Heart failure - a challenge to our current concepts of excitation-contraction coupling

Abstract

Development of novel therapeutic strategies for congestive heart failure (CHF) seems to be hampered by insufficient knowledge of the molecular machinery of excitation-contraction (EC) coupling in both normal and failing hearts. Cardiac hypertrophy and failure represent a multitude of cardiac phenotypes, and available invasive and non-invasive techniques, briefly reviewed here, allow proper quantification of myocardial function in experimental models even in rats and mice. Both reduced fractional shortening and reduced velocity of contraction characterize myocardial failure. Only when myocardial function is depressed in vivo can meaningful studies be done in vitro of contractility and EC coupling. Also, we point out potential limitations with the whole cell patch clamp technique. Two main factors stand out as explanations for myocardial failure. First, a basic feature of CHF seems to be a reduced Ca²⁺ load of the sarcoplasmic reticulum (SR) mainly due to a low phosphorylation level of phospholamban. Second, there seems to be a defect of the trigger mechanism of Ca²⁺ release from the SR. We argue that this defect only becomes manifest in the presence of reduced Ca²⁺ reuptake capacity of the SR and that it may not be solely attributable to reduced gain of the Ca²⁺-induced Ca²⁺ release (CICR). We list several possible explanations for this defect that represent important avenues for future research.

Congestive heart failure (CHF) is an important cause of mortality in Western countries (Ito et al. 2000). It is also a disabling condition. More than 200 years ago Withering systematically described alleviating effects of digitalis on the symptoms of CHF (Withering, 1785). Although digitalis to a large extent has been replaced by other drugs, alternative therapeutic principles are few and even fewer of them target the underlying pathophysiological mechanism directly (Hoshijima & Chien, 2002). Two hallmarks of CHF are cardiac hypertrophy and reduced contractility or function of viable myocardium. Thus the disease seems to directly affect two basic biological processes, control of cardiac growth and the signalling cascade that allows the action potential to trigger contraction (excitation-contraction coupling, or EC coupling). Interestingly, current concepts of these processes in the healthy individual do not comprise enough information to allow us to understand the pathological processes. On the contrary, recent insight into the pathology has in fact provided new understanding of normal biological mechanisms. For example activation of the Ca²⁺-calmodulin-activated phosphatase calcineurin was first associated with pathological cardiac hypertrophy, but is now recognized as a possible regulator of normal cardiac growth (Wilkins & Molkentin, 2002).

In this review we will therefore focus on the pathology of EC coupling in CHF since in our opinion this literature reveals that our understanding of how the heart beat is normally triggered may be incomplete. We will also point out some methodological restrictions with the patch clamp technique that may have concealed important properties of the EC coupling. More complete reviews of the mechanisms of heart failure have previously been published in entire journal issues (Cardiovascular Research 37(2), 277-548, 1998 and Basic Research in Cardiology 97 (suppl. 1), I/1-I/158, 2002).

Our current concept of the EC coupling in the heart, as recently reviewed by Bers (2002), has developed gradually since Ringer first discovered the dependency of the beating heart on extracellular Ca²⁺ (1882). Ebashi (1976) and Weber & Murray (1973) initially described the importance of the sarcoplasmic reticulum (SR) in skeletal muscle, and Fabiato (see for instance Fabiato, 1985) introduced the concept of CICR in the heart. Stimulated by the work of Reuter & Seitz (1968), the important relationship between intracellular Na⁺ and Ca²⁺ became clear, and it was possible to ascribe the effect of digitalis to inhibition of the Na⁺-K⁺ pump which had been discovered by Skou (1957).

The size and duration of the cytosolic Ca²⁺ transient are important determinants of the velocity, size and duration of cardiomyocyte shortening. The key elements of EC coupling are therefore the recyclable pool of intracellular Ca²⁺, the velocity of recycling and the mechanism that triggers Ca²⁺ release from the SR. However, before reviewing these aspects of EC coupling in normal and failing cardiomyocytes, a brief account of the clinical condition is required, since the great disparity of causes of CHF may be reflected in confusing inconsistencies in the literature.

Myocardial function in heart failure

The clinical condition of heart failure

Clinically heart failure is a manifestation of various conditions. A recent definition is that ‘Heart failure is a complex clinical syndrome that can result from any structural or functional cardiac disorder that impairs the ability of the ventricle to fill with or eject blood’ (Hunt et al. 2001). Thus, not all cases of heart failure are due to a reduced myocardial function, which in many cases is reduced contractility of otherwise seemingly healthy myocardium. However, heart failure is a progressive disease, and in many cases irrespective of the cause, myocardial function will eventually deteriorate in the face of unchanged or reduced haemodynamic load (Mann, 1999). It is important that during initial stages the heart failure is compensated and symptoms are scarce (Francis, 2001; Hunt et al. 2001). Also, cardiac performance deteriorates with age (Lakatta, 2002), which means that there could be important interaction between progress of the disease and age related alterations. At some point during the progression of the disease cardiac output can only be maintained at elevated diastolic filling pressure. This leads to pulmonary congestion. Hence, in this review we focus on CHF since it most often means that the disease has progressed to a state where the failure is no longer compensated and the function of seemingly healthy myocardium is depressed. In other words the progression of heart failure can often be linked to an increasing myocardial failure.

Causes of cardiac hypertrophy and heart failure

In principle heart failure has been thought to result from increased wall stress caused either by increased ventricular volume (increased preload) or by increased left ventricular systolic pressure (increased afterload) (Braunwald, 1997). The ensuing hypertrophy will tend to normalize the wall stress, but if hypertrophy and reduced myocardial performance are causally linked, the outcome will still be reduced cardiac performance (Mann, 1999). Such a causal link is difficult to accept since hypertrophy and remodelling of the heart take various shapes. On the cellular level a volume overload will cause serial addition of sarcomeres so the cells mainly become elongated, whereas a pressure overload will cause parallel addition of contractile elements. This means that the growth signals must differ with pressure and volume overload of the left ventricle. Training is a physiological stimulus for cardiac hypertrophy, which will also enhance cardiac performance and has in fact been shown to reduce the cardiac hypertrophy observed in rats after a myocardial infarction (MI) (Wisløff et al. 2002). However, Esposito et al. (2002) recently showed that in genetically modified mice that did not develop hypertrophy in response to pressure overload, cardiac function was better maintained with a high wall stress than in mice that developed hypertrophy. Thus, increased wall stress can be well tolerated for a prolonged time, and their study indicates that at least concentric hypertrophy and reduced contractility may be related. This is in some contrast to an earlier study that showed improved cardiac performance with hypertrophy subsequent to constriction of the transverse aorta (Nakamura et al. 2001). Also Crackower et al. (2002) recently showed that cardiac hypertrophy and contractility defects can be genetically uncoupled. We will not elaborate on the possible link between hypertrophy and contractile properties on the cellular level. This has been reviewed recently (Francis, 2001; Hoshijima & Chien, 2002). We will focus on CHF following MI. In this condition an eccentric hypertrophy develops and an extensive MI with dilatation of the left ventricle may be present, with little sign of congestion and reduced myocardial performance. However, with progression of the disease, CHF develops indicating attenuated contractility of the remaining myocardium (Sjaastad et al. 2000).

Myocardial failure

With echocardiography heart failure can be categorized as systolic and/or diastolic (Hunt et al. 2001). Systolic heart failure is characterized by a reduction in left ventricular performance, reduced contractility, enlargement of the left ventricle, reduction in ejection fraction, and pulmonary congestion. In post-infarction CHF the cardiac systolic function is primarily impaired, but diastolic dysfunction might also be present in the terminally failing post-infarction heart. Patients with primarily diastolic heart failure, such as in hypertensives, typically have pulmonary congestion, but the left ventricular systolic performance can be normal. Even enhanced systolic performance with increased contractility may be present in a diastolic dysfunctional heart. These hearts relax slowly due to low left ventricular compliance, leading to an abnormal left ventricular pressure-volume relationship and elevated left ventricular filling pressure (Zile & Brutsaert, 2002). Thus, the underlying cellular phenotypic changes may be different, underscoring the importance of using well-defined experimental models to unravel mechanisms of altered EC coupling in CHF.

Myocardial systolic dysfunction is characterized by both reduced velocity of contraction and attenuated magnitude of contraction. In vivo this becomes manifest as low ejection fraction and low rate of pressure development (dP/dt) (Carabello, 2002). On the cellular level many investigators only report reduced magnitude of shortening (Wasserstrom et al. 2000; Gomez et al. 2001). This is clearly only one parameter that reflects the in vivo contractility. Figure 1A shows isometric force during twitches of isolated papillary muscles from sham operated (SHAM) and CHF rats. Clearly peak force is reduced in CHF but in addition rate of force development is reduced. This is in agreement with some cellular studies that also report reduced shortening velocities (Ito et al. 2000; Sjaastad et al. 2002b). Thus, in terms of the classical force-velocity concept, power is reduced since both force and velocity are reduced (Fig. 1B) (Alpert et al. 2002). There are reports that fractional shortening (maximum shortening expressed as a percentage of resting cell length) of isolated cells is normal or even increased in CHF (Anand et al. 1997; Prahash et al. 2000) and in hypertrophy with little sign of failure (Litwin & Bridge, 1997; Ito et al. 2000). This of course raises the question of whether such cells are representative of a dysfunctional myocardium. One point to consider is that these cells may still exhibit reduced velocity of shortening, but data on velocity are rarely provided. If velocity of shortening is reduced, power or contractility will also be reduced even in the face of maintained magnitude of shortening.

Figure 1. Cardiomyocyte contractility.

A, contraction relaxation cycles in posterior left ventricular papillary muscles during isometric contractions. Muscles from sham-operated animals (thick line) developed higher force more rapidly than papillary muscles from CHF animals (thin line). B, shortening velocity-force relationship at a muscle length (L_max) that gives maximal force development. The two examples illustrate contractions in a normal muscle (thick line) and in a muscle with reduced contractility (thin line). The corresponding powers (force × velocity) are illustrated above.

The isometric force and the fractional unloaded shortening are both dependent on the magnitude of the Ca²⁺ transient. However, the velocity of shortening is closely correlated both to the myofibrillar ATPase activity, which is reduced in heart failure (Alpert et al. 2002), and to the cytosolic Ca²⁺ concentration (Vannier et al. 1996; Baker et al. 1998; Palmer & Kentish, 1998). Recently it has been pointed out that in heart failure, dyssynchronous release of Ca²⁺ sparks will lead to a slower rate of rise of the Ca²⁺ transient (Litwin et al. 2000; Sah et al. 2002), and in a mouse model of hypertrophy and failure both time to peak of the Ca²⁺ transient and the contraction were observed (Ito et al. 2000). The relative contribution of slowed rate of rise of the Ca²⁺ transient, reduced peak of the transient and reduced myofibrillar ATPase activity is unclear with regard to causing the reduced velocity of shortening in failing hearts. However, dyssynchronous release of Ca²⁺ will probably lead to dyssynchronous binding of Ca²⁺ to troponin C and to slower velocity of contraction. We will therefore focus on possible defects in well-established trigger mechanisms for CICR. We also consider whether a defect in SR function can account for the systolic dysfunction. Finally we discuss the possibility of a defect of other putative trigger mechanisms for SR Ca²⁺ release. However, some studies on post-infarction myocardial function have utilized animals that did not have CHF, which raises the possibility that these animals did not have myocardial failure (Cheung et al. 1994; Litwin & Bridge, 1997). For this reason we also try to distinguish between post-infarction studies performed on animals with or without a confirmed diagnosis of CHF and we include some comments on methods that have been applied to monitor myocardial function in experimental studies.

Methods for monitoring myocardial function

There is no simple method for evaluating myocardial function in vivo (Carabello, 2002). After an MI the function of the whole heart will not appropriately reflect the contractility of the viable myocardium due to the presence of scar tissue that has variable elasticity. Also experimental models in rat and mice have introduced new challenges with regard to miniaturization of the methods. Since both genetic modification and surgery may lead to great individual variation as to the degree of hypertrophy and failure in experimental models, we stress the importance of individual characterization of myocardial function in vivo before cells or tissue are examined in vitro. Techniques for evaluation of cardiac performance in rats and especially in mice have recently been reviewed so we will only mention a few common procedures (see for instance Christensen et al. 1997; Doevendans et al. 1998; Lorenz, 2002; Janssen et al. 2002; Roth et al. 2002b).

Invasive techniques include measurement of blood pressure, and segment and volume measurements of the heart using piezo-electric crystals or impedance measurements (Esposito et al. 2002). Lately echocardiography has become feasible even in the mouse allowing for monitoring myocardial function. Thus individual animals with various degrees of heart failure can be selected for experiments. Also, with proper consideration of both pre- and afterload, a diagnosis of myocardial failure can be achieved. In the rat MI model left ventricular end diastolic pressure above 15 mmHg correlated very well with reduced shortening velocity of the myocardium, left atrial dilatation and lung congestion (Sjaastad et al. 2000). Even so, direct comparison with the unloaded shortening of isolated cells should be made with great care.

Finally magnetic resonance imaging deserves to be mentioned as a very valuable method, but small bore magnets that provide adequate magnetic fields are still very expensive (Chacko et al. 2000; Weiss, 2001; Wiesmann et al. 2001).

Conclusion

Some of the variability in the literature as regards myocardial and cardiomyocyte function in hypertrophic and failing hearts undoubtedly stems from differences in experimental models. Both invasive and non-invasive techniques allow detailed analysis of cardiac and myocardial function in vivo, which may be valuable for classification of individual animals prior to single cell experiments. Velocity of shortening is equally important for proper function of the heart, as is the degree of shortening. It should be pointed out that myocardial failure (reduced contractility of seemingly intact myocardium) is not necessarily present in the hypertrophied and/or failing heart, especially not after a small myocardial infarction.

Cardiomyocyte contractility in heart failure

In post-infarction rats, shortening of isolated cardiomyocytes has been reported to be both normal (Anand et al. 1997; Prahash et al. 2000) and significantly depressed (Holt et al. 1998; Gomez et al. 2001; Sjaastad et al. 2002b), which could reflect variable experimental models, conditions and techniques. It is therefore difficult to judge on the basis of the literature whether reduced fractional shortening of isolated cells is an inherent property of the failing myocardium. One study showed that with high resistance pipettes and from a holding potential of −70 mV both fractional shortening and velocity of shortening were reduced in CHF cells compared to SHAM cells (Fig. 2A and B) (Sjaastad et al. 2002b). However, when cells were patched with low resistance pipettes using identical protocols there was no longer any difference between CHF and SHAM cells. The reason was that fractional shortening of SHAM cells was lower by 35 % under control conditions with patch pipettes as compared with high resistance pipettes (Fig. 2). Also velocity of shortening was different in patched cells. Thus in this study the patch clamp technique reduced the contractility and velocity of shortening of normal cells, but did not affect contractility of CHF cells. One explanation could be the dialysis of the cells that occurs through low resistance patch pipettes and that the contractile deficit of CHF cells is due to the absence of some substance that is easily dialysed out of SHAM cells, possibly some signalling molecule. With wide bore pipettes the patch clamp technique allows for dialysis of the cell interior, which is a great advantage of the method since it allows for introduction of substances into the cell and maintenance of ion concentrations at desired levels. Hence contractility can easily be manipulated.

Figure 2. Effects of experimental conditions on cell shortening and velocity of shortening.

A, fractional shortening and B, velocity of shortening in SHAM cells and CHF cells impaled with high resistance electrodes and low resistance patch pipettes. *Significantly different from SHAM; # significantly different from high resistance electrodes. C, fractional shortening at various voltages using patch pipettes and a post-conditioning potential of −50 mV. Solutions contained Cs⁺ instead of K⁺ and there was 0 Na⁺ in the pipette. D shows the same experiments repeated with K⁺ containing solutions and 6 mm Na⁺ in the pipette. From Sjaastad et al. (2002b) with permission.

Other investigators have reported a contractile deficit in cells from rats with myocardial infarction using the same patch clamp procedure (Gomez et al. 2001). This again points either to a true difference in phenotypes because of variable degrees of CHF or duration of the disease, or alternatively to subtle differences in experimental techniques such as different timing of recording after patch rupture, variable degree of dialysis or variable SR Ca²⁺ load.

Figure 2 also illustrates another aspect of experimental conditions (Sjaastad et al. 2002b). Figure 2C and D shows fractional shortening as a function of test potential using low resistance patch pipettes. With pipette solutions containing a high concentration of Cs⁺ and no Na⁺ in the pipette, the shortening-voltage relationship was bell-shaped (Fig. 2C), but when K⁺ replaced Cs⁺ and internal Na⁺ concentration was kept at 6 mm the relationship became sigmoidal (Fig. 2D). This has been ascribed to the effect of reverse mode Na⁺-Ca²⁺ exchange at positive potentials (Litwin et al. 1998; Wasserstrom et al. 2000), and contractions at step potentials to positive voltages are attenuated by the drug KBR-793 which inhibits reverse mode Na⁺-Ca²⁺ exchange (Piacentino et al. 2000). Importantly, no difference could be observed between CHF and SHAM cells with any of these protocols.

Another important aspect is cellular heterogeneity within the normal and diseased myocardium. The cellular phenotypes are different within the wall of the left ventricle, also in the rat (Gomez et al. 1997a). In addition, after myocardial infarction the wall stress is not homogeneously distributed within the remaining viable myocardium (Loennechen et al. 2001). This latter study also showed that expression of marker genes for hypertrophy and failure was associated with the degree of wall stress. Thus, the region of the left ventricle from which cells are harvested may be very important. Several studies have harvested cells from the border zone between scar tissue in MI hearts and seemingly normal, viable myocardium. These cells show larger hypertrophy and more pronounced deterioration of contractility than cells from a region more remote from the infarct zone (Olivetti et al. 1991; Melillo et al. 1996; Litwin & Bridge, 1997; Kim et al. 2002).

Conclusion

Contractile properties of isolated cardiomyocytes are highly dependent on experimental conditions. It is difficult to reconcile various reports on the contractile deficit of cardiomyocytes from rats with post-infarction CHF. Both reduced and normal fractional shortening have been reported, but few studies report velocities of contraction. There is a possible limitation with the whole cell patch clamp technique, which has been reported to depress contractility of normal cells thus obscuring the contractile deficit of CHF cells. This points to the necessity of using standardized disease models, and carefully monitor and report phenotypes, site within the ventricular wall for cell harvest and experimental conditions to allow comparison of data between studies.

Triggers of calcium release

L-type Ca²⁺ current (I_{Ca, L})

The inward Ca²⁺ current through the L-type Ca²⁺ channels is regarded as the single most important trigger of Ca²⁺ release (for an extensive treatise see Bers, 2002). Since many studies report proportionality between size of the current, the ensuing Ca²⁺ transient and contraction (Bers, 2001), reduced current density could in theory be an explanation for reduced contractility. However, as recently reviewed, peak I_Ca,L density is normal in most types of heart failure (Benitah et al. 2002a) including the post-infarction rat model (Zhang et al. 1995, 1996; Holt et al. 1998; Gomez et al. 2001; Sjaastad et al. 2002b). It has been argued that hypertrophy and failure have opposing effects on the current density (Benitah et al. 2002b) and that the net outcome therefore is unchanged current density. At the single channel level there is a puzzling report that in human heart failure availability and opening probability are both increased as if the channels were hyperphosphorylated (Schröder et al. 1998). However, this could fit with the observation that mRNA and protein expression are reduced and that the size of the gating charge is significantly depressed in a dog model of pacing-induced heart failure which may indicate that the number of channel protein copies is reduced (He et al. 2001). Other observations include slower Ca²⁺-dependent inactivation and absence of facilitation in human heart failure cells (Barrere-Lemaire et al. 2000). However, in the rat post-infarction model changes in inactivation kinetics could not be observed (Holt et al. 1998). In failing human hearts an isoform switch has been reported for the α_1c subunit (Yang et al. 2000), and Chen et al. (2002) recently reported that the density of L-type Ca²⁺ channels was reduced, but the current maintained due to increased phosphorylation of the channel protein. Clearly, the role of the L-type Ca²⁺ channel in CHF should be investigated more thoroughly, especially to reconcile a possible discrepancy between single channel and whole cell investigations.

Reverse mode Na⁺-Ca²⁺ exchange

Several reports show that reverse mode Na⁺-Ca²⁺ exchange can trigger contraction (Nuss & Houser, 1992; Levi et al. 1994; Wasserstrom & Vites, 1996). It has been reported that the exchanger is a less efficient trigger of Ca²⁺ release than I_Ca,L (Sipido et al. 1997). Several investigators point out that the exchanger may contribute to increased Ca²⁺ concentration in the dyadic cleft during the action potential, and hence contribute importantly to stimulate Ca²⁺ release (Litwin et al. 1998). In a recent simulation study, however, Weber et al. (2002) pointed out that Ca²⁺ mediated through I_Ca,L rapidly accumulates in the dyadic cleft, and for that reason the Na⁺-Ca²⁺ exchanger runs in reverse mode for a very short time during a normal action potential.

The direction of the Na⁺-Ca²⁺ exchange is highly dependent on intracellular Na⁺ concentration or possibly also a subsarcolemmal pool of Na⁺ (Lederer et al. 1990). There is evidence that when intracellular Na⁺ concentration is gradually raised in patch clamp experiments, the exchanger supports contractions at positive test potentials to an increasing extent (Litwin et al. 1998). Therefore the bell-shaped relationship between voltage and contraction that exists when only the L-type Ca²⁺ channels are providing Ca²⁺ for triggering release is changed so that contractions are present at positive voltages where L-type Ca²⁺ current is zero as illustrated in Fig. 2.

In a recent study Lipp et al. (2002) reported that the Ca²⁺ transients triggered by the L-type Ca²⁺ current and the Na⁺ current may not be identical. The L-type Ca²⁺ current triggered Ca²⁺ release composed of Ca²⁺ sparks. A Ca²⁺ spark is a well-defined quantal release of Ca²⁺ emanating from a group of Ca²⁺ release channels (RyRs) (see review by Guatimosim et al. 2002) and can be triggered by a single Ca²⁺ channel (Santana et al. 1996). However, in the presence of an inward Na⁺ current Lipp et al. (2002) also detected a Ca²⁺ release that seemed to be composed of much smaller unitary Ca²⁺ signals than the sparks and they speculate that this Ca²⁺ release is triggered by reverse mode Na⁺-Ca²⁺ exchange.

Several investigators report that the Na⁺-Ca²⁺ exchanger is upregulated by 50-100 % in various models of heart failure and in CHF patients (Studer et al. 1994; Gaughan et al. 1999; Wasserstrom et al. 2000; Barry, 2000) although some investigators have reported unchanged levels in both animal models and patients (Yao et al. 1998; Schwinger et al. 1999b; Piper et al. 2000). An increased expression of the Na⁺-Ca²⁺ exchanger can have several important effects: one is reduced Ca²⁺ load of the SR (see below), another is altered interplay between I_Ca,L and the exchanger in triggering contraction. Upregulation of the exchanger could make the EC coupling less dependent on I_Ca,L. In a rat CHF model the effect of verapamil was greatly reduced in CHF cells compared with SHAM at positive potentials above 20 mV (Wasserstrom et al. 2000). In fact verapamil had no significant effect on contractions triggered in this voltage range in CHF cells as opposed to about 60 % reduction of contraction in SHAM. It remains to be seen whether this is of importance when contractions are triggered by action potentials.

Also, upregulation of the Na⁺-Ca²⁺ exchanger will make the hearts more prone to Ca²⁺ overload for instance during ischaemia and reperfusion (Sjaastad et al. 2002a), and it provides a substrate for delayed afterpotentials and lethal arrhythmias (Pogwizd et al. 2001).

Other putative triggers of Ca²⁺ release

Ca²⁺ current through T-type Ca²⁺ channels, TTX-sensitive current through Ca²⁺ channels, Ca²⁺ current through Na⁺ channels and a voltage sensing mechanism have all been proposed as triggers of Ca²⁺ release (Aggarwal et al. 1997; Santana et al. 1998; Sipido et al. 1998; Zhou & January, 1998; Ferrier & Howlett, 2001). T-type Ca²⁺ channels are not normally expressed in the left ventricle, but seem to be upregulated in the hypertrophic heart (Martinez et al. 1999). Since they open at more negative voltages than the L-type channels they could contribute to a negative shift of the voltage dependence of contraction. However, this has not been reported in heart failure, and furthermore, these channels would probably contribute to maintain contractions, not reduce them. The TTX-sensitive Ca²⁺ current is only detectable in the absence of Na⁺, and the physiological relevance is therefore unclear (Aggarwal et al. 1997). The so-called slip-mode conductance of the Na⁺ channels allowing Ca²⁺ to pass only occurs during β-adrenergic stimulation (Santana et al. 1998). However, the existence of this current has been questioned (Piacentino et al. 2002), and there are no studies examining this current in cells from failing hearts. In a series of papers the voltage-sensitive release mechanism (VSRM) has been proposed to be a prominent trigger of contraction (Ferrier et al. 1998; Howlett et al. 1998; Zhu & Ferrier, 2000; Ferrier & Howlett, 2001). It is activated from voltages negative to −40 mV, and contributes to making the contraction-voltage relationship sigmoidal independent of the Na⁺-Ca²⁺ exchanger. The VSRM has been reported to be significantly attenuated in cardiomyopathic Syrian hamsters (Howlett et al. 1999). The existence of VSRM is debated (Piacentino et al. 2000), and so far a molecular substrate for such a mechanism has not been identified.

Conclusion

It seems reasonable to conclude that based on existing data the contractile deficit in CHF cannot be ascribed to reduced I_Ca,L and that in some instances of heart failure upregulation of the Na⁺-Ca²⁺ exchanger could in fact reinforce the trigger of contraction. There are conflicting data as to whether contractions triggered by I_Ca,L are reduced in CHF. It seems that expression of channel protein may be reduced in CHF but that this is compensated for by increased single channel current. Taken together with several reports on various alternative triggers of Ca²⁺ release than the I_Ca,L, most notably reverse mode Na⁺-Ca²⁺ exchange, it seems important to investigate more closely the role of the L-type Ca²⁺ channel in the EC coupling.

The intracellular calcium store

The SR Ca²⁺ store in CHF

For a given peak I_Ca,L fractional Ca²⁺ release from the SR increases almost exponentially with SR Ca²⁺ content (SR load), but no release can be triggered at an SR Ca²⁺ load of less than 50 % of the maximum load as shown in a study on isolated rabbit cardiomyocytes (Shannon et al. 2000). Therefore, reduced SR Ca²⁺ load might explain reduced contractility in CHF, primarily reduced fractional shortening. As discussed below, SR load is highly dependent on a number of factors, and it is difficult to distinguish between altered SR load and altered trigger of SR Ca²⁺ release as an explanation for reduced contractility in CHF. To our knowledge, no studies have really determined the SR Ca²⁺ content under conditions that prevail in vivo. Studies have been done in isolated cells under variable pacing frequencies by field stimulation or during current or voltage clamp. With a standardized stimulation protocol during voltage clamp in a rat post-infarction CHF model there seems to be no significant difference in SR Ca²⁺ load by caffeine application (Gomez et al. 2001; Sjaastad et al. 2002b). However, in human heart failure and other experimental models, reduced SR Ca²⁺ load has been reported (Beuckelmann et al. 1992; Lindner et al. 1998; Pogwizd et al. 2001; Hobai & O'Rourke, 2001). Recently Maier et al. (2002) reported reduced capacity to store Ca²⁺ in muscle strips from human failing hearts. It is therefore reasonable to assume that in vivo, the SR content of Ca²⁺ is probably low in CHF contributing to reduced fractional shortening and force development (Pieske et al. 2002).

The Ca²⁺ pump (SERCA)

Gwathmey et al. (1987) were the first to focus on the reduced capacity of the SERCA to load the SR with Ca²⁺ in a study of the contractile defect in failing hearts. Later numerous studies examined expression of SERCA at the mRNA and protein levels, as well as the capacity to transport Ca²⁺. As compiled by Bers (2001, Table 26, p. 319) and Houser et al. (2000) many studies show reduced expression of the SERCA 2 protein, but some also show unaltered expression. Also the results are variable with regard to the regulatory protein phospholamban (Houser et al. 2000; Bers, 2001). We recently showed in one study a nominal decrease of SERCA2 protein and in another study 30 % decrease of SERCA2 protein in the rat post-infarction CHF model (Sande et al. 2002; Sjaastad et al. 2002b). In none of these studies did we find any alteration in phospholamban expression, and the ratio between SERCA2 and phospholamban pentamer or monomer were not significantly changed in CHF.

The expression of SERCA2 and phospholamban in the heart is highly dependent on thyroid hormones. In isolated cardiomyocytes exposed to triiodothyronine for 48 h, phospholamban expression was profoundly reduced and Ca²⁺ transients were increased (Holt et al. 1999). With more prolonged exposure also SERCA2 level also goes up. The effects of thyroid hormones on the cardiovascular system have recently been reviewed (Klein & Ojamaa, 2001). The point to make here is that not only may thyroid hormones be low in heart failure patients, but also expression of thyroid hormone receptors in the heart is significantly altered (Kinugawa et al. 1999). Thus without careful control of the thyroid status an independent effect of the heart failure condition on expression of SERCA2 and phospholamban will be very difficult to judge. Variable thyroid status may therefore explain some of the variability in the literature regarding expression of these proteins in the failing heart.

Recent focus on SR function has shifted to the phosphorylation status of phospholamban in CHF. There are three phosphorylation sites, of which the serine-16 site and the threonine-17 site are most important. Several studies, both experimental and in humans, have shown a reduced level of serine-16 phosphorylation, suggesting that SERCA2 will be more inhibited than in normal cells (Schmidt et al. 1999; Schwinger et al. 1999a; Sande et al. 2002). This effect seems to predominate over altered expression level, and was also found in non-failing MI rats (Huang et al. 1999). The reduced phosphorylation is associated with significant slowing of Ca²⁺ reuptake by the SR (Sande et al. 2002). Interestingly in the same study the phosphatases PP1 and PP2A were found to be upregulated, whereas there was no change in the expression of the regulatory unit of protein kinase A. Finally, in line with the reduced level of phosphorylation the response to forskolin was increased in CHF cells. These findings point to an important aberration of intracellular signalling systems in heart failure. Accordingly, overexpression of adenylcyclase in cardiomyopathic mice increases survival (Roth et al. 2002a).

Factors that control the Ca²⁺ content of SR

SR Ca²⁺ load is controlled by a multitude of factors that are depicted in Fig. 3. The sarcolemma and the membrane of the sarcoplasmic reticulum are arranged in series so that all ions have to pass both membranes to alter SR Ca²⁺ load. The concentration of Ca²⁺ in the cytosol is maintained so low, even though it transiently increases during each beat, that the storage of Ca²⁺ in the cytosolic compartment is negligible. Altered net transport across the sarcolemma is therefore transmitted to the SR within a few beats. Ca²⁺ binding proteins and the mitochondria can temporarily store significant amounts of Ca²⁺, but will not be considered here.

Figure 3. Factors influencing SR Ca²⁺ load.

Examples of cellular proteins, cell functions and experimental conditions that influence sarcoplasmic reticulum load are illustrated. SOC, store operated Ca²⁺ channels; V_m, membrane potential; SERCA-2, sarcoplasmic reticulum Ca²⁺-ATPase; PLB, phospholamban; RyR, ryanodine receptor. Illustration made by Tore Taraldsen, MD.

The net Ca²⁺ transfer across the sarcolemma will be controlled by the balance between I_Ca,L, and reverse and forward modes of Na⁺-Ca²⁺ exchange. The outward Ca²⁺ transport by the sarcolemmal Ca²⁺-ATPase is very small (Choi & Eisner, 1999). In principle, therefore, it is only temporary mismatches between I_Ca,L and reverse mode Na⁺-Ca²⁺ exchange on one hand and forward mode on the other hand that will cause changes in SR Ca²⁺ load (Meme et al. 2001). Since cytoplasmic Ca²⁺ is a common substrate for the Na⁺-Ca²⁺ exchanger in forward mode and SERCA2, the two transporters compete for the same cytosolic Ca²⁺ pool (Bassani et al. 1994; Choi & Eisner, 1999). To put it simply: increased SERCA2 activity will increase SR Ca²⁺ load and increased Na⁺-Ca²⁺ exchange activity will decrease it. Species with high SERCA2 activity will recirculate a large fraction of the Ca²⁺ inside the cell with each contraction, whereas species with low SERCA2 activity will depend on Ca²⁺ fluxes across the sarcolemma to provide Ca²⁺ for the contractile machinery (Bers, 2001).

As cited above, in some studies of the failing and in the hypertrophied heart, independent of species, there is about a doubling of the Na⁺-Ca²⁺ exchanger protein (Pogwizd et al. 1999; Wasserstrom et al. 2000; Hobai & O'Rourke, 2000). Thus the slower rate of Ca²⁺ uptake into the SR (Beuckelmann et al. 1992; Sande et al. 2002) combined with increased expression of exchanger protein will both contribute to reduced SR Ca²⁺ load. In line with this Shillinger et al. (2000) and Ranu et al. (2002) recently reported that contractions were severely depressed in rabbit cardiomyocytes that overexpress the Na⁺-Ca²⁺ exchanger.

However, the picture is complicated by alterations of the driving forces for the Na⁺-Ca²⁺ exchanger. Figure 3 indicates that membrane potential and ion gradients play an important role. A depolarisation of 10 mV in right ventricular cardiomyocytes from rats with extensive myocardial infarction as reported by Kaprelian et al. (1999) will significantly shift the balance of the exchanger in favour of more reverse and less forward mode exchange. Furthermore, intracellular Na⁺ is probably slightly elevated in failing cardiomyocytes (Despa et al. 2002b), which will also favour reverse mode Na⁺-Ca²⁺ exchange. Hence, the upregulation of the exchanger protein may well be counterbalanced by reduced driving force for forward Na⁺-Ca²⁺ exchange.

Intracellular Na⁺ concentration is controlled by the Na⁺-K⁺ pump. Bundgaard & Keldsen (1996) have reviewed the literature regarding expression levels of the Na⁺-K⁺ pump in failing human hearts, and conclude that pump concentration seems to be reduced by about 25 %. One study ascribed this to downregulation of the α1 and β1 subunits as well as a small reduction in α3 (Schwinger et al. 1999b). In an experimental rat study a significant isoform switch was reported in CHF after MI (Semb et al. 1998). Expression of the α2 isoform was reduced and the α3 isoform appeared, whereas the expression of α1 was unchanged. This did not seem to affect the k_0.5, which was much higher than in other species. Altered level of pump protein expression relative to tissue weight or tissue protein does not necessarily reflect Na⁺-K⁺ pump efficacy in the tissue (Ellingsen et al. 1994). As recently pointed out by Gomez et al. (2002) with respect to the Na⁺-Ca²⁺ exchanger and by Semb et al. (1998) as regards the Na⁺-K⁺ pump, it is important to consider cell volume and surface to volume ratio when judging the capacity of these transporters to maintain the cytosolic concentrations of Ca²⁺ and Na⁺, respectively. In the latter study it was reported that the number of pump protein copies per cell was unchanged. However, since CHF cells were about 30 % larger than SHAM they contained more Na⁺ and the Na⁺-K⁺ pump could not adjust the concentration as rapidly as in SHAM cells. In the study by Ellingsen et al. (1994) it was pointed out that even though the tissue concentration of Na⁺-K⁺ pumps was reduced, the number of pumps in the whole left ventricle was unchanged. Despa et al. (2002a) did not find reduced capacity of the Na⁺-K⁺ pump in a rabbit heart failure model, but ascribed the higher cytosolic Na⁺ concentration to an increased leak. It seems safe to conclude that alterations of the Na⁺-K⁺ pump expression are not very pronounced in heart failure, but intracellular Na⁺ seems to be increased by a combined effect of reduced expression of pump protein in some cases, larger cell volume, and increased leak of Na⁺. Together these effects are large enough to significantly alter the driving force for the Na⁺-Ca²⁺ exchanger.

Another important determinant of SR Ca²⁺ load is the RyR. As neatly pointed out in several papers from Eisner's group (Eisner et al. 2000; Trafford et al. 2000, 2001), a sudden increase or a sudden decrease of release through the RyR for a given stimulus will, within a few beats, cause adjustments of the SR Ca²⁺ load so that release is again normalized. Thus, provided SERCA2 activity was maintained, a reduced release or gain of the EC coupling in CHF might cause increased SR Ca²⁺ load so that the size of Ca²⁺ transients would be maintained. This is clearly not the case, so that by this reasoning it seems that any change in EC coupling would not be sufficient to reduce fractional shortening in CHF unless it is accompanied by reduced capacity for Ca²⁺ reuptake by the SR. Thus, overexpression of SERCA may restore fractional shortening even in the face of continued inefficiency of the EC coupling (del Monte et al. 1999; Miyamoto et al. 2000; Davia et al. 2001). Synchronization of sparks is probably not related to SR Ca²⁺ load, and reduced velocity of shortening might therefore be a more sensitive indicator of defective EC coupling than the variable fractional shortening.

Conclusion

SR Ca²⁺ load is an important determinant of the size of the Ca²⁺ transient and cardiac contractility. Regulation of SR Ca²⁺ load is extremely complex, but it seems that in many instances of heart failure, and possibly also CHF, SR Ca²⁺ load is reduced. However, depolarisation and increased intracellular Na⁺ concentration might counteract upregulation of the Na⁺-Ca²⁺ exchanger and depressed phosphorylation of phospholamban and thus contribute to maintain an adequate Ca²⁺ content of the SR. Interestingly, a defect in the EC coupling might not be sufficient in itself to cause reduced Ca²⁺ transients, but requires concomitant reduction of Ca²⁺ reuptake capacity by the SR.

A variable gain?

Gain of the EC coupling in CHF

As alluded to above, efficiency of the trigger (the size of I_Ca,L) to cause Ca²⁺ release from the SR has been termed ‘gain’ of the EC coupling (Beuckelmann & Wier, 1988; Wier et al. 1994; Wier & Balke, 1999). Reduced gain in heart failure has been reported by several authors, as recently reviewed by Benitah et al. (2002b). Gain is dependent not only on the SR Ca²⁺ load, but also on membrane potential. At negative voltages, although I_Ca,L is reduced, the driving force for inward Ca²⁺ current is increased, which seems to make it more efficient in triggering release (Santana et al. 1996).

In many cases of heart failure, reduced gain of the EC coupling seems to be due to reduced SR Ca²⁺ load (Hobai & O'Rourke, 2001; Pogwizd et al. 2001). However, also under experimental conditions that are aimed at keeping the SR Ca²⁺ load constant, gain has been reported to be reduced in CHF (Gomez et al. 1997b; Gomez et al. 2001). In another study Sjaastad et al. (2002b) were not able to see this defect. At various temperatures there were no differences in I_Ca,L between CHF and SHAM rats, and contractions were not significantly different at any voltages between −40 and +60 mV. Also, the contraction- current relationship was almost linear from −40 to +60 mV when contractions were triggered by I_Ca,L, suggesting that I_Ca,L-triggered gain was not very high at negative voltages (Sjaastad et al. 2002b). However, possible differences between these studies (Gomez et al. 1997b, 2001; Sjaastad et al. 2002b) include duration of disease, severity of failure and possibly also SR Ca²⁺ load. Accordingly, some important considerations should be pointed out. First, SR Ca²⁺ load should be monitored under the exact same conditions that were used for the voltage clamp protocols. Second, contractions and Ca²⁺ release can be triggered at voltages below −40 mV where the macroscopic I_Ca,L is very small (Piacentino et al. 2002), or even in the absence of any I_Ca,L provided a VSRM mechanism exists (Ferrier & Howlett, 2001). Thus, gain may approach infinity and will be very difficult to quantify unless experimental conditions are carefully designed.

Possible causes of altered gain in CHF

The major players in the coupling between the sarcolemma and the SR are depicted in Fig. 4. They comprise first of all the L-type Ca²⁺ channels (no. 1), and the Na⁺-Ca²⁺ exchanger (no. 2) and possibly a VSRM (no. 3).

Figure 4. Summary of alterations in EC coupling in CHF.

An L-type Ca²⁺ channel (1), both t-tubular (2) and sarcolemmal Na⁺-Ca²⁺ exchangers, a sarcolemmal voltage sensor (3), the sarcoplasmic reticulum Ca²⁺-ATPase, phospholamban and a sarcolemmal Ca²⁺-ATPase are illustrated. The following structures have altered function in CHF compared to sham: the Na⁺-Ca²⁺ exchanger both in trigger and relaxation mode, the voltage sensor, phospholamban and the sarcoplasmic reticulum Ca²⁺-ATPase. Illustration made by Tore Taraldsen, MD.

Since the Na⁺-Ca²⁺ exchanger is upregulated it is not likely that reduced gain can be ascribed to altered function of this protein. However, there could be several reasons why L-type Ca²⁺ channels and RyRs do not communicate efficiently. First it has been reported that the structure of the t-tubules is altered in heart failure (He et al. 2001). Second, a reduced number of ryanodine receptors relative to L-type Ca²⁺ channels has also been observed (Milnes & MacLeod, 2001). Third, increased distance between the two proteins has been suggested (Gomez et al. 2001). Fourth, increased phosphorylation of the ryanodine receptor has been claimed to alter EC coupling, and cause a basal leak of Ca²⁺ from the SR (Marx et al. 2000; Marks, 2000; Ono et al. 2000; Yano et al. 2000; Reiken et al. 2001; Doi et al. 2002; Marks et al. 2002). Similarly, the L-type Ca²⁺ channels may be hyperphosphorylated (Schröder et al. 1998; Chen et al. 2002). This finding might imply very different signalling systems in CHF cells as compared with SHAM since it has now been shown that phosphorylation of phospholamban is reduced and not increased (Schwinger et al. 1999a; Sande et al. 2002). It should also be mentioned that the VSRM has been reported to be highly sensitive to the phosphorylation status of the cell, and seems to be attenuated with reduced phosphorylation levels (Hobai et al. 1997; Ferrier et al. 1998; Zhu & Ferrier, 2000). Finally, there is recently a growing interest in the role of the cytoskeleton in targeting both channel proteins and signalling systems that control EC coupling (Zile et al. 2001; Rueckschloss & Isenberg, 2001). This may be a new avenue for understanding the defects of EC coupling in heart failure.

Conclusion

Gain of the EC coupling in heart failure is clearly reduced by low SR Ca²⁺ content, but also, independently of load, some investigators have reported reduced gain of the EC coupling although this is not a universal finding. Both the L-type Ca²⁺ channel and the RyR seem to be hyperphosphorylated in CHF. Several explanations, some involving altered structure of t-tubules and/or the dyad, have been proposed. Very few studies have addressed the question of whether reduced velocity of shortening in CHF can be linked to altered gain or efficiency of the EC coupling. Clearly our current concepts of the molecular machinery that constitutes the link between the t-tubules and the SR need to be refined.

Perspectives

Cardiac hypertrophy and failure represent a multitude of phenotypes, and future studies should carefully monitor myocardial function in vivo to allow interpretation of data from in vitro experiments. Even with well-defined reduced contractility of the myocardium reflected on the cellular level, there is as yet no single explanation for the contractile deficit. Rather, an improved understanding of EC coupling calls for further investigations to unravel both the normal and the pathological trigger mechanisms of cardiomyocyte contraction.