Identification of Drosophila Genes Modulating Janus Kinase/Signal Transducer and Activator of Transcription Signal Transduction

Abstract

The JAK/STAT pathway was first identified in mammals as a signaling mechanism central to hematopoiesis and has since been shown to exert a wide range of pleiotropic effects on multiple developmental processes. Its inappropriate activation is also implicated in the development of numerous human malignancies, especially those derived from hematopoietic lineages. The JAK/STAT signaling cascade has been conserved through evolution and although the pathway identified in Drosophila has been closely examined, the full complement of genes required to correctly transduce signaling in vivo remains to be identified. We have used a dosage-sensitive dominant eye overgrowth phenotype caused by ectopic activation of the JAK/STAT pathway to screen 2267 independent, newly generated mutagenic P-element insertions. After multiple rounds of retesting, 23 interacting loci that represent genes not previously known to interact with JAK/STAT signaling have been identified. Analysis of these genes has identified three signal transduction pathways, seven potential components of the pathway itself, and six putative downstream pathway target genes. The use of forward genetics to identify loci and reverse genetic approaches to characterize them has allowed us to assemble a collection of genes whose products represent novel components and regulators of this important signal transduction cascade.

DEVELOPING cells in vivo are influenced by, and interact with, their surroundings via multiple mechanisms central to which are the signal transduction cascades. Activation of such cascades by extracellular ligands generally converts signals into changes in the gene expression profile of a responding cell. Although only a relatively limited number of such pathways have been identified, these signaling cascades have generally been conserved throughout evolution and are often active at multiple stages of development where they exert a wide range of pleiotropic effects, including cellular growth, proliferation, and differentiation.

The Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway is one such signaling cascade. The pathway was first identified in mammals where extensive analysis has led to the development of a canonical model for JAK/STAT signaling in which nonreceptor JAK tyrosine kinases are associated with the intracellular portion of trans-membrane receptors. Following ligand binding to dimerized cytokine receptors the associated JAK molecules become active and auto- and trans-phosphorylate one another and their receptors. The resulting phospho-tyrosine residues are recognized by the SH2 domain of normally cytosolic STAT proteins, which are recruited to these docking sites before being themselves phosphorylated on a C-terminal tyrosine residue by the JAKs. The activated STATs form homo- and hetero-dimers and translocate to the nucleus, bind to a palindromic DNA recognition sequence, and activate the transcription of pathway target genes (Zeidler et al. 2000a; Bromberg 2002; Kisseleva et al. 2002).

In mammals, four Jak molecules have been identified: Jak1, Jak2, Jak3, and tyrosine kinase 2(tyk2). STATs compose a family of seven structurally and functionally related proteins: Stat1–4, Stat5a, Stat5b, and Stat6. Stat proteins play a central role in transmitting cell surface cytokine signals into the nucleus and in induceing cellular proliferation, differentiation, and survival signals in multiple hematopoetic cell types. Under normal circumstances, ligand availability and negative feedback mechanisms tightly regulate the cytokine-mediated activation of Stats (Bowman et al. 2000). However, constitutive activation of multiple Jaks and Stats is associated with diverse leukemias and lymphomas (Friedmann et al. 1996; Lacronique et al. 1997; James et al. 2005), resulting in a radical alteration of the gene expression and ligand-independent survival/proliferation of these transformed cells (Sternberg and Gilliland 2004). Mechanisms required for the regulation of the JAK/STAT pathway and candidate downstream target genes, however, are comparatively poorly understood. The redundancy present in the human system with multiple activating cytokines, Jaks and Stats, makes a genetically based approach to identifying pathway interacting genes difficult (Levy and Darnell 2002). By contrast, the JAK/STAT pathway in Drosophila represents a less complex and genetically tractable system with which the pathway can be studied.

The canonical pathway in Drosophila consists of the homologous secreted ligands Unpaired (Upd), Upd2, and Upd3 (Harrison et al. 1998; Agaisse et al. 2003; Hombria et al. 2005) and a cytokine-like transmembrane receptor, Domeless (Dome; Brown et al. 2001), with homology to gp130 and the leukemia inhibitory factor receptor (Hombria and Brown 2002). Dome genetically interacts with stat92E and has been shown to associate with Upd when coexpressed in mammalian cells (Brown et al. 2001; H. W. Chen et al. 2002). A JAK tyrosine kinase called Hopscotch (Hop; Binari and Perrimon 1994) bearing 27% identity to human JAK2 and a single STAT-like transcription factor, STAT92E (also known as Marelle; Hou et al. 1996; Yan et al. 1996), with 37% identity to human STAT5b, are also encoded by the fly genome. Drosophila homologs of other components of the mammalian JAK/STAT pathway have also been identified. These include protein inhibitor of activated STAT (PIAS), suppressor of cytokine signaling (SOCS), and signal transducer adaptor molecule-like molecules. dPIAS [also known as suppressor of variegation 2-10 (Su(var)2-10) and zimp] 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). The Drosophila genome encodes three SOCS genes and the expression of socs36E, a member of the vertebrate SOCS4/5 class, is regulated by JAK/STAT pathway activity and functions to suppress JAK activity (Karsten et al. 2002; Rawlings et al. 2004), while SOCS44A, a member of the vertebrate SOCS6/7 class, is independent of JAK pathway activity but capable of repressing JAK-induced signaling (Rawlings et al. 2004).

In vertebrate systems biochemical approaches and gene-targeting experiments have identified a requirement for the pathway in diverse processes, including embryonic development, neuronal survival, development of the immune system, and hematopoesis (Levy and Darnell 2002; Boulay et al. 2003). In addition, STAT3 and STAT5 appear to represent the major STATs involved in promoting oncogenesis (Bowman et al. 2000) while Stat1 induces antiproliferative responses and functions as a potential tumor suppressor (Platanias 2005).

Many of these processes are mirrored in Drosophila. In the adult, the pathway is involved in stem cell renewal in the male germline (Kiger et al. 2001; Tulina and Matunis 2001) as well as in border cell migration and stalk cell development in oogenesis (Beccari et al. 2002; McGregor et al. 2002). In early stages of embryonic development, JAK/STAT signaling plays an important role in sex determination (Sefton et al. 2000) and regulates embryonic segmentation by controlling the expression of the pair-rule genes even-skipped, runt, and fushi tarazu (Binari and Perrimon 1994; Hou et al. 1996; Yan et al. 1996; Harrison et al. 1998). At later embryonic stages, roles in tracheal and posterior spiracle formation have been identified in dome mutants (Brown et al. 2001; H. W. Chen et al. 2002) along with a requirement in both fore- and hind-gut development (Johansen et al. 2003; Josten et al. 2004). During larval development, the JAK/STAT pathway is also required for hematopoiesis (Luo et al. 1997), ommatidial rotation in the eye (Zeidler et al. 1999a), and cellular proliferation in the wing disc. In this tissue, STAT92E exerts both early proproliferative and late antiproliferative functions (Mukherjee et al. 2005). To achieve this degree of complexity, it is probable that Drosophila JAK/STAT signaling is influenced by a range of both environmental and physical interactions, which act to modulate the consequences of its activation during different developmental processes.

A recent genetic screen to identify chromosomal regions interacting with the JAK/STAT pathway in vivo has identified a number of novel regulators and components of the pathway (Bach et al. 2003). However, such a screen allows only the identification of specific genes via a candidate approach and caveats associated with the availability of alleles and varying genetic backgrounds apply. By contrast, the ability to identify potentially interacting loci from among mutations generated by random mutagenesis represents a potentially more stringent approach. We therefore employed the P{w⁺, GMR-updΔ3′} transgenic strain previously described by Bach et al. (2003). In this stock a trans-gene containing multimerized binding sites for the eye-specific transcription factor Glass (Ellis et al. 1993) is used to drive expression of the pathway ligand Upd. Expression of Upd posterior to the morphogenetic furrow by the glass multimerised response (GMR) promoter results in increased levels of JAK/STAT pathway activity as shown by upregulation of the pathway target socs36E (Karsten et al. 2002) and increased levels of cellular proliferation shown by staining with the mitosis-specific marker phospho-Histone3 (Bach et al. 2003). Increased cellular proliferation occurs primarily in a region ahead of the morphogenetic furrow and corresponds to the first mitotic wave (Bach et al. 2003; Tsai and Sun 2004). The additional cells that result appear to differentiate normally and give rise to a greatly enlarged adult eye with overgrowth particularly apparent in dorsal regions (Figure 1, A and B). As required of a dosage-sensitive assay, the degree of eye overgrowth caused by P{w⁺, GMR-updΔ3′} is dependent on the ability of downstream JAK/STAT pathway components to transduce the overactivation and is significantly suppressed following the removal of a single copy of the stat92E locus (Figure 1C).

Figure 1.

The P{w⁺, GMR-updΔ3′} sensitized screen. One copy of the P{w⁺, GMR-updΔ3′} transgene inserted on the first chromosome results in overgrowth of the adult eye and the formation of dorsal folds (arrow in B). Removal of one copy of stat92E suppresses the enlarged-eye phenotype (C). Removal of one copy of jim results in a mild suppression (D) and loss of one copy of Brd results in a good suppression of the overgrown-eye phenotype (E). Note that the dorsal folds in B are missing in these eyes. (F) Generation of independent autosomal insertions of P{Mae-UAS.6.11} P-element (abbreviated as P{EPy⁺}). Genotypes: WT (Ore-R) eye (A); P{w⁺, GMR-updΔ3′}′/+ ; +/+ (B); P{w⁺, GMR-updΔ3′}′/+ ; stat92E⁰⁶³⁴⁶/+ (C); P{w⁺, GMR-updΔ3′}/+ ; jim^BG01625/+ (D); P{w⁺, GMR-updΔ3′}/+ ; Brd¹/+ (E).

Here we present our analysis of mutants identified in an F₁ genetic screen to identify potential modifiers of the Drosophila JAK/STAT pathway. We have screened 2267 independent autosomal P{Mae-UAS.6.11} (Crisp and Merriam 1997) insertions for their interaction with the P{w⁺, GMR-updΔ3′}-induced eye overgrowth phenotype (Bach et al. 2003). In addition to the initial identification, we further validated interacting loci using a combination of reverse genetic RNA interference (RNAi)-based approaches and in vivo expression studies. The screen identified 23 potential pathway interacting loci, including members of the Dpp and Notch signaling pathways, seven potential pathway components defined by RNAi knockdown, and six novel pathway target genes whose expression patterns appear to be modulated by changes in the JAK/STAT pathway activity.

MATERIALS AND METHODS

Genetic interactions:

For genetic interaction assays, y,w,P{w⁺,GMR-updΔ3′}/FM7, P{w⁺,Ubq-GFP} (Bach et al. 2003) virgins were crossed to males of the indicated genotypes (Tables 1 and 2). Each batch of interaction assays was grown at 25°, the “average” eye overgrowth in adult progeny of the next generation was scored in relation to positive and negative/neutral controls crossed to stat92E⁰⁶³⁴⁶ mutants and “wild type” (Ore-R or w¹¹¹⁸) lines, respectively. In general, lack of interaction (±) results in somewhat variable eye sizes while increasing levels of suppression or enhancement (indicated by − or +, respectively) are more uniform. The Ten-m alleles were a gift of Ron Wides and the details of other alleles used are available at http://flybase.bio.indiana.edu/. To screen for GMR modifiers, P{w⁺ GMR-rho} flies were crossed to males with genotypes listed in Tables 1 and 2 and, as controls, P{w⁺ GMR-rho} virgins were crossed to wild-type (Ore-R) males and stat92E⁰⁶³⁴⁶ males.

TABLE 1.

Interacting candidate genes from the P{w⁺, GMR-updΔ3′} screen

	P element					Interaction
Allele	Position	Location	EP	Cytology	Disrupted gene	GMR-upd	GMR-rho	Comments/references
F332.1	2.8 kb 5′ of first transcriptional start site	200,515 in AE003699.3	N	87E8-E9	diminished discs (did/alt1)	−−	+/−	Defects in cell cycle and chromosome condensation (Gatti and Baker 1989)
P7.7/03	8.4 kb 5′ of first transcriptional start site	101,671 in AE003544.4	N	68C12-13	Mob1	−	+/−	Required for exit from mitosis in yeast (Luca et al. 2001)
P7.9/02	13 bp 3′ of first transcriptional start site	73,903 in AE003591.4	N	77C1	tribbles (trbl)	−−	+/−	Negative regulation of mitosis (Seher and Leptin 2000)
F389.1	In 1st intron	56,888 in AE003581.4	N	23B1-2	lilliputian (lilli)	−	−−	Regulation of cell size (Wittwer et al. 2001)
F393.1	204 bp 5′ of first transcriptional start site	224,892 in AE003833.4	N	45F1-F3	mir-14	−−	+/−	Negative regulation of cell death (Xu et al. 2003)
P1.3/03	In 2nd intron	11,985 in AE003720.3	N	90D1-6	couch potato (cpo)	−	+/−	RNA binding (Bellen et al. 1992; Lasko 2000)
H469.2	7.9 kb 5′ of first transcriptional start site	32,337 in AE003599.2	N	80A3	jim	+	+/−	Zinc-finger TF activity (Drysdale et al. 2005)
F4.24/03	In 2nd intron	86,857 in AE003493.3	N	12A9-B2	NFAT	++	+/−	RNA polymerase II transcription factor activity (Drysdale et al. 2005)
F3.28/03	In 1st intron	245,143 in AE003808.4	N	52F5-7	CG8443	−	+/−	Translation initiation factor activity (Drysdale et al. 2005)
P10.3/12	In 2nd exon	36,214 in AE00–3721.3	N	90F6-7	Sequence-specific single-strand DNA-binding protein (Ssdp)	−−	+/−	Transcription regulator activity (L Chen et al. 2002)
H74.3	In 1st intron	84,204 in AE003800.4	N	55C1	poly (A)-binding protein (pAbp)	−	+/−	RNA binding (Lasko 2000)
F464.1	1 bp 5′ of first transcriptional start site	95,124 in AE003698.3	N	85B10	C- terminal Binding Protein (CtBP)	−	+/−	Negative regulation of transcription from Pol II promoter (Alifragis et al. 1997)
H186.1	33 bp 5′ of first transcriptional start site	434,6686 in AE03481.3	N	64B2-4	Cip4	−	+/−	Rho interactor activity (Drysdale et al. 2005)
P2.3/03	In 3rd intron	60,887 in AE003451.3	N	9D2	Rin1/Sprint (Spri)	+	+/−	Ras interactor activity (Szabo et al. 2001)
P1.24	In 3rd intron	109,964 in AE003636.4	N	33E5-9	bunched (bun)	−−	+/−	Component of dpp signalling (Dobenset al. 2000)
F716.1	305 bp 5′ of first transcriptional start site	50,764 in AE003532.3	N	71A4	Bearded (Brd)	−	+/−	Negative regulator of notch pathway (Laiet al. 2000)
F29.6	783 bp 5′ of first transcriptional start site	2,649 in AE003467.3	N	61A5	mth-like8 (mthl8)	−	+/−	G-protein-coupled receptor activity (Brody and Cravchik 2000)
F411.1	350 bp 5′ of first transcriptional start site	284,254 in AE003597.4	Y	79D2	Tenascin-major (Ten-m)	+	+/−	Secreted and transmembrane protein (Baumgartner et al. 1994)
F463.1	14 kb 5′ of first transcriptional start site	139,258 in AE003680.4	N	85B1	CG8351	−−	+/−	ATPase activity, coupled, involved in protein folding (Drysdale et al. 2005)
F595.1	In 1st exon	5,907 in AE003689.2	N	86C7	Sorbitol dehydrogenase-2 (Sodh-2)	−−	+/−	Carbohydrate metabolism (Luque et al. 1998)
P10.1/12	In 1st intron	30,952 in AE003781.5	N	39E2	CG3305	−	+/−	Contains Lamp/CD68 domains (Drysdale et al. 2005)
F294.3	In 2nd intron	46,839 in AE003820.3	N	49D4-6	CG17574	−−	+/−	Unknown, no homologs
F422.1	300 bp 5′ of first transcriptional start site	112,391 in AE003520.4	Y	75C6	CG4306	−	+/−	Unknown, no homologs
A1	11 kb 5′ of first transcriptional start site	192,436 in AE003623.3	N	30A2	CG32982	−−	+/−	Pleckstrin-like protein domains (Drysdale et al. 2005)

The degree of interaction observed was classified under “GMR-upd” and “GMR-rho” as +/−, no effect; +, mild enhancer; ++, moderate enhancer; −, mild suppressor; −−, moderate suppressor. EP indicates if the UAS sequence within the P element is oriented to drive expression of the associated gene. Cytology is computationaly derived.

TABLE 2.

Secondary validation of the P{w⁺, GMR-updΔ3′} candidate genes

Gene	Allele	GMR-upd	os1
did/alt1	F332.1	−−	ND
EY03597	−−	Y
l(3)05137	−−−	Y
CG8443	F3.28/03	−	N
KG02346	−−	N
Ssdp	P10.3/12	−−	Y
KG03600	−−−	Y
Mob1	P7.7/03	−	Y
KG00128	−−	Y
KG05879a	−−−	Y
KG05765	−−−	N
jim	H469.2	+	ND
BG01625	−−	ND
GE23777	−	N
GE16615	−−	N
trbl	P7.9/03	−−	N
NFAT	F4.24/03	++	ND
GE02173	−−	N
pAbp	H74.3	−	ND
EP310	+/−	Y
k10109	−−−	ND
CtBP	F464.1	−	ND
87De-10	+/−	Y
03463a	+	N
KG07519	−−	Y
mir-14	F393.1	−−	ND
k10213	−	Y
cpo	P1.3/03	−	ND
l(3)01432	+/−	N
Cip4	H186.1	−	ND
EY11321	−−−	Y
sprint	P2.3/03	+	ND
KG07279	−−	Y
bun	P2.7/03	−−	Y
P1.24	−−	Y
00255	−−	Y
KG00392	−−	Y
KG00456	−−	Y
Brd	F716.1	−	Y
1	−−−	Y
BG02319	−−	Y
mthl8	F29.6	−	Y
Ten-mb	F411.1	+	N
odz3	−−	ND
odz1	−−−	Y
odz29	−−	ND
CG8351	F463.1	−−	Y
KG01477	−−	Y
Sodh-2	F595.1	−−	Y
CG3305	P10.1/12	−	N
KG04055	−−−	Y
CG17574	F294.3	−−	ND
EY01345	−−	Y
CG4306	F422.1	−	ND
EY06480	−−−	Y
CG32982	A1	−−	N

The degree of interaction was classified as +/−, no interaction; +, mild enhancer; ++, moderate enhancer; −, mild suppressor; −−, moderate suppressor; −−−, good suppressor. For os¹: Y, enhancement of the small-eye phenotype observed; N, no effect; ND, not determined.

^aPreviously identified as a mild enhancer (Bach et al. 2003).

^bA deficiency removing this locus was previously identified as a suppressor (Bach et al. 2003).

For interaction with the loss-of-function os¹ allele (Verderosa and Muller 1954), homozygous females were crossed to Ore-R, w¹¹¹⁸, and stat92E⁰⁶³⁴⁶ as negative and positive controls, respectively. Crosses to potentially interacting alleles were set up in parallel with controls. Hemizygous os¹ mutant males were scored for an enhancement of their eye-size reduction in the next generation.

Inverse PCR:

Inverse PCR was performed essentially as described on the BDGP web page (http://www.fruitfly.org/). The PCR-amplified DNA was sequenced, and the resulting sequences were aligned to release 3.0 and release 4.0 genomic DNA using BLAST searches (Adams et al. 2000).

RNA interference:

Double-strand (ds)RNA targeting the various candidate genes was prepared from 400- to 500-bp PCR products, amplified from genomic DNA using primers containing a 5′ T7 promoter (GAATTAATACGACTCACTATAGGGAGA). The 18- to 20-bp gene-specific portion of primers was directed against single exons of the 18 candidate genes. Further details are available on request. PCR products were used as direct templates for in vitro transcription using the T7 RNA polymerase. To obtain dsRNA, in vitro-transcribed RNA was heated to 95° for 1 min and then allowed to cool slowly to room temperature.

For the paracrine assay, 5 × 10⁶ Kc₁₆₇ cells were seeded in 6-well dishes 1 day before transfection. Cells were batch transfected using Effectene (QIAGEN, Chatsworth, CA). For “signaling cells” (Figure 2B), 600 ng of pAct5c-upd-GFP was transfected per well in a 6-well dish, and for “receiving cells,” 500 ng of 6x2xDrafLuc(wt) reporter and 25 ng of pAct5c-RL was transfected. Plasmids are described in Müller et al. (2005). Following transfection, cells were grown for 24 hr, and subsequently the media were removed and the cells were mixed at a ratio of 1:1 in serum-free Schneider's Drosophila medium. A total of 25,000 cells were then aliquoted into 96-well plates containing 20–50 nm dsRNA/well. Following dsRNA treatment, the cells were grown for 72 hr and then lysed. Twenty microliters of the lysate was used to carry out the dual luciferase measurements. Firefly and Renilla luciferase activity was measured by dual-luciferase reporter assay (Promega, Madison, WI) on Wallac Victor Light 1420 luminescence counter (Perkin-Elmer, Norwalk, CT). Relative reporter activity is shown as a ratio between the firefly luciferase readout and that of the Renilla luciferase. Reporter activation values in the experimental samples were normalized to mock transfected cells.

Figure 2.

Identification of pathway modifiers using a cell-based RNAi assay. (A) Diagrammatic of the 6x2xDraf Luc STAT reporter construct containing 12 STAT92E binding sites (shown with essential residues underlined) and pAct5C-RL used as transfection control. (B) Outline of the paracrine assay system. (C) The effect of dsRNA treatment on the paracrine mode of pathway activation. JAK/STAT pathway activity in each case is shown as fold luciferase activity, and mock transfection corresponds to the basal level of pathway activity with full reporter activity defined as 1 (dashed line). dsRNA-targeting Rh5 and lacZ show no effect, stat92E dsRNA treatment results in an almost basal level of reporter activity, and dsRNA treatment knocking down socs36E results in threefold activation. The seven candidate genes identified in this screen show statistically significant reduction in reporter activity: cip4, P = 0.0002; CG4306, P = 0.001; CG8443, P = 0.0007; Mob1, P = 0.001; CtBP, P = 0.05; did, P = 0.015; and Ten-m, P = 0.012.

Statistics:

Statistical analysis of data sets was undertaken using Microsoft Excel (mean and standard deviation measurements) and U-tests (see http://faculty.vassar.edu.html).

Histology:

In situ hybridization was carried out as described in Lehmann and Tautz (1994). To prepare sense and antisense probes for the candidate genes, direct PCR products amplified from genomic DNA with T7-containing primers were used as templates for in vitro transcription using the digoxygenin labeling kit (Roche). Wild-type and P{w⁺,GMR-updΔ3′} eye discs were prepared and stained in parallel and the color reaction was developed for the same amount of time. The discs were dissected and mounted in 70% glycerol and photographed using a Zeiss Axioskop2 MOT microscope.

RESULTS

Design of a sensitized screen:

To identify dominant modifiers of P{w⁺, GMR-updΔ3′}, we generated new independent autosomal insertions of the P{Mae-UAS.6.11} P element using the crossing scheme in Figure 1F. In this scheme, the mutagenic transposon was mobilized from within the CyO balancer using the {Δ2-3}99B transposase source. Single, yellow⁺ males were selected in the F₂ generation to ensure unique transposition events. In the F₃ generation, the chromosome containing the reintegrated P element was identified, with those inserted in the X chromosome used for a genomics project (Beinert et al. 2004) and those present on autosomes crossed to P{w⁺, GMR-updΔ3′} females. The eye size of the resulting P{w⁺, GMR-updΔ3′}/+; P{Mae-UAS.6.11}/+ females was compared to controls out-crossed to the Ore-R or w¹¹¹⁸ “wild type” strains and the strong stat92E⁰⁶³⁴⁶ allele (Hou et al. 1996). Interaction strength was graded according to a scale in which those equivalent to the P {w⁺, GMR-updΔ3′}/+; stat92E⁰⁶³⁴⁶/+ control (Figure 1C) were classified as strong suppressors while those comparable to P{w⁺, GMR-updΔ3′}; +/+ (Figure 1B) were defined as no effect (±). Using this scoring system, P{Mae-UAS.6.11} insertions that modify the P{w⁺ GMR-updΔ3′}-induced eye overgrowth were identified. Although the degree of overgrowth induced by the P{w⁺ GMR-updΔ3′} transgene varies within any population of F_4a progeny, it was consistently observed that the range of variance within a population was reduced in backgrounds showing specific interactions and this phenomena was also used to identify potentially interacting loci.

In crosses showing probable interaction with P{w⁺, GMR-updΔ3′}, sibling males carrying the P{Mae-UAS.6.11} insertion were recovered and balanced. The resulting stocks were then retested over multiple rounds to ensure that the interaction was consistent and segregated with the y⁺ marked chromosome. Examples of representative eyes showing interaction between P{w⁺, GMR-updΔ3′} and the interacting mutations identified include the weak/moderate suppression shown by jim alleles (Figure 1D) and the strong suppression caused by the removal of a single copy of Bearded (Figure 1E).

We tested a total of 2267 independent, primarily autosomal insertions for interaction. During the initial round of screening, we retained a total of 91 (4%) interacting loci for further analysis and, of these, 23 (1%) candidates passed all subsequent rounds of retesting and were classified as potential interactors (Table 1).

Although the genetic interaction induced by the 23 P{Mae-UAS.6.11} insertions was reconfirmed during multiple rounds of rescreening with P{w⁺, GMR-updΔ3′}, it remains possible that the effect observed may result from a modulation of the strength of the GMR promoter rather than from any influence on the JAK/STAT pathway itself. We therefore crossed candidate insertions to a stock misexpressing the Rho GTPase within the developing eye under the control of the GMR promoter (Rebay and Rubin 1995; Häcker and Perrimon 1998). Insertions that interact with both P{w⁺, GMR-updΔ3′} and the dominant rough-eye phenotype induced by P{w⁺, GMR-rho} are likely to represent nonspecific interactions. Although most insertions showed interactions specific for only P{w⁺, GMR-updΔ3′} (Table 1), one insertion, subsequently identified as representing a putative lilli allele, also modified the P{w⁺, GMR-rho} rough-eye phenotype. This finding is consistent with previous studies that identified lilli as a transcriptional regulator of the GMR promoter (Tang et al. 2001).

Identification of interacting genes:

Having identified the P{Mae-UAS.6.11} insertions specifically interacting with P{w⁺, GMR-updΔ3′}, we then set out to determine the genes associated with these mutations. Genomic DNA flanking the P-element insertion site was therefore recovered by inverse PCR (Beinert et al. 2004), sequenced, and aligned to release 3.0 and release 4.0 genomic DNA (Adams et al. 2000; Celniker et al. 2002) using BLAST searches. Single, unambiguous P-element insertion positions could be determined for all interactors. The position of the insertion relative to the putatively mutated genes, the absolute position within AE clones of the Drosophila euchromatin release 4.0 sequence, and the direction of potential misexpression from the upstream activation sequences (UAS) present within the P{Mae-UAS.6.11} transposon are given in Table 1.

As expected from a genetic screen of this type, candidate interacting genes of various classes were identified. These include proteins poposed to be involved in the regulation of the cell cycle (did, Mob1, tribbles), transcription factors (jim, NFAT), DNA- and RNA-binding proteins (Ssdp, CG8443, couch potato, pAbp, CtBP), members of other signal transduction pathways (Cip4, Bearded, bunched, sprint, mth-like 8) as well as the cell adhesion protein Tenascin-M. In addition, a number of uncharacterized genes of unknown function were also identified (CG3305, CG4306, CG17574, CG32982).

Although no examples of multiple P-element insertions were detected by inverse PCR, it remains possible that the interacting mutations may be independent of the P{Mae-UAS.6.11} transposons characterized. In addition, the identity of loci potentially mutated by transposons inserted at a distance from currently annotated transcription units is not always unambiguous. To address these limitations, we therefore set out to test the interaction of other available alleles of the candidate genes. Using mutations obtained from the Bloomington stock center, as well as lines from individual labs, we were able to retest independently generated alleles of all 23 putatively interacting loci for their ability to modulate P{w⁺, GMR-updΔ3′}-induced eye overgrowth (Table 2).

The ability to validate putative mutations in this manner was particularly helpful in the case of the did^F332.1 and Mob1^P7.7/03 alleles where the putatively mutagenic P-element insertion was mapped between 2.8 and 8.4 kb upstream of the first annotated transcriptional start site (Table 1). Despite the separation between gene and transposon, independently generated alleles of each of these loci all demonstrated consistent interaction with P{w⁺, GMR-updΔ3′} (Table 2). It therefore seems likely that the P elements originally identified represent bona fide alleles and may affect enhancer regions, differential splice forms, or as-yet-unannotated 5′ exons.

Although most genes tested showed interactions consistent with those originally identified, a subset of the alleles that we identified, including Ten-m^F411.1, NFATF^4.24/03, and jim^H469.2, shows interactions opposite to those produced by independently generated loss-of-function alleles (Table 2). Given the presence of UAS sequences within the P{Mae-UAS.6.11} transposon, and potential cryptic promoters present within the long terminal repeats of the P element, it is possible that the mutations identified represent gain-of-function alleles whose misexpression may explain the converse interactions observed.

On the basis of our ability to consistently identify interactions with independently generated alleles, we subsequently focused our efforts on 18 candidate genes (underlined in Table 2) on the basis that these are most likely to represent bona fide interacting loci. These candidate modulators of JAK/STAT signal transduction form the basis of our further studies.

Interaction testing with os¹:

Having defined a set of loci that interact with the P{w⁺, GMR-updΔ3′} gain-of-function phenotype, we then tested for potential genetic interaction with a complementary loss-of-function phenotype. Previous reports have shown that the hypomorphic loss-of-function os¹ allele of the pathway ligand Upd results in a reduction in the size of the adult eye (Verderosa and Muller 1954), a phenotype that can be rescued by exogenous pathway activation (Bach et al. 2003; Tsai and Sun 2004). In addition, the degree of eye-size reduction caused by os¹ is significantly enhanced when an individual is simultaneously mutant for the downstream pathway components hop or stat92E (Tsai and Sun 2004).

Although the enhancement of the os¹ small-eye phenotype caused by the removal of one copy of stat92E is relatively subtle, a distinct and reproducible genetic interaction is observed (not shown). We therefore used this interaction to screen alleles of the genes previously identified as modifiers of P{w⁺, GMR-updΔ3′} and tested for enhancement of the eye-size reduction that may indicate an interaction between the tested loci and endogenous JAK/STAT signaling (Table 2).

Although not observed in all cases, many of the loci tested modify not only the gain-of-function, but also the loss-of-function eye phenotypes (Table 2) and are therefore likely to represent genes that are required for both ectopically activated and endogenous levels of JAK/STAT pathway signaling.

Characterization of modifiers—RNAi-based assays:

The modification of the P{w⁺, GMR-updΔ3′}-induced phenotype may be the consequence of mutations in genes encoding components of the JAK/STAT pathway, direct and indirect regulators of the pathway, or downstream target genes required to elicit the biological phenotype used in the initial screen—namely cellular overproliferation in the developing eye.

The Upd ligand is a secreted glycoprotein (Harrison et al. 1998) capable of activating the JAK/STAT pathway in cells located at some distance from the source of expression (Zeidler et al. 1999b; Tsai and Sun 2004). To fulfill this function, Upd must be post-translationally modified, secreted into the extracellular space, and must interact with receptors on the receiving cell. Given that the Upd expression domain is physically separate from the region of increased cellular proliferation (Tsai and Sun 2004), it is possible that genes identified in our screen may represent loci required for these upstream processes.

We therefore set out to devise a model with which such paracrine signaling can be mimicked in a tissue-culture-based system using an assay with which JAK/STAT pathway activity can be determined (Müller et al. 2005). The reporter used consists of a firefly luciferase gene upstream of 12 copies of a STAT92E binding site originally identified in the promoter of the pathway target gene Draf (Figure 2A; Kwon et al. 2000). To generate a paracrine model of pathway signaling, we transfected an Upd-expressing vector into one population of Kc₁₆₇ (Cherbas et al. 1977) “signaling cells” and transfected the reporter and transfection control into another population of Kc₁₆₇ “receiving cells” (Figure 2B). These two cell types were then mixed and co-cultured in sextuplicate parallel experiments carried out in 96-well plates before measurement of reporter gene activity. In this scenario, receptor stimulation must result from Upd originally expressed in another cell (Figure 2B) with the 6x2xDrafLuc reporter showing a robust induction in response to this form of paracrine activation (Figure 2C; Müller et al. 2005).

Given the effectiveness of the reporter system, we then set out to use RNAi-mediated knockdown (Clemens et al. 2000) to identify the candidate genes that might represent pathway components or modulators.

Under the experimental conditions used, Upd-dependent paracrine stimulation, in the absence of dsRNA, results in luciferase activity ∼60 times higher than that of unstimulated controls (Figure 2C, columns 1 and 2). This level of reporter activation was defined as 1 unit of luciferase activity and is not affected by dsRNA-targeting Rhodopsin-5 or lacZ used as negative controls. However, addition of dsRNA-targeting stat92E was sufficient to reduce reporter activity to almost basal levels. Conversely, knockdown of the negative regulator socs36E boosted reporter activity almost threefold (Figure 2C). These results serve to validate the assay and suggest that potential positive and negative regulators of the pathway can be identified using this technique.

We therefore synthesized dsRNA targeting the 18 candidate genes selected for further analysis (underlined in Table 2) and tested these in the assay described above. These assays identified seven dsRNAs, which statistically significantly (P < 0.05) reduced the level of reporter activity (Figure 2C), a result consistently reproduced in multiple experiments. Given this interaction, it is likely that the seven identified interacting dsRNAs target the transcripts of genes encoding potential pathway components and/or regulators. The remaining 11 loci for which no consistent effect was observed (not shown) may represent genes functioning at other levels.

Although this tissue-culture-based assay gives a clear result for interacting loci, lack of interaction is not sufficient to disprove a potential role. For example, it is possible that the noninteracting loci may constitute pathway components that are not required or expressed in Kc₁₆₇ cells but are necessary for full activity in the developing eye disc. Alternatively, it is also possible that the dsRNA treatment used did not adequately knock down the activity of all targeted transcripts.

Characterization of modifiers—in situ hybridization:

To represent credible candidates, the identified genes must be expressed within the developing eye imaginal disc during the stages when additional P{w⁺, GMR-updΔ3′}-induced overproliferation occurs. In addition, it is possible that the screen has identified pathway target genes whose expression is either directly or indirectly modulated by STAT92E activity and whose activity is required for the proliferative cellular response. To address both questions, we examined the expression pattern of the genes identified by in situ hybridization to their mRNAs in both wild type and P{w⁺, GMR-updΔ3′}/+ late third instar eye-antennal imaginal discs (Figures 3 and 4).

Figure 3.

Expression of candidate genes in wild-type third instar larval eye-imaginal discs. All the genes are expressed in wandering third instar eye discs; in some, expression is also detected in the morphogenetic furrow (MF; arrowheads). The expression pattern of these genes does not change in P{w⁺, GMR-updΔ3′} eye discs (data not shown). Discs are shown with anterior at the left and dorsal at the top. In situ antisense probes: did (A); CG8443 (B); Ssdp (C); jim (E); NFAT (E); cpo (F); Cip4 (G); Sprint (H); bun (I); Brd (J); CG17574 (K); CG4306 (L).

Figure 4.

Potential JAK/STAT pathway target genes. Expression patterns of candidate genes in wild-type (top row) and P{w⁺,GMR-updΔ3′}/+ (bottom row) eye discs identified six potential pathway target genes. CtBP, trbl, mthl-8, and CG3305 are expressed in wild-type eye discs and are upregulated in P{w⁺, GMR-updΔ3′}/+ eye discs (A and B, C and D, E and F, G and H). Ten-m is expressed uniformly in a wild-type eye disc with stronger staining in the MF (I); by comparison, expression in P{w⁺, GMR-updΔ3′}/+ eye discs is below detectable levels (J). Mob1 expression is also detected in wild-type eye discs with higher levels of expression ahead of the MF (K). The expression ahead of the MF is strongly reduced in P{w⁺, GMR-updΔ3′} eye discs (L). Anterior is at the left; dorsal is at the top. All experimental pairs of wild-type and P{w⁺, GMR-updΔ3′}/+ eye discs were stained for the same time under identical conditions.

As expected for genuinely interacting genes, antisense RNA probes used for in situ staining showed that all identified candidates are expressed in wild-type eye-imaginal discs. Although some interacting genes are expressed only at low levels, a large proportion are upregulated in or ahead of the morphogenetic furrow (Figures 3 and 4). By contrast, control sense RNA probes showed no or only low-level background staining (not shown). Of the loci examined, 12 are expressed at essentially equivalent levels in both wild-type and P{w⁺, GMR-updΔ3′}/+ eye discs (Figure 3). However, the expression of CtBP, trbl, mthl-8, CG3305, Ten-m, and Mob1 is clearly modified by ectopic JAK/STAT signaling identifying them as potential pathway target genes (Figure 4). Of this group, expression of CtBP, trbl, mthl-8, and CG3305 were upregulated in P{w⁺, GMR-updΔ3′} eye discs (Figure 4, A–H), an effect consistent with the known role of STAT92E as a transcriptional activator. By contrast, both Ten-m and Mob1 show a clear and consistent decrease in expression associated with upd misexpression (Figure 4, I–L). Although this apparent negative regulation is unexpected, it is possible that it may represent an indirect effect of pathway signaling.

It should, however, be noted that the ability of P{w⁺, GMR-updΔ3′} to alter the expression pattern of the six interacting genes does not prove that these loci are normally the target of endogenous pathway activity.

P {w⁺, GMR-updΔ3′} and signal transduction pathways:

One intriguing aspect of the genes identified was the interaction with bearded, a component of the Notch (N) pathway, and bunched, a positive regulator of the decapentalplegic (dpp) signal transduction pathway. These results suggest that cellular proliferation induced by ectopic upd expression is also sensitive to inputs from other signal transduction pathways. Such effects may result from direct interaction or may be the result of coregulation of common pathway target genes involved in cellular proliferation. While a genetic interaction between the P{w⁺, GMR-updΔ3′} eye overgrowth phenotype and Dpp pathway members has already been observed (Bach et al. 2003), the interaction with Notch signaling components has not been previously described. We therefore tested other members of the Notch pathway to determine if mutations in these components show similar interactions. Consistent with our original identification of an allele of bearded, mutations in the Notch receptor and Delta ligand also suppress eye overgrowth, although the Serrate ligand does not appear to interact (Table 3). Although this finding is consistent with reports from vertebrate systems in which activation of STAT3 by Notch has been demonstrated (Kamakura et al. 2004), it is also possible that the coregulation of common target genes may explain this interaction. One precedent for such coregulation is the expression of four-jointed in the developing eye disc, which not only requires JAK/STAT pathway activity but also integrates Notch and Wingless signaling (Zeidler et al. 2000b).

TABLE 3.

P{w⁺, GMR-updΔ3′} and Notch signaling pathway components

Gene	Allele	GMR-upd	Comments
stat92E	06346	−−−	Positive control
Notch (N)	I1N	−−	Temperature-sensitive allele raised at partially permissive temperature (25°)
Delta (Dl)	9P39	−−−
3	−−−
RevF10	+/−
Serate (Ser)	Bd-3	+/−
Bearded (Brd)	F716.1	−	Allele identified in this report
1	−−−
	BG02319	−−

+/−, no interaction; −, weak suppressor; −−, moderate suppressor; −−−, strong suppressor.

P{w⁺, GMR-updΔ3′} and cell cycle components:

Given the increase in cellular proliferation associated with the eye overgrowth phenotype used in the screen, we were intrigued that no mutations in components of the core cell cycle machinery had been identified. However, given the nonsaturating nature of our mutagenesis, it is possible that such alleles were not included in the collection of mutated chromosomes screened. To address this question, we therefore tested for potential interactions with alleles of known cell cycle components (Table 4). Although weak interaction was observed for some alleles, the majority of mutations removing diminutive, string, CyclinA, CyclinB, CyclinB3, CyclinD, CyclinE, E2f transcription factor, roughex, p53, dacapo, gigas, Dichaete, or the cyclin-dependent kinase cdc2 showed no consistent interaction with P{w⁺, GMR-updΔ3′} (Table 4). Furthermore, although the cdk4³ allele previously reported as interacting with STAT92E (Chen et al. 2003) was classified as a weak suppressor, other independently generated cdk4 mutations produced no consistent interactions (Table 4).

TABLE 4.

P{w⁺, GMR-updΔ3′} and cell cycle components

Gene	Allele	GMR-upd	Comments/ references
diminutive (dm)	1	−
cdc2	2	+/−
3	−
string (stg)	4	+/−	Drosophila Cdc25 Homolog
01235	+/−
CyclinA (CycA)	03946	+
C8LR1	+/−
CyclinB (CycB)	2	+/−
CyclinB3 (CycB3)	2	+/−
CyclinD (CycD)	KG04817	+/−
Cyclin E (CycE)	AR95	+/−
05206	+/−
k05007	+/−
Cyclin-dependent kinase 4 (Cdk4)	3	−	cdk4 has been previously reported as interacting with stat92E (Chen et al. 2003)
k06503	+/−
s4639	+/−
DP transcription factor (Dp)	49Fk-1	+/−
E2F	07172	+/−
roughex (rux)	1	+/−
2	−
p53	5A-1-4	+/−
dacapo (dap)	4	+/−
04454	+/−
07309	−
gigas (gig)	109	+/−	Cell-size regulator
Dichaete (D)	r8	+/−

+/−, no interaction; +, mild enhancer; −, mild suppressor.

DISCUSSION

We have used a genetic approach to identify regulators of the Drosophila JAK/STAT signal transduction pathway. Using an in vivo eye overgrowth assay, we screened 2267 independent P-element insertions and identified 23 loci responsible for the modification of the overgrown eye phenotype associated with P{w⁺, GMR-updΔ3′}. Using a quantitative cell-based STAT92E activity assay, we have determined that seven of these candidates are likely to be potential pathway components/regulators. In addition, in situ hybridization was used to show that expression of six of these genes is modulated in response to pathway activity and is likely to represent direct or indirect pathway target genes.

The loci identified represent new components and modulators of the JAK/STAT signal transduction pathway, most of which have not previously been identified as such. Our analysis of the genes, in conjunction with their known biological roles, allow the candidates to be subdivided into a number of classes.

Cell cycle proteins:

The eye overgrowth induced by P{w⁺, GMR-updΔ3′} results from additional rounds of mitosis in eye-imaginal disc cells anterior to the morphogenetic furrow (Bach et al. 2003). Despite the ectopic JAK/STAT pathway activation caused by the misexpression of upd, these cells are patterned essentially normally and go on to form an increased number of ommatidia in the P{w⁺, GMR-updΔ3′} eye disc (Bach et al. 2003). Despite this proliferation-dependent phenotype, core cell cycle regulatory proteins failed to show consistent interactions when assayed as part of a candidate approach (Table 4). While unexpected, this result suggests that the core cell cycle regulatory proteins do not represent components that become rate limiting in the proliferative environment tested.

Despite the lack of interaction with core cell cycle components, alleles of did, trbls, and Mob1 were identified as modifiers of the overgrown eye phenotype. Indeed, homozygous did mutants have been described as having small imaginal discs (Gatti and Baker 1989), and a phenotype similar to that is observed in hop^M13 mutant third instar larval discs (Perrimon and Mahowald 1986; Mukherjee et al. 2005). While not central to cell cycle progression, these loci appear to be involved in its regulation and may imply that the interaction between JAK/STAT signaling and cellular proliferation is indirect.

Of particular interest are the inconsistent interactions observed between Cdk4 alleles. Although cdk4 represents the only Drosophila component of the cell cycle machinery proposed to interact with the JAK/STAT pathway (Chen et al. 2003), our assay identified only one of the three alleles tested as a weak suppressor of the eye overgrowth phenotype (Table 4). Previous studies did not utilize loss-of-function experiments but rather utilized the converse approach. When misexpressed by a P{w⁺, GMR-Gal4} driver, the coexpression of P{w⁺, UAS-CycD}, P{w⁺, UAS-Cdk4}, and P{w⁺, UAS-upd} dramatically enhanced the eye overgrowth phenotype over that mediated by P{w⁺, UAS-upd} or P{w⁺, UAS-CycD} and P{w⁺, UAS-Cdk4} alone (Chen et al. 2003). Although it is possible that loss of a single copy of the cdk4 locus does not reduce protein levels below a rate-limiting threshold, the inconsistency of interactions produced by multiple cdk4 alleles is puzzling and true existence or nature of any potential interaction between JAK/STAT signaling and endogenous Cdk4 remains to be established.

Transcription factors and coregulators:

We have identified a number of transcription factors as interacting loci in our screen. One of these is the Drosophila homolog of the nuclear factor of activated T-cells (NFAT), a locus originally identified as an inducer of cytokine gene expression (Shaw et al. 1988). Intriguingly, it has been shown that human NFAT, in conjunction with NF-κB, AP-1, and STATs, represents factors involved in mediating cytokine and T-cell-receptor-induced interferon-γ signaling (Malmgaard 2004). Intriguingly, activation of these transcription factors results in the production of numerous intrinsic antiviral factors in the vertebrate system, a role that has also been shown to depend on JAK/STAT signaling within Drosophila fat-body cells (Agaisse et al. 2003). Although further analysis of this interaction is required, this is the first report that suggests an evolutionarily conserved link between NFAT and JAK/STAT signaling in Drosophila.

C-terminal binding protein (CtBP), a transcriptional corepressor previously characterized as an enhancer of the Drosophila JAK/STAT pathway (Bach et al. 2003), has also been identified in our screen. While not all alleles of CtBP show consistent interaction with P{w⁺, GMR-updΔ3′} (Table 2), cell culture assays utilizing dsRNA-mediated knockdown imply that CtBP is a component of the JAK/STAT pathway, which acts as a positive regulator of signaling (Figure 2C). In addition, an independent genomewide RNAi-based screen for JAK/STAT pathway interactors also identified dsRNAs targeting CtBP as a suppressor of pathway signaling (Müller et al. 2005). Finally, an upregulation of CtBP transcript is observed in P{w⁺, GMR-updΔ3′} eye discs compared to wild-type eyes (Figure 4, A and B). Given the results from cell-based assays and in situ analysis, it appears most likely that CtBP does indeed represent a positive regulator of JAK/STAT pathway activity. This finding is particularly surprising, given the previously identified role for CtBP as a transcriptional repressor, which, in combination with the Groucho corepressor, is involved in repressing Su(H)-mediating expression of Notch pathway target genes (Barolo et al. 2002). The significance of our result, however, remains to be determined and it is conceivable that the observed interaction with the eye overgrowth phenotype represents an indirect effect, possibly via interaction with Notch pathway signaling activity.

Extracellular proteins:

One aspect of the screen undertaken is the paracrine mode of Upd signaling required for cellular overproliferation. In the P{w⁺, GMR-updΔ3′} eye, the region of upd expression is spatially separate from the domain in which increased levels of cellular proliferation are observed (Bach et al. 2003; Tsai and Sun 2004) and the ligand must therefore be able to move to and activate the pathway in neighboring cells. Although it has been shown that Unpaired represents a secreted extracellular signaling molecule that is both post-translationally glycosylated and able to associate with the extracellular matrix (ECM) (Harrison et al. 1998), very little is known regarding the mechanisms regulating these processes.

One class of molecules previously shown to be involved in the extracellular trapping and movement of signaling ligands is the heparan sulfate proteoglycans (HSPGs) Dally, Dally-like, Perlecan, and Syndecan (Princivalle and de Agostini 2002). These molecules, and their extensive post-translational modifications, not only play important roles in providing shape and biomechanical strength to organs and tissues, but also have been shown to be required for the transduction of signaling by the Wingless, Hedgehog, and the FGF-like ligands Heartless and Breathless (Lin et al. 1999; The et al. 1999; Baeg et al. 2004). Despite the significance of HSPGs for the transduction of these ligands, mutations in the HSPGs themselves, as well as mutations in the HSPG-modifying enzymes sugarless and sulphateless, do not appear to interact with the eye overgrowth phenotypes associated with P{w⁺, GMR-updΔ3′} (not shown; E. Selva, personal communication) and suggest that Upd is likely to interact with the ECM via different mechanisms. One potential component of this alternative mechanism identified in our screen is Tenascin-major (Ten-m). Ten-M, also known as odd Oz (Dgany and Wides 2002), encodes an extracellular adhesion molecule that was also classified as a component of the JAK/STAT pathway in the tissue-culture-based paracrine signaling assay (Figure 2C). Although the tissue culture results imply a direct function of the molecule in pathway signaling, further analysis of the role of Ten-m in controlling the secretion and/or movement of Upd remains to be determined in vivo.

Signaling pathways:

The Drosophila eye is dispensable in a laboratory environment and sensitized genetic screens that compromise its function have proven to be powerful tools for the identification of signal transduction pathway components (Dickson et al. 1996; Karim et al. 1996; Bach et al. 2003). Drosophila eye development is, however, a complex process involving multiple signal transduction pathways including EGFR, Hh, Notch, Dpp, and Wingless. A number of examples of interactions between these pathways and JAK/STAT signaling have been described. For example, a gradient of four-jointed in the developing eye disc is determined by the coordinated activities of Notch, Wingless, and JAK/STAT pathways (Zeidler et al. 1999a). Also, at the posterior dorso/ventral border of the eye, Notch and eye gone (eyg) have been shown to cooperatively induce expression of upd, which then acts to promote cell proliferation (Chao et al. 2004). Consistent with these complex interactions, our screen has identified Bunched (bun), a member of the Dpp signal transduction pathway (Dobens et al. 2000), and Bearded (brd), a member of the Notch signaling pathway (Lai et al. 2000). bunched is a transcription factor that genetically interacts with dpp (Treisman et al. 1995; Dobens et al. 2000). Strikingly, Dpp pathway components have previously been reported as modulators of the P{w⁺, GMR-updΔ3′} eye phenotype, with hypomorphic alleles of dpp and Mothers against dpp (Mad) representing strong suppressors of eye overgrowth (Bach et al. 2003). Similar interactions in mammalian systems have identified the synergistic activity of STAT3 and Smad1 in the differentiation of astrocytes from their progenitor cells. These proteins, however, do not physically interact, but bind to p300/CBP to promote the transactivation of target genes (Nakashima et al. 1999).

Finally, our screen also identified mth-like8, a seven-pass trans-membrane protein with predicted G-protein-coupled receptor activity. Although expression of mth-like8 changes in response to JAK/STAT pathway activation (Figure 4), an in-depth analysis of its interaction remains to be undertaken.

Summary:

As is no doubt the case for all signaling pathways required during development, the JAK/STAT cascade does not function in isolation, and cross-talk between multiple interacting loci is likely to be involved in generating developmental responses by this apparently “simple” pathway. However, the identity of many of these interacting partners is as yet largely unknown. Although traditional forward genetic analysis using transposon-mediated mutagenesis is almost impossible to drive to saturation, the rapidity with which mutated genes can be identified makes this approach appealing. In particular, when combined with reverse genetic approaches such as RNAi, candidate interactions can be rapidly validated. The combination of forward and reverse genetic techniques used here has identified a number of diverse loci involved in transducing and regulating JAK/STAT signaling in vivo. Given the significance of the pathway during development and its implication in human malignancies, it is hoped that a future detailed analysis of these gene products will provide the foundation for a better understanding of this signal transduction cascade.