Dong Hyun Song

Jie-Oh Lee

Sensing of microbial molecular

patterns by Toll-like receptors

Authors addresses
Dong Hyun Song1, Jie-Oh Lee1,2
1
Department of Chemistry, KAIST, Daejon, Korea.
2
Graduate School of Nanoscience & Technology (WCU),
KAIST, Daejon, Korea.

Summary: Toll-like receptors (TLRs) sense structural patterns in microbial molecules and initiate immune defense mechanisms. The structures
of many extracellular and intracellular domains of TLRs have been
studied in the last 10 years. These structures reveal the extraordinary
diversity of TLR-ligand interactions. Some TLRs use internal hydrophobic pockets to bind bacterial ligands and others use solvent-exposed
surfaces to bind hydrophilic ligands. The structures suggest a common
activation mechanism for TLRs: ligand binding to extracellular domains
induces dimerization of the intracellular domains and so activates intracellular signaling pathways. Recently, the structure of the death domain
complex of one of the signaling adapters, myeloid differentiation factor
88 (MyD88), has been determined. This structure shows how aggregation of signaling adapters recruits downstream kinases. However, we
are still far from a complete understanding of TLR activation. We need
to study the structures of TLR710 in complex with their ligands. We
also need to determine the structures of TLR-adapter aggregates to
understand activation mechanisms and the specificity of the signaling
pathways. Ultimately, we will have to study the structures of the complete TLR signaling complexes containing full-length receptors, ligands,
signaling, and bridging adapters, and some of the downstream kinases
to understand how TLRs sense microbial infections and activate
immune responses against them.

Correspondence to:
Jie-Oh Lee
Department of Chemistry, Graduate School of Nanoscience
& Technology (WCU), KAIST
Daejon, Korea
Tel.: +82 42 350 2839
Fax: +82 42 350 2810
e-mail: jieoh@kaist.ac.kr
Acknowledgement
We gratefully acknowledge the support of the World Class
University Program (R3110071) of the Ministry of
Education, Science and Technology of Korea. The authors
have no conflicts of interest to declare.

This article is part of a series of reviews

covering Insights from Structure appearing
in Volume 250 of Immunological Reviews.

Keywords: Toll-like receptor, innate immunity, pattern recognition, MyD88

Toll-like receptors

Immunological Reviews 2012

Vol. 250: 216229
Printed in Singapore. All rights reserved

2012 John Wiley & Sons A/S

Immunological Reviews
0105-2896

216

Toll-like receptors (TLRs) provide efficient and immediate

immune responses to bacterial, fungal, and viral infections
by recognizing diverse molecules released from them.
Because they recognize common patterns in microbial molecules, they are often referred to as pattern recognition
receptors (1, 2). Humans have 10 TLRs, and each of them
binds a distinct family of microbial molecules (3, 4). TLR2
forms a heterodimer with either TLR1 or TLR6 and binds
bacterial lipoproteins. TLR4 in complex with MD-2 recognizes the lipopolysaccharide (LPS) that is a major component of the outer membrane of Gram-negative bacteria.
TLR5 can be stimulated by a depolymerized form of flagellin. TLR3 and TLR79 respond to various forms of bacterial
or viral nucleic acid. The engagement of TLRs with ligand
2012 John Wiley & Sons A/S
Immunological Reviews 250/2012

Song & Lee
Structure of Toll-like receptors

Overview of the TLR activation mechanism

Like other single transmembrane receptors, TLRs are activated by ligand-induced multimerization (Fig. 1). The structures of the key intermediates in these activation processes
have been mainly studied by x-ray crystallography (1321).
The extracellular domain of each TLR has the characteristic
horseshoe-like structure of the leucine-rich repeat (LRR)
family of proteins (13, 14, 22, 23). The concave surface
consists of a large central b sheet, and the convex surface is
decorated by a combination of helices and irregular loops
(Fig. 2). Binding of agonistic ligands to TLRs induces dimerization of their extracellular domains. We now have structures for the extracellular domains of six of the 10 human
TLRs in the ligand-induced dimeric state. Although the
ligand recognition mechanisms of the TLRs vary substantially, all the TLR dimers have a similar overall arrangement:
the C-termini of the two extracellular domains converge in
the middle, separated by distances ranging from 20 to 40
angstrom. This structural observation has led to the proposal
that ligand binding to the extracellular domains enhances

LPS
Binding

MyD88

N
x

Concave

xL

Descending lateral

Hydrophobic
core

Extracellular
LRR domain
LRRNT

LRRCT

Transmembrane
helix

Intracellular
TIR domain

Fig. 2. Overall structure of a Toll-like receptor (TLR). The

extracellular and intracellular domains of TLR4 are drawn as a blue
surface. A Ca trace of one of the LRR modules and the side chains of
the conserved leucines and asparagine is shown. The LRRNT and
LRRCT modules are colored pink and green, respectively.

dimerization of the intracellular domains because the latter

are connected to the C-termini of the extracellular domains
by transmembrane helices (15). The intracellular domains of
TLRs, as well as of several signaling adapters, share a
domain named the Toll/interleukin-1 receptor (TIR)
domain. Its structure is composed of a parallel b sheet surrounded by a helices (2429). Ligand-induced aggregation
of the receptors and adapters is mediated by homotypic and
heterotypic interaction between these TIR domains.
Although the structures of quite a few monomeric TIRs have
already been reported, the structures of their homo- and
heteromultimers remain unsolved.
All these structural studies have been conducted with
truncated extracellular or intracellular domains because the
full-length receptors including the transmembrane helices
are difficult to produce and crystallize. Therefore, the structural information of the full-length TLRs under membrane
embedded condition is not available. There is evidence that

TLR signaling complex

LPS

TLR_TIR

Convex
Ascending lateral

molecules induces dimerization of the receptors, which

leads to recruitment of intracellular signaling adapters and
kinases. The activation of TLR signaling pathways eventually
results in the synthesis of pro-inflammatory cytokines and
therefore prepares the entire immune system for infection.
TLRs are important targets for drug design because excess
activity of TLRs is involved in various inflammatory diseases
(5). Antagonists of TLR4 are being developed as targets of
anti-sepsis drugs (69); agonists of TLR7 and TLR8 are
being used as anti-viral and anti-cancer drugs, and many
more are in clinical trials (5). A partial agonist of TLR4 has
recently been approved as a general vaccine adjuvant, and
agonists targeted to several TLRs are being developed for
this purpose (1012).

TLR4-MD-2
Dimerization
Adaptor-Kinase
recruitment

MAL
IRAK4

Receptor
clustering (?)

IRAK2

Fig. 1. Overall mechanism of TLR4 activation. The plasma membrane is shown schematically by pink horizontal bars. The TLR4 extracellular
domains are shown as black arcs and the MD-2s as green balls.
2012 John Wiley & Sons A/S
Immunological Reviews 250/2012

217

Song & Lee
Structure of Toll-like receptors

activated and dimerized TLRs form higher order clusters in

the cell membrane (25, 30). However, structures of these
proposed clusters have never been characterized in detail
because of difficulties reconstituting the complete receptor
signaling complexes that includes the full-length receptors,
ligands, signaling, and bridging adapters and some of the
downstream kinases. This is a common problem for all
receptors with single transmembrane helices, and future
research in receptor biology should address this fundamental
problem.
Structures of the monomeric TLRs without bound
ligands
The extracellular domains
The available x-ray structures of the extracellular domains of
TLRs in the monomeric state reveal the conformations of
inactive TLRs prior to binding ligand (14, 17, 31). In addition to this, several structures of TLR2 and TLR4 bound to
non-agonistic or antagonistic ligands have been determined
(16, 17). The extracellular domains of TLRs belong to the
LRR family. The amino acid sequences of LRR family proteins are composed of multiple copies of repeat modules
named LRR modules (22, 23) (Fig. 2). Each module is 20
30 amino acids long and has a characteristic LxxLxLxxN
sequence motif. The central LxL part of the module forms
the core of a b strand. The two leucines point towards the
interior of the protein, making up the hydrophobic core,
whereas the variable 9 residues within the motif are
exposed to solvent, and some are involved in interactions
with ligands. The conserved asparagines in the motifs form
a continuous hydrogen bonding network with backbone
carbonyl oxygens in neighboring LRR modules throughout
the protein. The overall structures of the LRR proteins
resemble a horseshoe. The b strands provided by the each
LRR module assemble into a large b sheet making up the
entire concave surface of the horseshoe. Variable amino
acids outside the conserved LRR motif of each module generate the convex part of the structure. The LRR modules are
protected by two special modules named LRRNT and LRRCT
in the N- and C-termini of the proteins. These modules do
not conform to the sequence conservation pattern of the
LRR modules and often contain an anti-parallel b hairpin
stabilized by disulfide bridges. The majority of the LRR
family proteins are involved in proteinprotein interactions,
and their protein-binding sites are located in their concave
surfaces. TLRs are atypical members of the LRR family: with
the exception of TLR5, their ligands are not proteins and

218

their ligand-binding sites are not located in their concave

surfaces (4).
The LRR family is divided into seven subfamilies, each of
which has a different module length and sequence conservation pattern in the convex part of the structure (22, 23).
Some of them have parallel a helices, and others have 310
helices, polyproline II helices, or irregular loops. Their
sequence conservation patterns suggest that the TLRs belong
to the typical subfamily (17, 32, 33). However, they deviate substantially from the canonical pattern of the subfamily.
The LRR modules of the typical subfamily contain 24 amino
acids and have 310 helices in the convex area. On the other
hand, the lengths of TLR modules vary between 20 and 28
amino acids, and the convex parts of their structures are
composed of mixtures of a helices, 310 helices, and irregular
loops. Because of this, the convex surface of a TLR is rough
rather than smooth and contains multiple grooves and shallow pockets that can serve for ligand interaction. Furthermore, the conserved asparagines are missing in some of the
LRR modules in lipid-interacting TLRs, TLR1, TLR2, TLR4,
and TLR6; this allows unusual distortions of the horseshoelike structures, giving rise to sharp transitions in the torsion
and twist angles of the central b sheet (4, 15, 17). Such
structural distortion permits the formation of hydrophobic
pockets in TLR1 and TLR2 and also has a role in the interaction of TLR4 with MD-2.
Structures of the intracellular domains
Intracellular TIR domains are found not only in the TLRs
but also in the adapters involved in TLR signaling pathways
(34, 35). Five signaling and bridging adapters, MyD88,
MyD88-adapter-like protein (MAL), TIR-domain-containing
adapter-inducing interferon-b (TRIF), translocating chainassociating membrane protein (TRAM), and sterile-a and
Armadillo motif-containing protein (SARM), have been
found to be important in human TLR responses. MAL, TRIF,
and TRAM are also called TIR-domain-containing adapter
protein (TIRAP), TIR-containing adapter molecule-1 (TICAM-1), and TICAM-2, respectively. All TIR domains have a
similar overall structure (Fig. 3), with a central five-stranded
parallel b-sheet surrounded by five a helices (24, 2629).
Sequence conservation in the TIR family is relatively low, in
the 2030% range, and domain lengths vary between 135
and 160 residues. This is because of large deletions and
insertions in the loop regions.
The structures of monomeric TIR domains have been
determined in the cases of TLR1, TLR2, and TLR10 (26,
2012 John Wiley & Sons A/S
Immunological Reviews 250/2012

Song & Lee
Structure of Toll-like receptors

Fig. 3. TIR domain structures. The PdTIR, AtTIR, and L6 TIR domains are from the proteins of Paracoccus denitrificans, Arabidopsis thaliana, and Linum
usitatissimum, respectively. The figure was drawn using coordinate files with PDB codes, 1FYV (TLR1), 1FYW (TLR2), 2J67 (TLR10), 2Z5V
(MyD88), 2Y92 (MAL), 1T3G (IL-1R), 3H16 (PdTIR), 3JRN (AtTIR), and 3OZI (L6). The a helices are green and the central b sheet is blue. The
BB and DD loops important in TLR signaling are shown in red. The AtTIR and L6 TIR domains have additional helices in the aD region.

28). They share substantial structural homology with Ca

rms differences of less than 2 A. Mutagenesis studies have
identified several regions that are important for signaling
activity. The most prominent region identified is the BB
loop that connects strand bB and helix aB (26, 28, 3639).
The BB loop extends away from the rest of the TIR domain,
forming a protrusion from the surface. A P712H substitution in TLR4 abolishes the host response to LPS. The corresponding residue in TLR2 is located in the BB loop and
does not have any clear structural role, suggesting that it is
functionally but not structurally important. The TLR10 TIR
domain has been crystallized as a homodimer with an
extensive dimer interface, although it exists as a monomer
in solution (26). The dimerization interface is symmetrically
shaped, involves the BB loop, and shows substantial
sequence conservation. Because of this, the dimeric arrangement of TLR10 TIR has been used as the template for TIR
homodimerization in several modeling studies (38). Some
mutations in the DD loop that connects strand bD with helix
aD also have significant effects on TLR activation, suggesting
that the DD loop is part of the TIRTIR interaction interface
along with the BB loop (26, 28, 3639).
The structures of the TIR domains of the signaling adapters MyD88 and MAL have also been reported (27, 29).
Human MyD88 consists of 296 amino acids and is composed of two signaling domains. The C-terminal TIR
domain mediates interaction with the receptor, and the Nterminal death domain is responsible for recruiting the
2012 John Wiley & Sons A/S
Immunological Reviews 250/2012

IRAK4 and IRAK1/2 kinases that initiate downstream signaling. The structure of the MyD88 TIR domain has been
determined by nuclear magnetic resonance (NMR) spectroscopy (27). Its overall structure is similar to those of the TLR
TIR domains. The largest structural difference is in the BB
and DD loop regions; the aB helix becomes much shorter
because some of the N-terminal residues of the helix are
unwound to be part of the BB loop. The aD helix is completely changed to an irregularly shaped loop. Using a
NMR-based method, they have performed binding studies
between the MyD88 and MAL TIR domains. They found that
these two TIR domains interact weakly, with a Kd value of
approximately 7 micromoles. Alanine substitution of R196
of MyD88 reduced MAL binding twofold, and mutations of
the residues R288 and R217 also substantially reduced MAL
binding. These observations suggest that the continuous
interface containing the BB loop and these arginine residues
mediates the TIRTIR interaction with MAL.
MAL is required for MyD88-dependent signaling by TLR2
and TLR4 (34, 35). Other TLRs do not need MAL for
MyD88-dependent signaling. Human MAL has 221 amino
acid residues that make up a phosphatidyl inositol-binding
motif and a TIR domain. Although the TIR domain has a
similar overall fold composed of a central b sheet surrounded by a helices, its structure is rather different from
other TLR TIR domains (29). Its sequence identity is in the
low twenties and the Ca rms deviation is greater than 2 A.
Furthermore, the aB and aD helices are changed to irregular

219

Song & Lee
Structure of Toll-like receptors

loops. Cysteines play important roles in the structure and

function of MAL, forming two disulfide bridges, C89C134
and C142C174. Disulfide bridges are rare in intracellular
proteins, although not unprecedented. The two other cysteines, C91 and C157, are not involved in forming disulfide
bridges with protein residues. However, they appear to be
chemically reactive and form simultaneous disulfide bridges
with a single DTT molecule in the buffer. Mutagenesis
experiments have shown that these cysteines are crucial for
MyD88 binding. Interestingly, the C157 position is highly
conserved in the TLR family and the corresponding residue
in TLR4 is the target site of TAK-242 (40). This is an experimental drug that blocks TLR4 signaling by inhibiting its
interaction with MAL and TRAM (9).
TIR domains are found not only in the mammalian proteins but also in several bacterial and plant proteins. The
bacterial TIR-containing proteins are thought to interfere
with the host immune system by hijacking MyD88 through
direct TIRTIR interaction (4144). Plants contain a large
number of related proteins which are involved in resistance
to infection (45, 46). These proteins contain single TIR
domains that are thought to be involved in multimerization.
The structures of the bacterial TIR domain of Paracoccus denitrificans and the plant TIR domains of Linum usitatissimum and
Arabidopsis thaliana have been determined (36, 47, 48). These
non-animal TIR domains have the similar overall fold
expected from the sequence conservation, although the two

plant TIRs have additional helices in the aD helix region

(Fig. 3).
Dimerization and activation of TLRs
Agonistic ligands induce homo- or heterodimerization of
TLR extracellular domains. The structures of these TLRligand complexes have been extensively studied by x-ray
crystallography (15, 16, 18, 20, 21) (Fig. 4). Their interactions with ligands are extraordinarily diverse (Fig. 5A). TLR2
forms heterodimers with TLR1 and TLR6 after binding lipopeptides (15, 16). The lipid chains of the ligands are
inserted into hydrophobic chambers within TLR2 and TLR1.
For binding to the lipid chains of LPS, TLR4 use a hydrophobic chamber not present in the TLR itself but in the
accessory protein MD-2 (17, 19). MD-2 interacts with the
N-terminal and central domains of TLR4, and LPS induces
dimerization of the TLR4-MD-2 heterodimers, forming heterotetrameric complexes of two TLRs and two MD-2s via a
second TLR4 interaction site in MD-2 (20). Hydrophilic
interactions play major roles in ligand recognition by TLR3
and TLR5. TLR3 has two ligand-binding sites located near
the N-terminus and C-terminus, respectively, of its extracellular domain, and flagellin binds to a concave surface
formed by the N-terminal nine LRR modules of TLR5 (18,
21). In each of these cases, the ligands play direct roles in
TLR dimerization by interacting simultaneously with two
TLRs. All the resulting TLR structures have strikingly similar

Triacyl lipopeptide

Diacyl lipopeptide

dsRNA

TLR2

TLR2

TLR3

TLR1

TLR6

LPS

TLR4

Flagellin

TLR5

Fig. 4. Structures of Toll-like receptor (TLR)-ligand complexes. The transmembrane helices and intracellular domains are not included in the
crystal structures. Synthetic lipopeptides that contain the di- or triacylated cysteine and several peptide residues were used in the TLR2-TLR1 and
TLR2-TLR6 crystallographic studies. The 11 C-terminal modules of TLR5 are deleted in the crystal structure.

220

2012 John Wiley & Sons A/S

Immunological Reviews 250/2012

Song & Lee
Structure of Toll-like receptors

Triacyl lipopeptide
Central

LRRNT

C-terminal

N-terminal

TLR2

Primary Dimerization
interface interface

LPS

TLR1

LRRCT

MD-2

MD-2

Diacyl lipopeptide
TLR4

TLR2

TLR6
C

D3

D3

D2 D2

Flagellin

D1
D3

dsRNA

D3

TLR5

TLR3

O specific
chain

Core

Protein

Protein

O
O
-

O P

O
O

HO

O-

O
O
O

HN
O

O
O

HO
O

NH

Cys

P OO

Cys

NH
O

NH

HO

NH
NH3+

O
O

O
O

O
O

Lipid A

LPS

Triacyl lipoprotein

Diacyl lipoprotein

Fig. 5. Ligand induced homo- and heterodimerization of the extracellular domains of Toll-like receptors (TLRs). (A) Diversity of the TLRligand interaction. The extracellular domains of TLRs are shown as arcs. The C-terminal fragment of TLR5 deleted in the crystal structure is shown
by dashed lines. The D3 domain of flagellin is not visible in the electron density map, presumably because it is very flexible. The boundaries of
the LRRNT and LRRCT modules are drawn with solid lines. Some TLRs, namely TLR1, TLR2, TLR4, and TLR6, have structural transitions that
divide them into three subdomains, N-terminal, central, and C-terminal. The boundaries of these subdomains are shown by dashed lines. (B)
Chemical structures of LPS and lipoproteins.

shapes resembling the letter m. In these dimeric structures,

the two C-termini of the extracellular domains converge in
the center while the N-termini stretch outwards (15, 16,
18, 20, 21). This finding suggests the hypothesis that
dimerization of the extracellular domains leads to juxtaposition of the intracellular domains, and the signaling adapters
are believed to be recruited into such dimerized forms of
the TLR intracellular domains (15). The structural changes
2012 John Wiley & Sons A/S
Immunological Reviews 250/2012

induced by ligands are minimal, and only minor adaptations

of flexible loops or amino acid side chains are involved in
ligand binding.

Dimerization of TLR4-MD-2-LPS complexes

The principal ligand of TLR4 is lipopolysaccharide (LPS),
which is the main component of the outer membrane of

221

Song & Lee
Structure of Toll-like receptors

Gram-negative bacteria (4951). LPS is a macromolecular

glycolipid composed of a hydrophobic lipid A region
attached to a long and branched carbohydrate chain
(Fig. 5B). The lipid A has two glucosamine sugars to which
typically four to seven lipid chains are attached (5254).
Lipid A is negatively charged because two phosphate groups
are attached to the glucosamine backbone. Significant structural diversity has been reported in the structure of lipid A.
For example, the number of lipid chains can vary, and the
phosphate groups can be deleted, or modified by several
other chemical groups (52, 54). The carbohydrate chain of
LPS can be divided into two areas, the core and the O-specific chain (50). The relatively conserved core oligosaccharide chain contains multiple copies of the unusual sugars, 3deoxy-D-manno-oct-2-ulosonic acid (Kdo) and L-glycero-Dmannoheptose (Hep). The core region is frequently modified
by negatively charged chemical groups such as phosphates.
The O-specific chain is composed of many copies of repeating units containing galactoses, galactosamines, and other
sugars. The structure and composition of the repeating units
differ between bacterial species and even between different
growth conditions in the same species. Complete removal of
the core and O-specific chain has only a minor effect on the
immunological activity of LPS. Therefore, the lipid A region
is believed responsible for most of its inflammatory activity.
Crystallographic investigations have confirmed this and
have shown that hydrophobic interaction of lipid A with a
pocket in MD-2 makes the main contribution to LPS binding
(20).
LPS is recognized by TLR4-MD-2 heterodimers expressed
on the surface of immune cells (5557). MD-2 is approximately 18 kDa glycoprotein that functions as the LPS-binding subunit of the TLR4-MD-2 heterodimer (5760). It has
two anti-parallel b sheets sandwiched together, like immunoglobulin fold proteins. But the disulfide bond connecting
the two sheets in the latter is absent from MD-2 (17, 19).
Because of this, the two b sheets in the sandwiched structure of MD-2 can be separated to generate a large internal
pocket that is ideally shaped for binding flat hydrophobic
ligands like LPS. MD-2 has two TLR4 binding sites, named
the primary interface and dimerization interface, respectively
(17, 20) (Fig. 5A). The primary interaction interface is
formed between one edge of MD-2 and part of the concave
surface provided by the N-terminal and central domains of
TLR4. This well-conserved interaction interface is mostly
composed of ionic and hydrogen bonds and does not
require binding of LPS. The dimerization interface of MD-2
is located opposite the primary interface and interacts with a

222

convex surface provided by a small hydrophobic patch in

the C-terminal domain of TLR4. One partially exposed lipid
chain of LPS is located in the core of the dimerization interface. This lipid chain and surrounding hydrophobic amino
acid residues of MD-2 interact with the hydrophobic patch
of TLR4. Several hydrophilic interactions between protein
residues of MD-2 and TLR4 support this core hydrophobic
interface. The two phosphate groups of lipid A form multiple charge and hydrogen bond interactions with both MD-2
and the two TLR4s and therefore contribute to the stability
of the dimerization interface.
The structure-activity relationships of LPS have been studied for decades using natural and chemically modified LPS
(5052, 61). The results show that the total number of lipid
chains and the presence of the two phosphate groups in
lipid A are the most important factors affecting its inflammatory activity. The crystal structure provides an explanation
of this observation (20). The six lipid chains are optimal
because partial exposure of one of the lipid chains is critical
for forming the dimerization interface. When the LPS has
less than six lipid chains, part of the MD-2 binding pocket
becomes empty, and this will destabilize the overall interaction (62). When the LPS has more than six lipid chains, the
additional lipid chain may interfere with the interaction
between TLR4 and MD-2 because it is likely to prevent the
approaching TLR4 from achieving the optimal interaction
distance. The phosphate groups are essential because they
provide the charge and hydrogen bonding network. Deletion of either of these phosphates reduces inflammatory
activity more than a 100-fold (54, 61, 63). A lipid A derivative with one of the phosphate groups deleted retains more
or less, all of its immune stimulatory activity, but loses most
of its inflammatory toxicity. Because of this, it has recently
been approved as a general vaccine adjuvant (10).
Formation of TLR2-TLR1 and TLR2-TLR6-lipoprotein
complexes
TLR2 forms heterodimers with TLR1 or TLR6 and interacts
with bacterial lipoproteins (15, 16, 64, 65). Lipoproteins
are a family of proteins that are anchored to the bacterial
membrane by lipid chains covalently attached to conserved
N-terminal cysteines (6670) (Fig. 5B). The protein parts of
the lipoproteins have little sequential and functional homology. The lipoproteins found in the Gram-negative bacteria
have three lipid chains; two of them are connected to the
glycerol group, which is in turn attached to the sulfur atom
of the N-terminal cysteine. These lipid chains are referred to
as ester-bound lipid chains. The third lipid chain is
2012 John Wiley & Sons A/S
Immunological Reviews 250/2012

Song & Lee
Structure of Toll-like receptors

connected to the N-terminal NH2 via an amide bond. Therefore, it is referred to as an amide-bound lipid chain. The
lipoproteins and lipopeptides of Gram-positive bacteria and
mycoplasmas were thought to have only two lipid chains
because they lack the enzyme responsible for attaching the
amide-bound lipid chain (7173). However, this view has
recently been challenged because several lipoproteins purified from Staphylococcus aureus were shown to have three lipid
chains (74, 75). In addition to lipoproteins, many bacterial
and fungal molecules have been proposed to function as
agonistic ligands for TLR2, and many, but not all, of these
candidate ligands also contain diacylglycerol groups (3, 4).
TLR2-TLR1 heterodimers are the principal receptors for
triacylated lipoproteins (65). The three lipid chains of the
ligand bridge the TLRs by interacting simultaneously with
TLR2 and TLR1 via the three lipid chains; two lipid chains
are inserted into the large hydrophobic pocket in TLR2, and
the third, the amide-bound chain, is inserted into the
narrow hydrophobic channel in TLR1 (15) (Fig. 5A). The
entry of the TLR2 pocket lies near the boundary between
the central and C-terminal domains and extends into an
internal hydrophobic pocket formed by LRR modules 912.
The glycerol and the peptide backbone atoms of the two
N-terminal residues of the lipoprotein form extensive
hydrogen bonds with amino acids in both TLR2 and TLR1.
Direct TLR1TLR2 interaction also contributes to the stability of the heterodimer. The core of the interface is formed
by several hydrophobic residues. It is surrounded by hydrophilic residues forming ionic and hydrogen bonds. The
Pro315 SNP of TLR1 frequently found in Caucasian populations appears to block TLR1 signaling by perturbing this
heterodimerization interface (15, 76).
Diacylated lipoproteins from Gram-positive bacteria or
mycoplasmas mainly activate TLR2-TLR6 heterodimers (64).
The two ester-bound lipid chains are inserted into the same
TLR2 pocket as TLR2-TLR1 triacyl lipopeptide complexes,
and the hydrophilic glycerols and peptide backbones of the
lipoproteins form hydrogen bonds with amino acid residues
of both TLR2 and TLR6 (16) (Fig. 5A). The TLR2-TLR6
complex cannot bind to the triacylated lipoproteins because
the hydrophobic channel responsible for interaction with
the amide-bound lipid chain in TLR1 is blocked by two
bulky phenylalanines in TLR6. Mutation of these residues
into the ones found in TLR1 renders TLR6 capable of
responding to both diacylated and triacylated lipoproteins.
In the TLR2-TLR1-triacylate lipopeptide complex, the
amide-bound lipid chain plays an important role in TLR
heterodimer formation. Without the amide-bound lipid
2012 John Wiley & Sons A/S
Immunological Reviews 250/2012

chain in the diacylated lipopeptide, TLR2 and TLR6 can

still form stable heterodimers because the size of the hydrophobic core of the proteinprotein interface is increased,
and this can compensate for the missing lipid channel
interaction.
Homodimerization of the TLR3-dsRNA complex
The principal ligand of TLR3 is the double-stranded RNA
that can be generated during virus replication (3). The
receptor is localized mainly in endocytic organelles and its
activation requires TRIF instead of MyD88 (Fig. 6A). The
crystal structure of the complex between the TLR3 extracellular domain and a 46 bp double-stranded RNA has been
determined by x-ray crystallography (18). TLR3 has two
RNA binding sites located in its N-terminal and C-terminal
regions, respectively (Fig. 5A). The distance between the two
sites is approximately 70 A, which corresponds to roughly
two helical turns of RNA. The backbone phosphates and
sugars of dsRNA make major contributions to TLR3 binding
at both sites. Ligand binding and TLR3 dimerization require
an acidic environment below pH 6.0 (77). The TLR3 structure can explain this strong pH dependence because several
histidines make crucial bonds with the phosphate backbones
of the RNA, and their protonation states appear essential for
stability of the interaction. The nucleic acid bases make minimal contact with the TLR, which accounts for the
sequence-independent binding of RNA. Although approximately 40 bp of RNA appears to be the minimum required
to induce TLR3 dimerization, a robust immune response
requires RNA of more than approximately 100 bp. Furthermore, binding of RNA by TLR3 shows strong positive cooperativity (77). These observations suggest that clustering of
more than one TLR3 dimer may be necessary for an adequate immune response. A recent study supports this model
(30). They produced antibodies that can block activity of
TLR3. Interestingly, two of these antibodies did not directly
interfere with neither of the two ligand-binding sites. Based
on crystallographic and molecular modeling study, they proposed that these antibodies interrupt TLR3 activation by
inhibiting lateral clustering of the TLR3 dimers.
Homodimerization of the TLR5-flagellin complex
Flagellae are large protein complexes that confer motility on
bacteria. They are made up of multiple copies of approximately 20 different protein subunits (78, 79). Electron
microscopy shows that they consist of three substructures, a
basal body generating torque, a hook connecting the basal

223

Song & Lee
Structure of Toll-like receptors

Diacyl
lipoprotein

Flagellin

TLR2
TLR6

TLR5
TLR5

MAL

MAL

Unmethylated
CpG DNA

LPS

dsRNA

ssRNA

MD-2
TLR4
TLR4
MD-2

Triacyl
lipoprotein

TLR1
TLR2

Plasma membrane
MAL

TLR3
TLR3

Endosome

TRIF

TLR4-MD-2

MAL

TLR4-MD-2
LPS

TIR

TIR

TIR

TIR

MAL

TIR

TIR

TIR

TIR

MyD88

IRAK4 death domain

Myddosome

IRAK2 death domain

Fig. 6. Toll-like receptor (TLR) signaling complexes. (A) A variety of signaling and bridging adapters are recruited to active TLRs. (B)
Proposed structure of the TLR4-MD-2-LPS-Myddosome complex. The structures of the multimerized TIR complexes between receptors and
adapters are not known and are shown schematically as boxes.

body to a helical fiber, and the long flagellin fiber. The latter
is assembled from a large number of fliC flagellins. A highresolution structure of a fliC fragment has been determined by
x-ray crystallography (80). To obtain monomeric proteins
suitable for crystallization, the D0 domain formed by the Nterminal and C-terminal fragments was removed by proteolysis and the resulting approximately 45 kD fragment containing the remaining middle domains, D1, D2, and D3, was
used for structure analysis. The highly conserved D1 domain
plays the main role in polymerization of the fliC subunits into
a helical fiber. It is composed of three long a helices and a b
hairpin. Two of the helices are formed by the N-terminal part

224

of the protein, and the third helix by the C-terminal region.

The helices and the b hairpin are arranged in an anti-parallel
bundle-like structure. The D2 domain is composed of several
b hairpin structures and connects the D1 and D3 domains.
The D3 domain is formed by the middle part of fliC, and its
sequence is highly diverse in different bacterial species. It is
exposed to the surface of the flagellin fiber and mediates the
reversible lateral interactions between flagellin fibers. Because
neither vertebrates nor plants contain flagellae, the presence
of these structures serves as an early sign of bacterial infection
in both classes of organism (81). TLR5 is the major innate
immune receptor for flagellin in vertebrates.
2012 John Wiley & Sons A/S
Immunological Reviews 250/2012

Song & Lee
Structure of Toll-like receptors

Recently, the crystal structure of zebrafish TLR5 has been

determined in a complex with a truncated fragment of
Salmonella fliC (21). As with TLR2 and TLR4, several hybrid
proteins consisting of TLR fragments fused with VLR
proteins from jawless fishes were used for crystallization
(1517). The TLR5 of fish and humans have substantial
sequence homology, suggesting that the two proteins have
similar structures and ligand recognition mechanisms. The
structure obtained shows that the flagellin D1 domain plays
the dominant role in binding and dimerization of TLR5
(Fig. 5A). One face of the D1 domain interacts with an
extended surface area encompassing the LRR modules from
LRRNT to LRR9 of TLR5. At the same time, the D1 domain
forms a bond with the second TLR5 molecule in the dimer
and therefore bridges the two TLRs. The overall shape and
curvature of the horseshoe-like structure formed differ in
only minor ways from those of TLR5 without bound ligand.
This observation applies to all TLR structures, thus underlining the high structural rigidity of the LRR fold. The spatial
arrangement of the two TLR5 molecules in the dimer also
resembles closely that of the other TLR dimers. The two Cterminal regions join in the center and the two N-termini
splay outwards. Because the TLR5 fragment that they used
for crystallization does not contain native TLR sequences
beyond LRR14, it is not clear from the structure alone if the
C-terminal part of the receptor contributes to dimerization
or to ligand interaction. However, mutagenesis experiments
suggest that the deleted C-terminal region probably has only
a minor role (21).
Dimerization of intracellular domains and recruitment of
signaling adapters
Binding of agonistic ligands initiates intracellular signaling
by recruiting specific adapters to the intracellular TIR
domains of TLRs (Fig. 6A). In humans, TLR activation
involves five proteins, MyD88, MAL, TRIF, TRAM, and
SARM (34, 35). MyD88 is required for signaling by all TLRs
except TLR3. The latter uses TRIF instead of MyD88. The
bridging adapter, MAL, is necessary for MyD88-dependent
signaling by TLR2 and TLR4. The MyD88-MAL-dependent
signaling by TLR4 is activated mainly on the plasma membrane, whereas TRIF-TRAM-dependent signaling by TLR4 is
activated on endocytic organelles (82). TRAM is the bridging adapter for TRIF. It is necessary for signaling by TLR4
but not by TLR3. SARM negatively regulates TLR signaling
by interacting with TRIF and blocking its function (83). All
these adapters contain TIR domains that are essential for signal activation. Mutagenesis experiments have shown that
2012 John Wiley & Sons A/S
Immunological Reviews 250/2012

recruitment of adapters to receptor TIRs is mediated by TIR

TIR interactions. In addition to TIR domains, MAL contains
an N-terminal phosphatidyl inositol-binding motif and
TRAM has a myristoylation site (84, 85). These additional
motifs are necessary for membrane localization of these
bridging adapters.
An important aim of research has been to determine the
structures of TIR homo- and heterocomplexes. However, the
TLR TIR domains produced as truncated forms by genetic
manipulation interact only weakly or not at all with the TIR
domains of signaling adapters (28). This makes the direct
structural study of the TIR complexes very difficult. Hence,
identification of interaction interfaces has been attempted by
site-directed mutagenesis. Ronni et al. (86) introduced
approximately 70 alanine substitution and short deletion
mutations into the human TLR4 TIR domain and measured
the effect on the IL-7 response in transfected HEK293 cells.
To differentiate the signal of the introduced TLR4 from the
endogenous TLR4, they used a constitutively active chimeric
receptor with the extracellular domain of CD40 and the
intracellular domain of TLR4. Using this chimeric receptor,
they found that mutations clustered in two regions, the BB
and DD loops, resulted in substantial attenuation of the
TLR4 signaling. However, it is not clear whether these
regions are involved in the homotypic aggregation of the
TLR TIRs or the heterotypic interactions with adapter TIR
domains.
Recently, Bovijn et al. (87) advanced research on TIRTIR
interaction by identifying mutations that specifically affect
homotypic interactions of TLR4 TIR domains using a technique called MAPPIT, which detects protein interactions in
mammalian cells using chimeric receptors. They discovered
several interesting aspects of TIR interaction. First, the
MyD88-TLR4 interaction requires as bridging adapter, the
MAL TIR domain. Second, the TIR domains of the bridging
adapters MAL and TRAM can interact directly with the TLR4
TIR domain without the help of signaling adapters, MyD88
or TRIF. Third, they found that the regions surrounding the
BB loops are crucial for TLR4 TIR homomultimerization.
Finally, they also found that all mutations that affect the
MAL or TRAM interaction block TLR4 TIR multimerization,
suggesting that TLR4 TIR multimerization is a prerequisite
for the MAL/TRAM interaction.
In addition to the TIR domain, MyD88 contains a death
domain that is required for activation of downstream effectors. The downstream kinases that are directly recruited to
the TLR-MyD88 complex are IRAK4 and IRAK1/IRAK2. Lin
et al. (25) reported a crystal structure of the death domain

225

Song & Lee
Structure of Toll-like receptors

complex of MyD88-IRAK4-IRAK2 (Fig. 6B). The TIR domain

of MyD88 and kinase domains of IRAKs were removed for
the crystallization. These workers found that the death
domains form a large helical complex that they have named
as Myddosome. In this complex, six MyD88, four IRAK4,
and four IRAK2 death domains are arranged as a left-handed
helix composed of four distinct layers. The top two layers
have six MyD88s, the middle layer, four IRAK4s, and the
bottom layer, four IRAK2s. Some deletion or substitution
mutations that cause life threatening bacterial infections
mapped in the binding interface, demonstrating an indispensable role for the complex in TLR signaling. Based on
this structure, it was proposed that formation of the Myddosome brings the kinase domains of the IRAKs into proximity
so that they can be phosphorylated and activated.
Because the Myddosome complex has six MyD88 death
domains, it appears that the complete TLR signaling complex contains more than one set of TLR dimers. If we
assume that the TIR domain of MyD88 interacts 1:1 with
the TLR TIR, then six MyD88 molecules in the Myddosome
are capable of binding three TLR dimers (Fig. 6B). This kind
of higher order clustering of receptor multimers has been
proposed for quite a few membrane receptors (30, 8892).
However, characterization of high-resolution structures of
the clusters has not been possible, mostly due to difficulties
in making full-length receptors and reconstituting complete
signaling complexes in a membrane environment.

phospholipids on the concave surface of the boomerang. As

the sequences of LBP and BPI can be aligned over their
entire lengths, LBP is expected to share the general structural
features of BPI: its elongated shape and pseudosymmetry as
well as the three structural elements, the two barrels, central
sheet, and apolar phospholipid-binding pockets. However,
the two proteins have clear functional difference: CD14
transfers LPS to TLR4-MD-2, but BPI cannot. This difference
need to be explained by future structural study of LBP.
CD14 is an LRR family protein that is involved in the
transfer of LPS from LBP to the TLR4-MD-2 complex. It is
expressed on the surface of myelomonocytic cells as a glycosylphosphatidylinositol (GPI)-linked glycoprotein or in
soluble form in serum (97). In addition to the LPS of
Gram-negative bacteria, CD14 can bind other amphipathic
microbial products, such as peptidoglycan, lipoteichoic acid,
lipoarabinomannan, and lipoproteins (98, 99). The monomeric subunit of CD14 contains 11 LRR modules and a single LRRNT module (100) (Fig. 7B). CD14 does not contain
an LRRCT module, which is required to cover the hydrophobic core of the protein from exposure to solvents.
Instead, CD14 uses the exposed hydrophobic core to form
a homodimer. Each CD14 monomer contains an N-terminal
hydrophobic pocket that is formed between the LRRNT and
the first LRR modules. The convex surface between these

Structures of functionally and structurally related

proteins, LBP, CD14, RP105, and MD-1
LPS and lipoproteins are amphipathic macromolecules, and
they form aggregated micelle structures in water solution.
Because of this, efficient extraction of these molecules from
bacterial membrane requires accessary proteins, lipopolysaccharide-binding protein (LBP), and CD14. LBP is an acutely
induced plasma protein that binds avidly to LPS and delivers
it to CD14 (93). Addition of purified LBP enhances the
sensitivity of macrophages to LPS 100- to 1000-fold (94).
LBP belongs to the lipid transfer/lipopolysaccharide-binding
protein (LT/LBP) family (95). Other members of the family
include bactericidal/permeability-increasing protein (BPI),
cholesterol ester transfer protein (CETP), phospholipid
transfer protein (PLTP), and a few poorly characterized proteins. LBP and BPI share 48% sequence identity and bind to
and regulate the biological effects of LPS. BPI is a boomerang-shaped protein formed by two domains of similar size
connected by a proline-rich linker (96) (Fig. 7A). Each
domain has a deep hydrophobic pocket that can bind

226

BPI
C

RP105

CD14
RP105

RP105-MD-1

Fig. 7. Structures of proteins involved in Toll-like receptor

activation. Structures of BPI (A), CD14 (B), and the RP105-MD-1
complex (C). CD14 forms a homodimer. The monomeric subunits of
CD14 are colored in green and blue, respectively. The MD-1 molecules
bound to RP-105s are schematically drawn in grey.
2012 John Wiley & Sons A/S
Immunological Reviews 250/2012

Song & Lee
Structure of Toll-like receptors

two modules is cracked open, exposing the hydrophobic

core residues inside. The size of the pocket is large enough
to accommodate the lipid chains of a single LPS molecule.
Therefore, the function of CD14 appears to be monomerization of the aggregated ligands for efficient binding to
TLR4.
RP105 and MD-1 are structurally and sequentially homologous to TLR4 and MD-2, respectively (101103). However, unlike TLR4, RP105 does not contain a sizable
intracellular domain, suggesting it cannot activate signaling
pathways by itself. The exact function and the physiological
ligands for the RP105-MD-1 complex have not been clearly
identified yet, although it has been proposed that this protein complex regulates TLR2 and TLR4 signaling by directly
interacting with them (104107). Several structures of the
RP105-MD-1 complex have recently been determined (108
111) (Fig. 7C). The individual structures of RP105 and MD1 closely resemble those of TLR4 and MD-2, respectively,
and the primary interaction interface between RP105 and
MD-1 shows a homology with that of TLR4 and MD-2.
However, unlike TLR4-MD-2, which requires LPS for the
formation of a heterotetramer, the RP105-MD-1 can form a
stable heterotetramer without bound ligands. Furthermore,
the overall arrangement of the two RP105 in the complex
shows a clear difference from the arrangement of TLR4 in
the TLR4-MD-2-LPS complex: the N-terminal LRR modules,
instead of the C-terminal modules, of the two RP105 chains

converge in the center of the complex (Fig. 7C). This headto-head arrangement of RP105 contrasts with the tail-to-tail
mode, which is highly conserved in the ligand-activated
TLR homo- and heterodimers (Fig. 5A). The structure of the
MD-1 pocket is homologous with that of MD-2, suggesting
that MD-1 may be able to interact with flat hydrophobic
ligands like LPS.
Conclusion
Over the last 10 years, important progress has been made in
structural understanding of the ligand recognition and activation mechanisms of TLRs. The structures of the extracellular domains of TLRs 16 have been determined in complex
with their cognate ligands. The structures of the intracellular
TIR domains of several TLRs and adapters have also been
determined. These structures show us how TLRs recognize
their ligands and how ligand binding induces structural
rearrangements of the receptor. For the next 10 years, the
most important goal should be determining structures of the
complete signaling complexes comprising the full-length
receptors, ligands, signaling adapters, bridging adapters, and
downstream kinases. To achieve this ambitious goal, we
must first find ways to produce enough amounts of structurally and functionally intact receptors and to reconstitute the
complete receptor-ligand adapter-kinase complexes from the
purified components.

References
1. Janeway CA, Jr. Approaching the asymptote?
Evolution and revolution in
immunology. Cold Spring Harb Symp Quant
Biol 1989;54:113.
2. Medzhitov R. Toll-like receptors and
innate immunity. Nat Rev Immunol
2001;1:135145.
3. Gay NJ, Gangloff M. Structure and function of
Toll receptors and their ligands. Annu Rev
Biochem 2007;76:141165.
4. Kang JY, Lee JO. Structural biology of the Tolllike receptor family. Annu Rev Biochem
2011;80:917941.
5. Hennessy EJ, Parker AE, ONeill LA. Targeting
Toll-like receptors: emerging therapeutics? Nat
Rev Drug Discov 2010;9:293307.
6. Bennett-Guerrero E, et al. A phase II, doubleblind, placebo-controlled, ascending-dose study
of Eritoran (E5564), a lipid A antagonist, in
patients undergoing cardiac surgery with
cardiopulmonary bypass. Anesth Analg
2007;104:378383.
7. Tidswell M, et al. Phase 2 trial of eritoran
tetrasodium (E5564), a toll-like receptor 4
antagonist, in patients with severe sepsis. Crit
Care Med 2010;38:7283.

2012 John Wiley & Sons A/S

Immunological Reviews 250/2012

8. Ii M, et al. A novel cyclohexene derivative, ethyl

(6R)-6-[N-(2-Chloro-4-fluorophenyl)sulfamoyl]
cyclohex-1-ene-1-carboxylate (TAK-242),
selectively inhibits toll-like receptor 4-mediated
cytokine production through suppression of
intracellular signaling. Mol Pharmacol
2006;69:12881295.
9. Kawamoto T, Ii M, Kitazaki T, Iizawa Y,
Kimura H. TAK-242 selectively suppresses Tolllike receptor 4-signaling mediated by the
intracellular domain. Eur J Pharmacol
2008;584:4048.
10. Casella CR, Mitchell TC. Putting endotoxin to
work for us: monophosphoryl lipid A as a safe
and effective vaccine adjuvant. Cell Mol Life Sci
2008;65:32313240.
11. Stevceva L. Toll-like receptor agonists as
adjuvants for HIV vaccines. Curr Med Chem
2011;18:50795082.
12. Mizel SB, Bates JT. Flagellin as an adjuvant:
cellular mechanisms and potential. J Immunol
2010;185:56775682.
13. Bell JK, Askins J, Hall PR, Davies DR, Segal DM.
The dsRNA binding site of human Toll-like
receptor 3. Proc Natl Acad Sci USA
2006;103:87928797.

14. Choe J, Kelker MS, Wilson IA. Crystal structure

of human toll-like receptor 3 (TLR3)
ectodomain. Science 2005;309:581585.
15. Jin MS, et al. Crystal structure of the TLR1-TLR2
heterodimer induced by binding of a tri-acylated
lipopeptide. Cell 2007;130:10711082.
16. Kang JY, et al. Recognition of lipopeptide
patterns by Toll-like receptor 2-Toll-like receptor
6 heterodimer. Immunity 2009;31:873884.
17. Kim HM, et al. Crystal structure of the TLR4MD-2 complex with bound endotoxin antagonist
Eritoran. Cell 2007;130:906917.
18. Liu L, et al. Structural basis of toll-like receptor 3
signaling with double-stranded RNA. Science
2008;320:379381.
19. Ohto U, Fukase K, Miyake K, Satow Y. Crystal
structures of human MD-2 and its complex with
antiendotoxic lipid IVa. Science 2007;316:1632
1634.
20. Park BS, Song DH, Kim HM, Choi BS, Lee H, Lee
JO. The structural basis of lipopolysaccharide
recognition by the TLR4-MD-2 complex. Nature
2009;458:11911195.
21. Yoon SI, et al. Structural basis of TLR5-flagellin
recognition and signaling. Science
2012;335:859864.

227

Song & Lee
Structure of Toll-like receptors

22. Bella J, Hindle KL, McEwan PA, Lovell SC. The

leucine-rich repeat structure. Cell Mol Life Sci
2008;65:23072333.
23. Kobe B, Kajava AV. The leucine-rich repeat as a
protein recognition motif. Curr Opin Struct Biol
2001;11:725732.
24. Khan JA, Brint EK, ONeill LA, Tong L. Crystal
structure of the Toll/interleukin-1 receptor
domain of human IL-1RAPL. J Biol Chem
2004;279:3166431670.
25. Lin SC, Lo YC, Wu H. Helical assembly in the
MyD88-IRAK4-IRAK2 complex in TLR/IL-1R
signalling. Nature 2010;465:885890.
26. Nyman T, Stenmark P, Flodin S, Johansson I,
Hammarstrom M, Nordlund P. The crystal
structure of the human toll-like receptor 10
cytoplasmic domain reveals a
putative signaling dimer. J Biol Chem
2008;283:1186111865.
27. Ohnishi H, et al. Structural basis for the multiple
interactions of the MyD88 TIR domain in TLR4
signaling. Proc Natl Acad Sci USA
2009;106:1026010265.
28. Xu Y, et al. Structural basis for signal
transduction by the Toll/interleukin-1 receptor
domains. Nature 2000;408:111115.
29. Valkov E, et al. Crystal structure of Toll-like
receptor adaptor MAL/TIRAP reveals the
molecular basis for signal transduction and
disease protection. Proc Natl Acad Sci USA
2011;108:1487914884.
30. Luo J, et al. Lateral clustering of TLR3:dsRNA
signaling units revealed by TLR3ecd:3Fabs
quaternary structure. J Mol Biol 2012;421:112
124.
31. Bell JK, et al. The molecular structure of the Tolllike receptor 3 ligand-binding domain. Proc Natl
Acad Sci USA 2005;102:1097610980.
32. Jin MS, Lee JO. Structures of the toll-like receptor
family and its ligand complexes. Immunity
2008;29:182191.
33. Jin MS, Lee JO. Structures of TLR-ligand
complexes. Curr Opin Immunol 2008;20:414
419.
34. Kenny EF, ONeill LA. Signalling adaptors used
by Toll-like receptors: an update. Cytokine
2008;43:342349.
35. Watters TM, Kenny EF, ONeill LA. Structure,
function and regulation of the Toll/IL-1 receptor
adaptor proteins. Immunol Cell Biol
2007;85:411419.
36. Chan SL, et al. Molecular mimicry in innate
immunity: crystal structure of a bacterial
TIR domain. J Biol Chem 2009;284:21386
21392.
37. Gautam JK, Ashish, Comeau LD, Krueger JK,
Smith MF, Jr. Structural and functional evidence
for the role of the TLR2 DD loop in TLR1/TLR2
heterodimerization and signaling. J Biol Chem
2006;281:3013230142.
38. Nunez Miguel R, et al. A dimer of the Toll-like
receptor 4 cytoplasmic domain provides a
specific scaffold for the recruitment of signalling
adaptor proteins. PLoS ONE 2007;2:e788.
39. Underhill DM, et al. The Toll-like receptor 2 is
recruited to macrophage phagosomes and

228

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.
51.

52.

53.

54.

55.

56.

discriminates between pathogens. Nature

1999;401:811815.
Takashima K, et al. Analysis of binding site for
the novel small-molecule TLR4 signal
transduction inhibitor TAK-242 and its
therapeutic effect on mouse sepsis model. Br J
Pharmacol 2009;157:12501262.
Newman RM, Salunkhe P, Godzik A, Reed JC.
Identification and characterization of a novel
bacterial virulence factor that shares homology
with mammalian Toll/interleukin-1 receptor
family proteins. Infect Immun 2006;74:594
601.
Salcedo SP, et al. Brucella control of dendritic cell
maturation is dependent on the TIR-containing
protein Btp1. PLoS Pathog 2008;4:e21.
Cirl C, et al. Subversion of Toll-like receptor
signaling by a unique family of bacterial Toll/
interleukin-1 receptor domain-containing
proteins. Nat Med 2008;14:399406.
Spear AM, Loman NJ, Atkins HS, Pallen MJ.
Microbial TIR domains: not necessarily
agents of subversion? Trends Microbiol
2009;17:393398.
Chisholm ST, Coaker G, Day B, Staskawicz BJ.
Host-microbe interactions: shaping the evolution
of the plant immune response. Cell
2006;124:803814.
Rafiqi M, Bernoux M, Ellis JG, Dodds PN. In the
trenches of plant pathogen recognition: role of
NB-LRR proteins. Semin Cell Dev Biol
2009;20:10171024.
Chan SL, Mukasa T, Santelli E, Low LY, Pascual J.
The crystal structure of a TIR domain from
Arabidopsis thaliana reveals a conserved helical
region unique to plants. Protein Sci
2010;19:155161.
Bernoux M, et al. Structural and functional
analysis of a plant resistance protein TIR domain
reveals interfaces for self-association, signaling,
and autoregulation. Cell Host Microbe
2011;9:200211.
Beutler B, Rietschel ET. Innate immune sensing
and its roots: the story of endotoxin. Nat Rev
Immunol 2003;3:169176.
Raetz CR. Biochemistry of endotoxins. Annu Rev
Biochem 1990;59:129170.
Raetz CR, Whitfield C. Lipopolysaccharide
endotoxins. Annu Rev Biochem 2002;71:635
700.
Erridge C, Bennett-Guerrero E, Poxton IR.
Structure and function of lipopolysaccharides.
Microbes Infect 2002;4:837851.
Galanos C, et al. Synthetic and natural Escherichia
coli free lipid A express identical endotoxic
activities. Eur J Biochem 1985;148:15.
Rietschel ET, et al. Bacterial endotoxin: molecular
relationships of structure to activity and function.
FASEB J 1994;8:217225.
Medzhitov R, Preston-Hurlburt P, Janeway CA,
Jr. A human homologue of the Drosophila Toll
protein signals activation of adaptive immunity.
Nature 1997;388:394397.
Poltorak A, et al. Defective LPS signaling in C3H/
HeJ and C57BL/10ScCr mice: mutations in Tlr4
gene. Science 1998;282:20852088.

57. Shimazu R, et al. MD-2, a molecule that confers

lipopolysaccharide responsiveness on Toll-like
receptor 4. J Exp Med 1999;189:17771782.
58. Nagai Y, et al. Essential role of MD-2 in LPS
responsiveness and TLR4 distribution. Nat
Immunol 2002;3:667672.
59. Schromm AB, et al. Molecular genetic analysis of
an endotoxin nonresponder mutant cell line: a
point mutation in a conserved region of MD-2
abolishes endotoxin-induced signaling. J Exp
Med 2001;194:7988.
60. Viriyakosol S, Tobias PS, Kitchens RL, Kirkland
TN. MD-2 binds to bacterial lipopolysaccharide. J
Biol Chem 2001;276:3804438051.
61. Rietschel ET, et al. The chemical structure of
bacterial endotoxin in relation to bioactivity.
Immunobiology 1993;187:169190.
62. Ohto U, Fukase K, Miyake K, Shimizu T.
Structural basis of species-specific endotoxin
sensing by innate immune receptor TLR4/
MD-2. Proc Natl Acad Sci USA 2012;109:
74217426.
63. Qureshi N, Takayama K, Ribi E. Purification and
structural determination of nontoxic lipid A
obtained from the lipopolysaccharide of
Salmonella typhimurium. J Biol Chem
1982;257:1180811815.
64. Takeuchi O, et al. Discrimination of bacterial
lipoproteins by Toll-like receptor 6. Int Immunol
2001;13:933940.
65. Takeuchi O, et al. Cutting edge: role of Toll-like
receptor 1 in mediating immune response to
microbial lipoproteins. J Immunol 2002;169:10
14.
66. Hantke K, Braun V. Covalent binding of lipid to
protein Diglyceride and amide-linked fatty acid at
the N-terminal end of the murein-lipoprotein of
the Escherichia coli outer membrane. Eur J
Biochem 1973;34:284296.
67. Belisle JT, Brandt ME, Radolf JD, Norgard MV.
Fatty acids of Treponema pallidum and Borrelia
burgdorferi lipoproteins. J Bacteriol
1994;176:21512157.
68. Braun V. Covalent lipoprotein from the outer
membrane of Escherichia coli. Biochim Biophys
Acta 1975;415:335377.
69. Mizuno T. A novel peptidoglycan-associated
lipoprotein found in the cell envelope of
Pseudomonas aeruginosa and Escherichia coli. J
Biochem 1979;86:9911000.
70. Zlotnick GW, Sanfilippo VT, Mattler JA, Kirkley
DH, Boykins RA, Seid RC, Jr. Purification and
characterization of a peptidoglycan-associated
lipoprotein from Haemophilus influenzae. J Biol
Chem 1988;263:97909794.
71. Hutchings MI, Palmer T, Harrington DJ, Sutcliffe
IC. Lipoprotein biogenesis in Gram-positive
bacteria: knowing when to hold em, knowing
when to fold em. Trends Microbiol
2009;17:1321.
72. Muhlradt PF, Kiess M, Meyer H, Sussmuth R,
Jung G. Isolation, structure elucidation, and
synthesis of a macrophage stimulatory
lipopeptide from Mycoplasma fermentans acting
at picomolar concentration. J Exp Med
1997;185:19511958.

2012 John Wiley & Sons A/S

Immunological Reviews 250/2012

Song & Lee
Structure of Toll-like receptors

73. Shibata K, Hasebe A, Into T, Yamada M,

Watanabe T. The N-terminal lipopeptide of a 44kDa membrane-bound lipoprotein of
Mycoplasma salivarium is responsible for the
expression of intercellular adhesion molecule-1
on the cell surface of normal
human gingival fibroblasts. J Immunol
2000;165:65386544.
74. Kurokawa K, et al. The triacylated ATP binding
cluster transporter substrate-binding lipoprotein
of Staphylococcus aureus functions as a native
ligand for Toll-like receptor 2. J Biol Chem
2009;284:84068411.
75. Tawaratsumida K, et al. Characterization of Nterminal structure of TLR2-activating lipoprotein
in Staphylococcus aureus. J Biol Chem
2009;284:91479152.
76. Omueti KO, Mazur DJ, Thompson KS, Lyle EA,
Tapping RI. The polymorphism P315L of human
Toll-like receptor 1 impairs innate immune
sensing of microbial cell wall components. J
Immunol 2007;178:63876394.
77. Leonard JN, et al. The TLR3 signaling complex
forms by cooperative receptor dimerization. Proc
Natl Acad Sci USA 2008;105:258263.
78. Berg HC. The rotary motor of bacterial flagella.
Annu Rev Biochem 2003;72:1954.
79. Kojima S, Blair DF. The bacterial flagellar motor:
structure and function of a complex molecular
machine. Int Rev Cytol 2004;233:93134.
80. Samatey FA, et al. Structure of the bacterial
flagellar protofilament and implications for a
switch for supercoiling. Nature 2001;410:331
337.
81. Zipfel C. Pattern-recognition receptors in plant
innate immunity. Curr Opin Immunol
2008;20:1016.
82. Gangloff M. Different dimerisation mode for
TLR4 upon endosomal acidification? Trends
Biochem Sci 2012;37:9298.
83. Carty M, Goodbody R, Schroder M, Stack J,
Moynagh PN, Bowie AG. The human adaptor
SARM negatively regulates adaptor protein TRIFdependent Toll-like receptor signaling. Nat
Immunol 2006;7:10741081.
84. Kagan JC, Medzhitov R. Phosphoinositidemediated adaptor recruitment controls
Toll-like receptor signaling. Cell 2006;125:943
955.
85. Rowe DC, et al. The myristoylation of TRIFrelated adaptor molecule is essential for Toll-like
receptor 4 signal transduction. Proc Natl Acad Sci
USA 2006;103:62996304.

2012 John Wiley & Sons A/S

Immunological Reviews 250/2012

86. Ronni T, Agarwal V, Haykinson M, Haberland

ME, Cheng G, Smale ST. Common interaction
surfaces of the toll-like receptor 4 cytoplasmic
domain stimulate multiple nuclear targets. Mol
Cell Biol 2003;23:25432555.
87. Bovijn C, et al. Identification of interaction sites
for dimerization and adapter recruitment in Toll/
interleukin-1 receptor (TIR) domain of Toll-like
receptor 4. J Biol Chem 2012;287:40884098.
88. Abulrob A, et al. Nanoscale imaging of
epidermal growth factor receptor clustering:
effects of inhibitors. J Biol Chem
2010;285:31453156.
89. Germain RN. T-cell signaling: the importance of
receptor clustering. Curr Biol 1997;7:R640
R644.
90. Himanen JP, et al. Architecture of Eph receptor
clusters. Proc Natl Acad Sci USA 2010;107:
1086010865.
91. Stock J. Receptor signaling: dimerization and
beyond. Curr Biol 1996;6:825827.
92. Han SH, Kim JH, Martin M, Michalek SM, Nahm
MH. Pneumococcal lipoteichoic acid (LTA) is not
as potent as staphylococcal LTA in stimulating
Toll-like receptor 2. Infect Immun
2003;71:55415548.
93. Weiss J. Bactericidal/permeability-increasing
protein (BPI) and lipopolysaccharide-binding
protein (LBP): structure, function and regulation
in host defence against Gram-negative bacteria.
Biochem Soc Trans 2003;31:785790.
94. Martin TR, et al. Lipopolysaccharide binding
protein enhances the responsiveness of alveolar
macrophages to bacterial lipopolysaccharide
Implications for cytokine production in normal
and injured lungs. J Clin Invest 1992;90:2209
2219.
95. Mulero JJ, et al. Three new human members of
the lipid transfer/lipopolysaccharide binding
protein family (LT/LBP). Immunogenetics
2002;54:293300.
96. Beamer LJ, Carroll SF, Eisenberg D. Crystal
structure of human BPI and two bound
phospholipids at 2.4 angstrom resolution. Science
1997;276:18611864.
97. Tobias PS, Soldau K, Gegner JA, Mintz D,
Ulevitch RJ. Lipopolysaccharide binding proteinmediated complexation of lipopolysaccharide
with soluble CD14. J Biol Chem
1995;270:1048210488.
98. Dziarski R. Recognition of bacterial
peptidoglycan by the innate immune system. Cell
Mol Life Sci 2003;60:17931804.

99. Gregory CD, Devitt A. CD14 and apoptosis.

Apoptosis 1999;4:1120.
100. Kim JI, et al. Crystal structure of CD14 and its
implications for lipopolysaccharide signaling.
J Biol Chem 2005;280:1134711351.
101. Miyake K, et al. Mouse MD-1, a molecule that is
physically associated with RP105 and positively
regulates its expression. J Immunol
1998;161:13481353.
102. Fugier-Vivier I, et al. Molecular cloning of
human RP105. Eur J Immunol 1997;27:1824
1827.
103. Miyake K, Yamashita Y, Ogata M, Sudo T,
Kimoto M. RP105, a novel B cell surface
molecule implicated in B cell activation,
is a member of the leucine-rich
repeat protein family. J Immunol
1995;154:33333340.
104. Divanovic S, et al. Negative regulation of
Toll-like receptor 4 signaling by the Toll-like
receptor homolog RP105. Nat Immunol
2005;6:571578.
105. Divanovic S, et al. Regulation of TLR4 signaling
and the host interface with pathogens and
danger: the role of RP105. J Leukoc Biol
2007;82:265271.
106. Ogata H, et al. The toll-like receptor protein
RP105 regulates lipopolysaccharide signaling in B
cells. J Exp Med 2000;192:2329.
107. Nagai Y, et al. Requirement for MD-1 in cell
surface expression of RP105/CD180 and B-cell
responsiveness to lipopolysaccharide. Blood
2002;99:16991705.
108. Harada H, Ohto U, Satow Y. Crystal structure of
mouse MD-1 with endogenous phospholipid
bound in its cavity. J Mol Biol 2010;400:838
846.
109. Yoon SI, Hong M, Wilson IA. An unusual
dimeric structure and assembly for TLR4
regulator RP105-MD-1. Nat Struct Mol Biol
2011;18:10281035.
110. Yoon SI, Hong M, Han GW, Wilson IA. Crystal
structure of soluble MD-1 and its interaction
with lipid IVa. Proc Natl Acad Sci USA
2010;107:1099010995.
111. Ohto U, Miyake K, Shimizu T. Crystal
structures of mouse and human RP105/MD-1
complexes reveal unique dimer organization of
the toll-like receptor family. J Mol Biol
2011;413:815825.

229

This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy.
Users should refer to the original published version of the material.