Dual Activation of the Drosophila Toll Pathway by Two Pattern Recognition Receptors

See allHide authors and affiliations

Science  19 Dec 2003:
Vol. 302, Issue 5653, pp. 2126-2130
DOI: 10.1126/science.1085432


The Toll-dependent defense against Gram-positive bacterial infections in Drosophila is mediated through the peptidoglycan recognition protein SA (PGRP-SA). A mutation termed osiris disrupts the Gram-negative binding protein 1 (GNBP1) gene and leads to compromised survival of mutant flies after Gram-positive infections, but not after fungal or Gram-negative bacterial challenge. Our results demonstrate that GNBP1 and PGRP-SA can jointly activate the Toll pathway. The potential for a combination of distinct proteins to mediate detection of infectious nonself in the fly will refine the concept of pattern recognition in insects.

A hallmark of the host response of Drosophila to infection is the expression of antimicrobial peptide genes. Two intracellular signaling pathways regulate this expression. The Toll pathway is activated primarily by Gram-positive bacterial or fungal infections and, in particular, directs the expression of the antimicrobial peptide gene Drosomycin. In contrast, the immune deficiency (IMD) pathway responds predominantly to infections by Gram-negative bacteria and induces expression of several antibacterial peptide genes, including Diptericin (1). Fungal and Gram-positive infections lead to the proteolytic processing of a cytokine, the growth factor–like polypeptide Spätzle, which in its cleaved form binds directly to the transmembrane receptor Toll (28). This binding triggers an intracellular signaling cascade that culminates in the up-regulated transcription of multiple target genes, through the nuclear factor κB–related protein Dorsal-related immune factor (DIF) (2, 914).

How bacterial infection leads to processing of Spätzle and subsequent activation of Toll has remained unclear. In the course of a large-scale screen designed to isolate mutants of the innate immune response of Drosophila, we identified a new mutation that affected the inducibility of Drosomycin by Gram-positive bacterial infection. We named the mutation osiris (osi). Osi flies exhibited a significant decrease in the expression of the Drosomycin gene in response to several Gram-positive bacterial strains (Fig. 1A and fig. S1) (15). This was not observed in the majority of osi flies infected with fungi (Fig. 1C); Diptericin expression was normal after a challenge with Gram-negative bacteria (Fig. 1E). Although variable, data from these flies did not differ significantly from results obtained with wild-type flies (figs. S2 and S3). Survival rates of osi flies after injection of the Gram-positive bacteria Enterococcus faecalis (Fig. 1B) and Streptococcus pyogenes (16) were severely compromised and were similar to those of Dif loss-of-function mutants (6, 7). However, in contrast to these phenotypes, osi flies displayed levels of survival similar to those of wild-type flies in response to natural infection by the fungus Beauveria bassiana (Fig. 1D) and to infection by the Gram-negative bacteria Enterobacter cloacae (Fig. 1F) and Erwinia carotovora subsp. carotovora (16). Thus, the osi mutation primarily affected the defense pathway initiated against infection by Gram-positive bacteria rather than that against Gram-negative strains or fungi.

Fig. 1.

GNBP1 is required for the host response against Gram-positive bacteria and not for that against Gram-negative bacteria or fungi. (A and C) Representative values for Drosomycin mRNA steady-state levels as measured by quantitative–reverse transcription–polymerase chain reaction (Q-RT-PCR) (supporting online material). The panel shows the induction of Drosomycin mRNA after an immune-challenge with (A) Micrococcus luteus (for 24 hours, at 20°C) or (C) B. bassiana (for 36 hours, at 25°C) in GNBP1osi (GNBP1[osi]), PGRP-SAseml (PGRP-SA[seml]), and wild-type (wt) control flies. The level of inducibility measured in the wild type was arbitrarily set to 100, and the measures obtained in the mutants were expressed as a percentage of this value. (B, D, and F) Survival experiments. We monitored the survival rate (expressed in percentage) after immune challenge. As expected, Dif mutant flies died more rapidly than wild-type (w) controls after (B) Gram-positive (E. faecalis, at 25°C) and (D) natural fungal (B. bassiana, at 29°C) infections, whereas (F) IMD pathway mutants kenny (key) (25) and PGRP-LCE12 (26) succumbed rapidly to E. cloacae (at 25°C) infection. The Dif and key mutations were originally generated in the cn bw genetic background, which serves as a wild-type control. (E) Diptericin mRNA levels 6 hours after an immune challenge at 20°C with Escherichia coli in GNBP1osi, PGRP-SAseml, and wild-type flies.

The osi mutation corresponds to the insertion of a modified piggyBac transposon in the coding region of the Gram-negative binding protein 1 gene (GNBP1), a potential pattern recognition receptor (17). Members of the GNBP/β-(1,3)-glucan recognition proteins (βGRPs) were originally isolated from large-sized lepidopterans (1821) (supporting online text). The Drosophila genome encodes three canonical members of this family. GNBP1 has been reported to bind a commercial Gram-negative bacterial lipopolysaccharide, fungal β-(1,3)-glucan, but not to bind Gram-positive bacterial peptidoglycan (17). The transposon is inserted in the second exon of GNBP1 at nucleotide 511, which corresponds to amino acid 48 of the preprotein (or residue 29 of the predicted mature protein) (Fig. 2A). Transcripts of GNBP1 were barely detectable (Fig. 2B), suggesting that either the transcription of the locus or the stability of the mRNA are strongly affected by the insertion. No phenotypic difference was observed when the osi mutation was in homozygous or hemizygous conditions, indicating that this mutation is genetically null (16). The osi phenotype is due to the transposon insertion in GNBP1, because revertant lines, in which this transposon had been excised with a piggyBac transposase, were as resistant as wild-type flies to E. faecalis infection (Fig. 2C). Expression in osi flies of a pUAS-GNBP1 cDNA transgene driven by the ubiquitous heat-shock protein (hsp) 70–GAL4 construct rescued the inducibility of Drosomycin in response to Gram-positive bacterial infection, further demonstrating that the osi phenotype resulted from the insertion of the transposon (Fig. 2D). Because the osi mutant phenotype is due to the disruption of the GNBP1 gene, we designated the mutation as GNBP1osi.

Fig. 2.

The osiris insertion of a modified piggyBac transposon disrupts the GNBP1 gene. (A) Scheme of the GNBP1 region. The upper line shows the GNBP1 genomic region with exons drawn as boxes. The osiris (e03371) transposon insertion is shown on top. The lower part of the panel shows the schematic structure of the GNBP1 protein. The grey box indicates the signal sequence for amino acids 1 to 17; bold stripes show the β-(1,3)-glucan binding domain [amino acids 18 to 111; PFAM-B domain 4799 (27)]; thin stripes show the domain of homology to bacterial β-glucanases (amino acids 239 to 447, PFAM-A domain glyco-hydro16); the black box indicates the 10 C-terminal amino acids required for GPI anchoring (17). The asterisk shows the position where the open reading frame is disrupted by the insertion. (B) Steady-state GNBP1 mRNA levels measured by Q-RT-PCR in wild-type and GNBP1osi mutant flies, either unchallenged, 24 hours after a challenge with M. luteus, or 48 hours after a challenge with B. bassiana. (C) GNBP1osi transposon excision (exc) strains were challenged with E. faecalis; w, wild-type controls. (D) Rescue of the GNBP1osi phenotype by a pUAS-GNBP1 transgene after an immune challenge with M. luteus. Drosomycin and GNBP1 mRNA levels were quantified by Q-RT-PCR. Flies of the indicated genotype were submitted to a heat-shock procedure (HS). Nonheat-shocked flies (NHS) were used as a control. However, the lower yet significant expression of GNBP1 transcripts, due to a leaky expression of the hsp promoter, was sufficient for the rescue of Drosomycin inducibility in the (b) control flies. The GNBP1 steady-state transcript levels measured after heat-shock [in the (c) flies] were arbitrarily set to 100%; in this case, wild-type levels reached about 1% of this value.

Fungal-dependent Toll activation requires the persephone (psh) gene, which encodes a serine protease and whose overexpression leads to challenge-independent Toll activation and Drosomycin gene expression (5). This effect was not abolished when psh was overexpressed in a GNBP1osi genetic background (Fig. 3A), indicating that psh-dependent activation of Toll does not require GNBP1.

Fig. 3.

GNBP1 function is required upstream of Toll in the blood compartment. (A) GNBP1 is not epistatic to psh. We overexpressed the PSH protease using the GAL4/UAS system with a yolk protein 1 promoter that drives the expression of GAL4 specifically in the female fat body. Challenge-independent Drosomycin expression was detected in females, whether (a) heterozygous or (c) homozygous for the GNBP1osi mutation. No expression was detected in (b) or (d) males. Challenged wild-type flies are shown. Similar results were obtained with an hsp-GAL4 driver. M. l., M. luteus; Unc., unchallenged. (B) Hemolymph transfer experiments. The transfer of wild-type hemolymph into a GNBP1osi recipient (f) restores the M. luteus inducibility of Drosomycin as determined by Q-RT-PCR. In contrast, transfer of GNBP1osi hemolymph into a GNBP1osi recipient does not significantly rescue the GNBP1osi phenotype (g). Results obtained with PGRP-SAseml (i) were similar to those previously reported (6). Drosomycin transcript levels in challenged GNBP1osi (e) and PGRP-SAseml (h) mutants are shown for reference. Hemolymph collected from PGRP-SAseml flies could rescue the GNBP1osi mutant phenotype; the converse experiment, transfer of GNBP1osi hemolymph into PGRP-SAseml recipients, yielded the same result (16).

GNBP1 encodes a 55-kD protein that has been shown in cell-culture studies to exist as both secreted and glycosylphosphatidylinositol (GPI)–linked form in the cytoplasmic membrane (17). Transfer of hemolymph from wild-type flies to GNBP1osi mutants restored the inducibility of Drosomycin by Gram-positive bacterial challenge (Fig. 3B), indicating that GNBP1-dependent activity within the hemolymph was sufficient to rescue the GNBP1osi mutant phenotype. The membrane-tethered form of GNBP1 appears, therefore, to be dispensable in the induction of Drosomycin.

The phenotypes described here for the GNBP1osi mutation are very similar to those observed with the semmelweis (seml) loss-of-function mutants of a gene that encodes a member of an unrelated protein family, peptidoglycan recognition protein PGRP-SA (Figs. 1, 2C, and 3B) (6). We generated PGRP-SAseml;GNBP1osi double-mutant flies and found that they were not more sensitive than the single mutants in survival assays with Gram-positive bacteria. Similarly, the reduction of the levels of induced Drosomycin by various Gram-positive bacterial strains was equivalent in the double and single mutants (Fig. 4, A to C). These data indicate that GNBP1 and PGRP-SA are involved in a common process. Indeed, the concomitant overexpression of both genes (but not of the single genes) did lead to a significant, challenge-independent, expression of Drosomycin (Fig. 4D). As expected, this expression required the spätzle (spz) gene product (Fig. 4E), demonstrating that both GNBP1 and PGRP-SA act upstream of spz. These data provide genetic evidence to suggest that GNBP1 and PGRP-SA may interact to trigger the activation of the Toll pathway. To provide biochemical support for this interaction, we separated protein complexes from fly extracts by electrophoresis under native conditions and performed Western blot analysis for PGRP-SA and GNBP1. In extracts derived from wild-type flies, GNBP1 (17) and PGRP-SA antibodies recognized a band at exactly the same position (Fig. 4F). Because both proteins have markedly different sizes and amino acid compositions, this suggested comigration of both proteins as part of a complex. In flies mutant for GNBP1 or PGRP-SA, this band was either absent (GNBP1osi) or strongly reduced (PGRP-SAseml). Importantly, seml represents a point mutation that is assumed to affect the folding of the protein (22). Thus, both GNBP1 and PGRP-SA are required for the formation of the detected band. Finally, concomitant overexpression of the GNBP1 and PGRP-SA genes resulted in the appearance of an additional intense band detected by both antibodies, which may have corresponded to an activated complex.

Fig. 4.

The interaction of GNBP1 and PGRP-SA leads to Toll pathway activation. (A) Survival of PGRP-SAseml;GNBP1osi mutant flies after E. faecalis infection. The survival assay was performed as in Fig. 1. w, wild-type; DD1, wild-type reference stock (15). (B and C) Expression of Drosomycin in PGRP-SAseml;GNBP1osi double mutants 24 hours after (B) M. luteus or (C) E. faecalis challenge. (D) The concomitant overexpression of PGRP-SA and GNBP1 leads to the challenge-independent expression of Drosomycin. Drosomycin steady-state transcript levels were measured as in (B) and (C). The value measured in wild-type samples (wt M. l.) 24 hours after an immune challenge was arbitrarily set to 100%. The GAL4/UAS system was used to overexpress GNBP1 and/or PGRP-SA with an hsp-GAL4 driver. The third-chromosome balancer MKRS allows discrimination between UAS-PGRP-SA and the wild-type third chromosome. Because we used a UAS-GNBP1 transgene inserted on the X chromosome, flies that did not carry this transgene were males (sets a and b), and flies that overexpressed GNBP1 were females (sets c and d). Significant Drosomycin expression was also observed when the ubiquitous daughterless-GAL4 driver was used. (E) Epistatic analysis between the GNBP1–PGRP-SA and spätzle genes. GNBP1 and PGRP-SA were overexpressed in a spätzle (spz/spz) homozygous mutant background (set e). TM6b (set f) is another balancer of the third chromosome and carries a wild-type copy of the spz gene. (F) Western blot of whole fly extracts that were separated by gel electrophoresis under native conditions. The blot was first incubated with an antiserum raised against a PGRP-SA peptide (right), stripped, and then reprobed with an antibody against GNBP1 (left). The specific band recognized by both antibodies (open triangles) is absent in GNBP1osi and PGRP-SAseml mutants (6). The lower band observed with the antibody to GNBP1 (arrow) is not specific, because it is observed in the GNBP1osi null mutant. (G) Scheme of Toll pathway activation by Gram-positive bacteria and fungi. NEC, necrotic serine protease inhibitor; TL, Toll receptor.

Our results demonstrate that GNBP1 is required specifically for the Toll-dependent response to Gram-positive bacterial infections, but not for responses to fungi or Gram-negative bacteria (supporting online text) (Fig. 4G). Taken in conjunction with data on the PGRP-SAseml mutation (6), our results are compatible with a model in which the GNBP1 and PGRP-SA proteins form a complex that activates a proteolytic cascade culminating in the cleavage of Spätzle and the activation of Toll. The x-ray structure of Drosophila PGRP-LB reveals the existence of a conserved surface at the back of PGRP proteins that could serve for protein-protein interactions (22). We assume that the complex would be activated on detection of microbial elicitors during infection. In this context, it may be relevant that a GNBP/βGRP from the coleopteran Tenebrio molitor has been reported to be associated with a serine protease (23). Together, our data imply that Toll-dependent activation by Gram-positive bacteria requires the cooperation of at least two distinct pattern recognition proteins, although the precise mechanisms of recognition and activation of these proteins during infection remain to be established. Recognition of infection in insects appears more complex than the initial model of microbial pattern recognition has postulated (24).

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S3

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article