The calcium sensitizer drug MCI-154 binds the structural C-terminal domain of cardiac troponin C

The compound MCI-154 was previously shown to increase the calcium sensitivity of cardiac muscle contraction. Using solution NMR spectroscopy, we demonstrate that MCI-154 interacts with the calcium-sensing subunit of the cardiac troponin complex, cardiac troponin C (cTnC). Surprisingly, however, it binds only to the structural C-terminal domain of cTnC (cCTnC), and not to the regulatory N-terminal domain (cNTnC) that determines the calcium sensitivity of cardiac muscle. Physiologically, cTnC is always bound to cardiac troponin I (cTnI), so we examined its interaction with MCI-154 in the presence of two soluble constructs, cTnI1–77 and cTnI135–209, which contain all of the segments of cTnI known to interact with cTnC. Neither the cTnC-cTnI1–77 complex nor the cTnC-cTnI135–209 complex binds to MCI-154. Since residues 39–60 of cTnI are known to bind tightly to the cCTnC domain to form a structured core that is invariant throughout the cardiac cycle, we conclude that MCI-154 does not bind to cTnC when it is part of the intact cardiac troponin complex. Thus, MCI-154 likely exerts its calcium sensitizing effect by interacting with a target other than cardiac troponin.


Introduction
In heart failure, cardiac output is insufficient to satisfy the metabolic needs of the body. In acute decompensated heart failure, cardiac output becomes so compromised that, unless something is done imminently, clinical deterioration and death will ensue. Positive inotropes such as dobutamine and milrinone can increase cardiac output, but these are also associated with systemic hypotension, arrhythmias, and increased mortality [1].
Dobutamine and milrinone activate β 1 -adrenergic signaling pathways via direct binding to G-protein-coupled receptors in the case of dobutamine or dual phosphodiesterase-3 and − 4 inhibition by milrinone [2,3]. This leads to increased cyclic AMP concentrations and activation of protein kinase A, which phosphorylates many different targets within the cardiac myocyte including L-type calcium channels and phospholamban, which regulates sarcoplasmic reticulum Ca 2+ -ATPase (SERCA) [4]. Increased calcium fluxes enhance cardiac output, but at the expense of increased myocardial oxygen consumption and increased risk of arrhythmias. An alternative approach is to use "calcium sensitizers", drugs that impact muscle function not by altering calcium fluxes, but by enhancing the contractile response to calcium. The best known cardiac calcium sensitizer is levosimendan, which has undergone many trials for the treatment of acute decompensated heart failure [5].
Many calcium sensitizers are purported to act on the cardiac troponin complex (cTn), the key calcium dependent switch that turns muscle contraction on and off with every heartbeat. It consists of three protein subunits: C, I, and T. The calcium dependence of the complex derives from troponin C (cTnC), which has two calcium-binding EFhand domains. The N-terminal regulatory domain (cNTnC) determines the contractile state of the heart. Its calcium affinity is tuned to calcium fluctuations in the cardiac myocyte, so that it becomes significantly occupied only during systole, when the cytoplasmic free calcium concentration rises to about 1 μM (for a review, see [6]). Calcium binding stabilizes the open conformation of cNTnC, allowing it to bind the switch region of cardiac troponin I (cTnI 147-163 ) [7]. This removes adjacent inhibitory cTnI segments from actin, favoring activation of actin-myosin ATPase and cardiac muscle contraction.
In theory, calcium sensitizers could act by stabilizing the calciumbound activated form of cNTnC. This has been convincingly demonstrated for bepridil, a drug originally designed to treat angina by blocking calcium channels, but serendipitously shown to also act as a calcium sensitizer [8]. It was later demonstrated that bepridil binds cNTnC to stabilize its calcium-bound open conformation [9]. Bepridil, however, is not an effective troponin activator because it displaces cTnI switch peptide from cNTnC [10]. We thus searched for compounds that bind to cNTnC without displacing cTnI switch peptide and discovered that derivatives of diphenylamine are effective for this purpose [11]. A structurally similar compound is MCI-154 ( Fig. 1), which was previously developed by Mitsubishi Chemical Corporation as a calcium sensitizing agent. It was shown to increase the calcium sensitivity of cardiac myofibrils as well as isolated cardiac troponin C [12,13]. Thus, out of all the calcium sensitizers developed to date, MCI-154 had some of the strongest evidence for a mechanism of calcium sensitization through direct binding to cardiac troponin C [14]. We thus set out to more thoroughly investigate the interaction of MCI-154 with cTnC.
Reactions were carried out in flame-dried glassware. Transfer of anhydrous solvents and reagents was accomplished with oven-dried syringes. Dimethylformamide (DMF) was distilled before use from calcium hydride; N-methylpyrrolidone (NMP) was dried over 4 Å molecular sieves. Thin layer chromatography was performed on glass plates pre-coated with 0.25 mm Kieselgel 60 F254 (Merck). Flash chromatography columns were packed with 230-400 mesh silica gel (Silicycle). 1 H NMR were measured on Agilent/Varian DD2 MR two channel 400 MHz spectrometer and are reported in ppm relative to tetramethylsilane (0.00 ppm) standard; coupling constants (J) are reported in Hertz (Hz). Standard notation is used to describe the multiplicity of signals observed in 1 H NMR spectra: singlet (s), doublet (d), triplet (t), etc. Please see the Supplementary Materials for the NMR spectra of MCI-154 and compounds 3-5.

4-(4-aminophenyl)-4-oxobutanoic acid (4)
Into the round bottom flask charged with 4-(4-acetamidophenyl)-4oxobutanoic acid 3 (3. 102 g, 13.19 mmol) was added conc. hydrochloric acid (14 mL) and the reaction mixture was refluxed for 30 min. After cooling down to room temperature the mixture was diluted with water (28 mL) and the acidity was adjusted to pH= 4 at 10°C by using aqueous saturated solution of sodium carbonate. The product was collected by filtration,washed with water and dried. The product was obtained as off-white powder (0.869 g, 4.46 mmol, 34% yield). 1

Protein sample preparation
Recombinant human cardiac cTnC (residues 1-161) with the mutations C35S and C84S, cTnI (residues 1-77), and cTnI (residues 135-209) were used in this study. The expression and purification of 15 N-cTnC in E. coli were as described previously [18]; the expression and purification of cTnI (residues 1-77) and cTnI (residues 135-209) were as described in recent publications [19,20]. cTnI switch peptide (residues 147-163) was purchased from GL Biochem (Shanghai) Ltd. Stock solutions of MCI-154, in d 6 -DMSO, were prepared, and the vials containing the solutions were wrapped in aluminum foil to protect the molecules from light-catalyzed degradation. For NMR sample preparations, solid 15 N-labeled cTnC was dissolved into 500 μL NMR buffer containing 100 mM KCl, 10 mM imidazole, 0.5 mM DSS in 90% H 2 O/ 10% D 2 O to generate a 450 μM NMR sample. 5 μL of 1 M CaCl 2 was added to ensure that the protein was Ca 2+ -saturated and the pH was adjusted by 1 M NaOH and 1 M HCl to~6.7. For cTnC-cTnI peptide complexes, a 1:1 molar ratio of solid unlabeled cTnI 1-77 or cTnI  peptide was added to solid 15 N-labeled cTnC prior to dissolution in buffer.

NMR Spectroscopy
All NMR experiments were run on a Varian Inova 500 MHz spectrometer at 30°C. The spectrometer is equipped with a triple resonance 1 H, 13 C, 15 N-probe and z-axis pulsed field gradients. Both 1D 1 H and 2D 15 N}-HSQC NMR spectra were acquired at every titration point. The dissociation constant (K D ) for MCI-154 binding to cTnC•3Ca 2+ was calculated by plotting chemical shift changes as a function of the ligand-to-protein ratio and then fitting the values to a function using our in-house curve-fitting software, xcrvfit (www.bionmr.ualberta.ca/ bds/software/xcrvfit). The function relating the predicted change in chemical shift ( ) to total protein (P) and total ligand concentrations (L) is as follows: where max is the change in chemical shift expected at 100% saturation and K D is the dissociation constant for the 1:1 protein-ligand complex. A χ 2 function measuring the sum of differences between observed and predicted values was minimized, using K D and max as fitting parameters.

Titration of MCI-154 to 15 N-cTnC•3Ca 2+
For NMR sample preparation, 4.8 mg of solid 15 N-cTnC were dissolved into 500 μL NMR buffer containing 100 mM KCl, 10 mM imidazole, 0.5 mM DSS in 90% H 2 O/10% D 2 O to generate a 450 μM NMR sample (protein concentration was estimated based on weight and integration of 1D 1 H NMR spectroscopy). 5 μL of 1 M CaCl 2 was added to ensure that the protein was Ca 2+ -saturated and the pH was adjusted by 1 M NaOH and 1 M HCl to~6.7. Aliquots of 1, 1, 1, 1, 0.5, 1.5, 2, 5 μL of 90.2 mM MCI-154 in d 6 -DMSO were added consecutively to the 15 N-cTnC•3Ca 2+ sample. The sample was mixed thoroughly with each addition. The change in protein concentration due to volume increase was taken into account for data analyses. The pH changes from ligand additions were compensated by 1 M NaOH or 1 M HCl. Both 1D 1 H and 2D { 1 H, 15 N}-HSQC NMR spectra were acquired at every titration point.

Isothermal Titration Calorimetry
Isothermal titration calorimetry (ITC) was carried out on a Malvern VP-ITC MicroCalorimeter at 25°C (note that the NMR experiments were performed at 30°C), and data were analyzed using the original ITC analysis software package (MicroCal). 2 mM MCI-154 was titrated into 200 μM purified cTnC protein, both in buffer containing 50 mM imidazole at pH 6.8, 100 mM KCl, 10 mM CaCl 2 , and 1% DMSO. The initial injection volume was 4 μL, followed by 15 μL injections spaced 300 s apart. Data were corrected and fitted to a single binding site model, from which apparent molar reaction enthalpy (ΔH), entropy (ΔS), and dissociation constant (K D ) were generated.

MCI-154 binds to the structural C-terminal domain of cTnC
MCI-154 was titrated into isolated calcium-saturated 15 N-enriched cTnC, and binding was followed by solution 2D { 1 H, 15 N} HSQC NMR spectra. Chemical shift changes were observed in residues from the cCTnC domain, but none was observed in the cNTnC domain (Fig. 2). This is somewhat surprising, since one would expect a calcium sensitizer to interact with the regulatory cNTnC domain, which determines the calcium sensitivity of cardiac muscle. (It should also be noted that titration of cTnC with DMSO yields no chemical shift changes.) With each titration point, cCTnC signals progressively move from the unbound state to an MCI-154-bound state without significant broadening, indicating that MCI-154 binding occurs with fast kinetics relative to the NMR frequency differences between free and bound cCTnC states. The MCI-154-induced total chemical shift changes (Δδ) of each affected residue was calculated as follows: Δδ= [(Δδ 1H ) 2 + (Δδ 15N /5) 2 ] 1/2 , then averaged for a group of residues that underwent significant chemical shift changes. MCI-154 appears to bind to the central hydrophobic cavity of the cCTnC domain (where all small molecules have been found to bind, see Discussion section), as shown by chemical shift perturbation mapping (Fig. 1). The linear movement of the signals as the titration progressed is suggestive of 1:1 stoichiometry. The dissociation constant (K D ) of MCI-154 for cTnC was calculated as described in the Methods section. Fitting of the titration points yields a binding dissociation constant, K D , of about 0.5 mM (Fig. 3). This affinity is much weaker than the observed calcium sensitizing activity of MCI-154, which was~50% active at concentrations of 1-10 μM [21], suggesting that the interaction between MCI-154 and cTnC cannot account for its biological activity.
Isothermal titration calorimetry (ITC) was used to confirm the binding affinity. A K D value of 0.5 mM ± 0.1 mM was obtained, in perfect agreement with the value obtained by NMR. In addition, ΔH of binding was determined to be −2.5 kcal/mol, with ΔS= +6.6 cal/ mol K (see Fig. 4). The stoichiometry of binding could not be precisely determined using ITC.  cTnC is strongly tethered to the rest of the cardiac troponin complex by a tight interaction with residues 39-60 of cTnI, which form an amphipathic helix that occupies the large hydrophobic patch in the cCTnC domain [22]. This is the strongest interaction within the troponin complex (nanomolar affinity). We previously demonstrated that residues 19-37 of cTnI interact electrostatically with the regulatory cNTnC domain, and this interaction fixes the positioning of the cNTnC domain relative to cCTnC and the rest of the troponin complex [23].

Binding of full-length cTnC to cTnI 147-163 switch peptide
It is very well established that the binding of cTnI switch region, cTnI 147-163 , to cNTnC constitutes the key event that triggers cardiac muscle contraction. It was somewhat surprising that binding of cTnI 135-209 to cTnC should eliminate binding of MCI-154 to the cCTnC domain (see Results Section 3.3 above), since this segment of cTnI was not known to bind the cCTnC domain. To gain more insight into this, we titrated a shorter peptide corresponding to the switch region, cTnI 147-163 , into full-length cTnC. Interestingly, large chemical shift changes were induced in both domains of cTnC (Fig. 5). Given the similar magnitude of chemical shift changes in both domains, one possible explanation is that the cCTnC domain is able to bind the switch region as well as the cNTnC domain. Another possibility is that the cNTnC-cTnI 147-163 complex interacts with the cCTnC domain. In either case, we have previously shown that when the cCTnC domain is bound to cTnI 37-71 , addition of cTnI 147-163 into full-length cTnC impacts only the cNTnC domain, and not the cCTnC domain [24]. Apparently, the cCTnC domain is able to bind promiscuously to a variety of peptides and small molecules in the absence of its physiologic binding partner: cTnI residues 39-60.

Discussion
The binding of the positive inotropic compound MCI-154 to cTnC has not previously been studied by NMR. Previous studies indicated that it directly increased the calcium binding affinity of free cTnC [25][26][27][28]. It is possible that it enhances the affinity of the cCTnC domain by binding to its central hydrophobic cavity, thus stabilizing the calcium-bound open form of the domain. The lack of binding to the cNTnC domain is consistent with our recent study of diphenylamine analogs binding to a cNTnC-cTnI 136-163 chimeric construct [11]. Bulky substituents in the para position of diphenylamine (as is present in MCI-154, Fig. 1) were not well tolerated. Hence, the positive inotropic effect of MCI-154 in experimental models is most likely related to a mechanism distinct from cTnC binding. MCI-154 has potent inhibitory activity towards type 3 phosphodiesterase, and there is some indication that it may also bind cardiac myosin [29]. These two proteins appear to be among the most promiscuous proteins in cardiomyocytes when it comes to binding small organic molecules.
The structural cCTnC domain of cardiac troponin C is like a "hyperfunctioning" version of the regulatory cNTnC domain, binding the primary physiologic target sequence of the cNTnC domain, the switch region of cTnI. It also binds to molecules that cNTnC cannot (like MCI-154, EMD57033, EGCg, and resveratrol), in addition to some that cNTnC can also bind (like trifluoperazine, bepridil, W7, and levosimendan) (for reviews, see [6,30]). Of the compounds that bind exclusively to the cCTnC domain, MCI-154 binds with a similar affinity (K D~5 00 μM for MCI-154, compared with 10 μM for EMD 57033 [31,32], 1120 μM for EGCg [33], and 243 μM for resveratrol [34]). All of these compounds appear to be displaced from the central hydrophobic cavity of the cCTnC domain in the presence of cTnI. We therefore recommend caution in the interpretation of binding and functional studies involving free uncomplexed cTnC [35]. Moreover, it is possible that studies using reconstituted cardiac troponin complex could be influenced by the presence of free cTnC due to stoichiometric excess of cTnC or incomplete incorporation into the troponin complex.