Interleukin (IL)-6 is an important humoral mediator of fever following infection and inflammation and satisfies a number of criteria for a circulating pyrogen. However, evidence supporting such a role is diminished by the moderate or even absent ability of the recombinant protein to induce fever and activate the cyclooxygenase-2 (COX-2) pathway in the brain, a prerequisite step in the initiation and maintenance of fever. In the present study, we investigated the role of endogenous circulating IL-6 in a rodent model of localized inflammation, by neutralizing its action using a specific antiserum (IL-6AS). Rats were injected with LPS (100 μg/kg) or saline into a preformed air pouch in combination with an intraperitoneal injection of either normal sheep serum or IL-6AS (1.8 ml/rat). LPS induced a febrile response, which was accompanied by a significant rise in plasma IL-6 and nuclear STAT3 translocation in endothelial cells throughout the brain 2 h after treatment, including areas surrounding the sensory circumventricular organs and the median preoptic area (MnPO), important regions in mediating fever. These responses were abolished in the presence of the IL-6AS, which also significantly inhibited the LPS-induced upregulation of mRNA expression or immunoreactivity (IR) of the inducible form of COX, the rate-limiting enzyme for PGE2-synthesis. Interestingly, nuclear signal transducer and activator of transcription (STAT)3-positive cells colocalized with COX-2-IR, signifying that IL-6-activated cells are directly involved in PGE2 production. These observations suggest that IL-6 is an important circulating pyrogen that activates the COX-2-pathway in cerebral microvasculature, most likely through a STAT3-dependent pathway.
- local inflammation
- nuclear factor-κB
- circumventricular organs
fever is a brain-regulated sickness response, which can be triggered by a multitude of peripheral signaling pathways (46), including circulating endogenous pyrogens (15). These humoral mediators, which belong primarily to the cytokine family, are induced in the periphery following systemic infection or inflammation, and gain entry to the brain where they act on specific receptors. A number of proinflammatory cytokines have been shown to fit the criteria proposed for circulating pyrogens (26), such as IL-1, IL-6, and TNF-α. Of these, only IL-6 was demonstrated to consistently increase in the circulation of febrile animals (8, 10, 38) and humans (42) in a manner that correlates significantly with the magnitude and duration of the fever response to exogenous pathogens, such as LPS (34, 47, 49). Furthermore, absence of endogenous IL-6 activity following neutralization with a specific antiserum (10) or in IL-6-deficient mice (11, 28) demonstrated unequivocally that IL-6 is essential for the fever response following infection. Conversely, administration of recombinant IL-6 in different species of rodents and using different routes of administration resulted in rather moderate (3, 16, 22, 23, 52) or absent (10, 34, 59) increases in body temperature.
These apparent discrepancies are further confounded by observations, showing that exogenously systemically administered IL-6, unlike IL-1β, does not trigger PGE production in the brain as measured by the induction of cyclooxygenase (COX)-2 (30, 58), the rate-limiting enzyme in PGE2 synthesis (35, 66). The induction of COX-2 and PGs in the brain represents a critical step in the generation of fever and is an important criterion for the definition of a circulating pyrogen. The major intracellular signaling pathway for COX-2 (40, 56) and fever (29, 33, 59) induction is believed to be NF-κB. Unlike IL-1β, which signals through this transcription factor, IL-6 triggers a different signaling pathway, namely JAK-signal transducer and activator of transcription (STAT)3 (2). Recent evidence, however, suggests that COX-2 induction, at least in peripheral tissue, such as the myocardium, can indeed be activated through this pathway (63). In the brain, studies have now shown that systemically administered IL-6 can induce nuclear STAT3 translocation in structures implicated in the fever response (22, 51), namely endothelial cells lining the vasculature within the blood-brain barrier (BBB) and the sensory circumventricular organs (sCVOs), which lack an intact BBB. The same studies, however, did not address the precise functional consequence of IL-6-mediated genomic activation of these structures, including whether this humoral mediator can trigger PG synthesis.
The aim of the present study, therefore, was to address the question of whether circulating IL-6 can trigger the brain mechanisms associated with the fever response by assessing whether it induces COX-2 expression in the brain. For this, we targeted the contribution of the endogenous circulating protein by using a specific neutralizing antiserum in rodents treated with LPS, a potent pyrogen and activator of systemic IL-6 and COX-2 in the brain.
MATERIALS AND METHODS
Adult male Sprague-Dawley rats (Charles River, Saint Constant, Canada) with a body weight of 290–320 g on the day of experiment were used in all studies. Animals were housed individually in a controlled environment at an ambient temperature of 21 ± 2°C on 12:12-h light-dark cycle (lights off at 7:00 PM) with free access to standard laboratory chow and water. All animals were handled extensively for at least 2 days before each experiment for habituation. The Animal Care Committee of McGill University pursuant to the Canadian Council of Animal Care guidelines approved all experimental procedures.
Surgery and Measurement of Body Temperature
Rats were anesthetized with ketamine/xylazine/acepromazine (50, 5, and 0.5 mg/kg, respectively) via intramuscular injection and implanted with intra-abdominal temperature-sensitive radiotransmitters. During the same procedure, each animal received 20 ml of sterile air injected into the subcutaneous tissue of the dorsal midline at the level of the scapulae to form an “air pouch” for administering the inflammatory agent, as previously described (10, 17). Three days before the experiment, animals were briefly reanesthetized (3% isoflurane in oxygen) until they demonstrated complete areflexia and the air pouches reinflated with 10 ml of sterile air to maintain the cavities. On day 6 following the initial surgery, injections were performed between 1:00 and 2:00 PM directly into the subcutaneous air pouches in lightly restrained (hand-held), conscious animals. Changes in core body temperature were monitored in these animals using remote biotelemetry (Data Sciences, St. Paul, MN), as described earlier (9).
Treatment and Experimental Protocols
All animals including controls were pretreated (−4 h) either intraperitoneally with 1.8 ml normal sheep serum (NSS: n = 19; Sigma) or 1.8 ml of sheep anti-rat IL-6 serum [n = 5; IL-6 antiserum (AS); National Institute for Biological Standards and Control (NIBSC) Potters Bar, UK]. The IL-6AS serum recognizes both recombinant (Escherichia coli-derived) and natural rat IL-6, but does not cross react with rat recombinant (rr)IL-1α, IL-1β, or TNF-α (45). In a previous study, we reported that pretreatment with the same IL-6AS had no effect on body temperature in saline-treated animals (10). Some animals (n = 10) were injected subcutaneously with bacterial LPS (derived from E. coli, suspended in sterile pyrogen-free 0.9% saline; 0111:B4, Lot 42k4120; L-2630; Sigma) at 100 μg/kg (prepared as 100 μg/ml solution). This dose of LPS was previously shown to induce fever and to significantly increase circulating IL-6 concentrations in rats (9). Another group of rats (n = 4) received an intraperitoneal injection of rrIL-6 (E. coli-derived, nonglycosylated, specific activity 250,000 IU/μg as measured in a B9 mouse hybridoma bioassay; NIBSC) at a dose of 45 μg/kg and a volume of 1 ml/kg diluted in sterile 0.9% saline. Equivalent volumes of the vehicle (0.9% saline, 1 ml/kg, each n = 5) were injected subcutaneously or intraperitoneally, respectively, in control groups. In all cases, core body temperature was continuously monitored for 12 h before treatment and for 2 h after treatment. The final time point was based on our earlier studies showing that LPS-induced (100 μg/kg) plasma IL-6 levels in the air pouch model peak at 2 h posttreatment (10). At this time, animals were euthanized by using terminal anesthesia (ketamine/xylazine/acepromazine; 100, 10, and 1 mg/kg ip, respectively), blood samples were collected via cardiac puncture, and lavage fluid was collected from each air pouch by injecting, then quickly withdrawing 1 ml saline using a sterile syringe. Following the fluid collection, rats were transcardially perfused with 150 ml ice-cold 0.9% NaCl solution; their brains were quickly removed, frozen in powdered dry ice, and kept at −80°C until analysis. The brains of all animals were prepared for immunohistochemistry as previously described (51).
To determine IL-6 concentration, heparinized (10 IU/ml) blood and lavage samples were centrifuged (5,300 g, 10 min, 4°C), the supernatant was collected and stored at −80°C until assays were performed. IL-6 levels were measured in duplicates using a two-site, rat-specific ELISA (NIBSC) as described previously (45).
Coronal cryostat sections (20 μm, prepared with a model CM 3050 S cryostat; Leica Microsystems, Nussloch, Germany) were prepared from two brain areas encompassing the forebrain [including the vascular organ of the lamina terminalis (OVLT), the subfornical organ (SFO), and the median eminence (ME); bregma 0.5 to −3.5 mm] and the brain stem [area postrema (AP) and surrounding tissue, bregma −14.28 to −12.28 mm] according to the atlas of Paxinos and Watson (44). To perform RT-PCR and immunohistochemistry from the same brain, every third consecutive frozen section was thaw-mounted on poly-l-lysine-coated glass slides for immunohistochemistry and stored at −80°C until processing. The remainder of the serial whole brain sections were collected for RNA extraction and stored in a cryotube at −80°C for mRNA extraction. Because IL-6 was recently shown to activate endothelial cells throughout the entire brain, including the cortex (51), whole brain sections were used for RT-PCR analyses.
Total RNA was extracted using Trizol (Invitrogen, Burlington, ON, Canada) according to the manufacturer's instructions. The first-strand cDNA was synthesized from 1 μg total RNA using 200 units of Molony murine leukemia virus RT (Invitrogen), 5 μM of random hexamers (Applied Biosciences, Streetsville, ON, Canada), and 1 mM of deoxyribonucleotide triphosphate mix (Sigma) in a total reaction volume of 20 μl. Following RT, the product was stored at −20°C. The cDNA product (0.9 μl) was added to 15 μl PCR reaction mix containing ReadyMix Taq PCR (Sigma) and 6 pmol of gene-specific primer sets for COX-2 and suppressor of cytokine signaling 3 (SOCS3) and β-actin. In each case, amplification was performed with a thermal cycler (GeneAmp PCR System 9700; Applied Biosciences) using the following cycling parameters: 1) denaturing, 95°C for 5 min; 2) amplification cycle, 95°C for 30 s, annealing temperature for 30 s and 72°C for 1 min; and 3) final extension, 72°C for 10 min. Primers were designed to span a sequence derived from different exons (separated by an intron in genomic DNA sequence) to minimize amplification from non-mRNA-derived templates. Inappropriate amplification from genomic DNA was negligible when amplification was performed with a template without RT. The gene accession numbers, primer sequences, annealing temperatures, and cycle numbers used are listed as follows: COX-2 (NM_017232) sense 5′-TGATAGGAGAGACGATCAAGA-3′, anti-sense 5′-ATGGTAGAGGGCTTTCAACT-3′, 57°C, 32 cycles; SOCS3 (NM_053565) sense 5′-CCAGCGCCACTTCTTCAC-3′, anti-sense 5′-GTGGAGCATCATACTGGTCC-3′, 60°C, 40 cycles; β-actin (NM_031144) sense 5′-GCCGTCTTCCCCTCCATCGTG-3′, anti-sense 5′-TACGACCAGAGGCATACAGGGACAAC-3′, 60°C, 20 cycles. PCR products were separated by 2% agarose gel electrophoresis and visualized with ethidium bromide staining. The band density was obtained for quantification using GeneTool image analysis software (Syngene, Frederick, MD). To normalize the expression level of genes between different samples, the levels were estimated as the ratio of geneX/β-actin. In a pilot experiment, the amount of PCR product (on a log scale) vs. the number of cycles was plotted, and the linear range of template amplification was determined for two samples from each treatment group. The cycle numbers were determined to be within the exponential phase of amplification for all treatment groups.
On the day of the experiment, the frozen brain sections were air dried at room temperature for 10 min and then immersion fixed in 2% paraformaldehyde diluted in 0.1 M PBS (pH 7.4) for 10 min at room temperature. After three successive washes in PBS, the sections were treated for 1 h at room temperature with a blocking solution consisting of PBS containing 10% normal donkey serum (NDS; Chemicon International, Temecula, CA) and depending on the type of applied antibodies, 0.1% or 0.25% Triton X-100. Details of the primary antibody concentrations as well as fixation and blocking procedures are shown in Table 1.
The sections were incubated with the primary antibodies for 36–48 h at 4°C followed by visualization with CY3-conjugated anti-rabbit IgG (for STAT3 and COX-2; 1:500 dilution; cat. no. 711–165-152) or CY3-conjugated anti-goat IgG (for NF-κB; 1:500; cat. no. 705–165-147). All secondary antibodies were raised in donkey and all were multiple labeling-grade (Jackson ImmunoReaserch, West Grove, PA). Sections were counterstained with the nuclear 4,6-diamidino-2-phenylindole (DAPI) stain (Molecular Probes, Eugene, OR) for 15 min, using a 1:4,000 dilution in PBS to assess the nuclear localization of the STAT3 signals and to give better visualization of the tissue surrounding the signal. Finally, all sections were cover slipped using an equal mix of glycerol/PBS and stored at 4°C until microscopic analysis was performed.
A rabbit anti-STAT3 antibody was used to detect STAT3 signals in rat brain. This antibody was raised to detect both the phosphorylated (activated) and the nonphosphorylated (constitutive) forms of the transcription factor and was previously demonstrated to be specific for both (25). In some studies (colocalization), a specific antibody (rabbit polyclonal anti-phopho-STAT3 antibody) was used for the detection of the phosphorylated (P) form of STAT3.
The rabbit anti-COX-2 antibody was raised against a peptide of murine COX-2. The antibody shows specific staining of this enzyme typically at a perinuclear subcellular localization.
As for STAT3, the goat anti-NF-κB antibody used in our studies recognizes both cytoplasmic (inactive) and nuclear (active) forms of the peptide. Specifically, this antibody targets the inactive p65 subunit in the cytoplasm (bound to p50 and inhibitor κB), as well as the active monomeric or dimeric form in the nucleus.
Specificity of each antibody was shown previously [STAT3, (25); COX-2, (64); NF-κB, (39)] and was confirmed by preabsorption with respective blocking peptides (data shown only for NF-κB, see ⇓⇓⇓⇓⇓⇓⇓Fig. 8, O–P).
In studies involving double immunostaining for PSTAT + COX-2 and PSTAT + NF-κB, the rabbit anti-PSTAT antibody (see Table 1) was mixed with a goat COX-2 antibody (dilution 1:2,000, raised against the COOH terminus of murine COX-2; cat. no. sc-1747; Santa Cruz Biotechnology, Santa Cruz, CA) or the goat anti-NF-κB antibody (see Table 1), respectively. Primary antibodies were visualized with CY3-conjugated anti-rabbit IgG (for PSTAT, 1:500 dilution; cat. no. 711–165-152) and FITC-conjugated anti-goat IgG (for COX-2 and NF-κB; 1:500 dilution). The specificity of the COX-2 staining was confirmed by preincubation of the antibody with the specific blocking peptide (10-fold excess; cat. no. sc-1747 P, Santa Cruz Biotechnology). In control experiments, the sections were incubated with the primary antibody mixture in which one of the primary antibodies was substituted with nonimmunized animal IgG (rabbit IgG for STAT3, PSTAT, and COX-2). Inappropriate cross-reaction between antibodies was confirmed to be negligible.
To identify endothelial cells, double immunohistochemistry was performed using the specific cell marker antibodies against von Willebrand (vW) factor. The antibodies for STAT3, PSTAT, or COX-2 were mixed with a sheep anti-rat vW factor antibody (1:3,000; cat. no. SARTW-IG; Affinity Biologicals, Ancaster, ON, Canada), and brain sections were incubated with the primary antibodies at 4°C for 36 to 48 h. STAT3, PSTAT, and COX-2 were visualized with CY3-conjugated anti-rabbit IgG (1:500 dilution) and vW with FITC-conjugated anti-sheep IgG (1:500 dilution; cat. no. 713–095-147) as secondary antibodies.
Temperature data were analyzed using a two-way repeated measures ANOVA followed by Newman-Keuls multiple comparisons post hoc test to compare treatment groups and controls. Circulating levels of IL-6 and the relative density of SOCS3 and COX-2 expression in the forebrain and the brain stem were compared separately between the experimental groups (LPS, LPS + IL-6AS) by one-way ANOVA followed by Newman-Keuls multiple comparisons post hoc test. A parametric unpaired two-tailed Student’s t-test was used to compare IL-6 and vehicle-treated groups. We also investigated potential correlations between individual values of plasma IL-6 and the respective relative density of mRNA signals in the brain. Each brain area (forebrain, brain stem) was evaluated separately, with an n = 9 (IL-6 vs. NaCl; n = 4/5), respectively. Datasim software (Desktop Press, Lewiston, ME) was used for two-way ANOVA analysis and GraphPad Prism (GraphPad Software, San Diego, CA) was used for the other statistical analyses. P < 5% was deemed significant. All data are presented as means ± SE.
IL-6-AS Significantly Inhibits LPS-Induced Fever and Circulating IL-6 Levels
LPS (100 μg/kg) induced a significant increase (90–120 min, P < 0.005 vs. saline) in body temperature, which peaked at 110 min after injection (LPS sc, NSS ip; 38.19 ± 0.13°C; see Fig. 1A). IL-6AS (ip) pretreatment completely abolished this response. The mean IL-6 levels in the blood and lavage were 0.16 and 0.15 ng/ml, respectively, in saline-treated animals. LPS treatment induced a 50-fold increase of IL-6 in the lavage and a threefold increase in plasma IL-6 (lavage: 3.36 ± 0.69 ng/ml; plasma: 0.53 ± 0.10 ng/ml; see Fig. 1, B and C). In both cases, this was significantly attenuated in the presence of IL-6AS (P = 0.0172 for plasma; P < 0.0001 for the lavage).
Exogenous administration of rat recombinant IL-6 (45 μg/kg ip) induced a significant rise (40–50 and 70–80 min, P < 0.05 vs. saline) in body temperature, which peaked 40 min after treatment (IL-6 ip, NSS ip; 37.99 ± 0.17°C; see Fig. 2A). rrIL-6 treatment was accompanied with a fourfold significant rise of circulating IL-6 2 h after injection (0.57 ± 0.13 ng/ml, P = 0.0081; Fig. 2B).
IL-6-AS Significantly Inhibited LPS-Induced Nuclear STAT3 Translocation and SOCS3 mRNA Expression in Sensory CVOs and Brain Endothelial Cells
LPS induced a dramatic increase in nuclear STAT3 translocation in the brain 2 h after treatment (Table 2). Double immunohistochemical studies revealed that the majority of this staining was localized in vW factor positive endothelial cells throughout the entire brain (see Fig. 5M for example), including fever-relevant areas, such as the MnPO (data not shown). One exception was observed in the sCVOs in which nuclear STAT3-positive IR was detected in other cell types in addition to the endothelial cells (OVLT, SFO, AP; Fig. 3, B, E, and H). This response was almost totally abolished in the presence of IL-6AS in the forebrain and the brain stem (data not shown), including all three structures of the sCVOs (Fig. 3, C, F, and I), although a few remnant STAT3 nuclear-positive cells could still be seen in the SFO (Fig. 3F). Saline-treated controls (Fig. 3, A, D, and G) only exhibited cytoplasmic (inactive) STAT3.
To further assess the contribution of IL-6 to the activation of this pathway by LPS, we measured the change in expression of SOCS3. This is a major end product of the STAT3 signaling pathway, which is involved in autoregulatory feedback for activated STAT3 and is broadly used as a marker for STAT3 activation (32). Changes in the levels of SOCS3 were measured by semiquantitative RT-PCR on mRNA obtained from two different regions of the brain: forebrain (including the OVLT, SFO, and ME; bregma 0.5 to −3.5 mm) and the brain stem (including the AP and surrounding tissue, bregma −14.28 to −12.28 mm). LPS treatment significantly increased SOCS3 mRNA levels in both of these regions, a change that was significantly inhibited by IL-6AS (Fig. 4A, forebrain, P = 0.0053; Fig. 4B, brain stem, P = 0.0002).
In a separate experiment, systemic (45 μg/kg ip) rrIL-6 treatment alone induced STAT3 translocation in endothelial cells (data not shown) throughout the entire brain (Table 2) and SOCS3 mRNA expression both in the forebrain (Fig. 4C, P = 0.0141) and in the brain stem (Fig. 4D, P = 0.0019).
IL-6-AS significantly inhibited LPS-induced COX-2 mRNA expression in the brain and COX-2-IR within brain endothelial cells.
Constitutive, neuronal COX-2-IR, was detected in the brains taken from all treatment groups (Fig. 5, A–D), as previously reported by others (5, 7, 27). LPS treatment clearly induced COX-2-IR in blood vessels throughout the entire brain, including the cortex (see Fig. 5B for an example) and structures implicated in fever, such as the MnPO (Fig. 5F) or the sCVOs (see Fig. 5J, SFO for example). For practical reasons, only examples for this broad and consistent effect throughout the brain are shown. Double immunolabeling with vW factor-IR indicated that this staining was localized to endothelial cells (Fig. 5, B, F, and J). Coadministration of IL-6AS (Fig. 5, C, G, and K) almost completely abolished this response to a comparable level seen in saline-treated animals (Fig. 5, A, E, and I).
In a separate experiment, exogenous administration of rrIL-6 (45 μg/kg ip) was followed by a similar induction of COX-2 in the brain. COX-2-IR could be clearly seen throughout the brain (Fig. 5, D, H, and L), including areas in close proximity to the sCVOs (Fig. 5L, SFO, for example) and the MnPO (Fig. 5H). Colocalization studies with vW factor-IR indicated that this was specific to endothelial cells (Fig. 6, D, H, and L). In general, control saline-treated animals did not show significant induction of COX-2-IR; however, in some cases, single COX-2-IR positive cells were detected in endothelial cells, most likely reflecting constitutive expression of this enzyme (data not shown).
The levels of COX-2 mRNA were measured by semiquantitative RT-PCR. LPS treatment significantly increased COX-2 levels only in the brain stem, but not in the forebrain. The LPS-induced rise in COX-2 mRNA in this brain area was attenuated by IL-6AS (Fig. 6B, P = 0.0112). Similarly, exogenous administration of rrIL-6 resulted in a significant induction of COX-2 mRNA in the brain stem (Fig. 6D, P = 0.0296). The lack of significant change in COX-2 mRNA expression between the control and treated groups in the forebrain is most likely due to the very high constitutive expression of COX-2 in cortical neurons (see Fig. 5, A–D), which probably masked any intricate changes that were detected with the immunohistochemical analysis of this brain region.
PSTAT-IR Colocalizes with COX-2-IR After rrIL-6 or LPS Stimulation
On the basis of the observation that both PSTAT (Fig. 5M) and COX-2 (Fig. 5, B, D, F, H, J, and L) were induced in endothelial cells after LPS or rrIL-6 stimulation, colocalization of PSTAT and COX-2 was examined in these animals by double immunohistochemistry. LPS-induced perinuclear COX-2-IR was found to clearly colocalize with PSTAT-IR in blood vessel-like structures in the cortex (Fig. 5N) and other areas (data not shown). Similar observations were made in the brains of rrIL-6-treated animals (Fig. 5, O–T).
rrIL-6-Induced SOCS3 Correlates with COX-2 Expression
To further clarify the relationship between IL-6 activity in the brain and the induction of COX-2, we studied the correlation between SOCS3 mRNA induction as a measure of cellular action of IL-6 and COX-2 mRNA expression. The analysis of these data indicated a clear and significant correlation between the two in the brain stem (Fig. 7B; r2 = 0.8642, P = 0.006), but not in the forebrain (Fig. 7A; r2 = 0.1693, P = 0.6632). This provides additional support for the hypothesis that exogenous rrIL-6 directly activates the COX-2-pathway within the rat brain through a STAT3-related mechanism.
LPS-Induced Transcription Factor NF-κB Is not Affected by IL-6-AS Treatment
Nuclear translocation of p65, a subunit of NF-κB, was investigated to monitor its activation within immunostained rat brain sections 2 h after stimulation. LPS injection (Fig. 8, G and H, cortex for example) induced nuclear translocation of NF-κB within endothelial-like cells throughout the entire brain. Coadministration of IL-6AS did not affect this response (Fig. 8, C and D). Nuclear translocation was absent in control animals (Fig. 8, A and B) and in those treated with rrIL-6 alone (Fig. 8. E and F), where only cytoplasmic (inactive) NF-κB-IR was observed (Table 2). Interestingly, nuclear translocation of NF-κB in LPS-treated animals was seen to colocalize with PSTAT (Fig. 8, I–N), indicating a possible association of these two different intracellular signaling pathways.
The results of the current study confirm our earlier observations (10) and those of others (6, 11, 28, 34), demonstrating that IL-6 is an important neuroimmune mediator in fever following infection or inflammation. An important aspect of the role of IL-6 in fever that some studies (10, 11, 28) did not address and others did not show (30, 58) was a direct link between the circulating cytokine and the activation of the COX pathway, a critical step in the generation of fever. In the present study, we provide evidence demonstrating that circulating IL-6 does indeed activate this pathway. This is in agreement with an earlier study demonstrating that COX-2 induction was absent in the brains of IL-6 knockout mice injected with turpentine (a potent inflammatory agent) (57). Other studies using recombinant IL-6 administered systemically in rats failed to demonstrate COX-2 induction in the brain and had no effect on core body temperature (30, 58). The discrepancy between these observations and ours from the present study is most likely because of the bioactivity of the different recombinant protein used, whether as a result of species difference, i.e., human as opposed to rat (10, 30), or amounts injected (10, 58). In our experiments, we used a dose of 45 μg/kg of rrIL-6, which reflects better biological concentrations found in the circulation of rats in response to a systemic injection of LPS (21) and thus enough to reach a threshold concentration necessary for inducing fever. This was confirmed by our recent study showing that nuclear STAT3 translocation following LPS treatment in guinea pigs occurs only after plasma concentrations of IL-6 reach a critical level (50), also suggesting that LPS-induced STAT3 activation is IL-6 dependent. We have now confirmed these observations and those of others (19, 22, 24, 51) by demonstrating unequivocally that this is indeed the case. Most importantly, the present study showed that STAT3 activation using the paradigms described occurred in endothelial cells lining the vasculature of the brain, adding support to previous observations that these cells constitute important targets for circulating pyrogens (36). We also observed IL-6-induced STAT3 activation in the sCVOs, in cells other than endothelial cells. Although previously identified as important areas in fever, based primarily on lesion studies (48), the significance of a direct action of IL-6 on nonendothelial cells within the sCVOs in fever remains undetermined. One possibility is that IL-6 could be acting on thermoregulatory neurons, especially in the SFO in a similar manner to IL-1β, which was previously shown to trigger the firing of isolated neurons in this area (14). Other than nuclear STAT3 translocation, however, we have no evidence indicating that IL-6 is acting directly on this area to induce fever, especially since the only COX-2-IR we detected in this region was in endothelial cells in close proximity to this area of the brain. Thus, the most likely contribution of this brain region to the fever response remains to be that of a gateway for circulating pyrogens to access their targets in the brain through a “leaky” BBB. In contrast, the action of endogenous pyrogens on endothelial cells is well documented, and these cells are known to be the major source of PGE2 in the brain (36). Our study, however, is the first to document that systemic circulating IL-6 can act on these cells to induce COX-2, confirming findings of an earlier study in which this cytokine was injected directly into the brain (6). Other than endothelial cells, we also observed COX-2-IR in perivascular macrophage-like cells, which, along with endothelial cells (27), have previously been reported to express the inducible form of COX after inflammatory stimuli (55). The proportional contribution of each cell type to PGE2 production in the brain remains undetermined; however, our current observations based on COX-2-IR show that endothelial cells are by far the major source, at least at the time point (2 h) investigated in our study.
Furthermore, our results show that circulating IL-6-mediated COX-2-induction in endothelial cells occurred throughout the entire brain. However, PGE receptors, especially the EP3 receptor (EP3R) subtype are mainly expressed in specific brain structures, such as the MnPO (18), in which PG binding was reported to be highest (37), an observation that corresponds with the importance of this region in fever (31, 43). In addition, using Fos immunohistochemistry, Oka et al. (43) showed that activated neurons during LPS-induced fever within this brain structure colocalized with prostaglandin E receptors. This and other studies (53, 54, 60) demonstrated that despite PG production throughout the brain, the action of PGE in fever is limited to specific thermoregulatory hypothalamic nuclei. In the present study, by demonstrating that IL-6 induces COX-2 production in the MnPO, we provide strong evidence that this cytokine is involved in triggering the COX pathway in the hypothalamic regions mostly associated with regulating the febrile response. This is further supported by observations made in another study, in which the number of LPS-activated (Fos-IR positive) cells in the preoptic area was significantly reduced in IL-6 knockout mice (58).
Our investigation of the link between the IL-6-triggered signaling pathway, namely JAK-STAT and COX-2 induction, revealed a close association between the two. This was indicated by the significant correlation between circulating rrIL-6-induced SOCS3 mRNA and COX-2 mRNA levels (Fig. 7B). In addition, we show clear colocalization of COX-2-IR with PSTAT-IR, thus supporting a functional link between STAT3 and COX-2 induction. Our observations point to a somewhat unexpected, but possible role of the JAK-STAT pathway in COX-2 gene regulation. The difficulty with this rationale is that, to our knowledge, there is no consensus region reported for STAT3 in the COX-2 promoter region pointing to the possibility that IL-6-activated transcription factors other than STAT3 could play a role. A plausible candidate for this could be NF-IL-6, which along with STAT3, has been reported to mediate IL-6 signaling, also by phosphorylation followed by nuclear translocation (41). The role of this transcription factor in COX-2 expression was previously reported with several studies demonstrating its activation following different inflammatory stimuli (61). However, the majority of these studies were conducted in vitro to assess cell-specific molecular mechanisms in COX-2 induction, for example, in LPS-stimulated macrophages or in fibroblasts (20, 62), but to our knowledge not in brain endothelial cells and not in vivo. In the present study, using immunohistochemistry we failed to detect LPS- or IL-6-induced NF-IL6 activity in any of the brain regions and cell types analyzed at the 2-h time point, indicating that its importance during the early febrile response is questionable (data not shown).
Our proposed hypothesis that JAK-STAT is involved in COX-2 induction, contradicts the established dogma that the activation of the transcription factor NF-κB is the main intracellular pathway for fever induction (29, 33). Although this is still valid, it most likely reflects the involvement of other pyrogenic mediators such as IL-1β, which induces COX-2 through an NF-κB-dependent mechanism (40). However, unlike IL-6, IL-1β is not detected in the circulation of febrile animals (8, 38), at least not in models of localized inflammation, such as the one used in our current study. The other possibility is TNF-α, which like IL-1β also signals via NF-κB but unlike IL-1β is readily detectable in the circulation of LPS-treated animals. A role for this cytokine in fever is, however, still a subject of some debate with evidence for and against its role as a pyrogen (12). Nevertheless, although NF-κB was clearly induced in our studies, its activation was not affected in the presence of IL-6AS in clear contrast to STAT3, suggesting that the COX-2 induction observed in our experiments is independent of the NF-κB-pathway. Interestingly, however, we did detect PSTAT-IR with activated NF-κB-IR in blood vessel-like structures an observation that offers basis for some speculation. A recently described functional network between transcription factors (“crosstalk”) suggests that STAT3, NF-κB, and NF-IL-6 physically interact with one another. They mediate synergistic effects for instance by stabilizing their respective binding to genomic DNA or even transactivate promotors in proximity (1, 65). In line with these findings, COX-2 induction in myocardial preconditioning was identified to be both STAT3- and NF-κB-dependent (4). Interestingly, these authors also suggested that COX-2 can be upregulated via an IL-6 activated, JAK-STAT-dependent pathway in the myocardium (13, 63), a similar conclusion to ours in the brain.
In summary, we provided new and strong evidence demonstrating that IL-6 is directly involved in the activation of at least some components of the COX pathway in fever. Our findings do not, however, preclude the synergistic actions of this cytokine with other mediators working through different signaling pathways; indeed, our observations add insight into mechanisms of complex cytokine networks that may be operating at the level of transcription factors.
This work was supported by the Canadian Institutes of Health Research and Natural Sciences and Engineering Research Council of Canada.
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