Emerging evidence for specific neuronal functions of auxiliary calcium channel α₂δ subunits

Abstract

In nerve cells the ubiquitous second messenger calcium regulates a variety of vitally important functions including neurotransmitter release, gene regulation, and neuronal plasticity. The entry of calcium into cells is tightly regulated by voltage-gated calcium channels, which consist of a heteromultimeric complex of a pore forming α₁, and the auxiliary β and α₂δ subunits. Four genes (Cacna2d1-4) encode for the extracellular membrane-attached α₂δ subunits (α₂δ-1 to α₂δ-4), out of which three isoforms (α₂δ-1 to -3) are strongly expressed in the central nervous system. Over the years a wealth of studies has demonstrated the classical role of α₂δ subunits in channel trafficking and calcium current modulation. Recent studies in specialized neuronal cell systems propose roles of α₂δ subunits beyond the classical view and implicate α₂δ subunits as important regulators of synapse formation. These findings are supported by the identification of novel human disease mutations associated with α₂δ subunits and by the fact that α₂δ subunits are the target of the anti-epileptic and anti-allodynic drugs gabapentin and pregabalin. Here we review the recently emerging evidence for specific as well as redundant neuronal roles of α₂δ subunits and discuss the mechanisms for establishing and maintaining specificity.

1. Introduction

In the central nervous system (CNS) calcium entering nerve cells through voltage-gated calcium channels (Ca_Vs) mediates and regulates a variety of neuronal functions ranging from neurotransmitter secretion and postsynaptic signal integration to gene regulation and neuronal plasticity. For example, presynaptic Ca_Vs trigger neurotransmitter release (Stanley, 1993) and postsynaptic Ca_Vs are involved in the transcriptional regulation of CREB (cAMP-responsive element-binding protein) and NFAT (nuclear factor of activated T cells) (Deisseroth et al. 2003; Dolmetsch 2003) and thus modulate the excitability and likely play a crucial part in the formation of new memory (Moosmang et al. 2005). Over the recent years a detailed picture on the distribution and function of pre- and postsynaptic calcium channel types has begun to emerge (Obermair and Flucher 2013) and the importance of Ca_Vs is emphasized by the existence of channelopathies caused by loss-of-function as well as gain-of-function mutations (Pietrobon 2010; Striessnig et al. 2010). Calcium channels are organized in heteromultimeric complexes consisting of the pore-forming α₁ subunit and the auxiliary β and α₂δ subunits. The α₁ subunit defines the basic biophysical, pharmacological and physiological properties of the channels while the β and α₂δ subunits are involved in the localization, trafficking and stabilization of the channel complex (reviewed in Arikkath and Campbell 2003; Obermair et al. 2008; Dolphin 2009; Buraei and Yang 2010). The majority of today’s knowledge on the role of the individual channel subunits is based on studies in heterologous expression systems such as Xenopus laevis oocytes or human embryonic kidney (HEK) cells. However, the recent development of powerful neuronal expression systems and the characterization of Ca_V knockout and mutant animal models have provided novel insights into the physiological importance of auxiliary β and α₂δ subunits in neuronal functions such as the regulation of gene expression as well as synaptic function. While these studies provide evidence for specific roles of individual Ca_V subunits, they also indicate that some of their basic functions may be rather common and redundantly shared between subunit isoforms. Therefore the challenge of ongoing and future investigations is to elucidate whether and how auxiliary Ca_V subunit isoforms and their respective splice variants contribute to the various neuronal functions. This is not merely relevant for our understanding of neuronal signaling, but may open up new therapeutic strategies for treating CNS diseases involving Ca_Vs such as neurodegenerative diseases, epilepsy, or anxiety and mood disorders. Here we review recent evidence for both specific and redundant functions of auxiliary Ca_V subunits with a focus on the extracellular α₂δ subunits.

2. Neuronal calcium channel complexes

2.1. Subunit composition

The core complex of neuronal voltage-gated calcium channels consists of an α₁ subunit, which contains the ion-conducting pore, and the auxiliary β and α₂δ subunits (Fig. 1A). In skeletal muscle also a γ subunit is part of the complex; however, in the nervous system various γ subunits act primarily as transmembrane AMPA receptor regulatory proteins (TARPs; Jackson and Nicoll 2011).

Figure 1.

A. The voltage-gated calcium channel complex. Voltage-gated calcium channels are composed of a transmembrane pore-forming α₁ subunit (red) which forms a macromolecular complex together with extracellular α₂δ (blue) and cytoplasmic β (yellow) subunits. The β subunit binds the intracellular I-II linker of α₁ with high affinity. The α₂δ subunit consists of posttranslationally cleaved highly glycosylated α₂ and δ peptides, which are associated to each other by a disulfide bond and are most likely linked to the plasma membrane (indicated in grey) via a GPI-anchor. Note that the folding structure of the individual subunits is simplified. B. Domain structure of auxiliary α₂δ subunits. Amino acid positions refer to the mouse α₂δ-1 isoform [uniprot: O08532- CA2D1_MOUSE], however all isoforms have a similar topology. Amino acid residues 1—24 encode the signal peptide (SP). The α₂ peptide contains a von Willebrand factor type A (VWA) domain and two sequence stretches homologous to extracellular domains of bacterial chemosensing proteins (Cache I and II). In α₂δ-1 and α₂δ-2 an arginine (RRR) motif proximal to the VWA domain represents the potential gabapentin (GBP) and pregabalin (PG) binding site. Two cysteine residues (AA 404 and 1059) were identified to be important for the formation of an intermolecular disulfide bond between α₂ and δ. The δ peptide contains predicted ω amino acids, to which the GPI anchor can attach, and a C-terminal hydrophobic sequence (c-HS). Potential glycosylation sites are indicated by asterisks (*).

The α₁ subunit:

The α₁ subunits of voltage-gated calcium channels are a family of 10 genes involving the high-voltage-activated classes Ca_V1 and Ca_V2 (α_1A to α_1F and α_1S), and the low-voltage-activated class Ca_V3 (α_1G to α_1I) (Catterall 2011). Each α₁ subunit consists of four homologous repeats (I to IV), each with six membrane spanning domains (S1 to S6). Specific functions have been assigned to particular structures within this sequence, like the voltage sensor in the four S4 segments and the ion selectivity filter of the conductance pore in the loops between S5 and S6 of each repeat. Depolarization of the neuronal plasma membrane leads to a conformational change within the α₁ subunit that gates the channel and activates the calcium current. Domains involved in protein-protein interaction with the accessory channel subunits, regulatory proteins, and other channels have been localized in the large intracellular loops connecting the homologous repeats and in the long C-terminal tail. The α₁ subunit also defines the basic biophysical and pharmacological properties due to important drug binding sites such as dihydropyridines, phenylalkylamines, as well as benzothiazepines (Catterall 2011). α₁ subunit splicing increases the functional heterogeneity and the spectrum of pharmacological characteristics (Koschak 2010; Flucher and Tuluc 2011). Out of the four members of the dihydropyridine-sensitive L-type (Ca_V1) calcium channels, Ca_V1.2 and Ca_V1.3 are expressed in the CNS and play important roles in pace-making (Ca_V1.3), postsynaptic signal integration, and excitation-transcription coupling in neurons. L-type calcium channels have mainly been associated with postsynaptic functions with two notable exceptions. Ca_V1.3 and Ca_V1.4 regulate presynaptic glutamate release in two highly specialized synapses, namely auditory hair cells and retinal photoreceptor cells, respectively. The P/Q-, N-, and R-type calcium channels (Ca_V2.1, 2.2, and 2.3) can be specifically blocked with a number of invertebrate toxins (Catterall 2011), and their chief function is controlling neurotransmitter release in synapses (Nimmervoll et al. 2013). The members of the low-voltage activated, T-type calcium channels (Ca_V3 family) are highly expressed during early development of many cell types, can be blocked by Ni²⁺, and contribute to the excitability and pace-making in neurons (reviewed in Perez-Reyes 2003).

The β subunit:

The cytoplasmic β subunit consists of a conserved SH3 protein interaction domain and a nucleotide kinase-like domain (Chen et al. 2004; Opatowsky et al. 2004; Van Petegem et al. 2004) and thus resembles in structure the membrane-associated guanylate kinase proteins (Dolphin 2003; Takahashi et al. 2005). However, the SH3 domain of β subunits differs from that of canonical polyprolin-binding pockets and the guanylate kinase fold is modified so that it lacks kinase activity. Instead it binds the intracellular I-II linker of α₁ subunits at the so-called α-interaction-domain (AID) with nanomolar affinity (De Waard et al. 1995; Van Petegem et al. 2008). The SH3 and the GK-like domain are highly conserved among the four genes encoding β subunits (Cacnb1-b4). The sequences connecting these domains as well as the N- and C-termini vary between isoforms and are subject to alternative splicing (Colecraft et al. 2002; Dolphin 2003). In the channel complex β subunits serve two roles: They have a chaperon function regulating the export of the calcium channel from the endoplasmic reticulum and thus membrane expression of functional channels (Obermair et al. 2010; Fang and Colecraft 2011). Moreover, they modulate gating properties of the channel directly as well as by interaction with other regulatory proteins like Rab binding proteins or G-proteins. β itself is subject to PKA mediated phosphorylation (reviewed in Buraei and Yang 2010). The β_2a isoform is palmitoylated at two N-terminal cysteines and therefore membrane-associated even in the absence of an α₁ subunit. Nevertheless, the association of β subunits with the channel complex entirely depends on their binding to the AID in the α₁ subunit. This binding site in the cytoplasmic loop between repeats I and II of the α₁ subunit is a unique feature of the Ca_V1 and Ca_V2 subclasses of Ca_Vs. Accordingly, at least in heterologous expression systems all β subunits can associate with any of the Ca_V1 or Ca_V2 members but not with the low-voltage-activated calcium channels of the Ca_V3 subclass (Dolphin 2003). Because of their central role in regulating functional expression and biophysical properties of calcium channels, and because of the well-defined interaction site (see above), interference with the AID-β interaction is an attractive strategy for designing specific calcium channel antagonists. Indeed, a competing AID peptide induced functional uncoupling of Ca_V1.2 and β_2a subunits (Hohaus et al. 2000). Nevertheless, until today no isoform-specific inhibitors could be validated.

The α₂δ subunit:

Four genes (Cacna2d1-4) encode for α₂δ subunits (α₂δ-1 to α₂δ-4). They display distinct tissue distribution and three isoforms (α₂δ-1 to -3) are strongly expressed in the developing and mature CNS (Arikkath and Campbell 2003; Schlick et al. 2010). Mature α₂δ subunits consist of posttranslationally cleaved highly glycosylated α₂ and δ peptides (Fig. 1B), which are associated to each other by a disulfide bond (Calderon-Rivera et al. 2012). Glycosylation of α₂δ subunits is required for the functional membrane expression of calcium channels, as deglycosylation and glycosylation-site-directed mutagenesis resulted in strongly reduced current densities without affecting the kinetic properties (Gurnett et al. 1996; Sandoval et al. 2004). α₂δ subunits are most likely linked to the plasma membrane via GPI-anchors (Davies et al. 2010). Alternatively, the δ subunit may constitute a single-pass membrane protein (Robinson et al. 2011). The vast majority of the α₂δ protein is extracellular, ideally situated to interact with constituents of the extracellular matrix or extracellularly exposed proteins. In the domain structure of α₂ a von Willebrand factor type A (VWA) domain and two Cache domains were identified by sequence homology in all α₂δ subunits (Fig. 1B; Anantharaman and Aravind 2000; Canti et al. 2005; Davies et al. 2007). VWA-domains are found in a variety of extracellular matrix proteins and integrin receptors and are well known for their role in cell-cell adhesion involving a metal ion-dependent adhesion site (MIDAS). The integrity of the MIDAS motif in α₂δ-2 has been shown to be necessary for calcium current enhancement and Ca_V channel trafficking (Canti et al. 2005). It has been hypothesized that these domains may be regulated by small endogenous ligands, such as the amino acid isoleucine (reviewed in Dooley et al. 2007), and that they are involved in binding the anti-epileptic and anti-allodynic drugs gabapentin (GBP) and pregabalin (PG) (Davies et al. 2007), which have also proven clinical efficacy in the treatment of generalized anxiety disorders.

2.2. Basic principles of neuronal functions mediated by voltage-gated calcium channels

The strong buffering of the second messenger calcium requires a close temporal and spatial vicinity of effector proteins to the calcium channel complex. Accordingly, neuronal Ca_Vs are closely associated with upstream modulators, like protein kinases and phosphatases, and downstream effectors, as well as adapter and scaffold proteins. In neurons two highly specialized complexes have been intensively characterized: the synaptic vesicle fusion apparatus mediating excitation-secretion coupling and the postsynaptic calcium channel complex mediating excitation-transcription coupling (Fig. 2). At the synaptic vesicle fusion apparatus presynaptic calcium channels (mainly Ca_V2.1 and Ca_V2.2) are located in the close vicinity of the calcium sensor synaptotagmin, which triggers vesicle fusion and thereby neurotransmitter release. In the postsynaptic compartment activation of L-type channels (Ca_V1.2 and Ca_V1.3) initiates a signaling cascade to the nucleus that regulates gene expression. In both specialized compartments calcium channels need to be tightly organized in macromolecular complexes with upstream and downstream signaling molecules (Fig. 2B and C). The molecular organization of these signaling complexes is expected to be influenced by the subunit composition of the Ca_V complex in several ways. First, modulation of current properties (activation and inactivation kinetics, voltage-dependence) by auxiliary subunits will affect the local calcium transient and thereby downstream signaling. Second, a role of auxiliary Ca_V subunits in channel trafficking, targeting, and scaffolding will affect the composition of the entire complex and thereby determine specificity. Because different subunit isoforms and splice variants may differ with respect to their modulatory properties, protein-protein interactions, and subcellular targeting, the diversity of the auxiliary subunits may determine specific cellular functions in neurons. Therefore, in order to understand the role of the various distinct auxiliary Ca_V subunits in neurons it is necessary to address 1) which subunit isoforms and splice variants are actual components of specific signaling complexes, 2) whether different isoforms and splice variants serve distinct functions, 3) whether and to what degree can specific functions be compensated by other isoforms, and finally 4) do auxiliary subunits also act in a molecular context which is independent of the calcium channel complex.

Figure 2.

Neuronal calcium channel complexes. A. In neurons voltage-gated calcium channels are located in presynaptic boutons, where they trigger neurotransmitter release, and in postsynaptic as well as extrasynaptic positions along dendrites and dendritic spines (Obermair et al. 2004; Jenkins et al. 2010). B. In the presynaptic active zone calcium channels are organized in a dense network of presynaptic proteins (depicted are RIM, RBP, Rab, bassoon, SNARE complex) and G-protein coupled receptors (GPCR) in the vicinity of the calcium sensor synaptotagmin (stg). Together this macromolecular complex orchestrates the tight regulation of synaptic vesicle (SV) fusion and thus neurotransmitter release (reviewed in Südhof 2012). C. In the postsynaptic/somato-dendritic compartment L-type channels can be found in complexes with G-protein coupled receptors (GPCR, e.g. the β₂ adrenergic receptor), an adenylyl cyclase (AC), protein kinases (PKA) and phosphatases (PP2B), and scaffolding proteins (e.g. AKAP150) (reviewed in Dai et al. 2009). Downstream signaling initiates a signaling cascade which leads to the modulation of transcription such as CREB or NFATc4. Lateral mobility of L-type calcium channels in the plasma membrane also suggests regulation via the association or dissociation with distinct signaling complexes (Di Biase et al. 2011). Considering the membrane diffusion of GPI-anchored proteins, such a mechanism may also apply to α₂δ subunits. ic, intracellular; ec, extracellular.

2.3. Establishing and maintaining signaling specificity

In order to elucidate the physiological neuronal functions of calcium channels and their auxiliary subunits, understanding the specificity of Ca_V subunit interactions and complex formation in native differentiated cell systems is a prerequisite. Three distinct mechanisms may contribute to the specificity of neuronal Ca_V complexes (Obermair and Flucher 2013). First, the calcium subunit complement in a specific cell type at a given time will yield the formation of particular complexes. Such preferential expression patterns of α₁, β, and α₂δ isoforms can indeed be observed in skeletal and cardiac muscle as well as highly specialized neuronal cell types like retina photoreceptor cells (Knoflach et al. 2013). In the CNS various brain regions including cortex, hippocampus, cerebellum, and even single cell types like cultured hippocampal pyramidal cells simultaneously express physiologically relevant mRNA levels of five out of seven high-voltage activated α₁ subunits, all four β subunit isoforms, and three of four α₂δ subunits (Schlick et al. 2010). The cerebellum, however, represents one notable exception as it preferentially expresses one set of calcium channel subunits (Ca_V2.1/β₄/α₂δ-2) (Ludwig et al. 1997; Brodbeck et al. 2002; Schlick et al. 2010). Altogether, in the majority of brain regions a restricted expression of auxiliary subunit isoforms may not be the prime strategy for establishing specific Ca_V complexes.

Second, distinct subcellular targeting properties of individual subunits or splice variants and/or possible interactions with other proteins may also yield specificity. We have recently identified such a mechanism for splice variants of the auxiliary β₄ subunits. Whereas all β₄ splice variants could increase the expression of presynaptic Ca_V2.1 by directly interacting with the channel, only β₄ splice variants targeted into the nucleus (β_4b, β_4a) did specifically repress expression of synaptic proteins including Ca_V2.1 in a channel-independent manner (Etemad et al. 2014a, 2014b). Recently the differential up-regulation of an α₂δ-1 splice variant in DRG neurons after spinal nerve ligation was identified (Lana et al. 2014). This finding supports a functional heterogeneity of splice variants also with α₂δ subunits.

Finally it is also possible that specific stable complexes may not exist in all neuronal compartments and calcium channels could be regulated by reversible interactions with pools of functionally diverse β or α₂δ subunits. An elegant FRAP study in skeletal muscle cells recently confirmed the presence of a stable calcium channel complex between the homologous Ca_V1.1 and β_1a isoforms, while the heterologous β_2a and β_4b subunits formed dynamic complexes with the channel (Campiglio et al. 2013). The existence of a mobile pool of the L-type channel Ca_V1.2 in hippocampal neurons may be a first hint towards regulation of neuronal Ca_Vs by dynamic subunit interactions (Di Biase et al. 2011).

3. Neuronal roles of auxiliary α₂δ subunits

3.1. Channel trafficking and current modulation

The classical roles of auxiliary α₂δ subunits on voltage-gated calcium channels, namely regulating the functional membrane expression and modulating the calcium currents, are widely recognized (Fig. 3A and B). When heterologously expressed all α₂δ isoforms can enhance the functional membrane expression of α₁ subunits (reviewed in Arikkath and Campbell 2003; Davies et al. 2007; Obermair et al. 2008). α₂δ-1 co-expression studies in tsA201 cells additionally identified a role in modulating the biophysical channel properties such as activation and inactivation kinetics as well as the voltage-dependence of activation (Felix et al. 1997). Similar to α₂δ-1 also α₂δ-3 increased the functional membrane expression and altered the biophysical properties of L-type calcium channels (Klugbauer et al. 1999). In contrast, expression of α₂δ-2 in Xenopus oocytes did not reveal biophysical modulation of Ca_V1.2 and Ca_V2.2 channels, although it was also found to increase the channels’ current density (Gao et al. 2000; Brodbeck et al. 2002). α₂δ-4 is the least well studied isoform among α₂δ subunits with respect to its modulatory effects on Ca_Vs. In HEK293 it has been shown to increase calcium influx via Ca_V1.2 channels (Qin et al. 2002); however, a detailed functional characterization is still missing. Altogether existing evidence indicates some isoform specificity in modulating the biophysical current parameters, while increasing the functional membrane expression of Ca_Vs seems to be a rather general mechanism. On the other hand isoform specificity in channel trafficking has been described in non-neuronal excitable cells. In skeletal muscle cells were only one isoform (α₂δ-1) is expressed, experiments utilizing an α₂δ-1 shRNA approach demonstrated that this subunit is not essential for targeting of calcium channels or for their primary physiological role in activating skeletal muscle excitation-contraction coupling (Obermair et al. 2005). However, α₂δ-1 was identified as the major determinant of the characteristic L-type calcium current kinetics (Obermair et al. 2005; Tuluc et al. 2007). Together this suggests that α₂δ-1 does not regulate the membrane trafficking and targeting of L-type calcium channels in their native environment. In consistence with this hypothesis α₂δ-1 had only a small trafficking effect on Ca_V1.2 in a model system for cardiac myocytes (Tuluc et al. 2007), whereas it strongly enhanced functional expression of neuronal N- (Ca_V2.2) and P/Q-type (Ca_V2.1) channels (Obermair et al. 2008). In neurons of the CNS, where three α₂δ isoforms (α₂δ-1, -2, -3) are expressed, co-expression of all three subunits with P/Q or N-type channels causes an increase in current density (Davies et al. 2007). Thus regulating membrane trafficking and thereby calcium channel abundance may be the primary mechanism for modulating specific neuronal functions, for example synaptic transmission, by α₂δ subunits. This is supported by reduced presynaptic Ca_V2.1 targeting as well as reduced synaptic release probability in neurons were α₂δ-1 was knocked down by shRNA (Hoppa et al. 2012). On the other hand, the largely mild CNS phenotypes of the individual knockout mice (see below) suggest compensatory actions of α₂δ isoforms. The propensity of individual α₂δ-isoforms to regulate channel targeting is one important aspect in controlling Ca_Vs, the other aspect is their capacity to fine tune the current kinetics of individual calcium channels. This was not only observed for skeletal and cardiac muscle Ca_Vs (see above). Constitutive overexpression of α₂δ-1 in neuronal tissues, for instance, resulted in enhanced currents as well as altered kinetics and voltage-dependence of activation in sensory neurons (Li et al. 2006).

Figure 3.

Sketch summarizing the neuronal functions of auxiliary α₂δ subunits. A. Channel trafficking: The export from the endoplasmic reticulum (ER) is mediated by β subunits; however, α₂δ subunits can further enhance the trafficking. This may involve trafficking from the endoplasmic reticulum (ER) via the Golgi apparatus in transport vesicles (TV), recycling from recycling endosomes (RE), or stabilization of the channel complex at the plasma membrane. α₂δ subunits can also be transported to the plasma membrane independent of the α₁ subunit. B. Current modulation: Besides their role in channel trafficking α₂δ subunits are important modulators of the calcium current. α₂δ subunits can shift the voltage-dependence of channel activation to more negative potentials (IV curve, middle graph) and can modify the activation and inactivation kinetics (current traces, lower graph). This may be of particular relevance for somato-dendritic L-type calcium channels. C. Synaptogenesis: Accumulating evidence suggests an important role of neuronal α₂δ subunits in synapse formation. Whether α₂δs are necessary for the initial contact between the axon and the corresponding postsynaptic membrane (top) or for the differentiation into a fully aligned and mature synapse (bottom) remains to be answered.

3.2. α₂δ subunits as targets for gabapentinoid drugs

Although the chemical backbone of GBP and PG was derived from GABA the mechanism of action of both drugs was initially unclear. The only outcome was that they did not have any notable effect on GABA receptors, metabolism, or transport (reviewed in Taylor et al. 2007; Silverman 2008; Bauer et al. 2010). However, both drugs were shown to reduce the cellular influx of calcium via Ca_Vs in synaptosome fractions and hippocampal synapses (Fink et al. 2002; van Hooft et al. 2002). Accordingly it was a surprising result as it turned out that the 3H-gabapentin binding site corresponded to α₂δ-1 and finally a selective binding to α₂δ-1 and α₂δ-2 was confirmed by the availability of newly established mouse models (see below). Both GBP and PG were found to be effective treatments for various forms of neuropathic pain and are also used as an adjunctive therapy of certain forms of epilepsy. The effect of these drugs in neuropathic pain was found to be state-dependent in that they have little effect on acute pain perception in animals or humans, but they are effective in chronic neuropathic pain. In line with this long term effects GBP inhibits calcium currents when applied chronically, both in heterologous expression systems and in dorsal root ganglion neurons (e.g. Hendrich et al. 2008). The most likely explanation for this effect is that chronic GBP treatment interferes with Ca_V trafficking to the cell surface as well as to presynaptic terminals (Bauer et al. 2009; Tran-Van-Minh and Dolphin 2010). Besides affecting trafficking during chronic application, these drugs have also been shown to exert acute effects on the current amplitude and the biophysical properties of calcium currents as well as on Ca_V-dependent functions, such as synaptic transmission (e.g. Uchitel et al. 2010; Farrell et al. 2014). However, acute effects of GBP and PG have not been observed in all studies and may depend on the composition of the calcium channel complex (Martin et al. 2002).

3.3. α₂δ subunits and synapse formation

Over the recent years accumulating evidence suggests an important role of α₂δ subunits in synapse formation (Fig. 3C). α₂δ-1 has been shown to act as a receptor for thrombospondin, an astrocyte-secreted protein that promotes CNS synaptogenesis (Eroglu et al. 2009). In this study overexpression of α₂δ-1 strongly promoted, and shRNA knockdown inhibited excitatory synapse formation in cultured retinal ganglion cells. Drosophila mutants of the α₂δ-3 homologue (straightjackt; stj) show defects in presynaptic calcium channel localization and synaptic function (Dickman et al. 2008), which also coincides with a failure in normal synapse development in motoneurons (Kurshan et al. 2009). Interestingly, this study provided evidence that the defect in synapse formation was independent from the Drosophila pore forming α₁ subunit (cacophony), because in contrast to the stj mutants, cacophony null mutants did not show a defect in synapse formation. α₂δ-2 mutant (ducky, ducky^2J) and knockout mice display altered Purkinje cell morphology and reduced calcium currents (Barclay et al. 2001; Ivanov et al. 2004; Donato et al. 2006), which suggests a role in maintaining normal cellular physiology and likely also the morphology of synapses. A spontaneous mouse mutation of α₂δ-4 (Cacna2d4) causes structural and functional abnormalities of retinal ribbon synapses associated with the loss of rods (Wycisk et al. 2006a). Unc-2 and unc-36, the C. elegans homologues of Ca_V2 and α₂δ, respectively, were identfied as regulators of C. elegans synaptogenesis (Caylor et al. 2013). While for unc-2 the study revealed a role in regulating the dynamic changes in the size and morphology of synapses that occur during development, unc-36 was demonstrated to regulate presynaptic morphology. α₂δ-1 upregulation in a nerve injury model for chronic pain induced abnormal excitatory synapse formation and enhanced neurotransmitter release (Li et al. 2014). Finally, α₂δ-3 knockout mice have a defect in presynaptic Ca_V2.1 channel targeting and synapse formation of auditory nerve fiber terminals contacting cochlear nucleus bushy cells (Pirone et al. 2014).Taken together, evidence is accumulating suggesting a role of α₂δ subunits in synapse formation, which may be in part or entirely independent of the calcium channel complex. The mechanistic background for an independent function of α₂δ subunits could be that these proteins protrude far into the extracellular space and therefore have the potential to contribute to signaling events that coordinate synaptic development.

3.4. Insights from animal models and human disease mutations

Within the past years, there has been great progress in the generation and description of mouse models with targeted deletions and spontaneous mutations for all α₂δ subunits. These knockout and mutant mice display various disease phenotypes, a part of which share similarities with recently identified human diseases linked to α₂δ subunits. The involvement of α₂δ subunits in disease gives important insight into possible functions of the distinct isoforms (summarized in Tab. 1).

Table 1.

α₂δ subunit isoforms and mouse and human disease phenotypes

	Mouse		Human
Isoform	Mouse models	Phenotype	Potential cause	Disease
α2δ-1	knock-in1 (GBP insensitive)	no known CNS phenotype1	aberrations of CACNA2D15	epilepsy and intellectual disability5
	knockout2	reduced pain sensitivity3, no known CNS phenotype1
	α2δ-1 over-expressing4	hyperalgesia, tactile allodynia4
α2δ-2	spontaneous loss-of-function mutations6,7,8,9 (ducky, ducky2j, entla), knockout10	ataxia, epilepsy and paroxysmal dyskinesia6,7,8,9,10	homozygous mutations (point11 and frameshift12)	epileptic encephalopathy11,12
			genomic deletion of CACNA2D213,14	potential tumor suppressor gene13,14
			single-nucleotide polymorphisms (SNPs)15	candidate gene in childhood absence epilepsy15
α2δ-3	knockout16	deficits in pain16 and auditory/acoustic startle processing17	SNPs16	reduced sensitivity to acute noxious heat and chronic back pain16
			splice site mutation in CACNA2D318	autism spectrum disorders18
α2δ-4	spontaneous mutation (premature stop)19	structural and functional abnormalities in ribbon synapses, loss of rods19	mutation (premature stop) of CACNA2D420	slowly progressive cone dystrophy and night blindness20
			partial deletion of CACNA2D421	psychiatric disorders21

¹Field et al. 2006,

²Fuller-Bicer et al. 2009,

³Patel et al. 2013,

⁴Li et al. 2006,

⁵Vergult et al. 2014,

⁶Barclay et al. 2001,

⁷Brodbeck et al. 2002,

⁸Brill et al. 2004,

⁹Donato et al. 2006,

¹⁰Ivanov et al. 2004,

¹¹Edvardson et al. 2013,

¹²Pippucci et al. 2013,

¹³Carboni et al. 2003,

¹⁴Hesson et al. 2007,

¹⁵Chioza et al. 2009,

¹⁶Neely et al. 2010,

¹⁷Pirone et al. 2014,

¹⁸Iossifov et al. 2012,

¹⁹Wycisk et al. 2006a,

²⁰Wycisk et al. 2006b,

²¹Van Den Bossche et al. 2012

α₂δ-1:

As α₂δ-1 (and α₂δ-2) were identified targets for GBP and PG, the need for mouse models was soon recognized for studying the involvement of the distinct α₂δs in neuropathic pain and epilepsy. By generating a knock-in mouse expressing a mutated α₂δ-1, that is not able to bind GBP and PG, Field et al. (2006) revealed α₂δ-1 being the in vivo target for gabapentinoid drugs. This was ultimately confirmed by the lack of high affinity GBP binding in the brain of a conventional α₂δ-1 knockout mouse (Fuller-Bicer et al. 2009). As previously expected from studies in skeletal and cardiac myocytes (Obermair et al. 2005; Tuluc et al. 2007), these mice display decreased cardiac L-type currents with altered activation kinetics. In somatosensory neurons α₂δ-1 null mice showed a reduced Ca_V2.2 level and calcium channel current density, which was associated with a deficit in mechanical and cold sensitivity (Patel et al. 2013). Nevertheless, despite the widespread and predominant expression of α₂δ-1 in the brain, so far neither gross-anatomical alterations nor signs for neuronal disease have been identified. Very recently disruption and deletion of the CACNA2D1 gene in three human patients with epilepsy and intellectual disability has been described (Vergult et al. 2014). Interestingly similar gene aberrations have also been identified in healthy individuals; therefore the causal relation of α₂δ-1 disruption and human disease is still elusive. Senatore et al. (2012) identified an interaction of α₂δ-1 with mutant prion proteins, putting forward the idea that disrupted neurotransmission in prion diseases may be linked to impaired synaptic Ca_V trafficking via α₂δ subunits. Taken together the role of α₂δ-1 for CNS disease remains unclear and the resulting phenotypes and severities may be associated with alterations in the overall neuronal excitability, which can be influenced by a variety of other factors.

α₂δ-2:

There are several strains of naturally occurring mice with mutations in the Cacna2d2 gene (ducky, ducky^2j, and entla) (Barclay et al. 2001; Brodbeck et al. 2002; Brill et al. 2004; Donato et al. 2006) as well as a targeted α₂δ-2 knockout mouse (Ivanov et al. 2004), which result in a loss of the full-length α₂δ-2 protein (ducky, ducky^2j, knockout) or a structurally altered protein (entla). Affected homozygous mice show a decreased life span, infertility, and reduced body size when compared to heterozygous or wildtype littermates. The mice suffer from ataxia, paroxysmal dyskinesia, as well as epilepsy. α₂δ-2 null mice (ducky, ducky^2j, knockout) also exhibit altered cerebellar purkinje cell morphology and reduced calcium currents, as well as brainstem dysgenesis and spinal cord and myelination defects. In humans CACNA2D2 has been discussed as a potential tumor suppressor gene (Hesson et al. 2007) and a single nucleotide polymorphism (SNP)-based study identified CACNA2D2 as a candidate gene in childhood absence epilepsy (Chioza et al. 2009). Indeed a point mutation in CACNA2D2 resulting in reduced current density and slowed inactivation in neuronal calcium channels has recently been associated with early infantile epileptic encephalopathy (Edvardson et al. 2013). Pippucci et al. (2013) identified a frameshift mutation in CACNA2D2 in a patient with epilepsy, dyskinesia, cerebellar atrophy, psychomotor delay and dysmorphic features, a disease strikingly similar to the above-mentioned phenotypes of α₂δ-2 mouse models.

α₂δ-3:

Mice with a targeted deletion of the Cacna2d3 gene and the concomitant insertion of a bacterial β-galactosidase under its promoter were generated by Deltagen, Inc. These mice exhibit deficits in nociceptive pain processing and a delay in inflammatory heat hyperalgesia (Neely et al. 2010). Interestingly, α₂δ-3 knockout mice display a reduced acoustic startle reflex and distorted auditory brainstem responses, which are likely caused by defects in presynaptic Ca_V2.1 channel targeting and synapse formation of auditory nerve fiber terminals contacting cochlear nucleus bushy cells (Pirone et al. 2014). Drosophila embryos lacking the α₂δ-3 homologue straightjacket display altered heat nociception, a decrease in synaptic transmission and Ca_V2 channel abundance, as well as morphologically altered synaptic structure (Dickman et al. 2008; Kurshan et al. 2009; Neely et al. 2010). The C. elegans α₂δ-3 mutant unc-36 revealed a role in Ca_V2 (unc-2) channel trafficking (Saheki and Bargmann 2009) which is associated with altered presynaptic morphology (Caylor et al. 2013). However, so far the only link between α₂δ-3 and human disease is the identification of SNPs within the large genomic region of CACNA2D3 in patients with reduced sensitivity to acute noxious heat and chronic back pain (Neely et al. 2010).

α₂δ-4:

Although expression of α₂δ-4 is negligible in CNS neurons (Schlick et al. 2010), α₂δ-4 is the major α₂δ subunit in retinal photoreceptor cells (Knoflach et al. 2013). Mice with a spontaneous mutation of Cacna2d4 resulting in a truncated α₂δ-4 protein show structural and functional abnormalities in mouse ribbon synapses associated with the loss of rods (Wycisk et al. 2006a). A similar phenotype was reported in humans, where a comparable mutation of the CACNA2D4 gene causes slowly progressing cone dystrophy associated with night blindness (Wycisk et al. 2006b). In a genome-wide association study a rare partial deletion in CACNA2D4 in two patients with late onset bipolar disorder was identified and also other genetic studies provide links between the α₂δ-4 gene and psychiatric disorders (Van Den Bossche et al. 2012; see discussion therein). Any causal and mechanistic explanations for these associations remain as for now speculative, but may involve the expression of α₂δ-4 in a specific and low abundant CNS cell type. Other causes could be provided by the role of α₂δ-4 in vision or the close genetic proximity to the CACNA1C gene encoding for Ca_V1.2, which was shown to be associated with psychiatric disorders (Splawski et al. 2004).

4. Conclusion and outlook

Over the last years progress in unraveling the specific neuronal functions of neuronal α₂δ subunits has been made at several levels. The availability of novel knock-out and mutant animals together with sophisticated heterologous expression studies in a variety of cell systems opened the possibility studying the specific roles of α₂δ subunit isoforms. The findings showed that all α₂δ subunits have classical roles dependent on the Ca_V complex, such as the modulation of membrane trafficking or current properties. Interestingly, more recent studies suggested functions of α₂δ subunits in synapse formation, which may be in part or entirely independent of the Ca_V complex. Because, however, any defect or loss-of-function of α₂δ subunits will inevitably affect the entire calcium channel complex, a clear distinction between independent or dependent functions is inherently difficult.

An emerging common feature in loss-of-function models of single α₂δ isoforms is that the loss of one isoform only induces a strong phenotype in cells or tissues expressing predominantly this respective isoform. Such a pattern can be observed in the retina (α₂δ-4; Wycisk et al. 2006a), in retinal ganglion cells (α₂δ-1; Eroglu et al. 2009), in auditory nerves (α₂δ-3; Pirone et al. 2014), and is also supported by studies in invertebrates (Kurshan et al. 2009; Caylor et al. 2013). Quite contrary, in CNS brain regions or neurons which simultaneously express three different α₂δ isoforms the consequences of loss-of-function of a particular isoform may be compensated by the other isoforms thereby obscuring any potential phenotype. Thus, the severe consequences of α₂δ-2 mutations in animal models and human disease are likely caused by the predominant expression of this particular isoform in the cerebellum. There the low protein levels of α₂δ-1 and α₂δ-3 do not suffice to compensate the loss of α₂δ-2. Alternatively, Ca_V2.1 may exclusively form a complex with the auxiliary α₂δ-2 and β₄ subunits. Another possibility is that α₂δ-2 itself mediates unique channel-independent functions, which cannot be compensated by the other isoforms. This possibility can be tested by analyzing calcium channels and Ca_V-dependent neuronal functions in cortical neurons from α₂δ-2 knockout mice, as these neurons predominantly express α₂δ-1 and α₂δ-3 (Schlick et al. 2010). This endeavor is, however, complicated by the severe overall phenotype of α₂δ-2 knockout and mutant mice. In a nutshell the available animal models, experimental co-expression studies, as well as identified human disease mutations at the moment do not provide a clear cut answer as to whether the individual α₂δ isoforms have specific roles in CNS neurons.

Taken together, for the future the main challenge in elucidating the roles of neuronal α₂δ subunits will be sorting out the functions of individual isoforms in cells and tissues simultaneously expressing more than one isoform. In order to clearly distinguish between potential redundant and specific functions it will be necessary to generate knockout mouse models lacking different combinations of α₂δ isoforms (e.g. double knockout models). Such mouse models need to be complemented by native neuronal expression systems lacking two if not all three α₂δ isoforms for studying the consequences on pre- and postsynaptic Ca_V-dependent functions. Characterization of the behavioral phenotype and the brain anatomy of such mouse models as well as analyzing neuronal functions could strongly contribute to a better understanding of the molecular mechanism of α₂δ signaling and potentially uncover novel pathophysiological mechanisms.