Inositol-1,4,5-trisphosphate induced Ca²⁺ release and excitation–contraction coupling in atrial myocytes from normal and failing hearts

Abstract

We studied excitation–contraction coupling (ECC) and inositol-1,4,5-triphosphate (IP₃)-dependent Ca²⁺ release in normal and heart failure (HF) rabbit atrial cells. Left ventricular HF was induced by combined volume and pressure overload. In HF atrial myocytes diastolic [Ca²⁺]_i was increased, action potential (AP)-induced Ca²⁺ transients (CaTs) were larger in amplitude, primarily due to enhanced Ca²⁺ release from central non-junctional sarcoplasmic reticulum (SR) and centripetal propagation of activation was accelerated, whereas HF ventricular CaTs were depressed. The larger CaTs were due to enhanced IP₃ receptor-induced Ca²⁺ release (IICR) and reduced mitochondrial Ca²⁺ buffering, consistent with a reduced mitochondrial density and Ca²⁺ uptake capacity in HF. Elementary IP₃ receptor-mediated Ca²⁺ release events (Ca²⁺ puffs) were more frequent in HF atrial myoctes and were detected more often in central regions of the non-junctional SR compared to normal cells. HF cells had an overall higher frequency of spontaneous Ca²⁺ waves and a larger fraction of waves (termed arrhythmogenic Ca²⁺ waves) triggered APs and global CaTs. The higher propensity of arrhythmogenic Ca²⁺ waves resulted from the combined action of enhanced IICR and increased activity of sarcolemmal Na⁺–Ca²⁺ exchange depolarizing the cell membrane. In conclusion, the data support the hypothesis that in atrial myocytes from hearts with left ventricular failure, enhanced CaTs during ECC exert positive inotropic effects on atrial contractility which facilitates ventricular filling and contributes to maintaining cardiac output. However, HF atrial cells were also more susceptible to developing arrhythmogenic Ca²⁺ waves which might form the substrate for atrial rhythm disorders frequently encountered in HF.

Key points

• Impaired calcium (Ca²⁺) signalling is the main contributor to depressed ventricular contractile function and occurrence of arrhythmia in heart failure (HF).

• Here we report that in atrial cells of a rabbit HF model, Ca²⁺ signalling is enhanced and we identified the underlying cellular mechanisms. Enhanced Ca²⁺ transients (CaTs) are due to upregulation of inositol-1,4,5-trisphosphate receptor induced Ca²⁺ release (IICR) and decreased mitochondrial Ca²⁺ sequestration.

• Enhanced IICR, however, together with an increased activity of the sodium–calcium exchange mechanism, also facilitates spontaneous Ca²⁺ release in form of arrhythmogenic Ca²⁺ waves and spontaneous action potentials, thus enhancing the arrhythmogenic potential of atrial cells.

• Our data show that enhanced Ca²⁺ signalling in HF provides atrial cells with a mechanism to improve ventricular filling and to maintain cardiac output, but also increases the susceptibility to develop atrial arrhythmias facilitated by spontaneous Ca²⁺ release.

Introduction

Excitation–contraction coupling (ECC) refers to the process that links action potential-dependent membrane depolarization, elevation of cytosolic Ca²⁺ levels ([Ca²⁺]_i) and force development in cardiac myocytes. Disturbances of ECC play a crucial role in cardiac disease resulting in contractile dysfunction and arrhythmogenesis. Atrial contraction is an important determinant of ventricular filling (‘atrial kick’) and it can contribute up to 40% of the end-diastolic ventricular filling during periods of increased haemodynamic demands, e.g. during exercise (Rahimtoola et al. 1975; Nicod et al. 1986). In patients with systolic heart failure (HF), the atrial contribution to the diastolic filling of an impaired ventricle becomes increasingly important to maintain cardiac output (Bonow et al. 1983; Raymond et al. 1998). Furthermore, such patients are especially prone to develop rhythm disorders like atrial fibrillation, the most common form of cardiac arrhythmia, which not only increase the risk of stroke, but also eliminate the atrial kick. The consequence of diminished atrial contractility is a reduced overall blood pumping capacity of the heart by up to a third, which in turn worsens symptoms of HF and facilitates its progression (Alpert et al. 1988). At the single cell level atrial Ca²⁺ signalling during ECC shows properties distinctly different from ventricular cells. These differences can largely be attributed to structural and functional differences in the organization of the couplon (Bootman et al. 2006) and the Ca²⁺ release units (CRUs) (Chen-Izu et al. 2006; Li et al. 2012). The couplon is the basic unit of ECC and refers to the ultrastructural arrangement of the sarcoplasmic reticulum (SR) membrane hosting ryanodine receptor (RyR) Ca²⁺ release channels and the sarcolemmal membrane containing L-type Ca²⁺ channels (LTCCs), only separated by a narrow gap of a few nanometres. The fact that the LTCCs directly face clusters of RyRs in CRUs assures activation of RyRs despite their rather low cytosolic Ca²⁺ sensitivity (Cannell & Kong, 2012). These SR regions are referred to as junctional SR (j-SR) where Ca²⁺ entry through LTCCs allows for an efficient activation of a cluster of RyRs forming a CRU by a process termed Ca²⁺ induced Ca²⁺ release (CICR; Cannell et al. 1987; Sheehan & Blatter, 2003). In ventricular cells, invaginations of the sarcolemmal membrane organized in a sarcomeric pattern and known as t-tubules create couplons throughout the entire volume of the cell and allow for rapid synchronous release of Ca²⁺ from CRUs. Atrial cells of many species, however, are either devoid of or show very sparse t-tubules, resulting in a rather inhomogenous distribution of the structures involved in Ca²⁺ handling and consequently the spatio-temporal organization of the cytosolic Ca²⁺ signal during ECC. In the subsarcolemmal space, the sarcolemmal membrane is located in close vicinity to the SR (j-SR). Here LTCCs directly face CRUs, resembling the ultrastructural arrangement found in couplons of ventricular cells (Stern et al. 1999) and are referred to as peripheral couplings (Kockskamper et al. 2001). However in more central regions of the cytosol where t-tubules are absent, abundant RyRs of the non-junctional SR (nj-SR) are found where CRUs are not associated with LTCCs and activation of RyRs relies on centripetal Ca²⁺ diffusion and propagation (Coombes & Timofeeva, 2003; Thul et al. 2012; Trafford et al. 2013). As a consequence, during ECC striking differences in the shape, magnitude and kinetics of subcellular Ca²⁺ transients (CaTs) are observed: upon action potential-dependent membrane depolarization fast and high-amplitude CaTs in the subsarcolemmal space are observed first that result from CICR from j-SR elicited by Ca²⁺ currents through LTCCs. Centripetal Ca²⁺ diffusion activates CICR from nj-SR CRUs resulting in propagation of CICR towards the cell centre in a Ca²⁺ wave-like fashion (Huser et al. 1996; Kockskamper et al. 2001; Sheehan & Blatter, 2003). Consequently, central CaTs have a slower rise time, a delayed peak and are smaller in amplitude.

The scarcity of t-tubules in atrial cells also affects beat-to-beat Ca²⁺ removal that occurs through reuptake into the SR by SERCA and extrusion across the sarcolemma predominantly via Na⁺–Ca²⁺ exchange (NCX). In atrial myocytes Ca²⁺ released from nj-SR CRUs occurs at greater distances from NCX and therefore the contribution of NCX to beat-to-beat Ca²⁺ removal is likely to differ from ventricular cells. Furthermore, increased NCX activity is a hallmark of myocytes from failing hearts and the electrogenic nature of its activity as cause for arrhythmogenic disturbances in ventricular tissue is well established (Pogwizd et al. 1999); however, the role of NCX in atrial arrythmogenesis still needs to be determined.

Atrial myocytes are also unique compared to ventricular cells in their significantly higher expression levels of a second type of SR Ca²⁺ release channel, the inositol-1,4,5-trisphosphate receptor (IP₃R; Lipp et al. 2000). IP₃Rs are part of a G_q-protein coupled receptor (GPCR) signalling pathway and release Ca²⁺ from the SR upon binding of IP₃ in a [Ca²⁺]_i-dependent fashion. IP₃R-induced Ca²⁺ release (IICR) has been hypothesized to lower the threshold for the activation of RyRs by CICR and may directly contribute to Ca²⁺ transients (Kockskamper et al. 2008). In cardiac disease such as hypertrophy, HF (Wang et al. 2003) and atrial fibrillation (Yamda et al. 2001) IP₃R expression levels are further increased. Furthermore, patients with cardiac hypertrophy or HF have elevated levels of Angiotensin II (Ang II; Nattel, 2002), a hormone that binds to GPCRs and exerts its effect, at least in part, by enhancing IICR. Taken together, evidence is growing that IP₃R-mediated Ca²⁺ signalling plays a critical role in cardiac disease, but the specific ramifications for atrial function are poorly understood.

Mitochondria contribute to Ca²⁺ cycling during ECC at different levels, from providing the fuel (ATP) for ion pumps (including SERCA) and substrate for phosphorylation of Ca²⁺ handling proteins, to control of the redox environment and to Ca²⁺ buffering (Dedkova & Blatter, 2012, 2013). While it is generally agreed that mitochondria are capable of sequestering Ca²⁺, even in large quantities, it has remained controversial whether significant amounts of Ca²⁺ buffering can occur during ECC on a beat-to-beat basis and thus whether mitochondria can shape the CaT in space and time (Bootman et al. 2006; O'Rourke & Blatter, 2009; Dedkova & Blatter, 2013). Again, the scarcity of t-tubules and the spatially inhomogeneous patterns of Ca²⁺ release in atrial cells, raise the question of a specific role of mitochondria for the propagation of activation, and specifically, whether the high density of mitochondria (Barth et al. 1992; Mackenzie et al. 2004; Hohendanner et al. 2013) can impede the propagation of Ca²⁺ release through the network of the nj-SR, and whether this changes in HF.

The goal of this study was a comprehensive characterization of atrial Ca²⁺ cycling during ECC in single myocytes from normal and from failing hearts. We focused on the interplay between IICR, mitochondrial Ca²⁺ buffering and altered NCX activity and its consequences for atrial arrhythmogenesis in HF. We found that in HF enhanced IICR and reduced mitochondrial Ca²⁺ uptake had positive inotropic effects mediated by enhanced CaTs during ECC, but also led to a higher propensity of spontaneous proarrhythmic Ca²⁺ release mediated by enhanced NCX activity.

Methods

Ethical approval

All protocols were approved by the Institutional Animal Care and Use Committee of Rush University Chicago, and comply with US and UK regulations on animal experimentation (Drummond, 2009).

Myocyte isolation and HF induction

Left atrial and left ventricular myocytes were isolated from male New Zealand White rabbits (Myrtle's Rabbitry, Thompsons Station, TN, USA). Rabbits were killed by intravenous injection of pentobarbital sodium (100 mg kg⁻¹). Hearts were excised, mounted on a Langendorff apparatus, and retrogradely perfused with nominally Ca²⁺-free Tyrode solution for 5 min followed by minimal essential medium Eagle (MEM) solution containing 20 μm Ca²⁺ and 22.5 μg ml⁻¹ Liberase Blendzyme TH (Roche Applied Science, Indianapolis, IN, USA) for 20 min at 37°C. The left atrium was removed from the heart and digested for an additional 5 min in the enzyme solution at 37°C. Ventricular or atrial tissue was minced, filtered, and washed in MEM solution containing 50 μm Ca²⁺ and 10 mg ml⁻¹ BSA. Isolated cells (n = 69 normal hearts) were kept in MEM solution with 50 μm Ca²⁺ at room temperature (20–24°C) until indicator dye loading and subsequent experimentation or cultured and used for experimentation after 24–48 h.

In addition, left atrial and ventricular myocytes were isolated from male rabbits with non-ischaemic HF (n = 17 hearts) induced by combined volume and pressure overload. Rabbits underwent two survival surgeries. Volume overload was induced by creation of aortic valve insufficiency, followed 2–4 weeks later by induction of pressure overload by abdominal aorta banding. Surgeries were conducted under isoflurane inhalation anaesthesia after ketamine–xylazine–acepromazine premedication. HF developed gradually (over 5–8 months, monitored by serial echocardiography), and was characterized by depressed systolic ventricular function, and arrhythmogenesis in an animal that has human-like Ca²⁺ handling and elecrophysiological properties (Pogwizd, 1995). Atrial cells were isolated from hearts that showed marked left ventricular dilatation (54 ± 5% increase in left ventricular diastolic internal dimension) and systolic dysfunction (decrease of left ventricular fractional shortening by 37 ± 4%). Ventricular cells from this model have been studied extensively and are well characterized at structural, biochemical, molecular, Ca²⁺ handling, and electrophysiological levels (Pogwizd, 1995; Pogwizd et al. 2001; Bers et al. 2002; Despa et al. 2002; Shannon et al. 2003; Domeier et al. 2009; Zima et al. 2010).

Cell culture and adenoviral gene transfer

For adenoviral transfections myocytes were kept in short-term culture (≤48 h) using Corning cellgro Medium 199 (VWR, Radnor, PA, USA) supplemented with penicillin and streptomycin (50 μg ml⁻¹). The medium was changed daily. Adenoviral gene transfer (multiplicity of infection of 500) was used to express FIRE1-cyt (fluorescent IP₃ responsive element 1 targeted to the cytosol; Remus et al. 2006), an IP₃ affinity trap (Domeier et al. 2008), mouse 43 kDa inositol polyphosphate 5′-phosphatase (m43 phosphatase; Kapoor et al. 2014), GFP or Mitycam (Kettlewell et al. 2009). Experiments were performed after 24 h culture for FIRE1-cyt, IP₃ affinity trap, m43 phosphatase and GFP, and after 48 h culture for Mitycam.

Solutions and chemicals

All chemicals were obtained from Sigma-Aldrich (St Louis, MO, USA) unless noted otherwise. SEA-0400 was obtained from ChemScene (Monmouth Junction, NJ, USA). Fluorescent calcium indicator dyes and MitoTracker Red FM were obtained from Molecular Probes/Life Technologies (Grand Island, NY, USA). Tyrode solution contained (in mm): 130 NaCl, 4 KCl, 2 or 7 CaCl₂ (as indicated), 1 MgCl₂, 10 d-glucose, 10 Hepes; pH 7.4 (adjusted with NaOH). For experiments on permeabilized myocytes saponin (0.01%) was used for cell membrane permeabilization and cells were kept in an ‘internal solution’ containing (in mm): 100 potassium aspartate, 15 KCl, 5 KH₂PO₄, 5 MgATP, 0.35 EGTA, 0.067 CaCl₂, 0.75 MgCl₂, 10 phosphocreatine, and 10 Hepes with 5 U ml⁻¹ creatine phosphokinase and 4% dextran (molecular weight: 40000); pH 7.2 (adjusted with KOH). Free [Ca²⁺]_i and [Mg²⁺]_i of this solution were 150 nm and 1 mm, respectively (calculated using Maxchelator, Stanford University). All cells were plated on laminin-coated glass coverslips and superfused with 2 mm Ca²⁺ Tyrode solution. To prevent movement of cells during confocal imaging blebbistatin (10 μm; Tocris Bioscience, Bristol, UK) was added to all Tyrode based solutions. ‘Mitochondrial solution’ contained (in mm): 135 KCl, 5 NaCl, 20 Hepes, 5 pyruvate, 2 glutamic acid, 2 malic acid, 0.5 KH₂PO₄, 1 MgCl₂, 5 EGTA, 15 BDM. Free [Ca²⁺] of this solution was set to 0 or 5 μm by addition of CaCl₂ (as determined with Maxchelator).

[Ca²⁺]_i and [Ca²⁺]_SR measurements

Indo-1 AM (5 μm, loading for 30 min at room temperature, followed by 15 min wash) was used to ratiometrically (excitation at 360 nm, emission collected at 405 (F₄₀₅) and 485 (F₄₈₅) nm) measure averaged whole-cell [Ca²⁺]_i. Background signals were subtracted and [Ca²⁺]_i changes are expressed as the ratio R = F₄₀₅/F₄₈₅. Fluo-4 and Fluo-5N were used for confocal cytosolic and SR [Ca²⁺] measurements. Changes in [Ca²⁺] measured with single wavelength dyes are expressed as F/F₀ where F₀ refers to resting fluorescence in unstimulated cells measured at the begin of an experiment or diastolic fluorescence levels measured under control steady-state conditions during electrical stimulation, respectively. CaT amplitudes were quantified as ΔF/F₀ and ΔR, respectively, where ΔF = F − F₀ and ΔR = R_peak − R_diast. For confocal [Ca²⁺]_i measurements cells were loaded with Fluo-4 AM (5 μm, excitation at 488 nm, emission at 515 nm) for 20 min. For [Ca²⁺]_SR measurements cells were loaded with Fluo-5N AM (10 μm, excitation at 488 nm, emission at 515 nm) for 1 h 45 min. All dye loading procedures were followed by a washing period (>10 min) in Tyrode solution. Confocal line scan images were recorded at 512 lines s^–1 using a ×60 oil-immersion objective lens (NA = 1.49) with a Nikon A1R system (Nikon Corporation, Melville, NY, USA). The scan line was placed along the longitudinal or transversal axis (as indicated) of the cell using pixel dimensions of 0.03–0.24 μm. Subcellular CaTs reflecting Ca²⁺ release from j-SR (SS, subsarcolemmal) and nj-SR (CT, central) were derived from 1 μm wide regions. The centripetal propagation velocity of CICR in atrial myocytes was derived from transverse confocal line scan images and calculated as the distance between two subcellular locations of interest and the interval between the times when the local CaTs reached half-maximal amplitude at these respective locations. The empirically measured point-spread function of the system had full width at half-maximum (FWHM) values of 0.28, 0.29 and 0.7 μm in the x, y and z (axial) dimensions (Maxwell & Blatter, 2012). Action potentials and global CaTs were elicited by electrical field stimulation (0.5 Hz) of intact atrial myocytes using a pair of platinum electrodes (voltage set at 50% above the threshold for contraction). Caffeine (10 mm) was used to deplete SR Ca²⁺ stores and measure the SR Ca²⁺ content. Experiments were conducted at room temperature (20–24°C).

IP₃ and IICR

A cytosol-targeted FRET-based IP₃ biosensor (FIRE1-cyt; for a detailed description see Remus et al. 2006) was used to dynamically study [IP₃]_i in atrial cells. Briefly, FIRE1-cyt reports fluorescence resonance energy transfer (FRET) between cyan and yellow fluorescent protein (CFP and YFP) upon binding of IP₃. FIRE1-cyt is a fusion protein of the IP₃ binding domain of the IP₃ receptor type 1 with CFP and YFP at the amino and carboxyl termini, respectively. Upon excitation of CFP at 457 nm, an increase in [IP₃] leads to increased FRET and a detectable change in the YFP/CFP fluorescence ratio (emissions were measured at >530 and 488 nm).

Inhibition of IICR was achieved by (i) adenoviral expression of FIRE1-cyt lacking CFP and YFP sequences which acts as a buffer of cytosolic IP₃ (IP₃ affinity trap; Domeier et al. 2008); (ii) adenoviral transfer of mouse inositol polyphosphate 5′ phosphatase (m43; Kapoor et al. 2014) which rapidly and efficiently degrades the biologically active IP₃ by removing the 5′ phosphate (Majerus et al. 1999; Vasilevski et al. 2008); and (iii) application (5 min) of the IP₃R blocker 2-aminoethoxydiphenyl borate (2-APB, 10 μm).

Elevation of [IP₃]_i in intact myocytes was achieved by agonist stimulation with angiotensin II (Ang II; 0.25 μm in freshly isolated myocytes; 10 μm in cultured cells) or by photolysis of caged IP₃ (cag-IP₃ PM; SiChem, Bremen, Germany). Cells were loaded with 100 nm cag-IP₃ for 20 min, and liberation of biologically active IP₃ was achieved by 405 nm laser illumination in user-defined cellular regions using the Mosaic illumination tool integrated in the Nikon A1R confocal imaging system (Shkryl et al. 2011).

Elementary IP₃R-mediated Ca²⁺ release events (Ca²⁺ puffs) were recorded in permeabilized atrial cells after stimulation with IP₃ (d-myo-inositol 1,4,5-trisphosphate, hexapotassium salt; Tocris Bioscience, Bristol, UK) and inhibition of RyR-mediated Ca²⁺ release with tetracaine (4 mm; ‘Ca²⁺ puff protocol’; Zima & Blatter, 2004).

Mitochondria

Cultured (48 h) left atrial myocytes were adenovirally transfected with the mitochondria-targeted fluorescent Ca²⁺ probe Mitycam (Kettlewell et al. 2009) to measure mitochondrial distribution and density as well as mitochondrial Ca²⁺ uptake. Mitycam is an inverse pericam that is excited at 488 nm and emission is measured at 530 ± 15 nm. Changes of [Ca²⁺]_mito are expressed as 1 – (F/F₀). For structural studies (density and distribution of mitochondria) with Mitycam or MitoTracker Red FM (25 μm, 30 min loading at 37°C, excitation at 543 nm, emission at >600 nm) atrial cardiomycoytes were kept in Tyrode solution and two-dimensional (2D) confocal images were recorded from a central plane. Mitochondrial Ca²⁺ uptake via mitochondrial Ca²⁺ uniporter (MCU) was studied in permeabilized cells expressing Mitycam. For the inhibition of the MCU, the ruthenium compound Ru360 (10 μm, EMD Chemicals, Philadelphia, PA, USA) was used.

Data analysis and statistics

The Liscana algorithm (Hohendanner et al. 2013) was used to analyse mitochondrial density and cell area.

The frequency of Ca²⁺ puffs and the frequency, amplitude, spatial width (FWHM), full duration (full duration at half-maximum, FDHM) and time to peak (TTP) of Ca²⁺ sparks were analysed with SparkMaster (Picht et al. 2007). Because of the small amplitude of the signal, Ca²⁺ puff amplitude, FWHM, FDHM and TTP were further verified by manual analysis.

The relative contribution of NCX to Ca²⁺ removal during CaTs (referred to as ‘NCX activity’) was calculated as:

1

where the time constants τ_CaT and τ_Caffeine were obtained by mono-exponential fits to the decay of electrically and caffeine-evoked CaTs in the same cell. The assay is based on the notion that the decline of an AP-induced CaT results from Ca²⁺ extrusion via NCX and reuptake into the SR via SERCA pump, whereas during caffeine (10 mm) exposure [Ca²⁺]_i declines solely by NCX extrusion. By considering the time constants τ_CaT and τ_Caffeine (cf. Fig. 7Ea) according to eqn 1 a measure of NCX activity can be derived that estimates the percentage contribution of NCX to the decline of an electrically evoked CaT.

Figure 7.

Arrhythmogenic Ca²⁺ waves and NCX activity in normal and HF atrial myocytes

A, protocol to monitor spontaneous Ca²⁺ wave activity. Cells were electrically paced (arrowheads) at 0.5 Hz in 7 mm extracellular [Ca²⁺] to enhance SR Ca²⁺ loading. Ca²⁺ waves were quantified during a 2-min rest period after cessation of pacing. The stars mark spontaneous Ca²⁺ waves without global cell-wide Ca²⁺ release and arrows mark arrhythmogenic Ca²⁺ waves. B, spontaneous Ca²⁺ wave in a single atrial myocyte observed during a rest period after pacing. The Ca²⁺ wave led to subsequent global Ca²⁺ release by triggering an AP and SR Ca²⁺ release typical for atrial ECC (‘arrhythmogenic Ca²⁺ wave’). C, confocal 2D image from a central focal plane revealing the ‘U-shaped’ wave front (left) and normalized fluorescence profiles (F/F₀) recorded along the dashed line (right). Da, frequency of spontaneous Ca²⁺ waves in normal and HF cells under control (CTRL) conditions (black), after the application of 2-APB (white) and SEA (grey). *P < 0.05 versus normal control; **P < 0.05 versus HF control, Student's t test. Db, latency of spontaneous Ca²⁺ waves. *P < 0.05, Student's t test. Dc, fraction of Ca²⁺ waves that induce a spontaneous AP and global CaTs (‘arrhythmogenic Ca²⁺ waves’) in HF atrial cells before (black) and after application of SEA (grey). In panel Dc square brackets indicate significant differences at P < 0.05, one-way ANOVA. Dd, absolute frequency of arrhythmogenic Ca²⁺ waves in normal and HF atrial myocytes in the absence and presence of SEA. To calculate absolute wave frequencies (panels a and d), also cells showing no waves were included. **P < 0.05 versus normal cells in the absence of SEA, Student's t test. *P < 0.05 versus control, Student's t test. Ea, protocol used to determine τ_CaT and τ_Caffeine for quantification of NCX activity. Eb, NCX activity in normal atrial cells (left), after inhibition of NCX in normal atrial cells with SEA, and in HF atrial cells. *P < 0.05 against normal cells, Student's t test.

Statistical comparisons were made using Student's t test or one-way ANOVA for paired or unpaired data, with statistical significance set at P < 0.05. Data are presented as individual observations or as means ± SEM of n measurements, where n is the number of cells.

Results

Ca²⁺ transients and SR Ca²⁺ depletion during ECC in normal and HF myocytes

We compared cytosolic CaTs between left atrial myocytes from normal rabbits and rabbits with systolic heart failure (Maxwell et al. 2012). Cytosolic CaTs were elicited by electrical field stimulation and recorded with Fluo-4 by line scan confocal imaging (Fig.1Aa and d). As expected for a cardiac cell type lacking transverse tubules (Huser et al. 1996; Kockskamper et al. 2001) [Ca²⁺]_i rose first and to the highest level in the cell periphery (SS in Fig. 1Aa, corresponding to Ca²⁺ release from j-SR), followed by propagation of Ca²⁺ release in centripetal direction where [Ca²⁺]_i peaked with a delay at significantly (Fig. 1Ab and C) lower levels (CT in Fig. 1Aa, corresponding to Ca²⁺ release from nj-SR). In atrial cells from failing hearts, CaTs were prolonged and significantly higher in amplitude in central regions of the cell (Fig. 1Ae and C). This is in stark contrast to ventricular myocytes where CaTs were significantly depressed in HF (Fig. 1Af and E) compared to normal myocytes (Fig. 1Ac and E). Furthermore, the centripetal propagation velocity of Ca²⁺ release was significantly accelerated in atrial HF myocytes (Fig. 1D; HF: 205 ± 19 μm s⁻¹; normal: 147 ± 13 μm s⁻¹). Consistent with the larger amplitude of cytosolic CaTs in HF atrial myocytes, the beat-to-beat SR Ca²⁺ depletion transients (Fig. 1B) were significantly larger in HF atrial myocytes (Fig. 1F).

Figure 1.

Ca²⁺ cycling during ECC in normal and HF left atrial myocytes

A, control line scan images (a, d) and CaTs (b, e) from central (CT) and subsarcolemmal (SS) cytosolic regions of normal and HF atrial myocytes and from control (c) and HF (f) ventricular cells. Arrowheads indicate electrical stimulation. B, [Ca²⁺]_SR traces obtained from whole-cell imaging of Fluo-5 N fluorescence in normal (left) and HF (right) atrial myocytes. Arrowheads indicate electrical stimulation. C, summary data of local CaT amplitudes in CT and SS regions of normal and HF atrial myocytes obtained from transversal line scans. Square brackets indicate significant differences at P < 0.05, one-way ANOVA. D, summary data for the centripetal Ca²⁺ propagation velocity in normal and HF atrial cells derived from transversal line scans. E, summary data for the CaT amplitude in normal and HF ventricular myocytes obtained from longitudinal line scans. F, summary data for the [Ca²⁺]_SR depletion amplitude in normal and HF atrial myocytes as obtained from 2D confocal measurements of Fluo-5 N fluorescence. Panels D–F: *P < 0.05, Student's t test. Here and subsequent figures: numbers in bar graphs indicate n = number of cells.

IICR induced by angiotensin II

Ang II is known to increase [IP₃]_i and to exert a positive inotropic effect in cardiac mycoytes (Meissner et al. 1998; Remus et al. 2006). Ang II levels as well as Ang II receptor expression on cell surfaces are increased in vivo in different models of cardiac disease (e.g. atrial fibrillation; Nattel, 2002); however, Ang II has also been shown to depress left ventricular contractile performance in HF (Cheng et al. 1996). We compared the effects of Ang II stimulation on diastolic [Ca²⁺]_i, CaTs and SR Ca²⁺ load in normal and HF atrial myocytes. CaTs were elicited by field stimulation and in each experiment SR Ca²⁺ load was estimated from the amplitude of the [Ca²⁺]_i signal elicited by rapid application of caffeine (10 mm; Fig. 2A). We confirmed the Ang II-dependent intracellular IP₃ production using the FRET-based IP₃ biosensor FIRE1-cyt (Remus et al. 2006). As shown in Fig. 2B, Ang II stimulation increased [IP₃]_i significantly in normal and HF atrial myocytes, but there was no difference in magnitude of this increase between normal and HF cells, although the sensitivity and dynamic range of the IP₃ biosensor may not be sufficient to detect these potentially small changes (Fig. 2F). In atrial myocytes from failing hearts, diastolic [Ca²⁺]_i levels, measured ratiometrically with Indo-1, were significantly increased compared to normal cells (Fig. 2C), and treatment with Ang II increased diastolic [Ca²⁺]_i in normal and HF myocytes even further. The whole-cell CaT amplitudes measured with Indo-1, were significantly increased in HF cells (+32%; Fig. 2A and D), thus confirming the results obtained with Fluo-4 (Fig. 1C). However, Ang II significantly decreased the CaT amplitude in HF cells by 36% (Fig. 2D, P < 0.05). Additionally, the SR Ca²⁺ content was unchanged in HF atrial cells compared to normal cells. However, exposure of cells to Ang II for >5 min decreased the SR Ca²⁺ load significantly in HF, whereas it had no effect in normal atrial cells (Fig. 2E).

Figure 2.

Effect of Ang II on Ca²⁺ signalling during ECC in atrial cells

A, electrically and caffeine-evoked CaTs in normal (top) and HF (bottom) atrial cells before (a, c; CTRL) and after (b, d) application of Ang II. B, changes of [IP₃]_i in normal (a) and HF (b) atrial myocytes revealed by measurement of FRET between CFP and YFP in cells expressing the IP₃ biosensor FIRE1-cyt. C, diastolic [Ca²⁺]_i during field stimulation in normal and HF myocytes under control conditions (black) or after treatment with Ang II (grey). D, CaT amplitudes during field stimulation in normal and HF myocytes under control conditions (black) or after treatment with Ang II (grey). E, SR Ca²⁺ load assessed from changes of [Ca²⁺]_i induced by rapid application of 10 mm caffeine, in normal and HF myocytes under control conditions (black) or after treatment with Ang II (grey). In panels C–E square brackets indicate significant differences at P < 0.05, one-way ANOVA. F, grey bars show average increase of FIRE1-cyt FRET signal (F₅₃₀/F₄₈₈) after application of AngII in normal and HF myocytes. Black bars represent changes of control FRET signal over the same time interval in the absence of AngII. *P < 0.05, Student's t test.

IICR induced by cytosolic uncaging of IP₃

To further elucidate the role of IP₃R-mediated Ca²⁺ release during ECC in normal and HF atrial myocytes, we loaded atrial myocytes with the caged IP₃ compound cag-IP₃. Upon UV-light exposure, cag-IP₃ instantaneously released biologically active IP₃ which led to an increase of [IP₃]_i levels as confirmed by FIRE1-cyt measurements (Fig. 3A and B). Global IP₃ uncaging significantly increased the CaT amplitude in regions of the nj-SR (CT in Fig. 3C) in normal atrial cells with little effect on SS CaTs. In HF atrial myocytes, however, IP₃ uncaging led to a similar decrease of Ca release from j-SR and nj-SR (Fig. 3E). This decrease in CaT amplitude in HF myocytes was due to a combined effect of reduced peak CaT (by 17 ± 2 and 19 ± 3% in CT and SS regions) and an increase in diastolic [Ca²⁺]_i (by 57 ± 7 and 70 ± 8% in CT and SS regions). Averaged over the entire cells the increase in diastolic [Ca²⁺]_i 20 s after IP₃ uncaging was almost twice as large in HF as in normal cells ([Ca²⁺]_diast (F/F₀): 1.36 ± 0.06 in normal cells (n = 12); 1.67 ± 0.13 in HF cells (n = 6); P < 0.05). The IP₃R blocker 2-APB (10 μm) abolished the IP₃ effects (Fig. 3D), confirming that the observed effects on CaT amplitude were mediated by IICR, and actually decreased CaT amplitudes below control levels, suggesting that basal IP₃ activity made a contribution towards the magnitude of CaTs under control conditions.

Figure 3.

Effect of IP₃ uncaging on Ca²⁺ signalling during ECC in atrial cells

A, change of [IP₃]_i in an atrial myocyte expressing FIRE1-cyt before and after IP₃ uncaging. B, average percentage change of the [IP₃]_i-dependent FIRE1-cyt FRET signal (F₅₃₀/F₄₈₈) averaged over 3 s after uncaging. C and E, effect of global whole-cell IP₃ uncaging on local CaTs in the subsarcolemmal (SS) and central (CT) cytosol in normal (C) and HF (E) atrial myocytes. D, same experiment as in panel C but in the presence of the IP₃R blocker 2-APB. Panels C–E, bottom: summary data of IP₃ uncaging effects on CaT amplitudes. *P < 0.05, Student's t test. 100% corresponds to the average [Ca²⁺]_i levels before IP₃ uncaging.

Effect of basal [IP₃]_i on CaTs

We investigated the effect of basal IP₃ activity on CaTs further, using three different methodological approaches. First, we used the pharmacological IP₃R blocker 2-APB. In the presence of 2-APB the CaT amplitude was reduced in atrial myocytes; however, this basal IP₃ effect was significantly more pronounced in HF (Fig. 4E; cf. also Fig. 3D). In two additional series of experiments we adenovirally expressed the IP₃ binding domain of the IP₃R type 1 (IP₃ affinity trap) or the m43 phosphatase in the cytosol. Both strategies aimed to reduce [IP₃]_i levels. As a control we transfected cells with an adenovirus encoding GFP and determined in control measurements that CaT amplitudes were identical in the presence or absence of the virus (data not shown). Figure 4Aa and Ba shows CaT in virus infected normal and HF atrial cells. As in uninfected cells (Figs 1 and 2) the CaT amplitude was increased in HF atrial myocytes. In atrial cells expressing the IP₃ affinity trap (Fig. 4Ab and Bb) or m43 phosphatase (Fig. 4Ac and Bc) the CaT amplitude was reduced in normal cells (albeit statistically not significant) and to a significantly larger degree in HF atrial myocytes. For comparison, m43 phosphatase expression in normal ventricular myocytes had no effect on CaT amplitudes, suggesting a lower basal IP₃ activity in ventricular cells. The effect of inhibition of IP₃ signalling with three distinctly different methods yielded extremely consistent results (Fig. 4C–E) and validated the use of 2-APB as a valuable tool to inhibit IP₃Rs in cardiac cells. As shown in Fig. 4F the effect of basal [IP₃]_i levels on CaT amplitude affected predominantly release of Ca²⁺ from j-SR in normal cells, whereas in HF atrial myocytes the effect was overall more pronounced but affected j-SR and nj-SR to a similar degree.

Figure 4.

Effect of basal IICR on CaT amplitude

A and B, CaTs in field stimulated normal (A) and HF (B) atrial myocytes after 24 h in culture, expressing a control (CTRL) virus (a), IP₃ affinity trap (b) or m43 phosphatase (c). C and D, summary data of the percentage change of CaT amplitudes in normal and HF atrial myocytes expressing an IP₃ affinity trap (C) or m43 phosphatase (D). *P < 0.05 against normal cells, Student's t test. In D, comparison of m43 phosphatase effect in normal ventricular cells. E, summary data of the effect of 2-APB on CaT amplitudes in normal and HF cells. Panels C–E: *P < 0.05, Student's t test. F, same as panel E but spatially resolved for the subsarcolemmal (SS) and the central (CT) cytosol. Square brackets indicate significant differences at P < 0.05, one-way ANOVA. In panels C–F, 100% corresponds to the average CaT amplitudes before inhibition of IP₃ signalling.

Elementary IP₃R-dependent Ca²⁺ release events

Elementary IP₃R-dependent Ca²⁺ release events from clusters of IP₃Rs (termed ‘Ca²⁺ puffs’, cf. Tovey et al. 2001) have been observed in a variety of cell types including atrial myocytes. In cat atrial myocytes their amplitude has been reported to be 70–80% smaller, the duration three to four times longer and the rise time approximately twofold longer than RyR-mediated Ca²⁺ release events (Ca²⁺ sparks; Zima & Blatter, 2004). In order to quantify Ca²⁺ puffs from rabbit atrial myocytes, localized non-propagating elementary Ca²⁺ release events were recorded in permeabilized left atrial myocytes from normal (Fig. 5A) and HF hearts (Fig. 5B). To abolish RyR-mediated Ca²⁺ release from the SR, the RyR blocker tetracaine (4 mm) was added which resulted in an almost complete cessation of local Ca²⁺ release event activity (Fig. 5Ab and Bb). For further analysis cells were only included if they displayed no more than 1 event s⁻¹ and 100 μm of confocally scanned distance (s⁻¹ (100 μm)⁻¹) in the presence of tetracaine. Cells were subsequently exposed to tetracaine + IP₃ (5 μm) to record elementary Ca²⁺ release events that originated essentially exclusively from IP₃Rs (Fig. 5Ac and Bc), thus representing Ca²⁺ puffs. In comparison to RyR-mediated Ca²⁺ sparks (Fig. 5C), Ca²⁺ puffs recorded in normal (Fig. 5D) and HF atrial rabbit myocytes were >50% smaller in amplitude, their duration was more than twofold longer and the rise time of [Ca²⁺]_i was about 40% longer. The IP₃R origin of Ca²⁺ puffs was confirmed with the cessation of Ca²⁺ release after the addition of 2-APB (Fig. 5Ad and Bd). The spatio-temporal properties of Ca²⁺ puffs were strikingly similar between normal and HF atrial myocytes: amplitude (ΔF/F₀: 0.38 ± 0.04 in normal cells vs. 0.32 ± 0.03 in HF cells), FDHM (62 ± 7 ms normal vs. 60 ± 7 ms HF), FWHM (2.7 ± 0.1 μm normal vs. 2.7 ± 0.2 HF) and rise time (29 ± 2 ms normal vs. 28 ± 2 ms HF). The overall frequency of Ca²⁺ puffs was significantly increased in HF (Fig. 5E; Ca²⁺ puffs: 0.8 ± 0.3 s⁻¹ (100 μm)⁻¹ vs. 1.8 ± 0.3 s⁻¹ (100 μm)⁻¹). The subcellular distribution of Ca²⁺ puffs revealed interesting inhomogeneities (Fig. 5F). The highest Ca²⁺ puff frequencies were found within a distance of ≤2 μm from the sarcolemma with no obvious difference between normal and HF myocytes. In contrast, a robust increase in Ca²⁺ puff frequency in HF was found in the central cytosol (Fig. 5F; defined as regions with a minimal distance of 2 μm to the sarcolemmal or nuclear membrane) which increased from 0.4 ± 0.2 in normal cells to 1.9 ± 0.3 Ca²⁺ puffs s⁻¹ (100 μm)⁻¹. Interestingly, in normal cells Ca²⁺ puffs were more frequent in the perinuclear region (i.e. at distances <2 μm from the nuclear membrane) compared to the central cytosol; however, in HF myocytes central and perinuclear frequencies were similar. In summary, these results indicate that overall in HF IICR became more prevalent and shifted spatially towards more central regions of the cell, indicative of substantial structural and/or functional remodelling of IP₃-dependent Ca²⁺ signalling in HF.

Figure 5.

Ca²⁺ puff activity in normal and HF atrial myocytes

A and B, examples of longitudinal line scan images from permeabilized normal (A) and HF (B) myocytes under control (CTRL) conditions (a), after application of tetracaine (4 mm) (b), tetracaine + IP₃ (5 μm) (c), and tetracaine + IP₃ + 2-APB (10 μm) (d). nuc, nucleus; c, cytosol; e, extracellular space. C, line scan image of a Ca²⁺ spark (top) and Ca²⁺ spark profile (bottom) under control conditions in a normal atrial cell. D, line scan image of a Ca²⁺ puff (top) and Ca²⁺ puff profile (bottom) under control conditions in a normal atrial cell. E, overall frequencies of Ca²⁺ release events under the four conditions of panels A and B in normal (black) and HF (grey) cells. F, Ca²⁺ puff frequencies in the subsarcolemmal, central cytosolic and perinuclear compartments of normal (black) and HF (grey) atrial cells. Panels E and F: *P < 0.05, Student's t test.

Effects of mitochondrial Ca²⁺ uptake on Ca²⁺ cycling during ECC

Mitochondria can take up significant amounts of Ca²⁺ and have been shown to influence the spread of activation of Ca²⁺ release during ECC in rat atrial myocytes (Mackenzie et al. 2004). This was confirmed here for rabbit atrial myocytes. In cells exposed to the MCU blocker Ru360 CaT amplitudes were increased, but no significant effects on diastolic [Ca²⁺]_i and SR Ca²⁺ load were observed (Fig. 6A). These results suggest that mitochondrial Ca²⁺ uptake can contribute to [Ca²⁺]_i regulation on a beat-to-beat basis.

Figure 6.

Mitochondrial density and mitochondrial Ca²⁺ uptake

A, average effects of Ru360 on diastolic [Ca²⁺]_i, amplitude of AP-induced CaTs and SR Ca²⁺ load (CaT evoked by 10 mm caffeine). B, normal (top) and HF (bottom) atrial cell stained with Mitotracker Red FM. Raw fluorescence images and binary images used to calculate mitochondrial density. C, estimated cell volume occupied by mitochondria using Mitotracker (a) or Mitycam (b). D, effect of rapidly increasing [Ca²⁺]_i from 0 to 5 μm on mitochondrial Ca²⁺ uptake in permeabilized normal and HF atrial cells. E, average effect of increasing [Ca²⁺]_i to 5 μm on amplitude and the time to peak of [Ca²⁺]_mito. *P < 0.05, Student's t test.

Since changes in mitochondrial density in ventricular cells associate with cardiac disease – whereby both, increases and decreases have been reported (Coleman et al. 1987; Barth et al. 1992; Hohendanner et al. 2013) – we measured mitochondrial density in rabbit atrial myocytes. For this purpose we used two different mitochondria-targeted fluorescent probes and measured mitochondrial density in normal and HF atrial myocytes. From 2D confocal images of dye-loaded myocytes, binary images were constructed (Fig. 6B) that then allowed us to determine the fraction of a cellular cross-section occupied by mitochondria (for details and quantitative evaluation of the algorithm used see Hohendanner et al. 2013). Using the probe Mitotracker (Fig. 6Ca), HF atrial myocytes revealed a significantly reduced mitochondrial content (estimated cell volume occupied by mitochondria: 33 ± 0.5% in normal cells vs. 31 ± 0.7% in HF; P < 0.05). Because accumulation of Mitotracker is membrane potential dependent and disease-related potential differences between normal and HF cannot be excluded, we used the fluorescent Ca²⁺-sensitive mitochondria targeted protein Mitycam (Kettlewell et al. 2009) as a second approach to quantify mitochondrial density (Fig. 6Cb). Using Mitycam the estimated mitochondrial volume was 38 ± 0.8% in normal and 36 ± 0.7% in HF myocytes (P < 0.05). Although the values obtained with Mitycam tended to be higher, both probes reported independently a significantly decreased density of mitochondria in HF. The consequences of reduced mitochondrial density for mitochondrial Ca²⁺ uptake were measured directly using the Mitycam probe. As shown in Fig. 6D, exposing Mitycam-loaded mitochondria in permeabilized normal atrial cells to 5 μm extramitochondrial Ca²⁺ resulted in a fast (<1 s) increase of mitochondrial [Ca²⁺] ([Ca²⁺]_mito) that rapidly levelled off; 5 μm is approximately the [Ca²⁺] for half-maximal activation of the MCU (Sedova et al. 2006). In HF atrial myocytes the Ca²⁺ uptake response was blunted. As summarized in Fig. 6E, the [Ca²⁺]_mito levels reached in HF cells amounted to only about half of that observed in normal cells, and the uptake kinetics were slowed in HF myocytes. These data indicate that mitochondrial Ca²⁺ uptake occurs on a time scale similar to CaTs and therefore has the potential to buffer Ca²⁺ released during ECC. The reduced mitochondrial density and Ca²⁺ uptake in HF may therefore contribute to the robust increase of the CaT, particularly in the centre of atrial cells (Fig. 1).

Arrhythmogenic spontaneous Ca²⁺ release in HF atrial myocytes

Heart failure patients are significantly more prone to develop atrial fibrillation (Cohn et al. 1991; Johnstone et al. 1992; Tsang et al. 2002). We therefore tested the hypothesis that in HF atrial myocytes spontaneous pro-arrhythmic Ca²⁺ release events in the form of Ca²⁺ waves occur more frequently, and we explored the effects of spontaneous Ca²⁺ waves on the electrical stability of these cells. Spontaneous Ca²⁺ waves and spontaneous (not electrically triggered) APs were quantified during a 2-min period of rest after >1 min of electrical pacing (arrowheads in Fig. 7A) in elevated extracellular Ca²⁺ (7 mm) to facilitate Ca²⁺ loading of the SR (Domeier et al. 2012). Spontaneous Ca²⁺ waves were observed in both ventricular (after stimulation at 1 Hz) and atrial (after stimulation at 0.5 Hz) myocytes at frequencies of 0.017 ± 0.003 (n = 11) and 0.025 ± 0.005 waves s^–1, respectively. (No waves were observed in ventricular cells stimulated at 0.5 Hz). However, the propensity to produce spontaneous waves was significantly increased in HF atrial myocytes. Not only was the fraction of cells revealing waves increased (normal: 75% (30/40 cells); HF: 86% (18/21 cells)), but wave frequency also increased (Fig. 7Da: increase to 0.050 ± 0.010 waves s^–1 in HF) and the latency (the time interval between cessation of pacing and the onset of the first spontaneous Ca²⁺ wave) was reduced from 56 ± 6 s to 32 ± 7 s (Fig. 7Db). The spontaneous Ca²⁺ waves in atrial cells revealed a unique spatio-temporal organization. As illustrated in Fig. 7B, typically the waves propagated from the site of origin (t = 0) along the longitudinal axis of the cell; however, upon closer examination it became evident that in atrial cells the leading wave front propagated through the subsarcolemmal space (thus presumably via release of Ca²⁺ from the j-SR) giving rise to a ‘U’-shaped wave front (t = 368 ms), illustrated in more detail in Fig. 7C. Furthermore, 39 ± 7% of all Ca²⁺ waves observed in normal atrial cells and 55 ± 8% of all Ca²⁺ waves in HF atrial cells at some point during propagation induced global Ca²⁺ release, presumably as a consequence of inducing an AP (Fig. 7Dc). These events showed the typical [Ca²⁺]_i pattern of an atrial CaT with an initial ring of elevated [Ca²⁺]_i in the cell periphery (t = 385 ms) that subsequently propagated towards the cell centre (t = 403 ms) (Huser et al. 1996; Kockskamper et al. 2001; Sheehan et al. 2006). We will refer to this type of Ca²⁺ wave here as an ‘arrhythmogenic Ca²⁺ wave’ (arrows in Fig. 7A). The overall absolute frequency of arrhythmogenic Ca²⁺ waves was significantly higher in HF atrial myocytes (Fig. 7Dd; 0.030 ± 0.007 vs. 0.008 ± 0.002 in normal). In contrast to atrial cells, no arrhythmogenic Ca²⁺ waves were observed in ventricular myocytes.

In the next set of experiments we investigated the mechanisms underlying spontaneous and arrhythmogenic Ca²⁺ waves. As shown in Fig. 7Da in the presence of the IP₃R blocker 2-APB the frequency of spontaneous Ca²⁺ waves was reduced in atrial myocytes with a more pronounced effect in HF myocytes. In normal cells 2-APB reduced the wave frequency by nearly a third (statistically not significant), whereas in HF the reduction amounted to over half and became statistically significant. Since 2-APB has been reported to potentially also affect Ca²⁺ pumps (Peppiatt et al. 2003), we checked for this possibility by assessing the SR Ca²⁺ content with caffeine (10 mm) in normal cells and found no significant difference (amplitude of the caffeine-induced CaT (ΔR): 0.81 ± 0.03 (n = 19) control vs. 0.75 ± 0.07 (n = 10) in 2-APB). The data indicate that basal IP₃-dependent Ca²⁺ signalling facilitated spontaneous SR Ca²⁺ release (Ca²⁺ waves; see also Li et al. 2005; Domeier et al. 2008), a phenomenon that was more pronounced in HF atrial cells, consistent with enhanced IP₃ activity in HF atrial cells (Figs 2–5).

We further investigated the involvement of NCX in the generation of spontaneous Ca²⁺ waves. The electrogenic nature of NCX action has the potential to depolarize the membrane potential and to initiate Ca²⁺ release. Addition of the NCX inhibitor SEA-0400 (SEA) had no significant effect on the overall Ca²⁺ wave frequency in normal and HF atrial cells. However, NCX inhibition had a very pronounced effect when considering exclusively arrhythmogenic Ca²⁺ waves (Fig. 7Dc and Dd). SEA significantly reduced the fraction of cells revealing arrhythmogenic Ca²⁺ waves (by 62%) as well as their absolute frequency in normal atrial myocytes, and almost completely abrogated this wave type in HF atrial myocytes, suggesting an enhanced role of NCX for the occurrence of arrhythmogenic Ca²⁺ waves, especially in HF. We quantified NCX activity from the time constants of the decline of electrically (τ_CaT) and caffeine (τ_Caffeine)-evoked CaTs (Fig. 7Ea) as outlined in the Methods section (eqn 1). The experiments revealed two important results. First, as shown in Fig. 7Eb, SEA efficiently reduced NCX activity. Second, in HF atrial cells NCX activity was significantly increased (by 76 ± 30%; P < 0.05). The latter results are similar to the increase of NCX activity found in rabbit HF ventricular cells (Bassani et al. 1994; Pogwizd et al. 1999). Together these data indicate that HF atrial cells have a lower Ca²⁺ wave threshold, are more likely to show spontaneous Ca²⁺ waves and are more prone to develop Ca²⁺ waves with subsequent cell-wide depolarization due to an increased NCX activity compared to normal atrial myocytes.

Discussion

In this study, we investigated IP₃-dependent Ca²⁺ release and its role for ECC in normal atrial cells and atrial myocytes from a left ventricular heart failure model. The key findings of our investigation were as follows: first, in HF atrial myocytes, AP-dependent CaTs were enhanced in amplitude and duration, whereas ventricular CaTs were depressed. Primarily more robust Ca²⁺ release from nj-SR and faster propagation of activation were responsible for this positive inotropic response. Second, HF atrial myocytes revealed enhanced IICR that preferentially facilitated Ca²⁺ release from central nj-SR and was, at least in part, the cause of enhanced inotropy in HF atrial myocytes. Third, HF atrial myocytes revealed a reduced mitochondrial density that resulted in reduced Ca²⁺ sequestration, with the consequence of larger amplitude CaTs. Finally, HF atrial myocytes revealed a higher propensity of proarrhythmic Ca²⁺ waves that resulted from the combined effect of enhanced IICR and increased NCX activity.

Enhanced IICR in HF: consequences for atrial ECC

Heart failure is characterized by depressed ventricular contractility and arrhythmogenesis. At the cellular level the combination of increased RyR-mediated SR Ca²⁺ leak (Maxwell et al. 2012), lower SERCA activity and higher NCX activity (Pogwizd et al. 1999), resulting in diminished CaTs and contractility, are hallmarks of systolic HF. The observation of depressed CaTs in single ventricular myocytes presented here (Fig. 1Af) confirmed the impaired Ca²⁺ cycling in ventricular myocytes in our rabbit HF model, induced by volume and pressure overload. Patients with systolic heart failure increasingly rely on the atrial contribution to diastolic ventricular filling to maintain cardiac output, but are also more prone to develop atrial fibrillation (AF; Bonow et al. 1983; Raymond et al. 1998). The irregular rhythm of AF resulting in the loss of the atrial systole and the neurohumoral activation of the renin–angiotensin–aldosterone system due to depressed ventricular contractility are contributing factors to adverse haemodynamic changes (Rahimtoola et al. 1975; Linderer et al. 1983; Daoud et al. 1996; Wijffels et al. 1997). Furthermore, atrial (Cao et al. 2002; Zhao et al. 2007) and ventricular (Go et al. 1995; Ai et al. 2005) cells show an increase of cytosolic IP₃R expression under conditions of HF and AF. Although IP₃Rs are outnumbered by the much more abundant RyRs by one to two orders of magnitude, they can provide an inotropic reserve by increasing diastolic [Ca²⁺]_i in the vicinity of the Ca²⁺-sensitive RyR. This is particularly the case in atrial myocytes that have been shown to express up to six times as many IP₃Rs as ventricular cells (Lipp et al. 2000). IICR has been postulated to have positive inotropic effects by sensitizing RyRs and bringing them closer to threshold for activation by Ca²⁺ and to facilitate CICR (Domeier et al. 2008; Kockskamper et al. 2008; Harzheim et al. 2009). Furthermore, Ca²⁺ released through IP₃Rs may contribute directly to systolic CaTs (Foskett et al. 2007). A role for IICR directly contributing to systolic CaTs could be explained by the need of IP₃Rs for binding of both Ca²⁺ and IP₃ to release Ca²⁺ from the SR. For ‘sensitized’ IP₃Rs (with pre-bound IP₃) Ca²⁺ release is expected to be augmented during systolic CaTs, when Ca²⁺ fluxes through LTCC or RyRs are maximal. In our study, photorelease of caged IP₃ or neurohumoral stimulation with Ang II to activate the endogenous IP₃ production confirmed that increased IICR enhanced CaT amplitudes in field-stimulated normal rabbit atrial myocytes. The previously reported upregulation of IP₃Rs in HF and AF (Go et al. 1995; Yamda et al. 2001; Ai et al. 2005) is consistent with our functional results. In HF atrial cells the CaT amplitude was significantly increased, predominantly resulting from enhanced Ca²⁺ release from the central nj-SR. This is consistent with the observed overall increased frequency of elementary Ca²⁺ release events originating from clusters of IP₃Rs (Ca²⁺ puffs) in HF atrial myocytes that was particularly pronounced in areas of the nj-SR. The latter may suggest that the reported increased IP₃R expression in HF is preferentially located to the central cytosolic regions in atrial cells. Interestingly, the spatial and temporal properties of Ca²⁺ puffs were unchanged in HF, only the frequency was increased. This would suggest that IP₃R upregulation in HF may not occur through changes in IP₃R cluster size, but through formation of new clusters that are preferentially located in the nj-SR. This is consistent with the finding that pharmacological IP₃R inhibition (2-APB) predominantly affected j-SR Ca²⁺ release in normal cells whereas j-SR and nj-SR were equally affected in HF atrial cells (Fig. 4F).

The notion that IICR significantly contributed to SR Ca²⁺ release during ECC is underscored by the fact that IP₃R inhibition (2-APB), enhanced IP₃ buffering (IP₃ affinity trap) or IP₃ degradation (m43 phosphatase) all decreased the amplitude of CaTs in normal atrial cells and did so even more in HF atrial cells, providing additional experimental support for IICR becoming increasingly more important in HF. The fact that enhancing cytosolic IP₃ degradation by expressing m43 phosphatase had no effect on the CaTs in ventricular cells further underscored the relevance of IP₃ signalling for atrial ECC.

A common finding in patients with developing cardiac hypertrophy and HF is an increased level of Ang II, a hormone that exerts its effects in part by increasing IICR (Li et al. 2005; Bootman et al. 2007). In our experiments, in normal atrial cells Ang II enhanced Ca²⁺ signalling during ECC, similar to IP₃ uncaging. However, both Ang II and IP₃ uncaging had the opposite effect in HF atrial myocytes. Under steady-state pacing conditions, elevating [IP₃]_i levels decreased the CaT amplitude, an observation that was paralleled by a significant decrease in the SR Ca²⁺ content. Our results, together with the previous observation of enhanced SR Ca²⁺ leak in HF and the sensitivity of Ca²⁺ leak to stimulation of IICR (Zima et al. 2010), suggest, that in HF higher IP₃R activity and enhanced IICR ultimately result in Ca²⁺ depletion of the SR. The data suggest that in HF atrial myocytes basal IICR is activated to the extent that it boosts atrial contractility to optimize ventricular filling. However, further stimulation of IP₃ production via humoral pathways (Ang II) or by directly elevating [IP₃]_i (IP₃ uncaging) converts this compensatory positive inotropic effect of IICR into adverse consequences for contractility by depleting the SR of Ca²⁺ and therefore depressing atrial CaTs and subsequently ventricular filling.

Mitochondrial Ca²⁺ uptake

Mitochondria occupy approximately 30% of the cell volume of a cardiac myocyte and play a crucial role in energy metabolism, redox regulation and Ca²⁺ homeostasis under normal and pathophysiological conditions. Although the question of whether they contribute to Ca²⁺ signalling during ECC on a beat-to-beat basis (O'Rourke & Blatter, 2009) is still a matter of controversy, the organelle is equipped for efficient and rapid Ca²⁺ uptake and release (Wendt-Gallitelli & Isenberg, 1989; Miyata et al. 1991; Sedova et al. 2006) and is able to buffer significant amounts of Ca²⁺ (Solaro & Briggs, 1974; Dedkova & Blatter, 2012, 2013). Mitochondria further contribute (indirectly) to shaping CaTs by providing ATP to SERCA, influencing RyR activity via ATP-dependent regulation, and generally by setting the redox environment that in turn regulates the activity of Ca²⁺ handling proteins (Zima et al. 2013). Mitochondria have been suggested to attenuate the propagation of release through the nj-SR by sequestering cytosolic Ca²⁺ to the extent that mitochondria can act as a ‘firewall’ against activation of Ca²⁺ release in central regions of atrial myocytes (Mackenzie et al. 2004; Bootman et al. 2011). When mitochondrial Ca²⁺ uptake was blocked with Ru360 (Matlib et al. 1998) in our experiments, CaT amplitudes during ECC were increased without affecting SR Ca²⁺ content. Changes in mitochondrial function and/or structure are of great interest in cardiac disease (Dominic et al. 2014). In several disease models increases as well as decreases in mitochondrial density have been reported, suggesting that mitochondrial remodelling in cardiac disease is complex (Coleman et al. 1987; Barth et al. 1992; Hohendanner et al. 2013). We observed a decreased mitochondrial density in HF atrial cells, suggesting a smaller capacity to buffer [Ca²⁺]_i while the cell size was not significantly increased (1017 ± 50 vs. 1255 ± 118 μm² in normal (n = 29) vs. HF (n = 36) cells; P = 0.09). Even though the difference in mitochondrial density between normal and HF atrial myocytes appeared rather small at first glance, the consequences for mitochondrial Ca²⁺ uptake were significant. When testing mitochondrial Ca²⁺ uptake directly, not only was the rate of uptake slowed, the maximum uptake was significantly reduced to approximately half (Fig. 6). Our data suggest that atrial remodelling during HF involves a reduction in mitochondrial density that also turns out to have beneficial effects for ECC. By attenuating mitochondrial Ca²⁺ uptake during ECC, SR Ca²⁺ release became more robust and efficient which ultimately resulted in larger and prolonged CaTs and a positive inotropic effect that would facilitate ventricular filling.

Arrhythmogenic SR Ca²⁺ release in HF

Our data show that HF atrial cells reveal a higher frequency of spontaneous Ca²⁺ waves (Fig. 7). Our experiments indicate that increased IICR plays a critical role. While IP₃R inhibition decreased the Ca²⁺ wave frequency in both normal and HF atrial cells, the effect was significantly more pronounced in HF cells. This is consistent with the upregulation of IP₃ signalling in HF and the more pronounced effect of IICR on electrically evoked CaTs (Fig. 3). In addition, atrial cells in general were prone to develop arrhythmogenic Ca²⁺ waves (around 40% of all Ca²⁺ waves in normal atrial cells), i.e. waves that led to subsequent global Ca²⁺ release by triggering an AP and SR Ca²⁺ release typical for atrial ECC. Arrhythmogenic Ca²⁺ waves were not observed in ventricular cells, thus appear to be a unique feature of atrial cells and occurred more often in HF. This is consistent with findings in atrial cells from hearts with myocardial infarction, where the frequency of spontaneous depolarizations was significantly increased (Kettlewell et al. 2013). The spatial propagation pattern of Ca²⁺ waves in atrial cells revealed a unique feature where the front of the propagating wave advanced through the subsarcolemmal space via j-SR Ca²⁺ release, thus preceding the centripetal propagation of CICR from nj-SR. In rabbit atrial cells NCX is exclusively located in the cell periphery due to the lack of transverse tubules. Thus, the subsarcolemmal propagation of the Ca²⁺ wave front exposed NCX for a prolonged period of time to high [Ca²⁺]_i, which might play a critical role in the generation of arrhythmogenic Ca²⁺ waves. Because of the electrogenic nature of NCX, extrusion of elevated [Ca²⁺]_i will lead to membrane depolarization to the extent that it can trigger an AP. We tested this hypothesis by inhibiting NCX pharmacologically (SEA). NCX inhibition decreased the frequency of arrhythmogenic Ca²⁺ waves significantly in normal atrial cells and did so even more in HF atrial cells, indicating that NCX in forward mode can trigger a depolarization of the cell during spontaneous Ca²⁺ waves. Consistent with our finding of an increased propensity of arrhythmogenic Ca²⁺ waves in HF, NCX activity was indeed enhanced in HF atrial myocytes (Fig. 7E).

Conclusions

In hearts with ventricular failure and decreased cardiac output, atrial cells might compensate to maintain ventricular filling despite increased end-diastolic pressure, with a positive inotropic response that results from enhanced atrial CaTs due to increased IICR. The positive inotropic response of atrial cells is aided by reduced Ca²⁺ buffering by mitochondria during ECC. Upregulation of IICR, on the other hand, also has undesirable consequences in form of a significantly higher propensity for arrhythmogenic Ca²⁺ release. The combined action of enhanced IICR together with a higher NCX activity are likely contributors to the arrhythmogenicity found in HF and may form the substrate for the frequently encountered AF in this disease state.

Glossary

AF

atrial fibrillation

Ang II

angiotensin II

AP

action potential

2-APB

2-aminoethoxydiphenyl borate

BSA

bovine serum albumin

[Ca²⁺]_i

cytosolic free Ca²⁺ concentration

[Ca²⁺]_mito

mitochondrial free Ca²⁺ concentration

[Ca²⁺]_SR

sarcoplasmic reticulum free Ca²⁺ concentration

CaT

Ca²⁺ transient

CICR

Ca²⁺-induced Ca²⁺release

CRU

Ca²⁺ release unit

ECC

excitation–contraction coupling

FDHM

full duration at half-maximum

FIRE

fluorescent IP₃-responsive element

FRET

florescence resonance energy transfer

FWHM

full width at half-maximum

GFP

green fluorescent protein

GPCR

G_q-protein coupled receptor

HF

heart failure

IICR

IP₃-induced Ca²⁺release

IP₃

inositol-1,4,5-trisphosphate

IP₃R

IP₃ receptor

j-SR

junctional sarcoplasmic reticulum

LTCC

L-type Ca²⁺ channel

MCU

mitochondrial calcium uniporter

NCX

Na⁺–Ca²⁺ exchange

nj-SR

non-junctional sarcoplasmic reticulum

RyR

ryanodine receptor

SEA

SEA-0400, 2-[4-[(2,5-diuorophenyl) methoxy]phenoxy]-5-ethoxyaniline

SERCA

sarcoplasmic–endoplasmic reticulum Ca²⁺ ATPase

SR

sarcoplasmic reticulum

TTP

time to peak

Additional information

Competing interests

None declared.

Author contributions

Conception and design of the experiments: F.H, L.A.B. Collection, analysis and interpretation of data: F.H., S.W., J.T.M. S.K., S.A., G.L.S., V.A.L., L.A.B. Drafting the article: F.H., L.A.B. All authors have approved the final version of the manuscript.

Funding

This work was supported by National Institutes of Health grants HL62231, HL80101 and HL101235 and the Leducq Foundation (to L.A.B.).