Constitutive Activation of Toll-Mediated Antifungal Defense in Serpin-Deficient Drosophila

See allHide authors and affiliations

Science  17 Sep 1999:
Vol. 285, Issue 5435, pp. 1917-1919
DOI: 10.1126/science.285.5435.1917


The antifungal defense of Drosophila is controlled by the spaetzle/Toll/cactus gene cassette. Here, a loss-of-function mutation in the gene encoding a blood serine protease inhibitor, Spn43Ac, was shown to lead to constitutive expression of the antifungal peptide drosomycin, and this effect was mediated by thespaetzle and Toll gene products. Spaetzle was cleaved by proteolytic enzymes to its active ligand form shortly after immune challenge, and cleaved Spaetzle was constitutively present inSpn43Ac-deficient flies. Hence, Spn43Acnegatively regulates the Toll signaling pathway, and Toll does not function as a pattern recognition receptor in the Drosophilahost defense.

Genetic analysis has established that the Toll signaling cascade controls the antifungal host defense of flies (1). In particular, Toll mediates the expression of the antifungal peptide drosomycin in the fat body cells by way of Rel-Cactus complexes, which are structurally and functionally equivalent to the vertebrate NF-κB–IκB complexes (1,2). Toll (Tl) was initially identified as a gene that controls dorsoventral patterning in theDrosophila embryo (3). The Tl gene encodes a transmembrane receptor with extracellular leucine-rich repeats and an intracellular domain exhibiting marked similarities with the cytoplasmic domain of the interleukin-1 receptor (4, 5). A proteolytically cleaved form of the spaetzle(spz) gene product is thought to be the extracellular ligand of Toll both in embryonic development and in the immune response (1, 5). In the embryo, a proteolytic cascade, involving the Gastrulation defective, Snake, and Easter proteases, cleaves Spaetzle, a cysteine-knot growth factor, cytokine-like polypeptide (5). The genes encoding these three proteases are dispensable for induction of a Toll-mediated immune response (1). A human homolog of Toll was recently cloned and shown to activate signal transduction by way of NF-κB, leading to the production of pro-inflammatory cytokines (6). Studies performed with human cell lines suggest that a lipopolysaccharide (LPS)-binding and signaling receptor complex is assembled at the cell membrane where human Toll, in association with the MD-2 protein (7), interacts with LPS bound to the peripheral membrane protein CD14 (8–10). The LPS signal is probably transduced across the membrane by Toll, as mutations in this gene in mice lead to an LPS-unresponsive state (11).

Here, we have addressed the activation of the Toll receptor during the immune response of Drosophila. For this, we have used flies carrying ethylmethane sulfonate–induced mutations in thenecrotic (nec) locus (12). The locus, which maps at position 43A, generates three transcripts encoding putative serine protease inhibitors of the serpin family (13). The nec mutants exhibit brown spots along the body and the leg joints, corresponding to necrotic areas in the epidermis. This mutant phenotype is rescued by a single transgenic copy of one of the serpin genes, Spn43Ac (13).

Because the absence of a functional Spn43Ac serpin may affect proteolytic cascades involved in the host defense ofDrosophila, we examined the level of expression of the antimicrobial peptide genes in nec mutants. All genes were induced 6 hours after challenge in wild-type (WT) flies; however, innec mutants the gene encoding drosomycin was strongly expressed in the absence of immune challenge (Fig. 1A). The expression was further enhanced by immune challenge. The gene encoding the peptide metchnikowin, which has both antibacterial and antifungal activities (14), also exhibited constitutive expression innec mutants, although the response was less marked than fordrosomycin. In contrast, we observed no constitutive expression of the genes encoding diptericin and cecropin A1, whose expression is either independent of the Toll signaling pathway or requires a signal from an additional pathway, depending on theimmune deficiency (imd) gene (1).

Figure 1

Expression of genes encoding antimicrobial peptides and serpin Spn43Ac in various mutant backgrounds inDrosophila adults (27). (A) Expression of drosomycin, diptericin, cecropin A1, metchnikowin genes, and rp49 loading control in unchallenged (−) and 6-hours immune-challenged (+) WT (Or R) and mutant (nec) flies. (B) The constitutive expression of drosomycin in unchallengednec mutants (lane 1) is abolished by introducing theSpn43Ac (lane 2), but not the Spn43Aa (lane 3), transgene into this mutant background (28). Flies were as follows: lane 1: nec1/nec2 ;P{w+ UAS-Spn43Ac +}/+; lane 2: nec1/nec2 ;P{w+pda-GAL4}/P{w+UAS-Spn43Ac +}; lane 3:nec1/nec2 ;P{w+pda-GAL4}/P{w+UAS-Spn43Aa +}. (C) The constitutive expression of drosomycin in unchallenged necmutants is abolished in Toll and spaetzleloss-of-function, but not in snake and gastrulation defective, mutants. Diptericin expression is induced only by immune challenge and is independent of the above mutations. The levels of drosomycin induction by bacterial challenge inTl and spz flies are markedly reduced but not abolished, in agreement with previous findings (1). PhosphoImager (Becton Dickinson) quantification of several independent Northern blots, corrected for rp 49 loading controls, show that in nec;Tl double mutants, the level of immune induction of drosomycinby bacterial challenge was between 10 and 23% that of WT values. Mutant flies were as follows: nec:nec1/nec2 , nec;spz :nec1/nec2 ;spzrm7/spzrm7 ,nec; Tl :nec1/nec2 ;Tlr632/Tl9QRE (29°C), nec; snk :nec1/nec2 ;snk073/snk073 ;gd , nec:gd8/gd8 ;nec1/nec2 . (D) The expression of the serpin gene Spn43Ac is up-regulated by 6-hours immune challenge (+) in WT flies (Or R) and in imdmutants, but not in Toll loss-of-function (Tl ) mutants. In Tollgain-of-function mutants (TlD ), expression of Spn43Ac was five times higher than in WT, as determined by phosphoimaging, taking into account the correction for the rp49 loading control. Flies were as follows: Tl :Tlr632/Tl9QRE (29°C); TlD :Tl10b /+; imd: imd/imd.

Overexpression of the Spn43Ac gene in nec flies abolished the constitutive expression of drosomycin, whereas overexpression of a different serpin gene from the same cluster,Spn43Aa, had no effect on this phenotype (Fig. 1B). In aTl or spz loss-of-function background, thenec-mediated constitutive expression ofdrosomycin was abolished (Fig. 1C), indicating that Spn43Ac acts upstream of spz and Tl. However, when thenec mutation was combined with gastrulation defective (gd) or snake(snk) loss-of-function mutations, constitutive expression of drosomycin was still observed (Fig. 1C), confirming that these proteases are not necessary for the Toll-controlled antifungal response. Furthermore, the constitutive expression ofdrosomycin was not affected when the nec mutation was in an imd mutant background (15), suggesting that the imd-mediated expression of the antibacterial peptide genes (1) is independent of the proteolytic cascade controlled by Spn43Ac.

The expression of the Tl gene and that of the downstream genes in the signaling cascade is up-regulated by immune challenge (1). We similarly found that the transcription of the Spn43Ac gene is up-regulated by immune challenge (Fig. 1D). This up-regulation is not observed in a Tlloss-of-function background. Conversely, Tl gain-of-function mutants exhibit a constitutive expression of Spn43Ac. Inimd mutants, the up-regulation of Spn43Ac by immune challenge is similar to that in wild-type flies. Thus,Spn43Ac is an immune-responsive gene, and its expression is under the positive control of the Toll pathway. This could represent a negative feedback mechanism to shut down the activation of Toll by inhibiting the upstream proteolytic cascade.

To function as a negative regulator of the Toll pathway upstream of Spaetzle and Toll, Spn43Ac should be present in the hemolymph of adult flies. Indeed, immunoblotting with an antiserum directed against recombinant Spn43Ac revealed a band of ∼60 kD in the blood of WT flies. This band was absent from the hemolymph of flies deficient for the Spn43Ac gene (Fig. 2A). The size of the mature Spn43Ac protein predicted from the cDNA sequence (13) is smaller (52 kD) than the size of the immunoreactive protein, possibly reflecting posttranslational modifications (because serpins are generally glycoproteins) (16). After immune challenge, a band of ∼50 kD was observed (Fig. 2A), which may correspond to the Spn43Ac serpin that had undergone cleavage by activated protease or proteases.

Figure 2

Protein immunoblot analysis demonstrating the presence of the Spn43Ac protein in the hemolymph of WT flies (A) and the cleavage of the Spaetzle protein in WT andnec mutants (B). (A) Hemolymph was extracted from control (−) and 6-hours bacteria challenged (+) WT (Or R) flies and from unchallenged (−) flies carrying a transheterozygous combination of deficiencies, Df(2R)sple-J1/Df(2R) nap2 (12), that uncovers the Spn43Ac gene (Df). Total hemolymph protein (5 μg) was probed with an antiserum directed against a GST-Spn43Ac fusion protein (26). Arrows 1 and 2 point to protein bands present in the hemolymph of WT flies and absent from flies deficient for theSpn43Ac gene. (B) Total protein extracts (40 μg) of embryos or adult flies (29) were probed with two independently raised antisera directed against the COOH-terminal part of the Spaetzle protein (18). Arrow (a) indicates band of ∼40 to 45 kD detected in unchallenged (−) WT flies (Or R), embryos (em), and nec flies. Two additional bands corresponding to proteins of 16 to 18 kD are observed in WT flies after bacterial challenge (arrows b) and in unchallenged (−) nec mutants. The asterisk denotes an irrelevant band that was recognized only by one of the two antisera. Molecular size markers are indicated on the left (in kilodaltons).

During dorsoventral patterning of the embryo, the 382-residue Spaetzle protein is cleaved to a 106-residue COOH-terminal active ligand form (17, 18). Experiments on the putative proteolytic cleavage of Spaetzle in the host defense have not been reported so far, and we therefore analyzed protein extracts from naı̈ve and immune-challenged flies by protein immunoblotting, using two polyclonal antisera directed against recombinant COOH-terminal Spaetzle (18). In extracts of embryos these antisera recognize the full-length Spaetzle protein and a smaller COOH-terminal fragment of 16 to 18 kD (Fig. 2B). In experiments with unchallenged flies, a band corresponding to a protein of 40 to 45 kD was detected in denatured extracts (Fig. 2B). It was also present in extracts of hemolymph (19). One hour after immune challenge, the 40- to 45-kD band had disappeared, whereas an immune-reactive doublet of ∼16 to 18 kD was apparent (Fig. 2B), which we assume to correspond to the processed form of Spaetzle protein. The Spaetzle protein has glycosylation sites, which may account for slightly larger molecular sizes than predicted from the cDNA sequences. The 16- to 18-kD doublet was detected in unchallenged nec flies, together with the 40- to 45-kD protein corresponding to uncleaved Spaetzle. This result is in agreement with our working hypothesis that in nec mutants the absence of the functional serpin leads to the constitutive cleavage of Spaetzle. Finally, the strong signal of the 40- to 45-kD form of Spaetzle together with that of the 16- to 18-kD form in necmutants confirms at the protein level that the expression of thespz gene is regulated by a positive-feedback loop.

Our data indicate that in the absence of a functional product of the Spn43Ac serpin gene in the blood of adult flies, the Spaetzle protein is spontaneously cleaved, leading to constitutive activation of the Toll signaling pathway. This phenotype can be rescued, either by a functional Spn43Ac transgene or by aspz- or Tl-deficient background. It is not known whether the protease, which cleaves Spaetzle, is a direct target of the serpin.

Conceptually, the activation of Spaetzle by blood protease zymogens is similar to the coagulation cascade in the horseshoe crab, which can be activated by binding of LPS to an upstream multidomain recognition protein (20). Several serpins, which fall into the same class as Spn43Ac, can specifically inhibit the proteases of the coagulation cascade.

Our results, and the parallels with the horseshoe crab coagulation cascade, imply that non-self recognition is an upstream event. Toll does not qualify as a bona fide pattern recognition receptor inDrosophila, in contrast to what has been proposed for Toll-like receptors in mammals (9, 10). The actual pattern recognition receptor, which initiates the cascade leading to the cleavage of Spaetzle and activation of Toll, remains to be identified.

Genetic aberrations and deficiencies of mammalian serpin genes have been correlated with clinical syndromes, such as pulmonary emphysema, angioedema, and coagulopathies, as a result of inappropriate inhibition of their respective target proteases (21). Our demonstration that a serpin functions in the regulation of the Drosophilaimmune response highlights the similarities between innate immunity in insects and mammals and reinforces the idea of a common ancestry of this system (8).

  • * These authors contributed equally to this study.

  • Present address: European Molecular Biology Laboratory, Meyerhofstrasse 1, Heidelberg 67117, Germany.

  • Present address: Wellcome/CRC Institute, Tennis Court Road, Cambridge CB2 1QR, UK.

  • § To whom correspondence should be addressed. E-mail: reichhart{at}


View Abstract

Navigate This Article