Association of a Variant of CNR1 Gene Encoding Cannabinoid Receptor 1 With Gilles de la Tourette Syndrome

Background Gilles de la Tourette syndrome (GTS) is a neuropsychiatric disorder of unknown etiology, although a major role of genetic factors has been established. Cannabis-based medicines may alleviate GTS-associated tics and variants of CNR1 gene encoding central cannabinoid receptor (CB1) are believed to be a risk factor for the development of some neurodevelopmental diseases. Our aim was to test the association of selected CNR1 gene variants with GTS. Material and Methods The cohort of GTS cases comprised 262 unrelated patients aged 3–53 years (mean age: 18.3 ± 9.1 years; 204 males (77.9%), 126 (48.1%) adults defined as ≥18 years). As a control group we enrolled 279 unrelated, ethnically and gender matched individuals with no diagnosed mental, neurological or general disorder, aged 13–54 years (mean age: 22.5 ± 3.0 years; 200 males, (74.1%). Both study and control groups were selected from Polish population, which is ethnically homogenous subgroup of Caucasian population. Four single nucleotide polymorphisms (SNPs) in CNR1 were selected: rs2023239, rs2180619, rs806379, and rs1049353 based on minor allele frequency in general population >15%. These variants were genotyped using a real-time quantitative polymerase chain reaction system (TaqMan SNP genotyping assay). Results We found significant association of GTS clinical phenotype with rs2023239 variant. Minor allele C and CT+CC genotypes were found significantly more often in GTS patients compared to controls (17.4 vs 11.1%, p=0.003 and 32.8 vs 20.4%, p=0.001, respectively), and the difference remained significant after correction for multiple testing. C allele of rs2023239 polymorphism of the CNR1 gene was associated with the occurrence of tics. There were no statistically significant associations for rs806379, rs1049353 or rs2180619 variants. Conclusion Our findings suggest that C allele of rs2023239 polymorphism of the CNR1 gene is a risk factor of GTS in Polish population. The variant can be potentially associated with abnormal endocannabinoid transmission, which is suspected to be one of the causes of GTS.

The two main cannabinoid receptors are CB1 and CB2, cannabinoid receptors type 1 and 2. CB1 receptor is mostly present in the central nervous system, with a high density in basal ganglia (Pertwee, 2006), the brain region implicated in movement control, while CB2 receptor is mainly present in the immune system and in hematopoietic cells (Russo, 2013) where it modulates cytokine release (Suárez-Pinilla et al., 2015). Variants of the CNR1 gene encoding CB1 receptor have been associated with various psychiatric and neurological diseases (Gadzicki et al., 1999;Gadzicki et al., 2004;Domschke et al., 2008;Evans et al., 2016;Ruiz-Contreras et al., 2017;Yao et al., 2018) (see Table 1). Gadzicki et al. (1999) were the first to investigate polymorphisms in a coding exon of the CNR1 gene finding the 1359(G/A) polymorphism in German GTS patients. Posteriorly they focused on CNR1 variants in a larger group of GTS patients (Janik et al., 2014). However, they failed to find a correlation between those variants and GTS.
therefore tested four such variants as possible risk factors of GTS in Polish GTS patients.

Ethics Statement
The study has been approved by the Ethics Committee of the Medical University of Warsaw (KB/2/2007, KB/53/A/2010, KB/ 63/A/2018) and has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. All participants aged ≥18 years signed an informed consent form prior to their inclusion in the study, and legal guardians gave a written consent on behalf of individuals under 18.

Study Participants
The study group comprised 262 unrelated GTS patients aged 3-53 years (mean age: 18.3 ± 9.1 years; 204 males (77.9%), 126 (48.1%) adults defined as ≥18 years). The family history of tics was positive in 142 (54.8%) patients, and was unknown for three patient. The family history of OCD or obsessive-compulsive behavior (OCB) was available from 125 patients, and was positive for 37 (29.6%) patients. The mean age of tic onset was 6.9 ± 3.0 years. Two hundred and three (77.8%) patients had at least one of the following co-morbidities: OCD/OCB n = 127 (48.7%), ADHD n=90 (34.5%), non-OCD anxiety disorder n = 84 (32.2%), learning disorder n=64 (24.5%), depression n = 37 (14.2%), conduct disorder n=16 (6.1%), while 58 (22.2%) had none of the above. Yale Global Tic Severity Scale (YGTSS) was used to assess the severity of tics, but could only be performed for 127 (48.5%) patients. The control group comprised 279 unrelated, ethnically and gender matched individuals with no diagnosed mental, neurological or general disorders, aged 13-54 years (mean age: 22.5 ± 3.0 years; 200 males, (74.1%). The age of the controls was slightly higher (mean age of the two groups differed only by four years) as we wanted to be sure that the controls have passed the age of tic onset (the onset of GTS after 13 years is very rare). The patients were evaluated for the clinical diagnosis of GTS and co-morbid mental disorders according to DSM-IV-TR, most of them before the DSM-5 criteria were published. Patients who fulfilled the DSM-5, but not DSM-IV-TR diagnostic criteria for GTS were excluded to ensure homogeneity of group. OCB was diagnosed if obsessions and compulsions were egosyntonic in contrast to egodystonic symptoms which characterized OCD. The diagnosis of co-morbid mental disorders was also made based on earlier psychiatric examinations that had been performed before the time of patients' evaluation. This included psychiatric disorders that were usually diagnosed in the childhood (e.g., attention deficit hyperactivity disorder or oppositional defiant disorder) and symptoms of which were not yet present in adult patients at the time of examination. All the patients were referred to a neurologist experienced in tic disorders and were personally interviewed by an author of the study (PJ). The study was designed as a one-time registration study and no new clinical data obtained on follow-up visits have been included in the analysis.

Genetic Analysis and Selection of SNPs
DNA samples were collected between 2007 and 2018. Genomic DNA was extracted either from peripheral blood leukocytes using a standard salting-out procedure (Miller et al., 1988) or from buccal cells collected with Oragene OG-500 DNA collection kit and using Prep IT L2P purification kit (DNA Genotek Inc., Ottawa, Ontario, Canada). The latter method was used with young patients that were unable or unwilling to comply with blood collections. DNA samples obtained with both methods were of same quality and applicability. Genotyping of selected SNPs was performed using TaqMan SNP genotyping assays (Life Technologies, Carlsbad, California, USA) on a StepOne Plus Real-Time PCR system (Life Technologies) (Janik et al., 2014).
The rs2023239, rs2180619, rs806379, and rs1049353 CNR1 polymorphisms were chosen on the basis of their minor allele frequency (MAF) estimated above 15% in both general and Polish populations, and their well-documented association with various neurological and psychiatric disorders ( Table 1). As indicated by data from the Exome Aggregation Consortium (ExAC) and the Genome Aggregation Database (GnomAD) rs1049353 is the only known polymorphism in the coding sequence of the CNR1 gene with a MAF above 1%; it is synonymous. All other published variants in the coding sequence are private withfrequency of at most hundreds per 276,000 alleles (MAF below 0.4%) and therefore were not selected for the study. Also in our in-house database of whole exome/genome sequencing results for >200 Polish patients not affected by GTS, the rs1049353 variant is the only one located in the coding sequence of CNR1. Thus, except for rs1049353, the selected SNPs are located in non-coding sequences. For clarity we note that rs1049353 is different from the 1359(G/A) polymorphism studied by Gadzicki et al. (Mackie, 2006).

Statistical Analysis
Chi-square test was used for both allelic and genotypic association studies. Association of genotypes with age of tic onset and YGTSS was analyzed with Kruskal-Wallis test. Multivariate logistic regression model was used to find independent predictors of GTS risk. The study sample size was sufficient to detect with 80% probability the true effect size measured as odds ratio (OR) between 1.62 and 1.97 for positive association or between 0.39 and 0.61 for negative association (depending on the actual MAF which ranged from 0.11 to 0.47) for the differences in allele frequencies of the four polymorphisms between the GTS and the control group. Statistica 13 program was used for statistical calculations. The significance level was set at p < 0.05, and Bonferroni-corrected significance criterion was used where indicated in the text.

RESULTS
The genotyping success rate was 100% and the consensus rate (on the basis of 10% duplicates) was 100% for DNA isolated from both leukocytes and buccal cells. The quality of genotyping was the same regardless of the biological source and method used for DNA isolation. The genotype frequencies of all four SNPs were in accordance with the Hardy-Weinberg equilibrium in the control group (p>0.9 for rs806379, p>0.3 for rs2023239, p>0.6 for rs2180619 and p>0.6 for rs1049353). In the patient group the genotype frequencies of three SNPs were in accordance with the Hardy-Weinberg equilibrium (p>0.2 for rs2023239, p>0.1 for rs2180619 and p>0.2 for rs1049353) while the distribution of the rs806379 genotype deviated slightly from the Hardy-Weinberg equilibrium (p=0.047) due to a deficit of heterozygotes.
A significant association with the GTS was found for the rs2023239 variant. The MAF of CNR1 rs2023239 was significantly higher in GTS patients compared to control group. When the TT genotype was used as a reference, genotypes CC+CT (dominant model, p = 0.001) and CT alone (p = 0.0009) were significantly associated with a higher risk of GTS. This association remained significant after Bonferroni correction for multiple tests (Bonferroni-corrected significance level for four SNPs and six genetic models is 0.05/(4*6)=0.002). There were no statistically significant associations between the rs2180619, rs806379 rs1049353 variants and the GTS ( Table 2). Multivariate logistic regression analysis adjusted for gender and age showed that the presence of CNR1 rs2023239 allele C (genotypes CC+CT, dominant model) is an independent factor associated with a higher risk of GTS (Table 3).
Genotypic and allelic association analysis of all four CNR1 polymorphisms with most common co-morbid disorders in patients with GTS are shown in Table 4. The only significant association was found for rs806379. The minor allele T of this SNP was more frequent in GTS patients with ADHD than in the patients without ADHD (OR 1.45, 95% CI: 1.01-2.08, p=0.045). No associations were found between the CNR1 genotype and age of tic onset (p>0.4), total YGTSS or YGTSS Total Tic Score (p>0.15), family history of tics (p>0.15) or family history of OCB/OCD (p>0.2).

DISCUSSION
So far the only comprehensive study investigating genetic variation in the CNR1 gene in GTS (Janik et al., 2014) was aimed to localize pathogenic mutations in CNR1 and failed to   show differences with healthy controls. Our study is the first to examine common SNP variants of CNR1 gene in a group of GTS patients. A significant association was found both in the allelic and genotypic analysis between the risk of GTS and the rs2023239 polymorphism. Multivariate analysis proved that CC+CT genotypes are gender-independent factors associated with a higher GTS risk. Our study shows that GTS is associated with rs2023239 polymorphism along with various others neuropsychiatric disorders, such as the cannabis dependence (Benzinou et al., 2008), eating disorders (Yu et al., 2013), schizophrenia (Wiskerke et al., 2012), impulsivity (Juhasz et al., 2009), depression (Ruiz-Contreras et al., 2017, and migraine (Ketcherside et al., 2017). We can only speculate on the molecular mechanism underlying the role of this variant in the pathogenesis of GTS. rs2023239 allele C was found to be associated with increased CB1 receptor density (Janik et al., 2015) which could modulate neurotransmission. It is unlikely that this intronic variant directly affects the splicing, as it is too distant the adjacent exon (122bp). Human Splicing Finder predicts that the minor allele C does not create any new potential splice sites or new potential branch points compared to the common T allele. Additionally, this variant could affect gene expression either directly or through a linkage to functional variant(s) located in regulatory regions.
Our earlier association studies in Polish patients with GTS have revealed a relationship between some SNPs of BTBD9, ADORA1 and ADORA2A genes and co-morbid psychiatric disorders (Muller-Vahl et al., 1999;Martiny, 2017). Therefore, we hypothesized that CNR1 variants could also contribute to the disease complexity predisposing the patients to psychiatric comorbidities as all the CNR1 polymorphisms studied here had been found to be associated with various psychiatric disorders (Table 1). However, both genotypic and allelic association analyses of all the CNR1variants examined, including rs2023239 associated with a higher risk of GTS, failed to demonstrate a significant association with any psychiatric comorbidity. The only statistically significant association found, that of co-morbid ADHD with rs806379, should probably be regarded as a false positive because of its marginal significance (0.05>p>0.04) in multiple statistical tests performed (4 SNPs x 4 disorders x 6 models).
A number of open as well as randomized controlled studies and case reports support the premise that the use of CBM improves not only tics, but also ADHD (Abi-Jaoude et al., 2017), OCB/OCD (Hasan et al., 2010;Denys et al., 2013), depression (Yoon et al., 2007) and anxiety . Our study does not confirm that this beneficial effect of CMB on co-morbid disorders in GTS could be related to CNR1 SNPs as we did not find any association of the CNR1 variants with co-morbidity. Additionally, we did not find any correlation of CNR1 variants with any clinical data, such as age of tic onset, tic severity measured with YGTSS or familial history of tics and OCD/OCB. It can therefore be speculated that while rs2023239 contributes to the pathogenesis of tics per se, other genetic and environmental factors contribute to further clinical features of GTS. The present results on the CNR1 variants could also be interpreted in relation to an earlier SNP study by our group, in which we found association of ADORA1 and ADORA2A variants with a higher risk of GTS (Muller-Vahl et al., 1999). In this context we note that abnormality of the dopaminergic system has been confirmed to be one of the primary causes of GTS (Singer et al., 1982;Sandyk and Bamford, 1987;Ferré et al., 2009;Ferré et al., 2010;Filbey et al., 2010), both adenosine and endocannabinoids act as modulators of dopamine neurotransmission in the striatum (Müller et al., 2007) and it has been suggested that CB1 receptors form heteromers with dopamine D2 and adenosine A2A receptors (Ehlers et al., 2007). Although we do not know the functional significance of the minor variant of rs2023239 in GTS, we put forward the hypothesis that it could be related to faulty cannabinoid transmission due to lower expression of the CB1 receptor or reduced receptor's affinity for the endocannabinoid ligand. This, in turn, could lead to reduction of adenosinergic signaling and over-activity of dopaminergic transmission, and finally aggravate tics. Another possible mechanism is a direct enhancement of dopaminergic transmission as a consequence of limited endocannabinoid activity, without an involvement of adenosine signaling.
In conclusion, the obtained results indicate that the C allele of the rs2023239 polymorphism of CNR1 gene is a risk factor of GTS in the Polish population, associated with the occurrence of tics, but not with the co-existing psychiatric symptoms. The results also support the hypothesis of an ECS involvement in the pathogenesis of GTS. Nevertheless, our finding is preliminary and needs to be replicated in an independent cohort. Further research is needed to determine functional significance of the variant.

LIMITATIONS
The following limitations of the study have to be addressed: 1) the analyzed variants can be in linkage disequilibrium with true risk conferring variants located outside analyzed regions; 2) the study group was relatively small; 3) the patients were evaluated for co-morbid disorders only once (one-time registration) and it cannot be excluded that psychiatric disorders could develop over time and additionally it was impossible to measure the YGTSS in all patients; there is a possibility of false negative results in rs2180619, rs806379, and rs1049353 CNR1 polymorphisms; 4) some clinical data, especially on adult patients, were subject to a recall bias, ; 5) we have not included the comparison with the variants previously associated with GTS and this should be taken into consideration in the future studies; and 6) our finding needs to be replicated using study groups from different populations.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher. These are also included in Supplementary Material.

ETHICS STATEMENT
The study has been approved by the Ethics Committee of the Medical University of Warsaw (KB/2/2007, KB/53/A/2010, KB/ 63/A/2018) and has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. All participants aged ≥18 years signed an informed consent form prior to their inclusion in the study, and legal guardians gave a written consent on behalf of individuals under 18.