Regulatory Domain Conformational Exchange and Linker Region Flexibility in Cardiac Troponin C Bound to Cardiac Troponin I*

Previously, we utilized 15N transverse relaxation rates to demonstrate significant mobility in the linker region and conformational exchange in the regulatory domain of Ca2+-saturated cardiac troponin C bound to the isolated N-domain of cardiac troponin I (Gaponenko, V., Abusamhadneh, E., Abbott, M. B., Finley, N., Gasmi-Seabrook, G., Solaro, R.J., Rance, M., and Rosevear, P.R. (1999) J. Biol. Chem.274, 16681–16684). Here we show a large decrease in cardiac troponin C linker flexibility, corresponding to residues 85–93, when bound to intact cardiac troponin I. The addition of 2 m urea to the intact cardiac troponin I-troponin C complex significantly increased linker flexibility. Conformational changes in the regulatory domain of cardiac troponin C were monitored in complexes with troponin I-(1–211), troponin I-(33–211), troponin I-(1–80) and bisphosphorylated troponin I-(1–80). The cardiac specific N terminus, residues 1–32, and the C-domain, residues 81–211, of troponin I are both capable of inducing conformational changes in the troponin C regulatory domain. Phosphorylation of the cardiac specific N terminus reversed its effects on the regulatory domain. These studies provide the first evidence that the cardiac specific N terminus can modulate the function of troponin C by altering the conformational equilibrium of the regulatory domain.

Cardiac muscle contraction is regulated by Ca 2ϩ -dependent interactions between members of the troponin complex and other thin filament proteins including actin and tropomyosin. The troponin complex is associated with tropomyosin and consists of troponin (Tn) 1 I, the inhibitory subunit, troponin C (TnC), the Ca 2ϩ -binding subunit, and troponin T (TnT), which makes primary protein-protein contacts with tropomyosin. TnC is required to confer Ca 2ϩ sensitivity on the actin-myosin interaction. The cardiac isoforms of TnC and TnI differ structurally and functionally from fast skeletal isoforms. Calcium binding to the regulatory domain of skeletal TnC results in movement of the B and C helices away from the N, A, and D helices producing a conformational "opening." The newly exposed hydrophobic pocket allows additional interactions with TnI (1)(2)(3)(4)(5). Calcium binding site I in cTnC is naturally inactive and in the presence of Ca 2ϩ the major conformer of cTnC was found to maintain a "closed" regulatory domain conformation with minimal exposure of the hydrophobic pocket (5,6). However, the Ca 2ϩ -saturated regulatory domain of cTnC exhibits chemical exchange consistent with an equilibrium between open and closed conformations (6,7). NMR studies on the isolated regulatory domain and fluorescence resonance energy transfer studies utilizing intact cTnC and a variety of synthetic cTnI peptides indicated that a structural opening of the cTnC regulatory domain similar to that observed in the skeletal isoform requires both Ca 2ϩ and cTnI-(147-163) (8,9). The cardiac isoform of TnI is also unique in that it contains a cardiac specific extension of approximately 32 residues containing two adjacent Ser residues, 23 and 24, that are phosphorylated in vivo in response to ␤-adrenergic stimulation (10). Cardiac TnC and TnI are known to interact in an antiparallel manner such that the C-domain of TnC interacts with the N-domain of TnI (11). An N-terminal segment of cTnI, corresponding to residues 33-80, was found to be sufficient for interaction with the C-domain of cTnC in a manner identical to that observed for intact cTnI (11,12). This region of cTnI is homologous to sTnI residues 1-47, which was shown to bind as an ␣-helix to the C-domain of sTnC by x-ray crystallography (13). In this complex, the D/E linker of sTnC was partially unwound and bent by 90°in contrast to the fully helical extended D/E linker seen in the x-ray crystal structures of sTnC (3,13,14).
Two low resolution models based on neutron scattering studies utilizing the intact skeletal TnIC complex have been published. In the first model, derived from data collected on the sTnI⅐sTnC complex in the presence of 2-3 M urea, extended structures for both sTnC and sTnI were proposed (15,16). Troponin I was found to wrap around TnC in a coiled manner, and cap each end of an extended TnC molecule (15,16). Neutron scattering data, obtained in the absence of urea using the intact troponin complex, was more consistent with a two-domain model for TnI (17). The structure of TnI is predicted to * This work was supported by Grants AR 44324 (to P. R. R.), GM 40089 (to M. R.), and HL63377 (to R. J. S.) from the National Institutes of Health. 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.
High resolution solution NMR studies of TnC and TnI interactions have largely utilized isolated domains and synthetic peptides due to the large molecular mass (approximately 42 kDa) and limited solubility of the intact cTnIC complex. Findings from these studies, by their nature, fail to reveal structural mechanisms that rely on interdomain interactions and must be verified in more sophisticated model systems. Low resolution structural techniques such as fluorescence and neutron scattering have provided important information on the intact complex. However, they do not contain sufficient detail to verify the findings of the high resolution single-domain experiments or completely elucidate the structural mechanisms of cTnC regulation and cTnI inhibitory function.
In an effort to correlate functional regions of cTnI with structural mechanisms in the cTnIC system and bridge the gap between available high resolution structures of isolated domains and the low resolution data of the intact complex, a series of cTnC⅐cTnI complexes were created for multidimensional heteronuclear NMR analysis. Strategic deletion of functionally significant regions cTnI was utilized to investigate the structural interactions that underlie the switch mechanism within the cTnIC complex. The dynamics of cTnC in the intact cTnIC complex were monitored by 15 N transverse relaxation rates and chemical shift mapping. Chemical shift data for cTnC bound to cTnI-(1-80), cTnI-(1-80)pp, cTnI-(1-80)DD, cTnI-(33-211), and cTnI-(1-211) were compared and used to monitor conformational equilibrium in the cTnC regulatory domain. Surprisingly, binding of both the cardiac specific N terminus and the C-domain of cTnI altered the cTnC regulatory domain conformation. The absence of the cardiac specific N terminus (cTnI-(33-211)), mutation of Ser 23 and Ser 24 to Asp (cTnI-(1-80)DD) or phosphorylation at Ser residues 23 and 24 (cTnI-(1-80)pp), shifts the conformational equilibrium toward that observed in free Ca 2ϩ -saturated cTnC. This provides the first evidence that the cardiac specific N terminus modulates the function of cTnC by altering the conformational equilibria within the regulatory domain. These results also demonstrate that a simple model based on Ca 2ϩ -dependent binding of cTnI residues 147-163 to the regulatory domain of cTnC is insufficient to account for the molecular details of cardiac troponin I-troponin C interactions.
Fluorescence Assay-Calcium binding to cTnC was measured as described previously (24). Binding measurements were made with the aid of the fluorescent probe, 2-(4Ј-iodoacetamidoanilino)-naphthalene-6sulfonic acid (IAANS), which reports Ca 2ϩ binding to the regulatory site. Cardiac troponin C (1.5 mg/ml) was labeled in 10 mM MOPS, 90 mM KCl, 2.6 mM CaCl 2 , 2 mM EGTA, 6 M urea, pH 7.0 and a 5-fold molar excess of IAANS, prepared fresh as a stock solution in Me 2 SO. Labeling proceeded for 8 -10 h at room temperature with constant shaking. The solution was then dialyzed against two changes of 4 liters of 10 mM MOPS, 90 mM KCl at pH 7.0. The molar ratio of the incorporated probe to cTnC was approximately 1.5. Labeled cTnC (cTnC IA ) was complexed with cTnI as described previously (24). Fluorescence measurements were carried out in a Perkin-Elmer SS-5B luminescence spectrometer using an excitation wavelength of 330 nm. The peak emission wavelength (445-455 nm) was determined prior to Ca 2ϩ titrations. Ca 2ϩ titrations were carried out with and without 2 M urea in 3 ml of a solution containing 1 M of the cTnI⅐cTnC IA complex, 20 mM MOPS, 90 mM KCl, 3 mM MgCl 2 , pH 7.0, 1 mM EGTA. A range of free Ca 2ϩ concentrations was achieved by sequential addition of CaCl 2 as calculated using binding constants reported by Godt and Lindley (25). Changes in fluorescence intensities were normalized to the maximum change, and the data fitted to the Hill equation as described previously (24).

RESULTS
Previously, we have shown that the N-domain of cTnI containing the cardiac specific N terminus, cTnI-(1-80), interacts with both domains of cTnC and decreases chemical exchange in the regulatory domain (7). Binding of cTnI-(1-80), however, does not significantly affect flexibility in the D/E linker region. To further elucidate dynamic relationships within the cTnIC complex, we examined the binding of cTnI-  (7,18). When necessary, assignments for cTnC⅐cTnI-(1-211) and cTnC⅐cTnI-(33-211) were confirmed using H N to H N and H N to H ␣ sequential NOEs obtained from three-dimensional NOESY-HSQC spectra. The 800-MHz 1 H-15 N TROSY spectrum for [ 15 N, 2 H]cTnC bound to cTnI-(1-211) with crosspeaks assigned is shown in Fig. 1. Multiple conformations were observed for Ala 31 , Glu 32 , Gly 34 , Gly 42 , Glu 66 , Gly 68 , Val 72 , Asp 73 , and Val 79 in the 1 H-15 N TROSY spectra of the 42-kDa cTnC⅐cTnI-(1-211) complex. For these residues, the most intense cross-peak representing the predominant conformation of cTnC was utilized for chemical shift comparison and to obtain rate information (Fig. 2). A number of resonances throughout the regulatory domain were broadened beyond detection or shifted upon binding either cTnI-(1-80) or cTnI-(33-211). Interestingly, cross-peaks for Ala 31 , Asp 33 , and Gly 34 in inactive calcium binding site I, cross-peaks for Leu 48 and Gly 49 in the B/C loop region, and cross-peaks for Glu 66 , Asp 67 , and Thr 71 in calcium binding site II were significantly perturbed only when cTnC was bound to cTnI-(1-80) or cTnI-(1-211).
To further explore the conformational exchange observed in the regulatory domain of cTnC, 15 N transverse relaxation rates were measured in the intact [ 15 N, 2 H]cTnC⅐cTnI-(1-211) complex. 15 N transverse relaxation rates, R 2 , depend on the rotational correlation time of the molecule, internal motions, conformational exchange, dipole-dipole interactions, and chemical shift anisotropy. Internal motion and conformational exchange are indicated by 15 N R 2 rates lower and higher than the average 15 N R 2 value for the molecule, respectively. The previously published 15 N R 2 values for free cTnC are included for comparison (7). 15 N R 2 values for free Ca 2ϩ -saturated cTnC demonstrate distinct averages for the N-and C-domains, indicating some independence in the tumbling of the two domains resulting in distinct rotational correlation times (7). Additionally, lower than average rates are observed in the D/E linker, consistent with rapid internal motions due to flexibility in this region (7). Higher than average rates within defunct site I were previously shown to result from conformational exchange by comparison of 15 N transverse cross-correlation rates and 15 N transverse relaxation rates (7).
The average 15 N R 2 values for the N-and C-domains of cTnC in the cTnC⅐cTnI-(1-211) are 34 Ϯ 7 s Ϫ1 and 35 Ϯ 7 s Ϫ1 , respectively, suggesting a uniform rotational correlation time across the molecule (Fig. 2B). Larger errors for the R 2 values in the cTnC⅐cTnI-(1-211) complex are a consequence of the large molecular mass (42 kDa) and lower solubility of the intact cTnIC complex. Despite larger errors for the intact complex, trends in rates through the molecule are discernable. Collection of 15 N transverse cross-correlation rates was not practical due to the lower protein concentration of the complex, the larger molecular mass (42 kDa) of the complex, and the inherent insensitivity of the experiment. Further, collection of a complete set of relaxation parameters and calculation of order parameters (S 2 ) would be impractical given the physical limitations of the more biologically relevant systems utilized in this study. Determination of S 2 values in the absence of detailed tertiary structures is ambiguous due to the uncertainty of the diffusion tensors (26 -28). Significantly, R 2 values for residues 85-93, which comprise the linker region, are nearly as large as those for the N-and C-domains, indicating that full-length cTnI considerably restricts motion of these residues in the intact complex (Fig. 2). 15 N R 2 values for Asp 25 , Ile 26 , and Gly 34 in defunct Ca 2ϩ binding site I, as well as Val 72 and Asp 73 in Ca 2ϩ binding site II were equal to or greater than the average rate in the intact cTnIC complex (Fig. 2). In contrast, the  31 and Asp 73 , shown in boxes at the appropriate chemical shifts, are included as insets as they are specifically discussed below. Gly 70 , also shown boxed at the appropriate position, is not visible in this TROSY spectrum, but is clearly visible in HSQC spectra.
cTnC⅐cTnI-(1-80) complex has below average 15 N R 2 values in site I as well as the linker region, indicating internal motion in both regions, as previously shown (7).
Relaxation parameters of cTnC in the intact complex were examined in the presence of urea to mimic the conditions used for neutron scattering studies of the skeletal complex (15,16). In the presence of urea, the average 15 N transverse relaxation rate increases to approximately 46 s Ϫ1 . The increase in average rate likely results from an increase in rotational correlation time due to either interactions of urea with amino acid side chains or partial disruption of cTnC⅐cTnI interactions. It has been shown that the residence times of urea molecules in solvation sites near methyl groups of Val, Leu, and Ile are significant, possibly leading to a decrease in the hydrophobic effect within the protein and stabilizing solvent-exposed hydrophobic groups (29,30). Urea, in the range of 2-4 M has been shown to increase the radius of gyration of other polypeptides (31). 15 N transverse relaxation rates in the linker region are reduced to about 60% of the average rate and indicate increased mobility for these residues in the presence of urea. The observed change in linker flexibility may have important implications in analysis of data collected on the sTnI⅐sTnC complex in the presence of 2-3 M urea. In addition, 2 M urea affects Ca 2ϩ binding to the regulatory site of cTnC, as reported by fluorescence changes in IAANS (Fig. 3). A small change in the correlation between pCa and percentage of fluorescence intensity in the cTnC IA ⅐cTnI complex is observed upon addition of 2 M urea. The shift of half-maximal pCa to a higher value in the presence of 2 M urea indicates Ca 2ϩ binding affinity of the regulatory domain in the cTnIC complex is increased, suggesting that urea might alter the conformation of the regulatory domain. These results, along with comparison of amide 1 H and 15 N chemical shift differences in the presence and absence of 2 M urea (data not shown), suggest, that although urea does not cause major perturbations in the structure of cTnC in the intact cTnIC complex, it significantly alters the dynamics within the D/E linker region and the regulatory domain.
Conformational exchange in Ca 2ϩ -saturated cTnC has been previously identified in defunct Ca 2ϩ binding site I (6, 7). In addition, multiple amide cross-peaks for a number of residues (Leu 29 , Asp 32 , and Gly 34 ) have been detected in the cTnC⅐cTnI-(1-80) complex (7). Initially, we analyzed interactions of cTnC with cTnI-(1-211) by comparing combined amide 1 H and 15 N chemical shift differences for cTnC bound to cTnI-(1-80) and cTnI-(1-211) (Fig. 4). Residues showing significant chemical shift perturbations were located in defunct Ca 2ϩ binding site I (residues 28 -38), Ca 2ϩ binding site II (residues 65-76), and the linker region (residues 85-93) (Fig. 4). In general, chemical shift differences were considerably smaller than those observed for binding of the N-domain of cTnI to the C-domain of cTnC (11,18). Observed chemical shift differences could result from either a direct interaction between cTnC and the C-terminal region of cTnI, comprising residues 81-211, or as a consequence of an altered chemical exchange rate between solution conformations.
Residues Ala 31 , Glu 32 , Gly 42 , Glu 66 , and Asp 73 , were shown to exhibit some of the largest 1 H and 15 7). B, resonance assignments in 1 H-15 N correlation spectra of the cTnC⅐cTnI-(1-211) complex could not be confirmed due to a lack of sequential NOEs in NOESY-HSQC spectra for residues 1  equilibrium toward a second conformation as judged by induced 1 H and 15 N chemical shift perturbations. Here we examine in detail the conformational transitions of two residues, Ala 31 and Asp 73 , in Figs. 5 and 6, respectively. These residues were chosen since their amide 1 H and 15 N chemical shift values are unique and assignment of multiple resonances for each of these residues is unequivocal (Fig. 1).
Alanine 31 is located in defunct Ca 2ϩ binding loop I. A single cross-peak was found for Ala 31 in the 1 H-15 N HSQC spectrum of cTnC bound to cTnI-(1-80)pp, cTnI-(1-80)DD or cTnI-(33-211) (Fig. 5, A, B, and D). In the presence of cTnI-(1-80) containing the cardiac specific N terminus (Fig. 5C), the resonance for Ala 31 was observed downfield from the cross-peak observed in either cTnI-(1-80)pp, cTnI-(1-80)DD, or cTnI-(33-211). Two cross-peaks for Ala 31 in cTnC bound to cTnI-(1-211) were found, suggesting the presence of multiple conformations in slow chemical exchange for this region of the regulatory domain (Fig. 5E). In the presence of 2 M urea, two resonances for Ala 31 are observed with an additional downfield shift (Fig.  5F). Ligand-induced chemical shift changes result from perturbations of the magnetic environment of the resonance caused directly by the bound peptide ligand, or indirectly due to ligand-induced conformational changes, or alteration of chemical equilibria within the complex. These data demonstrate that the cardiac specific N terminus, as well as elements within cTnI-(81-211), can induce chemical shift changes consistent with a more open structure of the regulatory domain. Further, mutation of Ser 23 and Ser 24 to mimic phosphorylation or actual phosphorylation of Ser 23 and Ser 24 eliminates the effect of the cardiac specific N terminus on the conformation of the regulatory domain.
A similar pattern of chemical shifts is observed for Asp 73 , located in the short ␤-sheet between Ca 2ϩ binding sites I and II, in Ca 2ϩ -saturated cTnC complexes with cTnI (Fig. 6). A single cross-peak is found for Asp 73 in cTnC bound to either cTnI-(1-80)pp or cTnI-(1-80)DD (Fig. 6, A and B). In the presence of cTnI-(1-80), having the non-phosphorylated cardiac N terminus, two downfield shifted cross-peaks are observed for Asp 73 (Fig. 6C). In the presence of cTnI-(33-211), lacking the cardiac N terminus and containing residues 81-211, multiple cross-peaks are observed for Asp 73 in the regulatory domain (Fig. 6D). Addition of the cardiac N terminus in the cTnC⅐cTnI-(1-211) complex shifts the Asp 73 cross-peaks downfield in both the 1 H and 15 N dimensions toward a second regulatory domain conformation as observed for cTnC bound to cTnI-(1-80) (Fig.  6, D and E). These resonances appear to be in intermediate to slow exchange between two or more conformations. Finally, addition of 2 M urea to the cTnC⅐cTnI-(1-211) complex shifts the equilibrium for Asp 73 toward the second regulatory domain conformation (Fig. 6F).
These comparisons demonstrate that not only is the region 81-211 of cTnI capable of inducing conformational changes in the regulatory domain of cTnC, but the cardiac N terminus can also influence the cTnC regulatory domain conformation, either independently or in conjunction with the inhibitory motifs. Phosphorylation of Ser 23 and Ser 24 or mutation of Ser 23 and Ser 24 to mimic phosphorylation removes the effect of the cardiac specific N terminus on the cTnC regulatory domain conformation. The pattern of chemical shift changes for Ala 31 and Asp 73 (Figs. 5 and 6), as well as those for Glu 32 , Gly 34 , Gly 42 , Gly 68 , Glu 66 , Val 72 , and Val 79 (data not shown) closely resembles the pattern of chemical shift changes observed in the isolated N-domain of cTnC upon titration with the peptide corresponding to cTnI-(147-163) (8). In general, cTnC regulatory domain chemical shift perturbations attributable to the presence of residues 81-211 of cTnI were more prominent for residues in calcium binding site II, whereas chemical shift changes in site I were largely due to the presence of the unmodified cardiac specific N terminus of cTnI.
Many of the cross-peaks for residues that constitute the hinges of the cTnC regulatory domain, Glu 40 -(1-211). The horizontal line represents the average chemical shift difference-plus one standard deviation (0.04 Ϯ 0.03 ppm). Filled diamonds mark residues for which chemical shift data is unavailable. Cross-peaks for 28,29,30,36,37,38, and 39 were broadened beyond detection in both spectra. Multiple crosspeaks were observed for residues 31, 32, 34, 42, 66, 68, 73, and 77. For these residues, the most intense cross-peak, representing the predominant conformation, was selected.

DISCUSSION
In molecular switch proteins consisting of multiple domains, an understanding of both interdomain and intradomain interactions is essential for elucidation of a detailed structural mechanism. However, structure determination in complexes with molecular mass larger than 25 kDa can be problematic, particularly in poorly soluble, largely ␣-helical proteins. Studies on isolated domains have provided high resolution structures and revealed domain level mechanisms, such as the opening of the globular domains of cTnC upon binding the appropriate segment of cTnI (8,12). However, the study of single domain model systems prevents investigation of interdomain mechanisms within the troponin complex. Investigation of proposed mechanistic roles of the linker region of cTnC and the cardiac specific N terminus of cTnI require a model system based on the intact cTnC molecule.
Despite the rigid ␣-helical conformation of the D/E linker in x-ray crystal structures, considerable evidence suggesting conformational flexibility in solution exists (1,7,33,34). Mutational studies of the central helix in TnC have previously shown that both a required amount of flexibility and a critical length are necessary for optimal activity, suggesting that the linker region may function as a molecular ruler, maintaining a proper relationship between the N-and C-domains of cTnC (35)(36)(37). In the present study, we utilized 15 N R 2 values to measure flexibility within the linker region of cTnC in the presence of full-length cTnI. There was a significant loss of flexibility for residues 85-93 in cTnC upon binding cTnI-(1-211), suggesting that residues 81-211 may interact with or immobilize the linker region in cTnC (Fig. 2). Alternatively, interactions between C-domain of cTnI and the regulatory domain of cTnC may fix the relationship between cTnC domains, resulting in reduced mobility in the linker region. Alterations in linker flexibility related to changes in interdomain relationships may play a crucial role in the conformational switch that signals muscle contraction (36 -38).
Our data further demonstrate that urea significantly increases interdomain flexibility within cTnC (Fig. 2). The lack of large chemical shift differences and the small change in the pCa 2ϩ -fluorescence relationship in IAANS-labeled cTnC⅐cTnI complex in 2 M urea (Fig. 3) suggest minimal perturbation of the globular domains in cTnC bound to cTnI in the presence of urea. However, regulatory domain chemical shifts (Figs. 5 and 6) and the pCa 50 shift (Fig. 3) are consistent with urea stabilizing a more open structure of the regulatory domain, possibly by solvating exposed hydrophobic surfaces (39). Taken together, the data suggest that urea may alter interactions between the regulatory domain of cTnC and the C-domain of cTnI.
The amide cross-peaks of Ala 31 and Asp 73 (Figs. 5 and 6), as well as those for Glu 32 , Gly 34 , Gly 42 , Gly 68 , Glu 66 , Val 72 , and Val 79 (data not shown) of Ca 2ϩ -saturated cTnC shift in the same direction and by approximately the same magnitude in the presence of intact cTnI-(1-211) as was observed when the isolated regulatory domain bound cTnI peptide corresponding to residues 147-163 was bound to cTnC-(1-89) (8). These chemical shift changes were shown to correlate with a conformational opening of the isolated regulatory domain (8). sion is further supported by urea stabilization of the second or open state. It has been shown that urea can decrease the hydrophobic effect within the protein and may stabilize solvent-exposed hydrophobic surfaces (29,30). Interestingly, addition of the well known cTnI inhibitory peptide, comprising residues 129 -147, to the cTnC⅐cTnI-(1-80) complex had little effect on conformational exchange or the number of conformers detected in residues Ala 31 , Glu 32 , Gly 34 , Gly 42 , Glu 66 , Gly 68 , Val 72 , Asp 73 , and Val 79 (40). However, addition of cTnI-(129 -166) to the cTnC⅐cTnI-(1-80) complex resulted in regulatory domain chemical shifts similar to those observed in cTnC-(1-89) bound to cTnI-(147-163) (8). 2 Surprisingly, binding of cTnI-(1-211) to full-length cTnC did not induce a single open conformation in the regulatory domain as judged by multiple NMR detectable conformers for residues Ala 31 , Glu 32 , Gly 34 , Glu 66 , Gly 68 , Val 72 , Asp 73 , and Val 79 located within this region (Figs. 5 and 6). Although chemical shift changes and the presence of multiple conformations do not necessarily indicate a large structural transition from a closed to open conformation, the observed multiple H N chemical shifts for residues in the regulatory domain do show a number of different magnetic environments for each amide resonance. These magnetically different states presumably result from conformational differences in the regulatory domain of cTnC when bound to cTnI. The structural uniqueness of these multiple isoforms is at present unclear. However, based on the available data for the isolated regulatory domain of cTnC binding to cTnI-(147-163) (8), and our R 2 values for the various cTnIC complexes (7), it is likely that the observed conformational isomerism represents exchange between a closed and a more open state for the regulatory domain of cTnC.
The possibility remains that inhibitory motifs within cTnI-(81-211) fail to open the regulatory domain to the degree observed in the skeletal system. Comparison of the structural opening induced by binding cTnI-(147-163) to cTnC-(1-89) with the Ca 2ϩ induced opening of sTnC has led Li et al. (8) to suggest that the regulatory domain in the cardiac complex is less open than observed for the skeletal system. Effects of urea on exchange in the regulatory domain support either further stabilization or an additional opening of the hydrophobic pocket by urea.
The effects of the cardiac specific N terminus of cTnI on the cTnC regulatory domain conformation have not previously been recognized. The unexpected finding that the cardiac specific N terminus of cTnI can also shift the conformational equilibrium of the cTnC regulatory domain suggests a novel and more complex mechanism for activation of cardiac muscle troponin. Based on the data presented in Figs. 5 and 6 and the observation of the open-state specific H N to H N NOE between Ser 37 and Ile 61 in the cTnC⅐cTnI-(1-80) complex, the cardiac specific N terminus may play a role in shifting the equilibrium toward a more open conformation. Phosphorylation or mutation to Asp of Ser 23 and Ser 24 in cTnI does not significantly affect the conformational states observed for the regulatory domain compared with binding cTnI-(33-211), demonstrating that the phosphorylated cardiac specific N terminus does not directly interact with the regulatory domain of cTnC. This was first proposed based on R 2 values for cTnC bound to cTnI-(1-80) and cTnI-(1-80)DD (7).
These results are consistent with a variety of in vitro and functional studies. Phosphorylation of cTnI reduces the fluorescence-detected Ca 2ϩ affinity of site II in the cTnIC complex (41). Replacement of cTnI with cTnI-(33-211) in skinned fiber bundles from rat induces a desensitization to activation by 2 M. B. Abbott, unpublished data. Ca 2ϩ (42). Desensitization to Ca 2ϩ was also observed upon cAMP-dependent protein kinase phosphorylation of cTnI-(1-211) (43). Finally, reconstitution with cTnI-(1-211)DD, mimicking the phosphorylated state, led to a reduced Ca 2ϩ sensitivity in skinned cardiac muscle fibers compared with reconstitution with wild type cTnI (44).
The available experimental evidence supports a novel mode of action for the cardiac specific N terminus during cardiac muscle contraction. Interaction of the cardiac specific N terminus appears to facilitate a "partial opening" of the regulatory domain by altering either the exchange rate and/or the equilibrium between forms. This could result from stabilization of the defunct Ca 2ϩ binding loop as well as portions of helices A and B. Stabilization of the open state strengthens interactions between the C-terminal region of cTnI and the regulatory domain of cTnC. Phosphorylation of the cardiac specific N terminus eliminates the stabilizing effect on the regulatory domain resulting in reduced calcium sensitivity of the troponin switch. This novel mechanism utilizes the unique isoform differences in both cardiac TnC and TnI and suggests a structural mechanism for modulating the cTnC and cTnI interactions that regulate cross-bridge cycling. Chemical exchange in the regulatory domain of Ca 2ϩ -saturated cTnC bound to cTnI indicates that a simple closed to open transition is insufficient to completely describe the conformational isomerism observed in the intact system.
In summary, we have demonstrated that binding of cTnI-(1-211) to Ca 2ϩ -saturated cTnC results in a significant decrease in linker region flexibility as well as a decrease in conformational exchange and a shift toward an open regulatory domain facilitating binding of the C-terminal domain of cTnI. We also describe a novel role for the cardiac N terminus of cTnI. We propose that a region of the cardiac specific N terminus of cTnI interacts with the regulatory domain and shifts the conformational equilibrium toward a more open form, possibly by stabilizing defunct Ca 2ϩ binding site I. Phosphorylation of the cardiac specific N terminus at Ser 23 and Ser 24 results in a loss of interaction between the regulatory domain and the phosphorylated N terminus, resulting in an additional entropic energy barrier to the activation of cardiac muscle contraction. This is completely consistent with available physiological data showing a decrease in Ca 2ϩ affinity upon phosphorylation or deletion of the cardiac specific N terminus (45). Structure determination of cTnC in the presence of the cardiac specific N terminus and the inhibitory region of cTnI will better define the molecular basis for the observed conformational isomerism within the regulatory domain of cTnC.