The central domain of cardiac ryanodine receptor governs channel activation, regulation, and stability

Structural analyses identified the central domain of ryanodine receptor (RyR) as a transducer converting conformational changes in the cytoplasmic platform to the RyR gate. The central domain is also a regulatory hub encompassing the Ca 2 1 -, ATP-, and caffeine-binding sites. However, the role of the central domain in RyR activation and regulation has yet to be defined. Here, we mutated five residues that form the Ca 2 1 activation site and 10 residues with negatively charged or oxygen-containing side chains near the Ca 2 1 activation site. We also generated eight disease-associated mutations within the central domain of RyR2. We determined the effect of these mutations on Ca 2 1 , ATP, and caffeine activation and Mg 2 1 inhibition of RyR2. Mutating the Ca 2 1 activation site markedly reduced the sensitivity of RyR2 to Ca 2 1 and caffeine activation. Unexpectedly, Ca 2 1 activation site mutation E3848A substantially enhanced the Ca 2 1 -independent basal activity of RyR2, suggesting that E3848A may also affect the stability of the closed state of RyR2. Mutations in the Ca 2 1 activation site also abolished the effect of ATP/caffeine on the Ca 2 1 -independent basal activity, suggesting that the Ca 2 1 activation site is also a critical determinant of ATP/caffeine

Structural analyses identified the central domain of ryanodine receptor (RyR) as a transducer converting conformational changes in the cytoplasmic platform to the RyR gate. The central domain is also a regulatory hub encompassing the Ca 21 -, ATP-, and caffeine-binding sites. However, the role of the central domain in RyR activation and regulation has yet to be defined. Here, we mutated five residues that form the Ca 21 activation site and 10 residues with negatively charged or oxygencontaining side chains near the Ca 21 activation site. We also generated eight disease-associated mutations within the central domain of RyR2. We determined the effect of these mutations on Ca 21 , ATP, and caffeine activation and Mg 21 inhibition of RyR2. Mutating the Ca 21 activation site markedly reduced the sensitivity of RyR2 to Ca 21 and caffeine activation. Unexpectedly, Ca 21 activation site mutation E3848A substantially enhanced the Ca 21 -independent basal activity of RyR2, suggesting that E3848A may also affect the stability of the closed state of RyR2. Mutations in the Ca 21 activation site also abolished the effect of ATP/caffeine on the Ca 21 -independent basal activity, suggesting that the Ca 21 activation site is also a critical determinant of ATP/caffeine action. Mutating residues with negatively charged or oxygen-containing side chains near the Ca 21 activation site significantly altered Ca 21 and caffeine activation and reduced Mg 21 inhibition. Furthermore, diseaseassociated RyR2 mutations within the central domain significantly enhanced Ca 21 and caffeine activation and reduced Mg 21 inhibition. Our data demonstrate that the central domain plays an important role in channel activation, channel regulation, and closed state stability.
Ryanodine receptor type 2 (RyR2) is an intracellular Ca 21 release channel that is expressed predominantly in the heart and brain and plays an essential role in many cellular processes, including muscle contraction, learning, and memory, by governing the release of Ca 21 from intracellular Ca 21 stores (1)(2)(3)(4)(5)(6)(7)(8)(9). RyR2 is normally activated through a mechanism known as Ca 21 -induced Ca 21 release (CICR), in which an elevation of cytosolic Ca 21 opens the RyR2 channel, leading to a large Ca 21 release from the sarcoplasmic reticulum and the endoplasmic reticulum (10)(11)(12). Thus, activation of RyR2 by cytosolic Ca 21 is a critical step in the mechanism of CICR (13)(14)(15). Consistent with its important physiological roles, impaired Ca 21 activation of RyR2 has been associated with diseases in both the heart and brain, such as cardiac arrhythmias, cardiomyopathies, and intellectual disability (16)(17)(18)(19). However, despite its physiological and pathological significance, the molecular mechanism underlying Ca 21 activation of RyR2 is not well defined.
It has long been recognized that RyR2 contains a high-affinity cytosolic Ca 21 activation site that mediates Ca 21 activation of RyR2 and CICR (10)(11)(12)(20)(21)(22)(23)(24). Major efforts have been focused on identifying residues that are responsible for Ca 21 activation of RyR2. Through site-directed mutagenesis and single-channel analyses, we showed that a point mutation E3987A in RyR2 dramatically reduced the sensitivity of RyR2 to activation by Ca 21 (25). We also reported that point mutation E3885A in RyR3 (corresponding to E3987A in RyR2) markedly decreased the Ca 21 sensitivity of RyR3 (26). Furthermore, point mutation E4032A in RyR1 (corresponding to E3987A in RyR2) abolished Ca 21 -dependent activation of RyR1 (27,28). These findings indicate that residue Glu-3987 in RyR2 and the corresponding residue Glu-4032 in RyR1 and Glu-3885 in RyR3 play an important role in Ca 21 activation of RyRs.
Comparisons of the open and closed states of RyR1 and RyR2 reveal that the central domains of RyR1 (amino acids 3668-4251) and RyR2 (amino acids 3613-4207) play an important role in channel activation (32)(33)(34)(35)(36). The putative Ca 21 binding site in RyR1 formed by residues Glu-3893, Glu-3967, Gln-3970, His-3895, and Thr-5001 in the core solenoid domain of RyR1, and the putative Ca 21 binding site in RyR2 formed by residues Glu-3848, Glu-3922, Gln-3925, His-3850, and Thr-4931 in the central domain of RyR2 have been identified (34)(35)(36). Interestingly, the RyR1-Glu-4032 or RyR2-Glu-3987 residue is located just beside the Ca 21 activation site. Although the RyR1-Glu-4032 or RyR2-Glu-3987 residue does not directly contribute to Ca 21 coordination, it is involved in H-bonding between the central domain and the C-terminal domain (CTD) that may stabilize the Ca 21 binding pocket. Furthermore, the putative ATP-and caffeine-binding sites have also been localized to regions very close to the Ca 21 activation site (34)(35)(36). Recent functional studies are consistent with the proposed locations of the Ca 21 and caffeine activation sites (37). For instance, mutations in the Ca 21 activation site (E3848A or E3922A in RyR2) diminished Ca 21 activation of RyR2, whereas mutation W4644A or W4644R in the caffeine-binding site abolished caffeine activation of RyR2 (37). Thus, the central domain is critically involved in RyR2 channel activation and regulation. It is also known that RyR is inactivated by Mg 21 (20,(22)(23)(24)38), but the molecular basis of Mg 21 -dependent inhibition of RyR is not well understood.
In addition to those Ca 21 -coordinating residues, the central domain also contains a number of residues with negatively charged or oxygen-containing side chains clustered near the Ca 21 activation site (29)(30)(31)(32)(33)(34)(35)(36). The significance of these residues in channel activation and regulation is unknown. The central domain also harbors a number of RyR2 mutations associated with catecholaminergic polymorphic ventricular tachycardia (CPVT) and sudden unexplained death (17,39), but their functional effect has yet to be determined. It is also unclear how ATP and caffeine modulate Ca 21 activation of RyR2. To address these important questions, in the present study, we performed a systematic site-directed mutagenesis and structure-function relationship analysis of the central domain of RyR2. Our results revealed that the central domain plays a critical role not only in channel activation by Ca 21 , ATP, and caffeine, but also in channel regulation by Mg 21 and in determining the stability of the closed state of the channel in the near absence of Ca 21 . Our data also showed that disease-associated RyR2 mutations located within the central domain enhanced Ca 21 activation and reduced Mg 21 inhibition of RyR2. Thus, the central domain controls the activation and regulation of the RyR2 channel.

Results
Contribution of Ca 21 -coordinating residues to Ca 21 and caffeine activation and basal activity of RyR2 Recent 3D structural analyses revealed the RyR2 Ca 21 activation site that is formed by residues Glu-3848, Glu-3922, Gln-3925, His-3850, and Thr-4931 (35, 36) (Fig. 1A). To assess the role of these Ca 21 -coordinating residues in RyR2 function, we mutated each of these residues to alanine (i.e. E3848A, E3922A, Q3925A, H3850A, and T4931A) in the mouse RyR2 and determined the Ca 21 -dependent [ 3 H]ryanodine binding to each of these RyR2 mutants with a wide range of Ca 21 concentrations (;0.1-100 mM). Note that the mouse and human RyR2 proteins share .97% amino acid sequence identity. Because ryanodine only binds to the open state of RyR, [ 3 H]ryanodine binding assay has widely been used to monitor RyR channel activity (27,(40)(41)(42). As shown in Fig. 1 1, B, C, and E and Table 1). This is consistent with previous studies (25,43). Unlike WT, the E3848A mutant displayed a complex Ca 21 dependence of [ 3 H]ryanodine binding. In the near absence of Ca 21 , the E3848A mutant exhibited a substantially higher level of [ 3 H]ryanodine binding than WT (p , 0.0001) (Fig. 1, C and E). This indicates that the E3848A mutation markedly increases the Ca 21 -independent basal activity of RyR2. Interestingly, different from WT, elevating Ca 21 concentration from ;1 mM to ;10 mM decreased (rather than increased) [ 3 H]ryanodine binding to the E3848A mutant. This suggests the existence of a putative Ca 21 inactivation site(s) in RyR2 that is independent of residue Glu-3848. However, at Ca 21 concentrations .10 mM Ca 21 , [ 3 H]ryanodine binding to the E3848A mutant increased but did not saturate even at 100  Table 1). Because we could not estimate the maximal (saturated) [ 3 H]ryanodine binding to E3848A, we cannot accurately determine the EC 50 of Ca 21 -dependent activation of E3848A, but it is likely to be greater than 10 mM ( Fig. 1E and Table 1). We propose that this increase in [ 3 H]ryanodine binding at Ca 21 concentrations .10 mM reflects the Ca 21 -dependent activation of the E3848A mutant, which would be dramatically reduced compared with that of the WT. The E3922A mutant also displayed a significantly higher level of basal [ 3 H]ryanodine binding than the WT in the near absence of Ca 21 (;0.1 nM). In addition, [ 3 H]ryanodine binding to E3922A was activated at ;10mM Ca 21 ( Fig. 1, B, C, and E and Table 1). This indicates that, like the E3848A mutation, the E3922A mutation also significantly increases the Ca 21 -independent basal activity and dramatically decreases the Ca 21dependent activation of RyR2. We also assessed the effect of the E3838A/E3922A double mutation on the Ca 21 (Fig. 1, B, C, and E and Table  1). The Q3925A and H3850A mutations also significantly increased the EC 50 (158 mM for Q3925A and 0.78 mM for H3850A) of Ca 21 -dependent activation of [ 3 H]ryanodine binding to RyR2 but did not significantly affect Ca 21 -independent basal [ 3 H]ryanodine binding ( Fig. 1, B, C, and F and Table 1). The T4931A mutation did not significantly alter the EC 50 or the basal activity of [ 3 H]ryanodine binding ( Fig. 1, B, C, and F). Taken together, consistent with 3D structural analyses, our data indicate that residues Glu-3848, Glu-3922, and Gln-3925 are critical for Ca 21 -dependent activation of RyR2. Our results also show, unexpectedly, that the E3848A and E3922A mutations could alter the stability of the closed state of RyR2 in the near absence of Ca 21 .
We also assessed the effect of mutations of the Ca 21 -coordinating residues on channel function by measuring caffeineinduced Ca 21 release in RyR2 WT or mutant expressing HEK293 cells (Fig. 1, D, G, and H, Fig. 2, and Table 2). The amplitude of caffeine-induced Ca 21 release in HEK293 cells transfected with RyR2-WT increased progressively with each cumulative addition of caffeine (from 0.05 to 1.0 mM) with an apparent EC 50 Table  2). Interestingly, the T4931A mutation also markedly inhibited caffeine activation (Fig. 1, D and H, Fig. 2, and Table 2), despite  Data are presented as mean 6 S.D. The significance of differences in EC 50 and basal activity between WT and mutants was evaluated by performing one-way ANOVA with Dunnett's multiple comparisons post hoc testing. A p value ,0.05 was considered statistically significant.
its small (insignificant) inhibitory effect on the Ca 21 -dependent [ 3 H]ryanodine binding. This suggests that the activation of RyR2 by Ca 21 and caffeine is not always the same.

Effect of ATP and caffeine on [ 3 H]ryanodine binding to RyR2 WT and Ca 21 activation site mutants
Structural analyses also revealed that the Ca 21 activation site is located near the ATP-and caffeine-binding sites and that the Ca 21 -, ATP-, and caffeine-binding sites are interconnected through the CTD (Fig. 1A). Thus, it is possible that mutations in the Ca 21 activation site may alter the actions of ATP and caffeine in RyR2 channel gating. To test this idea, we assessed the effect of ATP/caffeine on [ 3 H]ryanodine binding to Ca 21 activation site mutants. We found that ATP (3 mM) plus caffeine (3 mM) markedly increased the sensitivity of RyR2 WT to Ca 21 activation (Ca 21 sensitivity) and the basal activity (in the near absence of Ca 21 , ;0.1 nM) of RyR2 WT (Fig. 3, A, H, and I and Table 3). Similarly, ATP/caffeine enhanced both the Ca 21 sensitivity and basal activity of the T4931A mutant ( Fig. 3, B, H, and I and Table 3). ATP/caffeine also dramatically increased the basal activity (at ;0.1 nM Ca 21 ) of the E3848A, E3922A, or E3848A/E3922A mutants. The effect of ATP/caffeine on Ca 21 activation of these mutants could not be accurately determined because of their nonsaturated [ 3 H]ryanodine binding. Nevertheless, the estimated thresholds of Ca 21 activation of [ 3 H]ryanodine binding to these mutants in the absence and presence of ATP/caffeine appeared to be similar (;10 mM) (Fig. 3, C-E, H, and I and Table 3). On the other hand, ATP/caffeine significantly enhanced the Ca 21 sensitivity of the H3850A and Q3925A mutants without significantly altering their basal activity (Fig. 3, F-I and Table 3). These observations indicate that ATP/caffeine significantly affects both the Ca 21 independent basal activity and the Ca 21 -dependent activation (i.e. Ca 21 sensitivity) of RyR2. Our data also indicate that mutations of the Ca 21 -coordinating residues can exert different effects on the actions of ATP and caffeine.
Effect of mutating residues with negatively charged or oxygen-containing side chains near the Ca 21 activation site on Ca 21 and caffeine activation and basal activity of RyR2 In addition to the Ca 21 -coordinating residues, there are a number of residues with negatively charged and oxygen-containing side chains that are clustered near the Ca 21 activation site (Fig. 4A). These include Thr-3929, Gln-3932, Gln-3933, Ser-3984, Asn-3989, Glu-4146, Tyr-4149, Thr-4934, Gln-4936, and Glu-4937. The functional significance of these residues is unclear. To this end, we mutated each of these residues and determined their effect on Ca 21 activation and Mg 21 inhibition of RyR2 using the [ 3 H]ryanodine binding assay. As shown in  Data are presented as mean 6 S.D. The significance of differences in caffeine activation between WT and mutants was evaluated by performing one-way ANOVA with Dunnett's multiple comparisons post hoc testing. A p value ,0.05 was considered statistically significant.
RyR2 central domain in activation, regulation, and stability Fig. 4, the T3929A, Q3932A, E4146A, T4934A, and E4937A mutations significantly suppressed the Ca 21 activation but without significantly affecting the basal activity of RyR2 (Fig. 4, B, C, and E-H and Table 1), whereas mutations Y4149S and Q4936A significantly increased the Ca 21 sensitivity and basal activity of RyR2 (Fig. 4, B, C, G, and H and Table 1). The S3984A mutation increased the Ca 21 sensitivity but not the basal activity of RyR2 (Fig. 4, B, C, and F and Table 1). On the other hand, mutations Q3933A and N3989A had no significant effect on the Ca 21 acti-vation or basal activity of RyR2 (Fig. 4, B, C, E, and F and Table 1). Thus, these Ca 21 noncoordinating residues Thr-3929, Gln-3932, Glu-4146, Thr-4934, Glu-4937, Tyr-4149, Gln-4936, and Ser-3984 are also important for Ca 21 activation of RyR2, whereas residues Tyr-4149 and Gln-4936 are important for stabilizing the closed state of RyR2 in the near absence of Ca 21 (;0.1 nM) (Fig.  4, B and C and Table 1). We also performed caffeine-induced Ca 21 release assays in HEK293 cells transfected with these mutations to determine their effect on the activation of RyR2 by    (Fig. 6A) (Fig. 6A). The effect of mutations H3850A and Q3925A on Mg 21 inhibition of RyR2 cannot be determined, because they displayed little or no [ 3 H]ryanodine binding under these conditions. Taken together, these data indicate that residue Glu-3848, located in the Ca 21 activation site, and residues Ser-3984, Tyr-4149, Gln-4936, and Glu-4937, located near the Ca 21 activation site, are important for Mg 21 -dependent inhibition of RyR2 (Fig. 6B).

Disease-associated mutations in the central domain affect Ca 21 and caffeine activation and Mg 21 inhibition of RyR2
The central domain harbors a large number of disease-causing RyR2 mutations (Fig. 7A). RyR2 mutations K3997E (39,44), F4020L (45), R4157Q (39,46), T4196A (47), and Q4201R (48) were identified in individuals presenting with syncope, CPVT, RyR2 central domain in activation, regulation, and stability and/or sudden unexplained death. The L4188P mutation was found in a 15-year-old girl with a history of seizure-like episodes often triggered by anxiety (49). The N4097S mutation was found in a sudden unexplained death individual through molecular autopsy (50). The M3999V mutation was found in the CPVT post-mortem panel in ClinVar (National Center for Biotechnology Information. ClinVar; VCV000201320.2). To understand the effect of these disease-associated RyR2 mutations on channel function, we generated each of these mutations. Ca 21 -dependent [ 3 H]ryanodine binding was carried out to determine their effect on Ca 21 activation and Mg 21 inhibition of RyR2. All of these mutations significantly enhanced the sensitivity of RyR2 to Ca 21 activation (Fig. 7, B and E-G and Table 1). Furthermore, M3999V, L4188P, T4196A, and Q4201R mutations, but not K3997E, F4020L, N4097S, and R4157Q mutations, increased the basal activity of RyR2 (Fig. 7, C and E-G and Table 1). This is consistent with the effect of other disease-associated RyR2 mutations located in the central domain (51). We also determined their effect on the activation of RyR2 by caffeine. Consistent with their effects on Ca 21 -dependent [ 3 H]ryanodine binding to RyR2, mutations M3999V, F4020L, N4097S, R4157Q, L4188P, T4196A, and Q4201R significantly enhanced caffeine activation of RyR2. Interestingly, the K3997E mutation had no significant effect on caffeine activation of RyR2, although it slightly reduced the EC 50 of Ca 21 activation of [ 3 H]ryanodine binding to RyR2, (Fig. 7, D and H-J, Table 2, and Fig. 8). The effect of these disease-associated mutations on Mg 21 -dependent inhibition of RyR2 is shown in Fig. 9. Mutations M3999V, F4020L, T4196A, L4188P, Q4201R, N4097S, and R4157Q significantly reduced the Mg 21 -dependent inhibition of RyR2, whereas mutation K3997E had no effect on the inhibition of RyR2 by Mg 21 . These results suggest that enhanced Ca 21 and caffeine activation and reduced Mg 21 inhibition are common defects of central domain disease-associated RyR2 mutations.

Discussion
The overall 3D structure of RyR consists of a large cytoplasmic assembly and a channel domain, which are connected by the central domain. Thus, the central domain is believed to be the primary transducer that integrates structural changes in the cytoplasmic assembly to the gating of the channel pore (29)(30)(31)(32)(33)(34)(35)(36). Furthermore, recent structural studies mapped the Ca 21 -, ATP-, and caffeine-binding sites within or near the central domain (34)(35)(36). This suggests that the central domain also serves as a signaling hub that controls the activity of RyR. Indeed, mutations of the putative Ca 21 activation site within the central domain abolished or markedly diminished Ca 21 activation of RyR2 (37). Consistent with these studies, we also found that mutating the Ca 21 -coordinating residues (Glu-3848, Glu-3922, and Gln-3925) of RyR2 to alanine dramatically reduced Ca 21 activation of RyR2. The double mutation E3848A/E3922A nearly completely abolished Ca 21 activation of RyR2. Mutating the other two residues involved in Ca 21 coordination (His-3850 and Thr-4931) has a relatively lower effect on Ca 21 activation of RyR2. Surprisingly, mutation E3848A also markedly enhanced the basal activity of RyR2 in the near absence of Ca 21 . This suggests that the E3848A mutation may destabilize the closed state of RyR2, resulting in spontaneous channel opening in the absence of activating Ca 21 , but the underlying mechanism is unknown. It is possible that in the absence of Ca 21 , the negatively charged side chains of the Ca 21 -coordinating residues would tend to move away from each other because of electrostatic repulsion, which may contribute to the stabilization of the closed state of the channel. Removal of the negative charge from Glu-3848 (in the E3848A mutant) would permit the E3848A mutant reside to move closer to other Ca 21 -coordinating residues, which may mimic the Ca 21 -bound Glu-3848 state and thus destabilize the closed state. Further studies are needed to test this hypothesis. Interestingly, the enhanced basal activity in the E3848A mutant channel was inhibited by increasing concentrations of Ca 21 (;1 mM-1 mM). This suggests the presence of a putative Ca 21 -dependent inactivation site(s) that is different from the Glu-3848/Glu-3922-based high-affinity Ca 21 activation site. However, the identity, molecular mechanism, and physiological relevance of this putative Ca 21 -dependent inactivation site(s) have yet to be investigated. Taken together, these The close proximity of the Ca 21 -, ATP-, and caffeine-binding sites within the central domain suggests that these ligands may interact with one another and interdependently modulate the activity of RyR2. Indeed, we found that ATP/caffeine increased both the basal activity and the Ca 21 sensitivity of the RyR2 WT and T4931A mutant channels. On the other hand, ATP/caffeine increased mainly the Ca 21 -independent basal activity but had a relatively lower effect on the threshold or sensitivity of Ca 21 activation (by high concentrations of Ca 21 ) of [ 3 H]ryanodine binding to the E3848A and E3922A mutants (Fig. 3, C, D, and H). This suggests that the action of ATP/caffeine in the Ca 21 -dependent activation of RyR2 depends on Glu-3848 and Glu-3922 but that the action of ATP/caffeine in the Ca 21 -independent basal activity of RyR2 does not. In contrast, ATP/caffeine increased predominantly the Ca 21 sensitivity, but not the basal activity, of the H3850A and Q3925A mutants. This suggests that the H3850A and Q3925A mutations strongly stabilized the closed state of the channel in the near absence of Ca 21 , opposite to the effect of the E3848A mutation. Collectively, these data indicate that ATP and caffeine affect both the Ca 21 -dependent activation and Ca 21 -independent basal activity of RyR2 and that, vice versa, the Ca 21 activation site also has a major role in determining the actions of ATP/caffeine. These results also suggest that the Ca 21 activation site not only plays a critical role in Ca 21 activation but is also involved in controlling the stability of the closed state of RyR2.
It is of note that adjacent to the Ca 21 activation site, there is a cluster of residues with negatively charged or oxygen-containing side chains. The exact roles of these residues in RyR2 function are unknown. We previously reported that mutating one of these residues (E3987A) in RyR2 and the corresponding residue in RyR3 (E3885A) markedly reduced the Ca 21 activation of the channel (25,26). Structural analysis suggested that although the Glu-3987 residue is not directly involved in Ca 21 coordination, it may play a role in the formation of the Ca 21 binding pocket by stabilizing the interface between the central domain and the CTD (34)(35)(36). Our current work showed that mutations T3929A, Q3932A, E4146A, T4934A, and E4937A near residue Glu-3987 also significantly decreased, whereas S3984A, Y4149S, and Q4936A significantly increased Ca 21 activation of RyR2, but their effect on Ca 21 activation is much lower compared with that of the Ca 21 coordination site mutations. Thus, these residues are unlikely to be directly involved in Ca 21 binding.
Although the residues with negatively charged or oxygencontaining side chains adjacent to the Ca 21 activation site may not play a major role in Ca 21 activation of RyR2, they may be  involved in regulation of the channel by other cations, such as Mg 21 . It is well established that Mg 21 inhibits the RyR channel (20,23,24,38,52), but the molecular basis of Mg 21 -dependent inhibition of RyR is unknown. It has been proposed that Mg 21 inactivates RyR by binding to the Ca 21 activation site (20,23,24,38,52). Interestingly, the E3848A mutant that markedly reduced the Ca 21 activation site remained sensitive to both Ca 21 and Mg 21 inhibition. This suggests that Mg 21 may bind to another site, in addition to the Ca 21 activation site, to inactivate the RyR2 channel. To test this idea, we systematically assessed the effect of mutating the residues with negatively charged or oxygen-containing side chains located in the central domain on Mg 21 inhibition of RyR2. We found that mutations Y4149S, E4937A, E3848A, Q4936A, and S3984A significantly reduced Mg 21 -dependent inhibition of RyR2. Interestingly, residues Tyr-4149, Glu-4937, Gln-4936, and Ser-3984 are located in close proximity and could potentially form a binding pocket (Fig. 6B). These results suggest that the central domain may also be involved in Mg 21 inhibition and Ca 21 activation of RyR2. However, it is possible that these mutations could alter Mg 21 -dependent inhibition of RyR2 via an allosteric mechanism. High-resolution structural analysis will be needed to determine whether these residues are involved in Mg 21 binding.
We have previously shown that disease-associated RyR2 mutations located in the central domain increase the Ca 21 activation of RyR2 (51). We have now characterized additional RyR2 central domain mutations that are associated with CPVT and sudden death. Consistent with our previous work, we found that CPVT-associated mutations M3999V, F4020L, N4097S, R4157Q, L4188P, T4196A, and Q4201R (except for K3997E) significantly enhanced Ca 21 and caffeine activation and reduced Mg 21 -dependent inhibition of RyR2. However, the exact molecular mechanisms by which these disease-associated RyR2 mutations alter Ca 21 and caffeine activation and Mg 21 inhibition of RyR2 have yet to be defined. A close examination of the locations of these mutations in the 3D structure of RyR2 revealed some clues. Among the central domain mutations characterized, the M3999V mutation exerted the strongest effect on RyR2 function by markedly increasing Ca 21 and caffeine activation and the basal activity of the channel. We speculate that this Met-3999 residue may potentially form hydrophobic interactions with Leu-3986, Leu-3982, Trp-3941, and Leu-4105, which may stabilize the adjacent Ca 21 binding pocket (Fig. 10A). Mutating Met-3999 to valine (M3999V) may strength these hydrophobic interactions, which may favor Ca 21 activation. The K3997E mutation that is located in the same helix as M3999V may alter the interaction among Lys-3997, Asn-3992, and Met-4109, and the Ca 21 activation of RyR2 in a similar manner but to a different extent (Fig. 10B). We further speculate that the Phe-4020 residue may potentially interact with Arg-4086 via cation-p interactions (53) and with Phe-4016 and Leu-4023 via hydrophobic interactions (Fig.  10C). Mutation F4020L could disrupt the cation-p interactions, which may allosterically alter the confirmation of the central domain and the confirmation of the Ca 21 activation site. We also hypothesize that the N4097S mutation may alter potential interactions among Asn-4097, Lys-3976, and Val-3979 and thus the confirmation of the adjacent CTD and Ca 21 activation site (Fig. 10D). Mutations R4157Q, L4188P, T4196A, and Q4201R are all located in the U-motif (Fig. 10E). This U-motif is the part of the central domain that grasps the CTD, forming a direct pathway transducing conformational changes from the central domain to the channel gate through the CTD and the S6 helix. We speculate that mutation R4157Q may alter potential interactions among residues Arg-4157, Phe-4920, and Ile-4152 (Fig. 10F), whereas mutation L4188P may affect potential interactions among residues Leu-4188, Val-4176, and Phe-4172 (Fig. 10G). Furthermore, we speculate that mutation T4196A may disturb potential interactions among residues Thr-4196, Phe-4192, and Leu-4919, whereas potential interactions with residue Q4201R are unclear (Fig. 10H). Thus, each of these mutations may alter the confirmation of the U-motif or the interactions between the U-motif and CTD, which may in turn affect the transduction of Ca 21 -and caffeine-induced conformational changes to the channel gate or the stability of the channel gate itself, thereby altering Ca 21 and caffeine activation or the basal activity of RyR2. Clearly, further structural and functional analyses are required to validate these speculations and to define the molecular mechanisms of actions of each of these disease-associated RyR2 mutations.
Our current and previous studies (51) consistently demonstrate that enhanced Ca 21 activation and/or reduced Mg 21 inhibition of RyR2 represent a common defect of RyR2 mutations located in the central domain. However, the exact mechanisms by which RyR2 mutations in the central domain increase the propensity for CPVT and sudden death are unknown. Because activation of RyR2 by cytosolic Ca 21 is a critical component of the CICR mechanism that controls cardiac muscle contraction, enhanced RyR2 sensitivity to cytosolic Ca 21 activation or reduced sensitivity to Mg 21 inhibition would result in increased CICR sensitivity. An increased CICR sensitivity would enhance the propensity for the generation and propagation of arrhythmogenic spontaneous Ca 21 waves. Thus, it is possible that by increasing Ca 21 activation or decreasing Mg 21 inhibition of RyR2, the RyR2 central domain mutations may enhance CICR sensitivity and thus the propensity for spontaneous Ca 21 waveevoked delayed afterdepolarizations, and triggered activity, and thus CPVT and sudden death.
The effect of the central domain mutations on caffeine activation was assessed and compared with their effect on the Ca 21 activation of RyR2. Notably, mutations that suppressed Ca 21 activation also suppressed caffeine activation, whereas mutations that enhanced Ca 21 activation also enhanced caffeine activation of RyR2. On the other hand, the effect of the central domain mutations on the Ca 21 -independent basal activity of RyR2 has no correlation with that on caffeine activation. Because the caffeine-binding site differs from the Ca 21 activation site, some mutations may affect caffeine activation but not Ca 21 activation of RyR2. Indeed, we found that the T4931A mutation markedly suppressed caffeine activation of RyR2 but had little or no effect on Ca 21 activation of [ 3 H]ryanodine binding. This indicates that although there is a close relationship between Ca 21 and caffeine activation, caffeine activation does not always reflect Ca 21 activation of RyR2.
In summary, our present study demonstrates that the central domain is a pivotal signaling hub that not only controls Ca 21 activation and basal activity of RyR2 but also determines Mg 21 inhibition of RyR2. The central domain also controls the modulation of the channel by ATP and caffeine and the stability of the closed state of RyR2 in the near absence of Ca 21 . We also reveal that ATP and caffeine can enhance RyR2 channel activity in a Ca 21 -dependent and Ca 21 -independent manner. Our work also provides novel insights into the mechanism of RyR2 central domain mutations linked to cardiac arrhythmias and sudden death.

Experimental procedures
Materials [ 3 H]ryanodine was purchased from PerkinElmer. Ryanodine was purchased from Abcam. Caffeine was obtained from Sigma-Aldrich. ATP was obtained from EMD Millipore.

Preparation of HEK293 cell lysates
HEK293 cells were transfected with RyR2 WT or central domain mutant cDNAs using the calcium phosphate precipitation method as described previously (25,55). Twenty-four hours after transfection, the cells were harvested and resuspended in the lysis buffer containing 25 mM Tris, 50 mM HEPES, pH 7.4, 137 mM NaCl, 1% CHAPS, 0.5% egg phosphatidylcholine, 2.5 mM DTT, and a protease inhibitor mix (1 mM benzamidine, 2 mg/ml leupeptin, 2 mg/ml pepstatin A, 2 mg/ml aprotinin, and 0.5 mM PMSF) on ice for 60 min. Cell lysates were obtained after removing insolubilized materials by centrifugation.

Caffeine-induced Ca 21 release in HEK293 cells
The free cytosolic Ca 21 concentration in transfected HEK293 cells was measured using the fluorescence Ca 21 indicator dye Fluo-3 AM (Molecular Probes). HEK293 cells grown on 100-mm tissue culture dishes for 18-20 h after subculture were transfected with 12-16 mg of RyR2 WT or mutant cDNAs. Cells grown for 18-20 h after transfection were washed four times with PBS and incubated in KRH (Krebs-Ringer-Hepes, 125 mM NaCl, 5 mM KCl, 1.2 mM KH 2 PO 4 , 6 mM glucose, 1.2 mM MgCl 2 , 2 mM CaCl 2 , and 25 mM HEPES, pH 7.4) buffer without MgCl 2 and CaCl 2 at room temperature for 40 min and at 37°C for 40 min. After being detached from culture dishes by pipetting, the cells were collected by centrifugation at 1,000 rpm for 2 min in a Beckman TH-4 rotor. Cell pellets were loaded with 5 mM Fluo-3 AM in high glucose DMEM at room temperature for 30 min, followed by washing with KRH buffer plus 2 mM CaCl 2 and 1.2 mM MgCl 2 (KRH1 buffer) three times and resuspended in 150 ml of KRH1 buffer plus 0.1 mg/ml BSA and 250 mM sulfinpyrazone. The Fluo-3 AM loaded cells were added to 2 ml (final volume) of KRH1 buffer in a cuvette. The fluorescence intensity of Fluo-3 AM at 530 nm was measured before and after repeated cumulative additions of various concentrations of caffeine (0.025-5mM) in an SLM Aminco series 2 luminescence spectrometer with 480 nm excitation at 25°C (SLM Instruments). The peak levels of each caffeine-induced Ca 21 release were determined and normalized to the highest level (100%) of caffeine-induced Ca 21 release for each experiment. The normalized data of the ascending part of the caffeine dose response curve were fitted with the Hill equation to calculate the apparent EC 50 value of caffeine activation for each construct using the curve fitting module of Prism 8 (GraphPad Software).

Statistical analysis
All data shown are means 6 S.D. One-way analysis of variance (ANOVA) followed by Dunnett's multiple comparisons test or Student's t test (two-tailed) was performed using GraphPad Prism version 8 to assess the difference between mean values. A p value ,0.05 was considered statistically significant.

Data availability
All data are contained within the article.
Funding and additional information-This work was supported by the Canadian Institutes of Health Research Grant PJT-155940 (to S. R. W. C.), the Heart and Stroke Foundation of Canada Grant G-19-0026444 (to S. R. W. C.), and the Heart and Stroke Foundation Chair in Cardiovascular Research (to S. R. W. C.). W. G. is a recipient of the Alberta Innovates-Health Solutions Graduate Studentship Award, and B. S. is a recipient of the Heart and Stroke Foundation of Canada Junior Fellowship Award and the AIHS Fellowship Award.
Conflict of interest-The authors declare that they have no conflicts of interest with the contents of this article.