Specific Sequences Determine the Stability and Cooperativity of Folding of the C-terminal Half of Tropomyosin*

Tropomyosin is a flexible 410 Å coiled-coil protein in which the relative stabilities of specific regions may be important for its proper function in the control of muscle contraction. In addition, tropomyosin can be used as a simple model of natural occurrence to understand the inter- and intramolecular interactions that govern the stability of coiled-coils. We have produced eight recombinant tropomyosin fragments (Tm143–284(5OHW), Tm189–284(5OHW), Tm189–284, Tm220–284(5OHW), Tm220–284, Tm143–235, Tm167–260, and Tm143–260) and one synthetic peptide (Ac-Tm215–235) to investigate the relative conformational stability of different regions derived from the C-terminal region of the protein, which is known to interact with the troponin complex. Analytical ultracentrifugation experiments show that the fragments that include the last 24 residues of the molecule (Tm143–284(5OHW), Tm189–284(5OHW), Tm220–284(5OHW), Tm220–284) are completely dimerized at 10 μm dimer (50 mm phosphate, 100 mmNaCl, 1.0 mm dithiothreitol, and 0.5 mm EDTA, 10 °C), whereas fragments that lack the native C terminus (Tm143–235,Tm167–260, and Tm143–260) are in a monomer-dimer equilibrium under these conditions. The presence of trifluoroethanol resulted in a reduction in the [θ]222/[θ]208 circular dichroism ratio in all of the fragments and induced stable trimer formation only in those containing residues 261–284. Urea denaturation monitored by circular dichroism and fluorescence revealed that residues 261–284 of tropomyosin are very important for the stability of the C-terminal half of the molecule as a whole. Furthermore, the absence of this region greatly increases the cooperativity of urea-induced unfolding. Temperature and urea denaturation experiments show that Tm143–235 is less stable than other fragments of the same size. We have identified a number of factors that may contribute to this particular instability, including an interhelix repulsion between g and e′ positions of the heptad repeat, a charged residue at the hydrophobic coiled-coil interface, and a greater fraction of β-branched residues located at d positions.

The coiled-coil motif mediates the process of oligomerization of many proteins. This structure is a consequence of a heptapeptide repetition (abcdefg) in the chemical nature of the residues in the primary structure of the polypeptide chain (1)(2)(3). An ␣-helical conformation places hydrophobic residues at positions a and d on the same side of the helix, creating a nonpolar interface that promotes dimerization through the burial of hydrophobic surface. Residues at positions e and g are often charged and may form salt bridges with residues at positions g and e, respectively, of the other helix. The maximization of favorable ionic attractions and the minimization of unfavorable repulsions probably influence the particular alignment of the ␣-helical chains (4). The folding of synthetic peptides that adopt a coiled-coil structure has been investigated extensively (4 -11). Synthetic peptide models for coiled-coils may not always be taken as representative because they often employ regular patterns of residues at positions a and d and lack i, iϩ3 and i, iϩ4 intrahelical ionic interactions (7,12). Natural coiled-coil sequences, such as tropomyosin (Tm), 1 are more complex, with irregular patterns of intrahelix and interhelix side chain-side chain interactions and a greater variety of residues at the hydrophobic core. Tm can be used as a natural model for the study of the principal interactions that maintain the coiled-coil structure and ␣-helix stability.
Muscle ␣-Tm is a symmetric coiled-coil composed of two parallel and in register ϳ410 Å, 284-residue ␣-helices (2,13,14). The C-terminal half of the Tm molecule is less thermally stable than the N-terminal half (15,16). A series of studies (17,18) have suggested that Tm fragments with less than 94 residues were unable to form stable secondary structures at low micromolar concentrations. Recently, Holtzer et al. (19) showed that a 65-amino acid fragment (residues 190 -254) presents a significant amount of ␣-helix (ϳ43%) when present in high micromolar concentrations (115 M) at 10°C but is essentially unfolded at 25°C.
Microcalorimetric analysis of ␣-Tm thermal denaturation has been interpreted as a multistep process in which specific Tm segments ("cooperative blocks") denature independently of one another (20). However, the sum of the denaturation curves for two Tm subfragments corresponding to predicted cooperative blocks (190 -254 and 253-280) is not equal to the experimental curve for the fragment 190 -284, arguing for long range cooperativity along the Tm structure (19). Kinetic studies of the Tm folding process show that the rate-limiting step is not dimerization, but rather a slow first order isomerization of the preformed and correctly aligned dimer (21). Recently, Kammerer et al. (22) proposed the importance of a minimal consensus "trigger" sequence responsible for the formation of autonomous folding units and partially folded coiled-coil dimers. However, Lee et al. (23) have suggested that a complex combination of stabilizing effects along the sequence is a more general indicator of protein folding in coiled-coils than the identification of a specific sequence. In the case of Tm, the fact that fragments derived from different parts of the molecule form stable coiled-coils argues against the importance of a specific trigger sequence in Tm folding. Whether or not Tm folds via independent cooperative blocks or is nucleated by one or more specific sequences, it seems clear that Tm contains regions that are more intrinsically stable than others (15,17,20,24,25). During thermal denaturation, the Tm region near residue Cys 190 unfolds before strand separation (24), and high pressure produces partially unfolded Tm intermediates in which the polypeptide chains remain associated (26). Furthermore, the chemical environment of a pyrene probe bound to Cys 190 is sensitive to the presence or absence of the last C-terminal residues of the Tm chain (27). Therefore, long range communication between regions of variable stability may contribute to the folding and dimerization processes.
In this work we have focused on the conformational stability of fragments derived from the C-terminal half of Tm and attempted to determine the contributions of these different regions to the stability of Tm as a whole. We present evidence that residues 261-284 confer a particular stability to the C-terminal region of Tm. Furthermore, the presence or absence of this region affects the cooperativity of the ureainduced denaturation process. In contrast, the region between residues 143 and 235 is relatively unstable, and we have identified a number of factors, which together may explain this observation. Our results may contribute to the understanding of the folding process in naturally occurring coiled-coils with relatively complex primary structures and may be relevant to the physiological function of Tm in the regulation of muscle contraction.
Circular Dichroism Experiments-Circular dichroism (CD) spectra were recorded on a Jasco 720 spectropolarimeter. Buffer conditions and protein concentrations are indicated in the figure legends, and cell temperature was maintained with a circulating water bath. Far UV CD (190 -260 nm) spectra were collected at 0.5-nm intervals, using a response time of 0.25 s and a velocity of 100 nm/min. At least four scans were collected for the urea and thermal denaturation curves (at 10°C), and at least eight scans were collected for the experiments at 25°C. The ␣-helix content was calculated based on the algorithm developed for coiled-coil proteins by Holtzer et al. (33).
where, ⌽ h ϭ fraction of residues in an ␣-helical conformation, hϱ ϭ mean residual ellipticity at 222 nm for a helix of infinite length (Ϫ386 degrees⅐cm 2 ⅐mmol of peptide bonds Ϫ1 ), c ϭ mean residual ellipticity for a random coil at 222 nm (Ϫ10 degrees⅐cm 2 ⅐mmol of peptide bonds Ϫ1 ), I ϭ the average number of helical segments per chain (in this case equal to 1), n ϭ the number of peptide bonds, and ‫؍‬ mean residual ellipticity at 222 nm. Urea and Thermal Denaturation Monitored by CD-Urea denaturation studies were carried out in 50 mM sodium phosphate (pH 7.0), 100 mM KCl, 0.5 mM DTT, and 0.5 mM EDTA. A fresh stock solution of ϳ9 M urea (Sigma ultra pure grade) was prepared in the above buffer and mixed with the proteins to achieve the desired final concentration of urea and protein (10 M dimer). Urea concentration was confirmed by refractive index (34). Samples were incubated at 10°C for 3 h to reach equilibrium before recording the CD spectra. The urea denaturation curve was analyzed as described by Pace and Scholtz (34), using Equation 2.
⌬G is the free energy of unfolding at each point of the denaturation curve, Y is the mean residual ellipticity at 222 nm, [urea] 1/2 is the urea concentration at which 50% of the molecules are unfolded (assuming a two-state transition). Y f and Y u are the linear equations that define the pre-transition and post-transition base lines (urea sensitivities of the CD spectra of the folded and unfolded states, respectively). For all fragments, Y u was obtained directly from the denaturation curves. For Tm 143-284(5OHW) and Tm 189 -284(5OHW) , equations for Y f were also estimated directly from the denaturation curve. For some relatively unstable fragments, Y f could not be estimated accurately from the denaturation data and was therefore assigned values based on the similarity of the shape of the denaturation curve to that of other fragments. In this manner, Y f for Tm 143-260 was assumed to be equal to that of Tm 143-284(5OHW) . For Tm 220 -284(5OHW) , Tm 143-235 , and Tm 167-260 we used the Y f equation for Tm 189 -284(5OHW) . Curve fitting simulations showed that the calculated values of m and [urea] 1/2 were relatively insensitive to the specific equation for Y f employed. Thermal denaturation experiments were performed incubating a 10 M (dimer) protein solution in the above buffer for 5 min at each temperature before recording the CD spectra as described above.
Urea Denaturation Monitored by 5-Hydroxytryptophan Fluorescence-Fluorescence experiments were performed on an AVIV Automated Titrating Differential/Ratio spectrofluorometer at 10°C. The emission fluorescence spectra were collected at 337 nm (5-mm bandwidth) using an excitation wavelength of 305 nm (1-mm bandwidth). Buffer conditions were the same as for the urea denaturation experiments monitored by circular dichroism (above). Aliquots of a ϳ9 M urea solution were added automatically into the sample and reference cuvettes, and solutions were equilibrated for 5 min before each measurement. At the end of each titration, reversibility was confirmed by measuring the fluorescence of the sample after dilution with 20 volumes of buffer without urea. The data were analyzed in a manner similar to that described above for the CD experiments. Y u was obtained directly from the denaturation curves. Y f for Tm 189 -284(5OHW) and Tm 220 -284(5OHW) was set to be equal to Y f estimated from the denaturation curve of Tm 143-284(5OHW) .
Analytical Ultracentrifugation (AUC)-Protein solutions were dialyzed against 50 mM sodium phosphate (pH 7.0), 100 mM NaCl, 0.5 mM EDTA, 1 mM DTT. Sedimentation equilibrium experiments were carried out at 10°C in a Beckman XL-I analytical ultracentrifuge using absorbance optics following the procedures described by Laue and Stafford (35). The fragments were analyzed at two different protein concentrations, in the absence or presence of 5 M urea and in the presence of 30% TFE. 110-l aliquots of sample solution were loaded into six-sector charcoal-filled epon sample cells. Runs were performed at a minimum of two different speeds, and each speed was maintained until there was no significant difference in r 2 /2 versus absorbance scans taken 2 h apart to ensure that equilibrium was achieved. The sedimentation equilibrium data were evaluated using the NONLIN program, which incorporates a nonlinear least squares curve-fitting algorithm described by Johnson et al. (36). Data were fit to either a single ideal species model or models containing two associating species. The protein partial specific volume and the solvent density were estimated using the SEDNTERP program, which incorporates calculations detailed by Laue et al. (37).
Gluteraldehyde Cross-linking Experiments-All proteins were equilibrated for 30 min under the conditions indicated in the figure legends. Gluteraldehyde was added to a final concentration of 0.02%, and after 15 min the cross-linking reaction was stopped by the addition of SDS-PAGE sample buffer. The products of the cross-linking reaction were analyzed in 15% SDS-PAGE.
Peptide Synthesis, Purification, and Characterization-Ac-Tm 215-235 (Ac-SQKEDKYEEEIKVLTDKKLEA) was prepared manually by the stepwise solid phase method using Boc (tert-butyloxycarbonyl) chemistry as described previously (38,39). Analysis and purification were performed by reversed phase high performance liquid chromatography using C 18 columns. The chemical identity of the purified peptide was confirmed by electrospray ionization mass spectrometry and amino acid analysis.

RESULTS
The recombinant fragments, derived from the C-terminal half of skeletal muscle ␣-Tm, are shown schematically in Fig. 1. They were designed so that all of the N-terminal residues fall in solvent-exposed positions of the heptad repeat (c, g, or f) to avoid electrostatic repulsion between the protonated ␣-amino groups, which may destabilize the coiled-coil structure (30). On the other hand, the C termini all fell in an internal position of the heptad repeat, to mimic the C-terminal of native Tm, where Ile 284 occupies a d position. The peptide Ac-Tm 215-235 ( Fig. 1) was synthesized with an acetylated N terminus and a free C terminus. Fragments Tm 143-284(5OHW) , Tm 189 -284(5OHW) , and Tm 220 -284(5OHW) have a fluorescent probe (5-hydroxytryptophan) at position 278, occupied by a leucine in the native protein. The fluorescence intensity of this probe is sensitive to the conformational state of the polypeptide (see below) as well as troponin binding and head-to-tail interactions. 2 Two fragments, Tm 189 -284 and Tm 220 -284 , with the native leucine residue at position 278, were also produced (not shown in the figure).
We initially characterized the relative helix propensities of the Tm fragments (all 10 M dimer) by analyzing their circular dichroism spectra in phosphate buffer (50 mM phosphate (pH 7.0), 0.5 mM EDTA, 0.5 mM DTT) at 25°C under three conditions: in the presence and absence of 100 mM KCl and in the presence of 100 mM KCl and 50% TFE. The calculated helix content of all of the fragments increased in phosphate buffer upon the addition of 100 mM KCl (data not shown). The stability of Tm is known to increase as a function of ionic strength (40,41). The CD spectra in the presence of 100 mM KCl with and without TFE are shown in Fig. 2, and the ␣-helix contents of these fragments, based on their residual ellipticities at 222 nm, are presented in Table I.
We found that in the absence of TFE, the ␣-helix contents of the fragments derived from the C-terminal half of Tm are not only dependent on their size, but also on the specific Tm region encompassed by the fragment. Tm 220 -284(5OHW) and Tm 220 -284 , despite their relatively small size (65 residues), present CD spectra typical of an ␣-helix with minima at 208 and 222 nm (Table I and Fig. 2). On the other hand Tm 143-235 , which has 93 amino acids, is only 20% ␣-helical at 25°C (Table I) and has a CD spectrum that is more characteristic of a random-coil structure (Fig. 2). Another fragment, Tm 167-260 (94 residues) also has a low ␣-helix content (28%) at 25°C. All of the other recombinant Tm fragments present spectra typical of ␣-helical proteins in phosphate buffer plus 100 mM KCl. Therefore, for all fragments with 96 or fewer amino acids, the presence of residues 261-284 was essential to observe substantial helix content under these conditions. Lowering the temperature to 10°C increases the helix content of all the fragments except Tm 143-284(5OHW) . At this temperature, all of the recombinant fragments have calculated helix contents of at least 60% (Table I).
The presence of 50% TFE promotes an increase in the ␣-helix content of Tm 220 -284(5OHW) , Tm 220 -284 , Tm 143-235 , Tm 167-260 , and Tm 143-260 ( Fig. 2 and Table I). A comparison of the CD spectra in the presence and absence of TFE shows that all fragments experience an inversion in the relative magnitudes of their ellipticity minima at 208 and 222 nm ( Fig. 2 and Table  I). The 222 / 208 ratio was found to be equal or greater than 1.0 in the absence of TFE for all fragments except for Tm 143-235 and Tm 167-260 , which do not present much secondary structure under these conditions (Table I). In the presence of 50% TFE, 222 / 208 is reduced to values below unity. The 222 / 208 ratio has been proposed to be a function of the oligomerization state of a coiled-coil protein (see "Discussion"). We investigated the oligomerization state of the recombinant Tm polypeptides by AUC under a variety of conditions. Fig. 3 shows the sedimentation equilibrium data collected for Tm 220 -284(5OHW) in phosphate buffer (ϩ100 mM NaCl) with or without TFE or 5 M urea. For this fragment, the data in each case could be fit to a single species model: dimers in phophate buffer, trimers in the presence of TFE, and monomers in the presence of urea. The apparent molecular masses obtained through sedimentation equilibrium experiments show that all fragments tested which possess the last 24 C-terminal residues (261-284) of tropomyosin are completely dimerized at 10 M dimer in phosphate buffer plus 100 mM KCl, whereas the fragments that do not have this region are in a monomer-dimer equilibrium under these conditions (Table II). These latter fragments were found to be fully dimerized (or almost fully dimerized in the case of Tm 143-260 ) at higher concentrations (between 100 and 140 M) (Table II). In the presence of 5 M urea, all fragments tested dissociated to monomers (Table II), which appear to be completely unfolded (see Fig. 4). AUC experiments performed in the presence of 30% TFE at 10 M dimer concentration showed that the fragments possessing residues 261-284 form stable trimers. Under these conditions, the fragments lacking this region present apparent molecular masses significantly less than that expected for a trimer (Table  II). The data for Tm 143-235 are best fit by a dimer model, whereas the data for Tm 143-260 and Tm 167-260 are best fit by 2 A. A. Paulucci and C. S. Farah, manuscript in preparation. assuming a mixture of monomers and either dimers or trimers. For these two fragments, the AUC data are fit equally well by monomer-dimer and monomer-trimer models, so trimer formation cannot be ruled out. The tendency to form trimers in the presence of TFE is not solely a function of coiled-coil stability because the second largest and relatively stable fragment Tm 143-260 forms trimers to a much lesser extent, if at all (Table  II). Trimer formation is also not attributable to the leucine to 5-hydroxytryptophan substitution at residue 278 because both Tm 220 -284(5OHW) and Tm 220 -284 form trimers in 30% TFE (Table II). The significance of TFE-induced trimer formation is unclear at the moment, but we may speculate that this higher order association may be mimicking tertiary interactions with other thin filament proteins (actin and troponin). For example, troponin T (TnT) is known to possess a sequence with a high ␣-helix content (ϳ80%) between residues 71 and 151, which may be interacting with the C-terminal of Tm in an antiparallel fashion (42). Furthermore, the last 9 residues of Tm are able to interact with the N-terminal of a second Tm molecule to form a head-to-tail complex responsible for Tm polymerization (2). The AUC and CD results also indicate that at least in the case of polypeptides derived from the C-terminal half of Tm, the TFE-  1) and (2), residual ellipticity (222 nm) of Tm fragments at 25°C without and with 50% TFE. (3) and (4), residual ellipticity ratios of Tm fragments without and with 50% TFE. (5) and (6), % ␣-helix was calculated using the ellipticity at 222 nm as described under "Materials and Methods." (7) and (8)   induced inversion in the 222 / 208 ratio is not associated with a coiled-coil to ␣-helix monomer transition, but may in fact be the result of more subtle conformational changes in ␣-helical structure (see also Ref. 25).
To analyze the conformational stability of the recombinant Tm fragments in more detail, we performed urea-induced unfolding followed by CD at 222 nm in phosphate buffer plus 100 mM KCl at 10°C. We chose to carry out these experiments at this lower temperature because the less stable fragments present more helical structure at 10°C than at 25°C (Table I). The denaturation curves are presented in Fig. 4. Folding of Tm and its fragments is known to be fully reversible ((21) and see Figs. 5 and 6 below). The urea denaturation data were analyzed quantitatively to calculate [urea] 1/2 , the urea concentration at which 50% of the molecules are unfolded and m, a constant associated with the cooperativity of the denaturation transition (Table I and see "Materials and Methods"). When we compare the denaturation curves of Tm 143-284(5OHW) (142 residues), Tm 189 -284(5OHW) or Tm 189 -284 (96 residues), and Tm 220 -284(5OHW) or Tm 220 -284 (64 residues), we observe that their relative stabilities, as reflected by [urea] 1/2 , are sizedependent. These fragments all possess residues 261-284, which we previously identified as important for dimerization and stability (above). A small but significant effect on stability was observed for the leucine to 5-hydroxytryptophan substitution at position 278. In the two cases in which direct comparison was possible, the wild-type sequence denatured with a [urea] 1/2 0.4 -0.5 M greater than that observed for the corresponding fragment containing 5-hydroxytryptophan ( Fig. 4 and Table I). It is not clear whether 5-hydroxytryptophan stabilizes or destabilizes ␣-helices with respect to leucine. Furthermore, residue 278 corresponds to an e position within the heptad repeat which, in a canonical coiled-coil structure, is often occupied by charged residues involved in interhelical contacts with g position residues (i-5), in this case residue Glu 273 . Because both leucine and 5-hydroxytryptophan are uncharged and because the C-terminal region may in fact not be a canonical coiled-coil (43; see also "Discussion"), the precise molecular basis for the observed differences in stability is not evident.
Comparison of the denaturation curves of the similar-sized fragments Tm 143-235 (93 residues), Tm 167-260 (94 residues), and Tm 189 -284(5OHW) or Tm 189 -284 (96 residues) clearly illustrates that the specific amino acid sequence is very important for fragment stability, and that stability increases as the sequence moves toward the C terminus of the molecule. Consistent with the helix content data obtained at 25°C, these denaturation  Table I).
Inspection of the denaturation curves in Fig. 4 shows that the cooperativity of the denaturation transitions varies from fragment to fragment. This cooperativity is reflected in the value for m ( Table I). Comparison of the m values obtained for similar fragments containing either leucine or 5-hydroxytryptophan at position 278 were essentially the same (Table I). When we compare the data for Tm 143-284(5OHW) and Tm 143-260 we note that the absence of residues 261-284 promotes an increase in the m value (from 1.5 to 3.0 kcal⅐mol Ϫ1 ⅐M Ϫ1 ), indicating an increase in the cooperativity of the denaturation curve (Table I). Tm 143-235 also denatures with high cooperativity (m ϭ 2.4 kcal⅐mol Ϫ1 MϪ1 ), whereas Tm 167-260 has a low m value (1.3 kcal⅐mol Ϫ1 MϪ1 ). It should be noted that the m values for the two least stable fragments (Tm 143-235 and Tm 167-260 ) may not be very precise because these fragments are already partially unfolded (Fig. 4) and not fully dimerized (Table II) in the absence of urea. Therefore, the data necessary to estimate m accurately in these cases is limited. For the fragments with well defined pre-and post-transition base lines, however, it is clear that those containing residues 261-284 denature with less cooperativity than fragments lacking this region ( Fig. 4 and Table I) and may also indicate the presence of one or more intermediates in the denaturation profile (see "Discussion").
The urea denaturation of Tm 143-284(5OHW) , Tm 189 -284(5OHW) , and Tm 220 -284(5OHW) was also monitored by observing changes in the fluorescence intensity of the 5-hydroxytryptophan probe at position 278 (Fig. 5A). The progress of the transition at each urea concentration was calculated and compared with that obtained in the CD experiments (Fig. 5B). Unlike the CD experiments, which reflect the average conformational state of the polypeptide chain, the fluorescence data report on the average microenvironment of the probe under specific conditions. In agreement with the CD data, denaturation monitored by fluorescence shows that fragment Tm 143-284(5OHW) is more stable than Tm 189 -284(5OHW) and Tm 220 -284(5OHW) . The fluorescence data did not, however, reflect the differences in stability observed for Tm 189 -284(5OHW) and Tm 220 -284(5OHW) in the CD urea denaturation experiments ( Fig. 5 and Table I). In addition, the m values obtained from the fluorescence experiments were similar but not equal to those obtained from the CD denaturation experiments (Table I). In the case of a two-state transition one would expect the CD and fluorescence data to be superimposable. Nonsuperimposable curves monitored by different methods can be used as evidence for a stable or metastable intermediate(s). At the moment, we are reluctant to arrive at this conclusion based on the fluorescence data alone because the pre-and post-transition base lines in the fluorescence experiments are significantly steep, partially masking the actual transition and reducing the accuracy with which we    As mentioned above, Tm 143-235 is particularly unstable compared with other fragments derived from the C-terminal half of the Tm molecule. Furthermore, the helix content of Tm 143-235 is very sensitive to the presence of TFE (Table I). The ellipticity of this fragment at 222 nm varies from Ϫ8,113 degrees⅐cm 2 ⅐dmol Ϫ1 in the absence of TFE to Ϫ16,527 degrees⅐cm 2 ⅐dmol Ϫ1 in the presence of 50% TFE at 25°C (Fig.  2). Thermal denaturation experiments monitored by CD show that this fragment (10 M dimer) has a very low melting temperature (Fig. 6A), being extensively folded (78%) at 5°C and completely unfolded at 30°C. Gluteraldehyde cross-linking experiments performed with Tm 143-235 at 5°C intervals between 10 and 30°C (Fig. 6C) are consistent with the CD data. Glut-araldehyde-induced covalent dimers of this fragment could be detected up to 20°C (in the middle of the thermal denaturation transition), whereas little or no dimers were observed at and above 25°C where the fragment is unfolded (compare Fig. 6, A  and C). Furthermore, the ellipticity of this fragment is highly dependent on its peptide concentration. At 10°C, a 25-fold increase in polypeptide concentration results in a significant increase in helix content (Fig. 6B).
The relative instability of Tm 143-235 is interesting because its C-terminal contains the first 10 of 13 residues of a putative minimal consensus region, which has been proposed to be responsible for the autonomous folding of coiled-coils (22). This putative trigger sequence (XXLEXchXcXccX), where X is any residue, h is a hydrophobic residue, and c is a charged residue, corresponds to amino acids 226 -238 of Tm (Fig. 7A) and is also found in our other five recombinant fragments. We therefore synthesized peptide Ac-Tm 215-235 to test whether the concentration and TFE dependence of the helical content of Tm 143-235 could be the result of a short well folded structure that associates to nucleate and stabilize the coiled-coil. CD analysis of Ac-Tm 215-235 shows that it assumes very little helical structure in phosphate buffer plus 100 mM KCl at 10°C at all polypeptide concentrations between 0.1 and 2 mM (Fig. 7B). Under these conditions, the parental peptide Tm 143-235 has significant helix content (60%; Fig. 6 and Table I). The addition of TFE increases the helix content of Ac-Tm 215-235 from 0.4 to 33% (Fig. 7, B and  C). Therefore, although residues 215-235 possess an intrinsic helix forming potential, we have no evidence that they fold "autonomously" in the context of a larger fragment. Tm 220 -284(5OHW) , and Tm 220 -284 ) are stable dimers. Comparison of helix content at 25°C and [urea] 1/2 at 10°C shows that Tm 220 -284(5OHW) and Tm 220 -284 are significantly more stable than other Tm fragments of up to 94 amino acids which lack residues 261-284. In previous studies, Tm fragments containing 95 or more amino acids were found to form stable, fully folded coiled-coils (15,16). On the other hand, the only two fragments studied to date with sizes comparable with that of Tm 220 -284(5OHW) and Tm 220 -284 were both found to be particularly unstable (15,19). A fragment corresponding to residues 183-244 presented a CD spectrum characteristic of a random coil at 27°C and only 17% helix content at 10°C (15). Holtzer et al. (19) showed that a 65-amino acid fragment (residues 190 -254) presents a significant ␣-helix content (ϳ43%) when present in high concentrations (115 M) in 50 mM phosphate (pH 7.4), 100 mM NaCl, 10 mM DTT at 10°C but is almost completely denatured at 25°C even at this high concentration. These results are consistent with our observations for Tm 143-235 and Tm 167-260 , where low stabilities cannot be explained by their sizes but rather must be a consequence of the intrinsic instability of the central region of the Tm polypeptide chain (residues 143-235). Other studies have also presented evidence that the region around Cys 190 is particularly unstable (20,24).
Our AUC analysis indicates that TFE does not result in strand dissociation of our Tm fragments, but in fact induces the formation of trimers in some fragments and a concomitant decrease in the 222 / 208 ratio. Therefore, we can conclude that inversion of the 222 / 208 ratio is not the result of strand dissociation. Quadrifoglio and Urry (44) observed a similar TFEinduced inversion with copoly-L-Glu-Ala-Lys, which does not adopt a coiled-coil structure. Recent studies have shown that some peptides are induced to associate in the presence of TFE, forming dimers, trimers, and even tetramers (45,46). Therefore, the ellipticity inversion may be caused by subtle and poorly understood conformational changes in helical structure or solvent effects rather than a coiled-coil to ␣-helix transition (25).
Our results show that residues 261-284 clearly play an important role in the unfolding mechanism of the C-terminal half of Tm because their presence or absence has clear effects on conformational stability, TFE-induced trimer formation, and the cooperativity of the unfolding transition (m values). The denaturation of fragment Tm 143-260 was found to be especially cooperative. The cooperativity of unfolding of this 116-amino acid fragment is not related to any particular instability because it is quite highly helical in the absence of denaturant (82% at 10°C, Table I) and begins to denature only at urea concentrations greater than 0.5 M (Fig. 4). Why then, does the presence of these C-terminal residues (261-284) reduce the cooperativity of unfolding? If the C-terminal region of Tm forms a particularly stable structure that may remain folded even as neighboring sequences begin to unfold, then the unfolding transition of fragments containing this structure would be a stepwise process and be expected to occur over a wider range of urea concentrations, even though an intermediate species is not clearly detected. We cannot precisely define the limits of this putatively stable region because partial deletions could destabilize it; but the evidence we presented above seems to indicate that it may lie between residues 220 and 284 and include at least part if not all of the last 24 amino acids. A 31-amino acid fragment that included Tm residues 253-280 was nearly completely unfolded at 2°C and concentrations of up to 1.1 mM (25), indicating that the stability conferred by residues 261-284 is dependent on neighboring sequences.
Model peptides with primary structures of relatively low complexity are useful to test hypotheses about the contribution of specific side chain-side chain contacts toward the stability of coiled-coils (7,8,11). The application of conclusions drawn from these studies to naturally occurring proteins such as Tm, however, is seldom straightforward. This is because the highly complex sequences of natural proteins almost invariably create a unique microenvironment for every pair of potentially interacting residues. It is to be expected that the topological and electrostatic details of this microenvironment will be of great importance in determining whether and to what degree two or more side chains will interact and the contribution of these interactions to the overall stability of the protein.
Despite the above restrictions, several groups (see below) have developed a number of algorithms, based on extensive thermodynamic and structural data, to predict ␣-helix and coiled-coil stabilities. We therefore ventured to analyze the amino acid sequence of the C-terminal half of Tm (residues 143-284) in an attempt to identify specific patterns that could possibly explain the differential stabilities of the Tm fragments tested. First we analyzed factors that could stabilize or destabilize the helical conformations of the Tm fragments. For example, ␣-helices can be stabilized or destabilized by attractive or repulsive interactions between charged side chains at position i and iϩ3 or iϩ4 (47). Although many potentially destabilizing ion pairs could be identified, we could not detect any significant differences in the number (normalized with respect to fragment size) of possible intrahelix repulsions between likecharged side chains in the fragments. We found that these repulsive pairs were evenly distributed throughout the studied region. We also used the program AGADIR (47)(48)(49) to predict the propensity for ␣-helix formation for the polypeptide chains used in this study. This algorithm, based on helix-coil transition theory, considers several types of interactions including main chain hydrogen bonds, side chain-side chain interactions, intrinsic helical propensities, capping interactions, and intermediate range electrostatic interactions. Analysis by AGADIR found no significant differences in helix propensity among the recombinant Tm fragments. All fragments, when analyzed as single chains, present low propensity to adopt an ␣-helical structure (3-7% at pH 7, I ϭ 0.1), which is consistent with the analysis of Holtzer et al. (50) who used Zimm-Bragg helix-coil theory to demonstrate that even full-length Tm polypeptide chains would not be very helical on their own if confined to the monomeric state. Therefore, it is evident that the observation that full-length Tm is almost fully helical (15,41) must be explained in large part by stabilizing tertiary interactions of both hydrophobic and electrostatic natures.
Several programs have been designed to predict the tendency of a polypeptide sequence to adopt a coiled-coil conformation. The algorithm STABLECOIL (www.pence.ca/stablecoil/) estimates coiled-coil stability using free energy values for relative ␣-helix propensity of the amino acids at solvent-exposed positions and differential stability contributions at positions a and d of the dimer interface obtained from extensive amino acid substitution data in model polypeptides (7,8,11,51). Analysis of the Tm sequence using STABLECOIL clearly indicates a strong coiled-coil potential for residues 220 -284, whereas Tm 143-235 scores below the threshold for coiled-coil formation (data not shown) and Tm 167-260 give intermediate values.
These predictions are consistent with our CD, urea denaturation, and AUC results, which point to a particularly unstable region between residues 143 and 235.
An important factor for coiled-coil stability is the hydrophobic interactions between residues at positions a and d of the heptad repeat, which result in the side-by-side association of two or more helices (1, 12, 52). Tripet et al. (51) compared the relative contributions of each amino acid at position a and d of the heptad repeat. In this manner they were able to rank the amino acids at each position as thermodynamically stabilizing or destabilizing (with respect to alanine, which we may consider neutral). Table III shows the residues found at positions a and d of the C-terminal half of Tm. These positions are occupied predominantly by "stabilizing" amino acids (26 positions occupied by Leu, Ile, Val, Tyr, Met) or neutral amino acids (10 Ala and 2 Gln). The only two clearly destabilizing residues at positions a or d are Ser 186 and Glu 218 (Table III; no data were presented by Tripet et al. (51) for Cys residues). These two residues are found in the two most unstable fragments that we analyzed (Tm 143-235 and Tm 167-260 ) but are not found in Tm 220 -284(5OHW) or Tm 220 -284 . Furthermore, the only charged residue at a core position is Glu 218 . This is especially intriguing because Glu 218 also participates in a possibly disruptive intrahelical (i, iϩ4) charge-charge repulsion with residue Glu 222 . Therefore, this specific site in the tropomyosin structure may be particularly destabilized with respect to both ␣-helix formation and coiled-coil dimerization. Tripet et al. (51) also found that position a of the heptad repeat is relatively insensitive to whether or not the side chain is ␤-branched; that is, very little difference in stability was observed for Leu, Ile, and Val.  (Table III).
The helix-helix interaction of coiled-coils can also be stabilized or destabilized by interhelical charge-charge interactions between a residue (i) at position g and a residue (iϩ5) at position e of the heptad repeat (2). Table III lists the amino acids at positions e and g of the C-terminal half of Tm. Interestingly, a single putative interhelical repulsion between a pair of like charges was found, involving the negatively charged residues Asp 175 and Glu 180 at positions g and e, respectively. This repulsion may therefore contribute to the relative instabilities of fragments Tm 143-235 and Tm 167-260 compared with similar sized or smaller fragments lacking these residues (Tm 189 -284(5OHW) , Tm 189 -284 , Tm 220 -284(5OHW) , and Tm 220 -284 ) (Table III). Therefore, despite the difficult task to identify the main interactions responsible for the low stability in the central region of the molecule of Tm, we believe that we have identified a combination of factors which together may go a long way to account for our observations. These hypotheses could be tested in the future by site-directed mutagenesis.
The troponin complex is constituted of three subunits (TnC, TnI, and TnT), each of which can be divided into two functional domains. One domain is responsible for the regulatory properties, and the other is responsible for the contacts that maintain the subunits associated and bound to the actin filament (53,54). TnT mediates the binding of the troponin complex with Tm; the N terminus of TnT is extended along the C-terminal region of Tm including the overlap (head-to-tail) region between adjacent Tm molecules, whereas the C-terminal half of TnT interacts with TnI and TnC, which together interact with actin and the central region of Tm in a Ca 2ϩ -dependent manner (53)(54)(55)(56). Tm is a flexible molecule (13), and this flexibility is diminished when it interacts with the troponin complex, being partially recovered upon Ca 2ϩ binding (57). Ca 2ϩ -induced movement of Tm across the actin surface has been observed by electron microscopy (58), and we have previously demonstrated that when Ca 2ϩ binds to troponin, a change occurs in the microenvironment of a 5-hydroxytryptophan probe at position 122 of Tm (31), a residue not expected to be in direct contact with troponin. Therefore, a correlation between conformational stability and function of Tm may exist: local structural transitions may play a role in the binding of Tm to the Ca 2ϩ -sensitive domain of Tn as well as switching positions on the actin filament surface in response to Ca 2ϩ and myosin binding (13).
If the movement of Tm across the surface of actin in response to Ca 2ϩ is in fact part of the regulatory mechanism of muscle contraction, it may be expected that the last residues of the C-terminal region necessitate a relatively more stable coiledcoil structure to fulfill their role of association with TnT in a Ca 2ϩ -independent manner and perhaps in the maintenance of Tm-Tm head-to-tail interactions. On the other hand, the structural lability of the central part of the Tm molecule, which binds troponin in a Ca 2ϩ -dependent manner, may promote conformational changes in the coiled-coil which facilitate its repositioning along the thin filament (13). In fact, based on electron cryomicroscopy data, Narita et al. (59) recently presented a model in which Tm molecules adopt a kinked conformation along the thin filament in the absence of Ca 2ϩ . In their model, a major kink, stabilized by troponin binding to actin, was localized to half of the fourth and the whole fifth Tm ␣/␤ repeat, that is, somewhere between residues 145 and 204 of Tm. This is precisely the region we found to be particularly unstable in this study.
The results presented in this work suggest that the last 50 residues of Tm confer stability to the C-terminal half of the molecule, whereas the region comprised by residues 143-235 is particularly unstable. In addition to differences in stability, the presence or absence of residues 261-284, which are encoded by the alternatively spliced, striated muscle-specific exon 9a (258 -284) (60), has a strong influence on the mechanism of unfolding of this coiled-coil. The conformational states available to Tm may well be of importance to its physiological role in the control of muscle contraction.
After submission of this work, we became aware of the crystal structure of a C-terminal fragment of striated muscle ␣-tropomyosin determined by Li et al. (43). They found that although Tm residues 254 -262 (as well as the N-terminal GCN4 fusion) adopt a canonical coiled-coil structure, the helices formed by Tm residues 263-284 splay apart to such an extent that after position 267 the two helices do not contact one another but instead make tail-to-tail contacts with a symmetry-related molecule in the crystal. It is interesting to interpret our data in the light of this structure because the splayed region corresponds almost exactly to the residues we found to be particularly important for stability and the cooperativity of folding of Tm C-terminal fragments. These properties cannot be attributed to the formation of the tail-to-tail tetramers observed in the crystal because our AUC data for all fragments possessing residues 261-284 indicate the formation of stable dimers under the benign conditions from which the denaturation experiments were performed (and trimers in the presence of TFE). The Tm sequence between residues 261 and 269 shows significant deviations from the polarity pattern normally observed in a coiled-coil (43,61), and this may initiate the splaying of the two helices observed in the crystal structure (43). However, in the absence of tail-to-tail tetramerization, it is still not clear whether residues 270 -284 remain splayed in solution or reassociate to form a parallel coiled-coil. To remain splayed in solution in a manner similar to that observed in the crystal structure, the ␣-helical conformation of residues 261-284 would have to be particularly stable, as our data appear to suggest. In any case, it is difficult to explain the effect of these residues on stability and dimerization without considering their participation in a coiled-coil structure.