Ionic bases for electrical remodeling of the canine cardiac ventricle

Abstract

Emerging evidence suggests that ventricular electrical remodeling (VER) is triggered by regional myocardial strain via mechanoelectrical feedback mechanisms; however, the ionic mechanisms underlying strain-induced VER are poorly understood. To determine its ionic basis, VER induced by altered electrical activation in dogs undergoing left ventricular pacing (n = 6) were compared with unpaced controls (n = 4). Action potential (AP) durations (APDs), ionic currents, and Ca²⁺ transients were measured from canine epicardial myocytes isolated from early-activated (low strain) and late-activated (high strain) left ventricular regions. VER in the early-activated region was characterized by minimal APD prolongation, but marked attenuation of the AP phase 1 notch attributed to reduced transient outward K⁺ current. In contrast, VER in the late-activated region was characterized by significant APD prolongation. Despite marked APD prolongation, there was surprisingly minimal change in ion channel densities but a twofold increase in diastolic Ca²⁺. Computer simulations demonstrated that changes in sarcolemmal ion channel density could only account for attenuation of the AP notch observed in the early-activated region but failed to account for APD remodeling in the late-activated region. Furthermore, these simulations identified that cytosolic Ca²⁺ accounted for APD prolongation in the late-activated region by enhancing forward-mode Na⁺/Ca²⁺ exchanger activity, corroborated by increased Na⁺/Ca²⁺ exchanger protein expression. Finally, assessment of skinned fibers after VER identified altered myofilament Ca²⁺ sensitivity in late-activated regions to be associated with increased diastolic levels of Ca²⁺. In conclusion, we identified two distinct ionic mechanisms that underlie VER: 1) strain-independent changes in early-activated regions due to remodeling of sarcolemmal ion channels with no changes in Ca²⁺ handling and 2) a novel and unexpected mechanism for strain-induced VER in late-activated regions in the canine arising from remodeling of sarcomeric Ca²⁺ handling rather than sarcolemmal ion channels.

ventricular electrical remodeling (VER) is a persistent change in the electrophysiological properties of the myocardium in response to a change in the pattern of ventricular electrical activation. Numerous human (22, 32, 33) and animal (6, 14) studies have shown that an alteration of ventricular activation by pacing produces persistent and marked ECG T-wave changes termed “T-wave memory,” providing clear evidence that the electrophysiological properties of the human ventricular myocardium exhibit the plasticity needed to induce VER. Altered electrical activation is a common sequelae of a variety of cardiac pathologies, including conduction system disease, ischemia, hypertrophy, and heart failure (30). Similarly, alteration of electrical activation by cardiac pacing reduces the mechanical efficiency of contraction and increases mortality (34). In contrast, restoration of synchronized electrical activation by biventricular pacing improves mechanical function and reduces mortality (4). However, the ionic mechanisms that govern VER remain poorly understood.

Interestingly, unlike atrial electrical remodeling, VER is characterized by prolongation rather than shortening of the action potential (AP) duration (APD) (14, 25). Another characteristic feature of VER is attenuation of the AP phase 1 notch due to reduced expression of transient outward K⁺ current (I_to) and its α (Kv4.3)-/β (KChIP2)-subunits (21). Using transmural optical imaging from multiple left ventricular (LV) regions, we identified two distinct types of AP remodeling induced by altered activation (14). Interestingly, the type of AP remodeling observed was dependent on whether myocytes were located in regions proximal to (early activated) or remote from (late activated) the site of pacing. Specifically, the early-activated region (i.e., region adjacent to the site of altered activation) exhibited minimal changes in APD. In contrast, the most significant APD prolongation occurred in the late-activated region (i.e., region farthest from the site of altered activation), which accounted for the repolarization changes underlying VER, including T-wave memory (14). Notably, marked APD remodeling in the late-activated region was attributed to focally increased mechanical strain (14), whereas remodeling of the AP shape but, to a lesser extent, APD in early-activated regions occurred without increased strain. A similar response was later reported in the dyssynchronous pacing model of heart failure, which exhibits marked AP prolongation in the late-activated lateral LV (1).

Despite its implications to cardiac function in health and disease, the ionic basis for both strain-dependent and strain-independent long-term VER remains unclear. Based on the aforementioned discussion, a key to understanding the mechanisms for VER is ascertaining the distinct ionic bases for AP remodeling in early-activated (low strain) versus late-activated (high strain) ventricular regions. These mechanisms were investigated in the present study by comparing ion channel densities and Ca²⁺ transients between regions of the myocardium isolated from early- and late-activated regions.

METHODS

Experimental model of VER.

The Institutional Animal Care and Use Committee of Case Western Reserve University (Cleveland, OH) approved all animal protocols used in the present study. A previously validated canine model of VER induced by anterior LV epicardial pacing for 4 wk (n = 6) was compared with unpaced controls (n = 4) (14). Briefly, adult male mongrel dogs were anesthetized with propofol (10 mg/kg), intubated, ventilated, and maintained on inhaled isoflurane. After a lateral thoracotomy, a unipolar lead was implanted on the right atrium and on the epicardial surface of the LV (n = 6). Atrial and ventricular leads were connected to a pulse generator (Discovery II, Guidant, Minneapolis, MN), which was implanted subcutaneously. We used a model of VER induced by a change in ventricular activation independent of the heart rate [VDD mode pacing, i.e., atrial sensing was used to trigger ventricular pacing using a short atrioventricular (AV) delay] for 4 wk. This resulted in 99% ventricular paced beats at the animal's intrinsic sinus rate. The pacemaker was temporarily switched off (5- to 10-min intervals) at a weekly interval to monitor the development of VER as evidenced by T-wave inversion on the surface ECG. As previously reported (14), VER was documented by a 127 ± 35° (P = 0.009) change in the ECG T-wave vector after 28 ± 3 days of ventricular pacing. After the successful induction of VER, hearts were harvested, and basal LV myocardial regions were analyzed proximal to the site of pacing (early activated) and distal to site of pacing (late activated).

Isolated myocyte experiments of APs and ionic currents.

Epicardial myocytes were used for further experiments because we (14) have previously demonstrated that VER induced APD remodeling to a comparable degree in myocytes that reside across all layers of the transmural wall (i.e., the epicardium, midmyocardium, and endocardium). Canine epicardial myocytes were isolated using a standard enzymatic dispersion technique (16). Myocytes were resuspended in medium 199 and used within 12 h of isolation. The amphotericin perforated-patch technique was used to obtain AP recordings at physiological temperature (31). Briefly, myocytes were bathed in a chamber continuously perfused with Tyrode solution of the following composition (in mmol/l): 137 NaCl, 5.4 KCl, 2.0 CaCl₂, 1.0 MgSO₄, 10 glucose, and 10 HEPES (pH 7.35). Patch pipettes (0.9–1.5 MΩ) were filled with electrode solution of the following composition (in mmol/l): 120 aspartic acid, 20 KCl, 10 NaCl, 2 MgCl₂, and 5 HEPES (pH 7.3) supplemented with 240 μg/ml amphotericin B (A-4888, Sigma, St. Louis, MO). After gigaseal formation, amphotericin lowered the access resistance within 10 min sufficiently to perform current-clamp recordings. The conventional whole cell mode was used for all other current recordings. Inward rectifying K⁺ currents (I_K1) were elicited with hyperpolarizing voltage steps to −100 mV from a holding potential of −40 mV. Rapidly activated delayed rectifying K⁺ currents (I_Kr) were activated with voltage steps to 60 mV for 750 ms (holding potential: −40 mV) and isolated as E4031-sensitive tail current components upon the return to −40 mV. Slowly activated delayed rectifying K⁺ currents (I_Ks) were elicited with voltage steps to 60 mV for 2.5 s (holding potential: −40 mV), and tail currents were measured on the return to −40 mV in the presence of 5 μM E4031 to block I_Kr. L-type Ca²⁺ currents (I_Ca) were blocked with 1 μM nisoldipine. Na⁺/Ca²⁺ exchanger (NCX) current (I_NCX)-voltage relationships were determined by applying ramp pulses (change in voltage over time: 680 mV/s, holding potential: −40 mV). In these experiments, K⁺ channels, Ca²⁺ channels, and Na⁺-K⁺ pump currents were blocked with tetraethylammonium (TEA), Cs⁺, Ba²⁺, nisoldipine, and ouabain. I_NCX was defined as the 5 mM Ni²⁺-sensitive current recorded under these experimental conditions. The experimental temperature for the above currents was 36–37°C. I_to was recorded in Tyrode solution with 200 μM Cd²⁺ at room temperature. Currents were elicited with depolarizing voltage steps of 500 ms from −50 to +60 mV after a 500-ms prepulse to −140 ms (holding potential: −40 mV). I_Ca was blocked with 1 μM nisoldipine, and I_to amplitudes were quantified at +50 mV to minimize contamination by cardiac Na⁺ currents. Cardiac I_Ca was recorded at room temperature using extracellular solution of the following composition (in mM) 137 NaCl, 5.4 CsCl, 1.8 MgCl₂, 1.8 CaCl₂, 10 glucose, and 10 HEPES (pH 7.4). The intracellular solution contained (in mM) 130 CsMeSO₄, 20 TEA Cl, 1 MgCl₂, 10 EGTA, 10 HEPES, 4 Mg-ATP, 14 Tris-phosphocreatine, and 0.3 Tris-GTP with 50 U/ml creatine phosphokinase (pH 7.2). I_Ca was elicited using 300-ms depolarizing voltage steps from −30 to +60 mV (holding potential: −40 mV). For quantitative comparisons, current amplitudes were measured at +10 mV. All current amplitudes were normalized to cell capacitance and converted into current densities.

Ca²⁺ transient measurements.

Ratiometric intracellular Ca²⁺ transients were measured using the fluorescent Ca²⁺ indicator indo-1 AM and pluronic F-127 as previously described (13). Cells were loaded with indo-1 AM by incubating them in Tyrode solution containing indo-1 AM (2 μM, Molecular Probes) and 0.025% (wt/wt) pluronic F-127 (Molecular Probes) for 20 min at room temperature. Intracellular indo-1 was excited at 355 nm, and fluorescence emitted at 405 and 485 nm was collected by two matched photomultiplier tubes. The emission field was restricted to a single cell with the aid of an adjustable window. Data were filtered at 200 Hz and sampled at 1 kHz. Ca²⁺ concentrations ([Ca²⁺])were determined using the following equation: [Ca²⁺] = K_d × b × [(R − R_min)/(R_max − R)], where K_d is the dissociation constant of indo-1 AM (equal to 250 nM), b is (intensity at a wavelength of 485 nm at R_min)/(intensity at a wavelength of 405 nm at R_max), R is the measured ratio, R_max is the ratio when indo-1 AM is saturated with Ca²⁺ (10 mM), and R_min is the ratio with no free Ca²⁺ (containing 10 mM EGTA) present. Background fluorescence recorded from a control cell without loaded indicator was subtracted from the test signal before the fluorescence ratio was calculated. Diastolic Ca²⁺ was defined as the cytosolic Ca²⁺ level just before the onset of a Ca²⁺ transient or just before the AP upstroke. Amplitudes of intracellular Ca²⁺ transients were calculated from the difference between peak and diastolic Ca²⁺ levels. The rate of reuptake of intracellular Ca²⁺ was measured by fitting the decay portion of the Ca²⁺ transient (from 30% to 100% of the decline phase) to a single-exponential function.

Computational modeling.

A model of the canine epicardial myocyte (7) with recent updates (8) was used to study the effects of ion channel and Ca²⁺ transient remodeling on the AP. Simulations to examine the effects of ion channel/Ca²⁺ transient remodeling were conducted by the incorporation of experimentally measured changes into the model. The experimentally observed percent change in ion channel densities (Table 1) was simulated relative to control (I_to: control 100%, VER early-activated 48% reduction and no changes in the late-activated region; I_Ks: control 100%, VER early-activated 38% reduction and late-activated 37% reduction; and I_Ca: control 100%, VER early-activated 19% reduction and no changes in the late-activated region). Simulated steady-state APs and ionic currents were obtained after 2,400 s of pacing at a cycle length of 600 ms. To examine the effect of Ca²⁺ transient remodeling on APD, digitized and scaled (by 0.55) Ca²⁺ transient recordings from control and VER myocytes were used to clamp model intracellular [Ca²⁺] for a single beat in the presence or absence of ionic current remodeling.

Table 1.

Summary data of AP duration, ionic currents, and Ca²⁺ transient parameters from the early- and late-activated regions of control and VER hearts

	Early-Activated Region			Late-Activated Region
Parameter	Control	VER	P value	Control	VER	P value
AP remodeling
AP duration, ms	271 ± 7 (7)	270 ± 20 (6)	NS	248 ± 22 (6)	325 ± 13 (6)	0.015
Ionic current remodeling
Ito, pA/pF	11.36 ± 0.1 (8)	5.88 ± 0.7 (12)	0.001	11.53 ± 0.5 (9)	9.95 ± 0.9 (15)	NS
ICa, pA/pF	−2.62 ± 0.2 (19)	−2.11 ± 0.2 (11)	NS	−1.79 ± 0.2 (14)	−1.86 ± 0.1 (13)	NS
IKs, pA/pF	0.69 ± 0.09 (10)	0.43 ± 0.08 (12)	0.03	0.67 ± 0.09 (10)	0.42 ± 0.09 (11)	0.06
IKr, pA/pF	0.26 ± 0.07 (6)	0.28 ± 0.07 (8)	NS	0.22 ± 0.03 (6)	0.25 ± 0.04 (9)	NS
INCX, pA/pF	0.24 ± 0.09 (4)	0.22 ± 0.13 (3)	NS	0.2 ± 0.12 (4)	0.25 ± 0.16 (3)	NS
IK1, pA/pF	−19.7 ± 3 (4)	−17.4 ± 5 (6)	NS	−16.5 ± 3 (8)	−17.2 ± 2 (5)	NS
Ca2+ transient remodeling
Diastolic Ca2+, nM	372 ± 39 (7)	454 ± 30 (8)	NS	369 ± 31 (6)	677 ± 97 (6)	0.02
Amplitude, nM	153 ± 19 (7)	186 ± 55 (8)	NS	121 ± 24 (6)	633 ± 228 (6)	0.002
Time constant, ms	272 ± 22 (7)	262 ± 16 (8)	NS	253 ± 42 (6)	326 ± 25 (6)	NS

Data are means ± SE; n = 2 independent control and remodeled animals, with parantheses showing the number of cells recorded. AP, action potential; VER, ventricular electrical remodeling; I_to, transient outward K⁺ current; I_Ca, L-type Ca²⁺ current; I_Ks, slowly activated delayed rectifying K⁺ current; I_Kr, rapidly activated delayed rectifying K⁺ current; I_NCX, Na⁺/Ca²⁺ exchanger current; I_K1, inward rectifying K⁺ current; NS, not significant.

Real-time PCR quantification of mRNA.

Myocardial regions from the early- and late-activated epicardium (1–2 mm) were dissected using a tissue dissection block, washed in PBS, and flash frozen in liquid nitrogen. The frozen tissue was homogenized in TRIzol (Invitrogen, Carlsbad, CA), and RNA was extracted following the manufacturer's instructions. RNA was reverse transcribed using Moloney murine leukemia virus reverse transcriptase (New England Biolabs, Ipswich, MA), and Taqman probe-based real-time PCR assays (ABI) were performed for KCND3 (Cf 02623099) and KCNIP2 (Cf 02624497) on an ABI StepOnePlus real-time PCR machine. Results were normalized to 18S rRNA (Hs 99999901), and fold changes were calculated using the 2^−ΔΔC_t method (where C_t is cycle threshold).

Western immunoblot analysis.

Western blots were performed on membrane preparations made from rapidly frozen tissues obtained from myocardial regions of the late-activated epicardium and controls. Briefly, tissues were homogenized in 5 volumes of 0.3 M sucrose and 10 mM sodium phosphate (pH 7.4) in the presence of protease and phosphatase inhibitors. Samples were centrifuged at 2,800 g for 10 min to pellet nuclei and debris. The supernatant was collected and centrifuged at 8,800 g for 10 min to pellet mitochondria. The supernatant was collected and centrifuged at 20,000 g for 60 min to pellet the membranes. The membrane pellet was then solubilized 10 min on ice in 1% Triton X-100 lysis buffer with protease and phosphatase inhibitors.

All samples were run on 10% bis-Tris precast gels (Bio-Rad, Hercules, CA). The following primary antibodies were used: ryanodine receptor (RyR; MA3-916, Affinity BioReagents), sarcoplasmic reticulum Ca2+-ATPase 2a (SERCA2a; ab2861, Abcam), phospholamban (GTX22865, Gene Tex), and NCX (R3F1, Swant). Relative band densities were quantified using ImageQuant software (Molecular Dynamics). To normalize for protein loading, we used β-actin (A4700, Sigma).

Apparatus and experimental protocols for in vitro muscle experiments.

Solution compositions for mechanical experiments were calculated using a computer program (11) and known stability constants (12) corrected to pH 7.0 and 22°C. All solutions contained (in mM) 100 N,N-bis-(2 hydroxy-ethyl)-2-aminoethanesulfonic acid, 15 creatine phosphate, 5 DTT, 1 free Mg²⁺, and 4 MgATP. pCa 9.0 solution contained 7 mM EGTA and 0.02 mM CaCl₂, pCa 4.5 solution contained 7 mM EGTA and 7.01 mM CaCl₂, and preactivating solution contained 0.07 mM EGTA. The ionic strength of all solutions was adjusted to 180 mM with potassium propionate. Solutions containing different amounts of free [Ca²⁺] ([Ca²⁺]_free) were prepared by mixing the appropriate volumes of pCa 9.0 and 4.5 solutions.

On the day of the experiment, myocardial tissue from control and VER regions were isolated for preparation of the multicellular myocardium for mechanical experiments (27). Myocardial preparations were skinned in a solution containing Triton X-100 for 30 min, and the ends of the preparations were then attached to the arms of a position motor and force transducer as previously described (27). Motor position and force signals were sampled using SL Control software (11) and saved to computer files for later analysis.

Force-pCa analysis.

Each myocardial preparation was set to a sarcomere length of 2.25 μm before activation and allowed to develop steady force in solutions of varying [Ca²⁺]_free. The difference between steady-state force and the force baseline obtained after the 20% slack step was measured as the total force at that [Ca²⁺]_free. Active force was then calculated by subtracting Ca²⁺-independent force in solution of pCa 9.0 from the total force and was normalized to the cross-sectional area of the preparation, which was calculated from the width of the preparations assuming a cylindrical cross section. Force-pCa relationships were constructed by expressing submaximal force (P) at each pCa as a fraction of maximal force (P_o) determined at pCa 4.5, i.e., P/P_o. The apparent cooperativity in the activation of force development was inferred from the steepness of the force-pCa relationship and was quantified using a Hill plot transformation of the force-pCa data (27). Force-pCa data were fit using the following equation: P/P_o = [Ca²⁺]^n_H/(k^n_H + [Ca²⁺]^n_H), where n_H is the Hill coefficient and k is the [Ca²⁺] required for half-maximal activation (i.e., pCa₅₀).

Myofilament gel electrophoresis.

Myofilament protein preparation from frozen heart tissue was performed as previously described (15). Profiling of total and phosphorylated myofilament proteins by gel electrophoresis and phosphoprotein-specific staining were performed as previously described (39). Briefly, purified myofilament proteins were separated by one-dimensional SDS-PAGE using 4–12% gradient gels (Invitrogen). Gels were fixed in a 10% acetic acid and 10% methanol solution. Phosphorylated proteins were detected by Pro-Q Diamond staining (Invitrogen) according to the manufacturer's instructions. Pro-Q Diamond-stained gels were imaged with a Typhoon scanner (GE Healthcare). Subsequently, gels were stained with Coomassie blue (Bio-Rad) to reveal total proteins.

Statistical analysis.

Data are plotted as means ± SE. The nonparametric Wilcoxon rank-sum test was used to assess significance between groups. P values of <0.05 were considered significant.

RESULTS

AP changes underlying VER in early- versus late-activated regions.

Figure 1 shows the two distinct types of AP remodeling observed at early- versus late-activated regions after a 4-wk period of pacing-induced altered ventricular activation. In the early-activated region, there was attenuation of the AP phase 1 notch with minimal changes in APD [control: 271 ± 7 ms vs. VER: 270 ± 20 ms, P = not significant (NS)]. In contrast, VER in the late-activated region was characterized by marked APD prolongation (325 ± 13 ms) compared with control (248 ± 22 ms, P < 0.01). These data are consistent with our previous report (14) and reaffirm the diverse manifestation of VER determined by anatomic proximity of the myocardium to the origin of altered activation.

Fig. 1.

Action potential (AP) changes after ventricular electrical remodeling (VER). A: representative APs from early- and late-activated VER regions compared with unpaced controls. B: representative transient outward K⁺ current (I_to) recordings from VER and control hearts.

Ionic changes underlying VER in early-activated regions.

I_to and its molecular determinants (Kv4.3/KChIP2) have been previously suggested to be critical to the development of cardiac memory (21). Therefore, to examine if I_to changes could underlie the AP changes observed in VER (Fig. 1A), we assessed I_to in early- and late-activated VER regions. In agreement with a previous report (38), the AP phase 1 notch amplitude in the early-activated region was significantly reduced (control vs. early activated regions: 55.7 ± 5.8 vs. 30.3 ± 3.3 mV, P = 0.008). Consistent with this observation, I_to density was reduced by ∼50% in the early-activated region (Fig. 1B and Table 1) and not in the late-activated region (Fig. 1, A and B). To confirm the focal remodeling of I_to, we also assessed mRNA expression levels of the α (Kv4.3)- and β (KChIP2)-subunits. There was an ∼5-fold reduction in Kv4.3 (control: 1.03 ± 0.14 vs. VER: 0.27 ± 0.08, P = 0.02) and a >10-fold reduction in KChIP2 (control: 1.02 ± 0.12 vs. VER: 0.09 ± 0.02, P = 0.02); however, expression levels of the transcripts were unchanged compared with controls in the late-activated region.

Next, to determine if changes in other ionic currents could account for the AP remodeling, we measured I_Kr, I_Ks, I_K1, I_Ca, and I_NCX from control and VER hearts. Table 1 shows ion channel densities measured from early- and late-activated regions of control (unpaced) and VER hearts. After VER, in the early-activated region, in addition to changes in I_to, there was a 19% reduction in I_Ca that did not reach significance and a 38% reduction in I_Ks (Table 1). No significant changes in other ionic currents were observed (Table 1). To examine if the measured ionic current densities could account for the AP remodeling, we incorporated the ionic current changes in a computational model of the canine AP. In the early-activated region, consistent with our experimentally measured attenuation of the AP phase 1 notch, we observed a simulated epicardial AP with a reduced phase 1 notch amplitude (Fig. 2). Therefore, the measured ionic current changes, and I_to in particular, could account for the AP morphological remodeling observed in the early-activated region.

Fig. 2.

Ionic remodeling underlies the AP remodeling in the early-activated region. Computational simulations were based on measured ionic current changes in the canine epicardial AP model. I_Ca, L-type Ca²⁺ current; I_Ks, rapidly activated delayed rectifying K⁺ current.

Ionic changes underlying VER in late-activated regions.

In contrast to the early-activated region, the AP phase 1 notch was not significantly different from controls in the late-activated region (control vs. late-activated regions: 70.4 ± 4.5 vs. 51.4 ± 13.1 mV, P = NS). Furthermore, the 37% reduction in I_Ks was the only ionic change in the late-activated region that was close to being significant (Table 1). Incorporating this change of I_Ks into the canine epicardial AP model failed to recapitulate the measured AP prolongation observed in the late-activated region (see Fig. 4, left). This suggested that sarcolemmal ionic current changes could not account for the AP remodeling in the late-activated region.

Fig. 4.

Ca²⁺ transient remodeling underlies the AP remodeling in late-activated VER regions. Measured ionic and Ca²⁺ transients from late-activated regions were incorporated into the canine epicardial AP model. The incorporation of ionic changes only (left) had minimal effects on the AP. In contrast, simulation of the Ca²⁺ transient changes (middle) caused AP prolongation due to a marked increase in inward Na⁺/Ca²⁺ exchanger (NCX) current (I_NCX) despite a compensatory increase in I_Ks. Right: exaggerated AP duration (APD) prolongation when both ionic and Ca²⁺ transient changes were simulated. Ca_i, intracellular Ca²⁺.

We (14) have previously reported an important mechanistic role for mechanical strain in AP remodeling in the late-activated region. Since mechanoelectrical coupling is mediated through changes in intracellular [Ca²⁺], we reasoned that Ca²⁺ transient changes could account for the AP remodeling in VER. To measure Ca²⁺ transient changes independent of the AP remodeling, we measured Ca²⁺ transients during an identical AP clamp from control and VER myocytes (Fig. 3). In these experiments, we did not observe significant changes in the Ca²⁺ transient amplitude or diastolic levels of Ca²⁺ in the early-activated region. In sharp contrast, there were profound changes in Ca²⁺ transients from the late-activated region (Fig. 3). Specifically, there were significantly elevated levels of diastolic Ca²⁺ and increased amplitude of the Ca²⁺ transient (data shown in Table 1). Although a trend for a slower Ca²⁺ reuptake was observed, this did not reach statistical significance (Table 1).

Fig. 3.

Ca²⁺ handling is altered in late-activated regions but not in early-activated regions. Late-activated regions exhibited greater Ca²⁺ transient amplitudes and increased diastolic levels of Ca²⁺. No significant changes were observed in Ca²⁺ transients from identical control regions (gray traces, early-activated regions; black traces, late-activated regions).

To examine whether remodeling of Ca²⁺ transients could account for the AP remodeling, the experimentally measured Ca²⁺ transient changes were incorporated in the canine computational AP model. Interestingly, simulation of the Ca²⁺ transient remodeling (Fig. 4, middle) caused ∼30% APD prolongation, which was congruent with the measured APD prolongation of ∼25% in the late-activated region (Fig. 1A). This APD prolongation was due to enhanced forward-mode I_NCX as changes in other ionic currents (I_Ca, I_to, or I_Ks) could not account for the APD prolongation (Fig. 4, middle). When both measured ionic (reduced I_Ks) and Ca²⁺ transient changes were incorporated in the canine epicardial AP model (Fig. 4, right), APD was further prolonged by ∼35%. To examine if altered Ca²⁺ handling in the late-activated region arises from altered expression of Ca²⁺-handling proteins, we measured the expression of several components of the sarcoplasmic reticulum Ca²⁺-handling machinery. As shown in Fig. 5, there were no significant changes in the expression of RyR, SERCA, or phospholamban. However, a modest increase in NCX expression was noted in the late-activated region (Fig. 5).

Fig. 5.

Ca²⁺-handling proteins after VER in late-activated regions. A: representative Western immunoblots of Ca²⁺-handling proteins from late-activated regions of control and VER hearts. B: quantitative analysis of Ca²⁺-handling proteins using densitometry (n = 4/group). RyR, ryanodine receptor; SERCA2a, sarco(endo)plasmic reticulum Ca²⁺-ATPase 2a.

Finally, since there were minimal alterations in the expression of Ca²⁺-handling proteins, we examined if changes in myofilament Ca²⁺ sensitivity were evident after remodeling in the late-activated region. To examine this, we measured myofilament Ca²⁺ sensitivity from skinned fibers at a fixed sarcomere length by measuring force development at various extracellular [Ca²⁺]. Interestingly, the force-pCA relationship was shifted to the left in late-activated VER regions, consistent with an increase in the sensitivity of myofilaments to Ca²⁺ (Fig. 6). Since these changes could be either a direct response to stretch or due to an alteration in the phosphorylation status of myofilaments, we measured gross changes in myofilament phosphorylation using Pro-Q Diamond stain. Interestingly, we did not observe significant changes in myofilament phosphorylation, which suggests a primary change in myofilament Ca²⁺ sensitivity (Fig. 7). In summary, these data support the notion that changes in myofilament Ca²⁺ may trigger changes in intracellular Ca²⁺ transients that underlie AP remodeling in the late-activated ventricular region.

Fig. 6.

Increased myofilament Ca²⁺ sensitivity in late-activated VER regions. A: skinned myocardium isolated from paced, late-activated segments displayed increased Ca²⁺ sensitivity of force compared with skinned myocardium isolated from control segments (n = 4 control segments and 4 VER segments). Forces measured at submaximal free Ca²⁺ concentration ([Ca²⁺]_free) were expressed relative to the maximal force obtained at pCa 4.5. The smooth lines were fit using the following Hill equation: P/P_o = [Ca²⁺]^n_H/(k^n_H + [Ca²⁺]^n_H)], where P is the force measured at submaximal [Ca²⁺]_free, P_o is the force measured at maximal [Ca²⁺]_free (pCa 4.5), n_H is the Hill coefficient, and k is the [Ca²⁺] required for half-maximal activation (i.e., pCa₅₀). B: summary data of steady-state mechanical properties of fibers isolated from control and VER hearts. F_min, minimal force; F_max, maximal force.

Fig. 7.

Myofilament phosphorylation is not altered in VER. Left: Coomassie blue staining of myofilament proteins. Right: phosphoprotein using Pro-Q Diamond stain. CON, control; MHC, myosin heavy chain; MyBP-C, myosin-binding protein-C; TnT, troponin T; Tm, tropomyosin; TnI, troponin I; MLC1 and MLC2, myosin light chain 1 and 2, respectively.

DISCUSSION

Altered cardiac electrical activation from heart disease or pacing is an important but poorly understood marker for morbidity and mortality. In a previous study (14), we identified regionally heterogenous AP remodeling after altered activation. Specifically, the most significant AP remodeling occurred in myocardial regions that were far from site of pacing where mechanical strain was greatest (14). However, the cellular and ionic mechanisms underlying this form of mechanoelectrical remodeling of the ventricular myocardium were heretofore unknown. In the present investigation, we performed detailed electrophysiological assessment in conjunction with computer simulations in a clinically relevant animal model. Using this approach, we identitied two distinct remodeling responses that occurred after altered activation, both providing insights into mechanisms that underlie regionally heterogeneous remodeling of myocardial repolarization. In summary, ionic currents play a central role in AP remodeling in the early-activated, low-strain region, whereas remodeling in the late-activated, high-strain region was attributable to alterations in myocyte Ca²⁺ handling.

Mechanism of VER in the early-activated region.

We found that AP remodeling in the early-activated region was characterized by attenuation of the epicardial phase 1 notch, as previously reported (Fig. 1A) (9, 38). This was due to a localized reduction in I_to and its molecular components in this region (Fig. 1B) (19, 38). These findings are consistent with a previous study (18) in both long and short-term models of cardiac memory. Two key observations from previous studies have led to the hypothesis that changes in I_to are central to the development of cardiac memory. First, in isolated canine myocardial preparations, pharmacological blockade of I_to inhibits the accumulation of memory (9). Second, neonatal dogs with reduced expression of I_to subjected to ventricular pacing for 2 h have a diminished accumulation of memory (20). In the present study, we extend prior observations by demonstrating that remodeling of I_to and its molecular components are limited to the early-activated region (Fig. 1B). Hence, VER in response to altered electrical activation is not homogeneous throughout the ventricle but rather involves several distinct, but consistent, electrophysiological responses.

In addition to remodeling of I_to, there were modest reductions in I_Ca and I_Ks in the early-activated region with no significant changes in Ca²⁺ transients (Table 1). The reduction in I_Ca in the early-activated region could arise from a reduction in KChIP2, the β-subunit for I_to, which has recently been identified to interact with the L-type Ca²⁺ channel and enhance its activity (29). In contrast to changes in I_to and I_Ca, which were limited to the early-activated region, changes in I_Ks were common to both early and late-activated regions. Thus, I_Ks changes may reflect a broader mechanism operative in VER. In summary, remodeling of sarcolemmal ionic currents in the early-activated region account for the AP morphological remodeling that occurs after altered activation.

Mechanism of VER in the late-activated region.

AP remodeling in the late-activated region was most notably characterized by marked APD prolongation (Fig. 1A). In contrast to the early-activated region characterized by ion channel remodeling, the electrogenic driving force for AP remodeling in the late-activated region was enhanced forward-mode I_NCX (Fig. 4) due to changes in the diastolic levels of Ca²⁺ and Ca²⁺ transient amplitudes. The enhanced protein expression of I_NCX in the late-activated region also provides support for this observation (Fig. 5). Increased I_NCX is a well-described compensatory response during mechanical remodeling of the heart (2). A common mechanism is thought to arise from the increased forward-mode I_NCX in a compensatory attempt to counteract the Ca²⁺ overload, thereby generating an inward depolarizing current causing AP prolongation. For example, the chronic AV-block dog develops significant AP remodeling secondary to increased intracellular Ca²⁺ and enhanced NCX expression (26). Similarly, the dyssynchronous model of canine heart failure (LBBB with rapid pacing at 190–200 beats/min) also exhibits significant AP prolongation and increased NCX expression in the late-activated lateral LV (1). In contrast to the AV-block dog and our model, the dyssynchronous heart failure model develops reduced Ca²⁺ transient amplitudes resembling the Ca²⁺ transients observed in the failing myocardium. Interestingly, in contrast to both models with significant structural remodeling, i.e., hypertrophy in the AV-block canine model and heart failure in dyssynchrony model, we (14) previously reported that the physiological pacing model used in our study at 1 mo exhibits no overt evidence of structural remodeling. Finally, a recent transmural optical imaging study (17) in human heart failure reported heterogenous transmural remodeling of Ca²⁺ transients as well as the expression of SERCA2a. It would be of interest to examine the relative contribution of transmural versus regional changes in AP and Ca²⁺ transient remodeling in modulating the electrophysiological changes in various disease states. In summary, our present study identifies remodeling of myocyte Ca²⁺ handling in late-activated, high-strain regions to play a central role in driving the AP remodeling after long-term altered activation.

The degree of Ca²⁺ remodeling in our study, in particular the diastolic levels of Ca²⁺ that are higher than systolic levels of controls, is of concern as it could cause contracture. Interestingly, prior studies under physiological conditions (24) and pathological conditions such as heart failure (36) have reported similar changes in the diastolic levels of Ca²⁺. Furthermore, all measurements in the present study were from viable myocytes with normal resting membrane potentials with no overt signs of cell shortening. Despite marked changes in myocyte Ca²⁺ handling, we were surprised to find minimal changes in sarcomeric Ca²⁺-handling proteins in the late-activated region. In search for alternate mechanisms that could link mechanical strain to myocyte Ca²⁺ handling (5), we explored myofilament Ca²⁺ sensitivity, a well-described adaptive response to changes in sarcomere length, which underlies the Frank Starling response (3). Thus, the enhanced myofilament Ca²⁺ sensitivity in late-activated regions observed in our study is reflective of an increased affinity for Ca²⁺ to troponin C (Fig. 6). Interestingly, a recent murine model of troponin mutation with increased myofilament Ca²⁺ sensitivity also exhibited increased diastolic levels of Ca²⁺ (23). Myofilament Ca²⁺ sensitivity has been reported to demonstrate similar changes in a canine model of heart failure (35). Furthermore, the absence of changes in myofilament phosphorylation changes (Fig. 7) are suggestive of primary remodeling in myofilament Ca²⁺ sensitivity. Future studies are needed to examine if myofilament Ca²⁺ sensitivity is a cause of cardiac electrical remodeling or a consequence of the AP/Ca²⁺ transient remodeling. Finally, indepth genomic and proteomic changes in the early- and late-activated regions are likely to provide insights into the upstream signaling and transcriptional mechanisms that underlie the ionic basis for cardiac electrical remodeling. These experiments may eventually pave the way for the development of novel therapies for the prevention and treatment of cardiac electrical remodeling.

Limitations.

In the present study, we focused on analyses of repolarization, as prior clinical and experimental models of remodeling or T-wave memory have demonstrated minimal changes in conduction as evidenced by QRS changes on the surface ECG (14). Next, we principally studied early- and late-activated basal epicardial LV regions to obtain insights into the role of mechanical strain in cardiac electrophysiology. However, remodeling of important gradients such as transmural, apical to basal, or interventricular gradients was not evaluated in the present study. Thus, future studies that analyze remodeling in a more global fashion are likely to provide a better perspective on regional VER. Furthermore, we did not conduct detailed mechanical activation and strain mapping in the present study, but this has been previously reported (14). With regard to major ionic currents, we studied all major cardiac membrane currents with only two exceptions (Table 1): the Ca²⁺-dependent component of the I_to (I_to2) and the late Na⁺ current. Possible interactions of the sympathetic system in modulating I_Ks were not evaluated in the present study. Furthermore, we measured whole cell changes in [Ca²⁺] but did not measure Ca²⁺ changes in other myocyte microdomains, such as the sarcoplasmic reticulum or mitochondria. Since Ca²⁺ transient changes were only observed in late-activated regions, our Ca²⁺-handling protein and myofilament analyses were limited to these regions. In the present study, we have proposed altered myofilament Ca²⁺ sensitivity as one plausible mechanism in causing the accumulation of cytosolic Ca²⁺. However, we do not provide evidence showing that the changes in myofilament Ca²⁺ sensitivity could fully account for the Ca²⁺ transient changes or changes in the AP observed in the late-activated region. Other possible mechanisms, such as ANG II-mediated signaling, activation of stretch-activated ion channels, or posttranslational modifications of proteins involved in sarcomeric Ca²⁺ handling, were not analyzed and are potential avenues for future investigation. Future studies that examine these mechanisms may provide insights into the novel mechanisms mediating stretch-mediated VER.

Clinical implications.

In the present study, we report two distinct forms of VER that are highly dependent on the pattern of mechanical strain during altered activation. Remodeling in the early-activated, low-strain region is largely due to sarcolemmal ion channel changes altering AP morphology. In contrast, remodeling in the late-activated, high-strain region is due to increased levels of cytosolic Ca²⁺ causing AP prolongation by enhancing depolarizing I_NCX. One important implication to such distinctly differing electrophysiological responses is that it explains how altered electrical activation can induce electrophysiological heterogeneities. For example, selective APD prolongation distal but not proximal to the site of pacing amplifies repolarization gradients in the heart (14). This may explain, in part, why clinical evidence of aberrant activation (10, 28, 37) from heart disease is independently associated with cardiovascular mortality. Furthermore, our study also provides the first insights into how alterations of myofilament Ca²⁺ sensitivity may affect cytosolic levels of Ca²⁺, thus serving as an interface between changes in mechanics and cardiac electrophysiology. Consequently, future experimental and modeling studies should incorporate effects of mechanical stretch as an important dimension to the electrophysiological function of the heart.

GRANTS

This work was supported by an American Heart Association postdoctoral fellowship award (to D. Jeyaraj) and by National Heart, Lung, and Blood Institute Grants K08-HL-094660 (to D. Jeyaraj), RO1-HL-071789 (to E. Ficker), R01-HL-049054 (to Y. Rudy), and RO1-HL-054807 (to D. S. Rosenbaum).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: D.J. and D.S.R. conception and design of research; D.J., X.W., E.F., J.E.S., H.L., L.D.W., K.F.D., T.H.S., and D.S.R. performed experiments; D.J., X.W., E.F., J.E.S., I.D., H.L., K.F.D., and D.S.R. analyzed data; D.J., X.W., E.F., J.E.S., L.D.W., K.F.D., M.K.J., Y.R., and D.S.R. interpreted results of experiments; D.J., E.F., J.E.S., K.F.D., and D.S.R. prepared figures; D.J. and D.S.R. drafted manuscript; D.J., E.F., and D.S.R. edited and revised manuscript; D.J., X.W., E.F., J.E.S., I.D., H.L., L.D.W., K.F.D., T.H.S., M.K.J., and Y.R., approved final version of manuscript.