of development, and have found that wg acts to restrict the initiation of photoreceptor differentiation to the posterior margin of the eye disc.

When wg function was inactivated by a shift to the restrictive temperature (25°C), photoreceptor differentiation progressed medially from the dorsal and ventral margins as well as anteriorly from the posterior margin (Fig. 1). Hedgehog (hh) is expressed in differentiated photoreceptors and is thus normally only present posterior to the morphogenetic furrow (Ma et al., 1993; Fig. 1A) but, within 24 hours of a temperature shift, it could also be seen at the dorsal and ventral edges of the eye disc (Fig. 1C). When larvae were shifted at earlier times in development, hh expression moved progressively inward, accompanied by a decrease in the size of the third instar disc (Fig. 1E,G). The expression pattern of decapentaplegic (dpp) revealed that the morphogenetic furrow curved anteriorly at the edges in the absence of wg (Fig. 1D), becoming semicircular 48 hours after a temperature shift (Fig. 1F) and disappearing by 72 hours after a shift (Fig. 1H). Using an antibody to the neuronal protein Elav (Robinow and White, 1991) to reveal the stage of differentiation of the photoreceptor clusters, it was clear that, in the regions of ectopic differentiation, the least differentiated clusters were closest to the center of the disc and to the region of dpp expression, confirming that differentiation was progressing in a lateral-tomedial direction (Fig. 1I).

There always appeared to be more ectopic furrow movement on the dorsal side of the disc than the ventral side, correlating with the stronger dorsal than ventral expression of wg. However, we have observed ectopic furrow movement initiating predominantly from the ventral margin in the wg¹ mutant (data not shown), a semiviable allele that causes a deletion of the ventralmost part of the eye (Morata and Lawrence, 1977) in addition to transforming the wing into notal tissue (Sharma and Chopra, 1976; Couso et al., 1993). Thus wg does appear to be required at both dorsal and ventral margins.

These results show that in the absence of wg the lateral margins are competent to initiate a morphogenetic furrow and wg is required to prevent such a furrow from forming. Before furrow initiation, dpp is expressed at the dorsal and ventral margins of the eye disc as well as at the posterior margin (Masucci et al., 1990; Fig. 6B). The function of wg may be to inhibit its activity at the lateral margins. wg itself, as judged by the expression of an enhancer trap inserted in the gene that has been shown to reproduce the pattern of wg RNA (Kassis et al., 1992; Couso et al., 1993), is never present at the posterior margin even at early stages (Fig. 6D). To determine whether hh and dpp were required for the ectopic furrow movement caused by the loss of wg, we combined wg^IL114 with the eye-specific mutations hh¹ and dpp^d-blk. hh¹ arrests furrow movement after differentiation of 8-10 rows of photoreceptors (Heberlein et al., 1993; Fig. 2B, 3C), while dpp^d-blk allows furrow movement only in the central region of the eye disc (Masucci et al., 1990; Fig. 2C, 3A). The hh¹ mutation completely abolished the effects of loss of wg function (Fig. 2E). In the wg^IL114, dpp^d-blk combination, initiation occurred over a wider region than in dpp^d-blk alone, but the extensive inward movement seen in wg^IL114 alone (Fig. 2D) did not occur (Fig. 2F). Thus the ectopic movement resembles normal furrow movement in its requirement for functional hh and dpp.

In the dpp^d-blk mutant, furrow initiation appears to be inhibited in part of its normal domain (Fig. 2C). This inhibition is alleviated by loss of wg function (Fig. 2F), suggesting that wg is contributing to the repression. Indeed, even the loss of one copy of wg is sufficient to increase the size of the dpp^d-blk eye (compare Fig. 3A and B). However, wg expression does not expand into the inhibited regions of the margin (Fig. 3F), so its effect on furrow movement is felt at a distance. This might reflect either diffusion of the WG protein or its indirect action through other factors. It is not clear whether this effect of wg, like its effect on the lateral margins of wild-type eye discs, could be achieved by blocking dpp function. Although dpp RNA is not detectable in third instar dpp^d-blk eye discs (Masucci et al., 1990), several observations lead us to believe that there is some remaining weak or early expression. Firstly, large clones of cells mutant for dpp fail to differentiate (Heberlein et al., 1993), and yet there is differentiation in a region of the dpp^d-blk eye. Secondly, if dpp^d-blk mutants did not express dpp in the eye disc, we would expect them to be unaffected by the hh¹ mutation, which causes the loss of dpp expression and thus arrests the furrow (Heberlein et al., 1993; Fig. 3C). However, in doubly mutant hh¹; dpp^d-blk eye discs, there is also an arrest of furrow movement within the central region (data not shown), reducing the size of the adult eye (Fig. 3D). Finally, the enhancer region deleted in the dpp^d-blk mutant (coordinates 106–110, St. Johnston et al., 1990) does not include all the regulatory sequences active in the eye disc, as it is not sufficient to direct expression in the full dpp pattern when fused to a heterologous reporter (see Fig. 5E, F). It is therefore likely that the loss of wg suppresses the dpp^d-blk phenotype by allowing a small amount of dpp to function more effectively.

The effects of loss of normal wg function showed that wg is necessary to inhibit abnormal furrow movement. We next wished to determine whether wg was sufficient to inhibit normal furrow movement. We did this by ectopically expressing wg, using both the flp-out (Struhl and Basler, 1993) and the GAL4 (Brand and Perrimon, 1993) systems. In the adult eye, induction of wg expression in random clones of cells using the flp-out system led to scar formation; scars always ran in the anterior-posterior direction and usually extended to the anterior margin of the eye (Fig. 4A). These scars were associated with bristle loss in the surrounding ommatidia, an effect that has been shown to be produced by expression of wg under the control of the sevenless promoter (R. Nusse, personal communication); it is therefore likely that the scars do result from the presence of wg-expressing cells. Sections revealed that the scars were devoid of ommatidia but appeared to contain pigment cells. In addition, the dorsal-ventral polarity was reversed in the row of ommatidia next to the scar on the side further from the equator (Fig. 4B). These scars appeared to result from regions in the eye disc where development was delayed or inhibited. In these regions, expression of neuronal markers and of dpp and hh was delayed relative to the surrounding tissue and there appeared to be overproliferation (Fig. 4C-E and data not shown). The extent of the adult scars suggests that once blocked, the furrow does not reinitiate beyond the position of the block. We assume that these blocked regions result from the presence of wg-expressing cells, as they are never observed in wild-type eye discs. However, since the clones of wg-expressing cells are not marked, we cannot evaluate the range of wg action. No activation of the wg enhancer trap was associated with the blocks, suggesting that the endogenous wg gene is not required (Fig. 4E).

Similar blocks to furrow progression were caused by loss of shaggy (sgg) function in clones of cells (Fig. 4F). Since sgg is inhibited by wg (Siegfried et al., 1992), sgg⁻ cells behave as though they have received the wg signal. Almost all such mutant cells failed to differentiate (Fig. 4G,H) and were not present in the adult eye (data not shown). In addition, wild-type cells anterior to the sgg⁻ cells did not express neuronal markers (Fig. 4H), confirming that the furrow is unable to reinitiate beyond a block caused by the wg pathway.

Occasionally, regions in the disc where the furrow had failed to initiate were observed after induction of clones of wgexpressing cells or sgg mutant cells (Fig. 5A,B). This resulted in lack of expression of the Elav and Glass proteins. However, dpp-lacZ continued to be expressed at a high level (Fig. 5A). We hypothesized that such blocks resulted from the presence of wg pathway function at the posterior margin, which would interfere with initiation. To confirm this, we also misexpressed wg specifically at the posterior margin of the eye disc, using a disc-specific enhancer from the dpp gene to drive expression of GAL4 at the margins of the disc (Staehling-Hampton et al., 1994) and a GAL4-responsive UAS sequence to drive a temperature-sensitive form of wg (Wilder and Perrimon, 1995). When shifted to the permissive temperature (18°C), ectopic wg appears to delay normal furrow initiation. The central region of the disc is least affected, so that initiation occurs in the center ahead of more lateral regions, resembling the phenotype of dpp^d-blk. However, the expression of dpp-lacZ is not affected (Fig. 5D), suggesting that wg interferes with the function of dpp rather than with its expression.

The eye does eventually develop almost normally in flies carrying the dpp-GAL4/UAS-wg^ts combination (data not shown), so the block imposed by wg is not permanent in this case. The temperature-sensitive form of WG may be less active, even at the permissive temperature, than wild-type WG present at the lateral margins. It has been shown to be less efficiently secreted than the wild-type protein (van den Heuvel et al., 1993), although it is sufficiently active to direct normal embryonic development (Baker, 1988a). There is also a difference between the expression pattern of the dpp-GAL4 construct, which contains the enhancer region deleted in the dpp^d-blk mutant, and that of dpp itself. When crossed to UASlacZ, this dpp-GAL4 construct did induce early lacZ expression at the posterior margin as well as at the lateral margins, reflecting the normal pattern of dpp expression at this stage, but this expression remained at the posterior margin rather than progressing anteriorly with the furrow (Fig. 5E,F). However, when used in conjunction with UASdpp, the same construct is sufficient to rescue the dpp^d-blk phenotype (Staehling-Hampton et al., 1994). Despite these caveats, we can conclude that the presence of wg at the posterior margin has at least a mild inhibitory effect on furrow initiation.

We have shown that wg acts at the dorsal and ventral edges of the eye disc to prevent these marginal regions from initiating neuronal differentiation. This ensures that differentiation initiates only at the posterior margin, in spite of the early expression of dpp around the entire margin. Ectopic wg can also inhibit both initiation from the posterior margin and progression in the center of the disc. In these instances, the expression of dpp is not reduced, although it fails to advance normally.

The results in combination suggest a model dividing the margin of the eye disc into three zones (Fig. 6). In the posterior central zone, surrounding the optic stalk, initiation is robustly driven by an unidentified signal, which does not require hh, can overcome the presence of wg and functions even with greatly reduced levels of dpp. This mechanism accounts for the differentiation seen in dpp^d-blk and in discs in which wg is ectopically expressed at the posterior margin. Initiation only from this zone is also seen in weak alleles of dac (Mardon et al., 1994) or eya (Bonini et al., 1993); these genes might control the expression of wg or act downstream to block its function. There does appear to be some ectopic wg expression at the margin of dac mutant discs (data not shown). In the posterior zones lateral to this central zone, the furrow initiates at a time when very little hh is present, and initiation still occurs in the hh¹ mutant but not in dpp^d-blk. Thus initiation in this region requires dpp but probably does not require hh. Ectopic wg has an inhibitory effect on initiation from this zone. There may not be a sharp boundary between the posterior central and posterior lateral zones; an alternative would be a gradient of a molecule conferring initiation capability (X in Fig. 6), perhaps derived from a source at the optic stalk. Finally, in the lateral zones, which normally express wg, initiation requires the presence of both dpp and hh and the absence of wg. The lateral zone is distinguished from the posterior-lateral zone by its lack of any initiation in the absence of both wg and hh, as seen in the wg^ts; hh¹ double mutant.

Although wg prevents initiation from the lateral margins (Fig. 1), and wg expressed in the central region can block furrow progression (Fig. 4), the furrow is able to progress adjacent to these wg-expressing regions. The most likely explanation of this is that hh allows the furrow to move past wgexpressing cells but not through them. As WG protein is thought to act over a short range, it may produce only a narrow zone of cells unable to respond to hh. Another possibility is that hh may act indirectly on the transcription of wg to inhibit its expression at the edges. The wg-expressing clones that we generated using the flp-out system express wg under the control of the Actin5C promoter (Struhl and Basler, 1993), which would presumably not respond to the hh signal and would therefore result in a more permanent block to differentiation.

The effect of wg on furrow initiation could be described as a decision between head cuticle fate and eye fate; regions determined by wg to become cuticle would be unable to respond to signals directing eye morphogenesis. Alternatively, the primary function of wg could be to prevent cells from responding to the dpp signal, with cuticle formation being the default fate. The effect of clones of ectopic wg seems to support the latter model. Centrally located cells expressing wg fail to respond to dpp and the furrow is unable to progress through them. However, their later fate appears to be the formation of pigment cells rather than cuticle; pigment cells may be the default state of cells in the central region of the eye disc that do not become photoreceptors or cone cells, since these cells are determined synchronously at a stage after passage of the morphogenetic furrow is complete (Cagan and Ready, 1989).

wg may influence eye development in ways other than directly affecting differentiation. Discs in which wg has been inactivated reach a smaller final size than wild-type discs. This could be due to stimulation of cell division by wg, an effect that has been observed in other tissues (Skaer and MartinezArias, 1992; Wilder and Perrimon, 1995). Induction of ectopic wg in clones produces outgrowths of disc tissue (Fig. 4C), supporting this hypothesis. Alternatively, the earlier differentiation of cells anterior to the furrow in wg^ts discs may simply leave insufficient time for the normal number of cell divisions. However, when differentiation anterior to the furrow is induced by ectopic hh expression, it is accompanied by disc overgrowth (Heberlein et al., 1995). The reverse effect seen when such differentiation is induced by loss of wg suggests that wg may have an active role in directing proliferation.

wg may also influence dorsoventral polarity, which is manifested in the eye as the rotation of ommatidia in opposite directions on the dorsal and ventral sides of the equator (Ready et al., 1976). The direction of ommatidial rotation is reversed in ommatidia adjacent to a scar caused by ectopic wg expression (Fig. 4B). Possibly wg expressed at the edges of the disc is one of the signals directing ommatidial rotation. This might explain the polarity defects seen in dishevelled (dsh) mutant eyes (Theisen et al., 1994), since dsh has been shown to act downstream of wg (Noordemeer et al., 1994; Siegfried et al., 1994).

Although dpp and wg both affect the development of many tissues in the fly, the relationship between them is not invariant. During embryogenesis, wg acts as a segment polarity gene, determining cell identity along the anterior-posterior axis of each segment (Baker, 1988a); dpp acts dorsally to determine the perpendicular dorsoventral axis (Ferguson and Anderson, 1992). In wing development, wg and dpp again direct development along mutually perpendicular axes, but wg establishes the dorsoventral and dpp the anterior-posterior axis (Williams et al., 1993; Couso et al., 1993, 1994; Basler and Struhl, 1994). Gut development requires dpp and wg to act on the common target labial (Immergluck et al., 1990). During leg development, specification and outgrowth are both induced by the presence of cells expressing wg adjacent to cells expressing dpp (Cohen, 1990; Campbell et al., 1993; Basler and Struhl, 1994). In the optic lobes of the brain, wg activates expression of dpp in an adjacent domain (Kaphingst and Kunes, 1994). Finally, we show here that, in the eye disc, wg interferes with the function of dpp but not with its expression. In vertebrates, preliminary indications also favor a variety of relationships between members of the Wnt and TGF-β gene families (Nusse and Varmus, 1992; Kingsley, 1994). These signaling molecules may have been co-opted for many different uses, with their relationship to each other remaining flexible.

We thank Alfonso Martinez-Arias, Phil Beachy, Ron Blackman, Ulrike Heberlein, Gary Struhl, Karen Staehling-Hampton, Elizabeth Wilder, Esther Siegfried, Roel Nusse and Barry Dickson for fly stocks and reagents. We are grateful to Ulrike Heberlein, Françoise Chanut, Bruce Hay, Tanya Wolff and Alfonso Martinez-Arias for helpful discussions, and to Kevin Moses for communicating results prior to publication. The manuscript was improved by the critical comments of Ulrike Heberlein, Henry Chang and Tanya Wolff. J. E. T. was supported by a Jane Coffin Childs Memorial Fund postdoctoral fellowship. G. M. R. is a HHMI investigator. This work was supported in part by NIH grant GM33135 to G. M. R.