Allosteric Transmission along a Loosely Structured Backbone Allows a Cardiac Troponin C Mutant to Function with Only One Ca2+ Ion*

Hypertrophic cardiomyopathy (HCM) is one of the most common cardiomyopathies and a major cause of sudden death in young athletes. The Ca2+ sensor of the sarcomere, cardiac troponin C (cTnC), plays an important role in regulating muscle contraction. Although several cardiomyopathy-causing mutations have been identified in cTnC, the limited information about their structural defects has been mapped to the HCM phenotype. Here, we used high-resolution electron-spray ionization mass spectrometry (ESI-MS), Carr-Purcell-Meiboom-Gill relaxation dispersion (CPMG-RD), and affinity measurements of cTnC for the thin filament in reconstituted papillary muscles to provide evidence of an allosteric mechanism in mutant cTnC that may play a role to the HCM phenotype. We showed that the D145E mutation leads to altered dynamics on a μs-ms time scale and deactivates both of the divalent cation-binding sites of the cTnC C-domain. CPMG-RD captured a low populated protein-folding conformation triggered by the Glu-145 replacement of Asp. Paradoxically, although D145E C-domain was unable to bind Ca2+, these changes along its backbone allowed it to attach more firmly to thin filaments than the wild-type isoform, providing evidence for an allosteric response of the Ca2+-binding site II in the N-domain. Our findings explain how the effects of an HCM mutation in the C-domain reflect up into the N-domain to cause an increase of Ca2+ affinity in site II, thus opening up new insights into the HCM phenotype.

Hypertrophic cardiomyopathy (HCM) is one of the most common cardiomyopathies and a major cause of sudden death in young athletes. The Ca 2؉ sensor of the sarcomere, cardiac troponin C (cTnC), plays an important role in regulating muscle contraction. Although several cardiomyopathy-causing mutations have been identified in cTnC, the limited information about their structural defects has been mapped to the HCM phenotype. Here, we used high-resolution electron-spray ionization mass spectrometry (ESI-MS), Carr-Purcell-Meiboom-Gill relaxation dispersion (CPMG-RD), and affinity measurements of cTnC for the thin filament in reconstituted papillary muscles to provide evidence of an allosteric mechanism in mutant cTnC that may play a role to the HCM phenotype. We showed that the D145E mutation leads to altered dynamics on a s-ms time scale and deactivates both of the divalent cationbinding sites of the cTnC C-domain. CPMG-RD captured a low populated protein-folding conformation triggered by the Glu-145 replacement of Asp. Paradoxically, although D145E C-domain was unable to bind Ca 2؉ , these changes along its backbone allowed it to attach more firmly to thin filaments than the wildtype isoform, providing evidence for an allosteric response of the Ca 2؉ -binding site II in the N-domain. Our findings explain how the effects of an HCM mutation in the C-domain reflect up into the N-domain to cause an increase of Ca 2؉ affinity in site II, thus opening up new insights into the HCM phenotype.
Hypertrophic cardiomyopathy (HCM) 6 is the most common familial heart disorder. Macroscopic features include myocardial disarray and left ventricle thickening accompanied by significant calcium imbalance at a molecular level (1,2). The relationship between a myosin heavy chain (MYH7) gene defect and HCM was a groundbreaking discovery that identified HCM as the first cardiac disease linked to an inherited defect. Subsequent reports identified mutations encoding primarily for sarcomeric proteins (3). New mutations in TNNC1 (A8V, A31S, and D145E), the gene that codifies cardiac troponin C (cTnC), have been observed in patients with HCM (4,5).
Cardiac troponin C, a subunit of the troponin complex (Tn), is a key player in regulating muscle contraction at the level of the thin filament (6,7) and belongs to the EF-hand family of Ca 2ϩ -binding proteins. The EF-hand motif is known for its highly conserved helix-loop-helix structure consisting of 12 amino acids organized spatially to coordinate Ca 2ϩ or Mg 2ϩ in a pentagonal bipyramidal arrangement. cTnC comprises two globular domains with a pair of EF-hand motifs in each domain. The N-domain Ca 2ϩ affinity (ϳ10 5 M Ϫ1 ) is lower than that of the C-domain (ϳ10 7 M Ϫ1 ) (8). Moreover, the N-domain site I has lost a Ca 2ϩ -coordinating carboxylate moiety (9), so that myofilament contraction is triggered by the binding of a single Ca 2ϩ ion to site II. The binding of Ca 2ϩ to this site leads to the exposure of hydrophobic patches and a slight opening that favors interaction with troponin I (7).
HCM-causing mutations can lead to changes in the cTnC three-dimensional structure and tend to impair its ability to regulate the Ca 2ϩ binding process and interact with biological partners on the thin filament (10). The single-residue substitution at position 8 in the N-helix of cTnC (A8V) confers an increase of the ␣-helical content (11). Although the overall 1 H- 15 N HSQC spectrum appears similar to that of the wild-type (WT) form, this mutation seems to increase the population of the N-domain open conformation (12). In the case of A31S, the subject carrying this mutation is prone to develop ventricular fibrillation. Although little is known about the structure of A31S, the secondary structure content is the same as that observed for WT, and in reconstituted fibers, Ca 2ϩ affinities and force development are increased (5). The cTnC D145E mutation carries an extra methylene at the Zϩ position of the coordinate sphere at site IV. This mutation increases Ca 2ϩ affinity of the N-domain site and force recovery in reconstituted fibers (4) but drastically decreases Ca 2ϩ binding to the C-domain. To our knowledge, no studies have addressed, at the structural level, the mechanism of how decreased Ca 2ϩ affinity in the D145E C-domain would lead to the paradox of increased affinity of the Ca 2ϩ -binding site II in the N-domain and its associated effects on the HCM phenotype. The structural consequences of HCM-causing mutations not specifically associated with alterations in the Ca 2ϩ coordination sphere (i.e. A8V in the N-helix and A31S in the loop of the defunct Ca 2ϩ -binding site I) were explored for comparison.
Among the main physiological changes associated with cardiomyopathies, the altered concentration of intracellular Ca 2ϩ is a common feature, reflecting the fact that cTnC is one of the major Ca 2ϩ buffers in the cardiomyocyte. A lower affinity of the C-domain for Ca 2ϩ in the cTnC variant (D145E) compared with the WT protein has been reported (13). Here, using an ensemble of thermodynamic and structural approaches including steady-state fluorescence, circular spectropolarimetry, NMR spectroscopy, and electron-spray ionization mass spectrometry (ESI-MS) we describe how altered Ca 2ϩ binding to the C-domain of D145E impacts cTnC function in solution and in permeabilized fibers. Based on ESI-MS and nuclear spin relaxation data, the results suggest that D145E impairs both of the Ca 2ϩ -binding sites of the cTnC C-domain: site IV, due to a direct effect of the mutation at the coordination sphere, and site III as a side effect of the altered dynamics triggered by eliminating Ca 2ϩ at site IV. Carr-Purcell-Meiboom-Gill relaxation dispersion (CPMG-RD) measurements have captured a low populated protein-folding intermediate that is not present in the WT isoform and promotes tighter cTnC binding to permeabilized fibers. Based on our results and previous functional studies with D145E (4, 13), we provide evidence for an allosteric mechanism occurring between the two lobes of cTnC. Our findings explain how the effects of an HCM mutation in the C-domain reflect up into the N-domain to cause an increase of Ca 2ϩ affinity in site II, thus opening up new insights into the HCM phenotype.

D145E Lacks Ca 2ϩ -induced Hydrophobic Changes-To
examine the overall profile of Ca 2ϩ binding, we performed Ca 2ϩ titrations on WT cTnC and three HCM-related mutants (A8V, A31S, and D145E) using bis-ANS (4,4Ј-dianilino-1,1Јbinaphthyl-5,5Ј-disulfonic acid) fluorescence. As shown in Fig.  1A, all curves except that for D145E were biphasic. Following an abrupt increase from pCa ϳ7.0 to 5.5 or 5.0, there was a second transition from pCa ϳ5.0 to 4.0. The magnitudes of these transitions were the same for WT and A8V (F/F 0 ϭ 1.69 and 1.73, respectively), showing that conformational changes associated with hydrophobic exposure during Ca 2ϩ binding were similar (Fig. 1A). In contrast, the magnitude for A31S was much greater (F/F 0 ϭ 2.20), indicating higher hydrophobic exposure upon Ca 2ϩ binding to the C-domain (Fig. 1A). The first transition for A31S was also greater than for the other proteins. Although the first transition was not observed for D145E, the fluorescence magnitude of the second transition was comparable to that for WT and A8V, showing that the hydrophobic exposure of the N-domain was not affected by the mutation (Fig. 1A). Evaluation of bis-ANS emission curves at pCa 9.0, the Ca 2ϩ -deficient form (apo), and 4.0, the Ca 2ϩ -saturated form (holo), revealed a greater hydrophobic exposure of D145E compared with WT at pCa 9.0, suggesting that the apo-D145E C-domain is already in an open conformation (Fig. 1B). A8V and A31S led to only modest exposure of hydrophobic patches at pCa 9.0 and 4.0 when compared with the WT. All curves except that for D145E were fit to two classes of binding sites to obtain K 1 and K 2 ( Table 1).

D145E Inactivates the Binding of Two Ca 2ϩ Ions in Cardiac
TnC-In a previous report (13), similar data obtained with bis-ANS binding to WT, A8V, and D145E proteins were interpreted as reporting first, Ca 2ϩ binding to C-domain sites (lacking D145E), and second, nonspecific Ca 2ϩ binding to weak Ca 2ϩ /Mg 2ϩ sites (non-EF-hand) of the C-domain. None of the bis-ANS signal was assigned to the N-domain, and the question of whether the D145E mutation inactivates site IV was left open (13). To answer this question, we first analyzed the chemical shift perturbations (CSP) obtained by comparing pCa 2.5 against pCa 7.0 (N-domain sensitivity) and pCa 6.0 against pCa 7.0 (C-domain sensitivity) through 1 H-15 N peak assignments (Figs. 2 and 3). CSP plots for WT and D145E revealed differences in perturbed residues in both proteins (Fig. 4A). At low Ca 2ϩ concentration (pCa 6.0 -7.0) CSP values that differed by Ն1 S.D. from the average across all evaluated residues were mapped onto the cTnC structure. There were several differences in the C-domain between WT and D145E (Fig. 4B), distinguishing the two isoforms and pinpointing structural changes triggered by Ca 2ϩ binding. Of note, the N-domain response was also different. For the Ca 2ϩ range of high-affinity sites (pCa 6.0 -7.0), residues Lys-17, Leu-48, Gln-58, Asp-75, and Met-80 are mapped onto the WT N-domain, whereas Ala-23, Phe-27, and Glu-66 are shown at right for the mutant (Fig.  4B). It is evident that structural changes in the N-domain due to Ca 2ϩ in the WT C-domain are different from the structural changes produced in the D145E N-domain. For the Ca 2ϩ range of low-affinity sites (pCa 2.5-7.0), several N-domain residues affected in WT were also affected in D145E, including Asp-25, Gly-30, Gly-34, Gly-42, and Gly-70, but perturbation of Gly-140 and Asp-149 in the C-domain was only observed for the mutant (Fig. 4B). The structural differences mapped here reveal sharp differences in the effect of Ca 2ϩ on the structure of the C-domain of D145E.
To confirm our hypothesis that Ca 2ϩ binding to cTnC C-domain is drastically affected in D145E, we performed high-resolution mass spectrometry analyses of both the WT and D145E prepared in the absence and presence of Ca 2ϩ . First, the accurate molecular mass of WT Table 2). Our results show that D145E has lost the ability to bind two Ca 2ϩ ions.
Increase of Thermal Stability of Cardiac TnC Is Governed by Ca 2ϩ Binding to C-domain Sites-To characterize the thermal stability of WT and HCM variants, using circular dichroism we monitored the loss of secondary structure generated by rising temperatures. The apo-states of all three mutants exhibited a similar sigmoidal transition with an increase in temperature that could be fit as a three-state process. In the holo-state, only D145E maintained the three-state behavior observed for its apoform. Reversibility was observed for all constructs, and the melting temperatures at the midpoint for the first and second transitions (T m ) were extracted (Fig. 6). In the holo-state, because of increased stability for WT, A8V, and A31S, the second transition was not observed. Thus only one T m value was extracted for this condition. Statistical significance for these differences was observed for WT against A31S (p ϭ 0.041) and D145E (p ϭ 0.0003) (Fig. 6). There was no difference in the thermostability of D145E in the absence or presence of Ca 2ϩ . These results show that the binding of Ca 2ϩ to the C-domain sites of cTnC governs the increase of thermostability. The mutant, unable to bind Ca 2ϩ at these sites, retains the twotransition profile of apo-cTnC.
Lower Stability of D145E Does Not Affect Overall Cardiac TnC Structure-A detailed overview of how the mutation at Asp-145 affects the cTnC global conformation was obtained using dihedral angle restraints based on HN, H ␣, H␤, and 15 N chemical shifts and the software TALOSϩ to provide backbone phi and psi angles (14). The ␣-Helix and ␤-sheet contents were similar for D145E and WT, revealing that although the mutation impairs Ca 2ϩ coordination, it does not affect protein secondary content (Fig. 7A). The interatomic function P(r) describes the distribution of distances r between pairs of points within a given volume (the protein particle). The expected P(r) distribution for a bilobular protein such as cTnC is a major peak at low values of r followed by a prominent shoulder at higher values of r, as clearly observed for all three holovariants. There were no significant differences in dimensional values extracted from P(r) distributions when comparing WT and D145E in the holo-state (Fig. 7, B-D). In contrast, higher D max values were obtained for the apo-states, as summarized in Table 3. Tertiary contacts were investigated by small-angle X-ray scattering (SAXS), creating a scattering profile of each protein in its apoand holo-states (Fig. 8). Our results indicate that the mutations studied in the holo-state (A8V, A31S, and D145E) do not affect the overall conformation of cTnC. The effect of Ca 2ϩ -binding impairment on D145E stability is better observed when the protein is challenged by 5 M urea, in the holo-and apo-states. As shown on the Kratky plots, in the presence of 5 M urea, the bell shape of the WT holo-state contrasts with the WT apoform, whereas D145E is similar in both states (Fig. 7E). A continuous increase in I(s)*s 2 values as a function of s in the Kratky plots is consistent with an unfolding process. Altogether our data sug- gest that the D145E mutation does not affect the global secondary and tertiary contacts but has lower stability because of local changes that arise from impaired Ca 2ϩ binding. D145E Affects C-domain Local Dynamics-Because D145E does not affect the overall secondary and tertiary structure of the protein but does affect Ca 2ϩ binding and stability, we designed a protocol for obtaining a complete set of backbone dynamics by measuring amide 15 N longitudinal (R 1 ) and transverse (R 2 ) relaxation times and 1 H- 15 -137 and Gly-140), and ␣H (Phe-153, Glu-155, Met-157, and Gly-159) revealed values of R 2 /R 1 more than 1 S.D. above the average (4.08 Ϯ 0.71) (Fig. 9, B and C). This means that there are distinct regions along the backbone from ␣D to ␣H where N-H motions are slower for the mutant, presumably because of repercussions from loss of Ca 2ϩ binding and lower stability. Of note, helices G and H flank the last EF-hand-binding site (site IV) in the C-domain. Moreover, several ␣D residues in the N-domain sense these slower dynamics occurring in the C-domain (Fig. 9B).
Another type of information resides in the 1 H-15 N heteronuclear NOE values for each residue, which are related basically to internal motions of 1 H-15 N bonds on a much faster (ps-ns) time scale. These values were similar for WT and D145E, with average values across all residues of 0.55 Ϯ 0.18 and 0.53 Ϯ 0.22, respectively (Fig. 9A, lower panel). The N-and C-terminal segments of both proteins presented lower NOE values, a sign of increased flexibility at these sites. In addition, residues 87-91, forming the loop that connects the N-and C-domains, displayed negative values consistent with its flexible behavior (31). Interestingly, the NOE values for these residues tended to be more negative for D145E compared with WT, indicating a more flexible linker (Fig. 9D). Conformational exchange motions on a s-ms time scale were not present in the WT, and the R 2 /R 1 values were not uniformly distributed among residues (Fig. 9A, upper panel), which is consistent with the presence of different correlation times ( c1 and c2 ) for tumbling of the N (residues 1-90)-and C-domains (residues 91-161) of WT cardiac TnC (Fig. 9E).
To quantify the conformational changes in D145E, we turned to 15 N CPMG-RD measurements, which are sensitive to millisecond exchange dynamics. The effective decay constant of magnetization (R 2eff ) evaluated at 25 and 37°C with refocusing pulses of 50 and 1000 Hz revealed intermediate exchange motions for D145E at both temperatures on the order of milliseconds that were not observed for WT cTnC. Of note, these movements, visible for residues 125-157 in the D145E panels of Fig. 10A, were even more pronounced for D145E at 37°C. Additionally, at 37°C some residues of WT acquired fast-exchange motions (s time scale), observed as virtually identical values of R 2eff for the two refocusing pulses of 50 and 1000 Hz. These fast-exchange motions indicate that some residues in the WT N-domain (Asp-3, Glu-14, Gly-30, Glu-32, and Gly-42) do develop fast-exchange motions when challenged at physiologi-   Table 2). All samples were mixed in similar molar concentrations, and direct injection was performed at a constant flow rate (0.4 ml ⅐ min Ϫ1 ) using 50% acetonitrile. Intensity (%) is shown as a function of m/z for 15 N-cTnC WT apo (C) and holo (D) and D145E apo (E) and holo (F).
filament, experiments were carried out with skinned papillary muscles depleted of their native cTnC and reconstituted with one of the two cTnC isoforms. Dissociation of D145E from the thin filament was much slower compared with the WT, indicating a higher affinity for the thin filament (Fig. 11A). The cTnC concentration required for half-maximal binding to the thin filament was about 3-fold greater with the WT protein (0.3 versus 0.1 M), again indicating a higher affinity of D145E for the thin filament (Fig. 11B).

Discussion
TNNC1 is now included in routine genetic screening for HCM mutations (15), but up to this point there has appeared very little structural information showing how these mutations affect the binding affinity of cTnC for Ca 2ϩ and for its protein partners on the thin filament. Cardiomyopathic mutations can be found in both domains of the protein, but they all share a unifying molecular mechanism: accompanied by minimal changes in overall structure, they affect site II Ca 2ϩ binding affinity and kinetics when cTnC is incorporated into the thin filament. It is widely recognized that both domains of cTnC can communicate with each other; however, we still do not know how the information is transmitted from one domain to the other.
CPMG-RD measurements have proven useful for probing the population and exchange rates of marginally stable and/or destabilized mutants (16). The short lifetimes and small populations of these excited conformers present researchers with a difficult task when only conventional structural biology techniques are available. Innovative CPMG-RD applications, either alone or in combination with physical approaches such as high hydrostatic pressure (17), are currently emerging as tools for understanding the molecular mechanisms of protein folding and misfolding (18,19) and protein functionalities (20). Here, our CPMG-RD measurements captured a low populated, high energy conformer of marginal stability caused by a disease-related mutation in site IV that derails Ca 2ϩ coordination at both C-domain sites III and IV. The occurrence of this excited state intermediate does not cause global unfolding but favors a more open C-domain conformation with lower stability and slower internal dynamics in specific segments. As a result, there is an allosteric communication of increased affinity at Ca 2ϩ -binding site II of the N-domain, a characteristic feature of the HCM phenotype (21). These features may explain why D145E has a higher affinity for the thin filament and dissociates more slowly from skinned fibers (Fig. 11), as discussed below.
Our first experiments targeted the question of Ca 2ϩ binding stoichiometry in the isolated D145E protein, an issue that had been left open in previous reports (4,13). Here, a clear response to Ca 2ϩ was obtained for WT but not D145E in measurements of hydrophobic surface exposure (Fig. 1) and thermal stability (Fig. 6). SAXS experiments in the presence of a subdenaturing urea concentration also revealed a Ca 2ϩ response for the WT but not the mutant (Fig. 7E). These results are explained by direct measurements of Ca 2ϩ binding by ESI-MS showing that both C-domain sites of D145E are inoperative (Figs. 4C and 5).
A major distinction between the WT and mutant is evident in the backbone dynamics, captured by conventional relaxation measurements using NMR. Notably, although the fast-exchange (ps-ns) dynamics are for the most part not affected by the mutation (Fig. 9A, lower panel), a slower exchange process appears and may contribute locally toward the loss of Ca 2ϩ binding in site III. Evidence for this contribution comes from D145E C-terminal residues with R 2 /R 1 ratios that are systematically higher than those observed for WT, consistent with s-ms conformational exchange motions (Fig. 9A, upper  panel). In addition, there are repercussions of the mutation for  the dynamics in ␣D of the N-domain (Fig. 9A) and greater flexibility of the central helix (Fig. 9D). These data are important; they provide evidence for transmission of this altered dynamics from the C-to the N-domain. For skeletal TnC, an inverse correlation exists between protein stability and Ca 2ϩ affinity, i.e. an increase in protein stability leads to a decrease in Ca 2ϩ affinity (22,23) and vice versa. Intramolecular coupling between the N-and C-domains has been observed for skeletal TnC (24), and several lines of evidence support this cross-talk mechanism (25)(26)(27)(28), which is probably aided by the bending of the DE-linker helix (29,30). Bending at the DE linker is even more pronounced for the cardiac protein (31,32), and we see the consequences of this N-/C-domain cross-talk in the form of an allosteric event in the N-domain, mitigating in part the potentially catastrophic effects of an HCM-related mutation in the C-domain.
Other studies have analyzed aspects of the internal dynamics of cTnC mutants, restricted to isolated C-or N-domains (33,34). It is worthwhile comparing the results for D145E with data from the dilated cardiomyopathy mutation G159D located at the end of the C-domain. Unlike D145E, this mutation presents a decrease of Ca 2ϩ sensitivity in fibers and myofibrils. Moreover, the G159D mutation triggers different kinetics of conformational transitions (35,36); it is noteworthy that it also had no   Fig. 9A (33). The CPMG analysis adds an important dimension to R 2 /R 1 relaxation rates. Although R 2 /R 1 rates provide initial evidence of an altered local dynamics in the C-domain of D145E, CPMG is sensitive enough to quantify the exchange rates and size of both populations: the major conformer and the low populated excited state present in D145E. Thus D145E appears to be unique in that it inserts a Ca 2ϩ binding flaw in the C-domain that has a local impact on protein dynamics through the population of a kinetically trapped excited state. The open conformation, slower motions, and local unfolding of ␣H in the D145E C-domain, together with the allosteric effect at Ca 2ϩ -binding site II of the N-domain, are events that plausibly favor cTnC contacts with other proteins of the thin filament, thereby accounting for this mutant having tighter binding and slower dissociation from skinned fibers (Fig. 12).
Overall, the D145E mutant functions remarkably well in the laboratory; it binds to the thin filament and contributes to fullfledged regulation (both activation and inhibition, including cross-bridges) with only a small increment in Ca 2ϩ sensitivity of the N-domain. With regard to the question of global structural changes caused by the D145E mutation, our experiments with skinned fibers provided unexpected evidence for some degree of loose folding of the C-domain. Sites III and IV are traditionally considered "structural" sites, always occupied by Ca 2ϩ or Mg 2ϩ and essential for maintaining cTnC bound to the thin filament but not involved in beat-to-beat regulation (8). Experiments like that of Fig. 11 provide a means of comparing the efficacy of the C-domain structure in anchoring cTnC to the filament when sites III and IV are altered. Both experiments demonstrated that the D145E mutant has a higher affinity for the thin filament than the WT isoform, even though its C-domain does not bind cations. The D145E C-domain is demonstrably rather open, with an exposed hydrophobic region equivalent to that found for Ca 2ϩ -bound WT (Fig. 1B). Normally, TnC in skinned fibers requires the presence of Ca 2ϩ or Mg 2ϩ to induce the formation of a stable complex with its partners on the thin filament. We infer that this mutant has acquired a C-domain conformation that already favors this complex. Thus the C-domain may provide an attractive target for small molecules that have the potential to exert a long distance effect on the Ca 2ϩ sensitivity of the N-domain. Furthermore, the proposed allosteric modulation could be extended to TnC orthologs and other EF-hand proteins such as calmodulin or myosin essential light chain. An understanding of how Ca 2ϩbinding proteins change during evolution, preserving (or losing) their ability to regulate biological function, can provide information for the engineering of new molecules. Also, targeting this allosteric mechanism seems uniquely suited to the task of exploring the dynamics and activity of these Ca 2ϩ sensors (37)(38)(39).

Experimental Procedures
Protein Expression and Purification-The pET-3d vector containing the sequences of human cardiac TnC WT and the constructs (A8V, A31S, and D145E) were inserted into BL21 DE3 Escherichia coli strain, plated on semi-solid Luria-Bertani (LB) medium and incubated at 37°C for ϳ16 h. The colonies were transferred to 25 ml of LB and placed in a shaker at 37°C at 180 rpm for 12 h. Next day, 5 ml of the culture was transferred to an Erlenmeyer flask containing 800 ml of LB, and cells were grown at 37°C at 180 rpm until absorbance (A 600 nm ) reached 0.8 -1.0. All cultures contained ampicillin at 50 g⅐ml Ϫ1 . Induction was achieved with 1 mM isopropyl ␤-D-1-thiogalac-topyranoside followed by additional growth at 37°C and 180 rpm for 7 h. Samples were centrifuged for 10 min at 8000 ϫ g and 4°C and stored at Ϫ20°C until needed. Lysis was per- formed by sonication in buffer A containing 50 mM Tris-Cl (pH 7.8), 6 M urea, 1 mM EDTA, and 1 mM fresh dithiothreitol (DTT) followed by 20 min at 15,000 ϫ g and 4°C. The supernatant was loaded onto a Q-Sepharose column (coupled to an Ä kta Purifier system) equilibrated previously in buffer A. Elution was performed with a linear gradient of 300 ml from 0 to 0.6 M KCl added to buffer A. The most enriched fractions observed at A 280 nm were collected and dialyzed against 50 mM Tris-Cl (pH 7.5), 50 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , and 1 mM DTT and loaded onto a Superdex 75 prep grade column (GE Life Sciences) equilibrated previously in the same buffer. Eluted fractions were concentrated when necessary in a 10,000 MWCO Amicon (EMD Millipore, catalog No. 901024) and stored at Ϫ80°C. 15 N-and 15 N-13 C-labeled proteins were produced as described previously (40) and purified as above.
Fluorescence Spectroscopy-Fluorescence emission was recorded using an ISSK2 spectrofluorometer (ISS Inc., Champaign, IL). For Ca 2ϩ titrations, 2 M of bis-ANS fluorescence was excited at 360 nm and monitored at 490 nm and 15°C. Microliter amounts of Ca 2ϩ (ϳ0.5-2 l) were added to buffer containing 2 M protein and 200 mM MOPS (pH 7.0), 150 mM KCl, 2 mM EGTA, and 1 mM DTT (final reaction volume, 3 ml). Titrations between pCa 2.5 and 9.0 at free 2 mM [Mg 2ϩ ] were calculated using the pCa Calculator suite (41). The pH was 7.0 at the end of titrations. Results were fitted using OriginPro software (Northampton, MA) with the following equation, which describes two classes of binding sites (42), where F is the fluorescence at each [Ca 2ϩ ], F 0 is the fluorescence at pCa 9.0, A 1 and A 2 are the magnitudes of transitions, n1 is the Hill coefficient, and K 1 and K 2 are [Ca 2ϩ ] at the midpoints of the two transitions. Experiments are shown as average Ϯ S.E. of three independent experiments using at least two different protein batches.
NMR Spectroscopy-WT and D145E peak assignments were carried out at the Brazilian Biosciences National Laboratory (LNBio), Campinas, Brazil. We acquired a set of three-dimensional experiments (HNC ␣ C ␤ , C ␤ C ␣ (CO)NH, HN(C ␣ )CO, and HNCO) using 15 N-13 C-labeled proteins (ϳ0.5-1 mM) at 25°C in a Varian Inova, 600 MHz, equipped with a triple resonance cryogenic probe. Buffer conditions were 20 mM MOPS (pH 7.0), 6 mM CaCl 2 , 1 mM MgCl 2 , 100 mM KCl, and 10 mM DTT containing 10% D 2 O. 1 H-15 N peak assignments are depicted in Figs. 2 and 3. The CARA 1.8.4 platform (Computer Aided Resonance Assignment) was used for assignment. To compare the secondary structure predictions for WT and D145E using the TALOSϩ suite (14), we acquired H ␤ H ␣ (CO)NH spectra for both proteins, thus completing the H ␣ and H ␤ resonance assignments.
For NMR Ca 2ϩ titrations, heteronuclear 1 H-15 N HSQC NMR spectra were acquired at 25°C in a Bruker Avance III 800-MHz or 500-MHz spectrometer equipped with a 5-mm TXI probe at the National Center of Nuclear Magnetic Resonance Jiri Jonas (Rio de Janeiro, Brazil). WT and D145E were recorded at pCa 9.7 (apoform), 7, 6.5, 6, 5.8, 5.6, 5.2, 4.5, 3.9, 3.6, 3.3, 3, 2.7, 2.5, and 2. The 1 H-15 N HSQC spectra were acquired with 1024 increment points for the 1 H dimension and 128 points for the 15 N dimension, with 16 scans collected at each increment. CSP analysis was performed at pCa 2.5 and 6.0 against pCa 7.0 for WT and D145E using the following equation, where ⌬␦ H and ⌬␦ N are the chemical shift variations between Mass Spectrometry-Samples were analyzed with a high capacity ion trap LC/MSn system HCT Ultra ETD (Bruker Daltonik, Germany) equipped with a standard ESI ion source with full scan data in UltraScan mode between m/z 400 and 3000, in positive mode and with a fixed accumulation time of 30 ms. The ionization parameters of nebulizer gas, dry gas, and dry temper-ature were 230 psi, 5 liters ⅐ min Ϫ1 , and 300°C, respectively. The ion spray voltage was set to 2500 V, and the maximal accumulation time was set to 200 ms. Extracted ion chromatograms were used for identification of the expected molecular weights. All data were acquired and processed using Compass 1.3 software (Bruker Daltonik, Bremen, Germany). Calculations including calibration curve regressions, sample concentrations, and statistics were performed using GraphPad Prism 5.0 software (GraphPad, San Diego, CA). Protein concentration was 100 M in 200 mM MOPS (pH 7.0), 150 mM KCl, and 2 mM EGTA (for the apocondition) and 4 mM CaCl 2 (for the holocondition).
Circular Dichroism Spectropolarimetry-To measure the effects of cTnC mutations on protein stability, thermograms ranging from 20 to 90°C were performed in a Chirascan spectropolarimeter (Applied Photophysics). Samples were heated at a rate of 1°C ⅐ min Ϫ1 in a quartz cell of 1-mm path length, and ellipticity was recorded at 222 nm every 0.2°C. Mean residue ellipticities () MRE in deg ⅐ cm 2 ⅐ dmol Ϫ1 ⅐ res Ϫ1 were calculated using the following equation, where is the measured ellipticity in millidegrees, MW is the molecular weight in Daltons, n is the number of residues in the primary sequence, l is the path length in cm, and c the concentration in mg ⅐ ml Ϫ1 . Reversibility was observed for all transitions, and experimental points were fitted using the following biphasic dose-response function in OriginPro, X-ray Scattering Data Collection and Analysis-SAXS measurements were collected on the SAXS1 beamline of the National Synchrotron Light Laboratory (Campinas, Brazil) using a parallel mica sample holder, a Pilatus 300K detector (Dectris, Switzerland), and a wavelength of 0.155 nm at 20°C. Isolated cTnC mutants were scattered at an s range of 0.11 to 4.07 nm Ϫ1 . Scattering was performed by taking six frames of 5 s with a time delay of 2 s. In buffer alone, a single frame of 100 s was taken. Individual frames from scattered proteins were superimposed and visually inspected prior to averaging. Because no radiation effects were observed among successive frames, averaged values were used to enhance the signal-to- FIGURE 11. Dissociation from and binding of cTnC D145E to thin filaments in skinned rat cardiac fibers. A, the D145E mutation significantly slows the dissociation of cTnC from the thin filament. Native cTnC was extracted and fibers were reconstituted with human recombinant cTnC WT or D145E. The dissociation of cTnC is represented by the fall in tension after cTnC extraction and reconstitution. RECON, tension recovery following incubation with the same protein after the dissociation curve. B, the D145E mutation significantly increases cTnC binding affinity to the thin filament. Native cTnC was extracted, and fibers were successively incubated with increasing concentrations of human recombinant cTnC WT or D145E (see "Experimental Procedures" for further details). *, p Ͻ 0.05 for D145E versus WT at the same time point (A) or at same cTnC concentration (B). noise ratio. Parasitic scattering from the mica sample holder and buffer was properly subtracted from the corresponding protein scattering curve. Buffer contained 200 mM MOPS (pH 7.0), 100 mM KCl, and 1 mM EGTA (for the apocondition) and in addition 2.075 mM MgCl 2 and 1.096 mM CaCl 2 to yield a free [Ca 2ϩ ] of 10 Ϫ4 M and free [Mg 2ϩ ] of 2 mM (for the holocondition) without or with 5 M urea. Protein concentrations ranged from 2 to 4 mg ⅐ ml Ϫ1 .
Subtracted intensities were fitted using autoGNOM (PrimusQt suite) in which an indirect Fourier transformation routine is implemented to obtain the pair distance distribution function, P(r). The radii of gyration (R g ) were computed from Guinier plots using the angular coefficient (␣) of the low s region (s Ͻ 1.3/R g ) and checked against those derived from the P(r) function.
Ab Initio Shape Restorations from Scattering Data-The GASBOR software (43) provided by the ATSAS package is suitable for ab initio reconstructions using dummy residues and was chosen for modeling cTnC variants. For each model, 10 runs were generated in GASBOR considering 161 dummy residues in the asymmetric part, no symmetry (P1), and scattering intensity as the mode selected for fitting. Models were superimposed by SUPCOMB and averaged using DAMAVER. The normalized spatial discrepancy (NSD) was used as a filter to include models in the averaging procedure. The inclusion criterion derived from NSD was NSD Ͻ mean NSD ϩ 2* variation.
Protein Dynamics-The diffusion properties of cTnC WT and D145E were obtained at 37°C from the longitudinal (R 1 ) and transverse (R 2 ) relaxation rates. R 1 and R 2 values were measured from 1 H- 15 where T is the relaxation time, The exchange rate constant, K ex (i.e. K ex ϭ k N 3 I ϩ k I 3 N ), was obtained by the following equation.
Optimization of the theoretical model was carried out using the software CPMG_FIT9 (19) and minimizing the equation, where superscripts calc and exp refer to the theoretical and experimental relaxation rates, respectively. All experiments were carried out in 200 mM MOPS (pH 7.0), 150 mM KCl, 2 mM EGTA, 2 mM free Mg 2ϩ , and 1 mM DTT at pCa 2.5. Experiments with Fibers-The rates of association and dissociation of cTnC to and from the thin filaments were evaluated in permeabilized papillary muscles dissected from the left ventricle of Wistar rats, prepared as described (45) with a protease inhibitor mixture (1% w/v) and Triton X-100 (1% w/v) dissolved in relaxing solution (20 mM imidazole (pH 7.0), 152 mM potas- FIGURE 12. Illustration of the structural effects of D145E on thin filaments. The proposed allosteric mechanism of Ca 2ϩ binding with higher affinity to the N-domain site II due to a loose D145E C-terminal domain is presumably transferred through the connecting elements, linker D/E and ␣D, and is extended to a higher affinity of D145E to the thin filament (dark green). The structure of actin-tropomyosin was obtained from PDB 4A7F. The structure of the core domain of troponin complex was obtained from PDB 1J1E. sium propionate, 2.5 mM magnesium acetate, 5 mM EGTA, and 2.5 mM K 2 Na 2 ATP) made up in 50% glycerol. After 24 h at Ϫ20°C, fibers were transferred to a solution without Triton X-100 and stored at Ϫ20°C for up to 2 weeks. Prior to performing assays for binding or dissociation of cTnC, a bundle (ϳ200-m diameter) was attached to the transducer and stretched to 1.2ϫ rest length for an initial force measurement at pCa 4.4 (P(o)). All subsequent measurements were made with temperature control (15°C) and rapid stirring. Native cTnC was extracted using 15-s exposures to Rigor solution containing 0.5 mM trifluoperazine, 20 mM imidazole (pH 6.9), 165 mM potassium propionate, and 10 mM EDTA alternating with 15 s in relaxing solution (see above but without glycerol) for 40 min. After extensive washing to remove trifluoperazine, extraction of native TnC was verified by recording the residual tension at pCa 4.4 (20 mM imidazole (pH 7.0), 152 mM potassium propionate, 6 mM magnesium acetate, 5 mM Ca-EGTA, and 4.4 mM K 2 Na 2 ATP). For the binding experiments, the residual force values post-extraction were 10.5 Ϯ 2.0 and 6.3 Ϯ 1.9% (P/P(o)) for fibers reconstituted with WT and D145E, respectively. For the dissociation experiments, the residual force values postextractions were 11.9 Ϯ 4.7 and 21.0 Ϯ 5.7% (P/P(o)) for fibers reconstituted with WT and D145E, respectively.
For estimates of HcTnC affinity for the thin filament, binding curves were constructed using 0.025-2 M WT or D145E cTnC to reconstitute papillary fibers in relaxing solution adjusted to 2 mM free Mg 2ϩ without Ca 2ϩ (20 mM imidazole (pH 7.0), 152 mM potassium propionate, 6.4 mM magnesium acetate, 5 mM EGTA, 4.4 mM K 2 Na 2 ATP, and 1 mM DTT). Successive 20-min incubations at a low concentration of cTnC were followed by tests at pCa 4.4 until no further increment in tension occurred. The procedure was repeated with a second, higher concentration of protein and, finally, with a drop of protein (55 M) applied directly to the fiber for 1 min (in air). These values were no greater than those obtained by incubation with 2 M protein, indicating saturation of the TnC-binding sites at 2 M. On average, force recovery after this step was 67.4 Ϯ 2.8% (WT) and 70.1 Ϯ 2.9% (D145E) of the original P(o) value with native cTnC and was not different for the two recombinant proteins.
For dissociation of cTnC from the thin filament, fibers were depleted of native cTnC and reconstituted immediately with a drop of protein (55 M) applied directly to the fiber. The reconstitution values prior to the dissociation curves were 72.1 Ϯ 4.7 and 86.7 Ϯ 4.4% (P/P(o)) for WT and D145E, respectively. When a test with pCa 4.4 revealed saturation of the cTnCbinding sites, the fiber was exposed to a relaxing solution buffered with EDTA to contain 0.1 mM MgATP and 1 M free Mg 2ϩ (20 mM imidazole (pH 7.0), 43 mM potassium propionate, 2.3 mM magnesium acetate, 10 mM EDTA, and 1.6 mM K 2 Na 2 ATP). This Mg 2ϩ concentration is well below that required to occupy the C-domain sites. This solution was used because cTnC binds to the cardiac thin filament with a very high affinity, and in the absence of divalent cations bound to C-domain sites, cTnC dissociation can be measured in shorter time spans preventing any effect of fiber rundown. Dissociation of WT and D145E cTnC in this solution was tested using brief applications of pCa 4.4 to record the loss of isometric tension over a period of 60 min followed by a final reconstitution with a drop of the recombi-nant protein to ascertain recovery. On average, this value was 91.5 Ϯ 3.4% (WT) and 86.7 Ϯ 4.4% (D145E) of the original force recorded initially with cTnC in the fiber, and it was not different for the two recombinant proteins.
Statistical Analysis-The results for statistical analysis are shown as the average Ϯ S.E. of at least three independent experiments. Statistical significance of affinity constants (K 1 and K 2 ) extracted from the fit of bis-ANS and skinned fiber binding and dissociation assays were tested using unpaired Student's test against the WT protein. Melting temperatures (T m ) obtained from the fit of circular dichroism thermograms were tested for statistical significance using one-way analysis of variance followed by Dunnett's test for post-hoc comparisons. p values lower than 0.05 were considered significantly different and are described throughout this article.