SHOCK, Vol. 26, No. 2, pp.

174 Y 179, 2006

HMGB1 SIGNALS THROUGH TOLL-LIKE RECEPTOR (TLR) 4 AND TLR2

Man Yu,* Haichao Wang,* Aihao Ding,† Douglas T. Golenbock,‡ Eicke Latz,‡
Christopher J. Czura,* Matthew J. Fenton,§ Kevin J. Tracey,* and Huan Yang*
From the *Laboratories of Biomedical Science, The Feinstein Institute for Medical Research, Manhasset,
NY; † Cornell University, 1300 York Avenue, New York, NY; ‡ Department of Infectious Diseases and
Immunology, University of Massachusetts Medical School, 364 Plantation Street, LRB 370M, Worcester,
MA; and §Division of Pulmonary and Critical Care, University of Maryland School of Medicine, 685 West
Baltimore Street, Baltimore, MD

Received 2 Dec 2005; first review completed 19 Dec 2005; accepted in final form 2 Mar 2006

ABSTRACT—In response to bacterial endotoxin (e.g., LPS) or endogenous proinflammatory cytokines (e.g., TNF and
IL-1"), innate immune cells release HMGB1, a late cytokine mediator of lethal endotoxemia and sepsis. The delayed
kinetics of HMGB1 release makes it an attractive therapeutic target with a wider window of opportunity for the treatment of
lethal systemic inflammation. However, the receptor(s) responsible for HMGB1-mediated production of proinflammatory
cytokines has not been well characterized. Here we demonstrate that in human whole blood, neutralizing antibodies
against Toll-like receptor 4 (TLR4, but not TLR2 or receptor for advanced glycation end product) dose-dependently
attenuate HMGB1-induced IL-8 release. Similarly, in primary human macrophages, HMGB1-induced TNF release is dose-
dependently inhibited by anti-TLR4 antibodies. In primary macrophages from knockout mice, HMGB1 activates
significantly less TNF release in cells obtained from MyD88 and TLR4 knockout mice as compared with cells from
TLR2 knockout and wild-type controls. However, in human embryonic kidney 293 cells transfected with TLR2 or TLR4,
HMGB1 effectively induces IL-8 release only from TLR2 overexpressing cells. Consistently, anti-TLR2 antibodies dose-
dependently attenuate HMGB1-induced IL-8 release in human embryonic kidney/TLR2-expressing cells and markedly
reduce HMGB1 cell surface binding on murine macrophage Y like RAW 264.7 cells. Taken together, our data suggest that
there is a differential usage of TLR2 and TLR4 in HMGB1 signaling in primary cells and in established cell lines, adding
complexity to studies of HMGB1 signaling which was not previously expected.
KEYWORDS—HMGB1, TLR2, TLR4, signal transduction, receptor

INTRODUCTION HMGB1 and its role as a late mediator in lethal systemic

inflammation make it a potential therapeutic target. However,
Vertebrate animals have evolved an innate response to
the molecular basis for HMGB1-induced cytokine signaling in
infection and/or trauma. The response is initiated upon ac-
macrophage/monocyte is poorly understood.
tivation of pro-inflammatory cytokines, including HMGB1,
Previous studies implicated receptor for advanced glycation
released either actively from macrophage/monocyte or
end product (RAGE) as an HMGB1 receptor that mediates
passively from necrotic and damaged cells (1 Y 4). The early
neurite outgrowth during brain development and migration of
macrophage/monocyte response is associated with the release
smooth muscle cells. Binding of RAGE to HMGB1 leads to
of TNF, IL-1", and other early inflammatory mediators of
activation of NF-.B and ERK1/2, p38 and SAPK/JNK kinases
shock and tissue injury. The late responses, occurring after
(13 Y 15). In addition to RAGE, Toll-like receptor (TLR) 2 and
the early pro-inflammatory response, lead to the release
TLR4 have been implicated in HMGB1 signaling (16 Y 18).
of HMGB1 into the serum which mediates systemic in-
TLRs are highly conserved proteins that activate innate
flammation, tissue damage, and lethality (1, 4). HMGB1,
immune cells in response to a variety of endogenous and
released passively by injured or necrotic cells, acts as a pro-
exogenous stimuli (19). TLR proteins transduce signals
inflammatory cytokine and functions as a major stimulus of
through a family of cytosolic Toll adapter proteins that link
necrosis-induced inflammation (2). Administration of HMGB1
to downstream signaling cascades highly similar to those
to normal animals causes inflammatory responses including
activated by IL-1 and IL-18 (20 Y 22). Recent findings
fever, weight loss and anorexia, acute lung injury, epithelial
implicated that HMGB1 can signal via both TLR2 and
barrier dysfunction, arthritis, and death (5 Y10). Anti-HMGB1
TLR4 in neutrophils and macrophage-like RAW 264.7 cells;
treatment with antibodies or other strategies to inhibit
RAGE plays only a minor role in macrophage activation by
HMGB1 rescues mice from lethal bacterial toxin Yinduced
HMGB1 (16). Others have also shown that HMGB1 acts as a
endotoxemia or sepsis and ameliorates the severity of
mediator in hepatic ischemia-reperfusion Y induced injury via
collagen-induced arthritis (5, 11, 12). The delayed release of
a process that is TLR4 dependent (17). Mice defective in
TLR4 (C3H/Hej) developed less hepatic injury as compared
Address reprint requests to Dr. Huan Yang, The Feinstein Institute for Medical with wild-type (C3H/HeOuj) controls. However, we do not
Research, Manhasset, NY 11030. E-mail: hyang@nshs.edu. know whether TLR2 or TLR4 is involved in HMGB1
Supported by grants from The Feinstein Institute for Medical Research, GCRC, M01 signaling independently or together. These results prompted
RR018535, NIH (NIGMS), and Critical Therapeutics, Inc., (Lexington, MA) (to K.J.T.).
DOI: 10.1097/01.shk.0000225404.51320.82 us to investigate the role of TLR2 and TLR4 in HMGB1
Copyright Ó 2006 by the Shock Society signaling in primary cells and established cell lines.
174
Copyright @ 2006 by the Shock Society. Unauthorized reproduction of this article is prohibited.
SHOCK AUGUST 2006
Our findings showed that in human whole blood, HMGB1-
induced IL-8 release is dose-dependently attenuated by neutral-
izing antibodies against TLR4 but not TLR2 or RAGE. Similarly,
in primary human macrophages, HMGB1-induced TNF release
is dose-dependently inhibited by anti-TLR4 antibodies. In
macrophages from knockout mice, HMGB1 stimulated signifi-
cantly less TNF release in cells from MyD88 and TLR4 knockout
mice as compared with TLR2 knockout or wild-type controls. In
contrast, in human embryonic kidney (HEK) 293 cells trans-
fected with TLRs, HMGB1 effectively induces IL-8 release from
TLR2, but not TLR4 overexpressing cells. Consistently, anti-
TLR2 antibodies dose-dependently attenuate HMGB1-induced
IL-8 release in HEK/TLR2-expressing cells and markedly reduce
HMGB1 cell surface binding on murine macrophage-like RAW
264.7 cells. Our study demonstrates that HMGB1 exhibits a
differential usage of TLR2 and TLR4 in different cell types.
MATERIALS AND METHODS
Materials
Cell culture media (DMEM and RPMI 1640), L-glutamine, penicillin/strepto-
mycin, and phosphate-buffered saline (PBS) solutions were obtained from
BioWhittaker, Inc. (Walkersville, MD). Opti-MEM I medium was purchased from
Invitrogen (Carlsbad, CA). Fetal bovine serum (FBS) was obtained from Hyclone
(Logan, UT). Geneticin (G418) and !-MEM medium were from Gibco-BRL
(Grand Island, NY). Polymyxin B, Triton X-114, LPS (endotoxin, Escherichia coli
0111:B4), phenylmethylsuphonyl fluoride (PMSF), macrophage-colony stimulat-
ing factor, and nonimmune rabbit IgG and 3 FLAG peptide affinity gel were
purchased from Sigma Chemicals (St. Louis, MO). Pam2CSK4 was obtained from
EMC Microcollections (Catalog no. L2020, Tuebingen, Germany). Isopropyl-D-
thiogalactopyranoside was purchased from Pierce Fine Chemicals (Rockford, IL).
LB broth and Chinese Hamster Ovary (CHO)-S-SFM II medium were obtained
from Life Technologies (Grand Island, NY). Mouse antihuman TLR2 (TL2.1, cat
no. 16-9922) and TLR4 (HTA125, cat no. 16-9917) antibodies were from
eBioscience (San Diego, CA), and mouse antihuman RAGE antibodies (cat no.
MAB5328) were from Chemicon International, Inc., (Temecula, CA). Affinity-
purified polyclonal antibodies against mouse TLR2 (sc-10739) and TLR4
antibodies (sc-10741) were from Santa Cruz Biotech, Inc., (Santa Cruz, CA).
Functional grade Y purified antibodies against mouse TLR2 (cat no. 16-9024) and
TLR4 (cat no. 16-9924) were from eBioscience and were used in confocal
microscopy studies. Antihuman TLR2 monoclonal antibodies (sc-21759) were
used for in vitro neutralization studies in HEK 293 cells. Antibodies were dialyzed
extensively against PBS to remove sodium azide before use in cell culture.
Cell culture
Murine macrophage Y like RAW 264.7 cells and CHO cells (CHO-AA8, CRL-
1859) were obtained from American Type Culture Collection (ATCC, Rockville,
MD). RAW 264.7 cells were cultured in RPMI 1640 medium, and CHO cells were
cultured in !-MEM media, both supplemented with 10% heat-inactivated FBS,
2 mM glutamine, and 1 penicillin/streptomycin and grown in a humidified
incubator with 5% CO2. Cells were used at 90% confluence, and treatment was
carried out in Opti-MEM I medium.
HEK 293 cells stably transfected with fusion proteins consisting of cyan fluorescent
protein fused to TLR2 or TLR4 at the C-terminus (or pcDNA alone) were described
previously (23). The stably transfected HEK 293 cells were cultured in DMEM
medium supplemented with 10% FBS, 1 penicillin/streptomycin, 2 mM glutamine,
and 500 2g/mL geneticin and used for experiments at 90% confluence.
Human primary monocytes were purified by density gradient centrifugation
through Ficoll from blood donated by normal individuals to the Long Island Blood
Bank (New York Blood Center, Melville, NY) (24). Cells were allowed to
differentiate into macrophages for 7 days in complete DMEM medium containing
macrophage-colony stimulating factor (1 ng/mL) and then used in the experiments.
MyD88, TLR2, or TLR4 knockout mice were obtained from Dr. S. Akira
(Osaka University, Osaka, Japan). Bone marrow Y derived macrophages were
isolated and differentiated matured to macrophages in 20% L-cell medium as
previously described (25). Cells were incubated with HMGB1 (10 2g/mL) for 24 h,
and TNF released in the conditioned medium was measured.
Copyright @ 2006 by the Shock Society. Unauthorized reproduction of this article is prohibited.
HMGB1 SIGNALS VIA TLR4 AND TLR2 175
MyD88 dominant/negative cell lines
RAW 264.7 macrophages were transiently co-transfected with an expression
plasmid encoding a murine MyD88 Y dominant-negative mutant (corresponding to
amino acids 146-296), or empty vector, plus a luciferase reporter plasmid under the
control of the NF-.B Y dependent ELAM promoter.
CHO reporter cell lines that constitutively express human TLR2 or TLR4 were
described previously (26, 27). These reporter lines also contain a stably transfected
ELAM-CD25 reporter gene and express human CD25 on their surface as a
consequence of NF-.B activation.
FITC labeling of HMGB1
HMGB1 protein was labeled with fluorescein isothiocyanate (FITC) using a
labeling kit according to the manufacturer’s instructions (Pierce).
Generation of CHO-psF-HMGB1 cell line clone and
production of HMGB1 in CHO cells
Methods for HMGB1 isolation from mammalian CHO cells were described
previously (28). Briefly, CHO cells were transfected with plasmid psF-HMGB1
using the calcium phosphate method according to the manufacturer’s instructions
(Gibco-BRL). The best HMGB1 secreting cell line was adapted to grow in
suspension by culturing in CHO-S-SFM II media supplemented with 2 mM
glutamine, 300 2g/mL geneticin, and 1 penicillin/streptomycin. HMGB1
secretion was approximately 5 2g/mL in the medium. HMGB1 protein was
isolated from conditioned medium by affinity purification using FLAG antibody
(ANTI-FLAG 2 affinity gel) according to the manufacturer’s instructions (Sigma).
Confocal microscopy
Macrophage-like RAW 264.7 cells were plated in 8-well slide chambers (Lab-
Tek, Nalge Nunc Internationals, Naperville, IL) and used at 70% confluence. Cells
were preincubated with anti-TLR2 or TLR4 antibodies of 25 2g/mL for 30 min at
4-C in Opti-MEM I medium, and FITC-labeled HMGB1 of 5 2g/mL was added
and incubated for an additional 6 h at 4-C. Cells were then washed three times with
PBS (pH 7.2) and fixed for 15 min at room temperature in 10% paraformaldehyde-
PBS. After fixing, cells were washed three times with PBS and mounted for
fluorescent confocal microscopy analysis.
LPS content in HMGB1
Contaminating LPS in protein preparations was removed by phase separation
using Triton X-114 as previously described (28). Briefly, 1/20 volume of Triton
X-114 was added to HMGB1 solution. After 10 min of gentle rotation at room
temperature, the samples were centrifuged for 10 min (8,000  g) at room
temperature, and the top layer (containing HMGB1 proteins) was carefully
aspirated and saved. The LPS content in HMGB1 preparations was measured by
the Chromogenic Limulus Amebocyte lysate assay according to the manufac-
turer’s instructions (BioWhittaker, Inc.). LPS content in HMGB1 expressed in E.
coli. is typically 0.6 T 0.3 pg/2g (28). For stimulation experiments, polymyxin B
was added to cell culture medium at 6 U/pg LPS (5).
Cytokine measurements
The concentrations of TNF and IL-8 were determined using commercially
obtained enzyme-linked immunosorbent assay (ELISA) kits according to the
manufacturer’s instructions (R&D System, Inc., Minneapolis, MN).
Antibody production
Polyclonal antibodies against HMGB1 B box were raised in rabbits (Cocalico
Biologicals, Inc., Reamstown, PA) and assayed for titer by Western blotting. IgG
was purified from anti-HMGB1 antiserum using Protein A agarose according to the
manufacturer’s instructions (Pierce). Neutralizing activity of anti-HMGB1 anti-
bodies was tested as previously described (11).
Statistical analysis
Data are presented as mean T SEM, unless otherwise stated. Differences
between treatment groups were determined by Student’s t test. P value less than
0.05 is considered statistically significant.
RESULTS
HMGB1-stimulated IL-8 release from human whole blood
and TNF from primary macrophages are dose-dependently
inhibited by anti-TLR4 antibodies
To examine the role of TLRs in HMGB1 signaling, we first
used whole blood obtained from healthy volunteers. Neutralizing
antibodies against TLR2, TLR4, or RAGE were preincubated
176 SHOCK VOL. 26, NO. 2 YU ET AL.

FIG. 1. HMGB1-stimulated IL-8 release from human whole blood and TNF from primary macrophages are dose-dependently inhibited by anti-
TLR4 antibodies, but not by anti-TLR2 antibodies. Blood from healthy volunteers was drawn into heparinized syringes (15 U/mL blood). Whole blood
(200 2L) was preincubated with neutralizing anti-TLR2, TLR4, or RAGE antibodies or control IgG as indicated for 30 min at 37-C. HMGB1 (5 2g/mL) was then added
and incubated for an additional 4 h in the presence of polymyxin B (350 U/mL). After incubation, plasma was harvested and IL-8 release was measured. Data are
presented as mean T SEM from three donors. *P G 0.05 versus HMGB1 alone. Peritoneal macrophages from human blood (2  105 cells/well in 48-well culture
dish) were stimulated with HMGB1 (3 2g/mL) in the presence of neutralizing anti-TLR2 and TLR4 antibodies or control IgG as indicated for 16 h. TNF release was
measured in the conditioned media. Data shown are means T SEM of five to seven experiments. *P G 0.05 versus HMGB1 alone.

with whole blood (200 2L) for 30 min, and then HMGB1 (5 2g/
mL) was added for 4 h at 37-C. We observed that HMGB1-
mediated IL-8 release was dose-dependently inhibited by anti-
TLR4 antibodies, but not by anti-TLR2 or RAGE antibodies
(Fig. 1A). Similarly, in isolated primary human macrophages,
HMGB1-induced TNF release was dose-dependently attenuated
by anti-TLR4 antibodies (Fig. 1B), suggesting that HMGB1
signals via TLR4 in human whole blood and in macrophages.
HMGB1 induces TNF release on macrophages from
wild-type and TLR2 knockout mice, not from cells obtained
from MyD88 or TLR4 knockout mice
Previous studies indicated that TLR proteins transduce
signals through an intermediate protein MyD88 (19, 29). To
determine the role of MyD88 and TLRs in HMGB1 signaling,
we next used bone marrow cells obtained from wild-type,
MyD88, TLR2, or TLR4 knockout mice (25). Cells were
matured to macrophages in 20% L-cell medium and were then
incubated with HMGB1 (10 2g/mL) for 24 h, and TNF release
was measured (Fig. 2). HMGB1 stimulated TNF release from
wild type and TLR2 knockouts, but significantly less TNF
release occurred in the MyD88 and TLR4 knockout cells.
These data confirmed the results from human whole blood and
indicated that MyD88 and TLR4 are important in HMGB1
signaling in primary cells.
CHO cells also triggered IL-8 release in a dose-dependent
manner, suggesting that HMGB1 signals via TLR2 in HEK
293 cells overexpressing TLR2 (Fig. 3). Positive controls
using LPS, a known TLR4 agonist, stimulated IL-8 release in
HEK/TLR4 cells but not in TLR2 or pcDNA-transfected HEK
293 cells (Fig. 3). In contrast, Pam2CSK4, a synthetic
lipopeptide originated from mycoplasm and a specific TLR2
agonist, significantly stimulated IL-8 release only from TLR2
overexpressing cells, indicating that the recombinant TLR
proteins expressed by the HEK 293 cells are capable of
responding to their ligands specifically (Fig. 3). These data
also showed discrepancy using TLRs in HMGB1 signaling
between primary cells and established cell lines.
HMGB1-stimulated IL-8 release is dose-dependently
inhibited by anti-TLR2 antibodies in HEK/TLR2
overexpressing cells
We used anti-TLR2 antibodies to ascertain the specificity
of TLR2-dependent HMGB1 signaling as observed in HEK/

HMGB1 stimulates IL-8 release from HEK 293 cells

overexpressing TLR2, but not from HEK 293 cells
overexpressing TLR4 or vector protein
To further examine the role of TLRs in HMGB1 signaling,
we used the cultures of HEK 293 cells engineered to
overexpress either TLR2, TLR4, or control vector alone
(23). Unexpectedly, HMGB1 (produced in E. coli) dose-
dependently stimulated IL-8 release in HEK 293 cells over-
expressing TLR2 but not from cells overexpressing TLR4 or
control vector alone (Fig. 3). To eliminate the possibility of
bacterial contamination from E. coli, we used HMGB1
expressed in mammalian CHO cells. HMGB1 purified from
FIG. 2. HMGB1 stimulates TNF release from macrophages obtained
from wild-type, but not MyD88, TLR2, or TLR4 knockout mice. Bone
marrow cells were obtained from wild-type, MyD88, TLR2, or TLR4 knockout
mice and were matured to macrophages in 20% L-cell medium. Cells were
incubated with HMGB1 (10 2g/mL) for 24 h, and TNF released in the
conditioned medium was measured. Data are presented as percent of TNF
stimulated by wild-type cells with n = 3 separate experiments. (The range of
HMGB1-stimulated TNF release varies from 250 to 450 pg/mL.) *P G 0.05
versus wild-type control.

Copyright @ 2006 by the Shock Society. Unauthorized reproduction of this article is prohibited.
SHOCK AUGUST 2006
TLR2 overexpressing cells. Neutralizing anti-TLR2 anti-
bodies dose-dependently inhibited HMGB1-induced IL-8
release. This suppression was specific because nonimmune
control IgG did not significantly inhibit HMGB1-induced IL-8
release (Fig. 4).
HMGB1 interacts with TLR2 in RAW 264.7 cells
To further study the interaction between TLRs and
HMGB1, we examined the effects of anti-TLR2 and anti-
TLR4 antibodies on HMGB1 cell surface binding in macro-
phage cultures using confocal microscopy. Cell surface
binding of FITC-HMGB1 was observed in macrophage-like
RAW 264.7 cells incubated with FITC-HMGB1 for 6 h at 4-C
(Fig. 5). Pretreatment with anti-TLR4 antibodies or IgG did
not have any significant effect, whereas anti-TLR2 antibodies
markedly reduced HMGB1 cell surface binding as revealed by
the decrease in cell-associated fluorescence compared with
IgG-treated controls (Fig. 5), confirming the physical inter-
action of HMGB1 with TLR2 in RAW 264.7 cells.
HMGB1 signaling Is dependent on MyD88 in RAW 264.7
and TLR2 in CHO cells transfected with TLRs
We next examined the possible role of MyD88 in
mediating HMGB1 signaling in cell lines. RAW 264.7 cells
were transfected with plasmids encoding a dominant negative
mutant MyD88 protein and an NF-.B Y dependent luciferase
reporter. The dominant negative MyD88 mutant significantly
inhibited the activation of NF-.B luciferase activity in-
duced by HMGB1 indicating a MyD88-dependent response
(Fig. 6A). To address the specificity of the MyD88-dependent
response, HMGB1 was added to cultures of CHO cells
engineered to constitutively express either human TLR2 or
TLR4 (Fig. 6B). These cells also contain a NF-.B Y dependent
CD25 reporter gene as previously described (26, 27).
Compared to IL-1 which stimulated CD25 expression from
both TLR2 and TLR4 expressing cells, HMGB1 failed to
increase CD25 expression in TLR4-expressing cells but
significantly stimulated CHO cells expressing TLR2. These
data indicate that HMGB1 signals via TLR2, but not TLR4 in
established cell lines. Taken together, our findings show that
there is a differential usage of TLR2/TLR4 in HMGB1
signaling, depending on the cell types. HMGB1 acts as an
endogenous ligand for TLR4 on primary cells, whereas it acts
mainly through TLR2 in established cell lines (Fig. 7).
Copyright @ 2006 by the Shock Society. Unauthorized reproduction of this article is prohibited.
HMGB1 SIGNALS VIA TLR4 AND TLR2 177
FIG. 3. HMGB1 (expressed in E. coli or CHO cells)
induces IL-8 release from HEK 293 cells overexpress-
ing TLR2, but not from cells overexpressing TLR4 or
vector. HEK 293 cells overexpressing TLR2, TLR4, or
vector alone (pcDNA3) were plated in 96-well culture
plates and were used at 90% confluence. Cells were
stimulated with recombinant HMGB1 expressed in either
E. coli or mammalian CHO cells, positive controls using
LPS (ligand for TLR4, 100 ng/mL) or Pam2CSK4 (ligand
for TLR2, 1 ng/mL) for 18 h; the conditioned media were
assayed for IL-8 release by ELISA. Data represent mean T
SEM; n = 3-6 independent experiments. *P G 0.05 versus
unstimulated control.
DISCUSSION
HMGB1 released by monocyte/macrophage after exposure to
TNF, IL-1", or bacterial toxins causes delayed lethality and
inflammation (5, 24). HMGB1 can interact with TLR2, TLR4,
and RAGE in established cell lines and in animal models (8,
16Y18, 30). TLRs recognize specific molecular patterns that
are present in microbial products (19, 29). TLR4 recognizes the
Gram-negative product LPS and participates in the recognition
of several endogenous ligands including heat shock proteins,
whereas TLR2 recognizes various fungal, Gram-positive, and
mycobacterial components (19, 29, 31, 32). The purpose of this
study was to further examine what receptors are involved in
HMGB1-induced activation of inflammatory signaling using
both primary and established cell lines. We observed unex-
pected difference in requirement for TLRs in HMGB1 signal-
ing in different cell types: (a) HMGB1 acts through TLR4 in
human whole blood and primary macrophages, and (b) in
contrast, HMGB1 signals via TLR2 in macrophage-like RAW
264.7 cells and CHO cells overexpressing TLRs, although the
involvement of another TLR4-dependent cell type cannot be
excluded (Fig. 7). Our data immediately raise the question of
why there is differential usage of TLRs in HMGB1 signaling in
different cell types?
Although previous studies showed that RAGE is involved
in HMGB1 signaling (8, 13, 15, 30), we found that anti-
RAGE antibodies did not inhibit HMGB1-induced IL-8
FIG. 4. Anti-TLR2 antibodies dose-dependently inhibit HMGB1-
induced IL-8 release in HEK/TLR2 cells. HEK 293 cells overexpressing
TLR2 were plated in 96-well culture plates. Cells were stimulated with
HMGB1 for 18 h in the presence or absence of anti-TLR2 antibodies or
control IgG at the concentrations indicated. Conditioned media were
collected, and IL-8 release was measured by ELISA. *P G 0.05 versus
HMGB1 alone. Data are presented as mean T SEM from three or more
separate experiments.
178 SHOCK VOL. 26, NO. 2 YU ET AL.

FIG. 5. Anti-TLR2 antibodies inhibit FITC-HMGB1 cell

surface binding. Macrophage-like RAW 264.7 cells were
plated in 8-well slide chambers and were used at 70%
confluence. Cells were preincubated with anti-TLR2 and TLR4
antibodies or control IgG of 25 2g/mL for 30 min at 4-C in Opti-
MEM I medium. FITC-labeled HMGB1 (and unlabeled HMGB1)
was added at 5 2g/mL followed by an additional incubation of
6 h at 4-C. Cells were then washed three times with PBS and
were fixed in 4% paraformaldehyde-PBS solution (pH 7.2) for
20 min at room temperature. After fixing, cells were washed
with PBS three times and were mounted for viewing by
fluorescent confocal microscopy. Data are representative
from five to eight experiments. Magnification, 40.

release in human whole blood assays. However, our results do
not exclude the possibility that RAGE signaling is involved in
other cell types. Additional experiments are needed to
determine the relative contribution of each receptor in
HMGB1 signaling.
Recent findings showed that streptococcal cell wall-induced
joint inflammation is dependent on TLR2-signaling and
MyD88, demonstrating the critical role of TLR2 and MyD88
in sepsis and in rheumatoid arthritis where excessive amounts
of HMGB1 are produced (7, 31). Scaffidi et al. demonstrated
that necrotic cells from wild-type mice cause inflammation,
whereas necrotic cells from HMGB1 knockout mice do not
(2). It is plausible that this putative nuclear factor is HMGB1.
These results are consistent with our current findings which
indicate that HMGB1 is a nuclear factor that causes the
inflammatory responses and HMGB1 acts, at least partially,
through TLR4 and TLR2. Our results are complementary to
the findings by Park et al. showing that HMGB1 signals
through both TLR2 and TLR4 in human neutrophils (16) and
macrophages (18) and further revealed a differential usage of
TLR2 and TLR4 in different cell types. It remains theoret-
ically possible that HMGB1 binds endotoxin and bacterial
lipoproteins and functions to synergistically increase the
responses via TLR4 and TLR2. The recent study by Tsung
et al. suggested that even circulating HMGB1 requires intact
TLR4 for effective signaling in vivo (17). Taken together, our
observations demonstrate that HMGB1 signals via distinct
pathways through TLR4 in primary cells or TLR2 in
established cell lines such as RAW 264.7 cells, CHO, or
HEK 293 cells overexpressing TLR2 and TLR4. The reasons
are not clear, but caution should be taken when interpreting
HMGB1 signaling results from cell lines or overexpressing
systems. Understanding the nature of molecular signaling by
HMGB1 may prove therapeutically beneficial in the treatment

FIG. 6. MyD88 and TLR2 are required in HMGB1-mediated cellular responses. A, MyD88 is required for HMGB1 signaling. Subconfluent RAW 264.7
cells (overexpressing either wild-type or mutant MyD88 and a NF-.B Y dependent luciferase reporter) in 24-well plates were stimulated with HMGB1 (100 ng/mL)
for 5 h. Cells were harvested and luciferase activity was measured as described. All transfections were performed in triplicate, repeated at least three times, and a
single representative experiment is shown. Data are expressed as the ratio (fold-activation) of average luciferase values from unstimulated to stimulated cells
(subtracted for background) T SD. *P G 0.05 versus control. B, TLR2 is required for HMGB1 signaling. CHO cells overexpressing TLR2 or TLR4 were seeded into
24-well tissue culture plates and used at 90% confluence. Cells were stimulated with IL-1 (10 ng/mL) or purified HMGB1 (100 ng/mL) for 18 h. Following
stimulation, cells were stained with a phycoerythrin-labeled anti-CD25 monoclonal antibody, and surface expression of CD25 was measured by flow cytometry.
Data are expressed as the ratio (fold-activation) of the percent of CD25 positive cells in unstimulated and stimulated cell populations that were gated to exclude the
lowest 5% of cells based on mean FL1 fluorescence. Data shown are from a single experiment and are representative from three separate experiments.

Copyright @ 2006 by the Shock Society. Unauthorized reproduction of this article is prohibited.
SHOCK AUGUST 2006
FIG. 7. HMGB1 uses different TLRs in signaling for stimulation of
cytokine release in primary cells and established cell lines. HMGB1
uses different TLRs for signaling in different cell types. HMGB1 signals
through TLR4 in human whole blood and primary macrophages obtained
from MyD88, TLR2, or TLR4 knockout mice. In contrast, HMGB1 signals via
TLR2 in macrophage-like RAW 264.7 cells and in CHO cells overexpressing
TLR2 or TLR4.
of human diseases where excessive amounts of HMGB1 are
produced.
REFERENCES
1. Yang H, Wang H, Czura CJ, Tracey KJ: The cytokine activity of HMGB1.
J Leukoc Biol 78:1 Y 8, 2005.
2. Scaffidi P, Misteli T, Bianchi ME: Release of chromatin protein HMGB1 by
necrotic cells triggers inflammation. Nature 418:191 Y 195, 2002.
3. Tracey KJ, Fong Y, Hesse DG, Manogue KR, Lee AT, Kuo GC, Lowry SF,
Cerami A: Anti-cachectin/TNF monoclonal antibodies prevent septic shock
during lethal bacteraemia. Nature (Lond) 330:662 Y 664, 1987.
4. Tracey KJ. Fatal sequence, the killer within. Washington D.C.: Dana Press,
2005.
5. Wang H, Bloom O, Zhang M, Vishnubhakat JM, Ombrellino M, Che J, Frazier A,
Yang H, Ivanova S, Borovikova L, et al: HMG-1 as a late mediator of endotoxin
lethality in mice. Science 285:248 Y 251, 1999.
6. Abraham E, Arcaroli J, Carmody A, Wang H, Tracey KJ: HMG-1 as a
mediator of acute lung injury. J Immunol 165:2950 Y 2954, 2000.
7. Kokkola R, Sundberg E, Ulfgren A-K, Palmblad K, Li J, Wang H, Ulloa L,
Yang H, Yan XJ, Furie R, et al: High mobility group box chromosomal protein
1 (HMGB1)-a novel proinflammatory mediator in synovitis. Arthritis Rheum
46:2598 Y 2603, 2002.
8. Sappington PL, Yang R, Yang H, Tracey KJ, Delude RL, Fink MP: HMGB1 B
box increases the permeability of Caco-2 enterocytic monolayers and impairs
intestinal barrier function in mice. Gastroenterology 123:790 Y 802, 2002.
9. Taniguchi N, Kawahara K, Yone K, Hashiguchi M, Goto M: High
mobility group box chromosomal protein 1 plays a role in the pathogenesis of
rheumatoid arthritis as a novel cytokine. Arthritis Rheum 48:971 Y 981, 2003.
10. Agnello D, Wang H, Yang H, Tracey KJ, Ghezzi P: HMGB1, a DNA-binding
protein with cytokine activity, induces brain TNF and IL-6 production, and
mediates anorexia and taste aversion. Cytokine 18:231 Y 236, 2002.
11. Yang H, Ochani M, Li J, Qiang X, Tanovic M, Harris HE, Susarla SM, Ulloa L,
Wang H, DiRaimon R, et al: Reversing established sepsis with antagonists of
endogenous HMGB1. Proc Natl Acad Sci U S A 101:296 Y 301, 2004.
12. Kokkola R, Li J, Sundberg E, Aveberger AC, Palmblad K, Yang H, Tracey KJ,
Andersson U, Harris HE: Successful treatment of collagen-induced arthritis in
mice and rats by targeting extracellular high mobility group box chromosomal
protein 1 activity. Arthritis Rheum 48:2052 Y 2058, 2003.
Copyright @ 2006 by the Shock Society. Unauthorized reproduction of this article is prohibited.
HMGB1 SIGNALS VIA TLR4 AND TLR2 179
13. Hori O, Brett J, Slattery T, Cao R, Zhang J, Chen JX, Nagashima M, Lundh ER,
Vijay S, Nitecki D, et al: The receptor for advanced glycation end products
(RAGE) is a cellular binding site for amphoterin. J Biol Chem 270:
25752 Y 25761, 1995.
14. Huttunen HJ, Rauvala H: Amphoterin as an extracellular regulator of cell
motility: from discovery to disease. J Intern Med 255:351 Y 366, 2004.
15. Degryse B, Bonaldi T, Scaffidi P, Muller P, Resnati M, Sanvito F, Arrigoni G,
Bianchi ME: The high mobility group (HMG) boxes of the nuclear protein
HMG1 induce chemotaxis and cytoskeleton reorganization in rat smooth
muscle cells. J Cell Biol 152:1192 Y 1206, 2001.
16. Park JS, Svetkauskaite D, He Q, Kim J, Strassheim D, Ishizaka A, Abraham E:
Involvement of TLR2 and TLR4 in cellular activation by high mobility group
box 1 protein (HMGB1). J Biol Chem 279:7370 Y 7376, 2004.
17. Tsung A, Sahai R, Tanaka H, Nakao A, Fink MP, Lotze MT, Yang H, Li J,
Tracey KJ, Geller DA, et al: The nuclear factor HMGB1 mediates hepatic injury
after murine liver ischemia-reperfusion. J Exp Med 201:1135 Y 1143, 2005.
18. Park JS, Gamboni-Robertson F, He Q, Svetkauskaite D, Kim JY, Strassheim
D, Sohn JW, Yamada S, Maruyama I, Banerjee A, et al: High Mobility Group
Box 1 protein (HMGB1) interacts with multiple Toll like receptors. Am J
Physiol Cell Physiol 290:C917 Y C924, 2006.
19. Aderem A, Ulevitch RJ: Toll-like receptors in the induction of the innate
immune response. Nature 406:782 Y 787, 2000.
20. Bowie A, O’Neill LA: The interleukin-1 receptor-Toll-like receptor super-
family: sign generators for pro-inflammatory interleukins and microbial
products. J Leukoc Biol 67:508 Y 514, 2000.
21. O’Neill LA: Signal transduction pathways activated by the IL-1 receptor/toll-
like receptor superfamily. Curr Top Microbiol Immunol 270:47 Y 61, 2002.
22. Martin MU, Wesche H: Summary and comparison of the signaling mecha-
nisms of the Toll/interleukin-1 receptor family. Biochem Biophys Acta 1592:
265 Y 280, 2002.
23. Latz E, Visintin A, Lien E, Fitzgerald KA, Monks BG, Kurt-Jones EA,
Golenbock DT, Espevik T: Lipopolysaccharide rapidly traffics to and from the
Golgi apparatus with the Toll-like receptor 4-MD-2-CD14 complex in a
process that is distinct from the initiation of signal transduction. J Biol Chem
277:47834 Y 47843, 2002.
24. Andersson U, Wang H, Palmblad K, Aveberger A-C, Bloom O, Harris HE,
Janson A, Kokkola R, Yang H, Tracey KJ: HMG-1 stimulates proinflamma-
tory cytokine synthesis in human monocytes. J Exp Med 192:565 Y 570, 2000.
25. Shi S, Nathan C, Schnappinger D, Drenkow J, Fuortes M, Block E, Ding A,
Gingeras TR, Schoolnik G, Akira S, et al: MyD88 primes macrophages for
full-scale activation by interferon-gamma yet mediates few responses to
Mycobacterium tuberculosis. J Exp Med 198:987 Y 997, 2003.
26. Yoshimura AE, Lien E, Toumanen R, Dziarski D, Golenbock DT: Recognition
of Gram-positive bacterial cell wall components by the innate immune system
occurs via Toll-like receptor 2. J Immunol 163:1 Y 5, 1999.
27. Delude RL, Yoshimura A, Ingalls RR, Golenbock DT: Construction of a
lipopolysaccharide reporter cell line and its use in identifying mutants
defective in endotoxin, but not TNF-alpha signal transduction. J Immunol
161:3001 Y 3009, 1998.
28. Li J, Wang H, Mason JM, Levine J, Yu M, Ulloa L, Czura CJ, Tracey KJ,
Yang H: Recombinant HMGB1 with cytokine-stimulating activity. J Immunol
Methods 289:211 Y 223, 2004.
29. Takeuchi O, Akira S: Toll-like receptors; their physiological role and signal
transduction system. Int Immunopharmacol 1:625 Y 635, 2001.
30. Huttunen HJ, Fages C, Kuja-Panula J, Ridley AJ, Rauvala H: Receptor for
advanced glycation end products-binding COOH-terminal motif of amphoterin
inhibits invasive migration and metastasis. Cancer Res 62:4805 Y 4811, 2002.
31. Joosten LA, Koenders MI, Smeets RL, Heuvelmans-Jacobs M, Helsen MM,
Takeda K, Akira S, Lubberts E, van de Loo FA, van den Berg WB: Toll-like
receptor 2 pathway drives streptococcal cell wall-induced joint inflamma-
tion: critical role of myeloid differentiation factor 88. J Immunol 171:
6145 Y 6153, 2003.
32. Vabulas RM, Braedel S, Hilf N, Singh-Jasuja H, Herter S, Ahmad-Nejad P,
Kirschning CJ, Da Costa C, Rammensee HG, Wagner H, et al: The endoplasmic
reticulum-resident heat shock protein Gp96 activates dendritic cells via the Toll-
like receptor 2/4 pathway. J Biol Chem 277:15107 Y 15112, 2002.