MIF is a noncognate ligand of CXC chemokine receptors in inflammatory and atherogenic cell recruitment

Article (PDF Available)inNature Medicine 13(5):587-96 · June 2007with102 Reads
DOI: 10.1038/nm1567 · Source: PubMed
Abstract
The cytokine macrophage migration inhibitory factor (MIF) plays a critical role in inflammatory diseases and atherogenesis. We identify the chemokine receptors CXCR2 and CXCR4 as functional receptors for MIF. MIF triggered G(alphai)- and integrin-dependent arrest and chemotaxis of monocytes and T cells, rapid integrin activation and calcium influx through CXCR2 or CXCR4. MIF competed with cognate ligands for CXCR4 and CXCR2 binding, and directly bound to CXCR2. CXCR2 and CD74 formed a receptor complex, and monocyte arrest elicited by MIF in inflamed or atherosclerotic arteries involved both CXCR2 and CD74. In vivo, Mif deficiency impaired monocyte adhesion to the arterial wall in atherosclerosis-prone mice, and MIF-induced leukocyte recruitment required Il8rb (which encodes Cxcr2). Blockade of Mif but not of canonical ligands of Cxcr2 or Cxcr4 in mice with advanced atherosclerosis led to plaque regression and reduced monocyte and T-cell content in plaques. By activating both CXCR2 and CXCR4, MIF displays chemokine-like functions and acts as a major regulator of inflammatory cell recruitment and atherogenesis. Targeting MIF in individuals with manifest atherosclerosis can potentially be used to treat this condition.
MIF is a noncognate ligand of CXC chemokine receptors
in inflammatory and atherogenic cell recruitment
Ju
¨
rgen Bernhagen
1,11
, Regina Krohn
2
, Hongqi Lue
1
, Julia L Gregory
3
, Alma Zernecke
2
, Rory R Koenen
2
,
Manfred Dewor
1
, Ivan Georgiev
1
, Andreas Schober
4
, Lin Leng
5
, Teake Kooistra
6
,Gu
¨
nter Fingerle-Rowson
7
,
Pietro Ghezzi
8
, Robert Kleemann
6,9
, Shaun R McColl
10
, Richard Bucala
5
, Michael J Hickey
3
&
Christian Weber
2,11
The cytokine macrophage migration inhibitory factor (MIF) plays a critical role in inflammatory diseases and atherogenesis. We
identify the chemokine receptors CXCR2 and CXCR4 as functional receptors for MIF. MIF triggered G
ai
- and integrin-dependent
arrest and chemotaxis of monocytes and T cells, rapid integrin activation and calcium influx through CXCR2 or CXCR4. MIF
competed with cognate ligands for CXCR4 and CXCR2 binding, and directly bound to CXCR2. CXCR2 and CD74 formed a receptor
complex, and monocyte arrest elicited by MIF in inflamed or atherosclerotic arteries involved both CXCR2 and CD74. In vivo,
Mif deficiency impaired monocyte adhesion to the arterial wall in atherosclerosis-prone mice, and MIF-induced leukocyte
recruitment required Il8rb (which encodes Cxcr2). Blockade of Mif but not of canonical ligands of Cxcr2 or Cxcr4 in mice with
advanced atherosclerosis led to plaque regression and reduced monocyte and T-cell content in plaques. By activating both CXCR2
and CXCR4, MIF displays chemokine-like functions and acts as a major regulator of inflammatory cell recruitment and
atherogenesis. Targeting MIF in individuals with manifest atherosclerosis can potentially be used to treat this condition.
Small chemotactic cytokines termed chemokines orchestrate the
activation and recruitment of leukocytes during immune surveillance
and inflammation. The structural classification of chemokines, their
G protein–coupled receptors (GPCRs), expression patterns and bio-
logical functions in leukocyte traffic and inflammatory disease have
been well defined
1–4
. Genetic deletion and antibody inhibition studies
have implicated chemokines and their receptors in the pathogenesis of
atherosclerosis
2,5
. An impressive body of evidence supports the con-
cept that atherosclerosis is a chronic inflammatory disease of the
arterial wall, characterized by an influx of immunocompetent mono-
nuclear cells
6–8
. The atherogenic recruitment of monocytes and T cells
is governed by specialized functions of chemokines
6
.Beyondtheirrole
in chemotaxis, chemokines presented on the endothelial surface are
instrumental in triggering integrin-mediated arrest of rolling leuko-
cytes
9
. In the context of atherosclerosis, this is best established for the
platelet-derived CC chemokine CCL5, and the CXC chemokines and
CXCR2 ligands CXCL1 and CXCL8 (ref. 5).
The group of ‘chemokine-like function (CLF) or ‘micro chemo-
kines includes chemotactic polypeptides such as the b-defensins,
which cannot be classified into known chemokine subfamilies but
share structural or functional features and can signal through
chemokine receptors, for example, CCR6 (ref. 10,11). The evolution-
arily ancient cytokine MIF in its monomeric form exhibits
remarkable homology to dimers of CXCL8 (ref. 12). MIF plays an
important role in acute and chronic inflammatory diseases such as
septic shock, rheumatoid arthritis and colitis
5,13–16
. Unique among
cytokines, MIF serves as an endogenous counter-regulator of gluco-
corticoids in inflammation
14
and intracellularly interacts with c-Jun
activation domain binding protein-1 (JAB1) (ref. 17). CD74 (also
known as invariant chain) binds MIF on the surface of cells, can
mediate activation of extracellular-regulated mitogen-activated pro-
tein (ERK-MAP) kinases and associates into a signaling complex with
MIF, CD44 and Src-tyrosine kinases
18,19
. However, not all cells
targeted by MIF express surface CD74 (neutrophils are one example;
ref. 16). Thus, despite recent progress in understanding MIF-mediated
signaling pathways, the molecular modes of MIF action and the
functional receptor(s) underlying its role in inflammatory diseases
remain unclear.
Historically, MIF was discovered as an inhibitor of random
macrophage migration
20
, but whether this observation can be linked
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
Received 3 December 2006; accepted 5 March 2007; published online 15 April 2007; doi:10.1038/nm1567
1
Department of Biochemistry and Molecular Cell Biology, Institute of Biochemistry, and
2
Institute of Molecular Cardiovascular Research, University Hospital Aachen,
Rheinisch-Westfa
¨
lische Technische Hochschule (RWTH) Aachen University, D-52074 Aachen, Germany.
3
Centre for Inflammatory Diseases, Monash University,
Victoria 3168, Australia.
4
Medical Policlinic, Ludwig-Maximilians-University, D-80336 Munich, Germany.
5
Department of Medicine, Pathology, Epidemiology and
Public Health, The Anlyan Center, Yale University School of Medicine, New Haven, Connecticut 06520, USA.
6
TNO Quality of Life, Gaubius Laboratory/Biosciences,
Department of Vascular and Metabolic Diseases, 2301 CE Leiden, The Netherlands.
7
Medical Clinic I, Department of Hematology and Oncology, University Hospital
Cologne, D-50937 Cologne, Germany.
8
Mario Negri-Institute for Pharmacological Research, 20157 Milan, Italy.
9
Department of Vascular Surgery, Leiden University
Medical Center, 2333 CK Leiden, The Netherlands.
10
School of Molecular and Biomedical Science, University of Adelaide, Adelaide SA 5005, Australia.
11
These
authors contributed equally to this work. Correspondence should be addressed to J.B. (jbernhagen@ukaachen.de) or C.W. (cweber@ukaachen.de).
NAT URE MED ICINE ADVANCE ONLINE PUBLICATION 1
ARTICLES
to disease states involving MIF is largely unknown. Recently, MIF-
mediated recruitment of mononuclear cells has been implicated in
glomerulonephritis and arthritis
21,22
; however, its promigratory func-
tions have been ascribed to its pleiotropic properties in cell activa-
tion
23
. Thus, the mechanisms underlying MIF-regulated cell migration
and the receptors involved have been elusive for over four decades.
MIF has emerged as a key element in vascular processes giving rise
to atherosclerosis
5,14
. Its expression is upregulated in endothelial cells,
smooth muscle cells (SMCs) and macrophages during the develop-
ment of atherosclerotic lesions in humans, rabbits and mice
24–26
.This
upregulation is induced by proatherogenic stimuli, such as oxidized
low-density-lipoprotein (oxLDL), is associated with plaque instability
and occurs after arterial injury
26,27
. The retardation of diet-induced
atherogenesis in Ldl receptor–deficient (Ldlr
–/–
)micebygenetic
deletion of Mif is manifested as a decrease in intimal thickening,
lipid deposition and protease expression
28
. Blocking Mif diminishes
macrophage infiltration in the intima of atherosclerosis-prone apo-
lipoprotein E–deficient (Apoe
–/–
) mice and in the neointima after
endothelial denudation, whereas SMC and collagen content are
increased
26
, reflecting a more stable plaque composition. These data
imply a pathway for unstable lesion formation involving MIF and
underscore MIF’s importance in arterial macrophage accumulation.
MIF may directly affect endothelial-monocyte adhesion by a mode
resembling that of immobilized chemokines, as MIF on aortic
endothelial cells exposed to oxLDL triggers monocyte arrest under
flow conditions
26
.
Here we have unraveled the molecular machinery that underlies
the regulation of leukocyte migration by MIF. We report that MIF
is a functional noncognate ligand for the chemokine receptors CXCR2
and CXCR4, and thereby controls inflammatory and atherogenic
leukocyte recruitment.
RESULTS
Surface-bound MIF induces monocyte arrest through CXCR2
Given the structural resemblance of MIF monomers
12
to CXCL8
dimers (Supplementary Fig. 1 online) and the fact that CXCL1 and
CXCL8 and their receptor CXCR2 have been identified as major
ligand-receptor pairs promoting monocyte arrest
5,29,30
,weused
monoclonal antibodies and pertussis toxin (PTX) to explore whether
MIF-induced monocyte arrest depends on G
ai
-coupled activities of
CXCR2. Human aortic endothelial cells that had been pretreated with
MIF for 2 h substantially increased the arrest of primary human
monocytes under flow conditions, an effect blocked by an antibody to
MIF (Fig. 1a). Notably, MIF-triggered, but not spontaneous, mono-
cyte arrest was ablated by an antibody to CXCR2 or by PTX,
implicating G
ai
-coupled CXCR2. We confirmed the ability of MIF
to induce monocyte arrest through CXCR2 using monocytic Mono-
Mac6 cells (Supplementary Fig. 2 online) and found that this activity
was associated with an immobilization of MIF on aortic endothelial
cells (Fig. 1b), indicating that MIF presented on the endothelial cell
surface exerts a chemokine-like arrest function as a noncognate
CXCR2 ligand. Blocking classical CXCR2 agonists (CXCL1/CXCL8)
did not interfere with these effects of MIF (Fig. 1a).
To dissect the mechanisms used by MIF to promote integrin-
dependent arrest, we used Chinese hamster ovary (CHO) transfectants
expressing the b
2
integrin ligand intercellular adhesion molecule
(ICAM)-1, which support the arrest of monocytic cells under flow
conditions. The exposure of CHO transfectants to MIF for 2 h resulted
in its surface presentation (Fig. 1b)and,likeexposureofthe
transfectants to CXCL8, increased monocytic cell arrest (Fig. 1c).
This effect was fully sensitive to PTX and an antibody to b
2
integrin
(Fig. 1c), confirming a role of G
ai
in b
2
integrin–mediated arrest
induced by MIF. Primary monocytes and MonoMac6 cells express
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
ab
Monocyte arrest
(cells/mm
2
)
MIF
MIFControl
CHO/
ICAM-1
HAoEC
Anti-CXCR2
Anti-MIF
PTX
Anti-CXCL1/8
0
10
20
30
P < 0.05 P < 0.05
+
+
+
+
+
+
––
–+
+
+
+
e
T-cell arrest
(cells/mm
2
)
MIF
CXCL12
Anti-CXCR4
Control lgG
PTX
0
20
40
P < 0.01
+
+
+
+
––
––
+
+
+
+
+
P < 0.05
P < 0.05
f
Jurkat T-cell arrest
(cells/mm
2
)
MIF
CXCL12
Anti-CXCR4
Control lgG
0
20
60
+
+
+
+
+
––
––
+
+
+
+
40
h
L1.2 cell arrest
(cells/mm
2
)
Cxcl12
MIF
AMD3465
0
50
200
+
––
+–
+
+
150
100
P < 0.05
P < 0.05
Percent control
CXCL8
MIF
CXCL10
AMD3465
0
100
+
+
––
––
+
+
200
300
+
+
++++++ +++
CXCR3
+
CXCR2
+
CXCR1
+
g
Jurkat T-cell arrest
(% vector control)
MIF
+ pcDNA3 + pcDNA3
-CXCR2
Control MIFControl
0
200
100
P < 0.05
P < 0.05
c
MonoMac6 cell arrest
(cells/mm
2
)
CXCL8
MIF
Control lgG
PTX
Anti-β
2
0
50
100
P < 0.001 P < 0.05
+
+
+
+–
+
+
+
+
P < 0.05
d
MonoMac6 cell arrest
(cells/mm
2
)
MIF
Anti-CXCR1
Anti-CXCR2
Anti-CD74
Control lgG
0
50
75
P < 0.05
+
+
+
+
+
+
++
+
+
+
+
25
P < 0.01
Figure 1 MIF-triggered mononuclear cell arrest is mediated by CXCR2, CXCR4 and CD74. Human aortic endothelial cells (HAoECs), CHO cells stably
expressing ICAM-1 (CHO/ICAM-1) and mouse microvascular endothelial cells (SVECs) were preincubated with or without MIF (together with antibody to MIF,
antibodies to CXCL1 and CXCL8, or isotype control), CXCL8, CXCL10 or CXCL12 for 2 h as indicated. Mononuclear cells were pretreated with antibodies to
CXCR1, CXCR2, b
2
integrin, CXCR4, CD74 or isotype controls for 30 min, or pertussis toxin (PTX) for 2 h as indicated. (a) HAoECs were perfused with
primary human monocytes. (b) Immunofluorescence using antibody to MIF revealed surface presentation of MIF (green) on HAoECs and CHO/ICAM-1 cells
after pretreatment for 2 h, but not 30 min (not shown); in contrast, MIF was absent in buffer-treated cells (control). Scale bars, 100 mm. (c,d) CHO/ICAM-1
cells were perfused with MonoMac6 cells. (e) HAoECs were perfused with T cells. (f,g) CHO/ICAM-1 cells were perfused with Jurkat T cells (f), and with
Jurkat CXCR2 transfectants or vector controls (g). In c,d,f and g, background binding to vector-transfected CHO cells was subtracted. (h) Mouse SVECs were
perfused with L1.2 transfectants stably expressing CXCR1, CXCR2 or CXCR3, and with controls expressing only endogenous CXCR4, in the presence of the
CXCR4 antagonist AMD3465. Arrest is quantified as cells/mm
2
or as percentage of control cell adhesion. Data in a and cg represent mean ± s.d. of 3–8
independent experiments; data in h are results from one representative experiment (n ¼ 3 measurements; mean ± s.d.) of four experiments.
ARTICLES
2 ADVANCE ONLINE PUBLICATION NATU RE M EDI CINE
both CXCR1 and CXCR2 (ref. 30). Whereas blocking CXCR1 had no
effect, blocking CXCR2 substantially but not fully impaired MIF-
triggered and CXCL8-triggered monocytic cell arrest; addition of
antibodies to both CXCR1 and CXCR2 completely inhibited the arrest
functions of MIF or CXCL8 (Fig. 1d and Supplementary Fig. 2). The
use of antibodies to CD74 implicated this protein, along with CXCR2,
in MIF-induced arrest (Fig. 1d). Spontaneous arrest was unaffected
(Supplementary Fig. 2). Thus, CXCR2 assisted by CD74 mediates
MIF-induced arrest.
The arrest function of MIF extends to CXCR4 in T cells
Because T-cell arrest can be triggered by the CXCR4 ligand CXCL12
through the Src kinase p56 (ref. 9) and MIF can activate T cells
through Src-family kinases
16,19
, we evaluated whether MIF can target
CXCR4 in T cells. Either MIF or CXCL12 immobilized on aortic
endothelial cells triggered the arrest of primary human effector T cells
(Fig. 1e). MIF-induced, but not spontaneous, T-cell arrest was
sensitive to PTX and was inhibited by an antibody to CXCR4
(Fig. 1e). Although less pronounced than in monocytes expressing
CXCR2 (Fig. 1d), presentation of MIF (or CXCL12) on CHO
transfectants expressing ICAM-1 elicited a
L
b
2
-dependent arrest of
Jurkat T cells, an effect mediated by CXCR4 (Fig. 1f).
Ectopic expression of CXCR2 in Jurkat T cells increased MIF-
triggered arrest (Fig. 1g), corroborating the idea that CXCR2 imparts
responsiveness to MIF in leukocytes. L1.2 pre-B lymphoma trans-
fectants expressing CXCR1, CXCR2 or CXCR3, and controls using
cells expressing endogenous Cxcr4 only were used in the presence of
the CXCR4 antagonist AMD3465. MIF triggered the arrest of CXCR2
transfectants and Cxcr4-bearing controls on endothelial cells with a
similar efficacy to that of the canonical ligands CXCL8 and CXCL12,
whereas CXCR1 and CXCR3 transfectants were responsive to CXCL8
and CXCL10, respectively, but not to MIF (Fig. 1h). Our data establish
that CXCR2 and CXCR4, but not CXCR1 or CXCR3, support
MIF-induced arrest.
MIF stimulates leukocyte chemotaxis through CXCR2/4
Chemokines have been eponymously defined as inducers of chemo-
taxis
1,5
. Paradoxically, MIF was initially thought to interfere with
‘random migration
16
. Although this may be attributable to active
repulsion or desensitization of directed emigration, specific mechan-
isms evoked by MIF to regulate migration remain to be clarified. As
cell activation by MIF may rather stimulate migration
23
, our results
showing that MIF promotes G
ai
-mediated functions of CXCR2 and
CXCR4 prompted us to test whether MIF can directly elicit leukocyte
chemotaxis through these receptors.
Using a Transwell system, we compared the promigratory effects of
MIF and CXCL8 on primary human peripheral blood mononuclear
cell–derived monocytes; CCL2 was also used as a prototypic chemo-
kine for monocytes. Similar to CXCL8 and CCL2, adding MIF to the
lower chamber induced migration, which followed a bell-shaped
dose-response curve typical for chemokines, with an optimum at
25–50 ng/ml, albeit with a lower peak migratory index (Fig. 2a). To
demonstrate specificity, we used heat treatment or a neutralizing
antibody to MIF, which abolished MIF-induced transmigration; in
contrast, isotype-matched immunoglobulin (IgG) had no effect
(Fig. 2b). When added to the upper chamber, MIF dose-dependently
desensitized migration toward MIF in the lower chamber (Fig. 2c)
but did not elicit migration when present in the upper chamber only
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
abcd
efgh
Neutrophil transmigration
(chemotactic index)
Log (MIF) ng/ml
1.0
1.5
0
0.001
0.01
0.1
1
10
100
1,000
0
1
2
3
4
5
Neutrophil transmigration
(chemotactic index)
P < 0.01
MIF
Anti-CXCR2
Anti-CXCR1
CXCL8
P < 0.0002
0
1
P < 0.05
+
++
+
+
+
+
+
+
+
+
T-cell transmigration
(chemotactic index)
P < 0.05
MIF
Anti-CXCR4
CXCL12
Anti-MIF
P < 0.02
0
1
2
3
P < 0.02
+
++
+
+
+
+
+
+
Monocyte transmigration
(% input)
P < 0.05
0
5
10
P < 0.002
P < 0.02
MIF
Anti-CD74
Control IgG
Anti-CXCR1/2
++
+
+
+
+
+
+
+
Monocyte transmigration
(% input)
P < 0.05
P < 0.005
0
5
10
15
20
MIF
PTX A + B
PTX B
Ly294002
P < 0.0005
++
+
+
+
+
+
Percent of maximal MIF-
induced transmigration
–110
0
50
100
Log [MIF] (ng/ml) in top chamber
Monocyte transmigration
(% input)
P < 0.0005
MIF
Anti-MIF
Control IgG
Boiled
Anti-MIF
0
5
10
15
20
P < 0.0005
––
––
––
––
––
––
––
++
+
+
+
+
+
Monocyte transmigration
(chemotactic index)
Log [cytokine] (ng/ml)
MIF
CXCL8
CCL2
1
2
3
4
0
0.001
0.01
0.1
1
10
100
1,000
2
100
Figure 2 MIF-triggered mononuclear cell chemotaxis is mediated by CXCR2, CXCR4 and CD74. Primary human monocytes (ae), CD3
+
Tcells(f)and
neutrophils (g,h) were subjected to transmigration analysis in the presence or absence of MIF. CCL2 (a), CXCL8 (a,g,h) and CXCL12 (f) served as positive
controls or were used to test desensitization by MIF (or by CXCL8, h). The chemotactic effects of MIF, CCL2 and CXCL8 on monocytes (a)orofMIFon
neutrophils (g) followed bell-shaped dose-response curves. MIF-triggered chemotaxis of monocytes was abrogated by an antibody to MIF, boiling (b), or
by MIF at indicated concentrations (in the top chamber; c). (d) MIF-triggered chemotaxis was mediated by G
ai
/phosphoinositide-3-kinase signaling, as
evidenced by treatment with pertussis toxin components A and B (PTX A + B), PTX component B alone or Ly294002. (e) MIF-mediated monocyte
chemotaxis was blocked by antibodies to CD74 or antibodies to CXCR1 and CXCR2. (f) T-cell chemotaxis induced by MIF was blocked by antibodies to MIF
or CXCR4. (g) Neutrophil chemotaxis induced by MIF. (h) MIF-induced versus CXCL8-induced neutrophil chemotaxis, effects of antibodies to CXCR2 or
CXCR1, and desensitization of CXCL8 by MIF are shown. Data in a and fh are expressed as chemotactic index; data in c are expressed as percent of
control; and data in b,d and e as percent of input. Data represent mean ± s.d. of 4–10 independent experiments, except for panels a,c and g,boiledMIF
in b, and the anti-MIF, anti-CD74, anti-CXCR1 and anti-CXCR2 antibody controls in b and e, which are means of 2 independent experiments.
ARTICLES
NAT URE MED ICINE ADVANCE ONLINE PUBLICATION 3
(data not shown), suggesting that MIF evokes true chemotaxis rather
than chemokinesis. Consistent with G
ai
-dependent signaling through
phosphoinositide-3-kinase, MIF-induced monocyte chemotaxis was
sensitive to PTX and abrogated by Ly294002 (Fig. 2d). Both CXCR2
and CD74 specifically contributed to MIF-triggered monocyte che-
motaxis (Fig. 2e). We confirmed a role for CXCR2 by showing MIF-
mediated cross-desensitization of CXCL8-induced chemotaxis in
CXCR2-transfected L1.2 cells (data not shown). We verified the
chemotactic activity of MIF in RAW264.7 macrophages (S upplemen-
tary Fig. 2) and THP-1 monocytes (data not shown). These data
demonstrate that MIF triggers monocyte chemotaxis through CXCR2.
To substantiate functional MIF-CXCR4 interactions, we evaluated
the transmigration of primary CD3
+
T lymphocytes devoid of CXCR1
and CXCR2. Similar to CXCL12, a known CXCR4 ligand and T-cell
chemoattractant, MIF dose-dependently induced transmigration, a
process that was chemotactic and transduced through CXCR4, as
shown by antibody blockade and cross-desensitization of CXCL12
(Fig. 2f and Supplementary Fig. 2). Thus, MIF elicits directed T-cell
migration through CXCR4. In primary human neutrophils, a major
cell type bearing CXCR2, MIF exerted CXCR2- but not CXCR1-
mediated chemotactic activity, exhibiting a bell-shaped dose-response
curve and cross-densensitizing CXCL8 (Fig. 2g,h). The moderate
chemotactic activity of neutrophils towards MIF is likely to be related
to an absence of CD74 on neutrophils, as its ectopic expression in
CD74
promyelocytic HL-60 cells enhanced MIF-induced migration
(Supplementary Fig. 2). Although MIF, like other CXCR2 ligands,
may preferentially function as an arrest chemokine
29,30
,thesedata
reveal that MIF also has appreciable chemotactic properties on mono-
nuclear cells and neutrophils.
MIF triggers rapid integrin activation and calcium flux
Arrest functions of MIF may reflect direct MIF/CXCR signaling, but it
cannot be entirely excluded that MIF may induce other arrest
chemokines during the time required for MIF immobilization. To
consolidate evidence that MIF directly induces leukocyte arrest
(Fig. 1), we performed real-time PCR and ELISA and found that
2-h-long preincubation of human aortic (or venous) endothelial cells
with MIF did not upregulate typical arrest chemokines known to
engage CXCR2 (Fig. 3a and data not shown).
Short-term exposure to chemokines present in solution or immo-
bilized in juxtaposition to integrin ligands (for example, vascular cell
adhesion molecule (VCAM)-1) can rapidly upregulate integrin activ-
ity, which mediates leukocyte arrest
9
.Thisisaccomplishedbycluster-
ing (for example, a
4
b
1
) or conformational changes (for example,
a
L
b
2
) immediately preceding ligand binding. Stimulation of mono-
cytic cells with MIF (or CXCL8) for 1–5 min triggered a
L
b
2
-
dependent arrest on CHO/ICAM-1 cells (Fig. 3b). We obtained
evidence for a direct stimulation of monocyte integrins in assays
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
0
10
0
10
1
10
2
10
3
10
4
Events
0.3
0.1
MonoMac6 cell arrest
(% control)
250
200
150
100
50
0
MIF
Isotype control
Anti-CXCR2
Anti-CD74
Anti-α
4
+
+
+
+
+
+
+
+
+
+
+
300
350
P < 0.01
CXCL8
(ng/ml)
CXCL7
(ng/ml)
MIF (ng/ml)
1 10 100 1,000 1 10 100 1,000 1 10 50 100 250 1,000
0
20
MFI
40
60
80
P < 0.01
P < 0.05
100
50
0
Control MIF CXCL8
0.7
0.5
0.9
0102030
Time (min)
0 min
2 min
10 min
30 min
MFI
0 60 120 180 240
Time (s)
80
60
40
MFI
CXCL8 CXCL8
80
60
40
MFI
0 60 120 180 240
Time (s)
0 60 120 180 240
Time (s)
0 60 120 180 240
Time (s)
0 60 120 180 240
Time (s)
0 60 120 180 240
Time (s)
MIF MIF
80
60
40
MFI
CXCL8 MIF
80
60
40
MFI
MIF CXCL8
80
60
40
MFI
CXCL7 MIF
80
60
40
MFI
MIF CXCL7
Control MIF
Cxcl1
Cxcl8
100
50
0
70
50
30
10
0
Percent maximal activation
15 2010525
Time (min)
25
mRNA expression
(normalized to control)
TNF-α
Control
MIF
TNF-α
CXCL8 (pg/ml)
2,000
1,000
0
MonoMac6 cell
arrest (cells/mm
2
)
Fluorescence intensity
abcd
e
g
f
Figure 3 MIF triggers rapid integrin activation and
calcium signaling. (a) Human aortic endothelial cells
were stimulated with MIF or TNF-a for 2 h. Cxcl1
and Cxcl8 mRNAs were analyzed by real-time PCR
and normalized to control. Supernatant-derived
CXCL8 was assessed by ELISA (mean ± s.d. of three
independent experiments performed in duplicate).
(b) MonoMac6 cells were directly stimulated with
MIF or CXCL8 for 1 min and perfused on
CHO-ICAM-1 cells for 5 min (mean ± s.d. of
8 independent experiments). (c) MonoMac6 cells
were stimulated with MIF for the indicated times.
LFA-1 activation (detected by the 327C antibody)
was monitored by FACSAria, and expressed as the
increase in mean fluorescence intensity (MFI).
(d)Asinc but for primary monocytes; data are
expressed relative to maximal activation with Mg
2+
and EGTA. (e) MonoMac6 cells were pretreated with
antibodies to a
4
integrin, CD74 or CXCR2, stimulated
with MIF for 1 min, and perfused on VCAM-1.Fc
protein for 5 min. Adhesion is expressed as a
percentage of controls. Arrest data in ce represent
mean ± s.d. of 5 independent experiments.
(f) Calcium transients in Fluo-4 AM–labeled
neutrophils were elicited with MIF, CXCL8 or CXCL7.
Calcium-derived MFI was recorded by FACSAria for 0–240 s. For desensitization, stimuli were added 120 s before stimulation. Traces shown are
representative of 4 independent experiments. (g) Dose-response curves of calcium-influx triggered by CXCL8, CXCL7 or MIF, at indicated concentrations, in
L1.2-CXCR2 transfectants. Data are expressed as the difference between baseline and peak MFI (mean ± s.d. of 4–8 independent experiments).
ARTICLES
4 ADVANCE ONLINE PUBLICATION NATU RE M EDI CINE
using the reporter antibody 327C, which recognizes an extended high-
affinity conformation of a
L
b
2
(ref. 31). These assays revealed that a
L
b
2
activation in MonoMac6 cells (Fig. 3c) and human blood monocytes
(Fig. 3d) occurred as early as 1 min after exposure to MIF and
persisted over 30 min. To evaluate whether MIFs effects were restricted
to a
L
b
2
,westudieda
4
b
1
-dependent monocytic cell arrest on VCAM-1.
Exposure to MIF for 1–5 min induced marked arrest, which was
mediated by CXCR2, CD74 and a
4
b
1
(Fig. 3e). Similarly to the effect of
CXCL12, stimulation of Jurkat T cells with MIF for 1–5 min triggered
CXCR4-dependent adhesion on VCAM-1 (Supplementary Fig. 2).
As CXCR2 can mediate increases in cytosolic calcium elicited by
CXCL8 (ref. 32), we tested whether MIF stimulates calcium influx and
densensitizes CXCL8 signals. Indeed, like CXCL8, MIF induced
calcium influx in primary human neutrophils and desensitized cal-
cium transients in response to either CXCL8 or MIF (Fig. 3f),
confirming that MIF activates GPCR/G
ai
signaling. The partial desen-
sitization of CXCL8 signaling by MIF seen in neutrophils parallels
findings with other CXCR2 ligands
32
and reflects the presence
of CXCR1: in L1.2 transfectants expressing CXCR2, MIF fully
desensitized CXCL8-induced calcium influx (data not shown),
and in neutrophils, MIF desensitized transients induced by the
selective CXCR2 ligand CXCL7 (and CXCL7 desensitized transients
induced by MIF) (Fig. 3f). In CXCR2 transfectants, MIF dose-
dependently induced calcium influx, and was slightly less potent
and effective than CXCL8 or CXCL7 (Fig. 3g). In conclusion, MIF
acts on CXCR2 and CXCR4 to elicit rapid integrin activation and
calcium influx.
MIF interacts with CXCR2 and CXCR4
To assess the physical interactions of MIF with CXCR2 and CXCR4,
we performed receptor-binding competition and internalization
studies. In HEK293 cells ectopically expressing CXCR2, MIF
strongly competed with
125
I-labeled CXCL8 for CXCR2 binding
under equilibrium conditions. Binding of the CXCL8 tracer to
CXCR2 was inhibited by MIF with an effector concentration for
half-maximum response (EC
50
) of 1.5 nM (Fig. 4a). The affinity
of CXCR2 for MIF (K
d
¼ 1.4 nM) was close to that for CXCL8
(K
d
¼ 0.7 nM) and within the range of the MIF concentration
that induced optimal chemotaxis (2–4 nM). To confirm binding
to CXCR2, we used a receptor internalization assay that reports
specific receptor-ligand interactions. FACS analysis of sur-
face CXCR2 on stable HEK293 transfectants showed that MIF
induced CXCR2 internalization with a dose response resembling
that of CXCL8 (Fig. 4b). We obtained comparable data in CXCR2-
transfected RAW264.7 macrophages (inset in Fig. 4b, and data
not shown).
To verify an interaction of MIF with CXCR4, we performed
receptor-binding studies in Jurkat T cells, which endogenously express
CXCR4. MIF competed with
125
I-labeled CXCL12 for CXCR4 binding
(K
d
for CXCL12 ¼ 1.5 nM; EC
50
¼ 19.9 nM, K
d
for MIF ¼ 19.8 nM)
(Fig. 4c). The K
d
was in accordance with MIF concentrations that
induce T-cell chemotaxis. Consistently, MIF, like CXCL12, elicited
CXCR4 internalization in a dose-dependent fashion (Fig. 4d).
MIF-induced internalization of CXCR2 and CXCR4 was specific to
these receptors, as MIF, unlike the cognate ligand CCL5, was unable to
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
WT
CXCR2
CD74
CXCR2
IgG
IgG
-CXCR2
Anti-CXCR2 IP
Anti-CXCR2 WB
Anti-CD74 WB
IP
Anti-CD74 WB
h
IgG Anti
Input Beads
Anti-
CXCR2
His-CD74
CXCR2
CXCR2 IP
Input Beads
Anti-
His
CXCR2
His-CD74
His-CD74 IP
Anti-CXCR2 WB
Anti-CD74 WB
Anti-CXCR2 WB
Anti-CD74 WB
gf
CXCR2
HEK-CXCR2
10
0
10
1
10
2
10
3
10
4
Events
FL1
HEK
HEK
e
bd
Log [ligand] (M)
Buffer
MIF
CXCR4 internalization
(% of control)
Con –11 –10 –9 –8 –7 –6
20
40
60
80
100
MIF
CXCL12
CXCR2 internalization
(% of control)
Buffer
Con –11 –10 –9 –8 –7 –6
40
60
80
100
MIF
CXCL8
MIF
10
0
10
1
10
2
10
3
FL1
Events
10
0
10
1
10
2
10
3
Events
Log [ligand] (M)
ac
Bound [I
125
]
CXCL8 (c.p.m.)
Log [competitor] (M)
–14 –13 –12 –11 –10 –9 –8 –7 –6 –5
0
20,000
40,000
60,000
80,000
MIF
CXCL8
Bound [I
125
]
CXCL12 (c.p.m.)
Log [competitor] (M)
–14 –13 –12 –11 –10 –9 –8 –7 –6 –5
0
25,000
50,000
75,000
MIF
CXCL12
Anti-CXCR2 WB
HEK-CXCR2
CXCR2
CXCR2
CD74
CD74
Hoechst Overlay
Overlay
Figure 4 MIF interaction with CXCR2 or CXCR4 and formation of a CXCR2-CD74 complex. (a) HEK293-CXCR2 transfectants were subjected to receptor
binding assays, in order to analyze competition of [I
125
]CXCL8 with MIF or cold CXCL8 (mean ± s.d., n ¼ 6–10). (b) MIF- and CXCL8-induced CXCR2
internalization in HEK293-CXCR2 or RAW264.7-CXCR2 transfectants (inset shows representative histograms) as indicated. Assessment was by FACS
analysis of surface CXCR2 expression (percentage of buffer (Con), mean ± s.d., n ¼ 5). (c)Asina, but for CXCR4-bearing Jurkat T-cells, to analyze
competition of [I
125
]CXCL12 with MIF or cold CXCL12. (d) MIF- and CXCL12-induced CXCR4 internalization in Jurkat T-cells as in b (mean ± s.d.,
n ¼ 4–6). (e) Binding of fluorescein-MIF to HEK293-CXCR2 transfectants or vector controls analyzed by FACS. Inset shows binding of biotin-MIF to CXCR2
assessed by western blot using antibodies to CXCR2 after streptavidin pull-down from HEK293-CXCR2 transfectants versus vector controls. (f) Colocalization
of CXCR2 and CD74 (orange-yellow overlay) in RAW264.7-CXCR2 transfectants stained for CXCR2, CD74 and nuclei (Hoechst), analyzed by fluorescence
microscopy (top) or confocal laser scanning microscopy (bottom). Scale bar, 10 mm. (g) Coimmunoprecipitation of CXCR2/CD74 complexes in CHAPSO-
extracts of HEK293-CXCR2 transfectants expressing His-tagged CD74. Anti-His immunoprecipitation (IP) followed by anti-CXCR2 or anti–His-CD74 western
blotting (WB; top) or anti-CXCR2 immunoprecipitation followed by anti–His-CD74 or anti-CXCR2 western blotting (bottom). Controls: lysates without
immunoprecipitation or beads alone. (h)Asing for L1.2-CXCR2 transfectants. Anti-CXCR2 immunoprecipitation from L1.2-CXCR2 transfectants followed
by anti-CD74 or anti-CXCR2 western blotting (top). Immunoprecipitation from CXCR2-negative L1.2-cells (WT; bottom) served as a control. Data are
representative of 3 independent experiments (eh).
ARTICLES
NAT URE MED ICINE ADVANCE ONLINE PUBLICATION 5
induce CCR5 internalization in L1.2 CCR5 transfectants (data
not shown).
To corroborate its interactions with CXCRs, we labeled MIF with
biotin or fluorescein, which, in contrast to iodinated MIF, allows for
direct receptor-binding assays. We found that CXCR2 transfectants,
but not vector controls, supported direct binding of labeled MIF, as
evidenced by flow cytometry (Fig. 4e), pull down with streptavidin
beads (inset in Fig. 4e) and fluorescence microscopy (data not
shown). In addition, the specific binding of fluorescein-MIF to
CXCR4-bearing Jurkat cells was inhibited by the CXCR4 antagonist
AMD3465 (data not shown).
Complex formation between CXCR2 and CD74
CD74 has been implicated as an MIF-binding protein; therefore, we
addressed the possibility that a functional MIF receptor complex
involves both GPCRs and CD74. We studied the colocalization of
endogenous CD74 and CXCR2 by confocal fluorescence microscopy
in RAW264.7 macrophages expressing human CXCR2. We detected
prominent colocalization, in a polarized pattern, in B50% of cells
(Fig. 4f).
Notably, coimmunoprecipitation assays revealed that CXCR2 phy-
sically interacts with CD74. We detected CXCR2/CD74 complexes in
HEK293 cells stably overexpressing CXCR2 and transiently expressing
His-tagged CD74. We observed these complexes by precipitation with
an antibody to CXCR2 and by detecting coprecipitated CD74 by
western blot against the His-tag; coprecipitation was also seen when
the order of the antibodies used was reversed (Fig. 4g). We also
detected complexes with CD74 in L1.2 transfectants stably expressing
human CXCR2, as assessed by coimmunoprecipitation with an anti-
body to CXCR2; in contrast, we observed no complexes with L1.2
controls or the isotype control (Fig. 4h). Our data are in accord with a
model in which CD74 forms a signaling complex with CXCRs to
mediate MIF functions.
CXCR2 mediates MIF-induced monocyte arrest in arteries
MIF promotes the formation of complex plaques with abundant cell
proliferation, macrophage infiltration and lipid deposition
5,14
.This
has been related to the induction of endothelial MIF by oxLDL,
triggering monocyte arrest
26
. The CXCR2 ligand CXCL1 can also elicit
a
4
b
1
-dependent monocyte accumulation in ex vivo–perfused carotid
arteries of mice with early atherosclerotic endothelium
29
. Therefore,
we tested whether MIF acts via CXCR2 to induce recruitment in this
system. Monocyte arrest in carotid arteries of Apoe
–/–
mice fed a high-
fat diet was inhibited by antibodies to CXCR2, CD74 or Mif (Fig. 5a,
and Supplementary Fig. 3 and Supplementary Video 1 online),
indicating that MIF contributed to atherogenic recruitment via
CXCR2 and CD74. Following the blockade of Mif, CXCR2 and
CD74 for 24 h, we observed a similar pattern for monocyte arrest
in arteries of wild-type mice treated with tumor necrosis factor
(TNF)-a, mimicking acute vascular inflammation (Fig. 5b). In
arteries of TNF-a–treated Mif
–/–
mice, inhibitory effects on CD74
were attenuated and blocking Mif was ineffective, whereas there was
residual CXCR2 inhibition, implying the involvement of other indu-
cible ligands (Fig. 5c). Compared to the effect of Mif deficiency
observed with TNF-a stimulation, monocyte accumulation was more
clearly impaired by Mif deficiency in arteries of Mif
–/–
Ldlr
–/–
mice
(compared to atherogenic Mif
+/+
Ldlr
–/–
mice; Fig. 5d,e). In the
absence of Mif, there was no apparent contribution of CXCR2;
moreover, blocking Mif had no effect (Fig. 5d,e). The inhibitory
effects of blocking CXCR2 were restored by loading exogenous
MIF (Fig. 5f).
To provide further evidence for the idea that CXCR2 is required for
MIF-mediated monocyte recruitment in vivo, we performed intravital
microscopy on carotid arteries of chimeric wild-type Mif
+/+
and
Mif
–/–
mice reconstituted with wild-type or Il8rb
–/–
bone marrow
(Il8rb encodes Cxcr2; Fig. 5g,h and Supplementary Video 2 online).
After treatment with TNF-a for 4 h, we found that the accumulation
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
abc de
fgh
0
5
10
15
P < 0.05
+
+
+
+––
0
5
10
15
+
+
+
+––
20
30
0
10
P < 0.01
P < 0.01
P < 0.05
+
+
+
+––
+
+
+
+––
WT
Mif
+/+
Mif
–/–
Il8rb
–/–
WT Il8rb
–/–
10
0
20
WT
Mif
+/+
Mif
–/–
WT Il8rb
–/–
P < 0.05
0
10
20
30
P < 0.01
P < 0.001
+
+
+
+––
0
10
30
20
P
+
+
+
+––
P < 0.05
+
+
+
+––
0
10
30
20
P < 0.05
+
+
+
+–––
+
+
+
+–––
Anti-Mif Anti-Mif
Anti-Mif
Anti-CXCR2 Anti-CXCR2
Anti-CXCR2
Anti-CD74 Anti-CD74
Anti-CD74
Control IgG Control IgG
Control IgG
Control IgG Control IgGControl IgG
Anti-CD74 Anti-CD74Anti-CD74
Anti-CXCR2 Anti-CXCR2Anti-CXCR2
Anti-Mif Anti-MifAnti-Mif
P < 0.05
MonoMac6 cell
arrest in Apoe
–/–
(cells/carotid artery)
MonoMac6 cell
arrest in Mif
+/+
(cells/carotid artery)
MonoMac6 cell
arrest in Mif
–/–
(cells/carotid artery)
MonoMac6 cell
arrest in Mif
+/+
Ldlr
–/–
(cells/carotid artery)
MonoMac6 cell
arrest in Mif
–/–
Ldlr
–/–
(cells/carotid artery)
MonoMac6 cell arrest
in Mif
–/–
Ldlr
–/–
(cells/carotid artery)
Monocyte arrest
(cells/field)
Il8rb
–/–
Figure 5 MIF-driven monocyte arrest in inflamed or atherosclerotic arteries involves CXCR2. (a) MonoMac6 cell arrest in carotid arteries from Apoe
–/–
mice
fed a western diet for 6 weeks. (b,c) MonoMac6 cell arrest in carotid arteries from Mif
+/+
and Mif
–/–
mice 4 h after intraperitoneal injection of TNF-a.
(df) MonoMac6 cell arrest in carotid arteries from Mif
+/+
Ldlr
–/–
and Mif
–/–
Ldlr
–/–
mice fed a western diet for 6 weeks (n ¼ 3each).Inf, carotid arteries
were loaded with MIF for 2 h before perfusion with MonoMac6 cells. After 10 min, adherent cells in 5–6 fields per carotid artery were counted. Data
represent mean ± s.d. of 3 independent experiments. Data in af correspond to Supplementary Figure 4 and Supplementary Video 1.(g,h) For intravital
microscopy, Mif
+/+
and Mif
–/–
mice reconstituted with wild-type or Il8rb
–/–
bone marrow (n ¼ 3 each) were stimulated by intraperitoneal injection of TNF-a
for 4 h, and the accumulation of leukocytes labeled by intravenous injection of rhodamine G was studied after 30 min in carotid arteries in vivo. Scale bar,
50 mm. Data in g are expressed as mean ± s.d. Representative segments are shown in h. For corresponding movies, please see Supplementary Video 2.
ARTICLES
6 ADVANCE ONLINE PUBLICATION NATU RE M EDI CINE
of rhodamine G–labeled leukocytes was attenuated in Mif
–/–
mice
reconstituted with wild-type bone marrow compared to that in wild-
type mice reconstituted with wild-type bone marrow. The reduction in
leukocyte accumulation due to deficiency in bone marrow Cxcr2 was
more marked in chimeric wild-type mice than in chimeric Mif
–/–
mice
(Fig. 5g,h and Supplementary Video 2).
MIF-induced inflammation in vivo relies on CXCR2
We aimed to corroborate the importance of CXCR2 for MIF-mediated
leukocyte recruitment under atherogenic or inflammatory conditions
in vivo. The adhesion of monocytes to the luminal surface of aortic
roots was reduced in Mif
–/–
Ldlr
–/–
versus Mif
+/+
Ldlr
–/–
mice with
primary atherosclerosis, and this was mirrored by a marked decrease
in lesional macrophage content (Fig. 6a). Intravital microscopy of
microcirculation in the cremaster muscle revealed that injecting MIF
adjacent to the muscle caused a marked increase in (mostly CD68
+
)
leukocyte adhesion and emigration in postcapillary venules (Supple-
mentary Video 3 online), which was inhibited by an antibody to
Cxcr2 (Fig. 6b,c). Circulating monocyte counts were unaffected (data
not shown).
We next used a model of MIF-induced peritonitis in chimeric mice
reconstituted with wild-type or Il8rb
–/–
bone marrow. Intraperitoneal
injection of MIF elicited neutrophil recruitment after 4 h in mice with
wild-type bone marrow, which was abrogated in mice with Il8rb
–/–
bone marrow (Fig. 6d). Collectively, our results reveal that MIF
triggers leukocyte recruitment under atherogenic and inflammatory
conditions in vivo through CXCR2.
Targeting MIF results in regression of atherosclerosis
As MIF can act through both CXCR2 and CXCR4, and given the role
of MIF and CXCR2 in the development of atherosclerotic lesions
5
,we
reasoned that targeting MIF, rather than CXCL1 or CXCL12, may be
ideally suited for modifying advanced lesions and their content of
CXCR2
+
monocytes and CXCR4
+
T cells. We treated Apoe
–/–
mice,
which had received a high-fat diet for 12 weeks and had developed
severe atherosclerotic lesions, with neutralizing antibodies to Mif,
Cxcl1 or Cxcl12 for 4 weeks. To verify the specificity of the Mif
antibody, we used immunoblotting and adhesion assays, and found
that it blocked MIF-induced, but not CXCL1- or CXCL8-induced,
arrest (Supplementary Fig. 4 online).
Blockade of Mif, but not Cxcl1 or Cxcl12, resulted in a
reduced plaque area in the aortic root at 16 weeks and a significant
(P o 0.05) plaque regression compared to baseline at 12 weeks
(Fig. 6e,f). In addition, blockade of Mif, but not Cxcl1 or
Cxcl12, was associated with a less inflammatory plaque phenotype
at 16 weeks, as evidenced by a lower content of both macrophages and
CD3
+
T cells (Fig. 6g,h). By targeting MIF, it might therefore be
possible to achieve therapeutic regression and stabilization of
advanced atherosclerotic lesions.
DISCUSSION
Our data reveal that MIF is a noncognate CXCR ligand; this may
qualify MIF as a pseudo-CXC or DC chemokine (given its lack of
N-terminal cysteines), extending the spectrum of binding partners
within the existing classification of chemokine receptors
3
. Alterna-
tively, MIF may, like CLF chemokines, have chemokine function
although lacking the primary structural features of chemokines. For
instance, human b-defensins bind and activate CCR6 to induce
chemotaxis of dendritic cells and T cells
11
; a tyrosyl-tRNA-synthetase
fragment binds to CXCR1, stimulating neutrophil chemotaxis;
and thioredoxin, a redox enzyme released during inflammation, is
chemoattractive for leukocytes
33,34
. As the inflammatory cytokine that
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
e
abcd
fg h
Monocyte adhesion to
lumen (cells/section)
Macrophage content (%)
0
20
10
0
40
20
Ldlr
–/–
Mif
–/–
Ldlr
–/–
Ldlr
–/–
Mif
–/–
Ldlr
–/–
Ldlr
–/–
Mif
–/–
Ldlr
–/–
P < 0.05
Monocyte adhesion in
venules (cells/100 µm)
0
10
5
Control lgG
12 weeks Control Anti-Cxcl12
Anti-MifAnti-Cxcl1
Anti-Cxcr2
P < 0.05
Neutrophils (%)
0
5
Control
WT
MIF
P < 0.05
10
P < 0.05
Monocyte emigration in
venules (cells/field)
0
20
10
Control lgG Anti-Cxcr2
P < 0.05
P < 0.01
Control
Il8rb
–/–
MIF
CD3
+
T-cell content
(cells/mm
2
)
0
80
40
120
P < 0.05
Anti-Cxcl12
Anti-Mif
Anti-Cxcl1
––
+–
+
+
MOMA-2
+
area
(× 10
3
µm
2
)
0
40
20
60
P < 0.05
Anti-Cxcl12
Anti-Mif
Anti-Cxcl1
12 weeks 16 weeks
––
+–
+
+
Plaque area
(% aortic root surface)
0
20
10
30
P < 0.05
Anti-Cxcl12
Anti-Mif
Anti-Cxcl1
––
+–
+
+
P < 0.01
Figure 6 MIF-induced atherogenic and microvascular inflammation through CXCR2 in vivo and effects of MIF blockade on plaque regression. (a) Monocyte
adhesion to the lumen in vivo and lesional macrophage content in native aortic roots were determined in Mif
+/+
Ldlr
–/–
and Mif
–/–
Ldlr
–/–
mice (n ¼ 4) fed a
chow diet for 30 weeks. Representative images are shown. Arrows indicate monocytes adherent to the luminal surface. Scale bar, 100 mm. (b,c)Exposure
to MIF induced Cxcr2-dependent leukocyte recruitment in vivo. Following intrascrotal injection of MIF, the cremasteric microvasculature was visualized by
intravital microscopy. Pretreatment with blocking Cxcr2 antibody abrogated adhesion and emigration, as compared to IgG control (n ¼ 4) (see also
Supplementary Video 3). (d) Intraperitoneal injection of MIF or vehicle elicited neutrophil recruitment in wild-type mice (n ¼ 3) reconstituted with wild-type,
but not Il8rb
–/–
,bonemarrow.(eh) Blocking Mif but not Cxcl1 or Cxcl12 resulted in regression and stabilization of advanced atherosclerotic plaques.
Apoe
–/–
mice received a high-fat diet for 12 weeks and were subsequently treated with antibodies to Mif, Cxcl1 or Cxcl12, or with vehicle (control) for an
additional 4 weeks of (n ¼ 6–10 mice). Plaques in the aortic root were stained using Oil-Red-O. Representative images are shown in e (scale bars, 500 mm).
Data in f represent plaque area at baseline (12 weeks) and after 16 weeks. The relative content of MOMA-2
+
macrophages is shown in g and the number of
CD3
+
T cells per section in h. Data represent mean ± s.d.
ARTICLES
NAT URE MED ICINE ADVANCE ONLINE PUBLICATION 7
has been known and studied for the longest period of time, MIF may
thus represent a prototypic member of the CLF family.
Some CLF polypeptides have been postulated to share tertiary
structural features with corresponding canonical ligands, enabling
them to use chemokine receptors. MIF monomers exhibit consider-
able three-dimensional architectural similarity to CXCL8 dimers
12
.
Recent studies have proposed that hetero-oligomerization of chemo-
kines can modify their activity to fine-tune leukocyte responses in
chemokine-rich microenvironments: for example, CXCL8 (hetero-)
dimers have been implicated in functional interactions with CXCR2
(ref. 35). The CXCL8 dimer–like architecture of MIF may thus have a
particular aptitude to bind the arrest receptor CXCR2. Activation of
CXCR2 by its cognate ligands requires an N-terminal Glu-Leu-Arg
(ELR) motif
32
. Notably, MIF features a pseudo-ELR motif, composed
of two nonadjacent but adequately spaced residues (Asp and Arg) in
exposed neighboring loops, mimicking that in ELR
+
chemokines
(Supplementary Fig. 1). However, these characteristics do not
preclude activation of other receptors, namely CXCR4. Identi-
fication of a MIF/CXCR4 axis challenges the dogma of a CXCL12/
CXCR4 monogamy.
The results reported here indicate that MIF, a highly conserved and
evolutionarily ancient molecule, is involved in leukocyte recruitment
in a more universal sense than hitherto appreciated. MIF has a similar
efficacy as ‘specialist’ cognate ligand chemokines. However, in contrast
to the canonical ligands CXCL8 and CXCL12, which act in a more
cell-restricted manner, MIF promotes the recruitment of both mono-
cytes and T cells by interacting with CXCR2 and CXCR4, respectively.
Through CXCR2, the chemotatic activity of MIF also extends to
neutrophils. By analogy to complement-based defense mechanisms,
MIF may represent an archaic ‘master regulator’ or source code for
leukocyte arrest and chemotaxis.
We found that atherogenic or inflammatory monocyte recruitment
induced by MIF relied not only on CXCR2 binding but also involved
the MIF-binding protein CD74, which colocalized with CXCR2. The
interaction of CXCR2 and CD74 suggests that MIF signals via a
functional CXCR/CD74 complex. CD74 lacks a signal-transducing
intracellular domain; however, MIF-induced signaling via CD74
involves the proteoglycan CD44 and Src kinases
19
. CD74, CD44 and
Src could form a functional receptor tyrosine kinase (RTK)-like
complex. These results suggest further complexities in cross-talk
regulation between RTK and GPCR pathways
36
. Although MIF can
bind to CXCR2 alone, as apparent in cells devoid of CD74, accessory
binding to CD74 (ref. 18) might facilitate GPCR activation and
formation of a signaling complex with Src kinases, resembling the
use of CD44 as an auxiliary receptor by CCL5 (ref. 37). The inhibitory
effects of blocking CD74 on CXCR2-dependent monocyte arrest
in Mif
/
arteries imply its participation, independently of MIF, in
CXCR2 signaling in order to accomplish atherogenic cell recruitment.
This role of CD74 is underscored by its preferential expression
on mononuclear cells relevant to atherosclerosis
18
but may also reflect
a more general involvement of CD74 in chemokine receptor signaling.
Assuming that CD74 is important for amplifying MIF signals,
its absence may explain the moderate recruitment activity of MIF
in neutrophils.
Using a peritonitis model as well as intravital microscopy in carotid
arteries with early atherosclerosis and in postcapillary venules, we have
provided several lines of evidence for the functional involvement of
CXCR2 in mediating MIF-triggered monocyte recruitment in vivo.For
both MIF and CXCR2, roles in the initiation and progression of
atherosclerosis have been established in various models
26,28,38
.How-
ever, an additional function of MIF as a T-cell agonist and its potential
to promote plaque progression have not been taken into account.
Notably, our data show that blocking Mif (as a dual Cxcr2/Cxcr4
agonist), but not the cognate ligands Cxcl1 and Cxcl2, leads to a
regression of pre-existing atherosclerotic plaques in Apoe
–/–
mice, and
that both macrophage and T-cell content are reduced, resulting in a
more stable plaque phenotype. Genetic deletion of Cxcl1 in Ldlr
–/–
mice reduces atherosclerosis to a smaller extent than does Il8rb
deficiency in bone marrow in Ldlr
–/–
mice; in both cases, macrophage
accumulation is affected in established, rather than early, lesions
38
.
This has prompted speculation that as yet unidentified CXCR2 ligands
may partially compensate for a lack of CXCL1. Our data identify MIF
as a crucial CXCR2 ligand in advanced atherosclerosis. Furthermore,
blockade of MIF impaired CXCR4-supported T-cell recruitment,
which is important for driving lesion development. Beyond strategies
to promote reverse cholesterol efflux and raise HDL cholesterol
levels
39
, targeting MIF may represent an approach to initiating
therapeutic regression and stabilizing advanced atherosclerosis.
Our data establish that MIF functions as a noncognate CXCR ligand
in microvascular and atherogenic recruitment of mononuclear cells,
and demonstrate its importance in the context of inflammation and
atherosclerosis. The growing number of chemokine receptor antago-
nists shown to interfere with mononuclear cell recruitment add to
existing evidence in favor of this therapeutic strategy despite the
apparent redundancy in the chemokine system
40
. Given that MIF is
becoming increasingly recognized as a target for treating inflammatory
disease
14
, and in light of intense efforts aimed at validating therapeutic
benefits of chemokine receptor antagonists, our findings in turn offer
the option of targeting MIF in human disease through such agents.
METHODS
Cells. Human aortic
26
and umbilical vein
30
endothelial cells (PromoCell),
MonoMac6 cells
41
and Chinese hamster ovary (CHO) ICAM-1-transfectants
42
were used as described. Jurkat cells and RAW264.7 macrophages were trans-
fected with pcDNA3-CXCR2. HL-60 cells were transfected with pcDNA3.1/V5-
HisTOPO-TA-CD74 or vector control (Nucleofector Kit V, Amaxa). L1.2 cells
were transfected with pcDNA3-CXCRs or pcDNA-CCR5 (UMR cDNA
Resource Center) for assays on simian virus-40–transformed mouse micro-
vascular endothelial cells (SVECs). Peripheral blood mononuclear cells were
prepared from buffy coats, monocytes by adherence or immunomagnetic
separation (Miltenyi), primary T cells by phytohaemaglutinin/interleukin-2
(Biosource) stimulation and/or immunomagnetic selection (antibody to CD3/
M-450 Dynabeads), and neutrophils by Ficoll gradient centrifugation. Human
embryonal kidney–CXCR2 transfectants (HEK293-CXCR2) have been
described previously
43
.
Reagents. Recombinant MIF was expressed and purified
15
. Chemokines were
from PeproTech. Human VCAM-1.Fc chimera, blocking antibodies to CXCR1
(42705, 5A12), CXCR2 (48311), CXCR4 (44708, FABSP2 cocktail, R&D),
human MIF and mouse Mif (NIHIII.D.9)
21
, CD74 (M-B741, Pharmingen),
b
2
integrin (TS1/18), a
4
integrin (HP2/1)
44
and CXCR2 (RII115), and antibody
to a
L
integrin (327C)
31
were used. PTX and B-oligomer were from Merck.
Adhesion assays. Arrest of calcein-AM (Molecular Probes)-labeled monocytes,
T cells and L1.2 transfectants was quantified in parallel-wall chambers in flow
(1.5 dynes/cm
2
, 5 min)
26,42,44
. Confluent endothelial cells, CHO-ICAM-1 cells,
VCAM-1.Fc-coated plates and leukocytes were pretreated with MIF, chemo-
kines or antibodies. CHO-ICAM-1 cells incubated with MIF (2 h) were stained
with antibody to MIF Ka565 (ref. 18) and FITC-conjugated antibody.
Chemotaxis assays. Using Transwell chambers (Costar), we quantified primary
leukocyte migration toward MIF or chemokines by fluorescence microscopy or
using calcein-AM labeling and FluoroBlok filters (Falcon). Cells were pretreated
with PTX/B-oligomer, Ly294002, MIF (for desensitization), antibodies to
CXCRs or CD74, or isotype IgG. Pore sizes and intervals were 5 mmand3h
(monocytes), 3 mm and 1.5 h (T cells), and 3 mm and 1 h (neutrophils).
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
8 ADVANCE ONLINE PUBLICATION NATU RE M EDI CINE
Q-PCR and ELISA. RNA was reverse-transcribed using oligo-dT primers. RT-
PCR was performed using QuantiTect Kit with SYBRGreen (Qiagen), specific
primers and an MJ Opticon2 (Biozym). CXCL8 was quantified by Quantikine
ELISA (R&D).
a
L
b
2
integrin activation assay. Monocytes stimulated with MIF or Mg
2+
/EGTA
(positive control) were fixed, reacted with the antibody 327C and an FITC-
conjugated antibody to mouse IgG. LFA-1 activation analyzed by flow
cytometry is reported as the increase in mean fluorescent intensity (MFI) or
relative to the positive control
31
.
Calcium mobilization. Neutrophils or L1.2 CXCR2 transfectants were labeled
with Fluo-4 AM (Molecular Probes). After the addition of the first or a
subsequent stimulus (MIF, CXCL8 or CXCL7), we monitored MFI as a measure
of cytosolic Ca
2+
concentrations for 120 s using a BD FACSAria. L1.2 controls
showed negligible calcium influx.
Receptor-binding assays. Because iodinated MIF is inactive
18,45
, we performed
competitive receptor binding
46
using radioiodinated tracers (Amersham):
[I
125
]CXCL8, reconstituted at 4 nM (80 mCi/ml) to a final concentration
of 40 pM; [I
125
]CXCL12, reconstituted at 5 nM (100 mCi/ml) to a
final concentration of 50 pM. For competition of [I
125
]CXCL8 with MIF for
CXCR2 binding or competition of [I
125
]CXCL12 with MIF for CXCR4
binding in equilibrium binding assays, we added cold MIF and/or CXCL with
tracers to HEK293-CXCR2 or CXCR4-bearing Jurkat cells, respectively. We
performed the analysis by liquid scintillation counting. To calculate EC
50
and
K
d
values, we assumed a one-site receptor-ligand binding model and used the
Cheng/Prusoff-equation and GraphPad Prism.
For pull-down of biotin-MIF-CXCR complexes, we incubated HEK293-
CXCR2 transfectants or controls with biotin-labeled MIF (ref. 45), and
then washed and lysed these with coimmunoprecipitation (CoIP) buffer
(see Supplementary Methods online for details). Complexes were isolated
from cleared lysates by streptavidin-coated magnetic beads (M280, Dynal)
and analyzed by western blotting with antibody to CXCR2 or streptavidin-
peroxidase. For flow cytometry, HEK293-CXCR2 transfectants or Jurkat
cells pretreated with AMD3465 and/or 20-fold excess of unlabeled
MIF were incubated with fluorescein-labeled MIF and analyzed using a
BD FACSCalibur.
CXCR internalization assays. We treated HEK293-CXCR2 or Jurkat cells with
CXCL8 or CXCL12, respectively, and with MIF, washed them with acidic
glycine-buffer, stained them with antibodies to CXCR2 or CXCR4, and
analyzed the cells by flow cytometry. We calculated internalization relative
to surface expression of buffer-treated cells (100% control) and isotype control
staining (0% control): geometric MFI[experimental]–MFI[0% control]/
MFI[100% control]–MFI[0% control] 100.
Colocalization of CXCR2 and CD74. RAW264.7-CXCR2 transfectants were
costained with CXCR2 and rat antibody to mouse CD74 (In-1, Pharmingen),
followed by FITC-conjugated antibody to rat IgG and Cy3-conjugated anti-
body to mouse IgG, and were analyzed by confocal laser scanning micro-
scopy (Zeiss).
Coimmunoprecipitation of CXCR2 and CD74. HEK293-CXCR2 cells tran-
siently transfected with pcDNA3.1/V5-HisTOPO-TA-CD74 were lysed in non-
denaturing CoIP buffer. Supernatants were incubated with the CXCR2
antibody RII115 or an isotype control, and were preblocked with protein G-
sepharose overnight. Proteins were analyzed by western blots using an antibody
to the His-tag (Santa Cruz). Similarly, CoIPs and immunoblots were performed
with antibodies to the His-tag and CXCR2, respectively. L1.2-CXCR2 cells were
subjected to immunoprecipitation with antibody to CXCR2 and immunoblot-
ting with an antibody to mouse CD74.
Ex vivo perfusion and intravital microscopy of carotid arteries. Mif
–/–
Ldlr
–/–
mice and Mif
+/+
Ldlr
–/–
littermate controls, crossbred from Mif
–/–
(ref. 47) and
Ldlr
–/–
mice (Charles River), and Apoe
–/–
mice were fed an atherogenic diet
(21% fat; Altromin) for 6 weeks. All single knockout strains had been back-
crossed in the C57BL/6 background ten times. Mif
+/+
and Mif
–/–
mice were
treated with TNF-a (intraperitoneally (i.p.), 4 h). Explanted arteries were
transferred onto the stage of an epifluorescence microscope and perfused at
4 ml/min with calcein-AM-labeled MonoMac6 cells treated with antibodies to
CD74 or CXCR2, isotype control IgG, or left untreated
29
. Untreated monocytic
cells were perfused after blockade with antibody to MIF for 30 min. For
intravital microscopy, rhodamine-G (Molecular Probes) was administered
intravenously (i.v.), and carotid arteries were exposed in anesthetized mice.
Arrest (430 s) of labeled leukocytes was analyzed by epifluorescence
microscopy (Zeiss Axiotech, 20 water immersion). All studies were approved
by local authorities (Bezirksregierung Ko
¨
ln), and complied with German
animal protection law Az: 50.203.2-AC 36, 19/05.
Mouse model of atherosclerotic disease progression. Apoe
–/–
mice fed an
atherogenic diet for 12 weeks were injected (3 injections per week, each 50 mg)
with antibodies to Mif (NIHIIID.9), Cxcl12 (79014) or Cxcl1 (124014, R&D)
(n ¼ 6–10 mice) for an additional 4 weeks. Aortic roots were fixed by in situ
perfusion and atherosclerosis was quantified by staining transversal sections
with Oil-Red-O. We determined relative macrophage and T-cell contents by
staining with antibodies to MOMA-2 (MCA519, Serotec) or to CD3 (PC3/
188A, Dako) and FITC-conjugated antibody. In Mif
–/–
Ldlr
–/–
and Mif
+/+
Ldlr
–/–
mice fed a chow diet for 30 weeks, we determined the abundance of luminal
monocytes and lesional macrophages in aortic roots
48
.
Cremaster microcirculation model. We injected human MIF (1 mg) intra-
scrotally and exteriorized the cremaster muscle in mice treated with antibody to
Cxcr2 (100 mg i.p.). After 4 h, we performed intravital microscopy (Zeiss
Axioplan; 20) in postcapillary venules (ref. 22,49). Adhesion was measured as
leukocytes stationary for more than 30 s, emigration as the number of
extravascular leukocytes per field.
Bone marrow transplantation. We aseptically removed femurs and tibias from
donor Il8rb
–/–
(Jackson Laboratories) or BALB/c mice. The cells flushed from
the marrow cavities were administered i.v. into Mif
+/+
or Mif
–/–
mice 24 h
after ablative whole-body irradiation
50
.
Model of acute peritonitis. Mice repopulated with Il8rb
+/+
or Il8rb
–/–
bone
marrow were injected i.p. with MIF (200 ng). After 4 h, we performed
peritoneal lavage and quantified Gr-1
+
CD115
F4/80
neutrophils, using the
relevant conjugated antibodies, by flow cytometry.
Statistical analysis. To compare data, we used either a one-way analysis of
variance (ANOVA) and Newman-Keuls post-hoc test or an unpaired Student’s
t-test with Welchs correction (using GraphPad Prism).
Note: Supplementary information is available on the Nature Medicine website.
ACKNOWLEDGMENTS
We thank E. Liehn, S. Knarren and L. Verschuren for assistance with the
atherosclerotic mouse models; S. Kraemer for help with internalization
assays; A. Ben-Baruch (Department of Cell Research and Immunology,
Tel Aviv University) for HEK293-CXCR2 cells; H.W.L. Ziegler-Heitbrock
(University of Leicester) for MonoMac6 cells; M. Locati (Istituto Clinico
Humanitas) for L1.2 cells; H. Hengel (University of Du
¨
sseldorf) for
SVECs; A. Ludwig and E. Brandt (Department of Immunology and Cell
Biology, Forschungszentrum Borstel) for CXCR2 antibody RII115; D. Staunton
(Department of Biomedical Engineering, Genome and Biomedical Sciences
Facility, University of California at Davis) for the antibody to 327C; Anormed
Inc. (Genzyme Corporate Offices) for AMD3465; and A. Ludwig, E. Morand,
M. Thelen and A. Kapurniotu for helpful discussions. Supported by the Deutsche
Forschungsgemeinschaft grants BE 1977/2-1, BE 1977/4-1 (J.B. and C.W.),
WE 1913/7-1 (C.W.), SFB542-A7 (J.B.) and SFB542-C12 (C.W.); US National
Institutes of Health grants AI43210 and AR49610 (R.B.); Australian National
Health and Medical Research Council (NHMRC) program grant 334067
(M.J.H. and S.R.M.); US NIH grant AR51807-01 (M.J.H. and S.R.M.); and the
Netherlands Organization for Scientific Research grant VENI 016.036.061 (R.K.).
COMPETING INTERESTS STATEMENT
The authors declare competing financial interests: details accompany the full-text
HTML version of the paper at www.nature.com/naturemedicine/.
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
NAT URE MED ICINE ADVANCE ONLINE PUBLICATION 9
Published online at http://www.nature.com/naturemedicine
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions
1. Baggiolini, M., Dewald, B. & Moser, B. Interleukin-8 and related chemotactic
cytokines–CXC and CC chemokines. Adv. Immunol. 55, 97–179 (1994).
2. Charo, I.F. & Ransohoff, R.M. The many roles of chemokines and chemokine receptors
in inflammation. N. Engl. J. Med. 354, 610–621 (2006).
3. Murphy, P.M. et al. International union of pharmacology. XXII. Nomenclature for
chemokine receptors. Pharmacol. Rev. 52, 145–176 (2000).
4. Rot, A. & von Andrian, U.H. Chemokines in innate and adaptive host defense: basic
chemokinese grammar for immune cells. Annu. Rev. Immunol. 22, 891–928 (2004).
5. Weber, C., Schober, A. & Zernecke, A. Chemokines: key regulators of mononuclear cell
recruitment in atherosclerotic vascular disease. Arterioscler. Thromb. Vasc. Biol. 24,
1997–2008 (2004).
6. Libby, P. Inflammation in atherosclerosis. Nature 420, 868–874 (2002).
7. Hansson, G.K. Inflammation, atherosclerosis, and coronary artery disease. N. Engl.
J. Med. 352, 1685–1695 (2005).
8. Tedgui, A. & Mallat, Z. Cytokines in atherosclerosis: pathogenic and regulatory path-
ways. Physiol. Rev. 86, 515–581 (2006).
9. Laudanna, C. & Alon, R. Right on the spot. Chemokine triggering of integrin-mediated
arrest of rolling leukocytes. Thromb. Haemost. 95,511(2006).
10. Degryse, B. & deVirgilio, M. The nuclear protein HMGB1, a new kind of chemokine?
FEBS Lett. 553, 11–17 (2003).
11. Yang, D. et al. b-defensins: linking innate and adaptive immunity through dendritic and
TcellCCR6.Science 286, 525–528 (1999).
12. Sun, H.W., Bernhagen, J., Bucala, R. & Lolis, E. Crystal structure at 2.6 A
˚
resolution of
human macrophage migration inhibitory factor. Proc. Natl. Acad. Sci. USA 93,
5191–5196 (1996).
13. Donnelly, S.C. et al. Regulatory role for macrophage migration inhibitory factor in acute
respiratory distress syndrome. Nat. Med. 3, 320–323 (1997).
14. Morand, E.F., Leech, M. & Bernhagen, J. MIF: a new cytokine link between rheumatoid
arthritis and atherosclerosis. Nat. Rev. Drug Discov. 5, 399–410 (2006).
15. Bernhagen, J. et al. MIF is a pituitary-derived cytokine that potentiates lethal
endotoxaemia. Nature 365, 756–759 (1993).
16. Calandra, T. & Roger, T. Macrophage migration inhibitory factor: a regulator of innate
immunity. Nat. Rev. Immunol. 3, 791–800 (2003).
17. Kleemann, R. et al. Intracellular action of the cytokine MIF to modulate AP-1 activity
and the cell cycle through Jab1. Nature 408, 211–216 (2000).
18. Leng, L. et al. MIF signal transduction initiated by binding to CD74. J. Exp. Med. 197,
1467–1476 (2003).
19. Shi, X. et al. CD44 is the signaling component of the macrophage migration inhibitory
factor-CD74 receptor complex. Immunity 25, 595–606 (2006).
20. David, J.R. Delayed hypersensitivity in vitro: its mediation by cell-free substances
formed by lymphoid cell-antigen interaction. Proc. Natl. Acad. Sci. USA 56,7277
(1966).
21. Lan, H.Y. et al. The pathogenic role of macrophage migration inhibitory factor in
immunologically induced kidney disease in the rat. J. Exp. Med. 185, 1455–1465
(1997).
22. Gregory, J.L. et al. Reduced leukocyte-endothelial cell interactions in the inflamed
microcirculation of macrophage migration inhibitory factor-deficient mice. Arthritis
Rheum. 50, 3023–3034 (2004).
23. Schrans-Stassen, B.H.G.J., Lue, H., Sonnemans, D.G.P., Bernhagen, J. & Post, M.J.
Stimulation of vascular smooth muscle cell migration by macrophage migration
inhibitory factor (MIF). Antioxid. Redox Signal. 7, 1211–1216 (2005).
24. Burger-Kentischer, A. et al. Expression of macrophage migration inhibitory factor in
different stages of human atherosclerosis. Circulation 105, 1561–1566 (2002).
25. Lin, S.G. et al. De novo expression of macrophage migration inhibitory factor in
atherogenesis in rabbits. Circ. Res. 87, 1202–1208 (2000).
26. Schober, A. et al. Stabilization of atherosclerotic plaques by blockade of macrophage
migration inhibitory factor after vascular injury in apolipoprotein E-deficient mice.
Circulation 109, 380–385 (2004).
27. Kong, Y.Z. et al. Macrophage migration inhibitory factor induces MMP-9 expression:
implications for destabilization of human atherosclerotic plaques. Atherosclerosis 178,
207–215 (2005).
28. Pan, J.H. et al. Macrophage migration inhibitory factor deficiency impairs athero-
sclerosis in low-density lipoprotein receptor-deficient mice. Circulation 109, 3149–
3153 (2004).
29. Huo, Y. et al. The chemokine KC, but not monocyte chemoattractant protein-1, triggers
monocyte arrest on early atherosclerotic endothelium. J. Clin. Invest. 108, 1307–1314
(2001).
30. Weber, K.S., von Hundelshausen, P., Clark-Lewis, I., Weber, P.C. & Weber, C.
Differential immobilization and hierarchical involvement of chemokines in monocyte
arrest and transmigration on inflamed endothelium in shear flow. Eur. J. Immunol. 29,
700–712 (1999).
31. Shamri, R. et al. Lymphocyte arrest requires instantaneous induction of an extended
LFA-1 conformation mediated by endothelium-bound chemokines. Nat. Immunol. 6,
497–506 (2005).
32. Jones, S.A., Dewald, B., Clark-Lewis, I. & Baggiolini, M. Chemokine antagonists that
discriminate between interleukin-8 receptors. Selective blockers of CXCR2. J. Biol.
Chem. 272, 16166–16169 (1997).
33. Bertini, R. et al. Thioredoxin, a redox enzyme released in infection and inflammation, is
a unique chemoattractant for neutrophils, monocytes, and T cells. J. Exp. Med. 189,
1783–1789 (1999).
34. Wakasugi, K. & Schimmel, P. Two distinct cytokines released from a human aminoacyl-
tRNA synthetase. Science 284, 147–151 (1999).
35. Weber, C. & Koenen, R.R. Fine-tuning leukocyte responses: towards a chemokine
‘interactome’. Trends Immunol. 27, 268–273 (2006).
36. Waters, C.M., Connell, M.C., Pyne, S. & Pyne, N.J. c-Src is involved in regulating signal
transmission from PDGFb receptor-GPCR(s) complexes in mammalian cells. Cell.
Signal. 17, 263–277 (2005).
37. Roscic-Mrkic, B. et al. RANTES (CCL5) uses the proteoglycan CD44 as an auxiliary
receptor to mediate cellular activation signals and HIV-1 enhancement. Blood 102,
1169–1177 (2003).
38. Boisvert, W.A. et al. Up-regulated expression of the CXCR2 ligand KC/GRO-a in
atherosclerotic lesions plays a central role in macrophage accumulation and lesion
progression. Am.J.Pathol.168, 1385–1395 (2006).
39. Navab, M. et al. Human apolipoprotein A-I and A-I mimetic peptides: potential for
atherosclerosis reversal. Curr. Opin. Lipidol. 15, 645–649 (2004).
40. Proudfoot, A.E. Chemokine receptors: multifaceted therapeutic targets. Nat. Rev.
Immunol. 2, 106–115 (2002).
41. Weber, C., Aepfelbacher, M., Haag, H., Ziegler-Heitbrock, H.W. & Weber, P.C. TNF
induces enhanced responses to platelet-activating factor and differentiation in human
monocytic Mono Mac 6 cells. Eur. J. Immunol. 23, 852–859 (1993).
42. Ostermann, G., Weber, K.S., Zernecke, A., Schroder, A. & Weber, C. JAM-1 is a ligand
of the b(2) integrin LFA-1 involved in transendothelial migration of leukocytes. Nat.
Immunol. 3, 151–158 (2002).
43. Ben-Baruch, A. et al. IL-8 and NAP-2 differ in their capacities to bind and chemo-
attract 293 cells transfected with either IL-8 receptor type A or type B. Cytokine 9,
37–45 (1997).
44. Weber, C., Alon, R., Moser, B. & Springer, T.A. Sequential regulation of a4b1anda5b1
integrin avidity by CC chemokines in monocytes: implications for transendothelial
chemotaxis. J. Cell Biol. 134, 1063–1073 (1996).
45. Kleemann, R., Grell, M., Mischke, R., Zimmermann, G. & Bernhagen, J. Receptor-
binding and cellular uptake studies of macrophage migration inhibitory factor (MIF):
use of biologically active MIF derivatives. J. Interferon Cytokine Res. 22, 351–363
(2002).
46. Hayashi, S. et al. Synthetic hexa- and heptapeptides that inhibit IL-8 from binding to
and activating human blood neutrophils. J. Immunol. 154, 814–824 (1995).
47. Fingerle-Rowson, G. et al. The p53-dependent effects of macrophage migration
inhibitory factor revealed by gene targeting. Proc. Natl. Acad. Sci. USA 100,
9354–9359 (2003).
48. Verschuren, L. et al. Effect of low dose atorvastatin versus diet-induced cholesterol
lowering on atherosclerotic lesion progression and inflammation in apolipoprotein
E*3-Leiden transgenic mice. Arterioscler. Thromb. Vasc. Biol. 25, 161–167 (2005).
49. Keane, M.P., Belperio, J.A., Xue, Y.Y., Burdick, M.D. & Strieter, R.M. Depletion of
CXCR2 inhibits tumor growth and angiogenesis in a murine model of lung cancer.
J. Immunol. 172, 2853–2860 (2004).
50. Zernecke, A. et al. SDF-1a/CXCR4 axis is instrumental in neointimal hyperplasia and
recruitment of smooth muscle progenitor cells. Circ. Res. 96, 784–791 (2005).
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
ARTICLES
10 ADVANCE ONLINE PUBLICATION NATURE ME DICI NE
    • "The invariant chain or CD74 was the first MIF surface receptor described [3]. Chemokine receptors CXCR2 and CXCR4 have also been shown to be MIF receptors161718. In murine models of human colorectal adenoma [16] and metastatic breast cancer [19], inhibition of MIF expression by genetic deletion or RNA interference decreased tumor progression and metastasis. "
    [Show abstract] [Hide abstract] ABSTRACT: Macrophage migration inhibitory factor (MIF) is a pro-inflammatory cytokine implicated in acute and chronic inflammatory diseases. MIF is overexpressed in various tumors. It displays a number of functions that provide a direct link between the process of inflammation and tumor growth. Our group recently identified the MIF-receptor CD74 as an independent prognostic factor for overall survival in patients with malignant pleural mesothelioma. In the present study, we compared the levels of expression of MIF and CD74 in different human mesothelioma cell lines and investigated their physiopathological functions in vitro and in vivo. Human mesothelioma cells expressed more CD74 and secreted less MIF than non tumoral MeT5A cells, suggesting a higher sensitivity to MIF. In mesothelioma cells, high MIF levels were associated with a high multiplication rate of cells. In vitro, reduction of MIF or CD74 levels in both mesothelioma cell lines showed that the MIF/CD74 signaling pathway promoted tumor cell proliferation and protected MPM cells from apoptosis. Finally, mesothelioma cell lines expressing high CD74 levels had a low tumorigenic potential after xenogeneic implantation in athymic nude mice. All these data highlight the complexity of the MIF/CD74 signaling pathway in the development of mesothelioma.
    Full-text · Article · Feb 2016
    • "Biochemically, MIF is comprised of 115 aminoacids with a molecular weight of 12.5 kDa (Weiser et al., 1989), whereas in its crystal structure MIF reveals an active form a 37.5 kDa protein expressed in a monomer (44%), homodimer (33%) or homotrimer (23%) form (Bendrat et al., 1997; Mischke et al., 1998). CXCR2 and CD74 have been demonstrated to play the role of MIF receptor in monocyte recruitment in early atherosclerotic endothelium (Bernhagen et al., 2007). In another series of studies, phosphorylation of p44/p42 (ERK-1/2) "
    [Show abstract] [Hide abstract] ABSTRACT: Invariant chain (Ii) or CD74 is a non-polymorphic glycoprotein, which apart from its role as a chaperone dedicated to MHCII molecules, is known to be a high-affinity receptor for Macrophage Migration Inhibitory Factor (MIF). The present study aimed to define the roles of CD74 and MIF in the immune surveillance escape process. Towards this direction, the cell lines HL-60, Raji, K562 and primary pre-B leukemic cells were examined for expression and secretion of MIF. Flow cytometry analysis detected high levels of MIF and intracellular/membrane CD74 expression in all leukemic cells tested, while MIF secretion was shown to be inversely proportional to intracellular HLA-DR (DR) expression. In the MHCII-negative cells, IFN-γ increased MIF expression and induced its secretion in HL-60 and K562 cells, respectively. In K562 cells, CD74 (Iip33Iip35) was shown to co-precipitate with HLA-DOβ (DOβ), inhibiting thus MIF or DR binding. Induced expression of DOα in K562 (DOα-DOβ+) cells in different transfection combinations decreased MIF expression and secretion, while increasing surface DR expression. Thus, MIF could indeed be part of the antigen presentation process.
    Article · Feb 2016
    • "Following secretion, MIF binds to extracellular receptors, in particular the CD74 receptor, and is also internalized where it interacts with intracellular proteins. It has also been shown that MIF is a non-cognate ligand of the CXC chemokine receptors CXCR2 and CXCR4 [7,8] . MIF exists as a homotrimeric protein and possesses ketoeenol tautomerase activity (against substrates L-dopachrome and phenylpyruvate), which is catalyzed by its N-terminal proline [9,10]. "
    [Show abstract] [Hide abstract] ABSTRACT: Macrophage migration inhibitory factor (MIF) is a pleiotropic cytokine that has roles in the innate immune response, and also contributes to inflammatory disease. While the biological properties of MIF are closely linked to protein-protein interactions, MIF also has tautomerase activity. Inhibition of this activity interferes with the interaction of MIF with protein partners e.g. the CD74 receptor, and tautomerase inhibitors show promise in disease models including multiple sclerosis and colitis. Isothiocyanates inhibit MIF tautomerase activity via covalent modification of the N-terminal proline. We systematically explored variants of benzyl and phenethyl isothiocyanates, to define determinants of inhibition. In particular, substitution with hydroxyl, chloro, fluoro and trifluoro moieties at the para and meta positions were evaluated. In assays on treated cells and recombinant protein, the IC50 varied from 250 nM to >100 μM. X-ray crystal structures of selected complexes revealed that two binding modes are accessed by some compounds, perhaps owing to strain in short linkers between the isothiocyanate and aromatic ring. The variety of binding modes confirms the existence of two subsites for inhibitors and establishes a platform for the development of potent inhibitors of MIF that only need to target one of these subsites. Copyright © 2015 Elsevier Masson SAS. All rights reserved.
    Full-text · Article · Feb 2015
Show more