Calcium signalling in developing cardiomyocytes: implications for model systems and disease

Abstract

Adult cardiomyocytes exhibit complex Ca²⁺ homeostasis, enabling tight control of contraction and relaxation. This intricate regulatory system develops gradually, with progressive maturation of specialized structures and increasing capacity of Ca²⁺ sources and sinks. In this review, we outline current understanding of these developmental processes, and draw parallels to pathophysiological conditions where cardiomyocytes exhibit a striking regression to an immature state of Ca²⁺ homeostasis. We further highlight the importance of understanding developmental physiology when employing immature cardiomyocyte models such as cultured neonatal cells and stem cells.

Introduction

In heart muscle cells, called cardiomyocytes, contraction is triggered by the binding of Ca²⁺ to the myofilaments. However, Ca²⁺ signals also regulate the activity of kinases, phosphatases, ion channels, exchangers and transporters. Through these pathways, Ca²⁺ is known to regulate cardiomyocyte growth, gene expression, differentiation and development. This multifunctional role requires high dynamic gain, fast propagation, and accurate spatial control of Ca²⁺ signals.

Cardiomyocyte contraction is initiated by rapid elevation of the cytoplasmic free [Ca²⁺]. In the adult mammalian heart, this Ca²⁺ transient is initiated by the opening of plasmalemmal L-type Ca²⁺ channels during the action potential, which triggers additional release of Ca²⁺ stored in the sarcoplasmic reticulum (SR) via ryanodine receptors (RyRs, Fig.1C). This sophisticated amplification process is referred to as Ca²⁺-induced Ca²⁺ release, and it enables an impressive dynamic range of Ca²⁺ signals, with an approximately 10-fold increase in free cytosolic Ca²⁺ concentration in only tens of milliseconds. This process is regulated by numerous ion channels, transporters, Ca²⁺ buffers, and regulatory enzymes whose activities are tightly coordinated in time and space. In addition to having a high gain system for Ca²⁺ release, adult myocytes also possess specialized structures to enable rapid propagation of cytosolic Ca²⁺ signals. These structures are essential to overcoming the rather slow diffusion of Ca²⁺ in the cytosol, which results from significant Ca²⁺ buffering.

Figure 1.

Established and hypothesized changes in cardiomyocyte structure during maturation and disease development

A, t-tubules are absent in prenatal cardiomyocytes, although the sarcoplasmic reticulum (SR) is present, at least in rudimentary form. Pacemaking is initiated by spontaneous release of Ca²⁺ from ryanodine receptors (RyRs) and IP₃ receptors (IP₃Rs), predominantly from the perinuclear SR. Resulting Ca²⁺ extrusion by the NCX triggers action potentials, Ca²⁺ influx via L-type Ca²⁺ channels and reverse-mode NCX, and release of SR Ca²⁺ via peripheral RyRs. B, during postnatal development, budding t-tubules project inwards forming dyadic junctions with the SR, which are anchored by junctophilin-2. Maturation of dyads is believed to involve their progressive ‘packing’ with Ca²⁺ channels and RyRs. C, this enables Ca²⁺ entry through Ca²⁺ channels to efficiently trigger Ca²⁺ release from RyRs across adult cells. Relaxation occurs as Ca²⁺ is recycled into the SR by SERCA and extruded from the cell by NCX. D, during diseases such as heart failure, t-tubules are lost and/or disorganized causing some RyRs to be ‘orphaned’ without adjacent RyRs. Our data indicate that RyRs are also lost and dispersed in this condition, suggesting that dyadic ‘unpacking’ may occur (Louch et al. 2013). This process can be viewed as an effective reversal of those that occur during development. Dyadic disassembly is proposed to reduce the efficiency of Ca²⁺-induced Ca²⁺ release and weaken contraction. A reactivation of Ca²⁺-dependent pacemaking (the ‘Ca²⁺ clock’) in diseased cardiomyocytes is believed to promote arrhythmia.

During development, cardiomyocytes undergo gradual maturation towards the efficient state of Ca²⁺ homeostasis present in adult cells. This process includes progressive structural development and increasing capacity of Ca²⁺ sources and sinks. In this review, we outline our current understanding of these developmental changes. A recurrent theme in this discussion is the notion that throughout development, cardiomyocyte Ca²⁺ signalling relies on unique mechanisms and cellular structures that are sufficient to fulfil the signalling tasks at a given developmental stage, but which are quite immature compared to those of adult cells (Sasse et al. 2007; Rapila et al. 2008; Korhonen et al. 2009, 2010). We emphasize the importance of understanding developmental physiology when performing experiments with neonatal cell cultures and stem cells that exhibit immature Ca²⁺ homeostasis. We also examine Ca²⁺ handling in pathophysiological conditions, as accumulating evidence indicates that diseased cardiomyocytes exhibit de-differentiation, with a reversal of changes that occur during development.

The early embryonic cardiomyocyte is a Ca²⁺ oscillator

The heart is the first organ that becomes functional in the vertebrate embryo. In mouse, the precardiac mesoderm is induced at embryonic day 7.25 (E7.25), and approximately 24 h later it forms a primitive but beating tubular heart (E8). An equivalent stage of development in man occurs at about 3 weeks of gestation (Brand, 2003). During subsequent stages of maturation, the tubular heart is progressively remodelled towards the four-chambered, adult organ.

Contractile activity of the developing heart is initiated when the first cardiomyocytes begin producing oscillations in their intracellular Ca²⁺ concentration, which then trigger the contractile protein machinery to generate a heartbeat. Contractile activity in the embryonic heart initiates at early stages, long before the formation of specialized structures for excitation–contraction coupling which are present in adult cardiomyocytes. This suggests that different mechanisms are involved in triggering the first contractions in developing cardiac cells. However, even the simple Ca²⁺ signals present in embryonic cells require a variety of proteins and specialized structures. Indeed, embryonic cardiomyocytes express most Ca²⁺-handling proteins from very early stages (Fig.1A). By embryonic day 9, immature cardiomyocytes exhibit functional SR with two types of SR Ca²⁺ release channels, inositol-3-phosphate receptors (IP₃Rs) and ryanodine receptors (RyRs), and express the SR Ca²⁺ ATPase (SERCA2a) for SR Ca²⁺ uptake (Rapila et al. 2008). The whole-cell Ca²⁺ oscillations that serve pacemaking originate from spontaneous IP₃R- and RyR-mediated Ca²⁺ release, predominantly from the perinuclear SR (Rapila et al. 2008) (Fig.1A). IP₃Rs cannot produce Ca²⁺ oscillations in the absence of active RyRs (Sasse et al. 2007), but instead produce Ca²⁺ leak, which in turn stimulates co-localized RyRs to release Ca²⁺ (Rapila et al. 2008) (Fig.1A). Consequently, in early embryonic cardiomyocytes the frequency of the oscillations and therefore spontaneous contractile activity depends on [IP₃]_i (Sasse et al. 2007; Rapila et al. 2008). Further support for this view comes from the recent observation that endothelin 1-dependent production of IP₃ controls the early embryonic heart rate in vitro and in vivo (Karppinen et al. 2014).

In addition to functional SR, early embryonic cardiomyocytes also have an excitable plasmamembrane containing voltage-activated Na⁺ and Ca²⁺ (L- and T-type) channels and Na⁺/Ca²⁺ exchanger (NCX, Figs1A and 2) (Korhonen et al. 2008). Spontaneous SR Ca²⁺ release triggers Ca²⁺ removal from the cell by NCX, resulting in inward current (Reppel et al. 2007). This depolarization activates the voltage-dependent Na⁺ and Ca²⁺ channels, and thereby triggers an action potential (Sasse et al. 2007; Rapila et al. 2008). Thus, the same cells that produce spontaneous Ca²⁺ oscillations respond to electrical stimulation by generating action potentials and Ca²⁺ influx, and this incoming Ca²⁺ can trigger Ca²⁺-induced Ca²⁺ release from the SR (Rapila et al. 2008) (Fig.3A and C). With further development, pacemaking becomes less dependent on SR Ca²⁺ release – the ‘Ca²⁺ clock’ (Lakatta & DiFrancesco, 2009) – and additionally begins to rely on the pacemaker current, I_f. This current is carried by hyperpolarization-activated cyclic nucleotide-gated channels, which are expressed from approximately E9 onwards in mice (Stieber et al. 2003) (Fig.2).

Figure 2.

Development of ventricular cardiomyocyte Ca²⁺ signalling

Summarized data from available literature (see main text) illustrate semi-quantitative alterations in Ca²⁺ homeostasis during maturation of the heart through embryonic, fetal, postnatal and adult stages. Ventricular myocyte development is associated with progressive cardiomyocyte growth and cessation of cell division toward adulthood. Pacemaking is predominantly Ca²⁺ dependent (the ‘Ca²⁺ clock’) in the early embryo, with I_f playing a larger role at later stages. Neither pacemaking mechanism is prominent in adult ventricular cardiomyocytes. t-tubules containing L-type Ca²⁺ channels develop postnatally. A resulting progressive increase in L-type Ca²⁺ current density is paralleled by loss of T-type Ca²⁺ current and a shift in NCX function favouring Ca²⁺ extrusion. Close proximity between L-type channels and an increasing density of RyRs in the SR enables greater reliance of the Ca²⁺ transient on Ca²⁺-induced Ca²⁺ release during development. SR function is progressively augmented by increasing SERCA expression and greater Ca²⁺ buffering capacity due to expression of calsequestrin. Cystolic Ca²⁺ buffering is also progressively increased. During diseases such as heart failure (right panel), there is a striking disruption of Ca²⁺ homeostasis consistent with regression toward an immature phenotype. Changes include re-emergence of pacemaker activity and T-type Ca²⁺ current, and loss of t-tubules, RyRs and SERCA which weakens Ca²⁺-induced Ca²⁺ release.

Figure 3.

Comparison of spatiotemporal characteristics of intracellular Ca²⁺ dynamics in developing and adult cardiomyocytes in silico

A and B, both the action potential (blue line) and Ca²⁺ transient (green line) exhibit much faster dynamics in the adult compared to developing myocyte. C and D, indeed, Ca²⁺ fluxes through the L-type Ca²⁺ channel (J_CaL; blue line) and RyR (J_RyR; green line) reach their peaks and decay much more rapidly in the adult myocyte. SERCA and NCX fluxes (J_SERCA and J_NCX, respectively) exhibit more rapid time courses in mature cells, and J_NCX exhibits markedly less Ca²⁺ influx component than is present in the developing myocyte. The presented data were obtained from in silico models published by Korhonen et al. (2010) and Morotti et al. (2014) for the mouse embryonic and adult cardiomyocyte, respectively.

Development of ventricular cardiomyocyte Ca²⁺ homeostasis

In the later stages of embryonic and fetal development, Ca²⁺ homeostasis becomes specialized in different regions of the hearts. In the atria and ventricles cardiomyocytes become increasingly dependent on action potential-driven Ca²⁺-induced Ca²⁺ release, and generally exhibit minimal pacemaking behaviour. However, not all regions of the heart exhibit similar maturation. For example, in the adult sinoatrial node, pacemaking resembles the early embryonic phenotype, as it is reliant on spontaneous IP₃- and RyR-mediated Ca²⁺ release as well as I_f current (Vinogradova et al. 2000; Bogdanov et al. 2001; Lakatta & DiFrancesco, 2009; Ju et al. 2011; Sirenko et al. 2014) (Fig.2). Due to such divergent maturation of Ca²⁺ homeostasis across the heart we will, for the sake of clarity, focus on developmental changes in Ca²⁺ handling in ventricular cardiomyocytes in the following sections. These cells undergo dramatic growth and differentiation, with structural specialization and altered expression of Ca²⁺-handling proteins.

Maturation of SR Ca²⁺ release units

An increasing role of Ca²⁺-induced Ca²⁺ release in the developing ventricular myocyte is accompanied by the development of specialized cisternae of SR, called ‘junctional SR’, which bear RyRs (Fig.1B). While SR Ca²⁺ release is largely derived from the perinuclear SR in the early embryo, some couplings have been observed between junctional SR and the plasmalemma even at these early stages (Franzini-Armstrong et al. 2005). These junctions, called ‘dyads’ or ‘Ca²⁺ release units’ (CRUs), contain RyRs in close apposition to sarcolemmal Ca²⁺ channels (Fig.1). Peripheral CRUs are quite rudimentary in the early embryo (Franzini-Armstrong et al. 2005), but probably enable action potential-mediated SR Ca²⁺ release observed in these cells (described above). During embryonic maturation, the peripheral CRUs are enriched with an increasing number of RyRs and Ca²⁺ channels (Franzini-Armstrong et al. 2005). Junctophilin-2 anchors the junctional SR to the plasmamembrane (Fig.1A), to tightly control the geometry of this dyadic junction, and this protein has been shown to be present from early embryonic stages (Takeshima et al. 2000; Takeshima, 2002; Franzini-Armstrong et al. 2005).

While CRUs are restricted to the cell periphery in the embryo and fetus, internal CRUs develop after birth. Rudimentary SR cisternae containing RyRs are positioned at regular intervals in embryonic and neonatal cardiomyocytes, in anticipation of the inward growth of t-tubules (Korhonen et al. 2010; Ziman et al. 2010) (Fig.1A). These invaginations of the surface sarcolemma begin to form around 10 days after birth in mice, and subsequent inward propagation occurs simultaneously with the appearance of junctophilin-2 (Ziman et al. 2010) (Fig.1B). Interestingly, t-tubule density continues to increase until surprisingly late periods of adulthood (Louch et al. 2006). As with the formation of peripheral CRUs in the embryo, maturing internal CRUs exhibit progressive packing with L-type Ca²⁺ channels and RyRs (Franzini-Armstrong et al. 2005). In the fully mature cardiomyocyte, electron and super-resolution microscopy have shown that, while there is significant variation between species, CRUs contain on average about 14 RyRs in rat cardiomyocytes (Baddeley et al. 2009), with a stoichiometry of 4–10 RyRs per L-type Ca²⁺ channel (Bers & Stiffel, 1993). The critical role of junctophilin-2 in establishing the formation of these CRUs was recently demonstrated in a mouse model with junctophilin-2 knockdown (Chen et al. 2013; Reynolds et al. 2013). Inhibited t-tubule formation in these mice was associated with heart failure development and death within a few weeks following birth (Chen et al. 2013; Reynolds et al. 2013).

Developmental changes in sarcolemmal Ca²⁺ fluxes

Consistent with the progressive maturation of peripheral and internal CRUs, L-type Ca²⁺ channel expression and current increase steadily from the early embryonic to neonatal to postnatal phases (Vornanen, 1996; Liu et al. 2002). However, it is important to consider that cell capacitance also increases dramatically during development due to growth of the cell and t-tubules. In normalizing Ca²⁺ currents to cellular capacitance, most reports indicate that Ca²⁺ current density also increases during development (Osaka & Joyner, 1991; Wetzel et al. 1991, 1993; Huynh et al. 1992; Liu et al. 2002) (Fig.2), although Vornanen (1996) observed a modest decline in the late stages of rat development. The progressive formation of CRUs during development also somewhat limits the magnitude of Ca²⁺ current as RyR-mediated Ca²⁺ release enhances Ca²⁺-dependent inactivation of L-type channels (Osaka & Joyner, 1991; Vornanen, 1996). This finding is supported by mathematical modelling (Fig.3C and D).

An increased reliance on L-type Ca²⁺ current (I_CaL ) for triggering SR Ca²⁺ release is associated with progressive loss of T-type Ca²⁺ channels and current (Fig.2), which are present in embryonic (Korhonen et al. 2008), fetal (Cribbs et al. 2001) and neonatal cardiomyocytes (Nuss & Marban, 1994). Even at developmental stages where they are present, T-type currents (I_CaT) are never large due to fairly low channel expression (E9–11: I_CaT, 0.59 ± 0.21 vs. I_CaL, 4.5 ± 0.6 pA pF⁻¹; Korhonen et al. 2008) and negative voltage activation requirements, in combination with less negative resting membrane potential in developing cells (mouse E8.5–10.5: −33.2 ± 1.5 mV, Sasse et al. 2007; mouse E9–11: −57.2 ± 0.9 mV, Rapila et al. 2008; rabbit 21-day fetus: −68 ± 0.5 mV vs. 28-day fetus, −73 ± 0.9 mV, Huynh et al. 1992). Indeed, Leuranguer et al. observed that by the neonatal stage, Ca²⁺ influx via T-type channels in rat cardiomyocytes amounted to less than 20% of the flux through L-type channels, and was undetectable 3 weeks after birth (Leuranguer et al. 2000). A low contribution of this current (7.7%) was also estimated by mathematical modelling (Korhonen et al. 2009). Furthermore, when present, T-type channels have been observed in caveolae, suggesting that their role in Ca²⁺ homeostasis is primarily important for regulation of signalling pathways rather than triggering contraction (Markandeya et al. 2011).

In both developing and adult cardiomyocytes, Ca²⁺ is predominantly extruded by the plasmalemmal NCX (Koban et al. 1998). Thus, as L-type Ca²⁺ current increases during development, greater NCX-mediated Ca²⁺ efflux is required. However, unlike most other Ca²⁺-handling proteins, which exhibit increased expression during development, NCX protein levels decrease after the early embryonic phase (Liu et al. 2002; Reppel et al. 2007). Artman et al. (2000) proposed that this apparent paradox may be explained by changes in the directionality of NCX function, with a progressive dominance of the Ca²⁺ extrusion (forward) mode over Ca²⁺ entry (reverse) mode during development (Artman et al. 2000) (Fig.2). Indeed, marked NCX-mediated Ca²⁺ influx has been observed in immature cells (Escobar et al. 2004; Reppel et al. 2007; Huang et al. 2008), and Korhonen et al. (2009) calculated that 30% of all Ca²⁺ influx can be traced to NCX function in neonatal myocytes. Huang et al. (2008) observed that NCX-mediated Ca²⁺ influx even triggers SR Ca²⁺ release in neonatal rabbit cardiomyocytes. By comparison, in adult cardiomyocytes much less Ca²⁺ influx is attributed to NCX, and NCX-mediated SR Ca²⁺ release has been shown to be only a minor contributor to the Ca²⁺ transient (Sipido et al. 1997; Lines et al. 2006). Thus, despite declining NCX protein levels during development, forward-mode function appears to become predominant in remaining exchangers (Fig.2). This mode of NCX operation is favoured by an increasingly negative resting membrane potential in maturing cells (Huynh et al. 1992; Reppel et al. 2007). Artman et al. (2000) have speculated that progressive reduction in cytosolic [Na⁺] might also occur, but experimental evidence is lacking. Interestingly, Gershome et al. (2011) recently showed high expression of both skeletal and cardiac muscle isoforms of Na⁺ channels (Na_v1.4 and 1.5, respectively) in neonatal cells, with close proximity to NCX (Gershome et al. 2011). Earlier work has illustrated that Na⁺ channel localization near the CRU is a prerequisite for NCX-mediated triggering of SR Ca²⁺ release (Lines et al. 2006).

Maturation of SR function

While there is general agreement that SR Ca²⁺ release is a contributor to the Ca²⁺ transient of embryonic and neonatal myocytes (Haddock et al. 1999; Seth et al. 2004), the dependence of contraction on SR Ca²⁺ release increases markedly after birth (Bassani & Bassani, 2002; Escobar et al. 2004). Indeed, depending on species, between 70 and 90% of the Ca²⁺ transient originates from the SR in adult mammals (Bers, 2001). This increasing role for SR Ca²⁺ release is facilitated by the progressive assembly of RyRs in both peripheral and internal CRUs, as discussed above, and an augmenting ability of L-type Ca²⁺ current to trigger SR release (i.e. the ‘gain’ of Ca²⁺-induced Ca²⁺ release; Huang et al. 2008) (Fig.2). An increasing contribution of SR Ca²⁺ cycling is also supported by a near-linear increase in the expression of other SR components such as SERCA2a from the early embryo until adulthood (Liu et al. 2002) (Fig.2). Increasing SERCA expression is paralleled by elevation of releasable SR Ca²⁺ content. In immature cells, caffeine-elicited Ca²⁺ transients are observed to be somewhat larger (embryo ∼1.5-fold, neonate ∼1.25-fold) than the electrically stimulated Ca²⁺ transient (Rapila et al. 2008; Korhonen et al. 2009). By comparison, caffeine transients in adult cells are ∼2- to 3-fold larger than the amplitude of the Ca²⁺ transient, relative to cytosolic volume (Delbridge et al. 1997; Díaz et al. 1997; Korhonen et al. 2009). However, it is likely that this progressive increase in SR content is not solely due to increased SERCA expression, as there is a simultaneous and dramatic increase in the SR Ca²⁺ storage capacity (Fig.2). The level of calreticulin, which may buffer SR Ca²⁺ in the embryo, is progressively reduced during development (Mesaeli et al. 1999) as it is replaced by calsequestrin (Ioshii et al. 1994). Calreticulin is a relatively high affinity buffer with low capacity (K_d = 1 μm, 1 mol Ca²⁺ (mol protein)^–1) whereas calsequestrin has low affinity but high capacity (K_d = 1 mm, 50 mol Ca²⁺ (mol protein)^–1) (Lee  &  Michalak, 2010). Postnatal calsequestrin expression coincides with increased expression of triadin and junctin (Wetzel et al. 2000; Franzini-Armstrong et al. 2005). These proteins together form a complex that allows precise control of the amount of Ca²⁺ available for RyR-dependent Ca²⁺ release, the Ca²⁺ sensitivity of the RyR, and the set-point for termination of Ca²⁺ release (Györke I et al. 2004; Györke S et al. 2009). The key role for calsequestrin in these processes in adult cardiomyocytes is illustrated by the fact that mutations or changes in the amount of calsequestrin predispose cells to abnormal spontaneous SR Ca²⁺ release and triggered arrhythmias (Viatchenko-Karpinski et al. 2004; Tavi et al. 2005). By comparison, altered calsequestrin expression in neonatal cardiomyocytes does not alter (Gergs et al. 2011) or modestly alters (Hanninen et al. 2010) myocyte Ca²⁺ signalling, suggesting that the effect of calsequestrin depends on the stage of differentiation.

While formation of internal CRUs occurs quite late during ventricular cardiomyocyte development, it is important to point out that from early embryonic stages internal, ‘orphaned’ RyRs are already present in corbular SR, which is not associated with the sarcolemma (Fig.1A). These SR structures occur at regular ∼2 μm intervals throughout the cytosol, and allow propagation of the cytosolic Ca²⁺ signal by the triggering of consecutive Ca²⁺ sparks (Korhonen et al. 2010). This ‘fire-diffuse-fire’ mechanism (Lipp et al. 1990; Trafford et al. 1995; Cheng et al. 1996; Keizer et al. 1998) ensures rapid propagation of the Ca²⁺ signal over large cytosolic distances to meet the demands of contraction (Korhonen et al. 2010). In the absence of such adaptations, [Ca²⁺] at the front of the propagating Ca²⁺ signal would dilute, the velocity of diffusion would decrease in proportion to the diffusion distance, and the Ca²⁺ signal would have limited range. Thus, with a regular arrangement of internal release sites, a regenerative wave of propagating Ca²⁺ release is established, as indicated by the in silico line scan illustrated in Fig.4A (top). With initiation of SR Ca²⁺ release on both sides of the cell, a stereotypical ‘U-shaped’ Ca²⁺ transient is observed in confocal line scan images of embryonic myocytes (Rapila et al. 2008), neonatal cardiomyocytes (Haddock et al. 1999; Seki et al. 2003) (Fig.4A, bottom), and cardiomyocytes derived from embryonic stem cells (Kapur & Banach, 2007; Satin et al. 2008). The delay inherent in propagation of Ca²⁺ release to the centre of the cell results in a spatially averaged Ca²⁺ transient with a prolonged time course (Fig.3A). Loose functional coupling between sarcolemmal and internal CRUs also creates distinct cytosolic Ca²⁺ gradients across immature cells. For example, in early embryonic myocytes perinuclear SR Ca²⁺ release creates larger transients in these regions than in other parts of the cytosol (Rapila et al. 2008). Furthermore, ‘orphaned’ RyRs in corbular SR are also prone to trigger spontaneous Ca²⁺ events ranging from local Ca²⁺ sparks to propagating Ca²⁺ waves, which consequently predispose cells to irregular membrane voltage changes (Sasse et al. 2007; Rapila et al. 2008). While U-shaped Ca²⁺ transients are sufficient to meet the contractile demands of small, prenatal cardiomyocytes, more synchronized Ca²⁺ release is required for growing postnatal cells. Such synchronization of activation of CRUs in maturing cells is enabled by t-tubule development (Fig.1B and C), which carries the depolarizing action potential into the interior of the cardiomyocyte, and minimizes dependence on the ‘fire-diffuse-fire’ paradigm (Figs 2 and 4A, right). The regularity of CRU organization in adult cardiomyocytes varies between species, and shows an apparent correlation with animal size and heart rate. In small species with high heart rates such as mice, a dense t-tubule network in adult cells ensures near simultaneous RyR Ca²⁺ release, while a somewhat less homogeneous Ca²⁺ transient is observed in larger species such as pig and human (Louch et al. 2004).

Figure 4.

An organized t-tubule network enables synchronous Ca²⁺ release

In neonatal cardiomyocytes, which lack t-tubules, SR Ca²⁺ release is initiated at the cell periphery and then propagates into the cell interior as a propagating Ca²⁺ wave, the ‘fire-diffuse-fire’ paradigm. Mathematical modelling predicts delayed release of Ca²⁺ at the cell interior, manifested as ‘U-shaped’ Ca²⁺ transients recorded by experimental line scan imaging (A, left). Growth of t-tubules during postnatal development places CRUs across the cell interior, which enables more synchronous Ca²⁺ release during the action potential (A, right). Confocal images of t-tubule structure (di-8-ANEPPS staining) show that the well-organized t-tubules of healthy adult mouse cardiomyocytes are markedly disrupted during heart failure (B, top). This disorganization includes the appearance of abnormal gaps between tubules. Simultaneous imaging of t-tubules and Ca²⁺ with longitudinally drawn line scans (B, bottom) shows that Ca²⁺ release occurs early following the stimulus at locations near t-tubules, which are visible as horizontal lines. In failing cells, delayed Ca²⁺ release occurs at abnormal gaps between t-tubules, which de-synchronizes and slows the Ca²⁺ transient. Experimental data in A and B reproduced from Haddock et al. (1999) and Louch et al. (2013), respectively; SL, sarcolemma.

While increasing importance is attached to RyR-mediated Ca²⁺ release during development, the important role of IP₃-mediated Ca²⁺ release observed in the embryo is not completely abandoned. A functional interaction between IP₃Rs and RyRs remains robustly present in neonatal and juvenile cardiomyocytes (Janowski et al. 2010) (Fig.1B), and recent evidence suggests that such an interaction may also remain into adulthood, at least in atrial cells (Zima & Blatter, 2004). In adult cardiomyocytes, IP₃ receptors play a particularly important role in modulating nuclear Ca²⁺ signals, and recent evidence indicates that the nuclear envelope, which also contains RyRs, is a Ca²⁺ storage organelle which is contiguous with the SR (Hohendanner et al. 2014). While IP₃-mediated Ca²⁺ signals are known to play an important role in non-contractile signalling, it is possible that these signals may also fine-tune RyR function and contractility.

Developmental changes in cytosolic Ca²⁺ buffering

The above discussion has illustrated that cardiomyocytes at all stages of development have specialized structures that facilitate propagation of Ca²⁺ signals. The importance of these changes becomes even more apparent when considering the cytosolic buffering of Ca²⁺. While the diffusion of most ions (K⁺, Na⁺, Cl⁻) in the cytosol is relatively fast, with rates only 2-fold slower than that in water, the presence of Ca²⁺ buffers dramatically slows diffusion of this ion. Even at short distances (<15 μm) the average speed of Ca²⁺ diffusion is approximately 30 ms μm⁻¹ (Korhonen et al. 2010) and is quadratically slowed upon an increase in distance. The largest cytosolic Ca²⁺ buffers in the adult cardiomyocyte are the Ca²⁺-binding proteins of the contractile element troponin-C (Creazzo et al. 2004) and the SR (Kushmerick & Podolsky, 1969). Thus, as the contractile elements assemble during development to fulfil the demand for more forceful contraction, Ca²⁺ buffering is greatly augmented (Fig.2). This increase in cytosolic buffering both shapes the Ca²⁺ transient and defines how much Ca²⁺ is needed to elevate cytosolic [Ca²⁺]. It has been estimated that raising free [Ca²⁺]_i from 100 nm to 1 μm in neonatal myocytes requires 88 μmol Ca²⁺ (l cytosol)^–1, whereas a similar magnitude Ca²⁺ transient in an adult cardiomyocyte requires a 2-fold larger total amount of Ca²⁺ (149 μmol (l cytosol)^–1) (Bassani et al. 1998). Based on a similar comparison of buffering capacity between embryonic and neonatal cardiomyocytes (Korhonen et al. 2008, 2009), it appears that the cytosolic buffering remains relatively unchanged throughout development until the drastic increase after birth (Fig.2).

Ca²⁺ signals for cardiomyocyte differentiation

The above discussion has illustrated the complex evolution of cardiomyocyte Ca²⁺ homeostasis that occurs through embryonic, fetal and postnatal stages of development. The paradigm that the collection of cellular structures and Ca²⁺-handling proteins present at each stage are required for further development is supported by data from genetically manipulated mouse models. Genetic deletion or suppression of any of the key components of Ca²⁺ signalling produces phenotypes with severe heart malformations and functional impairments that induce early embryonic lethality. For example, deletion of the genes that code for the L-type Ca²⁺ channel was observed to reduce cytosolic Ca²⁺ signals, slow heart rate, and compromise heart function as early as E10.5, with mortality occurring soon afterward (Weissgerber et al. 2005). Similarly, the central role of NCX in embryonic pacemaking is illustrated by the fact that NCX-deficient mice lack spontaneous heartbeats and organized contractile myofibrils during embryogenesis, with lethality observed before E11 (Koushik et al. 2001). In addition to the expression of plasmalemmal ion channels, it seems that functional SR is also required for cardiogenesis. Mice deficient in the cardiac RyR isoform RyR2 did not exhibit Ca²⁺ transients in embryonic cardiomyocytes, developed abnormal SR and mitochondrial structure, and died at approximately E10 (Takeshima et al. 1998). Similarly, systemic disruption of SERCA2 is lethal in the embryonic stage (Periasamy et al. 1999), and inducible deletion of SERCA2 leads to death of the embryo very soon after activation of the transgene (Andersson et al. 2009). Further supporting the role of SR in embryonic cardiac Ca²⁺ signalling, mice deficient in calreticulin, the major embryonic SR Ca²⁺ buffer, showed embryonic malformations, impaired Ca²⁺ signalling, and compromised Ca²⁺-dependent transcription (Mesaeli et al. 1999).

While it is clear that Ca²⁺ transients are critical for triggering the heartbeat in the developing embryo, precise control of Ca²⁺ handling is also essential for regulation of transcription, protein modifications, metabolism and growth necessary for cardiogenesis. A plethora of signalling molecules are regulated by Ca²⁺, and these Ca²⁺ ‘decoders’ are known to be sensitive to the location, size and frequency of Ca²⁺ signals (Dolmetsch et al. 1997; Tavi et al. 2004; Saucerman & Bers, 2008; Tavi & Westerblad, 2011). One such Ca²⁺ sensor is Ca²⁺–calmodulin-dependent protein phosphatase 2B (calcineurin). Calcineurin activates nuclear factor of activated T-cells (NFAT) to interact with other transcription factors in the nucleus (Molkentin et al. 1998), and the calcineurin–NFAT pathway is specifically required for morphogenesis of cardiac valves and septa (de la Pompa et al. 1998; Ranger et al. 1998). During mouse cardiogenesis NFAT is first activated at E9 when it transforms endothelial cells into mesenchymal cells; a second activation of NFAT in the endocardium takes place at E11 when it is required for valvular elongation and refinement (Chang et al. 2004). Throughout a mammal's lifespan the calcineurin–NFAT pathway is essential for the regulation of muscle cell growth and the expression of muscle-specific genes (Crabtree & Olson, 2002), and lack of NFAT results in embryonic lethality due to impaired Ca²⁺ signals, compromised mitochondrial oxidative phosphorylation and retarded cardiac growth (Bushdid et al. 2003).

Ca²⁺ signals also activate calmodulin-dependent kinases (CaMK) that are known to stimulate several transcription factors, such as CaMK-dependent myocyte-enhancing-factor 2 (MEF2a, -b, -c and -d isoforms). These transcription factors control myogenesis, muscle hypertrophy and mitochondrial function (Frey et al. 2000). Most mice lacking MEF2a die suddenly within the first week of life and exhibit pronounced dilatation of the right ventricle, myofibrillar fragmentation, mitochondrial disorganization and activation of a fetal cardiac gene programme (Naya et al. 2002). This is perhaps not surprising since MEF2 activity is required early in the differentiation of cardioblasts into cardiomyocytes, and later during morphogenesis of the left and right ventricles (Srivastava & Olson, 2000).

In addition to activation of individual transcription factors, cardiac development is also guided by epigenetic modifications which control expression patterns of genes such as those associated with cell commitment to cardiac lineage (Wamstad et al. 2012). Epigenetic modifications are mediated in cardiac myocytes by various chromatin-modifying enzymes such as Brg1/Brm-associated factor complex, poly(ADP-ribose) and histone deacetylases (HDACs) (Hang et al. 2010). Of these, HDACs have been shown to be to particularly important targets of Ca²⁺-activated enzymes such as CaMK (McKinsey et al. 2000), thus further establishing a link between Ca²⁺ signals and control of genetic programmes. The effects of HDACs depend on their cellular localization; when they are in the nucleus they condense chromatin by deacetylating the histones, and repress the expression of their target proteins (Kuo & Allis, 1998). Phosphorylation of HDAC enables chaperone protein (14-3-3) binding and subsequent nuclear export of the HDAC–chaperone complex relieves HDAC-mediated gene repression (Backs & Olson, 2006; Bossuyt et al. 2008). In adult cardiomyocytes, class IIa HDACs are mainly localized in the nucleus where they act to suppress transcription, and thus fetal gene expression. These genes are activated upon HDAC exportation from the nucleus, leading to development of cardiac hypertrophy (Ito et al. 1991; Sucharov et al. 2006). Similarly, HDAC nuclear export and subsequent de-repression of the same fetal gene pattern is a prerequisite for differentiation of developing cardiomyocytes (Karamboulas et al. 2006). While the precise manner in which Ca²⁺ homeostasis shapes HDAC-dependent gene expression remains unclear, it appears that these Ca²⁺ signals are insulated from global Ca²⁺ transients, at least in adult cells (Zima et al. 2007). Indeed, these pathways can be triggered even if global Ca²⁺ transients are normal (Louch et al. 2012). Instead, HDAC5 nuclear export is reported to be regulated by IP₃-receptor-induced Ca²⁺ release from perinuclear SR (Zima et al. 2007) which increases nuclear [Ca²⁺] and induces HDAC5 phosphorylation (Wu et al. 2006). It has also been speculated that elevation of dyadic Ca²⁺ levels may activate HDAC and NFAT signalling, as these pathways are triggered following increases in SR Ca²⁺ leak or L-type Ca²⁺ current in adult cells (Goonasekera & Molkentin, 2012; Kalyanasundaram et al. 2013; Louch & Lyon, 2013). Although, it remains unknown whether similar local Ca²⁺ signals regulate transcription during development, it should be pointed out that differentiation is associated with progressive upregulation of both the L-type Ca²⁺ channel and the RyR, as discussed above. Deciphering the precise nature by which local Ca²⁺ signals code cardiomyocyte development represents a considerable challenge, but may be enabled by emerging techniques for measuring Ca²⁺ handling in microdomains such as the dyad (Shang et al. 2014).

Cardiac disease: reversion to an embryonic/fetal phenotype

The above discussions have illustrated that much work is required to precisely identify the mechanisms by which Ca²⁺ signalling controls both the heartbeat and gene expression during development. These issues also have central relevance for understanding cardiac disease, in which fetal genes are re-expressed and adult genes repressed. This altered genetic profile is induced as a response to biomechanical stress, including pressure or volume overload, and other stimuli such as reactive oxygen species which may contribute to disease progression (Hafstad et al. 2013). The details of these mechanisms are largely beyond the scope of this review, but are known to include transcriptional, post-transcriptional and epigenetic regulation of the cardiac genome. Key players include the HDAC and NFAT pathways, as discussed above, but also regulators of DNA and histone methylation, and a host of microRNAs (Duygu et al. 2013).

With re-expression of fetal genes during disease, it is perhaps not surprising that there are striking similarities in the phenotype of diseased and developing cardiomyocytes (Fig.2). As described above, dyadic structure is assembled gradually during development; t-tubules develop after birth and are anchored by junctophilin-2 to the SR to form CRUs. Recent evidence indicates that junctophilin-2 is lost during heart failure (Xu et al. 2007; Wei et al. 2010), causing t-tubules to ‘drift’ out of CRUs (Louch et al. 2006; Song et al. 2006; Swift et al. 2008; Wei et al. 2010; Oyehaug et al. 2013) and t-tubule density to be reduced (Heinzel et al. 2008; Swift et al. 2008; Wei et al. 2010) (Figs 2, 1D and 4B). Similar t-tubule disruption has been reported in atrial fibrillation (Lenaerts et al. 2009) and diabetic cardiomyopathy (Stolen et al. 2009). Mechanisms proposed to underlie junctophilin-2 downregulation or mislocalization during disease include increased expression of microRNA-24 (Xu et al. 2012; Li et al. 2013), altered activity of phosphoinositide 3-kinases (Wu et al. 2011) and increased mechanical load (Ibrahim & Terracciano, 2013). With disorganization and/or loss of t-tubules, some RyRs become ‘orphaned’, and are activated only by Ca²⁺ diffusing from intact CRUs (Louch et al. 2004, 2006; Song et al. 2006) (Fig.4B). This ‘fire-diffuse-fire’ paradigm of Ca²⁺ release resembles that present in immature cardiomyocytes (Korhonen et al. 2010). In the setting of heart failure, the de-synchronized and slowed Ca²⁺ transient contributes to slowing of contraction and reduced power of the heartbeat (Bokenes et al. 2008; Mørk et al. 2009).

In addition to t-tubule reorganization, recent data indicate that more subtle changes in CRU structure also occur during disease. We have observed that there is a marked slowing of Ca²⁺ spark kinetics during heart failure, which was predicted to result from drift and/or loss of RyRs from the CRU (Louch et al. 2013). Furthermore, loss of L-type Ca²⁺ channels has been reported in heart failure (He et al. 2001). Thus, there may be an ‘unpacking’ of RyRs and Ca²⁺ channels during this disease (Fig.1D), which is a reversal of the CRU filling process that occurs during development (Franzini-Armstrong et al. 2005). Such changes would be expected to contribute to reduced gain of Ca²⁺-induced Ca²⁺ release observed in failing cardiomyocytes (Gomez et al. 2001), and de-synchronization of Ca²⁺ release across the cell (Louch et al. 2013). Of note, L-type Ca²⁺ current densities are generally reported to be of normal magnitude in heart failure (Louch et al. 2010b) (Fig.2), despite loss of Ca²⁺ channels. This discrepancy has been attributed to increased single channel activity (Schröder et al. 1998).

While there is a distinct loss of the striated, transverse pattern of t-tubule structure during disease, it is interesting to note that an increased fraction of longitudinally oriented tubules is frequently reported (Louch et al. 2006; Song et al. 2006; Swift et al. 2012) (Fig.4B). Recent evidence indicates that these longitudinal tubules are actually grown during disease and are not simply repositioned transverse elements (Swift et al. 2012). Such t-tubule growth may result from re-expression of prenatal genes since longitudinal tubules are overtly present in fetal skeletal muscle (Franzini-Armstrong, 2002). We observed that longitudinal tubules contain NCX but not L-type Ca²⁺ channels (Swift et al. 2012). Indeed, NCX upregulation is a common feature of diseased cardiomyocytes (Ottolia et al. 2013) (Fig.2). Interestingly, dyadic junctions are formed between newly grown longitudinal tubules and SR, with co-localization of NCX and RyRs (Swift et al. 2012). Based on these data, mathematical modelling suggested a role for NCX-triggered Ca²⁺ release at these sites. In both diseased and developing cells, lower expression of the Na⁺–K⁺ ATPase, particularly the α2 isoform, is associated with elevated intracellular Na⁺ levels, which favours the Ca²⁺ entry mode of NCX function (Fozzard & Sheu, 1980; Herrera et al. 1994; Artman et al. 2000; Swift et al. 2008; Louch et al. 2010a; Li et al. 2012).

Reduced gain of Ca²⁺-induced Ca²⁺ release during heart failure does not solely result from t-tubule disruption and loss of L-type Ca²⁺ channels, but also from reduction in SR Ca²⁺ content (Louch et al. 2010b). Although calsequestrin expression declines during heart failure (Yeh et al. 2008; Hu et al. 2011), Guo et al. (2007) observed no change in SR buffering (Fig.2). Instead, reduction in SR content has been attributed to decreased SERCA expression, and increased RyR-mediated Ca²⁺ leak due to increased phosphorylation of the channel (Bers, 2014). Indeed, the ‘pacemaking phenotype’ of immature cardiomyocytes is revisited in ventricular cardiomyocytes during disease, as IP₃-receptor expression is increased (Go et al. 1995) and the pacemaker current, I_f, is also re-expressed (Suffredini et al. 2007) (Fig.2). As in the embryonic heart, spontaneously released Ca²⁺ is removed by NCX, generating depolarizing current and action potentials. While this mechanism is central to pacemaking in the embryonic heart, it underlies increased arrhythmogenesis during heart failure (Bers, 2014). Thus, inhibition of the pacemaking phenotype during heart failure is an important therapeutic goal.

Interestingly, T-type Ca²⁺ channels are re-expressed in heart failure, which is reminiscent of the prenatal heart (Korhonen et al. 2008; Cribbs, 2010). Proposed triggers for T-type channel re-expression include insulin-like growth factor-1 (Larsen et al. 2005), endothelin-1 (Furukawa et al. 1992) and angiotensin II (Bkaily et al. 2005). Although probably a minor contributor to overall Ca²⁺ influx in diseased cardiomyocytes, T-type Ca²⁺ current has been suggested to promote arrhythmogenesis (Kinoshita et al. 2009) and contribute to local Ca²⁺ pools responsible for activating the NFAT pathway, triggering hypertrophy (Chiang et al. 2009).

Developing cardiomyocytes as disease models

The similarity between developing and diseased cardiomyocytes has been used as a rationale for employing immature cells as disease models. Neonatal cardiomyocytes and immortalized cardiomyocyte cell lines such as HL-1 cells are commonly used model systems, since a dividing cell population can be maintained and manipulated relatively easily in cell culture. However, it is important to point out that drawing comparisons between diseased and developing cells is primarily an approach to help understand the mechanisms underlying pathophysiology. In reality, cardiomyocyte cell lines exhibit a phenotype that is much less mature than that of adult cardiomyocytes at even the most advanced disease stages. There are, for example, striking differences in cellular geometry; neonatal and HL-1 cardiomyocytes are nearly spherical while diseased ventricular cells remain rod-shaped. During disease, changes in CRU structure also exhibit a reversal of the developmental process (t-tubule loss, dyadic ‘unpacking’) but retain a far more mature phenotype than that present in the prenatal or neonatal cardiomyocyte, where internal CRUs are absent. Overall, the diseased Ca²⁺-handling paradigm is reminiscent of a stage intermediate between the neonatal and adult phenotype, such as that occurring 1–2 weeks after birth in rodents. Unfortunately, cardiomyocytes at this stage no longer divide, limiting their utility as a model system. Indeed, regular isolation of cardiomyocytes from 1- to 2-week-old hearts is probably as labour-intensive as employment of an adult disease model, and of less relevance for investigation of disease pathophysiology. It seems likely, therefore, that dividing cell lines will remain a more commonly used disease model. We suggest that while these cell lines present a valuable investigational tool, these results are best interpreted alongside data from adult models of cardiac disease.

With recent developments in stem cell research, cardiomyocytes are currently being produced from embryonic stem cells, induced pluripotent stem cells (Takahashi et al. 2007), and even through direct reprogramming of somatic cells (Miki et al. 2013). These stem cell-derived cardiomyocytes are gaining popularity as they may offer accessible models for drug screening and for studying inherited human cardiac diseases (Priori et al. 2013). Despite ongoing efforts to improve differentiation of these cells, the attained phenotype remains immature (Poon et al. 2011). Furthermore, there appear to be notable differences in the physiology of naturally developing and stem cell-derived cardiomyocytes, particularly regarding pacemaking mechanisms. Data from embryonic cardiomyocytes suggest that whole cell Ca²⁺ oscillations (the ‘Ca²⁺ clock’) result from co-activation of both RyRs and IP₃Rs (Sasse et al. 2007; Rapila et al. 2008; Wang et al. 2013). In contrast, results from stem cell-derived cardiomyocytes point exclusively to IP₃R-driven oscillations (Mery et al. 2005; Kapur & Banach, 2007; Satin et al. 2008), and indicate that RyR- and IP₃R-dependent Ca²⁺ release rely on separate sarcoplasmic reticulum and endoplasmic reticulum Ca²⁺ stores, respectively (Kapur & Banach, 2007; Kapoor et al. 2014). This difference may explain why RyR-induced Ca²⁺ release induces robust NCX current in early embryonic cardiomyocytes (Sasse et al. 2007; Rapila et al. 2008; Wang et al. 2013), whereas similar NCX current activation in stem cell-derived cardiomyocytes requires Ca²⁺ release from IP₃ receptors (Kapoor et al. 2014). Thus, different Ca²⁺ pools and release channels may control the Ca²⁺ clock in these two cells types. Based on these differences, we urge caution in extrapolating data from embryonic stem cells to naturally developing cells, particularly when drawing comparisons to mature cardiomyocytes.

Conclusions

The above discussion has highlighted our evolving understanding of cardiomyocyte Ca²⁺ signalling during development, and emphasized the critical role that cytosolic Ca²⁺ fluxes play in regulating a plethora of processes during maturation. Understanding the details of these processes is important not only for those studying development, but also those employing immature cardiomyocyte models such as neonatal cell cultures and stem cells. Diseased cardiomyocytes exhibit a striking regression to an immature state of Ca²⁺ homeostasis, indicating that greater insight into disease mechanisms can also be gained by investigating developmental physiology and the fetal gene programme. Recent developments in stem cell research have also put stem cell-derived cardiomyocytes forward as potential and accessible models for drug screening and for studying inherited human cardiac diseases. As large challenges remain in differentiation of these cells, they should be used as disease models with great caution, taking into account the functional consequences of the phenotypic differences.

Abbreviations

CaMK

calmodulin-dependent kinase

CRU

Ca²⁺ release unit

HDAC

histone deacetylase

I_f

pacemaker current

IP₃R

inositol-3-phosphate receptor

MEF2

myocyte-enhancing factor 2

NCX

Na⁺/Ca²⁺ exchanger

NFAT

nuclear factor of activated T-cells

RyR

ryanodine receptor

SERCA

sarcoplasmic reticulum Ca²⁺ ATPase

SR

sarcoplasmic reticulum

Biographies

William E. Louch received his PhD in Pharmacology in 2001 from Dalhousie University in Halifax, Canada, and is currently Research Associate Professor at Oslo University Hospital/University of Oslo in Norway. His research examines structure and function of normal and diseased cardiac myocytes,with particular focus on cellular Ca²⁺ homeostasis.

Jussi T.Koivumäki received his PhD in Biophysics in 2009 at the University of Oulu, Finland. In his present position as a postdoctoral fellow at Simula Research Laboratory, Oslo, Norway, he employs mathematical modelling to investigate disease mechanisms of atrial fibrillation and heart failure.

Pasi Tavi received a PhD in Physiology and Biophysics in 1999 from the University of Oulu, Finland. He is currently Professor in Cardiovascular Cell Physiology at the University of Eastern Finland, A.I. Virtanen Institute for Molecular Sciences, Kuopio, Finland. His research is in the areas of cardiac cell physiology, focusing on alterations in cardiac myocyte excitation–contraction coupling in development and disease progression.

Additional information

Competing interests

None declared.

Funding

William E. Louch received generous funding from the South-Eastern Norway Regional Health Authority, The Research Council of Norway, Anders Jahre's Fund for the Promotion of Science, Oslo University Hospital Ullevål, University of Oslo and European Union Project [Grant number FP7-HEALTH-2010.2.4.2-4 (MEDIA-Metabolic Road to Diastolic Heart Failure)].