The Importance of the Carboxyl-terminal Domain of Cardiac Troponin C in Ca2+-sensitive Muscle Regulation*

The interactions between troponin I and troponin C are central to the Ca2+-regulated control of striated muscle. Using isothermal titration microcalorimetry we have studied the binding of human cardiac troponin C (cTnC) and its isolated domains to human cardiac troponin I (cTnI). We provide the first binding data for these proteins while they are free in solution and unmodified by reporter groups. Our data reveal that the C-terminal domain of cTnC is responsible for most of the free energy change upon cTnC·cTnI binding. Importantly, the interaction between cTnI and the C-terminal domain of cTnC is 8-fold stronger in the presence of Ca2+ than in the presence of Mg2+, suggesting that the C-terminal domain of cTnC may play a modulatory role in cardiac muscle regulation. Changes in the affinity of cTnI for cTnC and its isolated C-terminal domain in response to ionic strength support this finding, with both following similar trends. At physiological ionic strength the affinity of cTnC for cTnI changed very little in response to Ca2+, although the thermodynamic data show a clear distinction between binding in the presence of Ca2+and in the presence of Mg2+.

The interactions between troponin I and troponin C are central to the Ca 2؉ -regulated control of striated muscle. Using isothermal titration microcalorimetry we have studied the binding of human cardiac troponin C (cTnC) and its isolated domains to human cardiac troponin I (cTnI). We provide the first binding data for these proteins while they are free in solution and unmodified by reporter groups. Our data reveal that the C-terminal domain of cTnC is responsible for most of the free energy change upon cTnC⅐cTnI binding. Importantly, the interaction between cTnI and the C-terminal domain of cTnC is 8-fold stronger in the presence of Ca 2؉ than in the presence of Mg 2؉ , suggesting that the Cterminal domain of cTnC may play a modulatory role in cardiac muscle regulation. Changes in the affinity of cTnI for cTnC and its isolated C-terminal domain in response to ionic strength support this finding, with both following similar trends. At physiological ionic strength the affinity of cTnC for cTnI changed very little in response to Ca 2؉ , although the thermodynamic data show a clear distinction between binding in the presence of Ca 2؉ and in the presence of Mg 2؉ .
Cardiac and skeletal muscle contraction is regulated by the troponin complex and tropomyosin. The troponin complex consists of three proteins: the Ca 2ϩ binding subunit, troponin C (TnC), 1 the inhibitory subunit, troponin I (TnI), and troponin T (TnT) which anchors the troponin complex to the thin filament. The binding of Ca 2ϩ to TnC results in conformational changes and altered interactions within the thin filament which ultimately lead to muscle contraction (1). Central to the transmission of the Ca 2ϩ signal are altered interactions between TnC and TnI.
Skeletal and cardiac TnC share ϳ70% sequence identity and consist of two globular domains (N-and C-terminal) connected by a central helix (2). Both proteins have four EF hand divalent cation binding sites, sites I and II in the N-terminal domain and sites III and IV in the C-terminal domain. All four sites bind Ca 2ϩ in skeletal TnC (skTnC), but in cardiac TnC (cTnC) site I is unable to bind divalent cations (3).
Sites I and II of skTnC and site II of cTnC are low affinity Ca 2ϩ -specific binding sites (K a(Ca) ϭ 5 ϫ 10 6 M Ϫ1 and 2 ϫ 10 6 M Ϫ1 for skTnC and cTnC, respectively) (4,5). It is Ca 2ϩ binding to these sites which is proposed to act as the trigger for the initiation of muscle contraction. In skTnC this involves an "opening" of the structure with increased exposure of an extensive hydrophobic patch to which skeletal troponin I (skTnI) binds, ultimately relieving inhibition on actin (6). There is, however, evidence that the details of interaction between TnC and TnI may differ in cardiac and skeletal muscle as the Ca 2ϩsaturated N-terminal domain of cTnC exists in a closed conformation (7), and a recent report suggests that the binding of cTnI 147-163 is required for the "opening" of this domain (8).
Sites III and IV in cTnC are high affinity Ca 2ϩ binding sites that can also bind Mg 2ϩ with a lower affinity (K a(Ca) ϭ 3 ϫ 10 8 M Ϫ1 and K a(Mg) ϭ 3 ϫ 10 3 M Ϫ1 ) (5). Under physiological conditions these sites are always occupied by Mg 2ϩ or Ca 2ϩ . The C-terminal domain of TnC has been proposed as a Ca 2ϩ -independent structural binding site for TnI and TnT, helping to anchor the troponin complex to the thin filament (9).
Although the crystal and solution structures of skTnC and cTnC are well characterized (2,7,10,11) only low resolution data are available for the structure of TnI (12)(13)(14). Several lines of evidence indicate that TnI lies antiparallel to TnC and that the two proteins interact at several sites along their length (1). The structures of skTnC bound to skTnI 1-47 and cTnC C-domain bound to the equivalent region of cTnI (residues  show this region of TnI to be ␣-helical and to bind to the hydrophobic pocket of the C-terminal domain of TnC by both polar and Van der Waals interactions (15,16). Both of these studies were carried out with Ca 2ϩ -saturated TnC, although it has not been determined if the C-terminal sites are bound with Mg 2ϩ at all times, or if the increase in [Ca 2ϩ ] on activation is of sufficient duration to displace Mg 2ϩ in the contractile state. In relaxed muscle, at low [Ca 2ϩ ], interactions of the C-terminal domain of TnC with the N-terminal region of TnI will still persist, but the inhibitory region of TnI (residues 104 -115 in skTnI, residues 136 -147 in cTnI) and a region C-terminal to this (residues 140 -148 in skTnI, residues 152-199 in cTnI) are thought to bind to actin, inhibiting the actomyosin ATPase. On activation, when the regulatory site(s) of TnC are occupied by Ca 2ϩ , these sites on TnI bind to TnC, relieving inhibition (for review, see Ref. 17). Although much is known about the interaction of TnI with TnC, further information is required for us to understand fully their role in muscle regulation.
In this study isothermal titration microcalorimetry (ITC) has been used to investigate the interaction of cTnI with cTnC and its isolated domains. We have for the first time investigated the * This work was supported in part by the Wellcome Trust. 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.
‡  1 The abbreviations used are: TnC, troponin C; TnI, troponin I; TnT, troponin T; cTnI, human cardiac troponin I; cTnC, human cardiac troponin C; cTnC N-domain, recombinant human cardiac troponin C N-terminal domain (residues 1-91); cTnC C-domain, recombinant human cardiac troponin C C-terminal domain (residues 91-161); skTnI, troponin I from skeletal muscle; skTnC, troponin C from skeletal muscle; ITC, isothermal titration microcalorimetry; MOPS, 3-(N-morpholino)propanesulfonic acid; lnf, logarithm of fringe displacement; Pipes, 1,4-piperazinediethanesulfonic acid. effect of ionic strength and Ca 2ϩ on the free energy, enthalpy, and entropy changes associated with the binding of cTnC and its isolated domains to cTnI. Our results suggest that the C-terminal domain of cTnC may play a more important role in cardiac muscle regulation than previously thought.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonucleases, DNA-modifying enzymes, and deoxynucleotides were purchased from New England Biolabs, Taq polymerase was purchased from Roche Diagnostics Ltd., and oligonucleotide primers were produced by Alta Bioscience, University of Birmingham. pET11c(cTnC) and pET11c(cTnI) DNA constructs were kindly provided by Dr. E. Al-Hillawi. Inorganic chemicals were of Analar quality from BDH Laboratory Supplies. DEAE-Sepharose Fast Flow, CM-Sepharose Fast Flow, and Superdex 75 were obtained from Amersham Pharmacia Biotech. BCA protein assay reagent (bicinchoninic acid) was obtained from Pierce Chemical Company and used as per the manufacturer's instructions.
Expression and Purification of cTnC, cTnC N-domain, and cTnC C-domain-Polymerase chain reaction was used to generate DNA fragments encoding the isolated N-terminal domain of cTnC (cTnC Ndomain, residues 1-91 inclusive of N-terminal Met), and cTnC C-domain (residues 91-161). The 5Ј-polymerase chain reaction primer (5Ј-GGGAATTCATATGGATGACATCTACAAGGCTGC-3Ј) for the cTnC N-domain was designed to amplify pET11c(cTnC) DNA from the first amino acid and also encoded an NdeI restriction site. The 3Ј-primer (5Ј-AAGGATCCCTACCCTTTGCTGTCGTCCTT-3Ј) corresponded to amino acids 86 -91 and also encoded a stop codon upstream of a BamHI site. To generate the cTnC C-domain a 5Ј-amplification primer (5Ј-GGGAATTCATATGGGGAAATCTGAGGAGGAG-3Ј) was designed to amplify the pET11c(cTnC) DNA sequence from Gly-91 to the 3Ј-end of the coding sequence. This 5Ј-primer also encoded an NdeI restriction site. The 3Ј-primer (5Ј-GGGGATCCCTACTCCACACCCTTCATGAAC-3Ј) was complementary to pET11c(cTnC) DNA encoding the C terminus of cTnC; this primer also encoded a stop codon and BamHI site.
The resulting polymerase chain reaction products were purified to a single band by gel electrophoresis on a 7.5% polyacrylamide gel and electroeluted. The clean products were digested with NdeI and BamHI, ligated into NdeI/BamHI-digested pET11c vector, and used to transform competent Escherichia coli JM101 cells. Successful clones were screened for by restriction analysis and verified by DNA sequencing and by amino acid sequencing of the first 10 residues of the purified protein.
The cTnC, cTnC N-domain, and cTnC C-domain pET11c constructs were transformed into and expressed in E. coli BL21(DE3) cells. The cells were grown at 37°C in NZCYM medium with 0.3 mM ampicillin until they reached an A 600 of 0.8-1. They were then induced with 0.5 mM isopropyl ␤-D-thiogalactopyranoside and incubated for a further 4 -5 h. The cells were harvested by centrifugation at 5,000 ϫ g for 10 min. Harvested cells were homogenized into 25 mM triethanolamine hydrochloride, pH 7.5, 8 M urea, 2 mM EDTA, 1 mM dithiothreitol. The homogenate was then sonicated for 9 ϫ 20 s at power level 6 using a Heatsystems Inc. sonicator. Insoluble cell debris was pelleted by centrifugation at 75,000 ϫ g for 40 min at 4°C. The cell extract was loaded onto a DEAE-Sepharose Fast Flow column (2.5 ϫ 20 cm), equilibrated with 25 mM triethanolamine hydrochloride, pH 7.5, 8 M urea, 2 mM EDTA, 1 mM dithiothreitol buffer, and eluted with a linear 0 -0.5 M NaCl gradient. The fractions containing cTnC or its domains were identified by absorption at 280 nm and SDS-polyacrylamide gel electrophoresis. These fractions were dialyzed against 1 mM ammonium bicarbonate and freeze dried. The lyophilized protein was dissolved in 25 mM triethanolamine hydrochloride, pH 7.5, 150 mM NaCl, 1 mM dithiothreitol, loaded on a Superdex-75 column (2.5 ϫ 50 cm), and eluted in the same buffer. The fractions containing cTnC or its domains were pooled and dialyzed against 1 mM ammonium bicarbonate, freeze dried and stored at Ϫ20°C. The isolated proteins were judged to be ϳ95% pure by SDS-polyacrylamide gel electrophoresis analysis.
Expression and Purification of cTnI-The expression of cTnI was as described by Al-Hillawi et al. (18). Ammonium sulfate (30% w/v) was added to cTnI cell extract in 25 mM triethanolamine hydrochloride, pH 7.5, 8 M urea, 2 mM EDTA, 1 mM dithiothreitol buffer, and the solution was centrifuged at 15,000 ϫ g for 10 min. The cTnI remained in the supernatant and was dialyzed against 25 mM triethanolamine hydrochloride, pH 7.5, 8 M urea, 2 mM EDTA, 1 mM dithiothreitol. The cTnI was then loaded onto a CM-Sepharose Fast Flow column (2.5 ϫ 20 cm) equilibrated with the same buffer and eluted with a linear 0 -0.5 M NaCl gradient. The fractions containing cTnI were identified by absorption at 280 nm and SDS-polyacrylamide gel electrophoresis, pooled, dialyzed against 1 mM HCl, freeze dried, and stored at Ϫ20°C.
Analytical Methods-12% SDS-polyacrylamide gel electrophoresis gels were run at 35 mA and stained with Coomassie Blue (19). DNA and N-terminal protein sequencing were carried out on Applied Biosystems 373A and 437A automated sequencers respectively by Alta Bioscience, School of Biosciences, University of Birmingham. Protein concentrations were determined by the bicinchoninic acid method calibrated with bovine serum albumin (Pierce Chemical Company).
Isothermal Titration Microcalorimetry-Experiments were carried out using a Microcal Inc. isothermal titration microcalorimeter. Proteins were dialyzed extensively against 20 mM MOPS, pH 7.0, 3 mM MgCl 2 , 0.5 mM EGTA, Ϯ 1 mM CaCl 2 , with 0 -0.3 M KCl and experiments performed at 30°C. Protein samples were clarified by centrifugation at 86,000 ϫ g for 20 min with a TL-100 ultracentrifuge (Beckman Instruments Co.) prior to protein concentration determination. The sample cell was filled with 1.4 ml of cTnI solution and titrated with cTnC, cTnC N-domain, or cTnC C-domain. Protein concentrations were 3 M (ϩCa 2ϩ ) or 6 M (ϪCa 2ϩ ) cTnI solution in the cell titrated with 60 M (ϩCa 2ϩ ) or 160 M (ϪCa 2ϩ ) cTnC or cTnC C-domain. When cTnI was titrated with cTnC N-domain in the presence of Ca 2ϩ concentrations were 6 and 160 M, respectively. The injection size was 5 l, with duration of 10 s, at 210-s intervals with stirring at 350 rpm. Control titrations of buffer with cTnC or its isolated domains and titrations of cTnI with buffer indicated that heats of dilution were small and constant. In each experiment the heat of dilution was obtained from additional injections following complete saturation and was subtracted from the binding isotherm. Origin TM ITC data analysis software (Microcal Inc.) was used in the "one set of sites" mode to analyze all binding isotherms.
Sedimentation Equilibrium-These studies were performed with a model E analytical ultracentrifuge (Beckman Instruments Co.) equipped with Raleigh interference optics. The method used was the long column meniscus depletion technique as described by Chervenka (20). An AN-D rotor was used together with a 12-mm double-sector synthetic boundary cell fitted with sapphire windows. All of the samples were dialyzed overnight against a buffer containing 20 mM MOPS, pH 7.0, 3 mM MgCl 2 , 0.5 mM EGTA Ϯ 1 mM CaCl 2 and either 0, 0.05, 0.1, or 0.3 M KCl. Protein samples were centrifuged at 86,000 ϫ g for 20 min with a TL-100 ultracentrifuge prior to analytical ultracentrifugation. Ultracentrifugation data were obtained with a rotor speed of 35,000 rpm at 30°C over a range of cTnI concentrations from 10 to 30 M. The interference patterns were photographed on Kodak TSK 400 film. The developed film was then read on a Nikon two-dimensional microcomparator. The fringe displacement as a function of radial distance was then measured for two fringes and the average obtained. Equilibrium was established when no further fringe displacement occurred with time and was attained after a 6-h ultracentrifugation. The average molecular mass for the whole solution was calculated from the slope of the plot of the natural logarithm of the fringe displacement (lnf) against the distance from the center of rotation squared (r 2 ). All fringe displacements greater than 100 m were included in the least squares analysis. The relationship between the M r and slope of the lnf versus r 2 plot is given by Equation 1, where R is the gas constant expressed as 8.313 J K Ϫ1 mol Ϫ1 , is the angular velocity (2/60 * rpm (rad. s Ϫ1 )), is the solvent density, and the partial specific volume of the protein (0.72 g ml Ϫ1 ). The was calculated from the known amino acid composition of cTnI (21). The density was estimated using standard tables (22).

RESULTS
The Effect of Temperature on the ⌬H of cTnC⅐cTnI Binding-To determine the optimum temperature at which to monitor cTnC⅐cTnI binding, the effect of temperature on the binding enthalpy was studied between 16 and 37°C (Fig. 1). In the absence of Ca 2ϩ (but in the presence of Mg 2ϩ ) the ⌬H was ϩ4 kJ mol Ϫ1 at 16°C. As the temperature was increased, ⌬H decreased linearly to Ϫ46 kJ mol Ϫ1 indicating exothermic binding at physiological temperatures. The slope of this plot yields ⌬C p , the heat capacity of the binding process. ⌬C p was Ϫ2.4 Ϯ 0.1 kJ mol Ϫ1 K Ϫ1 in the absence of Ca 2ϩ . These results also indicate that the minimum temperature at which ⌬H was large enough to obtain good data in the absence of Ca 2ϩ was 30°C. This temperature was therefore used for all further experi-ments. In the presence of Ca 2ϩ the binding process was considerably more exothermic across the entire temperature range, although ⌬C p was similar (Ϫ2.2 Ϯ 0.1 kJ mol Ϫ1 K Ϫ1 ). The large negative value of ⌬C p upon cTnC⅐cTnI binding is probably accounted for by a decrease in the solvent-exposed nonpolar surface area (23). This could be the result of movement of apolar residues within cTnC and/or cTnI on interaction or, more probably, the burial of hydrophobic residues at the binding interface. The fact that ⌬C p is very similar in the presence and absence of Ca 2ϩ is consistent with the binding of the cTnC N-domain binding to cTnI being thermally neutral under these conditions (see Fig. 5a). This suggests that the area of nonpolar residues buried at the cTnC C-terminal domain/cTnI interface is similar in the presence of Mg 2ϩ or Ca 2ϩ .
Proton Release during cTnC⅐cTnI Binding-If the cTnC⅐cTnI binding process results in a net release or absorption of protons then an equivalent number of protons must be absorbed or released by the buffer. This process will contribute to the observed ⌬H for cTnC⅐cTnI binding and therefore has to be accounted for. To assess whether there is a significant absorption or release of protons upon cTnC⅐cTnI complex formation ⌬H was measured in experiments using buffers with enthalpies of ionization ranging from Ϫ2.35 to ϩ36.51 kJ mol Ϫ1 (cacodylate, Pipes, MOPS, and imidazole) (24). The results in Fig. 2 show that the observed ⌬H changed very little, either in the presence or absence of Ca 2ϩ , and was not dependent on the buffer, demonstrating that there was little or no net release or absorption of protons during binary complex formation.
Binding of cTnC to cTnI -Initially the experiments were performed in the presence of 0.3 M KCl to allow comparison with previous work (25,26). The result of a typical ITC experiment is illustrated in Fig. 3. A trace of the calorimetric titration of cTnI with cTnC in the absence of Ca 2ϩ is shown (Fig.  3a). The negative peaks show that the interaction is exothermic. Each deflection represents the heat released by cTnC binding to cTnI with each injection. The binding isotherm derived from these data is plotted in Fig. 3b. This graph shows the integrated heats for each cTnC injection versus the molar ratio of cTnC to cTnI. From these data the stoichiometry (n), binding constant (K a ) and enthalpy (⌬H) of binding were obtained directly, and the changes in Gibbs free energy (⌬G) and entropy (T⌬S) of binding were calculated using Equation 2.
The average K a , ⌬H, ⌬G, and T⌬S values from a minimum of five independent ITC binding experiments are given in Table I.
As expected, the K a was ϳ 6-fold higher (17.4 ϫ 10 7 M Ϫ1 ) in the presence of Ca 2ϩ than in the absence of Ca 2ϩ (2.7 ϫ 10 7 M Ϫ1 ). The binding stoichiometry was essentially 1:1 under both sets of conditions. Binding of cTnI to the cTnC N-and C-domains-Initial studies of the binding of the cTnC N-and C-domains were carried out under conditions identical to those employed for intact cTnC. The results summarized in Table I show striking similarities for the thermodynamic properties of whole cTnC and cTnC C-domain binding to cTnI. The large negative enthalpy change drives complex formation in both cases although there is a small entropic contribution in the absence of Ca 2ϩ . The affinities of cTnC and cTnC C-domain for cTnI were very similar in the presence of Ca 2ϩ but higher than those in the absence of Ca 2ϩ . This indicates that the C-terminal domain of cTnC is the major contributor to the overall binding affinity of cTnC to cTnI. We find that Ca 2ϩ -saturated cTnC C-domain binds to cTnI with an 8 fold higher affinity than the Mg 2ϩsaturated cTnC C-domain. Control binding assays with increased concentrations of Mg 2ϩ (up to 10 mM) gave binding parameters identical to assays performed under standard conditions, indicating that the cTnC C-domain was fully saturated with Mg 2ϩ in the presence of 3 mM Mg 2ϩ .
Attempts to monitor the binding of the isolated cTnC Ndomain to cTnI in the presence of Ca 2ϩ proved unsuccessful in the presence of 0.3 M KCl, even at different temperatures (data not shown). Consequently studies were undertaken to investigate the effect of ionic strength on the binding of whole cTnC, cTnC N-and C-domains to cTnI.

Effect of Ionic Strength on cTnC and Its Isolated Domains
Binding to cTnI-The effect of increasing ionic strength on the binding of cTnC to cTnI is shown in Fig. 4a. In the absence of Ca 2ϩ (Mg 2ϩ only) the affinity of cTnC for cTnI decreased as the ionic strength was increased, whereas in the presence of Ca 2ϩ the affinity increased as ionic strength was increased. Similar trends were observed when cTnI was titrated with cTnC Cdomain (Fig. 4b). In the absence of Ca 2ϩ the affinity of the cTnC C-domain for cTnI decreased dramatically as the ionic strength was increased, and in the presence of Ca 2ϩ the affinity of cTnC C-domain for cTnI increased slightly as ionic strength was increased.
cTnC N-domain bound to cTnI in both the presence and absence of Ca 2ϩ in the 0 -0.1 M KCl range. In the presence of Ca 2ϩ , binding of the cTnC N-domain to cTnI was exothermic, equimolar, decreased slightly as ionic strength was increased, and was ϳ8-fold weaker than that of the cTnC C-domain under similar conditions (Fig. 4b). In the absence of Ca 2ϩ , binding was endothermic and therefore must be entropically driven. However, conditions could not be found to increase the cTnC N-domain/cTnI stoichiometry much beyond ϳ0.5, making the data difficult to interpret. The endothermic binding of cTnC N-domain to cTnI in the absence of Ca 2ϩ also appeared to contribute to the binding of whole cTnC to cTnI, at concentrations of KCl less than 0.1 M, giving rise to a two-site binding curve that was difficult to interpret, hence these data are not shown.
Effect of Ionic Strength on Enthalpy and Entropy of Binding-The effect of varying the ionic strength on ⌬H and T⌬S of cTnC and its isolated domains binding to cTnI resulted in the linear plots shown in Fig. 5, a and b, respectively. The thermodynamic parameters of cTnC C-domain binding to cTnI closely paralleled those obtained with whole cTnC both in the presence and absence of Ca 2ϩ and were quite dissimilar to those observed for the cTnC N-domain. The ⌬H observed for cTnC N-domain binding to cTnI approached zero as the ionic strength was increased. Because ITC monitors the heat change as a result of complex formation a very small ⌬H made it difficult to measure cTnC N-domain/cTnI binding affinities at ionic strengths greater than 0.1 M. The ⌬H and T⌬S results show that changes in enthalpy were the driving force for the interaction between cTnC and cTnC C-domain with cTnI both in the presence and absence of Ca 2ϩ , as both ⌬H and T⌬S values were negative over most of the range of salt concentrations studied. The interaction between cTnC N-domain and cTnI is entropically favorable at low salt concentrations and entropy would be the only driving force at physiological ionic strength (as determined from extrapolation of Fig. 5, a and b).
Sedimentation Equilibrium Studies on cTnI-Because of the potential of cTnI to aggregate it was important to show that aggregation was not the cause of differences in cTnC⅐cTnI binding observed at different ionic strengths. Sedimentation equilibrium experiments were therefore performed on cTnI under the conditions used for ITC. A typical plot illustrating the relationship between the lnf and the square of the distance from the axis of rotation (r 2 ) is shown in Fig. 6. The linear relationship of this analysis over the entire length of the solution column demonstrates that the cTnI is monodisperse in solution. Furthermore, this linearity was observed over all of the KCl and protein concentrations used and in the presence and absence of Ca 2ϩ . The estimated molecular mass obtained for cTnI by sedimentation equilibrium, by averaging the molecular mass at all KCl conditions, was 22.6 Ϯ 1.1 kDa (S.E.). This agrees with the known molecular mass of cTnI (23,906 Da) calculated from the amino acid composition and observed in this laboratory by mass spectrometry. 2

DISCUSSION
Fluorescence and surface plasmon resonance studies have been used previously to establish the affinity of cTnC for cTnI, e.g. (25)(26)(27)(28). Although these methods have provided much valuable information, fluorescence techniques require modification of the proteins with reporter groups, and surface plasmon resonance produces orientation constraints because of protein immobilization. We have used ITC, a technique for directly measuring K a and ⌬H values in biological systems, to examine the binding of cTnC to cTnI under a variety of ionic conditions. No protein modification was required, and the experiments were performed with the proteins free in solution. Furthermore, we have studied the binding of the recombinant N-and C-terminal domains of cTnC to cTnI. Previous studies, using a variety of techniques, have shown that the isolated domains of both skeletal and cardiac TnC have structures and properties very similar to those in the intact proteins (16,29,30).
Because it is known that TnI has a tendency toward selfassociation, it was important to ascertain whether it was monomeric or forming aggregates in solution, under the conditions used in the ITC experiments. From the sedimentation equilibrium studies the linearity of the graphic analysis was consistent with sample homogeneity. No indication of dimerization or higher order aggregation was ever observed, even in the absence of KCl, in the presence or absence of Ca 2ϩ , with protein concentrations higher than those used for ITC. Furthermore, previous studies from our laboratory (31) have demonstrated that both cTnC and the cTnC⅐cTnI complex show no tendency toward aggregation at and above the protein concentrations used in these ITC experiments. These data indicate that the protein solutions used for the ITC experiments were all monodisperse and that effects of ionic strength on binding were not caused by protein aggregation. The fact that all observed binding ratios of cTnC (or its isolated domains) with cTnI were essentially 1:1 under all conditions studied is consistent with this finding.
The results show that the affinity of cTnC for cTnI is strongly dependent on ionic strength. All previously reported measurements of the affinity of cTnC for cTnI were obtained at concentrations of 0.3 M KCl or greater (25)(26)(27)(28). At these high ionic strengths the affinity of cTnC for cTnI was significantly higher in the presence of Ca 2ϩ than in the presence of Mg 2ϩ , as we have found here. At physiological ionic strength, however, we found no increase in the affinity of cTnC for cTnI upon addition of Ca 2ϩ . These data suggest that around physiological ionic strength the cTnC/cTnI interaction in the presence and ab-TABLE I Binding parameters for complexation between cTnI and cTnC or cTnC C-domain The binding parameters were obtained from data similar to those shown in Fig. 3. The parameters are given as the mean Ϯ S.E. obtained from a minimum of five independent determinations. Assay conditions were: 20 mM MOPS, pH 7.0, 0.3 M KCl, 3 mM MgCl 2 , 0.5 mM EGTA (ϪCa 2ϩ ), with or without 1 mM CaCl 2 (ϩCa 2ϩ ). 6 M (ϪCa 2ϩ ) or 3 M (ϩCa 2ϩ ) cTnI was titrated with 5 l of 160 M (ϪCa 2ϩ ) or 60 M (ϩCa 2ϩ ) cTnC or cTnC C-domain.  sence of Ca 2ϩ is finely balanced, although Ca 2ϩ clearly changes the mode of interaction as demonstrated by changes in ⌬H and T⌬S. It should be noted, however, that we have of necessity investigated only the isolated binary interaction between cTnC and cTnI in these studies. It is possible that the other thin filament protein components could influence the effect of ionic strength on the Ca 2ϩ sensitivity of the cTnC/cTnI interaction.
We find that the effect of Ca 2ϩ and ionic strength on the affinity and thermodynamic parameters of the cTnC C-domain binding to cTnI closely parallel those found with whole cTnC. These data suggest that the C-terminal domain of cTnC dominates the steady-state interaction with cTnI in both the presence and absence of Ca 2ϩ . These results contrast sharply with the binding of cTnC N-domain to cTnI over the limited ionic strength range where we were able to obtain data. Even in the presence of Ca 2ϩ the affinity of cTnI for the cTnC N-domain was significantly lower than that measured for the cTnC Cdomain. It is interesting to note that where data for the binding of cTnI to the isolated domains of cTnC can be compared with whole cTnC, considerable constraints occur in the latter interaction. For example, at 0.1 M KCl in the presence of Ca 2ϩ , if we consider that cTnI binds simultaneously to both domains of TnC and that the overall affinity is the product of the individual K a values for both domains, then one would expect to find a K a for cTnC⅐cTnI in the 10 14 M Ϫ1 range (Fig. 5b). Clearly the interactions of the two domains of whole cTnC with cTnI are not independent and influence one another.
We observed that the binding of cTnC and cTnC C-domain to cTnI were dramatically affected by the addition of Ca 2ϩ over a range of ionic strengths. In the absence of Ca 2ϩ , the reduction in K a as the ionic strength was increased, both for the Cterminal domain of cTnC and whole cTnC, may suggest that electrostatic interactions dominate the binding when Mg 2ϩ occupies sites III and IV. On the other hand, when Ca 2ϩ is bound to these sites, increasing the ionic strength increased the affinity, especially between the intact proteins, a characteristic of interactions in which hydrophobic forces dominate. This agrees with NMR (16) and crystallographic data (15) that show extensive Van der Waals interactions between TnC and TnI in the presence of Ca 2ϩ . It also suggests a difference in the structure of the C-terminal domain of cTnC when it is occupied by Mg 2ϩ or Ca 2ϩ . It may be that Ca 2ϩ occupancy of sites III and IV "opens" up the paired EF hands exposing hydrophobic sites as seen in the crystal structures of skTnC (2) and skTnC bound to the N-terminal region of skTnI (15), whereas Mg 2ϩ is not able to do this fully. There is presently no high resolution structure for TnC with Mg 2ϩ bound in sites III and IV, which would be particularly valuable to compare with the Ca 2ϩ -bound state. Several previous studies are consistent with our findings. Different fluorescence changes were found within the C-terminal domain of skTnC depending upon whether Mg 2ϩ or Ca 2ϩ were bound (32,33). Analysis of the binding of Mg 2ϩ and Ca 2ϩ to cTnC by microcalorimetry suggests that Mg 2ϩ and Ca 2ϩ induce different conformations of the C-terminal domain of cTnC (34). 3 Microcalorimetry has also been used to show that conformational changes occur in the C-terminal domain of skTnC when Ca 2ϩ replaces Mg 2ϩ in sites III and IV, although these differ somewhat from those observed in the cardiac isoform (35).
Our data demonstrate, for the first time, that there are structural differences within the C-terminal domain of cardiac TnC, depending on whether Ca 2ϩ or Mg 2ϩ is bound, which affect the characteristics of the cTnI/cTnC interaction. Experiments with skeletal isoforms also suggest Ca 2ϩ sensitivity in the binding of skTnC C-terminal domain to skTnI. An increase in the affinity of a skTnC tryptic fragment, corresponding to the C-terminal domain of skTnC, for skTnI in the presence of Ca 2ϩ has been noted previously (36). Additionally, Pearlstone and Smillie (37) found that a higher ionic strength was required to elute the C-terminal domain of skTnC from a skTnI peptide affinity column in buffer containing Ca 2ϩ rather than Mg 2ϩ . The difference in cTnC⅐cTnI affinity would be important if Ca 2ϩ can exchange for Mg 2ϩ at sites III and IV in working cardiac muscle. It has not been determined if the C-terminal sites of TnC are bound with Mg 2ϩ at all times or if the Ca 2ϩ signal is of sufficient duration to displace Mg 2ϩ in the contractile state. Although modeling studies (38) suggest that Mg 2ϩ dissociation from sites III and IV is likely to be too slow to be completed in a single contraction-relaxation cycle, they also predict that Ca 2ϩ occupancy of the C-terminal domain of cTnC is a measure of the intensity and frequency of muscle contraction and could therefore increase with increased heart rate. This would lead to a change in the structure of cTnC and enhanced affinity for cTnI, allowing the C-terminal domain of cTnC to play a modulatory role in the regulation of the contractile cycle. Our results give no indication as to whether the change in affinity of cTnC C-domain for cTnI, in the presence and absence of Ca 2ϩ , reflects a different mode of binding to residues 33-80 of cTnI or whether a different region of cTnI displaces the cTnI helix at positions 33-80 in response to Ca 2ϩ as in the model proposed by Tripet (39).
The thermodynamic results obtained by ITC give some interesting insights into the cTnC/cTnI interaction. Previously the ⌬G difference between Ca 2ϩ -and Mg 2ϩ -bound states of cTnC binding to cTnI, at 0.4 M KCl, was attributed solely to Ca 2ϩ binding to "regulatory" site II in the N-terminal domain of cTnC (40). This change in ⌬G was thought to be responsible for Ca 2ϩ activation in cardiac muscle. Our data suggest that at high ionic strengths most of the Ca 2ϩ -dependent increase in ⌬G of cTnC binding to cTnI arises from Ca 2ϩ binding to sites III and IV. Our results demonstrate that any differences between the steady-state binding of the Mg 2ϩ and Ca 2ϩ forms of cTnC to cTnI cannot solely be attributed to Ca 2ϩ binding to the Nterminal domain of cTnC.
Parallels can be drawn between our cTnC⅐cTnI work and published ITC data on calmodulin-target peptide recognition (41,42). Quantitatively similar ⌬H, T⌬S, and ⌬C p values and negligible proton release are seen in both cases. The negative ⌬C p values indicate internalization of hydrophobic surfaces upon complex formation, a process that is entropically favorable. Wintrode and Privalov (41) considered that burial of hydrophobic surfaces could not be the major driving force in the enthalpically driven binding of calmodulin to target peptide. However, it cannot be excluded that the entropy associated with the burial of hydrophobic surfaces is important in offsetting unfavorable entropy changes arising from noncovalent bond formation and increased order in the cTnC⅐cTnI complex. Both domains of calmodulin appear to contribute similarly to the overall ⌬H of target peptide binding (41), whereas tissuespecific adaptations of the cardiac homolog, cTnC, to muscle regulation result in very different contributions of the cTnC N-terminal and C-terminal domains to the overall ⌬H.
This study has revealed the importance of ionic strength on the cTnC/cTnI interaction, which must be considered when interpreting data. Large changes in the ⌬H of cTnC⅐cTnI binding, but little change in affinity at physiological ionic strength, could imply that the mode but not the strength of the cTnC/ cTnI interaction changes in response to Ca 2ϩ . We have also demonstrated that the interaction between cTnC C-domain and cTnI is Ca 2ϩ -sensitive and that cTnC C-domain is responsible for most of the energy difference observed between cTnC⅐cTnI binding in the presence and absence of Ca 2ϩ . Further studies are required to clarify the importance of the cTnC C-domain in regulation. It will be interesting to compare the results obtained for cTnC and its isolated domains binding to cTnI with data obtained using the skeletal isoforms, where significant differences in sequence and binding occur.