Structure and Dynamics of the C-domain of Human Cardiac Troponin C in Complex with the Inhibitory Region of Human Cardiac Troponin I*

Cardiac troponin C is the Ca 2 (cid:1) -dependent switch for heart muscle contraction. Troponin C is associated with various other proteins including troponin I and troponin T. The interaction between the subunits within the troponin complex is of critical importance in understanding contractility. Following a Ca 2 (cid:1) signal to begin contraction, the inhibitory region of troponin I comprising residues Thr 128 –Arg 147 relocates from its binding surface on actin to troponin C, triggering movement of troponin-tropomyosin within the thin filament and thereby freeing actin-binding site(s) for interactions with the myosin ATPase of the thick filament to generate the power stroke. The structure of calcium-satu-rated cardiac troponin C (C-domain) in complex with the inhibitory region of troponin I was determined using multinuclear and multidimensional nuclear magnetic resonance spectroscopy. The structure of this complex reveals that the inhibitory region adopts a helical conformation spanning residues Leu 134 –Lys 139 , with a novel orientation between the E- and H-helices of troponin C, which is largely stabilized by electrostatic interactions. By using isotope labeling, we have studied the dynamics of the protein and peptide in the binary complex. The structure of this inhibited complex provides a framework for understanding into interactions within the troponin complex upon heart contraction. The binding of Ca contacts between protein-protein, peptide-peptide, and protein-peptide were derived from various three-dimensional 13 C/ 15 N-NOESY experiments. The three-dimensional 13 C-edited NOESY experiment per- formed on 13 C/ 15 N-cCTnC complexed with 13 C/ 15 N-cIp-r was used to define protein-peptide contacts with use of symmetrical peaks within the three-dimensional planes for use in unambiguous NOE assignments. All protein-peptide NOEs were calibrated to 4 (cid:3) 2 Å. An initial set of 100 structures of both cCTnC and cIp was first generated sepa-rately using NOE distance restraints only. NOEs with a distance vio- lation of 0.2 Å or greater were closely examined prior to further rounds of structure refinements. In the latest stages of refinement, (cid:3) angles of cCTnC were added for residues located in well defined regions, as determined using the program Procheck (49). In the first round of structure calculations of the cCTnC-cIp complex, an initial set of 100 structures was determined using pre-folded structures of cCTnC and cIp respectively, with unambiguous protein-peptide NOEs being intro-duced. Further refinement of structures of the complex using the initial set of 100 structures was performed using the program Procheck to inspect structures, as well as closely examining NOE distance viola- tions greater than 0.1 Å. To ensure independent folding of the complex from pre-folded cCTnC and cIp structures, final set of 100 structures derived from both cCTnC and cIp starting in extended conforma-tions. annealing Ca (cid:1) -distance cCTnC the linker region the

The binding of Ca 2ϩ to the troponin complex initiates cardiac muscle contraction (1)(2)(3)(4)(5). The troponin complex is composed of three subunits: troponin C (TnC), 1 troponin I (TnI), and tropo-nin T (TnT) (6). The three subunits of the troponin complex are necessary for Ca 2ϩ -induced regulation of cardiac muscle contractility. TnC, the Ca 2ϩ -sensitive component of the complex, is a member of the EF-hand family of Ca 2ϩ -binding proteins and contains two high affinity Ca 2ϩ /Mg 2ϩ -binding sites (sites III and IV) in the C-terminal domain and one low affinity Ca 2ϩbinding site (site II) in the N-terminal domain (7)(8)(9). At physiological conditions during muscle relaxation, the two C-terminal Ca 2ϩ /Mg 2ϩ -binding sites are occupied, and the N-terminal Ca 2ϩ -binding site is unoccupied. During the onset of muscle contraction, a transient increase in cytosolic Ca 2ϩ concentrations allows the low affinity N-terminal domain to bind Ca 2ϩ , resulting in the initiation of heart contraction (10). TnI is the subunit that in the presence of tropomyosin inhibits myosin Mg 2ϩ -ATPase activity. TnI inhibition is removed by the binding of Ca 2ϩ to the N-terminal domain of the TnC subunit (11). TnT is the subunit that binds tropomyosin, TnC, and TnI, anchoring the troponin complex to the thin filament and aids in the propagation of Ca 2ϩ -induced conformational changes (12,13).
The structure of cardiac TnC, sharing sequence and structural similarities to the skeletal isoform, has been solved by both x-ray crystallography and NMR spectroscopy (7,(15)(16)(17)(18)(19). Upon Ca 2ϩ binding, there is an opening of the N-domain of the skeletal isoform, whereas the cardiac isoform remains closed, opening only upon binding to cTnI-(147-163) (cSp) (16). The structure of the complete troponin complex is of critical importance in understanding molecular interactions during muscle contraction, yet only low resolution structures of the complex and several structures of TnC in complex with different TnI peptides are currently available (16,18,(21)(22)(23)(24), giving insights into muscle contraction and regulation. The numbering of the residues of cTnI in this study are as presented previously (25), in contrast to the numbering of the wild type protein, which has an extra N-terminal Met residue (26).
TnI-TnC interactions have been studied extensively for over 3 decades. The functionality of domains of TnC was determined early by Head and Perry (27) by using various proteolytic/ cleavage techniques. These studies of rabbit skeletal muscle yielded a region of TnI that has been referred to as the "inhibitory region." Subsequent work by Talbot and Hodges (28) identified a refined inhibitory region, which has been mapped to a central region of its primary amino acid sequence, corresponding to residues Thr 128 -Arg 147 in the cardiac system (cIp). The inhibitory region contains the minimum sequence required * This work was supported in part by the Canadian Institutes of Health Research and the Heart and Stroke Foundation of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
to fully inhibit the myosin ATPase activity on the thin filament (29). Interactions of the cardiac TnI inhibitory region are of key importance in complete understanding of muscle regulation. It has been shown that the inhibitory region interacts with the protein actin on the thin filament during muscle relaxation and that movement of cIp off actin interacts with TnC upon Ca 2ϩinduced muscle contraction. Previously, it has been shown that the inhibitory region of TnI binds to the C-domain of TnC and the central linker region of the N-and C-terminal domains of TnC (30). Additional interactions of TnC with other regions of TnI have been elucidated (31,32), resulting in an intertwined system with multiple contacts between the subunits of the troponin complex.
Whereas elucidation of the structure of the inhibitory region of TnI in complex with the troponin complex is of key importance in the understanding of muscle contraction, the structure has remained elusive, and consequently many models have been proposed. Early 1 H NMR study of the skeletal TnI isoform involving transferred NOEs by Campbell and co-workers (33,34) led to the proposal that the inhibitory region, sTnI-(104 -115) (cTnI-(136 -147)), adopts a short helix, distorted around two central proline residues, and this structure was subsequently docked within the hydrophobic cleft on sCTnC (35). The crystal structure of sTnC in complex with the N-terminal regulatory peptide of sTnI-(1-47) (Rp40) presented by Vassylyev et al. (21) showed that Rp40 adopts a helical structure and binds in the hydrophobic cleft of sCTnC, and these workers modeled the inhibitory region as adopting a helical conformation away from the hydrophobic cleft. The helical conformation of the inhibitory region was directly challenged by Hernandez et al. (36), who proposed an extended conformation of the inhibitory region, with a two-stranded ␤-hairpin away from the hydrophobic cleft of the C-domain (37).
Electron spin labeling work by Brown et al. (38) has shown that a section of the inhibitory region of the cardiac isoform (cTnI-(129 -137)) displays a helical profile, with the C-terminal residues 139 -145 displaying no discernible secondary structural characteristics. The inhibitory region possesses a large number of basic residues, and Tripet et al. (39) predicted that this highly basic region of cTnI makes numerous electrostatic interactions with the acidic TnC in the troponin complex. Recent work on the cardiac isoform using residual dipolar coupling by Dvoretsky et al. (19) determined the orientations of the domains of cTnC within a cTnI⅐cTnC complex, and a small angle scattering study by Heller et al. (40) has determined relative domain orientation within a cTnI⅐cTnC⅐cTnT-(198 -298) complex, which suggests that interactions between TnC and the TnI⅐TnT components differ significantly between the skeletal and cardiac isoforms. A preliminary crystallographic structure of a TnT⅐TnC⅐TnI ternary complex by Takeda et al. (41) indicated a coiled-coil region of cTnI⅐cTnT with multiple interactions with cTnC; however, the inhibitory region spanning cIp was not visualized within the structure.
We have been successful in elucidating the NMR solution structure of the cardiac isoform of the inhibitory region of TnI (cTnI-(128 -147)) in complex with the Ca 2ϩ -saturated C-terminal domain of TnC. The inhibitory region displays a helical secondary structure from residues Leu 134 -Lys 139 , with several stabilizing electrostatic interactions with cCTnC. The structure correlates well with previous NMR chemical shift mapping of the interactions of the inhibitory region with TnC (22,30,42). The ability to isotopically label the inhibitory region has given us the unique ability to utilize 15 N NMR relaxation to study dynamics of the bound inhibitory region. NMR relaxation indicates that the central core of cIp is rigid when bound to cCTnC, giving validation to the structure. This is the first high resolution structure determined for the inhibitory region of TnI in complex with TnC and provides a framework for understanding interactions within the troponin complex during heart contraction.

EXPERIMENTAL PROCEDURES
Preparation of cCTnC Protein-The engineering of the expression vector for the cCTnC-(90 -161) protein was as described by Chandra et al. (43). The expression and purification of cCTnC, 15 N-cCTnC, and 13 C/ 15 N-cCTnC proteins in Escherichia coli followed the procedure as described previously (44) for sNTnC. CCTnC-labeled proteins were further purified using a gravity flow Superdex 75 column (Amersham Biosciences) as described previously (42) for cCTnC.
Preparation of cIp Protein-cIp-s peptide acetyl-TQKIFDLRGK-FKRPTLRRVR-amide was prepared as described for the sIp peptide in Tripet et al. (32). The engineering of the expression vector for expression and purification of cIp-r, 15 N-cIp-r, and 13 C/ 15 N-cIp-r via a fusion protein approach has been described previously (42,45). The procedure for purification of the GB-1-cIp-His fusion protein was modified as follows. Fusion proteins were eluted in 100 mM EDTA, 500 mM NaCl, 20 mM Tris-HCl, pH 7.9. Eluted proteins were then desalted by loading on a Sephadex G-25 medium (Amersham Biosciences) column in 10 mM NH 4 HCO 3 , pH 8.0, and lyophilized to dryness.
Titration of 15 N-cCTnC⅐2Ca 2ϩ with cIp-The titration is as described previously (42). Both one-dimensional 1 H and two-dimensional { 1 H, 15 N}-HSQC spectra were acquired at every titration point. In order to check if the cIp-s and cIp-r peptides bind the same way to cCTnC⅐2Ca 2ϩ , 8.1 mg of solid cIp-r was added to a 0.92 mM 15 15 N}-HSQC spectrum was acquired and superimposed with that of the cCTnC⅐2Ca 2ϩ ⅐cIp complex.
Titration of 15 N-cIp-r with cCTnC⅐2Ca 2ϩ -5.11 mg of recombinant 15 N-cIp-r was dissolved in 550 l of NMR buffer, pH 6.7, containing 25 l of 1 M CaCl 2 , and 500 l was transferred to an NMR tube. Solid cCTnC was added in 1.2-mg additions, with thorough mixing after each addition. Both one-dimensional 1 H and two-dimensional { 1 H, 15 N}-HSQC spectra were acquired at every titration point. After every titration point, 1 l of the resulting titrated solution was removed and used for amino acid analysis. Solid cCTnC was added until no further changes were observed in the HSQC spectra, ensuring complete cIp-r saturation. The change in cIp-r concentration due to changes in volume during the titration was taken into account for data analysis, and the change in pH for cCTnC addition was corrected by addition of NaOH during each addition. A small amount of white precipitate accumulated as increasing amounts of solid cCTnC was added. Structural Studies on cCTnC⅐2Ca 2ϩ ⅐cIp Complex-Three samples were prepared for structural studies on the complex, with subsequent structural determination carried out using the NMR experiments listed in Table I. 10.13 mg of 13 C/ 15 N-cCTnC was dissolved in 600 l of NMR buffer, pH 6.7, containing 5 l of 1 M CaCl 2 to which 6.23 mg of cIp-s was added, and 500 l was added to an NMR tube. 3.34 mg of 13 C/ 15 N-cIp-r was dissolved in 600 l of NMR buffer, pH 6.7, containing 5 l of 1 M CaCl 2 to which 25.5 mg of cCTnC was added to ensure complete cIp-r saturation, and 500 l was added to an NMR tube. 5.06 mg of 13 C/ 15 N-cIp-r was dissolved with 16.44 mg of 13 C/ 15 N-cCTnC to ensure a 1:1 molar ratio in 575 l of NMR buffer (100% D 2 O, pH 6.7) containing 5 l of 1 M CaCl 2 , and 500 l was added to an NMR tube.
NMR Spectroscopy-All NMR data used in this study were acquired at 30°C using Varian INOVA 500 MHz, Unity 600 MHz, and INOVA 800 MHz spectrometers. All three spectrometers are equipped with triple resonance probes and z axis pulsed field gradients (xyz gradients for the 800 MHz). For the cCTnC⅐cIp complex, the chemical shift assignments of the backbone and the side chain atoms and NOE interproton distance restraints were determined using the two-dimensional and three-dimensional NMR experiments described in Table I.
Data Processing and Peak Calibration-All two-dimensional and three-dimensional NMR data were processed using NMRPipe (46), and all one-dimensional NMR data were processed using VNMR (Varian Associates). All spectra were analyzed using NMRView (47). For cCTnC and cIp in the complex, intramolecular distance restraints obtained from the NOESY experiments were calibrated according to Gagne et al. (48). Dihedral angle restraints were derived from data obtained from HNHA and NOESY-HSQC experiments according to Sia et al. (7).
Structural Calculations for Binary Complex cCTnC⅐2Ca 2ϩ ⅐cIp-NOE contacts between protein-protein, peptide-peptide, and protein-peptide were derived from various three-dimensional 13 C/ 15 N-NOESY experiments. The three-dimensional 13 C-edited NOESY experiment performed on 13 C/ 15 N-cCTnC complexed with 13 C/ 15 N-cIp-r was used to define protein-peptide contacts with use of symmetrical peaks within the three-dimensional planes for use in unambiguous NOE assignments. All protein-peptide NOEs were calibrated to 4 Ϯ 2 Å. An initial set of 100 structures of both cCTnC and cIp was first generated separately using NOE distance restraints only. NOEs with a distance violation of 0.2 Å or greater were closely examined prior to further rounds of structure refinements. In the latest stages of refinement, angles of cCTnC were added for residues located in well defined regions, as determined using the program Procheck (49). In the first round of structure calculations of the cCTnC-cIp complex, an initial set of 100 structures was determined using pre-folded structures of cCTnC and cIp respectively, with unambiguous protein-peptide NOEs being introduced. Further refinement of structures of the complex using the initial set of 100 structures was performed using the program Procheck to inspect structures, as well as closely examining NOE distance violations greater than 0.1 Å. To ensure independent folding of the complex from pre-folded cCTnC and cIp structures, a final set of 100 structures was derived from both cCTnC and cIp starting in extended conformations. The default CNS (50) built-in annealing protocol was modified to allow the introduction of 12 Ca 2ϩ -distance restraints preceding the second cooling step using Cartesian dynamics. The structural statistics of the family of the 30 calculated lowest energy structures is shown in Table II. A comparison of interhelical angles of the binary complex with previously solved EF-hand structures is shown in Table III.
Backbone Amide 15 N Relaxation Measurements of cCTnC⅐2Ca 2ϩ ⅐cIp Complex-Two samples of 15 N-cCTnC were prepared, one of cCTnC⅐2Ca 2ϩ free in solution and the other with binary complex cCTnC⅐2Ca 2ϩ ⅐cIp-s. A first sample of 11.82 mg of 15 N-cCTnC was dissolved in 600 l of NMR buffer, pH 6.7, containing 5 l of 1 M CaCl 2 , and 500 l was added to an NMR tube. A second sample of 12.49 mg of 15 N-cCTnC was dissolved in 600 l of NMR buffer, pH 6.7, containing 5 l of 1 M CaCl 2 to which 6.69 mg of cIp-s peptide was added to ensure complete cCTnC saturation, and 500 l was added to an NMR tube. Two samples of 15 N-cIp-r were prepared, one of cIp-r free in solution and the other with cIp-r⅐cCTnC⅐2Ca 2ϩ in complex. 0.8 mg of recombinant 15 N-cIp-r was dissolved in 550 l of NMR buffer, pH 6.7, containing 5 l of 1 M CaCl 2 , and 500 l was added to an NMR tube. A second sample of 1.88 mg of recombinant 15 N-cIp-r was dissolved 600 l of NMR buffer, pH 6.7, containing 5 l of 1 M CaCl 2 to which 14.5 mg of 1 H-cCTnC was dissolved to ensure complete cIp-r saturation, and 500 l was added to an NMR tube.
All relaxation data were acquired on Varian INOVA 500 MHz and Unity 600 MHz spectrometers at 30°C. Sensitivity enhanced pulse sequences developed by Farrow et al. (51) were used to measure 15 N-T 1 , 15 N-T 2 , and { 1 H}-15 N NOE (where T 1 is longitudinal (spin-lattice) and T 2 is transverse (spin-spin) relaxation). By using two-dimensional spectroscopy a set of backbone 15 N-T 1 , 15 N-T 2 , and { 1 H}-15 N NOE experiments were collected for each sample with parameters described in Table IV. The delay between repetitions of the pulse sequence was set to 3 s for both the T 1 and T 2 experiments. { 1 H}-15 N NOE measurements were made in the absence (relaxation delay incorporation of 5 s between spectrometer pulses) and the presence or proton saturation (incorporation of 3 s of 1 H saturation, with a delay of 2 s between spectrometer pulses). All relaxation data were processed using NMRPipe (46) and analyzed using NMRView (47).
Coordinates-The coordinates for the structure of cCTnC⅐2Ca 2ϩ ⅐cIp have been deposited in the Protein Data Bank (code 10ZS).

RESULTS
Titration of cCTnC⅐2Ca 2ϩ with cIp-The titration of 15 N-cCTnC⅐2Ca 2ϩ with cIp-s, as monitored by two-dimensional { 1 H, 15 N}-HSQC NMR spectral changes, is shown in Fig. 1A. We also performed the reverse titration of 15 N-cIp-r titrated with cCTnC to monitor chemical shift changes in cIp-r, as shown in Fig. 1B. The two-dimensional { 1 H, 15 N}-HSQC NMR spectra of both cCTnC⅐2Ca 2ϩ and cIp-r have been completely assigned and were used as starting points to monitor protein-peptide chemical shift changes. All of the chemical shift changes for both titrations fall into the fast exchange limit on the NMR time scale, so that each cross-peak corresponds to the weighted average of the bound and free chemical shifts. Linear movement of the cross-peaks indicates that only two species exist in solution; also the binding of cCTnC and cIp occurs in a 1:1 stoichiometry, as described previously (42). Resonances undergoing large backbone amide 1 H and/or 15 N chemical shifts were followed to monitor peptide-protein binding. The normalized chemical shift data were fit to Equation 1, which yielded a dissociation constant (K D ) of 31 Ϯ 11 M for 15 N-cCTnC⅐2Ca 2ϩ titrated with cIp-s, as reported previously (42), which is in the same order as the equivalent titration for the skeletal isoform (42 Ϯ 7 M) which our laboratory has reported previously (22). The titration of 15 N-cIp-r with cCTnC is expected to produce the same K D ; however, when solid cCTnC was added to the NMR tube containing 15 N-cIp-r, some white precipitate was observed after the initial addition of cCTnC, and the level of precipitate increased as increasing levels of cCTnC were added; thus an accurate K D value was not obtained for this titration. cIp induces significant chemical shift perturbations of the amide resonances of 15 N-cCTnC, with residues in the E-helix, H-helix, and the linker region showing the greatest changes. Superimposition of the { 1 H, 15 N}-HSQC NMR spectra of cCTnC⅐2Ca 2ϩ ⅐cIp-s and cCTnC⅐2Ca 2ϩ ⅐cIp-r revealed that the spectra were identical, indicating that both recombinant and synthetic cIp peptides used in this study perturb cCTnC in an identical manner. Monitoring of the reverse titration of 15 N-cIp-r with cCTnC also reveals large chemical shift perturbation, with the largest degree of perturbation in the central region of cIp-r peptide. Specified residues undergoing significant chemical shift perturbation are labeled on Fig. 1, A and B. Two species of hSer are observed in Fig. 1B, showing resonances of hSer and hSer-lactone species present on the C termini of the peptide cIp-r, neither of which undergo chemical shift perturbation, indicating that the C termini residues of cIp-r are not involved in binding of the complex.
Overall Structure of the cCTnC⅐2Ca 2ϩ ⅐cIp Complex-The NMR experiments performed to obtain structural data for the TnC⅐TnI complex are summarized in Table I. A sample of 13 C/ 15 N-cCTnC in complex with cIp-s gave a two-dimensional { 1 H, 15 N}-HSQC NMR spectra that was highly resolved (Fig.  1A), allowing the assignment of chemical shifts of the cCTnC backbone and side chain nuclei. Previously, our laboratory has obtained data for protein-peptide/ligand complexes using 13 C/ 15 N-labeled protein in complex with unlabeled peptides/ligands (16 -18), using two-dimensional 13 C/ 15 N-filtered DIPSI and NOESY experiments (52) to obtain chemical shift data of the bound peptide/ligand in the complex. Production of 13 C/ 15 N-cIp peptide was essential for this study as information obtained from standard filtered experiments produced very limited information, suggesting few close hydrophobic contacts, which would contribute to NOE cross-peaks, occur between the protein and the peptide in the binary complex (data not shown). Edited experiments of 13 C/ 15 N-cIp complexed with unlabeled cCTnC allowed for complete chemical shift assignment of cIp atoms in the bound binary complex. Three samples were used in this study to obtain NMR structural restraint data for the binary complex: 13 C/ 15 N-cCTnC⅐2Ca 2ϩ ⅐cIp-s, cCTnC⅐2Ca 2ϩ ⅐ 13 C/ 15 N-cIp-r, and 13 C/ 15 N-cCTnC⅐2Ca 2ϩ ⅐ 13 C/ 15 N-cIp-r. Distance restraints for cCTnC and cIp-r in the complex were obtained by analyzing three-dimensional 15 N-, 13 C-, and/or 13 C/ 15 N-edited NOESY experiments. Dihedral angle restraints for cCTnC and cIp-r in the complex were obtained from three-dimensional HNHA experiments and 15 N-edited NOESY experiments. Sample 13 C/ 15 N-cCTnC⅐2Ca 2ϩ ⅐ 13 C/ 15 N-cIp-r was used for assignment of symmetrical peaks in the 13 C-edited NOESY experiment for unambiguous protein-peptide side chain NOE crosspeak assignments.  A total of 1313 experimental distance restraints were obtained for the complex and used to calculate the high resolution solution structure of the complex of cCTnC⅐2Ca 2ϩ ⅐cIp: 1046 intramolecular NOE distance restraints for cCTnC (ϳ15 re-straints per residue), 267 intramolecular NOE distance restraints for cIp (ϳ13 restraints per residue), 23 intermolecular NOE distance restraints between cCTnC and cIp, 29 dihedral restraints for cCTnC, and 12 cCTnC distance restraints to Ca 2ϩ . Fig. 2A depicts the stereo view ensemble of the 30 lowest energy structures of the complex with superimposition of the backbone heavy atoms of cCTnC-(95-155). The overall conformational energies and structural statistics for the ensemble are provided in Table II. A ribbon diagram of the lowest energy structure of the ensemble is provided in Fig. 2B, and a surface map depicting the perturbation of amide resonances of cCTnC upon cIp binding is provided in Fig. 2C. The structures of cCTnC in the ensemble are of higher quality than those of cIp, as a result of more restraints per residue for the protein than for the peptide. The majority of assigned cIp intramolecular NOEs in the complex are of intraresidue origin, with no long range (i Ϫ j Ն 5) NOEs observed for cIp.
Structure of cCTnC in the Complex-The overall fold of cCTnC in the complex resembles other Ca 2ϩ -bound domains in the EF-hand family, such as the C-domain of skeletal TnC (53). The secondary structure elements of cCTnC in the complex are identical to unbound cCTnC⅐2Ca 2ϩ (7), displaying four well defined helices (helices E-H) and two well defined anti-parallel ␤-sheets. This helix-loop-helix structural motif is common in many Ca 2ϩ -binding proteins (Ca 2ϩ atoms are not shown in The N-terminal residues (residues 89 -91) are not as well defined, yet unexpectedly the C-terminal residues (residues 159 -161) are well ordered when compared with the unbound structure (7). This ordering of the C-terminal residues is due to NOE contacts with cIp in the complex, with contacts between cIp-Phe 138 and cCTnC-Val 160 minimizing the flexibility of the C-terminal residues, yielding a more ordered H-helix near residues 159 -161.
EF-hand structures can be described as "opened or closed," based upon the interhelical angles that the E-F and G-Hhelices make with respect to one another. A comparison of interhelical angles of cCTnC in the binary complex with previously solved EF-hand structures is presented in Table III. The binding of cSp to calcium-saturated cNTnC reveals a conformational opening of over 20 degrees, yet binding of cIp to calcium-saturated cCTnC reveals only a slight conforma-  Structure of cIp in the Complex-The structure of cCTnC⅐ 2Ca 2ϩ ⅐cIp is presented in Fig. 2. The structure of cIp in the binary complex displays a region of helical content in the central region of the peptide, flanked by a non-helical structured region on the N termini and an unstructured random coil region on the C termini. The overall fold of cIp in complex with cCTnC is novel, possessing no structural similarities to known Ca 2ϩ -binding protein/peptide interactions. cIp runs anti-parallel to cCTnC in the complex, with residues Ile 131 -Asp 133 making contacts with the hydrophobic cleft of cCTnC. Residues Ile 131 -Asp 133 possess no definitive helical secondary structure, yet superimposition of the 30 lowest energy ( Fig. 2A) structures reveals a structured region. This structured region makes a near 90°turn along the hydrophobic cleft of cCTnC to begin the helical region beginning at residue Leu 134 . This abrupt turn in the sequence is evidenced by multiple side chain inter-peptide NOE contacts of Ile 131 to Leu 134 and Arg 135 . Multiple hydrophobic NOE contacts were observed within this region, with cIp-Phe 132 making contacts with cCTnC-Thr 127 /Ile 128 /Ile 133 (linker region of cCTnC). Residues Leu 134 -Lys 139 adopt a helical secondary structure, with NOE contacts of cIp-Phe 138 making hydrophobic contacts with cCTnC-Val 160 /Leu 100 (E-and H-helices), with E-helix contacts closely matching those predicted by Cachia Fig. 2A) reveals this region as very labile in solution. No long range peptide-peptide NOE contacts were observed for this region, with only i, i ϩ 1 contacts observed between residues, and no protein-peptide NOE contacts were observed for this region. Arg 147 begins the segment of the ''switch peptide'' (cSp, residues 147-163), which has been shown to start a helical secondary structure when complexed with cNTnC⅐Ca 2ϩ (16,18,24). Arg 147 makes multiple contacts with cNTnC and thus is not expected to make interactions with cCTnC in the complex. Fig. 2C displays the chemical shift surface map of cCTnC upon cIp binding, which we have reported previously (42). Changes in the two-dimensional { 1 H, 15 N}-HSQC spectra of 15 N-cCTnC were monitored during a titration of cIp (Fig. 1A) and were mapped on the surface of the binary complex of cCTnC⅐2Ca 2ϩ ⅐cIp. largest chemical shift perturbations of cCTnC upon the binding surface area of cIp, within the linker region and the E-and H-helices. A large chemical shift was observed for the linker region, as cIp-Phe 132 is within close proximity to the linker region, with aromatic ring effects predicted to be responsible for the large shifts observed.
Chemical shift values for 13 C ␣ and 1 H ␣ nuclei are important indicators for predicting secondary structure (55)(56)(57). NMR studies have shown that upon the initiation of helical secondary structure formation, there is a downfield shift observed for 13 C ␣ and an upfield shift observed for 1 H ␣ (thus ⌬␦(ppm) ϭ ␦ bound Ϫ ␦ free ). Incorporation of 13 C label in cIp allows for measurement of both 13 C ␣ and 1 H ␣ chemical shifts of free and bound peptide. cIp undergoes 13 C ␣ chemical shift perturbation toward a more helical conformation (downfield shift (␦ bound Ϫ ␦ free ) Ͼ 0) in the region of residues Leu 134 -Lys 137 , with significant shift (Ͼ1 ppm) for residue Arg 135 . Residues Ile 131 -Asp 133 experience small upfield changes in 13 C ␣ chemical shift upon cCTnC binding, corresponding to the N-terminal region before the helical region of Leu 134 -Lys 139 . 13 C ␣ chemical shift changes for residues Arg 140 -Arg 147 indicate that this region possesses no helical characteristics, corresponding to the flexibility of this region as shown in Fig. 2A. 1 H ␣ perturbation measurements show minimal changes for all residues for cIp-r upon cCTnC binding, with only residues Gln 129 , Asp 133 , Arg 135 , Arg 145 , and Arg 147 undergoing changes greater than 0.05 ppm. For the helical region Leu 134 -Lys 139 , residues Leu 134 and Arg 135 undergo an upfield shift, with all others undergoing minimal downfield shifts less than 0.05 ppm. Comparison with induced 1 H ␣ chemical shift changes of the skeletal inhibitory peptide isoform (sIp) upon binding to sTnC as reported by Hernandez et al. (36) yields similar results of no changes greater than 0.1 ppm for the majority of residues, with Arg 135 undergoing the largest upfield chemical shift for both the cardiac and skeletal isoforms upon TnC binding. cIp and sIp share a conserved sequence with only a mutation of Thr 142 in cIp to that of a Pro residue in sIp, thus it is predicted that both cIp and sIp bind to TnC in a similar manner.
The electrostatic surface map of the binary complex cCTnC⅐ 2Ca 2ϩ ⅐cIp and its components are shown in Fig. 3. The highly basic cIp peptide has the potential for many favorable electrostatic interactions with the highly acidic cCTnC. At physiological conditions, pH 6.7, cIp (pI ϭ 11.0) will have an overall charge of 6.9, whereas cCTnC (pI ϭ 4.1) will possess an overall charge of Ϫ14.8 (58). It is supposed that the highly basic C-terminal end of cIp (Arg 140 , Arg 144 , Arg 145 , and Arg 147 ) will make beneficial interactions with the acidic E-helix of cCTnC (Glu 94 , Glu 95 , and Glu 96 ), thereby stabilizing the binary complex (54). As we have shown previously, mutation of R145G of cIp results in a 4-fold reduction of binding affinity with cCTnC when compared with wild type cIp. The resulting mutation resulted in a change of a basic residue for one that is uncharged, thereby effectively reducing the charge on cIp-R145G and thus affecting binding. Similar results were seen for phosphorylation of Thr 142 of cIp (42).
Backbone Amide 15 N Relaxation Studies of cCTnC-Backbone amide 15 N NMR relaxation data for 15 N-cCTnC⅐2Ca 2ϩ and 15 N-cCTnC⅐2Ca 2ϩ ⅐cIp were obtained at 500 and 600 MHz (Fig. 4). Backbone resonances for Met 90 -Lys 92 were not observed due to rapid amide proton exchange with water for both samples. Resonances Met 131 , Lys 137 , and Arg 147 were not included in 15 N-cCTnC⅐2Ca 2ϩ ⅐cIp data, and resonances Ala 99 and Arg 147 were not included in 15 N-cCTnC⅐2Ca 2ϩ , due to partial peak overlap in the { 1 H, 15 N}-HSQC spectra.
The T 1 , T 2 , and NOE values of 15 N-cCTnC⅐2Ca 2ϩ are shown in Fig. 4A. The measured values at 500 MHz for 15 N-cCTnC⅐2Ca 2ϩ show a decrease in T 1 , a decrease in T 2 , and an increase in NOE values for the first 5 N-terminal residues on the E-helix when compared with calcium-saturated skeletal C-domain isoform (sCTnC), indicating that the N-terminal residues on the E-helix of cCTnC are more ordered when compared with the skeletal isoform (59), as well as small differences in calcium-binding sites III and IV. Residues whose internal motions affect their measured relaxation values were excluded from the calculation of the averages, as determined using NOE criteria (NOE 500 Ͼ 0.6 and NOE 600 Ͼ 0.65). As expected, the average T 1 600 /T 1 500 ratio is Ͼ1; the ratio of T 2 600 / T 2 500 is approximately equal to 1, and the average NOE 600 / NOE 500 ratio is approximately equal to 1 for the majority of the residues studied. These ratios are within accepted theoretical calculations, taking into account effects from dipole-dipole relaxation and chemical shift anisotropy at the two magnetic field strengths studied.
The relaxation values for the 8. values upon cIp binding, indicating that this region is becoming more ordered upon cIp binding. These results correspond well with the structure of the binary complex of 15 N-cCTnC⅐ 2Ca 2ϩ ⅐cIp, as multiple NOE contacts are observed between cIp and cCTnC⅐2Ca 2ϩ in the linker region and the H-helix. 15 N NMR Relaxation Studies of 15 N-cIp-Backbone amide 15 N NMR relaxation data for 15 N-cIp-r and cCTnC⅐2Ca 2ϩ ⅐ 15 N-cIp-r were obtained at 500 and 600 MHz (Fig. 5). Backbone resonances for Thr 128 and Gln 129 were not observed due to rapid amide proton exchange with water for both samples. Residue Pro 141 was not observed due to the absence of a residual amide proton. The T 1 , T 2 , and NOE values observed at the two frequencies are typical for an unstructured 2.5-kDa domain in solution. Addition of cCTnC⅐2Ca 2ϩ produces large effects on the relaxation rates of all residues present on cIp. As expected for an increase in total molecular mass to 10.7 kDa (assuming a 1:1 ratio), decreases of both T 1 and T 2 values are observed at both magnetic field strengths, along with a large increase in the recorded values for NOE. Interpretation of the data reveals that cIp is becoming increasingly rigid upon cCTnC binding.
Comparison of the total molecular weights of both 15 N-cCTnC⅐2Ca 2ϩ ⅐cIp and cCTnC⅐2Ca 2ϩ ⅐ 15 N-cIp-r indicates that both systems should produce equivalent experimental correlation times ( C ), assuming isotropic tumbling. Predicted C values for a 1:1 binary complex of 15 N-cCTnC⅐2Ca 2ϩ ⅐cIp and cCTnC⅐2Ca 2ϩ ⅐ 15 N-cIp-r are 5.5 and 5.5 ns, respectively, whereas experimentally measured values of 5.8 and 5.7 ns are observed, enforcing that the binary complex of cCTnC and cIp is binding in a 1:1 ratio. The measured T 2 values of 15 N-cCTnC⅐ 2Ca 2ϩ ⅐cIp mirror the values measured for cCTnC⅐2Ca 2ϩ ⅐ 15 N-cIp-r (ϳ120 ms) also supporting a 1:1 binary complex, as well implying that cIp is becoming increasingly rigid upon cCTnC binding even though the structure of cIp in the binary complex is not as well defined as cCTnC. DISCUSSION The general mechanism of contraction involving temporal interactions of the thin and thick filaments in muscle fibers upon a Ca 2ϩ signaling cascade is now quite well understood (1)(2)(3)(4)(5). However, no high resolution structures of the troponin complex are available to date. In the last few years, successful determination of various structures of both cardiac and skeletal TnC isoforms in various states and in complex with various TnI segments have become available (7, 16 -19, 21, 24, 41, 53), yielding insights into the molecular interactions of muscle contraction. Coupled with these studies has been the advent of backbone and side chain NMR relaxation data (59,62,63), which have provided details into the dynamics and energetics of peptide and Ca 2ϩ binding to TnC. Several research groups  (21,32,37,64) have proposed structural models of the TnC⅐TnI⅐TnT ternary complex, involving specific interactions of the domains during muscle contraction. Within the cardiac system, the calcium-binding sites in the C-domain of TnC are believed to always be occupied by Ca 2ϩ (65), and the C-domain is thus designated the structural domain. The N-domain of TnC is believed to be the regulatory domain, undergoing large structural changes upon the onset of Ca 2ϩ -induced contraction. In the resting state the Ca 2ϩ -binding site in the N-domain of TnC is unoccupied. The ability of the N-domain to sense the Ca 2ϩ signal in order to propagate muscle contraction has led to the widely accepted designation as the molecular switch for muscle contraction. The classification of the N and C domains of TnC as the regulatory and structural domains are, however, somewhat arbitrary as physiological roles of the two domains may vary within intact muscle fibers. The C domain of TnC has been shown to interact with specific regions of both TnT and TnI, with a high resolution structure presented by Vassylyev et al. (21), showing interactions of sCTnC with an N-terminal fragment of TnI- , revealing a helical structure of the TnI fragment that interacts with the hydrophobic cleft of the C-domain. Recent functional studies (32) have required researchers to re-evaluate the role of the C domain as one that actively participates in the muscle contraction signaling process. Following Ca 2ϩ binding to NTnC, the inhibitory region of TnI-(128 -147) (cardiac isoform) is believed to relocate from its binding partner actin to TnC on the thin filament. This reorganization of the inhibitory region allows for myosin binding to actin, which allows for completion of the power stroke during muscle contraction. The binding of cIp to cTnC has been shown to have interactions with the C domain, as well as with the central linker region between the D-and E-helices (16).
Early NMR work by Campbell and co-workers (33, 34) using transferred NOEs in the skeletal system (sTnI-(104 -115)) predicted the inhibitory region to comprise two turns of a helix, surrounding two central proline residues, and subsequent modeling of a binary complex docked the bound inhibitory region within the hydrophobic core of sCTnC (35). The structure of a N-terminal regulatory peptide Rp40 in complex with sTnC presented by Vassylyev et al. (21) conflicted with this model of the inhibitory region, as Rp40 was shown to bind to the hydrophobic core of sCTnC. Vassylyev et al. (21) have proposed a model of the troponin complex with the inhibitory region having a helical secondary structure, away from the hydrophobic cleft of sCTnC that binds Rp40. NMR chemical shift mapping data presented by Mercier et al. (22) agreed with results by Vassylyev et al. (21) that Rp40 binds within the hydrophobic cleft of sCTnC, yet predicted that the inhibitory region may bind across the top of the hydrophobic patch within the ternary troponin complex. Hernandez et al. (36) challenged the idea of a helical inhibitory region and suggested that the inhibitory region adopts an extended conformation in the binary complex, with a structural model presented by Tung et al. (37) predicting the inhibitory region to possess a ␤-hairpin secondary structure, away from the hydrophobic cleft of sCTnC.
The purpose of this study was to explore and solve the NMR solution structure of cIp when bound to the troponin complex. We have shown previously that cIp binds to the C-domain of cTnC, with large chemical shift perturbations in the { 1 H, 15 N}-HSQC spectrum (30,42). Considering these data, and to minimize the total molecular weight of the complex for ease of assignment in the NMR spectrum, it was decided to pursue the structure of cIp in complex with cCTnC⅐2Ca 2ϩ . By using isotope labeling strategies of recombinant peptide production as a costeffective alternative to synthetic peptide labeling (see under ''Experimental Procedures''), we have been successful in elucidating the structure of the binary complex of cCTnC⅐2Ca 2ϩ ⅐cIp, which is presented in Fig. 2. In the binary complex, cIp adopts a helical conformation, making NOE contacts with the linker region of cCTnC, as well as with both the E-and H-helices. Residues Leu 134 -Lys 139 adopt a helical arrangement within the binary complex, with residues Arg 140 -Arg 147 adopting an extended conformation, void of any secondary structure elements.
Reflection on the structure of the binary complex reveals that all previous predicted models are in part incorrect. The inhibitory region has no ␤-sheet secondary structure present, as is indicated by the presence of traditional helical NOEs (i.e. d ␣␤(i, i ϩ 3) and d ␣N(i, i ϩ 3) ) and by chemical shift data for 1 H ␣ and 13 C ␣ atoms in the protein backbone. No long range NOE ( i Ϫ j Ն 5) contacts are observed, which would be predicted if the inhibitory region possessed ␤-sheet characteristics. The model predicted by Vassylyev et al. (21) correctly predicted a helical region of cIp, yet positioning of the helix in reference to cTnC was incorrect. The model by Campbell et al. (33) predicted a helical orientation of sTnI-(104 -115), which we have shown to be unstructured in the cardiac isoform. Our results correlate well with recent results published by Brown et al. (38) in which the structure of the inhibitory region was proposed by spin labeling EPR. Within this study it was proposed that the inhibitory region in a ternary troponin complex possesses a helical region from residues Gln 129 -Lys 137 , with residues Phe 138 -Arg 145 showing no secondary structure elements, which is in good agreement with the preliminary ternary troponin structure by Takeda et al. (41). As well, it was predicted that residues Lys 130 -Arg 135 are important in making contacts with TnT, which are immediately N-terminal to the region which comprises the 1.5 turns of the helix presented in Fig. 2.
It is observed that numerous stabilizing electrostatic interactions occur between the acidic C domain and the highly basic inhibitory region, as there is only 23 cCTnC⅐cIp NOEs in the binary complex. Electrostatic NOE contacts are traditionally difficult to measure by NMR, due to proton exchange and NOE distance limitations (Ͻ5 Å). The probability of stabilizing electrostatic interactions within the complex is high, as inspection of the structure yields close association of basic/acidic residue overlap, with the potential of the unstructured region containing residues Arg 140 , Arg 144 , Arg 145 , and Arg 147 in cIp to interact with the acidic residues Glu 94 , Glu 95 , and Glu 96 present on the E-helix (Fig. 3). Recent work by Tripet et al. (39) has shown a decrease in the affinity of cIp to cTnC with increasing concentrations of KCl, suggesting that electrostatics play a part in binding. Previous work from our laboratory has shown that mutation R145G and phosphorylation of Thr 142 diminishes the binding affinity of cIp for cCTnC by ϳ4and ϳ14-fold (42), respectively. The mutation and phosphorylation events of cIp effectively change the charge of the residue side chain by Ϫ1, which diminishes electrostatic interactions when compared with the wild type inhibitory region. In this regard, it is not surprising that the phosphorylation of Thr 142 has a large reduction of binding when compared with R145G, as Thr 142 is in much closer proximity to cCTnC in the presented structure.
The structure of the inhibitory region may be altered within the full-length TnC⅐TnT⅐TnI troponin complex when compared with the presented binary complex of cCTnC⅐cIp, wherein additional interacting subunits may play a factor in ternary assembly. Comparison of the structure of cCTnC⅐2Ca 2ϩ ⅐cIp with the structure of sCTnC⅐2Ca 2ϩ in complex with the N-terminal domain of sTnI-(1-47) presented by Vassylyev et al. (21) reveals that there would be steric clashes between the two cTnI subunits when bound to the C-domain of TnC. Specifically, there is steric overlap of the N-terminal residues of cIp with Rp40 in the hydrophobic cleft of the C-domain. Previously, it has been shown that the regulatory peptide Rp40 will displace sIp in a titration with sCTnC⅐2Ca 2ϩ (22), implying that the Rp40 is always bound to sCTnC⅐2Ca 2ϩ during both muscle relaxation and during contraction. Thus a rationale must be presented for validation of the presented structure. There are notable differences in primary sequence between the cardiac and skeletal isoforms, specifically large differences in the regulatory region of TnI (cTnI-(32-79) and sTnI-(1-47)), thus the regulatory peptide might have altered binding in the cardiac system. As well, EPR results from Brown et al. (38) have shown cIp to be in contact with cTnT, as well as cTnC. The observed binary structure (Fig. 2) may of course be altered somewhat in the context of the intact troponin complex. In particular, a section of cTnI-(128 -139) has been predicted to be involved in a coiled-coil with TnT (66). The residues of steric clash of superimposition of sTnC⅐Rp40 and cCTnC⅐cIp are those of Thr 128 -Arg 135 , of which Brown et al. (38) have shown Lys 130 -Arg 135 to be in contact with TnT. As well, preliminary data from the 2.6 Å x-ray structure by Takeda et al. (41) of the complex of cTnC⅐cTnI-(34 -160)⅐cTnT-(181-288) revealed a coiled-coil of TnT with Thr 128 -Lys 137 of cIp, with an undefined region spanning Phe 138 -Arg 147 , implying this region as unstructured or flexible.
It is possible that Thr 128 -Lys 139 of cIp makes a coiled-coil interaction with cTnT, with Arg 135 -Lys 139 of cIp making contacts with the E-and H-helices of cCTnC up to Lys 139 with an unstructured region of Arg 140 -Arg 147 , followed by cSp which has been shown to have interactions with cNTnC. The region Arg 140 -Arg 147 makes stabilizing electrostatic interactions with the acidic E-helix and the linker region of cTnC, where cSp begins. As well the coiled-coil interaction of cTnT to Thr 128 -Arg 135 of cIp will bring the N-terminal residues of cIp out of the hydrophobic cleft of cCTnC, allowing the regulatory peptide Rp40 to bind. Supporting this, in the binary complex presented (Fig. 2) Leu 134 was found to possess no NOE contacts to cCTnC. In this model, modulation of the interaction between cSp and cNTnC during intracellular calcium efflux is sufficient to excise cIp from its binding site on actin to that of cCTnC, thereby releasing the inhibition of heart contraction.
We have been successful in elucidating the NMR solution structure of the binary complex of cCTnC⅐2Ca 2ϩ ⅐cIp, which reveals a helical conformation of the inhibitory region. The helical region of cIp (Leu 134 -Lys 139 ) binds to cCTnC in a novel orientation, with contacts to regions of the E-and H-helices and the linker region of cCTnC. The structure further reveals the potentiality of stabilizing electrostatic interactions in the binary complex. Coupled with 15 N NMR relaxation data, we have shown that the binary complex forms a 1:1 complex in solution. Further studies on the ternary complex are encouraged as the interactions predicted between TnC⅐TnI⅐TnT might currently be too simplistic.