Loss of the mucosal barrier alters the progenitor cell niche via Janus kinase/signal transducer and activator of transcription (JAK/STAT) signaling

The mucous barrier of our digestive tract is the first line of defense against pathogens and damage. Disruptions in this barrier are associated with diseases such as Crohn's disease, colitis, and colon cancer, but mechanistic insights into these processes and diseases are limited. We have previously shown that loss of a conserved O-glycosyltransferase (PGANT4) in Drosophila results in aberrant secretion of components of the peritrophic/mucous membrane in the larval digestive tract. Here, we show that loss of PGANT4 disrupts the mucosal barrier, resulting in epithelial expression of the IL-6–like cytokine Upd3, leading to activation of JAK/STAT signaling, differentiation of cells that form the progenitor cell niche, and abnormal proliferation of progenitor cells. This niche disruption could be recapitulated by overexpressing upd3 and rescued by deleting upd3, highlighting a crucial role for this cytokine. Moreover, niche integrity and cell proliferation in pgant4-deficient animals could be rescued by overexpression of the conserved cargo receptor Tango1 and partially rescued by supplementation with exogenous mucins or treatment with antibiotics. Our findings help elucidate the paracrine signaling events activated by a compromised mucosal barrier and provide a novel in vivo screening platform for mucin mimetics and other strategies to treat diseases of the oral mucosa and digestive tract.

The mucous barrier that lines our respiratory and digestive tracts is the first line of defense against pathogens and provides hydration and lubrication (1)(2)(3)(4). This unique membrane separates the delicate epithelia from factors present in the external environment. In mammals, the mucous lining of the intestine allows nutrient penetration while conferring protection from both bacteria and the mechanical damage associated with digestion of solid food (reviewed in Refs. 2 and 5). The principal components of mucous membranes are mucins, a diverse family of proteins that are expressed in tissue-specific fashions (2,(5)(6)(7)(8). Whereas secreted mucins vary greatly in sequence and size, they all share highly O-glycosylated serine-and threonine-rich regions that confer unique structural and rheological properties, allowing the formation of hydrated gels. Deletion of specific components of the mucous membranes present in the lung (Muc5b (9)) or the intestinal tract (Muc2 (10)) resulted in defects in mucociliary clearance or an increased incidence colorectal cancer, respectively. Changes in mucus production or the glycosylation status of mucins have also been associated with oral pathology in Sjogren's syndrome (11,12), the development of colitis and intestinal tumors in mice (10,(13)(14)(15), and the progression of ulcerative colitis and colon cancer in humans (16,17). However, the detailed mechanisms by which loss/alteration of this protective layer results in epithelial pathology remain unknown.
Many components and factors that confer the unique properties of this protective lining, including the mucins and the enzymes that mediate their dense glycosylation, are conserved across species (18 -21). Indeed, Drosophila melanogaster contains a similar protective lining known as the peritrophic membrane, consisting of a chitin scaffold that is bound by highly O-glycosylated mucins (8,22). During larval development, where animals ingest solid food and undergo massive growth in preparation for metamorphosis, specialized secretory cells (PR cells) of the anterior midgut produce this membrane, which is thought to protect epithelial cells of the digestive tract from mechanical and microbial damage (22,23). Previous work in Drosophila has shown that loss of one component of the peritrophic membrane resulted in a thinner and more permeable membrane, where adult flies were viable yet more susceptible to oral infections (24). However, the exact role of the entirety of this lining in the integrity and protection of cells of the digestive tract remains unknown.
Extensive research elucidating the development and function of the many cell types that comprise the Drosophila digestive tract has provided insights into mammalian digestive system formation and function (23,(25)(26)(27). The Drosophila larval midgut is composed of specialized epithelial cells (enterocytes (ECs) 3 and enteroendocrine cells) for digestion and nutrient absorption, as well as progenitor cells that will eventually form the adult midgut epithelium (26,28,29). These adult midgut progenitor cells (AMPs) reside in a protected niche, formed by peripheral cells (PCs) that wrap and shield them from external signaling (28). PCs are characterized by a unique crescent shape, with long processes that surround AMPs, restricting proliferation and differentiation until metamorphosis (25). The larval digestive tract therefore represents an ideal system to interrogate the role of the mucous layer in protection of both the epithelium and the progenitor cell niche at a stage when the mechanical and microbial stresses associated with the ingestion of solid food are abundant.
Previous work from our laboratory has shown that loss of a conserved UDP-GalNAc:polypeptide N-acetylgalactosaminyltransferase responsible for initiating O-linked glycosylation (PGANT4) resulted in aberrant secretion of components of the peritrophic membrane in the larval digestive tract (30). Here, we show that pgant4 mutants are devoid of a peritrophic membrane, resulting in epithelial cell damage and expression of the IL-6 -like inflammatory cytokine, unpaired 3 (Upd3). Upd3 expression resulted in increased JAK/STAT signaling in the progenitor cell niche, causing niche cell differentiation and aberrant progenitor cell proliferation. These effects were dependent on Upd3 and could be rescued by deleting upd3 or partially rescued by feeding animals antibiotics or exogenous mammalian intestinal mucins. Moreover, overexpression of the conserved extracellular matrix cargo receptor, Tango1 (transport and Golgi organization 1), in secretory cells of the digestive tract resulted in restoration of the peritrophic membrane and rescue of niche integrity. Our results elucidate new mechanistic details regarding how a compromised mucous lining can influence epithelial integrity and the progenitor cell niche and provide an in vivo screening platform for compounds and strategies that could restore mucosal barrier function.

Results
To address the role of the mucosal barrier in normal digestive system health, we examined third instar Drosophila larvae deficient for pgant4, which were previously shown to have defects in the formation of secretory vesicles in the PR cells that are responsible for the synthesis and secretion of the peritrophic membrane (30). Loss of pgant4, via conventional mutations (pgant4 f02186 /Df2L) or in vivo RNAi specifically in PR cells (c135Ͼpgant4 RNAi ) resulted in complete loss of the peritrophic membrane throughout the larval digestive tract ( Fig. 1A and Fig. S1). The lectin Helix pomatia, which detects GalNAc linked to serine or threonine, revealed specific loss of O-glycosylated proteins/mucins along the surface of the epithelium (Fig. 1A). Staining with the chitin-binding protein (Chitin) revealed loss of all chitin normally present as part of the peritrophic membrane along the digestive tract (Fig. S1A). Additionally, larvae fed fluorescently labeled dextran showed that in WT, food was encased in the protective peritrophic membrane, but no food encasement was seen in pgant4 mutants (Fig. S1B). Loss of the peritrophic membrane resulted in epithelial irregularities within the midgut (as detected by actin, which outlines cells), where cells no longer form a single cell layer but rather a multicell layer in places, with some cells detaching (Fig. 1, A and B). Additionally, loss of the peritrophic membrane resulted in the induction of epithelial apoptosis (as detected by the caspase marker, Dcp-1) (Fig. 1B) and the upregulation of certain genes encoding antimicrobial proteins (Fig. 1E).
In addition to epithelial apoptosis, increased cell proliferation within the midgut was detected by EdU and phosphohistone H3 (PH3) staining upon loss of the peritrophic membrane ( Fig. 1C and Fig. S2). Co-staining with PH3 and markers for the various cell types of the digestive tract, including PCs (detected by the marker Su(H)-lacZ) and AMPs (detected by the marker Delta), revealed that the proliferating cells were the AMPs ( Fig.  1D and Fig. S3). During normal larval development, AMPs undergo limited proliferation and remain in an undifferentiated state within the unique progenitor cell niche formed by the PCs (Fig. 2A). Interestingly, niche morphology was altered in pgant4 mutants (Fig. 2B). The normally crescent-shaped PCs were found to be in various stages of flattening/rounding and unwrapping AMPs. Once unwrapped, AMPs, which normally exist as tight groups of cells that stain brightly for the marker Armadillo (Arm) around their periphery in WT, were no longer present in tight groupings in pgant4 mutants (Fig. 2C). In addition to changes in PC morphology, we also observed changes in PC fate in pgant4 mutants. PCs are normally positive for the marker Su(H), whereas differentiated ECs are positive for the marker Pdm-1 (Fig. 2D). However, upon loss of the peritrophic membrane, many Su(H)-positive PCs were also positive for Pdm-1 (Fig. 2, D and E), suggesting that PCs are undergoing differentiation to EC-like cells. Taken together, these data indicate that loss of the peritrophic membrane affects PC differentiation and morphology, suggesting important roles for this protective membrane in maintenance of the progenitor cell niche.
We next performed qPCR analysis to determine which signaling pathways might be responsible for the changes in PC fate. Interestingly, a specific and dramatic increase in the expression of the gene encoding the IL-6 -like inflammatory cytokine (and ligand for JAK/STAT signaling) unpaired 3 (upd3) and its downstream target socs36E were observed (Fig.  3A). Other signaling pathways known to be involved in response to damage within the digestive system of Drosophila did not show consistent and significant changes (Fig. S4). These results suggest that loss of the protective lining is activating JAK/STAT signaling through the production of Upd3 (but not Upd or Upd2). To determine the source of Upd3, we performed RNA in situ hybridizations to upd3 in WT and pgant4 mutant midguts. As shown in Fig. 3B, upd3 expression is not normally detected in WT but is up-regulated in the large ECs (characterized by large nuclei) of the midgut upon loss of pgant4 and the peritrophic membrane.
Upd3 production from ECs upon loss of the peritrophic membrane could activate JAK/STAT signaling in an autocrine or paracrine manner. To determine in which cells Upd3 is activating JAK/STAT signaling, we next performed RNAi to pgant4 in a reporter line that contains 10 STAT-binding sites driving GFP expression (Stat92E-GFP). Using this line, cells in which JAK/STAT signaling is activated will fluoresce green. Interestingly, PCs of the progenitor cell niche showed a dra-Mucosal barrier loss disrupts the progenitor cell niche matic increase in JAK/STAT signaling in the absence of the peritrophic membrane (c135Ͼpgant4 RNAi ) relative to WT (Fig.  3C). In WT midguts, islands of cells consist of JAK/STAT-negative AMP clusters (small nuclei) surrounded by JAK/STATpositive PCs (medium-sized, green nuclei) (Fig. 3C). Upon loss of the peritrophic membrane, large clusters of cells consisting of both medium-sized nuclei (PCs) and small nuclei (AMPs) fluoresced brightly for JAK/STAT signaling (Fig. 3C). ECs with large nuclei remained negative for JAK/STAT signaling in both WT and c135Ͼpgant4 RNAi midguts (Fig. 3C), indicating that Upd3 is not acting in a cell-autonomous manner. Notably, some cells with small nuclei that appeared to remain surrounded by PC processes remained JAK/STAT negative, suggesting that residual PC wrapping might protect some AMPs from external Upd3 signals (Fig. 3C, red arrow). These data suggest that a dramatic up-regulation in JAK/STAT signaling

Mucosal barrier loss disrupts the progenitor cell niche
takes place within PCs and in some groups of AMPs in the absence of the peritrophic membrane. Taken together, our results suggest a model where loss of the protective mucinous peritrophic membrane results in epithelial cell damage and upregulation of upd3 expression specifically in EC cells, which in turn up-regulates JAK/STAT signaling in PCs, thereby altering the progenitor cell niche cell and causing an increase in progenitor cell proliferation.
To test this model, we first overexpressed upd3 in ECs of WT larvae to determine if we could recapitulate the alteration of the progenitor cell niche and proliferation of the AMPs. As shown in Fig. 4, overexpression of upd3 specifically in ECs (Myo1AϾUAS-upd3) resulted in increased JAK/STAT signaling, aberrant PC morphology, and increased proliferation of AMPs relative to control (Myo1A) (Fig. 4, A, B, and D). By also using a UAS-GFP reporter (Myo1AϾUAS-upd3, UAS-GFP), we found evidence for PC differentiation (as detected by the induction of expression of GFP in PCs that are still partially wrapping AMPs) when upd3 was overexpressed (Fig. 4C). Expression of GFP under the control of the EC-specific promoter Myo1A is seen in PCs (white arrows) in Myo1AϾUAS-upd3, UAS-GFP larvae but not in the absence of upd3 overexpression (Myo1AϾUAS-GFP). We next tested whether we could rescue the niche morphology and AMP proliferation seen upon loss of the peritrophic membrane by deleting upd3. In pgant4 RNAi flies that carry a deletion for upd3 (⌬upd3; c135Ͼpgant4 RNAi ), JAK/STAT signaling was reduced (Fig. 5, A  and C). Additionally, niche morphology was rescued (Fig. 5, B

Mucosal barrier loss disrupts the progenitor cell niche
and C), and AMP proliferation was significantly decreased (Fig.  5D).
To test whether the effects on JAK/STAT signaling and niche integrity were dependent upon the secreted peritrophic membrane, we next attempted to genetically restore the peritrophic membrane. Previous work from our group demonstrated that the defects in secretory vesicle formation and secretion of peritrophic membrane components in pgant4 mutants were due to Dfur2-mediated proteolysis of the cargo receptor Tango1 and that secretory vesicle formation could be rescued by overexpression of Tango1 in PR cells of pgant4 mutants (30). As shown in Fig. 6 (A-C), overexpression of tango1 in PR cells in the pgant4 RNAi background (c135Ͼpgant4 RNAi , tango1 OE ), restored the protective mucinous lining and rescued epithelial irregularities and apoptosis. Moreover, genetic restoration of the peritrophic membrane restored JAK/STAT signaling to wild-type levels and rescued aberrant proliferation (Fig. 6, B and C).
To determine whether this in vivo system could be used to test exogenous compounds that may function as mucin mimetics and restore epithelial cell integrity, we next fed purified mammalian mucins to pgant4 mutant larvae. Larvae were fed either water (H 2 O) or carboxymethylcellulose (CMC; a soluble derivative of cellulose that has viscous properties but lacks mucin-type glycans) as controls or mucins purified from pig intestine (Muc2) or stomach (Muc5AC). Muc2 is known to be crucial for digestive system function in mice (10). Interestingly, only Muc2 was able to reduce upd3 expression and expression of its downstream target, socs36E (Fig. 6D and Fig. S5), suggesting that this particular mucin conferred protective properties to the digestive system. No reduction in upd3 or JAK/STAT signaling was seen when using water or carboxymethylcellulose alone. Unexpectedly, Muc5AC actually increased upd3 expression and JAK/STAT signaling, indicating that different mucins have distinct properties in terms of interaction with and/or protection of epithelial surfaces of the digestive tract. A, qPCR analysis shows a dramatic increase in expression of upd3 and its downstream target socs36E in flies without peritrophic membrane (c135Ͼpgant4 RNAi ) relative to WT (c135ϾVDRC60000). Error bars, S.D. ***, p Ͻ 0.001. B, RNA in situ hybridization to upd3 (red) reveals that it is expressed in ECs (large nuclei) in animals without peritrophic membrane (c135Ͼpgant4 RNAi ). No detectable expression of upd3 was seen in WT. Nuclear staining is shown in blue. Scale bars, 20 m. C, JAK/STAT signaling (as detected by the reporter Stat92E-GFP; green) is seen in the PCs (but not the AMPs) of the progenitor cell niche in the WT (Stat92E-GFP, c135ϾVDRC60000) third instar midgut. Upon loss of the peritrophic membrane (Stat92E-GFP, c135Ͼpgant4 RNAi ), JAK/STAT signaling is greatly increased in PCs, which now have a rounded morphology. Additionally, JAK/STAT signaling is now seen in some clusters of AMPs (white circles and arrows). The red arrow shows an AMP cluster that is still wrapped by a PC and is not JAK/STAT-positive. Cells with large nuclei (DNA, blue) are ECs. Scale bars, 20 m.

Mucosal barrier loss disrupts the progenitor cell niche
Finally, we tested the role of microbes in the increased JAK/ STAT signaling and cell proliferation seen upon loss of the peritrophic membrane by treating pgant4 RNAi larvae with antibiotics. Newly hatched WT or c135Ͼpgant4 RNAi first instar larvae were cultivated on food containing antibiotics and grown to third instar. As shown in Fig. 6E, treatment of pgant4 RNAi larvae with antibiotics reduced the expression of antimicrobial genes to that of WT. Additionally, antibiotic treatment partially reduced upd3 and socs36E expression as well as cell proliferation, suggesting that there is a microbial contribution to the cellular damage/changes seen upon loss of the peritrophic membrane (Fig. 6, F and G).
Taken together, our results demonstrate a crucial protective role for the mucinous peritrophic membrane of the Drosophila digestive tract. In the absence of this protective barrier, induction of upd3 expression from epithelial cells increases JAK/ STAT signaling within the progenitor cell niche that causes differentiation of PCs to EC-like cells, which alters the niche morphology and releases AMPs from their protective environment. AMPs are subsequently exposed to JAK/STAT signaling, which causes them to undergo aberrant cell proliferation (Fig.  7). Genetic restoration of this membrane rescues epithelial damage, JAK/STAT signaling, and niche integrity. Moreover, our studies suggest that this in vivo system could be used to screen for compounds and strategies that restore the functions of this protective barrier and/or modulate conserved signaling pathways.

Discussion
Here, we demonstrate the crucial importance of the protective mucinous membrane in the integrity of the digestive epithelium and the maintenance of the progenitor cell niche. Whereas previous studies have investigated roles for components of this membrane in Drosophila (24), here we are able to completely eliminate it and monitor the ensuing cellular changes and signaling responses. We show the peritrophic membrane is essential to protect the integrity of the epithelial cell layer and maintain an appropriate environment for the progenitor cell niche. Moreover, we show a dynamic and specific response to the loss of this membrane via the production of the IL-6 -like cytokine Upd3 from epithelial cells, which in turn signals to niche cells in a paracrine fashion, causing differentiation and morphological changes. This demonstrates the multipotent nature of PCs, which can respond to specific cytokines to alter their fate. Once the PC morphology and fate were altered, JAK/STAT signaling was activated in AMPs exposed to Upd3, causing aberrant cell proliferation/DNA replication (Fig.  7). This represents the first example where loss of the protective

Mucosal barrier loss disrupts the progenitor cell niche
mucous lining activates signaling from epithelial cells to alter the fate of niche cells and change the behavior of progenitor cells. These studies highlight the importance of this membrane in both epithelial and progenitor cell biology and elucidate the paracrine signaling cascade that is specifically activated when this barrier is compromised.
Interestingly, the mucinous peritrophic membrane could be restored by overexpression of the conserved cargo receptor Tango1. Tango1 is an essential protein that functions to package large extracellular matrix proteins, such as collagen and mucins, into secretory vesicles (30,31). Loss of the mammalian ortholog of Tango1 (Mia3) in a murine model resulted in lethality with global defects in collagen secretion and extracellular matrix composition (31). Alterations in Tango1/Mia3 expression have also been associated with colon and hepatocellular carcinomas in humans (32). Previous work in Drosophila demonstrated that PGANT4 glycosylates Tango1, protecting it from Dfur2-mediated proteolysis in the digestive tract (30). Here, we show that Tango1 overexpression specifically in the secretory PR cells of the digestive tract can restore the mucinous membrane throughout the midgut to rescue epithelial viability and niche integrity, further demonstrating the crucial role  of the peritrophic membrane in digestive system homeostasis and health. These results suggest the possibility of exogenous Tango1 expression as a potential strategy to restore secretion, mucous membranes, and/or extracellular matrix composition and confer epithelial protection. The larval digestive system offers unique opportunities to investigate the role of the individual components of the mucinous membrane and restorative strategies in epithelial biology. Unlike the adult stage, the larval portion of the life cycle is devoted to continuous feeding and digestion to orchestrate the massive growth of cells and tissues in preparation for metamorphosis (23). As such, larvae consume many types of solid food and will readily ingest various compounds. Indeed, oral supplementation with an intestinal mucin (Muc2) partially rescued JAK/STAT signaling, suggesting that this could serve as a strategy for epithelial protection. Muc2 is a major component of the protective mucous membrane that lines the small intestine and colon of mammals (2,3). Muc2 is thought to confer lubrication for food passage as well as to form a barrier between microbes and epithelial cells of the digestive tract (2, 3). Our results suggest that Muc2 supplementation could be providing similar properties in the Drosophila digestive tract. Interestingly, supplementation with the gastric mucin (Muc5AC) dramatically exacerbated JAK/STAT signaling, suggesting that the different structural, rheological, or binding properties of each mucin are mediating distinct cellular responses in this system. Current work is focused on deciphering the specific functional regions of various secreted mucins and testing their ability to confer epithelial protection using this in vivo system.

Mucosal barrier loss disrupts the progenitor cell niche
Human diseases of the digestive tract are associated with disrupted mucinous linings, and disease severity is often correlated with the severity of barrier disruption (3,16,17). Interestingly, these diseases are also characterized by increased levels of the mammalian ortholog of Upd3 (IL-6), increased JAK/STAT activation, and increased cell proliferation (33)(34)(35), similar to what we see in Drosophila, suggesting conserved mechanisms for responding to mucosal disruption/injury. It is widely known that immune cells are one source of IL-6 in mammals, but recent studies have demonstrated that mechanically damaged epithelial and endothelial cells also produce IL-6 (36, 37). Here, we demonstrate that epithelial expression of Upd3 is both nec-essary and sufficient for the changes in PC fate and AMP proliferation, as disruption of the niche could be recapitulated by overexpression of upd3 from ECs and rescued by deletion of upd3. How peritrophic membrane loss is signaling to up-regulate upd3 expression in ECs is currently unknown. However, previous studies in the adult Drosophila digestive system have shown up-regulation of upd, upd2, and upd3 in response to enteric infection or damage-inducing agents, such as bleomycin or dextran sulfate sodium (38 -41), suggesting roles for both microbial insults and physical/mechanical damage to epithelial cells. Indeed, our results also suggest roles for microbial and mechanical damage in the absence of the peritrophic membrane, as both antibiotics and mucin supplementation were able to reduce upd3 expression and cell proliferation. This study demonstrates that the larval midgut can serve as a model system to study how cells/tissues sense and respond to damage as well as to decipher how upd3 is specifically activated in epithelial cells under various conditions.
As a mucous layer is present across most internal epithelial surfaces of our bodies, understanding the mechanisms by which it confers protection and epithelial homeostasis will inform us in treating various diseases affecting the integrity of this layer. Mucosal healing has been proposed as a treatment option for inflammatory bowel disease and other diseases of the digestive tract that are characterized by destruction of the mucosa and epithelial surfaces (42). Likewise, mucins are a component in some oral treatments for dry mouth caused by head and neck irradiation or Sjogren's syndrome (43). Other therapeutics for various autoimmune and inflammatory diseases include JAK inhibitors (Jakinibs) and drugs directed against particular cytokines (44,45). In this study, we show that genetic restoration of the peritrophic membrane can restore digestive system health and that antibiotic treatment or mucin supplementation can partially rescue damage-induced signaling cascades, suggesting that this Drosophila system may be a viable platform for testing compounds to remediate epithelial damage. Future studies will focus on testing newly emerging mucin mimetics (designed to confer epithelial protection and appropriate rheology/hydration), synthetic mucins (where the extent of glycosylation can be specifically modified) (46), glycan-based hydrogels (47), and drugs that target conserved steps In the absence of the peritrophic membrane (red), ECs are exposed to mechanical and microbial insults, resulting in damage and the up-regulation of the IL-6-like cytokine, Upd3. Upd3 from ECs increases JAK/STAT signaling in PCs, resulting in changes in PC morphology and fate and disruption of the progenitor cell niche. Once the niche is disrupted, AMPs are exposed to Upd3, resulting in activation of JAK/STAT signaling and abnormal cell proliferation.

Mucosal barrier loss disrupts the progenitor cell niche
in the JAK/STAT signaling cascade (44). Lessons learned in Drosophila may inform future strategies for functional restoration of mucosal protection.

Quantitative real-time PCR
Primers used are from the DRSC FlyPrimerBank (http:// www.flyrnai.org/flyprimerbank) 4 (56). To examine gene expression levels, midguts of third instar larvae were used to isolate RNA and perform real-time PCR. Briefly, RNA was isolated from 15 midguts of the proper genotype or condition using the RNAqueous-Micro kit (Invitrogen). cDNA synthesis was performed using the iScript cDNA synthesis kit (Bio-Rad). Quantitative RT-PCR was performed on a MyiQ real-time PCR thermocycler (Bio-Rad) using the SYBR Green PCR Master Mix (Bio-Rad). Analyzed products were assayed in triplicate and in multiple independent experiments. Data are presented as mean values, and error bars represent S.D.

In situ hybridization
upd3 RNA probes were prepared as described previously (52) and labeled using the DIG labeling kit (Roche Applied Science). Drosophila third instar larval midguts were dissected and fixed in 4% PFA/PBS with 0.6% Triton X-100. After fixation and washing, whole-mount midgut in situ hybridization was performed. Briefly, fixed guts were rinsed in hybridization buffer/ PBST (PBS with 0.1% Tween 20) and then incubated with hybridization buffer at 60°C for 1 h. Hybridization was performed by incubating with denatured 100 ng/ml DIG RNA probes at 60°C overnight. Guts were washed the next day and incubated with anti-DIG-POD (horseradish peroxidase) antibody (Roche Applied Science; 1:100) at 4°C overnight. Guts were washed and incubated with tyramide signal amplification plus Cy3 (PerkinElmer Life Sciences; 1:50 by amplification buffer) for 1 h. Finally, samples were washed with PBST and mounted with Vectashield mounting medium with DAPI (Vector Laboratories) and stored at 4°C.

EdU staining analysis
Briefly, midguts were dissected from third instar larvae and incubated in medium (Schneider's ϩ 10% FBS) with 10 M EdU (Invitrogen) for 45 min at room temperature. The following steps were performed according to the manufacturer's instructions. After incubation, tissues were fixed in 4% paraformaldehyde, washed in 3% BSA, and permeabilized in 0.5% Triton X-100. Then guts were incubated in Click-iT reaction mixture. After washing, samples were mounted in aqueous mounting medium with Vectashield mounting medium with DAPI (Vector Laboratories) on a slide with a spacer and imaged on a Zeiss LSM 510 confocal microscope or Nikon A1R confocal microscope. Images were processed using the LSM Imager Browser and ImageJ.

Cell counting
Cell proliferation was quantified by counting PH3 antibody or EdU-stained cells in five third instar larval midguts of each genotype under the appropriate conditions. Individual data points and mean values were shown. For the niche cell differentiation analysis, doubly positive Su(H) and Pdm-1 cells and singly positive Su(H) cells were counted in five third instar larval posterior midguts of each genotype. We analyzed the region of the posterior midgut ϳ40 m from the junction of the midgut and hindgut. Cells were scored in three areas (each of 100 ϫ 100 m) within the posterior midgut of each larvae of each genotype to generate the data shown in Fig. 2E. Data are presented as the percentage of doubly positive Su(H) and Pdm-1 cells over singly positive Su(H) cells.

Mucin purification and feeding
Porcine intestinal and gastric mucins were isolated by the previously described protocol (54,55) but excluding the cesium chloride density gradient ultracentrifugation. Briefly, mucus was removed from the epithelial surface of pig tissues by gentle scraping and then solubilized in the presence of saline and protease inhibitors. Insoluble debris was removed by centrifugation, and the high-molecular weight, periodic acid-Schiffpositive glycoproteins were isolated in the void volume fractions from Sepharose CL2B size-exclusion chromatography, followed by ultrafiltration and lyophilization. Muc2 and Muc5AC glycoproteins were verified by mass spectrometry as the predominant mucin species in the isolated intestinal and gastric mucins preparations, respectively. Fly crosses were performed on egg lay plates as described (52). Eggs were then transferred to tubes containing food mixed with either purified mucins (600 l of 5 mg/ml mucin added to 1 g of MM media from KD Medical), carboxymethylcellulose (600 l of 5 mg/ml CMC added to 1 g of MM media), or water (600 l added to 1 g of MM media). After 3 days of feeding, midguts were dissected from 15 larvae, and RNA was extracted for real-time PCR.

Antibiotic feeding
Crossed flies were set up in vials containing food (MM media, KD Medical) with antibiotics (penicillin (100 units/ml) and streptomycin (0.1 mg/ml)). After egg laying, parents were removed, and progeny were incubated in the bottle for an additional ϳ3-4 days. 15 midguts from each treatment were dissected from third instar larvae, and cell proliferation and gene expression were examined. Four independent feeding experiments were performed.

Statistics
Experiments were performed three or more times, and averages for each experiment were calculated. Error bars represent S.D., and significance values were calculated using Student's t test or analysis of variance for comparison of data from three or more groups. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.
Author contributions-L. Z. and K. G. T. H. designed all experiments. L. Z. carried out all experiments shown. K. R. and B. T. purified and characterized the mucins. K. G. T. H. wrote the manuscript, with assistance from L. Z. and K. R.