Hyperactive ryanodine receptors in human heart failure and ischaemic cardiomyopathy reside outside of couplons.

Aims: In ventricular myocytes from humans and large mammals, the transverse and axial tubular system (TATS) network is less extensive than in rodents with consequently a greater proportion of ryanodine receptors (RyRs) not coupled to this membrane system. TATS remodelling in heart failure (HF) and after myocardial infarction (MI) increases the fraction of non-coupled RyRs. Here we investigate whether this remodelling alters the activity of coupled and non-coupled RyR sub-populations through changes in local signalling. We study myocytes from patients with end-stage HF, compared with non-failing (non-HF), and myocytes from pigs with MI and reduced left ventricular (LV) function, compared with sham intervention (SHAM). Methods and results: Single LV myocytes for functional studies were isolated according to standard protocols. Immunofluorescent staining visualized organization of TATS and RyRs. Ca2+ was measured by confocal imaging (fluo-4 as indicator) and using whole-cell patch-clamp (37°C). Spontaneous Ca2+ release events, Ca2+ sparks, as a readout for RyR activity were recorded during a 15 s period following conditioning stimulation at 2 Hz. Sparks were assigned to cell regions categorized as coupled or non-coupled sites according to a previously developed method. Human HF myocytes had more non-coupled sites and these had more spontaneous activity than in non-HF. Hyperactivity of these non-coupled RyRs was reduced by Ca2+/calmodulin-dependent kinase II (CaMKII) inhibition. Myocytes from MI pigs had similar changes compared with SHAM controls as seen in human HF myocytes. As well as by CaMKII inhibition, in MI, the increased activity of non-coupled sites was inhibited by mitochondrial reactive oxygen species (mito-ROS) scavenging. Under adrenergic stimulation, Ca2+ waves were more frequent and originated at non-coupled sites, generating larger Na+/Ca2+ exchange currents in MI than in SHAM. Inhibition of CaMKII or mito-ROS scavenging reduced spontaneous Ca2+ waves, and improved excitation-contraction coupling. Conclusions: In HF and after MI, RyR microdomain re-organization enhances spontaneous Ca2+ release at non-coupled sites in a manner dependent on CaMKII activation and mito-ROS production. This specific modulation generates a substrate for arrhythmia that appears to be responsive to selective pharmacologic modulation.

1. Introduction

Abnormal cellular Ca2+ handling and excitation-contraction coupling contribute to reduced contractile function and arrhythmogenesis in heart failure (HF). Ca2+ uptake into the sarcoplasmic reticulum (SR) by SERCA, Ca2+ release through ryanodine receptors (RyRs), Ca2+ extrusion across the sarcolemma by the Na+/Ca2+ exchanger (NCX), and Ca2+ influx via L-type Ca2+ channels (LTCCs) all contribute to this process. Novel Ca2+ pathways also emerge such as flux through TRP channels. Functional changes in Ca2+ handling occur through alteration in protein expression or through post-translational modifications. Signalling pathways involved in remodelling of Ca2+ handling include those mediated by reactive oxygen species (ROS), Ca2+/calmodulin-dependent kinase II (CaMKII), and nitric oxide (NO). Loss of SR Ca2+ through ‘leaky’ RyRs is attributed to post-translational modifications of RyRs that lead to alterations in RyR Ca2+ sensitivity and open probability (Po). In animal models of disease, increased SR Ca2+ leak is observed after myocardial infarction (MI) compared with control as well as in HF, diabetes, and atrial fibrillation (AF). In human myocytes, greater Ca2+ leak was reported in HF and AF, although one study found that spontaneous release events, Ca2+ sparks, were less frequent in HF. Ca2+ loss through RyRs reduces contraction by decreasing the SR Ca2+ content, and can be a trigger to induce arrhythmogenic Ca2+ waves through activation of NCX.

Changes in the subcellular architecture of cardiac myocytes with disease further modulate Ca2+ handling. Reduced density and disorganization of the transverse and axial tubular system (TATS) is seen in post-MI remodelling. These and other data suggest that TATS remodelling is a general feature in the progression to HF. In contrast, RyR clusters appear to remain distributed at the Z lines throughout the cell, resulting in a large fraction of RyRs that are no longer in close proximity to the sarcolemmal (SL) membrane and TATS, i.e. outside couplons. Due to the increased distance to LTCCs, these non-coupled RyRs cannot be activated directly by Ca2+ entry via LTCC but are activated with a delay through Ca2+-induced-Ca2+-release propagating from coupled sites. This dys-synchrony reduces the rise time and amplitude of the global Ca2+ transient, and contributes to impaired contractile function.

The architecture of TATS results in different signalling microdomains around coupled and non-coupled RyRs. In healthy pig myocytes, coupled RyRs have a unique NOX2- and CaMKII-dependent regulation during increased heart rate and β-adrenergic stimulation., However, the link between TATS remodelling in disease and the modulation of the larger population of non-coupled RyRs is not well studied.

To date, changes in RyR function in human HF have been exclusively studied at the myocyte level,, without information on subpopulations of RyRs, and mostly using HF myocytes only. Comparisons between non-HF and HF are scarce. Recent work studied RyRs extracted from left ventricular (LV) tissues samples from HF and non-HF and reconstituted in bilayers. In the pig with MI, we reported that TATS remodelling affected properties of the RyR subpopulations but did not investigate the signalling mechanisms involved. In a dog model of HF, Belevych et al. found that RyR subpopulations had different characteristics. However, little is known of the intrinsic properties, modulation, and arrhythmogenic contributions of coupled vs. non-coupled RyRs in post-MI remodelling or in human HF. Yet, selective targeting of modulators controlling RyR Po has become of interest in the treatment of cardiac disease where impaired Ca2+ handling plays a role.

Pilot data from our earlier work indicated that after MI, increased spontaneous RyR activity was exclusively a feature of non-coupled sites. However, no mechanistic explanation could be offered at the time. This study was designed: (i) to explore and validate the relevance of differences in coupled and non-coupled RyR properties in human HF; (ii) to study underlying mechanisms of the modulation of RyR subpopulations and the arrhythmogenic consequences in post-MI remodelling. Unlike in earlier work, here we induced MI in the anterior wall, which has a larger impact on LV remodelling than MI in the posterior wall.,

2. Methods

Permission for the use of tissue from human hearts in the heart transplantation programme, explants, and non-used donor hearts, was obtained from the Ethical Committee of the University Hospital (UZ Leuven) with permit number S58824. The study conforms to the Helsinki declaration. Patients received usual care HF medication and data can be found in Supplementary material online, .

Animals were housed and treated according to the Guide for the Care and Use of Laboratory Animals (National Institute of Health, USA), the European Directive 2010/63/EU and experimental protocols were approved by the in-house ethical committee (Ethische Commissie Dierproeven, KU Leuven), with permit numbers P10139, P14176, and P16110.

Cell isolation, confocal microscopy, and electrophysiological recordings were performed as previously described. Single LV myocytes were enzymatically isolated from explanted human hearts (failing, HF, and non-failing, non-HF) and from pigs with MI in the anteroseptal region, with reduced LV function and cellular remodelling, (see Supplementary material online, ). In human hearts, myocytes were obtained from the territory of the left anterior descendens coronary artery (LAD), cannulating a diagonal side-branch. If the LAD was not accessible, the left circumflex branch was used. In MI pigs, myocytes from the peri-infarct regions were prepared by cannulating a LAD side-branch and the same region was selected in SHAM animals. We used low (0.5 Hz) and high (2 Hz) frequency stimulation in the whole-cell voltage-clamp mode (−70 to +10 mV); the Ca2+ indicator fluo-4 was included in the pipette solution. Ca2+ sparks were recorded for 15 s after stimulation during confocal line-scan imaging. Cells were scanned along the longitudinal axis, orthogonal to the Z lines, avoiding scanning through nuclei. All experiments were performed at 37°C. Ca2+ sparks were assigned to coupled/non-coupled release sites based on an established algorithm. A per-pixel spark frequency was calculated and normalized to give a measure in sparks/100 μm/s. Spark frequency was normalized to the length of the line in coupled/non-coupled sites. Ca2+ waves were induced at 2 Hz stimulation in the presence of isoproterenol (ISO; 10 nM). Ca2+ wave incidence was normalized to number of waves/100 µm/s and wave origins (pixel-based) were used to assign waves to coupled/non-coupled sites.

Immunofluorescent imaging was performed in fixed isolated cells or tissue sections. NCX was used as a surrogate marker to report on TATS membranes and distance masks were constructed as previously described. From these masks, the fraction of coupled/non-coupled RyRs was calculated and co-localization of coupled/non-coupled RyRs with CaMKII was quantified according to the Manders’ coefficient. To study TATS in relation to subcellular Ca2+ release, non-fixed myocytes were stained with wheat-germ-agglutinin-Alexa594 as described previously.

Measurements of FAD/NAD(P)H for estimating redox potential were performed at 37°C under electrical field stimulation at various frequencies by wide field fluorescence imaging using a Nikon TE2000 inverted microscope equipped with a 40 × 1.3 NA oil immersion objective and Hamamatsu Flash 4 CMOS camera. Endogenous fluorescent FAD and NAD(P)H signals were measured: NAD(P)H was excited at 360/10 nm and emission was collected at 480/40 nm (500 ms), FAD was excited at 485/20 nm and emission was collected at 535/30 nm (500 ms). Excitation and emission filters were selected using filter wheels (Sutter) and wavelengths discriminated using a triple band pass dichroic filter. Images were collected at a frequency of a ratio pair every 20 s.

2.1 Statistics

All data are presented as means ± SEM. For statistical analysis, we used a nested design taking into account n cells and N animals (freeware programme R, R Core Team (2013). R: A language and environment for statistical computing.), and Student’s t-test or a two-way ANOVA with Bonferroni post hoc testing as applicable. The Fisher’s exact test was used to compare incidence of Ca2+ waves between different groups. Data were considered significantly different when the probability value was < 0.05.

Detailed methods are provided in the Supplementary material online.

3. Results

3.1 Myocyte hypertrophy and increased fraction of non-coupled release sites in human HF

HF myocytes were longer than non-HF, without difference in cell width (Figure ). The density of TATS was significantly lower in HF myocytes and axial tubules were more abundant (Figure ).

Myocyte hypertrophy and increased fraction of non-coupled release sites with an increased spark frequency in human HF. (A) Cell size and TATS characteristics in non-HF (ncells = 27–31; Nhumans = 4) and HF (ncells = 83–91; nhumans = 5). Data are presented as mean value per subject. (B) Left: Examples of confocal line-scan recording of Ca2+ transients in non-HF and HF. Middle and right: Distribution and quantification of TF50 of the Ca2+ transient in coupled (<14.5 ms) and non-coupled (>21.5 ms) sites in non-HF (ncells = 18; Nhumans = 4) and HF (ncells = 32; Nhumans = 5). (C) Left: Example of a confocal line-scan recording using TF50 of the Ca2+ transient to assign sparks to different release sites. Middle: Spark frequency in coupled and non-coupled sites in non-HF (ncells = 22; Nhumans = 4) and Right: HF (ncells = 32; Nhumans = 6) (*P-value < 0.05; **P-value < 0.01; ***P-value < 0.001).

To identify RyR subpopulations in human myocytes, we used a previously described algorithm, establishing criteria as shown in the Supplementary material online, . These pre-established criteria allow studying of coupled and non-coupled release sites in a more rigorous way than used previously,, excluding intermediate release sites (between > 0.5 and < 2 µm) (see Supplementary material online, ). This approach still includes on average 80% of all release sites in all groups. As shown by its rightward shift, the distribution of the time to half-maximum of the Ca2+ transient (TF50) for all release sites was significantly increased in HF myocytes (Figure , middle). Quantification of early (coupled) and late (non-coupled) sites confirmed the decrease in the number of coupled sites and the increase of non-coupled sites in HF (Figure , right). These data indicate that in human HF, hypertrophy is associated with an altered TAT network and an increased proportion of non-coupled release sites.

3.2 Increased activity of non-coupled release sites in HF, dependent on CaMKII

In non-HF myocytes, high-frequency stimulation caused a three-fold increase in spark frequency in coupled sites compared with that observed under low-frequency stimulation, whereas the spark frequency was only increased by 1.5-fold in non-coupled release sites (Figure , middle). In HF myocytes, this frequency-dependent response was absent at coupled sites (Figure , right), despite a substantial increase in SR Ca2+ load at 2 Hz in both HF and non-HF (Figure ); in HF SR Ca2+ content tended to be lower. In contrast, frequency of sparks at non-coupled sites in HF myocytes was significantly increased at 2 Hz (Figure , right). The specific CaMKII inhibitor autocamtide-2-related-inhibitory-peptide (AIP) significantly reduced this hyperactivity in non-coupled regions without affecting spark frequency at coupled sites (Figure ).

CaMKII inhibition normalized both RyR activity and Ca2+ transients in human HF myocytes. (A) The effect of AIP in coupled and non-coupled sites in non-HF (ncells = 19; Nhumans = 4) and HF (ncells = 23; Nhumans = 6). (B) Ca2+ transient amplitude in non-HF (ncells = 19; Nhuman = 4) and HF (ncells = 25; Nhuman = 6) at 2 Hz stimulation with examples of line-scan and reduced synchrony. The fraction of the line-scan with at least half-maximal Ca2+ release as a function of time. SR Ca2+ content in non-HF and HF at 0.5 Hz (ncells = 8; Nhumans = 4 in non-HF; ncells = 16; Nhumans = 6 in HF) and 2 Hz (ncells = 11; Nhumans = 4 in non-HF; ncells = 20; Nhumans = 6 in HF). (C) Ca2+ transient amplitude in non-HF (ncells = 19; Nhuman = 4) and HF (ncells = 25; Nhuman = 6) at 2 Hz and the effect with AIP (non-HF: ncells = 19; Nhuman = 4 and HF: ncells = 23; Nhuman = 6) (*P-value < 0.05; **P-value < 0.01; ***P-value < 0.001).

3.3 Slower and smaller global Ca2+ transients in HF with beneficial effect of CaMKII inhibition

A reduced synchrony of Ca2+ release has previously been shown to lead to a slow rise of the Ca2+ transient in human HF myocytes. Here we confirm that an increased fraction of non-coupled sites in HF myocytes in comparison to non-HF leads to a slower upstroke and reduced amplitude of the Ca2+ transient (Figure ). This dys-synchrony is further illustrated by the delay in the fraction of the line-scan that reached half-maximal Ca2+ release in HF myocytes compared with non-HF. Interestingly, inhibition of CaMKII that reduces diastolic release, removes the differences between HF and non-HF for amplitude and rate of upstroke (Figure ).

3.4 Specific modulation of coupled sites is lost after MI, whereas a de novo regulation appears at non-coupled release sites

Having established remodelling of TATS and spark properties in human HF, we used the post-MI pig model to gain further mechanistic understanding. The pig shares many features of cardiac physiology with humans, and ischaemic cardiomyopathy is the major cause of human HF. The relevance of the pig MI model is further underscored by earlier pilot data indicating that coupled and non-coupled release sites have properties similar to those seen here in human HF and CaMKII dependence was similar to human HF.

In MI myocytes in the present study, likewise the CaMKII inhibitor AIP did not affect the spark frequency in coupled sites (Figure ), but reduced the spark frequency in non-coupled release sites (Figure , see Supplementary material online, for baseline data). Global Ca2+ transients in MI myocytes were smaller with reduced rate of upstroke, as in HF, and consistent with the loss of TATS and reduced synchrony of Ca2+ release (see Supplementary material online, ).

CaMKII-dependent and ROS-dependent signalling examined at coupled vs. non-coupled sites after MI in pig ventricular myocytes. (A) Left: The effect of AIP in coupled sites in MI (ncells = 16; Npigs = 7) and SHAM (ncells = 15; Npigs = 4). Right: The effect of gp91 ds-tat peptide in coupled sites in MI (ncells = 14; Npigs = 4) and SHAM (ncells = 13; Npigs = 4). (B) Left: The effect of AIP in non-coupled sites in MI (ncells = 16; Npigs = 7) and SHAM (ncells = 15; Npigs = 4). Right: The effect of NAC in non-coupled sites in MI (ncells = 10; Npigs = 5) and SHAM (ncells = 10; Npigs = 3) (*P-value < 0.05; **P-value < 0.01).

We further examined signalling and microdomains in the RyR subpopulations focussing on ROS as a candidate for post-MI signalling. NOX2 signalling is known to change in cardiac disease and could be affected because of the TATS remodelling, while changes in metabolism and mitochondrial function could affect mito-ROS production. In contrast to SHAM myocytes, inhibition of NOX2 with gp91 ds-tat peptide, did not affect the spark frequency in coupled sites after MI (Figure ). Since non-coupled sites lack TATS, NOX2-generated ROS at the TAT membrane is unlikely to affect non-coupled RyRs as confirmed by the lack of an effect of gp91 application (see Supplementary material online, ). Using the global ROS scavenger, NAC, we probed for ROS-mediated modulation from other sources. NAC significantly reduced the spark frequency in non-coupled sites (Figure ). None of these interventions affected the SR Ca2+ load (see Supplementary material online, ).

3.5 CaMKII and NOX2 localization is intact after MI

The lack of frequency-response in coupled release sites after MI could be attributed to the specific loss of CaMKII and/or NOX2 in coupled regions. In both SHAM and MI myocytes, there was however a similar degree of co-localization between CaMKII and coupled or non-coupled RyRs (Figure ). No change in CaMKII expression in SHAM and MI tissue was detected (see Supplementary material online, ). The localization of NOX2 at the TATS was intact after MI (Figure , see Supplementary material online, ).

Ca2+ and ROS in the dyadic cleft after MI in pig myocytes. (A) a. Example of a confocal image of a pig ventricular myocyte stained for RyR (blue), CaMKII (green), and NCX (red). The co-localization of CaMKII in coupled and non-coupled RyRs is shown using the Manders’ coefficient in SHAM (ncells = 26; Npigs = 6) and MI (ncells = 32; Npigs = 7). b. Example of tissue samples in SHAM and MI stained for gp91 (red), NCX (green) and overlay (yellow). Scale bar equals 5 µm. (B) Example and mean values of ICaL inactivation in SHAM (ncells = 19; Npigs = 3) and MI (ncells = 24; Npigs = 5). (C) The effect of L-NAME (ncells = 14; Npigs = 4) and L-VNIO (MI: ncells = 15; Npigs = 4; SHAM: ncells = 12; Npigs = 3) in coupled sites in MI and SHAM (*P-value < 0.05; **P-value < 0.01; ***P-value < 0.001).

3.6 Reduced [Ca2+] and ROS scavenging in the dyadic cleft after MI

A potential mechanism for the lack of CaMKII activation in coupled regions after MI remodelling is the loss of high Ca2+ microdomains in the dyadic cleft. We analysed local [Ca2+] by measuring ICaL inactivation and local Ca2+ transient kinetics. The early inactivation of ICaL was decreased (larger Taufast), indicating reduced [Ca2+] near LTCC (Figure ). In addition, the amplitude and rate of upstroke of Ca2+ transients in regions of coupled sites were reduced (see Supplementary material online, ).

We tested the hypothesis that changes in NOS1 could reduce ROS in the dyadic cleft, as e.g. produced by local NOX2 (Figure ). Recently, local ROS scavenging mechanism at the SR have been described, whereby NOS1-derived NO can scavenge xanthine oxidase (XO)-produced ROS. During disease, NOS1 has been shown to translocate from the SR to the TATS. Translocated NOS1-derived NO at the TATS could thus mask a local ROS microdomain by scavenging ROS produced by NOX2 in MI myocytes. Inhibition of NOS1 with L-NAME, and with the more specific L-VNIO produced a significant increase in the spark frequency in coupled sites in MI myocytes to a similar level as in control. NOS inhibition did not affect the frequency-response in coupled sites in SHAM myocytes (Figure ).

3.7 ROS from mitochondria modulate non-coupled RyRs after MI

The data in Figure imply that there must be an alternative ROS source to NOX2 affecting the non-coupled sites in MI. We hypothesised that the mitochondria were responsible. Scavenging mitochondrial ROS (mito-ROS) using mitoTEMPO significantly reduced spark frequency at non-coupled regions in MI, without affecting coupled sites (Figure ); there was no effect in SHAM. SR Ca2+ content was unchanged by this manoeuver in SHAM or MI (see Supplementary material online, ). Inhibition of XO with oxypurinol had no effect on sparks in either SHAM or MI myocytes (see Supplementary material online, ) indicating that this enzyme did not contribute to ROS generation. These observations raised the question whether pacing at 2 Hz selectively induced mito-ROS in MI but not in SHAM. To test this possibility, we studied cellular redox potential through measuring the FAD/NAD(P)H ratio. During stimulation at 2 Hz, the NAD(P)H signal decreased while the FAD signal increased. Comparison of the ratio of these changes between SHAM and MI revealed a greater increase in FAD/NAD(P)H in MI myocytes (Figure ).

Mito-ROS microdomain affects non-coupled sites after MI in pig myocytes. (A) The effect of mitoTEMPO on spark frequency in coupled and non-coupled sites in MI (ncells = 14; Npigs = 4) and SHAM (ncells = 13; Npigs = 5). (B) a. Example traces of NAD(P)H and FAD signals, measured over time as the mean fluorescence intensity, in SHAM and MI at 2 Hz. b. For each cell, the FAD/NAD(P)H ratio measured after pacing at 2 Hz was normalized to the rest value. The graph represents the frequency distribution with the fitted Gaussian curves in SHAM and MI (*P-value < 0.05; ***P-value < 0.001).

3.8 After MI, Ca2+ waves are more frequent and originate in non-coupled release sites

Increased RyR activity/sensitivity may result in spontaneous Ca2+ waves that promote arrhythmias. We investigated the susceptibility for Ca2+ waves in MI during β-adrenergic stimulation at 2 Hz, mimicking the in vivo adrenergic effects.

MI myocytes exhibited more Ca2+ waves than SHAM, and these Ca2+ waves generated significantly larger NCX currents (INCX) (Figure ). This was reflected by an increased number and amplitude of delayed-afterdepolarizations (DADs) as well as an increase in number of spontaneous action potentials (APs) (Figure ). We related the source of Ca2+ waves to either coupled and non-coupled sites based on the TF50 of the Ca2+ transient. A distribution plot shows that after MI, Ca2+ waves originated more frequently from non-coupled sites (Figure ).

Ca2+ wave frequency increases and Ca2+ wave origin shifts towards non-coupled release sites after MI in pig myocytes. (A). Examples of Ca2+ waves in SHAM and MI are shown. (B) Wave frequency (SHAM: ncells = 59; Npigs = 13, MI: ncells = 62; Npigs = 16), wave-induced Ca2+ transient amplitude (SHAM: nwaves = 30 in ncells = 25 in Npigs = 9; MI: nwaves = 49 in ncells = 38 in Npigs = 11) and wave-induced INCX (SHAM: nwaves = 59 in ncells = 32 in Npigs = 9; MI: nwaves = 91 in ncells = 38 in Npigs = 11) for SHAM and MI. (C) DAD incidence, DAD amplitude and spontaneous APs in SHAM (ncells = 28; Npigs = 6) and MI (ncells = 34; Npigs = 7). (D) Left: Waves are assigned to coupled or non-coupled release sites by using the wave origin (pixel-based) and TF50 of the Ca2+ transient. Middle: The distribution of wave origins and Right: wave frequency in SHAM (ncells = 39; Npigs = 9) and MI (ncells = 62; Npigs = 16). (E) a. Example and quantification of colliding and single waves in SHAM and MI. b. Ca2+ wave peak INCX and integrated INCX in coupled and non-coupled sites in SHAM and MI. Colliding waves are indicated in green. Only colliding waves are included when originating from only coupled or non-coupled sites. (nwaves = 11–28, ncells = 32–38, Npigs = 9–11) (*P-value < 0.05; **P-value < 0.01; ***P-value < 0.001).

To induce triggered activity, and ultimately arrhythmias, membrane depolarization by the inward INCX generated by the Ca2+ waves must reach the threshold for AP generation. Larger INCX is generated when multiple Ca2+ waves collide. In MI, more cells exhibited colliding Ca2+ waves compared with SHAM (Figure ). These colliding waves had a larger amplitude and INCX compared with single Ca2+ waves in MI myocytes (see Supplementary material online, ). Colliding waves originating from non-coupled sites had the highest peak and integrated INCX (Figure ).

3.9 Inhibition of CaMKII and mito-ROS reduces Ca2+ waves from non-coupled sites in MI

In MI myocytes, mitoTEMPO and AIP reduced the increased wave frequency originating from non-coupled sites without significantly affecting those arising from coupled regions (Figure ). Direct inhibition of RyR opening with flecainide or S107 reduced Ca2+ waves in both MI and SHAM.

Non-coupled release sites are more sensitive to RyR blockers and inhibition of RyR modulators after MI in pig myocytes. (A) Effects of flecainide (SHAM: ncells = 27; Npigs = 5; MI: ncells = 26; Npigs = 6), S107 (SHAM: ncells = 19; Npigs = 3; MI: ncells = 27; Npigs = 4), AIP (SHAM: ncells = 22; Npigs = 3; MI: ncells = 25; Npigs = 9) and mitoTEMPO (MI: ncells = 23; Npigs = 5) on (A) wave frequency and (B) global Ca2+ transient in SHAM and MI (*P-value < 0.05; **P-value < 0.01).

We assessed the impact of these interventions on excitation-contraction coupling and the Ca2+ transient. Flecainide, CaMKII and mitoTEMPO did not affect the Ca2+ transient amplitude, whereas S107 significantly reduced the amplitude in both SHAM and MI myocytes (Figure ).

4. Discussion

TATS remodelling in ventricular myocytes after MI and in HF, leads to an increase in non-coupled release sites. Unlike in the healthy heart, these sites, outside of couplons, exhibit increased spontaneous activity during diastole at higher heart rates. This arises through distinct alterations in Ca2+ and ROS microdomains around the different RyR populations. After MI, non-coupled release sites are at the origin of Ca2+ waves. These Ca2+ waves generate larger depolarizing currents by simultaneously activating multiple Ca2+ release sites, thereby highlighting the arrhythmogenic potential of these sites.

4.1 TATS remodelling in human HF impacts on RyR activity and global Ca2+ transient properties

We observed a significant reduction in TATS density, predominantly of the T-tubules (TTs) in HF, extending earlier studies in human tissue. Early electron microscopy studies described dilated TTs. Other features previously described include an irregular distribution,, an increase in TT diameter and a reduction in TT density.

TATS remodelling results in an increased fraction of non-coupled RyR cluster sites, in line with data from a canine model of HF and human HF myocytes. This remodelling is associated with loss of synchrony and reduced amplitude of the Ca2+ transient through delayed opening of the non-coupled RyRs. Similar findings were reported in animal models,, but data from human HF are scarce or did not include information from non-HF myocytes.,

This study also adds new information regarding the intrinsic properties of RyRs clusters at coupled and non-coupled sites. Indeed, previous studies have only reported on overall RyR activity recorded as sparks in diastole and in lipid bilayer experiments, without distinguishing subpopulations.,, They mostly report an increase in total spark frequency, although Lindner et al. reported a decrease in HF. An important difference between the aforementioned study and ours is that we studied cells under voltage-clamp. The voltage-clamp mode with controlled membrane potential and equal duration of membrane depolarization achieves a more equal SR load and Ca2+ fluxes allowing better dissection of RyR properties with less influence of confounding factors (including disease associated remodelling of the AP and resting membrane potential). Our finding that the spark activity in non-coupled sites is sensitive to CaMKII inhibition is in line with previous reports on RyR properties in HF, although a comparison to non-HF was not included in those studies., Recently, lipid bilayer experiments on RyR activity support these findings, with a major role for RyR phosphorylation and thiol oxidation in HF conditions. In line with the study from Fischer et al., no differences in response to CaMKII inhibition were found between myocytes from ischaemic and dilated cardiomyopathy (see Supplementary material online, ). The data complement earlier work on the global assessment of Ca2+ transients in HF vs. non-HF human myocytes. Our data further show that in HF, CaMKII inhibition can reduce diastolic events with favourable effects on the global Ca2+ transient.

4.2 Loss of dyadic cleft CaMKII signalling after MI

Despite its presence at both coupled and non-coupled RyR clusters, we did not detect CaMKII activation in the dyadic cleft near coupled sites in MI, an effect possibly due to reduced [Ca2+] in this microdomain. This could be a primary event after MI with several potential causes. The SR Ca2+ content is not reduced, excluding this possibility. Reduced coupling gain between LTCC and RyR is another possibility, as in the rabbit with HF., Litwin et al. proposed a different molecular configuration of LTCC as the underlying mechanism whereas Bryant et al. reported a redistribution of ICaL away from the TT., Another possibility is the structure of the RyR cluster. Cluster size and organization affect Po, and may change in disease as recently reported for AF.

Furthermore, Erickson et al. recently reported that exposure to NO prior to Ca2+/CaM binding to CaMKII strongly suppresses CaMKII activity. This could explain the loss of CaMKII activation in coupled regions after MI, in line with the increased spark frequency with NOS inhibition. NO production may mask the ROS microdomain through translocation of NOS1 from the SR to the TATS. This may be through ROS in the dyadic cleft or ROS from mitochondrial sources.

Another possible mechanism may be through dysregulation of localized antioxidant mechanisms. Glyceraldehyde 3-phosphate dehydrogenase, located at the TATS, potentially regulates the redox state at SR–SL junctions by the production of NADH. In support of this notion, ischaemic and dilated cardiomyopathies are known to cause a shift in metabolic substrate utilization from free fatty acids to glucose,, thereby increasing the rate of glycolysis and associated NADH production, resulting in ROS scavenging at the dyadic cleft.

4.3 Non-coupled release sites have increased spontaneous activity

Our data show a de novo form of modulation of non-coupled sites, dependent on CaMKII activation and mito-ROS. Increased CaMKII activity in the regulation of RyRs has been associated with several cardiac diseases,,, but its selective activation at non-coupled regions was so far not reported. The low rate of upstroke of triggered Ca2+ release in non-coupled regions makes it unlikely that CaMKII is activated by Ca2+ in non-coupled regions. Instead, our data suggest a more prominent role for mito-ROS, rather than Ca2+, in the activation of CaMKII, as recently shown in diabetic mouse heart. Moreover, increased levels of mito-ROS manifest in patients and animals with advanced HF.

The increased activity of RyRs at non-coupled sites in MI occurs without apparent change in SR Ca2+ content. We could not yet study the full relationship between SR leak and content. It is conceivable that our experiments occur at the exponential part of this relationship together with a shift in the curve due to RyR post-translational modifications. Therefore, one can have increased leak with small or no changes in SR Ca2+ content.

4.4 Non-coupled sites are hotspots for arrhythmogenic Ca2+ waves

The increased wave frequency that we observe make non-coupled RyR clusters a potential source of deleterious Ca2+ handling in MI. In contrast, Belevych et al. recently proposed that potential arrhythmogenic Ca2+ waves arise from coupled sites in myocytes from the dog with pacing-induced HF. This opposing finding may reflect species differences and different types of remodelling, as there was no loss of TT in this HF model. Also, differences in experimental conditions may help to explain the different findings between the two studies. Belevych et al. recorded at room temperature and 0.5 Hz pacing, in contrast to 37°C and 2 Hz pacing frequency in this study. Of note, at 0.5 Hz conditioning, we did not see the distinguishing activity between coupled and non-coupled RyRs.

Inhibition of diastolic RyR Ca2+ leak could be a therapeutic strategy with direct RyR block or modulation. In this study, Ca2+ waves were reduced by both RyR blockade and inhibition of RyR modulation. The S107 and flecainide-dependent reduction in Ca2+ wave frequency is in line with data previously reported under various conditions,, although how flecainide reduces Ca-wave dependent arrhythmias is debated.

The role of CaMKII in arrhythmias has been previously investigated, including ROS-induced CaMKII activation., Here we show that CaMKII inhibition in MI selectively reduced waves originating from non-coupled RyRs without adverse effects on coupled RyRs and excitation-contraction coupling. CaMKII inhibition or mito-ROS scavenging would thus be better strategies than direct RyR inhibition, which is less discriminating.

Last, the impact of waves on depolarizing currents was larger in MI. This may at first glance be counterintuitive as non-coupled sites are more distant from the membrane. However, this distance may also cause a temporary increased re-circulation of Ca2+ in the stores with eventually larger cell-wide Ca2+ release, colliding waves, larger INCX, and ultimately, triggered activity. In a pilot study, in vivo monophasic AP from the peri-infarct region during 2 Hz pacing and isoproterenol support the occurrence of high amplitude DADs (see Supplementary material online, ), whereas only low amplitude DADs were observed in SHAM animals. In three out of five MI pigs, this pro-arrhythmic pacing protocol triggered premature ventricular complexes in vivo. This was rarely observed in SHAM animals.

4.5 Limitations and future directions

Human samples are limited in nature and quantity and not available for all experiments. Nevertheless, our key observations in pig suggest comparable mechanisms after disease remodelling in human HF myocytes.

In this study, only indirect and global ROS measurements were used although development of local ROS measurements could provide further insights. For local [Ca2+], targeted probes are available but cell culture complicates the application as culture affects TT ultrastructure.

Our biochemical characterization of RyR and CaMKII modifications remains limited. No differences were found for the direct oxidation of RyR between SHAM and MI tissues (see Supplementary material online, ). Immunostainings for both phosphorylation and oxidation of CaMKII showed a significant increase in MI myocytes (see Supplementary material online, ). These approaches are however not sufficient to detect dynamic changes during live cell experiments at the subcellular level. In the current experiments, we have relied on pharmacological tools and despite the use of different agents and controls, there is always the potential for non-selective or indirect effects. This limitation was preferred over gene manipulation that requires cell culture, introducing other potential confounders.

In this work, a fraction of nearly 20% of non-coupled RyRs was observed in SHAM pig myocytes using functional and immunostaining methods (see Supplementary material online, ). This is high compared with studies in human tissues, including present data, and dog myocytes., Species-specific and methodological differences might account for some of this apparent disparity. Here and in our previous work, we predominantly rely on the relationship between SR Ca2+ release and TATS distance to identify RyR subgroups. If not all RyRs are active this may introduce discrepancies.

5. Conclusion

In HF and after MI, RyR microdomain organization enhances spontaneous Ca2+ release and sparks from non-coupled sites, in a manner that is dependent on CaMKII activation and mito-ROS production. This specific modulation may generate a substrate for arrhythmia that appears to be responsive to selective pharmacologic modulation.

Click here for additional data file.