Structures of T Cell Immunoglobulin Mucin
of Immune Responses by the TIM Receptor Family
Ce ´sar Santiago,1Angela Ballesteros,1Cecilia Tami,2Laura Martı ´nez-Mun ˜oz,1Gerardo G. Kaplan,2
and Jose ´ M. Casasnovas1,*
1Centro Nacional de Biotecnologia, CSIC, Campus Universidad Auto ´noma, 28049 Madrid, Spain
2Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, MD 20892, USA
The T cell immunoglobulin mucin (TIM) recep-
tors are involved in the regulation of immune re-
sponses, autoimmunity, and allergy. Structures
of the N-terminal ligand binding domain of the
murine mTIM-1 and mTIM-2 receptors revealed
an immunoglobulin (Ig) fold, with four Cys resi-
dues bridging a distinctive CC0loop to the GFC
b-sheet.The structuresshowed twoligand-rec-
ognition modes in the TIM family. The mTIM-1
sion interaction, whereas the mTIM-2 domain
ing. Biochemical, mutational, and cell adhesion
analyses confirmed the divergent ligand-bind-
features characteristic of mTIM-1 appear con-
served in human TIM-1, which also mediated
domain enhanced binding through the Ig do-
results explain the divergent immune functions
described for the murine receptors and the
role ofTIM-1 as acell adhesion receptor in renal
regeneration and cancer.
The genes coding for cellular receptors of the TIM family
(also known as the Tim family) locate in an airway hyperre-
activity regulatory locus linked to asthma and allergy sus-
ceptibility in mice (McIntire et al., 2001). Up to eight genes
have been described in mice and three in humans (Kuch-
roo et al., 2003), which code for at least four cellularrecep-
tors in mice (mTIM-1, mTIM-2, mTIM-3, and mTIM-4 [also
known as Tim-1, Tim-2, Tim-3, and Tim-4]) and for three
receptors in humans (TIM-1, TIM-3, and TIM-4). mTIM-2
is the only murine receptor that does not have a human
The TIM receptors are type I cell-surface glycoproteins
with an N-terminal Cys-rich region followed by a mucin
domain at the extracellular region, a single transmem-
brane region, and a cytoplasmic tail with phosphorylation
motifs except in the TIM-4 receptors. Sequence identity
among the N-terminal Cys-rich region of TIM homologs
thologs, it is around 60%. There are, however, marked dif-
ferences in the length of the threonine-, serine-, and pro-
line-rich mucin domain, with the number of O-linked
glycosylation sites ranging from 43 in mTIM-4 to 1 in
TIM-3 (Kuchroo et al., 2003).
The TIM gene family is involved in the regulation of im-
mune responses. mTIM-1 is preferentially expressed in
Th2 cells and delivers a signal that enhances T cell activa-
tion and proliferation, increasing airway inflammation and
allergy (Meyers et al., 2005; Umetsu et al., 2005). In con-
trast, mTIM-2 is an inhibitory molecule of Th2 immune
responses (Chakravarti et al., 2005; Rennert et al., 2006).
mTIM-3 is mainly expressed in Th1 cells and provides a
negative costimulatory signal that leads to immune toler-
ance (Sabatos et al., 2003; Sanchez-Fueyo et al., 2003).
Polymorphisms in mTIM-1 and mTIM-3 confer suscepti-
bility to the development of asthma and allergy (McIntire
et al., 2001).
Different ligands have been described for the murine
members of the TIM family. mTIM-2 binds to Semaphorin
4A and H-ferritin (Kumanogoh et al., 2002; Chen et al.,
2005). mTIM-4 expressed in antigen-presenting cells
binds to mTIM-1 and triggers T cell proliferation, prevent-
Similar phenotype was observed by crosslinking of
mTIM-1 receptor molecules with the 3B3 mAb, which ac-
tivated T cells and prevented the induction of respiratory
tolerance (Umetsu et al., 2005). There is less information
about the functions of the human TIM receptors. The
HAVCR1 gene that codes for TIM-1 was the first member
of the TIM family identified initially in monkeys and subse-
quently in humans as the hepatitis A virus cellular receptor
HAVCR1 is an important asthma determinant gene in
humans (McIntire et al., 2003), and its expression is upre-
gulated in acute kidney diseases and renal carcinoma
Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc. 299
(Han and Bonventre, 2004; Vila et al., 2004). The N-termi-
nal Cys-richdomainiscriticalforbindingof theTIMrecep-
tors to their ligands (Thompson et al., 1998; Sabatos et al.,
2003; Sanchez-Fueyo et al., 2003; Meyers et al., 2005),
and the mucin domain is required for binding of mTIM-1
to mTIM-4 (Meyers et al., 2005) and for the neutralization
of HAV particles by TIM-1 (Silberstein et al., 2003).
Currently, there is no structural information on the TIM
receptor domains, and it is unclear how they bind to their
ligands. The predicted immunoglobulin fold for the N-ter-
minal Cys-rich domain required further verification be-
cause of the unusually high number of conserved Cys res-
idues (six) for a single Ig domain. We pursued structure
determination of the Cys-rich region for several receptors
and determined the crystal structures of the N-terminal
domains of mTIM-1 and mTIM-2. Our work provides
a structural view on ligand-binding domains in the TIM
gene family. In spite of their high sequence identity
(62%), these two receptors displayed striking differences
in oligomerization and presentation of ligand-binding epi-
topes, which explain the reported divergence in ligand
recognition. The structures and derived functional data
defined two distinct ligand-recognition modes in the re-
ceptor family and identified a TIM-TIM intercellular inter-
action in mice and humans.
Structure of the N-Terminal Cys-Rich Domain
of TIM Receptors
We used X-ray crystallography to determine the structure
of the N-terminal Cys-rich domain of TIM family members.
We obtained functional domains of the mTIM-1 and
mTIM-2 receptors by using bacterial expression systems
and raised crystals diffracting at high resolution (Experi-
mental Procedures). The crystal structure of the N-termi-
nalCys-rich regionofmTIM-2was solvedfirstat1.5A˚res-
olution, and the mTIM-1 structure was subsequently
determined to a resolution of 2.5 A˚(Table 1). The struc-
tures revealed an Ig domain belonging to the V set (IgV),
related to the N-terminal domains of the CD4 and CAR
(coxsackievirus and adenovirus receptor) cellular recep-
tors (highest Z score in DALI search) (Holm and Sander,
1993). The IgV domains have two antiparallel b sheets
with particularly short b strands B, E, and D in one face
(BED b sheet) and the A, G, F, C, C0, and C00b strands in
the opposite one (GFC b sheet) (Figures 1A and 1B). A
Pro residue found prior to the first Cys in all TIM receptor
domains was responsible for the short length of the
b strand B (Figure 1D). The first and last Cys residues in
the N-terminal domain of the TIM receptors bridged the
Table 1. Data Collection and Refinement Statistics
Native mTIM-2Se-Met mTIM-2Native mTIM-1
Cell dimensions a, b, c (A˚)61.5, 60.4, 69.144.5, 55.7, 75.5
a, b, g (?)90, 115.9, 9090, 90, 90
7.4 (23.5)5.0 (7.4) 8.4 (19.0)
I/sI6.0 (3.0)9.9 (9.1)7.0 (4.0)
Completeness (%)98.6 (99.6)100 (100) 99.8 (100)
Redundancy3.5 (2.9)15.9 (14.4)5.3 (5.5)
No. atoms Protein17101818
Rms deviations Bond lengths (A˚) 0.00520.0078
Bond angles (?)1.621.75
Diffaction data statistics at the highest resolution shell are shown in parentheses. B factors for the molecules A and B found in the
asymmetric unit of the crystals are included.
Structure-Function of TIM Receptors
300 Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc.
b sandwich as in most Ig domains, whereas the other four
Cys residues characteristic of the TIM family formed two
disulphide bonds that link a long CC0loop to the GFC
b sheet (Figure 1).
The mTIM-1 and mTIM-2 N-terminal domains share
66% sequence identity and high structural similarity
(Figure 1C). The rms deviation between the two structures
was0.9A˚,andthedeviationbetween thetwomolecules in
the asymmetric unit of the crystals was about 0.5 A˚. The
superimposed mTIM-1 and mTIM-2 structures showed
just three misaligned regions (Figures 1C and 1D): the
BC and FG loops and the interdisulphide region of the
CC0loop. The mTIM-1 BC loop was one residue shorter
than the mTIM-2 loop and it did not have a helical confor-
mation. The extended and hydrophobic mTIM-1 FG loop
structure is more representative of the family than that of
mTIM-2 (Figure 1D).
The mTIM-1 and mTIM-2 structures defined a common
architecture for the related N-terminal domains of TIM re-
ceptors, with structural variability circumscribed to the
loops at the top of the domain. The two disulphide bonds
on the GFC b sheet are a distinctive structural feature of
TIM IgV domains.
CC0Loop Conformation Variability in the TIM
and Related Ig Receptors
The interdisulphide region of the CC0loop was remarkably
different in the homologous mouse receptors (Figure 1C).
In mTIM-2, the tip of the loop folds down and had a helical
conformation, whereas in mTIM-1 it extended up onto the
GFC b sheet. These differences arise from distinct con-
tacts between residues in the loop and the b sheet (Fig-
ure 2). In mTIM-2, the aromatic ring of Tyr39 was located
phobic side chains of Val89 and Phe98, whereas the pre-
ceding residues projected toward the solvent (Figure 2A).
The van der Waals interactions were not sufficient to fix
the conformation of the interdisulphide region, so that
Figure 1. Crystal Structures of the N-Terminal Cys-Rich Domain of TIM Receptors
(A and B) Ribbon diagrams of the mTIM-1 and mTIM-2 structures are shown in (A) and (B), respectively. b strands of one face are red and those in the
opposite side are magenta, coil is orange, 310helix is light blue, a helix in the BC loop is blue, and the loop between C and C0b strands is light green.
Cys residues and disulphide bonds are in green. Strands are labeled with uppercase letters and terminal ends (n and c) are in lowercase.
(C) Stereo view of superimposed mTIM-1 (red) and mTIM-2 (magenta) structures, with the regions showing structural variability labeled.
(D) Structural alignment of the mTIM-1 and mTIM-2 structures with residues closer than 3 A˚aligned. b strands and helical regions defined by the pro-
gramdssp(Wolfgangand Sander,1983)arecoloredasin(A)and(B).TheotherTIMreceptor domainswerealignedbysequence.Conservedresidues
in most TIM receptors are colored in yellow and the six Cys residues in green. N-linked glycosylation sites are underlined and sequences of mTIM-1
and mTIM-2 numbered. Green lines join the disulphide-linked Cys residues.
Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc. 301
Structure-Function of TIM Receptors
the tip of the solvent-exposed mTIM-2 CC0loop remained
flexible and poorly defined in the electron density maps
In mTIM-1, the conformation of the CC0loop tip was
fixed by interactions with the Arg88 and Lys99 residues
at the b strands F and G, respectively (Figure 2B). Their
side chains hydrogen bonded to main-chain oxygen
atoms of Pro35, Ser36, and Ala38 in the two molecules
of the asymmetric unit. So, the disulphide-bridged CC0
loopwas additionallylinkedto theupperhalfofthebsheet
bytheconserved Arg88 andLys99residues inthe mTIM-1
structure. These basic residues are conserved in all pri-
mate and murine TIM receptors, but they are absent in
mTIM-2 (Figure 1D), which has a distinct CC0loop confor-
mation from mTIM-1.
In addition to the described variations between the CC0
loop of mTIM-1 and mTIM-2, there were also significant
differences in the conformation of the neighboring FG
loop (Figure 1C). The FG loop of mTIM-1 is extended by
two additional aromatic residues, Trp94 and Phe95 (Fig-
ure 1D). The side chain of Phe95 came close (?4A˚) to the
tip of the CC0loop in the mTIM-1 structure (Figure 2B). In-
terestingly, the mTIM-1 FG loop is conserved in TIM-1 and
in the human and mouse TIM-4 receptors, and the length
of the loop in human TIM-3 and its mouse ortholog
ities suggest that the conformational epitope built by the
CC0and FG loops (CC0FG epitope) on the GFC face of the
mTIM-1 domain could be conserved in the TIM receptor
family, except for mTIM-2 (Figure 2A). Flexibility in the
long FG loop could affect the epitope conformation.
The loops connecting the C and C0b strands are largely
divergent both in length and conformation among V do-
mains of Ig receptors (Tan et al., 2002). The CC0loop in
related CD4 receptor (not shown), and it had similar length
but different conformation than in the homologous CAR
domain (Figure 2C). In CAR the CC0loop adopts an ex-
tended conformation, similar to other IgV domains where
the GFC face is engaged in ligand recognition (Jones
et al., 1992; Wang et al., 1999; van Raaij et al., 2000; Kos-
trewa et al., 2001), whereas in the TIM and CEA V do-
mains, the loop folded back onto the b sheet (Figure 2C).
In the CEA domain, the CC0loop packs against an aro-
matic residue on the C strand (Tan et al., 2002), whereas
in the TIM receptors, it is bridged by two external disul-
phide bonds to the GFC b sheet.
of the TIM IgV domains will prevent extended face-to-face
intermolecular interaction through this side of the domain,
observed with related Ig receptors (Jones et al., 1992;
Wang et al., 1999; van Raaij et al., 2000; Kostrewa et al.,
Figure 2. Conformation of the Loops Connecting C and C0
b Strands in the TIM Structures and in Related IgV Domains
(A and B) Ribbon diagram of the GFC face of the mTIM-2 (A) and
mTIM-1 (B) domain structures. Insets show lateral views. Residues
between the two external disulphide bonds in the CC0loop and inter-
acting residues at the F and G b strands and at the FG loop are yellow.
Hydrogen bonds are shown as pink dashed cylinders. Oxygens and
nitrogen atoms are in red and blue, respectively. Residues are labeled
according to Figure 1D. Blue dots on the C0C00loop of (B) indicate
location of N-linked glycosylations in human and monkey TIM-1 and
(C) Stereo view of the superimposed mTIM-1 (red) and CAR (green,
1f5w) homologous domains. The CC0loop region of the CEA IgV do-
main (1L6Z) is blue.
302 Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc.
Structure-Function of TIM Receptors
such as that of mTIM-2 could lead to differences on ligand-
binding specificity among receptors of the TIM family.
A Dimeric Structure for mTIM-2
Two mTIM-2 IgV domains built up the asymmetric unit of
the crystals (Figure 3A). The angle between the two
mTIM-2 domains was ?60?(Figure 3A), similar to intermo-
lecular angles reported in structures showing dimerization
in cis of receptors linked to the same cell surface (Casas-
novas et al., 1998). The buried surface area by the domain
association was of 775 A˚2per monomer, close to that
reported for antigen-antibody complexes (800 A˚2) (Janin,
tion revealed also self-association of the mTIM-2 IgV do-
mains at high protein concentration in solution (Figure 3C
and see Figure S1 in the Supplemental Data available
online). Crosslinked dimers were observed with both
mTIM-2 and mTIM-1 in conditions where dimerization
was also detected for an ICAM-1 protein known to dimer-
ize at high protein concentrations (Miller et al., 1995; Ca-
sasnovas et al., 1998). No such dimers were seen with
tion of the crystallized mTIM-2 protein (13 kDa) provided
solution molecular weights of 16 and 21 kDa at low and
high protein concentrations, respectively (Figure S1),
Figure 3. N-Terminal Domain Interactions in the TIM Receptors
neighboring molecules are in yellow. Side chains of residues contributing to the dimer interfaces are included and some central residues are labeled.
Acetateligand found in the mTIM-2 structure is black, water molecules are red spheres, and hydrogen bonds are pink dashed cylinders. Asn residues
to which glycans link in mTIM-2 are green. Arrows represent the hypothetical interaction of O-linked glycans from the C-terminal mucin domain with
residues at the b strand A, BC, and FG loops of the interacting mTIM-1 domains (see also Figure S3).
(C) Self-association of the N-terminal IgV domains in solution. SDS-PAGE under reducing conditions of mTIM-1, mTIM-2, and mTIM-4 domains un-
and asoluble fragmentof CD46are also included. Sizeand migration of themolecular weightmarker isindicated. Crosslinked dimers arelabeled with
an asterisk. No dimerization of the mTIM-4 IgV domain is seen here or in the protein crystals (not shown).
(D) Structural alignment with residues at the dimer interface in yellow and those at the center of the interacting molecules in blue. b strands are rep-
resented by lines.
Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc. 303
Structure-Function of TIM Receptors
indicating IgV domain oligomerization with increasing pro-
tein concentration. The determined dimerization KDfrom
sedimentation was ?300 mM.
the IgV domain, with the helical BC loop of one molecule
embracing the G strand of the neighboring domain
(Figure 3A). Main intermolecular contacts included resi-
dues following the b strand B, such as the conserved
Pro15, the helical BC and CC0loops from the two interact-
ing molecules, and residues on the FG loop and upper half
of the b strand G (Figure 3D). The His97 residue that be-
gins the b strand G was about the center of the dimer in-
terface (labeled in Figure 3A and blue in Figure 3D). The
hydrophobic cavity below the His97 side chain was occu-
pied by an acetyl ligand, hydrogen bonded to the two
neighboring histidine residues (black in Figure 3A, see
Figure S2), and a network of water molecules fill up the
cavity over His97 (Figure 3A). Almost 50% of the total di-
merization surface was buried by the helical BC loop
(365 A˚2) that sits on the b strand G and approached the
CC0loop of the neighboring molecule (Figure 3A). The
His21 side chain on the BC loop stacked over the long
Arg36 side chain in the CC0loop, and the hydrophobic
Leu22 at the tip of the BC loop inserted into a hydrophobic
pocket built by Cys35, Tyr39, and Phe98 (Figure 3A). Po-
tential N-linked glycosylation sites lay below the interdo-
main interface and might contribute to the stability of the
mTIM-2 dimer. Therefore, IgV domain dimerization cre-
ates an extended glycan-free surface at the top of the
macromolecular complex highly accessible to ligands.
Interdomain Interactions in the mTIM-1
mTIM-1 IgV domain interactions were observed in the
crystals and in solution, in contrast to the homologous
mTIM-4 domain (Figure 3 and Figure S1). The association
of the two IgV domains in the asymmetric unit of the
mTIM-1 crystals was remarkably different from mTIM-2.
The mTIM-1 domains were related by a rotation angle of
about 180?and had their C-terminal ends extending to-
ward opposite directions (Figure 3B), which was sugges-
tive of an intermolecular interaction between cell-surface
receptors on opposite cell surfaces (Wang et al., 1999).
The domains interact through the upper half of the BED
face, including residues following the short b strand B
and at the DE loop (Figures 3B and 3D). Two Thr17 resi-
dues from opposite molecules were hydrogen bonded at
the center of the dimer interface. Hydrophobic contacts
included Thr13 and Pro15 with the bulky Tyr21 side chain
at the BC loop of a neighboring domain. Additional inter-
acting sites between the two mTIM-1 domains engaged
several residues at the long DE coil (Figures 3B and 3D).
Asp69 bound to Ser19 on the BC loop of the opposite do-
main. Also, His64 and Glu67 were involved in interdomain
interactions. Although these two residues at the helical DE
loop are not conserved, some charged residues alternate
at the aligned positions in the TIM receptors (Figure 1D).
They could provide certain specificity for binding of TIM
receptors in trans.
The interaction through the BED side of the mTIM-1 do-
mains buried around 450 A˚2per monomer, significantly
less than that observed for the mTIM-2 IgV domains.
These differences correlated with the lower dimerization
constant inferred from the sedimentation experiments,
where a lower solution molecular weight was determined
for the mTIM-1 IgV domain at high protein concentration
(Figure S1). The buried surface was also lower than the
600 A˚2observed in intermolecular interactions between
IgV domains of receptors binding through the GFC face
(Jones et al., 1992; Wang et al., 1999; Kostrewa et al.,
2001). Occupancy of a cavity between the b strand A,
FG, and BC loop of the interacting domains could further
stabilize the receptor interaction (Figure S3). In the crys-
tals, the cavity was occupied by the C-terminal end of
a symmetry-related domain. In the native mTIM-1 recep-
tor, O-linked glycans from the contiguous mucin domain
could penetrate in the interdomain cavity and bridge inter-
acting receptors from opposite cells (Figure 3B). Potential
glycan-interacting residues such as the basic Lys5 and
Arg22, and the conserved Pro92, Trp94, and Asn96 on
the FGloop lay on the upperedge ofthe cavity(Figure S3).
Figure 4. Oligomerization of the mTIM-2 Receptor
(A) Size exclusion chromatography of soluble mTIM-1 (open triangles)
and mTIM-2 (open squares) molecules with the complete extracellular
regions. mTIM2-BCt1 corresponds to a mutant mTIM-2 receptor
where BC loop residues (HLG) were replaced by aligned residues in
the mTIM-1 BC loop (YR). Percentage of the total optical density
(OD) is plotted for each elution fraction. Exclusion volume and size of
the molecular weight markers are indicated. A representative experi-
ment is shown.
(B) Immunoblot of BS3treated (+) or untreated (?) cell supernatants
lacking (mock) or having the indicated TIM receptor (see Experimental
Procedures). The arrows mark the three mTIM-2 oligomeric species.
Size and migration of the molecular weight marker are shown.
304 Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc.
Structure-Function of TIM Receptors
Although intercellular binding of mTIM-1 to mTIM-4 has
been described in the past (Meyers et al., 2005), the
mTIM-1 structure provided an indication of homophilic
binding in the TIM family.
mTIM-2 Receptor Oligomerization
To understand further the relevance of the mTIM-2 dimer
structure and the organization of the mTIM-2 receptor on
the cell surface, we analyzed oligomerization of the com-
plete extracellular region (IgV and mucin domains) of the
receptor molecule. Soluble mTIM-2 receptors secreted
to cellsupernatantswereanalyzed bysizeexclusion chro-
matography and chemical crosslinking (Figure 4). The
mTIM-2 molecules eluted with an apparent molecular
weight ?160 kDa (Figure 4A), whereas the size of the re-
clusion sizeof the homologous mTIM-1 receptor molecule
loop residues (HLG, 25% of the buried surface) of mTIM-2
for the aligned residues of mTIM-1 (YR) reduced signifi-
cantly (about 70%) the amount of mTIM-2 oligomer (Fig-
ure 4A), showing contribution of the BC loop to the mTIM-2
omerization of the mTIM-2 receptors was confirmed by
crosslinking experiments, where BS3-treated samples
periment was done under nonsaturating crosslinker con-
centration, most of the receptor molecules migrated as
monomer (40 kDa) in the denaturing gel. Heterogeneity re-
lated to O- and N-linked glycosylation could account for
the broad bands of the soluble TIM proteins. Crosslinked
mTIM-2 receptor oligomers that had molecular weights
around 80, 120, and 150 kDa were seen in the BS3-treated
ate oligomeric forms (80 and 120 kDa) could come from
main of mTIM-2 dimerized in the crystals and in solution, it
appears that the formation of the mTIM-2 receptor tetra-
mer required the mucin domain. The absence of mTIM-1
oligomers suggested a divergence in the organization of
the cellular receptors on the cell surface.
Homophilic mTIM-1 Receptor Interaction
To analyze the relevance of the TIM-TIM interaction
showed by the mTIM-1 structure, we carried out both pro-
Figure 5. Homophilic mTIM-1 Receptor Interaction
(A) Binding of soluble Fc fusion proteins to plastic-coated mTIM-1 pro-
tein with the complete extracellular region. Binding to the isolated IgV
domain is shown in Figure S4. TIM-Fc (mTIM-1 and mTIM-2) and con-
trol ICAM-1-Fc (IC1-2D) proteins used are included in the legend.
Binding at the indicated protein concentration was determined from
the optical density (OD) at 492 nm (see Experimental Procedures).
Mean ± SD of three experiments is shown.
(B) Normalized binding of mTIM-1-Fc protein to plastic-coated
mTIM-1 IgV in the absence (mTIM-1) or presence of mTIM-1 (T1.4
and T1.10) and mTIM-2 (T2.1) antibodies, or EDTA (10 mM). Binding
of a mutant mTIM-1-Fc protein where His64 was replaced by a gluta-
ments carried with 20 and 10 mg/ml of Fc protein is shown.
(C) 293T cells transfected with the indicated TIM protein tagged with a
cyan fluorescent protein (CFP) at their C-terminal ends. Fluorescence
images of individual (white framed insets) and adhered cells were
acquired 2 days after transfection (see Experimental Procedures).
DIC image of the cell cluster with three transfected cells (t) and one
cell untransfected (ut) or lacking mTIM-1 is shown in the inset.
(D) Fluorescence and DIC images of adhered 300.19 preB-cells trans-
fected with CFP-tagged mTIM-1 are on the top. Individual transfected
cells untreated or treated with 1 mg/ml of ionomycin for 30 min at 37?C
are in the bottom (see also Figure S4).
Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc. 305
Structure-Function of TIM Receptors
protein binding assay, we monitored binding of soluble
mTIM-1-Fc fusion proteins with the IgV and mucin do-
mains to plastic-coated mTIM-1 proteins having either
the isolated IgV domain used in crystallization (Figure S4)
or the complete extracellular region of the receptor
shown). Homophilic mTIM-1 binding was specifically
blocked by the T1.10 mAb that recognizes the IgV domain
and by the addition of EDTA (Figure 5B), which indicated
requirement of divalent cations for high-affinity binding
cin domain. Interestingly, the structure-guided His64Glu
significantly the homophilic mTIM-1 binding (Figure 5B),
showing a critical contribution of the DE loop and agree-
ment with the mTIM-1 structure (Figure 3B). This mutation
decreased moderately (40%) binding of mTIM-1 to
mTIM-4 (not shown), suggesting binding through the
BED face as well. Determined affinity for the homophilic
103times higher than that inferred from sedimentation ex-
periments with the isolated IgV domain. These differences
ions for homophilic mTIM-1 binding, although preliminary
observations indicated that the contribution of the mucin
domain was more critical for mTIM-1 binding to mTIM-4
than for the homophilic interaction.
The involvement of the mTIM-1 and mTIM-2 receptors
in cell-cell interactions was studied with receptor mole-
cules that had a fluorescent protein tagged to their C-ter-
minal intracellular domains (Figures 5C and 5D). mTIM-2
fluorescent proteins were on the surface of 293T and
lymphoid cells, whereas most mTIM-1 accumulated at
intracellular vesicles. However, treatment of the 300.19
lymphocytes with ionomycin or PMA enhanced localiza-
tion of mTIM-1 on the cell surface (Figure 5D and Fig-
ure S4). mTIM-2 distributed quite homogenously on the
surface of both isolated and interacting cells, and additive
fluorescence was seen atintercellular junctions. However,
fected cells(Figures 5C and5D). No increase in cellular re-
ceptor was observed at the junctions of transfected and
showed a relevant homophilic intermolecular interaction
for the mTIM-1 receptor at intercellular junctions, not
seen with mTIM-2. The efficient trafficking of mTIM-2 to
the cell surface could be related to its oligomeric nature,
whereas productive transfer of mTIM-1 to the cell surface
must require rearrangements in the receptor domains in-
duced by increasing calcium concentration.
Conservation of the mTIM-1 Structure
and the Homophilic Interaction in Humans
be conserved in the human TIM-1 IgV domain. The Arg88
structure are conserved in TIM-1 (Figures 1D and 2B).
Moreover, the sequence of the FG loop is conserved
also in the human and mouse TIM-1 receptors.
The presumed structural similarity between the human
and mouse TIM-1 IgV domains suggested conservation
of the receptors functions. Therefore, homophilic TIM-1
binding was studied both with soluble and cell-surface-
expressed receptor molecules. TIM-1-Fc proteins bound
to TIM-1 molecules containing the complete extracellular
portion of the receptor (Figure 6A). As shown with the
murine receptor, mutations in the DE loop of the TIM-1
IgV domain affected the homophilic interaction, suggest-
ing binding through the BED face as well. Interestingly,
TIM-1 bound with lower affinity than mTIM-1. This could
be related to residue substitutions at the interacting sur-
faces, such as the Tyr21 in mTIM-1 for Ala in the human
receptor (Figure 1D), which will reduce hydrophobic
contacts. Beads coated with the human TIM-1-Fc protein
bound specifically to TIM-1 receptors expressed on the
cell surface (Figure 6B). High-affinity binding required
both IgV and mucin domains, as seen with mTIM-1.
HAV Binding to TIM-1 Receptors
HAV binds specifically to the N-terminal IgV domain of the
human and monkey TIM-1 receptors (Kaplan et al., 1996;
Feigelstock et al., 1998b), although no binding to mTIM-1
has been detected (unpublished results). The expected
structural similarity between the primate and mouse N-
terminal domains allowed us to define a virus-binding
Figure 6. Homophilic TIM-1 Receptor Interaction in Humans
(A) Binding of soluble Fc fusion proteins to plastic-coated TIM-1 pro-
teins with the complete extracellular receptor as in Figure 5A. Binding
of the Arg65Glu TIM-1 mutant or a double mutant Asp62His-Arg65Glu
(TIM1-DEt1) in the DE loop of the IgV domain are included.
(B) Binding of latex beads coated with the indicated protein to cells ex-
pressing GFP, the complete TIM-1 receptor, or a mutant lacking the
IgV domain (TIM-1DIgV).
306 Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc.
Structure-Function of TIM Receptors
surface based on a gene polymorphism in monkey TIM-1
that abolished binding of a protective mAb (190/4) (Fig-
ure 7; Feigelstock et al., 1998a). This antibody blocks
HAV receptor binding and protects cells from infection.
The antigenic variant Lys to Gln aligns with Glu90 in
mTIM-1 (blue in Figure 7), nearby the CC0FG epitope,
and located the virus-binding surface on the GFC face of
the IgV domain. A Glu residue is also found in human
TIM-1, which binds to HAV but is not recognized by the
190/4 antibody. The protruding conformation of the
mate TIM-1 receptors, which have aromatic residues both
in the FG and CC0loops (Figure 1D), could be suited for
HAV recognition. Moreover, the conservation of the FG
loop in the primate and mouse TIM-1 IgV domains indi-
cated that the enhanced hydrophobicity of the CC0loop
in TIM-1 could determine its virus-binding specificity
(Figure 1D). The unique Phe residue in the primate recep-
could in fact be a critical virus-binding residue, as de-
scribed for a hydrophobic residue at the homologous
loop in the CEA coronavirus receptor (Tan et al., 2002).
The crystal structures of the N-terminal ligand-binding
domain of two TIM family members presented here have
provided relevant insights for understanding the immune
functions of these receptors. The structures of the related
mTIM-1 and mTIM-2 showed marked differences in pre-
sentation of ligand-binding epitopes, suggestive of two
distinct modes of ligand recognition. The lack of a human
ortholog for mTIM-2 and the unique structural features of
its IgV domain suggest an evolutionary divergence in mice
and the conservation of mTIM-1-like structures in
Dimerization of the N-terminal domain of mTIM-2 buries
the domain surface engaged in homophilic mTIM-1 inter-
actions, preventing mTIM-2 binding to mTIM-1 as well as
homophilic mTIM-2 binding. Preliminary observations
showed that disruption of the mTIM-2 dimer allowed bind-
ing to mTIM-1 (not shown). Oligomerization of the mTIM-2
receptor on the cell surface will facilitate binding to multi-
valent ligands (Kumanogoh et al., 2002; Chen et al., 2005).
Differences on mTIM-1 and mTIM-2 organization on the
cell surface could deliver distinct intracellular signals that
lead to either activation or inhibition of Th2 immune re-
The ‘‘extra’’ four Cys residues characteristic of the IgV
domain in the TIMs fix the folded conformation of the
long CC0loop onto the GFC b sheet, defining a distinctive
structural feature of the TIM family. The conformation of
the loop in mTIM-2 appears to be unique within the TIM
family, whereas that of mTIM-1 could be shared by other
TIM receptors. Conformation of the CC0loop in several
mTIM-4 IgV domain crystal structures was almost identi-
cal to that described here for mTIM-1 (not shown). The
CC0loop appears structurally connected to the FG loop
in mTIM-1, building up a protruding CC0FG epitope that
partially covers the GFC b sheet. This feature suggests
a divergence in ligand-recognition modes between the
TIMs and related Ig receptors such as CAR, where a flat
GFC face is engaged in intermolecular interactions (Jones
et al., 1992; Wang et al., 1999; van Raaij et al., 2000; Kos-
trewa et al., 2001).
The crystal structure of mTIM-1 identified a new homo-
philic TIM-TIM receptor interaction in mice and humans
that could be relevant for the regulation of immune func-
tions by these receptors. Moreover, the observed traffick-
ing of mTIM-1 to the cell surface upon lymphocyte stimu-
lation revealed a new regulatory mechanism of TIM
receptor functions. Engagement of mTIM-1 on the T cell
surface bydifferent ligands triggers a cell regulatory signal
that has been linked to critical immune reactions (Meyers
et al., 2005; Umetsu et al., 2005; Mesri et al., 2006). The
observed clustering of mTIM-1 by homophilic binding at
intercellular junctions could facilitate phosphorylation of
its cytoplasmic tail, which provides a costimulatory signal
for T cell activation (de Souza et al., 2005). Therefore, the
homophilic mTIM-1-binding interaction described here
could have important implications in the regulation of im-
mune processes both in mice and humans and could be
Figure 7. Ligand-Binding Surfaces in the IgV Domain of TIM-1
Surface representation of the mTIM-1 domain structure. Surface in-
volved in the homophilic interaction is pink. Residues in a conforma-
tional epitope built by the tip of the long CC0loop and the FG loop
onto the GFC b sheet are colored red and orange, respectively. The
surface where an mkTIM-1 polymorphism (Lys88Gln) has been map-
ped is in blue. The mutation identified the side of the domain recog-
nized by a mAb blocking HAV binding to its mkTIM-1 receptor (Feigel-
stock et al.,1998a).Surface corresponding to the Asnresidueto which
glycans will be linked in the primate TIM-1 receptors is green.
Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc. 307
Structure-Function of TIM Receptors
responsible of the hyperproliferation of T cells observed in
mice treated with soluble mTIM-1 molecules (Meyers
et al.,2005). mTIM-1isexpressed on the surface of Bcells
and activated T cells (Meyers et al., 2005), so it is feasible
that the homophilic mTIM-1 interaction mediates B-T cell
adhesion interactions and has important implications in
the regulation of immune responses. The conservation of
the homophilic mTIM-1 receptor interaction both in mice
and humans supports a conserved role in B-T cell cross-
talk and its relevance in the immune system. Moreover,
murine mTIM-1 and human TIM-1 receptors are overex-
pressed after ischemic kidney injury (Han and Bonventre,
2004), and human TIM-1 is overexpressed in renal carci-
noma (Vila et al., 2004). The homophilic interaction de-
sion interactions relevant for renal regeneration and tumor
development. mTIM-1-related functions must be regu-
lated also by receptor trafficking to the cell surface, which
we show here is enhanced by increase of intracellular cal-
cium amounts. This regulatory step could be shared by
other receptors of the TIM family.
The homophilic mTIM-1 receptor binding pictured by
the crystal structure revealed a striking difference with
those mediated by related receptors. The IgV domains
contact through their BED faces, opposite the GFC face
commonly used for ligand binding by Ig receptors, which
displayed a distinctive CC0FG epitope in mTIM-1. Al-
though homophilic binding engaged the N-terminal IgV
domain, experiments shown here suggested that carbo-
hydrates from the contiguous mucin domain contribute
to the interaction. O-linked glycans from the mucin do-
main could participate in TIM receptor interactions by oc-
cupying cavities generated upon N-terminal domain bind-
ing, such as that seen between the interacting mTIM-1
domains. Furthermore, polymorphisms in the mucin do-
main near the end of the IgV domain (McIntire et al.,
2001) might have also some influence on the contribution
of the O-linked glycans to the homophilic binding interac-
tions, whereas those close to the membrane could modu-
late receptor oligomerization on the cell surface as well as
intracellular receptor trafficking.
In summary, the crystal structures of the related murine
mTIM-1 and mTIM-2 receptors presented here provided
the structural basis for understanding ligand-binding di-
versity and the immune regulatory role of the TIM gene
family. In contrast to the structural and functional redun-
dancy observed in other receptor families (Wang and
Springer, 1998), the structural divergence between the
ences in their ligand-binding specificities and functions.
The lack of a human mTIM-2 ortholog suggests that
mTIM-2-related functions in humans must be carried out
by a cellular receptor from a related family. In contrast, se-
quence similarity in the mTIM-1 structural motifs between
primate and murine TIM receptors suggests a conserved
ligand-binding mode in the family, as well as possible ho-
mophilic TIM-TIM interactions with other TIM receptors.
Conservation of the homophilic TIM-1 binding in mice
and humans suggests a critical implication in the regula-
tion of immune responses, although further investigation
is now needed to determine the biological significance
of this adhesion process. Moreover, structural insights
coming from glycosylated TIM receptor are required for
understanding the precise role of glycosylation in ligand
recognition. Research extending the current knowledge
on this relevant receptor family will guide the design of
small molecules capable of regulating their functions in
the immune system, preventing development of asthma,
autoimmune diseases, and hepatitis.
Antibodies and cDNAs
TIM mAbs were obtained from eBioscience, Inc. The full-length cDNA
coding for mTIM-1 was obtained from mouse EST #AA547594 derived
from a Knowles Solter mouse 2-cell embryo cDNA library (IMAGE con-
sortium, ATCC).The cDNA coding for full-length mTIM-2 was obtained
from EST #AA509542 derived from a C57BL/6J mouse mammary
gland cDNA library (IMAGE consortium, ATCC).
Protein Sample Preparation for Crystallization
Bacterial expression of the Cys-rich domains cloned into the unique
NdeI and XhoIsitesof thepET-27b vector (Novagen) gave insoluble in-
clusion bodies. However, soluble receptor domains were prepared by
in vitro refolding of the inclusion bodies as described elsewhere (Jime-
nez et al., 2005). The mTIM-1 and mTIM-2 domains had an N-terminal
Met and residues 20 to 130 and 129 of the precursor proteins, respec-
tively (McIntire et al., 2001), a thrombin recognition site, and two pro-
tein tags included in the vector. The soluble proteins eluted from
a Superdex-75 column (Amersham Biosciences) with the expected re-
tention volume(15–20 kDa)and wererecognizedbythecorresponding
TIM monoclonal antibodies (not shown). The recombinant proteins
were thrombin treated to release C-terminal tags and further purified
by ion-exchange chromatography.
mTIM-1 and mTIM-2 Crystallization
and Structure Determination
Crystals were initially raised with the mTIM-2 protein at 12 mg/ml by
the hanging drop method and with a crystallization condition of 30%
PEG-2000 methylether, 5% PEG-400, 0.2 M ammonium sulfate,
0.1 M sodium acetate (pH 4.6) and ?4% 1,2,3 heptanetriol. The
mTIM-2 crystals belong to the monoclinic C2 space group, and they
have two molecules in the asymmetric unit and 45% solvent content.
Plate-like crystals were raised with the mTIM-1 protein domain by
very similar crystallization conditions to those used for mTIM-2. The
crystals belong to the orthorhombic P212121 space group and have
two independent molecules in theasymmetric unit and about 37%sol-
vent content. Details on structure determination are included in Sup-
plemental Data; diffraction data and refinement statistics are shown
in Table 1. The refined models contain all 116 amino acid residues of
the mTIM-1 protein construct and all 115 amino acid residues of the
crystallized mTIM-2 protein for molecule B, but the five N-terminal
and the three C-terminal residues are missing for molecule A. The N-
terminal residue of mTIM-1 in Figure 1D corresponds to Tyr4 in the
pressed mTIM-2 receptor protein corresponds to His4 in the Pdb file.
Protein Expression in Mammalian Cells
DNAs coding forthe complete extracellularregion of theTIM receptors
followed by a thrombin recognition site were cloned upstream of
a hemagglutinin A epitope (TIM-HA) or the IgG1-Fc region (TIM-Fc)
in the pEF-BOS expression vector (Jimenez et al., 2005). Serum-free
cell supernatants with HA and Fc-tagged soluble receptor proteins
were prepared by transient expression in 293T cells and protein con-
centration (10–50 mg/ml) determined by ELISA (Jimenez et al., 2005).
308 Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc.
Structure-Function of TIM Receptors
Mutagenesis was done by the overlapping PCR technique and con-
firmed by DNA sequencing. Fluorescent-tagged proteins at their cyto-
plasmic tails were expressed in 293T cells and in the murine 300.19
pre-B cell line by transfection with recombinant mTIM-1 and mTIM-2
cDNAs cloned in-frame with a cyan fluorescent variant of GFP (CFP)
pus IX81 confocal microscope. Confocal fluorescence and differential
interference contrast (DIC) images were acquired and superimposed
with the FV10-ASW 1.4 software. Fluorescent proteins were localized
by excitation with a 405 nm line.
Analysis of receptor oligomerization by size-exclusion chromatogra-
phy was carried with TIM-HA proteins in serum-free media. Cell super-
natants collected 1 day after 293T transfection with the pEF-TIM-HA
construct were concentrated five times and run through a Super-
dex200 column with HBS buffer (20 mM HEPES and 100 mM NaCl
[pH 7.5]) and 2.5 mM CaCl2. The proteins in the elution fractions
were detected byELISA withHAand TIM antibodies.Molecular weight
markers were run under the same conditions. Analysis of TIM-HA olig-
omerization by chemical crosslinking was done by BS3((Bis(sulfosuc-
cinimidyl) suberate) (Pierce) treatment of the proteins in cell superna-
tants collected 3 days after transfection. After overnight treatment at
4?C, the reaction was quenched with 50 mM Tris (pH 7.5), and the pro-
teins were immunoprecipitated with anti-HA mAb (12CA5) and protein
A-Sepharose and resolved by 8% SDS-PAGE under reducing condi-
tions. Proteins were detected by immunoblot with the HA antibody
and the ECL detection system (Amersham Biosciences). Chemical
crosslinking of the isolated TIM IgV domains and control proteins at
5 mg/ml was carried with BS3for 1 hr at 4?C in HBS buffer. The sam-
ples were analyzed by 12% SDS-PAGE.
TIM-TIM Binding Assays
Binding of soluble Fcfusion proteins to plastic-coated IgVdomain pre-
pared in bacteria and TIM-1-HA proteins prepared in mammalian cells
wascarriedinduplicatewellsof 96-well platesas described elsewhere
(Jimenez et al., 2005). Binding data were corrected by the background
binding monitored in wells without coated proteins. A control ICAM-1-
in cell supernatants supplemented with 5% FCS were diluted with
binding buffer (20 mM Tris [pH 7.5], 100 mM NaCl, 2.5 mM CaCl2,
and 5% FCS). Blocking antibodies were used at 30 mg/ml.
Protein A-purified TIM-1-Fc fusion proteins were covalently coupled
to 6 micron blue carboxilated microparticles as recommended by the
manufacturers (Polyscience, Inc.). The poliovirus receptor (PVR) pro-
tein was included as control. 293H cells transfected with the plasmids
containing cDNAs for the indicated proteins were incubated with the
beads24–48 hrafter transfection inPBSwith2%FCS atroomtemper-
ature. After 15–30 min, unbound beads were washed and cell mono-
layers in culture media examined under an inverted microscope
(2003) for micrograph acquisition.
Supplemental Data include five figures and Experimental Procedures
and can be found with this article online at http://www.immunity.
We acknowledge the European Synchrotron Radiation Facility for pro-
vision of synchrotron radiation facilities through the BAG-Madrid and
core instrument, to B. Souto and G. Nurani for mTIM-2 crystallization,
to C. Alfonso for ultracentrifugation experiments and to S. Prat for
comments. This work was supported by grants from the Ministerio
de Educacion y Ciencia of Spain (BIO2002-03281, BFU2005-05972)
to J.M.C. and from the US National Institutes of Health (PO1
AI54456) and Food and Drug Administration to G.G.K. A grant from
the CAM-CSIC (200520M028) to J.M.C. is also acknowledged. This
paper is dedicated to the memory of J. Tormo, who started protein
crystallography at CNB.
Received: November 3, 2006
Revised: January 9, 2007
Accepted: January 29, 2007
Published online: March 15, 2007
Casasnovas, J.M., Stehle, T., Liu, J.-H., Wang, J.-H., and Springer,
T.A. (1998). A dimeric crystal structure for the N-terminal two domains
of intercellular adhesion molecule-1. Proc. Natl. Acad. Sci. USA 95,
Chakravarti, S.,Sabatos, C.A., Xiao,S.,Illes, Z.,Cha,E.K.,Sobel,R.A.,
Zheng, X.X., Strom, T.B., and Kuchroo, V.K. (2005). Tim-2 regulates T
helper type 2 responses and autoimmunity. J. Exp. Med. 202, 437–
Chen, T.T., Li, L., Chung, D.H., Allen, C.D., Torti, S.V., Torti, F.M., Cys-
ter, J.G., Chen, C.Y., Brodsky, F.M., Niemi, E.C., et al. (2005). TIM-2 is
expressed on Bcells and in liver and kidney and is a receptor for H-fer-
ritin endocytosis. J. Exp. Med. 202, 955–965.
de Souza, A.J., Oriss, T.B., O’Malley, K.J., Ray, A., and Kane, L.P.
(2005). T cell Ig and mucin 1 (TIM-1) is expressed on in vivo-activated
T cells and provides a costimulatory signal for T cell activation. Proc.
Natl. Acad. Sci. USA 102, 17113–17118.
Feigelstock, D., Thompson, P., Mattoo, P., and Kaplan, G.G. (1998a).
Polymorphisms of the hepatitis A virus cellular receptor 1 in African
green monkey kidney cells result in antigenic variants that do not react
with protective monoclonal antibody 190/4. J. Virol. 72, 6218–6222.
Feigelstock, D., Thompson, P., Mattoo, P., Zhang, Y., and Kaplan,
G.G. (1998b). The human homolog of HAVcr-1 codes for a hepatitis
A virus cellular receptor. J. Virol. 72, 6621–6628.
Han, W.K., and Bonventre, J.V. (2004). Biologic markers for the early
detection of acute kidney injury. Curr. Opin. Crit. Care 10, 476–482.
ment of distance matrices. J. Mol. Biol. 233, 123–138.
Janin, J. (1997). Specific versus non-specific contacts in protein crys-
tals. Nat. Struct. Biol. 4, 973–974.
Jimenez, D., Roda, P., Springer, T.A., and Casasnovas, J.M. (2005).
Contribution of N-linked glycans to the conformation and function of
intercellular adhesion molecules (ICAMs). J. Biol. Chem. 280, 5854–
Jones, E.Y., Davis, S.J., Williams, A.F., Harlos, K., and Stuart, D.I.
(1992). Crystal structure at 2.8 A˚resolution of a soluble form of the
cell adhesion molecule CD2. Nature 360, 232–239.
Kaplan, G., Totsuka, A., Thompson, P., Akatsuka, T., Moritsugu, Y.,
and Feinstone, S.M. (1996). Identification of a surface glycoprotein
EMBO J. 15, 4282–4296.
Kostrewa, D., Brockhaus, M., D’Arcy, A., Dale, G.E., Nelboeck, P.,
Schmid, G., Mueller, F., Bazzoni, G., Dejana, E., Bartfai, T., et al.
(2001). X-ray structure of junctional adhesion molecule: structural
basis for homophilic adhesion via a novel dimerization motif. EMBO J.
Kuchroo, V.K., Umetsu, D.T., DeKruyff, R.H., and Freeman, G.J.
(2003). The TIM gene family: emerging roles in immunity and disease.
Nat. Rev. Immunol. 3, 454–462.
Kumanogoh,A.,Marukawa,S., Suzuki, K., Takegahara, N.,Watanabe,
C., Ch’ng, E., Ishida, I., Fujimura, H., Sakoda, S., Yoshida, K., and
Kikutani, H. (2002). Class IV semaphorin Sema4A enhances T-cell ac-
tivation and interacts with Tim-2. Nature 419, 629–633.
McIntire, J.J., Umetsu, S.E., Akbari, O., Potter, M., Kuchroo, V.K.,
Barsh, G.S., Freeman, G.J., Umetsu, D.T., and DeKruyff, R.H. (2001).
Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc. 309
Structure-Function of TIM Receptors
Identification of Tapr (an airway hyperreactivity regulatory locus) and Download full-text
the linked Tim gene family. Nat. Immunol. 2, 1109–1116.
McIntire, J.J., Umetsu, S.E., Macaubas, C., Hoyte, E.G., Cinnioglu, C.,
Cavalli-Sforza, L.L., Barsh, G.S., Hallmayer, J.F., Underhill, P.A.,
Risch, N.J., et al. (2003). Hepatitis A virus link to atopic disease. Nature
Mesri, M., Smithson, G., Ghatpande, A., Chapoval, A., Shenoy, S.,
Boldog, F., Hackett, C., Pena, C.E., Burgess, C., Bendele, A., et al.
(2006). Inhibition of in vitro and in vivo T cell responses by recombinant
Meyers, J.H., Chakravarti, S., Schlesinger, D., Illes, Z., Waldner, H.,
Umetsu, S.E., Kenny, J., Zheng, X.X., Umetsu, D.T., DeKruyff, R.H.,
et al. (2005). TIM-4 is the ligand for TIM-1, and the TIM-1-TIM-4 inter-
action regulates T cell proliferation. Nat. Immunol. 6, 455–464.
Miller, J., Knorr, R., Ferrone, M., Houdei, R., Carron, C.P., and Dustin,
M.L. (1995). Intercellular adhesion molecule-1 dimerization and its
consequences for adhesion mediated by lymphocyte function associ-
ated-1. J. Exp. Med. 182, 1231–1241.
Rennert, P.D., Ichimura, T., Sizing, I.D., Bailly, V., Li, Z., Rennard, R.,
McCoon, P., Pablo, L., Miklasz, S., Tarilonte, L., and Bonventre, J.V.
(2006). T cell, Ig domain, mucin domain-2 gene-deficient mice reveal
a novel mechanism for the regulation of Th2 immune responses and
airway inflammation. J. Immunol. 177, 4311–4321.
Sabatos, C.A., Chakravarti, S., Cha, E., Schubart, A., Sanchez-Fueyo,
A., Zheng, X.X., Coyle, A.J., Strom, T.B., Freeman, G.J., and Kuchroo,
V.K. (2003). Interaction of Tim-3 and Tim-3 ligand regulates T helper
type 1 responses and induction of peripheral tolerance. Nat. Immunol.
Sanchez-Fueyo, A., Tian, J., Picarella, D., Domenig, C., Zheng, X.X.,
Sabatos, C.A., Manlongat, N., Bender, O., Kamradt, T., Kuchroo,
V.K., et al. (2003). Tim-3 inhibits T helper type 1-mediated auto- and
alloimmune responses and promotes immunological tolerance. Nat.
Immunol. 4, 1093–1101.
Silberstein,E.,Xing, L.,vandeBeek,W.,Lu,J.,Cheng,H.,and Kaplan,
G.G. (2003). Alteration of hepatitis A virus (HAV) particles by a soluble
form of HAV cellular receptor 1 containing the immunoglobin-and
mucin-like regions. J. Virol. 77, 8765–8774.
Tan, K., Zelus, B.D., Meijers, R., Liu, J.-h., Bergelson, J.M., Duke, N.,
Zhang, R., Joachimiak, A.,Holmes, K.V., and Wang, J.-h. (2002). Crys-
tal structure of murine sCEACAM1a[1,4]:a coronavirus receptor in the
CEA family. EMBO J. 21, 2076–2086.
Thompson, P., Lu, J., and Kaplan, G.G. (1998). The Cys-rich region of
hepatitis A virus cellular receptor 1 is required for binding of hepatitis A
virus and protective monoclonal antidody 190/4. J. Virol. 72, 3751–
Umetsu, S.E., Lee, W.L., McIntire, J.J., Downey, L., Sanjanwala, B.,
Akbari, O., Berry, G.J., Nagumo, H., Freeman, G.J., Umetsu, D.T.,
and DeKruyff, R.H. (2005). TIM-1 induces T cell activation and inhibits
the development of peripheral tolerance. Nat. Immunol. 6, 447–454.
van Raaij, M.J., Chouin, E., van der Zandt, H., Bergelson, J.M., and
Cusack, S. (2000). Dimeric structure of the coxsackievirus and adeno-
virus receptor D1 domain at 1.7 A˚resolution. Structure 8, 1147–1155.
Vila, M.R., Kaplan, G.G., Feigelstock, D., Nadal, M., Morote, J., Porta,
R., Bellmunt, J., and Meseguer, A. (2004). Hepatitis A virus receptor
blocks cell differentiation and is overexpressed in clear cell renal cell
carcinoma. Kidney Int. 65, 1761–1773.
noglobulin superfamily members for adhesion to integrins and viruses.
Immunol. Rev. 163, 197–215.
Wang, J.-h., Smolyar, A., Tan, K., Liu, J.-h., Kim, M., Sun, Z.-Y.,
Wagner, G., and Reinherz, E.L. (1999). Structure of a heterophilic
adhesion complex between the human CD2 and CD58 (LFA-3) Coun-
terreceptors. Cell 97, 791–803.
Wolfgang, K., and Sander, C. (1983). Dictionary of protein secondary
structure: pattern recognition of hydrogen-bonded and geometrical
features. Biopolymers 22, 2577–2637.
mTIM-1 and mTIM-2 coordinates have been deposited in the Protein
Data Bank with access codes 2OR8 and 2OR7, respectively.
310 Immunity 26, 299–310, March 2007 ª2007 Elsevier Inc.
Structure-Function of TIM Receptors