2-Arachidonoylglycerol Modulates CXCL12-Mediated Chemotaxis in Mantle Cell Lymphoma and Chronic Lymphocytic Leukemia

Simple Summary Most lymphoma patients relapse after therapy, mainly due to malignant cells hidden in tissues and protected by different components from the surrounding microenvironment. In this study, we describe the interaction between two important factors from the microenvironment, the biolipid 2-arachidonoylglycerol and the chemokine CXCL12. We specifically aimed at better understanding the role of 2-arachidonoylglycerol in the migration of lymphoma cells, and how it affects the migration towards CXCL12 and specific downstream signaling pathways. 2-arachidonoylglycerol signals through the cannabinoid receptors type 1 and type 2 (CB1 and CB2). CB1 and CB2 are variably expressed in B-cell lymphoma. We demonstrate that 2-arachidonoylglycerol induces migration of B-lymphoma cells via CB1 and CB2. We also provide evidence that the signaling pathways activated by 2-arachidonolglycerol and CXCL12 are cross-talking. Abstract To survive chemotherapy, lymphoma cells can relocate to protective niches where they receive support from the non-malignant cells. The biolipid 2-arachidonoylglycerol (2-AG), an agonist for the cannabinoid receptors CB1 and CB2, is released by stromal cells in the bone marrow. To investigate the role of 2-AG in lymphoma, we analyzed the chemotactic response of primary B-cell lymphoma cells enriched from peripheral blood of twenty-two chronic lymphocytic leukemia (CLL) and five mantle cell lymphoma (MCL) patients towards 2-AG alone and/or to the chemokine CXCL12. The expression of cannabinoid receptors was quantified using qPCR and the protein levels visualized by immunofluorescence and Western blot. Surface expression of CXCR4, the main cognate receptor to CXCL12, was analyzed by flow cytometry. Phosphorylation of key downstream signaling pathways activated by 2-AG and CXCL12 were measured by Western blot in three MCL cell lines and two primary CLL samples. We report that 2-AG induces chemotaxis in 80% of the primary samples, as well as 2/3 MCL cell lines. 2-AG induced in a dose-dependent manner, the migration of JeKo-1 cell line via CB1 and CB2. 2-AG affected the CXCL12-mediated chemotaxis without impacting the expression or internalization of CXCR4. We further show that 2-AG modulated p38 and p44/42 MAPK activation. Our results suggest that 2-AG has a previously unrecognized role in the mobilization of lymphoma cells by effecting the CXCL12-induced migration and the CXCR4 signaling pathways, however, with different effects in MCL compared to CLL.


Introduction
B-cell lymphomas are caused by genetic aberrations in the lymphoma cells, but the tumor tissue microenvironment contributes to the growth and survival of lymphomas, such

Cell Lines and Primary Samples
The MCL cell lines JeKo-1, Granta519, JVM-2 and the T-cell line Jurkat were obtained from DSMZ (Braunschweig, Germany). All cell lines were maintained in culture in RPMI 1640 medium, GlutaMAX supplement, and HEPES (RPMI-GM, Gibco, UK), supplemented with 50 µg/mL of gentamicin and 10% fetal bovine serum (FBS) (Gibco) under the conditions of 5% CO 2 at 37 • C. The glioblastoma cell line U-251 was purchased from ECACC and was kept in culture in MEM complete medium supplemented with 10% FBS (Sigma-Aldrich/Merck, Darmstadt, Germany) at 5% CO 2 at 37 • C, until confluency before splitting them using trypsin-EDTA (0.25%, ThermoFisher Scientific, Paisley, UK) to detach the cells. MCL and CLL blood samples were collected at Karolinska University Hospital, Huddinge. The study was approved by the Regional Central Ethical Review Board at Karolinska Institutet and patients gave informed consent, in compliance with the Declaration of Helsinki.

Lymphoma Cell Isolation from Peripheral Blood Samples
Blood samples were enriched for B cells by negative selection using RosetteSep™ containing a cocktail of antibodies (StemCell Technologies, Saint Egreve, France) according to the manufacturer's protocol. Thereafter, cells were collected using Ficoll-Paque PLUS (GE Healthcare Life Science, Chicago, IL, USA), washed with PBS and directly used for functional assays or viability frozen in 50% RPMI-GM, 40% FBS, 10% DMSO, and preserved at −150 • C. After B cell enrichment by RosetteSep™, the lymphoma cell purity (68.4-99.8%; median 97.1%, Supplemental Table S1) was assessed by double expression of CD19 and CD5 within the CD45 positive population (Supplemental Figure S1A), using flow cytometry.

siRNA Introduction by Electroporation
Transient downregulation of the CNR1 and CNR2 genes was obtained using small interference RNA (siRNA). JeKo-1 cells were transfected with commercial gene-specific siRNAs (CNR1: s3261; CNR2: s3265 and s3263; ThermoFisher) or negative siRNA by electroporation method using the AMAXA machine and the Nucleofector Kit C (Lonza, Basel, Switzerland). Briefly, 1 µM of the siRNA was introduced together with the Nucleofector solution, the program X-01 from the electroporation machine was applied, 20% FBS in RPMI-GM was added to the cells immediately after transfection and then the cells were transferred to a 6-well plate containing 10% FBS in RPMI-GM and incubated at 37 • C, 5% CO 2 until functional assay.

qPCR
mRNA expression levels of genes encoding for CB1 (CNR1) and CB2 (CNR2) were assessed by real-time quantitative PCR using Platinum SYBR Green qPCR Supermix-UDG (Invitrogen). Custom-made primers were purchased from Invitrogen, and sequences for respective genes were as follows: CNR1 forward: 5 -CATTAAGACGGTGTTTGCATTCT-3 , reverse: 5 -CGTGTCGCAGGTCCTTACTC-3 ; CNR2 forward: 5 -GACACGGACCCCTTTTTGCT-3 , reverse: 5 -CCTCGTGGCCCTACCTATCC-3 ; β-actin (ACTB) forward: 5 -AAAGACCTGTA CGCCAACACA-3 , reverse: 5 -AGTACTTGCGCTCAGGAGGA-3 . Since false positive signals may appear during the quantification of CNR2 after introduction of specific siRNA [33], we used a different set of primers to quantify CNR2 expression after silencing only: CNR2_2 forward: 5 -ATGAGGCCTCTTCCCAATTT-3 ; CNR2_2 reverse: 5 -CAGGCTGTCTTCCAGGAGTC-3 . Each sample was prepared in triplicates in a 96-well plate (BioRad, Leipzig, Germany) and the reactions were performed with the C1000 Thermal Cycler (BioRad). An initial denaturation step was performed at 95 • C for 3 min, followed by 15 s at 95 • C, then by 30 s 57 • C step to allow primer annealing and elongation. The two latest steps were repeated 39 times for the amplification. The results were analyzed, and cycle threshold (Ct) values of transcripts were quantified using CFX manager software (BioRad). ∆Ct values (meaning that a low value corresponds to a high expression, a high value to a low expression, and a level at 0 to no expression) were determined using ACTB as reference. ∆Ct value from either the control condition or from enriched B cells from a buffy coat was used to calculate ∆∆Ct values and the change in gene expression levels was determined by relative fold increase (RFI; 2 −∆∆Ct ).

Activation of p38, AKT and p44/42 Signaling Pathway and Western Blot
Cells were rested for 1 to 3 h in 0.1% fatty acid free bovine serum albumin (BSA, Sigma-Aldrich)/RPMI-GM at 37 • C, then resuspended in 0.1% BSA/RPMI-GM at 2 × 10 6 cells per mL, and treated with CXCL12 (200 ng/mL) and/or 2-AG (100 nM) for 2 min. AMD3100 (25 µg/mL), SR141716 (10 nM), SR144528 (10 nM), AM6545 (10 nM), LY294002 (50 µM), SB202190 (25µM) and PD98059 (50µM) pre-treatments were done for 20 min to 1 h prior to 2-AG +/− CXCL12 treatments. Cells were washed with cold PBS to stop the reaction, pelleted, placed on ice and cell lysates were prepared using RIPA buffer supplemented with phosphatase and protease inhibitors cocktails (all from Sigma-Aldrich). For each specific pathway, proteins were resolved on two parallel 12% NuPAGE gels (Invitrogen) (one for phospho-protein, one for the corresponding total protein) and transferred using a semi-dry transfer system onto PVDF membranes (Millipore, Burlington, MA, USA). Non-specific binding sites were blocked with 10% BSA TBS-T solution for 1 h at room temperature (RT), then probed overnight at 4 • C with the respective primary antibodies: anti-phospho-AKT (S473), anti-phospho-p44/42 (T202/Y204), anti-phospho-p38 (T180/Y182), total-AKT, total-p44/42, total-p38 or anti-GAPDH (all from Cell Signaling) in 5% BSA in TBS-T. Membranes were then washed in TBS-T and probed with secondary antibodies (HRPconjugated anti-rabbit or anti-mouse; GE Healthcare). Blots were developed using Western Lightning Plus ECL (PerkinElmer, Groningen, The Netherlands) and visualized using LiCor machine. For re-probing of membranes, HRP was blocked using the SG substrate kit (Vector Laboratories, Newark, CA, USA). Bands intensity was measured using ImageJ software. The quantification was done by first normalizing bands intensities for total or phospho to their respective GAPDH control on the respective blot. Then, we could normalize phospho/GAPDH to total/GAPDH, giving the activation status for each pathway. The same Western blot protocol was used for detection of cannabinoid receptors.

Flow Cytometric Analysis of CXCR4 Surface Expression
Basal CXCR4 surface expression was analyzed in MCL cell lines and in primary MCL and CLL cells. JeKo-1 cell line or freshly isolated cells were treated with CXCL12 (200 ng/mL) and/or 2-AG (100 nM) for 10 min or 4 h, washed twice with ice-cold PBS and stained with IgG control or anti-CXCR4 antibody for 20 min on ice. Stained cells were analysed by BD FACSCanto II (BD Biosciences, Erembodegem, Belgium). Data analysis was conducted using FlowJo for Windows version 10.

Immunofluorescence Staining
For immunofluorescence staining, cytospins from MCL cells in culture were prepared (2-5 × 10 4 cells/slide) or the cell line U-251 was cultured on an 8-well slide (2 × 10 4 cells/well) for 48 h prior to the staining protocol. Cells were first fixed with 4% paraformaldehyde for 10 min at RT. After two washes in PBS-T, permeabilization was achieved with 0.5% TritonX100 for 10 min at RT. Unspecific sites were then blocked with 4% donkey serum for 1 h at 37 • C, before adding the primary antibody anti-CB1 (rabbit) or anti-CB2 (mouse), which was incubated for 2 h at RT in humidity chambers. Respective secondary antibodies (donkey anti-mouse coupled to FITC or donkey anti-rabbit coupled to TRITC) were applied for 1 h at RT after washes with PBS-T. Excess of antibodies was washed with PBS-T and slides were dried before mounting the coverslip with mounting medium containing DAPI. Slides were then visualized under a confocal inverted microscope Nikon Eclipse Ti-E, A1R, using X20 magnification objective. Cell lines U-251 and Jurkat were used as positive controls for CB1 and CB2 staining, respectively (Supplementary Figure S5). Analysis of the images obtained from ICC/IF was performed using CellProfiler (version 4.07), Broad Institute of MIT and Harvard, USA. Pipelines were created in CellProfiler to determine mean fluorescence intensity of CB1 (TRITC) and CB2 (FITC) immunofluorescence staining per cell. Either CB1 or CB2 were assigned as child objects to DAPI-stained nuclei which were used as surrogates of total cell counts. Percentage of CB1+ or CB2+ cells were calculated by dividing the number of fluorescently labelled cells by the total number of cells per slide. (Pipelines used for image analysis are uploaded in Supplementary Files).

Chemotaxis
Chemotaxis experiments were performed using a Boyden chamber-based method in which inserts shield the fluorescence in the upper chamber and thus allow detection of only migrated fluorescently labelled cells (FluoroBlok technology, Corning, Durham, NC, USA). FluoroBlok™ inserts were used for MCL cell lines in 8 µm pore-size, in 24-well plates with fluorescent cell number as the readout using Nikon Eclipse Ti confocal microscopy system, or in 96-well plate format with fluorescence intensity readout using spectrofluorometer (FLx800, BioTek, Winooski, VT, USA), which corresponded to the cell number (see Supplementary Figure S2A). Both settings were used for JeKo-1 cell line and the latter would be used for Granta519 and JVM-2 since they form aggregates making it difficult to quantitate migrated cells by microscopy. In the case of primary MCL and CLL cells, 5 µm inserts were used (Costar, Washington, DC, USA), in 24-well plates using fluorescent cell number by microscopy as the read out. Before the chemotaxis assay, cultured cells were washed in order to remove FBS. Frozen isolated primary cells were thawed and kept in 10% FBS in RPMI-1640-GlutaMAX for 1 h. Primary and cultured cells were then stained with the cell-permeable green fluorescent dye calcein-AM (1 µM) and placed at the top on the inserts. The bottom chamber contained RPMI-1640-GlutaMAX medium supplemented with 0.1% fatty acid free BSA, and/or CXCL12 (200 ng/mL) and/or 2-AG (100 nM). Since 2-AG was dissolved in DMSO, DMSO was used as vehicle control at the same final concentration. When using the 24-well plates and confocal microscopy as the read out, cells were allowed to migrate for 6 h, images of the bottom well in focus were captured every 30 min and fluorescent cells were counted using NIS-Elements AR software. When using 96-well plates and fluorescence intensity as the read out, cells were allowed to migrate for 4 h and fluorescence at 485/528 nm was measured at that time point.

Statistics
Analyses were performed using GraphPad Prism 8.3.0. Paired sample t, Mann-Whitney, Wilcoxon signed-rank, Friedman for multiple comparison and Spearman tests were used. p < 0.05 was considered significant.

Chemotaxis of Primary MCL and CLL Cells towards 2-AG
Lymphoma cells isolated from blood of patients with MCL (n = 5) and CLL (n = 22) were subjected to a chemotaxis assay. The clinical characteristics of the patients are described in Table 1. Fluorescently labelled lymphoma cells isolated from MCL ( Figure 1A) and CLL ( Figure 1B) were allowed to migrate for 4 h towards either vehicle or towards the endocannabinoid 2-AG used at the physiological concentrations found in human plasma (100 nM) [34]. Twenty out of twenty-two CLL samples (91%) migrated towards 2-AG with a range of 1.2-15-fold increase chemotaxis towards 2-AG compared to chemotaxis towards medium alone (median of 2.2 fold; p < 0.0001, Figure 1B). In MCL, 2-AG enhanced the chemotaxis of three out of five cases, with an increase of 1.4-to 3.9-fold compared to medium (median of 1.5 fold; p > 0.05, ns) ( Figure 1A). endocannabinoid 2-AG used at the physiological concentrations found in human plasma (100 nM) [34]. Twenty out of twenty-two CLL samples (91%) migrated towards 2-AG with a range of 1.2-15-fold increase chemotaxis towards 2-AG compared to chemotaxis towards medium alone (median of 2.2 fold; p < 0.0001, Figure 1B). In MCL, 2-AG enhanced the chemotaxis of three out of five cases, with an increase of 1.4-to 3.9-fold compared to medium (median of 1.5 fold; p > 0.05, ns) ( Figure 1A). normal B cells from buffy coat taken as reference in the qPCR; Mann-Whitney test was performed to compare expression between MCL and CLL samples, ns: non-significant, *** p < 0.001. (E,F) Correlation between chemotaxis towards 2-AG (relative to medium) and (E) CNR1, or (F) CNR2 expression in MCL and CLL primary samples combined with MCL cell lines; Spearman correlation test.

Cannabinoid Receptors Are Differentially Expressed in MCL and CLL
Because 2-AG is agonist to both cannabinoid receptors, we analyzed CNR1 and CNR2 expression in all samples ( Figure 1C,D). All samples but two (both from CLL patients) expressed CNR1 above the detection level. Six out of nineteen CLL had a higher CNR1 expression than normal B cells from buffy coat (median of 0.4 relative fold increase; range 0.06-8.5), while all MCL samples expressed CNR1 at higher levels than normal B cells (median of 5.6 relative fold increase; range 3.3-27.6), which was significantly higher than in CLL cells (p = 0.0007). CNR2 was expressed in all patients' samples, with a high range from 3.8 to 568.4 relative fold increased (median of 53.8). CNR2 was not expressed significantly higher in CLL cells compared to MCL cells despite the different median of 61.6 and 22.9, respectively (p = 0.08). Considering that the chemotaxis pattern towards 2-AG was also different between MCL and CLL samples, we hypothesized that this could be due to the different expression of the cannabinoid receptors. Correlation analysis showed a moderate but significant inverse correlation between the CNR1 levels and chemotaxis towards 2-AG when data from CLL and MCL were combined ( Figure 1E; r = −0.42; p = 0.03). Thus, high CNR1 expression was associated with lower migration towards 2-AG. Such correlation was not found with CNR2 gene expression ( Figure 1F). These results indicate involvement of CB1 in 2-AG mediated chemotaxis in lymphoma.

Both CB1 and CB2 Receptors Contribute to the 2-AG Mediated Chemotaxis in MCL Cell Line
In order to investigate via which cannabinoid receptor 2-AG was inducing chemotaxis, we used, as a model, the JeKo-1 MCL cell line, which expresses both CNR1 and CNR2. First, we verified that JeKo-1 cells could migrate towards 2-AG ( Figure 2A). After 4 h, the cells had a significantly increased chemotaxis index towards the same concentration of 2-AG as the primary cells (100 nM; 1.7-fold increase compared with medium alone, p = 0.03). The 2-AG induced chemotaxis was concentration dependent (Figure 2A). To elucidate whether CB1 or CB2 was mediating the migratory response to 2-AG we used two approaches: silencing the genes and targeting receptors with specific inhibitors. For silencing we used gene-specific siRNAs, then verified the efficacy of silencing by quantifying the mRNA expression with qPCR (Supplementary Figure S2B). After siRNA for CNR1, CNR1 was downregulated by 75% at 24 h and by 62% after 48 h. CNR2 downregulation was only 60% after 24 h and the levels were back to the basal levels at 48 h. Viability was not affected by the CNR-specific siRNAs (Supplementary Figure S2C). However, at the 48 h time point, downregulating CNR2 gene significantly increased CNR1 expression (p = 0.02). We performed chemotaxis assay towards 2-AG with the transfected cells and could see a tendency towards an inhibition of chemotaxis towards 2-AG after downregulation of CNR1 and CNR2 genes separately (Supplementary Figure S2D). However, these changes were not significant, perhaps due to the potential effect of gene cross-regulation.
We therefore decided to use specific antagonists for CB1 (AM6545 and SR171416, both at 10 nM) and a CB2 inverse agonist (SR144528, 10 nM). After treatment with the CB1 antagonist SR171416 or AM6545, chemotaxis towards 2-AG was not any more significantly different from chemotaxis towards medium at 4 h (p = 0.2 and p = 0.14, respectively; Figure 2B,C). After treating the cells with CB2-specific inverse agonist SR144528, the chemotaxis towards 2-AG was reduced by 50% compared with chemotaxis towards medium alone (p = 0.04; Figure 2D). However, it is important to note that SR144528 treatment increased chemotaxis towards medium alone compared with untreated cells (p = 0.008, Figure 2D). Figure 2B,C). After treating the cells with CB2-specific inverse agonist SR144528, the chemotaxis towards 2-AG was reduced by 50% compared with chemotaxis towards medium alone (p = 0.04; Figure 2D). However, it is important to note that SR144528 treatment increased chemotaxis towards medium alone compared with untreated cells (p = 0.008, Figure 2D).

Chemotaxis towards 2-AG and CXCL12 in Combination
CXCL12 and its receptor CXCR4 are among the main factors regulating lymphoma cell trafficking from blood to lymphoid tissue. Because 2-AG is also present within the microenvironment, we wondered whether 2-AG could modulate CXCL12-induced chemotaxis. We subjected MCL and CLL primary cells to chemotaxis assay towards CXCL12 alone or towards CXCL12 in combination with 2-AG ( Figure 3A,B, respectively). CXCL12 alone increased chemotaxis of all samples but two (1 MCL and 1 CLL) from 1.4-to 95-fold increase compared with medium (median of 3.5-fold increase, p < 0.01). However, when CXCL12 was combined with 2-AG in the bottom chamber, MCL and CLL cells behaved differently. In CLL, the combination CXCL12 + 2-AG increased migration compared with CXCL12 alone (median of 4.0-fold increase; p = 0.015, Figure 3B), while in MCL 3/5 samples had a reduced chemotaxis when CXCL12 was combined with 2-AG in bottom chamber (up to 50% reduction, Figure 3A).

Chemotaxis towards 2-AG and CXCL12 in Combination
CXCL12 and its receptor CXCR4 are among the main factors regulating lymphoma cell trafficking from blood to lymphoid tissue. Because 2-AG is also present within the microenvironment, we wondered whether 2-AG could modulate CXCL12-induced chemotaxis. We subjected MCL and CLL primary cells to chemotaxis assay towards CXCL12 alone or towards CXCL12 in combination with 2-AG ( Figure 3A,B, respectively). CXCL12 alone increased chemotaxis of all samples but two (1 MCL and 1 CLL) from 1.4-to 95-fold increase compared with medium (median of 3.5-fold increase, p < 0.01). However, when CXCL12 was combined with 2-AG in the bottom chamber, MCL and CLL cells behaved differently. In CLL, the combination CXCL12 + 2-AG increased migration compared with CXCL12 alone (median of 4.0-fold increase; p = 0.015, Figure 3B), while in MCL 3/5 samples had a reduced chemotaxis when CXCL12 was combined with 2-AG in bottom chamber (up to 50% reduction, Figure 3A).

Effect of 2-AG on CXCL12-Mediated Chemotaxis in MCL Cell Lines
We aimed at investigating the chemotaxis towards 2-AG and the effect on CXCL12mediated chemotaxis in JeKo-1 and two additional MCL cell lines. JeKo-1, Granta519 and JVM-2 were chosen due to their establishment from peripheral blood. They all expressed CXCR4 at the cell surface (Supplementary Figure S1C), and as previously described [35,36], they all expressed CNR2 at the mRNA level, with a higher expression in Granta519 cells ( Figure 1B). Of the three cell lines, JVM-2 had the lowest CNR1 expression, at the limit of the detection level. These cell linese also expressed the cannabinoid receptors CB1 ( Figure

Effect of 2-AG on CXCL12-Mediated Chemotaxis in MCL Cell Lines
We aimed at investigating the chemotaxis towards 2-AG and the effect on CXCL12mediated chemotaxis in JeKo-1 and two additional MCL cell lines. JeKo-1, Granta519 and JVM-2 were chosen due to their establishment from peripheral blood. They all expressed CXCR4 at the cell surface (Supplementary Figure S1C), and as previously described [35,36], they all expressed CNR2 at the mRNA level, with a higher expression in Granta519 cells ( Figure 1B). Of the three cell lines, JVM-2 had the lowest CNR1 expression, at the limit of the detection level. These cell linese also expressed the cannabinoid receptors CB1 ( Figure

Effect of 2-AG on CXCL12-Mediated Chemotaxis in MCL Cell Lines
We aimed at investigating the chemotaxis towards 2-AG and the effect on CXCL12mediated chemotaxis in JeKo-1 and two additional MCL cell lines. JeKo-1, Granta519 and JVM-2 were chosen due to their establishment from peripheral blood. They all expressed CXCR4 at the cell surface (Supplementary Figure S1C), and as previously described [35,36], they all expressed CNR2 at the mRNA level, with a higher expression in Granta519 cells ( Figure 1B). Of the three cell lines, JVM-2 had the lowest CNR1 expression, at the limit of the detection level. These cell linese also expressed the cannabinoid receptors CB1 (   MCL cell lines were then subjected to chemotaxis towards 2-AG +/− CXCL12 ( Figure 5A-C). In Figure 2A, we showed that 2-AG induced chemotaxis in JeKo-1, but to a lower level than towards CXCL12. We confirmed this data with the fluorescence intensity read out, in which chemotaxis towards CXCL12 was increased by 2.3-fold compared to vehicle control (p = 0.001, Figure 5A). Neither CB1 nor CB2 antagonists affected the chemotaxis towards CXCL12 in JeKo-1 (Supplementary Figure S3). When Granta519 cells were subjected to chemotaxis towards 2-AG, it showed a 1.5-fold increase of chemotaxis of (p = 0.03, Figure 5B), while JVM-2 did not migrate more towards 2-AG than towards medium alone (p = 0.2, Figure 5C). CXCL12 also induced chemotaxis in both Granta519 and JVM-2, by 5.4-fold (p = 0.002) and 4.6-fold (p = 0.03), respectively. towards CXCL12 in JeKo-1 (Supplementary Figure S3). When Granta519 cells were subjected to chemotaxis towards 2-AG, it showed a 1.5-fold increase of chemotaxis of (p = 0.03, Figure 5B), while JVM-2 did not migrate more towards 2-AG than towards medium alone (p = 0.2, Figure 5C). CXCL12 also induced chemotaxis in both Granta519 and JVM-2, by 5.4-fold (p = 0.002) and 4.6-fold (p = 0.03), respectively. When 2-AG was combined with CXCL12, chemotaxis of JeKo-1 and JVM-2 cells was significantly reduced compared with CXCL12 alone by 1.6-fold for JeKo-1 (p < 0.05) and 1.3-fold for JVM-2 (p = 0.007). This reduction of the CXCL12-mediated chemotaxis by 2-AG was similar to the majority of MCL primary cells. However, the chemotaxis of Granta519 towards the combination of 2-AG and CXCL12 was not significantly different from chemotaxis towards CXCL12 alone (p = 0.13), which was also the case for some of the primary samples (MCL and CLL) ( Figure S4). ; Paired t-test, ns: non-significant, * p < 0.05, ** p < 0.01, *** p < 0.001, ****p < 0.0001.

Effect of 2-AG on CXCR4 Surface Expression
One possible explanation of how 2-AG could modulate the chemotaxis to CXCL12 is to impair the ability of the CXCR4 receptor to internalize and signal upon CXCL12 binding, as described in breast and prostate cancer [37]. To investigate this possibility, we quantified CXCR4 membrane expression in MCL and CLL samples by flow cytometry analysis. There was no difference in the basal CXCR4 expression (median fluorescence intensity, MFI) between MCL and CLL (Supplementary Figure S5A). All samples expressed functional CXCR4 at the cell surface, as shown by the expression being reduced by 30 +/− 20% after 10 min incubation with CXCL12 (Supplementary Figure S5B), concordant with previous reports [37][38][39][40]. CXCR4 surface expression was also measured after 10 min incubation with 2-AG, and we concluded that 2-AG did not affect the expression of CXCR4 compared with incubation with medium alone in either disease type (p = 0.11 for MCL and p = 0.77 for CLL). Furthermore, 2-AG + CXCL12 did not significantly change CXCR4 expression as compared with CXCL12 alone (p = 0.41 for MCL and p = 0.47 for CLL; Supplementary Figure S5B). ; Paired t-test, ns: non-significant, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. When 2-AG was combined with CXCL12, chemotaxis of JeKo-1 and JVM-2 cells was significantly reduced compared with CXCL12 alone by 1.6-fold for JeKo-1 (p < 0.05) and 1.3-fold for JVM-2 (p = 0.007). This reduction of the CXCL12-mediated chemotaxis by 2-AG was similar to the majority of MCL primary cells. However, the chemotaxis of Granta519 towards the combination of 2-AG and CXCL12 was not significantly different from chemotaxis towards CXCL12 alone (p = 0.13), which was also the case for some of the primary samples (MCL and CLL) ( Figure S4).

Effect of 2-AG on CXCR4 Surface Expression
One possible explanation of how 2-AG could modulate the chemotaxis to CXCL12 is to impair the ability of the CXCR4 receptor to internalize and signal upon CXCL12 binding, as described in breast and prostate cancer [37]. To investigate this possibility, we quantified CXCR4 membrane expression in MCL and CLL samples by flow cytometry analysis. There was no difference in the basal CXCR4 expression (median fluorescence intensity, MFI) between MCL and CLL (Supplementary Figure S5A). All samples expressed functional CXCR4 at the cell surface, as shown by the expression being reduced by 30 +/− 20% after 10 min incubation with CXCL12 (Supplementary Figure S5B), concordant with previous reports [37][38][39][40]. CXCR4 surface expression was also measured after 10 min incubation with 2-AG, and we concluded that 2-AG did not affect the expression of CXCR4 compared with incubation with medium alone in either disease type (p = 0.11 for MCL and p = 0.77 for CLL). Furthermore, 2-AG + CXCL12 did not significantly change CXCR4 expression as compared with CXCL12 alone (p = 0.41 for MCL and p = 0.47 for CLL; Supplementary Figure S5B).
In addition, we measured the CXCR4 surface expression in JeKo-1 cell line after 10 min incubation ( Figure 6A). We did not observe any difference in the CXCR4 expression at the cell surface upon pre-incubation with CXCL12 alone or in combination with 2-AG. We also investigated the re-expression of the receptor after CXCL12 treatment (recycling of CXCR4) in JeKo-1 cell line ( Figure 6B). This was done by first incubating the cells with CXCL12 for 10 min, then washing the cells and measuring the CXCR4 expression at 30 min and 1 h after incubating the cells in medium alone. We could first confirm that the CXCR4 expression increased after washing off, and that this re-expression was not significantly different in cells exposed to the combination of CXCL12 + 2-AG. Lastly, we verified that the CXCR4 at the cell surface was not differentially expressed in migrated cells harvested after 4 h chemotaxis ( Figure 6C). Altogether, the data suggests that the inhibition of CXCL12mediated chemotaxis by 2-AG is not due to an impaired internalization or re-expression of the CXCR4 receptor.
In addition, we measured the CXCR4 surface expression in JeKo-1 cell line aft min incubation ( Figure 6A). We did not observe any difference in the CXCR4 expre at the cell surface upon pre-incubation with CXCL12 alone or in combination with 2 We also investigated the re-expression of the receptor after CXCL12 treatment (recy of CXCR4) in JeKo-1 cell line ( Figure 6B). This was done by first incubating the cells CXCL12 for 10 min, then washing the cells and measuring the CXCR4 expression min and 1 h after incubating the cells in medium alone. We could first confirm tha CXCR4 expression increased after washing off, and that this re-expression was not s icantly different in cells exposed to the combination of CXCL12 + 2-AG. Lastly, we ve that the CXCR4 at the cell surface was not differentially expressed in migrated cells vested after 4 h chemotaxis ( Figure 6C). Altogether, the data suggests that the inhib of CXCL12-mediated chemotaxis by 2-AG is not due to an impaired internalization expression of the CXCR4 receptor.
Inhibition of the phosphorylation of Akt was reversed after CB1 antagonist SR14 and CB2 inverse agonist SR144528 ( Figure 8A), while only CB1 antagonist SR14171 stored fully p38 phosphorylation ( Figure 8B). 2-AG did not change phosphorylation s in Granta519, although it induced 1.5-fold increased chemotaxis compared to me alone ( Figure 5B). 2-AG did not significantly change the phosphorylation status of
Inhibition of the phosphorylation of Akt was reversed after CB1 antagonist SR141716 and CB2 inverse agonist SR144528 ( Figure 8A), while only CB1 antagonist SR141716 restored fully p38 phosphorylation ( Figure 8B). 2-AG did not change phosphorylation status in Granta519, although it induced 1.5-fold increased chemotaxis compared to medium alone ( Figure 5B). 2-AG did not significantly change the phosphorylation status of these three investigated pathways in JVM-2 ( Figure 7C), concordant with the chemotaxis data that was unchanged towards 2-AG alone in this cell line ( Figure 5C).
The combination of 2-AG and CXCL12 had a contrasting effect on the phosphorylation status of ERK1/2, in JeKo-1 and JVM-2. In JeKo-1, ERK1/2 phosphorylation significantly changed after incubation with the two compounds together (1.8-fold decrease, p = 0.04, Figure 7A), while in JVM-2, 2-AG combined with CXCL12 significantly increased the phosphorylation status of ERK1/2 (2-fold increase, p = 0.01; Figure 7C) and decreased p38 phosphorylation (1.4-fold decrease, p = 0.04) compared with CXCL12 alone. The combination of CXCL12 and 2-AG seemed to increase the phosphorylation of Akt pathway in Jeko-1 and JVM-2; however, this was not statistically significant. The activation of the signaling pathways in Granta519 were fluctuating and did not show any significant differences between the different conditions. that was unchanged towards 2-AG alone in this cell line ( Figure 5C).
The combination of 2-AG and CXCL12 had a contrasting effect on the phosphorylation status of ERK1/2, in JeKo-1 and JVM-2. In JeKo-1, ERK1/2 phosphorylation significantly changed after incubation with the two compounds together (1.8-fold decrease, p = 0.04, Figure 7A), while in JVM-2, 2-AG combined with CXCL12 significantly increased the phosphorylation status of ERK1/2 (2-fold increase, p = 0.01; Figure 7C) and decreased p38 phosphorylation (1.4-fold decrease, p = 0.04) compared with CXCL12 alone. The combination of CXCL12 and 2-AG seemed to increase the phosphorylation of Akt pathway in Jeko-1 and JVM-2; however, this was not statistically significant. The activation of the signaling pathways in Granta519 were fluctuating and did not show any significant differences between the different conditions.  ng/mL) or 2-AG (100 nM) or the combination, and the activation of the signaling pathways ERK1/2, Akt and p38 was assessed by Western blotting, normalizing phospho/GAPDH band intensity to total/GAPDH as described in Section 2, bars represent an average of the ratio of at least four repeats and error bars are standard error of the mean; paired t-test, * p < 0.05, ** p < 0.01, *** p < 0.001, ns: non-significant, a representative blot for each experiment is shown. Original blot see Supplementary File S1. signaling pathways ERK1/2, Akt and p38 was assessed by Western blotting, normalizing phospho/GAPDH band intensity to total/GAPDH as described in Section 2, bars represent an average of the ratio of at least four repeats and error bars are standard error of the mean; paired t-test, * p < 0.05, ** p < 0.01, *** p < 0.001, ns: non-significant, a representative blot for each experiment is shown. Original blot see Supplementary File S1. Figure 8. Effects of inhibiting the cannabinoid receptors on Akt and p38 activation. JeKo-1 cells were treated for 20 min with the CB1 antagonist SR141716 (10 nM), or with the CB2 inverse agonist SR1445282 (10 nM), prior to be incubated for 2 min with 2-AG. Activation of the signaling pathways (A) Akt and (B) p38 was assessed by Western blotting, normalizing phospho/GAPDH band intensity to total/GAPDH as described in Section 2, bars represent an average of the ratio of three repeats and error bars are standard error of the mean; 2-AG condition is shown as normalized to incubation with vehicle set as 1, unpaired t-test with Welch's correction, # p < 0.05.
In addition, we could investigate the activation of the pathways ERK1/2 and p38 in two CLL samples that displayed different chemotaxis patterns towards the combination CXCL12 + 2-AG ( Figure 9). Both samples displayed an increased chemotaxis towards 2-AG and towards CXCL12 compared with medium alone. However, sample 40 displayed an increased chemotaxis towards the combination of 2-AG and CXCL12 compared to CXCL12 alone (3-fold increase), while sample 41 had a decreased chemotaxis towards the combination (2-fold decrease). Both samples expressed CNR1 and CNR2 ( Figure 1C,D), with sample 40 expressing CNR1 at lower levels than B cells from buffy coat and sample 41 expressing CNR1 at a 2.8-fold increase compared to buffy coat B cells. The activation of the signaling pathways seems to be affected by the combination of 2-AG and CXCL12 in the case of Sample 40, but not in sample 41. Figure 8. Effects of inhibiting the cannabinoid receptors on Akt and p38 activation. JeKo-1 cells were treated for 20 min with the CB1 antagonist SR141716 (10 nM), or with the CB2 inverse agonist SR1445282 (10 nM), prior to be incubated for 2 min with 2-AG. Activation of the signaling pathways (A) Akt and (B) p38 was assessed by Western blotting, normalizing phospho/GAPDH band intensity to total/GAPDH as described in Section 2, bars represent an average of the ratio of three repeats and error bars are standard error of the mean; 2-AG condition is shown as normalized to incubation with vehicle set as 1, unpaired t-test with Welch's correction, # p < 0.05.
In addition, we could investigate the activation of the pathways ERK1/2 and p38 in two CLL samples that displayed different chemotaxis patterns towards the combination CXCL12 + 2-AG ( Figure 9). Both samples displayed an increased chemotaxis towards 2-AG and towards CXCL12 compared with medium alone. However, sample 40 displayed an increased chemotaxis towards the combination of 2-AG and CXCL12 compared to CXCL12 alone (3-fold increase), while sample 41 had a decreased chemotaxis towards the combination (2-fold decrease). Both samples expressed CNR1 and CNR2 ( Figure 1C,D), with sample 40 expressing CNR1 at lower levels than B cells from buffy coat and sample 41 expressing CNR1 at a 2.8-fold increase compared to buffy coat B cells. The activation of the signaling pathways seems to be affected by the combination of 2-AG and CXCL12 in the case of Sample 40, but not in sample 41.
signaling pathways ERK1/2, Akt and p38 was assessed by Western blotting, normalizing phospho/GAPDH band intensity to total/GAPDH as described in Section 2, bars represent an average of the ratio of at least four repeats and error bars are standard error of the mean; paired t-test, * p < 0.05, ** p < 0.01, *** p < 0.001, ns: non-significant, a representative blot for each experiment is shown. Original blot see Supplementary File S1. Figure 8. Effects of inhibiting the cannabinoid receptors on Akt and p38 activation. JeKo-1 cells were treated for 20 min with the CB1 antagonist SR141716 (10 nM), or with the CB2 inverse agonist SR1445282 (10 nM), prior to be incubated for 2 min with 2-AG. Activation of the signaling pathways (A) Akt and (B) p38 was assessed by Western blotting, normalizing phospho/GAPDH band intensity to total/GAPDH as described in Section 2, bars represent an average of the ratio of three repeats and error bars are standard error of the mean; 2-AG condition is shown as normalized to incubation with vehicle set as 1, unpaired t-test with Welch's correction, # p < 0.05.
In addition, we could investigate the activation of the pathways ERK1/2 and p38 in two CLL samples that displayed different chemotaxis patterns towards the combination CXCL12 + 2-AG ( Figure 9). Both samples displayed an increased chemotaxis towards 2-AG and towards CXCL12 compared with medium alone. However, sample 40 displayed an increased chemotaxis towards the combination of 2-AG and CXCL12 compared to CXCL12 alone (3-fold increase), while sample 41 had a decreased chemotaxis towards the combination (2-fold decrease). Both samples expressed CNR1 and CNR2 ( Figure 1C,D), with sample 40 expressing CNR1 at lower levels than B cells from buffy coat and sample 41 expressing CNR1 at a 2.8-fold increase compared to buffy coat B cells. The activation of the signaling pathways seems to be affected by the combination of 2-AG and CXCL12 in the case of Sample 40, but not in sample 41.

Discussion
In this study we demonstrate that 2-AG induces chemotaxis of MCL and CLL primary cells isolated from blood, and in the MCL cell lines JeKo-1 and Granta519, however, to a lesser extent than CXCL12. Other reports also showed that 2-AG induces chemotaxis in different cell types, such as NK cells [43], monocytes [44] and eosinophils [45] from peripheral blood, as well as CD34+ hematopoietic stem cells [27] and mesenchymal stromal cells [46], from bone marrow. 2-AG was also reported to increase the motility of the T cell line Jurkat in a co-culture system with stromal cells pre-incubated with 2-AG [47]. In mice, 2-AG has a role in the innate immune response by increasing the recruitment of dendritic cells to the lymph node upon immunization [48] and has a different chemotactic potency according to the activation state of B lymphocytes [49]. Importantly, the cells used in these previous reports express mainly CB2 receptor but none or low levels of CB1. Our study is the first to scrutinize this in primary lymphoma cells, which express both cannabinoid receptor types at higher levels than normal B lymphocytes. Both CB1 and CB2, demonstrated to be expressed at the protein level in MCL cell lines, are involved in the chemotaxis towards 2-AG as shown by using specific CB1 and CB2 inhibitors ( Figure 1). In our hands, silencing the genes was not a useful tool as downregulation of CNR2 induced a change in CNR1 gene expression, which suggests a crosstalk between the regulation of these two genes. We therefore chose to use specific antagonists to CB1 and CB2 which demonstrated that both receptors are involved in 2AG induced chemotaxis. We also show that the selective CB2 inverse agonist SR144528 increases the basal chemotaxis of B lymphocytes. The fact that blocking the CB2 receptor induces spontaneous chemotaxis is concordant with previous reports that show the importance of CB2 in the retention of the cells in lymphoid compartments [20][21][22].
Endocannabinoids, chemokines and adhesion molecules participate in the communication between the tumor cells and their microenvironment. Each of these have specific functions. CXCR4 and CXCR5 are implicated in the homing of cells into lymphoid tissues where their respective ligands CXCL12 and CXCL13 are found in high concentrations [50,51]. Within the tissue, adhesion molecules are involved in the engraftment and migration of the tumor cells beneath the stromal cells [1], enhancing the risk of drug resistance and relapses. Endocannabinoids are released in the bone marrow microenvironment by mesenchymal cells [27], including stromal cells and osteoblasts, which might influence the response of T lymphocytes in that microenvironment [46,51]. In the blood circulation, chemokines are present in lower concentration than in lymphoid tissues, but 2-AG is elevated in the plasma of cancer patients [25,26]. In this context, we investigated whether the combination of 2-AG and CXCL12 would have a synergistic effect on chemotaxis, as some cytokines, such as interleukin-5 and interleukin-3, do when combined with 2-AG in hematopoietic cells, B lymphocytes and neutrophils [52][53][54]. We show here that the combination of 2-AG and CXCL12 had a different effect on chemotaxis in CLL than in MCL as it reduced chemotaxis in most MCL cases and increased it in CLL. A previous report from Coopman et al. has also investigated the combination of CXCL12 and 2-AG. Although their experimental set-up was different from ours, as they used T lymphocytes that were incubated with 2-AG prior to subjecting them to chemotaxis towards CXCL12, a similar conclusion as for MCL cells from this study was reached, although T cells do not express CB1 receptor [55].
CB2 has been described to heterodimerize with the CXCR4 receptor in breast cancer and prostate cancer cells when both receptors are simultaneously activated by respective agonists. This heterodimerization blocks CXCL12-induced CXCR4 internalization and downstream signaling [37]. Our results suggest different mechanisms behind the effects of 2-AG on CXCL12-induced migration in lymphoma cells. In contrast to findings described in breast cancer, in MCL and CLL cells CXCR4 was still internalized even when CXCR4, CB1 and CB2 were stimulated simultaneously by their endogenous ligands. The fact that the CXCR4 receptor could still internalize upon activation and that the recycling of CXCR4 was not affected by incubation with 2-AG or antagonists to CB1 or CB2, indicates that the reduced migration is rather caused by alteration in the downstream signaling or by the activation of alternative pathways than by affecting CXCR4 receptor functionality and localization.
In our hands, 2-AG did not induce activation of ERK1/2 pathway in lymphoma cells, in contrast to reports on other cell types such as CB2-transfected CHO cells and CB1 and CB2 expressing microglial cells [47,56,57]. This could be due to different cell types studied and/or due to different concentrations of 2-AG used in our study and studies by others. We used 100 nM, which is comparable to the physiological concentration in blood and tissues [25,27], similarly used in a study by Wang et al., in which low 2-AG concentrations in astrocytes do not induce ERK1/2 activation [58] but induced chemotaxis. In fact, it seems that ERK1/2 activation requires micromolar (µM) of 2-AG, as shown in T lymphocytes [55]. However, Akt phosphorylation was found to be inhibited with the presence of 2-AG, which was reverted with CB1 and CB2 inhibitors. Similar results have been described in B-acute lymphoblastic leukemia cells after CB2 stimulation with the specific agonist JWH-133 [59]. The effects of 2-AG on CXCL12-mediated chemotaxis and on the downstream signaling pathways are different in the three MCL cell lines used in this study, reflecting the variation in chemotaxis in primary samples. However, it seems that for JeKo-1 and JVM-2, 2-AG is affecting CXCL12-mediated ERK1/2 phosphorylation, which could explain the inhibitory effect on the chemotaxis ( Figure 4B,C). The fact that JeKo-1 and JVM-2 had a different pattern on the ERK1/2 and p38 signaling after the combination of 2-AG and CXCL12 could be due to the different expression of the cannabinoid receptors, as suggested by Kleyer et al. and Callén et al. [60,61]. CNR1 mRNA and its protein expression are very low/negative in JVM-2, and we see less effect on the chemotaxis towards the combination of CXCL12 and 2-AG in that cell line compared with JeKo-1, which has higher CNR1 mRNA and protein levels. We hypothesize that both receptors are implicated in the chemotaxis response to the combination of CXCL12 and 2AG. This could be one explanation for the difference in 2-AG modulation of CXCL12 response between JeKo-1 and JVM-2 regarding pERK signaling. We could then further hypothesize that the differences in clinical presentation between MCL (being mainly restricted to lymph nodes and bone marrow) and CLL (leukemic disease) could in part be due to their differential expression of CNR1 and CNR2.

Conclusions
Here, we describe a novel player in lymphoma biology. 2-AG on its own is inducing the chemotaxis of most lymphoma cells and it is modulating the CXCL12-mediated chemotaxis that, based on our findings, is dependent on CB1 and CB2 expression levels. These novel findings in hematological malignancies strengthen the potential role of the endocannabinoid system as playing an important role in the positioning of lymphoma cells in tissues and their release in the blood circulation. Our results suggest that, together with other chemokines, 2-AG is involved in the regulation of circulating cells in the blood. Further research is, however, still needed to confirm that targeting 2-AG synthesis in lymphoma would be a potential therapeutic strategy.