Allosteric Modulation of Ryanodine Receptor RyR1 by Nucleotide Derivatives

Abstract

The coordinated release of Ca²⁺ from the sarcoplasmic reticulum (SR) is critical for excitation-contraction coupling. This release is facilitated by ryanodine receptors (RyRs) that are embedded in the SR membrane. In skeletal muscle, activity of RyR1 is regulated by metabolites such as ATP, which upon binding increases channel open probability (P_o). To obtain structural insights into the mechanism of RyR1 priming by ATP, we determined several cryo-EM structures of RyR1 bound individually to ATP-γ-S, ADP, AMP, adenosine, adenine, and cAMP. We demonstrate that adenine and adenosine bind RyR1, but AMP is the smallest ATP-derivative capable of inducing long-range (>170 Å) structural rearrangements associated with channel activation, establishing a structural basis for key binding site interactions that are the threshold for triggering quaternary structural changes. Our finding that cAMP also induces these structural changes and results in increased channel opening suggests its potential role as an endogenous modulator of RyR1 conductance.

Introduction

Under resting conditions, intracellular Ca²⁺ levels are sustained at ~100 nM and below, yet hundreds of cytosolic proteins are regulated by Ca²⁺ signaling from nanomolar to millimolar concentrations¹. Various organelles, including the endoplasmic reticulum (ER) and sarcoplasmic reticulum (SR), form intracellular Ca²⁺ stores that provide Ca²⁺ to initiate signaling pathways². In muscle fibres, the coordinated release of Ca²⁺ from the SR facilitates excitation-contraction (EC) coupling which is facilitated by ryanodine receptors (RyRs), 2.2 MDa Ca²⁺ release channels embedded in the ER and SR membranes^3–5. The activity of sarcoplasmic endoplasmic reticulum ATPase (SERCA) restores basal Ca²⁺ levels as it transports Ca²⁺ back into the ER and SR from the cytoplasm². Three isoforms of RyRs are expressed in human tissue: RyR1 is primarily expressed in skeletal muscle, RyR2 in cardiac muscle, and RyR3 has low expression levels in many tissue types. However, RyRs are expressed in multiple other cell types. With the dynamic balance of cytoplasmic Ca²⁺ levels being so critical for cellular homeostasis, it is not surprising that there are hundreds of known disease-causing mutations in RyRs, the majority of which result in a gain-of-function phenotype^5–8.

Various single base pair substitutions in RyR1 can result in malignant hyperthermia (MH), an acute pharmacogenetic disorder marked by elevated body temperature and hypermetabolism upon exposure to volatile anesthetics or certain muscle relaxants^9,10. Without intervention, the mortality rate of an MH episode is ~75%. In contrast, central core disease (CCD) is a chronic disorder resulting from various RyR1 mutations that lead to either gain or loss of function. Patients with CCD experience congenital muscle weakness that is often progressive and are additionally more susceptible to MH. Similar disorders arise from interaction partners of RyR1, such as STAC3, where mutations lead to both myopathy and sensitivity to MH^11,12. In addition, 25% of individuals with ages ≥ 70 are faced with sarcopenia, a disease associated with progressive loss of skeletal muscle mass connected with aberrant RyR1 function^13–15. The only clinically approved drug for the treatment of RyR1 disorders is Dantrolene, an RyR1-specific inhibitor^16,17. Administration of Dantrolene during a MH episode reduces the mortality rate to 5%; however, this drug is not suitable for the treatment of chronic disorders due to the myriad of side effects, including liver toxicity. Thus, RyR1 remains an important target for the development of novel drugs to treat CCD and sarcopenia in an aging population^10,15.

The mechanism underlying RyR1 activation is only beginning to be unraveled. In 2016, des Georges et al. reported cryo-EM structures of RyR1 in three different states: closed, primed, and opened¹⁸. They also identified the binding sites of activators Ca²⁺, caffeine, and ATP, in the C-terminus of the protein. An additional pocket capable of binding ATP was localized located in 2022 by Melville et al. in RyR1 and Miotto et al. in RyR2. This pocket resides in the Repeat12 domain at the corner of the cytosolic cap, and this observation supports previous studies which indicated that 2 ATP molecules can bind per RyR1 protomer^19–22. Although caffeine is capable of activating RyR1, the high concentration of this exogenous molecule required for activation suggests that there may be another endogenous molecule capable of occupying this pocket, commonly hypothesized to be a nucleotide derivative²³. Regardless, communication between the nucleotide binding pocket and the caffeine binding site with the Ca²⁺ binding site has long been hypothesized between these synergistic activators²⁴. In 2021, Chirasani et al. used molecular dynamics simulations to suggest that the binding of ATP alone can activate RyR1. Such activation produces structural rearrangements in RyR1 regions known as the central solenoid (CSol), C-terminal domain (CTD), the loop connecting transmembrane helices S2 and S3 (S2S3), and the ‘thumb-and-forefingers’ (TaF) domain that interacts with the CTD. These rearrangements result in pore dilation and the remodeling of the Ca²⁺ binding pocket, thereby increasing the channel’s sensitivity to Ca²⁺ activation. Planar lipid bilayer recordings also show that ATP can result in higher channel open probability (P_o) with low nanomolar Ca²⁺ concentrations and in the absence of caffeine^25,26.

In this study we investigated the binding of ATP and its shorter fragments ADP, AMP, adenosine, adenine, as well as cAMP to RyR1 by cryo-EM. We show that all these nucleotide building blocks occupy the same C-terminal binding pocket resulting in varying degrees of structural rearrangements and provide evidence that cAMP has the potential to act as an endogenous effector.

Results

ATP-derivative bound RyR1 structures delineate important interactions for ATP binding

RyR1 was purified from rabbit skeletal muscle (>97% sequence homology as compared to human RyR1) as a complex with recombinant FKBP12.6. Despite the nucleotide binding pocket of RyR1 being non-catalytic, ATP-γ-S was used as a slowly hydrolysable ATP analog to prevent spontaneous degradation. ATP-γ-S was supplemented at 5 mM, and the sample contained 2.5 mM EGTA to limit the activating effects imparted by Ca²⁺.

Figure 1A depicts a derived atomic model for ATP-γ-S bound to RyR1. A locally refined map (3.2 Å local resolution) of the C-terminus and transmembrane domain reveal the precise binding mode of ATP-γ-S within the protein (Figure 1B and C, Supplementary Figure S1, Supplementary Table 1). As shown in previous cryo-EM reconstructions of ATP-bound RyR1^18,27, the phosphate tail of ATP-γ-S is coordinated by residues K4211, K4214, and R4215, while the adenine base is located in a hydrophobic cleft containing residues M4954, F4959, T4979, and L4985 (Figure 1C). Further coordinating the positioning of the adenine base, the main-chain carbonyl oxygen of H4983 is within hydrogen bonding distance of the 6-amine of the adenine base. Coordination between T4979 and the 2’ hydroxyl of the ribose sugar of ATP-γ-S also contributes to nucleotide binding. Interestingly, the binding of ATP-γ-S does not alter the size of the nucleotide binding pocket (as measured between residues K4211-F4975), relative to the apo structure solved at 3.5 Å (Supplementary Table 2). This finding is consistent with a previously reported ATP, caffeine, and Ca²⁺ bound RyR1 structure and a Ca²⁺ inactivated RyR1 structure in which the dimensions of the nucleotide binding pocket remain unchanged regardless of the ligands bound (Supplementary Table 2) (des Georges et al., 2016; Nayak and Samsó, 2022).

Figure 1. CryoEM structure of ATP-bound RyR1 highlights the molecular interactions underlying ATP binding.

A Atomic model of ATP-γ-S-bound RyR1 (Left: side view, Right: top view). Each protomer of RyR1 is shown as a different shade of green or blue. Each bound FKBP12.6 protein in the RyR1-FKB12.6 complex is coloured with the associated RyR1 protomer. Lines delineate the RyR1 transmembrane region, with cytoplasmic and lumenal sides labeled. B Structure of the nucleotide binding pocket derived from a 3.2Å cryo-EM density map. RyR1 is shown in cyan, and the oxygen, nitrogen, sulfur and phosphorus atoms in bound ATP-γ-S- are shown in red, blue, yellow, and orange, respectively. The grey and blue mesh represent cryo-EM density for RyR1 and the nucleotide, respectively, contoured at 0.7 sigma. C As in C but without mesh for the cryo-EM density map. Dashed lines indicate key polar interactions within 3.9 Å of heteroatoms.

Binding modes of ATP derivatives within the nucleotide binding pocket are conserved

Following analysis of ATP-γ-S-bound and apo RyR1, further cryo-EM structures of RyR1 were determined in the presence of ATP derivatives to investigate the structural consequences and the critical interactions made by truncated ATP derivatives that allow them to bind within the pocket. Structures for five additional conditions: (1) ADP, (2) AMP, (3) adenosine, (4) adenine, and (5) cAMP were solved at global resolutions ranging from 3.4 to 4.0 Å (Supplementary Table 3). Figure 2 presents the binding poses of ATP-derivatives within the RyR1 nucleotide binding pocket. We observed that ADP is bound with the same conserved binding mode as ATP-γ-S within the pocket. The adenine base remains positioned in the same hydrophobic cleft surrounded by residues M4954, F4959, T4979, and L4985, while T4979 coordinates the 2’ hydroxyl of the ribose sugar. Additionally, the main-chain carbonyl oxygen of H4983 is maintained within proximity of hydrogen bonding with the 6-amine of the adenine base. With the absence of a γ-phosphate in ADP, only K4214 is within range of coordinating the diphosphate tail. These conserved binding features remain true for AMP: the adenine base anchors the molecule, T4979 is within hydrogen bonding distance of the ribose sugar, and K4214 coordinates the phosphate tail. Of note, K4214 interacts with the γ-phosphate of ATP-γ-S, the β-phosphate of ADP, the single α-phosphate of AMP, and the 5’ hydroxyl of adenosine. Despite resolved density for K4211 and R4215, they remain >3.9 Å away from the phosphate tail of the shorter ATP derivatives.

Figure 2. Conserved binding poses of ATP derivatives within the RyR1 nucleotide binding pocket.

Atomic models after local refinement of the RyR1 nucleotide binding pocket in the absence of ligand (Apo, 3.1 Å), or bound with adenine (3.0 Å), adenosine (3.0 Å), AMP (2.9 Å), ADP (2.9 Å), or cAMP (2.9 Å). RyR1 is shown in cyan, and the ATP derivatives are shown in magenta, with the blue mesh showing the cryo-EM density, contoured at 0.7 sigma for apo, adenine, adenosine, and ADP and 0.4 sigma for AMP and cAMP. Dashed lines indicate key polar interactions within 3.9 Å of heteroatoms. Color scheme: oxygen atoms (red), nitrogen (blue), sulphur (yellow), and phosphorus (orange).

cAMP can indirectly increase the activity of RyR1 through activation of protein kinase A (PKA), which is known to phosphorylate RyR1 residue S2843 (or the corresponding S2808 in RyR2)^28,29. There are some important differences in the binding pose of cAMP as compared to AMP (Figure 2 and Supplementary Figure S2 and S3). In contrast to the other ribose-containing ligands bound to RyR1 that we report here, the 2’ hydroxyl moieties of bound adenosine and cAMP are both at a greater distance from T4979, at distances of 4.2 Å and 4.5 Å, respectively. Additionally, interaction of the α-phosphate with K4214 is diminished for cAMP as the cyclic phosphate is positioned ~2 Å further away from K4214 than the α-phosphate in AMP. Notably, hydrogen bonding between a ligand and K4214 doesn’t appear to be necessary for pocket occupancy. This is further supported by the finding that adenine is sufficient for binding within the nucleotide binding pocket when present at 5 mM (Figure 2). Given that caffeine is an exogenous molecule with a median effective concentration (EC50) in the mM range^23,30, it is commonly believed that an alternative endogenous molecule is capable of binding and allosterically modulating RyRs through the caffeine binding site. Due to the similarities in chemical structure and abundance in the cell, it has been hypothesized that nucleotide derivatives may bind within the caffeine binding site^18,23 but we did not observe density for adenine - or any other nucleotide derivative - at the known caffeine binding site under the conditions of our experiment (Supplementary Figure S4).

The addition of nucleotide phosphate groups is sufficient to induce global structural rearrangements of RyR1

Despite the similar binding modes observed for each ATP derivative, they generated markedly different effects on the overall conformation of the receptor. Figure 3A depicts aligned atomic models of Apo (shown in blue) and ATP-γ-S-bound RyR1 (shown in orange). While density for an additional transmembrane helix may be observed near residues ~4321–4337, the helix was not modeled due to resolution limitations. In the presence of bound ATP-γ-S, the pore of the channel remains closed with the inter Cα distance between opposing I4937 gating residues being ~10.7Å apart (Figure 3A IV, Supplementary Table 3). Outer regions of the cytosolic cap, including the bridging solenoid (BSol), and Repeat12 domain exhibit outward (away from the pore) and downward (towards the transmembrane domain) movements (Figure 3A I, II and V). Even with the lower resolutions observed at the periphery of the cytosolic cap, these large rearrangements are readily observable (Supplementary Figure S5 A). To quantify the relative changes in global RyR1 structure between different ligand conditions, pairs of global atomic models were aligned across residues 4500–5037, and the distance between the α-carbon of each residue was measured (Figure 3B–G). Relative to the Apo condition, binding of adenine and adenosine show limited displacement across the entire protein chain (Figure 3B, 3C and Supplementary Movie M1), despite their capacity to occupy the nucleotide binding pocket (Figure 2). The presence of an α-phosphate, as observed in AMP binding, induces two primary regions of movement: first, near the N-terminal solenoid (NSol), SPRY domains, and the Repeat12 domain (residues ~800–1400); and second, near the BSol and Repeat34 domain (residues ~2600–3600) (Figure 3D and Supplementary Movie M2). These same displacements are also seen for ADP- and ATP-γ-S-bound RyR1 (Figure 3E, 3F, and Supplementary Movie M3). Interestingly, cAMP binding also results in similar shifts in these same regions of RyR1 (Figure 3G, Supplementary Movie M4). cAMP is distinct from the other ATP-derivatives capable of inducing global structural movements as it is uniquely unable to hydrogen bond with K4214 suggesting that more subtle changes within the pocket, not resolved at the current resolution, may be leading to the observed global conformational changes (Figure 2, Supplementary Figure S2).

Figure 3. Structural rearrangements of RyR1 as a consequence of differential ATP-derivate binding.

A Least-squares superposition of atomic models for Apo RyR1 and ATP-γ-S RyR1 aligned across residues 4500–5037. The apo and ATP-bound models are shown in blue and orange, respectively. Two protomers were removed for clarity. An enlarged view of a segment of the BSol region, spanning residues ~2370–2390 (I), another enlarged view of a segment of BSol, spanning residues ~3210–3350 (II), a side view of the transmembrane domain and pore including residues ~4200–4950 (III), a cross section of the pore with missing protomers included in grey for clarity (IV), and an enlarged view of the Repeat12 domain across residues ~860–920 (V) are shown as insets. B-G Cα deviation measured between Cα atoms per amino acid residue for indicated aligned RyR1 conditions pairs. Single chains were aligned across residues 4500–5037 by least-squares superposition. Roman numerals correspond to the residue ranges presented in the respective insets in panel A.

cAMP can act as a direct activator of RyR1

Previously, cAMP was shown to indirectly activate RyR1 by increasing the activity of cAMP-dependant kinase, which in turn phosphorylates RyR1, albeit with low catalytic turnover²⁹. Our structural finding that the binding of cAMP to RyR1 induces global rearrangements suggests that cAMP may act as a direct activator of RyR1 (Figure 2 and 3G). To functionally test the effect of cAMP on channel opening, a [³H]ryanodine binding assay was performed in which the amount of [³H]ryanodine bound to open RyR1 channels was quantified via a scintillation filter-binding assay (Figure 4). At a low concentration of Ca²⁺ (0.1 μM) the serial addition of cAMP increases the proportion of open channels, achieving a maximum fold change of ~38 at 10 mM cAMP relative to conditions without the addition of cAMP (Figure 4A). In the presence of an activating concentration of Ca²⁺ (10 μM), supplementing cAMP again results in an increase in the proportion of open RyR1 channels (Figure 4B), albeit to a lower extent than in the 0.1 μM Ca²⁺ condition (a maximal fold change of ~2 was observed in the 10 μM Ca²⁺ condition at both 3 mM and 10 mM cAMP). Even at the lowest cAMP concentration tested (0.1 mM), we observed increases in [³H]-ryanodine binding at both Ca²⁺ concentrations. As shown in Figure 4C, the relative effect of cAMP on channel opening is smaller as compared to activating levels of Ca²⁺, yet the activating effects of cAMP are observed both in sub-activating (0.1 μM) and activating (10 μM) concentrations of Ca²⁺. Taken Together, these structural and functional results provide support for a role of cAMP as a direct activator of RyR1, likely through its nucleotide binding pocket.

Figure 4. cAMP increases RyR1 opening in a [³H]ryanodine binding assay.

A. Conditions with low activating levels of Ca²⁺ (0.1 μM) show increased [³H]ryanodine binding with increasing concentrations of cAMP (0.1 – 10 mM) as measured in fmol of [³H]ryanodine per mg of RyR1. B. Conditions with high activating levels of Ca²⁺ (10 μM) show increased [³H]ryanodine binding with increasing concentrations of cAMP (0.1 – 10 mM) as measured in pmol of [³H]ryanodine per mg of RyR1. C. Comparison of [³H]ryanodine binding at low (0.1 μM) and activating (10 μM) levels of Ca²⁺ measured in pmol of [³H]ryanodine per mg of RyR1. Statistics: t-test vs 0 mM cAMP, * p<0.05, ** p<0.01.

Discussion

In this study, we demonstrate that an adenine base alone is capable of binding in the nucleotide binding pocket of RyR1 (Figure 2). Regardless of the ATP-derivative bound, the adenine base always remains in the same binding orientation in a hydrophobic pocket surrounded by residues M4954, F4959, T4979, and L4985, while T4979 coordinates the ribose sugar (Figure 2 and Figure 5A–F). K4214 interacts with the 5’ hydroxyl of adenosine and the terminal phosphate of ATP-γ-S, ADP, and AMP. However, this 5’ hydroxyl interaction is diminished with cAMP binding due to the cyclic phosphate being constrained closer to the ribose sugar and adenine base (Figure 1, Figure 2, and Supplementary Figure S2). Although the phosphate of cAMP appears too far from K4214 for a direct salt bridge interaction, one possibility is that there are one or more water molecules, not observed at current resolutions, that link the phosphate of the K4214 side chain. These findings are consistent with a 2021 study by Yuan et al. indicating that mutating residues that interact with the adenine base of ATP abolishes nucleotide binding; while mutating residues that interact with the triphosphate tail only affects channel activation³¹.

Figure 5. The adenine base orients ATP analogs within the nucleotide binding pocket and AMP is the smallest fragment of ATP capable of inducing global structural rearrangements.

A-F Aligned atomic models of RyR1 nucleotide binding pockets are shown in each panel. The apo RyR1 mainchain is shown as a surface model in grey. Adenine is coloured in brown, adenosine in purple, AMP in orange, ADP in teal, ATP in pink, and cAMP in green. Models were aligned to the apo structure across residues 4953–4985. G Ligplots⁴⁸ of the key interacting residues within the nucleotide binding pocket for adenosine and AMP. Dashed lines indicate key polar interactions within 3.9 Å of heteroatoms. ATP derivatives are shown in pink with nitrogen in blue, oxygens in red, and phosphates in purple. Hydrophobic residues are shown as dashed red arches.

Despite the ability of all the ATP derivatives to occupy the same pocket, not all analogs induce the same global structural movements in RyR1. Binding of AMP, ADP, ATP-γ-S, and cAMP are associated with overall movements of peripheral regions of the cytosolic cap outwards, away from the pore, and down towards the transmembrane domain (Figure 3, 5G, and Supplementary Movies M1–4). Even with relatively lower resolution in regions towards the periphery of the cytosolic cap, these ATP derivative dependant conformational shifts are distinct. Such movements have been previously identified in RyR1 priming, a conformational continuum that is correlated with higher P_o¹⁸. Further, similar conformational shifts can be observed in RyR1 structures bound to ATP, caffeine, and Ca²⁺ in the closed conformation (Supplementary Figure S5 C and D). Despite functional results suggesting that binding of ATP alone is sufficient to increase the probability of channel opening^25,26, we show that the ATP-γ-S -bound protein is in a closed conformation, a finding consistent with previous structures of RyR1 bound with both ATP and caffeine¹⁸. The conserved structural movements induced by AMP, ADP, and cAMP binding, as compared to ATP, suggests the potential for similar effects on priming by ATP-derivatives. A study by Lindsay et al. (2018) analyzed the effect of ATP-derivative binding on RyR2 channel opening (as measured by [³H]ryanodine binding and planar lipid bilayer recordings)³². It was observed that although the addition of ATP resulted in the greatest increase in P_o, ADP, AMP, and adenosine also increase P_o to a lesser degree. While our results show no structural movements when adenosine is bound, movements upon AMP, ADP, and ATP binding match greater binding of [³H]ryanodine under these conditions.

Although ADP and AMP directly activate RyR1, it is likely that under normal physiological conditions ATP is typically bound to the nucleotide binding pocket of RyR1. In skeletal myocytes, the physiological concentration of ATP is ~5.9 mM³³ which is well above the single channel EC50 of 0.22 mM for ATP in ryanodine receptors³⁴. Further, cellular ATP concentrations are largely homeostatic, meaning that despite large fluctuations in ATP turnover during exercise, the concentration of ATP remains relatively constant³⁵. However, ATP levels can drop by nearly 20% during episodes of highly elevated exertion^36,37. In contrast, during moderate to high intensity exercise, cellular ADP concentrations have been reported to increase from 37 μM to 213 μM while AMP concentrations increase from 0.2 μM to 9.3 μM³⁸. Activation of RyRs was previously reported below 1 mM ADP and below 100 μM AMP, suggesting that while under normal physiological conditions the nucleotide binding pocket of RyR1 is primarily bound by ATP, during intense exercise ADP and AMP may have partial activating effects³².

Previous studies of RyR1 and RyR2 have shown binding of ATP within the Repeat12 domain^19,22. In our cryo-EM density maps, we did not observe density in the Repeat12 domain that could be attributed to any of the nucleotide derivatives (Supplementary Figure S5 B). Although the reasons for not observing signal corresponding to these nucleotide derivatives is not clear, differences in the preparation of the RyR1, or differences in local resolution could explain this discrepancy.

Several important observations suggest that ATP binding to the RyR1 C-terminal binding site results in activation, and not the binding to the Repeat12 domain. Molecular dynamics studies report that ATP binding to the C-terminal binding site results in RyR1 activation as well as communication with the Ca²⁺ binding site, increasing sensitivity to Ca²⁺ induced Ca²⁺ release²⁴. Additionally, point mutations in the C-terminal nucleotide binding site were found to completely abolish the activating effects of ATP on RyRs both in the hydrophobic cleft surrounding the adenine base (T4979F) and with triple mutations affecting phosphate tail interactions (K4211S, K4214S, and R4215S)³¹. Further, T4980M in human RyR1 (equivalent to T4979M in rabbit RyR1) is implicated in congenital myopathy³⁹, highlighting the importance of the interaction of T4979 with the 2’ hydroxyl of ATP, ADP, and AMP (Figure 1 and Figure 2). Finally, the related Ca²⁺ release channel IP3R contains a highly similar C-terminal nucleotide binding site to RyR1 and is also activated by ATP despite lacking a Repeat12 domain^5,40.

It is well established that phosphorylation of RyRs via kinases including protein kinase A (PKA, otherwise known as cAMP dependant protein kinase), prime these channels for Ca²⁺ efflux. Several sites have been proposed as PKA targets, including RyR2 residues S2808 (S2843 in RyR1) and S2030^28,29,41. Although there is uncertainty around which site is more important, aberrant phosphorylation of RyRs has been implicated in various pathophysiological mechanisms including heart failure, atrial fibrillation, and sarcopenia^9,28,42–44. Our study indicates that while cAMP is a well-known indirect activator of RyRs, its ability to bind within the nucleotide binding pocket (Figure 2) and induce displacement at the periphery of the cytosolic cap (Figure 3G) suggest its potential to act directly as an endogenous allosteric modulator. Our experiments show activation of RyR1 at cAMP concentrations as low as 100 μM (Figure 4). However, the ability of cAMP to activate RyRs under physiological conditions will highly depend on the available cAMP concentrations. Although overall cytosolic concentrations of cAMP are low, ~1.2 μM⁴⁵, nanodomains with locally increased cAMP have been shown to exist. For example, Brandenburg et al. observed nano-domains of elevated adenylyl cyclase activity - and therefore increased cAMP levels - at the cytosolic surface of RyR2s in arterial myocytes, resulting in increased Ca²⁺ efflux⁴⁶. Although this remains to be further explored, our structural and functional results suggest the possibility that elevated local concentrations of cAMP may have a dual effect: 1) increasing RyR phosphorylation through activation of PKA, and 2) allosteric modulation of RyRs through direct binding to the nucleotide binding pocket.

AICAR, an orally available prodrug, has been demonstrated to prevent heat-induced sudden death in malignant hyperthermia mouse models and is believed to act as an RyR1 inhibitor⁴⁷. Given that AICAR - and its phosphorylated form ZMP - are nucleotide analogs, we hypothesize that both of these compounds have the potential to occupy the RyR1 nucleotide binding pocket and outcompete the activating effects of ATP to act as a RyR1 inhibitor. As such, we hypothesize that the independent binding of AICAR or ZMP would not induce global structural rearrangements associated with channel activation. Evidence for this mechanism of RyR1 inhibition are supported by in vitro studies reporting that AICAR diminishes the activating effects of ATP⁴⁷.

The mechanisms of communication between the nucleotide binding pocket and the caffeine and Ca²⁺ binding sites remain unclear. Despite MD simulations suggesting a change in caffeine and Ca²⁺ binding site conformations when ATP is bound²⁴, such changes were not observed in our structures regardless of which ATP-derivative was bound (Supplementary Table 2, Supplementary Figure S5 E). It is likely that local subtle changes within the nucleotide binding pocket – not readily resolved at the current resolutions – lead to gradually larger rearrangements further away from the binding site observed in the cytosolic cap. The non-physiological, high concentrations of caffeine required for in vitro RyR1 activation raises the question of whether there is an alternative endogenous molecule capable of binding the caffeine binding site. Absence of adenine in the caffeine binding site at 5 mM, despite binding of adenine in the nucleotide binding pocket, suggests that this is not the endogenous molecule that this allosteric site evolved to bind (Supplementary Figure S4). Recently, xanthines were shown to bind the caffeine binding pocket , and single-channel recordings showed an increase in channel opening at concentrations as low as 10 μM²².

Understanding the environment of the nucleotide binding pocket and the diversity of molecules that can fit within this site are important for efforts aimed at structure-guided therapeutic targeting of RyRs. The identification of molecules such as adenosine and adenine that can occupy the nucleotide binding pocket without resulting in the structural shifts of RyR1 associated with activation suggests that such molecules compete with ATP for binding RyR1. Our structural analysis also provides a structural foundation to design small molecules that can occupy the nucleotide binding pocket with higher affinity and function as potent allosteric inhibitors.