Evaluation of the Expression Levels of Cannabinoid Type-1 (CB1) Receptor and Phosphorylated Extracellular Signal-Regulated Kinase (p-ERK) in Gliomas

Gliomas are the most frequent primary brain tumors and one of the most aggressive forms of cancer. Recently, numerous studies have focused on cannabinoids as a new cancer-therapeutic approach due to their antineoplastic effects through activation of the cannabinoid receptors, CB1 and CB2. The aim of this study was to investigate the expression levels of cannabinoid type-1 (CB1) receptors and phosphorylated extracellular signal-regulated kinase (p-ERK) in human glioma samples and evaluate their clinicopathologic signicance. We analyzed the expressions of CB1 receptors (CB1R) and p-ERK in 61 paran-embedded gliomas and 4 normal brain tissues using automated immunohistochemical assay. CB1R and p-ERK expressions were categorized into high versus low expression levels. Statistical analyses were performed to evaluate the association between CB1R and p-ERK expression levels and the clinicopathologic features and overall survival. association


Introduction
Gliomas or glial tumors represent the majority of primary brain tumors in adults accounting for nearly 70% of these tumors [1]. They are amongst the most lethal cancers with an increasing incidence rate over the years [2]. Despite advancements in cancer treatment, gliomas are still characterized by low survival rates compared to other tumors. This instigates a dire need for identifying novel target-directed therapies based on the increased knowledge of the molecular and cellular biology of gliomas [3].
From the hemp plant, Cannabis sativa, more than 400 chemical entities can be extracted, of which more than 60 of them are cannabinoid compounds. The effects of cannabis administration involve impairment of the motor coordination, impairment of short-term memory, hypothermia, catalepsy, analgesia, vasodilation, tachycardia, and bronchodilation among others [4]. The most abundant psychotomimetic plant-derived cannabinoid is Δ9-tetrahydrocannabinol (THC). Other cannabinoid compounds incorporate cannabidiol (CBD) and cannabinol (CBN) and have psychoactive properties similar to THC [5].
Cannabinoids have been recently utilized adjunctively in cancer treatment to mitigate the adverse effects of chemotherapy and improve quality of life. Numerous studies have shown that cannabinoids can inhibit cancer growth [6][7]. A number of clinical trials involving cannabinoids have announced a signi cant reduction in tumor size and prolonged survival time of patients with brain tumors [8]. The antineoplastic effects of cannabinoids involve different signaling pathways such as the mitogenactivated protein kinase (MAPK)/(ERK1/2) pathway [9][10].
The biological effects of cannabinoids are mediated through speci c G-protein coupled receptors (GPCR), CB1 and CB2 receptors [11]. The cloning of the cannabinoid receptors, CB1 receptors (CB1Rs) in 1990 [12] and CB2 receptors (CB2Rs) in 1993 [13], prompted recognizable proof of the presence of endogenous ligands, called endocannabinoids including Anandamide (N-arachidonoyl ethanolamine, AEA) and 2arachidonoyl glycerol (2-AG) [14]. CB1Rs are highly expressed in the brain and mediate many neuronal effects produced by endocannabinoids and cannabinoid drugs [15]. CB2Rs are also distributed in many areas in the brain at lower levels than CB1Rs, yet they have a more pronounced expression in peripheral immune and hematopoietic cells [16].
Cannabinoid CB1Rs are also expressed in many cancers; however, their expression levels vary. De Jesús et al. reported a decrease in the expression of CB1R in human glioma, with direct correlation with tumor grades [17]. In contrast, another study showed overexpression of CB1Rs in human glioma, which was positively associated with tumor grade [18][19]. Another study revealed that CB1Rs' overexpression was not associated with tumor grade in glioma [20]. Nonetheless, no difference in the expression level of CB1R was noted between normal brain tissues and glioblastoma multiforme cells as reported by Schley et al. [21], and weakly positive expression of CB1R was shown in both normal brain tissues and gliomas by Calatozzolo et al. [22]. In other tumors, CB1R have been found to be overexpressed at the protein level in prostate cancer [23], pancreatic cancer [24], hepatocellular carcinoma [25], melanoma [26], breast cancer [27], lung cancer [28], lymphoma [29] and ovarian cancer [30]. However, CB1 receptors downregulation was observed in renal tumor tissues [31], and no change in the expression of CB1R was reported in bone cancer [32]. To date, expression levels of CB1R in human gliomas remain controversial.
Therefore, the aim of the present study was to examine CB1R and p-ERK expression levels in human gliomas compared to that in normal brain tissues and reveal its clinicopathologic and prognostic signi cance.

Patients selection and tissue specimens:
A total of 61 formalin-xed para n-embedded glioma samples from patients who underwent surgical resection at different hospitals in Lebanon between June 2005 and October 2018 were provided by INP. Based on the World Health Organization (WHO) classi cation system, tumors were divided according to pathological grade as follows: 24 cases with low-grade gliomas and 37 cases with high-grade gliomas.
Out of the 61 specimens, 19 were resected from the frontal lobe, 18 from the temporal lobes, 13 from the parietal lobes, and one specimen from the occipital lobe. The localization was not documented for the remaining ten glioma specimens. The average age of the patients was 48.84 years (range, 12-80) and male-to-female ratio was 37:24. None of the patients received any type of therapy prior to surgery. As a control, 4 normal cerebral cortex brain tissues were obtained from patients who received epilepsy surgery and veri ed for the absence of any tumor. Three of these 4 normal brain tissues were from the temporal lobes and the remaining one was from the frontal lobe. In addition, 38 normal brain tissues were found adjacent to tumors (internal controls) and were veri ed for the absence of any in ltration of tumor cells by a pathologist.

Immunohistochemical procedure:
Serial sections of 4 µm thickness from each para n block were mounted on charged slides and dried in an oven at 60ºC for about 30 min. The immunostaining was carried out using an automated immunohistochemical staining system Ventana BenchMark XT autostainer with Ultraview Universal DAB detection kits (Ventana Medical Systems, Tucson, AZ). All the sections were stained in runs following one another. The automated staining technique is widely used in histology laboratories and research studies as it allows monitoring for errors such as inadequate volumes of reagents and unadjusted temperature.
The solutions used were from Ventana Medical System, Inc., USA. Antigen retrieval was achieved using heat-induced epitope retrieval (HIER) for 10 min at a temperature of 100 ºC. Prior to starting our experiments, we assessed two protocols for antigen retrieval: the rst was based on enzymatic retrieval by protease I, and the other was HIER, at different concentrations. Results of HIER were compared with those obtained by protease digestion showing clearly superior in uence of heat. The appropriate staining was achieved by HIER at 1:40 dilution. The used protocol in our study was approved by two independent pathologists, taking into consideration that HIER was performed in a convenient buffer and in a short period of time to prevent tissue damaging. The antibodies used were anti-cannabinoid receptor I rabbit polyclonal antibody (diluted 1:40; ab23703; Abcam; Cambridge, UK), anti-ERK1 (phospho Y204) + ERK2 (phospho Y187) rabbit polyclonal antibody (diluted 1:100; ab47339; Abcam; Cambridge, UK) and Ki67 anti-human mouse monoclonal antibody (ready to use dilution; PA0118; Leica Biosystem, UK) with an incubation time of 30 min at 36 ºC. Antibodies used were previously validated, such as anti-cannabinoid receptor I rabbit polyclonal antibody (ab23703; Abcam; Cambridge, UK) [33] [34], and shown to produce an appropriate pattern of staining in para n-embedded formalin-xed sections. CB1 receptor speci city has been veri ed using IHC on normal human cerebral cortex brain tissue slides purchased from Abcam, Cambridge, UK (ab4296) (Fig. 1A). Normal cerebral cortex brain tissues from our cohort of patients were also stained in order to compare the staining of tumor tissues with that in normal ones (Fig. 1B). Besides, positive staining of antibodies was compared to an appropriate negative control to avoid the false results of background staining (Fig. 1C). The omission of the primary antibody served as a negative control by which it was substituted with non-immune serum. The visualization system was OptiView DAB. The counterstaining with Hematoxylin II and Bluing reagent followed the immunostaining step. The prepared slides were dehydrated and covered by coverslips. Normal brain tissues used as negative and positive controls were included in each run.

IHC evaluation:
All the immunostained sections were examined for CB1R, p-ERK and Ki67 expressions under a light microscope and assessed by two independent pathologists in a blinded manner without any knowledge of the clinicopathologic characteristics of the samples. The evaluation of the immunostained sections was based on the staining intensity. The CB1R and p-ERK staining was scored in normal and tumor tissues from 0 to 3 (tissues with no staining were scored as 0; weak staining as 1, moderate staining as 2, and strong staining as 3). For statistical analysis, 0 and 1 scores were counted as low expression, whereas 2 and 3 scores were counted as high expression [33]. Ki67 scores were determined by assessing the percentage of cells expressing Ki67 immunoreactivity. The percentages ranged from 0-70%. The division of data as ≤ 30% and > 30% represents a division around the median and considered as low and high Ki67 expression respectively [35].

Analysis of mRNA expression patterns among the different publicly available online datasets:
We surveyed different publicly available datasets using data retrieved from the online database Oncomine (URL: http://www.oncomine.org; RRID:SCR_007834). Those datasets are comprised of human glioma and brain tumor tissues of different stages and types to better understand the expression pattern of CNR1 (CB1R) gene. Expression within tumor tissues was presented by fold-change expression and pvalues were obtained using t-tests. Data from the different datasets is presented as box and whiskers plots indicating median and interquartile range, and p-values were obtained using t-tests.

Statistical analyses
Statistical analysis was conducted using SPSS 24.0 statistical analysis software (SPSS, Inc., Chicago, IL, USA). The association between the expression of CB1R and clinicopathological variables was analyzed by Chi-square test. Fisher's exact test was used when necessary. Independent sample t-test was used to assess the difference in expression between groups. Pearson's correlation coe cient was performed to quantify CB1R and p-ERK expressions. Survival analysis was performed by the Kaplan-Meier test, and groups were compared by the log-rank test. The factors related to overall survival were determined by the Cox proportional hazard regression model. The differences were considered statistically signi cant at P < 0.05.

CB1R and p-ERK expressions in normal brain samples
The expressions of CB1R and p-ERK were assessed in normal brain tissues including 4 normal tissues and 38 normal tissues adjacent gliomas. CB1R positive staining was detected in all the normal brain tissues. High immunoreactivity was shown in all normal tissues (100%) and in 32/38 (84.2%) normal tissues adjacent to tumors. The staining intensity of CB1R did not vary between normal brain tissues (p = 0.429, independent sample t-test). The expression of CB1R was mainly located at the cell membrane and in the cytoplasm of neurons and glial cells and clearly found on the axons ( Fig. 2A and 1B).
With respect to the intensity of p-ERK, positive staining of p-ERK was also detected in all the normal brain tissues. However, weak p-ERK immunoreactivity was observed in all normal tissues (100%) and in 17/38 (44.7%) normal tissues adjacent to tumors. Statistical analysis showed no difference in the intensity of p-ERK expression in normal tissues compared to normal tissues adjacent tumors (p = 0.083; independent sample t-test). Neurons and glial cells exhibited weak positive cytoplasmic and nuclear staining for p-ERK (Fig. 2B).

Comparison of CB1R and p-ERK expressions in glioma
and normal brain tissues CB1R and p-ERK expression levels in glioma were also evaluated. Positive IHC staining of CB1R was detected in 59 patients (96.7%) out of 61. The expression of CB1R was found at the cell membranes, in the cytoplasm and nuclei of tumor cells. The tumors exhibited distinct degrees of CB1R immunoreactivity. Zero immunoreactivity was observed in 2 specimens (3.3%), weak immunoreactivity in 11 specimens (18.0%), moderate immunoreactivity in 21 specimens (34.4%), and high immunoreactivity in 27 specimens (44.3%). A signi cant difference in the staining intensity of CB1R was noted between glioma and normal tissues (p < 0.001; independent sample t-test) and between glioma and the normal tissues adjacent tumors (p < 0.001; independent sample t-test) with more pronounced expression in normal tissues in both cases (Fig. 2C).
With respect to the intensity of p-ERK, positive staining of p-ERK was also found in 59 (96.7%) glioma cases out of 61 in which p-ERK expression was observed in the cytoplasm and nuclei of tumor cells. High p-ERK immunoreactivity was detected in 49 (80.3%) of glioma samples. Statistical analysis revealed a signi cant difference between glioma and normal tissues (p < 0.001; independent sample t-test) and between glioma and normal areas adjacent tumors (p < 0.001; independent sample t-test) with p-ERK higher expression in glioma in both cases. Representative images are presented in Fig. 2D.

Clinicopathologic signi cance of the expression of CB1R in glioma
Investigation of the clinicopathologic signi cance of CB1R expression in glioma is summarized in Table 1. Samples with scores 0-1 and 2-3 were categorized as having low and high expression of CB1R and p-ERK (Fig. 3). No obvious relationships were noted between the expression of CB1R and clinicopathologic parameters, including patient gender (P = 0.539), age (P = 0.063), tumor grade (P = 0.941), resection size (P = 0.433), necrosis (P = 0.571), vessel density (P = 0.055) and Ki67 expression (P = 0.910). Yet, the expression of CB1R was signi cantly associated with p-ERK expression (p = 0.028). In addition, Pearson's correlation coe cient indicated that there was a weak positive correlation between CB1R and p-ERK expressions (r = 0.210, P < 0.01).

Factors affecting overall survival
Kaplan-Meier analysis showed that there was no signi cant evidence of a statistical difference in the survival times for patients with low and high expression of CB1R (P = 0.554) and with low and high p-ERK expression (P = 0.883) (Fig. 4). Moreover, multivariate analysis using Cox's proportional hazards model revealed that there was no evidence of a greater risk of death in association with age, histological grade, CB1R expression, p-ERK expression and Ki67 expression ( Table 2).

CNR1 (CB1R) mRNA expression patterns in human glioma tissues
In the present study, we aimed at better understanding the expression pattern of CNR1 (CB1R) gene in human glioma tumor tissues, so we surveyed different publicly available datasets (data retrieved from the online database Oncomine.org) comprised of human glioma tumor tissues of different stages and types. Interestingly, the analysis revealed that CNR1 (CB1R) gene was down-expressed in glioma tissues among the different datasets (fold change ranged between − 1.521 and − 9.886) (Figs. 5 and 6).

Discussion
Cannabinoids, the active component of cannabis, have been widely used for medical purposes for years [36], and proposed for treatments of numerous diseases such as multiple sclerosis, neurodegenerative disorders, epilepsy, schizophrenia and cancer [37]. Many studies have considered cannabinoids as a promising drug for glioma due to their antiproliferative effect [20], apoptotic potential and inhibition of angiogenesis [38]. So far, cannabinoids exert their effects by activation of speci c receptors, CB1 and CB2.
The present study was motivated by the studies done before on human gliomas that indicated contradictory results regarding the expression level of CB1R [10][11]. We have found that CB1Rs are mainly located at the cell membrane and in the cytoplasm of normal neurons and in glial cells. The same localization of CB1Rs was found in tumor cells including some mitotic cells, in addition to the nucleus. Our result is consistent with previous ndings that reported the presence of CB1R at the cell membrane [39], in the mitochondria, ER and nucleus of neurons [40], as well as in astrocytes [41] and in vascular endothelial cells [21] [42]. In cancer, CB1Rs were also detected in the cytoplasm of pancreatic tumor cells [24], and at the cell membrane, in the cytoplasm, and nuclei of colorectal cancer cells [33].
In our study, we found a down expression of CB1R in glioma tissues compared to normal tissues and normal areas adjacent tumors. Previous studies on CB1Rs in different types of cancers showed discrepancy results, speci cally in glioma. Our results are consistent with that of De Jesús et al. where CB1Rs were reported to be down expressed in gliomas [17]. However, overexpression of CB1Rs in gliomas was stated by Ciaglia et al. and Wu et al. [11][12]. Whereas, Held-Feind et al. showed a little increase in CB1R expression in glioma compared to normal brain tissue [20]. Schley et al. found no difference in CB1R expression between normal brain tissues and gliomas [21], and Calatozzolo et al. reported weak positive expression of CB1R in both normal brain tissues and gliomas [22]. Regarding other types of cancer, down-expression of CB1R was reported in renal and colorectal tumors [19][20][21]. In contrast, overexpression of CB1R was stated in prostate cancer [23], pancreatic cancer [24], hepatocellular carcinoma [25], melanoma [26], breast cancer [27], lung cancer [28], lymphoma [29] and ovarian cancer [30]. While the expression of CB1R showed no change in bone cancer [32].
The contradictory results regarding the expression of CB1R in gliomas in different studies could be explained by the variation in the source of control tissues, where CB1R normally show distinct distribution in brain areas [10,31]. However, in our study, the evaluation of the expression of CB1R was assessed in gliomas based on the comparison with normal brain tissues in one hand, and normal tissues adjacent tumors on the other hand (double control). Both revealed down-expression of CB1R in glioma. The different results in IHC assay could be sometimes due to the variation in the used antibodies and antigen retrieval techniques. In our study, the down-expression of CB1R in gliomas could be explained by receptor phosphorylation and endocytosis [44]. Endocytosis is a major mechanism for signal attenuation via the degradation of signaling receptors [45]. It is considered one of the impaired processes in cancer and plays a critical role in cancer progression [46].
We further analyzed the association between the expression of CB1R and the clinicopathologic characteristics of patients. The analysis showed no signi cant association between the expression of CB1R and many clinicopathologic characteristics including gender, age, tumor grade, resection size, vessel density, necrosis, and Ki67 expression. However, the expression of CB1R was associated with p-ERK expression. Several studies have reported the activation of ERK by CB1R [27]. Even though we did not see an overexpression in CB1R in tumors, the activation of ERK by phosphorylation can be due to the activation of several other signaling pathways [47]. Previous studies have described the activation of MAPK/ERK pathway and its involvement in the development of many types of cancers including breast cancer [48], gastric cancer [49], non-small cell lung cancer [50], gallbladder tumors [51] and gliomas [52]. Future studies on glioma could tackle the activation of ERK and role of CB1R in this regard.

Limitations
We believe that our study has some limitations. First, although our aim in this research paper was to evaluate the expression patterns of CB1R and p-ERK in human glioma tumor tissues, the sample size is relatively small and hence, the results obtained require conducting subsequent studies on a larger cohort.
In accordance, we believe that more data and follow up is required to assess the correlation of CB1R and p-ERK expression with clinical outcomes and to compare this expression among the different types of glioma tumors as well. Total ERK expression could also be determined in future studies to assess the activation status of ERK (by comparing total ERK with p-ERK expressions). Second, we used in our study a small number of normal brain specimens as a control. In fact, only 4 specimens were acquired since obtaining normal cerebral cortex brain tissue is indeed challenging and brain resection is usually done only in limited cases as in epilepsy patients or patients with brain tumors who need surgical resection. So, we collected in our study 4 brain specimens from epilepsy patients, and these tissues were veri ed for the absence of any tumor by a pathologist. Third, although the cannabinoid eld lacks reliable antibodies to precisely detect expression of CB1R, we used in our study anti-cannabinoid receptor I rabbit polyclonal antibody (diluted 1:40; ab23703; Abcam; Cambridge, UK) that is validated in Western blotting, immunohistochemistry, immunocytochemistry, and immuno uorescence, cited in more than 35 publications, and independently reviewed in 17 reviews (Source: https://www.abcam.com/cannabinoidreceptor-i-antibody-ab23703.html). Immunohistochemistry has been previously performed and results have been published using this antibody on human gastric carcinoma tissues [53] heart tissues [54], and articular cartilage samples [55] among others. In future studies, results shall be con rmed using an additional CB1 receptor antibody other than Abcam. Lastly, double immuno uorescence analyses could be also used to stain for neural cells (NeuN) versus glial cells (GFAP) to improve the quality of the study.

Conclusions
The present study demonstrated that CB1R is down expressed in glioma. This decrease in the expression of CB1R was not related to malignancy grades and other clinicopathologic features of glioma. Yet, our results showed a weak positive correlation between the expression of the CB1R and p-ERK. However, this study utilized a small number of samples and the results were only obtained from IHC assay. Therefore, a larger scale tumor sample size of well-characterized patients is indeed needed to con rm the obtained results. Future studies are required to clarify the molecular mechanisms of CB1R and other endocannabinoid components, and to identify whether the CB1/endocannabinoid system might serve as a promising therapeutic target for brain tumors.

ACKNOWLEDGMENTS
We would like to thank all members of the Neuroscience Research Center at the Faculty of Medical Sciences, Lebanese University, for their support. In addition, we would like to thank all members of Institut National de Pathologie for their help and support.
Funding statement: This work was supported by funding from the Neuroscience Research Center, Faculty of Medical Sciences, Lebanese University, Beirut, Lebanon. The funders had no role in study design, data collection, and analysis, decision to publish, or preparation of the manuscript.
Con ict interest statement: The authors declare no con ict of interest.
Ethics approval and consent to participate: Ethics Committee approval was obtained from the Faculty of Medical Sciences of the Lebanese University (LU) and the Institute National de Pathologie (INP). Ethical clearance was accomplished as per the norms and in accordance with relevant guidelines and regulations of INP and LU. The need for informed consent from patients was waived.
Consent for publication: Not applicable.
Availability of data and material: Not applicable.