The transverse-axial tubular system of cardiomyocytes

Abstract

A characteristic histological feature of striated muscle cells is the presence of deep invaginations of the plasma membrane (sarcolemma), most commonly referred to as T-tubules or the transverse-axial tubular system (TATS). TATS mediates the rapid spread of the electrical signal (action potential) to the cell core triggering Ca²⁺ release from the sarcoplasmic reticulum, ultimately inducing myofilament contraction (excitation–contraction coupling). T-tubules, first described in vertebrate skeletal muscle cells, have also been recognized for a long time in mammalian cardiac ventricular myocytes, with a structure and a function that in recent years have been shown to be far more complex and pivotal for cardiac function than initially thought. Renewed interest in T-tubule function stems from the loss and disorganization of T-tubules found in a number of pathological conditions including human heart failure (HF) and dilated and hypertrophic cardiomyopathies, as well as in animal models of HF, chronic ischemia and atrial fibrillation. Disease-related remodeling of the TATS leads to asynchronous and inhomogeneous Ca²⁺-release, due to the presence of orphan ryanodine receptors that have lost their coupling with the dihydropyridine receptors and are either not activated or activated with a delay. Here, we review the physiology of the TATS, focusing first on the relationship between function and structure, and then describing T-tubular remodeling and its reversal in disease settings and following effective therapeutic approaches.

Physiological properties of the TATS

T-tubules are transverse invaginations of the surface sarcolemma (SS) that occur in mammalian heart ventricles at each Z-line and branch within the cell to form a complex network with predominantly transverse elements but also longitudinal (axial) components running from one Z-line to the next (see Fig. 1a) [1]. This complex tubular structure is commonly referred as the T-system or, also highlighting the presence of the longitudinal components, the transverse-axial tubular system (TATS). TATS ensures rapid and homogeneous propagation of the action potential (AP) into the cell interior [2], eventually inducing myofilament contraction (excitation–contraction coupling, ECC) [3]. During the AP, Ca²⁺ enters the cell through depolarization-activated Ca²⁺ channels known as dihydropyridine receptors (DHPRs). Ca²⁺ entry triggers Ca²⁺ release (calcium-induced calcium release, CICR) from the sarcoplasmic reticulum (SR), where a large amount of Ca²⁺ is stored [4]. The ryanodine receptor (RyR) is the main SR Ca²⁺ release channel [5]. The combination of external Ca²⁺ influx and release from the SR raises the free intracellular Ca²⁺ concentration allowing Ca²⁺ to bind to the myofilament protein troponin C, which then switches on the contractile machinery. Afterwards, the SR Ca²⁺-ATPase (SERCA) and sarcolemmal Na⁺/Ca²⁺ exchanger (NCX) lower the intracellular Ca²⁺ concentration [4] allowing muscle relaxation.

Fig. 1.

TATS in cardiac myocytes. a Three-dimensional reconstruction of TATS from confocal microscopy images in a rat ventricular cardiomyocyte. Modified from [108]. b Confocal images of myocytes labeled with membrane dyes and transverse line scans showing intracellular Ca²⁺. Loss of functional T-tubules (lower panels), obtained with formamide osmotic shock in isolated myocytes, determines an inhomogeneous Ca²⁺ release throughout the cell, with delayed Ca²⁺ transient rise in the cell core. Modified from [73]. c Two-photon image of a portion of sarcolemma labeled with di-4-ANEFFPTEA. The membrane is rapidly scanned using the RAMP system at the three sites marked and voltage variations are recorded while pacing the myocyte. Of note, SS, TT and AT do not show any differences in the shape of locally recorded APs. d Average amplitude of voltage variations at SS, TT and AT. Modified from [23]. The fluorescence TATS images are shown inverted with contrast and threshold modified from the original version to make the T-system uniform and comparable among different panels

Immunohistochemical studies have recently shown that many of the key sarcolemma proteins involved in ECC, including DHPRs and NCX, are predominantly located at the tubular network [6, 7]. DHPRs are in very close proximity to RyR clusters in the terminal cisternae of the SR whilst NCX co-localization with RyRs is still debated [8, 9]. This organization is crucial in ensuring a homogeneous Ca²⁺-release throughout the cell, providing synchronous myofilaments contraction. Technically, however, the resolution of confocal microscopy is not adequate for mapping the complex geometry of TATS, preventing a quantitative interpretation of immunofluorescence data. Additionally, immunohistochemistry is unable to report protein function, which depends not only on the presence of the protein, but also on membrane insertion, local environment, accessory proteins and protein regulation. A complementary approach to probe the channel distribution exploited the restricted diffusion space within the tubular lumen. When applying a rapid change of the bathing solution, some currents showed an initial fast change, attributable to the channels located in the SS, followed by a slower phase representing the contribution of channels in the TATS [10]. Such experiments suggest that 64 % of Na⁺ and Ca²⁺ currents (I _Na and I _Ca) are within the TATS. An alternative method to determine currents distribution is provided by detubulation, a technique recently described and validated in isolated ventricular myocytes to physically and functionally uncouple the TATS from the SS. This procedure has shown that acute uncoupling of TATS from SS alters the amplitude and timing of Ca²⁺-release (Fig. 1b) highlighting the crucial role of TATS in guaranteeing a synchronous and complete activation of all Ca²⁺-release units. Comparison of the loss of cell capacitance (a function of membrane area) with the loss of membrane currents following detubulation enables to calculate the fraction of the TATS membrane currents. This investigation points out that many membrane currents, including L-type Ca²⁺ channel (LTCC) and NCX, appear to occur predominantly in TATS [11, 12]. The electrochemical gradient for AP-relevant ions may also be different across TATS and SS because the tubular network represents a restricted diffusional space, thus ion concentration in the tubular lumen may vary with respect to bulk extracellular concentration [13, 14]. In line with these considerations, the AP of detubulated ventricular myocytes was shorter than that of intact myocytes [15], suggesting that the loss of tubules is associated with a reduction of inward currents, such as I _Ca and inward I _NCX.

TATS was modeled as a single compartment separated from the SS by the mean resistance of the tubular system (R _st) [16]. The contribution of one tubule to R _st was expressed as the resistance of a cylindrical conductor, whose length, radius, and specific resistivity corresponded to one-half of the tubular effective length (l/2), its average radius (r), and the specific resistivity of the extracellular solution (ρ), respectively. TATS represents a parallel combination of all (n) tubules in the model cell, the mean resistance of the tubular system can be calculated from the relationship:

$$R_{\text{st}}=\mathit{\rho }l/\phantom{\mathit{\rho }l2\mathit{\pi }{r}^{2}n}2\mathit{\pi }{r}^{2}n$$ 1

Starting from this simple application of the second Ohm law, a biophysically realistic computer model of a cardiac cell was created [17], incorporating the TATS compartment and the current distributions determined using detubulation [18]. The model highlighted a negligible difference of AP amplitude and kinetics between SS and TATS. This model, however, exhibits some limitations due to the simplistic assumption that a tubule is a single isolated compartment and not a portion of a complex network, profoundly affected by events occurring elsewhere in the system itself. Such mathematical model does not address important issues. For instance, it is not possible to mimic how the AP is propagated when a tubule has a stricture or when it branches to form an axial component.

The experimental investigation of the electrical properties of different portions of the sarcolemma requires a direct measure of tubular AP. Although current optical techniques for probing membrane potential allow recording of voltage changes at subcellular level [19], most approaches lack the spatial–temporal resolution and signal-to-noise ratio needed for regional assessment of AP profile in multiple positions. Recently, a new imaging method (RAMP) [20, 21] has been developed to simultaneously record APs at multiple sarcolemmal sites with submillisecond temporal and submicrometer spatial resolution [22, 23]. The RAMP microscope was used to rapidly scan linear segments of different membrane domains and perform multiplexed measurements of the two-photon fluorescence (TPF) signal. Fig. 1c shows an example of real-time and simultaneous optical recording of ten elicited APs from three different membrane sites: surface sarcolemma (SS, red), T-tubule (TT, green), and axial tubule (AT, blue). Figure 1d shows that AP amplitude measured in the TATS is not statistically different from that on SS. Measurements at a high stimulation frequency were also performed to test possible effects of local AP alterations due to cumulative changes in TATS luminal ion concentration (e.g., Ca²⁺ depletion and K⁺ accumulation). The uniformity of AP in all sarcolemma domains is maintained even at high pacing rates, proving the tight electrical coupling between membrane domains assumed in theoretical modeling.

The structure of the TATS

The first observations of cardiac T-tubules date back to the 1950s [1]. Studies on mammalian ventricular myocytes confirmed that structure and function of cardiac T-tubules is similar to those of skeletal muscle. Some structural differences can, however, be highlighted. The cardiac T-tubules have a wide lumen, they contain a visible extension of the basal lamina, and have wide rather than restricted openings. The skeletal muscle T-tubules, at least in vertebrates, have a narrow lumen, particularly in the segments between triads, they have an apparently empty lumen, and they are connected to the SS mostly thorough narrow restricted passages [24].

An increasing body of observations has revealed a very high complexity of the cardiac tubules in terms of structural diversity and dynamics (both in physiological and pathological conditions). Firstly, cardiac tubules have a complicated topology. Electron microscopy (EM), in combination with the use of horseradish peroxidase to label the extracellular space [25], revealed that, in rat ventricular myocytes, tubules run not only transversally to the myofilaments’ direction but also longitudinally (Figs. 1a, 2a), with numerous branchings leading to the formation of a complex network of transverse and axial elements, the TATS. The topographical extension of cardiac tubules beyond the Z-line profiles potentially creates a large variability in the location and morphology of tubule-SR junctions. Figure 2b shows the ultrastructure of a typical T-SR junction. The relationship between SR and the tubular network will be discussed further below.

Fig. 2.

Cardiac tubule structure, junctions and network. a Electron micrograph of rat right ventricle, showing the extracellular space (ES), tubules running transversally (asterisk) and longitudinally (T3b). The arrows point at subsarcolemmal cisternae either labeled (double arrows) or unlabeled (single arrows) by the extracellular marker. Calibration bar 1.0 μm. Modified from [25]. b Electron micrograph of mouse left ventricle, showing an internal junction between SR and a T-tubule. Two SR elements (probably belonging to the same cisterna but separated in the image due to the sectioning plane) are apposed to a T-tubule. Calibration bar 0.25 μm. Modified from [109]. c–e Confocal images of rat myocytes, stained with di-8-ANEPPS and showing the dense tubular network in ventricular cells (c) and the low density or almost absence of tubules in atrial cells (d). In (d), a large longitudinal component can be noticed. Calibration bar 10 μm. Modified from [30]. f–g Confocal images of sheep myocytes, stained with FITC-conjugated wheat germ agglutinin and showing the presence of a developed tubular network (albeit with lower density than in rat) in the ventricle (f) and a less developed network in the atrium (g). Modified from [27]. The fluorescence images are shown inverted with contrast and threshold modified from the original version to make the tubular system uniform and comparable among different panels

Beyond the first EM observations, the investigation of the topology of the whole 3D tubular network in living myocytes has become accessible with the use of confocal or TPF microscopy, in combination with fluorescent membrane dyes (for example, 6-ANNEPS; see Fig. 1a). This method has allowed the systematic investigation of the TATS in different compartments of the heart and in different species, revealing, once again, a high degree of variability [26–32]. Figure 2c–g shows the variable density of tubules in ventricular myocytes across different species and the scarcity (or absence altogether) of tubules in atrial cells. The figure shows examples limited to two mammalian species, since all other phyla (amphibians, reptiles and birds) seem to lack tubules in both atrium and ventricle. Within mammals, the topology and ultrastructure of the tubular system have been extensively sampled and characterized, leading to a clear picture of TATS structure in small mammals (especially rodents). Larger mammals, on the other hand, present more anecdotic literature pointing anyway to a large spectrum of T-system structures and density in both ventricle and atrium, ranging from rare or absent tubules (for example, sheep atrium; see Fig. 2g, and pig ventricle [33]) to fairly dense and complex systems (most notably ventricles of dog [34] and sheep; see Fig. 2f), but never reaching the densities and structural complexity typical of rodent TATS.

Ventricular T-system density appears, by and large, correlated with the heart rate of each species. Functionally, it is reasonable to assume that those species with a higher heart rate require faster propagation of the cardiac AP from the sarcolemma into the cell interior, thus experiencing an evolutionary pressure for a higher T-tubules density. Indeed, the dynamics of Ca²⁺ propagation inside the myocyte depending on SS-tubular continuity (Fig. 1b) provide a clear demonstration of the effects of tubular density on the synchronicity and efficacy of myofilament contraction across the myocyte.

Image analysis on ventricular rat myocytes allowed the quantification of tubular size, as well as the relative cell volume occupied by tubules and of the relative abundance of tubules at the Z-lines and elsewhere [35]. The former was estimated at 3.6 %, somewhat higher than previous estimates derived from EM images; the latter ratio was estimated at around 60 % at the Z-line and 40 % elsewhere along the sarcomere. Cardiac tubule diameter could also be estimated at 100–300 nm, significantly larger than skeletal tubules (20–40 nm). The three-dimensional reconstruction obtained with two-photon microscopy also allowed mapping of the topography of tubular invagination from the SS, indicating large areas of SS characterized by a regular rectangular pattern of T-tubules mouths at about 1.8 μm distance from each other. This distribution of T-tubules mouths was first observed using the freeze-etch technique [36] and recently confirmed by scanning ion conductance microscopy [33].

Axial components have been described in both skeletal and cardiac muscle [24, 37]. Although in skeletal muscle axial tubules represent only 3 % of the T-system, in cardiac ventricles they have been estimated around 10–20 %, bridging between up to 4–5 consecutive Z-lines.

The principal function of tubules in cardiac ECC takes place at the junctions between the tubules themselves and the SR. Such junctions (dyads or couplons) are characterized by the apposition of the two membranes (with a typical distance of 12 nm), which brings in close proximity the sarcolemmal LTCCs with the SR RyRs (Fig. 2b). Immunostaining for RyRs [38] has shown that they are located primarily on the junctional SR, at the level of T-tubules and on the cell surface near the Z-lines, but some RyRs are on junctional SR adjacent to longitudinal tubules. Thus, the landscape of tubule-SR junctions mirrors the complexity of the topology of the TATS in cardiac myocytes, presenting contacts in three regions: in couplons on the surface, on T-tubules (both of which are near the Z-line), and in junctions on most of the longitudinal tubules–axial junctions. The axial junctions average 510 nm in length, sometimes spanning an entire sarcomere; in some studies, they were found to contain as much as 19 % of a cell’s RyRs [39]. In the same study, tomographic analysis confirmed the architecture of axial junctions as being indistinguishable from those on T-tubules or on the surface. Also, a complex tubule structure was observed, with a lumen of only 26 nm at its narrowest point. RyRs on axial junctions colocalize with CaV1.2, suggesting that it plays a role in ECC. In conclusion, longitudinal elements, although difficult to image properly, have been recently proved as an essential component of TATS, whose exact function still needs to be elucidated.

The complex structure–function relationship of cardiac T-tubules is highlighted by the dependence of the presence of tubules themselves on the cardiac performance: absent in the embryo, tubules develop at birth with the rising of left ventricular pressure and working volumes. The TATS formation occurs in parallel with the formation of other membrane structures, particularly the caveolae and the SR. The relationship between caveolae and T-tubules is similar to that seen in skeletal muscle [40, 41]. The relationship is strongly confirmed by the presence of caveolin-3, a specific marker of caveolae in T-tubules of developing skeletal and cardiac muscle [42] as well as in adult heart [43]. Typically, the initial development of T-tubules occurs at the cell edge in parallel to caveolae proliferation and to the caveolar multiple complexes formation to which T-tubules are clearly connected [44]. This suggests that caveolae formation is a necessary step in the invagination process of T-tubules, and that caveolin plays an essential role in this process by allowing curving of the membrane [42].

The formation of the TATS appears to derive from the generation of repetitive caveolae, which form “beaded tubules”. The penetration of T tubules into the cell has been envisaged as originating at the cell periphery and extending towards the center [40, 45–49], with a progressive increase in the number of connections between SR and growing T-tubules. Accordingly, DHPR foci, initially present only at the cell periphery, then also appear in the cell interior, where they co-localize with RyR clusters [45]. This would indicate that, during the T-tubule formation, the two proteins migrate independently through the SR and T-tubules and become associated when the two sets of membranes get close to each other and form the calcium release unit. However, observations in skeletal and in cardiac muscle suggest the possibility that distinct vesicles (expressing RyR or DHPR at their surface) associate in the cell interior due to RyR–DHPR interaction. Subsequently, each vesicle would be targeted to developing T-tubules or SR and fuse, contributing to their development. A combination of the two mechanisms is probably the most feasible [50].

The network organization reached at the end of tubular development is not static but rather responds to the cardiac functionality, with loss of tubules in pathological conditions and myocyte culture being the most evident examples. A number of proteins have been identified in the processes responsible for the formation and maintenance of the tubular network and of the tubule–SR molecular contacts.

Recent investigations found loss and disorganization of T-system in a number of pathological conditions including human chronic heart failure (HF) [33, 51], dilated and hypertrophic cardiomyopathies [33], and in animal models of chronic HF [52], chronic ischemia [26], atrial fibrillation [53], and mechanical unloading [54]. Additionally, the tubular network is gradually lost when myocytes are kept in culture, indicating a highly dynamic equilibrium involved in the maintenance of the intact network. To date, one of the most effective methods employed for the study of the role of the tubular network, as well as the consequences of its disruption, is the detubulation by osmotic shock [15, 55]. As described before, in this method, the osmotic volume changes induced by sequential addition and removal of formamide cause the physical detachment of the T-tubules from the SS. The effectiveness of the method is clearly demonstrated by staining myocytes before and after detubulation (Fig. 3); in the latter case, the loss of continuity between SS and tubular membrane, due to detubulation, prevents diffusion of the fluorescent dye in most of the tubular elements. Staining the tubules before detubulation, on the other hand, ensures uniform labeling of the tubular network, thus allowing AP measurements into the network after detubulation [23]. Figure 3a–c demonstrate that the loss of continuity between SS and the TTs abolishes the propagation of APs from the SS into the TATS. To investigate the behavior of the tubules that remain connected to the SS after detubulation, we stained cells after the osmotic shock so that only tubules still continuous with the surface were labeled. Figure 3e–f highlights a large variability in the responsiveness of different tubular elements to the propagation of AP. Even though the tubules are connected to the surface, AP propagation can fail. This indicates that connection of a single tubular element with the SS is not sufficient to ensure its electrical coupling to the surface.

Fig. 3.

TATS integrity and AP propagation. a Two-photon fluorescence (TPF) image of a myocyte with TATS stained before detubulation (S/D) by formamide-induced osmotic shock. Scale bar 5 μm. b Normalized fluorescence traces from the scanned lines indicated in (a): no electrical activity is detected in the scanned TATS regions (green and blue traces). Arrowheads point at the time of electrical stimulation. c Average of the ten sequential episodes shown in (b). d TPF image from a myocyte stained after detubulation (D/S): only a subpopulation of TTs is well labeled. Scale bar 5 μm. g Normalized fluorescence traces from the scanned lines indicated in (d): SS and TT 1 show regular APs, whereas TT 2 and TT 3 display non-regenerative electrical responses. Red asterisks in TT 3 highlight AP failures. f Average of ten episodes for SS, TT 1, and TT 2. Separate averaging of six APs and four subthreshold events in TT 3. Modified from [23]. The fluorescence images are shown inverted with contrast and threshold modified from the original version to make the T-system uniform and comparable among different panels

Structural and functional alterations of the TATS in cardiac diseases

Pathological alterations of the T-system were first identified in a limited number of EM studies performed on human ventricular tissue from patients with cardiac hypertrophy or HF [56–59]. Employing confocal microscopy and fluorescent membrane labeling, patchy T-tubular loss and disorganization have been described in a number of animal models of cardiac diseases, as illustrated in Fig. 4. Most of these results have been obtained from ventricular myocytes of HF models developed on small animals with high heart rates (such as mouse or rat) showing a rich and well-organized TATS before the pathological insults. In larger species with lower heart rates and a slower speed of contraction–relaxation, the requirement for such a highly developed T-tubular structure is not so clear. Although both pig and dog models of HF showed significant reduction in T-tubule density [52, 60], the most striking feature in these large animals was the wide number of areas devoid of T-tubules even in the normal hearts [61].

Fig. 4.

TATS remodeling in cardiac diseases. The table shows a collection of representative images obtained with membrane labeling and confocal microscopy from different animal and human models of cardiac disease. For each reference, representative images of diseased myocytes are in the red squares, while the respective controls are in the blue squares. Images modified from [34, 52, 53, 63, 64, 67, 68, 77, 110, 111]. The images are shown inverted with contrast and threshold modified from the original version to make the T-system uniform and comparable among different panels

In humans, early reports based on histological examinations in failing heart tissue sections showed T-tubular dilation with either increased [62] or decreased [58, 59] density of T-tubules, leaving the open question of whether low T-tubule density was failure-related or not. Recent studies in failing human hearts agree on the reduction of T-tubule density, which was two to three times lower in failing ventricular myocytes from HF patients with different aetiologies than in healthy donors [33, 63]. In a rat model of left ventricular pressure overload, the loss of T-tubules marks the transition from compensated hypertrophy to HF and thus detubulation appears likely as an “early event” in disease-related myocardial remodeling [64] (Fig. 5a). Consistently, in septal myocytes from myectomy samples of non-failing hypertrophic cardiomyopathy (HCM) patients with obstructive hypertrophy, we have recently observed that T-tubules are completely lost [65], indicating that T-tubule remodeling likely precedes the end-stage of the disease.

Fig. 5.

Abnormal TATS function in HF. a Representative T-tubule images (FM4–64 membrane staining) from the left ventricle of sham-operated, hypertrophic, early HF, and advanced HF rat hearts, showing the progression of TATS remodeling at different stages of the disease. Rats were subjected to pressure overload by thoracic aortic banding, leading first to left ventricular hypertrophy and then failure. Modified from [64]. b Live-cell STED images (di-8-ANEPPS membrane staining) showing TATS structures from sham, 4 weeks post-MI (4pMI), and 8 weeks post-MI (8pMI) cells (bottom triangles indicate position of striations); TATS appears enlarged, misaligned, and with increased longitudinal components in 4 weeks post-MI and, more pronouncedly, 8 weeks post-MI. Scale bar 1 μm. Modified from [69]. c Cell from a pig model of chronic ischemia, 6 weeks after inducing severe stenosis of the circumflex coronary artery. T-tubular (WGA-Alexa594) and corresponding line-scan Ca²⁺ (Fluo 3) images, showing that the regions of delayed Ca²⁺ release are related to areas of T-tubule rarefaction. Right horizontal scale bar 10 μm; vertical scale bar = 100 μm; left horizontal scale bar 50 ms. Modified from [26]. d Representative TPF image of a HF myocyte. e Normalized fluorescence traces from the scanned line indicated in (d): SS and TT 1 show that regular APs, TT 2, and TT 3 display non-regenerative electrical responses, and TT 4 highlights local arrhythmic events (blue asterisks). d, e Modified from [23]. The fluorescence TATS images in (a, c, and d) are shown inverted with contrast and threshold modified from the original version to make the T-system uniform and comparable among different panels

In addition to standard quantification of T-tubular loss with di-8-ANEPPS surface staining, detailed changes of cardiomyocyte surface topology were detected using a unique method (scanning ion conductance microscopy, SICM) to form topographic images of live myocytes [33, 66]. With this method, the loss of T-tubular openings in ventricular myocytes from failing human hearts was confirmed. In failing human cardiomyocytes, in addition to T-tubule loss, disruption of Z-groove structure has been observed [33].

Besides a reduction in the number of T-tubular openings and transverse components, disease-related T-system structural abnormalities include: (1) a greater proportion of tubules running in the longitudinal and oblique directions [67, 68], (2) increased mean T-tubular diameter and length [56, 59], and (3) tubular proliferation, with increased T-tubular tortuosity (i.e. number of constrictions and branches) [58]. These geometrical changes were recently characterized with nanometric resolution by using STED microscopy [69, 70] (Fig. 5b).

In HF, T-tubular remodeling goes hand-in-hand with spatial alterations of protein expression. Membrane fractionation studies were used to assess the redistribution of membrane-bound proteins among the different membrane sub-compartments in HF cells, showing reduction in LTCCs and β-adrenergic receptors, but increased levels of NCX protein, in both SS and residual T-tubular fractions [34].

In addition, SICM was used to assess the spatial localization of functional β2-ARs, which are concentrated in the TATS in normal cells. However, in heart failure, the spatial localization is lost, with a relative redistribution of β2-ARs to the cell surface [71]. The tight spatial localization of the β2-AR response depends on its co-localization with protein kinase A (PKA) and phosphodiesterases, which limit the spread of the cAMP response. The normally striated pattern of PKA, reflecting T-tubular periodicity, is lost in HF. This ‘globalization’ of the β2-AR response may partly mediate the loss of its cardioprotective effects and its role in driving the disease process in HF.

Asynchronous CICR and loss of Ca²⁺ release regulation

Since T-tubules are essential for fast coupling between the electrical activation of the membrane and the initiation of contraction, the consequences of a partial loss of T-tubules is expected to influence the amplitude and kinetics of Ca²⁺ fluxes and contraction parameters. The effects of T-tubule alterations on ECC were initially investigated by experimentally promoting T-tubule loss by either cell culture [60] or formamide-induced detubulation [72]. In both cases, T-tubule loss was associated with desynchronization of Ca²⁺ release across the cell. When cells were detubulated with formamide, wave-like propagation of the Ca²⁺ transient from the sarcolemma to the cell interior was observed [73], which resembled the pattern of Ca²⁺ release reported in cells with very low T-tubule density, such as atrial myocytes from rodents or Purkinje cells [74, 75]. With a less dramatic TATS reduction observed during cell culture, a fragmented Ca²⁺ release pattern was found [60]. This indicated that SR Ca²⁺ release was initially triggered at sites where T-tubules were present, followed by propagation into regions devoid of T-tubules. This situation is similar to that observed in normal myocytes characterized by a moderate T-tubular density, such as sheep atrial myocytes [28]. Spatially summating Ca²⁺ release in cells that had partially lost the T-system organization always resulted in an overall Ca²⁺ transient with slower kinetics and reduced magnitude.

Asynchronous Ca²⁺ release was first observed in myocytes from failing hearts by Litwin et al. [76]. Studies by Louch et al. [60] and Song et al. [77] highlighted a spatial link between such alterations of Ca²⁺ release and abnormalities of T-tubular structure. In both studies, by simultaneously visualizing the T-tubule network and intracellular [Ca²⁺] (Fig. 5c), small regions with delayed Ca²⁺ release were observed following T-tubule disorganization, occurring at irregular gaps among adjacent T-tubules. Song et al. [77, 78] observed, however, that the RyR distribution remained intact in failing myocytes, suggesting that T-tubule disorganization resulted in some orphaned RyRs, i.e. RyRs cluster that became physically separated from their DHPR partners. These orphaned RyRs might respond differently to local Ca²⁺ changes, with loss of normal local control. Importantly, a number of investigations [33, 51] have shown that reduced Ca²⁺ release synchrony in failing cells (as in case of experimental T-tubule loss [73]) promotes slowing and broadening of the overall Ca²⁺ transient, a hallmark of the failing condition [79]. In HF, prolongation of the AP and loss of an early repolarization notch in failing cells reduce the driving force for Ca²⁺ entry, resulting in decreased peak Ca²⁺ current [80]. In addition, the time course of Ca²⁺ entry is prolonged during the failing AP, which reduces efficiency for triggering Ca²⁺ release from the SR [81].

Whether and how alterations in RyR function in HF are related to T-tubular changes is still unclear. In cardiomyocytes that underwent acute formamide-induced detubulation, the frequency of spontaneous Ca²⁺ sparks was not increased, indicating that functional uncoupling of T-tubules from the surface does not affect RyR open probability [73]. Conversely, Meethal et al. [82] recently reported that, in failing cells, spontaneous Ca²⁺ sparks occur more frequently at the irregular gaps between T-tubules created by T-tubule loss. Resting Ca²⁺ levels were also higher at these locations. These last observations suggest that orphaned RyR could exhibit greater activity than those in intact dyads, and that this over-activity may contribute importantly to increased SR leak and arrhythmic potential in failing cells. If groups of orphaned RyRs are also functionally uncoupled and therefore unregulated, such alterations might theoretically exacerbate Ca²⁺ release asynchrony by slowing propagation of CICR into regions where T-tubules are absent. Similarly, SR leak and reduced SR content might also influence synchrony by affecting the speed of Ca²⁺ propagation into detubulated cell regions [73]. Moreover, in failing myocytes, the observed expansion of the dyadic cleft, resulting in a greater distance between Ca²⁺ channels and RyR [83], would contribute to the lower Ca²⁺ release amplitude and synchrony, further reducing ECC efficiency.

AP propagation in remodeled TATS: implications for arrhythmias

Ventricular cardiomyocytes in HF are characterized by loss and disorganization of the TATS [33], determining orphaned RyR clusters, which remain non-recruited or are activated with a significant delay [83]. We have previously observed that, within the diseased myocyte, many TATS elements fail to conduct AP despite being physically connected to the SS [23] (Fig. 5d, e). In dysfunctional TATS elements, no active responses are observed upon cell stimulation. TATS elements that do not conduct AP cannot activate Ca²⁺ current and Ca²⁺ release. Therefore, the presence of dysfunctional tubules represent an additional mechanism contributing to the reduced RyR recruitment and lower ECC efficiency in HF. Morphological analysis of TATS density may then overestimate the number of recruitable Ca²⁺ release units in HF.

How electrical conduction may fail in physically coupled TATS elements remains to be investigated. Mathematical modeling, as described above, may help clarify this issue. Local electrical uncoupling even in the presence of the membrane continuity may arise from the increased electrical resistance between the TATS element and the surrounding sarcolemma. The morphological correlate of this increased resistance in HF-related tubular remodeling, as in myocytes from rats with post-myocardial infarction systolic dysfunction, may rely on the changes of T-tubular architecture discussed above, in particular the increased average length and tortuosity of T-tubules. Interestingly, the model predicts that, when the number of T-tubules is reduced, slight variations of T-tubular length can cause failure of AP conduction.

Moreover, we have shown that, in cells from infarcted rats, T-tubules that do not display normal APs in response to stimuli sometimes show spontaneous electrical activity. We can speculate that these spontaneous depolarizations may be related to spontaneous Ca²⁺ releases from the SR. Local Ca²⁺ releases may give rise to activation of inward NCX current within the T-tubules. If the resistance R _st (see Eq. 1) is very high, T-tubules are electrically disconnected from SS. The capacitance of electrically-uncoupled TATS elements is smaller than SS capacitance; therefore, even small Ca²⁺ releases from the SR can potentially generate large local NCX-driven inward currents and give rise to important variations of membrane voltage. Similar currents would be too small to affect global membrane potential. The high R _st would keep these responses “local” and prevent their spread; however, their arrhythmogenic potential may be relevant. In fact, local spontaneous depolarizations may be capable of amplifying Ca²⁺ sparks via activation of the nearby RyR clusters, thereby turning local spontaneous Ca²⁺ release into generalized Ca²⁺ waves [84] spreading throughout the whole myocyte. In our view, two complementary mechanisms may cooperate in determining HF-related arrhythmogenicity: unregulated RyRs, increasing the rate of Ca²⁺ sparks, and uncoupled TATS elements, generating a positive feedback that turns local sparks into generalized Ca²⁺ waves, which are triggers for delayed after depolarizations (DADs) and arrhythmias. Besides the reduced TATS density, the presence of tubules with untimely electrical responses, by triggering local asynchronous Ca²⁺ release, may also contribute to non-uniform myofilament activation [84] and SR Ca²⁺ content depletion [83], promoting contractile dysfunction in HF. Besides spatial inhomogeneities, asynchronous activation of unregulated RyR may desynchronize their refractoriness leading to Ca²⁺ alternans, beat-to-beat variability, and, thus, arrhythmogenicity. This hypothesis, although rather speculative, deserves further investigation.

Reverse T-tubular remodeling and molecular determinants of TATS assembly/disorganization

Is the pathological loss of T-tubules a one-way course of events or can the T-system structure be recovered? What happens when the primary source of damage to the diseased myocardium is resolved? And what happens to the T-system when sinus rhythm is recovered after successful cardioversion of atrial fibrillation or when HF is efficaciously treated?

Reversal of pathological myocardial remodeling (reverse remodeling) has been extensively studied in a number of human models of HF. Mechanical unloading of the failing heart, obtained either by heterotopic heart transplantation or by virtue of left-ventricular assist devices (LVAD) [85], seems to reverse some of the HF-related changes in diseased cardiomyocytes. In patients with terminal HF, LVAD treatment leads to reduction in cell hypertrophy [86], shorter AP duration, increased LTCC current with faster inactivation [87], higher SR Ca²⁺ content related to increased SERCA protein [88], and augmented amplitude of Ca²⁺ release, ultimately determining improved contractility of cardiac muscle [89]. Similarly, in a rat model of ischemic HF, heterotopic transplantation led to beneficial effects on cell size, Ca²⁺ transient amplitude and kinetics, and developed force [90]. Interestingly, in that model, unloading counteracted the HF-related loss and disorganization of T-tubules: the density of T-tubules, the number of Z-grooves and T-tubule openings on the surface, and T-tubular dimensions were restored by unloading [91] (Fig. 6b). As a consequence of T-tubular recovery, the synchrony of CICR was also normalized. Although the effects of mechanical unloading on T-tubules remain to be studied on human failing cardiomyocytes, the reported beneficial effects on Ca²⁺ release kinetics [87] suggest that a partial recovery of T-system structure also occurs in human tissue. Furthermore, in a canine model of HF T-system restoration has been observed after 3 weeks of cardiac resynchronization therapy [92] (simultaneous pacing of the right and the left ventricles), largely used in HF patients [93].

Fig. 6.

Molecular complexes implicated in TATS turn-over and reverse remodeling. a Junctophilin-2 (JPH2) knockdown causes disorganization and loss of TATS: confocal images of di-8-ANEPPS stained cardiomyocytes. Modified from [101]. b HF myocytes show lower TATS density and T-tubular dilation, which are partially reversed by mechanical unloading (HF-unloading). Modified from [91]. c Retubulation of failing cardiomyocytes after rescue by AAV9SERCA2a gene therapy. Confocal images of di-8-ANNEPPS stained healthy, failing (HF) and failing AAV9SERCA2a-treated (HF + S) hearts. Modified from [97]. d Exercise training in HF (MI-TR) increases TATS density, compared to sedentary heart failure (MI-SED): confocal images of di-8-ANEPPS stained cardiomyocytes. Modified from [100]. The fluorescence images are shown inverted with contrast and threshold modified from the original version to make the T-system uniform and comparable among different panels

Overexpression of SERCA protein induced by targeted viral gene transfer of SERCA2a gene has been widely studied as a valuable therapy for HF [94]. Owing to the faster Ca²⁺ transient decay, the reduced diastolic Ca²⁺, and the higher SR Ca²⁺ content, SERCA overexpression leads to improved myocardial contraction [89]; in rat models of chronic pressure overload [95] and ischemic HF [96], SERCA2a gene transfer determined ameliorated cardiac function and reduced frequency of arrhythmias. Interestingly, in the rat HF model, functional improvement due to SERCA2a transfection was accompanied by a reversal of T-tubular remodeling [97]; the density of TATS was restored and the spatial heterogeneity of Ca²⁺ transient was reduced, resulting in larger amplitude and faster kinetics of Ca²⁺ release (Fig. 6c).

Pharmacological approaches have also been employed in disease model to reduce myocardial remodeling and TATS disorganization. In a rat model of pulmonary hypertension, treatment with phosphodiesterase-5 inhibitor sildenafil reduced the hypertrophic remodeling of right ventricular cardiomyocytes and reversed the impairment of T-tubule integrity [98]. In a HF rat model, reduction of heart rate with ivabradine reduced T-tubular changes in combination with unloading [99]. In a rat model of HF, exercise training promotes TATS regeneration [100] (Fig. 6d).

In conclusion, T-tubular remodeling in cardiac diseases is a highly plastic phenomenon, which reverts upon effective treatments of the underlying cause of disease.

Despite the fact that loss of T-tubules occurs in most cardiac diseases, the molecular pathways determining TATS disorganization are still largely unknown. A recent advancement has been made by two seminal works highlighting the role of junctophilin [101] and telethonin [102] in the correct assembly of TATS-SR-Z disc connections. Junctophilin (JPH2) expression is reduced in human and animal models of hypertrophy and failure [103, 104] and inherited mutations of JPH2 and telethonin gene (Tcap) have been identified as responsible for hypertrophic and dilated cardiomyopathies [105, 106]. Of note, all these conditions involve loss of T-tubules. Interestingly, specific cardiac knockout of JPH2 or Tcap genes in transgenic mouse models leads to severe reduction of TATS density, orphaned and unregulated RyRs, and abnormal ECC leading to global cardiac dysfunction [101, 102] (Fig 6a).

Junctophilin-2 undergoes Ca²⁺-dependent proteolysis, and junctophilin-2 levels are reduced following cardiac ischaemia–reperfusion [107]. Junctophilin proteolysis, destabilizing the close apposition of the T-system and SR membranes, may disrupt communication between the dihydropyridine and RyR and thus contribute to cardiac dysfunction.

These observations suggest that JPH2 and Tcap play an important role in determining a correct T-tubular structure, and changes in the expression of these proteins might be a determinant of T-system remodeling in disease settings.