Expression and Functional Relevance of Cannabinoid Receptor 1 in Hodgkin Lymphoma

Background Cannabinoid receptor 1 (CB1) is expressed in certain types of malignancies. An analysis of CB1 expression and function in Hodgkin lymphoma (HL), one of the most frequent lymphomas, was not performed to date. Design and Methods We examined the distribution of CB1 protein in primary cases of HL. Using lymphoma derived cell lines, the role of CB1 signaling on cell survival was investigated. Results A predominant expression of CB1 was found in Hodgkin-Reed-Sternberg cells in a vast majority of classical HL cases. The HL cell lines L428, L540 and KM-H2 showed strong CB1-abundance and displayed a dose-dependent decline of viability under CB1 inhibition with AM251. Further, application of AM251 led to decrease of constitutively active NFκB/p65, a crucial survival factor of HRS-cells, and was followed by elevation of apoptotic markers in HL cells. Conclusions The present study identifies CB1 as a feature of HL, which might serve as a potential selective target in the treatment of Hodgkin lymphoma.


Introduction
The Endocannabinoid system consists of cannabinoid receptors, their endogenous, exogenous or synthetic ligands and the enzymes responsible for synthesis and degradation of endogenous ligands. So far, two types of cannabinoid receptors, namely CB 1 and CB 2 have been identified. Both belong to the superfamily of G-proteincoupled receptors [1,2]. Activation of Cannabinoid receptors inhibit adenylate cyclase and cAMP production via Gi/o coupling but also activate phospholipase C, MAPK and phosphoinositide 3kinase (PI3K) signaling pathways (Gq coupling) [3]. Cannabinoid receptor 1 (CB 1 ) represents one of the most abundant G-proteincoupled receptors (GPCR) in the brain [4]. It binds exogenous and endogenous cannabinoids and is thereby associated with several physiological and pathological processes within the central nervous system [5][6][7][8] but is also linked to a variety of peripheral disorders, such as obesity, liver fibrosis [9][10][11] and cancer [2,3,12].
Recent evidence points to the involvement of CB 1 in growth of human breast cancer cells [13][14][15][16]. High CB 1 -protein expression in prostate cancer cases was associated with an increased dedifferentiation of tumor cells and poor prognosis [17]. In contrast, patients suffering from hepatocellular carcinoma with high mRNA-expression of CB 1 reportedly have a better prognosis [18]. SR141716 (RimonabantH), an antagonist/inverse agonist of CB 1 , was shown to induce apoptosis in an in vivo model of colon carcinoma [19,20]. The expression level of cannabinoid receptors in astrocytoma-cells has been reported as crucial for downstream signaling processing with consequences on cell-viability [21]. Thus, the effects of either activation or blocking CB 1 depends on the tissue investigated and the local expression level of the receptor.
Hodgkin lymphoma (HL) is one of the most frequent lymphomas in the Western world mainly affecting young adults. Although the majority of HL cases at any clinical stage have a good prognosis under adequate therapy, still about 20% of patients develop highly mortal relapse [22]. HL is classified into classical HL (cHL) representing the largest subtype (95% of HL) and nodular lymphocyte predominant HL (NLPHL) which accounts for around 5% of the cases. The largest subentities of cHL are the nodular sclerosis (NS) and mixed cellularity (MC) types with 70% and 20% of the cases, respectively [23]. Histologically, cHL consists of B-cell derived mononuclear Hodgkin and multinucleated Reed-Sternberg (HRS-) cells surrounded by a vast non-neoplastic infiltrate [24]. p65 (also known as RelA) is a member of the NF-kB transcription factor family which is a key mediator in the TNFsignaling pathway contributing to a variety of cellular processes such as survival, proliferation and immune response. In HRS-cells, several mutations were identified leading to a loss of endogenous inhibitors of p65 such as I-kB [25][26][27][28] and A20 [29,30]. Via bypassing apoptosis with subsequent cell death, high expression and activity of p65 in HRS cells [31] is regarded as a key mechanism in the pathogenesis of cHL [32].
Thus far, the functional relevance of CB 1 has not been elucidated in HL. We therefore investigated the expression of CB 1 in primary cases of different HL entities. We further determined the impact of CB 1 specific agonist ACEA and the inverse agonist / antagonist AM251 on signal transduction cascades such as NF-kB/p65-or PI3K/Akt-pathway and on cell fate in HL-derived cell lines.

Tissue samples
All tissues samples were studied in accordance with the Helsinki declaration. Specimens, which were originally submitted for diagnostic purposes, were retrieved from the files of the Department of Pathology of the University of Frankfurt.

Immunohistochemical staining
For immunohistochemical staining, 3 mm thick sections of fixed (5% [w/v] buffered formalin) and paraffin-embedded tissue samples were generated and deparaffinized. Antigen retrieval was performed by incubation in a microwave oven for 10 min in 1 mM EDTA (pH 8.0). Sections were exposed to a 3% (v/v) H 2 O 2 -methanol solution for 10 min, washed in water and blocked with Tris-buffered saline (TBS, 3% [w/v] bovine serum albumin, BSA) for 20 min at room temperature. Antibodies against Nterminal (1.65 mg/mL, #101500, Cayman Chemical, Ann Arbor, USA) or C-terminal CB 1 (5 mg/mL, #10006590, Cayman Chemical) were added in TBS containing 3% (w/v) BSA for 16 h at 4uC. After washing with TBS, sections were incubated with rabbit specific biotinylated secondary antibody (DAKO, Hamburg, Germany) followed by horseradish peroxidase conjugated streptavidin (DAKO) with TBS-wash steps in between. Staining was developed with diaminobenzidine (DAB, DAKO) for 3 min and subsequent counterstain of nuclei was performed using Meyer's haematoxylin (Applichem, Darmstadt, Germany).
Specific signals for N-and C-terminal CB 1 in human hippocampus and in a case of nodular sclerosing HL were absent when antibody was preabsorbed using the corresponding CB 1 immunizing peptides in equimolar concentrations confirming antibody specificity ( Figure S1). Picture acquisition was performed using a Zeiss microscope (Zeiss, Göttingen, Germany) equipped with an Axio-Cam digital camera (Zeiss) at 100, 200 and 400 fold magnification. The analysis of CB 1 immunoreactivity was independently performed by three pathologists. A case was rated positive when more than 30% of its tumor cells displayed immunoreactivity for CB 1 as generally accepted.

Cell culture experiments
Cell lines were obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (DSMZ, Braunschweig, Germany) maintained in RPMI 1640 containing fetal bovine serum (FBS, PAA, Pasching, Austria) and penicillin-streptomycin mix (Gibco) at 37uC and 5% CO 2 . Peripheral blood Blymphocytes were isolated from healthy donors and isolated by Ficoll-Paque PREMIUM (GE Healthcare, München, Germany) density centrifugation. CD19 + cells were separated by magnetic cell separation using the MACS system (Miltenyi Biotec, Bergisch Gladbach, Germany) and were maintained in RPMI with 10% (v/ v) FBS containing penicillin-streptomycin.
Signal intensities of the digitized images were analyzed using a combination of densitometry and volumetry as implemented in the QuantiScan software (BioSoft, Cambridge, UK) as described in detail before [33]. Each area/density value for a specific protein band was normalized against the corresponding ß-Actin signal of each extract.

Flow cytometric analyses
100,000 cells were cultured in RPMI 1640 containing 10% (v/v) FBS in a 6 well culture plate and treated as indicated. Flow cytometric analysis was performed using a FACSCanto II Flow Cytometer (BD Bioscience, San Jose, USA). For each measurement, 10,000 events within the live gate were counted. Fluorescence distribution was displayed as dot plot analysis and the percentage of fluorescent cells in each quadrant was determined using DIVA-software (BD Bioscience).
To characterize externalization of phosphatidyl-serine and plasma membrane permeability, the AnnexinV-PE/7-AAD Apoptosis Detection kit (BD Pharmingen, Franklin Lakes, USA) was used according to the manufacturer's instructions.
Cell cycle analysis was performed using Click-iT EdU Alexa-647 Flow Cytometry Assay (Invitrogen) according to the manufacturer's instructions. Briefly, DNA synthesis was measured by incorporation of the Alexa-647 labeled thymidine analog 5ethynyl-2'-deoxyuridine (EDU) and the DNA content was examined using a cell cycle sensitive dye.

Statistical analysis
Data were analyzed using the GraphPad Prism 5.0 software (GraphPad, San Diego, CA, USA). For multiple group comparison one-way ANOVA algorithm was used, followed by the Bonferroni post hoc test. For comparison of two groups Student's t test was performed. The criterion of significance was p,0.05.

Results
Expression of CB 1 in cases of classical Hodgkin lymphoma and non-neoplastic lymphatic tissues To determine occurrence and localization of CB 1 protein in Hodgkin lymphoma and normal lymphatic tissue, immunohistochemical staining with CB 1 -specific antibody was performed. Abundant CB 1 protein was found in CD30 + HRS cells of cHL whereas the surrounding reactive, non-neoplastic lymphatic infiltrate was largely negative. The CB 1 -specific signal was located inside HRS cells, mainly with a perinuclear staining pattern ( Figure 1A, B). In cases of tonsillitis and lymphadenitis, few cells displayed CB 1 -positivity in some of the germinal centers and interfollicular zones ( Figure 1C, D). To further characterize the positive cells in the reactive tissues, CB 1 -counterstaining was performed in a case of tonsillitis. CB 1 -specific signal was localized in the cytoplasm of CD138 + plasma cells and within branches of CD68 + macrophages while both CD3 + and CD20 + lymphocytes were found negative for CB 1 (Figure 2).  Figure 3A,B). In PCR analyses, CB 1 was detected in the neuroblastoma cell line SHSY as well as in all investigated lymphoma derived cells ( Figure 3A). Interestingly, a moderate signal for the ''peripheral'' cannabinoid receptor CB 2 , was obtained in lymphoma cells, which was weaker in SHSY cells. All investigated cell extracts showed strong expression of GPR55. Since Hodgkin-and Reed-Sternberg cells of HL originate from B-cells, we included isolated CD19 + B-lymphocytes in the Western blot analyses. A band for CB 1 at approximately 60 kDa was most prominent in L428 cells, lower in L540, L1236 and KM-H2. Two other bands were detected at around 50 and 80 kDa, both being most intense in KM-H2, followed by L1236, L428, L540 and to a lesser extent in Karpas422 cells. HDLM2, BJAB, SUDHL8, Farage cell lines and CD19 + cells were negative ( Figure  3B). Different sizes obtained in Western blot analyses might be due to post-translational modification of the 50 kDa core protein, specificity of the used antibodies was verified by preabsorption. Next, functional relevance of CB 1 was tested in L428, L540, KM-H2 as representative cell lines for HL and in Karpas 422 cell line representing B-NHL.

AM251 impairs viability of HL cell lines
To test the functional relevance of detected CB 1 on cell fate, cell lines were kept in culture medium containing 10% (v/v) FBS to provide optimal growth condition. The cell lines L428, L540, KM-H2 and Karpas 422 were treated with CB 1 antagonist AM251 and viability was assessed using MTT-assay. Gallotta and colleagues observed a decrease in viability of Jurkat cells with an IC 50 of around 12 mM using SR141716, a CB 1 -antagonist with an affinity to CB 1 similar to AM251 [34]. Therefore, we tested viability of lymphoma cells using CB 1 ligands at a maximum of 10 mM each.
Next, we confirmed that AM251 induced effects on cell viability in L428 were due to inhibition of CB 1 and not to activation of the GPCR GPR55, another target of AM251 [35] which was also detected in HL cell lines at mRNA level ( Figure 3A). Therefore, GPR55-agonist LPI was applied as control. In comparison to vehicle treated cells, LPI had no significant effect on cell viability at 3 mM (101.568.7%, p.0.05, n = 12) but a small significant inhibitory effect at 10 mM (93.965.8%, p,0.05, n = 12) ( Figure  S4A).
To uncover the mechanisms behind the described decreased viability, further experiments were carried out in L428 cells, which displayed both, a strong CB 1 immunosignal and a remarkable response in the viability assays.
To resolve the relative changes of G1, early-S, late-S and G2M phases of cell cycle after AM251 treatment (10 mM) of L428 cells, EdU and DNA-specific staining was performed followed by flow cytometric analysis. After 72h and 120h L428 cells were stained with EdU and pacific blue (DNA stain). After 72 h, the number of cells in early-S changed from 20.2% (vehicle) to 12.3% (AM251), in late-S from 22.4% to 13.5%, in G1 from 41.9% to 37.1% and in G2M from 15.4% to 37.0%, respectively. After 120 h the population of cells changed in early-S from 11.8% (vehicle) to 0.7% (AM251), in late-S from 21.1% to 0.2%, in G1 from 47.7% to 52.8% and in G2M from 19.3% to 46.3%, respectively ( Figure  4B). Compared to the effects of AM251, the distribution of cells in all four phases was only slightly changed after treatment with CB 1 selective agonist ACEA for 72 h and 120 h ( Figure S4B).
To prove whether down stream members of apoptosis pathways were activated, the effects of AM251 treatment on caspase-3 cleavage were evaluated. In comparison to vehicle controls, application of AM251 (10 mM) for 96 h resulted in a decline of full length caspase-3 (33.9613.8%, p,0.01) and in parallel to an induction of cleaved caspase-3 (136.469.2%, p,0.05, Figure 4D). Discussion Involvement of CB 1 in cell survival has been described in several types of cancer models but the functional relevance of CB 1 in Hodgkin lymphoma has not been studied to date. In the present study, we report the abundance and anti-apoptotic role of CB 1 in classical Hodgkin lymphoma.
The presence of functional CB 1 -protein has been reported in prostate cancer and hepatocellular carcinoma [18,36]. In prostate cancer cells, higher amounts of CB 1 were found when compared to their benign counterparts and high CB 1 -immunoreactivity correlated with severity of the disease [17]. Contrarily, elevated cannabinoid receptor expression in hepatocellular carcinoma was associated with improved prognosis [18]. Prognostic relevance of CB 1 expression levels in lymphoid neoplasms such as HL remains to be determined. CB 1 protein is located at the plasma membrane, as well as in intracellular vesicles such as lysosomes. Intracellular CB 1 receptors are associated with heterotrimeric G proteins, functional and able to mediate signal transduction [37]. As shown in other malignancies such as prostate carcinoma, we also found an intracellular and perinuclear CB 1 -staining pattern in tumor cells of HL. The subcellular localization of CB 1 might further hint to internalization upon binding highly lipophilic endogenous ligands such as arachidonoylethanolamide (AEA) or 2-arachidonoylglycerol (2-AG) produced by HRS cells themselves or the surrounding reactive infiltrate.
Since blocking of CB 1 resulted in a decline in viability of L428 cells, one might hypothesize that this GPCR is a survival factor for HRS cells and, in conclusion, its activation promotes tumor cell growth. In fact, involvement of activated CB 1 signaling in liver regeneration was recently demonstrated in mice after partial hepatectomy via upregulation of cell cycle regulators [38]. In HL cells, however, application of CB 1 agonist ACEA showed only marginal effects on apoptotic parameters in HL cells and even led to a slight decrease of cell viability. CB 1 might act as a promoter of cell growth in HL cells. Endogenous agonists (endocannabinoids) such as AEA or 2-AG in the culture media might bind and activate CB 1 in these cells, leaving the addition of synthetic agonists with adverse effects. Such a possible role for paracrine or even autocrine action of endogenous cannabinoids in HL is yet to be determined.
In colorectal cancer, the effects of CB 1 -specific treatment on cell viability are controversial. It was shown that CB 1 -activation of colon carcinoma cells resulted in inhibition of growth [39]. Santoro and colleagues reported on increased cell death of colon cancer cells using the CB 1 -antagonist SR141716 [19]. In rhabdomyosarcoma, increased CB 1 expression was associated with enhanced proliferation and invasion which was blocked by application of CB 1 antagonist/inverse agonist AM251 [40].
In breast cancer cell lines treated with the plant derived cannabinoid and CB 1 -agonist delta-9-tetrahydrocannabinol (THC), no alteration of cell viability was detected in vitro, but when the same cells were transplanted into mice, enhanced tumor growth was observed. In that case, the THC-mediated suppression of Th1-specific immune response in the animals was proposed to be responsible for enhanced growth in vivo [41]. The functional relevance of the endocannabinoid system in reactive immune cells surrounding HRS-cells needs further investigation in HL, with special regard to the fact that these specific reactive immune cells surrounding HRS-cells represent CB 1 negative immune cells in vivo. Putative endocannabinoids might act via cannabinoid receptor 2 (CB 2 ) since CB 2 is known to be predominantly located in immune cells modulating immune cell migration and cytokine release [42].
Since we found a predominant expression of CB 1 protein in lysates of HL derived cell lines, we subsequently analyzed the effects of pharmacological activation and inhibition of CB 1 in HL derived cells expressing a relatively high amount of CB 1 . ACEA did not alter cell cycle or apoptotic parameter in flow-cytometric analyses. However, a striking decrease of cell viability down to 11% (L428), 44% (L540) and 13% (KM-H2) compared to control levels was observed upon application of 10 mM of CB 1 inverse agonist AM251. A Striking reduction of L428 cells in S-phase was seen after inhibition of CB 1 . Although the effects of highest doses of CB 1 -agonist ACEA were significant on cell viability, they were not compelling, since 83% of L428 cells were still viable.
Expression of Cnr1 in B-and T-cell NHL has earlier been described [43][44][45]. In line with these data, we also demonstrate Cnr1 in B-NHL cell line Karpas 422 at mRNA as well as CB 1 protein level. Unlike in HL cells, viability was not impaired after pharmacological inhibition of CB 1 . We conclude that the effects of AM251 on viability used in HL cells are not of unspecific toxic nature and hypothesize that the B-NHL cell line Karpas 422, compared to HL tumor cells, might use other intrinsic mechanisms bypassing CB 1 dependent cell death.
Recently, another target of several CB 1 -antagonists was uncovered as AM251 was demonstrated to bind and activate the orphan receptor GPR55 [35]. To exclude GPR55 as a mediator of the effects observed after AM251 treatment, we performed viability assays using LPI, a ligand highly specific to GPR55 [46]. A significant decrease of L428 cell viability of 6% was detected with LPI which was marginal when compared to the effects of AM251 (89% reduction). Hence, the observed effects of AM251 on viability of L428 cells were most probably due to inhibition of CB 1 rather than activation of GPR55.
Previously, we reported on aberrant expression and activation of certain receptor tyrosine kinases (RTK) in cases of HL. The constitutive activation of downstream signaling cascades was found to be an important survival-factor for HRS-cells [47,48]. To determine whether RTK-signaling is involved in the observed reduction of cell viability after CB 1 -inhibition, we analyzed the effects of AM251 on phosphorylation of Akt and Erk1/2, two downstream targets within RTK-signaling pathways [49]. However, no significant change in phosphorylation of Akt at Ser473 or Erk1/2 at Thr202/Tyr204 was observed.
A crucial survival factor of HRS-cells in HL cases is the transcription factor p65 [50]. Aberrant activation of this transcription factor is a central mechanism to bypass apoptosis [32]. Activation of CB 1 by THC was shown to increase the activity of p65 [51]. After CB 1 antagonization, we found a remarkable decrease of p65-levels in L428 cells. Since others showed an induction of apoptosis of HRS-cells after knock-down of p65 [31], increased cell demise after inhibition of CB 1 may be due to decreased p65-levels.
In conclusion, our data reveal that CB 1 expression is a common feature of HRS-cells in cHL and suggest its antagonization as a possible novel strategy for specific pharmacological treatment of HL. Figure S1 Preabsorption of CB1-specific antibody. Staining of human hippocampus slices and a case of NS with CB 1 -antibody and CB 1 -antibody incubated for 3 hours with CB 1immunizing peptide. In the Cornu ammonis region and in the hilar zone, some neurons displayed perinuclear positivity. Further, the neuropil of the hilar zone showed strong granulated CB 1 abundance. The cytosol of HRS cells was stained positive for CB 1  Analysis of 153 B-cell lymphoma cases stained with a N-terminal CB1-antibody. In total, cHL cases were positive in 83.7%. The cHL sub-entities NS and MC were positive in 90.1% and 75%, respectively. None of the NLPHL cases were found positive. In B-NHL subentities, 0% of MCL, 5.3% of MZL, 11.5% of DLBCL, 0% of FL and 0% of B-CLL cases were positive for CB1. Cases of NLPHL, DLBCL, FL, MCL, MZL and B-CLL were stained against CB 1 (brown). Note that tumor cells of each entity (arrows) are mostly negative for CB 1 whereas only a few non-neoplastic reactive cells (arrow heads) show a positive immunoreaction for CB 1 . Bars = 20 mm (TIF) Figure S4 Effects of CB1 agonist ACEA and GPR55 agonist LPI on lymphoma derived cell lines. A) Cell viability was determined in L428 and Karpas 422 cells treated with the indicated concentrations of ACEA for 120 h using the MTT-assay. When compared to vehicles, ACEA did not reduce the number of vital cells at 3 mM significantly (p.0.05) but at 10 mM (p,0.05). Administration of maximal dose of ACEA did not change the viability of Karpas 422 (p.0.05). The GPR55 specific agonist LPI slightly reduced viability of L428 cells at 10 mM (p,0.01). Values represent means 6 SD. B) ACEA treated L428 cells and cell cycle proportions after 72 h and 120 h as revealed from EdU/nuclear stain and subsequent flow cytometric analysis. C) L428 cells were stained with AnnexinV/7-AAD. Subsequent flow-cytometric analysis revealed slight changes after 72 h of treatment with 10 mM ACEA. (TIF)