N-Palmitoylethanolamine Prevents the Run-down of Amplitudes in Cortical Spreading Depression Possibly Implicating Proinflammatory Cytokine Release

Cortical spreading depression (CSD), a wave of neuronal depolarization in the cerebral cortex following traumatic brain injury or cerebral ischemia, significantly aggravates brain damage. Here, we tested whether N-palmitoylethanolamine (PEA), a substance that effectively reduces lesion volumes and neurological deficits after ischemic stroke, influences CSD. CSD was elicited chemically in adult rats and occurrence, amplitude, duration and propagation velocity of CSD was determined prior to and for 6 hours after intraperitoneal injection of PEA. The chosen systemic administration of PEA stabilized the amplitude of CSD for at least four hours and prevented the run-down of amplitudes that is typically observed and was also seen in untreated controls. The propagation velocity of the CSD waves was unaltered indicating stable neuronal excitability. The stabilization of CSD amplitudes by PEA indicates that inhibition or prevention of CSD does not underlie PEA’s profound neuroprotective effect. Rather, PEA likely inhibits proinflammatory cytokine release thereby preventing the run-down of CSD amplitudes. This contribution of PEA to the maintenance of neuronal excitability in healthy tissue during CSD potentially adds to neuroprotection outside a damaged area, while other mechanisms control PEA-mediated neuroprotection in damaged tissue resulting from traumatic brain injury or cerebral ischemia.

velocities 7,8 . Despite this preclinical evidence the pathologic role and mechanistic involvement of CSD in neurological disorders and traumatic brain injury has been ignored in clinical practice until recently, when events resembling both PID and CSD were identified in stroke patients by the Co-Operative Studies on Brain Injury Depolarizations consortium 10 .
We have previously described the neuroprotective effects of N-acylethanolamines in a rat model of ischemic stroke using temporary middle cerebral artery occlusion (MCAO) to generate ischemia/reperfusion injury 11,12 . N-Palmitoylethanolamine (PEA) belongs to the group of N-acylethanolamines (NAEs), and is an endogenous signaling lipid involved in cellular signaling and neuroprotection [13][14][15][16] . While several NAEs, such as anandamide, are ligands for cannabinoid receptors, we and others have shown that the neuroprotective effects of PEA are mediated by an intracellular mechanism independent of cannabinoid and vanilloid receptor activation 11,12,14,[16][17][18] .
Here, we investigated whether the neuroprotective properties of PEA mediated by intracellular signaling pathways and resulting in reduced lesion size and improved neurological deficit score after experimental stroke, also contribute mechanistically to CSD/PID events and their deleterious impact on stroke.

Animals.
The in vivo experiments in the present study were approved by the Thuringian Government Surgical preparation of the rats. Adult male Wistar rats (n = 9, 350-450 grams) were deeply anesthetized with sodium thiopental (Trapanal ® , Inresa GmbH, Freiburg, Germany; initially 100 mg/kg i.p.). During dissection depth of anesthesia was checked by testing the corneal blink reflex and reflexes to noxious squeezing the tail tip. During the experiments, absence of the corneal blink reflex was maintained by supplemental doses of 20 mg/kg i.p. The trachea, the right femoral vein and artery were cannulated. The mean arterial blood pressure and the electrocardiogram were continuously monitored. Body temperature was kept at 37 °C by a feedback controlled heating system.
After stereotactic fixation of the head a trephination was made over the left hemisphere (spanning from the coronal suture over a length of 5-8 mm, 3-4 mm wide) using a mini-drill under cooling with artificial cerebrospinal fluid (ACSF, in mmol/L: NaCl 138.4, KCl 3.0, CaCl 2 1.3, MgCl 2 0.5, NaH 2 PO 4 0.5, urea 2.2, glucose 3.4, warmed to 37 °C and equilibrated with 5% CO 2 in O 2 ). The underlying dura and arachnoidea were removed; the exposed cortex was kept moist with ACSF. A wall was built with dental acrylic on the skull around the trephination thereby forming a trough in which the exposed cortex was superfused with ACSF.

Recording of intracortical DC potentials.
We used an Ag/AgCl reference electrode (containing 2 M KCl) on the nasal bone. The electrode arrays for elicitation of CSD and recordings of intracortical DC potentials at sites 1-4 are displayed in Fig. 1A. Electrodes for DC recordings had tip diameters of about 5 μm, resistance < 10 MΩ, and were filled with 150 mmol/L NaCl. The distance between the electrode tips was 400-500 μm. The electrode for CSD elicitation contained 1 mol/L KCl; SDs were elicited by injection of 0.5 μL KCl with a pressure of 100 kPa for 1 second using a microinjector (picoinjector PLI-100; Harvard Apparatus, Holliston, MA, USA). The signals were recorded using a four-channel high-impedance amplifier (Meyer, Munich, Germany) and stored on PC. CSDs were accepted if they had a steep onset, exceeded amplitudes ≥ 5 mV, and migrated over the whole recording area. At the end of an experiment the animals were euthanized by an overdose of the anesthetic (100 mg per rat, intravenously).
PEA was dissolved in ethyl alcohol and administered intraperitoneally at a dose of 20 mg/kg body weight. The recording protocol consisted of at least two CSD elicitations at intervals of 30 min prior to the injection of PEA forming a baseline. After injection of PEA, CSD were elicited at intervals of 30 min for two hours and then every hour. Four hours after the injection the rats received a second dose of PEA (20 mg/kg body weight), and CSD elicitations were continued 30 min, one hour and two hours after the second injection.
In a subset of three adult rats, two instead of one trephinations were made over the left hemisphere (1 frontal, spanning from 2 mm in front of bregma over a length of 5-6 mm, 3-4 mm wide, and 1 caudal, with a diameter of 3-4 mm in front of lambda), and a wall was built with dental acrylic on the skull around the frontal trephination thereby forming a trough with a capacity of 150 to 200 μL. The same sets of recording electrodes (DC 1 and DC 2 in the untreated area; DC 3 and DC 4 in the treated area) and the same stimulation device as in the experiments with PEA were used.
After testing for CSD under control conditions (both threphinations were superfused with warmed equilibrated ACSF), in the trough around the frontal trephination 1 μg of the IL-1ß receptor antagonist Anakinra (recombinant N2-L-methionyl-interleukin 1 receptor antagonist; Sobi, Stockholm, Sweden) was applied until the end of the experiment. CSD were elicited following the same schedule as in the experiments with PEA. CSD amplitudes in untreated and treated areas were evaluated and CSD propagation velocity was calculated from the time interval between microinjection of KCl and the peak of the CSD in the treated area.
Data are reported as means ± S.E.M. Statistical analysis was performed using the Wilcoxon matched pairs signed rank test. Bonferroni adjustment was performed as necessary. Significance was accepted at P < 0.05.

Results
In vivo CSD recordings. Microinjection of KCl elicited single propagating CSD with typical parameters (cf. Somjen) 19 in all rats at all recording electrodes. Typically observed delays between the DC potential peaks (as shown in Fig. 1B,C, traces DC2 and DC3/DC4), are due to the distance between the recording electrode arrays. In rats that were not dosed with PEA (vehicle control animals), we observed a continuous decline of CSD amplitudes at all three recording sites (Fig. 1B). At the end of the experiments after four hours, we were still able to elicit CSD with similar slopes, but CSD amplitudes were reduced significantly to 79.5 ± 2.7% of amplitudes at the beginning of the experiments ( Fig. 2A In contrast in PEA-treated rats, amplitudes of the experimentally induced CSD remained constant at all recording sites over the entire experimental observation period (Fig. 1C) with a statistically not significant decline to 92.1 ± 2.0% of baseline after four hours ( Fig. 2A,B; n = 5; P = 0.262). After a second systemic dosing with PEA this statistically not significant decrease continued, but reached only 83.6 ± 6.4% of the amplitudes at the beginning of the experiment two hours after that ( Fig. 2A,B; n = 5; P = 0.283). In the same time, CSD propagation velocity was unchanged statistically with a decline from 3.4 ± 0.3 mm/min to 3.0 ± 0.5 mm/min in the first four hours after the first application of PEA ( Fig. 2C; n = 5; P = 0.321) and to 2.8 ± 0.5 mm/min in the two hours after the second injection with PEA ( Fig. 2C; n = 5; P = 0.167). CSD duration at half maximal amplitude increased significantly from 23.12 ± 1.2 s to 49.2 ± 8.5 s (Fig. 2D; n = 5; P < 0.05) in the first four hours. In the two hours after the second injection with PEA, CSD duration at half maximal amplitude further increased significantly to 74.3 ± 11.8 s (Fig. 2D).
In a subset of experiments, we tested the effects of the IL-1β blocker, Anakinra. As evident from the sample traces (Fig. 3A), CSD amplitudes slowly decreased to 90.1 ± 2.8% of control in the untreated cortical area. In the treated area, however, such decrease did not occur and CSD amplitudes were stable. Summary data (Fig. 3B) indicate that blockade of IL-1ß receptors stabilizes CSD amplitudes, confirming that IL-1ß contributes to the run-down of CSD amplitudes in the cortical area superfused with ACSF. The slight increase in CSD amplitudes in the untreated area 240 min after beginning of the treatment might be due to diffusion of the drug below the skull and a drug effect in remote brain areas as well.

Discussion
The in vivo study showed that systemically applied PEA is able to keep the amplitude of single experimentally induced CSD events at a stable level for at least four hours, whereas in untreated controls, we measured a significant run-down of CSD amplitudes in the same time interval. In parallel, CSD propagation velocity was unaltered This novel finding that dosing with PEA prevents the run-down of CSD amplitudes differs from all observations on CSD in untreated control animals or untreated cortical areas that were used for example to determine the effects of topically applied cytokines 20 .
An inflammatory process in the exposed cortex of the healthy rat could be a reason for the rundown of the CSD amplitudes in the native control animals. Though the cortex surface in our experiments was superfused with warmed, equilibrated ACSF, this cannot prevent an activation of the microglia as a source of cytokines. In previous work we have shown, that the proinflammatory cytokine TNFα dose-dependently diminishes amplitudes of CSD 20 . TNFα is released after brain trauma (TBI) or after stroke [21][22][23][24][25] . Similarly, the proinflammatory cytokine IL-1ß is able to diminish CSD amplitudes 26 and also released after TBI and/or stroke 22,27 .
It has been shown previously that induction of repetitive CSD by KCl results in a 24-fold increase in IL-1β mRNA ipsilaterally within a 4 hr time window 28 . Importantly, such increase could not be induced by other stimuli, and was predominantly localized to ramified microglia in the cortex 28 . Our results provide evidence that blockade of the IL-1β receptor using Anakinra prevents CSD amplitude rundown in our experimental paradigm (Fig. 3), directly implicating cytokine signaling as the underlying physiological mechanism. There is evidence from the literature, that activation and secretion of IL-1ß is mediated via the inflammasome 29,30 which is formed from intracellular protein complexes that activate caspases needed to convert pro IL-1ß into bioactive IL-1ß 31,32 . The activators of the inflammasome are potassium efflux 33 and/or decrease of the intracellular potassium concentration either caused by the potassium efflux itself or by the influx of water from the extracellular space together with an influx of sodium ions 34 . All three elements needed for activation of the inflammasome occur during CSD: potassium ions transiently leave the neurons and glial cells, and in the same time sodium ions and water enter the cells 35 , leading to shrinkage of the extracellular space down to 35% of the normal values 36 . It is likely therefore, that activation of the inflammasome by CSD is the reason for a release of IL-1ß causing the decline in CSD-amplitudes. Indeed, this decrease was prevented, when the IL-1ß pathway was blocked by Anakinra.
Several previous reports have shown reduced levels of IL-1β following PEA administration. For example, Di Paola and colleagues measured significantly reduced IL-1β levels after PEA in their splanchnic artery occlusion (SAO) shock model 37 . Similarly, intrathecal administration of PEA attenuated the increase of IL-1β levels in the rat spinal cord 38 , and analogously, the fatty acid amide hydrolase (FAAH) inhibitor URB597 resulted in an up-regulation of endogenous PEA and an attenuation of induction of IL-1β in a model for Toll-like receptor 3-mediated neuroinflammation 39 . Altogether, it appears logical to speculate that PEA treatment counteracts the upregulation of IL-1β during CSD, thereby stabilizing CSD amplitudes.
Our previous studies have excluded cannabinoid receptors as mediators of PEA's neuroprotective properties in a preclinical model for stroke 11,12,14,16 . It has been established that activation of CB1 receptors by WIN 55212-2 dose-dependently suppressed CSD amplitude, duration and propagation velocity 40 . The same compound also decreased the spatial spread of neocortical excitation by interfering with glutamatergic neurotransmission in an in vitro assay 41 . By contrast, local application of WIN 55212-2 together with the AMPA receptor blocker CNQX facilitated the propagation of CSD in rat brain slices from neocortex into the hippocampus 42 . Therefore, a CB1 receptor-mediated mechanism underlying the prevention of a run-down of CSD amplitudes by PEA is unlikely.
Another potential mechanism for the action of PEA is a reduction of CSD-induced astrogliosis and reduced astrocyte-mediated cytokine release. It has previously been shown that potassium concentrations, similar to those occurring during CSD, result in reactive astrogliosis in the rat cortex 43 , hippocampus 44 , and spinal cord 45 . This research served as the basis for our own previous work, in which we concluded that changes in the extracellular space volume and geometry after CSD are the result of astrogliosis 36 . Similarly, astrogliosis has more recently been described after status epilepticus 46 , which shares many aspects of hyperexcitability with acute CSD. Ahmad and colleagues 47 have reported that PEA blocked the infiltration of astrocytes after ischemic brain injury in mice and inhibited the expression of pJNK, NF-κ B and degradation of Iκ B-α 47 . Signaling pathways controlled by these three mediators are linked to the release of proinflammatory cytokines and/or to the control of their actions after receptor binding.
In summary, our data support the notion that systemic application of PEA in our experimental paradigm prevented or reduced the release of proinflammatory cytokines. Such cytokines are released in our experimental model as a consequence of trephination of the skull and exposure of the brain surface, leading to a slow rundown of the CSD amplitude.