Abstract
We have screened the third chromosome of Drosophila melanogaster for mutations that prevent the normal immune response. We identified mutant lines on the basis of their failure to induce transcription of an antibacterial peptide gene in response to infection or their failure to form melanized clots at the site of wounding. These mutations define 14 genes [immune response deficient (ird) genes] that have distinct roles in the immune response. We have identified the molecular basis of several ird phenotypes. Two genes, scribble and kurtz/modulo, affect the cellular organization of the fat body, the tissue responsible for antimicrobial peptide production. Two ird genes encode components of the signaling pathways that mediate responses to bacterial infection, a Drosophila gene encoding a homolog of IκB kinase (DmIkkβ) and Relish, a Rel-family transcription factor. These genetic studies should provide a basis for a comprehensive understanding of the genetic control of immune responses in Drosophila.
INSECTS use sophisticated humoral and cellular immune responses to protect themselves from microbial infection (Hultmark 1993; Hoffman and Reichhart 1997; Khush and Lemaitre 2000). The fat body, an organ homologous to the mammalian liver, is responsible for the principal humoral response to infection, the secretion of a battery of antimicrobial peptides into the hemolymph. The transcription of the antimicrobial peptide genes in the fat body is induced within minutes after microbial infection. Different classes of microbes elicit the production of different spectra of antimicrobial peptides, indicating that complex regulation mediates the humoral responses (Lemaitreet al. 1997). Several cell types present in the insect hemolymph are also important in controlling infection (Rizki 1978; Elrod-Ericksonet al. 2000; Lebetskyet al. 2000). Macrophage-like cells phagocytose microorganisms, lamellocytes encapsulate larger invaders, and crystal cells release granules containing components required for clotting and melanization at the site of wounding.
Until recently, genes important for the Drosophila immune response were identified on the basis of their roles in other processes or by reverse genetic approaches. Normal induction of transcription of the gene encoding the antifungal peptide drosomycin depends on activation of Dif, one of the three Drosophila Rel proteins (Manfruelliet al. 1999; Menget al. 1999; Rutschmannet al. 2000a). Dif, identified molecularly as a Rel family member, is present in an inactive cytoplasmic complex prior to stimulation and moves into fat body nuclei within minutes after infection like other Rel proteins (Ipet al. 1993). In adult flies, Dif is activated via the Toll signaling pathway, which was first identified on the basis of its role in embryonic patterning. Mutations in Toll, spätzle (which encodes a putative ligand for Toll), pelle, or tube (downstream signaling components in the Toll pathway) all prevent normal induction of drosomycin (Lemaitreet al. 1996). Activation of Spätzle appears to require proteolytic processing that is inhibited by a serpin family protease inhibitor encoded by the necrotic gene (Levashinaet al. 1999).
Other Toll-like receptors and Rel proteins are important in other aspects of the Drosophila humoral immune response. The Toll-like receptor 18-wheeler is necessary for full induction of attacin and cecropin, two antibacterial peptide genes (Williamset al. 1997), but not for other antibacterial peptide genes, including diptericin. Mutations in Relish, another Drosophila Rel gene, reduce the expression of all of the known antimicrobial peptides: diptericin and cecropin are not induced, and induction of attacin, metchnikowin, and drosomycin expression are reduced (Hedengrenet al. 1999).
One mutation affecting the immune response, immune deficiency (imd; Lemaitreet al. 1995; Corbo and Levine 1996), was identified fortuitously. imd mutants fail to induce several antibacterial peptides in response to infection (Lemaitreet al. 1996). The imd phenotype demonstrated it should be possible to identify genes important in the immune response on the basis of their mutant phenotypes. To screen directly for genes required for the immune response, we looked for mutations that prevent the normal induction of an antibacterial peptide gene in the larval fat body and for mutations that prevent melanization at the wound site, which requires components secreted by crystal cells. We screened EMS-mutagenized lines for third chromosomal mutations affecting the induction of an antibacterial peptide reporter gene (diptericin-lacZ; Reichhartet al. 1992; Wu and Anderson 1998). We have identified 56 mutant lines that prevent diptericin reporter gene induction and 11 mutant lines that prevent the melanization response. We named the genes responsible for these phenotypes ird genes, for immune response deficient (ird).
Third chromosomal ird genes
We have mapped 14 ird genes to defined chromosomal regions. Northern blot analysis revealed that the ird mutations have a range of effects on the endogenous antimicrobial peptide genes, which allows the genes to be classified into groups. We have characterized four of the ird genes molecularly and find that some genes are required for the normal structure of the fat body organization and that other genes encode components of immune response signaling pathways.
MATERIALS AND METHODS
Genetic screen and Drosophila stocks: A three-generation crossing scheme was used to produce larvae homozygous for newly induced mutations (Figure 2). Homozygous claret (ca) males carrying the diptericin-lacZ (dipt-lacZ) reporter gene (Reichhartet al. 1992) on the third chromosome were mutagenized with 12.5 mm ethyl methanesulfonate (EMS; Lewis and Bacher 1968). The mutagenized males were crossed to virgin females heterozygous for a dominant temperature-sensitive (DTS4) mutation (Holden and Suzuki 1973) and a balancer (TM6B) that carries the dominant larval marker Tubby (Tb). Single F1 males (dipt-lacZ /DTS4 or dipt-lacZ /TM6B) were backcrossed to virgin females of the same stock (DTS4/TM6B). These crosses were kept at room temperature for 2 days and then shifted to 29° for 4 days to kill off the F2progeny carrying the DTS4 mutation. The remaining F2 flies (dipt-lacZ ca*/TM6B) were transferred to new vials and the F3 progeny that were homozygous for the mutagenized chromosome and the dipt-lacZ reporter gene (the non-Tubby larvae) were screened for their immune responses by the assays described below. Lines were maintained as balanced stocks as ird dipt-lacZ ca/TM6B.
Deficiency stocks and alleles of modulo and kurtz were obtained from the Bloomington Drosophila stock center and Gregg Roman and Ron Davis. Relish alleles were a gift from Svenja Stöven and Dan Hultmark.
Assays for immune function: For each mutagenized line, five homozygous (non-Tubby) larvae were placed in a well of a 24-well plate. The plate wells were coated with Sylgard to make the bottom of the wells rubbery so that the injection of larvae did not break the glass needles. An overnight culture of Escherichia coli was resuspended in phosphate-buffered saline (PBS) and injected into the larvae at a position just anterior to the anal pads (see Figure 1C) using a micro-injection needle and holder attached to a 10-ml syringe. The pressure in the needle was adjusted so that it would release solution if it touched the body wall. A small strip of Whatman filter paper moistened with 1% sucrose was placed in each well and the 24-well plate was sealed with Parafilm to prevent larvae escaping to other wells of the plate. After 2.5 hr, the Parafilm was removed and the larvae were visually inspected for melanization at the wound site. Melanization was also evident in most lines by the darkening of the filter paper in the well. The larvae from each line were then dissected in 1% glutaraldehyde in PBS for 5 min, rinsed in PBS, and then placed in a well of a 96-well plate with β-galactosidase (β-gal) stain solution (0.2% X-gal in 0.01 m sodium phosphate buffer, pH 7.2, 0.15 m NaCl, 1 mm MgCl2, 3.1 mm K4[FeII(CN)6], 3.1 mm K3[FeIII(CN)6], 0.3% Triton X-100) at 37° overnight. This screening procedure required an average of 2–3 min per line. Lines that failed to stain were retested several times to confirm the presence of an immune response defect.
Assays for immune response function. (A) The fat body of an infected wild-type larva carrying the diptericin-lacZ reporter gene. Three hours after E. coli were injected into the animal, the fat body was dissected and stained for β-galactosidase activity. (B) The fat body of an infected ird71 homozygous larva carrying the diptericin-lacZ reporter gene treated in the same way failed to express the reporter gene. (C) Melanization at the site of wounding (arrow) in a wild-type third instar larva, 1 hr after wounding.
Complementation tests and mapping: To determine whether the failure to express β-gal was due to inactivation of the dipt-lacZ reporter gene, heterozygous larvae (Tubby) from each mutant line were screened for induction of lacZ in response to infection. The expression of lacZ in response to infection indicated that the mutation was recessive and not in the reporter gene.
To map each ird mutant it was necessary to establish recombinant lines and then test larvae from the lines for the ability to induce diptericin-lacZ. The ird mutations were roughly mapped by meiotic recombination relative to the st cu sr e ca markers on the mwstipe (mwh th st ri roe p cu sr e) chromosome. Lines were set up for at least five recombinants in each interval. The dipt-lacZ reporter gene is in the sr-e interval. Homozygous recombinants that also carried the dipt-lacZ gene were assayed for lacZ induction. Recombinants that lost the dipt-lacZ gene by recombination were crossed to their original ird mutation and the non-Tubby F1 were assayed for lacZ induction. A number of ird lines were not adult viable, and we tested whether the lethality mapped to the same interval as the immune defect (see Table 1).
Complementation tests were done between mutations that mapped to the same region. Allelism was established if the dipt-lacZ reporter was not induced in the trans-heterozygotes.
Northern blot analyses of immune-challenged larvae: For each mutant line, 30 homozygous third instar larvae were placed in a well of a 24-well tissue culture plate (Falcon). E. coli injections were carried out as described above. After a 3-hr incubation, healthy animals were collected and homogenized in an Eppendorf tube by a pestle (Kontes) in RNA STAT-60 solution (TEL-TEST). Subsequent RNA purification steps were done following the manufacturer's instruction. Northern hybridization was done as described in Sambrook et al. (1989). Ten micrograms of total RNA was loaded in each lane, and the blot was hybridized at 68° overnight with indicated probes previously radiolabeled by random priming. The final washing was done in 0.1× SSC and 1% SDS at 68° for 1 hr. Quantification of the signals was done using a Fujifilm phosphoimager. The loading control was the transcript of the ribosomal protein 49 gene (rp49).
Molecular characterization of ird15: Genetic mapping located the ird15 mutation to 97B on the basis of its failure to complement Df(3R)Tl9QRX but complementation by Df(3R)ro80b. Mutations that map in that region were tested for complementation, which allowed the identification of three additional ird15 alleles: l(3)673, l(3)882 (two EMS alleles; K. V. Anderson, unpublished results), and l(3)J7B3j7B3 (a P-element allele; these stocks have been named scrib673, scrib882, and scribj7B3, respectively; Bilder and Perrimon 2000). Three genomic DNA fragments covering a 12.5-kb region surrounding the P element were obtained from a plasmid rescue with EcoRI and also from a genomic DNA library screening (Tamkunet al. 1992) using the plasmid-rescued fragment as a probe. A 2.3-kb fragment located 1.7 kb proximal to the P element contained a fragment of a transcript expressed in the third instar larval stage. Using this fragment as a probe, an embryonic cDNA library was screened (LD library, Berkeley Drosophila Genome Project), which allowed identification of an ird15 cDNA.
Immunohistochemical staining of fat bodies: Coracle antibody was used to stain septate junctions (Fehonet al. 1994), and FITC-labeled phalloidin (no. P-5282, Sigma, St. Louis) was used to stain adherens junctions. The third instar larvae of an appropriate genotype were dissected to expose the fat body in 4% paraformaldehyde in PBS, incubated for 15 min, washed three times in PBS for 5 min, and then permeabilized in 0.1% Triton X-100 and 1% bovine serum albumin (BSA) in PBS for 1 hr at room temperature. A 1:100 dilution of the mouse monoclonal anti-Coracle antibody and 1:150 dilution of the FITC-labeled phalloidin were added to the sample and incubated overnight at 4°. The samples were washed three times in 0.1% Triton X-100 and 0.1% BSA for 30 min and incubated with a 1:300 dilution of rhodamine-labeled goat anti-mouse IgG antibody (no. 115-025-100, Jackson ImmunoResearch, West Grove, PA), at room temperature for 1 hr. The samples were washed four times in 0.1% Triton X-100 and 0.1% BSA in PBS, and the fat bodies were dissected and mounted in 90% glycerol in PBS for microscopy.
RESULTS
Identification of mutations that prevent the normal immune response: We used two simple assays to identify mutations that affect the immune response, induction of a diptericin-lacZ reporter gene and melanization at the wound site. The diptericin-lacZ reporter gene, in which the expression of β-galactosidase is controlled by the regulatory region of the diptericin gene (Reichhartet al. 1992), is strongly and rapidly induced in response to injection of E. coli. The fat body of larvae carrying the diptericin-lacZ reporter gene stains dark blue at 2–3 hr after infection (Figure 1A). Wild-type larvae form a melanized clot at the injury site within 15 min after wounding (Figure 1C). It is believed that melanization depends on the secretion of proteins from the crystal cells, one of the two prominent cell types in the hemolymph that activate the prophenoloxidase cascade (Rizki 1978); melanin or some by-product of its synthesis appears to inhibit microbial growth (Nappi and Vass 1993).
Using the diptericin-lacZ and melanization assays, we carried out a systematic genetic screen for EMS-induced immune response mutations on the third chromosome. The crossing scheme for the screen is shown in Figure 2. Third instar larvae that were homozygous for newly induced mutations were injected with E. coli and then tested for melanization at the wound site and induction of the diptericin-lacZ reporter gene. Because we tested larvae rather than adults, it was possible to recover mutations that cause lethality at later developmental stages.
From 6429 F3 lines generated, 3627 lines produced homozygous viable third instar larvae that could be screened. Eleven mutations that inactivated the diptericin-lacZ reporter gene were identified and discarded. Homozygous larvae from 56 lines failed to induce diptericin-lacZ normally in response to infection (e.g., Figure 1B), and homozygous larvae from 11 lines failed to melanize at the wound site. All of the melanization mutants identified were also defective in diptericin-lacZ induction, suggesting that a common defect was responsible for both the abnormal melanization and antimicrobial responses (diptericin-lacZ) in these lines.
Crossing scheme used in the genetic screen. To identify new recessive mutations on the third chromosome, adult males homozygous for a third chromosome diptericin-lacZ reporter gene were treated with EMS. Single F1 males carrying the mutagenized third chromosome were used to establish lines. F2 male and female progeny from each line were intercrossed. F3 third instar larvae homozygous for the mutagenized diptericin-lacZ third chromosome were identified on the basis of the absence of the dominant marker Tubby on the TM6B balancer chromosome.
Mapping and complementation analysis: Of the 56 lines that failed to induce diptericin-lacZ, it was not possible to map the mutation responsible to a single locus in 8 lines. In an additional 27 lines, the homozygous larvae in the balanced stocks became less healthy and smaller over time, making analysis difficult. We therefore chose to focus on the remaining 21 lines and mapped the mutations responsible for the failure to induce the diptericin-lacZ reporter gene to defined chromosomal regions. Mutations that mapped to the same chromosomal interval were tested for complementation. On the basis of this analysis, 14 genes required for the induction of diptericin-lacZ were identified in the screen (Table 1). Nine genes were defined by more than one allele from the screen and/or a deficiency that failed to complement the immunity phenotype, and 5 genes were defined by single mutations.
Some of the ird mutants had additional phenotypes related to immune responses. Mutations in two of the mapped ird genes prevented normal melanization at the wound site (ird1 and ird15). Homozygous ird1 larvae also showed a high frequency of melanotic tumors. Melanotic tumors are melanized masses within the larva that arise when activated blood cells encapsulate self-tissue or when blood cells are activated spontaneously (Sparrow 1978). Mutations in four additional ird genes were associated with melanotic tumors but did not have melanization defects (ird6, ird9, ird16, and ird27). The penetrance of the melanotic tumor phenotype in ird1, ird6, and ird16 was ~80% and lower in ird9 and ird27. In the course of the screen, we also identified a number of melanotic tumor lines that did not affect melanization or diptericin induction, indicating that melanotic tumor formation does not prevent diptericin induction.
Induction of transcripts of antimicrobial peptide genes in response to infection in ird mutant larvae, as assayed on Northern blots. All RNA samples are from third instar larvae 3 hr after infection, except for uninfected wild-type larvae, labeled wt (uninf). Lanes grouped together are from the same experiment. The rp49 loading control is from the same blot as the line immediately above in each case. Duplicate independent experiments for each mutant gave results similar to those shown here. Note that ird1 and ird6 homozygous larvae are small, so the data may not be directly comparable to the other mutants shown, in which the larvae tested were comparable to wild-type larvae in size. dipt, diptericin; cecA1, cecropin A1; attA, attacin A; mtk, metchnikowin; drsm, drosomycin.
Mutations in four of the mapped ird genes were homozygous viable and fertile; these lines therefore have specific defects in aspects of the immune response. Animals homozygous for the other ird mutations died at the end of third instar or at the larval/pupal transition. The lethality mapped to the same region as the immune defect in these lines, suggesting that a single mutation was responsible for lethality and the immunity defect.
Classification of ird genes based on antimicrobial gene expression patterns: We assayed the induction of five antimicrobial peptide genes that are transcriptionally induced by infection in the ird mutants. Diptericin and attacin are antibacterial peptides (Samakovliset al. 1990; Hoffmann and Reichhart 1997), drosomycin is an antifungal peptide (Fehlbaumet al. 1994), and cecropin and metchnikowin are active against both bacteria and fungi (Levashinaet al. 1995; Ekengren and Hultmark 1999). The expression of each of these antimicrobial peptide genes in infected ird mutants is shown in Figure 3 and summarized in Figure 4.
Although all of the ird mutations were identified on the basis of the failure to induce the diptericin-lacZ reporter gene in the fat body to normal levels, the ird mutants produced a range of effects on endogenous diptericin expression from whole animals (Figure 3 and Table 1). Mutations in eight genes (ird1, ird4, ird5, ird6, ird7, ird16, ird24, and ird25) had a strong effect on the induction of the endogenous diptericin gene, expressing 0–30% of the wild-type level of diptericin RNA at 2–3 hr after infection. Mutations in the six other ird genes (ird8, ird9, ird15, ird21, ird26, and ird27) had milder effects on diptericin RNA induction, inducing diptericin to 30–80% of wild-type levels. It is possible that some of the mutants that had clear effects on reporter gene expression and weak effects on the levels of diptericin RNA isolated from whole larvae specifically affected expression in the fat body.
The expression patterns of the other antimicrobial peptide genes revealed that individual ird genes control different sets of antimicrobial peptide genes (Figure 4). For example, mutations in ird5 or ird7 cause similar profiles of antimicrobial peptide gene induction. Mutation in these genes prevented almost all induction of diptericin, cecropin, and metchnikowin and reduced induction of attacin, but allowed significant induction of drosomycin. In contrast, mutations in ird4 or ird16 had strong effects on all of the antimicrobial peptide genes. A few genes show different patterns of expression. For example, ird27 mutants had a specific effect on metchnikowin induction: no metchnikowin transcription was detected after infection, while diptericin, cecropin, attacin, and drosomycin were induced at 30–100% of the levels seen in wild type. Similarly, drosomycin expression was decreased to 40% of normal in ird21 mutants, without a significant effect on the other antimicrobial peptide genes.
Genes that encode Rel/NF-κB signaling pathway components: The map positions of two ird genes led to the discovery that these genes encode homologs of proteins that are important in mammalian Rel/NF-κB signaling. Of the three Drosophila Rel/NF-κB genes, only Relish mutants have a strong effect on diptericin expression (Hedengrenet al. 1999). Because the Relish gene is on the third chromosome, we tested whether Relish mutants could complement the immune response defects of the ird mutants that mapped between st and cu, the Relish interval. RelishE20 is a deletion that removes the transcription start sites of the Relish gene and therefore is a null allele of Relish (Hedengrenet al. 1999; Stövenet al. 2000). RelishE20 failed to complement the failure to induce diptericin-lacZ seen in the four ird4 alleles, indicating that these four mutations are alleles of Relish. The immune response phenotypes of the ird4 homozygotes are weaker than that of null alleles of Relish (Hedengrenet al. 1999), suggesting that ird4 alleles may be partial loss-of-function mutations.
Profiles of antimicrobial peptide gene induction in ird mutants. The data from Northern analysis, including that shown in Figure 3, were quantitated by phosphorimager analysis and are presented in bar graph form as the percentage of level of induction compared to wild-type animals in the same experiment. Each bar represents the average data from duplicate or triplicate experiments. Mutants are arranged into groups with similar profiles of antimicrobial gene expression.
We mapped the ird5 gene to region 89B of the third chromosome, the location of a Drosophila melanogaster homolog of the IκB kinase gene (DmIkkβ). In mammals, IκB kinases are required for the activation of NF-κB in response to inflammatory signals (Q. Liet al. 1999; Z. W. Liet al. 1999). We found that both ird5 alleles encode mutations that should inactivate DmIkkβ function (Luet al. 2001). Thus the activity of Ird5/DmIkkβ is essential for the induction of diptericin and other antibacterial peptides.
Fat body integrity is required for full induction of antimicrobial peptide genes: Genetic mapping allowed us to identify two additional ird genes that had been studied previously because of other phenotypes. The ird6 mutation causes a general decrease in the induction of antimicrobial peptide genes (Figures 3 and 4), as well as melanotic tumors, and maps to the distal tip of 3R. We therefore tested the ability of ird6 to complement mutations in two genes in that region that have been associated with melanotic tumors, modulo (mod) and kurtz (krz). The mod and krz genes lie immediately adjacent to one another and appear to share regulatory elements (Romanet al. 2000). mod mutations are recessive lethal and are associated with melanotic tumors and male sterility and act as dominant suppressors of position-effect variegation (Garzinoet al. 1992). kurtz mutations are also recessive lethal and are associated with melanotic tumors (Romanet al. 2000). We found that ird6 failed to complement the lethality of both mod [Df(3R)A4-4 L2 and Df(3R)A4-4 L3] and krz (krz1) alleles. In addition, ird6, like mod, acted as a dominant suppressor of position-effect variegation (data not shown). Mod protein binds DNA and may play a role in regulating gene expression through effects on chromatin structure (Garzinoet al. 1992; Perrinet al. 1999). The kurtz gene encodes a nonvisual arrestin, which is presumably required for inactivation of a G-protein-coupled receptor (Romanet al. 2000). In both mod and krz mutants, the fat body cells have abnormal morphology and are surrounded by lamellocytes that form melanotic tumors (Garzinoet al. 1992; Romanet al. 2000). Although we have not determined whether the molecular lesion in ird6 affects the kurtz or modulo genes specifically, it is likely that the immune response defects are a consequence of the abnormal fat body cells in the mutant.
Deficiency complementation mapping localized the mutation responsible for both the immune response phenotype and the late larval lethality of the ird15 to 97B (see materials and methods). Three additional mutations that map in the region failed to complement ird15, l(3)673, l(3)882, and l(3)j7B3. In ird15/ird15, ird15/Df, l(3)673/Df, l(3)882/Df, and l(3)j7B3/Df larvae all of the antimicrobial peptides were induced to ~50% of the wild-type level after infection (Figure 3 and data not shown). The l(3)882 and l(3)j7B3 alleles appeared to be hypomorphic: some l(3)882/l(3)j7B3 animals survived to adult stages, and larvae of this genotype showed apparently normal induction of the antimicrobial peptide genes in response to infection. The strong allele combinations also prevented normal melanization at the wound site: melanization did occur, but appeared weaker than in wild-type animals. In addition, melanized spots developed on the surface epidermis of the larvae in the absence of injury in the strong allele combinations.
We cloned DNA flanking the P-element insertion in l(3)j7B3 and identified ird15 as a large novel protein with leucine-rich repeats and PDZ domains. The same gene was identified on the basis of its requirement for epithelial organization and named scribble (Bilder and Perrimon 2000; Bilderet al. 2000). Homozygous scribble mutants display defects in epithelial structure and the abnormal epithelia show tumor-like overgrowth. These findings raised the possibility that structure of the fat body might be affected by the ird15/scribble mutations. The wild-type fat body is a sheet of cells one cell thick, in which the cells have a flattened morphology (Rizki and Rizki 1983). In contrast, the cells in the ird15/scribble larval fat body were more rounded (Figure 5), suggesting that the normal junctional structures were not formed. Coracle protein, which is homologous to band 4.1 protein and marks septate junctions (Fehonet al. 1994), and phalloidin staining, a marker for adherens junctions, were present in junctions in the fat body, but at substantially lower concentrations in the ird15/scribble than in wild type (Figure 5), indicating that the mutant fat body does not have the proper cellular organization.
Fat body morphology of wild-type and of ird15/scribble mutants. In unstained samples, the large fat body cells of ird15/scribble mutants assume a rounded shape, while wild-type fat body cells form a flat sheet. Immunohistochemical staining for Coracle, a septate junction protein, and for phalloidin, a marker for adherens junctions, marked cell junctions clearly in wild type but was much weaker in the ird15/scribble fat body.
Like scribble, discs-large (dlg) is a Drosophila tumor suppressor gene and encodes a PDZ-domain protein required for the formation of septate junctions and epithelial integrity (Woods and Bryant 1991; Bilderet al. 2000). We found that larvae homozygous for dlg mutations induced the antimicrobial peptides to ~50% of wild-type levels (data not shown), very similar to the ird15/scribble phenotype. Thus the data suggest that the proper structure of junctions in the fat body is important for full responsiveness to immune stimuli.
DISCUSSION
Identification of genes controlling the immune response: We have identified 14 genes on the Drosophila third chromosome that are required for the normal induction of antimicrobial peptide gene expression in response to infection. The identification of immune response deficient mutants and the molecular identification of several ird genes validates the forward genetic approach as an efficient means to identify novel genes and dissect signaling pathways important for the Drosophila immune system. Two or more alleles were isolated in 5 of the genes, and single alleles of the other 9 genes were identified in the screen. On the basis of this relatively low allele frequency, Poisson statistics suggest that there are likely to be 20–25 third chromosomal genes required for the normal humoral antibacterial response, either directly or indirectly. Extrapolating to the whole genome and considering the humoral antifungal and cellular immune responses, it is likely that >1% of the 14,000 genes in the Drosophila genome are required for the full immune response, although only a small fraction of those genes are likely to be directly involved in the signal transduction pathways that mediate the transcriptional induction of the antibacterial peptide genes.
Several ird genes affect induction of all antimicrobial peptide genes: Mutations in six ird genes (ird1, ird4, ird6, ird16, ird24, and ird25) decrease induction of all the antimicrobial peptide genes to 10–50% of wild-type induction levels. The ird4 mutations fail to complement Relish and map to the chromosome interval including the Relish gene (st-p). The ird4 phenotypes are weaker than the phenotype of Relish null mutants (Hedengrenet al. 1999), suggesting that the ird4 mutations are partial loss-of-function alleles of Relish. It should be noted, however, that the null alleles of Relish also inactivate a flanking gene, Nmdmc-B. Additional studies of the ird4/Relish mutations should establish how they alter Relish activity. It seems likely that some of the other genes in this group act in the pathway that activates Relish to allow induction of all of the antimicrobial peptide genes. Mutations in four additional genes, ird8, ird9, ird21, and ird26, have weaker effects on the induction of all the antimicrobial peptide genes. We have not identified deficiencies that remove these genes, so it is possible that these mutations represent partial loss-of-function alleles of genes required for activation of all the antimicrobial peptides.
We isolated single alleles of ird27 and ird21, which have unusual patterns of antimicrobial peptide gene induction. There is no detectable induction of metchnikowin in the ird27 mutants, while the induction of diptericin is ~40% of wild type. Metchnikowin is an unusual peptide that is active against both fungi and bacteria, and it therefore seems possible that it may be regulated differently from the other antimicrobial peptide genes. The ird21 gene appeared to specifically affect the induction of drosomycin, although drosomycin expression was still induced to 40% of normal levels in ird21 mutants. Because drosomycin induction is affected by mutations in the Toll pathway (Lemaitreet al. 1996), future studies can test whether ird21 acts in that pathway.
Fat body integrity is required for the normal humoral response: The ird mutants were identified because of lowered expression of the diptericin-lacZ reporter gene in the larval fat body, so it is not surprising that some ird mutants prevent a normal immune response because of changes in fat body organization. This mesodermal organ has an unusual structure: it is a single-cell-thick sheet that spreads through much of the body cavity, where it is exposed to the hemolymph (Rizki and Rizki 1983). The surfaces of the fat body cells that face the hemolymph are covered by a fibrillar basement membrane. The membranes of the cells are highly polarized: those cell surfaces that face the basement membrane are highly convoluted and have a high density of lectin binding sites, while the membranes between fat body cells are not convoluted and do not bind wheat germ agglutinin (Rizki and Rizki 1983).
Both the humoral response deficit and the melanotic tumor phenotype of ird6 mutants are likely to be the result of fat body degeneration. In ird6 and kurtz/modulo mutants, the fat body degeneration is associated with encapsulation of the abnormal fat body cells by activated lamellocytes and leads to melanotic tumors around the fat body.
The altered structure of the ird15 fat body is also likely to be responsible for its immune response-deficient phenotype. However, in contrast to ird6, the fat body of ird15/scribble mutants does not appear to be degenerating and is not associated with fat body melanotic tumors. The ird15 fat body is metabolically active and can induce heat-shock genes to normal levels in response to heat shock (data not shown). Animals that carry any of the four ird15/scribble alleles in trans to deficiency induce all of the antimicrobial peptide genes to only ~50% of wild-type levels. The ird15/scribble gene encodes a large cytoplasmic protein containing an N-terminal block of leucine-rich repeats (LRRs) and a central region including three PDZ domains required for the formation of septate junctions. Animals homozygous for mutations in dlg, which also encodes a PDZ-domain protein required for the formation of septate junctions (Woods and Bryant 1991; Bilderet al. 2000), have a similar immune response phenotype. Thus ird15/scribble and dlg are not essential for induction of the antimicrobial peptides, but are important for normal high levels of induction. The ird15/scribble and dlg genes are required for the assembly of septate junctions in epithelia, and the requirement for all of these genes for the full humoral response to infection indicates that septate junctions and cellular organization of the fat body are essential for full induction of antimicrobial peptides.
The function of ird15/scribble and dlg in the immune response could be indirect: lack of proper junctions could have an indirect effect on the physiology of fat body cells. It is also possible that Ird15/Scribble and Dlg may act more directly to control transmembrane signaling. PDZ-domain proteins can promote clustering of transmembrane receptors (Kim 1997). An LRR-PDZ protein, ERBIN, was identified on the basis of its interaction with the receptor tyrosine kinase ERBB2 and is important for restricting the receptor to the basolateral membrane of epithelial cells (Borget al. 2000). The human homolog of Dlg is expressed in T and B lymphocytes (Lueet al. 1994; Hanadaet al. 1997). In T cells, human Dlg becomes localized to sites of ligated CD2 and can associate with the Src family tyrosine kinase Lck (Hanada et al. 1997, 2000). Human Toll-like receptor 2 (TLR2) is recruited to phagosomes during phagocytosis by macrophages (Underhillet al. 1999), which presumably requires interaction between receptors mediated by other protein components. Future experiments will test whether Ird15/Scribble, Lgl, and Dlg affect localization of the transmembrane receptors that mediate immune responses in both Drosophila and mammals.
The ird5/DmIkkβ and ird7 genes encode components of a signaling pathway that activates Relish: On the basis of the profiles of antimicrobial gene expression in the ird mutants, two genes stand out as having similar, strong phenotypes, ird5/DmIkkβ and ird7 (Figure 4). Mutations in either gene block induction of diptericin and cecropin show no or decreased induction of metchnikowin and attacin, but have only mild effects on drosomycin induction. Two alleles of the two genes were isolated in the screen; in each case, both alleles show very similar effects on antimicrobial gene expression (Luet al. 2001 and data not shown). We identified deficiencies that fail to complement the immunity phenotypes of these two genes (Table 1). In each case, the antimicrobial gene expression profile of the ird/Df was very similar to the ird homozygous phenotype, indicating that the ird mutations represent strong loss-of-function alleles of these genes. Thus ird5/DmIkkβ and ird7 are specifically required for the induction of antibacterial peptide genes and not for the induction of drosomycin, which encodes an antifungal peptide.
The ird5/DmIkkβ and ird7 mutant animals are homozygous viable and fertile. The ird5/DmIkkβ and ird7 mutants melanize normally at wound sites and do not have melanotic tumors. Macrophages from mutant animals appear to phagocytose bacteria normally (B. Soper, personal communication). Because the ird5/DmIkkβ and ird7 mutations appear to be null alleles, we conclude that these genes are not required for development or general physiology and are specifically required for the humoral immune response to bacterial infection.
Like ird5/DmIkkβ and ird7, the Drosophila Rel/NF-κB gene Relish is absolutely required for the induction of diptericin and cecropin. It is therefore likely that ird5/DmIkkβ and ird7 are required for the activity of Relish in the immune response. Relish is a compound Rel-Ank protein, like mammalian p100 and p105. In response to immune stimulation, Relish undergoes a proteolytic processing event that separates the N-terminal Rel domain from the C-terminal ankyrin repeat domain, allowing the Rel domain to move into the nucleus (Stövenet al. 2000). We found previously that the ird mutations affect the nuclear localization of the Rel-protein Dif in the immune response (Wu and Anderson 1998). The results presented here suggest that Relish/Dif heterodimers may play a role in the induction of antibacterial peptide genes. On the basis of the profiles of antimicrobial gene expression in the ird mutants, it should be possible to design experiments that define the genes responsible for the activation of particular Rel dimers and to define the Rel dimers that activate specific antimicrobial peptide genes.
The ird5 gene encodes a Drosophila IκB kinase (DmIkkβ; Luet al. 2001) and appears to act by phosphorylating Relish and promoting its proteolytic processing (Silvermanet al. 2000). Mutations in the Ikkγ subunit of the IκB kinase complex also block the induction of antibacterial peptide genes (Rutschmannet al. 2000b). Mutations in Dredd also cause the failure to induce antibacterial peptide genes but allow induction of drosomycin (Elrod-Ericksonet al. 2000; Leulieret al. 2000). Dredd encodes a member of the caspase family (Chenet al. 1998), which could be the protease that activates Relish. The imd and ird7 gene products are also likely to play essential roles in the activation of the same Relish-containing dimers. Thus a combination of classical and molecular genetics has identified six different genes that act in this signaling pathway.
Genetic analysis of the Drosophila immune response: This work reports the findings from the first genetic screen that looked directly for mutations that affect the immune response. Phenotype-based genetic screens for immune response mutants are also underway for the first and second chromosomes in several labs (Elrod-Ericksonet al. 2000; Leulieret al. 2000; Rutschmann et al. 2000a,b; J. Delaney and K. V. Anderson, unpublished results). These screens, combined with the resources of Drosophila genomics, should define the mechanisms that lead to microbial recognition and response in Drosophila. This analysis should lead to the same in-depth understanding of the genetic control of immune response signaling that Drosophila genetics has provided in studies of the genetic control of development.
Acknowledgments
We thank Gregg Roman, Ron Davis, Svenja Stöven, Dan Hultmark, and the Drosophila Stock Centers in Bloomington and Umeå for stocks and Katie Brennan for helpful comments on the manuscript. This work was supported by National Institutes of Health grant AI-45149 and the Lita Annenberg Hazen Foundation. L.P.W. was a Leukemia and Lymphoma Society Special Fellow.
Footnotes
-
Communicating editor: T. Schüpbach
- Received April 9, 2001.
- Accepted June 13, 2001.
- Copyright © 2001 by the Genetics Society of America
