In animal experiments, some of the sensory abnormalities associated with neu- ropathic pain—allodynia and hyperalgesia—have been reproduced by injury to peripheral nerves (Bridges et al. 2001b). CB treatment has been demon- strated to reduce some of these behavioural effects of neuropathic pain in animal models. After a unilateral CCI to the rat sciatic nerve, behavioural hypersensitivity to cold, mechanical and thermal stimuli can be measured by timing hind paw withdrawal latencies. At doses (2.14 mg/kg) that did not affect withdrawal latencies of hind paws contralateral to the injury, systemic application of WIN55,212-2 was effective at reducing heat and mechanical hyperalgesia and mechanical and cold allodynia following the induction of neuropathy (Herzberg et al. 1997). The same rodent neuropathy model was
used to show that intrathecal injections of Δ9THC were also effective at length- ening withdrawal latencies to thermal stimuli (Mao et al. 2000). In contrast, the dose-response curve produced for the thermal anti-nociceptive effects of
intrathecally injected morphine, underwent a rightward shift after CCI. In an alternative neuropathic pain model, partial ligation of the rat sciatic nerve pro- duces similar changes to hind limb sensory stimulus thresholds, that manifest
as allodynia and hyperalgesia (Seltzer et al. 1990). Systemic treatment with WIN55,212-2, CP55,940 or HU210, 12–15 days after nerve injury, produced a dose-related reversal of mechanical hyperalgesia measured from 1 h to 6 h
after drug administration (Fox et al. 2001). With higher doses of these CBs, anti-nociceptive effects on contralateral withdrawal thresholds and side-effects of catalepsy and sedation were also observed. Of these three CBs, WIN55,2212-
2 had the best side-effect proﬁle and was also reported to reverse mechanical
allodynia (0.3–3 mg/kg) and reduce thermal hyperalgesia (3 mg/kg). In addi- tion, intrathecal injections of WIN55,212-2 (as well as local injections into hind paw tissue) were effective at reducing mechanical hyperalgesia. AM404, an in- hibitor of anandamide uptake, has been shown to enhance the anti-nociceptive and hypotensive effects of anandamide in vivo (Beltramo et al. 1997; Calignano et al. 1997). This drug also decreases spinal cord Fos immunoreactivity pro- duced in the CCI model (Rodella et al. 2005). In support of these studies, 7 days after a ligation injury to the L5 spinal nerve in rat, systemic administration of WIN55,212-2 (0.5–2.5 mg/kg) dose-dependently reduced cold allodynia and thermal hyperalgesia, with higher doses reducing hypersensitivity to mechan- ical stimuli (Bridges et al. 2001a). In a subsequent study, tactile and thermal hypersensitivity produced by the ligation of both L6 and L5 spinal nerves was
dose-dependently inhibited by systemic administration of the CB2 receptor agonist AM1251. This effect could be reproduced in CB1 −/− and CB1+/+ mice with the same nerve injury, thus claiming the advantage of circumventing
side-effects mediated by CBs active at CB1 receptors in the CNS (Ibrahim et al.
2003; Malan et al. 2001). The analgesic action of AM1241 injected directly into
the hind paw skin, has been demonstrated in both acute and inﬂammatory pain models (Malan et al. 2001, 2003; Nackley et al. 2003a; Quartilho et al.
2003) but not so far in neuropathic animals. In an electrophysiological study, however, local injection of the CB2-selective agonist JWH-133 into the hind paw is reported to directly reduce the mechanically evoked responses of spinal cord neurones in the same L5/L6 spinal nerve ligation (SNL) neuropathy model (Elmes et al. 2004). Intrathecal injection of CP55,940 was also effective at in- hibiting tactile allodynia in this model, although treatment was associated with signiﬁcant behavioural toxicity that could be alleviated by treatment with CB1 but not CB2 receptor antagonists (Scott et al. 2004).
In an interesting study on CCI-induced neuropathy (Costa et al. 2004c), rats
were given daily systemic injections of a low, sub-effective (0.1 mg/kg) dose of WIN55,212-2 from day 1 to days 7 or 14 after injury. This dose does not produce CNS side-effects or have any effect on sensory thresholds in the con- tralateral paw, nor does it affect ipsilateral sensory thresholds when injected after the development of neuropathic pain (Costa et al. 2004c; Bridges et al.
2001a). Repeated administration of WIN55,212-2 during the development of neuropathy, effectively normalised mechanical hypersensitivity and reduced thermal hyperalgesia compared to vehicle-treated nerve-injured controls. In- dicators of inﬂammatory injury, like plasma elevations of PGE2 and increased levels of neuronal NOS and NO production in sciatic nerve tissue, were also attenuated by WIN55,212-2 treatment (Costa et al. 2004c). Furthermore, CBs have been shown to have analgesic properties in a model of peripheral nerve demyelination-associated pain (Wallace et al. 2003). Intrathecal administration of WIN55,212-2 reverses, in a CB1-mediated manner, the thermal hyperalgesia and mechanical allodynia in mice that had been rendered neuropathic fol- lowing injection topical application of lysolecithin into peripheral nerves of the hind limb. This CB is also reported to be effective against behavioural signs of neuropathic pain in a model of diabetic neuropathy induced in mice. Both systemic and local hind paw skin injections of WIN55,212-2 were ef- fective at reducing mechanical sensitivity in this model (Dogrul et al. 2003; Ulugol al. 2004).
Mechanisms of CB-Mediated Analgesia
Targeted application of CB receptor agonists and antagonists to supraspinal
(Martin et al. 1998), spinal (Richardson et al. 1998a) and peripheral sites
(Malan et al. 2001) in the pain pathway has provided evidence for CB-mediated analgesic signalling mechanisms operating at these loci. Interestingly, other intrinsic pain control signalling systems like the GABAergic (Naderi et al. 2005), nor-adrenergic (Gutierrez et al. 2003) and opioidergic systems (Manzanares et al. 1999) are also known to operate at these sites. The cellular mechanisms by which CBs mediate their analgesic effects have yet to be fully elucidated, as has the degree to which the operations of parallel analgesic signalling systems in these regions might be integrated. In addition, a recent study has suggested that endogenous CBs mediate the non-opioid component of stress-induced analgesia (Hohmann et al. 2005)
The central pain modulatory systems operating in mammals are thought to consist of neural circuits in the forebrain, brain stem regions like PAG and RVM and the spinal cord. Major evidence for the involvement of these regions in pain processing was derived from experiments that demonstrated their role in the anti-nociceptive effects of systemically administered analgesics such as morphine. Repetitions of these experiments have recently shown that some of these same neural circuits are also involved in the anti-nociceptive effects of systemically applied CBs.
CB Receptors in Brain
The anti-nociceptive effects of CBs, as measured using the tail ﬂick assay, can be reproduced by brain area-speciﬁc microinjection into the RVM, PAG and central and basolateral nuclei of the amygdala (Martin et al. 1998, 1999b; Mon- hemius et al. 2001; Meng et al. 1998). The amygdala region of the limbic fore- brain is implicated in the processing of emotional information. Its functional repertoire also includes modulation of pain sensations such as the induction of stress-induced anti-nociception. Targeted inactivation of the central nucleus region of the amygdala (CeA) in rhesus monkeys and rodents was shown to attenuate the anti-nociceptive effects of systemically administered morphine (Manning et al. 2001, 2003). In rats, this was assessed by two outcome measures: an increase to tail ﬂick latencies and a decrease in the level of Fos protein in the spinal cord evoked by the injection of formalin into the hind paw. Chemical activation of nociceptors in hind paw skin by formalin produces a bi-phasic nociceptive behaviour that is strongly correlated to the level of Fos protein induced in the lumbar spinal cord (Presley et al. 1990). The activity of spinal cord neurones receiving nociceptive input from the hind paw is modulated by descending inputs. Inactivation of the CeA region results in dysfunction of these inputs, as indicated by a decrease in the extent to which morphine is able
to reduce Fos levels. Repetition of these experiments with systemic admin- istration of WIN55212-2 similarly demonstrated that inactivation of the CeA region was also effective at reducing the full anti-nociceptive effects of this CB. WIN55,212-2 reduction of formalin-induced Fos was similarly affected, impli- cating the role of descending inputs in the systemic anti-nociceptive effects of CBs, as well as opioids, in the spinal cord (Manning et al. 2003). CB1 receptors are enriched in the limbic system and are the likely targets for CB-mediated ef- fects on both the anxiolytic and analgesic functions of this region (Herkenham et al. 1990; Glass et al. 1997).
As further evidence for the involvement of descending nociceptive path- ways, the selective depletion of noradrenaline in the terminals of descending projections to the rat lumbar spinal cord was effective at suppressing the anti- nociceptive efﬁcacy of both opioids and WIN55,212-2 in both acute and tonic nociception assays (Martin et al. 1999a; Gutierrez et al. 2003). Consistent with these results, CB-mediated anti-nociception can be attenuated by antagonis-
ing α2-noradrenaline receptors (Martin et al. 1991). The microinjection of
WIN55,212-2 directly to the A5 noradrenergic nucleus in the brain stem also
produces anti-nociceptive effects in the tail ﬂick test (Martin et al. 1999b). This microinjection study also implicated the superior colliculus and lateral pos- terior and submedius regions of the thalamus in CB-mediated analgesia. The presence of injury-regulated CB1 receptor expression in the thalamus lends support to this evidence (Siegling et al. 2001).
Neural circuits located in the RVM and PAG regions of the brain stem also contribute to opioid and CB-mediated analgesia. Microinjections of CBs into the RVM suppress pain-related behaviour (Martin et al. 1998; Monhemius et al. 2001) and targeted inactivation of the RVM (by pharmacological activa- tion of GABAA receptors) prevents the analgesic action of systemic CBs (Meng et al. 1998; Meng and Johansen; 2004). In vitro recordings from brain stem slices have demonstrated a CB1-mediated pre-synaptic inhibition of GABAer- gic neurotransmission in the RVM (Vaughan et al. 1999). Extracellular single unit recordings, conducted during local microinfusion of WIN55,212-2 to the RVM, has determined its effect on the activity of individual neurones involved in withdrawal reﬂexes (Meng and Johansen 2004). Before a withdrawal re- ﬂex, the activity of RVM ‘on-cells’ increases whilst ‘off-cell’ activity decreases. CB-mediated activation of CB1 receptors prolongs tail withdrawal latencies by inhibiting the activity of on-cells and disinhibiting off-cell ﬁring. Endoge-
nous opioids released in the RVM operate via μ-opioid receptors to inhibit limb withdrawals by the same mechanism. However, pretreatment with μ-opioid re- ceptor antagonists have demonstrated that the analgesic effects of WIN55,212-2 in the RVM occur independently (Meng et al. 1998). Pre-treatment with the
selective CB1 receptor antagonist SR141716a also provided evidence that RVM CB1 receptors are tonically activated to suppress nociceptive behaviours. In a separate study, microinfusions of WIN55,212-2 or SR141716a into the nucleus reticularis gigantocellularis pars α (GiA) region of the RVM also demonstrated
CB1-mediated inhibition of nociceptive behaviours in the tail ﬂick and forma- lin tests (Monhemius et al. 2001). The results suggested that sustained input from ascending nociceptive pathways (provided by peripheral formalin injec- tion or nerve injury) is required to stimulate endocannabinoid production in the RVM and drive descending inhibition.
Further evidence that endocannabinoid production in the brain stem mod- ulates pain responses, is provided by a microdialysis study in rats (Walker et al.
1999). Release of AEA in the periaqueductal grey was measured after formalin injection into the hind paw or local electrical stimulation of the dorsal and lateral PAG. Enhancement of AEA levels in the PAG, by direct electrical stim-
ulation, produced an analgesic effect that was mediated by CB1 receptors. In support of this mechanism, formalin-evoked Fos protein, induced in the PAG, was attenuated by a local microinjection of HU210 (Finn et al. 2003). In vitro electrophysiological studies have revealed a CB1-dependant modulation of ex- citatory and inhibitory post-synaptic currents in the PAG. The pre-synaptic inhibitory mechanisms mediating this action were also exhibited by μ-opioids (Vaughan et al. 2000).