Lower troponin expression in the right ventricle of rats explains interventricular differences in E–C coupling

The right ventricle (RV) propels blood into the pulmonary circulation, where the hydraulic impedance is significantly lower than that of the systemic circulation. Consequently, despite the same amount of cardiac output from the two ventricles, the thickness of the RV free wall is lower than that of the left ventricle (LV). As a whole, the RV uses approximately one-fifth of the energy of the LV (Friedberg and Redington, 2014; Foschi et al., 2017). In terms of the work of cardiac cycle, the pressure–volume relationship (P-V loop) of the RV is described as a triangular or trapezoidal shape, with a short isometric period (Redington et al., 1988), different from the square shape in the LV.

The contractile properties of isolated cardiomyocytes from LV (LVCM) and RV (RVCM) have rarely been investigated, while demonstrating variable results in sarcomere length (SL) changes depending on the animal species and research conditions. A study of mice showed significantly smaller shortening of SL (ΔSL) in RVCM than in endocardial LVCM (Kondo et al., 2006). In rats, the analysis of SL changes revealed the tendency of slower kinetics in RVCM (Carneiro-Júnior et al., 2013). In contrast, canine RVCM showed no significant difference from LVCM in basal SL and ΔSL (Carneiro-Júnior et al., 2013; Molina et al., 2014).

Previous studies of the interventricular differences in the rodent cardiac action potential duration (APD) showed shorter APD in RVCM than LVCM (Watanabe et al., 1983; Knollmann et al., 2001). In addition, canine RVCM showed a deeper notch of phase I in their action potential (AP) than LVCM (Di Diego et al., 1996). Consistent with the shorter APD, the peak amplitudes of cytosolic Ca²⁺ concentration changes triggered by a single AP (change in intracellular Ca²⁺ concentration [Δ[Ca²⁺]_i] or Ca²⁺ transient) were lower in the RVCM than in the endocardial LVCM of mice (Kondo et al., 2006). However, in rat cardiomyocytes, the Δ[Ca²⁺]_i was not different between the RV and LV (Sathish et al., 2006; Carneiro-Júnior et al., 2013).

In addition to the upstroke amplitude, the decay kinetics of the Ca²⁺ transient determine the contractile properties of cardiomyocytes. In an early study of rat RVCM, slower decay of Ca²⁺ transient was observed along with the experimental evidence suggesting a slower rate of Ca²⁺ sequestration by the sarcoplasmic reticulum Ca²⁺-ATPase (SERCA). However, in more recent studies of rodent hearts, the decay of Ca²⁺ transient in RVCM was not different from that in LVCM (Carneiro-Júnior et al., 2013; Molina et al., 2014). The controversial results from previous studies suggested that another Ca²⁺ buffering component, such as Ca²⁺ binding troponin C (TnC), might be different between RVCM and LVCM.

In cardiac myocytes, Ca²⁺ regulates contraction by binding to the thin filament regulatory protein TnC. Along with TnC, the troponin complex (cTn) comprises TnI, which prevents the interaction between myosin and actin filaments, and TnT, which links the TnI–TnC complex to tropomyosin in the thin filament. Ca²⁺ binding with TnC causes a conformational shift in TnI and TnT, allowing myosin to bind to actin (Sweeney and Hammers, 2018). In addition to the regulation of contraction, Ca²⁺-binding TnC could also contribute to the cytosolic Ca²⁺ buffering capacity (κ_S) of cardiomyocytes. However, the comparison of myofilament expression between RV and LV cardiomyocytes is very rare, let alone the expression of TnC. A recent study of isolated myosin molecules suggested that the mechanical characteristics of working myosin (kinetic rates and the distribution of spatial orientation of myosin lever arm) were the same in both ventricles (Duggal et al., 2017). However, there have been few reports comparing the expression of myofilament components between the two ventricles except 2-D electrophoresis data from porcine and human heart tissues (Phillips et al., 2011; Su et al., 2015).

In this study, we aimed to elucidate the functional differences in excitation–contraction (E–C) coupling along with the expression of myofilament proteins between RVCM and LVCM from rats, focusing on (1) the AP shapes and relevant ionic currents, (2) AP-triggered Ca²⁺ transients with contraction–relaxation kinetics, (3) the expression of Ca²⁺-binding TnC and other cTn proteins, and (4) the Ca²⁺ buffering capacity and Ca²⁺ sequestration rate by the SR. The results demonstrate intriguing discordant properties between the levels of E–C coupling (APD, Δ[Ca²⁺]_i, and ΔSL) in terms of biventricular differences. The immunoblot assay revealed lower expression of cTn; in addition, rigorous in situ analysis revealed a lower κ_S, consistent with the lower TnC in RVCM. The SERCA activity was found to be lower in RVCM, consistent with the slower decay rate of the Ca²⁺ transient. The novel findings regarding cTn and cytosolic κ_S in RVCM could explain the biventricular gaps between the APD, Δ[Ca²⁺]_i amplitude, and ΔSL, which were successfully simulated by using a computational model of rodent cardiomyocytes.

Ventricular cardiomyocytes from Sprague-Dawley rats (RRID:RGD_1566457) were used. The study protocol followed the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (Publication No. 85-23, revised 1996) and also conformed to the Institutional Animal Care and Use Committee of Seoul National University (approval no. SNU-160128-1). 46 male rats were used. At the time of the experiments, the rats were 12 wk old and had mean (± SD) body weight of 348.4 ± 4.68 g.

Cardiomyocytes were isolated using a standard enzymatic dispersion technique, as described previously (Jin et al., 2012), with minor modifications. Briefly, rats were anesthetized with a mixture of ketamine (90 mg ⋅ kg⁻¹, i.p.) and xylazine (10 mg ⋅ kg⁻¹, i.p.). After monitoring the anesthesia depth with a toe pinch, the hearts were extracted and rapidly mounted onto the Langendorff apparatus at 37°C. Cells were dissociated by a protocol consisting of a constant-flow perfusion (8–10 ml ⋅ min⁻¹) for (1) 8 min with Ca²⁺-free Tyrode solution; (2) 8 min with Ca²⁺-free Tyrode solution containing enzyme (collagenase 1 mg ⋅ ml⁻¹ [Worthington Biochemical Co.], protease, 0.133 mg ⋅ ml⁻¹, BSA 1.65 mg ⋅ ml⁻¹, and Ca²⁺ 0.05 mM); and (3) a further 10-min digestion period with collagenase-containing Tyrode’s solution. In this study, the septum between the two ventricles was not used. Cardiomyocytes were separately harvested from both ventricles and resuspended in cardioplegic storage solution (for IonOptix, in mM; 120 NaCl, 5.4 KCl, 10 HEPES, 5 MgSO₄, 5 Na-pyruvate, 5.5 glucose, 20 taurine, 29 mannitol, and 0.2 CaCl₂, pH 7.4) or Kraft-Brühe (K-B) solution (for electrophysiology, in mM; 50 L-glutamate, 70 KOH, 55 KCl, 10 HEPES, 0.5 EGTA, 20 KH₂PO₄, 20 taurine, 20 glucose, and 3 MgCl₂, pH 7.3). The proportion of rod-shaped cardiomyocytes ranged from 60 to 80%. Cardiomyocytes for physiological assays were used immediately, and cardiomyocytes for biochemical assays were centrifuged and stored at −80°C.

Freshly isolated cardiomyocytes were transferred into a bath mounted on the stage of an inverted microscope (Ti; Nikon). The bath (0.15 ml) was continuously perfused at a constant rate (5 ml ⋅ min⁻¹), and voltage-clamp experiments were performed at room temperature (22–25°C). Borosilicated glass pipettes with a free-tip resistance of ∼2.5 MΩ were used for whole-cell and perforated whole-cell patch clamp. The series resistance, estimated by dividing the time constants of the capacitive current, was kept <10 MΩ in the whole-cell configuration and <16 MΩ in the perforated whole-cell configuration. To correct for cell size, the absolute current amplitudes were divided by the cell capacitance and expressed as pA ⋅ pF⁻¹. The pipettes were connected to the CV 203BU head stage of the Axopatch 200B, a patch-clamp amplifier (Molecular Devices). pCLAMP software v10.6.2 and Digidata-1440A (Molecular Devices) were used to acquire data and apply command pulses. The recorded currents were sampled at 10 kHz and low-pass Bessel filtered at 5 kHz. The recorded data were processed using Clampfit v11.0.1 (Molecular Devices).

The physiological properties of cardiomyocytes were measured using the IonOptix system (IonOptix). The contractility was measured by detecting the length of two edges with a contractility recording system at a 2–6-Hz frequency of field stimulation. Soft-edge software (IonOptix) was used to capture and analyze the changes in SL, and the results were given as fractional shortening. For [Ca²⁺]_i measurement, cardiomyocytes were incubated with 2 µM fura-2-acetoxymethyl ester (fura-2-AM) for 15 min at room temperature. After sedimentation, the supernatant was removed, and cardiomyocytes were reintroduced into a perfusion solution containing 500 µM Ca²⁺ for 10 min. Then, the fura-2-AM–loaded cardiomyocytes were placed in a bath and excited by 360- and 380-nm filtered fluorescent light. The emitted signal was measured simultaneously with photomultiplier tubes. [Ca²⁺]_i was calculated using the in situ calibration described below and the estimated dissociation constant (K_d) of fura-2-AM (K_d,fura-2). Measurements from ≥20 steady-state contractions were averaged for each cardiomyocyte for each stage of the experimental protocol. All experiments were performed in a temperature-controlled bath solution at 37 ± 1°C.

The sampling for Western blotting (immunoblotting) was performed in two different ways, as described previously (Jang et al., 2015). The membrane/cytosol fraction was extracted from cardiomyocytes using lysis buffer containing 0.5 mM EGTA, 25 mM Tris-HCl, 150 mM NaCl, and 1% Triton X-100 with phosphatase and protease inhibitor cocktail (Roche) following centrifugation. To obtain the myofilament fraction, cardiomyocytes were washed with cold PBS and transferred to screw-cap tubes containing 2.5 mm zirconia/silica beads and cold buffer (60 mM KCl, 30 mM imidazole, 2 mM MgCl₂, and 1% Triton X-100 with protease and phosphatase inhibitor cocktail). The cells in the lysis buffer were homogenized with a bead beater (Mini-Beadbeater-8; BioSpec Products) for 20 s, and the homogenates were centrifuged at 8,000 rpm for 10 min at 4°C. The supernatant was discarded, and the process was repeated at least six times. The remaining pellets (myofilament-enriched fraction) were resuspended in the high-salt buffer (Bio-Rad) and used in subsequent experiments.

Lysates were quantified using the Bradford method (Bio-Rad) and heated at 95°C for 5 min. Lysates were fractionated by SDS-PAGE and transferred to PVDF membranes (Bio-Rad) in 25 mM Tris, 192 mM glycine, 0.01% SDS, and 20% methanol. Membranes were blocked in 1× TBS containing 1% Tween-20 and 5% BSA (blocking solution) for 1 h at room temperature with gentle rocking. Membranes were then incubated overnight at 4°C with TnC (RRID:AB_306435; Abcam), TnI (RRID:AB_2206278; Cell Signaling), TnT (RRID:AB_261723; Sigma-Aldrich), tropomyosin (RRID:AB_261817; Sigma-Aldrich), desmin (RRID:AB_306653; Abcam), actinin (RRID:AB_476766; Sigma-Aldrich), and α-actin (RRID:AB_476695; Sigma-Aldrich) primary antibodies, SERCA2a (Abcam), phospholamban (RRID:AB_2617049; Badrilla), and phosphor-PLB (RRID:AB_310352; Millipore) followed by secondary antibodies after washing. Blots were developed using ECL Plus Western blotting detection reagents (Merck). The relative densities were calculated by normalizing each blot to actin. The number of experiments means each different animal. We excluded the results that showed large differences in internal control (10%) between LV and RV protein samples.

For these experiments, a whole-cell patch clamp was conducted with simultaneous fura-2 fluorescence ratiometry (PTI). The Cs-pipette solution contained 50 µM fura-2 pentapotassium salt (Table 1, Cs-pipette). Light from a xenon arc was collected and filtered at 340 and 380 nm with 10 nm full width at half-maximum. The filtered light of the region of interest surrounding the cardiomyocyte was passed through the epifluorescence port of the inverted microscope and the CFI Super Fluor objective lens (Nikon). The emitted light was filtered at 510 nm and measured simultaneously using a photomultiplier tube (PTI).

To estimate the endogenous cytosolic buffer components, we defined the endogenous (intrinsic) Ca²⁺ binding ratio (κ_S) in terms of a hypothetical Ca²⁺ binding species “S” (Mathias et al., 1990). The κ_S and Ca²⁺-binding ratio of the total buffer including fura-2 (κ_B) were defined as follows:

The properties of fura-2, such as its K_d,fura-2 value when associated with Ca²⁺, are known to change in the cytosolic environment due to its binding with proteins (Konishi et al., 1988). Therefore, we conducted an in vitro calibration according to the method described by Berlin et al. (1994). The free Ca²⁺ concentrations of the calibration solutions were calculated with Slider, v2.0 (written by Chris Patton, Pacific Grove, CA), which used the equation from Fabiato and Fabiato (1979). To obtain the K_d,fura-2 value in the cytosolic environment, we used Ca²⁺-buffered (10 mM EGTA) KCl solution containing 283 mM sucrose or 20 mg/ml aldolase to reflect the viscosity and the fura-2–binding property of cytosolic proteins, respectively (Ca²⁺-calibration solution in Table 1). The viscosity–protein and –sucrose relationships were measured using an Ubbelohde-type viscometer under the assumption of a Newtonian fluid (Fig. S1 A). To minimize the error factor from the depth of the bath solution, a calibration solution constrained between two round coverslips (diameter 18 mm) was used (Fig. S1 B). From the amount of the calibration solution and the size of coverslips, we estimated the distance between the coverslips to be 120–140 µm, which was ∼10 times the average thickness of cardiomyocytes, 13 µm (Satoh et al., 1996). To obtain a fluorescence light intensity similar that of fura-2 loaded into cardiomyocytes, the concentration of fura-2 in the in vitro calibration was lowered to 5 µM. We then measured K_d,fura-2 in three types of calibration solutions (Fig. S1 C): (1) standard calibration solution (normal), (2) viscosity-corrected calibration (sucrose), and (3) protein-containing calibration solution (protein). Because the K_d,fura-2 values of the standard and viscosity-corrected calibration solutions were significantly lower than those of the protein-containing solutions (Fig. S1 D), we used the K_d,fura-2 value of the protein-containing calibration, in accordance with previous measurement of Berlin et al. (1994). In this study, the in situ K_d,fura-2 of fura-2 was 381.8 nM.

It is well known that there are several problems in calibrating Ca²⁺ concentration in cardiomyocytes, including hypercontraction of cells at [Ca²⁺]_i >1 µM, which distorts the path length and membrane integrity (Cheung et al., 1989). To overcome this problem, we measured the calibration parameters, R_min, R_max, and β, in separate experiments according to the lanthanum-method described by Borzak et al. (1990). Conversion factors were measured using a cuvette-type spectrophotometer with an identical right source and calculated as C₁ = R_max,Ca/R_max,La and C₂ = F_380max,Ca/F_380max,La (Fig. S2). These La³⁺-derived proportionality constants were determined to convert the fluorescence signal of La³⁺ to that of Ca²⁺.

Solutions for electrophysiology, Ca²⁺ measurement, and the IonOptix system are listed in Table 1. We used three external solutions: NT, 0 Na/0 Ca, and 0 Na/2 Ca. Four types of pipette solutions were used depending on the experimental conditions (AP-, VOCC_L-, K-, and Cs-pipette). All drugs and chemicals were obtained from Sigma-Aldrich.

For analysis of the activation and inactivation kinetics of the current of VOCC_L (I_Ca,L) and transient outward K⁺ current (I_to), the conductance was normalized to the maximal conductance and fitted by a Boltzmann equation. The time constants of inactivation, τ, were obtained by fitting the current trace with a dual exponential function. Data are presented as mean ± SD, with n denoting the number of cells and N denoting the number of animals. Data were compared using a nested t test was where appropriate, taking into account n cells and N animals (Sikkel et al., 2017; Eisner, 2021). Current amplitudes of I_to and I_Ca,L were assessed using MANOVA in Origin software (MANOVA app, v1.22). Differences were considered significant when P < 0.05. Asterisks were used in the graphs to indicate significance of statistical inference (*, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001). Curve fitting with the least-squares method was performed using Origin (Microcal Software) or hand-written Python software.

Fig. S1 shows the methods of in vitro calibration for measuring K_d,fura-2. Fig. S2 shows the spectra of fura-2-Ca²⁺ and fura-2-La³⁺ and converting factors between two measurements. Fig. S3 shows the diagram of troponin states used in this study. Fig. S4 shows the AP analysis for each animal. Fig. S5 shows the inactivation time constants of I_to and I_ca,L. Fig. S6 shows the Ca²⁺ transient analysis for each animal. Fig. S7 shows the immunoblotting of cTn and tropomyosin while varying the total loading protein amount. Fig. S8 shows the sarcomere shortening analysis for each animal. Fig. S9 shows the changes in sarcomere shortening pattern according to the stimulation frequency. The box-and-whisker plots in Figs. S4, S6, and S8 display the five-number summary of a set of data including the minimum, first quartile, median, third quartile, and maximum.

The APs of RVCM and LVCM were triggered by a 6-ms square pulse (∼1.0 nA of current injection, 2 Hz) under the nystatin-perforated mode of current-clamp (Fig. 1 A). In separate experiments under the conventional whole-cell mode, electrical membrane capacitances (C_m) were measured, which showed a larger C_m in LVCM than in RVCM (Fig. 1 B). All the parameters analyzed in the AP recordings were obtained in the nystatin-perforated condition (Fig. 1, C–M). Peak amplitudes, time to reach peak depolarization (peak time), resting membrane potential (RMP), maximum increasing speed, and time of the maximum increasing speed were not different between LVCM and RVCM (Fig. 1, C–G). In contrast, the repolarization was more rapid in RVCM; and higher maximum decreasing speed (Fig. 1 H); shorter time of maximum decreasing speed from the stimulus (MdT) and from the peak time (MdT from peak; Fig. 1, I and J), and shorter APD at 10, 50, and 90% of repolarization (APD₁₀, APD₅₀, APD₉₀) were seen in RVCM than in LVCM (Fig. 1, K and M). Six and five Sprague-Dawley rats were used for measuring capacitance and AP, respectively. The statistical significance between RVCM and LVCM was analyzed by nested t test (Fig. S4).

Because the faster repolarization suggested larger voltage-dependent K⁺ conductance, we measured the K⁺ currents activated by depolarizing step pulses from −90 mV holding voltage under the conventional whole-cell voltage-clamp. A VOCC_L blocker (nifedipine, 10 µM) was added to the bath solution. Fast activating transient outward currents (I_to) and sustained outward currents were observed (Fig. 2 A). The amplitudes of I_to—that is, the inactivating component of outward current—were normalized to C_m and plotted against the test voltages (I/V curves; Fig. 2 B). I_to was larger in RVCM than LVCM (131 ± 5.8% at 60 mV, P < 0.0001). Voltage-dependent inactivation of I_to was analyzed using a two-step voltage-clamp protocol. A common test pulse to 30 mV (200 ms) was applied at the end of the conditioning step (1 s) to different potentials (inset of Fig. 2 C). Inactivation was determined as the ratio of the test current amplitude to the maximum test current. The voltage dependence of inactivation was fitted by a Boltzmann function, providing a half-inactivation voltage (V_1/2,inact) and slope factor (k). The V_1/2,inact and k values were not significantly different between RVCM and LVCM (Fig. 2 C). The kinetics of the inactivating components were further analyzed by fitting to double exponential functions. The time constants for the fast and slow components did not differ between RVCM and LVCM (Fig. S5, A and B).

The current of VOCC_L (I_Ca,L) generates prolonged depolarization of the cardiac AP, which determines the APD; therefore, we measured the I_Ca,L using Cs-pipette solution under the nystatin-perforated patch clamp condition to minimize the decay of I_Ca,L. Steplike depolarizations from −40 mV holding voltage were applied. Voltage-dependent activation of inward currents with relatively slow inactivation was observed (Fig. 2 D). The peak amplitudes normalized to C_m were plotted against the test voltages, and the I/V curves did not differ between RVCM and LVCM (Fig. 2 E). Steady-state voltage-dependent inactivation was analyzed using a two-step voltage clamp protocol. The pulses were applied to a test potential of 10 mV (300 ms) from various conditioning pulses (−80 mV to 20 mV, 300 ms; inset of Fig. 2 F). The voltage dependence and kinetics of I_Ca,L inactivation did not differ between RVCM and LVCM (Fig. 2 F and Fig. S5, C and D).

Using a mathematical electrophysiology model of rat ventricular cardiomyocytes (Pandit et al., 2001), we simulated the effects of the higher I_to in RVCM (Fig. 2, dotted lines). The Pandit model simulating the cardiac AP of the “left” ventricular AP was modified according to the experimental data of the RV. As the measured parameters representing the kinetics of I_to did not differ from those of the Pandit model, we modified only the conductance of I_to to reflect the measured values. It was also notable that the I_Ca,L of the Pandit model, 7.8 ± 0.2 pA ⋅ pF⁻¹ at 10 mV (Katsube et al., 1998), was smaller than our experimental data, 9.0 ± 0.78 pA ⋅ pF⁻¹ at 10 mV (Fig. 2, D and E). Therefore, to reflect this difference, we modified the maximum conductance for I_Ca,L (g_CaL in the original model) in our RVCM and LVCM models. The calculated APD of rat RVCM was shorter than that of LVCM (Fig. 2 G), similar to the experimental data (Fig. 1 A and Table 3).

The amount of net Ca²⁺ influx would also be limited by the shorter APD inducing faster termination of I_Ca,L. However, the peak amplitude of I_Ca,L during the actual AP might not be smaller or could even be larger in RVCM with larger I_to, owing to the electrical driving force for the Ca²⁺ influx through the not-yet-deactivated VOCC_L (Harris et al., 2005). Consistent with the latter possibility, it has been reported that a loss of the notch due to reduction of I_to contributes to the reduced Ca²⁺ influx in failing hearts of larger mammals (Cooper et al., 2010). To elucidate the actual Ca²⁺ influx via VOCC_L, we compared the I_Ca,L induced by the AP-clamp protocol in RVCM and LVCM. In fact, the peak amplitude of inward I_Ca,L became higher, and the decay became faster, when applying the RVCM-type versus the LVCM-type AP clamp. Overall, the cumulative Ca²⁺ influx—that is, mathematical integration of the inward current per membrane capacitance (∫I_Ca,Ldt ⋅ C_m⁻¹)—was smaller when applying the RVCM-type AP (Fig. 2 H).

The Ca²⁺ transient of intact cardiomyocytes triggered by repetitive electrical field stimulation (2 Hz) was observed in cells loaded with fura-2-AM (Fig. 3 A, representative traces). Summarized results from LVCM and RVCM are shown as bar graphs (Fig. 3, B–G). The [Ca²⁺]_i during the diastolic period did not differ between the RVCM and LVCM (Fig. 3 B). Considering the shorter APD and the smaller Ca²⁺ influx via VOCC_L (Figs. 1 and 2), it was anticipated that Δ[Ca²⁺]_i would be smaller in RVCM. However, the peak [Ca²⁺]_i and Δ[Ca²⁺]_i were not different from those of LVCM, indicating a mismatch between AP and Ca²⁺ transient (Fig. 3, C and D). Furthermore, despite the shorter APD, the decay of the Ca²⁺ transient was slower in the RVCM than in the LVCM (Fig. 3, E–G). 14 animals were used for measuring Ca²⁺ transient, and the four parameters analyzed in each animal are depicted in Fig. S6.

In rat ventricular cardiomyocytes, SERCA is mainly responsible for the decay of the Ca²⁺ transient (Bassani et al., 1994). Therefore, we analyzed SERCA activity in RVCM using whole-cell patch clamp combined with fura-2 ratiometry. Under the current-clamp configuration, AP-triggering current injections (∼1.0 nA, 1 Hz) were applied to confirm that the steady-state Ca²⁺ transient reflected the responses of the Ca²⁺-loaded state of the SR (Fig. 4 A, stimulation). Three Ca²⁺ removal mechanisms, SERCA, NCX, and PMCA, eliminated intracellular Ca²⁺ into SR and extracellular space in this condition, and the time constant τ_tran reflected the sum of the rates of Ca²⁺ removal by the three mechanisms (Fig. 4 B, stimulation). After stopping stimulation, we immediately applied 20 mM of caffeine to induce Ca²⁺ transient (Fig. 4 A, caffeine). In this condition, where SERCA activity was nullified, NCX and PMCA were involved in the decay of caffeine-induced calcium transient, which could be fitted to a single exponential function (τ_caff; Fig. 4 B, caffeine). As the SR was depleted by the previous caffeine application, the cardiomyocytes were then restimulated by current injection until steady state was achieved (Fig. 4 A, second stimulation). The cardiomyocytes were then superfused with 0 Na/0 Ca solution to inhibit the NCX, and the second application of caffeine induced the Ca²⁺ transient with a slow decay speed (τ_0Na0Ca, Fig. 4 A). As the SERCA and NCX were inhibited by caffeine and 0 Na/0 Ca bath solution, the decay kinetics of Ca²⁺ transient would be mainly affected by PMCA (Fig. 4 B, right panel). The decay of the Ca²⁺ transient of this experiment could be fitted to single exponential function (Fig. 4 B).

The time constant reflecting all the three Ca²⁺ removal mechanisms (τ_tran) was significantly larger in the RVCM (Fig. 4 C), while the time constant was almost identical when SERCA was neglected (τ_caff, Fig. 4 D), implying that the slower Ca²⁺ decay of RVCM was due to the lower SERCA activity. From the three rate constants (k_tran, k_caff, and k_0Na0Ca) obtained by taking the reciprocal of time constants (τ_tran, τ_caff, and τ_0Na0Ca), the relative contributions made to Ca²⁺ removal by SERCA, NCX, and PMCA (relative rate constant) were determined using the protocol described by Díaz et al. (2004), with correction for the plausible error from the difference in endogenous buffer (see In situ analysis of the Ca²⁺ binding ratio…). The total rate of Ca²⁺ removal was higher in LVCM than RVCM, mostly because of the 2.23-times-higher relative rate of SERCA in LVCM (Fig. 4 F). The relative NCX and PMCA activity were also lower in RVCM (Fig. 4 F). However, since >89% of the total calcium removal mechanism was performed by SERCA [(k_tran − k_caff)/k_tran], the difference of NCX and PMCA between two ventricles seems to have a partial effect on the total Ca²⁺ removal rate.

To investigate the underlying mechanisms of the observed difference in SERCA activity, we compared the protein amount of SERCA2a, which actively sequesters Ca²⁺ back into the SR lumen to remove intracellular Ca²⁺ (Sathish et al., 2006). When assessed by quantification of bands, the protein level of SERCA2a was not significantly different in the two ventricles (Fig. 5). Because the small difference that was not statistically significant could not account for the difference in SERCA activity, we focused on the endogenous inhibitor of SERCA, PLB, and its phosphorylated form (p-PLB), which relieves the inhibitory function of PLB (Tada and Katz, 1982; Simmerman et al., 1986). We compared the expression of PLB, which was normalized to the expression of internal loading control. The amount of PLB was not different in the two ventricles, while that of p-PLB was lower in RVCM. The ratio of p-PLB to PLB was 0.40 ± 0.039 versus 0.28 ± 0.065, also significantly lower in RVCM (Fig. 5 E).

Although the lower SERCA activity could explain the slower Ca²⁺ decay of RVCM, the similar Ca²⁺ peaks despite the shorter APD suggested additional differences in the Ca²⁺-buffering components of RVCM. We hypothesized that the Ca²⁺-binding myofilament proteins, such as TnC, might be different between RVCM and LVCM. TnC is the major Ca²⁺-binding protein in the cytosol of cardiomyocytes (Smith and Eisner, 2019). The immunoblot assay of the myofilament fraction revealed that the expression levels of TnC, TnI, and TnT were commonly lower in RVCM compared with LVCM (Fig. 5). In contrast, the levels of tropomyosin, desmin, actinin, and myosin binding protein C did not differ between RVCM and LVCM.

Overloading of samples is a widespread problem, termed membrane saturation, that may compromise the reliability of quantitative analysis, especially in enriched samples (Ghosh et al., 2014; Pillai-Kastoori et al., 2020). To overcome this limitation, we performed immunoblot experiments while loading various amounts of protein (1–5 µg) to find the range of protein amount that does not cause membrane saturation (Fig. S7). In the case of TnI, TnC, and tropomyosin, the signal of protein bands was not saturated up to 5 µg of protein, while that of TnT showed a saturated signal from 4 µg of protein (Fig. S7 B). The expression of cTn was lower in RVCM at the unsaturated ranges of protein amount. All the signals of blotting in this experiment were normalized to the total signal of protein using Ponceau S staining as an internal loading control (Gilda and Gomes, 2013; Sander et al., 2019).

Cytosolic Ca²⁺ buffers in cardiomyocytes are composed of various components, including proteins such as TnC, calmodulin, and myosin. The major Ca²⁺ buffers in the physiological range of Ca²⁺ concentration (100–1,000 nM) are regulatory Ca²⁺-specific binding sites of TnC and active sequestration by SERCA. The other constituents (calmodulin and ATP) are quantitatively small or have dissociation constants that are widely different from the physiological [Ca²⁺]_i (Ca²⁺/Mg²⁺-binding site of TnC and myosin light chain; Smith and Eisner, 2019). We analyzed the κ_S values of the RVCM and LVCM after excluding the factors from calcium-induced calcium release (CICR), SERCA, and NCX using the SERCA inhibitor (thapsigargin, 2.5 µM), RYR activator (caffeine, 20 mM), and Na⁺-free bath solution (Fig. 6 and Fig. 7 A; see Eq. 1 in Materials and methods).

When depolarizing pulses (from −40 to 10 mV, 100 ms, 1 Hz) were applied in 0 Na/2 Ca solution, a cumulative increase in [Ca²⁺]_i ≤2 µM was observed during the 8–10 consecutive pulses (upper panel of Fig. 7 B). Because the Ca²⁺ influx via VOCC_L was the sole source of Δ[Ca²⁺]_i, the integral of calcium currents was obtained (Fig. 7 B, bottom panel). The integrated Ca²⁺ influx was divided by the accessible cell volume (vol_ac, in Materials and methods) to obtain the calculated net increase in [Ca²⁺] (Δ[Ca²⁺]_tot,cal). The median values of [Ca²⁺]_i before and immediately after the individual pulses were measured. The relationship between κ_B and the median [Ca²⁺] was plotted (circles in Fig. 8, A and B), and the κ_B and [B]_tot values were obtained by fitting Eq. 2 (see Materials and methods).

In the above experimental measurement, it should be noted that when using fura-2, an “added-buffer” condition (Neher, 1995), the contribution from fura-2 (Fura_tot) has to be excluded from [B]_tot to deduce the endogenous calcium buffer, [S]_tot (Fig. 8, C–E). Because we know the dissociation constant of fura-2 (K_d,fura-2) from the in vitro calibration mimicking the cytosolic environment (see Materials and methods; Fig. S1) and the effective total concentration of fura-2 from Eq. 3 ([fura-2]_tot; Fig. 8 D), the correlation between the Ca²⁺-binding ratio of fura-2 (κ_fura-2) and the corresponding median [Ca²⁺] could be fitted to a differential form of the Michaelis–Menten equation (square symbols in Fig. 8, A and B). Finally, the differences between κ_B and κ_fura-2, i.e., κ_S could be obtained by subtraction, the result of which was fitted by Eq. 1 to calculate the two parameters of the endogenous buffer: the total concentration of the intrinsic buffer ([S]_tot) and the dissociation constant of the intrinsic buffer (K_d,S). In accordance with the expression level of myofilament proteins, [S]_tot of RVCM was actually smaller than that of LVCM (195.8 ± 19.61 µM versus 127.5 ± 21.09 µM; Fig. 8 E). The reported dissociation constant of TnC (K_d,TnC < 0.6 µM, from Smith and Eisner, 2019) is smaller than that of measured intrinsic buffer, which was 0.90 ± 0.156 µM in LVCM. Therefore, the calcium affinity of the endogenous buffer would be lower in RVCM owing to the lower proportion of troponin among the total intrinsic buffer. Consistently, the K_d,S values were larger in RVCM than in LVCM (K_d,S of RVCM = 1.21 ± 0.239 µM). Using the parameters obtained from the experiments (κ_B, κ_S, κ_fura-2, [B]_tot, [fura-2]_tot, [S]_tot), we calculated the bound Ca²⁺ (calculated bound Ca²⁺) from the median [Ca²⁺] using the Michaelis–Menten equation. The calculated buffering power of RVCM was lower than that of LVCM (Fig. 8, G and H).

The above differences in the factors determining the Ca²⁺ transient kinetics (e.g., lower SERCA activity and endogenous Ca²⁺ buffer) suggested differential features of contraction–relaxation between RVCM and LVCM. In the IonOptix experiment, we analyzed the SL changes induced by repetitive electrical field stimulation (2 Hz; Fig. 8 A). 20 animals were used for measuring SL, and 6 parameters analyzed for each animal are depicted in Fig. S8. The basal SL of RVCM during the diastolic phase appeared to be longer than that of LVCM, but not statistically significantly when nested t test was applied (Fig. 9 C, P = 0.190). The average maximum shortening in 10 consecutive pulses (ΔSL) was smaller in RVCM (Fig. 9 D). The percent normalized values (ΔSL/SL %) were also smaller in RVCM than in LVCM (Fig. 9 E). To clarify the difference in relaxation speed, the relative SL changes normalized to the peak SL were analyzed (Fig. 9 B). The relaxation speed was lower in RVCM (Fig. 9, F–H), which seems to be consistent with the slower decay of the Ca²⁺ transient in RVCM (Fig. 3, E–G).

The physiological heart rate of rat is faster (5–6 Hz) than the pacing rate in the present study (2 Hz). To confirm whether the different pacing rates might have induced the different contractile properties between RVCM and LVCM, the SL analysis was conducted at different pacing frequencies (2–6 Hz), where the smaller ΔSL and the slower relaxation speed in RVCM were consistently observed (Fig. S9).

Finally, we recapitulated the experimental findings from the Ca²⁺ transient and the single-cell contraction (half-SL change, ΔSL_half) by using a combinational computational model of rat ventricular cardiomyocytes constructed from two mathematical models (Pandit et al., 2001; Negroni and Lascano, 2008). Because the models were based on the experimental values obtained from LVCM, the three parameters were modified: the larger I_to, the lower SERCA activity, and the lower cTn expression (Fig. 10; for the AP simulation, see Fig. 2). To consider the kinetics of Ca²⁺ binding with TnC, we also calculated the differential parameter expressing the flux of Ca²⁺ to the cTn (dJ_TRPN/dt) according to the I_tropn in Negroni and Lascano (2008) (Fig. 10 B).

When a higher I_to (g_to ⋅ 1.31) was introduced, all three amplitudes (Ca²⁺ transient, dJ_TRPN/dt, and ΔSL_half) were decreased (Fig. 10 A, upper panels), which was consistent with the effect of the RVCM AP-clamp on I_Ca,L (Fig. 2 H). The combined lowering of SERCA activity (I_SERCA ⋅ 0.44) with the higher I_to slowed the decay of the Ca²⁺ transient and ΔSL_half, resulting in their amplitudes being only slightly increased (Fig. 10, middle panels, blue traces). In contrast, the combined lowering of cTn (cTn ⋅ 0.65) with a higher I_to significantly increased the amplitude of the Ca²⁺ transient while decreasing dJ_TRPN/dt and ΔSL_half (Fig. 10, middle panels, green lines). By combining the changes in the three parameters, I_to, SERCA, and cTn, the simulated Ca²⁺ transient and ΔSL_half were close to the experimental data of the RVCM when compared with the simulated LVCM (Fig. 10, lower panels).

In addition to the previously known differences in electrophysiology (larger I_to) and lower SERCA activity in RVCM (Afzal and Dhalla, 1992; Molina et al., 2016), our present study demonstrated novel aspects of biventricular differences regarding E–C coupling, the lower expression of cTn that seems to underlie the lower Ca²⁺ buffering capacity and smaller contraction in RVCM. The lower SERCA activity and cTn in RVCM could explain the apparent mismatches between the AP versus Ca²⁺ transient as well as the Ca²⁺ transient versus ΔSL, which was reproduced by mathematical simulations.

The whole-cell patch clamp showed shorter APD and larger I_to in RVCM than in LVCM (Fig. 1 and Fig. 2, A–C). These findings are consistent with those of previous studies on rodent cardiomyocytes (Di Diego et al., 1996; Brunet et al., 2004). Because the I_Ca,L was not different (Fig. 2, D–F), the higher I_to seems to be responsible for the faster repolarization in RVCM, which was also reproduced by the mathematical electrophysiological model of rat cardiomyocytes (Fig. 2 G). Heterogeneity of electrical properties has also been observed in LVCM of the epi- and endocardium (Watanabe et al., 1983; Kondo et al., 2006). In fact, a study comparing RVCM with epi- and endocardial LVCM showed that the AP configuration of RVCM was similar to that of epi-LVCM (Kondo et al., 2006). Regrettably, we did not rigorously divide the LVCM from the epi- and endocardium. Nevertheless, the consistent statistical significance of the different parameters in AP and I_to suggested that the LVCM were a relatively homogeneous group close to the endocardial myocytes (Figs. 1 and 2).

Despite the shorter APD and smaller net Ca²⁺ influx via VOCC_L, the amplitude of the Ca²⁺ transient was not different between RVCM and LVCM (Fig. 3). Because the amount of Ca²⁺ released from the SR (CICR) is mainly responsible for the final amplitude of Ca²⁺ transient in cardiomyocytes (Eisner et al., 2017), one has to also consider the efficiency of CICR when comparing the biventricular difference. Unfortunately, we could not directly evaluate the CICR efficiency or activity of RYR channels. Because the inactivation of VOCC_L is largely affected by the [Ca²⁺] of the junctional space near RYR, an indirect estimation from the inactivation speed of I_Ca,L in the perforated whole-cell patch clamp could be provided. The identical time constants of inactivation kinetics did not support the hypothesis that CICR efficiency was higher in RVCM. Therefore, the SR Ca²⁺-release via CICR could be smaller in the beating RVCM than in LVCM because of the smaller Ca²⁺ influx during the shorter AP.

Taking these findings together, the more plausible explanation for the mismatch between the APD and the Ca²⁺ transient was the lower level of endogenous Ca²⁺ buffer and removal mechanism. In cardiomyocytes, it is known that only 1% of the total calcium increase could be experimentally detected as free Ca²⁺; 99% is bound to the cytosolic buffers or removed by SERCA, NCX, and PMCA (Fabiato, 1983; Hove-Madsen and Bers, 1993). Among the latter Ca²⁺ removal mechanisms, SERCA predominantly operates in rodent cardiomyocytes (Bassani et al, 1994; Díaz et al, 2004). A previous study using rat myocardium demonstrated slower SERCA in the SR vesicles isolated from RVCM than in those isolated from LVCM (Sathish et al., 2006). A similar finding was also confirmed in the present study using intact cardiomyocytes (Fig. 4), and the underlying mechanism of difference in SERCA activity was due to the difference in the phosphorylation states of PLB, which was lower in RVCM (Fig. 5). However, the lower SERCA activity appears to be more responsible for the slower rate of the Ca²⁺ transient decay than the instantaneous increase, which could also be simulated by mathematical modeling (Fig. 10).

Among the cytoplasmic proteins of cardiomyocytes, TnC could provide quantitatively effective Ca²⁺ buffering capacity in the physiological range of Δ[Ca²⁺]_i (Pan and Solaro, 1987; Smith and Eisner, 2019). The Ca²⁺-binding site of TnC is classified into two categories: Ca²⁺-specific site and Ca²⁺/Mg²⁺-binding site (Robertson et al., 1981). Because the Ca²⁺/Mg²⁺-binding site with low K_d values (0.0195 µM, from Pan and Solaro, 1987) would be largely occupied by Ca²⁺ or Mg²⁺ at their resting levels, binding with the Ca²⁺-specific site in TnC would change dynamically during the cardiac cycle. In this study, the expression of TnC, along with TnI and TnT, was actually lower by ∼30% in RVCM than LVCM (Figs. 6 and S7). Based on the buffering properties of cTn, we propose that the differential Ca²⁺ buffering capacity could explain the mismatch between APD and the amplitude of the Ca²⁺ transient.

The apparently lower ratio of cTn over other myofilament proteins was an unexpected finding. According to the recent cryo-electron microcopy study of the cardiac myofilament structure, cTn attaches to a specific position of tropomyosin at every seventh actin monomer. Therefore, the ratio of actin, tropomyosin, and cTn would be a tight stoichiometry of 7:1:1 (Oda et al., 2020; Yamada et al., 2020). In this respect, the lower cTn in RVCM could imply a discordance between the amount of tropomyosin and cTn and putative instability of the myofilament complexes. Unfortunately, our data from immunoblotting experiments could not provide a reasonable explanation for the controversy. Nevertheless, a previous 2-D electrophoresis analysis of porcine heart also suggested the relatively lower level of TnC than tropomyosin in the RV than LV (Phillips et al., 2011). In addition, a differential in-gel electrophoresis combined with mass spectrometry study of human cardiac tissues suggested higher tropomyosin in comparison with TnI in LV than RV (Su et al., 2015). The studies including our present study suggested a difference in cTn/tropomyosin between RV and LV. Although it is not very likely, the regularity of association between cTn and tropomyosin in RVCM might not be as strict as in LVCM. Otherwise, the immunoblot results might include putative soluble components of the myofilament proteins that are different between RV and LV.

To precisely compare the intrinsic Ca²⁺-buffering capacity, we calibrated the in situ Ca²⁺ buffering properties using the added-buffer method (Berlin et al., 1994; Neher, 1995). For this purpose, it was necessary to define the buffer components as a collective Ca²⁺-binding species (S), described by a single dissociation constant (K_d,S) and concentration ([S]_tot). Although S partially reflects components other than TnC such as calmodulin, ATP, and myosin, it mostly depends on the Ca²⁺-binding affinity and concentration of TnC under physiological conditions (Smith and Eisner, 2019). The calibration and analysis actually revealed that the Ca²⁺-binding ratio, K_d,S, and [S]_tot were lower in RVCM than in LVCM (Figs. 7 and 8). Taken together, these data suggest that the similar amplitudes of Ca²⁺ transients despite their shorter APD were due to the lower expression of the Ca²⁺-binding TnC in RVCM. However, there might be another plausible difference in posttranslational modification of TnC affecting Ca²⁺-binding properties (Robertson et al., 1982; Blanchard and Solaro, 1984), which remains to be investigated in the future.

In rodent cardiomyocytes, SERCA is a major functional Ca²⁺ buffer along with cTn (Berlin et al., 1994; Smith and Eisner, 2019). The calculation of Ca²⁺ buffering using the methods by Berlin et al. (1994) could not take SERCA activity into account. To overcome the limitation, we calculated this effect of lower SERCA activity using mathematical in silico models (Fig. 10). Although the lower cTn increased the height of the Ca²⁺ transient to a greater extent, the lower SERCA activity also had an effect on the Δ[Ca²⁺]_i. It appeared that those two features of RVCM, the lower expression of cTn and the activity of SERCA, were acting in concert in Δ[Ca²⁺]_i of RVCM. Notably, the reduction in sarcomere shortening due to lower cTn expression became less significant by the combined lowering of SERCA (Fig. 10 C, green line versus gray line). Considering the unavoidable lag time for the Ca²⁺-dependent operation of myofilament proteins, the combined contribution of SERCA as well as cTn would determine the final amplitude of contraction, that is, ΔSL.

Although Kondo et al. (2006) reported no difference in SERCA2a gene expression and its inhibitory regulator PLB, Sathish et al. (2006) showed lower levels of the protein SERCA2a and PLB proteins in RVCM than LVCM. Functional studies have suggested lower SERCA activity in RVCM (Afzal and Dhalla, 1992; Sathish et al., 2006), consistent with the present study. However, there was a quantitative difference; the previous functional studies showed four to five times slower maximum velocity of SR Ca²⁺ uptake (V_max) in RVCM (Afzal and Dhalla, 1992; Sathish et al., 2006). Our analysis of SERCA from the analysis of rate constants showed a 2.23-fold larger rate constant in RVCM than LVCM (Fig. 4). Even though the measured variables were not the same (V_max versus τ), the difference was still significant. A plausible explanation is the discrepancy in measuring processes: SR-enriched homogenate (Sathish et al., 2006) versus in situ Ca²⁺ imaging (present study). Regardless of the quantitative difference, a lower SERCA activity and lower phosphorylation of PLB might be the key features of E–C coupling in RVCM.

Here, we found another notable point in the E–C coupling process: shorter ΔSL despite the similar Δ[Ca²⁺]_i in RVCM, which might be explained by the lower expression of cTn. Under the physiological environment, the RV is exposed to smaller afterload with more rapid pressure decline in cardiac cycle, and the P-V loop shows shorter isovolumic contraction time with slower decay of systolic pressure (Dell’Italia and Walsh, 1988). The macroscopic properties of RV such as the thinner wall and nonoblique geometry of myofibers are thought to be associated with the hemodynamic differences (Redington et al., 1988; Sengupta et al., 2006; Walker and Buttrick, 2013). Although it is difficult to extrapolate the properties of a single myocyte to the whole ventricle, the smaller ΔSL and slower relaxation of RVCM might partly explain the triangular shape of the P-V loop of RV.

Although not directly comparable with our present results, a previous study using a point mutation in TnI (K118C) with dominant-negative effects showed decreased contractility and shorter length of cardiomyocytes in transgenic mice (Wei and Jin, 2014). Another case was missense mutations in the Ca²⁺-specific binding site of TnC (E59D and D75Y), which was discovered from a patient diagnosed with idiopathic dilated cardiomyopathy. Rat cardiomyocytes with adenoviral expression of D75Y showed less cell shortening than wild-type cardiomyocytes, even without changes in the Ca²⁺ transient (Lim et al., 2008). In addition, the increased proteolytic activity and subsequent lowering of cTn were responsible for the impaired contractility and relaxation in ventricular hypertrophy (van der Laarse, 2002). These studies suggest that the expression, Ca²⁺-binding affinity, and Ca²⁺ sensitivity of cTn closely regulate cell shortening. Although we did not directly investigate the Ca²⁺-dependent contraction of permeabilized cardiomyocytes to analyze Ca²⁺ sensitivity, we cautiously interpret that the smaller ΔSL might be due to, at least partly, the lower amount of cTn in RVCM. To confirm this hypothesis, it is worth investigating the relationship between the role of lowered cTn and smaller ΔSL of RVCM.

In the computational simulations combining the previous mathematical models and the kinetics of Ca²⁺ binding to the cTn (Pandit et al., 2001; Negroni and Lascano, 2008), we calculated the Ca²⁺ flux to the myofilament fraction (J_TRPN) and its time-dependent change (dJ_TRPN/dt; Fig. 10 B). This approach was based on the assumption that the Ca²⁺ flux to the myofilament is totally engaged in the Tn-Ca²⁺ state cycle. In the Negroni and Lascano (2008) model, the force and length of contraction depend on the number of cross-bridges, that is, the amount of effective troponin. In this context, the cTn-dependent ΔSL in the virtual simulation would be a self-evident result. Nevertheless, the intriguing point of our present study is that the expression of cTn was actually lower in RVCM (Figs. 6 and S7).

Although the heart rate of rats is faster than that of humans, 5–6 Hz (300–350 bpm), we used a 2-Hz frequency for pacing cardiomyocytes due to constraints in the experimental system using isolated single cells. For instance, myocytes showed tendency of deterioration at high rates of stimulating frequencies, especially when loaded with fura-2, limiting the simultaneous rerecording of Ca²⁺ transient and ΔSL. In the IonOptix system solely recording SL, we could confirm that the differences in contractility (ΔSL and relaxation speed) between RVCM and LVCM were maintained at higher pacing frequencies (Fig. S9).

It was notable that both RVCM and LVCM showed a tendency toward decreasing ΔSL at the higher pacing rates (Fig. S9, B and C), suggesting a kind of negative force–frequency relation (FFR). In large animals, the ventricular myocytes exhibit positive FFR due to increased SR Ca²⁺ content (Kurihara and Sakai, 1985; Joulin et al., 2009). In contrast, rat cardiomyocytes exhibit very weakly positive or even negative FFR (Banijamali et al., 1991; Maier et al., 1998; Janssen and Periasamy, 2007) owing to their higher SR Ca²⁺ content, even at the low stimulation frequencies (Shattock and Bers, 1989; Maier et al., 2000).

While both RVCM and LVCM showed frequency-dependent acceleration of relaxation (FDAR), slower relaxation of RVCM was consistently observed at the physiological pacing frequency (Fig. S9, E and F). In physiological conditions, FDAR might be essential to refill more rapidly between beats (Schouten, 1990; Hussain et al., 1997). It was reported that the FDAR was SR dependent and modulated by SERCA (Bassani et al., 1995). Although calcium transient was not directly compared at the higher pacing frequencies, accelerated Ca²⁺ removal could be inferred from the recordings of ΔSL. Also, we cautiously concluded that slower recovery of Ca²⁺ transient of RVCM would be conserved at the physiological heart rate of rodents.

Along with the anatomic biventricular differences such as the ventricular wall thickness and the muscle fiber band folding pattern, the lower cardiac troponin and the smaller ΔSL in the RVCM might be related to the pathophysiological responses of the RV in vivo. Changes in Ca²⁺ sensitivity make a significant contribution to the contractility of cardiomyocytes during disease progression (Belin et al., 2007; Marston and de Tombe, 2008). Because the phosphorylation of TnI plays a critical role in the modulation of Ca²⁺ sensitivity, the lower cTn might be partly responsible for the less effective adaptation to the increased afterload.

The importance of Ca²⁺ buffering has been suggested in genetically engineered rats to express parvalbumin, a Ca²⁺-binding protein, in cardiomyocytes (Wahr et al., 1999). Although there was little effect on the peak Ca²⁺ transient due to the slow Ca²⁺ binding kinetics of parvalbumin, the gene-transferred cardiomyocytes showed accelerated decay of [Ca²⁺]_i, implying improved diastolic performance of cardiomyocytes. These effects of parvalbumin suggest the potential importance of slow buffers, including the Ca²⁺/Mg²⁺-binding site on TnC, in the development and consequences of heart failure, as recently mentioned by Smith and Eisner (2019). While we did not directly measure the slow-acting buffer power of cardiomyocytes, the lower expression of cTn in RVCM might be partly associated with the vulnerability of the RV.

Beat-to-beat alternation of cardiac AP (alternans) is regarded as an arrhythmia mechanism with an increased risk of sudden cardiac death (Kulkarni et al., 2019). The delayed Ca²⁺ decay or abnormal Ca²⁺ signal could underlie the alternans, as suggested by the reduced probability of alternans by injecting EGTA into cardiomyocytes (Baudenbacher et al., 2008). The increased Ca²⁺ buffer could suppress the pathological Ca²⁺ waves triggered by SR Ca²⁺ release (Nivala et al., 2012; Bovo et al., 2015). In addition to the buffering effects on Ca²⁺ sparks, there was an idea that incomplete recovery at the time of the next stimulus followed by the slow decay of the Ca²⁺ transient might be attributed to the induction of alternans (Baudenbacher et al., 2008). In fact, genetically up-regulated SERCA2a suppressed the genesis of alternans (Lyon et al., 2011; Cutler et al., 2012). Interestingly, ventricular arrhythmia was more frequent in RV myocardial infarction than in inferior or anterior wall LV myocardial infarction (Mehta et al., 2001; Ondrus et al., 2013). According to the present study, we cautiously suppose that the lower Ca²⁺ buffer by TnC and the slower Ca²⁺ sequestration by SERCA in RVCM implied higher susceptibility to ventricular arrhythmia in the above conditions.

In summary, from the comparative analysis of the E–C coupling between the RVCM and LVCM of rat, in addition to the electrical difference (higher I_to and shorter AP), we found lower Ca²⁺-buffering mechanisms in RVCM, that is, TnC and SERCA (Fig. 11). The lower expression of TnC could explain the dual mismatch phenomena in the biventricular differences of the E–C coupling. The pathophysiological implications of the present findings, such as the putative changes in TnC expression, require further investigation in right heart disease models.

David A. Eisner served as editor.

This work was supported by the National Research Foundation of Korea funded by the Ministry of Science and ICT, Republic of Korea to S.J. Kim (grants NRF-2018R1D1A1B07048998, NRF-2018R1A5A2025964, and NRF-2021R1A2C2007243), and by EDISON (Education-Research Integration through Simulation on the Net; NRF-2016M3C1A6936605). This work was also supported by a grant of the M.D., Ph.D./Medical Scientist Training Program through the Korea Health Industry Development Institute to Y.K. Jeon, funded by the Ministry of Health and Welfare, Republic of Korea.

The authors declare no competing financial interests.

Author contributions: conception and design: Y.K. Jeon, J. Jang, J.B. Youm, Y.H. Zhang, and S.J. Kim; collection of data: Y.K. Jeon, J.W. Kwon, J. Jang, S.W. Choi, J. Woo, S.H. Cho, and B.I. Yu; analysis and/or interpretation: Y.K. Jeon, J.W. Kwon, J. Jang, Y.S. Chun, J.B. Youm, Y.H. Zhang, and S.J. Kim; manuscript writing: Y.K. Jeon, J.B. Youm, Y.H. Zhang, and S.J. Kim. All the authors reviewed and approved the final version of the manuscript.