Biological Function and Site II Ca2+-induced Opening of the Regulatory Domain of Skeletal Troponin C Are Impaired by Invariant Site I or II Glu Mutations*

To investigate the roles of site I and II invariant Glu residues 41 and 77 in the functional properties and calcium-induced structural opening of skeletal muscle troponin C (TnC) regulatory domain, we have replaced them by Ala in intact F29W TnC and in wild-type and F29W N domains (TnC residues 1–90). Reconstitution of intact E41A/F29W and E77A/F29W mutants into TnC-depleted muscle skinned fibers showed that Ca2+-induced tension is greatly reduced compared with the F29W control. Circular dichroism measurements of wild-type N domain as a function of pCa (= −log[Ca2+]) demonstrated that ∼90% of the total change in molar ellipticity at 222 nm ([θ]222 nm) could be assigned to site II Ca2+ binding. With E41A, E77A, and cardiac TnC N domains this [θ]222 nm change attributable to site II was reduced to ≤40% of that seen with wild type, consistent with their structures remaining closed in +Ca2+. Furthermore, the Ca2+-induced changes in fluorescence, near UV CD, and UV difference spectra observed with intact F29W are largely abolished with E41A/F29W and E77A/F29W TnCs. Taken together, the data indicate that the major structural change in N domain, including the closed to open transition, is triggered by site II Ca2+ binding, an interpretation relevant to the energetics of the skeletal muscle TnC and cardiac TnC systems.

The troponin complex, present in skeletal and cardiac muscle in association with tropomyosin and actin of the thin filaments, is primarily responsible for the Ca 2ϩ regulation of contraction and relaxation in these tissues. Each of the three components of this complex, troponin C, I, and T (TnC, TnI, and TnT, respectively) 1 has distinct functional properties. Conformational changes in TnC associated with Ca 2ϩ binding to its regulatory N domain lead to a strengthening of its interaction with TnI and a weakening of the latter's interaction with actin. Inhibition of the actomyosin ATPase activity is thereby released. TnT fulfills the role of anchoring the complex to tropomyosin and potentiating actomyosin ATPase activity (for reviews, see Refs. [1][2][3][4][5][6][7][8]. As the molecule responsible for triggering this complex series of events, TnC has been the subject of intensive investigation over many years. X-ray crystallographic (9 -14) and NMR solution studies (15)(16)(17)(18)(19)(20)(21) have demonstrated the presence of two domains, N and C, joined in the crystals by a solvent-exposed ␣-helix, which is partially disrupted and flexible in solution (16,18). Each of the N and C domains consists of two intimately associated EF-hand or helix-loop-helix metal binding motifs. Those of C domain (sites III and IV) have high affinity for Ca 2ϩ (K d Х 10 Ϫ7 M) and bind Mg 2ϩ competitively (K d Х 10 Ϫ3 M). Sites I and II are of lower affinity with K d values of ϳ16 and ϳ1.7 M, respectively (22,23), and are specific for Ca 2ϩ at physiological Mg 2ϩ concentrations. Helices are designated A-H, corresponding to the four helix-loop-helix motifs of sites I-IV. In addition, an extension of the NH 2 terminus forms the N helix; the central helix seen in the crystals connecting site II to site III is designated D/E. Based on a substantial body of evidence, the C domain sites III/IV are believed to be occupied by Ca 2ϩ /Mg 2ϩ throughout the contraction/relaxation cycle and to serve a structural role in anchoring TnC to the other troponin components. Association and dissociation of Ca 2ϩ from sites I/II of N domain are considered to fulfill the regulatory function of TnC.
As originally postulated by Herzberg et al. (24) for avian skeletal TnC N domain, Ca 2ϩ binding to sites I/II has been shown to promote a structural transition from a closed (apo) to open (2Ca 2ϩ ) conformation (13,14,17,20). Involving significant changes in interhelical angles, helices B and C are seen to have altered their positions relative to those of N, A, and D. As a result, a number of previously fully and partially buried apolar residues are exposed at the protein surface to create a hydrophobic pocket or patch. The intact TnC (4Ca 2ϩ ) now possesses two nonpolar surface patches, one in each domain, with surrounding constellations of negatively charged residues. Evidence for these as interaction sites with regions of TnI has been reported (see Ref. 25 and references therein).
In each loop of sites I-IV of TnC, Ca 2ϩ complexes with seven ligands in a pentagonal bipyramidal arrangement. By convention, the six amino acids contributing to these liganding groups are designated X, Y, Z, ϪY, ϪX, and ϪZ or as residues 1, 3, 5, 7, 9, and 12 in the 12-residue loop. Depending on the position within the loop and the amino acid occupying a particular position, Ca 2ϩ ligation may be to a protein main or side chain oxygen or to H 2 O. In most EF-hands/loops, the highly conserved bidentate Glu residue at position 12 (ϪZ) provides two Ca 2ϩ ligands through its two side chain carboxyl oxygens.
Binding of 2Ca 2ϩ to sTnC N domain has been shown to occur in a stepwise manner with estimated K D values in the range of 0.8 -3.0 and 5-23 M (22). Structurally, site II is more preformed for liganding to Ca 2ϩ and, based on this and other considerations, is believed to have the higher affinity of the two (13,23,26). Ca 2ϩ titrations monitored by NMR spectral changes revealed that chemical shifts occur throughout the N domain upon binding of each Ca 2ϩ (22,23). An important structural feature involved in the tight coupling is a short antiparallel ␤-sheet comprised of a segment of each of the loops of sites I and II (for structural details, see Ref. 13). The Ca 2ϩinduced transition from closed to open structures can be understood in terms of two hinge or pivot regions located in the loop regions of the two sites. Opening and closing of the structure would thus involve the reorientation of helices B/C at these pivot points relative to that of the N/A/D unit whose helical dispositions one to the other are virtually unchanged.
Cardiac TnC, while highly homologous, differs in several important respects from its skeletal muscle counterpart. Amino acid sequence differences are largely confined to the first 41 residues corresponding to the N helix and site I EF-hand. Site I is deficient in Ca 2ϩ binding due to an amino acid insertion and the substitution of Ca 2ϩ liganding residues in its binding loop (27). Calcium binding to site II is relatively unaffected, however, with a dissociation constant similar to that of site II of the skeletal protein (23, 28 -30). Recently, the apo and Ca 2ϩsaturated NMR solution structures of human cardiac intact and N domain TnCs have been reported (18,19). Surprisingly, Ca 2ϩ binding is not accompanied by an "opening" of the regulatory domain and concomitant exposure of a hydrophobic surface area although Ca 2ϩ binding to site II leads to spectral shifts throughout the entire N domain sequence (23). In related NMR studies, Gagné et al. (26) have shown that with a mutated form of skeletal TnC in which the invariant Glu at position 41 of site I is replaced by Ala, the "closed" to "open" transition is impeded. An earlier study (31) had shown that mutation of the equivalent Glu in rabbit sTnC to either Asp or Gln led to a reduction in Ca 2ϩ binding and functional activity in a reconstituted skinned fiber assay. Gagné et al. (20,26) have suggested that in the wild-type protein initial Ca 2ϩ binding to preformed site II would involve only minor structural changes but set the stage for Ca 2ϩ binding to site I. The latter would involve large conformational changes including opening of the structure. The invariant Glu 41 at position 12 of site I was postulated to be the key residue in facilitating this structural transition.
In order to provide further insight into the molecular mechanism of this Ca 2ϩ -sensitive regulatory switch, we have replaced the invariant Glu residue 41 or 77 by Ala in intact sTnC F29W and in the recombinant N domains sNTnC and sNTnC F29W . The effects of mutating Glu 77 on the properties of sTnC have not to our knowledge been previously investigated. Both Glu to Ala mutations in intact TnC largely abolished Ca 2ϩ -induced tension development in muscle skinned fibers. Of the total Ca 2ϩ -induced ellipticity change (⌬[] 222 nm ) in isolated wild-type N domain, ϳ90% could be attributed to site II Ca 2ϩ binding. With sNTnC E41A , sNTnC E77A , and cNTnC, this [] 222 nm change associated with site II was reduced to Յ40% of that seen in sNTnC, a result consistent with impairment of the closed to open structural transition (26). In keeping with this interpretation, the Ca 2ϩ -induced changes in fluorescence, near UV CD, and difference spectra observed with sTnC F29W are largely abolished with sTnC F29W/E41A and sTnC F29W/E77A . Our interpretation is that sites I/II in sTnC are highly interdependent, that the closed to open structural change is associated with Ca 2ϩ binding to site II, and that the integrity of both sites I/II invariant Glu residues is essential for this important conformational transition. These deductions have important implications for interpretations of the intramolecular mechanisms for the opening of the skeletal structure and for energetic considerations of the skeletal and cardiac systems.

MATERIALS AND METHODS
Construction and Expression of TnC Mutants-To produce sTnC F29W/ E77A, the double mutant of intact TnC, site-directed mutagenesis of M13mp19CII.Fx.sTnC F29W was performed by the oligonucleotide-directed (32) double priming method (33) as described previously for sTnC F29W (34), using the 17-mer E77A oligonucleotide 5Ј-CTTCGAG-GCGTTCCTGG-3Ј (with base change underlined) and the 19-mer oligonucleotide 5Ј-AGGGCATCAAATTAAACCA-3Ј corresponding to part of the CII untranslated region. The purified BstXI/SacI fragment of M13mp19CII.Fx.sTnC F29W/E77A was cloned into the phosphatasetreated BstXI/SacI fragment of expression vector pLCII.Fx.sTnC F29W (34). Ligation, transformation into rubidium-competent Escherichia coli K12CI, and final plasmid sequencing have all been described previously (35). To produce sTnC E41A , the EcoRI fragment encompassing the complete coding region of sTnC was isolated from pLCII.Fx.sTnC and subcloned into phagemid vector pTZ18R (36,37) purchased from Amersham Pharmacia Biotech. Using the Kunkel method (38,39), uracil-containing single-stranded DNA was obtained by two rounds of transfection of the phagemid into competent E. coli CJ236 (ung Ϫ dut Ϫ ) and subsequent infection by helper phage M13KO7. A 17-mer oligonucleotide 5Ј-CACCAAGGCGTTGGGCA-3Ј (with base change underlined) was used to introduce the single mutation. After mutagenesis, ligation mix was transformed into competent DH5␣, and transformants were selected on ampicillin plates. Eight colonies were screened by the dot blotting method and the DNA from positive colonies transformed into competent JM109. Single-stranded DNA was isolated and sequenced by the chain termination method (40) to confirm the base change. Since suitable sites were not available for digestion by restriction enzymes, the intact double mutant sTnC F29W/E41A was subsequently produced by carrying out a second round of mutagenesis using the DNA from sTnC E41A as the template and the 17-mer oligonucleotide reported previously (34) for sTnC F29W . The EcoRI fragment with the entire coding region of sTnC F29W/E41A TnC was cloned into the phosphatased EcoRI fragment of protein expression vector pLCII.Fx.sTnC. The ligation mix was transformed into competent E. coli QY13 as described previously (34). Expression of recombinant chicken fusion TnC, cleavage with factor Xa, and column purification procedures have been described earlier (35).
Construct pET3a.sNTnC E77A was produced by the overlap extension method of Horton et al. (41) as follows: 1) PCR product I (269 base pairs) was made using pET3a.sNTnC as template, a 5Ј TnC outside 30-mer primer 5Ј-GAGATATACATATGGCGTCAATGACGGACC-3Ј with an NdeI site (underlined) juxtaposed to Ala 1 (in boldface lettering) and a 17-mer E77A reverse primer 5Ј-CCAGGAACGCCTCGAAG-3Ј (with base change underlined); 2) PCR product II (452 base pairs) was made using pET3a wild-type intact TnC as template, the 17-mer E77A forward primer used previously for pLCII.Fx.E77A intact TnC, and a 3Ј TnC outside reverse primer 5Ј-GGAATGTCTGGATCCTTACTGCA-CACCCTC-3Ј with a BamHI site (underlined) and stop codon (in boldface lettering) next to Gln 162 ; 3) PCR product III (E77A intact TnC, 518 base pairs) was subsequently made using PCR products I and II as template and the two 30-mer TnC 5Ј and 3Ј outside primers used previously; 4) PCR product IV (E77A/ND) was made by a subsequent round of PCR using PCR product III (E77A intact TnC) as template, the 5Ј TnC 30-mer primer with NdeI site used previously, and 5Ј-AATAT-GGATCCTAGGCGTCCTCTTTCACT-3Ј, a 30-mer 3Ј-reverse primer with stop codon (in boldface type) after Ala 90 , followed by a BamHI site (underlined). PCR product IV was then cloned into pET3a via the engineered Ndel and BamHI sites. Construct pET3a.sNTnC E41A was made using the Stratagene mutagenesis protocol with pET3a.sNTnC as template, the 17-mer forward primer used previously for intact sTnC E41A , and its complementary 3Ј-reverse 17-mer primer.
pET3a.cNTnC was constructed as follows: 1) the DNA region incorporating intact human slow skeletal TnC (equivalent in amino acid sequence to intact cardiac TnC) was transferred from clone TC1 (42) into the pET vector using PCR technology with TC1 as template, and the two primers 37-mer 5Ј-CCCGCGCCCCGCCATATGGATGACATC-TACAAGGCT-3Ј (with NdeI site underlined) and 30-mer 5Ј-GCTCTG-GATCCAGGTCAGCATCTACTCCAC-3Ј (with BamHI site underlined); 2) pET3a.cTnC (produced in step 1) was used as template in the next round of PCR to produce N domain construct pET3a.cNTnC by including in the PCR mixture the same 37-mer 5Ј-primer (with NdeI site) used initially and a 36-mer primer 5Ј-CCTCCTCAGGGATCCCCTAGCT-GTCGTCCTTCATGC-3Ј with a stop codon (in boldface type) after Ser 89 , followed by a BamHI site (underlined). Using the NdeI/BamHI sites, this second PRC product was cloned into pET3a.
Tension Measurements-Chemically skinned fibers were prepared from rabbit psoas muscle and stored in the presence of glycerol at Ϫ20°C (44,45). Segments of single fibers were dissected and mounted for isometric tension measurements as described by Metzger et al. (46) in chambers of 0.5-1.0 ml equipped with temperature control and rapid stirring. Before and after reconstitution with different proteins, maximum Ca 2ϩ -activated tensions were measured at pCa (Ϫlog[Ca 2ϩ ]) 4.4 under standard conditions (pH 7.0, 15°C; free Mg 2ϩ , 2 mM; MgATP, 4 mM) using a solution of 152 mM potassium propionate, 10 mM imidazole, 4.4 mM K 2 Na 2 ATP, 6 mM magnesium acetate, and 5 mM CaEGTA, pH 7.0. In the fibers of Tables I and II, initial maximum tensions (P o ) were 1.7 Ϯ 0.2 kg/cm 2 (ϮS.E., n ϭ 7). In some experiments, CaCl 2 was added in excess over EGTA to obtain higher free [Ca 2ϩ ] (to pCa 3.7). Relaxing solution contained 5 mM K 2 EGTA in place of CaEGTA. Each Ca 2ϩ activation was preceded by a brief exposure to relaxing solution containing 0.1 mM EGTA.
For extraction and reconstitution, endogenous TnC was extracted at 15°C using 10 mM EDTA, 10 mM imidazole, and 0.2 mM trifluoperazine at pH 7.2. Tests for completeness of extraction were performed at pCa 4.4 after washing to remove trifluoperazine (45). After prolonged extraction (15-30 min), tensions were reduced to 0 -2% of the original P o . These small residual tensions were subtracted from those obtained following reconstitution with the mutants.
As a control for reconstitution with mutants sTnC F29W/E41A and sTnC F29W/E77A , chicken skeletal muscle TnC was used. Tensions generated when fibers were first reconstituted with this protein ranged from 60 to 80% of P o . In a previous report, reconstitution with mutant sTnC F29W was shown to be equal to that obtained using recombinant sTnC protein, within experimental error (47). Control experiments have established that the same is true for chicken muscle TnC (Table I) and that 1 M is saturating for this protein (data not shown). In this report, fibers were reconstituted for 15 min in relaxing solution containing 0.1 mM dithiothreitol and either 1.1 M chicken muscle TnC or (following extraction of the chicken protein) 22 M mutant recombinant TnC. For the competition experiments, fibers were exposed for 15 min to a mixture of 22 M mutant recombinant and 1.1 M chicken muscle protein after exposure to the mutant alone. At the end, fibers were extracted and reconstituted again with the control protein. On average, the tension resulting from the second reconstitution with control was 80 Ϯ 3% of the first, and we were able to repeat the procedure with the second mutant. Fibers were discarded when the control reconstitution fell below 50% of the P o . Protein concentrations were determined by the method of Hartree (48).
Spectral Measurements-Protocols for preparation of the final 50 mM MOPS, 100 mM KCl, 1 mM EGTA, 1 mM dithiothreitol, pH 7.1, buffer and dialysis of lyophilized protein for spectral analyses have been described previously (35). For far UV CD and fluorescence intensity measurements, 2-4 mg/ml stock solutions of proteins were diluted using 0.22-m filtered outer dialysate. For near UV CD and UV difference absorption measurements, stock protein solutions were used directly. Protein concentrations were determined by amino acid analysis (based on Ala and Leu content) and given in the figure legends. Descriptions of the spectral instruments, conditions for Ca 2ϩ titrations, and calculation of free [Ca 2ϩ ] have all been described earlier (34,35). Tables I and II summarize the ability of the three intact recombinant F29W TnCs to function in a TnC-depleted and reconstituted muscle fiber system. Using the maximal Ca 2ϩ -activated tension obtained with endogenous TnC in native rabbit psoas fibers as 100% P o , the pCa tension curves produced by replacement with sTnC F29W or chicken muscle TnC in extracted fibers were similar, with pCa1 ⁄2 values of 5.66 and 5.57, respectively, compared with 5.78 in native fibers (Table I). Cooperativity was also preserved, with Hill coefficients of ϳ2 (Table I). With each extraction, the percentage of P o was reduced to 0.5-8.0% within 10 -20 min, enabling several reconstitution experiments to be carried out on one set of fibers. In fibers reconstituted with sTnC F29W , maximum tensions were 93% of those obtained with chicken muscle protein (Table I). Although neither of the two Glu 3 Ala mutants was very effective as a replacement for native TnC (see reconstitution a in Table II), the site I mutant was better (19%) than the site II mutant (5%). In some experiments, tension recovery for the mutants was tested with higher concentrations of Ca 2ϩ (up to pCa 3.7), with similar results.

Tension Measurements-
In order to ascertain whether or not the mutants were able to bind to the thin filament, competition experiments were performed. Following exposure to the mutant, fibers were incubated with a 20:1 mixture of the mutant and chicken muscle proteins. In most cases, some increase in tension occurred to 25% of the control for sTnC F29W/E41A and to 9% of the control for sTnC F29W/E77A ; see reconstitution b in Table II), with further increments if the incubation was repeated (data not shown). However, much higher tensions were recorded when the fibers were subsequently stripped of bound TnC and incubated with control protein for the same length of time in the absence of mutant (90% of the control for sTnC F29W/E41A and 89% of the control for sTnC F29W/E77A ; see reconstitution c in Table II).
Far UV CD Studies-To investigate the effects of the two Glu 3 Ala mutations on the ellipticities of the intact proteins, CD analyses at 222 nm in the presence and absence of Ca 2ϩ were carried out on sTnC, sTnC F29W , sTnC F29W/E41A , and sTnC F29W/ E77A TnCs. sTnC F29W was included to serve as a control for the two F29W Glu 3 Ala mutants, since the latter were also studied by fluorescence measurements (see below). The data, expressed as the percentage change in [] 222 nm relative to sTnC taken as 100% are given in Table III. While the F29W mutation appears to have increased the 222-nm ellipticity, both Glu 3 Ala mutations showed highly significant decreases of ϳ20%. Previous studies in this laboratory (49) have shown that of the total Ca 2ϩ -induced ellipticity in intact TnC, ϳ27% could be assigned to N domain and ϳ73% to C domain. Rather similar large contributions by C domain were reported by Johnson and Potter (50). To eliminate the large background of  (18,19), we have also prepared and examined the Ca 2ϩ -induced ellipticity change of cNTnC (residues 1-89). The data as given in Table IV demonstrate that, in comparison with sNTnC, sNTnC F29W shows an increase in negative ellipticity, whereas the change for each of the Glu 3 Ala mutants and for cNTnC is very significantly reduced. Since both sNTnC E41A and cNTnC fail to undergo the Ca 2ϩ -induced structural opening (18,19,26), the data indicate this to be true also for sNTnC E77A . These observations are consistent with the view that a significant proportion of the change in [] 222 nm elicited by Ca 2ϩ binding to sites I and II can be attributed to the reorientation of helices B and C relative to the three helices of the NAD structural unit, an interpretation previously proposed by Gagné et al. (15). Calcium titrations of the changes in [] 222 nm for sNTnC, sNTnC F29W , sNTnC E41A , sNTnC E77A , and cNTnC are shown in Fig. 1, a and b, expressed as a percentage of the total change observed with sNTnC. The curves for all five of these proteins are seen to be biphasic, corresponding to stepwise Ca 2ϩ binding to sites II/I or I/II. Derivation of Ϫlog K D values as measures of Ca 2ϩ affinity for the two sites were obtained by curve fitting procedures as described previously (35) and are presented in Table V. For sNTnC, sNTnC F29W , sNTnC E41A , and cNTnC, the Ϫlog K D value for tighter binding has been assigned to site II in accordance with previous evidence (13,22,23). In the case of NTnC E77A , the higher Ϫlog K D value (5.9) has been assigned to site I and the lower Ϫlog K D value (ϳ3) to site II. This latter assignment is based on the demonstrated dramatic effects of mutations of invariant Glu residues in positions 12 of other EF-hand Ca 2ϩ binding loops (51)(52)(53)(54)(55). In contrast to the two Glu 3 Ala mutants the effects of the F29W mutation are to alter to a modest degree the Ca 2ϩ affinity of sites I and II. This is evident from a comparison of the far UV CD data for sNTnC F29W and sNTnC in Fig. 1 and Table V as determined in the present study. The biphasic nature of the curve for cNTnC deserves comment. While site I of cTnC is often described as nonfunctional in its ability to bind Ca 2ϩ , undoubtedly true from a physiological perspective, the data of Fig. 1b indicate that at high Ca 2ϩ concentrations (in the millimolar range; pCa4 to pCa2), there is a structural change (increasing negative ellipticity) associated with a Ϫlog K D value for Ca 2ϩ binding of ϳ3.0. These observations are in good agreement with the quality fluorescence Ca 2ϩ titration data reported by Johnson et al. (30) for cTnC derivatized with 2-(4Ј-acetamidoanilino)naphthalene-6-sulfonic acid (see cTnC IA in Table V). They also observed a biphasic curve with Ϫlog K D values of 5.8 and 2.7 (see Table   V) corresponding to site II and site I Ca 2ϩ binding, respectively.
A striking feature of these data in the case of sNTnC is that ϳ90% of the Ca 2ϩ -induced total ellipticity change is associated with Ca 2ϩ binding to site II ( Fig. 1 and Table IV), similar to the 80% previously noted by Li et al. (22). For sNTnC E41A , sNTnC E77A , and cNTnC, it is this site II Ca 2ϩ -induced ellipticity change that is markedly reduced to 30 -40% of that observed with sNTnC (see Fig. 1 and Table IV). Thus, the effects of the invariant Glu mutations in both sites I and II as well as of the defunct site I in cNTnC are remarkably similar, observations consistent with failure of the closed to open structural opening in all three proteins. In contrast, the effects of the F29W mutation in sNTnC F29W are to increase the magnitude of the Ca 2ϩ -induced ellipticity change and, as noted above, to alter the Ca 2ϩ affinity to a modest degree.
Tryptophan Fluorescence Studies-Previous studies have demonstrated that the F29W mutation in intact sTnC is a useful fluorescence probe for monitoring Ca 2ϩ -induced structural transitions of its N domain (34) and that these were not influenced by the presence of C domain as in intact sTnC as compared with isolated sNTnC F29W (49). In the present study, we have examined the properties of TnC variants in which E41A or E77A mutation has been introduced into intact F29W. The fluorescence emission spectra and quantum yields for sTnC F29W and the two mutants sTnC F29W/E41A and sTnC F29W/ E77A in both the absence and presence of Ca 2ϩ are shown in Fig.  2 and Table VI, respectively. In the apo state, the emission spectra of sTnC F29W/E77A and the control sTnC F29W were similar (both with quantum yields of 0.12), while that of sTnC F29W/ E41A was ϳ1.5-fold greater (quantum yield ϭ 0.18). In response to Ca 2ϩ , both F29W/Glu 3 Ala mutants showed a slight quenching, with quantum yields decreasing by 0.01-0.02, in contrast to the 3-fold increase with sTnC F29W from 0.12 (apo) to 0.33 (ϩCa 2ϩ ). These results indicate that for both Glu 3 Ala mutants, the environment of Trp 29 is now altered in the Ca 2ϩsaturated state.
Although the Ca 2ϩ -induced quenching of fluorescence for both sTnC F29W/E41A and sTnC F29W/E77A was small in comparison with the large increase for sTnC F29W , it was possible to monitor the fluorescence change as a function of increasing Ca 2ϩ concentration. From the monophasic curves shown in Fig.  3, Ϫlog K D values of 4.71 Ϯ 0.08 and 3.51 Ϯ 0.09 were deduced for sNTnC F29W/E41A and sNTnC F29W/E77A , respectively. Because of the uncertainties arising from the influence of the F29W mutation on Ca 2ϩ affinities of sites I/II, we have not attempted to assign these to one site or the other.
Near UV CD and UV Difference Absorption Spectroscopic Studies-The near UV CD spectra for the apo and Ca 2ϩ -loaded states for intact sTnC F29W , sTnC F29W/E41A , and sTnC F29W/E77A are shown in Fig. 4. In the apo state, the control sTnC F29W has positive CD bands in the region of 270 -300 nm. With the addition of Ca 2ϩ , the amplitude of the bands is much reduced to negative values over the entire wavelength range. In contrast, both of the double mutants sTnC F29W/E41A and sTnC F29W/E77A in the apo state have slightly higher positive ellipticity values than sTnC F29W , and these are almost unchanged upon the addition of Ca 2ϩ .
The UV difference absorption spectra of Ca 2ϩ -saturated versus apo forms of intact sTnC F29W , sTnC F29W/E41A , and sTnC F29W/E77A are shown in Fig. 5. As described previously (34), the region of the difference spectrum of sTnC F29W above 275 nm can be considered as arising from a red shift in the Trp absorption bands; below this wavelength, the difference spectrum is largely dominated by the contribution of Phe residues. In the case of sTnC F29W/E77A , the contribution of both the Trp and Phe residues are absent, resulting in almost no difference  (5) Reconstitutions were performed sequentially (a to c) following an initial extraction and reconstitution with control protein (chicken skeletal muscle TnC, 1.1 M). Extractions preceded reconstitutions a and c (see "Materials and Methods"). Recovery was normalized to the average of the two tensions (% P) recorded for each fiber with control protein, before and after the mutant (22 M). The range of the S.E. is shown, followed by the number of experiments in parentheses. spectrum. Likewise with sTnC F29W/E41A , while Trp still contributes to the difference spectra, that region ascribable to Phe has been virtually eliminated. Thus taken together, the far UV CD, fluorescence, near UV CD, and UV difference spectral data are consistent with the absence of a Ca 2ϩ -induced structural transition from closed to open states when either of the invariant Glu residues 41 or 77 is mutated to Ala. DISCUSSION The present study has revealed several important new insights into the molecular mechanisms by which the binding and dissociation of Ca 2ϩ to and from sites I and II of TnC elicits the Ca 2ϩ -regulated events of contraction and relaxation of striated muscle. These include 1) the demonstration that mutation to Ala of either of the invariant Glu residues in position 12 of the site I and II Ca 2ϩ binding loops leads to virtual elimination of the Ca 2ϩ sensitivity of the contractile response in skinned muscle fibers; 2) the observation that by far the major fraction (ϳ90%) of the total Ca 2ϩ -induced ellipticity change of isolated N domain of wild-type sTnC is associated with binding to site II; 3) that this latter change is substantially reduced to 30 -40% in each of the two Glu 3 Ala mutants and in cTnC; 4) the Ca 2ϩ -induced fluorescence, near UV CD, and UV difference spectral changes observed with sTnC F29W are largely eliminated in both Glu 3 Ala mutants of sTnC F29W ; 5) in comparison with sTnC, sTnC F29W has minimal effect on biological function as assessed in skinned fibers, but in solution its isolated N domain shows an increase in both Ca 2ϩ -induced ellipticity change and altered affinity for Ca 2ϩ at sites I and II. In the following, we discuss the significance of these observations in terms of our current detailed structural knowledge of the apo and 2Ca 2ϩ states of sNTnC as well as in terms of suggestions for the intramolecular mechanism by which Ca 2ϩ binding is coupled to the structural transition from a closed to open conformation.
Recent NMR studies have demonstrated that the transition from closed to open structures in sNTnC E41A and in cNTnC does not occur in excess Ca 2ϩ (18,19,21,26). The present study shows that with both of these proteins (Table IV), there is significant reduction of the Ca 2ϩ -induced change in [] 222 nm in comparison with that seen in wild-type sNTnC. The structural basis for the Ca 2ϩ -induced ellipticity change in wild-type skeletal N domain has been discussed by Gagné et al. (15). Since secondary structural changes associated with Ca 2ϩ binding are minimal and confined to a short region at the NH 2 -terminal end of the B helix (residues 39 -41) (13,15,17) it was concluded that the ellipticity changes must arise from tertiary structural alterations. Based on various considerations as reviewed by Manning (56), the major contributing factor is likely to be the movement of helices B and C relative to those of A and D from a roughly antiparallel arrangement to approximately perpendicular. Other contributions, although probably less significant, could include changes in the environment of clustered Phe residues, Ca 2ϩ -induced aggregation, and the removal of   Fig. 1, a and b, the percentage of total [] 222 nm for each protein attributable to site II is given.  Tables III  and IV and Fig. 1). In addition, when the Glu 3 Ala mutations are incorporated into intact F29W, the Ca 2ϩ -induced changes in fluorescence, near UV CD spectra, and UV difference spectra seen with the sTnC F29W control are virtually eliminated (see Figs. 2, 4, and 5). These similarities lead to the important conclusion that, like E41A, the E77A mutation leads to disrup-FIG. 2. Fluorescence emission spectra of intact TnC mutants with or without Ca 2؉ . Excitation was at 282 nm. The data for the three mutants were normalized using a Trp quantum yield of 0.13 as described (34). Both excitation and emission slits were set at 5-nm bandwidth. Buffer conditions were the same as in Fig. 1 The calculated fitted curves (indicated by solid lines) were obtained as described previously (35). For purposes of comparison, all three proteins were normalized to their respective 100% change in fluorescence intensities. Buffer conditions were the same as in Fig. 1.  c In the present work data were derived from Ca 2ϩ titrations of far UV CD analyses. Data sets from 2 to 6 titrations were analyzed separately using a curve-fitting program as described previously (35). tion of the closed to open Ca 2ϩ -induced transition. Clearly the data demonstrate that the integrity of both of the invariant Glu residues in sites I and II is essential for this important conformational transition.
Evidence bearing on the question of cooperativity of Ca 2ϩ binding to sites I and II of sTnC has been contradictory. In direct Ca 2ϩ binding measurements using equilibrium dialysis (57) or a Ca 2ϩ ion-selective electrode (58), no cooperativity between sites I and II was observed. On the other hand, steep curves of fluorescence and [] 222 nm as a function of pCa attrib-utable to the titration of sites I/II were observed for intact and isolated N domains of sTnC F29W and sTnC (34,35). These corresponded to high Hill coefficients (n Х 2), indicating a high degree of cooperativity. Subsequently, however, Ca 2ϩ titration of the NMR spectral changes in 15 N-labeled sNTnC revealed that Ca 2ϩ binding to sites II and I was a stepwise process with little or no cooperativity of Ca 2ϩ binding indicated (22). It was also demonstrated that when the large changes in [] 222 nm attributable to C domain are eliminated by working with isolated N domain, a biphasic curve was observed corresponding to the filling of binding loops II and I (22), an observation confirmed in the present work. The suggestion that the steepness of the ellipticity and fluorescence versus pCa curves now attributable to site II Ca 2ϩ binding is not a reliable index of cooperativity is reinforced by the present titration data for cNTnC. The steepness of its [] 222 nm versus pCa curve over the pCa range 7.0 -4.0 (see Fig. 1b; calculated Hill coefficient, n ϭ 1.7) is inconsistent with the known Ca 2ϩ binding properties of this protein. We have also extracted and analyzed the data of Fig. 1 of Johnson et al. (30) in which the change in fluorescence of cTnC IA was monitored as a function of pCa. The steepness of the first phase of this curve (attributed to site II binding) over the same pCa range corresponds to a Hill coefficient of n ϭ 2. Although we do not presently understand the molecular basis of these excessively steep curves, it is clear that they cannot be considered as reliable measures of cooperative Ca 2ϩ binding to sites II/I of skeletal and cardiac TnCs.
These and other considerations (13,23) lead to the conclusion that Ca 2ϩ binding to sNTnC occurs stepwise to sites II and I, respectively, and that the higher affinity of site II versus site I is explicable in terms of several differences in their structural features. These include in site II versus site I a larger number of Ca 2ϩ -coordinating residues arranged in a more preformed geometry, higher net negative charge, more stabilizing hydrogen bonds, and a lesser degree of conformational flexibility (see Ref. 13). Pertinent to the present investigation are the side chain dispositions of the two invariant bidentate glutamic acid residues at positions 12 of the two sites. While Glu 77 of site II differs little in its position in both apo and Ca 2ϩ states, Glu 41 of site I in the apo state is directed completely out of the binding site and forms a salt bridge with Lys 40 (see Ref. 13 and references therein). Only site I has this latter residue at position 11 of the loop.
A further important deduction from the present observations is that the major change in [] 222 nm and therefore in the conformational transition is closely coupled to the binding of Ca 2ϩ to site II. Curve fitting analyses of the biphasic pCa versus [] 222 nm data for sNTnC of Fig. 1 provided Ϫlog K D values of 5.92 Ϯ 0 (ϳ90% of total [] 222 nm change) and 4.54 Ϯ 0.10. For reasons described above, these measures of Ca 2ϩ affinity have been assigned to sites II and I, respectively, and are seen to be in good agreement with other previous estimates and the present work (see Table V for a compilation of these). It is the very significant [] 222 nm change associated with site II Ca 2ϩ binding that is much reduced in both the sNTnC E41A and sNTnC E77A mutants.
Comparison of the percentage of [] 222 nm versus pCa curves for sNTnC E41A and sNTnC of Fig. 1 indicates that the Glu to Ala mutation at residue 41 reduces site I Ca 2ϩ affinity from Ϫlog K Х 4.5 to ϳ3. The same mutation has a more modest to minimal effect on site II Ca 2ϩ affinity with reported values of ϳ4.8 (NMR; Ref. 23) and ϳ5.9 (present study). A similar comparison of sTnC E77A and sNTnC from the data of Fig. 1 provided site I and site II Ϫlog K D values of ϳ5.9 and ϳ3, respectively. Thus, while Ca 2ϩ affinity to site II was dramatically reduced (ϳ1000 fold), that to site I was apparently modestly FIG. 4. Near UV CD spectra of the three intact F29W mutants in the presence and absence of Ca 2؉ . Spectra were run in ϩCa 2ϩ conditions as described previously (35). Buffer conditions were the same as in Fig. 1 increased (ϳ10 fold). These large reductions in site I and site II Ca 2ϩ affinities by mutation to Ala of the invariant Glu residues in each of these loops are consistent with the reported dramatic effects of mutation of invariant Glu to Gln, Lys, or Ala in the Ca 2ϩ binding loops of calmodulin and calbindin (51)(52)(53)(54)(55). The increase in site I Ca 2ϩ affinity by the mutation at the site II invariant Glu as in E77A was, however, surprising. At our present level of understanding of the interplay of the multiple factors affecting Ca 2ϩ affinity to EF-hands and without a high resolution structural analysis of main and side chain dispositions for E77A, we presently have no explanation for this apparently anomalous observation.
Previous studies (22, 34, 47, 49, 59, 60 -62) have shown that the F29W mutation (immediately adjacent to the 1 (or X) position of the site I EF-hand) in intact sTnC serves as a useful fluorescence probe for monitoring Ca 2ϩ binding and associated structural changes of N domain. Comparison of Ca 2ϩ -induced changes in [] 222 nm of sTnC F29W and sTnC proteins detected no significant differences in properties of the two, although, as pointed out (34), "it would be unrealistic not to anticipate some local disruption of the immediate environment when Phe is replaced by Trp." Subsequently, comparison (22) of the far UV CD properties of isolated N domain bearing the F29W mutation with that of sNTnC showed that in the apo state both far UV CD spectra were the same; however, the increase in negative ellipticity at 222 nm upon saturation with Ca 2ϩ was significantly greater for sNTnC F29W . As well, a shift in the pCa versus [] 221 nm curve to lower [Ca 2ϩ ] was noted. The data thus indicated some perturbation of the Ca 2ϩ binding properties of the N domain. In the present investigation, these observations have been largely confirmed and extended. Of particular note is the biphasic character of the pCa versus [] 222 nm relationship for sNTnC F29W , a feature very similar to that of sNTnC (see Fig. 1) and not previously detected, presumably because of the small contribution to total [] 222 nm by the second phase. Curve fitting analyses of the biphasic curve provided Ϫlog K D values of ϳ6.2 (site II) and 3.7 (site I). Comparison of these with the corresponding values obtained in the present work for sNTnC (see Table V) indicates that Ca 2ϩ affinity to site I has been decreased, while that to site II has been modestly increased by the F29W mutation. Pertinent to the present observations and the effects of the F29W mutation on the properties of TnC N domain is the recent demonstration of decreased stability of the mutated protein as assessed by NMR spectral shifts elicited by both high pressures and elevated temperatures (61,62).
The data of Fig. 1 for sNTnC F29W as with sNTnC demonstrate that by far the largest percentage of Ca 2ϩ -induced change in [] 222 nm is elicited by site II binding. Since the Ca 2ϩ -induced fluorescence change in intact sTnC F29W and sNTnC F29W is monophasic (34,49) with a Ϫlog K D value corresponding closely to that of the change in [] 222 nm attributed to site II Ca 2ϩ binding, we may conclude that the ϳ3-fold increase in fluorescence seen in sTnC F29W and sNTnC F29W is closely coupled to the closed to open structural transition triggered by site II Ca 2ϩ binding. This is true although the F29W mutation at the COOH-terminal end of the A helix is immediately adjacent to the site I loop.
A further feature of the sNTnC F29W data is the significantly larger Ca 2ϩ -induced [] 222 nm change (see Fig. 1) when compared with sNTnC (129 versus 100%). If, as we propose, a significant proportion of this [] 222 nm change in sNTnC is attributable to the reorientation of the B/C unit relative to that of helices N, A, and D, this larger [] 222 nm change in sNTnC F29W may arise from a more extensive opening of the N domain structure upon site II Ca 2ϩ binding. While this suggestion is speculative, there is precedence for variability in the degree of opening based on both NMR and x-ray crystallographic studies (reviewed in Ref. 20; see also Houdusse et al. (14)).
Based on a comparison of the x-ray diffraction structures of the apo and 2Ca 2ϩ states of N domain, Strynadka et al. (13) have detailed their structural differences and proposed plausible intramolecular events ensuing from Ca 2ϩ binding. Initial Ca 2ϩ liganding at the more preformed site II would lead to a compaction of its loop in which six or seven residues in the COOH-terminal half of the loop are pivoted relative to those of the NH 2 -terminal half. The pivot or hinge point appears to be in the region of Ile 73  This series of events ensuing from site II calcium binding is thus seen to involve a complex and interdependent rearrangement of intramolecular structural elements. The degree to which the mutation of an individual residue in this chain would affect the closed to open structural transition would depend on the importance of that residue in the series of events leading ultimately to the transition. Both of the invariant Glu residues in position 12 of the two loops would be expected to play critical roles in this, as shown in the present study. In a temporal sense, Glu 77 can be considered as playing a key initial role in the compaction of loop II. Its replacement by Ala would largely eliminate Ca 2ϩ binding and prevent this compaction. This in turn would negate the transmission of the signal through the ␤-strand region and subsequent structural rearrangements in site I. The closed to open structural transition would be blocked and functionality would be impaired as observed in the present investigation.
As demonstrated by Gagné et al. (26) and in the present 2 Gagné et al. (26) on the other hand have concluded, based on their NMR structure, that Val 65 , positioned at the C-terminal end of helix C and immediately adjunct to loop II, is the pivot point associated with site II. Houdusse et al. (14), based on their structural analyses of the rabbit crystals, have identified small and larger hinge regions for each of EF-hands I and II The small regions involve residues at the Cterminal ends of helices A and C. The larger in the case of EF-hand I would include residues Val 39 , Lys 40 , and Glu 41 (chicken residue numbering) at the NH 2 terminus of helix B as well as ␤ strand residues Gly 35 , Ile 37 , and Ser 38 . The larger hinge in the EF-hand II is largely centered at Ile 73 occupying position 8 in loop II and adjacent to the NH 2 terminus of helix D. study, the substitution of Glu 41 by Ala also impairs the closed to open structural transition as well as functionality (31) as assessed by tension measurements in the skinned muscle fiber system. This is so although the data presented in Fig. 1 indicate that for the wild-type sNTnC, calcium binding to site II is sufficient to elicit the closed to open structural transition. This could presumably only occur with the assumption of normal helical conformation of residues 37-41 at the NH 2 -terminal end of helix B, the latter's redirection, and the repositioning of the Glu 41 side chain carboxyl from its ionic interaction with Lys 40 to an orientation more compatible with calcium ligation. Subsequent calcium binding to site I, requiring higher calcium concentration, would be accompanied by more minor structural alterations, consistent with the small change in [] 222 nm (see Fig. 1), including compaction of loop I (13) and associated changes throughout the N domain as detected by NMR (22). Since Glu 41 is thus seen as an important element in this series of events leading to the opening of the structure, the effects of its replacement by Ala on this transition are understandable.
The effects of mutating other liganding residues in sites I and II of sTnC on its Ca 2ϩ binding and functional properties have been previously reported (31,(63)(64)(65)(66)(67)(68). These have been largely concerned with loop position 1 and/or 3 and include in loop I the mutations D30V and D30A (position 1) and D32G and D32A (position 3). Loop II mutations have included D66V, D66E, D66N, and D66A (position 1) and D68G, D68E, D68A, and D68N (position 3). Ca 2ϩ binding presumably in the mutated loop in all of these was seriously compromised (with the exception of D68N), and functional activity was impaired to varying degrees as assessed by calcium regulation of actomyosin ATPase or skinned fiber tension measurements. It is presently not known whether the closed to open structural transition is impeded in these mutations.
Our proposal in the present study that the major structural transition from closed to open forms of sNTnC is associated with site II Ca 2ϩ binding clearly has relevance to considerations of the energetics of the skeletal and cardiac systems. The free energy changes (G 0 ) associated with site II Ca 2ϩ binding to the sNTnC and cNTnC proteins as calculated from their dissociation constants have been estimated to be Ϫ8.0 Ϯ 0.4 and Ϫ7.7 Ϯ 0.02 kcal mol Ϫ1 , respectively (20,23). While this small difference could conceivably account in part for the failure of the cardiac protein to undergo opening, a more important consideration is almost certainly the structural features of the cNTnC in comparison with those of the skeletal isoform. The most important of these are likely to be those in the site I region and include in the cardiac protein the Val 28 insertion and the substitution of Leu 29 and Ala 31 for the two Asp residues at positions 1 and 3 of the skeletal isoform. Recent NMR relaxation measurements of sNTnC and cNTnC (69,70) have provided important insights into this question. These have demonstrated that while sites II of the two isoforms in their apo states indicate little difference in their conformational entropies, site I of cNTnC shows a decrease of 0.9 Ϯ 0.3 kcal mol Ϫ1 in comparison with skeletal site I, indicating overall increased rigidity. Estimation of main chain order parameters (S 2 ) and conformational entropies on a per residue basis indicate that residue positions 1-3 of site I in cTnC contribute inordinately to this entropy difference. This increased rigidity is explicable in terms of the structural interactions and environments of Val 28 and Leu 29 . Val 28 , having an S 2 value of 0.83 and indicating little flexibility, is described as being largely buried and having several hydrophobic contacts with Ala 31 (position 3) and positions 6 -8 of site I. Hydrophobic contacts are also evident between Leu 29 and Ile 36 /Ser 37 (␤-sheet residues of site I). These stabilizing structural features would be expected to create an energy barrier to the conformational changes involving the straightening of helix B and the closed to open structural transition. Additional energy input provided by TnI peptide/ protein binding would be required to force this transition (see below). From this perspective, the failure of cNTnC to undergo full opening upon site II Ca 2ϩ binding, as we propose for sNTnC, can be understood not in terms of its inability to bind Ca 2ϩ at site I but rather in terms of the hydrophobic nature of the amino acid alterations that lead to this loss of Ca 2ϩ binding. Pertinent to these considerations are the experiments carried out by Putkey, Sweeney, and collaborators a decade ago (66 -68) in which Ca 2ϩ binding to site I of cTnC was activated by deletion of Val 28 and substitution of residues 29 -32 (positions 1-4) of site I by their skeletal TnC counterparts. This construct was fully active as a replacement for cTnC in slow skeletal muscle fiber tension measurements and as active as wild-type cTnC in a fast skeletal muscle system. It would be of great interest to establish whether these site I substitutions are sufficient to promote site II Ca 2ϩ -induced opening of the N domain structure.
As indicated above, NMR structural studies have shown that the Ca 2ϩ -induced structural opening observed with sNTnC does not occur with the E41A mutant (26) or the cardiac protein (18,19,21). Most recently, however, opening of the structure in the presence of Ca 2ϩ has been observed when either of these is complexed with peptides corresponding to one of the two inhibitory region(s) of the corresponding TnI (i.e. skeletal TnI residues 115-131 and cardiac residues 147-163) (71,72). With both the skeletal and cardiac TnC N domain, these TnI peptides have been shown to interact in the region of their hydrophobic pockets (Refs. 25 and 73 and references therein). Based on the recent data of Dong et al. (74) working with the cardiac TnC-TnI complex, it seems likely that as with the TnI peptides, the respective intact TnIs would also induce the opening of the cardiac and E41A mutant proteins in the presence of Ca 2ϩ . However, in the case of the E41A mutant, and in contrast to cTnC, the present data (Table II) show that this is insufficient to reverse the effects of the mutation on functional activity. In the case of mutant sTnC E77A , it is presently unclear whether complexation with inhibitory peptide or intact TnI can bring about opening of the structure. In any case, functional activity as measured by tension measurements in skinned fibers is essentially eliminated with this mutant.
Finally, we wish to express the view that although we consider the evidence presented in this paper to be fully consistent with our interpretation that site II Ca 2ϩ -binding is responsible for triggering the closed to open structural transition of sNTnC, it is possible that other experimental approaches in the future may show otherwise. In particular, it should in principle be feasible to address this issue by the application of appropriate NMR or other spectral measurements at different levels of N domain Ca 2ϩ saturation.