Heart failure (HF) is one of the major causes of death worldwide. It is characterized by the failure of the cardiac pump to maintain a sufficient blood flow to oxygenize and carry nutrients to the whole body. According to left ventricular systolic function, HF has been divided into two major groups: HF with reduced ejection fraction (HFrEF) and HF with preserved ejection fraction (HFpEF). HFrEF generally occurs after cardiac injury (myocardial infarction) or under chronic stress (hypertension), leading to the alteration of contractile function of the heart. It is now well established that alteration of cardiomyocyte Ca²⁺ homeostasis plays a critical role in the development of the pathology, leading to cardiac remodeling, failure of the cardiac pump, and cardiac arrhythmias.

Ca²⁺ plays a key role in cardiomyocyte contraction. In each heartbeat, the membrane depolarization during an action potential (AP) activates L-type Ca²⁺ channels (LTCC), which are located in the sarcolemma and are more concentrated at the transverse (T) tubules (Figure 1). The local increase in [Ca²⁺]_i in the dyads that follows activates the Ca²⁺ release channels, the ryanodine receptors (RyR2) located in the neighborhood, on the junctional sarcoplasmic reticulum (SR), resulting in the coordinated and global increase in the cell [Ca²⁺]_i, by the Ca²⁺-induced Ca²⁺-release (CICR) mechanism, activating contractile myofibrils. Relaxation happens when the cytosolic [Ca²⁺]_i decreases, mainly by Ca²⁺ re-uptake into the SR through the SarcoEndoplasmic Reticulum Ca²⁺ ATPase (SERCA) and extrusion out of the cell through the Na⁺/Ca²⁺ exchanger (NCX). Other systems, such as the plasmalemmal Ca²⁺ pump and the mitochondrial Ca²⁺ uniporter, play a minor role in cytosolic Ca²⁺ removal. The RyR2 are not so sensitive to Ca²⁺ so their activation probability depends on their proximity to the LTCCs, which determines the local Ca²⁺ concentration (Stern, 1992).

During diastole, some Ca²⁺ leaks outside the SR through the RyR2s, which have very low open probability at low cytosolic [Ca²⁺]_i, but are also sensitive to luminal Ca²⁺, what is high during diastole. The opening of a RyR2 cluster can be viewed visualized with a confocal microscope and adequate fluorescence dyes, as rapid (~10 ms), brief (~30 ms), and local (~1.5 μm) elevations in cytosolic Ca²⁺, named Ca²⁺ sparks (Cheng and Lederer, 2008). Ca²⁺ sparks analyses provide insight into RyR2 function in situ, while the incorporation of RyR2 into lipid bilayers provides direct information on the channel function in a non-cellular context. During excitation-contraction coupling (ECC), Ca²⁺ release units (CRU) formed by a cluster of RyR2s are activated by the LTCCs located in the T tubules in front of them, which open during the AP. The spatial and temporal summation of these coordinated Ca²⁺ sparks results in the whole [Ca²⁺]_i transient. The efficacy of the LTCCs to activate CRUs has been referenced as CICR gain, and it has been studied in HF combining patch-clamp to measure L-type Ca²⁺ current (I_Ca) and confocal microscopy to visualize changes in ≪Ca²⁺ⁱ.

In HF, there are alterations in contraction and relaxation, which may be related to alterations in cardiomyocyte Ca²⁺ cycling. Alterations have been reported in several cardiomyocyte Ca²⁺ cycling elements, such as SERCA, ratio phosphlamban (PLN)/SERCA, and single LTCCs, during both experimental and human HF. In this review, we will focus on alterations in the RyR2 function on left ventricular HF, although alterations in Ca²⁺ handling have also been found in right ventricular failure (Hautefort et al., 2019).

The first study of Ca²⁺ sparks in HF showed fewer Ca²⁺ sparks for a given amount of Ca²⁺ entry through LTCCs (I_Ca). This decrease in CICR gain was observed in experimental HF, due to chronic hypertension (Gomez et al., 1997) and after myocardial infarction (Gomez et al., 2001), and was reminiscent of pioneer work by Beuckelmann in human ventricular myocytes from explanted failing hearts, where they found a decrease in the global [Ca²⁺]_i transient but normal I_Ca density (Beuckelmann et al., 1992). One explanation of this decrease in CICR gain was that there could be an ultrastructural disarrangement, which physically increased the distance between CRU and LTCC. In fact, while I_Ca density is generally maintained in HF, higher activity of at the single channel level has been reported (Schroder et al., 1998). This finding suggests fewer LTCC in the T-tubules participating in CICR or a decrease in the T-tubules density, as observed in HF (Wagner et al., 2012; Guo et al., 2013; Jones et al., 2018). But this “uncoupling” of the ECC was also suggested in larger mammals, in which the T-tubular network is less developed. Following myocardial infarction, cardiomyocytes from pigs with HF had relatively more uncoupled RyR2s, which could underlie the decreased gain of CICR (Dries et al., 2018).

Besides this decrease in CICR gain, alterations in the RyR2 itself have been reported in different models of HF.

During HF progression, the sympathetic nervous system and renin-angiotensin-aldosterone system are activated to compensate for the reduced contractile function of the heart and thus maintain a sufficient blood flow (Lymperopoulos et al., 2013). In this early stage of cardiac remodeling, ECC is enhanced, with an increased [Ca²⁺]_i transient amplitude and contraction, together with a faster relaxation (Ohkusa et al., 1997). However, chronic activation of both systems has deleterious effects on cardiomyocytes and is responsible for the decompensation of cardiac function. This decrease in cardiac contraction is associated with an alteration of ECC (Gomez et al., 1997, 2001), the [Ca²⁺]_i transient amplitude is decreased and its duration is increased. Besides, or on top of the ultrastructural remodeling mentioned above, the reduction in [Ca²⁺]_i transient amplitude can be explained by a reduction of Ca²⁺ stored in the sarcoplasmic reticulum (Piacentino et al., 2003; Lehnart et al., 2009). While the first analyses explained this decrease in SR Ca²⁺ load by a lower SERCA expression and function and/or a higher expression of its natural inhibitor PLN, or a decrease in its phosphorylation (Kranias and Hajjar, 2012), some authors attributed it to RyR2 hyperactivity, which enhances diastolic SR Ca²⁺ leak. Indeed, an increase of the spontaneous Ca²⁺ leak through RyR2 has been observed in HF (Fischer et al., 2014; Ho et al., 2014; Grimm et al., 2015; Fu et al., 2016; Uchinoumi et al., 2016; Walweel et al., 2017; Dries et al., 2018). This SR Ca²⁺ leak, together with the reduced SERCA activity, contributes to a decrease in SR Ca²⁺ stores and an increase of the propensity of Ca²⁺ waves (Cheng et al., 1996; Venetucci et al., 2008; Curran et al., 2010; Dries et al., 2018). This contrasted with a significant decrease in the Ca²⁺ spark frequency reported in isolated cardiomyocytes from patients with terminal HF compared with non-failing individuals (Lindner et al., 2002). As spontaneous Ca²⁺ sparks depend on the amount of SR Ca²⁺ stored and [ATP], which are both reduced in cardiomyocytes from failing hearts, these can mask the increased activity of RyR2, whereas this higher RyR2 activity can still favor Ca²⁺ waves once the Ca²⁺ spark is produced, as the neighboring RyR2 will be more sensitive (Ruiz-Hurtado et al., 2015). Extrusion of Ca²⁺ constituting the Ca²⁺ waves during diastole via the NCX can generate delays after depolarization and trigger new action potentials that can propagate and induce arrhythmias (Pogwizd et al., 2001; Venetucci et al., 2008; Belevych et al., 2011). All these modifications of Ca²⁺ release have not been associated with alterations of RyR2 expression during HF (Hasenfuss and Pieske, 2002), but with post-translational modifications such as phosphorylation and oxidation (Houser, 2014). We should also keep in mind that while post-translational alterations in RyR2 tends to increase its activity, other alterations in Ca²⁺-handling proteins (such as SERCA reduction) and metabolic changes (that may curse with lower ATP levels) tend to reduce its activity (Ruiz-Hurtado et al., 2015).

In 2000, it was shown that hyperphosphorylation of RyR2 by PKA is responsible for the leaky RyR2 in failing hearts in humans. The authors claimed that PKA-mediated phosphorylation dissociates the FK506 binding protein (FKBP12.6) from the RyR2 leading to an increase in the channel open probability (Marx et al., 2000). FKBP12.6 is known to stabilize the RyR2 in a closed state and avoid aberrant Ca²⁺ leak (Brillantes et al., 1994; Gellen et al., 2008), but its removal may not affect RyR2 open probability (Xiao et al., 2007). In the early 90s, serine 2808 (S2808) was identified as the unique phosphorylation site of RyR2 (Witcher et al., 1991) and appeared to be a prominent target for PKA. In following publications, Marks' group highlighted the role of S2808 phosphorylation in the destabilization of the complex RyR2-FKBP12.6. They developed RyR2-S2808A knock-in mice, which prevent phosphorylation at this site. They showed that these mice are protected against RyR2-FKBP12.6 dissociation and leaky RyR2 in response to catecholamine stimulation (Shan et al., 2010a,b). Moreover, RyR2-S2808A knock-in mice seem to have a better cardiac function after myocardial infarction, which has been related again to the stabilization of the RyR2-FKBP12.6 complex in the absence of PKA phosphorylation (Wehrens et al., 2006). They also confirmed this major role of S2808 using RyR2-S2808D knock-in mice, which mimicked hyperphosphorylation at that site and had a reduced association of FKBP12.6 to RyR2 (Shan et al., 2010a,b). However, other groups have not been able to confirm this major role of S2808 phosphorylation in the progression of HF (Jiang et al., 2002). Valdivia and Houser's groups also developed RyR2-S2808A knock-in mice. In these mice, response to β-adrenergic stimulation on [Ca²⁺]_i transient and ECC gain (MacDonnell et al., 2008) and progression of HF after myocardial infarction (Zhang et al., 2012) or aortic banding (Benkusky et al., 2007) were not affected no matter the genetic background of the mice (Alvarado et al., 2017). Moreover, it has also been shown that FKBP12.6 binding to RyR2 is not modified by acute β-adrenergic stimulation on wild-type mice or FKBP12.6 overexpressing mice in the heart (Gellen et al., 2008). Recently, evidences that S2808 is not the only target for RyR2 phosphorylation by PKA have been put forward; S2030 seems to be a potent target of PKA (Potenza et al., 2019) and can be involved in the progression of HF (Xiao et al., 2006; Benkusky et al., 2007). Taken together, these studies are in contradiction with Marks' group publications, and the controversy continues through the years (Bers, 2012; Alvarado and Valdivia, 2020; Dridi et al., 2020a,b). However, independently on RyR2 phosphorylation status, FKBP12.6 has been more consistently found decreased in HF (Ono et al., 2000; Gomez et al., 2009), which can by itself promote diastolic Ca²⁺ leak.

The Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) phosphorylation of S2814 (2815) has also been highlighted for its role in the increase in SR Ca²⁺ leak through RyR2 in HF. CaMKII is activated by catecholamine and by oxidation, which is high during HF (Erickson et al., 2008). It has been shown that in failing rabbit hearts, CaMKII phosphorylation of RyR2 decreases SR load and increases Ca²⁺ leak by increasing RyR2 open probability (Ai et al., 2005). In these rabbits, inhibition of CaMKII, but not of PKA, reduced SR Ca²⁺ leak (Ai et al., 2005). In human heart failure due to ischemic or dilated cardiomyopathy, CaMKII expression has been shown to increase, contributing to augmented SR Ca²⁺ leak (Fischer et al., 2014). Rats subjected to myocardial infarction treated with exendin-4, which reduced CaMKII activity, showed a decreased SR Ca²⁺ leak (Chen et al., 2020). This major effect of CaMKII in the development of HF has been confirmed in mice lacking CaMKII. After transverse aortic constriction, these mice were protected against HF progression together with a reduced SR Ca²⁺ leak and RyR2 phosphorylation at S2815 (Ling et al., 2009). Similarly to CaMKII Knock out mice, RyR2-S2814A knock-in mice were protected against abnormal SR Ca²⁺ leak and HF after transverse aortic constriction (TAC) (van Oort et al., 2010) but surprisingly not after myocardial infarction (MI) (Respress et al., 2012). To explain how CaMKII increases SR Ca²⁺ leak, Uchinoumi et al. hypothesize that phosphorylation of S2814 reduces CaM affinity for the RyR2, leading to a RyR2 conformational change and leakiness of the channel (Uchinoumi et al., 2016). Supporting this hypothesis, Dantrolene, which restores RyR2 CaM affinity, suppressed SR Ca²⁺ leak in RyR2 S2814D knock-in mice (Uchinoumi et al., 2016).

Among the 90 cysteines present in each of the four subunits of RyR2, 21 are in a free thiol state and accessible for redox modifications during oxidative stress (Xu et al., 1998). Oxidative stress is associated with the development of several pathologies, and excessive production of reactive oxygen species has deleterious effects on protein function and cell viability. To counter these deleterious effects, cells have their own specific antioxidants machinery, including superoxide dismutase, catalase, and glutathione peroxidase, and non-specific antioxidants reduced glutathione, which is ubiquitous and the most important antioxidant system in cardiac cells (Nikolaienko et al., 2018).

During HF ROS production is chronically increased due to the uncoupling of the mitochondrial electron transport chain, the increase in energy demand, the switch from fatty acids to glucose as an energy substrate in cardiomyocytes (Mak and Newton, 2001; Ventura-Clapier et al., 2004), upregulation of nitric oxide synthase (NOS), xanthine oxidase and NADPH oxidase (NOX), and decreased reduced glutathione (Zima and Mazurek, 2016; Nikolaienko et al., 2018). ROS are also known to activate SR Ca²⁺ leak (Terentyev et al., 2008) by increasing spontaneous Ca²⁺ sparks frequency (Yan et al., 2008; Prosser et al., 2011) and Ca²⁺ waves (Bovo et al., 2012). Indeed, part of the beneficial effect of carvedilol on HF has been attributed to its antioxidant action on RyR2 (Mochizuki et al., 2007). ROS oxidation of RyR2 during HF induces the dissociation of FKBP12.6 leading to the aberrant Ca²⁺ release (Shan et al., 2010a), although other groups showed that RyR2 oxidation induces CaM dissociation without changing FKBP12.6 binding (Ono et al., 2010; Oda et al., 2015). Increased NOX activity induced by tachycardia led to an increase in RyR2 S-glutathionylation associated with an increased SR Ca²⁺ leak (Sanchez et al., 2005), which has been suggested to be cardio protective during preconditioning with exercise. However, the role of NOX during HF is not well-documented. The increased activity of NOX during end-stage HF (Heymes et al., 2003) can participate in the increase in diastolic SR Ca²⁺ leak and SR Ca²⁺ depletion. Even if hypernitrosylation of RyR2 is supposed to increase Ca²⁺ leak, decreased S-nitrosylation has been associated with increased SR Ca²⁺ leak during HF (Gonzalez et al., 2007; 2010). In fact, it has been proposed that hyponitrosylation during HF favors the oxidation of RyR2 by ROS, leading to this aberrant Ca²⁺ release. In these studies, inhibition of xanthine oxidase decreased ROS production restoring S-nitrosylation and cardiac function (Gonzalez et al., 2010). Evidence of the cardio protective effect of RyR2 S-nitrosylation has been also highlighted in a mouse model where NOS1 overexpression prevents cardiac dysfunction and delays HF in response to pressure overload (Loyer et al., 2008). On a canine model of HF, it has been suggested that the disulfide oxidation is the predominant form of redox-sensitive modulation of RyR2 compared to S-nitrosylation and S-glutathionylation to influence the RyR2-mediated leak. On a rabbit model of HF, increased SR Ca²⁺ leak (Mazurek et al., 2014) and Ca²⁺ waves (Bovo et al., 2015) could be attributed to an increase in RyR2 intersubunit disulfide cross-linking (Mazurek et al., 2014; Bovo et al., 2015, 2018).

By accessing data from different laboratories, we can gather that the RyR2 itself may be more active in HF due to several post-translational modifications such as oxidation, enhanced phosphorylation at the CaMKII site, as well as a depressed FKBP12.6 expression/binding to RyR2.

Figure 1 summarizes the key elements in Ca²⁺ handling in ventricular cardiomyocytes, as well as a representation of some of the reported findings in those elements, in different models of HF, which contribute to depressed contraction and pro-arrhythmogenic DADs production. The failing cell has a disruption in the T-tubular network, and a decreased SERCA expression (or augmented ratio PLN/SERCA), which together with more active RyR2 due to post-translational modifications, resulting in a drop of Ca²⁺ stored in the SR. As a result, and despite normal I_Ca density and prolonged AP, the triggered [Ca²⁺]_i transient is of lower amplitude, contributing to diminished cell contraction. During diastole, the higher RyR2 activity favors the probability of producing Ca²⁺ waves, generating an inward current through the NCX, when extruding Ca²⁺, which produces DADs. When DADs reach the threshold, a triggered AP is generated, which may initiate arrhythmias.

HF with preserved ejection fraction (HFpEF) has been identified as a subtype of HF in addition to HFrEF. HFpEF is more frequently found in older patients with comorbidities such as hypertension, obesity, diabetes, chronic kidney diseases (Redfield et al., 2003; Bhatia et al., 2006; Owan et al., 2006). HFpEF patients present normal left ventricular systolic function with impaired left ventricular relaxation and filling. Although HFpEF represents approximately 50% of HF patients (Dunlay et al., 2017), little is known about their specific mechanisms, in comparison to HFrEF, notably regarding RyR2 and Ca²⁺ release. A lack of knowledge is probably related to the difficulty in generating accurate experimental models or the diversity of associated comorbidities. As described previously, in HFrEF, cardiac systolic and diastolic changes are linked to Ca²⁺ mishandling. In HFpEF, Ca²⁺ alterations are not as clearly defined and seem to differ depending on the associated comorbidity. In a rat model in HFpEF with chronic pressure overload, a significant increase in diastolic SR Ca²⁺ release through RyR2s leading, as in HFrEF, to higher diastolic [Ca²⁺] and irregular [Ca²⁺]_i transients during pacing has been observed (Rouhana et al., 2019). This SR Ca²⁺ leak has been associated with an increase in PLN/SERCA ratio responsible for a delayed [Ca²⁺]_i transient decay and diastolic dysfunction, also seen in HFpEF with chronic kidney disease (Primessnig et al., 2016) and in human HF with EF > 45% (Hohendanner et al., 2013; Ljubojevic et al., 2014). Unlike HFrEF, a SR Ca²⁺ leak in HFpEF depends on PKA-dependent phosphorylation of RyR2 at S2808 (Durland, 2021) rather than CaMKII-dependent phosphorylation, alteration of FKBP12.6/RyR2 ratio, or RyR2s S-nitrosylation (Adeniran et al., 2015; Frisk et al., 2021). However, phosphorylation of RyR2 at S2808 was not significantly changed in Dahl salt-sensitive rats, with HFpEF, presenting both hypertension and insulin-resistance (Kilfoil et al., 2020). This discrepancy could be explained by insulin-resistance comorbidity not present in Durland's model (Durland, 2021). Although in HFrEF an increase in SR Ca²⁺ leak leads to SR Ca²⁺ depletion and lower [Ca²⁺]_i transient, in HFpEF, diastolic SR Ca²⁺ release does not induce SR Ca²⁺ load depletion or defective ECC. In fact, in HFpEF models, systolic Ca²⁺ release by the RyR2 is increased, probably as a compensatory mechanism to overcome the increased left ventricular stiffness associated with HFpEF (Selby et al., 2011; Adeniran et al., 2015; Durland, 2021). This enhanced SR Ca²⁺ release could be attributed to a more effective RyR2s recruitment by Ca²⁺ during the Ca²⁺ influx through the LTCC (Kilfoil et al., 2020) as well as an absence of T-tubules alteration (Durland, 2021). Indeed, in a Dahl-sensitive rat model of HFpEF (Kilfoil et al., 2020), a higher couplon recruitment improved ECC by increasing Ca²⁺ release synchronicity and lowering Ca²⁺ release latency. This should allow the heart to maintain an effective contraction besides higher ventricular wall stiffness. Moreover, T-tubule disorganization might also affect Ca²⁺ release effectiveness. However, the integrity of T-tubule structure in HFpEF depends on the comorbidities. Indeed, while T-tubule density is unchanged in Dahl salt-sensitive rats, in obese and diabetic Zucker rats or diabetic HFpEF patients, their density is increased in the ischemic model (post-MI with preserved EF but reduced E/A) (Frisk et al., 2021). This etiology-dependence is also found in proteins regulating SR Ca²⁺ release such as SERCA and NCX. In diabetic HFpEF, NCX expression and activity of SERCA decrease, which is not the case in ischemic or hypertensive disease models (Frisk et al., 2021). In type 1 diabetes, with subclinical diastolic dysfunction and normal systolic function, diastolic SR Ca²⁺ release and amplitude (measured as Ca²⁺ sparks frequency) decrease and is associated with a drop of SR Ca²⁺ load and [Ca²⁺]_i transient without affecting systolic function (Lagadic-Gossmann et al., 1996; Lacombe et al., 2007; Hamouda et al., 2015). In these diabetic models, SR Ca²⁺ uptake by the SERCA pump appears as the main responsible for the decrease in SR Ca²⁺ release with, in some models, an alteration of the SERCA2/PLN ratio (Miranda-Silva et al., 2020). It is deduced from the literature that the alterations in the RyR2s and Ca²⁺ release diverge between HFpEF depending on their diabetic or non-diabetic etiology, with a clear implication of SERCA alteration in the Ca²⁺ homeostasis underlying the diastolic function in diabetes.

Interestingly, in myocardial strips from patients with hypertrophy associated with hypertension and coronary artery disease, but normal ejection fraction (LVEF ≥ 50%), diastolic dysfunction is associated with an increase in SR Ca²⁺ load at high pacing rates (Selby et al., 2011). However, the impact of cellular changes in Ca²⁺ handling does not always transduce into in vivo alterations. Indeed, in isolated myocardial strips from animal models of hypertrophy induced by aortic banding, even though cellular Ca²⁺ extrusion was increased due to higher NCX and SERCA activities, in vivo relaxation remained slower (Roe et al., 2017). Similarly, exacerbated cellular Ca²⁺ handling has been described in a rat strain with cardiac hypertrophy despite the lack of hypertension, where the increased Ca²⁺ influx through the LTCCs seems to underlie the increased [Ca²⁺]_i transient amplitude and myocyte shortening. In this model, an increase in RyR2 phosphorylation by CaMKII leads to higher susceptibly to in vivo arrhythmia and spontaneous SR Ca²⁺ release although RyR2 expression was decreased (Curl et al., 2018).

In summary, alterations in RyR2 posttranslational modifications, location due to T-tubule remodeling, expression, and binding to accessory proteins have been all found in HF with variable conclusions depending on the experimental models or human etiology. What is clear is that RyR2 alteration contributes to the pathology of HF by participating in the depressed [Ca²⁺]_i transient, which is important for depression of cell contraction and in favoring arrhythmogenic Ca²⁺ waves during diastole.

All authors contributed to the conception of the manuscript, wrote, edited the manuscript, contributed to the article, and approved the submitted version.

This work was funded by Inserm, University Paris-Saclay, and grants from ANR (ANR-15-CE14–0005 and ANR-19-CE14-0031-01) and NIH (2R01HL055438-22).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.