Fluorinated Cannabidiol Derivatives: Enhancement of Activity in Mice Models Predictive of Anxiolytic, Antidepressant and Antipsychotic Effects

Cannabidiol (CBD) is a major Cannabis sativa constituent, which does not cause the typical marijuana psychoactivity. However, it has been shown to be active in a numerous pharmacological assays, including mice tests for anxiety, obsessive-compulsive disorder, depression and schizophrenia. In human trials the doses of CBD needed to achieve effects in anxiety and schizophrenia are high. We report now the synthesis of 3 fluorinated CBD derivatives, one of which, 4'-F-CBD (HUF-101) (1), is considerably more potent than CBD in behavioral assays in mice predictive of anxiolytic, antidepressant, antipsychotic and anti-compulsive activity. Similar to CBD, the anti-compulsive effects of HUF-101 depend on cannabinoid receptors.


Introduction
Cannabidiol (CBD) is a major cannabinoid present in Cannabis sativa, which does not cause the typical effects of the psychoactive component, Δ 9 -tetrahydrocannabinol (THC). CBD was isolated from marijuana in 1940 by Adams et al [1] in the US and from Egyptian hashish by Jacob and Todd [2] in the UK. Its structure was elucidated in 1963 [3] and its absolute configuration was established in 1967 [4].
Numerous preclinical studies indicate that CBD exerts therapeutic effects in animal models of a wide range of health disorders, including neuropsychiatric conditions. Several mechanisms have been suggested to be involved in the actions of CBD: activation of TRPV1 channels, inhibition of uptake and metabolism of the endocannabinoid anandamide, inhibition of adenosine refluxed over sodium and phosphorous pentoxide, respectively, and freshly distilled prior to use. 1 H-NMR spectra were obtained using a Bruker AMX 300 MHz apparatus using the deuterated chloroform (CDCl 3 , δ = 7.25 ppm) with tetramethylsilane (TMS) as internal standard. Optical rotations were measured on II polarimeter in a 2.00 dm cell and 25°C. Elemental analysis was performed using a Perkin-Elmer (Boston, MA, USA) 2400 series II analyzer at the Hebrew University Microanalysis Laboratory. Thin-layer chromatography (TLC) was run on silica gel 60F 254 plates (Merck). Column chromatography was performed on silica gel 60 Å (Merck). Compounds were located using a UV lamp at 254 nm. 2.1.1 Synthesis of 4'-fluoro-cannabidiol, HUF-101 (1) Fig 1. CBD was isolated from hashish following the procedure described by Gaoni and Mechoulam [32].
To a stirred solution of 8 (500 mg, 1.506 mmol) in CH 2 Cl 2 (12 ml) at -78°C was added DAST (0.238 ml, 1.807 mmol, 1.2 equiv) at -78°C, stirred for 1 hr at the same temperature under N 2 atmosphere, and the resulting mixture was slowly warmed to room temperature and stirred for 4 hrs monitoring by TLC. Solid Na 2 CO 3 (160mg, 1.506 mmol) was added, the organic phase was washed with 1M aqueous Na 2 CO 3. The solution was extracted with CH 2 Cl 2 . The combined extracts were washed with water and dried over MgSO 4 . After evaporation of the solvent under reduced pressure, the resulting residue was chromatographed using silica gel (1% ether-petroleum ether) to give 3 as light yellow oil (250 mg, 50%). 1

Biological Assays
Male Swiss mice (30-45g) originated from the Central Animal Farm of the School of Medicine of Ribeirão Preto (FMRP-USP) were maintained in groups of five animals per box (41x33x17 cm) in a temperature controlled room (24±2°C) with a 12x12 h light-dark cycle. They received CBD (THC Pharm, 15-60 mg/kg), fluoxetine (Pharmaceutica, Brazil, 10 mg/kg), AM251 (Tocris, USA, 1 mg/kg) [34,35], AM630 (Tocris, USA, 1 mg/kg) [35], 1 (1-10 mg/kg), 2 (1-60 mg/kg), 3 (1-30 mg/kg) and 5a (1-30 mg/kg) were administered intraperitoneally (i.p.). All drugs but AM251 and AM630 were dissolved in 2% Tween 80 in sterile saline. The other two drugs were dissolved in 10% DMSO in sterile saline [34,35].  in the open arms and the number of entries in the enclosed arms. Animals were considered to enter an open or enclosed arm when 90% of their bodies were inside the region. All experiments were performed in the morning period (5 to 12 a.m.). 2.2.1.2 Forced swimming test (FST): Immediately after the EPM test the animals were submitted to 6 min of forced swimming in glass cylinders (height 25 cm, diameter 17 cm) containing 10 cm of water at 23-25°C. Immobility time (characterized by slow movements necessary to avoid drowning) was measured during the last 4-min period. The water was changed after each trial to prevent the influence of alarm substances [24]. During the tests no animal drowned or struggled to keep their heads above water. Therefore, there was no need for intervention by the experimenter. 2.2.1.3 Prepulse inhibition (PPI): The PPI experiment was performed in independent groups of animals. The test was conducted simultaneously in two identical startle response systems (Med Associates, USA). A continuous acoustic signal provided a background white noise level of 65 dB. The pulse consists of a 105 dB white noise burst with a rise/decay of 5 ms and duration of 20 ms. The prepulse comprised pure 7000 Hz tones, 10 ms duration, with intensities set at 80, 85, and 90 dB. The setups were daily calibrated to ensure equal sensitivity throughout the experiments. Calibration was performed by adjusting the gain on the load cell amplifier to 150 arbitrary units (AU) at a standard weight appropriated for 40 g mice. The limits of the load cell were −2047 to +2047 AU. Thirty min after the injection of the tested compounds mice received an i.p. injection of amphetamine 10 mg/kg or vehicle. Animals were submitted to the PPI test 20 min after amphetamine or vehicle injection. After a 5 min acclimatization period in which the animal did not listen to any stimuli except the 65 dB background noise, mice were presented with a series of 10 stimuli (pulse alone). The first 10 pulse-alone trials allow for the within-session habituation to the startle stimulus and are not considered for PPI statistical analysis. The test consisted of 64 pseudo-random trials divided into eight different groups presented with an inter-stimulus interval of 30 s, and consisting of pulse alone (105 dB), prepulse alone (80, 85, or 90 dB), prepulse + pulse with 100 ms interval between prepulse and pulse, and no stimulus presented [18,19]. Prepulse stimulus did not elicit an acoustic startle response. Mean acoustic startle response to pulse-alone (P) trials and each prepulse + pulse (PP + P) trial was recorded for each subject. PPI was calculated by expressing the prepulse + pulse startle amplitude as a percentage of decrease from pulse-alone startle amplitude, according to the following formula: %PPI = 100-[100 × (PP + P/P)]. This transformation reduces statistical variability attributable to differences between animals and it is a direct PPI measure [18,19]. 2.2.1.4 Marble burying test (MBT): Independent groups of animals were submitted to the MBT. For this test twenty-five green clear glass marbles used were evenly spaced over the 5 cm sawdust layer-covered floor of a squared box (38 x 32 x 28 cm). Thirty minutes before the test the animals were pre-exposed for 5 min to the box. They are then placed in the center of marble-containing box. Thirty minutes later the number of buried marbles was recorded following the criteria for buried marbles proposed by Njung'e and Handley [37], namely that at least two-thirds were under sawdust.
2.2.2 Statistical Analysis. Results were analyzed by one-way or two-way ANOVAs. The Duncan test was used for posthoc analysis. Significant level was set at p<0.05.
The synthesis of the allylic fluoro derivative 2 was done in 2 steps (Fig 2) First CBD diacetate was oxidized with selenium dioxide and t-butyl peroxide to the known 10-hydroxy-CBD diacetate (4) [39]. Then the allylic hydroxyl group was replaced with fluorine with the nucleophylic fluorinating reagent DAST (diethyl amino sulfur trifluoride) [30], which is widely used for the direct transformation of aliphatic hydroxyl groups to C-F groups, to yield 2. The exact procedure followed the experimental details reported by Boukerb et al [40].

Biological Assays
A summary of our results can be seen in Table 1.
3.2.1 Elevated plus maze (EPM) and forced swimming (FST) tests. The EPM is currently the most widely used animal model of anxiety. Initially proposed by Handley and Mithani [41], it is based on the conflict generated in rodents by their natural tendency to explore novel environments versus the innate fear of bright and elevated places [42]. Factorial analysis indicates that the number of entries into the enclosed arm can be used to measure locomotor activity, while the percentage of entries and time spent in the open arms reflects the anxiety level of the animal [36]. Typically, CBD and other cannabinoids induce bell-shaped dose-response curves in this model, decreasing anxiety at small doses and being ineffective (or sometimes anxiogenic) at higher doses [43,44].
The FST is also the most extensively used animal model to predict antidepressant activity [45]. In this test rodents are exposed to force swimming in a cylinder filled with water. After an initial period of active escape behavior, the animal become immobile. Antidepressant-like drugs typically decrease the immobility time [45,46]. An initial study from our group showed that systemic administration of CBD induces antidepressant-like effects in mice submitted to the FST without changing locomotor behavior [24]. The antidepressant-like effects of CBD have been recently confirmed [47]. The authors showed, using another animal model, that a    (Fig 5).
HU-577 (5a): The starting material for the synthesis of 1 and of 2 was CBD. However, the starting material for the synthesis of 3 was dihydro-CBD (5a). Hence we also evaluated its  activity. At the dose of 3 mg/kg 5a increased the percentage of entries and time spent in the open arms entries (F 3,29 = 5.42, p = 0.004 and F 3,29 = 6.82, p = 0.001 respectively, Duncan, p<0.05). No change was observed in the number of enclosed arms. Although there was a trend, the drug also failed to change immobility time in the FST (F 3,29 = 2.5, p = 0.079) (Fig 6).

PPI (prepulse inhibition) test.
PPI is a phenomenon that reflects the response attenuation induced by a weaker, non-startling stimulus (prepulse) that precedes a loud, startling acoustic stimulus (pulse) [48]. Schizophrenia patients often present a disruption of this normal inhibitory process [48]. Since psychotomimetic drugs such as amphetamine or glutamate NMDA receptor antagonists also disrupt this response in rodents and normal subjects, this model has been used to investigate the sensorimotor gating impairment found in these patients [49]. Acute or repeated CBD treatment attenuates the PPI impairment induced by dopamine agonists or NMDA receptor antagonists in rodents [18,19,50].

Discussion
The EPM, FST, PPI and MBT are tests employed to unveil anxiolytic, antidepressant, antipsychotic and anticompulsive drug properties, respectively. All these properties have been shown to be associated with the CBD molecule (see above). In the present study we show that the fluorination of CBD leading to 1 enhances CBD potency in all these animal assays (Figs 4,7,11,12 and Table 1). Compound 5a was only effective in the EPM test (although there was a trend for an effect in the FST) (Fig 6). Compound 3 induced an effect, in the models for anxiety and depression, though with lower potency. However, the drug failed to attenuate PPI impairment induced by amphetamine, suggesting that it lacks antipsychotic-like effects (Fig 5). Compound 2 failed to cause any effect in the behavioral tests.
Although the knowledge of the molecular mechanisms responsible for CBD effects in the animal models used by us would have been an ideal starting point for the development of more potent compounds, our present poor understanding of these mechanisms prevents this approach. As mentioned in the introduction, numerous mechanisms have been suggested to be involved in the different actions of CBD [5][6][7][8][9][10]. Considering that these actions (e.g., anxiolytic,  antidepressant, anticompulsive and antipsychotic) cannot clearly be explained by a single mechanism, to test all possibilities would be out of the scope of this paper. However, since we have previously found that CBD effects in the MBT, a single and reproductive test used to detect anticompulsive-like drugs effects, is prevented by pretreatment with a CB1 antagonist [25], we decided to test if a similar mechanism was involved in HUF-101 effects. This was, indeed, the case. Moreover, we expanded our previous results showing that a CB2 receptor antagonist also blocks CBD and HUF-101 effects in the MBT. CBD effects are probably indirect, since it has a low affinity for these receptors and is unable to produce the characteristic tetrad observed with high doses of CB1 agonists [8,16]. It can, however, inhibit the FAAH enzyme, responsible for anandamide metabolism [8]. This indirect activation of the endocannabinoid system has been proposed to explain several (but not all) effects of CBD, including facilitation of adult hippocampal neurogenesis [22], impairment of aversive memories [52] and anticompulsive [25].
Presumably the actions of the fluorinated CBD derivatives, particularly those of 1, which parallel those of CBD, are based on the same mechanisms. The present findings in the MBT assay corroborate this proposition. However, this assumption clearly needs to be further investigated. Comparative pharmacokinetics studies are also required to determine if the greater potency of HUF-101 depends solely on pharmacodynamics changes induced by fluorination of the CBD molecule.

Conclusion
We describe the synthesis of 3 fluorinated CBD derivatives 1, 2 and 3. Compound 1, prepared by fluorination on the aromatic ring of CBD, is considerably more potent than CBD in behavioral assays in mice predictive of anxiolytic, antidepressant, antipsychotic and anticompulsive Fig 12. CBD (30 mg/kg, n = 7-9 animals/group) and HU-101 (10 mg/kg, n = 9-10 animals/group) decreased the number of buried marbles in the MBT. This effect was prevented by pre-treatment with the CB1 (AM251 1 mg/kg, n = 9 animals/group) or CB2 (AM630, 1 mg/kg, n = 7 animals/group) receptor antagonist. *indicates difference from all other groups. +indicates difference from V+V group (ANOVA followed by the Duncan test, p<0.05).
doi:10.1371/journal.pone.0158779.g012 activity (Table 1). We also found that, similar to CBD, HUF-101 anticompulsive effects depend on CB1 and CB2 cannabinoid receptors. Fluorinated derivative 2, in which the fluorine atom is on the propylidene entity (C-10) was not active (in doses of 1-10 mg/kg) in any of these assays, while 3, in which the fluorine substitution is on the C-7 methyl group was less active than 1 in the anxiolytic and antidepressant assays and not active in the antipsychotic assay. CBD is already being evaluated as a therapeutic agent for these conditions, though at relatively high doses (see Introduction). In view of the higher potency of 1, compared to CBD, this new CBD derivative may possibly be further developed as a therapeutic entity.