Stress mediators regulate brain prostaglandin synthesis and peroxisome proliferator-activated receptor-gamma activation after stress in rats.

Stress exposure leads to oxidative/nitrosative and neuroinflammatory changes that have been shown to be regulated by antiinflammatory pathways in the brain. In particular, acute restraint stress is followed by cyclooxygenase (COX)-2 up-regulation and subsequent proinflammatory prostaglandin (PG) E2 release in rat brain cortex. Concomitantly, the synthesis of the antiinflammatory prostaglandin 15d-PGJ(2) and the activation of its nuclear target the peroxisome proliferator-activated receptor (PPAR)-gamma are also produced. This study aimed to determine the possible role of the main stress mediators: catecholamines, glucocorticoids, and excitatory amino acids (glutamate) in the above-mentioned stress-related effects. By using specific pharmacological tools, our results show that the main mediators of the stress response are implicated in the regulation of prostaglandin synthesis and PPARgamma activation in rat brain cortex described after acute restraint stress exposure. Pharmacological inhibition (predominantly through beta-adrenergic receptor) of the stress-released catecholamines in the central nervous system regulates 15d-PGJ(2) and PGE(2) synthesis, by reducing COX-2 overexpression, and reduces PPARgamma activation. Stress-produced glucocorticoids carry out their effects on prostaglandin synthesis through their interaction with mineralocorticoid and glucocorticoid receptors to a very similar degree. However, in the case of PPARgamma regulation, only the actions through the glucocorticoid receptor seem to be relevant. Finally, the selective blockade of the N-methyl-d-aspartate type of glutamate receptor after stress also negatively regulates 15d-PGJ(2) and PGE(2) production by COX-2 down-regulation and decrease in PPARgamma transcriptional activity and expression. In conclusion, we show here that the main stress mediators, catecholamines, GCs, and glutamate, concomitantly regulate the activation of proinflammatory and antiinflammatory pathways in a possible coregulatory mechanism of the inflammatory process induced in rat brain cortex by acute restraint stress exposure.

T HE STRESS RESPONSE is a complex and conserved mechanism that affects many different systems when an organism is threatened by a stressor, including cardiovascular and central nervous system (CNS). In particular, the response to restraint stress is characterized by three related events: 1) activation of the sympathetic nervous system, which leads to an acute release of catecholamines (CATs) (epinephrine and norepinephrine) (1); 2) a slower but more sustained increase in plasma glucocorticoid (GCs) levels (1); and 3) the release of large amounts of the excitatory amino acid glutamate (GLUT) to the extracellular space in different brain areas (cortex, preferentially; striatum; nucleus accumbens, hippocampus) (2).
These main stress mediators can induce an acute phase response, which is similar to the inflammatory response elic-ited by an organism in response to an infection or acute trauma. In this way, repeated or chronic episodes of stress may result in inflammatory diseases (3). Interestingly, inflammation may also activate the hypothalamic-pituitaryadrenal (HPA) axis by acting as a stressful stimulus (3). Several clinical and experimental studies have described a close relation between the exposure to several stressful events and the onset, evolution, and resolution of inflammation-related diseases in multiple body systems (cardiovascular, endocrine, digestive, immune, etc.) (4,5).
Specifically, several neurological and neuropsychiatric diseases, including some related to stress exposure (neurodegenerative diseases, depression, posttraumatic stress disorder, and schizophrenia), have a strong neuroinflammatory component. These pathologies include key overlapping features as glial activation, major histocompatibility complex expression, invasion of immune cells, synthesis of inflammatory mediators (cytokines, free radicals. and pro-and antiinflammatory prostaglandins) (6), and the activation of inflammatory-related nuclear transcription factors, such as the nuclear factor-B (NFB) (7) or the peroxisome proliferator activated receptor (PPAR)-␥ (8).
In this vein, a growing body of evidence identified some endogenous mediators such as the free radical nitric oxide (NO) (9,10) and prostaglandins (PGs) (10,11) as key mediators of the accumulation of oxidative and nitrosative agents and the neuroinflammatory process elicited by stress exposure in rat brain.
Whereas NO and other free radicals have been demonstrated to have an important role on stress-induced neurodegeneration (12), a more complex regulatory role has been demonstrated for cyclooxygenases (COXs) and its by-products in these conditions (13,14). PGs are a family of autocrine and paracrine molecules implicated in the regulation of numerous physiological and pathophysiological processes (15). In the CNS, PGs have been related to the control of neuroinflammatory processes, synaptic transmission, brain plasticity, regulation of fever, HPA axis activation, and sleep (16). After inflammatory/neuroendocrine challenges, different brain cellular types (neurons; glia; and vasculature-associated cells such as endothelial, leukocyte, and perivascular cells) have been identified as responsible for the prostaglandin synthesis (17).
All these compounds are generated via a complex multienzymatic pathway: PGs are produced by metabolism of arachidonic acid by COXs, this generating the intermediate substrate PGH 2 . COXs are bifunctional enzymes, with fatty acid cyclooxygenase and hydroperoxidase activities that are present in three isoforms: COX-1, COX-2, and COX-3, a splice variant of COX-1 (17,18). Whereas COX-1 is constitutively expressed as a housekeeping enzyme mediating physiological responses in nearly all tissues, COX-2 is rapidly and transiently induced and is primarily responsible for the synthesis of prostanoids involved in pathological inflammatory processes. COX-2 is constitutively expressed in only certain tissues (such as the brain cortex) but is up-regulated in a wide variety of cells by cytokines, mitogens, growth factors, and bacterial lipopolysaccharide (19). Subsequently PGH 2 is further metabolized by tissue-specific prostaglandin synthases into different prostanoids, such as PGE 2 , PGD 2 , PGF 2 , PGJ 2 , and thromboxane A 2 . In the case of PGE 2 synthesis, microsomal PGE 2 synthase-1 (m-PGES-1) is the most important and studied E synthase-type enzyme in brain cortex (20), and for 15d-PGJ 2 synthesis, lipocalin-type PGD 2 synthase (L-PGDS) is the main one (21).
COX-2 up-regulation in different brain areas after stress exposure was first described by Yamagata et al. (22). We have shown that the PG synthesis pathway is activated by simple exposure (6 h) to restraint stress in rat cerebral cortex through COX-2 overexpression and subsequent PGE 2 production in a NFB-dependent mode (13). This up-regulation has been described to occur in different physical, psychological, or mixed stress protocols, such as crowding, immobilization, or cold restraint (23)(24)(25). Restraint stress also induces a COX-2-dependent increase of the antiinflammatory prostaglandin 15d-PGJ 2 and in the expression and activity of its nuclear target, the PPAR␥, in brain cortex (26,27).
PPAR␥ is a transcription factor that belongs to the superfamily of hormone nuclear receptors (28) widely expressed in brain cortex (29), exerting antiinflammatory activities in brain cells by reducing the release of proinflammatory cytokines, inducible NO synthase (NOS)-2, and COX-2 expression or interfering with inflammatory transcription pathways, such as activator protein-1 or NFB (30 -32). Thus, activation of PPAR␥ is considered as a mechanism able to halt the inflammatory response. Indeed, in the past 2 yr, an increasing number of studies have reported on the protective effects of PPAR␥ agonists in animal models of inflammatory-related neurologic diseases (33). The compensatory, antiinflammatory role of the COX-2-derived 15d-PGJ 2 -PPAR␥ pathway has been demonstrated to occur in brain prefrontal cortex by using natural, endogenous, and pharmacological ligands of PPAR␥ in rats acute or chronically stressed by restraint (26,27,34).
Taking into account the inherent limitations of any systemic pharmacological manipulation over the complex actions of the stress mediators in vivo, the present study was aimed to determine the possible role of CATs, GCs and GLUT in the regulation of prostaglandin synthesis and PPAR␥ activation previously observed in an single exposure to restraint stress model in rat brain cortex, a brain area with high levels of COXs and PPAR␥ proteins and very susceptible to the neuroinflammatory process elicited by restraint stress exposure.

Materials and Methods Animals
Adult male Wistar: Hann (Harlan, Barcelona, Spain) rats weighing 225-250 g were used. All experimental protocols adhered to the guidelines of the Animal Welfare Committee of the Universidad Complutense following European legislation (November 24, 1986; DC 86/609/EEC). The rats were housed individually under standard conditions of temperature and humidity and a 12-h light, 12-h dark cycle (lights on at 0800 h) with free access to food and water. All animals were maintained under constant conditions for 7 d before stress.

Restraint stress
Rats were exposed to stress between 0900 and 1500 h in the animal homeroom. The restraint was performed using a plastic rodent restrainer that allowed for a close fit to rats during 6 h (35) in their home cages. Restraint is a stress model classified as physical-psychological, predictable, and no escapable suitable to study neurochemical effects of stress on brain function. It is a very complex challenge in which multiple elements such as heat, novelty, and forced immobility are minimized (reviewed in Ref. 36). In our model some of these elements were controlled: 6 h of restraint stress did not produce significant changes in body temperature, body weight, and respiration (35). Stressed animals were killed immediately after restraint (still in the restrainer) using sodium pentobarbital (320 mg/kg ip). Control animals were not subjected to stress but were handled at 0900 h for few seconds; food and water were removed and were killed at the same time of the stressed animals (1500 h). Blood for plasma determinations was collected by cardiac puncture and anticoagulated in the presence of trisodium citrate [3.15% (wt/vol), 1 vol citrate per 9 vol blood]. After decapitation, the brain was removed from the skull, and after careful removing of meninges and blood vessels, the prefrontal cortical areas from both brain hemispheres were excised and frozen at Ϫ80 C until assayed.
This single stress exposure was chosen according with the time course of activation of pro-and antiinflammatory pathways in brain cortex after stress: previous studies using this protocol have shown that soon after the onset of stress (20 -60 min), GLUT and cytokines (TNF␣ and IL-1␤) increase in many brain areas including cortex; NFB and COX2 after 4 h; and inducible NOS after 6 h (reviewed in Ref. 10). The levels of proinflammatory PGE 2 increase after 4 h from the onset of stress (4 h: 128%; 6 h: 208% from control, 27.8 Ϯ 2.6 pg/mg protein). The time course of 15d-PGJ 2 release after single restrain exposure also shows an increase after 4 h from the onset in parallel with COX-2 activation (4 h: 153%; 6 h: 205% from control, 489.89 Ϯ 62.5 pg/mg protein).

Western blot analysis
To determine COX-1, COX-2, L-PGDS, m-PGES-1, and PPAR␥ protein expression levels, tissues were homogenized in the same buffer used for quantification of prostaglandins (see above), except for microsomal PGE 2 synthase in which Nonidet P-40 (0.2%) was added, and the supernatants resulting after centrifugation of 12,000 ϫ g for 20 min were used. After centrifugation in a microcentrifuge for 15 min, the proteins present in the supernatants were loaded (20 g) and size separated in 10% SDS-PAGE (90 -100 mV). The proteins were electroblotted onto polyvinylidene difluoride membranes (Millipore, Bedford, MA). Afterward the membranes were blocked in 10 mm Tris-buffered saline containing 0.1% Tween 20 and 5% skim milk, the membranes were, respectively, incubated with specific goat polyclonal COX-1 (raised against a epitope mapping near the C-terminus of COX-1 of human origin); COX-2 (peptide placed at the C terminus of COX-2 of mouse origin); L-PGDS (epitope situated near the N terminus of PGD 2 synthase of mouse origin) (Santa Cruz Biotechnology, Santa Cruz, CA; Ann Arbor, MI; 1:1000); rabbit polyclonal m-PGES-1 (raised against the peptide of human microsomal PGE synthase amino acids 59 -75 (CRSDPDVER-SLRAHRND) (Cayman Chemicals, Ann Arbor, MI; 1:500); and mouse monoclonal PPAR␥ [raised against a sequence mapping at the C terminus of PPAR␥ of human origin (identical with corresponding murine sequence)] (Santa Cruz, 1:1000) primary antibodies. After washing the membranes with 10 mm Tris-buffered saline containing 0.1% Tween 20, the membranes were incubated with the respective horseradish peroxidase-conjugated secondary antibodies for 90 min at room temperature. Proteins recognized by the antibody were visualized on x-ray film by chemiluminescence following the manufacturer's instructions (Amersham, lbérica, Spain). Autorradiographs were quantified by densitometry (Scion Image for W2000; Scion Corp., Frederick, MD), and several exposition times were analyzed to ensure the linearity of the band intensities. All densitometries are expressed in arbitrary units (AU). In all the Western blot analyses, the house keeping gene ␤-actin was used as loading control (blots shown in the respective figures).

PPAR␥ transcription factor assay
PPAR␥ transcription factor activity was determined on nuclear extracts by using an ELISA-based kit, which allows to detect and quantify the specific transcriptional activity of PPAR␥ (Cayman Chemicals). Briefly, nuclear extracts were incubated in a multiwell plate coated with specific peroxisome proliferator response element probes, and PPAR bounded to the peroxisome proliferator response element probe was detected using a specific antibody against the ␥-isoform. Horseradish peroxidase-labeled secondary antibody was added and the binding was detected by spectrophotometry. Measurement was performed according to the manufacturer's instructions. This assay is specific for PPAR␥ activation, and it does not cross-react with other PPAR isoforms such as PPAR␣ or PPAR␤-␦.

Plasma corticosterone levels
Plasma was obtained from blood samples (see above) by centrifuging the sample at 1000 ϫ g for 15 min immediately after stress. All plasma samples were stored at Ϫ20 C before assay by using a commercially available kit by RIA of 125 I-labeled rat corticosterone (Diagnostic Products Corp., Los Angeles, CA). A ␥-counter was used to measure radioactivity of the samples. The values obtained in control animals (192 Ϯ 7 ng/ml) are in accordance with the kit manufacturer's expected values in adult male Wistar rats at the time of blood extraction (1500 h).
MK was dissolved in saline, MET, and RU in 10% propylenglycol and SP, PHE, and PRO in 5% ethanol. All drugs and vehicles were injected ip 1 h before the onset of stress (0800 h). None of the parameters studied were modified in the three different vehicle-treated groups of rats when compared with noninjected animals. To simplify figures, these three groups were unified in only one vehicle-injected group in control (CϩVEH) and stress (S6ϩVEH) conditions. Each experimental group included at least six animals.

Adrenalectomy
A number of animals were also bilaterally adrenalectomized (ADX) under general anesthesia using isoflurane. A small incision was made along the midline of the back just below the rib cage, connective tissue and fat was displaced, and a small hole was made through the muscle using blunt cut scissors. The adrenals were excised with curved forceps. The integrity of each removed gland was examined after its excision. Incisions were closed and treated with surgical antiseptic solution. After surgery, all ADX animals were immediately given 0.9% saline to drink to maintain their salt balance. Surgery for sham animals was identical, but the adrenals were not removed. Animals were allowed 5 d postoperative recovery.

Protein assay
Proteins were measured using the bicinchoninic acid method (52).

Chemicals and statistical analyses
Unless otherwise stated, the chemicals were from Sigma (Madrid, Spain). Data in text and figures are expressed as mean Ϯ sem of the indicated number of experiments. For multiple comparisons, a one-way ANOVA followed by the Newman-Keuls post hoc test to compare all pairs of means between the control group and the rest of the groups. Furthermore, all the different stressϩtreatments groups were compared with the stressϩvehicle group using the same statistical analysis. P Ͻ 0.05 was considered statistically significant.

Regulation of cortical 15d-PGJ 2 and PGE 2 production in control and stress conditions by GCs, CATs, and GLUT
Single exposure (6 h) to restraint produces a marked increase in 15d-PGJ 2 (P Ͻ 0.05 vs. control, 427.83 Ϯ 70 pg/mg protein) and PGE 2 (P Ͻ 0.05 vs. control, 38 Ϯ 5 pg/mg prot) levels in rat cortex, compared with controlϩvehicle conditions (Fig. 1, A and B).
Preadministrations (1 h before the onset of stress) of the glucocorticoid synthesis inhibitor MET, the GR antagonist RU, the MR antagonist SP, the ␤-blocker PRO, or the NMDA glutamate receptor blocker MK prevents the overproduction of 15d-PGJ 2 observed after single-exposure stress (6 h) (Fig. 1A). Blockade of the ␣-receptor with PHE did not prevent this effect (Fig. 1A).
The stress-induced PGE 2 production was reduced by all treatments (Fig. 1B).

COX expression in control and stress conditions by GCs, CATs, and GLUT
To test whether the above-mentioned effects in PGs synthesis were correlated with changes in COX protein expres-sion levels, Western blot studies were made on cortical samples from control and stressed animals: Single exposure to restraint stress produces an increase of COX-2 protein levels in rat brain cortex. However, previous administration of MK and PRO prevented this increase ( Fig.  2A). Cortical COX-1 protein levels were not affected by stress or any particular drug pretreatments (Fig. 2B).

L-PGDS and m-PGES-1 expression in control and stress conditions by GCs, CATs and GLUT
To check whether the aforementioned effects in PG synthesis were correlated with modifications in the expression levels of the respective PG synthases, Western blot analysis was carried out. Both stress and the respective pretreatments do not modify the protein levels of the particular synthases, m-PGES-1 and L-PGDS (Fig. 3), suggesting that modification in 15d-PGJ 2 and PGE 2 levels (Fig. 1) are mainly due to the effect of the drugs on COX-2 protein.

PPAR␥ expression in control and stressed rat brain cortex by GCs, CATs, and GLUT
To further investigate the effects of the respective treatments in the COX-2/L-PGDS/15d-PGJ 2 /PPAR␥ antiinflammatory pathway, PPAR␥ protein expression was studied in brain cortex. Single exposure to restraint stress enhanced PPAR␥ protein expression (Fig. 4). Furthermore, all the treatments except for SP and PHE completely blocked this stressinduced PPAR␥ overexpression (Fig. 4). None of the treatments modified PPAR␥ expression in control, nonstressed animals.

Regulation of PPAR␥ activity in control and stressed rat brain cortex by GCs, CATs, and GLUT
As in the case of PPAR␥ expression, single exposure to restraint stress (S6) produced a significant up-regulation of PPAR␥ transcriptional activity in nuclear extracts obtained from rat brain cortex samples. This increase was prevented by the respective pretreatment with MK, MET, RU, or PRO (Fig. 5). Interestingly, all these pharmacological tools prevented the stress-induced increase in PPAR␥ activity. In stress conditions, MK, MET, RU, and PRO also decreases basal (vehicle-treated control animals) transcriptional activity (Fig. 5). Similarly to PPAR␥ expression, SP and PHE did not modify PPAR␥ activity. Basal PPAR␥ transcriptional activity remained unchanged in all groups when compared with vehicle-treated control group (Fig. 5).

Effects of adrenalectomy in prostaglandin levels and COX-2 and PPAR␥ expression in control and stressed rat brain cortex
To confirm the results obtained with the pretreatment of MET (pharmacological adrenalectomy), a complementary study with ADX rats in control and acute stress conditions was carried out. Adrenalectomy (S6ϩADX) prevented the overproduction of 15d-PGJ 2 and PGE 2 observed after single exposure stress (S6 SHAM) (Fig. 6A). Furthermore, the surgical ablation of the adrenal glands blocked the stress-induced increase in COX-2 and PPAR␥ expression, analyzed by Western blot (Fig. 6B).

Effects of the drugs used on plasma corticosterone levels
To determine the interactions between the aforementioned pharmacological tools and the stress response, corticosterone levels were measured in the plasma obtained from rats treated with the different drugs. As shown in Fig. 7, stress exposure caused an increase in corticosterone as expected by such stimulus.
The administration of the GR antagonist RU resulted in increased corticosterone concentration after stress when compared with the value obtained in the group of animals stressed in the absence of RU (Fig. 7). The inhibition of glucocorticoid synthesis by MET had the opposite effect, causing a reduction of about 50% in stress-induced corticosterone levels (Fig. 7), indicating that, in the conditions used, MET-induced reduction of glucocorticoids is able to interfere with the HPA pathway and the subsequent synthesis of corticosterone. Similarly, the blockade of ␤-adrenergic receptors by PRO reduced the stress-induced corticosterone levels.
The plasma corticosterone levels of control and stressed adrenalectomized rats are under the detection limits of the RIA kit (data not shown), validating the surgical ablation of the adrenal glands in these two groups of animals.

Discussion
The synthesis of prostaglandins and the activation of PPAR␥ detected in the cortices of stressed animals are regulated by CATs, GCs, and GLUT. The data presented here indicate that the blockade of CAT actions in the CNS after a single stress exposure can inhibit the overexpression of   6. A, 15d-PGJ 2 and PGE 2 levels in brain cortex of control (CϩADX) and stressed adrenalectomized (S6ϩADX) rats and their respective controls without adrenalectomy (sham). Data represent the mean Ϯ SEM of six rats. *, P Ͻ 0.05 vs. C; #, P Ͻ 0.05 vs. S6 (Newman-Keuls test). B, Western blots and densitometric analyses of the protein COX-2 and PPAR␥ in control and acute stressed rats with (ϩADX) and without adrenalectomy (SHAM). Each band is representative of the mean of three independent experiments Ϯ SEM. *, P Ͻ 0.05 vs. control SHAM; #, P Ͻ 0.05 vs. S6 SHAM (Newman-Keuls test).
COX-2, which leads to a reduction of 15d-PGJ 2 and PGE 2 synthesis and the activation of PPAR␥. This is mainly a ␤-adrenergic-dependent process, and in the case of PGE 2 , it is also ␣-adrenergic receptor dependent. Whereas glucocorticoids binding to both MR and GR type of receptors acts as a stimulus for the synthesis of PGs, only the actions through GR seem to be relevant for the regulation of PPAR␥. Finally, the key role of excitatory amino acids in stress-induced consequences in brain is also demonstrated here: the overactivation of the NMDA glutamate receptor caused by stress participates in the induction of COX-2 and PPAR␥ and the production of 15d-PGJ 2 and PGE 2 .
During the early phases of stress response, catecholamines have been identified as regulators of the production of proinflammatory-related molecules, such as cytokines (TNF-␣, IL-1␤), NFB, and NO (reviewed in Ref. 3). Our results are in agreement with previous studies showing that CATs (norepinephrine) stimulate the synthesis of PGE 2 via mechanisms dependent on both types of adrenergic receptors (53)(54)(55), although the opposite role of ␣-adrenergic receptors in control conditions remains controversial. Certainly further work is necessary to explain this unexpected emerging role of ␣-adrenergic receptors, but individual differences explain similar findings of the ␣-blocker prazosin on corticosterone levels in control and stressed animals (56). The different levels of other inflammatory-related molecules as NFB, cytokines, or GCs in control and after stress conditions could be a possible explanation.
Interestingly, CATs also regulate the antiinflammatory prostaglandin 15d-PGJ 2 overrelease previously shown in this model of stress (26) through a ␤-adrenergic-related mechanism. Emerging literature suggests a predominant antiinflammatory role of CATs in several models of neurotoxicity, a role that seems to be exerted through ␤-adrenergic-type receptors, further confirming our results (55). In this way, CATs could control COX-2-dependent proinflammatory and antiinflammatory prostaglandin synthesis in a balanced mechanism. In support of this idea, some authors recently identified 15d-PGJ 2 as a potent blocker of m-PGES-1 mRNA induction by IL-1␤ or lipopolysaccharide in chondrocytes, synovial fibroblasts, and macrophages (57)(58)(59). In this vein, we previously demonstrated that the administration of supraphysiological doses of 15d-PGJ 2 can prevent the release of PGE 2 and COX-2 up-regulation after a single exposure to restraint stress (26).
The regulation of the different prostaglandins synthesis by GCs during a single exposure to restraint stress seems to be a relatively complex process. Although the data presented here represent an advance over ex vivo or in vitro studies, it is necessary to address the limitations of the in vivo approaches to interpret the results. For instance, the pharmacological modulation of GRs after systemic administration of drugs may cause changes in the stress system (GR-dependent negative feedback on the HPA axis) (60). Furthermore, all these changes may affect the prostaglandin synthesis when trying to extrapolate these results to chronic stress exposure or chronic treatments.
With these limitations in mind, we report here that the glucocorticoids secreted after restraint stress exposure seem to up-regulate the synthesis of the proinflammatory prostaglandin PGE 2 in brain via both types of receptors. This is a paradoxical result, taking into account the classical therapeutic use of GCs in diverse pathologies (autoimmune diseases, chronic inflammation, or transplant rejection), acting as immunosuppressor and antiinflammatory (mainly due to their capacity to inhibit COX activity) in several body systems, including the brain (22,61,62).
In this vein, we have demonstrated that, in the presence of high GCs levels, a strong inflammatory response is induced in brain along with the accumulation of oxidative and nitrosative mediators (11). This inflammatory response is characterized by the activation of proinflammatory nuclear factors (NFB), up-regulation of NOS-2 and COX-2 with the subsequent NO and PGE 2 production, and the release of proinflammatory cytokines (TNF␣ and IL-1␤). All these effects decisively contribute to an increase in the susceptibility to cerebral damage after exposure to high GC levels (63).
Adding to the complexity of the GC effects on inflammatory processes, we demonstrate here that the same stress protocol is able to enhance the synthesis of antiinflammatory mediators as 15d-PGJ 2 and its nuclear target, PPAR␥ (26,27), through a mechanism related to both types of glucocorticoid receptors.
An interesting hypothesis derived from our results is that the glucocorticoids secreted after stress exposure could regulate the balance between proinflammatory (PGE 2 ) and antiinflammatory mediators (15d-PGJ 2 and PPAR␥) in a possible endogenous mechanism of brain plasticity against the cellular damage caused by excessive inflammation (27). Further studies are necessary to demonstrate the status of this balance in chronic stress conditions and other stress protocols.
Unexpectedly, in our model, the pharmacological inhibition of the synthesis of GCs and the blockade of MR-and GR-dependent actions inhibit prostaglandin synthesis after stress in the brain without affecting COX-2 or specific prostaglandin synthases expression, results that indirectly indicate possible effects of GCs in the catalytic activity or protein stability of these enzymes (64). Using a different model, García-Fernandez et al. (65) identified the synthetic glucocor- ticoid dexamethasone as a potent inducer of L-PGDS in mouse neurons. The apparent disparity between both studies may be due to the differences among dexamethasone and endogenous GCs in blood-brain barrier permeability and receptor-type affinity (66). Nevertheless, in our stress model, a possible effect of GCs in the expression or activity of other prostaglandin synthases, such as cytosolic PGE synthase or hematopoietic PGD 2 synthase cannot be completely ruled out.
Interestingly, adrenalectomy prevents the COX-2 overexpression observed after acute stress exposure. In line with the results obtained with the other pharmacological modulators of the GC pathway used in our study, we can speculate that this differential effect shown after adrenalectomy could be related with the stronger suppression of GC synthesis, compared with the treatment with MET and the synergic inhibition of the medullar catecholamines release.
The fact that RU administration increased corticosterone concentration after stress may be attributed to the blockade of the self-inhibitory feedback of corticosterone due to the effect of RU on the GRs (61). The interference between noradrenaline and its ␤-receptors is known to have an inhibitory effect on not only the secretion of corticosterone or stressassociated behaviors (67) but also the synthesis of the GRs (68). According to this, the alteration of corticosterone levels by PRO described here confirms the interaction of the two systems and allows us to attribute some of the effects of PRO to an inhibition not only of the ␤-adrenergic-dependent pathway but also to some extent to other mechanisms regulated by corticoids.
Finally, the stress-induced stimulation of the NMDA type of glutamate receptor is implicated in prostaglandin synthesis, as demonstrated by using MK. Previously our group and others demonstrated the control of PGE 2 production after stress by NMDA receptors (13). Here we show that the activation of NMDA receptors also up-regulates 15d-PGJ 2 synthesis in control and stress conditions in a COX-2-dependent mechanism.
Several animal models of neurodegeneration, including single exposure to restraint stress, demonstrated that the up-regulation of COX-2 is implicated in the excitotoxic processes generated by NMDA excessive activation (13). The mechanisms through which COX-2 might exert its neurotoxic effects are still partially unknown, but they seem to be due to the large quantities of free radicals generated by cyclooxygenase activity and also to its principal products like PGE 2 , which are neurotoxic per se (69 -71), although other authors have found certain neuroprotective action for this compound, depending on the type of PGE 2 receptor activated (72,73). In our single exposure to restraint stress paradigm, glutamate actions through its NMDA-type receptor regulate the brain synthesis of typically proinflammatory prostaglandins like PGE 2 (74) and also antiinflammatory ones such as 15d-PGJ 2 (26) in what might constitute a regulatory mechanism of the inflammatory process generated after NMDA excessive activation. In this vein we and other authors (34,75,76) found the administration of supraphysiological doses of 15d-PGJ 2 to be neuroprotective against excitotoxic insults in in vivo and in vitro models.
As in the case of its endogenous ligand 15d-PGJ 2 , PPAR␥ expression and transcriptional activity are modulated by all stress mediators studied: glutamate (NMDA dependent), catecholamines (␤-adrenergic type), and glucocorticoids (GR type). PPAR␥ regulation is not yet completely understood, but one of the main mechanisms may be the ligand availability (77). This kind of regulation could be important in our stress model because we found a 3-to 4-fold increase in 15d-PGJ 2 levels, compared with control conditions. In addition to this, all the drugs that inhibit PPAR␥ activation also reduce 15d-PGJ 2 synthesis (78). A tight relation exists between GC type I receptor and PPAR because it has been demonstrated in in vitro studies in which GCs induce PPAR␣ mRNA synthesis in isolated human adipocytes in a GR-dependent manner (as our data show) (79) and in vivo with the synthetic glucocorticoid dexamethasone as a potent inducer of PPAR␣ mRNA binding GR in hepatic tissue (80).
Some authors have shown that norepinephrine is able to up-regulate PPAR␥ expression in a ␤but not ␣-adrenergic receptor-type-dependent mechanism, an effect considered as antiinflammatory and neuroprotective in animal models of Alzheimer's disease in which PPAR␥ expression is reduced (81).
To our knowledge, this is the first time that a relation between stress-induced NMDA receptor activation and PPAR␥ activation has been demonstrated. As in the case of its natural ligand 15d-PGJ 2 , this activation could represent a possible endogenous neuroprotective mechanism against the excitotoxic process generated. Of course, further studies are required to explore the precise mechanism implicated. Accordingly, recent studies from our group show that PPAR activation by 15d-PGJ 2 or rosiglitazone provide neuroprotection through up-regulation of the main glutamate transporter in glia excitatory amino acid transporter-2 in animal models in which its expression and activity are reduced such as subchronic restraint stress (34) or ischemic preconditioning (82).
In conclusion, main stress mediators CATs, GCs, and GLUT concomitantly modulate the activation of proinflammatory (COX-2/m-PGES-1/PGE 2 ) and antiinflammatory (COX-2/L-PGDS/15d-PGJ 2 /PPAR␥) pathways in a possible complex mechanism regulating the inflammatory process induced by single exposure to restraint stress exposure in rat brain cortex. Furthermore, more detailed studies are necessary to explain the possible interaction among these three stress mediators in the regulation of brain cortex prostaglandin synthesis and to identify the origin (vascular or within brain parenquima) and the precise role of each cellular type implicated in our stress model.
The study of the balance and time course among these proand antiinflammatory pathways in specific brain areas related to pathological conditions, in which the inflammatory overreaction contributes to the cellular damage observed, seems to be critical to find new possible therapeutic strategies.