A Sensitized Genetic Screen to Identify Novel Regulators and Components of the Drosophila Janus Kinase/Signal Transducer and Activator of Transcription Pathway

Corresponding author: Erika A. Bach, New York University School of Medicine, 550 First Ave., New York, NY 10016., erika.bach{at}med.nyu.edu (E-mail)

Communicating editor: K. ANDERSON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The JAK/STAT pathway exerts pleiotropic effects on a wide range of developmental processes in Drosophila. Four key components have been identified: Unpaired, a secreted ligand; Domeless, a cytokine-like receptor; Hopscotch, a JAK kinase; and Stat92E, a STAT transcription factor. The identification of additional components and regulators of this pathway remains an important issue. To this end, we have generated a transgenic line where we misexpress the upd ligand in the developing Drosophila eye. GMR-upd transgenic animals have dramatically enlarged eye-imaginal discs and compound eyes that are normally patterned. We demonstrate that the enlarged-eye phenotype is a result of an increase in cell number, and not cell volume, and arises from additional mitoses in larval eye discs. Thus, the GMR-upd line represents a system in which the proliferation and differentiation of eye precursor cells are separable. Removal of one copy of stat92E substantially reduces the enlarged-eye phenotype. We performed an F₁ deficiency screen to identify dominant modifiers of the GMR-upd phenotype. We have identified 9 regions that enhance this eye phenotype and two specific enhancers: C-terminal binding protein and Daughters against dpp. We also identified 20 regions that suppress GMR-upd and 13 specific suppressors: zeste-white 13, pineapple eye, Dichaete, histone 2A variant, headcase, plexus, kohtalo, crumbs, hedgehog, decapentaplegic, thickveins, saxophone, and Mothers against dpp.

THE Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway is a phosphotyrosine-driven signaling system that responds to extracellular cues and triggers specific responses in the nucleus within minutes of activation (LEVY and DARNELL 2002 ). Extracellular ligands bind to and induce multimerization of cell-surface cytokine receptors, which constitutively associate with nonreceptor protein tyrosine kinase JAKs. Upon receptor activation, the JAKs are activated by auto- or transphosphorylation, and they in turn phosphorylate and activate a class of latent cytosolic transcription factors, STATs, at the plasma membrane. Activated STATs translocate to the nucleus and induce transcription of target genes. The JAK/STAT pathway is evolutionarily conserved and plays important roles in many biological processes in both vertebrates and invertebrates (ZEIDLER et al. 2000 ; LEVY and DARNELL 2002 ). Moreover, mutations in JAK and STAT genes cause cancer and immune deficiency in humans (RUSSELL et al. 1995 ; LACRONIQUE et al. 1997 ). Discovered as a key signaling pathway of cytokine receptors, the JAK/STAT pathway has been extensively characterized biochemically in mammalian tissue culture systems (BACH et al. 1997 ; LEVY and DARNELL 2002 ; O'SHEA et al. 2002 ). However, a systematic genetic approach to identify new components and regulators of the JAK/STAT pathway has lagged behind biochemical ones. The redundancy of this pathway in mammals, which have four JAK and seven STAT genes, makes a genetic approach difficult in this system (LEVY and DARNELL 2002 ). However, in the fruit fly Drosophila, which has only one JAK and one STAT gene, a genetic approach is feasible (ZEIDLER et al. 2000 ).

There are currently four key members of the Drosophila JAK/STAT pathway: a secreted ligand, Unpaired (Upd), also called Outstretched (Os; HARRISON et al. 1998 ; SEFTON et al. 2000 ); a cytokine-like receptor, Domeless (Dome; BROWN et al. 2001 ), also called Master of marelle (Mom; CHEN et al. 2002 ); a nonreceptor, cytosolic tyrosine Janus kinase Hopscotch (Hop; BINARI and PERRIMON 1994 ); and a STAT Stat92E (formerly known as Marelle; HOU et al. 1996 ; YAN et al. 1996 ). Upd biochemically activates and genetically interacts with Hop (HARRISON et al. 1998 ). Dome has similar overall structure and low but significant homology to gp-130 and leukemic inhibitory factor receptor (HOMBRIA and BROWN 2002 ). Dome interacts genetically with stat92E and has been shown to associate with Upd when both are expressed in mammalian cells (BROWN et al. 2001 ; CHEN et al. 2002 ). In mammals, protein inhibitors of activated STATs (PIAS) and suppressor of cytokine signaling (SOCS) proteins negatively regulate the JAK/STAT pathway (LEVY and DARNELL 2002 ). Drosophila possess one PIAS homolog, DPIAS [also called Suppressor of variegation 2-10 (Su(var)2-10) and zimp], that interacts genetically and biochemically with the JAK/STAT pathway (CHUNG et al. 1997 ; MOHR and BOSWELL 1999 ; BETZ et al. 2001 ; HARI et al. 2001 ). Drosophila also have three SOCS genes, but no mutations in any of them have been reported (HOMBRIA and BROWN 2002 ; HOU et al. 2002 ). The expression of one of them, SOCS36E, depends on the activity of the JAK/STAT pathway, thus making it a reporter for activation of the pathway (CALLUS and MATHEY-PREVOT 2002 ; KARSTEN et al. 2002 ).

In Drosophila, the JAK/STAT pathway is involved in sex determination, stem cell renewal in the male germline, border cell migration and stalk cell development in oogenesis, embryonic segmentation, tracheal development, larval hematopoiesis, and ommatidial rotation (BINARI and PERRIMON 1994 ; HARRISON et al. 1995 ; HOU et al. 1996 ; YAN et al. 1996 ; LUO et al. 1999 ; ZEIDLER et al. 1999 ; SEFTON et al. 2000 ; BROWN et al. 2001 ; KIGER et al. 2001 ; SILVER and MONTELL 2001 ; TULINA and MATUNIS 2001 ; BECCARI et al. 2002 ; CHEN et al. 2002 ; MCGREGOR et al. 2002 ).This plethora of biological outcomes is mirrored in the mammalian system, where biochemistry and gene targeting experiments have demonstrated a role for this pathway in numerous processes, including embryonic development, neuronal survival, and development of the immune system and immune responses (reviewed in LEVY and DARNELL 2002 ; O'SHEA et al. 2002 ).

To identify regulators and components of the Drosophila JAK/STAT pathway, we have generated a transgenic Drosophila line (GMR-upd) that ectopically overexpresses the ligand Upd in the developing eye-imaginal disc. Overexpression of Upd in the developing eye results in an enlarged eye, which is a phenotype that is easy to score visually and that can be used to screen enhancers and suppressors of the activation of the JAK/STAT pathway. To verify this, we found that the hyperactive JAK/STAT pathway in GMR-upd can be modulated by changes in the genetic dose of other known components of the pathway, making GMR-upd a sensitized genetic background for this pathway. The methodology we have used has proven highly successful in the dissection of signal transduction pathways, for example, the sevenless and the ras pathways (SIMON et al. 1991 ; THERRIEN et al. 2000 ). We performed a sensitized screen to identify dominant modifiers of the GMR-upd, enlarged-eye phenotype using a set of overlapping deficiencies of the Drosophila genome. We found 20 regions that suppress and 9 regions that enhance the enlarged-eye phenotype. Within these deficiencies, we identified 10 suppressors and two enhancers. We also found 3 suppressors of GMR-upd not covered by these deficiencies. In addition, we characterized the enlarged-eye phenotype to aid in understanding the mechanism of the interactions. Interestingly, we found that the GMR-upd phenotype is due to an increase in cell number and not cell size and can be modulated by the dpp pathway.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Stocks:
The deficiency kit, a set of overlapping deletions of the Drosophila genome, was obtained from the Bloomington Stock Center and has been estimated to cover 70–80% of the euchromatin of the Drosophila genome. Flies were grown on standard food at 25° unless mentioned otherwise. GMR-upd/Balancer flies were crossed to flies carrying a specific deficiency or mutation. The parents were allowed to lay eggs for 4 days and then were transferred to a new vial. In general, at least 15 progeny of the correct genotype were scored, and an interaction was significant only if most of the progeny exhibited the same phenotype (i.e., suppression or enhancement of the enlarged-eye phenotype). All stocks were crossed to GMR-upd three independent times.

Constructs:
The GMR-upd transgene was made by ligating a PCR fragment of the entire coding region of upd with EcoRI (5') and StuI (3') ends into BSKS at the EcoRI and HincII sites to generate BSKSupd3'. The lack of mutations in the upd3' insert was verified by sequencing the entire region amplified by PCR. The upd3' insert was excised from BSKS by digestion with BssHII. The 3' recessed termini were filled in with Klenow and then the blunted insert was digested with EcoRI to generate a upd3' insert with EcoRI (5') and blunt (3') ends. This fragment was ligated into pGMR at the EcoRI and StuI sites (HAY et al. 1994 ). The resulting pGMR-upd3' plasmid was verified by restriction digest and sequencing. To obtain the GMR-upd transgenic line, the pGMR-upd3' plasmid, together with a plasmid encoding the 2-3 transposase, was coinjected into w¹¹¹⁸ embryos by standard protocol (RUBIN and SPRADLING 1983 ). The G₀ generation was crossed to w¹¹¹⁸ flies and grown at 16° until eclosion. The resulting transgenic lines, yw P[w* GMR-upd3']¹⁹/FM7 and w; P[w* GMR-upd3']²⁸/TM3, Sb¹, resulted from an insertion of the transgene into the X and third chromosomes, respectively. We utilized the yw P[w* GMR-upd3']¹⁹/FM7, hereafter called GMR-upd19, most extensively. However, to examine genetic interactions between GMR-upd and alleles on the X chromosome, we utilized the w; P[w* GMR-upd3']²⁸/TM3, Sb¹ transgene, hereafter called GMR-upd28.

Flip-out clones:
y w UAS-upd52/y w UAS upd52; hh^P30/hh^P30 were crossed to w; flipout actin Gal4, UAS-eGFP/CyO; hs-flp, MKRS/TM6B, Tb (BASLER and STRUHL 1994 ). Larvae were subjected to heat shock for 1 hr at 37° during first or second instar, and green fluorescent protein (GFP)-positive larvae were dissected 24 or 48 hr after heat shock and stained with an anti-ß-galactosidase antibody to mark hh-LacZ.

Stainings:
Dissections were performed in 1x PBS and tissues were stained with rabbit anti-ß-galactosidase (ICN; 1:200, preadsorbed), rat anti-Elav (1:50), mouse anti-Prospero (1:4), rabbit anti-phospho-histone3 (1:200; Upstate Biotechnology, Lake Placid, NY) or Alexa Fluor 568-conjugated phalloidin (1:100; Molecular Probes, Eugene, OR). Elav and Prospero antibodies were obtained from the Developmental Studies Hybridoma Bank. Secondary antibodies (1:200) were obtained from the Jackson lab. Stained tissues were mounted by the SLOWFADE light antifade kit (Molecular Probes) and analyzed on a Leica LSM NT confocal microscope (Department of Genetics, Harvard Medical School) or an LSM510 Zeiss confocal microscope (Pharmacology Department, NYU School of Medicine). In situ hybridization was performed as described in HAUPTMANN and GERSTER 2000 . X-gal staining was performed as described in HAZELRIGG 2000 . Samples for in situ and X-gal stainings were developed on the same day, using the same probe and for the same length of time and were analyzed on an Axiophot 2 compound microscope.

Adult sections:
Newly eclosed flies were fixed in osmium tetroxide as described in WOLFF 2000 . Sections of 1 µm were cut and mounted on microscope slides. The sections were analyzed using a phase 3 condenser on an Axiophot compound microscope at x63 under immersion oil.

Scanning electron microscopy:
Adult flies were dehydrated in ethanol, subjected to drying and sputter coating, and analyzed on an Amray 1000a SEM (Cambridge Instruments) or a Leo SEM (Zeiss), both at the Harvard School of Public Health, or a JEOL 840 model (Department of Cell Biology, NYU School of Medicine).

Inverse PCR:
Inverse PCR was performed as described in HUANG et al. 2000 . PCR products were sequenced by the Biopolymer Facility at the Howard Hughes Medical Institute, Harvard Medical School, and aligned with Drosophila genomic sequences using BLAST.

Flow cytometry:
Collections of embryos and staining and flow cytometric analysis of the cell cycle were performed as described in NEUFELD et al. 1998 using a Becton Dickinson FACSvantage. We isolated GFP-positive larvae, dissected the eye-antennal discs, removed the antennal discs, and dissociated and stained only eye-imaginal disc cells. The statistics for each fluorescence-activated cell sorter (FACS) experiment are independent (see NEUFELD et al. 1998 ) and hence are presented separately, rather than as a meta-analysis. The results in Fig 6 are representative of three individual experiments.

View larger version (62K): In this window In a new window Download PPT slide   Figure 1. The JAK/STAT pathway controls size of the adult eye. Scanning electron micrographs of a WT eye (A). Heteroallelic combinations of upd (w os/y w os1A) result in a small eye (B). Ectopic misexpression of Upd using ey-Gal4II (C) or directly using a transgene GMR-Upd (D) results in an enlarged eye. In A–D, anterior is to the left and posterior to the right; dorsal is up and ventral is down. Scanning electron micrographs taken at x100 magnification.

View larger version (194K): In this window In a new window Download PPT slide   Figure 2. The upd small eye is rescued by ectopic expression of Upd. Genotypes: w os/y w os1A (A); w os/y w os1A; ey-Gal4/UAS-GFP (B); w os/y w os1A; ey-Gal4/UAS-upd (C); ey-Gal4/UAS-upd (D); w os/y w os1A; ey-Gal4/UAS-dome (E); ey-Gal4/UAS-dome (F); w os/y w os1A; ey-Gal4/UAS-hop (G); ey-Gal4/UAS-hop (H); w os/y w os1A; ey-Gal4/UAS-stat92E (I); ey-Gal4/UAS-stat92E (J); w os/y w os1A; ey-Gal4/UAS-SOCS36 (K); and ey-Gal4/UAS-SOCS36 (L). This small-eye phenotype associated with w os/y w os1A (A) can be rescued by ectopic misexpression of upd (C) to the developing eye disc but not by ectopic misexpression of GFP (B). The small eye is partially rescued by ectopic misexpression of Hop (G) but not of Stat92E (I). The small eye is exacerbated by ectopic misexpression of Dome (E) and Socs36 (K) . Ectopic misexpression of Upd (D) and Hop (H) in wild type using the ey-Gal4 II driver results in enlarged eyes, while Dome (F) generated a small eye and Stat92E (J) and Socs36 (L) had no effect. All crosses were performed at 25°, except B, which was done at 16° and H, which was performed at 20°. In A–L, anterior is to the left and posterior to the right; dorsal is up and ventral is down. Scanning electron micrographs were taken at 100x.

View larger version (96K): In this window In a new window Download PPT slide   Figure 3. GMR-upd is a sensitized genetic background. Genotypes: WT (A); GMR-upd19/+ (B); GMR-upd28/+ (C); GMR-upd19/+; stat92E06346/+ (D); GMR-upd19/+; glass3/+ (E); GMR-upd28/stat92E06346 (F); hopC111/+; GMR-upd28/+ (G); and dome217/+; GMR-upd28/+ (H). One copy of the GMR-upd transgene inserted on the first chromosome GMR-upd19 (B) or on the third GMR-upd28 (C) results in an enlarged eye. Removal of one copy of stat92E (D and F) or glass (E) suppresses the enlarged-eye phenotype. Removal of one copy of hop (G) or dome (H) moderately suppresses the enlarged-eye phenotype. Scanning electron micrographs, dorsal view, taken at x200.

View larger version (39K): In this window In a new window Download PPT slide   Figure 4. Expression of upd, dome, hh, dpp, wg in WT and GMR-upd eye discs. Expression patterns of upd (A–C), dome (D and E), dpp (H and I), and wg (J and K) were examined by in situ hybridization using RNA probes. hh (F and G) expression was monitored by X-gal staining using an enhancer trap hhP30 (hh-LacZ). WT discs (A, B, D, F, H, J) and GMR-upd19 discs (C, E, G, I, K). In WT larvae, upd is expressed at the posterior margin in first instar (disc, top left) and second instar eye discs (disc, center) (A), but not highly expressed in third instar eye discs (B). In third instar GMR-upd eye discs, upd is expressed in all cells posterior to the morphogenetic furrow (C). Dome expression is barely detectable in WT third instar eye discs (D), but is greatly upregulated in all cells anterior to the morphogenetic furrow in GMR-upd (E). We observed normal expression of hh in both WT (F) and GMR-upd (G). dpp is expressed in cells of the furrow in WT third instar eye discs (H), and its expression is slightly enhanced in GMR-upd (I). wg is expressed at the lateral margins in WT third instar eye discs (J) and is still expressed there in GMR-upd (K); however, the staining pattern is slightly enhanced. Note that in A–K, GMR-upd third instar discs are larger than WT. The positive staining observed posteriorly in B and I indicates macrophages. In A–K, anterior is to the left and posterior to the right; dorsal is up and ventral is down.

View larger version (90K): In this window In a new window Download PPT slide   Figure 5. GMR-upd eye discs have more cells than WT (A and B). Third instar eye-antennal discs were stained with an antibody to Elav (in green), which marks photoreceptors, and rhodamine-conjugated phalloidin (in red), which marks filamentous actin and hence the morphogenetic furrow. There are more Elav-positive clusters in GMR-upd (B) compared to WT (A). (C–F) Third instar eye-antennal discs were stained with an antibody to PH3 (in green), which marks cells in mitosis, and with rhodamine-conjugated phalloidin (in red) at 96 hr (C and D) and at 110 hr (E and F) AED. (C–F) Misexpression of Upd does not lead to extra rounds of cell divisions in the second mitotic wave, i.e., posterior to the furrow. However, GMR-upd discs (D) contain more mitotic cells in the region anterior to the furrow compared to WT (C). Older GMR-upd discs (F) are substantially larger than WT (E). Images of third instar eye-antennal discs taken on a confocal microscope at x20 magnification of WT (A, C, and E) and GMR-upd19/+ (B, D, and F) discs. In A–F, anterior is to the left and posterior to the right; dorsal is up and ventral is down.

View larger version (28K): In this window In a new window Download PPT slide   Figure 6. Ectopic misexpression of Upd leads to more cells in G2/M. Cell-cycle analysis by FACS on live GFPlo eye-imaginal disc cells from WT (ey-Gal4, UAS-GFP/+; thin line) or ey-upd (ey-Gal4, UAS-GFP/UAS-upd; thick line). (A) At 90 hr the cell-cycle profile and total number of cells are roughly the same in WT and ey-upd. (B and E) At 96 hr AED eye discs from ey-upd have more cells (1.5-fold more) but no distinct increase in a particular portion of the cell-cycle profile in GFPlo cells. However, there is a small but reproducible increase in the number of cells in G2/M in ey-upd discs compared to WT. (C and E) At 110 hr AED, ey-upd discs have more GFPlo cells in G2/M and have 4-fold more cells than WT (E). All FACS profiles contained at least 20,000 events. M1 represents cells in G1 phase, M2 in S, and M3 in G2/M. (D) eyGal4, UAS-GFP early third instar eye disc stained with phalloidin (red). GFP is strongly expressed in cells posterior to the furrow and faintly and in a fading pattern in cells anterior to the furrow. (E) Numeric representation of FACS profiles, percentage of GFPlo cells in G1, S, and G2/M from WT and ey-upd discs at the indicated time AED. These data were obtained from experiments repeated three independent times with similar results. In D, anterior is to the left and posterior to the right; dorsal is up and ventral is down.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The JAK/STAT pathway is involved in the establishment of eye size:
A hetero-allelic combination (w os/y w os1A, hereafter called os/os1A) of a viable upd allele (os) and a small deletion that removes the upd locus (os1A) results in a normally patterned eye that is considerably smaller than that of wild type (WT; Fig 1A and Fig B). In contrast, increased expression in the eye of an upd ortholog, the Om1E gene, in the closely related species D. anannasae, leads to an enlarged-eye phenotype (JUNI et al. 1996 ). Thus, we reasoned that ectopic misexpression of upd in the developing eye in D. melanogaster would also result in an enlarged eye. We used the Gal4-UAS system to ectopically misexpress upd in the developing eye-imaginal disc (BRAND and PERRIMON 1993 ). We employed four Gal4 drivers: eyeless-Gal4 (ey-Gal4), elav-Gal4, GMR-Gal4, and dpp-Gal4. ey-Gal4 is expressed throughout the eye disc very early in larval development and, in third instar, at high levels in cells posterior to the morphogenetic furrow and in a faint and fading pattern anterior to the furrow (HALDER et al. 1995 ; HAUCK et al. 1999 ; see also Fig 6D). elav-Gal4 and GMR-Gal4 are both expressed in cells posterior to the morphogenetic furrow (HAY et al. 1994 ; JONES et al. 1995 ). dpp-Gal4 is expressed only in the cells in the morphogenetic furrow (STAEHLING-HAMPTON et al. 1995 ). We observed enlarged eyes in flies expressing UAS-upd under the control of all four Gal4 driver lines (Fig 1C and Table 1). In all cases, the enlarged eyes have prominent outgrowths, primarily in the dorsal portion of the eye.

We also compared Gal4-mediated upd misexpression with that of upd directly under the control of the GMR promoter, since GMR has been used in many modifier screens (HARIHARAN et al. 1995 ). We therefore generated a transgene in which the coding region of upd was placed directly under the control of the GMR promoter, which contains multiple tandem binding sites for the eye-specific transcription factor Glass and which is expressed in cells posterior to the morphogenetic furrow (HAY et al. 1994 ). Animals expressing one copy of the GMR-upd transgene have greatly enlarged adult compound eyes, with dramatic dorsal outgrowths (Fig 1D). In addition, the eyes of GMR-upd flies do not appear rough, and the external morphology of the eye and the position of interommatidial bristles is relatively normal. Taken together, these data indicate that ectopic expression of Upd in the developing eye leads to a substantial increase in the size of the eye. Since we observe the same enlarged-eye phenotype using either the Gal4-UAS system or the GMR promoter, we used these two systems interchangeably in the characterization of the enlarged-eye phenotype described below, depending on which line was most convenient.

We next asked whether ectopic expression of upd in the developing eye could rescue the small-eye phenotype of os/os1A using the ey-Gal4 driver. Importantly, we rescued the small-eye phenotype in os/os1A animals using UAS-upd (Fig 2C) but not using UAS-GFP (Fig 2B). These results demonstrate that upd regulates the size of the developing eye.

Upd is a secreted molecule that can act in a cell- nonautonomous manner (HARRISON et al. 1998 ; ZEIDLER et al. 1999 ). Therefore, we wanted to determine if ectopic misexpression of cytosolic components of the JAK/STAT pathway, which presumably act cell autonomously, could also rescue the small eye in os/os1A and could generate a phenotype when expressed in wild-type flies. Using the Gal4-UAS system, we expressed UAS-dome, UAS-domeCyt, UAS-hop, UAS-hop^Tum-l, UAS-stat92E, and UAS-SOCS36E using the Gal4 drivers mentioned above. Misexpression of full-length Dome using ey-Gal4 in an os/os1A mutant does not rescue the small-eye phenotype (Fig 2E). In fact, os/os1A; ey-Gal4/UAS-dome flies actually have smaller eyes than os/os1A flies do. Expression of a full-length Dome or a cytoplasmically truncated and presumably inactive Dome (DomeCyt) in the wild-type eye discs resulted in a small-eye phenotype that looked similar to the small eye observed in os/os1A flies (Fig 2F; Table 1; data not shown). This result indicates that full-length Dome can act as a dominant-negative molecule, an observation that has been made after expressing UAS-dome in other tissues (E. A. BACH, unpublished data; S. BROWN and J. C.-G. HOMBRIA, personal communication). However, after coexpression of Upd and full-length Dome together in the developing eye, we still observed an enlarged eye (data not shown). Presumably, full-length Dome does not act as a dominant-negative when Upd is also misexpressed in the eye disc. Expression of wild-type Hop in os/os1A partially rescued the small-eye phenotype (Fig 2G), although not as well as Upd (Fig 2C). Expression of the wild-type Hop or the activated Hop^Tum-l resulted in an enlarged eye in all combinations (Fig 2H and Table 1; HARRISON et al. 1995 ; LUO et al. 1995 ). These data indicate that the growth observed by misexpression of Upd to the developing eye results from signals downstream of Hop. Ectopic misexpression of the negative regulator SOCS36E exacerbated the small-eye phenotype in os/os1A animals (Fig 2K). However, when misexpressed in wild-type animals, SOCS36E does not lead to a small-eye phenotype, which has been observed previously (Fig 2L; CALLUS and MATHEY-PREVOT 2002 ). In contrast, ectopic expression of stat92E does not rescue the small-eye phenotype in os/os1A flies (Fig 2I). In fact, ectopic misexpression of stat92E to the developing eye, using any of the Gal4 drivers, failed to produce a phenotype (Fig 2J and Table 1). This is presumably due to the misexpression of stat92E not leading to the activation of this transcription factor. This has also been observed in mammalian tissue culture experiments where overexpression of wild-type full-length STATs do not result in their activation without the addition of a stimulating ligand (BACH et al. 1997 ; DARNELL 1997 ). Nonetheless, these data indicate that the JAK/STAT pathway can control the size of the developing eye.

Similarly, we addressed whether the GMR-upd phenotype was dependent on activation of the JAK/STAT pathway. We established two independent transgenic lines, GMR-upd19/FM7 and GMR-upd28/TM3, Sb. In either line, the expression of the GMR-upd transgene does not result in embryonic lethality, and homozygous animals exhibit pupal lethality (data not shown). Animals expressing one copy of the GMR-upd transgene have a greatly enlarged adult compound eye, with significant dorsal outgrowths in GMR-upd19 and GMR-upd28 (Fig 3B and Fig C, respectively). We predicted that reduction in the dose of stat92E would modify (i.e., suppress) the GMR-upd phenotype. When we reduce by 50% the dose of stat92E, using the hypomorphic alleles stat92E⁰⁶³⁴⁶ or stat92E^jC68, there is a dramatic suppression of the enlarged-eye phenotype in both GMR-upd19 and GMR-upd28 (Fig 3D and Fig F, and data not shown). In addition, when we reduce the dose of glass, which drives the GMR promoter, using the viable glass³ allele, we also suppress the phenotype (Fig 3E and data not shown). We reduced the dose of hop, dome, and upd to assess if this would modify the enlarged-eye phenotype. The GMR-upd phenotype is moderately suppressed when we remove a copy of hop, using the null allele hop^C111, or dome, using the hypomorphic alleles dome²¹⁷ or dome⁴⁶⁸, although not to the same extent as when the dose of stat92E is reduced (Fig 3G and Fig H, respectively, and data not shown). However, a weak allele of hop, hop^msv1, does not modify the phenotype (data not shown). Reduction in the dose of upd, using the null allele upd^yc43, the strong hypomorph upd^ym55, or the os1A deficiency, does not modify the phenotype (data not shown). This is presumably because Upd is so highly expressed in GMR-upd that a reduction in the amount of endogenous upd does not modify the phenotype. Therefore, the GMR-upd phenotype is specific to activation of the JAK/STAT pathway in the developing eye.

Characterization of GMR-upd transgenic line:
In wild-type eye discs, upd is expressed in first and second instar at the posterior margin (Fig 4A). By third instar, endogenous upd expression has largely disappeared, and the observed staining in the furrow indicates macrophages (Fig 4B). In contrast, in third instar eye discs from GMR-upd19 animals, upd is expressed in all cells posterior to the morphogenetic furrow (Fig 4C). Importantly, third instar GMR-upd eye discs are larger than those of wild type (compare Fig 4C with 4B). However, first and second instar eye discs from GMR-upd are the same size as wild type (data not shown). These data demonstrate that the overgrowth observed in GMR-upd begins in third instar. Interestingly, dome is strikingly upregulated in cells anterior to the furrow in third instar GMR-upd discs (Fig 4E). In wild-type third instar eye discs, dome expression is not observed or is barely detectable (Fig 4D). These data suggest that dome is a target of the JAK/STAT pathway in the eye.

Secreted factors Hedgehog (Hh), Decapentaplegic (Dpp), and Wingless (Wg) have been shown to induce proper morphogenesis and to influence proliferation in the eye-imaginal disc (HEBERLEIN and TREISMAN 2000 ). Therefore, we wanted to investigate whether these molecules are expressed normally in third instar GMR-upd discs. In wild-type eye discs, Hh is produced by differentiated photoreceptors posterior to the furrow (HEBERLEIN et al. 1993 ; MA et al. 1993 ). We analyzed hh expression using an enhancer trap (hh^P30) and found that its expression in differentiating photoreceptors is normal in both wild-type and GMR-upd discs (Fig 4F and Fig G). In third instar, dpp is expressed in the cells of the furrow (HEBERLEIN et al. 1993 ; MA et al. 1993 ; HEBERLEIN and TREISMAN 2000 ). dpp is expressed at the correct place in GMR-upd but at slightly elevated levels compared to wild type (Fig 4H and Fig I; data not shown). The observed staining in the posterior part of GMR-upd disc is not dpp but rather macrophages (Fig 4I). In wild-type third instar eye disc, wg is expressed at the dorsal and ventral margins (HEBERLEIN and TREISMAN 2000 ). In both wild-type and GMR-upd discs, wg is expressed in its normal pattern. However, there appears to be more wg in the GMR-upd discs compared to wild type (Fig 4J and Fig K). The increased dpp and wg expression may be the by-product of a greater number of cells in GMR-upd discs. However, our previous work has shown that upd does not regulate wg expression and vice versa (ZEIDLER et al. 1999 ).

GMR-upd eyes have more cells due to increased mitoses:
We reasoned that the increased size of GMR-upd eyes could be due to an increase in cell number. This is supported by the observation that GMR-Gal4, UAS-upd/+ animals exhibit more facets than wild type exhibit (CHEN et al. 2002 ). In addition, we stained third instar eye discs from wild-type and GMR-upd animals with an antibody to Elav to mark neuronal cell fate and with phalloidin to mark filamentous actin. GMR-upd discs have more Elav-positive clusters than wild type (compare Fig 5A and Fig B). These data support the hypothesis of an increase in cell number in GMR-upd discs.

The increased numbers of cells in GMR-upd discs could arise from a decrease in apoptosis or an increase in cell division. To investigate the former, we removed one copy each of hid, reaper, and grim using the H99 deficiency (WHITE et al. 1994 ). If Upd prevents apoptosis, then removal of these apoptotic genes should result in an enhancement of the GMR-upd phenotype. However, we observed no modification of the GMR-upd phenotype when the dose of hid, reaper, and grim is reduced by 50% (data not shown). Similarly, when we ectopically misexpressed the baculovirus p35 or the caspase inhibitor DIAP1, using GMR-p35 or GMR-DIAP1, the GMR-upd phenotype was not modified (DAVIDSON and STELLER 1998 ; GOYAL et al. 2000 ; data not shown). These data suggest that a reduction in apoptosis does not account for the enlarged-eye phenotype.

We next investigated whether the enlarged-eye phenotype could be due to increased mitoses induced by Upd. In eye-imaginal disc development, there are two waves of mitosis (WOLFF and READY 1993 ). In the first mitotic wave, cells anterior to the morphogenetic furrow undergo asynchronous rounds of cell division. In the second mitotic wave, cells immediately posterior to the furrow undergo one more round of mitosis as they adopt specific cell fates. To investigate if the first or second wave of mitosis in the eye was affected by ectopic expression of upd, we stained GMR-upd or wild-type third instar eye discs with an antibody to phospho-histone 3 (PH3), which marks cells in mitosis. We examined PH3 expression at 96 and 110 hr after egg deposition (AED), which under our culture conditions corresponds roughly to middle and late third instar as assessed by the position of the furrow. At both time points, wild-type and GMR-upd discs had similar numbers of mitotic cells posterior to the furrow (Fig 5, C–F). These data indicate that the second mitotic wave is not affected by ectopic misexpression of upd to the developing eye. However, at 96 hr AED, there are more total cells in GMR-upd discs than in wild-type discs, and, importantly, there are more mitotic cells anterior to the furrow in GMR-upd discs compared to wild type (data not shown and Fig 5D). Thus, there are more undifferentiated cells to be patterned by the morphogenetic furrow in GMR-upd eye discs. At 110 hr AED, GMR-upd discs contain two to four times more cells and have more PH3-positive than wild-type cells (Fig 5F). These data suggest that in GMR-upd eye discs, Upd produced by cells posterior to the furrow can diffuse away from its production site and induce proliferation in the Dome-expressing, unpatterned cells anterior to the furrow.

We performed cell-cycle analysis by flow cytometry on live eye-imaginal disc cells (NEUFELD et al. 1998 ). We expressed upd in the developing eye disc using an ey-Gal4, UAS-GFP recombinant that we made. In this line, cells posterior to the furrow are strongly GFP-positive cells, and cells anterior to the furrow, which correspond to the more mitotic population mentioned above, are largely GFP negative and are referred to as GFP^lo (Fig 6D; HALDER et al. 1995 ; HAUCK et al. 1999 ). Thus, in discs from ey-Gal4, UAS-GFP/UAS-upd animals, the GFP-positive cells posterior to the furrow produce Upd, and we assume that Upd induces proliferation of the GFP^lo, Dome-expressing cells anterior to the furrow. We examined the cell-cycle distribution at 90, 96, and 110 hr AED. At 90 hr AED, histograms of GFP^lo cells showed similar cell-cycle distribution in ey-Gal4, UAS-GFP/+ and ey-Gal4, UAS-GFP/UAS-upd (Fig 6A and Fig E). At 90 hr AED, there are similar numbers of total eye disc cells in both genotypes (data not shown). At 96 hr AED, cell-cycle profiles of GFP^lo cells still appear similar between the two genotypes; however, there is a reproducible increase in the number of cells in G₂/M in ey-Gal4, UAS-GFP/UAS-upd compared to ey-Gal4, UAS-GFP/+: 50 vs. 55%, respectively (Fig 6B and Fig E). By 110 hr AED, GFP^lo cells from ey-Gal4, UAS-GFP/UAS-upd eye discs have more cells in G₂/M than do those from ey-Gal4, UAS-GFP/+: 46 vs. 34%, respectively (Fig 6C and Fig E). Therefore, we conclude that Upd increases the number of cycling cells in the eye disc.

GMR-upd larval eye discs and adult eyes are patterned normally:
When cells "exit" the morphogenetic furrow in wild-type third instar larvae, they receive specific signals to assume cell fates and positions within the ommatidia (WOLFF and READY 1993 ). The differentiating photoreceptors rotate 90° toward the equator, and eventually the dorsal and ventral halves of the eye form mirror images relative to the equator (Fig 7C). We used the position of the R7 cell, which expresses both Propero and Elav, within the ommatidium to assay ommatidial rotation. In wild-type and GMR-upd genotypes, the yellow R7 cell is in its expected position within the ommatidium, indicating normal rotation (Fig 7A and Fig B). We also examined adult sections to look at ommatidial rotation and photoreceptor differentiation. In wild-type discs, we observed the expected complement of photoreceptors and normal rotation of ommatidial clusters toward the equator (Fig 7D and Fig F). In GMR-upd adult sections, photoreceptor differentiation appears to be normal, although occasionally we observed the loss or gain of a photoreceptor within an ommatidium (Fig 7E). However, we did not observe a consistent loss or gain of any particular photoreceptor or support cell after analyzing eye sections of several GMR-upd animals (data not shown). We did observe abnormal ommatidial rotation in both dorsal and ventral halves of the adult eye in GMR-upd (Fig 7G), which is consistent with a previously observed role of the JAK/STAT pathway in ommatidial rotation (LUO et al. 1999 ; ZEIDLER et al. 1999 ). In addition, the adult sections have allowed us to examine the contribution, if any, of changes in cell volume to the GMR-upd phenotype. We observed no increase in cell volume of photoreceptors or their support cells in eyes from GMR-upd animals (Fig 7E). In fact, there appears to be a slight decrease in their cell volume compared to wild type, perhaps due to competition among cells for nutrients and space.

View larger version (81K): In this window In a new window Download PPT slide   Figure 7. Larval discs and adult eyes in GMR-upd animals are patterned normally. Positioning of the R7 photoreceptor in GMR-upd19 third instar eye discs occurs normally. WT (A) and GMR-upd (B) third instar eye discs were stained with antibodies to Prospero in green and Elav in red. The cells in yellow are R7 cells and cone cells and the equator has been marked manually in blue. In WT (A) and GMR-upd (B), rotation of the R7 cells occurs normally. (C) Schematic representation of larval ommatidial rotation. Sections of adult WT (D) and GMR-upd (E) animals reveal that misexpression of Upd in the developing eye does not perturb photoreceptor and secondary cell fates. Importantly, cell volume is not increased in GMR-upd (E) compared to WT (D). However, ommatidial rotation is abnormal in GMR-upd compared to wild type, which is best assessed in the schematics of the WT (F) and GMR-upd (G) adult sections. Dorsal ommatidia are represented by red and ventral by green in E and F. The equator is red in D and E and black in F and G.

Taken together, these data indicate that Upd acts as a growth factor in the developing Drosophila eye. Loss-of-function mutations in upd are associated with a small eye. Misexpression of upd to the developing eye results in a greatly enlarged eye-imaginal disc and compound eye. The enlargement is a result of an increase in the number of cells within the eye and not an increase in their volume. Moreover, although there are more cells in GMR-upd eyes, these cells appear to be patterned normally.

A deficiency screen to identify dominant modifiers of GMR-upd:
To determine how many loci in the Drosophila genome contain modifiers of the GMR-upd phenotype, we used a set of deficiency stocks from the Bloomington Stock Center that contain overlapping deletions in the Drosophila genome and crossed them to GMR-upd. Although initially we used the GMR-upd28/TM3, Sb line for our screen, the majority of the screen was conducted using the GMR-upd19/FM7. GMR-upd19/Y are observed at a low frequency, and they are sterile as they are defective in the proper development/morphogenesis of the male reproductive tract, preventing release of motile sperm (E. A. BACH and A. A. KIGER, unpublished observations). Because we used the GMR-upd19/FM7 line for most of this study, we have screened only those deficiencies on the X chromosome that are covered by a duplication on the Y (e.g., Df/DpY). To date, we have tested 166 deficiencies that together uncover 60% of the genome, almost all of the euchromatin on the autosomes, and a small portion of that on the X. We have identified 20 regions that suppress and 9 regions that enhance the GMR-upd phenotype (Table 2 and Table 3). We have also identified 21 regions that, when heterozygous in the GMR-upd background, result in lethality (synthetic lethals) prior to adult stages (data not shown). Importantly, the deficiency Df(3R)H-B79 (92B3; 92F13) that uncovers stat92E (92E11-12) behaved as a suppressor of GMR-upd, thus validating the screen (Table 2). One prediction from these results is that reduction in the genetic dose of the negative regulators DPIAS or SOCS would enhance the GMR-upd phenotype. However, a DPIAS allele Su(var)2-10⁰³⁶⁹⁷ does not interact in our screen and there are no mutations in SOCS genes (HARI et al. 2001 ; data not shown). There may be buffering of the GMR-upd phenotype at the level of feedback loops, and thus it is possible that a 50% reduction in the dose of DPIAS does not modify the enlarged-eye phenotype.

Testing candidate genes:
We tested mutations of several genes uncovered by deficiencies that control growth or survival in the imaginal eye, including ras85D, epidermal growth factor receptor, raf, corkscrew, chico, Pten, Insulin Receptor (InR), frizzled, wg, Toll, and spaeztle. However, mutations in these genes did not modify the GMR-upd phenotype (Table 4).

We then tested whether other genes uncovered by the interacting deficiencies could modify the GMR-upd phenotype. To date, we have tested >500 mutations that map to the interacting deficiencies. Df(1)64c18 (2E1-2; 3C2) uncovers l(1)3Ag, a mutation in zeste-white 13 (zw13), which also strongly suppressed GMR-upd (Table 2). Tp(3;Y)ry⁵⁰⁶-85C (87D1-2; 88E5-6; Y) acts as an enhancer in the screen and uncovers the C-terminal Binding Protein (CtBP) gene, which encodes a transcriptional corepressor. We tested two hypomorphic mutations in CtBP, one from the Bloomington Stock Center, CtBP⁰³⁴⁶³, and the other identified in a screen for epithelial morphogenesis that will be described elsewhere (M. SCHOBER and N. PERRIMON, unpublished observations). Interestingly, both mutations enhance the GMR-upd phenotype. Df(2L)J2 (31B-32A) acts as a suppressor in our screen and uncovers the pineapple eye (pie). A viable allele, pie^EB3, also suppresses the GMR-upd phenotype (Table 3). Df(3L)fz-M21 (70D2-3; 71E4-5) acts as a suppressor of GMR-upd and uncovers Dichaete (D), also called fish hook (fish; Table 2). Hypomorphic mutations in D, fish⁸⁷, and fish⁹⁶ suppress the GMR-upd phenotype. In addition, D¹, a dominant mutation, enhances the phenotype (Table 2).

In the course of trying to identify the gene(s) responsible for the enhancer activity of Df(3R)Tl-P (97A; 98A1-2), we identified a mutation, His2Av⁰⁵¹⁴⁶, in the Histone 2A variant gene at 97D2 that suppresses the enlarged-eye phenotype (Table 2 and Table 3). Therefore, we assume that Df(3R)Tl-P contains both an enhancer and suppressor of GMR-upd. We also identified a novel P-element insertion l(3)B4-3-20¹ that suppressed GMR-upd. Inverse PCR showed that this P-element was inserted in the headcase (hdc) gene at 99E. hdc is a nuclear factor required for imaginal cell development, and its expression is regulated by the transcription factor escargot (esg; STENEBERG et al. 1998 ). Interestingly, an esg allele, esg^k00606, also suppressed GMR-upd (data not shown). Df(3L)kto2 (76B1-2; 76D5) acts as a suppressor in the screen and uncovers the kohtalo (kto) gene. A hypomorphic mutation, kto¹, acts as suppressor in the screen (Table 2). Df(2R)Egfr5 (57D2-8; 58D1) suppresses the GMR-upd phenotype, and we identified two hypomorphic mutations in plexus (px), px¹ and px^k08613, which strongly suppressed the GMR-upd phenotype (Table 2). Df(3R)crb-F89-4 and Df(3R)crb87-5 act as suppressors in the screen and uncover 95D7-D11; 95F7 and the crumbs (crb) gene (Table 2). Mutations in crb, crb¹, and crb^j1B5 act as suppressors of the GMR-upd phenotype (Table 2).

Dpp pathway genes modulate GMR-upd:
Df(2L)cl-h3 (25D2-4; 26B2-5) and Df(2R)cn9 (42E; 44C) suppress the GMR-upd phenotype and uncover type I Dpp receptors thickveins (tkv) and saxophone (sax; Table 2) (BRUMMEL et al. 1994 ). Notably, hypomorphic tkv (tkv^k16713, tkv¹, tkv^04535a) and sax (sax¹, sax², sax⁴) alleles also suppressed the GMR-upd phenotype (Fig 8F and Fig G, and Table 5). Given these data, we tested other alleles in dpp pathway genes. Seven hypomorphic dpp alleles suppressed the enlarged-eye phenotype, as did a hypomorphic mutation in a type II Dpp receptor punt (put), put¹³⁵ (Fig 8E; Table 5; LETSOU et al. 1995 ; data not shown). Importantly, a null mutation in Mothers against dpp (Mad), Mad^k00237, the Co-Smad in Drosophila that transduces dpp signals, strongly suppresses the enlarged-eye phenotype to the level observed with stat92E (Fig 8D; Table 5; WIERSDORFF et al. 1996 ). However, Df(2L)JS17 (23C1-2; 23E1-2), which removes the Mad gene, did not interact in our screen and may also contain an enhancer. Importantly, another interacting deficiency, Df(3R)DG2 (89E1-F4; 91B1-B2), acts as an enhancer in our screen and uncovers the Daughters against dpp (Dad) gene (TSUNEIZUMI et al. 1997 ). Dad is a negative regulatory SMAD in Dpp signal transduction, and mutations in Dad should enhance the GMR-upd phenotype (Table 2). As expected, a hypomorphic allele of Dad, Dad¹, enhanced the enlarged-eye phenotype (Fig 8H and Table 5). Since Hh induces dpp expression in third instar eye discs, it was interesting to observe that Df(3R)23D1 (93F; 94F), which uncovers hh, acts as a suppressor in the screen (Table 2). Hypomorphic alleles of hh, hh^IJ35, and hh^G31 moderately suppressed the GMR-upd phenotype (Table 2). We also noted that Df(2R)en-B (47E3; 48A) enhances the GMR-upd phenotype and uncovers the en gene (Table 3). However, an overlapping deficiency Df(2R)en-A (47D3; 48B2) that also removes the en gene does not modify the GMR-upd phenotype (data not shown). Therefore, we assume that the enhancer uncovered by Df(2R)en-B is not en.

View larger version (163K): In this window In a new window Download PPT slide   Figure 8. dpp pathway genes modify the GMR-upd phenotype. Genotypes: WT (A); GMR-upd19/+ (B); GMR-upd19/+; stat92E06346/+ (C); GMR-upd19/+; Madl(2)k00237/+ (D); GMR-upd19/+; dpp10638/+ (E); GMR-upd19/+; tkvl(2)k16173/+ (F); GMR-upd19/+; sax1/+ (G); GMRupd19/+; Dad1/+ (H); WT (I); w os/y w os1A (J); w os/y w os1A; ey-Gal4/UAS-GFP (K); w os/y w os1A; ey-Gal4/UAS-dpp (L). Mutations in dpp (E), its receptors tkv (F) and sax (G), or the Dpp pathway positive signal transducer Mad (D) all suppress the GMR-upd phenotype. (C). Removing a copy of the Dpp pathway negative regulator Dad enhances the GMR-upd phenotype (H). The small eye in os/os1A (J) can be rescued by ectopic misexpression of upd (see Fig 2C) to the developing eye disc but not by ectopic misexpression of GFP (K) and can be only slightly rescued by ectopic misexpression of dpp (L). In A–H, scanning electron micrographs were taken at x200 magnification; in I–L, at x100. See Table 5 for more details on the genes in the dpp pathway that modify GMR-upd.

These data raise the possibility that Upd induces the hh gene. We tested this hypothesis directly by making flip-out clones of UAS-upd in a hh-lacZ genetic background. Ectopic expression of upd did not induce hh in any region of the eye disc or in the wing disc (supplemental Fig 1 available at http://www.genetics.org/supplemental/; data not shown). These data indicate that hh is not a direct target of the JAK/STAT pathway.

The GMR-upd modifiers do not alter Glass-mediated phenotypes:
We performed a secondary screen to determine whether the modifiers of GMR-upd also affected Glass-mediated transcription (supplemental Fig 2 available at http://www.genetics.org/supplemental/). GMR-hid 1M/+ flies have a small eye that is two-thirds the size of wild type and is rough and glassy in the posterior half (supplemental Fig 2A available at http://www.genetics.org/supplemental/). This phenotype is strongly suppressed by reduction in the dose of glass (supplemental Fig 2B available at http://www.genetics.org/supplemental/). Importantly, neither stat92E allele modified GMR-hid (supplemental Fig 2C and Fig D, available at http://www.genetics.org/supplemental/). Moreover, none of the enhancers and suppressors of GMR-upd behaved in a similar manner with GMR-hid. For example, mad strongly suppresses GMR-upd; however, it did not modify GMR-hid. In addition, fish alleles, which both suppress GMR-upd, actually enhance GMR-hid (supplemental Fig 2 available at http://www.genetics.org/supplemental/). The same results were obtained using another Glass-dependent eye phenotype (i.e., GMR-Gal4). Taken together, these data indicate that the modifiers identified in our screen are likely to modify JAK/STAT-dependent phenotypes rather than Glass-dependent ones.

Ectopic expression of Dpp does not rescue the upd small-eye phenotype:
We observed a consistent genetic interaction between GMR-upd and dpp pathway genes. Since dpp is slightly increased in GMR-upd discs (Fig 4I), we reasoned that Upd may directly induce expression of dpp. We found one consensus optimal Stat92E binding site in the dpp locus; however, the functional significance of this site is unknown (YAN et al. 1996 ; data not shown). We attempted to rescue the os/os1A small-eye phenotype by ectopically misexpressing dpp using UAS-dpp, an activated form of its receptor tkv using UAS-tkv^QD, or activated hh using UAS-hh-N driven by ey-Gal4 (WIERSDORFF et al. 1996 ). Ectopic misexpression of dpp or tkv^QD resulted in more eye tissue in os/os1A when compared to GFP. However, neither rescued to the extent observed with UAS-upd or UAS-hop (compare Fig 8L with Fig 2C or Fig G, and data not shown). In contrast, UAS-hh-N resulted in a smaller eye than did os/os1A with extra bristles (data not shown). Although it is possible that we did not express dpp, tkv^QD, or hh-N at the appropriate time to engender rescue of the small-eye phenotype, these results demonstrated that neither dpp nor hh-N can substitute for upd in the developing eye.

We assessed whether mutations in JAK/STAT pathway genes can modify an eye phenotype dependent on hyperactivation of the Dpp pathway. GMR-Gal4/+; UAS-tkv^QD/+ flies have rough, glassy eyes (supplemental Table 1 available at http://www.genetics.org/supplemental/). Reducing the dose of glass strongly suppressed the roughness in the eye, while reduction in the dose of mad partially modified the eye phenotype. The GMR-Gal4/+; UAS-tkv^QD/+ phenotype was not modified by reduction in the dose of stat92E, hop, upd, or dome (supplemental Table 1 available at http://www.genetics.org/supplemental/). These data indicate that the JAK/STAT pathway is not a direct target of the dpp pathway.

We also assessed whether visible dpp and upd mutants interacted genetically. Homozygous dpp^blk flies have small eyes (STAEHLING-HAMPTON et al. 1995 ). We compared the eye size in the following genotypes: os/os, os/Y, os/+, dpp^blk/dpp^blk, dpp^blk/+, os/+; dpp^blk/+, and os/Y; dpp^blk/+ (supplemental Table 1 available at http://www.genetics.org/supplemental/). As expected, os/os, os/Y, and dpp^blk/dpp^blk flies had a small-eye phenotype, while dpp^blk/+, os/+; dpp^blk/+ flies had wild-type eyes. os/Y; dpp^blk/+ flies have a small-eye phenotype identical to that observed in os/Y flies, indicating that the reduction in dose of dpp does not modify the os phenotype (supplemental Table 1 available at http://www.genetics.org/supplemental/). Taken together, these data indicate that the JAK/STAT and Dpp pathways do not directly regulate each other.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The JAK/STAT pathway controls eye size:
Our results indicate that Upd and the JAK/STAT pathway control the size of the Drosophila eye. Heteroallelic hypomorphic combinations of upd result in a small adult eye, while ectopic misexpression of upd in the developing fly eye results in a greatly enlarged eye. This phenotype is specific to activation of the JAK/STAT pathway in the developing eye because reduction in the dose of stat92E or the eye-specific transcription factor glass results in suppression of the enlarged eye. Our results suggest that ectopic misexpression of upd in the developing eye results in additional mitoses of precursor cells in the region of the eye disc anterior to the furrow. These additional cells are patterned normally by the morphogenetic furrow, resulting in increased numbers of ommatidia in GMR-upd discs.

The GMR-upd phenotype is distinct from other enlarged-eye phenotypes:
The enlarged-eye phenotype observed by ectopic misexpression of an activated form of ras85D using the ey enhancer, ey-ras^V12, is the result of ectopic R7 cells and also appears very rough (KARIM and RUBIN 1998 ). Our results indicate that the GMR-upd phenotype is distinct from the ey-ras^V12 because GMR-upd eyes are patterned normally, are not rough, and are not modified by ras85D mutations. The enlarged eyes observed with misexpression of the Drosophila InR using GMR-Gal4 results primarily from increased cell volume (BROGIOLO et al. 2001 ; BRITTON et al. 2002 ). Our results indicate that in the Drosophila eye the JAK/STAT and InR pathways do not interact, at least when ectopically misexpressed. Reduction in doses in InR pathway genes, such as InR, Pten, and chico, do not modify the GMR-upd phenotype. Moreover, the GMR-upd phenotype results from increased cell numbers, not from increased cell volume. In fact, cells in GMR-upd adult eyes actually exhibit decreased cell volumes when compared to wild type. Interestingly, the enlarged-eye phenotype in GMR-upd shares similarities with that produced as a nonautonomous effect of expression of an activated form of Notch (N^intra) in the eye, with prominent dorsal outgrowths (GO et al. 1998 ; KURATA et al. 2000 ). This observation is also interesting in light of the fact that we identify CtBP, which represses N pathway activity, as an enhancer of GMR-upd. It is possible that CtBP represses Stat92E itself or negatively regulates transcriptional coactivation by Stat92E.

Identification of modifiers of GMR-upd:
We established that the GMR-upd line is a sensitized genetic background and performed an F₁ screen for dominant modifiers of the GMR-upd phenotype using a set of overlapping deletions of the Drosophila genome. We identified 20 loci that suppress and 9 that enhance the enlarged-eye phenotype. The gene(s) in these deficiencies that are responsible for the modification of the phenotype may represent new components of or new interactors with the JAK/STAT pathway. We identified 13 mutations as Su(GMR-upd): zw13, crb, pie, D, His-2Av, kto, hdc, px, hh, dpp, tkv, sax, and Mad. In addition, we identified two mutations as En(GMR-upd): CtBP and Dad.

Identification of suppressors of GMR-upd:
zw13 interacts genetically with the meiotic kinesin-like genes nod and ncd and encodes a poorly characterized protein with RNA-recognition motifs. Therefore, Zw13 may be important in regulating upd expression. We also identified crb as a suppressor of GMR-upd. Crb is a PDZ-containing protein involved in the establishment and maintenance of apical-basal polarity in epithelia (PELLIKKA et al. 2002 ). crb may suppress the GMR-upd phenotype by altering the localization of Dome and/or Upd or the signaling output of the JAK/STAT pathway in the eye.

We identified several transcription factors as suppressors of GMR-upd: pie, D, His2Av, kto, px, and hdc. Pie is a nuclear protein that contains a PHD finger, which is a C4HC3 zinc-finger-like motif thought to facilitate chromatin-mediated transcriptional regulation (AASLAND et al. 1995 ). Eyes from pie homozygotes show irregular spacing of ommatidia, although the ommatidia have the normal array of photoreceptors (BAKER et al. 1992 ). Notably, pie homozygous flies also have held-out wings, a phenotype shared by os flies and flies that overexpress full-length Dome (LINDSLEY and GRELL 1968 ; E. A. BACH, unpublished observation). In embryonic segmentation, D directly regulates the expression of the pair-rule gene, even-skipped (eve), by binding to multiple sites located in downstream regulatory regions that direct formation of eve stripes 1, 4, 5, and 6 (MA et al. 1998 ). This overlaps with the function at Stat92E, which is needed for proper expression of eve stripes 3 and 5 (HOU et al. 1996 ; YAN et al. 1996 ). Interestingly, fish and upd share related expression patterns and phenotypes. The early expression pattern of fish is almost identical to that of upd (NAMBU and NAMBU 1996 ). Like upd, fish is also required in the hindgut, and the D held-out wing phenotype is very similar to that of os (LENGYEL and IWAKI 2002 ). His2Av belongs to the H2AZ variant subclass, which is involved in chromatin stability, chromatin remodeling, and transcriptional control (REDON et al. 2002 ). Given that mammalian STATs have been shown to mediate transcriptional changes within seconds of activation, it is possible that histone modification must be coordinated with transcriptional coactivation. Kto is the homolog of thyroid-hormone receptor associated protein (TRAP230), which was originally identified as part of the trithorax group, a large transcriptional coactivation complex (KENNISON and TAMKUN 1988 ). kto is involved in photoreceptor differentiation because homozygous mutant clones in the eye disc fail to develop into photoreceptors, although mutant cells can respond to Hh by expressing dpp (TREISMAN 2001 ). hdc encodes a nuclear factor involved in tracheal development, where it acts nonautonomously in an inhibitory signaling mechanism to determine the number of cells that will form unicellular sprouts in the trachea (STENEBERG et al. 1998 ). Interestingly, it has been recently noted that stat92E is also required in tracheal development (BROWN et al. 2001 ; CHEN et al. 2002 ). However, whether hdc and stat92E interact, if at all, in this tissue is not known, nor is it understood whether any interaction exists in the eye disc. Px is a nuclear protein that, like Pie, contains a PHD zinc finger and is involved in venation in the wing (MATAKATSU et al. 1999 ). It is not known if px mutants exhibit an eye phenotype. Clearly, future work must focus on the elucidation of any biochemical interaction between Stat92E and these transcription/nuclear factors and also whether they regulate the transcription of a common set of genes required for growth of the eye disc.

The Dpp pathway genes modify GMR-upd:
The other modifiers identified in our modifier screen are genes in the Dpp pathway, specifically dpp, tkv, sax, mad, hh, and Dad. We initially reasoned that upd may exerts its proliferative effects through hh or dpp. However, we show that hh and dpp are expressed normally in GMR-upd. In addition, we demonstrate that ectopic misexpression of hh or dpp in the os/os1A flies does not rescue the small-eye phenotype whereas upd does and that ectopic expression of upd in flip-out clones does not induce hh. These results suggest that upd may not directly regulate dpp or hh expression. These data also suggest that Upd and Dpp and/or Hh may coregulate genes involved in the proliferation of eye precursor cells. This hypothesis is supported by observations in mammalian systems. The cytokines leukemic inhibitory factor and bone morphogenic protein 2 activate Stat3 and Smad1, respectively, and act synergistically in fetal neuroepithelial cultures to promote the differentiation of astrocytes from progenitor cells. The synergism requires functional Stat3 and Smad1. However, these proteins do not physically interact; rather, they both bind to p300/CBP to promote transactivation of target genes, such as glial fibrillary acidic protein, a marker of astrocyte differentiation (NAKASHIMA et al. 1999 ).

The role of the JAK/STAT pathway in proliferation and growth control:
In both mammals and flies, the JAK/STAT pathway plays an important role in the control of organ/tissue size. Stat5 knock-out mice are runted due to impaired growth-hormone signaling (LEVY and DARNELL 2002 ). Similarly, Socs-2 knock-out mice are significantly larger than their wild-type littermates, due to a lack of negative regulation of the gr