Profiling of Eicosanoid Production in the Rat Hippocampus during Kainic Acid-induced Seizure

Kainic acid (KA)-induced seizure in rat involves eicosanoid production in the brain, but their production mechanism and biological functions are poorly understood. We profiled the eicosanoid production during KA-induced seizure by a comprehensive lipidomics analysis using liquid chromatography-tandem mass spectrometry. Systemic KA administration caused production of large amounts of prostaglandin (PG) F2α and PGD2 in the hippocampus, with smaller amounts of other PGs and hydroxyeicosatetraenoic acids. The production was biphasic, which consisted of an initial burst in the first 30 min and a sustained late phase production. The initial phase was specific to the hippocampus and was blocked by intracerebroventricular administration of KA receptor antagonists. A selective cyclooxygenase (COX)-2 inhibitor, NS398, completely inhibited the initial phase productions, except for PGD2 and thromboxane B2, whose productions were also dependent on COX-1. These results suggest that KA signals directly stimulate the arachidonic acid cascade in the initial phase and that COX-1 and COX-2, both constitutively expressed at low levels, differentially contribute to PG productions. In the late phase, a sustained PG production in hippocampus appears due to the increased COX-2 levels even with a limited arachidonic acid supply. The present study demonstrates a dual phase regulatory mechanism of eicosanoid production during KA-induced seizure, providing a biochemical basis for understanding the biosynthesis and roles of eicosanoids in the brain.

The pathology of intractable temporal lobe epilepsy involves glutamate excitotoxicity: an activation of glutamate receptors by high concentration of glutamate causes neuronal damages in the brain (1). An excitotoxic amino acid, kainic acid (KA), 4 stimulates KA receptors, which are the members of non-N-methyl-D-aspartate ionotropic glutamate receptors. When systemically administrated to rats, KA induces a rapid and recurrent epileptic seizure, followed by a neuronal degenera-tion in specific brain areas, including hippocampus (2). KA-induced seizure has been thus studied as one of the animal models of human temporal lobe epilepsy (3).
Eicosanoids such as prostaglandins (PGs) and leukotrienes (LTs) are arachidonic acid-derived lipid mediators produced from membrane phospholipids, which exert diverse biological activities as intercellular signaling molecules through their cognate G protein-coupled receptors (4,5). Involvement of eicosanoids in the KA-induced seizure has been suggested; KA stimulates eicosanoid production in the rat brain (6), and expression levels of enzymes such as cyclooxygenase (COX)-2 (7)(8)(9)(10)(11)(12) and 5-lipoxygenase (8) are affected by KA. However, the roles of eicosanoids as well as their production mechanisms are poorly understood. In the present study, we profiled the eicosanoid production in the rat hippocampus during KA-induced seizure. Taking advantages of the simultaneous quantification method using liquid chromatographyelectrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) (13,14), we carried out a comprehensive analysis of lipid mediators in the rat hippocampal tissues. The time course of COX-2 up-regulation and the profile of lipid mediator production suggested a dual phase regulatory mechanism; a direct KA action may cause a transient burst PG production in the initial phase, whereas increased COX-2 levels mediate the late phase production under limited arachidonic acid supply. We also demonstrate a contribution of COX-1 in the production of PGD 2 and TxB 2 in the hippocampus, as well as a role of constitutively expressed COX-2, suggesting a diversity of PG production pathways in the brain.

EXPERIMENTAL PROCEDURES
Animals-Three-week-old male Wistar rats, 3-day-old male rats, and embryonic day 18 rats (SLC, Hamamatsu, Japan) were used in the study. All animal studies were conducted in accordance with the Guidelines for Animal Research at the University of Tokyo and were approved by the Animal Ethics Committee of the University of Tokyo.
Drug Administration, Tissue Collection, and Lipid Extraction-Rats were euthanized at indicated periods after intraperitoneal KA administration (10 mg/kg, dissolved in saline). Hippocampi (70 -80 mg) and cerebral cortex (70 -100 mg) were dissected, immediately frozen under liquid nitrogen, and stored at Ϫ80°C until use. The tissues were crushed to powder with an SK-100 mill (Tokken, Chiba, Japan) without thawing, and lipids were extracted with 1 ml of ethanol with a mixture of deuterium-labeled internal standards of eicosanoids. The extracts were further cleaned up with Oasis HLB extraction cartridges (30 mg, Waters, Milford, MA) as described (14).
For the intracerebroventricular administration of KA receptor antagonists, CNQX, UBP296, and UBP301 were dissolved in phosphate-buffered saline (PBS)/Me 2 SO (95:5, v/v) at a concentration of 10 mM (50 nmol in 5 l). 5 l of the solutions was injected into both of the lateral ventricles by a 10-l Hamilton microsyringe fitted with a 27-gauge ϫ 3/4-inch length needle with the following coordinates: anterior/posterior ϭ Ϫ1 mm behind bregma, medial/lateral ϭ 1 mm, and an injection depth of 3.5 mm. For COX inhibitor administration, indomethacin and NS398 were dissolved in Me 2 SO (5 mg/ml) and intraperitoneally injected at a dose of 10 mg/kg, 30 min before KA administration.
Quantification of Eicosanoids-Eicosanoids in the brain were quantified as described (14). Briefly, a TSQ-7000 triple quadrupole mass spectrometer equipped with an electrospray ionization (ESI) ion source (Thermo Electron, Waltham, MA) was operated in negative-ESI and selected reaction monitoring (SRM) mode. A reversed phase high performance liquid chromatograph system, consisting of four Shimadzu (Kyoto, Japan) LC-10A pumps and a Shimadzu CTO-10 column oven, an electrically controlled 6-port switching valve (Valco, Houston, TX), and a 3033 HTS autosampler (Shiseido, Tokyo, Japan), was connected to the MS instrument and used for the rapid resolution of PGs with Shiseido Capcellpak C18 MGS3 (1 ϫ 100 mm) column within 10 min. For the accurate quantification, an internal standard method was used. As internal standards, a mixture of deuterium-labeled eicosanoids was used. Automated peak detection, calibration, and calculation were carried out by the use of Xcalibur 1.2 software package (Thermo Electron).
Western Blot Analysis-Hippocampi were homogenized in a lysis buffer (0.34 M sucrose, 150 mM NaCl, 80 mM potassium phosphate buffer, 10 mM MgCl 2 , pH 7.4) containing a Complete TM EDTA protease inhibitor mixture (1 tablet/20 ml, Calbiochem, San Diego, CA). 8 g/lane of lysates was subjected to a 10% SDS-polyacrylamide gel electrophoresis and transferred onto an Immun-Blot polyvinylidene difluoride membrane (Bio-Rad). The membrane was blocked overnight at 4°C with BlockACE (Yukijirushi, Sapporo Japan) and incubated for 1 h at room temperature with rabbit polyclonal antibody for cPLA 2 ␣ or COX-2 diluted 1:1000 in 0.1% Triton X-100/PBS (T-PBS). After wash- ing with T-PBS, the membrane was incubated with donkey anti-rabbit IgG conjugated with horseradish peroxidase for 1 h. After washing, immunoreactive signals were visualized with an ECL Plus Western blotting detection system (Amersham Biosciences).
Statistical Analysis-Statistical analyses were carried out with Graph-Pad Prism 4.0.3 package (GraphPad Software, San Diego, CA), and p values of Ͻ0.05 were considered statistically significant.

Biphasic PG Production in the Rat Hippocampus after Systemic KA
Administration-A comprehensive analysis of eicosanoids in the rat hippocampal tissue was carried out with a LC-ESI-MS/MS multiplex quantification system. Fig. 1 shows typical chromatograms for hippocampal tissues 1 h after intraperitoneal KA injection and those from saline-injected control rats. In our system, five major PGs, 6-keto-PGF 1␣ (a stable metabolite of PGI 2 ), TxB 2 (a stable metabolite of TxA 2 ), PGF 2␣ , PGE 2 , and PGD 2 , were readily detected even in the untreated rats (Fig. 1A) and dramatically increased after KA administration (Fig.  1B). Three major LTs, LTB 4 , LTC 4 , and LTD 4 , were undetectable. Among hydroxyeicosatetraenoic acid (HETE) isomers, two major peaks of 11-HETE and 12-HETE were observed, which increased after KA administration.
Using deuterium-labeled internal standards, PGs were quantified up to 24 h after KA treatment. As shown in Fig. 2A, hippocampal tissue produced PGF 2␣ and PGD 2 as two major PGs, followed by PGE 2 , TxB 2 , and 6-keto-PGF 1␣ in this order. The PG levels reached rapidly a peak within 30 min to 1 h after KA administration and then decreased to lower levels. The production levels remained higher than the basal levels at 3 h and sustained up to 24 h. All PG species monitored showed a similar time course (Fig. 2A). These results demonstrated that hippocampal PG production after systemic KA administration was biphasic; an initial burst was followed by a sustained late phase production. Interestingly, the initial response to KA within 30 min was not observed in the cerebral cortex, suggesting that PG production at this phase is specific to hippocampus (Fig. 2B).
Primary eicosanoids produced in the hippocampal tissue can be further converted to biologically active or inactive metabolites. To evaluate this, metabolites were quantified for three major eicosanoids produced in the hippocampus, PGF 2␣ , PGD 2 , and PGE 2 . As shown in Fig. 2 (C and D), considerable amounts of 13,14-dihydro-15-keto-PGF 2␣ and 13,14-dihydro-15-keto-PGE 2 were detected proportionally to respective precursors both in the initial phase and in the late phase, suggesting that PGF 2␣ and PGE 2 are readily metabolized to these inactive compounds by 15-hydroxyprostaglandin dehydrogenase (PGDH). The levels of 13,14-dihydro-15-keto metabolite of PGD 2 as compared with its precursor PGD 2 were much smaller than those of PGF 2␣ and PGE 2 , suggesting that PGD 2 is rather stable against PGDH-mediated PG inactivation (Fig. 2, C and D). PGD 2 can also be converted to PGJ 2 , ⌬ 12,14 -PGJ 2 , and/or 15-deoxy-⌬ 12,14 -PGJ 2 , which are reportedly bioactive metabolites (15), or 11␤-PGF 2␣ (16). The amounts of these metabolites were almost undetectable or very small throughout the time course, suggesting they are not major metabolites of PGD 2 in the rat hippocampus (data not shown). Inhibition of the Initial Phase PG Production by KA Receptor Antagonists-To examine if the initial phase hippocampal PG production was mediated by KA receptors in the hippocampus, rats were pretreated (intracerebroventricularly) with CNQX (a non-N-methyl-D-aspartate receptor antagonist), UBP296 (a selective antagonist for GluR5 subunitcontaining KA receptors), or UBP301 (a selective antagonist for KA receptors) 30 min prior to the systemic KA administration (intraperitoneally). UBP296 and UBP301 have been reported as potent selective antagonists for KA receptors (reported K d values of ϳ1 M and ϳ6 M, respectively; see Ref. 17). As shown in Fig. 3 (A-E), all of the PG productions were significantly suppressed by UBP296. CNQX and UBP301 were less potent than UBP296, because they inhibited only PGF 2␣ production with statistical significance (Fig. 3, A-E). When much smaller amounts of KA (1 nmol/rat) were intracerebroventricularly administrated, the rats exhibited immediate epileptic response within several minutes, much earlier than those observed by systemic administration (45-60 min). PG profiles after intracerebroventricular KA administration were similar to those by systemic KA administration (data not shown). Collectively, these results suggest that immediate burst PG productions in the hip-pocampus were mediated by direct activation of KA receptors in the brain, probably by those expressed in the hippocampus.
Changes of Enzyme Levels in the Hippocampus after Systemic KA Administration-To understand the mechanism of KA-induced PG productions, we measured the levels of enzymes involved in the arachidonic acid cascade at mRNA and protein levels. Reverse transcription-PCR analysis of hippocampal tissues after KA treatment demonstrated an increase in the COX-2 and mPGES-1 mRNAs after KA treatment (Fig. 4A). Northern blot analyses were performed for cPLA 2 ␣, COX-1, and COX-2 (Fig. 4B). COX-2 mRNA increased transiently, reaching a peak 3-6 h after KA treatment. By contrast, cPLA 2 ␣ and COX-1 mRNAs did not show significant changes. Western blot analysis showed a low level expression of COX-2 protein in unstimulated control rats, and a significant increase was observed 3 h after KA administration, which lasted for 24 h (Fig. 4C). The amount of cPLA 2 ␣ protein did not show significant changes (Fig. 4C). These results demonstrated that KA administration stimulates COX-2 up-regulation through KA receptors; however, the time course was delayed from the initial burst PG production ( Fig. 2A). Rather, the COX-2 up-regulation was concomitant with the late phase PG production, suggesting a role of increased COX-2 in the late phase production.
Effects of COX Inhibitors on the Initial Phase PG Production-To further analyze COX-isoform dependence of the initial phase PG production, effects of COX inhibitors were examined. As shown in Fig. 5, pretreatment of rats with a COX-2-selective inhibitor, NS398, effectively suppressed the initial phase PG production. NS398 inhibited the production of PGD 2 and TxB 2 only by 50 -70%, whereas other PGs (PGF 2␣ , PGE 2 , and 6-keto-PGF 1␣ ) were almost completely inhibited. Indomethacin, a non-selective COX inhibitor, inhibited all of the PG productions. These results demonstrate that COX-2 is constitutively present in the brain and plays a major role in PG productions, and that COX-1 contributes to some extent in the production of PGD 2 and TxB 2 .

DISCUSSION
In the present study, we profiled the eicosanoid production of rat hippocampal tissues during KA-induced seizure using the LC-ESI-MS/MS multiple quantification method (13,14). One of the advantages  of our strategy is that multiple eicosanoids can be measured simultaneously. From the method, a novel insight is now provided for PG production in KA-induced seizure model.
Dual Phase Regulation of Eicosanoid Productions in the Hippocampus-Systemic KA administration caused biphasic PG production (Fig. 2). Initially, an immediate burst of PG production occurred within 30 min after KA administration. Subsequently, late phase PG production sustained to 24 h, with levels being lower than the initial phase but higher than the untreated basal levels. The prominent productions of PGF 2␣ and PGD 2 in the initial phase (Fig. 2) agree well with a previous report (6). In addition, the lack of LT synthesis in our observation (Fig. 1) illustrates that the COX pathway is dominant in the hippocampus during this period. Kawaguchi et al. (12) reported an increased PGE 2 production in the rat hippocampus 24 h after KA administration. Our results, however, demonstrate that PGF 2␣ and PGD 2 rather than PGE 2 are major metabolites even 24 h after KA stimulation.
Analysis of the metabolic fate of major PGs showed that PGF 2␣ and PGE 2 are converted to the 13,14-dihydro-15-keto form in the hippocampus, whereas PGD 2 is rather resistant to this inactivation pathway (Fig. 2, C and D). This result agrees with the substrate specificity of NAD ϩ -dependent 15-PGDH (18 -20). It is not likely that PGD 2 is converted to other metabolites such as PGJ 2 , ⌬ 12,14 -PGJ 2 , 15-deoxy-⌬ 12,14 -PGJ 2 , and/or 11␤-PGF 2␣ , because they were hardly or very weakly detected in the hippocampal tissue (data not shown). The time course of the production of 13,14-dihydro-15-keto metabolites was similar to those of precursor PGs, which reached a peak at 30 min to 1 h after KA administration (data not shown). Thus, we suppose that both PGs and their metabolites can be removed from the hippocampal tissue, possibly by blood flow or diffusion.
The biphasic profile of PG production can be explained by the mode of KA action and the enzyme levels of the arachidonic acid cascade (Fig.  6). In the initial phase, large amounts of PGs are produced rapidly without enzyme levels apparently changed (Figs. 2 and 4). In addition, this initial robust production was not observed in the cortex, suggesting that it is hippocampus-specific (Fig. 2B). Further, intracerebroventricular  Our results suggest that hippocampal PG production consists of two distinct phases: the initial phase and the late phase productions that occur within 30 min and within several hours to 24 h after systemic KA administration, respectively. Induction of COX-2 occurs concomitantly with the late phase PG productions, suggesting that COX-2 up-regulation is involved in the late phase mechanism, whereas the initial phase mechanism is independent of COX-2 up-regulation. B and C, possible mechanisms for the initial phase (B) and the late phase (C) productions. In B, KA receptor signals in hippocampal PG-producing cells (e.g. neurons or astrocytes) directly stimulate arachidonic acid cascade through intracellular calcium increase. Both COX-1 and COX-2 contribute to PG production in this phase, with differential coupling to the production of each PG species; PGD 2 and TxA 2 are produced by both COX-1 and COX-2, whereas PGI 2 , PGF 2␣ , and PGE 2 are mostly COX-2-dependent. In C, increased COX-2 levels determine the late phase PG production profiles. With lower or no KA receptor signals, arachidonate supply is much limited. Increased COX-2 activity pulls up the PG production from the basal production levels; unchanged COX-1 levels suggest that COX-1 has minimal contribution to PG productions in the late phase. administration of KA receptor antagonists blocked the PG productions in the hippocampus. All these results suggest that this process is mediated by KA receptor in the hippocampus (Fig. 3), consistent with the high density distribution of KA receptors in the hippocampus (21,22). The most probable mechanism is that activation of KA receptor increases intracellular [Ca 2ϩ ] (23), causing translocation of cPLA 2 ␣ to the membrane, as reported in various types of cells (24 -26) (Fig. 6B).
The mechanism of the late phase PG production may much differ from that of the initial phase. In the late phase, the COX-2 levels were increased by KA treatment (Fig. 4), as reported (7)(8)(9)(10)(11)(12). The COX-2 up-regulation can result in an increased production of PGs, but the levels of late phase productions were much lower than the initial phase. This can be due to the limited supply of arachidonic acid: KA receptor signals might no longer stimulate arachidonic acid release as in the initial phase (Fig. 6C). The constitutive PG productions observed in the untreated rats (Figs. 1 and 2A) suggest that, in the hippocampus, there is a significant supply of arachidonic acid in the steady state. Thus, increased COX-2 levels with this limited arachidonic acid supply may illustrate the nature of the late phase production (Fig. 6C). In contrast to the initial phase PG production that was specific to the hippocampus, we observed the late phase PG production also in the cortex (data not shown). Since it has been reported that COX-2 up-regulation is seen in the cortex as well as hippocampus (7,10), the late phase PG production mechanism may be common in hippocampus and cortex. The mechanism of KA-induced COX-2 up-regulation may involve an increased transcription and/or a stabilization of COX-2 mRNA (27). KA-induced intracellular signaling processes activate extracellular signal-regulated kinase (28,29) and p38-mitogen activated protein kinase (29,30) in the brain; these can cause COX-2 up-regulation via transcriptional activation (27,31,32) and mRNA stabilization (27,33,34), respectively. In addition, cAMP-response element-binding protein phosphorylation is affected in KA-stimulated (28) and epileptic (35) hippocampus, which may also modulate COX-2 mRNA transcription (27). Further investigations are required to elucidate the precise mechanism and physiological roles of COX-2 up-regulation by KA in the brain.
Differential Involvement of COX-1 and COX-2 in the Initial Phase PG Productions-A critical and important finding from this study is that COX-1 and COX-2 differentially contribute to PG productions in the initial phase. Our results with a COX-2-selective inhibitor, NS398, showed that COX-1 might also contribute to productions of PGD 2 and TxB 2 during KA-induced seizure (Figs. 5 and 6B). The mechanism of the differential COX-1/COX-2 contribution can be explained by the coupling efficiency of terminal PG synthases to each COX isoform. A functional coupling of COX-2 to PGE synthase (36,37) as well as COX-1 to PGD and thromboxane synthases has been reported (38,39). However, PGD 2 and TxB 2 productions seem dependent also on COX-2 in the brain, because NS398 significantly reduced their production (Fig. 5). It is possible that platelets and immune cells also produce PGD 2 and thromboxane via COX-1. In such cases, it is unlikely that KA directly stimulates these cell types, because KA effects are localized to hippocampus (Fig. 2B). Rather, KA may indirectly stimulate these cells through activating neurons or glial cells.
In Vivo Significance of the Dual Phase PG Production-Although the present study was not aimed to directly address the function of eicosanoids in epileptic seizure or neuronal degeneration, our results provide an important concept on such studies, that there are distinct mechanisms of PG productions serially occurring, i.e. the initial phase and the late phase productions. There are controversial reports on the effect of COX inhibitors; in some COX inhibition was protective to KA-induced seizure (40), and in others COX inhibition rather aggravated symptoms (41,42). Most of the studies were performed by using pretreatment with inhibitors, which blocked the initial phase PG production. Because the effective periods of inhibition may vary depending on the inhibitor and the route of administration, it is difficult to evaluate the significance of the late phase production in KA-induced seizure by a single-dose pretreatment with inhibitors. To discriminate the initial phase and the late phase PG productions, administration method of inhibitors should be well designed. Such methods may include a post-treatment with inhibitors, which conserves the initial phase intact. A recent report by Gobbo et al. (43) demonstrated that post-treatment with COX-2-selective inhibitors enhanced the functional recovery from KA-induced neuronal degeneration, whereas pretreatment aggravated the seizure symptoms, suggesting the diversity of PG roles at the beginning and at the later periods. Similar experiments with detailed monitoring of PG amounts will help in understanding the causal relations of each phase of PG production and the outcomes.
In conclusion, the present study demonstrated a dual phase regulatory mechanism of eicosanoid production in the rat hippocampus during KA-induced seizure. In addition to the temporal and spatial regulation, our results also revealed the diversity of PG-producing pathways to which COX-1 and COX-2 differentially contribute. A detailed understanding of the mechanism of lipid mediator production will provide a biochemical basis for understanding the roles of lipid mediators in seizure-related pathologies. Furthermore, recent studies reporting novel lipid mediators in the brain (44 -46) raise the possible involvement of non-eicosanoid lipid mediators in excitotoxic degeneration. Thus, our strategy using LC-ESI-MS/MS will serve as a universal tool for understanding the roles of lipid mediators in the brain.