Postsynaptically Synthesized Prostaglandin E2 (PGE2) Modulates Hippocampal Synaptic Transmission via a Presynaptic PGE2 EP2 Receptor

Introduction

Cyclooxygenases (COXs) are the rate-limiting enzymes that convert arachidonic acid (AA) to prostaglandins. Of three COX isozymes (Vane et al., 1998; Bazan and Flower, 2002; Chandrasekharan et al., 2002), COX-2 has become the focus of growing attention, because selective COX-2 inhibitory drugs target its high inducibility by inflammation. Moreover, this enzyme is important to neuronal signaling and has been implicated in epilepsy, Alzheimer's, and other neurologic diseases. Expression of COX-2 is regulated by NMDA receptor-dependent synaptic activity and by a high-frequency stimulation (HFS) associated with long-term potentiation (LTP) induction (Yamagata et al., 1993). Furthermore, COX-2 is localized in excitatory neuronal dendrites (Kaufmann et al., 1996). These previous studies suggested that COX-2 is involved in synaptic transmission and plasticity. Recently, direct evidence that COX-2 participates in hippocampal long-term synaptic plasticity has been provided (Chen et al., 2002). Selective COX-2 inhibitors, but not those of COX-1, reduce HFS-induced LTP in hippocampal perforant path-dentate granule cell synapses. Importantly, the COX-2 inhibitor-induced reduction of LTP can be reversed by exogenous addition of prostaglandin E2 (PGE2) but not by prostaglandin D2 (PGD2) or prostaglandin F2α (PGF2α) (Chen et al., 2002). Because PGE2 is derived mainly from the COX-2 pathway (Brock et al., 1999; Vidensky et al., 2003; Chen and Bazan, 2005), it is likely that PGE2 may be a key messenger in COX-2-mediated regulation of hippocampal synaptic transmission and plasticity (Bazan, 2001; Chen et al., 2002; Chen and Bazan, 2005a,b). However, mechanisms by which PGE2 participates in synaptic signaling are not well understood.

Four subtypes of PGE2 receptors (EP1-4) have been cloned (Boie et al., 1997; Narumiya et al., 1999; Breyer et al., 2001), all displaying seven-hydrophobic-transmembrane-segment architecture typical of G-protein-coupled receptors and evoking cellular responses via distinct intracellular signaling (Narumiya et al., 1999; Breyer et al., 2001). EP2 and EP4 receptors couple with the G_s-adenylyl cyclase (AC)-cAMP pathway and increase cAMP levels. In contrast, activation of EP3 receptors inhibits cAMP generation via a pertussis toxin-sensitive G_i-coupled mechanism. EP1 receptors couple with the G_q-phospholipase C-IP₃ pathway. All four subtypes of EPs are expressed heterogeneously in the hippocampus (Zhu et al., 2005), and the expression of EP2 and EP3 is more abundant in hippocampal neurons. Because EP2 and EP3 are coupled to G_s and G_i, this implies that EP2 and EP3 may be functionally relevant in the hippocampus.

The purpose of the present work was to examine how PGE2 elicits its action as a messenger in synaptic transmission and its relationship with COX-2-mediated synaptic modifications. Our results indicate that PGE2 is a retrograde messenger in excitatory synaptic transmission through a presynaptic EP2 receptor that increases the probability of glutamate release at hippocampal synapses. Moreover, endogenous PGE2 is an important modulator of synaptic physiology, and its accumulation (e.g., increased COX-2 expression) may contribute to synaptic dysfunction and lead to pathologic conditions.

Materials and Methods

Hippocampal slice preparation. Hippocampal slices were prepared from 7- to 14-week-old FVB mice (Chen et al., 2001, 2002; Chen and Bazan, 2005). Briefly, after decapitation, brains were removed rapidly and placed in cold oxygenated (95% O₂, 5% CO₂) low-Ca²⁺/high-Mg²⁺ slicing solution composed of the following (in mm): 2.5 KCl, 7.0 MgCl₂, 28.0 NaHCO₃, 1.25 NaH₂PO₄, 0.5 CaCl₂, 7.0 glucose, 3.0 pyruvic acid, 1.0 ascorbic acid, and 234 sucrose. Slices were cut at a thickness of 400 μm and transferred to a holding chamber in an incubator containing oxygenated artificial CSF (ACSF) composed of the following (in mm): 125.0 NaCl, 2.5 KCl, 1.0 MgCl₂, 25.0 NaHCO₃, 1.25 NaH₂PO₄, 2.0 CaCl₂, 25.0 glucose, 3 pyruvic acid, and 1 ascorbic acid at 36°C for 0.5-1 h and maintained in an incubator containing oxygenated ACSF at room temperature (∼22-24°C) for >1.5 h before recordings. Slices were then transferred to a recording chamber where they were perfused continuously with the 95% O₂, 5% CO₂-saturated standard ACSF at ∼32-34°C. Individual dentate granule neurons or CA1 pyramidal neurons were viewed with a Zeiss (Oberkochen, Germany) Axioskop microscope, fitted with a 60× (Olympus, Melville, NY) water-immersion objective and differential interference contrast optics.

Primary hippocampal neuron culture. Primary hippocampal neurons were grown in culture as described previously with some modifications (Chen and Bazan, 1999). Briefly, hippocampi were dissected out from FVB mouse pups [postnatal day 0 (P0) to P1]. Tissue was incubated in oxygenated trypsin for 10 min at 37°C and then mechanically triturated. Cells were spun down and resuspended in Neurobasal/B27 medium (Invitrogen, San Diego, CA) supplemented with 0.5 mm l-glutamine, penicillin/streptomycin, and 25 μm glutamate. Cells (1 × 10⁶) were loaded into poly-d-lysine-coated 35 mm culture dishes for electrophysiological recordings and into six-well plates for real-time PCR analysis. Cells (4 × 10⁴) were plated on poly-d-lysine-coated glass coverslips for immunocytochemistry. In certain experiments in which the proliferation of glial astrocytes was inhibited, cultures were treated with 5-10 μm cytosine arabinoside (Arac). One-third to one-half of the culture medium without glutamate was changed every 2-3 d. Cultures were used between 10 and 15 d in vitro (DIV).

Electrophysiological recordings. Whole-cell patch-clamp recordings were made using an Axoclamp-2B patch-clamp amplifier in bridge mode for the current-clamp recordings. Recording pipettes (3-5 MΩ for CA1 pyramidal neurons and 5-7 MΩ for dentate granule neuron recordings) were pulled from borosilicate glass with a micropipette puller (Sutter Instruments, Novato, CA). The internal pipette solution contained the following (in mm): 120 potassium gluconate, 20 KCl, 4 NaCl, 10 HEPES, 0.5 EGTA, 0.28 CaCl₂, 4 Mg₂ATP, 0.3 Tris₂GTP, and 14 phosphocreatine, pH 7.25 with KOH. The resting membrane potential for recorded cells was between -62 to -74 mV for CA1 pyramidal neurons and -75 to -86 mV for dentate granule neurons. EPSPs were recorded in response to Schaffer collateral or perforant path stimulus at a frequency of 0.05 Hz via a bipolar tungsten electrode. Paired-pulse stimulation was induced by delivering two pulses with an interpulse interval of 80-100 ms (Zucker, 1989; Chen et al., 2001). Paired-pulse ratio (PPR) was calculated as P2/P1 (P1, the amplitude of the first EPSP; P2, the amplitude of the second EPSP). Series resistance was monitored during recordings by injection of a hyperpolarizing current (50 pA) before delivery of a stimulus (Chen, 2004). Miniature EPSCs (mEPSCs) were recorded in primary hippocampal neurons in culture under voltage clamp using an Axopatch-200B amplifier. The membrane potential was held at -70 mV. The internal pipette solution contained the following (in mm): 120 Csgluconate, 20 KCl, 4 NaCl, 10 HEPES, 0.5 EGTA, 0.28 CaCl₂, 4 Mg₂ATP, 0.3 Tris₂GTP, and 14 phosphocreatine, pH 7.25 with KOH. The external solution contained the following (in mm): 130.0 NaCl, 2.5 KCl, 1.0 MgCl₂, 10.0 HEPES, 1.25 NaH₂PO₄, 2.0 CaCl₂, 25.0 glucose, pH 7.4 with NaOH. TTX (0.5-1 μm) and bicuculline (10 μm) were added in the external solution. The frequency, amplitude, and rising and decay kinetics were analyzed using the MiniAnalysis program (Synaptosoft, Fort Lee, NJ).

RNA isolation and DNase treatment. Total RNA was prepared from harvested cells with the RNeasy Mini kit (Qiagen, Hilden, Germany) and treated with RNase-free DNase (Qiagen) according to the manufacturer's instructions. The RNA concentration was measured by spectrophotometer (DU 640; Beckman Instruments, Fullerton, CA). RNA integrity was verified by electrophoresis in a 1% agarose gel.

Reverse transcription and real-time PCR. The iScript cDNA synthesis kit (Bio-Rad, Hercules, CA) was used for the reverse transcription reaction. We used 1 μg of total RNA with 4 μl of 5× iScript reaction mix and 1 μl of iScript reverse transcriptase. The total volume was 20 μl. Samples were incubated for 5 min at 25°C. All samples were then heated to 42°C for 30 min, and reactions were stopped by heating to 85°C for 5 min.

Real-time reverse transcription-PCR (RT-PCR)-specific primers for COX-2, microsomal PGE synthase-1 (mPGES-1), mPGES-2, cytosolic PGES (cPGES), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were selected using Beacon Designer software (Bio-Rad) and synthesized by IDT (Coralville, IA). They are listed as follows: name, forward primer, reverse primer (amplicon size), GenBank accession number: COX-2, 5-aagcgaggacctgggttcac-3, 5-acacctctccaccaatgacctg-3 (142 bp), BC052900; mPGES-1, 5-gatgaggctgcggaagaagg-3, 5-gcgaaggcgtgggttcag-3 (271 bp), NM_022415; mPGES-2, 5-aagccaggacggaggagatg-3, 5-atcactcgcagcacaccatac-3 (355 bp), NM_133783; cPGES, 5-cagttgtcttggaggaagcg-3, 5-ttattgaagtccacactgagcc-3 (201 bp), AY281130; and GAPDH, 5-accacagtccatgccatcac-3, 5-accttgcccacagccttg-3 (134 bp), M32599. The PCR amplification of each product was further assessed using 10-fold dilutions of mouse brain cDNA library as a template and found to be linear over five orders of magnitude and at >95% efficiency. All the PCR products were verified by sequencing. The reactions were set up in duplicate in total volumes of 25 μl containing 12.5 μl of 2× iQSYBR green Supermix (Bio-Rad) and 5 μl of template (1:10 dilution from RT product) with a final concentration of 400 nm of the primer. The PCR cycle was as follows: 95°C for 3 min, 45 cycles of 95°C for 30 s, 58°C for 45 s, and 95°C for 1 min, and the melt-curve analysis was performed at the end of each experiment to verify that a single product per primer pair was amplified. Furthermore, the sizes of the amplified DNA fragments were verified by gel electrophoresis on a 3% agarose gel. The amplification and analysis were performed using an iCycler iQ Multicolor Real-Time PCR Detection System (Bio-Rad). Samples were compared using the relative cycle threshold (CT) method. The fold increase or decrease was determined relative to a vehicle-treated control after normalizing to a housekeeping gene using 2^-ΔΔCT, where ΔCT is (gene of interest CT) - (GAPDH CT), and ΔΔCT is (ΔCT treated) - (ΔCT control).

Immunocytochemistry. Immunostaining was performed in primary hippocampal neurons in culture. The neurons were rinsed with PBS after removal of the culture medium and fixed with prewarmed (37°C) 4% paraformaldehyde, 4% sucrose in 0.1 m phosphate buffer, pH 7.2, and incubated for 20 min at room temperature. Then, the cells were washed with ice-cold PBS four times and incubated with blocking buffer (1% BSA and 10% normal goat serum) in PBS at room temperature for 1 h. Primary antibodies at different dilutions (in PBS containing 1% BSA) were applied for 1 h at 37°C. These were rabbit anti COX-2, mPGES-1 and -2, cPGES (1:200; Cayman Chemical, Ann Arbor, MI), mouse anti-PSD-95 (postsynaptic density-95) (12.5 μg/ml), and synaptophysin (1: 1000; Chemicon, Temecula, CA). After four 10 min washes with PBS containing 1% BSA, species-specific and highly cross-adsorbed secondary antibodies coupled to Alexa 488 and cyanine 3 (Invitrogen), diluted 1:1000 in PBS containing 1% BSA, were applied for 1 h at 37°C. Cells were washed four times, 10 min each, with PBS containing 1% BSA and rinsed with PBS. Negative controls without primary antibody or secondary antibody for second staining in double staining were checked. No apparent cross-reactions from the second staining were noticed. The specificity of antibodies was checked by adsorption with corresponding blocking peptides (ratio of primary antibody/blocking peptide, 1:1) from the same company. The images were taken by a Zeiss 510 Meta laser confocal microscope with LSM 510 Meta software.

Plasmid construction. The PGSU6-GFP plasmid (GeneSilencer shRNA Vectors kit; Gene Therapy Systems, San Diego, CA), which contains the human U6 RNA pol III promoter for the expression of small hairpin RNA (shRNA) and the cytomegalovirus (CMV) promoter carried with the green fluorescent protein (GFP) reporter gene were used for construction of the shRNA-encoding plasmids. These were generated by inserting annealed oligonucleotides (EP2 sense plus EP2 antisense) (supplemental Table 1, available at www.jneurosci.org as supplemental material) into GeneSilencer shRNA vectors between the BamHI and NotI restriction sites. The shRNA inserts were designed based on the following criteria: (1) the 19 nt target sites were immediately downstream of an AA dinucleotide; (2) the G/C content of target sites ranged between 49 and 60%; (3) the sequence did not contain a repeat of more than three Gs or Cs; and (4) target regions were at least 70-100 nt away from the translational initiation site of the transcript (Caplen and Mousses, 2003). Selected sequences were submitted to a basic local alignment search tool search against the human genome sequence to ensure that the human genome was not targeted. Two pairs of candidate target sequences satisfied the criteria. We designed three different oligonucleotides (number 1, nucleotides 495-513; number 2, nucleotides 593-611; and number 3, nucleotides 1438-1456; GenBank accession number NM_008964). It appeared that the resultant plasmids that expressed three different shRNA targeting the EP2 worked, as evident from the decreases in mRNA and protein in RAW264.7 macrophages transfected with the plasmids. Annealed oligonucleotides (luciferase sense and antisense provided with the GeneSilencer shRNA Vectors kit; Gene Therapy Systems) were cloned into the vectors between the BamHI and NotI restriction sites as negative control plasmid. Similar procedures were used for designing EP4 oligonucleotides (number 1, nucleotides 1118-1136; number 2, nucleotides 1146-1164; and number 3, nucleotides 1506-1524; GenBank accession number NM_008965) and constructing EP4-shRNA plasmid.

Transfection of the shRNA expression plasmid. Transfection with the shRNA-encoding plasmid into primary hippocampal neurons in culture was conducted using a calcium phosphate precipitate protocol (Xia et al., 1996). Hippocampal neurons (8 × 10⁵) from P0 mice were plated onto 35 mm dishes for electrophysiological recordings, and 1 × 10⁵ cells were plated onto four-well chamber slides for immunostaining and transfected on 9 DIV. Neuronal culture medium (2 ml) was added into each dish for making conditioned medium on 8 DIV, and one-half of the conditioned medium in each dish was removed and saved before addition of the transfection precipitate. The DNA/calcium phosphate precipitate was prepared by mixing 1 vol of EP2- or EP4-shRNA plasmid DNA in 125 mm CaCl₂ solution with an equal volume of 2× HBS [composed of the following (in mm): 42 HEPES, 274 NaCl, 1.4 Na₂HPO₄·7H₂O, 10 KCl, and 15 glucose, pH 7.05]. The precipitate was incubated at room temperature for 30 min and shielded from light before addition of the transfection mixture onto the neurons. EP2-shRNA or EP4-shRNA plasmid DNA (10 μg) or luciferase-shRNA plasmid was used for 35 mm dishes and 2 μg of DNA for four-well chamber slides. After addition of the transfection precipitate mixture onto cells, the dishes were returned to the incubator for 90 min. Neurons were then washed twice with neuronal culture medium and returned to the incubator after addition of the saved conditioned medium back to each plate.

Hippocampal PGE2 assay. Quantitative analysis of PGE2 by liquid chromatography-mass spectrometry-mass spectrometry (LC-MS-MS) was performed in mouse hippocampal slices (Marcheselli et al., 2003; Chen and Bazan, 2005). Hippocampal slices were cut the same as for electrophysiological recordings and pretreated with or without N-[2-cyclohexyloxy-4-nitrophenyl]methanesulfonamide (NS398) (20 μm). Hippocampal slices were homogenized immediately in 1 ml of cold chloroform:methanol (1:1, v/v) and kept under nitrogen at -80°C until purification. To increase expression of COX-2, mice were injected with lipopolysaccharide (LPS; 3 mg/kg, i.p.). Animals were killed 4 and 12 h after LPS injection by head-focused microwave radiation that delivers 10 kW, 80% direct power, for 0.4 s duration to denature tissue proteins, and hippocampi were rapidly dissected out and homogenized in 1 ml of chloroform:methanol (1:1, v/v). Purification was performed by a solid-phase extraction technique (Marcheselli et al., 2003). In short, samples pre-equilibrated at pH 3.0 were loaded onto C18 columns (Varian, Palo Alto, CA) and eluted with 10 ml of 1% methanol in ethyl acetate (EM Science, Gibbstown, NJ). Samples were concentrated by nitrogen-stream evaporator before LC-MS analysis. Samples were loaded into a Surveyor MS pump (Thermo-Finnegan, Altrincham, UK) equipped with a C18 discovery column (inner diameter, 10 cm × 2.1 mm; internal phase, 5 μm; Supelco, Bellefonte, PA). Samples were eluted in a linear gradient [100% solution A (60:40:0.01 methanol/water/acetic acid) to 100% solution B (99.99:0.01 methanol/acetic acid)] and run at a flow rate of 300 μl/min for 45 min. LC effluents were diverted to an electrospray ionization probe on a TSQ Quantum (Thermo-Finnegan) triple quadrupole mass spectrometer running on negative ion-detection mode. PGE2 and PGD2 standards were used for calibration and optimization. The instrument was run on full-scan mode to detect parent ions and selected-reaction mode for quantitative analysis to detect daughter ions simultaneously.

PGE2, 17-phenyl trinor prostaglandin E2 (17-P-PGE2), butaprost, and sulprostone were purchased from Cayman Chemical, and N-[2-(P-bromocinnamylamino)ethyl]-5-isoquino-linesulfonamide dihydrochloride (H-89) and KT5720 ((9R,10S,12S)-2,3,9,10,11,12-hexahydro-10-hydroxy-9-methyl-1-oxo-9,12-epoxy-1H-diindolo[1,2,3-fg-3′,2′,1′-K1]pyrrolo[3,4-1][1,6]benzodiazocine-10-carboxylic acid) (KT) were purchased from Tocris (Ellisville, MO). All other drugs and chemicals were obtained from Sigma (St. Louis, MO), unless stated otherwise.

Data were presented as mean ± SEM. Unless stated otherwise, Student's t test and ANOVA with Student-Newman-Keuls test were used for statistical comparison when appropriate. Differences were considered significant when p < 0.05. The care and use of the animals reported in this study were approved by the Institutional Animal Care and Use Committee of Louisiana State University Health Sciences Center.

Results

Discussion

Our results reveal that postsynaptically synthesized PGE2 may serve as a retrograde messenger in synaptic transmission via a presynaptic EP2 receptor. This is supported by the following evidence: (1) the EP2 receptor is expressed in presynaptic terminals in hippocampal neurons (Zhu et al. 2005). (2) PGE2 increases the amplitude of EPSPs and reduces the PPR both at perforant path and Schaffer collateral synapses in hippocampal slices and enhances the frequency, but not the amplitude, of mEPSCs in hippocampal neurons in culture. These observations provide evidence that PGE2 increases the presynaptic probability of the release of the neurotransmitter glutamate. (3) An EP2 agonist mimics PGE2 actions, whereas EP1 or EP3 agonist does not elicit effects either on synaptic stimulus-evoked EPSPs or mEPSCs, and the PGE2-induced enhancement of the frequency of mEPSCs is attenuated by PKA inhibitors. (4) COX-2 and mPGES-1 and -2 are present in postsynaptic dendritic spines, suggesting that postsynaptic dendritic spines are the site of PGE2 synthesis. (5) Silencing EP2 gene expression, but not EP4, eliminates the PGE2-induced increase in the frequency of mEPSCs, confirming that EP2 mediates PGE2 signaling in synaptic transmission. (6) IL-1β and LPS enhance expression of COX-2 and mPGES-1, which produce PGE2, and augment the frequency of mEPSCs, and this enhancement is inhibited by a COX-2 inhibitor. These data demonstrate that the increased COX-2 expression causes enhanced production of PGE2, which in turn augments synaptic transmission via a presynaptic EP2-PKA pathway.

Three COX isozymes have been identified in brain (Vane, 1971; Kujubu et al., 1991; Hla and Neilson, 1992; O'Banion et al., 1992; Chandrasekharan et al., 2002) (for review, see Vane et al., 1998; Smith et al., 2000; Simmons et al., 2004). COX-1 is constitutively expressed in most tissues and is thought to mediate housekeeping functions. COX-3 is made from the COX-1 gene but retains intron 1. Thus, it is a COX-1 variant and the possible target of analgesic/antipyretic drugs such as acetaminophen. On the other hand, COX-2 expression is constitutive as well as inducible (Smith et al., 1996; Vane et al., 1998). The hippocampus expresses both forms of COX-2 (Vane et al., 1998). PGE2 is derived mainly through the initial conversion of AA by COX-2 (Yamagata et al., 1993; Brock et al., 1999; Chen et al., 2002; Vidensky et al., 2003; Chen and Bazan, 2005a). In the present study, we found that COX-2 and microsomal PGE synthases-1 and -2 are colocalized with PSD-95, indicating that postsynaptic dendritic spines are a site of PGE2 synthesis that in turn may be released to the synaptic cleft. In addition, IL-1β or LPS not only increases the expression of COX-2 but also significantly potentiates microsomal PGES-1. Colocalization of COX-2 and mPGES-1, the sequential biosynthetic enzymes for PGE2, in the same subcellular compartment (postsynaptic dendritic spines), allows efficient conversion of the short-lived, unstable substrate prostaglandin H2 (PGH2) to PGE2, indicating that PGE2 availability is tightly regulated by COX-2. Therefore, blockade of the microsomal PGES-1 may also reduce PGE2 production, particularly in conditions in which COX-2 expression and PGE2 are elevated (Murakami and Kudo, 2004).

Expression of G-protein-coupled EP1-4 receptors in hippocampus and their physiologic significance are not well understood (Zhu et al., 2005). All four EP1-4 receptors are heterogeneously expressed in the hippocampus. EP2 and EP4 are located at presynaptic terminals, in addition to their presence in paranuclear sites (Zhu et al., 2005). The present study shows that a presynaptic EP2 receptor may mediate PGE2 signaling at hippocampal synapses, whereas EP1 and EP3 may not be involved in the PGE2 modulation of basal synaptic transmission. However, the level of EP3 mRNA is significantly reduced in neurons treated with TTX or DNQX plus (+)-5-methyl-10,11-dihydro-5H-dibenzo

[a,d] cyclo-hepten-5,10-imine maleate (Zhu et al., 2005). EP3 is also the most abundant of the EPs in the hippocampus. Therefore, the role of EP3 in activity-dependent synaptic signaling cannot be ruled out. It is still not clear what functional role EP4 plays in synaptic transmission, because it is expressed at presynaptic terminals. Although we did not examine the effect of EP4 activation on synaptic transmission, EP4 expression level is relatively low in the hippocampus, and the decrease of EP4 expression through transient silencing of the EP4 gene did not significantly affect the PGE2-induced actions, suggesting that EP4 does not significantly contribute, if at all, to the PGE2 signaling in basal synaptic transmission. However, we cannot exclude the role of EP4 in mediating PGE2 signaling in activity-dependent events, such as occur in inflammation and synaptic plasticity. For instance, application of IL-1β or LPS increases EP2 and EP4 expression in brain (Zhang and Rivest, 1999). EP4 expression is enhanced in hippocampal neurons in culture exposed to IL-1β, high K+ (90 mm), or 4-AP and in hippocampal slices receiving HFS associated with LTP induction (Zhu et al., 2005).

Interestingly, we found that PGE2 significantly increases evoked EPSPs and the frequency of mEPSCs, whereas it slightly reduces the GABA receptor-mediated mIPSCs (including frequency and amplitude) in hippocampal neurons (data not shown). In rat spinal dorsal horn neurons, however, PGE2 decreases amplitudes of synaptic stimulus-evoked glycine receptor-mediated IPSCs and mIPSCs, whereas GABA-, AMPA-, and NMDA-receptor-mediated responses are unaffected (Ahmadi et al., 2002). The effect appears to be mediated via a postsynaptic EP2-like receptor. In addition, application of PGE2 produces a direct membrane depolarization through a postsynaptic EP2-like receptor in spinal dorsal horn neurons, whereas excitatory synaptic transmission remains unchanged (Baba et al., 2001). On the other hand, PGE2-produced inhibition of GABAergic synaptic transmission is attributed to presynaptic EP3-like receptors in rat supraoptic neurons (Ibrahim et al., 1999). These studies, together with our present observations, suggest a diversity of EP receptor distributions and functional roles of PGE2 in synaptic transmission.

RNA interference has become a powerful and widely used tool for the analysis of gene function in mammalian cells (Sharp, 2001; Meister and Tuschi, 2004; Mello and Conte, 2004). In the present study, we adopted the GeneSilencer shRNA vectors containing the U6 RNA polymerase III promoter and the CMV promoter carried with the GFP reporter gene, allowing optimal expression in a wide variety of cell types, including neurons, and easy determination of vector transfection efficiency and localization. Our results indicate that the decrease in EP2 expression through transfection with EP2-shRNA plasmid, silencing the EP2 gene, significantly reduces the level of mRNA and protein in RAW264.7macrophage cells (supplemental Fig. 1, available at www.jneurosci.org as supplemental material), and functionally eliminates PGE2-induced increase of the probability of release of the neurotransmitter. The recorded neurons mostly did not display green fluorescence, indicating that EP2 in these neurons remains intact. However, these neurons are synapsed with a few neurons transfected with the EP2-shRNA plasmid (showing the green fluorescence). We made similar recordings in luciferase-shRNA transfected cultures. PGE2 does not elicit significant changes either in the frequency or amplitude of mEPSCs in the EP2-shRNA-transfected cultures, whereas it does increase the frequency of mEPSCs in the luciferase-shRNA- and EP4-shRNA-transfected cultures, similar to that observed in normal cultures. This information provides evidence that the EP2-induced increase in synaptic activity is mediated via a presynaptic EP2 receptor.

Increasing evidence suggests that PGE2 and its receptors have multiple functions in the nervous system, including their roles in fever, pain, inflammation, sleep, regulation of membrane excitability, sexual behavior, and synaptic transmission, integration, and plasticity (Breyer et al., 2001; Chen et al., 2002; Amateau and McCarthy, 2004; Hayaishi and Huang, 2004; Murakami and Kudo, 2004; Simmons et al., 2004; Chen and Bazan, 2005a). A striking role of COX-2-derived PGE2 is in neuronal apoptosis and degenerative processes (Miettinen et al., 1997; Dubois et al., 1998; Nakayama et al., 1998; Ho et al., 1999; Hewett et al., 2000; Iadecola et al., 2001; McCullough et al., 2004). COX-2 expression or activity is enhanced in inflammatory and ischemic brain and in neurologic disorders such as epilepsy and Alzheimer's disease. Our previous studies demonstrate that PGE2, but not PGD2 or PGF2α, participates in hippocampal synaptic transmission and plasticity (Chen et al., 2002). This means that COX-2-derived PGE2 not only plays an important role as a messenger in synaptic physiology but also is involved in seizure and neurodegenerative diseases as a result of excessive activation of COX-2 and elevated PGE2. The results obtained from the present study indicate that PGE2 stimulates the release of the neurotransmitter glutamate. This may be one of the central mechanisms responsible for the neurodegeneration caused by enhanced COX-2 expression. Because COX-2 and mPGES are expressed in postsynaptic dendritic spines, the AA can be quickly converted to PGH2 and PGE2 (Bazan, 2003; Sun et al., 2004). In particular, both constitutive and inducible forms of COX-2 are present in dendritic spines. PGE2 derived from the constitutive COX-2 may dynamically regulate basal synaptic transmission, whereas PGE2 derived from the inducible COX-2 may contribute to activity-dependent synaptic plasticity, and, if COX-2 is overexpressed, to neurotoxicity. In addition, glial and endothelial cells are important cellular sources of PGE2 that contributes to neurodegenerative processes during inflammation. Therefore, our findings are relevant to understanding the roles of COX-2 and its derived PGE2 in synaptic physiology and raise an opportunity to explore mechanisms underlying neurologic disorders such as epilepsy and Alzheimer's disease.