Septate junctions regulate gut homeostasis through regulation of stem cell proliferation and enterocyte behavior in Drosophila

The intestinal epithelium serves as a physical barrier that prevents infiltration of food-derived harmful substances, microbial contaminants and digestive enzymes into the body. To constitute an effective intestinal barrier, specialized cell–cell junctions, namely occluding junctions, play a crucial role in regulating free diffusion of solutes through the paracellular route. Septate junctions (SJs) are occluding junctions in invertebrates and act as the functional counterparts of vertebrate tight junctions (Anderson and Van Itallie, 2009; Furuse and Tsukita, 2006; Lane et al., 1994; Tepass and Hartenstein, 1994). In arthropods, morphological variation of SJs has been observed among different types of epithelia (Lane et al., 1994; Tepass and Hartenstein, 1994). Most ectodermally derived epithelia and the perineural sheath have pleated SJs (pSJs), while endodermally derived epithelia, including the midgut, have smooth SJs (sSJs) (Lane et al., 1994; Tepass and Hartenstein, 1994). Genetic and molecular analyses in Drosophila have revealed a number of molecular components and functional properties of pSJs (Tepass et al., 2001; Wu and Beitel, 2004; Banerjee et al., 2006; Izumi and Furuse, 2014). In contrast, few studies have been carried out on sSJs, and their physiological role is not well understood. Recently, we identified three sSJ-specific membrane proteins: Ssk, Mesh and Tsp2A. Ssk is a four transmembrane domain-containing protein (Yanagihashi et al., 2012). Mesh is a single-pass membrane protein containing a large extracellular region (Izumi et al., 2012). Tsp2A belongs to the tetraspanin family (Izumi et al., 2016). Zygotic loss of any of these proteins results in embryonic lethality just before hatching or at the first-instar larval stage with impairment of sSJ formation as well as epithelial barrier function, suggesting that sSJs have critical roles (Yanagihashi et al., 2012; Izumi et al., 2012, 2016; Izumi and Furuse, 2014; Furuse and Izumi, 2017).

The Drosophila adult midgut epithelium is composed of absorptive enterocytes (ECs), secretory enteroendocrine cells (EEs), intestinal stem cells (ISCs), EC progenitors (enteroblasts; EBs) and EE progenitors (enteroendocrine mother cells; EMCs) (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006; Guo and Ohlstein, 2015). The sSJs are formed between adjacent ECs, and between ECs and EEs (Resnik-Docampo et al., 2017). ECs and EEs are continuously renewed by proliferation and differentiation of the ISC lineage through the production of intermediate differentiating cells, EBs and EMCs. This renewal of adult midgut epithelial cells is critical for maintenance of homeostasis in the adult midgut. Recent studies have suggested that sSJs influence proliferation and differentiation of the ISC lineage in the adult midgut. Experimental suppression of Gliotactin, a tricellular junction-associated protein, in ECs leads to epithelial barrier dysfunction, increased ISC proliferation and blockade of EC differentiation in young flies (Resnik-Docampo et al., 2017). Moreover, loss of mesh and Tsp2A in clones causes defects in polarization and integration of ECs in the adult midgut (Chen et al., 2018). Thus, it will be interesting to examine the role of sSJs in the adult midgut in terms of the regulation of ISC proliferation and tissue homeostasis by genetic ablation of the sSJ-protein-encoding genes throughout ECs.

Here, we describe that depletion of the sSJ proteins Ssk, Mesh and Tsp2A from ECs causes remarkably reduced lifespan and midgut barrier dysfunction in flies. The sSJ-protein-deficient flies show intestinal hypertrophy accompanied by accumulation of morphologically aberrant ECs and increased stem cell proliferation. Interestingly, we show that the interleukin-6-like cytokines Unpaired 2 (Upd2) and/or Unpaired 3 (Upd3) are involved in this intestinal hypertrophy. We conclude that sSJs are crucial for the regulation of stem cell proliferation and EC behavior in the Drosophila adult midgut.

To investigate the effect of sSJ protein depletion on the Drosophila adult midgut, inducible RNAi for the sSJ proteins was expressed using the GAL4/UAS system with an EC-specific driver Myo1A-GAL4 and the temperature-sensitive GAL4 repressor, tubGal80^ts (McGuire et al., 2004; Jiang et al., 2009). The UAS-RNAi lines used for the sSJ proteins were UAS-ssk-RNAi, mesh-RNAi (NIG-Fly stock 12074R-1) and Tsp2A-RNAi (NIG-Fly stock 11415R-2), which had been previously confirmed to effectively reduce the expression of their respective sSJ proteins (Yanagihashi et al., 2012; Izumi et al., 2012, 2016). Myo1A-GAL4 tubGal80^ts UAS-CD8-GFP and UAS-ssk-RNAi, -mesh-RNAi or -Tsp2A-RNAi (Myo1A^ts>ssk-RNAi, mesh-RNAi or Tsp2A-RNAi) flies were raised to adults at 18°C (permissive temperature) and then shifted to 29°C (non-permissive temperature) to inactivate GAL80, leading to activation of the GAL4/UAS system to express each UAS-driven transgene. Almost all flies expressing the RNAi that targets the sSJ protein transcripts (hereafter referred to as sSJp-RNAis) died within 10 days after transgene induction, while more than 95% of control flies (Myo1A^ts>CD8-GFP) survived until 15 days after induction (Fig. 1A; Fig. S1A,B). Thus, reduced expression of sSJ proteins in ECs results in remarkably shortened lifespan in adult flies. Next, we examined whether the barrier function of the midgut was disrupted in sSJp-RNAis flies. According to the barrier integrity assay (Smurf assay) method (Rera et al., 2011, 2012), flies were fed a non-absorbable 800-Da blue food dye in sucrose solution and observed for leakage of the dye from the midgut. At 3 days after transgene induction, reduced expression of any of the sSJ proteins in ECs led to a significant increase in flies with blue dye throughout their body cavity, indicating defective midgut barrier function (Fig. 1C). The proportion of flies with midgut barrier dysfunction was further increased at 5 days after transgene induction, compared with age-matched controls (Fig. 1B,C). Thus, we confirmed that sSJ proteins are required for the barrier function in the adult midgut, similar to the observations in the larval midgut (Yanagihashi et al., 2012; Izumi et al., 2012, 2016).

The shortened lifespan and midgut barrier dysfunction in sSJ-protein-deficient flies prompted us to examine the organization of their midgut epithelium. At 5 days after transgene induction, a typical simple epithelium in which ECs expressed CD8–GFP driven by Myo1A-GAL4 was observed in the control midgut (Fig. 2A,E,J,J′). Intriguingly, the organization of the epithelium was severely disrupted in the sSJp-RNAis midgut; a large number of ECs failed to become integrated into the epithelium and instead accumulated throughout the midgut lumen (Fig. 2B–D). In addition, the ECs exhibited a variety of aberrant appearances, implying a polarity defect (Fig. 2B–D,J–M′). Indeed, abnormal distributions of F-actin and Dlg, a cell polarity protein, were observed in sSJp-RNAis ECs (Fig. 2B–D,K–M,K′–M′). The most posterior part of the midgut had a hypertrophic phenotype; the lumen was filled with a large number of morphologically aberrant ECs and the diameter was severely expanded (Fig. 2F–I). We confirmed that expression of sSJp-RNAis in ECs led to decreased levels of the respective target proteins and mislocalization of other sSJ proteins in the midgut (Fig. S2). Additional RNAi lines for mesh and Tsp2A showed essentially the same phenotypes (Fig. S3). Toluidine Blue staining of semi-thin sections and ultrastructural analysis by transmission electron microscopy confirmed that morphologically aberrant cells were stratified in the sSJp-RNAis midgut (Fig. 3B–D,F–H), while a monolayer of ECs with developed microvilli facing the lumen was observed in the control midgut (Fig. 3A,E). Notably, microvilli-like structures were often observed between stratified ECs in the sSJp-RNAis midgut (Fig. 3F,H–J). Thus, depletion of any of the sSJ proteins from ECs causes intestinal hypertrophy accompanied by dysplasia-like accumulation of ECs in the midgut lumen, suggesting that sSJs are required for homeostasis of the midgut organization.

We speculated that the EC accumulation was caused by overproduction of ECs in the sSJp-RNAis midgut. Because regeneration of ECs depends on proliferation and differentiation of the ISC linage, we examined whether proliferation of ISCs was increased in the sSJp-RNAis midgut. We stained the midgut with an antibody against phospho-histone H3 (PH3), a mitotic marker, and found that PH3-positive cells were markedly increased in the sSJp-RNAis midgut compared with the control midgut (Fig. 4A–D,E). Furthermore, immunostaining of the midgut with an antibody against Delta, an ISC marker (Ohlstein and Spradling, 2007), revealed that the number of ISCs were increased in the sSJp-RNAis midgut compared with the control midgut (Fig. 4A–D). We also confirmed that PH3-positive cells were Delta positive (Fig. 4A–D). Furthermore, the number of cells expressing Escargot (Esg)-LacZ, an ISC and EB marker (Micchelli and Perrimon, 2006), was increased in the ssk- or mesh-RNAi midgut (Fig. 4F–H). These results indicate that reduced expression of sSJ proteins in ECs leads to increased ISC proliferation. Of note, Esg–LacZ signals were often observed in cells expressing CD8–GFP driven by Myo1A-GAL4 in the ssk- or mesh-RNAi midgut (Fig. 4J–K″,L–N), suggesting that sSJ protein depletion causes mis-differentiation of the ISC linage in the midgut.

In the adult midgut, the Ras-MAPK and Jak-Stat signaling pathways are required for activation of ISC proliferation during the regeneration of epithelial cells (Beebe et al., 2010; Buchon et al., 2010; Karpowicz et al., 2010; Shaw et al., 2010; Biteau and Jasper, 2011; Jiang et al., 2009, 2011; Osman et al., 2012; Zhou et al., 2013). Therefore, we examined whether these signaling pathways were activated in the sSJp-RNAis midgut. To monitor Ras-MAPK pathway activity, we examined the levels of di-phosphorylated ERK (dpERK; in Drosophila the antibody recognizes the phosphorylated form of the Rolled protein) (Gabay et al., 1997). In control flies, the dpERK signals were barely detectable in the midgut (Fig. 5A,E). In contrast, intense dpERK signals were observed in the sSJp-RNAis midgut, not only in cells residing on the basal side of the epithelium, but also in some ECs (Fig. 5B–D,F–H), strongly suggesting that the Ras-MAPK pathway is activated in ISCs, EBs and certain ECs. To monitor Jak-Stat pathway activity, we used a Stat92E reporter line driving expression of DGFP (10×STAT-DGFP). In the control midgut, few DGFP-positive cells were observed (Fig. 5I,M,Q), while DGFP-positive cells were markedly increased on the basal side of the epithelium throughout the sSJp-RNAis midgut (Fig. 5J–L,N–P,R–T). In addition, dpERK- and DGFP-positive signals were detected in Esg–LacZ-positive cells in the mesh-RNAi midgut (Fig. S4), indicating that the MAPK and Jak-Stat pathways were activated in the progenitor cells in the sSJ-disrupted midgut. Collectively, these results demonstrate that depletion of sSJ proteins from ECs results in activation of both the Ras-MAPK and Jak-Stat signaling pathways in the midgut.

In ISC proliferation and EB differentiation, the Jak-Stat signaling pathway is activated by cytokines known as Unpaired (Upd) ligands (Jiang et al., 2009; Osman et al., 2012; Zhou et al., 2013). Upd2 and Upd3 have been reported to contribute to increased ISC division in the midgut upon aging or exposure to stress (Osman et al., 2012). To examine whether Upd2 and Upd3 were involved in the increased ISC proliferation and abnormal accumulation of ECs observed in the sSJp-RNAis midgut, we suppressed mesh expression in the midgut of the upd2 and upd3 double-mutant (upd2,3^Δ) flies through expression of mesh-RNAi. In the mesh-RNAi upd2,3^Δ midgut, a large number of mitotic cells were still observed, at a similar level to the mesh-RNAi midgut (Fig. 6B,C,G), while only a few mitotic cells were detected in the control and upd2,3^Δ midgut (Fig. 6A,G). Meanwhile, the expansion of the midgut observed in mesh-RNAi flies was significantly reduced in mesh-RNAi upd2,3^Δ flies. At 3 days after RNAi induction, the diameter of the most posterior region of the midgut in mesh-RNAi upd2,3^Δ flies was significantly smaller than that in mesh-RNAi flies and was similar to the upd2,3^Δ flies and control flies (Fig. 6H). At 5 days after RNAi induction, the suppressive effect of upd2,3^Δ in mesh-RNAi flies in the midgut diameter appeared to be more substantial, but it may include the influence of upd2,3^Δ alone on the midgut because the diameter of the most posterior region of the midgut in upd2,3^Δ flies is smaller than that in control flies in this condition (Fig. 6D–F,H). Accumulation of ECs was still observed in mesh-RNAi upd2,3^Δ flies (Fig. 6F). The upd2,3^Δ flies expressing mesh-RNAi in ECs had a shortened lifespan, resembling that of mesh-RNAi flies (Fig. S5A–C). The midgut barrier dysfunction seen in mesh-RNAi flies was also observed in mesh-RNAi upd2,3^Δ flies (Fig. S5D), suggesting that Upd2 and Upd3 do not contribute to the loss of barrier function in the midgut. Taken together, our observations suggest that induction of Upd2 and/or Upd3 expression is involved in the aberrant behavior of ECs in the sSJp-RNAis midgut.

To further validate the effects of sSJ protein depletion on the adult midgut, we generated mitotic clones that lacked Tsp2A and simultaneously expressed GFP in the ISC lineage through the mosaic analysis with a repressible cell marker (MARCM) system (Lee and Luo, 2001). The clone size, indicated by the number of GFP-positive cells per clone, of Tsp2A-mutant clones was comparable to that of control clones (Fig. 7A,B,G). However, an increased number of PH3-positive cells outside the mutant clones was observed in Tsp2A-mutant clones compared with control clones (Fig. 7A,B,H). These results indicate that loss of Tsp2A in ECs has a non-cell-autonomous effect on ISC proliferation. In addition, immunostaining with anti-Pdm1 and anti-Prospero (Pros) antibodies, which label ECs and EEs, respectively (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006), revealed that Tsp2A-mutant clones contained differentiated ECs and EEs (Fig. 7C–F). These results suggest that loss of Tsp2A does not block differentiation of the ISC lineage.

In the Drosophila midgut epithelium, the paracellular barrier is mediated by specialized cell–cell junctions known as sSJs (Tepass and Hartenstein, 1994). Our previous studies revealed that three sSJ-associated membrane proteins, Ssk, Mesh and Tsp2A, are essential for the organization and function of sSJs (Yanagihashi et al., 2012; Izumi et al., 2012, 2016). In this study, we depleted the sSJ proteins from ECs in the Drosophila adult midgut and showed that they are also required for the barrier function in the adult midgut epithelium. Interestingly, the reduced expression of sSJ proteins in ECs led to remarkably shortened lifespan in adult flies, increased ISC proliferation and intestinal hypertrophy accompanied by accumulation of morphologically aberrant ECs in the midgut. The intestinal hypertrophy caused by mesh depletion was reduced by loss of upd2 and upd3 without profound suppression of ISC proliferation, without recovery of the shortened lifespan, and without recovery of the midgut barrier dysfunction. We also found that Tsp2A mutant clones promoted ISC proliferation in a non-cell-autonomous manner. Taken together, we propose that sSJs play a crucial role in maintaining tissue homeostasis through regulation of ISC proliferation and EC behavior in the Drosophila adult midgut. The adult Drosophila intestine provides a powerful model to investigate the molecular mechanisms behind the emergence and progression of intestinal metaplasia and dysplasia, which are associated with gastrointestinal carcinogenesis in mammals (Li and Jasper, 2016). Given that Drosophila intestinal dysplasia is associated with over-proliferation of ISCs and their abnormal differentiation, the intestinal hypertrophy observed in the present study should be categorized as a typical dysplasia in the Drosophila intestine. In this study, depletion of sSJ proteins from the ECs was mostly performed through RNAi, which may raise a concern about off-target effects. Nevertheless, it is safe to say that the midgut phenotypes were derived from specific effects of sSJ protein depletion because the essentially same phenotypes were observed in the RNAi lines for different sSJ proteins and additional RNAi lines for mesh and Tsp2A.

Based on our observations, we hypothesize the following scenario for the hypertrophy generation in the sSJ-protein-deficient midgut. First, depletion of sSJ proteins from ECs leads to disruption of sSJs in the midgut. Second, the impaired midgut barrier function caused by disruption of sSJs results in leakage of harmful substances from the intestinal lumen, thereby inducing the expression of cytokines and growth factors, such as Upd and EGF ligands, in the midgut. Alternatively, disruption of sSJs causes direct activation of a particular signaling pathway that induces expression of cytokines and growth factors by ECs. Third, proliferation of ISCs is promoted by activation of the Jak-Stat and Ras-MAPK pathways. Fourth, EBs produced by the asymmetric division of ISCs differentiate into ECs with impaired sSJs in response to cytokines such as Upd2 and/or Upd3. Consistent with this scenario, increased mRNA expression of upd3 in the ssk- and Tsp2A-deficient midgut has been reported very recently (Salazar et al., 2018; Xu et al., 2019). Finally, the ECs fail to integrate into the epithelial layer and thus become stratified in the midgut lumen to generate hypertrophy. Interestingly, loss of upd2 and upd3 reduced the intestinal hypertrophy caused by depletion of mesh, but not the increased ISC proliferation. These findings imply that Upd2 and/or Upd3 preferentially promote EB differentiation rather than ISC proliferation. Considering that Upd-Jak-Stat signaling is required for both ISC proliferation and EB differentiation (Jiang et al., 2009), Upd2 and/or Upd3 may predominantly promote EB differentiation and accumulation of ECs, while other cytokines such as Upd1 and/or EGF ligands may activate ISC proliferation in the sSJ-disrupted midgut. Meanwhile, Upd-Jak-Stat signaling-mediated ISC proliferation also occurs during experimental induction of apoptosis of ECs (Jiang et al., 2009). Given that apoptosis of ECs is thought to cause the disruption of epithelial integrity followed by midgut barrier dysfunction, it may induce ISC proliferation by the same mechanism as that observed in the sSJ-protein-deficient midgut. Thus, we cannot exclude a possibility that depletion of sSJ proteins initially causes apoptosis of ECs and thereby leads to increased ISC proliferation. This possibility needs to be examined in future studies. Interestingly, the ISC proliferation induced by the loss of sSJ proteins was non-cell-autonomous. Mechanistically, disruption of the intestinal barrier function caused by impaired sSJs may permit the leakage of particular substances from the midgut lumen, which would induce particular cells to secrete cytokines and growth factors, such as Upd and EGF ligands, and stimulate ISC proliferation. Alternatively, sSJs or sSJ-associated proteins may be directly involved in the secretion of cytokines and growth factors through the regulation of intracellular signaling in the ECs.

In this study, we observed abnormal morphology and aberrant F-actin and Dlg distributions in ssk-, mesh- and Tsp2A-RNAi ECs. Consistent with our results, Chen et al. (2018) recently reported that loss of mesh and Tsp2A in clones causes defects in polarization and integration of ECs in the adult midgut. In contrast, no remarkable defects in the organization and polarity of ECs were observed in the ssk-, mesh- and Tsp2A-mutant midgut in first-instar larvae (Yanagihashi et al., 2012; Izumi et al., 2012, 2016), suggesting that sSJ proteins are not required for establishment of the initial epithelial apical-basal polarity. This discrepancy may be explained by the marked difference between the larval and adult midguts: ECs in the larval midgut are postmitotic, while those in the adult midgut are capable of regeneration by the stem cell system (Lemaitre and Miguel-Aliaga, 2013). In the sSJ protein-deficient adult midgut, activated proliferation of ISCs generates excessive ECs. These ECs may lack sufficient cell–cell adhesion, because of impaired sSJs, fail to become integrated into the epithelial layer and detach from the basement membrane, leading to loss of normal polarity. Because sSJs seem to be the sole continuous intercellular contacts between adjacent epithelial cells in the midgut (Tepass and Hartenstein, 1994; Baumann, 2001), it is reasonable to speculate that sSJ-disrupted ECs have a reduced ability to adhere to other cells.

A recent study revealed that depletion of the tricellular junction protein Gliotactin from ECs leads to epithelial barrier dysfunction, increased ISC proliferation and blockade of differentiation in the midgut of young adult flies (Resnik-Docampo et al., 2017). In contrast to the findings after depletion of sSJ proteins presented here, the gliotactin-deficient flies did not appear to exhibit intestinal hypertrophy accompanied by accumulation of ECs throughout the midgut. Furthermore, the lifespan of gliotactin-deficient flies was found to be longer than that we find for sSJ-protein-deficient flies. The difference in phenotypes found between the Resnik-Docampo et al. (2017) study and our present study may reflect the difference in the degrees of sSJ deficiency – disruption of entire bicellular sSJs or tricellular sSJs only. Aging has also been reported to be correlated with barrier dysfunction, increased ISC proliferation and accumulation of aberrant cells in the adult midgut (Biteau et al., 2008; Rera et al., 2012). The hypertrophy formation in the sSJ-disrupted midgut accompanied by increased ISC proliferation and accumulation of aberrant ECs raise the possibility that disruption of sSJs is the primary cause of the alterations in the midgut epithelium with aging.

During the preparation of our manuscript, two groups published interesting phenotypes of the sSJ-protein-deficient adult midgut in Drosophila that are highly related to the present study. Salazar et al. (2018) reported that reduced expression of ssk in ECs leads to gut barrier dysfunction, altered gut morphology, increased stem cell proliferation, dysbiosis and reduced lifespan. They also showed that upregulation of Ssk in the midgut protects flies against microbial translocation, limits dysbiosis and prolongs lifespan. Meanwhile, Xu et al. (2019) reported that depletion of Tsp2A from ISCs and EBs causes accumulation of ISCs and EBs and a swollen midgut with multilayered epithelium, similar to our observations. They also showed that knockdown of ssk and mesh in ISCs and EBs results in accumulation of ISCs and EBs. Importantly, they demonstrated that Tsp2A depletion from ISCs and EBs causes excessive aPKC-Yki-JAK-Stat activity and leads to increased stem cell proliferation in the midgut. They further showed that Tsp2A is involved in endocytic degradation of aPKC, which antagonizes the Hippo pathway. Their results strongly suggest that sSJs are directly involved in the regulation of intracellular signaling for ISC proliferation. In their study, Tsp2A knockdown in ISCs and EBs caused no defects in the midgut barrier function, in contrast to what we show in the present study. This discrepancy may be due to differences in the GAL4 drivers used in each study or the conditions for the barrier integrity assay (Smurf assay). In addition, Xu et al. (2019) mentioned that MARCM clones generated from ISCs expressing Tsp2A-RNAi grow much larger than control clones, while we found no remarkable size difference between Tsp2A-mutant clones and control clones. Such discrepancies need to be reconciled by future investigations. To further clarify the mechanistic details for the role of sSJs in stem cell proliferation, it will be interesting to analyze the effects of sSJ protein depletion on the behavior of adult Malpighian tubules, which also have sSJs and a stem cell system (Singh et al., 2007).

In this study, we have demonstrated that sSJs play a crucial role in maintaining tissue homeostasis through regulation of ISC proliferation and EC behavior in the Drosophila adult midgut. Our sequential identification of the sSJ proteins Ssk, Mesh and Tsp2A has provided a Drosophila model system that can be used to elucidate the roles of the intestinal barrier function by experimental dysfunction of sSJs in the midgut. However, as described in this study, simple depletion of sSJ proteins throughout the adult midgut causes phenotypes that are too drastic, involving not only disruption of the intestinal barrier function but also intestinal dysplasia and subsequent lethality. To investigate the systemic effects of intestinal barrier impairment throughout the life course of Drosophila, more modest depletions of sSJ proteins are needed for future studies.

Fly stocks were reared on a standard cornmeal fly medium at 25°C. w¹¹¹⁸ flies were used as wild-type flies unless otherwise specified. The other fly stocks used were: y w; Myo1A-GAL4 [#112001; Drosophila Genetic Resource Center (DGRC), Kyoto, Japan], tubP-GAL80^ts [#7019; Bloomington Drosophila Stock Center (BDSC), Bloomington, IN], y w; CD8-GFP (#108068; DGRC), y w; Pin^Yt/CyO; UAS-mCD8-GFP (#5130; BDSC), w; 10xStat92E-DGFP/CyO (#26199; BDSC), w; 10xStat92E-DGFP/TM6C Sb Tb (#26200; BDSC), y w; esg-lacZ/CyO (#108851; DGRC) and FRT19A; ry (#106464; DGRC).

The RNAi lines used were: ssk-RNAi (Yanagihashi et al., 2012), mesh-RNAi [#12074R-1, 12074R-2; Fly Stocks of National Institute of Genetics (NIG-Fly), Mishima, Japan] (Izumi et al., 2012), Tsp2A-RNAi (#11415R-2; NIG-Fly Tsp2AIR1-2) (Izumi et al., 2016) and w-RNAi (#28980; BDSC).

The mutant stocks used were: Tsp2A^1-2 (Izumi et al., 2016) and w upd2^Delta upd3^Delta (#55729; BDSC).

The following stocks were used to generate positively marked MARCM clones: tub^P-GAL80 w FRT19A; Act5C-GAL4, UAS-GFP/CyO (#42726; BDSC) and FRT19A tub^P-GAL80 hsFLP w; UAS-mCD8GFP (#108065; DGRC).

Flies were crossed and grown at 18°C until eclosion. Adult flies at 2–5 days after eclosion were collected and transferred to 29°C for inactivation of GAL80. All analyses for these experiments were performed on female flies, because their age-related gut pathology is well established (Lemaitre and Miguel-Aliaga, 2013).

Flies were crossed at 18°C. At 2–5 days after eclosion, adult flies were heat-shocked at 37°C for 1 h twice daily. Adult flies were transferred to fresh vials every 2–3 days and maintained at 25°C for 10 days after clone induction before being dissected.

Adult flies were dissected in Hanks' balanced salt solution and the midgut was fixed with 4% paraformaldehyde in PBS with 0.2% Tween-20 for 30 min. The fixed specimens were washed three times with PBS with 0.4% Triton X-100 and blocked with 5% skim milk in PBS with 0.2% Tween-20. Thereafter, the samples were incubated with primary antibodies at 4°C overnight, washed three times with PBS with 0.2% Tween-20 and incubated with secondary antibodies for 3 h. After another three washes, the samples were mounted in Fluoro-KEEPER (12593-64; Nakalai Tesque, Kyoto, Japan). Images were acquired with a confocal microscope (Model TCSSPE; Leica Microsystems, Wetzlar, Germany) using its accompanying software and HC PLAN Apochromat 20× NA 0.7 and HCX PL objective lenses (Leica Microsystems). Images were processed with Adobe Photoshop^® software (Adobe Systems Incorporated, San Jose, CA).

The following primary antibodies were used: rabbit anti-Mesh (955-1; 1:1000) (Izumi et al., 2012); rabbit anti-Tsp2A (302AP; 1:200) (Izumi et al., 2016); rabbit anti-Ssk (6981-1; 1:1000) (Yanagihashi et al., 2012); mouse anti-Dlg [4F3; Developmental Studies Hybridoma Bank (DSHB), Iowa City, IA; 1:50]; mouse anti-Delta (C594.9B; DSHB; 1:20); mouse anti-Pros (MR1A; DSHB; 1:20); rabbit anti-Pdm1 (a gift from Dr Yu Cai, TEMASEK, Singapore; 1:500; Yeo et al., 1995); rabbit anti-PH3 (06-570; Millipore, Darmstadt, Germany; 1:1000); rabbit anti-dpERK (4370S, Cell Signaling, Danvers, MA; 1:500); rat anti-GFP (GF090R; Nakalai Tesque; 1:1000); rabbit anti-GFP (598; MBL, Nagoya, Japan; 1:1000) and mouse anti-β-galactosidase (Z3781; Promega, Madison, WI; 1:200) antibodies. Alexa Fluor 488-conjugated (A21206; Thermo Fisher Scientific, Waltham, MA) and Cy3- and Cy5-conjugated (Jackson ImmunoResearch Laboratories, West Grove, PA) secondary antibodies were used at 1:400 dilution. Actin was stained with Alexa Fluor 568 phalloidin (A12380; Thermo Fisher Scientific; 1:1000) or Alexa Fluor 647 phalloidin (A22287; Thermo Fisher Scientific; 1:1000). Nuclei were stained with propidium iodide (Nakalai Tesque; 0.1 mg ml⁻¹).

Adult control, ssk-, mesh- and Tsp2A-RNAi flies at 5 days after transgene induction were dissected and fixed overnight at 4°C with a mixture of 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M cacodylate buffer (pH 7.4). The specimens including the midgut were prepared as described previously (Izumi et al., 2012). Ultrathin sections (50–100 nm) were double-stained with 4% hafnium (IV) chloride and lead citrate and observed with a JEM-1011 electron microscope (JEOL, Tokyo, Japan) at an accelerating voltage of 80 kV. For Toluidine Blue staining, sections (0.5–1 µm) mounted on glass slides were placed in 0.05% Toluidine Blue in distilled water for 2–30 min, rinsed in water for 1 min and allowed to air dry. The stained sections were observed with an optical microscope (BX41; Olympus, Tokyo, Japan).

For the barrier integrity assay (Smurf assay), flies at 2–5 days of age were placed in empty vials containing a piece of paper soaked in 1% (wt/vol) Blue Dye No. 1 (Tokyo Chemical Industry, Tokyo, Japan)/5% sucrose solution at 50–60 flies/vial. After 2 days at 18°C, the flies were placed in new vials containing paper soaked in BlueDye/sucrose and transferred to 29°C. Loss of midgut barrier function was determined when dye was observed outside the gut (Rera et al., 2011, 2012). Flies were transferred to new vials every 2 days.

Statistical significance was evaluated by means of the Mann–Whitney U-test, a Student's t-test, one-way ANOVA with Tukey's multiple comparisons test (KaleidaGraph software; Synergy Software, Reading, PA) and the Fisher's exact test. Values of P<0.05 were considered significant.

We are grateful to all members of the Furuse laboratories and Kazutaka Akagi (NCGG, Japan) for helpful discussions. We thank the Bloomington Drosophila Stock Center, Drosophila Genetic Resource Center at Kyoto Institute of Technology and Fly Stocks of National Institute of Genetics (NIG-Fly) for fly stocks. We also thank Alison Sherwin, PhD, from Edanz Group (www.edanzediting.com/ac) for editing a draft of this manuscript.