Genome-Wide RNAi Screen Identifies Genes Involved in Intestinal Pathogenic Bacterial Infection

See allHide authors and affiliations

Science  17 Jul 2009:
Vol. 325, Issue 5938, pp. 340-343
DOI: 10.1126/science.1173164


Innate immunity represents the first line of defense in animals. We report a genome-wide in vivo Drosophila RNA interference screen to uncover genes involved in susceptibility or resistance to intestinal infection with the bacterium Serratia marcescens. We first employed whole-organism gene suppression, followed by tissue-specific silencing in gut epithelium or hemocytes to identify several hundred genes involved in intestinal antibacterial immunity. Among the pathways identified, we showed that the JAK-STAT signaling pathway controls host defense in the gut by regulating stem cell proliferation and thus epithelial cell homeostasis. Therefore, we revealed multiple genes involved in antibacterial defense and the regulation of innate immunity.

Drosophila melanogaster provides a powerful model that allows the dissection of the innate immune response at the organism level. In Drosophila, innate immunity has a humoral and a cellular immune response. The majority of our knowledge of Drosophila immunity is based on injection of nonpathogenic bacteria (13); however, this bypasses the initial steps of naturally occurring infections—namely, the physical barriers and the local, mucosal immune response. Intestinal immunity is currently the focus of intense research (4). In contrast to the human digestive tract, Drosophila lacks mammalian-like adaptive immunity and so relies entirely upon an innate immune system for protection against invading pathogens.

The intestinal infection model using pathogenic Serratia marcescens allows for the detailed analysis of local intestinal immunity and phagocytosis (5). S. marcescens is a gram-negative, opportunistic pathogen that can infect a range of hosts including Drosophila, Caenorhabditis elegans, and mammals (6, 7). Using ubiquitous RNA interference (RNAi)–mediated suppression, we performed an inducible genome-wide in vivo screen in Drosophila for novel innate immune regulators after S. marcescens infection (8) [Fig. 1A, fig. S1, A and B, and supporting online material (SOM) text]. To confirm our experimental approach we assayed various members of the Immune deficiency (IMD) and Toll pathways, the two major fly immune signaling cascades (Fig. 1B) (13). RNAi lines targeting several IMD members resulted in significantly reduced survival on infection with S. marcescens, whereas suppression of Toll pathway components had a less dramatic effect, which supports previous reports that the immune response to S. marcescens is IMD-dependent and Toll-independent (Fig. 1B) (5). Notably, not all members of the IMD pathway, such as imd, rel, and ird5, were picked up by our screening criteria, most likely because of inefficient RNAi silencing (Fig. 1B) (9).

Fig. 1

Analysis of genome-wide in vivo RNAi screen. (A) Total data of all RNAi lines screened for survival after S. marcescens infections. Data were analyzed as the time, in days, when 50% of the total number of flies had died. All data were normalized to the daily median time-to-death (LT50) mean of an experimental cohort. In all experiments, the cohort ranged from 80 to 200 lines. Hits were defined by susceptible (red dashed line) and resistant (blue dashed line) cut-offs, i.e., 1.5 SD below the mean and 2 SD above the mean, respectively, based on the pilot screen and controls. (B) Effect of RNAi knockdown of IMD and Toll pathway components on their survival against S. marcescens infection. SCOREs are shown for each line as described in (8). The dashed lines indicate the cut-offs used for resistance (+2 SD) and susceptibility (–1.5 SD) candidates. (C) Percentage distribution of GO annotated genes to biological processes for susceptible candidates.

We assayed 13,053 RNAi lines (9) representing 10,689 different genes (78% of the genome) against intestinal infection with S. marcescens (fig. S2A and tables S1 and S2). Of these, 8.3% (885 genes) were defined as hits, the majority of which (89.3%; 790 genes) were susceptible candidates (fig. S2A and table S3). On the basis of gene ontology (GO) annotations, susceptible candidates were classified according to their predicted biological processes. Genes involved in signaling, intracellular protein transport, and transcriptional regulation were overly represented among the entire data set (Fig. 1C). We also found marked enrichment for genes that regulate phagocytosis, defense responses, vesicle trafficking, and proteolysis. Several candidate RNAi lines represented genes that have been previously implicated in mounting an effective immune response (1019) (table S3).

Our approach also allowed us to identify negative regulators of Drosophila host defense (Fig. 1A). We identified 95 genes (10.7% of the total hits) that confer resistance to S. marcescens infections when silenced (fig. S2, A and B, table S4), none of which had previously been characterized as negative regulators of innate immunity. Thus, our genome-wide screen revealed previously known genes associated with Drosophila immunity and more than 800 additional candidate genes implicated in innate immunity, 40% of which had unknown function.

We retested some of our susceptible and resistant RNAi hits in the gut epithelium and the macrophage-like hemocytes, the two major cell types associated with our infection model, using cell type–specific driver lines, NP1-GAL4 and HML-GAL4, respectively (5, 20). We prioritized genes of interest by selecting the primary hits that have mammalian (mouse and/or human) orthologs. Of the 358 susceptible hits tested with the HML-GAL4 driver, RNAi against 98 genes (27%) resulted in significantly reduced survival as compared with RNAi controls, which indicated that these genes function in hemocytes to combat intestinal S. marcescens infections (Fig. 2A, fig. S3A, and table S5). When we used the NP1-GAL4 driver (fig. S4) to test 337 genes, RNAi against 129 genes (38%) resulted in significantly reduced survival, which suggested that these genes play an important role in host intestinal defense (Fig. 2B, fig. S3B, and table S6). Of the resistance hits, 37 HML-GAL4 RNAi candidates (79%) and 28 NP1-GAL4 RNAi candidates (61%) exhibited markedly enhanced survival (Fig. 2, C and D, fig. S3, and tables S7 to S9). Of the candidate genes, 79 functioned in both hemocytes and gut (fig. S3). Multiple susceptibility and resistance genes were tested 3 to 15 independent times, using ≥2 RNAi transformants to exclude position effects and second independent RNAi hairpins to confirm the target gene when available (Fig. 2, A to D, fig. S3, and tables S5 to S8). To exclude a potential developmental phenotype, we have tested most candidate lines by feeding flies on a sugar diet in the absence of bacteria (table S9). Thus, we have identified multiple regulators in hemocytes and/or gut epithelium that confer susceptibility or resistance to S. marcescens infections.

Fig. 2

Mapping and validation of conserved hits in the gut and hemocytes. (A and B) Survival graphs showing susceptible hits, (A) HML-susceptible genes and (B) NP1-susceptible genes, tested 3 to 15 times with several transformants and hairpins in hemocytes and gut epithelium, respectively. The kenny mutant line (key Mutant) is shown as a positive control. Means ± SEM, n ≥ 3 experiments with 20 flies in each. *P < 0.05 (Welch t test). (C and D) Survival graphs showing resistant hits, (C) HML-resistant genes and (D) NP1-resistant genes, tested 3 to 15 times with several transformants and hairpins in hemocytes and gut epithelium, respectively. Means ± SEM, n ≥ 3 experiments with 20 flies in each. *P < 0.05 (Welch t test). (E) Statistically enriched biological processes superimposed on a sketch depicting a gut epithelial cell, with the corresponding P value in the gut associated with S. marcescens infection. Green indicates processes to which susceptible candidates are exclusively attributed. Red indicates processes to which resistant candidates are exclusively attributed. Blue indicates processes to which both susceptible and resistant candidates can be attributed. See also table S10 for annotation of genes involved in each process. All processes shown display P < 0.05 (Fisher’s exact test).

Using GO enrichment analysis, we classified our tissue-specific candidates into statistically significant biological processes. In the intestinal tract, intracellular processes such as endocytosis and exocytosis, proteolysis, vesicle-mediated transport, and stress response all appeared significantly enriched (Fig. 2E, figs. S5 to S7, and table S10). We also observed a marked enhancement of genes associated with immune system development, growth, stem cell division, and cell death, which suggested an important role for these processes in the gut during S. marcescens infection. In hemocytes, ontology enrichment analysis revealed a strong enrichment in several processes linked to phagocytosis including endocytosis, response to external stimuli, and vesicle trafficking (figs. S8 to S10 and table S11). In both cell types, deregulation of the stress response, as well as amine and/or nitrogen metabolism, resulted in enhanced resistance to S. marcescens challenge (Fig. 2E and fig. S8).

We next performed Kegg pathway analysis to identify enriched gene sets that might be involved in S. marcescens infections. Kegg profiling on the susceptible genome-wide candidates (table S12) showed the importance of the IMD pathway in our infection model and also pointed to a possible role of Notch and transforming growth factor-β signaling pathways, which have previously been difficult to study in an infection setting because of a lack of adult viable mutants (21, 22). Moreover, our analysis revealed prominent involvement of the Janus kinase–signal transducers and activators of transcription (JAK-STAT) pathway during S. marcescens infection. In Drosophila, the JAK-STAT pathway plays an important role in hematopoiesis, stress responses, stem cell proliferation, and antiviral immunity, but its role in the defense against natural bacterial pathogens is unknown (2326). We therefore sought to validate our analysis and focused on how JAK-STAT signaling regulates the host response during S. marcescens infection.

To investigate whether the JAK-STAT pathway is activated during S. marcescens infection, we used transgenic reporter lines (24, 27, 28) in which green fluorescent protein (GFP) is expressed under the control of unpaired (upd) and upd-3, which encode two ligands for Domeless (the receptor of the JAK-STAT pathway). We observed upd-GFP and upd3-GFP expression in the gut of S. marcescens–infected flies (Fig. 3A and figs. S11 and S12). Moreover, we demonstrated intestinal activation of the JAK-STAT pathway by using a stat92E–binding site–GFP reporter line (Fig. 3B) (27, 28). On ligation of UPD or UPD3 to Domeless, Stat92E translocates to the nucleus and activates reporter GFP gene expression (27). To confirm the relevance of JAK-STAT activation for S. marcescens infections, we performed global (Fig. 3, C and D) and gut-specific (Fig. 3E) RNAi-mediated silencing of PIAS [also called Su(var)-10] and PP1α96A, two negative regulators of JAK-STAT signaling (29, 30). In both RNAi lines, we observed significantly earlier death compared with that of control flies (Fig. 3, C to E). The role of PP1α96A in intestinal immunity was also validated using a sensitized background (fig. S12). In contrast, partial pathway inhibition via gut-specific overexpression of PIAS (NP1-UAS-pias), dominant-negative domeless (NP1-UAS-domeDN), or RNAi-mediated silencing of the domeless ligand, UPD (NP1-RNAi-upd) significantly increased the survival of Serratia-challenged flies (Fig. 3F). Thus, the JAK-STAT pathway activation in the gut negatively regulates survival in response to an intestinal S. marcescens infection.

Fig. 3

The JAK-STAT pathway controls S. marcescens susceptibility in the gut. (A) GFP (green) and 4′,6′-diamidino-2-phenylindole (DAPI, blue) expression in the gut of transgenic upd-GFP flies on day 4 after infection with S. marcescens at 25°C compared with control, nonpathogenic conditions. Also shown is nuclear DAPI (blue) staining. (B) GFP expression in the gut of transgenic stat92E-GFP flies under S. marcescens–infected and control conditions on day 4 at 25°C. (C) Survival curves of S. marcescens–infected RNAi lines against the negative JAK-STAT pathway regulator PIAS driven by the ubiquitously expressed HSP-GAL4 driver compared with control and key mutant flies. (D) Survival curves of S. marcescens–infected RNAi lines targeting the negative JAK-STAT regulator PP1α96A driven by the ubiquitously expressed HSP-GAL4 driver compared with control and key mutant flies. (E) Survival graph representing individual tests of RNAi-mediated silencing of PIAS and PP1α96A specifically in the gut (NP1 driver) after S. marcescens challenge at 29°C, compared with control and key mutant flies. (F) Survival curves of lines shown at 29°C compared with control and key mutant flies following S. marcescens feeding. *P ≤ 0.05; ***P ≤ 0.0001 (logrank test). upd, unpaired; DN, dominant-negative.

To elucidate a possible mechanism in which JAK-STAT is involved in host defense against S. marcescens, we analyzed the effects of infection on gut epithelium. Infected flies exhibited massive death of intestinal epithelial cells (fig. S14A) and compensatory proliferation (fig. S14, B and C). Enhanced JAK-STAT signaling, through the use of NP1-RNAi-pp1α96A flies, resulted in a marked reduction in the number of large, polyploid nuclei, which signify differentiated enterocytes (31), after 5 days of infection (Fig. 4A). Epithelial morphology (fig. S15A) and survival on sucrose solution under nonpathogenic conditions (fig. S15B) were comparable for control, NP1-RNAi-pp1α96A, NP1-UAS-pias, and NP1-UAS-domeDN fly lines. We next assessed whether JAK-STAT signaling affected cellular proliferation of the epithelium. We found that DNA synthesis in epithelial cells was reduced when JAK-STAT signaling was impaired and significantly increased by silencing pp1α96A in the gut, both in the presence and absence of infection (Fig. 4A and fig. S16). Thus, JAK-STAT signaling enhances epithelial cell death and positively regulates compensatory proliferation of intestinal cells, also after S. marcescens infection.

Fig. 4

Impaired epithelial integrity and control of intestinal stem cell homeostasis upon S. marcescens challenge. (A) Analysis of gut epithelium integrity using DAPI (blue) and intestinal proliferation using EdU (green) staining; EdU was injected into flies just 3 hours before dissections. Samples were assayed on day 5 after S. marcescens infection at 25°C. (B) Representative confocal image showing JAK-STAT pathway activation in EdU-positive nuclei in an intestinal stem cell using the stat92E-GFP reporter line. Data are from day 5 following S. marcescens challenge. In a total of three experiments and 39 gut dissections, we detected 13 cells with small nuclei (DAPI) that were positive both for 10xSTAT-GFP (red) and positive for EdU staining in the region anterior to the copper cells, although no such cells were observed in 42 noninfected control guts (P < 0.003, Student’s t test). EdU was injected 3 hours before dissection. (C) Survival curves of S. marcescens–infected Drosophila in which PP1α96A is specifically silenced in intestinal stem cells of adult flies using Esg-GAL4;tubulinGal80ts at 25°C. Control and key mutant lines are shown for comparison. ***P < 0.0001 (logrank test). (D) Analysis of gut epithelium integrity in Esg-pp1a96A-RNAi lines kept for 5 days after S. marcescens infections at 25°C (top) or under nonpathogenic conditions (bottom). Nuclei were visualized with DAPI, and actin was visualized with phalloidin (green).

We next examined whether the JAK-STAT pathway was affecting intestinal cell homeostasis specifically through the resident stem cell compartment (32). Basal intestinal stem cells (ISCs) can be distinguished from apical enterocytes on the basis of a characteristic smaller nuclear morphology (31, 33). By using the stat92E-GFP reporter line to image JAK-STAT activation, the JAK-STAT pathway was selectively induced in the ISCs but not in mature enterocytes (fig. S17). Moreover, on infection of stat92E-GFP flies with S. marcescens, we observed GFP expression also in small, 5-ethynyl-2′-deoxyuridine (EdU)–positive cells, which suggests that JAK-STAT signaling regulates ISC proliferation during S. marcescens infection (Fig. 4B). To definitively demonstrate that this pathway acts in gut stem cells and that this compartment controls susceptibility to S. marcescens infections, we silenced pp1α96A in adult ISCs using an escargot-GAL4 driver line. Escargot is a specific marker of ISCs (31). ISC-specific suppression of PP1α96A resulted in early lethality in response to S. marcescens infection, whereas flies remained viable under nonpathogenic conditions (Fig. 4C and fig. S18). Furthermore, the guts of infected escargot-GAL4-pp1α96A-RNAi flies showed a phenotype similar to that obtained using the gut-specific NP1 driver, namely, severely depleted mature enterocytes (Fig. 4, A and D). Thus, our data demonstrate that JAK-STAT signaling is required for ISC homeostasis and implicates ISCs as a critical component of host defense to mucosal S. marcescens infections.

Our global experimental approach allows a comprehensive dissection of the biological processes that may regulate host defense against a bacterial infection at the organism level. Besides revealing previously known immune pathways, we uncovered more than 800 additional genes, many of which were of unknown function. Furthermore, our data demonstrate that host defense may involve many processes that are not limited to classical innate immune response pathways, as exemplified here by the role of the JAK-STAT pathway in the regulation of epithelial homeostasis in response to infection. In addition, we validate and map conserved candidates to intestinal cells and hemocytes, which allows us to propose a blueprint of the processes involved in host defense against S. marcescens infection. As all genes analyzed here are conserved during evolution, it is likely that some of the processes that are important in flies are also relevant to mammalian host defense (34, 35).

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S19

Tables S1 to S12


  • * These authors contributed equally to this work.

  • This work is based on equal contributions from the laboratories of the last two authors.

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank all members of our laboratories and the Vienna Drosophila RNAi Center for helpful discussions and technical support. We thank M. Novatchakova and M. Lafarge for expert technical help, the Drosophila Resource Center of the National Institute of Genetics of Japan for midgut Gal4 driver stocks, B. Mathey-Prévot for the HML-GAL4 line, and J. Mutterer for help with confocal microscopy. This work is supported financially by the CNRS, an NIH Program grant P01 AI44220, and a DROSELEGANS grant from the Programme Microbiologie, Immunologie, et Maladies Emergentes (MIME) of the Agence Nationale de la Recherche. The D.F. laboratory is an “Équipe FRM,” awarded by the Fondation pour la Recherche Médicale. J.M.P. is supported by IMBA, EuroThymaide, an Austrian Science Fund–Science Research Program (FWF-SFB) grant, an advanced European Research Council grant, and the Austrian Ministry of Science. R.M.S. and A.v.H. are supported by the Vienna Science and Technology Fund (WWTF) and the German Research Foundation (DFG) (Ha-1628/8-1). The screen was supported by Boehringer Ingelheim.
View Abstract

Stay Connected to Science

Navigate This Article