The Arrhythmogenic Calmodulin p.Phe142Leu Mutation Impairs C-domain Ca2+ Binding but Not Calmodulin-dependent Inhibition of the Cardiac Ryanodine Receptor

A number of point mutations in the intracellular Ca2+-sensing protein calmodulin (CaM) are arrhythmogenic, yet their underlying mechanisms are not clear. These mutations generally decrease Ca2+ binding to CaM and impair inhibition of CaM-regulated Ca2+ channels like the cardiac Ca2+ release channel (ryanodine receptor, RyR2), and it appears that attenuated CaM Ca2+ binding correlates with impaired CaM-dependent RyR2 inhibition. Here, we investigated the RyR2 inhibitory action of the CaM p.Phe142Leu mutation (F142L; numbered including the start-Met), which markedly reduces CaM Ca2+ binding. Surprisingly, CaM-F142L had little to no aberrant effect on RyR2-mediated store overload-induced Ca2+ release in HEK293 cells compared with CaM-WT. Furthermore, CaM-F142L enhanced CaM-dependent RyR2 inhibition at the single channel level compared with CaM-WT. This is in stark contrast to the actions of arrhythmogenic CaM mutations N54I, D96V, N98S, and D130G, which all diminish CaM-dependent RyR2 inhibition. Thermodynamic analysis showed that apoCaM-F142L converts an endothermal interaction between CaM and the CaM-binding domain (CaMBD) of RyR2 into an exothermal one. Moreover, NMR spectra revealed that the CaM-F142L-CaMBD interaction is structurally different from that of CaM-WT at low Ca2+. These data indicate a distinct interaction between CaM-F142L and the RyR2 CaMBD, which may explain the stronger CaM-dependent RyR2 inhibition by CaM-F142L, despite its reduced Ca2+ binding. Collectively, these results add to our understanding of CaM-dependent regulation of RyR2 as well as the mechanistic effects of arrhythmogenic CaM mutations. The unique properties of the CaM-F142L mutation may provide novel clues on how to suppress excessive RyR2 Ca2+ release by manipulating the CaM-RyR2 interaction.

Point mutations in one of the three extremely conserved calmodulin (CaM) 2 -encoding genes, CALM1-3, result in lifethreatening ventricular arrhythmias likely due to altered CaMregulation of the ion channels that govern cardiac excitationcontraction (1)(2)(3)(4)(5)(6)(7)(8). The CaM-N54I and -N98S mutations (numbering includes start-Met) were identified in individuals with catecholaminergic polymorphic ventricular tachycardia (CPVT). The CaM-D96V, -D130G, and -F142L mutations were found in individuals with long QT syndrome (LQTS) (1,2). Interestingly, the CaM-N98S mutant also apparently causes LQTS or a mixed phenotype, depending on the genetic background (4). Not only do these various mutations impose different cardiac arrhythmias, but it appears that their disease mechanisms differ at the molecular level for each CaM target, even within the same arrhythmia type (6 -11). One such CaM target is the cardiac Ca 2ϩ release channel/ryanodine receptor (RyR2). RyR2 mediates Ca 2 release from the sarcoplasmic reticulum (SR) in cardiomyocytes (12,13). The RyR2 protein forms homotetrameric channels in the SR membrane with a large cytosolic domain that interacts with numerous proteins and ligands, which regulate RyR2 Ca 2ϩ release (12,13). During cardiac excitation-contraction coupling, RyR2 channels are activated by

Results
The CaM-F142L Mutation Slightly Decreases the Termination Threshold for Store Overload-induced Ca 2ϩ Release-To test whether the CaM-F142L mutation affects the regulation of RyR2 during SOICR, we transfected RyR2-expressing HEK293 cells with CaM-WT or -F142L and then monitored the endoplasmic reticulum (ER) Ca 2ϩ concentration using the D1ER Ca 2ϩ probe (27). Perfusion of the transfected cells with increasing extracellular Ca 2ϩ concentrations induced SOICR in the form of spontaneous ER Ca 2ϩ oscillations ( Fig. 2) (27,30). The oscillating D1ER signal was then used to determine the ER Ca 2ϩ level at which SOICR occurred (activation threshold) and the ER Ca 2ϩ depletion at which SOICR ended (termination threshold; Fig. 2A and "Experimental Procedures"). The difference between the activation and termination thresholds was specified as the fractional ER Ca 2ϩ release. Fig. 2, B and D, shows that expression of the CaM-F142L mutant decreased the termination threshold by 5% compared with the control with only endogenous CaM-WT (F142L 55% versus control 60%, p Ͻ 0.001). This in turn increased the fractional ER Ca 2ϩ release during Ca 2ϩ release oscillations by 6% (Fig. 2E, F142L 38% versus control 32%, p Ͻ 0.001). Note that the percentages listed  Sites of arrhythmogenic CaM mutations are highlighted as green stick representations (non-mutated residues) and Ca 2ϩ ions as black spheres. RyR1 CaMBD residues (Trp-3620, Leu-3624, and Phe-3636) corresponding to RyR2 CaMBD residues Trp-3587, Leu-3591, and Phe-3603 are highlighted as magenta stick representations.

CaM-F142L Impairs Ca 2؉ Binding but Not RyR2 Inhibition
here refer to the unit for ER Ca 2ϩ load and not the relative effects of the CaM variants. On the other hand, expression of CaM-WT increased the termination threshold by 4% (WT 64% versus control 60%, p Ͻ 0.01) and minutely reduced the fractional ER Ca 2ϩ release (WT 30% versus control 32%, p Ϸ 0.1), although the latter was not statistically significant (Fig. 2, A and  E). Neither CaM-WT nor CaM-F142L expression affected the activation threshold (Fig. 2, A-C). The control and CaM-WT results here are highly consistent with those reported previously (27). Taken together, CaM-F142L slightly reduced the SOICR termination threshold or, in other words, had a slightly less inhibitory action on RyR2-mediated Ca 2ϩ release compared with CaM-WT. For comparison, we previously reported equivalent experiments showing that CPVT-causing (N54I), CPVT-and LQTS-causing (N98S), and LQTS-causing CaM mutations (D96V and D130G) all dramatically alter the Ca 2ϩ release termination threshold (8). Specifically, the CaM-N54I, -D96V, -N98S, and -D130G mutations reduced the termination threshold by 12, 17, 18, and 16%, respectively. These variants also minimally, but significantly, reduced the activation threshold (ϳ5%) (8). Thus, the action of the CaM-F142L mutant on RyR2 function appeared very different from that of CaM-N54I, -D96V, -N98S, and -D130G.
The CaM-F142L Mutation Enhances Inhibition of Single RyR2 Channels-Next, we tested the action of the CaM mutations on single RyR2 channels incorporated into lipid bilayers. Luminal [Ca 2ϩ ] free was kept at 1 mM, and the cytosolic [Ca 2ϩ ] free was set at 10 M. The cytosolic solution also contained 1 mM [Mg 2ϩ ] free and 5 mM ATP to approximate the levels present in cells. The 10 M cytosolic [Ca 2ϩ ] free approximates the level that RyR2 encounter during systole (see "Experimental Procedures"). Single RyR2 channel function was measured before and after the addition of 1 M CaM-WT, -F142L, -N54I, -D96V, -N98S, or -D130G to the cytosolic solution (Fig.  3). Control (Ctrl) recordings in the absence of CaM were also included. The addition of CaM-WT significantly lowered the RyR2 open probability (P O ) from 0.47 to 0.35 compared with control, consistent with CaM-WT inhibition of RyR2-mediated Ca 2ϩ release (Fig. 3D). CaM-WT reduced the mean open time (MOT) and increased the mean closed time (MCT), although these changes individually were not statistically significant (p ϭ 0.18 and 0.05 against Ctrl) (Fig. 3, E and F). In contrast, the C-domain mutations, CaM-D96V, -N98S, and -D130G, did not lower RyR2 P O compared with the control (P O 0.54, 0.46, and 0.52, respectively). The CaM-D96V, -N98S, and -D130G mutations all significantly decreased the RyR2 MCT compared with both CaM-WT and the control. The CaM-N98S mutation significantly decreased MOT compared with the control. These results suggest that: 1) these CaM mutations do interact with RyR2 as also shown in previous studies (7,8,10); and 2) CaM mutations D96V, N98S, and D130G, in contrast to CaM-WT, do not appear to inhibit RyR2 function (but this is due to almost equal decreases in both MOT and MCT compared with no CaM present). In contrast, the CaM N-domain N54I mutation did not significantly affect RyR2 function (P O , MCT or MOT) under these experimental conditions, suggesting that aberrant RyR2 regulation by CaM-N54I is mechanistically distinct from that caused by the CaM C-domain mutations (8,17).
Strikingly, the C-domain CaM-F142L mutation had an opposite action on single RyR2 channel function. CaM-F142L significantly lowered the RyR2 P O , even more so than CaM-WT (P O 0.18 versus 0.35) (Fig. 3D). The CaM-F142L mutation caused a significant MOT decrease compared with both CaM-WT and the control, yet appeared to have approximately the same effect on MCT as CaM-WT (22 versus 24 ms). Hence, the CaM-F142L mutation resulted in a stronger RyR2 inhibitory action (compared with that of CaM-WT). In other words, the CaM F142L mutation generates a GoF defect with regard to RyR2 inhibition, which is in stark contrast to CaM-D96V, -N98S, and -D130G, none of which inhibited RyR2 function. Hence, CaM-WT, CaM-F142L, and the other C-domain mutations (D96V, N98S, and D130G), as well as the CaM N-domain N54I mutation tested here, have distinctive actions on RyR2 modulation.
The   mutation causes a loss of function (LoF) in terms of Ca 2ϩ binding to the CaM C-domain when complexed with its RyR2 target, and this LoF is at least as severe as for CaM-D96V or -N98S. Thermodynamically Distinct Interactions of CaM-F142L and CaM-WT with the RyR2 CaMBD-To further probe the interaction between the CaM-F142L mutant and the RyR2 CaMBD, we followed the titration of CaMBD (i.e. the RyR2(R3581-P3607) peptide) with CaM variants under Ca 2ϩ -free (apoCaM) and saturating Ca 2ϩ conditions (CaCaM) using ITC (Fig. 5). This measurement compares the differences in the CaMBD interaction aside from those caused by differences in CaM Ca 2ϩ binding affinity. The binding of apo-or CaCaM-WT to the CaMBD peptide represent two thermodynamically distinct reactions. The apoCaM-WT/CaMBD interaction is comparatively low affinity (M K D ) and is entropy-driven (⌬H o Ͼ 0, ϪT⅐⌬S o Ͻ 0). The CaCaM-WT/CaMBD interaction is high affinity (nM K D ) and is enthalpy- The ITC measurements showed a fundamental difference in the thermodynamics of apoCaM-F142L binding to the CaMBD compared with the apoCaM-WT (Fig. 5, A-C). The binding reaction between apoCaM-F142L and the CaMBD peptide was exothermic (⌬H o Ͻ 0) in marked contrast to the endothermic (⌬H o Ͼ 0) interaction between apoCaM-WT and the CaMBD. Moreover, titration curve analysis showed that apoCaM-F142L binding affinity for the CaMBD was ϳ3-fold greater compared with that of apoCaM-WT (K D 9 versus 28 M) (Fig. 6A).

CaM-F142L Impairs Ca 2؉ Binding but Not RyR2 Inhibition
ApoCaM-F142L binding also displayed a small negative ⌬H o (Ϫ1.8 kJ/mol)) compared with the larger, positive ⌬H o (11.8 kJ/mol) for apoCaM-WT binding (Fig. 6B). A comparison of the fitted ⌬H o values to the ϪT⅐⌬S o values (Ϫ38 and Ϫ27 kJ/mol for apoCaM-WT and -F142L) indicated that the binding of both apoCaM-WT and -F142L to the CaMBD peptide remained governed by entropy (ϪT⅐⌬S o ) relative to enthalpy (⌬H o ) (Fig. 6C). The change in ⌬H o conferred by the CaM-F142L mutation translated into a significant 11% decrease (Ϫ29 versus Ϫ26 kJ/mol) in ⌬G o for the apoCaM-F142L interaction with the CaMDB peptide (Fig. 6D). Generally, an enthalpydriven reaction is indicative of specific molecular bonding, whereas an entropy-driven reaction indicates hydrophobic interactions and solvent effects (32). Thus, these ITC results indicated that the CaM-F142L mutation transformed the apoCaM interaction from mainly entropy-driven to one more dominated by molecular bonds. Under saturating Ca 2ϩ conditions, the interaction between CaCaM-F142L and the CaMBD peptide was indistinguishable from that for the CaCaM-WT (Figs. 5, C and D, and 6). Interestingly, the increase in affinity comparing Ca 2ϩ -free to saturating Ca 2ϩ conditions was still 3 orders of magnitude (K D 8 M versus 14 nM) for CaM-F142L, attributable to an increased ⌬H o contribution under the CaCaM condition. Thus, the thermodynamic difference between apo and CaCaM binding to the CaMBD peptide remained similar for CaM-WT and -F142L.
The interactions of CaM-D96V and -N98S with the RyR2 CaMBD peptide were also investigated using ITC, and for both Ca 2ϩ conditions neither of the thermograms (not shown) was visibly different from those for the CaM-WT. However, detailed titration curve analysis indicated a slightly decreased affinity of CaM-D96V for CaMBD binding with and without Ca 2ϩ present (Figs. 5, C and F, and 6A). Small changes to ⌬H o were also detected for apoCaM-D96V and -N98S and for CaCaM-D96V and -N98S (Fig. 6B). Interestingly, the minute effects observed for CaM-D96V and -N98S were all significantly distinct from those observed for CaM-F142L under both Ca 2ϩ conditions. Thus, CaM-F142L generally showed CaMBD binding properties different from not only the CaM-WT but also from the other arrhythmogenic CaM mutations, D96V and N98S.
Motivated by the results from the ITC experiment, we used 2D NMR ( 15 N-HSQC) to measure the effect of Ca 2ϩ on the chemical shifts from a CaMBD peptide with 15 N-labeled Val-3586 and Phe-3603 in a complex with CaM-WT, -N98S, or -F142L. The chemical shifts of these labeled residues depend on their immediate structural surroundings (i.e. the binding of the CaM C-and N-domain, respectively). The peaks corresponding to Val-3586 and Phe-3603 were easily discernable under apoCaM conditions (Fig. 7A). Unfortunately, their chemical shifts overlapped under saturating Ca 2ϩ conditions (Fig. 7B). Nonetheless, a clear difference under apoCaM conditions was observed for CaM-CaMBD complexes containing CaM-F142L compared with CaM-WT or -N98S. ApoCaM-F142L displayed a higher chemical shift for the Val-3586 H N but showed no differences for the Phe-3603 H N or N. The latter observation may reflect that the N-domains of apoCaM-F142L and -WT do not bind to the CaMBD peptide. Under saturating Ca 2ϩ conditions, no differences in the spectra were observed, albeit the addition of Ca 2ϩ clearly affected the structural surroundings of the labeled residues (i.e. chemical shifts for both Val-3586 and Phe-3603 changed markedly). Opposite CaM-F142L, the spectra recorded using CaM-N98S were identical to those for CaM-WT, with and without Ca 2ϩ (Fig. 7, A and B). Taken together, these results support the notion that the apoCaM-F142L C-domain binds to the RyR2 CaMBD peptide close to the Val-3586 residue (next to Trp-3587) in a unique conformation structurally distinct from that of the apoCaM-WT and -N98S. Further, Ca 2ϩ binding to CaM-F142L changes this conformation into one that is indistinguishable from that of the CaCaM-WT and -N98S.

Discussion
CaM is a constitutive Ca 2ϩ sensor of the RyR2 macromolecular complex, where it inhibits RyR2 Ca 2ϩ release in a [Ca 2ϩ ] cyt dependent, allosteric manner (3,13,34). This inhibition is critical for maintaining a low RyR2 activity at diastole (i.e. at low [Ca 2ϩ ] cyt ) and also for a sufficient termination of the Ca 2ϩ release during cardiac excitation (i.e. as [Ca 2ϩ ] cyt increases) (  . This is similar to what was observed for the CaM-N98S and -D96V mutants (appK D 10 and 31 M), albeit less than for CaM-D130G (appK D 84 M) (2,17). Here, we found that this reduced affinity for the CaM-F142L C-domain Ca 2ϩ binding was retained in the presence of the RyR2 CaMBD peptide, even to the extent that CaM-F142L displayed a lower affinity than CaM-D96V (appK D 0.32 versus 0.14 M) (Fig. 4). Thus, we expected that in HEK293 cells where [Ca 2ϩ ] cyt oscillates between ϳ0.1 and 2 M, CaM-F142L would abnormally regulate RyR2 function (compared with CaM-WT) to a similar extent as CaM-N98S and -D96V (Fig. 2) (37, 38). Curiously, the action of CaM-F142L on RyR2 function in the HEK293 SOICR assay was relatively benign and distinctly different from the actions of the CaM-N54I, -D96V, -N98S, and -D130G mutants (8). Moreover, these other CaM mutants were expressed using low expressing plasmids, resulting in a CaM ratio of ϳ1.4 relative to endogenous CaM (8). The expression plasmid used in this study, and also in Tian et al. (27), results in a CaM ratio of ϳ4 relative to endogenous CaM as judged from Western blotting analysis (Fig. 8). This further supports that the decrease in RyR2 inhibition caused by CaM-F142L was strikingly less than the decreases caused by CaM-N54I, -D96V, -N98S, and -D130G. Lastly, we observed no difference between CaM-F142L and -WT when the low expressing plasmid was used (data not shown  (Fig. 3). In other words, CaM-F142L displayed a GoF action in that it increased the CaM-dependent RyR2 inhibition. How can the F142L mutation confer both a LoF in terms of Ca 2ϩ binding with seemingly little consequence to the RyR2 inhibitory action in cells and a robust RyR2 inhibitory action at the single channel level? Our ITC experiments hint at a potential explanation. ApoCaM-F142L showed increased affinity for binding the RyR2 CaMBD peptide and bound in a manner thermodynamically in between that of the apoCaM-WT and CaCaM-WT interactions (Figs. 5 and 6). The NMR HSQC spectra also support this view demonstrating that the apoCaM-F142L C-domain bound around Val-3586 in the CaMBD in a conformation that was structurally distinct from that of the apoCaM-WT. Based on our functional, biophysical, and structural results, we propose that the F142L mutation has two opposing actions on RyR2 regulation.
First, the F142L mutation enhances the CaM C-domain interactions with the RyR2 CaMBD, thus increasing RyR2 inhibition (GoF). Second, the CaM-F142L C-domain has impaired Ca 2ϩ binding (LoF), thus decreasing RyR2 inhibition. Under the comparatively low [Ca 2ϩ ] cyt conditions in HEK293 cells or the permeabilized cardiomyocytes (7), the effect of reduced CaM-F142L C-domain Ca 2ϩ binding would be partially offset by the enhanced CaM binding to the RyR2 CaMBD, explaining why little aberrant RyR2 regulation was observed. However, increasing Ca 2ϩ saturates the CaM-F142L C-domain (appK D : 0.32 M), in effect ablating any LoF action from the reduced C-domain Ca 2ϩ affinity. Accordingly, at high Ca 2ϩ concentrations only the GoF action would remain and may explain why CaM-F142L was a more potent inhibitor than the CaM-WT in our single RyR2 channel studies. The molecular basis for this GoF effect was not clear from these experiments, as no differences between CaCaM-WT and CaCaM-F142L binding to the RyR2 CaMBD peptide were detected. This implies perhaps that interactions between CaM and RyR2 not recapitulated in our biophysical experiments, e.g. RyR2 regions outside the CaMBD studied here, are responsible for the GoF action. Structurally delineating this GoF action may provide a molecular guide for how to manipulate the CaM-dependent RyR2 inhibition. Also, increasing this inhibition reduces SR Ca 2ϩ release and/or leak and thus represents a therapeutic approach for treating heart failure and arrhythmia (28, 31, 39 -41).

CaM-F142L Impairs Ca 2؉ Binding but Not RyR2 Inhibition
Aside from the novel insights into the effects of CaM-F142L on RyR2 regulation, our single RyR2 channel experiments support our previous finding that both CPVT-and LQTS-causing CaM mutations (N98S, D96V, and D130G) result in aberrant RyR2 regulation (8). Using different experimental conditions (sheep cardiomyocytes, 0.1 mM luminal [Ca 2ϩ ] free , 2 mM ATP cytosolic, and no Mg 2ϩ ), Hwang et al. (7) report that CaM-N54I and -N98S increased RyR2 P O at 0.1 and 1 M cytosolic [Ca 2ϩ ] free , whereas CaM-D96V did not. The apparent CaM-D96V discrepancy between our study and the Hwang et al. study (7) may be explained by differences in the experimental conditions. Our study shows that CaM mutations N54I, D96V, N98S, D130G, and F142L alter the RyR2 CaM regulation via distinct molecular mechanisms. The CaM-D96V, -N98S, and -D130G mutations diminish C-domain Ca 2ϩ binding and thereby the inhibitory interaction of CaM with the RyR2 CaMBD. The N-domain CaM-N54I mutation likely affects CaM-RyR2 interactions that are outside the canonical CaMBD and/or increases Ca 2ϩ binding kinetics (17). The CaM-F142L mutation diminishes C-domain Ca 2ϩ binding but also enhances CaM interaction with the RyR2 CaMBD and possibly other regions (7,8,10,17). Because LQTS-causing CaM-D96V, -N98S, and -D130G mutations diminish CaM-dependent RyR2 inhibition, increased RyR2 Ca 2ϩ release may contribute to the LQTS phenotypes in individuals with these CaM mutations. Generally, it is thought that spontaneous diastolic RyR2 Ca 2ϩ release causes CPVT, and augmented Ca 2ϩ influx through Ca V 1.2 channels causes LQTS (see Refs. 42 and 43 for details). Studies using recombinant Ca V 1.2 and patch clamping show that Ca 2ϩ -dependent inactivation of Ca V 1.2 is reduced when CaM-D96V, -N98S, -D130G and -F142L are present (compared with CaM-WT) (6,9).
Thus, some mechanistic overlap between CPVT and LQTS caused by CaM mutations likely exist, and how arrhythmogenic CaM mutations manifest probably depends on their relative effects on RyR2, Ca V 1.2, and other CaM-regulated targets. The best example of this is CaM-N98S, which affects both RyR2mediated Ca 2ϩ release and Ca V 1.2 Ca 2ϩ -dependent inactivation, likely contributing to both CPVT and LQTS. Another scenario is represented by the strictly CPVT-causing CaM-N54I mutation, which does not affect Ca V 1.2 Ca 2ϩ -dependent inactivation (1,4,6,44). In addition, some CaM C-domain mutations (N98S, D132E, and Q136P) are reported to cause a LQTS phenotype with some features of CPVT overlapping (4,44). ␤-Blockers are the common drug treatment for both CPVT and LQTS (42,43). Given the differential influence of CaM mutants on Ca V 1.2 and RyR2, drugs that preferentially alter RyR2 or Ca V 1.2 function might provide better treatments for CaM-mediated arrhythmias (42,43,45). Novel drugs affecting RyR2 and Ca V 1.2 are being sought (45)(46)(47), and as antiarrhythmic therapies advance, defining their differential action on Ca V 1.2 and RyR2 function will almost certainly become increasingly relevant.

Experimental Procedures
Plasmid Constructs-Plasmid constructs for the recombinant expression and purification of CaM (pMAL, New England Biolabs) or for overexpression of CaM in HEK293 cells (pcDNA3.1, Invitrogen) were prepared as described previously (8). Sanger sequencing verified the CaM-encoding inserts in all plasmids.
Endoplasmic Reticulum Luminal Ca 2ϩ Imaging of HEK293 Cells Expressing RyR2-Stable expression of murine RyR2, or a RyR2 variant with the CaMBD deleted (murine ⌬K3583-F3603), in HEK293 cells co-transfected with plasmids encoding CaM and the D1ER Ca 2ϩ probe was done as described previously (27). Briefly, D1ER FRET signals reflecting ER luminal [Ca 2ϩ ] free in individual cells were monitored by using an epifluorescent microscope (27,48). Each FRET signal trace was used to measure the Ca 2ϩ release properties of the RyR2 channels relative to the ER Ca 2ϩ store capacity: the activation and termination thresholds and their difference taken as the fractional Ca 2ϩ release. The ER Ca 2ϩ store capacities were calculated from the difference between maximum and minimum FRET signal (F max Ϫ F min ) (Fig. 2). The measured RyR2 Ca 2ϩ release properties were compared using one-way ANOVA with Tukey's multiple comparisons test for all possible combinations, with adjusted p Ͻ 0.05 taken as significant. The experiments included a control without plasmid expression of CaM.
Estimation of CaM-WT and RyR2 Expression Levels in HEK293 Cells-HEK293 cells were cultured as described above with CaM-WT overexpression from a low (8) or high expressing plasmid (this study and Ref. 27) and without overexpression (Ctrl). Overexpression plasmids differed in their Kozak sequences upstream of the CALM1 cDNA inserts. 40 g of total protein from cell lysates (protein assay, Bio-Rad) was subjected to SDS-PAGE (1 h at 20 A) in a gradient gel (Bio-Rad, Mini-Protean TGX 4 -20%) alongside a molecular weight marker (Bio-Rad, catalog No. 161-0309), and the electrophoretic separated proteins were blotted (0.5 h at 100 V, ϳ0.33 A) to a nitrocellulose membrane in Tris-glycine buffer (Bio-Rad) with 10‰ SDS. The membrane was transiently stained with Ponceau S and cut into three regions: RyR2 (ϳ500 kDa), ␤-actin (ϳ42 kDa), and CaM (ϳ16 kDa). The membrane pieces were blocked in PBS with 1% casein (Bio-Rad), washed in PBS, and then incubated overnight with different primary antibodies (Ab) against CaM (05-173, EMD Milipore), ␤-actin (A5316, Sigma), or RyR2 (MA3-925, Pierce) in PBS with 1‰ Ab, 20% fetal bovine serum (Sigma), and 17 mM NaN 3 . After another wash, the pieces were incubated for 0.5 h with a secondary Ab (in-house anti-mouse IgG conjugated to horseradish peroxidase) and then washed again. The amount of bound secondary Ab was detected using luminol reagent (detection reagent 1-2, Thermo Scientific) with the resulting chemiluminescence imaged (ImageQuant LAS 4000, GE Healthcare Life Sciences). As a measure of protein expression levels, the protein band area intensities were quantified in ImageJ, and the expression levels of RyR2 and CaM in the individual samples were normalized to that of ␤-actin (49) (Fig. 8, normalized area). Western blotting analysis was done in at least duplicate.
Bilayer Recordings of Single RyR2 Channels-Native SR microsomes isolated from rat ventricular muscle were incorporated into bilayers using a modification of the method described by Chamberlain et al. (47,50,51). Briefly, planar lipid bilayers (50 mg/ml in a 5:4:1 mixture of bovine brain phosphatidylethanolamine, -serine, and -choline in n-decane) were formed across a 100-m-diameter hole in a Teflon partition separating two compartments with cytosolic (114 mM Tris-HEPES, 5 mM ATP, 1 mM free Mg 2ϩ , 1 mM EGTA, and 10 M free Ca 2ϩ at pH 7.4) and luminal (cytosolic solution plus 200 mM Cs-HEPES and 1 mM free Ca 2ϩ ) recording solutions. Single RyR2 activity was measured before and 20 min after the addition of CaM variants (1 M) to the cytosolic solution. Recordings were made at room temperature (20 -22°C) with currents sampled at 50 s/point and filtered at 0.75 kHz (4-pole Bessel). Analysis was done using pCLAMP9 software (Molecular Devices, Sunnyvale, CA). Recapitulating the cytosolic and intra-SR cellular milieu in vitro during planar lipid bilayer studies is impossible. Consequently, experimental compromises were necessary, and here the solutions approximated those in cardiomyocytes during systole. Low cytosolic Ca 2ϩ (0.1-1 M) reduces RyR2 activity to a level unsuitable for reliable measurements. Some researchers have overcome this issue by omitting Mg 2ϩ , but this causes a very non-physiological RyR2 Ca 2ϩ dependence, as in cells Mg 2ϩ normally competes with Ca 2ϩ for occupancy of RyR2 cytosolic Ca 2ϩ activation and inactivation sites (51). Differences in single channel parameters (P O , MOT, and MCT) extracted from time traces were compared using two-tailed t tests against the values following the addition of CaM-WT with p Ͻ 0.05 considered significant. Also, comparisons with control (no addition of CaM) were done with the same p value criteria.
Protein Expression and Purification-CaM was expressed from the pMAL vectors and purified as described previously (8). The identity, purity, and integrity of each protein preparation was confirmed by SDS-PAGE and MALDI-TOF mass spectrometry of trypsin-digested proteins.
CaM Ca 2ϩ Titrations in the Presence of the RyR2(R3581-L3611) Peptide-A peptide corresponding to the RyR2 CaMBD (human RyR2 3581 RSKKAVWHKLLSKQRKRAVVACFRMA-PLYNL 3611 ) was purchased from Peptide 2.0 Inc. (Chantilly, VA) at Ͼ 98% purity. Titration experiments were done as described previously (8). Briefly; pH-and Ca 2ϩ -buffered solutions (50 mM HEPES, 100 mM KCl, 0.5 mM EGTA, and 2 mM NTA at pH 7.2 (25°C)) with or without 7 mM CaCl 2 were mixed to obtain different [Ca 2ϩ ] free levels (52). CaM (15 M), RyR2(R3581-L3611) peptide (16.5 M), 16.5 M TCEP, and Fura-2 (Invitrogen) Ca 2ϩ probe (0.8 M) were added to double distilled water for dilution of 1.5ϫ concentrated buffers. A 15% error for the [Ca 2ϩ ] free was included in data fitting procedures based on measuring [Ca 2ϩ ] free using Fura-2 and Ca 2ϩ binding to CaM-WT. The intrinsic protein fluorescence from each CaM domain was monitored during CaM/RyR2 CaMBD complex Ca 2ϩ titrations. The titration curves were fitted to a two-site Adair function as described previously (8,19,53,54). Briefly, fluorescence intensities (FI) from the N-and C-domains of CaM were measured as partial Phe and Tyr emission spectra, and the fractional saturations (Y) for each domain were fitted to the raw FI according to Equation 1, where b and a are the initial FI and the span in FI, respectively. Y is the fractional saturation of the monitored CaM domain binding two Ca 2ϩ as described by the two-site Adair model, Y ϭ K 1 ϫ ͓X͔ ϩ 2 ϫ K 2 ϫ ͓X͔ 2 2 ϫ ͑1 ϩ ϫ ͓X͔ ϩ K 2 ϫ ͓X͔ 2 ) (Eq. 2) where K 1 is the sum of the microscopic equilibrium constants, and K 2 is the equilibrium constant for the domain binding to two Ca 2ϩ . The apparent dissociation constant (appK D ) for either domain was then calculated as the reciprocal square root of K 2 . The fitted K 2 values were compared using one-way ANOVA with Dunnett's multiple comparison test against the value for CaM-WT/RyR2(R3581-L3611) titrations.
Isothermal Titration Calorimetry of RyR2(R3581-P3607) Peptide with CaM-A peptide corresponding to the CaMBD (human RyR2 3581 RSKKAVWHKLLSKQRKRAVVACFR-MAP 3607 ) was purchased from GenScript (Piscataway, NJ) at Ͼ95% purity. Titration of the RyR2(R3581-L3607) peptide with CaM was investigated under both Ca 2ϩ -free (apo) and Ca 2ϩsaturating conditions and followed the use of ITC. Purified CaM variants were dialyzed (10 K Slide-a-lyzer TM , Thermo Scientific) against the titration buffer (10 mM HEPES, 150 mM KCl, pH 7.2) with either 10 mM EDTA or 10 mM CaCl 2 added for 40 h at 4°C (with buffer changed at 24 h). The spent dialysis buffer was filtered (0.22 m) and used for diluting peptide and CaM preparations. 5 mM final TCEP was added to all solutions. Protein and peptide concentrations were estimated using absorption at 280 nm (NanoDrop ND-1000). Using an Auto-ITC200 (Malvern Instruments Ltd.) isothermal titration calorimeter, the peptide (apo 0.1 mM, Ca 2ϩ 10 M) in the sample cell was titrated 20 times at 25°C with CaM (apo 1 mM, Ca 2ϩ 100 M) in 2-l increments at 2-min intervals and with 750 rpm stirring (c values: apo 2-10 and Ca 2ϩ 550 -850). Two to three titrations were done for each combination of CaM variant and Ca 2ϩ condition. ITC curves (change in instrument heat effect as a function of time, DP) were analyzed using MicroCal PEAQ-ITC software, and the extracted curves of the cumulative reaction heat as a function of total CaM concentration were fitted to a two-component binding equation (32). The reaction stoichiometry (n), standard enthalpy change (⌬H o ), and dissociation constant (K D ) were estimated from this fitting, and the parameters were compared using one-way ANOVA and Tukey's post hoc test for all possible comparisons. The binding reaction change in standard Gibb's free energy (⌬G o ) and entropy contribution (ϪT⅐⌬S o ) were calculated from Equation 3.
NMR Spectra of 15 N-Labeled RyR2(R3581-L3611) Peptide in Complex with CaM-A version of the RyR2 CaMBD peptide, 15 N-labeled at Val-3586 and Phe-3603 ( 15 N-RyR2(R3581-L3611)), was purchased from ProteoGenix (Schiltigheim, France) at Ͼ98% purity. The complex of this peptide with CaM-WT, -F142L, or -N98S was investigated under both Ca 2ϩ -free (apo) and Ca 2ϩ -saturating conditions using NMR. All four samples contained 20 mM HEPES, 100 mM KCl, 2 mM TCEP, 1 mM PMSF, 5% D 2 O, 40 M TSP-d 4 , 100 M 15 N-RyR2(R3581-L3611), and 200 M CaM-WT or -F142L along with either 10 mM CaCl 2 or 10 mM EDTA. Protein and peptide concentrations were estimated using absorption at 280 nm (NanoDrop ND-1000) and quantitative NMR. The pH of the samples was adjusted to 6.50 Ϯ 0.03 with either 1 M KOH or 1 M HCl, and the final sample volume was 550 l. 15 N-HSQC spectra were recorded on a 600-MHz Bruker AVIII at 298.1 K. TopSpin 3.2. was used for the acquisition and processing of the spectra. To assign the peaks in the 15 N-HSQC, a 2D 15 N-TOCSY-HSQC with a DIPSI2 mixing of 60 ms and a ␥B1/2 of 9.6 kHz was performed.