Origin of Contractile Dysfunction in Heart Failure
Calcium Cycling Versus Myofilaments
Background—Chronic congestive heart failure is a common, often lethal disorder of cardiac contractility. The fundamental pathophysiology of the contractile failure remains unclear, the focus being on abnormal Ca2+ cycling despite emerging evidence for depressed myofilament function.
Methods and Results—We measured intracellular Ca2+ concentration ([Ca2+]i) and contractile force in intact ventricular muscle from SHHF rats with spontaneous heart failure and from age-matched controls. At physiological concentrations of extracellular Ca2+ ([Ca2+]o), [Ca2+]i transients were equal in amplitude in the 2 groups, but [Ca2+]i peaked later in SHHF muscles. Twitch force peaked slowly and was equivalent or modestly decreased in amplitude relative to controls. Steady-state analysis revealed a much greater (53%) depression of maximal Ca2+-activated force in SHHF muscles, which, had other factors been equal, would have produced an equivalent suppression of twitch force. Phase-plane analysis reveals that the slowing of Ca2+ cycling prolongs the time available for Ca2+ to activate the myofilaments in failing muscle, partially compensating for the marked dysfunction of the contractile machinery.
Conclusions—Our results indicate that myofilament activation is severely blunted in heart failure, but concomitant changes in [Ca2+]i kinetics minimize the contractile depression. These results challenge prevailing concepts regarding the pathophysiology of heart failure: the myofilaments emerge as central players, whereas changes in Ca2+ cycling are reinterpreted as compensatory rather than causative.
Heart failure is a systemic illness in which cardiac output is diminished. Excitation-contraction coupling is abnormal in human heart failure: diastolic intracellular [Ca2+] ([Ca2+]i) is elevated, whereas [Ca2+]i transients are blunted and slow to decay.1 The function of the myofilaments has been argued not to be much altered,2 although carefully controlled steady-state analysis has revealed marked depression of the maximal force-generating capacity in a rat ischemic injury model of heart failure.3 Multiple biochemical alterations of the contractile proteins have been described, including switching of the isoforms of troponin T,4 suppression of α-myosin heavy chain expression,5 and decreased basal cAMP-dependent phosphorylation.6 The realization that heart failure increases the likelihood of myocardial apoptosis7 gives us further reason to wonder whether depression of the contractile apparatus (by loss of functional units) might play a major unanticipated role.
To assess the relative importance of abnormal Ca2+ cycling versus depressed myofilaments in heart failure, we measured [Ca2+]i and force simultaneously in intact ventricular muscle from rats with spontaneous heart failure. These SHHF/Mccfacp (spontaneous hypertension and heart failure) rats recapitulate many of the features of the human disease, including systemic vascular congestion, ventricular dilatation, and decreased ejection fraction.8 9 10 We find that myofilament force-generating capacity is strongly depressed in heart failure, far out of proportion to the observed depression of force during physiological twitch contractions. The salient abnormality of Ca2+ cycling turns out to be slowing of the [Ca2+]i transients, including a 3-fold increase in the time to peak [Ca2+]i. Phase-plane analysis reveals that [Ca2+]i comes closer to equilibrium with the myofilaments in the SHHF rats than in age-matched controls. The kinetic changes in [Ca2+]i can thus be construed as adaptive, insofar as they minimize the contractile failure, which would otherwise track the marked depression of maximal force production. This integrative analysis of excitation-contraction coupling in failing heart muscle highlights the importance of the myofilaments, necessitating a fundamental revision of prevailing concepts regarding the pathophysiology of heart failure.
In Vivo Characterization of Rats
Lean male SHHF rats were bred and housed at The Ohio State University. Animals were chosen for study when they exhibited clinical signs of heart failure (tachypnea, dyspnea, cyanosis, peripheral edema). Age-matched (17 to 18 months) male Wistar-Furth rats (Harlan, Indianapolis, Ind) were purchased at ≈9 months of age and housed under conditions identical to those for the SHHF rats until use. M-mode echocardiography was performed under light ketamine/xylazine anesthesia, as described previously.9 Afterward, rats were transported to Baltimore, Md, and used within 2 weeks for studies of isolated trabeculae.
Rats were anesthetized by intra-abdominal injection of sodium pentobarbital 0.1 to 0.2 mL, and the hearts were quickly removed and perfused retrogradely with a high-potassium (20 mmol/L) Krebs-Henseleit (K-H) solution equilibrated with a mixture of 95% O2/5% CO2 at room temperature (21°C to 22°C). The K-H solution was composed of (in mmol/L) NaCl 120, NaHCO3 20, KCl 20, MgSO4 1.2, glucose 10, and CaCl2 0.5, pH 7.35 to 7.40. Trabeculae from the right ventricles of these hearts were dissected and mounted between a force transducer and a micromanipulator in a perfusion chamber placed on the stage of an inverted microscope according to methods already described.11 Not every heart yields geometrically suitable (long, thin, nonbranching) trabeculae: 6 of 11 control hearts and 5 of 6 SHHF hearts did so. The dimensions of the trabeculae were not significantly different in the 2 groups (in mm: control, 3.25±0.10 long, 0.20±0.02 wide, and 0.10±0.01 thick [n=6]; SHHF, 3.14±0.25 long, 0.17±0.04 wide, and 0.11±0.02 thick [n=5]). The cross-sectional area was calculated by multiplying thickness and width and was corrected by a factor of 0.75, assuming an ellipsoidal shape, and a reduction in thickness of ≈5% because of the stretch to a sarcomere length of ≈2.2 μm. All muscles were superfused with a K-H solution of the same composition as the one described above, but in this case with normal [KCl] (5 mmol/L), at a flow rate of ≈10 mL/min, and stimulated at 0.5 Hz. All experiments were performed at room temperature. Force was measured as described previously11 by a silicon strain gauge (model AEM 801, SensoNor) and expressed as mN/mm2 of cross-sectional area. All experiments were carried out at the length at which the muscles developed maximal twitch force (end-diastolic sarcomere length of 2.2 to 2.3 μm).
Measurement of Intracellular [Ca2+] in Trabeculae
[Ca2+]i was measured with fura 2 pentapotassium salt, according to the method described by Backx and ter Keurs.12 Briefly, after 40 to 60 minutes of stabilization at 0.5-Hz stimulation frequency, pacing was stopped, and fura 2 pentapotassium salt was microinjected iontophoretically into 1 cell and allowed to spread throughout the muscle via gap junctions. After fura 2 loading, stimulation was resumed, and [Ca2+]i was determined by epifluorescent illumination at 380 and 340 nm. The fluorescence was collected at 510 nm by a photomultiplier tube (R1527, Hamamatsu). The output of the photomultiplier was filtered at 100 Hz, collected by an analog-to-digital converter, and stored in a computer for later analysis. [Ca2+]i was calculated by the following equation, after subtraction of the corresponding background fluorescence of the trabeculae: [Ca2+]i=K′d(R−Rmin)/(Rmax−R), where R is the observed ratio of fluorescence (340/380), K′d is the apparent dissociation constant, Rmax is the ratio 340/380 nm at saturating [Ca2+], and Rmin is the ratio 340/380 nm at zero [Ca2+]. The values of K′d, Rmax, and Rmin were 2.95, 9.55, and 0.50, respectively, as determined by in vivo calibrations in the muscles.13 14 15 The apparent Kd is the result of multiplying the true Kd of fura 2 by a correction factor obtained from the ratio of fluorescence of the Ca2+-free to Ca2+-bound forms of fura 2 at 380 nm (Sf2/Sb2); for more details see Reference 1616 .
To characterize excitation-contraction coupling, muscles were subjected to the following conventional experimental protocols.14 15 We first studied the response to extracellular [Ca2+] ([Ca2+]o) (0.5, 1.0, 1.5, and 2.0 mmol/L) during twitch contractions elicited by field stimulation (pulse duration, 5 ms) at a rate of 0.5 Hz. Thereafter, the muscles were exposed to 5 μmol/L ryanodine for 30 minutes and stimulated periodically (≈1 minute−1) at 10 Hz to elicit tetani of 4- to 5-second duration. By varying extracellular [Ca2+], different levels of steady-state activation were achieved during tetani until maximal force was reached.
Steady-state force-[Ca2+]i relations were fit with a function of the following form (Hill equation): F=Fmax[Ca2+]n/(Ca50n+[Ca2+]n), where Fmax is the maximal Ca2+-activated force, Ca50 is the [Ca2+]i required for 50% of maximal activation, and n is the Hill coefficient.14 15
Student’s t test was used for simple comparisons. Two-way ANOVA was used to analyze differences between the SHHF and Wistar-Furth rats. ANOVA with repeated measures was used to analyze [Ca2+]i and force data for trabeculae superfused at varying [Ca2+]o. Appropriate post hoc tests were then applied where indicated. Data are expressed as mean±SEM.
We studied 17- to 18-month-old SHHF rats, just after the animals had developed overt heart failure. This was confirmed by echocardiography in the living animals. Figure 1⇓ shows pooled data for ejection fraction from echocardiography; this was significantly lower in the heart failure group than in age-matched controls. To further characterize the clinical phenotype of these failing rats, we determined the ratios of heart weight to body weight and lung weight to body weight; both were increased in the heart failure group (Figure 1⇓), consistent with previous findings (eg, see Reference 1010 ).
Force-[Ca2+]i Relationships During Twitch Contractions in Control Versus Failing Trabeculae
When the trabeculae were homogeneously loaded with fura 2, we measured [Ca2+]i transients and the corresponding force development at various different [Ca2+]os. Figure 2⇓ shows [Ca2+]i transients and force in typical experiments from the control (left) and failing (right) groups in 1 mmol/L (A and B) or 1.5 mmol/L [Ca2+]o (C and D). The amplitudes of [Ca2+]i transients were similar in both muscles. At the lower [Ca2+]o (1 mmol/L, A and B), the forces developed by the 2 muscles were also comparable, but a divergence in force development is evident at higher [Ca2+]o (1.5 mmol/L, C and D). The pooled data in Figure 3⇓ confirm that these results were typical. At more physiological [Ca2+]o (0.5 to 1 mmol/L), [Ca2+]i transient amplitude and developed force were not significantly different in the 2 groups. Thus, a latent abnormality of contractile activation distal to Ca2+ cycling becomes apparent at higher [Ca2+]o. Nevertheless, the results are notable for the overall similarities in Ca2+ transient amplitude and for the striking preservation of force development at physiological [Ca2+]o in the failing muscles.
The [Ca2+]i transient amplitudes in the 2 groups are not superimposed on major differences in diastolic [Ca2+]i. Figure 4⇓ shows pooled data for end-diastolic [Ca2+]i (top) and end-diastolic force (bottom). Although there was a tendency for end-diastolic [Ca2+]i to be higher in the heart failure group at any given [Ca2+]o, the differences were not significant.
Time Courses of [Ca2+]i Transients and Twitch Force
Despite the similarity in the values of [Ca2+]i achieved during twitch contractions in the 2 groups, we do find marked differences in the time courses of both [Ca2+]i transients and force. Such differences are apparent even from a casual inspection of the results in Figure 2⇑: the times to peak [Ca2+]i and peak force are prolonged in the failing muscle, as are the decay phases. These differences were consistent and significant. Figure 5⇓ shows pooled data for time to peak (left panels) and time from peak to 50% relaxation (right panel) of [Ca2+]i transients (top panels) and twitch force (bottom panels). At any given [Ca2+]o, the times to peak [Ca2+]i and peak force were prolonged in the failing group. The time to 50% decay of the [Ca2+]i transient was also prolonged, but the increase in the time to 50% decay of twitch force did not reach statistical significance. The physiological importance of the striking prolongation of the rising phase of [Ca2+]i transients will be considered later.
Steady-State Force-[Ca2+]i Relationship in Control Versus Failing Trabeculae
To determine whether myofilament Ca2+ sensitivity and/or maximal Ca2+-activated force is decreased, we achieved steady-state activation by tetanizing muscles in the 2 experimental groups.13 17 Figure 6⇓ shows pooled steady-state [Ca2+]i-force relationships derived from all muscles in the control and failing groups. Data were normalized to the maximal force in each muscle, then expressed with respect to the averaged maximal force in each of the 2 experimental groups.13 15 17 The steady-state force-[Ca2+]i relationship revealed a lower maximal Ca2+-activated force in the failing group (128±6 versus 61±7 mN/mm2, P<0.0001). The 2 curves overlap at low [Ca2+]i but diverge as [Ca2+]i rises; as a consequence, the [Ca2+]i required to activate 50% of the maximal force is lower in the failing muscles (Kd, 0.56±0.017 versus 0.42±0.02 μmol/L, P<0.001), as is the Hill coefficient (n, 4.2±0.4 versus 2.3±0.3, P<0.02). Although both are significant, the 53% decrease of Fmax outweighs the modest (25%) increase in Ca2+ sensitivity; a further depressant effect arises from the decrease in myofilament cooperativity, which is reflected by the drop in the Hill coefficient. In general terms, the depression of Fmax (by 53% relative to control) is unmatched by any decrease of twitch force at physiological [Ca2+]o, and the extent of the depression far exceeds the relatively modest dysfunction during twitches at higher [Ca2+]o. To investigate the basis for these discrepancies, we performed phase-plane analysis.
The fact that [Ca2+]i transients are slowed in SHHF muscles (Figure 5⇑) might offset the marked depression of Fmax by providing more time for Ca2+ to interact with the contractile proteins during each cardiac cycle. We tested this idea by plotting [Ca2+]i versus force on a point-by-point basis during representative twitch contractions, then superimposing the resultant phase-plane loop on the steady-state [Ca2+]i-force relationship obtained from the same muscle (Figure 7⇓). Normally, the rise of [Ca2+]i transients is far too rapid to allow [Ca2+]i to reach equilibrium with the contractile proteins11 ; this is evident in Figure 7⇓ as the discrepancy between the peak [Ca2+]i in each loop and the [Ca2+]i that produces the same force at steady state in the same muscle. This discrepancy of [Ca2+]i (Δ[Ca2+]i), highlighted graphically in both panels of Figure 7⇓, is much greater in the control example (left panel) than in the SHHF (right panel). On average, Δ[Ca2+]i was 5 times greater in controls than in SHHF (0.20±0.07 versus 0.04±0.02 μmol/L, P<0.05). This analysis confirms that the phase lag between [Ca2+]i and force is much smaller in SHHF, giving Ca2+ more time for interaction with the myofilaments during each cardiac cycle.
This study is the first to measure [Ca2+]i and force simultaneously in intact muscles from a well-controlled model of spontaneous congestive heart failure. The results indicate that although Ca2+ cycling is not normal, the dysfunction of the myofilaments turns out to predominate quantitatively. Moreover, the Ca2+ cycling changes can be construed as adaptive, insofar as they largely negate the profound myofilament depression at low stimulation frequencies and physiological [Ca2+]o.
Comparison With Previous Studies
Many previous studies of normal and failing cardiac muscle used relatively thick preparations (eg, rat papillary18 or human trabecular muscle2 ), in which core ischemia can be problematic. These also tended to rely on aequorin loading of surface myocytes, which may not be representative of the entire cell population. Our data are therefore the first that enable a reasonably quantitative dissection of the relative importance of Ca2+ cycling versus myofilament abnormalities in a single heart failure model. In agreement with the present study, Bing et al,18 using electrically stimulated aequorin-loaded papillary muscles from aged spontaneously hypertensive rats with heart failure, noted depressed force development in the setting of unaltered peak [Ca2+]i. Their studies, however, failed to detect an altered time to peak tension in the failing preparations: time to peak in both control and failing muscles was quite prolonged (ie, ≈200 ms). Conversely, studies of human trabeculae from transplant recipients failed to detect a decline in maximal force production.2 With human studies, an additional difficulty can be the severely limited availability of truly normal control subjects.
The conclusions of the present study differ from those of Gómez et al,9 who used the same SHHF heart failure model to investigate Ca2+ sparks in voltage-clamped ventricular myocytes. These investigators argued that coupling between sarcolemmal L-type Ca2+ channels and the Ca2+ release channels of the sarcoplasmic reticulum (SR) is a primary cause of altered contractile performance in hypertrophy and failure. In their experiments, [Ca2+]i transients and isotonic shortening were markedly blunted in failing SHHF myocytes, but the relation between [Ca2+]i and fractional shortening was not altered relative to that in cells from age-matched controls. There are several reasons for the apparent discrepancies. In the voltage-clamp experiments, the duration of depolarization was controlled (200 ms) and was identical for the normal and failing myocytes. The present study, however, used field stimulation. Previous work has shown that the action potential is markedly prolonged in cells from failing SHHF hearts,19 and action potential duration has been shown to regulate the Ca2+ content of the SR.20 In addition, changes in Ca2+ release from the SR tend to be offset by compensatory changes in Ca2+ filling of the SR in physiologically contracting myocytes.21 Thus, the more normal [Ca2+]i transient amplitudes in SHHF trabeculae in the present study probably reflect the longer action potentials in the SHHF versus control trabeculae as well as other homeostatic mechanisms that regulate the Ca2+ content of the SR.
The present study also differs from that of Gómez et al9 in that they measured unloaded cell shortening, whereas this study measured genuine contractile force. Force per cross-sectional area is exquisitely sensitive to the number of contractile units, whereas isotonic shortening is relatively insensitive to this parameter. In this regard, it should be noted that histological and ultrastructural studies of failing SHHF ventricles have noted marked interstitial and perivascular fibrosis, a reduction in myofibrils, myofibrillar disorganization, and streaking of Z bands.10 These ultrastructural changes could well lead to a decrease in functional force-producing units and account for at least a part of the decline in maximal Ca2+-dependent force observed in the present study. Nevertheless, the fact that single skinned myocytes from a rat pressure-overload model of heart failure also exhibit a marked reduction of maximal Ca2+-activated force22 hints that the predominant change resides within the myofilaments themselves, not in extracellular factors or myocyte loss.
A decline in maximal Ca2+-activated force production similar to that described in the present study has been observed in skinned trabeculae from a postinfarct rat model of heart failure3 and from failing SHHF hearts (P.J. Reiser, MD et al, oral personal communication). Although 85% to 100% of the myosin in the failing SHHF heart is the slow, V3 (β-myosin heavy chain) subtype,23 this isoform switch should contribute only to the slowed rate of force development, not to the depressed maximal Ca2+-activated force.
Limitations of the Study
In interpreting the results presented here, the following limitations should be borne in mind. First of all, we did not measure myofilament properties in single skinned cells. Such measurements are desirable in the SHHF model to establish conclusively whether the depressed maximal force represents myofibrillar depression or a simple decrease in the number of myofilaments per cross-sectional area (eg, due to loss of functional units and replacement fibrosis). The single-cell results showing unequivocal depression of maximal force in other heart failure models provides some reassurance that the myofibrils themselves are depressed; in any case, the changes in Kd and Hill coefficient that we have observed cannot be rationalized by nonmyofibrillar alterations. Second, we examined only the terminal stage of heart failure; thus, our results may shed light on the mechanism of the dysfunction, but they do not tell us how the heart failure developed in the first place. For such insights, further studies in the premorbid and in the hypertrophic, prefailure stages of this model may prove to be useful. Third, all our experiments have been performed at a stimulation frequency of 0.5 Hz. It would be of interest in the future to examine the force-frequency relation, given the evidence that this is altered in heart failure.24 25
Relationship to Reversible, Postischemic Contractile Failure
The changes of myofilament Ca2+ responsiveness revealed in this study are reminiscent of those previously described in postischemic “stunned” myocardium.14 15 26 This reversible form of contractile failure occurs immediately, without compensatory changes in Ca2+ cycling. Thus, the profound depression of maximal force (typically 50% to 60%, comparable to that seen here) is matched by an equally profound depression of twitch contractions. In chronic heart failure, Ca2+ transients become slower and somewhat blunted in amplitude (particularly at higher [Ca2+]o). The slowing of calcium transients may reflect broadening of the subspace surrounding the SR Ca2+ release sites, resulting in reduced coupling between L-type Ca2+ channels and ryanodine receptors9 ; however, whatever the underlying mechanism, the changes of Ca2+ cycling will tend to mitigate the functional depression during physiological excitation-contraction coupling. The kinetic changes enable [Ca2+]i to approach equilibrium with the contractile proteins during each cardiac cycle; therefore, a lower Ca2+ transient amplitude suffices to achieve equivalent (or greater) fractional activation of the myofilaments. Nevertheless, these “adaptive” changes of Ca2+ homeostasis exact their own toll, impairing relaxation and predisposing to diastolic dysfunction.26
This study was supported by NIH grants RO1-HL-44065 (to Dr Marbán) and RO1-HL-48835 (to Dr Altschuld) and by a fellowship from the CONICET of Argentina (to Dr Pérez).
- Received August 17, 1998.
- Revision received September 30, 1998.
- Accepted October 9, 1998.
- Copyright © 1999 by American Heart Association
Beuckelmann DJ, Näbauer M, Erdmann E. Intracellular calcium handling in isolated ventricular myocytes from patients with terminal heart failure. Circulation. 1992;85:1046–1055.
Gwathmey JK, Copelas L, MacKinnon R, Schoen FJ, Feldman MD, Grossman W, Morgan JP. Abnormal intracellular calcium handling in myocardium from patients with end-stage heart failure. Circ Res. 1987;61:70–76.
de Tombe PP, Wannenburg T, Fan DS, Little WC. Right ventricular contractile protein function in rats with left ventricular myocardial infarction. Am J Physiol. 1996;271(Heart Circ Physiol 40):H73–H79.
Anderson PAW, Malouf NN, Oakeley AE, Pagani ED, Allen PD. Tropin T isoform expression in humans: a comparison among normal and failing adult heart, fetal heart, and adult and fetal skeletal muscle. Circ Res. 1991;69:1226–1233.
Li Q, Li BS, Wang XW, Leri A, Jana KP, Liu Y, Kajstura J, Baserga R, Anversa P. Overexpression of insulin-like growth factor-1 in mice protects from myocyte death after infarction, attenuating ventricular dilation, wall stress, and cardiac hypertrophy. J Clin Invest. 1997;100:1991–1999.
Gerdes AM, Onodera T, Wang XJ, McCune SA. Myocyte remodeling during the progression to failure in rats with hypertension. Hypertension. 1996;28:609–614.
Gómez AM, Valdivia HH, Cheng H, Lederer MR, Santana LF, Cannell MB, McCune SA, Altschuld RA, Lederer WJ. Defective excitation-contraction coupling in experimental cardiac hypertrophy and heart failure. Science. 1997;276:800–806.
Backx PH, Gao WD, Azan-Backx MD, Marban E. The relationship between contractile force and intracellular [Ca2+] in intact rat cardiac trabeculae. J Gen Physiol. 1995;105:1–19.
Backx PH, ter Keurs HE. Fluorescent properties of rat cardiac trabeculae microinjected with fura-2 salt. Am J Physiol. 1993;264:H1098–H1110.
Gao WD, Backx PH, Azan-Backx M, Marban E. Myofilament Ca2+ sensitivity in intact versus skinned rat ventricular muscle. Circ Res. 1994;74:408–415.
Gao WD, Liu Y, Marban E. Selective effects of oxygen free radicals on excitation-contraction coupling in ventricular muscle: implications for the mechanism of stunned myocardium. Circulation. 1996;94:2597–2604.
Gao WD, Atar D, Backx PH, Marban E. Relationship between intracellular calcium and contractile force in stunned myocardium: direct evidence for decreased myofilament Ca2+ responsiveness and altered diastolic function in intact ventricular muscle. Circ Res. 1995;76:1036–1048.
Grynkiewicz G, Poenie M, Tsien RY. A new generation of Ca2+ indicators with greatly improved fluorescence properties. J Biol Chem. 1985;260:3440–3450.
Marban E, Kusuoka H, Yue DT, Weisfeldt ML, Wier WG. Maximal Ca2+-activated force elicited by tetanization of ferret papillary muscle and whole heart: mechanism and characteristics of steady contractile activation in intact myocardium. Circ Res. 1986;59:262–269.
Bing OH, Brooks WW, Conrad CH, Sen S, Perreault CL, Morgan JP. Intracellular calcium transients in myocardium from spontaneously hypertensive rats during the transition to heart failure. Circ Res. 1991;68:1390–1400.
Groden DL, Guan Z, Biagi BA, Altschuld RA, Stokes BT. Calcium transients and electrophysiology in failing rat myocytes. Circulation. 1989;80(suppl II):II-504.
Eisner DA, Trafford AW, Díaz ME, Overend CL, O’Neill SC. The control of Ca release from the cardiac sarcoplasmic reticulum: regulation versus autoregulation. Cardiovas Res. 1998;38:589–604.
Fan DS, Wannenburg T, de Tombe PP. Decreased myocyte tension development and calcium responsiveness in rat right ventricular pressure overload. Circulation. 1997;95:2312–2317.
McCune SA, Park S, Radin MJ, Jurin RR. SHHF/Mcc-facp rat model: a genetic model of congestive heart failure. In: Singal PK, Dixon IMC, Reamish RE, Dhalla NS, eds. Mechanisms of Heart Failure. Dordrecht, Netherlands: Kluwer Academic Publishers; 1995:91–106.