A subset of neurons controls the permeability of the peritrophic matrix and midgut structure in Drosophila adults

Maintenance of the proper structure and function of the gastrointestinal tract is central to host homeostasis in metazoan animals. Aside from its main role in digestion and nutrient absorption, the gut must also protect an animal from harmful substances and microorganisms, leading to a strong immune system and robust physical/structural barriers against invaders (Sansonetti, 2004). The intestinal tract senses external cues, such as a nutrient availability, via the enteric endocrine or nervous system and sends systemic signals though hormonal or neuronal means to change both metabolism and behavior (Furness and Costa, 1987). These functions of the intestinal tract are also consistent for most insects, including the model organism Drosophila melanogaster (Kuraishi et al., 2013; Lemaitre and Miguel-Aliaga, 2013).

Complex and highly organized tissue structures ensure the realization of these important tasks of the gut. Compartmentalization – the sequential organization of regions that vary histologically and functionally – is an important feature of the intestinal tract (Karasov et al., 2011). In Drosophila adults, the gut is divided into three distinct domains based on developmental origin: the foregut, midgut and hindgut. The midgut, which is the main region responsible for intestinal functions, comprises a single layer of epithelium, surrounded by visceral muscles, nerves and tracheae, and is subdivided into six major anatomic regions with distinct functions (Buchon et al., 2013b).

The peritrophic matrix and septate junctions between epithelial cells have a central role as a physical barrier against external invaders (Hegedus et al., 2009; Tepass et al., 2001). The peritrophic matrix is an acellular structure that forms a layer of chitin polymers and glycoproteins, such as peritrophins, lining the insect midgut lumen (Lehane, 1997). The peritrophic matrix seems to be formed by either the midgut epithelium (type I) or the proventriculus (type II), which is a specialized structure located at the foregut/midgut junction that regulates passage of food to the midgut. In type I peritrophic matrix, delamination of successive concentric lamellae occurs along the length of the midgut. Diptera such as Drosophila have a type II peritrophic matrix that is continuously produced by specific cells of the proventriculus (King, 1988). The protective role of the peritrophic matrix against abrasive food particles and pathogens, as well as in sequestering ingested toxins, has been studied in many insects (Edwards and Jacobs-Lorena, 2000; Wang and Granados, 2000). In Drosophila adults, mutation of the drosocrystallin (dcy) gene, which encodes a structural component of the peritrophic matrix, results in its reduced thickness and higher permeability (Kuraishi et al., 2011). The dcy mutant flies show greater susceptibility to ingested entomopathogenic bacteria or pore-forming toxins. Septate junctions are functionally related to mammalian tight junctions and participate in epithelial barrier function, protecting the fly from oral infection by pathogenic bacteria. Bonnay et al. demonstrated that the big bang gene (bbg) encodes a PDZ domain-containing protein that is expressed in septate junctions (Bonnay et al., 2013). A mutation in bbg results in the loosening of septate junctions, and is associated with acute susceptibility to invasive enteric pathogens such as Pseudomonas aeruginosa and Serratia marcescens. The compartmentalization, peritrophic matrix and septate junctions of the gut are maintained throughout adult life by rapid turnover of the epithelium within 1 to 2 weeks under steady-state conditions (Buchon et al., 2013a,b). However, the cellular and molecular processes required to maintain these cellular and acellular structures of the intestinal tract are poorly understood.

A major function of the stomatogastric nervous system is to control peristalsis of the muscles surrounding the intestinal tract (Huizinga and Lammers, 2009) and to sense external conditions to regulate metabolism. A recent study in Drosophila adults revealed that enteric neurons also govern fluid homeostasis and sex peptide-induced changes in intestinal physiology, indicating an indispensable role for the brain–gut axis in maintaining host homeostasis (Cognigni et al., 2011; Talsma et al., 2012). We hypothesized that the enteric nervous system also has a role in maintaining the structural integrity of the gut, which is important for its barrier function. In this study, we identified a subset of neurons required for maintaining gut impermeability against enteric pathogens, providing evidence for the neural control of gut integrity in Drosophila adults.

Oregon R flies were used as wild-type flies. The GAL4 lines screened in this study (Table S1) were obtained from the Bloomington Stock Center (Indiana University, Bloomington, IN, USA), Drosophila Genetic Resource Center (Kyoto Institute of Technology, Japan) and 384-GAL4, Bx-GAL4, elav-GAL4, tubP-GAL80^ts, L¹/CyO; UAS-DenMark, syt.eGFP and UAS-mCD8::GFP/CyO were from Bloomington Stock Center. UAS-dTrpA1 was a gift from P. Garrity (Hamada et al., 2008). elav-GAL80 was a gift from Y. Jan (Yang et al., 2009). UAS-lacZ (Bloomington Drosophila Stock Center) or w¹¹¹⁸ was used as a control. Drosophila stocks and crosses were maintained at 18°C or 25°C in tubes containing standard cornmeal-agar medium. To inhibit or activate neural activity, 5- to 9-day-old flies of NP3253-GAL4/tubP-GAL80^ts; UAS-Kir2.1-EGFP (Baines et al., 2001)/+ (NP3253>Kir2.1 flies) or NP3253-GAL4/UAS-dTrpA1 were maintained at 30°C for 2 days prior to use in all experiments. The NP3253 line was subjected to standard mitotic recombination over yw chromosomes to eliminate possible second-site mutations.

The Ecc15-GFP strain was described previously (Basset et al., 2000) and was grown in Luria Bertani broth for all experiments. Flies were grown at 29–30°C and allowed to reach the stationary phase. Cells were then concentrated at OD₆₀₀=200 with 2.5% sucrose solution. For oral infection, flies were starved for 2 h at 30°C and then placed in a fly vial with food solution. The food solution was made by mixing a pellet of bacteria, added to a filter disk that completely covered the surface of standard fly medium. Flies were maintained at 30°C and survival was monitored at different time points.

Drosophila adults were dissected into cold phosphate-buffered saline (PBS) and the guts or brains were immediately fixed with 4% paraformaldehyde in PBS for 30 min at room temperature. The samples were rinsed in 0.5% Triton X-100 in PBS and then incubated with primary antibodies: mouse anti-GFP (1:500, Invitrogen, cat. no. A-6455), rabbit anti-RFP (1:500, Clontech, cat. no. 632496), rabbit anti-PH3 (1:500, Cell Signaling, cat. no. 9701), mouse anti-Discs-large 4F3 (1:100, Developmental Studies Hybridoma Bank), Alexa Fluor 647-conjugated goat anti-HRP antibody (1:500, Jackson ImmunoResearch, cat. no. 123-605-021) in 0.5% Triton X-100 in PBS at 4°C overnight. The samples were washed twice with 0.2% Triton X-100 in PBS and primary antibodies were labeled with Alexa Fluor 488-, 546- or 647-coupled secondary antibodies (Invitrogen). Actin filaments were stained with Rhodamine-Phalloidin (1:100, Sigma) and nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; 3 μg ml⁻¹, Dojindo, Japan). The samples were then washed with 0.2% Triton X-100 in PBS, incubated with 50% glycerol (Wako, Japan) in PBS, and mounted in 80% glycerol in PBS or in Vectashield. For anti-PH3 antibody staining, the samples were fixed with 3.7% formaldehyde in PBS, permeabilized with 99.5% pre-chilled ethanol at −30°C for 5 min. The samples were visualized with a Leica TCS-SPE confocal microscope, and images were reconstructed using Photoshop (Adobe) and ImageJ (NIH).

PH3-positive cells in the whole midgut of 10 to 12 female flies were counted under a confocal microscope. Confocal images that showed the maximum measured area were calculated by ImageJ in the luminal region of the proventriculus or in the anterior midgut (from the top of the proventriculus to the 200 µm point).

Flies were starved for 2 h at 30°C, fed with FITC-labeled beads (50 nm diameter, Polysciences, Warrington, PA, USA) to monitor the permeability of the peritrophic matrix, as described previously (Kuraishi et al., 2011). Images were captured with a Zeiss conventional fluorescence microscope or a Leica confocal microscope with a 1.5 AU pinhole. For quantification, the guts were dissected out 10 min after feeding and observed under a conventional fluorescence microscope using 20 to 30 female flies.

Five pairs of salivary glands were dissected out from adult flies in 50 µl of PBS, homogenized with a pestle and mortar in 350 µl of PBS and 50 µl of PBS containing 5% Triton X-100. Samples were then incubated for 10 min at room temperature, and diluted 100× with PBS. Fifty microliters of diluted sample was then mixed with 10× diluted β-glo reagent (Promega), incubated for 30 min and emission at 570 nm measured using a luminometer. Assays were performed on triplicate samples.

Assays were performed largely based using a published method (Cognigni et al., 2011). For quantification of feeding, three female flies were starved for 3 h at 30°C, fed with 0.5% Bromophenol Blue (BPB) sodium salt/cornmeal-agar for 1 or 2 h, and then each fly was placed into 50 µl MilliQ water, homogenized with a pestle and centrifuged twice (16,000 g, 2 min) to remove debris. Absorption at 594 nm was measured using a NANODROP2000 (Thermo Scientific, Waltham, MA). For quantification of the excretion rate, five female flies were starved for 3 h at 30°C, fed with 0.5% BPB sodium salt/cornmeal-agar for 1 h at 30°C, moved to a new vial containing normal food and maintained there for several hours. Each fly was placed into 80 µl MilliQ water, homogenized with pestle and centrifuged twice to remove debris. Absorption at 594 nm was measured using a Nanodrop2000 spectrophotometer.

All flies were raised at 18°C for 5 to 6 days after eclosion. Three vials (each containing 30 flies) were moved to 30°C. After 2 days, lifespan analysis was started (this day was set to day 0) at 30°C. Live flies were counted every day and transferred to new vials every 2 days.

Statistical analyses were performed using Student's t-test or the log-rank test, and P values less than 0.05 were considered significant.

To identify neurons important for gut integrity, we screened GAL4 enhancer trap lines with Kir2.1, a mammalian inwardly rectifying K⁺ channel, to block neural activity (Baines et al., 2001), and examined their susceptibility to oral bacterial challenge as a measure of gut integrity. Sensitivity to bacterial infection is a complex phenomenon (Ayres and Schneider, 2012; Lemaitre and Hoffmann, 2007) because not only resistance mechanisms, such as the expression of antimicrobial peptides, but also tolerance mechanisms, such as permeability of the epithelial barrier, feeding behavior, excretion of ingested materials and damage repair after infection, are required for normal survival upon infection (Buchon et al., 2013a; Kuraishi et al., 2013). Therefore, if some lines are susceptible to oral infection, the underlying mutations are expected to be involved in some aspect of gut function, including structural integrity.

We selected 350 GAL4 enhancer trap lines (Table S1) known to induce expression in neurons based on the FLYBRAIN and Flytrap databases (Kelso et al., 2004; Shinomiya et al., 2011). Kir2.1 expression was repressed by temperature-sensitive GAL80 (GAL80^ts) until adulthood and then induced by shifting the flies to a restrictive temperature for 2 days (Fig. 1A). Several enhancer trap lines that expressed Kir2.1 were susceptible to Ecc15 oral infection (Fig. 1B). Of those, a fly line expressing Kir2.1 using NP3253-GAL4, designated NP3253>Kir2.1, exhibited strong susceptibility to Ecc15 oral infection, but not to normal fly foods (Fig. 1C,D). In contrast, flies with hyperactive NP3253-positive cells upon expression of dTrpA1 (Rosenzweig et al., 2005) were not sensitive to Ecc15 oral infection (Fig. 1E). To exclude the possibility that these phenotypes resulted from the genetic background, the NP3253 line was backcrossed with the yw strain and Kir2.1-induced susceptibility was tested upon oral infection with Ecc15. The findings demonstrated that these flies are also susceptible to infection (Fig. 1C,D), indicating that the activity of NP3253-positive cells is specifically required for gut defense upon bacterial infection.

We next examined expression patterns in the tissues of adult flies to evaluate whether NP3253-GAL4 could induce expression in enteric neurons. Many green fluorescent protein (GFP)-positive cells driven by NP3253-GAL4 were detected in the brain, proventriculus and anterior midgut, as well as in the posterior midgut (Fig. 2A,B), salivary glands, trachea, and reproductive organs (data not shown). A previous study (Tanaka et al., 2008) reported that NP3253 labels neurons in the mushroom body. To analyze whether the NP3253-positive cells in the gut are neurons, they were stained with horseradish peroxidase (HRP), a neural marker protein, together with anti-GFP. The GFP-positive cells driven by NP3253-GAL4 in the proventriculus and anterior midgut were HRP positive, whereas those in posterior midgut were not (Fig. 2C). NP3253-positive cells in the proventriculus and anterior midgut were positive for the synaptic vesicle marker Syt.eGFP and the dendrite marker DenMark (Fig. 2D,E). These findings suggest that NP3253-positive cells in the proventriculus and anterior midgut are neurons, and indicate that not all NP3253-positive cells are neurons. This led us to examine whether NP3253-positive neurons are responsible for the survival phenotype upon Ecc15 oral infection. We analyzed NP3253>Kir2.1 flies in combination with elav-GAL80 to inhibit GAL4 activity in all neurons (Rideout et al., 2010). Survival analysis revealed that susceptibility to Ecc15 oral infection was partially rescued by co-expression with elav-GAL80 (Fig. 2F). Both NP3253-GAL4 and elav-GAL80 drive gene expression in the salivary gland (Fig. 2G); therefore, to rule out the possibility that the salivary gland is responsible for the survival phenotype, we used Bx-GAL4 and 384-GAL4 to drive Kir2.1. Both drivers induced reporter expression in the salivary gland as strong as NP3253-GAL4 (Fig. 2G,H), but neither Bx>Kir2.1 flies nor 384>Kir2.1 flies were susceptible to oral infection with Ecc15 (Fig. 2I,J). Together, these results suggest that a subset of neurons driven by NP3253-GAL4 is partly involved in the survival phenotype.

Next, we examined why the NP3253>Kir2.1 flies exhibit sensitivity to bacterial oral infection. After feeding the flies GFP-labeled Ecc15 (Ecc15-GFP), GFP signals were observed in the whole body of NP3253>Kir2.1 flies, in contrast to wild-type flies, which expressed the GFP signal only in the abdomen (Fig. 3A). GFP signals were observed throughout the whole body in ∼10% of the NP3253>Kir2.1 flies at 6 h after Ecc15-GFP feeding and in up to 20% at 24 h (Fig. 3B). This observation indicated that the bacteria intruded into the hemolymph of NP3253>Kir2.1 flies, suggesting that gut barrier function was compromised in these flies. The peritrophic matrix is an acellular layer that protects the gut epithelium and its permeability can be assessed by feeding adults fluorescein isothiocyanate (FITC)-labeled beads (Kuraishi et al., 2011). Conventional fluorescence microscopy revealed that the 50 nm FITC-labeled beads remained in the lumen of wild-type flies after feeding (Fig. 3C,D). In contrast, FITC signals were diffuse in the gut of NP3253>Kir2.1 flies (Fig. 3C,D). Close examination using a confocal microscope with the focal plane on the epithelial cells (Fig. 3E) revealed FITC signals outside the peritrophic matrix in the NP3253>Kir2.1 flies (Fig. 3F). Consistent with these observations, staining with the mitotic marker PH3 revealed that upd3-dependent stem cell proliferation, an indicator of gut damage, was increased in the midgut of NP3253>Kir2.1 flies (Fig. 3G,H). Furthermore, NP3253>Kir2.1 flies had a shorter lifespan, and began to die 1 to 2 weeks after emergence (Fig. 3I). Indeed, Rera et al. (2012) reported that increased gut permeability is a cause and predictor of imminent death. These findings indicate that the peritrophic matrix of NP3253>Kir2.1 flies is more permeable or damaged, providing an explanation for the susceptibility of NP3253>Kir2.1 flies to oral infection.

We then performed histological analysis of the gut of NP3253>Kir2.1 flies. The proventriculus, the organ responsible for secretion of the peritrophic matrix in Drosophila adults, was stained with Phalloidin and DAPI to visualize the actin filaments and nuclei, respectively. As shown in Fig. 4A,B, the proventriculus morphology in NP3253>Kir2.1 flies differed from that in wild-type flies: the bulge formed by the inner cells (indicated by arrowheads) was lost in NP3253>Kir2.1 flies, whereas the top of the inner part of the proventriculus was expanded (indicated by the arrows). This observation was supported by visualizing the tissue structure following staining with the marker for cell junctions, discs large (Dlg; Fig. 4C). These findings indicated that a part of the proventriculus of NP3253>Kir2.1 flies was flattened (Fig. 4D). We also observed a morphological abnormality of the midgut of NP3253>Kir2.1 flies. The diameter of the anterior part of the midgut, especially the R1 region of the midgut (Buchon et al., 2013b), was increased without a significant change in the number and shape of epithelial cells (Fig. 4E,F). This phenotype was also observed in starved NP3253>Kir2.1 flies. Notably, the increased diameter of the anterior midgut was also observed in the upd3 mutant background (Fig. 4G), suggesting that the increased diameter is not due to damage-induced stem cell proliferation. We further examined the feeding and excretion of the NP3253>Kir2.1 flies. As shown in Fig. 5, neither the feeding nor the excretion rate of NP3253>Kir2.1 flies, quantified by the amount of BPB food dye that flies ate, was compromised. These results suggest that the increased diameter was not due to defective excretion of the foods they had eaten, but rather to the homeostatic dysfunction of the NP3253>Kir2.1 flies to maintain normal gut morphology.

The stomatogastric nervous system controls peristalsis, fluid homeostasis and sex peptide-induced changes in intestinal physiology in adult Drosophila. Here, we describe a role of the nervous system in maintaining the impermeable gut physical barrier and organized epithelial structure of the anterior midgut. Several questions remain, however, as discussed below.

We demonstrated that the peritrophic matrix of NP3253>Kir2.1 flies is more permeable than that of wild-type flies. This phenotype is much stronger than that of dcy¹ mutant flies. The peritrophic matrix of dcy¹ mutant flies is not permeable to FITC-labeled beads with a size >70 kDa (Kuraishi et al., 2011). The peritrophic matrix of the NP3253>Kir2.1 flies, however, was permeable not only to latex beads, but also to bacteria, implying that the nature of the peritrophic matrix defects of NP3253>Kir2.1 flies differs from that of the dcy¹ mutant.

What is the defect that occurs in the epithelial barrier? We observed that the epithelial structure of the proventriculus and anterior midgut was disorganized and expanded in NP3253>Kir2.1 flies. We speculate that ingested bacteria augment the epithelial expansion and might affect the septate junctions between epithelial cells, resulting in a leaky epithelial barrier in the flies. This possibility should be examined in future studies.

NP3253-GAL4 drives expression in neuronal subsets in the brain and the anterior midgut in Drosophila adults. It is unclear which NP3253-positive neurons are involved in the observed phenotype and whether efferent or sensory neurons are responsible. We cannot rule out the possibility that NP3253-positive neurons only in the brain and not enteric neurons, are responsible for the observed phenotype. Further screening of enhancer trap lines is needed to identify drivers that have a similar phenotype as NP3253-GAL4.

Our study does not address the mechanisms of the NP3253-positive neurons that maintain the impermeability of the epithelial barrier and morphology. A possible mechanism by which NP3253-positive neurons control gut integrity is endocrine/paracrine regulation. Gut patterning is primarily achieved through interactions between the pan-midgut and region-specific transcription factors, together with spatial activities of morphogens (Buchon et al., 2013b). It is thus possible that secreted factors from NP3253-positive neurons affect morphogen expression or the activities of transcriptional factors in the anterior midgut. Although we showed that peristalsis is not severely compromised in the NP3253>Kir2.1 flies, another possibility is that NP3253-positive neurons control the pumping action of the proventriculus. The secreted components of the peritrophic matrix from the cells of the proventriculus are squeezed to form the peritrophic matrix sleeve and conveyed throughout the midgut by pumping of the proventriculus (Lehane, 1997). Therefore, if the activity of NP3253-positive neurons is inhibited, the peritrophic matrix does not form correctly or smoothly, and thus may accumulate around the anterior midgut, leading to an enlarged and abnormal structure with compromised permeability of the gut barriers. Because constitutive activation of NP3253-positive cells does not induce susceptibility to Ecc15 oral infection (Fig. 1D), the latter explanation is more plausible.

We are grateful to Dr B. Lemaitre for Ecc15-GFP. We thank Drs P. Garrity and Y. Jan at the Bloomington Stock Center (NIH P40OD018537), the Drosophila Genomics Resource Center at Indiana University, the Drosophila Genetic Resource Center at the Kyoto Institute of Technology, the Drosophila RNAi Screening Center, the Genetic Strain Research Center of National Institute of Genetics, and the Vienna Drosophila RNAi Center for fly stocks. We also thank the Developmental Studies Hybridoma Bank for antibodies; and FlyBase, FLYBRAIN, Flytrap and FlyAtlas for use of the databases.