Regulation of voltage-dependent calcium channels by RGK proteins

Abstract

RGK proteins belong to the Ras superfamily of monomeric G-proteins, and currently include four members– Rad, Rem, Rem2, and Gem/Kir. RGK proteins are broadly expressed, and are the most potent known intracellular inhibitors of high-voltage-activated Ca²⁺ (Ca_V1 and Ca_V2) channels. Here, we review and discuss the evidence in the literature regarding the functional mechanisms, structural determinants, physiological role, and potential practical applications of RGK-mediated inhibition of Ca_V1/Ca_V2 channels.

1. INTRODUCTION

RGK (Rad, Rem, Rem2, Gem/Kir) proteins are a four-member subfamily of the Ras superfamily of monomeric G-proteins. Rad was first discovered as a protein over-expressed in skeletal muscle of type II diabetic humans [1]; Gem as a mitogen-induced gene in human T cells [2]; Rem was originally cloned using a degenerate PCR strategy based on similarity to Rad and Gem [3]; and Rem2 was cloned from a rat brain cDNA library [4]. Functionally, individual RGKs have been linked to diverse functions in different cell types and tissues including (but not limited to): promotion of cell shape remodeling via regulation of cytoskeletal dynamics (Gem) [5-9]; induction of apoptosis in cardiac myocytes (Rad) [10]; regulation of synapse development and dendritic morphology (Rem2) [11, 12]; control of neuronal proliferation and apoptosis during embryogenesis, and survival of human embryonic stem cells (Rem2) [13, 14].

All RGK proteins powerfully inhibit high-voltage-activated Ca²⁺ (Ca_V1/Ca_V2) channels [15-17]. In this review, we discuss the experimental evidence that underlies current understanding of the mechanisms and structural determinants underlying RGK regulation of Ca_V1/Ca_V2 channels, its potential physiological role, and the state of efforts to exploit this channel regulation for practical applications.

2. BASIC STRUCTURE-FUNCTION OF RGK PROTEINS

Similar to other Ras superfamily proteins, RGKs contain a guanine nucleotide binding domain (G-domain) [18-23]. In comparison to Ras, RGK proteins have relatively large N– and C–termini extensions, and non-conservative substitutions in the G-domain of residues critical for GTP binding and hydrolysis [1-4] (Fig. 1). Within the RGK family, the N-termini extensions have variable lengths and display low sequence conservation. The functional significance of the N-terminus extensions in RGKs is unknown. The C-terminus extensions consist of a distal conserved region (~40 residues) separated from the G-domain by a relatively short (12 – 22 residues) variable linker sequence. The C-termini of RGK proteins lack the CAAX (C= cysteine, A= aliphatic, X= any amino acid) prenylation motif that is common to many Ras superfamily proteins, and directs their anchoring to membranes [18]. Nevertheless, RGKs target to the plasma membrane using basic and hydrophobic residues in their C-terminus extensions [24, 25]. The membrane-targeting region in the C-termini of RGK proteins overlaps with a calmodulin (CaM) binding domain (CBD) that mediates RGK interactions with Ca²⁺-CaM [26-29] (Fig. 1A). 14-3-3 proteins bind as dimers to all four RGKs, and this requires phosphorylation of two distinct serines in the N- and C-termini extensions, respectively [6, 26-28, 30] (Fig.1A). The functional significance of CaM and 14-3-3 binding to RGKs is unknown, though it has been suggested that these interactions may regulate the subcellular localization or stability of RGK proteins [6, 27, 28].

Figure 1. Structural features of RGK proteins.

(A) Sequence alignment of human RGK proteins and H-Ras. Identical residues in RGKs are shaded green, and similar residues shaded in cyan. ★ residues homologous to Ras^S17; ★ hydrophobic residues important for CaM binding and membrane targeting; ★serine residues important for 14-3-3 binding. (B) Crystal structures of GTP-bound H-Ras (PDB ID: 5P21) and GNP-bound Rem2 G-domain (PDB ID: 3Q85).

The G-domains of all RGKs have been functionally demonstrated to be bona fide guanine nucleotide binding proteins, although there may be quantitative differences within the family [1-4, 22, 23]. For example, Rad displays higher affinity (K_d ≈ 100 nM) for guanine nucleotides compared to Gem (K_d ≈ 1 – 20 μM) [23]. Moreover, Gem displays a 5- to 10-fold higher affinity for GDP vs GTP, whereas Rad displays no such preference [23]. Crystal structures of RGK proteins confirm that their G-domains adopt a fold comprised of a six-stranded β-sheet surrounded by five α-helices [20-22], similar to that found in other Ras superfamily members [18, 19] (Fig. 1B). In Ras superfamily proteins, five loops within the G-domain (G1 – G5) are highly conserved and form the guanine nucleotide-binding site (Fig. 1). The G1-box (or P-loop) with consensus sequence GXXXXGKS/T is conserved in RGKs, and similar to Ras, engages in interactions that co-ordinate the α/β phosphates and Mg²⁺ [20-22]. Similarly, G4 (NKXD) and G5 (ETSA) motifs are conserved in RGKs and participate in recognition of the guanine base. In Ras, G2 (XTX) and G3 (DXAG) motifs reside within regions referred to as Switch I and Switch II, respectively. In Ras-GDP, Switch I and Switch II are disordered, suggesting a high flexibility of these regions. The Thr³⁵ and Gly⁶⁰ residues in Ras G2 and G3, respectively, act as sensors for the γ phosphate of GTP. Hence, in Ras, GTP binding leads to a conformational change in which switches I and II are stabilized [31]. Since RGKs do not have the equivalent of Thr³⁵ in switch I and show non-conservative substitutions in G3 (DXWE instead of DXAG), it has long been suggested that they may not display the canonical GTP-regulated switch mechanism evident in most Ras superfamily G-proteins [31]. Indeed, crystal structures of Rad and Rem2 bound to a non-hydrolyzable GTP analog showed little structural changes from the GDP-bound forms, with no evidence of a GTP-mediated stabilization of switch I and II [23]. A caveat that must be mentioned is that all RGK protein crystal structures to date are of truncated proteins in which the N- and C-termini extensions have been removed (Fig. 1B). In some other G-proteins such as Ran and Arf, the canonical switch mechanism is modified by additional conformation changes in C- and N-termini extensions, respectively [31]. To this point, a crystal structure of GDP-bound Gem that includes the initial part of the C-terminus extension has been reported [22]. In this structure, the proximal C-terminus extension forms a helix that contacts with an interswitch region between the β2 and β3 strands of the G-domain, as well as the α5 helix. Moreover, in functional assays, the presence of the N- and C-termini extensions in Gem increases its GTPase activity by 20-fold [22].

The rate of intrinsic GTP hydrolysis has been measured for Rad [23, 32] and Gem [22, 23]. Purified Rad has a low rate of intrinsic GTPase activity which can be dramatically increased by cytosolic fractions derived from different tissues [32]. Sub-fractionation of liver cytosol led to the identification of nucleoside diphosphate kinase (nm23 or NDPK) as a GTPase activating protein (GAP) for Rad and Gem (but not Rem) [33]. In an unusual twist, nm23 also acted as a functional guanine nucleotide exchange factor (GEF) for Gem due to its diphosphate kinase catalyzing the transfer of phosphate from ATP to a bound GDP in Gem [33]. To date, nm23 is the only known GAP (and GEF) for any RGK protein.

3. BASIC STRUCTURE-FUNCTION OF Ca_V CHANNELS

Ca²⁺ influx through voltage-dependent Ca²⁺ (Ca_V) channels plays a critical role in various biological functions including muscle contraction, synaptic transmission, hormone secretion, and gene expression [34, 35]. Ca_V channels are divided into two main families based on their threshold for activation: low-voltage-activated (LVA) Ca²⁺ channels and high-voltage-activated (HVA) Ca²⁺ channels. There are three types of LVA Ca²⁺ channels (T-type: Ca_V3.1-3.3) [36], and seven types of HVA Ca²⁺ channels (L-type: Ca_V1.1–1.4; P/Q-type: Ca_V2.1; N-type: Ca_V2.2; R-type: Ca_V2.3) [34, 37].

HVA Ca_V channels are hetero-multimeric proteins comprised of pore-forming α₁ subunits and auxiliary β, α₂δ, and sometimes γ subunits. So far, seven genes encoding HVA α₁ subunits (Ca_V1.1 [α_1S]; Ca_V1.2 [α_1C]; Ca_V1.3 [α_1D]; Ca_V1.4 [α_1F]; Ca_V2.1 [α_1A]; Ca_V2.2 [α_1B]; Ca_V2.3 [α_1E]) [37], four genes encoding Ca_Vβ (Ca_Vβ1-4), and four genes encoding α₂δ (α₂δ1-4) have been identified [34]. Pore-forming α₁ subunits are all comprised of four homologous domains (I–IV) each with six transmembrane segments (S1–S6). Ca_Vβs are required for efficient targeting of α₁ subunits to the plasma membrane [38-42], enhancing channel open probability (P_o) [41], normalizing the voltage-dependence of channel activation [43-46], and modulating inactivation [43, 47-50]. Ca_Vβs bind with high affinity to a conserved 18-residue sequence (the α interaction domain, or AID) located in the loop connecting domains I and II (I-II loop) of α₁ subunits [51-53].

4. RGK INHIBITION OF VOLTAGE-DEPENDENT CALCIUM CHANNELS

4.1. Discovery and basic properties

The separate fields of RGK proteins and Ca_V channels intersected when a yeast two hybrid screen of insulin-secreting MIN6 cells identified Gem/Kir as a Ca_Vβ₃-binding protein [15]. Co-expression of recombinant Ca_V1.3 or Ca_V1.2 channels with Gem in Xenopus oocytes resulted in a complete and constitutive inhibition of both channel types [15]. Subsequently, it was found that the ability to inhibit Ca_V1.2 channels applied to all RGK proteins [16, 54]. Since these seminal studies, RGK proteins have been shown to indiscriminately and potently inhibit all high-voltage-activated channels tested including, Ca_V1.1 [55], Ca_V1.2 [17, 26, 28, 54, 56-60], Ca_V1.3 [15], Ca_V2.1 [15, 61, 62], and Ca_V2.2 [15, 56, 63]. By contrast, low-voltage-activated T-type (Ca_V3.1 – Ca_V3.3) channels are unaffected by RGK proteins [16, 63]. RGK inhibition of Ca_V1 and Ca_V2 channels takes place in all cell types studied to date, including heterologous expression systems (e.g. HEK cells, Xenopus oocytes), cell lines containing endogenous Ca_V channels (e.g. PC12 cells, MIN6 cells), and primary cells (heart, skeletal muscle and neurons).

4.2. RGKs use multiple mechanisms to inhibit Ca_V channels

The question of how RGKs inhibit Ca_V1 and Ca_V2 channels has been intensely studied by several groups. The whole-cell calcium current (I_Ca) is related to microscopic channel properties by the relation: I_Ca = N × F_A × i × P_o, where N is the total number of channels in the surface membrane, F_A is the fraction of activatable channels, i is the unitary current amplitude, and P_o is the single-channel open probability. In principle, RGKs could inhibit I_Ca by reducing any one of the four parameters or a combination of them. To address whether RGKs reduce N, Beguin and colleagues used a Ca_V1.2 pore-forming α_1C subunit harboring a hemagluttinin (HA) epitope tag in an extracellular loop. Combining immunofluorescence and confocal microscopy, they found that all four RGKs prevent surface expression of HA-tagged Ca_V1.2 channels reconstituted in either PC12 or HEK 293 cells [15, 26, 27]. By contrast, using a surface biotinylation/Western blot detection approach, Finlin et al, found that Rem2 inhibited I_Ca in mouse insulinoma MIN6 cells without reducing the number of endogenous Ca_V1.2 channels at the membrane [54]. Similarly, Ikeda and colleagues found that Rem2 inhibits Ca_V2.2 channels stably expressed in tsA201 cells without decreasing channel surface density as determined by radio-labeled ω-conotoxin GVIA binding assays [63]. These two latter reports suggested that Rem2 inhibited Ca_V1.2 and Ca_V2.2 channels directly at the cell surface, although the exact mechanisms were not investigated.

By combining optical detection of surface epitope-tagged α_1C subunits with quantum dot and high throughput flow cytometry measurements, our group discovered that Rem partially decreases (by 60%) the surface density of recombinant Ca_V1.2 channels reconstituted in HEK 293 cells [57]. This decrease was completely prevented by co-expressing dominant negative dynamin, suggesting that Rem increased the rate of dynamin-dependent Ca_V1.2 endocytosis, rather than interfered with forward trafficking of the channel [57] (Fig. 2, mechanism I). Interestingly, even when Rem-induced decrease in channel surface density was completely reversed with dominant negative dynamin, the inhibition of I_Ca was not rescued, suggesting that Rem could also block the activity of surface channels. We distinguished two separate mechanisms Rem used to block Ca_V1.2 channels at the cell surface. First, Rem could decrease I_Ca by diminishing channel P_o without accompanying reductions in voltage sensor movement (Fig. 2, mechanism II). Targeting the Rem G-domain to the membrane either by the Rem C-terminus or a generic membrane targeting module is sufficient to reconstitute mechanism II [57]. Second, we discovered that Rem reduced maximal gating charge (Q_max) of Ca_V1.2 channels in a manner that was not accounted for by a decrease in N. This result suggested that Rem partially immobilizes Ca_V1.2 channel voltage sensors (Fig. 2, mechanism III). On the assumption that all four voltage sensors are required to move for the channel to open, the decreased Q_max suggests that Rem reduces the fraction of activatable (F_A) Ca_V1.2 channels on the cell surface. Overall, this study established that within the same experimental system Rem utilized at least three separable mechanisms to inhibit recombinant Ca_V1.2 channels [57, 64].

Figure 2. Rem inhibits Ca_V1.2 channels using multiple mechanisms.

(A) Left, exemplar whole-cell currents from HEK 293 cells expressing Ca_V1.2 α_1C + β_2a. Right, representative α_1C + β_2a channel currents in the presence of Rem. (B) Rem inhibits I_Ca using three independent mechanisms: by reducing channel surface density via enhanced dynamin-dependent endocytosis (mechanism I); by reducing channel P_o independently of channel voltage sensor movement (mechanism II); by partially immobilizing channel voltage sensors as reported by a decrease in maximal gating charge (Q_max; mechanism III). Mechanisms I and II persist with Rem^T94N, suggesting they do not require GTP to be bound to the Rem nucleotide binding domain (represented as a red hexagon). By contrast, mechanism III is selectively lost with Rem^T94N suggesting a requirement for GTP binding to Rem (represented as a green hexagon). Figure modified from [57] and [17].

Several groups have reported that over-expressing distinct RGKs in cardiac myocytes or skeletal myotubes dramatically decreases endogenous I_Ca,L [55, 58-60]. In three of these studies the impact of RGKs on Q_max was also measured. Murata et al [60] found that over-expressing Gem in adult guinea pig ventricular myocytes profoundly inhibited Ca_V1.2 channel Q_max (70% reduction). By contrast, over-expressing Rem in guinea pig ventricular myocytes [59] or skeletal myotubes [55] yielded smaller reductions in Ca_V1.2 (33% reduction) and Ca_V1.1 (44% reduction) Q_max, respectively. RGK mediated decrease in Ca_V1.1/Ca_V1.2 channel Q_max could result from a decrease in the surface density of channels or an ability of RGKs to partially immobilize channel voltage sensors. The impact of RGKs on I_Ca,L in cardiac myocytes has often been interpreted to reflect a decrease in the surface density of Ca_V1.2 channels [58, 60, 65]. However, we discovered that acute treatment of control and Rem-over-expressing myocytes with the Ca_V1.2 channel agonist, BAY K 8644, eliminated the inhibitory effect of Rem and resulted in I_Ca,L of comparable amplitude between the two conditions [59]. This result not only demonstrated that Rem-inhibited I_Ca,L can be rescued pharmacologically, but also indicated that the majority of Ca_V1.2 channels must be present at the cell surface in cardiac myocytes over-expressing Rem. This contrasts with the finding that Rem significantly reduces the surface density of recombinant Ca_V1.2 channels reconstituted in PC12 or HEK 293 cells [17, 57], suggesting that although RGKs can inhibit Ca_V1.2 channels using multiple distinct mechanisms, only particular subsets of these may be available and used in different cellular contexts.

5. STRUCTURAL DETERMINANTS ON RGKs IMPORTANT FOR I_Ca INHIBITION

5.1. Role of the RGK C-terminus

Much work has focused on defining the important structural elements on RGKs that mediate their potent inhibition of Ca_V1/Ca_V2 channels. Deleting the N-terminus extension from different RGKs does not abolish their ability to inhibit I_Ca [61-63], suggesting that this non-conserved feature is not critical for this effect. By contrast, several investigators have demonstrated that deleting the distal C-terminus of individual RGKs generates truncated versions (e.g. Rem₂₆₅, Rad₂₇₆, Gem₂₆₄) that do not block I_Ca [16, 56, 57, 59, 62, 63, 66]. Hence, the distal C-terminus of RGKs is clearly critical for the mechanism of Ca_V channel block. Nevertheless, the precise manner in which the C-termini of RGKs participate in I_Ca inhibition is not completely clear, though the available evidence suggests multiple mechanisms may be in play. Ambiguities arise in part due to the overlapping roles of RGK C-termini– as membrane-targeting modules, as CaM binding sites, and as bearers of nuclear localization signals (NLS) (Fig. 1A).

5.2. Is RGK C-terminus sufficient to inhibit I_Ca?

Several investigators have sought to determine whether the C-terminus of different RGK proteins is sufficient to inhibit I_Ca with mixed results. Ikeda and colleagues found that over-expressing the C-terminus of Rem2 in SCG neurons did not block endogenous Ca_V2.2 channels, whereas full-length Rem2 strongly inhibited I_Ca [63]. Similarly, expression of the Rem distal C-terminus (final 32 residues) did not inhibit recombinant Ca_V1.2 channels in tsA201 cells [66]. By contrast, Leyris et al found that the final 75 residues of Gem, which includes the entire C-terminus extension, was sufficient to fully inhibit recombinant Ca_V2.1 channels reconstituted in Xenopus oocytes [67]. Recently, a 12-amino acid peptide containing residues K265–K276 in the Gem C-terminus was shown to be sufficient to acutely inhibit Ca_V2.1 channels expressed in Xenopus oocytes, albeit at high concentrations [62]. In this study, the Gem K265–K276 peptide was directly applied to Ca_V2.1 in the inside-out patch configuration, explicitly demonstrating that the inhibitory effect was achieved at the level of surface channels. Remarkably, a scrambled Gem K265–K276 peptide, which contained the same amino acid residues as the original but in different positions, also effectively inhibited Ca_V2.1 in inside-out patches [62], suggesting that the amino acid content of this region rather than its sequence was important for channel inhibition. The region corresponding to Gem residues K265-K276 is quite well conserved amongst RGK proteins (Fig. 1A). Therefore, the inability of Rem2 and Rem C-termini to inhibit Ca_V2.2 channels in SCG neurons [63] and Ca_V1.2 in tsA201 cells [66], respectively, may indicate that the sufficiency of this region to block I_Ca, as observed with Gem K265–K276 inhibition of Ca_V2.1 channels [62, 67] may not be a general property. Alternatively, one possible explanation for the discrepancies could be that the effective concentration of the expressed C-terminal fragment was different in these studies-higher in oocyte studies and lower in cell lines. Further work is needed to clarify this issue, as well as the mechanism K265-K276 peptide uses to decrease Ca_V2.1.

5.3 Role of RGK membrane targeting

In many cell types, RGK proteins autonomously target to the inner leaflet of the plasma membrane via their C-termini interacting with membrane phosphatidylinositol lipids [24, 25, 56, 63, 66]. Deleting the distal C-terminus of RGKs eliminates both their ability to inhibit I_Ca and their membrane targeting [56, 63, 66]. The potential importance of RGK membrane targeting in the mechanism of I_Ca inhibition has been investigated by several groups. Replacing the C-termini of either Rem2 [63] or Rem [66] with the tail of K-Ras4B which consists of a polybasic region and a CAAX prenylation motif, restored both membrane targeting and inhibition of Ca_V2.2 and Ca_V1.2 channels, respectively. For Rem2, the polarity of the membrane targeting domain appeared to be important, as attaching the first 10 residues from Gα_i1, which contains myristoylation and palmitoylation motifs, to the N-terminus of truncated Rem2 (Rem2ΔC) restored membrane targeting but not inhibition of Ca_V2.2 channels [63]. By contrast, attaching a palmitoylated peptide to the N-terminus of truncated Rem (Rem₂₆₅) restored both membrane targeting and inhibition of Ca_V2.2 channels stably expressed in tsA201 cells [56]. Overall, the available data suggest that one way the RGK C-terminus participates in I_Ca inhibition depends on its ability to target the RGK G-domain to the plasma membrane. We exploited this feature to create a small-molecule-inducible Ca_V1/Ca_V2 channel inhibitor by fusing the C1 domain from protein kinase Cγ to either the N- or C-terminus of YFP-Rem₂₆₅ [17, 56, 57]. The resulting constructs, C1_PKCγ-YFP-Rem₂₆₅ or YFP-Rem₂₆₅-C1_PKCγ, are cytosolic when expressed in HEK 293 cells, but are rapidly recruited to the membrane upon exposure to phorbol-12,13-dibutyrate (PdBu) [56, 57]. Inhibition of Ca_V1.2 and Ca_V2.2 channels occurs concomitantly with the dynamic translocation of C1_PKCγ-YFP-Rem₂₆₅ or YFP-Rem₂₆₅-C1_PKCγ to the plasma membrane [17, 56, 57]. These molecules were termed genetically encoded molecules for inducibly inactivating Ca_V channels (GEMIICCs) [56]. GEMIICCs acutely inhibited I_Ca without affecting Q_max, suggesting they lowered current solely by decreasing channel P_o [56, 57] (Fig. 2, mechanism II). Beyond providing insights into the mechanism by which membrane targeting of RGKs results in I_Ca inhibition, GEMIICCs provide a proof-of-concept that new functionalities (in this case inducible inhibition) can be engineered into RGKs.

5.4. Role of CaM binding and nuclear localization of RGKs

In addition to mediating targeting to the membrane, the C-terminus of RGKs binds CaM and possesses nuclear localization signals (Fig. 1A). A useful mutation widely used in the field converts a hydrophobic residue in the RGK C-terminus (corresponding to Gem^W269 Rem^L271 Rad^L281 Rem2^L317) to glycine (Fig. 1A). The effects of this mutation are complex since it diminishes CaM binding [26-28], prevents membrane targeting [57], and dramatically increases nuclear localization of RGKs [26-28, 57]. Hence, the functional consequences of this mutation in RGKs need to be interpreted carefully. Gem^W269G consistently shows a diminished ability to inhibit I_Ca compared to wild-type Gem. This effect has been demonstrated for Gem^W269G inhibition of Ca_V1.2 channels in PC12 cells [15, 28] and cardiac myocytes [60], and Ca_V2.2 channels in SCG neurons [6]. Similarly, Rad^L281G displayed a diminished ability to inhibit Ca_V1.2 channels in PC12 cells, whereas Rem^L271G retained full inhibitory activity in the same system [26]. The precise reason for the functional differences between Gem^W269G/Rad^L281G and Rem^L271G is not clear. Because these mutations markedly increase the nuclear localization of the respective RGK protein, it was suggested that nuclear sequestration of Ca_Vβ by Rad^L281G and Rem^L271G could represent a mechanism to regulate surface Ca_V1.2 channels [26]. An ambiguity with these experiments is that though the mutant RGKs are enriched in the nucleus, a significant portion remains in the cytosol. Hence, in the case of Rem^L271G it was unclear whether inhibition of Ca_V1.2 channels was achieved via the nuclear or cytosolic pools of the mutant protein. We examined this question by examining the effect of Rem^L271G on Ca_V1.2 channels reconstituted in HEK 293 cells [57]. YFP-Rem^L271G in HEK cells was present in both the nucleus and cytosol and caused a partial inhibition of I_Ca,L when compared to wild type Rem. Attaching a nuclear localization signal (NLS) to YFP-Rem^L271G localized it exclusively to the nucleus. Interestingly, NLS-YFP-Rem^L271G was completely inert with respect to I_Ca,L inhibition. Conversely, attaching a nuclear export signal (NES) to YFP-Rem^L271G targeted it exclusively to the cytosol, and NES-YFP-Rem^L271G completely blocked I_Ca,L, explicitly demonstrating that the cytosolic pool is the active component for channel inhibition. These results support three consequential conclusions. First, the ineffectiveness of Gem^W269G and Rad^L281G to inhibit I_Ca may be because their nuclear localization reduces their active concentration in the cytosol. This interpretation is consistent with the finding that RGK inhibition of I_Ca is dose-dependent [68]. Second, that CaM binding is not necessary for RGK inhibition of I_Ca. Third, that membrane targeting of RGKs is not an absolute requirement for I_Ca inhibition, since the non-membrane-targeted NES-YFP-Rem^L271G is an effective blocker of Ca_V1.2 channels.

5.5 Role of nucleotide binding and hydrolysis in RGK inhibition of I_Ca

A canonical feature of Ras superfamily G-proteins is that they function as guanine nucleotide-regulated molecular switches, cycling between inactive GDP-bound and active GTP-bound conformations [18]. There is tremendous ambiguity as to whether and how this basic defining property plays a role in RGK regulation of Ca_V1/Ca_V2 channels. All RGKs have been demonstrated to be bona fide guanine nucleotide binding proteins though there may be quantitative differences among them with respect to relative affinities for GDP and GTP, and the intrinsic rate of GTP hydrolysis. Comparing crystal structures of GTP- and GDP-bound Rad and Rem2 suggest RGKs do not undergo the classical switch mechanism observed in Ras, which involves a GTP-mediated stabilization of otherwise disordered Switch I and Switch II regions [23, 31]. An important approach used to investigate the potential role of GTP binding and hydrolysis of RGKs is to introduce point mutations that decrease their affinities for guanine nucleotides. In Ras, a S17N mutation locks the protein in a GDP-bound state [69]. Ras^S17N has a dominant negative effect on Ras signaling in cells because it has a higher affinity for, and sequesters, Ras-GEFs [69]. The residue corresponding to Ras S17 is conserved in all RGKs (Rad^S105, Gem^S89, Rem^T94, Rem2^S129) (Fig. 1A). Similar to the S17N mutation in Ras, GTP binding is abolished in Rad^S105N and Gem^S89N, although the affinity to GDP is also reduced [23, 32].

Over a series of papers, Beguin and colleagues demonstrated using pull-down assays that Rad^S105N, Gem^S89N, Rem^T94N, and Rem2^S129N displayed decreased binding to Ca_Vβ subunits compared to their wild-type counterparts [26-28]. This was interpreted as indicating that GTP binding was necessary for RGKs to bind Ca_Vβs. However, the impact of these mutations on the ability of the distinct RGKs to inhibit I_Ca was not evaluated in these studies. A different conclusion regarding the role of nucleotide binding in RGKs was reached in a study where the Rem2/Ca_Vβ_2a interaction was unaffected when Rem2 was loaded with either GDP or GTP [24].

A number of studies have focused on evaluating the functional impact of the nucleotide binding state of RGKs on their ability to inhibit I_Ca, with mixed results Chen et al found that Rem2^S129N down-regulated I_Ca to the same extent as wild type Rem2 in sympathetic neurons [63]. However, the same group also found that Gem^S89N lost the ability to inhibit I_Ca in sympathetic neurons [6]. This discrepancy hints at a fundamental difference between these two RGKs with respect to the role of nucleotide binding in I_Ca inhibition in neurons. Rad^S105N has been reported to have no effect on recombinant Ca_V1.2 channels reconstituted in HEK 293 cells, but to exert a dominant negative effect in cardiac myocytes, increasing endogenous I_Ca,L [70]. By contrast, we found that Rem^T94N potently inhibited Ca_V1.2 channels in HEK 293 cells, but with an interesting difference from wild-type Rem. Whereas wild-type Rem significantly decreased gating currents, Rem^T94N blocked I_Ca without impacting Q_max [57]. This result suggested that the ability of Rem to inhibit voltage sensor movement is selectively dependent on GTP binding (Fig. 2, mechanism III). Surprisingly, over-expressing Rem^T94N in heart cells had no impact on endogenous I_Ca,L, in contrast to wild-type Rem which markedly decreased I_Ca,L [59]. Given that Rem^T94N does inhibit Ca_V1.2 channels in HEK 293 cells, this result suggests the existence of a cardiac specific mechanism that inactivates the ability of GDP-bound Rem to inhibit I_Ca,L in heart. A caveat of all the functional studies described here is they rely on mutations predicted to lock RGK proteins in a GDP-bound state. It is possible that these mutations may also have some unanticipated effects that could confound interpretation of results. To circumvent this problem, Chen et al attempted to reverse Rem2 inhibition of I_Ca in sympathetic neurons by dialyzing in GDP-β-S via the patch pipette [63]. They observed no reversal of I_Ca inhibition over 20 minutes of GDP-β-S dialysis, suggesting that GTP binding may not be necessary for Rem2 inhibition of I_Ca in sympathetic neurons.

6. STRUCTURAL DETERMINANTS ON Ca_V1/Ca_V2 CHANNELS IMPORTANT FOR RGK INHIBITION

6.1. Role of auxiliary β subunits in RGK regulation of Ca_V channels

All RGK proteins bind Ca_Vβ subunits [15-17, 26, 56, 71-73]. The affinity of RGK/Ca_Vβ association is about an order of magnitude lower than the interaction of Ca_Vβ with the AID present in the I-II loop of Ca_V1/Ca_V2 pore-forming α₁ subunits [56]. The relatively low affinity of the RGK/Ca_Vβ interaction may explain the lack of a high resolution crystal structure for this complex. Beguin et al. conducted an extensive mutagenesis screen of Ca_Vβs and RGKs to identify mutations that interrupted their mutual interaction without disrupting their global tertiary structures [72]. Mutations on Ca_Vβ that eliminated interaction with RGKs clustered at a hotspot region that was distinct from the α-binding pocket that binds the AID [74-76] (Fig. 3A). Based on results from this mutagenesis screening and the known structures of RGKs and Ca_Vβs, the authors generated a homology model in which Gem was docked to the identified hotspot on Ca_Vβ (Fig. 3B).

Figure 3. Role of Ca_Vβ-binding in Rem inhibition of Ca_V1.2 channels.

(A) Mutations on Ca_Vβ that disrupt binding to RGK proteins aggregate at a hotspot that is separate from the α₁-binding pocket that binds AID. Modified from [72]. (B) Computational model showing docking of Gem to the Ca_Vβ RGK-binding hotspot. Modified from [72]. (C) Co-immunoprecipitation assay demonstrating that a mutated β_2a, β_2aTM, no longer interacts with Rem. (D) Cartoon showing two possible modes of RGK inhibition of Ca_V1.2 channels– β-binding-dependent and α₁-binding-dependent. (E) Exemplar currents from HEK 293 cells expressing α_1C + β_2a in the absence (left) and presence (right) of Rem. (F) Population current-voltage (I-V) relationships for α_1C + β_2a in the absence (■) and presence (▲) of Rem. (G, H) Data for α_1C + β_2aTM ±Rem. Same format as E and F. Figure reproduced from [17].

It was initially suggested that RGKs bind to Ca_Vβs and prevent their interaction with α₁ subunits, thereby compromising their chaperone function and severely limiting channel trafficking to the membrane [15, 26, 77]. However, subsequent work has shown that RGKs do not disrupt the α₁-β interaction, and there is consensus that RGK inhibition of I_Ca involves a ternary α₁/β/RGK complex [24, 54, 56, 57, 63]. Within this ternary complex framework, there are two possible ways in which Ca_Vβ could play a role in RGK inhibition of I_Ca. In the first scenario, a direct RGK/Ca_Vβ interaction is not necessary for I_Ca inhibition. However, Ca_Vβ could play a role by promoting a permissive conformation of the channel complex that is necessary for functional interaction with RGKs. Alternatively, a direct RGK/Ca_Vβ interaction could be obligatory for the mechanism of RGK inhibition of I_Ca. Evidence has been provided for both scenarios, and there are indications that there may be specificity for different RGK/Ca_V channel combinations as discussed below. An important tool that made these advances possible was the discovery of specific mutations in Ca_Vβs that selectively eliminated binding to RGKs without compromising functional regulation of Ca_V α₁ subunit trafficking and gating [17, 61, 72] (Fig. 3C).

Jian Yang’s group used a mutated Ca_Vβ that no longer binds RGKs to demonstrate that direct interaction with Ca_Vβ was not necessary for Gem to acutely inhibit Ca_V2.1 channels in Xenopus oocytes [61]. Nevertheless, they further showed that β binding to α_1A was necessary for Gem to downregulate Ca_V2.1. This latter effect was demonstrated using a β with weakened affinity for α_1A, which could be readily washed off in inside-out patches. With the β washed off, Gem no longer inhibited the channel, though it could still bind α_1A. On the basis of these results, they proposed a model where the presence of β exposes an inhibitory site on the channel complex that is then engaged by Gem to block Ca_V2.1 [53, 61]. The location of this putative inhibitory site on the Ca_V2.1 channel complex that binds Gem is currently unknown.

We investigated whether the RGK/β interaction has any role in the mechanism of I_Ca inhibition, or merely represents an unrelated epiphenomenon, by examining Rem inhibition of Ca_V1.2 channels in HEK 293 cells [17]. We found that Ca_V1.2 channels containing a β_2a mutant that selectively loses binding to RGK proteins (Fig. 3C), are less potently inhibited (74% inhibition) by Rem than channels containing wild-type β_2a (96% inhibition), suggesting the prevalence of both β-binding dependent and independent modes of inhibition [17] (Fig. 3, E-H). We further found that two mechanistic signatures of Rem inhibition of Ca_V1.2 channels (decreased N and P_o), but not a third (reduced Q_max), depend on Rem binding to Ca_Vβ. Surprisingly, we discovered a functional dichotomy amongst the RGKs– while Rem and Rad used both β-binding-dependent and independent mechanisms to inhibit Ca_V1.2, Gem and Rem2 solely utilized a β-binding-dependent method to do so [17] (Fig. 4). These findings may explain why Ca_V1.2 channels expressed in the absence of β subunits are partially blocked by Rem in HEK 293 cells [78], but minimally affected by Gem or Rem2 in Xenopus oocytes [15, 68, 79]. Finally, we found that Rem inhibition of Ca_V2.2 channels was completely dependent on the Rem/Ca_Vβ interaction, in contrast with Ca_V1.2 [17]. Therefore, the mechanisms of RGK inhibition of Ca_V1/Ca_V2 channels are customized at the levels of both the RGK and channel type. Overall, these data suggest a dualistic view for RGK inhibition of Ca_V channels. First, all RGKs can inhibit all Ca_V1/Ca_V2 channels via mechanisms that depend on direct RGK/Ca_Vβ interactions. Because Ca_Vβs are necessary for the formation of all functionally mature Ca_V channels, this could explain the promiscuity of RGKs in blocking all Ca_V1/Ca_V2 channel types. Second, distinct RGKs may initiate β-binding-independent channel inhibition through selective interaction with specific Ca_V1/Ca_V2 channel pore-forming α₁ subunits [17].

Figure 4. Distinct RGKs differentially use β-binding-dependent and β-binding-independent mechanisms to inhibit Ca_V1.2 channels.

(A) Impact of distinct RGKs on wild-type (α_1C+β_2a) and mutant (α_1C+β_2aTM) Ca_V1.2 channels. *, #, $ P < 0.05 when compared to α_1C+β_2a, α_1C+ β_2aTM, or α_1C+β_2a+RGK, respectively, using two-tailed unpaired Student’s t test. (B) Cartoon showing dichotomy in determinants used by distinct RGKs to inhibit Ca_V1.2 channels. Rem can inhibit Ca_V1.2 by both β-binding-dependent and α₁-binding-dependent mechanisms. Gem and Rem2 inhibit Ca_V1.2 solely using a β-binding-dependent mechanism. The stoichiometry of Rem binding to the Ca_V1.2 complex has not been investigated. One possibility is that two Rem proteins can simultaneously associate with a single Ca_V1.2 channel. Alternatively, it is also possible that within a single Ca_V1.2 channel complex, the interaction of Rem with α_1C N-terminus or β subunit is mutually exclusive, thus ensuring a 1:1 binding stoichiometry. Figure reproduced from [17].

6.2. Direct interactions of RGKs with Ca_V α₁ subunits

A logical explanation for the observation that specific RGKs can inhibit Ca_V2.1 and Ca_V1.2 channels without binding to β subunits is that particular RGKs directly interact with individual Ca_V1/Ca_V2 α₁ subunits. Consistent with this idea, Gem co-immunoprecipitates with Ca_V2.1 α_1A in the absence of Ca_Vβ [61]. Nevertheless, the exact Gem binding sites on α_1A that underlie channel inhibition are unknown. By exchanging corresponding fragments between P/Q- (α_1A) and T-type channels (Ca_V3.1 α_1G), Fan et al. showed that the region comprised of the S1-S3 transmembrane segments of domain II conferred Gem sensitivity [61]. However, since Gem is an intracellular protein, it is most likely that IIS1-IIS3 may be involved in the transduction of the effect, rather than being a binding site for Gem [61, 79].

There have also been efforts to identify direct RGK binding sites on Ca_V1.2 α_1C subunit. Pang et al. [80] reported that Rem (as well as Rad and Rem2) directly binds to proximal and distal regions C-terminus of α_1C. The proximal α_1C C-terminus contains structural elements, including a CaM binding domain and an EF hand motif, that are essential for Ca²⁺-dependent regulation (inactivation and facilitation) of Ca_V1.2 [81-86]. Ca²⁺-CaM was found to block the Rem/α_1C C-terminus interaction in vitro. This interplay was not due to Ca²⁺-CaM interaction with the Rem C-terminus, suggesting that Rem directly competes with Ca²⁺-CaM for α_1C C-terminus [80]. Furthermore, co-overexpression of CaM was found to partially relieve Rem inhibition of Ca_V1.2 channels in tsA201 cells. The authors concluded that direct binding to the proximal α_1C C-terminus is important for Rem-mediated regulation of CDI and I_Ca,L blockade in Ca_V1.2 [80]. Results from our group support a different conclusion. Based on the idea that the putative Rem binding site was most likely localized within an intracellular region of the channel, we conducted an unbiased screen for potential Rem interaction sites on α_1C (N- and C-termini, I-II, II-III, and III-IV loops) using three independent approaches (fluorescence resonance energy transfer, co-localization analyses, and co-immunoprecipitation assays) [17]. We found that Rem interacts solely with the α_1C N-terminus in the region immediately upstream of transmembrane segment I in domain I (IS1). Moreover, over-expressing α_1C N-terminus completely rescued Rem-mediated β-binding-independent inhibition of Ca_V1.2 channels, suggesting that Rem/α_1C N-terminus interaction underlies this mode of regulation. Neither Rem2 nor Gem bound α_1C N-terminus, providing an explanation for why they exhibit only a β-binding-dependent mechanism of Ca_V1.2 inhibition. Although the distal N-terminus shows homology among distinct Ca_V1/Ca_V2 α₁-subunits (60% identical residues or conservative substitutions), Rem does not bind Ca_V2.2 α_1B N-terminus [17]. This explains why Rem inhibits Ca_V2.2 channels solely through a β-binding-dependent mechanism.

7. CROSSTALK OF RGK AND PROTEIN KINASE SIGNALING ON Ca_V1.2 CHANNELS

In heart, up-regulation of Ca_V1.2 channels by protein kinase A (PKA) is an important physiological modulation that contributes to sympathetic regulation of the heartbeat, a critical component of the flight-or-fight mechanism [87, 88]. Two reports have documented an interesting cross talk between RGKs and PKA signaling at the level of Ca_V1.2 channels in heart. Specifically, in primary cardiac myocytes over-expressing either Rad [58] or Rem [59], the remaining I_Ca,L was completely insensitive to PKA modulation. By contrast, the Ca_V1.2 channel agonist BAY K 8644 robustly and acutely rescued I_Ca,L in myocytes over-expressing Rem [59]. The mechanism by which RGKs prevent PKA-mediated modulation of Ca_V1.2 channels in heart is unknown.

A recent study found that α₁-adrenergic receptor stimulation could attenuate Rem inhibition of Ca_V1.2 channels in both HEK 293 cells and cardiac myocytes [65]. The mechanism was found to involve activation of protein kinase D1 (PKD1) and subsequent phosphorylation of Rem at residue Ser18. The authors propose that this leads to sequestration of Rem by 14-3-3 proteins and permits Ca_V1.2 channels trapped intracellularly to traffick to the cell surface [65].

8. PRACTICAL APPLICATIONS OF RGK INHIBITION OF Ca_V CHANNELS

Inhibition of Ca_V1/Ca_V2 channels is an important or potential therapy for many cardiovascular and neurological diseases including: hypertension, cardiac arrhythmias, neuropathic pain, Alzheimer’s disease, and Parkinson’s disease [89-92]. Under certain circumstances, intracellular genetically encoded Ca_V channel inhibitors may have advantages over traditional small-molecule blockers [93]. For example, localized expression of RGKs may permit a more restricted inhibition of Ca_V channels, permitting a targeted therapeutic result while minimizing unwanted off-target effects [93]. In a proof of concept demonstration of this principle, focal gene delivery of Gem to the atrioventricular (AV) node slowed AV nodal conduction and reduced the heart rate in a porcine atrial fibrillation model [60].

The potential practical applications of RGKs may be widely expanded if new functionalities such as inducibility and selectivity could be engineered into them. To this end, we have developed engineered derivatives of Rem which are inert but can be acutely activated by small molecules to inhibit I_Ca [56]. Recently, a caveolae-targeted Rem has been developed that selectively inhibits caveolae-localized Cav1.2 channels in heart cells [94]. This molecule selectively inhibits hypertrophic Ca²⁺ signaling pathways in cardiac myocytes without impairing contractility.

9. PHYSIOLOGICAL SIGNIFICANCE OF RGK INHIBITION OF I_Ca

In contrast to the steady progress in evaluating the mechanisms by which RGKs inhibit Ca_V1/Ca_V2 channels, the physiological role of this channel regulation remains somewhat mysterious. Many different excitable cell types co-express RGKs and specific Ca_V1/Ca_V2 channel isoforms. For example, Rad and Rem are expressed in cardiac and skeletal muscle [3, 16, 95], Rem2 is expressed in neurons and pancreatic β-cells [4, 11, 54, 96], and Gem is present in mitogen-activated T cells and pancreatic β cells [2, 15, 97]. This leads to the expectation that RGK inhibition of Ca_V channels plays a physiological role in these cells. However, this is difficult to prove for two main reasons.

First, RGKs have been shown to interact with many other signaling molecules besides Ca_V channel subunits. Therefore, it is difficult to specifically assign any (patho)physiological consequences that arise from knockout of individual RGKs to their effect on Ca_V channels. Knockout mice for Rad [95], Gem [97], and Rem [98] have been generated. Rad knockout mice display an increased susceptibility to transverse aortic constriction (TAC) induced hypertrophy, and down-regulation of Rad is correlated with development of heart failure in humans [95]. I_Ca,L has not been measured in these mice so it is unknown whether this contributes to the observed phenotype. However, deciphering the potential role of I_Ca,L is further complicated by the fact that Rad functionally interacts with CaMKII [29, 95] and Rho kinase [99, 100], two signaling molecules well known to be involved in the development of pathological cardiac hypertrophy [101, 102]. Gem knockout mice displayed glucose intolerance, impaired insulin secretion in response to high glucose, and abnormal Ca²⁺ handling in pancreatic β-cells [97]. I_Ca in β-cells was not measured, precluding any inferences about a possible role of Ca_V channel dysregulation in these abnormalities. Rem knockout mice exhibit a moderately increased I_Ca in cardiac myocytes, without showing any overt cardiac phenotype [98]. Interpretation of the results may be confounded by compensatory mechanisms as well as the fact that other RGKs present in heart may provide a redundant pathway for I_Ca inhibition. To date, the most direct evidence that an RGK may mediate constitutive inhibition of I_Ca in an excitable cell comes from a study which demonstrated that knockdown of Rad with shRNA resulted in an increase in I_Ca,L in cultured cardiac myocytes [58]. Similarly RNAi knockdown of Rem2 in neurons demonstrated a role in synaptic development and dendritic morphology [11, 12], and reduced the frequency of miniature excitatory postsynaptic currents [96]. However, the Rem2 knockdown in hippocampal neurons had no impact on I_Ca, suggesting that other effectors underlie the observed functional effects.

A second reason for the difficulty in evaluating the physiological role of RGK inhibition of Ca_V channels in different systems has to do with the likely constitutive, background nature of this regulation. Development of approaches that can selectively and acutely relieve RGK inhibition of Ca_V channels would appear to be necessary for progress in determining the physiological role of this form of Ca_V1/Ca_V2 channel regulation.

10. CONCLUSION

All RGKs powerfully and promiscuously inhibit all Ca_V1 and Ca_V2 channels. Surprisingly, this seemingly homogenous and simple phenomenon is underlain by a rich variety of mechanisms and structural determinants [57, 68]. The mechanisms and structural determinants of RGK inhibition of Ca_V channels appear to be customized based on the RGK type, Ca_V1/Ca_V2 channel isoform, and cellular context. This may have contributed to apparently contradictory reports in the field. Further work is needed to define the precise mechanisms that different RGKs use to inhibit distinct Ca_V1/Ca_V2 channel isoforms in specific cell types. Nailing down the physiological role and importance of RGK inhibition of Ca_V channels has proven difficult due to the constitutive nature of this inhibition, the fact that RGKs functionally interact with other important signaling molecules, and the existence of compensatory mechanisms and redundant pathways in knockout mice. New tools that permit acute relief of RGK inhibition of Ca_V channels are needed to probe the physiological significance of this phenomenon. Finally, RGKs and engineered derivatives have potential utility as therapeutics and useful molecular tools. Exploring these dimensions of the RGK/Ca_V channel functional interaction is an exciting area for future research.

Highlights.

• RGK proteins inhibit voltage-dependent Ca_V1 and Ca_V2 channels.

• RGKs use multiple mechanisms and determinants to inhibit Ca_V1/Ca_V2 channels.

• The mechanisms of channel inhibition are customized for different RGK/Ca_V channel types.

• New functionalities can be engineered into RGKs for practical applications.