Nanometric Targeting of Type 9 Adenylyl Cyclase in Heart

Abstract

Adenylyl cyclases (ACs) convert ATP into the classical second messenger cyclic adenosine monophosphate (cAMP). Cardiac ACs, specifically AC5, AC6, and AC9, regulate cAMP signaling controlling functional outcomes such as heart rate, contractility and relaxation, gene regulation, stress responses, and glucose and lipid metabolism. With so many distinct functional outcomes for a single second messenger, the cell creates local domains of cAMP signaling to correctly relay signals. Targeting of ACs to A-kinase anchoring proteins (AKAPs) not only localizes ACs, but also places them within signaling nanodomains, where cAMP levels and effects can be highly regulated. Here we will discuss the recent work on the structure, regulation and physiological functions of AC9 in the heart, where it accounts for less than 3% of total AC activity. Despite the small contribution of AC9 to total cardiac cAMP production, AC9 binds and regulates local PKA phosphorylation of Yotiao-I_Ks and Hsp20, demonstrating a role for nanometric targeting of AC9.

Introduction

Adenylyl cyclases (ACs) generate cAMP from ATP in response to hormone stimulation of G protein coupled protein receptors (GPCRs) or crosstalk with other cellular pathways. The cAMP second messenger regulates numerous functional outcomes in the heart such as contractility, relaxation, and stress responses (1–5). These outcomes depend on cAMP interactions with effectors, including protein kinase A (PKA), hyperpolarization-activated cyclic nucleotide-gated channels (HCN), exchange protein activated by cAMP (EPAC), Popeye domain containing proteins (Popdc), and a subset of phosphodiesterases (PDEs), which degrade cAMP. Gs-coupled GPCRs can stimulate at least 4 distinct AC isoforms in cardiomyocytes, leading to activation of multiple cAMP effectors and a myriad of downstream signaling targets. The fact that all of this is controlled by a single second messenger dictates the need for local control and assembly of macromolecular complexes to organize these systems (reviewed in (6–10)). For ACs, this is achieved through A-kinase anchoring proteins (AKAPs). AKAPs bring together ACs, PKA, phosphatases, and in some cases PDEs and other kinases, allowing for the spatial and temporal regulation of cAMP synthesis, its degradation, and tight coupling to downstream effectors. For example, AKAPs can facilitate feedback inhibition of ACs (AC 5, 6, 8) via PKA, feedforward regulation via protein kinase C (PKC, AC2), and sensitization of bound PKA effectors by local cAMP (6, 11–15). Here we highlight the recent work on the structure, regulation, and physiological functions of AC9 in heart and how these findings emphasize key properties for nanometric targeting of AC.

AC9 structure and regulation

AC9 is the most divergent of the mammalian transmembrane AC isoforms, and unlike other ACs is largely forskolin insensitive. Until recently, AC9 regulation was not well understood, but new advancements in structural and functional studies have begun to shed light on AC9 regulatory mechanisms. AC9 is greatly stimulated by Gαs, although with a slightly right-shifted dose response curve (16). Gβγ regulates AC9 indirectly through stimulation of PKCβII in neutrophils to promote chemotaxis (19, 20). Whole-cell studies had initially suggested that AC9 may be directly stimulated by PKCβII and calcium-calmodulin kinase II (CaMKII) (17, 18) and inhibited by Gαi/o, a novel PKC isoform, and calcineurin (CaN or PP2B) (18, 21). Baldwin et al. (2019) sought to clarify which of these regulators act directly or indirectly on AC9. Of these, only Gαs could regulate the enzymatic activity of AC9 in vitro (Baldwin et al 2019). Other regulators, including PKC, CAMKII, Gβγ, and Gαi are incapable of altering AC9 activity in vitro, rather they appear to act indirectly, possibly due to the ability of AC9 to form heterodimers with AC5 and AC6 upon overexpression (16). The lack of direct regulation of AC9 by Gαi/o was not surprising, since AC9 lacks the key amino acid residues identified in AC5 as necessary for Gαi regulation (22).

Recently, the full-length structure of AC9 bound to activated Gαs was resolved using cryo-electron microscopy (23). The AC catalytic core is formed by the interface of two large cytoplasmic domains (C1 and C2), with Gαs binding to the C2 domain (24). The larger structure reveals the molecular details of the two sets of six transmembrane domains (TM1–6 and TM7–12) and the helical domain that, together with the catalytic core, form an overall pseudo-two-fold symmetry. Surprisingly, the structure also reveals a unique C-terminal extension (C2b) that occupies the catalytic and allosteric sites, suppressing AC9 activity (23). This extension is unique to AC9, as no other ACs show homology in this region (23, 25). The presence of this C2b region in AC9 results in a 45-fold increase in the K_m for ATP when activated by Gαs, compared to the basal activity of AC9 or Gαs-stimulation of other ACs (23, 25). The response to GPCR stimulation in cells expressing AC9 without the C2b domain is several fold higher than cells expressing full length AC9 (25). Interestingly, both human and rodent myocardial tissue contain AC9 protein lacking this C-terminus. In vertebrates, the C-terminus is generated by a single exon with no apparent alternative splice sites, suggesting the C2b region is cleaved in vivo. It is interesting, therefore, to conjecture that AC9 activity may be regulated by proteolytic processing. AC9 activity may also be regulation by its interactions with scaffolding proteins which facilitate physiological functions, but the mechanisms remain elusive.

Targeting of AC9 to Yotiao-I_Ks

AC9 binds the AKAP Yotiao in both brain and heart (26, 27). Yotiao is the smallest splicing variant of AKAP9 at 250 kDa. In the heart, Yotiao interacts with the KCNQ1 alpha-subunit of the slow delayed rectifier current, I_KS. PKA phosphorylation of KCNQ1 increases the I_KS current to enhance relaxation of muscle in response to increased heart rate. Mutations in KCNQ1 and Yotiao are associated with long QT syndrome 1 and 11 (LQT1 and LQT11), a potentially lethal arrhythmia. A subset of KCNQ1 or Yotiao mutations reduce binding of Yotiao and KCNQ1, disrupting sympathetic regulation of I_KS (28). The recruitment of PKA to the Yotiao-I_KS complex facilitates not only Yotiao phosphorylation but also that of KCNQ1, while bound protein phosphatase 1 (PP1) opposes the actions of PKA for tight temporal control (29–31). Similarly, association of the phosphodiesterase PDE4D3, degrades cAMP near the complex to both temporally and spatially regulate channel function (32).

Yotiao interacts with AC isoforms 1, 2, 3, and 9, but only AC9 is expressed in adult cardiac myocytes (26, 27). AC9 association with KCNQ1 is completely dependent on the presence of Yotiao, as the N-terminus of AC9 directly binds to Yotiao in a region distinct from the KNCQ1 interaction site (27). Although adult mice do not express the I_Ks current (29), a KCNQ1-Yotiao-AC9-PKA complex can be detected by western blot from hearts of transgenic mice that have cardiac expression of the human KCNQ1-KCNE1 subunits of I_KS (IKs mice). The AC9-Yotiao-KCNQ1 complex is also detected in guinea pig heart extracts that express endogenous AC9, Yotiao, and I_KS (27).

For AKAP5 (a.k.a. AKAP79 or AKAP150), anchoring of AC to an AKAP-PKA-TRPV1 complex sensitizes the bound effector (TRPV1) to regulation by cAMP and PKA, shifting the forskolin dose-response curve necessary for channel activity by ~100 fold (12). Similarly in CHO cells expressing both AC9 and Yotiao, KCNQ1 phosphorylation is sensitized to lower concentrations of beta-adrenergic stimulation compared to cells expressing only AC9 or Yotiao (27). Furthermore, a catalytically inactive AC9 (AC9-D399A) resulted in complete loss of beta-adrenergic stimulated phosphorylation of KCNQ1 (33). Thus targeting AC to an AKAP-scaffolded complex can lead to greater local increases in cAMP to sensitize the system to the regulatory effects of PKA.

Recently, Li et al. (2019) showed an in vivo requirement for AC9 in the phosphorylation and regulation of I_Ks. Utilizing AC9 knockout mice (AC9KO) crossed with the transgenic IKs mice (AC9KO-IKs), the amount of AC activity associated with Yotiao and KCNQ1 immunoprecipitates could be measured. Yotiao and KCNQ1 pull-down significant Gαs-stimulated AC activity from heart extracts of IKs mice, but this activity is eliminated in hearts from AC9KO-IKs mice (33). Therefore AC9 is the only AC isoform that interacts with the Yotiao-KCNQ1 complex in the heart. Activation of PKA within the complex leads to phosphorylation of KCNQ1, increasing I_Ks currents, which in vivo shortens the action potential for maintaining diastolic intervals in response to increased heart rate. In IKs mice injected with beta-agonists, in vivo phosphorylation of KCNQ1 is increased 2-fold, while in hearts isolated from AC9KO-IKs mice, the increase in phosphorylation is lost. Phosphorylation of KCNQ1 decreases the voltage dependence for activation of the channel, increasing the rate of activation and decreasing the rate of deactivation (34–36). In adult cardiomyocytes isolated from IKs and AC9KO-IKs mice, deletion of AC9 results in a blunted response of I_Ks currents to beta-adrenergic stimulation as well as a reduced shift in the voltage dependence of activation (−10 mV vs −4 mV for IKs versus AC9KO-IKs) (33).

Although AC9 mRNA and protein is present in cardiac fibroblasts and myocytes, the extent to which AC9 contributes to total AC activity was unknown (27, 37, 38). Total basal or Gαs-stimulated AC activity in cardiac membranes from AC9KO mice show no discernible difference from wild-type. Based on this and use of a P-site inhibitor (SQ 22,536) with higher selectivity for AC5/6 over AC9, AC9 is estimated to contribute to less than 3% of total AC activity in heart (33, 38). Despite the undetectable contributions of AC9 to total AC activity, deletion of AC9 eliminates all Yotiao-associated AC activity. Functional analysis of anesthetized wild-type and AC9KO mice reveal a significant reduction in heart rate in AC9KO mice, while structural and myocardial function remain unchanged (38). The bradycardia in AC9KO mice suggests a role for AC9 in the sinoatrial node (SAN). Yotiao associated-AC9 activity is strongly detected in SAN from wild-type mice (5-fold increase over controls), while this activity is lost in SAN from AC9KO mice (38). Some long QT patients also show bradycardia, however, since mice do not have an endogenous I_Ks current, the mechanism for the bradycardia in mice is unknown, possibly arising from an unidentified AC9-Yotiao-associated K⁺ channel (38).

Targeting of AC9 to Hsp20

Although a functional role for Yotiao-associated AC9 activity exists, it was unknown whether AC9 had global effects on cAMP signaling. Injection of mice with a beta-agonist (isoproterenol), increases PKA phosphorylation of numerous cardiac proteins. Deletion of AC9 does not alter global patterns of PKA phosphorylation or the phosphorylation of specific targets such as PKA, CREB, troponin I, and phospholamban, suggesting only local roles for AC9. However, one PKA target, heat shock protein 20 (Hsp20), shows a 70% decrease in basal phosphorylation in AC9KO hearts (38). Hsp20 phosphorylation is unaltered in lysates from brain, suggesting cardiac specific regulation by AC9. Although Hsp20 is found in the left ventricle and right atrium of heart, significant basal Hsp20 phosphorylation is only present in left ventricle.

If AC9 only contributes to local cAMP, then AC9 should be part of an Hsp20 complex to regulate basal PKA phosphorylation levels. Indeed, immunoprecipitation of Hsp20 brings down significant Gαs-stimulated AC activity in heart from wild-type mice, while AC9KO mice show 35% less Hsp20-associated AC activity. Surprisingly, the association of AC9 with Hsp20 is independent of Yotiao, as Yotiao is undetected in Hsp20 immunoprecipitates. Moreover, Yotiao and Hsp20 compete for AC9 binding when overexpressed in HEK293 cells. Cellular interactions between AC9 and Hsp20 are confirmed by proximity ligation assay (PLA), although whether this interaction is direct or forms via an unidentified AKAP scaffold is unclear (38). Displacement of AC9 with a catalytically dead AC9 (AC9-D399) in rat neonatal cardiomyocytes, results in a 77% decrease in beta-adrenergic-stimulated phosphorylation of Hsp20 compared to controls, suggesting that AC9 plays a key role in the generation of local cAMP levels within an Hsp20 complex in cardiomyocytes (38).

PKA phosphorylated Hsp20 plays a cardioprotective role in heart (39–41), therefore a decrease in basal phosphorylation of Hsp20 in AC9KO mice suggests a loss of this protective function leading to increased stress, particularly within the left ventricle. Pulsed-wave Doppler echocardiography measures early (E) and late (A) blood flow velocities through the mitral valve into the left ventricle. In AC9KO mice, the early filling velocity and the E/A ratio are both significantly reduced (38), indicative of an overall diastolic dysfunction and a stiffening of the left ventricle. Although the exact nature of the Hsp20 complex is still unknown, it appears that AC9 association is a critical component regulating spatial and temporal Hsp20 phosphorylation and overall cardiac function.

Cyclic AMP nanodomains: local versus global control

Over the last four decades, there is clear evidence for the formation of local cAMP nanodomains that are distinct from the bulk cytosolic pool of cAMP. These pools are differentially regulated and control separate cellular functions, with many physiologically relevant cAMP functions occurring within local nanodomains (9, 10, 42–50). This implies that compartmentalization of cAMP function must also be tightly regulated to propagate signals in a spatial and temporal manner. This can be achieved through creation of AKAP-scaffolded macromolecular complexes that bring together PKA, effector proteins, ACs, PDEs, and possibly receptors, in addition to other regulatory proteins. AC9 containing macromolecular complexes in heart serve as examples of two differentially regulated types of cAMP nanodomains.

As discussed above, AC9 forms two distinct complexes: I_Ks-Yotiao-AC9 and Hsp20-AC9 (Figure 1). To facilitate efficient PKA phosphorylation of the I_Ks subunit KCNQ1 and Hsp20, both complexes require anchoring of AC near the PKA substrate. For example, AC9 activity is required for phosphorylation of Hsp20 under basal conditions, where anchoring of AC9 to Hsp20 presumably facilitates activation of PKA near the Hsp20 substrate at these lower cAMP concentrations. However, upon isoproterenol stimulation, this complex can “detect” global increases in cAMP that are AC9 independent to increase Hsp20 phosphorylation. Therefore, this AC9-containing nanodomain is not highly restrictive and can detect both local and bulk cAMP production (Figure 1B). In contrast, I_Ks-Yotiao-AC9 responds to only local cAMP production and appears completely isolated from global increases in cAMP. In the absence of AC9, isoproterenol stimulation of KCNQ1 phosphorylation in vivo and I_Ks channel function in adult cardiomyocytes are nearly eliminated (33). This is particularly surprising in light of the fact that AC9 contributes to less than 3% of total AC activity in heart (33, 38). Therefore, cAMP diffusion into the nanodomain from the other 97% of AC activity must somehow be slowed, either by physical barriers or the degradation or buffering of cAMP to protect and isolate the complex (Figure 1A).

Figure 1. Cartoon model depicting two distinct AC9 complexes in the heart: I_KS-Yotiao-AC9 (A) and Hsp20-AC9 (B).

Note that both complexes require anchoring of AC near the PKA substrate to facilitate efficient PKA phosphorylation of the I_Ks subunit KCNQ1 and Hsp20. However, the mechanisms governing cAMP responses are vastly different. (A) The I_Ks-Yotiao-AC9 complex also includes PDE4D3 (PDE) and PP1. This complex is capable of responding to only local cAMP production and appears completely isolated from global increases in cAMP, represented by the cloud that separates this nanodomain from global cAMP. (B) The Hsp20-AC9 complex with a yet unknown scaffold. AC9 activity is required for phosphorylation of Hsp20 under basal conditions, where anchoring of AC9 to Hsp20 presumably facilitates activation of PKA at low cAMP concentrations. However, upon stimulation, this complex can “detect” global increases in cAMP that are AC9 independent to increase Hsp20 phosphorylation. This ability is depicted by the more diffuse cloud allowing free exchange of cAMP between domains. Therefore, the AC9-Hsp20 nanodomain is not highly restrictive and can detect both local and bulk cAMP production.

Local cAMP levels are controlled at least in part by anchoring PDE within the I_Ks-Yotiao-AC9 nanodomain. PDE4D3 binds to Yotiao and associates with the Yotiao-I_Ks complex, independent of PDE4D5 complexes organized by the beta2-adrenergic receptor (51). Inhibition of PDE4 enhances I_Ks currents in cardiomyocytes at baseline, indicating an important contribution by PDE4 family members to regulate cAMP signaling within the nanodomain (32). However, it is unclear if upon stimulation, anchored PDE4D3 activity would be sufficient to block cAMP diffusion from “bulk cytosol” into the nanodomain. Although patch-clamp and imaging studies confirm the existence of cAMP compartmentalization and the general need for localized PDEs, experimental studies combined with mathematical modeling question whether the kinetics of cAMP hydrolysis by PDEs is adequate to create and/or maintain nanodomains, since theoretical and experimental reaction rates for PDEs necessary to form a nanodomain differ by up to 4 orders of magnitude (47, 48, 52–57).

Other possible mechanisms that might contribute to decreased cAMP diffusion include physical barriers, such as the membranous architecture of cardiomyocytes, the viscous nature of the cellular environment, or buffering of cAMP molecules. Although the cellular geometry of cardiomyocytes is almost certainly a contributing factor to decreased diffusion, cAMP nanodomains are formed and maintained in cell-lines that do not form extensive t-tubule and sarcoplasmic reticulum networks. The buffering hypothesis has gained significant traction of late. It was first proposed in the late 1970’s (43, 58), as early evidence showed that cAMP bound to the regulatory subunits of PKA is protected from degradation by phosphodiesterases (59, 60). More recently, experimental and mathematical evidence further strengthen this hypothesis, as a means to explain slow cAMP diffusion (61–63). The presence of 10–25 fold excess PKA regulatory subunits over catalytic subunits in various cell types gives further credence to this concept (64).

Finally, a local pool of beta-adrenergic receptor or other Gs-coupled GPCR must be available to activate AC9 within the nanodomain. Whether this is directly coupled to the AKAP scaffold is unclear, but to date there is no evidence for beta-adrenergic receptor binding to Yotiao. However, lipid-raft and non-raft regions of the plasma membrane can also drive compartmentalization of cAMP signaling (65). Both beta-adrenergic receptors and a sub-set of AC isoforms cluster within the plasma membrane of cardiomyocytes, although the mechanism driving this organization is debated (66–68).

In summary, nanodomains are critical aspects of cAMP signaling but the regulation within each complex will be dictated by the types of anchored regulatory enzymes, and potentially the level of cAMP buffering proteins or clustering within the plasma membrane. The presence of ACs, PDEs, PKA, and phosphatases all aid in establishing feed-forward and feedback loops to control kinetics within the complex and the overall response to signals from outside the nanodomain. Despite the low levels of AC9 in heart, its nanometric targeting to distinct complexes controls important aspects of heart function, including repolarization and stress responses. However, AC9 is also implicated in other systems, including immune function, while polymorphisms of AC9 point to roles in allergies/immune response, colon cancer, and Rubinstein-Taybi syndrome (69–72). Therefore, the nanometric targeting of AC9 is likely important in many cell types.

Perspectives:

(i). Importance of the field:

The cAMP pathway is essential for the chronotropic, inotropic and lusitropic effects during the ‘fight-or-flight’ response. Several distinct adenylyl cyclase-containing complexes are formed in cardiomyocytes to spatially and temporally regulate cAMP signaling, each controlling a unique cellular function. For example, the major cardiac adenylyl cyclase (AC) isoforms associate with AKAPs to control Ca²⁺ cycling and cardiac ionotropy (AC5/6-AKAP5 complex) or cardiac remodeling and pathological cardiac hypertrophy (AC5-mAKAP complex). Although AC9 contributes to only 3% of cAMP production, it forms distinct complexes, creating nanodomains of cAMP that control repolarization of cardiac muscle (via AC9-Yotiao-I_KS-PKA-PDE complex) or cardiac stress (via Hsp20-AC9 complex).

(ii). Summary of current thinking:

There are multiple types of cAMP nanodomains that exist in cells. For some of these, AC must be anchored within the nanodomain to generate a local pool of cAMP to activate anchored PKA. As cAMP concentrations within the cell rise, some nanodomains can sense bulk cAMP in the cell (e.g. AC9-Hsp20), while more insulated domains (i.e. AC9-Yotiao-I_Ks-PKA-PDE) are shielded from effects of PKA or cAMP generated outside the nanodomain.

(iii). Future directions:

A number of important questions remain with respect to targeting of AC9 and other ACs to functional nanodomains. It is currently unknown what the stoichiometry of the macromolecular complex components are, particularly with respect to AC, PDE, and PKA, and whether GPCR-association with the complex is required. The amplitude and kinetics of the cAMP/PKA signals within each nanodomain are beginning to be measured, but it is unclear how different ACs in the complex alter these kinetics. Moreover, the dynamics of formation, dissolution, and size of a complex are completely unknown. Further research is needed to examine how compartmentalization is disrupted in disease states.