Parallel Measurement of Ca2+ Binding and Fluorescence Emission upon Ca2+ Titration of Recombinant Skeletal Muscle Troponin C

Calcium binding to chicken recombinant skeletal muscle TnC (TnC) and its mutants containing tryptophan (F29W), 5-hydroxytryptophan (F29HW), or 7-azatryptophan (F29ZW) at position 29 was measured by flow dialysis and by fluorescence. Comparative analysis of the results allowed us to determine the influence of each amino acid on the calcium binding properties of the N-terminal regulatory domain of the protein. Compared with TnC, the Ca2+ affinity of N-terminal sites was: 1) increased 6-fold in F29W, 2) increased 3-fold in F29ZW, and 3) decreased slightly in F29HW. The Ca2+titration of F29ZW monitored by fluorescence displayed a bimodal curve related to sequential Ca2+ binding to the two N-terminal Ca2+ binding sites. Single and double mutants of TnC, F29W, F29HW, and F29ZW were constructed by replacing aspartate by alanine at position 30 (site I) or 66 (site II) or both. Ca2+ binding data showed that the Asp → Ala mutation at position 30 impairs calcium binding to site I only, whereas the Asp → Ala mutation at position 66 impairs calcium binding to both sites I and II. Furthermore, the Asp → Ala mutation at position 30 eliminates the differences in Ca2+ affinity observed for replacement of Phe at position 29 by Trp, 5-hydroxytryptophan, or 7-azatryptophan. We conclude that position 29 influences the affinity of site I and that Ca2+ binding to site I is dependent on the previous binding of metal to site II.

Structural modifications resulting from this association are thought to be transmitted to other components of the thin filament, which ultimately allows the effective association of myosin to actin and force generation (for review, see Refs. [1][2][3][4]. TnC has a dumbbell-shaped structure with the globular Nterminal and C-terminal domains separated by a nine-turn ␣-helix (5,6). Calcium binds with high affinity (K d Ϸ 10 Ϫ7 M) to the two EF-hand Ca 2ϩ /Mg 2ϩ binding sites located at the TnC C-terminal domain (sites III and IV, the C-terminal sites) and with low affinity (K d Ϸ10 Ϫ5 M) to the two EF-hand Ca 2ϩspecific binding sites located in the TnC N-terminal domain (sites I and II, the N-terminal sites) (7). Because of the affinity and kinetic properties of the C-terminal sites they are believed to be always occupied by Ca 2ϩ or Mg 2ϩ in vivo and are involved in the stable TnC binding to TnI (8). Ca 2ϩ binding to the two TnC N-terminal sites results in a greatly increased affinity for the TnI C-terminal and inhibitory regions, resulting in their dissociation from actin and disinhibition of the actomyosin interaction.
The affinity and cooperativity of calcium binding to TnC have been studied extensively. These parameters are determined through direct measurement of free calcium concentration in association systems (direct techniques) (7, 9 -11) or through spectroscopically monitored titrations (indirect techniques) (12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31). These studies largely agree on the value of the binding constants but disagree with regard to the cooperativity parameter. Cooperativity between N-terminal sites has been detected only by indirect techniques, raising the possibility that potentially damaging conditions, the extensive time needed to prepare samples, or the intrinsically high errors involved in direct techniques disturb or conceal the N-terminal site cooperativity. In one of the studies using indirect techniques (25) we constructed a spectral probe mutant of recombinant skeletal chicken TnC, named TnCF29W (or simply F29W), in which Phe-29 was replaced by Trp. This position is immediately adjacent to Ca 2ϩ binding site I (residues 30 -41), and the substitution was expected to cause only minor alterations in both apo and Ca 2ϩ -saturated forms of the wild type N-terminal domain structure. Upon Ca 2ϩ binding to F29W N-terminal sites there is a 3-fold increase in the intensity of * This work was supported in part by the Fundaçã o de Amparo à Pesquisa do Estado de Sã o Paulo and the Howard Hughes Medical Institute. 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 1 The abbreviations used are: TnC, recombinant chicken troponin C; TnI, troponin I; DTT, dithiothreitol; MOPS, 4-morpholinepropanesulfonic acid; 5-HW, 5-hydroxytryptophan; 7-ZW, 7-azatryptophan; F29, Phe-29, used to indicate broad spectrum of Phe-29 mutations; F29W, F29HW, and F29ZW, recombinant TnCs with phenylalanine at position 29 replaced by tryptophan, 5-hydroxytryptophan, or 7-azatryptophan, respectively; F29X (where X ϭ W, HW, or ZW), any single mutant with phenylalanine at position 29 replaced by tryptophan, 5-hydroxytryptophan, or 7-azatryptophan; D/A, any protein with phenylalanine at position 29 and at least one replacement of aspartate by alanine; F29X D/A, any double or triple mutant with tryptophan or an analog at position 29 and at least one replacement of aspartate by alanine; Ndomain, N-terminal domain sites I and II of TnC or its mutants; Cdomain, C-terminal domain sites III and IV of TnC or its mutants; pCa 1/2 , value of -log[Ca 2ϩ ] free corresponding to 50% of the maximal associated transition. fluorescence emission at 340 nm. This mutant binds Ca 2ϩ with a Hill coefficient of 1.96, reflecting a high degree of cooperativity between sites I and II. High values of cooperativity for N-terminal sites were also detected by far-UV CD measurements of wild type chicken recombinant TnC (23,26). Later reports, however, indicated that the properties of the F29W N-terminal domain are different from those of TnC (29,32,33), and NMR studies using the isolated N-terminal domain of TnC indicated that sites I and II are sequentially occupied (with a nearly 10-fold affinity difference) with no cooperativity between them (32).
To resolve this discrepancy and to understand the influence of position 29 of chicken TnC on the properties of N-terminal sites, we introduced the amino acid analogs 5-hydroxytryptophan (5-HW) and 7-azatryptophan (7-ZW) into position 29 of TnC (35)(36)(37)(38)(39)(40). Ca 2ϩ binding by these new mutants (F29HW and F29ZW, respectively), as well as F29W and TnC was analyzed using a direct flow dialysis calcium binding assay and indirectly by fluorescence. We have also constructed fluorescent and nonfluorescent mutants with Asp 3 Ala replacements at position X of sites I or II or both sites to assert the dependence of N-domain calcium binding properties on each site. These mutants were submitted to flow dialysis and/or fluorescence assays. Our data indicate that: 1) tryptophan at position 29 increases the N-terminal site calcium affinity nearly 6-fold compared with TnC; 2) 5-HW at position 29 lowers the Nterminal site calcium affinity to levels slightly below those of TnC; 3) 7-ZW at position 29 increases the N-terminal site affinities to an intermediate level, between the F29W and TnC N-terminal site affinities; 4) the flow dialysis data obtained with Asp 3 Ala mutants indicate that when Asp at the first position (X) of site II is replaced by Ala, site I does not bind Ca 2ϩ . Conversely, calcium binding to site II is maintained by Asp 3 Ala mutation at position X of site I. 5) The fluorescence data indicate lower values of cooperativity between F29W Nterminal sites than reported previously and a still lower Nterminal site cooperativity value for F29HW. 6) 7-ZW probe at position 29 senses the binding of calcium to the higher affinity N-terminal site (site II) by increasing fluorescence and to the lower (in the case of F29ZW) affinity N-terminal site (site I) by decreasing the fluorescence in a bimodal fluorescence change. This observation indicates stepwise calcium binding to the regulatory N-terminal sites.
In conclusion, the data obtained with isolated TnC and its mutants support a model for stepwise Ca 2ϩ binding to Nterminal sites, suggesting that site I does not bind Ca 2ϩ when site II is unoccupied. Our results also indicate that mutations in position 29 alter the properties of the regulatory sites.

MATERIALS AND METHODS
Construction of TnC Mutants-The D30A and D66A (8) and F29W mutants (25) were used to obtain the mutants D30A/D66A, F29W/ D30A, F29W/D66A, and F29W/D30A/D66A using the overlap extension PCR method (41). The primers to generate the Asp 3 Ala mutation at position 30 on the F29W gene were 5Ј-C ACC ACC GTC CGC AGC CCA CAT GTC AAA-3Ј and 5Ј-TTT GAC ATG TGG GCT GCG GAC GGT GGT G-3Ј. The Asp 3 Ala mutation at position 66 was generated with the primers 5Ј-C GCT GCC ATC CTC GGC CAC CTC CTC GAT GA-3Ј and 5Ј-TC ATC GAG GAG GTG GCC GAG GAT GGC AGC G-3Ј. The PCR products were purified from low melting agarose gels and used in the overlap extension reactions. These final products were digested with BamHI and NdeI and inserted to pET3a (42). Mutations were confirmed by sequencing.
Production of TnC and TnC Mutants-The expression and purification of TnC and the mutant F29W have been described (10,25,43). The same procedure was used to express and purify D/A and F29 D/A mutants. To incorporate 5-HW or 7-ZW into TnC mutants a derivative of W3110TrpA33 (44) was constructed by infecting this strain with phage DE3 (42) and transforming with plasmid pLysS (45). The new strain, named W3110TrpA33(DE3)pLysS, carries the auxotrophic marker for tryptophan from the mother strain, the T7 RNA polymerase gene from phage DE3 under control of the isopropyl-1-thio-␤-D-galactopyranoside-inducible promoter lacUV5, and the lysozyme gene from plasmid pLysS. By transforming this strain with plasmid pET-F29W or pET-F29W D/A we could strictly regulate the production of large amounts of the desired mRNA. To incorporate the Trp analogs we grew transformed W3110TrpA33(DE3)pLysS cells in M9 minimal medium containing 200 g/ml ampicillin, 20 g/ml chloramphenicol, 5 mg/liter thiamin, and 20 g/ml L-tryptophan. When the culture reached an A 600 nm Ϸ1.0, cells were centrifuged at 5,000 rpm for 10 min at 4°C and resuspended in M9 minimal medium plus 200 g/ml ampicillin, 20 g/ml chloramphenicol, 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside, and 100 g/ml 5-HW or 7-ZW instead of L-tryptophan, followed by 3 h of additional incubation at 37°C with vigorous shaking (37). The protein purification process was the same of that of TnC (10), yielding 10 -20 mg of protein/liter of culture.
Estimate of Tryptophan Analog Incorporation into F29HW and F29ZW-Trp analog incorporation was estimated based on a comparative analysis of the absorbance and emission spectra of the proteins. The absorbance spectra of proteins (18 -24 M) and free amino acids (100 M) were obtained in 5 M guanidine HCl. The F29W, F29HW, and F29ZW absorbance spectra were normalized by subtracting the TnC absorbance spectrum to discount the contribution of the phenylalanines. The percentage of 5-HW incorporation into mutant F29HW (H%) was estimated using the equation where S A N 5HW, S A N SF29W, and S A N F29HW are the areas between 269 and 324 nm of the normalized absorbance spectra of 5-HW, F29W, and F29HW, respectively. The 7-ZW incorporation into F29ZW is more straightforward because the emission spectra is red shifted with a maximum near 380 nm. We subtracted the two normalized F29ZW emission spectra obtained by excitation at 287 or 315 nm, and the difference (seen as a small peak near 340 nm) was transformed back to absolute values. This area was compared with that of a sample with a known F29W concentration and assumed to represent the amount of F29W contaminating the F29ZW sample.
Ca 2ϩ Binding to TnC and TnC Mutants Determined by Flow Dialysis-Flow dialysis experiments were performed as described previously (11) with some modifications. The buffers used in dialysis to remove bound Ca 2ϩ from proteins were: buffer A, which contained 20 mM MOPS, 100 mM KCl, 1 mM DTT, 5 mM EDTA, pH 7.0; 2) buffer A with 2 mM EDTA; 3) buffer A but with no EDTA. The proteins were dialyzed with each buffer for 12 h with two changes. We have found that using Spectra/Por dialysis tubing in all steps gives the most reproducible results. Protein concentration was determined as described previously (46) using as standard a TnC sample whose concentration was determined through amino acid analysis (Met and Val content). Protein concentration in the dialysis chamber was 13 M and was adjusted with the same buffer used for the flow dialysis experiment. The Marquardt-Levenberg nonlinear regression algorithm of Jandel Scientific Sig-maPlot 2.0 package was used for curve fitting. The number of calcium sites of each protein was inferred from the saturation plateau obtained with pCa values between 4.5 and 4.0. This avoided the introduction of another parameter in the curve fitting associated with nonspecific calcium binding at lower pCa values. The data from TnC or F29X (where X ϭ W, HW, or ZW) were fitted using Equation 2.
The data from D30A and F29X/D30A were fitted using Equation 3. [ The data from D66A, D30A/D66A, F29X/D66A, and F29X/D30A/D66A were fitted using Equation 4, where K d1 , n 1 , K d2 , and n 2 are the apparent dissociation constant and Hill coefficients, respectively, for the C-terminal sites and the N-terminal sites.
Fluorescence Measurements of Calcium Binding to F29X and F29X/ D30A-The fluorescence experiments were performed in a Hitachi F-4500 spectrofluorometer with 2.5 nm excitation slit and 5 nm emission slit for F29W and 5 nm for excitation and emission slits for all other mutants. The excitation wavelength used was 287 nm for F29W and F29W/D30A and 315 nm for F29X and F29X/D30A. The temperature of the cell was kept at 25°C. We used the calcium-free protein batch used in the flow dialysis experiments. The volume of the protein solution in the fluorescence assays was 2.5 ml, and calcium was added to maintain its concentration the same as in the flow dialysis experiments. In this way, we could use the mean pCa values obtained in the flow dialysis experiments to correlate to the data obtained in the fluorescence experiments. Curve fitting for F29W, F29HW, and F29ZW/D30A (negative variation for F29ZW/D30A) was performed adjusting the experimental data to Equation 5.
Curve fitting for F29W/D30A and F29HW/D30A was performed adjusting the experimental data to Equation 6, where F is the relative fluorescence change, K d2 the apparent dissociation constant for calcium binding to the N-terminal sites (or the microscopic dissociation constant to the remaining N site for the F29/ D30A mutants) and n 2 the Hill coefficient reflecting the cooperativity between the two N sites. Curve fitting for F29ZW was performed adjusting the experimental data to Equation 7 where F is the relative fluorescence change, f 21 is the relative fluorescence change associated with the filling of the site II, K d21 is the microscopic dissociation constant for site II, f 12 is the relative fluorescence change associated with the filling of the site I, and K d12 is the microscopic dissociation constant for site I. Fig. 1A shows the normalized absorbance spectra of the three isolated amino acids, Trp, 5-HW, and 7-ZW. Both 5-HW and 7-ZW have red shifted absorbance compared with tryptophan. This characteristic is maintained in the absorbance spectra of the corresponding recombinant TnC mutants, F29W, F29HW, and F29ZW (Fig. 1B). Fig. 1, C-E, demonstrates the similarity between the free amino acid and the corresponding protein absorbance spectra.

Incorporation of Tryptophan Analogs into F29HW and F29ZW Mutants-
The normalized excitation and emission spectra of the three amino acids and the three mutants are shown in Fig. 2. The fluorescence of F29W increases nearly three times upon addition of Ca 2ϩ (pCa 2.0), as reported previously (25). When excited at 315 nm, the fluorescence of F29HW increases nearly seven times upon addition of Ca 2ϩ (pCa 2.0) (this increase drops to nearly three times when the excitation wavelength is 287 nm). The fluorescence of F29ZW displays a bimodal response to Ca 2ϩ , increasing nearly 40% from pCa 9.0 to pCa 5.0 and decreasing nearly 20% from pCa 5.0 to pCa 3.0.
Quantitative analysis of the absorbance and emission spectra of the recombinant TnC indicates 87-96% analog incorporation efficiency for F29HW and nearly 96% for F29ZW (see "Materials and Methods"). Because of this incorporation ratio the results obtained with both analog mutants are not significantly distorted by contamination with F29W. We used the same procedure in producing all of the other F29HW D/A and F29ZW D/A mutants.
Calcium Binding to TnC and Its Mutants- Fig. 3A shows the calcium binding curves to TnC and to the three fluorescent mutants, F29W, F29HW, and F29ZW as determined by flow dialysis. Calcium binding curves indicate that all four proteins bind four Ca 2ϩ /molecule and that near saturation is achieved between pCa 4.5 and 4.0.
At pCa values less than 4, there is an increase in the calcium binding ratio which is most possibly related to nonspecific calcium binding to TnC. Thus, pCa 4 was assumed to be the maximum Ca 2ϩ concentration to which we could relate any dependent parameter. The TnC and F29HW Ca 2ϩ binding curves are similar regarding the overall transition. Between pCa 5.5 and pCa 4.0, the most evident difference among the binding curves is the displacement of the F29W and F29ZW curves toward higher pCa values compared with TnC and F29HW. This reflects an increase in the affinity of the low affinity N-terminal sites. The C-terminal sites have affinities nearly 100 times higher than the N-terminal sites and are represented by the lower part of curves (between pCa 7 and 5.5). Fig. 3B compares the relative fluorescence changes of F29W, F29HW, and F29ZW. The data were obtained under the same conditions as the flow dialysis experiments, that is, in absence of EGTA. The fluorescence experiments showed that the halftransition point of F29HW occurs 0.7 pCa unit after the F29W transition. This confirms the lower calcium affinity of F29HW N-terminal sites with respect to the F29W N-terminal sites. The F29ZW fluorescence change is bimodal, with the maximum value at ϷpCa 5 and a final decrease of nearly 20%, which plateaus at pCa 3.5.
In Fig. 3, C-E, we compare flow dialysis and fluorescence data for each mutant to correlate the fluorescence changes with calcium saturation level and to determine the point in the free calcium titration which corresponds to the fluorescence change.
In the case of F29W and F29HW (Fig. 3, C and D), the greater part of the fluorescence change occurs above the pCa where there are two bound calcium ions/molecule. This limit is better defined for F29HW than for F29W. The difference in affinity between F29HW C-terminal and N-terminal sites is nearly 0.7 pCa unit higher than between C-terminal and N-terminal sites of F29W. These Ca 2ϩ -dependent fluorescence changes provide direct evidence that calcium binding to the C-terminal sites does not alter the fluorescence of the tryptophan or 5-HW incorporated at position 29 of TnC and that these probes selectively report calcium binding to the regulatory N-terminal sites of TnC. Fig. 3E shows fluorescence of F29ZW during titration with Ca 2ϩ . The maximal fluorescence value occurs at pCa 5.0 and is associated in the calcium binding curve with three bound Ca 2ϩ / molecule. Therefore, in the case of this mutant, the probe can be used to distinguish between the binding of the third and the fourth Ca 2ϩ ions. This pattern is also found in the calcium binding curves of the fluorescent D/A mutants (Fig. 4, B-D). The difference is that the presence of the fluorescent probes affects the pCa at which the curves for the non-D/A and the corresponding D30A mutant diverge (pCa Ϸ 6 for F29W and F29W/D30A, pCa Ϸ 4.6 for F29HW and F29HW/D30A, and pCa Ϸ 5.4 for F29ZW and F29ZW/D30A). The corresponding fluorescent and nonfluorescent D66A and D30A/D66A calcium binding curves largely superimpose over the entire pCa scale (Fig. 4). Fig. 5 shows the fluorescence change upon calcium binding to the fluorescent D/A mutants, which bind only one or zero Ca 2ϩ ions in their N-terminal domains. Fig. 5A compares the fluorescence change of each mutant upon calcium binding (relative to the apo form ϭ 100%). For all D30A mutants, the Ca 2ϩinduced fluorescence increase is much less than that observed in the original fluorescent mutants that possess two functional N-terminal Ca 2ϩ binding sites. In the case of F29ZW/D30A the fluorescence intensity decreases slightly upon Ca 2ϩ binding. For all D66A and D30A/D66A mutants, the fluorescence change was very small or nonexistent. Therefore, these latter mutants were not used in the subsequent fluorescence following calcium titrations. indicates that Ca 2ϩ binding to the single functioning N-terminal site in these proteins (site II) is the event associated with the fluorescence change. Fig. 6 further highlights the similarity between fluorescent and nonfluorescent forms of D30A and compares the fluorescence changes during calcium titration of the fluorescent mutants with and without the D30A substitution. These results are summarized in Table II. Fig. 6A compares the calcium binding curves obtained by flow dialysis for fluorescent and nonfluorescent D30A mutants. The four binding curves are similar, indicating that the effect of the three different probes at position 29 on Ca 2ϩ binding to site II is very small in the D30A mutants. Fig. 6B compares the fluorescence changes during calcium titration of the fluorescent D30A mutants. The three pCa 1/2 values are very similar: ϳ5.5 for F29W/D30A and F29HW/D30A and ϳ5.3 for F29ZW/D30A. Fig. 6, C-E, compares the calcium titrations followed by fluorescence of the fluorescent mutants with and without the D30A mutation. Both F29W and F29W/D30A have a pCa 1/2 close to 5.5, and the curves differ only slightly at low free calcium concentrations (Fig. 6C). Interestingly, when 5-HW instead of tryptophan is present at position 29, the D30A mutation has a much greater effect: the pCa 1/2 of F29HW is ϳ4.8 whereas the pCa 1/2 of F29HW/D30A is ϳ5.5. The slopes of both curves are similar (predicted Hill coefficient of 1.11 for F29HW and fixed to 1 for F29HW/D30A). Fig. 6E compares the F29ZW and F29ZW/ D30A curves. The bimodal shape of the F29ZW curve is replaced by a monotonic decrease in F29ZW/D30A, with pCa 1/2 Ϸ 5.3 (see Table II).
It is important to highlight that the mid-transition points of Fig. 3, B and E, and Fig. 6E (mutant F29ZW) are only apparent because the absolute increase and decrease are not known. Hence, the apparent mid-transition points are not reflected directly by the K d values. We chose the K d values from the best fitting, and small variations on the assumed absolute increase/ decrease led to significant variation on K d values. We therefore emphasize that the main indication of sequential binding is the bimodal profile of the fluorescence curve and that the affinities of site II and site I (when site II is filled) do not need to be very different to be in accordance with sequential binding.

Recombinant skeletal chicken TnC is devoid of Tyr and Trp.
Pearlstone et al. (25) replaced Phe at position 29 with Trp, allowing Ca 2ϩ binding to regulatory (or N-terminal sites) to be followed using fluorescence techniques. Initial tests indicated that the most important properties of TnC were maintained in F29W, including the Ca 2ϩ affinity of the N-terminal sites. In addition, the fluorescence data revealed a high degree of cooperativity (Hill coefficient of 1.96), consistent with an all-or-none mechanism of Ca 2ϩ binding to the two N-terminal sites. Chandra et al. (29) found that the F29W mutant is deficient in its ability to relieve TnI inhibition of actomyosin S1 ATPase in a Ca 2ϩ -dependent manner. Using the same system, a small increase in N-domain F29W calcium affinity was detected. Li et al. (32) confirmed this affinity increase (0.5 pCa unit, determined by far-UV CD-Ca 2ϩ titration) as well as an increased negative ellipticity of the Ca 2ϩ -loaded state at 221 nm. Yu et al. (33) showed that the apo N-domain of F29W is conformationally less stable at lower temperatures compared with the isolated TnC N-domain.

FIG. 5. Fluorescence change upon calcium binding to F29 mutants and calcium titration comparing the calcium saturation level of each mutant with the fluorescence change.
A shows the fluorescence increase of F29 mutants upon calcium binding relative to the calcium free value, considered as 100%. Only the D/A mutants whose fluorescence intensity varied more than 5% upon calcium addition (the F29 D30A mutants) were used for parallel flow dialysis/ fluorescence following calcium titrations. B, C, and D correlate the fluorescence change (black symbols) of mutants F29W/D30A, F29HW/ D30A, and F29ZW/D30A, respectively, with the calcium saturation level (white symbols) obtained in the correspondent flow dialysis experiments, as in Fig. 3, C-E, for the F29 single mutants. The buffer contained 20 mM MOPS, 100 mM KCl, and 1 mM DTT.
Here we used direct binding assays to show that the F29W N-terminal site calcium affinity is at least 6-fold greater than that of TnC N-terminal sites ( Table I). The addition of a hydroxyl at position 5 of Trp side chain (F29HW mutant) reverts this affinity increase. When 7-ZW is incorporated at position 29, the N-terminal site affinity increases nearly 3-fold compared with TnC. Therefore, the N-domain calcium binding properties of TnC are indeed quite sensitive to the nature of the amino acid side chain at this position (Phe in the native protein; Trp, 5-HW, or 7-ZW in the mutants).
This influence of position 29 on TnC N-domain Ca 2ϩ affinity is not surprising because in the crystal structure of chicken 2Ca 2ϩ -TnC (i.e. unloaded N-domain) an extensive network of interactions involving Phe-29, Met-48, Glu-41, Thr-44, and Val-45, can be observed. In both the x-ray and NMR-determined calcium-saturated structures (34,47,48) the Phe-29 side chain is adjacent to Phe-75, Ile-37 and Val-45. Phe-29 and the residues cited above form a "hydrophobic patch" that is exposed upon calcium binding to the N-terminal sites, as TnC goes from the "closed" to the "open" form. It is therefore very likely that the structure of TnC should be sensitive to the introduction, at position 29, of a bulkier indole side chain with or without potential hydrogen bond donors and acceptors in the six-membered ring. It is generally accepted that in the thin filament, this exposed hydrophobic patch interacts with TnI in the presence of Ca 2ϩ . In the absence of TnI, calcium binding to TnC would force the hydrophobic patch to become solvent-exposed. This explains the reduced calcium affinity of the N-terminal sites in free TnC compared with that measured in the TnC⅐TnI complex (7). Pearlstone et al. (25) and da Silva et al. (11) reported TnC mutants with increased N-terminal site Ca 2ϩ affinity by replacing, one by one, the five highly hydrophobic residues Val-45, Met-46, Met-48, Leu-49, and Met-82 from the hydrophobic patch by Thr, Gln, Ala, Thr and Gln, respectively. These substitutions were expected to lower the energetic barrier between the closed and open forms and, thus, increase the calcium affinity of the N-terminal sites. The increased Ca 2ϩ affinity observed for the F29W N-terminal sites could be the result of a destabilization of the closed form (because of steric clashes involving Trp-29) and/or a stabilization of the open form because of the presence of the N 1 -H (indole) hydrogen bond donor while maintaining at least some of the original interactions of the six-membered ring in the open form. The observation that the Ca 2ϩ affinity of the F29HW N-terminal sites was very similar to that of native TnC (dissociation constants 2.4 ϫ 10 Ϫ5 M and 1.9 ϫ 10 Ϫ5 M, respectively, Table I) suggests that the stabilization or destabilization of the Nterminal domain structure by 5-HW is similar for both the closed and open forms. The mutant F29ZW displayed an intermediate N-terminal site Ca 2ϩ affinity, between that of F29W and TnC (nearly 2 times smaller than that of F29W and 3 times greater than that of TnC, Table I). These data indicate that the decrease in hydrophobic character of the residue at position 29, initially from Phe to Trp, should not be extended as a general rule to explain the influence of that position on N-terminal sites Ca 2ϩ affinity, as highlighted by Tikunova et al. (49) for a series of 27 F29W mutants with hydrophobic residues replaced by glutamine. In the case of F29HW another possibility is an interaction involving the hydroxyl group of 5-hydroxytryptophan residue and the carboxylic group of Glu-41. This could hamper the reorientation of the Glu-41 side chain which is necessary for calcium to bind to site I (48). This interaction could mimic, to a lesser degree, the replacement of Glu at position 41 by Ala, a mutation reported to raise the site I calcium dissociation constant from 16 M to 1.3 mM and impairs opening of N-domain even with site I filling (50).
The fluorescence data obtained for F29W, F29HW, and F29ZW (Table II) were all compatible with those from flow dialysis experiments. The observation of disparate fluorescence behavior for F29ZW was expected because the photophysics of 7-ZW are the most complex of the three chromophores used, with high quantum yield variation and maximum emission sensitivity upon environment changes (51). Rosenfeld and Taylor (52) reported a bimodal fluorescence change using skeletal rabbit TnC labeled at Cys-98 with N,NЈ-dimethyl-N-(iodoacetyl)-NЈ-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)ethylenediamine which occurred at low ionic strength (0 M KCl) and was absent at 0.1 M KCl. The present F29ZW titrations were performed at 100 mM KCl, which argues that the bimodal form of the Ca 2ϩ titrations followed by fluorescence is not the result of subtle variations in the ionic strength which accompany the additions of Ca 2ϩ . Furthermore, the presence of 2 mM Mg 2ϩ does not significantly alter the pCa 1/2 of the transition (data not shown). Although a bimodal curve poses statistical problems to the fitting procedure, the first and second phases of the emission curve may be related to the calcium binding to the N-terminal sites II and I, respectively. The F29ZW fluorescence increase is not caused by calcium binding to C-terminal sites because it occurs principally at pCa values lower than 6.
D30A and F29X/D30A bind three Ca 2ϩ ions, and the Ca 2ϩ binding curves reveal equal affinities among the remaining N-domain site (site II of each mutant) ( Table I) (Fig. 4). In mutants with a modified site I the saturation level of three Ca 2ϩ /molecule is reached at the same pCa as TnC and F29X mutants. We conclude that the sites I and II contribute differently to the Ca 2ϩ binding properties of the N-domain. One possible explanation is that the replacement of aspartate at position 66 by alanine disturbs both sites, whereas the same replacement at position 30 is functionally restricted to site I. Another explanation is that previous filling of site II is essential for calcium binding to site I at physiological calcium concentrations, as suggested by Gagné et al. (53) and Strynadka et al. (48). According to this proposal, Ca 2ϩ binding to site II would "set the stage" for calcium binding to site I. An unoccupied site I would thus exist in two conformations: one previous to Ca 2ϩ binding to site II and other after Ca 2ϩ binding to site II. The data presented here indicate that site I is essentially defunct without previous site II filling. Strong site I functional dependence on site II filling is compatible with the experimentally determined stepwise filling and lack of cooperativity of the N-terminal domain (32). The different Ca 2ϩ binding curves of TnC and fluorescent TnCs may be a consequence of differences exclusively among the sites I or among the stabilities of the open forms, which would be sensitive to residue at position 29.
The F29W and F29W/D30A fluorescence changes during Ca 2ϩ titration shown in Fig. 6 occur in the same pCa range, but F29W transition displays a greater Hill coefficient (Table II). This observation is compatible with the hypothesis that binding to site II activates site I (obligatory sequential binding model) and that Ca 2ϩ binding to F29W site I causes the large fluorescence increment. On the other hand, the F29HW fluorescence transition occurs nearly 0.7 pCa unit after the F29HW/D30A fluorescence transition. Therefore, site II in F29HW has an affinity nearly five times lower than in proteins with only site II functional. Furthermore, the site I affinity (when site II is filled) is nearly 2-fold lower than site II affinity in the F29HW mutant. Fig. 6E compares the F29ZW and F29ZW/D30A fluorescence following Ca 2ϩ titration. The F29ZW/D30A fluorescence change has a pCa 1/2 of nearly 5.6, corresponding to a dissociation constant nearly 5 ϫ 10 Ϫ6 M, which is of the same order of magnitude as that observed for the F29W/D30A and F29HW/D30A mutants. The bimodal shape of the F29ZW fluorescence change may be related to an increase in fluorescence caused by Ca 2ϩ binding to site II and a fluorescence decrease associated with Ca 2ϩ binding to site I (that is, the fluorescence of the saturated form is lower than the fluorescence of the form with only site II filled). The affinity of the site I (when site II is filled) would be close to that of site II (Table II).
Previous results related to cooperativity between TnC sites I and II or III and IV are contradictory. Earlier reports (7,12,13) excluded the cooperativity parameter from the fitting process. Some later studies (15) found no cooperativity between the sites in each domain, even though data were allowed to fit to the Hill equation. Nevertheless, some reports detected cooperativity between C-terminal sites but no cooperativity between N-terminal sites (9, 16, 54, 56). All but one of these studies (9)   a Although F29ZW/D30A has an unique functional N-site, the fluorescence curve showed an abnormal slope. The fitting procedure maintained the parameter n 2 , but with no cooperativity significance. utilized EGTA buffers to control the free calcium concentration. Pearlstone et al. (25) and Tikunova et al. (49) have reported the highest value for cooperativity between N-terminal sites. However, Pearlstone et al. (57) highlighted the difficulty in conferring biological significance to cooperativity inferred from Ca 2ϩ titrations using indirect spectroscopic methods. This is mainly because of Hill coefficients greater than unity which were predicted for processes known to be related to the filling of only one site (57). Although our fluorescence data led in some cases to predicted Hill coefficients different from unity (Table II), we consider that there is no secure reason to assume all those values as real indications of positive or negative cooperativity. Also, direct Ca 2ϩ binding data from flow dialysis experiments involve such variability (reflected in Hill coefficients from 0.75 to 1.51, Table I) that we do not consider them to confirm the existence or nonexistence of cooperativity between sites I and II or III and IV. Furthermore, the obligatory sequential binding model does not maintain the correspondence between Hill coefficients and cooperativity.
The near-UV CD spectra changes associated with calcium binding (data not shown) indicate that the F29W D/A and F29HW D/A mutants do not suffer large decreases in CD bands at wavelengths above 260 nm upon calcium binding. Pearlstone et al. (57) suggested that this decrease is associated with opening of the N-domain, based on the characteristics of cardiac TnC (cTnC) and the E41A mutant of TnC. These proteins show neither the CD band decrease nor the opening of N-domain upon calcium binding (53,55). In our study, the data suggest that the replacement of Asp by Ala at position 30 impairs opening of the N-domain of mutants F29W/D30A and F29HW/ D30A. The calcium-associated fluorescence increase of these mutants may be related to the effect of metal binding to site II on the environment of the probe at position 29 and reinforces the opening of N-domain as the cause of the large fluorescence increase of F29W and F29HW mutants upon calcium binding. Because there is no large increase in fluorescence of F29W/ D30A and F29HW/D30A mutants even at pCa values near 2 (data not shown), calcium binding to the mutated sites I which would eventually occur at these high calcium concentrations does not lead to N-domain opening (as in E41A mutant and in cardiac TnC).
Our results indicate that position 29 strongly influences the Ca 2ϩ affinity of TnC N-terminal sites. F29W has N-terminal sites with higher affinity than TnC, and the addition of 5-HW reverts this increase. Incorporation of 7-ZW results in intermediate N-terminal site affinities between those of F29W and of TnC. The direct binding data of D/A mutants suggest that site I is essentially defunct without previous Ca 2ϩ binding to site II and that the differences among TnC and F29X N-domain site affinities are the consequence of differences in the site I affinities only (when site II is filled). The fluorescence data of F29W/D30A and F29HW/D30A mutants may be explained assuming obligatory sequential Ca 2ϩ binding (first to site II) with the Ca 2ϩ binding to site I as the cause of the fluorescence increase and N-domain opening. The bimodal F29ZW fluorescence change is the result of a fluorescence increase caused by Ca 2ϩ binding to site II and a subsequent fluorescence decrease associated with Ca 2ϩ binding to site I.