JAK/STAT Pathway Signalling in Drosophila Melanogaster

Unpaired

Probably the first mutations in a JAK/STAT pathway component to be described were the outstretched group of regulatory mutations identified on the basis of small eye and held out wing phenotypes.¹⁴ Amorphic loss-of-function alleles were subsequently identified during a mutagenesis screen to identify loci required for the proper segmentation of the larval cuticle (Fig. 1). These were classed as a novel group of atypical gap genes on the basis of their striking phenotypes and named Unpaired (Upd) on this basis.¹⁵ Subsequent genetic analysis suggested that Upd must represent a non-autonomously acting factor¹⁶ while subsequent cloning of the locus showed that Upd encodes a secreted, glycosylated protein that leads to the tyrosine phosphorylation of Hop in cultured Drosophila S2 cells¹⁷ and represents the major physiological ligand responsible for the majority of JAK/STAT pathway signalling during many stages of Drosophila development.

Following the sequencing of the Drosophila genome,¹⁸ sequence homology searches identified two further upd-like genes located adjacent to upd on the X chromosome.^19,20 In vivo analysis of Upd2 suggests that it acts as a semi-redundant ligand being expressed in the embryo in an identical pattern to Upd. While null upd2 mutations are completely rescued by endogenous Upd, the phenotypes arising from the complete loss of upd are only partially compensated by upd2.²⁰ Upd and Upd2 are both secreted and have several potential N-glycosylation sites,^20,21 however, they display different biophysical properties with Upd closely associating to the extracellular matrix (ECM) while Upd2 appears to be more freely diffusible²⁰—the significance of this with respect to different in vivo activities is however less clear.

In addition, a third upd-like gene, upd3, has also been identified.¹⁹ upd3 is expressed by the hemocytes of adult flies in response to immune challenge after septic injury and is required for JAK/STAT-mediated signalling in the fat body to produce anti-microbial peptides.²² The role of upd3 in other tissues has not yet been established.

Domeless

The JAK/STAT receptor Domeless was identified by a number of independent groups using P-element mutagenesis to look for defects in the posterior spiracle development,⁸ gut morphology²³ or modulation of eye size¹⁰ and was ultimately named on the basis of the characteristic dome-shaped posterior spiracle in mutants. Molecularly, Dome resembles a Type 1 cytokine receptor of the JAK associated IL-6 superfamily with homology to gp130 and IL-6R. The extracellular region of the receptor contains five fibronectin type III (FNIII) domains with the first two being homologous to the vertebrate cytokine binding modules (CBM).^8,19

Elegant in vivo assays using a bipartite -galactosidase complementation assay have also demonstrated the requirement for Dome homo-dimerisation in multiple developmental tissues as a prerequisite of JAK/STAT pathway signalling,⁹ while expression of dome itself appears to be at least partly dependent on JAK/STAT pathway activity. Analysis of a dome intronic enhancer region shows clear pathway dependent mesodermal expression²⁰ suggesting the possibility of a positively acting feedback loop of Dome activity in these tissues.

Searches of the Drosophila genome identified CG14225 as a distant homologue of dome. CG14225 is also a transmembrane molecule with two extracellular fibronectin domains forming a CBM and with a striking similarity in domain structure to the vertebrate IL-6Ra receptor.¹⁹ As yet no role for CG14225 has been suggested, although it is tempting to speculate on the existence of functional CG14225/Dome as well as Dome/Dome homo- and hetero-dimeric receptor complexes in vivo.

Evidence also exists suggesting noncanonical and potentially Dome-independent, activation of STAT92E under certain gain-of-function scenarios. For example, removal of dCsk, which is a negative regulator of the dSrc tyrosine kinase, leads to cellular overgrowth phenotypes which are strongly suppressed by reducing STAT92E activity.²⁴ It is not, however, clear if the overgrowth phenotype in dCsk mutants, which ultimately leads to elevated Src signalling, is a direct activation of STAT92E or an indirect mechanism that activates JAK/STAT and JNK pathways. A further example of interplay/cross talk between pathways is also suggested by gain-of-function mutations in the Drosophila Torso (Tor) receptor tyrosine kinase, a receptor most homologous to the mammalian PDGFR. Although potentially offering an insight into misregulation of JAK/STAT signalling in human malignancies, the significance of this interaction for normal development is less clear as STAT92E activity is required for gain-of-function mutation of TOR to activate ectopic gene expression, whereas wild-type Tor signalling appears to be mediated by the Ras/Raf/MAPK signalling pathway.^25,26

Hopscotch

Initially identified on the basis of a segmentation phenotype (Fig. 1) strikingly similar to that of Upd, mutations in hopscotch (hop) also reduce the size of imaginal discs, ovaries, testis, hindgut, the foregut imaginal ring and proliferation regions of the brain.²⁷ Following its molecular characterisation, hop was found to encode a nonreceptor tyrosine kinase of the JAK family¹¹ and is most similar to mammalian JAK1 and 2.¹⁰ Hop contains one functional and one pseudo-kinase domain, a SH2-like domain and a FERM domain.¹⁹ Intriguingly, subsequent genetic study identified the long standing l(1)Tumours lethal (Tum-l) mutation as a temperature sensitive dominant gain-of-function allele of hop²⁸ (henceforth termed hop^Tum-l). Intriguingly, these mutations result in the over-proliferation of Drosophila blood cells and the formation of melanotic blood cell tumours. Molecular characterisation of hop^Tum-l and a second gain-of-function allele hop^T42, identified lesions in the JH2 domain which result in constitutive kinase activity^29,30—findings with particular parallels to the more recent identification of JAK2 activating V617F mutations recovered in a wide range of human meyoproliferative diseases.^31-33

STAT92E

The transcription factor of the Drosophila JAK/STAT pathway, STAT92E, was identified by two independent screens, a cDNA library screening approach on the basis of homology to human STATs¹³ and a genetic approach screening for P-elements producing segmentation defects similar to upd and hop mutants (Fig. 1).¹² STAT92E is most similar to mammalian STAT3 and STAT5 and has an essentially conserved domain structure including an N-terminal domain, coiled-coil region, a DNA binding domain, SH2 domain and conserved tyrosine residue likely to represent the target of Hop phosphorylation. Following activation Drosophila STAT92E has been shown to translocate to the nucleus^10,22,34-36 and bind to a palindromic DNA target sequence¹³ to activate transcription.^34,37 In addition, stat92E is expressed as a number of splice variants including an N-terminal 133 amino acid truncated form termed ΔNSTAT92E suggested to function as a dominant negative regulator of JAK/STAT pathway signalling in vivo.³⁸

Intriguingly, recent developments also suggest other roles for STAT92E in addition to its canonical role as a pathway activated transcription factor. For example, mutational analysis of STAT92E have separated DNA binding from the transcriptional activation activities³⁴ while recent studies have also shown a role for unphosphorylated, DNA-associated STAT92E in regulation of heterochromatin protein 1 (HP1) and the progression of heterochromatinisation³⁹ (see also below).

SOCS36E

The socs genes represent the best-characterised negative regulators of the JAK/STAT pathway, both in mammals and Drosophila. Many are themselves target genes of pathway signalling and form a potent negative feedback loop which functions to down-regulate JAK/STAT pathway activity. Three socs-like genes have been identified in the Drosophila genome, socs16D, socs36E and socs44A, all of which contain an SH2 domain and SOCS box (SB). SOCS36E is most similar to mammalian SOCS5 with 29.7% identity whereas SOCS44A and SOCS16D are most similar to mammalian SOCS6 and 7. Of the three Drosophila homologues, SOCS36E is the only JAK/STAT pathway repressor that is also a direct target gene of the pathway.^35,37,40,41 In addition SOCS36E is not only a negative regulator of JAK/STAT pathway activity,^35,37 but has also been reported to suppress EGFR signalling activity in the wing disc suggesting that it can interact with multiple pathways.⁴¹ Although SOCS44A is not a direct target of the JAK/STAT pathway, it can inhibit its activity in some tissues while it also appears to up-regulate EGFR signalling.⁴² It is not yet clear, however, if this is a direct or indirect effect. The function of SOCS16D has yet to be characterised.

dPIAS/Su(var)2-10

PIAS proteins were originally identified in vertebrate systems and termed Protein Inhibitors of Activated STAT. PIAS1 and 3 physically interact with STAT1 and 3 respectively and their over-expression in transfected cells inhibits STAT-induced transcription.^43,44 However, the physiological relevance of PIAS in vivo was first established with its Drosophila homolog dPIAS (originally known as zimp⁴⁵). Reduction or over-expression of dPIAS in flies leads to increases and decreases in JAK/STAT activity, respectively, suggesting that dPIAS acts as a negative regulator of the pathway in vivo.^37,46 Intriguingly, more recent studies in mammalian systems suggest that PIAS proteins have diverse functions including roles as SUMO-ligases, while genetic studies in Drosophila have shown that dPIAS is in fact allelic to the Su(var)2-10 locus. Su(var)2-10 is a modifier of the position effect mottling of eye pigmentation in the adult fly eye thought to represent a readout of heterochromatinisation with mutants in Su(var)2-10 having defects in nuclear structure and chromatin segregation.⁴⁷ It is not yet however clear if these STAT regulation and Su(var) effects are caused by a failure of PIAS-mediated SUMOlation .

Ken and Barbie

The interaction of mutations in ken and barbie (ken) with the JAK/STAT pathway was first shown in a genetic interaction screen suggesting that ken acts as a negative regulator of the pathway.⁴⁸ The ken gene encodes a DNA-binding protein containing three zinc fingers and an N-terminal BTB/POZ domain commonly found in transcriptional repressors and appears to be a homologue of human B-cell lymphoma 6 (BCL6).^49,50 Strikingly, the core DNA binding sequence of Ken identified in vitro overlaps half of the palindromic STAT92E DNA binding sequence.⁴⁹ Luciferase reporter assays carried out in cultured cells and using sites to which either STAT92E and Ken or only STAT92E can bind show that Ken acts as a repressor of the JAK/STAT pathway. Ken, however, only suppresses a subset of JAK/STAT target genes in vivo with the JAK/STAT targets vvl, trh and kni being repressed while socs36E expression is not affected.⁴⁹ This result underlines the observation that modulation of signalling pathways via subtle changes to transcription factor binding sites can have significant effects on cell-specific transcriptional responses to ligand stimulation. Intriguingly, in mammalian systems BCL6 has also been shown to bind to STAT6 DNA binding sites where it can act as a repressor of STAT6 dependent target gene expression^51,52—however a potential link between the modulation of JAK/STAT signalling and the association of BCL6 with B-cell lymphomas remains to be investigated.

PTP61F

ptp61F is one of the 28 predicted protein tyrosine phosphatases encoded in the fly genome⁵³ and was identified as a JAK/STAT pathway regulator in two independent genome-wide RNAi screens.^35,37 PTP61F is a homologue of the human phospho-tyrosine phosphatase B1 (PTPB1) and is implicated in the de-phosphorylation of both JAKs and STATs in both the cytoplasm and nucleus of mammalian cells.⁵⁴ However some confusion remains regarding the true in vivo substrate of PTP61F that regulates JAK/STAT signalling. A mechanism invoking direct de-phosphorylation of activated, nuclear localised STAT92E has been proposed by one group³⁷ while knock down by RNAi results in increased phosphorylation of both Drosophila Hop and STAT92E.³⁵ Even more confusingly, recent proteomics approaches have failed to identify interaction of any JAK/STAT pathway components to a substrate trapping variant of PTP61F.⁵⁵ As such, the precise mechanism of action remains unclear. None the less, in vivo studies underline the repressing activity of PTP61F on the JAK/STAT pathway^35,37 and its embryonic expression in a pattern highly reminiscent of upd indicates that ptp61F expression is likely to be a target of the JAK/STAT pathway suggesting the existence of yet another negative feedback loop level of pathway regulation.³⁵

Embryonic Development

The Drosophila JAK/STAT pathway plays a number of key roles at multiple stages during embryogenesis being activated in diverse tissues and at multiple stages. Its roles start with sex determination and segmentation and subsequently the formation of the tracheal pits, elongation of intestinal tracks and formation of the posterior spiracles. Other, as yet less well defined roles include a role in the normal development of axons within the central nervous system⁵⁶ and embryonic dorsal/ventral patterning.⁵⁷ Here we will discuss the better-studied roles of the JAK/STAT pathway in each tissue in greater detail.

Sex Determination

In Drosophila, sex is determined by the ratio of X chromosome(s) to autosomes. This ratio information is supported by X-linked signal elements (XSE), which turn on the expression of sex-lethal (sxl) in females but not in males. Upd, among others, is one of these XSEs.^58,59 While other XSEs induce transcription of sxl by directly binding to its promoter, Upd is a secreted ligand, which must activate Sxl promoter indirectly. In support of this, the sxl promoter contains two STAT92E binding sites and loss-of-function of hop and stat92E mutations reduce sxl expression. Intriguingly and consistent with this model, genome wide RNAi screening data also indicates that the Drosophila X-chromosome contains a number of positively activating JAK/STAT pathway components and regulators but no identified negatively acting factors.³⁷

Segmentation

Study of the processes regulating the segmentation of the Drosophila embryo has a long history¹⁵ and elucidation of this process has provided profound insights into multiple developmental processes. A combination of molecular, genetic and embryological approaches has led to a hierarchical model describing how a segmented pattern of three thoracic and eight abdominal segments are generated along the anterior-posterior axis. Mutations removing upd, dome, hop and stat92E all lead to deletion or fusion of the fourth and fifth abdominal segment and partial loss of the eigth abdominal segment, as well as defects in the head skeleton and posterior spiracles (Fig. 1).^11,12,27 This distinctive phenotype, however, is not consistent with those of the classical maternal, gap, pair rule or segment polarity phenotypic classes (reviewed in ref. ⁶⁰). Rather, mutants of the JAK/STAT pathway show an atypical gap gene phenotype due to a localised reduction in expression of the pair-rule genes even-skipped (eve), runt and fushi tarazu¹¹ Direct molecular evidence to support this finding has been established for eve where two crucial STAT92E binding sites are located within the eve stripe 3-7 promoter.^11,61

Trachea

The tracheal system supplies oxygen to all organs of the fly via a network of air filled tubules radiating throughout the body with the posterior spiracles representing the external opening during larval stages. The embryonic tracheal system is an epithelial tubular network established from ectodermal precursor cells. During embryonic development tracheal pits are formed which define a population of cells that divide, migrate and fuse to form the complete tubular network. The JAK/STAT ligand upd is expressed in tracheal pits¹⁷ and activates its downstream transcription factor STAT92E.⁵⁶ Mutants that lack maternal and zygotic contribution of hop or stat92E do not form any tracheal system.^8,27 Tracheal pits and posterior spiracles are marked at early stages of embryogenesis by trachealess (trh), knirps and ventral veins lacking (vvl), expression of which is lost in JAK/STAT pathway mutants.⁸ Trh is required early for trachea specification and later to promote invagination to form tubes. The JAK/STAT pathway therefore seems to play a dual role in the trachea, namely activating downstream target genes, which are crucial for cell movement and elongation as well as specifying these specialised cells for invagination.⁶²

Gut

The Drosophila hindgut is a single-layered ectodermally derived epithelium which elongates and narrows during a 10 h period of late embryogenesis (reviewed in ref. 63) and provides an excellent system to study cell-shape changes and cell rearrangements. The process of hindgut elongation during embryogenesis requires polarized cell rearrangements, which are supported by local expression of signalling molecules including hedgehog, wingless and upd.^64-66 Upstream of these pathways, transcription factors including drumstick (drm), bowel (bowl) and lines (lin) are required. Whereas in drm and bowl mutants expression of upd is reduced, in lin mutants upd expression is expanded suggesting that localised Upd protein might be an important cue for hindgut cells.⁶⁷ Inactivation of the JAK/STAT pathway leads to shorter but wider hindgut phenotype, a change not due to reduced cell number, but rather due to the loss of spatially localised JAK/STAT signalling required to trigger cellular rearrangement within the hindgut via a process of convergent extension.⁶⁷

In Drosophila, the activity of Hedgehog, Wingless and the TGF-β homologue Decapentaplegic (Dpp) are also required for the development of the foregut-associated proventriculus.^68,69 This organ regulates the passage of food into the midgut and is composed of endodermal, mesodermal and ectodermal layers. Localised JAK/STAT and Notch signalling are required for proper cellular migratory events in the foregut, which also involve cytoskeletal reorganisation.^23,70

Ultimately, understanding of how JAK/STAT pathway signalling is able to trigger and control cellular movement and rearrangement, not only in the hind gut and the foregut, but also as part of tracheal development and border cell migration (discussed below), will be an important avenue for future research.

Imaginal Discs

The JAK/STAT signalling pathway plays several important roles in the development of both the eye and wing imaginal discs. Imaginal cells are set aside during embryogenesis and proliferate throughout larval life before differentiating and everting during pupal stages to form a large proportion of the adult fly. One of the first roles for JAK/STAT signalling in this process is a requirement for the pathway in cellular proliferation. Regulatory loss-of-function alleles of Upd with small eye phenotypes are amongst the first pathway mutations identified¹⁴ and constitutive expression of dominant negative pathway components in the eye primordia are sufficient to significantly reduce eye size.⁴⁸ Conversely, ectopic pathway activation within the developing eye imaginal disc leads to cellular over-proliferation ahead of the morphogenetic furrow—a zone of multipotent eye progenitor cells.^48,71 This JAK/STAT pathway-induced over proliferation and the large increase in adult eye size that results, is dependent on the normal diploid dose of downstream pathway components. As a consequence, mutation of a single copy of positively acting pathway components is sufficient to cause a relative decrease in eye over-proliferation—an effect subsequently used to undertake a number of genetic screens for previously unidentified pathway regulators.^48,72

A detailed in vivo analysis of the role of JAK/STAT pathway signalling in the regulation of cellular proliferation has also been undertaken in the developing wing imaginal disc.⁷³ Intriguingly, clonally related groups of marked cells mutant for stat92E loss-of-function alleles show different proliferative profiles at differing developmental stages. During early larval stages, loss of the pathway reduces the rate of cellular proliferation. This is consistent with our understanding in the eye and genetic studies implicating the Drosophila Cdk4 and Cyclin D-cell cycle proteins as physical interaction partners of STAT92E.⁷⁴ However, during later stages, when upd expression in the wing disc is more localised, removal of stat92E causes an increase in the rate of cellular proliferation.⁷³ While the mechanisms underlying this change in pathway function remain to be determined, it is intriguing that the single Drosophila stat92E locus can mediate both pro- and anti-proliferative functions in the same tissue during the course of development.

In addition to roles in regulating cellular proliferation, JAK/STAT pathway signalling is also an important regulator of a number of additional stages of eye imaginal disc development. It has recently been shown that the activation of STAT92E in the eye is required to repress wingless (wg) expression and so allows the proper initiation of the morphogenetic furrow,^75,76 while at later stages the JAK/STAT pathway is also required for correct ommatidial rotation⁷⁷ via a process mediated (at least in part) by the pathway dependent expression of the planar polarity regulator four-jointed.⁷⁸ In addition, repression of both wg and the Dpp expression in the developing leg and antennal discs has also been shown to be at least partly under the control of JAK/STAT signalling where the pathway regulates the proximal distal growth of the appendages.⁷⁹

Intriguingly, expression of upd appears to be under the control of the Notch signalling pathway in multiple developmental contexts, including in the eye,^80,81 the embryonic foregut²³ and ovarian follicle cells.⁸² Indeed, ectopic activation of Notch signalling in the eye caused by mutations in endocytic pathway components such as vps25^83,84 and tsg101⁸⁵ lead to Upd mediated non-autonomous eye overgrowth. As such, it is clear that Upd/JAK/STAT signalling processes are targets of upstream factors and regulators of diverse downstream signalling pathways at many stages of development. The recent development of sensitive in vivo reporters of JAK/STAT pathway activity⁸⁶ will no doubt help to unravel these complex interactions in the future.

Hematopoiesis

Drosophila blood cells, termed hemocytes, perform two significant functions. They search for and engulf pathogens and apoptotic cell debris (especially during pupal stages) and they monitor the environment for bacterial, fungal, parasitic and viral infection in order to signal to the fat body to mount appropriate innate immune responses (reviewed in refs. 87, 88).

The three types of blood cells constitute macrophage like plasmatocytes (90-95% of hemocytes), crystal cells (~5%) and lamellocytes. Lamellocytes are large cells that are rarely observed in healthy larva, however large numbers are induced in larval stages upon challenge with parasitic wasp eggs⁸⁹ where they function to encapsulate objects that are too large to be engulfed by plasmatocytes.⁹⁰

Several signal transduction pathways, including the JAK/STAT, Toll, Ras/Raf, Hedgehog and Notch cascades, have been shown to be crucial for Drosophila hematopoiesis.^29,91-95 In particular, two temperature sensitive gain-of-function mutations of the Drosophila Hop kinase, named hop^{Tum-l 28,29} and hop^T42,30 illustrate the key role played by JAK/STAT signalling in haemocyte development. At low temperatures, where hop^Tum-l is only moderately over active, larvae contain 5-20 fold more plasmatocytes than wild type,⁹³ however, when raised at higher temperatures the more strongly signalling gain-of-function Hop^Tum-l mutation drives haemocyte differentiation into lamellocytes, which contribute to auto-encapsulation and black melanotic tumours (Fig. 2A and B).⁹³ The formation of Hop^Tum-l-induced tumours can be suppressed by the removal of one copy of downstream pathway components such as stat92E or over expression of dPIAS,³⁷ while transplantation of the larval haematopoietic organ from a Hop^Tum-l mutant into a wild-type host results in the appearance of melanotic masses, implicating invasive properties of the tumours.⁹⁶

The primary source of larval hemocytes is the haematopoietic organ or lymph gland, a lobed structure located along the dorsal vessel (Fig. 2C). The most anterior lobe of the lymph gland is divided into three distinct regions, the posterior signalling centre (PSC), the medullary zone (MZ), which is marked by the expression of the Domeless receptor and the outer cortical zone. The PCS acts as a signalling centre by expressing the ligands of a number of signalling pathways including Hedgehog,⁹⁴ the Notch ligand Serrate (Ser)⁹⁷ and probably Upd3⁹⁸ which is thought to induce the JAK/STAT pathway in the MZ to maintain the undifferentiated state of pro-hemocytes within this region.⁹⁹

Although the precise relationship between Hedgehog, Notch and JAK/STAT pathway signalling in controlling pro-haemocyte development and maintenance is not yet clear, thin cytoplasmic fillopodial extensions from the PSC suggest that direct cellular contact may be involved.^94,99

While the JAK/STAT pathway is important for lamellocyte differentiation, haemocyte over proliferation has been described to be STAT92E independent.²⁹ This suggests that hop may crosstalk with other pathways to mediate plasmatocyte over-proliferation observed in hop^Tum-l mutants at low temperatures.⁹³ Indeed, Hop has been shown to physically interact and activate both STAT92E and dRaf. dRaf is a key component of the Ras/Raf/Erk pathway, a cascade that has itself been shown to play a key role in haemocyte proliferation^93,100 with the added twist that expression of the draf gene is itself at least partly under the control of the STAT92E transcription factor in vivo.¹⁰¹ Overall, it seems likely that both Ras/Raf and JAK/STAT signalling pathways play key roles in Drosophila blood cell development although the details of a model in which the differentiation of hemocytes into lamellocytes via the JAK/STAT pathway and cell proliferation via Erk mediated signalling remain to be determined.

Immunity

Although the innate insect immune response to bacterial and fungal infections involves the signalling pathways Toll and Imd (reviewed in ref. ¹⁰²), evidence that the JAK/STAT pathway also plays a role in insect immunity was shown in the mosquito Anopheles ganbiae where AgSTAT translocates into the nucleus of fat body cells in response to bacterial infection.¹⁰³ In Drosophila, STAT92E also translocates into the nucleus of fat body cells upon immune challenge, the result being expression of several anti-microbial peptides including the Tep and Tot protein families.²²

Tep1 is one of the four members of the Tep family each of which contain thiolester motifs and have high similarity to thiolester-containing proteins of the complement C3/alpha2-macroglobulin super family. Tep1 expression is highly induced in fat body cells upon immune challenge.¹⁰⁴ This expression is dramatically reduced in hypomorphic hop loss-of-function mutant larvae and is constitutive in Hop^Tum-l individuals. In vertebrates, the complement system mediates inflammatory reactions. Activation of the C system can occur by the lectin pathway, among others, leading to the development of the complement C3 protein.¹⁰⁵ Intriguingly, the JAK/STAT ligand upd has been suggested to be weakly homologous to leptin¹⁰⁶ although the potential significance of this is as yet unclear.

The Tot family encodes eight small peptides, whose transcription is activated in response to bacterial challenge or environmental stress, such as UV-radiation, high temperature, or oxidizing agents.^22,107,108 TotA expression is reduced in challenged flies carrying hypomorphic hop loss-of-function alleles and is dependent on the Domeless receptor in the fat body. Flies carrying an upd hypomorphic allele or upd2 null allele continue to show a normal totA expression in response to challenge. However upd3 expression in the hemocytes (but not in the fat body) is necessary to activate the JAK/STAT pathway in the fat body leading to totA expression.²² The closely related TotC and TotM are also controlled by JAK/STAT pathway during septic injury.^22,109

An innate defence against viral infection also seems to involve the Drosophila JAK/STAT pathway, which is required, but not sufficient, for the antiviral response. Infection of flies with Drosophila C virus induces the transcription of virus-induced RNA 1 (vir-1). Its expression requires JAK/STAT pathway activation and its promoter contains STAT92E DNA binding sites to which the transcription factor is able to bind, while expression is diminished in hypomorphic hop loss-of-function backgrounds.¹¹⁰

Ultimately, although our understanding of the interplay between the Toll, Imd and JAK/STAT pathways in triggering the appropriate responses to sterile and septic injuries, as well as parasitic and viral infections is already partly understood,¹⁰⁹ this area will no doubt represent an important field for future research.

Stem Cell Niches

The importance of the stem cells niche, a microenvironment that provides external signals to maintain surrounding stem cells, is becoming increasingly appreciated across many species (reviewed in refs. 111-114). Intensive work on niches reveal that while primary signals are crucial for germline stem cell (GSC) homeostasis, hierarchies among multiple pathways are also important (reviewed in ref. 115). Adult stem cell niche studies in invertebrates have shed light on the nature of the signals which mediate stem cell regeneration capacity in vivo with the Drosophila GSCs representing one of the best studied adult stem cell niches. In males, tightly packed somatic hub cells at the tip of the testis signal to adjacent GSCs by producing the Upd ligand.^116-118 Asymmetric division of stem cells produces one stem cell close to the hub and one displaced cell further away; secretion of Upd from the hub activates the JAK/STAT pathway, promoting self-renewal in the nearby GSC, while the one further from the source of ligand subsequently differentiates,^119,120 a process that is further supported by the MAP kinase pathway.^121,122 Consistent with this model, constitutive JAK/STAT pathway activation results in a testis filled with stem cell-like cells, while loss of pathway components causes the differentiation and loss of the GSCs.^117,118,123 Reductions in the levels of upd and cadherin expression in the niche cells over time correlates with a decrease in the number of GSCs, indicating that aging of the stem cell niche may well be a factor in decreased stem cell activity.¹²⁴ In addition to Upd, Dpp signalling is also required for male GSC niche.^116,125

The female ovary consists of multiple ovarioles each of which consists of strings of individual egg chambers at progressively more mature developmental stages. The youngest, most anterior end of the ovariole contains two populations of stem cells, the GSC and the somatic escort stem cells (ESC). It was long thought that the maintenance of GSC is performed through a microenvironment formed by somatic cap cells which secrete Dpp to maintain the niche.¹²⁶ However, several studies have now shown that the JAK/STAT pathway lies at the top of a hierarchical signalling model. The pathway not only positively regulates Dpp expression in the cap cells^127,128 but is also needed in ESCs to maintain the structural integrity of the germarium.¹²⁹ The requirement of JAK/STAT activity in the GSC niche of both sexes is however different. Whereas cell autonomous JAK/STAT activity is required in male GSCs, JAK/STAT signalling in female ovary is needed only for the maintenance of Dpp signalling in the cap cells and is dispensable for female GSC self renewal.

In addition to the relatively well characterised roles in germ line stem cell maintenance, the activity of JAK/STAT pathway signalling is also required for the maintenance of haematopoietic stem cells in the lymph gland⁹⁹ and multipotent stem cells in the malpighian tubules (Drosophila kidney).¹³⁰ These findings suggest a recurring theme in which pathway activity is a central component in the establishment and maintenance of stem cell niche environments in vivo. A potential link between this role and the recently popular concept of ‘cancer stem cells’^131,132 is particularly intriguing, especially given the prevalence of inappropriate JAK/STAT pathway activation in a range of human malignancies.¹³³

Oogenesis

As discussed above, JAK/STAT signalling is required to maintain the stem cells that will ultimately generate the somatic support cells required for normal oogenesis.¹²⁹ The Drosophila ovary consists of multiple ovarioles, each of which contains a string of developing egg chambers. Egg chambers are surrounded by a monolayer of somatic epithelial follicle cells, which undergo a highly stereotypical series of movements to enclose the developing oocyte (Fig. 3). Epithelial cells in the Drosophila ovary offer a powerful system to model the study of the development of invasive behaviour and have been shown to actively require JAK/STAT signalling for this process.¹³⁴ These epithelial follicle cells differentiate into three distinct cell populations^135,136 to establish the egg chamber: polar cells at each end of the egg chamber, stalk cells connecting individual chambers and main-body follicle cells overlying the egg chamber itself. Induction of Notch signalling from the germline cells to neighbouring follicle cells induces polar cell fate.^137-139 However, after stalk induction, Upd signalling extends further resulting in a gradient of JAK/STAT signalling across the main-body follicle cells located at the posterior end of the ooctye. Upd ligand secreted from the anterior polar cells signals to adjacent follicle cells defining their fate to border cells, which delaminate from their normal position, undergo an epithelial to mesenchymal transition (EMT) and acquire the ability to migrate between the supporting nurse cells to contact the maturing oocyte (Fig. 3 - dark cells at stage 9-10).^134,140 Over-expressing Upd or Hop is sufficient to recruit surrounding follicle cells into a border cell fate, which then exhibit invasive ‘metastatic’ behaviour.^134,140

This final requirement of the JAK/STAT pathway for the EMT and movement of polar cells has parallels in other Drosophila tissues such as the tracheal pits and gut where JAK/STAT dependent cell movements are crucial to normal development²³ (reviewed in ref. 141). The significance of the recurring theme of JAK/STAT pathway mediated cellular movement may be of particular significance given the poor prognosis of patients with metastatic human cancers. Investigation of the role for JAK/STAT signalling in this process represents a potentially important theme for future research.

Chemical and Genome-Wide RNAi Screens

One of the greatest strengths of the Drosophila experimental system is its tractability as a ‘screenable’ genetic tool. Such screening approaches have also been applied to the study of JAK/STAT pathway signalling in flies and a number of classical genetic screens have been undertaken.^48,72 These have identified a number of novel pathway regulators including the transcriptional repressor ken and Barbie.⁴⁹ The publication of the Drosophila genome in 2000¹⁸ and the discovery of RNA interference (RNAi) as a tool for genome-wide screening^142,143 have further widened the possibilities for screening. Genome wide libraries of double stranded (ds)RNAs targeting all predicted Drosophila transcripts have been generated¹⁴⁴ and large scale cell-based screening has made possible forward genetic studies to identify genes that modulate signalling pathways (reviewed in refs. 145, 146).

Two genome-wide, cell-based screens for JAK/STAT pathway modulating loci have been undertaken in recent years.^35,37 Using STAT92E-responsive luciferase reporters in cultured Drosophila cells, a range of both positively acting genes, whose knock down results in a decrease in JAK/STAT pathway reporter activity and negatively acting loci, whose knock down results in an increase in signalling, have been identified. Following cell-based validation, both screens identified around 100 enhancer or suppressor loci whose knock down was sufficient to modulate JAK/STAT pathway activity. These included core pathway components, the pathway phosphatase ptp61F (described above) and a number of genes whose human homologues have been previously associated with leukaemia,^35,37 (reviewed in ref. 147). While the further analysis of these JAK/STAT pathway modulating loci, both in Drosophila and in vertebrate model systems, will undoubtedly represent an important future research direction, it is also likely that further genome-wide RNAi screens will be undertaken as the screening technology itself matures. Ultimately, the increasing ease with which such screens can be performed is likely to allow ‘smaller’ questions to be addressed. For example, high-content visual screens for loci that change the sub cellular localisation of STAT92E, or screens to identify the factors required for receptor predimerisation⁹ can be devised. Screens to identify factors required in Upd-secreting cells versus receiving cells or to identify differences in the factors that interact with the three Upd-like ligands are all possible. As such, assays will not only identify pathway interacting genes, but will start to characterise them on the basis of their function as well.¹⁴⁸ Such functional screening, in combination with the in vivo developmental genetic analysis of loci with respect to the known roles of JAK/STAT pathway signalling, are likely to represent a major future endeavour.

Another potential approach to identify regulators of JAK/STAT signalling is the use of chemical genetic studies utilising small molecules that modulate protein function in cell-based assays.¹⁴⁹ Although this has yet to be undertaken for Drosophila JAK/STAT signalling, complementary approaches combining chemical and genome-wide RNAi screens have the potential to rapidly identify the target through which a small molecule inhibitor of pathway signalling acts. By carrying out essentially identical small molecule and RNAi screens, the loss of protein activity elicited chemically should have similar readouts to RNAi depletion of the mRNA encoding the same protein. Given that the target of the dsRNA is known, the likely mechanism of action of small molecule inhibitors in whole cell conditions can be rapidly ascertained. Ultimately, practical experience will be needed to determine exactly how effective such cell scale drug discovery efforts prove to be (reviewed in ref. 150).