Cannabinoids and cancer: pros and cons of an antitumour strategy

Authors


Dipartimento di Scienze Farmaceutiche, Università degli Studi di Salerno, Via Ponte Don Melillo, Fisciano 84084, Salerno, Italy. E-mail: maubiful@unina.it or maubiful@unisa.it

Abstract

In the last two decades, research has dramatically increased the knowledge of cannabinoids biology and pharmacology. In mammals, compounds with properties similar to active components of Cannabis sativa, the so called ‘endocannabinoids’, have been shown to modulate key cell-signalling pathways involved in cancer cell growth, invasion and metastasis. To date, cannabinoids have been licensed for clinical use as palliative treatment of chemotherapy, but increased evidences showed direct antiproliferative actions of cannabinoid agonists on several tumour cells in vitro and in animal models. In this article, we will review the principal molecular pathways modulated by cannabinoids on cancer and summarize pros and cons evidence on the possible future use of endocannabinoid-based drugs in cancer therapy.

British Journal of Pharmacology (2006) 148, 123–135. doi:10.1038/sj.bjp.0706632

Abbreviations:
2-AG

2-arachidonoylglycerol

2-LG

2-linoleoyl-glycerol

AA-5-HT

arachidonoyl-serotonin

AEA

anandamide or N-arachidonoyl-ethanolamine

Ang-2

angiopoietin-2

AR

androgen receptor

BRCA

breast cancer associated antigen

CB

cannabinoid receptor

CBD

cannabidiol

COX2

cyclooxygenase-2

CRC

colorectal cancer cells

CYP1A1

carcinogen-metabolizing enzyme

EGF

epidermal growth factor

EGF-R

epidermal growth factor receptor

FAAH

fatty acid amide hydrolase

HBCC

human breast cancer cell

HUVEC

human umbilical vein endothelial cells

MAPK

mitogen-activated protein kinase

MET

R-(+)-methanandamide

Met-F-AEA

met-fluoro-anandamide

MMP

matrix metalloproteinase-2

MPTK-6

rat thyroid carcinoma lung metastasis cells

OEA

N-oleoylethanolamine

PEA

N-palmitoylethanolamine

PG-EAs

prostaglandin-ethanolamides

PI3K

phosphatidylinositol 3-kinase

PIGF

placental growth factor

PKA

phospho-kinase A

PKC

phospho-kinase C

PSA

prostatic-specific antigen

SEA

N-stearoylethanolamine

SR141716A

rimonabant

TGFα

transforming growth factor α

THC

Δ9-tetrahydrocannabinol

TKF

trifluoromethyl-ketone moiety

VEGF

vascular endothelial growth factor

VR

vanilloid receptor

Introduction

The endocannabinoid system, that is, the cannabinoid receptors, endogenous cannabinoid ligands and endocannabinoid-metabolizing enzymes, has drawn a great deal of scientist attention during the past 15 years. The use of cannabinoids in the treatment of cancer chemotherapy side effects was the most studied potential therapeutic application. Powerful chemotherapy side effects can be very severe and intolerable: reported beneficial effects from cannabinoids use, in chemotherapy patients, are a reduced incidence and severity of emesis, appetite stimulation, improvement of cachexia and pain inhibition. Marijuana's major active principle, Δ9-tetrahydrocannabinol (THC), has been licensed for clinical use as palliative treatment for cancer patients, in two preparations, dronabinol and its analogue nabilone. Moreover, mammals produce at least two endogenous compounds anandamide (AEA, N-arachidonoyl-ethanolamine) and 2-arachidonoylglycerol (2-AG) selectively acting on the same receptors as THC.

The ‘endocannabinoid’ system seems to be involved in an increasing number of diseases and to hold promise for development of new therapeutic drugs without psychoactive effects peculiar to THC. Increasing evidence showed a direct antitumour activity of cannabinoid agonists in a plethora of tumour cells including breast, brain, skin, thyroid, prostate and colorectal. This effect was due to the inhibition of tumour growth mediated by cell-cycle arrest or apoptosis, as well as reduction in neovascularization and metastases. When these findings will be supported by in vivo studies, beside their therapeutical implication, they might open new insight on endogenous mechanisms of tumour suppression.

The endocannabinoid system

The discovery of a family of endogenous cannabinoids, named endocannabinoids (Devane et al., 1992; Sugiura et al., 1995), have focused much attention on cannabinoids during the past years. Two different cannabinoid receptors have been cloned from mammalian tissues: cannabinoid receptor 1 (CB1), originally named ‘central’ receptor (Matsuda et al., 1990) and CB2, also incorrectly known as ‘peripheral’ receptor (Munro et al., 1993), and an increasing number of reports and pharmacological evidence suggest that endocannabinoids might also exert their biological effects through non-CB1/CB2 receptors (Di Marzo et al., 2000; Kunos et al., 2000; Maccarrone et al., 2000).

Both the CB1 and CB2 genes encode a seven-transmembrane-domain protein belonging to the Gαi protein-coupled receptor family (Munro et al., 1993): the signal transduction pathway downstream cannabinoid receptors includes adenylate cyclase (Howlett et al., 1986), mitogen-activated protein kinase (MAPK) (Bouaboula et al., 1995) and, in the case of CB1, ion channels (Mackie & Hille, 1992). Whereas CB1 is preferentially expressed in the central nervous system (Matsuda et al., 1990), CB2 has been described as the predominant form expressed in peripheral immune cells (Munro et al., 1993; Galiegue et al., 1995).

The central and most of the peripheral effects of cannabinoids rely on CB1 activation. This receptor is detectable in several brain areas, at very high levels in the basal ganglia, hippocampus, cerebellum and cortex, where it mediates cannabinoid psychoactive effects; its expression during brain development is significantly different from the one observed in the adult stage (Berrendero et al., 1999). CB1 receptors are also present in peripheral nerve terminals, as well as in extra-neural tissues such as testis, uterus, vascular endothelium, eye, spleen, ileum and in adipocytes (Matsuda et al., 1990; Munro et al., 1993; Felder & Glass, 1998; Straiker et al., 1999; Liu et al., 2000; Pertwee, 2000; Cota et al., 2003). The CB2 receptor is believed to be expressed in immune cells and it is unrelated to cannabinoid psychoactive effects (Felder & Glass, 1998). The CB2 is normally expressed in areas enriched of B lymphocytes such as the spleen marginal zone, the lymph node cortex, the nodular corona of Peyer patches and the mantle zones of secondary follicles in tonsils (Munro et al., 1993; Lynn & Herkenham, 1994; Galiegue et al., 1995; Howlett et al., 2002). CB2 receptors were found in microglia cells (Kearn & Hillard, 1997; Walter et al., 2003; Nunez et al., 2004), in glioma and in skin tumour cells (Casanova et al., 2001; Di Marzo et al., 2004). The CB2 receptor is involved in B-cell differentiation and migration of splenic B lymphocytes, suggesting a role for this receptor in the immune response (Galiegue et al., 1995; Carayon et al., 1998). A recent study (Jorda et al., 2004) showed that CB2 was overexpressed in several human myeloid leukaemia cell lines; interestingly, in retrovirus-induced myeloid leukaemia models, the Cb2 gene was located in a common virus integration site, EVI1, suggesting that Cb2 could be a proto-oncogene involved in transformation (Valk et al., 1997). Endogenous ligands for the cannabinoid receptors are lipid molecules containing long-chain polynsatured fatty acids, amides, esters and ethers, with different selectivity for the two receptor types (Mcallister & Glass, 2002; Mechoulam et al., 2002). The best-known endogenous cannabimimetics are AEA (also called Anandamide) and another arachidonate derivative, 2-AG (Devane et al., 1992; Mechoulam et al., 1995; Sugiura et al., 1995). Moreover, N-palmitoylethanolamine (PEA), N-oleoylethanolamine (OEA) and N-stearoylethanolamine (SEA) compounds called ‘endocannabinoid-like’ are present in human, rat and mouse brain (Di Marzo, 1998; Maccarrone & Finazzi-Agrò, 2002) where they might inhibit the degradation of AEA or 2-AG and, consequently, increase their activity (Mechoulam et al., 2002). In the central nervous system, endocannabinoids act as neuromodulators or retrograde messengers (MacDonald & Vaughan, 2001) which inhibit the release of various neurotransmitters (Schlicker & Kathmann, 2001); in the peripheral and neural tissues, they modulated the effects of proteins and nuclear factors involved in cell proliferation, differentiation and apoptosis, as paracrine or autocrine mediators. These data suggested that endocannabinoids could play a role in the control of cell fate (Guzman et al., 2001b).

The most exciting studies reported the potential use of cannabinoids as therapeutic agents (Piomelli et al., 2000; Porter & Felder, 2001). It is now unquestionable that cannabinoids are effective as antiemetic agents in vomiting induced by anticancer drugs (Joy et al., 1999) and increasing evidence suggests the efficacy of cannabinoids for treatment of various diseases such as glaucoma, multiple sclerosis, brain injuries, cardiovascular disorders, chronic inflammation diseases (Mechoulam et al., 2002; Baker et al., 2003; Guzman, 2003; Mendizabal & Adler-Graschinsky, 2003; Kunos & Pacher, 2004; Tomida et al., 2004). Hopes for these possible applications encouraged the development of new synthetic cannabinoid-related drugs capable of a more selective activation of cannabinoid receptors. Principal compounds and their actions are summarized in Table 1. To date, these substances have been extensively used, both in vitro and in vivo, as pharmacological tools to obtain more detailed insight of cannabinoid action, in order to evaluate their potential clinical use. There is mixed evidence on the effects of cannabinoids on cancer: in vitro and in vivo studies and clinical data showed both antineoplastic and protumoral activity, depending on type of agonist, target tissues, route of administration, doses and duration of the treatment.

Table 1.  Properties of cannabinoid-related drugs
CompoundTarget(s)Potential therapeutic applications
CP-55,940Nonselective agonist (CB1=CB2)Analgesic, antiemetic, appetite stimulant, tumour growth inhibitor, multiple sclerosis
WIN 55,212-2Nonselective agonist (CB1=CB2)Analgesic, antiemetic, appetite stimulant, tumour growth inhibitor, multiple sclerosis
HU-210Nonselective agonist (CB1=CB2)Analgesic, multiple sclerosis, neuroprotective
Δ9-THCNonselective agonist (CB1>CB2)Analgesic, antiemetic, appetite stimulant tumour growth inhibitor
AnandamideNonselective agonist (CB1≫CB2)Analgesic, antiemetic, appetite stimulant, tumour growth inhibitor
(R)-methanandamideNonselective agonist (CB1≫CB2) metabolically stableAnalgesic, antiemetic, appetite stimulant, tumour growth inhibitor
2-AGNonselective agonist (CB1>CB2)Analgesic, antiemetic, appetite stimulant, tumour growth inhibitor
O-1269Partial CB1 agonist 
Noladin etherSelective CB1 agonistAnalgesic
AM-1241Selective CB2 agonistsTumour growth inhibitor (in glioma, skin carcinoma, lymphoma and leukaemia); multiple sclerosis immune diseases peripheral analgesia
HU-308  
JWH-133  
JWH-015  
BML190Nonselective agonist (CB2≫CB1) 

In this article, we will review the principal molecular pathways modulated by cannabinoids in cancer cells and summarize pros and cons evidence for a possible use of cannabinoid-based drugs in cancer therapy in the future.

Cannabis smoke intake and cancer

Several studies produced exciting new leads in the search for anticancer treatments using cannabinoid-related drugs. Plant-derived (THC), synthetic (HU210, WIN-55,212-2), and endogenous (2-AG, AEA) cannabinoids modulate tumour growth, apoptosis, migration and neoangiogenesis in various types of cancer (Bifulco & Di Marzo, 2002; Guzman et al., 2002). However, studies performed to investigate marijuana-smoking effects on carcinogenesis and tumour growth produced contradictory results (Table 2): THC failed to induce mutagenicity in the Ames test (Hall & MacPhee, 2002) and in skin test in mice (Chan et al., 1996), whereas cannabis smoke was mutagenic in vitro (MacPhee, 1999; Marselos & Karamanakos, 1999). The Ames test is a sensitive biological method for measuring the potentially carcinogenic effect of chemical substances on microrganisms, cells and tissue cultures. This test by itself does not demonstrate cancer risk; however, mutagenic potency evaluated by Ames test does correlate with the carcinogenic potency for chemicals in rodents. These results show that THC have no carcinogenic properties, at least as purified compound. Moreover, evidence showed that smoking of cannabis preparations caused cancer of the respiratory and oral tracts or, at least, potentiated tobacco smoke-induced damages. Various mechanisms have been involved in these processes: direct THC-induced damage of the bronchial epithelium (Barsky et al., 1998), induction and regulation of the carcinogen-metabolizing enzyme CYP1A1 (Roth et al., 2001), alteration of the balance between apoptotic and necrotic cell death (Sarafian et al., 2001), increase of cellular oxidative stress (Sarafian et al., 1999), CB2-mediated immune suppression (Srivastava et al., 1998; Zhu et al., 2000). Recently, Hall et al. (2005) extensively reviewed the results of epidemiological studies reporting inconsistent association between cannabis smoking and lung cancer. The author highlighted the need of a case–control cohort larger than those previously examined, excluding concomitant risk factors as alcohol use or tobacco smoke. Furthermore, the cannabis smoking and the medical use of cannabinoids have been largely mistaken in public debate: the recreational long-term cannabis smoking, potentially but to date ambiguously connected with respiratory and oral cancer, is not univocally associated with pharmaceutical cannabinoids exploitable for medical purposes.

Table 2.  Potential use of cannabinoids in cancer treatment: pro and cons evidence
Tumour (cell type)Cannabinoid (concentration or dose)Anticancer effectProcancer effectMechanism of actionReferences
Bronchial epitheliumTHC +Molecular abnormalities and histopatological alterationsBarsky et al. (1998)
Murine hepatoma cell line (Hepa)THC (2–10 μg/ml) +Induction of CYP1A1Roth et al. (2001)
Lung cancer cell line (A549)THC +Inhibition of Fas-induced caspase-3 activitySarafian et al. (2001)
Endothelial cell lineTHC (1.77 or 3.95%) +Increased ROS generationSarafian et al. (1999)
Murine Lewis lung carcinoma (3LL); alveolar cell carcinoma (L1C2)THC (5–40 mg/kg) +In vivo, decreased production of cytokines and/or CB2-mediated immune suppressionZhu et al. (2000)
 CBD (geqslant R: gt-or-equal, slantedμg/ml) + Srivastava et al. (1998)
Human breast cancer cell lines (MCF7; EFM-19)AEA (2–10 μM)
2-AG (2–10 μM)
HU210 (geqslant R: gt-or-equal, slantedμM)
+ Inhibition of the mitogen-induced stimulation of the G0/G1–S phaseDe Petrocellis et al. (1998)
 AEA (geqslant R: gt-or-equal, slantedμM)+  Melck et al. (2000)
 2-AG, HU210 (geqslant R: gt-or-equal, slantedμM)    
Human breast cancer cell lines (MCF7; MDA-MB-231)
Mouse mammary carcinoma (4T1)
THC (leqslant R: less-than-or-eq, slantμM) +Increased tumour growth and metastasis; in vivo, decreased antitumour immune responseMcKallip et al. (2005)
Androgen-independent prostate cancer cells (PC3, DU145)AEA, R-(+)-MET (geqslant R: gt-or-equal, slantedμM)+ Inhibition of mitogen-induced proliferation, G1 arrestMimeault et al. (2003)
Melck et al. (2000)
 THC (1 μM)+ ApoptosisRuiz et al. (1999)
Androgen-dependent prostate cancer cells (LNCaP)AEA, R-(+)-MET (geqslant R: gt-or-equal, slantedμM)+ Inhibition of mitogen-induced proliferation, G1 arrestMimeault et al. (2003)
Androgen-dependent prostate cancer cells (LNCaP)WIN-55,212-2 (geqslant R: gt-or-equal, slanted2.5 μM)+ Dose- and time-dependent induction of apoptosis; decreased expression of AR and PSASarfaraz et al. (2005)
Androgen-dependent prostate cancer cells (LNCaP)R-(+)-MET (0.1–0.2 μM) +Increased proliferation and AR expressionSanchez et al. (2003)
Rat glioma cell line (C6)THC (1 μM)+ Apoptosis via ceramide de novo synthesis
In vivo, regression of C6-derived glioma
Galve-Roperh et al. (2000)
 JWH133, WIN-55,212-2 (0.1 μM)+ Apoptosis via ceramide de novo synthesisSanchez et al. (2001a, 2001b)
 WIN-55,212-2 (15 μM)+ Apoptosis via activation of caspase cascadeEllert-Miklaszewska et al. (2005)
Human astrocitoma (grade IV)JWH-133 (50 μg/die)+ In vivo, inhibited growth of tumours induced in deficient miceSanchez et al. (2001a, 2001b)
Human glioblastoma multiforme cell line (GBM)THC (1 μM)
WIN-55,212-2
+ Decreased proliferation and increased cell deathMcAllister et al. (2005)
K-ras-transformed FRTL-5 thyroid cells (KiMol)Met-F-AEA (0.5 ng/kg/dose)+ In vivo, inhibited growth of tumours induced in nude miceBifulco et al. (2001)
Mouse skin carcinoma cells (PDV-C57)JWH-133, WIN-55,212-2 (1.58 μg)+ In vivo, inhibited growth of tumours induced in nude miceCasanova et al. (2003)
Human umbilical vein endothelial cells (HUVEC)JWH-133 (25 nM)+ Induction of apoptosis, inhibited migrationBlazquez et al. (2003)
Lung cancer cells (NCI-H292)THC (0.1–0.3 μM) +Increased proliferationHart et al. (2004)
Glioblastoma cell line (U373-MG)     
Human breast cancer cell line (MDA-MB-231)Met-F-AEA (10 μM and+ Inhibition of adhesion and migrationGrimaldi et al. (2006)
Mouse breast cancer cell line (TSA-E1)0.5 mg/kg/dose)  In vivo, reduction of number and dimension of metastatic nodes 

Effects of cannabinoids on tumour biology: modulation of key cell-signalling pathways involved in control of cell fate

Breast and prostate cancer and cannabinoids

Studies performed in order to understand the role of endocannabinoids and their receptors in the control of cell fate raised great interest (Table 2). In 1998, De Petrocellis et al. investigated the possible antimitogenic effects of AEA on epithelial human breast cancer cell (HBCC) lines EFM-19 and MCF-7, expressing oestrogen and prolactin receptors and proliferating in response to steroid or lactogenic hormones treatments (Simon et al., 1985; Clevenger et al., 1995). In these models, treatment with submicromolar concentration of AEA (as well as of 2-AG or HU-210) significantly inhibited the G1–S transition of mitotic cell cycle. Moreover, anandamide inhibited the expression of prolactin receptor, induced downregulation of the brca1 gene product (De Petrocellis et al., 1998), and of trk proteins, the high-affinity neurotrophin receptors (Melck et al., 1999b, 2000). The antiproliferative CB1-receptor-mediated effect was AEA dose-dependent and proportional to the degree of hormone dependency of the used HBCC line (De Petrocellis et al., 1998). The block of the G1–S transition was ascribed to the inhibition of adenylyl cyclase and, consequently of cAMP-protein kinase A pathway and to the activation of MAPK (Melck et al., 1999b). Cannabinoids prevented the inhibition of RAF1 (caused by protein kinase A-induced Raf phosphorylation) and induced prolonged activation of the RAF1-MEK-ERK signalling cascade, leading to downregulation of PRLr and Trk (Melck et al., 2000).

On the other hand, a recent report (McKallip et al., 2005) demonstrated that HBCC lines MCF-7 and MDA-MB-231, and the mouse mammary carcinoma 4T1, are resistant to THC-induced cytotoxicity. The authors hypothesized that the degree of tumour sensitivity to THC may be related to the level of CB1 and CB2 expression, and that THC exposure may lead to an increase in growth rate and metastatic potential of tumours with low to no expression of cannabinoid receptors. It is an unsurprising data that different clones of the same cell lines, as well as of breast cancer cells, showed very variable levels of receptors and a different responsivity to hormone and growth factors (Hamelers et al., 2003). Cannabinoid receptors expression could be at least in part modulated by the culture conditions and the number of subculturing passages, even in the absence of specific ligands (Melck et al., 2000). In addition, McKallip et al. specify that 4T1 cells expressed high levels of vanilloid receptor (VR1), a nonselective cation channel, activated by capsaicin, which is also a characterized target for AEA. This observation could be very interesting because these breast cancer cells may be more sensitive to AEA (Melck et al., 1999a; Smart et al., 2000; Zygmunt et al., 2000), rather than to THC.

Proliferative disorders of the prostatic gland involve multistep process and sequential changes in the responsiveness of prostate epithelial cells to steroid hormones, growth factors and neuropeptides (Marker et al., 2003). Several intraepithelial or invasive prostatic cancers showed increased expression of epidermal growth factor receptor (EGF-R) tyrosine kinase, EGF and transforming growth factor α (TGFα) (Liu et al., 1993; Ware, 1993; Kim et al., 1999). Moreover, androgen-independent human prostate cancer cell lines PC3 and DU145 overexpressed EGF-R, which, via a selective interaction with autocrine and paracrine-secreted EGF and TGFα, promoted cell proliferation. In these models, androgen and EGF downregulated p27kip, an inhibitor of cyclin-dependent protein kinases (Peng et al., 1996; Wu et al., 1996; Ye et al., 1999). Mimeault et al. (2003) showed that a micromolar concentration of AEA inhibited EGF-induced proliferation of DU145 and PC3 cells, as well as of androgen-stimulated LNCaP, via G1 arrest, and downregulated EGF-R levels. Both phenomena were CB1-mediated. Similar growth arrest and receptor modulation were also reported for prolactin- and nerve growth factor-stimulated DU145 (De Petrocellis et al., 1998; Melck et al., 2000). It is important to remark that longer AEA-incubation times (5–6 days) were able to induce massive apoptosis in DU145 and PC3 cells. This effect was mediated by CB1/2 via cellular ceramide accumulation, and was absent in LNCaP cells (Mimeault et al., 2003). Furthermore, micromolar WIN-55,212-2 treatment significantly decreased LNCaP cells viability and androgen receptor (AR) expression in a dose- and time-dependent manner, with maximal effect at 72 h (Sarfaraz et al., 2005). The authors described also a decrease in intracellular as well as in secreted levels of prostatic-specific antigen (PSA), an androgen-receptor-regulated glycoprotein (Montgomery et al., 1992) that currently is the most-accepted marker for assessment of prostate cancer progression (Stamey et al., 1987). Their results showed that treatment of LNCaP cells with WIN-55,212-2 also inhibited vascular endothelial growth factor (VEGF) protein expression, an ubiquitous cytokine with a key role in angiogenesis (Blazquez et al., 2003). Dose- and time-dependent effects of cannabinoids are a crucial issue to debate. It is puzzling that a 4-day treatment with R-(+)-methanandamide (MET) or exogenous cannabinoids, at submicromolar concentrations, increased the proliferation rate of LNCaP cells and the expression of AR, whereas longer incubation periods led to differentiation (Sanchez et al., 2003). Apparently, MET-induced mitogenic effect was phospho-kinase C (PKC)- rather than cAMP-pathway dependent; furthermore, in this cellular model, the androgen receptor expression was CB1- and, partially, CB2-mediated (Sanchez et al., 2003; Sarfaraz et al., 2005).

Depending on drug concentration, cannabinoids may either inhibit or stimulate cancer cell proliferation. Hart et al. (2004) found that treatment of several cancer cell lines with nanomolar concentration of THC, AEA, HU-210 or WIN-55,212-2 induced increased proliferation that was dependent on EGF-R phosphorylation. These data suggested that in a variety of human cancer cell lines CB1/CB2 receptors are linked to MAPK and AKT/PKB activation, and that cannabinoid concentrations could have dramatic effects in the cellular choice between proliferation and cell growth arrest.

Glioma and cannabinoids

The antitumoral action of cannabinoids on glioma may be exerted either via the CB1 or the CB2 receptor. THC induced apoptosis of C6 glioma cells by a pathway involving CB1 receptor, sustained generation of the proapoptotic lipid ceramide and prolonged activation of Raf1/MEK/ERK cascade (Galve-Roperh et al., 2000). A role for BCL-2 family members, such as Bad, have also been hypothesized (Ellert-Miklaszewska et al., 2005). Galve-Roperh et al. (2000) showed that cannabinoids induced regression of gliomas in vivo. In their model, intratumour administration of THC and WIN-55,212-2 induced regression of C6-derived glioma in Wistar rats and in RAG-2-deficient mice. In this study, they showed that cannabinoid administration induced no substantial modification in behavioural parameters, in food and water intake or in body weight; neurotoxicity nor markers of tissue damage have been revealed for at least 2 months after cannabinoid treatment. Moreover, selective CB2 agonists showed good in vivo efficacy on regression of highly malignant human astrocitoma (grade IV) (Sanchez et al., 2001a). Ramer et al. (2003) demonstrated that cannabinoids induced the expression of cyclooxygenase-2 (COX-2) in human neuroglioma cells via a cannabinoid-receptors independent pathway, probably linked to lipid raft microdomains (Hinz et al., 2004). Since COX-2 can inhibit apoptosis (Tsujii & Dubois, 1995), these findings could demolish the promising effects of potential cannabinoids use in human gliomas, but additional studies showed that COX-2 induction may sensitize cells to apoptotic death (Corasaniti et al., 2000; Na & Surh, 2002) or rather finely regulate the cell choice between proliferation and death (Ramer et al., 2003). Cannabinoid receptors could have a protective role against programmed cell death, as reported in human neuroblastoma and C6 cells, where AEA induced apoptosis, via vanilloid receptors, increasing intracellular calcium concentration, activating COX, releasing cytochrome c and activating caspase 3 (Maccarrone et al., 2000). The mechanism through which AEA induces apoptosis in cells expressing both functional cannabinoid and vanilloid receptors is still controversial and might depend on the experimental conditions used. In fact, Jacobsson et al. (2001) showed that in rat glioma C6 cells, the AEA antiproliferative effect was associated with a combined activation of cannabinoid and vanilloid receptors and it was difficult to exclude a cannabinoid receptor role in the AEA-induced apoptotic cell death.

Cannabinoid receptor expression and endocannabinoid levels in transformed versus normal cells

Cannabinoid receptor levels seem to be a fundamental element for growth inhibitory effects. It has been documented that the expression of CB1 receptor was regulated in an opposite way in normal or transformed cells. Bifulco et al. (2001) demonstrated that met-fluoro-anandamide (Met-F-AEA) increased the levels of CB1 receptors in both K-ras-transformed FRTL-5 (KiMol) cells and in KiMol-derived tumours in nude mice, whereas in FRTL-5 cells, a thyroid-differentiated epithelial cell line, Met-F-AEA produced downregulation of CB1 receptors. Furthermore, cannabimimetic substances inhibited the proliferation of KiMol cells more strongly than of FRTL-5 cells; in vivo, Met-F-AEA inhibited growth of KiMol-induced tumours in athymic mice. These effects were accompanied by reduction of p21ras activity.

Apparently, an opposite regulation of CB1 expression in transformed versus normal cells was a common mechanism: THC induced apoptosis in several human cancer cell lines but showed less efficacy in nontransformed cell counterparts (Sanchez et al., 1998; Ruiz et al., 1999; Galve-Roperh et al., 2000; Guzman et al., 2001a; McAllister et al., 2005). Finally, cannabinoids protected oligodendroglial cells from various proapoptotic stimuli (Molina-Holgado et al., 2002) and astrocytes from ceramide-induced sensitization to oxidative damage (Carracedo et al., 2004), whereas they induced apoptosis of glioma cells (Galve-Roperh et al., 2000; Sanchez et al., 2001b; Gomez del Pulgar et al., 2002). A recent study showed a different endocannabinoid metabolism in human glioblastoma and meningiomas (Petersen et al., 2005): glioblastoma were characterized by increased levels of AEA and decreased fatty acid amide hydrolase (FAAH) activity, while meningiomas showed enhanced levels of 2-AG compared to human nontumour brain tissue. The authors suggested that modulation of endocannabinoids in these tumour tissues could be an endogenous antiproliferative mechanism acting through selective cannabinoid receptor activation. Even if this hypothesis is not demonstrated yet, similar mechanisms have been suggested in colon cancer cells by Ligresti et al. (2003). The different endocannabinoid metabolism in normal compared to tumour cells and the different effects exerted by endocannabinoids is an unquestionable issue, probably connected with physiological fundamental properties, which could be a possible means to control tumour growth. Finally, it is interesting to remark that cannabinoids cannot induce significant changes in the survival of nontransformed epidermal cell lines MCA3D, HaCat and of primary human keratinocytes, whereas they block in vivo the growth of highly malignant PDV.C57-derived tumours (Casanova et al., 2003).

Cannabinoid hydrolysis and reuptake inhibitors

A series of compounds, such as palmitoylethanolamine, might act as ‘entourage’ substances enhancing cannabinoid biological actions. Di Marzo et al. (2001) reported that chronic treatment with PEA enhanced the AEA-induced inhibition of HBCC proliferation decreasing the expression of FAAH, the enzyme mainly responsible for AEA degradation. Similar results were obtained with HU210, which cannot be hydrolysed by FAAH, suggesting that PEA could also enhance the vanilloid VR1 receptor-mediated effects of AEA on calcium influx into cells (De Petrocellis et al., 2000, 2002; Di Marzo et al., 2002). Recent studies in colorectal cancer cells in vitro (Ligresti et al., 2003), and in thyroid carcinoma cells in vitro and in vivo (Bifulco et al., 2004), argue for a therapeutic anticancer strategy aimed at raising the levels of endocannabinoids by preventing their cellular reuptake and enzymatic degradation. VDM11, a selective inhibitor of endocannabinoid cellular reuptake, and arachidonoyl-serotonin (AA-5-HT), a blocker of endocannabinoid enzymatic hydrolysis, both inhibited the in vitro growth of rat thyroid-transformed cells (KiMol), and in vivo of tumour xenografts induced by subcutaneous injection in mice of the same cell line (Bifulco et al., 2004). Other evidence demonstrated that a decreased 2-AG hydrolysis inhibited invasion of androgen-independent cancer cells (Nithipatikom et al., 2005) and Ben-Shabat et al. (1998) showed that 2-acyl-glycerol esters, such as 2-linoleoyl-glycerol (2-LG), potentiated the central biological activity of 2-AG in various normal murine tissues. Given that cannabinoid receptors expression and/or endocannabinoids levels are altered in certain malignancies, as in gliomas, astrocytomas and transformed thyroid epithelium, it would be plausible to argue that endocannabinoids exert a tonic control of tumour growth. Thus the inhibitors of cannabinoid inactivation and reuptake might be considered as new tools for therapeutic intervention.

Effects of cannabinoids on tumour progression

Modulation of angiogenesis

Angiogenesis, providing nutrients to proliferating cancer cells, is a critical event involved in the progression of solid tumours. Positive and negative regulators of angiogenesis could be produced by cancer cells, by vascular endothelial cells, by infiltrating inflammatory cells and by the extracellular matrix (Kuroi & Toi, 2001; Distler et al., 2003).

Increasing evidence suggests that antitumour effect of cannabinoid-related drugs could be at least in part ascribed to inhibition of tumour neoangiogenesis in animal models. The nonpsychoactive CB2-agonist cannabinoid JWH-133 inhibited in vitro human umbilical vein endothelial cells (HUVEC) migration and survival (Blazquez et al., 2003); in vivo JWH-133 treatment of C6 glioma- and grade IV astrocytoma-derived tumours reduced expression levels of angiopoietin-2 (Ang-2), VEGF, and matrix metalloproteinase-2 (MMP) (Blazquez et al., 2003), three proangiogenic factors that destabilize vessel integrity, facilitate vessel sprouting and endothelial cells growth, disrupte the extracellular matrix organization, respectively. These findings were confirmed by cDNA array analysis showing that JWH-133 administration to mouse downregulated in gliomas genes related to angiogenesis, hypoxia and metastasis and increased the expression of metalloproteinase substrates involved into matrix remodelling, probably via ceramide de novo synthesis (Blazquez et al., 2004).

Several authors (Rak et al., 1995; Casanova et al., 2002) suggested that oncogenes, such as mutant ras, may have an impact on tumour growth and progression through upregulation of VEGF, a common element of the ras-dependent angiogenic phenotype (Grunstein et al., 1999). Casanova et al. (2003) evaluated the potential antiangiogenetic power of cannabinoids in mouse skin carcinoma cell line (PDV-C57) expressing high levels of activated ras and EGF-R and showed that WIN-55,212-2 or JWH-133 were able to arrest in vivo the growth of highly malignant PDV-C57 cells-derived tumours: in this model, cannabinoid treatment decreased the expression of proangiogenetic factors VEGF, Ang2 and placental growth factor (PIGF). Similarly, Met-F-AEA, by inhibiting p21ras activity, prevented the growth of v-K-ras-transformed rat thyroid cells both in vitro and in vivo (Bifulco et al., 2001). Furthermore, it inhibited growth of already established tumours by reducing the expression of both VEGF and its receptor Flt1, and upregulating the levels of the cyclin-dependent kinase inhibitor p27kip (Portella et al., 2003).

Modulation of cancer cell migration and metastasis

Cell migration plays important role in many physiological and pathological processes, including angiogenesis, tissue repair, metastasis and inflammation (Lauffenburger & Horwitz, 1996). The ability to mediate cell migration may be shared by many Gi protein-coupled receptors (Neptune & Bourne, 1997).

Cannabinoid variable effects on cell migration seem to be dependent on both cellular differentiation levels and specific activation of different receptors. Song & Zhong (2000) demonstrated that cannabinoid agonists (HU210, WIN 55212-2, AEA) induced migration of human embrionic kidney 293 cells. The anandamide-induced cell migration was CB1-mediated in human embrionic kidney 293 cells and it was blocked by PD98059 (MAPK inhibitor), suggesting that ERK, rather than adenylate cyclase, was crucial for CB1-mediated migration. On the other hand, the antitumour effects of cannabidiol (CBD), a nonpsychoactive cannabinoid, could be ascribed, beside to the antiproliferative action on U87 and U373 human glioma cells in vitro and in vivo (Massi et al., 2004), to inhibition of migration. In U87 cells, such inhibition did not involve classical Gi/o protein-coupled cannabinoid receptors (Vaccani et al., 2005). Moreover, Met-F-AEA was able to inhibit proliferation of a metastasis-derived thyroid cancer cell line, MPTK-6, more efficaciously than of the primary thyroid cancer-derived TK-6 cells (Portella et al., 2003). To test the in vivo effects of Met-F-AEA on induction of metastatic foci, the authors used the Lewis lung carcinoma model of metastatic spreading and demonstrated that Met-F-AEA efficaciously interfered with the formation of lung metastatic nodules by acting on CB1 receptors. Recently, our group demonstrated that Met-F-AEA treatment inhibited both adhesion and migration of the highly invasive metastatic breast cancer cell lines MDA-MB-231 and TSA-E1, by in vitro testing in an adhesion and migration assay on type IV collagen, the major component of the basement membrane. Furthermore, Met-F-AEA treatment significantly reduced number and dimension of metastatic nodes induced by TSA-E1 cell injection in syngenic mice (Grimaldi et al., 2006).

In androgen-independent prostate cancer cell lines PC3 and DU145, 2-AG reduced invasion through the CB1-dependent inhibition of adenylyl cyclase, decreasing phospho-kinase A (PKA) activity (Nithipatikom et al., 2004). Compounds containing a trifluoromethyl-ketone moiety (TKF), by blocking 2-AG hydrolysis, were able to efficaciously decrease prostate cancer cells spreading (Nithipatikom et al., 2005).

The cannabinoid-modulated migration could finely regulate immunological antitumour responses. Interestingly, in differentiated HL-60 leukemia cells, 2-AG induced a significative production of chemokine (Kishimoto et al., 2004; Sugiura et al., 2004), caused rapid actin rearrangement and morphological changes, such as extension of pseudopods and increased migration (Kishimoto et al., 2003; Gokoh et al., 2005). Moreover, 2-AG stimulated migration of NK cells (Kishimoto et al., 2005), splenocytes, B lymphoid cells and myeloid leukaemia cells (Jorda et al., 2002). 2-AG-induced migration was CB2 receptor-dependent and in B cells was enhanced by CD40 costimulation (Rayman et al., 2004).

Most studies in vitro and in vivo indicated that THC is immunosuppressive on macrophages, NK cells and T lymphocytes (Bhargava et al., 1996; Klein et al., 1998; McCoy et al., 1999): in murine lung cancer models, THC could promote, rather than suppress, tumour growth inhibiting antitumour immunity by a CB2 receptor-mediated cytokine-dependent patway (enhanced IL-10 and TGFβ, reduced IL2 and IFN-γ) (Zhu et al., 2000).

Noteworthy, AEA alone had no effect on the migration of leukocytes, HL-60 and monocytes (Kishimoto et al., 2003), whereas stimulated embryonic kidney, microglial and myeloid leukaemia cells transfected with the CB-2 receptor gene (Jorda et al., 2003). Moreover, AEA could inhibit chemokine-induced migration of CD8+ T lymphocytes and of SW480 colon carcinoma cells through activation of distinct cannabinoid receptors: CB2 in lymphocytes and CB1 in colon carcinoma cells, respectively, suggesting that specific inhibition of tumour cells migration could be obtained without significant effect on the immune system at least in colon cancer (Joseph et al., 2004).

Multifaceted role of COX-2 and cannabinoids in tumour progression

The enzyme COX catalyses the conversion of arachidonic acid to PGH2, an endoperoxide that functions as precursor of prostaglandins (PGs) and tromboxane (TX). The constitutive isoform, COX-1, is ubiquitous and responsible for physiological functions; COX-2, the isoform expressed by cells involved in inflammation (macrophages, monocytes, platelets) can be dramatically induced by a variety of stimuli (Morita, 2002).

Recent data showed that COX-2-derived prostaglandins modulated the production of proangiogenetic factors in colon cancer cells (Hinz & Brune, 2002), and that COX-2 overexpression could be a common mechanism, identified in a number of epithelial cancer cells (for a review see Prescott & Fitzpatrick, 2000; Romano & Clària, 2003; Zha et al., 2004), resulting in resistance to apoptosis (Tsujii & Dubois, 1995), increased invasiveness (Tsujii et al., 1997) and tumour angiogenesis (Tsujii et al., 1998). Increasing evidence suggested that selective COX-2 inhibitors may represent novel chemopreventive tools (for a review, see Ruegg et al., 2003). A very intriguing hypothesis for the possible role of endocannabinoids on the control of tumour angiogenesis has been proposed by Ligresti et al. (2003). They found that 2-AG and AEA concentration was increased in colorectal cancer cells (CRC) compared to normal mucosal tissue and they proposed that these compounds might act as endogenous growth inhibitors through two distinct mechanisms: that is, by stimulation of cannabinoid receptors and by reducing prostaglandins production, since they efficaciously competed with COX-2 substrates (Marnett, 2002). The CB receptor independent effect of anandamide was investigated in colon cancer cell by Patsos et al. (2005). They showed that AEA significantly reduced the growth of COX-2-expressing HT29 and HCA7/C29 CRC cell lines and that COX-2 produced metabolites of AEA, prostaglandin-ethanolamides (PG-EAs), which induced apoptosis in CRC cells. Since PG-EAs production was increased in AEA-treated cells and COX-2 selective inhibitors partially attenuated AEA-induced cell death, the authors suggested that a combination of factor, including PG-EAs and COX-2 metabolites, could play a role, at least in part, in the antiproliferative properties of AEA (Patsos et al., 2005). On the other hand, an increased COX-2 expression has been associated with poor prognosis in lung cancer (Achiwa et al., 1999), and it is induced by Methanandamide in murine lung cancer via a cannabinoid receptor-independent pathway (Gardner et al., 2003).

Causal relationship between overexpression of COX-2 and carcinogenesis has been demonstrated in breast, colon and in lung cancer. In human breast cancer cell lines, COX-2 and PGE2, the major COX-2 products, were poorly expressed in the MCF-7 cell line and overexpressed in the metastatic cell line MDA-MB-231 (Liu & Rose, 1996); in human breast tumours, a significant correlation between COX-2 expression and aromatase (the enzyme catalysing oestrogen production from androgens) expression has been found (Brueggemeier et al., 1999); finally, clinical studies showed a strong association of high metastatic potential and lack of oestrogen and progesterone receptors with high PGE2 concentration (Rolland et al., 1980).

COX-2 overexpression was detected in human colorectal carcinoma compared with normal epithelium (Eberhart et al., 1994; Elder et al., 2002) and PGE2 was increased in human colorectal carncer tissue (Rigas et al., 1993). Evidence of the role played by COX-2 and PGE2 in precancerous lesions and in cancer growth was provided from clinical (Kune et al., 1988; Steinbach et al., 2000) and animal (Kawamori et al., 1998; Oshima et al., 2001) studies.

The nonsteroidal anti-inflammatory drugs (NSAIDs), particularly the selective COX-2 inhibitors (coxibs) have been proposed as anticancer agents. In fact selective COX-2 inhibitors suppressed the growth of human colon and epithelial cancers (for a review, see Koki et al., 2002) and they could enhance the response to conventional anticancer therapies (Moore et al., 2000; Trifan et al., 2002).

Taken together, these data suggested that specific COX-2 inhibitors might be used as adjuvants in the treatment of tumours as well as in cancer prevention. However, COX-2-derived products have a variety of protective properties: the prostacyclin PGI2 exert antioxidant effect which may retard atherogenesis (Pratico et al., 1998) and contributes to the atheroprotective effect of oestrogen (Egan et al., 2004); PDG2 and PGE2 showed hepato-protective function in a murine model of pharmacological-induced acute liver injury (Reilly et al., 2001); adiponectin induced COX-2-dependent synthesis of PGE2 protects the heart from ischemia–reperfusion injury (Shibata et al., 2005). In this scenario, some advantage could be offered by the use of endocannabinoids compared to selective COX-2 inhibitors: (a) AEA induced nonapoptotic cell death in high COX-2-expressing colorectal tumour cells (Patsos et al., 2005) and in prostate carcinoma cells (Mimeault et al., 2003); these properties could be beneficial in treating tumour cells that have become resistant to induction of apoptosis; (b) normal cells which do not express COX-2 were resistent to endocannabinoid induced cell death (Patsos et al., 2005); (c) AEA neither increased COX-2 levels nor inhibited its activity, at least in CRC cells, but acting as substrate (Marnett, 2002; Ligresti et al., 2003; Patsos et al., 2005) it might preserve the protective effects of COX-2-derived products.

Conclusions

Presented findings suggest that cannabinoids exert a number of effects depending on cell types, activation of signal transduction pathways, route of drug administration, timing of drug delivery and, last but not least, responsivity of tumour and normal cells.

Epidemiological studies reported inconsistent association between cannabis smoke and cancer, and administration of high oral doses of THC in rats or mice did not increase tumour incidences in a 2-year study (Chan et al., 1996). In animal models, cannabinoids exert a direct antiproliferative effect on tumours, but they could indirectly enhance tumour growth via inhibition of immunogenicity (for immunosuppressive effect of cannabinoids, see Klein, 2005). The typical immunosuppressive effect of THC is an unquestionable topic imposing caution in the dosage and administration timing of CB2-receptor-selective compounds (Klein et al., 2000; Salzet et al., 2000).

The immunosuppressive properties of plant-derived cannabinoids could enhance tumour cell proliferation (Zhu et al., 2000; McKallip et al., 2005) and accelerate cancer progression in patients, but the biological response to cannabinoids critically depends on drug concentration and cellular context (Hart et al., 2004). Nevertheless, different therapeutic strategies could be developed on the basis of peculiar characteristics expressed by several malignancies. Jones & Howl (2003) suggested as therapeutic target for tumour intervention some distinctive properties: (1) in cancer, such as malignant astrocytomas, gliomas, breast, thyroid, prostate, where cannabinoid receptor expression is enhanced, strategies aimed at raising levels of endocannabinoids could be a successfully treatment; (2) in colorectal carcinoma, the increased expression of endocannabinoids suggests that inhibitors of endocannabinoid metabolism could be used as therapeutic tools; (3) upregulation of CB2 receptor expression in malignant astrocytomas and gliomas and/or the increased CB2/CB1 ratio in tumours of immune origin could suggest the use of cannabinoid-based drugs devoid of psycotropic effects. Moreover, there is at present no obvious universal mechanism whereby cannabinoids affect cell viability and proliferation; furthermore, the immunosuppressive properties of cannabinoids or their effects on COX-2 expression, even if incompletely demonstrated to date, could represent cons evidence for medical use of cannabinoids, at least in lung carcinoma.

Indeed, cannabinoids have the advantage of being well tolerated in animal studies and they do not present the generalized toxic effects of most conventional chemotherapeutic agents (Guzman et al., 2003). Routes of cannabinoids administration have been recently studied. THC is rapidly absorbed after inhalation and its effects become apparent within minutes. Grotenhermen (2001) showed that THC oral administration was associated with slow onset of action and with accidental overdosage. In fact, maximum THC serum concentration measured after smoke intake (Huestis et al., 1992) is 2–3-fold higher than maximum serum concentration achievable with oral or rectal THC administration (Consroe et al., 1991; Brenneisen et al., 1996). The inhalation may have pharmacokinetic advantages, but it requires use of higher potency cannabinoids and strategy aimed at eliminating carcinogenic products combustion: for this purpose, Gieringer (2001) proposed the vaporization lacking the carcinogenic compounds formed during combustion. The trans-dermal route could be eligible for pain, nausea and vomiting treatment in chemotherapy patients giving a continuous steady dose (Stinchcomb et al., 2001).

In rats, THC and WIN-55,212-2 administered by infusion at the site of tumour showed a good efficacy, but so far, only preliminary results from one clinical study applying a strategy of local THC administration in patients with recurrent glioblastoma multiforme has been reported (Blazquez et al., 2004). Moreover, long-term effects of chronically administered cannabinoids have not been studied. To date, the prescription of cannabinoids is provided for medical conditions that are not adequately controlled by standard treatments, but considering their potentiality in clinical practise the Clinical Cannabinoid Group, chaired by Dr Peterwee, encourage properly conducted clinical trials to evaluate the further potential therapeutic uses of cannabinoids alone or in combination with other drugs.

Even if the use of cannabinoids in clinical practice needs further preclinical research, in order to confirme safety, efficacy, doses and administration protocols, the cannabinoids could provide unquestionable advantages compared to current antitumoural therapies: (1) cannabinoids selectively affect tumour cells more than their nontransformed counterparts that might even be protected from cell death; (2) systematically administered selective inhibitors of endocannabinoid degradation would be effective only in those tissues where endocannabinoid levels are pathologically altered, without any significant psychotropic or immunosuppressive activity; (3) selective CB1 agonists unable to cross the blood–brain barrier would be deprived of the immunosuppressive and psychotropic effects of cannabinoids and therefore could be efficaciously used as antineoplastic drugs in a large number of tumours, with the exception of glioma; (4) cannabinoids could represent an efficacious therapy in COX-2-expressing tumours that have become resistant to induction of apoptosis: acting as COX-2-substrates with no effect on the protective properties of COX-2-derived products, they could offer some advantage with respect to the NSAID in order to enhance the sensibility to conventional anticancer therapies.

Even if further in vivo research are required to clarify cannabinoids action in cancer and especially to test their effectiveness in patients, the cannabinoid system represent a promising target for cancer treatment.

Acknowledgments

We thank the Associazione ‘Educazione e Ricerca Medica Salernitana’ (ERMES) and Sanofi-Aventis Research for supporting our studies on this subject.

Ancillary