Neuropathic pain is caused by lesion or dysfunction of the peripheral sensory nervous system. Up-regulation of the voltage-gated Ca2+ channel subunit α2δ-1 in DRG (dorsal root ganglion) neurons and the spinal cord correlates with the onset of neuropathic pain symptoms such as allodynia in several animal models of neuropathic pain. The clinically important anti-allodynic drugs gabapentin and pregabalin are α2δ-1 ligands, but how these drugs alleviate neuropathic pain is poorly understood. In the present paper, we review recent advances in our understanding of their molecular mechanisms.
Unlike nociceptive or acute pain, neuropathic pain occurs without continuous noxious peripheral input. Patients with neuropathic pain experience spontaneous pain that is described as ‘electric-shock-like, burning and tingling’. Further symptoms of neuropathy are the painful response to normally innocuous stimuli (allodynia) and the increased response to noxious stimuli (hyperalgesia). In addition, patients also suffer from depression, anxiety and insomnia as a result of their chronic pain condition. A recent survey showed that up to 8% of the population of the U.K. may endure chronic pain of predominantly neuropathic origin .
Neuropathic pain is caused by damage to primary sensory afferent neurons of the peripheral nervous system that relay nociceptive and non-nociceptive peripheral stimuli to the spinal cord and the brain  as a result of, e.g., trauma, diabetes, cancer and chemotherapy . Diabetes is the most common cause of neuropathic pain in the U.K. The estimated annual NHS (National Health Service) expenditure on the treatment of peripheral diabetic neuropathy and its associated complications is in the range of 250 million and is expected to rise .
The axons of primary sensory afferents form the spinal nerves and dorsal roots, with their cell bodies residing within the DRGs (dorsal root ganglia) . Damaged DRG neurons become hyperexcitable (increased probability to fire action potentials) and show ectopic activity (spontaneous firing of action potentials). Hyperexcitability and ectopic activity lead to increased transmitter release in the spinal cord and this causes central sensitization (increased excitability of neurons within the central nervous system). These changes in neuronal activity together with increased activation of descending pathways from the brain are the mechanistic components in the development and maintenance of neuropathic pain .
The role of the voltage-gated Ca2+ channel subunit α2δ-1 in neuropathic pain
Voltage-gated Ca2+ channels are heteromultimeric complexes consisting of the pore-forming Cavα1 subunit that determines the main biophysical properties of the channel and the auxiliary subunits β and α2δ (for a review, see ). α2δ increases Ca2+ currents by increasing the number of functional channels at the plasma membrane through enhancement of Cavα1 trafficking to the plasma membrane, and by stabilization of the channels at the cell surface . The α2δ subunit consists of two proteins that are derived from a single gene product by proteolytic cleavage, namely the extracellular α2- and the δ-protein that is thought to be transmembrane. The proteins α2 and δ remain linked via disulfide bonds. The α2-protein is heavily glycosylated and harbours the vWF-A (von Willebrand factor-A) domain . An intact vWF-A domain is a prerequisite for the positive effect of α2δ on Ca2+ channel forward trafficking . So far, four genes encoding α2δ subunits (α2δ-1–α2δ-4) have been identified. α2δ-1 was cloned from skeletal muscle, but was found to be relatively ubiquitously expressed, whereas α2δ-2 and α2δ-3 are more restricted to the brain and α2δ-4 to certain endocrine tissues and retina .
Accumulating evidence from numerous groups points to an important role of α2δ in neuropathic pain. α2δ-1 mRNA and protein levels are dramatically up-regulated in affected DRGs in several models of neuropathic pain, and this increase in α2δ-1 correlates with the onset of allodynia [7–11]. In contrast with α2δ-1, α2δ-2 and α2δ-3 were found to be down-regulated in the unilateral lumbar SNL (spinal nerve ligation) model of neuropathic pain , demonstrating the dominant role played by α2δ-1 in neuropathy . Furthermore, transgenic mice overexpressing α2δ-1 show allodynic symptoms even in the absence of nerve damage, indicating that increased levels of α2δ-1 are sufficient to cause neuropathic pain .
In a recent study , we performed a detailed analysis of the distribution of α2δ-1 in DRG neurons in the SNL model (Figure 1). DRG neurons are heterogeneous in morphology and their function correlates with the size of their somata. Small and medium-sized somata belong to mainly C- and Aδ-nociceptors, whereas non-nociceptive Aβ fibres have a large soma . This study showed that α2δ-1 protein levels were elevated in all three size groups of DRG neuron somata at the level of ligation, regardless of their function (Figure 1A) . The augmentation occurred in the endoplasmic reticulum and at the plasma membrane.
Distribution of α2δ-1 following SNL
α2δ-1 protein levels were also found to be increased in the dorsal roots that are formed by the central axon branches of DRG neurons (Figure 1A) . This increase intensified over days following ligation (experimental time points were 2, 4 and 7 days post-SNL). Within the dorsal roots, α2δ-1 was localized to tubular-vesicular structures that are implicated in protein trafficking . A subset of DRG neurons involved in touch and proprioception do not form synapses with spinal cord dorsal horn neurons, but their central axons directly project up to the brainstem. These axons form the fasciculus gracilis as part of the spinal cord dorsal column, and transection of the dorsal column prevents tactile allodynia symptoms in SNL rats . α2δ-1 was increased in the fasciculus gracilis starting at the level of ligation, and this augmentation continued up to the brain stem .
The majority of DRG neurons, however, form glutamatergic synapses on to second-order sensory neurons in the spinal cord dorsal horn. α2δ-1 levels are increased in the spinal cord dorsal horn [10,11] and this up-regulation of α2δ-1 is crucial for the aetiology of neuropathic pain . The increase occurred in the superficial and to some extents also in the deeper layers of the dorsal horn (Figure 1A) . This increase in α2δ-1 protein levels was not accompanied by an increase in α2δ-1 mRNA in the spinal cord dorsal horn neurons , but was due to an increase in α2δ-1 in the presynaptic terminals of the DRG neurons . Such an increase in α2δ-1 in presynaptic terminals of affected DRG neurons is thought to enhance Ca2+ influx at the nerve terminals and therefore increases synaptic transmitter release into the spinal cord which then causes central sensitization .
The α2δ ligands and anti-allodynic drugs pregabalin and gabapentin inhibit anterograde trafficking of α2δ
The current first-line treatment of neuropathic pain comprises the anticonvulsant gabapentinoid drugs gabapentin and pregabalin. These drugs have an analgesic effect on neuropathic neuronal activity without affecting baseline nociception . Gabapentin was originally designed as an analogue of GABA (γ-aminobutyric acid) with increased membrane permeability, but, like its more potent derivative pregabalin, showed little affinity for the relevant GABA-binding sites (for a review, see ). Various molecular targets have been proposed (for a review, see ), but, to date, just one high-affinity binding site has been found, namely α2δ .
Of the four known isoforms of α2δ, only α2δ-1 and α2δ-2 interact with the gabapentinoid drugs, and α2δ-1 has a higher affinity than does α2δ-2 . The arginine residue of α2δ-1 at position 217 close to the vWF-A domain is critical for the binding . Mutating this arginine residue to alanine (R217A) greatly reduces drug-binding affinity and R217A knockin mice develop neuropathic pain that is insensitive to pregabalin and gabapentin . These observations indicate that the binding of the gabapentinoid drugs to α2δ-1 is required for their analgesic effect.
Several studies have shown that pregabalin and gabapentin reduce neurotransmitter release , but the molecular mechanism of this inhibition is unclear. Transmitter release by means of stimulus-coupled Ca2+-dependent exocytosis depends on a tight spatial and temporal interplay of a plethora of processes and proteins . Transmitters are stored in specialized vesicles and released in a highly controlled manner. Fusion of the vesicle with the plasma membrane is the final step of secretion when Ca2+ that enters the cell through Ca2+ channels triggers the fusion of the vesicle with the cell membrane. Therefore an obvious mechanism would be that the gabapentinoid drugs inhibit Ca2+ channels directly by binding to α2δ-1, which would then reduce presynaptic Ca2+ influx and subsequent transmitter release. However, evidence for a direct inhibitory effect of gabapentinoid drugs on native Ca2+ currents and synaptic transmission is inconclusive . Whereas some studies reported a small reduction of Ca2+ current when the drugs were applied acutely, others did not observe such an acute inhibitory effect on either heterologously expressed channels or endogenous Ca2+ currents in DRG neurons in vitro [22,23]. Only a prolonged and chronic application of pregabalin and gabapentin was able to reduce Ca2+ influx [22,23]. Thus it seems unlikely that gabapentinoid drugs inhibit Ca2+ channels directly. However, chronic application of gabapentin and pregabalin reduced the amount of the Ca2+ channel subunits α2δ and Cavα1 at the cell surface without affecting their rate of endocytosis [11,22]. This led to the conclusion that chronic application of gabapentinoid drugs in vitro reduced Ca2+ influx owing to a reduction of the forward trafficking of α2δ and Cavα1 to the plasma membrane. Interestingly, to reduce Ca2+ channels at the plasma membrane, gabapentin had to be taken up into the cell via the system L amino acid transporter and possibly displaced an endogenous ligand that is a positive modulator of α2δ forward trafficking .
A classical approach to study neuronal protein trafficking in vivo is to perturb trafficking by nerve ligation or nerve section. Proteins that are being trafficked accumulate at the site of the obstruction, and the level of accumulation is a measure of trafficking activity. In the SNL model of neuropathic pain, α2δ-1 was found to accumulate proximal to the SNL site . These results showed for the first time that endogenous α2δ-1 is indeed subject to anterograde trafficking from the DRG somata to both peripheral and central terminals (Figure 1A). Chronic treatment of SNL animals with repetitive injections of pregabalin had a profound anti-allodynic effect on neuropathic pain symptoms and it inhibited the accumulation of α2δ-1 at the SNL site . Moreover, it also reduced α2δ-1 in ascending DRG axons of the fasciculus gracilis and, most importantly, reduced the increase of α2δ-1 in the presynaptic terminals of DRG neurons in the spinal cord dorsal horn. Because chronic pregabalin and gabapentin had no effect on the up-regulation of α2δ-1 in DRG somata (Figure 1B) [11,24], these results strongly suggest that the anti-allodynic effect of chronic gabapentinoid drug treatment was due to an inhibition of anterograde trafficking of α2δ-1 in vivo.
Conclusion and outlook
Recent advances in our understanding of the molecular mechanisms of gabapentinoid drugs indicate that pregabalin alleviates neuropathic pain by impairing the trafficking of α2δ-1 to presynaptic terminals of DRG neurons which would reduce Ca2+ influx and transmitter release in the spinal cord and subsequently reduce spinal sensitization. This intracellular effectiveness of pregabalin is a novel molecular mechanism to explain the anti-allodynic effect of pregabalin and is in clear contrast with other analgesic drugs that influence their targets at the cell surface. Further studies are needed to unravel how and where in the cell gabapentinoid drugs affect the trafficking machinery of α2δ-1.
Synaptopathies: Dysfunction of Synaptic Function: A Biochemical Society Focused Meeting held at The Hotel Victoria, Newquay, U.K., 2–4 September 2009. Organized and Edited by Nils Brose (Max Planck Institute for Experimental Medicine, Göttingen, Germany), Vincent O'Connor (Southampton, U.K.) and Paul Skehel (Centre For Integrative Physiology, Edinburgh, U.K.)
This work was supported by the Biotechnology and Biological Sciences Research Council [grant number BB D018250 (to A.H.D. and A.C.D.)], the Wellcome Trust [grant number GR077883MA (to A.C.D.)], Epilepsy Research UK (to A.C.D.), Junta de Comunidades de Castilla la Mancha [grant number PA108-0174-6967 (to R.L.)] and a University College London Ph.D. scholarship to A.T.-V.-M.