Remote Control of Intestinal Stem Cell Activity by Haemocytes in Drosophila

The JAK/STAT pathway is a key signaling pathway in the regulation of development and immunity in metazoans. In contrast to the multiple combinatorial JAK/STAT pathways in mammals, only one canonical JAK/STAT pathway exists in Drosophila. It is activated by three secreted proteins of the Unpaired family (Upd): Upd1, Upd2 and Upd3. Although many studies have established a link between JAK/STAT activation and tissue damage, the mode of activation and the precise function of this pathway in the Drosophila systemic immune response remain unclear. In this study, we used mutations in upd2 and upd3 to investigate the role of the JAK/STAT pathway in the systemic immune response. Our study shows that haemocytes express the three upd genes and that injury markedly induces the expression of upd3 by the JNK pathway in haemocytes, which in turn activates the JAK/STAT pathway in the fat body and the gut. Surprisingly, release of Upd3 from haemocytes upon injury can remotely stimulate stem cell proliferation and the expression of Drosomycin-like genes in the intestine. Our results also suggest that a certain level of intestinal epithelium renewal is required for optimal survival to septic injury. While haemocyte-derived Upd promotes intestinal stem cell activation and survival upon septic injury, haemocytes are dispensable for epithelium renewal upon oral bacterial infection. Our study also indicates that intestinal epithelium renewal is sensitive to insults from both the lumen and the haemocoel. It also reveals that release of Upds by haemocytes coordinates the wound-healing program in multiple tissues, including the gut, an organ whose integrity is critical to fly survival.


Author Summary
Innate immunity acts as the primary line of defense to overcome invading organisms. This response starts through sensing of microbe-specific molecules such as lipopolysaccharide and peptidoglycan by host receptors. Metazoans can also recognize signals that are associated with tissue damage and wounding, to then activate specific pathways involved in tissue repair or inflammation. Previous studies have suggested a link between JAK/STAT activation and tissue repair in Drosophila, but the mode of activation and the precise function of this pathway remain unclear. In this article, we analyze the role of the JAK/STAT Introduction Innate immunity provides the first line of defense against invading pathogens. This response is initiated by host pattern recognition receptors (PRRs), which sense specific and conserved motifs called microbe associated molecular patterns (MAMPs) found in microbes, but not in the host. MAMPs include lipopolysaccharide, peptidoglycan, lipoproteins, and CpG motifs of DNA [1]. Upon the recognition of MAMPs, PRRs activate signaling cascades that trigger the expression of immune effectors and regulators. Over the years, many PRRs, their ligands and the downstream signaling cascade components have been identified [2]. Metazoans have also acquired the capacity to recognize signals called damage-associated molecular patterns (DAMPs). DAMPs are associated with tissue damage and wounding, and activate specific pathways involved in tissue repair or inflammation [3]. As most infections are associated with tissue damage, signaling pathways induced by both DAMPs and MAMPs participate in a coordinated manner to mount an effective host defense response. Several DAMPs have been identified in mammals, including cellular components such as ATP, uric acid, nucleic acids and actin [4]. However, the pathways and effector mechanisms that are activated by DAMPs and how these pathways are integrated with other facets of the innate immune response are poorly understood.
The JAK/STAT signaling pathway plays a pivotal role during development and immunity in both mammals and insects [5,6]. It is activated upon binding of a secreted ligand to a receptor, leading to the recruitment of the Janus kinase (JAK), and the subsequent activation of the transcription factor STAT (Signal Transducer and Activator of Transcription). STAT, which is activated through phosphorylation, then translocates into the nucleus where it regulates the expression of target genes. While the genome of mammals encodes multiple combinatorial JAK/STAT pathways, only one canonical pathway is found in Drosophila. The Drosophila JAK/STAT pathway was originally identified through its role in embryonic segmentation [7]. It has three main cellular components: the receptor Domeless, the JAK Hopscotch, and the transcription factor STAT92E [8], and is activated by three secreted proteins of the Unpaired (Upd) family-Upd1, Upd2 and Upd3 [9]. Several studies have revealed a role of JAK/STAT in the insect systemic immune response. The first evidence for a role of the JAK/STAT pathway in insect immunity was the observation that bacterial infection triggers the phosphorylation and translocation of STAT92E in the fat body of the mosquito Anopheles gambiae [10]. Subsequent gene expression profiling studies identified a subset of Drosophila immune-responsive genes that are regulated by JAK/STAT upon injection of bacteria in the body cavity (referred to as a septic injury). This includes the genes encoding a complement-like protein Tep2, the cytokine Diedel, a JAK/STAT negative regulator Socs36E, and Turandots [11][12][13][14][15][16]. Turandots are Drosophila-specific genes of unknown function that are induced under various stress conditions, especially septic injury [17]. Studies done in larvae have shown that septic injury induces the release of the cytokine Upd3 from haemocytes, leading to the activation of the JAK/STAT pathway in the fat body [18].
In addition to its role in the humoral response, the JAK/STAT pathway also plays a pivotal role in Drosophila haematopoiesis. Constitutive activation of the pathway in haemocytes leads to a 5-20 fold increase in plasmatocyte numbers in larvae [19]. Many of the plasmatocytes differentiate into large flat cells called lamellocytes, which play a role in the encapsulation and destruction of aberrant self-tissue [20]. The JAK/STAT pathway also participates in the control of tumors or aberrant self in flies. The Upd3 ligand is expressed either at a wound site or by tissues with damaged basement membrane due to tumor outgrowth [20,21]. Circulating Upd3 then triggers JAK/STAT activation in both haemocytes and fat body, resulting, among others, in an increased number of haemocytes, which contribute to wound healing or tumor reduction [20]. Although many studies have suggested a link between JAK/STAT activation and tissue repair, the mode of activation and the precise function of this pathway in Drosophila systemic immunity remain unclear. In this article, we analyze the role of the JAK/STAT pathway in the resistance to wounding and septic injury in Drosophila. By using previously generated upd2 and upd3 mutants [22], we show that this pathway contributes to fly resistance to wounding and bacterial infection. Our study reveals that haemocytes express the three Upds and that injury induces the production of Upd3, which then activates the JAK/STAT pathway in the fat body and in the gut. One of the most surprising findings of the study is that Upds released from haemocytes can remotely stimulate intestinal stem cell proliferation. Thus, we uncover an unexpected interaction between haemocytes and intestinal stem cell proliferation that contributes to fly survival after injury.

Upd2 and Upd3 contribute to survival after septic injury
The Drosophila genome contains three unpaired genes that encode ligands capable of activating the JAK/STAT pathway receptor Domeless. Since upd1 null mutants are embryonic lethal [8], we focused our attention on the role of upd2 and upd3 in the systemic immune response [22]. To assess their role in the antimicrobial response, we monitored the survival rate of flies lacking upd2 and/or upd3 after systemic infection with either the Gram-negative bacterium Erwinia carotovora carotovora 15 (Ecc15) or the Gram-positive bacterium Enterococcus faecalis. Fig 1A shows that upd2 Δ , upd3 Δ and upd2,3 Δ flies were significantly more susceptible than wild-type flies to septic injury with Ecc15. Compared to Imd deficient mutants, which usually die within 48h from the same challenge, the susceptibility of the upd mutants was mild with about 50% mortality after 10 days. Fig 1B shows that only upd2,3 Δ flies displayed a higher susceptibility to septic injury with E. faecalis. The absence of a differentiated survival phenotype for the upd2 Δ or upd3 Δ single mutants could be a result of the faster killing rate induced by this bacterium (Fig 1B).
The Toll and Imd pathways are two NF-kB signaling pathways that contribute to resistance to septic injury by regulating many immune genes, notably those encoding antimicrobial peptides [23]. One possibility is that loss of upd2 and upd3 disrupts one or both of these NF-kB pathways, resulting in lower survival to septic injury. To test this hypothesis, we monitored the expression of the antimicrobial peptide genes Diptericin (Dpt) and Drosomycin (Drs) as readouts of the Imd and Toll pathways, respectively, in upd mutants. Both antimicrobial peptide genes remain fully inducible in upd2 Δ , upd3 Δ and upd2,3 Δ mutants (S1A and S1B Fig). This indicates that Upd ligands contribute to host survival to wounding and septic injury in a Toll and Imd independent manner. of upd2 Δ , upd3 Δ and upd2,3 Δ male flies as compared to wild-type (w 1118 ). In (A) upd2 Δ , upd3 Δ and upd2,3 Δ flies are more susceptible to infection with Ecc15. Flies were maintained at 29°C. Log-rank test comparing upd2 Δ , upd3 Δ and upd2,3 Δ infected to w 1118 infected: *** = p < 0.001. (B) upd2,3 Δ flies are more susceptible to infection with E. faecalis. Infections with E. faecalis were done at 25°C. Log-rank test comparing upd2,3 Δ infected to w 1118 infected: *** = p, 0.0001. (C) Mortality of male flies injured with a clean needle in the thorax indicates that upd2,3 Δ flies are sensitive to wounding. Results in (A-C) are pooled data from at least four independent experiments, where 20 flies were used per genotype. Log-rank test comparing upd2,3 Δ injured to w 1118 injured: *** = p < 0.001 (D) Bacterial persistence in wild-type and upd2,3 Δ male flies at 10 h and 50 h post septic injury At this stage, it was unclear whether the higher susceptibility of upd mutants to septic injury was caused by the presence of the bacterium itself or by the wounding associated with this mode of infection. To distinguish between the two, we monitored the survival rates of upd deficient flies to sterile wounding and indeed, upd2,3 Δ and to a lesser extent the single mutant flies showed an enhanced susceptibility to a clean wound (Fig 1C). We conclude that upd2,3 Δ flies are more susceptible to injury and that the presence of bacteria in the wound simply accelerates the mortality rate in both wild-type and upd2,3 Δ flies. Resistance and resilience mechanisms both contribute to the survival to bacterial infection, where resistance involves the activation of the immune responses to eliminate pathogens, whereas resilience encompasses the capacity to survive infection without eliminating the pathogen [24]. We observed that the number of E. faecalis in flies both 10 h and 50 h after infection (when flies start to die) was roughly similar in the upd2,3 Δ flies and in wild-type ( Fig 1D). Hence, resistance mechanisms were unaffected in the upd2,3 Δ flies in response to bacterial infection. In the following step, we investigated how Upd ligands contribute to survival by restricting our analysis to two wounding protocols, clean or septic injury with Ecc15 (referred to below as septic injury).

Survival to wounding requires the production of Upd2 and Upd3 by haemocytes
Previous studies have shown that upd1, upd2 and upd3 are induced in various intestinal cell types upon oral infection to stimulate stem cell proliferation and differentiation. Upd3 has also been shown to be induced in haemocytes of third instar larvae upon septic injury [15,22,25]. To investigate the spatial expression pattern of upds in adults, we monitored their expression in the three major immuno-responsive tissues, fat body, gut and haemocytes. In the haemocytes, upd1 and upd2 are expressed, but their expression is not consistently altered upon septic injury (Fig 2A and 2B). In contrast, the transcription of upd3 is significantly up-regulated in haemocytes 2h post-infection ( Fig 2C). In the gut, all three upds are expressed at basal levels. However, the expression of upd2 was up-regulated at 2h following septic injury ( Fig 2B). In contrast, the fat body did not significantly up-regulate any of these three genes (Fig 2A-2C). We next investigated whether Upds are required in adult haemocytes for optimal survival to wounding. For this, we knocked down individually each upd gene in haemocytes using an in vivo RNAi approach with the haemocyte driver hmlΔGAL4 [26,27] and monitored the survival of flies to wounding as well as to septic injury. We discovered that knockdown of either upd2 or upd3 in haemocytes increased the susceptibility of flies to wounding, while only the knockdown of upd3 resulted in higher susceptibility to septic injury with Ecc15 (Fig 2D and 2E). To reinforce our conclusion, we performed survival to clean and septic injury in flies that express both the upd2 and upd3 RNAi using another GAL4 driver, peroxidasin (pxn)GAL4 (S1C Fig). While hmlΔGAL4 is a haemocyte specific driver [26,27], pxnGAL4 is expressed in haemocytes and weakly expressed in the fat body [28]. Simultaneous silencing of upd2 and upd3 with pxnGAL4 markedly increased susceptibility to both clean and septic injury (S1C Fig). These experiments support the notion that the release of Upds by haemocytes promotes host survival to clean and septic injury. In agreement with this, flies lacking most of their haemocytes due to the expression of the pro-apoptotic factor Bax in the plasmatocyte lineage (genotype: hmlΔGAL4 > UAS-Bax) were susceptible to both wounding and septic injury showing similar with E. faecalis (OD 600 = 10), expressed as the number of CFUs per fly. No difference was observed between upd2,3 Δ and w 1118 . Each histogram corresponds to the average of three independent experiments. Student's t test shows P value = NS (0.93) for upd2,3 Δ vs wt flies at 10 h and P value = NS (0.31) for upd2,3 Δ vs wt flies at 50 h. NS: Non-Significant.
doi:10.1371/journal.pgen.1006089.g001 Upd genes are differentially expressed in the adult Drosophila haemocytes and midgut in response to septic injury. (A) RT-qPCR experiments show that upd1 is expressed in haemocytes but its expression remains unchanged upon septic injury (SI). NS: P value = 0.479 between UC and 2 h SI for haemocytes. (B) Upd2 is expressed in haemocytes in unchallenged flies (UC) but its expression remains unchanged upon septic injury (UC). NS: P value = 0.856 (C) Upd3 is also expressed in haemocytes and its expression is increased upon septic injury. Mean values of at least three experiments ± SE are shown and data was analyzed by Student's t test. (D-E) Survival analysis shows that flies with reduced upd3 expression in haemocytes display an increased susceptibility to clean injury (CI) and septic injury. RNAi flies with reduced expression of upd2 or upd3 show higher mortality to clean injury survival kinetics (S1D Fig). Collectively, these results provide strong evidence that the production of Upd2 and Upd3 ligands by haemocytes is crucial for fly survival to wounding and septic infection.

Septic injury activates JAK/STAT transcriptional activity in the gut
The precise function of the JAK/STAT pathway in the systemic immune response is still poorly defined. The pathway regulates a small subset of immune inducible genes, notably genes encoding stress peptides of the Turandot family, the opsonin Tep2 as well as Socs36E, a negative regulator of the JAK/STAT pathway [10][11][12]. Having shown that Upds are expressed in haemocytes and that some of them are required for resistance to infection, we next determined the impact of upd2,3 Δ deficiencies on JAK/STAT transcriptional activity. As expected, we observed a reduced expression of the JAK/STAT target genes Socs36E, TotA and TotM in upd2 Δ , upd3 Δ and upd2,3 Δ deficient flies upon septic injury ( Fig 3A, 3C and 3E). Next we wanted to explore where the JAK/STAT pathway was activated in wild-type flies upon septic injury, by monitoring the expression of TotM and Socs36E in three immune compartments: the gut, the carcass (comprising of the fat body) and haemocytes. Surprisingly, we observed that Socs36E was strongly induced in the gut and to a lower extent in the carcass upon septic injury ( Fig 3B). TotA and TotM were induced mostly in the fat body, consistent with a previous study on Drosophila larvae [15] (Fig 3D and 3F). To further characterize JAK/STAT pathway activity in vivo, we used flies carrying the reporter gene 10XSTAT92E-eGFP, which drives GFP expression under the control of ten STAT92E binding sites [29]. Since the same 10XSTAT92E-eGFP transgene is induced in larval fat body upon septic injury (S2A Fig) and since a previous study has already shown that the transcription factor STAT is activated in the fat body of adult flies after septic injury [15], we would have expected the same in adults. However, we cannot exclude that a specific feature of the adult fat body hinders an accurate observation of GFP activity in this tissue. Interestingly, both clean and septic injury induces a significant expression of 10XSTAT92E-eGFP in the gut ( Fig 4A).
Altogether, our data indicate that Upd2 and Upd3 contribute to the activation of the JAK/ STAT pathway following septic injury, and reveal that the gut is one of the main sites of JAK/ STAT activation upon septic injury in adults.

Body injury stimulates stem cell proliferation and immune response in the intestine
The JAK/STAT pathway has two main functions in the intestine. The first one is to promote epithelium renewal by stimulating the activity of intestinal stem cells and their differentiation into new enterocytes to rebuild the gut. The JAK/STAT pathway is activated through the release of Upd2 and Upd3 by stressed or damaged enterocytes, establishing a compensatory homeostatic loop [23,30,31]. Interestingly, the JAK/STAT pathway is also activated in the visceral muscles that surround the epithelium, regulating the production of the epidermal growth factor Vein, which stimulates the EGF-Receptor pathway in intestinal stem cells [30,32]. The second function of the JAK/STAT pathway is to regulate a subset of peptides in the anterior part of the gut, with homology to the antimicrobial peptide Drosomycin: Drsl2, Drsl3 and Drl4. The expression of these Drosomycin-like peptides also depends on the JAK/STAT ligands Upd2 and Upd3, which are released by the intestinal epithelium following damage (D) and only upd3 RNAi flies show a greater susceptibility to Ecc15 septic injury (E). Log-rank test was used to determine statistical significance. Male flies were used for experiments.
doi:10.1371/journal.pgen.1006089.g002  [22]. Our observation that clean and septic injury induces STAT activity in the gut suggests that a body wound can remotely stimulate stem cell proliferation and an anti-microbial response in the intestine.
To investigate this, we further analyzed the site of 10XSTAT92E-eGFP expression in the midgut of challenged adult flies. We noticed that the STAT reporter gene is induced both in the epithelium and in the visceral muscle that surrounds the epithelium (Figs 4A, 4B and S2B). In the epithelium, 10XSTAT92E-eGFP signals are restricted to progenitor cells (stem cells and enteroblasts) in unchallenged flies, but the signal expands to enteroblasts and enterocytes after wounding and septic injury ( Fig 4B). An expansion of 10XSTAT92E-eGFP is suggestive of an increased epithelium renewal rate and is known to occur when flies are fed with infectious bacteria or corrosive agents, two conditions that stimulate stem cell activity [33,34]. Supporting the hypothesis of increased epithelium turnover, the gene encoding the EGF ligand Vein is induced in the visceral muscles surrounding the gut following wounding and septic injury (Figs 4C and S4A). Intestinal epithelium renewal can easily be monitored in Drosophila by counting the number of dividing stem cells along the midgut using an anti-phosphohistone H3 (anti-PH3) antibody. Guts derived from flies collected after clean and septic injury exhibit an increased level of PH3 counts (Fig 4C-4E). Importantly, the midgut PH3 counts were strongly reduced, but not abolished, in upd2,3 Δ flies upon both clean and septic injury (Fig 4E for  females and S3A Fig for males). We next investigated whether body injury can trigger the induction of Drosomycin-like peptides in the gut. Drosomycin-like 3 (Drsl3) is indeed induced in the anterior part of the gut 6 h after septic injury (S4B Fig). In all these experiments we noticed that septic injury triggers a stronger response than a clean injury, indicating that the presence of bacteria increases the activation of JAK/STAT in the gut.
Taken together, our study uncovers a signaling pathway linking wound healing in the thorax to activation of stem cell activity and Drosomycin-like expression in the intestine. It also implies that Upds are required as a link between the wound and intestinal stem cell proliferation.

The JAK/STAT activity in the gut is induced by the release of Upds from haemocytes
The results above show that JAK/STAT activity can be induced in the gut upon wounding and that this activation requires Upd2 and Upd3. However, they do not inform us in which tissue Upds are expressed. The increased stem cell proliferation and Drs-like induction upon body injury could be a secondary consequence of damage to the gut, which is known to produce unpaired ligands. This hypothesis is supported by a recent paper by Takeishi and colleagues, which showed that wounding induces apoptosis in the gut, which in turn activates the stem cell division pathway [21]. The authors postulated the existence of a 'lethal' factor that is released in the haemolymph upon wounding, which is responsible for intestinal damage. This prompted us to investigate whether clean or septic injury induces cell death in the gut that could explain the increase of STAT activity and epithelium renewal we observed. For this, we monitored apoptosis in the intestine of flies 6 h septic injury and 16 h after oral infection using an anti-caspase 3 staining and inspect caspase activation by monitoring the cleavage of DCP-1. We did not observe any increase of caspase 3 signal or cleavage of DCP-1 in flies following septic injury (S5A and S5B Fig  two apoptotic assays [30]. In light of these findings, we hypothesized that it might be rather the release of Upd ligands from circulating haemocytes that contributes to intestinal regeneration. To test this notion, we monitored stem cell activity upon septic injury in flies lacking haemocytes due to the hmlΔGal4 driven expression of Bax in plasmatocytes. Flies lacking haemocytes have a significantly lower intestinal mitotic index after septic injury as compared to their wild-type counterparts ( Fig 5A). This strongly suggests that haemocytes do contribute to the signal that activates stem cell proliferation upon systemic wounding. We then tested whether over-expression of Upd ligands in haemocytes was sufficient to activate intestinal stem cell proliferation. Over-expression of upd2 or upd3 but not upd1 using the haemocyte specific driver hmlΔGal4, stimulates an increased stem cell activity in the gut in the absence of wounding ( Fig  5B). In contrast, we confirmed that over-expression of each of upd2, upd3 and upd1 (but a lower extent) in enterocytes (genotype: MyoTS-Gal4> UAS-upd) stimulates stem cell proliferation in agreement with previous studies [32]. The observation that the expression of upd2 and upd3 but not upd1 in haemocytes can remotely activate stem cells is consistent with in vivo and in vitro data, showing that Upds have distinct diffusing properties and biological activities as suggested by previous studies [22,35,36]. We next tested whether the expression of Upds in haemocytes is sufficient to stimulate intestinal stem cell proliferation in the absence of any intestinal source of Upd2 and Upd3. Consistent with our hypothesis that haemocytes remotely activate intestinal stem cells, we observed that the over-expression of either upd2 or upd3 by haemocytes can stimulate intestinal stem cell proliferation in upd2,3 Δ flies (Fig 5C). In this experiment, we noticed that over-expression of Upd2, but not of Upd3, led to the accumulation of small nucleate cells in the gut of upd2,3 Δ flies, suggesting a differentiation defect (S4C Fig). This indicates that while hemolymphatic production of Upd2 or Upd3 can trigger gut stem cell proliferation, only Upd3 ensured proper epithelium renewal with differentiation of the proliferating stem cells, pointing to a specific requirement of Upd2 to ensure proper enterocyte differentiation. Finally, to confirm that Upds derived from haemocytes play a role in septic injury induced intestinal stem cell proliferation, we used an in vivo RNAi approach to silence upd1, upd2 or upd3 in haemocytes using the hmlΔGal4 driver, and monitored stem cell activity after septic injury. Fig 5D shows that reduction of upd2 or upd3 in haemocytes (but not upd1) resulted in decreased PH3 counts in the gut 8 h after a septic injury, compared to wild-type. We conclude that production of Upd2 and Upd3 by haemocytes upon systemic wounding is necessary for JAK/STAT activation in the gut. This uncovers a new mechanism where haemocytes can remotely activate intestinal stem cell activity.

The JNK pathway regulates upd3 expression upon septic injury in haemocytes
Previous studies have established a link between JNK activity and the expression of Upds in the intestine [25,31,37]. We therefore investigated a possible role of the JNK pathway in the regulation of Upds in haemocytes upon septic injury. Interestingly, RT-qPCR analysis revealed that Immunostaining against LacZ and PH3 of midgut of vein-lacZ flies reveals that the EGF-R ligand encoding gene, vein (nuclear signal), is induced upon clean and septic injury with Ecc15 in the circular visceral muscles surrounding the intestine of female flies. Cuticle injury also increases the number of mitotic intestinal stem cells as revealed by an increased number of Phospho-Histone-3-positive cells (PH3+, white arrowheads) as seen clearer in the yellow boxed panels below that are zooms of the panels above.  the JNK pathway is activated in both haemocytes and gut upon septic injury, as illustrated by the induction of puckered, a reporter gene of JNK pathway activity [38] (Fig 6A). We then monitored upd2 and upd3 expression and intestinal mitotic activity in flies expressing a dominant-negative form of the Drosophila JNK Basket in haemocytes [39]. Inactivation of JNK signaling reduces both haemocyte expression of upd3 and intestinal PH3 counts upon septic injury, revealing that the stress-pathway JNK regulates upd gene expression in haemocytes ( Fig  6B and 6C).

Inhibition of intestinal epithelium renewal increases susceptibility to septic injury
Many studies have shown that dysfunction of the Drosophila intestine impacts both aging and resistance to stress in flies [40][41][42]. Along this line, Takeishi et al. have shown that a viable mutation in the caspase activator Apaf-1/dark, dpf-1 K1 , which leads to defects in gut epithelial renewal, results in lethality to wounding [21]. Based on these and our results, we hypothesized that the susceptibility of upd2 Δ , upd3 Δ and upd2,3 Δ mutants to systemic wounding could be linked to an impairment in gut epithelium renewal. Recently, a method named the 'Smurf assay' has been developed to probe gut integrity. In this assay, flies are fed on food colored with a non-toxic water-soluble dye, and gut integrity is estimated by measuring the diffusion of the blue dye from the gut lumen to the hemocoel [41,42]. 'Smurf' flies, which show systemic diffusion of blue dye due to intestinal barrier dysfunction, exhibit a lower life expectancy. We used this assay to assess gut barrier integrity of upd2,3 Δ flies. Upd2,3 Δ flies display a higher proportion of 'Smurf' flies over time upon wounding and septic injury (Fig 6D and 6E). The kinetics of 'Smurf' occurrence correlates with the survival to clean and septic injury. A higher proportion of 'Smurf' flies over time was also observed in hmlΔGal4 > UAS-Bax haemocyte-less flies compared to wild-type (S3B Fig). Hence, an accelerated loss of gut integrity in upd2,3 Δ flies is likely underlying their enhanced susceptibility to injury. To further test this hypothesis, we knocked down the JAK/STAT pathway specifically in the gut progenitor cells using esg ts GAL4, and monitored survival to septic injury. In these experiments, a dominant-negative form of the receptor Domeless (Domeless DN ) or a STAT RNAi construct (STAT-IR) were expressed in 3 day old adults using the thermo-sensitive driver esg ts GAL4. Flies with reduced JAK/STAT activity in progenitor cells failed to survive clean and septic injury (Figs 6F and S4D). Thus, our results identify intestinal epithelium renewal as a critical event for survival to septic injury.

Activation of intestinal stem cells upon oral infection does not require haemocytes
Our study is not the first to identify a role for haemocytes in intestinal epithelium renewal. Ayyaz et al. (2015) have recently shown that adult haemocytes are required for proper stem cell proliferation upon ingestion of paraquat and oral bacterial infection via the release of the Dpp ligand [43]. They also reported some haemocytes are recruited to the intestine upon stresses such as oral bacterial infection. This study does not preclude that haemocytes could also be a source of Upds in those conditions. To test this, we first investigated whether haemocytes respond to oral bacterial infection by expressing upds. Upd3 was slightly induced in haemocytes upon oral infection with Ecc15, but the level of increase did not reach statistical significance (Fig 7A-7C). The relative gene expression level of upd3 over RpL32 indicates that upd3 expression after oral infection in haemocytes was much weaker as compared to septic injury with Ecc15 (Fig 2C). This indicates that haemocytes are more reactive to systemic infection than oral infection. We next monitored intestinal stem cell proliferation in the midgut of flies expressing upd3 RNAi in haemocytes, or of flies lacking nearly all haemocytes due to the  [43]). Nevertheless, we did not observe any obvious change in haemocytes number dye. The right Y-axis curve represents the cumulative proportion of Smurfs in the population. (E) Proportions Smurfs in wild-type (w 1118 ) male flies and upd2,3 Δ male flies from day 10 and day 30 of adulthood. upd2,3 Δ flies display a higher proportion of Smurf phenotype after clean injury (CI) and septic injury (SI) as compared to their wild-type counterparts. n = 60 flies/condition. **: p< 0.01, *: p< 0.05 as determined by Student's t-test. (F) Knockdown of JAK/ STAT signaling specifically in intestinal progenitors using the escargot TS driver leads to a higher mortality rate in flies subjected to either clean injury or septic injury. n = 60 flies per genotype that is pooled from three independent experiments. Log-rank test comparing esg TS > UAS-STAT-IR Clean Injury and esg TS > UAS-STAT-IR Septic Injury to esg TS > Cs Clean Injury and esg TS > Cs Septic Injury respectively: *** = p, 0.001. RT-qPCR and survival analysis were done on adult male flies while PH3 counts were done with adult female flies.  around the gut upon septic injury or oral bacterial infection with Ecc15. We thus conclude that the intestinal repair response to oral bacterial infection can occur in absence of haemocytes.

Discussion
Although it is recognized that three signaling pathways in Drosophila, Toll, Imd, and JAK/ STAT, mediate the bulk of the transcriptional response to septic injury, functional studies on the JAK/STAT pathway have been hampered by the fact that loss-of-function mutations in genes encoding components of this pathway cause severe developmental defects. In this study, we have used viable mutations in upd2 and upd3 and tissue-specific inactivation of JAK/STAT components to assess their functions in systemic immunity. We demonstrate that Upd-JAK/ STAT signaling is required to resist wounding and septic injury in adults. The upd2,3 Δ deficient flies show a mild but clear susceptibility ten days after wounding. This is consistent with the common view that this pathway responds to damage, while Toll and Imd pathways are more responsive to microbial infection. Thus, the Upd-JAK/STAT axis should be seen as an integral part of the systemic wound response. Agaisse et al. 2003 initially reported that JAK/STAT activation in the larval fat body is mediated by a ligand, Upd3, which is secreted by haemocytes [15]. Our data confirm and extend this finding, showing that the production of Upd3 by haemocytes activates JAK/STAT signaling in the adult fat body (See model Fig 8). It also shows that haemocytes produce two other upds, Upd1 and Upd2, whose expression is less sensitive to septic injury. However, the observation that haemocytes contribute to intestinal epithelium renewal in response to septic injury was unexpected. Takeishi et al. (2013) were the first to establish a link between systemic wounding and intestinal homeostasis. They showed that body wounds remotely control caspase activity in enterocytes, which in turn activates the intestinal regeneration pathway in the gut [21]. However, the authors of this paper did not identify the link between body injury and the induction of apoptosis in the gut, they postulated the existence of a haemolymph 'lethal' factor that contributes to intestinal damage. Our study demonstrates that the wounding of the cuticle can activate the production of Upds by haemocytes, and that these secreted ligands are sufficient to induce stem cell proliferation in the intestine. Our observation that Upds released by haemocytes can remotely stimulate stem cell activity indicates that epithelium renewal is influenced by insults from both the luminal (microbial infection, ingested toxins) and the hemolymphatic compartments (septic injury). Along the same line, it is well known that gut dysfunction and subsequent loss of appetite in humans commonly occurs following surgery or major injury [44], although the molecular mechanisms of this process are poorly understood.
The reason why septic injury stimulates intestinal stem cells remains enigmatic. Takeishi et al. (2013) and our study suggest that the intestine is an organ sensitive to systemic wounding. In agreement with this notion, the 'Smurf' assay reveals an increased intestinal permeability in upd2,3 Δ flies upon septic injury. Stem cell proliferation is still activated, albeit at a much lower rate, in upd2,3 Δ and in haemocyte-depleted flies. This suggests that additional factors beyond Upds, can stimulate intestinal stem cell proliferation. A first hypothesis would be that septic injury with a needle directly damages the intestine [21]. We do not favor this notion because the intestine did not appear damaged upon septic injury as assessed by phalloidin staining of the entire gut (S2B1 and S2B4 Fig) and did not show an increase of apoptosis at early time points (S5A and S5B Fig). A second hypothesis is that wounding of the cuticle indirectly alters gut integrity. For instance, reactive oxygen species (ROS) induced during the systemic wounding reaction could damage the intestinal basement membrane. Thus, the production of Upds by haemocytes could be a way to stimulate intestinal repair as an anticipatory tissue repair mechanism.
The role of the JAK-STAT pathway in the systemic wound response is not limited to the fat body and the gut. The JAK/STAT also plays a major role in larval haemocytes where it contributes to haemocyte recruitment or the differentiation into lamellocytes [19,20]. A recent study has shown that infestation by the parasitoid wasp Leptopilina boulardi triggers the release of Upd2 and Upd3 from haemocytes, which in turn activate the JAK/STAT pathway in the larval somatic muscles [45]. Collectively, haemocyte-derived Upds stimulate a broad array of responses in multiple tissues, and future studies should decipher how each organ contributes to the systemic wound response. Blocking the JAK/STAT pathway in the gut makes flies susceptible to a simple injury, suggesting that proper gut function is essential for recovery from systemic injury. Therefore, it cannot be excluded that several roles initially attributed to the JAK/STAT pathway are linked to its ability to turn on epithelium renewal in the intestine. For instance, the involvement of the JAK/STAT pathway in the resistance to Drosophila C Virus and cricket paralysis viruses in Drosophila or its implication in the defense against Plasmodium in Anopheles [46][47][48][49] could be linked to a defect in epithelium renewal.
Our study reveals that haemocytes are dispensable to epithelium renewal upon oral infection with Ecc15. Thus, haemocytes promote epithelium repair in response to a systemic threat but are less important in the defense against oral bacterial infection. Our results diverge from those of a recent study, which pointed to a contribution of adult haemocytes to stem cell proliferation, via their release of the Dpp ligand upon ingestion of paraquat or oral bacterial infection [43]. Consistent with the work of Ayyaz et al., (2015), we did observe aggregation of haemocytes close to the midgut at the level of the loop. Nevertheless, we did not observe any obvious recruitment of haemocytes to the gut upon septic injury or oral bacterial infection with Ecc15. The continuous presence of haemocytes in close proximity to the midgut could allow a local enrichment of the diffusible JAK/STAT ligands upon septic injury [43]. The presence of haemocytes in close proximity to many tissues including gut and imaginal discs [20,50] underlines their function in tissue homeostasis as local tissue repairers.
Our study delineates a pathway in which damages associated with injury induce the release of Upd3 by haemocytes to activate JAK/STAT in the intestine both in the visceral muscles and stem cells (Fig 8). We further show that the JNK pathway regulates the expression of upd3 as previously described for the gut [25,31,41]. This is in line with a former study describing a role for the JNK pathway component Mekk1 in the expression of the JAK/STAT target genes of the Turandot family, although the tissue where Mekk1 is required had not been identified [18]. Another recent study already implicated the JNK pathway for the regulation of Upd3 production in haemocytes in response to the ingestion of a high-lipid diet [51]. The observation that septic injury triggers upd3 expression in haemocytes more effectively than clean injury suggests that the Imd and Toll pathways, which sense microbial determinants, could promote the induction of upds in haemocytes possibly through a convergence on the JNK pathway.
Systemic wounding induces a complex series of events that includes the entry of oxygen, rupture of basement membranes, ROS burst, production of melanin, as well as aggregation of clotting factors and haemocytes around the wound [52,53]. Wounding also leads to the release of cytoplasmic material from dead cells such as cytoplasmic actin, which functions as a DAMP in mammals [54]. A major challenge is now to identify the underlying mechanism that senses damage and activates the pathway. Previous studies have shown that H 2 O 2 is a primary local signal responsible for the homing of haemocytes to the wound site in embryos [28,55]. An attractive hypothesis is that ROS may be the signal activating the JNK pathway in the haemocytes close to the wound site. Alternatively, it cannot be excluded that haemocytes harbor a receptor capable of recognizing a component normally restricted to the cytoplasm of living cells and unable to cross the basement membrane. This would explain why they are activated only by the concomitant destruction of both the basement membrane and the underlying epithelium. Our study emphasizes the central role of haemocytes via the production of Upds in an integrated systemic wound response. Future studies should now address how damage is sensed by haemocytes and how the JAK/STAT pathway targets genes contributes to repair programs and host survival.

Bacterial strains and analysis of bacterial counts
Erwinia carotovora carotovora 15 (Ecc15) is a Gram-negative bacterium described in [59], while Enterococcus faecalis (E. faecalis) and Microccus luteus (M. luteus) are both Gram-positive bacteria [14]. Ecc15 and M. luteus were cultured overnight in LB medium at 29°C, and E. faecalis was cultured at 25°C.
For bacterial counts, w 1118 and upd2,3 Δ male flies were infected with E. faecalis, and the number of bacteria was determined as follows at 10 h and 50 h post-infection [60]. Flies were surface sterilized in 95% ethanol for 1 min, and then 5 flies were homogenized using a Precellys 24 instrument (Bertin Technologies, France), and then five 1/10 serial dilutions were made and plated on LB culture medium by spotting the serial in duplicates and incubated overnight at 25°C. Colonies were counted from spots containing more than 10 colonies. The experiment was repeated three times.  1) which was deposited on a filter disk that completely covered the surface of standard fly medium [59]. Flies were incubated for one day at 29°C on the contaminated filter, after which they were transferred to fresh vials containing standard medium without living yeast. In survival experiments, flies were maintained on medium without fresh yeast following infection and survivors were counted daily.

RT-qPCR
10-15 adult male flies or 20 dissected guts (including crop, midgut and hindgut but without Malpighian tubules), or 15 carcasses (that comprise mostly fat body) were collected in 500 μl of Trizol (Invitrogen). For collecting haemocytes, 20 individuals were placed on a 30 μM filter of an empty Mobicol spin column (MOBITEC), then covered with glass beads and centrifuged for 20 minutes at 4°C, 10'000 r.p.m. This was done twice for a total of 40 adult male flies, and then the haemolymph recovered was pooled and collected in 300 μl of Trizol. Total RNA was extracted according to the manufacturer's instructions. Quality of the RNA was determined using a NanoDrop ND-1000 spectrophotometer. The purified RNA was then treated with DNAse (Ambion) according to the manufacturer's instructions. The quantity of RNA was determined using the NanoDrop and then 1 μg of RNA was used to generate cDNA using SuperScript II (Invitrogen, Carlsbad, California, United States). RT-qPCR was performed using dsDNA dye SYBR Green I (Roche Diagnostics, Basel, Switzerland). Expression values were normalized to RpL32. Primer sequences used are provided in S1 Table. Imaging and immunohistochemistry For immunofluorescence, guts from 3 to 5 day old females were dissected in 1X PBS, fixed for 20 minutes in PBS, 0.1% Tween 20 (PBT), and 4% paraformaldehyde; then stained with primary antibody [1/1000 mouse anti-GFP (Roche); 1/500 rabbit anti-PH3 (Upstate/Millipore)]; 1/200 rabbit anti-Caspase 3 activity (ThermoFischer); 1/200 rabbit anti-cleaved DCP1 (Cell Signaling Technology) in PBT + 2% BSA]. Secondary staining was performed with Alexa594 anti-rabbit and Alexa488 anti-mouse antibodies (Invitrogen). Visceral muscles were stained using 1/500 Phalloidin-Rhodamine or 1/500 Phalloidin-FITC (Life Technologies). DNA was stained with 1/15000 dilution of 4',6-diamidino-2-phenylindole DAPI (Sigma). The stained gut tissue was mounted in the antifading agent Citifluor AF1 (Citifluor Ltd.). The mitotic index was determined by counting the number of PH3 positive cells along the midgut with Axioplot imager (Zeiss). For determination of PH3 counts, at least 10 guts were counted per condition in each experiment and data was pooled from 3 independent experiments. The mean number of mitoses per midgut and S.E.M. are shown for each genotype or treatment in the graphs.

Statistics
Each experiment was repeated independently a minimum of three times (unless otherwise indicated), error bars represent the standard error of the mean of replicate experiments (unless otherwise indicated). Statistical significance was determined using Student's t test, Mann-Whitney U test or log-rank test on GraphPad Prism, and P values of <0.05 = Ã , <0.01 = ÃÃ and <0.001 = ÃÃÃ for the t test and Mann-Whitney test while P <0.0001 = ÃÃÃ for log-rank test were considered significant.
Supporting Information S1 Fig. Upd2 Δ , upd3 Δ and upd2,3 Δ flies have wild-type Toll and IMD pathway activity. (A and B) RT-qPCR analysis of Diptericin (Dpt) and Drosomycin (Drs) expression in adults either unchallenged (UC) or collected at 6 h and 24 h after septic injury with Ecc15 (A) or M. luteus (B). Data are the mean of three repeats ± SE. The P value was determined NS (non-significant) using pair-wise comparison between each time-point of upd2 Δ , upd3 Δ and upd2,3 Δ infected to w 1118 infected using Student's t-test. (C) Knockdown of both upd2 and upd3 in haemocytes using the PxnGAL4 driver leads to a higher mortality rate in flies subjected to either clean injury or septic injury with Ecc15. Data is pooled from three independent experiments, n = 60. (D) 'Hemoless' flies (hmlΔGAL4 > UAS-BAX) display an increased susceptibility to wounding as well as septic injury with Ecc15. Data is pooled from three independent experiments, n = 60. The log-rank test was used to determine statistical significance. P value < 0.001 = ÃÃÃ as determined by log-rank test for hmlΔGAL4 > UAS-BAX CI and Ecc15 SI as compared to hmlΔGAL4 > Cs CI and Ecc15 SI; P value < 0.001 = ÃÃÃ for pxnGAL4 > UAS-upd2,3IR CI and Ecc15 SI compared to pxnGAL4> Cs CI and Ecc15 SI. Male flies were used for experiments done in panels A-D. (TIF) S2 Fig. Upd2 and Upd3 ligands contribute to JAK/STAT pathway in the fat body and the intestines upon injury. (A) A 10XSTAT-GFP (green) reporter allele was used to monitor JAK/ STAT activity in a living third-instar larva. Live-imaging of larvae at 6 h post-injury revealed an increase of GFP signal in the fat body and the gut. No increase in GFP signal was observed in the upd2,3 Δ larvae upon injury. Representative images from unchallenged (UC) flies or flies collected 6 h after clean injury or septic injury. The white arrowheads in wild-type clean injury and septic injury indicate GFP signal in the gut of larvae while white arrows indicate GFP in the fat body of the head. (B1-B4) Immunostaining on 10XSTAT-GFP (green) reporter gene and visceral muscle using Rhodamine-phalloidin (red) in the intestine region R2 of adult female flies either unchallenged (UC) or collected 6 h after septic injury with Ecc15. Nuclei are stained using DAPI (blue) and co-staining of 10XSTAT-GFP expression with the visceral muscle is seen in yellow. (TIF) Visualization of gut-associated haemocytes using hmlΔGAL4>UAS-RFP flies. Intestines were fixed with 4% paraformaldehyde and stained with FITC-phalloidin to label visceral muscle and DAPI to label nuclei. While the number of haemocytes attached to the loop of the midgut was variable, we did not observe any marked variation in haemocytes number upon septic injury. (B) 'Hemoless' female flies due to the overexpression of the pro-aptototic gene hid in hemocytes have wild-type level of intestinal stem cell proliferation 8 h after oral infection with Ecc15 as determined by counting PH3+ cells. Genotype: hmlΔGAL4 > UAS-hid. (TIF) S1 Table. List of qPCR primer sequences used in this study. (DOCX) Alfred Chng and Claudine Neyen for critical reading of the manuscript. We thank Bianca Petrignani for useful discussions.