Arrhythmogenic Calmodulin Mutations Affect the Activation and Termination of Cardiac Ryanodine Receptor-mediated Ca2+ Release*

Background: Mutations in the Ca2+ sensing protein calmodulin (CaM) cause lethal cardiac arrhythmias. Results: CaM mutations impair the activation and termination of store overload-induced Ca2+ release via the cardiac ryanodine receptor (RyR2). Conclusion: CaM mutations alter RyR2-CaM interaction, thereby affecting RyR2-mediated Ca2+ release. Significance: Aberrant regulation of RyR2 store Ca2+ sensing is a potential component of calmodulin-mediated cardiac arrhythmias. The intracellular Ca2+ sensor calmodulin (CaM) regulates the cardiac Ca2+ release channel/ryanodine receptor 2 (RyR2), and mutations in CaM cause arrhythmias such as catecholaminergic polymorphic ventricular tachycardia (CPVT) and long QT syndrome. Here, we investigated the effect of CaM mutations causing CPVT (N53I), long QT syndrome (D95V and D129G), or both (CaM N97S) on RyR2-mediated Ca2+ release. All mutations increased Ca2+ release and rendered RyR2 more susceptible to store overload-induced Ca2+ release (SOICR) by lowering the threshold of store Ca2+ content at which SOICR occurred and the threshold at which SOICR terminated. To obtain mechanistic insights, we investigated the Ca2+ binding of the N- and C-terminal domains (N- and C-domain) of CaM in the presence of a peptide corresponding to the CaM-binding domain of RyR2. The N53I mutation decreased the affinity of Ca2+ binding to the N-domain of CaM, relative to CaM WT, but did not affect the C-domain. Conversely, mutations N97S, D95V, and D129G had little or no effect on Ca2+ binding to the N-domain but markedly decreased the affinity of the C-domain for Ca2+. These results suggest that mutations D95V, N97S, and D129G alter the interaction between CaM and the CaMBD and thus RyR2 regulation. Because the N53I mutation minimally affected Ca2+ binding to the C-domain, it must cause aberrant regulation via a different mechanism. These results support aberrant RyR2 regulation as the disease mechanism for CPVT associated with CaM mutations and shows that CaM mutations not associated with CPVT can also affect RyR2. A model for the CaM-RyR2 interaction, where the Ca2+-saturated C-domain is constitutively bound to RyR2 and the N-domain senses increases in Ca2+ concentration, is proposed.


the disease mechanism for CPVT associated with CaM mutations and shows that CaM mutations not associated with CPVT can also affect RyR2. A model for the CaM-RyR2 interaction, where the Ca 2؉ -saturated C-domain is constitutively bound to RyR2 and the N-domain senses increases in Ca 2؉ concentration, is proposed.
During cardiac excitation, Ca 2ϩ entry into the cytoplasm of cardiomyocytes through sarcolemmal voltage-gated Ca 2ϩ channels (Ca V 1.2) activates RyR2 2 channels in the SR, giving rise to the so-called Ca 2ϩ -induced Ca 2ϩ release (1)(2)(3). Ca 2ϩ released from the SR eventually leads to increases in cytosolic free Ca 2ϩ ([Ca 2ϩ ] cyt ) throughout the cardiomyocyte, where binding of Ca 2ϩ to myofilaments results in contraction (4). RyR2 channels are, however, not only sensitive to [Ca 2ϩ ] cyt but also the SR luminal free Ca 2ϩ ([Ca 2ϩ ] SR ), and both calcium concentrations modulate the activation and termination of Ca 2ϩ release (1,(5)(6)(7)(8)(9).
In the SR membrane, RyR2s arrange as homotetrameric channels extending into the cytosol, where the interaction with numerous proteins and ligands regulate the Ca 2ϩ release activity of the channel (4,10). Among these RyR2 modulators, CaM is a cytosolic inhibitor of Ca 2ϩ release both at diastolic and systolic [Ca 2ϩ ] cyt and may also serve additional regulatory purposes (11)(12)(13)(14). CaM is a ubiquitous Ca 2ϩ sensing protein in vertebrates and confers its sensing of intracellular Ca 2ϩ signals onto a multitude of protein targets, including ion channels and pumps responsible for excitation and contraction in cardiomyocytes (15). Although a cytosolic protein, CaM does affect the response of the RyR2 channel to [Ca 2ϩ ] SR , most likely as part of an extensive allosteric regulation of RyR2 (14).
CaM consists of two domains (C-and N-domains, respectively) with a flexible linker between them that enables independent and correlated functions. Each domain is comprised of two structurally integrated EF hand motifs that each binds one Ca 2ϩ ion (Fig. 1). Although highly homologous with extensive sequence identity, the two domains of CaM display distinct Ca 2ϩ binding properties and both independent and correlated interactions with protein targets. CaM binds to RyR2 with a stoichiometry of four per channel, primarily via the CaM binding domain (CaMBD) (Arg 3581 -Pro 3607 , human RyR2) (15)(16)(17)(18). In the Ca 2ϩ saturated form, the CaM C-domain appears to bind RyR2 around Trp 3587 in the CaMBD, and the CaM N-domain appears to bind RyR2 in the vicinity of Phe 3603 , although the CaM N-domain interaction is more promiscuous, especially at low free Ca 2ϩ concentrations ([Ca 2ϩ ] free ) (18 -20). Even in the absence of Ca 2ϩ , CaM can bind to the RyR2 CaMBD, most likely via the C-domain (18,19). Also, an engineered CaM mutant (CaM E31A/E67A/E104A/E140A, CaM 1234 ) defective in Ca 2ϩ binding is a competitive inhibitor of the native regulation of RyR2 by the CaM WT (21). Hence, CaM binding to RyR2 and Ca 2ϩ binding to CaM are each critical determinants of RyR2 regulation by CaM.
Three genes (CALM1-3) in the human genome encode the exact same CaM protein, and we previously found that two separate mutations in the CALM1 gene (N53I and N97S, mature CaM numbering) each lead to dominantly inherited CPVT (22). Subsequently, Crotti et al. (23) identified a CaM mutation in CALM2 (D95V) and two in CALM1 (D129G and F141L) each dominantly causing long QT syndrome (LQTS). More recently, Makita et al. (24) found another five mis-sense mutations in CALM2, three in individuals with severe LQTS (N97I, D133H, and, interestingly, N97S) and two in individuals showing features of both LQTS and CPVT (D131E and Q135P). In addition, a CaM F89L mutation was identified in a family with idiopathic ventricular fibrillation (25).
Both CPVT and LQTS lead to perturbations of excitationcontraction cycles in cardiomyocytes, which in turn can lead to ventricular fibrillation and sudden cardiac death. CPVT is characterized by syncope or sudden cardiac death following exercise or acute emotion, whereas LQTS also affects the resting heart and with increasing effect upon adrenergic stimulation (25,26). Despite the universal function of CaM, all mutations identified so far confer arrhythmia phenotypes with few or no other symptoms observed in carriers (1,(22)(23)(24)27).
The RyR2 channel accounts for the bulk of Ca 2ϩ introduced into the cardiomyocyte cytosol during contraction, and mutations in RyR2 and auxiliary proteins cause CPVT (28 -30). Thus, the aberrant regulation of RyR2 by CaM is a likely convergence point for the CaM N53I and N97S disease mechanisms causing CPVT (1,15). This is also consistent with the observation that these CaM mutations increase spontaneous Ca 2ϩ release in permeabilized cardiomyocytes (15,31). Conversely, LQTS is mainly associated with dysfunction of sarcolemmal voltage-gated Na ϩ , Ca 2ϩ , or K ϩ channels in control of the action potential. Notably the activities of a majority of these channels are regulated by CaM, for example, CaM mutations N97S, D95V, F141L and D129G all confer reduced Ca 2ϩdependent inhibition of the cardiac L-type voltage-gated Ca 2ϩ channel (Ca V 1.2) (15,26,32).
In this study, we investigated the impact of CaM mutations linked to CPVT (N53I), LQTS (D95V and D129G), or both (N97S) on the CaM-dependent regulation of RyR2 channels. To this end, we monitored the endoplasmic reticulum (ER) Ca 2ϩ dynamics in RyR2-expressing HEK293 cells transfected with CaM WT or mutants. The RyR2-mediated Ca 2ϩ release in this system is triggered by increasing the Ca 2ϩ load in the ER, which mimics the store overload-induced Ca 2ϩ release (SOICR) model for CPVT (14,33). Furthermore, we investigated domain-specific interactions of CaM with Ca 2ϩ in the presence of a peptide corresponding to the CaMBD of RyR2. Our results demonstrate that both CPVT and LQTS-associated CaM mutations alter RyR2-mediated Ca 2ϩ release.

Experimental Procedures
Plasmid Constructs-Plasmid constructs based on the pMAL vector (New England Biolabs) for recombinant expression and purification of native, full-length CaM were prepared as described previously (1,22). For expression of CaM variants in HEK293 cells, CaM coding sequences from the pMAL vectors were PCR-amplified and the products ligated into pcDNA3.1 vectors (Invitrogen). Chemically competent Escherichia coli DH5a cells (in-house stock) were transformed with pcDNA3.1 vectors, and overnight cultures were used to prepare purified plasmid preparations using the Qiagen plasmid maxi kit. pcDNA3.1 vectors were eluted in double-distilled water, and Sanger sequencing verified the sequence of CaM encoding inserts in all plasmids.
Model Fitting and Statistical Analysis-All fitting of data to mathematical models and statistical analyses was done using GraphPad Prism 6 for Mac (version 6.0f). Models and statistical method details are described below. Endoplasmic Reticulum Luminal Ca 2ϩ Imaging of HEK293 Cells Expressing RyR2 during Store Overload-induced Ca 2ϩ Release-Stable expression of murine RyR2 in HEK293 cells co-transfected with plasmids encoding CaM, and the D1ER Ca 2ϩ probe was done as previously described (14). Briefly, D1ER FRET signals reflecting ER luminal [Ca 2ϩ ] free were monitored for individual cells in an epifluorescent microscope set-ting with perfusion (6, 14) ( Fig. 2A). Each FRET signal trace was used to measure the Ca 2ϩ release properties of the RyR2 channels relative to the ER Ca 2ϩ content: the activation and termination thresholds ( Fig. 2A) and their difference, the fractional Ca 2ϩ release. The ER Ca 2ϩ store capacities were calculated as the difference between maximum and minimum FRET signal (F max Ϫ F min ). Experiments were also done with HEK293 cells Stepwise increase in Ca 2ϩ concentration elicited RyR2 SOICR oscillations, tetracaine blocked Ca 2ϩ release filling ER to maximum [Ca 2ϩ ] free , and finally caffeine opened RyR2 channels depleting ER Ca 2ϩ . Tetracaine and caffeine were used to establish maximum and minimum ER [Ca 2ϩ ] free as measured using D1ER (F max and F min ), respectively. The activation and termination thresholds were calculated relative to F max and F min . Example traces for transfection with CaM WT (A), N53I (B), and N97S (C) are shown. D-G, the activation threshold (D), termination threshold (E), fractional release (F), and store capacity (G) averaged from multiple traces are shown as bar graphs. The error bars show S.D., and asterisks indicate significant changes compared with CaM WT (p Ͻ 0.01).
expressing a RyR2 mutant with the CaMBD deleted (RyR2 ⌬CaMBD, murine RyR2 ⌬Lys 3583 -Phe 3603 ). Measured parameters were compared using one-way analysis of variance with Holm-Sidak multiple comparison test for all possible combinations and with p Ͻ 0.01 (against the CaM WT values) chosen as a conservative measure of significance.
Protein Expression and Purification-CaM was expressed from the pMAL (CaM N53I and N97S) or the pET15b (D95V and D129G) vector in E. coli Rosetta B cells (EMD Chemicals) or E. coli BL21 (DE3) cells (Novagen), respectively, and purified as previously described (1,22,23). The identity, purity, and integrity of each protein preparation were confirmed by SDS-PAGE and MALDI-TOF mass spectrometry of trypsin-digested proteins.
Titration Buffers and Verification of Free Ca 2ϩ Concentrations-Titration experiments were performed by mixing different volumes of pH-and Ca 2ϩ -buffered solution (50 mM HEPES, 100 mM KCl, 0.5 mM EGTA, and 2 mM nitrilotriacetic acid at pH 7.2 (25°C)) with the same buffer spiked to 3, 7, or 22 mM CaCl 2 to reach precalculated [Ca 2ϩ ] free levels (5). In practice, ϫ1.5 concentrated buffer stocks were prepared and proteins, peptide, Ca 2ϩ probes (final concentration, 0.75 M), and reducing agent (tris(2-carboxyethyl)phosphine; final concentration, 16.5 M) were added to the double distilled water used for diluting concentrated buffers. The calculated buffer ionic strength, which affects Ca 2ϩ binding to CaM, was stable at 0.15 M. For all titration experiments, the [Ca 2ϩ ] free was followed by including the Ca 2ϩ probe Fura-2 or Fura-6F in solutions (Invitrogen) and indirectly via measuring of Ca 2ϩ binding to CaM WT (1). Based on these measurements, a 15% error for the [Ca 2ϩ ] free was included throughout data sets and fitting procedures.
Titrations of CaM/ RyR2(Arg 3581 -Leu 3611 ) and Free CaM with Ca 2ϩ -15 M CaM in the presence of 16.5 M RyR2(Arg 3581 -Leu 3611 ) was titrated with Ca 2ϩ as previously described (1,5). Briefly, intrinsic protein fluorescence from the N-and C-domain of CaM were measured as partial Phe and Tyr emission spectra, respectively (Table 1), using a spectrofluorometer (HORIBA Jobin Yvon, FluoroMax-4P) (1,34,35). In addition, the Trp fluorescence from the RyR2(Arg 3581 -Leu 3611 ) peptide was also measured ( Table 1). Titrations of free CaM WT, D95V, and D129G with Ca 2ϩ were done by mixing discontinuous titration points in an automated liquid handler (Hamilton, Microlab STARlet) and serially transferring these to a 2-mm cuvette for emission spectra recording. Measurements were done in triplicate with 10 M CaM. Each of the fractional saturations (Y) for the N-and C-domains of CaM were fitted to the raw fluorescence intensity (FI) signals from the partial Phe and Tyr spectra at 280 and 320 nm, respectively, according to the following, where the constants b and a indicate the initial FI and the span in FI from low to high [Ca 2ϩ ] free , respectively. Y is the fractional saturation of the monitored CaM domain binding to two Ca 2ϩ as described by a two-site Adair model (36,37), where K 1 is the sum of the microscopic equilibrium constants, and K 2 is the equilibrium constant for the particular domain binding to two Ca 2ϩ . The apparent dissociation constants for either domain in the free CaM or in the presence of RyR2(Arg 3581 -Leu 3611 ) (appK D free and appK D bound , respectively) were calculated as the reciprocal square root of K 2 . When CaM binds to RyR2(Arg 3581 -Leu 3611 ), the peptide Trp fluorescence shows a peak shift from ϳ350 to 340 nm, and furthermore the FI increases markedly upon Ca 2ϩ binding to CaM (22). The raw FI for Trp fluorescence at 340 nm was also fitted to the model described above. Titration curves were normalized using the fitted a and b parameters for figure plotting purposes only. Statistical significances of differences in K 2 were evaluated via nonoverlapping 95% confidence intervals or oneway analysis of variance with Dunnett's post hoc test against values measured for the CaM WT and p Ͻ 0.05 considered significant. Furthermore, the K 2 values were also used to calculate the mutation-induced change in Gibb's free energy of Ca 2ϩ binding to the domains of CaM (⌬⌬G o free and ⌬⌬G o bound respectively) according to the following, using standard conditions of 1 M and 298.15 K (25°C).

Arrhythmogenic CaM Mutations Decrease the Activation and Termination Thresholds for Spontaneous Ca 2ϩ
Release-Spontaneous Ca 2ϩ release can occur in cardiomyocytes during increased Ca 2ϩ load in the SR lumen. This spontaneous SOICR is arrhythmogenic, because it may lead to delayed afterdepolarizations and triggered activity, a hallmark of CPVT (25,38). To determine whether CaM mutations affect this arrhythmogenic SOICR, we transfected RyR2-expressing HEK293 cells with CaM and monitored the ER Ca 2ϩ dynamics using a FRETbased ER luminal Ca 2ϩ probe D1ER (14). Perfusion of these transfected cells with increasing extracellular Ca 2ϩ concentrations induced SOICR in the form of ER Ca 2ϩ oscillations ( Fig.  2A), as reported previously (14,39). The oscillating D1ER FRET signal was then used to calculate the ER Ca 2ϩ level required for activating SOICR, the activation threshold, and the ER Ca 2ϩ depletion required for terminating SOICR, the termination threshold ( Fig. 2A and "Experimental Procedures"). Perfusion with the RyR2 inhibitor tetracaine and then the agonist caffeine established maximum (F max ) and minimum (F min ) ER Ca 2ϩ content. The ER capacity for storing Ca 2ϩ , the store capacity, was calculated as F max Ϫ F min .
In the SOICR experiments, the CPVT-linked CaM mutation, N53I, lowered the SOICR activation threshold (5%) relative to CaM WT, and so did the CPVT-and LQTS-linked mutation CaM N97S (4%) (Fig. 2, B-D, and Table 2). Similarly, the two LQTS-linked CaM mutations, D95V and D129G, also lowered the SOICR activation threshold by 6 and 4% (Fig. 3, B-D, and Table 2), respectively. The effects on the activation thresholds are modest, however, in cardiomyocytes SR Ca 2ϩ release increases with increasing SR luminal Ca 2ϩ concentrations in a steep and nonlinear fashion (40). Hence, even modest effects of CaM mutations on the response of RyR2 to SR luminal Ca 2ϩ may have major impacts on SR Ca 2ϩ release.
More strikingly, all of these CaM mutations markedly affected the ER Ca 2ϩ level at which SOICR terminated. The CaM mutation N53I decreased the termination threshold by 21% relative to CaM WT (Fig. 2, B and E), and, likewise, the CaM mutations N97S, D95V and D129G also decreased the termination threshold by 30, 33, and 28% (Figs. 2, C and E, and 3, B, C, and E), respectively.
The amount of Ca 2ϩ released from the ER during SOICR was calculated by subtracting the termination threshold from the activation threshold. As shown in Figs. 2 (B-E) and 3 (B-E), the disease-causing CaM mutations decreased the termination thresholds proportionally more than the activation thresholds and hence increased ER fractional Ca 2ϩ release (Figs. 2F and 3F). CaM N53I increased the ER fractional Ca 2ϩ release by 20% relative to the CaM WT, and CaM D95V, D129G, and N97S increased this release by 38, 38, and 34%, respectively (Table 2). Interestingly, the effect of CaM-N53I, located in the N-domain of CaM, on the ER fractional release (20%) was significantly less than that of the three CaM mutations located in the C-domain (average 33%), (p Ͻ 0.001 for each multiple comparison).
Notably, none of these CaM mutations affected the SOICR activation or termination thresholds in HEK293 cells expressing a RyR2 mutant with a deletion of the CaMBD (Fig. 4, A-D). This observation demonstrates that the CaMBD in RyR2, and not secondary effects of CaM overexpression (e.g. altered Ca 2ϩ buffering), mediate the observed effects (Fig. 4, C-E, all individual data sets not shown). Furthermore, the deletion of the CaMBD in RyR2 increased the ER fractional Ca 2ϩ release by 65% relative to that of WT RyR2 (Table 2), whereas the maximum increase in the ER fractional Ca 2ϩ release caused by the CaM mutations was less than 40% (Figs. 2F and 3F and Table 2). It is noteworthy that this comparison clearly shows that the arrhythmogenic CaM mutations did not suppress CaM-mediated inhibition of Ca 2ϩ release to the same extent as ablating the interaction between CaM and RyR2 by deleting the CaMBD in RyR2. In other words, the arrhythmogenic CaMs did interact with and inhibit RyR2, only not to the same extent as the CaM WT.
Ca 2ϩ homeostasis in HEK293 cells differs from that in cardiomyocytes. The [Ca 2ϩ ] cyt in HEK293 cells before RyR2 Ca 2ϩ release (60 nM) is similar to that of diastolic cardiomyocytes (100 nM), but ER Ca 2ϩ release is unlikely to increase [Ca 2ϩ ] cyt to the same peak levels as within the cardiomyocyte dyadic clefts during systole (Ͼ100 M) (41,42). High Ca 2ϩ concentrations mitigate the effects of mutations in the CaM C-domain (see below), and therefore their effect on RyR2 Ca 2ϩ release termination may be less at peak systole [Ca 2ϩ ] cyt in cardiomyocytes than in the HEK293 cells. However, the CaM mutations will on average lead to a suppressed inhibition of RyR2 by CaM and consequently increase the fractional Ca 2ϩ release by RyR2 in vivo (32). No significant differences in the HEK293 cell ER Ca 2ϩ store capacities were observed (Figs. 2-4), and Ca 2ϩ release in cardiomyocytes terminates at ϳ60% of diastole [Ca 2ϩ ] SR , similar to the RyR2 termination thresholds in HEK293 cells (Table  2), which supports similar ER and SR luminal [Ca 2ϩ ] free (7,8).
Taken together, each of the CaM mutations N53I, N97S, D95V, and D129G conferred a slightly increased propensity for SOICR and reduced the capability for terminating SOICR. The difference in termination threshold and ER fractional Ca 2ϩ release between CaM N53I and the other CaM mutations may support a mechanistically distinct effect of the CaM N53I mutation.

Arrhythmogenic Calmodulin Mutations Affect Binding of Ca 2ϩ to Calmodulin in the Presence of the RyR2 CaM Binding
Domain-Regulation of RyR2 by CaM critically depends on (a) CaM binding to the CaMBD in RyR2 and (b) Ca 2ϩ binding to CaM (43,44). Furthermore, this regulation of RyR2 by CaM is dependent on the characteristics of binding of Ca 2ϩ to each of the CaM domains (11,14,16,21). Hence, we investigated whether the mutations would affect Ca 2ϩ binding to CaM in the presence of a RyR2(Arg 3581 -Leu 3611 ) peptide corresponding to part of the CaMBD in RyR2. Ca 2ϩ binding was investigated by monitoring changes in CaM protein fluorescence (Phe for the N-domain and Tyr for the C-domain) as Ca 2ϩ was titrated to the CaM-peptide complex (34). Although available from a previous study, Ca 2ϩ binding to CaM in the absence of RyR2(Arg 3581 -Leu 3611 ) was measured for D95V and D129G as reported for CaM N53I and N97S to ensure the reliability of comparisons made here (1,23). Average [Ca 2ϩ ] cyt varies between ϳ0.1 and 1 M in cardiomyocytes during each heartbeat, but there is a large spatial heterogeneity of systolic [Ca 2ϩ ] cyt . For example systolic [Ca 2ϩ ] cyt can exceed 100 ⌴ in the dyadic cleft in vicinity of RyR2 channels (41,45). Thus, a wide range of Ca 2ϩ concentrations (1 nM to 2 m⌴) was used in our Ca 2ϩ titration experiments with RyR2(Arg 3581 -Leu 3611 ).

TABLE 2 Quantified change in Ca 2؉ release properties for RyR2 channels expressed in HEK293 cells transfected with CaM
Numbers in parentheses indicate the percentages of statistically significant change in the measured properties for RyR2 Ca 2ϩ release relative to RyR2 expressing cells transfected with CaM WT (p Ͻ 0.01, one-way analysis of variance). Transfections of RyR2 ⌬CaMBD expressing cells with CaM variants did not change the Ca 2ϩ release properties (Fig. 4, A-E). For comparison to RyR2 WT expressing cells, an average for all RyR2 ⌬CaMBD experiments was calculated (Average RyR2 ⌬CaMBD). The presence of the RyR2(Arg 3581 -Leu 3611 ) increased the apparent Ca 2ϩ affinity of the CaM N-domain and CaM C-domain ϳ20and ϳ80-fold, respectively ( Fig. 5A and Table 3). These increases clearly demonstrate the thermodynamic coupling between the Ca 2ϩ -CaM and the CaM-RyR2(Arg 3581 -Leu 3611 ) binding events (  C-domain of CaM in complex with RyR2 is inherently Ca 2ϩloaded at diastolic levels of [Ca 2ϩ ] cyt . Conversely, the N-domain of CaM in the same protein is poised for sensing increases in [Ca 2ϩ ] cyt . Hence, the N-domain but not the C-domain will switch between apo and Ca 2ϩ -loaded states during diastole to systole cycles and alter its interactions with RyR2.

CaM variant Activation Termination Fractional release
An additional observation was made during the titrations of the CaM-peptide complex with Ca 2ϩ . During the titrations, we also monitored the change in RyR2(Arg 3581 -Leu 3611 ) Trp fluorescence (data not shown), which coincided with the change in Tyr fluorescence from the CaM C-domain, but not with the change in Phe fluorescence from the CaM N-domain. This strongly supports that the C-domain of CaM WT binds around RyR2 Trp 3587 and its Ca 2ϩ binding induces a structural shift of the C-domain position on RyR2(Arg 3581 -Leu 3611 ) as is also hinted by previous studies (18,19,22).
The CPVT-and/or LQTS-linked CaM mutations differentially perturbed this domain-specific interaction of Ca 2ϩ with CaM. The CPVT-linked CaM-N53I mutation slightly decreased (appK D bound 1.2 M) the Ca 2ϩ affinity of the N-domain in the presence of RyR2(Arg 3581 -Leu 3611 ) and showed no measurable effect on Ca 2ϩ binding to the C-domain ( Fig. 5B and Table 3). Conversely, the CaM-N97S mutation linked to both CPVT-and LQTS, had no measurable effect on Ca 2ϩ binding to the N-domain, but markedly decreased (appK D bound 0.15 M) the Ca 2ϩ affinity of the C-domain (Fig. 5C and Table 3).
The LQTS-linked CaM mutations, D95V and D129G, both increased (appK D bound 0.48 and 0.22 M) the Ca 2ϩ affinity of the N-domain in the presence of RyR2(Arg 3581 -Leu 3611 ) and both also markedly decreased the Ca 2ϩ affinity of the C-domain (appK D bound 0.14 and 4 M) (Fig. 4, D and E, and Table 3). The effect of the D129G mutation was so marked that the Ca 2ϩ affinity of the C-domain was weaker than that of the N-domain. This suggests that the N-domain of CaM D129G may bind the CaMBD in RyR2 at a lower Ca 2ϩ concentration than the C-domain. Our data support the notion that mutations in the CaM C-domain (D95V, N97S, and D129G) all markedly perturb the interaction between Ca 2ϩ and CaM bound to the CaMBD in RyR2. Because CaM will bind to the CaMBD even in the absence of Ca 2ϩ , the reduction in C-domain Ca 2ϩ affinity for the mutants may result in binding to the CaMBD in RyR2 in a partially saturated or even Ca 2ϩ -unbound form (18). The small increase in the Ca 2ϩ affinity of the N-domain in the presence of RyR2(Arg 3581 -Leu 3611 ), as conferred by the LQTS-linked CaM mutations D95V and D129G, suggests that the N-domain CaM mutations will still respond to an increase in [Ca 2ϩ ] cyt from diastole to systole. In contrast to the marked perturbations of Ca 2ϩ binding by the CaM C-domain mutations, the CPVTlinked CaM N53I mutation showed only a small decrease in the CaM N-domain affinity for Ca 2ϩ in the presence of RyR2(Arg 3581 -Leu 3611 ), yet a substantial effect on the interaction with RyR2 is expected because there is a substantial effect on SOICR.

Discussion
The domain-specific affinities of CaM for Ca 2ϩ in the presence of the RyR2(Arg 3581 -Leu 3611 ) offer novel insight into the distinct roles of the two domains of CaM in regulating the Ca 2ϩ release from RyR2 (Fig. 6, A-C). Our results support an interaction between CaM and the CaMBD in RyR2 where the Ca 2ϩsaturated C-domain of CaM is constitutively tethered around RyR2 Trp 3587 even at the low diastole level of 0.1 M Ca 2ϩ free (Fig. 6B) and thus not releasing Ca 2ϩ during cardiac contraction cycles (Fig. 6, B and C). This implies a critical role for Ca 2ϩ binding to the C-domain in inhibiting Ca 2ϩ release from RyR2. In line with this view, Tian et al. (14) showed that a CaM double mutant (D93A/D129A) with loss of C-domain Ca 2ϩ binding is defective in regulating the termination of RyR2 SOICR. Both the work of Tian et al. and this study also demonstrate that the interaction between the Ca 2ϩ -unbound CaM C-domain with the CaMBD in RyR2 inhibits Ca 2ϩ release (Fig. 6A), but not to nearly the same extent as the Ca 2ϩ -saturated C-domain. Thus, the novel observation in this study is that the CaM C-domain under physiologically relevant conditions binds to the RyR2 CaMBD in a Ca 2ϩ -saturated form.
The pivotal role of the tethered CaM C-domain for inhibiting RyR2 Ca 2ϩ release is consistent with studies showing that Ͼ70% of RyR2 channels in intact cardiomyocytes have CaM bound and that dissociation of CaM leads to excessive Ca 2ϩ release through RyR2 channels (13,43,49,50). In addition, several studies show that single mutations in the binding site for the CaM C-domain in the CaMBD of RyR2 (W3587A or L3591D) decrease the affinity for CaM to RyR2 at 0.4 and 1 M [Ca 2ϩ ] free , and that this decrease also leads to impaired inhibition of RyR2-mediated Ca 2ϩ release (14,16,44,51).
With the region of the RyR2 CaMBD around Trp 3587 occupied by the C-domain of CaM, the binding site for the N-domain would be adjacent and likely covering RyR2 Phe 3603 , similar to the structure of CaM binding to the CaMBD in RyR1   (52,53). We further propose that the Ca 2ϩunbound N-domain interacts with regions of RyR2 outside the canonical CaMBD and then binds in the vicinity of Phe 3603 upon cardiomyocyte excitation (Fig. 6, B and C) (18,20). This proposed interaction would be consistent with the CaMBD in RyR2 not being the only RyR2 region reported to intact with CaM (11,18,54,55). Moreover, at low [Ca 2ϩ ] free , the N-domain may interact with a region in the skeletal muscle SR Ca 2ϩ release channel, ryanodine receptor (RyR1) that is noncontiguous with the RyR1 CaMBD (18 -20, 56). On the other hand, cryo-electron microscopy models of RyR2 channels indicate that CaM does not change position on RyR2 when comparing samples prepared in low and high Ca 2ϩ (55,57). However, because activation of the RyR2 channel involves substantial movements in the channel structure, it cannot be ruled out that the N-domain may bind different RyR2 regions in the open and closed channel, respectively, without CaM markedly shifting position.
In summary, we propose that the interaction between CaM and RyR2 involves constitutive tethering of the Ca 2ϩ -saturated CaM C-domain to the CaMBD in RyR2, which serves to inhibit Ca 2ϩ release by stabilizing the RyR2 channel closed state throughout Ca 2ϩ release (14,21,43,58). In our model, N-do-main binding to RyR2 is a Ca 2ϩ sensing step that responds to increased [Ca 2ϩ ] cyt upon cardiomyocyte excitation. However, determining the exact function of Ca 2ϩ binding to CaM N-domain in regulating RyR2 Ca 2ϩ release will require further investigation.
Ca 2ϩ binding to the CaM N-domain does not appear to affect the SOICR activation and termination thresholds of RyR2 Ca 2ϩ release, because no effect was detected in experiments using a CaM mutant with dramatically decreased N-domain Ca 2ϩ affinity (CaM D20A/D56A) (14). Also, in the current study only minor effects of the CaM N53I mutation on the binding of Ca 2ϩ to the N-domain in the presence of RyR2(Arg 3581 -Leu 3611 ) were observed (see below). The binding of Ca 2ϩ to the N-domain of CaM, when CaM is tethered to RyR2, may instead be part of inhibiting Ca 2ϩ -induced Ca 2ϩ release during the high systolic [Ca 2ϩ ] cyt . Alternatively, Ca 2ϩ binding to the N-domain could be part of the triggering mechanism for the activation of RyR2 channel Ca 2ϩ release during Ca 2ϩ -induced Ca 2ϩ release. Both of these functions would be in combination with endogenous cytosolic and SR luminal Ca 2ϩ sensors in RyR2 (52,59). An inhibiting or facilitating function of CaM on Ca 2ϩ -induced Ca 2ϩ release would not be detected in the SOICR experiments performed in this study (6,14).
Our results support the notion that arrhythmogenic CaM mutations adversely affect the native interaction between CaM and RyR2. We show that the C-domain mutations (D95V, N97S, and D129G) perturb the interactions between CaM, Ca 2ϩ and the CaMBD in RyR2. Notably, this perturbation occurs within physiologic relevant Ca 2ϩ concentrations. We suggest that at diastole [Ca 2ϩ ] cyt , CaM with either of these C-domain mutations will bind to the RyR2 CaMBD with the CaM C-domain in a partially saturated or Ca 2ϩ -unbound form as opposed to the native Ca 2ϩ -saturated state (transition from Figs. 6B to 6A). This in turn leads to an insufficient inhibition of the RyR2 Ca 2ϩ release during SOICR in intact cells, similar to the aberrant regulation of RyR2 by CaM 1234 and CaM D93A/D129G mutants (14,21). In vivo CaM D95V, N97S, and D129G are likely to still tether to RyR2 because each mutated CaM retains a significant affinity for intact RyR2 channels, despite reduced affinity for Ca 2ϩ (31).
Interestingly, the N53I mutation in the N-domain of CaM adversely affected the regulation of RyR2-mediated Ca 2ϩ release during SOICR, without any changes to the affinity of the CaM N53I C-domain for binding Ca 2ϩ in the presence of RyR2(Arg 3581 -Leu 3611 ). This strongly suggests that the integrity of the CaM N-domain is important in inhibiting RyR2 Ca 2ϩ release but not dependent on Ca 2ϩ binding to the N-domain, as discussed above (14). In strong support of this hypothesis is that CaM D20A/D56A, with markedly decreased N-domain Ca 2ϩ affinity, does not affect activation or termination thresholds for RyR2-mediated SOICR, whereas CaM N53I does. A likely explanation is that the N53I mutation affects an interaction of the Ca 2ϩ -unbound N-domain of CaM with a region of RyR2 that is outside the CaMBD. The significantly different impact of the CaM N53I mutation on RyR2 Ca 2ϩ release, compared with the C-domain mutations, also supports this notion. Furthermore, we previously found divergent effects of the N53I and N97S mutations on protein properties and the binding of Ca 2ϩ to CaM, which supports different molecular disease mechanisms for the two mutations (1).
Overexpression of CaM in HEK293 cells could theoretically lead to buffering of [Ca 2ϩ ] cyt , which in turn may affect RyR2 Ca 2ϩ release through mechanisms independent of the interaction between CaM and RyR2, and lead to differences between the CaM mutations caused by differences in Ca 2ϩ buffering capacities. This was, however, not the case in the SOICR experiments as judged from three observations: (a) CaM N53I had marked effects on RyR2-mediated Ca 2ϩ release, although it displayed Ca 2ϩ binding properties highly similar to those of CaM WT both in the RyR2-bound and free form (Fig. 2, A and B, and Table 3); (b) CaM WT and CaM D129G had no effect on Ca 2ϩ release from the RyR2 ⌬CaMBD mutant, although they exhibited widely different Ca 2ϩ binding properties (Fig. 4, A and B) (1,14); and (c) because the Ca 2ϩ affinity (appK D free ) of the CaM C-domain not bound to a protein target is ϳ2.5 M, the buffering capacity of free CaM is highly limited at the ϳ60 nM [Ca 2ϩ ] cyt in HEK293 cells.
The combined results of this study show that both CPVT-and LQTS-linked CaM mutations can lead to excessive Ca 2ϩ release from RyR2 channels, primarily from insufficient termination of RyR2-mediated Ca 2ϩ release. Also, the same CaM mutations lower the activation thresholds for SOICR, which would increase the propensity for SOICR in vivo during conditions with increased SR Ca 2ϩ load. Excessive Ca 2ϩ release and leaky RyR2 channels are a well documented hallmark of CPVT-linked RyR2 mutations. It follows that dysregulation of RyR2 likely underlies the disease mechanisms of CaM N53I and N97S mutation-associated CPVT (38). Taken together, we propose that the regulation of RyR2 Ca 2ϩ release is highly sensitive to CaM and that aberrant regulation of RyR2 may be a common component of both CPVT and LQTS arrhythmias caused by CaM mutations.