The identification of the Drosophila melanogaster Toll pathway cascade and the subsequent characterization of TLRs have reshaped our understanding of the immune system. Ever since, Drosophila NF-κB signaling has been actively studied. In flies, the Toll receptors are essential for embryonic development and immunity. In total, nine Toll receptors are encoded in the Drosophila genome, including the Toll pathway receptor Toll. The induction of the Toll pathway by Gram-positive bacteria or fungi leads to the activation of cellular immunity as well as the systemic production of certain antimicrobial peptides. The Toll receptor is activated when the proteolytically cleaved ligand Spatzle binds to the receptor, eventually leading to the activation of the NF-κB factors Dorsal-related immunity factor or Dorsal. In this study, we review the current literature on the Toll pathway and compare the Drosophila and mammalian NF-κB pathways.
The Toll pathway was initially identified in a series of genetic screens for genes involved in early Drosophila embryonic development. These screens were based on the revolutionary saturation mutagenesis screen developed by C. Nüsslein-Volhard and E.F. Wieschaus, who identified 15 genes that control embryonic segmentation (1). This approach earned them, together with E.B. Lewis, the Nobel Prize in Medicine in 1995 (http://nobelprize.org/nobel_prizes/medicine/laureates/1995/). Subsequent genetic screens led to the discovery of genes important in the dorsal-ventral (DV) patterning of the embryo (i.e., the dorsal group of genes, including Toll, tube, pelle, cactus, the NF-κB homolog dorsal, and seven genes upstream of Toll) (2).
Because NF-κB was implied to be involved in mammalian immunity, and because the moth Hyalophora cecropia expresses an NF-κB–like immunoresponsive factor (3), it gradually became evident that parallels between the signaling pathways in Drosophila embryonic development and activation of the immune system exist (4). Hultmark and colleagues (5) first identified Toll (Toll-1) as an activator of the immune response in a Drosophila cell line in 1995. Around the same time, a human homolog of Toll was identified and mapped to chromosome 4p14 (6). Soon after this, a compelling in vivo study in Drosophila demonstrated that the DV regulatory gene cassette signaling from the Toll ligand spatzle to cactus is involved in the antifungal response in Drosophila (7). The first mammalian TLR was described 1 y later in 1997 (8). This was shortly followed by the characterization of five human TLRs (9) establishing the role of the Drosophila Toll pathway as an evolutionarily conserved signaling cascade. However, mammalian TLRs are believed to have no role in development (10), whereas the Drosophila Toll pathway is involved both in immunity (7) and developmental processes (2, 11, 12).
The Toll pathway in the immune response
The Drosophila immune system is composed of humoral and cellular components. A Gram-positive or fungal infection triggers the activation of the Toll pathway, which leads to the systemic production of antimicrobial peptides (AMPs) (13, 14). The antifungal peptide Drosomycin appears to be the principal target of the Toll humoral response. The Toll pathway also plays a role in the cellular immune response, which includes the phagocytosis of microbes, and the encapsulation and killing of parasites (15). Infecting Drosophila with the parasitic wasp Leptopilina boulardi activates a cellular immune response (16), which is manifested by increased production of circulating plasmatocytes, and the differentiation of a group of plasmatocytes into another specialized class of hemocyte, the lamellocyte. Lamellocytes participate in the encapsulation and killing of the parasite. Mutations in the gene cactus, a gain-of-function mutation in the Toll receptor gene, or the constitutive expression of dorsal can induce lamellocyte differentiation and cause the formation of melanotic tumor phenotypes (12, 17). Moreover, the Toll signaling pathway together with other pathways has been found to control hemocyte proliferation and hemocyte density (16, 18). In Drosophila larvae, Toll signaling is required for melanization (19). In gain-of-function Toll mutant flies, or cactus mutant flies that exhibit melanotic tumor phenotypes, the Toll-responsive NF-κB factor Dorsal is constitutively nuclear (17). However, this melanotic tumor phenotype is independent of Dorsal, suggesting a redundant role for Dorsal and the Dorsal-related immunity factor (Dif) in this context (17).
The cellular response can also affect the activation of the Toll pathway. Under normal conditions, Spn77Ba, a protease inhibitor of the serpin family, inhibits a phenol oxidase protease cascade. It was reported that tracheal melanization resulting from Spn77Ba disruption induces the systemic expression of the antifungal Drosomycin via the Toll pathway (20). Such signaling between local and systemic immune responses may be an alarm mechanism that prepares the host in case a pathogen breaches the epithelial barrier (20).
Drosophila Toll receptors
To date, nine genes encoding Toll-related receptors have been identified in the Drosophila genome. Toll, or Toll-1, was the first Toll identified and is responsible for AMP induction via the Toll pathway. All Drosophila Toll receptors share a similar molecular structure, with an ectodomain mainly composed of leucine-rich repeat and cystein-rich flanking motifs. Phylogenetically, Toll-5 is the closest relative to Toll (21). In contrast to other Tolls, Toll-9 has only one cystein-rich motif between the transmembrane domain and leucine-rich repeats, a structure very similar to mammalian TLRs (22). Drosophila Tolls and the IL-1Rs in mammals share a cytosolic homology domain called Toll/IL-1R (TIR) domain, which interacts with adaptor molecules, thereby activating downstream events (22).
As all mammalian TLRs are involved in the immune response, it is tempting to speculate the involvement of other Drosophila Tolls in immunity. Some Tolls could well play roles in immunological events; for example, Toll-5 may induce Drosomycin and Metchnikowin expression (21, 23, 24). In addition, Toll-5 has been shown to interact with the intracytoplasmic domains of Toll and Pelle, leading to the activation of Dorsal-dependent transcription in a synergistic manner with Toll (24). Also, Toll-9 has been reported to activate the constitutive expression of Drosomycin (25), and for this, Toll-9 may take advantage of the Toll signaling pathway components (26).
To activate the Drosophila Toll pathway either in development or in immunity, extracellular recognition factors initiate protease cascades leading to the activation of the Toll receptor ligand Spatzle ( or Spaetzle, Spätzle [Spz]) (27, 28). In nonsignaling conditions, the prodomain of Spz masks a predominantly hydrophobic C-terminal Spz region. Activation induces proteolysis, which causes a conformational change exposing determinants that are critical for binding of the Toll receptor (29). Interestingly, the prodomain remains associated with the C terminus and is only released when the Toll extracellular domain binds to the complex (30). Two models for the binding of Spz to Toll have been suggested, the first of which implies that one Spz dimer binds to two Toll receptors (31). In a newer model, two Spz dimers, each binding to the N terminus of one of the two Toll receptors, trigger a conformational change in the Tolls to activate downstream signaling (32) (Fig. 1).
Spz is synthesized and secreted as an inactive precursor consisting of a prodomain and a C-terminal region (C-106) (33). In DV patterning, Spz is processed into its active C-106 form by a serine protease cascade including Nudel, Gastrulation Defective, Snake, and Easter (34, 35). In addition, sulfotransferase Pipe is required independently of the protease cascade to activate Easter (36). In microbe recognition, Spz-processing enzyme (SPE) is responsible for Spz cleavage (37). The current model for activation of SPE contains three upstream cascades depending on the activating microorganism (Fig. 1). Two protease cascades leading to the activation of Gram-positive–specific serine protease (Grass) are initiated by cell wall components of both fungi (β-glucan) and Gram-positive bacteria (Lysine-type peptidoglycan) (38). Grass was originally identified to be specifically involved in the recognition of Gram-positive bacteria (39), but was later shown to be important also for the recognition of fungal components (38). In addition, four other serine proteases, namely spirit, spheroide, and sphinx1/2, were identified in response to both fungi and Gram-positive bacteria (39). Upstream of Grass, a modular serine protease (ModSP), conserved in insect immune reactions, plays an essential role in integrating signals from the recognition molecules Gram-negative binding protein (GNBP) 3 and PGN recognition protein (PGRP)-SA to the Grass-SPE-Spatzle cascade (40). A third protease cascade leading to the activation of SPE is mediated by the protease Persephone, which is proteolytically matured by the secreted fungal virulence factor PR1 (41) and Gram-positive bacterial virulence factors (38). Similar detection mechanisms have been suggested to occur in mammals, in which TLRs or Nod-like receptors directly detect virulence factors or endogenous proteins released by damaged cells (42, 43).
The recognition of the Gram-positive bacterial lysine-type peptidoglycan and/or the β-glucan from fungal cell walls is mediated by extracellular recognition factors. GNBP3 is responsible for yeast recognition (41). The other identified factors, namely GNBP1, PGRP-SA, and PGRP-SD, appear to mainly recognize Gram-positive bacteria. Upon Gram-positive bacterial recognition, PGRP-SA and GNBP1 physically interact and form a complex (44–46). Thereafter, activated GNBP1 hydrolyzes the Lys-type PGN and produces new glycan reducing ends, which are presented to PGRP-SA (47). In contrast, Buchon et al. (40) showed that full-length GNBP1 had no enzymatic activity. They suggested a role for GNBP1 as a linker between PGRP-SA and ModSP. PGRP-SD functions as a receptor for Gram-positive bacteria with partial redundancy to the PGRP-SA–GNBP1 complex (48). It appears that PGRP-SD can also recognize diaminopimelic acid (DAP)-type PGNs from Gram-negative bacteria, thereby activating the Toll pathway (49).
The core Toll signaling pathway
After binding the processed Spz, the activated Toll receptor binds to the adaptor protein MyD88 via intracellular TIR domains (50–52). Upon this interaction, MyD88, an adaptor protein, Tube, and the kinase Pelle are recruited to form a MyD88-Tube-Pelle heterotrimeric complex through death domain (DD)-mediated interactions (52–54). MyD88 and Pelle do not come into contact with each other; instead, two distinct DD surfaces in the adaptor protein Tube separately bind MyD88 and Pelle (52). Recently, a highly conserved Pelle/IL-1R–associated kinase (IRAK) interacting protein Pellino was shown to act as a positive regulator of Toll signaling (55). Drosophila Pellino mutants have impaired Drosomycin expression and reduced survival against Gram-positive bacteria (55). As all Pellinos contain a RING domain, it is tempting to speculate that Drosophila Pellino may ubiquitinate Pelle in a similar fashion as mammalian Pellinos polyubiquitinate IRAK1 (56).
From the oligomeric MyD88-Tube-Pelle complex, the signal proceeds to the phosphorylation and degradation of the Drosophila IκB factor Cactus. In nonsignaling conditions, Cactus is bound to the NF-κB transcription factor(s) Dorsal and/or Dif in a context-dependent manner, inhibiting their activity and nuclear localization. So, the nuclear translocation of both Dorsal and Dif requires Cactus degradation (57). To be degraded, Cactus needs to be phosphorylated, and although it has not been directly shown, it is possible that this is achieved by Pelle, because its kinase activity is required for Cactus phosphorylation (58). Also, in a recent screening (59) in which 476 dsRNA were targeted against all the known and predicted Drosophila kinases, Pelle was found to be the only kinase implicated in Cactus phosphorylation. Cactus needs to be phosphorylated in two distinct N-terminal motifs (60) that resemble IκB kinase (IKK) targets, yet the Drosophila IKK-β (Ird5) or IKK-γ (Kenny) are not involved in the Toll/Cactus pathway (61, 62). After phosphorylation, nuclear translocation of Dorsal/Dif leads to activation of the transcription of several sets of target genes. The Drosophila core Toll signaling pathway is shown in Fig. 2.
The Drosophila Dorsal is a Rel protein originally identified as an important morphogen in DV polarization. In larvae and adult Drosophila, Dorsal is expressed in the fatbody, and both its expression level (63) and nuclear localization (17) are enhanced upon microbial challenge. Dorsal interacts with Pelle, Tube, and Cactus (64–66), and, upon pathway activation, Dorsal translocates to the nucleus and binds to the κB-related sequence of AMP genes (63). Dorsal can activate the diptericin promoter in vitro (67), and, moreover, bacterial culture supernatants can stimulate nuclear translocation of Dorsal in vivo in dissected fatbodies in a hemolymph-dependent manner (19). Also, Dorsal activity is required to restrict the infectivity of Pseudomonas aeruginosa in adult Drosophila, providing evidence for Dorsal function in resistance against microorganisms (68).
Dif was identified in Drosophila as a dorsal-related immune responsive gene that does not participate in DV patterning. Instead, it mediates an immune response in Drosophila larvae (69) and interacts with Cactus in vitro (70). Dif (71), but not Dorsal (7), mediates Toll-dependent induction of the antifungal peptide gene Drosomycin in Drosophila adults, whereas Dorsal and Dif seem to be redundant in larvae (71, 72). Furthermore, Dif and Dorsal can form heterodimers in vitro (67), and in a Drosophila macrophage-like S2 cell line, Dorsal seems to play a more important role in Drosomycin promoter activation than does Dif (73).
RNA interference screening for new components of the Toll pathway
Drosophila cells are ideal for large-scale in vitro RNA interference (RNAi) screens (74, 75). Long dsRNA fragments up to several kilobases are readily internalized and processed by Drosophila S2 cells, which makes RNAi in S2 cells a very feasible tool for identifying genes involved in various processes (76, 77). In general, robust degradation of target RNA is obtained without a need for any transfection reagents. RNAi screening strategies have revealed several new important findings related to Drosophila Toll signaling. Recently, deformed epidermal autoregulatory factor-1 (DEAF-1), which was first identified as a transcription factor that binds to Metchnikowin and Drosomycin promoters (78), was confirmed to be required for full Drosomycin expression as well as for defending against fungal infections (79). Moreover, endocytic machinery components, including Myopic, were indicated to play a role in the endocytosis of the Toll receptor upon pathway activation (59).
In a recent genome-wide RNAi screen in S2 cells, G Protein-coupled receptor kinase 2 (Gprk2) was identified as a regulator of the Toll pathway (73). Gprk2 was found to be important against a Gram-positive bacterial infection as well as in Toll pathway-mediated hemocyte activation in Drosophila in vivo. Gprk2 interacts with Cactus in S2 cells, but is not involved in Cactus degradation, adding a new level of complexity to Drosophila Toll/Cactus regulation (73). Other genes identified in the screen include a Friend of GATA factor U-shaped and Toll activation mediating protein (TAMP; CG15737), with a previously unknown function. RNAi knockdown of ush or TAMP was shown to reduce the activity of the Drosomycin reporter in S2 cells in vitro as well as Drosomycin expression in vivo in infected flies. However, the molecular mechanisms of the effect of these components on the Toll pathway remain to be investigated (73).
Synergistic activation of the Drosophila immune-responsive pathways
It is clear that the Drosophila Toll pathway plays a key role in Gram-positive bacterial and fungal infections (80). In turn, Imd signaling is initiated by the PGRP-LC–mediated recognition of mainly a DAP-type PGN from Gram-negative bacteria (74, 81). Imd pathway activation ultimately leads to the activation of the NF-κB factor Relish (82–84), its translocation to the nucleus, and the transcriptional activation of a group of target genes including AMPs (13, 14) (Fig. 2). Although the Drosophila immunity pathways get selectively activated to a certain degree (85), synergistic interactions between the Toll and the Imd pathways have gradually become evident (73, 86–88). For example, although the bacterial branch of the Toll pathway is mainly activated by a Lys-type PGN, the crystal structure of the Toll pathway mediator PGRP-SD suggests binding to a DAP-type PGN rather than a Lys-type one (49). Moreover, in a Drosophila cell line, Relish RNAi reduces the expression of the Toll10b-induced Drosomycin reporter gene (86), and the Drosomycin reporter can be synergistically activated by Toll10b and Gram-negative bacteria (73). Furthermore, the expression of Drosomycin and Defensin are best induced by the Relish/Dif and the Relish/Dorsal heterodimers, respectively (89). In vivo, postinfection with Escherichia coli, the double mutants for Dif and the Imd pathway component kenny die earlier than kenny mutants (90). The same holds for the Relish,spz, and Relish,Toll double mutants compared with Relish mutants (88).
In addition to κB binding sites for Rel proteins, the transcriptional regulation of many Drosophila AMP genes depends on GATA binding sites in their promoter proximal regions (91). Drosophila has five GATA factors, namely Pannier (dGATAa), Serpent (dGATAb), Grain (dGATAc), dGATAd, and dGATAe. Pannier and a Friend of GATA factor U-shaped were recently identified as regulators of the Toll pathway in S2 cells (73). Serpent is the major GATA transcription factor in the larval fat body, and synergy between Relish and Serpent in the activation of the full immune response in larvae has been shown (92). Moreover, evidence is presented for dGATAe-mediated immune responses in the gut (93). It appears that, in most cases, Rel proteins and GATA factors act in concert to activate immune responses. Also, at least full Metchnikowin expression requires DEAF-1 (78).
Comparison of the Drosophila Toll and Imd pathways to mammalian TLR signaling
To date, 10 TLRs have been identified in humans and 12 in mice. The significance of TLRs was unknown until the mouse Tlr4 gene was identified as essential for LPS signaling (94). TLRs have since been shown to act as pattern recognition receptors for bacterial-, viral-, and self-derived products (reviewed in Ref. 95). When the signal is transduced, Tolls and TLRs associate with MyD88 via their intracytoplasmic TIR domains, activating the homologous protein kinases Pelle (in Drosophila) and IRAK (in mammals) (22). A recent study provides evidence for orthology between Tube and IRAK4 as well as Pelle and IRAK1 (96). In contrast, it has also been suggested that Drosophila Tube is at least functionally equivalent, and maybe distantly related in sequence, to the human TLR pathway adaptor protein MyD88 adaptor-like (97). In mammals, six MyD88, four IRAK4, and four IRAK2 DDs form a helical oligomer complex called Myddosome for downstream signaling (98). A similar three-component system, albeit with a different stoichiometry, is used in Drosophila: dimers of MyD88, Tube, and Pelle are needed for complex formation (54). In mammals, signal transmission downstream of MyD88 triggers the cooperation of several IRAKs, after which the IRAK complex interacts with TNFR-associated factor (TRAF) 6, which mediates the signal forward, via ubiquitination events, to the TGF-β–activated kinase 1 (TAK1) and TAK1 binding protein (TAB) complexes. TRAF homologs have been identified in the Drosophila genome, but they do not appear to participate in immune signaling (52, 86).
It appears that downstream from TAK1/TAB, the mammalian TLR pathway and the Drosophila Imd pathway, rather than the Toll pathway, share homologous components (95, 99). In mammals, the signal bifurcates at the level of a complex containing TAK1 and TABs, where one signal leads to the phosphorylation of the IKK complex and another via MAPKs to the activation of the JNK pathway and the eventual nuclear translocation of AP-1. The IKK complex phosphorylates IκB, leading to its ubiquitination and degradation. This results in the nuclear translocation of NF-κB factor(s) and the activation of transcription (95). Similarly, in the Drosophila Imd pathway, two signals from a complex containing Tak1, Tab2, and inhibitor of apoptosis 2 (86) are transmitted, one to the JNK pathway and one to the IKK complex, which phosphorylates the Rel protein Relish. After this, the caspase death-related Ced-3/Nedd2-like protein (DREDD) cleaves the C-terminal inhibitory domain of Relish (100). As was recently reported, DREDD is also involved in the cleavage of the Imd protein (101). The Drosophila Toll and Imd pathways are compared with related mammalian TLR pathways in Fig. 2.
Events downstream of MyD88 in the Drosophila Toll pathway appear somewhat different from the mammalian MyD88-dependent TLR pathways. The IKK complex is not involved in the phosphorylation and degradation of the IκB protein Cactus. However, conserved mechanisms in downstream parts of the Toll pathway and mammalian NF-κB signaling are evident. The Drosophila Gprk2 protein, which was shown to be involved in Toll pathway regulation and to interact with Cactus (73), is homologous to the murine GRK5, which was recently implicated in TNF-α–induced NF-κB signaling via direct interaction with IκB (102). Furthermore, the GRK5 knockout mice have attenuated LPS response, suggesting an evolutionarily conserved role for Gprk2/GRK5 (103).
In mammals, TLR7/TLR7, TLR9/TLR9, and TLR7/TLR8 act on the endosomal membrane also in a MyD88-dependent way to recognize nucleic acids from, for example, viruses (95, 104). Upon activation, the signal is propagated via several cytoplasmic IRAK proteins leading to the phosphorylation and nuclear translocation of IFN regulatory factor 7 (95). Interestingly, it was recently reported (59) that the Drosophila Mop (myopic) and Hrs (hepatocyte growth factor-regulated tyrosine kinase substrate), which are critical components of the endocytosis complex, colocalize with the Toll receptor in endosomes. Also, the Bro1 domain in Mop, which points to endosomal localization, is required for Toll signaling. So, it is plausible that endocytosis has an evolutionarily conserved role in Drosophila Toll and mammalian TLR signaling (59).
Since the initial discovery of the Toll pathway in fruit fly development 25 y ago, research in the field has firmly established the role of Toll signaling in immunity as well. In recent years, studies on microbe recognition and events upstream of Spz activation have revealed new components of the pathway. In addition, large-scale RNAi screens on the core intracellular pathway have revealed new essential components, putative conserved mechanisms, and cooperation of the fly immune pathways.
Mammalian TLR signaling mechanisms share similarities with the Drosophila Toll pathway, but also important differences exist; for example, the Toll receptor is a cytokine receptor, whereas TLRs are pattern recognition receptors. Also, among the nine Drosophila Tolls, a clear immunological role has only been assigned to Toll, whereas the others have putative roles in development. In contrast, all mammalian TLRs appear to have roles in immunity. Future work on the Drosophila Toll and other immune response pathways will undoubtedly continue to increase our understanding of these conserved NF-κB mechanisms in mammals.
We thank Dr. Helen Cooper for revising the language of the manuscript.
Disclosures The authors have no financial conflicts of interest.
This work was supported by grants from the Academy of Finland, the Foundation for Pediatric Research, the Sigrid Juselius Foundation, and the Emil Aaltonen Foundation (to M.R.), the Foundation of the Finnish Anti-Tuberculosis Association (to S.V.), the Tampere Tuberculosis Foundation, Competitive Research Funding of the Pirkanmaa Hospital District, and Biocenter Finland (to M.R. and S.V.).
Abbreviations used in this article:
deformed epidermal autoregulatory factor-1
dorsal-related immunity factor
death-related Ced-3/Nedd2-like protein
Gram-negative binding protein
G protein-coupled receptor kinase 2
Gram-positive–specific serine protease
modular serine protease
peptidoglycan recognition protein
TGF-β–activated kinase 1 binding protein
TGF-β–activated kinase 1