The effect of Mg²⁺ on Ca²⁺ binding to cardiac troponin C in hypertrophic cardiomyopathy associated TNNC1 variants

Abstract

Cardiac troponin C (cTnC) is the critical Ca²⁺ sensing component of the troponin complex. Binding of Ca²⁺ to cTnC triggers a cascade of conformational changes within the myofilament that culminate in force production. Hypertrophic cardiomyopathy (HCM) associated TNNC1 variants generally induce a greater degree and duration of Ca²⁺ binding, which may underly the hypertrophic phenotype. Regulation of contraction has long been thought to occur exclusively through Ca²⁺ binding to site II of cTnC. However, work by several groups including ours, suggest that Mg²⁺, which is several orders of magnitude more abundant in the cell than Ca²⁺ may compete for binding to the same cTnC regulatory site.

We previously used isothermal titration calorimetry (ITC) to demonstrate that physiological concentrations of Mg²⁺ may decrease site II Ca²⁺-binding in both N-terminal and full length cTnC. Here, we explore the binding of Ca²⁺ and Mg²⁺ to cTnC harbouring a series of TNNC1 variants thought to be causal in HCM. ITC and thermodynamic integration (TI) simulations show that A8V, L29Q, and A31S elevate the affinity for both Ca²⁺ and Mg²⁺. Further, L48Q, Q50R, and C84Y that are adjacent to the EF hand binding motif of site II have a more significant effect on affinity and the thermodynamics of the binding interaction.

To the best of our knowledge, this work is the first to explore the role of Mg²⁺ in modifying the Ca²⁺ affinity of cTnC mutations linked to HCM. Our results indicate a physiologically significant role for cellular Mg²⁺ both at baseline and when elevated on modifying the Ca²⁺ binding properties of cTnC and the subsequent conformational changes which precede cardiac contraction.

Introduction

Cardiac troponin C (cTnC) is a dumbbell-shaped protein with 4 EF hand domains. High affinity structural sites III and IV in the C-terminal domain bind Ca²⁺ (K_A ^~ 10⁷ M⁻¹) and Mg²⁺ (K_A ^~10⁴ M⁻¹) to tether cTnC to the other components of the cardiac troponin (cTn) complex [1–3]. The N-terminal domain contains site I (which is dysfunctional in cardiac muscle) and site II which plays the critical regulatory role. Low affinity (K_A ^~10⁵ M⁻¹) site II binds Ca²⁺ at elevated concentrations during systole (400 – 1000 nM) and is largely unbound during diastole (100 nM) [4]. cTnC is composed of nine α-helices: helices N, A, B, C, and D in the N terminal-domain are linked through the flexible D-E linker to the C-domain which contains helices E, F, G, and H. Binding of Ca²⁺ to site II acts as a conformational switch, causing helices N, A and D to move away from helices B and C and expose a hydrophobic cleft (Figure 1). This region is then bound by the cTnI switch peptide (TnI_sw), causing further perturbation within the cTn complex and the rest of the thin filament (TF) to expose actin binding sites allowing for contact with myosin heads, ultimately resulting in force production [5–7].

Figure 1:

N-cTnC within cTnC and the Troponin complex. A; the cTn complex is shown containing the core domain with some components excluded from the final protein structure to facilitate x-ray crystallography [108]. This complex includes the Ca²⁺ binding cTnC in red, the inhibitory cTnI in blue, and the tropomyosin binding cTnT in green. Black spheres depict the bound cations that interact with sites III/IV in the C-terminal domain and site II in the N-terminal domain. B; cTnC shown in rainbow colors with the N-terminal domain in blue and the C-terminal domain in red. C; the N-domain of cTnC (N-cTnC) with 6 mutations of interest labelled. Helices N through D are labelled. For this figure, PyMol was used to generate the cTnC structures.

Hypertrophic Cardiomyopathy (HCM) afflicts ^~1 in 200 in the general population [8, 9]. Over 1,000 HCM-associated mutations have been found in a variety of sarcomeric proteins, of which over 100 are located in the cTn complex [10–13]. Despite a wide range of molecular precursors, the disease phenotype consistently entails hypertrophy, myocyte disarray, and fibrosis [14, 15]. This devastating disease often manifests as sudden cardiac arrest secondary to ventricular tachycardia/fibrillation and is the most common cause of sudden cardiac death in young athletes [16, 17].

Hypertrophy has been posited to result from increased myofilament Ca²⁺ sensitivity which prolongs systole and shortens diastole [4]. Alternatively, hypertrophy may result from changes in maximum tension enacted by changes in tropomyosin displacement [18]. The role of cTnC as the Ca²⁺-sensing component has made it a target of study with multiple HCM-associated mutations identified in the regulatory N-terminal domain [19, 20]. While large scale studies show a sparce number of pathogenic HCM associated mutations in cTnC [21–24], this may be due to the central role this molecule plays in excitation-contraction (EC) coupling, whereby significant functional changes may be incompatible with physiological viability.

After potassium, Mg²⁺ is the second most abundant cellular cation with a total concentration of ^~15 - 20 mM. Mg²⁺ is also tightly regulated through extensive buffering by cytosolic components such as ATP. A wide range of free Mg²⁺ concentrations between 0.2 – 3.5 mM have been reported in different systems with the majority of Mg²⁺ believed to exist in complex with ATP. The consensus is the free [Mg²⁺]_i is ^~0.5 - 1 mM making this cation approximately 3 orders of magnitude more abundant than systolic Ca²⁺ [25–27]. Short term ischemia (^~15 minutes) elevates free Mg²⁺ up to 10-fold due primarily to a loss of [ATP]_i [28, 29]. Studies on isolated TnC [30–35], the cTn complex [30, 32], and reconstituted fibers [32, 36] have demonstrated that Mg²⁺ decreases the ability of Ca²⁺ to expose the hydrophobic cleft and cause subsequent structural changes in the rest of the cTn complex. Previous studies utilizing physiological systems similarly demonstrated an inverse correlation between Ca²⁺ sensitivity of force production and availability of Mg²⁺ in skinned skeletal and cardiac fibers [37–39]. 3 mM Mg²⁺ decreased the Ca²⁺ affinity in isolated cTnC [3] and skinned psoas muscles [40], but did not seem to cause structural changes in cTnC [3, 41].

Equilibrium dialysis studies suggested that Mg²⁺ does not compete with Ca²⁺ for binding to the N-terminal cTnC, and exclusively binds to the two C-terminal sites [42]. This notion has endured for decades. In the time since, however, several studies have shown through a number of different experimental techniques that the Mg²⁺ binding affinity of site II of skeletal and cardiac TnC is physiologically relevant [31, 33, 39, 43–49]. The Mg²⁺ affinity of site II was estimated through fluorescence to have a K_A of ^~0.6 ×10³ M⁻¹ and through competition experiments, site II was posited to be 33 – 44% saturated with this cation at diastolic concentrations of Ca²⁺ [3]. More recent studies show that these results also hold in TnC variants of similar sequence from other species where Mg²⁺ binding affinity of is an order of magnitude lower than Ca²⁺ [50]. Our recently published ITC and thermodynamic integration simulations corroborated and expanded upon these same findings [51].

A prevalent idea posits that contractile protein variants which destabilize the closed conformation of the protein prior to Ca²⁺ binding and/or those that favor the open, Ca²⁺-bound state confer an increase in affinity [24]. Sequence variations outside the coordinating residues of the cTnC EF hands may induce alterations in Ca²⁺ affinity allosterically [52, 53]. These changes in Ca²⁺ binding have previously been linked to HCM and DCM-associated variants [41, 54, 55]. In contrast, variants outside the binding residues of each EF hand are not thought to allosterically modify Mg²⁺ binding [56–58], therefore the role of this cation in HCM is currently unclear. Here, we further explore the effects of Mg²⁺ on Ca²⁺ binding to the regulatory domain of cTnC and possible modifying affects on the previously listed series of HCM-associated variants.

In this study, we focus on the HCM associated TNNC1 variants A8V [59, 60], L29Q [61], A31S [62], and C84Y [63, 64], the engineered mutation L48Q [3] and the Dilated Cardiomyopathy (DCM) -associated TNNC1 variant Q50R [65] (Figure 1). We have previously studied this series of mutations through Isothermal Titration Calorimetry (ITC) and Molecular Dynamics (MD) Simulations [55]. Our findings supported the notion that L48Q, Q50R, and C84Y destabilize the closed-state of cTnC or stabilize the interaction with the TnI_SW in the open-state leading to an increase in Ca²⁺-binding affinity. In contrast, A8V, L29Q, and A31S were found to alter Ca²⁺-coordination through a more subtle local structural perturbations [55].

We previously used ITC and thermodynamic integration (TI) simulations to show that physiological concentrations (1 mM) of free Mg ²⁺ significantly reduce Ca²⁺-binding to site II in both full-length and N-terminal cTnC [51]. It is important to establish whether cardiomyopathy mutations affect binding of Ca²⁺, Mg²⁺, or both. If the binding of these ions is affected unequally by a specific variant, then this could exacerbate or attenuate the effect when considering Ca²⁺ alone. This study compares Mg²⁺-induced modifications in the Ca²⁺ binding properties of the regulatory site of cTnC with cardiomyopathy associated TNNC1 variants with Ca²⁺ and Mg²⁺ binding affinity shown to be different for each mutant. In particular, the affect of L48Q on the binding of these cations was significantly different from the WT, with Q50R and C84Y also showing some small, insignificant differences. These findings underline the importance of considering background Mg²⁺ levels which were studied through competition experiments. We show that affinity of the five HCM-associated variants (including L48Q which is engineered) and a single DCM associated TNNC1 variant for Mg²⁺ is higher than WT cTnC which is now established to have a physiological significant affinity for Mg²⁺. Therefore, the presence of baseline Mg²⁺ may contribute to the dynamics which govern cardiac EC coupling. Further, higher than baseline concentrations of Mg²⁺ such as may occur during ischemia may accentuate the effects of inherited TNNC1 variants and further exacerbate dysfunction in the diseased heart.

Results

Each N-cTnC construct was independently titrated by Ca²⁺ and Mg²⁺ in the apo-state. Each construct was also independently pre-incubated with 1 mM Mg²⁺ and 3 mM Mg²⁺ and titrated by Ca²⁺ to measure the relative binding affinity (Figures 2 – 7).

Figure 2:

The thermodynamic properties of the binding interactions with WT N-cTnC and each of the mutants are listed. The Ca²⁺ and Mg²⁺ experiments were the titration of each cation into apo-state protein while +1/3 mM Mg²⁺ indicates the concentration of Mg²⁺ in each sample cell prior to titration with Ca²⁺. Each parameter is displayed the as mean ± SEM, with the exception of N which was fixed to 1.00 in the Mg²⁺ binding and preincubation experiments. ANOVA and subsequently Tukey’s test were independently used to find difference across constructs for each parameter and titration. For each parameter, the mean of constructs not connected by the same letter were found to be statistically different, p<0.05.

Figure 7:

Representative isotherms for each titration condition for L48Q, Q50R, and C84Y N-cTnC. In the 3 most N-terminal mutations are shown; L48Q, Q50R, and C84Y from top to bottom. On the left most panel, titration of Ca²⁺ into apo-state protein is shown with the next panels showing the titration of Ca²⁺ into 1 mM and 3 mM Mg²⁺ pre-incubated N-cTnC. The right most panel show the titration of Mg²⁺ into apo-state N-cTnC. These three mutants caused the greatest deviation in thermodynamic properties form the WT titration conditions. The Ca²⁺ into apo-protein titration is endothermic for Q50R and C84Y but exothermic for L48Q. The Mg²⁺ into apo-protein titration is endothermic for all 3 mutants. The pre-incubation condition with both 1 mM and 3 mM Mg²⁺ resulted in an exothermic interaction with Ca²⁺.

All studied TNNC1 variants increased the binding affinity of both Ca²⁺ and Mg²⁺ relative to the WT construct (Figure 2). The smallest increase was associated with A8V and the greatest with L48Q. In general, the mutations adjacent to site II (L48Q, Q50R, and C84Y) increased the affinity for each cation to a greater extent and altered the thermodynamic profile of the binding interaction more dramatically; the changes associated with the other mutants were more subtle.

The K_A associated with Mg²⁺ binding was orders of magnitude lower than that determined for Ca²⁺ binding for all constructs (Figures 2 and 3). The addition of 1 mM Mg²⁺ decreased both the amount of binding and affinity for Ca²⁺ at site II with 3 mM Mg²⁺ further accentuating the trend (Table 1).

Figure 3:

Comparing the affinity for Ca²⁺/Mg²⁺ in each titration condition between all N-cTnC constructs. The x-axis labels indicate the tiration conditions: “Ca2+” indicates the titration of this cation into apo-state protein, similarly “Mg2+” indicates titration into apo-state protein. “+1 mM Mg2+” and “+3 mM Mg2+” are the pre-incubation conditions for the construct into which Ca²⁺ was titrated; the affinity obtained is therefore seen in a sytem with both Ca²⁺ and Mg²⁺ present. SEM error bars are used to depict where significant differences exist in the mean values. Sample size for each condition was between 6-9 independent tirations. With the exception of L48Q, the highest affinity is seen in the Ca²⁺ titration and the lowest in the Mg²⁺ titrations. L48Q has the highest Mg²⁺ binding affinity by over an order of magnitude. Increasing Mg²⁺ from 0 to 1 to 3 mM lowered Ca²⁺ binding affinity in a graded manner. ANOVA and subsequent Tukey’s post-hoc test indicate a number of differences in the mean K_A between the constructs and titration conditions. For the titrations Ca²⁺ in the apo-state, L48Q, C84Y, and Q50R were significantly different from the WT. For the Mg²⁺ titrations in the apo-state, L48Q was significantly different from the WT. For pre-incubation with 1 mM Mg²⁺, A31S, L48Q, and C84Y were significantly different from the WT. For pre-incubations with 3 mM Mg²⁺ L48Q was significantly different form the WT.

Table 1 –

Mg²⁺ restraint distances for thermodynamic integration

	WT N-cTnC		L48Q		Q50R		C84Y
Restraint 1	ASP67 CG Atom 1029	2.3 Å	ASP67 CG Atom 1027	2.3 Å	ASP67 CG Atom 1036	2.3 Å	ASP67 CG Atom 1029	2.3 Å
Restraint 2	SER69 OG Atom 1048	3.7 Å	SER69 OG Atom 1046	3.7 Å	SER69 OG Atom 1055	3.7 Å	SER69 OG Atom 1048	3.7 Å
Restraint 3	THR71 OG1 Atom 1067	4.5 Å	THR71 OG1 Atom 1065	4.5 Å	THR71 OG1 Atom 1074	4.5 Å	THR71 OG1 Atom 1067	4.5 Å

Ca²⁺ Binding

The interaction of Ca²⁺ with each construct was endothermic and entropically driven with the exception of L48Q in which the reaction was exothermic. Ca²⁺ bound to WT N-cTnC with a K_A of 64.5 ± 3.1×10³ M⁻¹, A8V had moderately higher binding affinity. Relative to WT N-cTnC, Ca²⁺ bound to each cTnC variant with higher affinity: L29Q and A31S (^~2-fold), L48Q (^~6-fold), Q50R (^~3.5-fold), and C84Y (^~4-fold). The ΔG reflects the ΔH and the ΔS and as such demonstrates that the most favorable interaction occurs in L48Q, then Q50R/C84Y, followed by the other mutants (Figures 2, 8, 9). These results are in agreement with our previously published work [51, 55].

Figure 8:

Comparing the enthalpy for each Ca²⁺/Mg²⁺ titration condition between all N-cTnC constructs. Ca²⁺ and Mg²⁺ experiements are titration of each cation into apo-state protein while +1/3 mM Mg²⁺ indicates the concentration of Mg²⁺ in each sample cell prior to titration with Ca²⁺. SEM error bars are used to depict where significant differences exist in the mean values. Sample size for each condition was between 6-9 independent tirations. ANOVA and subsequent Tukey’s post-hoc test indicate a number of differences in the mean ΔH between the constructs and titration conditions. For the titrations Ca²⁺ in the apo-state, A31S>Q50R>C84Y>>L48Q were significantly different from the WT. For the Mg²⁺ titrations in the apo-state, only L48Q was significantly different from the WT. For pre-incubation with 1 and 3 mM Mg²⁺, all mutants were significantly different from the WT with A8V/L29Q most similar and L48Q most dissimilar from the WT.

Figure 9:

Comparing the entropy for each Ca²⁺/Mg²⁺ titration condition between all N-cTnC constructs. Ca²⁺ and Mg²⁺ experiements are titration of each cation into apo-state protein while +1/3 mM Mg²⁺ indicates the concentration of Mg²⁺ in each sample cell prior to titration with Ca²⁺. SEM error bars are used to depict where significant differences exist in the mean values. Sample size for each condition was between 6-9 independent tirations. ANOVA and subsequent Tukey’s post-hoc test indicate a number of differences in the mean T*ΔS between the constructs and titration conditions. For the titrations Ca²⁺ in the apo-state, A31S>Q50R>C84Y>>L48Q were significantly different from the WT. For the Mg²⁺ titrations in the apo-state, all conditions were significantly different and less than the WT. For pre-incubation with 1 and 3 mM Mg²⁺, all titrations were significantly different and less than the WT.

Mg²⁺ Binding

The interaction of Mg²⁺ with WT N-cTnC was endothermic and entropically driven with a K_A of 1.4 ± 0.1 ×10³ M⁻¹; more than an order of magnitude lower binding affinity than seen with Ca²⁺. Relative to the WT, Mg²⁺ binding to each mutant occurred in a similar energetic profile and with higher affinity: A8V and L29Q (^~2-fold), A31S (^~3-fold), L48Q (^~120-fold), Q50R (^~15-fold), and C84Y (^~20-fold) (Figure 2). The ratio between Mg²⁺ and Ca²⁺ binding affinity was significantly greater than seen in the WT in L48Q and greater but not significantly so in Q50R and C84Y; these findings stress the need for competition experiments which allow for the study of Ca²⁺-affinities in the presence of physiologically relevant Mg²⁺ concentrations (Figure 10).

Figure 10:

Ratio between Mg²⁺ and Ca²⁺ binding affinity for each N-cTnC construct indicating the realtive change in affinity. Standard deviation error bars are also depicted to provide a measure of the confidence in each ratio. Sample size for each set of titrations for the constructs was between 6-9. A greater ratio indicates that the Mg²⁺-binding affinity was greater in the construct relative to the baseline Ca²⁺-binding affinity. ANOVA and subsequent Tukey’s post-hoc test indicated that L48Q was significantly different from the other constructs, p<0.05 with the other constructs not statistically distinguishable. The ratio is not equal to one in any of the constructs highlighting the importance of considering background cellular Mg²⁺ in these studies and stressing the importance of competition experiments.

Ca²⁺ and Mg²⁺ Competitive Binding

Pre-incubation of WT N-cTnC with 1 mM Mg²⁺ lowered Ca²⁺-binding affinity by 1.6-fold and 3 mM Mg²⁺ further accentuated the effect. Pre-incubation of A8V with 1 and 3 mM Mg²⁺ reduced the Ca²⁺ affinity ^~3 and 7-fold, respectively and in L29Q the change was ^~3 and 9-fold, respectively. 1 and 3 mM Mg²⁺ decreased the Ca²⁺-affinity of L48Q by ^~2 and 5-fold, respectively. Competition altered the reaction kinetics in A31S, Q50R, and C84Y making each interaction exothermic and enthalpically driven. In A31S 1 mM had a higher than apo-state affinity for Ca²⁺ (1.4-fold) yet 3 mM Mg²⁺ reduced Ca²⁺-binding affinity (^~2-fold). In both Q50R and C84Y N-cTnC, 1 and 3 mM Mg²⁺ lowered the Ca²⁺-binding affinity of 3 and 5-fold, respectively. L48Q, Q50R, and C84Y affected Mg²⁺ and Ca²⁺-binding to a different degree, increasing Mg²⁺-binding more than they increased Ca²⁺-binding (Figure 10).

Mg²⁺ binding affinities from Thermodynamic Integration (TI)

The binding affinity for each protein structure determined by Thermodynamic Integration was averaged over 5 independent runs and was $-5.139\pm 2.308kcal/mol$, $-5.481\pm 0.719kcal/mol$, $-6.205\pm 2.112kcal/mol$, $-6.364\pm 1.372kcal/mol$, respectively. The TI calculated Mg²⁺ binding affinities were in good agreement with the ITC data, except for L48Q N-cTnC. The ITC data showed a much stronger increase in Mg²⁺ sensitivity for the L48Q N-cTnC. However, all mutated structures were shown to have increased Mg²⁺ sensitivity compared to WT cTnC. The $\Delta \Delta G_{Q50R}$ values were similar for TI and ITC (1.066 $kcal/mol$ and 1.63 $kcal/mol$, respectively). The $\Delta \Delta G_{C84Y}$ values also showed good agreement for TI and ITC (1.225 $kcal/mol$ and 1.79 $kcal/mol$, respectively) (Table 2).

Table 2 -.

Mg²⁺ thermodynamic integration free energy of WT and mutant N-cTnC structures

N-cTnC Structure	ΔGTI(kcal/mol) ± std. dev.	ΔΔGTI
WT	−5.139 ± 2.308
L48Q	−5.481 ± 0.719	−0.342
Q50R	−6.205 ± 2.112	−1.066
C84Y	−6.364 ± 1.372	−1.225

^*Statistically significant differences not found

Discussion

The binding of Ca²⁺ to site II within the N-terminal domain of cTnC is the fundamental molecular precursor to a series of conformational changes that culminate in cross bridge formation and force production. As such, changes in the sequence of this highly conserved protein often has grave consequences for the force production capabilities of the heart [75] and likely leads to cardiac remodeling. The six variants examined in this study all occur outside the EF hand binding regions and must allosterically alter Ca²⁺ affinity. Given the location of each mutation of interest and the desire to focus on changes in the binding interaction, we exclusively studied the N-terminal domain of cTnC. The mutations in question have also been studied at various levels of complexity by numerous groups, whose findings are in general agreement with our own [3, 55, 59, 61–63, 65]. In this study we found that the Ca²⁺ and Mg²⁺ binding affinity of the five HCM and single DCM variants were variable and different from the WT. The Mg²⁺ affinities measured here, with the WT serving as a point of comparison are physiologically significant and indicate a potential modulatory role for this cation in EC coupling.

The recently published Cryo-EM structure of the cardiac thin filament has shown that cardiomyopathy associated variants in troponin overwhelmingly occur in regions that interface with the actin-tropomyosin complex [76]. Variants which occur at a distance from these interfaces, are still most likely to affect changes through altered interactions with other proteins of the contractile complex [77]. These interactions may be affected by mutations occurring at the interface between proteins within the complex [78]. Thin filament loss-of-function variants are associated with HCM as they increase the Ca²⁺ sensitivity of contraction by decreasing the steric hinderance of myosin binding sites. In contrast, gain-of-function mutations in N-cTnC have a sensitizing effect given that in the holo-state, this domain displaces the TnI_SW from actin, moves tropomyosin, and unhinders myosin binding sites [21].

ITC directly measures the binding interaction and subsequent conformational change as an alternative to the introduction of naturally occurring fluorophores such as F27W [79ⁱ] or synthetic fluorophores such as IAANS [41, 80]. These reporters can be used to quantify the structural changes that proceed the binding interaction and thus, indirectly measure affinity. Fluorescence can also be used to report on the dissociation of Ca²⁺ from cTnC, the cTn complex, or the TF by utilizing chelators such as EDTA and rapid fluid changes through stopped flow experiments [53]. ITC has the sensitivity to detect minute enthalpic fluctuations which accompany these interactions, accurately detecting changes to within 0.1 μcal [81]. That is, not withstanding the inherent limitations associated with every experimental technique. For this assay a limitation results from the absence of other components of the cTn complex, particularly cTnI which plays a central role to the regulation of TF Ca²⁺ handling [41, 76, 82–84]; thus care must be taken when translating these findings to more complex systems.

The binding of Ca²⁺ to N-terminal cTnC is driven through a balance between the conformational strain resulting from the interaction and the energetics of exposing a hydrophobic cleft to the aqueous environment [85]. We posit that the changes in affinity seen in each TNNC1 variant result from either the destabilization of the apo-state protein or the stabilization of the solvent exposed state [55]. We found that all of the HCM associated cTnC variants studied had a more negative ΔG compared to the WT, consistent with our previously published MD Simulations [55]. Work by Bowman and Lindert corroborates these findings and suggests a unifying theory that increased frequency of opening may result from the lowered energetic cost of exposing the N-terminal domain of cTnC. The placement of more hydrophilic amino acids that destabilize hydrophobic packing in the closed state and stabilize the open, solvent-exposed state that follows, may allow for this mechanism of action [86].

A8V was only moderately different from the WT, consistent with previous findings that suggest this mutation alters the interaction of cTnC with other TF proteins rather than altering the Ca²⁺ binding affinity directly [63, 82]. The locus of valine near the interface with N-cTnI and strengthens interaction with the switch peptide making this a distinct possibility [55, 87]. Nuclear Magnetic Resonance (NMR) data suggest a slight increase in the opening frequency in the apo-state relative to the WT [59]. In our study, the binding affinity of this mutant for both Ca²⁺ and Mg²⁺ was higher than the WT. In pre-incubation experiments in which both cations were present, the A8V construct had lower than WT affinity for Ca²⁺ (Figures 2, 3, and 6).

Figure 6:

Representative isotherms for each titration condition into A8V, L29Q, and A31S N-cTnC. The 3 most N-terminal mutations are shown; A8V, L29Q, and A31S from top to bottom. On the left panel, titration of Ca²⁺ into apo-state protein is shown with the next panels show the titration of Ca²⁺ into 1 mM and 3 mM Mg2+ pre-incubated N-cTnC. The right most panel shows the titration of Mg²⁺ into apo-state N-cTnC. The majority of titrations are characterized by an endothermic interaction with the exception of A31S, where pre-incubation with Mg²⁺ resulted in an exothermic interaction with Ca²⁺.

L29Q nearly doubled the K_A for Ca²⁺ compared to WT, though the difference was not statistically significant. It also had greater than WT Mg²⁺ binding affinity but similar thermodynamic parameters and isotherm characteristics. Fluorescence (F27W) studies on isolated cTnC harboring the L29Q mutation exhibited an nearly 2-fold increase in Ca²⁺ affinity [41]. A complex system containing the entire contractile apparatus with L29Q cTnC had similar to WT Ca²⁺ sensitivity [88]. This mutation also changes sensitivity of force generation in a length- and phosphorylation-dependent manner [52]. Changing a hydrophobic residue to one that is polar, sensitizes site II Ca²⁺ sensitivity [89, 90]. Mechanistically, solvent exposure of an uncharged glutamine may facilitate a greater extent of opening than a hydrophobic leucine [55]. However, our previous work suggests that this variant has the highest closed probability among the seven constructs studied and lowers opening frequency [55]. In contrast, it has been shown through NMR that this mutation may cause a more open N-domain in the cTnC in both the apo- and holo-states [91]. L29Q may open in the holo-state with similar frequency as the WT and have similar energetic requirements as the WT for opening in both the apo- and holo-states [86]. Therefore, it is likely that the effects of this mutation in isolated N-cTnC are minimal and that changes are enacted through modification of the interaction with other cTn complex proteins [41].

The A31S TNNC1 variant is the result of a change of a hydrophobic amino acid for an uncharged one within the site 1 EF hand. In skeletal tissue, there is a great deal of cooperativity between binding to the two N-terminal sites of TnC [92, 93]. This mutant was found to have a greater K_A for both Ca²⁺ and Mg²⁺ compared to the WT and significantly higher affinity for Ca²⁺ in both pre-incubation conditions. A31S was found through MD simulations to sample a greater number of interhelical angles and to have a lower average angle between helices A and B [55]. Interestingly, pre-incubation with Mg²⁺ completely changed the reaction dynamics (Figure 6). ΔH reflects the strength of hydrogen bonds, van der Waals interactions, and electrostatic forces between the titrant and the target ligand. Optimal placement of hydrogen bond donors and acceptors balances de-solvation of polar groups to contribute to the enthalpy change [94]. The significant change in enthalpy suggests alteration of the number of bonds formed by the side chains of binding site residues or those exposed to the environment following the conformational change. This mutation may stabilize binding site I between helices A and B through formation of an additional hydrogen bond causing local changes that minimally alter global structure [62].

L48Q significantly altered affinity and thermodynamics of the binding interactions studied. The combination of high Ca²⁺ and Mg²⁺ affinity resulted in the highest observed K_A in both competition conditions. This is not unexpected, as the mutant is located within the BC helical bundle and was strategically engineered to increase the Ca²⁺ sensitivity of force production [3]. Our previously published MD simulations suggest that the absence of a hydrophobic residue disrupts hydrophobic packing in the AB domain. We also reported that the L48Q mutant opens more frequently than the other constructs [55, 86]. The changes in ΔH are likely due in part to the presence of an additional hydrogen bonds resulting from the introduction of a polar amino acid in a key domain of cTnC. In vivo, this would increase the opportunities for interaction with the TnI_sw [17, 95, 96]. Tikunova et al. originally suggested that despite a shift towards the Ca²⁺-bound state, resulting from a reduction in hydrophobic contact between helices NAD and BC, the solvent exposure of the N-domain is minimized by numerous side chain contacts. Their hypothesis regarding the disruption of hydrophobic interactions and minimization of exposure to the surrounding solvent is reconcilable with our findings and explains the much lower ΔS associated with this set of titrations [3].

Q50R is a relatively recently identified mutation that has yet to be fully explored. This mutant replaces a polar side chain with one that is bulky and charged. Given the results in L48Q and the vicinity of these residues, it is conceivable that the packing of helices NAD and BC is also disrupted by this mutant. This mutant had a much higher affinity than WT for both Ca²⁺ and Mg²⁺. Similar to L48Q, the interaction with Ca²⁺ was exothermic in the pre-incubation condition. Our previous work suggests that Q50R is more frequently open than the WT cTnC [55]. The reduced entropic cost of exposing a more charged residue to the aqueous environment may explain the decreased ΔS of the system, yet the adjacent residues would also be exposed to variable degrees in this mutant. Further, the energetic cost of opening the hydrophobic patch is increased in this mutant in comparison to the WT protein that has a less stable closed conformation [86].

C84Y places a bulky hydrophobic side chain in the region immediately preceding the flexible DE linker that is bound and stabilized in the open state by the TnI_SW. This bulky tyrosine may act as a wedge to reduce interaction with the TnI_SW and thus increase the Ca²⁺ sensitivity of force development in skinned fibers [63, 82]. C84Y was thermodynamically similar across all titrations with Q50R; given the location of these mutants, this does not necessarily suggest a similar mode of action. Interestingly however, our MD Simulations previously showed that a hydrophobic interaction between C84 and Q50 may be disrupted by this tyrosine. The bulky tyrosine in helix D may reduce the entropic cost of opening associated with the binding interaction, this is consistent with the observed, lower than WT ΔS values in C84Y N-cTnC [55].

The calculated Mg²⁺ binding affinity results using TI were in good agreement with the ITC values. We observed increased ion sensitization for all mutant N-cTnC structures. In particular the ${\Delta \Delta G}_{TI}$ values for Q50R and C84Y mutations aligned very well with the experimental data. The absolute binding affinities calculated for WT, Q50R and C84Y were overestimated compared to the ITC values by less than 1 kcal/mol. Parameterization of cations, especially Mg²⁺, in simulating biological systems has proven to be difficult [97, 98]. This could offer a potential explanation for the overestimation and relatively high standard deviations observed in the averaged absolute binding affinities. Another source of discrepancy between the in silico and ITC results, arises from the fact that there is no PDB structure of Mg²⁺ bound to site II of N-cTnC. In order to create the starting structure for TI simulations the PDB 1AP4 served as the base model, and Mg²⁺ was substituted in place of the Ca²⁺ ion. If crystal structures of Mg²⁺-bound WT N-cTnC and mutants were to exist, the use of these structures could potentially improve the IT results. TI of the L48Q mutant did not produce nearly as strong an increased Mg²⁺ sensitization as observed in the ITC data. We speculate that this could possibly be attributed to large conformational changes that were unable to be captured using TI. The timescale of the TI simulations was only 5 ns, which was insufficient to properly sample any large protein conformational changes.

Our data clearly demonstrate that each of the N-cTnC variants, including the WT, responded significantly but variably to the presence of Mg²⁺. Except for A31S, each mutant site II has a significantly lower Ca²⁺ binding affinity in the presence of 1 mM Mg²⁺. The degree of desensitization is best described as A8V>L29Q>Q50R>C84Y>L48Q. Given these observations, it is possible that Mg²⁺ binding dampens the presupposed sensitizing effect of HCM-associated mutations (Chang and Potter 2005); at the very least, the role of background Mg²⁺ in modifying Ca²⁺ sensitivity of force production cannot be ignored.

Given the high concentration of free Mg²⁺ in the cytosol and its similarities as a divalent cation and the small difference in atomic radius in comparison to Ca²⁺, this ion is a candidate for binding to site II of cTnC. A polar serine at residue 69 and a negatively charged glutamic acid at residue 76 in the EF hand binding site II of N-cTnC, create a domain that is amicable to Mg²⁺ binding [99, 100]. Despite previous work in this field, the central dogma in the literature is largely dismissive of the possibility that physiologically relevant concentrations of Mg²⁺ bind to site II. We previously explored Mg²⁺ binding to site II in full length and N-terminal cTnC and established competition with Ca²⁺ at physiological concentrations of each cation in the cell [51]. In this work, we explore the hypothesis that Ca²⁺ and Mg²⁺ compete, where affinity for each cation is allosterically modified by single amino acid changes outside the binding domain.

Tikunova and Davis have shown that Mg²⁺, unlike Ca²⁺, does not cause a structural change in the troponin complex upon binding but does significantly alter the affinity of cTnC for Ca²⁺ with 3 mM Mg²⁺ causing a more than 3-fold reduction in Ca²⁺ binding affinity. Moreover, Mg²⁺ reverses the fluorescence change of Ca²⁺ saturated cTnC; that is to say, their data supports the concept that 3 mM Mg²⁺ competes for binding to site II [3].

The molecular mechanisms which underpin the role of cellular Mg²⁺ in cardiac contractility are yet to be fully understood and require further exploration. We measured a 47-fold difference in the affinity of WT N-cTnC for Ca²⁺ (64.48 ± 3.09 ×10³ M⁻¹) in comparison to Mg²⁺ (1.43 ± 0.07 ×10³ M⁻¹). However, the free Mg²⁺ concentration is 3 orders of magnitude more abundant in the cytosol at systole than Ca²⁺ (1 μM vs. 1000 μM) [101, 102] and may compete for binding to site II in addition to the structural sites III and IV [51]. Mg²⁺ deficiency has been linked to cardiac disease including arrhythmias, hypertension, and congestive heart failure [103–106]. It is possible that Mg²⁺ modulates the role of Ca²⁺ and alters activation of contractile pathways that are governed by this messenger.

Less than 15 mins of ischemia can substantially decrease [ATP]_i resulting in a three-fold increase in free [Mg²⁺] [107]. This elevated Mg²⁺ may compete with Ca²⁺ for binding to cytosolic buffers such as cTnC. Overall, our study suggests that the effect of cellular Mg²⁺ on the Ca²⁺ binding properties of site II within N-cTnC is not negligible. This effect may be even more pronounced in HCM- and DCM-mutant N-cTnC, where both cytosolic concentrations of free Mg²⁺ (1 mM) and elevated Mg²⁺ that may accompany energy depleted states (3 mM) causing a more significant reduction in affinity compared to the WT through alterations in structural dynamics and the energetic landscape of each interaction.

Conclusions

The interaction of Ca²⁺ with mutant N-cTnCs occurred with higher than WT affinities, with the highest affinity seen in the L48Q mutant. In general, A31S, L48Q, Q50R, and C84Y had the highest affinities for both Ca²⁺ and Mg²⁺. Thermodynamic, structural, and simulation work by our group and others suggests a common mechanism whereby mutants destabilize hydrophobic interactions between helices NAD and BC elevate binding affinity.

We found that the affinity for Mg²⁺ (^~1.5 × 10³ M⁻¹) was at least an order of magnitude lower than that seen for Ca²⁺ (^~60 × 10³ M⁻¹). The change in affinity observed when comparing the Mg²⁺ pre-incubated N-cTnC and apo-state protein was variable in each mutant and significantly different from the WT. Moreover, 1 mM and 3 mM Mg²⁺ caused a graded decrease in the amount of binding and affinity for Ca²⁺. In contrast to Ca²⁺, cellular Mg²⁺ does not cause a conformational change upon binding to site II of cTnC and thus cannot initiate contraction. However, Mg²⁺ has been shown, both here and in numerous previous studies to interact with the same N-terminal locus. Cellular Mg²⁺ may be altered in disease states; for example, it may be elevated in ischemic stress or decreased as in hyperparathyroidism. Moreover, Mg²⁺-binding to cTnC may alter the already skewed Ca²⁺-cTnC binding interaction which exists in diseases such as HCM or DCM, further affecting significant changes in cardiomyocyte EC coupling.

Materials and Methods

Construct Preparations

Recombinant proteins were expressed and purified as described previously [66]. In brief, the human cTnC gene (TNNC1) within the pET-21a(+) vector was ordered from Novagen and the Phusion site directed mutagenesis kit (Thermo) was used to introduce a stop codon at residue 90, followed by single base pair changes to introduce all 6 variants of interest (A8V, L29Q, A31S, L48Q, Q50R, and C84Y) on separate N-terminal constructs (cTnC_1-89). Mutagenesis was carried out with preliminary steps using the DH5α E. coli strain to house the plasmids. Following the mutagenesis and confirmation by sequencing, the constructs were transformed into the BL21(DE3) expression strain and stored as glycerol stocks.

Protein Expression

100 mL of Lysogeny Broth (LB) supplemented with 50 μg/mL of Ampicillin and a stab of the glycerol stock was grown overnight at 37^⁰ C for 16 – 20 hrs with shaking at 225-250 rpm. 1 L flasks of LB were induced with 1-5% of the overnight culture and supplemented with the same concentration of antibiotic and grown under the same conditions for ^~3 hrs (until the OD₆₀₀ was between 0.8-1.0). The culture was then supplemented with 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) and grown for a further 3 – 4 hrs. Cells were then harvested by centrifugation and resuspended in the Lysis Buffer (50 mM Tris-Cl and 100 mM NaCl at pH 8.0). The suspended pellet was stored at −80° C until purification.

Protein Purification

The pellet was thawed and sonicated at ^~80% amplitude in 30 second intervals for a total time of 3 – 4 mins with each intermittent period spent on ice. The cells were then spun 2 times, for 15 minutes each at 30,000 ×g and the supernatant kept, and the pellet discarded. The supernatant was filtered as needed and applied to a 15 mL fast-flow DEAE or Q Sepharose column (GE Healthcare), pre-equilibrated with Buffer A (in mM 50 tris-Cl, 100 NaCl, and 1 dithiothreitol (DTT) at pH 8.0). Buffer B (Buffer A + 0.55 M NaCl) was applied over a 180 mL protocol, in which the concentration was ramped up from 0 to 100% to elute the proteins of interest. Fractions containing the N-terminal cTnC construct were identified by SDS PAGE and pooled. An Amicon centrifugal concentrator (Millipore) with a 3 KDa cut-off was used to concentrate the pooled samples to a volume of 3 – 5 mL. The pooled samples were then applied to a HiPrep 26/60 Sephacryl S-100 column (GE Healthcare) equilibrated with Buffer A. The fractions were again analyzed by SDS PAGE and those containing the protein of interest, free of contaminants were pooled, concentrated, and stored at −80° C.

ITC Protocol

The protein was dialyzed against 3 exchanges of 2 L for at least 6 hrs each with ITC Buffer 1 containing (in mM) 50 HEPES, 150 KCl, 2 EDTA, and 15 β-mercaptoethanol (BME) at pH 7.2. ITC Buffer 2 was identical to the first but did not contain EDTA; this was the second buffer used for equilibration and was used after the first to remove the EDTA contained in Buffer 1. ITC Buffer 3 was identical to the second but contained significantly reduced BME (2 mM). Buffer 3 was used to dilute the protein and Ca²⁺/Mg²⁺ prior to ITC experiments. A Nanodrop instrument was used to gain a preliminary measure of the protein concentration using an extinction coefficient of 1490 M⁻¹*cm⁻¹ and a molecular weight of 10.4 kDa. An initial ITC run was used to determine the molar ratio (N). Given that the concentration of the titrant is known and the number of cation binding sites in N-cTnC is 1, the concentration of folded, functional protein was determined and adjusted in subsequent runs to give an N of 1.0.

The protein was diluted in the final dialysis buffer to a final concentration of 100 μM. The titrating solutions were prepared from 1.00 M Ca²⁺ and Mg²⁺ stocks (Sigma) by serial dilution in the final dialysis buffer. For the apo-state experiments, 2 mM Ca²⁺ and 20 mM Mg²⁺ were titrated into 100 μM N-cTnC with the exception of the L48Q in which 2 mM Mg²⁺ was used. For competition experiments, the Apo-state construct was pre-incubated with Mg²⁺ to a final concentration of 1 mM or 3 mM Mg²⁺ prior to titration with 2 mM Ca²⁺. Titrant into buffer blank experiments were carried out to gauge the impact of these experiments and indicate minimal heat change resulting from the interactions (Figure S1). For all experiments, 19 titrations, 60 seconds apart were performed with the first being 0.8 μL and each subsequent injection 2 μL. The cell contents were mixed at 750 rpm throughout the titration. All titrations were carried out at 25° C.

Data Processing and Statistical Analysis

Data were imported and analyzed in Origin 8.0 software for Microcal ITC₂₀₀ (Northampton, MA). After saturation, the final 2-3 data points were averaged, the heat was subtracted from all injections as a control for heat of dilution and non-specific interactions. Least-squares regression was used to fit each titration after the first (dummy) injection was removed with minimization of chi-square and visual evaluation used to determine the goodness-of-fit for a single binding site model. Following establishment of the protein concentration based on the obtained N value for each apo-state Ca²⁺ titration for each construct, the same dilution of protein was used for each other titration and the N-value fixed to 1.00 to facilitate data fitting. The various thermodynamic parameters were averaged and reported as a mean ± SEM. The difference between the means was compared using a one-way ANOVA. This was followed by Tukey’s post-hoc test to determine where significant (p<0.05) differences existed (Figure 2).

Thermodynamic Integration (TI)

The structure of the N-terminal domain of cardiac troponin C (N-cTnC) was obtained from PDB:1AP4 [67], this structure contained N-cTnC with a single Ca²⁺ ion bound. Since there was no model of Mg²⁺ bound N-cTnC in the protein databank, we made use of the Mg²⁺ substituted structure as outlined in our previous work [51]. The model was then solvated using the tLeap module of AMBER 16 [68] in a 12$Å$ TIP3P water box and neutralized with Na⁺ ions; the forcefield used to describe the protein was ff14SB [69]. In order to complete the thermodynamic cycle, a system containing the Mg²⁺ ion was prepared using the tLeap module, referencing the ${\Delta G}_{solvation}$ optimized Mg²⁺ parameters from Li et. al. [70], and solvated in a 12$Å$ TIP3P water box. Simulations were conducted under NPT conditions using the Berendsen barostat and periodic boundary conditions. The system was minimized for 2000 cycles and heated to 300 K using the Langevin thermostat over 500 ps prior to the 5 ns production with a time step of 2 fs. The SHAKE algorithm was employed to constrain all bonds involving hydrogen atoms, and the Particle Mesh Ewald method [71] was utilized to calculate electrostatic interactions of long distances with a cutoff of 10$Å$.

The alchemical thermodynamic cycle used for ligand binding has been detailed previously by Leelananda and Lindert [72]. In this implementation of TI, the method consisted of three steps for ligand (Mg²⁺) in protein: 1) introduction of harmonic distance restraints; 2) removal of electrostatic interactions, and 3) removal of van der Waals forces. TI consisted of two steps for the ligand in water system: removal of electrostatic interactions and removal of van der Waals forces. The coupling parameter $\left(\lambda \right)$ increased incrementally by 0.1 from 0.0 – 1.0 for each transitional step of the thermodynamic cycle. During each simulation dV/d $\lambda$ values were collected every 2 ps resulting in 5000 data points per transitional step of $\lambda$ for further analysis. The Multistate Bennett Acceptance Ratio (MBAR) [73] was used to calculate the relative free energies of the simulations across all values of $\lambda$. Free energy $\left(\Delta G\right)$ corrections were made due to the introduction of the distance restraints and to correct for the charge of the system as described previously [51]. For each system, 5 independent runs were performed, and the results averaged. The specific distance restraints for all protein structures are shown in Table 1.

The cTnC variants (L48Q, Q50R, C84Y) were constructed using the protein mutagenesis tool in PyMOL[74][69] and the Mg²⁺ substituted representative model of N-cTnC serving as the base model. TI simulations were performed on the mutant structure as detailed above for the wild type.

Figure 4:

Representative isotherms are presented to quantify the heat signal from the interaction with each of Ca²⁺ and Mg²⁺ with the buffer. N-values do not present any utility and are included for completeness. The buffer titrations involve only transient binding interactions such as hydrogen bonds and Van der Waals forces if indeed they are quantifiable, between the solutes and solvents; this system contains no protein and no true binding sites. The scale may be used as measure of the heat change in the system and a proxy to compare with the titrant into N-cTnC conditions. The values measured when a single binding site model was used to fit each titration are presented in the table.

Figure 5:

Representative isotherms for each titration condition in the WT N-cTnC construct. From the right, the titration of Ca²⁺ into apo-state N-cTnC is shown, the next two panels show the titration of Ca²⁺ into 1 mM and 3 mM Mg²⁺ incubated WT N-cTnC. The right-most panel shows titration of Mg²⁺ into apo-state protein. Each titration is similarly endothermic with the scales indicating differences in absolute value of change in enthalpy.

Abbreviations:

BME

β-mercaptoethanol

cTnC

Cardiac troponin C

cTn

Cardiac troponin

TNNC1

cTnC gene

TnI_sw

cTnI switch peptide

DCM

Dilated Cardiomyopathy

DTT

Dithiothreitol

EC

Excitation-contraction

HCM

Hypertrophic Cardiomyopathy

IPTG

Isopropyl β-D-1-thiogalactopyranoside

ITC

Isothermal titration calorimetry

LB

Lysogeny Broth

N

Molar ratio

MD

Molecular Dynamics

MBAR

Multistate Bennett Acceptance Ratio

N-cTnC

N-terminal domain of cardiac troponin C

NMR

Nuclear Magnetic Resonance

TI

Thermodynamic Integration

TF

Thin filament