Pharmaceutical drugs

Non-steroidal anti-inflammatory agents (NSAIDs)
NSAIDs inhibit two cyclooxygenase (COX) enzymes, COX1 and COX2, and thereby block the conversion of arachidonic acid (AA) into inflammatory prostaglandins. Ibuprofen, ketorolac, and flurbiprofen also block the hydrolysis of AEA into arachidonic acid and ethanolamine [27]. See Figure 2. A binding site for some NSAIDs on FAAH has also been identified [28]. NSAID inhibition of COX2 blocks the metabolism of AEA and 2-AG into prostaglandin ethanolamides (PG-EAs) and prostaglandin glycerol esters (PG-GEs), respectively [29]. PG-EAs and PG-GEs increase the frequency of miniature inhibitory postsynaptic currents (mIPSCs) in primary cultured mouse hippocampal neurons, an effect opposite to that of their parent molecules [30].
10.1371/journal.pone.0089566.g002 Figure 2  Anandamide (top figure) is metabolized into arachidonic acid and ethanolamine (bottom figures). Prostaglandin E2 glycerol ester (PGE2-GE), a COX2 metabolite of 2-AG, induced mechanical allodynia and thermal hyperalgesia in rat paws [31]. PGF2α-EA, a COX2 metabolite of AEA, was found in the spinal cord of mice with carrageenan-induced knee inflammation. PGF2α-EA contributed to pain perception and dorsal horn nociceptive neuron hyperactivity, thus providing a rationale for the combined use of COX2 and FAAH1 inhibitors against inflammatory pain [32].
Electrophysiology studies of rat hippocampal cells showed that meloxicam and nimesulide prolonged and increased DSI; that is to say, the COX2 inhibitors potentiated synaptic 2-AG release and CB1 signaling [33]. Consistent with this, intrathecally applied indomethacin enhanced eCB-mediated antinociception in mice that was blocked by the CB1 antagonist AM251 [34]. Intrathecally applied flurbiprofen produced a similar eCB-dependent antinociception in the rat formalin test [35].
Combining NSAIDs with cannabinoids (either eCBs or exogenous cannabinoids) produces additive or synergistic effects. A sub-effective dose of WIN55,212-2 became fully antinociceptive following administration of indomethacin in rats [36]. A local injection of ibuprofen plus AEA in the rat formalin test produced synergistic antinociceptive effects involving both CB1 and CB2 [37]. The FAAH inhibitor URB937, when coadministered to mice with indomethacin, produced a synergistic reduction in pain-related behaviors [38]. Furthermore, URB937 reduced the number and severity of gastric lesions produced by indomethacin. One contrary study showed that THC's decrease in intraocular pressure was partially blocked by indomethacin in rabbits [39].
In a small human study, the administration of indomethacin antagonized marijuana effects [40]. Yet a high dose of ibuprofen may cause sedation, possibly a cannabimimetic effect [41]. Clinical anecdotes of NSAIDs eliciting cannabimimetic effects have been reported; the individuals are usually familiar with the effects of cannabis, and usually females [42].
In summary, preclinical studies indicate that some NSAIDs inhibit FAAH and enhance the activity of eCBs, phytocannabinoids, and synthetic cannabinoids. Combinational effects may be particularly relevant at peripheral sites, such as the peripheral terminals of nociceptors.

Acetaminophen
Acetaminophen (paracetamol), following deacetylation to its metabolite p-aminophenol, is conjugated with AA to form N-arachidonoylphenolamine (NAP, aka AM404). It is likely that deacetylation takes place mainly in the liver, and conjugation occurs in the central nervous system. NAP blocks the breakdown of AEA by FAAH, inhibits COX1 and COX2, and acts as a TRPV1 agonist [43]. The analgesic activity of acetaminophen in rats is blocked by CB1 or CB2 antagonists [44], [45]. Analgesic activity is also lost in CB1 −/− knockout mice [46]. A sub-effective dose of the synthetic cannabinoid WIN55,212-2 became effective following intracisternal administration of acetaminophen in rats [36]. A sub-effective dose of AEA in mice became anxiolytic in the Vogel conflict test and the social interaction test when co-administered with acetaminophen; the effect was blocked by the CB1 antagonist AM251 [47].
Small amounts of acetaminophen are also metabolized via the cytochrome P-450 pathway into N-acetyl-p-benzoquinone imine (NAPQI). Intrathecal administration of NAPQI activates TRPA1 and imparts antinociception in the mouse hot-plate test, and a similar action is found for Δ9-tetrahydrocannabiorcol. These effects are lost in Trpa1(−/−) mice [48].
In summary, preclinical studies indicate that acetaminophen enhances the activity of eCBs and synthetic cannabinoids in rodents. Why acetaminophen fails to elicit cannabimimetic effects in humans is unknown. Acetaminophen-cannabinoid drug interactions may be species-specific; Gould et al. [49] demonstrated strain-specific differences in mice. They suggested that other indirect actions of acetaminophen, including 5-HT receptor agonism, may outweigh any CB1 mediated effects in some mouse strains.

Glucocorticoids
The distribution of glucocorticoid receptors (GRs) and CB1 overlap substantially in the central nervous system and other tissues, as do GRs and CB2 in immune cells. Dual activation of GRs and CBs may participate in glucocorticoid-mediated anti-inflammatory activity, immune suppression, insulin resistance, and acute psychoactive effects. In a rat model of spinal nerve injury (sciatic nerve constriction with suture loops), the GR receptor agonist dexamethasone increased CB1 density after spinal nerve injury, which suggests that CB1 is a downstream target for GR actions [50]. Glucocorticoid administration also induced CB1 expression in bone in mice [51] and rats [52].
The acute administration of glucocorticoids may shift AA metabolism toward eCB synthesis in parts of the brain. Electrophysiological studies of rat hypothalamic slices demonstrated that adding dexamethasone or corticosterone to slice baths caused a rapid suppression of synaptic activity, characterized as glucocorticoid-induced, eCB-mediated suppression of synaptic excitation (GSE). GSE was blocked by CB1 antagonists, indicating that eCB release mediated GSE [53]. A follow-up study demonstrated that GSE correlated with increased levels of AEA and 2-AG [54]. The same group found no changes in AEA and 2-AG after exposure of cerebellar slices to dexamethasone. In hypothalamic slices, GSE could be blocked by leptin, suggesting that GSE is a nutritional state-sensitive mechanism [55]. Dexamethasone enhanced eCB-mediated GSE by inhibiting COX2 in dorsal raphe serotonin neurons [56].
Corticosterone administration increased AEA levels in several rat limbic structures (amygdala, hippocampus, hypothalamus), but not the prefrontal cortex. 2-AG levels were only elevated in the hypothalamus [57]. The same group conducted an ex vivo study of the rat medial prefrontal cortex (mPFC). Bath application of corticosterone to mPFC slices suppressed GABA release onto principal neurons in the prelimbic region, which was prevented by application of the CB1 antagonist AM251 [58]. This indicates local recruitment of eCB signaling, probably through 2-AG. A previous study of rats receiving a single dose of corticosterone detected no change in 2-AG and a reduction of AEA in hippocampal homogenates [59]. Corticosterone increased hippocampal levels of 2-AG in rats; the impairment of contextual fear memory by corticosterone was blocked by the CB1 antagonist AM251 [60].
Chronic exposure to glucocorticoids downregulates the eCB system. Chronic corticosterone administration decreased CB1 densities in rat hippocampus [59] and mouse hippocampus and amygdala [61]. Chronic corticosterone administration in male rats led to visceral hyperalgesia in response to colorectal distension, accompanied by increased AEA, decreased CB1 expression, and increased TRPV1 expression in dorsal root ganglia. Co-treatment with the corticoid receptor antagonist RU-486 prevented these changes [62].
In summary, preclinical rodent studies indicate that acute glucocorticoid administration enhances the activity of eCBs. The clinical phenomenon of acute “corticosteroid mania” may have a cannabimimetic component. Chronic exposure to glucocorticoids downregulates the eCB system, a scenario consistent with chronic stress, which we review below.

Opiates
Naloxone, a μ-opioid receptor antagonist, inhibited THC-induced Fos immunoreactivity in several regions of the rat central nervous system, including the ventral tegmental area, hypothalamus, caudate-putamen, and periaqueductal grey. Conversely, naloxone and THC had an additive effect on Fos immunoreactivity in the amygdala, stria terminalis, insular cortex, and paraventricular nucleus of the thalamus [63].
Short-term co-administration of morphine with THC caused an upregulation of CB1 protein in the spinal column of rats, far greater than THC or morphine given alone [64]. A rodent study of chronic but voluntary intake of opiates (rats self-administering heroin) enhanced [3H]CP55,940 binding in the amygdala and ventral tegmental area, plus a marked increase in cannabinoid-stimulated [35S]GTPγS binding in the nucleus accumbens, caudate putamen, and amygdala [65]. Superperfusion of ex vivo rat nucleus accumbens slices with 4-aminopyridine and NMDA released glutamate and GABA, respectively, and either morphine or the CB1 agonist HU210 predictably inhibited these responses. Combining HU210 and morphine caused a synergistic inhibition of GABA release, but a non-additive response in glutamate release [66].
Chronic morphine exposure in rats caused a reduction in hippocampal and cerebellar CB1 density measured with [3H]CP55,940, and a strong reduction in CP55,940-stimulated [35S]GTPγS binding; 2-AG contents were also reduced [67]. Another rat study showed that chronic morphine exposure caused variable, regionally-specific modulations in [3H]CP55,940 binding and CB1 mRNA levels; CB1 upregulated in some regions and dowregulated in other regions [68]. In human CB1-transfected HEK293 cells, morphine induced a desensitization of the μ-opioid receptor and heterologous desensitization of CB1, demonstrated by a reduction in WIN55212-2-induced [Ca2+]i release [69]. μ-opioid receptor knockout mice showed a dramatic reduction in WIN55212-2-stimulated [35S]GTPγS binding [70]. In human SH-SY5Y neuroblastoma cells, sequential activation of CB1 and δ-opioid receptor produced synergistic elevations of intracellular Ca2+, a response that each receptor alone did not trigger in an efficacious way [71].
In behavioral studies, heroin reinstated “drug-seeking” behavior for WIN55,212-2 in rats [72]. Morphine did the same for THC in monkeys [73]. The rewarding effects of THC, measured by conditioned place-preference, were reversed by naloxone in rats [74]. In rats trained to discriminate THC, morphine administration markedly potentiated the THC discriminative stimulus [75]. Morphine or codeine potentiated THC-induced antinociception and analgesia in mice and rats [76]–[79]; inactive doses of the drugs in combination produce potent, synergistic analgesia [80]. Synergistic analgesia was confirmed in an isobolographic analysis [64]. Historically this is the first isobolographic analysis of a cannabinoid since the days Walter Siegfried Loewe, who invented the isobologram to test drug combinations for synergy [81]. Loewe demonstrated synergy generated by cannabis extracts combined with other drugs [82], [83], as well as synergy generated amongst the individual components within cannabis itself [84], [85].
Normal men subjected to a thermal pain stimulus did not experience analgesia from a low dose of nabilone (a synthetic THC analogue), or a low dose of morphine. But co-administration of the drugs produced an analgesic effect [86].
Endorphins (endogenous opioids) enhance the effects of cannabinoids: Administering a low dose of THC to rats produced an anxiolytic response in the light-dark box test, which was abolished by beta-funaltrexamine, a μ-opioid receptor antagonist [87]. In rats trained to discriminate THC, microinjection of β-endorphin into the ventral tegmental area potentiated the THC discriminative stimulus [75]. Enkephalins (endogenous opioids) also enhance the effects of THC: the inhibition of encephalin-degrading enzymes augmented THC-induced antinociception in mice, an effect blocked by either rimonabant or naloxone [88]. Naltrexone, a μ- and κ-opioid receptor antagonist, significantly increased many of the “positive” subjective effects of oral THC [89] and smoked cannabis [90] in marijuana smokers. These results suggest that endogenous opioids contribute to the effects of cannabis.
In summary, preclinical studies and clinical trials indicate that acute opiate administration enhances the activity of eCBs, phytocannabinoids, and synthetic cannabinoids. Acute opiates may also upregulate CB1 expression. Chronic opiate administration, however, may have a deleterious effect on the eCB system.

Antidepressant drugs
Serotonin selective uptake inhibitors (SSRIs), tricyclic antidepressants (TCAs) and monoamine oxidase inhibitors (MAOIs) are the most commonly prescribed antidepressant drugs. Treatment with fluoxetine, the archetypal SSRI, potentiated THC-induced hypothermia in rats [91], but did not change THC-induced behavioral effects—freezing behavior, social interaction or exploration, and preference for outer or inner zones [92]. Fluoxetine increased CB1 binding density in the prefrontal cortex, without altering AEA or 2-AG levels in rat brains [93]. Chronic fluoxetine also increased WIN55212-2-stimulated [35S]GTPγS binding in the rat prefrontal cortex [94]. Conversely, treatment with citalopram reduced HU210-stimulated [35S]GTPγS binding in the rat hypothalamus and hippocampus [95].
Treatment with fluoxetine prevented synaptic defects in mice induced by chronic unpredictable stress (the CUS protocol included inversion of day/night light cycle, 45° tilted cage, cage rotation, tube restraint, predator sounds, strobe lights, food and water deprivation, cold environment, and wet bedding), and CUS preserved eCB- and WIN55,212-2-stimulated CB1 signaling [96]. In the hands of Mato et al. [97], fluoxetine in rats enhanced the inhibition of adenylyl cyclase by WIN55212-2, but did not alter WIN55212-2-stimulated [35S]GTPγS binding or CB1 density measured with [3H]CP55,940. They proposed that fluoxetine enhanced WIN55212-2 signaling through Gαi2 and Gαi3 subunits and not through Gαo subunits.
Treatment with the TCA desipramine increased CB1 binding density in the hippocampus and hypothalamus, without significantly altering AEA or 2-AG levels in rat brains [98]. The CUS protocol altered CB1 density in rat brains, and these changes were attenuated by concurrent treatment with imipramine [99]. Desipramine-induced weight gain was reduced by cotreatment with SR141716A, suggesting an eCB pathway [100].
Treatment with the MAOI tranylcypromine increased CB1 binding density in the prefrontal cortex and hippocampus, and increased 2-AG but decreased AEA levels in the prefrontal cortex [93]. Repeated electroconvulsive shock treatment (EST) for depression produced complex and regionally specific effects. Generally EST downregulated CB1 binding density and AEA levels in the cortex, but enhanced cannabinoid-stimulated [35S]GTPγS binding in the amygdala [101].
In summary, the effects of antidepressant drugs or treatments upon the eCB system are not definitive, but likely result in CB1 upregulation, at least in some brain regions. Preclinical studies suggest agonist trafficking may be responsible for variable responses.

Antipsychotic drugs
First-generation antipsychotic drugs, such as haloperidol and chlorpromazine (thorazine), are dopamine D2 receptor inverse agonist. Second-generation “atypical” antipsychotics (e.g., risperidone, olanzapine, clozapine, and aripiprazole) antagonize D2 and 5-HT2A, and also target other neuroreceptors. Acute administration of chlorpromazine enhanced the hypothermic response to THC [102]. Subchronic administration of haloperidol increased CB1 density in rat brains, indicated by increased binding of [3H]CP55,940 in the substantia nigra>globus pallidus>striatum. Subchronic haloperidol also potentiated CP55,940-stimulated [35S]GTPγS binding in the substantia nigra [103]. Sundram et al. [104] confirmed haloperidol's effects on [3H]CP55,940 binding, and obtained similar results with chlorpromazine and olanzapine. In monkeys trained to discriminate THC, haloperidol sensitized the THC discriminative stimulus [105]. Risperidone increased [3H]CP55,940 binding in rat brain without altering CB1 mRNA levels [106]. Four weeks of aripiprazole upregulated CB1 in rat frontal cortex [107]. Clozapine decreased [3H]CP55,940 binding in rat brain [104], and attenuated THC-induced disruption of spatial working memory in the rat radial maze task [108].
Several researchers have proposed that CB1 upregulation during antipsychotic drug treatment may explain appetite enhancement, weight gain, and CB1 supersensitivity. D'Souza et al. [109] conducted a double-blind study on the effects of adding haloperidol to THC. Compared to THC alone, the combination of drugs significantly worsened verbal recall, distractibility, and vigilance scores. The drug combination did not affect other testing parameters, such as euphoric effects and motor outcomes. Another double-blind study showed that haloperidol reversed THC-induced increases in the Positive and Negative Syndrome Scale (used for measuring symptom severity in schizophrenia), but did not affect the THC-induced “high” in healthy male volunteers [110]. A double-blind study in healthy male volunteers showed that olanzapine reduced the effects of THC as measured on the positive and negative syndrome scale, and the visual analogue scale for psychedelic effects, but the reduction fell short of statistical significance, p = 0.67 [111].
In summary, antipsychotic drugs likely upregulate CB1 expression in parts of the rodent brain. In human clinical studies, antipsychotic drugs do not affect THC-induced “high” or “euphoria,” but dampen dysphoria and worsen verbal recall and distractibility.

Anxiolytics, sedatives, and anesthetics
Diazepam is used for treating anxiety, insomnia, muscle spasms, and seizure disorders. Combining diazepam with WIN55212-2 produced a supra-additive anxiolytic effect in the rat elevated plus maze test; combining diazepam with the FAAH inhibitor URB597 also led to a supra-additive effect; coadministration of diazepam with the CB1 antagonist AM251 attenuated diazepam's anxiolytic effect [112]. These findings might be explained by the observation that both chronic and, particularly, acute administration of diazepam to mice is accompanied by strong elevation of brain eCB levels [113].
The anxiolytic and sedative effects of alprazolam were also attenuated by AM251 in mouse behavioral assays (light-dark box test, neurological severity score, and step-down inhibitory avoidance test) [114]. Surprisingly, however, the administration of alprazolam reduced WIN55212-2-stimulated [35S]GTPγS binding in mouse amygdala and hippocampus [114]. CB1 −/− knockout mice showed impaired anxiolytic responses to both buspirone and bromazepam in light/dark box, elevated plus maze, and social interaction tests [115]. N-arachidonoyl-serotonin (AA-5-HT), a dual FAAH/TRPV1 blocker, imparted anxiolytic effects in the mouse elevated plus maze assay [116].
A sub-effective dose of THC given to mice caused catelpsy in the horizontal bar test after sub-effective doses of either flurazepam or baclofen were added [117]. The beta-adrenergic blocking agent propranolol causes mild sedation, but pretreatment with propranolol blocked cannabis-induced cardiovascular effects and learning impairment in a small clinical trial [118].
General anesthesia (midazolam, sufentanil, isoflurane, and sufentanil) resulted in decreased serum AEA in patients stressed by the anticipation of cardiac surgery [119]. The dissociative anesthetic phencyclidine (PCP) impairs rotarod performance and open-field behavior in rats, effects shared by THC; combining the two caused supra-additive results [120]. Low-grade Mexican marijuana was adultered with PCP and marketed as “superweed” in the 1970s [121]. Nitrious oxide and THC both increase pain threshold in the tail-flick and hot-plate test, and their combination caused supra-additive effects [122].

Anticonvulsants
Combining diazepam with WIN55212-2 produced a supra-additive anticonvulsant effect in rats; combining diazepam with the FAAH inhibitor URB597 also led to a synergistic effect; coadministration of diazepam with the CB1 receptor antagonist AM251 attenuated the anticonvulsant effect of diazepam [123]. Chronic administration of valproate in rats increased CB1 binding of the PET scan tracer [18F]MK-9470; this was not seen with levetiracetam [124]. Tiagabine, an anticonvulsant GABA reuptake inhibitor, augmented THC-induced catalepsy [117] but not antinociception or hypothermia [125]. In a human study, tiagabine augmented THC discrimination and enhanced THC effects in other outcomes [126].
Pregabalin is a Ca2+ channel antagonist used for treating epilepsy and neuropathic pain. Isobolographic analysis demonstrated that combining WIN 55,212-2 with pregabalin exerted synergistic antinociceptive effects in the mouse hot-plate test [127]. Vagus nerve stimulation (VNS) is used as an add-on treatment to patients with drug-resistant epilepsy. Implantation of a vagus nerve stimulator in rats significantly decreased AEA and 2-AG in mesenteric adipose tissue, but increased PEA [128]. Chemical VNS by administration of the peptide hormone cholecystokinin 8 to fasted rats decreased expression of CB1 in vagal afferent neurons [129].