Selective Modulation of BV-2 Microglial Activation by Prostaglandin E2

The influence of prostaglandins on glial functions and, more specifically, on glial activation is not well understood. We report here that prostaglandin E2(PGE2), one of the major prostaglandins produced in the brain, acts as a potent and selective inhibitor of tumor necrosis factor α (TNF-α) production in lipopolysaccharide-stimulated primary microglia and the microglial cell line BV-2. The IC50 for this effect is 1 nm, and 100 nm PGE2 suppresses TNF-α production by >95%. More detailed studies of BV-2 cells show that PGE2also prevents the secretion of interleukin (IL)-6 but does not significantly modify lipopolysaccharide-stimulated expression of cyclooxygenase-2, pro-IL-1β, or inducible nitric oxide synthase. PGE2 appears to act primarily at the level of translation or protein stability, because TNF-α and IL-6 mRNA levels were only modestly decreased at high PGE2 concentrations; concomitantly with this inhibition, PGE2 up-regulated the levels of IL-1β mRNA. The effects of PGE2 could be largely mimicked by 8-bromo-cAMP, suggesting that, as in other cell types, PGE2 action is mediated at least in part by a rise in intracellular cyclic AMP. However, the protein kinase A inhibitor H89 only partially reversed the inhibition of TNF-α production by PGE2, implying that the PGE2 effect in BV-2 cells is mediated through both protein kinase A-dependent and -independent pathways.

One prominent feature of neurodegenerative diseases such as Alzheimer's disease is the presence of activated microglial cells (1)(2)(3). Transition of microglia from the resting state to the activated one in response to a pathological stimulus is characterized in part by the production of various cytokines and the induction of inflammation-related enzymes such as inducible nitric oxide synthase (iNOS) 1 and cyclooxygenase-2 (COX-2) (4). Transient activation of microglia is most likely beneficial for brain-repairing processes; however, a chronic reactive state of microglia or an abnormally high proportion of activated microglia may become dangerous by increasing the inflammatory burden.
Recent data indicate that the prolonged treatment of individuals with nonsteroidal anti-inflammatory drugs decreases the probability of development of Alzheimer's disease (5,6). This in turn suggests that inhibition of COX-2 activity may be beneficial in the treatment of neurodegenerative conditions. Although the induction of COX-2 in glia (4,7,8) and neurons (9,10) is well documented, the mechanisms underlying the effect of nonsteroidal anti-inflammatory drugs in the brain as well as the effects of increased prostaglandin production on glial functions are not well understood and need to be investigated at the molecular level. The goal of this study was to determine how prostaglandin E 2 (PGE 2 ), one of the major prostaglandins produced in the brain, affects the production of activation-related molecules in microglia. We studied the proteins produced both at early (TNF-␣ and IL-1␤) and later (IL-6, iNOS, COX-2) stages of microglial activation in vitro, and we report here that PGE 2 acts as a potent and selective inhibitor of TNF-␣ and IL-6 production. The mechanism of this inhibition and the role of cAMP were also investigated.

EXPERIMENTAL PROCEDURES
Chemicals-PGE 2 was purchased from Cayman Chemical (Ann Arbor, MI) and prepared from powder as ethanol solutions. Bacterial lipopolysaccharide (LPS) from Salmonella typhimurium (Sigma; L-2622) was resuspended in sterile phosphate-buffered saline at 10 mg/ml and stored at Ϫ20°C. Mouse macrophage-colony stimulating factor (R&D Systems) was dissolved in phosphate-buffered saline containing 0.1% fatty acid-free bovine serum albumin (Sigma) to make a 20 g/ml stock solution and stored in aliquots at Ϫ80°C. H89 (Alexis) and 8-bromo-cAMP (Sigma) were prepared as 25 and 100 mM stock solutions, respectively, in sterile distilled water and stored in aliquots at Ϫ20°C.
Cell Culture and Treatment-Primary rat microglia were prepared and maintained as described elsewhere (11). The purity of microglial cultures was assessed using OX42 antibody (Serotec Inc., Raleigh, NC). and Ͼ95% of cells stained positively. After shake-off, cells were cultured for 2 days before treatment with LPS and prostaglandins. Immediately before treatment, serum-containing medium was removed, cells were washed two times with warm ␣-minimum Eagle's medium, and then ␣-minimum Eagle's medium containing N2 supplements (Life Technologies, Inc.) and 10 ng/ml macrophage-colony stimulating factor were added.
BV-2 (a murine microglial cell line generously provided by Dr. Michael McKinney; Mayo Clinic, Jacksonville, FL) cells were cultured in ␣-minimum Eagle's medium containing 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were washed twice with warm ␣-minimum Eagle's medium and then treated in serum-free medium. In all experiments, cells were pre-treated with PGE 2 or 8-bromo-cAMP for 30 min before the addition of LPS. PGE 2 was added to cells as an ethanol solution, with the final concentration of ethanol never exceeding 0.1%. Control samples contained the same concentration of diluent as experimental samples. This concentration of ethanol did not have any influence on cell viability or responses to LPS.
Immunostaining-Microglia were plated at 1 ϫ 10 5 cells/well in 4-well chamber slides (Falcon). For immunostaining, cells were washed twice with cold phosphate-buffered saline, fixed in 4% formaldehyde * This work was supported in part by National Institutes of Health Grants AG13939 and AG15501 (to L. J. V. E.), National Institutes of Health Training Grant GM08061 (to K. T. A.), and a postdoctoral fellowship from the Swiss National Science Foundation (to T. V. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
(Tousimis; Rockville, MD) for 10 min, permeabilized with 0.1% Triton X-100 for 5 min, and blocked with 10% normal donkey serum for 30 min. Slides were incubated for 1 h at room temperature with goat anti-rat IL-1␤ (R&D Systems) primary antibody at a 1:200 dilution, washed, and then incubated for 30 min with fluorescein isothiocyanite-labeled donkey anti-goat IgG secondary antibody (Jackson ImmunoResearch) at a 1:400 dilution in the presence of 10% normal donkey serum. Slides were mounted and viewed with a Zeiss Axioplan microscope. No punctate staining or very weak punctate staining of microglial cells was observed when goat anti-rat TNF-␣ antibodies were used instead of goat anti-rat IL-1␤. The lack of TNF-␣ staining is most likely because this cytokine is rapidly secreted from microglial cells, and the immunofluorescence method is not sensitive enough to detect the low levels of remaining intracellular and membrane-bound TNF-␣ protein. The absence of TNF-␣ staining provides a specificity control for the immunofluorescence method.
Protein Kinase A Assays-BV-2 cells were pre-treated with 5 M H89 or control buffer for 30 min and then stimulated with 1 M PGE 2 or vehicle for 3 min. Protein kinase A activity was determined in digitonin-permeabilized cells as described previously (12), except that cells were plated at 5 ϫ 10 4 cells/well in 48-well plates and grown for 2 days before assay.
ELISA and Western Blotting-TNF-␣ and IL-1␤ levels in cell conditioned media were determined using Duo-Set TNF-␣ and IL-1␤ kits (Genzyme) according to the manufacturer's instructions. Cell lysates were prepared, and Western blotting was performed as described previously (11) using the LumiGlo ® chemiluminescence detection system (New England BioLabs). To determine IL-6 levels, bovine serum albumin was added to 300 -400 l of conditioned media to a final concentration of 0.2 mg/ml, and proteins were precipitated in 5% trichloroacetic acid. After a 5-min incubation on ice, precipitates were recovered by centrifugation for 5 min at 12,000 rpm at 4°C, and pellets were dissolved in 40 l of 0.1 M Trizma base. Aliquots were saved for determination of protein concentration, and the levels of IL-6 in the samples were determined by Western blotting. IL-6, pro-IL-1␤, iNOS, and COX-2 protein levels were determined by Western blotting using the following antibodies: goat anti-mouse IL-6 (R&D Systems), rabbit antimouse IL-1␤ (Genzyme), monoclonal anti-murine mac-NOS (Transduction Laboratories), and goat anti-human COX-2 (Biotechnology, Inc., Santa Cruz). All primary antibody dilutions were 1:1000. Secondary goat anti-rabbit, goat anti-mouse, and rabbit anti-goat antibodies conjugated to horseradish peroxidase (Jackson ImmunoResearch) were used at 1:5000 (goat anti-rabbit and rabbit anti-goat) and 1:2000 (goat anti-mouse) dilutions.
Total RNA was isolated from the cells, and Northern blots were performed as described previously (13). Equal loading of RNA was verified by stripping the membranes and reprobing with rat pTRI-glyceraldehyde-3-phosphate dehydrogenase probe (Ambion, Austin, TX).
Statistics-Statistics have been calculated using Student's t test with significance established at a level of p Ͻ 0.05.

PGE 2 Suppresses TNF-␣ but not IL-1␤ Production in
Primary Rat Microglia-Primary microglia show a strong increase in TNF-␣ production at 3 h after stimulation with LPS, whereas pre-treatment with 1 M PGE 2 significantly inhibited production of this cytokine (Fig. 1A). To examine the generality of this response, we tested whether the production of another pro-inflammatory cytokine, IL-1␤, was affected by PGE 2 . The levels of IL-1␤ were detected by immunostaining and Western blotting, based on previous findings that IL-1␤ in activated microglia, unlike TNF-␣, remains primarily cell-associated (16). In contrast to TNF-␣, levels of IL-1␤ were not affected by treatment with PGE 2 (Fig. 1B); on the contrary, PGE 2 even seemed to stimulate the production of IL-1␤ somewhat. PGE 2 Inhibits TNF-␣ and IL-6 Protein Production in LPSstimulated BV-2 Cells-The murine microglial cell line BV-2 was shown to reproduce many microglial responses in culture and has been used as a model microglial system in other studies (17,18). Stimulation of BV-2 cells with LPS leads to a robust increase of TNF-␣ levels in the cell-conditioned media, which becomes evident by 3 h and reaches a maximum by 6 h (data not shown). Pre-treatment of BV-2 cells with PGE 2 inhibited LPS-induced TNF-␣ production in a dose-dependent manner, with an IC 50 of ϳ1 nM ( Fig. 2A). Treatment of BV-2 cells with PGE 2 also prevented LPS-stimulated accumulation of another cytokine, IL-6, 18 -24 h after the addition of LPS (Fig. 2B). PGE 2 Does Not Affect the Production of Other Inflammatory Response Proteins, Such as pro-IL-1␤, iNOS, and COX-2 in BV-2 Cells-We tested the influence of PGE 2 on the production of other activation-related cytokines or enzymes in BV-2 cells. Stimulation of BV-2 cells with LPS leads to a strong increase in the production of cell-associated pro-IL-1␤, COX-2, and iNOS (Fig. 3). Although BV-2 cells produced high amounts of pro-IL-1␤, processed IL-1␤ was not detected in conditioned medium by ELISA, even at later time points (data not shown). PGE 2 treatment did not inhibit LPS-induced pro-IL-1␤ or COX-2 and only weakly blocked (ϳ50% inhibition) iNOS induction (Fig. 3). In addition, in the case of weak inhibition of iNOS, even high concentrations of PGE 2 (up to 10 M) were unable to further suppress the production of this protein or the production of NO as measured by the accumulation of nitrite in the conditioned media (11). These data demonstrate that the action of PGE 2 in BV-2 microglia is highly selective, inhibiting TNF-␣ and IL-6 but not IL-1␤, COX-2, or iNOS.
Influence of PGE 2 on TNF-␣, IL-6, and IL-1␤ mRNA Levels-We have investigated at what level the inhibition of TNF-␣ and IL-6 may occur in BV-2 cells. Stimulation with LPS leads to ϳ20-fold increase in TNF-␣ mRNA compared with control; pre-treatment with 1 M PGE 2 results in a 40% decrease in LPS-induced TNF-␣ mRNA levels (Fig. 4). Although a small decrease in TNF-␣ mRNA was also observed in cells treated with lower doses (100 and 1 nM) of PGE 2 , these differences did not reach statistically significant levels (Fig. 4). IL-6 mRNA accumulation was affected by PGE 2 in the same way as TNF-␣ mRNA, i.e., IL-6 mRNA was decreased by ϳ50% by 1 M PGE 2 and only weakly decreased at lower concentrations of the prostaglandin (Fig. 4). The ϳ40 -50% inhibition of TNF-␣ and IL-6 mRNA by high concentrations of PGE 2 was evident at all time points examined (data not shown). The modest inhibition of TNF-␣ and IL-6 mRNA levels is in striking contrast to the strong inhibition observed on the protein level, i.e. 1 M, 100 nM, and 1 nM PGE 2 inhibit TNF-␣ protein production by 99%, 95%, and 50%, respectively, and IL-6 protein levels are affected in a similar way.
While inhibiting TNF-␣ and IL-6 mRNA accumulation, PGE 2 treatment increased the production of IL-1␤ mRNA (Fig. 4), demonstrating that PGE 2 has a differential effect on the mRNA accumulation of these cytokines. IL-1␤ mRNA levels were ϳ2.5fold higher in cells treated with LPS in the presence of 1 M PGE 2 in comparison with cells treated with LPS alone (Fig. 4).

Role of cAMP and PKA in Inhibition of TNF-␣ Production by PGE 2 -
The effect of PGE 2 on cells can be modulated by four subtypes of prostanoid receptors, EP1-EP4, with EP2 and EP4 receptors positively coupled to adenylate cyclase (19). A recent study (20) has shown that in microglial cells, the effects of PGE 2 are mediated by the EP2 subtype, thus implicating the importance of a rise in intracellular cAMP for the activity of PGE 2 in microglia. In agreement with these data, we found that the activity of PGE 2 can be largely mimicked by a cAMP analog, 8-bromo-cAMP. Indeed, pre-treatment of BV-2 cells with 8-bromo-cAMP for 30 min completely abolished the LPSinduced accumulation of TNF-␣ and IL-6 proteins, whereas the production of pro-IL-1␤ was not affected, and iNOS was only weakly inhibited (Fig. 5). Unlike PGE 2 , however, 8-bromo-cAMP significantly inhibited COX-2 production (Fig. 5).
We also investigated whether PKA mediates PGE 2 effects in BV-2 cells. Pre-treatment with the PKA-selective inhibitor H89 (21) only partially reversed the PGE 2 inhibition of TNF-␣ (Fig.  6A). At the same time, this concentration of H89 almost completely inhibited PGE 2 -induced activation of PKA (Fig. 6B). These data suggest that although activation of PKA by PGE 2 may represent one of the important steps in the action of PGE 2 in microglia, PKA-independent pathways are also likely to be involved.

DISCUSSION
This study addresses the question of how PGE 2 , one of the major prostaglandins produced in the brain, influences microglial activation. Here we report for the first time that in LPSstimulated BV-2 microglial cells, PGE 2 selectively suppresses the production of TNF-␣ and IL-6, whereas it has little or no effect on the accumulation of other activation-related proteins, such as pro-IL-1␤, iNOS, or COX-2. These results demonstrate the potential for modulation of specific signal transduction pathways in activated glia and provide a precedent for future attempts to develop selective therapeutics for neurodegenera- tive diseases in which glial activation and inflammatory responses are directly linked to neuropathology.
Suppression of TNF-␣ production by PGE 2 in peripheral macrophages is a well-described process; PGE 2 acts primarily by suppressing TNF-␣ transcription (22). Based on our data, it appears that transcriptional suppression is not the mechanism in microglia, or at least in the BV-2 microglial cell line. Despite the complete inhibition of TNF-␣ and IL-6 protein production, PGE 2 at high concentrations decreased the TNF-␣ and IL-6 mRNA levels by only 40 -50%. These data suggest that PGE 2 , in addition to the modest effect on TNF-␣ and IL-6 mRNA accumulation at high concentrations, acts principally by regulating protein accumulation at the level of translation or protein stability. The production of TNF-␣ in LPS-stimulated macrophages and microglia has been shown to be regulated both at the level of transcription and translation, with p38, extracellular signal regulated kinase 1/2, and c-Jun NH 2 -terminal kinase being implicated in the translational control (23,24). Interestingly, the suppression of IL-6 accumulation by PGE 2 in BV-2 cells was significant only at later time points, when the viability of LPS-treated cells is impaired (11), thus suggesting that the action of PGE 2 on IL-6 production may be mediated indirectly through a cytoprotective effect rather than a direct regulation of IL-6 protein synthesis.
As in other cell systems, PGE 2 appears to act by increasing intracellular cAMP levels, because the cAMP-mimicking agent, 8-bromo-cAMP, has the same effect as PGE 2 on TNF-␣ and IL-6 production. PGE 2 also rapidly activates PKA in BV-2 cells, and the selective inhibitor of protein kinase A, H89, partially reversed the inhibitory action of PGE 2 on TNF-␣ production. Nevertheless, the modest effect of PKA inhibition on the TNF-␣ levels suggests that additional pathways are likely to be involved. Recently, two new guanine-nucleotide-exchange factors that bind cAMP directly were described (25,26). The binding of cAMP to these factors leads to the activation of the small GTPase Rap1 that is abundantly expressed in the brain (26). One could speculate that PGE 2 stimulation of microglia might affect the production of some cytokines through modulation of Rap1 activity; however, additional studies are necessary to evaluate this and other possibilities. PGE 2 has been reported to inhibit iNOS (27) and IL-1␤ (28) production in LPS-stimulated rat microglia. In agreement with the former study (27) that found ϳ40% inhibition of iNOS, we also observed partial (ϳ30 -50%) inhibition of NO production (11) and iNOS protein (this study) by PGE 2 in BV-2 cells. However, in contrast to the latter study (28), we detected no changes in LPS-induced pro-IL-1␤ accumulation in BV-2 cells. Our findings on the protein level in BV-2 cells are substantiated by the fact that PGE 2 increased IL-1␤ mRNA levels in BV-2 cells, thus demonstrating an opposite effect of PGE 2 on the production of mRNA for pro-inflammatory cytokines. In rat primary microglia, we also did not observe inhibition of IL-1␤ production at a 3 h time point; however, the small number of microglial cells available precluded a detailed time course analysis of the IL-1␤ expression pattern. Clearly, more detailed investigation of PGE 2 action in primary microglia, including future comparative studies on human microglia, is needed to clarify this apparent discrepancy.
Overall, our data demonstrate that PGE 2 is a potent and selective inhibitor of TNF-␣ and IL-6 production in activated BV-2 cells and that the mechanism of PGE 2 action is primarily post-transcriptional, perhaps at the level of translation or protein stability. We have previously shown that a cyclopentanone derivative of prostaglandin D 2 , 15-deoxy-⌬ 12,14 -PGJ 2 , acts as an inhibitor of iNOS transcription in activated glia (11). Our additional studies 2 indicate that cyclopentanone prostaglandins also inhibit the production of TNF-␣, COX-2, IL-6, and IL-1␤. Taken together, these data suggest that there is a remarkable specificity and complexity in the effects of prostaglandins that matches the complexity of responses of glia to activating stimuli and that prostaglandins may play the role of both fine-tuning agents (PGE 2 ) and general inhibitors of glial activation (cyclopentanone prostaglandins).
TNF-␣ and IL-6 appear to play a dual role in brain injury and neurodegeneration that can include both neurotrophic and neurotoxic effects. For example, increased levels of TNF-␣ and IL-6 are observed after acute brain insult, such as stroke, in AIDS dementia, and in a number of neurodegenerative diseases, such as multiple sclerosis, Parkinson disease, and Alzheimer's disease (reviewed in Refs. 29 and 30). IL-6 was shown to promote neuronal survival and protect select populations of neurons against N-methyl-D-aspartate-induced cell death both in vivo and in vitro (31,32); on the other hand, mice deficient in IL-6 are less susceptible to experimental allergic encephalitis (33)(34)(35)(36), and overexpression of IL-6 in astrocytes leads to profound astrogliosis and neurological impairment (37). With regard to TNF-␣, some studies suggest a neurotoxic role for TNF-␣ (30), whereas recent data indicate that it might also have a neuroprotective function. For example, stimulation of neuronal nuclear factor B by TNF-␣ protects neurons against ␤ amyloid-induced cell death (38), and TNF-␣ receptor-deficient mice demonstrate diminished microglial activation and exacerbated neuronal damage after focal cerebral ischemia and epilepsy (39,40). It is tempting to speculate that PGE 2 in some circumstances might be detrimental to neuronal survival in vivo by inhibiting TNF-␣ and IL-6 production by activated microglia, and that in this context, inhibition of PGE 2 production by nonsteroidal anti-inflammatory drugs would prove beneficial. PGE 2 has been shown to play multiple roles in the central nervous system such as mediation of nociception, induction of fever, and regulation of blood flow (reviewed in Refs. 10 and 41). The implications of our findings of differential effects of PGE 2 on the production of cytokines and inflammation-related enzymes in microglial cells are that in addition to the "traditional" functions, glia-or neuron-derived PGE 2 may profoundly affect the responses of microglia to activating stimuli by modifying a panel of cytokines produced by microglia. It is likely that TNF-␣ and IL-6 are not the only proteins in activated microglia whose production is affected by PGE 2 ; however, the no-tion of selective modification by prostaglandins of glial responses to activation deserves serious attention and more detailed investigation because it may provide a rationale for the development of treatments of neurodegenerative diseases in which glial activation is considered to be a pathological component.