JAK/STAT signaling promotes regional specification by negatively regulating wingless expression in Drosophila

In metazoans, homogeneous cell populations acquire regional differences and form specific domains that ultimately give rise to distinct organs. This process of regional specification occurs through a cascade of cell fate decisions, in which broad regional specification is followed by differentiation of increasingly smaller domains. In both Drosophilaand vertebrates, a limited number of conserved signaling pathways, used reiteratively throughout development, regulate these processes. Drosophila imaginal discs represent an excellent model system to dissect the signals that control regional specification. Imaginal discs begin as groups of 20 to 50 cells that are formed as a single layer of epithelium that invaginates from the embryonic ectoderm(Cohen, 1993). Two opposing layers, the peripodial membrane and the disc proper, enclose the disc lumen and give the disc a sac-like appearance. Discs grow and differentiate during the three larval instars and evert during pupal development to become functional in the adult.

The eye imaginal disc is part of the compound eye-antennal disc and differentiates into the adult retina and head capsule(Haynie and Bryant, 1986). Early in development, there is no regional identity in the eye-antennal primordium (Dominguez and Casares,2005). The eye field emerges during the second larval instar by the action of the selector genes eyeless (ey) and twin of eyeless (Czerny et al.,1999; Quiring et al.,1994). These factors are required for the subsequent expression of the nuclear factor Eyes absent (Eya) in the posterior region of the eye-antennal disc, which is the presumptive eye field(Bonini et al., 1993). Eya is the first of a group of nuclear factors called the `early retinal genes',which also includes Sine oculis and Dachshund (Dac), that function together in a network to lock in eye fate (Pichaud et al., 2001). Decapentaplegic (Dpp), a secreted factor and BMP family member, is required for Eya expression and is expressed in a similar pattern to Eya (Kenyon et al.,2003). Wingless (Wg), a secreted Wnt factor induced by the GATA transcription factor Pannier (Pnr), acts antagonistically to Dpp and is expressed in the anterior domain of the eye in a pattern complementary to Dpp(Cho et al., 2000; Maurel-Zaffran and Treisman,2000). By negatively regulating the expression of early retinal genes, Wg controls the specification of retinal territory(Baonza and Freeman, 2002). In early eye development, when most of the cells in the eye disc receive both the Wg and the Dpp signal, retinal determination cannot take place(Cavodeassi et al., 1999; Pichaud and Casares, 2000; Royet and Finkelstein,1996).

During second instar, the eye disc becomes divided into dorsal and ventral compartments, which promotes growth of the entire disc by localized activation of Notch at the dorsoventral midline. Notch induces expression of Unpaired(Upd), a secreted mitogenic factor that activates the JAK/STAT pathway and promotes growth of the eye disc (Bach et al., 2003; Chao et al.,2004; Reynolds-Kenneally and Mlodzik, 2005; Tsai and Sun,2004). This growth separates wg and dpp domains,enabling anterior `head' and posterior `eye' regions to form in response to Wg and Dpp signaling, respectively (Kenyon et al., 2003). Following partition, wg is localized to the dorsal and ventral anterior margins of the eye disc and is maintained by Pnr and the homeodomain transcription factor Homothorax (Hth), respectively(Maurel-Zaffran and Treisman,2000; Pichaud and Casares,2000).

Retinal differentiation begins in third instar at the posterior midline of the eye disc, where the morphogenetic furrow forms and moves in an anterior direction across the eye disc (Ready et al., 1976). The antagonistic relationship between Dpp and Wg restricts the onset of differentiation to a narrow region at the posterior margin in the eye disc (Chanut and Heberlein, 1997; Ma and Moses,1995; Pignoni and Zipursky,1997; Treisman and Rubin,1995). Progression of the furrow requires Hedgehog, which is expressed in differentiated photoreceptors posterior to the furrow(Heberlein et al., 1993; Ma et al., 1993).

We have examined the contribution of the JAK/STAT pathway to regional specification. In mammals, this pathway is activated by cytokines and growth factors that bind to specific cell-surface receptors and lead to the activation of JAK tyrosine kinases that phosphorylate and activate latent cytosolic STAT transcription factors (reviewed by Levy and Darnell, Jr, 2002). Activated STATs dimerize and accumulate in the nucleus where they alter transcription of target genes. Studies have elucidated evolutionarily conserved roles for JAK/STAT signaling, including hematopoiesis, immunity and proliferation (Arbouzova and Zeidler,2006; Levy and Darnell, Jr,2002).

Drosophila serves as an excellent model system to study this pathway, as flies are highly amenable to genetic manipulation and possess only one JAK and one STAT gene, compared with the four JAKs and the seven STATs in mammals (Arbouzova and Zeidler,2006; Levy and Darnell, Jr,2002). Genetic studies in Drosophila have uncovered several components of this pathway, including Upd and two other Upd-like ligands called Upd2 and Upd3 (Agaisse et al., 2003; Harrison et al.,1995; Hombria et al.,2005); the transmembrane receptor Domeless (Dome), which is distantly related to the gp130 family of cytokine receptors(Brown et al., 2001; Chen et al., 2002); the JAK Hopscotch (Hop) (Binari and Perrimon,1994); and the STAT protein Stat92E(Hou et al., 1996; Yan et al., 1996). The JAK/STAT pathway is crucial for diverse processes in Drosophila,including proliferation and planar polarity in the developing eye(Bach et al., 2003; Luo et al., 1999; Tsai and Sun, 2004; Zeidler et al., 1999).

In this paper we report that the JAK/STAT pathway regulates regional specification of the eye disc by promoting the formation of the eye field via repression of the wg gene. Specifically, we demonstrate that JAK/STAT signaling negatively regulates wg expression in the eye disc epithelium and in the peripodial membrane and promotes the formation of eye tissue. This is the first demonstration that the repression of wg by the JAK/STAT pathway is an important mechanism in regional specification of the eye imaginal disc. Furthermore, this is the first report of a functional interaction between the JAK/STAT and Wnt signaling pathways, and suggests the possibility that a similar interaction exists in higher organisms.

The following stocks are described in FlyBase: stat92E^85C9; stat92E³⁹⁷; stat92E⁰⁶³⁴⁶; wg^CX3 dpp-lacZ; wg-lacZ (wg^P); UAS-upd; UAS-hop; pnr-GAL4, UAS-gfp (pnr>GFP); UAS-gfp; yw; P{AyGAL4}25 P{UAS-GFP.S65T}T2; hs-flp MKRS/TM6B; upd-GAL4; ey-GAL4; hop^M13; ey-FLP; UAS-flp; P{neoFRT}^82B P{Ubi-GFP(S65T)_nls}3R/TM6B, Tb¹; P{neoFRT}^82B P{arm-lacZ.V}83B/TM6C,Sb¹ Tb¹; P{neoFRT}^82B M(3)96C,arm-lacZ; P{neoFRT}^82B RpS3(1), Ubi-GFP; and P{neoFRT}^82B ry⁵⁰⁶. We used 10XSTAT92E-GFP (Bach et al.,2006) and wg2.11Z(Pereira et al., 2006). All crosses were maintained at 25°C.

We used PCR to generate a full-length stat92E cDNA with BamHI (5′) and NotI (3′) ends using the following primers (restriction sites underlined):5′-CGCGGATCCATGAGCTTGTGGAAGCGCATCGCCAGCC-3′ and 5′-TTTCCTTTGCGGCCGCAAAGTTCTCAAAGTTGTAATCGTATCG-3′. After digestion with BamHI and NotI, the insert was ligated into the pBluescript KS-based vector C5HA3. The 5′ end of the C5HA3 polylinker includes an ATG immediately upstream of sequence encoding three HA epitopes, followed by an in-frame BamHI site and the costal2cDNA (a kind gift of Kent Nybakken, Harvard Medical School, Boston, MA, USA). At the 3′ end, a NotI site immediately precedes and is in frame with a stop codon. The C5HA3 plasmid was digested with BamHI and NotI to remove the costal2 sequence. The 3HA-stat92E insert was excised from C5HA3 by digestion with BssHII. The 3′ recessed termini were filled in with Klenow. This fragment was ligated into UASp(Rorth, 1996) that had been cut with BamHI and filled in with Klenow. UASp-3HA-stat92Etransgenic flies were generated as described(Bach et al., 2003).

Clones were generated using the FLP/FRT system(Xu and Rubin, 1993). ey-flp or ey-GAL4, UAS-flp (EGUF) transgenic animals express FLP in the eye disc(Newsome et al., 2000). Clones were marked using the following chromosomes: P{neoFRT}^82BP{Ubi-GFP(S65T)_nls}3R/TM6B,Tb¹ or P{neoFRT}^82BP{arm-lacZ.V}83B/TM6C, Sb¹ Tb¹. Minute clones were marked by P{neoFRT}^82B M(3)96C, arm-lacZ or P{neoFRT}^82B Ubi-GFP. P{neoFRT}^82Bry⁵⁰⁶ was used as the control. hop-expressing or flip-out clones were generated using UAS-hop and P{AyGAL4}25 P{UAS-GFP.S65T}T2; hs-flp MKRS/TM6B, in which FLP is under the control of the heat-shock promoter (Ito et al.,1997). Flip-out clones express both Hop and GFP.

Sense and antisense upd riboprobes were generated as previously described (Bach et al.,2003).

Antibody stainings were performed as previously described(Bach et al., 2003). We used the following primary antibodies: rat anti-Elav (1:50), mouse anti-Wg (1:20),mouse anti-β-galactosidase (1:50), mouse anti-Discs large (Dlg) (1:50),mouse anti-Dac (1:100) (all from Developmental Studies Hybridoma Bank), rabbit anti-Hth (1:300) (a kind gift of H. Sun)(Pai et al., 1998), rabbit anti-β-galactosidase (1:100) (Cappel) and rat anti-HA (1:100) (Roche). We used fluorescent secondary antibodies (Jackson Laboratories) at 1:250. We collected fluorescent images (at 25 × magnification) using a Zeiss LSM 510 confocal microscope, scanning electron micrographs (at 100×) using a Leo SEM (Zeiss) (Harvard School of Public Health), and bright field pictures of adults (at 2.5× or 5×) and of discs (at 20×) using a Zeiss Axioplan microscope with a Nikon Digital Sight DL-UL camera.

To test whether the sole function of JAK/STAT signaling is to promote growth, or whether it also regulates patterning of the eye disc, we examined endogenous Upd expression, assessed the range of Upd activity and misexpressed Upd using the UAS/GAL4 system (Brand and Perrimon, 1993). An upd-GAL4, UAS-GFP(upd>GFP) reporter revealed that upd was expressed in cells residing next to the optic stalk in first and second instar eye discs(Fig. 1A,B). However, upd mRNA was not detected after early third instar, suggesting a function in the early eye disc (Fig. 1C,D). Upd is likely to be the only Upd-like molecule involved in eye development. A upd2-null mutant is viable and fertile and has no eye defects, whereas upd3 is not expressed in the eye disc (data not shown).

Upd is a secreted molecule that acts cell non-autonomously(Bach et al., 2003; Tsai and Sun, 2004). To assess the range of Upd activity, we generated a reporter called 10XSTAT92E-GFP that specifically reflects JAK/STAT pathway activity,as evidenced by the loss of reporter expression in stat92E clones(Bach et al., 2006). Although Upd is synthesized only by cells at the posterior midline, it has long-range effects. The 10XSTAT92E-GFP reporter was activated uniformly throughout the posterior domain of the eye to the lateral margins in first and second instar eye discs (Fig. 1E-H). Using this reporter, we demonstrated that the JAK/STAT pathway is not active in the eye disc after early third instar(Bach et al., 2006). These data indicate that JAK/STAT signaling is active only during the early stages of larval eye disc development. We next assessed the spatial and temporal relationship between JAK/STAT signaling and other pathways, such as Dpp and Wg, that are also active in early eye development. We used a dpp-LacZreporter and a wg-lacZ enhancer trap (wg^P) that faithfully recapitulate expression of the endogenous genes(Blackman et al., 1991; Kassis et al., 1992). In early larval eye development, the 10XSTAT92E-GFP reporter partially overlapped with dpp but abutted the wg expression domain,suggesting functional interactions between these pathways(Fig. 1E-H).

To test whether the restricted expression of Upd is functionally important,we ectopically misexpressed upd throughout the early eye disc using ey-GAL4. ey>upd adults exhibited overgrowth of the eye field,presumably due to the mitogenic functions of Upd, as well as aberrant patterning of the head cuticle (Fig. 1J). Specifically, lateral head cuticle was replaced by photoreceptors, thereby reducing the distance between the eyes, a phenotype previously reported for loss of wg signaling(Fig. 1J)(Ma and Moses, 1995). These data indicate that the early, restricted expression of Upd to the posterior midline is essential for proper formation of the eye field.

The consequences of loss of Stat92E during eye development have not been reported. To address this issue, we generated mutant stat92E clones in the eye using the stat92E^85C9 allele, which results from an R to P substitution at position 442 in the DNA-binding domain(Silver and Montell, 2001). The Stat92E^85C9 protein is nonfunctional, as evidenced by the lack of activation of the 10XSTAT92E-GFP reporter in stat92E^85C9 clones (data not shown)(Bach et al., 2006). We induced stat92E^85C9 mutant clones in the eye disc using ey-flp, in which the yeast FLP recombinase is expressed under the control of the eye-specific enhancer of ey(Newsome et al., 2000). To generate large patches of stat92E mutant tissue we also used a Minute mutation, which causes slow growth and recessive lethality in cells possessing the wild-type chromosome(Morata and Ripoll, 1975). When used with ey-flp and a mutation on an FRT chromosome, the Minute technique produces an eye composed almost entirely of homozygous mutant tissue. Adults carrying a control chromosome in a Minute background (hereafter referred to as + M⁺)did not exhibit pupal lethality or eye or head cuticle defects(Fig. 2A,I). By contrast, less than 10% of animals carrying stat92E^85C9 clones in a Minute background (hereafter referred to as stat92E M⁺) eclosed from their pupal cases(Fig. 4G). The few adults that eclosed exhibited a small or ablated eye and abnormal head cuticle(Fig. 2B,C). Instead of the smooth head cuticle seen in control animals, stat92E^85C9M⁺ adults displayed a ridged cuticle that had a disorganized bristle pattern (Fig. 2A-C). Non-eclosed stat92E^85C9 M⁺ pupae were frequently headless or exhibited severe eye defects and expansion of the head cuticle (Fig. 2J,K). Similar phenotypes and significant hatching defects were observed with another stat92E hypomorphic allele, stat92E³⁹⁷, whereas less severe phenotypes were observed with stat92E⁰⁶³⁴⁶(Fig. 2F,G). We also observed defects in stat92E clones in a non-Minute background. Frequently, stat92E clones within the eye field developed into ectopic head cuticle or formed ambiguous outgrowths(Fig. 2H). The stat92Emutant eye and head cuticle phenotypes were due specifically to the loss of Stat92E function. Expression of a wild-type stat92E transgene(UAS-3HA stat92E) fully rescued all stat92E mutant phenotypes (Fig. 2D). However,misexpression of this transgene (ey>3HA-stat92E) in a wild-type background did not have a phenotype(Fig. 2E). Taken together,these data suggest that Stat92E, activated by a Upd/Dome/Hop signal, promotes the formation of the eye field. In the absence of Stat92E, cells which are normally part of the eye field form other tissues.

To determine if the stat92E phenotypes observed in the adult arise during larval eye development, we examined the effect of loss of stat92E activity in the eye imaginal disc. Indeed, stat92E^85C9 M⁺ eye discs display morphological defects consistent with adult stat92E mutant phenotypes. The dorsal domain in stat92E^85C9 M⁺ eye discs was morphologically abnormal and was marked by protrusions, an elongated appearance and an ectopic eye field (Fig. 3B,C,E and data not shown). In 5% of stat92E^85C9M⁺ discs, the morphogenetic furrow failed to initiate (see Fig. S1 in the supplementary material). In about 85% of these discs, the furrow moved only through the ventral portion of the eye disc(Fig. 3B,C). Ommatidia in stat92E^85C9 M⁺ adult eyes are almost entirely of ventral origin. In the eye disc, dorsal cells express the homeodomain protein Mirror (Mirr) (McNeill et al.,1997). The lack of dorsal ommatidia in stat92E mutants is not due to altered expression of mirr, but due rather to an inability of the furrow to move through the dorsal, mirr-positive cells in the stat92E^85C9 M⁺ eye disc (see Fig. S1 in the supplementary material).

Many larval and adult phenotypes observed in the absence of stat92E are characteristic of either loss of dpp or increased wg expression. As the expression patterns of dppand wg are consistent with regulation by the JAK/STAT pathway(Fig. 1E-H), we next asked if the loss of stat92E alters expression of either gene. In wild-type third instar eye discs, dpp is expressed in the morphogenetic furrow(Heberlein et al., 1993; Ma et al., 1993). However, in stat92E M⁺ eye discs, dpp was lost from the dorsal region of the disc, whereas its ventral expression appeared normal(Fig. 3A-C). wg is expressed in dorsal and ventral margin cells, located between the peripodial membrane and the disc proper, and in cells adjacent to them in wild-type eye discs (Fig. 3D)(Cho et al., 2000). By contrast, wg was expanded throughout the disc epithelium in stat92E M⁺ eye discs, as well as in eye discs hemizygous for the amorphic mutation hop ^M13(Fig. 3E,G and data not shown). wg was also ectopically expressed in mosaic stat92E clones(Fig. 3F). The increased expression domain of the wg gene in discs lacking JAK/STAT signaling led to a dramatic increase in Wg protein(Fig. 3C). In stat92E M⁺ eye discs expressing a stat92E transgene, ectopic Wg protein was not observed (data not shown). The presence of a ventral furrow in stat92E M⁺ eye discs is consistent with high dorsal levels of Wg, which is observed with ectopic clonal expression of wgin the eye disc (data not shown) (Treisman and Rubin, 1995).

In wild-type adults, wg is expressed around the periphery of the eye and in the lateral head cuticle (Fig. 3H) (Heslip et al.,1997). However, ectopic wg was apparent in stat92E clones within the adult eye field, often at the base of ectopic outgrowths (Fig. 3I). Similar outgrowths are observed in shaggy clones, which behave as if they have transduced the Wg signal (Heslip et al., 1997). The hypothesis that JAK/STAT signaling functions upstream of Wg is further supported by the observation that the stat92E ablated eye phenotype was identical to that observed in ey>wg adults, in which wg was expressed throughout the developing eye (compare Fig. 2C with 2L). Although these results indicate that JAK/STAT signaling is an important regulator of wg expression, wg was not ectopically expressed in all cells lacking stat92E(Fig. 3E-G,I). This may reflect the time of induction of stat92E clones, which are constantly generated in an ey-flp background. Alternatively, another factor may be acting redundantly with Stat92E to regulate wg.

stat92E mutants exhibit phenotypes characteristic of expanded wg expression. However, the antagonistic relationship between dpp and wg raises the possibility that Stat92E regulates dpp, and wg expansion occurs as a secondary effect. We therefore looked at the ability of ectopic JAK/STAT signaling to either induce dpp or repress wg. Ectopic expression of hop using the flip-out technique resulted in ligand-independent, autonomous activation of Stat92E. Ectopic hop clones did not induce dpp in the eye disc (Fig. 4A). However, they did autonomously repress wg in both the dorsal and ventral eye(Fig. 4B). Therefore, JAK/STAT signaling regulates wg, and dpp is altered as a consequence. This is further supported by the ability of upd expressed throughout the eye disc (ey>upd) to repress wg. In ey>upd discs, dorsal wg was completely eliminated and ventral wg was substantially repressed(Fig. 4C). Furthermore, ey>upd discs exhibited precocious furrow initiation from the lateral margins, a phenotype also observed in wg mutants(Fig. 4C)(Treisman and Rubin, 1995). This result is consistent with the reduction in the inter-eye distance displayed by ey>upd adults(Fig. 1J). To determine if JAK/STAT regulation of wg is specific to the eye disc, we examined hop flip-out clones in other imaginal discs. Indeed, hopexpressing clones also repressed wg in the notal region of the wing disc, indicating that the JAK/STAT pathway can repress wg in other tissues (Fig. 4D).

To test whether the effects of stat92E loss-of-function are due to ectopic wg, we asked whether reducing the dose of wg, using the hypomorphic mutation wg^CX3, could rescue stat92E mutant phenotypes. As predicted, reducing the dose of wg significantly increased the hatching rate of stat92E^85C9 M⁺ flies from 9.7% to 31.2%(Fig. 4G). Furthermore, eclosed wg^CX3/+; stat92E^85C9 M⁺flies showed less severe phenotypes when compared with eclosed +/CyO;stat92E^85C9 M⁺ flies, which carry two functional copies of the wg gene (Fig. 4E,F). Taken together, these data indicate that Stat92E normally represses wg and that ectopic wg expression significantly contributes to the stat92E mutant phenotype.

Although the majority of functions attributable to STATs involve transcriptional activation, at least one STAT protein, Dd-STATa, acts as a functional repressor (Mohanty et al.,1999). We therefore assessed the ability of Stat92E to directly repress wg. A reporter called wg2.11Z, in whichβ-galactosidase is driven by a 263 base pair enhancer from the 3′ cis wg genomic region, was sufficient to recapitulate wgexpression in the dorsal margin of the disc proper and in the dorsal peripodial membrane (Fig. 5A,B)(Pereira et al., 2006). This reporter was ectopically expressed in mosaic stat92E clones as well as in stat92E M⁺ clones in a manner similar to that observed for wg^P (Fig. 5C,D). Moreover, wg2.11Z was repressed autonomously in hop-expressing clones (Fig. 5E). These data indicate that Stat92E can regulate dorsal wg expression through the wg2.11Z enhancer. wg2.11Zdoes not contain any Stat92E binding sites (TTC(N)₃GAA), strongly suggesting that Stat92E does not repress dorsal wg directly, but rather regulates another factor which represses wg(Hou et al., 1996; Yan et al., 1996).

Expression of the wg2.11Z enhancer is maintained by pnrin dorsal peripodial cells (Pereira et al., 2006). We therefore assessed whether pnr expression is altered in the absence of JAK/STAT signaling using a pnr>gfpreporter, which mimics expression of the endogenous pnr gene(Singh and Choi, 2003). In wild-type eye discs, pnr was expressed in dorsal anterior peripodial cells and in dorsal margin cells in the disc proper(Fig. 6E). In hop^M13 eye discs, pnr expanded throughout the dorsal posterior peripodial membrane, whereas it remained unchanged in the disc proper (Fig. 6F). These data suggest that Stat92E negatively regulates expression of pnr in peripodial cells. To test this hypothesis, we assessed pnr>gfpexpression when JAK/STAT signaling was ectopically activated. We used the GMR-upd transgenic line, in which ectopic Upd activates the 10XSTAT92E-GFP reporter in anterior cells adjacent to the furrow(Bach et al., 2006; Bach et al., 2003). In mid-third-instar eye discs, pnr was expressed in peripodial cells that reside `above' dorsal cells both anterior and posterior to the furrow(Fig. 6G). By contrast, in GMR-upd discs, the pnr expression domain was significantly repressed and now resided only `above' dorsal cells anterior to the furrow(Fig. 6H). These data suggest that when stat92E function is lacking in peripodial cells, pnr is ectopically expressed and may induce wg expression in these cells via the wg2.11Z enhancer.

To determine whether other known regulators of wg are altered in stat92E mutant discs, we examined the expression patterns of Dac, Hth and the Paired-domain protein Eyegone (Eyg). Although Eyg negatively regulates wg expression, we found that Eyg expression was normal in mosaic stat92E clones (data not shown)(Hazelett et al., 1998). Dac and Wg mutually repress each other's expression(Baonza and Freeman, 2002; Treisman and Rubin, 1995). Dac expression was reduced in stat92E M⁺ discs; however, its loss did not coincide with the boundary of stat92E clones but rather with regions of the eye disc that contained high levels of ectopic wg(Fig. 6A,B). This suggests that the loss of Dac expression is an indirect consequence of ectopic wg,as has been reported previously (Baonza and Freeman, 2002). The observation that Dac was not induced by clones expressing ectopic hop, strongly supports the conclusion that Dac expression is lost as a result of the ectopic Wg expressed in stat92E M⁺ discs (data not shown). Wg positively regulates expression of the hth gene. In addition, ectopic Hth can increase expression of existing wg in both the dorsal and ventral margins of the eye disc(Pichaud and Casares, 2000). Hth was expressed at the anterior edge of the third instar eye disc(Fig. 6C). By contrast, in stat92E M⁺ discs, Hth expanded into the posterior dorsal and ventral domains (Fig. 6D). Clones expressing ectopic hop cannot induce Hth expression,suggesting that the effects on hth are an indirect consequence of the ectopic Wg observed when JAK/STAT signaling is impaired (data not shown).

In this paper, we report a new role for the Drosophila JAK/STAT pathway. Our study demonstrates for the first time that JAK/STAT signaling promotes formation of the eye field through repression of wg gene transcription in both the dorsal and ventral halves of the eye disc epithelium(Fig. 6I). By monitoring Upd expression and activity, we have shown that the JAK/STAT pathway is normally activated early in eye development, during first and second instar. Ectopic activation of this pathway leads to abnormal patterning of the head capsule and a reduction in the inter-eye distance through an increase in dorsal ommatidia. By contrast, loss of activity of this pathway, using strong hypomorphic stat92E mutations, frequently resulted in the development of a rudimentary head. When the head capsule did form, stat92Emutants often had small or ablated adult eyes and excessive head cuticle. wg was ectopically expressed in stat92E clones and hop mutant eye discs, and was repressed by ectopic activation of the JAK/STAT pathway. Reduction in the dose of wg partially rescued stat92E mutants by increasing the rate of eclosion and by mitigating the phenotypes of stat92E mutant animals. Lastly, we showed that wg regulation by the JAK/STAT pathway is independent of the known wg regulators Eyg, Dac, Hth and Pnr.

Our results conflict with those of a previous study, which reported that JAK/STAT signaling does not repress wg in the eye disc. This conclusion was reached on the basis of wild-type Wg protein expression in eye discs that contained ectopic upd-expressing clones(Zeidler et al., 1999). However, we found that in the absence of stat92E, wg was ectopically expressed in both dorsal and ventral halves of the eye disc. It is likely that our examination of the wg gene using the wg^Penhancer trap is a more sensitive measure of wg expression than monitoring Wg protein. Zeidler and colleagues also reported that the JAK/STAT pathway negatively regulates mirr expression. This conclusion was drawn after finding a preponderance of dorsal, mirr-positive ommatidia in adult eyes containing unmarked upd loss-of-function clones (Zeidler et al., 1999). However, using marked clones, we showed that Mirr is expressed normally in eye tissue that is largely homozygous mutant for stat92E. Moreover, we demonstrated that stat92E M⁺ adult eyes are largely composed of Mirr-negative ommatidia, which indicates their ventral origin. Thus, our data indicate that mirr is not regulated by JAK/STAT pathway activity.

Previous work has shown that the 3′ cis region of the wg gene regulates its expression in imaginal discs. Several wg mutations that specifically affect imaginal disc development, as well as discspecific enhancers, map to this region(Baker, 1988; Couso et al., 1993; Neumann and Cohen, 1996; Pereira et al., 2006; van den Heuvel et al., 1993). In this study, we showed that Stat92E negatively regulates dorsal wgthrough a small enhancer (wg2.11Z) in the 3′ cisgenomic region of the wg gene. This enhancer is ectopically expressed in stat92E and hop mutants and is autonomously repressed by ectopic activation of Stat92E. The DNA binding preferences of Stat92E and other STAT proteins have been well characterized(Hou et al., 1996; Seidel et al., 1995; Yan et al., 1996). Because there are no Stat92E binding sites in the wg2.11Z enhancer, we favor the interpretation that Stat92E does not directly repress dorsal wgbut rather acts through another factor. This repressor may be encoded by a direct Stat92E target gene, because wg is autonomously repressed by the JAK/STAT pathway. However, we cannot rule out the possibility that Stat92E regulates wg through other transcription factors, such as Dorsal or vHNF-4, which have putative sites in wg2.11Z(Pereira et al., 2006). It is also possible that there are cryptic Stat92E binding sites in this wgenhancer, through which Stat92E may directly repress wg. Additional experiments will be needed to test these possibilities.

We also demonstrated that Stat92E represses ventral wg in the eye disc epithelium. This is presumably independent of the wg2.11Zenhancer, which recapitulates wg expression in the dorsal but not the ventral eye disc (Pereira et al.,2006). Moreover, we showed that Stat92E negatively regulates pnr in peripodial cells. In the absence of JAK/STAT signaling, pnr is dramatically expanded into the posterior peripodial membrane. However, we stress that because pnr is an intracellular protein, the ectopic pnr in the peripodial membrane cannot account for the ectopic wg observed in the disc proper of stat92E mutants. Currently, we do not know whether Stat92E regulates wg in the ventral eye disc epithelium and the peripodial membrane in the same manner as in the dorsal eye. All three wg expression domains may be regulated by the same as yet unidentified factor. Alternatively, Stat92E may regulate wg expression domains through different mechanisms. For example,dorsal wg may be regulated indirectly, whereas ventral and peripodial wg may be regulated directly by Stat92E. The wg gene 3′ cis genomic region contains one putative Stat92E binding site, which resides downstream of the wg2.11Z enhancer. Therefore, it is possible that Stat92E regulates ventral and peripodial wg through this site. Future work will be needed to address these issues.

Our study raises the possibility that the JAK/STAT pathway negatively regulates expression of vertebrate Wnts. Interestingly, both pathways are active in the developing vertebrate eye. One recent study indicates that activation of canonical Wnt signaling in the developing chick eye inhibits retinal progenitor gene expression and the differentiation of retinal neurons(Cho and Cepko, 2006). This led to the proposal that canonical Wnt signaling is required for peripheral eye development, which is similar to the role of wg in peripheral eye tissue in Drosophila. In contrast to Wnt signaling, the roles of the JAK/STAT pathway in the development of the vertebrate eye are poorly understood. However, numerous cytokines that activate this pathway, such as ciliary neurotrophic factor, are expressed in the developing retina and induce proliferation of lens cells and the expression of glial intermediate filament protein (Potts et al., 1998; Wang et al., 2002). It is therefore possible that, in mammals, activation of the JAK/STAT pathway in the developing retina inhibits Wnt expression, thus promoting retinal progenitor gene expression and the differentiation of retinal neurons.

We thank Jessica Treisman, Fernando Casares, Craig Micchelli, Doug Harrison, Martin Zeidler, Kwang-Wook Choi, Amit Singh, Norbert Perrimon, James Hombria, Herve Agaisse, Utpal Banerjee and the Bloomington Stock Center for fly stocks; Kent Nybakken for the C5HA3 plasmid; and Rebecca Stearns (Harvard School of Public Health) for use of the Leo SEM. We are grateful to Henry Sun for Hth antibodies and for communicating results prior to publication. We are indebted to two anonymous reviewers, to Jessica Treisman, Ram Dasgupta and members of the Bach and Dasgupta labs for insightful comments. L.A.E. was supported in part from an NIH institutional training grant (T32 GM066704-03);E.A.B. was supported in part by a Young Investigator Award from the Breast Cancer Alliance and a Whitehead Fellowship for Junior Faculty.