Bladder cancer cell growth and motility implicate cannabinoid 2 receptor-mediated modifications of sphingolipids metabolism

The inhibitory effects demonstrated by activation of cannabinoid receptors (CB) on cancer proliferation and migration may also play critical roles in controlling bladder cancer (BC). CB expression on human normal and BC specimens was tested by immunohistochemistry. Human BC cells RT4 and RT112 were challenged with CB agonists and assessed for proliferation, apoptosis, and motility. Cellular sphingolipids (SL) constitution and metabolism were evaluated after metabolic labelling. CB1-2 were detected in BC specimens, but only CB2 was more expressed in the tumour. Both cell lines expressed similar CB2. Exposure to CB2 agonists inhibited BC growth, down-modulated Akt, induced caspase 3-activation and modified SL metabolism. Baseline SL analysis in cell lines showed differences linked to unique migratory behaviours and cytoskeletal re-arrangements. CB2 activation changed the SL composition of more aggressive RT112 cells by reducing (p < 0.01) Gb3 ganglioside (−50 ± 3%) and sphingosine 1-phosphate (S1P, −40 ± 4%), which ended up to reduction in cell motility (−46 ± 5%) with inhibition of p-SRC. CB2-selective antagonists, gene silencing and an inhibitor of SL biosynthesis partially prevented CB2 agonist-induced effects on cell viability and motility. CB2 activation led to ceramide-mediated BC cell apoptosis independently of SL constitutive composition, which instead was modulated by CB2 agonists to reduce cell motility.


Orthotopic bladder cancer model
The orthotopic animal model was performed as follows: Luciferase-expressing RT112 (RT112-luc) cells were generated by using ready-to-use lentiviral particles (Redifect, Perkin Elmer). Immunodeficient nude mice were then anesthetized, laparatomized to expose the bladder, and injected transmurally with RT112-luc cells (2.5x10 5 cells in PBS/matrigel 1:2, 50 L)Three days post tumor implantation, 15 min after luciferase substrate administration, animals underwent to in vivo imaging for assessing tumor take and started intravesical instillation of therapy. Drug (JWH015,20 µM) or vehicle (PBS) instillations were repeated at day 7, and 11 and 15 post tumor injection. Mice were carefully monitored daily for health status and hematuria and tumor burden was observed by luminescence scan at day 10, 28, and 48 post tumor implantation. Animals were terminally euthanized to allow bladder histopathological evaluation.
SUPPLEMENTAL RESULTS:

3.
Genomic and bioinformatics analysis of CB2 in urothelial bladder cancer.
Genetic profiling of CB2 (CNR2) was carried out on the TCGA data sets encompassing either 413 cases (http://bit.ly/2fDFqSi) or 131 samples (http://bit.ly/2fZzrvs) of urothelial bladder cancer by using the cBioPortal analysis tools 1 . The CNR2 gene was found to be mutated in 3 patients (0.7%), whereas an alteration of the CB2 copy number (gain, amplification, and shallow deletion) was present in 154 samples (35.6%), with the majority of them being T3-T4 stage. No clear correlation was found between CNR2 expression levels and either survival or diagnosis subtype (papillary or non-papillary). Finally, by evaluating the expression of CNR2 gene within the recently identified mRNA cluster subtypes in BC 2 , in all the cases available in the public domain, we found that higher CNR2 mRNA expression significantly associated with cluster mRNA subtype IV (49.1±30 expr. units, n=13, p<0.05 1-way ANOVA with Tuckey post-test vs cluster 1: 0.81±0.4, 2: 4.6±1.8, and 3: 1.6±0.7). Additionally, we found weak but significant positive correlation between CNR2 gene expression and miR-100-5p (Pearson 0.22, p=0.03) and an anti-correlation with E-Cadherin protein expression (Pearson -0.25, p=0.01), both features of type IV cluster. Further cases on mRNA clustering are not presently available, as confirmed by personal communication from TCGA group.

Relative contribution of CB1 and CB2 to agonist-induced cytotoxicity
A dual CB1-CB2 agonist (anandamide, AEA) was used. As expected, AEA (20 μM) displayed a cytotoxic effect that peaked 48 hrs after exposure. Blockade of CBs with rimonabant (CB1 antagonist) or SR144528 (CB2 antagonist) in AEA-treated RT112 cells suggested that both receptors contributed to the early CB agonist-induced cytotoxic effects, since both CB1 and CB2 antagonists partially rescued cell viability to 50% 24 hrs after AEA exposure. Only CB2 receptor blockade, however, prevented cell death at longer time points, with a 75% increase of cell viability measured 48 hrs after AEA. Cells exposed to AEA spontaneously regained normal proliferation activity after 5 days (data not shown). Moreover, when cells were co-treated with AEA plus oleoyl ethyl amide (OEtA), a specific blocker of FAAH enzyme, the cytotoxic effect of AEA was potentiated (data not shown), suggesting that the reduced effects of AEA measured 5 days after exposure were mainly due to its degradation.

5.
In vivo effect of CB2 agonist instilled othotopically in xenotrasplanted immunodeficient mice The antitumor effect of cannabinoid agonist JWH015 was tested in vivo in mice xenotrasplanted with RT112-luc cells, implanted intramurally. Four intravesical instillation of JWH015 (20 M), carried out 4 days apart, significantly reduced the tumor burden 24 and 48 days after tumor implantation. JWH015 treatment, however, did not completely eliminate the tumor, which 33 days after last intravesical instillation, displayed re-growth (Fig. s4A). The histological image of the bladder (Fig. s4C) shows a proliferating neoplastic focus within the muscle layer thickness.

6.
The apoptotic phenotype of BC cells after CB2 agonist exposure
Cannabinoids can promote cell death in tumours by inducing autophagy and modulating membrane sphingolipids 3,4 . JWH015 induced conversion from the soluble form of LC3 (LC3-I) to autophagosome-associated form (LC3-II) and mediated inhibition of mTORc1, evaluated as reduced phosphorylation of p70S6, was detected already 6 hrs after agonist stimulation and counteracted by the autophagosome fusion inhibitor bafilomycin. The data suggest that some molecular events, such as mTORC1 inhibition, predisposing the autophagy machinery, are occurring before the cannabinoid-mediated caspase 3 activation, confirming that cellular and possibly defective autophagy, is preceding the JWH015-induced cytotoxicity of BC cells.
Suppl. Fig. s6. Autophagy occurring in RT112 upon exposure to CB2 agonist. 8. Sphingolipids composition of RT4 and RT112 cells before and after CB2 agonist challenge

Suppl. Fig. s7. Representative TLC showing lipid composition of BCa cells before and after JWH015 treatment.
After metabolic labeling with 2h pulse of [1-3H]sphingosine, RT4 (lanes 1-2) and RT112 (lanes 3-4) cells were chased with vehicle (lanes 1 and 3) or 20μM JWH015 for 24 h (lanes 2 and 4). Radioactive lipids were identified by using the appropriate standards: ceramide (Cer), glucosylceramide (GlcCer), phosphoethanolamine (PE), globotriaosylceramide (Gb3), sphingomyelin (SM) 9. Functional phenotype of low-(RT4) and high-(RT112) aggressive BC cell lines As shown in suppl. Fig. S7, the levels of the globotriesosylceramide, Gb3, a cell motility-promoting sphingolipid, were 15-fold higher (p<0.001) in RT112 than in RT4 cells. Conversely, RT4 cells showed 7-fold higher levels (p< 0.01) of glucosylceramide (GlcCer). We also detected increased amounts of gangliosides typically associated with reduced invasiveness, particularly GM2 and GM3, in RT4 cells. The SL profile was also evaluated in T24, HT1376, and 5637 BC cell lines and found similar to RT112 (not shown). The different SL pattern between RT112 and RT4 cell lines were, in turn, associated to different cellular motility, which was independently of their comparable proliferation rates (not shown). In a wound-healing assay, RT112 cells reconstituted more than 70% of the scratched area within 48 hrs (suppl. Fig.8 B-C), while RT4 cells only populated 5%. The metastatic features of RT112 cells were also markedly different from the RT4 cells, as demonstrated by a classical matrigel-based invasion assay (suppl. Fig.8D). In addition to these evidences, along with the recent findings demonstrating the strong link between plasma membrane SL and the activation of ERM proteins in driving cancer progression mechanisms 27,28 , we showed that the levels of phospho-ERM detected in RT112 cells were significantly higher than RT4 (suppl fig Fig.8E), suggesting a cytoskeletal rearrangement in high-grade BC cells. Previous data 5 showed that the phosphorylation level of ERM was strictly linked to the level of S1P. Consistently with this, the content of S1P we measured in high-motility RT112 cells was 3-fold higher than the one in low-motility RT4 cells (3.8±0.56 vs 1.43±0.36 nCi mg -1; p<0.01). Furthermore, the sphingosine kinase 1 (SK1) was upregulated in both gene expression (5-fold, not shown) and enzymatic activity (891±94 vs 283±71 pmoles hour -1 mg -1; p<0.01) in RT112 cells compared to RT4 cells.
Finally, in order to directly relate the SL metabolism with cell motility, RT112 cells were pretreated with an inhibitor of SL synthesis (AMP-dNM), before measuring the cell migration in the scratch assay. As shown in suppl. Fig.8F, AMP-dNM was able to significantly reduce the motility of RT112 cells.