The C-terminus of Rad is required for membrane localization and L-type calcium channel regulation

L-type Ca_V1.2 current (I_Ca,L) links electrical excitation to contraction in cardiac myocytes. I_Ca,L is tightly regulated to control cardiac output. During stress or exercise when higher cardiac output is demanded, sympathetic input via the β-adrenergic receptor/adenyl cyclase/cAMP/protein kinase A (PKA) cascade results in positive inotropy and lusitropy in contractile cardiomyocytes (Bers, 2002). PKA modulates I_Ca,L where modulation of Ca_V1.2 is defined as increased I_Ca,L amplitude, shifted activation midpoint to a more negative potential, and faster current decay (Bers, 2008).

The L-type Ca²⁺ channel is a heteromultimeric protein complex. Members of the RGK (Rad, Rem, Rem2, and Gem/Kir) subfamily of small Ras-related G-proteins bind to auxiliary Ca_Vβ subunits and govern L-type calcium channel function (Béguin et al., 2001; Finlin et al., 2003). Ras associated with diabetes (Rad, gene name Rrad) was first identified in the skeletal and cardiac muscle of humans with type II diabetes mellitus (Reynet and Kahn, 1993). In heterologous expression systems, a shared property among all RGK proteins is interaction with Ca_Vβ-subunits and inhibition of I_Ca,L (Béguin et al., 2001, 2005; Finlin et al., 2003; Murata et al., 2004; Ward et al., 2004; Chen et al., 2005) via several potential mechanisms including voltage sensor immobilization (Yang et al., 2010), reduced channel open probability, and reduced cell surface expression of Ca_V1 channels (Finlin et al., 2003; Béguin et al., 2006; Correll et al., 2008b; Pang et al., 2010; Sasson et al., 2022, Preprint). Rad has been shown to use both Ca_Vβ-binding-dependent and -independent mechanisms for channel regulation (Yang et al., 2012). Early heterologous expression studies showed that the polybasic C-terminus of Rem is required for membrane targeting and I_Ca,L regulation (Correll et al., 2007; Pang et al., 2010). The RGK C-terminus is highly conserved among family members, suggesting parallel mechanisms for Rad regulation of I_Ca,L in cardiomyocytes.

Heterologous expression studies have proven invaluable for understanding RGK-regulation of I_Ca,L, but data interpretation is limited because of the impossibility of recapitulating native stoichiometries of the multitude of proteins with overlapping interaction domains located within the cardiac L-type Ca²⁺-channel complex (Satin, 2017). We recently demonstrated that myocardial-restricted Rad knockout results in a sustained increase of inotropy without structural or functional remodeling of the heart (Ahern et al., 2019), consistent with a role for Rad as an endogenous negative regulator of Ca_V1.2 voltage-gated calcium channels (Ahern et al., 2019). Using proximity proteomics, Liu et al. (2020) reported that Rad is adjacent to the Ca_V1.2 channel complex in mouse hearts, but is depleted in response to acute β-adrenergic stimulation, suggesting that PKA phosphorylation leads to dissociation of Rad from the Ca_V1.2 complex. The purpose of this study was to address unanswered questions: (1) what is the native subcellular distribution of Rad in cardiomyocytes, and is this distribution altered in response to adrenergic signaling; and, (2) what role does the C-terminus play in control of Rad function. To address these issues, we created two Rad transgenic mouse models: the first was engineered with an N-terminal Flag affinity tag to permit facile immunohistological detection of endogenous Rad protein in cardiomyocytes. The second transgenic mouse model expresses endogenous levels of Flag-tagged Rad bearing a truncation of Rad in the C-terminus to allow analysis of the contribution of this domain to subcellular localization and channel regulation. Using these engineered mouse models, we find that Rad is localized in a striated pattern in cardiomyocytes, in a manner that requires the conserved polybasic C-terminus. Moreover, loss of membrane localization results in a phenotype which mirrors that arising from complete Rad knockout.

All experimental procedures and protocols were approved by the Animal Care and Use Committee of the University of Kentucky Cincinnati Children’s Hospital and conformed to the National Institute of Health’s “Guide for the Care and Use of Laboratory Animals.”

Genome editing was performed by the staff at the Transgenic Animal and Genome Editing Core (TAGE) at Cincinnati Children’s Hospital. CRISPR/Cas9 gene engineering was used to introduce first a 3xFlag epitope tag to the N-terminus of the endogenous mouse Rrad gene, with offspring screened by genomic PCR analysis, and insertion of the 3xFlag epitope confirmed by Sanger sequencing, resulting in a mouse expressing 3xFlag-Rad under control of its native promoter (Flag-Rad). A second round of CRISPR/Cas9 genome-editing was used to modify the 3xFlag-tagged Rad mouse, introducing a stop codon at Ala277 within the extended C-terminus of the RRad gene to generate a second transgenic mouse expressing 3xFlag-Rad bearing a C-terminal truncation (see Fig. 1 A; herein abbreviated Flag-RadΔCT). The sgRNA’s for the A277stop insertion were engineered to contain an AfeI site, with offspring screened by genomic PCR analysis/AfeI digestion, and the inclusion of both the 3xFlag tag and A277 stop codon confirmed by Sanger sequencing (Fig. 1 B). The authenticity of the models was further supported by Western blot analysis of heart lysates from Flag-Rad and Flag-RadΔCT mice (Fig. 1 C). For a Flag-negative control, cardiomyocytes were dispersed from the hearts of cardiomyocyte-restricted Rad-knockout mice as previously described (Ahern et al. 2019).

Ventricular cardiomyocytes were prepared as previously described (Magyar et al., 2012). Prior to heart excision, mice were euthanized by first anesthesia with ketamine + xylazine (90 + 10 mg/kg intraperitoneal) followed by cervical luxation. Hearts were excised from adult mice (3–7 mo of age) and immediately perfused on a Langendorff apparatus with a high-potassium Tyrode buffer and then digested with 5–7 mg liberase (05401135001; Sigma-Aldrich). After digestion, atria were removed and ventricular myocytes were mechanically dispersed. Calcium concentrations were gradually restored to physiological levels in a stepwise fashion, and the criteria for inclusion/exclusion was that only healthy quiescent ventricular myocytes were used for electrophysiological analysis within 8 h. Researchers were blinded to genotype for data collection and analysis. For β-adrenergic receptor agonist studies using immunocytochemistry studies, drug treatment was randomized across aliquots of cells from each dispersal. In lieu of power analysis, sample sizes were determined a priori based on published literature for electrophysiological recordings and adult cardiomyocyte immunocytochemistry. Female mice were used for cellular experiments. In experiments using these isolated cells, mice were considered as biological replicates. In the main and supplemental figures, every mouse and cell as individual data points are shown, and nested statistics were used to treat mice as biological replicates (see Statistics). For anti-Flag microscopy, two to four technical replicate images of different regions of each cell were collected and averaged into one cell average. Detailed sample sizes for each experiment and replicates are reported in figure legends.

I_Ca,L was recorded in the whole-cell configuration of the patch-clamp technique as previously described (Magyar et al., 2012). All recordings were performed at room temperature (20°–22°C). Borosilicate glass pipettes fire-polished to a resistance of 800–3,000 kΩ were filled with a solution consisting of (in mmol/liter) 125 Cs-methano-sulfonate, 15 TEA-Cl, 1 MgCl₂, 10 EGTA, 5 HEPES, 5 Mg-ATP, and 5 phosphocreatine, pH 7.2. Physiological Tyrodes bath solution contained (in mmol/liter) 140 NaCl, 5.4 KCl, 1.2 KH₂PO₄, 5 HEPES, 5.55 glucose, 1 MgCl₂, and 1.8 CaCl₂, pH 7.4. After whole-cell access was achieved, zero sodium bath solution was introduced into the chamber consisting of (in mmol/liter) 150 NMDG, 2.5 CaCl₂, 1 MgCl₂, 10 glucose, 10 HEPES, 0.0313 tetrodotoxin citrate, and 5 4-AP, pH 7.2. L-type calcium currents were evoked by 300-ms depolarizing voltage steps starting from V_rest of −80 mV in +5 mV increments up to +40 mV at 5 s intervals. β-Adrenergic receptor agonist response was recorded in zero sodium bath solution containing 300 nM isoproterenol (ISO) using a 200-ms voltage ramp from −80 to +60 mV repeated once every 3 s. Currents were sampled at 10 kHz. A low-pass filter was applied using an Axopatch 200B (Molecular Devices), digitized with an Axon Digidata 1550B (Molecular Devices), acquired with Clampex 11.2 (Molecular Devices). Clampfit 11.2 (Molecular Device) was used for analysis. Activation voltage dependence parameters were obtained by fitting the current–voltage slope conductance transform to a Boltzmann distribution of the form $G(V)=Gmax/[1+exp(V½/k)]$⁠, where G_max is the maximal conductance and V½ is the activation midpoint. GraphPad Prism 10.1 was used to fit the current–voltage slope conductance transform to a Boltzmann distribution of the form $G(V)=Gmax/[1+exp(V½/k)]$⁠, where G_max is the maximal conductance and V½ is the activation midpoint to determine activation voltage dependence parameters. For current decay kinetics, expressed as r30 and r150, traces were normalized to peak I_Ca,L, and then mean amplitude was measured at 30 and 150 ms after the peak. The remaining current was then calculated by subtracting the mean amplitude from 1.

Transthoracic echocardiography was performed using the Vevo 3100 high-resolution imaging system equipped with a MX550D (25–55 MHz) linear transducer (FujiFilm, VisualSonics, Inc.). Mice were first anesthetized with 2% isoflurane in an induction chamber, chest hair was removed prior to imaging, then a nose cone was placed with inhaled isoflurane, (0.5–1% + 0.5–1.0 liter/min 100% O₂) to maintain a light anesthesia level, with heart rate (350–500 beats per minute) and core temperature (37°C) continuously monitored and maintained by a heated platform. Male and female mice were evaluated. The heart was visualized from the modified parasternal long-axis and short-axis views. The left ventricular dimensions and calculated left ventricular EF were measured from the short-axis M-mode view (Zacchigna et al., 2021). For ISO echocardiography stress testing, mice were injected intraperitoneally with ISO (10 mg/kg) after the baseline long- and short-axis M-mode images and B-mode cine loops of the left ventricle were acquired. The peak stress short-axis M-mode images and B-mode cine loops of the left ventricle were acquired 1–3 min after ISO was given, allowing for a maximal stable increase in heart rate and contractility. Each mouse consisted of three technical replicates that were averaged. Power analysis was not used prior to the study. The sonographer was blinded to animal genotype, and data analysis was performed with animal genotype blinded.

Freshly excised hearts were flash-frozen, and the tissue was pulverized before sonication in cell lysis buffer containing (in mmol/liter) 50 Hepes, pH 7.5, 150 NaCl, 0.5% Triton X 100, 1 EDTA, 1 EGTA, 10 NaF, 2.5 NaVO4, complete Protease Inhibitor Cocktail (11697498001; Sigma-Aldrich), Phosphatase Inhibitor Cocktail 2 (P5726; Sigma-Aldrich), and Phosphatase Inhibitor Cocktail 3 (P0044; Sigma-Aldrich). Lysates were centrifuged at 14,000 rpm and protein concentrations of supernatants were measured. Lysates were analyzed on 4–20% (anti-Ca_V1.2) or 15% SDS PAGE gels. Immunoblotting was performed with anti-Rad (1:2,000, EB11418; Everest Biotech), anti-Flag (1:500, F1804, RRID:AB_262044; Sigma-Aldrich), and anti-Ca_V1.2 (1:500, ACC-003, RRID:AB_2039771; Alomone) antibodies. Proteins were detected using SuperSignal enhanced chemiluminescence (Pierce) or luminol/enhancer solution: peroxide solution (anti-Ca_V1.2) (Bio-Rad). Immunoblots were developed and quantified using the Bio-Rad ChemiDoc MP Imaging System. Total protein loading was assessed by staining the membranes with Ponceau S staining (data not shown). Ca_V1.2 protein quantification was determined with Bio-Rad Image lab software and quantified relative to total protein for each lane (stain free total protein measurement).

No. 1.5H coverslips were coated with poly-L lysine before plating freshly dispersed ventricular murine cardiomyocytes that were allowed to adhere in an incubator (37°C) for 30 min. Cells were then fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) for 10 min; a subset of coverslips were treated with ISO, (1 µM) for 3 min before fixing. Cells were washed thrice before permeabilization with 0.5% triton-X100 in PBS for 10 min. Cells were washed again before blocking for at least 1 h with 3% bovine serum albumin with 0.5% triton in PBS. After washing, cells were incubated with primary antibody in 3% bovine serum albumin in PBS overnight at 4°C. Primary antibodies used were Sigma-Aldrich F1804 mouse IgG1 anti-Flag at 1 µg/ml, Alomone Labs ACC-003 Rabbit IgG anti-Ca_V1.2 at 2.67 µg/ml, and Invitrogen (701914) α-actinin 2 Rabbit IgG (clone: 7H1L69) at 1 µg/ml. The next day, after washing, cells were incubated for 1 h with secondary antibodies: Invitrogen A21202 Alexa Fluor 488 donkey anti-mouse IgG (H+L) at 2 µg/ml and Invitrogen Alexa Fluor A11011 568 goat anti-rabbit IgG (H+L) at 4 µg/ml. After washing, Invitrogen ProLong Glass Antifade Mountant (P36982) 1.52 refractive index hard setting mounting media was used to mount coverslips onto glass slides. The mounting media was allowed to cure for at least 60 h before imaging. A Nikon A1R confocal microscope (Nikon) and Nikon Super Resolution Microscope equipped with a Structured Illumination Microscopy (SIM) module (Nikon) with an oil-immersion CFI Apo 100×/1.49 NA TIRF objective (v) was used. Excitation wavelengths 488 and 561 nm were used to excite Alexa Fluor 488 and Alexa Fluor 568 secondary antibodies, respectively. To quantify the regularity in the periodic signal of Flag, fluorescence intensity plot profiles of confocal micrographs were analyzed via fast Fourier transform (FFT), following methodology performed by other groups analyzing t-tubule and t-tubular protein expression in cardiomyocytes (Hong et al., 2010; Guo and Song, 2014). Intensity plot profiles of rectangular regions of interest of the same size free of nuclear signal were collected from raw 16-bit confocal images using ImageJ (Schneider et al., 2012). A FFT analysis was written in MATLAB (MathWorks) and applied to the exported regions of interest. Individual images of a cell were considered technical replicates and averaged to produce the peak FFT power for a given cell. For live cell t-tubular staining, unladdered isolated ventricular cardiomyocytes were loaded with 10 µM Di-8-ANEPPS (sc-214873; Santa Cruz) for 10 min then washed thrice. Cells were imaged on glass (#1.5) bottom dishes at room temperature with a Nikon AX-R confocal microscope (Nikon) with a Nikon oil-immersion Plan Apo 60×/1.42 NA using an excitation wavelength of 488 nm. T-tubular periodicity was quantified with FFT analysis of fluorescence intensity profiles using MATLAB (RRID:SCR_001622) as done with the anti-Flag FFT analysis. Resting sarcomere length of live cells was measured with brightfield microscopy utilizing a Nikon Eclipse TE200 (Nikon), a Nikon oil-immersion plan fluor 40×/1.30 NA objective, IonOptix MyoCam-S3, MyoPacer Field Stimulator, and Fluorescence System Interface, and analyzed via IonWixard v7.5.3.165. Cells at room temperature were paced at 1 Hz to induce a steady state. Researchers were blinded to genotype when scoring confocal micrographs for t-tubular patterns of anti-Flag staining, and a minimum of five regularly spaced vertical signals was the criteria in consideration of an observed t-tubular-like expression. Mice and cell sample sizes are in figure legends. Only images in which both blinded researchers independently from each other identified expression were counted as true expression.

AlphaFold (AlphaFold2_mmseqs2 notebook of ColabFold; Deepmind) (Jumper et al., 2021; Mirdita et al., 2022) was used to generate a PDB file of a C-termini truncation of the polybasic α-helix that corresponds to replacing amino acid 277 alanine with a stop codon of murine Rad. The generated PDB file was then processed using PPM 3.0 web tool to visualize the orientations of proteins in membranes (Lomize et al., 2012, 2022) and to determine predicted residue interactions with a mammalian plasma membrane. The tool was also used for the AlphaFold full-length Mus musculus Rad PDB file. As per PPM 3.0 authors’ recommendation, AlphaFold regions with very low confidence (pLDDT < 50) were removed before modeling with PPM 3.0 (residues 1–89; 296–308 were excluded). The graphics of the PDB files outputted from PPM 3.0 were generated using the National Center for Biotechnology Information tool, iCn3D (Wang et al., 2020, 2022).

In figures, data are presented as the mean ± SEM unless represented by a box and whisker plot in which case the medians (line), mean (+), and min to max (whiskers) are reported. Linear mixed models were used in experiments that had a hierarchical/nested structure of cells being sampled from the same mouse and treated mice as a random factor with genotype as a fixed factor. Normality and equality of variances were assessed visually and via normality (Shapiro-Wilk) and variance (Levene’s) tests. The P values of pairwise comparisons were corrected for multiple comparisons using Holm adjustment of estimated marginal means. The means and their 95% confidence intervals are reported in the figure legends. In experiments that did not have nested structure, parametric two-sample t-tests or nonparametric Mann–Whitney U test was used for comparisons of two groups, unpaired. For the repeated-measures echocardiography ISO stress test, a three-way ANOVA considering basal (before) versus (after) ISO, sex, and genotype was performed. Since sex as a main effect was not significantly different, separate two-way repeated measures ANOVA’s were used for males and females to focus on the primary questions of genotype and ISO treatment effects. Sidak’s multiple comparison test was used post-hoc to correct for multiple comparisons. For the microscopy FFT data that contained cells fixed with or without drug exposure thus having fixed factors of genotype and drug treatment, the data was right-skewed, not normally distributed, and had unequal variances, and thus was log10 transformed for statistical analysis to mitigate violations of linear mixed model assumptions. For all statistical tests, α was 0.05. Statistical analysis was performed using GraphPad Prism 10.1 (RRID:SCR_002798) and JASP 0.18.3. (JASP Team, 2024) RRID:SCR_015823.

Studies in heterologous expression systems showed that the polybasic C-terminal domain of Rad is necessary for membrane anchoring and regulation of the L-type Ca²⁺ channel; however, non-native systems cannot recapitulate cardiomyocyte L-type Ca²⁺ channel constituents and stoichiometries. To permit analysis of the subcellular distribution of endogenous Rad protein, CRISPR/Cas9 technologies were used to genetically target RRad, introducing three copies of the Flag epitope to the N-terminus of the RRad gene (3xFlag-Rad, see Fig. 1 A). To examine the importance of the conserved C-terminus to Rad function, a second round of CRISPR/Cas9 engineering was used to introduce a stop codon at amino acid position Ala277 using embryos from the 3xFlag-Rad mouse, generating the 3xFlag-RadΔCT mouse model lacking the conserved C-terminal domain (Fig. 1 A). Successful generation of the mouse models was confirmed by Sanger sequencing and Western blot analysis performed on heart ventricle tissue lysates (Fig. 1 B). Importantly, both Flag-Rad and Flag-RadΔCT were expressed at levels indistinguishable from that of endogenous Rad protein. As expected, the Flag-Rad protein migrated more slowly than native Rad on SDS-PAGE, and the 31-amino acid C-terminal deletion in Flag-RadΔCT resulted in a protein that migrated closer to that of endogenous Rad (Fig. 1 C).

Rad is known to bind to Ca_Vβ and regulate I_Ca,L, but to date, there have been no investigations of Rad localization in cardiomyocytes. Furthermore, if the C-terminus tail of Rad plays a role in membrane localization allowing for interaction with Ca_V1.2 in t-tubules, we hypothesize that truncated Rad (Flag-RadΔCT) will display significantly reduced t-tubular localization. Super-resolution structured illumination microscopy (SR SIM) micrographs of primary cardiomyocytes isolated from Flag-Rad mice exhibit regularly spaced signals consistent with t-tubular periodicity (Fig. 2 A and Video 1) as reported in the literature of stained t-tubules or junctional dyad proteins such as Ca_V1.2 (Soeller and Cannell, 1999; Hong et al., 2010; Maleckar et al., 2017; Ito et al., 2019). By contrast, Flag-RadΔCT cells presented with diffuse immunofluorescent staining (Fig. 2 A and Video 2). To quantify the regularity in the periodic signal of Flag-Rad staining (Fig. S2 A), fluorescence intensity plot profiles of confocal micrographs were analyzed via FFT, following a methodology applied by other groups for analyzing t-tubule and t-tubular protein expression in cardiomyocytes (Hong et al., 2010; Guo and Song, 2014) (Fig. 2 B). Median Flag-Rad FFT power was significantly higher (476%) than Flag-RadΔCT (Fig. 2 C). 79% of Flag-Rad cells were classified by blinded researchers as having organized t-tubular expression in at least one region of the cell, an observation corroborated by 76% of Flag-Rad cells showing well-powered fundamental peaks >10 at a periodicity of 1.6 μm (Fig. 2 C; Fig. S1; and Fig. S2 C). In contrast, 7% of Flag-RadΔCT cells were classified as displaying organized t-tubular expression. 22% showed FFT power over 10; however, notably, 0% of Flag-RadΔCT cells showed Flag-expression throughout the entire cell in contrast to 55% of Flag-Rad cells.

Proteins found in the junctional dyad, such as Ca_V1.2, show a periodic signal that repeats with the average murine t-tubular periodicity of 1.8 μm (Soeller and Cannell, 1999; Hong et al., 2010; Maleckar et al., 2017; Ito et al., 2019). The signal in Flag-Rad, but not in Flag-RadΔCT cardiomyocytes, qualitatively colocalizes with anti-Ca_V1.2 staining (Fig. 2 D). Taken together these data suggest that the Rad C-terminus is required for Rad localization to cardiomyocyte membrane domains in common with Ca_V1.2.

The loss of organized, t-tubular pattern expression of anti-Flag-Rad signal in Flag-RadΔCT could result from a general loss of dyad organization. To address this question, primary cardiomyocytes were stained for α-actinin. Cardiomyocytes from Flag-Rad and Flag-RadΔCT presented with highly structured order (Fig. S4 A). T-tubules were probed using live cell staining with di-8-ANEPPS. FFT produced a strong fundamental peak in the power spectra at ∼1.8 μm periodicity in cells of both genotypes, suggesting a regular transverse tubule network that was not significantly different between Flag-Rad and Flag-RadΔCT mice (Fig. S4 B). Brightfield microscopy of live cells demonstrated that both genotypes had mean resting sarcomere lengths of 1.9 μm, suggesting that the 1.6 μm periodicity derived from the anti-Flag cell immunocytochemistry is likely a fixation artifact and not representative of live cell resting sarcomere length (Fig. S4 C) or t-tubular periodicity (Fig. S4 B).

Flag-RadΔCT showed reduced membrane localization suggesting that binding to Ca_Vβ is alone insufficient to generate stable Rad:LTCC complexes, but it is unknown whether durable Rad interaction is required for Rad-mediated channel regulation. L-type channel complexes without Rad show modulated I_Ca,L under basal conditions (Ahern et al., 2019). Fig. 3 A shows representative families of I_Ca,L traces from cardiomyocytes isolated from Flag-Rad and Flag-RadΔCT hearts. The I(V) curve for Flag-RadΔCT shows increased peak current density compared with Flag-Rad (Fig. 3 B). Flag-RadΔCT showed a 5.2-fold increase in maximal conductance (Fig. 3 C) and a hyperpolarizing shift of activation midpoint (shifted −11 mV) compared to Flag-Rad (Fig. 3 D). Western blotting from whole hearts found no significant change in Ca_V1.2 expression between Flag-Rad and Flag-RadΔCT mice (Fig. 3 E).

I_Ca,L in cardiomyocytes isolated from Flag-RadΔCT hearts exhibit enhanced decay kinetics relative to I_Ca,L from Flag-Rad myocytes (Fig. 4). Representative traces normalized to maximum current and superimposed show that Flag-RadΔCT myocytes display faster current decay than that of Flag-Rad (Fig. 4 A). The early component of decay, largely reflecting Ca²⁺-dependent inactivation, was analyzed by assessing the percentage of remaining current 30 ms after peak current. Flag-RadΔCT I_Ca,L showed a significant reduction in mean remaining current (faster decay) across multiple voltages (linear mixed model; genotype P = 0.002, F = 55.5362) (Fig. 4 B and Fig. S3 A). The difference between Flag-RadΔCT and Flag-Rad decay at 30 ms diminished for a more positive V_test as expected for Ca²⁺-dependent inactivation (Ahern et al., 2021). The late I_Ca,L decay was quantified by measuring the percentage of remaining current 150 ms after peak current. Flag-RadΔCT again showed a significant reduction in mean remaining current (faster decay) across multiple voltages (linear mixed model; genotype P = 0.001, F = 80.672) (Fig. 4 C and Fig. S3 B). Taken together, these data suggest that cardiomyocytes expressing Flag-RadΔCT have modulated I_Ca,L under basal conditions similar to that of I_Ca,L seen following complete genetic Rad knockout or in WT cardiomyocytes under acute β-adrenergic receptor agonist stimulation (Manning et al., 2013; Ahern et al., 2019, 2021).

We next tested whether the modulated I_Ca,L observed in Flag-RadΔCT cardiomyocytes under basal conditions recapitulates that of β-adrenergic receptor modulated I_Ca,L. Flag-Rad cardiomyocytes were acutely treated with ISO. To capture the acute, real-time change, the experiment was performed in a paired fashion with a continuous ramp protocol (Fig. 5, A–C). ISO increased Flag Rad I_Ca,L significantly by a mean of threefold with modulated Flag-Rad I_Ca,L resembling that observed in untreated Flag-RadΔCT cardiomyocytes. In contrast to Flag-Rad, ISO stimulation did not increase I_Ca,L in Flag-RadΔCT cardiomyocytes (Fig. 5, A–C).

Recent studies suggested that PKA phosphorylation of Rad releases L-type calcium channel inhibition by loss of Rad within close proximity to Ca_V1.2 and Ca_Vβ (Liu et al., 2020; Papa et al., 2022); therefore, we assessed the localization pattern of Rad after β-adrenergic receptor agonist administration. As in Fig. 2, confocal micrographs from Flag-Rad cardiomyocytes without, or 3 min following ISO stimulation, were analyzed with FFT (Fig. 5 D). The ISO treatment main effect was significant (P = 0.032, F = 5.147) and the median FFT power of Flag-Rad cardiomyocytes treated with ISO was significantly reduced by 62% relative to untreated Flag-Rad myocytes; however, 66% of ISO-treated Flag-Rad cardiomyocytes still showed FFT power over 10, only 10% less than untreated Flag-Rad (Fig. 5 E and Fig. S2). Blinded researchers classified less ISO-treated Flag-Rad cells as having organized t-tubular expression: 60% ISO-treated Flag-Rad versus 79% untreated.

As reported in Fig. 2, which focused on the genotype difference, the main effect of genotype (P = 0.002, F = 16.502) was significant considering all 12 mice and 311 cells ±ISO. Flag-Rad and ISO-treated Flag-Rad cardiomyocytes had mouse median FFT powers in the double-digits while both medians of Flag-RadΔCT ±ISO groups were <10 (Fig. 5 E and Fig. S2 C). Flag-Rad cardiomyocytes untreated and ISO-treated had three- to fourfold greater percentages of cells with fundamental FFT power peaks >10.

Similarly, blinded researchers classified 79 and 60% of Flag-Rad ± ISO and 7 and 0% of Flag-RadΔCT ± ISO cells as having organized t-tubular expression in at least one region of the cell.

We previously showed that constitutive, global Rad-knockout phenocopies β-AR stimulation at multiple levels, from I_Ca,L to in vivo heart function. We next performed an ISO stress echocardiography test in Flag-Rad and Flag-ΔCT mice. The left ventricular (LV) ejection fraction was significantly lower in Flag-Rad compared with Flag-RadΔCT in the basal condition (Fig. 6, A and B). Acute ISO resulted in a significant increase of ejection fraction in Flag-Rad but not Flag-RadΔCT (Fig. 6, A and B). Similar results were obtained from male and female mice (Fig. 6 and Table 1) with the exception that LV mass was greater in male Flag-Rad (Table 1). These results suggest that Flag-RadΔCT heart function phenocopies that of myocardial Rad knockout.

The major finding of this study is that the polybasic C-terminus of Rad is required for membrane localization and for L-type calcium channel regulation in cardiomyocytes. To preserve native protein complexes, we engineered a transgenic mouse model containing an N-terminal Flag epitope, allowing the subcellular distribution of Rad protein to be assessed for the first time at native levels of expression using microscopy. We now report that Rad is enriched in a striated pattern in adult ventricular cardiomyocytes, consistent with the distribution of Ca_V1.2, the main pore-forming subunit of the cardiac L-type calcium channel. In contrast, truncation of the C-terminal results in a significantly reduced striated cellular distribution of Flag-RadΔCT, with a corresponding loss in correlation with Ca_V1.2 staining. In cardiomyocytes expressing Flag-RadΔCT, I_Ca,L is modulated under baseline conditions consistent with properties of myocytes genetically deleted for Rad (Manning et al., 2013; Ahern et al., 2019, 2021), or those expressing full-length Rad following β-adrenergic receptor modulation.

Several independent lines of evidence now support the importance of the evolutionarily conserved C-terminal domain in the biological function of all RGK subfamily G-proteins (Puhl et al., 2014). Heo and colleagues (Heo et al, 2006) first demonstrated that membrane targeting by the isolated RGK C-terminal membrane targeting domain involves the interaction of conserved polybasic amino acids with negatively charged phosphatidylinositol 4,5-bisphosphate and phosphatidylinositol 3,4,5-trisphosphate lipids. Studies in immortalized cell lines support the importance of this conserved domain as an intact C-terminal domain is necessary for Rad- and Rem-dependent I_Ca,L inhibition and plasma membrane association (Finlin et al., 2003; Correll et al., 2007). The C-terminus of Rem and Rem2 is required for phosphatidyl-inositol lipids—RGK association (Correll et al., 2007, 2008a). Furthermore, C-terminal Rem and Rem2 deletion mutants retain their ability to bind Ca_Vβ subunits but fail to inhibit I_Ca,L, indicating that Ca_Vβ subunit binding is alone not sufficient for Rem/Rem2-dependent LTCC regulation (Correll et al., 2007, 2008a). However, fusion of the CAAX membrane targeting motif from Ras proteins to C-terminal truncated Rem (Correll et al., 2007) restored membrane localization and LTCC regulation (Correll JBC 2007), suggesting that Ca_Vβ binding and membrane localization are independent events required for RGK-dependent LTCC regulation. Consistent with results in heterologous cell models, we find that Flag-tagged Rad at endogenous expression levels is localized in repeating t-tubular pattern that colocalizes with Ca_V1.2 staining in cardiomyocytes and that this subcellular distribution depends upon the presence of an intact Rad C-terminal membrane targeting domain.

We used in silico tools to graphically represent the consequence of deleting the Rad C-terminus. Amphipathic α-helixes are common secondary structure motifs that can mediate weak, reversible binding of plasma membranes (Hatzakis et al., 2009). AlphaFold predicts in Rad an amphipathic α-helix from amino acids 273–288 (Fig. 7 A) (Jumper et al., 2021). In this model, helix 8 of Rad contains six Lys or Arg residues that alternate coordinately with hydrophobic residues. AlphaFold (Jumper et al., 2021; Mirdita et al., 2022) was used to generate a model for RadΔCT (Fig. 7, A and B; and Video 3). The PPM3.0 computational tool for predicting membrane interaction was used to calculate the interaction of both proteins with a mammalian plasma membrane (Lomize et al., 2012, 2022) (Fig. 7 A). The predicted ΔG_transfer for Flag-RadΔCT from water to a lipid bilayer was reduced twofold from −7.3 to −2.8 kcal/mol (Fig. 7 C), a change consistent with a loss of membrane association (Fig. 7). In summary, this analysis strongly suggests that the C-terminal domain of Rad is essential for membrane localization.

Regulation of I_Ca,L downstream of activated PKA involves multiple proteins within the L-type calcium channel complex. Our early findings showed that the absence of Rad results in a tonic-modulated I_Ca,L in sinoatrial nodal (Levitan et al., 2021) and ventricular cardiomyocytes (Ahern et al., 2019a), even in the absence of β1- and β2-adrenergic receptor expression (Ahern et al., 2021). We now show that Rad retains membrane localization with acute β-AR stimulation, albeit with significantly reduced power. A proximity proteomic study suggested that Rad was lost from the proximal LTCC microenvironment following β-AR stimulation (Liu et al., 2020; Papa et al., 2022). Acute β-AR stimulation promotes recycling/reinsertion of Ca_V1.2 in a Rab4a- and Rab11a-dependent fashion (Del Villar et al., 2021). Future work will be needed to address whether Rad participates in early endosome-mediated L-type calcium channel recycling (Ito et al., 2019; Westhoff and Dixon, 2021). In Xenopus oocytes, multiple proteins can be expressed approximating tunable stoichiometric ratios (Weiss et al., 2013). Interestingly, Rad-dependent and Rad-independent PKA regulation of I_Ca,L was discovered (Katz et al., 2021). Recent work suggests added complexity of PKA regulation of I_Ca,L arises from multiprotein interactions including tripartite interactions of PKA catalytic subunits with the C-termini of Ca_V1.2. In this vein, it is worth noting that evidence exists for Rem—Ca_V1.2-proximal–C-terminal interaction in a Ca²⁺-calmodulin-dependent fashion, thus raising the specter of penta-partite interactions conferring modulation of I_Ca,L.

We previously reported that Rad slows I_Ca,L kinetics in cardiomyocytes (Ahern et al., 2021). The loss of Rad-regulation of the LTCC increases peak trigger Ca²⁺, and yet, the faster decay kinetics precludes elevation of late I_Ca,L. As shown previously by others, late I_Ca,L is proarrhythmic (Madhvani et al., 2011, 2015). Previous work suggests that disruption of the Rad:CaVβ interaction could be inotropic by increasing Ca²⁺ entry. In this context, it is interesting that deletion of the Rad C-terminus phenocopied Rad-deletion kinetics. This suggests a new therapeutic direction for targeting Rad–C-terminal interaction with the plasma membrane (PIP lipids) as a means for boosting trigger Ca²⁺ for excitation–contraction coupling without necessarily promoting pathologies related to Ca²⁺ overload caused by overactive L-type Ca²⁺ channels.

In conclusion, we show that full-length Flag Rad is expressed in a regular, striated subcellular distribution in cardiomyocytes consistent with a t-tubule expression pattern and localization to the junctional dyad. Deletion of the C-terminus ablates the plasma membrane targeting of Rad in cardiomyocytes, resulting in similar results to that in earlier heterologous expression system studies, showing a lack of membrane expression and instead nuclear and diffuse cytosolic signal. Flag-RadΔCT I_Ca,L is modulated under basal conditions similarly to that of I_Ca,L of β-AR–treated cardiomyocytes, both recapitulating the increased open probability, negative shift of activation midpoint, and hastened inactivation observed in β-adrenergic receptor modulation of the L-type calcium channel. With no C-terminus to anchor it to the membrane in t-tubules, we observed no evidence of a Rad interaction with Ca_V1.2; thus, a similar phenotype to that of Rad knockout occurs (Manning et al., 2013; Ahern et al., 2019, 2021).

Fig. S1 contains 16 confocal micrographs of fixed adult murine ventricular cardiomyocytes immunostained with anti-Flag antibody in addition to control micrographs (e.g., a Rad knockout mouse, a no primary control). Fig. S2 contains 16 confocal micrographs similarly treated except the cells were incubated with 1 µM ISO for 3 min to supplement Fig. 5 D’s two representative images. Fig. S2, B and C show the non-log₁₀-transformed mice means, and log₁₀ and non-transformed individual cell data is shown in Fig. S2 C and Fig. 5 E. Fig. S3 and Fig. S4 show averaged data with individual mouse and cell data and the 95% confidence intervals of the genotype differences at each V_test. Fig. S4 displays microscopy assessing t-tubule integrity and quantitates resting sarcomere length in live cardiomyocytes. Video 1 SR SIM of maximal intensity projection (multiple z-slices) of Flag-Rad stained with Flag antibody. Scale x = 32.9 μm; y = 32.9 μm; z = 4.1 μm. Video 2 SR SIM of maximal intensity projection (multiple z-slices) of Flag-RadΔCT stained with Flag antibody. Scale x = 32.9 μm; y = 32.9 μm; z = 4.9 μm. Video 3 Alphafold full-length Rad (Uniprot accession no. O88667) PPM 3.0 computational model of interaction with the plasma membrane.

The data are available from the corresponding author upon reasonable request.

David A. Eisner served as editor.

This research was funded by an American Heart Association pre-doctoral fellowship to B.M. Ahern (19PRE34380909), NIH/National Heart, Lung and Blood Institute grant HL166280, and Department of Defense grant W81WXH-20-1-0418.

Author contributions: G. Elmore: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing—original draft, Writing—review & editing, B.M. Ahern: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Visualization, Writing—review & editing, N.M. McVay: Formal analysis, Investigation, Writing—review & editing, K.W. Barker: Formal analysis, Investigation, Methodology, Validation, Visualization, Writing—original draft, S.S. Lohano: Data curation, Formal analysis, Investigation, Visualization, N. Ali: Data curation, Formal analysis, Investigation, Methodology, Writing—review & editing, A. Sebastian: Investigation, Methodology, Resources, Writing—review & editing, D.A. Andres: Conceptualization, Funding acquisition, Investigation, Methodology, Supervision, Validation, Writing—review & editing, J. Satin: Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing, B.M. Levitan: Data curation, Formal analysis, Investigation.