CB₂: a cannabinoid receptor with an identity crisis

Abstract

CB₂ was first considered to be the ‘peripheral cannabinoid receptor’. This title was bestowed based on its abundant expression in the immune system and presumed absence from the central nervous system. However, multiple recent reports question the absence of CB₂ from the central nervous system. For example, it is now well accepted that CB₂ is expressed in brain microglia during neuroinflammation. However, the extent of CB₂ expression in neurons has remained controversial. There have been studies claiming either extreme-its complete absence to its widespread expression-as well as everything in between. This review will discuss the reported tissue distribution of CB₂ with a focus on CB₂ in neurons, particularly those in the central nervous system as well as the implications of that presence. As CB₂ is an attractive therapeutic target for pain management and immune system modulation without overt psychoactivity, defining the extent of its presence in neurons will have a significant impact on drug discovery. Our recommendation is to encourage cautious interpretation of data that have been presented for and against CB₂'s presence in neurons and to encourage continued rigorous study.

This article is part of a themed issue on Cannabinoids. To view the editorial for this themed issue visit http://dx.doi.org/10.1111/j.1476-5381.2010.00831.x

The therapeutic potential of cannabis as well as its psychoactive effects has been known for thousands of years (Adams and Martin, 1996). However, it was not until the discovery of cannabinoid binding sites in brain (Devane et al., 1988; Herkenham et al., 1990; Herkenham et al., 1991; Matsuda et al., 1993) and the subsequent cloning of the CB₁ receptor (Matsuda et al., 1990; Alexander et al., 2008) that cellular mechanisms for these effects began to be elucidated. A second cannabinoid receptor (CB₂) was identified and first cloned from HL60 cells by Munro et al. in 1993 (Munro et al., 1993). CB₂ was dubbed the ‘peripheral cannabinoid receptor’ as a result of in situ hybridization analysis that revealed high levels of CB₂ mRNA in spleen and its absence from brain. CB₂ receptors were cloned from mouse and rat in later years (Shire et al., 1996; Griffin et al., 2000; Brown et al., 2002).

CB₂ and cellular signalling

Around the time that CB₁ and CB₂ were cloned, anandamide and 2-arachidonylglycerol (2-AG) were identified as endogenous cannabinoid agonists (Devane et al., 1992; Felder et al., 1993; Sugiura et al., 1995; Hanus et al., 2001). 2-AG is a high efficacy agonist at CB₂ (Lynn and Herkenham, 1994; Slipetz et al., 1995; Gonsiorek et al., 2000; Sugiura et al., 2000; Shoemaker et al., 2005b). However, anandamide has a low efficacy at CB₂, often functioning as a weak partial agonist (Showalter et al., 1996; Gonsiorek et al., 2000; Sugiura et al., 2000). Similar to CB₁, CB₂ is a G_i/o coupled G protein coupled receptor and as such inhibits adenylyl cyclase (Bayewitch et al., 1995; Felder et al., 1995; Slipetz et al., 1995; Gonsiorek et al., 2000; Sugiura et al., 2000; Shoemaker et al., 2005b). Furthermore, it can also promote MAPK activation (p38 and p42/44), PI3K, ceramide production and gene transcription (Bouaboula et al., 1996; Bouaboula et al., 1999a; Bouaboula et al., 1999b; Howlett, 2002; Howlett et al., 2002; Herrera et al., 2005; Herrera et al., 2006; Grimaldi et al., 2009; Romero-Sandoval et al., 2009). A key difference, however, is that unlike CB₁, CB₂ appears to poorly modulate calcium channels or inwardly rectifying potassium channels (Felder et al., 1995). Studies using SR144528, a CB₂ antagonist/inverse agonist, have revealed that the receptor possesses a high degree of constitutive activity in expression systems (Bouaboula et al., 1999a; Bouaboula et al., 1999b). Of further interest is that CB₂ receptors from different species often have different pharmacological responses to identical drugs, complicating the drug discovery process (Mukherjee et al., 2004; Yao et al., 2006; Bingham et al., 2007). Therefore, despite coupling to the same family of G proteins and sharing some ligands, CB₁ and CB₂ appear to differ significantly from one another in their signalling.

CB₂ as a therapeutic target

CB₂ is an attractive therapeutic target. The abundant CB₂ expression in immune cells presents a plausible explanation for cannabinoid immunomodulatory activity (Lynn and Herkenham, 1994; Berdyshev, 2000; Howlett, 2002; Costa, 2007). Indeed, CB₂ activation affects a myriad of immune responses from inflammation to neuroprotection (Cabral and Griffin-Thomas, 2009). Additionally, numerous reports indicate that CB₂ activation is analgesic and CB₂ agonists suppress responses in many animal models of pain, from acute to neuropathic (Anand et al., 2009), although these effects may involve CB₁ activation as well. CB₁ is abundant within the brain, where it appears responsible for mediating the psychoactive effects of cannabis (Mackie, 2005). Thus, the scarcity of central nervous system (CNS) CB₂ receptors makes CB₂ selective drugs attractive as therapeutics as they would presumably lack psychoactivity. In support of this notion, mice with ‘knockout’ of CB₂ had typical behavioural responses to Δ⁹-tetrahydrocannabinol (THC) but lost their normal immune responsiveness to THC (Buckley et al., 2000). CB₂ levels can also be increased under certain conditions and disease states further adding to its attractiveness as a therapeutic target (Zhang et al., 2003; Wotherspoon et al., 2005; Yiangou et al., 2006).

CB₂ agonists and antagonists

As CB₂ is such an attractive therapeutic target, much effort has been made to synthesize selective CB₂ agonists and antagonists. Some of the early cannabinoid agonists such as CP55940, WIN55212-2 and HU210 demonstrate high affinity at CB₂ and are considered full agonists, but are not selective for CB₂ over CB₁[see Miller and Stella (2008) for a summary of the binding data]. JWH015 was one of the first potential CB₂ selective agonists (Showalter et al., 1996; Griffin et al., 2000), but has since been shown to also be an agonist at GPR55 (Ryberg et al., 2007; Lauckner et al., 2008). Numerous other compounds have been synthesized with the aim of making CB₂ selective agonists. AM1241 (Malan et al., 2001) and JWH133 (Huffman et al., 1999) are two of the most commonly used ‘selective’ CB₂ agonists. Other early ones include HU308 (Hanus et al., 1999) and GW405833 (L-768242) (Gallant et al., 1996; Valenzano et al., 2005). More recently synthesized compounds include GW833972A (Belvisi et al., 2008), MDA7 (Naguib et al., 2008), A-796260 (Yao et al., 2008) and A-836339 (Yao et al., 2009). Cannabilactones have also been suggested as potential CB₂ selective compounds (Khanolkar et al., 2007). For the interested reader, a review by Whiteside et al. contains detailed analysis of many of these compounds as well as numerous others (Whiteside et al., 2007). SR144528 (Rinaldi-Carmona et al., 1998) and AM630 (Pertwee et al., 1995; Ross et al., 1999) are the two of the most commonly used CB₂ selective antagonists and have been frequently used to demonstrate specificity of many of these other CB₂ selective agonists. However, a fundamental problem with designing selective agonists and antagonists is possible interactions with other unforeseen targets. These compounds may exhibit a strong preference for CB₂ over CB₁, but as evidenced by JWH015, other non-CB₁/CB₂ binding sites may still exist. It is extremely difficult to conclusively establish an agonist (or antagonist) is specific for CB₂ and no other receptors. This caveat must be kept in mind when evaluating studies that solely use a pharmacological approach and allege CB₂ involvement in a process. Furthermore, as demonstrated for AM1241, CB₂ agonists may produce very different effects at CB₂ receptors from different species (Bingham et al., 2007). Here, a racemic mixture of AM1241 was an agonist at human CB₂ but functioned as an inverse agonist at rodent CB₂. R-AM1241 has a higher affinity for CB₂ than S-AM1241, but functions similar to the racemate. On the other hand S-AM1241 was an efficacious agonist at both human and rodent CB₂. Furthermore naloxone, a µ opioid receptor antagonist, can block the analgesic effects of AM1241, but this appears not to be the case for other CB₂ agonists (Ibrahim et al., 2005; Yao et al., 2009). It also important to bear in mind that selective agonists and antagonists for a particular receptor may differentially alter coupling to distinct signalling pathways, a concept known as functional selectivity (Urban et al., 2007). Thus, CB₂ agonists that share identical binding characteristics may have different potencies in activating different signalling pathways and evoke substantially different physiological responses. For example, with CB₂ expressed in Chinese hamster ovary cells, 2-AG, CP55940 and noladin ether had different rank orders of potency depending on the signalling pathway analysed: inhibition of adenylyl cyclase, MAPK activation and stimulation of calcium transients (Shoemaker et al., 2005a). This concept of functional selectivity mandates caution in comparing pharmacological studies that employ different agonists and antagonists for CB₂. This may explain the differences previously mentioned between AM1241 and other CB₂ agonists and their ability to produce analgesia and may even extend to other agonist-specific effects obtained with other compounds. Thus, it is important that functional evidence obtained using CB₂‘selective’ agonists and antagonists be balanced with careful controls and supported by additional genetic and anatomical analyses to assure that some other unanticipated target is not the true site of action.

CB₂: the ‘peripheral’ cannabinoid receptor?

CB₂ and evidence for absence from CNS

In addition to the data obtained during the CB₂ knockout mouse characterization, earlier reports also arrived at the conclusion that CB₂ is absent from the CNS. (Of course, ‘absence’ just means below the level of detection of the particular assay being employed.) When CB₂ was first cloned, in situ hybridization demonstrated a lack of CB₂ mRNA signal in rat brain (Munro et al., 1993). While characterizing CB₁ and CB₂ in immune cells, Schatz et al. performed Northern blots on mouse brain and rat cerebellums and could not detect the presence of CB₂ in these tissues (Schatz et al., 1997). However, RT-PCR demonstrated the presence of CB₂ mRNA, at levels too low to be quantified. Northern blot analysis performed by Galiegue et al. is in agreement with the Schatz et al. data (Galiegue et al., 1995). However, in RT-PCR experiments performed as a part of this study, CB₂ was undetectable in human cortex, cerebellum, whole brain and pituitary gland. McCoy et al. also did not detect CB₂ mRNA in mouse brain using RT-PCR/Southern blot analysis (McCoy et al., 1999). As part of the characterization of SR144528 as a CB₂ antagonist, rat brain radioligand binding and GTPγS binding analyses were performed (Griffin et al., 1999). The authors found little SR144528 binding in whole brain and cerebellum and the results of their GTPγS binding analysis supported this. Furthermore, Northern blotting did not detect CB₂ mRNA in cerebellum, cortex or spinal cord. Rat cortex has also been reported to lack CB₂ mRNA (Beltramo et al., 2006). Included in the review by Howlett et al. (Howlett et al. 2002), Herkenham and Hohmann replicated the in situ hybridization results of Munro and colleagues. Derbenev and colleagues did not detect CB₂ mRNA or protein in rat brainstem (Derbenev et al., 2004). As part of an initial characterization of cannabinoid receptors in dorsal root ganglia (DRG), in situ hybridization revealed the presence of CB₁ but not CB₂ receptors (Hohmann and Herkenham, 1999a,b;). Price et al. could also not detect CB₂ mRNA in rat trigeminal ganglia (Price et al., 2003).

Based on all these data, CB₂ was informally referred to as the peripheral cannabinoid receptor. However, a number of more recent reports have suggested that, in contrast to these previous claims of its absence, CB₂ may in fact be expressed in the CNS (see below). This finding has had a significant impact on drug discovery and our understanding of the biology of the endocannabinoid system. This review will focus on reports of CB₂ in neurons and in the brain and the implications of that presence.

CB₂ and the immune system

CB₂ research continues to have a large focus on its role in the immune system. Analysis of the presence and function of CB₂ in the brain necessitates a discussion concerning CB₂ in immune cells. CB₂ mRNA has been identified in many immune tissues (Munro et al., 1993; Lynn and Herkenham, 1994; Galiegue et al., 1995; Schatz et al., 1997; Berdyshev, 2000; Buckley et al., 2000). Of specific immune cell types, the highest levels of CB₂ are found in macrophages, CD4+ T cells, CD8+ T cells, B cells, natural killer cells, monocytes and polymorphonuclear neutrophils (Derocq et al., 1995; Galiegue et al., 1995; Schatz et al., 1997; Carayon et al., 1998; McCoy et al., 1999; Buckley et al., 2000; Carlisle et al., 2002; Maresz et al., 2007; Dittel, 2008). Of particular relevance for the role of CB₂ in the CNS, CB₂ mRNA and protein have been found in microglia (Carlisle et al., 2002; Klegeris et al., 2003; Walter et al., 2003; Beltramo et al., 2006; Maresz et al., 2007). Microglia are derived from macrophages and can be viewed as the resident immune cells of the brain where they monitor the brain for pathological damage. In response to specific signals within the brain they transition between different states of activity (Ashton and Glass, 2007; Hanisch and Kettenmann, 2007). The expression levels of CB₂ in microglia vary depending on the activation state of the cell (Carlisle et al., 2002; Walter et al., 2003; Stella, 2004; Maresz et al., 2007; Cabral et al., 2008; Pietr et al., 2009). CB₂ modulates microglial migration and infiltration into brain areas with active neuroinflammation and degeneration (Walter et al., 2003; Ashton et al., 2007; Fernandez-Ruiz et al., 2008; Miller and Stella, 2008; Price et al., 2009). In healthy brain microglia do not appear to express CB₂ (Stella, 2004). However, in Alzheimer's brain tissue, CB₂ can be detected in neuritic plaque-associated microglia (Benito et al., 2003). Similarly, in models of neuropathic pain (but not inflammatory pain) CB₂ mRNA increases in association with activated microglia in the spinal cord (Zhang et al., 2003). In addition during amyotrophic lateral sclerosis and multiple sclerosis, CB₂ microglial expression increases in the spinal cord (Yiangou et al., 2006). According to this evidence it is clear that under certain conditions brain microglia are capable of expressing CB₂.

CB₂ and tissue distribution

Despite being initially described as an immune cell cannabinoid receptor, CB₂ has been identified molecularly and pharmacologically in numerous other cell types. Evidence for the presence of CB₂ receptors has been found in pulmonary endothelial cells (Zoratti et al., 2003). In these cells, CB₂ activation by anandamide results in phospholipase C-mediated calcium release from smooth ER with subsequent increases in mitochondrial calcium. CB₂ can also be found in bone (in osteocytes, osteoblasts and osteoclasts) where it modulates bone formation and turnover (Ofek et al., 2006). The gastrointestinal system appears to express CB₂ receptors as well (Storr et al., 2002; Hillsley et al., 2007; Duncan et al., 2008). 2-AG affects meiosis in spermatogonia via CB₂ (Grimaldi et al., 2009) as well as a number of other aspects of reproductive function (Maccarrone, 2008; Grimaldi et al., 2009). Keratinocytes release beta-endorphin in response to CB₂ selective agonist stimulation (Ibrahim et al., 2005), although this result is controversial (Whiteside et al., 2007; Anand et al., 2008; Yao et al., 2008; Yao et al., 2009). These cells have also been reported to have CB₂ immunoreactivity. In the eye, trabecular meshwork cells have been shown to have functional CB₂ receptors (Zhong et al., 2005; He and Song, 2007). Mature and precursor adipocytes express functional CB₂ receptors that are negatively coupled to adenylyl cyclase (Roche et al., 2006). In cirrhotic liver, CB₂ receptors are expressed in hepatic myofibroblasts, but are absent in normal liver (Julien et al., 2005). THC protects cardiomyocytes from hypoxic damage by acting at CB₂ receptors resulting in nitric oxide production (Shmist et al., 2006).

CB₂ and nociception

To better understand the possible presence of CB₂ in neurons it is helpful to consider the role of CB₂ in nociception. Cannabinoids have long been known to possess analgesic activity, but evidence for CB₂ having a role in analgesia was not presented until 1998 (Calignano et al., 1998). Shortly thereafter, HU308, a CB₂ selective agonist, was shown to have analgesic activity without typical cannabinoid CNS side effects (Hanus et al., 1999). AM1241, another CB₂ selective agonist was also shown to promote analgesia when injected peripherally and this did not produce CNS side effects, suggesting that CB₂ receptors modulate nociception (Malan et al., 2001; Ibrahim et al., 2003; Malan et al., 2003; Ibrahim et al., 2006). It has since been shown that a number of different CB₂ agonists can modulate many types of pain: acute, inflammatory, neuropathic, post-surgical and cancer pain (Khanolkar et al., 2007; Whiteside et al., 2007; Jhaveri et al., 2008; Naguib et al., 2008; Ohta et al., 2008; Yao et al., 2008; Anand et al., 2009; Yao et al., 2009). It is still unclear as to where these CB₂ agonists exert their analgesic activity. The site could be microglia, astrocytes, neurons, another cell type or a combination of these. Furthermore, as discussed above, the specificity of these CB₂‘selective’ compounds may not be as specific as previously thought. Additional work must be performed to state with confidence that these agonists produce analgesia solely via activation of CB₂ receptors.

CB₂ and peripheral neurons

The first step to determine if CB₂ activation has a direct effect on neural mechanisms is to determine whether CB₂ is expressed in neurons. The existence of functional CB₂ receptors in peripheral neurons has been suggested by a number of studies. The first evidence for CB₂ function in peripheral neurons came in 1997 when CB₂ mRNA was identified in mouse vas deferens tissue (Griffin et al., 1997). In support of a functional role for CB₂, JWH015 and JWH051 (agonists preferring CB₂ over CB₁) produced concentration dependent inhibition of evoked contractions presumably via a prejunctional site. However, a submicromolar concentration of AM630, a CB₂ antagonist, could not block this effect. Further, JWH015 is also an agonist for GPR55 (Ryberg et al., 2007; Lauckner et al., 2008), so the involvement of CB₂ cannot be unequivocally asserted.

CB₂ and sensory neurons

Functional studies have hinted at the presence of CB₂ receptors on sensory neurons. In these studies, it is necessary to consider the possible involvement of CB₂-expressing immune cells as microglia can affect synaptic properties (Cullheim and Thams, 2007; Abbadie et al., 2009). Patel and colleagues provided some of the first functional evidence of CB₂ in sensory neurons (Patel et al., 2003). Using isolated guinea pig and human vagus nerve preparations, they demonstrated that the CB₂ agonist JWH133 inhibited nerve depolarizations in response to capsaicin, PGE₂ and hypertonic saline. These three treatments activate vagal C and/or Aδ fibres. SR144528 blocked the effects of JWH133. A follow-up study with another putative CB₂ agonist, GW833972A, produced similar results (Belvisi et al., 2008). Neither study was designed to determine a specific site or mechanism of action. While CB₂ does not appear to play a role in myenteric contractions, it does seem to play a role in activation of mesenteric sensory nerves. AM1241 administered intravenously inhibits bradykinin induced activation of isolated mesenteric afferents in mice (Hillsley et al., 2007). This effect was blocked by AM630 and absent in CB₂ knockout mice. Interestingly, while CB₂ agonists do not affect normal enteric contractility, JWH133 can prevent lipopolysaccharide (LPS) induced increases in evoked contractions (Mathison et al., 2004; Duncan et al., 2008). JWH133 also blocks LPS stimulation of Fos expression in enteric neurons. AM630 antagonizes these effects. CB₂ receptors in myenteric neurons were identified as the most likely target of this drug effect (see below). CB₂ mRNA has also been identified in rat and mouse retina using RT-PCR as well as within specific layers of the retina (ganglion, inner nuclear and photoreceptor inner layers) using in situ hybridization (Lu et al., 2000). Additionally, Burdyga et al. identified low, barely detectable levels of CB₂ mRNA in rat nodose ganglion, but were unable to detect CB₂ in the human vagal nerve trunk (Burdyga et al., 2004).

CB₂ and nociceptive neurons

Further functional studies point to a role for CB₂ in sensory neuron function, particularly nociceptive neurons. A study was performed to address AM1241's ability to prevent windup of wide dynamic range (WDR) neurons in spinal cord (Nackley et al., 2004). Here, AM1241 administered locally or systemically reduced the activity of C-fibres synapsing onto WDR neurons and this was reversed by SR144528, but not SR141716A. Significantly, suppression occurred in the presence and absence of inflammation. This, combined with the time course observed suggests long-term changes in presynaptic facilitation, makes the effects less likely to be due to CB₂ targeting immune cells. The authors speculate a direct effect of CB₂ activation on C-fibre neurons. Elmes et al. performed a similar study using JWH133 as a CB₂ agonist to test WDR spinal neuron responses in models of inflammatory and neuropathic pain as well as in non-inflammatory and sham-operated conditions (Elmes et al., 2004). Like the Nackley study, they also found that peripherally administered CB₂ agonist inhibits WDR activity in both naïve and inflammatory conditions as well as following neuropathic injury. Once again, the data are suggestive of a non-immune function of CB₂, possibly in peripheral neurons. A follow-up study analysed JWH133's ability to inhibit capsaicin-induced calcium increases in DRG neurons cultured from sham-operated and neuropathic rats (Sagar et al., 2005). JWH133 slightly inhibited calcium increases in DRG cultured from both neuropathic and sham rats in a SR144528-sensitive fashion, consistent with the presence of functional CB₂ receptors in peripheral neurons. However, spinally administered JWH133 inhibited mechanically evoked responses of dorsal horn neurons from laminae V and VI only in neuropathic rats, but not in sham-operated animals. This points to an up-regulation of CB₂ in intrinsic spinal cord neurons in pain states, although does not provide evidence of the site of up-regulation. AM1241 and L768242 (another CB₂ agonist) can also decrease capsaicin-induced calcitonin gene-related peptide release, a pain biomarker, from neurons in spinal cord slices and this can be blocked by SR144528 (Beltramo et al., 2006). These studies are most consistent with CB₂ participating in neural mechanisms rather than via immune cells, but do not directly answer the question of whether or not CB₂ is expressed in neurons.

The initial support for the presence of CB₂ protein in neurons came from Ross and colleagues. Using fluorescence-activated cell sorting analysis, they determined that DRG cultures and F-11 cells (DRG neuron × neuroblastoma hybrid) express both CB₁ and CB₂, but could not conclude that CB₂ was functionally expressed in DRG neurons (Ross et al., 2001). To further address the site of CB₂ expression in DRG, Wotherspoon and colleagues used immunhistochemistry on DRG and spinal cord of naïve and nerve damaged rats and mice (Wotherspoon et al., 2005). No CB₂ could be detected in normal rat or mouse spinal cord and DRG neurons. However, upon nerve sectioning or ligation, CB₂ immunoreactivity could be detected in the ipsilateral dorsal horn. This immunoreactivity was strongly reduced in CB₂ knockout mice and was blocked by incubation with the immunizing peptide, suggesting specificity of the primary antibody used. Of great interest, and in contrast to what would be expected based on Zhang et al.'s (2003) study, was that the CB₂ signal did not co-localize with markers of astrocytes (GFAP) or microglia (OX-42). Instead it co-localized with markers of damaged sensory neuron terminals (GAP-43 and galanin). CB₂ immunoreactivity also accumulated in axons proximal to the ligation site. They could not identify CB₂ in cell bodies in tissue sections, but were able to identify CB₂ in isolated DRG neurons grown in culture from lesioned mice. They also did not observe CB₂ immunoreactivity in skin, in contrast to other studies (Stander et al., 2005; Kress and Kuner, 2009) and Ibrahim et al. who found it in keratinocytes (Ibrahim et al., 2005). A few studies have detected CB₂ mRNA in DRG and spinal cord using quantitative RT-PCR. Here levels increased following nerve ligation, but this does not necessarily implicate a neuronal source (Zhang et al., 2003; Beltramo et al., 2006).

A more recent study by Anand and colleagues is consistent with the above findings (Anand et al., 2008). Specifically, they found CB₂ positive, small diameter neurons in human DRG and peripheral nerves using three different CB₂ antibodies. The specificity of the antibodies was assessed using peptide block. CB₂ levels increased following nerve injury. They further extended the analysis by demonstrating CB₂ colocalization with neuronal (GAP-43), axonal (neurofilament) and nociceptive neuronal markers (TRPV1). Similar staining was observed in mouse, rat and guinea pig DRG. This study also replicated the functional data reported by Sagar et al. in that a CB₂ agonist (GW833972) inhibited capsaicin-induced calcium increases in DRG sensory neurons. They further sought to identify a mechanism for this activity and determined that CB₂-mediated cAMP depletion attenuated TRPV1 activation. This presumably decreased PKA-mediated phosphorylation of TRPV1, analogous to the effects of µ opioid receptor activation.

CB₂ and the enteric nervous system

Despite earlier findings (Griffin et al., 1997), several studies suggest CB₂ is expressed in the enteric nervous system. Duncan et al. and Storr et al. found CB₂ mRNA in the enteric nervous system (Storr et al., 2002; Duncan et al., 2008). The site of action of JWH133 in preventing LPS-induced increases in ileum contractility was addressed using RT-PCR and immunohistochemistry (Duncan et al., 2008). CB₂ mRNA was detected in the full-wall thickness ileum, ileal muscle, submucosal and mucosal layers in normal rats. LPS treatment had no effect on the levels of expression. A number of different antibodies and knockout tissue were used for controls. CB₂ protein was detected in all the same tissues in which CB₂ mRNA was found. CB₂ colocalized with markers of enteric ganglia, pan-neuronal markers and synaptic terminals suggesting a strong presence in myenteric neurons. CB₂ immunoreactivity did not colocalize with glial markers.

CB₂: another central cannabinoid receptor?

CB₂ in the cerebellum

One of the earliest reports of the presence of CB₂ in the CNS came from a study performed by Skaper and colleagues (Skaper et al., 1996). In situ hybridization revealed the presence of CB₂ mRNA in cultured granule cells. In addition in situ hybridization localized CB₂ to the granule and Purkinje cell layers of mouse cerebellum. Radioligand binding analysis of cerebellar membranes revealed the presence of two WIN55212 binding sites. The affinities of WIN55212 at these sites were reported to be close to those of CB₁ and CB₂, although the exact identity of the binding sites could not be specifically determined. RT-PCR analysis has identified CB₂ mRNA in the rat cerebellum and Western blotting has revealed expressed CB₂ protein in rat and ferret cerebellum as well (Van Sickle et al., 2005). Peptide block was used as a control for those Western blots.

Additional studies have also attempted to localize CB₂ protein in the cerebellum (Ashton et al., 2006; Baek et al., 2008). Using an antibody directed against the C-terminus of CB₂, with peptide block control, they identified CB₂ protein expression in the granule, Purkinje and white matter layers of the rat cerebellum. The signal did not overlap with astrocytes markers and the staining pattern in the Purkinje layer and parts of the other layers appeared to be capillary endothelial in nature. There were fine fibres in the white matter and granule cell layers that were CB₂ positive but their origin remains to be determined. These could possibly arise from microglia or neurons. Onaivi et al. have also reported CB₂ expression in the Purkinje and molecular layers of the cerebellum using Western blot, immunohistochemistry and in situ hybridization techniques (Gong et al., 2006; Onaivi et al., 2008b).

CB₂ in the brainstem

CB₂ has been identified within the brainstem as well. A thorough analysis was performed that investigated CB₂ expression in brain, focusing on mRNA, protein and functional expression within the brainstem (Van Sickle et al., 2005). Quantitative RT-PCR showed that the rat brainstem contains CB₂ mRNA at significantly lower levels than spleen (1.5% of spleen levels). Western blotting confirmed this expression for rat as well as for ferret. Immunocytochemistry identified the dorsal motor nucleus of the vagus nerve (DMNX) of the mouse, rat and ferret as a brainstem nucleus containing CB₂ protein. The CB₂ knockout mouse did not show any immunostaining in the DMNX. The DMNX immunoreactivity colocalized with neuronal markers, but in contrast to what Ashton et al. found in the cerebellum (Ashton et al., 2006), the signal did not overlap with glial or blood vessel markers. The authors acknowledged the differences in results between their study and that of Derbenev et al. that did not find CB₂ in similar regions (Derbenev et al., 2004) and state that in the latter study a faint signal could be observed in Western blots consistent with low levels of expression. The authors also demonstrated that AM630 blocked the anti-emetic actions of 2-AG treatment in ferrets, suggesting CB₂ receptor involvement. Furthermore a sub-efficacious concentration of anandamide combined with AM1241 treatment produced anti-emetic effects. Another, more superficial study of the brainstem using immunohistochemistry was later performed to look for CB₂ in other brainstem nuclei (Baek et al., 2008). CB₂ immunoreactivity was found in the medial vestibular nucleus as well as the dorsal and ventral cochlear nuclei, but no attempts were made to identify cell types. Peptide block and secondary antibody controls were used to determine CB₂ antibody specificity. Viscomi et al. did not find CB₂ protein and only low levels of CB₂ mRNA in inferior olive and pontine nuclei using immunohistochemistry and quantitative PCR in normal rats (Viscomi et al., 2009). However, following a hemicerebellectomy, CB₂ expression dramatically increased in both mRNA and protein levels in these nuclei. The CB₂ immunoreactivity colocalized with neuronal markers but not with microglial or astrocytic ones. Further JWH015 had a neuroprotective effect, preventing cell death due to the hemicerebellectomy. This was likely operating through CB₂, although they did not report block of the neuroprotective effect with a CB₂ antagonist. Gong et al. reported the presence of CB₂ in many nuclei of the brainstem using RT-PCR and immunohistochemistry (Gong et al., 2006).

CB₂ and the hippocampal formation

Using several approaches, Onaivi and his collaborators have reported finding CB₂ immunoreactivity in many areas of the hippocampal formation (Gong et al., 2006; Onaivi, 2006; Onaivi et al., 2006; 2008a,b; Brusco et al., 2008). They report a predominately postsynaptic expression and an association with rough endoplasmic reticulum and Golgi structures. They have also demonstrated CB₂ staining in hippocampal cultures. In contrast to their immunohistochemical results, they have had mixed results in finding CB₂ mRNA in the hippocampus (Gong et al., 2006; Onaivi et al., 2008b).

Functional evidence for CB₂ expression in the cortex comes from recording spontaneous inhibitory postsynaptic currents (sIPSCs) in layers II and V of the medial entorhinal cortex. Here, 2-AG mediated suppression of sIPSCs was not blocked by LY320135, a CB₁ antagonist/inverse agonist, whereas they were blocked by AM630 and JTE907 (a structurally distinct CB₂ antagonist) (Morgan et al., 2009). Further JWH133 suppressed sIPSCs in a CB₂ antagonist sensitive fashion. The site of CB₂ agonist action remains to be conclusively demonstrated.

CB₂ and other brain regions

Evidence exists for CB₂ expression in other brain regions. While recording from the ventral posterior nucleus of the thalamus, Jhaveri et al. found that after spinal nerve ligation, JWH133 reduced spontaneous and evoked responses in a SR144528-sensitive fashion, but that this effect was absent in sham operated rats (Jhaveri et al., 2008). Gong et al. have also reported CB₂ immunoreactivity in many thalamic nuclei, but could not detect CB₂ mRNA using RT-PCR (Gong et al., 2006). Furthermore, this group has reported finding CB₂ mRNA in striatum and hypothalamus, but not in olfactory bulb, cortex and spinal cord and mixed results in midbrain (Gong et al., 2006; Onaivi et al., 2008b). Additionally they report CB₂ immunoreactivity in olfactory bulb, cortex, midbrain as well as the other areas already mentioned (Gong et al., 2006; Onaivi, 2006; Onaivi et al., 2006; 2008b;).

CB₂ and neurogenesis

CB₂ also appears to play a role in neurogenesis. Both CB₁ and CB₂ are expressed in stem cells (Jiang et al., 2007; Molina-Holgado et al., 2007). More specifically, RT-PCR, Western blot and immunohistochemical analyses have all revealed the presence of CB₂ in embryonic and adult neural progenitor cells (Palazuelos et al., 2006; Molina-Holgado et al., 2007). CB₂ blockade or genetic disruption impairs neurosphere formation and prevents progenitor cell proliferation, whereas CB₂ agonists promote these activities via ERK and Akt signalling (Palazuelos et al., 2006; Molina-Holgado et al., 2007). However, CB₂ expression seems to diminish as the cells differentiate, being nearly absent by the time neuronal and astrocytic markers appear (Palazuelos et al., 2006). Further, CB₁ agonists and antagonists have similar effects in neurosphere formation and in COR-1 neural stem cell cultures (Molina-Holgado et al., 2007; Goncalves et al., 2008) suggesting either functional interactions or redundant signalling. In contrast to these data, CB₁ agonists and antagonists had no effect on neurogenesis in the subventricular zone (SVZ) of either young or adult mice (Goncalves et al., 2008). On the other hand, JWH133 and WIN55212 stimulated SVZ neurogenesis, whereas AM630 and JTE907 decreased it (Goncalves et al., 2008).

CB₂: where is its real home and why do we care?

We feel careful analysis of the studies reviewed above allows us to reach the following conclusions: CB₂ is expressed by microglia, with levels increasing as they are activated, and CB₂ is present at detectable and functionally relevant levels in a subset of neurons, with increasing levels following injury. We care where CB₂ is expressed primarily for understanding pathology that involves CB₂ and to develop therapies that target difficult to treat conditions. To this end it is important to have a rigorous understanding of where and under what conditions CB₂ is expressed in the CNS.

Approaches aimed at identifying CB₂ receptor expression in the brain can be divided into functional (pharmacological), biochemical and anatomical techniques. All three have their strengths and weaknesses. The most convincing studies will incorporate a combination of these techniques. Table 1 summarizes the studies presented here, detailing the brain region analysed and whether or not CB₂ was detected and the techniques(s) used to detect it. Pharmacological studies rely on the specificity of the drugs used. When interpreting these studies it is necessary to recall that specificity is never absolute – at sufficiently high concentrations any drug will interact with additional targets. Thus, it is important to relate the concentration of the drug being used to the binding affinity of the CB₂ receptor for that drug. The second consideration is that drugs considered to be ‘specific’ or ‘selective’ based on our current understanding may soon be found to interact with other receptors. Examples of this in the cannabinoid system include AM251, often used as a ‘selective’ CB₁ receptor antagonist, but it is also a GPR55 agonist (Henstridge et al., 2009; Kapur et al., 2009) and JWH015, sometimes used as a ‘selective’ CB₂ agonist, but it, too, is a GPR55 agonist (Ryberg et al., 2007; Lauckner et al., 2008). Approaches to circumvent this issue include using several structurally diverse agonists and antagonists (presumably decreasing the likelihood of having the same ‘off-target’ actions) and knockout or ‘knockdown’ controls, when appropriate.

Table 1.

CB₂ distribution in the peripheral and central nervous systems

Location	Presence	Detection method	Species	Reference
Whole brain
Whole brain	Absent	ISH	Mouse	Munro et al., 1993
Whole brain	Absent	NB/RT-PCR	Human	Galiegue et al., 1995
Whole brain	Absent	NB/RT-PCR	Mouse	Schatz et al., 1997
Whole brain	Absent	SB	Mouse	McCoy et al., 1999
Whole brain	Absent	RLB/GTPγS	Rat	Griffin et al., 1999
Whole brain	Absent	ISH	Mouse	Howlett et al., 2002
Whole brain	Present	WB	Rat	Gong et al., 2006
Whole brain	Present	WB/RT-PCR	Mouse	Onaivi, 2006; Onaivi et al., 2008b
Brainstem
DMNX	Present	RT-PCR/WB/IHC	Rat	Van Sickle et al., 2005
DMNX	Present	ICC	Mouse	Van Sickle et al., 2005
DMNX	Present	WB/ICC/functional	Ferret	Van Sickle et al., 2005
Cochlear nuclei	Present	IHC	Rat	Baek et al., 2008
Medial vestibular nuclei	Present	IHC	Rat	Baek et al., 2008
Inferior olive and pontine nuclei	Present	IHC/ICC/RT-PCR	Rat	Viscomi et al., 2009
Brainstem	Present	RT-PCR/IHC	Rat	Gong et al., 2006
Brainstem	Present	RT-PCR	Mouse	Onaivi et al., 2006; Liu et al., 2009
Cerebellum
Cerebellum	Absent	NB/RT-PCR	Human	Galiegue et al., 1995
Granule and Purkinje cell layers	Present	ISH/RLB	Mouse	Skaper et al., 1996
Cerebellum	Absent	NB/RT-PCR	Rat	Schatz et al., 1997
Cerebellum	Absent	NB/RLB/GTPγS	Rat	Griffin et al., 1999
Cerebellum	Present	RT-PCR/WB	Rat	Van Sickle et al., 2005
Cerebellum	Present	WB	Ferret	Van Sickle et al., 2005
Cerebellum	Present	ICC	Rat	Ashton et al., 2006
Cerebellum	Present	IHC	Rat	Baek et al., 2008
Cerebellum	Present	IHC/ISH	Rat/mouse	Onaivi, 2006; Onaivi et al., 2008b
Cerebellum	Present	RT-PCR	Human	Liu et al., 2009
Cortex
Cortex	Absent	RT-PCR/NB	Human	Galiegue et al., 1995
Cortex	Absent	NB	Rat	Griffin et al., 1999
Cortex	Absent	RT-PCR	Rat	Beltramo et al., 2006
Cortex	Present/absent	IHC/RT-PCR	Rat	Gong et al., 2006
Cortex	Present	IHC	Mouse	Onaivi, 2006; Onaivi et al., 2006; 2008b;
Cortex	Present	RT-PCR	Mouse/human	Liu et al., 2009
Hippocampus
Hippocampus	Present/absent	IHC/RT-PCR	Rat	Gong et al., 2006; Brusco et al., 2008
Hippocampus	Present	IHC	Mouse	Onaivi, 2006; Onaivi et al., 2006; 2008b;
Hippocampus	Present	IHC	Rat	Onaivi et al., 2008b
Hippocampus	Present	RT-PCR	Human	Liu et al., 2009
Other brain regions
Thalamus	Present/absent	IHC/RT-PCR	Rat	Gong et al., 2006
Hypothalamus	Present	RT-PCR		Gong et al., 2006
Midbrain	Present/absent	IHC/RT-PCR	Rat	Gong et al., 2006
Olfactory bulb	Present/absent	IHC/RT-PCR	Rat	Gong et al., 2006
VPN of thalamus	Present	Functional	Rat	Jhaveri et al., 2008
Entorhinal cortex	Present	Functional	Rat	Morgan et al., 2009
Peripheral neurons/spinal cord
Vas deferens	Present	Functional	Mouse	Griffin et al., 1997
Spinal cord	Absent	NB	Rat	Griffin et al., 1999
DRG neurons	Absent	ISH	Rat	Hohmann and Herkenham, 1999b
Retina	Present	ISH/RT-PCR	Mouse	Lu et al., 2000
DRG neurons	Present	FACS	Rat	Ross et al., 2001
Enteric system	Present	RT-PCR	Rat	Storr et al., 2002
Trigeminal ganglia	Absent	ISH	Rat	Price et al., 2003
Spinal cord	Present	ISH	Rat	Zhang et al., 2003
Vagus nerve	Present	Functional	Guinea pig/human	Patel et al., 2003
DRG neurons	Present	Functional	Rat	Nackley et al., 2004
DRG neurons	Present	Functional	Rat	Elmes et al., 2004
Nodose ganglion	Present	RT-PCR	Rat	Burdyga et al., 2004
Vagus nerve trunk	Absent	RT-PCR	Human	Burdyga et al., 2004
DRG neurons	Present	Functional	Rat	Sagar et al., 2005
DRG neurons/spinal cord	Inducible	ICC	Rat/mouse	Wotherspoon et al., 2005
Skin sensory neurons	Present	IHC	Human	Stander et al., 2005
DRG neurons/spinal cord	Present	Functional/RT-PCR	Rat	Beltramo et al., 2006
Mesenteric sensory neurons	Present	Functional	Mouse	Hillsley et al., 2007
Vagus nerve	Present	Functional	Guinea pig	Belvisi et al., 2008
Myenteric neurons	Present	Functional/ICC/RT-PCR	Rat	Duncan et al., 2008
DRG neurons	Present	IHC/WB/functional	Gp/human/rat/mouse	Anand et al., 2008
Neural progenitor cells
Hippocampal neural progenitor cells	Present	RT-PCR/ICC/functional	Mouse	Palazuelos et al., 2006
Neurospheres	Present	WB/ICC/functional	Mouse	Molina-Holgado et al., 2007
SVZ neural progenitor cells	Present	WB/ICC/functional	Mouse	Goncalves et al., 2008

Summary of studies investigating CB₂ expression in the nervous system. Details the location studied, whether CB₂ was detected or not, the method(s) used to detect it, the species analysed and the reference(s) for the studies.

FACS, fluorescence-activated cell sorting; ICC, immunocytochemistry; IHC, immunohistochemistry; ISH, in situ hybridization; NB, Northern blot; RLB, radioligand binding; RT-PCR, reverse transcriptase polymerase chain reaction; SB, Southern blot; WB, Western blot.

Biochemical studies include Western blotting and PCR-based approaches. For Western blotting, the key limitations are the sensitivity and specificity of the antibody used. At a minimum, blots from knockout (assuming the antibody is recognizing an epitope present in mouse CB₂) and positive control tissues (e.g. spleen) should be shown. Blindly trusting an antibody to ‘work’ without concurrent controls is unacceptable. Block with the immunizing antibody is desirable, but will not rule out a fortuitous interaction of the antibody with an unintended epitope on another protein. The sensitivity of Western blotting will depend on the abundance of CB₂ as well as the affinity of the antibody. The lack of detection of CB₂ on the blot can only be interpreted as that the level of CB₂ in the brain is below a certain level. (This level, relative to a CB₂-expressing tissue like spleen, can be determined by serial dilution.) PCR-approached tissues are the most sensitive. Their high sensitivity makes their interpretation subject to several considerations (Suzuki et al., 2000; Lion, 2001). These include amplification of CB₂ mRNA from immune cells trapped in the cerebral vasculature and amplification of CB₂ mRNA from a very small subset of activated microglia. In order to rationally interpret results from PCR-based experiments it is necessary that they be performed in a quantitative fashion, preferably calculating copy number, to facilitate comparisons.

Anatomical studies need to be conducted and interpreted with a similarly critical approach. These studies fall into three categories: autoradiography, in situ hybridization and immunocytochemistry. As above the issue of sensitivity needs consideration – it is possible to show CB₂ is present, but it is very hard to conclusively demonstrate that it is not present, just that it is present at a level below the limit of detection. However, this information, coupled with a lack of functional response, can be very valuable in sorting out the role of CB₂ receptors in a particular physiological response. The caveats of autoradiography include the pharmacological considerations discussed above as well as specific technical issues (Frey and Albin, 2001). As this technique has not been widely applied to directly identifying CB₂ receptors in the brain, it will not be further discussed here. In situ hybridization studies can yield useful information on which cells express CB₂ and thus can complement PCR-based studies. However, the lower sensitivity of in situ hybridization may make this difficult. A necessary control for in situ hybridization includes lack of hybridization in knockout tissues (when possible).

Immunocytochemistry studies have the powerful potential to identify the precise localization of CB₂. However, for meaningful information to be drawn from them it is essential that proper controls are followed [as reviewed by Bussolati and Leonardo (2008); Lorincz and Nusser (2008); Saper and Sawchenko (2003)]. Briefly, some of the controls are: adsorption with the immunizing peptide, parallel, blinded staining of wild-type and knockout tissue, the use of two or more antibodies raised against distinct epitopes, antibody titration, omitting the primary antibody from the staining procedure and supporting these findings with RT-PCR, in situ hybridization and other such detection methods. Using just one control for one experimental setup is usually insufficient proof of specificity. These basic controls must be remembered when interpreting the data presented from any study cited in this review and future studies as well.

In conclusion, despite originally being thought of as the ‘peripheral’ cannabinoid receptor, considerable functional and anatomical evidence suggests that CB₂ is expressed in the nervous system – certainly in activated microglia and very likely in some neurons. In addition, this raises the point that any report that identifies CB₂ in neurons of the nervous system must incorporate careful controls to ensure that the CB₂ signal found originates from neurons and not from microglia or immune cells associated with brain blood vessels. Given the importance of determining the functional role of CB₂ in the CNS, under what conditions it is up regulated, and the potential therapeutic applications of CB₂ agonists it is vital to understand where in the CNS CB₂ receptors are expressed. We encourage those working in the field and those reviewing manuscripts to conduct and review these studies in a careful, thoughtful and rigorous fashion.

Glossary

Abbrevations:

2-AG

2-arachidonylglycerol

CB_1/2

type1/2 cannabinoid receptor

DMNX

dorsal motor nucleus of the vagus

DRG

dorsal root ganglion

FACS

fluorescence activated cell sorting

GFAP

glial fibrillary acidic protein

ICC

immunocytochemistry

IHC

immunohistochemistry

IPSC

inhibitory post synaptic current

ISH

in situ hybridization

MAPK

mitogen-activated protein kinases

NB

northern blot

PGE2

prostaglandin E2

PI3K

phosphoinositide 3-kinase

RLB

radioligand binding

RT-PCR

reverse transcriptase polymerase chain reaction

SB

Southern blot

THC

Δ⁹-tetrahydocannabinol

TRPV1

transient receptor potential cation channel V1

WB

western blot

Conflicts of interest

None.