The influx of Ca2+ ions through voltage-dependent calcium (CaV) channels links electrical signals to physiological responses in all excitable cells. Not surprisingly, blocking CaV channel activity is a powerful method to regulate the function of excitable cells, and this is exploited for both physiological and therapeutic benefit. Nevertheless, the full potential for CaV channel inhibition is not being realized by currently available small-molecule blockers or second-messenger modulators due to limitations in targeting them either to defined groups of cells in an organism or to distinct subcellular regions within a single cell. Here, we review early efforts to engineer protein molecule blockers of CaV channels to fill this crucial niche. This technology would greatly expand the toolbox available to physiologists studying the biology of excitable cells at the cellular and systems level.
This article examines the rationale and prospects for developing novel genetically encoded voltage-dependent calcium channel blockers.
Electrical signals, or action potentials, generated by ionic fluxes through ion channel proteins residing in the plasma membranes of cells, constitute one of the most prevalent and important cell-signaling mechanisms in biology. Electrical signals coordinate the activity of the millions of cells required to generate the heartbeat, underlie the orchestrated firing of neurons that enable sight, speech, movement, and formation of memories, and regulate the release of hormones that control glucose homeostasis, growth, and development. Although the spectrum of biological responses dependent on electrical signals is impressively diverse, they all utilize a similar signal transduction paradigm: membrane depolarization leads to the opening of voltage-dependent Ca2+ (CaV) channels, permitting a Ca2+ influx that triggers the appropriate cell biological response. The central role CaV channels play in transducing electrical signals into biological responses position them as attractive targets for potentially regulating a wide range of physiological processes. Indeed, modulation of CaV channels by a variety of second messenger pathways and small molecules is widely exploited as a means to regulate physiology and as a therapy for various diseases. In this review, we will discuss nascent efforts to engineer protein molecules for custom inhibition of CaV channels. As a prelude to in-depth discussion of this topic, we will first briefly review the structure function of CaV channels, traditional CaV channel blockers, and the rationale for developing such novel protein inhibitors of CaV channels.
Structure Function of CaV Channels
CaV channels are divided into two main families depending on their threshold for activation. There are three types of low-voltage-activated (CaV,LVA) or T-type Ca2+ channels (CaV3.1–CaV3.3) encoded by distinct genes (CACNA1G, CACNA1H, and CACNA1I), each with multiple splice variants (93). Functionally, CaV,LVA channels activate at a relatively negative threshold of around − 60 mV, have a small conductance, and rapidly inactivate (111). CaV,LVA channels are found in many excitable cell types including sinoatrial node cells, smooth muscle, and neurons, where they contribute to pacemaking (91). CaV,LVA channels will not be discussed any further in this review.
High-voltage-activated calcium (CaV,HVA) channels currently include seven members (CaV1.1–CaV1.4, CaV2.1–CaV2.3) encoded by distinct genes each with multiple splice variants (19). CaV,HVA channels typically have an activation threshold of around − 30 mV, with the exception being CaV1.3, which activates at a threshold of around − 50 mV (75, 123). Structurally, CaV,HVA channels are hetero-multimeric proteins comprised of a main α1 subunit assembled with auxiliary β and α2δ subunits, calmodulin, and sometimes a γ subunit (FIGURE 1⇓).
CaV,HVA channel α1 subunits are the pore-forming proteins and define the identity of the channel complex. All CaV,HVA channel α1 subunits have a similar architecture, comprised of four homologous domains (I–IV), each with six membrane-spanning segments (S1–S6). The four domains are connected by intracellular loops of varying lengths, along with cytosolic NH2 and COOH termini. The S4 segment of each domain contains positively charged residues that are an integral part of the voltage sensor. The S5–S6 pore loops from each domain collaborate to form the selectivity filter, and the S6 segments line the channel pore (19).
There are four auxiliary CaVβ subunits (β1–β4) encoded by different genes, each with multiple splice variants (17, 18, 31, 92, 98, 101, 110). At the primary sequence level, the different CaVβs display two conserved domains separated by an alternatively spliced linker region and variable NH2 and COOH termini. Crystal structures revealed the conserved core of CaVβs contain src homology 3 (SH3) and guanylate kinase-like (GK) motifs that interact intramolecularly (23, 88, 117). This functional signature suggests a kinship to the membrane-associated guanylate kinase (MAGUK) super-family of scaffold proteins, which all contain an SH3-GK module, and organize intracellular signaling pathways by colocalizing diverse proteins (2, 48). Functionally, CaVβs are necessary for trafficking pore-forming α1 subunits to the plasma membrane, produce depolarizing shifts in the voltage dependence of channel activation, elevate single-channel open probability (Po), and impart characteristic inactivation properties to CaV,HVA channels (24, 29, 63, 70, 86, 92, 105). CaVβs bind with high affinity to α1 subunits using an “α binding pocket” (ABP) formed by noncontiguous residues in the GK motif and a conserved 18-residue sequence, the “α interaction domain” (AID), located in the intracellular loop connecting α1-subunit domains I and II (23, 88, 97, 117). Binding of CaVβ to α1 increases the helical propensity of the region from the AID to the end of IS6, suggesting formation of a rigid helix that spans the AID and IS6 (89). This rigid IS6-AID helix is important for the ability of CaVβs to modulate activation and inactivation gating (42, 118).
There are currently four α2δ subunit types (α2δ-1 to α2δ-4) encoded by different genes (34, 49, 68, 99). The α2δ subunit is generated in cells as a single gene product, which is post-translationally cleaved to generate separate α2 and δ proteins that are held together by disulfide bonds (61). Both the α2 and δ subunits are heavily glycosylated. Topologically, the α2 component is entirely extracellular, whereas the distal part of the δ peptide spans the plasma membrane. Functionally, α2δ subunits typically increase current amplitude by increasing the surface density of α1 subunits and also influence channel activation and inactivation gating (40, 104).
CaV,HVA channels are subject to rich positive and negative feedback regulation by Ca2+ (37). The Ca2+ sensor for the effects of intracellular Ca2+ on CaV,HVA channels is calmodulin (71, 94, 126), which associates in the basal state with an IQ motif in the cytoplasmic COOH terminus of CaV,HVA channel α1 subunits (66, 82, 116). There is also evidence that calmodulin binding is important for trafficking CaV,HVA channel α1 subunits (119).
The first γ subunit identified (α1) was originally isolated as one of the component subunits of CaV,HVA channel purified from skeletal muscle (14). Currently, there are eight members of this protein family, but only a subset has been shown to interact with some CaV,HVA channel α1 subunits (64). Topologically, γ subunits are predicted to have four transmembrane segments with cytoplasmic NH2 and COOH termini. Functionally, they appear to have inhibitory effects on some CaV,HVA channels (64).
Traditional CaV,HVA Channel Blockers
Small molecules that block CaV,HVA channels have historically played a critical role in advancing understanding of the different CaV,HVA channel subtypes and their respective biological functions. Furthermore, small molecule CaV,HVA channel blockers are used pharmacologically as an important therapy for various cardiovascular and neurological diseases (69, 114).
CaV1.1 to CaV1.4 channels, also referred to as L-type channels, are inhibited by three classes of drugs, dihydropyridines, phenylalkylamines, and benzothiazepines, in a state-dependent manner (108). Pharmacological blockade of L-type channels is an important therapy for cardiovascular diseases such as hypertension, angina, and some cardiac arrhythmias (114). The three drug classes inhibit CaV1 channels by binding to partially overlapping residues localized in domains III and IV of the respective pore-forming α1 subunits (53, 57, 95, 102, 108). Drug binding to the CaV1 α1 subunits is believed to couple allosterically to the channel pore and gating machinery (108).
Unlike CaV1 channels, there is a dearth of small organic blockers for CaV2.1–CaV2.3 channels. However, CaV2 channel family members are potently blocked by various peptide toxins isolated from predatory marine snails or spider venom (115). Specifically, CaV2.1 (P/Q-type) channels are selectively blocked by ω-aga-toxin IVA (80, 120), CaV2.2 channels are inhibited by ω-conotoxin GVIA (3, 96), and CaV2.3 channels are repressed by SNX-482 (85, 113). These toxins act by binding the respective pore-forming α1 subunit and either physically occluding the pore or modifying channel gating. Blockade of CaV2 channels is an effective or potential therapy for an assortment of disorders including neuropathic pain, epilepsy, stroke, and neurodegenerative conditions (114).
Gabapentin is efficacious in the treatment of neuropathic pain and seizures (106). Although this drug was originally designed as a γ-aminobutyric acid (GABA) derivative, its analgesic action stems from a high-affinity interaction with the α2δ-1 (and α2δ-2) subunit of CaV,HVA channels (13, 41, 51). Under control physiological conditions, gabapentin has only moderate effects on ICa. However, during nerve injury, there is a marked upregulation of α2δ-1 subunits in dorsal root ganglion (DRG) neurons and the spinal dorsal horn (73). Under this condition, gabapentin acutely inhibits ICa in DRG neurons, and this effect may underlie the analgesic effect (74). Chronic exposure to gabapentin suppresses CaV2.1 channel trafficking to the plasma membrane, and this may also be a contributing mechanism to the therapeutic effects of the drug (54). Overall, gabapentin provides a nice proof-of-concept that it is possible to design efficacious CaV,HVA channel-modulating molecules that target auxiliary subunits rather than the pore-forming α1 subunit (114).
Rationale for Engineering Proteins to Inactivate CaV Channels
There are many potential applications of CaV,HVA channel inhibition that cannot be achieved using the traditional CaV,HVA channel blockers described above. The limitations arise because it is difficult to specifically target these inhibitors to either a select group of cells in a tissue or organ or to CaV,HVA channels localized in spatially distinct regions within a single cell (FIGURE 2⇓). By contrast, these limitations may be overcome with engineered intracellular proteins that block CaV,HVA channels because these have the capacity to be deployed in defined cell types and may also be targeted to spatially distinct subcellular sites using appropriate addressing motifs. To illustrate the potential niches that can be uniquely filled by engineered protein blockers of CaV,HVA channels, we draw on specific examples from neuroscience and cardiac biology, although the potential applications extend to all excitable cells.
Macroscopic neuroscience applications
An important tool for neurophysiologists studying the intricacies of the mammalian brain or the function of neural circuits in model organisms is the ability to functionally eliminate specific neurons in a living animal and observe the resulting behavioral consequences (FIGURE 2A⇑) (77). Various tools have been developed to advance this capability, each with its own set of limitations (77). These include suppressing neuronal excitability by overexpressing potassium (62) or chloride ion channels (72), or a light-gated chloride pump (125), and inactivating synaptic transmission using small-molecule-mediated cross-linking of synaptic vesicle fusion proteins (65). In principle, blocking CaV2 channels should be highly effective in eliminating neurotransmission because synaptic vesicle fusion is steeply dependent on Ca2+ influx via these channels (transmitter release ∝ ICan, where n = 3–5) (12, 30). However, traditional peptide toxin blockers of CaV2 channels cannot be easily targeted to specific neurons in living animals, thus limiting their utility for this purpose. Genetically encoded intra-cellular blockers of CaV,HVA channels have the advantage that they can be expressed in specified neurons using a number of different approaches that have been developed including, utilizing appropriate cis-regulating elements (77).
Microscopic neuroscience applications
Neurons have a highly compartmentalized architecture. A single neuron usually has multiple CaV,HVA channel types that are differentially distributed in different subcellular compartments. A single CaV,HVA channel type expressed in a neuron may mediate different biological responses on Ca2+ influx depending on its subcellular localization within the cell (FIGURE 2B⇑). For example, Ca2+ influx through presynaptic terminus-localized CaV2 channels is the dominant trigger for neurotransmitter release in most neurons (20, 87). However, these same channels regulate neuronal excitability by coupling to Ca2+-activated K+ channels in axons and dendrites (39, 79, 122). Having the capacity to selectively inhibit a particular CaV,HVA channel type located in spatially distinct regions of a neuron would not only advance fundamental understanding of Ca2+ signaling mechanisms in neurons but also permit a more fine-tuned regulation of neuronal activity. Although such micro-scale targeting of CaV,HVA channels in single neurons is not possible with the currently available toxin blockers, it may be possible to achieve this objective using engineered protein inhibitors.
Macroscopic cardiac applications
In heart, CaV1.2 channels are critical for excitation-contraction (EC) coupling, membrane excitability, and conduction velocity through the atrio-ventricular (AV) node. Atrial fibrillation (AF) is a prevalent arrhythmia characterized by rapid and uncoordinated activation of the atria due to reentrant excitation or abnormal impulse formation from ectopic foci (84). A significant portion of the adverse effects associated with AF are due to ventricular tachycardia produced by abnormal activation of the ventricles as a result of the electrical activity in the atria. A treatment for AF is ablation of the AV node to electrically uncouple the atria and ventricles. This treatment is invasive and irreversible, and it has been proposed that inhibiting CaV1.2 channels in the AV node may represent a viable alternative that is reversible (FIGURE 2C⇑) (32). Protein inhibitors of CaV1.2 channels have the advantage over the small organic blockers because they can be focally expressed, and thus specifically targeted, to the AV node (32, 83).
Microscopic cardiac applications
In single ventricular myocytes, most CaV1.2 channels are localized in transverse tubules where they are apposed to nearby clusters of ryanodine receptors (RYR) in the junctional sarcoplasmic reticulum (SR; FIGURE 2D⇑) (16, 50, 103, 109). This spatial arrangement of CaV1.2 channels and RYRs is critical for the calcium-induced calcium release that underlies cardiac EC coupling (15, 38, 107, 121). However, a portion of ventricular CaV1.2 channels are found localized within caveolae (5). It has been hypothesized that caveolae CaV1.2 channels in heart locally activate Ca2+-sensitive molecules that signal to responses other than contraction, including, potentially, cardiac hypertrophy (52, 81). Testing the function of caveolae-localized CaV1.2 channels in heart requires the selective inactivation of this channel pool in single ventricular myocytes. This requirement is beyond the capabilities of organic CaV1.2 channel blockers but may be achievable with appropriately targeted protein inhibitors.
Engineering Proteins for Custom Inhibition of CaV,HVA Channels
Inducible inhibition of CaV1–2 channels is widely exploited as a mechanism to regulate diverse physiological processes in organisms (37, 67). For example, CaV2 channels are inhibited by many G-protein-coupled receptor (GPCR) ligands. In one form of modulation, Gβγ subunits, released from heterotrimeric G-proteins on GPCR activation, bind CaV2 channel α1 subunits and shift them into a reluctant gating mode where they require large depolarizations to open (6, 33, 35, 56, 59). Hallmarks of Gβγ-induced inhibition of CaV2 channels include a slowing of current activation kinetics and relief of inhibition by either large depolarizations or high-frequency action potential waveforms (25, 36, 76). In another paradigm, binding of GPCR ligands inhibit ICa by causing the removal of CaV,HVA channels from the cell surface (1, 112). In principle, the myriad physiological mechanisms that exist to inhibit CaV,HVA channels are potential candidate templates for engineering to create novel derivatives for custom applications. In one approach, the engineering is performed at the level of the GPCR to evolve forms that can be activated by pharmacologically inert ligands (4). In another approach, membrane-tethered ω-conotoxin MVIIA was used to selectively and potently block coexpressed CaV2.2 channels in Xenopus oocytes (58). We will, however, focus the rest of the review on the Rad/Rem/Gem/Kir (RGK) GTPases, which have several features that make them particularly well-suited for this purpose.
The RGK protein family currently consists of four members: Rem, Rem2, Rad, and Gem (mouse homolog also referred to as Kir). These proteins belong to the Ras superfamily of monomeric GTP binding proteins that function as GTP-regulated switches to regulate a wide variety of essential biological processes in cells (26). Structurally, RGK proteins have several unique features that distinguish them from other Ras GTPases including nonconservative substitutions in the GTP binding domain of residues involved in nucleotide binding and hydrolysis, a long NH2-terminus extension that is variable within the family, and a relatively conserved COOH-terminus extension (27, 43, 47, 78, 100). The COOH-terminus extension of RGK GTPases lack the CAAX prenylation motif found in many Ras-like GTPases (26). Nevertheless, the distal COOH-terminus extension effectively targets RGK proteins to the plasma membrane utilizing a combination of electrostatic and hydrophobic interactions involving basic and aromatic (or aliphatic) residues with the membrane (55). The COOH terminus of RGK GTPases binds calmodulin and 14-3-3 proteins in vitro, and these interactions may regulate the subcellular localization of these proteins (8, 9). Functionally, Gem and Rad regulate cytoskeleton remodeling via interactions with Rho kinase (27), siRNA knockdown of Rem2 in neurons inhibits synapse development (90), and loss of Rad in heart leads to heart failure (21).
Cross talk of RGK GTPases with CaV,HVA channels
A yeast two-hybrid screen using CaVβ3 as bait fished out Gem GTPase as an interaction partner (10). Electrophysiological experiments on recombinant channels reconstituted in Xenopus oocytes revealed that Gem effectively eliminated CaV1.2 and CaV1.3 channel currents (10). Subsequently, the property of dramatically inhibiting CaV,HVA channels was found to extend to all members of the RGK GTPase protein family (45). The mechanism of RGK protein inhibition of CaV,HVA channels was initially believed to involve disruption of the α1-β subunit interaction (10). However, there is an emerging consensus that this does not occur and that RGK GTPases rather form a ternary complex with α1 and β subunits to inhibit ICa (11, 44). RGK GTPases inhibit ICa by different mechanisms. In some studies, RGK proteins reduce the surface density of CaV,HVA channels (7–10), although this is not universally found (22, 46). A significant portion of inhibition involves channels that remain on the cell surface but are either held in a low open probability mode or else display immobilized voltage sensors.
Engineering RGK proteins for custom applications
RGK proteins are promising candidates for engineering custom CaV,HVA channel blockers given their extreme potency in inhibiting all CaV,HVA channel types. Indeed, overexpression of wild-type RGK GTPases in tissues can be used to achieve a constitutive block of native CaV,HVA channels. In a nice demonstration of this, viral-mediated expression of Gem in adult heart cells ablated native CaV1.2 channel currents, and its focal delivery to the AV node slowed AV nodal conduction and heart rate in an animal model of atrial fibrillation (83). One limitation of using wild-type RGK GTPases is that the inhibition is constitutive, with no facile way for temporal control of channel block. A second disadvantage is that the magnitude of channel block cannot be easily regulated. Whether these capabilities could be engineered into RGK GTPases provided a crucial initial test of the feasibility of exploiting these proteins for custom applications. An important finding that advanced this possibility is that deleting the distal COOH terminus of RGK GTPases eliminates their ability to block ICa (22, 45, 124). For both Rem and Rem2, deleting the distal COOH terminus results in a redistribution of the GTPase from the plasma membrane to the cytosol. Moreover, constitutively targeting the inactive truncated RGK GTPases to the plasma membrane, using generic membrane-targeting modules, recapitulated their capacity to inhibit ICa (22, 124). This suggested that membrane targeting is essential for the ability of RGK GTPases to inhibit ICa. We took advantage of this feature of RGK GTPases to develop a chemical genetic hybrid approach where Rem derivatives are cytosolic and inactive in the basal state but can be acutely activated to block ICa by induced translocation to the plasma membrane using a small molecule (124). These engineered proteins were termed genetically encoded molecules for inactivating CaV channels (GEMIICCs). The first generation GEMIICC featured the C1 domain from protein kinase Cγ fused to the NH2 terminus of a COOH-terminus-truncated Rem, termed Rem265. When expressed in HEK293 cells, C1PKCγ-YFP-Rem265 was predominantly cytosolic. Application of 1 μM phor-bol-12,13-dibutyrate (PdBu) resulted in a rapid translocation of C1PKCγ-YFP-Rem265 to the plasma membrane. In cells co-expressing recombinant CaV2.2 channels and C1PKCγ-YFP-Rem265, ICa was high in the basal state, reflecting the inability of a cytosol-localized truncated Rem derivative to block CaV,HVA channels. Upon addition of PdBu, ICa decreased concomitantly with translocation of C1PKCγ-YFP-Rem265 to the plasma membrane. Moreover, the magnitude of CaV2.2 channel inhibition was easily regulated simply by varying the concentration of PdBu (IC50 = 56 nM). One surprise with the C1PKCγ-YFP-Rem265 GEMIICC was that it showed selectivity in inhibition. Both CaV2.2 and CaV1.2 channels were significantly inhibited by C1PKCγ-YFP-Rem265 in a PdBu-inducible manner. However, CaV2.1 and CaV2.3 channels were unresponsive. This contrasts to wild-type Rem, which effectively inhibits all CaV,HVA channels. The reason for the selectivity of the GEMI-ICC is unknown but could reflect different geometric constraints for the distinct channel types. Ultimately, the selectivity of C1PKCγ-YFP-Rem265 may prove fortuitous since it suggests that it may be possible to develop GEMIICCs that are specific for individual CaV,HVA channels.
Many of the potential applications of GEMIICCs may rely on their inducible targeting to spatially distinct plasma membrane sites within a single cell. This level of spatial precision is not possible with the C1PKCγ-YFP-Rem265 GEMIICC. As a prelude to developing GEMIICCs that can be targeted with specificity to defined regions of the plasma membrane, we evaluated the feasibility of using a rapamycin-mediated heterodimerization strategy to facilitate membrane translocation of a Rem265 derivative (28). Rapamycin is an immunosuppressant drug that acts by heterodimerizing two proteins, FKBP and the kinase mTOR. To implement the approach, FKBP was fused to the N-terminus of Rem265, and the rapamycin binding domain of mTOR (FRB) was fused constitutively to the plasma membrane using the membrane-targeting module from Lyn kinase (60) (FIGURE 3⇓). In the basal state, cells co-expressing YFP-FKBP-Rem265 (YFR) and Lyn-FRB (LDR) displayed YFP fluorescence in the cytosol. Addition of 1 μM rapamycin caused a rapid translocation of YFR to the plasma membrane, with a concomitant inhibition of co-expressed CaV2.2 channels (FIGURE 3B–E⇓). The successful implementation of the heterodimerization approach greatly advances the prospect of developing GEMIICCs that target CaV,HVA channels with subcellular precision in single excitable cells.
In summary, we have described recent ongoing efforts to develop technologies that permit spatially restricted and temporally regulated inactivation of CaV channels in vitro and in vivo. It is anticipated that this capability would significantly expand the toolkit available to physiologists to probe and manipulate the biology of excitable cells at the cellular and system level. Potential uses of the technology include 1) manipulating neuronal excitability in vivo to evaluate the function of specific neural circuits, or as a treatment for neurological disorders such as epilepsy and stroke that are characterized by excessive neural activity and excitotoxicity; 2) selectively inactivating spatially distinct pools of CaV channels within single heart and neuronal cells to discover novel functional paradigms of CaV channel signaling in these cells; and 3) developing inducible animal models of diseases, such as atrial fibrillation, in which a diminished ICa is an important factor.
- Copyright © 2009 the American Physiological Society