Transforming growth factor-β (TGF-β), a pleiotropic cytokine, regulates cell proliferation, differentiation, and apoptosis, and plays a key role in development and tissue homeostasis. TGF-β functions as an anti-inflammatory cytokine because it suppresses microglia and B-lymphocyte functions, as well as the production of proinflammatory cytokines. However, we previously demonstrated that the intracisternal administration of TGF-β induces fever like that produced by proinflammatory cytokines. In this study, we investigated the mechanism of TGF-β-induced fever. The intracisternal administration of TGF-β increased body temperature in a dose-dependent manner. Pretreatment with cyclooxygenase-2 (COX-2)-selective inhibitor significantly suppressed TGF-β-induced fever. COX-2 is known as one of the rate-limiting enzymes of the PGE2 synthesis pathway, suggesting that fever induced by TGF-β is COX-2 and PGE2 dependent. TGF-β increased PGE2 levels in cerebrospinal fluid and increased the expression of COX-2 in the brain. Double immunostaining of COX-2 and von Willebrand factor (vWF, an endothelial cell marker) revealed that COX-2-expressing cells were mainly endothelial cells. Although not all COX-2-immunoreactive cells express TGF-β receptor, some COX-2-immunoreactive cells express activin receptor-like kinase-1 (ALK-1, an endothelial cell-specific TGF-β receptor), suggesting that TGF-β directly or indirectly acts on endothelial cells to induce COX-2 expression. These findings suggest a novel function of TGF-β as a proinflammatory cytokine in the central nervous system.
- prostaglandin E2
- body temperature
- central nervous system
- proinflammatory cytokine
- blood vessel
fever is a common manifestation of infectious disease. During infection, various types of pyrogen activate monocytes or macrophages and cause them to release proinflammatory cytokines as endogenous pyrogen. Proinflammatory cytokines induce cyclooxygenase-2 (COX-2) and microsomal PGE synthase-1 in the perivascular microglia, macrophages in the meninges and endothelial cells of brain blood vessels in the central nervous system (8, 50), and pulmonary and hepatic macrophages (16, 43, 56) in peripheral tissue. These enzymes ultimately synthesize PGE2 from arachidonic acid (17). PGE2 acts on the thermoregulatory center in the hypothalamus and causes a rise in body temperature (9, 58, 61). A high body temperature (fever) suppresses viral proliferation and bacterial growth and activates and induces the proliferation of immune cells (20). Thus, fever has a protective role against infection.
Transforming growth factor-β (TGF-β) was discovered by virtue of its capacity to induce anchorage-independent growth of normal rat kidney and fibroblast cell lines, that is, its capacity to induce transformation (36, 44). Later research has identified TGF-β as one of the cytokines that has multifunctional effects on various cell types. TGF-β promotes cell survival or induces apoptosis and stimulates cell proliferation or induces differentiation (6, 31). TGF-β also regulates the immune system. There have been studies demonstrating that TGF-β attenuates the production and activity of proinflammatory cytokines in peripheral tissues (6) and suppresses the activation of lymphocytes (5, 47) and microglia (2, 25). On the basis of such functions, TGF-β is considered to be an anti-inflammatory cytokine. However, it has been suggested that TGF-β both activates and inactivates macrophages (1) and activates nuclear factor-κB (NF-κB) (21, 29), a member of the proinflammatory signal transduction system. These reports indicate that the effects of TGF-β are bidirectional, that is, proinflammatory or anti-inflammatory, depending on the type and status of the cell that receives its action.
Recently, using transgenic mice with TGF-β-responsive luciferase reporter construct, Lin et al. (28) showed that the intraperitoneal administration of LPS strongly activates TGF-β signaling in the brain. TGF-β induces COX-2 in neurons and astrocytes in vitro (30). In addition, we observed that the intraperitoneal administration to rats of polyinosinic:polycytidylic acid (poly I:C), a synthetic double-stranded RNA, increases active TGF-β concentrations in cerebrospinal fluid (CSF) but not in the blood (33); the intracisternal administration of an anti-TGF-β antibody partially inhibited fever induced by poly I:C administration. Furthermore, intracisternal administration of TGF-β induces fever. This phenomenon of fever being induced by intracisternal administration of TGF-β is the same as that observed by the administration of proinflammatory cytokines such as IL-1β, TNF-α, and IL-6. On the basis of our previous results and the above-mentioned reports, we hypothesized that TGF-β in the brain acts as a proinflammatory cytokine and induces fever.
In the present study, we examined whether the intracisternal administration of TGF-β induces fever in a COX-2- and PGE2-dependent manner and, if so, where in the brain COX-2 is induced by TGF-β.
MATERIALS AND METHODS
This study was conducted in accordance with the ethical guidelines of the Kyoto University Animal Experimentation Committee and the Japan Neuroscience Society, and was in complete compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All efforts were made to minimize the number of animals used and to limit experimentation to what was necessary to produce reliable scientific information and prevent any suffering. All procedures were approved by the Kyoto University Animal Care and Use Committee. Eight-week-old male Sprague-Dawley rats were used. All animals were maintained under a 12:12-h light-dark cycle (0600, lights on; 1800, lights off) for at least 1 wk before surgery. They were allowed free access to water and standard lab chow in a room maintained at 22 ± 0.5°C and 50% humidity.
A telemetry sensor/transmitter for core body temperature (model TA10-TA; Data Science, St. Paul, MN) was intraperitoneally implanted in each rat anesthetized with pentobarbital sodium as described by Ohnuki et al. (40). After implantation of the telemetry sensor/transmitter, a cannula for the intracisternal administration of TGF-β was implanted into the cisterna magna as described by Yamazaki et al. (63). The anesthetized rat was fixed to a stereotaxic apparatus, and a stainless cannula was inserted 3.0 mm posterior to the lambda and 8.7 mm deep, inclined anteriorly at an angle of 60° to the horizontal plane. The cannula was fixed with Loctite 454 (Loctite Japan, Yokohama, Japan). The rats were allowed to recover for 1 wk after surgery.
All drugs were administered at 0800, 2 h after the onset of the light period. TGF-β (R&D Systems, Minneapolis, MN; LPS contamination <100 pg/μg) was dissolved in sterilized artificial cerebrospinal fluid (aCSF; 140 mM NaCl, 3 mM KCl, 1.5 mM Na2HPO4, 0.23 mM NaH2PO4, 1.5 mM CaCl2, 1.26 mM MgCl2, 3.4 mM d-glucose, 0.1% bovine serum albumin) at a final concentration of 1 μg/ml, and then 40 μl (i.e., 40 ng of TGF-β) was intracisternally administered. Heat-denatured TGF-β was prepared by boiling for 15 min. In control rats, the same volume of aCSF or aCSF containing 4 pg LPS (Sigma-Aldrich, St. Louis, MO) was administered.
To investigate the effect of COX-2-specific inhibitor on febrile response, we injected rats intraperitoneally with either nimesulide (4 mg/kg, Sigma-Aldrich) and NS398 (4 mg/kg, Sigma-Aldrich) or its vehicle (500 μl of 50% DMSO in saline) 1 h before the administration of TGF-β. Nimesulide and NS398 possess high selectivity for COX-2, as shown by several studies (26, 32, 57, 59).
Measurement of Body Temperature
Body temperature was measured as described by Ohnuki et al. (40) with some modifications. Measurement was started the day before the experiment. The rats were acclimated to the test environment (plastic rat cage, 33 × 23 × 14 cm), and the baseline body temperature of each rat was recorded. Signals from the sensor/transmitter were detected every 5 min by a flat-panel receiver (model RA1010; Data Science) and analyzed using the Dataquest IV program (Data Science).
Enzyme Immunoassay for PGE2 in CSF
CSF was collected following the method described by Inoue et al. (15). Five hours after the injection of TGF-β or its vehicle, rats were anesthetized with pentobarbital, and CSF was collected from the preimplanted cannula, which was connected to a polyethylene tube. After collection, CSF was centrifuged at 2,000 g and 4°C. The supernatant was then collected and stored at −70°C until measurement. CSF visibly contaminated with red blood cells was discarded. On the day of the assay for PGE2, the samples were thawed on ice, and PGE2 was extracted with an organic solvent (ethyl acetate). The extracted samples were assayed for PGE2 with an enzyme immunoassay kit (Cayman Chemical, Ann Arbor, MI), according to the manufacturer's instructions.
Rats were deeply anesthetized with pentobarbital and perfused via the aorta with 0.01 M PBS (pH 7.4). For immunohistochemistry, the brains were freshly frozen in dry-ice powder. The frozen brains were cut at a thickness of 16 μm in a cryostat and thaw-mounted on glass slides. After air drying at room temperature for 30 min, the sections were fixed with 2% paraformaldehyde in PBS for 10 min at room temperature. After rinsing once with PBS, the sections were treated with 0.3% Triton X-100 in PBS for 30 min and then incubated with 10% normal donkey serum for 1 h. The sections were incubated with goat anti-COX-2 antibody (1:4,000 dilution; Santa Cruz Biotechnology, Santa Cruz, CA), goat anti-type I TGF-β receptor (TβRI) antibody (1:100 dilution; Santa Cruz Biotechnology), and goat anti-activin receptor-like kinase-1 (ALK-1) antibody (1:200 dilution; Santa Cruz Biotechnology) overnight at room temperature. After removal of the primary antibody, the sections were incubated with Alexa Fluor 488-conjugated donkey anti-goat IgG (1:500 dilution; Molecular Probes, Eugene, OR) for 2 h. Sections were washed in PBS and coverslipped using an antifading glycerol-based mounting medium. Immunostained sections were observed under a confocal laser scan microscope (FV-300; Olympus, Tokyo, Japan) with argon and He-Ne laser sources.
In the case of double immunostaining, after COX-2 or TβR immunostaining, the sections were further incubated with rabbit anti-vWF antibody (1:5,000 dilution; Dako, Carpinteria, CA), rabbit anti-COX-2 antibody (1:500 dilution; Cayman Chemicals) and mouse monoclonal anti-α-actin antibody (1:10,000 dilution; Sigma-Aldrich) overnight and then with Alexa Fluor 546-conjugated donkey anti-rabbit IgG or Alexa Fluor 546-conjugated donkey anti-mouse IgG for 2 h.
Control staining was conducted without primary antibody or with preabsorbed antibody, in which the diluted antibody and the antigen peptide (1 μg/ml) had been mixed and incubated at room temperature. No labeling was observed on control sections.
Quantitative Assessment of Colocalization of COX-2 and vWF
Quantitative assessment of immunostaining was conducted on the coronal sections of the rat brain corresponding to −0.20 to −0.60 mm posterior from the bregma. To quantify the frequency of colocalization of COX-2 with vWF, we classified COX-2-immunoreactive cells from three to five sections from each individual as either COX-2 only or double-labeled with vWF. COX-2-immunoreactive cells were manually counted in each rat and averaged, and the percentage of COX-2-positive cells expressing vWF was calculated.
Western Blot Analysis
For Western blot analysis, the brains were removed as described above. Arachnoid meninges and leptomeninges containing blood vessels were separated from the brain with fine-tipped forceps under a dissecting microscope. Samples were homogenized by sonication in Laemmli buffer, and the protein concentration was determined using a Pierce bicinchoninic acid protein assay kit (Pierce Biotechnology, Rockford, IL). Each sample (10 μg of protein) was subjected to SDS-polyacrylamide gel electrophoresis (10% gel). Proteins in the gel were transferred to a polyvinylidene difluoride membrane (Immobilon-P; Millipore, Bedford, MA) by electroblotting. The membrane was blocked for 1 h with Tris-buffered saline (pH 7.4) containing 5% skim milk and 0.1% Tween 20 and incubated overnight at 4°C with anti-COX-2 antibody (1:1,000 dilution; Cayman). After several washings, a secondary antibody, horseradish peroxidase-conjugated goat anti-rabbit IgG was added. Antibody labeling was detected by chemiluminescence. The relative intensity of the bands was analyzed with National Institutes of Health Image software (NIH image software) and normalized by the relative intensity of the bands of the control group.
All values are presented as the means ± SE. Body temperature is expressed as the average of five temperatures measured every 5 min over a period of 30 min. The average baseline body temperature was calculated using data from 1 to 0 h before the intracisternal administration. The effects of the administration of TGF-β on increases in body temperature were examined by two-way repeated-measures ANOVA. The values at each time point were compared by an all-pairwise Bonferroni's multiple comparison post hoc test. The effect of COX-2 inhibitor (nimesulide and NS398) on TGF-β-induced fever was examined by two-way repeated-measures ANOVA followed by an unpaired Student's t-test as a post hoc test. The concentrations of PGE2 in CSF and COX-2 protein in blood vessels were examined using an unpaired t-test.
Fever Induced by the Intracisternal Administration of TGF-β
Figure 1A shows the time course changes of body temperature that were induced by the intracisternal administration of TGF-β. The average body temperature at the time of administration (time point 0) was 37.1 ± 0.1°C for the vehicle-administered group and 37.1 ± 0.1°C, 36.9 ± 0.1°C, and 37.1 ± 0.1°C for the 10, 20, and 40 ng dose of TGF-β administered group, respectively. There were no significant differences in the average body temperature among all groups. An increase in body temperature was observed in all groups immediately after administration, probably due to handling stress. The body temperature of the vehicle-administered group returned to baseline at 1.5 h after administration; however, the body temperature of the TGF-β-administered group rose dose dependently. A 20-ng dose of TGF-β significantly increased body temperature, but 10 ng of TGF-β did not. The body temperature of rats in the TGF-β-administered group (40 ng) reached its peak at 5 h after administration and then declined. The body temperature of the vehicle-administered group increased slowly and reached a maximum level only slightly higher than that observed under the untreated condition.
To determine whether endotoxin contamination of preparation causes fever, we administered 4 pg of LPS, which is the maximum content of LPS contamination within the recombinant TGF-β used in this study or heat-denatured TGF-β (40 ng). These treatments did not affect body temperature (Fig. 1B).
Effect of COX-2 Inhibitor on Fever Induced by the Intracisternal Administration of TGF-β
The COX-2 selective inhibitor nimesulide (4 mg/kg) or its vehicle (500 μl of 50% DMSO) was administered intraperitoneally 1 h before the intracisternal administration of TGF-β. The average body temperature at the time of TGF-β administration (time point 0) was 37.4 ± 0.1°C for the vehicle (ip) + TGF-β-administered group, 37.5 ± 0.1°C for the nimesulide (ip) + TGF-β-administered group, and 37.4 ± 0.1°C for the nimesulide (ip) (without TGF-β administration) group. Baseline body temperature (time point 0) was increased, probably because of handling stress associated with intraperitoneal administration of nimesulide or its vehicle compared with those without intraperitoneal administration (Fig. 1), and gradually decreased thereafter. However, TGF-β administration induced a further increase in body temperature from the baseline body temperature. In contrast, in the nimesulide-administered group, TGF-β administration showed no increase in body temperature (Fig. 2A). The intraperitoneal administration of nimesulide alone brought about no changes in body temperature.
We further examined the effect of another COX-2 selective inhibitor, NS398, on increases in body temperature induced by TGF-β. The average body temperature at the time of TGF-β administration (time point 0) was 37.2 ± 0.2°C for the vehicle (ip) + TGF-β-administered group, 37.2 ± 0.2°C for the NS398 (ip) + TGF-β-administered group and 37.2 ± 0.2°C for the NS398 (ip) (without TGF-β administration) group. Similar to nimesulide, NS398 (4 mg/kg) effectively suppressed TGF-β-induced fever (Fig. 2B).
PGE2 Concentrations in Cerebrospinal Fluid
It has been reported that there is a temporal relation between fever induced by LPS or by proinflammatory cytokines and PGE2 level in CSF (15, 32). Therefore, we collected CSF 5 h after TGF-β administration when the increase in body temperature had reached its peak. PGE2 concentrations in CSF were found to have increased significantly in the TGF-β group compared with those in CSF of the vehicle group (Fig. 3).
Induction of COX-2 Protein by the Intracisternal Administration of TGF-β
Figure 4 shows COX-2 immunostaining of the coronal brain sections of rats 5 h after the administration of TGF-β or its vehicle. Three different individuals from both groups are shown. It has been reported that the EP3 subtype of PGE receptor is localized in the preoptic area (POA) and mediates PGE2-induced fever (38, 64). The intracerebroventricular administration of LPS or proinflammatory cytokine induces COX-2 expression in brain blood vessels near the POA (3, 4). Therefore, we investigated COX-2 expression near the POA (illustrated in Fig. 4B) by immunohistochemistry. COX-2 immunoreactivity (COX-2-ir) was found most strongly in the subarachnoidal blood vessels and moderately in the brain parenchymal blood vessels (Fig. 4, A and C). COX-2-ir was hardly detectable in the vehicle group.
To investigate COX-2 expression levels by Western blot analysis using anti-COX-2 antibody, we collected subarachnoidal blood vessels 5 h after TGF-β administration. In the homogenate of the subarachnoidal blood vessels, the anti-COX-2 antibody recognized a protein band of 70 kDa, which is consistent with the molecular weight of rat COX-2 (Fig. 5A). Semiquantitative analysis revealed that COX-2 expression was significantly increased by TGF-β administration (Fig. 5B). These results indicate that TGF-β induces COX-2 expression in the brain blood vessels and increases PGE2 synthesis similar to the action of proinflammatory cytokines.
Identification of COX-2-Immunoreactive Cells
To identify the type of COX-2-immunoreactive cells found in the TGF-β-administered group, we performed double immunostaining of COX-2 and vWF, a well-known endothelial marker protein. Figure 6 (left, top) shows the colocalization of COX-2 and vWF in the subarachnoidal space, indicating that COX-2-immunoreactive cells were endothelial cells of the blood vessels. Furthermore, COX-2 was induced in the brain parenchymal blood vessels (Fig. 6, left, bottom). However, some COX-2-ir was also found in vWF-negative cells in the subarachnoidal space. The inconsistency of double immunostaining of COX-2 and OX-42 in these cells excluded the possibility that these were parenchymal microglia or perivascular/meningeal macrophages (data not shown).
The percentage of COX-2-immunoreactive cells expressing vWF was 89 ± 3.1% (1,279/1,512 cells; n = 6). These results indicate that TGF-β administered into the brain induced COX-2 expression mainly in the endothelial cells of the brain blood vessels.
Localization of TβRs
Three different types of TGF-β receptors have been identified to date (7). Among them, type I TGF-β receptor [TβRI, also termed activin receptor-like kinase-5 (ALK-5)] and type II TGF-β receptor (TβRII) are the signaling components. Signaling of TGF-β is initiated by binding to constitutively active TβRII. Upon binding, TβRI and TβRII form a heterometric complex, resulting in transphosphorylation of TβRI by TβRII, and signals are transduced to the cytoplasmic targets by TβRI. In most cells, TGF-β transmits its signal via TβRI and TβRII. In endothelial cells, however, TGF-β has been shown to bind and signal via both ALK-5 and ALK-1 (12, 39). Furthermore, ALK-1 is specifically expressed in the endothelial cell and functions as a type I receptor (42, 46).
Immunohistochemical examination using anti-TβRI antibody showed that TβRI immunoreactivities were observed primarily in the artery (Fig. 7). Double immunostaining revealed colocalization of TβRI and smooth muscle α-actin, indicating that TβRI is expressed in the smooth muscle cells. Weak TβRI immunoreactivity was also found in glial fibrillary acidic protein (glial cell marker)-positive cells (data not shown). However, TβRI immunoreactivity was not found in the endothelial cells (Fig. 7). These results agree with those of a previous study that utilized a knockin mouse line that carries a lacZ reporter in the TβRI gene locus (52).
ALK-1 immunoreactivities were specifically found in blood vessels (Fig. 8). Double immunostaining of vWF and ALK-1 revealed that some cells of vWF-positive cells express ALK-1, indicating that ALK-1 is specifically expressed in the endothelial cells in the rat brain.
Figure 9 shows that COX-2-ir is found in ALK-1-positive cells, but it is also found in ALK-1-negative endothelial cells. Although it has been reported that COX-2 is induced by immune stimuli in astrocytes (60) or macrophages and microglial cells, no obvious COX-2-ir was found in these cells in the present study.
In our previous study, we demonstrated that the concentration of active TGF-β in CSF was increased by poly I:C administration and identified the involvement of TGF-β in the mechanism underlying the development of fever (33). In the present study, we demonstrated that the intracisternal administration of TGF-β increased PGE2 levels in CSF and induced fever. Pretreatment with COX-2-selective inhibitors (nimesulide and NS398) suppressed TGF-β-induced fever. These results clearly indicate that COX-2 and its end product, PGE2, are involved in fever induced by the intracisternal administration of TGF-β. We also demonstrated that TGF-β induces COX-2 expression in the endothelial cells of brain blood vessels and promotes PGE2 biosynthesis.
The environment in which TGF-β acts is extremely interesting. The increase in the TGF-β concentration was found only in CSF, but not in the blood, after intraperitoneal administration of poly I:C, and TGF-β administration into the cisterna magna resulted in alterations of body temperature. These findings indicate that TGF-β is probably activated in the brain (most likely in the CSF) and that it acts on the endothelial cells of brain blood vessels from the basement membrane side rather than from the luminal side. In our previous study, we showed that an increase in TGF-β concentrations in CSF precedes that of other proinflammatory cytokines in the blood (33). This observation may suggest that the early phase of the pyrogenic reaction elicited by poly I:C administration progresses primarily in the brain. Although expression of the TGF-β receptor (ALK-1) and induction of COX-2 were found in the vascular endothelial cells, factors in blood do not appear to be involved in this case. Therefore, other factors may function to signal the brain regarding peripheral inflammation or infection. One candidate would be the vagus nerve (48). Thus, peripheral information may be transmitted via the vagus nerve, and TGF-β in the brain may act as a mediator.
PGE2 is synthesized via the arachidonic acid cascade. COX is a rate-limiting enzyme of this cascade and has two isoforms, COX-1 and COX-2. COX-1 is constitutively expressed in various tissues, while COX-2 is not expressed under normal conditions. Various stimuli, including growth factors, proinflammatory cytokines, endotoxins, and tumor promoters are known to stimulate its expression (54). Gene knockout studies have indicated that COX-2, rather than COX-1, plays an important role in the development of the fever induced by LPS or IL-6 and IL-1β (26, 27).
It has been demonstrated that proinflammatory cytokines (IL-1β, IL-6, TNF-α) induce the expression of COX-2 in the endothelial cells of brain blood vessels, promoting PGE2 synthesis (3, 4, 32). In the present study, we showed that intracisternal administration of TGF-β also induces COX-2 expression, mainly in the endothelial cells, but not in astrocytes, microglia, or macrophages. In addition, TGF-β increased the PGE2 level in CSF, indicating that TGF-β promotes PGE2 synthesis in the brain and elicits effects very similar to those elicited by proinflammatory cytokines. However, COX-2 expression was also found in nonendothelial cells and TGF-β receptor (ALK-1)-negative endothelial cells. These results indicated that, in the brain, TGF-β may act on another type of TGF-β receptor or may act indirectly on these cells. Further studies are needed to clarify the mechanism of induction of COX-2 protein by TGF-β.
COX-2 expression is regulated by multiple signaling pathways, and its regulation is dependent on cell type and the specific stimulus in question (45, 49, 51, 55). With respect to brain endothelial cells, several studies have demonstrated the involvement of NF-κB and the MAPK pathway in the regulation of COX-2 expression induced by LPS or proinflammatory cytokines (22, 24, 37). TGF-β also activates NF-κB (21) and MAPK (14) in various cell types, and Smad, one of the intracellular components of the TGF-β signaling pathway, positively regulates NF-κB activity (35). Furthermore, Harding et al. (13) and Rodriguez-Barbero et al. (45) recently demonstrated that TGF-β increases COX-2 expression by activating NF-κB and MAPK in glomerular mesangial cells. It is likely that TGF-β induces COX-2 expression in the endothelial cells of the brain blood vessels via activation of NF-κB and MAPK pathways.
Targeted deletion of the TGF-β gene results in excessive inflammatory responses (23, 53) and increases neuronal cell death and severe microgliosis (2). Several in vitro studies have demonstrated that TGF-β represses the activation of microglia (19, 25, 41). In all of these studies, the effects of TGF-β in the central nervous system appear to be in direct opposition to the effects of proinflammatory cytokines.
Contrary to previous reports, herein, we have demonstrated that TGF-β acts like a proinflammatory cytokine on the endothelial cells of the brain blood vessels in the central nervous system. TGF-β is a rare cytokine that changes its function to proinflammatory or anti-inflammatory depending on the cell type where it is acting or on variations in other factors; it remains unknown why such a phenomenon occurs. Three types of TGF-β receptor have been identified: TβRI (also known as ALK-5), TβRII, and TβRIII (7). Of these, the TβRI and TβRII receptors are the signaling components. On the other hand, ALK-1 is specifically expressed in endothelial cells during embryogenesis and in adult stages (42, 46). ALK-1 is reported to function as a type I TGF-β receptor (12, 39). However, in endothelial cells, activation of ALK-1 stimulates cell proliferation and migration, whereas activation of TβRI inhibits these responses (11, 12). These reports suggest that ALK-1 exhibits the opposite effect against TβRI on angiogenesis. In addition, we also observed that TGF-β-induced COX-2 expression in ALK-1-positive cells, but not in TβRI-positive cells. Thus, some of the bidirectional effects of TGF-β may be accounted for by the diversity of TGF-β receptor. Further studies are needed to elucidate this mechanism.
In conclusion, we have demonstrated that TGF-β induces fever via the mediation of COX-2-dependent PGE2. We also observed that TGF-β induces COX-2 expression in endothelial cells expressing ALK-1. These findings suggest that TGF-β, which is typically known as an anti-inflammatory cytokine, acts as a proinflammatory cytokine on endothelial cells, possibly through activation of ALK-1.
Perspectives and Significance
TGF-β has been considered to be an anti-inflammatory cytokine. Targeted deletion of the TGF-β gene causes serious inflammation throughout the whole body (2, 23, 53). However, our study clearly indicates the proinflammatory aspect of this pluripotent cytokine and presents a novel mechanism for the proinflammatory actions of TGF-β in the brain. In neurodegenerative diseases, such as ischemia, Alzheimer's, and multiple sclerosis, in which the inflammatory reaction is problematic, both TGF-β expression levels in the brain and the TGF-β concentration in CSF have been reported to be elevated (62). This may support the notion of proinflammatory actions of TGF-β. On the other hand, the facts that TGF-β plays a neuroprotective role in such diseases (10) and impairment of TGF-β signaling causes severe pathology have been demonstrated in both in an vitro study and gene knockout animal models, described above. Furthermore, patients with hereditary hemorrhagic telangiectasia, a human vascular disorder, have a gene deficiency in ALK-1 (18, 34), another type I receptor for TGF-β that is expressed in vascular endothelial cells. Taken together with previous reports and our data, TGF-β plays a highly divergent role in the central nervous system. Thus, elucidation of the mechanism of action of TGF-β is expected to lead not only to the understanding of physiological role of this cytokine but also the development of therapeutic regimens of various diseases.
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 © 2008 the American Physiological Society