Calcineurin in the Heart: New Horizons for an Old Friend

Abstract

Calcineurin, also known as PP2B or PPP3, is a member of the PPP family of protein phosphatases that also includes PP1 and PP2A. Together these three phosphatases carryout the majority of dephosphorylation events in the heart. Calcineurin is distinct in that it is activated by the binding of calcium/calmodulin (Ca²⁺/CaM) and therefore acts as a node for integrating Ca²⁺ signals with changes in phosphorylation, two fundamental intracellular signaling cascades. In the heart, calcineurin is primarily thought of in the context of pathological cardiac remodeling, acting through the Nuclear Factor of Activated T-cell (NFAT) family of transcription factors. However, calcineurin activity is also essential for normal heart development and homeostasis in the adult heart. Furthermore, it is clear that NFAT-driven changes in transcription are not the only relevant processes initiated by calcineurin in the setting of pathological remodeling. There is a growing appreciation for the diversity of calcineurin substrates that can impact cardiac function as well as the diversity of mechanisms for targeting calcineurin to specific sub-cellular domains in cardiomyocytes and other cardiac cell types. Here, we will review the basics of calcineurin structure, regulation, and function in the context of cardiac biology. Particular attention will be given to: the development of improved tools to identify and validate new calcineurin substrates; recent studies identifying new calcineurin isoforms with unique properties and targeting mechanisms; and the role of calcineurin in cardiac development and regeneration.

Graphical Abstract:

Calcineurin (CN) signaling plays diverse roles in cardiac development, from (A) early steps in lineage commitment to (B) critical processes during cardiac morphogenesis. (C) In the adult heart, calcineurin activity is heterogeneous across the ventricular wall following a gradient of increasing stress from the epicardium (Epi) to the endocardium (Endo), as well as displaying a robust 24-hour oscillation. Green arrows indicate steps dependent on CN. Red arrows indicate steps inhibited by CN. embryonic stem cells (ES), mesoderm (ME).

1. Introduction:

Calcineurin was first isolated as the most abundant Ca²⁺/CaM-binding protein present in the brain [1] and was soon shown to function as a Ca²⁺/CaM-activated protein phosphatase [2]. Calcineurin is also abundant in skeletal muscle and heart where it has been shown capable of dephosphorylating an array of sarcolemmal proteins that are phosphorylated by cAMP-dependent kinase (PKA) [3], suggesting a role for calcineurin in the coordination of Ca²⁺ and cAMP signaling cascades that mediate cardiac adaptation. Interest in calcineurin grew exponentially after seminal experiments by Jeff Molkentin and Eric Olson demonstrated that sustained calcineurin activity in cardiomyocytes was sufficient to drive hypertrophic growth of the heart that progressed to decompensated failure and fatal arrhythmias [4]. Numerous investigators have since validated these findings and demonstrated that inhibiting calcineurin will both blunt hypertrophic growth and preserve cardiac function under a variety of physiological and pathological conditions [5-11]. Despite thousands of papers describing a diversity of roles for calcineurin in the cardiovascular system, much remains unknown regarding how this Ca²⁺-responsive enzyme functions in the Ca²⁺-rich environment of a continuously contracting heart without pathological consequences. Multiple mechanisms mediate control and specificity. Some are inherent to the structure and regulation of calcineurin itself, others are mediated by selective targeting of calcineurin and its substrates to subcellular signaling domains.

2. Calcineurin structure and regulation:

Calcineurin is a heterodimer composed of a 62 kilodalton catalytic subunit (CnA) and a 19-kilodalton regulatory subunit (CnB) (Fig. 1). The catalytic domain at the N-terminus of CnA is followed by a binding domain for CnB (BBD) and a binding domain for a Ca²⁺/CaM complex (CBD). An autoinhibitory sequence (AIS) lies immediately adjacent to and overlapping with the CBD. In the absence of Ca²⁺/CaM the AIS blocks one of the substrate docking sites on calcineurin [12]. An autoinhibitory domain (AID) at the C-terminus of canonical CnA subunits engages and blocks the catalytic site when Ca²⁺ levels are low (<100nM) or availability of CaM is limited. Truncations of CnA that remove the CBD/AIS and AID yield CnA*, a constitutively active form that no longer requires Ca²⁺/CaM. It is this truncated form, under the control of the cardiomyocyte-specific alpha myosin heavy chain promoter (αMHC-CnA*) that was initially used to demonstrate the pro-hypertrophic properties of calcineurin [4].

Figure 1: Schematic shows the domain details of calcineurin catalytic and regulatory subunits, as well conserved domains in RCAN proteins using the RCAN1.4 isoform as an example.

The location of calpain cleavage sites are indicated by red arrow points below CnAα. CnAα* depicts the constitutively active calcineurin expressed in αMHC-CnA* transgenic mice. CnB binding domain (BBD), calmodulin binding domain (CBD), autoinhibitory sequence (AIS), Autoinhibitory domain (AID), polyproline rich domain (PP), membrane localization sequence 9MLS), LxVP calcineurin docking motif (LxVP), PxIxIT calcineurin docking motif (PxIxIT), EF-hand calcium-binding domain (EF), RCAN1 motif (SPPASPP), RCAN1 calcineurin inhibitory domain (TxxP).

Calcineurin binds Ca²⁺ both directly, through CnB, and indirectly, through CaM (Fi. 2). The structure of the regulatory subunit, CnB, is similar to CaM, containing four EF-hand Ca²⁺-binding sites. Two sites are high affinity structural sites that are always occupied, and two are lower affinity regulatory sites [13, 14]. Binding of Ca²⁺ to the regulatory sites on CnB causes a conformational change in the holoenzyme allowing subsequent binding of the Ca²⁺/CaM complex. Binding of Ca²⁺/CaM to CnA then displaces the AIS from the substrate-docking site and the AID from the catalytic site, allowing substrates access [12, 15]. On average, during the course of a single cardiac cycle, cytosolic Ca²⁺ raises from 0.1 μM to 1 μM [16], sufficient to activate calcineurin, which requires only 0.6 μM Ca²⁺ for half maximal activity in the presence of 20 μM CaM [17]. The concentration of Ca²⁺ in defined microdomains can raise even higher. Initial influx through L-type voltage gated Ca²⁺ channels (LTCC) increases Ca²⁺ levels in the dyadic cleft between the sarcolemma and the sarcoplasmic reticulum (SR) to as high as 10 μM [18]. Subsequent Ca²⁺-induced Ca²⁺ release from ryanodine receptors (RyR2) in the SR may then raise Ca²⁺ above 100 μM in the dyadic cleft [19].

Calcineurin’s ability to evade beat-to-beat activation likely depends on a number of factors. Among them are the rate of association with Ca²⁺/CaM, regional differences in CaM distribution, competition with other Ca²⁺/CaM-binding proteins, and association with endogenous inhibitors [20]. Signaling microdomains are determined by local concentrations of calcineurin, Ca²⁺, and CaM, as well as the availability of specific substrates [21-23]. Free CaM in adult cardiomyocytes averages between 50 to 100 nM [24] but variations in CaM across microdomains can alter the concentration of Ca²⁺ required to activate calcineurin. As an example, at 20 nM CaM, the level of Ca²⁺ required for half maximal activation more than doubles to 1.3 μM [17]. Thus, higher Ca²⁺ levels are required to activate calcineurin in a domain where CaM is limited or being competed away. Theoretical models of competitive binding between calcineurin and CaM Kinase II (CaMKII), another Ca²⁺/CaM-activated enzyme involved in pathological remodeling of the heart, predict that activation shifts away from calcineurin in favor of CaMKII as the frequency of Ca²⁺ transients increases [25, 26].

2.1. Genes encoding calcineurin:

Three mammalian genes encode CnA catalytic subunits: PPP3CA/CnAα, PPP3CB/CnAβ, and PPP3CC/CnAγ (Fig 1). Two genes encode CnB regulatory subunits: PPP3R1/CnB1 and PPP3R2/CnB2. Both CnAα and CnAβ are abundant in the heart, with transcript and protein levels of the CnAβ isoform increasing in response to hypertrophic agonists or mechanical stress in a calcineurin-dependent fashion [11, 27, 28]. Mice lacking CnAβ have smaller hearts and a blunted hypertrophic response to pressure overload [29], but sustain greater damage following ischemia reperfusion (I/R) [30]. Less is known regarding the cardiac phenotype of mice lacking CnAα as homozygous KO mice die between 2 and 4 weeks of age [31] and conditional alleles have not yet been studied. As a consequence, the CnAβ isoform is generally considered the primary isoform involved in pathological remodeling of the adult heart, potentially forming a feed-forward amplification loop.

The CnAβ isoform contains a unique, N-terminal proline-rich domain not present on CnAα or CnAγ. There are also isoform variants of CnAβ with distinct regulatory mechanisms. The structure and regulation of the CnAβ2 isoform is typical of other CnA subunits, whereas the CnAβ1 isoform lacks the canonical AID. The unique properties of CnAβ play important roles in substrate targeting and regulation of catalytic activity. These will be discussed later in greater detail.

The CnAγ subunit is often quoted as being testis-specific. However, it is involved in both early fate decisions in stem cells [32] and synaptic vesicle endocytosis in the brain [33]. Furthermore, polymorphisms in CnAγ are associated with responsiveness of the heart to exercise training [34] and post-translational regulation of proBNP [35]. However, the specific molecular events mediated by the CnAγ isoform in the adult heart are as yet unknown.

All of the CnA isoforms are susceptible to cleavage by Ca²⁺-activated proteases, or calpains, which can yield constitutively active forms (CnA*) lacking both the CBD/AIS and AID [36-38]. Various other forms generated by calpain cleavage may retain aspects of Ca²⁺ responsivity. Calpain-dependent cleavage of CnA occurs in cardiomyocytes in vitro in response to angiotensin II (AngII), followed by translocation of the cleaved product into the nucleus [39]. In vivo, CnA* has been detected in failing human myocardium [40] and hearts of patients with atrial fibrillation [41]. The vast majority of calcineurin is cytosolic. The role of nuclear calcineurin is not fully understood, however, it has been shown capable of responding to Ca²⁺-transients released from inositol-3-phosphate receptors (IP₃R) in the nuclear envelope [42]. Interestingly, caspase-3 has also been shown to cleave calcineurin to produce CnA* in immune cells [43] and in models of Alzheimer’s disease [44], raising the possibility that activation of caspase-3 in the heart is another avenue for generating constitutively active calcineurin.

CnB1 is the only calcineurin regulatory subunit expressed in the heart. Because of this, cardiomyocyte-specific deletion of CnB1 has been used to examine the consequences of eliminating calcineurin activity in the myocardium [45, 46]. Mice with a conditional deletion of CnB1 using an αMHC-Cre driver manifest metabolic dysregulation and declines in cardiac function, ultimately dying of arrhythmias around 3 months of age [45, 46]. These studies clearly demonstrate the essential nature of CnB, however, given that the catalytic subunits of calcineurin are still expressed and potentially subject to proteolytic cleavage, it would be wise to use caution extending interpretation of these studies to all calcineurin-dependent processes.

2.2. Unique properties of the CnAβ1 isoform:

The PPP3CB gene produces two different CnA proteins; CnAβ1 and CnAβ2, both of which contain an N-terminal, polyproline domain unique to CnAβ (Fig 1). This domain helps target CnAβ to specific subcellular compartments [47] and assists in the selection of some substrates [48]. The β1 and β2 isoforms have unique C-termini that are generated as a consequence of differences in the site chosen for polyadenylation at the 3’ end of their respective transcripts. The polyadenylation site for the β1 isoform is located upstream of the one used by β2 such that the last two exons in β2 are excluded from β1, which continues translation into the intronic sequence. As a consequence, the CnAβ1 isoform lacks an AID. Instead, the C-terminus of CnAβ1 contains a membrane localization sequence (MLS) that targets CnAβ1 to Golgi membranes through an interaction with COG8 (component of oligomeric goli complex 8) [49]. In addition, CnAβ1 contains an LxVP motif, that blocks access of substrates to one of the substrate -binding domains on calcineurin [50]. The canonical β2 isoform had a canonical AID, but does not carry either the MLS or LxVP motif found in the β1 isoform. Detailed biochemical analysis demonstrates that calcineurin enzyme complexes containing the CnAβ1 subunit have basal activity even in the absence of Ca²⁺/CaM [50]. Although CnAβ1 activity is still increased in response to Ca²⁺/CaM, maximal activity is limited by LxVP-mediated autoinhibition. CnAβ1 is highly expressed in progenitor cells and regenerating tissues, and its expression is induced by insulin-like growth factor 1 [51]. Developmental control of CnAβ isoforms is mediated by Muscleblind Like Splicing Regulator 1 (MBNL1) [49]. CnAβ1 levels are low in adult heart and skeletal muscle but are elevated in mouse models of diabetes [52].

The hearts of mice carrying an αMHC-CnAβ1 transgene are not hypertrophic at base line, demonstrating that the Ca²⁺-independent basal activity of CnAβ1 is not pathological. Remarkably, the αMHC-CnAβ1 transgene reduces pathological remodeling of the heart following myocardial infarction (MI), pressure overload, or ischemia reperfusion (I/R) [53-55]. Mice with targeted deletion of just the LxVP/MLS C-terminal domain unique to CnAβ1 develop age-related pathological hypertrophy and have an exaggerated response to pressure overload suggesting that the CnAβ1 mediates protective functions [55]. Recent studies link the mode of action of CnAβ1 with activation of protein kinase B (Akt) and mechanistic target of rapamycin (mTOR). Targeting of CnAβ1 to Golgi by its MLS domain is required for recruitment of mTORC2 to Golgi and its activation of an Akt/ATF4 cardioprotective, metabolic pathway [49, 53, 55]. Remarkably, a catalytically-dead CnAβ1 mutant was similarly able to activate Akt [51] suggesting that, in this setting, CnAβ1 may be acting as an adaptor scaffold, rather than a phosphatase. Many questions remain regarding this intriguing isoform, its mode of action, its potential substrates, and its relationship to canonical forms of calcineurin.

2.3. Motifs targeting substrates to calcineurin:

The catalytic cleft of calcineurin is shallow, allowing accommodation of a wide range of phospho - serine, -threonine and -tyrosine substrates [56, 57]. As a consequence, there are no conserved recognition motifs surrounding target residues that can be used to predict target sites, although there is a reported preference for sites with a basic residue in the −3 position and lacking an acidic residue immediately adjacent on the C-terminal side [57]. Crystal structures show that key residues in the AID of CnA (⁴⁸¹ERMP⁴⁸⁴) engage the catalytic cleft of calcineurin in such a way that the proline at P484 binds a hydrophobic pocket in the catalytic cleft and E481 mimics a phosphorylated substrate residue thereby blocking the active site [58]. Phospho-substrates with a proline in the +3 position show enhanced affinity for calcineurin over other PPP family phosphatases, which lack calcineurin’s hydrophobic, proline-binding pocket [59].

Instead of a defined phosphosite recognition sequence, the primary determinants for calcineurin substrates are the presence of two short linear motifs (SLIMs), PxIxIT and LxVP, located elsewhere on the protein, that act as recognition motifs to dock and position the substrate protein to facilitate access to the catalytic site of calcineurin [60, 61]. SLIMs are short stretches of protein sequence that mediate protein-protein interactions. Both PxIxIT and LxVP were initially identified as conserved features of the Nuclear Factor of Activated T Cells (NFAT) family of transcription factors, the best understood of calcineurin substrates. The PxIxIT motif binds to a docking domain on CnA, regardless of whether calcineurin is active or inactive [62-64]. Substrates compete with one another for access to calcineurin based on their local concentration and the binding affinity of their specific PxIxIT motif. Over expression of a PxIxIT containing peptide is sufficient to inhibit dephosphorylation of calcineurin substrates in vivo. Over expression of any protein containing a PxIxIT domain will compete with native proteins for docking with calcineurin, because of this, in the literature there are several examples of PxIxIT containing proteins probably erroneously proposed as calcineurin inhibitors based solely on co-immunoprecipitation and over expression studies. By extension, one might predict that cardiomyocyte-specific overexpression of any PxIxIT containing protein would have the potential to suppress calcineurin-dependent hypertrophy regardless of its normal in vivo function.

In contrast to the PxIxIT motif, the LxVP motif binds at the CnA/CnB interface, which is only accessible when calcineurin is in its active conformation [12, 65]. Cocrystal structures of calcineurin with selected substrates demonstrate docking of both motifs, which act together to mediate calcineurin substrate recognition and specificity [59]. Many calcineurin substrates contain both docking motifs, while others appear to carry only one.

2.4. Calcineurin inhibitors:

2.4.1. Pharmacological inhibitors:

The calcineurin inhibitors Cyclosporin A (CsA) and FK506 (or Tacrolimus) are used clinically as immunosuppressives. Each acts by forming a complex with a different class of endogenous proteins called immunophilins: cyclophilins and FK506 binding proteins (FKBPs) respectively [66, 67]. The drug-immunophilin complex engages calcineurin at the docking site for LxVP, thereby preventing docking of substrates [58, 65, 66, 68, 69]. They do not directly block the active site or inhibit calcineurin enzymatic activity. Recombinant calcineurin bound by a CsA/FK506-immunophilin complex is still capable of dephosphorylating small, non-proteinaceous molecules such as p-nitrophenyl phosphate (pNPP), which is often used as a substrate in calcineurin assays [66]. Whether calcineurin might encounter and act on a non-protein, small molecule substrate in vivo is not known, but if so, neither CsA nor FK506 would inhibit this activity.

CsA and FK506 have distinctly different off-target effects based on sequestration of their respective immunophilins. Notably, CsA inhibits opening of the mitochondrial permeability transition pore (MPTP) by binding to cyclophilin D, an essential component of the MPTP [70]. FK506, on the other hand, can impact Ca²⁺ release from RyR2 and IP3R [71, 72], although this occurs at FK506 concentrations well above those required for calcineurin inhibition. Best practice for determining whether a process is calcineurin-dependent is to test outcomes with each drug separately. Sanglifehrin A, which inhibits MPTP but not calcineurin, can also be used as a control for off target effects of CsA [73].

Short peptide inhibitors of calcineurin act by mimicking PxIxIT, LxVP, or AID motifs [74, 75]. A cell permeable version of a PxIxIT peptide was shown capable of inhibiting pressure overload hypertrophy in vivo [10, 76]. Several reviews focused on pharmacological inhibitors of calcineurin are available for further details [77, 78].

2.4.2. Endogenous inhibitors:

The AID of calcineurin itself is the primary endogenous inhibitor of calcineurin activity. Studies of posttranslational modifications that regulate calcineurin activity are limited. One notable exception is the recent finding that in resting T-cells a conserved lysine in the PxIxIT binding pocket of CnA is mono-ubiquitinated with a K-29-linkage, impairing recruitment of NFAT [79]. Upon Ca²⁺ stimulation, the deubiquitinase USP16 rapidly removes ubiquitin from both CnAβ and CnAγ, allowing substrate access. USP16 does not engage CnAα and none of the other over 40 deubiquitinases tested was shown to be capable of deubiquitinating calcineurin. Whether this form of regulation impacts calcineurin activity in the heart awaits further investigation.

2.4.2.1. Regulators of calcineurin (RCAN1, RCAN2, and RCAN3):

RCANs, also known as DSCR1/MCIP1/calcipressin [80], are endogenous proteins that bind and inhibit calcineurin [81-83]. There are three mammalian genes: RCAN1, RCAN2, and RCAN3 [84]. Both RCAN1 and RCAN2 are abundant in the heart. RCAN1 is the most widely studied of the three genes, in part, because expression of the RCAN1.4 isoform is under direct control of calcineurin/NFAT and thus provides feedback inhibition of calcineurin activity [82], but also because of its location on human chromosome 21 and the potential for contributing to Down syndrome pathologies. Recombinant RCAN1 inhibits the ability of calcineurin to dephosphorylate both peptide substrates and pNPP [85, 86], making it distinct from inhibition by CsA or FK506.

RCAN proteins use three independent mechanisms to inhibit calcineurin (Fig. 1 and 2). First, they contain both a PxIxIT and an LxVP motif that compete with substrates for docking to calcineurin. Second, they have a C-terminal TxxP motif that engages and blocks the catalytic site of calcineurin in a way that resembles binding of ⁴⁸¹ERMP⁴⁸⁴ in the calcineurin AID [85-88]. Third, their highly conserved SPPASPP motif is a calcineurin substrate that can function as a pseudosubstrate inhibitor [88, 89]. Various amino acids in RCAN1 have been shown to be phosphorylated in vivo and proposed to modify RCAN1 activity [90-96]. Of these, only phosphorylation of Thr¹⁵³ by the p38 mitogen-activated protein kinase has undergone rigorous biochemical assessment demonstrating that phosphorylation of this site, located adjacent to the PxIxIT motif in RCAN1, reduces the affinity of RCAN1 for calcineurin 30-fold [88].

Figure 2: Model of calcineurin domain interactions.

in the inactive state (A), activated following binding of Ca²⁺ and Ca²⁺/Calmodulin to disengage the AID from the catalytic site (B), NFAT docked with the activated form of calcineurin, small red circles indicate phosphorylated amino acids in the NFAT regulatory domain that are dephosphorylated by calcineurin (C), RCAN1 docked with calcineurin showing the TxxP motif blocking the catalytic site and inhibiting calcineurin activity (D), and docking of immunophilin bound drug complexes (E).

Mice carrying a cardiomyocyte-specific αMHC-Rcan1 transgene have a blunted hypertrophic response to both pathological and physiological stimuli [6]. They are also protected from I/R and pathological remodeling following MI [7, 97]. Damage from I/R is greater in mice with a whole-body knockout of Rcan1 [97, 98]. Paradoxically, these mice have smaller hearts and a blunted hypertrophic response to pressure overload [98, 99]. This has been used to support the proposal that at very low levels RCAN1 acts in some way to facilitate calcineurin signaling, perhaps as a chaperone or targeting protein, however, rigorous biochemical evidence of this is still lacking [100, 101]. A recent review provides in depth discussion of RCAN1 in cardiovascular disease and includes discussion of pathways, in addition to calcineurin suppression, through which RCANs may act [102].

Information regarding the impact of RCAN2 and RCAN3 on cardiac biology is much more limited. The hearts of Rcan1^−/−/Rcan2^−/− double knockout mice are phenotypically similar to those of Rcan1^−/− mice [98]. A recent intriguing study reports the generation of circular RCAN2 transcripts (circ-RCAN2) in cardiac progenitor cells and a significant decrease in circ-RCAN2 levels in pig heart following I/R in vivo [103]. Nothing is known regarding the function or physiological impact of these unique transcripts.

2.4.2.2. TBC1D10C (Carabin/EP164C):

TBC1D10C belongs to the EPI64/TBC1D10 GTPase-activating protein (GAP) family based on the presence of a conserved catalytic Tre2/Bub2/Cdc16 (TBC) at its N-terminus. It was first identified as a negative regulator of T-cell activation via its ability to inhibit both Ras and calcineurin [104]. Pressure overload-induced hypertrophy is enhanced in mice with either whole-body [105]or cardiomyocyte-specific deletion of TBC1D10C [105, 106]. Conversely, forced expression blunts hypertrophic growth and repress the activities of calcineurin, Ras/extracellular signal-regulated kinase (ERK), and CaMKII after either cardiac pressure overload or adrenergic stimulation. TBC1D10C levels are reduced in animal models of heart failure and in patients with either dilated or hypertrophic cardiomyopathy [106], suggesting that a reduction in TBC1D10C levels could contribute to severity of disease. The C-terminus of EPI64C binds to calcineurin near to where PxIxIT motifs dock [106] and therefore may act by competing with substrates for binding at this site. It is worth noting that the GTPase activity of TBC1D10C is specific for Rab35 and has been shown to impair exosome release from oligodendrocytes [107]. Given emerging evidence that Rab35 is involved in the release of extracellular vesicles from the heart in response to exercise [108], defining the interaction of this process with calcineurin-mediated signaling will be an important avenue for future research.

2.5. Methods for assessing calcineurin activity

Quantitative measures of in vivo calcineurin activity are challenging due to the dynamic nature of Ca²⁺-mediated activation. Biochemical assays of tissue extracts are carried out in the presence of excess Ca²⁺ and CaM [109]. Because of this they only assess the total amount of calcineurin in the extract, not what fraction of it was active in vivo. Co-immunoprecipitation of calcineurin with CaM is similarly limited as the outcome is entirely dependent on the level of Ca²⁺ present during extraction and washing steps. Transcript and protein levels of RCAN1.4 have proven to be one of the most reliable indications of calcineurin activity and have been used in a wide range of tissues. However, it is important to note that changes in RCAN1.4 are dependent on NFAT-mediated transcription, and therefore may not accurately reflect calcineurin activity directed toward other substrates.

Fluorescent-tagged NFAT proteins and NFAT-driven luciferase reporters have proven useful but are likewise directly tied to an NFAT response [110-113]. Genetically-encoded reporters, such as CaNAR, that use fluorescence resonance energy transfer (FRET) as a dynamic read-out of calcineurin activity, can be targeted to specific subcellular domains [20]. When combined with a Ca²⁺ sensor, these have been used to demonstrate that activation of calcineurin is impacted by differences in CaM availability and local rates of rephosphorylation even when the magnitude of the Ca²⁺ transient is the same, [20]. The CaNAR reporter is also based on NFAT and therefore still an indirect, substrate-specific assessment of calcineurin activity.

The FRET-based reporters DuoCaN and UniCaN provide a more direct readout of calcineurin activation [114]. Elegant experiments using these reporters reveal fundamental differences in the localization and activation of calcineurin in neonatal verses adult cardiomyocytes. In neonatal cardiomyocytes they distribute homogeneously throughout the cell and respond directly to individual Ca²⁺ pulses, achieving sustained activity at higher stimulation frequencies. In adult cardiomyocytes the reporters localize to T-tubules, consistent with the pattern described for endogenous calcineurin [115, 116]. Here, they are unresponsive to either single Ca²⁺ transients or low frequency pacing, activating only at pacing frequencies sufficient to raise diastolic Ca²⁺ levels. Therefore, studies from neonatal cardiomyocytes may not always be directly applicable to calcineurin signaling in the adult heart. During failure, adult cardiomyocytes develop a number of characteristics typical of immature cardiomyocytes and this may include aspects of calcineurin activation. It is important to be aware that adult cardiomyocytes isolated from mouse begin to dedifferentiate rapidly in culture, making the time required for adenoviral infection often problematic, although approaches have been developed that may help slow the rate of dedifferentiation [117]. Finally, the DuoCaN and UniCaN FRET reporters are based on CnAα and therefore may not accurately reflect properties unique to CnAβ or CnAγ.

3. Calcineurin substrates relevant to cardiac function and disease (Fig 3):

Figure 3: Schematic of calcineurin (CN) functional and physical distribution in adult cardiomyocytes.

AKAPs and other proteins that act as calcineurin scaffolds are depicted in green. Interactions that are isoform-specific are indicated. Direct and indirect calcineurin substrates are depicted in blue. Calcineurin inhibitors are yellow. Note that CN is primarily tethered near known sites of Ca²⁺ release. A kinase anchor proteins (AKAP), β-adrenergic receptor (β-AR), BCL2 associated agonist of cell death (BAD), calcineurin (CN), carabin/TBC1D10C (Carabin), cardiomyopathy associated 5 (CMYA5), component of oligomeric goli complex 8 (COG8), connexin 43 (Cx43), dynamin related protein 1 (DRP1), inhibitor 1 (I1), LIM and cysteine-rich domains 1 (LMCD1), L-type calcium channel (LTCC), mammalian target of rapamycin complex 2 (mTROC2), muscle LIM protein (MLP), Myozenin 2 (MYOZ2), nuclear factor of activated T cells (NFAT), nuclear pore complex (NCP), protein phosphatase 1 (PP1), regulator of calcineurin 1 (RCAN1.4), ryanodine receptors (RyR), sodium-calcium exchanger (NCX1), sodium-potassium exchanger (NHE1), thyroid hormone receptor interactor 10 (TRIP10), transient receptor potential canonical (TRPC).

Calcineurin signaling mediates both immediate and long-term responses. Rapid responses may involve changes in protein activity, subcellular location, and/or protein-protein interactions, many of which could be rapidly reversed by rephosphorylation. In contrast, calcineurin-dependent changes in transcription have the potential for mediating longer-lasting consequences.

3.1. Key transcription factors controlled by calcineurin:

3.1.1. NFAT:

Four of the five known NFATs are activated by calcineurin; NFATc1 (a.k.a. NFAT2), NFATc2 (a.k.a. NFAT1), NFATc3 (a.k.a. NFAT4) and NFATc4 (a.k.a. NFAT3). Dephosphorylation by calcineurin causes translocation of NFAT to the nucleus where it acts in conjunction with a diversity of transcription factors, thereby facilitating the generation of a specific transcriptional response to a more generalized signal initiated by a change in sytosolic Ca²⁺. Both the MEF2 [118] and GATA [119-121] family of transcription factors are activated by calcineurin, work cooperatively with NFAT, and play prominent roles in cardiac gene expression [122-124]. A direct interaction between NFAT and NFκB during hypertrophy has also been reported [125]. Deletion of either NFATc2 or NFATc3 blunts hypertrophic remodeling [126, 127], although the two proteins show clear differences in the dynamics of their response to calcineurin [128-131]. Despite the prominence of NFAT-dependent transcription in the cardiac literature, the identity of the most relevant target genes remains a critical question.

3.1.2. CRTC:

Similar to NFATs, dephosphorylation of cAMP response element binding protein (CREB)-regulated transcriptional coactivators (CRTCs) by calcineurin promotes their translocation to the nucleus where they act in conjunction with CREB to drive gene expression. In other tissues CRTCs have been shown to mediate metabolic responses to hormonal stimuli [132-134] and promote mitochondrial biogenesis [135-137]. β-adrenergic stimulation of neonatal cardiomyocytes increases phosphorylation of CRTC1, and this is further enhanced with pharmacological inhibition of calcineurin [138]. The hearts of mice lacking CRTC1 are hypertrophic, suggesting that CRTC1 activity may mediate compensatory mechanisms that delay disease progression [138].

3.1.3. FOXO1:

The forkhead box protein family of transcription factors mediate metabolic remodeling of the heart in diabetic cardiomyopathy and post-ischemic heart failure [139-141]. Dephosphorylation of FOXO1 by calcineurin triggers its translocation to the nucleus, where it can cooperate with NFAT to drive gene expression [142]. Sustained FOXO1 activity provides feedback inhibition of calcineurin activity in the heart [143-145], suggesting a close interdependence of these two pathways.

3.1.4. TFEB:

Transcription Factor EB (TFEB) is a master regulator of lysosomal biogenesis and autophagy [146, 147]. Calcineurin dephosphorylates TFEB triggering its translocation to the nucleus and activation of target genes [148]. Proteotoxic stress in either the ER or cytosol can lead to calcineurin-dependent activation of TFEB [149, 150]. Thus, calcineurin control over lysosomal biogenesis via TFEB may help coordinate degradative and growth processes during structural and metabolic remodeling of the myocardium [151, 152].

3.2. Integrated regulation with sodium exchangers NHE1 and NCX1.

Calcineurin participates in regulation of the sodium-potassium (NHE1) and the sodium-calcium (NCX1) exchangers as well as being activated downstream of their joint activities. The cytoplasmic tail of NHE1 contains both a PxIxIT and LxVP motif for calcineurin docking [153, 154]. NHE1 activity is increased by phosphorylation at Thr779 and calcineurin dephosphorylates this site, thereby reducing exchanger activity [153]. Calcineurin is also reported to inhibit NCX1, although the specific mechanism of action is not yet defined [155, 156]. In the adult myocardium NHE1 and NCX1 can be found colocalized in caveolae-positive lipid rafts [156-158] providing a microdomain platform through which NHE1/NCX1 activity can activate calcineurin/NFAT to promote hypertrophic growth [154, 159] and calcineurin in turn can regulate exchanger activity. Activation of NHE1 causes a sustained increase in cytosolic Ca²⁺ to activate calcineurin by either decreasing NCX1 activity or increasing its reverse mode of action [160]. During ischemia, in response to a build-up of lactic acid, the combined actions of NHE1 and reverse-mode NCX1 lead to cytosolic Ca²⁺ overload, however, calcineurin remains inactive under ischemic conditions due to the acidic environment. Upon reperfusion, physiological pH is restored allowing calcineurin activity to then reduce the activities of NHE1 and NCX1, as well as acting on other potential substrates.

3.3. Connexin 43 (Cx43) and gap junction conductance:

The phosphorylation status of Cx43 regulates cardiac gap junction conductance and propagation of the action potential (AP) between adjacent cardiomyocytes [161, 162]. A reduction in the speed at which APs are propagated (conduction velocity) can lead to re-entrant arrhythmias. In a healthy heart Cx43 is phosphorylated at Ser365 and localized to intercalated disks. Calcineurin inhibitors prevent dephosphorylation of Cx43 during ischemia in both cardiomyocytes and astrocytes [163, 164]. In this context calcineurin acts indirectly by dephosphorylating inhibitor 1 (I1) thereby releasing its inhibition of PP1 [165]. PP1 then dephosphorylates Ser365 of Cx43, causing disassembly of gap junction channels and promoting redistribution Cx43 to lateral membranes [161, 162, 166].

3.4. Potential mechanism for cross-talk with JNK:

The dual specificity mitogen-activated protein kinase kinase 7 (MKK7) in the c-Jun NH(2)-terminal kinase (JNK) pathway has been implicated in both hypertrophy and arrhythmogenesis [167-169]. Calcineurin specifically suppress MKK7γ activity via a PxIxIT motif specific to the γ isoform but not present on the MKK7α and β splice variants [170]. The distribution of individual MKK7 isoforms in the heart has not been reported, however, it is possible that this mechanism may in part mediate previously reported crosstalk between calcineurin and JNK during cardiac remodeling [171, 172].

3.5. Identifying new calcineurin substrates using SLIMs:

Recently, two groups undertook systematic approaches to identify new calcineurin substrates that both expand our knowledge of calcineurin-mediated processes and provide data-base resources for investigators to query potential new targets [173] http://slim.icr.ac.uk/motifs/calcineurin/. These groups combined in silico proteome-wide identification of potential PxIxIT and LxVP motifs with a variety of in vitro and in vivo affinity capture methods to identify a diversity of new calcineurin substrates, increasing the known calcineurin interactome by an order of magnitude in both cardiomyocytes and neurons [173, 174]. A key finding from both of these groups was that multiple components of the nuclear pore complex (NCP) are calcineurin substrates. In addition to gating nucleocytoplasmic transport, the NCP is involved in diverse functions including chromatin organization, regulation of gene expression and DNA repair [175]. In humans, it is a massive complex of ~125 megaDaltons and composed of multiple (8-64) copies of 30 different nuclear pore proteins called nucleoporins (NUPs). Several NUP proteins were verified as calcineurin substrates and pharmacological inhibition of calcineurin was shown to alter nuclear import of model proteins. Thus, calcineurin may influence nuclear translocation of a much broader range of proteins and transcription factors than previously thought. Other key proteins verified as calcineurin substrates include NOTCH1 and MDM2, although the functional consequences of calcineurin-mediated modification of these proteins remain to be explored. The extensive supplemental data linked to these three manuscripts are a valuable resource for investigators seeking to determine whether their protein of interest may be modified by calcineurin [173, 174, 176].

4. Targeting of calcineurin to subcellular domains:

Immunohistochemistry indicates that in adult cardiomyocytes the majority of calcineurin is localized to the T-tubules, which are deep invaginations of the sarcolemma (or plasma membrane) that bring voltage-activated LTCC in the plasma membrane into close association with RyR2 in the sarcoplasmic reticulum (SR). Ca²⁺-induced Ca²⁺-release in this microdomain links depolarization of the plasma membrane with activation of cardiomyocyte contraction (excitation-contraction coupling). In addition to LTCC, a diversity of membrane channels and receptors are also found at T-tubules. Thus, calcineurin is found in close proximity to a wide array of proteins involved in receptor-mediated signaling, trafficking of ions, and mechanotransduction. Although there is also evidence for calcineurin, either free or tethered, responding to specific Ca²⁺ microdomains elsewhere in cardiomyocytes.

4.1. A kinase anchor proteins (AKAPs):

AKAPs are scaffolding proteins that form multimolecular signaling complexes that by definition include PKA. Many, but not all, also bind calcineurin, facilitating reversal of specific PKA-mediated phosphorylation events by calcineurin. More than 30 AKAPs are expressed in the heart [177], however, direct interactions with calcineurin have been described for only a handful of these, and include; AKAP5 (AKAP79/AKAP150), at the plasma membrane of T-tubles; AKAP6 (AKAP6β/mAKAPβ), at the nuclear envelop; AKAP1 (AKAP121/AKAP84), at the outer mitochondrial membrane; and Cypher/ZASP (LIM domain binding 3, LDB3), at the plasma membrane associated with LTCCs. Here we will describe AKAP5 in depth to illustrate the challenges of understanding the mechanism and regulation of calcineurin-AKAP interactions; what is known and what is not known regarding this fundamental aspect of calcineurin localization and substrate targeting. For additional details we recommend several comprehensive reviews covering AKAP structure, function, and cardiovascular phenotypes [113, 178-181].

4.1. AKAP5 (AKAP79/AKAP150)

AKAP5 was the first AKAP shown to bind calcineurin [182]. In addition to acting as a scaffold for PKA and calcineurin signaling, AKAP5 binds adenylyl cyclases (AC), phosphodiesterase (PDE4D), and protein kinase C (PKC) [178, 183, 184]. It forms complexes with numerous plasma membrane channels and receptors, including LTCC, transient receptor potential channels (TRPV1 and 4)[185, 186], N-methyl-D-aspartic acid receptors (NMDAR), α-amino-3-hydroxy-5- methyl-4-isoxazoleproprionic acid receptors (AMPAR), β-adrenergic receptors 1 and 2 (β-AR) and voltage-gated potassium channels including A-type (Kv4.2) [187-189] and M-type (KCNQ2-5) [190]. Notably, AKAP5 doesn’t directly bind KCNQ1, mutation of which causes Long Qt Syndrome [191]. Instead, KCNQ1 associates with the Yotiano form of AKAP9, which is not known to interact directly with calcineurin [192, 193] but may interact indirectly through TRIP10 (see later section for details). AKAP5 complexes also contain postsynaptic density protein 95 (PSD-95) and synapse-associated protein 97 (SAP97), members of the membrane-associated guanylate kinase (MAGUK) family that are involved in trafficking of transmembrane receptors [194]. Caveolin 3 (CAV3), enriched in striated muscle, is also an integral component of many AKAP5 complexes in cardiomyocytes [184]. The specific composition of a AKAP5/calcineurin complex depends on cell type, subcellular location, and physiological state. Studies in neurons have yielded important insights into the mechanism of action and physiological function of these complexes, some of which may, or may not, be directly transferable to signaling in the heart.

4.1.1. Engagement of AKAP5 with calcineurin:

AKAP5 binds calcineurin using a PxIxIT motif [195]. This leaves the LxVP docking site of calcineurin still available to engage substrates with LxVP motifs. This is the case for LTCC, which has an LxVP but no PxIxIT motif. A direct interaction between LTCC and AKAP5 helps hold LTCC in proximity of calcineurin (and PKA) [196, 197] and mediates Ca²⁺-dependent inactivation of LTCC in neurons [198]. Similarly, NFAT also interacts directly with AKAP5 through a C-terminal leucine zipper (LZ) domain that is required for NFAT activation in response to membrane depolarization in neurons [199]. Thus, AKAPs provide a scaffold to maintain both calcineurin and its substrates in a specific subcellular domain, often close to a site of Ca²⁺ entry. Variability in the binding affinities of PxIxIT motifs plays an important regulatory role in substrate selection and release from the complex. Replacing the PxIxIT motif on AKAP5 with a higher affinity site increases recruitment of calcineurin to AKAP5 but impairs NFAT activation [195]. Conversely, too weak of a PxIxIT domain on AKAP5 also impairs NFAT activation, potentially by losing tethering of calcineurin near a site of Ca²⁺ release. Recently, AKAP5 has been shown to also contain an LxVP motif [200]. This may help explain how isolated AKAP5 complexes contain four copies of calcineurin per AKAP5 dimer [201]. Taken as a whole, the AKAP5 complex orchestrates the remarkable feat of facilitating calcineurin as a positive transducer of LTCC Ca²⁺ signaling to NFAT as well as providing feedback inhibition of LTCC through calcineurin-mediated reversal of PKA-mediated phosphorylation of LTCC.

4.1.2. AKAP5/calcineurin in adrenergic signaling:

AKAP5/CAV3 complexs link a subset of LTCC with β-ARs, thereby providing a platform for adrenergic enhancement of LTCC Ca²⁺ release, activation of NFAT-dependent transcriptional responses, and modification of β-AR activity by calcineurin. Neonatal cardiomyocytes lacking AKAP5, or expressing an AKAP5 lacking a PxIxIT motif, do not activate NFAT3 in response to an adrenergic stimulus [202]. Whether AKAP5 is specific for NFAT or can act as a platform for other calcineurin-responsive transcription factors is not known. There is evidence that only AKAP5 complexes associated with CAV3 are able to mobilize NFAT in response to an adrenergic stimulus, however, this remains controversial [203, 204]. In heart failure and ischemic heart disease the structure of the T-tubule system degrades, channel microdomains are remodeled, and spontaneous Ca²⁺ release events increase [205-210]. As a consequence, the nature of NFAT mobilization at LTCC/AKAP5/calcineurin complexes is likely very different in a failing heart than in a healthy one.

In adult cardiomyocytes lacking AKAP5, β-AR agonists do not increase either the magnitude of Ca²⁺ transients or the phosphorylation of LTCC, RyR2 and phospholamban [184, 211]. Only the pool of LTCC associated with CAV3 containing AKAP5 complexes responds to β-adrenergic stimulation in the heart [184]. In neurons, AKAP5-bound calcineurin inactivates neuronal LTCC following adrenergic activation [197, 198], however, whether calcineurin does this in cardiomyocytes has not yet been rigorously tested.

Persistent activation of β-ARs causes desensitization and recycling, which involves internalization of the receptor into endosomes followed by degradation or return to the plasma membrane. Recycling of β₁-AR, the primary cardiac β-AR subtype [212], is dependent on AKAP5 and has been shown to require binding of both PKA and calcineurin [213]. SAP97 targets AKAP5 to the C-terminal tail of β₁-AR, facilitating phosphorylation at Ser312, which is required for receptor cycling [214, 215]. Calcineurin promotes return of the receptor to the plasma membrane and may act by dephosphorylating this same site, but this remains to be definitively demonstrated. Treating cardiomyocytes with FK506 to inhibit calcineurin has no impact on internalization of either β₁-AR or β₂-AR, however, it prevents recycling of β₁-AR, but not β₂-AR, back to the cell membrane following the removal of agonist [216], thus, calcineurin appears to be involved in trafficking of β₁-AR but not β₂-AR in cardiomyocytes. It is relevant to note that, in a healthy heart, β₁-ARs are distributed throughout the sarcolemma, whereas, β₂-ARs are restricted to caveolae and found in association with CAV3 and LTCC [217-219]. AKAP5/calcineurin/CAV3 complexes [220] are required for β₂-AR activation of the LTCC [184, 217] and mediate β₂-AR activation of MAPK kinase [221].

4.1.3. AKAP5/calcineurin remodeling of potassium channels:

In pathological hypertrophy and failure the action potential is prolonged, in part, due to a reduction in repolarizing K⁺ currents, including the fast transient outward K+ current (I_to) carried by K_v4 channels [222]. In mice lacking AKAP5 I_to is not decreased following MI [202]. In both neurons and reconstituted Cos7 cells, AKAP5-anchored PKA promotes phosphorylation and internalization of K_v4.2, reducing current density. Conversely, AKAP5-anchored calcineurin promotes trafficking back to the surface [187]. If similar mechanisms control surface expression of K_v4-type channels in cardiomyocytes, it would suggest that during heart failure, PKA activity dominates AKAP5-dependent reductions in I_to, despite concurrent activation of AKAP5-achored calcineurin. I_to is increased in neonatal rat ventricular myocytes (NRVMs) overexpressing CnA* [223], consistent with calcineurin facilitating K⁺ channel activity. β-adrenergic-stimulation of NRVMs reduces I_to which is further reduced by the addition of a calcineurin inhibitor, also consistent with a positive effect for calcineurin [224]. Finally, I_to density is elevated in young αMHC-CnA* mice prior to a decline in cardiac function [223]. This suggests that, although I_to is reduced in older αMHC-CnA* mice [225], this is likely secondary to heart failure rather than a direct consequence of calcineurin’s impact on K⁺ channel activity.

NFAT-dependent repression of K⁺ channel gene expression has been proposed as a mechanism through which calcineurin activity could reduce I_to [202, 226, 227], and indeed, I_to is not reduced following MI in mice lacking NFATc3, similar to Akap5 KO mice [227]. However, the fact that calcineurin activity increases K⁺ channel transcript levels in NRVMs and in αMHC-CnA* transgenic mice prior to heart failure argues against transcriptional repression as a mechanism of action [223]. Thus, the specifics of a potential NFATc3-dependent mechanism remain to be deciphered.

4.1.4. AKAP5/PKA/calcineurin in vascular smooth muscle cells:

An AKAP5/PKA/calcineurin signaling complex also mediates fundamental signaling responses in vascular smooth muscle cells, interfacing with specific pools of LTCC, K⁺ channels, TRPV4 channels, purinergic receptors, and β₁-ARs (reviewed in [228, 229]). Many of these interactions are relevant to control of vascular tone and mice lacking AKAP5 are hypotensive [230].

Finally, it is important to note that deletion of AKAP5 in mice causes hypertrophy, pathological remodeling, and increased calcineurin/NFAT activity [211, 216, 231]. This would suggest that an association between calcineurin and AKAP5 is not required to mediate hypertrophic remodeling and may instead be cardioprotective by restraining pathological calcineurin signaling [213].

4.2. Other cardiac AKAPs anchoring calcineurin:

4.2.1. AKAP6 (AKAP6β/mAKAPβ),

located at the nuclear envelope, provides a platform for integrating calcineurin signaling with a diversity of proteins that mediate transcriptional responses to cardiac stress. Ca²⁺ released from a select pool of perinuclear RyRs provides localized activation of calcineurin and subsequent stimulation of NFAT and MEF2 dependent gene expression [232-234]. AKAP6 is required for agonist induced nuclear translocation of NFAT in neonatal cardiomyocytes [232, 235], however, this has also been demonstrated for AKAP5 [202, 203]. Whether the two AKAP complexes work in conjunction with one another or mobilize distinct pools of NFAT is not known. There is a precedent for AKAP cooperativity as AKAP6 and AKAP9 have been shown to work in conjunction to tether Golgi to the nucleus [236]. Mice with a cardiomyocyte-specific deletion of AKAP6 are resistant to both pressure overload and agonist induced heart failure [234]. Extensive details of the many signaling processes mediated through AKAP6 and their relevance to cardiac remodeling can be found in an excellent recent review [237].

4.2.2. AKAP1 (AKAP121/AKAP84),

located at the outer mitochondrial membrane (OMM), acts as a coordinating hub for mitochondrial dynamics driven by changes in phosphorylation of the Dynamin Related Protein 1 (DRP1/DNM1L). AKAP1-bound DRP1 is phosphorylated by PKA [238-240]. Dephosphorylated of DRP1 by calcineurin allows it to engage receptors on the OMM to initiate mitochondrial fission [241, 242]. Remarkably, preventing mitochondria fission blocks hypertrophic growth in both neonatal and adult cardiomyocytes [243, 244] suggesting that DRP1 is an essential calcineurin substrate in the hypertrophic response. Loss of AKAP1 increases dephosphorylation of Drp1 and mitochondrial fission [245]. Total loss of AKAP1, or disruption of its ability to bind mitochondria, increases hypertrophic growth in neonatal and adult cardiomyocytes [246-248]. PTEN-induced kinase (PINK1), which is stabilized upon loss of mitochondria membrane potential, reduces PKA’s affinity for AKAP1 [249], thereby increasing calcineurin/DRP1-mediated mitochondrial fission and providing a mechanism for coupling PINK1-mediated mitophagy with mitochondrial fission.

Calcineurin can activate intrinsic cell death pathways by dephosphorylating the pro-apoptotic protein BAD, sending it to the OMM where it promotes MPTP opening [250-252]. AKAP1 coordinates the opposing rephosphorylation of BAD by PKA [253]. In addition to coordinating PKA and calcineurin activities, AKAP1 contains an RNA binding domain that binds mRNAs encoding electron transport chain and TCA cycle proteins, thereby gating local translation of mitochondrial proteins at the OMM [254-256]. Many important questions remain unanswered, such as; Is calcineurin involved in AKAP1-dependent control over mitochondrial protein translation and import? Can Ca²⁺ released from mitochondria during MPTP opening activate calcineurin localized nearby? and Does AKAP1 act as a platform for mobilizing other calcineurin substrates?

4.2.3. LDB3 (LIM Binding Domain 3/Cypher/Zasp)

is a calcineurin binding AKAP, specific to striated muscle and required for sarcomere integrity [257, 258]. Mutations in LDB3 cause dilated heart failure in humans [259] and cardiomyocyte-specific KO of Ldb3 causes severe dilated failure [260]. In adult cardiomyocytes lacking LDB3, protein levels of LTCC and calcineurin are significantly elevated, yet, both basal and isoproterenol-stimulated LTCC currents are reduced [261], suggesting that LDB3 is involved in β-adrenergic regulation of LTCC. Calcineurin’s role in the LDB3 complex is yet to be fully defined.

4.2.4. CMYA5 (cardiomyopathy associated 5/Myospryn)

is an extremely large, 500 kDa calcineurin-binding AKAP found in association with the intermediate filament protein desmin [262, 263]. In adult cardiomyocytes CMYA5 co-localizes with RyR2 at junctional SR and when it is expressed in heterologous cells it promotes clustering of RyR2 [264]. CMYA5 polymorphisms are associated with increased cardiac wall thickness in hypertension [265] and mutations are associated with sporadic hypertrophic cardiomyopathy [266]. Little is known regarding calcineurin’s role in CMYA5 functions. It is intriguing to note that calpain-3 also binds CMYA5 in the same general region as calcineurin [263, 267], making it a potential platform for calpain cleavage of calcineurin to produce the unregulated, constitutively active CnA* form.

4.3. Calcineurin’s impact on PP1 via I1:

Calcineurin can indirectly increase the activity of other AKAP-associated phosphatases through its control of I1. Phosphorylation of I1 by PKA causes selective inhibition of PP1 [268], which is released when calcineurin dephosphorylates I1 [165]. Key cardiac AKAPs containing PP1, include AKAP7 (AKAP15/AKAP18)[269] and Yotiao[270].

4.4. Calcineurin’s impact on AKAP/PKA via RII:

The RII regulatory subunit of PKA contains an LxVP motif and was one of the first identified calcineurin substrates [65]. AKAPs have binding sites for either RI or RII regulatory subunits which act to tether the catalytic subunits of PKA to the complex in an inactive state. Binding of cAMP to the R subunits releases the now active catalytic subunit from the complex. Phosphorylation of RII at Ser-96 can both increase its affinity for AKAP [271] and decrease its affinity for the PKA catalytic subunits [272, 273]. Dissociation of the catalytic subunit allows phosphatases access to RII. Thus, in AKAP complexes containing both RII and calcineurin, the phosphatase is poised to facilitate reassociation of RII with the catalytic subunit and terminate PKA activity [274]. The abundance and phosphorylation level of RII is decreased in the hearts of patients with dilated cardiomyopathy [271].

4.5. Additional calcineurin scaffolds:

TRIP10 (thyroid hormone receptor interactor 10), also called CIP4 (CDC42 interacting protein 4) binds the polyproline domain unique to CnAβ, targeting it to the sarcolemma [47, 275]. TRIP10 is an adaptor protein involved in coordinating both actin polymerization and membrane curvature and is required for hypertrophic growth of neonatal cardiomyocytes [276]. Cardiomyocyte-specific deletion of TRIP10, or peptides designed to prevent binding of CnAβ to TRIP10, blunts pathological remodeling in response to pressure overload. Remarkably, this pool of calcineurin responds robustly to pacing but only minimally to neurohormonal stimulation [47]. TRIP10 is phosphorylated by AKAP9/PKA complexes localized to Golgi [277, 278], thereby indirectly associating calcineurin with AKAP9. Whether this also occurs in the Yotiao form of AKAP9 [192, 279] and if it is involved in the role of Golgi in cardiovascular disease [280] remain to be investigated.

5. Calcineurin responses to mechanical stress

5.1. Mechanical stress at the plasma membrane:

The concept of mechanosensitive Ca²⁺ microdomains is central to our understanding of how mechanical stress is translated to actionable signaling events in the heart [281-283]. Less well understood is how, where, and when calcineurin engages with this system. Stress receptors at the plasma membrane include transient receptor potential canonical (TRPC) channels composed of homo or heterotetramers of either TRPC1/4/5 (stretch activated) or TRPC3/6/7 (activated by diacylglycerol downstream of G protein coupled receptors) [284, 285]. Inhibition of either subtype blunts hypertrophy and failure in response to sustained agonist or pressure overload [286]. Activation of calcineurin/NFAT under conditions of pressure overload increases TRPC6 expression creating feed-forward amplification of hypertrophic signaling [287].

Polycystin-1 (PC-1), another member of the TRPC family of Ca2+ transporters is a calcineurin/NFAT coupled mechanosensor [288]. Pressure overload hypertrophy is blunted in mice lacking PC-1 [289]. TRPC-dependent stretch activation of calcineurin/NFAT also occurs in myofibroblasts [290] promoting production of extracellular matrix and differentiation into activated myofibroblasts, thereby increasing myocardial stiffness [291-293]. Syndacan-4, a transmembrane heparin sulfate, acts as an essential coordinating center for TRPC6/calcineurin-mediated activation of myofibroblasts [294]. In atrial myocytes calcineurin activity is also required to increase matrix metalloproteases in response to static stretch [295]

5.2. Mechanical stress at the Z-disk:

Muscle LIM protein (MLP/CSRP3) tethers calcineurin to the Z-disk and is required for both stretch and adrenergic activation of NFAT [296]. Glutaredoxin 3 (GLRX3/PICOT), a positive regulator of DNA damage repair [297], disrupts calcineurin’s interaction with MLP [298]. Overexpression of GLRX3/PICOT blunts pressure overload-induced hypertrophy, whereas mice with reduced GLRX3/PICOT have an exacerbated response to pressure overload [299]. The Z-band protein Myozenin 2 (MYOZ2/calsarcin) also interacts with calcineurin [116, 300]. Loss of MYOZ2 increases activation of calcineurin by mechanical stress but has no effect on adrenergic activation of calcineurin. LMCD1 (LIM and cysteine-rich domains 1/dyxin), is another Z-disc protein that signals through the calcineurin pathway [301]. Over expressing LMCD1 increases calcineurin activity in response to pressure overload and accentuates pathological remodeling in vivo, whereas depletion of LMCD1 blunts NFAT activation in vitro [302].

Finally, the muscle-specific RING finger 1 (MuRF1), an E3 ubiquitin ligase localized to the Z-disk, targets CnA for degradation via the proteasome [303]. Mice lacking MuRF1 have increased fibrosis and hypertrophy in response to pressure overload that can be normalized by inhibiting calcineurin.

6. Calcineurin signaling during development and regeneration.

Given the importance of Ca²⁺ signals during development, it is not surprising that calcineurin is activated at diverse steps over the course of differentiation from a single cell to a fully mature adult heart.

6.1. Early commitment:

Calcineurin signaling is involved at several steps during the progression from embryonic stem cells (ESCs) to cardiac progenitors. During early differentiation of ESCs calcineurin/NFATc3 promotes epithelial-to-mesenchymal transition in both mice and humans [32].[32] Knock down of CnAγ, the gamma isoform of the catalytic subunit, increases the capacity of ESCs for self-renewal, suggesting that a calcineurin-dependent signal is required to exit pluripotency. Multipotent, mesodermal Flk1⁺ (VEGF receptor 2) cells subsequently become committed to either hematopoietic or cardiac progenitor lineages. Calcineurin/NFATc3 again plays a role in this cell fate discission by driving expression of Etv2 (ETS Variant Transcription Factor 2), which promotes differentiation toward hematopoiesis away from a cardiac lineage [304]. Thus, calcineurin provides both positive and negative signals along the progression from ESC to committed cardiomyocyte depending on the developmental step.

6.2. Cardiac valve morphogenesis:

Calcineurin/NFAT signaling is involved in at least three steps during cardiac valve formation. Endocardial to mesenchymal transformation (EMT) gives rise to valve mesenchyme during cushion formation. Endocardial cells with high levels of NFATc1 remain as endocardium and then proliferate to form elongated valve leaflets. Severe valve defects occur in mice lacking Nfatc1 [305, 306]. Lineage tracing studies suggest that Nfatc1 suppresses EMT by repressing expression of the Snail family of transcriptional repressors, SNAI1 and SNAI2, key factors required for EMT [307]. In endothelial cells where NFATc1 remains high, VEGF activates calcineurin/NFATc1 to work in conjunction with MEK1-ERK1/2 and promote endothelial cell proliferation. Later in maturation, RANKL (receptor activator of NFκB ligand) activates calcineurin/NFATc1 to work in conjunction with JNK1/2 signaling to activate expression of cathepsin K, an enzyme involved in extracellular matrix remodeling [308].

6.3. CRELD1 and atrioventricular septal defects (AVSDs):

Mutations in Cysteine-Rich with EGF-Like Domains 1 (CRELD1) have been linked to AVSDs in humans [309-311]. CRELD1 forms a complex with the CnB subunit of calcineurin at the ER and forced expression of CRELD1 is sufficient to activate NFATc1 [312]. CRELD1 is required for VEGF-dependent proliferation of endocardial cells and expression of NFATc1 target genes at E10.5 [312]. Cardiomyocyte-specific deletion of Creld1 during development causes hypoplasia and trabeculation defects caused by altered NOTCH1 signaling [313]. Increased calcineurin and NOTCH1 activity have been linked to cardiac hypoplasia caused by inhibition of mitochondrial fusion in cardiomyocytes during development [314]. It is relevant to note that NOTCH1 was recently validated as a calcineurin substrate [173]. Therefore, understanding the mechanism of CRELD1/calcineurin/NOTCH1 function may yield new insights into AVSD, one of the most common congenital heart defects.

6.4. Calcineurin signaling is different in neonatal cardiomyocytes compared to adult cardiomyocytes:

There are significant structural, metabolic, and functional differences between neonatal and adult cardiomyocytes, many of which impact the character and duration of the Ca²⁺ signal encountered by calcineurin [315-317]. In embryonic cardiomyocytes Ca²⁺ is released primarily from the perinuclear SR through the cooperative action of IP3R and RyR2. Mature T-tubules and dyads are not established until many days after birth. As a consequence, there is a slower time to peak and decay of both the action potential and the Ca²⁺ transient in neonatal cardiomyocytes compared to adult. The capacity for buffering cytosolic Ca²⁺ also increases dramatically after birth, thereby increasing the potential for establishing confined Ca²⁺ microdomains where specific pools of calcineurin may be shielded. The shift in calcineurin localization from diffuse, cytosolic in neonatal cardiomyocytes to a T-tubule/Z-disk pattern adult cells is coincident with the postnatal development of T-tubules, but probably also depends on changes in expression of specific calcineurin isoforms such as CnAβ with its unique N-terminal polyproline domain. In general, the changes that occur during maturation increase the total potential capacity for calcineurin activity while establishing micro-environments that restrain activation and protect against uncontrolled signaling. During failure, we postulate that loss of the T-tubule system releases these domain constraints on calcineurin contributing to a pathological feed-forward signaling cascade.

6.5. Vascularization of the myocardium:

Calcineurin/NFATC1 signaling also contributes to formation of the cardiac vascular system. During embryogenesis RANKL activation of calcineurin/Nfatc1 in the epicardium drives expression of cathepsin K to degrade extracellular matrix, thereby facilitating migration of epicardial cells into the myocardium where they contribute to formation of coronary vessels and the fibrous matrix of the mature heart [318].

6.6. Postnatal cardiomyocyte proliferation:

Soon after birth mammalian cardiomyocytes exit cell cycle and lose the capacity to regenerate. We recently demonstrated that postnatal activation of calcineurin contributes to postnatal cell cycle withdrawal by dephosphorylating the homeodomain transcription factor HOXb13, promoting its nuclear translocation to act in conjunction with the TALE family protein MEIS1 to mediate cell cycle arrest [319]. Remarkably, there is evidence of increased proliferation of adult cardiomyocytes following MI in αMHC-RCAN1 transgenic mice compared to in wild type animals post-MI, suggesting that calcineurin actively contributes to suppression of mitosis in the adult myocardium following damage.

7. Special considerations:

7.1. Not all calcineurin activity is pathological:

Much of the study of calcineurin signaling in the heart has focused on pathological outcomes, yet the high level of calcineurin present in heart tissue is a strong indication that calcineurin is central to normal cardiac function. Indeed, cardiac metabolism and function are severely compromised in mice with a cardiomyocyte-specific deletion of CnB1, demonstrating that calcineurin activity is essential for maintaining heart health [45, 46]. Although some investigators have suggested that calcineurin signaling only participates in pathological hypertrophy and is not engaged during physiological, hypertrophic growth [112], it is more likely a matter of signal intensity and duration. Consistent with a role for calcineurin in physiological hypertrophic growth exercise-induced hypertrophy is blunted in αMHC-RCAN1 transgenic mice [6].

The mechanisms activating calcineurin during physiological verses pathological hypertrophy may be vastly different, particularly given the changes in the T-tubule system, dyadic junction, and Ca²⁺ handling that occur with failure. Some downstream substrates may be common to both processes while others are unique. Equally important are the activities of other cell signaling cascades with which calcineurin converges. An elegant illustration of this can be seen in studies of human cardiac troponin C (cTnC) mutations that lead either to hypertrophic cardiomyopathy (HCM) or dilated cardiomyopathy (DCM). Calcineurin was activated in the hearts of mouse models carrying either type of mutation [320]. However, the HCM mutations increased sarcomeric tension, leading to concentric hypertrophy, whereas the DCM mutations decreased sarcomeric tension, resulting in eccentric hypertrophy. Thus, even under pathological conditions, activation of calcineurin may occur in response to different inputs and have different outcomes. Similarly, we propose that, although there are certain to be some aspects of similarity, the behavior and targets of calcineurin signaling in a healthy heart are likely distinct from those set in motion by pathological stresses.

7.2. There is a gradient in calcineurin activity across the ventricular wall:

In a normal healthy heart there is evidence of a gradient in calcineurin activity across the ventricular wall such that calcineurin and NFAT activity are both lower in the epicardium (EPI) than in the endocardium (ENDO) [227]. This parallels the transmural gradients in mechanical stress [321], diastolic Ca²⁺, systolic Ca²⁺ [227], and action potential duration (APD) [322], each of which is lower in magnitude in EPI than in ENDO. The transmural APD gradient allows ventricular repolarization to proceed in a synchronized wave from EPI to ENDO, supporting efficient pump function and preventing arrhythmias. The APD gradient is blunted in both human heart failure [323] and mouse models of failure [324]. Studies in mice suggest that NFATc3 contributes to the APD gradient by suppressing Kv4.2 expression in the endocardium, thereby decreasing I_to to prolong APD in the endocardium [227, 325]. However, the mechanism of repression is not understood and the relevance to human pathology is not clear as Kv4.2 is negligible in human heart [326, 327]. Given the ability of calcineurin to impact the activity and/or trafficking of a diversity of membrane proteins, calcineurin is just as likely to shape transmural gradients and their remodeling through transcription-independent control over other target channels that contribute to APD, such as LTCC and the sodium/potassium ATPase. Alternatively, the transmural gradient in calcineurin activity could be a consequence rather than the cause of the gradients in the activities of these channels. This complicated relationship requires further study.

8.3. Calcineurin activity changes throughout the day in healthy hearts:

Many physiological processes of the cardiovascular system display robust circadian rhythms and disruption of normal circadian rhythms predisposes humans to cardiovascular disease [328-333]. We have shown that calcineurin activity cycles daily in the hearts of healthy mice, with a peak in activity at the animal’s transition to rest and the lowest activity at the transition to waking [334]. In wild type mice there is as much as a 20-fold change in nuclear translocation of NFAT, binding of NFAT to chromatin, and expression of Rcan1.4, over the course of twenty-four hours. This suggests that, although either sustained activation of calcineurin [4] or its complete absence [45] is damaging, daily, periodic activation of calcineurin is compatible with heart health. Damage to the heart from ischemia reperfusion (I/R) is also time of day dependent both in humans and mice, with the greatest damage occurring near the transition to waking [97, 335, 336]. Administering FK506 to at the time of reperfusion reduced damage when reperfusion occurred near the transition to waking, when the heart is at greater risk, but had no additional benefit if reperfusion occurred at the transition to rest, when the heart is already inherently protected [97]. This suggests not that all damage to the heart from I/R is calcineurin-dependent, but, that the component that is dependent on time-of-day.

Circadian activation of calcineurin is not unique to the heart and has also been documented in skeletal muscle [337], as evidenced by changes in Rcan1.4 transcript levels and activity of NFAT reporters. In a series of elegant experiments these investigators demonstrated that circadian activation of calcineurin/NFAT in skeletal was independent of the transcriptional circadian clock control and was driven instead by muscle innervation and contractile activity. By extension, we postulate that circadian activation of calcineurin in the heart is being driven by changes in cardiac demand.

9. Conclusion

The impact of calcineurin signaling on cardiac health and disease is far more complex than simple activation of NFAT-dependent hypertrophic gene expression. The calcineurin targets highlighted here are by no means exhaustive of all known calcineurin substrates and targeting molecules that may be relevant to cardiovascular biology, however, we have tried to cover the majority of those for which molecular studies have been carried out in the context of cardiomyocytes. However, calcineurin mediates process in all cell types and tissues of the cardiovascular system. Many of these may be equally important to maintaining a healthy heart. Key points to keep in mind are: (1) Calcineurin signaling in neonatal cardiomyocytes can be very different than that found in adult cardiomyocytes and changes in heart failure. (2) Targeting of calcineurin and its substrates to specific sites of Ca²⁺ release creates signaling microdomains. (3) Activation of calcineurin is no not always pathological. (4) In a healthy heart calcineurin activity is heterogeneous relative to location in the ventricular wall and time of day. (5) Sustained activation of calcineurin leads to fundamental changes in a number of substrates that often act in a feed-forward fashion that accelerates cardiac decline.

Highlights:

1. Calcineurin activity is not always pathological.

2. Calcineurin and its substrates are anchored near specific Ca²⁺ release microdomains.

3. Mechanisms of activation and substrate choice depend on developmental stage and cardiac health.

4. Calcineurin activity is heterogeneous relative to location and timing.

5. Sustained calcineurin activity accelerates cardiac decline.

Abbreviations

AID

autoinhibitory domain

AIS

autoinhibitory sequence

AKAP

PKA kinase anchor proteins

Akt

protein kinase B

Ang II

angiotensin II

AP

action potential

β-AR

β-adrenergic receptors

Ca²⁺/CaM

calcium/calmodulin

CamKII

CaM Kinase II

CAV3

Caveolin 3

CBD

CaM binding domain

CMYA5

cardiomyopathy associated 5

CnA

catalytic subunit of calcineurin

CnA*

constitutively active, truncated form of calcineurin

CnB

regulatory subunit of calcineurin

CREB

cAMP response element binding protein

CRELD1

Cysteine-Rich with EGF-Like Domains 1

CRTC

CREB-regulated transcriptional coactivators

CsA

Cyclosporin A

Cx43

connexin 43

DRP1

Dynamin Related Protein 1

ENDO

endocardium

EPI

epicardium

ERK

extracellular signal-regulated kinase

FOXO1

Forkhead box protein O1

IP3R

inositol-3-phosphate receptors

I1

inhibitor 1

I/R

ischemia reperfusion

I_to

fast transient outward K+ current

JNK

c-Jun NH(2)-terminal kinase

LDB3

LIM Binding Domain 3

LTCC

L-type voltage gated Ca²⁺ channel

LxVP

calcineurin docking motif

MI

myocardial infarction

MKK7

mitogen-activated protein kinase kinase 7

MPTP

mitochondrial permeability transition pore

MTOR

mechanistic target of rapamycin

NFAT

Nuclear Factor of Activated T-cells

NCX1

sodium-calcium exchanger

NHE1

sodium-potassium exchanger

NRVM

neonatal rat ventricular myocyte

OMM

outer mitochondrial membrane

PC-1

Polycystin-1

PxIxIT

calcineurin docking motif

PKA

cAMP-dependent kinase

RII

regulatory subunit II of PKA

RANKL

receptor activator of NFκB ligand

RCAN

Regulator of calcineurin

RyR

ryanodine receptor

SAP97

synapse-associated protein 97

SLIMs

short linear motifs

SR

sarcoplasmic reticulum

TBC1D10C

EPI64/TBC1D10 GTPase-activating protein

TFEB

Transcription Factor EB

TRIP10

thyroid hormone receptor interactor 10

TRPC

transient receptor potential canonical channels

TRPV

transient receptor potential channels