Structure and function of the β subunit of voltage-gated Ca²⁺ channels

Abstract

The voltage-gated Ca²⁺ channel β subunit (Ca_vβ) is a cytosolic auxiliary subunit that plays an essential role in regulating the surface expression and gating properties of high-voltage activated (HVA) Ca²⁺ channels. It is also crucial for the modulation of HVA Ca²⁺ channels by G proteins, kinases, Ras-related RGK GTPases, and other proteins. There are indications that Ca_vβ may carry out Ca²⁺ channel-independent functions. Ca_vβ knockouts are either non-viable or result in a severe pathophysiology, and mutations in Ca_vβ have been implicated in disease. In this article, we review the structure and various biological functions of Ca_vβ, as well as recent advances.

1. INTRODUCTION

Most voltage-gated ion channels are large protein complexes composed of a pore-forming subunit and one or more auxiliary subunits that regulate channel properties. Unlike the majority of voltage-gated channels, high voltage-activated (HVA) Ca_v1 and Ca_v2 Ca²⁺ channels absolutely require an auxiliary β subunit (Ca_vβ) for plasma membrane expression and proper gating [1]. Ca_vβ was first purified as part of the complex of skeletal muscle voltage-gated Ca²⁺ channels and was cloned in 1989 [2-3]. Subsequent cloning efforts revealed four subfamilies of Ca_vβs (β₁-β₄), encoded by four distinct genes and each with splice variants. In addition, early studies determined that Ca_vβ binds with high affinity to the pore-forming α₁ subunit (Ca_vα₁) of voltage-gated calcium channels (VGCCs). The high-affinity site is located in the cytoplasmic linker connecting the first two of the four homologous repeats of Ca_vα₁ (i.e., the I-II linker) and was named the α-interaction domain or AID (Fig. 1) [4-6]. Several crystal structures of Ca_vβ have been solved, providing great insights into the molecular mechanisms of Ca_vβ function. A comprehensive review on Ca_vβ is available [7]. Here, we highlight the most important features of Ca_vβ structure, function, and involvement in cell physiology and pathophysiology.

Figure 1.

VGCC topology and the structure of the Ca_vβ core in complex with the AID. (A) Schematic representation of the predicted transmembrane topology of the α₁ subunit of VGCC. The AID, marked in red, is located on the I-II linker. ‘P’ indicates the pore loops, located between transmembrane regions S5 and S6. ‘+’ indicates the charged amino acids in S4 – the voltage sensor. (B) Amino acid sequence alignment of the AID from the indicated calcium channel α₁ subunits. Residues involved in interactions with Ca_vβ are marked in red, with the most critical residues underlined. Residue numbers are indicated on both sides of the sequence. (C) Ca_vβ is organized into 5 regions represented schematically in the upper panel. The lower panel shows the crystal structure of the Ca_vβ₃ core in complex with the AID (PDB accession code 1VYT) with the following regions: N-terminus (light blue), the SH3 domain (gold), part of the HOOK region (purple, residues 121-169), and the GK domain (green). Residues 137-166 of the HOOK region were disordered and are not included. The AID region of Ca_v1.2 (residues 422 to 446) is colored in orange.

2. STRUCTURE OF Ca_vβ

Based on amino acid sequence alignment, biochemical and functional studies, and molecular modeling, it became clear that Ca_vβ consists of a conserved core region flanked by non-conserved N- and C-termini (Fig. 1) [8-10]. The core region is composed of two highly conserved regions homologous to the Src homology 3 (SH3) and guanylate kinase (GK) domains, connected by a weakly conserved HOOK region. This SH3-HOOK-GK core can recapitulate many key functions of Ca_vβ [8, 11-17]. In 2004, three independent research groups reported the crystal structures of the Ca_vβ core region of β_2a, β₃ and β₄, alone or in complex with the AID [11, 18-19]. The structures confirmed the existence of an SH3-HOOK-GK module in the Ca_vβ core.

2.1. The N- and C-termini

The amino and carboxyl termini of Ca_vβ (abbreviated as Ca_vβ-NT and Ca_vβ-CT) are highly variable. They have not been crystallized, but an NMR structure has been obtained for the N-terminus of β₄. This structure reveals a fold consisting of two α-helices and two anti-parallel β-sheets [20]. There are currently no structures of Ca_vβ-CT.

2.2 The GK domain

Guanylate kinases catalyze the formation of ADP and GDP from ATP and GMP. The general structural features of guanylate kinases [21-22] are preserved in the Ca_vβ GK domain (Ca_vβ-GK), but structural variations exist in the catalytic site, and many key catalytic residues are absent in Ca_vβ-GK [11, 18-19]. Thus, Ca_vβ-GK is catalytically inactive. Instead, Ca_vβ-GK has evolved into a protein interaction module, binding tightly to Ca_vα₁ through its high-affinity interaction with the AID (Fig. 1) [11, 18-19, 23]. Importantly, a large surface of the GK domain remains free to interact with other proteins, such as RGK GTPases (see review by Colecraft and colleagues in this issue) and BK channels [24].

2.3. The SH3 domain and the HOOK region

Classical SH3 domains mediate specific protein-protein interactions. They have a highly conserved hydrophobic surface - the PxxP-binding site, which binds to PxxP-sequences in target proteins. In general, SH3 domains have a well conserved and compact fold consisting of five sequential β-strands (β_{strand 1-5}) assembled into two orthogonally packed sheets [25]. However, in the Ca_vβ SH3 domain (Ca_vβ-SH3) the last two β-strands are non-continuous, separated by the HOOK region [11, 18-19] so that the SH3 domain has the following primary structure: SH3_{βstrand 1-4}-HOOK-SH3_{βstrand 5} (Fig. 1). The crystal structures show that the PxxP-binding site in Ca_vβ-SH3 is occluded by part of the HOOK region and a long loop connecting two of the four continuous β-sheets. It is conceivable that these two regions move and expose the PxxP-binding site, either when Ca_vβ is bound to full-length Ca_vα₁ and/or when it interacts with other partners. Nevertheless, while binding between Ca_vβ and PxxP-containing proteins, such as dynamin, has been demonstrated [26], the putative PxxP-binding site itself has yet to be implicated.

The HOOK region is variable in length and amino acid sequence among the Ca_vβ subfamilies. In the crystal structures, a large portion of the HOOK is unresolved due to poor electron density, indicating that it has a high degree of flexibility [11, 18-19]. As will be discussed later, the HOOK region plays an important role in regulating channel inactivation.

2.4. The SH3-GK intramolecular interaction

The crystal structures show that the SH3 and GK domains interact intramolecularly [11, 18-19]. This interaction is strong enough that NT-SH3_{βstrand 1-4}-HOOK module and the SH3_{βstrand 5}-GK-CT module can associate biochemically in vitro and reconstitute the functionality of full length Ca_vβs in situ [14-15, 17, 19, 27-29].

It has recently been proposed that the intramolecular SH3-GK interaction can be disrupted by dynamin and replaced by an intermolecular interaction resulting in the dimerization of two Ca_vβs – a proposed mechanism for dynamin-mediated Ca²⁺ channel endocytosis (discussed later) [30].

2.5. The AID-Ca_vβ interaction

All Ca_vβs bind to the 18 amino acid AID in the I-II linker of Ca_v1 and Ca_v2 α₁ subunits (Fig. 1) [5]. The affinity of the AID-Ca_vβ interaction ranges from 2 to 54 nM in vitro [5, 15, 31-38]. Single mutations of several conserved residues in the AID, including Y10, W13 and I14, greatly weaken the AID-Ca_vβ interaction in vitro, and reduce or abolish Ca_vβ-induced stimulation of Ca²⁺ channel current in heterologous expression systems [4, 12, 31, 35-36, 39-44], firmly establishing the role of the AID as the principal anchoring domain for Ca_vβ.

The Ca_vβ crystal structures reveal a big surprise in regard to the region of Ca_vβ that interacts with the AID. A 31-amino acid segment of Ca_vβ, referred to as the β-interacting domain or BID, had been described as the main binding site for the AID [8]. The BID was able to slightly enhance Ca²⁺ channel current and modulate gating [8], and several BID point mutations were able to weaken the Ca_vβ/Ca_vα₁ interaction and reduce BID-stimulated Ca²⁺ channel currents [8, 31]. Thus, it had been generally accepted that Ca_vβ interacted with Ca_vα₁ primarily through the BID. However, crystal structures of different AID-Ca_vβ core complexes reveal that the AID does not even contact the BID [11, 18-19, 23], which is buried inside Ca_vβ. Thus, mutating the BID most likely alters the folding and/or structure of Ca_vβ, which explains the disruptive effect of BID mutations [8, 31]. The current enhancement by the BID peptide, on the other hand, is likely a non-specific effect, since a random peptide had a similar effect [11].

The crystal structures show that the AID binds to a hydrophobic groove in the GK domain termed the AID-binding pocket or ABP (Fig. 1, α-helices 3, 6 and 9 and some of their flanking loops) [11, 19, 35]. The interactions between the AID and the ABP are extensive and predominantly hydrophobic. These interactions account for the nM affinity of the AID-Ca_vβ binding. Functional studies show that mutating two or more key residues in the ABP severely weakens or completely abolishes the AID-Ca_vβ interaction [12-13].

Binding to the AID does not significantly change the Ca_vβ structure, except for some small and localized changes near the ABP. On the other hand, the AID undergoes a dramatic change in secondary structure when it is engulfed by the ABP. When alone, the AID forms a random coil in solution, as shown by circular dichroism spectrum measurements [18]; when bound to Ca_vβ, the AID forms a continuous α-helix, as shown in the crystal structures. Importantly, the 22 amino acid region that links the first S6 segment of Ca_vα₁ (i.e., IS6) to the AID seems to form an α-helix [45]. Thus, a picture emerges that in the presence of Ca_vβ the entire region from IS6 to the AID adopts a continuous α-helical structure. Indeed, two recent crystal structures of a β₂ core in complex with large parts of the I-II linker of Ca_v1 or Ca_v2 channels show a continuous α-helical structure upstream of the AID (towards IS6), albeit with some differences between Ca_v1 and Ca_v2 channels [23]. This rigid structure is crucial for Ca_vβ regulation of Ca²⁺ channel gating, as will be discussed later.

3. THE FUNCTIONS OF Ca_vβ

Ca_vβ regulates multiple aspects of HVA channel physiology including surface expression, degradation, and gating (Fig. 2). Ca_vβ is also critical for the regulation of VGCC by lipids, G-proteins, RGK GTPases (Rem, Rem2, Rad and Gem/Kir - reviewed in this issue by Colecraft and colleagues), kinases, phosphatases, and other signaling proteins (Fig. 2) [for a comprehensive review see 7]. We highlight here the most important functions of Ca_vβ as well as recent advances.

Figure 2.

Major functions of Ca_vβ. (A) Ca_vβ enhances Ca_vα₁ localization to the plasma membrane by preventing Ca_vα₁ degradation and exposing ER export signals on Ca_vα₁. (B) Ca_vβ promotes VGCC gating, resulting in an overall enhancement of current. (C) Ca_vβ interacts with the ryanodine receptor (RYR) in the sarcoplasmic reticulum (SR) of muscle cells and is critical for excitation-contraction coupling. (D) Ca_vβ can be translocated into the nucleus where it may participate in transcriptional regulation. (E) Ca_vβ interacts directly with many intracellular proteins that regulate VGCC function. The strongest of those regulators are RGK proteins, which potently inhibit VGCCs (reviewed in this issue by Colecraft and colleagues, also see [7]). Other partners include ion channels (e.g., BK_Ca and bestrophin), synaptic proteins (e.g., synaptotagmin I and RIMI), and signaling proteins (such as kinases, phosphatases, dynamin, and Ahnak) [for extensive review see 7].

3.1. Membrane targeting of Ca_vβ

Ca_vβs are cytosolic proteins. This is based both on primary sequence analyses [3, 46] and subcellular localization when Ca_vβ is expressed alone, in the absence a Ca_vα₁ [16, 42] (a few important exceptions are discussed below). In the presence of Ca_vα₁, Ca_vβ switches its localization from cytosolic to membrane-bound. This translocation depends on the AID-Ca_vβ interaction. Single point mutations in the AID of Ca_v1.2 that disrupt binding with Ca_vβ abolish both membrane localization and dendritic clustering of Ca_vβ in hippocampal neurons [47].

Some Ca_vβs can independently be localized to the plasma membrane [48-49], most notably β_2a [50-53]. β_2a is tethered to the plasma membrane through dynamic palmitoylation of two cysteines (Cys 3, 4) in its N-terminus [50-53]. However, palmitoylation alone may not be sufficient for membrane localization because implanting the β_2a N-terminus into other Ca_vβs imparts palmitoylation but not membrane localization [51]. Thus, β_2a probably possesses some additional determinants that help target it to the plasma membrane. Importantly, membrane tethering of Ca_vβ coincides with many functional effects, especially slowed inactivation, as we discuss later.

3.2. Ca_vβ is required for normal channel expression

Ca_v1 and Ca_v2 α₁ subunits show little or no surface expression and produce very small or no currents when expressed without auxiliary subunits. Upon the coexpression of Ca β, Ca⁺² v currents are increased by orders of magnitude [9, 54-57] due to enhanced channel surface expression and open probability. The dramatic increase in surface expression of all Ca_v1 and Ca_v2 α₁ subunits can be observed both in native cells [58-60] and in various heterologous expression systems with any of the four subfamilies of Ca_vβ. Thus, Ca_vβ is required for HVA channel surface expression. The increased surface expression is dependent on Ca_vβ binding to the AID, as point mutations in the AID or the ABP that weaken or abolish the AID-Ca_vβ interaction severely reduce or abolish Ca_vβ-stimulated Ca²⁺ channel current [4, 12-13]. The GK domain itself can largely recapitulate the chaperone function of full length Ca_vβs, greatly increasing Ca²⁺ channel surface expression and current in Xenopus oocytes and mammalian cells [12, 61].

It should be noted that some expression systems, such as Xenopus oocytes, have endogenous Ca_vβs that can transport a small number of exogenously expressed Ca_vα₁ to the plasma membrane [62]. This yields small Ca²⁺ channel currents that can be suppressed by antisense oligonucleotides against endogenous Ca_vβ [34, 62].

How does Ca_vβ enhance Ca²⁺ channel surface expression? An early hypothesis was that Ca_vβ shields or disrupts one or more ER retention signals on the I-II linker of Ca_vα₁ [63]. However, several results are inconsistent with this hypothesis: (i) I-II linkers from several different Ca_vα₁ subunits do not cause ER retention of CD8 or CD4 peptides [64-65], (ii) In the absence of Ca_vβ, transplanting the I-II linker of different HVA Ca_vα₁ subunits (Ca_v2.2, Ca_v2.1 and Ca_v1.2) into a T-type channel (Ca_v3.1), which does not require Ca_vβ for its function, causes current upregulation instead of downregulation [45, 66], (iii) several labs implicated regions other than the I-II linker in Ca_vα₁ trafficking [64, 67-71].

These inconsistencies prompted a re-evaluation of the mechanism of Ca_vβ-mediated upregulation of HVA channel expression. In a recent study [66], all of the L-type Ca_v1.2 channel intracellular linkers were systematically transplanted into the T-type channel, individually or in combination. This was followed by careful examination of the linkers’ ER-export and ER-retention properties, in the presence or absence of a Ca_vβ, by monitoring channel surface expression. The results suggest that the I-II linker of Ca_v1.2 has an ER export signal composed of 9 acidic residues downstream of the AID. All other intracellular linkers, including the N- and C-termini, were found to contain overall ER retention signals. Thus, it was proposed that the intracellular regions of Ca_vα₁ form a complex that yields a prevailing ER retention signal, and when Ca_vβ binds to the I-II linker, it orchestrates a switch in the complex such that the ER export signal becomes dominant, enhancing Ca_vα₁ surface expression. In this process, the Ca_vα₁ C-terminus plays an essential role since it is absolutely required (but not sufficient) for Ca_vβ-dependent channel upregulation [66].

A few recent studies have proposed that Ca_vβ increases channel surface expression by preventing Ca_vα₁ ubiquitination and proteasomal degradation [71-73]. In the absence of Ca_vβ, a proteasome inhibitor (MG132) can rescue Ca_vα₁ surface expression. Ca_vβ coexpression, on the other hand, decreased Ca_v1.2 ubiquitination and association with proteins involved in proteasomal degradation, suggesting that Ca_vβ could be rerouting channels away from predestined proteasomal degradation. This mechanism was proposed for Ca_v2.2 channels [72]. However, Cav2.1 channels do not appear to be subject to this type of regulation [71]. For an extensive review on VGCC trafficking see [74].

3.3. Ca_vβ regulates Ca²⁺ channel gating

Besides enhancing channel surface expression, Ca_vβ regulates channel gating. Here we describe the effects of Ca_vβ on channel activation and voltage- and Ca²⁺-dependent inactivation. We also discuss a unifying model for the mechanism of Ca_vβ-mediated gating regulation. The effects of Ca_vβ on Ca²⁺ channel facilitation have been reviewed previously [7].

3.3.1 Ca_vβ enhances channel activation

All Ca_vβs facilitate channel opening by shifting the voltage-dependence of channel activation by ~10-15 mV to more hyperpolarized voltages [75-77]. This is reflected as an increase in the open probability at the single channel level [78-79], with β_2a producing the most dramatic increase in channel open probability [79-81]. Ca_vβ also often speeds channel activation [1, 82], which is observed as a shortened latency to first channel opening in single channel recordings [83-84]. Many of these effects can be reconstituted by the core region of Ca_vβ [12] and, in some cases, the GK domain alone [85].

3.3.2 Ca_vβ enhances inactivation, except β_2a

Calcium channels inactivate in a voltage- and Ca²⁺-dependent manner (VDI and CDI respectively). This process is modulated by Ca_vβ in at least 3 ways: (1) Ca_vβ generally shifts the voltage-dependence of inactivation to more hyperpolarized voltages by ~10-20 mV (Fig. 3), enhancing VDI. Similarly, Ca_vβ increases CDI [86]. β_2a, however, shifts the voltage-dependence of inactivation to more depolarized voltages by ~10 mV, reducing VDI [12, 87]. (2) Ca_vβ (except β_2a) promotes Ca_v2 channels’ ‘closed state’ inactivation [88-89]. (3) Ca_vβ generally accelerates inactivation kinetics, but β_2a and β_2e decelerate inactivation kinetics (Fig. 3).

Figure 3.

Modulation of HVA Ca²⁺ channel gating by Ca_vβ. (A) Activation: Ca_vβ shifts the current-voltage curve (left panel) and the activation curve (right panel) to more hyperpolarized voltages. (B) Inactivation: Ca_vβ shifts the voltage dependence of inactivation to more hyperpolarized voltages, except β_2a, which shifts it to more depolarized voltages (left panel). All Ca_vβ subunits speed the kinetics of inactivation, except β_2a, which slows the kinetics of channel inactivation (right panel). All traces are schematic representations.

The unique effects of β_2a on inactivation are largely due to its palmitoylation [90-91], but it seems that membrane anchorage rather than palmitoylation per se is critical [91-92]. Indeed, the non-palmitoylated but membrane-attached β_2e has properties similar to β_2a [49].

The molecular determinants on Ca_vβ that regulate channel inactivation are many. The N-terminus is clearly important, as indicated by the observation that natural β₂ and β₄ splice variants differing in their N-termini exhibit markedly different effects on VDI [49, 93-95]. The C-terminus, on the other hand, seems to play a limited or no role in regulating VDI [12, 96]. All Ca_vβ-GK domains, including that of β_2a, speed VDI and hyperpolarize the voltage-dependence of VDI [12, 61, 97]. The HOOK domain is also important as swapping the HOOK between the core regions (SH3-HOOK-GK) of β_1b and β_2a also swaps their effects on VDI [12].

3.3.3. How does Ca_vβ regulate voltage-dependent activation and inactivation of HVA channels?

To answer this question, it is necessary to mention the molecular determinants of voltage-dependent activation (VDA) and inactivation (VDI) of VGCCs. For VDA, they include (i) the external pore and the ion selectivity filter formed by the pore loop between the S5 and S6 transmembrane segments of Ca_vα₁ [98-101], (ii) the inner pore formed by all four S6 segments of Ca_vα₁ [102], and (iii) the activation gate, located at the cytoplasmic end of the S6 segments [103].

The molecular determinants of VDI include the cytosolic ends of the S6 segments, the I-II linker – considered to be the inactivation gate, and the N- and C-termini of Ca_vα₁ [for review, see 86, 104, 105].

Thus, the S6 segments are critical for both VDA and VDI. Notably, the AID, to which Ca_vβ binds, is connected to the IS6 segment through a short linker. Based on extensive studies [12, 18, 29, 35, 45, 55, 86, 106-107], a unified model for Ca_vβ regulation of VDA and VDI of VGCCs has emerged. First, when Ca_vβ is bound to the AID, the entire region starting with IS6 to the end of the AID becomes a continuous α-helix [18, 23, 45, 86]. This rigid structure allows Ca_vβ to regulate both activation and inactivation, most likely by changing the energetics of voltage-dependent movement of both IS6 and the inactivation gate. When the integrity of this rigid α-helix is disrupted by the insertion of glycine residues, the ability of Ca_vβ to regulate VDA, VDI and CDI are severely compromised, while Ca²⁺ channel surface expression remains unaffected [13, 86, 107]. These results underscore the essential role of a rigid IS6-AID linker in Ca_vβ regulation of VGCC gating. Further supporting this notion, the GK domain, which is the minimal part of Ca_vβ that can bind to the AID and presumably induce the formation of the rigid α-helix, can significantly impact (but not entirely normalize) activation and inactivation [12, 61, 97].

Second, and equally important for the model, the anchoring of Ca_vβ to Ca_vα₁ through the AID-GK interaction enables the formation of intrinsically low-affinity interactions between Ca_vβ and other parts of Ca_vα₁ that fully normalize channel gating [reviewed in 7].

An additional factor important for Ca_vβ regulation of gating is the orientation of Ca_vβ relative to Ca_vα₁ [13, 107-108]. Insertions or deletions in the IS6-AID linker, which are expected to maintain the α-helical structure of the linker but induce a 180^o rotation of Ca_vβ with respect to Ca_vα₁, diminish Ca_vβ regulation of activation and inactivation [13, 107]. These studies are consistent with the notion that additional contacts between Ca_vβ and Ca_vα₁ besides the AID-GK domain interaction are critical for VGCC gating.

4. Ca_vβ STOICHIOMETRY WITH Ca_vα₁

4.1. Ca_vα₁ and Ca_vβ are paired with a 1:1 stoichiometry

Early biochemical studies suggest that skeletal and neuronal VGCCs contain a single Ca_vα₁ and a single Ca_vβ [109-110]. Extensive recent studies indicate that this is indeed the case and that the 1:1 stoichiometry is determined by the AID-GK domain interaction. (i) Covalently linked Ca_v1.2 and Ca_vβ_2b (Ca_v1.2-Ca_vβ_2b) have the same gating properties as channels formed by the coexpression of Ca_v1.2 and β_2b. Moreover, when β_2a, which slows inactivation, is coexpressed with Ca_v1.2-β_2b, gating properties remain unchanged [111]. (ii) The simultaneous coexpression of β_2a and β₃ with Ca_vα₁ yields two biophysically distinct channel populations, rather than a single population with intermediate biophysical properties [92, 112]. (iii) The crystal structures of the AID-Ca_vβ core complexes clearly show that each Ca_vβ binds a single AID [11, 18-19, 23], and mutations of key residues in the AID or the ABP abolish both Ca_vβ-mediated Ca²⁺ channel surface expression and gating modulation [4, 12-13, 39, 44].

4.2. Dimerization of Ca_vβ

Several recent studies suggested that Ca_vβ fragments can associate to form GK-GK [85, 113] or SH3-SH3 domain dimers [30]. While the molecular mechanism of GK-GK domain dimerization is unclear [113], SH3 domain dimerization seems to depend on a cysteine residue that participates in forming an SH3-SH3 domain disulfide bond [30].

In addition to fragment dimerization, full-length Ca_vβ dimerization and oligomerization has been proposed, including homodimerization for β₃ and β_2a, and heterodimerization between β₃ and other Ca_vβs [30, 113]. Higher order Ca_vβ oligomers (3 or more Ca_vβs) have also been reported, based on limited data from co-immunoprecipitation and native gel analyses under reducing and non-reducing conditions [113]. The molecular mechanisms of full length Ca_vβ oligomerization are unknown. Mutating the cysteine that holds together the SH3 dimer disrupts SH3-SH3 dimerization but fails to prevent full length Ca_vβ_2a from dimerizing [30]. Similarly, mutations that can individually disrupt GK fragment dimerization fail to prevent full-length Ca_vβ₃ dimerization [113]. It is possible that both the GK and SH3 domains lend residues for Ca_vβ dimerization.

It was recently reported that Ca_vβ dimerization is critical for dynamin-mediated channel internalization [26, 30]. However, this was shown for Shaker K⁺ channels and Ca²⁺ channels with a deleted AID, while WT Cav1.2 channels prevent dynamin-mediated internalization. The steps in dynamin-mediate internalization are unclear but may involve the formation of a quaternary complex between two Ca_vβs and two dynamin molecules [30]. Interestingly, the formation of Ca_vβ dimers is suppressed in the presence of WT Ca_vα₁ [30], consistent with a 1:1 stoichiometry of the Ca_vα₁/Ca_vβ complex.

5. THE ROLE OF Ca_vβ IN G_βγ INHIBITION OF Ca_v2 CHANNELS

Ca_v2 channels are susceptible to several kinds of inhibition by hormones and neurotransmitters through the activation of G-protein coupled receptors (see Currie and Zamponi review in this issue). The most prominent type of inhibition is the membrane-delimited, voltage-dependent inhibition mediated by the direct binding of G protein G_βγ subunits to the channel’s α₁ subunit [114-115].

5.1. Ca_vβ is required for voltage-dependent G_βγ inhibition

Many studies indicate that Ca_vβ is essential for G_βγ-dependent channel inhibition. In COS-7 cells, G protein inhibition of N-type Ca²⁺ channels was markedly enhanced by coexpressed Ca_vβs [116]; in tsA-201 cells, a point mutation in the AID of Ca_v2.2 (W391A) that disrupted Ca_vβ binding abolished voltage-dependent G protein inhibition [43]. In a recent study that directly tested whether Ca_vβ is required for G_βγ inhibition [13], large populations of Ca²⁺ channels devoid of Ca_vβ were obtained by washing away a mutant Ca_vβ that was loosely bound to the AID but was still able to chaperone channels to the membrane. Such β-less channels were still inhibited by purified G_βγ protein applied to the cytoplasmic side of the macropatch; however, all the hallmarks of voltage-dependent inhibition were absent [13, 117-118]. When Ca_vβ was supplied, G_βγ inhibition became voltage-dependent [13]. These results suggest that in the absence of Ca_vβ, G_βγ can bind the channel and inhibit it in a voltage-independent manner [13, 43]. They also suggest that under physiological conditions, Ca_vβ remains bound to the channel during G_βγ inhibition, enhancing the dissociation of G_βγ from the channels and giving rise to the voltage-dependence of inhibition [119]. There is further evidence supporting the notion that Ca_vβ remains associated with Ca_vα₁ during G_βγ modulation. (1) Different Ca_vβs have different effects on voltage-dependent G_βγ inhibition, with β_2a being the least effective in promoting this inhibition [61, 120-123]. In addition, the efficacy of the four Ca_vβs to increase the rate of G_βγ dissociation from the channel is different [121-122]. (2) VDA and VDI, which are significantly modulated by Ca_vβ, remain unchanged before, during, and after G_βγ modulation [13, 117, 123].

5.2. An allosteric model for the voltage-dependent G-protein inhibition of VGCC

An allosteric model was recently proposed for the origin of the voltage-dependence of G_βγ inhibition of Ca_v2 channels [13]. There are several components in this model. First, although the G_βγ-binding pocket in the holo-channel is still unknown, it is likely formed by several regions including the I-II linker, the N-terminus, and the C-terminus of Ca_vα₁ [124]. Second, binding of Ca_vβ transforms IS6 and a large portion of the I-II linker, including the AID, into an α-helix. This allows movements in IS6, following a depolarization, to be efficiently transduced to dismantle the G_βγ binding pocket, causing G_βγ dissociation. In the absence of Ca_vβ, the AID is a random coil [18] and IS6 movements cannot be efficiently transmitted to the I-II linker. Thus, G_βγ stays on the channel, inhibiting it with no voltage dependence. Corroborating this model, mutations that disrupt the rigid α-helix encompassing IS6 and the AID, abolish the voltage dependence of G_βγ inhibition in the presence of Ca_vβ [13, 61].

5. ROLE OF Ca β IN THE REGULATION OF Ca²⁺ v CHANNELS BY PIP2 AND ARACHIDONIC ACID

5.1. β_2a dampens the inhibitory effect of PIP₂ depletion

Phosphatidylinositol-4, 5-biphosphate (PIP₂),a membrane phospholipid composed of two long fatty acid chains attached to a phosphoinositol head group, is necessary for the maintenance of HVA currents [125-128], and PIP₂ depletion following Gq-coupled receptor stimulation results in voltage-independent inhibition of HVA channels [92, 125, 129-130]. Recently, it was demonstrated that the coexpression of β_2a with Ca_v2.1, Ca_v2.2 and Ca_v1.3 channels can largely prevent channel inhibition upon PIP₂ depletion [92]. This effect was the direct result of β_2a palmitoylation since preventing palmitoylation abolished channel protection from PIP₂ depletion. Moreover, imparting palmitoylation to β₃, by fusion to an unrelated palmitoylated peptide, enabled the modified β₃ to protect the channels from PIP₂ depletion [92]. The proposed molecular mechanism for β_2a‘s action is that the two β_2a palmitoyl groups, which are long fatty acid chains, can stabilize Ca²⁺ channels by substituting for PIP₂. Although the PIP₂ binding site on VGCCs is unknown, it was proposed to be ‘bidentate’ – one region binds the PIP₂ fatty acid chains, and the other binds to the PIP₂ head group. When both sites are occupied, the channel is ‘stretched’ in a more active conformation. It was further suggested that β_2a can engage both sites to maintain channel activity even in the absence of PIP₂ [92].

5.2. β_2a suppresses channel inhibition by arachidonic acid

Many phospholipids, including PIP₂, can be metabolized by lipases to arachidonic acid (AA), an unsaturated fatty acid without a head group. The accumulation of AA inhibits HVA Ca²⁺ channels [131-133]. It was recently suggested that this inhibition was the result of occupying only a single site within the bidentate lipid binding site on the channels [92]. The inhibitory action of AA on Ca_v1.3 is attenuated in the presence of β_2a, but not other Ca_vβs [134-135]. This dampening effect critically depends on β_2a palmitoylation per se, rather than membrane anchorage. Thus, the palmitoyl groups of β_2a can both compete with AA to prevent VGCC inhibition, and also substitute for PIP₂ and protect channels from PIP₂ depletion, as discussed above. Both actions likely occur via the same bidentate lipid binding site on VGCCs [92, 134, 136]. Finally, the competition of the β_2a palmitoyl groups with AA can be prevented by manipulating the IS6-AID linker to change the orientation of Ca_vβ in relation to Ca_vα₁ [108], suggesting that β_2a palmitoyl groups have a precise binding site on the channel, likely the same site where PIP₂ and AA bind.

6. Ca_vβ MAY HAVE TRANSCRIPTIONAL ACTIVITY

Several short Ca_vβ isoforms have been cloned that do not contain a GK domain [83, 137-140]. In the first such report, a short Ca_vβ that lacks 90% of the GK domain and the entire C-terminus was cloned form chicken brain and named β_4c [140]. This cβ_4c has almost no effect on Ca_v2.1 channels expressed in Xenopus oocytes but it can dose-dependently attenuate the repressor function of heterochromatin protein 1 (HP1), a chromatin organizer. These findings suggest that cβ_4c may function as a transcription regulator. Similar results have been recently reported for a human β_4c isoform found in the nuclei of vestibular neurons [141]. Full length Ca_vβ has also been implicated in transcriptional regulation. β₃, for example, can directly interact with and suppress the transcriptional activity of Pax6(S), in vitro [142]. In HEK 293T cells, coexpression of β₃ and Pax6(S) results in the translocation of β₃ from the cytoplasm to the nucleus. Several other studies have shown nuclear targeting of Ca_vβ in native cells [79, 143-144]. It remains to be determined which specific genes are the targets of Ca_vβ transcriptional regulation. It is also unclear whether nuclear targeting of Ca_vβ is correlated with VGCC activity; one recent study suggests that L-type Ca²⁺ channel activity diminishes the nuclear targeting of β_4b in the cerebellum [143].

7. Ca_vβ knockouts and pathophysiology

Because of the essential role of Ca_vβ to enable the surface expression and functional modulation of HVA Ca²⁺ channels, Ca_vβ mutations have been implicated in human disease. Ca_vβ knockout animals and mutants have severe phenotypes that are in some cases lethal. Ultimately, what determines the phenotypical outcome of a knockout mouse is the ability of the remaining Ca_vβs to compensate for the lost functions.

7.1. β₁

Ca_vβ₁ is expressed in brain, heart, skeletal muscle, spleen, T-cells and other tissues [reviewed in 7]. The β_1a isoform, however, is exclusively expressed in skeletal muscle where it partners with skeletal muscle Ca_v1.1, and is irreplaceable for excitation-contraction coupling. Thus, β₁ knockout mice, similar to Ca_v1.1 knockouts, are born motionless and die immediately from asphyxiation [75]. A similar phenotype is observed in zebrafish [145].

Paradoxically, the increased expression of β_1a in aging mice was recently proposed to cause skeletal muscle weakness due to decreased levels of Ca_v1.1 channel expression [146]. Knockdown of β₁ in aging mice could increase muscle force and Cav1.1 expression levels to those observed in young mice. This is a surprising effect considering Ca_vβ normally enhances VGCC expression, as is the case in young mice [146]. It remains to be determined what age-related factors turn Ca_vβ from a positive to a negative regulator of Ca_v1.1 expression.

7.2. β₂

Ca_vβ₂ and its various splice variants are expressed in brain, heart, lung, nerve endings at the neuro-muscular junction, T-cells, osteoblasts and other tissues [7]. It is also the predominant Ca_vβ in the heart, especially Ca_vβ_2b [79]. β₂ knockouts die prenatally at embryonic day 10.5 due to lack of cardiac contractions [147-148]. Interestingly, when β₂ expression is restored to the heart of β₂-knockout animals using a cardiac muscle-specific promoter, the animals survive but are deaf due to several deficiencies in the inner ear, including a dramatic reduction in the expression of Ca_v1.3 channels [147, 149]. These ‘rescued’ mice also have defects in vision with a phenotype similar to human patients with congenital stationary night blindness [150].

It is not clear whether β₂ is essential only during certain stages of development or throughout life. A recent study in which the β₂ gene was conditionally knocked out in adult mouse cardiomyocytes gave unanticipated results [151]. Peak calcium currents were reduced by only ~ 30% and the mice had no obvious impairment, suggesting that β₂ may be more critical for the developing than the adult heart [151].

Two point mutations in β_2b have been implicated in cardiovascular human diseases. The S481L mutation, which occurs in the C-terminus of β_2b, contributes to a type of sudden death syndrome characterized by a short QT interval and an elevated ST-segment [152]. The other mutation, in the β_2b N-terminus (T11I), causes accelerated inactivation of cardiac L-type channels and is linked to the Brugada syndrome [153].

7.3. β₃

Ca_vβ₃ knockouts are viable [154-155], with reduced perception of inflammatory pain but unaltered mechanically or thermally induced pain. This is likely the result of reduced N-type calcium channel expression in dorsal root ganglia [77]. Ca_vβ₃ knockouts also have abnormally high insulin secretion at high blood glucose concentrations [156]. A high salt diet causes abnormally elevated blood pressure, a reduction in plasma catecholamine levels, and a hypertrophy of heart and aortic smooth muscle [155, 157]. These results point to a compromised sympathetic control in β₃ knockout mice, likely due to reduced N- and L-type channel activity [154]. Behaviorally, β₃-null mice exhibit impaired working memory, but some forms of hippocampus-dependent learning are enhanced [158-159]. Knockout mice also have lower anxiety, increased aggression, and increased nighttime activity [159]. Finally, both β₃ and β₄ knockout mice have abnormal T-cell signaling, revealing an unanticipated function of Ca_vβs in the immune system [160].

A recent study comparing epileptic patients with non-epileptic individuals identified 3 mutations in β₃ that were present only in patients; however, it is difficult to conclude whether these mutations are the cause of epilepsy [161].

7.4. β₄

Lethargic mice are naturally occurring β₄ knockouts [162-164]. They have an insertion that causes exon skipping and a premature stop codon in the gene for β₄. These mice exhibit ataxia, seizures, absence epilepsy, and paroxysmal dyskinesia [162, 165-166]. Some of this phenotype is contributed by a 50% upregulation of T-type Ca²⁺ channels in thalamic neurons [167]. Other characteristics of lethargic mice include reduced excitatory neurotransmission in the thalamus [168] and a modified electro-oculogram [169]. Interestingly, and perhaps indicative of Ca_vβ functions that are independent of VGCC, β₄ knockouts have aberrant splenic and thymic involution [163-164] and renal cysts [163]. Similar to β₃ knockouts, CD4⁺ T-cells have attenuated receptor-mediated Ca²⁺ responses [160].

In humans, an R468Q mutation in the gene encoding β₄ is associated with a history of febrile seizures, presumably due to the enhancement of Ca_v2.1 currents [170]. In addition, a juvenile myoclonic epilepsy patient was found to have a truncated β₄ (R482x) [171]. Interestingly, two families with different disease histories - one with episodic ataxia and the other with generalized epilepsy and praxis-induced seizures - share the same mutation in β₄ (C104F) [171], highlighting the importance of genetic background in disease penetrance.

8. FUTURE DIRECTIONS

Past studies have provided great insights into the structure, function, and physiology of Ca_vβ. They have also opened new avenues for future research and prompted many intriguing questions that remain to be answered, some of which we highlight below.

1. Knockout mice provide a useful tool in the study of Ca_vβ physiology. Conditional and tissue-specific knockouts are particularly useful as they circumvent compensatory mechanisms that are triggered in conventional knockout mice. This advantage was highlighted in several recent studies that found unexpected mouse phenotypes when Ca_vβ was conditionally knocked out [147, 150-151]. We anticipate that future conditional knockouts would similarly provide great insights into the physiological functions of Ca_vβ and the significance of its diversity.

2. Ca_vβ has functions that are independent of VGCC [reviewed in 7]. Most notable is a possible role in transcriptional regulation, which has been demonstrated for both short splice variants and full-length Ca_vβ. It is not clear, however, how and which tissues benefit from such regulation under physiological conditions. It would also be interesting to know which signaling events trigger the nuclear translocation of Ca_vβ.

3. There are currently two prevalent hypothesis that explain the role of Ca_vβ in VGCC trafficking: (1) Ca_vβ alters the balance of ER retention and ER export signals on Ca_vα₁ in favor of the latter, and (2) Ca_vβ protects channels from predestined proteasomal degradation. These two hypotheses are not mutually exclusive but their relationship remains unclear. It also remains to be determined whether the proposed mechanisms of Ca_vβ-mediated channel trafficking apply to all Ca_v1 and Ca_v2 channels.

4. The crystal structures of the Ca_vβ core have provided great insights into the molecular mechanisms of Ca_vβ function. It would be of great interest to obtain high-resolution structures of full length Ca_vβ and of Ca_vβ in complex with its various interacting partners, such as RGK proteins, dynamin or the ryanodine receptor. Such structures would shed significant light on how different cellular signals converge on Ca_vβ to regulate VGCC function.

5. Ca_vβs have many interacting partners (Fig. 2, also reviewed in [7]), some of which significantly impact the function of VGCCs. The recent findings that Ca_vβ interacts with synaptic proteins [172-173] uncover new roles for Ca_vβ in the organization of the synaptic vesicle release machinery and reveal a new facet of VGCC physiology. We anticipate that the search and study of new Ca_vβ partners will continue to be an interesting and productive area of research in the future.

Highlights.

• The Ca²⁺ channel β subunit is required for voltage-gated Ca²⁺ channel function.

• We review the structure and many functions of Ca_vβ

• Ca β also has functions independent of Ca²⁺ v channels

• Ca_vβ knockouts have severe defficienceis and Ca_vβ is implicated in human disease.

• We provide a perspective for future research in this field.