Cannabis Constituents and Acetylcholinesterase Interaction: Molecular Docking, In Vitro Studies and Association with CNR1 rs806368 and ACHE rs17228602

The study documented here was aimed to find the molecular interactions of some of the cannabinoid constituents of cannabis with acetylcholinesterase (AChE). Molecular docking and LogP determination were performed to predict the AChE inhibitory effect and lipophilicity. AChE enzyme activity was measured in the blood of cannabis addicted human subjects. Further, genetic predisposition to cannabis addiction was investigated by association analysis of cannabinoid receptor 1 (CNR1) single nucleotide polymorphism (SNP) rs806368 and ACHE rs17228602 using restriction fragment length polymorphism (RFLP) method. All the understudied cannabis constituents showed promising binding affinities with AChE and are lipophilic in nature. The AChE activity was observed to be indifferent in cannabis addicted and non-addicted healthy controls. There was no significant association with CNR1 SNP rs806368 and ACHE rs17228602. The study concludes that in silico prediction for individual biomolecules of cannabis is different from in vivo physiological action in human subjects when all are present together. However, for a deeper mechanistic insight into these interactions and association, multi-population studies are suggested. Further studies to explore the inhibitory potential of different cannabis constituents for intended AChE inhibitor-based drug are warranted.


Introduction
Cannabis is commonly referred as marihuana, marijuana, hashish and hash and is obtained from plant Cannabis sativa L. The medicinal use of cannabis has been documented from the Middle East and Asia since sixth century B.C. [1]. However, presently, it is one of the leading drugs of substance-abuse globally. Its use or possession comes under the criminal act in most of the countries, though legalized for medical and recreational use in some states of USA and European countries. More than five hundred biologically active molecules have been identified [2] from cannabis and are, categorized as cannabinoids and non-cannabinoids. Tetrahydrocannabinol (THC) is considered the most abundant For docking analysis, 3D structure of AChE enzyme (PDB accession codes:4PQE) [27] was downloaded from RCSB Protein Data Bank (PDB) (http://www.rcsb.org). Three pdb entries belonging to human, mouse and electric ray were identified. Chimera [28] was used for alignment and superimposing the pdb structure of AChE of these three species; and the conserved residues were scanned. The PRALINE multiple sequence alignment toolbox was used to align the sequences of AChE from these three species. The binding sites residues are crucial in the ligand-receptor binding hence, utilized for structure-based drug designing [29]. The binding site residues for human AChE are given in Table 1. 3D structures of all the ligands i.e., tetrahydrocannabinol, cannabielsoin, cannabicyclol, cannabidiol, cannabigerol, cannabinol, cannabitriol, cannabivarin, paraoxon and donepezil were obtained using PubChem and Chemspider [30,31]. Molecular docking between residues of binding site of the receptor and ligands was performed by Autodock Vina PyRx version 0.8 [32] for Windows (available free at http://pyrx.sourceforge.net). Parameters used in Vina Search space for docking are given in Table 2. Briefly, first PyRx of the receptor was loaded into the program and then ligand file was loaded to perform molecular docking. The grid-size was established such that all possible binding site interactions pertaining to each ligand would be covered. Furthermore, the examination extended to ensure each ligand was in fact at its appropriate location with reference to the structure of the receptor. A total of nine runs were performed for each docking. The docking results were analyzed by comparing the binding interactions and binding energies between ligand and AChE receptor. Various ligand-receptor interactions like hydrogen bond, π-π interactions and Van der Waals forces were calculated.

LogP Determination
All logP values were calculated using Pallas 3812 of Prolog P (ComInnex, Budapest, Hungary). Pallas 3812 is a computer-assisted tool to be applied in drug research. It is an advanced version of MetabolExpert [33] that is used to calculate logP, metabolic pathway and several other characteristics of drugs and drug candidates.

Sampling of Study Subjects
Cannabis addicted subjects were enlisted from different rehabilitation centers (New Roshni Center, Wada Rehab Center, Psychaid Hospital) in Islamabad, Pakistan. The study was approved by ethics review board (ERB) of the Department of Biosciences, COMSATS University Islamabad (CIIT/Bio/ERB/19/98). The study conformed to tenets of 1964 Declaration of Helsinki and its later amendments. After filling the consent form of patient, venous blood was taken in EDTA-vacutainer tubes (Atlas-Labovac Italiano, FL Medical, Torreglia PD, Italy). Data about age and gender were obtained. Forty-nine confirmed cannabis addicted individuals with average age (Mean ± SD, 29 ± 9) were included in this study. Apart from inclusion criteria, the fulfilment of exclusion criteria too was ensured which included any viral diseases, chronic diseases like diabetes, use of combination of drugs and drug abuse for less than three months. Age-matched forty-five non-addicted individuals were enlisted as controls. Only non-hemolysed blood were used for AChE measurements.

Primer Designing and Chemicals
Primer 3 version 0.4.0 software (http://bioinfo.ut.ee/primer3-0.4.0/primer3/) was used for designing of forward and reverse primer. NCBI Blast software (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) was used to check the specificity of primer. Further testing was done by using In Silico PCR Tool (https://genome.ucsc.edu/cgi-bin/hgPcr). Primers were prepared by Macrogen (Rockville, MD, USA). Sequences of primers for both SNPs rs806368 and rs17228602 with their Tm and product sizes are given in Table 3. Chemicals for DNA extraction, PCR, RFLP analysis, Gel electrophoresis and AChE estimation were obtained from Thermo Fisher Scientific (Waltham, MA USA) and Sigma-Aldrich (St. Louis, MO, USA).

Genomic DNA Extraction and SNP Genotyping
By using salting out method [35], genomic DNA was extracted from whole blood samples. Genotyping of CNR1 (rs806368) and ACHE (rs17228602) was carried out by polymerase chain reaction and restriction fragment length polymorphism (PCR-RFLP) method. Forward and Reverse primer used in this method is shown in Table 2. Quantities of reagents used for PCR amplification are given in Table 4. In case of CNR1 (rs806368) the PCR products were incubated at 55 • C for 4 h with restriction enzyme BseGI (BtsCI) (Cat # ER0871, ThermoFisher Scientific) which cleaves in the presence of major allele T into fragments of 248 bp and 154 bp size whereas in presence of C allele the 401 bp fragment remain uncut. Then these restriction fragments were visualized on 2% agarose gel in horizontal electrophoresis (Cleaver Scientific, Rugby, UK; catalog number MSMINI10) as shown in Figure 1. In the case of ACHE, the incubation of PCR product was done for 16 h at 37 • C in the presence of restriction enzyme Psp5II (PpuMI) (Cat # ER0761, Thermofisher Scientific). Cleavage of Psp5II occur in presence of major allele C which produces fragments of 224 bp and 141 bp size although in presence of T allele the fragment of 365 bp remains uncut. The restriction products were visualized on 2% gel as shown in    Table 4. Quantity of reagents and procedure used for the genotyping of rs806368 and rs17228602.

Statistical Analysis
Genotypes and allelic frequencies between addicts and non-addicts were analyzed by Fisher exact test. To assess the association effect in different inheritance models, the Odds ratio with 95% Confidence Interval for both SNPs was calculated. Genotype frequencies were evaluated for deviations from Hardy-Weinberg Equilibrium (HWE) in addicted and non-addicted groups by means of goodness of fit Chi-square test (http://www.had2know.com/academics/hardy-weinberg-equillibriumcalculator-2alleles.html). GraphPad Prism 5 and online available software (http://vassarstats.net/fisher2x3.html) were utilized for calculations.

Docking Analysis
The binding interactions and binding energies for ligands docked against acetylcholinesterase are summarized in Table 5. The free binding energies of donepezil and paraoxon were −8.3 and −6.1 kcal/mol, respectively. Tetrahydrocannabinol (THC) had the lowest binding energy (−9.3 kcal/mol) compared to all other investigated ligands of cannabis. Table 4 lists various kinds of molecular interactions of cannabis ligands with acetylcholinesterase. Figure 3 shows complex which was formed by loading all the ligand, one by one, on receptor using UCSF chimera. It is clearly visible from Figure 3 that all the ligands have taken the same binding pocket as reported residues. 2D binding interactions of all the ligands with acetylcholinesterase receptor are shown in Figure 4.  When the results were compared with the pre-reported interactions of binding residues, it was seen that tyr 341 and phe 297 showing H-bonding between ligand-receptor were present in binding pocket of acetylcholinesterase. Likewise, π-π interactions residue trp 286, tyr124 and 341 were also observed.    Table 6 shows the logP values that is predicted log of the octanol/water partition coefficient of the understudied biomolecules of cannabis. All of them are highly lipophilic in comparison to donepezil, a drug used for the treatment of Alzheimer.  Table 7. AChE enzyme activity in addicted and non-addicted was not different (0.16 μmol/L/min).
The statistical analysis of genotype and allele frequencies for rs806368 are summarized in Table  8. No significant difference in genotype frequencies was found between cannabis addicts and nonaddicts (ϰ 2 = 2.872, p = 0.2379). No homozygote of minor allele C was observed both in cannabis addicts and non-addicts and so no recessive model was evaluated. There was no statistically significant association of rs806368 with risk of cannabis addiction when detected in dominant and Table 7. AChE enzyme activity in addicted and non-addicted was not different (0.16 µmol/L/min).
The statistical analysis of genotype and allele frequencies for rs806368 are summarized in Table 8. No significant difference in genotype frequencies was found between cannabis addicts and non-addicts (κ 2 = 2.872, p = 0.2379). No homozygote of minor allele C was observed both in cannabis addicts and non-addicts and so no recessive model was evaluated. There was no statistically significant association of rs806368 with risk of cannabis addiction when detected in dominant and allelic models (DM: OR = 0.7403, 95%CI = 0.3072-1.784, p = 0.502; Allele OR = 0.9384, 95%CI = 0.4663-1.888, p = 0.8585). Table 9 shows the results of allele and genotype frequencies of AChE rs17228602 in cannabis addicts and non-addicts. There was no statistical difference for genotype and allele frequencies of rs17228602 in non-addicts and cannabis addicts (κ 2 = 1.180, p = 0.277, Consequently, no significant statistical association was found with addiction vulnerability in any of the inheritance models in cannabis addicts and non-addicts ( Table 8). The genotype frequency of CT was lower in addicts than in non-addicts while no TT genotype was observed in both addicts and non-addicts ( Table 8). The allele frequency of major allele C was observed to be 86.73% and 81.63% in cannabis addicts and non-addicts, respectively while T minor allele was found to be lower in cannabis addicts as compared to non-addicts (13.27% and 18.37%, respectively).  Table 9. Association analysis of ACHE rs17228602 between cannabis addicts and non-addicts.

Discussion
Acetylcholinesterase has been investigated for association with mental disorders and psychosis since long [36]. AChE is reported to be depressed in depression [37], and some other neurological disorders [38] and is known to interact with many drugs of abuse. It can hydrolyze heroin, a well known substance abuse drug [39], though morphine inhibits AChE [40]. Similarly, the psychoactive effect of cannabis has been attributed to its primary psychoactive constituent, delta-9-tetrahydrocannabinol [19] which is reported as potent AChE inhibitor [41]. Acetylcholinesterase, a hydrolyzing enzyme of the cholinergic neurotransmitter ACh has been a target for cannabis or its constituents since decades [18,[42][43][44] but still there is paucity of research in human subjects. Some studies have investigated the cannabis effect on AChE and have reported conflicting results. For instance, Abdel-Salam and Khadrawy [18] reported a decrease in serum AChE activity in rats when subcutaneously treated with cannabis resin. Sadaf et al. and Javed et al. [25,45] reported elevated AChE activity in hashish users. Luthra et al. [46] found that prolong treatment (180 days) in rats caused slight elevation of AChE compared to moderate inhibition by treatment up to 90 days in the brain of male rats. Ghosh et al. [47] reported the surge in AChE in rat brains after acute administration of THC.
Meanwhile, cannabis has been long known for its medicinal applications but uncharacterized physiological interactions of its constituents in the human body and, social stigma associated with substance-abuse has constrained its use in wide-spread clinical settings. In addition, studies mostly been limited to THC or cannabidiol (CBD) constituents of cannabis. Thus, identification and characterization of different bioactive molecules of cannabis is required for its intended application in therapeutics of pathophysiology. Furthermore, genetic heterogeneity has come to be recognized as a source of variable drug response in recent years and may be considered as one of the aetiology of chemical action. In present study, apart from THC and CBD, seven other cannabinoids; Cannabinol (CBN), Cannabicyclol (CBC), Cannabitriol (CBT), Cannabielsoin (CBE), Cannabigerol (CBG), Cannabichromene (CBC), Cannabivarin (CBDV) were evaluated to find their AChE inhibitory capabilities and lipophilicity and compared with donepezil and paraoxon, known AChE inhibitors. AChE activity was measured in human cannabis addicted subjects. In addition, tentative association of cannabis consumption/addiction and cannabinoid receptor 1 gene (CNR1} SNP) rs806368 and AChE gene SNP rs17228602 was assessed and found no association. To the best of our knowledge, these are studied for the first time which may open new direction of studies and discussion in future. Our results show that not only THC and CBD are AChE inhibitors rather all other studied cannabinoids have potency to inhibit AChE in this order THC > CBN = CBDV > CBL > CBT > CBE > CBC > CBD > CBG. The compounds rank in following order for lipophilicity; CBG > CBC > CBT > CBD > CBE > THC > CBDV > CBN > CBL. THC has been reported to cause inhibition of AChE by molecular docking previously [41,48]. Some of them (THC, CBN, CBDV, CBL, CBT) were found to be either more or, almost similar potent inhibitors when compared with donepezil; first line drug for treatment of Alzheimer.
Though inhibitory action of AChE was not observed in vivo as reported by some earlier studies and in contrast to in silico molecular docking here in this work. This apparent disparity between in silico prediction and in vivo finding might be due to antagonizing behavior of other components of cannabis like cannabidiol [20]. This inhibitory, stimulatory or no effect may be based on the dose and duration of cannabis exposure in addition to the quantity of bioactive constituents in cannabis. Moreover, various interactions among different neurotransmitters in the body could also contribute to that. For instance, THC if acted through CB1 receptors, would cause the release of dopamine [49]. Increase in dopamine may decrease the AChE or increase the acetylcholine [50]. Apart from these interactions of various neurotransmitters, SNP polymorphisms in genes may also complicate differential physiology. However, no polymorphism in the studied group was found for chosen ACHE and CNR1 SNPs, though patterns of allelic shift were noted in rs806378 of CNR1 gene. The overall findings suggest that cholinergic and cannabinoid interaction is very complex and cannot be interpreted with cholinergic enzymes only. From the literature it is evident that some interactions do exists [18,19,50]. Though previous conventionally research has focused on two major components of cannabis i.e., THC and CBD as potential candidate for therapeutic applications but in silico prediction could unravel many others. However, in silico prediction could not be perhaps translated to in vivo human physiology because of various antagonist constituents of the cannabis. Not only the antagonistic attributes natural physiological conditions, and pharmacokinetics of the molecules like absorption, distribution, metabolic fate and elimination also significantly influence the efficacy which may vary in different experimental models. Therefore, in silico and in vitro findings even for an individual component of cannabis should be cautiously translated for human applications. Future work with participants from multiple ethnic backgrounds and with well-documented drug-usage patterns need to be carried out for more tangible understanding. Despite certain limitations, the study will open the discussion and further investigation for the possible therapeutic application of AChE inhibitors molecules of cannabis other than THC and cannabidiol. For instance, if we look at the data, CBT has better AChE inhibitory potency than CBD and more lipophilic than THC and CBD, but no physiological study could be traced for CBT. The literature search of cannabitriol on PubMed retrieved only four papers related with isolation and structure only.

Conclusions
Cannabis addiction was not found to alter the acetylcholinesterase in the blood of cannabis users, though in silico molecular binding of its constituents predicted the inhibitory actions. This apparent predicted and observed discrepancy in AChE could be a masking effect of different biomolecules when present together. Association analysis did not show cannabis addiction vulnerability with SNPs rs806368 and rs17228602. However, variation trends in the allelic frequencies were found in the allelic model of rs806368 with possible pharmacogenetics potential and need more exploration in this regard. Further investigation on CBT for AChE inhibitor based therapeutic application for neuronal disorder is suggested. A more robust conclusion can be drawn by further studies with increased number of addicted and non-addicted individuals and different combination of cannabis biomolecules with predicted potential to inhibit AChE.