[ Clive Orchard (left) is Professor of Physiology at the University of Bristol. His research interests are in cardiac excitation–contraction coupling and its regulation. Most recently this has focused on the role of the t-tubules in ventricular myocyte function during normal and pathological conditions. Using a technique developed in his lab to detubulate ventricular myocytes, work from his lab has shown that t-tubules form a specialised microdomain for calcium handling, excitation–contraction coupling, and its regulation. Andrew James (right) is a Senior Lecturer at the University of Bristol. His research interests centre on the electrophysiology of the heart and the basis to arrhythmias, which his lab studies both at the cellular level and in intact perfused hearts. Work in his lab is focused on atrial electrophysiology and the disease remodelling that underlies susceptibility to atrial fibrillation. He also has an interest in the regulation of cardiac cellular electrophysiology by G-protein-coupled receptors.]
Abstract The transverse (t-) tubules of mammalian ventricular myocytes are invaginations of the surface membrane. The function of many of the key proteins involved in excitation–contraction coupling is located predominantly at the t-tubules, which thus form a Ca2+-handling micro-environment that is central to the normal rapid activation and relaxation of the ventricular myocyte. Although cellular arrhythmogenesis shares many ion flux pathways with normal excitation–contraction coupling, the role of the t-tubules in such arrhythmogenesis has not previously been considered. In this brief review we consider how the location and co-location of proteins at the t-tubules may contribute to the generation of arrhythmogenic delayed and early afterdepolarisations, and how the loss of t-tubules that occurs during heart failure may alter the generation of such arrhythmias, as well as contributing to other types of arrhythmia as a result of changes of electrical heterogeneity within the whole heart.
Transverse (t-) tubules constitute a specialised region of the cell membrane of mammalian cardiac ventricular myocytes, which plays a key role in excitation–contraction coupling and its associated ion fluxes (for reviews see Orchard & Brette, 2008; Orchard et al. 2009). However, although cellular arrhythmogenesis shares many ion flux pathways with normal excitation–contraction coupling, the role of the t-tubules in arrhythmogenesis in ventricular myocytes has not previously been considered. This brief review therefore analyses the contribution that t-tubules may make to arrhythmogenesis in normal ventricular myocytes and in myocytes from failing hearts.
The role of t-tubules in cardiac excitation–contraction coupling
Contraction of a cardiac myocyte is normally initiated by an action potential, which causes L-type Ca2+ channels (LTCCs) in the cell membrane to open. The resulting Ca2+ influx (ICa) activates ryanodine receptors (RyRs) in adjacent sarcoplasmic reticulum (SR) membrane, thereby triggering Ca2+ release from the SR via calcium-induced calcium release (CICR), resulting in a rapid increase of cytoplasmic [Ca2+] (the systolic Ca2+ transient), and hence contraction. Relaxation occurs as a result of Ca2+ re-uptake into SR via a Ca2+ ATPase (SERCA), and Ca2+ extrusion from the cell via Na+/Ca2+ exchange (NCX) and, to a lesser extent, a sarcolemmal Ca2+ ATPase. In mammalian ventricular myocytes, the presence of transverse (t-) tubules – invaginations of the surface membrane – ensures synchronous Ca2+ release, and rapid Ca2+ extrusion from throughout the cell. It is worth noting that t-tubules have also been described in atrial cells from larger mammals (Dibb et al. 2009); although the t-tubule network appears to be sparser in these cells, the following considerations may also be applicable to cellular arrhythmogenesis in atrial myocytes from such species (Dibb et al. 2013).
Transverse tubules occur at each Z-line and form a complex network within the cell, with both transverse and longitudinal elements. This network appears to be denser in small mammals, which have higher heart rates, commensurate with its role in ensuring rapid activation and relaxation. During the last few years it has become apparent that t-tubules are distinct from the surface sarcolemma both structurally, in terms of lipid and protein composition, and functionally, in terms of ion fluxes and excitation–contraction coupling. Using osmotic shock to detach the t-tubules physically and functionally from the membrane at the cell surface (detubulation), we have shown that they are specialised for trans-sarcolemmal ion flux, particularly of Ca2+, while immunohistochemical studies have shown that most RyRs occur in close juxtaposition to the t-tubules (Kawai et al. 1999; Brette et al. 2002; Jayasinghe et al. 2009).
Detubulation results in an ∼40% decrease of cell capacitance (a measure of membrane area), and an ∼80% decrease of ICa, an ∼65% decrease of NCX current (INCX), and complete loss of Ca2+ extrusion via the sarcolemmal Ca2+ ATPase (Kawai et al. 1999; Yang et al. 2002; Chase & Orchard, 2011). Thus the majority of Ca2+ influx via ICa occurs close to RyRs, which enables local control of Ca2+ release. Immunohistochemical studies suggest that SERCA is also located close to the t-tubules. The functional relevance of having Ca2+ removal pathways located close to the site of Ca2+ entry and release is unclear; NCX and sarcolemmal Ca2+ ATPase in the t-tubule membrane will help ensure rapid Ca2+ removal from throughout the cell, thus speeding relaxation (other consequences are discussed below). However, the presence of SERCA near the site of Ca2+ release would be expected to lead to futile Ca2+ cycling, and thus be energetically unfavourable, although the extent to which this occurs will depend on competition with other Ca2+ removal pathways and the rate at which Ca2+ diffuses away to the bulk cytoplasm.
The disproportionately high percentage of each Ca2+ flux across the t-tubule membrane compared with the fraction of the cell membrane located within the t-tubules could be due either to concentration in the t-tubule membrane of the proteins mediating ion fluxes and/or to local upregulation of protein function at the t-tubules. By inhibiting protein kinase A (PKA) in intact and detubulated myocytes we have shown that ICa and INCX, but not sarcolemmal Ca2+ ATPase, appear to be locally upregulated at the t-tubules by basal PKA activity, and that this can account, in part, for the localisation of these currents to the t-tubules (Chase et al. 2010; Chase & Orchard, 2011).
The role of t-tubules in spontaneous Ca2+ release and delayed afterdepolarisations
Although an increase of cytoplasmic Ca2+ concentration is normally triggered by the action potential, it is well known that spontaneous SR Ca2+ release can occur when the myocyte, and hence SR, becomes overloaded with Ca2+ (Orchard et al. 1983). This release normally arises in one part of the cell, and propagates throughout the cell, between clusters of RyRs, as a wave of CICR. Such Ca2+ release is associated with transient depolarisation of the cell membrane as a result of Ca2+-activated inward currents, including inward INCX (Kass et al. 1978; Mechmann & Pott, 1986; Schlotthauer & Bers, 2000). Such depolarisations can occur between action potentials (delayed afterdepolarisations; DADs) and, if sufficiently large, can trigger extra-systolic action potentials, and thus arrhythmias. The t-tubules may play an important role in the generation of such arrhythmias for the following three main reasons.
Firstly, Ca2+ sparks – spontaneous local Ca2+ release from a cluster of RyRs – occur predominantly at t-tubules (Cannell et al. 1995; Cheng et al. 1995). Acute detubulation results in loss of Ca2+ sparks from the centre of the cell, although they continue to occur at the cell edge (Brette et al. 2005). Why proximity of surface membrane (including t-tubules) increases the probability of Ca2+ sparks is unknown. It is unlikely to be due to local background Ca2+ entry because Ca2+ sparks occur in the absence of extracellular Ca2+. It is also unlikely to be due to regional differences in SR Ca2+ content, because Ca2+ appears to equilibrate quickly between different regions of the SR (Shannon et al. 2003), consistent with which, caffeine causes uniform Ca2+ release across the cell width in detubulated myocytes (Brette et al. 2005). In atrial cells (which have few or no t-tubules) a peptide representing the C-terminus of the LTCC can activate RyRs in the centre of the cell (Woo et al. 2003) so that physical interaction of the LTCC with RyRs may play a role. Whatever the mechanism, the t-tubules are an important site for the genesis of Ca2+ sparks because of the density and activation of adjacent RyRs. Such local Ca2+ releases are the initiating event for propagated waves of CICR that can cause arrhythmias.
Secondly, INCX occurs predominantly at the t-tubules, close to the site of SR Ca2+ release; indeed, NCX may exist within the dyad, although not exclusively concentrated there (Despa et al. 2003; Jayasinghe et al. 2009; Chase & Orchard, 2011). One consequence of this location is that the relationship between INCX and bulk cytoplasmic [Ca2+] during SR Ca2+ release shows hysteresis: INCX is larger for a given cytoplasmic [Ca2+] while Ca2+ is increasing than when it is declining (Trafford et al. 1995; Acsai et al. 2011), suggesting that NCX is exposed to a higher [Ca2+] than that in bulk cytoplasm during SR Ca2+ release. This is consistent with SR Ca2+ release occurring adjacent to t-tubules, where high local [Ca2+] stimulates INCX thereby generating arrhythmogenic inward current.
Thirdly, t-tubules are the major site for transmembrane Ca2+ fluxes, and might thus be expected to play a role in normal Ca2+ balance and, conversely, in the development of the Ca2+ overload that underlies spontaneous SR Ca2+ release. It has been suggested that t-tubules play an important role in ‘autoregulation’ of Ca2+ (Orchard & Brette, 2008), whereby an increase in SR Ca2+ release decreases Ca2+ influx by enhancing Ca2+-dependent inactivation of ICa, and increases efflux via NCX, thus restoring Ca2+ balance (Trafford et al. 1998). Since Ca2+-dependent inactivation of ICa by SR Ca2+ release is more marked at the t-tubules than at the surface membrane (Brette et al. 2004b) and Ca2+ extrusion via INCX appears to be efficiently stimulated by SR Ca2+ release via RyRs, and both occur predominantly at the t-tubules (above), it appears likely that the t-tubules are the major site for such autoregulation.
Although detubulation has little effect on SR Ca2+ load at relatively low stimulation rates and/or following a period of rest (i.e. when the SR has time to complete reloading; Brette et al. 2005), because Ca2+ influx and efflux decrease by approximately the same amount, the response to interventions that change the Ca2+ load of the cell by altering trans-sarcolemmal Ca2+ flux does appear to change. For example, the increase of intracellular Ca2+ caused by either the cardiac glycoside strophanthidin or by increasing stimulation rate, is reduced in detubulated myocytes (Fowler et al. 2004). Both of these interventions increase intracellular Na+ which acts via NCX to increase intracellular Ca2+. Following detubulation, the rise of Na+ is not significantly altered, but is not translated into a rise of intracellular Ca2+, presumably because of loss of NCX (Fowler et al. 2004). Since DADs can be associated with either of these interventions, loss of t-tubules may be protective (see below).
The role of t-tubules in early afterdepolarisations
Early afterdepolarisations (EADs) are abnormal depolarisations of the cell membrane that occur following phase 1 of the action potential (AP), but before complete repolarisation. They can be classified according to the membrane potential at which they occur (Volders et al. 2000). Those that arise near the plateau of the AP are produced by ICa‘window’ current as LTCCs recover from inactivation and become re-activated (January & Riddle, 1989); such EADs depend on ‘conditioning’ during the plateau, so that slowing of repolarisation and lengthening of the plateau increases the generation of such EADs (January & Riddle, 1989). Since inward ICa and INCX predominate during the AP plateau, these currents play an important role in such conditioning (Janvier & Boyett, 1996; Linz & Meyer, 1998; Volders et al. 2000). EADs can also arise late in phase 3 via a mechanism similar to that of DADs: they are facilitated by Ca2+ overload and are associated with spontaneous Ca2+ waves, and are probably generated by inward INCX (Volders et al. 1997; Burashnikov & Antzelevitch, 1998; Zhao et al. 2012).
Transverse tubules are likely to play an important role in the generation of both forms of EAD for a number of reasons. First, EADs depend on, and are due to, ICa and INCX, which are concentrated in the t-tubule membrane (Kawai et al. 1999; Yang et al. 2002). Secondly, EADs are dependent on local subsarcolemmal [Ca2+], which is determined by trans-sarcolemmal and SR Ca2+ fluxes, both of which occur predominantly at the t-tubules. Subsarcolemmal [Ca2+] is an important determinant of the driving force for ICa and INCX and will thus control the conditioning phase of EADs arising at plateau potentials and the generation of the inward currents producing both types of EAD (Volders et al. 2000). Notably, subsarcolemmal [Ca2+] also regulates the rate of Ca2+-dependent inactivation of LTCCs, thereby determining ICa during the plateau and the availability of LTCCs to generate the window current (Benitah et al. 2010; Morotti et al. 2012). Since SR Ca2+ release causes greater Ca2+-dependent inactivation (and ICa is larger) at the t-tubules than at the surface membrane (Brette et al. 2004b), it appears likely that the t-tubules will play an important role in determining LTCC availability and thus the occurrence of EADs: reduction in Ca2+-dependent inactivation of ICa is predicted to lead to EADs (Morotti et al. 2012; Johnson et al. 2013). It has also been suggested that SR Ca2+ depletion as a result of diastolic Ca2+ release, and the consequent reduction in Ca2+-dependent inactivation of ICa, prolongs action potential duration (APD) and increases the likelihood of EADs, leading to beat-to-beat variability in repolarisation associated with the genesis of ventricular tachyarrhythmias such as torsades de pointes, particularly in the context of a reduced delayed rectifier current, IKs (Johnson et al. 2013). It is well established that a decrease of delayed rectifier (IKr or IKs) and/or background inward rectifier (IK1) potassium currents during the plateau and repolarisation phases of the action potential, as occurs in some forms of long QT syndrome, may lead to generation of EADs (Volders et al. 2000; Weiss et al. 2010). However, there are important differences in the configuration of the ventricular action potential, and in particular, of repolarising K+ currents, between small rodents (i.e. rats and mice) and larger mammalian species that show a distinct plateau phase to the action potential. While available evidence from rat ventricular myocytes shows that delayed rectifier currents and IK1 appear to be uniformly distributed between the t-tubule and surface membranes (Komukai et al. 2002), to the best of our knowledge, to date there is no information regarding the subcellular distribution of K+ channels from species with a distinct plateau phase.
Transverse tubule loss and the genesis of arrhythmias
Given the importance of t-tubules in normal excitation–contraction coupling, the number of factors that determine their function, and their potential role in the genesis of EADs and DADs, it is worth considering the role that disruption or loss of t-tubules could have on the genesis of arrhythmias, since changes in the morphology of the t-tubule network have been widely reported in human heart failure (HF) and in animal models of HF (Lyon et al. 2009; Crossman et al. 2011). The changes reported include decreases and increases in t-tubule diameter, disruption of the normal ordered t-tubule structure, and the appearance of gaps in the t-tubule network.
Anti-arrhythmic effects Given the function and role of t-tubules discussed above, loss of t-tubules might be expected to decrease the probability of spontaneous Ca2+ release causing DADs and EADs, for the following reasons.
Firstly, it will decrease the frequency of Ca2+ sparks. In intact myocytes, Ca2+ sparks occur throughout the cell; following detubulation, they occur predominantly at the surface membrane, and the overall frequency is decreased (see above; Brette et al. 2005). In addition, however, it will reduce spontaneous Ca2+ release by reducing Ca2+ loading in two ways. (i) It will slow the replenishment of SR Ca2+ content, which is a determinant of Ca2+ spark frequency. The rate at which the SR reloads with Ca2+ following caffeine-induced depletion is slower in detubulated myocytes: the number of beats (at 0.5 Hz) required for 50% refilling is ∼6 in control myocytes compared with ∼10 in detubulated myocytes (Brette et al. 2005). Although such slowing appears to have little effect on SR Ca2+ content at slow stimulation rates (Brette et al. 2005), it may decrease SR refilling at physiological heart rates, thereby keeping SR Ca2+ content further away from the threshold for spontaneous Ca2+ release. (ii) Increased spontaneous Ca2+ release is frequently caused by interventions, including inotropic agents, that increase the Ca2+ load of the cell by acting on ion flux pathways that are found predominantly in the t-tubules. For example, the adrenergic agonist isoprenaline acts in part by increasing ICa, and this increase is greater at the t-tubules than at the surface membrane (Brette et al. 2004a); other interventions, such as cardiac glycosides and increasing stimulation rate, act via NCX to increase intracellular Ca2+, and the inotropic response to such interventions is decreased following detubulation (see above). Thus loss of t-tubules would decrease the rise of intracellular Ca2+ in response to such interventions, and thus the probability of spontaneous Ca2+ release.
Secondly, it will result in the loss of INCX, which normally occurs predominantly at the t-tubules (Yang et al. 2002). Thus, following loss of t-tubules, there may be insufficient activation of INCX by spontaneous Ca2+ release to cause arrhythmias unless NCX increases in the remaining cell membrane and/or its activity is increased, for example by PKA, and interestingly, an increase in NCX activity has been reported in HF (Hobai & O’Rourke, 2000; Wei et al. 2003). Loss of t-tubules may also disrupt the normal proximity of NCX to RyRs, which would also decrease the efficacy of spontaneous Ca2+ release as a stimulus for inward INCX..
Finally, loss of t-tubules decreases action potential duration (Brette et al. 2006), presumably due to the loss of ICa and INCX without marked changes in outward current density (Komukai et al. 2002; Pásek et al. 2008), which would be expected to reduce the generation of EADs during the plateau of the action potential.
Pro-arrhythmic effects The gain of the Ca2+ release process is unaltered by acute detubulation (Brette et al. 2004b), although other changes in cell ultrastructure may decrease the gain in HF (Gomez et al. 1997). However, loss of t-tubules results in loss of action potential propagation into the cell; this causes loss of synchronous Ca2+ release and slower Ca2+ extrusion, thus decreasing the amplitude and slowing the time course of the systolic Ca2+ transient (Louch et al. 2006; Heinzel et al. 2008), which – by affecting Ca2+-dependent currents – may lead to abnormal and unstable action potential morphologies, which are suggested to be arrhythmogenic. It may also provide a substrate for altered Ca2+ propagation and regenerating waves of intracellular Ca2+ (Li et al. 2012): these authors showed, for example, that the decreased Ca2+ removal produced by a 10% decrease in SERCA activity could increase Ca2+ wave propagation. Given that SERCA is normally responsible for ∼87% of Ca2+ removal in rat ventricular myocytes (Negretti et al. 1993), a 10% decrease in SERCA activity represents an ∼8.7% decrease in Ca2+ removal. NCX is normally responsible for ∼8.7% and the sarcolemmal (SL) Ca2+ ATPase for 2.6% of Ca2+ removal. Thus, because 63% of NCX and 100% of SL Ca2+ ATPase activity appears to occur in the t-tubules (Despa et al. 2003; Chase & Orchard, 2011), loss of the t-tubules, and the associated NCX and Ca2+ ATPase activity, would result in a similar ((0.63 × 8.7) + 2.6 ≈ 8.1%) decrease of Ca2+ removal, which would therefore be expected to increase Ca2+ wave propagation, and will be greater in other species, in which NCX is responsible for a greater fraction of Ca2+ removal.
In addition, the density of ICa, INCX and neuronal-type Na+ channels is greater in the t-tubules than in the peripheral membrane, whereas the density of transient outward current (ITO), delayed rectifier potassium current (IK) and IK1 is the same at the two sites, and cardiac-type Na+ channels are located predominantly in the peripheral membrane (Komukai et al. 2002; Brette & Orchard, 2006a,b). Changes in each of these currents associated with loss of t-tubules will alter excitability, but the net effect is difficult to predict. However, detubulation has little effect on resting membrane potential but decreases APD (Brette et al. 2006), probably due to loss of ICa and INCX (Pásek et al. 2008). Shortening of APD may increase susceptibility to late phase 3 EADs (Burashnikov & Antzelevitch, 2006); the resulting decreased refractory period may also increase susceptibility to arrhythmias and, if localised within the heart, altered heterogeneity of APD could lead to re-entry. It is unknown whether there are regional differences within the heart in t-tubule structure and function, in which case even uniform t-tubule loss could lead to the development of regional heterogeneity, and differences in t-tubule density and heart size between species may result in species variation in the propensity to arrhythmias arising at the t-tubules.
Unknown arrhythmogenic effects In addition to the effects of changes in the structure of the t-tubule network per se, there are also changes in protein expression, regulation and functional distribution in HF – some known, some unknown – which may also alter arrhythmogenic activity.
It is unclear whether there are changes in the t-tubular localisation of protein function in HF. Since PKA is involved in the localisation of ICa and INCX (above), changes of PKA activity during HF may cause local changes in protein, and thus t-tubule, function. In failing myocytes the β2-adrenergic pathway, which is normally localised to the t-tubules, becomes more evenly distributed throughout the cell (Nikolaev et al. 2010). If localised basal activity of this pathway underlies stimulation of ICa and INCX at the t-tubules, then this change in distribution may result in a more uniform distribution of these currents.
It is also unknown whether ion accumulation or depletion within the t-tubules is functionally important. Given the flux of ions – particularly Ca2+– across the t-tubule membrane, and that the t-tubule lumen represents a restricted diffusion space, it seems possible that changes of ion concentration could occur within the lumen, which would modulate t-tubular membrane currents. Thus, ICa within the t-tubules may be regulated by changes of [Ca2+] in the t-tubule lumen, as well as by intracellular Ca2+ (e.g. Ca2+-dependent inactivation), so that ICa may be limited by a decrease in luminal [Ca2+] that it causes. Although not currently amenable to experimental investigation, computer modelling has shown activity-dependent Ca2+ depletion adjacent to the t-tubule membrane, and a consequent decrease in ICa and the Ca2+ transient (Pásek et al. 2012). This work also suggested an activity-dependent increase of t-tubule luminal [K+]. Such changes may alter the susceptibility to arrhythmias, with the decrease in luminal Ca2+ affording protection from Ca2+ overload, and the increase in luminal [K+] changing excitability. Such changes in luminal [ion] may be modulated by changes in t-tubule morphology in HF, which may alter the rate of diffusion of ions between the t-tubule lumen and bulk extracellular space, and by contraction of the cell, which may increase the rate of ion exchange between the two compartments.
It appears likely that in normal cardiac ventricular myocytes, the t-tubules form an important site for the initiation of arrhythmias due to spontaneous SR Ca2+ release, and the subsequent activation of INCX. Simplistically, therefore, it might be expected that loss of t-tubules in HF might afford protection against such arrhythmias. However, the net effect of altered t-tubule morphology in conditions such as HF is unknown, and will depend on protein expression, regulation, distribution, co-localisation and activity of proteins normally located at the t-tubules, all of which may alter the propensity to generate arrhythmias but many of which are currently unknown, so that the net effect is unclear. Such considerations do, however, emphasise that the ultrastructure of cardiac myocytes, as well as the structure of the whole heart, is important in the genesis of arrhythmias.
We are grateful to the British Heart Foundation for support.