Windpipe Controls Drosophila Intestinal Homeostasis by Regulating JAK/STAT Pathway via Promoting Receptor Endocytosis and Lysosomal Degradation

The adult intestinal homeostasis is tightly controlled by proper proliferation and differentiation of intestinal stem cells. The JAK/STAT (Janus Kinase/Signal Transducer and Activator of Transcription) signaling pathway is essential for the regulation of adult stem cell activities and maintenance of intestinal homeostasis. Currently, it remains largely unknown how JAK/STAT signaling activities are regulated in these processes. Here we have identified windpipe (wdp) as a novel component of the JAK/STAT pathway. We demonstrate that Wdp is positively regulated by JAK/STAT signaling in Drosophila adult intestines. Loss of wdp activity results in the disruption of midgut homeostasis under normal and regenerative conditions. Conversely, ectopic expression of Wdp inhibits JAK/STAT signaling activity. Importantly, we show that Wdp interacts with the receptor Domeless (Dome), and promotes its internalization for subsequent lysosomal degradation. Together, these data led us to propose that Wdp acts as a novel negative feedback regulator of the JAK/STAT pathway in regulating intestinal homeostasis.


Introduction
The JAK/STAT pathway is evolutionarily conserved from Drosophila to mammals, and plays important roles in various developmental processes including cellular proliferation, innate immune response and stem cell development [1][2][3][4]. Dysregulation of the JAK/STAT pathway is associated with many human diseases, such as immune disorders and cancers [5][6][7]. Therefore, the JAK/STAT pathway is tightly controlled by various regulators and mechanisms to ensure proper signaling. While the core components of this pathway are well-documented, it is less understood how the duration of its signal activity is temporally regulated.
Drosophila is an excellent model to investigate the regulation of JAK/STAT signaling. Compared with various isoforms of the JAK/STAT pathway components in mammals [8][9][10], Drosophila has a relatively simple signal transduction cascade: a one-pass transmembrane receptor, Domeless (Dome) [11,12]; a tyrosine JAK kinase, Hopscotch (Hop) [13]; a transcription factor, STAT92E [14,15]; and three different ligands including Unpaired (Upd) [16], Upd2 [17], and Upd3 [18]. In the canonical pathway, binding of Dome receptor with its extracellular ligands induces Dome dimerization or oligomerization, which leads to juxtaposition of Hop. Hop molecules cross-phosphorylate each other and then phosphorylate Dome to generate docking sites for cytoplasmic STAT92E. Once bound to the Dome/Hop complex, STAT92E molecules are phosphorylated, form dimers, and then translocate into the nucleus, where they bind to defined STAT92E binding sites, and regulate the transcription of downstream target genes [1,19]. This signaling transduction is under tight control at multiple steps to avoid improper signal activation [1,9]. Several negative feedback regulators such as Socs36E and Ptp61F are identified to be involved in switching off JAK/STAT signaling during developmental processes [20][21][22].
Drosophila Windpipe (Wdp) is a single-pass transmembrane protein containing four leucine-rich repeats (LRR) in the extracellular domain, and is highly expressed in the developing trachea [47]. Currently, the biological function of this protein has not been defined. In this study, we have identified wdp as a novel component of the JAK/STAT pathway. We showed that Wdp is positively regulated by JAK/STAT signaling. Loss of wdp results in disruption of midgut homeostasis under physiological conditions, and potentiates tissue regeneration under damage conditions. Conversely, ectopically expressed Wdp negatively regulates JAK/STAT signaling. Importantly, we demonstrate that Wdp can promote Dome internalization and subsequent lysosomal degradation. Together, we propose that Wdp controls intestinal homeostasis by interfering with JAK/STAT signaling activity via a negative feedback mechanism.

Wdp expression is positively regulated by JAK/STAT signaling in Drosophila midguts
Although the JAK/STAT pathway is required for regulating midgut homeostasis under physiological conditions and damage-induced tissue regeneration, how JAK/STAT signaling executes its functions during these processes remains largely unknown. So far, only a few STAT92E target genes including Socs36E, zfh1, and chinmo, have been identified [48][49][50]. To further explore the regulatory mechanism of the JAK/STAT pathway in midguts, we performed ChIP-Seq experiments aimed at identifying novel downstream targets of the JAK/STAT pathway. ChIP-Seq experiments were carried out in intestines ectopically expressing Upd and STAT92E in the progenitor cells (esg ts >upd; STAT). In these experiments, we identified about 200 candidates with at least one peak (p<0.01) in the gene regulatory region. The previously well-characterized JAK/STAT downstream targets, including Domeless [51], Socs36E [20], and STAT92E [52,53], were recovered in our experiments (S1 Table), indicating that our ChIP approach is workable to identify potential novel targets.
From the candidate genes, wdp, a gene previously shown to be highly expressed in the developing trachea [47], was identified. Wdp was ranked in the top 10% through our bioinformatics analysis of ChIP results. At least 3 significant peaks (p<0.01) containing conserved STAT92E binding sites (TTCN3/4GAA) [15] were found in the 5' UTR and genomic region of wdp ( Fig 1A). Moreover, wdp mRNA levels as determined by RT-qPCR were increased in response to ectopic JAK/STAT signaling in esg ts >upd intestines ( Fig 1B). To further analyze the transcriptional regulation of wdp by JAK/STAT signaling, we identified four potential STAT92E binding sites (BS1-, BS2-, BS3-and BS4), and generated luciferase reporters that contain these potential binding sites. With the addition of Upd expressing S2 cells, the luciferase activity in cells transfected with BS2-, BS3-or BS4-luciferase constructs was obviously increased ( Fig 1C). These results suggest that the expression of Wdp might be regulated by JAK/ STAT signaling through BS2, BS3 and BS4 binding sites.
Next we wanted to determine whether the expression of Wdp is regulated by JAK/STAT signaling in vivo in Drosophila posterior midgut. First, the expression pattern of Wdp was examined by immunostaining with anti-Wdp antibody. The specificity of anti-Wdp antibody was verified (S1G-S1I Fig). Then we determined Wdp expression pattern in midguts and imaginal discs (S1 Fig). Wdp was ubiquitously expressed in wild-type intestines (Fig 1D, 1D' and S1A-S1D' Fig). When we used esg ts to overexpress Upd and STAT in progenitor cells, we found Wdp protein levels were increased around the GFP + clusters (Fig 1E and 1E'). Furthermore, we blocked JAK/STAT signaling activity by expressing dome RNAi or STAT RNAi, and found Wdp expression was reduced in dome RNAi/STAT RNAi expressing cells (Fig 1F-1G  ChIP analysis was performed to monitor the binding of STAT92E to wdp genomic regions with STAT92E antibody using adult intestines expressing Upd and STAT92E under the esg ts driver for 10 days at 29°C. The localization of four putative STAT92E-binding sites (BS1-4) is indicated by a black square frame. The square boxes with green, red or blue colors represent putative STAT92E binding sites localized in wdp genomic region with 2, 3 or 4 spacers respectively. (B) wdp mRNA expression was obviously increased in esg ts >upd intestines at 29°C for 7 days using RT-qPCR quantification. Mean ± SD are shown. **p<0.01. (C) The relative activity of the indicated luciferase vectors, which contain different putative STAT92E binding sites (BS1-4) from wdp genomic regions, upon addition of Upd expressing cells. Mean ± SD are shown. *p<0.1, **p<0.01. (D and D') Wdp (red, by Wdp) is ubiquitously expressed in both small progenitor cells and large nuclei ECs in control midguts at 29°C for 7 days. (E and E') Wdp expression (red, by Wdp) was significantly increased around the GFP+ clusters (arrowheads) in esg ts >upd, STAT midguts at 29°C for 7 days. (F and F') Wdp expression (red, by Wdp) was reduced in the Flip-out clones (arrowheads) knocking down Dome at 29°C for 7 days. Square box is the enlarged image of the position labeled by yellow arrowhead. (G and G') Wdp expression (red, by Wdp) was reduced in the Flip-out clones (arrowheads) knocking down STAT at 29°C for 7 days. Square box is the enlarged image of the position labeled by yellow arrowhead. Blue indicates DAPI staining. Scale bars, 20μm. in JAK/STAT signaling deficient ISCs. Taken together, these data indicate that wdp, as a putative target of STAT92E, is positively regulated by JAK/STAT signaling in Drosophila intestines.
Loss of wdp disrupts midgut homeostasis under physiological conditions and potentiates tissue regeneration under damage conditions Next, we examined the possible functions of wdp in midgut homeostasis. We generated 2 alleles of wdp mutants by imprecise excision of P{wHy}wdp DG23704 (S1E Fig) and selected wdp 1 for further experiment. wdp 1 is likely a functional null mutant, as half of the wdp coding sequence is removed. Consistently, wdp transcription in wdp 1/1 homozygotes was abolished compared with WT (S1F Fig). Homozygous wdp 1/1 flies are semi-lethal with a few escapers displaying no visual phenotypes.
To examine the function of Wdp in the posterior midgut, we used esg-lacZ, Dl-lacZ and Su (H)GBE-lacZ to mark progenitor cells, ISCs and EBs respectively. Compared with the controls, the number of esg-lacZ positive cells was significantly increased in wdp 1/1 mutant intestines (Fig  2A, 2B and 2Q). Similar phenotype in wdp 1/2 trans-heterozygotes was observed (S3E-S3G Fig), excluding the existence of possible background mutations. We also found an increased number of Dl-lacZ and Su(H)GBE-lacZ positive cells in wdp 1/1 intestines (Fig 2C-2F and 2Q). Moreover, the number of 10×STAT GFP positive cells was obviously increased (Fig 2G and 2H). In addition, 10×STAT GFP seems to appear in the large putative EC cells (arrows in Fig 2H). These results suggest that midgut homeostasis is disrupted upon wdp loss under normal conditions.
We further examined the roles of Wdp under damage conditions. Midgut regeneration of wdp 1/1 adults was monitored in response to dextran sulfate sodium (DSS) feeding, which is used to investigate ISC proliferation and tissue regeneration upon damage [54]. Adult flies aged at 3 or 4 days were treated with 3% DSS for 4 days. Under DSS treatment, wdp 1/1 adults showed dramatic hyperplasia and extensive multilayering of the midgut epithelium compared with controls ( Fig 2I-2P and 2R). Moreover, the number of progenitor cells and 10×STAT GFP positive cells was also increased (Fig 2I-2P), indicating that tissue damage induced midgut regeneration was abnormally enhanced in the absence of wdp. Collectively, these data indicate that loss of wdp disrupts midgut homeostasis under normal conditions and potentiates tissue regeneration under damage conditions.

Wdp inhibits ISC proliferation and restricts ISC overproliferation induced by ectopic JAK/STAT signaling
We further examined whether Wdp is involved in regulating ISC activity. First, mosaic analysis with repressible cell marker (MARCM) approach was used to generate GFP positively marked clones for wdp 1 mutants [55]. The control ISC clones contained an average of 7-8 cells per clone 6 days after clone induction (ACI) (Fig 3A, 3D and 3E). In contrast, the wdp 1 mutant ISC clones contained up to 30 cells per clone 6d ACI (Fig 3B, 3D and 3F). Moreover, the number of Dl/Pros positive cells, which mark ISC/ee respectively, was increased in wdp 1 mutant ISC clones compared with controls ( Fig 3A and 3B). In addition, Brdu incorporation within wdp 1 mutant clones was also enhanced (Fig 3E, 3F and 3H), suggesting that loss of Wdp led to the increased proliferation of ISCs.
We also examined the role of wdp in regulating ISC proliferation under damage conditions. Adult flies carrying MARCM clones of various genotypes were fed with DSS. Consistently, we found the size of wdp 1 mutant clones was obviously enlarged, and the number of Dl/Pros positive cells was also increased within wdp 1 clones compared with controls under damage conditions ( Fig 3I, 3J and 3L). Similarly, the number of PH3 positive cells per gut was enhanced in the intestines containing wdp 1 mutant clones (Fig 3M, 3N and 3P). It is important to mention  that the increased ISC proliferation of wdp 1 mutants under physiological conditions or damage conditions could be rescued by simultaneous wdp expression (Fig 3C, 3G, 3K and 3O), confirming that the observed defects were derived from loss of Wdp activity.
We also knocked down Wdp in progenitor cells by expressing wdp RNAi driven by esg ts . A mild increase in GFP + cells was observed when wdp was knocked down in the progenitor cells ( Fig 3Q and 3R). Moreover, when the wdp knockdown flies were treated with DSS, the number of GFP+ cells was increased, and the midgut epithelium exhibited extensive multilayering compared with controls ( Fig 3S-3V). This suggests that the tissue damage induced ISC proliferation was enhanced in the absence of wdp. Taken together, the above data derived from wdp mutant clones and RNAi experiments indicate that wdp restricts ISC proliferation under normal and regenerative conditions.
As wdp expression could be induced by ectopic JAK/STAT signaling in the intestines, we examined its role under high levels of JAK/STAT signaling. When upd was ectopically expressed using the esg ts driver, ISC proliferation was obviously increased (Fig 3W and 3X). Surprisingly, simultaneous knockdown of wdp enhanced the excessive ISC proliferation induced by ectopic Upd expression, as determined by the increase in GFP + cells and a thickened midgut epithelium ( Fig 3Y-3Z'). These results indicate that Wdp restricts ISCs from excessive proliferation caused by ectopic JAK/STAT signaling.

Wdp downregulates JAK/STAT signaling activity
To gain insights into the mechanistic role of wdp in the JAK/STAT pathway, we examined its potential regulation of JAK/STAT signaling in other developmental processes. Eye imaginal disc is a good model to investigate JAK/STAT signaling [56,57]. We detected the expression of Wdp in 3 rd instar eye discs (S1J Fig). 10×STAT GFP is used as the signaling readout [58], which is detected throughout the posterior part of early 3rd instar eye discs ( Fig 4A and 4A'). When wdp was ectopically expressed using mirror-Gal4 in the dorsal compartment, the levels of 10×STAT GFP were obviously reduced ( Fig 4B and 4B'). Consistently, the activity of 10×STAT GFP was also decreased in the flip-out clones overexpressing Wdp compared with surrounding WT cells (Fig 4C-4D'). On the contrary, we detected enhanced expression region of 10×STAT GFP in the wdp 1/1 homozygous eye discs (S4A- S4C Fig). Moreover, Wdp knockdown using mirror-gal4 also led to the enlarged 10×STAT GFP region in the dorsal compartment of 3 rd instar eye discs (S4D- S4D‴ Fig). These data indicate that Wdp negatively regulates JAK/STAT signaling in eye discs. In contrast, Wingless (Wg), Hedgehog (Hh), and Decapentaplegic (Dpp) signaling pathways were not affected when wdp was ectopically expressed (S5 Fig), suggesting that Wdp mainly regulates JAK/STAT signaling in imaginal discs.
As mentioned above, loss of Wdp could potentiate Upd induced ISC proliferation ( Fig 3W-3Z'), implying its regulation of JAK/STAT signaling in posterior midguts. To verify this containing different MARCM clones under DSS treatment. Mean±SD are shown. n = 11-15 intestines. **p<0.01. (Q and R) The number of GFP+ cells was mildly increased upon knockdown of wdp using Su(H) GBE-lacZ; esg ts driver (R) compared with controls (Q) at 29°C for 7 days. (S-V) Adult flies of esg ts or esg ts >wdp RNAi were treated with 3% DSS for 4 days at 29°C. Cross-section of midgut epithelium with the indicated genotypes was shown in T and V. Upon DSS treatment, the number of GFP+ clusters (S and U) as well as the thickness of midgut epithelium (T and V) from esgts >wdp RNAi midguts were significantly increased compared with controls. The intestinal lumen is indicated by white double-headed arrows and the intestinal wall by yellow brackets. (W-Z) The overproliferation of ISCs caused by upd expression (W and X) using the esg ts driver was strikingly enhanced with simultaneous wdp knockdown (Y and Z). Cross-section of midgut epithelium with the indicated genotypes was shown in X and Z. (Z') Quantification of midgut epithelium thickness (μm) with the indicated genotypes. Mean±SD are shown. n = 7-10 intestines. **p<0.01. Blue indicates DAPI staining. Scale bars, 20μm. hypothesis, we examined JAK/STAT signaling by detecting the activity of 10×STAT GFP in RFP positively marked intestinal MARCM clones. In wdp 1 mutant clones, the levels of 10×STAT GFP were mildly increased when compared with surrounding wild-type cells (Fig 4F and 4F'). Moreover, in wdp 1 mutant clones we detected 10×STAT GFP in the large cells (putative EC cells) as well as in small progenitors (arrows in Fig 4F'), implying JAK/STAT signaling maybe abnormally activated in ECs. However, in control clones there seems no obvious difference of 10×STAT GFP activity between RFP+ clones and RFP-cells. Besides, 10×STAT GFP was restricted in small progenitors (Fig 4E and 4E'). Furthermore, we examined the destabilized 10×STAT DGFP reporter [58] in wdp 1/1 intestines and found the activity of 10×STAT DGFP was obviously increased compared with controls (S4E-S4F ' Fig). Consistently, 10×STAT DGFP also appeared in large ECs in wdp 1/1 homozygotes (S4E-S4F' Fig). These results suggest that JAK/STAT signaling was upregulated in the absence of wdp. Meanwhile, we used the esg ts driver to overexpress wdp in the progenitor cells and found mild increase of EBs under normal conditions (Fig 4G-4I). However, when esg ts >wdp flies were treated with DSS, the damage-induced tissue regeneration was suppressed. In contrast to the large GFP+ clusters containing both small diploid progenitors and large polyploid cells in the controls, there were no large GFP+ cells observed in the clusters of the esg ts >wdp midguts (Fig 4J and 4K). This was reminiscent of intestines with deficient JAK/STAT signaling by STAT or Dome knockdown under DSS treatment (Fig 4L and 4M). Altogether, these data suggest that Wdp could interfere with JAK/STAT signaling in posterior midguts.
We further assessed the activity of 10×STAT luciferase reporter in S2 cells transfected with wdp. Consistently, we found Wdp expression was able to suppress the basal 10×STAT luciferase as well as Upd-induced upregulation of 10×STAT luciferase activity ( Fig 4N). Taken together, the above data derived from different tissues indicate that Wdp is a negative regulator of the JAK/STAT signaling pathway.

Wdp acts downstream of Upd but upstream of Hop
To determine the level at which Wdp modulates JAK/STAT signaling, we examined the epistatic relationship between Wdp and the JAK/STAT pathway components. When hop was ectopically expressed using mirrorGal4 in eye discs, we observed increased JAK/STAT activity, as determined by an expanded expression region of 10×STAT GFP in the dorsal compartment which is marked by CD8-mRFP ( Fig 5A and 5A', arrow). Simultaneous expression of wdp failed to suppress the elevated JAK/STAT signaling caused by hop overexpression (Fig 5B, 5B' and 5C), suggesting that Wdp acts upstream of Hop. Consistent with previous study [28], ectopic expression of Hop in midguts using the esg ts driver led to weak expansion of esg positive cells (Fig 5D and 5D'). ISC over-proliferation caused by hop expression was not affected in the presence of wdp (Fig 5E, 5E' and 5F). Thus, these epistatic experiments performed in both the midguts and eye discs placed Wdp upstream of Hop.
To further confirm the genetic epistasis between Wdp and Hop, we assessed the activity of 10×STAT luciferase reporter in S2 cells cotransfected with wdp and hop-V5 vectors. Consistently, the increased 10×STAT luciferase activity resulting from hop expression could not be blocked by cotransfection of wdp. However, ectopic expression of wdp was able to suppress the enhanced activity of 10×STAT luciferase caused by Upd expression, indicating that Wdp acts downstream of Upd. In addition, increased 10×STAT luciferase activity resulting from simultaneous transfection of UAS-upd, dome, and hop was significantly suppressed by cotransfection with wdp ( Fig 5G).

Wdp interacts with Domeless and promotes its endocytosis and lysosomal degradation
The JAK/STAT pathway is under tight control at various steps by different regulators and regulatory mechanisms. Since Wdp functions upstream of Hop and downstream of Upd, we examined the possible regulation of Wdp to the Dome receptor. Importantly, the total levels of Dome were reduced in S2 cells coexpressing Wdp (Fig 5H), implying that Wdp may affect the stability of Dome. Previous study showed that JAK/STAT signaling is negatively regulated by endocytic trafficking [59]. One possibility is that Wdp promotes Dome endocytosis for subsequent degradation. To test this, we performed the following experiments. First, when S2 cells were transfected with Dome-HA alone, Dome was mainly localized on the cell membrane, with a few punctates detected in the cytoplasm (Fig 6B, 6B' and 6D). However, when co-expressed with Wdp, the majority of Dome was present in the cytoplasm as vesicle-like punctates ( Fig  6C, 6C' and 6D), implying the endocytosis of Dome is enhanced. To further determine whether Wdp could promote Dome endocytosis, we carried out time-lapse imaging experiments. After the live S2 cells expressing Dome-GFP were incubated with endocytic dye FM 4-64 at room temperature for 1h, we examined the dynamics of Dome-GFP and chased its co-localization with FM 4-64 at different time points. As shown in Fig 6E, in the absence of Wdp the majority of Dome-GFP was localized on the cell membrane and little co-localization with FM 4-64 was detected. When cotransfected with wdp, Dome-GFP was mainly observed as intracellular particles, which were partially co-localized with FM 4-64 (see arrowheads in Fig 6F). Furthermore, we observed newly formed Dome-GFP endocytic vesicles trafficking from the cell membrane (S1 Movie). All of these tissue culture data based on the overexpressed Wdp suggest that Wdp can promote Dome internalization.
We further examined the co-localization of Dome with various vesicular markers in S2 cells. Little co-localization was observed between intracellular Dome and the cis-Golgi apparatus as marked by GM130 (Fig 6G and 6G'), implying the presence of intracellular Dome-GFP was not due to defects in exocytosis. However, a large amount of Dome intracellular particles were colocalized with the early endosome marker Rab5 (Fig 6H and 6H'). In addition, high levels of Dome-GFP were present in the lysosomes as labeled by lyso-tracker in live cells (Fig 6I and 6I'). Moreover, the appearance of Dome intracellular punctates observed in Wdp-coexpressing cells was partially suppressed by Rab5 dsRNA treatment (S7 Fig), indicating that the accumulation of Dome intracellular punctates was a result of Rab5-mediated endocytosis. Interestingly, the subcellular localization of other membrane proteins such as CD8-mRFP or GFP-GPI was not affected when coexpressed with Wdp (S8 Fig), suggesting that Wdp specifically promotes Dome endocytosis.
We also examined whether Wdp functions similarly in vivo. We generated flip-out clones overexpressing Dome alone or along with Wdp in the wing and eye imaginal discs. Consistent with previous reports [59][60][61], Dome-V5, as a transmembrane receptor, was mainly localized on the cell membrane, and also formed some intracellular punctate structures which could correspond to endocytic vesicles (Fig 7A-7A‴). Importantly, coexpression with Wdp caused a significant change in the subcellular localization of Dome (Fig 7B-7B‴). In the presence of Wdp, Dome was totally disappeared from the cell membrane, but was found as intracellular punctates (Fig 7B-7B‴), where they partially colocalized with the early endosome marker Rab5 (Fig 7C-7C‴). In addition, we also observed the same phenomena in the eye discs (S9 Fig). Therefore, these data suggest that enhanced Wdp expression could promote Dome internalization in the wing and eye discs.
To further determine whether Dome is degraded in the lysosomes after being internalized, S2 cells were treated with Chloroquine (CQ), a lysosomal inhibitor. Interestingly, we found the reduction of Dome levels caused by Wdp co-expression was restored upon Chloroquine treatment but not upon MG132 treatment, suggesting that the internalized Dome is undergoing lysosomal degradation rather than proteasome degradation (Fig 7D). This result is in agreement with the recently published paper showing that Dome undergoes lysosomal degradation [62]. Taken together, our data indicate that Wdp functions to promote Dome endocytosis through the endosomes, and subsequently to the lysosomes for degradation.
We then investigated whether Wdp interacts with Dome to promote its endocytosis. We transfected HA-tagged Dome and wdp (or V5-tagged wdp) into S2 cells and found HA-tagged Dome could co-immunoprecipitate with both Wdp and V5-tagged Wdp (Fig 7E). These data indicate that Wdp interacts with Dome in transfected cells. Taken together, our data suggest that Wdp interacts with Dome and then promote its internalization from the cell membrane into the early endosomes, and finally to the lysosomes for degradation. In this way, Wdp attenuates JAK/STAT signaling to avoid uncontrolled signaling activation.

Discussion
In this study we have provided evidence that the LRR protein Wdp is a novel component of the JAK/STAT pathway that acts in a negative feedback manner to modulate JAK/STAT signaling activity and control intestinal homeostasis. Our in vivo and in vitro data indicate that wdp expression levels are positively regulated by JAK/STAT signaling. Loss of wdp disrupts midgut homeostasis under both physiological and damage conditions. Conversely, ectopic expression of Wdp leads to the reduction of JAK/STAT signaling activity. Mechanistically, we show that Wdp can interact with Dome, and promote Dome internalization and lysosomal degradation, thereby reducing JAK/STAT signaling activity.

Wdp controls intestinal homeostasis through interfering with JAK/STAT signaling activity
Midgut homeostasis is tightly controlled by different signaling pathways. During tissue damage, JAK/STAT, EGFR, JNK and Hippo signaling pathways are required for ISC proliferation and midgut regeneration [26,30,32,33,46,[63][64][65]. On the other hand, other signaling pathways, such as BMP signaling, may negatively regulate intestinal homeostasis after injury, although there exists some controversy about the function of BMP signaling during Drosophila intestinal development [36][37][38][39]. However, the mechanism of how ISC activity returns to quiescence after injury remains largely unknown. Here, we demonstrate that Wdp controls intestinal homeostasis through interfering with JAK/STAT signaling activity to avoid tissue hyperplasia.
Our data indicate that loss of Wdp disrupts midgut homeostasis under normal conditions and potentiates tissue regeneration under damage conditions (Figs 2 and 3). The proliferation rate of ISCs mutant for wdp is increased, while the differentiation of EC and ee cells is not inhibited (Fig 3 and S3A-S3D Fig). In addition, ectopic Wdp expression suppressed the damage induced tissue regeneration. Our data further demonstrate that Wdp controls intestinal homeostasis through interfering with JAK/STAT signaling activity (Fig 4). First, Wdp acts as a JAK/STAT downstream target and its expression levels are positively regulated by JAK/STAT signaling (Fig 1 and S2 Fig). Second, Wdp functions in a negative feedback loop to modulate JAK/STAT signaling activity (Fig 4 and S4 Fig). It is interesting to note that JAK/STAT signaling is mainly activated in ISCs and EBs [26]. However, we found that Wdp expression levels seem higher in ECs compared with progenitor cells (S1A-S1D ' Fig). One explanation is that low levels of Wdp in progenitors may guarantee high levels of JAK/STAT signaling, while high levels of Wdp in ECs may serve to reduce Dome levels thereby making ECs insensitive to Upd ligands. Consistent with this view, previous work showed that Dome is mainly expressed in the progenitors but not in their progeny [26]. Moreover, we found Wdp knock down using EC specific Myo1A ts also leads to the disruption of midgut homeostasis and the presence of 10×STAT GFP in putative EC cells (S4G-S4H' Fig), suggesting that JAK/STAT signaling is activated upon wdp knockdown in ECs. On the other hand, we found Wdp expression was reduced but not totally eliminated in JAK/STAT signaling deficient cells (S2 Fig), suggesting that the basal level of Wdp in intestines (especially in ECs) may also be regulated by other regulatory mechanisms or signaling pathways. Further experiments are needed to clarify this issue.
It's important to mention that Wdp expression could be induced under injury conditions, such as DSS or bleomycin treatment (S2I Fig). Consistent with our results, two recent studies also identified wdp as an upregulated gene upon Ecc15 and Pseudomonas entomophila (P.e) infection through their microarray data respectively [44,66]. These stress conditions are also associated with the activation of JAK/STAT signaling [26,46]. Therefore, their findings are consistent with our view that Wdp can be induced by the JAK/STAT pathway and then restrict its signaling activity in restoring intestinal homeostasis after tissue damage.
We further demonstrated the regulation of Wdp to JAK/STAT signaling in eye discs and S2 cells. 10×STAT GFP activity was decreased in eye discs overexpressing Wdp (Fig 4A-4D') while increased in wdp mutant eye discs (S4A-S4D‴ Fig). Similarly, a reduction of 10×STAT luciferase activity was also observed in S2 cells transfected with Wdp ( Fig 4N). Thus, we propose that Wdp is also likely to modulate JAK/STAT signaling activity for proper development of other tissues.
Taken together, we conclude that Wdp is involved in controlling intestinal homeostasis through interfering with JAK/STAT signaling in a negative feedback manner.

Wdp inhibits JAK/STAT signaling through promoting Dome endocytosis
Previously, several studies have addressed the roles of endocytosis in regulating JAK/STAT signal pathway. The Noselli lab found blocking internalization led to an inhibition of JAK/STAT signaling activity [61], while the Zeidler group reported the opposite results [59]. Moreover, several recent studies demonstrate that loss of ept/tsg101 or Rabex-5, two endocytic tumor suppressor genes, also induced JAK/STAT signaling activation and tissue overgrowth [67,68]. Yet, the regulatory mechanism of how Dome receptors are internalized remains largely unknown. Here we demonstrate that Wdp promotes Dome endocytosis and subsequent lysosomal degradation. First, in S2 cells Wdp ectopic expression induces the formation of Dome endocytotic vesicles which were colocalized with the early endosome marker and lysosome marker (Fig 6). Second, we found Wdp expression can also promote Dome endocytosis in wing and eye imaginal discs. Furthermore, the decreased Dome levels caused by Wdp expression can be suppressed by CQ treatment (Fig 7). All of these data argue that Wdp acts to promote Dome endocytosis from the cell membrane, first into the early endosomes, and finally into the lysosomes for degradation. Previous work are mainly about Dome receptors undergo ligands induced endocytosis [59,61], while in this work we show that Wdp is able to promote Dome internalization in a Upd independent manner. Our coimmnoprecipitation data indicate Wdp can interact with Dome ( Fig 7E). Moreover, S1 Movie shows that Dome-GFP are aggregated on the cell membrane before they are internalized in the presence of Wdp. Therefore, one possible mechanism is that Wdp interacts with Dome, induces the aggregation of Dome on the cell membrane and then promotes Dome endocytosis. Further experiments are needed to define the detailed mechanism.
A model for the role of Wdp in regulating JAK/STAT pathway during tissue damage On the basis of our findings, the following model is proposed (Fig 8A and 8B): Wdp regulates intestinal homeostasis through its modulation of JAK/STAT signaling. Under physical conditions, low levels of Wdp in progenitors are needed to maintain proper levels of JAK/STAT signaling activity, while high levels of Wdp in ECs reduce Dome levels to ensure these cells are insensitive to JAK/STAT signaling. When midgut epithelium is damaged by environmental challenges, high levels of JAK/STAT signaling activity are induced to replenish the damaged midgut. Then Wdp expression is highly induced in the intestines to reduce Dome levels, thereby switching off the overactivated JAK/STAT signaling. Through this way, ISC proliferative rate returns to normal levels to avoid tissue hyperplasia. While other mechanisms or regulators are likely to be involved in regulating intestinal homeostasis, our data suggest that Wdp is one of the key regulators in this process through interfering with JAK/STAT signaling activity. During tissue damage or pathogen infection induced midgut regeneration, the JAK/STAT pathway is highly activated (A). In response to extracellular signaling, STAT92E dimers translocate into the nucleus, bind to its consensus binding sites at the genomic region of wdp and then promote its transcription. Newly synthesized Wdp protein is transported to the cell membrane, where it interacts with Dome and promotes Dome internalization from the cell membrane finally into the lysosomes for subsequent degradation (B). Through this negative feedback manner, Wdp restricts the signal duration and ensures JAK/STAT signaling returns to the normal levels after injury in Drosophila intestines. doi:10.1371/journal.pgen.1005180.g008

Constructs
pUAST-wdp, wdp-V5 and wdp-RFP were constructed by cloning the wdp cDNA into pUAST-attB, pUAST-attB-V5 and pUAST-RFP vectors respectively. Dome-V5 and hop-V5 were constructed by insertion of the coding region, from transgenic lines UAS-Dome (a gift from S. Hou) and UAS-Hop3w (a gift from Rongwen Xi), into pUAST-V5-attB vector. Dome was excised from pUAST-dome-V5 and inserted in pUAST-3HA or pUAST-GFP to generate pUASTdome-3HA or pUAST-dome-GFP vectors respectively. pAC5.1-upd-V5 was made by cloning upd cDNA into pAC5.1-V5 vectors. UAS-wdp RNAi was made by cloning annealed oligos ctag-cagtAGAGGAGAGCGATGTTAGACCtagttatattcaagcataGGTCTAACATCGCTCTCCTCT gcg and aattcgcAGAGGAGAGCGATGTTAGACCtatgcttgaatataactaGGTCTAACATCGCTC TCCTCTactg into EcoRI/ NheIsites of pWalium20 vector [69] and was confirmed to be functional (S1H and S1I Fig). 10×STAT luciferase vector was generated by subcloning firefly luciferase gene into 10×STAT Gal4 vector. To determine the binding sites of STAT92E in wdp genomic regions, we generated luciferase vectors containing putative binding regions based on the ChIP results. Primers used for constructing luciferase vectors can be found in S1 Text.

Wdp antibody generation
We generated polyclonal antibody specific for Drosophila Wdp protein by choosing the hydrophilic polypeptides 480-550aa and 591-661aa as the antigen. GST-tagged Wdp antigen was expressed in E. coli BL21 (DE3) and purified with GST affinity chromatography. Using this antigen, we generated and further separated mouse polyclonal antibody of Drosophila Wdp.

MARCM clone
MARCM clones in the adult midguts were induced by heat-shocking 3-4 day-old females for 75 min at 37°C. Adult guts were dissected and examined 6 days after clone induction.

Flip-out clone
For Flip-out clones in adult midguts, crosses were set up and cultured at 25°C. Flies were heatshocked at 37°C for 75 minutes 3 days after eclosion and then dissected 6 days later. For Flipout clones in wing or eye discs, crosses were kept at 25°C. Larvae were heat-shocked for 90 minutes at 37°C 48 hours after egg deposition and dissected at late 3rd instar larva stage.

Feeding experiments
Female adult flies at age 3 or 4 days were used to perform feeding experiments. Flies were cultured in an empty vial containing chromatography paper wet with 3% dextran sulfate sodium (MP Biomedicals) or 25μg/mL bleomycin (Sigma) dissolved in 5% glucose solution with heat inactivated yeast for 4 days at 29°C.
Antibodies used for immunostaining, immunoprecipitation, and western blotting Fixation and antibody staining in imaginal discs were performed as described [70]. Fixation and antibody staining in cultured cells were performed as described [71]. Fixation and antibody staining in midguts were performed as described [37]. Primary antibodies used for the immunostaining were: mouse anti-Wdp

Brdu incorporation
Adult flies with MARCM clones were reared on standard corn meal food with 0.2mg/ml BrdU (Sigma) at 25°C for 4 days before dissection. Then midguts were treated with 3M HCl at 37°C for 30 min, and the reaction was stopped by washing with PBT twice.

RT-qPCR
RNA was extracted from 20 intestines of female adults using RNA pre pure kit (TIANGEN) and complementary DNA (cDNA) was synthesized with the M-MLV Reverse transcriptase (Promega). qPCR was performed using GoTaq qPCR Master Mix kit (Promega) on CFX96 Real-time PCR system(Bio-Rad). Experiments were performed in 3 biological independent replicates, each also contained 3 repeats. All the results are shown as Mean±SD of the biological replicates. Ribosomal gene RpL11 was used as normalization control. Primers used for qPCR are listed in S1 Text.

ChIP-Seq
The identification of STAT92E target genes in adult intestines was carried out through ChIP assay and ChIP-high throughput sequencing technique. JAK/STAT signaling was activated using esg ts to overexpress Upd and STAT92E at 29°C for 10 days. Then about 400 adult intestines were dissected and cross-linked with 1% formaldehyde for 15 minutes. After washing process to remove the formaldehyde, intestinal tissue was lysed with RIPA buffer which contains 1% SDS on ice for 30 minutes. The sonication of chromatin was performed using the Covaris (AFA) system with 3% power output for 5 minutes each on 100μl lysate. STAT92E-bound chromatin fragments were enriched by immunoprecipitation with mouse raised STAT92E antibody. Most chromatin fragments resulting from sonication occurred between 200 and 400 bp. The process of dilution, antibody incubation, protein G pull down, beads washing, DNA complex elution, de-link, RNAse A / proteinase K digestion and DNA extraction are all performed according to standard protocols. The high throughput sequencing process was carried out using the Illumina solexa system.

Cell culture, transfection, coimmunoprecipitation and western blotting
Drosophila S2 cells were maintained at 25°C in HyQ SFX-insect cell culture medium. All transfection experiments were carried out using Effectene Transfection Reagent (QIAGEN).
For Wdp and Dome interaction experiments, S2 cells were transfected in 60mm dishes with 200ng Arm-Gal4, 200ng pUAST-dome-HA and 200ng pUAST-V5 control vector, or pUASTwdp (with or without V5 tag). Then S2 cells were lysed in 200μl RIPA buffer without SDS on ice for 30 minutes. RIPA buffer includes 50 mM Tris-HCl (pH 7.8), 150 mM NaCl, 5 mM EDTA (pH 8.0), 0.5% Triton X-100, 0.5% NP-40, 0.5%DOC, complete protease inhibitor cocktail tablets (Roche), and phosphatase inhibitor cocktail tablets (Roche). After centrifugation, the suspension of lysates was added with antibody and incubated for 3h at 4°C, and then added with BSA blocked protein G beads and rotated overnight at 4°C. The immunocomplexes were collected by centrifugation and washed with 1 ml of RIPA buffer three times.
For western blotting, immunoprecipitated proteins were separated in SDS-PAGE and then blotted onto PVDF membranes. The membranes were stained with primary antibody overnight at 4°C, as anti-V5, anti-HA, anti-Wdp to detect interaction between Wdp and Dome. Antibody HA was used to examine the effects of Wdp on Dome levels. Followed by washing, PVDF membranes were incubated with secondary antibodies carrying infrared fluorophore, and then analyzed using Odyssey system (GENE).

Luciferase assay
S2 cells were seeded in 24-wells plate. Cells in each well were transfected with 5ng Renilla-luciferase, 30ng 10×STAT-luciferase reporter (or other reporters) and 30 ng other vectors as shown in figures. After 12h, cells were mixed with Upd transfected cells. After an additional 48h, S2 cells were washed with PBS and then lysed using Passive Lysis Buffer (Promega). Firefly-luciferase and Renilla-luciferase activity were detected using GLOMA Multi Detection System (Promega). All the results are from twice independent experiments each containing 3 repeats.

Live cell imaging
For labeling of endocytic vesicles, S2 cells were treated with 5μg/ml FM4-64 (Molecular Probes, Inc.) at 25°C for 1h. S2 cells were then washed twice with medium and then incubated at 25°C for another 1h. Then 200ul of cell suspension was applied to a microscope slide. Images were captured by a Zeiss LSM780 inverted confocal microscope and movies were made from timelapse images using Corel Video Studio X4. For labeling of lysosomes, S2 cells were incubated with cell culture containing lyso-tracker (Invitrogen) at a final concentration of 50 nM at 25°C for 2h. In FRT 82B control MARCM clones (E and E'), Wdp was uniformly expressed between GFP+ clone cells (arrow in E') and GFP-cells (arrowhead in E'). However, Wdp expression was reduced in STAT92E 06346 clone cells (arrows in F') compared with surrounding WT cells (arrowheads in F'). In addition, we generated Notch 264-39 mutant clones and detected Wdp expression mainly on the plasma membrane of ISC clusters (G and G'). As shown in G', Wdp was also uniformly expressed between Notch 264-39 clones (arrow in G') and GFP-cells (arrowheads in G'). In STAT92E mutant clone cells with simultaneous Notch RNAi, Wdp expression levels (arrows in H') were reduced compared with Notch mutant clones (arrows in G'). Furthermore, Wdp expression was reduced in clone cells (arrows in H') compared with surrounding WT cells (arrowheads in H'). (I) The mRNA levels of wdp were increased under damage conditions using RT-qPCR quantification. w1118 flies aged at 3-4 days were treated with 3% DSS or 25ug/ml bleomycin at 29°C for 4 days. Mean ± SD are shown. GFP are indicated by white double-headed arrows. (C) Quantification of the expression region of 10×STAT GFP in WT and wdp 1/1 homozygous early 3 rd instar eye discs. Mean±SD are shown. n = 6-9. ÃÃ p<0.01. (D-D‴) The expression region of 10×STAT GFP was enlarged in the 3 rd instar eye discs upon wdp knockdown using mirrorGal4. CD8-mRFP was used to mark the dorsal compartment. Double headed arrows in D‴ show the expression region of 10×STAT GFP in the dorsal and ventral part. (E-F') The activity of unstable 10×STAT DGFP was obviously increased in wdp 1/1 intestines (F and F') compared with controls (E and E'). Moreover, 10×STAT DGFP was no longer restricted in small progenitor cells but also appeared in large ECs (arrows in F'). In eye discs bearing GFP positively marked clones overexpressing Dome-V5 (Act>y+>Gal4, UAS-GFP, UAS-dome-V5), Dome-V5 was mainly localized on the cell membrane (yellow arrows) despite some intracellular punctate structures (yellow arrowheads). A‴ is the enlarged image of the position labeled by square box in A". (B-B‴) In eye discs bearing GFP positively marked clones expressing Dome-V5 together with Wdp (Act>y+>Gal4, UAS-GFP, UASdome-V5, UAS-wdp), Dome-V5 was depleted from cell membrane but detected as cytoplasmic particles (B"), which were partially colocalized with early endosome marker Rab5 (B‴, white arrows). B‴ is the enlarged image of the position labeled by square box in B". All the eye discs shown here are oriented posterior right. Scale bars, 20μm. (TIF) S1 Table. Partial JAK/STAT targets identified from ChIP experiments with adult gut tissues. In this ChIP assay, we totally got 1487 peaks with p-value<0.01. The above table shows partial putative JAK/STAT downstream targets. Some of them have previously been reported as potential targets or components of the JAK/STAT pathway through microarray or RNAi screening methods. The binding sites of STAT92E around the ChIP peaks (±500bp) include TTCNNGAA, TTCNNNGAA and TTCNNNNGAA. The full ChIP-Seq data can be found in the GEO database with the accession number GSE67346. (DOC) S1 Movie. Wdp expression promotes Dome-GFP trafficking from the cell membrane into the cytoplasm in live S2 cells. (Part 1-3) The dynamics of Dome-GFP in live S2 cells co-transfected with dome-GFP and wdp. Dome-GFP was mainly detected as intracellular punctuates. In Part 1, the arrowheads marked by A, B, C, D and E show the newly formed endocytic vesicles trafficking Dome-GFP from the cell membrane. Part 2 is the enlarged movie of the endocytic vesicles marked by B and C. Part 3 shows the co-localization of Dome-GFP with endocytic dye FM4-64.(Part 4-5) The dynamics of Dome-GFP in live S2 cells transfected with dome-GFP alone. Dome-GFP was mainly present on the cell membrane and no newly formed vesicles trafficking Dome-GFP from the cell membrane were observed (Part 4). In addition, no obvious colocalization of Dome-GFP with FM4-64 was detected in Part 5. (AVI) S1 Text. Supplemental experimental procedures. Genotypes of flies used in Figs 1-8 and S1-S9 Figs are listed, followed by the information of primers used for RT-qPCR, constructing luciferase vectors and Rab5 dsRNA synthesis. In addition, the protocol of Rab5 dsRNA synthesis is also included. (DOC)