The skeletal muscle isoform of the ryanodine receptor Ca2+-release channel (RyR1) is regulated by Ca2+ and CaM (calmodulin). CaM shifts the biphasic Ca2+-dependence of RyR1 activation leftward, effectively increasing channel opening at low Ca2+ and decreasing channel opening at high Ca2+. The conversion of CaM from a RyR1 activator into an inhibitor is due to the binding of Ca2+ to CaM; however, which of CaM's four Ca2+-binding sites serves as the switch for this conversion is unclear. We engineered a series of mutant CaMs designed to individually increase the Ca2+ affinity of each of CaM's EF-hands by increasing the number of acidic residues in Ca2+-chelating positions. Domain-specific Ca2+ affinities of each CaM variant were determined by equilibrium fluorescence titration. Mutations in sites I (T26D) or II (N60D) in CaM's N-terminal domain had little effect on CaM Ca2+ affinity and regulation of RyR1. However, the site III mutation N97D increased the Ca2+-binding affinity of CaM's C-terminal domain and caused CaM to inhibit RyR1 at a lower Ca2+ concentration than wild-type CaM. Conversely, the site IV mutation Q135D decreased the Ca2+-binding affinity of CaM's C-terminal domain and caused CaM to inhibit RyR1 at higher Ca2+ concentrations. These results support the hypothesis that Ca2+ binding to CaM's C-terminal acts as the switch converting CaM from a RyR1 activator into a channel inhibitor. These results indicate further that targeting CaM's Ca2+ affinity may be a valid strategy to tune the activation profile of CaM-regulated ion channels.
Upon stimulation with a wide variety of signals, such as receptor binding and membrane depolarization, cells respond by elevating intracellular Ca2+ either globally throughout the cytoplasm or in local microdomains in the vicinity of the point of Ca2+ entry into the cytoplasm. Ca2+-binding proteins such as CaM (calmodulin)  decode these Ca2+ signals. CaM is a 148-amino-acid protein consisting of two globular domains connected by a flexible linker. Each of the two domains contains a pair of well-conserved canonical EF-hand Ca2+-binding motifs which bind Ca2+ in a co-operative manner . The canonical EF-hand consists of a helix-loop-helix motif [3–5]. Ca2+ is co-ordinated within the EF-hand by seven oxygen atoms from six amino acids, five of these residues are within the loop, and the sixth is located in the exiting helix. The co-ordinating residues (Figure 1) can be designated by their positions in the linear sequence within the loop (1, 3, 5, 7, 9 and 12) and by their locations in the tertiary geometry of the co-ordinating site (±X, ±Y and ±Z) . The residues of the co-ordinating site form a pentagonal bipyramid where the y- and z-axis pairs align along the vertices of a planar pentagon with the x-axis pair lying perpendicular to the y- and z-axis plane. The residue at position 1 (+X) is a conserved aspartate residue and position 3 (+Y) is most frequently an aspartate or asparagine residue. The residue at position 9 (−X) does not directly co-ordinate Ca2+, rather the interaction is mediated by a bridging water molecule. The residue at position 12 (−Z) is a conserved bidentate glutamate residue. Unlike the other co-ordinating residues which chelate Ca2+ via side-chain oxygens, the residue at position 7 (−Y) co-ordinates Ca2+ via its backbone carbonyl oxygen. Residues 7–9 form a small antiparallel β-sheet between the two coupled EF-hands within each domain.
The Ca2+ concentration range to which CaM responds is increased as a result of the 10-fold difference in the Ca2+ affinity of the two domains, the N-terminal domain (Ca2+-binding sites I and II) has approx. a 10-fold lower Ca2+ affinity than the C-terminal domain (Ca2+-binding sites III and IV) [7–9]. The differing Ca2+ affinities contribute to CaM's functional bifurcation which allows CaM's two domains to serve different functions [10,11]. For example, Ca2+-dependent facilitation of Cav2.1 voltage-gated Ca2+ channels requires Ca2+ binding to CaM's C-terminal, whereas Ca2+-dependent inhibition of these channels requires Ca2+ binding to CaM's N-terminal . Furthermore, the two lobes of CaM may separately decode global and local Ca2+ signals [13,14]. Therefore one strategy to manipulate cellular responses to Ca2+ signals may be to modify the Ca2+ affinity of the protein which decodes the signal. Because of its functional bifurcation, CaM is an attractive target. By manipulating the affinity of one domain of CaM it may be possible to selectively enhance a specific CaM function.
RyR (ryanodine receptor) Ca2+-release channels form the efflux pathway for the mobilization of Ca2+ stores from the SR (sarcoplasmic reticulum)/ER (endoplasmic reticulum) . The skeletal muscle isoform of the RyR (RyR1) is essential for Ca2+ release from the SR to activate contraction . RyR1 is a homotetramer composed of four 565 kDa subunits, each with a single high-affinity CaM-binding site [17–19]. In vitro, RyR1 channel opening exhibits a biphasic Ca2+-dependence attributed to high- and low-affinity Ca2+-binding sites which when occupied will, respectively, activate and inhibit channel opening . CaM shifts the Ca2+-dependence of RyR1 opening to lower Ca2+ concentrations, effectively increasing channel opening at sub-micromolar Ca2+ and inhibiting the channel in micromolar Ca2+ . The switch from a channel activator to a channel inhibitor is due to Ca2+ binding to CaM, as Ca2+-insensitive mutants of CaM activate, but do not inhibit, RyR1 . The Ca2+-induced conversion of CaM from a RyR1 activator into a channel inhibitor suggests that the RyR1 activation profile might be manipulated by tuning the CaM Ca2+ affinity. This strategy requires that the CaM Ca2+-binding site(s) acting as the RyR1 regulatory switch be identified and the Ca2+ affinity of the specific site be modified without significantly altering the RyR1–CaM functional interaction.
Initial work to ascribe particular regulatory functions to specific lobes of CaM abolished CaM inhibition of RyR1 via mutagenesis to knockdown the Ca2+ affinity of the individual lobes of CaM [21,22]. The authors concluded that Ca2+ binding to the C-terminal lobe of CaM serves as the switch converting CaM from a RyR1 activator into a channel inhibitor [22,23]. However, this view was challenged by Boschek et al.  who, based on the similar Ca2+-dependence of RyR1 activation and CaM conformational changes when bound to a putative RyR1 CaM-binding peptide, concluded that Ca2+ binding to the C-terminal of CaM activates RyR1, and Ca2+ binding to the N-terminal results in channel inhibition.
In the present study we tailor the Ca2+ activation profile of RyR1 by engineering a series of CaM variants designed to enhance the affinity of specific Ca2+-binding sites by increasing the number of acidic residues within individual EF-hands. The functional effects of these mutants were assessed via their effects on the Ca2+ activation profile of RyR1. Our results demonstrate the feasibility of tuning the RyR1 activation profile by fine-tuning the domain-specific Ca2+ affinity of CaM and support the hypothesis that Ca2+-binding to the C-terminal of CaM is the switch converting CaM from a RyR1 activator into a channel inhibitor. Thus tuning the Ca2+-binding affinities of CaM offers a potential strategy for tailouring the activation profile of CaM targets.
Pigs were obtained from Clemson University Research Farm Services. [3H]Ryanodine was purchased from PerkinElmer Life Sciences. Unlabelled ryanodine was obtained from Calbiochem. The RyR1 CaM-binding peptide RyR3614-3643 was synthesized by Anaspec.
Protein overexpression and purification
Mutations targeting specific CaM Ca2+-binding sites were introduced using QuikChange® mutagenesis kits (Stratagene) and verified by DNA sequencing. Recombinant rat CaM and its mutants were overexpressed in Escherichia coli strain BL21(DE3)pLysS using the pET-7 vector [25,26]. Proteins were purified by phenyl-Sepharose chromatography . Protein concentrations were determined using the molar absorption coefficient 3030 M−1·cm−1 at 277 nm .
CD spectra were obtained in the far-UV (190–260 nm) wavelength region on a Jasco-810 spectropolarimeter at room temperature (22 °C) using a Tris/KCl buffer [10 mM Tris/HCl (pH 7.4) and 100 mM KCl] in a 1 cm light-path quartz cuvette. All spectra were background subtracted and are the average of at least ten scans. To determine the structural effects of RyR3614-3643 on CaM structure, far-UV CD spectra of RyR1 peptide alone were subtracted from the spectra of CaM plus the peptide.
Equilibrium Ca2+ titrations
The CaM lobe-specific Ca2+ affinities were determined by monitoring the intrinsic phenylalanine (λex=253 nm, λem=280 nm) and tyrosine (λex=277 nm, λem=320 nm) fluorescence, which are sensitive indicators of Ca2+ binding to CaM's N- and C-terminal lobes respectively [29,30]. The lobe-specific Ca2+ affinities of CaM bound to the putative RyR1 CaM-binding domain, RyR13614-3643 , were similarly determined, but with tyrosine excited at 265 nm and emission monitored at 292 nm to avoid interference from the peptide intrinsic tryptophan fluorescence. All spectra were collected at room temperature using a PTI spectrofluorimeter with a 1 cm pathlength quartz cuvette. The Ca2+ stock solution (15 or 50 mM) was titrated into 5 or 10 μM CaM in a buffer containing 50 mM Hepes, 100 mM KCl, 5 mM NTA (nitrilotriacetate) and 0.05 mM EGTA (pH 7.4). The free Ca2+ concentration was determined by using Oregon Green 488 BAPTA [1,2-bis-(o-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid]-5N (0.2 μM) according to eqn (1):
Where F is the fluorescence at each titration point, and Fmin and Fmax are the fluorescence intensities at zero Ca2+ and saturating Ca2+ respectively. The Oregon Green dissociation constant for Ca2+ is 21.7 μM (λex=495 nm, λem=520 nm) . Titration data were fitted to a non-linear Hill equation (eqn 2):
where f is the relative fluorescence change, Kd is the dissociation constant of Ca2+ to CaM, h is the Hill coefficient and [Ca2+] is the free Ca2+ concentration.
Equilibrium [3H]ryanodine binding
Skeletal muscle HSR (heavy SR) vesicles were prepared from porcine longissimus dorsi muscle as described previously . [3H]Ryanodine binds to the open RyR with high affinity and specificity and is therefore a sensitive indicator of channel activity . [3H]Ryanodine binding was performed by incubating HSR vesicles (0.2 mg/ml) for 90 min at 37 °C in medium containing 150 mM KCl, 10 mM Pipes (pH 7.0), ± wild-type or mutant CaM (see Figure legends for details), 100 nM [3H]ryanodine, 1 mM EGTA and CaCl2 to obtain the free Ca2+ concentration indicated . HSR vesicles were then collected on Whatman GF/B filters and washed with 8 ml of ice-cold 100 mM KCl buffer. Radioactivity retained by the filters was determined by liquid-scintillation counting. Estimates of maximal [3H]ryanodine-binding capacity of each HSR vesicle preparation were determined in medium that in addition contained 500 mM KCl, 5 mM ATP and 100 μM Ca2+. Non-specific binding was measured in the presence of 20 μM non-radioactive ryanodine. [3H]Ryanodine binding is expressed as a percentage of maximal [3H]ryanodine binding. Owing to the number of mutations examined, it was not feasible to include all CaM variants in the same assay run. However, variants were always assayed with a control, either wild-type CaM in the concentration-dependence experiments or no CaM and wild-type CaM when examining the Ca2+-concentration-dependence. As there were no statistical differences between the subsets of control experiments, the control experiments were pooled.
The CaM-concentration-dependence of SR vesicle [3H]ryanodine binding was fitted to a four parameter Hill equation. Data relating the dependence of SR vesicle ryanodine binding to Ca2+ concentration were fitted to eqn (3), where Bo is the maximal Ca2+-activated SR [3H]ryanodine binding, EC50 is the Ca2+ concentration required to achieve half-maximal binding and IC50 is the Ca2+ concentration required to half-inhibit SR vesicle [3H]ryanodine binding. Curve fitting was performed using SigmaPlot 9.0 (Systat Software).
SR Ca2+-release channel protein was purified from SR membrane vesicles as described by Lai et al. . Briefly, SR membranes (1 mg/ml were solubilized in 1 M NaCl, 25 mM Pipes (pH 7.1), 1.0% Chaps, 0.5% PC (phosphatidylcholine), 2 mM dithiothreitol, 0.8 mM benzamidine, 0.1 mM PMSF, 0.6 μg/ml pepstatin, 1 μg/ml leupeptin and 1 μg/ml aprotinin for 2 h at 4 °C with gentle agitation. Insoluble material was removed by centrifugation at 38000 rev./min in a Beckman 70.1 Ti rotor for 30 min. The supernatant was applied to 5–20% linear sucrose gradients containing 1 M NaCl, 25 mM Pipes (pH 7.1), 0.75% Chaps and 0.25% PC plus protease inhibitors and centrifuged at 24500 rev./min in a Beckman SW28 rotor at 4 °C for 16 h. Gradients were fractionated and tubes containing the RyR were identified by [3H]ryanodine binding. Fractions containing the RyR were pooled, concentrated with an Amicon centriprep 30, aliquotted, frozen in liquid nitrogen and stored at −80 °C until use.
RyR1-mediated currents were recorded as described previously . Muller–Rudin planar lipid bilayers were formed by painting a lipid mixture (phosphatidylethanolamine, phosphatidylserine and PC in a 5:3:2 ratio by wild-type; 50 mg/ml dissolved in n-decane) across a 150–250 μm aperture in a Delrin cup. The recording solution consisted of symmetrical 200 mM KCl, 20 mM Mops (pH 7.0), 1.0 mM EGTA and 0.8087 mM CaCl2. The cis chamber, to which solubilized channels were added, was connected to the head stage input of an Axoclamp 200B patch-clamp amplifier (Axon Instruments). The trans chamber was held at virtual ground. Channels incorporate into the bilayer in a fixed orientation, with the cytoplasmic face towards the chamber into the channels were added, thus the cis chamber corresponds to the cytoplasmic side of the channel. To determine the effects of CaM on RyR1 channel activity, aliquots of concentrated wild-type or mutant CaM were added to the cis chamber. Single-channel currents were recorded at +70 mV (Clampex program, pCLAMP 9.2 software, Axon Instruments). Currents were filtered at 1 kHz and recorded at 5 kHz. Only those channels that had a conductance of at least 600 pS were used. Single-channel Po (open probability) was calculated (Clampfit program, pCLAMP 9.2 software, Axon Instruments) from at least 2 min of recording using a half-amplitude threshold. When two channels were present in the bilayer, indicated by current amplitudes of twice the expected magnitude, Po was estimated as the average Po of the two channels, calculated as [Polevel1+(Polevel2 × 2)]/2. Bilayers in which three channels had been incorporated were dealt with similarly; recordings were not made from bilayers containing more than three channels.
Data are means±S.E.M. Parameters derived from the Hill equation and eqn 3 were compared using a one-way ANOVA with a Holm–Sidak multiple comparison as a post-hoc test. Statistical analysis was performed using SigmaStat 3.1 (Systat Software). Statistical significance was set at P<0.05.
Design of CaM variants increasing the number of acidic residues at Ca2+ co-ordination sites
The key factors contributing to the Ca2+-binding affinity of EF-hand proteins, especially CaM, has been the subject of extensive work [6,36–38]. Thus CaM variants were designed based on considerations of site-specific Ca2+-binding affinities, co-operativity between EF-hands within a domain, interdomain co-operativity and impact on CaM regulatory function. First, increasing the number of negatively charged residues from three to four within the EF-hand co-ordination sphere stabilizes the protein–Ca2+ complex. Increasing the number of charged residues beyond four may decrease the stability of the complex due to electrostatic repulsion between nearby similarly charged residues . Four CaM variants, T26D, N60D, N97D and Q135D, were designed with one additional charged ligand residue (aspartate) introduced into EF-hand loops I, II, III and IV (Figure 1) respectively. Site I of wild-type CaM contains four acidic residues at co-ordination positions +X, +Y and ±Z. The −Y and −X positions are occupied by threonine residues. Because the residue at the −X position contributes to Ca2+ co-ordination via a water bridge, the −Y residue (Thr26) was replaced with an aspartate residue. Site II contains four acidic residues located at co-ordinating positions ±X, +Y and −Z, an asparagine residue at +Z and a threonine residue at −Y. Given the fact that the most common residue at the +Z location is aspartate , the asparagine residue at position 60 was replaced with aspartate. Site III contains three acidic residues at +X, +Y and −Z, along with an asparagine residue at +Z, a tyrosine residue at −Y and a serine residue at −X. Following the same reasoning as for site II, an aspartate residue was substituted for the asparagine at residue 97 (+Z). Site IV is similar to site I, containing four acidic residues at +X, +Y and ±Z, a glutamine residue at −Y and an asparagine residue at −X. Again, following the same logic as for site I, the glutamine residue at residue 135 (−Y) was replaced by an aspartate residue. Secondly, domain-specific Ca2+-binding affinity can be altered by changing the electrostatic interactions between coupled EF-hands within a domain . Changing residues within each Ca2+-binding site can have significant effects on interdomain co-operativity . Finally, by tailoring the affinity of individual CaM Ca2+-binding sites, Ca2+ loading of the target-specific functionally important lobe of CaM can be manipulated. This strategy will allow CaM regulation of the biological activity of specific targets, in this case the activation profile of RyR1, to be tuned.
Effects of charge variants on domain-specific Ca2+ affinity
Lobe-specific Ca2+-binding affinities of the CaM variants were determined by monitoring intrinsic phenylalanine and tyrosine fluorescence changes during equilibrium Ca2+ titrations [29,30]. Figure 1(A) shows the positions of the intrinsically fluorescent residues. Although three phenylalanine residues are located in the C-terminal lobe (C-domain), VanScyoc et al.  showed that these residues are non-reporting. Thus the decrease in phenylalanine fluorescence intensity as illustrated in Figure 2(A) reflects the conformational changes in the N-terminal lobe (N-domain) of CaM as a result of Ca2+ binding to sites I and II. The increase in the tyrosine fluorescence signal shown in Figure 2(B) reflects the conformational changes in the C-domain of CaM as a result of Ca2+ binding to CaM sites III and IV . The Ca2+-dependence of phenylalanine and tyrosine fluorescence of wild-type and mutant CaMs were fitted with the Hill equation and are illustrated in Figures 3(A)–3(D). CaM Ca2+ dissociation constants (Kd) and Hill coefficients (h) as well as the free energy of Ca2+ binding (ΔG) derived from these fits are summarized in Figures 3(E)–3(H) and in Table 1. As established previously [7,39], the C-domain of CaM has a significantly higher Ca2+ affinity than the N-domain (C-domain Kd=2.04±0.02 μM; N-domain Kd=11.5±0.68 μM, P<0.05). Overall, the free energy of Ca2+ binding to the C-domain was ~1 kcal/mol (1 kcal=4.184 kJ) lower than the free energy of binding to the N-domain (Table 1). The T26D mutation at −Y in EF-hand I significantly increased the Ca2+ affinity of both the N- and C- domains. In contrast, the N60D mutation at +Z in loop II had no significant effect on the Ca2+ affinity of either the N- or C- domain of CaM.
Intrinsic CaM fluorescence
Equilibrium Ca2+ titrations of wild-type and EF-hand CaM mutants monitored by phenylalanine and tryrosine fluorescence
|CaM variant||Kd (μM)||h||ΔG1||Kd (μM)||h||ΔG2|
|CaM variant||Kd (μM)||h||ΔG1||Kd (μM)||h||ΔG2|
Mutations in the C-domain had more significant effects on the Ca2+ affinity than mutations in the N-domain. The N97D mutation significantly increased the C-domain Ca2+ affinity, significantly decreased the N-domain affinity and thus increased further the difference in affinity between the two lobes of CaM. In addition, the free energy of Ca2+ binding to the C-domain of this CaM was the lowest of any lobe of any variant. In contrast with all of the other substitutions studied, the Q135D mutation significantly decreased the Ca2+ affinity of the domain in which it was located and increased the Ca2+ affinity of the opposing domain. This substitution narrowed the difference in Ca2+ affinity between the two lobes and reduced the free energy difference between the N- and C- domains to ~0.4 kcal/mol (compared with ~1.0 kcal/mol for wild-type). Furthermore, the Q135D mutation was the only substitution to significantly affect the co-operativity of Ca2+ binding, reducing the Hill coefficient for the N-domain (h: wild-type=1.72±0.08; Q135D=1.05±0.02) and thus uncoupling the co-operativity between sites I and II within the N-domain. Interestingly, this mutation did not significantly alter the co-operativity between sites III and IV in the C-domain.
Binding to target molecules can significantly alter the CaM Ca2+-affinity . Therefore we determined the lobe-specific Ca2+ affinities of wild-type CaM and the mutant N97D when the protein was bound to a peptide from the putative RyR1 CaM-binding domain (RyR13614-3643) (Figure 4) [18,41]. Binding of CaM to RyR13614-3643 decreased the N-domain Ca2+Kd of wild-type approx. 4.6-fold and the C-domain Kd approx. 2.7-fold. Compared with wild-type CaM, binding of the N97D variant to the RyR13614-3643 peptide had greater effects on the domain-specific Ca2+Kd, particularly the C-domain (N-domain, 6-fold decrease; C-domain, 18-fold decrease).
Domain-specific Ca2+-binding affinities to the wild-type CaM–RyR13614-3643 complex and the N97D-CaM–RyR13614-3643 complex
Functional effects of CaM EF-hand mutations on RyR1 activity
In medium containing sub-micromolar Ca2+, CaM increases RyR1 channel opening, whereas, in medium containing micromolar Ca2+, CaM is inhibitory. Therefore to determine whether our EF-hand mutations altered the ability of CaM to regulate RyR1, we determined the CaM-concentration-dependence of SR vesicle [3H]ryanodine binding in medium containing either 100 nM or 700 μM Ca2+. In medium containing 100 nM Ca2+ (Figure 5), wild-type CaM increased ryanodine binding 12.9±0.9% with an EC50 of 109.0±18.4 nM and a Hill coefficient of 1.07±0.17. All four CaM variants increased SR vesicle ryanodine binding to a similar extent as wild-type CaM (Table 2); however, the site II mutation, N60D, decreased the EC50 for CaM activation (Table 2). None of the remaining mutations altered the CaM EC50 for SR vesicle [3H]ryanodine binding and none of the mutations significantly altered the Hill coefficient.
CaM-concentration-dependence of activation (A,B) and inhibition (C,D) of SR vesicle [3H]ryanodine binding
|CaM variant||Extent of activation (% Bmax)||ha||EC50 (nM)||Extent of inhibition (% Bmax)||hi||IC50 (nM)|
|CaM variant||Extent of activation (% Bmax)||ha||EC50 (nM)||Extent of inhibition (% Bmax)||hi||IC50 (nM)|
In medium containing 700 μM Ca2+, wild-type CaM decreased SR vesicle ryanodine 16.7±0.8% with an IC50 of 30.9±3.0 nM and a Hill coefficient of 1.80±0.28 (Figure 5 and Table 2). Q135D was the only mutation to significantly alter the efficiency of CaM inhibition. At saturating concentrations inhibition by this mutant was only 59% of the inhibition induced by wild-type CaM. The reduced efficacy of inhibition suggests that this variant interacts with RyR1 in a mode that differs from that of wild-type CaM. Two variants, T26D and Q135D, increased the CaM IC50 (73.7±10.0 and 57.8±6.7 nM respectively). None of the mutations altered the Hill coefficient.
To determine whether manipulating the site-specific Ca2+ affinity of CaM modifies the RyR1 activation profile, we next examined the effects of CaM variants on the Ca2+-dependence of SR vesicle [3H]ryanodine binding (Figure 6 and Table 3). In the absence of CaM, ryanodine binding exhibited the expected biphasic Ca2+-dependence (EC50=1.69±0.25 μM; IC50=367.1±53.5 μM). With Ca2+ as the sole channel regulator, SR vesicle [3H]ryanodine binding reached 63.1±2.1% of the estimated maximal binding. Including 2 μM wild-type CaM in the incubation medium significantly reduced the extent of Ca2+ activation of RyR1 to 39.4±3.0% of maximal, significantly reduced the Ca2+ EC50 to 0.23±0.08 μM and reduced the Ca2+ IC50 to 70.6±25.6 μM. This resulted in a leftward shift in the Ca2+-dependence of RyR1 activation such that both channel activation and inhibition occurred at lower Ca2+ concentrations. The switch from RyR1 activator to inhibitor is attributed to Ca2+ binding to CaM, thus we define a Ca2+ switch-point as the Ca2+ concentration at which CaM is converted from a channel activator into an inhibitor. In practice, this point was the Ca2+ concentration at which the line from eqn (3) fitted to the ryanodine-binding data obtained in the presence of CaM intersected the curve fitted to the binding data obtained in the absence of CaM, i.e. the Ca2+ concentration at which the ratio of [3H]ryanodine bound in medium containing CaM to the [3H]ryanodine bound in medium lacking CaM was equal to one (Figures 6C and 6D). Thus the switch-point for wild-type CaM occurred at 3.8±0.7 μM Ca2+ (arrows labelled 1 in Figure 6). Including 2 μM CaM variants in the binding medium caused a similar leftward shift in the Ca2+-dependence of ryanodine binding significantly decreasing both the Ca2+ EC50 and Ca2+ IC50. Furthermore, mutations in CaM's N-terminal Ca2+-binding sites had no significant effect on the CaM-induced decrease in Ca2+-activated SR vesicle ryanodine binding. Thus the Ca2+-dependence of ryanodine binding in medium containing N-terminal CaM mutants virtually overlapped the Ca2+-dependence of binding in medium containing wild-type CaM.
Effects of wild-type and EF-hand mutant CaM on the Ca2+-concentration-dependence of SR vesicle [3H]ryanodine binding
|CaM variant||Bmax (pmol/mg)||Bo (% of Bmax)||EC50 (μM)||IC50 (μM)||Switch-point (μM)|
|CaM variant||Bmax (pmol/mg)||Bo (% of Bmax)||EC50 (μM)||IC50 (μM)||Switch-point (μM)|
In contrast, mutations in CaM's C-domain Ca2+-binding sites had substantial effects on CaM modulation of RyR1 activation. When N97D, the site III CaM mutant, was included in the binding medium the extent of Ca2+ activation was significantly reduced compared with wild-type CaM (27.1±1.7% compared with 39.4±3.0%). When compared with wild-type CaM, this site III CaM variant switched from activating RyR1 to inhibiting the channel at a significantly lower Ca2+ concentration (1.2±0.2 μM compared with 3.8±0.7 μM). In contrast, the site IV mutation Q135D increased the extent of Ca2+ activation to 77.0±7.3% of maximal. The switch-point for this CaM variant was significantly increased to 38.2±7.2 μM, approx. 10-fold higher than the switch-point for wild-type CaM.
To verify the effects of our N97D mutant on the Ca2+-dependence of RyR1 channel opening, we compared the effects of wild-type CaM and N97D CaM on the Po of RyR1 channels incorporated into planar lipid bilayers. On the basis of a switch-point for wild-type CaM of 3.8 μM Ca2+, in recording medium containing 4 μM Ca2+, the addition of wild-type CaM should not reduce channel Po. In contrast, because N97D CaM switches from a channel activator to an inhibitor at 1.2 μM Ca2+, N97D should decrease channel Po. Indeed, this was observed (Figure 7). In the representative experiment in Figure 7(A), the addition of 3 μM wild-type CaM increased channel Po from 0.032 to 0.058. Figure 7(B) shows an experiment in which two channels incorporated into the bilayer simultaneously. The addition of 3 μM N97D CaM reduced the average Po of the two channels from 0.028 to 0.017. Summary data is presented in Figure 7(C). In medium containing 4 μM Ca2+, wild-type CaM did not significantly alter RyR1 Po (no CaM=0.040±0.025, 3 μM wild-type CaM=0.033±0.017). In contrast, the addition of N97D significantly reduced channel Po (No CaM=0.032±0.013, 3 μM N97D CaM=0.011±0.004).
Effects of wild-type and N97D CaM on single channel open probability
Conformational analysis of CaM variants and peptide binding effects on the secondary structures
The decreased inhibitory efficacy of the Q135D CaM variant (Figure 5) suggests an altered mode of interaction of this CaM with RyR1. Therefore to determine the effect of the EF-hand charge mutations on the conformation of CaM in solution and when bound to the RyR13614-3643 peptide, the secondary structure of the CaM variants was examined using far UV CD. Consistent with the substantial α-helical content of both apo- and Ca2+-loaded CaM, wild-type CaM had large negative troughs at 208 and 222 nm in buffer containing 1 mM EGTA and in buffer containing 1 mM Ca2+ (Figure 8A). In EGTA-containing medium, CaM variants T26D, N60D and N97D (Figures 8B–8D) exhibited a 30–34% decrease in ellipticity at 222 nm. In contrast, the Q135D substitution resulted in only a 9% decrease in ellipticity (Figure 8E). In Ca2+-containing buffer, substitution in Ca2+-binding sites I and III caused a 30–40% decrease in ellipticity, whereas substitution in sites II and IV had only small effects (2–8% decrease).
CD analyses on CaM variants
Upon addition of the RyR13614-3643 peptide to wild-type CaM in EGTA-containing buffer, there was 22% decrease in ellipticity at 222 nm attributable to CaM (Figure 8A). In Ca2+-containing buffer, addition of the peptide caused a 9% increase in the ellipticity of CaM. Thus both the Ca2+-free and Ca2+-bound forms of wild-type CaM interact with the RyR13614-3643 peptide. CaM variants T26D, N60D and N97D showed similar ellipiticity changes (27–39% decrease) upon the addition of RyR1 peptide in EGTA-containing medium (Figures 8B–8D). Again, the Q135D (loop IV) CaM variant differed and exhibited a 64% decrease in ellipticity under these conditions (Figure 8E).
In Ca2+-containing buffer, peptide addition caused a 9% increase in the wild-type CaM ellipticity at 222 nm (Figure 8A). CaM variants T26D, N60D and N97D all exhibited similar increases (7–10%) in ellipticity upon the addition of peptide in Ca2+-containing medium (Figures 8B–8D). The Q135D CaM variant was again the exception. Upon addition of the RyR13614-3643 peptide to Q135D CaM in Ca2+-containing medium, there was a 22% decrease in ellipticity (Figure 8E).
These results, summarized in Figures 8(F) and 8(G), suggest that the addition of a charged residue to EF-loops I, II or III causes only minor conformational changes to the apo- and Ca2+-loaded forms of these CaM variants. Furthermore, these CaMs undergo similar conformational changes upon binding to the RyR13614-3643 peptide. In contrast, the conformations of the apo- and Ca2+-loaded forms of the Q135D CaM variant upon RyR13614-3643 peptide binding are likely to be significantly different from those of the bound wild-type CaM.
CaM regulation of many targets, including a number of ion channels is multimodal. For example, CaM can facilitate or inhibit Cav2.1 depending on which lobe of CaM is Ca2+-bound . Similarly, RyR1 activation can be enhanced or inhibited depending on the Ca2+ saturation of the C-terminal lobe. Thus by altering the Ca2+ affinity and thereby the Ca2+ saturation state of a specific lobe of CaM, it should be possible to selectively enhance (or reduce) a particular mode of CaM regulation of ion channels.
In the present study, we tuned the activation profile of the RyR1 channel by introducing a point mutation into CaM (N97D) that generated a CaM variant which when bound to a putative RyR1 CaM-binding peptide specifically increased the Ca2+ affinity of CaM's C-domain without altering the Ca2+ affinity of the N-domain or substantially altering the mode of interaction with the peptide. To identify this site as critical for the conversion of CaM from a channel activator into an inhibitor, CaM mutations were designed to increase the affinity of each individual CaM EF-hand Ca2+-binding motif by increasing the number of acidic residues in Ca2+-co-ordinating positions within each EF-hand. The Ca2+ affinity of each of CaM's lobes was then determined by monitoring the Ca2+-dependence of the intrinsic phenylalanine and tryptophan fluorescence, which has been shown previously to reflect Ca2+ binding to sites I and II of CaM's N-domain and sites III and IV of CaM's C-domain respectively . The functional effects of these substitutions were then determined by examining CaM modulation of Ca2+ activation of the RyR1 Ca2+-release channel.
Effects of CaM variants on the RyR1 activation profile
RyR1 channel opening exhibits a biphasic Ca2+-dependence. CaM shifts this activation profile to lower Ca2+ concentrations, enhancing channel opening at low Ca2+ concentration and inhibiting opening at higher concentrations. On the basis on the lack of channel inhibition by a Ca2+-insensitive CaM mutant, the RyR1 activation at low Ca2+ is ascribed to the Ca2+-free form of CaM and the inhibition is attributed to Ca2+ binding to CaM. Using domain-specific EF-hand knockdown mutations, Rodney et al.  and Fruen et al.  showed that Ca2+ binding to the C-domain of CaM is the switch to convert CaM from a RyR1 activator into a channel inhibitor. However, previously published work  conflicts with this view. Boschek et al.  compared the Ca2+-dependence of skeletal muscle SR [3H]ryanodine binding with a Ca2+-induced change in signals from CaM that had been fluorescently labelled on either the N- or C-domain and bound to a putative RyR1 CaM-binding peptide. The Ca2+-dependent conformational changes in the C-domain of CaM occurred in the same concentration range as the Ca2+ activation of SR vesicle ryanodine binding. Furthermore, the Ca2+-dependent conformational changes in the N- domain of CaM occurred in the same concentration range as the Ca2+ inhibition of SR vesicle [3H]ryanodine binding. Therefore the authors concluded that Ca2+ binding to CaM's C-domain caused activation of RyR1 and Ca2+ binding to CaM's N-domain resulted in channel closure .
We used our CaM variants to determine whether Ca2+ binding to CaM's N-domain or C-domain is the switch to convert CaM from a RyR1 activator into a channel inhibitor. We reasoned that increasing the Ca2+ affinity of the functionally important CaM domain, with no change in the affinity of the opposing domain, would lower the Ca2+ concentration at which CaM was converted from a channel activator into a channel inhibitor (switch-point). In this regard, the N-terminal variants were not informative as neither mutation changed the relative domain Ca2+ affinity and therefore did not alter the Ca2+ switch-point. The C-terminal CaM variants were more useful. The N97D mutation decreased the C-domain Kd and lowered the Ca2+ concentration at which CaM switched from activating RyR1 to inhibiting the channel. Conversely, the Q135D mutation increased the C-terminal Ca2+Kd and increased the Ca2+ concentration at which CaM was converted from a channel activator into a channel inhibitor. Thus the Ca2+ switch-point moved in the same direction as the change in C-domain Kd. Furthermore, there was an inverse relationship between the N-terminal Ca2+Kd and the Ca2+ concentration at which the switch-point occurred (see Q135D in Tables 1 and 2). We attribute the effects of the N97D mutant on the RyR1 activation profile to alteration in the variant's Ca2+ affinity. The concentration-dependences of CaM activation and inhibition by the N97D overlapped those of wild-type CaM, suggesting this variant binds RyR1 in a mode that is functionally indistinguishable from wild-type CaM. However, interpretation of the Q135D results is more complex. Although the concentration-dependence of activation by the Q135D variant was similar to wild-type CaM, the extent of inhibition by saturating concentrations of the mutant was significantly less than wild-type. Thus the large shift in the Ca2+ switch-point associated with this variant is the result of both the altered Ca2+ affinity of the CaM and the blunted maximal extent of inhibition. Furthermore, upon binding to the RyR13614-3643 peptide there was a decrease in the ellipticity at 222 nm of the Ca2+-loaded form of this CaM mutant rather than an increase, as seen in wild-type CaM. Therefore this mutant probably binds RyR1 in a mode that differs from wild-type CaM. However, in spite of these caveats, the present results, particularly those regarding the N97D CaM variant, are consistent with the initial proposal that Ca2+ binding to CaM's C-terminal is the inhibitory switch. We are currently carrying out detailed structural studies to examine the binding modes of these CaM variants.
Charged ligand residues contribute to Ca2+-binding affinity
Of the six Ca2+ co-ordinating residues within each of the four EF-hands of mammalian CaM, the aspartate residue at positions 1 and 3 and the glutamate residue at position 12 are conserved (Figure 1). The conservation of these residues suggests they are critical for CaM function and, indeed, replacing the charged glutamate residue with an uncharged glutamine or alanine residue at loop position 12 (−Z axis) in any of the EF-hands substantially decreased the Ca2+ affinity of the binding site [42,43]. The identity of the residue at each of the remaining positions is variable. Reid and Hodges  proposed the acid-pair hypothesis to relate the Ca2+ affinity of EF-hands to the location of negatively charged residues in chelating positions. This hypothesis predicted that the highest Ca2+ affinity would occur in an individual EF-hand when acid pairs are located in the X- and Z-planes and that additional carboxylate residues would decrease the affinity of the binding site due to electrostatic repulsion. However, additional factors contribute to EF-hand Ca2+-binding affinity, including the hydrophobicity of the loop and flanking residues [44–46], the nature of the chelating residues within the loop  and electrostatic interactions, especially those between charged liganding residues [48–50].
Work with synthetic peptides and uncoupled EF-hands in intact CaM supported the hypothesis [43,51]. Black et al.  directly tested the acid-pair hypothesis in coupled EF-hands by monitoring the N-domain Ca2+ affinity of intact CaM via the Ca2+-dependent change in fluorescence of a tryptophan residue which replaced a phenylalanine residue at residue 19 and systematically adding or removing acidic residues in Ca2+ co-ordinating positions within EF-hands I or II. Their findings generally supported the acid-pair hypothesis in that increasing the number of acid pairs within each hand from zero, to one and then to two, increased the affinity of N-domain Ca2+-binding sites. Adding an additional acidic residue to either EF-hand further increased Ca2+ affinity which then fell upon the completion of the third acid pair. Rather than introducing an additional mutation to monitor the concentration-dependence of CaM Ca2+ binding we took advantage of the method of VanScyoc et al.  and determined the domain-specific Ca2+ affinities by monitoring tyrosine and phenylalanine fluorescence within each domain (Figures 2 and 3) . The acid-pair hypothesis was insufficient to predict our results. In accordance with the hypothesis, increasing the number of acidic residues from three to four (site III, N97D) increased the Ca2+ affinity of the C-domain. However, rather than decreasing the domain Ca2+ affinity, further increases in the number of acidic residues had variable effects, including increasing the N-domain Ca2+ affinity (T26D), having no effect on the N-domain Ca2+ affinity (N60D) and decreasing the C-domain Ca2+ affinity (Q135D). Differences in the location of the acidic residues did not seem to account for their varying effects on the Ca2+ affinity. The T26D variant increased the Ca2+ affinity, whereas the Q135D variant had a decreased affinity, despite the fact that both mutations added a negative charge at the −Y position and resulted in acidic residues at +X,±Y and±Z.
These results can be explained in terms of a charge ligand-balanced model in which both the number of negatively charged ligand residues and the balanced electrostatic dentate–dentate repulsion by adjacent charged residues are major determinants of the Ca2+-binding affinities of EF-loops in CaM . In addition to acidic residues at positions 1, 3, 5 and 12, EF-hand I also has positively charged lysine residues at positions 2 and 11 which may reduce the electrostatic repulsion arising from closely situated carboxylates. Thus the T26D substitution increased the net charge from −2 to −3 and increased the N-domain Ca2+ affinity. In contrast, site IV has an additional negative residue at position 11. Thus the Q135D substitution increases the net charge within this site from −5 to −6, increases the charge repulsion within the EF-hand and decreases C-domain Ca2+ affinity.
Two CaM variants we created (T26D and N60D) were similar to those generated by Black et al.  (F19W/T26D and F19W/N60D). Both studies reported an increase in the N-domain Ca2+ affinity upon replacing Thr26 with aspartate. However, the results differ with regard to the N60D variant. We observed no change in the N-domain Ca2+ affinity of the N60D CaM, whereas Black et al.  reported a significant increase. Although substitution of tryptophan for Phe19 provided a highly sensitive measure of the Ca2+-induced conformational changes in the N-lobe of CaM, the substitution may subtly perturb the structure of CaM, and thus have also altered the Ca2+ affinities of sites I and II.
Mutation effects on Ca2+ affinity of opposing domain
Ca2+-induced conformational changes to the two domains of CaM are coupled phenomena [52,53]. Thus changes in the Ca2+ affinity of either the N- or C-domain could cause a change in Ca2+ affinity of the opposite domain. However, the direction of affinity change of the opposite domain does not appear to be linked to the direction of affinity change of the mutated domain. Indeed, in the CaM variants under study the direction of change could be the same in both lobes (T26D) or in opposite directions with the N97D mutation increasing the C-domain Ca2+ affinity and decreasing the N-domain affinity, whereas the Q135D mutation had the opposite effects.
The decreased C-domain Ca2+-binding affinity of the site IV Q135D mutation and the increased N-domain affinity support further the notion that EF-hand motif IV mutations often result in a large change in Ca2+-binding affinity, possibly due to alteration in inter- and intra-domain interactions. Fefeu et al.  reported that, similar to our Q135D mutation, the V136G mutation altered CaM's interdomain interaction, and substantially decreased the C-domain Ca2+ affinity, while increasing the N-domain affinity. The near-by very conservative D133Q substitution not only drastically decreased the site IV Ca2+ affinity, but also decreased the affinity of the coupled site III within the C-domain . Such significant alterations in site IV Ca2+ affinity are likely to be due to the clustering of a large number of negatively charged residues in both liganding positions 1, 3, 5 and 12 and in non-liganding position 11. This EF-hand motif also lacks the positively charged residues (lysine) at position 2 in EF-hand III and at positions 2 and 11 in EF-hand I.
In wild-type CaM, the difference in the Ca2+ affinity of the two domains is such that the C-domain Ca2+-binding sites may be nearly saturated, whereas the N-domain Ca2+-binding sites are only partially filled. Under the conditions used in the present study, the N-domain Ca2+Kd of wild-type CaM was approx. 5.6-fold higher than the C-domain Kd. Mutations in Ca2+-binding sites I and II had little effect on the relative difference in the domain Ca2+ affinities (N-domain Kd/C-domain Kd: T26D~6.4; N60D~5.7). Although the T26D CaM variant bound Ca2+ at lower concentrations than wild-type CaM, because of the similar increases in the N- and C-domain Ca2+ affinity, the relative Ca2+ saturation of N- and C-domains was not significantly altered. In contrast, the C-domain mutation significantly altered the ratio of N- to C-domain saturation. The site III mutation, N97D, doubled the N-domain Kd/C-domain Kd ratio to 11.8, whereas the site IV mutation Q135D reduced the ratio by half to ~2.1. Thus, on the basis of the fits to the Hill equation shown in Table 1 and Figure 3, at a free Ca2+ concentration of 1 μM, the N-domain saturation of wild-type, N97D and Q135D CaM for Ca2+ would be ~1.5, 1.3 and 9.8% respectively. The C-domain saturation of these variants for Ca2+ would be 16.2, 36.5 and 6.0% respectively. Binding CaM to the RyR13614-3643 peptide similarly increased the Ca2+ affinity of the N-domain of wild-type and N97D CaM (Kd with the peptide/Kd without the peptide: wild-type~5; N97D~6). Peptide binding had only a small effect on the wild-type C-domain Ca2+ affinity (Kd in the absence of peptide/Kd in the presence of peptide: wild-type~3), but dramatically increased the N97D affinity (Kd ratio~18). As a result in medium containing 1 μM Ca2+, the N-domain saturation of wild-type and N97D CaM was similar, ~25.0 and 25.9% respectively. In contrast, the C-domain saturation of these variants differed greatly, 56.7 and 99.9% respectively. Thus by selectively and substantially increasing the Ca2+ saturation of the C-domain, we were able to specifically enhance the RyR1 inhibitory function of CaM.
A number of skeletal muscle diseases including MH (malignant hyperthermia) and DMD (Duchenne muscular dystrophy) are genetic disorders leading to impaired cellular Ca2+ regulations. Mutations in RyR1 account for the majority of MH cases  and leaky RyR1 channels may contribute to DMD . RyR1 channels from MH-susceptible individuals exhibit an enhanced sensitivity to CaM activation [56,57] and possibly a decreased sensitivity to Ca2+/CaM inhibition . Therefore targeting these muscles with a CaM that will reduce channel opening at low Ca2+ may be of therapeutic benefit. Furthermore, the enhanced Ca2+/CaM inhibition of RyR1 may reduce the resting Ca2+ leak through the channels in DMD muscle. Because CaM regulates numerous cellular signalling pathways, an obstacle to this strategy will be to specifically target RyR1. As the CaM Ca2+-binding site that acts as the regulatory switch varies with the CaM-binding partner, we have taken an initial step towards achieving that specificity by targeting the Ca2+ affinity of a single EF-hand.
Jenny Yang and Edward Balog conceived the project and designed experiments. Jie Jiang, Yubin Zhou, Jin Zou, Yanyi Chen, Priya Patel and Edward Balog collected and analysed data. Jie Jiang, Jenny Yang and Edward Balog wrote the paper with input from Yubin Zhou, Jin Zou and Yanyi Chen.
This work was supported by the National Institutes of Health [grant number GM081749 (to J. J. Y.)].