Engineering selectivity into RGK GTPase inhibition of voltage-dependent calcium channels.

Genetically encoded inhibitors for voltage-dependent Ca2+ (CaV) channels (GECCIs) are useful research tools and potential therapeutics. Rad/Rem/Rem2/Gem (RGK) proteins are Ras-like G proteins that potently inhibit high voltage-activated (HVA) Ca2+ (CaV1/CaV2 family) channels, but their nonselectivity limits their potential applications. We hypothesized that nonselectivity of RGK inhibition derives from their binding to auxiliary CaVβ-subunits. To investigate latent CaVβ-independent components of inhibition, we coexpressed each RGK individually with CaV1 (CaV1.2/CaV1.3) or CaV2 (CaV2.1/CaV2.2) channels reconstituted in HEK293 cells with either wild-type (WT) β2a or a mutant version (β2a,TM) that does not bind RGKs. All four RGKs strongly inhibited CaV1/CaV2 channels reconstituted with WT β2a By contrast, when channels were reconstituted with β2a,TM, Rem inhibited only CaV1.2, Rad selectively inhibited CaV1.2 and CaV2.2, while Gem and Rem2 were ineffective. We generated mutant RGKs (Rem[R200A/L227A] and Rad[R208A/L235A]) unable to bind WT CaVβ, as confirmed by fluorescence resonance energy transfer. Rem[R200A/L227A] selectively blocked reconstituted CaV1.2 while Rad[R208A/L235A] inhibited CaV1.2/CaV2.2 but not CaV1.3/CaV2.1. Rem[R200A/L227A] and Rad[R208A/L235A] both suppressed endogenous CaV1.2 channels in ventricular cardiomyocytes and selectively blocked 25 and 62%, respectively, of HVA currents in somatosensory neurons of the dorsal root ganglion, corresponding to their distinctive selectivity for CaV1.2 and CaV1.2/CaV2.2 channels. Thus, we have exploited latent β-binding-independent Rem and Rad inhibition of specific CaV1/CaV2 channels to develop selective GECCIs with properties unmatched by current small-molecule CaV channel blockers.

High voltage-activated (HVA) Ca2+ (CaV) channels convert electrical signals into Ca2+ influx that controls myriad essential processes including neuronal communication, muscle contraction, hormone release, and activity-dependent gene transcription (1). HVA CaV channels are composed of a pore-forming α1- assembled with auxiliary β-, α2δ-, and γ-subunits and calmodulin. There are seven α1-subunits (CaV1.1 to 1.4 and CaV2.1 to 2.3), four CaVβs (β1 to β4), and three α2δs (α2δ1 to α2δ3), each with multiple splice variants. CaVα1-subunits contain the voltage sensor, selectivity filter, and channel pore, while auxiliary subunits regulate channel properties—CaVβs are obligatory for α1-trafficking to the plasma membrane and for modulating channel gating (2); α2δs enhance channel surface trafficking and also modulate channel gating (3); and calmodulin promotes channel trafficking, enhances basal open probability (Po), and confers feedback Ca2+ regulation of channel gating (4, 5).

CaV channels are also regulated by various intracellular signaling proteins and posttranslational modifications as a mechanism to control physiology. Pharmacological blockade of CaV1/CaV2 channels is an important treatment strategy for diverse diseases including hypertension, cardiac arrhythmias, chronic pain, and Parkinson’s disease (1). RGK proteins (Gem, Rad, Rem, and Rem2) are small Ras-like G proteins that bind CaVβ-subunits and profoundly inhibit all CaV1/CaV2 channels (6–8). Given their properties, RGKs straddle two worlds with respect to their impact on CaV1/CaV2 channels—they are (i) potentially powerful physiological regulators by virtue of their capacity to tune intracellular Ca2+ signals, and (ii) prototype genetically encoded CaV channel blockers with possible therapeutic and biotechnological applications (9). Consistent with important physiological roles, Rad knockout mice exhibit increased cardiac CaV1.2 currents and cardiac hypertrophy, while Gem-deficient mice display glucose intolerance and impaired glucose-stimulated insulin release. Regarding their use as potential therapeutics, expression of Gem in the atrioventricular node was effective at electrically uncoupling ventricular excitation from the fibrillating atria in a porcine model of atrial fibrillation (10). A Rem derivative engineered to selectively target and inhibit caveola-localized CaV channels effectively inhibited pacing-induced NFATc3-GFP translocation to the nucleus in adult feline ventricular cardiomyocytes, without affecting excitation–contraction coupling (11).

A major limitation for the use of RGKs as genetically encoded CaV channel blockers involves their lack of selectivity for particular CaV1/CaV2 isoforms. Rem inhibits CaV1.2 channels using multiple mechanisms including reduced channel surface density, diminished Po, and partial immobilization of voltage sensors (12). At least one of these mechanisms (decreased Po) involves the simultaneous association of Rem with the auxiliary CaVβ-subunit and the plasma membrane (13, 14). This mechanism likely accounts for the indiscriminate nature of RGK inhibition of CaV1/CaV2 channels, since all four RGKs bind CaVβ-subunits and the plasma membrane, and CaVβs are obligatory for forming functional channels. Beyond the β-binding mechanism, we previously showed that Rem can also inhibit CaV1.2 channels by directly binding to the pore-forming α1C-subunit (15). Potentially, such an α1-subunit–dependent mechanism could be exploited to develop genetically encoded CaV1/CaV2 isoform-selective inhibitors. Several outstanding questions need to be addressed to realize this potential. Does Rem inhibit other CaV1/CaV2 channels beyond CaV1.2 in a β-binding–independent manner? Do other RGKs beyond Rem inhibit CaV1/CaV2 channel isoforms in a β-binding–independent manner? Are both β-binding–dependent and β-binding–independent mechanisms of RGK inhibition of particular CaV1/CaV2 channels prevalent in native excitable cells? If so, do the two modes of inhibition display physiologically meaningful differences?

Here, focusing on four CaV channels (CaV1.2, CaV1.3, CaV2.1, and CaV2.2), we show that Rem uniquely blocks CaV1.2 using a β-binding–independent mechanism. Consistent with this finding, a mutant Rem that cannot bind β (Rem[R200A/L227A]) selectively blocked CaV1.2, with no effect on the closely related CaV1.3 channel. Further, Rad inhibited CaV1.2 and CaV2.2 (but not CaV1.3 or CaV2.1) channels via a β-binding–independent mechanism. Accordingly, a β-binding–deficient Rad mutant (Rad[R208A/L235A]) effectively blocked CaV1.2/CaV2.2, but not CaV1.3/CaV2.1, channels. Both Rem[R200A/L227A] and Rad[R208A/L235A] strongly inhibited endogenous CaV1.2 channels in adult ventricular cardiomyocytes. Finally, Rem[R200A/L227A] and Rad[R208A/L235A], but not Gem[R196A/V223A], inhibited HVA CaV channels in somatosensory dorsal root ganglion (DRG) neurons, albeit with different magnitudes reflecting their selectivity for either CaV1.2 alone or CaV1.2/CaV2.2, respectively. Altogether, we have exploited latent β-binding–independent inhibition of CaV1.2 and CaV1.2/CaV2.2 channels by Rem and Rad, respectively, to engineer genetically encoded isoform-selective CaV channel blockers.

Results

Differential Prevalence of β-Binding–Dependent and β-Binding–Independent Rem Inhibition Across Distinct CaV1/CaV2 Channels.

We profiled β-binding–dependent (BBD) and β-binding–independent (BBI) Rem inhibition of CaV channels by reconstituting distinct pore-forming α1-subunits with either wild-type β2a or a mutant β2a (β2a,TM) that does not bind RGKs. HEK293 cells expressing α1C + β2a expressed robust Ba2+ currents (IBa) that were virtually eliminated when Rem was coexpressed (Fig. 1 and ). Similarly, cells expressing α1C + β2a,TM displayed IBa that was significantly inhibited by Rem (Fig. 1 and ), indicating the incidence of both BBD and BBI Rem inhibition of CaV1.2 channels. These results confirm our previous report that both BBD and BBI mechanisms contribute to Rem inhibition of CaV1.2 (15). IBa influx through reconstituted CaV1.3 channels (α1D + β2a) was eliminated by Rem. However, CaV1.3 channels reconstituted with α1D + β2a,TM were refractory to Rem (Fig. 1 and ), indicating the absence of BBI inhibition, and revealing a fundamental difference from CaV1.2. Similar to CaV1.3, CaV2.1 (Fig. 1 and ) and CaV2.2 (Fig. 1 and ) channels were inhibited by Rem only when reconstituted with WT β2a, but not β2a,TM, a modulation profile consistent with exclusively BBD inhibition.

Fig. 1.

Rem uniquely inhibits CaV1.2 using both β-binding–dependent and β-binding–independent mechanisms. (A) Exemplar CaV1.2 Ba2+ currents elicited from HEK293 cells expressing α1C + β2a ± Rem (columns 1 and 2) or α1C + β2a,TM ± Rem (columns 3 and 4). Ba2+ currents were elicited by 25-ms test pulse depolarizations (from −50 to +100 mV in 10-mV increments) from a holding potential of −90 mV. (B) Population bar charts showing the impact of Rem on peak IBa from channels reconstituted with either α1C + β2a (Left) or α1C + β2a,TM (Right). *P < 0.01, Student’s unpaired t test. (C and D) Data for CaV1.3 channels reconstituted with either α1D + β2a ± Rem or α1D + β2a,TM ± Rem, same format as A and B. (E and F) Data for CaV2.1 channels reconstituted with either α1A + β2a ± Rem or α1A + β2a,TM ± Rem, same format as A and B. (G and H) Data for CaV2.2 channels reconstituted with either α1B + β2a ± Rem or α1B + β2a,TM ± Rem, same format as A and B. Data are means ± SEM.

Engineering a CaV1.2-Selective Inhibitor from Rem.

The finding that BBI Rem inhibition of IBa is a unique property of CaV1.2 suggested the possibility of engineering a CaV1.2-selective genetically encoded inhibitor by generating a Rem mutant that does not bind CaVβ. A previous mutagenesis study identified residues in RGKs that were critical for their interaction with CaVβs but did not disrupt their tertiary structure, as evaluated by GTP/GDP binding assays (16). Based on these findings, we introduced two point mutations (R200A, L227A) into Rem and used FRET to evaluate the association of Rem[R200A/L227A] with CaVβ (Fig. 2). HEK293 cells coexpressing CFP-WT Rem + YFP-β3 displayed a significantly elevated FRET (FRET efficiency 0.188 ± 0.006, n = 127) compared with negative control cells expressing CFP-FRB + β3-YFP (FRET efficiency 0.046 ± 0.002, n = 126) (Fig. 2), consistent with well-known Rem–CaVβ interaction (6, 7). By comparison, cells coexpressing CFP-Rem[R200A/L227A] + β3-YFP displayed a markedly lower FRET (FRET efficiency 0.058 ± 0.002, n = 138) that did not differ from control cells, consistent with reduced protein interaction (Fig. 2). Additional insights into the relative affinities of Rem and Rem[R200A/L227A] for CaVβ3 was provided from binding analyses of FRET efficiency vs. Afree scatterplots (Fig. 2), which indicated a fivefold decreased affinity of Rem[R200A/L227A] for CaVβ3 compared with WT Rem (Fig. 2).

Fig. 2.

Rem[R200A/L227A] does not bind CaVβ and selectively inhibits CaV1.2. (A) Crystal structure of Rem G domain with residues R200 and L227 highlighted. (B) FRET efficiency measurements in HEK293 cells coexpressing CFP-FRB + β3-YFP (control), CFP-Rem + β3-YFP, or CFP-Rem[R200A/L227A] + β3-YFP. Data are means ± SEM. *P < 0.01, one-way ANOVA. (C) Binding-curve analyses of FRET experiments. (D) Population Ipeak–V plots for cells expressing α1C + β2a (black squares; n = 9), α1C + β2a + Rem[L200A/L227A] (red squares; n = 10); α1D + β2a (black circles; n = 11), α1D + β2a + Rem[L200A/L227A] (red circles; n = 13); α1A + β2a (black triangles; n = 10), α1A + β2a + Rem[L200A/L227A] (red triangles; n = 12); and α1B + β2a (black diamonds; n = 11), α1B + β2a + Rem[L200A/L227A] (red diamonds; n = 12). *P < 0.05, Student’s t test. (E) Exemplar cultured adult cardiomyocytes expressing GFP (Top) or CFP-Rem[L200A/L227A] (Bottom). (F) Representative Ba2+ currents from adult guinea pig ventricular cardiomyocytes expressing either GFP (Left; control) or CFP-Rem[L200A/L227A] (Right). (G) Population Ipeak–V plots for cardiomyocytes expressing GFP (black squares; n = 8) or CFP-Rem[L200A/L227A] (red triangles; n = 10).

We next determined whether Rem[R200A/L227A] would function as a CaV1.2-selective inhibitor as hypothesized. Indeed, HEK293 cells coexpressing recombinant CaV1.2 (α1C + β2a) channels and Rem[R200A/L227A] displayed significantly lower IBa compared with control cells expressing CaV1.2 alone (Fig. 2; Ipeak,10mV = 62.3 ± 14.3 pA/pF, n = 9 for α1C + β2a compared with Ipeak,10mV = 24.9 ± 4.9 pA/pF, n = 10 for α1C + β2a + Rem[R200A/L227A]; P = 0.0194, Student’s t test; ). In sharp contrast, recombinant CaV1.3, CaV2.1, and CaV2.2 were refractory to Rem[R200A/L227A] (Fig. 2 and ), consistent with this engineered protein being a CaV1.2-selective blocker. The finding that CaV2.2 is not inhibited by a β-binding–deficient Rem recapitulates a previous similar finding by Beqollari et al. (17).

To determine whether Rem[R200A/L227A] could inhibit endogenous CaV1.2 channels, we assessed its efficacy in blocking IBa conducted through native CaV1.2 channels in guinea pig ventricular cardiomyocytes. We generated adenovirus enabling robust expression of YFP-Rem[R200A/L227A] (Fig. 2). Compared with control cells expressing GFP, cardiomyocytes expressing YFP-Rem[R200A/L227A] displayed a significantly reduced IBa at all test voltages (Fig. 2 ; Ipeak,0mV = 22.6 ± 4.6 pA/pF, n = 8 for GFP compared with Ipeak,0mV = 9.1 ± 2.3 pA/pF, n = 10, for YFP-Rem[R200A/L227A]), thus demonstrating BBI Rem inhibition of endogenous CaV1.2 channels in the heart.

Prevalence of BBD and BBI RGK Inhibition Across the CaV1/CaV2 Channel Family.

We wondered whether other RGKs display BBI inhibition of CaV1/CaV2 channels that could be similarly exploited to generate selective genetically encoded inhibitors for CaV channels (GECCIs). We profiled the occurrence of BBD and BBI inhibition across RGKs and CaV1/CaV2 channels by assessing the impact of Gem, Rad, and Rem2 on recombinant CaV channels reconstituted with either WT β2a or β2a,TM (Fig. 3 and ). CaV1.3 channels reconstituted with WT β2a (α1D + β2a) were uniformly inhibited by Gem, Rad, and Rem2, respectively (). By contrast, these three RGKs had no impact on IBa influx through α1D + β2a,TM channels (). Together, these results indicate that all RGKs inhibit CaV1.3 channels solely through a BBD mechanism. We obtained virtually identical results with reconstituted CaV2.1 channels—α1A + β2a channels were inhibited by Gem, Rad and Rem2, whereas α1A + β2a,TM channels were refractory to these RGKs (). Hence, CaV2.1 channels also display exclusively BBD RGK inhibition. The finding that Rem2 inhibits CaV2.1 in a solely BBD manner agrees with a previous result showing that Rem2 abolishes current through CaV2.1 channels reconstituted with WT β4 but not a mutant β4 lacking the capacity to bind Rem2 (18). Our finding that Gem requires binding to CaVβ to decrease CaV2.1 is in disagreement with a previous report that Gem binding to CaVβ3 was not necessary for its capacity to inhibit CaV2.1 current (19). The reasons for this discrepancy are unclear, though one possibility is the intrinsic differences between Xenopus oocytes (used in the previous study) and the mammalian cells used here. As expected, wild-type CaV2.2 channels (α1B + β2a) were robustly inhibited by Gem, Rad, and Rem2, respectively. Interestingly, while channels reconstituted with α1B + β2a,TM were unaffected by Gem and Rem2, they were significantly inhibited by Rad (Fig. 3). Therefore, Rad uniquely mediates both BBD and BBI inhibition of CaV2.2 channels. We previously reported that Rad (but not Gem or Rem2) also supports BBD and BBI inhibition of CaV1.2 (15). Together, these reports suggested that eliminating Rad binding to CaVβ would generate a selective inhibitor of CaV1.2/CaV2.2 channels.

Fig. 3.

Prevalence of β-binding–dependent and β-binding–independent RGK inhibition of IBa in CaV2.2 channels. (A) Schematic of HVA CaV channel pore-forming α1-subunit binding to β2a or β2a,TM with putative binding sites responsible for β-binding–dependent (solid arrow) and β-binding–independent (dashed arrows) RGK inhibition of current. (B) Bar charts showing impact of Gem, Rad, and Rem2 on CaV2.2 channels reconstituted with either α1B + β2a (Left) or α1B + β2a,TM (Right). Data are means ± SEM. *P < 0.05 compared with control, one-way ANOVA.

Engineering a CaV1.2- and CaV2.2-Selective Inhibitor from Rad.

Using an approach similar to the generation of Rem[R200A/L227A], we introduced equivalent mutations in Rad to create Rad[R208A/L235A]. Three-cube FRET experiments confirmed that cells expressing CFP-Rad[R208A/L235A] + YFP-β3 showed lower FRET efficiency (0.051 ± 0.002, n = 142) compared with CFP-Rad + YFP-β3 (0.123 ± 0.004, n = 174) (Fig. 4). Binding-curve analyses indicated an eightfold decrease in affinity of CFP-Rad[R208A/L235A] for YFP-β3 compared with CFP-Rad (Fig. 4). As predicted, Rad[R208A/L235A] significantly inhibited currents through recombinant CaV1.2 (α1C + β2a) and CaV2.2 (α1B + β2a) channels but had no impact on either CaV1.3 (α1D + β2a) or CaV2.1 (α1A + β2a) channels (Fig. 4 and ). Hence, Rad[R208A/L235A] is a CaV1.2/CaV2.2-selective inhibitor. When expressed in guinea pig ventricular cardiomyocytes, Rad[R208A/L235A] inhibited endogenous CaV1.2 channels to almost the same extent as WT Rad (Fig. 4 ), revealing a strong BBI Rad inhibition of CaV1.2 in the heart. It was previously shown that Rad-inhibited CaV1.2 channels are not up-regulated by activated protein kinase A (PKA) (20). We found that IBa through ventricular CaV1.2 channels inhibited by Rad[R208A/L235A] was robustly increased by 1 μM forskolin, in sharp contrast to the lack of modulation observed with WT Rad-inhibited channels (Fig. 4 and ). Hence, cardiac CaV1.2 channels undergoing either BBD or BBI Rad inhibition display fundamental differences in their sensitivity to PKA regulation. A caveat here is we cannot discount a contribution of CaV1.2 channels which are not bound to Rad[R208A/L235A] to the observed forskolin-induced increase in IBa. However, the finding that Rad[R208A/L235A] inhibits cardiac CaV1.2 to almost the same extent as WT Rad (Fig. 4 ) suggests Rad[R208A/L235A]-bound channels predominate over unbound channels, and is consistent with the interpretation that CaV1.2 channels undergoing BBI Rad inhibition are up-regulated by PKA activation.

Fig. 4.

Rad[R208A/L235A] does not bind CaVβ and selectively inhibits CaV1.2 and CaV2.2 channels. (A) Crystal structure of Rad G domain with residues R208 and L235 highlighted. (B) FRET efficiency measurements in HEK293 cells coexpressing CFP-FRB + β3-YFP (control), CFP-Rad + β3-YFP, or CFP-Rad[R208A/L235A] + β3-YFP. Data are means ± SEM. *P < 0.01, one-way ANOVA. (C) Binding-curve analyses of FRET experiments. (D) Population Ipeak–V plots for cells expressing α1C + β2a (black squares; n = 11), α1C + β2a + Rad[R208A/L235A] (red squares; n = 10); α1D + β2a (black circles; n = 10), α1D + β2a + Rad[R208A/L235A] (red circles; n = 13); α1A + β2a (black triangles; n = 10), α1A + β2a + Rad[R208A/L235A] (red triangles; n = 11); and α1B + β2a (black diamonds; n = 10), α1B + β2a + Rad[R208A/L235A] (red diamonds; n = 14). P < 0.05, Student’s t test. (E) Exemplar Ba2+ currents from cultured adult guinea pig ventricular cardiomyocytes expressing Rad (Top) and Rad[R208A/L235A] (Bottom). (F) Population Ipeak–V for cardiomyocytes expressing either CFP-Rad (red squares; n = 4) or CFP-Rad[R208A/L235A] (blue squares; n = 8). Dotted line is mean current density for control cardiomyocytes expressing GFP, reproduced from Fig. 2. (G) Exemplar currents (Top) and diary plot (Bottom) showing the impact of 1 μM forskolin on IBa in cardiomyocytes expressing CFP-Rad[R208A/L235A]. (H) Bar chart showing differential impact of forskolin in up-regulating CaV1.2 IBa in cardiomyocytes expressing CFP-Rad or CFP-Rad[R208A/L235A]. Data are means ± SEM.

Rem[R200A/L227A] and Rad[R208A/L235A] Inhibit HVA CaV Channels in Dorsal Root Ganglion Neurons.

Finally, we determined the performance of Rem[R200A/L227A] and Rad[R208A/L235A] as CaV channel inhibitors in primary cells with a complex expression of multiple CaV channel types. We chose dorsal root ganglion neurons which express multiple HVA CaV1/CaV2 channels as well as low voltage-activated (LVA) CaV3.2 channels. Mouse DRG neurons express mostly CaV2.1 and CaV2.2, with a smaller contribution of CaV1.2 and CaV2.3 channels (21). We used adenoviral vectors to robustly express GFP (control) or CFP-tagged RGKs in cultured mouse DRG neurons (Fig. 5). In control cells, a ramp protocol elicited two components of IBa, reflecting currents through LVA and HVA CaV channels, respectively. Overexpressing WT Rad essentially eliminated the HVA current component while leaving the LVA element intact (Fig. 5). We further assessed the impact of various WT and mutant RGKs on the HVA CaV channel currents using step depolarizations. For these experiments, LVA CaV channel currents were eliminated by 5 μM mibefradil and a −50-mV holding potential. Control DRG neurons displayed HVA IBa currents which were dramatically reduced by WT Rad (Fig. 5 ; Ipeak,−10mV = −76.5 ± 13.8 pA/pF, n = 10 for GFP compared with Ipeak,−10mV = −3.5 ± 1.3 pA/pF, n = 19 for CFP-Rad; ). DRG neurons expressing CFP-Rad[R208A/L235A] showed a significant 62% decrease in HVA IBa compared with control (Fig. 5 ; Ipeak,−10mV = −29.8 ± 5.8 pA/pF, n = 13; ). Similar to WT Rad, DRG neurons expressing either CFP-Rem or CFP-Gem showed a dramatically reduced HVA IBa amplitude (Fig. 5; Ipeak,−10mV = −8.7 ± 3.8 pA/pF, n = 8 for CFP-Rem, and Ipeak,−10mV = −4.8 ± 0.9 pA/pF, n = 4 for CFP-Gem; ). Expressing CFP-Rem[R200A/L227A] depressed HVA IBa by 25% compared with control (Fig. 5; Ipeak,−10mV = −56.5 ± 6.6 pA/pF, n = 13; ), substantially less than the reduction observed with CFP-Rad[R208A/L235A]. By contrast, Gem[R196A/V223A] had no impact on HVA IBa in DRG neurons (Fig. 5; Ipeak,−10mV = −74.5 ± 15.4 pA/pF, n = 12; ). Overall, the rank order of inhibition of HVA IBa by these mutant RGKs is consistent with the notion that CFP-Rad[R208A/L235A] inhibits both CaV1.2 and CaV2.2, CFP-Rem[R200A/L227A] inhibits only CaV1.2, and Gem[R196A/V223A] is inert against HVA CaV channels.

Fig. 5.

Differential block of high voltage-activated CaV channel currents in DRG neurons by WT and mutant RGK proteins. (A) Representative images of cultured DRG neurons expressing GFP, CFP-Rad, or CFP-Rad[R208A/L235A]. (B) Exemplar IBa waveforms elicited by voltage-ramp protocols in DRG neurons expressing GFP (Left) or CFP-Rad (Right). (C) Exemplar family of HVA IBa from DRG neurons expressing GFP (Left), CFP-Rad (Middle), or CFP-Rad[208A/L235A] (Right). Currents were elicited from a holding potential of −50 mV and in the presence of 1 μM mibefradil to eliminate LVA T-type currents. (D) Bar chart showing the relative impact of distinct WT and β-binding–deficient mutant RGKs on HVA CaV1/CaV2 channel currents in cultured DRG neurons. Data are means ± SEM. *P < 0.05, one-way ANOVA and post hoc Bonferroni test.

Discussion

Pharmacological blockade of distinct CaV1/CaV2 channel types is an important actual or potential therapy for many diseases, including hypertension (CaV1.2), angina (CaV1.2), cardiac arrhythmias (CaV1.2), chronic pain (CaV2.2), stroke (CaV2), and Parkinson’s disease (CaV1.3) (1, 22, 23). CaV1 channels are effectively blocked by dihydropyridines, benzothiazepines, and phenylalkylamines, while CaV2 channels are inhibited by various animal venoms: ω-agatoxin IVA (CaV2.1), ω-conotoxins GVIA and MVIIA (CaV2.2), and SNX-482 (CaV2.3) (24). Prialt (ziconotide), a blocker of CaV2.2 derived from a marine snail conotoxin, is Food and Drug Administration-approved for the treatment of chronic pain (25). The use of small-molecule CaV1/CaV2 channel blockers is mainly limited by two factors. First, CaV1/CaV2 expression in many types of excitable cells risks prohibitive off-target effects. Second, due to a high degree of similarity among pore-forming α1-subunits (e.g., the L-type channels, CaV1.1 to CaV1.4), currently available small-molecule blockers may not effectively distinguish between CaV channels of the same class. Difficulties encountered in developing CaV1.3-selective blockers as a potential treatment for Parkinson’s disease exemplify these challenges (26, 27). Efficacy of such a treatment approach was suggested by reports that the reliance of substantia nigra neurons on CaV1.3 for pacemaking made them sensitive to Ca2+ overload and vulnerable to cell death which drives the development of Parkinson’s disease (28, 29). Epidemiological studies suggest indeed some beneficial effects of L-type calcium channel (LTCC) blockers in Parkinson’s disease (30). However, because the currently available LTCC blockers are not selective for CaV1.3, off-target effects (e.g., on cardiovascular CaV1.2 channels) risk serious side effects such as hypotension, significantly narrowing the therapeutic window (31).

Genetically encoded CaV channel blockers could offer an alternative solution without the above-mentioned drawbacks of small-molecule inhibitors. Off-target effects might be avoided by restricted expression in target tissues or defined cell populations (9, 10). RGKs are promising candidates for such an alternative treatment approach, given their potency as CaV channel blockers. Their potential usefulness is twofold: (i) as endogenous GECCIs for therapeutic or biotechnological applications, and (ii) as natural prototypes that can help inform strategies to design novel GECCIs for targeted applications in diseases involving excitable cells. Regarding the former, the indiscriminate nature of RGK inhibition of all CaV1/CaV2 channels represents a potential obstacle for some applications. We tested here whether selectivity for particular CaV1/CaV2 isoforms could be engineered into RGKs. Based on the intuition that the indiscriminate manner with which RGKs inhibit all CaV1/CaV2 channel types is a consequence of their binding to auxiliary CaVβ-subunits, we mutated RGKs to eliminate their capacity of binding to CaVβ. This simple maneuver revealed Rem[R200A/L227A] as a CaV1.2-selective blocker and Rad[R208A/L235A] as a selective inhibitor for CaV1.2/CaV2.2. The selectivity of Rem[R200A/L227A] for CaV1.2 over CaV1.3 is noteworthy, given that currently available small-molecule LTCC blockers do not distinguish these channels. Hence, Rem[R200A/L227A] could be a valuable tool for differentially blocking CaV1.2- and CaV1.3-mediated signaling in excitable cells, such as many types of neurons, that coexpress both channel types. Similarly, Rad[R208A/L235A] could be applied to examine CaV1.2/CaV2.2-dependent signaling pathways. The effectiveness of both Rem[R200A/L227A] and Rad[R208A/L235A] in blocking HVA CaV channels in heart cells and DRG neurons demonstrates their utility as selective GECCIs. Additionally, our experiments revealed the existence of BBI mechanisms underlying Rem and Rad inhibition of CaV1/CaV2 channels in excitable cells. This raises the question of the biological significance of BBD versus BBI CaV channel inhibition by RGKs. Our findings in cardiac myocytes suggest indeed functionally relevant differences between the two inhibitory mechanisms. Cardiac CaV1.2 channels are acutely up-regulated pharmacologically by agonists such as BAY K 8644 or physiological activation of PKA initiated by β-adrenergic agonists. The latter contributes to the fight-or-flight response. Rem-inhibited CaV1.2 channels in heart cells can be overridden by BAY K 8644, indicating that the blocked channels remain at the cell surface (32). However, both WT Rem- and Rad-inhibited channels are insensitive to PKA-mediated regulation (20, 32). By contrast, we found that CaV1.2 channels inhibited by either Rem[R200A/L227A] or Rad[R208A/L235A] can be robustly up-regulated by PKA. These results suggest that cardiac CaV1.2 channels inhibited by Rem or Rad through the BBD mechanism are electrically silent, while those inhibited by the BBI pathway are coincidentally activated by membrane depolarization and PKA-mediated phosphorylation. This paradigm could solve the conundrum of how a subset of CaV1.2 channels in heart cells might be reserved for signaling functions other than contraction, given that these channels are voltage-gated and the cardiac sarcolemma is subject to excitation with each heartbeat (33, 34). In this regard, it is noteworthy that GDP-bound Rem and Rad have a lower affinity for CaVβ than their GTP-loaded counterparts (35). We speculate that endogenous Rad toggles between BBD and BBI mechanisms to inhibit cardiac CaV1.2 channels dependent on the G domain being bound to GTP or GDP. Testing this proposition will be an interesting concept for future experiments.

Materials and Methods

Detailed methods are provided in .

Cell Culture and Transfection.

Low‐passage‐number HEK293 cells were transiently transfected with CaVα (6 μg), CaVβ (4 μg), T antigen (2 μg), and RGKs (4 μg) using the calcium‐phosphate precipitation method.

Primary Cell Isolation and Culture.

Primary cultures of adult guinea pig heart cells and murine DRG neurons were prepared and infected with adenovirus as described (14, 36, 37). Procedures were in accordance with the guidelines of the Columbia University Animal Care and Use Committee.

Molecular Biology, Plasmids, and Adenoviral Vectors.

Generation of XFP-tagged RGKs (Rad, Rem, Rem2, and Gem) and β2a,TM has been previously described (12, 15). Adenoviral vectors were generated using the AdEasy XL System (Stratagene) as previously described (38).

Electrophysiology.

Whole‐cell recordings were carried out at room temperature on HEK293 cells 48 to 72 h after transfection as previously described (12). Whole-cell currents were recorded from isolated ventricular myocytes as described (36, 39).

Fluorescence Resonance Energy Transfer Imaging.

We used three‐cube FRET to probe protein–protein interactions (35). Relative Kd and Emax values were calculated as previously described (35, 40).