1. Introduction
The heart continuously balances the interplay of various signaling mechanisms in order to maintain homeostasis and respond to stress. One pathway that contributes to cardiac physiology and stress is the cyclic adenosine monophosphate (cAMP) pathway. cAMP is a universal second messenger that integrates input from G-protein-coupled receptors to coordinate subsequent intracellular signaling. Synthesis of cAMP from adenosine triphosphate (ATP) is controlled by the enzyme adenylyl cyclase (AC). In the heart, cAMP acts downstream on a variety of effectors including protein kinase A (PKA), hyperpolarization-activated cyclic nucleotide-gated channels (HCN), exchange protein directly activated by cAMP (EPAC), Popdc proteins, and a fraction of phosphodiesterases (PDEs). PKA is the most well-known and studied cAMP effector. PKA phosphorylation of intracellular targets coordinates a number of physiological outputs including contraction [
1,
2] and relaxation [
3]. HCN channel regulation by cAMP maintains basal heart rate [
4] while EPAC facilitates calcium handling and cardiac hypertrophy [
5]. PDEs degrade cAMP, further defining the temporal regulation of the signal. The most recently discovered cAMP effector, Popdc, is important for heart rate dynamics through regulation of the potassium channel TREK1 [
6].
The AC family is composed of nine membrane-bound isoforms (AC 1–9) and one soluble isoform (sAC). All of the isoforms can be found in the heart with the exception of AC8 [
7,
8]. Cardiac fibroblasts express AC 2–7 [
9], while in adult cardiac myocytes AC5 and AC6 are considered the major isoforms [
10,
11]. Lower levels of AC2, AC4, and AC9 are reported in myocytes [
12,
13].
2. Adenylyl Cyclases (ACs) and Their Role in Cardiac Function: Knockout Phenotypes
AC5 and AC6 are closely related isoforms that share similar regulatory mechanisms, including inhibition by Gαi as the hallmark of this group; however, physiologically they appear to play distinct roles in cardiac function [
8,
14]. Additional modes of regulation for AC5/6 are extensively reviewed elsewhere [
15]. AC5 and AC6 are differentially expressed in development, with age, and in a pressure overload model of cardiac hypertrophy where an increase in AC5 protein is observed in neonatal heart and models of heart disease [
16,
17]. Another potential distinction between these two isoforms is subcellular localization [
18].
Several overexpression and deletion studies have focused on roles of these isoforms in cardiac function. Two independent AC5-deletion (AC5
−/−) mouse lines have been generated. Overall, deletion of AC5 decreases total cAMP activity in cardiac membranes and isolated myocytes (~35–40%) under basal and stimulated (isoproterenol and forskolin) conditions [
19,
20]. The two studies reported varying results for changes in cardiac function. Okumura et al. [
19] observed a decrease in isoproterenol-stimulated left ventricular (LV) ejection fraction (LVEF) but no alterations in basal cardiac function (with intravenous isoproterenol). Conversely, Tang et al. [
20] noted basal changes in the contractile function of perfused isolated hearts in addition to a decreased sensitivity to β
1-adrenergic receptor agonist. The most notable finding of AC5
−/− mice was the effect on parasympathetic regulation of cAMP. Inhibition of cAMP production by Gi-coupled acetylcholine treatment is ablated and Ca
2+-mediated inhibition is significantly reduced upon AC5 deletion [
19]. Physiologically, this corresponds to a reduction in LVEF and heart rate in response to muscarinic agonists and an attenuation of baroreflexes [
19,
20]. Similarly, AC6 deletion results in a significant reduction of cAMP production in stimulated LV homogenates or cardiac myocytes (60–70%), with no changes to basal cAMP production [
21]. AC6 deletion revealed a number of unique contributions not observed in AC5
−/− including impaired calcium handling, which results in depressed LV function [
21]. In addition, levels of AC6, but not AC5, limit β-adrenergic receptor (βAR) signaling in heart [
22,
23].
In addition to cardiac contractility, AC5 and AC6 play important roles with regard to cardiac stress. Deletion of AC5 is protective in a number of models of cardiac stress, including transverse aorta constriction, chronic isoproterenol infusion, age-related cardiomyopathy, and high-fat diet, but not overexpression of Gq [
19,
24,
25,
26]. While knockout of AC5 can be beneficial to heart, overexpression of AC6 in heart infers protection in response to myocardial ischemia or dilated cardiomyopathy [
27,
28,
29], but not chronic pressure overload using transverse aorta constriction [
30]. However, the protection provided upon AC6 overexpression is independent of its catalytic activity as expression of catalytically inactive AC6 is also cardioprotective [
31], but requires proper localization via the N-terminus of AC6 [
32]. In fact, the expression of AC6 using adenoviral vectors for the treatment of heart disease is currently in clinical trials [
33]. Therefore, it is tempting to simplify the system and suggest that AC5 is largely associated with stress responses while AC6 is necessary for calcium handling and contractility. For these reasons, there has been considerable interest in AC5-selective inhibitors for the treatment of heart disease. However, deletion of AC6 can also be protective from chronic pressure overload in female but not male mice [
34]; therefore, roles for AC isoforms may depend on the type of heart disease model. AC inhibitors such as Ara-A (Vidarabine) do have benefits for the treatment of myocardial ischemia when delivered after coronary artery reperfusion in mice [
35]. However, Ara-A and related AC inhibitors are not selective for AC5 over AC6, although they show considerable selectivity over other AC isoforms [
36,
37]. Therefore, any benefits of Ara-A likely arise from inhibition of both AC isoforms. However, this could prove risky as AC6 deletion increases mortality during sustained catecholamine stress [
38].
Surprisingly, no polymorphisms that give rise to cardiovascular disease are known to occur in ACs [
39]. However, mutations in AC5 are linked to familial dyskinesia with facial myokymia (FDFM), a disease characterized by uncontrolled movement of limb and facial muscles [
40,
41]. These patients may also have a predisposition to congestive heart failure [
41]. Two FDFM mutations occur in a newly appreciated region of AC5, a helical domain that is present immediately after the transmembrane domain and precedes the catalytic cyclase domain (
Figure 1). In other nucleotidyl cyclases, this domain forms a tight hairpin to induce an active dimeric conformation of the catalytic domains [
42]. Thus, the helical domain may play a role in the stability of the catalytic core or direct regulation of activity.
Roles for additional AC isoforms in cardiac function have been largely overlooked. AC1 was proposed to function as the calcium-stimulated AC in the sinoatrial node that modulates the I(f) pacemaker current [
43,
44]. However, AC1 knockout mice are not reported to have a heart rate defect and RNA sequencing detects higher expression of AC1 in the right atrium versus the sinoatrial node [
45]. Roles for AC2 and/or AC4 are unknown. Currently, a knockout of AC4 is unavailable and AC2 knockout mice display no cardiac phenotype, although RNA for AC2 is elevated in pediatric dilated cardiomyopathy subjects [
46]. Cardiac functions for AC9 are discussed below.
3. The AKAP Connection: Generating Specificity for AC Function
Tissue distribution and regulation provide one mode for how the AC isoforms contribute to distinct physiological functions [
8,
15]. Another mode of signal specificity comes from the formation of AC macromolecular complexes through the scaffolding family of A-kinase anchoring proteins (AKAPs). AKAPs not only facilitate cellular localization of ACs but they also enhance temporal regulation of cAMP signaling. A number of AKAPs exist in heart including AKAP15/18, AKAP79/150, Yotiao, mAKAP, AKAP-Lbc, and Gravin [
47] (
Figure 2). In the heart, the spatial and temporal regulation by AKAPs provide an important mechanism to facilitate stress response. The associations of ACs with AKAPs facilitate regulation of PKA, downstream effectors, and ACs. This was shown in the dorsal root ganglion where the activation of the transient receptor potential vanilloid 1 (TRPV1) channel by forskolin or prostaglandin E2 is facilitated by AKAP79-AC5-PKA-TRPV1 complex formation and shifts the response to lower concentrations of forskolin by ~100 fold. Disruption of this complex attenuates sensitization of the channel, as anchoring of both PKA and AC5 were required to elicit the maximal effect on TRPV1 current [
48]. Anchoring of PKA and ACs to AKAPs can also regulate AC activity. Association of AC5/6 with AKAP79/150 creates a negative feedback loop where cAMP production is inhibited by PKA phosphorylation of AC5/6 [
49]. Although this complex feedback mechanism was defined in the nervous system, modulation of TRPV1 in heart is suggested to influence cardiovascular response to disease and injury [
50]. Fine tuning of the signal is important for modulating a number of effectors contributing to physiological function.
AC localization is assessed primarily through functional roles of associated complex members enriched at various cardiomyocytes substructures (
Figure 2). Association of both AC5 and AC6 with AKAP5 suggests localization at the t-tubule based upon functional association with calcium-induced calcium release [
51]. However, with respect to β adrenergic signaling, AC5 is enriched with β2ARin t-tubules whereas AC6 localizes outside the t-tubule [
18,
52]. Disruption of cAMP compartmentalization is potentially an underlying mechanism of heart failure [
52]. AC9 association with Yotiao and KCNQ1 suggests localization at intercalated discs, the sarcolemma, and t-tubules [
53]. These AC–AKAP complexes are discussed below.
3.1. AKAP5
The AKAP5 family of orthologs are named for their size on SDS-PAGE and for different species—for example, human AKAP79, mouse AKAP150, and bovine AKAP75. AKAP79/150 can associate with AC 2, 3, 5, 6, 8, and 9 as evaluated in tissue culture models [
49,
54]. In heart, AKAP79 primarily interacts with AC5/6 [
51]. The interaction site is located on the N-terminus of AC and in the second and third polybasic domain on AKAP79 (aa 77–153). In cells, AKAP79-scaffolded PKA phosphorylates AC5/6 to inhibit cAMP production [
49,
54]. This feedback loop allows for precisely timed activation and inactivation of the cAMP signal. Although AKAP79-anchored AC5/6 is inhibited by associated PKA, it is unclear how AKAP79 regulates AC2 activity in isolated plasma membranes.
Physiologically, AKAP79/150 has been studied in isolated cardiomyocytes from wild-type and AKAP150 knockouts. Deletion of AKAP150 significantly reduced stimulated calcium transients and calcium sparks in response to isoproterenol. Additionally, phosphorylation of the ryanodine receptor (RyR) and phospholamban (PLN) was eliminated in cardiomyocytes from knockout mice. It was further shown that AKAP150 forms a complex with AC5/6, PKA, protein phosphatase type 2 (PP2B or calcineurin), Ca
v1.2, and caveolin 3 (CAV3). This complex is found on t-tubules, while disruption of complex formation upon AKAP150 deletion alters CAV3 and AC6 localization [
51].
AKAP150 has additional roles that are independent of AC and PKA. AKAP150 localizes protein kinase C (PKC) and L-type calcium channels to the sub-sarcolemma in atrial myocytes enabling regulation of calcium sparklets [
55]. TRPV4 sparklets are also modulated by the AKAP150–PKC complex in a distance-dependent manner; a distance less than 200 nM between the TRPV4 and AKAP150–PKC is ideal for proper regulation [
56]. AKAP150 is also implicated in β
1AR recycling. Knockdown or knockout of AKAP150 in isolated myocytes inhibits recycling of β
1AR back to the membrane after isoproterenol stimulation, but not internalization. Isolated AKAP150
−/− cardiomyocytes have an enhanced contraction rate in response to isoproterenol and an increased cell size at basal and stimulated conditions. Based on these results it was postulated that AKAP150 is cardioprotective because the hypertrophy phenotype was enhanced in AKAP150
−/− [
57].
AKAP150 has been examined in a number of pathology models including myocardial infarction and pressure overload. AKAP150
−/− mice were subjected to transverse aortic constriction surgery (TAC) to induce pressure overload; AKAP150 expression significantly decreased in conjunction with a significant increase in hypertrophy, fibrosis, and cell death. Physiologically, deletion of AKAP150 increased left ventricular end diastolic size and impaired fractional shortening after TAC compared to sham animals. Physiological changes were mirrored by alterations in calcium signaling, as AKAP150 creates a complex between the PLN, RyR, and the sarcoplasmic endoplasmic reticulum calcium ATPase 2 (SERCA2). Phosphorylation of RyR and PLN in addition to calcium transients were impaired in response to isoproterenol [
58]. A model of myocardial infarction (MI) was also examined in AKAP150
−/− mice. Alteration in cardiac signaling that occurs after MI is well documented. MI causes an increase in NFATc3 activation and associated K
v channel downregulation. AKAP150
−/− cardiomyocytes displayed impaired NFAT translocation in response to phenylephrine, which was dependent on calcineurin activity, preventing downregulation of K
v channel currents [
59]. Cardiovascular disease is a co-morbidity associated with diabetes [
60]. Unlike the other models, in a model of diabetes mellitus, knockdown of AKAP150 ameliorates glucotoxicity-induced diastolic dysfunction in mice. In rat cardiomyocytes from diabetic animals or treated with high glucose, AKAP150 expression is enhanced combined with increased active PKC at the plasma membrane. This, in turn, promotes activation of NFκB and Nox [
61], players in the reactive oxygen species pathway that underlie diabetes-induced cardiovascular injury [
60]. Thus, while AKAP150 may play a cardioprotective role in some pathology models, this is not always the case.
3.2. mAKAP (AKAP6)
Anchored to the nuclear envelope, the cardiac splice variant of muscle AKAP (mAKAPβ) interacts with AC5 to facilitate cardiac signaling [
62,
63]. mAKAP is localized to the nuclear envelope through its interaction with nesprin [
64] while much lower levels of mAKAP are found at the sarcoplasmic reticulum (SR) [
65]. While mAKAP is intracellularly located primarily at the nuclear envelope and AC5 is membrane-bound, it is thought that localization of AC5 to the t-tubules allows for this interaction due to the close proximity of the nucleus and t-tubules at sites within the cardiomyocyte [
66,
67]. AC5 interacts with mAKAP through a unique binding site on the N-terminus (245–340). Similar to AKAP79, PKA binding to the mAKAP complex creates a negative feedback loop to inhibit AC5 activity [
63].
A number of molecules implicated in hypertrophy are anchored by mAKAP, including protein phosphatases 2A and 2B (PP2A/2B), PDE4D3, hypoxia-inducible factor 1α (HIF1α), phospholipase Cε (PLCε), myocyte enhancer factor-2 (MEF2) [
68], and p90 ribosomal S6 kinase 3 (RSK3) [
69,
70,
71,
72,
73]. Other proteins are associated indirectly with mAKAP complexes, including EPAC1 and the ERK5 and MEK5 mitogen-activated protein kinases via interactions with PDE4D3 [
74]. The interaction of mAKAP and the RyR at the SR promotes phosphorylation and enhances calcium release [
75], while RyR located within the nucleus promotes hypertrophy as discussed below.
A role for mAKAP in pathological hypertrophy was first described in mAKAP knockdown myocytes [
76] and subsequently shown in mAKAP knockout mice where knockout mice subjected to TAC had reduced hypertrophy, cell death, and did not display TAC-inducible gene expression [
77]. Further characterization of the mAKAP macromolecular complex highlights how multiple pathways converge on mAKAP to integrate hypertrophic signaling. The cAMP pathway is integrated through AC5–mAKAP–PDE4D3–EPAC binding to utilize and maintain local cAMP pools [
63,
69,
71,
74]. Activated calcineurin is recruited to the complex and is required for nuclear translocation of NFAT [
78]. PLCε binds to a complex containing mAKAP, EPAC, protein kinase D (PKD), and RyR2 contributing to PKD activity and nuclear calcium levels [
72,
79].
3.3. Yotiao (AKAP9)
Yotiao is a 250 kDa splice variant of AKAP9 that is present in heart. Yotiao interacts with the alpha subunit (KCNQ1) of the slowly activating delayed rectifier K
+ current (I
Ks), a critical component for the late phase repolarization of the cardiac action potential in humans [
80]. I
Ks is made up of four alpha subunits and accessory beta subunits, KCNE1. Beta-adrenergic control of I
Ks by PKA phosphorylation of KCNQ1 increases channel current to shorten the action potential and maintain diastolic intervals in response to an increase in heart rate. Mutations in KCNQ1 are associated with long QT syndrome type 1 (LQT1), a potentially lethal hereditary arrhythmia. Not only do mutations in the KCNQ1 lead to this disease, but mutations within Yotiao (LQT11) and KCNE1 (LQT5) can also give rise to LQT syndrome [
81,
82]. A subset of these mutations in either Yotiao (S1570L) or KCNQ1 (G589D) disrupts the KCNQ1–Yotiao interaction, resulting in altered regulation of the I
Ks channel [
81].
Yotiao creates a macromolecular complex between KCNQ1 and important regulators of KCNQ1 phosphorylation. Yotiao scaffolds both positive (PKA) and negative regulators, protein phosphatase 1 (PP1) and phosphodiesterase 4DE3 (PDE4D3), of KCNQ1 phosphorylation [
80,
83,
84,
85]. Loss of this scaffold decreases cAMP-dependent PKA phosphorylation of KCNQ1, eliminates the functional response by I
Ks, and prolongs the action potential [
85]. Yotiao is the key to maintaining a tightly regulated feedback loop for I
Ks-dependent cardiac repolarization and heart rate. Although Yotiao facilitates cardiac repolarization, it cannot overcome channel mutations that alter the capacity for phosphorylation. For example, an A341V mutation in KCNQ1 acts as a dominant negative that reduces basal channel activity and KCNQ1 phosphorylation with no alteration in Yotiao binding [
86].
Of the AC isoforms that Yotiao scaffolds (AC 1, 2, 3, and 9) [
87], AC9 is the only one present in cardiomyocytes. Unlike the other AKAPs that interact with AC in heart, Yotiao does not scaffold the major cardiac isoforms AC5/6. While Yotiao binds to the N-terminus of AC9, there appear to be multiple sites of interaction of AC9 on Yotiao with the primary site located within the first 808 amino acids and a second, weaker site that overlaps with the AC2 binding site on Yotiao (amino acids 808–956). The interaction of AC9 with Yotiao and KCNQ1 was shown by immunoprecipitation of the complex from cells co-expressing all three proteins, a transgenic mouse line with cardiac expression of KCNQ1–KCNE1, and from guinea pig hearts, which endogenously express the complex. Co-expression of AC9 and Yotiao in CHO cells stably expressing KCNQ1–KCNE1 sensitize PKA phosphorylation of KCNQ1 in response to isoproterenol compared to AC9 or Yotiao expression alone [
13]. Yotiao inhibits AC2 and AC3 activity but the mechanism of inhibition is unknown; no inhibition of AC9 activity is observed [
87]. Based on these results we would postulate that the AC9–Yotiao–PKA–KCNQ1 macromolecular complex generates a local pool of cAMP that is critical for cardiac repolarization in humans.