In the current study we show, first, that ethanol withdrawal increases both basal neuronal death and NMDA-stimulated neuronal death in primary cortical cultures, probably due to a change in the expression of NMDA receptor subunits. Second, we evidence for the first time that, during ethanol withdrawal, the stimulation of the endocannabinoid system protects neurons from excitotoxic insults, whereas its blockade aggravates neuronal death induced by NMDA. We show that the neuroprotective effect of HU-210 during ethanol withdrawal could be explained by a reduction in Ca²⁺ influx through NMDA receptors.

The increase in basal neuronal death found during alcohol withdrawal, which is concordant with previous results [11], suggests that chronic alcohol administration and subsequent withdrawal could decrease the threshold for triggering neurotoxic processes per se. This hypothesis is compatible with the enhanced brain damages observed in alcohol-dependent individuals who experienced successive periods of alcohol consumption, withdrawal and relapse [12], as well as in several in vivo experimental models of alcoholism [2]. In addition, we observed a significant increase in the sensitivity to NMDA toxicity in withdrawn neurons. This enhancement of excitotoxicity may reflect the neuronal adaptation to chronic ethanol exposure and could contribute to the generation of withdrawal symptoms such as seizures, pro-convulsive states and might also initiate excitotoxicity and loss of neurons in certain brain areas during alcohol withdrawal [12]. It is well accepted nowadays that changes in the glutamatergic system are one of the principal events responsible for the neuropathological alterations induced by chronic alcohol consumption in both laboratory animals and humans [2]. Our findings support the idea that NMDA receptor subunit composition changes after chronic alcohol administration and subsequent withdrawal, since we detected a decrease in GluN2A expression levels, whereas no changes were detected for GluN1 or GluN2B. It has been suggested that GluN2A-containing NMDA receptors promote neuronal survival and would exert a neuroprotective role against NMDA, while GluN2B-containing NMDA receptors would mediate neurotoxic processes [13]. If this hypothesis is correct, the decrease in GluN2A subunits induced by alcohol withdrawal would lead to an imbalance between the numbers of “trophic” versus “pro-neurotoxic” subunits and could explain the increase in neuronal death during ethanol withdrawal.

The implication of the endocannabinoid system in alcohol addiction has been largely studied in the last years. However, most studies have tried to reduce alcohol reinforcing effects or alcohol consumption [14], and less attention has been paid to the effects of cannabinoid stimulation or blockade on the neuropathological consequences of alcohol dependence and withdrawal. Our study shows that the stimulation of the endocannabinoid system protects ethanol-withdrawn neurons from excitotoxic cell death. This neuroprotection is present when the non-specific cannabinoid agonist HU-210 is administered both chronically (during ethanol administration and ethanol withdrawal) and acutely (only during ethanol withdrawal). Cannabinoid agonists have been proposed as neuroprotective molecules in several models of acute neuronal injury and chronic neurodegenerative diseases [15]. Among the proposed mechanisms of cannabinoid-receptor mediated neuroprotection, two are particularly relevant for the neurotoxicity caused by alcohol dependence and withdrawal: the inhibition of the presynaptic release of glutamate [16] and the inhibition of NMDA-induced Ca²⁺ influx [17]. It appears well-demonstrated that CB1 receptors are involved in these two neuroprotective properties of cannabinoid agonists and although HU-210 is a non-selective cannabinoid agonist, we assume that, in our experiments, HU-210 preferentially acts through the activation of CB1 receptors because -consistent with the suggestion that CB2 receptors are essentially present in glial cells [18] and considering the high purity of our neuronal cultures (thanks to the use of an anti-mitotic agent)- no mRNA encoding for CB2 receptors were detected in our cultures (data not shown).

As mentioned above, the neuroprotective effect of HU-210 against NMDA was exclusive of alcohol-withdrawn neurons, as it did not appear in control neurons (not exposed to alcohol). The absence of neuroprotection of HU-210 against NMDA excitotoxicity in cultured neurons has been previously reported [19], and the neuroprotective effect of HU-210 exclusively in ethanol-withdrawn neurons suggests that the neuroadaptive changes that occur during alcohol exposure sensitize cells to cannabinoid neuroprotective mechanisms, perhaps by enhancing CB1 receptor number and/or their signaling mechanisms. In agreement with this hypothesis, we found a 2-fold increase in the levels of CB1 receptor transcripts after alcohol cessation, which may explain the effects of HU-210 against NMDA excitotoxicity during alcohol withdrawal and its lack of efficacy in control neurons. To our knowledge, this is the first time that a neuroprotective role of HU-210 is found in conditions of alcohol withdrawal. In addition, we also found that the acute blockade of CB1 receptor-mediated signals during alcohol withdrawal with rimonabant increased NMDA-stimulated neuronal death, and again this neurotoxic effect was absent in control conditions. The data obtained with rimonabant also support the idea that the effects of HU-210 were CB1 receptor-mediated, as rimonabant is a selective CB1 receptor antagonist. The exacerbation of excitotoxin-induced neuronal death by inhibition of the cannabinoid receptors suggests that these neurons constitutively produce a cannabinoid receptor agonist. In accordance with this hypothesis, several authors have reported an increase in endocannabinoid production (anandamide and/or 2-arachidonoyl glycerol) after chronic alcohol exposure both in vitro and in vivo [20]–[24].

Excessive Ca²⁺ influx after the activation of NMDA receptors triggers neuronal death, whereas the suppression of the Ca²⁺ entry can protect cells from NMDA-induced cytotoxicity [25]. Given the inhibitory role played by CB1 receptors in Ca²⁺ homeostasis, the effects of CB1 receptor activation or blockade are likely followed by the corresponding reduction or increase in Ca²⁺ influx. Thus, we performed single cell calcium videomicroscopy analyses to determine the NMDA-evoked Ca²⁺ influx after the incubation with HU-210 or rimonabant. Consistent with its neuroprotective effect, the addition of HU-210 to alcohol-withdrawn neurons reduced NMDA-evoked Ca²⁺ influx. This effect was observed only in ethanol-withdrawn neurons, and not in control neurons (not exposed to ethanol; data not shown), in concordance with the results obtained in the excitotoxicity experiments. The addition of rimonabant to ethanol-withdrawn neurons, however, did not modify Ca²⁺ influx. Our data are supported by previous studies that described a modulation of NMDA-induced Ca²⁺ influx by CB1 agonists in different brain structures and through different mechanisms, although these studies were not conducted in conditions of alcohol withdrawal [17] [26].

In our study, the blockade of the endocannabinoid system with the CB1 antagonist rimonabant (both acutely and chronically) increased excitotoxic neuronal death in ethanol-withdrawn neurons. Moreover, the chronic blockade of the endocannabinoid system exacerbated excitotoxicity not only in ethanol-withdrawn neurons but also in control neurons (not exposed to ethanol). This suggests that the pharmacological blockade of the endocannabinoid transmission, especially during alcohol withdrawal, is highly counterproductive due to its ability to aggravate NMDA-stimulated cytotoxicity. This observation deserves further comments derived from its extrapolation to the in vivo situation. Increases in endocannabinoid levels have been described both in vivo and in vitro after acute [27] and more strikingly after chronic alcohol exposure [20] [22] [28], possibly in a cellular attempt to reduce damages due to alcohol [20] [22]. We can thus hypothesize that chronic CB1 receptor blockade, especially during alcohol exposure and withdrawal increases the sensibility to NMDA-induced cytotoxicity, which is concordant with the protective function assigned to the endocannabinoid generation in the above-mentioned studies. In any case, further experiments are needed to determine more precisely the cause of the deleterious effect of rimonabant during alcohol withdrawal.

Rimonabant has been proposed as a promising drug for the treatment of alcoholism, with positive effects on alcohol intake in laboratory animals under several administration protocols [29]. However, these preclinical data were not reproduced in the two clinical studies conducted to date with rimonabant (one in alcohol-dependent individuals and one in non treatment-seeking heavy alcohol drinkers) [30]–[31]. Moreover, further studies on the therapeutic potential of rimonabant in the treatment of alcohol dependence are apparently hindered, since the recent decision of the discontinuation of all clinical trials with this CB1 antagonist due to the occurrence of some psychiatric adverse effects. The results of the present work point in the same direction showing, for the first time, the potential neurotoxic effect of chronic administration of rimonabant in an in vitro model of alcohol withdrawal. It is also possible that the reason why rimonabant aggravated the deleterious effects of alcohol withdrawal in our in vitro model is related to its inverse agonist properties. If this is the case, it is likely that we will need the development of novel neutral antagonists, with no inverse agonist activity, that may be used efficaciously for treating alcohol craving with no additional neurotoxic effects.

In summary, these observations show, for the first time, that the stimulation of the endocannabinoid system could be protective against the hyperexcitability developed during alcohol withdrawal. By contrast, the blockade of the endocannabinoid system seems to be highly counterproductive during alcohol withdrawal due the ability of rimonabant to exacerbate NMDA-induced excitotoxicity.