International Immunology 2005 17(1):1-14; doi:10.1093/intimm/dxh186
© 2005 The Japanese Society for Immunology
Toll-like receptors in innate immunity
Kiyoshi Takeda1 and
Shizuo Akira2,3
1 Department of Molecular Genetics, Medical Institute of Bioregulation, Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, 2 Department of Host Defense, Research Institute for Microbial Diseases, Osaka University and 3 ERATO, Japan Science and Technology Agency, 3-1 Yamada-oka, Suita, Osaka 565-0871, Japan
Corresponding author: S. Akira; E-mail: sakira{at}biken.osaka-u.ac.jp
 |
Abstract
|
|---|
Functional characterization of Toll-like receptors (TLRs) has
established that innate immunity is a skillful system that detects
invasion of microbial pathogens. Recognition of microbial components
by TLRs initiates signal transduction pathways, which triggers
expression of genes. These gene products control innate immune
responses and further instruct development of antigen-specific
acquired immunity. TLR signaling pathways are finely regulated
by TIR domain-containing adaptors, such as MyD88, TIRAP/Mal,
TRIF and TRAM. Differential utilization of these TIR domain-containing
adaptors provides specificity of individual TLR-mediated signaling
pathways. Several mechanisms have been elucidated that negatively
control TLR signaling pathways, and thereby prevent overactivation
of innate immunity leading to fatal immune disorders. The involvement
of TLR-mediated pathways in autoimmune and inflammatory diseases
has been proposed. Thus, TLR-mediated activation of innate immunity
controls not only host defense against pathogens but also immune
disorders.
Keywords: adaptor, innate immunity, signal transduction, TIR domain, Toll-like receptor
|
Introduction
|
|---|
Host defense against invading microbial pathogens is elicited
by the immune system, which consists of two components: innate
immunity and acquired immunity. Both components of immunity
recognize invading microorganisms as non-self, which triggers
immune responses to eliminate them. To date, both components
have been characterized independently, and the main research
interest in the immunology field has been confined to acquired
immunity. In acquired immunity, B and T lymphocytes utilize
antigen receptors such as immunoglobulins and T cell receptors
to recognize non-self. The mechanisms by which these antigen
receptors recognize foreign antigens have been intensively analyzed,
and the major mechanisms, such as diversity, clonality and memory,
have been well characterized. However, these receptors are present
only in vertebrates, and accordingly we do not fully understand
the mechanism for non-self recognition in less evolved organisms.
In addition, the innate immune system in mammals has not been
well studied. As a result, although mammalian innate immune
cells such as macrophages and dendritic cells are known to be
activated by microbial components (non-self) such as lipopolysaccharide
(LPS) from Gram-negative bacteria, a receptor responsible for
the recognition remained unknown.
At the end of the 20th century, Toll was shown to be an essential receptor for host defense against fungal infection in Drosophila, which only has innate immunity (1). One year later, a mammalian homolog of the Toll receptor (now termed TLR4) was shown to induce expression of genes involved in inflammatory responses (2). In addition, a point mutation in the Tlr4 gene has been identified in a mouse strain that is unresponsive to LPS (3). These studies have made innate immunity a very attractive subject of research, and in recent years there has been rapid progress in our understanding that the innate immune system possesses a skillful system that senses invasion of microbial pathogens by Toll-like receptors (TLRs). Furthermore, activation of innate immunity is a critical step to the development of antigen-specific acquired immunity. In this review, we will describe the mechanisms by which innate immunity is activated through TLRs.
|
Identification of the TLR family
|
|---|
After the characterization of the first mammalian TLR, TLR4,
several proteins that are structurally related to TLR4 were
identified and named Toll-like receptors (
4). Mammalian TLRs
comprise a large family consisting of at least 11 members. TLR1–9
are conserved between the human and mouse. However, although
TLR10 is presumably functional in the human, the C-terminal
half of the mouse
Tlr10 gene is substituted to an unrelated
and non-productive sequence, indicating that mouse TLR10 is
non-functional (our unpublished observation). Similarly, mouse
TLR11 is functional, but there is a stop codon in the human
TLR11 gene, which results in a lack of production of human TLR11
(
5).
The cytoplasmic portion of TLRs shows high similarity to that of the IL-1 receptor family, and is termed a Toll/IL-1 receptor (TIR) domain. Despite this similarity, the extracellular portions of both types of receptors are structurally unrelated. The IL-1 receptors possess an immunoglobulin-like domain, whereas TLRs bear leucine-rich repeats (LRRs) in the extracellular domain. Functionally, a critical role of TLR4 in the recognition of the microbial component LPS was initially characterized (3). Subsequently, it has been rapidly established that individual TLRs play important roles in recognizing specific microbial components derived from pathogens including bacteria, fungi, protozoa and viruses (Fig. 1).
View larger version (37K):
[in this window]
[in a new window]
|
Fig. 1. TLRs and their ligands. TLR2 is essential in the recognition of microbial lipopeptides. TLR1 and TLR6 cooperate with TLR2 to discriminate subtle differences between triacyl and diacyl lipopeptides, respectively. TLR4 is the receptor for LPS. TLR9 is essential in CpG DNA recognition. TLR3 is implicated in the recognition of viral dsRNA, whereas TLR7 and TLR8 are implicated in viral-derived ssRNA recognition. TLR5 recognizes flagellin. Thus, the TLR family members recognize specific patterns of microbial components.
|
|
|
TLR1, TLR2 and TLR6
|
|---|
TLR2 recognizes a variety of microbial components. These include
lipoproteins/lipopeptides from various pathogens, peptidoglycan
and lipoteichoic acid from Gram-positive bacteria, lipoarabinomannan
from mycobacteria, glycosylphosphatidylinositol anchors from
Trypanosoma cruzi, a phenol-soluble modulin from
Staphylococcus epidermis, zymosan from fungi and glycolipids from
Treponema maltophilum (
6). In addition, TLR2 reportedly recognizes LPS
preparations from non-enterobacteria such as
Leptospira interrogans, Porphyromonas gingivalis and
Helicobacter pyroli (
7–
9).
These LPS structurally differ from the typical LPS of Gram-negative
bacteria recognized by TLR4 in the number of acyl chains in
the lipid A component, which presumably confers differential
recognition (
10). However, a recent report indicates that LPS
preparation from
P. gingivalis contaminates lipoproteins that
activate TLR2, and LPS from
P. gingivalis only poorly activates
TLR4 (
11). Therefore, more careful analysis will be required
to conclude that some LPS are recognized by TLR2, but not TLR4.
There are two aspects proposed for mechanisms that could explain why TLR2 recognizes a wide spectrum of microbial components. The first explanation is that TLR2 forms heterophilic dimers with other TLRs such as TLR1 and TLR6, both of which are structurally related to TLR2. Macrophages from TLR6-deficient mice did not show any production of inflammatory cytokines in response to mycoplasma-derived diacyl lipopeptides. However, these cells showed normal production of inflammatory cytokines in response to triacyl lipopeptides derived from Gram-negative bacteria (12). In contrast, macrophages from TLR1-deficient mice showed a normal response to mycoplasma-derived diacyl lipopeptides, but an impaired response to triacyl lipopeptides (13). Thus, TLR1 and TLR6 functionally associate with TLR2 and discriminate between diacyl or triacyl lipopeptides. Moreover, the involvement of TLR1 in the recognition of the outer surface lipoprotein of Borrelia burgdorferi has also been shown (14). The second explanation involves recognition of fungal-derived components by TLR2 (15). In this model, TLR2 has been shown to functionally collaborate with distinct types of receptors such as dectin-1, a lectin family receptor for the fungal cell wall component ß-glucan. Thus, TLR2 recognizes a wide range of microbial products through functional cooperation with several proteins that are either structurally related or unrelated.
|
TLR3
|
|---|
Expression of human TLR3 in the double-stranded RNA (dsRNA)-non-responsive
cell line 293 confers enhanced activation of NF-
B in response
to dsRNA. In addition, TLR3-deficient mice are impaired in their
response to dsRNA (
16). dsRNA is produced by most viruses during
their replication and induces the synthesis of type I interferons
(IFN-
/ß), which exert anti-viral and immunostimulatory
activities. Thus, TLR3 is implicated in the recognition of dsRNA
and viruses. However, TLR3-independent mechanisms of dsRNA recognition
exist, as discussed below.
|
TLR4
|
|---|
As described above, TLR4 is an essential receptor for LPS recognition
(
3,
17). In addition, TLR4 is implicated in the recognition of
taxol, a diterpene purified from the bark of the western yew
(
Taxus brevifolia) (
18,
19). Furthermore, TLR4 has been shown
to be involved in the recognition of endogenous ligands, such
as heat shock proteins (HSP60 and HSP70), the extra domain A
of fibronectins, oligosaccharides of hyaluronic acid, heparan
sulfate and fibrinogen. However, all of these endogenous ligands
require very high concentrations to activate TLR4. In addition,
it has been shown that contamination of LPS in the HSP70 preparation
confers ability to activate TLR4 (
20). LPS is a very potent
immuno-activator, and accordingly, TLR4 can be activated by
a very small amount of LPS, contaminating these endogenous ligand
preparations. Therefore, more careful experiments will be required
before we can conclude that TLR4 recognizes these endogenous
ligands.
|
TLR5
|
|---|
Enforced expression of human TLR5 in CHO cells confers response
to flagellin, a monomeric constituent of bacterial flagella
(
21). TLR5 has further been shown to recognize an evolutionarily
conserved domain of flagellin through close physical interaction
between TLR5 and flagellin (
22). TLR5 is expressed on the basolateral,
but not the apical side of intestinal epithelial cells (
23).
TLR5 expression is also observed in the intestinal endothelial
cells of the subepithelial compartment (
24). In addition, flagellin
activates lung epithelial cells to induce inflammatory cytokine
production (
25). These findings indicate the important role
of TLR5 in microbial recognition at the mucosal surface. A common
stop codon polymorphism in the ligand-binding domain of TLR5
has been shown to be associated with susceptibility to pneumonia
caused by the flagellated bacterium
Legionella pneumophila (
25).
|
TLR7 and TLR8
|
|---|
TLR7 and TLR8 are structurally highly conserved proteins, and
recognize the same ligand in some cases. Analysis of TLR7-deficient
mice revealed that murine TLR7 recognizes synthetic compounds,
imidazoquinolines, which are clinically used for treatment of
genital warts associated with viral infection (
26). Human TLR7
and TLR8, but not murine TLR8, recognizes imidazoquinoline compounds
(
27). Murine TLR7 has also been shown to recognize another synthetic
compound, loxoribine, which has anti-viral and anti-tumor activities
(
28,
29). Both imidazoquinoline and loxoribine are structurally
related to guanosine nucleoside. Therefore, TLR7 and human TLR8
were predicted to recognize a nucleic acid-like structure of
the virus. This prediction has recently been shown to be true
from the finding that TLR7 and human TLR8 recognize guanosine-
or uridine-rich single-stranded RNA (ssRNA) from viruses such
as human immunodeficiency virus, vesicular stomatitis virus
and influenza virus (
30–
32). ssRNA is abundant in the
host, but usually the host-derived ssRNA is not detected by
TLR7 or TLR8. This might be due to the fact that TLR7 and TLR8
are expressed in the endosome, and host-derived ssRNA is not
delivered to the endosome.
|
TLR9
|
|---|
Analysis of TLR9-deficient mice revealed that TLR9 is a receptor
for CpG DNA (
33). Bacterial DNA contains unmethylated CpG motifs,
which confer its immunostimulatory activity. In vertebrates,
the frequency of CpG motifs is severely reduced and the cysteine
residues of CpG motifs are highly methylated, leading to abrogation
of the immunostimulatory activity. There are at least two types
of CpG DNA, termed A/D-type CpG DNA and B/K-type CpG DNA. B/K-type
CpG DNA is conventional, which was identified first, and is
a potent inducer of inflammatory cytokines such as IL-12 and
TNF-
. A/D-type CpG DNA is structurally different from conventional
CpG DNA and has a greater ability to induce IFN-
production
from plasmacytoid dendritic cells (PDC), but less ability to
induce IL-12 (
34,
35). TLR9 has been shown to be essential for
the recognition of both types of CpG DNA (
36). The fact that
TLR9 recognition of A/D-type CpG DNA leads to induction of an
anti-viral cytokine IFN-
in PDC indicates that TLR9 is involved
in viral recognition. Indeed, in addition to bacterial CpG DNA,
TLR9 has been shown to recognize viral-derived CpG DNA in PDC
(
37,
38). Furthermore, TLR9-mutant mice have been shown to be
susceptible to mouse cytomegalovirus (MCMV) infection (
39).
TLR9-dependent recognition of MCMV in PDC or other types of
DC elicits an anti-MCMV response through activation of NK cells
(
40). In addition to bacterial and viral CpG DNA, TLR9 is presumably
involved in pathogenesis of autoimmune disorders. Sequential
engagement of IgG2a–chromatin complex by the B cell receptor
and TLR9 mediates effective production of rheumatoid factor
by auto-reactive B cells (
41). In this model, the IgG2a is bound
and internalized by the B cell receptor, and the chromatin,
including hypomethylated CpG motifs, is then able to engage
TLR9, thereby inducing rheumatoid factor (
42). Similarly, internalization
by the Fc receptor and subsequent exposure of the chromatin
to TLR9 mediates PDC induction of IFN-
by immune complexes containing
IgG and chromatin, which are implicated in the pathogenesis
of systemic lupus erythematosus (SLE) (
43). Thus, TLR9 appears
to be involved in the pathogenesis of several autoimmune diseases
through recognition of the chromatin structure. Chloroquine
is clinically used for treatment of rheumatoid arthritis and
SLE, but its mechanisms are unknown. Since chloroquine also
blocks TLR9-dependent signaling through inhibition of the pH-dependent
maturation of endosomes by acting as a basic substance to neutralize
acidification in the vesicle (
44), it may act as an anti-inflammatory
agent by inhibiting TLR9-dependent immune responses.
|
TLR11
|
|---|
The most recently identified TLR11 has been shown to be expressed
in bladder epithelial cells and mediate resistance to infection
by uropathogenic bacteria in mouse (
5). TLR11-deficient mice
are highly susceptible to uropathogenic bacterial infection.
Although the ligand has not yet been identified, these findings
indicate that mouse TLR11 mediates anti-uropathogenic bacterial
response. As described above, there is no functional TLR11 protein
in the human (
5). These findings may indicate that the human
TLR11 protein was futile in the human environment and became
lost through evolution.
|
Subcellular localization of TLRs
|
|---|
Individual TLRs are differentially distributed within the cell.
TLR1, TLR2 and TLR4 are expressed on the cell surface, as demonstrated
by positive staining of the cell surface by specific antibodies.
In contrast, TLR3, TLR7, TLR8 and TLR9 have been shown to be
expressed in intracellular compartments such as endosomes (
29,
45–
47).
TLR3-, TLR7- or TLR9-mediated recognition of their ligands has
been shown to require endosomal maturation (
30–
32,
44,
46,
48).
The TLR9 ligand CpG DNA is first non-specifically captured into
endosomes, where TLR9 is recruited from the endoplasmic reticulum
upon non-specific uptake of CpG DNA (
44,
47,
49). Thus, it can
be hypothesized that in the case of bacterial infection, macrophages
and dendritic cells engulf bacteria by phagocytosis. CpG DNA
is then exposed after degradation of bacteria in phagosomes/lysosomes
or endosomes/lysosomes, where TLR9 is recruited or expressed.
In the case of viral infection, viruses invade cells by receptor-mediated
endocytosis, and the viral contents are exposed to the cytoplasm
by fusion of the viral membrane and the endosomal membrane.
Occasionally, the viral particles are degraded in the endosomal
compartment, which results in exposure of TLR ligands such as
dsRNA, ssRNA and CpG DNA. Even TLR2, which is expressed on the
cell surface, is recruited to the phagosomal compartment of
macrophages after exposure to zymosan (
50). Thus, phagosomal/lysosomal
or endosomal/lysosomal compartments may be the main sites for
TLR recognition of microbial components.
|
TLR-independent recognition of micro-organisms
|
|---|
TLR-independent recognition of viruses and dsRNA
Although TLR3 is involved in the recognition of viral-derived
dsRNA, the impairment observed in TLR3-deficient mice is only
partial (
16,
51). In addition, introduction of dsRNA into the
cytoplasm of dendritic cells leads to the induction of type
I IFNs via a mechanism partially dependent on dsRNA-dependent
protein kinase (PKR), but independent of TLR3 (
52). These findings
indicate that molecules responsible for TLR3-independent recognition
of dsRNA and viruses do exist. Although PKR is implicated in
dsRNA recognition, it is still controversial whether PKR plays
a critical role in dsRNA-induced type I IFN expression (
53).
Recently, a key molecule was identified, which mediates the
TLR3-independent dsRNA recognition (
Fig. 2). Retinoic acid-inducible
gene I (RIG-I), which encodes a DExD/H box RNA helicase containing
a caspase recruitment domain, was identified from the screening
of a cDNA library that augments dsRNA-dependent activation of
the IRF-3-dependent promoter. Studies with ectopic expression
and RNA interference (RNAi)-mediated knockdown of RIG-I clearly
demonstrated that RIG-I is critical in dsRNA- and viral infection-induced
type I IFN expression (
54). It is interesting to analyze the
correlation between TLR3 and RIG-I in the recognition of dsRNA
and viruses.
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 2. TLR-dependent and -independent recognition of microbial components. TLR2 has previously been shown to mediate peptidoglycan (PGN) recognition. However, NOD1 and NOD2 have recently been shown to recognize motifs found in the layer of PGN. It is possible that TLR2 recognizes lipoprotein contamination in the PGN layer. Viral recognition is also mediated by TLR-dependent and -independent mechanisms. TLR3-mediated recognition of viruses or dsRNA results in TRIF-dependent activation of IRF-3 and NF-B. However, viruses or dsRNA are recognized in a TLR3-independent manner, since the impairment of the responsiveness to viruses or dsRNA in TLR3-deficient mice is only partial. RIG-I is identified as a molecule that is responsible for viral recognition and that mediates activation of IRF-3.
|
|
NOD1 and NOD2
TLRs are membrane-bound molecules that recognize microbial components
on the surface or within extracellular compartments of cells.
Accordingly, intracellular recognition of bacteria appears to
involve a TLR-independent system. Recent accumulating evidence
indicates that the nucleotide-binding oligomerization domain
(NOD) family of proteins plays an important role in the recognition
of intracellular bacteria (
Fig. 2). Peptidoglycan (PGN) has
previously been shown to be recognized by TLR2 (
55). However,
PGN is a thick rigid layer that is composed of an overlapping
lattice of two sugars that are crosslinked by amino acid bridges,
and the exact structure of PGN that is recognized by TLR2 remains
unclear. NOD1 was originally identified as a molecule that is
structurally related to the apoptosis regulator, Apaf-1. It
contains a caspase-recruitment domain (CARD), a NOD domain and
a C-terminal LRR domain. Recent studies have demonstrated that
overexpression of NOD1 enables 293 cells to respond to preparations
of PGN (
56,
57). Characterization of the PGN motif detected by
NOD1 revealed that
-
D-glutamyl-meso diaminopimelic acid (iE-DAP)
is the minimal structure required for NOD1 detection. NOD2 was
identified as a molecule that shows structural similarity to
NOD1, but which possesses two CARD domains in its N-terminal
region. Similar to NOD1, expression of NOD2 confers responsiveness
to PGN in 293 cells. Biochemical analyses identified the essential
structure recognized by NOD2 as muramyl dipeptide MurNAc-
L-Ala-
D-isoGln
(MDP) derived from PGN (
58,
59). Thus, NOD1 and NOD2 recognize
different structures within PGN. MDP is found in almost all
bacteria, whereas iE-DAP is restricted to Gram-negative bacteria.
Therefore, NOD1 may play an important role in sensing Gram-negative
bacterial infection inside cells. Although TLR2 has been reported
to recognize PGN, it is possible that TLR2 recognizes lipoprotein/lipopeptide
contaminants that are trapped within the layers of the PGN mesh.
Mutations in the NOD2 gene have been shown to be associated with Crohn's disease, an inflammatory bowel disease of unknown etiology (60,61). These mutations are found in the LRR domain of NOD2, and result in defective NF-B activation. However, the mechanisms by which NOD2 mutations result in an increased susceptibility to Crohn's disease are unclear. One of the answers to this question has recently been demonstrated. In the absence of NOD2 or the presence of a Crohn's disease-like Nod2 mutation, TLR2-mediated activation of NF-B, especially the cRel subunit, has been shown to be enhanced, which explains enhanced NF-B activity and Th1 responses in Crohn's disease patients (62). NOD2 mutations are also associated with Blau syndrome, a disease characterized by granulomatous arthritis, uveitis and skin rash (63). The NOD2 mutations in Blau syndrome patients are located in the NOD domain, leading to an increase in NF-B activity. Thus, NOD2 is associated with certain human diseases.
Recognition of PGN motifs by NOD1 and NOD2 results in their oligomerization, which induces the recruitment of Rip2/RICK, a serine/threonine kinase (64). Rip2/RICK has a CARD domain in its C-terminal portion and an N-terminal catalytic domain that shares sequence similarity with Rip, a factor essential for NF-B activation through the TNF receptor. NODs and Rip2/RICK interact via their respective CARD domains, and induce recruitment of the IKK complex to the central region of Rip2/RICK. This in turn leads to activation of NF-B. Rip2/RICK-deficient mice have been shown to be highly sensitive to infection with the intracellular pathogen Listeria monocytogenes (65). Introduction of NOD1 or NOD2 into Rip2/RICK-deficient embryonic fibroblast cells does not induce NF-B activation (66). Thus, Rip2/RICK is essential for NOD1- and NOD2-mediated responses, although its involvement in the recognition of PGN motifs needs to be more precisely analyzed in Rip2/RICK-deficient mice.
|
Phagocytosis and TLRs
|
|---|
Phagocytosis is an important step for host defense against microbial
pathogens, since it triggers both degradation of pathogens and
subsequent presentation of pathogen-derived peptide antigen.
TLR recognition of pathogens leads to expression of genes such
as inflammatory cytokines and co-stimulatory molecules. Phagocytosis-mediated
antigen presentation together with TLR-dependent gene expression
of inflammatory cytokines and co-stimulatory molecules, instruct
development of antigen-specific acquired immunity (
Fig. 3).
Therefore, it is of interest to characterize the relationship
between phagocytosis and TLRs. In the absence of TLR2/TLR4 or
MyD88, a common adaptor in TLR signaling, phagocytosis of bacteria
including
Escherichia coli, Salmonella typhimurium and
Staphylococcus aureus has been shown to be impaired due to impaired phagosome
maturation (
67). Further studies indicate that TLR-mediated
MyD88-dependent activation of p38 is required for phagosome
maturation (
67,
68). Thus, TLRs are linked to phagocytosis of
bacteria.
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 3. Innate and adaptive immunity. Innate immune cells, such as dendritic cells and macrophages, engulf pathogens by phagocytosis, and present pathogen-derived peptide antigens to naïve T cells. In addition, TLRs recognize pathogen-derived components and induce expression of genes, such as co-stimulatory molecules and inflammatory cytokines. Phagocytosis-mediated antigen presentation, together with TLR-mediated expression of co-stimulatory molecules and inflammatory cytokines, instruct development of antigen-specific adaptive immunity, especially Th1 cells.
|
|
|
TLR signaling pathways
|
|---|
Stimulation of TLRs by microbial components triggers expression
of several genes that are involved in immune responses. The
molecular mechanisms by which TLRs induce gene expression are
now rapidly being elucidated through analyses of TLR-mediated
signaling pathways (
69). Microbial recognition of TLRs facilitates
dimerization of TLRs. TLR2 is shown to form a heterophilic dimer
with TLR1 or TLR6, but in other cases TLRs are believed to form
homodimers (
70). Dimerization of TLRs triggers activation of
signaling pathways, which originate from a cytoplasmic TIR domain.
In the signaling pathways downstream of the TIR domain, a TIR
domain-containing adaptor, MyD88, was first shown to be essential
for induction of inflammatory cytokines such as TNF-
and IL-12
through all TLRs (
21,
26,
71–
74). However, activation of
specific TLRs leads to slightly different patterns of gene expression
profiles. For example, activation of TLR3 and TLR4 signaling
pathways results in induction of type I interferons (IFNs),
but activation of TLR2- and TLR5-mediated pathways does not
(
75–
77). TLR7, TLR8 and TLR9 signaling pathways also lead
to induction of type I IFNs through mechanisms distinct from
TLR3/4-mediated induction (
36,
78). Thus, individual TLR signaling
pathways are divergent, although MyD88 is common to all TLRs.
It has also become clear that there are MyD88-dependent and
MyD88-independent pathways (
Fig. 4).
View larger version (39K):
[in this window]
[in a new window]
|
Fig. 4. TLR signaling pathway. TLR signaling pathways originate from the cytoplasmic TIR domain. A TIR domain-containing adaptor, MyD88, associates with the cytoplasmic TIR domain of TLRs, and recruits IRAK to the receptor upon ligand binding. IRAK then activates TRAF6, leading to the activation of the IB kinase (IKK) complex consisting of IKK, IKKß and NEMO/IKK. The IKK complex phosphorylates IB, resulting in nuclear translocation of NF-B which induces expression of inflammatory cytokines. TIRAP, a second TIR domain-containing adaptor, is involved in the MyD88-dependent signaling pathway via TLR2 and TLR4. In TLR3- and TLR4-mediated signaling pathways, activation of IRF-3 and induction of IFN-ß are observed in a MyD88-independent manner. A third TIR domain-containing adaptor, TRIF, is essential for the MyD88-independent pathway. Non-typical IKKs, IKKi/IKK and TBK1, mediate activation of IRF-3 downstream of TRIF. A fourth TIR domain-containing adaptor, TRAM, is specific to the TLR4-mediated MyD88-independent/TRIF-dependent pathway.
|
|
|
MyD88-dependent pathway
|
|---|
A MyD88-dependent pathway is analogous to signaling pathways
through the IL-1 receptors. MyD88, harboring a C-terminal TIR
domain and an N-terminal death domain, associates with the TIR
domain of TLRs. Upon stimulation, MyD88 recruits IRAK-4 to TLRs
through interaction of the death domains of both molecules,
and facilitates IRAK-4-mediated phosphorylation of IRAK-1. Activated
IRAK-1 then associates with TRAF6, leading to the activation
of two distinct signaling pathways. One pathway leads to activation
of AP-1 transcription factors through activation of MAP kinases.
Another pathway activates the TAK1/TAB complex, which enhances
activity of the I
B kinase (IKK) complex. Once activated, the
IKK complex induces phosphorylation and subsequent degradation
of I
B, which leads to nuclear translocation of transcription
factor NF-
B.
As its name suggests, in the MyD88-dependent pathway, MyD88 plays a crucial role. MyD88-deficient mice do not show production of inflammatory cytokines such as TNF- and IL-12p40 in response to all TLR ligands (21,26,71–74). Thus, MyD88 is essential for inflammatory cytokine production through all TLRs.
A database search for molecules that are structurally related to MyD88 led to identification of the second TIR domain-containing molecule TIRAP (TIR domain-containing adaptor protein)/Mal (MyD88-adaptor-like) (79,80). Similar to MyD88-deficient macrophages, TIRAP/Mal-deficient macrophages show impaired inflammatory cytokine production in response to TLR4 and TLR2 ligands (81,82). However, TIRAP/Mal-deficient mice are not impaired in their response to TLR3, TLR5, TLR7 and TLR9 ligands, Thus, TIRAP/Mal has been shown to be essential for the MyD88-dependent signaling pathway via TLR2 and TLR4.
|
MyD88-independent/TRIF-dependent pathway
|
|---|
In MyD88-deficient macrophages, TLR4 ligand-induced production
of inflammatory cytokines is not observed; however, activation
of NF-
B is observed with delayed kinetics (
72). This indicates
that although TLR4-mediated production of inflammatory cytokines
completely depends on the MyD88-dependent pathway, a MyD88-independent
component exists in TLR4 signaling. Subsequent studies have
demonstrated that TLR4 stimulation leads to activation of the
transcription factor IRF-3, as well as the late phase of NF-
B
activation in a MyD88-independent manner (
83). TLR4-induced
activation of IRF-3 leads to production of IFN-ß.
IFN-ß in turn activates Stat1 and induces several
IFN-inducible genes (
75–
77). Viral infection or dsRNA
was found to activate IRF-3 (
84). Accordingly, the TLR3-mediated
pathway also activates IRF-3 and thereby induces IFN-ß
in a MyD88-independent manner. Hence, TLR3 and TLR4 utilize
the MyD88-independent component to induce IFN-ß.
Characterization of MyD88 and TIRAP/Mal prompted us to hypothesize that TIR domain-containing molecules regulate the MyD88-independent pathway, and also facilitated the search for such molecules. A database search led to identification of a third TIR domain-containing adaptor, TIR domain-containing adaptor inducing IFN-ß (TRIF) (85). This molecule was identified as a TLR3-associated molecule by two-hybrid screening and was named TIR domain-containing adaptor molecule (TICAM-1) (86). The physiological role of TRIF/TICAM-1 was then demonstrated by generation of TRIF-mutant mice. TRIF-deficient mice generated by gene targeting showed no activation of IRF-3 and had impaired expression of IFN-ß- and IFN-inducible genes in response to TLR3 and TLR4 ligands (52). Another mouse strain mutated in the Trif gene generated by random germline mutagenesis also revealed that they were defective in TLR3- and TLR4-mediated induction of IFN-ß- and IFN-inducible genes (87). Thus, TRIF has been demonstrated to be essential for TLR3- and TLR4-mediated MyD88-independent pathways.
Database searches further led to identification of a fourth TIR domain-containing adaptor, TRIF-related adaptor molecules (TRAM)/TICAM-2 (88–91). Studies with TRAM-deficient mice and RNAi-mediated knockdown of TRAM expression showed that TRAM is involved in TLR4-mediated, but not TLR3-mediated, activation of IRF-3 and induction of IFN-ß- and IFN-inducible genes (88–90). Thus, TRAM is essential for the TLR4-mediated MyD88-independent/TRIF-dependent pathway.
In TRIF- and TRAM-deficient mice, inflammatory cytokine production induced by TLR2, TLR7 and TLR9 ligands was observed, as well as TLR4 ligand-induced phosphorylation of IRAK-1 (52,89). These findings indicate that the MyD88-dependent pathway is not impaired in these mice. However, TLR4 ligand-induced inflammatory cytokine production was not observed in TRIF- and TRAM-deficient mice. Therefore, activation of both the MyD88-dependent and MyD88-independent/TRIF-dependent components is required for the TLR4-induced inflammatory cytokine production, but the mechanisms are unknown.
Key molecules that mediate IRF-3 activation have been revealed to be non-canonical IKKs, TBK1 and IKKi/IKK (92). Introduction of TBK1 or IKKi/IKK, but not IKKß, resulted in phosphorylation and nuclear translocation of IRF-3. RNAi-mediated inhibition of TBK1 or IKKi/IKK expression led to impaired induction of IFN-ß in response to viruses and dsRNA (92,93). Embryonic fibroblast cells obtained from TBK1-deficient mice showed impaired activation of IRF-3 and expression of IFN-ß and IFN-inducible genes in response to TLR3 and TLR4 ligands (94–96). In contrast, embryonic fibroblast cells from IKKi/IKK-deficient mice were not defective in their response to TLR3 and TLR4 ligands (95). However, TLR3-mediated activation of IRF-3 and expression of IFN-ß and IFN-inducible genes were almost completely abolished in embryonic fibroblast cells lacking both TBK1 and IKKi/IKK. Thus, TBK1 and IKKi/IKK are critical regulators of IRF-3 activation in the MyD88-independent pathway.
The mechanisms by which the TRIF-dependent pathway leads to activation of NF-B and IRF-3 are now under investigation. The TIR domain of TRIF is located in the middle portion of this molecule, flanked by the N-terminal and C-terminal portions. Both N-terminal and C-terminal portions of TRIF mediate activation of the NF-B-dependent promoter, whereas only the N-terminal portion is involved in IFN-ß promoter activation (85). Accordingly, the N-terminal portion of TRIF was shown to associate with IKKi/IKK and TBK1, which mediate IRF-3-dependent IFN-ß induction (93,97). The N-terminal portion of TRIF was also shown to associate with TRAF6 (97,98). Since TRAF6 is critically involved in TLR-mediated NF-B activation (99), TRAF6 may regulate NF-B activation derived from the N-terminal portion of TRIF. The C-terminal portion of TRIF was shown to associate with RIP1 (100). Embryonic fibroblast cells from RIP1-deficient mice showed impaired NF-B activation in response to the TLR3 ligand. Thus, RIP1 is shown to be responsible for NF-B activation that originates from the C-terminal portion of TRIF.
|
Negative regulation of TLR signaling
|
|---|
Stimulation of TLRs by microbial components triggers the induction
of inflammatory cytokines such as TNF-
, IL-6 and IL-12. When
all these cytokines are produced in excess, they induce serious
systemic disorders with a high mortality rate in the host. It
is therefore not surprising that organisms have evolved mechanisms
for modulating their TLR-mediated responses.
Exposure to microbial components such as LPS results in a severely reduced response to a subsequent challenge by LPS. This phenomenon was first described over 50 years ago and is now called endotoxin (or LPS) tolerance, but the precise mechanisms remain unclear (101). The mechanisms are now being analyzed in the context of TLR signaling, and several models are proposed. LPS stimulation of macrophages results in reduced surface expression of the LPS receptor complex composed of TLR4 and MD-2, a co-factor that facilitates LPS binding (102,103). TLR2, TLR7 and TLR4 ligands induce reduced expression of IRAK-1 (104–106). Several other mechanisms are also shown to be involved in LPS tolerance (107).
In addition, molecules that negatively regulate TLR signaling have been identified. IRAK-M, a member of the IRAK family of serine/threonine kinases, is induced by TLR stimulation in monocyte/macrophages, and lacks kinase activity (108). IRAK-M-deficient mice show increased production of inflammatory cytokines in response to TLR ligands and defective induction of LPS tolerance (109). Inhibitory activity of IRAK-M seems to be elicited by IRAK-M prevention of IRAK-1/IRAK-4 dissociation from MyD88, thereby preventing formation of the IRAK-1–TRAF6 complex.
An alternatively spliced variant of MyD88 that lacks the intermediary domain of MyD88 (MyD88s) is induced in monocytes upon LPS stimulation. Overexpression of MyD88s results in impaired LPS-induced NF-B activation through inhibition of IRAK-4-mediated IRAK-1 phosphorylation (110).
SOCS1 is a member of the SOCS family of proteins that are induced by cytokines and that negatively regulate cytokine signaling pathways (111). In addition to cytokines, TLR ligands such as LPS and CpG DNA induced expression of SOCS1 in macrophages (112,113). SOCS1-deficient mice were hypersensitive to LPS-induced endotoxin shock and showed defective induction of LPS tolerance (114,115). Ectopic expression of SOCS1 resulted in impaired LPS-induced NF-B activation in macrophages. These findings indicate that SOCS1 directly down-modulates TLR signaling pathways, although the precise mechanism by which SOCS1 inhibits TLR signaling remains unclear.
Membrane-bound proteins harboring the TIR domain, such as SIGIRR (single immunoglobulin IL-1 receptor-related molecule) and T1/ST2, have also been shown to be involved in negative regulation of TLR signaling. In both SIGIRR- and T1/ST2-deficient mice, the LPS-induced inflammatory response was enhanced (116,117).
Ubiquitination-mediated degradation of TLRs is also proposed as a mechanism to inhibit activation of the signaling pathway. A RING finger protein, Triad3A, is shown to act as an E3 ubiquitin ligase and enhance ubiquitination and proteolytic degradation of TLR4 and TLR9 (118). Thus, several molecules are postulated to modulate TLR signaling pathways (Fig. 5). Combination of these negative regulators may finely coordinate the TLR signaling pathway to limit exaggerated innate responses causing harmful disorders.
View larger version (38K):
[in this window]
[in a new window]
|
Fig. 5. Negative regulation of TLR signaling pathways. TLR signaling pathways are negatively regulated by several molecules. IRAK-M inhibits dissociation of IRAK-1/IRAK-4 complex from the receptor. MyD88s blocks association of IRAK-4 with MyD88. SOCS1 is likely to associate with IRAK-1 and inhibits its activity. TRIAD3A induces ubiquitination-mediated degradation of TLR4 and TLR9. TIR domain-containing receptors SIGIRR and T1/ST2 are also shown to negatively modulate TLR signaling.
|
|
Top
Abstract
Introduction