Designer genetically encoded voltage‐dependent calcium channel inhibitors inspired by RGK GTPases

Abstract

High‐voltage‐activated calcium (Ca_V1/Ca_V2) channels translate action potentials into Ca²⁺ influx in excitable cells to control essential biological processes that include; muscle contraction, synaptic transmission, hormone secretion and activity‐dependent regulation of gene expression. Modulation of Ca_V1/Ca_V2 channel activity is a powerful mechanism to regulate physiology, and there are a host of intracellular signalling molecules that tune different aspects of Ca_V channel trafficking and gating for this purpose. Beyond normal physiological regulation, the diverse Ca_V channel modulatory mechanisms may potentially be co‐opted or interfered with for therapeutic benefits. Ca_V1/Ca_V2 channels are potently inhibited by a four‐member sub‐family of Ras‐like GTPases known as RGK (Rad, Rem, Rem2, Gem/Kir) proteins. Understanding the mechanisms by which RGK proteins inhibit Ca_V1/Ca_V2 channels has led to the development of novel genetically encoded Ca_V channel blockers with unique properties; including, chemo‐ and optogenetic control of channel activity, and blocking channels either on the basis of their subcellular localization or by targeting an auxiliary subunit. These genetically encoded Ca_V channel inhibitors have outstanding utility as enabling research tools and potential therapeutics.

RGK proteins are small Ras‐like GTPases that potently inhibit voltage‐gated calcium (CaV) channels by binding their auxiliary b subunits. Mechanistic insights into how RGK proteins inhibit CaV channels has been exploited to develop novel genetically‐encoded CaV channel inhibitors that can be acutely activated by small molecules or light, or produce constitutive inhibition via targeted ubiquitination using CaVb‐binding nanobodies. Advantages of such genetically‐encoded CaV channel inhibitors include their ability to be selectively targeted to specific tissue, cell types, sub‐cellular localization, and distinct CaV channel macromolecular complexes.

Voltage‐gated calcium channels: basic structure, function and regulation

Ca²⁺ is a universal second messenger that regulates numerous biological functions in virtually all cells (Berridge et al. 2000). Cytoplasmic Ca²⁺ in cells is kept low (100 nM) but rises in response to diverse stimuli (to ∼1 μM) to initiate functional responses through the action of a variety of Ca²⁺‐dependent proteins. The source of signalling Ca²⁺ is from either intracellular stores or the extracellular milieu. There are a variety of integral membrane proteins on the plasma membranes of diverse cell types that permit the entry of Ca²⁺ in response to specific stimuli. Amongst these are the family of voltage‐dependent Ca²⁺ channels (VDCCs) which gate Ca²⁺ entry into cells in response to changes in membrane potential. VDCCs are sub‐divided into two categories based on the threshold voltage for activation – low‐voltage‐activated (LVA) and high‐voltage‐activated (HVA) Ca²⁺ channels, respectively. There are three distinct LVA (Ca_V3.1 – Ca_V3.3) and seven HVA (Ca_V1.1 – Ca_V1.4; Ca_V2.1 – Ca_V2.3) (Catterall, 2011; Zamponi et al. 2015). VDCCs play many essential roles in the biology of excitable cells. As examples, Ca²⁺ influx through VDCCs: contributes to pacemaking in many cell types including the sino‐atrial node of the heart and substantia nigra (Ca_V3; Ca_V1.3) (Guzman et al. 2010; Mesirca et al. 2015); regulates neuronal excitability by coupling to Ca²⁺‐activated K⁺ channels (Ca_V1.2; Ca_V2.1; Ca_V2.2) (Marrion & Tavalin, 1998; Womack et al. 2004); controls the heartbeat by coupling electrical excitation to muscle contraction in cardiomyocytes (Ca_V1.2) (Bers, 2002); enables communication among neurons by triggering presynaptic neurotransmitter release (Ca_V2.1‐Ca_V2.3) (Sudhof, 2012); promotes the release of hormones, e.g. insulin, adrenaline (epinephrine), essential for metabolic and physiological homeostasis (Ca_V1.2, Ca_V1.3, Ca_V2) (Braun et al. 2008); and engenders long‐term changes in cellular function by regulating gene expression (Ca_V1.2, Ca_V1.3) (Wheeler et al. 2012).

Functional HVA Ca²⁺ channels in vivo are multi‐subunit complexes comprising distinct pore‐forming α₁ subunits (α_1A – Ca_V2.1; α_1B – Ca_V2.2; α_1C – Ca_V1.2; α_1D – Ca_V1.3; α_1E – Ca_V2.3; α_1F – Ca_V1.4; and α_1S – Ca_V1.1) assembled with calmodulin and auxiliary β (Ca_Vβ₁ – Ca_Vβ₄), α₂δ (α₂δ‐1 – α₂δ‐3), and γ subunits (Zamponi et al. 2015). In heterologous expression studies, co‐expression with Ca_Vβ is necessary for efficient α₁‐subunit trafficking to the plasma membrane (Buraei & Yang, 2010). Consistent with an essential in vivo role, β₁‐null mice die at birth due to asphyxiation (Gregg et al. 1996) and β₂ knock‐out is embryonic lethal due to cardiac defects (Weissgerber et al. 2006). Nevertheless, recent in vivo data in adult cardiomyocytes indicate an exception to the absolute necessity for Ca_Vβ to enable trafficking of α_1C to the surface membrane of adult heart cells. Cardiac‐specific excision of Ca_Vβ₂, the dominant Ca_Vβ isoform in heart, reduced Ca_Vβ₂ protein by 96% while decreasing Ca_V1.2 current amplitude by only 26% (Meissner et al. 2011). Further, a transgenic mouse expressing a dihydropyridine‐resistant α_1C mutant that does not bind Ca_Vβ displayed ample DHP‐resistant Ca_V1.2 current, indicating a robust Ca_Vβ‐independent trafficking to the sarcolemma (Yang et al. 2019). It remains to be determined whether and to what extent Ca_Vβ‐independent trafficking happens in other cell types and other Ca_V1/Ca_V2 isoforms at different developmental stages. Beyond their impact on Ca_V1/Ca_V2 trafficking, Ca_Vβ isoforms alter multiple channel gating properties – shift the voltage dependence of channel activation to the left, increase single channel open probability, impart distinctive rates of inactivation, and endow different steady‐state inactivation profiles (Buraei & Yang, 2010). α₂δ subunits promote surface trafficking and can alter biophysical properties of particular Ca_V1/Ca_V2 channels (Dolphin, 2012). γ subunits are associated with Ca_V1.1 channels (Kang & Campbell, 2003; Wu et al. 2016); their association with other Ca_V1/Ca_V2 channels in vivo is unclear. Multiple CaM binding sites have been described at different locations in distinct Ca_V1/Ca_V2 channels (Van Petegem et al. 2005; Dick et al. 2008; Mori et al. 2008; Ben‐Johny & Yue, 2014). CaM binds to the C‐terminus of most Ca_V1/Ca_V2 channels in a fairly conserved region containing an IQ motif (Erickson et al. 2001; Kim et al. 2008, 2004, 2010; Mori et al. 2008). Binding of apoCaM to this region has been shown to enhance the open probability, P _o, of Ca_V1.3 channels (Adams et al. 2014). Cryo electron microscopy structures of Ca_V1.1 and Ca_V3.1 channels have yielded invaluable insights into Ca_V channel structure, three‐dimensional assembly and modulation by ligands (Wu et al. 2016; Zhao et al. 2019 a, b ).

An important feature of HVA Ca_V channels is that their activity is not static but is dynamically regulated both by stably associated proteins as well as transiently interacting signalling molecules. Typically, these regulatory mechanisms have profound physiological importance; their dysregulation can cause pathology, and they can be co‐opted or interfered with for therapy. Examples of these regulatory mechanisms include: Ca²⁺‐dependent inactivation of Ca_V1.2 channels mediated by preassociated CaM, a negative feedback mechanism which when disrupted leads to prolonged cardiac action potentials and life‐threatening cardiac arrhythmias (Peterson et al. 1999; Zuhlke et al. 1999; Alseikhan et al. 2002); protein kinase A mediated up‐regulation of cardiac Ca_V1.2, essential for the physiologically critical fight‐or‐flight response (Kamp & Hell, 2000); voltage‐dependent inhibition of Ca_V2 channels by G_βγ subunits (Dolphin, 2003), a mechanism for tuning synaptic strength that is important for the analgesic effects of opiates.

RGK GTPase inhibition of Ca_V channels: discovery and mechanisms

The seminal report of the functional interaction between RGK proteins and Ca_V1/Ca_V2 channels was in 2001 – a yeast two‐hybrid screen of MIN6 cells using Ca_Vβ₃ as bait fished out Gem/Kir as an interacting protein (Beguin et al. 2001). Co‐expressing Gem with recombinant Ca_V1.3 or Ca_V1.2 in Xenopus oocytes resulted in a marked inhibition of calcium channel current. Gem was initially discovered as a mitogen‐induced gene in human T cells (Maguire et al. 1994) and belongs to a sub‐family of Ras‐like monomeric G‐proteins with three other members: Rad (Ras associated with diabetes), originally discovered as a protein over‐expressed in skeletal muscle of diabetic patients (Reynet & Kahn, 1993); Rem, first identified using a degenerate cloning strategy based on homology to Gem and Rad (Finlin & Andres, 1997); and Rem 2, cloned from a rat brain cDNA library (Finlin et al. 2000). Subsequent to the original report of Gem inhibition of Ca_V1.2 and Ca_V1.3, it was shown that this phenomenon also extended to Rad and Rem, which both potently inhibited Ca_V1.2 channels (Finlin et al. 2003), and Rem 2 (Chen et al. 2005; Finlin et al. 2005). Over‐expressing any RGK protein markedly suppresses endogenous Ca_V1/Ca_V2 channels in native cells including cardiac myocytes, neurons and skeletal muscle (Murata et al. 2004; Chen et al. 2005; Bannister et al. 2008; Wang et al. 2010; Xu et al. 2010; Puckerin et al. 2018). A recent elegant study revealed that endogenous Rad in cardiomyocytes constitutively exerts a gating brake on a fraction of Ca_V1.2 channels. This inhibition is relieved by protein kinase A phosphorylation of Rad, and is the long sought‐after mechanism by which β‐adrenergic agonists increase cardiac Ca_V1.2 to enhance inotropy during the fight‐or‐flight response (Liu et al. 2020).

How do RGK proteins inhibit Ca_V1/Ca_V2 channels? The answer to this seemingly simple question turned out to be surprisingly complex. The whole‐cell current (I) is related to microscopic channel properties by the relation I = F _A × N × i × P _o; where F _A is the fraction of activatable channels, N is the total number of channels, i is the unitary current amplitude, and P _o is the open probability. In principle, RGK proteins could inhibit I by reducing any of the four parameters or a combination of them. We found that Rem inhibits Ca_V1.2 channels reconstituted in HEK293 cells in at least three distinct ways (Fig. 1) (Yang et al. 2010). First, in this system, Rem reduced Ca_V1.2 surface density (N) by 65%, an effect that was reversed by co‐expressing dominant negative dynamin. The second mechanism involved a reduction in channel P _o, which occurred without an impact on Ca_V1.2 voltage sensor movement, suggesting an impairment in coupling between voltage sensors and opening of the channel pore. This mechanism specifically required simultaneous association of Rem with the plasma membrane (mediated by a polybasic distal C‐terminus) and Ca_Vβ in the channel complex (via the guanine nucleotide binding domain). Finally, a third mechanism entailed a reduction in Ca_V1.2 maximal gating charge (Q _max) that was not accounted for by a change in channel surface density, suggesting an immobilization of one or more voltage sensors. This third mechanism required GTP bound to Rem and would have the practical effect of diminishing both F _A and P _o. While these three mechanisms of Rem inhibition of Ca_V1.2 can be observed in HEK293 cells, their relative prevalence may differ in other cell types. For example, over‐expression of Rem in cardiac myocytes markedly depresses Ca_V1.2 whole‐cell current without an apparent change in channel surface density as indicated by immunofluorescence, and the acute rescue of near‐maximal current with BAYK 8644 (Xu et al. 2010).

Figure 1. Rem inhibition of reconstituted Ca_V1.2 channels.

A, exemplar family of whole‐cell Ba²⁺ currents from recombinant Ca_V1.2 channels (α_1C + β_2a) reconstituted in HEK293 cells either without (left) or with (right) co‐expression of Rem. B, population I‐V curves from Ca_V1.2 channels in the absence (■) or presence () of co‐expressed Rem. C, schematic diagram showing three distinct mechanisms (I‐III) utilized by Rem to inhibit recombinant Ca_V1.2 channels. In mechanism I, co‐expressed Rem results in a decrease in the number of channels at the cell surface (N) due to enhanced Ca_V1.2 endocytosis. Mechanism II involves a reduction in the open probability (P _o) of channels residing on the plasma membrane without impacting on voltage sensor movement as measured by total gating charge (Q _max). This mechanism requires Rem simultaneously binding to the Ca_Vβ subunit (using the guanine nucleotide binding domain) and the plasma membrane (via the polybasic distal C‐terminus). Mechanism III involves an impaired movement of the voltage sensor movement of surface channels as measured by a decreased Q _max (observed even when N is completely rescued by co‐expressing dominant negative dynamin). Mechanism III is blocked by a mutation (T94N) that favours GDP over GTP binding to Rem, suggesting it requires GTP‐bound Rem.

From a macroscopic perspective all four RGKs profoundly inhibit all Ca_V1/Ca_V2 channels when over‐expressed. Nevertheless, underneath this apparent uniformity, there are important distinctions in the mechanisms of inhibition that extend to both the different RGKs as well as to individual channel types (Yang & Colecraft, 2013). Rem2 was found to inhibit Ca_V1.2 channels in mouse insulinoma MIN6 cells (Finlin et al. 2005) and also Ca_V2.2 channels in tsA201 cells without reducing the number of channels at the cell surface (Chen et al. 2005). By reconstituting channels with either wild type Ca_Vβ or a mutant Ca_Vβ that loses binding to RGK proteins, we found that Rem and Rad could inhibit Ca_V1.2 and Ca_V2.2 (but not the other Ca_V1/Ca_V2 channel types) using either β‐binding‐dependent or β‐binding‐independent mechanisms (Yang et al. 2012; Puckerin et al. 2018, 2016). In the particular case of Rem inhibition of Ca_V1.2, the β‐binding‐independent mechanism of inhibition is mediated by an interaction of the Rem distal C‐terminus with the α_1C N‐terminus region just upstream of the first transmembrane spanning segment of the channel (Yang et al. 2012). By contrast, Gem and Rem2 utilize solely a β‐binding‐dependent mechanism to inhibit Ca_V1/Ca_V2 channels. Overall, insights into the mechanisms and physical determinants of RGK inhibition of Ca_V1/Ca_V2 channels has proven invaluable to the broad objective of drawing inspiration from these proteins as prototype molecules to design next‐generation genetically encoded Ca_V channel inhibitors as research tools and potential therapeutics.

RGK‐inspired genetically encoded Ca_V channel inhibitors

Blocking Ca_V1/Ca_V2 channels with small molecules or toxins is a prevailing or prospective therapeutic strategy for many serious diseases including hypertension, chronic pain, cardiac arrhythmias, Parkinson's disease and stroke (Zamponi et al. 2015; Zamponi, 2016). While convenient, small molecule Ca_V channel blockers have limitations, some of which may be circumvented by genetically encoded inhibitors (Xu & Colecraft, 2009). First, they lack tissue specificity since small molecules are typically widely distributed in the body after administration, and distinct VDCCs are present across many different tissues, organs and cell types. Second, VDCCs show an immense molecular and functional diversity stemming from their organization into distinct macromolecular complexes, and sub‐cellular localizations that are poorly discriminated by small molecules. These two gap areas could potentially be filled by novel genetically encoded Ca_V channel inhibitors designed to target molecularly distinct VDCC macromolecular complexes in a tissue‐ or cell‐specific manner. While RGK proteins themselves are potent VDCC inhibitors, their usefulness as research tools or therapeutics is limited by several factors: (1) they are non‐selective, as they indiscriminately inhibit all Ca_V1/Ca_V2 channel types; (2) they are constitutive inhibitors, thus providing poor temporal and spatial control of channel block; and (3) they are non‐specific as they interact with and regulate other proteins such as enzymes and the cytoskeleton in cells (Yang & Colecraft, 2013). Over the last few years, using RGK proteins themselves as inspiration, we and others have explored different ways to engineer new genetically encoded Ca_V channel inhibitors that improve on various aspects of functional Ca_V channel block that are lacking in wild‐type RGK proteins.

Our finding that Rem specifically inhibits Ca_V1.2 using both a β‐binding‐dependent and α_1C‐binding‐dependent mechanism but used only a β‐binding‐dependent mechanism to block other Ca_V1/Ca_V2 channel types suggested a simple method to create a Ca_V1.2‐selective genetically encoded Ca_V channel inhibitor – introduce mutations in Rem that weaken its interaction with Ca_Vβ without altering the tertiary structure of the protein. Indeed, such mutations (Rem[R200A/L227A]) were identified by an extensive mutagenesis study (Beguin et al. 2007). Consistent with the hypothesis, Rem[R200A/L227A] selectively inhibited Ca_V1.2, but not other Ca_V1/Ca_V2 channels, reconstituted in HEK293 cells (Puckerin et al. 2018). The ability of Rem[R200A/L227A] to discriminate between Ca_V1.2 and Ca_V1.3 was especially notable given the difficulty of identifying small molecules that can effectively distinguish between these two L‐type channel subtypes (Zamponi et al. 2015). Using a similar logic, we found that Rad[R208A/L235A] selectively blocked Ca_V1.2 and Ca_V2.2, consistent with the finding that Rad inhibits these two channels using both β‐binding‐dependent and β‐binding‐independent mechanisms (Puckerin et al. 2018). Importantly, both Rem[R200A/L227A] and Rad[R208A/L235A] strongly inhibited Ca_V1.2 channels in cardiomyocytes, indicating that the β‐binding‐independent mechanism of inhibition is operational in this native environment. Similarly, the two proteins inhibited HVA Ca_V channel currents in dorsal root ganglion (DRG) neurons to different extents, reflecting their varying selectivity for Ca_V1.2 and Ca_V1.2/Ca_V2.2 channels, respectively (Puckerin et al. 2018).

Rem associates with the plasma membrane via the 32‐residue distal C‐terminus (DCT) using hydrophobic and electrostatic interactions. Deletion of the DCT abolishes both Rem membrane targeting and inhibition of Ca_V1/Ca_V2 channels (Finlin et al. 2003; Yang et al. 2007). The requirement for Rem binding to the plasma membrane for Ca_V channel inhibition has been exploited to engineer Rem derivatives that enable chemo‐ and optogenetic control of channel inhibition, and also subcellular specificity (Fig. 2). We replaced Rem DCT with the C1 domain from protein kinase γ, creating Rem_1‐265‐C1_PKCγ which when expressed in cells was primarily distributed in the cytosol but could be rapidly recruited to the plasma membrane with a small molecule, phorbol‐12,13‐dibuytrate (PdBu). The PdBu‐induced recruitment of Rem_1‐265‐C1_PKCγ caused a concomitant rapid inhibition of Ca_V1/Ca_V2 channel currents (Fig. 2B ) (Yang et al. 2013, 2007). The generality of this chemogenetic regulation was demonstrated by development of a FK506‐binding protein (FKBP)‐tagged Rem₂₆₅ version that could be recruited to the membrane to inhibit Ca_V1/Ca_V2 channels using rapamycin‐mediated heterodimerization in cells that also expressed constitutively membrane‐targeted FRB (a fragment of mTOR) (Crabtree & Schreiber, 1996; Inoue et al. 2005; Yang et al. 2007). Similarly, a 490 nm blue light‐mediated heterodimerization strategy was utilized to develop optogenetic control of Rem inhibition (Fig. 2C ). The approach is based on a light‐induced protein‐protein interaction created by inserting a bacterial peptide, ssrA, into a naturally occurring photoswitch, light‐oxygen‐voltage 2 (LOV2) domain from Avena sativa (Guntas et al. 2015). In the dark, SsrA is sterically obstructed from interacting with a binding partner, sspB. With blue light, this steric inhibition is relieved, allowing SsrA to bind SspB. Extensive bioengineering of LOV2‐SsrA yielded an improved light inducible dimer (iLID) in which the affinity of the photoswitch for SspB changes > 50‐fold with light illumination (Guntas et al. 2015). Ma et al. (2018) replaced Rem DCT with SspB (creating optoRGK) and anchored iLID constitutively to the plasma membrane using Lyn11, a plasma membrane‐tethering peptide from the tyrosine protein kinase, Lyn. Exposure of cells to blue light led to rapid recruitment of optoRGK to the plasma membrane and resulted in Ca_V channel inhibition that was quickly reversed in the dark (Ma et al. 2018) (Fig. 2C ). Finally, as a demonstration of inhibiting Ca_V channels with subcellular specificity, replacing the Rem C‐terminus with a caveolin‐targeting peptide enabled selective inhibition of caveolae‐localized Ca_V1.2 in cardiac myocytes, without significantly affecting non‐caveolae Ca_V1.2 channels responsible for excitation‐contraction coupling (Fig. 2D ) (Makarewich et al. 2012).

Figure 2. Replacing Rem distal C‐terminus for novel spatio‐temporal control of Ca_V channel inhibition.

A, Rem structure consists of a guanine nucleotide binding domain (G‐domain) flanked by N‐ and C‐termini. The Rem distal C‐terminus (DCT), comprising the last 32 residues of the protein, is a polybasic peptide that mediates binding to the plasma membrane and is necessary for Ca_V channel inhibition. B, replacing Rem DCT with C1 domain from protein kinase C γ (C1_PKCγ) enables acute recruitment of the engineered Rem to the plasma membrane with a small molecule phorbol ester, PdBu. Co‐expressed Ca_V1/Ca_V2 channels are inhibited concomitantly with Rem₂₆₅‐C1_PKCγ association with the plasma membrane. This chemogenetic configuration provides acute temporal control over Ca_V channel inhibition that is slowly reversible. C, optogenetic control of Rem inhibition of Ca_V channels was achieved using the photodimerizer pair, iLID (LOV2‐ssrA) and sspB. The Rem DCT was replaced with sspB via varying linkers (creating optoRGK) while iLID was constitutively anchored to the plasma membrane. Exposure of cells to blue light (470 nm) enabled acute recruitment of optoRGK to the plasma membrane and inhibition of Ca_V1.2 channels. Both plasma membrane association of optoRGK and Ca_V1.2 channel inhibition were reversed in the dark. D, replacing Rem DCT with a caveolae‐targeting peptide enabled selective inhibition of caveolae‐targeted Ca_V1.2 channels in cardiomyocytes while sparing dyadic Ca_V1.2 channels that mediate cardiac excitation‐contraction coupling.

The next conceptual advance came from further consideration of why Rem inhibition of Ca_V1.2 P _o had the dual requirement for Ca_Vβ binding and plasma membrane association? We hypothesized that Rem binding to the plasma membrane ‘pulled’ on the I‐II loop via the associated Ca_Vβ subunit and induced a conformation of the channel with a low P _o. This hypothesis led to a testable prediction that we could potentially evoke a similar low‐P _o channel conformational state by directly attaching a membrane‐targeting module to auxiliary Ca_Vβ subunits, thereby bypassing the need for an RGK altogether (Yang et al. 2013). To accomplish this, we fused the C1_PKCγ onto the C‐terminus of Ca_Vβ₃ (generating β₃‐C1_PKCγ) which enabled a PdBu‐induced association of β₃ with the plasma membrane (Yang et al. 2013). Channels reconstituted with β₃‐C1_PKCγ yielded robust baseline whole‐cell currents that were inhibited by exposure to PdBu. The kinetics and extent of inhibition could be tuned by serial truncations of the disordered β₃ C‐terminus (shortening the β₃ C‐terminus sped up the onset and deepened the extent of inhibition) (Yang et al. 2013). While this result was in accord with the stated hypothesis, it was, nevertheless surprising, because β_2a and β_2e subunits are naturally membrane‐associated via their N‐termini (Chien et al. 1998; Takahashi et al. 2003). β_2a is palmitoylated, while the N‐terminus of β_2e forms a helix that associates with the plasma membrane using electrostatic and hydrophobic interactions (Miranda‐Laferte et al. 2014). However, neither β_2a nor β_2e constitutively inhibit channels (rather, they both slow down voltage‐dependent inactivation of Ca_V1/Ca_V2 channels) (Takahashi et al. 2003). An apparent explanation for this discrepancy arose from the finding that placing the C1_PKCγ module on the β₃ N‐terminus yielded a construct that did not effectively inhibit Ca_V channels in response to PdBu, indicating that the phenomenon is sensitive to the polarity of the membrane‐targeting module on Ca_Vβ (Yang et al. 2013). This suggests a geometric constraint to this mode of inhibition. Based on these results, we probed whether other cytosolic proteins that bound other intracellular loops of Ca_V channels could be transformed into Ca_V1/Ca_V2 inhibitors simply by introducing a membrane binding module to them. Indeed, we found that 14‐3‐3, a protein previously reported to bind to Ca_V2.2 C‐terminus (Li et al. 2006), could be turned into either a PdBu‐inducible or constitutive inhibitor by attaching C1_PKCγ or a palmitoylated peptide, respectively (Yang et al. 2013). Unexpectedly, we found that 14‐3‐3‐C1_PKCγ also effectively inhibited Ca_V1.2 and Ca_V2.1 channels in a phorbol ester‐dependent manner, revealing that these other channels also interacted with 14‐3‐3. We termed this general mechanism ChIMP, an acronym for ‘channel inactivation by membrane‐tethering an associated protein’ (Yang et al. 2013). Beyond Ca_V1/Ca_V2 channels, ChIMP may also be used either as an investigational tool or method to develop genetically encoded modulators for other ion channels. In this regard, we exploited ChIMP to reveal that calmodulin is preassociated with TMEM16A and TMEM16B Ca²⁺‐activated chloride channels and mediates Ca²⁺‐dependent sensitization of activation as well as Ca²⁺‐dependent inactivation of particular splice variants (Yang et al. 2014).

Deployment of genetically encoded Ca_V channel inhibitors derived from endogenous proteins (such as Rem_1‐265‐C1_PKCγ, β₃‐C1_PKCγ, and 14‐3‐3‐C1_PKCγ) in vivo could potentially have unwanted effects owing to over‐expression of these modified natural proteins. As such, we sought to develop genetically encoded Ca_V channel inhibitors that would have limited off‐target effects relative to their inhibition of HVA Ca_V channels. Given the importance of Ca_Vβ‐binding in RGK‐mediated Ca_V1/Ca_V2 inhibition, we first isolated nanobodies targeted to auxiliary Ca_Vβ subunits. We immunized a llama with purified β₁ and β₃ subunits, isolated lymphocytes, amplified nanobodies by PCR, and cloned into a phagemid vector to generate a V_HHS phage library. Several nanobody binders to β₁ were isolated using phage display and an ELISA assay. One of these nanobodies, termed nb.F3, bound all four Ca_Vβ isoforms when expressed in cells (Morgenstern et al. 2019), which was not surprising given the high homology among these auxiliary subunits in their conserved src homology 3 (SH3) and guanylate kinase (GK) domains (Chen et al. 2004; Opatowsky et al. 2004; Van Petegem et al. 2004). Purified nb.F3 bound Ca_Vβ with high affinity (∼12 nM) and 1:1 stoichiometry as assessed by isothermal calorimetry. When expressed with reconstituted Ca_V2.2 and Ca_V1.2 channels in HEK293 cells, nb.F3 appeared functionally inert, as it had no impact on channel trafficking to the plasma membrane or on whole‐cell currents. Therefore, nb.F3 provided an ideal Ca_Vβ‐targeting module that could potentially be modified to generate a genetically encoded Ca_V channel inhibitor exploiting the mechanisms we had identified for RGK proteins. We first sought to mimic the impact of RGKs on decreasing the channel surface density by fusing the catalytic HECT domain of the ubiquitin ligase, Nedd4L, onto nb.F3. The rationale for this approach is that in many ion channels and membrane proteins, ubiquitination typically reduces surface density and, often, enhances protein degradation as well (Abriel & Staub, 2005; Jespersen et al. 2007; MacGurn et al. 2012; Kanner et al. 2017). In heterologous cells, nb.F3‐Nedd4L decreased the surface density of reconstituted Ca_V2.2 and Ca_V1.2 channels without enhancing the degradation of the pore‐forming α_1B and α_1C subunits, respectively (Fig. 3A and B ) (Morgenstern et al. 2019). Whole‐cell patch clamp experiments demonstrated that nb.F3‐Nedd4L essentially eliminated reconstituted Ca_V1.2, Ca_V1.3 and Ca_V2.1‐Ca_V2.3 channel currents (Fig. 3C ). Therefore, we named nb.F3‐Nedd4L as Ca_V‐aβlator, reflecting it's exceptional efficacy to inhibit HVA Ca_V channels by targeting auxiliary Ca_Vβ subunits. Ca_V‐ablator also proved effective in eliminating endogenous Ca_V1/Ca_V2 channels in pancreatic β‐cells, dorsal root ganglion (DRG) neurons and cardiac myocytes (Morgenstern et al. 2019). Examination of how Ca_V‐ablator eliminated Ca_V1.2 currents in ventricular cardiomyocytes indicated that pore‐forming α_1C subunits were re‐directed from dyadic junctions to intracellular compartments, specifically Rab 7‐positive late endosomes (Morgenstern et al. 2019).

Figure 3. Mimicking RGK‐mediated Ca_V inhibition mechanisms with an engineered Ca_Vβ‐targeted nanobody.

A, schematic diagram of experimental paradigm. Recombinant Ca_V2.2 (α_1B) with an extracellular bungarotoxin‐binding site (BBS) epitope is co‐expressed Ca_Vβ without (control) or with a Ca_Vβ‐targeting nanobody (nb.F3) fused to catalytic HECT domain of the E3 ubiquitin ligase, NEDD4L. Surface channels are measured by exposing non‐permeabilized transfected cells to Alexa 647‐conjugated bungarotoxin. B, histograms of surface BBS‐α_1B assessed by flow cytometry in cells expressing no nanobody (top), nb.F3‐NEDD4L (middle), or nb.F3‐NEDD4L^*, a catalytically dead variant (bottom). Results show a substantial decline in surface density when the channel is co‐expressed with nb.F3‐NEDD4L. C, exemplar (top) and population I‐V curves (bottom) in cells expressing α_1B + β_1b alone (■) or with either nb.F3‐NEDD4L () or nb.F3‐NEDD4L^* () co‐expression. D, schematic diagram of experimental paradigm. HEK293 cells are co‐transfected with recombinant Ca_V2.2 (α_1B + β₃) and nb.F3‐C1_PKCγ. E and F, exemplar currents and diary plot showing rapid and deep PdBu‐induced inhibition of Ca_V2.2 currents in cells expressing α_1B + β₃ + nb.F3‐C1_PKCγ.

We have also explored whether we could also use nb.F3 to create a small‐molecule‐inducible genetically encoded Ca_V channel inhibitor that exploited the ChIMP mechanism. We generated nb.F3‐ C1_PKCγ and co‐expressed it with recombinant Ca_V1.2 channels. Exposure of cells to phorbol ester resulted in a rapid decline in current that was not observed in control cells lacking nb.F3‐C1_PKCγ, indicating that nb.F3 permits inducible inhibition of Ca_V1/Ca_V2 channels via the ChIMP method (Fig. 3D–F ).

Conclusion

In summary, this review highlights work focused on understanding the mechanisms by which RGK proteins potently inhibit Ca_V1‐ and Ca_V2‐family channels and exploiting mechanistic insights to create novel genetically encoded Ca_V channel inhibitors. This work has led to the development of intracellular acting genetically encoded Ca_V channel inhibitors that can be controlled by either small molecules or light, and that have the capacity to block Ca_V1.2 channels in cardiac myocytes with subcellular specificity. Genetically encoded Ca_V channel inhibitors have potential utility as therapeutics for indications such as chronic pain, with the advantage that their expression can be restricted to target tissues or cell types of interest, thereby circumventing off‐target effects. The viability of such gene therapy approaches has been advanced by continually improved development of viral and non‐viral gene delivery methods in vivo. For such potential therapeutic applications, it would be important to develop variants whose potency can be controlled either through dosage or with a small molecule. The nanobody‐based approach offers opportunities to design novel genetically encoded Ca_V channel inhibitors that can eliminate or modulate Ca_V channel complexes on the basis of identity of the associated β subunit isoform. This would be a key enabling tool to probe the potential role of auxiliary β subunits in organizing distinct Ca_V channels into distinct signalling complexes that permit functional diversification of Ca²⁺ influx via Ca_V channels in individual cells. Finally, some of the approaches described here may be generalizable to develop genetically encoded inhibitors or modulators for other ion channels and membrane proteins. Indeed, we have previously shown that the nanobody‐based targeted ubiquitination approach can be used to inhibit KCNQ1 channels by eliminating them from the cell surface (Kanner et al. 2017).

Additional information

Competing interests

None declared.

Funding

This work was supported by grants from the National Institutes of Health (RO1‐GM107585, RO1‐HL121253, and 1RO1‐HL122421) to H.M.C.

Biography

Henry M. Colecraft, PhD, is the John C. Dalton Professor of Physiology and Cellular Biophysics at Columbia University Vagelos College of Physicians and Surgeons. Dr Colecraft's laboratory focuses on modulation of voltage‐gated ion channels (by intracellular signalling proteins, auxiliary subunits and posttranslational modifications) and understanding molecular/biophysical mechanisms underlying diseases caused by ion channel mutations (ion channelopathies). His lab has used protein engineering approaches to develop genetically encoded molecules to inhibit or modulate the activity of ion channels for customized applications.