Voltage-gated calcium channel α₂δ subunits: an assessment of proposed novel roles

Role for α₂δ-1 in calcium channel trafficking

The effect of α₂δ subunits to increase calcium current density can be partially explained by an increase in the trafficking of the channels to augment the amount on the cell surface⁵. The exact mechanism whereby α₂δ increases the density of Ca_V channels in the plasma membrane is still unclear. There was no effect of α₂δ-1 to reduce the internalization of Ca_V2.2⁵, indicating that the effect is likely to be on forward trafficking. Furthermore, the trafficking of α₂δ itself is blocked by a dominant-negative rab11 construct, suggesting the involvement of the recycling endosomes⁴⁵.

The VWA domain within the α₂ moiety of α₂δ is important for both trafficking of α₂δ and its associated effect on Ca_V channel trafficking and function^5,30,46,47. Furthermore, the presence of alternatively spliced exon 37a in the proximal C-terminus of Ca_V2.2, which is a minor splice variant expressed particularly in certain DRG neurons^48,49, increases Ca_V2.2 currents⁴⁸ and also increases its cell surface density via binding to adaptor proteins⁵⁰. We found that this increase was lost in the absence of α₂δ subunits, suggesting that this auxiliary subunit promotes particular steps in the forward trafficking process⁵⁰.

Proteomic study of Ca_V2 calcium channels

A comprehensive study of the Ca_V2 channel proteome was performed by using antibodies against Ca_V2.1 or Ca_V2.2, together with antibodies against β subunits, and cataloguing the associated proteins⁵¹. Many proteins were found to be part of this complex, although such studies do not indicate whether the interaction is direct or indirect. In contrast to initial purification studies of N-type channels⁵², and rather surprisingly to many in the field, the interaction of the channels with α₂δ proteins was found to be much less than 1:1; indeed, it depended on the mildness of the detergent used to solubilize the membranes, resulting in more or less α₂δ associated with the complex. Since we found that α₂δ subunits are present in lipid raft fractions⁵³ and subsequently identified that they are GPI-anchored²⁴, this supports the possibility that there is a rather mobile interaction between the α1 and α₂δ subunits^53,54 or that this interaction is more labile to disruption. Certainly, it also points to a pool of α₂δ which is not associated with calcium channels, which has also been identified by studies of calcium channel membrane mobility⁵⁴.

Importance of studies in knockout mouse models for elucidating potential novel roles for α₂δ subunits

The genetic ablation of particular α₂δ subunits has been found to affect neuronal and synaptic morphology in several systems^56–58, pointing to roles for α₂δ that may or may not involve calcium channels^22,59. Knockout mice have been generated for α₂δ-1³⁸, α₂δ-2²⁰, α₂δ-3⁶⁰, and α₂δ-4⁴¹. These have led to important findings regarding both calcium channel function in specific tissues and potential roles for the α₂δ proteins in neuronal and synaptic morphology and in physiological functions, especially in tissues such as cochlear hair cells⁴², spiral ganglion neurons⁵⁷, retinal photoreceptor cells⁵⁸, and Purkinje neurons^20,56, where one subtype of α₂δ predominates. However, complementary approaches are also required to elucidate the mechanisms of such effects.

Importance of α₂δ in disease states

Neuropathic pain. Cacna2d1, encoding α₂δ-1, is one of many genes whose expression is altered in experimental animals as a result of damage to sensory nerves, which may lead to chronic neuropathic pain. There is a consistent elevation of α₂δ-1 mRNA and protein^61–66 in every damaged DRG neuron^39,62. Furthermore, we have shown that, in α₂δ-1 knockout mice³⁸, there is a marked reduction in baseline responses to mechanical and cold stimulation, and a very retarded hyperalgesic response to sciatic nerve injury, in comparison with wild-type littermate mice³⁹.

Other diseases. CACNA2D1 mutations in humans have been identified to cause cardiac dysfunction, including short QT syndrome⁶⁷ and Brugada syndrome⁶⁸. Cacna2d1 knockout also resulted in a cardiovascular phenotype in mice involving a reduction in basal ventricular cardiac contractility and lower calcium current in ventricular myocytes³⁸. CACNA2D2 mutations in both humans and mice result in a recessive phenotype including epilepsy and ataxia^{20,56,69–73}, as well as a hearing deficit, related to aberrant trans-synaptic channel organization⁴². Furthermore, developmentally associated upregulation of α₂δ-2 expression suppressed axon regeneration in adult spinal cord, although the mechanism remains unclear⁷⁴. Cacna2d3 knockout mice have a hearing deficit⁵⁷ and a central pain phenotype^60,75. Finally, CACNA2D4 mutations in both humans and mice are associated with night blindness^76,77 and retinal degeneration⁵⁸.

Mechanism of action of gabapentinoid drugs which bind to α₂δ-1 and α₂δ-2

The α₂δ subunits are the target for gabapentinoid drugs⁷⁸, which bind to both α₂δ-1 and α₂δ-2 with similar affinity⁷⁹. However, from studies of mice with mutations in the gabapentin binding site within either α₂δ-1 or α₂δ-2, it was concluded that their therapeutic target both in alleviation of neuropathic pain and in epilepsy is α₂δ-1^4,80. We have found, from in vitro experiments, that incubation with gabapentin lowers the amount of α₂δ-1 and α₂δ-2 on the cell surface^5,45,81 by inhibiting their rab11-dependent recycling to the cell surface⁴⁵. In vivo, chronic administration of pregabalin to sensory nerve-injured rats reduced the elevation in the dorsal horn of pre-synaptic α₂δ-1, interpreted as being due to inhibition of trafficking⁶². Thus, gabapentin is likely to influence the function of the other proteins to which these α₂δ proteins have now been found to bind.

For the relevant Ca_V channels, we have also extensively examined the effects of gabapentin. They were initially found to have only small effects on calcium currents when applied acutely⁸². We found that longer-term incubation of cultured cells with gabapentin produced a clear reduction of calcium currents, both in transfected cells, when α₂δ-1 or α₂δ-2 was co-expressed, and in DRG neurons^45,81,83. We also observed a corresponding reduction in the expression of Ca_V2 α1 subunits on the cell surface^5,45.

Trafficking of α₂δ-1 by the multifunctional transport protein LRP1

The LRP family represents a large group of ligand-binding and trafficking proteins, including the LDL receptor and LRP1–6. They are multifunctional, multi-domain receptors, interacting with many protein ligands via their ligand-binding domains, mediating both forward trafficking and endocytosis of these ligands⁸⁵. They are also involved as co-receptors, affecting intracellular cell signaling processes^86,87.

LRP1 is a ubiquitous membrane protein with four ligand-binding domains (Figure 4a) and is involved in forward trafficking of proteins, including several TSPs^88–92, PrP⁹³, and NMDA receptors⁹⁴. LRP1 is also involved in clathrin-dependent endocytosis^85,95. It is present in synapses⁹⁴ and is implicated in neurite outgrowth⁹⁶. Whether different LRP proteins bind to overlapping sets of protein ligands is unclear, but LRP5/6 are also involved in Wnt signaling⁸⁷.

Figure 4. Protein domains involved in novel α₂δ interactions.

(a) Interaction of α₂δ-1 (and α₂δ-2/3) with the ligand-binding repeats II and IV of low-density lipoprotein receptor-related protein 1 (LRP1) (red). Other domains in LRP1 are epithelial growth factor (EGF)-like repeats (orange) and β-propeller domains (cyan)²⁷. i/c, intracellular; TM, transmembrane. (b) Interaction of neurexin 1α with α₂δ-3, via its laminin-like globular (LG) repeats (L, green) 1 and 5. E, EGF-like repeat (orange). Neurexin 1α is cleaved by a disintegrin and metalloprotease 10 (ADAM 10) (arrow) to have the observed effects on synaptic transmission, but it is not clear whether this is required for the interaction with α₂δ-3¹¹. (c) Interaction of the extracellular N-terminus of large conductance (big) potassium (BK) α subunits with α₂δ-1. The three blue arrows indicate the three alternative N-terminal translation initiation sites, the third being the most commonly used¹³. S0 is the additional transmembrane domain (red). (d) Interaction of α₂δ-1 von Willebrand factor A (VWA) domain with the EGF-like domains (black bars) of both pentameric (left) and trimeric (right) thrombospondins (TSPs)¹⁴. (e) Interaction of a C-terminal region of α₂δ-1 beyond its GPI-anchor site (dashed orange/white region) with the N-methyl-d-aspartate (NMDA) receptor GluN1, GluN2A, and GluN2B subunits¹⁵.

We recently showed that LRP1 binds to α₂δ-1²⁷ and the same is true for α₂δ-2 and α₂δ-3 (Ivan Kadurin and Annette Dolphin, preliminary results). For α₂δ-1, we showed this interaction is direct, involving the VWA domain of α₂δ-1 and LRP1 ligand-binding domains II and IV (Figure 4a)²⁷. The association is modulated by the LRP chaperone, receptor-associated protein (RAP), which is required for the correct folding of all LRP proteins and for their trafficking out of the endoplasmic reticulum^97,98. We found that the LRP1/RAP combination increases mature glycosylation, proteolytic processing, and cell-surface expression of α₂δ-1 and also increases plasma membrane expression and function of Ca_V2.2 when co-expressed with α₂δ-1²⁷. Since LRP1 is able to bind more than one ligand at different sites⁹⁹, it is possible that it forms a bridge between α₂δ-1 and other proteins, such as TSPs.

Sequestration of α₂δ-3 by interaction with α-neurexins

There are three vertebrate neurexin genes, and each can form α- and β-neurexins from different promoters. The α-neurexins have been found to be important for coupling calcium channels to synaptic transmission¹⁰⁰. Whereas in mammalian synapses the neurexins are pre-synaptic and bind to post-synaptic neuroligins, in Caenorhabditis elegans this polarity is reversed at many synapses. It has been found in the worm that post-synaptic neurexin 1α at the neuromuscular junction binds, via its laminin-like globular 1 (LG1) domain, to pre-synaptic unc-36 (similar to α₂δ-3), thus decreasing its availability to bind to the pre-synaptic unc-2 (a Ca_V2-like channel) that mediates neurotransmitter release¹¹. This was found to reduce synaptic transmission, an effect which required a proteolytically cleaved fragment of neurexin, shed from the post-synaptic plasma membrane (Figure 4b). In transfected cells, mouse neurexin 1α was found to bind α₂δ-3 and to decrease Ca_V2.2 current, whereas there was no effect on Ca_V2.2 currents in the presence of α₂δ-1 or α₂δ-2¹¹. An attractive suggestion is that this type of pre- to post-synaptic interaction may contribute to trans-synaptic nanoscale organization¹⁰¹. However, in view of recent results described below, it will be important in the future to identify the site of selective interaction on the α₂δ-3 protein of the LG1 domain (and LG5 in the mouse)¹¹ of neurexin 1α.

In contrast, a more recent article has identified positive effects of neurexin 1α in the presence of α₂δ-1 (but not α₂δ-3) on pre-synaptic Ca²⁺ transients in hippocampal neurons and in parallel on Ca_V2.1 calcium currents¹². Importantly, very carefully done experiments, designed to detect an interaction of neurexin 1α with α₂δ-1 or α₂δ-3, failed to find a specific association between the two proteins, as every protein tested (α-neurexin, neuroligin, and two forms of cadherin) was pulled down with α₂δ-1 (and also α₂δ-3 co-immunoprecipitated with neurexin 1α). The authors concluded that neurexin 1α does not form stable complexes with α₂δ subunits but nevertheless influences their function. Their results also provide a warning that α₂δ proteins may be rather prone to co-immunoprecipitation artefacts.

Sequestration of α₂δ-1 by interaction with BK channels

A recent study has identified that BK α subunits bind to α₂δ-1 subunits via the BK N-terminus¹³, and the authors suggest that this interaction sequesters α₂δ-1 from Ca_V channels. BK channels are important mediators of cell excitability, as they respond to both voltage and intracellular Ca²⁺ (for recent reviews, see 102,103). They consist of a tetrameric pore-forming α subunit, which is unusual compared with other voltage-gated K channels in that it has an additional transmembrane domain (S0), such that the N-terminus is extracellular. Furthermore, the N-terminus of BK α subunits contains an unusual sequence with three translation initiation methionines (M1, 25, and 66 in the human sequence below):

M¹AN³GGGGGGGSSGGGGGGGGSSLRM²⁵SSNIHANHLSLDASSSSSSSSSSSSSSSSSSSSSSVHEPKM⁶⁶DALIIPVTMEVPCDSRGQRM⁸⁶WWAFLASSMVTFFGGLFIILLWRTLKYLWTVCCHCGGKTK….

The third start methionine (M⁶⁶DAL) has generally been thought to be the main translation initiation site¹⁰⁴, and the underlined sequence was identified as a novel transmembrane segment S0. There is very good evidence that the existence of this additional transmembrane domain results in an extracellular N-terminus¹⁰⁴, although the exact mechanism driving this is unknown, as no signal peptide has been identified. In native rat brain, some mass spectrometry–mass spectrometry (MS-MS) peptide coverage of BK α was also seen from both the first (M¹ANG)¹⁰⁵ and the second (M²⁵SSN)¹⁰⁶ start methionines, indicating that they can also be used. BK channels are modulated by transmembrane β subunits which differentially interact with the different N-terminal isoforms of the BK α subunit and strongly affect BK voltage-dependent properties^107–109. BK channels also interact with γ subunits¹¹⁰.

In the study by Zhang et al.¹³, α₂δ-1 was found to associate with BK α subunits via their N-terminus (Figure 4c). This association was found to compete with both Ca_V1 and Ca_V2 channels for α₂δ-1 and therefore reduce the Ca_V channel function. Interestingly, the region of BK channels identified by pull-down experiments to interact with α₂δ-1 is within the N-terminal residues 1–86, which contain two unusual repetitive polyglycine and polyserine stretches (see above). If the sequence encoded from the first start methionine (residues 1–24) was truncated or if the asparagine (N) at position 3 was mutated to D, no effect of the BK channel on Ca_Vα1/β/α₂δ-1 currents was observed, whereas the in vitro binding also involved residues 66–86¹³. These results suggest that the effect of BK channels on Ca_V channel function would occur only for the full-length BK isoform, starting with MANG. It is also of interest that N3 in the BK channel potentially undergoes rapid deamidation in vivo which would abolish its interaction with α₂δ-1 in a time-dependent manner¹³, meaning that only a small subset of BK channels might be involved in this interaction with α₂δ-1. Moreover, in this study, no BK β or γ subunits were expressed and therefore it would be important to determine whether their interaction with the N-terminus or elsewhere would compete with α₂δ for interaction, which would represent an interesting means of reciprocal cross-talk between these channels.

Because the authors examine the potential role for this BK–α₂δ-1 interaction for neuropathic pain, in which α₂δ-1 is upregulated, it would also be of great interest to identify the relative expression from the different translation initiation sites used for the BK α protein in DRG neurons in control and neuropathic states. Furthermore, it should be noted that, in contrast to α₂δ-1 which is upregulated, BK channel mRNA is downregulated in DRGs following neuropathic nerve injury¹¹¹.

Surprisingly, in proteomic studies of native rat brain BK channels, α₂δ was not identified as co-purifying with these channels, although several Ca_V channel α1 subunits were well represented¹⁰⁶. Ca_V1.2, Ca_V2.1, and Ca_V2.2 as well as the Ca_Vβ subunits β1b, β2, and β3 were all found in this study¹⁰⁶. Indeed, Ca_V2.1 was the most abundantly represented protein that co-purified with BK channels, suggesting the possibility of a direct interaction. This finding would seem to contradict the model of Zhang et al.¹³, in which BK competes for α₂δ with the Ca_Vα1 subunit.

Sequestration of α₂δ-1 by interaction with a disease-associated mutant PrP

In an intriguing study, PrP was found to interact with α₂δ-1 proteins, and a Creutzfeldt–Jakob disease-causing mutant form of PrP resulted in intracellular retention of α₂δ-1 and disrupted synaptic transmission⁸⁴. It is of relevance in this regard that both PrP and α₂δ-1 are GPI-anchored and therefore would be likely to be in similar membrane domains. One confounding issue is that in overexpression studies, α₂δ-1 and PrP interfere with each other’s trafficking, at least partly because of competition for the limiting supply of GPI anchor²⁵. In this study²⁵, PrP disrupted the ability of α₂δ-1 to increase calcium currents, but a C-terminally truncated GPI-anchorless PrP did not²⁵. Thus, it remains unclear to what extent the α₂δ-1 interaction with cellular PrP has a physiological or pathophysiological role¹¹².

α₂δ-1 as a mediator of synaptogenesis via binding to TSPs

TSPs are extracellular matrix proteins which bind to a very large number of proteins, 83 being so far identified for TSP-1¹¹³; consequently, they have many functions^114–116. In the brain, they are produced by astrocytes and promote neurite outgrowth¹¹⁷, including the formation of silent excitatory synapses, lacking post-synaptic receptors¹¹⁸. It was then hypothesized that post-synaptic α₂δ-1 could be the sought-after post-synaptic binding partner of TSPs to mediate synaptogenesis, independent of any effects on calcium channels. This was first tested using co-immunoprecipitation to determine whether TSPs or individual domains of TSPs interacted with C-terminally tagged α₂δ-1¹⁴. An interaction which involved a key synaptogenic epithelial growth factor (EGF)-like domain was found (Figure 4d). As a note of caution, C-terminal tagging may interfere with trafficking of α₂δ-1 by disrupting the GPI anchor^24,26. Nevertheless, gabapentin was found to inhibit the interaction between α₂δ-1 and the EGF-like domain of TSP-2 and to disrupt synaptogenesis. Furthermore, in vivo, gabapentin was found to disrupt whisker barrel plasticity following whisker removal in some of the mice examined¹⁴.

TSP-4 is upregulated in rodent models of neuropathic pain¹¹⁹. Since α₂δ-1 is also upregulated in DRGs following peripheral sensory nerve injury, several studies have investigated whether an interaction between these two proteins is important in neuropathic pain or the effect of gabapentin. Interestingly, in a recent article, it was suggested that pre-synaptic, rather than post-synaptic, α₂δ-1 may be a synaptogenic binding partner for TSP-4 in the spinal cord¹²⁰.

We found (using overexpressed proteins) that TSP-4 modestly reduced the affinity for ³H-gabapentin binding to α₂δ-1, although the effect on ³H-gabapentin binding was not reproduced with the TSP-4 synaptogenic EGF-like domain. Furthermore, we found only very weak and unreliable co-immunoprecipitation of the two proteins, which again could not be reproduced with the synaptogenic EGF-like domain of TSP-4¹²¹. We also could not demonstrate any interaction between α₂δ-1 and TSP-4 on the cell surface of transfected cells, suggesting that the association between these two proteins to disrupt 3H-gabapentin binding is occurring intracellularly following co-transfection, when the two proteins are juxtaposed at high concentration¹²¹.

Nevertheless, there is evidence from other studies that α₂δ subunits are important for synaptic morphology in several different systems^{57,58,122,123}. Whether the role for α₂δ in calcium channel localization and function is responsible for these morphological changes has not always been investigated. However, α₂δ was shown to increase pre-synaptic localization of the relevant α1 subunit in Drosophila neuromuscular junction synapses¹²⁴ as well as in retinal⁵⁸ and hippocampal⁶ synapses.

α₂δ-1 as an NMDA receptor trafficking protein

It was recently shown that overexpression of α₂δ-1 administered intrathecally into the spinal cord potentiates pre-synaptic and post-synaptic NMDA receptor activity, and it was further shown that α₂δ-1 interacted with NMDA receptors, both in spinal cord and in overexpression studies¹⁵. The interaction was apparently specific for α₂δ-1, as it did not occur with α₂δ-2 or α₂δ-3. The authors identified the site of interaction as the C-terminus of α₂δ-1, surprisingly after the C-terminal GPI-anchor cleavage site (Figure 4e). This was determined using chimeras assembled from the different isoforms, swapping isoforms either between α2 and δ or with the C-terminus of δ⁶. However, it is important to note that such chimeras may have disrupted the primary sequences involved in proteolytic cleavage between α₂ and δ, a process which is important for function^6,28, or it might have affected the sequences involved in GPI anchoring²⁴. Nevertheless, this result suggests either that a transmembrane version of α₂δ-1 may be interacting with NMDA receptors, initially in the endoplasmic reticulum, or that the NMDA receptor interacts with the C-terminal peptide of α₂δ-1 that is cleaved off during GPI-anchor attachment¹²⁵.

The GluN1, GluN2A, and GluN2B subunits of NMDA receptors were found to interact with α₂δ-1, presumably via the transmembrane or intracellular domains of these subunits, since the identified interaction is with the C-terminus of α₂δ-1¹⁵. The C-termini of these NMDA receptors are rather different in both sequence and function^126–128, and determining the interaction site will be a key next step. It is of interest that α₂δ-1 has not been previously detected in proteomic studies of post-synaptic densities¹²⁹. In contrast, other calcium channel subunits (Ca_V1.2, Ca_V2.3, and a β) were identified. Another recent study also did not detect α₂δ-1 when purifying NMDA receptors from mouse brain¹²⁸, although α₂δ-1 is widely expressed in most brain regions^130,131. Therefore, it would be important to determine whether this interaction is for some reason observed only in the spinal cord. One possible reason is that it might be indirect (for example, via a scaffolding protein expressed in the spinal cord, interacting with both α₂δ-1 and NMDA receptors).