A Major Ingredient of Green Tea Rescues Mice from Lethal Sepsis Partly by Inhibiting HMGB1

Background The pathogenesis of sepsis is mediated in part by bacterial endotoxin, which stimulates macrophages/monocytes to sequentially release early (e.g., TNF, IL-1, and IFN-γ) and late (e.g., HMGB1) pro-inflammatory cytokines. Our recent discovery of HMGB1 as a late mediator of lethal sepsis has prompted investigation for development of new experimental therapeutics. We previously reported that green tea brewed from the leaves of the plant Camellia sinensis is effective in inhibiting endotoxin-induced HMGB1 release. Methods and Findings Here we demonstrate that its major component, (-)-epigallocatechin-3-gallate (EGCG), but not catechin or ethyl gallate, dose-dependently abrogated HMGB1 release in macrophage/monocyte cultures, even when given 2–6 hours post LPS stimulation. Intraperitoneal administration of EGCG protected mice against lethal endotoxemia, and rescued mice from lethal sepsis even when the first dose was given 24 hours after cecal ligation and puncture. The therapeutic effects were partly attributable to: 1) attenuation of systemic accumulation of proinflammatory mediator (e.g., HMGB1) and surrogate marker (e.g., IL-6 and KC) of lethal sepsis; and 2) suppression of HMGB1-mediated inflammatory responses by preventing clustering of exogenous HMGB1 on macrophage cell surface. Conclusions Taken together, these data suggest a novel mechanism by which the major green tea component, EGCG, protects against lethal endotoxemia and sepsis.


INTRODUCTION
Sepsis is a systemic inflammatory response syndrome resulted from a microbial infection. As a continuum of increasing clinical severity, severe sepsis is defined as sepsis associated with one or more acute organ dysfunctions [1]. Despite recent advances in antibiotic therapy and intensive care, sepsis is still the most common cause of death in the intensive care units, claiming approximately 225,000 victims annually in the U.S. alone. The pathogenesis of sepsis is attributable, at least in part, to dysregulated systemic inflammatory responses characterized by excessive accumulation of various proinflammatory mediators such as interleukin (IL)-1 [2], interferon (IFN)-c [3], nitric oxide [4,5], and macrophage migration inhibitory factor (MIF) [6].
Throughout human history, herbal medicine has formed the basis of folk remedies for various inflammatory ailments. The use of willow bark extract to reduce pain and fever was documented by a Greek physician (Hippocrates) in the 5 th century BC, and the subsequent discovery of salicylic acid as its pain/fever-relief active component gave rise to the first synthetic anti-inflammatory drug, aspirin, and the birth of the pharmaceutical industry. Brewed from the leaves of the plant, Camellia sinensis, tea has been one of the most popular beverages for almost fifty centuries. Its daily consumption (,120 ml/person) is second only to water [24], and has been associated with many important health benefits, such as reduction of risk of oxidative stress and damage [25], atherosclerosis [25], cancer [26], and cardiovascular diseases [27]. These healing properties of green tea are attributable to its abundant polyphenolic compounds known as catechins, such as (-)-epigallocatechin-3-gallate (EGCG), (-)-epicatechin-3-gallate (EG), (-)-epigallocatechin (EGC), and (-)epicatechin (EC). Among them, EGCG accounts for 50-80% of the total catechin, representing approximately 50 mg in a single cup (100 ml) of green tea [28]. However, it was previously unknown if green tea catechins can attenuate endotoxin-induced HMGB1 release or cytokine activities. In this study, we evaluated the capacity of tea catechins in inhibiting endotoxin-induced HMGB1 release and/or cytokine activities, and explored their therapeutic potential in animal model of sepsis.

LPS stimulation
Adherent macrophages or monocytes were gently washed with, and cultured in, serum-free OPTI-MEM I medium two hours before stimulation with bacterial endotoxin (lipopolysaccharide, LPS, E. coli 0111:B4, Sigma-Aldrich). At 16 hours after LPS stimulation, levels of TNF, nitric oxide, and HMGB1 in the culture medium were determined as previously described [8,12].

Animal models of endotoxemia and sepsis
This study was approved and performed in accordance with the guidelines for the care and use of laboratory animals at the Feinstein Institute for Medical Research, Manhasset, New York. Endotoxemia was induced in Balb/C mice (male, 7-8 weeks) by intraperitoneal injection of bacterial endotoxin (LPS, 15 mg/kg) as previously described [7,12,23]. Sepsis was induced in male Balb/C mice (7-8 weeks, 20-25 g) by cecal ligation and puncture (CLP) as previously described [12,23]. EGCG was administered intraperitoneally into mice at indicated doses and time points, and mice were monitored for survival for up to two weeks. In parallel experiments, mice were euthanized to collect blood at 52 h (following two doses of EGCG at +24 and +48 h) after CLP, and assayed for serum levels of TNF, HMGB1, and other cytokines. In other parallel experiments, blood was collected from 3-5 normal healthy mice, or septic mice appearing dying (i.e., in a moribund state, as judged by: 1) unresponsive to external stimuli; 2) inability to maintain upright position; and 3) agonal breathing] or non-dying (i.e., in a non-moribund state, as indicated by: 1) responsive to external stimuli; 2) ability to maintain upright position; and 3) normal breathing] at 52 h post CLP, and serum levels of cytokines were determined.

TNF ELISA
The levels of TNF in the culture medium or serum were determined using commercial enzyme linked immunosorbant assay (ELISA) kits (Catalog no. MTA00, R & D Systems, Minneapolis, MN) with reference to standard curves of purified recombinant TNF at various dilutions as previously described [12,23].

Nitric oxide assay
The levels of nitric oxide in the culture medium were determined indirectly by measuring the NO 22 production with a colorimetric assay based on the Griess reaction [12,23]. NO 22 concentrations were determined with reference to a standard curve generated with sodium nitrite at various dilutions.

HMGB1 Western blotting analysis
The levels of HMGB1 in the culture medium or serum were determined by Western blotting analysis as previously described [7,8,12,23]. The relative band intensity was quantified by using the NIH image 1.59 software to determine HMGB1 levels with reference to standard curves generated with purified HMGB1.

Cytokine antibody array
Murine cytokine antibody array (Cat. No. M0308003, RayBiotech Inc., Norcross, GA, USA), which detects 62 cytokines on one membrane, was used to determine the profile of cytokines in the culture medium or serum as previously described [12]. Briefly, the membranes were sequentially incubated with equal volume of cellconditioned culture medium, or murine serum (after 1:10 dilution), primary biotin-conjugated antibodies, and horseradish peroxidaseconjugated streptavidin. After exposing to X-ray film, the relative signal intensity was determined using the NIH image 1.59 software, and expressed as % of positive controls on the same membrane.

Cell Viability Assays
Cell viability was assessed by trypan blue exclusion assays as previously described [8]. Briefly, trypan blue was added to cell cultures at a final concentration of 0.08%. After incubation for 5 min at 25uC, cell viability was assessed by the percentage of dyeexcluding cells in five 406 microscope fields.

Expression and purification of recombinant HMGB1
The cDNA encoding for rat HMGB1 was cloned onto a pCAL-n vector, and the recombinant plasmid was transformed into E. coli BL21 (DE3) pLysS cells as previously described [7]. Recombinant HMGB1 containing a ,3 kDa calmodulin-binding peptide tag (CBP-HMGB1 fusion protein, 33 kDa) was expressed in E. coli, and purified to remove contaminating endotoxin using polymyxin B column as previously described [7,29,30]. Recombinant HMGB1 preparations were tested routinely for LPS content by the chromogenic Limulus amebocyte lysate assay (Endochrome; Charles River), and endotoxin content was below detection limit (,500 pg endotoxin per microgram of rHMGB1). Recombinant HMGB1 was biotinylated using a Pierce EZ-Link Sulfo-NHS-LC-Biotinylation Kit (Cat. # 21430) following the manufacturer's protocol. The sulfonated NHS esters are cell membrane-impermeable, and are therefore suitable for cell-surface binding/uptake studies. Subsequently, the biotinylated protein was purified by gel filtration chromatography using Sephadex G-25 column.
Fluorescence Immunostaining RAW 264.7 cells were grown to subconfluence, and incubated with biotinylated HMGB1, in the absence or presence of EGCG (10 mM) for various period of time. Subsequently, cells were fixed with 2% formalin for 10 min, and permeabilized with 0.1% Triton X-100 in PBS (1 min, room temperature). After extensive washing with PBS, cells were incubated sequentially with antigen-affinity-purified rabbit anti-HMGB1 antibodies, and goat anti-rabbit secondary antibodies conjugated with green Alexa fluor 488 (Molecular Probes, Eugene, OR). To visualize exogenous HMGB1, cells were co-incubated with streptavidin-conjugated Alexa fluor 594 or Alexa fluor 488 (Molecular Probes). Images were captured using a fluorescence microscope (Carl Zeiss Microimaging).

Streptavidin pull-down assays
Cell lysates were incubated with streptavidin agarose beads (Cat.# 15942-050, Invitrogen) for 1.5 h at 4uC on a rotating platform. After centrifugation, agarose beads were washed six times with 16PBS, and bound proteins were eluted with Laemmili sample buffer (Cat. # 161-0737, Bio-Rad), and analyzed by SDS-PAGE and Western blotting with anti-HMGB1 antibodies.

Statistical Analysis
Data are expressed as mean6SD of two independent experiments in triplicates (n = 2). One-way ANOVA was used for comparison among all different groups. When the ANOVA was significant, post-hoc testing of differences between groups was performed using Tukey's test. The Kaplan-Meier method was used to compare the differences in mortality rates between groups. A P,value less than 0.05 was considered statistically significant.

RESULTS
Tea epigallocatechin gallate (EGCG) dosedependently attenuated endotoxin-induced release of HMGB1, but not nitric oxide We previously discovered that green tea brewed from the leaves of the plant, Camellia sinensis, is effective in inhibiting endotoxininduced HMGB1 release [31]. To determine active components in green tea, we examined its components for HMGB1-inhibiting activities in murine macrophage-like RAW264.7 cells. A major tea catechin, EGCG, dose-dependently abrogated endotoxin-induced HMGB1 release, with an estimated IC 50 ,1.0 mM (Fig. 1A). In contrast, at concentrations that abrogated endotoxin-induced HMGB1 release, EGCG only partially attenuated endotoxininduced TNF secretion (Fig. 1B), but did not inhibit endotoxininduced nitric oxide release (Fig. 1C).
We further confirmed its HMGB1-inhibiting activities using primary murine peritoneal macrophages (MuMACs), as well as human peripheral blood mononuclear cells (huPBMCs). In primary MuMACs, EGCG also abrogated LPS-induced HMGB1 release, but similarly failed to inhibit LPS-induced nitric oxide release ( Fig. 2A). In primary huPBMCs, EGCG effectively abolished LPSinduced HMGB1 release (Fig. 2B), and partly attenuated LPSinduced TNF secretion (Fig. 2B). Taking together, these data suggest that EGCG is capable of effectively inhibiting LPS-induced HMGB1 release in both macrophage and monocyte cultures.

Delayed administration of EGCG still attenuated endotoxin-induced HMGB1 release
As compared with early proinflammatory cytokines (such as TNF), HMGB1 is released late following endotoxin stimulation [7]. It is intriguing to consider whether EGCG could inhibit HMGB1 release if added after LPS stimulation. Whereas concurrent administration of EGCG was most effective in inhibiting LPS-induced HMGB1 release, significant inhibition was still achieved when it was added 2 to 6 h after LPS (Fig. 3). It thus becomes feasible to attenuate lateacting proinflammatory mediators (such as HMGB1) by strategically administering EGCG in a delayed fashion.

Determination of structure-function relationships
As a class of biologically active polyphenols, catechins contain two or more aromatic rings (each carrying at least one aromatic hydroxyl) connected with a carbon bridge (consisting of five carbons and one oxygen, Fig. 4A). To gain insights into the structure-function relationships, we compared the HMGB1-inhibiting activities between EGCG and two relevant molecules: catechin and ethyl gallate (Fig. 4A). Even at concentrations up to 10 mM, catechin or ethyl gallate did not affect LPS-induced HMGB1 release (Fig. 4B), indicating that functional groups of both catechin and gallate are needed for EGCG's HMGB1-inhibiting properties.

EGCG protected mice against lethal endotoxemia
In light of the capacity of EGCG in attenuating LPS-induced HMGB1 release, we explored its efficacy in animal model of lethal endotoxemia. By treating animals with three doses of EGCG at 20.5, +24, and +48 hours post intraperitoneal administration of L.D. 50 dose of LPS, we observed a significant improvement in animal survival rate (from 50% to 76%, P,0.05, Fig. 5A), confirming a previous observation that mixture of tea catechins improved survival rate at 24 hour post onset of endotoxemia [32,33].

EGCG rescues mice from lethal sepsis
Although endotoxemia is useful for investigating the complex cytokine cascades, more clinically relevant animal models are necessary to explore therapeutic agents for the treatment of human sepsis. One well-characterized, standardized animal model of sepsis is induced by CLP. In light of the late and prolonged kinetics of HMGB1 accumulation in experimental sepsis [11], we reasoned that it might be possible to rescue mice from lethal sepsis even if EGCG is administered after the onset of sepsis. The first dose of EGCG was given 24 h after the onset of sepsis, a time point at which mice developed clear signs of sepsis (including lethargy, diarrhea, and piloerection). Repeated administration of EGCG beginning twenty-four hours after the onset of sepsis (followed by additional doses at 48, and 72 hours post sepsis) conferred a dose-dependent protection against lethal sepsis (N = 22-32 mice per group, Fig. 5B), significantly increasing animal survival rate from 53% to 82% (P,0.05), supporting a therapeutic potential for EGCG in the treatment of sepsis.
Although delayed administration of EGCG did not attenuate circulating TNF levels at 52 h after the onset of sepsis (Fig. 6C,  left panel), it did significantly attenuate circulating levels of HMGB1 (Fig. 6C, right panel, P,0.05), suggesting that EGCG confers protection against lethal sepsis partly by attenuating systemic HMGB1 accumulation.

EGCG inhibits HMGB1-induced cytokine release
To elucidate additional mechanisms underlying EGCG-mediated protection, we determined whether EGCG inhibits HMGB1mediated inflammatory response. Indeed, EGCG dose-dependently inhibited HMGB1-induced TNF release in murine macrophage-like RAW 264.7 cells (Fig. 7A, top panel). Despite the fact that EGCG failed to inhibit LPS-induced nitric oxide (Fig. 1C), it dose-dependently suppressed HMGB1-induced release of nitric oxide in RAW 264.7 cells (Fig. 7A, bottom  panel), or primary MuMACs (Fig. 7B), supporting the notion that LPS and HMGB1 use distinct mechanisms to activate innate immune cells [29,37]. Furthermore, EGCG effectively inhibited HMGB1-induced release of IL-6 release, even when it was given 2-4 hours after HMGB1 stimulation (Fig. 7C, and data not shown). Taken together, these data suggest that EGCG confers protection against lethal sepsis partly by inhibiting HMGB1 cytokine activities.
To further elucidate the mechanism by which EGCG attenuates HMGB1-mediated cytokine production, we determined whether  To determine the relative content of exogenous HMGB1, streptvidin-pulled down fraction or whole cell lysate were immunoblotted with HMGB1-specific antibodies (Panel C). Note EGCG dramatically decreased levels of exogenous HMGB1 (indicated by the 33 kDa band corresponding to CBP-HMGB1 fusion protein) (''rHMGB1''). doi:10.1371/journal.pone.0001153.g008 EGCG affects HMGB1-induced self accumulation/clustering on macrophage cell surface. Indeed, in the presence of EGCG (10 mM), HMGB1-induced cell surface clustering, as indicated by Alexa 488-associated cell surface fluorescence, was almost completely eliminated (Fig. 8B), suggesting that EGCG prevents HMGB1 accumulation on macrophage cell surface. To test this possibility, we assayed macrophage whole-cell lysate or streptavidin pull-down fraction for content of exogenous HMGB1 by Western blotting analysis. As expected, at 6 hours following HMGB1 incubation, levels of exogenous HMGB1 in macrophage whole-cell lysate or streptavidin-pulled-down fractions were increased in the presence of exogenous HMGB1 (''+rHMGB1'', Fig. 8C), but dramatically decreased in the presence of EGCG (''+rHMGB1+EGCG'', Fig. 8C).

DISCUSSION
We recently discovered that green tea, brewed from the leaves of the plant, Camellia sinensis, is effective in inhibiting bacterial endotoxin-induced HMGB1 release [31], but the active components responsible for its activities were previously unknown. Here we report that a major component, EGCG, recapitulated HMGB1-inhibiting activities of green tea, and dose-dependently inhibited LPS-induced HMGB1 release in macrophage/monocyte cultures. At concentrations that completely abrogated LPSinduced HMGB1 release, EGCG did not affect LPS-induced nitric oxide release, but only partially attenuated LPS-induced TNF secretion. These data contradict with some reports [39,40], but agree with several other observations [32,33]. The underlying causes for this discrepancy are unknown, and may be partly attributable to EGCG's chemical properties, which can spontaneously dimerize to liberate immunostimulatory products (such as hydrogen peroxide) [10,41].
The mechanism underlying EGCG-mediated suppression of HMGB1 release remains elusive. For instance, it is not known whether EGCG-mediated suppression of HMGB1 release is dependent on its antioxidant activities, because some antioxidants (e.g., catechin, ethyl gallate) failed to inhibit LPS-induced HMGB1 release. Similarly, it is not yet known whether EGCG inhibits LPSinduced HMGB1 release through inhibiting LPS-induced cytoplasmic translocation, or post-translational modification (such as acetylation or phosphorylation). Interestingly, it has been suggested that EGCG can bind to lipid raft-associated cell surface receptor (e.g., the 67 kDa laminin receptor, 67LR) to confer its anti-cancer properties [42,43], or anti-inflammatory allergic response [44,45]. LPS can induce clustering of ligand/receptor complexes (containing hsp70, hsp90, CXCR4, GDF5 and TLR4) within membrane macrodomain (lipid rafts), which transmit signals to active macrophages to produce various proinflammatory mediators [38]. Since receptor clustering-disrupting agents (such as nystatin or MCD) can prevent LPS-induced cytokine production [38], it will thus be interesting to determine whether EGCG inhibits HMGB1 release via similar mechanisms.
Once released, extracellular HMGB1 employs several cell surface receptors (such as TLR2, TLR4, or RAGE) to activate innate immune cells to produce pro-inflammatory cytokines [46][47][48][49]. Indeed, fluorescence resonance energy transfer (FRET) analysis has demonstrated a close physical interaction between HMGB1 and TLR2 or TLR4 on macrophage cell surface within 5-15 minutes of HMGB1 incubation [47], long before subsequent HMGB1-induced cytokine release. Intriguingly, we observed a time-dependent accumulation of exogenous HMGB1 clustering on macrophage cell surface within 2-6 hours of HMGB1 incubation, which correlates with the kinetics of HMGB1-induced release of proinflammatory cytokines [29]. On the other hand, EGCG dose-dependently inhibited cell surface clustering of exogenous HMGB1, and consequently attenuated HMGB1induced release of proinflammatory mediators (e.g., TNF, IL-6, and NO). Consistently, agents (e.g., catechin or ethyl gallate) incapable of inhibiting HMGB1-cell surface clustering uniformly failed to inhibit HMGB1-mediated cytokine production. We thus propose that HMGB1 may induce potential ligand/receptor (e.g., TLR2, TLR4, or RAGE) clustering on macrophage cell surface, which may be a prerequisite for HMGB1-mediated macrophage activation. Given the diverse range of receptors (e.g., TLR2, TLR4, or RAGE) involved in HMGB1 recognition [7,50,51], it is intriguing to investigate whether binding of HMGB1 to different receptors leads to combinational clustering of different receptors (such as TLR2 or TLR4) within the lipid rafts [38,52,53]. Nevertheless, our present study suggests a novel mechanism by which EGCG prevents HMGB1-mediated cytokine productionpotentially by interfering with HMGB1-induced ligand/receptor clustering. Although EGCG-mediated suppression of HMGB1 cell surface clustering may not account for its inhibitory effects on LPS-induced HMGB1 release, it likely underlies its inhibitory effects on cytokine activities of the secreted HMGB1.
In light of the capacity of EGCG in inhibiting LPS-induced HMGB1 release and cytokine activities, we explored its efficacy in animal models of lethal endotoxemia and sepsis (induced by cecal ligation and puncture). Consistent with a previous observation that green tea polyphenols confer protection at 24 h post onset of endotoxemia [32], we found that EGCG promoted significant, and long-lasting protection against lethal endotoxemia. More importantly, delayed and repeated administration of EGCG, beginning at 24 h after onset of sepsis, significantly rescued mice from lethal sepsis, supporting a therapeutic potential of EGCG in the clinical management of human sepsis.
The pathogenesis of lethal sepsis remains obscure, but is mediated in part by excessive release of early (e.g., TNF and IL-1) and late (e.g., HMGB1) proinflammatory cytokines. Appearing relatively early in the circulation, TNF plays a protective role in sepsis [54], and its circulating levels do not correlate with lethality of experimental sepsis [34,35,55]. In contrast, dys-regulated inflammatory response sustained by late-acting mediators (such as HMGB1) may be more pathogenic in lethal sepsis. Because EGCG could inhibit LPS-induced TNF release in vitro, we strategically administered EGCG in a delayed fashion (at 24 h post CLP) to preserve a potentially beneficial early TNF response. Consequently, delayed administration of EGCG did not affect circulating levels of TNF at late stage of sepsis, but specifically attenuated systemic accumulation of HMGB1, as well as IL-6 and KC-two most reliable surrogate markers of lethal sepsis [34,35]. In contrast to HMGB1, IL-6 and KC may not critically important in the pathogenesis of sepsis, because neither anti-IL-6 nor anti-KC antibodies confer long-lasting protection against lethal sepsis [56,57]. Therefore, we propose that EGCG rescues mice from lethal sepsis partly through inhibiting systemic HMGB1 accumulation, as well as HMGB1-induced release IL-6 and KC.
In conclusion, we demonstrated a major tea component, EGCG, recapitulated green tea's HMGB1-inhibiting activities, and dose-dependently abrogated LPS-induced HMGB1 release in macrophage/monocyte cultures. Its beneficial effects in experimental sepsis were partly attributable to: 1) attenuation of systemic accumulation of proinflammatory mediator (e.g., HMGB1) and surrogate markers (e.g., IL-6 and KC) of lethal sepsis; and 2) suppression of HMGB1-mediated inflammatory responses by preventing accumulation of exogenous HMGB1 on macrophage cell surface. The doses of EGCG given to septic mice (4 mg/kg, i.e., 10 mM) are much higher than those readily available in humans (up to 1 mM) after ingestion of 1 cup of green tea 58 . However, concentrated forms of de-caffeinated green tea extracts or purified EGCG are commercially available, and an individual does not need to drink multiple cups of tea everyday to enjoy the health benefits that green tea confers. It will be possible and important to evaluate the therapeutic potential of tea catechins (such as EGCG) for patients with lethal sepsis or other inflammatory diseases in future studies.