A 3′ cis-regulatory region controls wingless expression in the Drosophila eye and leg primordia

Authors


Abstract

The precise regulation of wingless (wg) expression in the Drosophila eye disc is key to control the anteroposterior and dorsoventral patterning of this disc. Here, we identify an eye disc-specific wg cis-regulatory element that functions as a regulatory rheostat. Pannier (Pnr), a transcription factor previously proposed to act as an upstream activator of wg, is sufficient to activate the eye disc enhancer but required for wg expression only in the peripodial epithelium of the disc. We propose that this regulation of wg by Pnr appeared associated to the development of the peripodial epithelium in higher dipterans and was added to an existing mechanism regulating the deployment of wingless in the dorsal region of the eye primordium. In addition, our analysis identifies a separate ventral disc enhancer that lies adjacent to the eye-specific one, and thus altogether, they define a 1-kb genomic region where disc-specific enhancers of the wg gene are located. Developmental Dynamics 235:225–234, 2006. © 2005 Wiley-Liss, Inc.

INTRODUCTION

In Drosophila, most structures of the fly head, including a set of sensory organs (eyes, antennae, palps, ocelli) and the head capsule, derive from a pair of larval eye–antennal discs (Haynie and Bryant, 1986). wingless (wg), the fly homologue of the mammalian Wnt-1 gene, plays essential roles during the development of the Drosophila head capsule and eye primordia by controlling the establishment of the dorsoventral (DV) and anteroposterior (AP) axes of the eye disc (Lee and Treisman, 2001). Thus, wg is expressed early on in dorsal cells of the first and second larval stage (L1 and L2) eye disc, both in the disc proper, which will give rise to the ptilinum and periocellar region of the head capsule, and in the peripodial epithelium, which also contributes to head capsule structures (Haynie and Bryant, 1986; Royet and Finkelstein, 1996; Cavodeassi et al., 1999; Cho et al., 2000; Pichaud and Casares, 2000). The dorsal expression of wg induces the expression of the Iroquois-complex (IroC) genes araucan, caupolican (Cavodeassi et al., 1999), and mirror (Heberlein et al., 1998). The dorsal IroC genes are required to restrict the expression of the glycosyl-transferase encoding gene fringe (fng) to the ventral disc cells. The resulting DV boundary between fng-nonexpressing (D) and fng-expressing (V) cells leads to the localized activation of the Notch signaling pathway along the boundary, which is in turn necessary for the growth of the eye primordium and for the further patterning of the developing retina (Cho and Choi, 1998; Dominguez and de Celis, 1998). During late L2, dorsal wg expression retracts to the anterior margins of the eye disc, and a new smaller domain appears on its anterior ventral margin (Cavodeassi et al., 1999; Ma and Moses, 1995). Removing wg transcription or blocking its signaling pathway in these anterior domains results in an anterior expansion of eye development along the disc's margins (especially along the dorsal one) at the expense of the head capsule, which normally derives from these regions (Ma and Moses, 1995; Treisman and Rubin, 1995). Therefore, the precise regulation of the spatiotemporal expression of wg is essential for the development of the eye disc.

Several genes and signaling pathways have been proposed to control wg expression in the eye disc. First, the GATA-binding transcription factor Pannier (Ramain et al., 1993) has been proposed to positively regulate wg expression in the dorsal eye disc, placing this gene at the top of the genetic cascade that establishes the DV subdivision of the disc (Maurel-Zaffran and Treisman, 2000). Thus, in L3 discs Pnr is coexpressed with wg along the dorsal eye disc margin, and large clones of Pnr mutant cells lose wg expression (Maurel-Zaffran and Treisman, 2000). Nevertheless, and although Pnr is expressed in the eye disc primordium in late embryos (Maurel-Zaffran and Treisman, 2000), it is not expressed in L1 eye discs (Singh and Choi, 2003), when wg expression is already detected in dorsal cells (Cho et al., 2000). In addition, wg expression domain is larger than Pnr's in the L3 disc: while wg is expressed in the outer dorsal margin and peripodial epithelium (PE, that overlays the main epithelium or disc proper), plus the dorsal region of the disc proper, Pnr is expressed just in the outer margin and PE, abutting the Iro-C domain of expression in the dorsal disc proper (Pichaud and Casares, 2000). These results suggest Pnr might not be the sole dorsal activator of wg expression. The ventral wg expression domain requires the transcription factor homothorax (hth; Pichaud and Casares, 2000), yet how this domain is initiated is still unknown. The decapentaplegic (dpp, a BMP-4 like gene) signaling pathway regulates wg transcription, preventing wg ectopic expression along the posterior eye margin and, thus, avoiding the wg-mediated repression of retinal differentiation. In this way, dpp controls the partition of the disc into eye competent regions, under Dpp influence, and head capsule competent region, under Wg's (Chanut and Heberlein, 1997; Dominguez and Hafen, 1997; Pignoni and Zipursky, 1997).

In this report, we identify a short and conserved eye-specific enhancer sequence located in the 3′ region of the wg gene that recapitulates its dorsal eye disc expression. Nevertheless Pnr is not required for the activity of the eye-specific enhancer or the transcription or translation of wg in the disc proper. Pnr is only required for wg expression in the PE of the eye disc. This role of Pnr in the peripodial epithelium might be a relatively recent acquisition during the evolution of insects. Sequences adjacent to the eye disc enhancer harbor additional wg enhancers for the wing and ventral (leg, antennal, and genital) discs and, together, constitute a disc-specific regulatory region of the wg gene.

RESULTS

A wg Eye Disc-Specific Regulatory Region Lies 3′ to the wg Transcription Unit, Within the wg-DWnt-6 Intergenic Region and Adjacent to a Separate Ventral Disc Enhancer

As a first step to molecularly dissect wg regulation during eye disc development, we set to identify the cis-regulatory elements (CREs) responsible for driving wg expression in this disc. We generated a series of transgenic strains harboring genomic fragments of the wg gene upstream of a heat-shock minimal promoter driving the lacZ reporter gene from previously unexplored regions (see Costas et al., 2004, and references therein). A ∼ 6.0-kb DNA fragment (wg2), derived from the wg 3′ genomic region (Fig. 1A) drives lacZ expression in imaginal discs but not in embryonic stages. Within the eye disc, wg2Z expression is confined to a large dorsal–anterior domain and to a smaller and weaker ventral domain, in a pattern reminiscent of wg expression (Fig. 1C,D). This expression is shared with an overlapping fragment, wg9, which extends further downstream (Fig. 1B and not shown). The overexpression of wg along the posterior margin of the eye disc, using a dpp-Gal4 driver line, does not extend the wg2 domain of expression (Fig. 1G), indicating that the eye disc enhancer is not an autoregulatory element but instead directly controlled by upstream wg regulators. The expression of Pnr with the same driver, though, leads to the expansion of wg2Z expression along the posterior margin (Fig. 1H), which indicates that Pnr has the potential to activate wg expression, in agreement with a previous report (Maurel-Zaffran and Treisman, 2000).

Figure 1.

A wg eye disc-specific enhancer lies within its 3′ noncoding region. A: Scheme of the genomic region of the wg gene indicating the position of the wg2 fragment and derived deletion fragments tested in this study. Bars indicate exons, and coding regions are in magenta. wgP-lacZ (wgZ) is a P-element insertion close to the wg promoter and used as a reporter of wg transcription (Kassis et al., 1992). Transgenic strains harboring wg2 and wg9 lacZ reporter constructs showed similar patterns of expression, although wg9 is weaker (not shown). B: Vista alignment (Brudno et al., 2004) of the sequence spanning wg2 and wg9 in Drosophila melanogaster and D. pseudoobscura, using a calculation and conservation window of 300 bp and a threshold of 70% for conservation identity labeling. Genome coordinates are indicated according to the Drosophila genome release r3.1. Peaks indicate sequence similarity, and regions shaded in pink correspond to highly conserved sequences. Blue bars indicate, to scale, the 3′ genomic regions tested in transgenic strains. Plus and minus symbols indicate the presence and absence, respectively, of eye disc expression. C,D,F: Similar patterns of expression in eye discs of wgZ (C), wg2 (D), and wg2.10 (F). E:wg2.4 larvae lose the eye-specific expression. This disc was overstained compared with those in C,D,F–H and, thus, shows the faint antennal expression (see Fig. 2). Dorsal (arrow) and ventral (arrowhead) expression domains are indicated. wgZ also shows strong antennal-specific expression (white arrowhead). G:dppblkGal4-driven wg expression along the posterior margin of a wg2-lacZ eye disc causes an abnormal development of the disc, but does not result in lacZ expansion along the margin (arrowhead). H: When Pnr is driven by dppblkGal4, the expression of wg2-lacZ now expands along the posterior margin (arrowhead). Signal in the antenna of wg2Z discs only appears if the X-Gal reaction is developed for longer times.

Figure 2.

Enhancer activities for the wing, leg, and antennal imaginal discs also lie within the wg2 fragment. A–R: Confocal images of the prospective notum region of wing (left column), leg (middle column), and antennal (right column) imaginal discs of wg2 (A–C), wg9 (D–F), wg2.5 (G–I), wg2.4 (J–L), wg2.9 (M–O), and wg2.10 (P–R) L3 larvae. Discs are stained for lacZ (green), Wg (red), and Hth (for leg and antennal discs, blue). LacZ expression is seen in a stripe of dorsal peripodial epithelial cells in the antennal discs (arrowhead in C) of all lines, except for wg2.4. Expression in the prospective notum is maintained in all lines, although it is reduced to a few cells in line wg2.10 (P). Signal in the leg and antennal discs is restricted to the medial and distal segments of the discs (in the leg these segments correspond to the ones not expressing Hth), except for wg2.10 discs, where the signal is lost (Q,R). S: Summary of the results, where the presence or absence of expression is indicated by a plus or a minus symbol, respectively; “(+)” indicates weak expression. Green and purple bars indicate the regions containing the wg notum (WNE) and leg/antennal (WVDE) enhancers. The span of WNE is ill defined on its 3′ region (discontinuous bar). T: Vista comparison of the 506-bp WVDE sequence (position 2L: 7315503-7316008 [release r3.1]) between Drosophila melanogaster and D. pseudoobscura, with calculation and conservation window of 50 bp and threshold of 70% for conservation identity labeling. Peaks indicate sequence similarity, and regions shaded in pink correspond to highly conserved sequences. Sites similar (seven of nine matches) to the Ci consensus binding sites (TGGG(A/T)GGTC; Zarkower and Hodgkin, 1993; Kwon et al., 2004) are marked as “Ci”. TAAT sites (“T”), which are the core of putative homeodomain binding sites (Hanes and Brent, 1991), lie clustered to the 3′ of WVDE. GATA sites are marked as “G”. None of them match the GATAA consensus sequence for GATA sites (Haenlin et al., 1997).

Deletion analysis of the wg2 fragment narrowed the eye disc cis-regulatory element down to a 502-bp fragment necessary and sufficient to drive the eye-specific pattern (wg2.10; Fig. 1E,F). The dorsal expression domain is strongly and reliably detected, whereas the late weaker ventral domain is more variable in intensity. While during late L3 and pupal stages wg expression extends around the posterior margin, so by the end of pupal development its expression encircles de eye (Tomlinson, 2003), wg2.10Z remains expressed strongly in the dorsal–anterior margin (not shown). This result indicates that further enhancer elements, responsible for the late pattern of wg expression in the eye disc, lie outside wg2.10.

In addition, wg2Z and wg9Z larvae show weak lacZ expression in all ventral discs—leg, antennal, and genital—in a pattern similar to that of wg (Baker, 1988; Couso et al., 1993; Casares et al., 1997b), plus a patch in the prospective notum of the wing disc (Fig. 2). The ventral discs and notum signals were very weak in histochemical X-Gal stainings, and were analyzed by immunofluorescence/confocal microscopy. The notum signal we detect in wg2 and wg9 is preserved in wg2.4 and wg2.9 larvae, but is very much reduced in wg2.10 (Fig. 2A,D,P). These results indicate that this wg notum-specific enhancer (WNE) maps within the wg2.9 fragment [2L: 7315503-7316510] and that sequences upstream to wg2.10 are required for full expression (Fig. 2). We note that the pattern of expression of the notum enhancer, although roughly similar to the prospective notum domain of wg, does not exactly replicate it (see Fig. 2A, for example). This enhancer might be part of the cis-regulatory regions controlling wg expression in the prospective notum, but binding sites for critical regulators shaping the final expression domain of the wg gene must lie outside the fragments analyzed.

In addition, the analysis of the ventral disc expression in different wg2 derivatives allows us to map a wgventral disc enhancer (WVDE) to the 506-bp fragment (adjacent to wg2.10) that is shared by wg2.4 and wg2.9 (7315503-7316008; Fig. 2S), as it is expressed by wg2.4 and wg2.9 but not by wg2.10 (Fig. 2K,L,N,O,Q,R). Vista analysis of this sequence shows regions of high conservation between Drosophila melanogaster and D. pseudoobscura (Fig. 2T). The expression of wg in legs and antennae is dependent upon hh signaling (Jiang and Struhl, 1995), which is mediated by the Cubitus interruptus (Ci) transcription factor (Von Ohlen and Hooper, 1997). Sequences with similarity to the Ci binding sites (Zarkower and Hodgkin, 1993; Kwon et al., 2004) are found in WVDE (Fig. 2T), consistent with a possible regulation of WVDE expression by hh. Therefore, enhancers for the eye, wing, and leg and antennal imaginal discs reside within ∼ 1 kb of wg 3′ intergenic region.

Deletion Analysis Suggests a Regulatory Rheostat Organization Within wg2.10

We next focused on the regulation of the dorsal domain of wg expression in the eye disc, as it is through this expression that wg controls the DV organization of the disc and has a major role in repressing dorsal eye fate (Ma and Moses, 1995; Treisman and Rubin, 1995; Von Ohlen and Hooper, 1997; Maurel-Zaffran and Treisman, 2000). Considering the previously described wg enhancers (Neumann and Cohen, 1996a, b; Von Ohlen and Hooper, 1997; Lessing and Nusse, 1998; Costas et al., 2004), and the noncoding wg gene regions surveyed in this study (FIG. 1), the 502-bp wg2.10 sequence is unique in its ability to direct expression in the eye disc.

Further comparative analysis of wg2.10 from D. melanogaster, D. pseudoobscura and D. virilis sequences reveals that it contains three sequence blocks (blocks 1, 2, and 3; Fig. 3A) that are highly conserved in all 11 Drosophila species for which preliminary genome sequences are available, including representatives of the Sophophora and Drosophila subgenuses (Trace Archive database at www.ncbi.nlm.nih.gov). In addition, these blocks are colinear in all species. This finding indicates an overall conservation at both the nucleotide and genomic organization levels of this cis-regulatory region. These blocks are no longer conserved in more distant species, not even in the dipteran Anopheles gambiae (not shown). The length of these conserved noncoding sequences suggests they might be the functional elements within the wg2.10 fragment. To test their regulatory contribution, we analyzed the effect of their removal on the reporter's expression in transgenic animals. Removal of block 1 (wg2.11) results in a significant increase of lacZ transcription levels, without altering its spatial domain, indicating that block1 down-regulates quantitatively the final transcriptional output (Fig. 3B,C). Deletion of the block3 region from wg2.11 results in a weakening of the expression pattern (wg2.12; Fig. 3B,C), although again without major changes in the domain of expression, whereas the block3 region is not able to drive lacZ expression on its own (wg2.13; Fig. 3B,C). Therefore, the wg2.12 fragment, containing block2, behaves as a wg eye disc-specific minimal enhancer (WEE), whereas the two flanking blocks in the 502-bp 2.10 fragment (blocks 1 and 3) modulate negatively and positively, respectively, the final output levels, acting together as a regulatory rheostat. The 47-bp block2 is remarkably conserved across approximately 40 million years of evolutionary distance, with a sequence identity of 93% or higher in the Drosophila sample species analyzed (Fig. 3A and not shown). Therefore, block2 should be the primary site of binding of the transcription factors regulating wg expression, and ultimately, the DV and AP organization of the eye disc. Blast analysis shows that none of these conserved blocks is found anywhere else in the D. melanogaster genome (not shown).

Figure 3.

The 502-bp wg2.10 cis regulatory region is organized as a conserved regulatory rheostat. A: Comparative sequence alignments of wg2.10 from Drosophila melanogaster, D. pseudoobscura, and D. virilis reveals three blocks (1 to 3) of conserved sequence with length–percentage of identity of 25 bp–96%; 47 bp–98%, and 35 bp–100%, respectively. The different regions assayed (wg2.10, 2.11, 2.12, and 2.13) are delimited by arrows. Within the blocks, nucleotides conserved in the three species are colored in black; those conserved in only two are colored in gray. D. viriliswg2.10 sequence was extracted from Trace Archive 392151478. B: Scheme of the wg2.10 regulatory region and of the smaller deletions derived from it. The position and size of the depicted conserved blocks is shown at scale. C: LacZ reporter gene expression of in parallel X-Gal stained eye discs of wg2.10Z, wg2.11Z, wg2.12Z, and wg2.13Z L3 larvae are shown, alongside with the fragments included in each construct. Fragment 2.12 retains the ability to direct eye-specific reporter gene expression and, therefore, contains a minimal enhancer. Regions containing blocks 1 and 3 regulate quantitatively the output of the regulatory region, acting as negative and positive elements, respectively.

GATA-Binding Transcription Factor Pnr Regulates wg Expression in the Peripodial Epithelium of Late Discs but Is Not Required for wg Expression in the Disc Proper

The early functions of wg in establishing the DV and AP axes of the eye disc rely on its regulated expression along dorsal eye disc and its later restriction to the dorsal anterior disc margin. Pnr has been proposed to act as an activator of this early wg expression (Maurel-Zaffran and Treisman, 2000; see also Fig. 1H). To test whether Pnr might be acting through WEE, we induced marked clones of Pnr mutant cells in both wg2.10-lacZ and wg2.11-lacZ reporter backgrounds. The two reporters behaved in an undistinguishable manner: most Pnr clones in dorsal margin did not affect wg reporter expression (Fig. 4A,A′; compare with 4H) or, in a few cases, show an expansion of WEE expression (Fig. 4C,D). Only Pnr clones spanning the dorsal peripodial epithelium caused a cell autonomous loss in reporter gene expression (not shown). To check if Pnr-activated wg enhancers lie outside WEE, we repeated the experiment with a wgZ reporter instead, which fully recapitulates wg expression (Kassis et al., 1992). Nevertheless, the results were the same as for wg2.10-lacZ and wg2.11-lacZ: Pnr-mutant cells maintain wgZ expression (Fig. 4B), except for some dorsal peripodial clones (Fig. 4E), which lose wgZ expression. Also, some clones in the anterior disc margin show ectopic wgZ expression (not shown). We also excluded a positive role of Pnr in the regulation of wg in the disc margin at the protein level, as Pnr mutant clones also maintain wild-type levels of Wg product (Fig. 4F,G). Therefore, Pnr is required for wg expression in the dorsal peripodial epithelium but not along the dorsal disc margin.

Figure 4.

Loss of Pnr does not alter the overall WEE, wgZ, or Wg protein expression domains but results in the loss of WEE or wgZ expression in the peripodial epithelium. Loss of function pannier mutant (PnrVX6) cell clones were induced by ey-flip and detected in L3 eye discs by the absence of green fluorescent protein (GFP). In all panels, GFP is shown in green, and different wg-lacZ reporters or Wg protein in red. Discs are oriented such that anterior is to the left and dorsal is up. A,B,H: The antibody 22C10 labels the developing photoreceptors (blue). E: Peripodial nuclei are stained with anti-Hth antibody (blue). A,B:PnrVX6 clones do not alter wg2.11-Z (A,A′) or wgZ (B,B′) eye disc expression (A′ is a blow up of the region marked in A, and B and B′ are two different focal planes of the same disc). Ectopic eye development is taking place in the clone shown in (B,B′), detected by 22C10-positive cells (asterisk). C,D: Eye disc with Pnr clones showing autonomous derepression of wg2.10 along the clone boundaries (arrows; C′ and C″ are blow ups of C, and D is a second focal plane of the same disc). E: A Pnr clone in the peripodial epithelium loses wgZ expression autonomously (arrow). F,G:Pnr clones, induced and marked as previously, maintain normal Wg protein pattern and levels (F shows the merged channel; F′ only the GFP channel; G is a high magnification view of a different dorsal clone). H: A wild type wg2.11Z pattern is shown for comparison.

Because Pnr is a GATA-binding transcription factor, it might regulate wg expression in the peripodial epithelium by directly binding to WEE. In fact, we identify two conserved putative GATA binding sites (Haenlin et al., 1997) in the enhancer (Fig. 5A). Nevertheless, transgenic strains carrying the wg2.11-lacZ reporter with GATA sites 1 and 2 mutated individually or in combination to CTGA sites (which have been shown to abolish Pnr binding; Garcia-Garcia et al., 1999) show a normal pattern of expression (Fig. 5B–E). Thus, Pnr does not regulate peripodial expression of wg through directly binding the two GATA sites in WEE. These results could also rule out other GATA-binding transcription factors as direct regulators of wg through WEE.

Figure 5.

Putative, conserved Pnr binding sites are not required for wg2.11 WEE activity. A: Putative transcription factor binding sites identified in the wg2.11 regulatory region are boxed. The sequence contains two possible Pnr binding sites (the core GATA sequence is highlighted), plus potential sites for dorsal/rel, vHNF-4, Pax6-homeodomain, and Pax-type paired domain transcription factor classes. We identify three copies of a seven-nucleotide long motif (marked as “repeat”). C–E: X-Gal staining of wg2.11(GATA#1mut) (C), wg2.11(GATA#2mut) (D), and wg2.11(GATA#1+2mut) (E) late L3 eye discs reveal that GATA sites are not required to establish the WEE pattern. B:wg2.11 is shown for comparison.

In addition to the putative GATA sites, we identify several other features in the wg2.11 region. These include potential binding sites for transcription factors of the dorsal/c-rel, vHNF-4, Pax6-type homeodomain and Pax-type paired domain classes, plus three repeated motifs with no match to known transcription factor binding sites (Fig. 5A).

DISCUSSION

A Small Disc-Specific Control Region of the wg Gene Lies in Its 3′ Intergenic Region

During embryogenesis and imaginal development, wg function relies on its highly complex and dynamic expression (reviewed in Klingensmith and Nusse, 1994). Regulatory elements controlling wg expression during embryogenesis have been located to the 5′ region upstream of the wg transcript and molecularly characterized (Von Ohlen et al., 1997; Lessing and Nusse, 1998). The activity of those regulatory elements ceases at the end of embryogenesis; therefore, imaginal-specific enhancers must be turned on. Nevertheless, much less is known about these wg imaginal-specific enhancers. Neumann and Cohen (1996a) and Costas and coworkers (2004) have characterized a wing disc margin/hinge enhancer in the wg 5′ intergenic region and a pupal enhancer within its third intron, respectively (Costas et al., 2004; Neumann and Cohen, 1996a). Therefore, other disc-specific cis-regulatory elements must exist to regulate wg expression. Although the span of intergenic region surrounding the wg transcript is vast, four wg mutations (wgCX3, wg1, wgP, wgSp) that specifically affect imaginal disc development, including eye, leg, and notum, have been associated with breakpoints in the region where we identify wg2 (Baker, 1987, 1988; Couso et al., 1993; van den Heuvel et al., 1993; Neumann and Cohen, 1996b), pointing to this area as harboring disc-specific enhancers. Here, we show that enhancer activity capable of directing wg-like expression in the eye, wing (prospective notum), and ventral (leg, antennal, and genital) discs resides in a surprisingly short 1-kb region within the wg2 fragment. It is, therefore, possible that all these mutations affect this disc-regulatory region. Although the enhancer activity of WVDE might be the composite result of several enhancers within it, it may as well be that a single enhancer works in all three ventral disc types: antennal, leg, and genital. This finding might be expected, because these discs develop into serially homologous structures that share basic regulatory mechanisms (Struhl, 1981; Casares and Mann, 1998; Sanchez and Guerrero, 2001; Estrada et al., 2003).

It is noteworthy that these wg disc cis-regulatory elements lie between wg and DWnt-6, two Wnt genes with very similar expression patterns in imaginal discs (Janson et al., 2001). This finding raises the possibility of both genes sharing these elements to control their expression in discs. Alternatively, the expression of one of the two Wnt genes might be dependent on the other, thus making their expression patterns resemble each other.

wg Expression in the Dorsal Eye Disc Does Not Depend Exclusively on Pnr

wg expression in the eye disc starts during L1 on its dorsal side and, later on, it comes up in the ventral side. The dorsal domain, which is larger than the ventral one, spans two territories in L3 discs: the outer dorsal margin and dorsal peripodial epithelium, where wg overlaps the Pnr realm, and the inner margin (already within the disc proper), that expresses genes of the Iroquois complex, and which is contiguous with the outer margin (Pichaud and Casares, 2000). These two territories will contribute to medial head structures (around the ocelli) and to frontal head structures (ptilinum), respectively (Royet and Finkelstein, 1996).

Our results indicate that Pnr function is required, during larval development, for the peripodial-specific expression of wg but not for the establishment of wg expression in the eye disc proper. A previous report showed that large clones of Pnr mutant cells, induced with the Minute technique (Morata and Ripoll, 1975), lose wgZ expression completely (Maurel-Zaffran and Treisman, 2000), contrary to the results we have obtained. One possibility to explain this discrepancy is that in large Minute+ clones the loss of wgZ expression could be an indirect result far downstream of the loss of Pnr, which induces ectopic eye differentiation and antennal duplication (Maurel-Zaffran and Treisman, 2000).

If the expression of wg in the disc proper of the eye disc does not require Pnr function, it must, therefore, be controlled by other factors. Nevertheless, analysis of the WEE sequence for putative binding sites of characterized transcription factors has not yielded any candidates for being direct regulators of this enhancer; therefore, these factors have yet to be identified. That the WEE enhancer is activated in both dorsal and ventral disc margins suggests that at least one of these factors should be present in both dorsal and ventral eye disc regions to activate wg expression through WEE.

Whereas the dorsoventral patterning of the eye by wg and its downstream target genes, the iroquois-complex, seems conserved in insects, in the locust Schistocerca wg expression precedes that of Pnr, and their expression patterns do not overlap (Dong and Friedrich, 2005), which rules out a direct regulation of wg by Pnr in this insect. Nevertheless, in Drosophila, Pnr plays a crucial role in establishing the dorsal territory of the eye disc (Maurel-Zaffran and Treisman, 2000, Singh and Choi, 2003). Here we show that Pnr is required for wg expression in the peripodial epithelium. This epithelial layer has been shown to be a major site for the production of signaling molecules (Cho et al., 2000; Gibson and Schubiger, 2000), and it might be that Pnr controls dorsal disc fates through its regulation of wg expression, and perhaps of other signaling molecules, in the peripodial epithelium. This epithelial layer is an advanced evolutionary specialization required for the development of the fully internalized imaginal discs of higher diptera; therefore, the role of Pnr in regulating peripodial wg expression might have arisen during the evolution of cyclorraphan flies, approximately 150 million years ago. Maurel-Zaffran and Treisman (2000) have proposed that the boundary between Pnr expressing and nonexpressing cells might trigger cell growth. We have noticed that, in some Pnr mutant clones, located in the dorsal anterior-most region of the disc, wg expression is autonomously de-repressed. It is possible, therefore, that it is wg itself that induces this proliferation, as it has been shown that wg is required for eye disc growth (Ma and Moses, 1995; Treisman and Rubin, 1995).

Our analysis of the eye disc enhancer does not yield any other clear candidate to be a positive regulator of wg expression in the eye disc. The presence of Pax-type homeodomain and paired binding sites, which could be potentially bound by the Pax6 proteins encoded by the eye-selector genes ey and toy (Gehring, 2002), is in agreement with the strong eye disc expression of the enhancer. Nevertheless, ey and toy do not seem sufficient to activate the enhancer. On the one hand, their expression is widespread within the eye disc (Czerny et al., 1999), whereas the expression of the eye enhancer is more restricted; and, on the other, ectopic expression of ey in other discs does not result in its activation (not shown). Eye gone (Eyg), another paired-homeodomain Pax-type transcription factor has been shown to repress wg transcription in the eye disc (Jang et al., 2003). This repression could be potentially achieved if Eyg binds to the predicted Pax homeodomain binding site, because an Eyg-type paired domain binding site in not present in the eye disc enhancer. No information regarding expression or function of the Drosophila HNF-4 in imaginal discs is available, although it has been implicated in embryonic gut development (Zhong et al., 1993). The putative Dorsal binding site, which could be bound by Dorsal or its related transcription factors Dif and Relish (Flybase: flybase.bio.Indiana.edu) falls in a nonconserved region and, therefore, is not likely to be of functional significance. Thus, further work is needed to test the involvement of Eyg and HNF-4 in the direct regulation of wg transcription through the eye disc enhancer as well as to look for further transcription factors involved in its regulation.

EXPERIMENTAL PROCEDURES

Plasmid Constructs

The wg2 and wg9 3′ genomic regions of the wg locus (Baker, 1987), were subcloned into the BamHI site of PCaSpeR-hs43-lacZ, and into the EcoRI site of pCAB70 (Bachmann and Knust, 1998) P-element transformation vectors, respectively, to generate the reporters wg2-lacZ and wg9-lacZ. All further described reporter constructs contain wg genomic sequences derived from the wg2 fragment. The internal deletion reporter construct wg2.1-lacZ was generated by removal of two internal BglII fragments reducing the wg genomic region to 3359 bp (corresponding to 2L:7310507…7311145 plus 7313792… 7316511, release 3.1). All the remaining deletion reporter constructs were generated by PCR amplifying wg2 fragments from the template wg2-lacZ using forward primers with terminal NotI sites and reverse primers mutating the wg2 terminal BamHI site to a BglII site. Reaction products were digested with NotI and BglII and inserted between those sites upstream of hs70 minimal promoter and the lacZ gene in the vector pCAB70. The length and genomic position of the several wg fragments used in these reporters, according to the Drosophila genome release 3.1, are as follows: wg2.3 (2,364 bp) 2L:7314147…7316510; wg2.5 (1,525 bp) 2L:7314986… 7316510; wg2.4 (1,428 bp) 2L:7314581…7316008; wg2.9 (1,008 bp) 2L:7315503…7316510; wg2.10 (502 bp) 2L:7316009… 7316510; wg2.11 (268 bp) 2L:7316243…7316510; wg2.12 (114 bp) 2L:7316243…7316356; wg2.13 (154 bp) 2L:7316357…7316510.

Mutations converting the Pnr/GATA factor binding sites from GATAA to inactive CTGAA sites (Garcia-Garcia et al., 1999) were generated by PCR. Transgenic lines were generated by standard methods and at least three independent lines for each construct were analyzed, except for wg2 and wg9, which were single insertions.

Prediction of Putative Transcription Factor Binding Sites

wg2.11 sequence was analyzed with Match (public version 1.1; www.biobase.de) and MatInspector (www.genomatix.de) algorithms, using default parameters. Both identify putative sites for dorsal/rel, vHNF-4, and Pax-paired transcription factors. The Pax6 homeodomain site was identified by MatInspector. The putative GATA-binding sites and the repeated motif were identified by direct inspection of the sequence.

X-Gal Histochemistry

Imaginal discs were dissected in cold phosphate buffered saline (PBS), fixed in 4% formaldehyde in PBS at room temperature and stained as described in Casares et al. (1997a).

Fly Strains and Genetic Manipulations

dppblk-Gal4, UAS-GFP-wg, UAS-Pnr, and wgP-lacZ (wgZ) stocks are described in FlyBase (flybase.bio.indiana.edu) and FRT 82B PnrVX6 in Maurel-Zaffran and Treisman (2000). For targeted gene expression, we used the UAS/Gal4 system (Brand and Perrimon, 1993).

Pnr Loss-of-Function Clones

To generate FRT-mediated mutant clones, flipase driven by the eyeless promoter was used to induce a high rate of mitotic recombination during the eye disc development (Newsome et al., 2000). PnrVX6 mutant cell clones were generated in larvae derived from the cross of y w ey-flp;; FRT82B Ubi-GFP stock females to y w; FRT82B PnrVX6/TM6B males carrying a lacZ reporter on the second chromosome when indicated. Pnr mutant cells were detected by the lack of GFP, and clones affecting the dorsal disc studied. To aid in locating the clones, Hth was often included in the stainings, as Hth is expressed specifically in disc margin and peripodial regions of the disc (Pichaud and Casares, 2000). The different wg reporters were introduced in the genotype by standard genetic procedures.

Immunostainings

The antibodies used were rabbit anti-βgal (Cappell) and guinea pig anti-Hth (Casares and Mann, 1998). Mouse anti-22C10 (Zipursky et al., 1984), mouse anti-Wg (Brook and Cohen, 1996), and rat anti-Elav (O'Neill et al., 1994) are from the Developmental Studies Hybridoma Bank. Appropriate fluorescently conjugated secondary antibodies are from Molecular Probes and Jackson Laboratories. Data were collected with Bio-Rad MRC600 and Leica TCS SP2 confocal systems.

Acknowledgements

We thank J. Treisman, M. Domínguez, S. Ricardo, the Bloomington Stock Center, and the Iowa University Hybridoma Bank for fly stocks and antibodies. P.S.P. is a Fundação para a Ciência e a Tecnologia (Portugal) postdoctoral fellow.

Ancillary