Atrial cardiomyocytes make an important contribution to the refilling of ventricles with blood, which enhances the subsequent ejection of blood from the heart. The dependence of cardiac function on the contribution of atria becomes increasingly important with age and exercise. We know much less about the calcium signals that link electrical depolarisation to contraction within atrial myocytes in comparison with ventricular myocytes. Nevertheless, recent work has shed new light on calcium signalling in atrial cells. At an ultrastructural level, atrial and ventricular myocytes have many similarities. However, a few key structural differences, in particular the lack of transverse tubules (`T-tubules') in atrial myocytes, make these two cell types display vastly different calcium patterns in response to depolarisation. The lack of T-tubules in atrial myocytes means that depolarisation provokes calcium signals that largely originate around the periphery of the cells. To engage the contractile machinery, the calcium signal must propagate centripetally deeper into the cells. This inward movement of calcium is ultimately controlled by hormones that can promote or decrease calcium release within the myocytes. Enhanced centripetal movement of calcium in atrial myocytes leads to increased contraction and a more substantial contribution to blood pumping. The calcium signalling paradigm within atrial cells applies to other cardiac cell types that also do not express T-tubules, such as neonatal ventricular myocytes, and Purkinje cells that aid in the spread of electrical depolarisation. Furthermore, during heart failure ventricular myocytes progressively lose their regular T-tubule expression, and their pattern of response resembles that of atrial cells.
Physiological role of atrial myocytes
The heart undergoes a cycle of events that cause blood to be propelled to the lungs and the body. These events can be divided into two stages: diastole and systole. During the diastolic stage, the atrial and ventricular myocytes are relaxed. Systole refers to the period of contraction and consequent ejection of blood from the ventricles to the pulmonary artery or aorta. The cardiac cycle is initiated by the sinoatrial node - a group of specialised non-contractile cardiac myocytes positioned in the wall of the right atrium. All of the cells in the heart have an intrinsic ability generate action potentials (electrical impulses). However, the sinoatrial node acts as the primary pacemaker because its cells naturally discharge at a high rate, and so override the other potential pacemaking sites. Hormonal modulation of the sinoatrial node is one of the principal mechanisms for altering the frequency of the heartbeat. From the sinoatrial node, the action potential spreads over the atria causing them to contract and push blood into the ventricles. As the wave of depolarisation sweeps across the heart, it reaches the atrioventricular node, which filters and relays the signal to the ventricles via specialised conduction tissue including the Purkinje fibres. The atrial chambers contract and relax before the ventricular systole, and their activation is evident as separate electrical activity in an electrocardiogram. The time course of contraction is marginally shorter in atria compared to ventricles. Both cell types reach peak contraction within a few tens of milliseconds (Luss et al., 1999).
The ventricles contract more forcefully than the atrial chambers and are predominantly responsible for forcing blood out of the heart. However, atria can play a significant role in altering the amount of blood that loads into the ventricles before to systole. When a person is at rest, the contribution of atria to the filling of ventricles with blood is relatively low. The majority (∼90%) of ventricular refilling occurs as a consequence of the venous pressure of blood returning to the heart. However, if the heart rate increases, such as during exercise or stress, then atrial contraction can account for ∼20-30% of the volume of blood in the ventricles. This contribution to ventricular refilling is known as `atrial kick' (Lo et al., 1999). Under some pathological conditions (such as mitral valve stenosis) and during ageing, the dependence on atrial kick can be critical (Nicod et al., 1986).
A common form of cardiac dysrhythmia known as `atrial fibrillation' occurs when electrical impulses do not solely arise from the sinoatrial node, but instead spontaneously occur with high frequency from sites around the atria (∼350 discharges per minute, compared with the normal sinoatrial rhythm of 60-80 beats per minute). The rapid and irregular electrical discharges during atrial fibrillation cause the atria to quiver and thereby prevent coordinated contraction. Under these conditions, atrial kick is missing and up to one-third of the blood pumping capacity of the heart is lost (Alpert et al., 1988). Atrial fibrillation is not immediately life threatening because the pressure of blood returning to the heart can still refill the ventricles, albeit to a lesser degree. However, concomitant with the loss of pumping capacity during fibrillation, blood can pool and clot within the atria. If a blood clot travels to the brain and becomes lodged in an artery it can result in a stroke. The incidence of atrial fibrillation increases with age, and there is an increased risk of stroke in people with atrial fibrillation (Wolf et al., 1998). It is therefore clear that co-ordinated atrial function is very important.
Ca2+ is a ubiquitous intracellular messenger
Cells use a variety of mechanisms to sense extracellular cues and transduce them into physiological actions. Alteration of the cellular Ca2+ concentration is an important transduction mechanism (Berridge et al., 2000; Bootman et al., 2001a). Indeed, Ca2+ is an almost ubiquitous messenger that has been demonstrated to regulate activities ranging from fertilisation to cell death (Berridge et al., 1998; Bootman et al., 2001b). The ability of Ca2+ to regulate such a wide assortment of processes is due to the array of mechanisms that cells use to modulate the duration, amplitude, frequency and spatial extent of Ca2+ signalling events (Thomas et al., 1996). For example, Ca2+ signals can occur over microseconds to trigger activities such as exocytosis (Bootman et al., 2001b; Neher, 1998). Alternatively, they can repeat over minutes or days to control cell growth (Berridge et al., 2003). Ca2+ transients can occur in nano- or microscopic domains (Bootman et al., 2001b; Rizzuto and Pozzan, 2006), be sequestered into organelles (Duchen, 2000), or pass through the cytoplasm and produce a global whole-cell Ca2+ signal. In addition, Ca2+ can co-ordinate the activity of neighbouring cells by diffusing through gap junctions (Paemeleire et al., 2000) or causing the paracrine release of messengers (Charles, 2005).
The versatility of Ca2+ signalling derives from a complex interplay of mechanisms that either induce or reverse rises in cellular Ca2+ concentration. Cells use an assortment of pathways to promote Ca2+ signals (Bootman et al., 2002), and an equally broad panoply of systems that terminate Ca2+ transients and restore tissues to their resting state (Berridge et al., 2003). The net result of these `on' and `off' mechanisms determines the nature of the Ca2+ event. Since different cell types display unique combinations of such on and off mechanisms, Ca2+ signals are tissue specific and compatible with the particular function of each cell. Furthermore, the proteomes responsible for cellular Ca2+ signalling are highly plastic and can be remodelled through Ca2+-dependent transcriptional/translational control mechanisms (Berridge et al., 2003).
Ca2+ and cardiac EC coupling
Ca2+ is a critical regulator of cardiac function, because it links the electrical depolarisation of cardiomyocytes with contraction (excitation-contraction coupling; EC coupling) (Bers, 2002). Elevation of Ca2+ within cardiac myocytes allows the actin and myosin contractile filaments to engage and slide past each other, thus shortening the cells and producing the force to propel blood. Ca2+ signals in cardiomyocytes are initiated as the action potential from the sinoatrial node sweeps over the heart. The consequent cellular depolarisation activates voltage-operated Ca2+ channels (VOCCs), which allows Ca2+ to flow across the sarcolemma and into the cytoplasm. This Ca2+ influx occurs into a narrow (∼10 nm) cleft (often referred to as the `dyadic' or `tryadic' cleft) (Ayettey and Navaratnam, 1978), which is formed by the sarcolemma and the membrane of the internal Ca2+ store (the sarcoplasmic reticulum; SR) (Fig. 1). The SR is studded with Ca2+ release channels known as ryanodine receptors (RyRs), which are activated by Ca2+ itself, in a process called Ca2+-induced Ca2+ release (CICR) (Fill and Copello, 2002; Meissner, 2004; Roderick et al., 2003). The density of RyRs is particularly high at the clefts, where the SR and sarcolemma are in close apposition, and they are therefore located directly opposite VOCCs. When Ca2+ floods into the cytoplasm through VOCCs it does not have far to diffuse before encountering RyRs. The rise in Ca2+ concentration on the cytosolic side of the RyRs causes them to open, thereby releasing Ca2+ sequestered within the SR and greatly amplifying the cytosolic Ca2+ signal. The Ca2+ concentration within this small space cannot be measured at present, but using mathematical modelling it has been estimated that the Ca2+ rise may exceed several millimolar in the centre of the cleft (Jones et al., 2006; Peskoff and Langer, 1998).
The essential effect of CICR in the dyadic cleft is to cause microscopic Ca2+-release events termed `Ca2+ sparks' (Cannell et al., 1995; Cheng et al., 1993). These tiny Ca2+ signals arise from the concerted activation of a cluster of RyRs. It has been suggested that as few as ∼15 RyRs underlie Ca2+ sparks (Soeller and Cannell, 2002), although the variability in their time course and amplitude suggests that the number and/or cooperation between channels can vary (Parker and Wier, 1997; Shen et al., 2004). The Ca2+ entry signal that activates the RyR cluster probably arises from a smaller group of closely apposed VOCCs (Inoue and Bridge, 2003). Depending on the condition of the myocyte, Ca2+ sparks can represent an amplification of ∼tenfold or more over the Ca2+ flux that originally entered through the VOCCs. This amplification often referred to as the `gain' in EC coupling. The degree of gain can be altered by environmental conditions, as described below.
Eventually, Ca2+ ions diffuse out of the cleft and trigger contraction by binding to troponin C within the myofibrils. Ca2+ recovers back to resting (diastolic) levels owing to sequestration in the SR by an ATP-dependent enzyme known as SERCA (sarcoendoplasmic reticulum Ca2+ATPase), and transport across the sarcolemma via the action of a Na+/Ca2+ exchanger. Myocytes also posses a sarcolemmal Ca2+ ATPase, but this is believe to make a lesser contribution to Ca2+ extrusion (Eisner and Sipido, 2004).
The cardiomyocytes within different regions of the heart vary in morphology, ultrastructure and molecular composition. Nevertheless, the process described above is used to initiate Ca2+ signals within all the electrically excitable cardiomyocyte cell types. Fig. 2 shows some of the structural aspects of ventricular, atrial and neonatal cardiac myocytes. A striking feature of these cells is their regular striations, evident from staining for structural proteins (e.g. α-actinin, an actin-binding protein found at Z-lines; Fig. 2Ai-Ci) Ca2+-release channels (RyRs; Fig. 2Aii-Cii) and plasma membrane proteins (the Na+/Ca2+ exchanger; Fig. 2Aiii). The stereotypic locations of structural and signalling proteins within a particular cardiac cell type means that depolarisation-induced Ca2+ signals are reasonably consistent for that tissue. However, the variation in cellular ultrastructure between different tissues within the heart gives rise to Ca2+ signals with dissimilar spatial and temporal properties.
Ventricular myocyte EC coupling
Cardiac Ca2+ signalling has been most extensively studied in myocytes isolated from the ventricles. It is well established that Ca2+ signals in mammalian ventricular myocytes take the form of homogenous whole-cell increases (Cannell et al., 1995; Guatimosim et al., 2002). Such global Ca2+ responses reflect the spatial and temporal summation of signals from many Ca2+ spark sites, which are synchronously recruited during depolarisation. The critical ultrastructural detail of mammalian ventricular myocytes that promotes homogenous global Ca2+ transients is the presence of an extensive `transverse tubular' system (T-tubules) (Song et al., 2005). These narrow (average diameter ∼200 nm) inwardly directed projections of the sarcolemma arise at each of the Z-lines within cardiac muscle and have a regular spacing (∼1.8 μm) (Brette and Orchard, 2003) (Fig. 1). The principal tubules running perpendicularly into the cells can also bear branches, giving rise to a complex network of longitudinal and axial tunnels (Ayettey and Navaratnam, 1978). The typical striated distribution of ventricular myocyte T-tubules is shown in Fig. 2Aiii, which depicts the distribution of the sarcolemmal Na+/Ca2+ exchanger. Many of the proteins involved in generating and reversing Ca2+ signals are concentrated at the T-tubules (Song et al., 2005). In particular, T-tubules serve to bring VOCCs and RyRs into close proximity. Indeed, immunostaining ventricular myocytes for VOCCs and RyRs reveals that these proteins largely appear in discrete clusters within a 3-dimensional matrix throughout the cells (Chen-Izu et al., 2006; Mackenzie et al., 2004a) (Fig. 3Ai-iii). The clusters of VOCCs and RyRs have substantially overlapping distributions, indicating that they are in very close proximity (Scriven et al., 2002). When ventricular myocytes are depolarised, the action potential is relayed to VOCCs within the T-tubule network. This means that even those Ca2+ spark sites deep within a ventricular cell are recruited during EC coupling, thereby giving rise to a synchronised global Ca2+ signal (Brette et al., 2005). If ventricular myocytes are imaged at high speed during the onset of EC coupling, the Ca2+ signal can be seen to arise with a striated appearance (Cleemann et al., 1998; Tanaka et al., 2003). This reflects the initiation of Ca2+ release adjacent to the T-tubules before the Ca2+ signal has managed to diffuse from the dyadic junctions (Isenberg et al., 1996).
Atrial myocyte EC coupling
Unlike ventricular myocytes, atrial cells do not possess an extensive T-tubule system (Fig. 2Biii; the Na+/Ca2+ exchanger immunostaining is around the outside of the cell only), although some atrial cells possess a more rudimentary transverse-axial tubular network (Ayettey and Navaratnam, 1978; Brette and Orchard, 2003; Gotoh, 1983; Tidball et al., 1991). In place of the T-tubules, atrial cells have prominent SR elements, which have been described as `Z-tubules' (Fig. 1) (Yamasaki et al., 1997). Just like T-tubules, these structures are perpendicular to the long axis of the cells. Atrial cells therefore contain a form of transversely oriented tubule, but it is formed from internal SR membrane and not the sarcolemma.
The SR in cardiac muscle has a highly convoluted organisation that can be classified into four major structural elements: junctional, corbular (corbula means `little basket' in Latin), cisternal and network (Fig. 1). Both ventricular and atrial cells possess all these SR forms, but their relative abundance is different (Yamasaki et al., 1997). For example, ventricular myocytes have a more extensive network SR running alongside the myofibrils and interweaving with the mitochondria. By contrast, atrial myocytes have a less complex network SR, but substantially more corbular compartments (Franzini-Armstrong et al., 2005; Sommer and Jennings, 1992). The corbular SR possesses both RyRs and the Ca2+ storage protein calsequestrin (Jorgensen et al., 1993) and is therefore a potential source for amplification of Ca2+ responses by CICR. Owing to its abundance in atrial myocytes, the corbular SR may contribute significantly to the generation of Ca2+ signals.
The pattern of RyR localisation in atrial myocytes has some similarity to that observed in ventricular cells, in that most RyRs lie within regularly spaced transverse striations corresponding to the positions of the Z-tubules (Figs 2Bii and 3Bii). However, there is one crucial difference between atrial and ventricular myocytes: that is the expression of additional RyR clusters around the periphery of the atrial cells (Carl et al., 1995; Hatem et al., 1997; Mackenzie et al., 2001) (Fig. 2Bii). Atrial myocytes therefore possess two populations of RyRs. One minor group (junctional RyRs), sit just beneath the sarcolemma. The other channels (non-junctional RyRs) are deeper inside the cell and constitute the bulk of the RyR population. Although the junctional RyRs represent a small fraction of the total number of channels, they are crucially important in atrial EC coupling (see below).
The localisation of VOCCs is entirely different in ventricular and atrial myocytes. Because of the general lack of T-tubules, the sarcolemma does not regularly protrude into atrial cells, and VOCCs only function around the outside of the myocytes (Mackenzie et al., 2004a). As a consequence, the close apposition of RyRs and VOCCs that is necessary for triggering EC coupling occurs only at the cell periphery (Fig. 3Bi-iii).
These differences in ultrastructure have a significant effect on the spatial properties of the Ca2+ signals. As described above, ventricular myocytes display homogenous responses, which arise from the simultaneous recruitment of Ca2+ sparks throughout a cell. However, in atrial myocytes, EC coupling is initiated around the periphery of the cells, because this is the only place where the VOCCs and junctional RyRs come together. High-speed imaging of atrial myocyte responses has shown that multiple Ca2+ sparks sites are simultaneously recruited around the edge of a cell immediately after depolarisation (Kockskamper et al., 2001; Mackenzie et al., 2001). Within a few tens of milliseconds, these microscopic Ca2+ transients appear to spread laterally and merge, thereby causing a shell of elevated Ca2+ concentration beneath the sarcolemma, with no immediate response deeper inside (Fig. 4A,B). From that point, the development of the Ca2+ signal depends on the status of the myocytes, factors such as the Ca2+ content of the SR and the presence of `positive inotropes' (agents that increase the force of heart contraction) determine whether the peripheral Ca2+ signal will pass further into the cells or not (Mackenzie et al., 2004a).
Under basal conditions, with cells electrically paced to their steady-state Ca2+ load, the Ca2+ signal beneath the sarcolemma does not propagate significantly into the centre of the cells (Fig. 4C) (Mackenzie et al., 2001; Mackenzie et al., 2004a). At the peak of the Ca2+ response, approximately 20 milliseconds after stimulation, the peripheral Ca2+ signal can reach ∼1 μM. However, owing to the lack of inward spreading, the Ca2+ concentration declines sharply with distance away from the sarcolemma, and the centre of the cell barely shows an increase over the normal prestimulated Ca2+ level of ∼100 nM. This spatially heterogeneous pattern is even maintained while the Ca2+ transients recover back to their pre-depolarisation (diastolic) condition (Fig. 4B). Similar results have been obtained using atrial cells from different mammalian species, including rat (Mackenzie et al., 2001; Tanaka et al., 2003; Woo et al., 2002), guinea pig (Berlin, 1995; Lipp et al., 1990), cat (Huser et al., 1996; Kockskamper et al., 2001; Sheehan and Blatter, 2003) and human (Hatem et al., 1997). These studies all agree that atrial Ca2+ responses originate at subsarcolemmal sites and that the Ca2+ signals have the greatest amplitude and rate of rise in the peripheral initiation zone. There is some disagreement about how much the subsarcolemmal Ca2+ signal is able to propagate into cells, but this probably reflects experimental variation in parameters such as SR Ca2+ content and the presence of cyclic AMP (cAMP), as discussed below.
Modulation of atrial Ca2+ responses
Under basal stimulation conditions, as described above, the elevation in Ca2+ concentration evoked by depolarisation of an atrial myocyte is largely restricted to the periphery of the cells. However, the majority of protein filaments that must sense the Ca2+ signal in order to cause contraction are located deeper inside the cell. What limits the inward propagation of the Ca2+ transient, and can it be overcome so that Ca2+ can reach the contractile machinery?
Given the regular distribution of RyRs inside atrial myocytes, it is perhaps a little surprising that Ca2+ signals can be restricted to the periphery at all. Clearly, the junctional RyRs are closest to the VOCCs, and that is why they respond. Although non-junctional RyRs are further away from the sarcolemmal VOCCs, they are not that remote. One might expect the non-junctional RyRs to sense the subsarcolemmal Ca2+ sparks and relay the Ca2+ signal deeper into the myocyte via successive rounds of CICR. Such a saltatoric conduction of Ca2+ waves between clusters of channels has been demonstrated to underlie the propagation of Ca2+ signals within cells (Bootman et al., 1997; Coombes et al., 2004; Keizer et al., 1998; Kockskamper et al., 2001). However, this centripetal movement of Ca2+ clearly does not occur under basal conditions.
Many studies of RyR distribution in atrial cells have shown that the junctional RyRs are separated from the non-junctional RyRs by a ∼2 μm gap (Carl et al., 1995; Chen-Izu et al., 2006; Kockskamper et al., 2001; Mackenzie et al., 2001; Mackenzie et al., 2004a) (Fig. 2Bii and Fig. 3Bii). Although cytosolic Ca2+ can diffuse further than 2 μm, the distance is sufficiently large for the Ca2+ concentration to fall below that required for the activation of a Ca2+ spark (Izu et al., 2006). Therefore, one possibility why the subsarcolemmal Ca2+ signal does not always trigger centripetal Ca2+ waves is that the gap in RyR distribution acts as a firebreak to diminish the progression of the Ca2+ signal. Beyond the gap, the non-junctional RyR clusters are spaced <1 μm apart in the transverse direction (they are ∼2 μm apart along the longitudinal axis) (Chen-Izu et al., 2006), so that centripetal propagation is more likely once the gap has been overcome.
In addition, the junctional RyRs that underlie the peripheral Ca2+ signal appear to be functionally distinct from their non-junctional counterparts. For example, the frequencies of spontaneous Ca2+ sparks (which can be observed when myocytes are not electrically paced) are significantly greater in the subsarcolemmal region than in the centre of atrial cells (Brette et al., 2005; Lipp et al., 2000; Woo et al., 2003a). It has been proposed that junctional RyRs interact directly with the C-terminal tail of L-type VOCCs, and that this physical coupling sensitises the RyRs to Ca2+ and thereby promotes CICR (Woo et al., 2003b). Since the non-junctional RyRs are distant from the VOCCs, they are not similarly sensitised and thus require a higher cytosolic Ca2+ concentration or larger SR Ca2+ load before CICR is initiated (Huser et al., 1996). These putative differences in the sensitivity of junctional and non-junctional RyRs to Ca2+ may contribute to the spatial patterning of Ca2+ signals during atrial myocyte EC coupling, because CICR will be more easily activated in the junctional region than in the centre of the cells.
A further reason why centripetal Ca2+ waves do not always occur is the interference of Ca2+ movement by cellular organelles. In particular, mitochondria and SERCAs restrict the Ca2+ signal to the periphery of atrial myocytes under basal conditions (Mackenzie et al., 2004a). Both of these Ca2+ transport mechanisms provide a powerful buffer against rises of cellular Ca2+ concentration (Diaz et al., 2005; Gunter et al., 2004; Landgraf et al., 2004). Inhibition of either of these Ca2+ sequestration pathways allows the peripheral Ca2+ signal to propagate significantly into the interior of atrial myocytes (Mackenzie et al., 2004a). Electron microscopy and immunofluorescence studies have confirmed that mitochondria and SERCAs are closely aligned with the junctional RyRs (Kockskamper et al., 2001). It therefore appears that activation of the junctional RyRs leads to Ca2+ sparks that have the capacity to recruit their neighbouring non-junctional RyR clusters by CICR, but under basal conditions a firewall composed of mitochondria and Ca2+ pumps prevents the inward diffusion of the signal. The contractile myofilaments may also contribute to the buffering of Ca2+ and hinder movement of the ion within myocytes because they contain Ca2+-binding proteins.
Artificially increasing the buffering capacity of the sarcoplasm within atrial myocytes by introducing exogenous Ca2+ chelators leads to a sharper definition of the peripheral Ca2+ signal and diminishes the response from non-junctional RyRs (Sheehan and Blatter, 2003). Diffusion of Ca2+ away from the subsarcolemmal initiation sites is therefore crucial to trigger propagating CICR deeper in the cell.
The conserved ultrastructural features of mammalian atrial myocytes dictate that the Ca2+ signals they produce are spatially heterogeneous. The expression of powerful Ca2+ sequestration pathways in proximity to the zones where EC coupling initiates disrupts the diffusion of Ca2+ from the periphery of the cells, so that the saltatoric centripetal propagation of a Ca2+ wave does not occur. A major function of atrial myocytes, however, is to promote the refilling of ventricles with blood, particularly under times of activity and stress. Given that the contractile machinery is deeper inside the myocytes than the junctional regions, the cells must have a mechanism to generate Ca2+ signals that overcome the firewall. Decreasing SERCA activity or the extent of mitochondrial Ca2+ uptake allows the propagation of the Ca2+ signal, but this does not happen in a physiological context. Instead, atrial myocytes use systems to enhance the trigger Ca2+ influx through VOCCs, elevate the Ca2+ content of the SR and increase the sensitivity of RyRs to CICR. Stimulation of any of these processes is sufficient to promote the centripetal propagation of Ca2+ into the cells and thereby promote contraction (Huser et al., 1996; Mackenzie et al., 2004a).
A major hormonal influence on the force of contraction of the heart occurs via the sympathetic nervous system (Bers, 2002). Stimulation of these nerves leads to the engagement of β-adrenergic receptors on the sarcolemma of myocytes, and the consequent production of cAMP and activation of protein kinase A (PKA). This kinase can phosphorylate cardiac L-type VOCCs, leading to an increase in mean channel open time and/or channel opening probability, thereby enhancing the trigger Ca2+ influx following depolarisation (van der Heyden et al., 2005). In addition, PKA phosphorylates phospholamban, a protein on the SR that interacts physically with the SERCAs to inhibit Ca2+ transport (Frank et al., 2003). Phosphorylation by PKA dissociates phospholamban from the SERCA pumps and increases Ca2+ sequestration, which can lead to a dramatic increase in SR Ca2+ content. The lumenal Ca2+ concentration within the SR has a substantial positive impact on the `gain' of CICR (i.e. the ratio of the resultant Ca2+ signal compared with the trigger Ca2+ concentration). Increasing the lumenal Ca2+ concentration has a steep non-linear effect on the sensitivity of RyRs to CICR, and promotes more substantial Ca2+ flux (Bassani et al., 1995; Diaz et al., 2005; Koizumi et al., 1999; Laver, 2005). This probably occurs through saturation of the Ca2+-binding protein calsequestrin, which provides a reservoir of releasable Ca2+ and allosterically regulates RyR opening (Beard et al., 2005; Gyorke et al., 2004). A further effect of PKA may be to phosphorylate RyRs and consequently increase their sensitivity to Ca2+ and enhance their open probability (Bers, 2004). Sympathetic stimulation of the heart therefore engages several molecular mechanisms to generate a positive inotropic effect.
β-adrenergic stimulation promotes the reliable centripetal propagation of the Ca2+ signal during EC coupling in atrial myocytes. Following a brief stimulation with a β-adrenergic agonist, the pattern of Ca2+ signalling changes dramatically from a peripheral ring of Ca2+ to a full-blown global response in which the non-junctional RyRs participate (Hatem et al., 1997). Perfusing atrial myocytes with cAMP via a patch pipette has a similar effect (Woo et al., 2002). Concomitant with the change in centripetal propagation of the Ca2+ signal, there is a significant increase in contraction (Mackenzie et al., 2004a). Hormones can thus control atrial contraction by modulating the spatial distribution of the Ca2+ signal. The interplay of several Ca2+ management systems dictates the spatial extent of the atrial myocyte Ca2+ signal. In turn, the spatial properties of the Ca2+ signal dictate the degree of cellular contraction. The non-junctional RyRs within atrial myocytes therefore constitute a positive inotropic reserve, which can be recruited when stronger cellular contraction is required.
In addition to RyRs, atrial myocytes express another form of intracellular Ca2+ release channel: inositol 1,4,5-trisphosphate receptors (InsP3Rs) (Li et al., 2005; Lipp et al., 2000; Mackenzie et al., 2002). Ca2+ release from InsP3Rs is triggered in response to stimulation of myocytes with hormones that activate phospholipase C, such as the potent vasoconstrictive peptide hormone endothelin-1. Although InsP3Rs are ∼100 times less abundant than RyRs within the heart, they appear to be able to subtly modulate cardiac functions. In particular, InsP3Rs can regulate gene transcription and stimulate positive inotropy (Li et al., 2005; Proven et al., 2006; Wu et al., 2006). For atrial myocytes, InsP3R activation enhances both the centripetal propagation of Ca2+ signals and cellular contraction (Mackenzie et al., 2004a). Pharmacological or genetic ablation of InsP3R function prevents the positive inotropic effect of agonists such as endothelin-1 on atrial myocytes (Li et al., 2005; Mackenzie et al., 2004b).
InsP3R activation has also been shown to cause spontaneous Ca2+ signals in many studies (Mackenzie et al., 2004b; Woodcock and Matkovich, 2005). The generation of spontaneous Ca2+ signals within myocytes has the potential to perturb the normal beating of myocytes during the cardiac cycle. For reasons that are presently not understood, atrial myocytes express substantially more InsP3Rs than ventricular cells (Li et al., 2005; Lipp et al., 2000). This means that atrial myocytes are more prone to the potential dysrhythmic effects of InsP3R activity. However, the expression of InsP3Rs within ventricular myocytes is increased during some forms of heart failure (Go et al., 1995). In this case, ventricular myocytes will become more like atrial cells - being more responsive to InsP3R activation - and could potentially suffer from greater hormone-induced dysrhythmia.
Atrial myocytes show a spatial continuum of Ca2+ signals
Under basal conditions, atrial Ca2+ signals are restricted to the subsarcolemmal periphery of cells (Fig. 4), whereas, following application of a positive inotropic stimulus, global Ca2+ transients are evoked (Mackenzie et al., 2004a). These responses represent the least and greatest Ca2+ signals that atrial myocytes can produce. However, these cells are not digital in their patterns of Ca2+ response. Instead, they show a spatially graded continuum of Ca2+ signals, which correlate with the extent of cellular contraction. This is illustrated in Fig. 5, which shows simultaneous recordings of Ca2+ concentration in the subsarcolemmal and nuclear regions of an atrial myocyte, and also the corresponding degree of cellular contraction. The cell was electrically paced at a constant frequency, just as if it was receiving a regular action potential from the sinoatrial node. To simulate the effect of a positive inotropic agonist, the extracellular Ca2+ concentration was elevated from 1 to 10 mM as indicated. The figure shows that the switch to 10 mM Ca2+ causes a substantial positive inotropic effect: the ability of the Ca2+ signal to invade the nuclear region increases with each depolarisation and the contraction of the cell is simultaneously augmented. The progressive increase in both the centripetal propagation of the Ca2+ signal and contraction shows that atrial cells can smoothly grade their responses over a wide range.
Rapid central Ca2+ responses in atrial myocytes
In the majority of studies examining the spatial properties of atrial Ca2+ signalling, peripheral Ca2+ signals have been found to peak tens of milliseconds before the central Ca2+ signals fully develop (Mackenzie et al., 2004a; Sheehan and Blatter, 2003). Furthermore, it has been demonstrated that Ca2+ release from the junctional RyRs can be abruptly terminated by rapidly repolarising an atrial myocyte, whereas once the non-junctional RyRs are activated their Ca2+ release persists for a while after repolarisation (Hatem et al., 1997; Sheehan and Blatter, 2003). These observations support the concept that the junctional RyRs are under the control of membrane potential via activation of VOCCs, whereas the non-junctional RyRs are recruited by a distinct CICR mechanism. However, a few studies have demonstrated a component of fast Ca2+ release within central regions of atrial cells (Kirk et al., 2003; Woo et al., 2002). As mentioned above, some atrial myocytes have a rudimentary T-tubule system (Ayettey and Navaratnam, 1978; Gotoh, 1983; Tidball et al., 1991), and this may be the basis for some rapid central Ca2+ responses. In atrial cells that have T-tubules, electrically evoked Ca2+ signals originate simultaneously at multiple independent locations within the cells (Kirk et al., 2003).
Why atrial myocytes possess T-tubules at all is unclear. Since T-tubules penetrate deep into the cells, they can activate Ca2+ signals beyond the firewall described above. It is plausible that they serve to trigger CICR within the cell centre and thereby increase the speed at which Ca2+ reaches the myofilaments. In addition, by acting as a conduit for extracellular Ca2+, the T-tubules may increase the SR Ca2+ content and speed SR refilling (Brette et al., 2005; Woo et al., 2005). These effects could lead to more homogenous Ca2+ transients within individual myocytes. However, the density and morphology of T-tubules not only differs between ventricular and atrial myocytes but can also vary in extent between cells of the right and left atrial chambers. Furthermore, within the different atrial chambers, the pattern of T-tubules varies from cell to cell (Kirk et al., 2003; Tidball et al., 1991; Woo et al., 2005). The inconsistency in T-tubule expression, density and morphology in atrial cells is therefore puzzling, because it could cause substantial variability in the response of adjacent cells within the working atria.
The relevance of atrial myocyte Ca2+ signalling to other cell types, heart failure and development
The Ca2+ signalling paradigm of atrial myocytes is also relevant to other tissues where excitation only occurs at the periphery of the cells. For example, cardiac Purkinje cells, which underlie the conduction of action potentials into the mass of ventricular myocytes, display patterns of Ca2+ signalling similar to those observed in atrial cells (Boyden et al., 2000; Cordeiro et al., 2001). Purkinje cells are non-contractile myocytes responsible for the transmission of electrical signals. Like atrial cells, they do not have T-tubules but express RyRs in regular striated patterns throughout their volume (Cordeiro et al., 2001; Stuyvers et al., 2005). Moreover, Ca2+ signals in Purkinje cells initiate in subsarcolemmal locations and show a weak centripetal propagation into the cell interior. Increasing the concentration of extracellular Ca2+ or application of a β-adrenergic agonist promotes the response of non-junctional RyRs, but the pattern of peripheral to central propagation is maintained (Cordeiro et al., 2001). The physiological consequences of modulating the centripetal propagation of Ca2+ within Purkinje cells are not known.
In some pathological conditions that lead to heart failure, the density of T-tubules in ventricular myocytes decreases (Balijepalli et al., 2003; He et al., 2001). Consequently, Ca2+ signals do not arise synchronously throughout ventricular cells, because there are cellular locations where EC coupling fails to occur (Louch et al., 2004; Song et al., 2006). A marked reduction in T-tubule density leads to non-simultaneous recruitment of the Ca2+ sparks within a ventricular myocyte, which decreases the amplitude and/or rate of rise of the Ca2+ signal (Brette et al., 2004; Louch et al., 2004). The functional consequences of detubulation can be mimicked to an extent by pharmacological agents that decrease Ca2+ entry through the L-type VOCCs (Cheng et al., 1995) or reduce the activation of RyRs by CICR (Diaz et al., 2002). The loss of global VOCC/RyR activation leads to patchy Ca2+ signals, where parts of the myocyte respond poorly or not at all (Shiels and White, 2005). The decline in the kinetics and amplitude of the Ca2+ signal has a detrimental effect on cellular contraction (Gomez et al., 1997). If a substantial number of T-tubules are lost, the Ca2+ signal within the ventricular myocytes begins to resemble that of atrial cells, displaying a peripheral-only initiation of EC coupling. Chemically-induced detubulation (Brette et al., 2005) or maintaining ventricular cells in long-term culture (where the lack of electrical pacing causes loss of T-tubules) produces a similar effect (Lipp et al., 1996).
The Ca2+ signals within detubulated cells can be substantially rescued by stimulation with positive inotropes. For example, application of β-adrenergic stimuli promotes the centripetal propagation of Ca2+ within detubulated ventricular cells and restores the kinetics and amplitude of the Ca2+ signal (Brette et al., 2004). It therefore appears that when ventricular myocytes deteriorate during heart failure, the loss of T-tubules makes their pattern of Ca2+ signalling increasingly like that in atrial cells. They turn out to be dependent on CICR to propagate Ca2+ signals away from the peripheral sites where EC coupling is initiated, and the successful centripetal movement of Ca2+ can be reliant on the presence of positive inotropes.
T-tubules develop after birth in mammals, and there is a progressive increase in the density and complexity of sarcolemmal invaginations (Chen et al., 1995; Sedarat et al., 2000). Neonatal ventricular myocytes are therefore not the same as their adult counterparts. Neonatal cells show some degree of striation, but are still far from being fully organised. Figs 2 and 3 demonstrate the variation in shape and structure within neonatal cells and, in particular, the absence of T-tubules (Fig. 2Ciii). As for atrial myocytes, the lack of T-tubules in neonatal cells means that RyRs and VOCCs co-localise around the periphery of the cells (Fig. 3Ci-iii). Consequently, neonatal myocytes display Ca2+ signals that are similar to the responses seen in atrial myocytes - they originate at the subsarcolemmal sites and show a diminishing centripetal movement (Haddock et al., 1999; Perez et al., 2005). Indeed, many features of neonatal ventricular myocytes responses are just like those of adult atrial cells. For example, when stimulated with the RyR agonist caffeine, neonatal cells give homogenous Ca2+ signals, which indicates that they have RyRs throughout the cell and replete Ca2+ stores. But the bulk of RyRs deep within the myocytes do not generally participate in EC coupling. Furthermore, again similarly to the situation in atrial cells, spontaneous Ca2+ sparks initiate predominantly in subsarcolemmal regions of newborn cells (Haddock et al., 1999). Since, in biochemical and electrophysiological assays, the RyRs in neonatal and adult ventricular myocytes are functionally equivalent (Perez et al., 2005), the differences in the spatial pattern of Ca2+ signalling can largely be ascribed to the lack of T-tubules. The physiology of neonatal cells is also different from that in adult tissue, however, because the SR is less well developed and has a smaller Ca2+ uptake capacity. Furthermore, neonatal cells express lower levels of L-type VOCCs and RyRs (Perez et al., 2005) (Figs 2 and 3).
T-tubules are actually a feature of mammalian cells only. Ventricular myocytes from non-mammalian vertebrates do not possess these structures. Characterisation of Ca2+ signalling in ventricular myocytes from animals such as fish (Shiels and White, 2005) has demonstrated that their responses resemble those of mammalian atrial cells. Non-mammalian ventricular cells largely display junctional couplings of VOCCs and RyRs around the cellular periphery (Protasi et al., 1996; Sommer and Jennings, 1992). Peripheral initiation and centripetal propagation is therefore the most commonly observed cardiac Ca2+ signalling paradigm within biology. It occurs within the majority of mammalian cardiac cell types and is the sole mechanism underlying EC coupling in other animals.
There are also electrically excitable cells outside the heart that display peripheral initiation of Ca2+ signalling around the cell soma, and rely on amplification by CICR from centrally located RyRs. Neurons, for example, display such Ca2+ profiles (McDonough et al., 2000). These cells do not possess the regularly arranged dyadic couplings and internal Ca2+ stores seen in adult cardiac cells, and perhaps are more akin to neonatal myocytes that have a more random arrangement of Ca2+ stores. However, as in atrial myocytes, the participation of central RyRs within the neuronal soma depends on the Ca2+ content of the intracellular stores or co-stimulation with agonists that sensitise internal Ca2+ release (Shmigol et al., 1995).
Atrial myocytes use spatially graded Ca2+ signals to control their contractile function. The Ca2+ response that occurs following depolarisation of an atrial myocyte depends on the interplay of multiple processes that serve to increase or reduce cytosolic Ca2+ levels. The channels, pumps and transporters that impinge on atrial Ca2+ signalling are expressed in regular patterns, so that the Ca2+ signals in these cells are stereotypic. Depolarisation triggers Ca2+ signals that occur at the periphery of the cells. The success or failure of this subsarcolemmal Ca2+ signal to penetrate into the centre of the cells depends substantially on the sensitivity of the non-junctional RyRs and their ability to perceive a triggering signal from behind a mitochondrial and/or SERCA firewall. Positive inotropic stimuli enhance the centripetal propagation of the Ca2+ signal, using mechanisms that enhance CICR. Negative inotropes (e.g. acetylcholine and adenosine), by contrast, would be expected to reduce the centripetal propagation of the Ca2+ signals and sharpen the subsarcolemmal ring of Ca2+. There is a roughly linear relationship between the inward movement of Ca2+ and atrial myocyte shortening. The extent of centripetal propagation of the Ca2+ signal therefore provides an analogue control over cell contraction. The spatial continuum of Ca2+ signalling from peripherally localised to global responses provides a mechanism by which atrial myocytes can precisely regulate their contribution to blood pumping in the heart. Owing to the paucity of T-tubules in myocytes outside the ventricles, the atrial EC coupling model is the most commonly encountered paradigm in mammalian cardiac tissues. It is the sole mechanism underlying activation of non-mammalian cardiac myocytes.
H.L.R. is a Royal Society University Research Fellow. This work was funded by the BBSRC, British Heart Foundation (grant number PG/06/034/20637) and EPSRC (grant number EP/C539184/1). D.R.H. is funded by an MRC PhD studentship.