Content uploaded by Marion Hofmann Bowman
Author content
All content in this area was uploaded by Marion Hofmann Bowman on Sep 04, 2018
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Cell, Vol. 97, 889–901, June 25, 1999, Copyright 1999 by Cell Press
RAGE Mediates a Novel Proinflammatory Axis:
A Central Cell Surface Receptor
for S100/Calgranulin Polypeptides
teraction might trigger intracellular signal transduction
mechanisms altering properties of vascular and inflam-
matory effector cells, thereby contributing to impaired
reparative responses in AGE-rich tissues, as occurs in
diabetes (King and Brownlee, 1996). Indeed, AGE liga-
Marion A. Hofmann,* Steven Drury,*
Caifeng Fu,* Wu Qu,* Akihiko Taguchi,*
Yan Lu,* Cecilia Avila,* Neeraja Kambham,*
Angelika Bierhaus,
‡
Peter Nawroth,
‡
Markus F. Neurath,
§
Timothy Slattery,
†
tion of RAGE activates p21
ras
, recruiting downstream
Dale Beach,
†
John McClary,
†
targets, such as MEK and MAP kinases, and activating
Mariko Nagashima,
†
John Morser,
†
David Stern,* the transcription factor NF-kB; this represents the first
and Ann Marie Schmidt*
k
receptor-dependent signal transduction pathway for
*College of Physicians and Surgeons AGEs (Lander et al., 1997). Subsequent studies demon-
Columbia University strating that RAGE could serve as a cell surface receptor
New York, New York 10032 for amyloid-bpeptide (Ab) (Yan et al., 1996, 1997), a
†
Berlex Biosciences cleavage product of the b-amyloid precursor protein
Richmond, California 94804 that accumulates and has been ascribed a pathogenic
‡
University of Tu
¨bingen role in Alzheimer’s disease (Selkoe, 1994), extends the
Tu
¨bingen 72076 concept of RAGE as a receptor that converts protein-
Germany aceous deposits, of glycoxidized adducts or Ab, into
§
University of Mainz bioactivespecies capableof modulating cellular proper-
Mainz 55101 ties. This view of the biology of RAGE contrasts with
Germany the effective uptake and disposal of AGEs, Ab, and other
ligands by the scavenger receptor (Krieger and Herz,
1994); RAGE is much less efficient in mediating endocy-
Summary tosis and degradation of bound ligands (Mackic et al.,
1998), but, rather, ligand–receptor interaction causes
S100/calgranulin polypeptides are present at sites of cellular perturbation (Wautier et al., 1996; Park et al.,
inflammation,likely releasedby inflammatorycells tar- 1998).
geted to such loci by a range of environmental cues. Expression of RAGE at high levels during central ner-
We report here that receptor for AGE (RAGE) is a cen- vous system development and the observation that lung
tral cell surface receptor for EN-RAGE (extracellular was a rich source of RAGE in adult animals suggested
newly identified RAGE-binding protein) and related quite different roles for the receptor than might be ex-
membersof theS100/calgranulin superfamily.Interac- pected if the molecule evolved solely to interact with
tion of EN-RAGEs with cellular RAGE on endothelium, AGEs and Ab. In fact, a nonglycated polypeptide, am-
mononuclear phagocytes, and lymphocytes triggers photerin, is a RAGE ligand present in developing brain
cellular activation, with generation of key proinflamma- in an overlapping distribution with cells expressing the
tory mediators. Blockade of EN-RAGE/RAGE quenches receptor (Hori et al., 1995). In contrast to this physiologic
delayed-type hypersensitivity and inflammatory colitis expression and, possibly, function of RAGE in develop-
in murine models by arresting activation of central ment, the presence of receptor in the lung raised the
signaling pathways and expression of inflammatory question of a contribution of RAGE in the response to
genemediators. Thesedata highlight anovel paradigm environmental challenge. Our analysis of lung tissue for
in inflammation and identify roles for EN-RAGEs and “natural” ligands of RAGE has yielded an intriguing an-
RAGE in chronic cellular activation and tissue injury. swer to this question and has provided insight into a
novel axis in the biology of inflammation; RAGE is a
signal transduction receptor for S100/calgranulin-like
Introduction molecules.
The S100/calgranulin family is comprised of closely
related polypeptides released from activated inflamma-
The receptor for advanced glycation end products tory cells, including polymorphonuclear leukocytes, pe-
(RAGE) is a multiligand member of the immunoglobulin ripheral blood-derived mononuclear phagocytes, and
superfamily of cell surface molecules (Neeper et al., lymphocytes (Zimmer et al., 1995; Schafer and Heinz-
1992; Schmidt et al., 1992). RAGE was originally identi- mann, 1996). Their hallmark is accumulation at sites of
fied and characterized based on its ability to bind ad- chronic inflammation. To date, however, specific means
vanced glycation end products (AGEs), adducts formed by which these polypeptides modulate the course of
by glycoxidation that accumulate in disorders such as inflammatory processes have not been elucidated. We
diabetes and renal failure (Brownlee et al., 1988; Miyata report here the characterization of an z12 kDa polypep-
et al., 1996). The proximity of cells expressing RAGE to tide, termed EN-RAGE (extracellular newly identified
lesional areas rich in AGEs, and the ensuing cellular RAGE-binding protein), which is in the S100/calgranulin
activation, suggested the possibility that AGE–RAGE in- family. Ligation of cellular RAGE by EN-RAGE and EN-
RAGE-like molecules (another S100 family member) me-
diates activation of endothelial cells, macrophages, and
k
To whom correspondence should be addressed (e-mail: ams11@
columbia.edu).
lymphocytes,cells centraltothe inflammatoryresponse.
Cell
890
Table 1. Amino Acid Sequence Analysis of p12, Later Termed “EN-RAGE,” and Comparisons with Homologous Polypeptides
1102030
P12 N TERM TKLEDHLEGIINIFHQYSVRVGHFDTLNKY
P12 CNBR
B-COAg TKLEDHLEGIINIFHQYSVRVGHFDTLNKR
B-CAAF1 TKLEDHLEGIINIFHQYSVRVGHFDTLNKR
31 40 50 60
P12 N TERM ELKQLGTKELPKTLQNXKDQ
P12 CNBR
B-COAg ELKQLITKELPKTLQNTKDQPTIDKIFQDL
B-CAAF1 ELKQLITKELPKTLQNTKDQPTIDKIFQDL
61 70 80 90
P12 N TERM
P12 CNBR DGAVSFEEFVVLVSRVLK
B-COAg DADKDGAVSFEEFVVLVSRVLKTAHIDIHK
B-CAAF1 DADKDGAVSFEEFVVLVSRVLKTAHIDIHK
The amino acid sequence analysis of EN-RAGE was compared with homologous polypeptides bovine corneal antigen and bovine CAAF1.
The latter sequences were obtained from Gottsch et al. (1997). A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M,
Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. X is an amino acid residue not identified at that position.
Consistent with the concept that EN-RAGE interaction the name EN-RAGE. Subsequently, sequences entered
in the database showed EN-RAGE to bear striking ho-with RAGE contributes importantly to a proximal step
in the cascade of events amplifying inflammation, block- mology to a polypeptide known as bovine calcium-bind-
ing protein in amniotic fluid-1 (CAAF-1; Hitomi et al.,ade of RAGE suppressed inflammation in acute and
chronic models. In parallel, inhibition of RAGE–S100/ 1996) or bovine corneal antigen/calgranulin C (Gottsch
et al., 1997), recently classified as S100A12 (Ilg et al.,calgranulin interaction decreased NF-kB activation and
expressionof proinflammatorycytokines in tissues,sug- 1996). Although the protein sequence of EN-RAGE dif-
fered at two positions (amino acids 30 and 36) from thatgesting that receptor blockade altered the course of
the inflammatory response. These data thus highlight a of bovine corneal antigen (Table 1), molecular cloning
studies firmly placed EN-RAGE in the calgranulin/S100novel paradigm in inflammation and identify roles for
EN-RAGEs and RAGE in chronic cellular activation and family (Dell’Angelica et al., 1994; Ilg et al., 1996; Wicki
et al., 1996; Gottsch et al., 1997) and led to isolation oftissue injury. clones identical to the deduced amino acid sequence
of bovine corneal antigen. Analysis of the sequence indi-Results cated that EN-RAGE was most likely the bovine counter-
part of human calgranulin C (or human corneal antigen/Characterization of the z12 kDa
RAGE-Binding Protein S100A12) (Ilg et al., 1996; Yamamura et al., 1996), dis-
playing .77% homology. Since the term EN-RAGE de-The evolution of an immunoglobulin superfamily mole-
cule solely to engage the products of nonenzymatic scribed a potentially central property of the z12 kDa
polypeptide, and, potentially other S100/calgranulins,glycoxidation or Abseemed improbable, especially in
view of the multiligand specificity of certain other mem- their interaction with RAGE, we have retained this name
at this time, at least until nomenclature for members ofbers of this family(Springer, 1990). We thus considered
the possibility that RAGE would recognize other physio- this family is standardized. We thus exploited implica-
tions of the term “EN-RAGE” to determine whether thelogically/pathophysiologically relevant ligands. As RAGE
was expressed at highest levels in the lung, this tissue EN-RAGE/RAGE axis provides a critical, previously un-
recognized link between the family of S100/calgranulinseemed a logical place to launch a search for natural
ligands. Indeed, bovine lung extract displayed binding proteins and a signal transduction receptor permitting
them to exert cellular effects.activity for RAGE, which, following column chromatog-
raphy, was resolved into two discrete polypeptides. The
first was identified as amphoterin, a high mobility group EN-RAGE Expressed by Inflammatory Cells
Stimulates Cellular Activation via RAGE1 nonhistone nuclear protein also present extracellularly
(Rauvala and Pihlaskari, 1987). Amphoterin participates Members oftheS100/calgranulin familyhavebeen asso-
ciatedwith a range ofinflammatorydisorders, especiallyin cell matrix interactions, mediated at least in part by
RAGE, and, because of its high levels in the embryonic those of a chronic nature. These polypeptides arepres-
ent at inflammatory loci, most likely due to their releaseperiod, amphoterin–RAGE interaction might prove im-
portant for guiding developmental processes (Hori et by inflammatory effector cells, such as polymorphonu-
clear leukocytes and peripheral blood-derived mononu-al., 1995). The second RAGE-binding species, an z12
kDa polypeptide, was subjected to N-terminal and inter- clear phagocytes. First, it was important to demonstrate
upregulation of EN-RAGE in stimulated inflammatory ef-nal sequence analysis, thelatter following digestion with
theendopeptidase Lys-C(Table 1).When firstcompared fector cells and then to show their release following
appropriate stimulation. Peripheral blood-derived mono-with sequences available in the database in 1995, the
sequence obtained was unique, leading us to assign it nuclear cells (PBMCs) or Jurkat cells, an immortalized
RAGE and Modulation of the Inflammatory Response
891
Figure 1. Expression of EN-RAGE Is Increased in Stimulated Inflammatory Cells, and EN-RAGE Binds RAGE
(A) Expression of EN-RAGE in stimulated PBMCs and Jurkat cells. PBMCs, Jurkat E6 cells, or HUVECs were cultured alone or in the presence
of the indicated stimuli. Cell lysates were prepared and electrophoresis/immunoblotting performed employing rabbit anti-EN-RAGE IgG (2 mg/
ml). Levels of IL-2 elaborated into cellular supernatant were determined by ELISA to control for extent of stimulation by cross-linking CD3/
CD28. In PBMCs (2/1stimulation), levels of IL-2 were 30 67 and 750 635 pg/ml, respectively. In Jurkat cells (2/1stimulation), levels of
IL-2 were 40 67 and 2100 670 pg/ml, respectively.
(B) Infusion of LPS into mice results in elaboration of EN-RAGE into plasma. LPS (30 mg/kg body weight) was infused into CF-1 mice. At the
indicated time, blood was retrieved and plasma was subjected to electrophoresis/immunoblotting for EN-RAGE using anti-EN-RAGE IgG as
above. In (A) and (B), results of densitometric analysis and mean 6SD of three experiments are shown.
(C) EN-RAGE binds purified RAGE. Murine soluble RAGE was immobilized onto the wells of plastic dishes. Radioligand binding assays were
performed employing the indicated concentration of
125
I-EN-RAGE in the presence or absence of excess unlabeled EN-RAGE (50-fold). Specific
binding to purified RAGE is demonstrated, with K
D
≈91 629 nM and capacity ≈21 62.9 fmol/well. Where indicated, radiolabeled EN-RAGE
(100 nM) was preincubated with either sRAGE, or wells were incubated with alternate RAGE ligand (50-fold molar excess in each case).
Alternatively, wells were preincubated with the indicated concentration of nonimmune IgG, or anti-RAGE IgG prior to binding assay. Results
are reported as percent of maximal specific binding 6SD.
T cell line, stimulated by cross-linking CD3/CD28, dis- IgG were without effect. Further, alternate ligands for
played increased EN-RAGE in cellular homogenates, RAGE, amphoterin, AGE albumin, and amyloid-bpep-
corresponding to z4.1-fold and 2.5-fold, respectively tide, similarly inhibited binding of EN-RAGE to immobi-
(Figure 1A). In contrast, stimulation of cultured human lized RAGE (Figure 1C, inset). The interaction of S100/
umbilical vein endothelial cells (HUVECs) with a proto- calgranulin polypeptides with RAGE was not limited to
typic stimulus, tumor necrosis factor a(Figure 1A), did EN-RAGE, as S100B also suppressed
125
I-EN-RAGE in-
not modulate expression of EN-RAGE. To address the teraction with RAGE (data not shown). Similar radioli-
second issue—the extracellular release of EN-RAGE— gand binding experiments were performed with RAGE
mice were infused with lipopolysaccharide (LPS), and in its natural cellular environment using cultured endo-
plasma was assayed for EN-RAGE. Time-dependent re- thelial cells that endogenously express cell surface
lease of EN-RAGE into plasma was noted with maximal RAGE. Cultured endothelial cells displayed saturable
levels by 12 hr of injection, corresponding to an z3.6- binding of
125
I-EN-RAGE, with K
D
≈90.3 634 nM; binding
fold increase (Figure 1B). was suppressed in the presence of either excess sRAGE
To directly test the concept that EN-RAGE bound or anti-RAGE IgG (data not shown).
RAGE, radioligand binding studies were performed. Upregulation of EN-RAGE, its release in response to
Specific binding of
125
I-EN-RAGE to purified RAGE on inflammatory stimuli, and its ability to bind RAGE-bear-
plastic wells was dose dependent, with K
D
≈91 629 ing cells important in the inflammatory response sug-
nM (Figure 1C). Specificity of binding was shown by gested the possible relevance of EN-RAGE–RAGE inter-
inhibition in the presence of excess soluble RAGE action in the pathogenesis of inflammatory lesions. We
(sRAGE), a truncated form of the receptor spanning the explored the effect of EN-RAGE on properties of RAGE-
extracellular domain, or anti-RAGE IgG. In contrast, un-
related proteins, bovine serum albumin, or nonimmune bearing cells in order to determine whether RAGE-
Cell
892
Figure 2. Ligation of RAGE by EN-RAGE and S100B Results in Activation of ECs (A–E), MPs (F and G), and PBMCs (H)
(A) Assessment of VCAM-1. HUVECs were incubated with the indicated mediators for 8 hr in the presence or absence of pretreatment with
nonimmune IgG, anti-RAGE IgG, or excess soluble RAGE. Cells were fixed, and cell surface ELISA for VCAM-1 was performed.
(B) Molt-4 adhesion assays. HUVECs were incubated with the indicated mediators for 8 hr. Varying concentrations (left panel) and incubation
times (middle panel) for EN-RAGE were employed. After incubation,
51
chromium-labeled Molt-4 cells were bound to the monolayer for 1 hr,
and cells were then disrupted in the presence of Triton X-100 (1%); the resulting material was counted in a beta counter. In the right panel,
HUVECs were treated with EN-RAGE in the presence or absence of pretreatment with the indicated F(ab9)
2
, excess sRAGE, or excess BSA.
Results are reported as fold increase above control (treatment with BSA). In (A) and (B), results are reported as mean 6SD.
(C) Assessment of ICAM-1. HUVECs were incubated with the indicated mediators in the presence or absence of pretreatment with nonimmune
IgG, anti-RAGE IgG, or excess soluble RAGE. Cell extract was obtained, and immunoblotting for ICAM-1 was performed. Mean 6SD of three
experiments is shown.
RAGE and Modulation of the Inflammatory Response
893
dependent responses underlying inflammation might to EN-RAGE (Figure 2D, lanes 4 and 3, respectively).
Supershift assays with anti-p50 and anti-p65 IgG dem-occur.
Endothelial Cells onstrated that the NF-kB complex activated upon liga-
tion of RAGE by EN-RAGE was composed of both p50Endothelium (HUVECs) incubated with EN-RAGE dis-
played induction of vascular cell adhesion molecule-1 and p65 (Figure 2D, lanes 11–13). EN-RAGE induction
of NF-kB nuclear translocation resulted from RAGE-(VCAM-1) (Figure 2A),a cell adhesion molecule thatteth-
ers mononuclear cells bearing VLA-4 to the vessel sur- mediated intracellular signaling, as shown by experi-
ments using a truncated form of the receptor, termedface (Li et al., 1993). The key role of RAGE in EN-RAGE-
mediated induction of endothelial VCAM-1 was shown dominant-negative RAGE (DN-RAGE), from which the
cytosolic tail was deleted. Endothelium transfected withby inhibition in the presence of anti-RAGE IgG or sRAGE
(Figure 2A). The functional significance of VCAM-1 ex- DN-RAGE displayed marked suppression of NF-kB acti-
vation (Figure 2D, lane 5) compared with those trans-pression by EN-RAGE-treated endothelium was shown
by enhanced binding ofVLA-4-bearing Molt-4 cells (Fig- fected with vector alone (Figure 2D, lane 6). Together,
these data suggested that EN-RAGE interaction withure 2B). Adherence of Molt-4 cells to EN-RAGE-treated
endothelium was dependent both on the concentration endothelial RAGEactivatedNF-kB, potentiallytriggering
expression of multiple gene products contributing to theand incubation time of EN-RAGE (left and middle panels,
respectively) and required interaction of EN-RAGE with inflammatory response, such as VCAM-1 and ICAM-1.
In view of the strong homology between membersthe receptor, as shown by the inhibitory effect of anti-
RAGE F(ab9)
2
and sRAGE (right panel). In addition to of the S100/calgranulin family, we sought to determine
whether another family member, human S100B, wouldmechanisms by which mononuclear cells might be tar-
geted to endothelium upon ligation of RAGE by EN- mediate RAGE-dependent endothelialactivation of NF-
kB. Using EMSA with
32
P-labeled NF-kB probe, nuclearRAGE, enhanced expression of intercellular adhesion
molecule-1 (ICAM-1) upon incubation of HUVECs with extracts from HUVECs incubated with S100B showed
an z3.2-fold increased intensity of the gel shift bandEN-RAGEwas notedin a RAGE-dependentmanner (Fig-
ure 2C), thereby providing a mechanism by which poly- (Figure 2E, lane 2) compared with cultures exposed to
BSA (Figure 2E, lane 1). That these findings were due tomorphonuclear leukocytesmight toobe attractedto EN-
RAGE-stimulated endothelium. Expression of VCAM-1 activation of RAGE was demonstrated by the inhibitory
effect of anti-RAGE IgG (Figure 2E, lane 3), overexpres-and ICAM-1 in response to inflammatory stimuli is sub-
ject, at least in part, to regulation at the transcriptional sion of DN-RAGE (Figure 2E, lane 5), and excess sRAGE
(datanot shown).Controlexperiments inwhich endothe-level by nuclear factor-kB (NF-kB) (Voraberger et al.,
1991; Neish et al., 1992). We previously demonstrated lial cultures were preincubated with nonimmune IgG
(Figure 2E, lane 4) or mock transfection was performedthat ligation of RAGE by AGEs and amyloid-bpeptide
enhanced nuclear translocation of NF-kB, as demon- (Figure 2E, lane 6) showed no effect on S100B-RAGE-
induced NF-kB activation. These observations empha-strated by electrophoretic mobility shift assay (EMSA)
(Yan et al., 1994, 1996; Lander et al., 1997). Nuclear size thelikelihood that a range of S100/calgranulin poly-
peptide ligands engage RAGE.extracts from endothelium exposed to EN-RAGE were
subjected to EMSA with a
32
P-labeled NF-kB probe and Mononuclear Phagocytes
Releaseof EN-RAGEat sitesof an ongoing inflammatoryshowed NF-kB activation (Figure 2D). Over the same
concentration range that induction of VCAM-1 expres- response could serve both to propagate and amplify the
cellularresponse byrecruiting mononuclearphagocytession and enhanced Molt-4 cell binding was observed,
endothelial cells exposed to EN-RAGE displayed an (MPs). In this regard, two issues were critical: EN-RAGE-
mediated induction of MP migration, and activation, fol-z5-fold increase in the intensity of the gel shift band
compared with cultures incubated with albumin (Figure lowing ligation of RAGE. Using modified chemotaxis
chambers, EN-RAGE placed in the lower compartment2D, lanes 1, 2, and 7, respectively; note densitometric
analysis of bands upon normalization for Sp1). That this stimulated migration ofRAGE-bearing peripheral blood-
derived human monocytes added to the upper compart-was largely mediated by EN-RAGE interaction with
RAGE was confirmed by the inhibitory effect of anti- ment in a dose-dependent manner (Figure 2F, left panel,
lines 2–4). In contrast, replacement of EN-RAGE withRAGE IgG or sRAGE added to the endothelium prior
(D and E) Electrophoretic mobility shift assay. HUVECs were treated with the indicated mediators for 8 hr. EN-RAGE (D) or S100B-treated (E)
cells were preincubated with anti-RAGE IgG; EN-RAGE was pretreated with excess sRAGE. Certain HUVECs were transiently transfected with
a construct encoding DN-RAGE or with vector control prior to treatment with EN-RAGE or S100B. Nuclear extract was prepared and EMSA
performed employing radiolabeled probes for NF-kB and Sp1. Supershift assays were performed by incubation of nuclear extract with the
indicated antibody prior to EMSA. Results of densitometric analysis after normalization for Sp1 are indicated; mean 6SD of three experiments
is shown.
(F) Modified chemotaxis assays. Mediators were placed in the upper or lower chamber, and human peripheral blood-derived monocytes were
placed in the upper chamber for 4 hr. Cells that had migrated through the membranes were stained and counted. Where indicated (right
panel), cells were pretreated with the indicated F(ab9)
2
fragments or EN-RAGE incubated with excess sRAGE prior to chemotaxis assay.
Mean 6SD is shown.
(G) Generation of IL-1band TNFa. BV-2 macrophages, either those transfected with DN-RAGE or mock-transfected cells, were incubated
with the indicated mediators for 8 hr. Supernatant was collected, and ELISA for IL-1bor TNFawas performed. Mean 6SD is shown.
(H) Generation of IL-2. PBMCs were incubated with the indicated mediators for 8 hr; supernatant was collected, and ELISA for IL-2 was
performed. Where indicated, cells were pretreated with the indicated IgG, or EN-RAGE was pretreated with excess sRAGE. Mean 6SD is
reported. In (G), results are reported as fold induction (incubation of cells with BSA alone).
Cell
894
Figure 3. EN-RAGE Mediates Cellular Activation and Inflammation In Vivo
(A) Expression of VCAM-1 in the lung. CF-1 mice were injected with EN-RAGE, BSA, or LPS in the presence or absence of RAGE blockade.
Twelve hours later, lungs were harvested, and immunoblotting for VCAM-1 was performed. Mean densitometric analysis 6SD of three
experiments is shown. (B–G) Injection of EN-RAGE into mouse footpad induces inflammation. (B–F) H&E analysis. Unilateral hind footpad was
injected with the indicated mediator; mice were treated with either sRAGE, MSA, or anti-RAGE/anti-EN-RAGE F(ab9)
2
and representative
H&E-stained sections shown. Bar, 75 mm. (G) Inflammation score. Twenty-four hours after injection, clinical and histologic score were
determined. In (G), score (maximal of 9; no inflammation 52) is defined as the sum of the clinical and histologic score. Clinical score: 1,
absence of inflammation; 2, slight rubor and edema; 3, moderate rubor and edema with skin wrinkles; 4, severe rubor and edema without
skin wrinkles; and 5, severe rubor and edema with toe spreading. Histologic score (H&E studies): 1, no leukocytic infiltration or subcutaneous
edema; 2, slight perivascular leukocytic infiltrate with slight subcutaneous edema; 3, severe leukocytic infiltrate without granulomata; and 4,
severe leukocytic infiltrate with granulomata. Mean 6SD of n 53/group is shown.
albumin was without effect on cell migration (Figure 2F, into cellular supernatants in a dose-dependent manner
(Figure 2G, right panel, filled bars).However, upon tran-line 1). EN-RAGE induction of cell migration was due
to true chemotaxis, as distortion of the chemotactic sient transfection with the RAGE-tail deletion construct,
BV-2 cell elaboration of TNFainto culture supernatantsgradient by addition of EN-RAGE to both upper and
lower compartments suppressed cell movement (Figure was significantly suppressed (Figure 2G, right panel,
hatched bars). In view of EN-RAGE-mediated induction2F, left panel, line 5). The central role of RAGE in EN-
RAGE-mediated cell migration was shown by the inhibi- of cytokine expression via RAGE in BV-2 cells, studies
wereperformed toanalyze NF-kBactivation underthesetory effect observed in the presence of sRAGE (Figure
2F, right panel, lines 2–3) and anti-RAGE F(ab9)
2
(Figure conditions.As inthe studiesdescribed abovewith endo-
thelial cells, EN-RAGEwas a strongagonist for induction2F, right panel, lines 5 and 6); in both settings, monocyte
migration was significantly suppressed compared with of nuclear translocation of NF-kB in mononuclear phago-
cytes; furthermore, blocking access of the ligand toincubation with either excess BSAor nonimmune F(ab9)
2
(Figure 2F, right panel, lines 1 and 4, respectively). At RAGE (with excess sRAGE or anti-RAGE F[ab9]
2
) or inhi-
bition of RAGE signaling (by transfection with the con-the siteof anongoing inflammatoryresponse, proinflam-
matory cytokines, such as IL-1band TNFa, have been struct encoding DN-RAGE) suppressed NF-kB activa-
tion (data not shown).shown to have a central role. Incubation of mock-trans-
fected (vector alone) cultured murine macrophage-like PBMCs
The emerging pattern of EN-RAGE-dependent cellularBV2 cells with EN-RAGE caused elaboration of IL-1b
into cellular supernatants in a dose-dependent manner activation, occurring via RAGE, was next explored on
PBMCs. PBMCs exposed to EN-RAGE displayed en-(Figure 2G, leftpanel, filled bars).That intact intracellular
signalingpathways triggeredbyRAGE wereessential for hanced elaboration of IL-2 into culture supernatant in a
RAGE-dependent manner (Figure 2H). Consistent withIL-1bexpression was demonstrated by the suppression
observed following transfection of a construct encoding these findings, an enhanced mitogenic response to
cross-linkingCD3/CD28 afterstimulation ofPBMCs withDN-RAGE (Figure 2G, left panel, hatched bars). Similar
results were observed when BV-2 cells were analyzed EN-RAGE wasnoted; compared with pretreatment with
albumin (BSA), significant uptake of [
3
H]thymidine wasfor the effect of EN-RAGE on expression of TNFa; EN-
RAGE–RAGE interaction increased elaboration of TNFaobserved in cells preincubated with EN-RAGE, 5 mg/ml
RAGE and Modulation of the Inflammatory Response
895
(39,285 62,323 [SD] versus 67,242 61,727 counts per (Figures 4A, 4B, and 4F). Injection of vehicle alone, or
administration of mBSA without prior sensitization didminute [cpm], respectively, p ,0.00001). That this was
largely due to interaction of EN-RAGE with RAGE ex- not elicit an inflammatory response (data not shown).
The contribution of EN-RAGE interaction with RAGEpressed by PBMCs was shown by the inhibitory effect
of anti-RAGE IgG (30 mg/ml) orexcess sRAGE (40-fold); to the pathogenesis of inflammation in this model was
tested by treating mice, prepared by presensitizationsuppression of uptake of [
3
H]thymidine was noted
(42,413 62,829 and 39,998 63,603 cpm, respectively; and challenge with mBSA, with intraperitoneally admin-
istered sRAGE or antibodies to RAGE and/or EN-RAGE.p ,0.00001 in both cases). In contrast, treatment with
nonimmune F(ab9)
2
was without effect (63,642 67,062 Control studies in which the intraperitoneal injection
contained vehicle (murine serum albumin [MSA]) dem-cpm; p .0.05).
Taken together, these studies in cell culture model onstrated a strong inflammatory response (score, 9.0 6
0.4) (Figure 4A, line 1; and Figure 4B) comparable tosystems demonstrated that engagement of RAGE on
thesurface ofcells criticaltopathogenesis ofthe inflam- that observed without an intraperitoneal injection (data
not shown). Histologic analysis of the affected footpadmatory response resulted in acquisition of a phenotype
propagating host effector mechanisms, including cell demonstrated a striking influxof inflammatorycells, with
granulomataand significantedema(Figure 4F).Adminis-migration, proliferation, and generation of cytokines. tration of murine sRAGE by intraperitoneal injection re-
sulted in dose-dependent suppression of inflammation;
Infusion of EN-RAGE Stimulates Cellular Activation at an sRAGE concentration of 100 mg/dose, the inflam-
As RAGE bound EN-RAGE and another S100 protein mation score was reduced to 2.7 60.3, (p ,0.001
(S100B), resulting in changes in activation of immune/ compared withMSA), virtuallyto backgroundlevels (Fig-
inflammatory effector cells in vitro, it was essential to ure 4A, lines 2–6, and Figure 4C). Consistent with these
extrapolate our observations to animal models. As a first results, H&E staining revealed striking abrogation of in-
step in evaluating the in vivo relevance of our observa- flammatory cell influx in animals treated with sRAGE
tions, RAGE-dependent cellular activation was analyzed (100 mg; Figure 4H).
following intravenous infusion. Systemic administration Because it was likely that sRAGE interrupted the EN-
of EN-RAGE (30 mg) to immunocompetent CF-1 mice RAGE/RAGE axis by binding up EN-RAGE-like species
resulted in an z2.4 increase in expression of VCAM-1 and preventing their interaction with cell surface RAGE,
antigen in the lung compared with infusion of albumin experiments were next performed with specific antibod-
(Figure 3A, lanes 2 and 1, respectively). That this was ies to block either EN-RAGE or RAGE. When mice sensi-
largely due to engagement of RAGE was demonstrated tized/challenged with mBSA were treated intraperitone-
by suppression of EN-RAGE-stimulated VCAM-1 ex- ally with nonimmune F(ab9)
2
, no effect was noted on
pression in the lung when animals also received sRAGE the observed inflammatory response (score, 9.0 60.2;
(Figure 3A, lane 3) or anti-RAGE IgG (Figure 3A, lane 4). Figure 4A, line 7). However, in the presence of anti-
Infusion of nonimmune IgG was without effect (Figure RAGE F(ab9)
2
(200 mg) or anti-EN-RAGE F(ab9)
2
(200 mg)
3A, lane 5). significant attenuation of the inflammatory response
Consistent with in vitro findings suggesting that EN- was evident (Figure 4A, lines 9 and 11, respectively,
RAGE might modulate inflammatory processes, local and Figures 4E and 4D, respectively), with inflammation
injection of EN-RAGE into murine footpad resulted in scores of 4.9 60.8 and 5.6 60.5, respectively (p ,0.05
influx of inflammatory cells (Figure 3C). That these were in both cases compared with treatment with nonimmune
largely due to interaction with cellular RAGE was evident F[ab9]
2
). Consistent with these results, a striking reduc-
upon intraperitoneal administration of either sRAGE or tion in the accumulation of inflammatory cells and
anti-EN-RAGE/anti-RAGE F(ab9)
2
(Figures 3D, 3F, and edema, as well as absence of granulomata, was ob-
3G, lines 2 and 4, respectively). In contrast, murine se- served in the presence of either anti-RAGE F(ab9)
2
or
rum albumin or nonimmune F(ab9)
2
was without effect anti-EN-RAGE F(ab9)
2
(Figures 4J and 4I, respectively).
(Figures 3E and 3G, lines 1 and 3, respectively). These In support of a central role for the EN-RAGE/RAGE axis
data suggested that EN-RAGE engagement of RAGE in pathogenesis of the inflammatory response, markedly
occurred in vivo. We next tested the pathophysiologic attenuated inflammation was observed when both anti-
significance of this interaction in models of inflam- EN-RAGE and anti-RAGE F(ab9)
2
were administered si-
mation. multaneously (Figure 4A, line 12); inflammation score
was reduced to 3.6 60.9 (p ,0.01, compared with
treatment with nonimmune F[ab9]
2
), with decreased
Role of EN-RAGE and RAGE in the numbers of inflammatory cells and edema (Figure 4K).
Inflammatory Response These data suggested that blockade of EN-RAGE in-
Acute Inflammation teraction with RAGE substantially quenched cellular ac-
Experimentally induced delayed-type hypersensitivity tivation in a model of delayed-type hypersensitivity. In
(Dunn et al., 1993) provided an ideal setting for evaluat- parallel with suppressed inflammation in mBSA-sensi-
ing the contribution of EN-RAGE interaction with RAGE tized/challenged mice treated with either sRAGE, anti-
to an important immune/inflammatory response. CF-1 RAGE F(ab9)
2
, or anti-RAGE/anti-EN-RAGE F(ab9)
2
, sup-
mice were sensitized with methylated BSA (mBSA; not pression of NF-kB activation was observed in nuclear
a ligand of RAGE). Twenty-one days later, mBSA or extracts prepared from the affected footpads. Com-
vehicle (PBS) was injected into the left hind footpad. pared with the contralateral footpad (the latter prepared
Micetreated inthis manner displayeda stronginflamma-
tory response as measured by the inflammation score bysensitization withmBSA, but without local challenge),
Cell
896
Figure 4. Blockade of EN-RAGE/RAGE Suppresses Inflammation in a Model of Delayed-Type Hypersensitivity
(A) Inflammation score. CF-1 mice were sensitized and challenged with mBSA. Mice were pretreated by intraperitoneal injection with sRAGE,
MSA, immune or nonimmune F(ab9)
2
fragments prior to and after local challenge with mBSA. Twenty-four hours after injection of footpad with
mBSA, clinical and histologic score of footpad were performed employing the criteria in Figure 3.
(B–E) Representative mice sensitized/challenged with mBSA are shown: (B), treatment with MSA; (C), treatment with sRAGE, 100 mg IP per
dose; (D), treatment with anti-EN-RAGE F(ab9)
2
, 200 mg IP per dose; and (E), treatment with anti-RAGE F(ab9)
2
, 200 mg IP per dose.
(F–K) H&E analysis of representative footpads from mice sensitized/challenged with mBSA is shown: (F), treatment with MSA; (G), contralateral
footpad, no DTH; (H), treatment with sRAGE, 100 mg IP per dose; (I), anti-EN-RAGE F(ab9)
2
, 200 mg IP per dose; (J), anti-RAGE F(ab9)
2
, 200
mg IP per dose; and (K), anti-EN-RAGE 1anti-RAGE F(ab9)
2
, 200 mg IP per dose. Bar, 150 mm.
(L) EMSA. Nuclear extracts were prepared from pooled hind footpads (n 53/condition), and EMSA was performed for NF-kB normalized to
Sp1. Mean 6SD of three experiments is shown.
(M) Measurement of TNFa. Extract of indicated footpad was prepared, and ELISA for murine TNFawas performed. Mean 6SD is reported.
nuclear extracts from the mBSA-injected footpad re- prepared/challenged with mBSA but treated only with
vehicle (MSA; Figure 4L, lanes 4 and 2, respectively).vealedan z6-foldincrease inactivation ofNF-kB (Figure
4L, lanes 1 and 2, respectively). In the presence of Consistent with the protective effect of interrupting EN-
RAGE interaction with RAGE, administration of anti-sRAGE (100 mg/dose; IP), the intensity of the gel shift
band was substantially reduced as contrasted with re- RAGE/anti-EN-RAGE F(ab9)
2
to mBSA-sensitized/chal-
lenged mice caused an z70% decrease in activation ofsults using nuclear extracts from footpads appropriately
RAGE and Modulation of the Inflammatory Response
897
Table 2. Histologic Examination of Rectosigmoid Tissue Retrieved from IL-10 Null Mice
Mouse No. Condition Cryptitis Cellular Infiltrate Comments
1 MSA Absent M Two clusters of mononuclear cells identified
2 MSA Present (base) L,P,E.M Patchy submucosal inflammation
3 MSA Absent L,P,E,M Patchy submucosal inflammation
4 MSA Present (base) L,P,E,M Focal submucosal inflammation
5 MSA Absent None No inflammation
6 sRAGE Absent None No inflammation
7 sRAGE Present (base) L,P,E,M Patchy submucosal inflammation
8 sRAGE Absent None No inflammation
9 sRAGE Absent None No inflammation
10 sRAGE Absent None No inflammation
Rectosigmoid tissue was retrieved from IL-10 null mice treated with either murine serum albumin (MSA) or sRAGE as indicated.Sections were
prepared as described and stained with H&E; histologic scoring was performed by a blinded investigator. Inflammatory cells are identified
as follows: M, monocyte/macrophage; L, lymphocyte; E, eosinophil; and P, plasma cell.
NF-kB (Figure 4L, lane 6) versus that seen in the pres- Indeed, assessment of serum levels of MRP8/14 (my-
ence of nonimmune F(ab9)
2
(Figure 4L, lane 7). These eloid-related protein), S100-like molecules, has been
dataindicate thatblocking bindingof EN-RAGEto RAGE suggestedto trackdisease activity inpatients withulcer-
potently quenched signaling via the NF-kB pathway. ative colitis,a chronic inflammatory disease ofthe bowel
An important consequence of RAGE ligation by EN- (Lugering et al., 1995). S100/calgranulin molecules have
RAGEwas increasedexpression of inflammatory media- in common structural features, most characteristic of
tors, likely due, at least in part, to NF-kB. Thus, expres- which are calcium-binding EF-hand domains. Based on
sion of the proinflammatory cytokine TNFawas compared these properties, a range of possible intracellular func-
in footpad protein extracted from mBSA-sensitized/ tions for these polypeptides has been postulated, such
challenged mice. Compared with mice receiving vehicle as alteration of the cytoskeleton and cell shape, signal
(MSA), animals treated with sRAGE (100 mg) displayed transduction, and, via modulation of intracellular cal-
an z3.1-fold decrease in levels of TNFaprotein in foot- cium, regulation of chemotaxis, phagocytosis, and gen-
pad tissue (Figure 4M). Similar beneficial effects were eration of reactive oxygen species (ROIs) (Schafer and
observed upon treatment with anti-EN-RAGE/anti-RAGE Heinzmann, 1996).
F(ab9)
2
(Figure 4M). Despite the fact that EN-RAGE and related family
Chronic Inflammation
A hallmark of S100/calgranulin polypeptides is their as-
sociation with chronic inflammation, such as human in-
flammatory bowel diseases (Lugering et al., 1995). To
determine whether S100/calgranulins were contributing
to pathogenesis of the inflammatory lesions via their
interaction with RAGE, a murine model of colitis, IL-10
null mice (Kuehn et al., 1993), was employed and RAGE
blockade was achieved by administration of sRAGE. IL-
10 null mice were treated with either MSA or sRAGE
(100 mg per day; IP) for 6 weeks, and, at the end of that
time, rectosigmoid colon was assessed for evidence of
inflammation. Although 4/5 mice receiving MSA dis-
played submucosal colonic inflammatory infiltrates com-
posed of lymphocytes, macrophages, eosinophils, and
plasmacells, 4/5mice treated with sRAGE showedvirtu-
ally no inflammation (Table 2). Consistent with these
histologic results, nuclear extracts prepared from co-
lonic tissue of IL-10 null treated with MSA and subjected
toEMSA withthe
32
P-NF-kBprobe displayeda strong gel
shift band in 5/6 mice (Figure 5A, lanes 7–12). Following
treatment with sRAGE, virtually complete suppression
of the gel shift band was observed in 3/6 animals (Figure
5A, lanes 3–4 and 6), and, taking all animals into account
Figure 5. Blockade of EN-RAGE/RAGE Suppresses Chronic Co-
(Figure 5A, lanes 1–12), the presence of sRAGE reduced
lonic Inflammation in IL-10 Null Mice
the intensity of the bands by z75% (p 50.04). Parallelling
(A) EMSA. Nuclear extracts were prepared from rectosigmoid colon
tissue of mice treated with either sRAGE (lanes 1–6) or MSA (lanes
these results, an z8.7-fold decrease in plasma levels of
7–12). Densitometric analysis after normalization for Sp1 was per-
TNFawas observedin micetreated with sRAGEcompared
formed. Mean densitometry pixel units for MSA-treated (n 56) ver-
with those receiving MSA (p 50.002; Figure 5B).
sus sRAGE-treated mice (n 56) were 7121.8 65359.6 versus
1,911 61155 U; p 50.04.
Discussion
(B) Assessment of plasma TNFa. Plasma was retrieved from IL-10
null mice upon sacrifice, and ELISA for TNFawas performed. Mean
The presence of S100/calgranulin polypeptides at sites
values for MSA-treated (n 56) versus sRAGE-treated mice (n 56)
were 190.5 689.0 versus 21.9 663.6 pg/ml, respectively; p 50.002.
of acute and chronic inflammation has long been noted.
Cell
898
Experimental Procedures
members lack signal peptides, there is definitive evi-
dence that these polypeptides readily gain access to
Protein Sequence Analysis
the extracellular space (Rammes et al., 1997) and medi-
To perform sequence analysis of RAGE-binding proteins from bo-
ate an array of inflammatory phenomena. For example,
vine lung extract, Coomassie blue visible bands (z12 kDa) were
following infusion or injection of CP-10 (chemotactic
eluted from SDS–PAGE gels. Automated Edman degradation was
peptide-10; a member of the S100 family) into mice,
carried out using an HP-G1005A sequencer (Hewlett Packard Ana-
lytical Instruments, Palo Alto, CA). Internal sequencing was per-
analysis of elicited peritoneal macrophages revealed in-
formed employing endoproteinase Lys-C (Boehringer Mannheim,
creased expression of the scavenger receptor and
Indianapolis, IN) digestion followed by microbore HPLC (Hori et al.,
TNFa,enhanced loadingof acetylatedLDL andfoam cell
1995).
formation, increased phagocytosis, and local footpad
inflammation (Geczy, 1996). The data presented here
Molecular Cloning
suggest a novel pathway through which S100/calgran-
Molecular cloning was performed using bovine and human lung
ulin molecules released into the extracellular space con-
libraries (Clontech, Palo Alto, CA) in order to obtain cDNAs for EN-
RAGE. The sequence encoding bovine cDNA for EN-RAGE is No.
tribute to the pathogenesis of the inflammatory re-
AF011757 (GenBank).
sponse via interaction with RAGE-bearing cells. The
binding of EN-RAGE and other EN-RAGE-like molecules
Protein Expression
(S100/calgranulin family members) to RAGE on critical
The cDNA encoding bovine EN-RAGE was placed into a baculovirus
cellular targets activates signaling pathways, thereby
expression system and expressed in Spodoptera frugiperda 9 (Sf9)
modulating gene expression and amplifying inflamma-
cells (Invitrogen, Carlsbad, CA). EN-RAGE was purified from cellular
tory effector mechanisms. Since both S100/calgranulins
pellets by sequential chromatography onto heparin and hydroxylap-
and RAGE are associated with chronic inflammation,
atite columns (Amersham Pharmacia, Piscataway, NJ) and eluted
with increasing concentrations of NaCl. Purified EN-RAGE, a single
this suggests possible roles for EN-RAGEs, via engage-
band on Coomassie-stained gels (M
r
≈12 kDa) was devoid of endo-
ment of RAGE, in sustaining the inflammatory response
toxin prior to experiments by chromatography onto Detox-igel col-
in a range of disorders beyond traditional inflammatory
umns (Pierce, Arlington Heights, IL) as documented using a kit from
conditions, such as inflammatory bowel disease, to ath-
Sigma (St. Louis, MO) (limulus amebocyte assay). Purified S100B
erosclerosis (Ross, 1999), Alzheimer’s disease (Marshak
from human brain was obtained from Calbiochem-Novabiochem
et al., 1992), and other situations in which an inflamma-
Corp. (San Diego, CA). Recombinant murine soluble RAGE was pre-
pared as described (Park et al., 1998).
tory component has gained increasing recognition. Al-
though it is not yet clear whether RAGE will bind all
Immunoblotting
S100/calgranulins, as EN-RAGE most closely resembles
In Vitro Studies
calgranulins and S100B is an S100 protein, it is likely
Human peripheral blood-derived mononuclear cells were isolated
that the receptor will interact with other family members.
from normal volunteers by using Histopaque 1077 (Sigma), and Jur-
Indeed, our studies indicate that anti-EN-RAGE IgG rec-
kat E6 cells were obtained from the American Type Tissue Corpora-
ognizes both bovine and human S100B in immunoblot-
tion (Rockville, MD). Where indicated, PBMCs or Jurkat E6 cells
ting studies (data not shown). Further, although our data
were stimulated by cross-linking CD3/CD28 (Gimmi et al., 1991).
Briefly, cells were incubated with purified mouse anti-human CD3
do not rule out a role for RAGE-independent mecha-
IgG (1 mg/ml) (PharMingen, San Diego, CA) for 24 hr followed by
nisms of S100/calgranulin-mediated cellular perturba-
washing in PBS, and then incubated with 1 mg/ml purified mouse
tion, the inhibitory effect of anti-RAGE F(ab9)
2
and
anti-human CD28 IgG (PharMingen) for an additional 24 hr. Superna-
sRAGE on the inflammatory response emphasizes an
tant was retrieved and assayed for levels of IL-2 by ELISA (R&D
important role for RAGE.
Systems, Minneapolis, MN). HUVECs were stimulated with TNFa
The present findings expand the biologic contexts
(10 ng/ml) (Genzyme, Cambridge, MA) for 12 hr. In all cases, after
of RAGE as a cell surface immunoglobulin superfamily
stimulation, cells (1 310
7
) were sonicated (Sonifer 250, Branson,
Danbury, CT) in PBS containing protease inhibitor mixture (Boeh-
memberwith adistinct panelofligands whosefunctional
ringer Mannheim) and centrifuged for 30 min at 14,000 rpm at 48C.
implications extend to both development and patho-
Protein concentration was measured using Bio-Rad protein assay
physiologically relevant states (Schmidt et al., 1998).
(Hercules, CA). To each lane of Tris-glycine gels (Novex, San Diego,
Indeed, assignment of RAGE to the major histocompati-
CA), 7 mg of protein was added; gel components were transferred to
bility complex on chromosome 6 (Sugaya et al., 1994)
nitrocellulose membranes (Bio-Rad) and immunoblotting performed
suggested a possible relationship of this receptor to
using polyclonal rabbit monospecific anti-EN-RAGE IgG prepared
against full-length recombinant bovine EN-RAGE. Goat anti-rabbit
immune/inflammatory responses. The presence of func-
IgG labeled with horseradish peroxidase (Sigma) and ECL system
tional NF-kB sites in the RAGE promoter, as well as
(Amersham-Pharmacia) were employed to indicate sites of primary
putative NF-IL-6, g-IRE, and AP2-binding sites further
antibody binding.
reinforced the likelihood that RAGE would contribute to
In Vivo Studies
the pathogenesis of inflammation (Li and Schmidt, 1997;
LPS (Sigma) (30 mg/g body weight) was injected intraperitoneally
Li et al., 1998). Our present findings close this loop by
into CF-1 mice. Mice were sacrificed at the indicated time points
demonstrating that RAGE engagement of S100/calgran-
and serum (0.015 ml) subjected to electrophoresis using Tris-glycine
gels (14%; Novex) and immunoblotting for detection of EN-RAGE
ulins (EN-RAGEs) at inflammatory loci propagates the
performed as above. Densitometry was performed using Im-
host response by perpetuating recruitment and activa-
ageQuant, Molecular Dynamics (Foster City, CA).
tion of cellular effectors. By placing RAGE at the center
of chronic inflammation, a novel paradigm can be pro-
Radioligand Binding Assays
posed and tested concerning cellular effects of S100/
Purified EN-RAGE was radiolabeled using
125
I and Iodobeads
calgranulins and their contribution to the pathogenesis
(Pierce) to a specific activity of z5000 cpm/ng. Radioligand binding
of diverse disorders, from inflammatory bowel disease
assays were performed in 96-well dishes to which had been ad-
and delayed-type hypersensitivity, to Alzheimer’s dis-
sorbed purified murine recombinant soluble RAGE (5 mg/well). Ra-
dioligand binding assays were performed in the presence of the
ease, atherosclerosis, and the complications of diabetes.
RAGE and Modulation of the Inflammatory Response
899
indicated concentration of radiolabeled EN-RAGE 6a 50-fold molar (500 mg). Twelve hours later, lungs were harvested, and extract
was subjected to electrophoresis/immunoblotting employing anti-excess of unlabeled EN-RAGE, amphoterin, amyloid-b-peptide
1–40, or AGE albumin in PBS containing calcium/magnesium and VCAM-1 IgG (Santa Cruz). In other studies, EN-RAGE (10 mg) or
vehicle PBS was injected into hind footpad. Certain animals wereBSA (0.2%) for 3 hr at 378C. Wells were washed rapidly, and elution
of bound material was performed in the presence of heparin (1 mg/ pretreated with sRAGE, MSA, or the indicated F(ab9)
2
fragments 12
hr prior injection, at the time of injection, and 12 hr after injection.ml). Eluate was counted in a gamma counter (LKB, Gaithersburg,
MD). Equilibrium binding data were analyzed as described (Klotz Mice were sacrificed 24 hr after injection; clinical and histologic
scores of H&E-fixed tissue were performed by two blinded investi-and Hunston, 1984):B 5nKA/1 1KA, where B 5specifically bound
ligand (total binding, wells incubated with tracer alone, minus non- gators.
specific binding, wells incubated with tracer in the presence of
excess unlabeled material), n 5sites/cell, K 5the dissociation Model of Delayed Hypersensitivity
constant, and A 5free ligand concentration) using nonlinear least- FemaleCF-1 mice, 6 weeksof age, were sensitizedby subcutaneous
squares analysis (Prism; San Diego, CA). Where indicated, pretreat- injection over the left inguinal lymph node of an emulsion (0.1 ml)
ment with either antibodies or soluble RAGE (2 hr) was performed. containing methylated BSA (mBSA; 25 mg/ml; Sigma), NaCl (0.9%),
dextran(5–40 310
6
MW;50 mg/ml; Sigma),andFreund’s incomplete
adjuvant (50%; ICN Biomedical; Aurora, OH). Three weeks later, the
Cellular Activation Studies left plantar hind paw was injected subcutaneously with mBSA (0.4
Endothelial Cells mg/ml; 0.050 ml). Where indicated, mice were pretreated by intra-
HUVECs were isolated and characterized as described (Schmidt et peritoneal injection with sRAGE, mouse serum albumin, immune or
al., 1995) and cultured in serum-free RPMI 1640 without endothelial nonimmune F(ab9)
2
fragments, 24 and 12 hr prior to, and 6 and 12
cell growth factor for 24 hr and then stimulated with the indicated hr after local challenge with mBSA. Twenty-four hours after injection
concentrations of EN-RAGE or other stimuli. Certain cells were pre- of footpad with mBSA, clinical score of footpad was performed by
treated with rabbit anti-human RAGE IgG or nonimmune rabbit IgG two blinded investigators; mice were humanely sacrificed and feet
for 2 hr, or EN-RAGE was pretreated with sRAGE for 2 hr prior to fixed in formalin (10%) or frozen. Histologic score was performed
stimulation with EN-RAGE. After 8 hr, cells were fixed with parafor- on sections of footpad stained with hematoxylin and eosin (Sigma)
maldehyde (2%) for 30 min, washed twice with PBS, and treated by two blinded investigators. Electrophoretic mobility shift assay
with PBS containing nonfat dry milk (5%) and BSA (2.5%) to block was performed employing 10 mg footpad nuclear extract added per
nonspecific binding sites on the cell surface. Cell surface ELISA lane and normalized to Sp1. Assessment of tissue extract prepared
employing anti-VCAM-1 IgG (4 mg/ml; Santa Cruz Biotechnologies, as above for levels of TNFawas performed.
Santa Cruz, CA) was performed and VCAM-1 activity determined
using
51
Cr-labeled Molt-4 cells (ATCC) (Schmidt et al., 1995). Levels
of ICAM-1 in EN-RAGE-treated HUVECs were determined by immu- Model of Chronic Colitis in IL-10 Null Mice
noblotting with anti-ICAM-1 IgG (Santa Cruz; 0.2 mg/ml) after 12 hr Four-week-old IL-10 null mice (C57BL/6 background) (Jackson Lab-
incubation. EMSA was performed with 10 mg nuclear extract loaded oratories, Bar Harbor, ME) were treated once daily by intraperitoneal
onto PAGE gels in the presence of
32
P-labeled probe for NF-kB from injection for 6 weeks with either MSA (100 mg/day) or sRAGE (100
the VCAM-1 promoter (Neish et al., 1992) or radiolabeled probe for mg/day). Mice were deeply anesthetized; plasma was removed, and
Sp1 (Santa Cruz). Supershift assays were performed by preincubat- rectosigmoid colon was retrieved for histologic analysis or prepara-
ing nonimmune anti-p50, anti-p65, or both IgG (2 mg/ml) (Santa Cruz) tion of nuclear extracts. Hematoxylin and eosin–stained sections of
with nuclear extract for 45 min prior to addition of radiolabeled rectosigmoid were evaluated by a blinded investigator. Plasma was
oligonucleotide probe. assessed for levels of TNFa(R&D Systems), and EMSA for NF-kB
PBMCs and MPs was performed on nuclear extracts.
Chemotaxis Assays. Chemotaxis assays were performed in 48-well
microchemotaxis chambers (Neuro-Probe, Bethesda, MD) con- Statistical Analysis
taining a polycarbonate membrane (8 mm; Nucleopore, Pleasanton, Statistical comparisons among groups was determined using one-
CA) (Miyata et al., 1996). The lower chamber contained the chemo- way analysis of variance (ANOVA); where indicated, individual com-
tactic stimulus. N-formyl-met-leu-phe (Sigma) was employed as parisons were performed using Student’s t test. Statistical signifi-
positive control. Human MPs were added to the upper chamber (10
4
cance was ascribed to the data when p ,0.05.
cells/well).After 4 hrat 378C, nonmigrating cellson the uppersurface
of the membrane were gently scraped and removed, the membrane Acknowledgments
was fixed in methanol (100%), and cells that had migrated through
the membrane were stained with Giemsa (Sigma) and counted in This work was supported, in part, by the Surgical Research Fund
nine high-powered fields. of the College of Physicians and Surgeons, ColumbiaUniversity, and
Mitogenic Assays. PBMCs were isolated from whole blood and by grants from the United States Public Health Service (DK52495,
suspended in RPMI containing FBS (10%) (1 310
6
cells/ml). Cells HL56881, HL60901, AG00602, and DE11561), Juvenile Diabetes
were seeded in 96-well tissue culture wells and treated with EN- Foundation International, American Heart Association, New York
RAGE for 12 hr prior to stimulation by cross-linking CD3/CD28 as affiliate, and theCouncilfor Tobacco Research. M.A. H. is arecipient
above for 12 hr. Wells were pulsed with [
3
H]thymidine (1 mCi/well) of a fellowship award from Deutsche Forschungsgemeinschaft
(New England Nuclear, Boston, MA) and incubated for an additional (DFG).
24 hr prior to harvesting and processing for liquid scintillation count-
ing using an LK betaplate (Wallac, Inc., Gaithersburg, MD). Received March 18, 1999; revised May 11, 1999.
Assessment of Cytokines. RAGE-bearing BV2 cells (Yan et al.,
1997) or PBMCs were incubated with the indicated mediators; cer- References
tain cells were preincubated with anti-RAGE F(ab9)
2
prior to stimula-
tion with EN-RAGE. In other cases, EN-RAGE was preincubated Brownlee, M., Cerami, A., and Vlassara, H. (1988). Advanced glyco-
with sRAGE prior to stimulation. ELISA for TNFa, IL-1b, or IL-2 was sylation endproducts in tissue and the biochemical basis of diabetic
performed (R&D Systems). Where indicated, cells were transfected complications. N. Engl. J. Med. 318, 1315–1320.
with a construct encoding human RAGE in which the cytosolic do-
main (tail) was deleted employing superfect (Qiagen, Valencia, CA) Dell’Angelica, E.C., Schleicher, C.H., and Santome, J.A. (1994). Pri-
(1 mg DNA/ml); pcDNA3 (Invitrogen) was employed as vector. Stimu- marystructure and bindingpropertiesof calgranulin C,anovel S100-
lation experiments were performed 48 hr after transfection. like calcium-binding protein from pig granulocytes. J. Biol. Chem.
269, 28929–28936.
Dunn, C.J., Galinet, L.A., Wu, H., Nugent, R.A., Schlachter, S.T.,Infusion Studies and Local EN-RAGE-Mediated Inflammation
Male CF-1 mice (Charles River), approximately 6 weeks of age, were Staite, N.D., Aspar, D.G., Elliott, G.A., Essani, N.A., Rohloff, N.A.,
and Smith, R.J. (1993). Demonstration of novel anti-arthritic andinjected via the tail vein with EN-RAGE (30 mg), BSA (30 mg), or LPS
Cell
900
anti-inflammatory effects of Diphosphonates. J. Pharmacol. Exp. human mononuclear phagocytes via an oxidant-sensitive pathway:
implications for the pathogenesis of dialysis-related amyloidosis. J.Ther. 266, 1691–1698. Clin. Invest. 98, 1088–1094.
Geczy, C. (1996). Regulation and proinflammatory properties of che-
motactic protein, CP-10. Biochim. Biophys. Acta 1313, 246–252. Neeper, M., Schmidt, A.M., Brett, J., Yan, S.D., Wang, F., Pan, Y.C.,
Elliston, K., Stern, D., and Shaw, A. (1992). Cloning and expression
Gimmi, C.D., Freeman, G.J., Gribben, J.G., Sugita, K., Freedman, of RAGE: a cell surface receptor for advanced glycosylation end
A.S., Morimoto, C., Nadler, L.M (1991). B-cell surface antigen B7 products of proteins. J. Biol. Chem. 267, 14998–15004.
provides a costimulatory signal that induces T cells to proliferate
and secrete interleukin-2. Proc. Natl. Acad. Sci. USA 88, 6575–6579. Neish, A.S., Williams, A.J., Palmer, H.J., Whitley, M.Z., and Collins,
T. (1992). Functional analysis of the human vascular cell adhesion
Gottsch, J.D., Stark, W.J., and Liu, S.H. (1997). Cloning and se- molecule-1 promoter. J. Exp. Med. 176, 1583–1593.
quenceanalysis of human andbovinecornealantigen (C0-Ag) cDNA:
identification ofhost-parasite protein calgranulin C. Trans.Am. Op- Park, L., Raman, K.G., Lee, K.J., Yan, L., Ferran, L.J., Chow, W.S.,
Stern, D., and Schmidt, A.M. (1998). Suppression of acceleratedthalmol. Soc. 14, 111–129. diabetic atherosclerosis by soluble receptor for AGE (sRAGE). Nat.
Hitomi, H., Yamaguchi, K., Kikuchi, Y., Kimura, T., Maruyama, K., Med. 4, 1025–1031.
and Nagasaki, K. (1996). A novel calcium-binding protein in amniotic
fluid, CAAF1: its molecular cloning and tissue distribution. J. Cell Rammes, A., Roth, J., Goebeler, M., Klempt, M., Hartmann, M.,
and Sorg, C. (1997). Myeloid-related protein (MRP) 8 and MRP14,Sci. 109, 805–815. calcium-binding proteins of the S100 family, are secreted by acti-
Hori, O., Brett, J., Slattery, T., Cao, R., Zhang, J., Chen, J., Naga- vated monocytes via a novel, tubulin-dependent pathway. J. Biol.
shima,M., Nitecki, D.,Morser,J., Stern, D.,and Schmidt, A.M. (1995). Chem. 272, 9496–9502.
The receptor for advanced glycation endproducts (RAGE) is a cellu-
lar binding site for amphoterin: mediation of neurite outgrowth and Rauvala, H., and Pihlaskari, R. (1987). Isolation and some character-
istics of an adhesive factor of brain that enhances neurite outgrowthcoexpression of RAGE and amphoterin in the developing nervous
system. J. Biol. Chem. 270, 25752–25761. in central neurons. J. Biol. Chem. 262, 16625–16635.
Ross, R. (1999). Atherosclerosis—an inflammatory disease. N. Engl.Ilg, E.C., Troxler, H., Buergisser, D.M., Kuster, T., Markert, M., Guig-
nard, F., Hunziker, P., Birchler, N., and Heinzmann, C.W. (1996). J. Med. 340, 115–126.
Amino acid sequence determination of human S100A12 (p6, Schafer, B.W., and Heinzmann, C.W. (1996). The S100 family of EF-
Calgranulin C, CGRP, CAAF1) by tandem mass spectrometry. Bio- hand calcium-binding proteins: functions and pathology. TIBS 21,
chem. Biophys. Res. Comm. 225, 146–150. 134–140.
King,G., and Brownlee, M.(1996). The cellular andmolecularmecha- Schmidt, A.M., Vianna, M., Gerlach, M., Brett, J., Ryan, J., Kao, J.,
nisms of diabetic complications. Endocrin. Metab. Clin. North Am. Esposito, C., Hegarty, H., Hurley, W., Clauss, M., et al. (1992). Isola-
25, 255–270. tion and characterization of binding proteins for advanced glycosyl-
ation endproductsfrom lung tissue which arepresent on the endo-Klotz, I., and Hunston, D. (1984). Mathematical models for ligand-
receptor binding. J. Biol. Chem. 259, 10060–10062. thelial cell surface. J. Biol. Chem. 267, 14987–14997.
Schmidt, A-M., Hori, O., Chen, J., Li, J., Crandall, J., Zhang, J., Cao,Krieger, M., and Herz,J. (1994).Structure andfunction ofmultiligand
lipoprotein receptors: macrophage scavengers and LDL receptor- R., Yan, S.D., Brett, J., and Stern, D. (1995). Advanced glycation
endproducts interacting with their endothelial receptor induce ex-related protein. Annu. Rev. Biochem. 63, 601–637. pression of vascular cell adhesion molecule-1 (VCAM-1): a potential
Kuehn, R., Loehler, J., Rennick, D., Rajewsky, K., and Mueller, W. mechanism for the accelerated vasculopathy of diabetes. J. Clin.
(1993). Interleukin-10-deficient mice develop chronic enterocolitis. Invest. 96, 1395–1403.
Cell 75, 263–274. Schmidt, A.M., Wautier, J.-L., Stern, D., and Yan, S.D. (1998). RAGE:
Lander, H.L., Tauras, J.M., Ogiste, J.S., Moss, R.A., and A.M. a receptor with a taste for multiple ligands and varied pathophysio-
Schmidt. (1997). Activation of the receptor for advanced glycation logic states. Horm. Signal. 1, 41–63.
endproducts triggers a MAP kinase pathway regulated by oxidant
stress. J. Biol. Chem. 272, 17810–17814. Selkoe, D. (1994). Alzheimer’s disease: a central role for amyloid. J.
Neuropathol. Exp. Neurol. 53, 438–447.
Li, J., and Schmidt, A.M. (1997). Characterization and functional
analysis of the promotor of RAGE, the receptor for advanced glyca- Springer, T. (1990). Adhesion receptors of the immune system. Na-
ture 346, 425–434.tion endproducts. J. Biol. Chem. 271, 16498–16506.
Li, H., Cybulsky, M., Gimbrone,M., and Libby, P. (1993). An athero- Sugaya, K., Fukagawa, T., Matsumoto, K., Mita, K., Takahashi, E.,
Ando, A., Inoko, H., andIkemura, T. (1994). Three genes in the humangenic diet rapidly induces VCAM-1, a cytokine-regulatable mononu-
clear leukocyte adhesion molecule in rabbit aortic endothelium. Ar- MHC class III region near the junction with the class II: gene for
receptor for advanced glycosylation end products, PBX2 homeoboxterioscler. Thromb. 13, 197–204. 2 gene and a notch homolog, human counterpart of mouse mam-
Li, J., Qu, X., and Schmidt, A.M. (1998). Sp1 binding elements in mary tumor gene int-3. Genomics 23, 408–419.
the promoter of RAGE are essential for amphoterin-mediated gene
expression in cultured neuroblastoma cells. J. Biol. Chem. 273, Voraberger, G., Schafer, R., and Stratowa, C. (1991). Cloning of the
human gene for intercellular adhesion molecule 1 and analysis of30870–30878. the 59-regulatory region. Induction by cytokines and phorbol ester.
Lugering, N., Stoll, R., Schmid, K.W., Kucharzik, T., Stein, H., Burg- J. Immunol. 147, 2777–2786.
meister, G., Sorg, C., and Domschke, W. (1995). The myeloic related
protein MRP8/14 (27E10 antigen) usefulness as a potential marker Wautier, J-L., Zoukourian,C., Chappey, O.,Wautier, M.P., Guillaus-
seau, P.J., Cao, R., Hori, O., Stern, D., and Schmidt, A.M. (1996).for disease activity inulcerative colitisand putativebiological func-
tion. Eur. J. Clin. Invest. 25, 659–664. Receptor-mediated endothelial cell dysfunction in diabetic vasculo-
pathy: soluble receptor for advanced glycation endproducts blocks
Mackic, J.B., Stins, M., McComb, J.G., Calero, M., Ghiso, J., Kim, hyperpermeability. J. Clin. Invest. 97, 238–243.
K.S.,Yan, S.D., Stern,D.,Schmidt, A.M., Frangione,B., and Zlokovic,
B.V. (1998). Human blood-brain barrier receptors for Alzheimer’s Wicki, R., Marenholz, I., Mischke, D., Schaefer, B.W., Heinzmann,
C.W. (1996). Characterization of the human S100A12 (calgranulin C,amyloid-b1–40: asymmetrical binding, endocytosis, and trans-
cytosis at the apical side of brain microvascular endothelial cell p6, CAAF1, CGRP) gene, a new member of the S100 gene cluster
on chromosome 1q21. Cell Calcium 20, 459–464.monolayer. J. Clin. Invest. 102, 734–743.
Marshak, D.R., Pesce, S.A., Stanley, L.C., and Griffin, W.S. (1992) Yamamura, T., Hitomi, J., Nagasaki, K., Suzuki, M., Takahashi, E.,
Saito, S., Tsukada, T., and Yamaguchi, K. (1996). Human CAAF1Increased S100 beta neurotrophic activity in Alzheimer’s disease
temporal lobe. Neurobiol. Aging 13, 1–7. gene-molecular cloning, gene structure, and chromosome mapping.
Biochem. Biophys. Res. Comm. 221, 356–360.
Miyata, T., O.Hori, J.H., Zhang, S.D., Yan, L., Ferran, Y., Iida, and
Schmidt, A.M. (1996). The receptor for advanced glycation endprod- Yan, S.D., Schmidt, A.-M., Anderson, G., Zhang, J., Brett, J., Zou,
Y.-S., Pinsky, D., and Stern, D. (1994). Enhanced cellular oxidantucts (RAGE) mediates the interaction of AGE-b2-microglobulin with
RAGE and Modulation of the Inflammatory Response
901
stress by the interaction of advanced glycation endproducts with
their receptors/binding proteins. J. Biol. Chem. 269, 9889–9897.
Yan, S.D., Chen, X., Fu, J., Chen, M., Zhu, H., Roher, A., Slattery,
T., Nagashima, M., Morser, J., Migheli, A., et al. (1996). RAGE and
amyloid beta peptide neurotoxicity in Alzheimer’s disease. Nature
382, 685–691.
Yan, S.D., Zhu, H., Fu, J., Yan S.F., Roher, A., Tourtellotte, W., Raja-
vashisth, T., Chen, X., Stern, D., and Schmidt, A.M. (1997). Amyloid-
beta peptide-RAGE interaction elicits neuronal expression of
M-CSF: a proinflammatory pathway in Alzheimer’s disease. Proc.
Natl. Acad. Sci. USA 94, 5296–5301.
Zimmer, D.B., Cornwall, E.H., Landar, A., and Song, W. (1995). The
S100 protein family: history, function, and expression. Brain Res.
Bull. 37, 417–429.
