Compartmental organization of the Drosophila genital imaginal discs

Imaginal discs are sac-like clusters of primordial cells in the larvae that are set aside during embryogenesis, proliferate during larval stages, and give rise to the adult cuticular structures during metamorphosis (reviewed by Cohen, 1993). Genetic studies of the thoracic wing and leg discs have revealed that the discs are divided into anterior and posterior (A/P) compartments (Garcia-Bellido et al., 1973). Each compartment represents a functionally distinct developmental unit. Cells in one compartment do not mix with those of the other. The compartment boundary, therefore, serves as a line of cell lineage restriction. Interactions between anterior and posterior cells which occur at the compartment boundary are believed to organize and pattern the entire disc (Meinhardt, 1983).

The genital imaginal disc gives rise to the adult terminalia, including the genitalia and the analia. Most imaginal discs are essentially two-dimensional, with folds in the plane of the disc epithelium (reviewed by Fristrom and Fristrom, 1993). In contrast, the genital disc epithelium is folded into a three-dimensional structure with distinct dorsal and ventral epithelia, which give rise to different adult structures, as shown on the threedimensional fate maps constructed by Ehrensperger and Epper (Ehrensperger, 1972; Epper, 1980, 1983) (Fig. 1a-c,e-g). The derivatives of the female genital disc include the female genitalia, the 8th tergite, the hind gut, and the dorsal and ventral anal plates (Fig. 1d), while the derivatives of the male genital disc include the male genitalia, the hind gut, and the left and right anal plates (Fig. 1h) (reviewed by Bryant, 1978; Lauge, 1982).

In addition to its three-dimensional structure, the genital disc has several unique properties which set it apart from the other imaginal discs (reviewed by Bryant, 1978; Lauge, 1982). First, it is the only single, unpaired imaginal disc. Second, it is the only imaginal disc which shows strong sexual dimorphism. This dimorphism, manifested most clearly in the adult structures, becomes evident in histological sections before the second larval molt and is quite pronounced in the third instar discs. Finally, the embryonic origin of the genital disc is different from that of the other discs. While the thoracic wing and leg discs are each derived from a single embryonic segment (Bate and Martinez Arias, 1991), the three primordia (the female genital, the male genital and the anal primordia) that make up the genital disc originate from within three embryonic tail segments, abdominal segments 8, 9, and 10, respectively (Dübendorfer and Nöthiger, 1982; Epper, 1980; Epper and Nöthiger, 1982; Nöthiger et al., 1977; Schüpbach et al., 1978). Cells from these three primordia are presumed to fuse together during late embryogenesis and form a transversely elongated cluster of genital disc precursor cells between the A8 denticle belt and the larval anal pads, which will later invaginate to develop into either a female or a male genital disc (Hartenstein and Jan, 1992). In a female larva, the female genital primordium will be activated, the male primordium will be repressed, and the anal primordium will adopt the female fate. Thus, the mature third instar female genital disc is composed of a well developed female genital primordium, a repressed male genital primordium, and a female anal primordium. The opposite is true for the mature male genital disc (Belote and Baker, 1982; Epper, 1980; Epper and Bryant, 1983; Epper and Nöthiger, 1982; Nöthiger et al., 1977; Schüpbach et al., 1978; Wieschaus and Nöthiger, 1982).

Although there is a clear compartmental boundary between the genital and anal primordia, the A/P compartmental organization of the genital disc has been unclear. Two models have been proposed to explain how the genital disc is divided into anterior and posterior compartments. Based on the phenotypes of the engrailed (en) mutant clones in the adult terminalia, Lawrence and Struhl (1982) suggested that the female genital primordium is of anterior origin, whereas the male genital and anal primordia are of posterior origin. Based on the phenotypes of heteroallelic combinations of en mutations, Epper and Sánchez (1983) suggested that all three terminal primordia have anterior and posterior compartments, and that the female and male genital primordia are mostly made of posterior compartment, with a very small anterior compartment.

Recent work on the thoracic wing disc has provided substantial information about the genes involved in the A/P compartmental organization. Genes that are involved in the A/P patterning of the wing disc include: (1) engrailed (en), which encodes a homeobox-containing transcription factor expressed in the posterior compartment (Kornberg et al., 1985); (2) hedgehog (hh), whose protein product is a secreted molecule also expressed in the posterior compartment (Lee et al., 1992; Tabata et al., 1992); (3) decapentaplegic (dpp), a member of the transforming growth factor β superfamily of signaling molecules (reviewed by Kingsley, 1994), which is expressed at the anterior side along the A/P compartment border, for several cell widths (these cells will be referred to as border cells hereafter) (Raftery et al., 1991); (4) cubitus interruptus (ci), which encodes a Zn-finger containing protein, expressed in the anterior compartment (Eaton and Kornberg, 1990); and (5) patched (ptc), which encodes a putative transmembrane protein (Hooper and Scott, 1989; Nakano et al., 1989), also expressed in the anterior compartment, but is expressed most predominantly in the border cells (Phillips et al., 1990).

Molecular and genetic studies have also revealed a regulatory hierarchy that controls the A/P patterning of the wing disc. In the anterior compartment, PTC represses the expression of dpp and ptc itself (Capdevila et al., 1994; Tabata and Kornberg, 1994). HH is secreted from the posterior compartment into the border cells and antagonizes the inhibitory activity of PTC on the expression of dpp and ptc, which results in the expression of dpp and higher expression level of ptc in the border cells (Lee et al., 1992; Tabata and Kornberg, 1994). Recently, the cAMP-dependent protein kinase A catalytic subunit gene (DC0, referred to as Pka in this paper) has been shown to be required for patterning the anterior compartment of the thoracic discs (Jiang and Struhl, 1995; Lepage et al., 1995; Li et al., 1995; Pan and Rubin, 1995). Pka mutant clones in the anterior compartment cause ectopic expression of dpp and ptc and result in adult pattern duplications and/or local overgrowth. However, Pka mutant clones in the posterior compartment do not show any abnormalities. In the posterior compartment, en, the posterior selector gene, activates the expression of hh, and represses the expression of ci, ptc and dpp (Eaton and Kornberg, 1990; Sanicola et al., 1995; Schwartz et al., 1995; Tabata et al., 1995; Zecca et al., 1995). It has been recently shown that invected (inv), a gene that contains a homeodomain almost identical to that of en (Coleman et al., 1987), contributes a non-essential and partially redundant function to en (Tabata et al., 1995). en-inv mutant clones in the posterior compartment cause loss of expression of hh, and ectopic expression of ci, dpp, and ptc, which result in posterior to anterior transformations in adult flies. However, en-inv mutant clones in the anterior compartment are normal. Thus, Pka has an anterior compartment-specific function in the wing disc, while en-inv have a posterior compartment-specific function.

In the leg disc, the relationship between domains of en, hh, ptc, and ci expression is analogous to those in the wing disc (reviewed by Perrimon, 1995). However, dpp is predominantly expressed in the dorsal half of the border cells, whereas wingless (wg), a member of the Wnt family of secreted molecules (reviewed by Nusse and Varmus, 1992), is expressed in the ventral half of the border cells (Williams et al., 1993). Correspondingly, Pka mutant clones cause ectopic expression of dpp in the dorsal anterior region, and ectopic expression of wg in the ventral anterior region of the leg disc (Jiang and Struhl, 1995; Lepage et al., 1995; Li et al., 1995).

As part of a long term study to elucidate imaginal disc development, and to understand how information from the sex determination hierarchy is integrated with input from other hierarchies to control developmental fates, we have undertaken molecular and genetic analyses to resolve the controversy over the A/P compartmental organization of the genital disc. We used genes which have A/P compartment-specific expression patterns and functions in wing and leg discs as molecular markers to localize the anterior and posterior domains in the genital disc. We show that the relationship between domains of en, hh, ptc, dpp, wg, and ci expression in the genital disc is analogous to those in the wing and leg discs, which suggests that the genital disc is also divided into anterior and posterior compartments. Moreover, mitotic clones induced at the beginning of the second instar in the genital disc do not cross the boundary between the en- expressing and ci-expressing cells, demonstrating that these two cell populations belong to two true genetic compartments. We also find that the spatial expression patterns of these genes are restricted to subsets of the cells that make up each of the three primordia of the genital disc, suggesting that each of the three primordia is composed of anterior and posterior compartments. Moreover, clonal analyses with Pka and en-inv mutants, which are known to play compartment-specific roles in the A/P patterning of other discs, show that the functions of these genes are conserved in the genital disc. The adult clonal phenotypes of Pka and en-inv mutants also provide a more detailed map of A/P compartmental divisions within the adult genitalia and analia. Based on these results we propose a new model for the A/P compartmental organization of the genital disc.

The enhancer trap and promoter lacZ lines used for antibody staining were as follows: Xho25 (en;Hama et al., 1990), P30 (hh;Lee et al., 1992), AT90 ptc (ptc; kindly provided by C. Goodman), BS3.0 (dpp;Blackman et al., 1991), P999 (wg; kindly provided by T. Tabata and T. Kornberg), 7.1-ci (ci;Schwartz et al., 1995).

Stocks used for Pka clonal analysis (Pan and Rubin, 1995) were: (1) y,w; Pka^B3.P[ry^+.; hs-neo; FRT]40A; BS3.0/T(2:3) SM6; TM6B,Tb. (2) y,w; Pka^B3.P999 P[ry^+.; hs-neo; FRT]40A/CyO. (3) y,w; Pka^B3.P[ry^+.; hs-neo; FRT]40A AT90 ptc/T(2:3) SM6; TM6B,Tb. The control stock used was y,w; P[ry^+.; hs-neo; FRT]40A; ry.

Stocks used for en-inv clonal analysis were: (1) y,w; P[ry^+.; hs-neo; FRT]42D Dfen^E./CyO. (2) y,w; BS 3.0 P[ry^+.; hs-neo; FRT]42D Dfen^E./CyO. (3) y,w; P[ry^+.; hs-neo; FRT]42D Dfen^E.; P30/T(2:3) SM6; TM6B,Tb. The control stock used was y,w; P[ry^+.; hs-neo; FRT]42D; ry.

For making mitotic clones and marking the posterior compartment of the genital disc, P[ry^+.; hs-neo; FRT]40A; P30/T(2:3) SM6; TM6B,Tb was used.

The following lines were used to introduce the flipase into the heterozygous flies and mark the clones in adults or discs: (1) y, w, hsFLP1; P[ry+; y+]25F, P[ry^+.; hs-neo; FRT]40A. (2)w, hsFLP1; P[mini-w+; hs-NM]31E, P[ry^+.; hs-neo; FRT]40A. (3) y, w, hsFLP1; P[ry^+.; hs-neo; FRT]42D, P[ry+; y+]44B.

Two types of wg mutants were used to examine the terminalia phenotypes (kindly provided by K. Cadigan and R. Nusse). One was a heteroallelic combination of wg^CX3/wg^CX4. The other was a wg temperature sensitive mutant, wg^IL.

Mitotic clones of wild type, Pka mutant, and en-inv mutant alleles were generated by the FLP/FRT system as described by Golic (1991) and Xu and Rubin (1993). FLP-mediated recombination was induced by heat shocking second instar larvae at 38oC for one hour. Adult clones were identified by the y marker. Clones in genital discs were identified by using either antibodies against the ubiquitously expressed NOTCH-MYC epitope (for wild-type and Pka clones), or antibodies against EN protein (for en-inv clones).

Double and triple immunofluorescence labeling of genital discs were done following the protocol of Xu and Rubin (1993). The primary antibodies used were: rabbit anti-β-galactosidase (β-gal) (Cappel), mouse anti-EN (Patel et al., 1989), rat anti-WG (gift from K. Cadigan and R. Nusse), mouse anti-MYC (gift from S. Carroll), rat anti-CI (gift from R. Holmgren), rabbit anti-DPP (gift from M. Hoffman), rabbit anti-HH (gift from T. Tabata), mouse anti-PTC (gift from I. Guerrero). The secondary antibodies used were from Jackson Immunologicals, including goat FITC anti-rabbit, goat Cy5 anti-rabbit, goat FITC anti-rat, goat Cy3 antimouse, goat rhodamine anti-mouse, goat Cy3 anti-rat.

Adult abdomens were dissected, boiled in 10% NaOH for 10 minutes, and washed in water four times. The cuticles of the abdomens were then washed in isopropanol 3-4 times and mounted in Gary Struhl’s Magic Mount for inspection under a compound microscope.

The adult terminalia were dissected and fixed in 4% glutaraldehyde for 20 minutes. The internal genitalia were then scored using a dissecting microscope.

Confocal images were collected using a Bio-Rad MRC-1024 system. Combined confocal images were made using Adobe Photoshop software.

To determine whether the six known A/P patterning genes function in the genital disc as they do in the thoracic discs, we examined the relative domains of expression of these genes by performing double and triple fluorescent labeling experiments. Since both female and male genital discs are bilaterally symmetrical, the expression patterns of these genes are bilaterally symmetrical as well. In the female genital disc, hh is differentially expressed in dorsal and ventral disc epithelia, while ci is expressed in a pattern complementary to that of hh (Fig. 2A,B). In the male genital disc, hh is also expressed in both dorsal and ventral disc epithelia. hh expression in the ventral anterior lobe is confined to the dorsal surface of the lobe, which can be observed from the dorsal view of the disc. Here too, ci is expressed in a pattern complementary to that of hh (Fig. 3A,B). There is no overlap between these two cell populations at the border. The en gene has a expression pattern similar to that of hh in both female and male genital discs (Figs 2C,D, 3C,D). The complementary patterns of en(hh) and ci expression are analogous to the situation in the wing and leg discs, where en(hh) define the posterior compartment and ci the anterior compartment. In both female and male genital discs, ptc is expressed at a high level along the entire border between en(hh)- and ci-expressing cells. Moreover, these cells overlap with ci-, but not en(hh)-, expressing cells at the border (Figs 2H,I, 3H). Again, this is very similar to the expression pattern of ptc in the wing and leg discs. Interestingly, dpp and wg are also expressed in non-overlapping stripes along the border between en(hh)- and ci-expressing domains in female and male discs (Figs 2E-G, 3E-G). For example, on the anterior lobe of the male disc, wg is expressed in the median region, while dpp is expressed in the more lateral positions along the border. This is analogous to what is seen in the leg disc, where dpp and wg are expressed in non-overlapping dorsal and ventral sectors along the A/P compartment border. Taken together, these results suggest that the genital disc, like the thoracic discs, is divided into anterior and posterior compartments.

A comparison of en(hh) and ci expression patterns with the known fate maps of the genital discs reveals that both the genital and the anal primordia contain complementary en(hh)- and ci-expressing regions (Fig. 4). For example, in the male genital disc, the en(hh)-expressing regions on the anterior lobe of the disc correspond to part of the male genital primordium, whereas the en(hh)-expressing regions on the dorsal posterior lobes of the disc correspond to part of the male anal primordium (Fig. 4E,F). Complementary en(hh) and ci expression patterns are also observed in the female genital and anal primordia of the female disc (Fig. 4A-C). en(hh) and ci are also expressed in non-overlapping cells in the repressed male and female primordia, respectively, of the female and male genital discs (Fig. 4A-D,F). These results suggest that each primordium of the genital disc may be divided into anterior and posterior compartments.

In order to test directly whether the genital disc is composed of anterior and posterior compartments, which are defined by ci- and en(hh)-expressing domains respectively, we made mitotic clones in the genital disc and asked if these clones could cross the border between the two domains. The mitotic clones were induced using the FRT-FLP system. Specifically, flies homozygous for an FRT and heterozygous for a ubiquitously expressed MYC epitope on the FRT-carrying chromosome arm were heat-shocked at the beginning of the second instar, when cell division normally resumes in the genital disc (Madhavan and Schneiderman, 1977). Clones were scored in the mature third instar discs. Homozygous Myc clones were identified by their elevated level of MYC epitope expression (above the heterozygous background), and their twin spots by the absence of MYC epitope expression. In 210 mosaic discs examined, over 20 large clones were found that abutted the border between hh-expressing and non-expressing cells. None of these clones crossed the border (Fig. 5).

This result demonstrates that there is indeed a clonal boundary between the en(hh)- and ci-expressing cells in the genital disc. We conclude that the genital disc has been subdivided into anterior and posterior compartments by the beginning of the second instar, if not earlier. To be consistent with the nomenclature in the wing and leg discs, we will refer to cells expressing en(hh) as posterior cells, and cells expressing ci as anterior cells in the genital disc.

Given that the relative domains of expression of the A/P patterning genes in the genital disc are similar to those in the leg disc, and that there is a clonal boundary between en(hh)- and ci-expressing cells, we tested whether the genetic interactions among these genes are conserved in the genital disc. Previous studies in thoracic discs have shown that in the anterior compartment, Pka activity represses the expression of genes that are normally expressed at the A/P compartment border, including dpp, wg and ptc. We generated Pka mutant clones in the genital disc and examined the expression of ptc, dpp, and wg by using lacZ lines that faithfully represent their endogenous expression patterns (Blackman et al., 1991; Chen and Baker, unpublished observation).

As shown above, ptc, dpp and wg are expressed along the A/P compartment border of the genital disc. While ptc is expressed along the entire length of the A/P border, dpp and wg are expressed in non-overlapping stripes along the border. In Pka mosaic discs, we observed that ptc was ectopically expressed in all the Pka mutant clones located in the anterior compartment of the genital disc (Fig. 6A,B). Likewise, dpp and wg were ectopically expressed in Pka mutant clones in the anterior compartment (Fig. 6C-F). Interestingly, the region in which dpp was ectopically expressed was distinct from the region in which wg was ectopically expressed. This is analogous to the situation in the leg disc where dorsal anterior Pka mutant clones express dpp, and ventral anterior clones express wg. We did not observe ectopic dpp, wg or ptc expression in Pka mutant clones in the posterior compartment of the genital disc (data not shown). Taken together, these results show that in the genital disc, Pka functions in the anterior compartment to repress the expression of genes that are normally expressed along the A/P compartment border. Thus, as in the thoracic discs, Pka exerts an anterior compartment-specific function in the genital disc.

Previous studies have shown that in the posterior compartment of the wing disc, en can activate the expression of hh, and repress the expression of dpp, ptc and ci. The invected (inv) gene encodes a protein that has a function partially redundant to that of en. Therefore, the en-inv double mutant has a more severe mutant phenotype than en mutant alone (Tabata et al., 1995). Homozygous en-inv mutant clones in the posterior compartment of the wing disc cause ectopic expression of dpp, ptc and ci, and the loss of hh expression. We have made en-inv mutant clones in the genital disc and examined the expression of dpp and hh. As in the wing disc, dpp was ectopically expressed in the en-inv mutant clones in the posterior compartment (Fig. 7A-D). However, ectopic dpp expression was never observed in the anterior compartment. As in the wing disc, en-inv mutant clones in the posterior compartment resulted in the loss of hh expression (Fig. 7E,F). Thus, en-inv function in the posterior compartment of the genital disc to negatively regulate dpp and positively regulate hh.

Since our previous results demonstrated that the genital disc is divided into anterior and posterior compartments, and the interactions among the A/P patterning genes are likely to be conserved in the genital disc, genes that have compartment-specific functions, such as Pka and en-inv, can be used in genetic mosaic analyses to localize the boundary between the anterior and posterior compartments in the adult structures. Previous studies have shown that Pka mutant clones can cause pattern duplications and/or local overgrowth in the anterior, but not posterior, structures in adult wing and leg. We reasoned that the adult terminalia structures showing pattern duplications and/or local overgrowth associated with Pka mutant clones are likely to belong to the anterior compartment, whereas structures not affected by Pka mutant clones are likely to belong to the posterior compartment.

We made Pka mutant clones by using the FRT-FLP system. These clones, in adult flies, were marked by y. We first examined the external terminalia of Pka mosaic animals. In females, Pka mutant clones were associated with duplications and/or local overgrowth in the vaginal thorn bristles, vulva, both the dorsal and ventral anal plates (Fig. 8C,D), and the sensillae in the ventral part of the 8th tergite (data not shown). Pka mutant clones often induced the overgrowth of the surrounding wild-type tissues. In males, duplications and/or local overgrowth were also frequently observed in most parts of the external terminalia, including the clasper teeth, the lateral plate bristles, the penis apparatus, and both the left and right anal plates (Fig. 9B-D).

The internal genitalia of Pka mosaic flies were also examined. In females, duplications were found in most parts of the internal genitalia, including the uterus, the oviduct, the seminal receptacles, and the spermathecae. In some flies, four spermathecae were observed instead of the two, or rarely three, found in wild type (Fig. 8E,F). However, duplicated parovaria were not observed (Table 1A). In males, most parts of the internal genitalia were also found duplicated in Pka mosaic animals, including the sperm pump, the ejaculatory duct, the paragonium, and the vas deferens (Table 1B).

Taken together, these data show that Pka mutant clones can induce duplications and/or local overgrowth in many parts of the female, male and anal structures, suggesting that each of the three primordia of the genital disc has an anterior compartment. In contrast, a few structures, such as the parovarium and the dorsal part of the 8th tergite, never showed duplications in Pka mosaic flies, suggesting that these structures belong to the posterior compartment. Our finding that most of the female and male genital structures showed duplications and/or local overgrowth associated with Pka mutant clones suggests that most of the genital primordia are of anterior, rather than posterior, origin. This is consistent with the observation that ci is expressed in larger regions of each genital disc primordium than is en (Fig. 4).

Previous studies in the wing disc have shown that en functions as the posterior selector gene. Homozygous en-inv mutant clones in the anterior compartment appear normal, whereas the mutant clones in the posterior compartment can cause posterior to anterior transformations. In order to further test which adult structures belong to the anterior or the posterior compartment, we have made en-inv mutant clones in the genital disc and examined the adult terminalia phenotypes in the en-inv mosaic flies. In females, most en-inv mutant clones appear normal. These clones are located within the anterior structures defined by Pka mosaic analysis, such as the vaginal thorn bristles (Fig. 8G). The regions that showed abnormalities in en-inv mosaic flies are much smaller than the areas that were affected by Pka mutant clones. en-inv mutant vaginal thorn bristles were found in the region between the ventral anal plate and the dorsal vaginal plate which is, in wild type, naked cuticle (Fig. 8H). In addition, en-inv mutant bristles were observed on the dorsal part of the 8th tergite, which is normally devoid of bristles (Fig. 8I). These regions were not affected by Pka mutant clones and were likely to be of posterior origin. We interpret the appearance of these bristles as posterior to anterior transformations, which is analogous to what happens to en-inv mutant clones in the wing discs.

Similar results were obtained in males. Most of the en-inv mutant clones appear normal. For example, no defects were associated with en-inv mutant clones in the clasper teeth, the lateral plate bristles or the genital arch, which have been defined as anterior structures by Pka clonal analysis (Fig. 9E). However, inside the abdomen and near the penis apparatus, where no cuticular structures are found in wild-type flies, ectopic anterior structures, such as clasper teeth and lateral plate bristles, were found associated with en-inv mutant clones (Fig. 9F). These ectopic anterior structures were also likely to result from posterior to anterior transformation. In addition, the anal plates were malformed in en-inv mutant animals (Fig. 9G), suggesting that at least some cells in the anal primordium are of posterior origin.

Take together, these results are in agreement with our Pka clonal analysis of the A/P compartmental division in the adults. The fact that en-inv clones affected structures derived from all three primordia demonstrates that each primordium has a posterior compartment, contradicting a previous proposal (Lawrence and Struhl, 1982) that the female genital primordium is purely of anterior origin. Our finding that en-inv clones only affected a small part of the female and male genitalia suggests that the posterior compartment only occupies a small portion of each genital primordium, and is inconsistent with a previous suggestion (Epper and Sanchez, 1983) that the genital primordia are largely of posterior origin. In terms of the genital structures that are affected by both Pka and en-inv mutant clones, such as the penis apparatus in males, and the 8th tergite in females, we suggest that the A/P compartment boundary passes through these structures.

Previous studies using wg^ts alleles or wg heteroallelic mutants failed to reveal any specific function of wg in genital disc development (Baker, 1988a,b). In these studies, only the external genitalia phenotypes were examined. Since we have observed the expression of wg along part of the A/P compartment border in the genital disc, as well as the ectopic expression of wg in the Pka mutant clones, we reasoned that wg activity might play a role in patterning the genital disc.

Comparison of the wg expression pattern with the fate maps of the genital discs suggests that the wg expression domains in both the female and male genital discs correspond to the internal, but not the external, genitalia. We therefore examined both the internal and the external genitalia phenotypes of wg mutants. The wg mutant allele we used was a heteroallelic mutant combination between wg^CX3 and wg^CX4, which specifically lacks wg activity during imaginal development (Baker, 1987, 1988b). When the internal genitalia of the pharate adults were examined, all internal structures were deleted in both females and males (Fig. 10). In addition, the male external genitalia showed deletions in certain structures, such as the clasper teeth, the lateral plate bristles and the penis apparatus (Fig. 9H). The female external genitalia were generally unaffected (Fig. 10D). Similar results were obtained by reducing wg activity at the end of the second instar using a wg^ts allele (data not shown). We note that the structures deleted in the internal genitalia correspond to larger regions than the wg expression domains in the third instar disc, which may reflect a cell non-autonomous function of this signaling molecule.

Our studies of the expression patterns of genes known to play roles in the A/P patterning of the thoracic discs provide molecular evidence that the genital disc is divided into anterior and posterior compartments. As in the thoracic discs, en(hh) and ci are expressed in complementary domains in the genital disc. ptc is expressed in a stripe of cells within the ci expression domain abutting the en-expressing cells. wg and dpp are expressed in non-overlapping subregions along the border between the en(hh)-expressing and ci-expressing domains. This is strikingly similar to the situation in the leg disc, where wg and dpp are expressed in complementary subregions along the A/P compartment border. These results are largely consistent with those of Freeland and Kuhn (1996) who characterized en, hh, dpp, wg, and ci expression in the genital disc at the light microscopic level. In addition, our results from the clonal analysis demonstrate that the en(hh)-expressing and ci- expressing domains are true genetic compartments, since mitotic clones induced in the genital disc do not cross the border between the two domains. We refer to cells expressing en and hh as posterior cells, and to cells expressing ci as anterior cells.

It is not clear, however, when the genital disc begins to be divided into anterior and posterior compartments. Since the mitotic clones were induced at the beginning of the second instar stage, we believe that the two compartments have already been established by this time. Genetic mosaic analysis has shown that the A/P compartment boundary in the thoracic discs already exists in the blastoderm stage embryo (Lawrence and Morata, 1977; Steiner, 1976; Wieschaus and Gehring, 1976). It has been proposed that the primordia of the two compartments originate on opposite sides of the parasegment boundary in the embryo, and that en specifies the identity of the posterior compartment in both the primordia and mature discs (reviewed by Cohen, 1993). However, the onset of the A/P compartment division in the genital disc may not be as simple, since en is not continuously expressed throughout the development of the genital disc. At the end of germ-band shortening stage, the expression patterns of en in the tail region of the embryo (A8, A9 and A10) go through dramatic changes. en-expressing cells on the ventral ectoderm move dorsally, resulting in the juxtaposition and fusion of ‘anterior’ A8, A9, and A10 cells that lack en expression on the ventral epithelium (DiNardo et al., 1985; Kuhn et al., 1992). This is the stage when one can first detect histochemically the genital disc precursor cells (GDPC), which are a group of transversely elongated cells located between the A8 denticle belt and the anal pads on the ventral ectoderm of the embryo. Not surprisingly, en is not expressed in the GDPC (Chen and Baker, unpublished observation). Thus, unlike the situation in the thoracic discs, there is a period of time in which en is not expressed in the genital disc primordia. The later expression of en in the genital disc likely results from the re-specification of posterior cells among the fused ‘anterior’ cells, which probably happens between late embryogenesis and the beginning of the second instar. It is not clear at this point how this re-specification occurs.

A comparison of the expression patterns of en(hh) and ci with the fate maps of the third instar genital disc provides insight into the A/P compartmental organization of each of the three primordia of the genital disc. In both the female and male genital primordia, en(hh) and ci are expressed in complementary subregions, suggesting that each primordium has anterior and posterior compartments. Moreover, in the repressed male and female primordia, en(hh) and ci are also expressed in non-overlapping cells. en(hh)-expressing regions are smaller than those of ci, indicating that the two genital primordia have relatively smaller posterior compartments, and larger anterior compartments. Results from the Pka and en-inv clonal analyses are consistent with this conclusion. Duplications and/or local overgrowth were observed in Pka mosaic animals in most structures of the female and male genitalia. In contrast, posterior to anterior transformations were found at a limited number of positions in en-inv mosaic flies, indicating that only a few structures belong to the posterior compartment.

Our results also demonstrate that there are anterior and posterior compartments within the anal primordium. en(hh) and ci are expressed in complementary patterns in the anal primordia in both female and male discs. Consistent with these expression patterns, malformed anal plates were found in both Pka and en-inv mosaic animals. Moreover, en(hh) are expressed in a smaller region than is ci in the anal primordium of the disc, suggesting that the anal primordium also has a relatively larger anterior and a smaller posterior compartment.

Based on our studies, we propose that each of the three primordia that make up the genital disc are composed of an anterior and a posterior compartment, and that most of the genital disc cells are of anterior origin (Fig. 11C). This model is different from the two previously proposed models in two regards. While Lawrence and Struhl’s model suggested that the female primordium is of anterior origin, and the male and anal primordia are made of posterior cells (Fig. 11A), our model proposes that each primordium contains both anterior and posterior compartments. In addition, while Epper and Sánchez’s model suggested that most of the genital primordial cells are of posterior origin (Fig. 11B), we propose that most cells of each primordium are of anterior origin.

In order to reconcile these disparate findings, it is useful to consider the following factors. The first model proposes that the anal primordium is purely of posterior origin based on the fact that the adult anal plates were affected by en mutant clones. However, if the anal plates contain both anterior and posterior compartments, as our results demonstrate, the anal plates would still be affected by en mutations. Second, it has been shown recently that inv contributes to the posterior selector function of en, and that the en-inv double mutant causes stronger posterior to anterior transformation phenotypes in the wing than do en mutants alone. Indeed, we have evidence that en-inv mutant clones have stronger phenotypes than en single mutant clones in the genital disc. Accordingly, we used an en-inv double mutant in our clonal analysis instead of en single mutants used in previous studies. Third, in both of the previous studies, en mutations were found to cause deletions in the genitalia, while we observed posterior to anterior transformations associated with en-inv mutant clones. A possible origin of these differences is the time at which en function was removed in these three studies. In the study by Lawrence and Struhl (1982),en mutant clones were induced at the first instar stage and in the study by Epper and Sánchez (1983) heteroallelic combinations of en mutations were used, while in our studies en-inv mutant clones were induced in second larval instar. It is conceivable that en has an early function that is distinct from its compartment selector function. For example, en could also be involved in cell survival/proliferation before the second larval instar, such that the reduction of this early activity would confound the interpretation of its compartment selector function. In fact, an involvement of en in cell survival/proliferation has been documented in the wing (Hidalgo, 1994). Thus, one cannot solely rely on such deletion phenotypes as indications of the compartment selector function for en, nor use such phenotypes to infer the position of the A/P compartment boundary. Finally, while both previous models were based on the mutant phenotypes of a single gene, en, our model is based on the mutant phenotypes of anterior compartment-specific gene Pka and posterior compartment-specific genes en-inv, as well as expression patterns of six molecular markers. Moreover, the expression patterns of the molecular markers are consistent with the mutant phenotypes of the compartment-specific genes. Such a multifaceted approach should reveal more reliably the compartmental organization of the genital disc.

Unlike the regularly patterned wing or leg, the adult terminalia are composed of many different sexually dimorphic tissues arranged in a more complex pattern. In theory, one could deduce the A/P compartment boundary in the terminalia by comparing the expression patterns of compartment-specific genes in the third instar disc to the fate maps. However, the low resolution of the fate maps makes such a deduction difficult. In this study, we have taken advantage of the clonal phenotypes of Pka and en-inv mutations to map the anterior and posterior compartments in adult terminalia more precisely. We found that Pka mutant clones caused duplications and/or local overgrowth in most of the terminalia structures, suggesting that these structures, or at least parts of them, belong to the anterior compartment. On the other hand, posterior to anterior transformations caused by en-inv mutant clones were only observed in a limited number of structures, indicating that few regions belong to the posterior compartment. Moreover, a few structures, such as the penis apparatus in male and the 8th tergite in female terminalia, are affected by both Pka and eninv mutant clones. We suggest that the latter structures are composed of both anterior and posterior compartments. Thus, the A/P compartment boundary in the adults does not simply lie between different structures. In some cases, the compartment boundary bisects certain structures.

Our results from Pka and en-inv clonal analyses have demonstrated that the interactions among the A/P patterning genes in the genital disc are analogous to those in the thoracic discs. We have shown that Pka is required to repress the downstream target genes dpp, wg and ptc in the anterior compartment, whereas en and inv are required to activate hh expression and repress dpp expression in the posterior compartment.

It has been shown that DPP controls growth and patterning of both the anterior and posterior compartments of the wing disc (Basler and Struhl, 1994; Capdevila and Guerrero, 1994; Lecuit et al., 1996; Nellen et al., 1996; Tabata and Kornberg, 1994; Zecca et al., 1995). It is also known that dpp^disk mutants cause deletions in both female and male genital disc derivatives (Blackman et al., 1991). In this study, we have demonstrated that, as in the thoracic discs, dpp is expressed along part of the A/P compartment border of the genital disc, and is ectopically expressed in Pka mutant clones in the anterior compartment and in en-inv mutant clones in the posterior compartment. We suggest that dpp is also likely to be an important organizer of the A/P developmental field in the genital disc.

Our studies have also revealed a function for wg during genital disc development. It has been suggested that WG acts as a gradient morphogen to pattern the leg disc (Struhl and Basler, 1993). However, the function of wg in genital disc development had not been uncovered previously, partly because only the external terminalia were examined (Baker, 1988a,b). In this study, we have demonstrated that wg is expressed along the A/P compartment border in the genital disc, in a pattern complementary to that of dpp. In addition, wg is ectopically expressed in Pka mutant clones. The expression domains of wg correspond to part of the regions that will give rise to the internal genitalia in both female and male discs. Consistent with its expression pattern, wg mutant flies show deletions of the internal genitalia. The fact that the deleted tissues correspond to a larger region than the wg expression domain is consistent with the cell non-autonomous function of wg.

We would like to thank K. Cadigan, J. Hooper, T. Kornberg, T. Laverty, K. Matthews, R. Nusse, D. Pan, G. Rubin, and T. Tabata for fly stocks, K. Cadigan, S. Carroll, C. Goodman, I. Guerrero, M. Hoffman, R. Holmgren, R. Nusse, and T. Tabata for antibodies, and R. Nöthiger for permitting us to modify the figures of the genital disc fate maps and the adult derivatives. We would also like to thank S. Ahmad, A. Franke, E. Keisman for their help with computer work and confocal microscopy, S. Ahmad, G. Bashaw, E. Keisman, H. Li, I. Mar ín, D. Pan, and S. Plump for their valuable comments on the manuscript, and G. Bohm for preparing the fly food. E. Chen would especially like to thank Duojia Pan for discussion and support during the work. This work was funded by an NIH grant to B. S. B.