An endogenous cannabinoid (2-AG) is neuroprotective after brain injury

Traumatic brain injury triggers the accumulation of harmful mediators that may lead to secondary damage(1),(2). Protective mechanisms to attenuate damage are also set in motion2. 2-Arachidonoyl glycerol (2-AG) is an endogenous cannabinoid, identi®ed both in the periphery(3) and in the brain(4), but its physiological roles have been only partially clari®ed(5)±(7). Here we show that, after injury to the mouse brain, 2-AG may have a neuroprotective role in which the cannabinoid system is involved. After closed head injury (CHI) inmice, the level of endogenous 2- AG was signi®cantly elevated.We administered synthetic 2-AG to mice after CHI and found signi®cant reduction of brain oedema, better clinical recovery, reduced infarct volume and reduced hippocampal cell death compared with controls. When 2-AG was administered together with additional inactive 2-acyl-glycerols that are normally present in the brain, functional recovery was signi®cantly enhanced. The bene®cial effect of 2-AGwas dosedependently attenuated by SR-141761A, an antagonist of the CB1 cannabinoid receptor.

An endogenous cannabinoid (2-AG) is neuroprotective after brain injury

David Panikashvili*Ç, Constantina Simeonidou*, Shimon Ben-ShabatÇ, LumõÂr HanusÏÇ, Aviva BreuerÇ, Raphael MechoulamÇ & Esther Shohami*

* Department of Pharmacology, and Ç Department of Medicinal Chemistry and Natural Products, Medical Faculty, Hebrew University, Jerusalem 91120, Israel

Traumatic brain injury is a major cause of mortality and morbidity, yet there is no effective drug to treat brain-injured patients. Understanding post-trauma events and developing neuroprotective agents are therefore important(8). Cannabinoids act through two distinct receptors, one of which (CB1) is found mainly in the brain. We have previously reported that 2-AG suppresses formation of reactive oxygen species (ROS) and tumour necrosis factor-a (TNF-a) by murine macrophages in vitro after stimulation with lipopolysaccharide (LPS), and have noted lower levels of TNF-a in the serum of LPS-treated mice after administration of 2-AG(9). Both classes of mediators, ROS(10) and TNFa(11), are major contributors to the pathophysiology of brain injury. On the basis of these observations, we assumed that 2-AG could have a neuroprotective role in the post-traumatic brain. We therefore investigated the dynamic changes in brain 2-AG levels after traumatic brain injury; the possibility that exogenous 2-AG may attenuate brain damage after CHI; and the involvement of the CB1 receptor in neuroprotection. Mice were subjected to CHI using a weight-drop device, or to sham surgery, as described elsewhere(12). They were decapitated at various time intervals (15 min, 1, 4, 8 or 24 h) and the brains were removed within 1 min and frozen in liquid nitrogen. The rapid freezing minimized the post-mortem production of 2-AG.

This was con®rmed by the similar levels of 2-AG obtained in brains extracted at 1, 3 or 5 min after decapitation (data not shown). The levels of brain 2-AG were determined in the traumatized hemispheres (Fig. 1) and compared with those in controls. One hour after CHI the levels of 2-AG were already signi®cantly higher than in sham, non-injured mice, peaking at a tenfold increase after 4 h, and declining thereafter. Even after 24 h, the levels of 2-AG were still higher (sixfold) than in controls. In the contralateral hemisphere, the changes in 2-AG were minor, and not signi®cant (data not shown). To test the role of 2-AG in post-traumatic pathophysiology, we synthesized 2-AG as previously described3 and intravenously administered doses of 0.1, 5 and 10 mg kg-1 15 min after CHI. One of the early manifestations of brain injury is oedema formation (accumulation of water in the tissue), leading to increased intracranial pressure. We therefore examined the effect of 2-AG on oedema formation by measuring water content 24 h after trauma, the time of maximal oedema(12). At all doses tested, 2-AG signi®cantly reduced water accumulation by about 50% (Fig. 2a). The clinical status of the mice was evaluated 1 and 24 h after CHI using a scoring system testing motor and behavioural functions(13) (neurological severity score, NSS: 0, healthy, to 10, fatally injured).

The initial NSS (at 1 h) re¯acts the severity of injury, and DNSS, the difference between the score at 1 and 24 h, re¯ects recovery.Whereas the NSS (1 h) was similar in the treated and control groups (8.256 0.36 and 7.6360.46, respectively), the mice treated with 2-AG recovered faster within the ®rst 24 h. All doses of 2-AG showed a trend towards better recovery (greater DNSS), but it reached signi®cance only at 5mg kg-1 (P,0.01, versus control; Fig. 2b). The effect of 2-AG was not long lasting. We injected 2-AG 1 h after CHI, and evaluated NSS 1, 4 and 7 days later, but found signi®cant protection on the ®rst day only, and not on subsequent days (Fig. 2c). Brain oedema and neurological de®cits are not necessarily interdependent, because they represent different manifestations of brain damage. Indeed, we found that 2-AG is less potent against the motor de®cits than in reducing oedema, as evidenced from the different effective doses needed to achieve protection. One of the manifestations of brain injury is neuronal cell death, which is grossly evidenced by an infarct around the site of injury. Mice were treated with either 2-AG (5mg kg-1, n = 7) or vehicle (n = 12), decapitated 24 h later and their brains were examined for changes in infarct volume(14). Infarct volume was signi®cantly reduced in 2-AG-treated mice compared with that in vehicle-treated mice (5.661.2% and 10.261.1%, respectively, of total brain volume; P = 0.021).

No infarct was observed in sham-operated mice (data not shown). In addition, we examined the effect of 2-AG on hippocampal cell death, which occurs speci®cally in the CA3 sub®eld(12). Mice were treated with vehicle or with 5mg kg-1 of 2-AG 1 h after CHI, and killed 7 days later by perfusion ®xation. Serial brain sections were stained with haematoxylin and eosin and Nissl. The ipsilateral hippocampus of vehicle- and 2-AG-treated mice showed ischaemic and dystrophic neurons, with large or small patchy acellular regions, in the CA3 sub®eld (Fig. 3). No neuronal pathology could be detected in the contralateral CA3 region (Fig. 3a, d). Neuronal loss was apparent in the injured side compared with the contralateral side of both vehicle- and 2-AG-treated mice (Fig. 4a); however, this loss was signi®cantly more obvious in the vehicle-treated animals. The number of neurons counted in the CA3 area in the injured hemisphere, expressed as a percentage of those counted in the CA3 area in the intact hemisphere, was signi®cantly lower in the vehicle-treated mice than in the 2-AG-treated mice (Fig. 4b).

2-AG is accompanied in mouse brains by an `entourage’ of congeners, such as 2-palmitoyl-glycerol and 2-linoleoyl-glycerol, which alone have no activity at the cannabinoid receptors. However, these glycerol esters facilitate the binding of 2-AG to the CB1 and CB2 receptors, probably through increasing the availability of 2-AG at the receptors by inhibiting its hydrolysis and uptake(15). To test the effect of the entourage after CHI, mice were injected with a combination of 2-AG (1mg kg-1), 2-palmitoyl-glycerol (5 mg kg-1) and 2-linoleoyl-glycerol (10 mg kg-1) 1 h after CHI. Although no effect was observed when these compounds were injected separately, the combination of the three acyl-glycerol esters led to a signi®cant improvement in functional recovery at 24 h (Fig. 5a). When the combination (2-AG plus entourage) was administered again 24 and 48 h after CHI, the improved clinical recovery was sustained up to 7 days after the trauma (Fig. 5b). To determine whether the protective effect of the exogenous 2-AG is mediated at least in part by the CB1 cannabinoid receptor, we injected SR-141716A, an antagonist speci®c to the CB1 receptor subtype, after CHI, along with 2-AG. Oedema was determined 24 h later. Five groups of mice were subjected to CHI and treated with either vehicle (control), 2-AG (1mg kg-1) alone or with 3, 10 and 20 mg kg-1 of SR-141716A. The protective effect of 2-AG was dosedependently attenuated by the antagonist (Fig. 6).

The water content in mice treated with 2-AG and 20 mg kg-1 of antagonist did not differ from that in the CHI control mice. The doses of the antagonist used were relatively high; however, the effective dose could depend upon the species or strain of animal used in the study. High doses (<20 mg kg-1) of the antagonist have previously been used(16),(17). It should be noted, however, that the antagonist alone did not worsen the outcome of traumatized mice not treated with 2-AG (data not shown). These results are, to our knowledge, the ®rst to demonstrate temporal and local changes in brain levels of the endogenous cannabinoid 2-AG after traumatic brain injury and to record its neuroprotective effects. We have previously reported, using the same model of CHI, that the arachidonic acid metabolic cascade is activated after trauma(18), and that there is massive accumulation of calcium at the site of injury(19).

Both events may be related to 2-AG release. It seems that 2-AG can be synthesized and released from neuronal and non-neuronal cells on demand by a non-synaptic release mechanism5. Endocannabinoids are taken up by cells using a speci®c transporter(20) or, in particular for 2-AG, also by passive diffusion(5) and are hydrolysed by fatty acid amide hydrolase (FAAH), which acts on both anandamide and 2-AG5,(21). Several cannabinoids that bind to CB1 have been assayed for neuroprotection (mostly using in vitro models). 2-AG has been shown to protect cerebral neurons of rats from ischaemia in vitro(22), and cannabinoid receptor agonists inhibit glutamatergic synaptic transmission in hippocampal cultures in vitro(23). The synthetic cannabinoid WIN 55212-2 decreases hippocampal neuronal loss after transient global cerebral ischaemia induced by occlusion of the middle cerebral artery in rats(24). Indeed, our present ®ndings show signi®cant protection of hippocampal cells after treatment with 2- AG. Tetrahydrocannabinol was also shown to protect against pathophysiological events, attenuating excitotoxicity and protecting the brain during development of secondary injury. Inhibition of this protective effect by SR-141716A suggests that 2-AG and the CB1 receptor may be important in the pathophysiology of traumatic brain injury. However, because of the high dose of the CB1 antagonist required to block the 2-AG protective effect, a non- CB1-mediated mechanism of 2-AG cannot be excluded. M

Methods

Animals

The study was performed according to the guidelines of the Institutional Animal Care Committee.Male Sabra mice (Hebrew University strain) weighing 32±35 g were used. The animals were divided into groups, treated with different doses of 2-AG and killed at different times after CHI or sham surgery.

Trauma model

Trauma was induced under ether anaesthesia, which was con®rmed by testing loss of pupillary and corneal re¯exes, as described previously in detail(12). To assess the functional impairment after trauma, the NSS scoring system was used based on the ability of themice to perform ten different tasks that evaluate motor ability, balancing and alertness(13). Cerebral oedema was evaluated by determining the tissue water content in the injured mbrain, as previously described(12). The percentage of water in the tissue was calculated as ((wet weight – dry weight)/wet weight) ´ 100.

Analysis of brain 2-AG levels

Brains were frozen at the designated post-CHI times. Tissue segments were homogenized in a mixture of chloroform and methanol (2 : 1) containing 10 ml of 1mM arachidinin as an internal standard. Vacuum ®ltration was carried out and the ®lter washed at least three times with the organic mixture. The solvents were evaporated to dryness. The solid residue was re-dissolved in chloroform and dried under nitrogen, and fatty acids were separated from other lipids using thin-layer chromatography (TLC) plates with ¯uorescent indicator. Synthetic 2-AG in chloroform was applied alongside the analysed fraction to help in the identi®cation of endogenous 2-AG. The solvent system for developing of sample was hexane, ether, acetone and acetic acid (40 : 20 : 7 : 1, v/v). Using a visualizing reagent (KMnO4), the 2-AG was identi®ed, scraped, removed from the silica gel and placed into tubes containing chloroform and methanol (19 : 1); the fractions were collected and evaporated to dryness. The solid residue was dissolved in chloroform and dried under a stream of nitrogen.

GC±MS analysis

We quantitatively analysed the samples by gas chromatography±mass spectrometry (GC±MS) in a Hewlett-Packard G1800A GCD system ®tted with a 5% phenylmethylsiloxane (HP-5MS) capillary column (30m ´ 0.25mm (internal diameter) ´ 0.25 mm), temperature programmed from 150 to 280 8C at 50 8Cmin-1. The selected quali®er ions were (m/z) 74, 103 and 129 for 2-AG and 57, 73 and 147 for arachidinin. Quanti®er ions were 218 for 2-AG and 427 for arachidinin, respectively. 2-AG amounts of 0.5±16 nmol were used for the calibration curve together with 10 nmol arachidinin. To the solid residue, which was separated by TLC, dissolved in chloroform and dried again, 10 ml of bis(trimethyl-silyl)tri¯uoroacetamide (BSTFA) was added to silylate the free hydroxyl groups. Twomicrolitres of this solution were injected into the GC-MS. 2-AG and the standard arachidinin were eluted after 540 and 620 s, respectively. The ratio between 2-AG and the standard was used to calculate the level of 2-AG in the brain.

Evaluation of infarct volume

Mice were traumatized and treated with vehicle or 2-AG, and after 24 h their brains were sliced every 2mm using a brain mould. The slices were placed in a 2% solution of 2,3,5- triphenyltetrazolium chloride (TTC) in PBS buffer(14) and photographed with a Stereoscope Stemi SV(11) (Zeiss) and digital photocamera Coolpix E990 (Nikon). Scion Image- Release Beta 4.0.2 program was used to quantify the infarct volume. Evaluation of hippopcampal cell death Mice were killed by perfusion ®xation with 4% phosphate-buffered formaldehyde under ether anaesthesia and brains were post-®xed with the same ®xative. Serial 6-mm brain cryostat sections, cut coronally, anterior to posterior, were stained with haematoxylin and eosin and Nissl, and two uninformed observers reviewed the preparations independently. Cell counting was performed under light microscope and camera lucida on the CA3 sub®eld of dorsal hippocampus in both hemispheres, in 12±15 randomly chosen sections from each brain. By de®nition, the CA3 area started at the CA3±CA2 border up to the point where the CA3 neurons enter the dentate gyrus, as de®ned by the lateral aspect of the granule cell layer’s dorsal and ventral leaves (Fig. 3a). Ischaemic and necrotic neurons were recognized according to morphological criteria. In an attempt to estimate the neuronal loss in the CA3 area of the injured side, the percentage of the number of neuronal cells counted in this area was evaluated against the contralateral control CA3 sub®eld. All neurons were included, no matter whether they were normal, ischaemic or necrotic. A paired t-test and t-tests for independent samples of groups were performed.

Acknowledgements

We thank the US National Institute of Drug Abuse and the Israel Science Foundation for support (grants to R.M.). E.S. and R.M. are af®liated with the David R. Bloom Center for Pharmacy at the Hebrew University of Jerusalem’s School of Pharmacy. Correspondence and requests for materials should be addressed to E.S. (e-mail: esty@cc.huji.ac.il).