The importance of supraspinal mechanisms to the anti-nociceptive action of systemically administered CBs, is highlighted by the ﬁnding that spinal transection reduces the anti-nociceptive effect of systemically administered
Δ9THC and CP55,940, as measured by tail ﬂick latency (Lichtman and Mar- tin 1991) and noxious heat-evoked activity in spinal wide dynamic range
neurones (Hohmann et al. 1999b). Yet there is evidence that CBs can also operate via local analgesic mechanisms sited in the spinal cord. CBs applied
directly to the spinal cord suppress the responses of dorsal horn neurones to nociceptive input from C-ﬁbres (Tsou et al. 1996; Hohmann et al. 1995,
1998; Drew et al. 2000; Harris et al. 2000) and have the ability to raise noci- ceptive thresholds in normal animals (Richardson et al. 1997; Pertwee et al.
2001). Analgesic effects can persist even after spinal transection (Smith and Martin 1992), suggesting that at least some of the anti-nociceptive actions of CBs are intrinsic to CB signalling mechanisms in the spinal cord (Licht-
man and Martin 1991). Intrathecal delivery of CBs also reduces pain be- haviours associated with tissue injury: inﬂammatory thermal hyperalgesia was decreased by intrathecal AEA in a manner that was reversed by intrathecal
administration of the selective CB1 antagonist SR141716a (Richardson et al.
1997). Intrathecal WIN55,212-2 attenuated pain behaviour following formalin
injection via a CB1—but not a CB2—receptor-dependent mechanism (Rice et al. 2002). Intrathecally administered HU210 was found to be more effec- tive than morphine in attenuating formalin-evoked pain behaviour (Guhring et al. 2001).
Spinal CBs also suppress mechanical hypersensitivity after inﬂammatory (Martin et al. 1999c) and neuropathic (Fox et al. 2001) injury. Furthermore, hyperalgesia (evoked by injections of capsaicin into the hind paw) can be sup- pressed by spinal CP55,940 (Johanek et al. 2001). Electrophysiological record- ings have shown that the sustained noxious input generated by capsaicin in this model potentiates the activity of spinal nociceptive neurones, allowing them to be activated by sub-threshold sensory stimuli. In this study, the sensi- tised responses of dorsal horn cells to mechanical stimuli could be attenuated by activation of spinal CB1 receptors with CP55,940. This suggests that the hyperexcitability of dorsal horn neurones (an underlying mechanism for hy- persensitivity after tissue injury) can be inhibited by CBs applied directly to the spinal cord. In support of this claim, short-term potentiation or ‘wind-up’ responses in these neurones and the induction of Fos protein are both sup- pressed by WIN55,212-2 applied directly to the spinal cord (Strangman and Walker 1999; Martin et al. 1999c).
The mechanism of action to explain the inhibitory effect of CBs at the cen- tral synapses of nociceptive C-ﬁbres has not yet been fully elucidated. The location of CB1 receptors in this area indicates that there could be several sites of action for exogenous CBs applied to the cord surface (Sect. 1.3.2). Functional evidence from patch clamp studies—recording from second or- der neurones in adult rat trigeminal caudal nucleus (Liang et al. 2004) and spinal cord slices (Morisset et al. 2001)—provides evidence for CB inhibition of glutamate transmission from pre-synaptic C-ﬁbre terminals. WIN55,212-2 is proposed to operate at CB1 receptors located pre-synaptically on the cen- tral terminals of sensory afferents by inhibiting calcium inﬂux via N-type voltage-gated calcium channels (Liang et al. 2004). This mechanism is fur- ther supported by peptide release studies from spinal cord tissue, which also serve to strengthen the evidence for a spinal site of CB analgesia. Using ex vivo spinal cord tissue mounted in a superfusion chamber, the excitation- mediated release of peptides found in the central terminals of C-ﬁbres can be measured in the superfusate ﬂuid in response to C-ﬁbre stimulation. CB application to the tissue reduces the release of CGRP and SP peptides after electrical or capsaicin stimulation of C-ﬁbres, an effect that can be attenuated by CB1 receptor antagonists (Richardson et al. 1998b; Tognetto et al. 2001; Lever et al. 2002; Brooks et al. 2004). Likewise, in cultured sensory neurones,
voltage-activated Ca2+ currents can be attenuated by CB1-mediated inhibition of N-type VGCC (Ross et al. 2001; Khasabova et al. 2004), and capsaicin-
evoked increases in [Ca2+]i are also suppressed by CBs (Millns et al. 2001; Oshita et al. 2005).
Collectively, the evidence suggests that CB1 receptor activity can reduce the increases in [Ca2+]i that drive transmitter exocytosis in sensory neurones. It is plausible that this mechanism also operates at the central terminals of
cutaneous sensory neurones in the spinal cord; however, there is a paucity of anatomical evidence for CB1 receptors in this location (Sect. 1.3.3).
Several studies also suggest that there maybe tonic inhibitory control over dorsal horn neurones receiving nociceptive input exerted by endogenous CBs acting at pre-synaptic CB1 receptors. Spinal administration of SR141716a has been shown to facilitate responses of dorsal horn neurones to noxious C-ﬁbre input (Chapman et al. 1999) and reduce the threshold for thermal nociceptive responses in rats (Richardson et al. 1997). This effect is pos- sibly due to the increased release of excitatory transmitters from stimu- lated C-ﬁbre terminals, as evidence for this was provided by the ability of SR141716a to increase capsaicin-evoked peptide release from both spinal cord tissue and cultured DRGs (Lever et al. 2002; Ahluwalia et al. 2003a). How- ever, the application of SR141716a to spinal cord slices had no effect on the excitability of dorsal horn neurones (Morisset et al. 2001). The produc- tion of endocannabinoids from C-ﬁbre-responsive dorsal horn neurones has been proposed to explain the ability of AMPA (S-alpha-amino-3-hydroxy-5- methyl-4-isoxazolepropionic acid) superfusion to reduce electrically evoked CGRP release from spinal cord slices, as this inhibitory effect can be reversed by SR141716a (Brooks et al. 2004). In this setting, endocannabinoids pro- duced by excitation of post-synaptic neurones may signal retrogradely back onto pre-synaptic CB1 receptors in order to reduce excitatory transmission from these ﬁbres. In support of this effect, the increased production of en- dogenous CBs by systemic treatment with the AEA uptake inhibitor AM404, reduces Fos protein induction (Rodella et al. 2005). Endocannabinoid sig- nalling at CB1 receptors in the dorsal horn is also cited to explain the anal- gesic effects of some nonsteroidal anti-inﬂammatory drug (NSAIDs) (Ates et al. 2003).
As a substantial proportion of dorsal horn CB1 receptors are expressed by interneurone populations, including GABAergic neurones and those express- ing opioid receptors (Sect. 1.3.2), the spinal analgesic action of CBs may operate via other intrinsic inhibitory systems. Interaction between CB and opioid sig- nalling is consistent with reports of synergism and cross-tolerance between their analgesic effects in the spinal cord (Manzanares et al. 1999). Delivery of intrathecal CBs has been used to enhance endogenous opioid levels measured in the spinal cord (see Welch and Eads 1999). In addition, opioid receptor an- tagonists can be used to reverse the anti-hyperalgesic effects of intrathecal CBs (Thorat and Bhargava 1994; Welch and Eads 1999). This is except in models of neuropathic pain (Fox et al. 2001; Bridges et al. 2001a; Mao et al. 2000), where the reduction of spinal opioid binding sites is more extensive compared to CB1 (Hohmann et al. 1999a) and may explain why, in contrast to opioids, the anti-nociceptive potency of CBs remains unchanged after a nerve injury (Mao et al. 2000; Dogrul et al. 2003). A recent study by Naderi et al. (2005) has demonstrated that spinal GABAB receptors do not mediate the analgesic effects of intrathecal CP55490 measured in the formalin test. Conversely, the analgesic effect of intrathecal baclofen (a GABAB receptor agonist) is prevented by pre- treatment with SR141716a. Interestingly, this suggests that endocannabinoid
signalling at CB1 receptors is required for baclofen-mediated analgesia in this model.
It has recently become clear that not only spinal neurones but also mi-
croglia play a role in the CNS responses to inﬂammation and peripheral nerve injury that underlie persistent pain (DeLeo et al. 2001; Watkins et al. 2001; Watkins et al. 2003; Tsuda et al. 2005). Evidence suggests that microglia ex- press functional CB2 receptors and synthesise endocannabinoids and that CBs can inﬂuence microglial migration (Walter et al. 2003). Furthermore, CBs block cytokine mRNA expression in cultured microglial cells (Puffenbarger et al. 2000) and are implicated in anti-inﬂammatory and neuroprotective ef- fects mediated by CB receptors on glial and neuronal cells (Molina-Holgado et al. 2003; see Walter and Stella 2004). The appearance of CB2 mRNA in the spinal cord after nerve injury coincident with microglial activation (Zhang et al. 2003), together with the analgesic efﬁcacy of selective CB2 agonists in inﬂammatory and neuropathic pain (Malan et al. 2001; Ibrahim et al. 2003), may implicate the function of spinal CB2 receptors in CB-mediated analgesia. Most of the experimental evidence so far indicates that the analgesic effects of spinally administered CBs are predominately mediated via CB1 receptors.
A number of recent studies have demonstrated that pain responses can be reduced by CBs applied locally to peripheral tissues at systemically inactive doses. These effects are locally reversed by CB receptor antagonists, implying that the analgesic mechanisms involve peripheral CB receptors and are distinct from the actions of CBs at receptors in the brain and spinal cord.
After inﬂammatory injury, both the effects of paw oedema and thermal hyperalgesia can be attenuated by a local injection of a low dose of AEA into the paw skin, but not when the same dose of AEA was applied systemically. Co-administration of the CB1 receptor antagonist SR141716a reversed the ef- fect of AEA on thermal hypersensitivity. It also attenuated the CB-mediated reduction of plasma extravasation after capsaicin inﬂammation, thus demon- strating both an anti-hyperalgesic and anti-inﬂammatory action for CBs op- erating via peripheral CB1 receptors (Richardson et al. 1998b). The peripheral anti-hyperalgesic actions of CBs, after capsaicin-induced inﬂammation, were also attenuated by CB1 receptor antagonists (Johanek et al. 2001; Ko et al. 1999). Local injection of WIN55,212-2, HU210, AEA or methanandamide in the for- malin model of the rat hind paw skin inﬂammation conﬁrmed a reduction in pain behaviour via the activation of peripheral CB1 receptors, excluding effects via CB2 or opioid receptors (Calignano et al. 1998). However, hind paw injection of PEA, reduced pain behaviour via a CB2-dependent mechanism, and co-injection of AEA and PEA produced a synergistic analgesic effect. This suggested the existence of two separate analgesic mechanisms operating via
peripheral CB1 and CB2 receptors. In support of this hypothesis, the peripheral analgesic effects of WIN55,212-2–mediating the reduction of hyperalgesia or spinal Fos induction after carrageenan inﬂammation–can be blocked sepa- rately by CB1 and CB2 receptor antagonists (Nackley et al. . 2003a; Kehl et al.
2003). A peripheral analgesic action of CBs mediated by CB2 receptors has been
conﬁrmed in several inﬂammatory pain models by the use of CB2-selective ag-
onists and antagonists (Hanus et al. 1999; Malan et al. 2001; Quartilho et al.
2003; Malan et al. 2003; Hohmann et al. 2004; Elmes et al. 2004).
Systemic CB2 receptor-selective agonists are also effective against paw tissue oedema (Clayton et al. 2002; Malan et al. 2003). In electrophysiological studies, local injection of CB2 agonists could inhibit both the normal and potentiated responses of dorsal horn neurones (after inﬂammation of the hind paw, CB2 agonists suppressed both the receptive ﬁeld expansion and wind-up responses in spinal cord neurones) (Sokal et al. 2003; Nackley et al. 2004; Elmes et al.
2004). Local injection of a CB1-selective agonist also suppresses responses to mechanical stimuli in carrageenan inﬂamed and non-inﬂamed rats (Kelly et al.
2003). Interestingly, however, the peripheral effects of AEA on these responses were attenuated by a CB2 antagonist (Sokal et al. 2003).
Peripheral CB1 and CB2 receptors have both been implicated in the anti- hyperalgesic effects of locally injected CBs after nerve injury (Fox et al. 2001; Ibrahim et al. 2003; Elmes et al. 2004; Ulugol et al. 2004; see Sect. 2.2). Thus, CBs have anti-inﬂammatory, anti-hyperalgesic and anti-nociceptive actions that can be mediated by peripheral CB1 and CB2 receptors.
Endogenous CBs have been measured in skin (the site of the peripheral
analgesic action of CBs) and their levels here are increased by tissue injury (Calignano et al. 1998; Oka et al. 2005). Therefore, CBs in the skin may interact directly with receptors on the peripheral endings of cutaneous nociceptive sen- sory neurones to modulate the transmission of nociceptive signals to the spinal cord. As evidence of this direct mechanism, functional CB1 receptors have been demonstrated on cultured sensory neurones (Ross et al. 2001; Khasabova et al.
2004; Millns et al. 2001; Oshita et al. 2005) where inhibition of excitatory cal- cium transients reduces the exocytosis of neuropeptides both from sensory neurone cell bodies (Ahluwalia et al. 2003a; Oshita et al. 2005) and the pe- ripheral terminals of these cells stimulated in skin (Richardson et al. 1998b; Ellington et al. 2002). The suppressive effects of topical CBs on efferent C-ﬁbre functions in vivo (Sect. 1.3.3) are also thought to be mediated by the inhibition
of excitatory neuropeptide release. The release of AEA—stimulated by [Ca2+]i increases in these neurones in culture (Ahluwalia et al. 2003b)—may provide the means for a negative feedback on excitatory transmitter release from sen- sory neurones. Nerve injury, involving transection of the nerve trunk and
disruption of distal axonal transport, is likely to reduce peripheral CB receptor levels (Bridges et al. 2002). This may explain why the anti-nociceptive pheno- type exhibited by FAAH−/− mice is retained by inﬂammatory injury but lost after nerve injury (Lichtman et al. 2004a). So far the anatomical evidence does
not fully support the hypothesis that direct activation of CB1 receptors on pe- ripheral nociceptive sensory neurones mediates the local anti-nociceptive and anti-hyperalgesic actions of CBs (see Sect. 1.3.3). Anatomical and functional evidence for CB2 receptors on these ﬁbres is also limited. Instead, stimulation of
β-endorphin release by AM1241 acting on CB2 receptors in skin keratinocytes
has been proposed as a mechanism to explain the peripheral analgesic actions of this compound, as these effects are critically dependent on μ-opioid receptor
activity (Ibrahim et al. 2005). In support of this mechanism, topical application of WIN55,212-2 and morphine were demonstrated to have synergistic effects in the mouse tail ﬂick test (Yesilyurt et al. 2003).
An alternative mechanism to explain the peripheral anti-hyperalgesic action of CBs is linked to anti-inﬂammatory effects of these compounds in periph- eral tissues. CB agonists may indirectly inhibit transmission in nociceptive afferents by reducing the release of by-products of inﬂammatory and exci- totoxic processes, which serve to sensitise peripheral nociceptive ﬁbres to mechanical and thermal stimuli. CB1 and CB2 receptor-mediated inhibition of cAMP formation could act to reduce the potentiating effects of PKA on TRPV1 receptor-mediated responses in these neurones (see Ross et al. 2004; Oshita et al. 2005). CBs are reported to reduce the production and release of pro-inﬂammatory signalling molecules, including TNF, NO and interleukin (IL)-1, and to enhance the release of anti-inﬂammatory cytokines like the IL-1 receptor antagonist, IL-4 and IL-10 (Molina-Holgado et al. 2003; Howlett et al.
2002; Walter and Stella 2004).
Mast cells synthesise and release inﬂammatory mediators that increase lo- cal vascular permeability, and others, like serotonin and NGF, can directly sensitise receptors involved in transducing noxious stimuli on sensory neu- rones. The ability of CB1 and CB2 receptor signalling to reduce inﬂammatory responses has been linked to their regulation of mast cell function (Sam- son et al. 2003; Mazzari et al. 1996). CB1 receptor activation on mast cell lines is linked to the suppression of serotonin secretory responses, whereas co-expressed CB2 receptors regulate the activation of signalling pathways that regulate gene expression. CBs are also reported to suppress immune cell proliferation and chemotactic processes that contribute to the estab- lishment of inﬂammatory sites in tissues (Howlett et al. 2002; Walter and Stella 2004). Speciﬁcally, PEA has been shown to reduce neutrophil accumu- lation in hind paw skin inﬂamed by NGF injections (Farquhar-Smith and Rice 2003). This is speculated to involve peripheral CB2-like receptor sig- nalling pathways that act to reduce the production and/or release of mast cell-derived neutrophil chemotactic factors, such as leukotriene B4 or even NGF itself. Components of the inhibitory action of CBs on oedema and plasma extravasation are also likely to be mediated by suppression of the pro-inﬂammatory functions of glial and immune cells (see Walter and Stella
2004). The reduced paw oedema observed in inﬂamed FAAH−/− mice (Licht- man et al. 2004a) was maintained in conditional (non-neuronal cell) FAAH
knock-outs, but was not sensitive to CB1 or CB2 receptor antagonists (Cravatt et al. 2004), thus implicating the involvement of other CB-sensitive receptors (Sect. 1.1).
The peripheral anti-inﬂammatory, anti-nociceptive and anti-hyperalgesic actions of CB1 and CB2 receptor agonists are of potential therapeutic im- portance because they may present a way of delivering the analgesic ben- eﬁt of CB compounds without their centrally mediated psychoactive side- effects. The CB2 receptor agonist AM1251 maybe useful in this context, as sys- temic administration produces no CNS side-effect in animal models (Malan et al. 2001).