Modulating cardiac conduction during metabolic ischemia with perfusate sodium and calcium in guinea pig hearts

Abstract

We previously demonstrated that altering extracellular sodium (Na_o) and calcium (Ca_o) can modulate a form of electrical communication between cardiomyocytes termed “ephaptic coupling” (EpC), especially during loss of gap junction coupling. We hypothesized that altering Na_o and Ca_o modulates conduction velocity (CV) and arrhythmic burden during ischemia. Electrophysiology was quantified by optically mapping Langendorff-perfused guinea pig ventricles with modified Na_o (147 or 155 mM) and Ca_o (1.25 or 2.0 mM) during 30 min of simulated metabolic ischemia (pH 6.5, anoxia, aglycemia). Gap junction-adjacent perinexal width (W_P), a candidate cardiac ephapse, and connexin (Cx)43 protein expression and Cx43 phosphorylation at S368 were quantified by transmission electron microscopy and Western immunoblot analysis, respectively. Metabolic ischemia slowed CV in hearts perfused with 147 mM Na_o and 2.0 mM Ca_o; however, theoretically increasing EpC with 155 mM Na_o was arrhythmogenic, and CV could not be measured. Reducing Ca_o to 1.25 mM expanded W_P, as expected during ischemia, consistent with reduced EpC, but attenuated CV slowing while delaying arrhythmia onset. These results were further supported by osmotically reducing W_P with albumin, which exacerbated CV slowing and increased early arrhythmias during ischemia, whereas mannitol expanded W_P, permitted conduction, and delayed the onset of arrhythmias. Cx43 expression patterns during the various interventions insufficiently correlated with observed CV changes and arrhythmic burden. In conclusion, decreasing perfusate calcium during metabolic ischemia enhances perinexal expansion, attenuates conduction slowing, and delays arrhythmias. Thus, perinexal expansion may be cardioprotective during metabolic ischemia.

NEW & NOTEWORTHY This study demonstrates, for the first time, that modulating perfusate ion composition can alter cardiac electrophysiology during simulated metabolic ischemia.

INTRODUCTION

Myocardial ischemia is one of the leading causes of cardiovascular death in the United States (39). Inability to support the metabolic demands of the working myocardium results in several acute changes including gap junctional uncoupling (20, 35) and ion channel remodeling (40, 42), which lead to functional consequences such as slow and aberrant conduction, potentially leading to fatal arrhythmias (22, 64).

Previous studies have explored modulating gap junctional coupling (GJC) during ischemia as a therapeutic strategy, and drugs that modulate connexin (Cx)43 coupling between ventricular myocytes have been developed for treating cardiac diseases (27, 29, 41, 55). Rotigaptide, for example, has previously demonstrated an ability to attenuate conduction slowing and prevent arrhythmias in various models of ischemia by altering Cx43 phosphorylation (3, 10, 27, 65). However, connexin proteins are complexly phosphorylated, and, furthermore, altered phosphorylation is only one method of Cx43 modification during ischemia, which can manifest as total protein downregulation, altered Cx43 subcellular localization, or other posttranslational modifications (17, 26, 30, 48).

Our recent studies have suggested that cardiac myocytes may also communicate electrically via an extracellular pathway termed “ephaptic coupling” (EpC). Specifically, we provided evidence that changing the concentration of extracellular cations such as sodium, potassium, and calcium (Na_o, K_o, and Ca_o, respectively) can modulate conduction velocity (CV) in response to genetic and pharmacologically induced gap junction uncoupling (13, 15, 16). We theorized that altering Na_o and K_o changes the driving force and sodium channel availability driving EpC (16). Furthermore, we observed that intermembrane separation in the gap junction-adjacent perinexus, a candidate structure for the cardiac ephapse, decreased when Ca_o increased (13, 14, 16).

Regardless of the mechanisms that underlie conduction sensitivity or insensitivity to loss of GJC, we hypothesized that altering extracellular cationic composition should modulate conduction slowing during ischemia. To test this hypothesis, we altered Na_o and Ca_o within a physiological range and determined whether these cations alter arrhythmic propensity in a model of metabolic ischemia.

Our findings indicate that varying perfusate ion composition has a significant effect on CV and arrhythmogenesis during metabolic ischemia without any significant effects during baseline or reperfusion conditions. More specifically, and surprisingly, theoretically reducing EpC by reducing Na_o or facilitating perinexal expansion may be cardioprotective during metabolic ischemia.

METHODS

All protocols were approved by the Institutional Animal Care and Use Committee of Virginia Polytechnic Institute and State University and were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Langendorff Heart Preparation

Male Hartley guinea pigs (13–15 mo old) were anesthetized with isoflurane (14, 19). Hearts were excised after thoracotomy, cannulated to a Langendorff system as previously described with atria removed to prevent competitive stimulation from endogenous pacemakers (60), and perfused with the various solutions shown in Table 1 bubbled with 100% oxygen and maintained at 37°C. HCl (1 N) was added to bring pH to 7.4 for baseline and 6.5 for simulated ischemia. Note that K_o is 6.9 mM in part to simulate an anticipated rise in K_o associated with more severe forms of no-flow ischemia. Hearts were immersed in a three-dimensional printed poly(lactic acid) bath (11) containing the same solution as it was perfused with, also maintained at 37°C. Perfusion pressure was maintained at 40–50 mmHg by adjusting flow rate.

Table 1.

Ionic composition of the solutions used in the ischemia study

	Solution A	Solution B	Solution C	Solution D	Solution E	Solution F
Concentration, mM
CaCl2	2.0	2.0	1.63	1.25	2.0	1.25
NaCl	141.5	149.5	149.5	149.5	149.5	149.5
NaOH	5.5	5.5	5.5	5.5	5.5	5.5
Total Nao	147	155	155	155	155	155
KCl	6.9	6.9	6.9	6.9	6.9	6.9
MgCl2	0.7	0.7	0.7	0.7	0.7	0.7
Glucose	5.5	5.5	5.5	5.5	5.5	5.5
HEPES	9.99	9.99	9.99	9.99	9.99	9.99
2,3-Butanedione monoxime	7.5	7.5	7.5	7.5	7.5	7.5
Mannitol	0	0	0	0	0	60.15
Albumin	0	0	0	0	45.3	0
Osmolarity, mOsm
Baseline	338.89	354.89	353.78	352.64	412.79	400.19
Ischemia	335.27	351.27	350.16	349.02	409.17	396.57

To simulate metabolic ischemia, glucose was replaced in the solution with sucrose, pH was reduced to 6.5 by addition of hydrochloric acid, and solutions were bubbled with N₂-CO₂ mixture for at least 1 h before perfusion.

Optical Mapping

After a 30-min stabilization period, hearts (n = 6 hearts/solution) were perfused with di-4-ANEPPS at a concentration of 7.5 μM for ~10 min followed by a 10-min washout period. An electromechanical uncoupler, 2,3-butanedione monoxime, was used to reduce motion artifacts. A unipolar silver wire pacing electrode was placed on the anterior epicardial surface of the heart, and the heart was paced by stimuli at 1-V amplitude, 5-ms duration, and 300-ms basic cycle length. A reference wire was placed at the back of the bath. The hearts were only paced during the specified time points for optical mapping; hearts were allowed to beat at their intrinsic rate throughout the rest of the protocol. To map transmembrane potential, the dye was excited by light passed through a 510-nm filter, and emission light was collected through a 610-nm filter by an Ultima L-type complementary metal-oxide-semiconductor (CMOS) camera at a sampling rate of 1,000 Hz. A tandem lens optical mapping system with a spatial resolution of 0.318 mm in the x- and y-dimensions after 2 × 2 binning was used in this study.

These data were then analyzed as previously described (13, 14, 16) to determine CV in the longitudinal (CV_L) and transverse (CV_T) directions, anisotropic ratio (AR = CV_L/CV_T), action potential duration (APD), and optical rise time (RT). Briefly, a parabolic surface was fit to activation times, defined as the maximum rate of rise of action potential, to determine CV. APD₃₀ and APD₉₀ were defined as the time interval between activation time and 30% and 90% repolarization, respectively. RT was defined as the time interval between 20% and 80% increases of the upstroke of the action potential.

Volume-conducted ECGs were obtained from three silver chloride electrodes placed in the bath. Signals were sampled at 1,000 Hz and filtered to remove noise (60).

Metabolic Challenge Protocol

The experimental protocol included a 15-min baseline period followed by 30 min of metabolic ischemia and, finally, 20 min of reperfusion. To simulate metabolic ischemia, the solutions shown in Table 1 were modified as follows: solutions were bubbled with an N₂-CO₂ mixture (instead of 100% O₂) for at least 1 h before and during perfusion (hypoxia), pH was reduced to 6.5 (acidosis), and glucose was substituted with sucrose (aglycemia). During reperfusion, hearts (n = 3 hearts/solution) were perfused with the same solution as during baseline. Importantly, the ionic composition of the solutions was maintained throughout the experiment.

Transmission Electron Microscopy

At the end of the baseline, metabolic ischemia, and reperfusion phases, tissue was fixed in 2.5% glutaraldehyde at 4°C overnight and then transferred to PBS and stored at 4°C. Samples were then processed and sectioned onto copper grids as previously described (13, 14, 16) and imaged using a JEOL JEM-1400 electron microscope at ×150,000 magnification. Images of the perinexi were then analyzed using ImageJ, and perinexal width (W_P) was determined. W_P values at 15-nm intervals between 30 and 105 nm from the edge of the gap junction plaque were averaged and are reported during baseline, metabolic ischemia, and reperfusion. W_P data are reported as means ± SE.

Western Blot Analysis

Left ventricular tissue was snap frozen at specific time points in the protocol, and Western blot analysis was performed as previously described (51). Briefly, samples were homogenized in RIPA lysis buffer [containing 50 mM Tris (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 2 mM NaF, 200 μM Na₃VO₄, and 5 mM N-ethylmaleimide] supplemented with Roche Protease Inhibitor Cocktail (no. 4693159001, Sigma-Aldrich). Protein concentration was determined by a Bio-Rad DC protein assay, and concentrations were normalized before analysis. Electrophoresis was performed to separate proteins, which were then transferred to a PVDF membrane, blocked with 5% BSA for 1 h at room temperature, and incubated overnight with primary antibodies against Cx43 phosphorylated at S368 (pS368-Cx43; 1:1,000, no. 3511, Cell Signaling Technology) at 4°C. Membranes were then washed and incubated with secondary antibody (1:5,000, goat anti-rabbit, horseradish peroxidase, Abcam) at room temperature for 1 h. After washing, bound antibody was detected using West Pico Plus chemiluminescent substrate (Thermo Scientific) and imaged using the Li-Cor Odyssey Fc system. Membranes were stripped with ReBlot Plus according to the manufacturer’s instructions and were blocked in Odyssey Blocking Buffer (Li-Cor) at room temperature for 1 h and incubated with primary antibodies against Cx43 (1:5,000, no. C-2619, rabbit, Sigma-Aldrich) and GAPDH (1:5,000, no. 101983-284, mouse, VWR). Membranes were then washed and incubated with secondary antibodies for 1 h (both 1:10,000, goat anti-rabbit IRDye 800-CW and goat anti-mouse IRDye 680-RD) and washed again. Membranes were again imaged using the Li-Cor Odyssey Fc system to determine protein expression. Total Cx43 was normalized to GAPDH, and pS368-Cx43 was normalized to total Cx43. Additionally, individual baseline, ischemia, and reperfusion values were further normalized to the average of the baseline values of pS368-Cx43/Cx43 or Cx43/GAPDH for each solution to detect relative changes.

Arrhythmia Incidence

Hearts were allowed to beat intrinsically over the duration of the experiment except during optical mapping recordings, at which time points hearts were electrically paced as described above. During ischemia, intrinsic rhythm ceased in all hearts at different time points, which were measured and reported as time to asystole. Some hearts went into arrhythmia upon pacing. Ventricular tachycardia (VT) and ventricular fibrillation (VF), the two observed types of arrhythmias, were defined as an electrocardiographic rhythm with a cycle length of <100 ms that was sustained for 30 s or longer. In such instances, arrhythmia incidence was assigned as 1, and a KCl bolus was applied to reset the heart rhythm. No recordings were taken immediately after the injection of KCl bolus to prevent the effect of increased potassium from confounding the observed CV results.

Statistics

Statistical significance within a group was assessed by ANOVA, individual comparisons within groups were made with Student’s t-tests, and the Bonferroni correction was applied to account for multiple comparisons. Log-rank tests were performed to determine significant differences in the Kaplan-Meier curves, and Fisher’s exact tests were used to test for statistical differences in arrhythmia incidence. All data are reported as means (SD) unless stated otherwise.

RESULTS

Conduction Velocity

In Fig. 1A, representative isochrone maps of optical data are provided at three distinct time points: t = 0 min (last time point of baseline measurements), t = 30 min (after 30 min of ischemia), and t = 50 min (after 20 min of reperfusion) for solutions A and D. Importantly, cardiac conduction maps from the two solutions appear similar during baseline and reperfusion conditions but different during ischemia simply on the basis of perfusate composition. Summary data for CV from all solutions tested are shown in Table 2, and two-way ANOVA with replication (time and solution as factors) revealed no significant differences in CV_T, CV_L, or AR among solutions A–D at baseline. Therefore, for the purposes of visualization and comparison between solutions, CV values in Fig. 1, B–D, were normalized to the average of the respective solution baseline values at t = −15 and 0 min. Note in Table 2 that after 30 min of metabolic ischemia, CV could not be repeatedly measured for all hearts under all conditions. Table 2 and Fig. 1, B–D, reveal that CV slowed at different time points in hearts perfused with solutions A, C, and D, with the relative percentage of CV slowing varying by solution. The response during the time course of metabolic ischemia is complex, and, as a result, the ischemia data are first discussed with respect to the last baseline measurement (t = 0 min) and then, second, compared with solution A, which served as a control for this study. Note that data from hearts perfused with solution A are replicated across all panels in Fig. 1, B−D, to illustrate the statistical comparisons at similar time points.

Fig. 1.

Conduction velocity (CV) during metabolic ischemia and reperfusion. A: representative optical isochrone maps of conduction in hearts perfused with solutions A and D, where solution A slows CV most and solution D slows CV least during simulated ischemia. B–D: percent change from average baseline (−15 and 0 min) for transverse CV (CV_T), longitudinal CV (CV_L), and anisotropic ratio (AR) in hearts perfused with solutions A–D. The yellow shaded region indicates the period of simulated ischemia. Student’s t-tests with Bonferroni correction were applied to determine statistical significance (#P < 0.05 relative to solution A at the same time point). Experimental numbers are shown in Table 2. Soln, solution.

Table 2.

Summary conduction velocity and anisotropic ratios

	Solution A (147 mM Nao and 2.0 mM Cao)		Solution B (155 mM Nao and 2.0 mM Cao)		Solution C (155 mM Nao and 1.63 mM Cao)		Solution D (155 mM Nao and 1.25 mM Cao)		Solution E (155 mM Nao and 1.25 mM Cao + Albumin)		Solution F (155 mM Nao and 2.0 mM Cao + Mannitol)
	Value	n	Value	n	Value	n	Value	n	Value	n	Value	n
Transverse conduction velocity, cm/s
Baseline
−15 min	22 (4)	6	20 (3)	6	20 (3)	6	23 (3)	6	22 (4)	6	20 (6)	8
0 min†	22 (3)	6	23 (5)	6	21 (4)	5	24 (2)	6	31 (5)	6	22 (7)	8
Ischemia
10 min	16 (3)*	6		2	18 (3)	4	19 (1)*	6	18 (3)*	6	16 (6)*	7
20 min	15 (3)*	6		2	16 (3)*	5	19 (1)*	6		0	16 (5)*	8
30 min	15 (3)*	6		2	15 (3)*	5	19 (1)*	6		0	15 (3)*	4
Reperfusion
40 min	23 (3)	3		2	25 (2)	3	24 (2)	3		0	19 (6)	5
50 min	24 (4)	3		2	26 (5)	3	24 (2)	3		0	21 (5)	5
Longitudinal conduction velocity, cm/s
Baseline
−15 min	54 (5)	6	53 (5)	6	55 (5)	6	59 (4)	6	49 (5)	6	48 (5)	8
0 min†	57 (4)	6	57 (9)	6	57 (9)	5	64 (3)	6	62 (3)	6	53 (4)	8
Ischemia
10 min	47 (5)*	6		2	52 (9)	4	54 (5)	6	37 (5)*	6	40 (3)	7
20 min	47 (7)*	6		2	48 (9)*	5	53 (4)*	6		0	39 (5)*	8
30 min	43 (5)*	6		2	46 (9)*	5	52 (7)*	6		0	41 (5)	4
Reperfusion
40 min	55 (9)	3		2	72 (2)	3	64 (2)	3		0	53 (8)	5
50 min	60 (9)	3		2	67 (8)	3	67 (2)	3		0	54 (8)	5
Anisotropic ratio
Baseline
−15 min	2.5 (0.4)	6	2.8 (0.3)	6	2.7 (0.2)	6	2.6 (0.2)	6	2.3 (0.5)	6	2.5 (0.4)	8
0 min†	2.6 (0.4)	6	2.6 (0.3)	6	2.7 (0.2)	5	2.8 (0.2)	6	2.0 (0.3)	6	2.6 (0.6)	8
Ischemia
10 min	3.1 (0.6)	6		2	3.0 (0.2)	4	2.9 (0.3)	6	2.0 (0.2)	6	2.8 (0.6)	7
20 min	3.2 (0.8)	6		2	3.0 (0.3)	5	2.8 (0.2)	6		0	2.6 (0.4)	8
30 min	2.9 (0.4)	6		1	3.0 (0.6)	5	2.8 (0.5)	6		0	2.9 (0.3)	4
Reperfusion
40 min	2.4 (0.4)	3		2	2.9 (0.2)	3	2.7 (0.3)	3		0	3.0 (0.4)	5
50 min	2.6 (0.4)	3		2	2.6 (0.2)	3	2.8 (0.2)	3		0	2.8 (0.6)	5

Values are means (SD); n, number of hearts. Missing data fields indicate that conduction velocity was not used for statistical comparisons because of the occurrence of arrhythmias, lack of optical signal, failure to capture, or hearts removed from the study at 30 min of ischemia for immunohistochemistry and transmission electron microscopy.

^*P < 0.05 by Student’s t-test and Bonferroni correction relative to t = 0 min within each solution.

^†Values at the last baseline measurement as perfusion was switched to metabolic ischemia at t = 0 min.

Solution A: 147 mM Na_o and 2.0 mM Ca_o.

In hearts perfused with solution A, CV_T and CV_L were significantly reduced during metabolic ischemia relative to what will henceforth be referred to as baseline (t = 0 min). Note that although it tended to rise, AR did not significantly change throughout the ischemic period. During reperfusion, CV_T, CV_L, and AR returned to baseline values. See Table 2 and Fig. 1, B–D (black lines).

Solution B: 155 mM Na_o and 2.0 mM Ca_o.

In solution B, we increased Na_o to increase the sodium driving force. Interestingly, this solution was highly arrhythmogenic during metabolic ischemia, with the pacing protocol producing VF in four of six hearts by t = 10 min of ischemia. Therefore, CV_T and CV_L are not reported during metabolic ischemia or reperfusion for solution B because the data lack statistical power. See Table 2 and Fig. 1B (red lines).

Solution C: 155 mM Na_o and 1.63 mM Ca_o.

We then decreased Ca_o to 1.63 mM to create solution C. Interestingly, CV_T and CV_L did not decrease significantly until 20 min of metabolic ischemia. Additionally, the significant CV_T and CV_L slowing at 20- to 30-min ischemia was not associated with a change in AR. Relative to control solution A, CV_T slowing was significantly attenuated only during the early phase of metabolic ischemia [10 min of ischemia vs. baseline CV_T: −16 (6)% vs. −29 (5)% for solutions C and A, respectively, P < 0.05]. By 30 min of metabolic ischemia, CV for solution C slowed to the same extent as solution A [CV_T slowing: −27 (7)% vs. −31 (5)% for solutions C and A, respectively]. Thus, during the early phase of metabolic ischemia, increasing Na_o and decreasing Ca_o with solution C preferentially attenuated CV_T slowing. During reperfusion, CV_T and CV_L returned to baseline values. See Table 2 and Fig. 1C (green lines).

Solution D: 155 mM Na_o and 1.25 mM Ca_o.

We further reduced Ca_o to 1.25 mM to create solution D to determine whether low Ca_o attenuates conduction slowing during metabolic ischemia. Relative to baseline, 10 min of metabolic ischemia significantly slowed CV_T and CV_L in hearts perfused with solution D, without altering AR. Importantly, relative to solution A, CV_T for solution D was elevated throughout the entire 30 min of the ischemic protocol. Upon reperfusion, CV_T and CV_L returned to baseline values. In other words, although hearts perfused with solution D maintained slow CV throughout the ischemic protocol relative to baseline, CV_T did not slow to the same extent observed with the other solutions investigated here, and AR remained unchanged relative to the preischemic period. See Table 2 and Fig. 1D (blue lines).

Summary of solutions A–D.

In summary, in hearts where CV could be quantified, metabolic ischemia slowed CV, but then conduction returned to baseline values during reperfusion. Among solutions, no significant differences were observed during baseline and reperfusion phases. However, during metabolic ischemia, the various combinations of Na_o and Ca_o slowed CV to different extents at distinct time points. Of the solutions compared here, CV slowed most or hearts were arrhythmogenic with solutions containing elevated Ca_o (solutions A and B). On the other hand, reducing Ca_o improved CV during metabolic ischemia. Finally, solution D, with 155 mM Na_o and the lowest Ca_o (1.25 mM), performed best with the least CV slowing observed during metabolic ischemia relative to solution A.

Action Potential Duration

Ischemia has been associated with alterations in the expression and function of several important sarcolemmal ion channels, pumps, and exchangers, and these changes can often be observed electrophysiologically by quantifying APD (43, 63). For example, ischemia is associated with opening of ATP-sensitive potassium channels, which can shorten APD (50). However, previous studies have demonstrated that ischemia can produce either no change in APD, APD shortening (50, 53), or APD prolongation (25, 62, 67) depending on the model of ischemia.

Representative action potentials during baseline and 30 min of metabolic ischemia are shown in Fig. 2A. In summary, APD₃₀ and APD₉₀ quantified from hearts that could be paced were not significantly different when comparing among solutions during the baseline period or after 30 min of metabolic ischemia (Fig. 2B, left). However, evaluating the time course of APD₃₀ during metabolic ischemia revealed that solution C was associated with a decrease in APD₃₀ relative to its own baseline APD₃₀ value after 30 min of metabolic ischemia, resulting in APD triangulation (Fig. 2C, left). Additionally, two solutions significantly altered APD₉₀ acutely during metabolic ischemia. Specifically, in hearts perfused with solutions A and D, APD₉₀ was prolonged at t = 10 min of metabolic ischemia relative to baseline (Fig. 2C, middle) but returned to baseline values as metabolic ischemia progressed.

Fig. 2.

Action potential characteristics. A: representative action potentials (left) and expanded action potential upstrokes (right) from hearts perfused with the study solutions. B: summary of the time interval between activation time and 30% and 90% repolarization (APD₃₀ and APD₉₀, respectively) and rise time (RT) during baseline (t = −15 to 0 min; solid boxes) and metabolic ischemia (t = 30 min; open boxes). During ischemia, APD₃₀ significantly decreased with solution C, and RT significantly increased with solutions A, C, and D relative to baseline (n = 6, *P < 0.05). C: temporal responses of APD₃₀, APD₉₀, and RT from hearts perfused with the study solutions. The yellow shaded region indicates the period of simulated ischemia. Student’s t-tests with Bonferroni correction were applied to determine statistical significance (n = 6, *P < 0.05 relative to baseline, t = 0 min). Hearts in arrhythmias during mapping time points were excluded, and only time points that had at least three measurable APD₉₀ values were included in the analysis. Soln, solution.

Additionally, metabolic ischemia has also been associated with decreased excitability that can manifest as a reduced maximum rate of rise of the action potential (dV/dt_max; 12, 50). Since dV/dt_max cannot be measured with optical mapping, we quantified the action potential RT as a correlate (14). Consistent with previous studies where acidosis/hypoxia/aglycemia reduced dV/dt_max (12, 50), optical RT increased at t = 30 min of metabolic ischemia relative to t = 0 min for solutions A, C, and D (Fig. 2B, right). Interestingly, examining the time course of changes revealed that solutions A, C, and D were associated with significant RT increases at 10, 20, and 30 min of metabolic ischemia (Fig. 2C, right). Increased RT was also associated with slow CV during metabolic ischemia, regardless of solution. No differences were observed among solutions.

Perinexal Width

To demonstrate that altering Ca_o changed W_P, gap junction-adjacent membrane separation within the perinexus was quantified. Representative transmission electron micrographs in Fig. 3A revealed noticeable W_P expansion during metabolic ischemia with solution D (155 mM Na_o and 1.25 mM Ca_o) relative to solution A (147 mM Na_o and 2.0 mM Ca_o). Summary data from baseline (t = 0 min), metabolic ischemia (t = 30 min), and reperfusion (t = 50 min) are shown in Fig. 3B.

Fig. 3.

Perinexal width (W_P) during metabolic ischemia and reperfusion. A: representative electron micrographs of perinexi from hearts perfused with solutions A and D. Solution A demonstrated no change in W_P during metabolic ischemia (t = 30 min) and reperfusion (t = 50 min), whereas solution D-perfused hearts demonstrated W_P widening during metabolic ischemia and returned to baseline values during reperfusion. B: summary of W_P values averaged over a distance of 30–105 nm away from the edge of the gap junction plaque. Student’s t-tests with the Bonferroni correction were applied to determine statistical significance (n = 3 hearts/solution with 15 images/heart, *P < 0.05 relative to baseline and #P < 0.05 relative to solution A for a given condition). Base, baseline; Isch, ischemia; Rep, reperfusion; Soln, solution.

At baseline, W_P was insensitive to perfusate composition. This is consistent with our previous study suggesting that small changes in Ca_o (1.25–2.0 mM) do not measurably alter W_P during baseline conditions (13). Despite the lack of effects of Ca_o on W_P during baseline conditions, these interventions had larger effects on W_P during metabolic ischemia. Specifically, during metabolic ischemia, W_P was wider in hearts perfused with lower Ca_o (solutions C and D) than in those perfused with solution A (solution C: 29.5 ± 4.0 nm and solution D: 28.8 ± 3.7 nm vs. solution A: 21.2 ± 0.6 nm, means ± SE). These effects during metabolic ischemia are on par with W_P modulation induced by much larger changes in Ca_o (1.0–3.4 mM) or treatment with mannitol (143.2 mM) during baseline conditions, as reported in previous studies (13, 16, 60).

Importantly, W_P was dynamically increased during metabolic ischemia when Ca_o was reduced to 1.25 or 1.63 mM, as evidenced by the finding that W_P returned to preischemic values during reperfusion with solutions C and D. Interestingly, the highly arrhythmogenic solution B (155 mM Na_o and 2.0 mM Ca_o) was associated with wider W_P during reperfusion (25.5 ± 2.2 nm, mean ± SE), suggesting that metabolic ischemia with this solution may produce some kind of persistent intercalated disk structural remodeling.

Cx43 Expression

Representative Western blots of total Cx43 and pS368-Cx43 are shown in Fig. 4A along with GAPDH, which was used as a loading control. The ratios of pS368-Cx43 to total Cx43 as well as total Cx43 to GAPDH in tissue perfused with solutions A–D during baseline, metabolic ischemia, and reperfusion are shown in Fig. 4, B and C. No statistically significant changes in total Cx43 and pS368-Cx43 were observed during metabolic ischemia with any solution. Total Cx43 significantly decreased with solution D (155 mM Na_o and 1.25 mM Ca_o) after 20 min of reperfusion, whereas the fraction phosphorylated on S368 remained constant. Total Cx43 and pS368-Cx43 appeared to be insensitive to other combinations of Na_o and Ca_o after reperfusion. Taken together with the data above, there does not appear to be a clear relationship between either total Cx43 or pS368-Cx43 and CV slowing during ischemia or recovery during reperfusion.

Fig. 4.

Connexin (Cx)43 expression and phosphorylation. A: representative Western blots probed for Cx43 phosphorylated at S368 (pCx43-S368), total Cx43, and GAPDH (loading control) are shown during baseline (Base; t = 0 min), metabolic ischemia (Isch; t = 30 min), and reperfusion (Rep; t = 50 min) conditions from hearts perfused with solutions A–D. B and C: summary of pS368-Cx43-to-total Cx43 (B) as well as total Cx43-to-GAPDH (C) ratios normalized to baseline values perfused with solution A. Student’s t-tests with the Bonferroni correction were applied to determine statistical significance (n = 3 hearts/solution, *P < 0.05 relative to baseline). Soln, solution.

Osmotic Modulation of Perinexal Width

Next, to determine whether the beneficial effect of reducing Ca_o was due to its effect on W_P or due to calcium’s many other physiological functions, perinexal expansion/reduction was facilitated by modulating solution osmolarity using mannitol and albumin, respectively, as previously described (59, 60). Briefly, these experiments demonstrated that albumin, which binds to the endothelial glycocalyx and reduces capillary hydraulic conductivity and filtration rate (21, 54, 57a), decreases interstitial volume. Mannitol, presumably acting mainly as a small-molecule osmotic agent (59, 60), increases interstitial volume and W_P.

Albumin at 60.15 mM was added to solution D, resulting in solution E (155 mM Na_o and 1.25 Ca_o + albumin). Similarly, 45.3 mM mannitol was added to solution B, resulting in solution F (155 Na_o and 2 mM Ca_o + mannitol). Metabolic ischemia significantly increased W_P both in hearts perfused with solution E and those perfused with solution F. Importantly, the albumin-containing solution resulted in significantly narrower perinexi (solution E: 22.07 ± 0.77 nm, mean ± SE) relative to the mannitol-containing solution (solution F: 27.4 ± 3.28 nm, mean ± SE, P < 0.05) after 30 min of metabolic ischemia (Fig. 5A). Additionally, total Cx43 and pS368-Cx43 were not significantly different between albumin and mannitol solutions (solutions E and F) during baseline conditions or after 30 min of metabolic ischemia (Fig. 5B).

Fig. 5.

Effects of albumin and mannitol as modulators of extracellular volume. A: summary of perinexal width (W_P) values averaged over a distance of 30–105 nm away from the edge of the gap junction plaque with solutions containing albumin and mannitol. Metabolic ischemia (t = 30 min) increased perinexal (W_P) expansion with solutions containing albumin and mannitol (n = 3 per group, *P < 0.05 relative to baseline), but perinexal expansion was greater with mannitol (n = 3 per group, #P < 0.05, solution F relative to solution E). B: summary of connexin (Cx)43 phosphorylated at S368 (pCx43-S368), total Cx43, and GAPDH demonstrating that 30 min of ischemia did not alter the pCx43-to-Cx43 ratio between solutions, but total Cx43 during ischemia was reduced with mannitol in solution F relative to solution E (n = 3 per group). C: percent change from baseline in transverse conduction velocity (CV_T), longitudinal conduction velocity (CV_L), and anisotropic ratio (AR) in hearts perfused with solution E (n = 6) and solution F (n = 7) demonstrating that 10 min of ischemia homogeneously decreased conduction but that conduction slowing was greatest with albumin relative to mannitol (#P < 0.05 solution F relative to solution E). D: albumin triangulated action potential morphology more than mannitol at 10-min simulated ischemia, as revealed by the time interval between activation time and 30% repolarization (APD₃₀; #P < 0.05), without significantly changing late repolarization at APD at 90% repolarization (APD₉₀), and the osmolytes did not significantly affect action potential rise time (RT). A.U., arbitrary units; Soln, solution.

It is important to note here that albumin rapidly decreases the signal-to-noise ratio of di-4-ANEPPS-transduced optical signals, so electrophysiological data with albumin and mannitol in Fig. 5 are presented after 10 min of metabolic ischemia. The percent changes in CV_L, CV_T, AR, APD₃₀, APD₉₀, and RT were also calculated after 10 min of metabolic ischemia relative to baseline and are shown in Fig. 5, C and D. Both CV_T and CV_L were significantly reduced to a greater extent in albumin-perfused hearts versus mannitol [CV_L: −41.01 (6.02)% vs. −23.30 (2.21)% and CV_T: −41.58 (6.48)% vs. −27.35 (6.57)%] without significantly altering AR. This was also accompanied by significant reduction in APD₃₀ with albumin relative to mannitol [−34.01 (9.85)% vs. 6.58 (11.97)%], but no changes in APD₉₀ or RT were measured between these solutions.

To summarize, whereas solution B (155 mM Na_o and 2 mM Ca_o) was highly arrhythmogenic, addition of mannitol (solution F, 155 mM Na_o and 2 mM Ca_o + mannitol) permitted conduction during metabolic ischemia. Additionally, solution D (155 Na_o and 1.25 Ca_o) was associated with the fastest CV during metabolic ischemia; however, addition of albumin to make solution E (155 mM Na_o and 1.25 mM Ca_o + albumin) resulted in significant CV slowing. Taken together, the data suggest that W_P expansion is beneficial during metabolic ischemia, whereas preventing W_P expansion can result in enhanced CV slowing, potentially leading to arrhythmias, irrespective of whether perinexal expansion during simulated ischemia is secondary to altered calcium or osmotic regulation.

Arrhythmias

Regardless of solution composition, all hearts went into asystole (cessation of intrinsic rhythm) during metabolic ischemia, and the time to asystole was measured. The Kaplan-Meier curves for the different solutions are shown in Fig. 6A. Notably, intrinsic rhythm ceased at approximately the same time with all solutions except for hearts perfused with solutions D and E. Specifically, hearts perfused with the solution containing 155 mM Na_o and 1.25 mM Ca_o (solution D) continued to exhibit intrinsic activity for a slightly longer period relative to solution A [median time to asystole: 8 (3) vs. 5 (2) min, P = 0.08]. Note that this perfusate was associated with improved CV during metabolic ischemia and a wider W_P relative to solution A. In contrast, perfusing hearts with solution D + albumin (solution E) significantly shortened time to asystole [median time to asystole: 3.6 (0.8) min, P < 0.05] relative to solution D. This once again suggests that increasing W_P by either low Ca_o or addition of osmolytes modulates cardiac electrophysiology beneficially during metabolic ischemia.

Fig. 6.

Arrhythmias. A: Kaplan-Meier curves demonstrating the onset of asystole during simulated ischemia in hearts perfused with the study solutions. Log-rank tests were performed to determine significance in Kaplan-Meier curves (n = 6 hearts/solution, P values noted relative to solution A). B: representative ECGs demonstrating normal-paced rhythm during solution A perfusion (top) and ventricular fibrillation (VF) during solution B perfusion (bottom). C: summary of the number of hearts that were in ventricular tachycardia/VF (solid color) after pacing at 10 and 30 min of ischemia. χ²-Analysis with Bonferroni correction was applied to determine statistical significance (n = 6, #P < 0.05 relative to solution A). Soln, solution.

VT and VF were also common during metabolic ischemia upon pacing. Solution A (147 mM Na_o and 2.0 mM Ca_o) never produced VT or VF during metabolic ischemia (Fig. 6, B, top, and C). In contrast, solution B (155 mM Na_o and 2.0 mM Ca_o) was highly arrhythmogenic (Fig. 6, B and C). Note that arrhythmias sometimes occurred after cessation of pacing during metabolic ischemia, explaining why CV could be quantified in Table 2 but the heart was considered arrhythmic at a similar time point.

Summary data in Fig. 6C reveal that the time course of arrhythmia risk was also solution dependent. Reducing Ca_o in solutions C and D progressively reduced the number of arrhythmic hearts at 10 and 30 min of ischemia relative to the most arrhythmogenic solution, solution B. Addition of mannitol (solution F) did not increase the proportion of arrhythmic hearts at 10 and 30 min of metabolic ischemia relative to solution A.

In contrast, the addition of albumin to hearts perfused with reduced Ca_o (solution E) increased the number of arrhythmic hearts at 10 min of ischemia, and this pattern was maintained through the duration of the experiment. It is also worth noting that elevating Na_o increased arrhythmic burden in hearts with the narrowest W_P (solution B compared with solution A).

DISCUSSION

This study demonstrates that during metabolic ischemia, perinexal expansion by reduced Ca_o (1.63 and 1.25 mM) during conditions of elevated Na_o (155 mM) is cardioprotective by attenuating cardiac conduction slowing and reducing arrhythmia vulnerability. On the other hand, increasing Ca_o (2 mM) during elevated Na_o was highly arrhythmogenic. This effect was further supported by osmotically modulating W_P. Specifically, addition of mannitol (perinexal expansion) to a high-Ca_o solution improved conduction and delayed the onset of arrhythmias, whereas addition of albumin (perinexal narrowing) to a low-Ca_o solution slowed conduction and increased arrhythmias. Alternatively, reducing Na_o in high-Ca_o solution also prevented arrhythmias; however, conduction was significantly reduced in this condition. Taken together, the data suggest that reducing determinants of EpC (i.e., perinexal expansion or reducing Na_o) during metabolic ischemia delays the onset of arrhythmias and, depending on the relative concentrations of sodium and calcium and degree of perinexal expansion, can attenuate conduction slowing.

GJC in Ischemia

Ischemia is a complex event that results in alterations of several parameters that can detrimentally affect cardiac conduction. For example, ischemia has been associated with gap junctional uncoupling, elevated K_o, acidosis, hypoxia, and aglycemia, among other factors (50). All of these factors have been shown to affect conduction in the heart and can cause CV to slow and become aberrant, which is a well-established substrate for increased arrhythmogenicity (64). Gap junctional remodeling has been extensively studied during ischemia (7, 9, 10, 27, 37, 49, 52) as gap junctional communication was historically considered to be the sole mechanism of cell-to-cell electrical communication between cardiomyocytes. In favor of this hypothesis, a number of studies have demonstrated that preserving GJC during ischemia can attenuate CV slowing and prevent arrhythmias (10, 18, 27, 35, 65).

Since GJC cannot be directly measured in intact tissue, expression levels as well as phosphorylation status have been used to infer changes in channel number and function. However, the relationship between Cx43 functional expression and cardiac conduction is complex. For example, one study demonstrated that pS368-Cx43 expression in ventricular myocytes only decreases after 8 h of exposure to hypoxia and reduced glucose, and no significant differences were observed at earlier time points (57), consistent with results obtained in the present study. This previous study also demonstrated that hypoxia alone increases pS368-Cx43 expression. On the other hand, other studies have also reported a reduction in pS368-Cx43 induced by acidosis (5, 23, 34). Additionally, dynamic responses in Cx43 expression and phosphorylation during ischemia have also been previously reported. For example, a significant reduction in Cx43 was reported with more aggressive or prolonged ischemia (>2 h) (20, 35). Therefore, several results have emerged that report varying Cx43 expression, phosphorylation patterns, and posttranslational modifications as a result of ischemia (4). The significance of changes in expression and S368 phosphorylation remains unclear, but our data suggest that neither explain our observations with respect to CV and arrhythmogenicity.

Interestingly, our results suggest that the solution most arrhythmogenic during metabolic ischemia (solution B) was not associated with a measurable change in either Cx43 protein expression or S368 phosphorylation after 30 min of metabolic ischemia, whereas a similar lack of Cx43 remodeling was observed with solution D with no VT/VF burden during ischemia and a longer duration of intrinsic electrical activity. Note that statistical differences were not observed between the albumin- and mannitol-containing solutions with respect to Cx43 expression and phosphorylation during simulated ischemia. Importantly, the lack of a relationship between Cx43 remodeling and phenotype may be more attributable to the choice of quantified Cx43 posttranslational modification state rather than pointing to the lack of Cx43 involvement in these results. Alternatively, many other ion channel-gating kinetics may also underlie these results at the relatively acute timescale applied in this study.

Ephaptic Transmission During Ischemia

Our previous studies have supported the theory of an alternative form of electrical coupling between myocytes called EpC, which is dependent on extracellular electric fields in nanodomains such as the perinexus (13, 16, 58, 59). We determined that altering the ionic composition of the perfusate, which we hypothesize can modulate EpC, can exacerbate CV changes in hearts with reduced GJC (15).

Importantly, the frame of “modulating EpC” is complex. On the one hand, we previously demonstrated that decreasing W_P is associated with increased CV, particularly during gap junction uncoupling (16). Yet, we also demonstrated that decreasing W_P can decrease CV when Na_o is reduced, by a mechanism termed “self-attenuation” (13, 28), which is also reported by many theoretical models of EpC (28, 33, 38). Briefly, ephaptic self-attenuation occurs when the driving force for the sodium current is significantly reduced in clefts within the intercalated disk by greatly increasing transmembrane potential and/or reducing the Nernst potential for sodium. The result of self-attenuation is a reduced rate and peak of sodium entry into the downstream cell by an extracellularly modulated mechanism. Therefore, the relationship between Na_o and W_P in this study warrants additional consideration.

We also previously demonstrated that increasing Na_o increases CV in mouse hearts with 50% reduced gap junction and wider W_P (16). In the present study, increasing Na_o and W_P also preserves CV during metabolic ischemia. Specifically, solutions with elevated Na_o were associated with faster CV during metabolic ischemia. If a mechanism of preserved conduction is reduced ephaptic self-attenuation, one might expect conduction to be even faster under conditions producing wider W_P. However, theoretical models of EpC demonstrate a biphasic relationship with intercalated disk width and GJC; therefore, it is difficult to precisely predict what the actual values of W_P, ionic cleft composition, or even GJC are.

Importantly, the data suggest that preventing W_P collapse and modulating ephaptic self-attenuation may be a mechanism for preserving electrophysiology during metabolic challenge. Without the theoretical mechanisms of EpC, we can only say that we observe some electrophysiologic improvements during metabolic ischemia when W_P is increased by lowering Ca_o or adding mannitol and when Na_o is elevated; yet the mechanism for those improvements is not entirely clear.

Alternative Mechanisms

Calcium is a key modulator of a multitude of cellular functions in the heart (66), and it could be argued that the above results are dependent on altered metabolic demand, for example. The finding of the complex, yet subtle changes in APD₉₀ found over time, and with different solutions, does not provide a clear view of what role Ca_o may be playing in this model of ischemia. Yet, the notion that the observed effects are mainly attributable to intracellular calcium homeostasis is not fully supported either. For example, solution B, with the greatest arrhythmic burden, also has the highest Ca_o used in this study, but Na_o is also elevated relative to solution A, which has elevated Ca_o and reduced Na_o. This is important because elevating Na_o decreases inotropy (44, 46, 61) and is theorized to reduce intracellular calcium load (1, 31, 47). Following this line of logic, one would expect that solution A, with reduced Na_o and elevated Ca_o, should produce more spontaneous arrhythmias during ischemia than solution B. Yet, solution A is relatively antiarrhythmic (0 arrhythmias in 6 hearts) compared with solution B (5 of 6 hearts) with similar times to asystole. Furthermore, reduced arrhythmia incidence and measurable CV in hearts with wider W_P, regardless of high or low Ca_o (solution F vs. solution B), further support the crucial role of W_P-modulated CV during ischemia.

It has also been demonstrated that altering Ca_o can affect other determinants of CV, including membrane excitability and GJC (24, 36, 56). In support of such an effect, the rate of action potential rise should be sensitive to both cellular excitability and GJC; however, we provide evidence that physiological changes in ionic composition result in similar prolongation of RT during metabolic ischemia that is indistinguishable among solutions. This could in part be because optical RT is not the same as measuring dV/dt_max (8) and/or the differences are below our detection limit.

Thus, we acknowledge that Ca_o may affect multiple cellular processes during baseline and ischemic conditions, but the data are more consistent with the interpretation that modulating W_P during metabolic ischemia by reducing Ca_o or adding albumin/mannitol can alter CV dependence on Na_o and arrhythmic burden.

Limitations

The data in this study are dependent on an established chemical model of ischemia that we term “metabolic ischemia” (2, 45, 50). Unlike a no-flow ischemia model, which simulates the end point of the ischemic disease, this model allows the study of cardiac structural and functional changes during the more general condition of metabolic stress. Comparing the effects of metabolic ischemia to no-flow ischemia would thus enable us to study the progression of the disease. Furthermore, components of this model are crucial symptoms of other underlying cardiac metabolic disorders.

The results presented here were also obtained during pharmacologic electromechanical uncoupling. It is possible that preserving ATP by electromechanically uncoupling the heart could result in a less severe form of ischemia, which may be why significant APD changes were not measurable with all solutions during ischemia; future studies need to be conducted to determine whether the effects are consistent in a working heart.

Optical mapping cannot measure absolute membrane voltage at present, and therefore conclusions about precise changes in action potential morphology and upstroke velocity are descriptive and best understood in relationship to time. Future studies using techniques such as microelectrode recordings may determine whether the inability to measure certain differences reported above was due to the quantification techniques limited to optical mapping and may also reveal more nuanced changes in parameters such as resting membrane potential.

The measurement of W_P is based on manual segmentation of a two-dimensional section of a three-dimensional structure obtained from chemically fixed tissues, but at present other three-dimensional electron microscopy methods lack the spatial resolution (7–20 nm/pixel) (32) necessary to estimate intercellular separation on the order of 10–20 nm. Therefore, the absolute values of W_P should not be interpreted as physiological values but rather the statistically significant changes should be viewed as directionally relevant effects. For example, reducing Ca_o or perfusing with mannitol expanded the perinexus during metabolic ischemia, consistent with our previous studies (16, 60).

Conclusions

Extracellular ionic composition and perinexal clefts are very important determinants of the conduction response during metabolic ischemia in the ventricular myocardium. Solutions that theoretically reduce EpC may be cardioprotective during metabolic ischemia. Further studies are needed to determine whether optimizing extracellular ionic composition is a possible therapy for conduction-mediated arrhythmias in different models of ischemia.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute (NHLBI) Grants R01-HL-102298 and R01-HL-138003 (to S. Poelzing), R01-HL-132236 (to J. W. Smyth), R01-HL-56728 (to R. G. Gourdie), and R01-HL-141855-01 (to R. G. Gourdie and S. Poelzing); an American Heart Association predoctoral fellowship (to P. J. Calhoun); a Virginia Tech Carilion Research Institute Medical Research Scholar award, an American Heart Association predoctoral fellowship, and the David W. Francis and Lillian Francis Scholarship Fund (to S. A. George); NHLBI Grant F31-HL-140873 (to T. B. Raisch); and Danish Heart Foundation Grant 16-R107-A6812-22015 (to M. S. Nielsen).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

S.A.G., J.W.S., M.S.N., and S.P. conceived and designed research; S.A.G., G.H., P.J.C., M.E., T.B.R., D.R.K., M.K., C.B., M.S.N., and S.P. performed experiments; S.A.G., G.H., P.J.C., M.E., T.B.R., D.R.K., M.K., C.B., M.S.N., and S.P. analyzed data; S.A.G., G.H., P.J.C., M.E., T.B.R., D.R.K., M.K., R.G.G., J.W.S., M.S.N., and S.P. interpreted results of experiments; S.A.G., T.B.R., D.R.K., M.K., J.W.S., M.S.N., and S.P. prepared figures; S.A.G., G.H., R.G.G., J.W.S., M.S.N., and S.P. drafted manuscript; S.A.G., G.H., P.J.C., M.E., T.B.R., D.R.K., M.K., C.B., R.G.G., J.W.S., M.S.N., and S.P. edited and revised manuscript; S.A.G., G.H., P.J.C., M.E., T.B.R., D.R.K., M.K., C.B., R.G.G., J.W.S., M.S.N., and S.P. approved final version of manuscript.