Cytosolic Phospholipase A2 alpha/Arachidonic Acid Signaling Mediates Depolarization-Induced Suppression of Excitation in the Cerebellum

Background Depolarization-induced suppression of excitation (DSE) at parallel fiber-Purkinje cell synapse is an endocannabinoid-mediated short-term retrograde plasticity. Intracellular Ca2+ elevation is critical for the endocannabinoid production and DSE. Nevertheless, how elevated Ca2+ leads to DSE is unclear. Methodology/Principal Findings We utilized cytosolic phospholipase A2 alpha (cPLA2α) knock-out mice and whole-cell patch clamp in cerebellar slices to observed the action of cPLA2α/arachidonic acid signaling on DSE at parallel fiber-Purkinje cell synapse. Our data showed that DSE was significantly inhibited in cPLA2α knock-out mice, which was rescued by arachidonic acid. The degradation enzyme of 2-arachidonoylglycerol (2-AG), monoacylglycerol lipase (MAGL), blocked DSE, while another catabolism enzyme for N-arachidonoylethanolamine (AEA), fatty acid amide hydrolase (FAAH), did not affect DSE. These results suggested that 2-AG is responsible for DSE in Purkinje cells. Co-application of paxilline reversed the blockade of DSE by internal K+, indicating that large conductance Ca2+-activated potassium channel (BK) is sufficient to inhibit cPLA2α/arachidonic acid-mediated DSE. In addition, we showed that the release of 2-AG was independent of soluble NSF attachment protein receptor (SNARE), protein kinase C and protein kinase A. Conclusions/Significance Our data first showed that cPLA2α/arachidonic acid/2-AG signaling pathway mediates DSE at parallel fiber-Purkinje cell synapse.


Introduction
Depolarization-induced suppression of excitation (DSE) was first reported at excitatory synapse in cerebellar Purkinje cells [1]. While DSE is a short-term retrograde plasticity associated with a change in paired-pulse ratio [1][2][3], it is initiated by the postsynaptic depolarization that activates local dendritic Ca 2+ spikes [1,4]. Both blocking dendritic Ca 2+ spikes by hyperpolarization and intracellular injection of 1,2-bis(o-aminophenoxy)ethane-N,N,N9,N9-tetraacetic acid (BAPTA) prevent DSE [1,4], indicating that the Ca 2+ elevation is critical for the DSE induction. DSE provides a means for altering the strength and properties of presynaptic inputs for tens of seconds during high postsynaptic activity [1]. It is postulated that DSE provides a neuroprotective effect because it reduces the glutamatergic transmission when Purkinje cells are subject to strong excitatory inputs in pathophysiological conditions [1][2][3].
It is known that DSE is mediated by a retrograde signaling that involves the production of postsynaptic endocannabinoid and the activation of presynaptic cannabinoid receptor 1 (CB1R) [2,3]. The synthesis and release of endocannabinoids are Ca 2+ -dependent [5,6]. Nevertheless, how Ca 2+ elevation leads to the production of endocannabinoid is unclear thus far. It is shown that a prolonged elevation of synaptic Ca 2+ activates Gq-coupled metabotropic receptors [7,8] and phospholipase-C (PLC) [6]. However, this may not be the case for DSE induction in Purkinje cells, because DSE is independent of PLC [9] and metabotropic glutamate receptor (mGluR) [1]. Thus, another PLC-independent pathway might contribute to the production of endocannabinoid and DSE induction in Purkinje cells.
The phospholipase A 2 (PLA 2 ) enzymes catalyze ester hydrolysis of fatty acids [10]. Of the PLA 2 s, cytosolic phospholipase A 2 alpha (cPLA 2 a) has a unique set of biochemical properties. It translocates to cellular membranes in response to micromolar intracellular Ca 2+ and produces arachidonic acid [11]. Arachidonic acid can be metabolized by a number of enzymes to create the eicosanoids [10,12] that play important roles in regulating cellular homeostasis, neurotoxicity and inflammation [11][12][13]. Since the brief depolarization during DSE exerts a rapid elevation of intracellular Ca 2+ with a peak level of 10-15 mM [8,14], we hypothesized that the elevated Ca 2+ triggers the activation of cPLA 2 a and causes the synthesis and release of endocannabinoid. Here, we examined the function of cPLA 2 a/arachidonic acid pathway in DSE at parallel fiber-Purkinje cell synapses derived from wild-type (WT) and cPLA 2 a knock-out (KO) mice. We also explored other unsolved mechanisms of DSE in Purkinje cells using various pharmacological treatments. In summary, our data showed that the cPLA 2 a/ arachidonic acid pathway is required for DSE induction.

Results
DSE is inhibited in cPLA 2 a KO mice DSE at the parallel fiber-Purkinje cell synapse was studied in sagittal cerebellar slices. Parallel fiber excitatory postsynaptic currents (EPSCs) were evoked with an extracellular electrode placed in the molecular layer. DSE was induced according to the previous work [1]. In brief, Purkinje cells were stimulated by a step voltage from 270 mV to 0 mV (50 ms) after 3 consecutive control EPSCs with an interval of 20 s were obtained in voltage-clamp mode ( Figure 1A). A test stimulus was set to 5 s after the depolarization to acquire the test EPSC. In WT mice, the amplitudes of test EPSCs were greatly reduced and the EPSC inhibition usually recovered within 90 s ( Figure 1B). We could reliably repeat this cycle every 90 s ( Figure 1B), similar to previous work [1]. To confirm the validity of our DSE experiment, we also perfused AM251, a CB1R antagonist, in cerebellar slices. We found that the inhibition of test EPSC was completely blocked by AM251 ( Figure S1), demonstrating that our DSE experiment was successful [1,4,15]. Interestingly, we found that DSE was significantly reduced in KO mice ( Figure 1C). Pretreatment of WT cells with the cPLA 2 a inhibitor arachidonyl trifluoromethyl ketone (AACOCF3; 10 mM) also resulted in a significant reduction of DSE ( Figure 1D). Finally, AACOCF39s effect was examined in KO cells. Similar to WT cells, we failed to observe DSE ( Figure 1E). The statistics are shown in Figure 1F, indicating that cPLA 2 a is required for DSE.
Arachidonic acid rescues DSE in cPLA 2 a KO mice cPLA 2 a selectively liberates arachidonic acid [11]. We next examined the effects of external application of arachidonic acid in parallel fiber EPSC and DSE. Parallel fiber-Purkinje cell synaptic EPSCs were evoked and recorded every 20 s. Bath application of arachidonic acid (10 mM) decreased EPSCs in WT Purkinje neurons (73.763.0% of baseline at t = 29 min; n = 25; Figure 2A). Similarly, 10 mM arachidonic acid induced a slow decrease of EPSCs that reached 74.562.5% of baseline at t = 29 min in KO cells (n = 17; Figure 2A). Representative recordings are shown in Figure 2B. These data implied that arachidonic acid, as the downstream factor of cPLA 2 a, is able to suppress parallel fiber EPSC. We next asked whether exogenous arachidonic acid is able to rescue the defect of DSE in KO Purkinje cells. In this experiment, DSE was evoked every 2 min using the protocol in Figure 1A. The ratio of test/control response was gradually reduced in the bath application of arachidonic acid (33.067.9% at t = 34 min; n = 19; Figure 2C), indicating that arachidonic acid is able to rescue DSE at parallel fiber-Purkinje cell synapse. Collectively, these experiments demonstrated that cPLA 2 a/ arachidonic acid signaling pathway is essential to the induction of DSE in Purkinje cells.

2-AG is the downstream factor of arachidonic acid in DSE
N-arachidonoylethanolamine (AEA) and 2-arachidonoylglycerol (2-AG) are two endogenous ligands that activate CB1Rs [16]. The biosynthetic processes and catabolic inactivation of AEA and 2-AG decide the onset and duration of CB1R-mediated signaling [6,16]. In central nervous system (CNS), AEA is selectively hydrolyzed by fatty acid amide hydrolase (FAAH) [17,18] and 2-AG is selectively hydrolyzed by monoacylglycerol lipase (MAGL) [19,20]. We added 0.7 mg/ml MAGL or 1 mg/ml FAAH in the pipette solution during the DSE experiments in WT cells, according to previous work [21]. These operations broke down 2-AG and AEA respectively in Purkinje cells [21]. As shown in Figure 3A and B, DSE was blocked in the presence of MAGL, but kept unchanged in the presence of FAAH, indicating that 2-AG but not AEA is required for DSE.
We showed that arachidonic acid is able to decrease parallel fiber EPSC and rescue DSE ( Figure 2). Next, we asked whether arachidonic acid's roles are mediated by 2-AG. A baseline of parallel fiber EPSCs was recorded in KO cells that were filled with 0.7 mg/ml MAGL and 1 mg/ml FAAH ( Figure 3C). We then applied exogenous arachidonic acid in order to examine whether it affects EPSC when 2-AG or AEA is broken down. EPSCs obtained from FAAH-filled KO cells decreased after 10 mM arachidonic acid was applied (73.963.7% of baseline at t = 29 min; n = 17; Figure 3C). On the contrary, arachidonic acid did not change parallel fiber EPSCs in MAGL-filled KO cells (98.763.6% of baseline at t = 29 min; n = 19; Figure 3C). These results suggested that the arachidonic acid's suppression of EPSC depends on the production of 2-AG. We continued to study whether MAGL or FAAH affects the rescue of DSE by arachidonic acid in KO cells. In this experiment, DSE was assessed every 2 min from either 0.7 mg/ml MAGL-filled or 1 mg/ ml FAAH-filled KO cells. After a baseline recording of DSE, 10 mM arachidonic acid was perfused in slices. We found that exogenous arachidonic acid failed to produce DSE in MAGLfilled KO cells (ratio of test/control response: 81.069.9% at t = 34 min; n = 20; Figure 3D). However, arachidonic acid effectively rescued DSE in FAAH-filled KO cells (ratio of test/ control response: 33.369.8% at t = 34 min; n = 18; Figure 3D). These results indicated that roles of cPLA 2 a/arachidonic acid in DSE are via 2-AG but not AEA. Interestingly, we found that DSE was gradually increased with the application of arachidonic acid in KO cells ( Figure 2C and Figure 3D). This implied that the depolarization not only activates cPLA 2 a, but facilitates the release of 2-AG. Indeed, the suppression ratios of test EPSC vs. control EPSC in DSE ( Figure 2C and Figure 3D) were much more prominent than the decreases of EPSC caused by arachidonic acid application (Figure 2A and 3C).
Ca 2+ -activated K + -channel inhibitor paxilline reverses the blockade of DSE by internal K + Our findings suggested that the depolarization-induced calcium increase is sufficient to activate the cPLA 2 a/arachidonic acid pathway, consistent with the characteristic low-threshold activation of cPLA 2 a [10,11]. Previous work showed that voltage-gated calcium channels (VGCCs) attribute the most of calcium influx [4], but how Ca 2+ influx is regulated in DSE is unclear. Although hyperpolarization in Purkinje cells prevents the induction of DSE cPLA2a/Arachidonic Acid Mediates Cerebellar DSE [4,14], the function of K + channels in DSE is unknown because Cs + -based internal saline was used in previous studies [1,4,14,15]. To examine the role of K + channels in cPLA 2 a activation and DSE, we switched the internal saline from Cs + -based to K + -based and examined DSE in WT cells. We found that DSE was completely blocked ( Figure 4A), suggesting that the activation of K + channels are sufficient to inhibit Ca 2+ influx and DSE.
We next studied which K + channel was involved in the inhibition of Ca 2+ influx. A number of conductances contribute to spike repolarization in Purkinje neurons, including large-conductance calcium-activated potassium channel (BK) and smallconductance calcium-activated potassium channel (SK). If the opening of BK or SK weakens Ca 2+ influx and inhibits DSE, then a pharmacologic disruption of BK and SK should reverse the blockade of DSE. Accordingly, we used Cs + -based internal saline in pipettes and perfused WT cells with either BK blocker, paxilline (1 mM), or SK blocker, apamin (100 nM). As shown in Figure 4B, co-application of apamin did not induce DSE. On the contrary, co-application of paxilline relieved the blockade of DSE by internal K + ( Figure 4C), indicating that BK opening is sufficient to block the internal Ca 2+ -induced cPLA 2 a inactivation and DSE induction.
Beyond cPLA 2 a/arachidonic acid: DSE is independent of SNARE, P2X7 receptor (P2X7R), protein kinase C (PKC) and protein kinase A (PKA) Although we showed that cPLA 2 a/arachidonic acid signaling was essential for DSE induction, several important questions in Purkinje cell DSE remain to be elucidated. Previous work showed that the production site of endocannabinoid and presynaptic CB1 receptors are in close proximity of postsynaptic neurons [22]. Thus, the first question is how endocannabinoid (2-AG) is released. Most retrograde messengers are stored in vesicles and released through exocytosis. Is 2-AG released via secretory vesicles at parallel fiber-Purkinje cell synapse? Botulinumtoxin (BoTx) destroys the stability of SNARE complex and prevents the release of secretory vesicles from synapstic plasma membrane [23]. Meanwhile, it does not affect the depolarization-evoked dendritic Ca 2+ transient [24]. Hence, 100 nM BoTx was added in Cs +based internal saline. We found that BoTx did not affect the induction of DSE ( Figure 5A), suggesting that the release of 2-AG is independent of SNARE.
Second question is whether astrocytes are involved in DSE. P2X7Rs are widely expressed in the cerebellum, including Purkinje cells and Bergmann glia cells [25]. Activation of P2X7Rs evokes a rapid and pronounced increase of endocannabinoid production in astrocytes [26]. Is it possible that depolarization-induced ATP release from Purkinje cells triggers the activation of astrocytic P2X7Rs, releases 2-AG and evokes DSE?    Figure 1A with Dt 5 s. Each data point represents the average percentage inhibition of test EPSC every 2 min. Arachidonic acid was applied in the bath as indicated by the bar. *, P,0.05. doi:10.1371/journal.pone.0041499.g003  To address this question, a broad-spectrum antagonist of P2X receptors, PPADS (pyridoxal-phosphate-6-azophenyl-29-49-disulfonic acid) (10 mM) was bath-perfused during recordings in WT cells. Our results showed that PPADS did not block DSE ( Figure 5B), indicating that P2X7R and astrocytes may be not involved in DSE.

(C) Time courses of percentage changes of parallel fiber EPSC amplitudes derived from KO cells filled with either MAGL (filled circles) or FAAH (open circles). Arachidonic acid (AA) was applied in the bath as indicated by the bar. Arachidonic acid depressed EPSCs in FAAH-filled cells but not MAGL-filled cells. (D) MAGL blocked the rescue of DSE by arachidonic acid in KO cells. KO cells filled with either MAGL (filled circles) or FAAH (open circles) Arachidonic acid restored DSE in KO cells. DSE was induced by the protocol indicated in
Some evidence suggests that DSE is mediated by a reduction in the presynaptic Ca 2+ [1,27]. However, another finding argues that DSE is unrelated to Ca 2+ entry [28]. Therefore, the CB1Rinduced presynaptic signaling underlying DSE is not clear. Presynaptic PKC and PKA are reported to regulate synaptic release and mEPSC frequency [29,30] and trigger presynaptic long-term potentiation at parallel fiber synapses [31]. We then examined the function of PKC and PKA in parallel fiber DSE. WT cells were continuously treated with PKC-selective inhibitor chelerythrine (10 mM) or PKA-selective inhibitor KT5720 (1 mM) before and during experiments. We found that DSE was successfully induced in the application of both chelerythrine ( Figure 5C) and KT5720 ( Figure 5D), indicating that presynaptic PKC and PKA are not involved in DSE.

Discussion
The main finding of the present study is that DSE at parallel fiber-Purkinje cell synapse was mediated by the cPLA 2 a/ arachidonic acid pathway. DSE was significantly inhibited in cPLA 2 a KO mice and rescued by the application of arachidonic acid in the bath. The action of arachidonic acid in DSE was prevented by MAGL, the degradation enzyme of 2-AG [17,18], but not FAAH that hydrolyzes AEA [19,20]. These data first demonstrated that cPLA 2 a/arachidonic acid/2-AG signaling induces DSE at parallel fiber-Purkinje cell synapse, as summarized by a model in Figure 6. As the explanation for this model, postsynaptic depolarization in Purkinje cell triggers Ca 2+ influx by activating voltage-gated Ca 2+ channels and causes a transient elevation of [Ca 2+ ] i . This internal Ca 2+ elevation is hindered by the presence of intracellular K + and the opening of BK channels [32]. Micromolar levels of [Ca 2+ ] i activate cPLA 2 a to liberate arachidonic acid, which produces 2-AG. The latter is released from Purkinje cells into the extracellular space independent of SNARE, diffuses retrogradely and binds to CB1Rs at the parallel fiber terminal. Finally, CB1R triggers a PKA and PKCindependent mechanism to suppress presynaptic glutamate release (DSE).
AEA and 2-AG are derivates of arachidonic acid [33]. They are highly distributed in the hippocampus and cerebellum [34] and considered to be the main endocannabinoids mediating DSE in CNS [35][36][37]. It is generally accepted that the productions of AEA and 2-AG are Ca 2+ -dependent [38], but it is unclear how elevated Ca 2+ leads to enhanced endocannabinoid production during DSE. PLC/diacylglycerol lipase (DAGL) signaling meditates the formation of 2-AG in hippocampus [38], implying that PLC/DAGL might control the production of 2-AG in DSE. Against to this hypothesis, strong evidence shows that DSE is independent of mGluR1s, PLC and DAGL [36,39]. The short depolarization in DSE is not strong enough to simulate the PLC/DAGL-dependent 2-AG production [38]. Eicosanoid biosynthesis is highly interactive and often changes among cell signaling pathways on demand [40]. Except the PLC/DAGL signaling, other pathways have been shown to generate the endocannabinoid production [34,41]. Indeed, glucocorticoid or cyclooxygenase stimulation directly shifts arachidonic acid metabolism toward endocannabinoid synthesis [33,40]. Although it is impossible to directly detect the biosynthesis of 2-AG from arachidonic acid in DSE that happens within seconds, our MAGL/FAAH experiments clearly showed that 2-AG is the downstream factor of arachidonic acid. Therefore, we conclude that, at parallel fiber-Purkinje cell synapse, the brief depolarization causes the cPLA 2 a activation and shifts the arachidonic acid metabolism towards promoting 2-AG production.
Although DSE has been extensively studied in hippocampal and cerebellar neurons [1][2][3][4]15,37,[42][43][44], several important questions are unsolved. (1) Most retrograde messengers are stored in vesicles and released through exocytosis. How is endocannabinoid released? Does it require any special apparatus? (2) What is the presynaptic mechanism after CB1R is activated? (3) ATP is released from neurons in response to depolarization [45], which may activate astrocytic P2X7Rs and evoke the endocannabinoid production and release [26]. Does P2X7R participate in DSE? Although the present work was mainly focused on the function of cPLA 2 a/arachidonic acid/2-AG signaling in DSE, these questions are also tentatively investigated. Using a series of inhibitors, including BoTX, chelerythrine and KT5720, we showed that DSE is independent of SNARE, PKC and PKA. Although these results were negative, they provide some evidence for future experiments studying precise mechanisms of DSE.
P2X7R is expressed in Purkinje cells and glial cells [25]. Since ATP is released from neurons upon depolarization, we hypothesized that released ATP might activate P2X7Rs on Purkinje cells and glia, and subsequently evoke endocannabinoid production [26]. Unexpectedly, we did not observe inhibition of DSE when we applied PPADS. A previous study reported that ATP release from neurons is crucially dependent on the stimulus frequency [45]. This leads us to propose that the depolarization protocol used in the present work might not be strong enough to stimulate ectopic endocannabinoid release. Alternatively, strong depolarization in Purkinje cells may recruit more ectopic endocannabinoid release and cause more profound inhibition at parallel fiber-Purkinje cell synapse. A current viewpoint suggests that DSE plays a neuroprotective role by suppressing presynaptic glutamate release in response to excitotoxicity and neuronal death [1][2][3], which is strengthened by findings that CB1R KO mice are much more subject than control mice to neurotoxic events [46] and CB1Rs are tonically activated in MAGL knock-out mice [47]. However, our result that DSE usually recovers within 90 s suggests that the neuroprotective role of DSE in neurotoxicity may be overestimated. Future work should be conducted to assess the function of ectopic endocannabinoid release from glia in the process of neurotoxicity.
Purkinje cells were visualized under an upright microscope (BX51; Olympus Optical, Tokyo, Japan) with a 406 waterimmersion objective and equipped with infrared differential interference contrast enhancement. Whole-cell recordings were obtained with an Axopatch 700B amplifier (Molecular Devices, Foster City, CA). Currents were filtered at 1 kHz and digitized at 10 kHz. For parallel fiber stimulation, standard patch pipettes were filled with ACSF and placed in the middle third of the molecular layer. Synaptic responses were evoked every 20 s using 12-16 mA pulses (100 ms duration).
Drugs were purchased from Sigma (St. Louis, MO) and Tocris (Bristol, UK) unless stated otherwise. Data analysis was performed using Excel 2003 (Microsoft, Chicago, IL), Clampfit 10 (Molecular Devices) and Igor Pro 6.0 (Wavemetrics, Lake Oswego, OR). All group data are shown as mean 6 SEM. Student's t tests were used to determine P values. n represents numbers of cells used in each experiment derived from at least three animals. Cells were excluded from the study if series resistance or input resistance varied by more than 15% over the course of an experiment.