Antitumor Cannabinoid Chemotypes: Structural Insights

Cannabis has long been known to limit or prevent nausea and vomiting, lack of appetite, and pain. For this reason, cannabinoids have been successfully used in the treatment of some of the unwanted side effects caused by cancer chemotherapy. Besides their palliative effects, research from the past two decades has demonstrated their promising potential as antitumor agents in a wide variety of tumors. Cannabinoids of endogenous, phytogenic, and synthetic nature have been shown to impact the proliferation of cancer through the modulation of different proteins involved in the endocannabinoid system such as the G protein–coupled receptors CB1, CB2, and GRP55, the ionotropic receptor TRPV1, or the fatty acid amide hydrolase (FAAH). In this article, we aim to structurally classify the antitumor cannabinoid chemotypes described so far according to their targets and types of cancer. In a drug discovery approach, their in silico pharmacokinetic profile has been evaluated in order to identify appropriate drug-like profiles, which should be taken into account for further progress toward the clinic. This analysis may provide structural insights into the selection of specific cannabinoid scaffolds for the development of antitumor drugs for the treatment of particular types of cancer.


INTRODUCTION
During these last years, significant research has been focused on the therapeutic potential of cannabinoids to manage palliative effects in cancer patients (Badowski, 2017;Guzmán, 2018). Cancer-induced emesis represents the initial target indication for Marinol ® (dronabinol) and Cesamet ® (nabilone), two cannabis-based medicines approved by various regulatory drug agencies. Management of cancer-induced neuropathic pain is also part of the palliative applications of cannabis-based medicines. Besides such palliative applications, some cannabinoids have shown anticancer properties (Guzmán, 2003;Guindon and Hohmann, 2011;Khan et al., 2016;Hinz and Ramer, 2019). As widely reported in the last decades, some cannabinoids are able to modulate different cellular signaling pathways implicated in cancer cell proliferation, migration, or death (Chakravarti et al., 2014). Even though the underlying mechanisms are not totally unraveled, there is significant evidence for the involvement of at least four mechanisms: direct inhibition of transformed-cell growth through the suppression of mitogenic signal, induction of apoptosis, inhibition of tumor angiogenesis, and metastasis (Velasco et al., 2016). The signaling pathways implicated in the activation of the endocannabinoid system may differ depending on specific cancers and/or experimental models, making more complex the understanding of these processes. So far, only few clinical data on the efficacy of cannabinoids as anticancer agents have been provided (Ladin et al., 2016). However, great efforts are currently being made to elucidate the potential utility of cannabinoids as anticancer therapeutics.
The physiological processes triggered by most of these cannabinoids are mediated by two G protein-coupled cannabinoid receptors (CBR), CB1R and CB2R. CB1R is predominantly and abundantly expressed in the central nervous system, with predominance in the hippocampus, cerebellum, basal ganglia, and cortical and olfactory regions, but CB1R is also present in many organs of the peripheral system. CB2R is mainly found in the immune system, in the spleen, thymus, lymph nodes, and tonsils, but it is also expressed in immune cells.
The overexpression of CBR and elevated endocannabinoid levels have been reported in different cancer types Pisanti et al., 2013). This expression in cancer cells is crucial for downstream signaling with implications on cell viability.
Focusing on a drug discovery approach, herein we have summarized the anticancer profiling of the cannabinoids reported thus far to impact cancer. Moreover, we have calculated their in silico pharmacokinetic profiles in order to predict appropriate drug-like profiles that may provide useful criteria for further development selection. In silico prediction of pharmacokinetic properties is a very useful approach that provides a great translational tool since absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties and bioavailability of drugs can strongly influence their development (Di et al., 2018).

CANNABINOIDS WITH ANTICANCER POTENTIAL
Molecules that modulate the endocannabinoid system are considered cannabinoids. These compounds generally have been classified following their structural nature or origin. Thus, they all belong to phytogenic-, endogenous-, or synthetic-derived families.

Endocannabinoids
Endogenous cannabinoids, called endocannabinoids, such as anandamide (AEA) and 2-arachidonoyl glycerol (2-AG), form a major family of cannabinoids (Ligresti et al., 2016). Structurally, they are lipid-based derivatives derived from arachidonic acid. They are involved in a number of physiological processes but are also easily degraded through enzymatic pathways. AEA is known to affect cancer cell proliferation; however, there are cell lines whose proliferation is more sensitive to anandamide than others. The molecular mechanism of action differs also from one cell line to another. For instance, AEA exerts a potent CB1R-mediated effect on the proliferation of MCF-7, and EFM-10 human breast cancer cells (Di Marzo et al., 1998), while in N18TG2 murine neuroblastoma cells, the effect is due to FAAHmediated degradation of AEA to ethanolamine (Matas et al., 2007). Another example concerns non-melanoma skin cancer, for which AEA induces endoplasmic reticulum stress and apoptosis mediated by oxidative stress and by CBR-independent endocannabinoid signaling .
(R)-Methanandamide (Met-AEA, Table 1) has been used in diverse biological assays as a metabolic stable anandamide analogue. One of the first assays in cancer concerns prostate LNCaP cells (Sánchez et al., 2003).
From a series of linolenic and arachidonic derivatives, 4g and 5c (Table 1) enhance AEA cytotoxicity on C6 glioma cell viability (Quintana et al., 2016). Both of them incorporate the same alkanolamine moiety in their structure.

Phytocannabinoids
The plant-derived family is exemplified by the phytocannabinoids tetrahydrocannabinol [(-)-Δ 9 -THC); THC] and cannabidiol [(-)-CBD; CBD], the two main components of Cannabis sativa. Structural modifications of THC mainly have been developed in the Makriyannis laboratory some years ago (Thakur et al., 2005). Concerning CBD, a recent review dedicated to CBD as scaffold provides an overview of the chemical structure of natural and synthetic derivatives, including their relative molecular targets (Morales et al., 2017b). Reports on anticancer effects of phytocannabinoids mainly have been focused on the activity of Δ 9 -THC, CBD, Δ 9 -tetrahydrocannabinolic acid (Δ 9 -THCA), cannabidiolic acid (CBDA), cannabigerol (CBG), and cannabichromene (CBC) ( Table 1) (Guzmán, 2003;Velasco et al., 2012;Fowler, 2015;Hinz and Ramer, 2019;Pellati et al., 2018). The effects of Δ 9 -THC have been tested in different cell lines of prostate cancer, breast cancer, colon cancer, pancreatic cancer, lymphoma, lung cancer, glioblastoma, and myeloma, among others (Fowler, 2015;Fraguas-Sánchez et al., 2016). Intracellular signaling through CBR has been shown to play an important role in these effects, involving complex signal transduction pathways, for instance, the ceramide pathway and/ or the PI3-K and ERK pathways. However, the mechanism of action of Δ 9 -THC can also result in being CBR-independent, as has been shown for leukemic cell lines (Powles et al., 2005).
The nonpsychoactive cannabinoid CBD revealed proapoptotic effects in different cancer cell lines (Hinz and Ramer, 2019). CBR activation is not required for CBD anticancer action since CBD shows very low affinity. Accumulation of reactive oxygen species (ROS) is one of the main reported mechanism responsible for apoptosis induced by CBD (Ligresti, 2006

Compound
Targeted tumor Antitumor effect/mechanism of action References
Quercetin, a flavonoid present in fruits and vegetables, inhibits the growth of human colon adenocarcinoma cells through CB1R (Refolo et al., 2015). Another flavonoid structurally closely related to quercetin, morin (Table 1), showed an apoptotic effect by a mechanism not fully resolved (Hyun et al., 2015), but interestingly, morin also showed analgesic effects mediated through CB2R (Jiang et al., 2017).

Arylpyrazoles
Rimonabant (SR141716 , Table 1), a CB1R inverse agonist, elicits alternative cell death pathways depending on the cell type affected. For example, Bifulco (Sarnataro et al., 2006) provides evidence for a lipid raft-mediated mechanism related to the CB1R in MDA-MB-231 cells, whereas it induces apoptosis in colon cancer through a CB1R-independent mechanism that involves the canonical Wnt/β-catenin pathway and β-catenin target genes (Santoro et al., 2009;Proto et al., 2017;Fiore et al., 2018). Rimonabant induces cell cycle arrest and programmed cell death in leukemia cell lines  (Capozzi et al., 2018) Lung cancer: A549 and H460 cells Anti-invasive and antimetastatic action via FAAH inhibition (Winkler et al., 2016) Name/number from the original articles has been preserved. Endogenous and phytogenic cannabinoids' anticancer activity has been further reviewed elsewhere (Fidyt et al., 2016;Fraguas-Sánchez et al., 2016;Hinz and Ramer, 2019).
by a mechanism unlikely to be CB1R-dependent due to a low expression of CB1R in the cell lines used (U937 and Jurkat cells) (Gallotta et al., 2010). Recently, rimonabant has been shown to reduce colon cancer stem cell proliferation, which may account for cancer initiation, progression, and metastasis (Fiore et al., 2018). Synergy with antineoplastic drugs has been explored. A synergic antitumor effect was observed when combining rimonabant and oxaliplatin in colon cancer . AM-251 (Table 1) is a CB1R antagonist structurally closely related to rimonabant. It often has been used as pharmacological tool. For instance, AM-251 allowed determining of the functional relevance of CB1R signaling in Hodgkin lymphoma (Benz et al., 2013) and in rhabdomyosarcoma (Marshall et al., 2011). As an antitumor agent, AM-251 has not been reported to have significant differences with rimonabant.
The naphthoylindole JWH-015 (Table 1) is characterized by high CB2R affinity, but it is not devoid of CB1R activity. CB2R activation has been reported to be involved in the antiproliferative effect of JWH-015 in different cancer cells, such as PC-3 prostate cancer cells (Olea-Herrero et al., 2009). In metastatic breast cancer MCF-7, crosstalk between CB2R and CXCR4 signaling seems to participate in the antiproliferative effect of JWH-015 . In lung cancer cell lines, the effect of JWH-015 is comparable to WIN55,212-2, with CB1R/CB2R agonistmediated antiproliferative effects (Preet et al., 2011). In cancer murine 4T1 and human MCF-7 mammary carcinoma cells, the action of JWH-015 seems to be complex, since it is not mediated either by CB1R or CB2R, or by GPR55, TRPV1, or TRPA1 receptors (Hanlon et al., 2016).

Quinones
Many quinones are cytotoxic through DNA intercalation, inhibition of DNA topoisomerase II enzyme, and free radical production. In this context, phytocannabinoids are interesting starting materials for preparing quinones. Thus, oxidation of CBD, ∆ 8 -tetrahydrocannabinol, and cannabinol leads to para-quinone derivatives respectively named HU-331 (Table 1), HU-306, and HU-345 (Kogan et al., 2004). They all exert antiproliferative activity for Burkitt's lymphoma, T-cell lymphoma, glioblastoma, breast cancer, prostate cancer, lung cancer, and colon cancer. Efforts have been focused on the mechanism of action of HU-331, whose antitumor effect has been shown as not being directly mediated by CB1R or CB2R receptors (Kogan et al., 2006). HU-331 was found to specifically be an inhibitor of topoisomerase II, while having no effect on topoisomerase I (Kogan et al., 2007).

Others
Apoptotic and necrotic cell death have been reported to be associated with elevated levels of AEA (Matas et al., 2007). Thus, inhibition of the enzymes involved in the biodegradation of the endocannabinoids has been shown to play a role in cancer May 2019 | Volume 10 | Article 621 Frontiers in Pharmacology | www.frontiersin.org cell viability, migration, and metastasis, as, for instance, does the FAAH inhibitor URB597 (Table 1) in lung cancer cells (Winkler et al., 2016). In colon cancer, FAAH inhibition (PF-3845, Table 1) seems to be a better strategy than MAGL (JZL184 ,  Table 1) or DAGL inhibition (RHC-80267) (Wasilewski et al., 2017). However, in other studies, JZL184 was shown to have antiproliferative activity in apoptotic LoVo, HCT116, and SW480 cells (Ma et al., 2016).
The 2-(3-hydroxycyclohexyl)phenol CP-55,940 (Table 1), a well-known CB1R/CB2R agonist, has been used as a pharmacological tool for comparing the antineoplastic activity induced by endogenous and synthetic cannabinoids on gastric cancer cells (Ortega et al., 2016). CP-55,940 and AEA induce similar apoptotic effects, whereas Meth-AEA is more effective at inducing necrosis through transient and rapid apoptosis.
The benzo[c]chromene JWH-133 (Table 1), structurally related to ∆ 8 -THC, has been chosen for studying the mechanism of action of synthetic nonpsychotic cannabinoids on breast cancer growth and metastasis due to its selectivity for CB2R (Qamri et al., 2009).
Interestingly, dexanabinol and nabilone (Table 1), synthetic analogues of THC, are the two synthetic molecules that have further progressed in the clinic. While nabilone is approved in certain countries for the treatment of chemotherapy-induced nausea and vomiting (Ware et al., 2008;Velasco et al., 2016;Badowski, 2017), dexanabinol is currently in clinical trials for the management of brain cancer (A Phase 1 Study of Dexanabinol in Patients With Advanced Solid Tumours, ClinicalTrials.gov; Dexanabinol in Patients With Brain Cancer, ClinicalTrials.gov). It is interesting to highlight that dexanabinol acts as an NMDA (N-methyl-D-aspartate) receptor antagonist and an inhibitor of the activity of nuclear factor kappa B (NF-kB) not binding CB1R and CB2R. Therefore, its antitumor molecular mechanisms could be mediated through the aforementioned targets (Striem et al., 1997;Jüttler et al., 2004).
Moreover, CB2R agonists have been the focus of molecular targets associated with photodynamic therapy (PDT) agents for developing target-specific PDT photosensitizers. In this sense, IR700DX-mbc94, a conjugate between a phthalocyanine dye and the CB2R inverse agonist SR144528, showed significant activity in the malignant astrocytoma cell line (Zhang et al., 2014). Another strategy is the co-administration of CB2R agonist and a PDT photosensitizer; synergic effects between the PDT agent IR700DX-6 T and JWH-133 have been observed in triplenegative breast cancer tumors (Zhang et al., 2018a).
Considering that GPR55 promotes cancer cell proliferation, peptide binders of GPR55 have been prepared and studied to inhibit the proliferation of EHEB and DeFew cells, two GPR55positive B-lymphoblastoid cell lines (Mangini et al., 2017). These peptide binders are used as substitute tools for an antibody-based therapy strategy, since there is a lack of humanized monoclonal antibodies for this receptor.

IN SILICO ADMET PROFILE
Besides their activity and antiproliferative profile, pharmacokinetic aspects should be considered in selecting cannabinoid scaffolds for further development towards the oncology scenario. In this context, we have estimated the drug-likeness of the previously listed molecules.
ADMET properties were predicted using QikProp, integrated in the Maestro software (Schrödinger, LLC, New York, 2019) and the admetSAR web server Dong et al., 2018;Yang et al., 2018). Selected parameters are shown in Table 2. These calculations provide a common parameterization of physicochemical descriptors that allows comparison of ADMET profiles, which is a useful criterion for chemical probe selection for further development.
According to our in silico calculations, most of the cannabinoids analyzed herein follow the Lipinski and Jorgensen pharmacokinetic rules (Lipinski, 2001;Jorgensen and Duffy, 2002). It is interesting to underline that our results are consistent with the experimental ADMET parameters published for some of these cannabinoids (Grotenhermen, 2003;Stout and Cimino, 2014;Zendulka et al., 2016). As shown in Table 2, the prediction of human oral absorption, blood-brain barrier permeability, bioavailability, human intestinal permeability, or binding to human serum albumin suggests that these cannabinoids have an appropriate drug profile. However, solubility as well as metabolic and toxicity parameters of specific compounds such as JZL184, LV50, JWH-015, or 3a fall outside the range predicted for FDAapproved small-molecule drugs (Hansen et al., 2009;Zhu et al., 2009;Xu et al., 2012;Li et al., 2014). Certain cannabinoids of phytogenic and synthetic nature may inhibit the activity of one or more cytochrome P450 isoforms ( Table 2). Since these enzymes are involved in over 70% of human drug metabolism (Guengerich, 2008), their interactions with cannabinoids can affect drug clearance, consequently enhancing toxicity. This should be especially taken into account when combining these cannabinoids with other chemotherapy agents. Consequently, when moving forward toward the clinic, selected cannabinoids could be discarded for pharmacokinetic issues.

PERSPECTIVE
The first report on the antitumor activity of phytocannabinoids was published over four decades ago (Munson et al., 1975). Nevertheless, it is only in recent years that interest in these properties has grown. In addition to the well-established palliative effects of cannabinoids in cancer therapy, cannabinoids have attracted attention as possible anticancer drugs. There is a growing body of evidence showing that endogenous, phytogenic, and synthetic cannabinoids, and modulators of endocannabinoid biosynthesis, inhibit proliferation of a wide spectrum of tumor cells. In this report, we aim to provide a perspective of the current drug development scenario of cannabinoid-based antitumor strategies and their potential pathway to the clinic.
Endogenous cannabinoids and their synthetic derivatives have widely exhibited their ability to modulate cell proliferation, angiogenesis, and metastasis in a number of cancer cell types. However, concerning their possible exogenous application May 2019 | Volume 10 | Article 621 Frontiers in Pharmacology | www.frontiersin.org TABLE 2 | Physicochemical descriptors of selected compounds as calculated using QikProp (integrated in Maestro, Schrödinger, LLC, New York, 2019) and the admetSAR web server Dong et al., 2018;Yang et al., 2018).  [-6.5/0.5]. b Predicted log of the brain/blood partition coefficient [-3.0/1.2]. c Prediction of binding to human serum albumin (-1.5-1.5 (Marchese Robinson et al., 2011). g AMES mutagenicity predictions are based on the previously published benchmark data set (Hansen et al., 2009;Xu et al., 2012). h Carcinogenic potency is divided into three classes, labeled as "danger," "warning," and "non-required," according to the TD 50 (median toxic dose) values. Carcinogenic compounds with TD 50 ≤ 10 mg/kg body wt/day were assigned as "danger," those with TD 50 > 10 mg/kg body wt/day were assigned as "warning," and non-carcinogenic chemicals were assigned as "non-required" (Lagunin et al., 2009;Li et al., 2015). i Compounds are classified into four categories based on the criterion of the US EPA (Category I contains compounds with LD 50 values less than or equal to 50 mg/kg; Category II contains compounds with LD 50 values greater than 50 mg/kg but less than 500 mg/kg; Category III includes compounds with LD 50 values greater than 500 mg/kg but less than 5000 mg/kg; Category IV consists of compounds with LD 50 values greater than 5000 mg/kg) (Li et al., 2014). j Predicted median lethal dose (LD 50 ) in rat model (acute toxicity in mol/kg) ). k Metabolism parameters from admetSAR: Molecules were classified as substrate or non-substrate, and inhibitor or non-inhibitor of the different CYP450 isoforms, according to the previously published classification (Carbon-Mangels and Hutter, 2011;Cheng et al., 2011a, Cheng et al., 2011b.

Compd
for cancer treatment, their lipid nature along with possible alterations of the metabolism of the endocannabinoid system may decrease the pharmaceutical interest for this family. In fact, eicosanoid degradation can trigger handling difficulties as well as chemical stability issues. On the other hand, phytocannabinoids and their synthetic analogues are further moving toward the bedside as potential antitumor agents. The natural occurring scaffold has been the most explored for their antiproliferative potential. As previously summarized, compounds isolated from Cannabis such as THC or CBD can reduce tumor growth in vitro and in vivo through different mechanisms depending on the cancer type. Remarkably, pilot studies and early-phase clinical trials indicate positive results regarding the survival of glioblastoma patients upon treatment with combinations of the aforementioned phytocannabinoids (A Pilot Study of Dronabinol for Adult Patients With Primary Gliomas, ClinicalTrials.gov; TN-TC11G (THC+CBD) Combination With Temozolomide and Radiotherapy in Patients With Newly-Diagnosed Glioblastoma, ClinicalTrials. gov; Guzmán et al., 2006).
Besides its promising anticancer potential, the pharmacokinetics of CBD should be taken carefully in the oncological field. This compound can inhibit the cytochrome isoform CYP3A4 (Zendulka et al., 2016), which may alter the metabolism of other drugs when used in combination. This is especially relevant due to the current use of drug cocktails for cancer treatment.
In line with these findings, it is interesting to underline the promising cannabis entourage effect. Synergic antitumor responses have been observed upon cancer treatment with cannabis botanical preparations (Blasco-Benito et al., 2018), exhibiting better results than pure phytocannabinoids administrated separately. Therefore, cannabinoid combinations may provide an improved antiproliferative strategy for cancer management.
Due to intellectual property pharmaceutical aspects, synthetic derivatives of phytogenic cannabinoids such as dexanabinol are also at advanced preclinical stages for the potential treatment of solid tumors (A Phase 1 Study of Dexanabinol in Patients With Advanced Solid Tumours, ClinicalTrials.gov; Dexanabinol in Patients With Brain Cancer, " ClinicalTrials.gov).
Moreover, the cannabinoid-quinones analyzed herein also represent a promising chemotype for anticancer research. In addition to their multitarget antitumor actions, they present a suitable pharmacokinetic profile, being, in our opinion, a good drug-like prototype for further development. As commented, the putative cannabinoid receptor GPR55 is considered an emerging target in cancer therapy (Andradas et al., 2011). Thus, GPR55 antagonists should be explored as antitumor drugs. Moreover, due to the overexpression of this receptor in specific tumors, compounds that specifically bind GPR55 might represent valuable tools as tumor-targeting agents for delivery of classical chemotherapeutic drugs. In this regard, the previously mentioned 1,4-naphthoquinones, such as 3a, or naphthylfenoterols, such as MNF, could be interesting candidates for the pursuit of this cannabinoid anticancer approach.
Arylpyrazoles, aminoalkylindoles, and other cannabinoid derivatives such as napththyridine need additional antiproliferative in vitro and in vivo assays to be considered for further antitumor drug discovery stages. Endocannabinoid enzyme inhibitors such as URB597 and JZL184 should also be further explored since they may provide complementary anticancer strategies.
A major concern when considering new molecules as antitumor agents is their selectivity for cancer cells versus normal cells. Interestingly, phytocannabinoids such as ∆ 9 -THC (Caffarel et al., 2006;Caffarel et al., 2008) and cannabinoid-quinones such as PM49 and 10 ( Morales et al., 2013;Morales et al., 2015) have shown selective toxicity toward cancer cells versus their nontransformed counterparts. This should be taken into account when selecting suitable entities for further development.
The role of the endocannabinoid system in carcinogenesis is not fully unraveled. Therefore, it is difficult to choose a specific cannabinoid chemotype for optimal anticancer drug development. Nowadays, major hopes are coming from phytocannabinoids and their synthetic derivatives since they are steps forward in the clinic race. However, parallel systematic exploration of promising scaffolds presenting optimized ADMET profiles along with diverse mechanistic antiproliferative effects will probably provide wider antitumor spectra.
In this perspective, we have been reporting cannabinoid-based scaffolds as single anticancer agents. However, over the last years, new anticancer strategies toward clinical translation of cannabinoids have been explored. Combinational therapy involving synergies between cannabinoids and other anticancer agents is one of these approaches Torres et al., 2011;Scott et al., 2017;López-Valero et al., 2018;Zhang et al., 2018a). Such combined therapies allow targeting of tumor progression at different levels. Another strategy will be the use of cannabinoids in preventive conditions (Liu et al., 2010;Khan et al., 2018). Since inflammation is a common risk factor for cancer, and some cannabinoids have shown antiinflammatory properties, they could play a role in chemoprevention.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the manuscript and/or the supplementary files.

AUTHOR CONTRIBUTIONS
The authors equally contributed to this manuscript. FUNDING N.J. thanks the Spanish Ministry for its support through MINECO/ FEDER SAF2015-68580-C2-2-R and RTI2018-095544-B-100. P.M. is grateful to the CAM program "Atracción de Talento, " number 2018-T2BMD-10819. We acknowledge support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI).