Drosophila melanogaster Is a Genetically Tractable Model Host for ...

INFECTION AND IMMUNITY, June 2003, p. 3540–3550 0019-9567/03/$08.00⫹0 DOI: 10.1128/IAI.71.6.3540–3550.2003 Copyright © 2003, American Society for Microbiology. All Rights Reserved.

Vol. 71, No. 6

Drosophila melanogaster Is a Genetically Tractable Model Host for Mycobacterium marinum Marc S. Dionne, Nafisa Ghori, and David S. Schneider* Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, California Received 6 November 2002/Returned for modification 15 January 2003/Accepted 6 March 2003

Mycobacterium marinum is a pathogenic mycobacterial species that is closely related to Mycobacterium tuberculosis and causes tuberculosis-like disease in fish and frogs. We infected the fruit fly Drosophila melanogaster with M. marinum. This bacterium caused a lethal infection in the fly, with a 50% lethal dose (LD50) of 5 CFU. Death was accompanied by widespread tissue damage. M. marinum initially proliferated inside the phagocytes of the fly; later in infection, bacteria were found both inside and outside host cells. Intracellular M. marinum blocked vacuolar acidification and failed to colocalize with dead Escherichia coli, similar to infections of mouse macrophages. M. marinum lacking the mag24 gene were less virulent, as determined both by LD50 and by death kinetics. Finally, in contrast to all other bacteria examined, mycobacteria failed to elicit the production of antimicrobial peptides in Drosophila. We believe that this system should be a useful genetically tractable model for mycobacterial infection. The past several years have seen growing interest in genetically tractable models of host-pathogen interactions. The first of these models resulted from the observation that Pseudomonas aeruginosa can cause a disease in Caenorhabditis elegans which in many ways resembles pseudomonal disease in humans (9, 30). More recently, it has been reported that Legionella pneumophila can infect and kill Dictyostelium discoideum (29), Salmonella enterica subsp. Typhimurium can kill C. elegans (2, 19), P. aeruginosa can kill Drosophila melanogaster (10), and Plasmodium gallinaceum can proliferate and develop within D. melanogaster, although the animal is not killed (27). These systems offer the potential of examining host-pathogen interactions at an unrivalled level of detail. The low cost of the model hosts allows screening methodologies which would be prohibitively expensive or laborious in vertebrate hosts (10, 15, 22, 31). Moreover, these hosts are also genetically tractable and permit rapid and potentially unbiased identification of novel host factors in pathogenesis (1, 9, 29). Given the proven utility of tractable model hosts in the study of other pathogens and the difficulty of working with established models of mycobacterial infection, we decided to examine the interaction of Mycobacterium marinum with the fruit fly D. melanogaster. M. marinum causes systemic disease in a wide variety of cold-blooded animals (8). This disease closely resembles human tuberculosis (24). Drosophila is perhaps the beststudied of all animals, and the genetic tools available in this system are unparalleled. Like vertebrates, Drosophila has bactericidal phagocytes, known as hemocytes (25; Schneider, unpublished observations). Adult flies or larvae can easily be injected with measured doses of bacteria (13). We report here that M. marinum kills Drosophila with a 50% lethal dose (LD50) of 5 CFU and that the initial stages of the infection closely resemble the early stages of M. marinum infections of frogs and fish.

MATERIALS AND METHODS Bacterial strains and culture. M. marinum strains M and L1D were cultured at 29°C in the dark without agitation in Middlebrook 7H9 broth supplemented with Middlebrook OADC (BD Bioscience) and 0.2% Tween. Middlebrook ADC was sometimes substituted for OADC, and there was no observable effect on bacterial growth. For plasmid maintenance, the culture medium was supplemented with 30 ␮g of apramycin per ml (msp12::GFP) or 30 ␮g of kanamycin per ml (mag24::GFP and mag85::GFP). Mycobacterium smegmatis was grown similarly; the strains used were mc2 and JCM89 (derived from mc2 by addition of a plasmid carrying kanamycin resistance and green fluorescent protein [GFP] under a constitutive promoter). For infection, bacteria were subcultured (1:100) from a stationary culture for 1 to 2 days. Before infection, the subculture was vortexed for 10 s, and then an aliquot was removed and quantified by determining the optical density at 600 nm. Flies. Flies were maintained on standard dextrose or molasses medium at 18 to 29°C and 65% humidity. Wild-type Oregon R flies were used except where indicated otherwise. The y w flies used correspond to Bloomington stock BL1495 (y1 w1). Fly infection. For infection, flies were anesthetized with CO2, and different numbers of bacteria in 50 nl (total volume) of medium were injected. Injection was carried out by using an individually calibrated pulled glass needle attached to a Picospritzer. Flies were always injected in the abdomen, close to the junction with the thorax and just ventral to the junction between the ventral and dorsal cuticles. Flies were never kept anesthetized for more than 15 min. After infection, flies were transferred to a new vial and maintained at 29°C and 65% humidity. Flies were transferred to fresh vials every other day (for mixed-sex cultures) or at least once a week (for all-male cultures). Bead injection. Dark red fluorescent 0.2-␮m-diameter polystyrene beads (F8807; Molecular Probes) were injected to block phagocytosis, as previously described (13, 27). Histology. Fly abdomens were dissected and fixed in 2% glutaraldehyde in phosphate-buffered saline (PBS) on ice and postfixed with OsO4. Next, samples were washed several times in PBS, dehydrated through an ethanol series, and embedded in Polybed 812. One-micrometer-thick sections were cut, stained with toluidene blue, and examined by standard light microscopy. Larval infections. Third-instar larvae were washed in distilled water or PBS and placed in a 15-cm-diameter petri dish, and bacteria were injected as described above. After injection, larvae were transferred to 15-cm-diameter petri dishes filled with dextrose Drosophila medium or with grape juice agar. For optimal expression of inducible GFP constructs, larvae were cultured at 29°C and a 65% humidity for 48 h. To stain acidified vesicles with Lysotracker, 50 nl of 1 mM Lysotracker Red DND-99 (Molecular Probes) in dimethyl sulfoxide was injected into each larva 15 min before bleeding. Larvae were bled by ripping them open in a 5- to 10-␮l drop of Schneider’s medium (Gibco) with 10% fetal bovine serum by using forceps. The drop was

* Corresponding author. Mailing address: Fairchild D333, Stanford University School of Medicine, Stanford, CA 94305-5124. Phone: (650) 724-8063. Fax: (650) 725-6757. E-mail: [email protected]. 3540

M. MARINUM INFECTIONS IN DROSOPHILA

VOL. 71, 2003

3541

then covered with a coverslip, and hemocytes were observed by using standard epifluorescence microscopy. Quantitative real-time reverse transcription (RT)-PCR. Flies were infected as described above and incubated at 29°C for the times indicated below. At given times, five flies were anesthetized and placed in 1.5-ml tubes, and total RNA was extracted with a Qiagen RNeasy Mini kit. The remaining genomic DNA was degraded by DNase I treatment. RT-PCR was carried out with a Bio-Rad iCycler by using TaqMan probes and rTth polymerase (Perkin-Elmer) as directed by the manufacturer. The following primers and were used: for attacin A, left primer CAA TGG CAG ACA CAA TCT GG, right primer ATT CCT GGG AAG TTG CTG TG, and Taqman probe AAT GGT TTC GAG TTC CAG CGG AAT G; for cecropin A1, left primer TCT TCG TTT TCG TCG CTC TC, right primer CTT GTT GAG CGA TTC CCA GT, and Taqman probe TTC TGG CCA TCA CCA TTG GAC AAT C; for diptericin, left primer ACC GCA GTA CCC ACT CAA TC, right primer CCC AAG TGC TGT CCA TAT CC, and Taqman probe CAG TCC AGG GTC ACC AGA AGG TGT G; and for ribosomal protein 15A, left primer TGG ACC ACG AGG AGG CTA GG, right primer GTT GGT GCA TGG TCG GTG A, and Taqman probe TGG GAG GCA AAA TTC TCG GCT TC. Data analysis. Survival data were graphed and analyzed by using GraphPad Prism (GraphPad Software, Inc.). Standard errors are calculated by the method of Greenwood. The 95% confidence intervals were computed as follows: 1.96 ⫻ standard error in each direction. All death curves were normalized at 2 days postinfection in order to exclude deaths resulting from wounding (at 2 days postinfection, the death rates for M. marinum-injected and medium-injected flies were identical).

RESULTS M. marinum kills D. melanogaster. To determine whether M. marinum can cause disease in Drosophila, various doses of bacteria were injected into flies (Fig. 1). All of the doses tested killed flies, while sibling flies that received sterile growth medium showed no significant mortality (Fig. 1). Animals infected with 500 CFU that died 5 to 7 days postinjection were extensively colonized (as determined by GFP fluorescence [see below]). The calculated LD50 was less than 5 CFU. In order to ensure that death was due to a pathogenic process and not to a generic host response to mycobacteria, we also injected M. smegmatis, a nonpathogenic fast-growing mycobacterium, into animals; 1,000 CFU of M. smegmatis caused lethality indistinguishable from that caused by injection of sterile medium in wild-type animals (Fig. 1). Also, 3,000 CFU of M. smegmatis could cause death comparable to that caused by 5 CFU of M. marinum, although the lethality was undependable (in some experiments, injection of 3,000 CFU of M. smegmatis did not cause more death than injection of medium [data not shown]). These findings support the idea that M. marinum kills flies via specific pathogenic interactions. M. marinum initially grows within hemocytes and then spreads systemically. In order to monitor the course of infection, 500 CFU of M. marinum expressing GFP under control of the constitutively active msp12 promoter (7) was injected into flies. The infection was monitored in individual animals by using fluorescence microscopy (Fig. 2). When samples were viewed under low power, there was little or no overt sign of bacterial growth until 96 h after infection. At 96 h, bacterial growth was evident near the dorsal midline at the anterior end of the abdomen (Fig. 2C). This was an invariant focus of infection in animals injected with 500 CFU; other foci of infection were usually present at this dose, but the location was variable and unpredictable. We attributed this apparent tissue tropism to the high number of hemocytes at this site (13). Once bacterial growth became apparent, its progression was relent-

FIG. 1. Survival of wild-type (Oregon R) flies infected either with M. marinum strain M carrying the msp12::GFP plasmid or with M. smegmatis strain JCM89. The error bars indicate 95% confidence intervals. Groups of 30 to 50 flies were injected with different quantities of M. marinum or M. smegmatis freshly diluted into sterile 7H9 medium containing OADC, 0.2% Tween 80, and 30 ␮g of apramycin per ml (or 30 ␮g of kanamycin per ml for M. smegmatis). Uninfected flies were injected under the same conditions with the same volume of sterile medium. Dead flies were counted at 24-h intervals. A total of 428 animals were infected with 500 CFU of M. marinum, 340 animals were infected with 50 CFU of M. marinum, 139 animals were infected with 5 CFU of M. marinum, 50 animals were infected with 3,000 CFU M. smegmatis, and 288 animals were not infected. (A) Survival after infection with 5 to 500 CFU of M. marinum. (B) Survival after infection with 500 CFU of M. marinum (Mm) compared with survival after infection with 1,000 CFU of M. smegmatis (Ms).

less, and animals typically died within 3 days. New foci typically continued to appear until late in the infection (Fig. 2D). Conspicuous sites of later bacterial growth usually included the legs, head, and wing veins. At a higher magnification, fluorescent bacteria were visible from the earliest times examined (Fig. 2E to G). The first stages of infection were characterized by transient spots which did not persist between 24-h time points; we believe that these spots corresponded to individual infected hemocytes (Fig. 2G). The number of these transient spots increased, but generally the size did not increase before 96 h (many such spots, as well as the first signs of apparent extracellular expansion, are visible

3542

DIONNE ET AL.

INFECT. IMMUN.

FIG. 2. y w flies, with the wings removed, infected with 500 CFU of msp12::GFP M. marinum and examined by GFP fluorescence at different times after infection. Panels A to D are photographs of one infected fly, panels E to J are photographs of a second fly, and panels K to M are photographs of a third fly. (A to D) Dorsal view of a whole fly. (A) Bright-field view before infection. The dashed rectangle indicates the approximate area seen more closely in panels E to M. (B) GFP fluorescence visible before infection. (C) The infection first became visible at 96 h at this magnification; fluorescence was initially visible near the dorsal midline at the anterior end of the abdomen (arrowhead). (D) The first visible focus of fluorescence expanded steadily through the rest of the infection. More fluorescent foci were visible at 144 h in the thorax (arrowhead), as well as in the legs and head (visible but out of the plane of focus). By 168 h, this fly was dead. (E to J) Close views of a second fly, showing the punctate fluorescence characteristic of the first stage of infection. (E) Bright-field view before infection. (F) GFP fluorescence visible before

VOL. 71, 2003

in Fig. 2J). Extracellular growth of bacterial colonies would not be expected to appear as individual spots of defined size; consequently, we believe that the early stage of infection is characterized largely by bacterial growth within hemocytes. Around 96 h, the bacterial growth pattern changed; individual spots began to expand into large patches (Fig. 2K). These bacterial patches corresponded to the infection as seen under low power. Severe tissue damage is present at late stages of M. marinum infection. M. marinum-infected flies with established disease (120 h postinjection) were examined histologically. These animals displayed severe tissue damage and cellular pathology (Fig. 3). There were many bacterium-filled abscesses in the fat body (Fig. 3D); in contrast, uninfected animals did not have such lesions (Fig. 3C). Many extracellular spaces in the infected animals were filled with bacteria (Fig. 3D), and there were numerous infected cells (Fig. 3D). Many fat body cells from infected animals appeared abnormal, with more numerous large vacuoles than healthy fat body cells (compare Fig. 3C and D). It is possible that these cells contained bacteria outside the plane of section or, alternatively, that the extracellular bacteria secreted some exotoxin. A third possibility is that the cellular pathology was caused by a systemic response to severe bacteremia. M. marinum expresses genes induced in vertebrate macrophages and does not colocalize with other phagocytosed particles or with vacuolar acidification. In larvae, hemocytes circulate and hence can be collected for examination in vitro, while in adults hemocytes are adherent and cannot be obtained from the animals by bleeding (13, 20; Schneider, unpublished observations). Conversely, larvae can only be cultured for a few days between molts, which makes examination of long-term disease possible only with adults. In order to better examine M. marinum infection at a cellular level, we took advantage of the fact that larvae contain many circulating phagocytes (20). For this experiment, dead Escherichia coli and live M. marinum were coinjected into larvae. The E. coli was labeled with tetramethyl rhodamine isocyanate (TRITC); the M. marinum carried either the mag24::GFP or the mag85::GFP plasmid. The mag promoters were previously isolated as promoters activated in M. marinum which had been phagocytosed by mouse macrophages (23). Hemocytes from animals bled at 6 h postinfection and examined immediately did not contain any visible green fluorescent particles (data not shown). However, by 48 h postinfection, bacteria carrying either plasmid were brightly fluorescent (Fig. 4A and D, green fluorescence), showing that the promoters were activated in Drosophila. At 48 h after infection, the M. marinum cells visible in hemocytes appeared to be intact. In contrast, phagocytosed dead E. coli had been completely broken down; their final locations in the cells were indicated by brightly fluorescent red vesicles (Fig. 4B and E, red fluorescence). Little or no colo-


3543

calization of the green and red signals was visible (Fig. 4C and F), confirming that the dead E. coli cells and the live M. marinum cells had not been sent to the same compartments. As expected, when cells which had phagocytosed green fluorescent dead E. coli cells were stained with Lysotracker, all vesicles which contained E. coli were also acidified (data not shown). To further examine the location of M. marinum in the phagocytes, larvae were injected with msp12::GFP M. marinum, aged for 24 h, and then injected with Lysotracker Red DND99, a membrane-permeable dye which is concentrated in acidic organelles. M. marinum did not colocalize with acidified compartments, suggesting that the bacteria are capable of blocking acidification of their vesicles in Drosophila, just as they are in vertebrate macrophages (4). We observed few or no extracellular M. marinum cells in this experiment (data not shown), supporting the idea that early in infection the vast majority of bacteria are within phagocytes. Prevention of phagocytosis alters the pattern of infection in the fly and demonstrates that most bacteria present at late stages of infection have passed through phagocytes. In order to confirm that the mag24 promoter was induced only upon phagocytosis, flies were injected with 0.2-␮m-diameter polystyrene beads at a dose which we had previously found to be sufficient to block phagocytic activity (13). Three days after injection of the beads, animals were injected with msp12::GFP (constitutive) M. marinum, mag24::GFP (inducible) M. marinum, or sterile 7H9 medium. If the mag24 promoter is activated only upon phagocytosis, little or no GFP fluorescence should have been visible in animals which were inoclated with beads and then infected with mag24::GFP M. marinum; in contrast, msp12::GFP M. marinum should still have been green fluorescent. This is what was found (Fig. 5E and F). Similar results were seen with bacteria carrying mag85::GFP (data not shown). Preinjection with beads altered the spatial pattern of infection. Most animals no longer showed strong infection in the anterior dorsal abdomen; instead, infections were found in many parts of the animal, especially in the head (Fig. 5E) and also in the wings and legs (data not shown). The change in the pattern of disease supports the idea that in the normal course of infection phagocytes capture bacteria but fail to kill them. The data described above indicate that there are two possible courses of M. marinum disease in the fly. Most of the bacteria present at late stages of infection could have been present by virtue of their parents having passed once or several times through the phagocytes of the fly, each time killing a phagocyte and releasing more bacteria, until all the phagocytes in the animal were killed and the bacteria were free to live in the extracellular space. Alternatively, the bacteria present at late stages of infection might have been descended from bacteria which had evaded phagocytosis, perhaps by lodging in

infection. (G to J) Foci of infection were visible as early as 8 h after bacterial injection (arrowhead in panel G). The number of foci steadily increased during the first 72 to 96 h of infection, and the foci were a different color than the autofluorescence visible in panel F. (K to M) Close views of a third fly, showing the spreading patches of fluorescence characteristic of the second stage of infection. Panel K is significantly dimmer than panel J because the brightness scale was changed for panels K to M in order to make the brighter fluorescence in panel M more interpretable. The dashed line in panel L shows the plane of section used for Fig. 3. Scale bar in panel A (for panels A to D), 1 mm. Scale bar in panel E (for panels E to M), 0.1 mm.

DIONNE ET AL.

FIG. 3. Histological examination of M. marinum-induced pathology. (A and B) Sections through uninfected (A) and infected (B) animals, roughly in the plane shown by the dashed line in Fig. 2L. The infected animal was examined 96 h after injection of 5,000 CFU of M. marinum. The cuticle (open arrowhead), dorsal vessel (solid arrowhead), muscle (solid arrow), and fat body (open arrow) are indicated. (C and D) Close-ups of the areas indicated by the rectangles in panels A and B. Extracellular (solid arrow) and intracellular (solid arrowhead) bacteria are evident, as is a small abscess of the fat body (open arrowhead).

3544 INFECT. IMMUN.

VOL. 71, 2003


3545

FIG. 4. Failure of internalized M. marinum to colocalize with phagocytosed E. coli or with acidified vesicles. (A, D, G, and J) Hemocyte from a larva injected with both mag24::GFP M. marinum and TRITC-labeled dead E. coli, bled and examined 48 h after injection. (B, E, H, and K) Hemocyte from an animal injected with both mag85::GFP M. marinum and TRITC-labeled dead E. coli, bled and examined 48 h after injection. (C, F, I, and L) Hemocyte from an animal injected with msp12::GFP, incubated for 24 h, injected with Lysotracker Red DND-99, bled, and examined. In each case, the red fluorescence (dead E. coli or Lysotracker) failed to colocalize with the internalized mycobacteria.

3546

DIONNE ET AL.

INFECT. IMMUN.

FIG. 5. Effect of blockage of phagocytosis on mag24 activation and progression of disease. The animals shown were all infected in parallel, as part of a single experiment. (A) Typical fly shown under bright-field illumination in order to orient the views in panels B to F. (B and C) Flies injected with water and then 3 days later with M. marinum carrying the msp12::GFP (B) or mag24::GFP (C) plasmid, examined for green fluorescence 5 days after infection. The patterns of infection in the two animals are apparently identical. (D to F) Flies injected with polystyrene beads to block phagocytosis and then 3 days later with sterile medium (D) or M. marinum carrying the msp12::GFP (E) or mag24::GFP (F) plasmid, examined for fluorescence 5 days after infection. In animals injected with beads first, the anterior dorsal abdomen was no longer the predominant focus of infection (E). Moreover, the mag24 promoter was not activated in these animals (F).

some protected part of the body, and divided there until the phagocytic resources of the fly were no longer capable of dealing with the infection. In order to establish which of these possibilities was true, flies were injected with 500 CFU of M.

marinum carrying either the msp12::GFP (constitutive) or mag24::GFP (inducible) plasmid. If the mag24 promoter were activated only by passage through phagocytes and the bacteria present late in infection were the descendants of bacteria

VOL. 71, 2003

FIG. 6. Survival of Oregon R flies infected with wild-type M. marinum (WT) or the L1D mutant of M. marinum. The error bars indicate 95% confidence intervals. Groups of 30 to 50 animals were injected with different quantities of M. marinum freshly diluted into sterile 7H9 medium containing OADC and 0.2% Tween 80. Uninfected flies were injected under the same conditions with the same volume of sterile medium. Dead flies were counted at 24-h intervals (i.e., 2 days postinfection [48 h postinfection]).

which had been phagocytosed earlier, then animals infected with mag24::GFP M. marinum should have displayed the same pattern of fluorescence at late stages of infection as animals infected with msp12::GFP M. marinum displayed. This is what was found (Fig. 5B and C). Identical patterns of fluorescence were seen whether animals were infected with msp12::GFP or mag24::GFP bacteria. msp12::GFP-infected animals were generally brighter than animals infected with mag24::GFP; we attributed the difference in brightness to the fact that most bacteria in both groups of animals had likely been extracellular (and hence the mag24 promoter was inactive) for at least 24 h at this point, as shown in Fig. 2. These data strongly support the idea that the majority of bacteria present late in infection have entered and proliferated within phagocytes. M. marinum L1D is less virulent in Drosophila. Since the mag24 promoter was activated in hemocytes, we wished to determine whether a mag24 mutant strain of M. marinum was less virulent in the fly. mag24 encodes a member of the PEPGRS family (23). The functions of this family of proteins are not yet completely clear but probably include roles in interactions both between individual bacteria (6) and with host factors (3, 14); it is notable that this family appears to be restricted to pathogenic mycobacteria (6). Mutant strain L1D, in which the mag24 locus is deleted, is less virulent in murine macrophages in vitro and in leopard frogs in vivo (23). We found that L1D bacteria were also less virulent in Drosophila. M. marinum L1D had an LD50 of 40 to 50 CFU, compared with an LD50 of 5 CFU for wild-type M. marinum strain M (compare Fig. 1 and 5). The death kinetics also supported a 10-fold defect in virulence (i.e., animals injected with 500 CFU of M. marinum L1D died with kinetics similar to those of animals injected with 50 CFU of wild-type M. marinum) (Fig. 6). This is comparable to what has been found in frogs, in which L1D displayed a 10- to 50-fold reduction in bacterial proliferation compared to the parental strain (23).


3547

M. marinum and M. smegmatis do not induce the production of antimicrobial peptides. The best-studied aspect of the Drosophila immune response is the production of circulating antimicrobial peptides. These peptides are produced rapidly (within a few hours of infection) in response to activation of one of three rel-related transcription factors in the fly (dif, dorsal, or relish). In order to examine peptide production in response to mycobacterial infection, we examined the expression of two antimicrobial peptides, metchnikowin and drosocin, using flies carrying promoter-GFP fusions. Animals were infected with 2,500 CFU of M. marinum, M. smegmatis, or Listeria monocytogenes (as a positive control) or with the same volume of sterile medium (as a negative control). At 24 h postinfection, L. monocytogenes induced expression of drosocin and metchnikowin (Fig. 7B and F); however, neither gene was induced by injection of M. marinum (Fig. 7C and G), M. smegmatis (Fig. 7D and H), or sterile medium (Fig. 7A and E). To further broaden the spectrum of analysis, we examined the expression of attacin A, cecropin A1, and diptericin in response to infection with 2,500 CFU of M. marinum, L. monocytogenes, or S. enterica serovar Typhimurium by real-time quantitative RT-PCR. Samples were taken and analyzed 6, 30, and 50 h after infection. M. marinum did not induce expression of any of these genes above the level induced by injection of sterile medium at any time (Fig. 7I). In contrast, injection of both Salmonella and Listeria resulted in long-lasting 10-fold induction (compared with the background level induced by wounding) of all three genes (Fig. 7I). Metchnikowin, drosocin, attacin A, cecropin A1, and diptericin are regulated differently by different pathogens; between them, these genes sample all the rel-related pathways in the fly (11, 12, 16). The fact that none of the antimicrobial agents examined was upregulated indicates strongly that there is little or no rel-mediated humoral immune response in the early stages of the mycobacterial infection of the fly. Flies carrying mutations in the imd and Toll pathways are not more susceptible than wild-type flies. The imd and Toll pathways are required for systemic induction of antimicrobial peptide production in response to gram-negative and grampositive bacterial infections, respectively (12). Since antimicrobial peptides were not induced by infection with mycobacteria, we infected flies carrying mutations in these two pathways to determine if there were other important immune responses induced by mycobacteria which might be mediated by imd or Toll signaling. To assay the imd pathway, we used the redshirt allele of kenny (key), which encodes a Drosophila homologue of IKK-␥ and which has previously been shown to be an indispensable component of the imd pathway (13, 26, 28). To assay the Toll pathway, we used animals lacking spaetzle (spz), which encodes a secreted protein required for activation of the Toll pathway (21). Neither of these mutants was more susceptible than the wild type to infection with M. marinum (Fig. 8). DISCUSSION M. marinum causes a lethal infection in Drosophila. Our goal was to establish a model for mycobacterial infection using the genetically tractable fruit fly D. melanogaster. Accordingly, we injected adult animals with M. marinum, a pathogenic mycobacterium closely related both to Mycobacterium ulcerans (the

3548

DIONNE ET AL.

INFECT. IMMUN.

FIG. 7. Antimicrobial peptide transcripts are not upregulated in response to M. marinum. (A to H) Animals injected with sterile Luria broth and 7H9 (A and E), L. monocytogenes (B and F), M. marinum (C and G), or M. smegmatis (D and H) and carrying GFP reporters for drosocin (A to D) and metchnikowin (E to H). M. marinum did not induce expression of these antimicrobial agents at levels above the levels induced by the process of injection (compare panels A to D with panels E to H). (I) Animals were injected with sterile medium (mock), S. enterica serovar Typhimurium (S.t.), L. monocytogenes (L.m.), or M. marinum (M.m.). The levels of mRNA encoding attacin A, cecropin A1, and diptericin were determined 6, 30, or 50 h after infection by quantitative real-time RT-PCR. The error bars indicate standard deviations. The experiment was repeated three times. In each case, antimicrobial agents were induced by L. monocytogenes and S. enterica serovar Typhimurium but not by M. marinum. The transcript levels were normalized to the expression level of Drosophila ribosomal protein 15A.

VOL. 71, 2003

FIG. 8. Survival of Oregon R, spaetzle, and kenny flies infected with M. marinum. The error bars indicate 95% confidence intervals. Groups of 30 to 50 animals were injected with different quantities of M. marinum freshly diluted into sterile 7H9 medium containing OADC and 0.2% Tween 80. Uninfected flies were injected under the same conditions with the same volume of sterile medium. Dead flies were counted at 24-h intervals (i.e., 2 days postinfection [48 h postinfection]). spz2/ spzrm7 flies were used because the chromosomes carrying each of the two spaetzle alleles also carry different recessive lethal mutations.

causative agent of Buruli ulcer) and to Mycobacterium tuberculosis (32). Injected M. marinum establishes an infection in Drosophila which kills the animal when inocula as small as 5 CFU are used. We believe that this infection is a specific pathogenic process rather than nonspecific bacteremia, for several reasons. First, M. smegmatis is more than 500-fold less virulent in the fly than M. marinum is, although M. smegmatis grows roughly twice as fast as M. marinum in vitro. Second, deletion of the mag24 gene of M. marinum reduces virulence in the fly, just as it does in frogs and mouse macrophages (5, 23); this mutant exhibits wild-type growth in vitro (23). Third, in hemocytes from infected larvae, internalized M. marinum does not colocalize either with acidified organelles or with internalized dead E. coli, suggesting that M. marinum successfully subverts fly macrophages just as it subverts the macrophages of vertebrates (4). We examined several aspects of this disease. At early stages of infection, the bacteria localize to phagocytes, and at later stages of infection they are found both inside and outside cells. Later stages of infection are characterized by widespread, severe tissue damage, which we believe to be the ultimate cause of death. Growth pattern of M. marinum in the fly. M. marinum grows in the phagocytes of the fly during early stages of infection. Four lines of data support this conclusion. First, early sites of infection are coincident with regions where there are many phagocytes (Fig. 2) (13). Second, for the first 4 days after infection with M. marinum that is constitutively expressing GFP, bacterial fluorescence is almost exclusively punctate, as would be expected if the bacteria are growing primarily within hemocytes (Fig. 2). Third, the apparent tropism of M. marinum for the anterior dorsal abdomen is disrupted in flies in which phagocytosis is blocked, which is consistent with a role for phagocytes in this region in collecting the bacteria (Fig. 5). Fourth, flies infected with bacteria which express GFP under


3549

the macrophage-inducible mag24 or mag85 promoter display a pattern of fluorescence indistinguishable from that of flies infected with bacteria which express GFP constitutively; this fluorescence is absent in animals in which phagocytosis is blocked, confirming that these promoters are activated only in the intracellular milieu (Fig. 5). Taken together, these data indicate strongly that M. marinum proliferates in Drosophila phagocytes early in infection. We believe that the shift from punctate growth to spreading patches of infection corresponds to the death of all or most of the phagocytes in the animal. Once the phagocytic response has been exhausted, the bacteria are able to proliferate freely in the hemolymph. We are currently testing this possibility. Mycobacteria do not activate antimicrobial peptide production in the fly. The Drosophila humoral immune system has been extensively studied as a model for innate immune signaling in mammals (18). The current models involve two parallel pathways, the imd pathway and the Toll pathway; the imd pathway has been implicated in responses to gram-negative bacteria, while the Toll pathway has been implicated in responses to gram-positive bacteria and fungi. Both pathways result in production of large quantities of secreted antimicrobial peptides via the activation of Rel-related transcription factors, as well as other less-characterized responses (11, 12, 16). Intriguingly, neither M. marinum nor M. smegmatis activates either pathway, as measured by known transcriptional responses to activation. Moreover, flies carrying mutations in either pathway displayed wild-type sensitivity to M. marinum infection. There are two possible explanations for this; either the fly fails to recognize mycobacteria as invaders, or some mycobacterial component blocks Rel activation. It has been suggested that macrophage apoptosis may be an important aspect of the immune response to M. tuberculosis. This suggestion is supported by the observation that M. tuberculosis-infected macrophages undergo apoptosis to an extent which is inversely proportional to the severity of disease caused by the infecting strain (17). In vertebrates, Rel signaling has a pronounced antiapoptotic effect. These observations raise the possibility that the lack of antimicrobial peptide induction may be a consequence of an apoptotic antimycobacterial defense in the fly. We are currently testing this hypothesis. The M. marinum-Drosophila model should help find host factors important for macrophage infection. Phagocyte-mediated killing is an important aspect of the host defense against mycobacterial infection. However, little is known about the mechanisms by which infectious mycobacteria evade this killing. The genetic tractability of Drosophila and the fact that cellular immunity is intrinsic to the disease process make this an ideal model system for studying the interaction of mycobacteria with phagocytes. ACKNOWLEDGMENTS We thank Bruno Lemaitre for Drosocin-GFP and MetchnikowinGFP flies; Ranjiv Khush, Kaman Chan, Igor Brodsky, Charlie Kim, Stan Falkow, Man-Wah Tan, and Karen Liu for helpful discussions; Linh Pham, Man-Wah Tan, Kaman Chan, and Stanley Falkow for comments on the manuscript; and Kaman Chan, Igor Brodsky, Jeff Cox, and Julie Theriot for bacterial strains. This work was supported in part by a Stanford Medical School Dean’s Postdoctoral Fellowship to M.S.D.

3550

DIONNE ET AL.

INFECT. IMMUN.

REFERENCES 1. Aballay, A., and F. M. Ausubel. 2001. Programmed cell death mediated by ced-3 and ced-4 protects Caenorhabditis elegans from Salmonella typhimurium-mediated killing. Proc. Natl. Acad. Sci. USA 98:2735–2739. 2. Aballay, A., P. Yorgey, and F. M. Ausubel. 2000. Salmonella typhimurium proliferates and establishes a persistent infection in the intestine of Caenorhabditis elegans. Curr. Biol. 10:1539–1542. 3. Banu, S., N. Honore, B. Saint-Joanis, D. Philpott, M. C. Prevost, and S. T. Cole. 2002. Are the PE-PGRS proteins of Mycobacterium tuberculosis variable surface antigens? Mol. Microbiol. 44:9–19. 4. Barker, L. P., K. M. George, S. Falkow, and P. L. Small. 1997. Differential trafficking of live and dead Mycobacterium marinum organisms in macrophages. Infect. Immun. 65:1497–1504. 5. Bouley, D. M., N. Ghori, K. L. Mercer, S. Falkow, and L. Ramakrishnan. 2001. Dynamic nature of host-pathogen interactions in Mycobacterium marinum granulomas. Infect. Immun. 69:7820–7831. 6. Brennan, M. J., G. Delogu, Y. Chen, S. Bardarov, J. Kriakov, M. Alavi, and W. R. Jacobs, Jr. 2001. Evidence that mycobacterial PE_PGRS proteins are cell surface constituents that influence interactions with other cells. Infect. Immun. 69:7326–7333. 7. Chan, K., T. Knaak, L. Satkamp, O. Humbert, S. Falkow, and L. Ramakrishnan. 2002. Complex pattern of Mycobacterium marinum gene expression during long-term granulomatous infection. Proc. Natl. Acad. Sci. USA 99: 3920–3925. 8. Clark, H. F., and C. C. Shepard. 1963. Effect of environmental temperatures on infection with Mycobacterium marinum (balnei) of mice and a number of poikilothermic species. J. Bacteriol. 86:1057–1069. 9. Darby, C., C. L. Cosma, J. H. Thomas, and C. Manoil. 1999. Lethal paralysis of Caenorhabditis elegans by Pseudomonas aeruginosa. Proc. Natl. Acad. Sci. USA 96:15202–15207. 10. D’Argenio, D. A., L. A. Gallagher, C. A. Berg, and C. Manoil. 2001. Drosophila as a model host for Pseudomonas aeruginosa infection. J. Bacteriol. 183:1466–1471. 11. De Gregorio, E., P. T. Spellman, G. M. Rubin, and B. Lemaitre. 2001. Genome-wide analysis of the Drosophila immune response by using oligonucleotide microarrays. Proc. Natl. Acad. Sci. USA 98:12590–12595. 12. De Gregorio, E., P. T. Spellman, P. Tzou, G. M. Rubin, and B. Lemaitre. 2002. The Toll and Imd pathways are the major regulators of the immune response in Drosophila. EMBO J. 21:2568–2579. 13. Elrod-Erickson, M., S. Mishra, and D. Schneider. 2000. Interactions between the cellular and humoral immune responses in Drosophila. Curr. Biol. 10:781–784. 14. Espitia, C., J. P. Laclette, M. Mondragon-Palomino, A. Amador, J. Campuzano, A. Martens, M. Singh, R. Cicero, Y. Zhang, and C. Moreno. 1999. The PE-PGRS glycine-rich proteins of Mycobacterium tuberculosis: a new family of fibronectin-binding proteins? Microbiology 145:3487–3495. 15. Garsin, D. A., C. D. Sifri, E. Mylonakis, X. Qin, K. V. Singh, B. E. Murray, S. B. Calderwood, and F. M. Ausubel. 2001. A simple model host for identifying Gram-positive virulence factors. Proc. Natl. Acad. Sci. USA 98: 10892–10897. 16. Irving, P., L. Troxler, T. S. Heuer, M. Belvin, C. Kopczynski, J. M. Reich-

Editor: B. B. Finlay

17. 18. 19. 20. 21.

22. 23. 24. 25. 26. 27. 28. 29.

30. 31.

32.

hart, J. A. Hoffmann, and C. Hetru. 2001. A genome-wide analysis of immune responses in Drosophila. Proc. Natl. Acad. Sci. USA 98:15119–15124. Keane, J., H. G. Remold, and H. Kornfeld. 2000. Virulent Mycobacterium tuberculosis strains evade apoptosis of infected alveolar macrophages. J. Immunol. 164:2016–2020. Khush, R. S., F. Leulier, and B. Lemaitre. 2001. Drosophila immunity: two paths to NF-kappaB. Trends Immunol. 22:260–264. Labrousse, A., S. Chauvet, C. Couillault, C. L. Kurz, and J. J. Ewbank. 2000. Caenorhabditis elegans is a model host for Salmonella typhimurium. Curr. Biol. 10:1543–1545. Lanot, R., D. Zachary, F. Holder, and M. Meister. 2001. Postembryonic hematopoiesis in Drosophila. Dev. Biol. 230:243–257. Levashina, E. A., E. Langley, C. Green, D. Gubb, M. Ashburner, J. A. Hoffmann, and J. M. Reichhart. 1999. Constitutive activation of toll-mediated antifungal defense in serpin-deficient Drosophila. Science 285:1917– 1919. Mahajan-Miklos, S., M. W. Tan, L. G. Rahme, and F. M. Ausubel. 1999. Molecular mechanisms of bacterial virulence elucidated using a Pseudomonas aeruginosa-Caenorhabditis elegans pathogenesis model. Cell 96:47–56. Ramakrishnan, L., N. A. Federspiel, and S. Falkow. 2000. Granulomaspecific expression of Mycobacterium virulence proteins from the glycine-rich PE-PGRS family. Science 288:1436–1439. Ramakrishnan, L., R. H. Valdivia, J. H. McKerrow, and S. Falkow. 1997. Mycobacterium marinum causes both long-term subclinical infection and acute disease in the leopard frog (Rana pipiens). Infect. Immun. 65:767–773. Ramet, M., P. Manfruelli, A. Pearson, B. Mathey-Prevot, and R. A. Ezekowitz. 2002. Functional genomic analysis of phagocytosis and identification of a Drosophila receptor for E. coli. Nature 416:644–648. Rutschmann, S., A. C. Jung, R. Zhou, N. Silverman, J. A. Hoffmann, and D. Ferrandon. 2000. Role of Drosophila IKK gamma in a toll-independent antibacterial immune response. Nat. Immunol. 1:342–347. Schneider, D., and M. Shahabuddin. 2000. Malaria parasite development in a Drosophila model. Science 288:2376–2379. Silverman, N., R. Zhou, S. Stoven, N. Pandey, D. Hultmark, and T. Maniatis. 2000. A Drosophila IkappaB kinase complex required for Relish cleavage and antibacterial immunity. Genes Dev. 14:2461–2471. Solomon, J. M., A. Rupper, J. A. Cardelli, and R. R. Isberg. 2000. Intracellular growth of Legionella pneumophila in Dictyostelium discoideum, a system for genetic analysis of host-pathogen interactions. Infect. Immun. 68:2939– 2947. Tan, M. W., S. Mahajan-Miklos, and F. M. Ausubel. 1999. Killing of Caenorhabditis elegans by Pseudomonas aeruginosa used to model mammalian bacterial pathogenesis. Proc. Natl. Acad. Sci. USA 96:715–720. Tan, M. W., L. G. Rahme, J. A. Sternberg, R. G. Tompkins, and F. M. Ausubel. 1999. Pseudomonas aeruginosa killing of Caenorhabditis elegans used to identify P. aeruginosa virulence factors. Proc. Natl. Acad. Sci. USA 96:2408–2413. Tonjum, T., D. B. Welty, E. Jantzen, and P. L. Small. 1998. Differentiation of Mycobacterium ulcerans, M. marinum, and M. haemophilum: mapping of their relationships to M. tuberculosis by fatty acid profile analysis, DNADNA hybridization, and 16S rRNA gene sequence analysis. J. Clin. Microbiol. 36:918–925.