Improvement in Survival and Cardiac Metabolism After Gene Transfer of Sarcoplasmic Reticulum Ca²⁺-ATPase in a Rat Model of Heart Failure

Construction of Recombinant Adenoviruses

To construct the adenovirus containing SERCA2a cDNA, we used the method described by He et al,¹⁰ whereby the backbone vector, containing most of the adenoviral genome (pAd.EASY1), is used and the recombination is performed in Escherichia coli. SERCA2a cDNA was subcloned into the adenoviral shuttle vector (pAd.TRACK), which uses the cytomegalovirus (CMV) long terminal repeat as a promoter. The shuttle vector used also has a concomitant green fluorescent protein (GFP) under the control of a separate CMV promoter. An adenovirus containing both β-galactosidase and GFP controlled by separate CMV promoters (Ad.βgal-GFP) was used as control. The adenoviruses were propagated in 293 cells. The titers of stocks used for these studies measured by plaque assays were 3 × 10¹¹ pfu/mL for Ad.βgal-GFP and 1.8 × 10¹¹ pfu/mL for Ad.SERCA2a, with particle/pfu ratios of 8:1 and 18:1, respectively. These recombinant adenoviruses were tested for the absence of wild-type virus by polymerase chain reaction of the early transcriptional unit E1.

Experimental Protocol

Four-week-old Sprague-Dawley rats (Charles River, Mass; 70 to 80 g) were anesthetized with pentobarbital (65 mg/kg IP) and placed on a ventilator. A suprasternal incision was made, exposing the aortic root, and a tantalum clip with an ID of 0.58 mm (Weck, Inc) was placed on the ascending aorta. Animals in the sham group underwent a similar procedure without insertion of a clip. The supraclavicular incision was then closed, and the rats were transferred back to their cages. The supraclavicular approach was performed because during gene delivery, a thoracotomy is necessary, and if the thorax is not opened during the initial aortic banding, adhesions are avoided when gene delivery is performed.

The animals were initially divided into 2 groups: 1 group of 45 animals with aortic banding and a second group of 42 animals that were sham-operated. Three animals in the aortic banding group did not survive the initial operation, and 2 animals in the sham-operated group did not survive. In the aortic-banded animals, we waited 26 to 28 weeks for the animals to develop left ventricular (LV) dilatation before cardiac gene transfer. In this last group as well as in the sham-operated group, 14 animals did not undergo gene transfer and were followed longitudinally. The rest of the animals underwent adenoviral gene transfer with either Ad.SERCA2a or Ad.βgal-GFP.

³¹P NMR Measurements

Hearts were retrogradely perfused from a 100-cm hydrostatic perfusion column with modified Krebs-Henseleit buffer (mmol/L: NaCl 116, KCl 4, CaCl₂ 1.5, MgSO₄ 1.2, NaH₂PO₄ 1.2, and NaHCO₃ 25, equilibrated with 95% O₂/5% CO₂ at 37°C) that contained 5 mmol/L glucose in a 2-L reservoir. Hearts beat spontaneously, contracting against a fluid-filled intraventricular balloon connected to a pressure transducer and inflated to an end-diastolic pressure of 5 mm Hg. A 10- to 15-mL volume of coronary effluent bathed the heart. A stable energetic state in rat hearts was confirmed from ³¹P NMR signals of PCr, ATP, and inorganic phosphate (P_i) as previously described.⁸ NMR data were collected on a Bruker 400-MHz spectrometer interfaced to a 9.4-T, vertical-bore, superconducting magnet. ³¹P spectra were obtained from isolated hearts perfused within a broadband, 20-mm NMR probe (Bruker Instruments). ³¹P NMR spectra were acquired in 128 scans with a 161-MHz, 45° excitation pulse, a 1.8-second repetition time, 35 ppm sweep width, and 8000 data points. Peak assignments were referenced to the well-established resonance signal of PCr at 0 ppm, with identification and assignment of the α-, β-, and γ-phosphate signals of ATP. Signal intensity was determined by NMR-dedicated data analysis.

Serial Echocardiographic Assessment

After 18 weeks of banding, serial echocardiograms were performed weekly in lightly anesthetized animals (pentobarbital 40 mg/kg IP). Transthoracic M-mode and 2D echocardiography was performed with a Hewlett-Packard Sonos 5500 imaging system with a 12-MHz broadband transducer. A mid–papillary level LV short-axis view was used, and measurements of posterior wall thickness, LV diastolic dimension, and fractional shortening were collected. Gene transfer was performed in all animals within 3 days of detection of a drop in fractional shortening of >25% compared with the fractional shortening at 18 weeks after banding. In the sham-operated rats, gene delivery was performed at 27 weeks.

Adenoviral Delivery Protocol

The group of animals subjected to aortic banding were further subdivided into 3 additional groups of 16, 12, and 14 receiving Ad.SERCA2a, Ad.βgal-GFP, or no adenovirus, respectively. The group of sham-operated animals was also subdivided into 3 groups of 14, 12, and 14 receiving Ad.SERCA2a, Ad.βgal-GFP, or no adenovirus. The adenoviral delivery system has been described previously by our group in detail.^11–13 Briefly, after the rats had been anesthetized and a thoracotomy performed, a 22-gauge catheter containing 200 mL of adenoviral solution (10¹⁰ pfu) was advanced from the apex of the LV to the aortic root. The aorta and main pulmonary artery were clamped for 20 seconds distal to the site of the catheter, and the solution was injected; then the chest was closed and the animals were allowed to recover.

Measurements of LV Volume and Elastance

Rats in the different treatment groups were anesthetized with 65 mg/kg of pentobarbital and mechanically ventilated. A 1.4F high-fidelity pressure transducer (Millar Instruments) was introduced into the LV. Four 0.7-mm piezoelectric crystals were placed over the surface of the LV along the short axis of the ventricle at the level of the mitral valve and at the apex of the LV to measure the intercrystal distances. The LV volume was derived by use of a mathematical model using Cardiosoft (Sonometrics Co). LV pressure-volume loops were generated under different loading conditions by clamping of the inferior vena cava. The end-systolic pressure-volume relationship was obtained by producing a series of pressure-volume loops and connecting the upper left corners of the individual pressure-volume loops to generate the maximal slope.

Western Blot Analysis and SERCA2a Activity

The preparation of lysates was described earlier. Briefly, lysates were prepared at 4°C in a lysis buffer containing (in mmol/L) NaCl 150, MgCl₂ 1, and CaCl₂ 1, plus detergents and proteinase inhibitors (pH 7.4). The tissue was homogenized and spun at 1400 rpm (Sorvall) for 30 minutes. The supernatant was then filtered through 4 layers of gauze and centrifuged at 15 000 rpm for 60 minutes (Beckman). SDS-PAGE was performed on the supernatant under reducing conditions on 7.5% separation gels with a 4% stacking gel in a Miniprotean II cell (Biorad). For immunoreaction, the blots were incubated with 1:2500 diluted monoclonal antibodies to SERCA2a (MA3-919; Affinity BioReagents), phospholamban (Upstate Biotechnology), or 1:1000 diluted anti-calsequestrin (MA3-913; Affinity Bioreagents) for 90 minutes at room temperature.

Crude membranes were prepared as described by Schwinger et al³ at 4°C in a buffer containing (in mmol/L) sucrose 300, PMSF 1, and PIPES 20, pH 7.4. The tissue was homogenized and spun at 8000 rpm (Beckman JA 20) for 20 minutes. The supernatant was then filtered through 4 layers of gauze and centrifuged at 35 000 rpm for 60 minutes (Sorvall). The pellet was resuspended in a 10% sucrose buffer containing (in mmol/L) KCl 400, MgCl₂ 0.5, CaCl₂ 0.5, EGTA 0.5, and PIPES 25, pH 7.0. SERCA2a activity assays were carried out on the basis of a pyruvate/NADH coupled reaction as previously described.^1,13,14 Ca²⁺-ATPase activity was calculated as Δabsorbance/(6.22 × protein × time) in nmol ATP/(mg protein × min).

Statistics

All values are presented as mean ± SD. A 2-factor ANOVA was performed to compare the different hemodynamic parameters among the different groups. For the echocardiography data, when the variables were examined at various intervals, ANOVA with repeated measures was performed. Comparison of survival in the different groups of animals was analyzed by a log-rank test with the Kaplan-Meier method. Statistical significance was accepted at the level of P < 0.05.

Survival

Figure 1 shows the survival curve for the 6 different groups studied. The sham-operated animals did not show any premature mortality. The sham-operated animals that were infected with either Ad.βgal-GFP or Ad.SERCA2a had early mortalities related to the surgical intervention, but then the survival curves leveled off for both sham + Ad.βgal-GFP and sham + Ad.SERCA2a. In the failing group, the noninfected animals had a survival curve that decreased steadily, and at 4 weeks the survival rate was only 18% (P < 0.0005 compared with sham). In the failing + Ad.βgal-GFP group, the survival curve also decreased, and at 4 weeks the survival rate was only 9% (P < 0.001 compared with sham + Ad.βgal-GFP). In the failing + Ad.SERCA2a group, however, the survival curve was significantly improved compared with failing + Ad.SERCA2a (P < 0.001 compared with failing + Ad.βgal-GFP).

Characterization of Animals

After 18 weeks of aortic banding, the animals showed echocardiographic signs of LV hypertrophy, including an increase in wall thickness (both posterior and septal), an increase in posterior wall thickness, a decrease in LV dimensions, and an increase in fractional shortening, as shown in Table 1. After 26 to 27 weeks of banding, these animals had uniformly (1) small pericardial effusions, (2) pleural effusions, (3) an increase in lung weight, (4) ascites, and (5) dyspnea at rest, all indicative signs of severe heart failure. Echocardiographically, LV end-diastolic dimensions increased and fractional shortening decreased.

Cardiac Gene Transfer and SERCA2a Expression

Protein levels of SERCA2a were decreased in failing compared with sham LVs, as shown in Figure 2A. Adenoviral gene transfer of SERCA2a in failing hearts increased SERCA2a protein expression, restoring it to levels observed in the nonfailing hearts. Calsequestrin did not change among the different groups, nor did phospholamban. As shown in Figure 2B, tabulated ratios of SERCA2a to phospholamban and SERCA2a to calsequestrin reveal a significant decrease in failing hearts and a restoration to control levels with gene transfer of SERCA2a.

SERCA2a Activity

We measured SERCA2a activity at a calcium concentration of 10 mmol/L in the (1) sham + Ad.βgal-GFP, (2) failing + Ad.βgal-GFP, and (3) failing + Ad.SERCA2a groups. There was a decrease in maximal ATPase activity in the failing group (27.4 ± 4.9 versus 62.2 ± 12.8 nmol · mg⁻¹ · min⁻¹). Gene transfer of SERCA2a restored ATPase activity back to normal levels in the failing group 4 weeks after gene transfer (61.0 ± 8.5 nmol · mg⁻¹ · min⁻¹).

NMR Spectroscopy

Representative ³¹P NMR spectra obtained from 3 groups of rats: (1) sham + Ad.βgal-GFP, (2) failing + Ad.βgal-GFP, and (3) failing + Ad.SERCA2a, are shown in Figure 3A. These spectra show that the ratios of total amounts of PCr to ATP are lower in the failing heart than the sham heart. The integrated area for P_i was also increased in the failing heart. The overexpression of SERCA2a in failing heart restored and normalized the content of both PCr and ATP (Figure 3B). Interestingly, we found that overexpression of SERCA2a in sham-operated animals induces a reduction in PCr/ATP ratio.

Effects of SERCA2a Overexpression on Pressure-Volume Relationship

Pressure-volume analysis was performed in a subset of animals. LV volumes were significantly increased in the failing rats (0.64 ± 0.05 versus 0.35 ± 0.03 mL, P < 0.02) and were decreased after SERCA2a gene transfer (0.46 ± 0.07 mL). The slope of the end-systolic pressure-volume relationship (Figure 4) was lower in failing hearts infected with Ad.βgal-GFP (n = 5) than in sham (n = 6), indicating a diminished state of intrinsic myocardial contractility: 450 ± 71 versus 718 ± 83 mm Hg/mL (P < 0.02). Gene transfer of SERCA2a restored the slope of the end-systolic pressure-volume relationship to control levels (691 ± 91 mm Hg/mL, n = 6, P < 0.03 versus failing + Ad.βgal-GFP; P < 0.1 versus sham + Ad.βgal-GFP).

Effect on Morphological Parameters

As shown in Table 2, the failing hearts had a significant increase in heart weight when normalized to either tibial length or body mass. Gene transfer of SERCA2a in the failing heart did not have a significant effect on LV mass whether normalized to tibial length or body mass.

Abnormal SR Function and Cardiac Gene Transfer of SERCA2a

Impaired SERCA2a activity is one of the main characteristics associated with abnormal calcium handling in heart failure.^1,2 A decrease in SERCA2a relative to phospholamban and a reduction of phosphorylation of phospholamban contribute to the contractile dysfunction in human heart failure.^3,4 More recently, the ratio of the Na/Ca exchanger to SERCA2a has been shown to be increased in failing hearts and to be predictive of diastolic function in these hearts.¹⁵ Gene transfer of SERCA2a corrects both the SERCA2a/phospholamban ratio and the SERCA2a/Na/Ca ratio and would contribute to restoring both systolic and diastolic function.

In our study, we showed that SERCA2a protein levels were restored to normal levels in the failing hearts and that this effect was sustained for up to 4 weeks. This seemed somewhat surprising, because first-generation adenoviruses induce transient expression peaking at 7 to 10 days and disappearing after 10 days.¹⁶ Endogenous turnover of SERCA2a, however, is ≈14 to 15 days in young rats and longer in older rats,¹⁷ which would explain the sustained levels of SERCA2a.

SERCA2a Expression and Cardiac Energetics

Decreased energy reserve via the creatine kinase reaction is a characteristic finding in both human and experimental heart failure.^9,18,19 This decrease in energy reserve contributes to the development of contractile dysfunction in heart failure.¹⁹ In addition, an increase in intracellular P_i has been shown to reduce SR calcium loading and to depress calcium-induced calcium release.²⁰ Local ATP regeneration by the creatine kinase system is one mechanism the cell can use to improve Ca²⁺ uptake by the SR in conditions in which an excessive increase in cytoplasmic [Ca²⁺] may have deleterious effects. Recently, Tian et al⁹ showed that pharmacological inhibition of creatine kinase resulted in altered energetics and induced abnormal Ca²⁺ handling and contractile dysfunction in the rat. In our experiments, restoring SERCA2a levels to normal induced an improvement in the ratio of PCr to ATP.

The findings of improved cardiac energetics in heart failure was somewhat surprising, because an increase in contractility by SERCA2a overexpression would be anticipated to increase ATP hydrolysis, thereby driving PCr down. Indeed, this increase in ATP hydrolysis is consistent with our observation of reduced PCr/ATP in the group of sham-operated hearts that were overexpressing SERCA2a. These results are also consistent with previous results showing that PCr/ATP was decreased in the phospholamban-deficient hearts relative to the wild-type hearts.²¹

In heart failure, however, elevated calcium levels would increase energy demand. To maintain low levels of diastolic Ca²⁺, a high level of free energy released from ATP hydrolysis (|ΔGp|) is necessary. To maintain the normal Ca²⁺ gradient (≈10 000-fold between the cytosol and the SR), the SERCA2a reaction requires a |ΔGp| of ≥52 kJ/mol, 85% to 90% of it from ATP.⁹ Therefore, of all the ATPase reactions in cardiac myocytes, the SERCA2a reaction is the most vulnerable to a decrease in |ΔGp|.

Survival After Gene Transfer: Therapeutic Implications

In this model of heart failure, SERCA2a overexpression improved parameters of inotropy and normalized contractile reserve. These effects translate into an inotropic intervention. Other inotropic interventions, however, have been shown clinically to increase mortality.²² There are, however, significant differences between enhancing inotropy with pharmacological agents that usually increase cAMP and gene transfer of SERCA2a. Unlike agents that increase cAMP, thereby increasing intracellular Ca²⁺, restoration of SERCA2a levels decreases diastolic Ca²⁺. Furthermore, it has been shown that sustained elevations of resting Ca²⁺ lead to activation of serine-threonine phosphatases, including calcineurin, inducing hypertrophy and cell death.²³ Therefore, a decrease in diastolic Ca²⁺ may in effect reduce the proapoptotic and prohypertrophy signaling. Heart failure is associated with an increased incidence of ventricular arrhythmias, and triggered activity is a probable mechanism of arrhythmogenesis in heart failure. The increase in intracellular calcium secondary to SERCA2a downregulation increases the arrhythmogenic potential. Preventing an increase in intracellular calcium by overexpression of SERCA2a prevents the induction of triggered activity. Furthermore, improvement in energetics is another important finding in this study that may have a direct influence on survival.

Conclusions

Our results demonstrate that restoring SERCA2a expression can improve not only systolic and diastolic performance in failing hearts but also survival and cardiac energetics. Furthermore, SERCA2a normalization halts the adverse remodeling that occurs with congestive heart failure. This study validates the feasibility of cardiac gene transfer in failing hearts as a therapeutic modality.