0β-Adrenergic induced SR Ca²⁺ leak is mediated by an Epac-NOS pathway

Abstract

Cardiac β-adrenergic receptors (β-AR) and Ca²⁺-Calmodulin dependent protein kinase (CaMKII) regulate both physiological and pathophysiological Ca²⁺ signaling. Elevated diastolic Ca²⁺ leak from the sarcoplasmic reticulum (SR) contributes to contractile dysfunction in heart failure and to arrhythmogenesis. β-AR activation is known to increase SR Ca²⁺ leak via CaMKII-dependent phosphorylation of the ryanodine receptor. Two independent and reportedly parallel pathways have been implicated in this β-AR-CaMKII cascade, one involving exchange protein directly activated by cAMP (Epac2) and another involving nitric oxide synthase 1 (NOS1). Here we tested whether Epac and NOS function in a single series pathway to increase β-AR induced and CaMKII-dependent SR Ca²⁺ leak. Leak was measured as both Ca²⁺ spark frequency and tetracaine-induced shifts in SR Ca²⁺, in mouse and rabbit ventricular myocytes. Direct Epac activation by 8-CPT (8-(4-chlorophenylthio)-2′-O-methyl-cAMP) mimicked β-AR-induced SR Ca²⁺ leak, and both were blocked by NOS inhibition. The same was true for myocyte CaMKII activation (assessed via a FRET-based reporter) and ryanodine receptor phosphorylation. Inhibitor and phosphorylation studies also implicated phosphoinositide 3-kinase (PI3K) and protein kinase B (Akt) downstream of Epac and above NOS activation in this pathway. We conclude that these two independently characterized parallel pathways function mainly via a single series arrangement (β-AR-cAMP-Epac-PI3K-Akt-NOS1-CaMKII) to mediate increased SR Ca²⁺ leak. Thus, for β-AR activation the cAMP-PKA branch effects inotropy and lusitropy (by effects on Ca²⁺ current and SR Ca²⁺-ATPase), this cAMP-Epac-NOS pathway increases pathological diastolic SR Ca²⁺leak. This pathway distinction may allow novel SR Ca²⁺ leak therapeutic targeting in treatment of arrhythmias in heart failure that spare the inotropic and lusitropic effects of the PKA branch.

Graphical abstract

1. Introduction

β-adrenergic receptor (β-AR) mediated myocyte Ca²⁺ mishandling is commonly described in heart failure (HF) and arrhythmia, which are increasing rapidly [1]. Therefore understanding fundamental mechanisms of β-AR effects on Ca²⁺ handling is critical. β-AR activation is an integral part of the cardiac fight-or-flight response, but its chronic activation (e.g. in HF) contributes to pathological hypertrophic remodeling, contractile dysfunction and arrhythmia.

Acute β-AR activation enhances cardiac contraction (ionotropy) and relaxation (lusitropy), in large part by increasing myocyte Ca²⁺ transients and accelerating [Ca²⁺]_i reuptake by the sarcoplasmic reticulum (SR). These effects are mainly produced by PKA-dependent phosphorylation of L-type Ca²⁺ channels which increases Ca²⁺ current (I_Ca), and phospholamban (PLB) which enhances SR Ca²⁺ uptake and content [Ca²⁺]_SRT (making more Ca²⁺ available for release; Fig 1, right). The higher Ca²⁺ transient causes stronger contraction, functionally offsetting myofilament Ca²⁺ desensitization by PKA (that otherwise participates in lusitropy) [2]. Studies have also shown that β-AR can also sensitizes ryanodine receptors (RyR) gating and SR Ca²⁺ leak, but while PKA can phosphorylate RyR, the functional effects on RyR are mediated via CaMKII activation and phosphorylation of RyR [3–18]. Chronic CaMKII activation and consequent increase of SR Ca²⁺ is now generally accepted as part of the HF syndrome [19–21]. Indeed, SR Ca²⁺ leak can contribute directly to both diastolic and systolic dysfunction in HF as well as β-AR triggered arrhythmias, via Ca²⁺ wave-induced inward Na⁺/Ca²⁺ exchange current (I_NCX) that causes delayed afterdepolarizations (DADs), triggered action potentials and premature ventricular contractions (PVCs) [3, 4, 22, 23]. Therefore, inhibition of SR Ca²⁺ leak is a valid therapeutic strategy in HF [19–21] which could improve systolic and diastolic function and limit β-AR-induced arrhythmias.

Figure 1. Proposed pathway for β-AR-induced increase of diastolic SR Ca²⁺ leak.

β-AR activation stimulates G-protein (G_s) dependent activation of adenylyl cyclase causing cAMP production that activates both Epac and PKA. The PKA branch enhances Ca²⁺ current (I_Ca) and phospholamban (PLB) sensitive SR Ca²⁺-ATPase (ATP). The Epac branch activates a cascade leading to NOS- and CaMKII-dependent RyR2 phosphorylation that promotes SR Ca²⁺ leak. Broken lines indicate a previously held idea that two parallel pathways might mediate Epac and NOS effects on RyR2.

Two pathways have been independently implicated to mediate β-AR activation of CaMKII and SR Ca²⁺ leak. One pathway involves exchange protein directly activated by cAMP (Epac), a cAMP target parallel to PKA, which may involve some downstream Epac targets leading to CaMKII autophosphorylation and RyR phosphorylation RyR (Fig 1, blue) [10–18]. Indeed, we have shown that this pathway specifically requires β₁-AR, Epac2 (which localizes at myocyte T-tubules), CaMKIIδ and RyR2 phosphorylation at S2814 [14, 16]. For this pathway, it is clear that cAMP is involved and that PKA only contributes indirectly (via PLB-dependent increase in SR Ca²⁺ load), but the details of the pathway from the Epac-Rap1 level to CaMKII are not well resolved (with several mediators implicated, Fig 1) [10–17].

The other β-AR to CaMKII-RyR pathway involving nitric oxide synthase 1 (NOS1), seemed cAMP-independent, but involving protein kinase B (PKB or Akt) as upstream activators of NOS1dependent CaMKII activation via S-nitrosylation (Fig 1, red) [3–8, 24]. But in this case, the steps upstream from the β-AR to Akt were less clearly defined. This pathway was thought to be independent of cAMP and Epac because neither forskolin (direct adenylyl cyclase activator) nor 8-CPT (selective Epac agonist) mimicked β-AR effects [4–6]. This raised the idea of β-arrestin mediated signaling to CaMKII [24], as a parallel pathway from (β-AR to Akt and NOS1; Fig 1). Recent studies that have revealed the molecular mechanism by which S-nitrosylation occurs and mediates CaMKII activation [7, 25], localization of NOS1 at the junctional SR domain,[9] and robust evidence for cardiac CaMKIIδ in regulating RyR2 have solidified our understanding of the bottom part of the NOS1-RyR pathway.

Here, two labs that helped characterize these two pathways have worked together to further test whether these apparently parallel pathways from β-AR to CaMKII-RyR might be related in series. Much of the work leading to the Epac pathway involved measurement of Ca²⁺ sparks as an index of SR Ca²⁺ leak, often used 8-CPT as an Epac agonist and did not explore NOS involvement. The studies leading to NOS1 involvement in β-AR-induced SR Ca²⁺ leak, more often used the Shannon-Bers method of tetracaine-induced Δ[Ca²⁺]_i and Δ[Ca²⁺]_SRT shifts to measure SR Ca²⁺ leak. Here, we find similar results regardless of SR Ca²⁺ leak method, that 8-CPT (when freshly prepared) mimics the β-AR effects on tetracaine-sensitive SR Ca²⁺ leak, and that NOS mediates Epac-dependent increase of Ca²⁺ sparks, and that Akt is involved. We conclude that these pathways are largely in series (Fig 1).

2. Materials and Methods

2.1. Myocytes Isolation

Cardiac myocytes were isolated from New Zealand white rabbits and C57BL6 mice using retrograde Langendorff perfusion using Liberase TM (0.075 mg/mL, Roche) and Trypsin (0.0138%, Gibco) (37°C) as previously described [26]. All procedures were approved by the University of California Davis Institutional Animal Care and Use Committee (IACUC) in accordance with the NIH Guide for the Care and Use of Laboratory Animals.

2.2. Ca²⁺ Spark Measurements

Spontaneous Ca²⁺ spark were measured in quiescent cardiomyocytes loaded with 5 μM Fluo-4 AM (Molecular Probes) for 30 min [14]. Images were recorded using confocal microscopy in line scan mode (Bio-Rad Radiance 2100, 40× oil immersion objective, 6 ms). Fluo-4 AM was excited at 488 nm using an Argon laser and the emission was collected at 505 nm. All experiments were done in Tyrode’s solution (in mM: 140 NaCl, 4 KCl, 1.1 MgCl₂, 10 HEPES, 10 glucose, 1.8 CaCl₂; pH7.4 with NaOH). Analysis was made using Firefly, a homemade routine in Python™, which uses analytical criteria similar to other widely used analysis methods [27].

8-CPT is a cAMP analog that is highly selective for Epac vs. PKA activation [28, 29]. The 8-CPT concentration used here (10 μM) has no effect on myocyte Ca²⁺ handling or Ca²⁺ sparks in mouse myocytes lacking Epac2 [14] and does not accelerate twitch [Ca²⁺]_i decline [15], in contrast to very potent effects of PKA to accelerate twitch [Ca]_i decline [14]. Moreover, direct measurements of PKA activity in intact ventricular myocytes showed no significant increase in PKA activation at 10–100 μM 8-CPT [16]. Thus, as used here 10 μM 8-CPT is selective for Epac-vs. PKA-mediated Ca²⁺ handling effects.

2.3. SR Ca²⁺ leak Measurements

The protocol used to measure SR Ca²⁺ leak in rabbit was as previously described (Fig 2) [5]. Briefly, [Ca²⁺]_i was measured using a fluo-4 (Invitrogen) signal in isolated myocytes in the presence and absence of SR Ca²⁺ leak. Images were recorded using confocal microscopy in line scan mode (Zeiss LSM DUO in Live Channel mode, 40× water immersion objective, 2 ms/line). Tetracaine was used to rapidly and reversibly block the RyR, thus disrupting the SERCA pump-leak balance (Figure 2). The tetracaine-dependent shift of Ca²⁺ from the cytosol to the SR (decrease in [Ca²⁺]_i and increase in SR Ca²⁺ content) is proportional to SR Ca²⁺ leak.

Figure 2. Exemplar Tetracaine-sensitive SR Ca²⁺ leak measurement in intact rabbit myocyte.

[Ca²⁺]_SRT is varied by stimulation rate. After the last field stimulated twitch in normal Tyrode’s (NT, left) tetracaine blocks RyR-mediated leak in Na⁺-free, Ca²⁺-free solution (0 Na, 0 Ca; to prevent sarcolemmal Ca²⁺ flux). Blocking leak causes net SR Ca²⁺ uptake such that [Ca²⁺]_i decreases and [Ca²⁺]_SRT increases (inset). Tetracaine wash-off reverses the shift. [Ca²⁺]_SRT is measured by rapid caffeine application and Δ[Ca²⁺]_i. Leak flux is proportional tetracaine-induced Δ[Ca²⁺] (purple arrows) and Δ[Ca²⁺]_SRT (black vs. green arrows) [32].

Myocytes were subjected to a protocol to load the SR in a graded manner: by emptying the SR with 10 mM caffeine followed either by 30 sec of rest, 30 sec of rest followed by one single stimulation, or field stimulation at 0.25 to 1.0 Hz. Field stimulations at given rates were performed for at least 20 beats to approach steady state cellular and SR Ca²⁺ content.

After one of the above loading protocols the bath solution was rapidly switched to 0 Na, 0 Ca²⁺ NT, 1 mM tetracaine. Without Na⁺ and Ca²⁺ in the bath, NCX, the primary Ca²⁺ efflux mechanism at rest, was blocked so that Ca²⁺ was entrapped in the resting cell [30]. The RyR (and therefore leak) is blocked by tetracaine and the measured resting fluorescence decreases as Ca²⁺ is taken up into the SR (Figure 2) [5]. Fluo-4 fluorescence was corrected for a 4% quench by tetracaine whenever it was present. Fluorescence was monitored for 30 s followed by another rapid solution switch to 0Na, 0Ca²⁺ NT with no tetracaine added. With the SR Ca²⁺ leak restored, diastolic [Ca²⁺]_i rises back to its resting value. Finally, 10mM caffeine in nominally Ca²⁺ free NT was added to cause SR Ca²⁺ release. The total SR [Ca²⁺] ([Ca²⁺]_SRT) was calculated as the difference between the basal and peak total cytosolic [Ca²⁺] ([Ca²⁺]_T) in the presence of caffeine. The difference in [Ca²⁺]_SRT in the presence and absence of tetracaine (the same as the difference in resting [Ca²⁺]_T) is due to the leak dependent shift of Ca²⁺ from the cytosol to the SR (i.e. the difference in basal [Ca²⁺] with and without tetracaine) and the leak rate is proportional to this shift.

Data were selected such that the average [Ca²⁺]_SRT was the same in all groups to ensure that the measured tetracaine-dependent SR Ca²⁺ leak was independent of SR [Ca²⁺].

2.4. Fluorescence Resonance Energy Transfer (FRET)

Adenovirus encoding Camui [31] was transfected in cultured cardiomyocytes for 2–4h (M.O.I 10–100)(5% CO2, 37 °C) in PC-1 medium. Experiments were performed 24h after transfection. FRET was measured using confocal microscopy in frame scan mode (Zeiss LSM5 Pascal, ×40 water immersion objective). FRET signal was measured as an increase of the F_CFP/F_YFP ratio with λ_ex set at 458 nm (Ar laser) for CFP (donor) and λ_ex set at 510 nm for YFP (acceptor) [31]. Donor fluorescence emission was detected at wavelengths of 470–500 CFP and acceptor fluorescence measured at ≥530 nm. Fluorescence images were analyzed using ImageJ software (NIH).

2.5. Western Blotting

Freshly isolated rabbit ventricular myocytes were plated on to laminin (Sigma-Aldrich; 1 mg/ml) coating the wells of 6-well tissue culture plates for further treatment in culture medium containing 2.0 mmol/L CaCl₂. 8-(4-Chlorophenylthio)-2″-O-methyladenosine 3′,5′-cyclic monophosphate-sodium salt (8-CPT) and Nω-Nitro-L-arginine methyl ester hydrochloride (L-NAME) were obtained from Sigma-Aldrich. LY924002 was from EMD Millipore.

Adhered ventricular myocytes were either pretreated with L-NAME (100 μM), LY924002 (10 μM) or dimethyl sulfoxide (DMSO) for 15 min followed by the addition of 8-CPT (10–25 μM) with gentle swirling to achieve the final desired dilution and incubated for 10–15 min.

Upon completion of experimental treatments, the overlying solution was aspirated and ventricular myocytes were lysed with the immediate addition of hot 1-X Laemmli sample buffer; without β-mercaptoethanol (β-ME) or bromophenol blue dye. Lysates were recovered to a microcentrifuge tube, heated to 95°C for 5 min and stored at −20°C. Sample protein determinations were made using a BCA protein assay kit (Pierce) followed by the addition of β-ME and dye to the samples in order to obtain an appropriate 1-X Laemmli buffer final-volume concentration for each and were again heated as before.

Utilizing standard electrophoresis protocols for SDS-PAGE and western blotting, cell lysates were separated using either pre-cast 4 % or 4–20 %-gradient tris-glycine mini-gels (Novex™, Invitrogen) and transferred to nitrocellulose. Equivalent sample loading of 20–60 μg of protein per well was required. Primary antibodies used for western blotting were directed against phospho-AKT (Ser473; #4690), and pan-AKT (#4690) from Cell Signaling Technology and anti-ryanodine receptor 2 (human RyR2; Alomone labs), or phospho-RyR2 (human, pSer2808 or pSer2814; Badrilla). Purified monoclonal anti-α-actinin immunoglobulin (EA-53; Sigma-Aldrich) was also used to evaluate sample loading. Species specific horseradish peroxidase-conjugated secondary antibodies (Jackson Immunoresearch Labs) were used and visualization was accomplished with Western Lighting™ chemiluminescence reagents (PerkinElmer) and directly imaged using a charge-coupled device (CCD) sensor with controlling software (Marconi Applied Technologies).

For the evaluation of Epac-mediated changes in phospho-RyR, blotted panels were first probed with either anti-pSer2808 or pSer2814 and then stripped completely using 2% SDS-buffer. The efficacy of antibody complex removal was verified by incubation of secondary antibody only for the typical time and antibody dilution with imaging. Next, each panel was re-probed with anti-RyR2 for evaluation of uniform protein loading. Protein bands were quantified (arbitrary units; a.u.) using ImageJ software (NIH) and phospho-RyR was normalized to the corresponding RyR2 band for each group. Statistical significance (P <0.05) was determined using unpaired Student’s t test to compare Control group values to 8-CPT-treated values (n=3) for each experimental group.

3. Results

3.1. Interdependence of Epac and NOS upon β-AR-induced SR Ca²⁺ Leak

Previous work studying the role of Epac in β-AR-induced SR Ca²⁺ leak via Ca²⁺ sparks had not tested nitric oxide or NOS involvement (although NOS1 inhibition was shown to prevent ISO-stimulated increase in total SR Ca²⁺ leak [6]). Figure 3A and 3C show that 100 nM ISO increased Ca²⁺ spark frequency by approximately 100% (0.22±0.048 in CTL vs. 0.46±0.12 in ISO, p<0.05). The NOS inhibitor, L-NPA (5 μM) did not significantly alter basal Ca²⁺ spark frequency, but prevented the ISO-induced increase in Ca²⁺ sparks (0.19±0.05 vs. 0.46±0.12). This is consistent with prior studies of total leak (using the tetracaine method in Fig 2) [6] and supports the idea that β-AR activation increased SR Ca²⁺ leak via NOS activation.

Figure 3. β-AR- and Epac-induced Ca²⁺ sparks.

A, B. Raw Ca²⁺ sparks in intact mouse ventricular myocytes. C. Ca²⁺ spark frequency (CaSpF) was increased by ISO (100 nM), but not by NOS-inhibitor L-NPA (5 mM, 10 min). However, L-NPA blocked ISO-induced increase in CaSpF (n=9, 6, 5 and 5 myocytes). D. 8-CPT (10 mM, 2 min) induced increased CaSpF, while neither NOS inhibition (100 μM L-NAME, 5 min) nor PI3K inhibition (10 μM LY294002) altered baseline CaSpF. Both L-NAME and LY294002 prevented 8-CPT-induced CaSpF (n=21, 10, 13, 13, 7 and 7 myocytes, respectively; * p<0.05). Resting fluorescence was not significantly altered by L-NPA, 8-CPT, L-NAME or LY94002, but was slightly increased (15%) by ISO, which may be secondary to higher CaSpF.

We next tested whether direct specific Epac activation of Ca²⁺ sparks (by 10 μM 8-CPT) [14, 15] would be sensitive to NOS inhibition. Figures 3B and D show that the robust Ca²⁺ spark activation by 8-CPT was prevented by 100 μM L–NAME (which inhibits NOS). Again, L-NAME alone did not alter basal Ca²⁺ sparks, indicating (as in Fig 3C) that this NOS signaling to Ca²⁺ sparks is minimal prior to activation by ISO or 8-CPT (CTL: 0.10±0.04, 8-CPT:0.29±0.11 vs. L-NAME; 0.12±0.06). These results indicate that NOS is, in fact activated downstream of Epac and mediates the Epac-induced activation of Ca²⁺ sparks, and may mediate the β-AR-induced and Epac-dependent activation of Ca²⁺ sparks.

Our previous work measuring tetracaine-sensitive SR Ca²⁺ leak also showed that ISO-induced increase was reliant upon NOS activity, but was not mimicked by 8-CPT [6]. We revisited this issue since full 8-CPT efficacy requires fresh solution preparation for each experiment, which we were unaware of during that study. Figure 4 shows the results of these experiments in rabbit ventricular myocytes stimulated to steady-state at varying frequencies to obtain a range of [Ca²⁺]_SRT (Figure 2). A rapid switch to Ca²⁺-and Na⁺-free Tyrode’s solution blocks the Na⁺/Ca²⁺ exchanger, keeping the total cellular [Ca²⁺] constant. Tetracaine (1 mM) blocks the RyR causing a shift of Ca²⁺ from the cytosol into the SR (lowering [Ca²⁺]_i and raising [Ca²⁺]_SRT, proportional to the SR Ca²⁺ leak) [32, 33]. We found that Epac activation with 8-CPT increased SR Ca²⁺ leak (Figure 4B 19.2±2.4 vs. 12.3±1.6 μmol/l cytosol ∆[Ca²⁺]_SRT, n=11,18, p<0.05) when considering only data for the same average [Ca²⁺]_SRT (Figure 4A, 120.9 vs. 120.6 μmol/l cytosol), consistent with the spark data above. Furthermore, 100 μM of the NOS inhibitor L-NAME completely prevented this increase (10.1±1.3 μmol/l cytosol ∆[Ca²⁺]_SRT, n=11), confirming the involvement of NOS is the pathway. We conclude that Epac can mimic the β-AR-induced SR Ca²⁺ leak, as seen with Ca²⁺ spark measurements, and that the prior study [6] missed this because of a technical limitation (see corrigendum link to be added in proof). Here we used freshly prepared perfusate, to which 8-CPT was added from frozen stock, just prior to the experiment.

Figure 4. 8-CPT and ISO-induced SR Ca²⁺ leak in rabbit myocytes.

The tetracaine-induced approach (Fig 2) was used to assess SR Ca²⁺ leak (Δ[Ca²⁺]_SRT, bottom) in myocyte records with matched SR Ca²⁺ content ([Ca²⁺]_SRT, top). SR Ca²⁺ leak induced by 10 μM 8-CPT was prevented by NOS inhibition (100 μM L-NAME; n= 18, 11, 11 myocytes; * p<0.05).

3.2. β-AR-dependent CaMKII activation is by a Epac/NOS-dependent pathway

In order to test that each pathway operates via CaMKII activation downstream of both Epac and NOS, we used the FRET-based activity reporter Camui [31] (Figure 5A) to measure the response of CaMKII to both ISO and 8-CPT treatment. Figures 5B & C show that ISO and 8-CPT significantly activate CaMKII, as measured by a reduction of FRET (shown as an increase in F_CFP/F_YFP). Furthermore, consistent with the SR Ca²⁺ leak data above, the NOS inhibitor L-NAME (100 μM) prevented the Epac activation of CaMKII (Figure 5B) and the NO-donor SNAP mimicked the CaMKII activation by both ISO and 8-CPT. This also shows that SNAP prevented further CaMKII activation by 8-CPT (Figure 5C), consistent with no further 8-CPT activation of already S-nitrosylated CaMKII [6, 7, 25]. Note also that PKA inhibition (H89) did not significantly reduce ISO-induced CaMKII activation (Figure 5C).

Figure 5. CaMKII activation via Epac.

A. Schematic of the FRET-based activity reporter Camui, where activation (opening of regulatory (Regul) and catalytic domains causes reduced FRET between cyan and yellow fluorescent proteins (CFP and YFP). Sites of autophosphorylation (P), S-nitrosylation (N), oxidation (O) and O-GlcNAcylation (GlcNAc) are indicated. B. Rabbit myocyte images of CFP and FRET (YFP emission) under indicated treatments. C. Mean Camui signals at baseline (CTL), with 10 μM 8-CPT alone (30 min) or after pretreatment with 100 μM L-NAME or 10 μM LY294002 (n=41, 44, 13 and 10, respectively). D. Mean Camui signals in control rabbit cardiomyocytes at baseline (CTL), ± 10 μM 8-CPT (30 min), ISO (100 nM) ± H89 (1 μM) or SNAP ± 8-CPT (n =33, 24, 13, 17, 18 and 14 for each bar, respectively).

We also tested whether the phosphorylation of RyR in response to Epac activation was dependent on NOS, using Western blots with site-specific antibodies against pRyR-S2809 and pRyR-S2815. Rabbit cardiomyocytes were stimulated ±8-CPT with and without L-NAME. Figure 6 shows that 8-CPT promoted RyR phosphorylation at both the site known to be a CaMKII target (S2815) as well as S2808 (best known as a PKA target, but can also be a CaMKII target [34]).

Figure 6. Epac-mediated increase in RyR phosphorylation is NOS-dependent.

Western blots for total RyR and RyR phosphorylated at S2809 (A) and S2815 (B). Right panels show an Epac-dependent increase in phosphorylation at both residues in response to 8-CPT treatment, which was prevented by 100 μM L-NAME (*p<0.05 vs. without 8-CPT, t-test, n=3).

All together these data (Figures 3–6) demonstrate that β-AR activation activates CaMKII-dependent increase SR Ca²⁺ leak via Epac-mediated NOS1 activation. The fact that Epac-dependent SR Ca²⁺ leak is also NOS-dependent clearly shows that these two independently developed pathways are, indeed, parts of the same pathway with NOS1 downstream of Epac and with CaMKII downstream of both. The use of the tetracaine protocol also confirms that these effects on the SR Ca²⁺ leak are not simply the result of an increase of SR Ca²⁺ load. Finally, the effect is not species-dependent, as it is present in both mouse and rabbit, and it is observable both in Ca²⁺ spark and tetracaine-sensitive SR Ca²⁺ leak measurements.

3.3. Epac/NOS-dependent SR Ca²⁺ leak is mediated by PI3K and Akt

The data above are consistent with our proposed unified series pathway for β-AR activation of SR Ca²⁺ leak in two distinct areas (Figure 1). This work indicates that β-AR activation and subsequent generation of cAMP lead to activation of Epac. Furthermore, farther down the pathway, activation of NOS1 results in nitrosylation and activation of CaMKII and generation of increased SR Ca²⁺ leak via RyR phosphorylation. But the pathway between Epac and NOS1 is still less clear. The remaining experiments aim to clarify this further.

Previous data suggested that ISO-induced Akt activation might be involved in the NOS1 activity that leads to increased SR Ca²⁺ leak [6]. In order to confirm and extend this idea, we performed Western blots with antibodies against activated pAkt-S473 in myocytes stimulated ± 8-CPT and flash frozen as above for the RyR. To our surprise, initial results revealed no difference in phosphorylation levels ± 8-CPT (Figure 7A). Noticing that the baseline Akt phosphorylation was rather high in control, we pretreated the myocytes with the phosphoinositide 3-kinase (PI3K) inhibitor LY924002. This reduced baseline Akt phosphorylation levels, and revealed an enhancement of Akt phosphorylation upon application of 8-CPT (Figure 7B & C, 100.0±7.6 vs. 373.2±63.4, n=3, p<0.05, t-test).

Figure 7. Epac activation by 8-CPT and Akt phosphorylation.

A. 8-CPT does not appear to increase phospho-473 Akt (p-Akt^Ser473). B. After 10 μM LY924002 pretreatment to decrease background phosphorylation, 8-CPT causes demonstrative phosphorylation of Akt. C. Mean data from B (P<0.05, t-test, n=3).

It is evident from the experiments above that inhibition of PI3K strongly influenced global Akt phosphorylation. However, the requirement for local PI3K stimulation in the Epac-dependent activation of SR Ca²⁺ leak is unknown. To test this, we performed Ca²⁺ spark experiments with the PI3K inhibitor LY294002 (Figure 3D). LY294002 inhibited the ability of 8-CPT to promote Ca²⁺ spark frequency. In contrast LY294002 alone had no effect on Ca²⁺ sparks. The data suggest that the activation of Akt, leading to increased SR Ca²⁺ leak, is dependent upon PI3K activity. Considering these Ca²⁺ spark results (Fig 3D) together with the need to inhibit global PI3K activity to detect the 8-CPT-induced effect on Akt phosphorylation might mean that local Akt near NOS1, CaMKII and RyR may be a more important readout, than is global Akt. In conclusion, Epac2 activation of NOS1 upon β-AR appears to be mediated by Akt activation via PI3K (Fig 1).

4. Discussion

Prior work had demonstrated two apparently independent parallel pathways by which cardiac β-ARs activate CaMKII-dependent diastolic SR Ca²⁺ leak involving Epac2 (Fig 1, blue) [10–18] and NOS1 (Fig 1, red) [4–9]. Here, we have demonstrated that these are part of a single series signaling cascade involving both Epac and NOS (Fig 1). Key novel findings here demonstrate such a link. Indeed, direct Epac activation by 8-CPT: 1) activates Ca²⁺ sparks, an effect prevented by inhibition of NOS or PI3K (as seen for ISO), 2) increases tetracaine-sensitive SR Ca²⁺ leak in rabbit (overturning prior conclusions [6]) and again this was NOS-dependent, 3) activates myocyte CaMKII in a NOS- and PI3K-dependent manner (as seen for ISO [6]), and 4) promotes NOS-dependent RyR S2815 phosphorylation. We conclude that β₁-AR activation triggers a series cascade via cAMP-Epac2-PI3K-Akt-NOS1-CaMKIIδ to cause phosphorylation of RyR at S2815 to increase pathological diastolic SR Ca²⁺ leak.

4.1. Epac activation mimics β-AR-induced SR Ca²⁺ Leak in rabbit and mouse

We had previously reported that direct Epac activation by 8-CPT failed to activate SR Ca²⁺ leak in rabbit ventricular myocytes using the tetracaine-sensitive SR Ca²⁺ leak method, and inferred that the Epac2- and NOS1-dependent pathways were independent in signaling from β-AR to RyR2 [6]. However, this was at odds with much of our prior mouse and rat work, where 8-CPT reliably increased SR Ca²⁺ leak assessed as Ca²⁺ sparks [14–17]. That included a study in which mice with genetic knockout of Epac1, Epac2 and CaMKIIδ, as well as knock-in of mutant RyR2 (S2814A and S2814D) were used for molecular dissection of these pathway components [14]. This left several possible explanations. First, the Epac-dependence seen in mouse and rat might be absent in rabbit. Second, the Epac-dependence might only be evident when SR Ca²⁺ leak is measured using Ca²⁺ sparks and not by total tetracaine-sensitive SR Ca²⁺ leak measurements. That is, Epac might promote more spark-mediated leak, but equally reduce spark-independent leak [35–37]). Third, there might be a technical problem. Working together we realized that 8-CPT storage conditions are important for efficacy, which sharply declines in aqueous solution at 23°C during an experimental day [14–17]. Here, we repeated the prior rabbit ventricular myocyte 8-CPT experiments [6], and found that freshly prepared 8-CPT induced a robust increase in SR Ca²⁺ leak that was prevented by NOS inhibition (Fig 4). These new rabbit results mesh well with our mouse data where the 8-CPT-induced Ca²⁺ spark enhancement was prevented by NOS inhibition (Fig 3). So the prior negative 8-CPT result [6] was likely a now-resolved technical issue, and we infer that this Epac-NOS-CaMKII-RyR pathway is likely universal in mammals (at least rabbit, mouse and rat).

Our novel data with the direct CaMKII activity reporter in rabbit myocytes shows that both ISO and 8-CPT similarly activate CaMKII and that that depends on both NOS and PI3K activity. This agrees with recent work demonstrating that CaMKII can be activated by direct S-nitrosylation of Cys-290 in the CaMKIIδ regulatory domain [7, 25]. This S-nitrosylation causes CaMKII to be trapped in the autonomous open (active) conformation, which can also expose the neighboring autophosphorylation site (T287). This may explain the previously reported 8-CPT-induced CaMKII autophosphorylation [15]. That is, the data in Fig 5B suggests that without the 8-CPT-induced S-nitrosylation of CaMKII, autophosphorylation does not occur (because that itself would strongly promote the active open conformation [31]). Thus S-nitrosylation may promote autophosphorylation which could further prolong CaMKII activation. Taken together, the data support the idea that Epac and NOS are in the same pathway, and that NOS is downstream of Epac2 (Fig 1).

4.2. Local Nature of the β -AR-induced SR Ca²⁺ Leak

Our prior work leading to NOS involvement in this pathway [4–6], indicated that bulk activation of adenylyl cyclase by forskolin did not mimic the ISO-induced SR Ca²⁺ leak. We interpreted this as evidence for a potentially cAMP-independent pathway to CaMKII activation [24]. The demonstrable involvement of Epac (which is cAMP-dependent) has caused us to re-interpret this observation. Forskolin certainly increases global [cAMP], but increasing evidence shows that β-AR can induce discrete local cAMP microenvironments relatively independent of global cAMP concentration [38]. Our findings that Epac2, which mediates the 8-CPT effect [14], is highly localized, along with RyR, along T-tubules [16], is consistent with this interpretation. That is, we hypothesize very local β-AR, Epac, NOS1 and CaMKIIδ signaling to RyR2 (where all 5 proteins are known to be concentrated). We think this is a plausible hypothesis, but details may require further testing (e.g. with targeted fluorescence reporters).

Another aspect of our present results that is consistent with highly localized signaling, is the PI3K-Akt results. Prior data [6] indicated that β-AR activation of SR Ca²⁺ leak was Akt-dependent. So, we expected that Epac activation would stimulate Akt. While we did find that 8-CPT increased Akt phosphorylation at S473 this was demonstrable only after inhibiting basal global Akt phosphorylation (Figure 7). This might be because the local pool of Akt participating in this pathway is a small (local) fraction of overall myocyte Akt. While precise details of the PI3K- and Akt-dependence of NOS1 activation merit further study, both kinases seem to be involved [6].

4.3. Remaining pathway gaps in the series from β-AR to RyR

This series arrangement is now molecularly well-defined at the top (β₁-AR, cAMP, Epac2) and bottom (Akt-NOS1-CaMKII-RyR) ends (Fig 1). However, some of the implicated intermediaries connecting Epac2 to Akt are less completely resolved molecularly. While Rap1 and PI3K seem critical at exactly this point, the involvement of other previously implicated components downstream of Epac (PLCε, PKCε and maybe InsP₃ receptors [12, 13, 17]) is less clear. There is also precedent for Epac-induced activation of PI3K in the regulation of Kv potassium channels in pancreatic β-cells Epac [39].

4.4. cAMP signaling via PKA to inotropy/lusitropy vs. via CaMKII to SR Ca²⁺ leak

The effects of β-AR to enhance Ca²⁺ transient amplitude and decline kinetics is mediated largely by PKA-dependent phosphorylation of L-type Ca²⁺ channels and phospholamban. This is undisputed and causes increased SR Ca²⁺ load, release and faster relaxation. PKA-dependent effects on myofilament proteins (troponin I, myosin-binding protein C and titin) the Na⁺/K⁺-ATPase and other ion channels also modulate the integrated β-AR effects on contractility and diastolic function [2]. An important distinction is that these well characterized cAMP effects via PKA operate in parallel to the cAMP-Epac-CaMKII pathway that we describe here. CaMKII phosphorylates many of the same protein targets as PKA (often at different sites). While cAMP-PKA has stronger functional effects than CaMKII on Ca²⁺ current and SR Ca-ATPase, the opposite is true for RyR2 (where CaMKII effects are much more functionally robust [40, 41]).

4.5 SR Ca²⁺ Leak as a Potential Therapeutic Target

Since elevated diastolic SR Ca²⁺ leak in HF (and other pathologies) can contribute to reduced systolic and diastolic function and also arrhythmogenesis, it is a potential therapeutic target [42]. We have shown that the enhanced SR Ca²⁺ leak in HF can be reversed by acute CaMKII inhibition but not PKA inhibition [3]. So inhibition of this Epac-dependent pathway to CaMKII-RyR2 may allow inhibition of β₁-AR-induced SR Ca²⁺ leak, while preserving the positive inotropic and lusitropic effects of PKA-dependent phosphorylation [43]. In principle, any of the identified components from Epac2 to RyR2 itself could be targets to suppress pathological SR Ca²⁺ leak. Critically, because this pathway is distinct from the PKA branch leading to increased inotropy and lusitropy, this might be done without reducing contractility, diastolic function or stroke volume that may limit β-AR blocker efficacy.

Epac activation has also been reported to increase myofilament Ca²⁺ sensitivity via CaMKII-dependent phosphorylation of myofilament proteins, resulting in stronger contraction for a given Ca²⁺ transient [18, 44]. To the extent that this Epac myofilament effect is beneficial in HF, it would need assessment when targeting the Epac induced SR Ca²⁺ leak therapeutically.

Highlights.

• Beta-adrenergic receptors (B-AR) induce arrhythmogenic SR Ca leak in adult rabbit and mouse ventricular myocytes.

• Previously proposed parallel pathways involving Epac or NOS signals are actually in series.

• This SR Ca leak is via a B1AR-cAMP-Epac2-PI3K-Akt-NOS1-CaMKII -RyR2 phosphorylation cascade.

• This parallels B-AR/cAMP inotropic and lusitropic PKA effects on Ca current and SR Ca uptake.