Stem-cell-specific endocytic degradation defects lead to intestinal dysplasia in Drosophila

ABSTRACT UV radiation resistance-associated gene (UVRAG) is a tumor suppressor involved in autophagy, endocytosis and DNA damage repair, but how its loss contributes to colorectal cancer is poorly understood. Here, we show that UVRAG deficiency in Drosophila intestinal stem cells leads to uncontrolled proliferation and impaired differentiation without preventing autophagy. As a result, affected animals suffer from gut dysfunction and short lifespan. Dysplasia upon loss of UVRAG is characterized by the accumulation of endocytosed ligands and sustained activation of STAT and JNK signaling, and attenuation of these pathways suppresses stem cell hyperproliferation. Importantly, the inhibition of early (dynamin-dependent) or late (Rab7-dependent) steps of endocytosis in intestinal stem cells also induces hyperproliferation and dysplasia. Our data raise the possibility that endocytic, but not autophagic, defects contribute to UVRAG-deficient colorectal cancer development in humans. Drosophila Collection: Intestinal-stem-cell-specific loss of the Drosophila ortholog of the tumor suppressor UVRAG, which is implicated in colorectal cancer, leads to endocytic defects and dysplasia.


INTRODUCTION
UV radiation resistance-associated gene (UVRAG) encodes a subunit of the Vps34 kinase complex, which is important for the generation of phosphatidylinositol 3-phosphate (PI3P). UVRAG promotes the biogenesis of this phospholipid by binding to the autophagic tumor suppressor Beclin-1 within this complex (Itakura et al., 2008;Liang et al., 2006). PI3P is enriched not only on endosomes but also on early autophagic structures, including phagophores and autophagosomes, and it mediates the recruitment of downstream effectors involved in autophagosome and endosome maturation (Kutateladze, 2012). UVRAG is also reported to bind to HOPS (Liang et al., 2008), a tethering complex that promotes multiple trafficking routes to lysosomes, including autophagy, endocytosis, and the biosynthetic transport of hydrolytic enzymes and membrane proteins. In cancer cell lines, overexpression of UVRAG is found to promote autophagy and reduce cell proliferation, raising the possibility that its tumor suppressor function involves the regulation of autophagy (Liang et al., 2006). In line with that possibility, transposon-induced Uvrag-mutant mice display impaired autophagic degradation in cardiomyocytes and fibroblasts . However, conflicting reports have also been recently published that have found no crucial role for UVRAG in autophagosome formation and fusion with lysosomes (Jiang et al., 2014;Takáts et al., 2014). A potential explanation for these discrepancies is cell-type-specific differences in UVRAG function.
It is worth noting that UVRAG might also act independently of its roles in vesicular trafficking. For example, it contributes to the maintenance of chromosomal stability by promoting repair after DNA damage (Zhao et al., 2012). In fact, UVRAG was originally discovered as a factor that partially complements the sensitivity to UV of xeroderma pigmentosum group C cells (Teitz et al., 1990). Thus, UVRAG function must be studied in the proper cellular and developmental context in order to understand its tissue-specific roles.
Progression of solid tumors is accelerated by genome instability owing to either chromosome instability or microsatellite instability. In the latter case, dysfunction of the mismatch repair pathway leads to somatic mutations in genes containing repetitive DNA sequences. One such gene is UVRAG, which contains an A10 exonic repeat. Mutations in UVRAG have indeed been found in colorectal cancer cells with microsatellite instability, indicating that the protein could function as a tumor suppressor in humans (Ionov et al., 2004). As a consequence, UVRAG loss has been suggested to contribute to the development of colorectal cancer, but there is still no experimental support for the in vivo relevance of this model.
There are remarkable similarities between the cell types and signaling pathways that are important for fly and mammalian gut physiology (Jiang and Edgar, 2012). For example, proliferation of intestinal stem cells (ISCs) in Drosophila ensures self renewal and generates progenitor cells called enteroblasts (EBs) that produce enterocytes and enteroendocrine cells, similar to the functions of stem cells residing at the base of Lieberkühn's crypts in the mammalian intestine. Ingested pathogens and toxins damage the gut and trigger a regeneration response through increased proliferation of stem cells and differentiation of progeny, both in flies and mammals (Jiang and Edgar, 2012). We thus decided to analyze whether the role of UVRAG as a tumor suppressor is evolutionarily conserved in the adult Drosophila intestine and to understand which of its diverse functions might be relevant in a setting similar to that during colorectal cancer development.

UVRAG is important for endosome maturation in ISCs
UVRAG mutations arise from microsatellite instability in human colorectal cancers. To understand the consequences of the adultonset loss of this gene, we induced RNA interference (RNAi)mediated silencing of Uvrag in midgut ISCs of adult Drosophila using a standard temperature-sensitive gene expression system. This method allows genetic manipulation of escargot (esg)-positive ISCs and differentiating progenitors (EBs) in adult flies, as gene silencing (or overexpression) and GFP expression can be triggered by shifting animals to 29°C (Micchelli and Perrimon, 2006). Knockdown of Uvrag in esg-GFP-positive cells (where GFP is expressed under the esg promoter) strongly decreased the number of GFP-tagged FYVE dots, which mark PI3P-positive vesicles, indicating efficient gene silencing (Fig. 1A).
Loss of UVRAG results in the accumulation of plasma membrane receptors and ligands in stalled endocytic compartments in cultured human cells, and also in cells of the developing Drosophila eye and wing, which might interfere with the activity of various signaling pathways (Jiang et al., 2014;Lee et al., 2011;Lorincz et al., 2014). During ISC proliferation and differentiation, Notch receptor and its ligand Delta traffic via endosomes (Montagne and Gonzalez-Gaitan, 2014). In line with this, RNAi against Uvrag (UVRAG RNAi) in esg-GFP-positive cells resulted in a striking intracellular accumulation of the Notch ligand Delta ( Fig. 1B; Fig. S1A). To confirm this finding, we generated mitotic clones in the gut that were homozygous mutants for previously described Uvrag-null alleles (Lee et al., 2011;Takáts et al., 2014). Both mutations in Uvrag also resulted in intracellular Delta accumulation (Fig. S1B), in line with our RNAi data. Basal levels of the Wnt signaling ligand Wingless/Wg can be detected in ISCs and EBs (Cordero et al., 2012;Lin et al., 2008), and the loss of Uvrag results in increased punctate intracellular Wg signals (Fig. S1C), again indicating impaired endolysosomal degradation.
UVRAG-or Vps34-dependent production of the phospholipid PI3P is important for endosome maturation. We thus analyzed Rab7-positive endosomes in control and UVRAG loss-of-function stem cells. Indeed, quantification of Rab7-GFP vesicles revealed a clear increase in the percentage of cytoplasmic area occupied in UVRAG-deficient cells (Fig. S1D).

UVRAG deficiency leads to intestinal dysplasia
Strikingly, the loss of UVRAG led to a remarkable expansion of both ISCs containing the endocytosed Notch ligand Delta and Delta-negative EBs (Fig. 1B). The overall cell number in the posterior midgut of these animals also increased (Fig. 1C), as well as the thickness of the intestinal wall (Fig. S1E).
Staining for the mitotic marker phosphorylated histone H3 revealed that ISCs proliferated more upon loss of UVRAG (Fig. 1D). This was further confirmed with the help of genetic mosaics in which Uvrag-mutant clones were induced by somatic recombination at the adult stage: GFP-positive ISCs that were homozygous mutant for either one of the two null alleles produced many more cells per clone than control cells during the same time window ( Fig. 1E; Fig. S1F). Importantly, the expansion of Uvragmutant clones was suppressed by expression of wild-type Uvrag in these cells in both cases ( Fig. 1E; Fig. S1F), confirming that the intestinal dysplasia is solely due to UVRAG deficiency. Uvrag overexpression on its own caused no obvious alteration of the esg-GFP compartment in the posterior midgut (Fig. S1G).

Loss of UVRAG causes differentiation defects and cell shape changes
ISCs have the ability to self renew and to generate EB daughter cells in which Notch signaling is activated by ISCs that are positive for its ligand Delta (Ohlstein and Spradling, 2007). In wild-type guts, most EBs differentiate into enterocytes. EBs become polyploid and downregulate both esg expression and Notch signaling during this differentiation process. In contrast, UVRAG-deficient cells retained both esg-Gal4-driven UAS-mCherry (esg-mCherry) expression and active Notch signaling even after becoming polyploid ( Fig. 2A). Moreover, the esg-GFP-positive cells with UVRAG RNAi failed to express the enterocyte differentiation marker Myo1A (Fig. S2A).
The activity of target of rapamycin (Tor) kinase is relatively low in ISCs, which maintains their stem cell identity. Notch-mediated activation of this important positive regulator of cell growth is reported to promote the differentiation of EBs into enterocytes (Kapuria et al., 2012). In line with the impaired differentiation of UVRAG RNAi cells, all esg-GFP-positive cells showed lower Tor activity than surrounding enterocytes based on the phosphorylation levels of its substrate 4EBP (Fig. S2B).
In addition, the size of both Delta-negative EBs and Deltapositive ISCs was increased by the loss of UVRAG (Fig. 2B), and cytoplasmic protrusions that extended towards the underlying muscle layer appeared on these cells (Fig. 2C). The number of enteroendocrine cells was unchanged in UVRAG-deficient intestines (Fig. S2C). As these cells represent another lineage that also differentiates from EBs (Jiang and Edgar, 2012), this result indicates that UVRAG loss specifically leads to the accumulation of EBs that are committed to an EC fate.
UVRAG is required for gut function and survival during normal conditions and in response to stress To understand the physiological relevance of intestinal dysplasia caused by the lack of UVRAG, we analyzed gut function in wildtype and esg-specific loss-of-function animals. Feeding flies a bluedye-containing food revealed that animals with dysplastic guts defecated significantly less 24 h after starting the experiments (Fig. 3A), even though their food intake was similar to that of controls ( Fig. S3A), indicating increased retention of ingested food. As proper gut function is a crucial determinant of Drosophila lifespan (Biteau et al., 2010), we next followed the long-term survival of control flies and those with knockdown of Uvrag in ISCs and EBs. These experiments revealed an obvious reduction of lifespan in the animals with UVRAG-deficient dysplastic guts (Fig. 3B).
We hypothesized that intestinal dysplasia sensitizes the flies to stressors that target the intestine. A common model for experimental colitis is to feed Drosophila or mice with the toxin dextran sodium sulfate (DSS) (Amcheslavsky et al., 2009). DSS treatment induces a regeneration response that is characterized by large-scale ISC proliferation and subsequent differentiation in wild-type guts; however, DSS treatment did not further enhance cell proliferation in dysplastic UVRAG-deficient guts (Fig. S3B). In line with this, DSS killed flies undergoing UVRAG RNAi in esg-GFP-positive cells much faster than it killed control flies (Fig. 3C).
Another commonly used experimental stressor of the gut is bacterial infection (Lemaitre and Miguel-Aliaga, 2013). Indeed, UVRAG loss-of-function animals were highly sensitive to oral administration of the pathogen Pseudomonas aeruginosa (Fig. 3D).

UVRAG regulates STAT and JNK signaling in the gut
We tested the activity of selected signaling pathways in UVRAGdeficient intestines that could play a crucial role in the proliferation of ISCs of the posterior midgut (Jiang and Edgar, 2012;Lemaitre and Miguel-Aliaga, 2013). Receptor tyrosine kinases such as epidermal growth factor receptor (Egfr) and insulin receptor (InR) signal through the Ras-MAPK cascade and Akt, respectively. We could not detect increased activation of these pathways in UVRAG RNAi cells, based on similar staining intensities using antibodies specific for the active and phosphorylated forms of MAPK and Akt, respectively (Fig. S4A,B).
Increased cytokine signaling through the JAK-STAT pathway has been shown to activate the proliferation of ISCs . Indeed, activation of a GFP reporter whose expression is controlled by ten Stat92E binding sites was obvious in all esg-mCherry-positive cells undergoing UVRAG RNAi (Fig. 4A). Accordingly, we found upregulation of cytokine expression using reporters for Unpaired1 (upd1) and Unpaired3 (upd3) (Fig. 4B,C).
Interestingly, loss of UVRAG triggered expression of the Unpaired1 reporter in esg-GFP-positive cells, whereas the Unpaired3 reporter was induced in the immediate neighbors of UVRAG-deficient cells clones of Uvrag LL null mutant cells contain more cells than control clones two weeks after clone induction, which is suppressed by transgenic expression of wild-type Uvrag. The quantification shows data from this panel as well as Fig. S1F, showing that mitotic clones for the independent null mutation Uvrag B21 produces the same phenotypes. Red lines indicate median in A-D, and distribution of data points in E, with dotted red lines indicating median. Because experiments were carried out in parallel, the control and UVRAG RNAi data in D is the same as that shown in Fig. 5C,D, and the control data in B,D is the same as in Fig. 6G. (Fig. 4B,C). These results indicate that both autocrine and nichederived paracrine signals contribute to activation of STAT signaling in Uvrag-knockdown cells.
We also looked at the Jun N-terminal kinase (JNK; also known as Basket in Drosophila) pathway, another positive regulator of ISC proliferation in Drosophila (Biteau et al., 2008;Tang et al., 2013). The activity of this signaling route is very low in stem cells of wild-type guts. One of the transcriptional targets of JNK signaling is puckered ( puc), a gene that encodes a phosphatase that antagonizes JNK activity. We used a puc-lacZ reporter line as a readout for JNK activation in the context of UVRAG loss. This transcriptional reporter is only active in enterocytes in wild-type guts (Biteau et al., 2008). In line with our data on phosphorylated JNK, the loss of UVRAG resulted in ectopic puc-lacZ expression in esg-GFP-positive cells (Fig. 4D). Moreover, immunolabeling using an antibody specific for the active phosphorylated form of JNK revealed a striking upregulation of phosphorylated JNK signal in all esg-GFP-positive cells undergoing UVRAG RNAi (Fig. S4C).
JNK and STAT signaling are required to sustain hyperproliferation induced by the loss of UVRAG We sought to establish the relevance of sustained activation of these signaling pathways during hyperproliferation induced by UVRAG loss by performing epistasis analyses. Expression of a dominantnegative form of Basket (the Drosophila homolog of JNK) in esg-GFP-positive cells undergoing UVRAG RNAi suppressed hyperproliferation and completely restored the number of ISCs and EBs to wild-type levels ( Fig. 5A-C). The knockdown of Stat92E also suppressed the increased mitotic activity and decreased the number of esg-GFP-positive cells in UVRAG-deficient guts ( Fig. 5A-C).
In control experiments, the expression of dominant-negative Basket did not affect the number of esg-GFP-positive cells and ISCs in 3-week-old animals, although it did reduce the number of mitotic cells at this time point (Fig. 5A,B,D). RNAi knockdown of Stat92E increased the number of ISCs and EBs in 3-week-old animals and also led to the accumulation of mitotic cells (Fig. 5A,B,D), in line with previous reports that loss of STAT function leads to the accumulation of progenitors as early as 5-8 days owing to a block of differentiation (Beebe et al., 2010;Jiang et al., 2009).
In contrast with these results, blocking Wingless signaling through the expression of a dominant-negative version of the downstream TCF transcription factor eliminated most esg-GFPpositive cells in UVRAG-deficient and also wild-type cells (Fig. S5A,B,E-G), indicating that the Wingless pathway is required for the maintenance of ISCs regardless of the genetic background.
We sought to test whether the above pathways are specifically involved in the dysplasia of UVRAG-deficient ISCs. We thus quantified the number of esg-GFP-positive cells using genetic backgrounds that were heterozygous for STAT, JNK and Wingless components. No statistically significant differences in the number of mitoses were observed in animals heterozygous for Wingless, Stat92E (STAT) or Basket (JNK) (Fig. S6). In contrast, heterozygosity for Stat92E or Basket, but not Wingless, suppressed the hyperproliferation of UVRAG-deficient cells (Fig. S6). These data indicate that STAT and JNK signaling are specifically required to sustain UVRAG-deficiency-induced intestinal dysplasia.

Activation of Notch promotes differentiation of UVRAG lossof-function cells
Deregulation of Notch signaling has been found to cause a male genitalia rotation defect in hypomorphic mutant Uvrag animals because Notch RNAi rescues this mutant phenotype (Lee et al., 2011). Notch is a major regulator of ISC fate in Drosophila: its inhibition causes stem cell tumors, whereas ectopic Notch activity promotes the differentiation of ISCs into enterocytes (Micchelli and Perrimon, 2006). We thus tested whether the dysplasia of UVRAGdeficient guts could be rescued by manipulating Notch. Simultaneous knockdown of Notch and Uvrag resulted in a striking accumulation of ISCs (Fig. S7A), instead of rescuing the dysplastic phenotype. Interestingly, overexpression of the active fragment of Notch could still induce the differentiation of UVRAGdeficient ISCs and EBs to enterocytes (Fig. S7B).
UVRAG-loss-induced dysplasia is independent of autophagy in the gut UVRAG is famous for its dual roles in autophagy as it has been implicated both in autophagosome formation as a Vps34 kinase complex subunit and in fusion with lysosomes because it binds to the HOPS tethering complex (Liang et al., 2006(Liang et al., , 2008. We thus tested whether its tumor suppressor function is associated with impaired autophagy in the gut. An established test for autophagic degradation (flux) is to measure the level of the selective cargo p62, which is also known as Ref(2)P in Drosophila (Nezis et al., 2008;Pircs et al., 2012). No statistically significant difference was detected in endogenous p62 levels between esg-GFP-positive wild-type and UVRAG RNAi cells (Fig. 6A,C). Similarly, the direct visualization of autophagosomes and autolysosomes using mCherry-tagged Atg8a, which labels both structures (Mauvezin et al., 2014;Nagy et al., 2015), did not show a statistically significant difference between control and UVRAGdeficient esg-GFP-positive cells (Fig. 6B,C). Knocking down the core autophagy genes Atg5 or Atg14 suppressed the dysplasia induced by UVRAG loss (Fig. 6D-F;  Fig. S5D). However, RNAi against Atg5 or Atg14 also led to loss of esg-GFP-positive cells and decreased the number of mitoses (Fig. 6D,E,G; Fig. S5C), suggesting that autophagy is required for ISC maintenance in general, instead of for the promotion of dysplasia specifically in the absence of UVRAG. Similarly, autophagy has recently been shown to maintain muscle stem cell function and survival in mice (García-Prat et al., 2016).

Inhibition of endocytic degradation leads to intestinal dysplasia
Our data suggested that it is not the autophagic but the endocytic function of UVRAG that is important for preventing intestinal dysplasia. We thus analyzed the consequences of disrupting endocytosis in stem cells by expressing a dominant-negative version of shibire (shi DN ), the Drosophila homolog of dynamin, which is a routinely used tool to block the endocytic downregulation of plasma membrane receptors (Moline et al., 1999). Indeed, shi DN expression in esg-GFP-positive cells led to a striking enhancement of mitotic activity and a major expansion of ISCs (Fig. 7A).
The organismal effects of inhibiting dynamin-dependent endocytosis in ISCs were much more severe than those seen upon UVRAG loss as these animals died at ages of 10 to 14 days. We thus tested another well-known endocytic regulator, Rab7, which acts downstream of shibire/dynamin and promotes endosome maturation and cargo degradation (Platta and Stenmark, 2011), similar to UVRAG. Importantly, knockdown of Rab7 in esg-GFP-positive cells perfectly phenocopied the loss of UVRAG, based on the increased number of ISCs and EBs seen in the midguts of 3-weekold animals (Fig. 7B). The JNK activity reporter puc-lacZ is activated in the majority of esg-GFP-positive Uvragknockdown cells (arrowheads), in addition to its normal expression in enterocytes. Note that β-gal, as expressed from lacZ, is in the nucleus owing to a nuclear localization signal.

DISCUSSION
UVRAG encodes a homolog of yeast Vps38 in metazoans. UVRAG/ Vps38 and Atg14 are mutually exclusive subunits of two different Vps34 lipid kinase complexes, both of which contain Vps34, Vps15 and Atg6/Beclin 1 (Itakura et al., 2008). It is well established that Vps38 is required for endosome maturation and vacuolar and lysosomal protein sorting, whereas Atg14 is specific for autophagy in yeast (Kihara et al., 2001). However, the function of UVRAG is much more controversial in mammalian cells. Although UVRAG was originally found to have dual roles in autophagy through promotion of autophagosome formation and fusion with lysosomes in various cultured cell lines based on, predominantly, overexpression experiments (Liang et al., 2006(Liang et al., , 2008, recent reports have described that autophagosomes are normally generated and fused with lysosomes in the absence of UVRAG in cultured mammalian (HeLa) cells and in the Drosophila fat body (Jiang et al., 2014;Takáts et al., 2014).
The discoveries of UVRAG mutations in colorectal cancer cells, and that its overexpression increases autophagy and suppresses the proliferation of certain cancer cell lines, altogether suggest that this gene functions as an autophagic tumor suppressor (Ionov et al., 2004;Liang et al., 2006). Such a role for UVRAG is thought to be related to its binding to Beclin 1, a haploinsufficient tumor suppressor gene required for autophagy (Qu et al., 2003;Yue et al., 2003). UVRAG appears to play roles similar to yeast Vps38 in the Drosophila fat body, and developing eye and wing: its loss leads to the accumulation of multiple endocytic receptors and ligands in an endosomal compartment, impaired trafficking of Lamp1 and Cathepsin L to the lysosome, and defects in the biogenesis of lysosome-related pigment granules (Lee et al., 2011;Lorincz et al., 2014;Takáts et al., 2014). However, whether this gene is also required for the maintenance of intestinal homeostasis in Drosophila was unclear because the loss of UVRAG did not lead to uncontrolled cell proliferation in the developing eye or wing according to these reports. Our results showing that Uvrag deficiency causes intestinal dysplasia suggest that this gene is also important for the proper functioning of the adult gut in Drosophila.
A surprising aspect of our work is that UVRAG appears to function independently of autophagy in the intestine. There are other lines of evidence that also support that UVRAG has a more important role in endocytic maturation than in autophagy. First, it has been shown that truncating mutations in UVRAG that are associated with microsatellite-unstable colon cancer cell lines do not disrupt autophagy (Knaevelsrud et al., 2010). Second, UVRAG depletion in HeLa cells does not prevent the formation or fusion of autophagosomes with lysosomes, but it does interfere with Egfr degradation (Jiang et al., 2014). Third, a very recent paper has shown that overexpression of the colorectal-cancer-associated truncated form of UVRAG promotes tumorigenesis independently of autophagy status, that is, both in control and Atg5-knockout cells (He et al., 2015). That paper, again, relied on the overexpression of full-length or truncated forms of UVRAG, rather than the analysis of cancer-related mutations of the endogenous locus. Fourth, the endocytic function of UVRAG has been found to be required for developmental axon pruning that is independent of autophagy in Drosophila (Issman-Zecharya and Schuldiner, 2014).
Our results indicate that UVRAG loss is accompanied with the sustained activation of JNK and STAT signaling in ISCs and EBs, and that these pathways are required for dysplasia in this setting. Sustained activation of these signaling routes is likely to be connected to the disruption of endocytic flux in the absence of UVRAG, because inhibiting endocytic uptake or degradation through dominant-negative dynamin expression or RNAi of Rab7, respectively, also leads to intestinal dysplasia (Fig. 8). It is worth noting that the effects of inhibiting Shibire/dynamin function led to a much more severe hyperproliferation of ISCs and early death of animals. In line with this, the loss of early endocytic regulators, such as Rab5, in the developing eye causes overproliferation of cells and lethality during metamorphosis (Lu and Bilder, 2005). Although eye development is not perturbed by the loss of the late endocytic regulators UVRAG or Rab7 (Cherry et al., 2013;Takáts et al., 2014), these proteins are clearly important for controlling ISC proliferation and differentiation.
A recent paper shows that hundreds of RNAi lines cause the expansion of the esg-GFP compartment in 1-week-old animals (Zeng et al., 2015), which might be due to an unspecific ISC stress response in some cases. However, several lines of evidence support that impaired UVRAG-dependent endocytic degradation is specifically required to prevent intestinal dysplasia. First of all, activation of JNK stress signaling in esg-GFP-positive cells induces short-term ISC proliferation (Biteau et al., 2008), and almost all stem cells are lost through apoptosis by the 2-to 3-week age (our unpublished data), the time when the Uvrag-mutant phenotype becomes obvious. In fact, UVRAG loss resembles an early-onset age-associated dysplasia that is normally observed in old (30-60 days) flies and involves the simultaneous activation of both JNK and STAT signaling (Biteau et al., 2008;Osman et al., 2012). Second, UVRAG RNAi in ISCs and EBs leads to paracrine week-old animals that is indistinguishable from that seen in UVRAG RNAi animals of a similar age. Note that the analysis of ISCs and EBs in Rab7 loss-offunction posterior midguts is compared to the same control animals as shown in Fig. 1A as experiments were performed in parallel.
activation of the cytokine Unpaired3 in enterocytes, one of the hallmarks of niche appropriation by Notch-negative tumors (Patel et al., 2015). However, autocrine expression of the Unpaired proteins and JNK activation is observed in Uvrag-knockdown cells, unlike in Notch-negative tumors, and EBs with active Notch signaling accumulate in the absence of UVRAG, so the two phenotypes are clearly different. Third, it is the loss of autophagy that could be expected to mimic a stress response and perhaps induce stem cell tumors, but this does not seem to be the case -ISCs with Atg5 or Atg14 RNAi proliferate less in 3-week-old animals and an overall decrease of the esg-GFP compartment is seen, as opposed to the Uvrag-deletion phenotype.
Taken together, our work indicates that endocytic maturation and degradation serves to prevent early-onset intestinal dysplasia in Drosophila, and its deregulation could be relevant for the development of colorectal cancer in humans.

DSS treatment and infections
At 10 days old, flies were transferred to empty vials containing 2.5×3.5 cm filter papers soaked in 800 μl of 5% sucrose solution or 3% DSS dissolved in 5% sucrose solution for 2 days. Flies were transferred to fresh medium after treatment twice a week, and their survival was monitored. Stem cell proliferation was assessed 1 day after treatment. Infection with Pseudomonas was performed as described previously ).