Local beta-adrenergic stimulation overcomes source-sink mismatch to generate focal arrhythmia

Abstract

Rationale

Beta-adrenergic receptor (β-AR) stimulation produces sarcoplasmic reticulum (SR) Ca²⁺ overload and delayed after-depolarizations (DADs) in isolated ventricular myocytes. How DADs are synchronized to overcome the source-sink mismatch and produce focal arrhythmia in the intact heart remains unknown.

Objective

To determine if local β-AR stimulation produces spatio-temporal synchronization of DADs and to examine the effects of tissue geometry and cell-cell coupling on the induction of focal arrhythmia.

Methods and Results

Simultaneous optical mapping of transmembrane potential (V_m) and Ca²⁺ transients (CaT) was performed in normal rabbit hearts during subepicardial injections (50µL) of norepinephrine (NE) or control (normal Tyrodes). Local NE produced premature ventricular complexes (PVCs) from the injection site, which were dose-dependent (low-dose [30–60µM]: 0.45±0.62 vs. high-dose [125–250µM]: 1.33±1.46 PVCs/injection, p<0.0001) and were inhibited by propranolol. NE-induced PVCs exhibited abnormal V_m-Ca²⁺ delay at the initiation site, and were inhibited by either SERCA inhibition or reduced perfusate [Ca²⁺], indicating a Ca²⁺-mediated mechanism. NE-induced PVCs were more common at RV vs. LV sites (1.48±1.50 vs. 0.55±0.89, p<0.01) which was unchanged following chemical ablation of endocardial Purkinje fibers, suggesting source-sink interactions may contribute to the greater propensity to RV PVCs. Partial gap junction uncoupling with carbenoxolone (25µM) increased focal activity (2.18±1.43 vs. 1.33±1.46 PVCs/injection, p<0.05), further supporting source-sink balance as a critical mediator of Ca²⁺-induced PVCs.

Conclusions

These data provide the first experimental demonstration that localized β-AR stimulation produces spatio-temporal synchronization of SR Ca²⁺ overload and release in the intact heart and highlight the critical nature of source-sink balance in initiating focal arrhythmias.

Introduction

Sudden cardiac death (SCD) due to ventricular arrhythmias is a major cause of death in patients with heart failure (HF). Despite decades of research, anti-arrhythmic therapy has failed to reduce this risk.¹ Beta-adrenergic receptor (β-AR) antagonists are a cornerstone of HF therapy² and one of the few treatments to reduce SCD.³ However, the role of β-AR stimulation in arrhythmogenesis is poorly understood. In isolated failing ventricular myocytes, β-AR stimulation produces increased sarcoplasmic reticulum (SR) Ca²⁺ load, spontaneous SR Ca²⁺ release, delayed after-depolarizations (DADs) and triggered activity,^4,5 providing a potential mechanistic link between β-AR stimulation and focal ventricular arrhythmia.^6,7 However, DADs in isolated cells cannot be directly extrapolated to focal arrhythmias in the intact heart where strong electrotonic coupling to the surrounding myocardium acts as a ‘sink’ for local depolarizing current.^8–10 Thus, a DAD occurring in a single cell (the ‘source’) cannot overcome the source-sink mismatch to produce propagating ectopic activity.^11,12 In order for a DAD to produce focal activity in the intact heart, SR Ca²⁺ release must be synchronized across many cells to produce sufficient depolarizing current to overcome the source-sink mismatch. Indeed, mathematical models predict that in well-coupled tissue, synchronized SR Ca²⁺ release must occur in many thousands of myocytes to generate a single premature ventricular complex (PVC).¹² The central questions as to how these cellular events are synchronized, both in space and time, and how many cells must be synchronized to produce focal arrhythmia in the intact heart remain unanswered.

We hypothesized that localized β-AR stimulation could produce spatio-temporal DAD synchronization in the intact heart. Experimental evidence suggests that hetero-geneous β-AR stimulation can occur in HF,¹³ as sympathetic nerve sprouting and remodeling,¹⁴ regional hyperinnervation,^15,16 and differential release of the neurotransmitter norepinephrine (NE)¹⁷ have all been identified. Activation of β-AR by NE increases SR Ca²⁺ load and thus increases the likelihood of spontaneous SR Ca²⁺ release. Therefore, localized β-AR stimulation may produce spatial synchronization of SR Ca²⁺ overload among neighboring myocytes. Temporal synchronization of the NE-induced SR Ca²⁺ release is dictated by the timing of the preceding action potential (AP), Ca²⁺ transient (CaT) and recovery from refractoriness of ryanodine receptors (RyR).^18,19 This combined spatio-temporal synchronization of SR Ca²⁺ overload and release is critical for producing a tissue-level DAD in the intact heart. If the DAD magnitude and local depolarizing current are large enough, the source-sink mismatch will be overcome and focal arrhythmia initiated.

We further hypothesized that, due to the critical nature of source-sink interactions in this scenario, tissue geometry and the degree of cell-cell coupling would contribute to focal arrhythmia propensity in response to localized β-AR stimulation. In order to test these hypotheses, simultaneous optical mapping of transmembrane potential (V_m) and CaT was performed during local subepicardial application of NE in normal rabbit hearts. The propensity of localized NE to initiate focal arrhythmia was quantified and the effects of tissue geometry and cell-cell coupling on the inducibility of PVCs were examined.

Methods

An expanded Methods section is available in the Online Data Supplement. All procedures involving animals were approved by the Animal Care and Use Committee of the University of California, Davis and adhered to the NIH Guide for the Care and Use of Laboratory Animals. Male New Zealand White rabbits (n=29) were anesthetized with an intravenous injection of pentobarbital sodium (50mg/kg). Hearts were excised, Langendorff-perfused at 37°C and loaded with RH237 and Rhod-2AM (Molecular Probes, Eugene, OR) for simultaneous fluorescent imaging of V_m and intracellular Ca²⁺. An electrocardiogram (ECG) was continuously recorded and pacing was from the basal left ventricle (LV). Blebbistatin (Tocris Bioscience, Ellisville, MO; 10–20µM) was used to eliminate motion artifact during optical recordings.²⁰ Fluorescent signals were recorded using two CMOS cameras (MiCam Ultima-L, SciMedia) with a sampling rate of 1kHz and 100×100 pixels with a 35x35mm field of view.

Subepicardial injections were delivered via 30G needles with a 90° bend 1.5mm from the tip. Injections of normal Tyrodes (NT, 50µL: control) and NE (50µL, low-dose [30–60µM] or high-dose [125–250µM]) were delivered at different anatomical locations (LV base/LV apex/RV base/RV apex) in 15 hearts. In a subset of hearts (n=8), co-injection of NE and the fluorophore rhodamine-6G (R6G, 50µM, 528/547nm ex/em, Sigma, St. Louis, MO) was performed to visualize the epicardial area and transmural depth of tissue exposed to NE. In another subset (n=8 hearts), 25µM carbenoxolone (CBX) was added to produce partial gap junction (GJ) uncoupling. A further 14 hearts were used to study the effect of the β-AR antagonist propranolol (5–10µM, n=3), low perfusate [Ca²⁺] (0.33mM, n=3), the SR Ca²⁺-ATPase (SERCA2a) inhibitor cyclopiazonic acid (CPA, 30µM, n=2) and ablation of the RV endocardium with Lugol’s solution (n=3) on the occurrence of NE-induced PVCs. Local caffeine injections (10–40mM, 50µL, n=4) and global perfusion of isoproterenol (1µM, n=4) were also studied.

Data analysis was performed using two analysis programs (BV_Analyze, Brainvision, Japan; and Optiq, Cairn, UK). Activation time was determined as 50% rise time. For APs, repolarization time at 80% return to baseline was used to calculate action potential duration (APD₈₀). For CaTs, duration was measured at 50% (CaTD₅₀) and the time course of decay was quantified using the time constant (τ) of a single exponential fit of the decline (30–100%).²¹ V_m activation time was subtracted from Ca²⁺ activation time to produce maps of V_m-Ca²⁺ delay. Phase plots of the V_m/Ca²⁺ relationship were generated by plotting the normalized V_m values (x-axis) against the normalized Ca²⁺ values (y-axis) for the time course of a single AP, where counterclockwise chirality indicates normal V_m-Ca²⁺ coupling, and clockwise chirality indicates abnormal V_m-Ca²⁺ coupling.²² Conduction velocity (CV) was calculated as in Bayly et al.²³ Continuous variables are presented as mean±SD. Comparisons between two groups of continuous data were made using a Student’s t-test, paired where appropriate, and categorical data using a Fisher’s exact test. Multiple comparisons were made using one- or two-way analysis of variance (ANOVA) with Bonferroni’s post-testing. P<0.05 was considered statistically significant.

Results

Local NE application induces focal arrhythmia

Baseline electrophysiological and Ca²⁺-handling parameters before injection and following complete washout of NE are detailed in the Online Data Supplement and in Supplemental Figure I. Local NE application resulted in PVCs in 15/15 hearts, whereas control injections (NT) produced PVCs in 4/13 hearts (p<0.001). Examples of focal arrhythmia induced by local NE application are shown in Figure 1A. PVCs are characterized by a broad QRS complex and ventriculo-atrial activation sequence, and arise from the injection site. Summary data are shown in Figure 1B. A range of NE doses were examined before grouping the data into high- and low-dose for statistical analysis (Supplemental Figure II). High-dose NE produced significantly more PVCs than NT (NE-high: 1.33±1.46 vs. NT: 0.17±0.43 PVCs/injection, p<0.0001, Figure 1B (i)) and a significantly higher proportion of NE applications resulted in at least one PVC (NE-high: 25/40 [63%] vs. NT: 8/48 [17%], p<0.001, Figure 1B (ii)). Low dose NE gave intermediate responses (NE-low: 0.45±0.62 vs. NE-high: 1.33±1.46 PVCs/injection, p<0.01, Figure 1B (i) and NE-low: 12/31 [39%] vs. NE-high: 25/40 [63%], p=0.058, Figure 1B (ii)).

Figure 1.

A) (i) ECG (black), atrial (red) and ventricular (blue) optical APs following subepicardial injection of 250µM norepinephrine (NE). V_m activation maps showing focal activation from the injection site. (ii) Alternate injection site. B) (i) PVCs/injection and (ii) proportion of injections resulting in at least one PVC for normal Tyrodes’ (NT, n=48 [13 hearts]), low dose NE (n=31 [10 hearts]) and high dose NE (n=40 [9 hearts]). C) The effect of 5µM and 10µM propranolol (PP) on responses to high dose NE (n=4). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Because NE stimulates α- as well as β-AR, the β-AR antagonist propranolol (PP 5–10µM, n=3) was added to the perfusate (Figure 1C). PP reduced PVCs/injection (NE-high: 1.00±0.95, +5µM PP: 0.32±0.48, +10µM PP: 0.06±0.25, p<0.001) and PVC occurrence (NE-high: 62%, +5µM PP: 32%, +10µM PP: 6%, p<0.05) in a dose-dependent manner, suggesting that PVCs were due to β- rather than α-AR stimulation. The propensity to NT-induced PVCs was unchanged with PP (25% vs. 21%, p=NS).

Local NE-induced PVCs are Ca²⁺-mediated

To investigate whether focal activity was due to local SR Ca²⁺ release, maps of V_m-Ca²⁺ delay were constructed. Spatial patterns of V_m-Ca²⁺ delay during NE-induced PVCs were compared to sinus rhythm, ventricular pacing and NT-induced PVCs, as shown in Figure 2. Mean epicardial V_m-Ca²⁺ delay was similar during sinus rhythm (7.04±0.94ms), ventricular pacing (7.02±0.95ms) and NT-induced PVCs (7.39±1.14ms, p=0.65). The presence of local NE did not shorten V_m-Ca²⁺ delay during sinus rhythm (Supplemental Figure IIIA) nor did V_m-Ca²⁺ delay exhibit restitution during shortening of coupling interval (Supplemental Figure IIIB). There were no significant differences in V_m rise times for sinus (14.1±4.1ms), pacing (18.4±4.0ms), NT-induced PVCs (17.0±2.6ms) or NE-induced PVCs (17.9±2.2ms).

Figure 2.

A) V_m activation maps during (i) sinus rhythm, (ii) ventricular pacing, (iii) NT-induced PVC, (iv) NE-induced PVC. B) Maps of V_m-Ca²⁺ delay indicate homogeneous V_m-Ca²⁺ delay during sinus rhythm, pacing, and NT-induced PVCs. During NE-induced PVCs, V_m-Ca²⁺ delay is short at the site of earliest V_m activation. C) Optical APs (black) and CaTs (red) from the earliest activated sites. D) V_m and Ca²⁺ upstrokes from the earliest activated sites. E) Phase plots from the traces shown in C. Chirality is reversed during the NE-induced PVC. F) Mean V_m-Ca²⁺ delay as a function of distance from the earliest activation site along the (i) longitudinal and (ii) transverse axes of conduction. * p<0.05, ***p<0.001.

During normal excitation-contraction (EC) coupling, Ca²⁺ activation followed V_m with a relatively uniform delay across the epicardial surface (Figure 2B (i)-(iii)). However, during the majority of NE-induced PVCs, areas of abnormally short V_m-Ca²⁺ delay were evident at the site of earliest V_m activation (Figure 2B (iv)). APs and CaTs sampled from the site of earliest V_m activation (Figure 2C (iv)) demonstrate near-simultaneous V_m and Ca²⁺ upstrokes (Figure 2D (iv)) and reversed chirality of the V_m-Ca²⁺ phase plot (Figure 2E (iv)). The mean V_m-Ca²⁺ delay over distance from the earliest activation site for NT- and NE-induced PVCs are shown in Figure 2F and demonstrate the significantly shorter V_m-Ca²⁺ delay associated with NE-induced PVCs compared to NT-induced PVCs.

Simultaneous V_m/Ca²⁺ upstroke profiles are consistent with APs induced by SR Ca²⁺ release in isolated myocytes,²⁴ in which caffeine (10mM) was used to rapidly activate RyR opening, and changes in V_m occurred instantaneously upon SR Ca²⁺ release. As subsarcolemmal Ca²⁺ instantaneously drives inward Na⁺/Ca²⁺ exchange (NCX) current, V_m/Ca²⁺ upstrokes are simultaneous.²⁴ To test whether the same phenomenon occurs in the intact heart, local injections of caffeine were applied (Figure 3A-C). Local caffeine-induced PVCs displayed simultaneous V_m/Ca²⁺ upstrokes, exactly as observed during NE-induced PVCs. Corresponding negative control experiments were also performed, in which the probability of SR Ca²⁺ overload and release was reduced, either by lowering perfusate [Ca²⁺] (0.33mM) or by administration of the SERCA2a inhibitor CPA (30µM). As shown in Figure 3D, in the presence of low [Ca²⁺] NE-induced PVCs were reduced (0.83±0.70 vs. 0.20±0.46 PVCs/application, p<0.0001; 67 vs. 16%, p<0.001). Partial SERCA inhibition with CPA (approximately 50% prolongation of τ, data not shown) also reduced NE-induced PVCs (1.21±0.80 vs. 0.42±0.66 PVCs/application, p<0.001; 87 vs. 50%, p<0.05, Figure 3E). In contrast, the occurrence of NT-induced PVCs was not changed by either intervention.

Figure 3.

Left panel – positive control using local caffeine injections. Maps of A) V_m activation time and B) V_m-Ca²⁺ delay during (i) NE-induced PVC and (ii) caffeine-induced PVC (40mM). C) V_m and Ca²⁺ upstrokes from the earliest activated sites, showing near-simultaneous upstrokes in both cases. Right panel – negative control experiments with reduced SR Ca²⁺ load. D) PVC occurrence after reducing perfusate [Ca²⁺] to 0.33mM. E) PVC occurrence after 30µM CPA to partially inhibit SERCA2a activity. *p<0.05, ****p<0.0001.

Subcellular imaging has shown that spontaneous SR Ca²⁺ release events occur at characteristic times after the prior beat, depending on the recovery of SR Ca²⁺ content, SR Ca²⁺ load and RyR refractoriness,^18,19 with a peak at 300–500ms.²⁵ To assess whether this same temporal synchronization was evident for NE-induced PVCs, we measured the coupling interval for each PVC with respect to the previous beat (Supplemental Figure IIIC). The latencies of NE-induced PVCs cluster at coupling intervals of 300–400ms, consistent with a temporal synchronizing effect of RyR restitution. When PVCs were not elicited by NE, subthreshold Ca²⁺ elevation could be observed near the injection site (Supplemental Figure IV). This suggests that local NE application resulted in SR Ca²⁺ release from a group of cells, but in these cases was not large enough to initiate a PVC. Taken together, these data strongly implicate that focal activity during localized β-AR stimulation is mediated by local and relatively synchronous diastolic SR Ca²⁺ release.

RV sites are more prone to NE-induced PVCs

To determine whether differences in tissue geometry or electro�physiological characteristics may play a role in the genesis of focal arrhythmia, PVC inducibility was compared across different sites. There were no significant differences in PVC inducibility between apex and base (NE-high-apex: 1.29±1.54 vs. NE-high-base: 1.35±1.47 PVCs/injection, p=NS; NE-high-apex: 64% vs. NE-high-base: 62%, p=NS). In contrast, local NE in the RV produced more PVCs than in the LV (NE-high-RV:1.94±1.66 vs. LV:0.82±1.10 PVCs/injection, p<0.001, Figure 4A), and a higher proportion of NE injections resulted in at least one PVC (81% vs. 50%, p<0.05). Similar RV/LV trends were observed for low-dose NE. To determine if the differences in PVC inducibility between LV and RV were due to underlying electrophysiological heterogeneity, AP and CaT parameters were compared. There were no differences in CV, AP rise time, APD₈₀, dispersion of APD₈₀ or CaT dynamics between LV and RV (Table 1). To determine whether functional heterogeneity in the response to β-AR stimulation may exist between the ventricles, 1µM ISO was added globally. As shown in Table 1, there were no differences between LV and RV in the response to global β-AR stimulation with ISO.

Figure 4.

A) PVCs/injection in LV and RV. ***p<0.001 vs. LV. B) V_m activation maps (i) and R6G fluorescence (ii) following co-injection of NE and R6G. C) PVC probability at LV and RV sites as a function of NE tissue exposure. RV injection sites had a higher probability of PVC induction. r²: LV, 0.46, RV, 0.88; regression slopes: LV, 0.0021±0.0008, p<0.05 and RV, 0.0025±0.0005, p<0.05, p=NS, y-axis intercepts: LV, 0.03±0.19 vs. RV, 0.33±0.12, * p<0.05. D) Transmural histological sections of: (i) LV and (ii) RV injection sites with R6G fluorescence in green.

Table 1.

AP and CaT characteristics

Mean±SD	LV	RV	P
Baseline	n=11	n=11
APD80 (ms)	185.3±18.6	183.3±17.7	NS
APD80 dispersion (ms/mm2)	0.23±0.21	0.22±0.17	NS
AP rise time (ms)	13.8±3.2	14.2±2.4	NS
Conduction velocity (cm/s)	58.5±9.7	56.8±12.3	NS
CaT rise time (ms)	20.4±1.5	21.3±2.4	NS
CaT decay (τ, ms)	65.5±9.8	66.3±7.9	NS
Global ISO	n=4	n=4
CaT rise time (ms)	18.5±2.8	17.4±1.5	NS
CaT decay (τ, ms)	42.3±10.9	41.4±5.9	NS

AP= action potential, APD₈₀=AP duration at 80% repolarization, CaT=calcium transient

Differences in tissue structure and local vasculature may result in non-uniform NE diffusion within tissue and may contribute to the regional differences in PVC propensity. In order to address this, the fluorophore R6G was added to NE to visualize the area of myocardium exposed. Figure 4B shows V_m activation time and R6G fluorescence following co-injection of NE and R6G. PVCs arise from the injection site and R6G fluorescence indicates the spatial extent of NE exposure. The area of epicardial tissue exposed to NE was correlated with PVC inducibility (Figure 4C). Both LV and RV injection sites displayed a similar increase in PVC probability with increasing NE exposure area (regression slopes: LV, 0.0021±0.0008, p<0.05 and RV, 0.0025±0.0005, p<0.05, LV vs. RV p=NS). However, for RV sites, PVC probability was higher across the range of exposure areas measured (y-axis intercepts: LV, 0.03±0.19 vs. RV, 0.33±0.12, p<0.05). Transmural histological sections at the injection site were then imaged to determine if the depth of NE exposure could account for the difference in PVC inducibility. Figure 4D shows LV and RV injection sites with similar epicardial areas of R6G fluorescence. In the LV, the R6G fluorescence extends to the midmyo�cardium, whereas in the RV, the R6G exposure is fully transmural. These differences in exposure may result in less source-sink mismatch in the RV compared to the LV (~2- vs. 3-dimensional), and therefore higher PVC probability.

In order to investigate whether transmural NE exposure in the RV was activating endocardial Purkinje fibers to initiate focal activity, NE injections were repeated before and after RV endocardial chemical ablation with Lugol’s solution. As shown in Figure 5A, Lugol’s produced QRS prolongation and delayed RV activation during sinus rhythm, consistent with disruption of the Purkinje network, which was subsequently confirmed with triphenyl tetrazolium chloride (TTC) staining. The occurrence of NE-induced PVCs in the RV was not affected by chemical ablation of the endocardium (1.0±0.6 vs. 1.1±0.6 PVCs/injection, p=NS; 85 vs. 85%, p=NS, Figure 5B), suggesting that the Purkinje fibers are not playing a large role in initiating focal activity during local β-AR stimulation.

Figure 5.

A) RV endocardial ablation with Lugol’s solution. (i) ECG, (ii) V_m activation map and (iii) endocardial TTC staining before (CON) and after (+Lugol’s) RV endocardial Lugol’s. Purkinje ablation resulted in QRS prolongation and delayed RV epicardial activation. Tissue ablation was confirmed by the white appearance on TTC. B) Occurrence of PVCs was not affected by endocardial ablation.

Partial gap junction uncoupling promotes NE-induced PVCs

To further investigate the role of source-sink interactions, NE application was repeated in the presence of CBX to induce partial GJ uncoupling. 25µM CBX resulted in a ~25% decrease in CV in both the longitudinal (CV_L) and transverse (CV_T) directions (CV_L 58.7±4.2 vs. 38.5±3.1cm/s, p<0.001; CV_T 43.9±3.2 vs. 28.2±3.1cm/s, p<0.001) (Supplemental Figure V). The addition of CBX did not affect CaTD₅₀ (140±9 vs.141±10ms, p=0.76) or V_m-Ca²⁺ delay during sinus rhythm (7.53±1.41 vs. 7.04±0.94ms, p=0.47) or ventricular pacing (7.12±1.58 vs. 7.02±0.95ms, p=0.89). With CBX, greater numbers of PVCs were induced by both low-dose NE (CBX: 1.48±1.33 vs. control: 0.45±0.62 PVCs/injection, p<0.01) and high-dose NE (CBX: 2.18±1.43 vs. control: 1.33±1.46 PVCs/injection, p<0.05) and a higher proportion of NE injections resulted in at least one PVC (NE-low: 76 vs. 39%, p<0.05; NE-high: 88 vs. 63%, p=0.06) (Figure 6).

Figure 6.

A) Number of PVCs/injection and B) proportion of injections resulting in at least one PVC for low (grey) and high dose NE (black) under control conditions (solid bars) and with CBX (hatched bars). *p<0.05 vs. control, **p<0.01 vs. control.

Source-sink interactions mediate the production of NE-induced PVCs

To determine the mechanism of increased propensity to focal arrhythmia during partial GJ uncoupling, the changes in V_m-Ca²⁺ dynamics were examined (Figure 7). β-AR activation produces more rapid SR Ca²⁺ uptake and thus faster CaT decline. Shortening of CaTD₅₀ around the injection site was observed during NE- but not NT-induced PVCs (Figure 7B). This area of abbreviated CaTD₅₀ can be used as a functional measure of the spatial extent of NE effects. However, due to electrotonic coupling, effects on V_m may not be seen across the same area. Thus, the area of abnormal V_m-Ca²⁺ delay was used as a functional estimate of the spatial extent over which Ca²⁺ release exerts an effect on V_m (Figure 7C). To quantify these parameters between hearts, normalized CaTD₅₀ and V_m-Ca²⁺ delay during NT and NE-induced PVCs were plotted as a function of distance from the earliest V_m activation (Figure 7D and E, respectively [longitudinal direction only, transverse profiles shown in Supplemental Figure VI]). During NE-induced PVCs, CaTD₅₀ was significantly shorter over the first 4mm, and this was unchanged with CBX (Figure 7D). V_m-Ca²⁺ delay was shorter over the first 4mm under control conditions, and this increased to 7mm with CBX (Figure 7E). Thus, after uncoupling, a larger effect on V_m is observed for the same NE effect on CaTD₅₀. This is consistent with an increased propensity to Ca²⁺-mediated focal arrhythmia induced by NE during partial GJ uncoupling.

Figure 7.

A) V_m activation during (i) NT-induced PVC, (ii) NE-induced PVC and (iii) NE-induced PVC with CBX. B) Corresponding maps of CaTD₅₀ and (C) V_m-Ca²⁺ delay. D) CaTD₅₀ as a function of distance during NT-induced PVCs (n=11 [6 hearts]), NE-induced PVCs under control conditions (n=11 [7 hearts]) and NE-induced PVCs with CBX (n=10 [6 hearts]). During NE-induced PVCs, CaTD₅₀ is significantly shorter (compared with NT-induced PVCs) for 4mm, and is not affected by CBX. E) Corresponding plot of V_m-Ca²⁺ delay. During NE-induced PVCs, V_m-Ca²⁺ delay is significantly shorter for 4mm in the longitudinal axis, increasing to 7mm with CBX. Similar results were observed in the transverse direction (Supplemental Figure VI). * p<0.05.

Discussion

This study tested the hypothesis that, in the intact heart, localized β-AR stimulation can provide spatio-temporal syn�chronization of SR Ca²⁺ release over many cells to overcome source-sink mismatch and produce focal arrhythmia. The results demonstrate that local NE application induces β-AR dependent, Ca²⁺-mediated focal arrhythmia in normal rabbit hearts. Quantification of V_m-Ca²⁺ dynamics established the mechanism of PVC induction. Abnormal V_m-Ca²⁺ delays were exclusively a feature of NE- and caffeine-induced PVCs, suggesting that focal arrhythmias induced by localized β-AR stimulation are mediated by SR Ca²⁺ overload and release. RV sites displayed a higher propensity to focal arrhythmia than LV. This difference could not be explained by electrophysiological differences or activation of Purkinje fibers and is thus likely due to the reduction in the source-sink mismatch in the thinner RV wall that more closely approximates a 2-dimensional geometry. Consistent with this finding, partial GJ uncoupling resulted in a reduction of the current sink and was associated with a higher propensity for focal arrhythmia in response to localized β-AR stimulation.

Short V_m-Ca²⁺ activation delay is indicative of Ca²⁺-mediated focal activity

During normal EC coupling, V_m-Ca²⁺ delay was 7–8ms, similar to that reported in rabbits²² and humans.²⁶ In contrast, during NE-induced PVCs, significantly shorter V_m-Ca²⁺ delays (0–4ms) were observed at the initiation site. Such short V_m-Ca²⁺ delays were never observed during normal EC coupling, indicating that during NE-induced PVCs SR Ca²⁺ release likely started before the AP upstroke and opening of L-type Ca²⁺ channels. We tested the possibility that local NE application might accelerate Ca²⁺ release dynamics and abbreviate the V_m-Ca²⁺ delay during normal EC coupling. However, localized NE did not alter V_m-Ca²⁺ delay during sinus beats (Supplemental Figure IIIA). V_m-Ca²⁺ delay was also unaltered over a broad range of coupling intervals, ruling out EC coupling restitution kinetics as a cause of the abnormally short V_m-Ca²⁺ delay during NE-induced PVCs (Supplemental Figure IIIB). The observations that partial SERCA inhibition with CPA and low [Ca²⁺] perfusate reduced NE- but not NT-induced PVCs (Figure 3D-3E) as well as local intracellular Ca²⁺ rises during subthreshold NE application (Supplemental Figure IV) all support the conclusion that NE-induced PVCs are Ca²⁺-mediated. As short V_m-Ca²⁺ delays were only observed during NE and caffeine-induced PVCs, these data suggest that very short V_m-Ca²⁺ activation delays can be used as an indicator of Ca²⁺-mediated excitation.

In optical mapping studies of pharmacological QT prolongation²² and post-shock arrhythmias²⁷ Ca²⁺ upstrokes preceding V_m upstrokes or slow diastolic Ca²⁺ elevation prior to V_m upstrokes have been reported. These models of Ca²⁺-mediated arrhythmia are characterized by ‘global’ SR Ca²⁺ overload and release, which results in gradual and spatially dyssynchronous Ca²⁺ elevation. In our experiments, we observed a different phenomenon, in that we saw rapid Ca²⁺ release in response to local NE. During NE- induced PVCs, Ca²⁺ and V_m upstrokes occurred nearly simultaneously at the site of earliest activation. This is highly analogous to rapid caffeine application in isolated cardiomyocytes, where synchronous SR Ca²⁺ release immediately activates NCX and depolarization occurs simultaneously with the Ca²⁺ rise.²⁴ Indeed, local application of caffeine in the intact heart recapitulated these results (Figure 3A-C). Thus, although the simultaneous V_m/Ca²⁺ upstrokes reported here differ from those previously reported for Ca²⁺-mediated arrhythmia,^22,27 this is likely attributable to the ‘local’ synchronous SR Ca²⁺ release produced in the current experiments as opposed to ‘global’ dyssynchronous Ca²⁺ release observed in other models.

Induction of a PVC requires temporal as well as spatial synchrony of SR Ca²⁺ release. We hypothesized that this temporal synchronization is dictated by the previous AP/CaT because of the time required for SR Ca²⁺ refilling and RyR restitution.^18,19 Indeed, in Ca²⁺ overload conditions there is an increased amplitude, frequency and synchrony of Ca²⁺ waves.²⁵ These factors increase the likelihood that neighboring Ca²⁺ overloaded myocytes will exhibit SR Ca²⁺ release in relative synchrony. We observed a clustering of NE-induced PVCs occurring 300–400ms after the previous beat (Supplemental Figure IIIC). In contrast, although the number of NT-induced PVCs was small, they appeared to show random coupling intervals. These data are consistent with a temporal synchronization of SR Ca²⁺ release in response to local NE through SR Ca²⁺ refilling and RyR restitution.

Regional heterogeneity in vulnerability to focal arrhythmia

Local β-AR stimulation was more likely to produce PVCs at RV than at LV sites. The finding of increased propensity to PVCs in the RV is congruent with clinical observations. Idiopathic VT/VF is often found to be triggered by PVCs arising from a focal RV origin, ablation of which has been shown to reduce subsequent VT/VF.²⁸ Although PVC inducibility was related to area of epicardial NE exposure as assessed by co-injection of NE with fluorescent R6G (Figure 4C), differences in the area of NE exposure did not account for higher arrhythmia propensity in the RV. The mechanisms underlying the increased arrhythmogenic potential of the RV are not well understood. If differences in sympathetic drive exist, then this may play an important role. Differences in I_K1 may influence the dominant frequency of VF in guinea-pig hearts.²⁹ Although we did not detect significant interventricular electrophysiological or Ca²⁺ handling differences at baseline or during global β-AR stimulation (Table 1), it is possible that differences in specific channels, transporters (e.g. I_K1, NCX, RyR), or β-ARs may contribute to this RV-LV difference. Varying degrees of interventricular heterogeneity in ion channel expression have been reported in different species,^29,30 but optical mapping studies in intact rabbit hearts are congruent with our observations that no significant differences in overall electrophysiology exist between RV and LV at baseline.³¹

Another possible explanation for the RV/LV difference in PVC propensity may be that transmural NE exposure in the RV results in stimulation of the Purkinje network. As Purkinje fibers may be more prone to focal activity,²⁷ activation with NE may result in greater PVC incidence in the RV. To test for this, we performed chemical ablation of the RV endocardial surface with Lugol’s solution and observed no significant difference in PVC propensity (Figure 5), suggesting that the higher incidence of PVCs in the RV is likely not attributable to Purkinje activation.

Tissue geometry and the source-sink balance

Another logical contributor to the observed RV/LV differences is inherent geometric source-sink balance. As illustrated in Figure 4D, for the same depth of NE exposure, the RV will experience 2-dimensional radial current sink around the circumference of the exposed area, whereas in the thicker-walled LV, the current sink is 3-dimensional. The importance of dimensionality of the surrounding current sink has been demonstrated in numerical simulations, which have estimated the number of cells with synchronized SR Ca²⁺ release required to produce a PVC.¹² Xie et al calculated that in normal well-coupled 3D tissue, approximately 820,000 cells with synchronized SR Ca²⁺ release were required to produce a PVC, 100-fold more than were required in the equivalent 2D model.

In simulations, the total current source can be precisely controlled by changing the number of cells with synchronized SR Ca²⁺ release. In the present experiments, however, we can neither exactly measure nor systematically vary the total current source, but we can make estimates from functional measures of V_m-Ca²⁺ dynamics. One way to estimate the current source is to use the area over which local NE abbreviates CaTD (i.e. the area of tissue affected by NE). With strong electrotonic coupling, however, V_m is ‘clamped’ at the borders of this region, thus the alternative estimate, the area over which SR Ca²⁺ release drives depolarization (i.e. where V_m-Ca²⁺ delay is short) may be different. Using the dimensions of these functional measurements, along with anatomical surface and depth estimates from R6G imaging and applying them to a hemi-ellipse (surface radii, 0.5–5mm and depth of 1–2mm), the volume of current source estimated in the present study is 0.5–105mm³. Assuming a cell volume of 30pL³² and 30% extracellular space this equates to 12,200–2,400,000 cells. While this is a relatively crude estimate, the upper and lower limits are very similar to the estimates derived from the mathematical predictions (2D tissue ~8,000 cells and 3D tissue ~820,000 cells).¹²

The effect of partial GJ uncoupling

Partial GJ uncoupling with CBX resulted in a slowing of CV with preserved anisotropy ratio and APD, as has previously been reported in rabbit ventricular myocardium.³³ Partial GJ uncoupling produced a two-fold increase in NE-induced PVCs (Figure 6). Partial uncoupling facilitates PVC propagation by ‘insulating’ the depolarized region (source) from the surrounding resting myocardium (sink),¹⁰ analogous to initiation of pacemaker activity in the sino-atrial node.^34,35 In the case of local β-AR stimulation, a reduction in electrotonic current flow through GJ reduces the hyperpolarizing current from surrounding cells, which allows the current carried by NCX in response to SR Ca²⁺ release to have a larger effect on V_m. This is demonstrated by the significantly larger area of abnormally short V_m-Ca²⁺ delay observed following uncoupling (Figure 7E) despite a similar area of NE effect (Figure 7D). Given that the area of NE effect is not changed, the larger area of abnormal V_m-Ca²⁺ delay with CBX likely reflects poorer voltage clamp at the periphery, due to reduced coupling of the depolarizing source to the surrounding sink.

Local β-AR stimulation as a pathophysiological experimental paradigm

In the present study, we examined how localized β-AR stimulation produces spatio-temporal synchronization of SR Ca²⁺ overload and release across many cells to initiate focal arrhythmia. The data demonstrate that local NE induces Ca²⁺-mediated PVCs in intact hearts, providing a critical link from cellular studies of focal arrhythmia initiation via DADs to the tissue level. Intrasynaptic NE concentrations at vascular nerve terminals have been measured at up to 100µM, and an inverse relationship between intrasynaptic NE concentration and junctional space was observed.³⁶ Because cardiac neuroeffector junctions are narrower, the NE concentrations used in the current study (30–250µM) may be similar to NE concentrations during physiological nerve activation, especially in HF when sympathetic drive is high. However, sympathetic stimulation does not typically result in PVCs in the normal rabbit or human heart and no exogenous NE application can fully recapitulate neuronal NE release.

Regionally heterogeneous sympathetic nerve sprouting, nerve remodeling, hyperinnervation^13–15 and heterogeneous NE release¹⁷ all occur in HF. Focal arrhythmias are known to occur in animals and humans with HF due to dilated cardiomyopathy.^6,7 Moreover, GJ uncoupling is a consistent feature of the electrophysiological phenotype in failing hearts.^{37, 38} Therefore, subepicardial NE injection is a useful model of the heterogeneous β-AR stimulation which occurs in HF and localized β-AR stimulation in the presence of GJ uncoupling represents a pathophysiologically relevant paradigm of focal arrhythmia induction. This is the first experimental study to determine a mechanistic link between altered local sympathetic stimulation and the generation of focal arrhythmia in the intact heart. Future studies should examine whether other features of HF remodeling also potentiate focal arrhythmias induced by local β-AR stimulation.

Study limitations

A small number of PVCs were induced by NT, likely due to activation of stretch-activated channels.³⁹ As demonstrated in Figures 2 and 7, these were clearly distinguishable from NE-induced PVCs on the basis of V_m-Ca²⁺ delay. R6G was used to indicate the area of NE exposure and, as the two have similar molecular weight (319 vs. 479), we expect that the diffusion kinetics in tissue may be similar. However, R6G⁴⁰ is more lipophilic⁴¹ and is not taken up by nerve terminals, thus does not demonstrate the time course of NE clearance from the tissue. There was variability in the area of tissue exposed to NE despite constant injection volume. The different patterns of R6G staining, both on the epicardium and transmurally, suggest this may be due to differences in local tissue architecture and microvasculature. Additional effects of β-AR stimulation (e.g. left-shift in Ca²⁺ current voltage-dependence, altered RyR properties), may further promote DADs and the development of focal arrhythmias, and these issues merit further study. In the current study, we did not determine the exact delay from NE injection to PVC appearance, partly because in dual V_m-Ca²⁺ imaging we could only record for several seconds. However, the first PVC typically occurred a few seconds after NE injection (as in Fig 1Ai and ii), consistent with the expectation that several beats would be required for β-AR activation to elevate SR Ca²⁺ load and release. As discussed above, focal epicardial application of NE only recapitulates some aspects of neuronal NE release, but does allow control and evaluation of exposed area. In future studies it would be valuable to assess the detailed time course of NE-induced PVCs and also to compare the responses to local injections to chemical or electrical stimulation of endogenous NE release.

Conclusions

Localized β-AR stimulation produced spatio-temporal synchronization of SR Ca²⁺ overload and release across many cells to produce focal activity in normal rabbit hearts. Source-sink interactions were found to be critically important in the generation of Ca²⁺-mediated focal arrhythmias. These data provide: (1) the first experimental demonstration of localized β-AR stimulation as a pathophysiologically relevant mechanism of focal arrhythmia initiation; (2) the first experimental evidence of a mechanistic link between sympathetic dysfunction and the generation of focal arrhythmias in the intact heart; (3) the first experimental estimation of the number of cells with synchronized SR Ca²⁺ release required to initiate focal arrhythmia and (4) a demonstration of the critical nature of the source-sink balance in the initiation of focal arrhythmias in the intact heart.

Novelty and significance.

What is known?

• In isolated cardiac myocytes, β-adrenergic stimulation can cause spontaneous Ca²⁺ release from the sarcoplasmic reticulum, which may depolarize the membrane and lead to triggered action potentials.

• In the intact heart, strong electrotonic coupling exists between cells, meaning that spontaneous Ca²⁺ release in a single cell cannot produce sufficient change in membrane potential to trigger propagating action potentials and arrhythmia (‘source-sink mismatch’).

• For spontaneous Ca²⁺ release to trigger arrhythmias in the intact heart, it must be synchronized across many cells, but little is known about how many are required, or the mechanism of synchrony.

What new information does this article contribute?

• Localized β-adrenergic stimulation by norepinephrine injection leads to Ca²⁺- mediated focal arrhythmia in the intact rabbit heart.

• Local β-adrenergic stimulation synchronizes spontaneous Ca²⁺ release from the sarcoplasmic reticulum across thousands of cells, overcoming the source-sink mismatch.

• Source-sink interactions are critically important in the generation of Ca²⁺-mediated focal arrhythmia.

This study addressed the important disconnect between our understanding of arrhythmia mechanisms in isolated cells and the intact heart. β-adrenergic receptor (β-AR) stimulation produces Ca²⁺ -mediated arrhythmias in cells, but electrotonic coupling in the intact heart means that spontaneous Ca²⁺ release in one cell is insufficient to significantly alter membrane potential (V_m). We examined how synchronization of spontaneous Ca²⁺ overload and release might occur in order for the source-sink mismatch to be overcome, and focal arrhythmia generated. Using dual optical mapping of V_m and Ca²⁺, we demonstrate that local norepinephrine application induces β-AR dependent, Ca²⁺-mediated focal arrhythmia in rabbit hearts. The source-sink balance was crucial to focal arrhythmia induction, which was potentiated when the source:sink ratio was increased, either by changes in tissue geometry or by reducing electrotonic conduction through gap junctions (both of which modify the sink). The data also allowed the first experimental estimation of the number of cells required to have synchronous Ca²⁺ release in order to initiate focal arrhythmia. These data provide the first demonstration of localized β-AR stimulation as a mechanism of focal arrhythmia initiation and evidence of a mechanistic link between sympathetic dysfunction and focal arrhythmias in the intact heart.

Non-standard abbreviations and acronyms

AP

Action potential

APD

Action potential duration

APD₈₀

Action potential duration at 80% repolarization

β-AR

Beta-adrenergic receptor

BCL

Basic cycle length

CaT

Calcium transient

CaTD

Calcium transient duration

CaTD₅₀

Calcium transient duration at 50% amplitude

CBX

Carbenoxolone

CPA

Cyclopiazonic acid

CV

Conduction velocity

DAD

Delayed afterdepolarization

DMSO

Dimethyl sulfoxide

ECG

Electrocardiogram

GJ

Gap junction

HF

Heart failure

LV

Left ventricle

NE

Norepinephrine

NT

Normal Tyrodes

PP

Propranolol

PVC

Premature ventricular complex

R6G

Rhodamine-6G

RyR

Ryanodine receptor

RV

Right ventricle

SCD

Sudden cardiac death

SERCA

Sarcoplasmic reticulum Ca²⁺-ATPase

SR

Sarcoplasmic reticulum

TTC

Triphenyl tetrazolium chloride

V_m

Transmembrane potential