Consumption of nutrients rich in hydroxystilbenes has been promoted because of their health benefits, including dampening of inflammatory responses. However, few studies have examined their effects in vivo. Here, we show that the hydroxystilbene oxyresveratrol (trans-2,3′,4,5′-tetrahydroxystilbene: o-RES) blocked hypothermia but caused no significant effect on the febrile response to the immune stimulus, bacterial LPS in rats. This was associated with a reduction in the LPS-induced plasma cytokine, tumor necrosis factor (TNF)-α, but not IL-6. Both IL-6-stimulated STAT-3 and LPS-induced cycoloxygenase-2 expression in the hypothalamus were not affected by o-RES. These data strongly suggest that the o-RES-induced dampening of neuroimmune responses is largely due to its inhibitory effect on TNF-α production. In contrast to in vitro experiments, o-RES has no direct effect on NF-κB signaling pathway in vivo. The specific inhibitory effect of o-RES on TNF-α opens new avenues for the clinical use of o-RES in pathological conditions where excessive production of TNF-α is deleterious.
- nuclear factor-κB
the consumption of nutrients rich in hydroxystilbenes (resveratrol and oxyresveratrol), such as grapes, red wine, peanuts and mulberry wood extracts, has beneficial effects on human health (8, 9, 56). The cellular and the molecular mechanisms that underlie these beneficial effects are largely unknown. Several in vitro studies suggest that resveratrol suppresses the expression of genes involved in the inflammatory response to bacterial and viral stimuli (41, 57). More specifically, resveratrol or its hydroxylated form, oxyresveratrol (o-RES; trans-2,3′,4,5′-tetrahydroxystilbene), reduces the immune activation of the inducible form of cyclooxygenase (COX-2), a key enzyme in the production of inflammatory prostaglandins (31, 38, 51). Such an effect is likely due to the resveratrol-induced inhibition of the NF-κB signaling pathway, a signaling pathway involved in cox-2 gene expression (26, 30).
However, the inflammatory responses in vivo are much more complex than can be modeled in cultured cells in vitro. As a consequence, there are conflicting data on the in vitro and in vivo effects of resveratrol (16), most likely because in vivo immune system activation sets in motion complex and intricate inflammatory responses not seen in vitro. For example, in the best-studied model of innate immune system activation (i.e., a systemic injection of LPS, a component of the outer coat of the gram-negative bacteria), LPS activates resident macrophages in a variety of tissues (17, 28, 45) to release proinflammatory cytokines such as tumor necrosis factor (TNF)-α, IL-1β, and IL-6, as well as anti-inflammatory cytokines and hormones with anti-inflammatory activity (12). A variety of transcription factors such as NF-κB and STAT-3 are consequently activated and lead to the induction of COX-2 in a variety of tissues in the body, including the brain.
One hallmark of innate immune system activation in vivo is fever, an important component of the host defense response against infection (22). At the molecular level, fever develops as a result of COX-2 induction in endothelial cells of the brain vasculature followed by a subsequent production of prostaglandin E2 (PGE2) (5, 32). PGE2 induces fever by acting on neurons located within organum vasculosum of the lateral terminalis (OVLT) and the ventral preoptic region of the hypothalamus (49, 50). In addition to fever, LPS, at higher doses, can also induce hypothermia. It is well established that LPS-induced hypothermia is mediated by the proinflammatory cytokine TNF-α (10, 11, 27, 47).
Hydroxystilbenes are available as nonprescription, complementary medications with largely undocumented claims of efficacy. It is important that the actions of such compounds be subjected to rigorous experimental assessment, so that both the accuracy of claims and the action of these nonproprietary medications are documented. The LPS fever model provides such an opportunity in light of its well-understood and characterized actions. Thus one objective of this study was to determine whether o-RES is able to alter LPS fever via an action on LPS-induced COX-2 in rat OVLT and preoptic area (OVLT/POA). We further explored whether o-RES alters levels of LPS-stimulated, blood-born inflammatory cytokines, such as TNF-α and IL-6 and the level of activation of the transcription factor NF-κB in the liver, one major target for bacterial LPS (28).
MATERIAL AND METHODS
Male Sprague-Dawley rats (Charles River) were individually housed at 22°C under a 12:12-h light-dark cycle (lights on 0700) with pellet chow and water accessible ad libitum. All experimental protocols were approved by the University of Calgary Animal Care Committee and were carried out in accordance with the Canadian Council of Animal Care guidelines.
Body temperature measurements.
Rats were anesthetized with halothane, and silicone-coated temperature data loggers (SubCue Dataloggers, Calgary, Alberta, Canada) were surgically implanted into the abdominal cavity. After 1 day of postoperative analgesia (Butorphanol tartrate, 2 mg/kg im) and a 7- to 9-day recovery period, different groups of rats received intraperitoneal injections of LPS (50 μg/kg ip, 250 μg/kg ip, or 1 mg/kg ip, E. coli, serotype 026:B6, Sigma) dissolved in sterile saline. At the same time, these rats were injected with either o-RES (dissolved in DMSO) at a dose of 20 mg/kg ip or with vehicle (control animals: 100 μl DMSO/100 g rat body wt ip). o-RES was extracted from mulberry wood (Morus alba L.), as previously described (29). The purity of the compound was confirmed by TLC, HPLC, and its melting point (mp = 199–200°C, mpLit. = 201°C), which were consistent with those of the literature (21, 42). All of the experiments described in the present paper were performed using the same batch of o-RES, the purity of which was >98%.
The dose of o-RES was chosen since it exerted the maximal neuroprotective effect on transient rat brain ischemia (1). Body temperature was recorded every 15 min for 8 h starting from 1 h before injections.
Two hours after injection of LPS (250 μg/kg ip), rats were perfused with PBS composed of the following chemicals (in mM): 137 NaCl, 2.7 KCl, 10 Na2HPO4, 1.8 KH2PO4, to remove the blood. This time point was chosen because the peak of NF-κB activity in the hypothalamus is at ∼2 h postimmune challenge (39) The brain region that consists of the OVLT/POA was cut in a trianglular shape where the upper angle is located at the midpoint of the anterior commissure, and the base is delimited by the cortex invaginations (with the tip of the piriform cortex excluded) (36). Both the OVLT/POA and a sample of liver [a source of circulating cytokines (3, 28)] were quickly dissected and put in lysis buffer [MOPS: 20 mM Mg (C2H3O2)2: 4.5 mM KCl: 150 mM, 1% Triton X-100] supplemented with a mixture of protease and phosphatase inhibitors as previously described (37).
Western blot analysis.
Western blot was performed as previously described (37) with slight modifications. Briefly, protein extracts from each individual rat were separated using 10% SDS-PAGE electrophoresis, then transferred onto a nitrocellulose membrane. The membranes were then incubated overnight at 4°C with one of these primary rabbit antibodies: anti-phospho-IκB (Santa Cruz Biotechnology, 1:500), anti-phospho-STAT-3 (pSTAT-3; Cell Signaling Technology, 1:2,000), anti-COX-2 (Cayman Chemical, 1:2,000). After several washes, the membranes were incubated for 1 h at room temperature with goat anti-rabbit antibody conjugated with horseradish peroxidase (Santa Cruz Biotechnology, 1:4,000). A chemiluminescence substrate was applied to the membrane (ECL Kit, Amersham Biosciences, Arlington, IL), and protein bands were visualized using Kodak X-Omat film (Eastman Kodak, Rochester, NY).
After protein detection, membranes were stripped with β-mercapto-ethanol (BDH) and reblotted with rabbit anti-IκB (Santa Cruz Biotechnology, 1:4,000), rabbit anti-STAT-3 (Santa Cruz Biotechnology, 1:5,000) or with rabbit anti-actin (Sigma, 1:20,000) antibodies as appropriate (see results) and processed as described above.
With rats under halothane anesthesia, a heparin-filled, silicone cannula was inserted into the jugular vein and externalized to the back of the neck. Four days later, the jugular vein cannula of each rat was connected to a longer cannula equipped with a swivel. This procedure allows for free movement of the rat in its home cage, while multiple blood samples (1 ml per time point) were collected as described before (14). Blood samples were collected in heparinized tubes before and at several time points after intraperitoneal injection of either LPS/o-RES [(250 μg/20 mg)/kg] or LPS/vehicle (250 μg/kg). Blood was centrifuged and plasma was snap-frozen, then stored at −80°C until cytokines were assayed.
Plasma levels of TNF-α and IL-6 were assayed using a specific rat ELISA kit (catalog # KRC3012 and KRC0062, respectively; Biosource International, Camarillo, CA). The minimum detectable concentration is 4 pg/ml for TNF-α and 8 pg/ml for IL-6.
Data collection and analysis.
Fever data were calculated as a thermal index (area under the curve) between 1 and 7 h after LPS challenge (a time after the stress-induced hyperthermia had subsided). The thermal index was calculated as follows: For each individual rat, the temperature values recorded every 15 min were subtracted from baseline temperature (average temperature during 1 h before injection), and the integrated values were summed to provide the area under the curve (thermal index). Thermal indices of the early phase that corresponds to the hypothermia (1 to 2.5 h post-LPS injection) and the later phase (2.75 to 7 h post-LPS injection) were separated, calculated, and statistically compared. The thermal indices of both LPS-oRES-injected and LPS-vehicle-injected rats were compared using an unpaired Student's t-test.
For Western blot analysis, the area and intensity of a given band was quantified using Quantity One quantitation software (Bio-Rad). The ratios of optical density values of either phospho-protein/protein or protein/actin were calculated. Immunoblot data were compared using ANOVA followed by Student-Newman-Keuls post hoc comparisons whenever possible.
Plasma cytokine levels were compared using repeated measures ANOVA followed by Student-Newman-Keuls post hoc comparisons whenever possible. The significance was accepted at P < 0.05.
o-RES effects on fever response.
o-RES (20 mg/kg) had no effect of its own on body temperature (Fig. 1C, a; o-RES/saline). All injections elicited the typical stress-induced hyperthermia due to the handling of the rat during the injection. In response to a febrile dose of LPS (50 μg/kg ip), this was immediately followed by a hypothermia that reached its lowest level at ∼1.5 h postinjection (Fig. 1A). The febrile response to LPS ensued after the hypothermic response and reached its maximum level at ∼3 h after LPS injection. o-RES (20 mg/kg ip) given concurrently with LPS blocked the LPS-induced hypothermia (Fig. 1A) but did not significantly affect LPS fever. The dose of o-RES was chosen as it exerted the maximal neuroprotective effect on transient rat brain ischemia (1). o-RES also blocked hypothermia induced by a relatively high dose of LPS (250 μg/kg) (Fig. 1B) but was ineffective on LPS-induced hypothermia resulting from a very high dose of LPS (1 mg/kg ip) (Fig. 1C).
o-RES effect on COX-2.
Because hypothalamic COX-2 plays a crucial role in the inflammatory response to LPS, we tested whether o-RES affects LPS-induced COX-2 expression in the OVLT/POA region of the hypothalamus. Immunoblots revealed a COX-2-immunoreactive band at the expected molecular weight (72 kDa). COX-2 was expressed at a low level in the OVLT/POA of rats injected simultaneously with o-RES and pyrogen-free saline. LPS alone induced a two-fold increase in the expression levels of COX-2 (P < 0.05). This LPS-induced increase in COX-2 was not affected by o-RES (Fig. 2, A and B).
o-RES effects on LPS-stimulated TNF-α and IL-6 secretion.
Because hypothermia is largely due to an action of TNF-α (27) and IL-6 is an essential cytokine in the fever response (6, 7), we explored whether the suppressive effects of o-RES on LPS-induced changes in body temperature (i.e., hypothermia and fever) were also associated with changes in circulating levels of TNF-α and IL-6. Plasma levels of TNF-α were measured before injection (time 0), 1 h, and 2 h after LPS injection (250 μg/kg ip). As seen in Fig. 3A, LPS induced a rapid increase in the plasma levels of TNF-α that reached a maximal level 1 h postinjection. o-RES significantly dampened this LPS-stimulated TNF-α production at both 1 h and 2 hrs postinjection. Plasma levels of IL-6 increased progressively after LPS administration (Fig. 3B). In contrast to the TNF-α response, LPS-induced IL-6 levels were not affected by o-RES.
o-RES effect on NF-κB signaling pathway.
As generation of cytokines by LPS involves the NF-κB signaling pathway, we asked whether o-RES reduced NF-κB activation in the liver, one of the major targets for bacterial LPS (28). As a well-accepted method of evaluating NF-κB signaling pathway activation (13, 20), we quantified phosphorylation of its binding protein IκBα. As expected, LPS induced phosphorylation of IκBα (pIκBα; molecular mass ∼37–38 kDa), but this was not affected by o-RES. (Fig. 4).
o-RES effect on STAT-3 signaling pathway.
Because LPS induced IL-6 and the STAT-3 signaling pathway is specifically activated by IL-6 (19), we tested whether o-RES can affect IL-6-induced pSTAT-3 (activated STAT-3). The immunoreactive band for pSTAT-3 was detected at the expected molecular weight (86–88 kDa). pSTAT-3 was undetectable in controls (o-RES/saline) in both the OVLT/POA (Fig. 5A) and the liver (Fig. 5B). LPS administration (250 μg/kg ip) induced pSTAT-3 in both the OVLT/POA and the liver. These LPS effects were not altered by o-RES.
A number of novel and intriguing findings emerge from this study. o-RES blocked specifically LPS-induced hypothermia but had no significant effect on LPS fever. In addition, o-RES specifically reduced the TNF-α response to immune challenge with no noticeable effect on either IL-6 production or IL-6-activated STAT-3 signaling pathway. Interestingly, o-RES did not alter the immune activation of NF-κB transcription factor in the liver. Finally, despite its possible inhibitory effect on TNF-α release, o-RES did not affect the COX-2 induction in the OVLT/POA.
Experimental studies have now established that the response to LPS consists of a sequence of thermoregulatory changes involving different mediators and different physiological processes (46). As is evident from our data, an initial response, particularly to higher doses of LPS, is hypothermia. Humans can become hypothermic when their immune response is activated by persistent, high titers of viral or bacterial load (24, 25, 33). Usually, the clinical outcome is worse in the hypothermic patient and appropriate measures, such as the use of antibiotic or, antiviral drugs, are taken to reduce hypothermia (25, 43). In the current study, we have demonstrated for the first time a blockade of the hypothermic response by o-RES. This observation suggests that o-RES could be clinically relevant to use for hypothermia that can occur as a result of postoperative infection or surgical complications. Interestingly, o-RES did not significantly affect the febrile response to LPS. This observation is particularly interesting, given that the fever associated with gram-negative infection is thought to be beneficial (22). Thus the potential exists for an o-RES-based therapy in infectious disease that preserves beneficial aspects of host defense (i.e., fever) while mitigating other possibly more deleterious aspects.
It is well accepted that TNF-α is the key mediator of LPS-induced hypothermia (27, 54). In this study, we showed that the suppressive action of o-RES on LPS hypothermia is very likely through an inhibition of the LPS-induced TNF-α. It is noteworthy that the o-RES action is cytokine specific, as the LPS induction of another proinflammatory cytokine, IL-6, was not affected by o-RES. To our knowledge, this is the first in vivo study to show that o-RES targets specifically TNF-α (and not IL-6) production. However, the observation of a selective inhibition of TNF-α production in vivo is in line with the recently published in vitro study where resveratrol, another hydroxyostilbene, inhibits TNF-α but not IL-6 release in immune-challenged human peripheral blood leukocytes (44). How o-RES alters specifically LPS-induced production of TNF-α, while sparing IL-6, is not yet known. o-RES may target specifically LPS-activated TNF-α production at either transcription, translation, or secretion phases. It is unlikely that o-RES affects TNF-α transcription via the NF-κB pathway as this pathway is not altered in LPS-stimulated liver (a target tissue for LPS and a major site for both NF-κB-dependent TNF-α and IL-6 production).
Because of the deleterious effect of exacerbated TNF-α production during different pathological conditions, therapeutic efforts have been directed toward the development of new TNF-α blockers (23, 48, 55). However, these TNF-α blockers are not without side effects (40). Our observation that o-RES can specifically reduce TNF-α levels in rats may represent an alternative therapeutic strategy if similar actions exist in humans.
In vitro studies showed that resveratrol and o-RES inhibit the expression of several genes involved in inflammatory responses by altering the NF-κB signaling pathways (26, 30, 41, 51, 52). In this in vivo study, we showed that o-RES had no significant effect on the LPS-activated NF-κB pathway in the liver. Similarly, recent published results showed that the dehydroxylated form of o-RES; resveratrol, also had no significant effect on NF-κB signaling pathway in lung tissue (2). Thus it appears that both resveratrol and o-RES possesses no noticeable effect on the immune-activated NF-κB pathway in peripheral tissues. We also noticed no apparent effect on LPS-induced COX-2 in the POA/OVLT. As NF-κB is a major modulator for transcription of the cox-2 gene (52), this again supports our contention that o-RES in vivo does not target NF-κB at least so as to reduce the inducible COX-2 involved in PGE2 production and the eventual fever phase of the inflammatory response. As we and others have previously shown that alterations in COX-2 induction correlate very well with brain levels of PGE2 (15, 35), this suggests that the hypothermic response to LPS is independent of COX-2-generated PGE2. The fact that o-RES affects LPS-induced hypothermia, but not LPS-induced COX-2, suggests that the hypothermia is dissociated from induction of COX-2. However, we cannot exclude the possibility that o-RES may alter LPS-induced hypothermia via an effect on PGE2 production independently of de novo synthesis of COX-2. Indeed, o-RES can cross the blood-brain barrier (4) and directly alter PGE2 production, likely by scavenging reactive oxygen species that are necessary for the activity of constitutively expressed COX-1 (53). Alternatively, o-RES might also promote the production of other antipyretic prostaglandins such as prostaglandin J2 (18, 34).
The o-RES effect on both the activity of NF-κB signaling pathway and COX-2 levels were measured 2 h post-LPS injection, a time where the hypothermic response was maximal (Fig. 1B). We cannot discard the possibility that o-RES might inhibit LPS-activated NF-κB signaling pathways at earlier time points.
High circulating levels of TNF-α are well known to be associated with hypothermia. In this in vivo study, we showed that o-RES specifically reduced circulating levels of TNF-α and that the reduced TNF-α levels explain all of the observed actions of o-RES. A major dichotomy exists between our in vivo findings and previously published in vitro work in cell lines, in that we see no effects of o-RES on LPS-induced NF-κB production. Future in vitro studies might be profitably directed at ascertaining how o-RES interferes with TNF-α production independent of NF-κB signaling. These mechanistic explorations of o-RES actions on immune response may open the door for a clinical use of o-RES compound in pathological conditions where TNF-α production is deleterious.
This study was supported by the Canadian Institutes of Health Research. Q. J. Pittman is an Alberta Heritage Foundation for Medical Research Medical Scientist.
We thank Mio Tsutsui for technical help and Dr. Peter Lorenz (Institute for Medical Neurobiology) and Otto-von-Guericke (University Magdeburg, Magdeburg, Germany) for their help in extracting and providing o-RES and Dr. Robert Newton for the critical reading of the manuscript.
Present address of T. F. Horn: BD Biosciences, Binningerstr. 94, CH-4123 Allschwil, Switzerland.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society