Upregulation of Na⁺/Ca²⁺ Exchanger Expression and Function in an Arrhythmogenic Rabbit Model of Heart Failure

Animal Preparations and Induction of HF

Studies were performed on healthy adult New Zealand White rabbits of either sex (2.9 to 3.5 kg; Doe Valley Farms, Inc, Bentonville, Ark). HF was produced by combined aortic insufficiency (AI) and aortic constriction. Induction of AI was performed by inserting a beveled catheter via the left carotid artery and repeatedly pushing it through the aortic valve to increase pulse pressure by at least 50%, as previously described.⁵ The severity of AI was assessed by 2D echocardiography with color flow mapping.¹⁶ After 14 to 28 days, a left thoracotomy was performed, allowing aortic constriction (to 2.42 mm) just distal to the left subclavian artery.⁵ The protocol was approved by the Animal Studies Committees of Washington University and the University of Illinois at Chicago.

At baseline and at approximately 3- to 6-month intervals, each rabbit underwent echocardiographic examination as previously described.⁵ M-mode echocardiographic measurements of left ventricular end-diastolic (LVEDD) and end-systolic (LVESD) dimensions were obtained.¹⁷ Fractional shortening (FS) was calculated as: FS(%)=(LVEDD−LVESD)/LVEDD.

At baseline and at an average of once every 2 to 4 weeks after aortic constriction, Holter monitor recordings were obtained for 24-hour periods from the rabbits while they were conscious, as previously described.⁵

RNA Isolation and Northern Blotting

RNA was isolated from LV tissue,¹⁸ size-fractionated by agarose gel electrophoresis, and then transferred to nylon membranes.¹⁹ Detection of cardiac NaCaX mRNA used 1.3 kb cDNA of guinea pig cardiac NaCaX (from Dr K.D. Philipson, University of California, Los Angeles). Detection of SERCA2 used 2.3 kb cDNA from rat cardiac SERCA2 (from Dr W. Dillmann, University of California, San Diego). Signal intensity was quantified by autoradiography (−80°C) and laser densitometry and was normalized to that of 18S rRNA and of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA.²⁰

Immunoblotting

For the quantitative analysis of NaCaX and SERCA2 protein, LV tissue (200 to 300 mg) was homogenized in 9 volumes of extraction buffer and diluted into electrophoresis sample buffer for separation on 7.5% SDS-polyacrylamide gels.¹² After transfer to nitrocellulose²¹ and blocking with nonfat milk (5% wt/vol), blots were incubated with rabbit anti-dog NaCaX antiserum (from Dr K.D. Philipson, University of California, Los Angeles) or with rabbit anti-rat SERCA2 antiserum (from Drs R. Hartong and W. Dillmann, University of California, San Diego). Primary antibody binding was detected with an ECL Western blotting kit (Amersham) and quantified by laser densitometry.

Myocyte Isolation

Left ventricular myocytes were isolated via Langenddorf perfusion with 0.5 mg/mL collagenase as described,²² except that back flow across the incompetent aortic valve in HF rabbits was blocked by a balloon-tipped catheter inserted across the aortic valve and inflated in the LV outflow tract. Also, additional incubations with 0.4 mg/mL collagenase plus 0.02 mg/mL pronase were done at 37°C for up to 12 minutes. Cells were studied at 23°C or 37°C in superfusion chambers on glass coverslip bases pretreated with laminin.

Contractions and Ca²⁺ Transients

Along with cell shortening, [Ca²⁺]_i was measured using indo 1 fluorescence (excited at 365 nm and measured at 405 and 485 nm, K_d=441 nmol/L) as described.²³ Normal Tyrode’s (NT) solution contained (in mmol/L): NaCl 140, KCl 4, MgCl₂ 1, CaCl₂ 2, glucose 10, and HEPES 5, pH 7.4 at 23°C. After steady-state twitches (0.5 Hz), stimulation was stopped and NT+10 mmol/L caffeine solution (NTCaff) was rapidly applied for 10 seconds (causing SR Ca²⁺ release and extrusion via NaCaX^{22 23} ).

Whole-Cell Voltage Clamp

Ion currents were measured with the whole cell ruptured-patch technique (pipette tip resistance=0.8 to 1.2 MΩ) with membrane capacitance measured by 5 mV voltage steps from −80 mV. I_Na/Ca was measured as described by Hobai et al.²⁴ Pipettes were filled with (in mmol/L) 45 CsCl, 55 Cs methanesulfonic acid, 10 ATP-tris, 0.3 GTP-tris, 20 HEPES, 10.8 MgCl₂ (1 mmol/L free Mg), 5 1,2-bis(2aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid, 5 DiBr-1,2-bis(2aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid, 2.21 CaCl₂ (100 nmol/L free Ca), and 14 NaCl (pH 7.3 with CsOH). To isolate I_Na/Ca, the NT (37°C) was modified (6 CsCl replaced KCl, and 10 μmol/L strophanthidin, 30 mmol/L BDM and sometimes 10 μmol/L nifedipine were added). Ni (5 mmol/L) was added to block I_Ca and I_Na/Ca.

Data Analysis

All results were expressed as means±SE and unpaired t tests were used to compare 2 groups (significance at P<0.05).

An expanded Materials and Methods section is available online at http://www.circresaha.org.

Development of Hypertrophy and Failure

HF was induced in 13 rabbits. Severe aortic insufficiency, confirmed by 2D color flow Doppler echocardiography, was first produced. Two to 6 weeks later, aortic constriction was performed. Over the subsequent 1 to 24 months, the rabbits developed progressive cardiac enlargement and decreased left ventricular systolic function. LVEDD increased by 40% (P<0.001), and LVESD increased by 68% (P<0.0001, Table 1⇓). Mean fractional shortening decreased from 36±2% to 23±1% (P<0.0001). The 6 age-matched control rabbits demonstrated LVEDD, LVESD, and fractional shortening that were comparable with baseline values for the rabbits that underwent induction of HF (Table 1⇓). There were significant increases in heart weight and heart-to-body weight ratio (77% and 79%, respectively), whereas body weight was not different between control and HF rabbits (Table 1⇓). There was also a corresponding significant increase in resting myocyte length (31%) and width (45%) in cells isolated from control and HF rabbits (Table 1⇓) resulting in an 89% increase in the length×width product.

Spontaneously Occurring Ventricular Arrhythmias

Holter monitoring in the conscious state was performed in 11 rabbits at baseline and, on average, every 1 to 2 months after the induction of HF in 10 of the 11 rabbits. At baseline, none of the rabbits exhibited any ventricular arrhythmias. However, over the course of the 1 to 24 months after the induction of HF, all of the HF rabbits exhibited spontaneously-occurring ventricular arrhythmias, with up to 13 339 premature ventricular complexes, 402 couplets, and 32 runs of nonsustained VT over the course of 24 hours. Nine of the 10 HF rabbits demonstrated nonsustained VT during serial Holter monitoring. The longest run of VT was 31 beats and the fastest was a 15 beat run of VT at a rate of 360 beats per minute (Figure 1⇓).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Spontaneously occurring ventricular arrhythmias in conscious HF rabbits. Lead II ECG tracings of spontaneous rhythm during Holter monitoring in the conscious state. Top, A 15- and a 12-beat run of nonsustained VT from a HF rabbit. Bottom, A 31-beat run of VT followed by an isolated premature ventricular contraction from a HF rabbit.

Arrhythmogenic Effects of Isoproterenol

To determine whether HF myocytes also exhibit spontaneous activity (as in the whole heart in vivo), the appearance of aftercontractions in isolated myocytes was also studied. Myocytes were field-stimulated at increasing frequency (0.2 to 1.8 Hz) in the absence and presence of 10 μmol/L isoproterenol. After the 20^th stimulated twitch, there was a 10-second interval with no stimulation to observe for the presence of aftercontractions. At baseline (without isoproterenol), no aftercontractions occurred in either control or HF myocytes (Figure 2⇓). After the addition of 10 μmol/L isoproterenol, aftercontractions were induced in 4 of 5 HF myocytes but in 0 out of 5 control myocytes studied with this protocol. Thus HF myocytes exposed to β-adrenergic agonist demonstrate spontaneous SR Ca²⁺ release and the induction of aftercontractions.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Aftercontractions induced by isoproterenol (10 μmol/L) in HF myocytes. A train of 20 field-stimulated contractions in a HF myocyte fails to induce an aftercontraction (top). After exposure to 10 μmol/L isoproterenol (Iso) the same protocol results in an aftercontraction (arrow). In both cases, the stimulation frequency was 0.5 Hz and [Ca²⁺]_o=2 mmol/L.

NaCaX and SERCA2 mRNA and Protein Levels

Figure 3⇓ shows Northern blot analysis of NaCaX mRNA in LV myocardium from control and HF rabbits. All control and HF rabbits demonstrated a 7-kb hybridization band, corresponding to the expected position for the NaCaX. Four of the 8 HF rabbits also demonstrated a high molecular weight transcript at ∼14 kb that was not evident in any of the controls. However, on prolonged exposure, the control rabbit with the highest NaCaX mRNA expression showed very low levels of the 14-kb transcript. Expression of total NaCaX mRNA, normalized to 18S rRNA, was increased 2.7-fold in the HF rabbits compared with controls (P<0.01). If only the 7-kb transcript is considered, there was still a 2.04-fold increase in NaCaX mRNA expression in the HF group (P<0.05, see Figure 9⇓). Similar values were obtained when NaCaX mRNA expression was normalized to GAPDH mRNA (eg, 2.9-fold increase in HF over controls for 7+14 kb).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Northern blot analysis of NaCaX mRNA in myocardial tissue from control and HF rabbits. In the 4 left lanes are samples from age-matched controls, and in the 4 right lanes are samples from HF rabbits. The uppermost gel was hybridized with a cDNA probe specific for cardiac NaCaX. All rabbits demonstrate expression of a 7-kb transcript (although one of the controls expressed low levels that were more apparent on prolonged exposures). However, HF rabbits 1, 3, and 4 also demonstrated a ≈14-kb transcript that was evident in only 1 of the 6 control rabbits. The bottom shows the same gel, which has been reprobed with cDNA specific for 18S rRNA to reflect comparable degree of loading. Quantitation of NaCaX mRNA was normalized to the levels of the 18S rRNA.

Western blots revealed a single band for NaCaX at 120kDa for both control and HF rabbits. Quantification indicated a 93% increase of NaCaX protein expression, consistent with the mRNA data (see Figure 9⇓).

Quantification of the levels of SERCA2 mRNA in LV was performed by Northern blot analysis using a 2.3-kb cDNA fragment specific for rat cardiac SERCA2. All control and HF rabbits exhibited a single hybridization band at 4.2 kb. SERCA2 expression of HF rabbits normalized to 18S rRNA or to GAPDH was not significantly different from that of control rabbits. (P=not significant).

Figure 4⇓ shows a Western blot for SERCA2, revealing a single band at 110 kDa for both control and HF rabbits. When normalized to the mean level of the control rabbits, the SERCA2 protein levels of the HF rabbits was not significantly different (see Figure 10⇓). However, as shown in Figure 4⇓, the level of SERCA2 protein did vary among the HF rabbits. Of note, the lowest level of SERCA2 protein in the HF rabbits (lanes 1 and 3 of HF) was evident in the 2 rabbits that had the most marked degree of LV enlargement and the most severe LV dysfunction. There was no significant relationship between the levels of NaCaX and SERCA2 protein levels.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

SERCA2 protein in control and HF rabbits. Immunoblot of myocardial SERCA2 protein for 6 age-matched controls (left) and 8 HF rabbits (right). Although there were no significant differences between HF and control (see Figure 10⇓), the lowest levels of SERCA2 protein among the HF were found in the first and third HF lanes. These same 2 animals had the most severe degree of LV enlargement and LV dysfunction.

There was good correlation between NaCaX mRNA and protein expression, especially among the HF group (Figure 5A⇓). We also examined the relationship of NaCaX (and SERCA) expression to LV systolic function in individual whole hearts. Figure 5B⇓ shows that the level of NaCaX mRNA and LV fractional shortening (FS) were inversely correlated (P=0.01). A similar relationship between NaCaX protein and FS was seen, but it did not attain statistical significance (P=0.067). There was no significant relationship between FS and SERCA2 mRNA (Figure 5C⇓) or protein (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Correlations with NaCaX and SERCA2 expression from individual hearts. A, Relationship between NaCaX mRNA (normalized to the levels of 18S rRNA) and NaCaX protein levels (normalized to total protein). B and C, Correlations between LV fractional shortening and NaCaX (B) and SERCA2 (C) mRNA expression (normalized to controls) for control and HF rabbits. Each point in these graphs is data from an individual heart.

Biochemical measurements (above) and isolated myocyte studies (below) were performed in 2 different groups of rabbits. However, in each case, HF rabbits demonstrated comparable levels of LV dilatation, depression of LV systolic function, and incidence of spontaneously-occurring ventricular arrhythmias (measured in both groups). In this regard, they were also comparable with the HF rabbits studied by 3D cardiac mapping.⁵ NaCaX and SERCA2 expression was assessed in left ventricular homogenates, rather than isolated myocytes. Because SERCA2 and NaCaX are so concentrated in ventricular myocytes, contamination from nonmyocytes in the homogenate were considered inconsequential. In previous studies in hypertrophied rat heart, changes in SERCA2 expression were comparable whether measured in left ventricular homogenate or isolated myocytes.¹²

Relaxation and [Ca²⁺]_i Decline in Isolated Cardiac Myocytes

The rate of myocyte relaxation and [Ca²⁺]_i decline during sustained application of 10 mmol/L caffeine has been shown to depend mainly on Ca²⁺ extrusion by NaCaX in rabbit ventricular myocytes.^{22 23} This is because net SR Ca²⁺ reuptake is prevented by caffeine, such that NaCaX only competes with the much slower sarcolemmal Ca²⁺-ATPase and mitochondrial uniporter. Thus relaxation and [Ca²⁺]_i decline during CafC provide functional data in the intact cell concerning Ca²⁺ transport by NaCaX.

Figure 6C⇓ shows CafC in control and HF myocytes and the normalized relaxation (Figure 6D⇓) is much faster in HF. Mean data (Table 2⇓) show that the half-time (t_1/2) of relaxation was 546±70 ms in control and 340±40 ms in HF. Considering relaxation rate to be inversely proportional to t_1/2, this constitutes a 61% increase in NaCaX-dependent relaxation in HF. Figure 7C⇓ shows comparable data for [Ca²⁺]_i during a CafC. In this case, [Ca²⁺]_i decline was approximately twice as fast in HF as in control. On average, [Ca²⁺]_i decline rate during CafC was 45% faster in HF than control (Table 2⇓).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Representative twitch contractions (A), normalized twitch relaxation (B), caffeine-induced contractures (C), and normalized relaxation of caffeine contractures (D) for myocytes from control (Ctl) and HF hearts.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Representative Ca²⁺ transients for twitches (A), normalized twitch [Ca²⁺]_i decline (B), and normalized [Ca²⁺]_i decline during caffeine-induced contractures (C) for myocytes from control and HF hearts.

Twitch relaxation in normal rabbit ventricular myocytes is dominated by the SERCA2 but only by a factor of 2 to 3 over the NaCaX.^{22 23} Thus the time course of relaxation and [Ca²⁺]_i decline of twitches should provide functional information about the SERCA2 in control versus HF. Figures 6B⇑ and 7B⇑ show normalized twitch relaxation and [Ca²⁺]_i decline in control and HF. The t_1/2 values were not significantly different between these groups (Table 2⇑). Superficially, this could be construed as evidence of unchanged SERCA2 function. However, some additional quantitative analysis may be warranted.

The issue here is that the SERCA2 and NaCaX are competing during the twitch. Thus, if NaCaX is upregulated and SERCA2 is unchanged (as the above data indicate), we would expect twitch relaxation to be accelerated in HF. If we consider simply that relaxation is attributed to parallel function of SERCA2 and NaCaX working at rates given by pseudo-rate constants (λ_SR and λ_NCX where λ=ln 2/t_1/2), then during a twitch, λ_twitch=λ_SR+λ_NCX and during a CafC, λ_CafC=λ_NCX. Applying this simple scheme to control contractions, λ_NCX is 1.41±0.49 s⁻¹ and λ_twitch is 4.77±0.18 s⁻¹, giving λ_SR=3.36 s⁻¹. This would suggest 29% of Ca²⁺ removal during a twitch in control rabbit myocytes is by NaCaX (in close agreement with more detailed quantitative analysis²³ ). In HF cells, λ_NCX is increased to 2.34±0.26 s⁻¹ and λ_twitch is comparable at 4.9±0.3 s⁻¹, giving a reduced λ_SR of 2.54 s⁻¹. This change in λ_SR (from 3.36 to 2.54) would imply a 24% reduction in SERCA2 function. Furthermore, the balance between the SERCA2 and the NaCaX is changed in HF (so λ_NCX∼λ_SR) such that they contribute about equally to twitch relaxation (ie, NaCaX component increases from control of 29% to 48% in HF). Analysis of the Ca²⁺ transient data yields a similar conclusion (17% reduction in SERCA2 function). Thus we cannot rule out a small reduction in SERCA2 function in the HF versus control rabbits.

Twitch Amplitude and SR Ca²⁺ Content

Twitches from HF myocytes were 26% smaller in amplitude compared with control (12.9±1.2% versus 17.4±1.3% of resting cell length, P<0.02; Table 2⇑). The amplitude of CafC provides an index of the SR Ca²⁺ content, and the mean value was only 8% smaller in HF (24.6% versus 26.5% of resting cell length, P=not significant). This raises the possibility that there is a reduction in fractional SR Ca²⁺ release in HF. However, using the ratio of twitch:CafC amplitude as a crude index of fractional SR Ca²⁺ release, we found no significant difference between HF and control.

There was no difference in resting [Ca²⁺]_I between the groups. The amplitude of twitch Ca²⁺ transients was 30% smaller in HF versus control cells (Table 2⇑). This is comparable with the depression of twitch contractions and LV fractional shortening but was not significant (possibly due to greater variance in [Ca²⁺]_i measurements). The 11% decrease in CafC Δ[Ca²⁺]_i was not significantly different (but again comparable with contraction results). The twitch:CafC ratio for Δ[Ca²⁺]_i was not significantly different between control and HF. Because the results for contraction and Ca²⁺ transients are relatively similar, we have no evidence of altered myofilament Ca²⁺ sensitivity in HF (although not measured directly). These results suggest that the reduction of twitch amplitude in HF is likely to be due to decreased Ca²⁺ transients. Reduced Ca²⁺ transients could, in principle, be due to reduced I_Ca trigger, lower SR Ca²⁺ load (possible based on the CafC data), or a depression of the E-C coupling process.

Measurement of I_Ca and NaCaX Current in Isolated Myocytes

Voltage clamp experiments were carried out to evaluate I_Ca (with respect to the issue of E-C coupling above) and also to measure I_Na/Ca under well controlled experimental conditions (to further verify the expression and functional results above). Figure 8A⇓ shows the central protocol used, where [Ca²⁺]_i was buffered to 100 nmol/L to minimize alterations in [Ca²⁺]_i during protocols. The conditions were also selected so that Ca, Na, and NaCaX currents should be the only ones available. From an initial holding potential of −90 mV, the cell was depolarized to −45 mV to activate and inactivate Na current. Then a step from −45 to 0 mV was used to activate and inactivate I_Ca. Finally, the voltage was stepped to 80 mV and ramped down to −140 mV to assess I_Na/Ca. The protocol was repeated in the presence of 5 mmol/L Ni (Figure 8A⇓) to obtain Ni-sensitive I_Ca and I_Na/Ca (Figure 8B⇓).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

I_Na/Ca characterization. A, Voltage protocol used to measure I_Na/Ca (see text) and typical membrane currents obtained in myocytes from a HF rabbit heart in the presence or absence of 5 mmol/L Ni. B, Ni-sensitive current from a representative HF and control myocyte. C, Current-voltage relationship for I_Na/Ca recorded in HF and control myocytes. D, I-V relationship for I_Na/Ca recorded in 9 HF and 9 control rabbit myocytes. E, Outward I_Na/Ca (at 60 mV) and inward I_Ca (at 0 mV) in control and HF myocytes. F, Current-voltage relationship for I_Ca recorded in 4 HF and 4 control myocytes from a holding potential of −40 mV.

Figures 8C⇑ and 8D⇑ show that both inward and outward I_Na/Ca are increased by 2.2-fold in HF. The apparent reversal potential is unchanged and close to the predicted value based on the pipette and extracellular solutions. This is consistent with the above data based on mRNA, protein, [Ca²⁺]_i decline, and relaxation of CafC (see Figure 9⇓).

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

NaCaX expression, activity, and current. Values are shown for NaCaX mRNA (7- and 14-kb transcripts) from Northern blotting, NaCaX protein from immunoblotting, pseudo-rate constant of relaxation of the caffeine contraction (λ_NCX=ln2/t_1/2), and I_Na/Ca (inward and outward, excluding values near the reversal potential, −70 to −50 mV). Results are normalized to values from control myocytes. Mean control values for λ_NCX was 1.41±0.49 s⁻¹ and for I_Na/Ca at 60 mV was 2.96±0.54 pA/pF. The ratio of I_Na/Ca for HF:control was 2.13 at 70 mV and 2.18 at −120 mV. *P<0.05 versus controls.

The amplitude of I_Ca was unchanged during the pulses to 0 mV (Figure 8E⇑). Some additional cells were also studied with standard I_Ca protocols to determine the voltage dependence of I_Ca (Figure 8F⇑). There was no difference in either I_Ca amplitude or voltage dependence. The I_Ca results suggest that depressed Ca²⁺ transients and contractions in HF were not attributable to altered I_Ca. Moreover, I_Ca density stayed remarkably constant despite the cellular hypertrophy in the HF cells.

Figures 9⇑ and 10⇓ summarize results from the different approaches used here to assess NaCaX and SERCA2 expression and function in control and HF. All of the data are consistent with an approximate doubling of NaCaX expression and function. In contrast, the results do not indicate any significant change in SERCA2 expression, although there might be a small decrease in function.

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

SERCA2 expression and activity. Plots of SERCA2 mRNA from Northern blotting, SERCA2 protein from immunoblotting, and pseudo rate constant of twitch relaxation (λ_twitch=ln2/t_1/2). Results are normalized to values from control myocytes. Mean control λ_twitch was 4.77±0.18 s⁻¹. No statistical differences were seen between HF and control.

Correlation of NaCaX Functional Expression and SR Ca²⁺ Content

An increase in NaCaX expression (with all other things remaining the same) might be expected to reduce the steady-state SR Ca²⁺ content (because NaCaX would compete better with the SERCA2 during [Ca²⁺]_i decline and diastole). Although hearts used for cell isolation were not used for Western blotting, we can use the t_1/2 of relaxation of CafC as a functional index of NaCaX level in a given cell. Figure 11⇓ shows that the greater the functional expression of NaCaX (smaller t_1/2 CafC values), the lower the SR Ca²⁺ content. Despite substantial cellular heterogeneity, the correlation was significant (P=0.015). Thus increased NaCaX expression may contribute to unloading the SR of Ca²⁺ in the steady state. There was no comparable correlation between twitch amplitude and functional expression of NaCaX.

• Download figure
• Open in new tab
• Download powerpoint

Figure 11.

NaCaX function versus SR Ca²⁺ load. Plot of t_1/2 of the CafC (as a functional index of myocyte NaCaX) versus amplitude of CafC (as an index of SR Ca²⁺ content) for control and HF myocytes. Each point is from an individual cell.

Based on the results above, the depression of twitch amplitude in HF myocytes does not seem to be due to alteration of I_Ca. There does seem to be a tendency toward lower SR Ca²⁺ content in HF. This by itself might explain the reduced twitch amplitude. However, altered E-C coupling cannot be completely ruled out because an altered SR Ca²⁺ load makes it more difficult to assess E-C coupling directly (see Discussion).

Functional Increase in NaCaX Activity in Failing Myocytes

In normal rabbit ventricular myocytes, relaxation and [Ca²⁺]_i decline are attributable to the following 4 competing Ca²⁺ removal systems: 1) SERCA2, 2) NaCaX, 3) sarcolemmal Ca²⁺-ATPase, and 4) mitochondria Ca²⁺ uniporter.^{22 23} Bassani et al²³ analyzed the relative contributions quantitatively, concluding that the SERCA2 removed 70%, NaCaX 28%, and the others totaled only ∼2%. The data from control hearts in the present study (29% NaCaX) agree well with those conclusions. However, the functional upregulation of NaCaX in HF shown here shifts this balance such that Ca²⁺ removal is closer to 50% SERCA2 and 50% NaCaX during twitch relaxation. The increase in NaCaX may well contribute to mechanical dysfunction and a larger inward I_Na/Ca could also contribute directly to arrhythmogenesis (see below).

Altered NaCaX expression and activity has been noted in experimental models of HF and in the failing human heart. Dogs with pacing-induced HF exhibit an increase in NaCaX protein and function (during [Ca²⁺]_i decline).²⁵ Rabbits with pacing-induced HF had a 44% increase in NaCaX protein but normal I_Na/Ca.²⁶ In contrast, Yao et al²⁷ found decreased NaCaX mRNA and I_Ca in a similar model.²⁷ Increased I_Na/Ca has been demonstrated in myocytes from cardiomyopathic Syrian hamsters²⁸ and infarcted rabbits.²⁹ Several studies have shown an increase in both NaCaX mRNA and protein in myocardium from patients with end-stage HF obtained at the time of cardiac transplantation.^{10 11} Although little human functional data are available, Pieske et al³⁰ recently showed a similar shift in NaCaX:SERCA2 function in human HF (from 25%:75% to ∼50%:50%).

Hasenfuss et al³¹ recently showed that in human HF, upregulation of NaCaX protein correlated with a relative lack of LV diastolic dysfunction. This is consistent with our results in which twitch relaxation and [Ca²⁺]_i decline were maintained and NaCaX was functionally increased in HF rabbits, but we did not directly evaluate diastolic function in vivo.

High Molecular Weight Transcript of NaCaX mRNA

In the present study, 4 of the 8 HF rabbits demonstrated a high molecular weight (≈14 kb) transcript of the NaCaX mRNA using a cDNA probe that recognizes cardiac NaCaX mRNA. We found this transcript at very low levels in only one of the controls, and in rabbit brain and kidney, but not in skeletal muscle (data not shown). A similar 14-kb transcript was detected in mouse heart and kidney³² ; rabbit heart, kidney, and brain³³ ; and human brain³⁴ but not in myocardium from patients with end-stage HF.^{10 11} It remains to be determined as to whether this 14-kb transcript is due to alternative splicing of NCX1 (the gene for NaCaX) or a long poly-A tail on some NaCaX mRNA and as to what its role is. However, the 2-fold increase in the 7-kb NaCaX transcript in the HF rabbits provides evidence that enhanced functional NaCaX activity is due to a generalized increase in NaCaX expression and not solely to increased expression of the high molecular weight transcript.

SERCA2 Expression

In this rabbit model of nonischemic HF, we found no change in SERCA2 mRNA or protein expression. Function of the SERCA2 based directly on the rate of twitch relaxation and [Ca²⁺]_i decline was apparently unaltered in HF versus control myocytes. However, more detailed analysis (see Results) indicated that there might be a modest functional depression of SERCA2 in the HF myocytes (but smaller than the change in NaCaX). Although downregulation of SERCA2 had been reported in several experimental models of HF^{35 36} and in myocardium from patients with HF,^{14 15} other studies have failed to demonstrate this effect.^{37 38 39}

Of course reduced SERCA2 function alone would tend to reduce SR Ca²⁺ content, lower SR Ca²⁺ release, and slow relaxation of twitches. However, the dynamic interplay between the SERCA2 and NaCaX compels one to consider Ca²⁺ balance in a more integrated context.

Mechanical Dysfunction

Diastolic dysfunction usually refers to either a slowing of relaxation or elevation of diastolic pressure. In isolated ventricular myocytes, there was no change in diastolic [Ca²⁺]_i. Neither twitch relaxation nor [Ca²⁺]_i decline were slowed in this model of HF. On one hand, this unaltered relaxation could be simply ascribed to an unaltered expression level of SERCA2. On the other hand, the large and compelling increase in NaCaX function would have been expected to accelerate [Ca²⁺]_i decline if SERCA2 did not change. Thus it is possible that the large increase in NaCaX expression compensates for a small depression of SR Ca²⁺ transport function (which could be hard to detect and could also be due to factors other than protein levels, such as phosphorylation state).

Systolic dysfunction can also result from alterations in Ca²⁺ transport mechanisms. In the HF cells, there was no change in I_Ca, which is expected to be the central trigger for SR Ca²⁺ release. However, there have been reports of depressed response of SR Ca²⁺ release to a given I_Ca trigger in certain models of hypertrophy and HF.^{40 41} The shift in balance between NaCaX and SERCA2 function would also mean that for a given Ca²⁺ transient, there would be greater extrusion from the cell and less refilling of the SR. This would be expected to result in a lower steady state level of SR Ca²⁺ loading, consistent with our results from CafC amplitudes. Although the decreases in CafC and Δ[Ca²⁺]_i in Table 2⇑ were not significant, there was a significant inverse correlation between NaCaX function and CafC amplitude (Figure 11⇑). A lower SR Ca²⁺ load would decrease SR Ca²⁺ release, and this could contribute to the lower twitch contractions and Ca²⁺ transients observed. Indeed, because fractional SR Ca²⁺ release depends steeply on SR Ca²⁺ content, a modest depression of SR Ca²⁺ content could decrease fractional release during E-C coupling to a greater extent.⁴² Although the twitch/CafC ratios did not uncover a significant E-C coupling defect, this issue should be more directly addressed in controlled voltage clamp experiments.^{40 41} Greater extrusion of Ca²⁺ via NaCaX during the twitch could also directly lower the twitch Ca²⁺ transient amplitude.²³

Thus, although numerous possibilities cannot be explicitly ruled out, our preferred working hypothesis is the following: systolic function may be depressed simply because of reduced Ca²⁺ transients that result from reduced SR Ca²⁺ load, which is due to shifts in the balance between NaCaX and SERCA2 (with unaltered I_Ca, intrinsic E-C coupling gain, or myofilament Ca²⁺ sensitivity). Causality is difficult to prove, and it may be that increased NaCaX expression is secondary to other factors in the failing heart, accounting for heterogeneity in NaCaX expression (Figure 5⇑ and Reference 31³¹ ). Thus further tests of this hypothesis will be required.

Spontaneous SR Ca²⁺ Release and Arrhythmogenesis

In the intact animal, the lowered Ca²⁺ transients (and SR Ca²⁺ load) could be partially offset by β-adrenergic activation of the SERCA2. This may be sufficient for the SR to reach an adequate Ca²⁺ load to produce relatively frequent spontaneous SR Ca²⁺ release events (Ca²⁺ sparks or waves).^{43 44} Indeed, we found in HF myocytes that treatment with isoproterenol increased the propensity for spontaneous aftercontractions (Figure 2⇑). Thus β-adrenergic agonists may offset a modest reduction in SR Ca²⁺ load in the myocytes and aftercontractions may be the cellular manifestation of the enhanced arrhythmias in the intact HF animals (Figure 1⇑).⁵

Increased NaCaX in HF could also allow more Ca²⁺ influx, as an outward current early in the action potential, which could even contribute to triggering of SR Ca²⁺ release.^{29 45 46} More importantly, for a given SR Ca²⁺ release during an aftercontraction, the elevated NaCaX activity in HF myocytes would result in greater inward I_Na/Ca, contributing to I_ti, DADs, and ultimately, nonreentrant VT as we see in the whole heart setting using 3D mapping.⁵ The specific role of upregulated NaCaX in the development of I_ti remains to be determined, but our present findings support the hypothesis that an upregulated NaCaX may itself mediate enhanced I_ti and the nonreentrant mechanism in this HF model.

Figure 12⇓ illustrates our working hypothesis based on increased arrhythmias and NaCaX in HF. Local SR Ca²⁺ release elevates local [Ca²⁺]_I, which stimulates Ca²⁺ extrusion via I_Na/Ca. This Ca²⁺ extrusion produces an inward, depolarizing current (I_ti) that can produce a DAD, bringing the diastolic membrane potential closer to threshold to trigger an inappropriately timed action potential. With higher NaCaX activity (as shown in the present study), any given amount of SR Ca²⁺ release will result in greater inward I_ti and increased probability that a triggered arrhythmia will result. It would be of interest to explore whether arrhythmias are more prevalent in the transgenic mouse that overexpresses NaCaX.^{45 47}

• Download figure
• Open in new tab
• Download powerpoint

Figure 12.

Schematic diagram of Ca²⁺ transport in the heart relevant to upregulation of NaCaX and the genesis of I_ti carried by NaCaX. Ca²⁺ influx via I_Ca induces SR Ca²⁺ release via the ryanodine receptor (RyR). Elevated [Ca²⁺]_i activates the myofilaments (MF). During relaxation, Ca²⁺ is transported back to the SR via the SR-Ca ATPase (regulated by phospholamban, PLB) or extruded via NaCaX. With upregulation of NaCaX in HF any spontaneous local SR Ca²⁺ release will produce greater Ca²⁺ efflux (inward current) via NaCaX than in control. This I_ti could cause a larger DAD, leading to triggered activity and, ultimately, nonreentrant initiation of VT in the failing heart.

Relevance of the Model of Nonischemic HF

In the present study, all of the HF rabbits had severe depression of LV function, and 9 of 10 rabbits with serial Holter monitoring demonstrated nonsustained VT. Additionally, we have found that overall, ∼10% of these HF rabbits have sudden death. These findings reflect what is observed in patients with end-stage nonischemic cardiomyopathy where 60% to 80% of patients exhibit nonsustained VT, and up to 40% to 45% will die suddenly.¹ This validates the use of this model to study the cellular and molecular mechanisms underlying arrhythmogenesis in the failing human heart. Indeed, this rabbit HF model seems to combine aspects of both mechanical dysfunction and arrhythmogenic potential that may be especially relevant to the failing human heart.

Implications

The results of the present study suggest that an upregulation of NaCaX expression and functional activity could contribute to both mechanical dysfunction and arrhythmogenesis in the failing heart. Upregulation of NaCaX could have an adaptive role in enhancing Ca²⁺ efflux from myocytes (or possibly increasing Ca²⁺ entry early in the development of HF). However, with the progression of HF, this enhanced NaCaX activity may limit SR Ca²⁺ loading and cellular Ca²⁺ transients and also play a direct role in the nonreentrant mechanisms underlying VT, which have been demonstrated by 3D cardiac mapping. In the setting of enhanced NaCaX activity, approaches to the treatment of HF in patients with drugs that either increase intracellular [Na] (such as digitalis glycosides) or agents that directly enhance NaCaX activity could have proarrhythmic effects in the failing heart that would greatly limit their efficacy.