Keywords

Introduction

Voltage-gated calcium (CaV) channels mediate and regulate a variety of functions ranging from muscle contraction, hormone secretion, and synaptic transmission to gene regulation. To provide precise timing and control of the calcium ions entering through CaV channels, they operate, with some exceptions, in heteromultimeric complexes with auxiliary and cytoplasmic β and extracellular α2δ subunits. At least in skeletal muscle, a transmembrane γ subunit is also part of the channel structure (Wu et al., 2016). The traditional concept of CaV channel function envisions the pore-forming subunit as the major determinant of the basic biophysical, pharmacological, and physiological properties (see chapters “Subunit Architecture and Atomic Structure of Voltage-Gated Ca2+ Channels” by William A. Catterall, “Voltage-Gated Calcium Channel Auxiliary β Subunits” by Sergej Borowik and Henry M. Colecraft, “Pharmacology of Voltage-Gated Calcium Channels at Atomic Resolution” by William A. Catterall and “Pharmacology and Structure-Function of Venom Peptide Inhibitors of N-Type (Cav2.2) Calcium Channels” by Md. Mahadhi Hasan et al.). Auxiliary subunits are thought to fine-tune CaV channel functions by regulating or modulating the membrane expression, localization, trafficking, and biophysical properties of the channel complex (reviewed in Arikkath & Campbell, 2003; Buraei & Yang, 2010; Dolphin, 2009, 2016; Obermair et al., 2008).

Among the auxiliary subunits, the α2δ subunit is unique in that the entire protein is extracellular and highly glycosylated, and hence, it interacts with the extracellularly exposed surface of the α1 subunit. Importantly, this distinctive position also theoretically enables α2δ subunits to link the CaV complex with other potential extracellular signaling or scaffolding proteins. Indeed, research over recent years has identified a number of extracellular interaction partners as well as physiological functions, some of which may be independent of the classical interaction with the CaV complex. A variety of disorders linked to defective α2δ subunit functions provide another line of evidence of key roles for these versatile proteins, and the widely prescribed anti-epileptic and anti-allodynic drugs gabapentin and pregabalin act by binding to specific α2δ isoforms. In this chapter, we provide a concise overview of the present state of knowledge about α2δ subunits, including α2δ structure and tissue distribution, the role of specific structural domains for protein function, and CaV channel-dependent and -independent functions. We particularly summarize the mechanisms of action of α2δ subunits and discuss recent developments concerning their pharmacology and disease association. Investigating and understanding the detailed pathophysiological mechanisms involving CaV complexes and α2δ subunits may open the path to the identification of new and specific treatment paradigms.

Discovery of α2δ Subunits

The CaV channel complex was purified, and its properties first investigated from skeletal and then cardiac muscle, by virtue of the ability of the channel to bind 3H-dihydropyridines (DHPs), which were known to block L-type calcium channels (Cooper et al., 1987; Curtis & Catterall, 1984; De Jongh et al., 1989; Leung et al., 1987; Takahashi et al., 1987; Tanabe et al., 1987). The subunit to which the DHP calcium channel blocker bound was identified to be the channel itself, and termed the α1 subunit (α1S, with the “S” referring to skeletal muscle). This co-purified with a similar molecular weight but glycosylated protein, later named α2δ (Sharp et al., 1987), and smaller molecular weight β and γ subunits (De Jongh et al., 1989, 1990; Takahashi et al., 1987; Tanabe et al., 1987).

The α2δ subunit was so called as it was found to consist of the two proteins α2 and δ that remain associated, except in reducing conditions when they are clearly separated (De Jongh et al., 1990). These two disulfide-bonded proteins were both highly glycosylated, suggesting they were largely extracellular proteins, and they were also membrane associated, and initially identified as transmembrane proteins, with δ containing a hydrophobic, potentially transmembrane, α-helix (Brickley et al., 1995; Gurnett et al., 1996).

Identification of α2δ Subunits

Cloning

The α2δ-1 subunit was the first α2δ whose gene was identified and cloned (Cacna2d1). This was achieved by first obtaining some peptide sequences from the α2 and δ proteins purified from skeletal muscle. Once obtained, the α2δ sequence then showed that both α2 and δ subunits were encoded by the same gene, which is therefore translated as a pre-protein, and then enzymatically cleaved into α2 and δ (De Jongh et al., 1990; Ellis et al., 1988). Two additional genes encoding α2δ-2 and α2δ-3 were identified by homology to α2δ-1 (Ellis et al., 1988), and α2δ-3 was then cloned (Klugbauer et al., 1999). α2δ-2 was also cloned by virtue of the fact that it is mutated in a spontaneously arising mouse model of absence epilepsy and ataxia called Ducky (Barclay et al., 2001; Barclay & Rees, 2000), and α2δ-4 was cloned by homology to other α2δ sequences (Qin et al., 2002).

Splice Variants

The Cacna2d1 gene for α2δ-1 was originally identified to contain three alternatively spliced regions (A, B, and C) within the α2 moiety (Angelotti & Hofmann, 1996). The rat α2δ-1 genomic sequence contains 39 exons, and further analysis of this sequence found that A is encoded by exon 18a, B is formed as a result of utilizing an alternative 3′ splice acceptor site for exon 19, while C represents the inclusion of exon 23 (Lana et al., 2014) (Fig. 1). In skeletal muscle, α2δ-1 (+A + B ΔC) was the only splice variant detected, whereas in heart, multiple different splice variants were found (Angelotti & Hofmann, 1996; Lana et al., 2014). By contrast, in neuronal tissue, including dorsal root ganglion (DRG) neurons, the predominant splice variant is ΔA + B + C (Angelotti & Hofmann, 1996; Lana et al., 2014) (Fig. 1). The mouse Cacna2d2 gene also contains 39 exons (Barclay & Rees, 2000), and splice region C has been found to be of functional relevance, as it regulates the trans-synaptic recruitment of postsynaptic GABAA receptors and axonal wiring (Geisler et al., 2019).

Fig. 1
Splice variants of alpha 2 delta one. Nine rectangles in a series with exon numbers 17, 18, 18 a, 19, 20, 21, 22, 23, and 24 are displayed with sizes of 75, 75, 57, 72, 62, 77, 21, and 61 respectively. 18 a is labeled as A, part of 19 is labeled as B, and 23 is labeled as C. 18 a and 19, 18 and 19, and 22 and 24 are connected externally via V-shaped connectors.

Splice variants in α2δ-1. Three alternatively spliced regions (A), (B), and (C) were originally described by Angelotti and Hofmann (1996), and the splice variants were further defined by Lana et al. (2014)

Full size image

Tissue Distribution of α2δ Subunits

α2δ-1

The distribution of α2δ-1 is widespread, being the main α2δ in skeletal, smooth, and cardiac muscle, and it is also present in brain (Angelotti & Hofmann, 1996). It is strongly expressed in DRG neurons (Newton et al., 2001) and is the major α2δ isoform in cortical brain regions (Schlick et al., 2010).

α2δ-2

This species was originally found in multiple tissues, including skeletal and cardiac muscle, pancreas, and brain (Klugbauer et al., 1999, 2003). Within the brain, it is found to be strongly expressed in cerebellum, particularly in Purkinje neurons (Barclay et al., 2001) and, at a lower expression level, also in the cortex (Schlick et al., 2010).

α2δ-3

This isoform is found mainly in brain (Klugbauer et al., 2003), such as the cortex and hippocampus, and it is the major isoform expressed in the striatum (Geisler et al., 2019).

α2δ-4

The α2δ-4 isoform is strongly expressed in the retina, particularly at photoreceptor ribbon synapses (Knoflach et al., 2013; Lee et al., 2015; Wycisk et al., 2006a); its expression in the brain, however, seems negligible (Schlick et al., 2010).

Structure of α2δ Subunits

Biochemical and Bioinformatic Studies

Following cloning, the primary α2δ sequence showed that the N-terminus of α2δ-1 had a predicted signal sequence; therefore signifying it was extracellular (Fig. 2a). The C-terminus of α2δ-1 had a hydrophobic domain, suggesting it is transmembrane, with a very short potentially intracellular sequence. Furthermore, the existence of multiple (16–18) glycosylation sites (Fig. 2a, b), and antibody mapping indicated the protein is mainly extracellular (Brickley et al., 1995; Gurnett et al., 1996). Experimentally, both reducing conditions and deglycosylation are important tools to distinguish between α2δ and α2 in immunoblots, particularly in expression studies (Fig. 2b), and the low molecular weight δ can also be identified separately (Davies et al., 2010; De Jongh et al., 1990; Jay et al., 1991). Although almost all α2δ in native tissues is proteolytically cleaved, uncleaved α2δ-1 was identified in DRG neuron cell bodies (Kadurin et al., 2016). Biochemical and other studies also showed disulfide bonding between α2 and δ (Calderon-Rivera et al., 2012; De Jongh et al., 1990).

Fig. 2
Two figures. N term signal sequence, C terminal G P I signal sequence, endopeptidase transamidase, multiple disguised bonds, multiple glycosylation sites, and proteolytic cleavage are displayed in the first figure. The second figure is a sketch of the alpha 2 delta 1 western blot.

Post-translational processing of α2δ subunits. (a) Diagram showing α2δ pre-protein (left) and subsequent post-translational processing steps (center), including removal of N-terminal signal sequence during translation (1), GPI anchor attachment within the ER (2, 3), formation of multiple disulfide bonds (4), glycosylation (5), and proteolytic cleavage of α2δ into α2 and δ (6, right). (b) Western blot of material from tsA-201 cells transfected with α2δ-1, showing that α2δ-1 and cleaved α2 cannot be clearly differentiated (left lane) unless deglycosylated with PNGase F (right lane). Primary antibody α2δ-1 monoclonal recognizing an epitope within α2. (Data courtesy of Dr. Ivan Kadurin)

Full size image

GPI Anchoring

It was found that α2δ subunits were strongly expressed in lipid rafts (Davies et al., 2006), and subsequently, it was identified that the α2δ subunits were not transmembrane, but rather glycosylphosphatidylinositol (GPI) anchored, using both bioinformatic and biochemical techniques (Davies et al., 2010). This GPI anchoring has been confirmed from the structure of α2δ-1 in the CaV1.1 and CaV2.2 channel complexes (Gao et al., 2021; Wu et al., 2016). GPI anchoring occurs immediately following translation in the endoplasmic reticulum (ER), when the pre-formed GPI anchor replaces the C-terminal hydrophobic sequence (Hooper, 2001) (Fig. 2a). GPI-anchored proteins have different properties from transmembrane proteins, for example, they are highly mobile and concentrated in cholesterol-rich membrane fractions (termed lipid rafts) (Hooper, 2001; Kadurin et al., 2012), and are rapidly endocytosed and recycled to the plasma membrane (Mayor & Riezman, 2004; Tran-Van-Minh & Dolphin, 2010).

VWA Domain

The presence of a von Willebrand factor-A (VWA) domain in the α2δ subunits was identified bioinformatically (Whittaker & Hynes, 2002). The VWA domains were then found to be essential for the function of α2δ to enhance CaV1 and CaV2 calcium currents (Canti et al., 2005) and for synaptic calcium channel localization (Schӧpf et al., 2021). The VWA domain was predicted to be involved in interacting with the α1 subunit, via its metal ion-dependent adhesion site (MIDAS) motif (Canti et al., 2005). This was shown to be the case from the structure of the CaV1.1 and CaV2.2 channel complexes, where the MIDAS motif in α2δ-1 interacts with an aspartate in the first extracellular loop of domain I of the α1 subunit (Gao et al., 2021; Wu et al., 2016) (Fig. 3a–c). In confirmation of this, mutation of this interaction site in CaV1.2 and CaV2.2 also prevented the effect of α2δ-1 (Bourdin et al., 2017; Dahimene et al., 2018).

Fig. 3
3 diagrams. A describes the domain structure that contains alpha 2 and delta. Alpha 2 contains cache 1, V W A, cache 2, cache 3, and cache 4. In B, alpha 1, beta 3, and alpha 2 delta 1 subunits are intertwined in the cryo E M structure of the C a v 2.2 complex. All these are helical shaped. C consists of a set of colored cylinders in series. This represents the interaction site in the V W A of alpha 2.

Domains in α2δ-1 and interaction with the α1 subunit. (a) Domain structure for α2δ showing four Cache domains, with the VWA domain inserted between Cache 1 and Cache 2. (b) Cryo-EM structure of the CaV2.2 complex (Gao et al., 2021), showing the α1 subunit (green), the β3 subunit (magenta), and the α2δ-1 subunit (rainbow). The MIDAS interaction site of the α2δ-1 VWA domain with the α1 subunit contains a co-ordinating divalent cation (red sphere). Image prepared from pdb 7miy using Pymol. (c) Diagram showing interaction site of the MIDAS motif in the VWA domain of α2 with an aspartate residue (D) in the first extracellular loop of the α1 subunit

Full size image

Cache Domains

Two Cache domains were identified bioinformatically in α2δ subunits (Anantharaman & Aravind, 2000); these Cache domains are very similar in structure to those identified in bacterial chemoreceptors and chemotransducers, where they are often involved in nutrient sensing and chemotaxis (Gumerov et al., 2022). The cryo-EM structure has identified four Cache domains in α2δ-1 (Fig. 3a, b).

Proteolytic Maturation of α2δ

The α2δ pre-protein encodes α2 at its N-terminus, followed by the shorter δ sequence. Following translation in the ER, the uncleaved α2δ begins to be cleaved in the Golgi apparatus (Kadurin et al., 2017), but the two “subunits” remain associated by pre-formed disulfide bonds, created as the protein folds in the ER. Proteolytic cleavage of α2δ appears to represent an activation step for CaV channel function (Ferron et al., 2018; Kadurin et al., 2016) (Fig. 2a).

Cachd1 Protein

Cachd1 was initially identified bioinformatically (although misnamed) (Whittaker & Hynes, 2002) and later confirmed as an α2δ-like protein (Cottrell et al., 2018; Dahimene et al., 2018) containing an imperfect VWA domain and Cache domains. However, the MIDAS motif in the VWA domain of Cachd1 is highly disrupted and non-functional in terms of enhancement of CaV2.2 function (Dahimene et al., 2018). Cachd1 also produced differential functional effects on particular calcium channels (see section “Cachd1 function”) (Cottrell et al., 2018; Dahimene et al., 2018).

Molecular Structure of α2δ-1

The predicted domains and post-translational modifications in α2δ proteins are all supported by the structure of α2δ-1 in the skeletal muscle calcium channel (CaV1.1) complex, which was obtained by cryo-electron microscopy (EM) (Wu et al., 2016). In summary of the salient points, the VWA domain was found to interact with an aspartate on the first extracellular loop on the α1S calcium channel, which coordinated divalent cation binding with the MIDAS motif in α2δ-1. No transmembrane domain was detected for α2δ, in agreement with the evidence that it is GPI anchored (Davies et al., 2010). Four Cache domains were identified in α2δ, with the VWA domain inserted between the first two (Wu et al., 2016). The structure of α2δ-1 was found to be very similar in the CaV2.2 complex obtained following expression of the subunits in HEK293T cells, and subsequent purification of the complex (Gao et al., 2021).

Functions of α2δ Subunits as Calcium Channel Subunits

Effects of Cloned α2δ Subunits on Calcium Currents

In most studies α2δ subunits have been found to increase CaV1- and CaV2-mediated currents, although this function is partly dependent on the α1 isoform, as knockdown of α2δ-1 in skeletal muscle cells did not reduce CaV1.1-mediated currents (Meyer et al., 2019; Obermair et al., 2005), whereas the effect on CaV2 channel current density was more pronounced (Barclay et al., 2001; Canti et al., 2005; Hendrich et al., 2008; Obermair et al., 2008). Co-expression studies showed that α2δ-1 increased CaV calcium currents that were mediated, for example, by CaV2.1/β4 (Gurnett et al., 1996, 1997), CaV1.2/β2a, and CaV2.3/β3 (Klugbauer et al., 1999; Yamaguchi et al., 2000). Moreover, for CaV1.2, CaV2.2, and CaV2.3, the different auxiliary β subunits were found to regulate CaV current properties synergistically with α2δ-1 in a β subunit-specific manner (Yasuda et al., 2004). Similarly, α2δ-2 increased CaV2.1/β4 calcium currents (Barclay et al., 2001; Brodbeck et al., 2002) as well as CaV1.2/β1b and CaV2.2/β1b currents (Canti et al., 2005). The α2δ-3 subunit also increased CaV1.2/β2a and CaV2.3/β3 (Klugbauer et al., 1999) and α2δ-4, the least well-studied α2δ subunit, increased CaV1.2/β3-mediated calcium influx (Qin et al., 2002). The results of these studies always need to be prefaced by the finding that some heterologous expression systems may contain endogenous calcium currents (Berjukow et al., 1996), as well as auxiliary β (Canti et al., 2001) and α2δ subunits (Kadurin et al., 2012).

Effects of α2δ Subunits on Biophysical Properties of Calcium Currents

No change in single-channel conductance has been observed that can be attributed to α2δ subunits (Brodbeck et al., 2002; Wakamori et al., 1999), indicating the main effects are not on permeation. However, α2δ-1 was found to reduce the percentage of null sweeps in single-channel recordings of CaV2.2 (Wakamori et al., 1999) and shRNA knockdown of α2δ-1 slightly reduced the open probability of CaV1.2 (Tuluc et al., 2007). The α2δ subunits also affect the kinetics of activation and inactivation (Canti et al., 2003; Obermair et al., 2005, 2008; Tuluc et al., 2007; Wakamori et al., 1999), and voltage sensor movement for CaV1.2 (Savalli et al., 2016) and result in a hyperpolarization of both current activation (Felix et al., 1997; Savalli et al., 2016) and inactivation (Canti et al., 2003).

Effects of α2δ Subunits on Calcium Channel Trafficking

α2δ subunits have been found to increase the plasma membrane expression of CaV2.2 (Cassidy et al., 2014). This is the case for both exon 37a and 37b-containing C-terminal splice variants of CaV2.2 (Macabuag & Dolphin, 2015), which show differential gating (Castiglioni et al., 2006), cell surface expression (Macabuag & Dolphin, 2015), and expression in the pain pathway (Bell et al., 2004). Furthermore, presynaptic clustering of CaV2.1 and CaV2.2 was strongly reduced in α2δ triple knockout/knockdown neurons (Schӧpf et al., 2021).

However, for CaV2.2, it has been shown that the increase in cell surface expression is generally not to the same extent as the increase in calcium current density, which may be up to 12-fold (Hoppa et al., 2012). The increase in cell surface expression has also been found to be greater for α2δ-1/2 than for α2δ-3, and is likely to stem from an increase in net forward trafficking of CaV2.2, which is much more pronounced for α2δ-1/2 than for α2δ-3 (Meyer & Dolphin, 2021), with no effect on endocytosis (Cassidy et al., 2014). CaV channels with mutations in their selectivity filter, such that they do not conduct Ca2+, show defective trafficking (Meyer et al., 2019). However, the trafficking of these permeation-defective channels is still enhanced by α2δ-1 (Meyer et al., 2019), indicating that α2δ subunits do not represent a checkpoint such that only functional CaV channels respond to them with increased trafficking.

The development of a knock-in mouse containing an epitope tag in endogenous CaV2.2 has allowed the effect of knockout of α2δ-1 to be investigated on the distribution of endogenous CaV2.2 (Nieto-Rostro et al., 2018). This study showed that CaV2.2 is strongly expressed on the cell surface of particular types of nociceptive DRG neurons and in their presynaptic terminals in the dorsal horn (Fig. 4). Knockout of α2δ-1 dramatically reduced the cell surface expression of CaV2.2 and presynaptic distribution (Fig. 4) (Nieto-Rostro et al., 2018), confirming the importance of α2δ-1 proteins in CaV2.2 trafficking within the pain pathway.

Fig. 4
D G R sections immunostained with H A, C G R P stain, and 4 prime, 6-diamidino-2-phenylindole. The distribution has combining circular structures.

Effect of α2δ-1 knockout on CaV2.2-HA distribution in mouse DRG neurons. DRG sections, showing immunostaining for HA (green) in CaV2.2_HAKI/KI DRGs, co-stained with CGRP (red). Nuclei are stained with 4′,6-diamidino-2-phenylindole (DAPI, blue). Left: α2δ-1+/+ DRGs, showing CaV2.2 presence at the cell surface; right: α2δ-1−/− DRGs, showing no CaV2.2 at the cell surface (e.g., yellow arrow). Scale bar 5 μm. (Images taken from Nieto-Rostro et al., 2018 under CC-BY 4.0 license)

Full size image

In contrast, the CaV1 channels appear to show a smaller trafficking response to α2δ subunits than the CaV2 channels, which may reflect their greater membrane stability in cardiac and skeletal muscle. For example, knockdown of α2δ-1 in muscle cells did not affect targeting and membrane expression of CaV1.1 and CaV1.2 and only modestly decreased CaV1.2 current density, while it strongly affected activation and inactivation kinetics (Obermair et al., 2005; Tuluc et al., 2007).

Cachd1 Function

The roles of the Cache domains in α2δ subunits remain unclear. Relevant to this, the α2δ-like Cachd1 protein was found to produce an increase in CaV2.2 but not CaV2.1 currents (Dahimene et al., 2018), and quite contrary to α2δ subunits, Cachd1 was also found to produce an increase in CaV3 T-type channel currents (Cottrell et al., 2018). The basis for this differential selectivity remains unclear and might relate to specific splice variants, or interactions with specific extracellular elements of the different CaV channels. Although Cachd1 was found to increase CaV2.2 currents, this was to a much smaller extent than observed with α2δ subunits (Dahimene et al., 2018), and it also increased CaV2.2 cell surface expression in parallel (Dahimene et al., 2018). However, unlike α2δ-1, the interaction with Cachd1 did not require its imperfect VWA domain interacting with the first extracellular loop of the CaV2.2 α1 subunit (Dahimene et al., 2018). Therefore, for CaV2.2, it is likely to involve other interactions, for example, between the Cache domains of Cachd1 and extracellular domains of the α1 subunit.

Synaptic Functions of α2δ Proteins Beyond Their Role as Calcium Channel Subunits

Importance of α2δ Proteins in Neuronal and Synaptic Functions

Several studies implicate an involvement of α2δ proteins in neuronal and synaptic functions, which partly go beyond their role as CaV channel subunits (Chen et al., 2018; Eroglu et al., 2009; Geisler et al., 2019; Kurshan et al., 2009; Schӧpf et al., 2021). Most importantly, α2δ proteins have emerged as critical regulators of synapse formation and differentiation in central nervous system neurons (Eroglu et al., 2009; Schӧpf et al., 2021) and expression of specific α2δ isoforms during distinct developmental phases may shape the structural and functional neuronal network connectivity (Bikbaev et al., 2020). While postsynaptically located α2δ-1 subunits have been found to be relevant for basic neuronal differentiation and for modulating postsynaptic signaling (Eroglu et al., 2009), for example, by serving as a receptor for the astrocyte-secreted and synaptogenic thrombospondins (TSPs), presynaptic α2δ proteins serve dual purposes: on one hand, they regulate the abundance of presynaptic calcium channels and hence can directly regulate the efficacy of synaptic transmission (Ferron et al., 2018; Hoppa et al., 2012; Schӧpf et al., 2021). On the other hand, they regulate synapse formation, differentiation, and the trans-synaptic recruitment of postsynaptic receptors (Fell et al., 2016; Geisler et al., 2019; Schӧpf et al., 2021).

Postsynaptic Functions of α2δ-1

In retinal ganglion cells the interaction of postsynaptic α2δ-1 and astrocyte-secreted thrombospondin-1 (TSP-1) mediates excitatory synapse formation in a mechanism, which is independent of the presence and function of α1 subunits (Eroglu et al., 2009) but involves the recruitment of NMDA receptors to the postsynaptic membrane (Risher et al., 2018). A role in synapse formation is further supported by impaired excitatory synaptogenesis and spine morphology in conditional (Risher et al., 2018) and constitutive (Bikbaev et al., 2020) α2δ-1 knockout mice. TSP-4 was also identified as an α2δ-1 ligand (Eroglu et al., 2009), although the interaction was subsequently found to be weak (El-Awaad et al., 2019; Lana et al., 2016), and less evident than for another known TSP ligand, low-density lipoprotein receptor-related protein 1 (LRP1) (Lana et al., 2016). However, and in contrast to TSP-1, this interaction likely occurs with presynaptic α2δ subunits (Yu et al., 2018).

Presynaptic and Trans-synaptic Roles of α2δ Proteins

We recently showed that in cultured hippocampal neurons presynaptic triple knockout/knockdown of all brain α2δ isoforms severely compromised synapse formation and pre- and postsynaptic differentiation (Fig. 5) (Schӧpf et al., 2021). This striking phenotype could be rescued by the expression of each individual α2δ isoform, suggesting a surprisingly redundant presynaptic and trans-synaptic role in this critical neuronal function. Trans-synaptic signaling via α2δ proteins is supported by other recent observations. Expression of an α2δ-2 splice variant lacking exon 23 (splice site C, see above) in hippocampal neurons is sufficient to trigger the aberrant wiring of presynaptic excitatory (glutamatergic) axons to inhibitory (GABAergic) postsynaptic sites (Fig. 6). Furthermore, this α2δ-2 splice variant, when expressed in presynaptic nerve terminals, regulates postsynaptic GABAA receptor (GABAAR) abundance, both in aberrantly wired synapses as well as normally wired inhibitory synapses (Geisler et al., 2019). Also, in inner hair cell synapses of the cochlea, α2δ-2 is necessary for the proper spatial alignment of presynaptic L-type CaV1.3 calcium channels and postsynaptic AMPA receptors (Fell et al., 2016). The α2δ-2 isoform is also involved in modulating axonal regeneration after injury and establishing neuronal circuits (Tedeschi et al., 2016). Synaptic functions of α2δ-3 so far have mainly been addressed in invertebrate model systems, where it was involved in regulating the size and morphology of motoneuron terminals (Caylor et al., 2013; Kurshan et al., 2009). A role for α2δ-3 in regulating presynaptic differentiation has also been confirmed in vertebrates, by studying the consequences of α2δ-3 knockout on auditory nerve fibers (Pirone et al., 2014) and the expression of α2δ-3 in GABAergic neurons (Geisler et al., 2019). In the context of α2δ functions independent of the calcium channel complex, it is noteworthy that a quantitative study of the neuronal CaV2 channel proteome suggested a considerably weaker association of α2δ with α1 subunits when compared to β subunits, although this was dependent on the detergent used for membrane solubilization (Müller et al., 2010). Furthermore, α1 and α2δ subunit interactions in the neuronal membrane are likely dynamic and association may only be transient, as suggested by single-molecule live-cell imaging (Brockhaus et al., 2018; Schneider et al., 2015; Voigt et al., 2016). In retinal photoreceptor cells, where the L-type channel CaV1.4 serves as the main presynaptic channel of ribbon synapses, α2δ-4 regulates functional membrane expression of α1 subunits and synaptic transmission of rods (Wang et al., 2017) and cones (Kerov et al., 2018; Schlegel et al., 2019). In line with a trans-synaptic role of α2δ proteins, knockout α2δ-4 affects not only presynaptic structure but also postsynaptic receptor clustering (Wang et al., 2017). Finally, the well-characterized mechanism of α2δ-1 in neuropathic pain and the development of hyperalgesia are also mediated by presynaptic mechanisms (Bauer et al., 2009; Nieto-Rostro et al., 2018; Yu et al., 2018).

Fig. 5
a and b have 2 sets of microscopic images and diagrams that illustrate the effects of presynaptic and postsynaptic triple knockout of alpha 2 delta 1,2 and 3.

Presynaptic α2δ subunits mediate excitatory synapse formation and trans-synaptic differentiation. Immunofluorescence labeling of eGFP-labeled axons with presynaptic boutons (a and b, left panels, arrows) and postsynaptic dendrites with dendritic spines (a and b, right panels, arrows) from presynaptic or postsynaptic α2δ triple knockout/knockdown hippocampal neurons. The sketches summarize the observed labelling patterns. (Images taken from Figure 4 of Schöpf et al., 2021 under a CC BY 4.0 license). Scale bars, 2 μm (selection) and 8 μm (overview). (a) Presynaptic triple knockout/knockdown of α2δ-1, α2δ-2, and α2δ-3 results in failed calcium channel and synapsin clustering (left panel, arrows and sketch). In contrast, dendritic spines opposite presynaptic boutons containing Cav2.1 and synapsin clusters develop normal when all α2δ subunits have been knocked out/down in the postsynaptic neuron (right panel, arrows and sketch). (b) Presynaptic triple knockout/knockdown of α2δ-1, α2δ-2, and α2δ-3 also affects postsynaptic differentiation as seen by missing PSD95 clustering (left panel, arrows and sketch) opposite defective presynaptic terminals. In contrast, postsynaptic α2δ subunit triple knockout/knockdown does not affect PSD95 clustering opposite presynaptic terminals containing Cav2.1 and synapsin clusters (right panel, arrows and sketch)

Full size image
Fig. 6
A is a microscopic image of neurons labeled m cherry and e G F P using two different colors. A part of the G F P positive dendrite is zoomed in to illustrate the positions of synapses on it. B is a set of four microscopic images arranged in two rows and two illustrations of control glutamatergic axons. The rows are labeled control and alpha 2 delta 2 respectively.

Presynaptic α2δ-2 induces aberrant wiring of excitatory synapses. (a) Experimental setup to analyze the position of synapses on postsynaptic dendrites of excitatory cultured hippocampal neurons: Presynaptic control or α2δ-2 expressing neurons were labeled with mCherry (red) and postsynaptic neurons with eGFP (green). The magnified inset shows the contact of presynaptic axonal boutons (mCherry) with postsynaptic dendritic spines (eGFP), as expected for excitatory spine synapses. Scale bars, 50 μm (overview), 3 μm (inset). (b) Super-resolution gSTED microscopy confirms the preferential location of excitatory synapses on dendritic spines of excitatory glutamatergic neurons (white arrowheads and sketch). Strikingly, presynaptic axons expressing α2δ-2_ΔE23 aberrantly wire to postsynaptic sites along dendritic shafts (blue arrowheads and sketch), a position typically observed for GABAergic synapses. Scale bars, 2 μm. (Images taken from Figure 11 of Geisler et al., 2019 under a CC BY 4.0 license)

Full size image

Diseases Associated with α2δ Subunits

It is evident from the above that α2δ proteins are abundantly expressed in various organs and particularly in the brain, and that they are involved in critical calcium channel-dependent as well as-independent functions. Hence, it is not surprising that calcium channel dysfunctions are associated with a large variety of disorders. Over recent years an increasing number of disorders in humans have been linked to the genes encoding α2δ subunits (reviewed in Ablinger et al., 2020), however, so far detailed insights into pathophysiological mechanisms are only beginning to emerge.

Ducky Mice and Human Mutations of the α2δ-2 Gene

A number of recessive mutations in Cacna2d2 underlie the cerebellar ataxia and epilepsy phenotypes seen in spontaneously arising mouse mutants ducky, ducky2J, and entla, as well as engineered Cacna2d2 knockout mice (Barclay et al., 2001; Brill et al., 2004; Brodbeck et al., 2002; Donato et al., 2006; Geisler et al., 2021; Ivanov et al., 2004). Furthermore, mutations in human CACNA2D2 (located on chromosome 3p21.31) are associated with recessive epileptic encephalopathy and mental retardation (Edvardson et al., 2013; Pippucci et al., 2013; Punetha et al., 2019) and may be also be linked to schizophrenia (Rodríguez-López et al., 2018).

Mutations in the α2δ-1 Gene Associated with Cardiac Phenotypes

CACNA2D1 (located on chromosome 7q21.11) mutations have been reported to be associated with cardiac dysfunction, including short QT syndrome (Templin et al., 2011) and Brugada syndrome (Burashnikov et al., 2010), although the effects of these mutations measured experimentally are rather inconsistent (Bourdin et al., 2015). In agreement, homozygous Cacna2d1 knockout causes a mild cardiac phenotype in mice (Fuller-Bicer et al., 2009) and additionally increases the susceptibility for diabetes (Mastrolia et al., 2017). However, several cases have been identified of epileptic encephalopathy involving copy number variants that usually contain multiple genes including CACNA2D1 (Mefford et al., 2011; Vergult et al., 2015). A recent study also found α2δ-1 was a target for autoimmune encephalitis (Lee et al., 2021), and another study also showed the presence of α2δ-1 auto-antibodies in cases of amyotrophic lateral sclerosis associated with type 2 diabetes (Shi et al., 2019).

Neuropathic Injury and the Role of α2δ-1

Peripheral sensory nerve injury in a variety of rodent models, including spinal nerve injury and administration of chemotherapeutic drugs, results in an increase of α2δ-1 mRNA in damaged DRG neurons (Bauer et al., 2009; Lana et al., 2014; Newton et al., 2001; Wang et al., 2002; Xiao et al., 2007). This gives rise to a corresponding increase of α2δ-1 protein within the injured DRG cell bodies, and also in their primary afferent terminals (Bauer et al., 2009; Luo et al., 2001). Furthermore, there was a differential upregulation of a splice variant of α2δ-1 that shows a lower affinity for 3H-gabapentin (Lana et al., 2014). In agreement with these results, Cacna2d1 knockout mice showed reduced sensitivity to mechanical stimulation and delayed onset of neuropathic mechanical hypersensitivity following peripheral nerve injury (Patel et al., 2013). Furthermore, α2δ-1 overexpressing mice have been generated, and they show increased baseline response to painful stimuli (Li et al., 2006).

Disorders Associated with α2δ-3

From knockout models of Cacna2d3, a clear role for α2δ-3 in hearing has been identified (Pirone et al., 2014), but no related human mutations in CACNA2D3 (located on chromosome 3p21.1) have yet been reported. However, independent studies qualify CACNA2D3 as risk gene for autism spectrum disorders (see below). Furthermore, a genome-wide Drosophila RNAi screen for heat nociception identified Cacna2d3 as a pain-related gene, which was supported by the phenotype of α2δ-3 knockout mice and human single-nucleotide polymorphisms (SNPs) associated with pain conditions, although potential human disease mechanisms are still unknown (Neely et al., 2010).

Disorders Associated with α2δ-4

Mutations in both mouse and human CACNA2D4 (located on chromosome 12p13.33 in humans) cause dysfunction of photoreceptors, resulting in certain forms of night blindness (Wycisk et al., 2006a, b).

Psychiatric Disorders

An interesting common feature of α2δ subunits is that over recent years SNPs or genomic variations in all CACNA2D (α2δ subunit) genes have been linked to a spectrum of psychiatric disorders (Ablinger et al., 2020; Consortium, 2013). Indeed, SNPs in CACNA2D2 and CACNA2D4 exhibited statistically significant associations with disease, across multiple psychiatric conditions. More recently, an excess of several rare disruptive mutations in CACNA2D1, CACNA2D2, CACNA2D3, and CACNA2D4 were observed in cases of schizophrenia (Purcell et al., 2014), as reviewed recently (Ablinger et al., 2020; Heyes et al., 2015). Furthermore, several mutations in CACNA2D3, including a splice site mutation (Iossifov et al., 2012), are potentially linked to autism spectrum disorders (De Rubeis et al., 2014; Guo et al., 2018). Similarly in CACNA2D1 a de novo mutation has been linked to autism (Iossifov et al., 2014), and a recent exome sequencing study suggests CACNA2D1 as candidate risk gene for neurodevelopmental disorders (Valentino et al., 2021). Furthermore, a partial CACNA2D4 deletion has been associated with rare cases of late onset bipolar disorder (Van Den Bossche et al., 2012).

Pharmacology Involving α2δ Subunits

Therapeutic Uses of Gabapentinoids

Gabapentin was first identified as an anti-epileptic drug (Crawford et al., 1987), and then as having efficacy in painful neuropathies (Wiffen et al., 2005). There are currently three therapeutically available gabapentinoid α2δ ligands which are used in chronic neuropathic pain conditions: gabapentin, pregabalin (Field et al., 1999; Li et al., 2011), and mirogabalin (Domon et al., 2018).

Gabapentinoid Drug Binding to α2δ Subunits and Potential Mechanisms of Action

The therapeutic target for gabapentin was identified as α2δ-1 by purifying a brain protein that binds to 3H-gabapentin (Brown & Gee, 1998; Gee et al., 1996). However, gabapentin has very little effect acutely on calcium currents (Hendrich et al., 2008; Kang et al., 2002; Martin et al., 2002), but chronic incubation with gabapentin for 24–48 h reduces calcium currents (Biggs et al., 2014; Hendrich et al., 2008), synaptic transmission (Hendrich et al., 2012; Lempel et al., 2017), and cell surface expression of α2δ-1, α2δ-2, and CaV2.2 (Cassidy et al., 2014; Tran-Van-Minh & Dolphin, 2010).

The binding site for gabapentin was found to include an RRR motif, which is just upstream of the VWA domain (Brown & Gee, 1998; Wang et al., 1999). This is present in α2δ-1/2 but not the other α2δ subunits which do not bind gabapentin (Gong et al., 2001; Marais et al., 2001). If the third R is mutated to A in α2δ-1 or α2δ-2, this reduces the affinity of gabapentinoid binding and α2δ function (Field et al., 2006; Hendrich et al., 2008; Tran-Van-Minh & Dolphin, 2010; Wang et al., 1999). It also reduces the response to gabapentinoids in chronic neuropathic pain models (Field et al., 2006) and in experimental epilepsy and anxiety models (Lotarski et al., 2011, 2014; Taylor et al., 2007). The skeletal muscle splice variant of α2δ-1 (+A + BΔC) bound to 3H-gabapentin with high affinity (Lana et al., 2014), in agreement with data from rat skeletal muscle α2δ-1 (Gee et al., 1996). Thus, the lack of effect of gabapentin on skeletal muscle function is not a result of its inability to bind to the skeletal muscle α2δ-1 isoform and could relate instead to the high stability of the calcium channel complex in skeletal muscle.

The First Double Cache Domain in α2δ-1 Has Structural Homology to a Universal Amino Acid Binding Domain

A ubiquitous extracellular double Cache domain has been identified in bacterial chemoreceptors and chemotransducers that contains a simple amino acid recognition motif. In bacteria and archaea this has been found to exclusively bind a variety of amino acids, whose identity and binding affinity depends on the exact residues in the binding pocket (Gumerov et al., 2022). In eukaryotes the same motif is found only in α2δ proteins and Cachd1, and in α2δ-1 the first double Cache domain (consisting of Cache 1 and Cache 2, Fig. 3a) contains the binding site for gabapentin (Gumerov et al., 2022). It has previously been shown that α2δ-1 also binds leucine and isoleucine, and these compete with gabapentin (Brown et al., 1998). Whether endogenous amino acids have a physiological role in binding to α2δ proteins, for example, as positive or negative modulators of trafficking, is yet unknown.

Effect of α2δ Subunits on Ziconotide Binding

The presence of α2δ proteins was found to reduce the on-rate and equilibrium inhibition of the CaV2.2 channel blocking ω-conotoxins (Mould et al., 2004). This study included ω-conotoxin MVIIA or ziconotide, which is licensed for use in neuropathic pain conditions. A recent structural study has identified the mechanism for α2δ interfering with the drug binding site, as ziconotide alters the orientation of α2δ-1 with respect to the channel (Gao et al., 2021).

Other Interactions of α2δ Proteins

The α2δs, being completely extracellular, and with structural domain similarities to other extracellular matrix and cell adhesion proteins (Whittaker & Hynes, 2002), are likely to have interactions with other proteins. For example, they may be associated with proteins involved in ion channel clustering. Indeed they have also variously been shown to either directly interact with or influence the function of, several other proteins, including the trafficking and endocytosis protein LRP1 (Kadurin et al., 2017), the extracellular matrix proteins, α-neurexins (Brockhaus et al., 2018; Tong et al., 2017), and other ion channels, such as BK channels (Zhang et al., 2018). Interactions of α2δ-1 with NMDA receptors (Chen et al., 2018) and certain AMPA receptors (Li et al., 2021) have also been identified. The interaction of α2δ-1 with these glutamate receptors was shown to involve the extreme C-terminus of α2δ-1 which would normally be cleaved off during the process of GPI anchoring in the ER (Davies et al., 2010; Guizzunti & Zurzolo, 2014); thus, any interaction must either occur in the ER and prevent the formation of the GPI-anchored α2δ-1, or the interaction occurs with the cleaved C-terminal GPI signal peptide, which has been shown to be rapidly degraded for other GPI-anchored proteins (Guizzunti & Zurzolo, 2014). This topic was recently reviewed (Dolphin, 2018). Generally, such potential interactions may also serve to co-locate multiple ion channels with neurotransmitter receptors and other signaling proteins within synaptic structures.

Summary and Outlook

Over the last 20 years, our appreciation of the CaV α2δ subunits has changed considerably: from a purely auxiliary channel subunit, which modulates some biophysical channel properties, to a ubiquitous, albeit enigmatic, signaling protein, which serves as an important drug target and regulates synaptic function. This remarkable evolution is based on several scientific developments: first, the study of calcium channels and hence α2δ subunits was conferred progressively from heterologous expression systems into native, differentiated cells and tissues. Second, knockout and mutant mouse models became available, which highlighted the functional importance of α2δ proteins in distinct tissues and physiological functions. Third and particularly owing to the increasingly employed OMICs technologies, our insights into disease associations are expanding at an accelerated speed. Because α2δ proteins have been linked to a variety of diseases, ranging from hormone secretion to neuropsychiatric disorders (and many new disease associations may be identified in future years), pharmacological targeting of α2δ proteins bears a tremendous therapeutic potential. Indeed, gabapentin is used as an anti-epileptic drug, and all gabapentinoids are widely prescribed for treating neuropathic pain. However, before the full theoretical therapeutic potential can be exploited, a number of questions need to be addressed:

  1. 1.

    Several studies have identified and proposed α2δ isoform and even splice variant-specific functions and disease associations. However, the physiological importance of the functional redundancy between different α2δ isoforms is not yet understood. In this context it is noteworthy that the synaptic phenotype in a presynaptic α2δ triple knockout/knockdown model can be rescued by the expression of each brain α2δ isoform (Schӧpf et al., 2021).

  2. 2.

    As discussed above, recent studies propose calcium channel-independent functions of α2δ proteins. However, any experimental condition or pathophysiological mechanism affecting the expression and function of α2δ subunits will also affect calcium channels; hence, the definitive distinction between calcium channel-dependent and -independent functions represents an experimental challenge. Yet, a thorough understanding is a prerequisite for novel therapeutic concepts, for example, for targeting trans-synaptic functions without altering/inhibiting calcium channels.

  3. 3.

    On one hand, OMICs technologies continue to identify novel disease associations for all α2δ isoforms (reviewed in Ablinger et al., 2020). On the other hand, detailed insights into the pathophysiological mechanisms are limited to a few examples, such as the key role of α2δ-1 upregulation in neuropathic pain (reviewed in Dolphin, 2016). The experimental challenge for future years thus lies in elucidating the mechanisms linking the α2δ proteins to disease, both in terms of calcium channel-dependent and -independent functions.

  4. 4.

    Our current understanding about the α2δ isoform and splice variant-specific expression, particularly during development and disease, is limited to tissues and brain regions. Investigating cell-type-specific expression patterns is another prerequisite for ultimately understanding specific functions, such as the establishment of network connectivity (Bikbaev et al., 2020). For example, in brain expression of the retinal α2δ-4 isoform is extremely low, basically negligible (Schlick et al., 2010). Nevertheless, α2δ-4 expression is strongly upregulated during development and disease (Schlick et al., 2010; van Loo et al., 2019). In theory this may be related to an important role of α2δ-4 in a single and rare neuron type. Hence, establishing conditional α2δ knockout mouse models may provide a first step into studying the cell-type-specific expression and function.

  5. 5.

    As discussed above, α2δ proteins are emerging as critical regulators of synapse formation, trans-synaptic signaling, and axonal/synaptic wiring. Future novel therapeutic paradigms affecting the expression and function of α2δ thus may have the potential to modulate synaptic wiring during and after development (Bikbaev et al., 2020; Geisler et al., 2019; Veroniki et al., 2017). Hence, the ethical basis for potentially interfering with neurodevelopment requires a discussion between scientists and affected individuals, for example, people with autism (Sanderson, 2021).

  6. 6.

    Finally, considering the newly discovered synaptic roles of α2δ, the consequences on human behavior (e.g., learning) of the extremely widespread used gabapentinoid drugs need to be more thoroughly investigated. Nevertheless, it should be noted that a meta-analysis of adverse effects of anti-epileptic drugs did not detect negative effects of gabapentin on cognitive development of children exposed during pregnancy or breastfeeding (Veroniki et al., 2017). Direct effects on synaptic transmission, synaptic wiring, and postsynaptic receptor clustering may ultimately qualify gabapentinoid drugs to be classified as a psychiatric medication. Because these drugs are predominantly prescribed for chronic pain conditions, they also bear a considerable abuse potential, which so far has not been sufficiently investigated (Goins et al., 2021).

Taken together, understanding the multiple functions of α2δ proteins, which are ubiquitously expressed in excitable cells, has come a long way. Future research efforts will be aimed at elucidating pathophysiological mechanisms of associated disorders and may lead to novel therapeutic paradigms.