Introduction

T-cell acute lymphoblastic leukemia (T-ALL) is an aggressive cancer, characterized by immature T-lymphoblasts (1, 2). In over 50% of cases, activating mutations in NOTCH1 were identified, generating much interest in targeting the Notch1 signaling pathway in this disease (3–5). The Notch1 protein is a single-pass transmembrane receptor that is expressed on the plasma membrane. It regulates evolutionarily conserved signaling that controls developmental processes, cell fate determination and tissue homeostasis. It is a member of the Notch receptor family, which consists of four receptors (Notch1-4) in humans (4, 6). The Notch1 receptor-signaling pathway is activated through a series of proteolytic cleavages at three different sites (S1–S3). S1 cleavage is carried out by a Furin-like convertase in the trans-Golgi apparatus, resulting in a heterodimeric receptor that is transported to the membrane where it can bind ligands from a neighboring cell. Upon binding of a ligand, cleavage is performed at the next site and the activation of the Notch1 receptor-signaling pathway is initiated, enabling the cleavage of S3 in the transmembrane region that causes the release of the Notch1 intracellular domain (NICD), which translocates to the nucleus where it promotes transcription of target genes involved in cell growth (7).

Aberrant Notch1 activity is associated with many other leukemia types and other cancers, including prostate, lung, ovarian, and renal (8–12). A new and promising emerging field in cancer treatment is medical Cannabis and its unique active compounds, the phytocannabinoids. Phytocannabinoids affect the body by means of the endocannabinoid system (eCBS), a ubiquitous neuromodulatory signaling system that has widespread functions throughout the body. Notable eCBS receptors include cannabinoid receptor type 1 (CB1) and CB2, which belong to the family of G protein-coupled receptors, and transient receptor potential vanilloid type 1 (TRPV1), which is an ion channel (13). The eCBS participates in many physiological activities, including the regulation of Ca²⁺ homeostasis (14–16).

Medical Cannabis is already commonly used by cancer patients for its palliative effects, it was shown to stimulate appetite while alleviating symptoms such as nausea, vomiting and pain (17). However, accumulating evidence demonstrated phytocannabinoids directly affect tumor development in cell lines and animal models by modulating key cell-signaling pathways (18–20). Studies demonstrated that purified phytocannabinoids can inhibit proliferation, metastasis, and angiogenesis and exert pro-apoptotic effects in a variety of cancer cell types such as lung, breast, prostate, skin, intestine, glioma, lymphoma, pancreas and uterine cancers (21–23). In preclinical models, treatment with phytocannabinoids led to tumor regression of different cancer types (24, 25).

Concerning T-ALL, some studies using phytocannabinoids and their respective formulations (such as Dronabinol) have shown anti-proliferative as well as proapoptotic effects in selected leukemia cell lines and native leukemia blasts cultured ex-vivo (26–28). Nevertheless, the mechanisms of phytocannabinoid-mediated antitumor effects are not yet fully understood. Most of the studies that examined the anti-cancer effects of Cannabis have focused mainly on the two major phytocannabinoids (−)-trans-Δ⁹- tetrahydrocannabinol (Δ⁹-THC) and Cannabidiol (CBD), but these are just two of more than 140 different phytocannabinoids that have been identified in different Cannabis chemovars (29, 30).

We have previously demonstrated that different Cannabis extracts with unique phytocannabinoid compositions impaired the survival and proliferation of specific cancer cell lines, suggesting that the effect of a particular Cannabis extract on a specific cancer cell line relies on its chemical composition (31). Following this framework, we successfully matched a specific CBD-rich Cannabis extract to T-ALL leukemia cells that harbor a NOTCH1 mutation (32). In the present study, we investigated which specific metabolites in the whole extract are responsible for T-ALL elimination, and elucidated the molecular mechanism by which they mediate their antitumor effects in this type of cancer.

Results

We have previously shown a specific CBD-rich Cannabis extract selectively induces apoptosis in T-ALL leukemia cell lines that harbor a NOTCH1 mutation. Treatment led to accumulation of the full-length immature form of Notch1 in the membrane, correspondingly with reduced NICD protein expression and transcription activity (32).

A combination of three phytocannabinoids induces apoptosis and reduces NICD expression

The CBD-rich extract was more potent than pure CBD or other high-CBD chemovars (32). This implied a specific combination of metabolites from the plant potentiate the full antitumor effect. Using a semi-preparative HPLC, we separated the whole extract into four 10-minute retention time fractions (Fig. 1A). A full list of the phytocannabinoids constituting each fraction is presented in supplementary table S1. We assessed the effect of each fraction on the viability of MOLT-4 cells (fig. S1A) and found fraction 2 reduced viability to a similar extent as the whole extract. Further analysis revealed that out of the four fractions, only fraction 2 induced apoptotic cell death (Fig. 1B), reduced the protein levels of the NICD (Fig. 1C, D) and elevated the levels of cleaved caspase-3 (Fig. 1D). Thus, we excluded all the phytocannabinoids present in the other three fractions, such as Δ⁹-THC in fraction 3.

Fig. 1.

To pinpoint the active phytocannabinoids in fraction 2, we further separated this fraction into individual phytocannabinoids using centrifugal partition chromatography (CPC) followed by semi-preparative HPLC (C1-C5 in Fig. 1E). We identified the specific compounds in the isolated peaks (Supplementary table S2), C2 was identified as cannabidivarin (CBDV) and C5 as CBD. C3 and C4 were identified as Cannabidiolic acid (CBDA), the precursor to CBD, and as CBD-C4, respectively; but their concentrations were very low. C1 was identified as 331-18A, a phytocannabinoid first identified by our team in decarboxylated CBD-rich Cannabis chemovars (29), we characterized 331-18A using ESI-LC/MS/MS as having a deprotonated m/z of 331.2279 and MS/MS fragments resembling those of CBD with the addition of one hydroxyl group. To confirm the identification, the purity of the three isolated compounds was determined by UHPLC/UV (fig. S2A-C).

Here, we further elucidated the chemical structure of 331-18A by nuclear magnetic resonance (NMR) with CBD as the reference (Supplementary Fig. S3 and table S3). The NMR data of CBD was in close agreement with the literature (33). The spectra of CBD and 331-18A were very similar with the exception of the disappearance of the external double bond (C8-C10) and the appearance of a tertiary alcohol instead. The absolute stereochemistry of 331-18A was confirmed by specific rotation determination with a polarimeter in which 331-18A showed [α]²⁰D −56.85 which is comparable with the [α]²⁰D −64.16 of both the CBD that was isolated from the fraction (C5) and a commercial standard of CBD. These results suggested that similarly to CBD, the two substituents at positions 1 and 6 of the cyclohexene ring are in a trans configuration (Fig. 1F).

To examine the cytotoxic properties of the isolated compounds and their combinations on MOLT-4 cells, the isolated phytocannabinoids were prepared in the same amounts and proportion as observed in the whole extract; for example, if 1 μg of the whole extract contained 50% CBD, we used 0.5 μg of the purified CBD and 1 μg of the whole extract in the respective treatments. Compared to the whole extract, the phytocannabinoids 331-18A and CBDV alone did not induce apoptosis or had minor effects on the cells, and CBD alone induced only 55% cell death compared to the whole extract (Fig. 1G). When CBD was combined with either 331-18A or CBDV, the percent of apoptotic cells increased to 58% and 61%, respectively. The combination of all three phytocannabinoids led to the greatest cytotoxic effect and only this combination was comparable to the whole extract (88.29% ± 4.31% and 93.85% ± 1.64%, respectively). We further evaluated the effect of the three phytocannabinoids separately or in different combinations on the protein expression of NICD (Fig. 1H-I) and found that all combinations of phytocannabinoids led to a significant reduction in NICD expression, but only when all the three were combined, NICD levels decreased to the same extent as with the whole extract.

331-18A, CBDV and CBD mediate their effect through CB2 and TRPV1 to trigger Ca²⁺ flux

Phytocannabinoids usually operate through specific cannabinoid receptors in the eCBS. To elucidate which endocannabinoid receptors mediate the antitumor effect, we first evaluated in MOLT-4 cells the mRNA expression of CNR1 and CNR2 (for CB1 and CB2, respectively), as well as G protein-coupled receptor 55 (GPR55), TRPV1, Transient receptor potential cation channel, subfamily A, member 1 (TRPA1) and Transient receptor potential cation channel subfamily M (melastatin) member 8 (TRPM8), (Fig. 2A). The ΔCT results showed low expression for CNR1, TRPA1 and TRPM8, medium expression for CNR2 and GPR55, and high expression for TRPV1. We tested antagonists for those receptors that had either high or medium expression: TRPV1 (AMG9810), CB2 (AM630 or SR-144,528), GPR55 (CID), and general Gq GPCR (BIM-46187) as a control, and evaluated whether they were able to rescue NICD expression upon treatment with the whole extract (fig. S4A-B). Only the CB2 and TRPV1 antagonists rescued the expression of NICD upon treatment with the whole extract, therefore we tested whether they were able to rescue NICD expression after treatment with the combination of 331-18A, CBDV and CBD (Fig. 2B-C). The CB2 antagonist rescued the expression from 56% to 87%, and the TRPV1 antagonist rescued NICD expression to 100% by itself.

Fig. 2.

TRPV1 is a Ca²⁺ channel and eCBS receptors are known to trigger Ca²⁺ flux (15, 16). Therefore, as the next step we tested the effect of the three cannabinoids on cytosolic Ca²⁺ levels. We used the Fluo-4 calcium probe to measure the effect of 331-18A, CBDV and CBD combination on MOLT-4. Each of the molecules separately caused some elevation in Ca²⁺, but only their combination had an effect on Ca²⁺ to the same extent as the whole extract (Fig. 2D). We tested Ca²⁺ levels when cells were pretreated with the antagonists to CB2 and TRPV1. Both antagonists inhibited the elevation in response to the whole extract (Supplementary fig. S3C, D). When 331-18A, CBDV and CBD were tested separately with antagonists to either CB2 or TRPV1 (Fig. 2E), CB2 but not TRPV1 antagonist diminished 331-18A induced Ca²⁺ increase, indicating this cannabinoid exerts its effect through CB2, while both antagonists significantly decreased the Ca²⁺ elevation by CBDV and CBD. Next, we assessed whether inhibition of the receptors rescues MOLT-4 cells from the cytotoxic properties of the cannabinoid combination (Fig. 2F). Treatment with the inhibitors by themselves, separately or both together, did not induce MOLT-4 cell death. Upon treatment with the combination of 331-18A, CBDV and CBD, inhibition of CB2 only partially rescued the cells from treatment-induced apoptosis, reducing the apoptotic rate from 75% of the combination-treated cells to 40%. However, the apoptotic rate was still significantly higher than that of the untreated control. Inhibition of TRPV1 only also partially rescued the cells, to ∼50%, but again the apoptotic rate remained significantly increased relative to the untreated control. Inhibition of both CB2 and TRPV1 simultaneously rescued the cells from treatment-induced apoptosis, confirming the effect of the three cannabinoids is mediated through both receptors.

Notch1 downregulation by cannabinoids is mediated through the integrated stress response pathway

To elucidate the mechanism by which the Notch1 pathway is down-regulated following treatment, we used RNA-seq analysis to measure and compare the gene expression profile of MOLT-4 cells after 3 hours of treatment with either vehicle or the whole extract (Fig. 3A). A list of the ten most increased- and decreased-abundance genes following treatment is presented in supplementary table S4. ChaC glutathione-specific γ-glutamylcyclotransferase 1 (CHAC1), a negative regulator of the Notch signaling pathway (34), was identified as the second most increased-abundance gene. CHAC1 acts by inhibiting the furin-like cleavage of Notch1, preventing the formation of the mature form of Notch1 (35), as we have previosly demonstrated upon treatment with the whole extract (32). C/EBP homologous protein (CHOP) and Activating transcription factor 4 (ATF4), ER-stress signaling pathway markers that are upstream to CHAC1 (36), had also a significant increase in their abundance. We confirmed the RNA-seq results by qRT-PCR showing that the mRNA of ATF4, CHOP and CHAC1 is indeed significantly increased after treatment with a combination of 331-18A, CBDV and CBD (Fig. 3B-D, respectively) or the whole extract (fig. S3E-G). A significant increase was apparent 60 min after treatment. Using specific antibodies to ATF4, CHOP and CHAC1, we verified the protein expression level of all three was also significantly increased at 60 min after treatment (Fig. 3E-H). The protein expression levels of CHAC1 in MOLT-4 cells were also significantly increased upon treatment with the whole extract (fig. S3H, I).

Fig. 3.

To further understand the mechanism, we evaluated how pretreatment with antagonists to the cannabinoid receptors through which 331-18A, CBDV and CBD exert their effect, CB2 and TRPV1, affected ATF4, CHOP and CHAC1 expression following treatment (Fig. 3I-K). Inhibition of CB2 reduced the treatment-induced increase in mRNA expression, but a trend of increase for ATF4 and CHOP remained at 60 minutes, and CHAC1 expression was significantly increased. Inhibition of TRPV1 rescued from treatment-induced induction of all three targets. Similar effects were found when pretreatment with antagonists to CB2 and TRPV1 was followed by treatment with the whole extract (fig. S3J-L); and the increase in the expression levels of CHAC1 protein was abolished when MOLT-4 cells were treated by either antagonist (fig. S3M). These results demonstrate that CB2 and TRPV1 participate in inducing CHAC1 expression and preventing Notch1 maturation, resulting in a decrease in NICD release and cell death.

We therefore speculated that the combination of the three cannabinoids caused the activation of ATF4-CHOP-CHAC1 signaling by Ca²⁺ depletion and ER-stress (37). To further verify the role of ATF4-CHOP-CHAC1 signaling in mediating the downregulation of the Notch1 pathway, we inhibited the activity of the upstream target eukaryotic Initiation Factor-2α (eIF2α) in MOLT-4 cells using integrated stress response inhibitor (ISRIB), which inhibits the activity of eIF2α without affecting its phosphorylation (38) (fig. S5A-B). Then, we treated the cells with either vehicle or the combination of the cannabinoids. After 24 hours, blocking the activity of eIF2α prevented the cytotoxic effect of 331-18A, CBDV and CBD combination (Fig. 4A) or of the whole extract (fig. S5C). Moreover, eIF2α inhibition prevented the reduction in NICD expression upon treatment with either the whole extract (fig. S5D-E) or the cannabinoid combination (Fig. 4B-C). Inhibition of eIF2α also rescued the cells from the induced increase in the mRNA expression of ATF4, CHOP and CHAC1 upon treatment with the combination of the cannabinoids (Fig. 4D-F). Thus, inhibition of the integrated stress response pathway, eIF2α-ATF4-CHOP-CHAC1 signaling, rescued MOLT-4 cells from the cytotoxic effect of the combination of the three cannabinoids and prevented the reduction in NICD expression, as well as the increased transcription of ATF4, CHOP and CHAC1 upon treatment.

Fig. 4.

To further investigate the interaction between 331-18A,CBDV and CBD, we assessed their effect on the viability of MOLT-4 cells in different relative proportions. First we tested the ability of each seaparate cannabinoid to inhibit the proliferation of MOLT-4 cells in doses ranging 0-2 µg/ml (Fig. 5A), which is below and above the overall concentration for which we found an effect for the cannabinoid combination. 331-18A and CBDV affected viability from a dose of 1 µg/ml and CBD from a dose of 0.5 µg/ml. Next, we calculated the combination index of every pair and all three cannabinoids in concentrations ranging from 0-2 µg/mL (Fig. 5B-E). The most synergistic area is displayed in red. We calculated the total synergy score for the combination of all three cannabinoids relative to the pairwise combinations (Fig. 5F). According to Highest Single Agent (HSA) reference model(39), a combination of CBDV with either 331-18A or CBD results in a likely antagonistic interaction (score < -10), a combination of 331-18A and CBD results in a likely additive interaction (score from -10 to 10), and only the combination of all three cannabinoids is likely synergistic (score > 10, Fig. 5G).

Fig. 5.

Inhibition of cancer progression in-vivo by 331-18A, CBDV and CBD combination

We further evaluated the anti-tumor properties of the three cannabinoids using several mouse models. The first model was a subcutaneous tumor model for which we have already demonstrated the effectivity of the whole extract (32). We engrafted subcutaneously Nonobese diabetic-severe combined immunodeficiency (NOD/Scid) mice with MOLT-4 cells, then treated with vehicle only or the combination of the cannabinoids from when tumors reached a volume of 100 mm³ (day 7), on alternating days, for a duration of four weeks. Treatment with a combination of 331-18A, CBDV and CBD significantly inhibited tumor growth (Fig. 6A-C). Significant differences were apparent approximately 3 weeks after treatment started (Fig. 6A), and at the endpoint, 34 days from the injection of cells, the average weight of tumors was significantly lower in the treated group compared to the control group (Fig. 6B). Excised tumors (Fig. 6C) were analyzed for their cannabinoid content according to a method recently developed by our group for biological tissues (40). Using ESI-LC/MS, we identified and quantified each of the three cannabinoids in the tumors [16.0 ± 6.7, 22.3 ± 10.4 and 265.8 ± 109.7 ng/g for CBDV, 331-18A and CBD, respectively (average ± S.E.M., N=3)]. Notably, the ratio of the cannabinoids analyzed in the tumors was similar to the ratio given to the mice in the treatment protocol. In addition, a significant decrease was observed in NICD expression in the tumors treated with the cannabinoids (Fig. 6D). To assess possible toxicity, the body weight of the mice was monitored throughout the experiment (Fig. 6E). There were no significant differences in the average weight of mice in the control group and the treatment group.

Fig. 6.

Next, we assessed the cannabinoids using an intravenous model which better resembles leukemogenesis (41, 42). We engrafted NOD-SCID IL2rγ-null (NSG) mice that are a better host for human immune cells with CCRF-CEM, Notch1 mutated T-ALL cells. The mice were treated with a combination of the three cannabinoids, and to verify the need for the combination, one group was also treated with pure CBD in the same overall dose. At the endpoint of the experiment, the amount of human CD45+ cells detected in the bone marrow (BM) was significantly decreased in the mice treated with the combination of the three cannabinoids compared to the control, and compared to pure CBD (Fig. 6F). In this model as well, there were no significant differences in the average weight of the mice (Fig. 6G). The combination of 331-18A, CBDV and CBD was as effective as the whole extract (fig. S6A-B). In addition, the administration of the cannabinoid combination significantly prolonged survival in treated mice compared to the survival of vehicle-treated control mice (Fig. 6H), to a similar degree to the whole extract (fig. S6C).

To examine the clinical relevance, we utilized a patient-derived xenografts (PDX) model with primary cells from a T-ALL patient. After allowing 35 days from transplantation for leukemic cell establishment, mice were treated for three weeks with either vehicle or whole extract containing the three cannabinoids. In mice that were treated with the extract, we found a significant reduction in the burden of leukemia in the bone marrow (fig. S7A), peripheral blood (fig. S7B) and spleen (fig. S7C). The weight of the spleen was also significantly lower in extract-treated mice (fig. S7D, E). Treatment with the whole extract did not affect the weight of the mice.

Discussion

Targeting Notch signaling components has generated much interest for its therapeutic potential. However, so far efforts to develop treatments that target Notch signaling have been unsuccessful (43). Cannabis, in its many forms, is already being prescribed to cancer patients all over the world, mainly as a palliative treatment for the inhibition of nausea and emesis associated with chemotherapy, appetite stimulation, pain relief and relief from insomnia (17). On top of that, recent studies have demonstrated that phytocannabinoids also possess anti-cancer potential (24, 25, 31, 44–46).

In the current study, we identified three unique phytocannabinoids that together selectively induce apoptosis in Notch1 mutated leukemia cells: CBD, CBDV and the novel phytocannabinoid 331-18A. CBD and CBDV are well-known and commercially available. However, to the best of our knowledge, the third phytocannabinoid 331-18A, denoted in the literature as 6,12-dihydro-6-hydroxy-cannabidiol (47), has not been previously identified in natural and/or decarboxylated Cannabis plants. We suggest that the need for all three cannabinoids for the optimal anti-tumor effect demonstrates the critical importance of elucidating the individual and polypharmacological effects of the entire molecular spectrum of Cannabis.

Although CBD by itself demonstrated some capacity to induce cell death, it was only when all three cannabinoids were combined that the same extent of response was achieved as treating with the whole extract. The Cannabis plant produces a vast array of over 140 phytocannabinoids and hundreds of other components, but the full extent of their biological effects and interactions is still not completely understood. Nevertheless, mounting evidence suggests that their combined effect as a whole is greater than that of each separate compound (48, 49). Recently, Blasco-Benito et al. (50), have shown that a Δ⁹- THC-rich extract was much more potent than pure Δ⁹-THC in producing antitumor responses in both cell culture and animal models of breast cancer. However, they were unable to identify the combinations of compounds that were responsible for this increased potency. In previous publications by our group, we showed that different CBD-rich Cannabis extracts, with equal amounts of CBD but varying concentrations of other minor compounds, lead to diverse anticonvulsant effects in a mouse model of epilepsy (29), and selective antitumoral effects in Notch1-mutated in T-ALL (32); suggesting the potential therapeutic effects of other compounds in the extracts in addition to CBD.

A possible mechanism previously suggested to explain the difference between the effects of purified phytocannabinoids versus full-spectrum extracts is the "entourage effect", where one compound may synergize the activity and efficacy of another on the same target (48, 49). This "entourage effect" was postulated by Russo (48) based on a phenomenon observed in the eCBS by Ben Shabbat et al. (51). While it is well-established for endocannabinoids, only very few studies demonstrated this for phytocannabinoids. We showed here that the presence of small amounts of the minor cannabinoids 331-18A and CBDV enhanced the effect of CBD by itself. Our success in isolating and characterizing a minimal number of cannabinoids that achieve the same antitumor effect as the whole extract is a significant advancement in the field of medical Cannabis and provides solid evidence for the “entourage effect”.

Subsequent reports have expanded the notion of the "entourage effect" to polypharmaceutical and/or polypharmacological effects, where several of the plant metabolites work simultaneously on the same receptors, and/or when the same molecule has several different targets within the same system (52–54). CBD has been shown to present complex signaling capabilities of many receptors in the extended eCBS. Specifically, it was shown to have a weak affinity to CB2, and a strong affinity to many of the TRPV family receptors, including TRPV1, TRPV2 and TRPV3 (55, 56).

We suggest that the Notch1-mutated T-ALL cells are affected through the engagement of the eCBS receptor CB2 and the channel TRPV1. More specifically, the known cannabinoids CBD and CBDV affected both, while the novel 331-18A exerted its effect only through CB2. Following the stimulation of the receptors, Ca²⁺ is depleted from the ER and there is an increase in cytosolic Ca²⁺ and activation of eIF2α followed by ATF4-CHOP-CHAC1 signaling. Notably, the inhibition of eIF2α upstream to ATF4 rescued T-ALL cells from reduced NICD expression by treatment with 331-18A, CBDV and CBD combination. The upregulation of CHAC1 inhibits the S1 furin-like cleavage of Notch1 at the Golgi apparatus (35), leading to the arrival of immature Notch1 to the plasma membrane as we have previously shown (32). This inhibition of the Notch1 pathway ultimately leads to apoptosis in T-ALL cells that rely on mutated Notch1 for propagation (see schematic representation in Fig. 7).

Fig. 7.

The potential antitumor effects of cannabinoids through the Notch1 signaling pathway were previously demonstrated in glioma, where THC induced the expression of p8 protein, upstream to ATF3/4 (57). The activation of Notch signaling by ER stress was demonstrated in additional systems, including Alzheimer’s disease (58), polycystic ovary syndrome (59) and other types of cancer (60, 61). A previous work that focused on chronic lymphocytic leukemia suggested a role for ER stress in Notch1 cleavage and leukemia cell death through the inhibition of proteasome activity (62). These findings suggest blocking Notch1 maturation through the eIF2α-ATF4-CHOP-CHAC1 pathway and inhibiting the S1 furin-like cleavage of Notch1 have the potential to affect all types of activation mutations in Notch1, and also all four human homologs of Notch (Notch1-4) (35, 63).

We demonstrated the effectiveness of the whole extract and the combination of the cannabinoids in several in-vivo models. To validate the clinical relevance of our findings for patients, we tested the extract containing the three cannabinoids in an in-vivo PDX model (64, 65) (fig. S7), and demonstrated treatment reduced leukemic burden. As medical Cannabis is already routinely prescribed to cancer patients, tailoring chemovars that contain the appropriate cannabinoids to T-ALL patients can provide them with optimal care that offers anti-cancer properties on top of the palliative ones. Our findings are potentially relevant for numerous diseases, as dysregulation of Notch signaling has been found in various other cancers including breast, prostate and colorectal; as well as in non-cancerous diseases (8, 9, 12). As such, they may pave the way for the establishment of new pharmacotherapies for the treatment of Notch1-dependent malignancies and other diseases.

Materials and Methods

Reagents

The analytical standards (>98%) for phytocannabinoids were purchased from Silicol Scientific Equipment Ltd. (Or Yehuda, Israel). The antagonists CID 16020046 (#4959/5), AM630 (#1120/10), SR-144,528 (#5039/10) and AMG9810 (#2316/10) were purchased from Tocris Bioscience (Bristol, UK). BIM-46187 (#5332990001) and ISRIB (#SML0843) were purchased from Sigma-Aldrich (Rehovot, Israel). Synthetic CBD, CBDV and 331-18A ware purchased from Cannasoul Analytics (Caesarea, Israel).

Phytocannabinoids analysis

Comprehensive phytocannabinoid analysis was performed using a UHPLC system coupled to a Q Exactive™ Focus Hybrid Quadrupole-Orbitrap MS (Thermo Scientific, Bremen, Germany) using a similar chromatographic method as previously described (29, 32). MS acquisition was carried out with a heated electro spray ionization (HESI-II) ion source operated in negative mode. Source parameters were as follows: sheath gas flow rate, auxiliary gas flow rate and sweep gas flow rate: 50, 20 and 0 arbitrary units respectively; capillary temperature: 350°C; heater temperature: 50°C; spray voltage: 3.00 kV. The scan range was 150–550 m/z for all acquisition events. MS was operated in full MS¹ mode at 70,000 resolution, and the AGC target was set to 10⁶ with a maximum IT of 100 ms. Identification and absolute quantification of phytocannabinoids was performed by external calibrations as described by Berman et al.(29). Absolute quantification of phytocannabinoids with analytical standards was performed by external calibrations. Standard mixes were prepared ranging from 1 to 1000 ng/mL for Δ⁹-THCA, Δ⁹-THC, CBDA and CBD, and 0.25 to 625 ng/mL for all the other components. The dynamic range for each component was determined as maximum deviation from expected concentrations of 20%, and minimum signal-to-noise ratios of 10. All other phytocannabinoids without analytical standards were semi-quantified according to calibration curves from the same phytocannabinoid family or average curves for unknown compounds (29). All samples were injected and analyzed by ESI-LC/MS in three concentrations (10, 1 and 0.1 µg/mL Cannabis extract to ethanol).

Isolation of single phytocannabinoids

Fractionation of the whole Cannabis extract into four fractions and isolation of single phytocannabinoids were performed using a semi-preparative HPLC/UV (Thermo Scientific, Bremen, Germany). The chromatographic separation was achieved using a Luna® C18 column (10 μm, 250 mm × 21.1 mm i.d.) and a two-solvent A/B multistep gradient (solvent A: 0.1% acetic acid in water and solvent B: 0.1% acetic acid in acetonitrile, the solvents and acetic acid were of HPLC and LC/MS grades, respectively). The multistep gradient program was established as follows: initial conditions were 50% B for 4 min, raised to 76% B until 6.5 min, held at 76% B for 16.5 min, and then raised to 90% B until 26 min, held at 90% B until 31 min, decreased to 50% B over the next 4 min, and held at 50% B until 40 min for re-equilibration of the system prior to the next injection. A flow rate of 20 ml/min and an injection volume of 500-750 µl were used. Data acquisition was performed at 220 nm. The crude Cannabis extract was prepared in a concentration of 100 mg/ml in ethanol. Fractions were collected in 10 min intervals and then lyophilized to dryness and analyzed by UHPLC/UV and ESI-LC/MS.

To improve the yield and purity of the fractions for isolation of single phytocannabinoids, a preliminary separation step was employed prior to semi-preparative HPLC fractionation. The crude Cannabis extract was dissolved to a concentration of 0.5 g/ml in ethanol, filtered using a PTFE 0.45 μm filter, and then purified using a PLC 2050 Purification System, equipped with a CPC-250 column and a photodiode array (PDA) detector (Gilson Inc.). Separations were achieved using a biphasic solvent system consisting of upper (hexane:acetonitrile in the ratio of 98:2 v/v) and mobile (acetonitrile:water in the ratio 50:50 v/v) phases in descending mode. The column was operated with a flow rate of 12.5 ml/min and a rotation speed of 2,200 rpm. The gradient program was established as follows: initial conditions were acetonitrile:water:hexane in the ratio 50:50:0 v/v for 25 min, and then gradually raised to acetonitrile:water:hexane in the ratio 64:34:2 v/v until 45 min. All the solvents were of HPLC grade. The collected fractions were lyophilized to dryness and then subjected to semi-preparative HPLC fractionation. The specific phytocannabinoids were collected according to retention time.

Structure elucidation

The structure of the isolated 331-18A was elucidated by ¹H and ¹³C one-dimensional NMR spectra (Avance-III-700, Bruker, Germany) with CBD as the reference. The samples were dissolved in CDCl3 and analyzed at 4°C with Tetramethylsilane as the internal reference. The optical rotation of 331-18A in comparison to CBD was determined in chloroform (10 mg/mL) using a Jasco P-1010 (Rudolph Research Analytical, US) polarimeter at 26°C.

Cell culture

The T-cell leukemia cell lines MOLT-4 (CVCL_0013, CRL-1582™) and CCRF-CEM (CVCL_0207, CCL-119™) were purchased from ATCC. Cell lines were grown in RPMI 1640 media (Sigma-Aldrich; R8758), and kept at 37°C in a 5% carbon dioxide incubator. Unless otherwise stated, all culture media were supplemented with 10% FBS (Biological Industries; 04-007-1A), penicillin (100 U/mL) and streptomycin (100 µg/mL) (Biological Industries; 03-031-1B).

Cell apoptosis assay

Cells were cultured in 12-well plates at 1×10⁶ cells/well with media containing 0.5% FBS and incubated for 24 h with either DMSO (vehicle), Cannabis extract 12, fractions 1-4 and different combinations of 0.06 µg/ml 331-18A, 0.06 µg/ml CBDV and 1.5 µg/ml CBD. Apoptotic cells were detected by an Annexin V/PI assay using flow cytometry or by detection of cleaved caspase-3.

Annexin V/PI assay

Apoptosis was assessed by Annexin V-FITC (BioVision; 1006-200, 1:500) and PI staining in an annexin binding buffer (BioVision; 1006-100; 1:500) according to the manufacturer’s instructions. Apoptosis assessed by flow cytometry used a BD™ LSR II digital four-laser flow cytometer (BD Biosciences) and was analyzed by BD FACSDiva™ software version 6.1.2. (BD Biosciences). Results were calculated as percentage of positive Annexin V-FITC cells out of total cells counted (30,000 events).

Western blot analysis

Cells were quickly washed twice with ice-cold PBS, and lysed in radioimmunoprecipitation assay buffer (RIPA) lysis buffer (Sigma Aldrich; R0278) freshly supplemented with protease and phosphate inhibitors (Cell Signaling Technology; 5872). Lysates were cleared by centrifugation at 13,000 g at 4°C for 10 min. Protein concentration was determined by a Bradford assay (Sigma Aldrich; B6916). Proteins were denatured in NuPAGE™ LDS Sample Buffer (4X) (Thermo Fisher; NP0008) and supplemented with β-mercaptoethanol (Sigma Aldrich; B6131). Typically, 20-50 mg of total protein were loaded per lane on a Novex™ 4-20% Tris-Glycine gradient gel (Invitrogen; XP04202BOX) and electrophoretically transferred to a nitrocellulose membrane (Bio-Rad; 1704159S). Membranes were blocked with TBS 0.1% Tween 20 buffer containing 5% bovine serum albumin (Sigma; A7906) for 1 h and incubated with the primary antibody overnight at 4°C. This was followed by incubation with horseradish peroxidase-labeled matching secondary antibodies. Immunoreactive bands were detected by Luminata™ HRP substrate (Millipore; WBLUR0500) and visualized using a MicroChemi imager (DNR Bioimaging Systems). Antibodies used: Cleaved Notch1 Val1744 (CST, 4147S; AB_2153348; 1:1,000) to assess NICD, cleaved caspase-3 (CST, 9664; AB_2070042; 1:1,000), β-tubulin (CST; clone D3U1W, 86298; AB_2715541; 1:1,000), GAPDH (CST, 2118; AB_561053; 1:5,000), ATF4 (Proteintech, 10835-1-AP; AB_2058600; 1:500), CHOP (Proteintech, 15204-1-AP; AB_2292610; 1:500), CHAC1 (abcam, ab76386; AB_1658635; 1:500), eIF2α (CST, 5324; AB_10692650; 1:1000) and Phospho-eIF2α (Ser51) (CST, 3398; AB_2096481; 1:1000). Secondary antibodies include: Peroxidase AffiniPure Goat Anti-Rabbit IgG (H+L) (Jackson ImmunoResearch, 111-035-144; AB_2307391; 1:5,000) and Peroxidase-AffiniPure Goat anti mouse IgG (HZ+L) (Jackson ImmunoResearch, 115-035-003; AB_10015289; 1:5000).

RNA extraction

Total RNA was isolated from MOLT-4 cells (1×10⁶ cells/sample) using TRIzol® (Thermo Fisher; 15596026) and RNeasy kit (Qiagen; 74104) according to the manufacturer’s instructions. Sample quality was assessed both spectrophotometrically (Nanodrop Technologies®) and by gel electrophoresis with 1% agarose.

RNA-seq

RNA sequencing of differentially expressed genes was performed on extracted RNA with an Affymetrix Clariom-S microarray for human genome RNA expression analysis and an Affymetrix Transcriptome Analysis Console software. The threshold set for increased- and decreased-abundance genes was a fold change ≥1.5 and p<0.05. The sequencing was performed by the Center for Genomic Technologies of The Hebrew University of Jerusalem.

Quantitative RT-PCR

cDNA was synthesized from 1 µg of extracted RNA. Purified RNA was reverse transcribed with the qScript™ cDNA synthesis kit (Quanta Biosciences; 95047). The mRNA expression levels of were quantified using qRT-PCR 7300 system (Applied Biosystems) using Human Taqman probes (Thermo Fisher Scientific) for: GAPDH (Hs02758991_g1); CNR1 (CB1) (Hs01038522_s1); CNR2 (CB2) (Hs00361490_m1); GPR55 (Hs00271662_s1); TRPV1 (Hs00175798_m1); TRPM8 (Hs01066596_m1); TRPA1 (Hs00175798_m1); ATF4 (Hs00909569_g1); DDIT3 (CHOP) (Hs00358796_g1); CHAC1 (Hs00225520_m1). Relative expression values were normalized using to the endogenous housekeeping gene GAPDH as the control and calculated using standard ΔCT methods.

Calcium imaging

MOLT-4 Cells were loaded with the fluorescein Ca²⁺ probe Fluo-4 AM (Invitrogen # F14217). Cells were incubated for 60 min in RPMI medium and supplemented with 1 µM Fluo-4/AM. Cells were then rinsed twice and incubated for another 30 min in a Fluo-4 AM-free HBSS buffer (Sigma Aldrich #H6648). During these steps the cells were kept in the dark. Fluo-4 signals were collected by a BD™ LSR II digital four-laser flow cytometer (BD Biosciences) and analyzed by BD FACSDiva™ software version 6.1.2. (BD Biosciences).

Cell viability assay

Cells were cultured in 96-well plates. At 1×10⁵ cells/well the media was replaced with RPMI containing 0.5% FBS and treatments as indicated. DMSO was used as the vehicle. The XTT-based cell proliferation kit (Sartorius, Beit Haemek, Israel) was used to quantitate viability according to the manufacturer’s instructions. Cytotoxicity was measured following 24 hrs of incubation using fluorescence spectrophotometry at excitation and emission wavelengths of 450-500 and 630-690 nm, respectively, and calculated as percent reduction between treated and control cells.

Synergy Scoring

The cell viability percentage for each concentration was assessed with an XTT-based cell proliferation kit (Sartorius, Beit Haemek, Israel), then entered into SynergyFinder web application (v 2.0). The HSA model was used to calculate synergy scores of the cannabinoid combinations. Estimation of outlier measurements was carried out with cNMF algorithm (66) implemented in SynergyFinder.

Mice and procedures

All procedures were performed according to protocols approved by the Technion Administrative Panel of Laboratory Animal Care (animal protocol number: IL_0470317). Female NOD.CB17-Prkdcscid/NCrHsd (IMSR_ENV:HSD-170) adult (6-7 weeks) mice were purchased from Envigo. Female NSG adult (6-7 weeks) mice were purchased from the Technion animal facility.

For the in-vivo tumorigenicity assay, MOLT-4 cells mixed with BD Matrigel^TM Matrix Growth Factor Reduced (BD; FAL356230) (1:1 v/v) were injected subcutaneously into the flanks of 7-week-old female NOD-scid mice. Tumor size was measured with a caliper, and the tumor volume was determined according to the formula: length x width² x 0.5. The mice were euthanized after the indicated days as none reached the maximal endpoint size (1 cm³). The xenograft tumors were dissected and weighed.

For the in-vivo leukemogenesis assay and survival analysis, CCRF-CEM cells were intravenously injected to 7-week female NSG mice. The bone marrow was harvested from each mouse and the infiltrated CCRF cells were identified by staining for a human CD45 antibody (Biolegend; AB_2687375; BLG-368522; 1:200).

For the Patient-Derived Xenograft (PDX) model, frozen aliquots of T-ALL patient samples in which Notch1 mutation was verified by next-generation sequencing were purchased from Dana-Farber Cancer Institute. The cells were expanded in mice and injected subcutaneously into the flanks of 7-week female NSG mice. To investigate the inhibitory effects of extract 12 on leukemogenesis, 30 mg/kg/day extract was intraperitoneally injected three times a week, starting three weeks after cell injection. The mice were euthanized after the indicated days. BM, peripheral blood and spleens were harvested from each mouse and the infiltrated primary cells were identified using flow cytometry by staining for human CD45 antibody (BioLegend; BLG-368522; 1:200).

Cannabinoids analysis from tumor tissue

Fresh excised tumors were accurately weighed and dissected to smaller sections. Hanks’ Balanced Salt Solution (Sigma Aldrich; H8264) was added in a ratio of 1 ml per 100 mg sample. The samples were further homogenized into single-cell suspensions with a gentleMACS™ Dissociator (Miltenyi Biotec) using a pre-set program for tumor tissues. The single cell suspensions were stored at -80°C until further extraction. CBD, CBDV and 331-18A were analyzed using the ESI-LC/MS system by retention time, accurate mass and spectral matching against our developed ESI-LC/MS/MS library of phytocannabinoids (29). Absolute quantification of the three compounds was performed by the stable isotope dilution method as recently described by our group (40). Briefly, the extraction solution (methanol:acetonitrile:acetic acid in the ratio 50:50:0.1 v/v) spiked with 20 ng/mL of CBD-d3 internal standard (Sigma-Aldrich; C-084) was added to the single cell suspensions in the ratio 3:1 v/v, respectively. Samples were thoroughly vortexed and then centrifuged for 20 min at 4°C for protein and cell precipitation. The supernatants were then mixed in a ratio of 1:3 v/v sample to 0.1% v/v acetic acid in water, and loaded onto Agela Cleanert C8 solid phase extraction cartridges (500 mg of sorbent, 50 μm particle size). Cannabinoids were eluted from the columns using 2 ml of 0.1% v/v acetic acid in methanol, evaporated to dryness by speedvac, and reconstituted in 100 µl ethanol. Since there was no analytical standard available for 331-18A, this compound was semi-quantified according to the calibration curve of CBD.

Statistics and reproducibility

Data were analyzed with GraphPad Prism software version 7.04. Statistical significance was determined by either an unpaired two-tailed Student’s t-test (with unequal variations if an F test ruled out the equal variation assumption) or one-way ANOVA followed by Sidak’s multiple comparisons test for multiple groups. Kaplan–Meier survival curves were analyzed with the Gehan–Breslow–Wilcoxon test, as appropriate for the dataset. Data are generally expressed as mean ± SEM, as denoted in the Figure captions, with p≤0.05 was considered statistically significant.

Supplementary Materials

Table S1. Phytocannabinoid concentrations by UHPLC/LC-MS of fractions relative to the whole extract.

Table S2. Phytocannabinoid concentrations by UHPLC/LC-MS of cannabis fraction 2 peaks.

Table S3. ¹H and ¹³C peak assignments and chemical shifts of 331-18A.

Table S4. Ten most increased- and decreased-abundance genes following treatment of MOLT-4 cells with the whole extract according to Affymetrix.

Fig. S1. Fraction 2 is as effective as the whole extract in reducing the viability of MOLT-4 cells.

Fig. S2. Specific identification of phytocannabinoids with anti-cancer properties by spectral matching and peak purity.

Fig. S3. CBD and 331-18A NMR spectra.

Fig. S4. Notch1 downregulation by Extract 12 is mediated through CB2 and TRPV1 followed by ATF4-CHOP-CHAC1 signaling pathway.

Fig. S5. Whole extract treatment activates eIF2α.

Fig. S6. Inhibition of tumor growth in-vivo by whole extract.

Fig. S7. Treatment with extract containing the three cannabinoids inhibits leukemic expansion in an in-vivo PDX model.

Acknowledgements

We thank the the Israel Cancer Research Fund, Evelyn Gruss Lipper Charitable Foundation and the Israeli Ministry of Agriculture and Rural Development for their financial support of this work. The authors would like to thank Cannasoul Analytics and especially Dr. Rotem Perry Feigenbaum and Yogev Nir for the isolation of the phytocannabinoid compounds; Dr. Hugo Gottlieb and Dr. Michal Afri from the NMR unit at Bar Ilan University for performing the NMR measurements and deciphering the structure of the 331-18A phytocannabinoid; Moshe Barkaly for the semi-preparative HPLC fractionations of the whole extract; and Dr. Dafna Antes for the graphic scheme illustration. The authors declare no competing financial interests.

Funding

Israel Cancer Research Fund grant 21-112-PG (DM)

Gruss Lipper Charitable Foundation grant 2027093 (DM)

Israeli Ministry of Agriculture and Rural Development 14370001 (DM)

Author contributions

Conceptualization: E.B., S.P. and D.M.

Methodology: E.B. and P.B.

Investigation: E.B., A.G. and P.B.

Analysis and interpretation of data: E.B., A.G., P.B and S.P.

Writing-original draft: E.B. and D.M.

Writing-review & editing: E.B., D.M., P.B. and S.P.

Study supervision: D.M.

Competing interests

A patent pertaining to the results of the paper has been applied for the use of cannabinoids in the treatment of Notch-related diseases. This patent was licensed to Cavnox Ltd. to test the combination of the molecules in human clinical trials.

Data and materials availability

The RNA-Seq data that support the findings of this study have been deposited in the Gene Expression Omnibus under the accession code GSE154287. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE154287.

All data generated or analyzed during this study are included in this published article and its supplementary information files.