Essential aspects of innate immune responses to microbial infections appear to be conserved between insects and mammals. In particular, in both groups, transmembrane receptors of the Toll superfamily play a crucial role in activating immune defenses. The Drosophila Toll family member 18‐Wheeler had been proposed to sense Gram‐negative infection and direct selective expression of peptides active against Gram‐negative bacteria. Here we re‐examine the role of 18‐Wheeler and show that in adults it is dispensable for immune responses. In larvae, 18wheeler is required for normal fat body development, and in mutant larvae induction of all antimicrobial peptide genes, and not only of those directed against Gram‐negative bacteria, is compromised. 18‐Wheeler does not qualify as a pattern recognition receptor of Gram‐negative bacteria.
The host defense of Drosophila shares significant similarities with mammalian innate immunity (Hoffmann et al., 1999). A hallmark of the Drosophila immune reaction is the infection‐induced transcription in the fat body of a battery of genes encoding small cationic peptides with potent activities against fungi and/or bacteria (Bulet et al., 1999). In the mid‐1990s, it was proposed that distinct signaling pathways control the resistance against fungi and bacteria: Toll, a transmembrane receptor protein, and several additional genes of the dorsoventral regulatory cassette of Drosophila regulate the antifungal response (Lemaitre et al., 1996), whereas resistance to Gram‐negative bacteria depends on the imd gene (for immune deficiency). This gene has since been shown to act upstream of a regulatory pathway independent of Toll (for a review, see Hoffmann and Reichhart, 2002). The discovery of the role of Toll in the host defense of Drosophila was shortly followed by the cloning of a Toll‐like receptor (TLR) in mammalian cells and the demonstration of its role in innate immunity (Medzitov et al., 1997). To date, 10 mammalian TLRs have been reported (Takeuchi and Akira, 2001).
Drosophila expresses, in addition to Toll, eight genes coding for transmembrane receptors with significant similarities to Toll (Tauszig et al., 2000). One of these, 18‐Wheeler (18w), was shown to serve, like Toll, as an adhesion molecule during various processes of morphogenesis (Eldon et al., 1994). In 1997, loss‐of‐function mutant 18w larvae were reported to exhibit severely reduced inducibility of the attacin gene in response to Gram‐negative infection (Williams et al., 1997). Attacin is considered to be predominantly active against Gram‐negative bacteria and the report raised the interesting prospect that the Toll‐like 18‐Wheeler sensed Gram‐negative infection and signaled to express an anti‐Gram‐negative peptide. This situation appeared to parallel that observed with Toll, which responds to fungal infection by directing expression of the antifungal peptide drosomycin. From these studies, the attractive paradigm emerged that distinct Toll receptors discriminate between different groups of microorganisms and control the selective expression of antimicrobial peptides with activity spectra directed against the corresponding invading microorganisms. This paradigm was indirectly supported by the accumulating evidence that mammalian TLRs are activated by distinct microbial patterns, e.g. TLR2 by peptidoglycan, TLR4 by lipopolysaccharide, TLR5 by bacterial flagellin, etc. (for a review, see Imler and Hoffmann, 2001; Takeuchi and Akira, 2001).
The observation that the Drosophila genome encodes nine distinct transmembrane receptors of the Toll family raised the tantalizing prospect that this battery of Tolls can serve to discriminate between a plethora of microbial patterns. This laboratory has taken up this question (Tauszig et al., 2000), which has also led to the re‐evaluation here of the role of 18w in the antimicrobial response. The data that we present provide different pictures for larvae and adults. A functional 18‐Wheeler product appears to be required for the larval fat body to develop into a fully responsive immunocompetent tissue. Mutant larvae are unable to express wild‐type levels of any of the antimicrobial peptide genes in response to immune challenge. In contrast, 18w mutant adults have a wild‐type response to bacterial challenge, and in particular express the attacin genes at wild‐type levels. These data are not in favor of a selective role for the 18W receptor in sensing Gram‐negative infection to mount an appropriate response.
The initial studies, which involved the 18w gene as a receptor of Gram‐negative sepsis, were based on the 18w7‐35 mutation (Eldon et al., 1994). This 18w allele had been generated by imprecise excision of a P element inserted 23 bp upstream of the transcription start site. The excision had removed a DNA fragment of ∼2.2 kb, comprising 1.7 kb of N‐terminal sequence of the 18w open reading frame. In molecular terms, the 18w7‐35 allele generates an N‐terminally truncated transcript, which results in expression of a protein lacking the extracellular and transmembrane domains of the wild‐type receptor and effectively corresponding to an N‐terminally truncated TIR (Toll‐interleukin‐1 receptor homology) domain. This protein is detected in the cytoplasm and is not localized to the cell membrane (Williams et al., 1997). 18w7‐35 mutants develop up to the third larval instar, and after a prolonged duration of this stage (up to 4 days, see also Methods) die prior to pupariation. Very few escapers are occasionally observed, which reach adulthood.
To minimize potential interferences of the truncated TIR domain with other signaling pathways and to eliminate possible contributions of other mutations in the genetic background of the 18w7‐35 allele, we combined this mutation with a chromosomal deletion (FlyBase: the database of the Drosophila genome projects and community literature; http://flybase.bio.indiana.edu/), Df(2R)017, which uncovers the 56F6‐9 locus to which 18w maps (Eldon et al., 1994) (see also Figure 1A). This combination produces viable adults (although in numbers significantly lower than the expected Mendelian ratios), indicating that the lethality of the 18w7‐35 larvae did not entirely result from the lack of a functional 18W product. In the following, we will address separately the possible roles of 18w in the antimicrobial response in adults and in larvae.
18wheeler adults have a wild‐type antimicrobial response
We have subjected 18w7‐35/Df(2R)017 flies to infection by the Gram‐negative bacteria Escherichia coli and have observed, as illustrated in Figure 1B, that all antimicrobial peptide genes are expressed as in wild‐type flies. In keeping with results obtained previously by this and other groups, the predominant reactants of this Gram‐negative challenge were the cecropin, diptericin, defensin, metchnikowin, drosocin and attacin genes. Induction of drosomycin was less marked. In contrast, infection with the Gram‐positive bacteria Micrococcus luteus resulted in a stronger induction of drosomycin, but again the levels induced for the various peptide genes were not significantly different between 18w mutants and wild type.
We were intrigued by a recent report suggesting that the attacinC gene, rather than the attacinA gene, was dependent on an input from the 18W receptor (Hedengren et al., 2000). This report was based on 18w7‐35 homozygous adult escapers. In our hands, this mutation did not give enough adult flies to perform northern blots and we therefore addressed the 18w7‐35/Df(2R)017 hemizygous strains. Four northern blots performed on total RNA extracted from these flies did not show any difference between the inducibility of attacinC in wild type and mutants. AttacinC was recently shown by Bulet and colleagues in this laboratory to give rise to a cleaved Pro‐domain, which is post‐translationally modified and stable (P. Bulet et al., unpublished results). This Pro‐domain form has a molecular weight of 2972 Da and is easily detected in hemolymph samples from single flies by MALDI‐TOF mass spectrometry. We have found that this molecule is present after Gram‐negative infection in the hemolymph of wild‐type and 18w7‐35/Df(2R)017 adults (Figure 2). Interestingly, it is totally absent in mutants for the Imd pathway, but is present in Dif mutants (data not shown).
Flies mutant for the Toll pathway are highly susceptible to fungal infection, while survival of Gram‐negative infection is severely reduced in flies mutant for the Imd pathway. We have subjected 18w7‐35/Df(2R)017 flies to fungal and Gram‐negative infection, but did not observe any difference in survival rates compared with wild‐type flies (data not shown).
18W has been suggested to regulate immune responses through infection‐induced nuclear translocation of the Rel protein DIF. However, in mutants for Dif, attacin induction is not impaired (Manfruelli et al., 1999; Meng et al., 1999; Rutschmann et al., 2000). As illustrated in Figure 3B, in adult fat body cells, bacteria‐induced translocation of DIF in 18w7‐35/Df(2R)017 mutants is similar to that observed in wild type. As described previously (Rutschmann et al., 2000), we find that in adult fat body cells, DIF translocation following immune challenge is controlled by the Toll pathway, since in spz mutants DIF cannot be targeted to the nuclei (data not shown).
18wheeler larvae exhibit an abnormal development of their fat body, which affects its immune function
We next analyzed the induction of antimicrobial peptides in third instar larvae lacking a functional copy of the 18w gene. We confirmed the earlier observation that in an 18w7‐35 homozygous context, attacin induction is reduced compared with wild‐type larvae (Williams et al., 1997). However, as illustrated in Figure 4, this reduction affected all antimicrobial peptides, including the antifungal peptide Drosomycin. We obtained similar results with 18w7‐35/Df(2R)017 hemizygous larvae.
Earlier studies from this laboratory had shown that the expression levels of the diptericin gene in bacteria‐challenged third instar larvae progressively increase during the instar (Reichhart et al., 1992), and that this increase is ecdysone dependent (Meister and Richards, 1996). We were unable to monitor a similar increase either in 18w7‐35 homozygous larvae or in 18w7‐35/Df(2R)017 hemizygous larvae. We reasoned that the low levels of inducibility of all antimicrobial peptide genes reflected developmental retardation of the fat body, and addressed this question with a fat body‐specific gene independent of the immune response, i.e. the larval fat body gene Fbp1. This gene encodes a receptor that mediates the uptake of hexamerins from the hemolymph (Brodu et al., 1999). It is exclusively transcribed in response to ecdysone in fat body cells during the second half of the third larval instar (Laval et al., 1993). We based our analysis on an Fbp1 promoter‐lacZ reporter transgene, which has an ecdysone‐responsive enhancer conferring a fat body‐specific ecdysteroid response (Laval et al., 1993). We observed that in wild‐type flies, expression of the transgene starts in third instar larvae 108 h after egg laying and reaches a maximum 12 h later, at the very end of the third instar. In contrast, we detected no reporter activity in 18w7‐35 larvae at similar time points and noted only a weak expression in 18w7‐35/Df(2R)017 larvae. These data confirm that 18w‐deficient larvae do not have a normal fat body physiology (Figure 5B). The result that the 18w7‐35/Df larvae do not express fbp1, but proceed to metamorphosis, suggests that ecdysone‐dependent signaling is perturbed only in the larval fat body. However, the fact that ultimately the 18w7‐35/Df adults are recovered in lower numbers than expected shows that the 18w7‐35 mutation influences viability. Taking in conjunction the facts that (i) expression of antimicrobial peptide genes is reduced, (ii) ecdysone‐dependent fat body maturation is affected, (iii) antimicrobial peptide gene expression maturation is ecdysone dependent and (iv) DIF is normally targeted to the nuclei of fat body cells in 18w7‐35 and 18w7‐35/Df(2R) larvae, we suggest that the reduction of peptide expression in these larvae is caused by a general problem of fat body maturation and is not due to impaired NF‐κB signaling.
The data presented in this study confirm the proposal made in 1997 that a functional 18wheeler gene is required for larvae to mount a wild‐type antimicrobial response (Williams et al., 1997). Rather than restricting this effect to the induction of attacin, our data show that the inducibility of all antimicrobial peptides is affected. As exemplified by the expression of the unrelated Fbp1 gene in mutant fat body, the data point to a general developmental delay of the fat body as a result of the mutation. A developmental role of 18W has already been postulated (Chiang and Beatchy, 1994; Eldon et al., 1994). It is an open question whether the effects that we observe in the fat body directly result from the absence of functional copies of the 18W protein in 18w7‐35 homozygous or 18w7‐35/Df(2R) hemizygous larvae. Indeed, the 18w7‐35‐encoded N‐terminal truncated TIR domain could have a dominant‐negative effect on a neighboring pathway by analogy with the effect of overexpression of the MyD88 TIR domain on interleukin‐1 receptor‐induced NF‐κB activation (Muzio et al., 1997). This would account for the less drastically affected development in 18w7‐35/Df(2R) hemizygous larvae, in which the concentration of the truncated protein is half that of 18w7‐35 homozygous larvae. The 18w7‐35 allele affects viability since 18w7‐35/Df(2R) flies are recovered in higher numbers than 18w7‐35 homozygous flies. This could designate 18w7‐35 as a neomorphic mutation indirectly affecting the larval immune response. Alternatively, other unrelated mutations on the 18w7‐35 chromosome, which lie outside of the deficiency used, could contribute to this difference.
We conclude that there is no stringent evidence to consider the transmembrane receptor 18Wheeler as a sensor of Gram‐negative infection. In fact, three independent studies point to a paramount role of the peptidoglycan recognition protein PGRP‐LC in sensing Gram‐negative infection (Choe et al., 2002; Gottar et al., 2002; Ramet et al., 2002). Data on the other receptors of the Toll family indicate that they do not signal to any of the bacterial peptide genes, and that only Toll‐5 and Toll‐9 are able to signal to the antifungal peptide gene drosomycin (Tauszig et al., 2000).
In the context of the valuable cross‐talk between the studies on Drosophila host defense and on mammalian innate immunity, it is relevant that we realize that we may not fully equate Toll signaling with that of the TLRs. The data available in Drosophila indicate that Toll is not activated by direct (or even indirect) interaction with microbial ligands, but rather responds to the cleavage product of the cytokine Spaetzle (Levashina et al., 1999). Cleavage of Spaetzle, in turn, depends on a proteolytic amplification cascade that is triggered when upstream proteins interact with microbial patterns in the hemolymph. This is in stark contrast to the situation described for the TLR family, whose members appear to directly interact with and discriminate between distinct microbial patterns. Furthermore, our data, taken in conjunction with those on other members of the Toll family, lead one to question whether Tolls are involved at all in the defense against Gram‐negative sepsis in Drosophila.
Fly strains and procedures.
See Supplementary data.
RNA preparation and northern blot analysis.
RNA was prepared and northern blots carried out as described previously (Lemaitre et al., 1996). The ribosomal protein 49 (rp49) probe was used as a loading control.
Mass spectrometry was performed as in Kussmann et al. (1997).
Flies or larvae immunized for 1 h or non‐immunized were dissected and stained as described previously (Rutschmann et al., 2000). Adult or larval fat body was incubated with a 1:500 dilution of a specific DIF rabbit antibody described previously (Rutschmann et al., 2000). Visualization of the nuclei was achieved with the use of a specific histone mouse antibody (Chemicon International Inc.). Secondary antibodies used were an anti‐rabbit coupled with the dye Alexa546 (red signal) and an anti‐mouse coupled with the dye Alexa448 (green signal), both at a dilution of 1:500 (Molecular Probes). Histochemical detection of β‐galactosidase activity was performed as previously described (Ashburner, 1989).
We thank Professor Jules Hoffmann for his continuous interest in our work; E. Eldon, D. Ferrandon, C. Antoniewski and the Bloomington Stock Center for fly stocks; M. Capovilla for help with confocal microscopy; and R. Klock and R. Syllas for technical help. Critical reading and comments by E. Eldon, D. Kimbrell and our colleagues from this laboratory are gratefully acknowledged. P.L. is supported by an EMBO Long Term Postdoctoral Fellowship. The LSM510 confocal microscope was co‐financed by the CNRS, the Université Louis Pasteur, the Région Alsace, the Association pour la Recherche contre le Cancer (ARC) and by the Ligue Nationale contre le Cancer. Research in this laboratory is supported by the CNRS, the NIH (1PO1 AI44220), Exelixis (South San Francisco) and Entomed (Strasbourg).
- Copyright © 2002 European Molecular Biology Organization