Requirement for a Peptidoglycan Recognition Protein (PGRP) in Relish Activation and Antibacterial Immune Responses in Drosophila

See allHide authors and affiliations

Science  12 Apr 2002:
Vol. 296, Issue 5566, pp. 359-362
DOI: 10.1126/science.1070216


Components of microbial cell walls are potent activators of innate immune responses in animals. For example, the mammalian TLR4 signaling pathway is activated by bacterial lipopolysaccharide and is required for resistance to infection by Gram-negative bacteria. Other components of microbial surfaces, such as peptidoglycan, are also potent activators of innate immune responses, but less is known about how those components activate host defense. Here we show that a peptidoglycan recognition protein, PGRP-LC, is absolutely required for the induction of antibacterial peptide genes in response to infection in Drosophila and acts by controlling activation of the NF-κB family transcription factor Relish.

In response to infection,Drosophila activates the transcription of a battery of antimicrobial peptide genes in cells of the fat body (the insect analog of the liver). Two major branches of this humoral response have been identified; as in mammals, these responses require NF-κB transcription factors (1). One branch activates antifungal responses and requires the receptor Toll and the NF-κB family transcription factor Dif (2–4). The second branch, which is primarily antibacterial, requires the NF-κB protein Relish, an IκB kinase (IKK), a caspase, a mitogen-activated protein kinase kinase kinase, and the death-domain protein Imd (5–11).

We have taken a genetic approach to identifying genes required for the antibacterial response (12, 13). One gene that is absolutely required for the induction of the antibacterial response isird7 (immune response deficient 7). Two mutations in ird7 identified in an ethylmethane sulfonate (EMS) mutagenesis screen (12, 13) prevented the induction of three antibacterial peptide genes, Diptericin,Cecropin, and Defensin, after infection by either Gram-negative or Gram-positive bacteria (Fig. 1, A and B). Three other antimicrobial peptide genes, Attacin, Metchnikowin, andDrosomycin, also failed to be induced to normal levels. The profile of antimicrobial gene expression observed in theird7 mutants was similar to that observed in imd,DmIkkβ/ird5, and Relish mutants after bacterial infection, but was distinct from that of Toll andDif mutants (Fig. 1A). This pattern suggests thatird7 is an essential component of the same signaling pathway that requires imd and Relish, but is not required for the Toll-Dif pathway. Both ird7 mutants are homozygous viable and fertile, and blood cells from ird7 mutants can phagocytose bacteria (14); these findings suggest thatird7 is required specifically for the humoral immune response.

Figure 1

Phenotypes of ird7 mutants. (A) In ird7 mutants, Diptericin(Dpt), CecropinA1 (CecA1),Defensin (Def), AttacinA(AttA), Metchnikowin (Mtk), andDrosomycin (Drs) transcription is not induced normally after E. coli infection, as assayed by Northern hybridization. ird71 is a very strong or null allele, whereas ird72 behaves like a strong hypomorph. RNA was prepared from adult flies 6 hours after infection as described (13). The loading control was Ribosomal protein49 (Rp49). Similar results were obtained in larvae (25). Genotypes: wt, wild type (the parental P{w+ Dpt-lacZ} ca stock);Df, Df(3L)29A6; imd,imd1 ; ird5/DmIkkβ,ird51 ; Dredd,DreddD55 ; Rel,RelishE20 ; Tl ,Df(3R)Tl9QRX/Df(3R)roXB3 ;Dif, Dif1 . For quantitation, see (17). (B) ird7 mutants fail to respond to both Gram-negative and Gram-positive bacteria. Adult flies were pricked with a sterile glass needle (wounding) or injected withMicrococcus luteus, Bacillus subtilis(Gram-positive), or Enterobacter cloacae (Gram-negative) and incu-bated for 6 hours, and total RNAs were prepared. Rp49was the loading control (25). The induction of other antibacterial peptide genes by these bacteria inird7 and imd mutants was also similar to that shown in (A) (25). (C) Relish is not endoproteolytically processed after infection in ird7mutants. Protein extracts from the wild-type parental stock (P[w+ Dpt-lacZ]ca),ird71 , and ird72 were prepared from uninfected (-) or infected (+) wandering third-instar larvae 30 min after E. coli injection (16). Protein from approximately 0.5 larva was loaded in each lane. After blotting, Relish processing was detected with a monoclonal antibody that recognizes the COOH-terminal ankyrin repeat domain of the protein. β-Tubulin was the loading control.

The transcription factor Relish directly activates antibacterial target genes in Drosophila. Relish is a compound protein similar to mammalian p100 and p105 (the precursors of the p52 and p50 subunits of NF-κB), with an NH2-terminal Rel homology and a COOH-terminal ankyrin repeat domain similar to that of the NF-κB inhibitor IκB (15). In response to immune challenge, full-length Relish (REL-110) is endoproteolytically clipped to generate the NH2-terminal REL-68 fragment, which translocates into the nucleus, and the COOH-terminal REL-49 ankyrin repeat fragment, which remains stable in the cytoplasm (16) (Fig. 1C). In contrast to wild-type animals, no processing of Relish was detected inird7 mutant larvae (Fig. 1C). The Rel domain of Relish failed to translocate to fat body nuclei in ird7 mutants (17). These results indicate that ird7 is required for Relish processing and nuclear translocation.

Recombination and deficiency mapping localized ird7 to a small interval on the third chromosome, 66F5-67A9 (Fig. 2A). The Drosophila genome sequence annotation indicates the presence of 12 genes in this region, including two genes encoding peptidoglycan recognition protein (PGRP) domains, PGRP-LA and PGRP-LC (18). Peptidoglycan is a strong activator of innate immune responses in insects and mammals, and a PGRP was first identified in a silk moth (Bombyx) on the basis of its ability to bind peptidoglycan and activate one aspect of the immune response, the prophenoloxidase cascade (19). Later studies have implicated PGRPs in innate immune responses from arthropods to mammals (20, 21).

Figure 2

Molecular identification of theird7 gene. (A) Genetic mapping ofird7. The ird7 mutation failed to complementDf(3L)29A6 but complementedDf(3L)Rdl-2 andDf(3L)AC1. Deficiency breakpoints were defined by single-embryo polymerase chain reaction (PCR) (26). P element–induced male recombination mapping (27) placed the ird7 locus betweenboule and EP(3)3043. Bars at bottom indicate the region that could include ird7. At all steps of mapping, X-Gal staining was used to monitor induction of Dpt-lacZafter E. coli infection. (B) Expression ofPGRP-LC in wild-type and ird7 mutants. Polyadenylated RNA (4 μg), prepared from wild-type (P[w+ Dpt-lacZ]ca) and ird7 adults, was loaded in each lane. Blots were hybridized with a radiolabeled probe from the second exon of PGRP-LC, which is common to both splice variants. α-Tubulin was the loading control. (C) Molecular lesions in PGRP-LC inird7 mutants. The ird71 allele is associated with an insertion of 858 bp in a common 5′ exon ofPGRP-LC that introduces a stop codon and would generate a truncated cytoplasmic protein of 105 amino acids. Theird72 is associated with a nonsense mutation in the x PGRP domain of the PGRP-LCx isoform, which would truncate this isoform. Light gray bars represent the transmembrane domain. Dark gray bars represent peptidoglycan recognition domains. For cloning of PGRP-LCx, a larval-pupal cDNA library (LP library from Berkeley Drosophila Genome Project) was screened using a random-primed probe for putative exonx (18).

We identified sequence changes that would disrupt the function ofPGRP-LC in both ird7 alleles. The gene was represented by several expressed sequence tag clones that encode a single splice form, designated PGRP-LCa. In addition, sequences encoding two additional exons encoding PGRP domains (“x” and “y”) were identified in an intron of PGRP-LC (18). We screened a larval-pupal cDNA library with the x and y exons and identified an alternatively spliced form of PGRP-LC that included the x exon; we call this isoformPGRP-LCx. Both PGRP-LC isoforms encoded type II transmembrane proteins with common NH2-terminal cytoplasmic and transmembrane domains but different extracellular domains. The extracellular PGRP domains of the two isoforms were only 38% identical (55 of 145 residues). Northern hybridization with a commonPGRP-LC exon probe revealed transcripts about 2.0 kb in size in wild-type larvae, but no transcript of that size inird71 mutant animals; instead, a larger transcript of lower abundance was detected (Fig. 2B). Sequence analysis revealed an insertion of 858 base pairs (bp) of single-copy sequence into exon 2, which is the first coding exon in both isoforms, in the ird71 allele (Fig. 2C). This insertion introduced a stop codon and would generate a truncated cytoplasmic protein. No sequence change in the PGRP-LCa isoform was identified in the ird72 allele. However, there was a G to A substitution in the x PGRP domain in thePGRP-LCx isoform of ird72 , which introduced a stop codon that makes a truncated protein lacking the last 107 amino acids of this isoform (Fig. 2C). Because theird72 allele alters only PGRP-LCx and has a profound effect on antimicrobial gene expression, this isoform must play a crucial role in vivo. The specific requirement for thePGRP-LCx isoform could be due to its ability to bind specific ligands or because its expression is limited to specific cell types by regulated RNA splicing. Overexpression of either of thePGRP-LC cDNAs rescued inducible expression of theDiptericin-lacZ reporter gene in homozygousird71 mutant animals (Fig. 3), confirming that the phenotype of ird7 mutants was the result of the lack of PGRP-LC activity.

Figure 3

Both PGRP-LCa andPGRP-LCx isoforms rescue induction of theDpt-lacZ reporter gene in ird7 mutants. Full-length PGRP-LCa and PGRP-LCx cDNAs were cloned into the pUAST (w+ ) transformation vector (28) and introduced into y w flies by P element–mediated transformation (29). The second chromosomec564-GAL4 line, which is expressed in the fat body and other tissues (30), was used to drive expression of the UAS construct. Flies of indicated genotypes were injected with E. coli, incubated for 6 hours, and assayed for β-galactosidase activity using X-Gal. (A) c564-GAL4/CyO; ird71 Dpt-lacZ/ird71 Dpt-lacZ (no UAS-cDNA) animals did not express the reporter gene. (B)UAS-PGRP-LCx/CyO; ird71 Dpt-lacZ/ird71Dpt-lacZ (no GAL4 driver) did not express the reporter gene. The same result was obtained for UAS-PGRP-LCa/CyO; ird71Dpt-lacZ/ird71 Dpt-lacZ animals. (C)c564-GAL4/UAS-PGRP-LCa; ird71Dpt-lacZ/ird71 Dpt-lacZ expressed the reporter gene at high levels after infection, as did c564-GAL4/UAS-PGRP-LCx; ird71 Dpt-lacZ/ird71 Dpt-lacZ animals (D). The GAL4-driven transgenes also showed a low level of constitutive expression of Dpt-lacZ withoutE. coli injection: (E)c564-GAL4/UAS-PGRP-LCa; ird71Dpt-lacZ/ird71 Dpt-lacZ. (F)c564-GAL4/UAS-PGRP-LCx; ird71Dpt-lacZ/ird71 Dpt-lacZ. In four repetitions of this experiment, the level of X-Gal staining in animals carrying bothc564-GAL4 and the UAS-PGRP-LC transgene was greater in infected than in uninfected animals.

We used RNA interference (RNAi) to test the role of PGRP-LCin the response to bacterial components. Treatment of blood cells from the mbn-2 line with peptidoglycan, Escherichia coli, or lipopolysaccharide (LPS) led to a robust induction of the antibacterial peptide genes. Introduction of double-stranded RNA (dsRNA) of PGRP-LC, but not PGRP-LA, effectively blocked induction of Diptericin, CecropinA1, andAttacinA in response to all three stimuli (Fig. 4). Thus, PGRP-LC is required for the response to both peptidoglycan and LPS in these cells.

Figure 4

Inactivation of PGRP-LC by transfection of dsRNA blocks induction of antibacterial gene expression in mbn-2 cells. Northern blot detection ofDiptericin, Cecropin A1, and Attacin Ain mbn-2 cells is shown after treatment with dsRNA fromPGRP-LC, PGRP-LA, or lacZ and induction with the indicated elicitors. Ethidium bromide staining of ribosomal RNA was used as a loading control. mbn-2 cells were plated at a density of 1 million cells/ml and transfected 1 day later with 10 μg of dsRNA (31). For PGRP-LA the dsRNA corresponded to 935 bp from exons 2 to 5; for PGRP-LCthe dsRNA corresponded to 861 bp from the common exons 2 and 3. Three days after transfection, the cells were induced with insoluble peptidoglycan from Micrococcus luteus for 6 hours, live E. coli (O55:B5) for 6 hours, LPS fromE. coli (O55:B5) for 2 hours, or sterile Ringer (-) as control. The pellet of an E. coli overnight culture was resuspended 1:100 in sterile Ringer, and 15 μl were used per induction. Peptidoglycan and LPS had a final concentration of 1 μg/ml. The cells were harvested after 2 or 6 hours, and total RNA was extracted. The loss of PGRP-LA and PGRP-LCmRNA due to RNAi was confirmed by reverse transcription PCR in a separate experiment. Drosomycin expression is not inducible in this mbn-2 cell line, so the effect of PGRP-LCRNAi on its expression could not be assessed in this experiment.

Because PGRP-LC is predicted to encode a transmembrane protein with an extracellular PGRP domain, PGRP-LC may act as a pattern recognition receptor that links recognition of microbial components with host immune responses (22). Because PGRP-LC is required for responses to both peptidoglycan and LPS, the extracellular domain of PGRP-LC may bind both peptidoglycan and LPS, and binding of either ligand may activate downstream signaling events. Alternatively, PGRP-LC may bind peptidoglycan (but not LPS) and may act as an essential subunit of a larger complex that includes other pattern recognition receptors that bind LPS. In mammals, signaling by Toll-like receptor 2 (TLR2) is activated by peptidoglycan (23). PGRP-LC might act in a complex with another transmembrane protein similar to TLR2.

Twelve PGRP genes have been identified in theDrosophila genome (18). AnotherDrosophila gene, PGRP-SA, encodes a soluble peptidoglycan recognition protein that is essential for activation of the Toll signaling pathway in response to infection by Gram-positive bacteria (21). Four PGRP genes have already been identified in the human genome (24). Given the evolutionary conservation of many proteins required for innate immune responses, it will be important to evaluate whether PGRPs function as a family of pattern recognition receptors in human innate immune responses.

  • * To whom correspondence should be addressed. E-mail: k-anderson{at}


View Abstract

Navigate This Article