Myocardial-restricted ablation of the GTPase RAD results in a pro-adaptive heart response in mice

Development of conditional cardiac-restricted RAD-deficient transgenic mice

Standard gene targeting of RRAD in the mouse results in a complex phenotype, including cardiac remodeling (21) and fibrosis (22), with increased cardiac output (18). To examine the role of RAD signaling selectively in cardiomyocytes, we used conditional gene targeting with the Cre-loxP system. To achieve cardiomyocyte-specific deletion, CRISPR/Cas9 techniques were used to genetically target RRAD, introducing loxP (flox: fl) recombination sites flanking exons 3 and 4 (RAD^fl/fl; Fig. 1A). These exons encode a large portion of the RAD GTPase core that directs both guanine nucleotide binding and hydrolysis, and the C-terminal membrane targeting motif required for LTCC regulation (27) such that any transcribed mRNA that is translated following recombination would produce a truncated protein lacking biological activity. Successful introduction of loxP sites was confirmed by genomic PCR (Fig. 1B). To permit cardiomyocyte-specific RAD deletion, RAD^fl/fl mice were crossed with a mouse line expressing a tamoxifen-inducible Cre recombinase under control of the α-myosin heavy chain promoter (αMHC-CreERT2) (28). RAD is abundantly expressed in the heart and other tissues (8, 13, 25, 26, 29), and administration of tamoxifen to RAD^fl/fl-MHC mice (Fig. 1A) resulted in RAD deletion from the myocardium (RAD^Δ/Δ) but not the spleen (Fig. 1C). Disruption of RAD expression in the heart was confirmed by qRT-PCR (Fig. 1D). RAD^fl/fl without αMHC-CreERT2 (RAD^fl/fl) mice served as littermate controls, with RAD expression unaffected following tamoxifen challenge (Fig. 1, C and D). Unless stated, all RAD^Δ/Δ and RAD^fl/fl mice were subjected to analysis >2 weeks following tamoxifen exposure.

Cardiac-restricted RAD-deficient mice (RAD^Δ/Δ) show improved function with no changes in heart dimensions

Global constitutive RAD knockout mice (gRAD^−/−) showed increased I_Ca,L (18) along with hypertrophic fetal gene program expression, structural remodeling (19), and fibrosis (22). By contrast, ANF and RCAN1, markers of the fetal gene program, were not altered in RAD^Δ/Δ (Fig. 2A). There was also no difference in the ratio of heart to body weight (Fig. 2B) and RAD^Δ/Δ shows no significant difference in fibrosis compared with RAD^fl/fl (Fig. 2, C and D). This is in stark contrast to gRAD^−/− hearts, in which RAD deficiency has been found to increase connective tissue growth factor (CTGF) expression, leading to greater extracellular matrix production and basal cardiac fibrosis (22). Importantly, cardiomyocyte-restricted RAD deletion failed to increase CTGF expression (Fig. 2, E and F). Global RADKO mice experience hypertension (50), and this may be a significant driver of the cardiac remodeling observed in these mice, as gRAD^−/− hearts have likely undergone adaptive alterations secondary to pressure overload. Importantly, RAD^Δ/Δ showed no change in aortic pressure (Fig. 2G). Taken together, these data demonstrate that myocardial-restricted RAD deficiency improves cardiac output without instigating structural remodeling.

To test the effect of myocardial-restricted RAD deficiency on heart function we longitudinally followed RAD^Δ/Δ mice from induced knockout in early adulthood for an additional 15 months (Fig. 3A). Ejection fraction (EF) significantly increased 7 days after tamoxifen administration and remained increased continuously for 15 months (Fig. 3B). There was no difference in heart chamber size (Fig. 3C) or in heart wall thickness (Fig. 3, D and E). Cardiomyocyte-restricted RAD deletion effects on in vivo heart function were not different between males and females (Fig. 3, F–H). EF has a shallow dependence on heart rate (HR) that is preserved in RAD^Δ/Δ. Regression analysis supports elevated EF in RAD^Δ/Δ across the range of HR measured (Fig. 3I).

RAD^Δ/Δ modification of voltage-gated calcium current, I_Ca,L

A straightforward explanation for enhanced heart function would be increased I_Ca,L in RAD^Δ/Δ. Previous studies of whole body, constitutive RAD knockout showed that LTCC conductance increased, voltage dependence shifted negatively, and the predicted I_CaL window increased (18); however, constitutive global RAD knockout hearts displayed structural remodeling. Thus, it is possible that changes in gRAD^−/− I_Ca,L were secondary to myocardial remodeling. To interrogate the contribution of RAD in structurally unaltered hearts, we measured I_Ca,L in cardiomyocytes from mature RAD^Δ/Δ hearts. Fig. 4A shows representative families of I_Ca,L traces from RAD^fl/fl and RAD^Δ/Δ for evaluating current/voltage (I(V)) relationships, and Fig. 4B shows representative current traces generated by the steady-state availability voltage protocol. These representative traces highlight the increased current density and accelerated kinetics of RAD^Δ/Δ. The I(V) curves for I_Ca,L show that RAD loss results in a greater current density compared with RAD^fl/fl myocytes (Fig. 4C). RAD deficiency results in greater maximal conductance (Fig. 4, D and E; maximal conductance: RAD^Δ/Δ = 254 ± 19 pS/pF, n = 15; RAD^fl/fl = 144 ± 12 pS/pF, n = 18; p < 10⁻⁴). The normalized conductance voltage curves superimposed on the steady-state inactivation curves (Fig. 4F) highlight the selective shift of steady-state activation with RAD ablation; activation midpoint is shifted toward more negative membrane potential (Fig. 4, F and G; RAD^Δ/Δ = −18.3 ± 1.0 mV, n = 15; RAD^fl/fl = −8.1 ± 1.9 mV, n = 18; p < 10⁻⁴). Steady-state availability shows no significant difference of midpoint between RAD^fl/fl and RAD^Δ/Δ (RAD^Δ/Δ = −24.2 ± 1.0 mV, n = 9; RAD^fl/fl = −22.1 ± 1.7 mV, n = 11; p = 0.11). There is no sexual dimorphism in the effect of RAD deletion on I_Ca,L (data not shown). The increase in I_Ca,L could stem from an increase in Ca_V1.2. To test this, we performed qRT-PCR for Ca_V1.2 from RAD^fl/fl and RAD^Δ/Δ hearts and found Ca_V1.2 mRNA did not change (Fig. 4H). Ca_V1.2 protein levels from RAD^Δ/Δ hearts was also not significantly different from that of RAD^fl/fl (Fig. 4I). Taken together these data show that myocardial-restricted RAD deficiency is sufficient to increase I_Ca,L by modulating LTCC function.

RAD^Δ/Δ I_Ca,L decay is biphasic with a prominent fast and slow decay phase compared with monophasic decaying RAD^fl/fl (Fig. 4A; Fig. S1). At −5 mV the fast and slow decaying amplitudes of RAD^Δ/Δ I_Ca,L are roughly similar (Fig. S1); the slow-decay component is significantly slower than RAD^fl/fl decay. The fast I_Ca,L decay is >10-fold faster than the monotonic I_Ca,L time constant observed in RAD^fl/fl (Fig. S1). In summary, myocardial RAD-reduction results in larger, but faster decaying I_Ca,L.

RAD^Δ/Δ enhances cellular calcium homeostasis

I_Ca,L provides trigger Ca²⁺ for myocardial Ca²⁺-induced Ca²⁺ release. Whole-cell cytosolic Ca²⁺ dynamics are significantly larger in amplitude for RAD^Δ/Δ compared with RAD^fl/fl (Fig. 5, A and B), and the upstroke velocity of the cytosolic Ca²⁺-transient is accelerated in RAD^Δ/Δ (Fig. 5C). To assess Ca²⁺ re-uptake we measured the time constant of the Ca²⁺ transient decay (τ). The value of the mean τ of RAD^Δ/Δ was faster than that for RAD^fl/fl but did not reach statistical significance using ratiometric imaging of fura2 in cardiomyocytes (Fig. 5D). We augmented ratiometric imaging with a separate series of experiments using fluo4 for high speed 2-dimensional cytosol-restricted Ca²⁺ dynamics. A benefit of fluo4 imaging is the elimination of contaminating, slower nuclear Ca²⁺ signals. Two-dimensional cytosolic Ca²⁺ decay was statistically significantly faster in RAD^Δ/Δ than RAD^fl/fl (Fig. 5E; RAD^Δ/Δ = 0.07 ± 0.003, n = 67, RAD^fl/fl = 0.10 ± 0.005, n = 69; p < 10⁻⁴). To probe mechanisms of re-uptake we assessed SERCA2a protein levels (Fig. 5F). RAD^Δ/Δ hearts expressed elevated SERCA2a protein levels (p < 0.05) but unchanged phospholamban (PLN) protein levels (see below). Taken together these data suggest that faster I_Ca,L trigger (Fig. 4) and increased SERCA2a expression conspire to contribute to accelerate Ca²⁺ transient dynamics following RAD knockout.

Sarcomere dynamics is an important determinant of myocardial function (30). We therefore measured sarcomere length and dynamics (Fig. 6A) simultaneously with calcium transients. Elevated calcium dynamics (Fig. 5) accompanying RAD deficiency coincided with unchanged resting sarcomere length (Fig. 6, A and B), increased fractional shortening (Fig. 6C; RAD^Δ/Δ = 16.1 ± 0.9, n = 24, RAD^fl/fl = 9.9 ± 1.0, n = 15; p < 10⁻⁴), and the rate at which the cell shortened (Fig. 6D; RAD^Δ/Δ = −4.8 ± 0.4, n = 24; RAD^fl/fl = −2.8 ± 0.3, n = 15; p = 0.0002). Overall, these results suggest an increase in contractility in RAD^Δ/Δ compared with RAD^fl/fl, consistent with cellular dynamics driving improved heart function with RAD^Δ/Δ.

Davis et al. (31) introduced a predictive model of cardiac growth based on integrated tension measured from mean twitch tensions or sarcomere shortening. A negative score of this tension-integral model predicts dilated cardiomyopathy; a positive score trends toward hypertrophic remodeling. Using this model, we calculated the index score for our RAD^Δ/Δ model to predict the direction and intensity of hypertrophy. The RAD^Δ/Δ tension index score is +0.03 relative to RAD^fl/fl (Fig. 6E), predicting only a modest tendency toward concentric hypertrophy. This provides novel evidence that increasing trigger calcium does not necessarily promote significant cardiac hypertrophy and pathological remodeling.

Basal I_Ca,L in RAD^Δ/Δ cardiomyocytes reflects a modulated I_Ca,L

The increase of I_Ca,L in response to β-adrenergic stimulation is atop a constellation of effectors that enhances cardiac output in the fight or flight response. We therefore next tested the effect of acute isoproterenol (ISO) treatment on I_Ca,L and calcium dynamics. RAD^fl/fl ventricular cardiomyocyte I_Ca,L conductance was significantly increased (Fig. 7, A and Bi, Fig. S2); however, ISO had no detectable effect on RAD^Δ/Δ I_Ca,L (Fig. 7, A and Bii, Fig. S2) resulting in ISO-treated RAD^fl/fl conductance rising to a level that was indistinguishable from that of unstimulated RAD^Δ/Δ cardiomyocytes (Fig. S2). Comparison of the voltage-dependence of activation of I_Ca,L reveals that the acute ISO effect in RAD^fl/fl, whereas shifted negative, remained more positive than that for RAD^Δ/Δ (Fig. 7, D and E, Fig. S2). Acute ISO had no detectable effect on I_Ca,L kinetics (data not shown). Two summary conclusions are apparent: 1) RAD^Δ/Δ I_Ca,L is maximally modulated; and 2) RAD^fl/fl–modulated I_Ca,L remains at a more positive activation midpoint potential than that for RAD^Δ/Δ. We conclude that basal RAD^Δ/Δ phenocopies β-AR–modulated RAD^fl/fl I_Ca,L.

RAD^Δ/Δ basal heart function is elevated in vivo, yet retains a partial ISO response

Given that RAD deficiency results in I_Ca,L resembling β-AR augmentation we next performed an echocardiography ISO stress test in RAD^fl/fl and RAD^Δ/Δ mice at 1 and 12 months after tamoxifen administration. Acute in vivo ISO resulted in a significant gain of EF in RAD^fl/fl and in RAD^Δ/Δ mice at 1-month post-tamoxifen treatment (Fig. 8). The ISO effect was preserved at 1 year (16-month-old mice). We conclude that myocardial-restricted RAD deletion provides stable inotropic support, without globally disrupting myocardial β-AR responsiveness.

RAD^Δ/Δ retains β-adrenergic receptor modulation of Ca²⁺ and sarcomere dynamics

To explore the source of ISO responsiveness in RAD knockout cardiomyocytes we next interrogated Ca²⁺ and sarcomere dynamical response to ISO. Acute ISO enhanced calcium dynamics in RAD^Δ/Δ and RAD^fl/fl ventricular cardiomyocytes (Fig. 9A). For cytosolic Ca²⁺ handling, RAD^fl/fl ISO response demonstrated an increase in amplitude, a faster upstroke velocity, and a faster rate of calcium reuptake compared with basal conditions, as expected (Fig. 9, B–D, Fig. S3). RAD^Δ/Δ ISO response also demonstrated an increase in amplitude, faster upstroke velocity, and a faster rate of Ca²⁺-reuptake. Compared with RAD^fl/fl with ISO, RAD^Δ/Δ with ISO had a relatively larger amplitude (RAD^Δ/Δ = 2.6 ± 0.1, n = 49; RAD^fl/fl = 2.0 ± 0.2, n = 37; p = 0.008), faster upstroke velocity (RAD^Δ/Δ = 63 ± 3.0, n = 49; RAD^fl/fl = 47 ± 5.0, n = 37; p = 0.009), and faster rate of calcium reuptake (RAD^Δ/Δ = 0.11 ± 0.003, n = 49; RAD^fl/fl = 0.13 ± 0.006, n = 38; p = 0.05; Fig. S3). To further test for active β-AR signaling in RAD^Δ/Δ we assessed PLN-serine 16 phosphorylation status. At baseline, PLN Ser-16^P and Thr-17^P relative to total PLN protein were not different between RAD^Δ/Δ and RAD^fl/fl (Fig. 9E). Acute ISO stimulation significantly increase phosphorylation of PLN–Ser-16^P protein from RAD^Δ/Δ hearts (Fig. 9F) providing a basis for accelerated Ca²⁺-reuptake.

To test for preservation of acute β-AR signaling to contractile function we measured sarcomere dynamics in response to isoproterenol (Fig. 10A). Fractional shortening (Fig. 10B, Fig. S4) and sarcomere relaxation was accelerated in RAD^Δ/Δ (Fig. 10C, Fig. S4). Acute ISO elevated the tension-integral index in RAD^fl/fl approaching the level observed for RAD^Δ/Δ (Fig. 10, D and E). To further investigate contractile mechanisms underlying the response to ISO, we assessed the phosphorylation status of sarcomeric proteins that modulate contractility (32, 33). Acute ISO significantly increased phosphorylation of cMyBPC and TnI (Fig. 10F). These data support the conclusion that RAD^Δ/Δ does not exhibit de-sensitized myocardial β-AR signaling. The reduction of RAD imposes a chronic modulated I_Ca,L while preserving β-AR modulation of critical targets, such as sarcoplasmic reticulum Ca²⁺-reuptake and contractile apparatus proteins.

Increased calcium does not necessarily promote cardiac hypertrophy

A major finding of this study is that myocardial RAD deletion, with increased I_Ca,L, did not demonstrate structural remodeling (Fig. 3). Previous studies have reported a connection between calcium influx through the LTCC and adverse growth of the heart wall (24, 36, 37). Increasing I_Ca,L through overexpression of the pore subunit α_1C or the β2 subunit of the LTCC led to cardiac hypertrophy, and eventual heart failure growth (38). An I_Ca,L phenotypic difference with these over-expression models and RAD^Δ/Δ is in the kinetics of I_Ca,L. RAD^Δ/Δ demonstrates a biphasic decay of current with fast and slow components (Fig. 4, Fig. S1). The time course of I_Ca,L is reminiscent of calcium-dependent fast inactivation contributed by the calcium load of the sarcoplasmic reticulum (39). Fast decay in RAD^Δ/Δ I_Ca,L attributed to a large calcium release from the SR might serve as a regulatory negative feedback mechanism to limit total LTCC Ca²⁺ flux, and in turn might not necessarily activate pathways to initiate pathological hypertrophy. Thus, even if dynamical total Ca²⁺ influx was increased in the intact myocardium, there may not be obligatory hypertrophic signaling by LTCC (40).

Over-expression of SERCA2a preserves myocardial function under conditions that otherwise would be expected to produce failure (41–44). RAD^Δ/Δ hearts display increased SERCA2a expression without a detectable change in PLN phosphorylation (Figs. 5 and 9), resulting in a change in the ratio of SERCA2a to PLN that is consistent with overall faster Ca²⁺ transient decay dynamics. This favorable alteration toward less regulated SERCA2a complexes in our model has also demonstrated cardiac protection in high-intensity exercise models (41) and PLN-null mice (47), as well as in phase 1 and phase 2 trials of patients suffering from advanced heart failure (43, 44). To summarize, we observe contemporaneous mechanistic changes in RAD^Δ/Δ myocardium that plausibly sum to a stable gain of heart function including: increased, faster-decaying trigger Ca²⁺ and increased SERCA2a.

Davis et al. (31) proposed a tension-integral index model that distinguishes between hypertrophic and dilated cardiac remodeling. A key insight of the tension-integral index study was that the tension-integral predicted myocardial structural changes, but there were poor correlations to changes in calcium handling indices (31). The RAD^Δ/Δ approximates a WT tension-integral as a consequence of larger amplitude contractions offset by faster kinetics. Regardless, the RAD^Δ/Δ heart showed elevated EF. Thus, how RAD^Δ/Δ accommodates the increased demands of chronic increased cardiac function will be an important area of follow-up studies.

In a similar vein to sarcomere dynamics, the faster kinetics of I_Ca,L in RAD^Δ/Δ could also potentially serve to attenuate arrhythmogenic potential. Mahajan et al. (46) proposed modifying I_Ca,L kinetics as a superior way to prevent arrhythmias, rather than through inhibiting I_Ca,L activation. This would allow early peak I_Ca,L to initiate normal SR calcium release, and diminish the effects of late I_Ca,L on repolarization. As encouraging as the present results are for suggesting myocardial RAD deletion as a therapeutic direction for cardioprotection, thorough analysis of arrhythmogenic potential will require further studies, including, but not limited to alternative model systems to the mouse.

The RAD^Δ/Δ model also differs from other models of increased I_Ca,L because responsiveness to β-adrenergic receptor (β-AR) stimulation is retained. A hallmark feature of HFrEF is insensitivity of β-AR to chronic stimulation (47) that ultimately limits the contractile reserve of the heart. RAD^Δ/Δ hearts retain β-AR responsiveness at the level of the cell (Figs. 9 and 10) and whole heart (Fig. 8), and did not develop heart failure. I_Ca,L did not change after treatment with isoproterenol in our RAD^Δ/Δ model (Fig. 7), similar to overexpression of the α_1C model (12). I_Ca,L was larger in RAD^Δ/Δ than in RAD^fl/fl stimulated with ISO. This suggests that RAD may play a critical role in the ability for the LTCC to be modulated by β-AR stimulation; when RAD is deleted, basal I_Ca,L is larger than currently seen following typical modulation. Further studies are warranted to define the molecular role for RAD in β-adrenergic Ca²⁺ current modulation. Finally, enhancing Ca²⁺ dynamics via targeting of β-AR signaling by G-protein coupling receptor kinase (48) or Raf kinase inhibition (49) also provides protection against pressure overload induced heart failure. In summary, increasing Ca²⁺ dynamics in a manner that bypasses chronic β-AR activity is a potentially attractive target of therapeutic strategies for heart failure therapy.

Myocardial-restricted RAD deletion differs from whole body RAD deletion

Key differences exist between RAD^Δ/Δ and the whole body deletion of RAD leading to the existing erroneous conclusion that RAD-reduction is pro-hypertrophic (19). In the global RAD knockout model (gRAD^−/−), channel function and cellular calcium dynamics were similar to RAD^Δ/Δ, and in both models the myocardium retained responsiveness to β-AR stimulation (18). However, myocardial RAD deficiency promotes enhanced heart function without structural remodeling or fibrosis. A major consideration is that gRAD^−/− mice experience hypertension (50), whereas RAD^Δ/Δ do not. This may result from loss of RAD in vascular smooth muscle (25, 45); increased I_Ca,L in vascular smooth muscle could lead to increased vascular resistance to progress toward hypertension. Regardless, gRAD^−/− hearts have likely undergone adaptive signaling and therefore interpretation of studies of Ca²⁺ dynamics in gRAD^−/− must consider the myriad of myocardial alterations secondary to pressure overload.

In summary, our findings suggest that increasing myocardial calcium handling does not necessarily promote pathological remodeling, in fact, by imposing a chronic modulated I_Ca,L phenotype in the basal state we show that heart function increases stably into senescence. These findings are consistent with the notion that decreased RAD in heart failure patients (17, 19) contributes to a compensatory response. There is a major unmet need for treatments targeting inotropy (1). We conclude that myocardial RAD deletion might provide a potential new therapeutic target for positive inotropy without introducing damaging effects to the heart.

Animal model

Transgenic animals were generated on a C57BL/6J background in the Transgenic Animal and Genome Editing Core at Cincinnati Children's Hospital Medical Center. The conditional RRAD allele was made via a sequential insertion strategy using CRISPR-Cas9 technology, in which exons 2 and 3 were flanked by Cre recombinase-dependent loxP (flox: fl) recognition sequences (see Fig. 1A). The sgRNA's were engineered to contain either HindIII (5′) or EcoRI (3′) restriction sites, with offspring screened by genomic PCR analysis, and loxP insertion confirmed by Sanger sequencing. These mice (RAD^fl/fl) were then crossed (or not, as controls) with mice expressing a tamoxifen-inducible Cre recombinase (MerCreMer) under α-myosin heavy chain promoter (28) to produce RAD^fl/fl-MHC animals. RAD deficiency was induced by a single intraperitoneal injection in control (RAD^fl/fl) and experimental mice with tamoxifen dissolved in sunflower seed oil (40 mg/kg body weight). All mice received tamoxifen and were used ≥2 weeks after tamoxifen treatment unless otherwise stated. This single tamoxifen injection protocol minimizes cardiomyopathological effects observed with multiday administration of tamoxifen (35). The age of the oldest mice used for study were limited to 18 months, but the cellular studies were from mice <12 months of age.

Immunoblotting

Hearts freshly excised from RAD^fl/fl and RAD^Δ/Δ were flash frozen, and tissue was pulverized using Freezer Mill (6775, SPEX SamplePrep) 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 Mixture (Roche 11697498001-1 tablet per 50 ml), phosphatase inhibitor mixture 2 (Sigma, P5726, ×100 solution), and phosphatase inhibitor mixture 3 (Sigma, P0044, ×100 solution) (500 μl of each inhibitor mixture). Lysates were centrifuged at 14,000 rpm and protein concentrations of supernatants were measured. Lysates were prepared in 4× SDS-PAGE loading buffer with 40.0 μg of total protein per lane resolved on 8, 15, or 4–20% SDS-PAGE gels and transferred to nitrocellulose/PVDF membranes. Immunoblotting was performed with: calsequestrin (1:3000, Abcam); RAD (1:2000, Everest Biotech); Cav1.2 (1:1000, Alomone); CaMKIId (1:1000, Santa Cruz Biotechnology); pCaMKII (1:1000, Santa Cruz Biotechnology); SERCA2a (1:1000, Badrilla); phospholamban phosphoserine 16 (1:1000, Badrilla); phosphor-threonine 17 (1:1000, Badrilla), and total PLN (1:1000, Invitrogen) antibodies, CTGF (1:1000, Santa Cruz Biotech) and GAPDH (1:1000, Cell Signaling Technology). Proteins were detected using SuperSignal enhanced chemiluminescence (Pierce), and immunoblots were developed and quantified using the Bio-Rad ChemiDoc MP Imaging System. The total protein loaded was quantified by staining the membranes with Coomassie staining after the detection of the desired proteins or using stain-free gels (Bio-Rad).

Quantification of myofibril protein phosphorylation

Myofibrils were prepared from frozen mouse ventricular tissue homogenized in relaxing solution containing protease and phosphatase inhibitors (PhosSTOP and cOmplete ULTRA Tablets, Roche Applied Science). Samples were skinned for 15 min using 1% Triton X-100, centrifuged at 10,000 × g for 5 min, and resuspended in fresh relax. Samples were reduced using 2-mercaptoethanol in Laemmli buffer and boiled for 5 min. 5 μg of the solubilized myofibrils were loaded and electrophoretically separated using 4–20% Tris glycine minigels at 160 V for 70 min. Gels were fixed and stained with Pro-Q diamond phosphoprotein stain (Invitrogen) per the manufacturer's protocol. The same gels were then stained with Coomassie Blue for total protein expression. Densitometry of bands in Pro-Q and Coomassie-stained gels was performed using ImageJ software (National Institutes of Health). The degree of phosphorylation of a given protein was determined by the ratio of the Pro-Q signal to the protein contented loaded.

Ventricular myocyte isolation

Isolated ventricular cardiomyocytes were prepared as previously described (15). Prior to heart excision mice were anesthetized with ketamine + xylazine (90 + 10 mg/kg intraperitoneal). Hearts were excised from adult RAD^fl/fl and RAD^Δ/Δ and immediately perfused on a Langendorff apparatus with a high-potassium Tyrode buffer and then digested with 5 to 7 mg of liberase (Roche Applied Science). After digestion, atria were removed and ventricular myocytes were mechanically dispersed. Calcium concentrations were gradually restored to physiological levels in a stepwise fashion, and only healthy quiescent ventricular myocytes were used for electrophysiological analysis or calcium imaging within 12 h.

Single-cell assays