Bepridil opens the regulatory N-terminal lobe of cardiac troponin C

Yu Li1,2, Michael L. Love2,3, John A. Putkey4 and Carolyn Cohen2

1Biophysics and Structural Biology Program, Brandeis University, Waltham, Massachusetts 02454-9110, USA. 2Rosenstiel Basic Medical Sciences Research Center, Brandeis University, Waltham, Massachusetts 02454-9110, USA. 3Biochemistry Graduate Program, Brandeis University, Waltham, Massachusetts 02454-9110, USA. 4Department of Biochemistry and Molecular Biology, University of Texas Medical School, Houston, Texas 77030, USA.

Correspondence should be addressed to C.C. email: ccohen@brandeis.edu

Cardiac troponin C (cTnC) is the calcium-dependent switch for contraction in heart muscle and a potential target for drugs in the therapy of congestive heart failure. We have determined the crystal structure of cTnC, with three bound Ca2+ ions, complexed with the calcium-sensitizer bepridil, to 2.15 Å resolution. In contrast to apo- and 3Ca2+-cTnC, the drug-bound complex displays an open regulatory N-terminal lobe, similar to those of 4Ca2+ skeletal troponin C (sTnC) and cTnC bound to a C-terminal fragment of cardiac troponin I (residues 147 to 163) . The closing of the N-terminal lobe is sterically hindered by one of the three bound bepridils. Our results provide a structural basis for the Ca2+-sensitizing effect of bepridil and reveal the details of a distinctive two-stage mechanism for Ca2+-regulation by troponin C in cardiac muscle.

Contraction in skeletal and cardiac muscles is triggered by the binding of Ca2+ to the troponin C (TnC) component of the troponin complex1,2. TnC, like other members of the calmodulin (CaM) superfamily, consists of two lobes joined by a flexible central linker (see Fig. 1a legend). The C-terminal or "structural" lobe (C-lobe) comprises two EF-hand motifs (domains III and IV) that bind either Ca2+ or Mg2+. The N-terminal or "regulatory" lobe (N-lobe) also contains two EF-hands (domains I and II) and in skeletal TnC (sTnC), both domains of the N-lobe bind Ca2+ specifically. By contrast, in the N-lobe of cardiac TnC (cTnC), only domain II binds Ca2+. Domain I is non-functional due to the mutation of Ca2+ ligands (D29L, D31A) and a V28 insertion. The other two subunits of troponin are troponin I (TnI), which binds to F-actin and inhibits actomyosin ATPase, and troponin T (TnT), which links TnC and TnI to tropomyosin. In the relaxed state of muscle, the interaction between myosin and F-actin is blocked by tropomyosin, whose position on F-actin is determined by its linkage with TnI. The binding of Ca2+ to the N-lobe of TnC produces a structural change so that TnC interacts strongly with TnI. As a result TnI is released from actin, and tropomyosin moves from its blocking position on F-actin. Myosin is then able to interact with F-actin and start its catalytic cycle3,4.

The binding of Ca2+ has strikingly different structural consequences in the regulatory N-lobes of cTnC and sTnC. In sTnC, the N-lobe switches from a "closed"5,6 to an "open"7,8,9 conformation upon the binding of 2 Ca2+ ions. Consequently, a hydrophobic pocket is exposed as the binding site for skeletal TnI (sTnI). This Ca2+-dependent structural transition was first proposed by Herzberg et al.10 (the so-called "HMJ model") and later confirmed by the crystal and NMR structures of sTnC7,8,9. By contrast, NMR studies have shown that the N-lobe of cTnC opens only partially upon the binding of a Ca2+ ion to the single regulatory site (domain II)11,12 (Figs 2a and 2b). (Although the conformation of the N-lobe of 3Ca2+-cTnC has been described as "closed" in ref. 11, we prefer the term "partially open", in order to distinguish this structure from the fully closed conformation of the N-lobe of 2Ca2+-cTnC). Moreover, Li et al.13 have recently shown that a C-terminal fragment of cardiac TnI (cTnI, residues 147-163) induces the fully open conformation in the N-lobe in which a single Ca2+ ion is bound (see Figs 2b and 2c). Thus TnC-based Ca2+-regulation appears to have distinctive mechanisms in skeletal and cardiac muscles. Although the structures of sTnC, in both open and closed states, have been determined to high resolution by X-ray crystallography, no crystal structure of cTnC has previously been reported.

The essential role of cTnC in the regulation of contraction suggests that this protein is a logical target for drugs in the therapy of congestive heart failure. "Calcium-sensitizers" are a new class of inotropic drugs that enhance the responsiveness of the cardiac myofibril to calcium14. The effect of these agents on short or long term treatment of heart failure is uncertain. In fact, no agent has yet been developed that has high specificity as a calcium sensitizer of the thin filament. Several CaM antagonists bind directly to cTnC and are thus good model compounds for the design of calcium sensitizers15. For example, although bepridil (Fig. 1b) is not used clinically for congestive heart failure, this drug has been shown to increase both the Ca2+-affinity of cTnC and the activity of actomyosin ATPase in vitro16,17.

We have determined the crystal structure of 3Ca2+-cTnC complexed with bepridil to 2.15 Å resolution. Our results enable us to visualize the details of the cTnC-bepridil interactions and have implications for the design of cTnC-specific calcium sensitizers. The current structure also reveals how the special properties of cTnC lead to a distinctive, stepwise mechanism for the opening of its N-lobe.

Overall description of the structure

Bepridil-3Ca2+-cTnC displays a globular (46 ¥ 38 ¥ 35 Å) shape resembling that of CaM-trifluoperazine (TFP) and CaM-peptide complexes (Fig. 1a)18,19,20. The central helical linker is unwound at residues 87-91, producing a 108° bend that brings the lobes close together. The two lobes are related by pseudo-2-fold symmetry so that their hydrophobic pockets merge to form a channel largely occupied by two bepridil molecules (Bep1 and Bep2). A third bepridil (Bep3) is located at one end of the hydrophobic channel, near the N-terminal linker and helices B, C and E, and partially blocks solvent access to the interior of the molecule. The structure contains 3 Ca2+ ions, bound to domains II, III and IV. Despite the absence of Ca2+ in domain I, the N-lobe of the current structure reveals a strong similarity to the 2 Ca2+-bound, fully open N-lobe of chicken sTnC9, with an r.m.s.d. of 0.621 Å for backbone overlay. The C-terminal lobe of Bepridil-3Ca2+-cTnC is similar to that of rabbit 4Ca2+-sTnC8 (PDB entry 1tn4) with an r.m.s.d. of 0.608 Å.

Bepridil-induced changes in the N-lobe

The most important effect of bepridil on 3Ca2+-cTnC is the full opening of the N-lobe. No crystal structure is presently available for 3Ca2+-cTnC, so that our analysis relies on the comparison between the current structure and the NMR structure of the N-lobe of chicken 3Ca2+-cTnC11. In that study, the N-lobe hydrophobic pocket of 3Ca2+-cTnC (without bepridil) is largely unexposed, in contrast to that of 4Ca2+-sTnC. In the current study, upon the binding of bepridil, the A-B and C-D interhelical angles decrease from 138° and 108° to 92° and 89°, respectively. The N-lobe opens as helices B and C move as a pair away from helices A and D (Figs 2b and 2d) so that this bepridil-induced transition in cTnC resembles the Ca2+-induced transition in the N-lobe of sTnC5,6,7,8,9.

As in sTnC8,9, the opening of the N-lobe in cTnC is accompanied by a backbone rearrangement in helix B, although domain I of cTnC is non-functional (Fig. 3). In the solution study of the partially open N-lobe of 3Ca2+-cTnC11, helix B displays a kink in its N-terminal region (residues 37-40) (Fig. 3a). By contrast, in the current structure, residues 37-40 adopt a helical conformation, and the C-terminus of helix B is swung away from helix A (Fig. 3b). Several changes lead to the straightening of helix B upon the binding of bepridil. In the small b-sheet between the two domains, the Thr 38N-Gly 70O backbone hydrogen bond is broken. The carbonyl oxygens of Thr 38, Lys 39 and Glu 40 then form three hydrogen bonds with the amino groups of Gly 42, Lys 43 and Val 44, respectively, and a carboxylate oxygen of Glu 40 also forms a strong hydrogen bond with the hydroxyl group of Ser 37 (Og -Oe distance=2.35 Å), providing stabilization to the newly formed helical region. As shown in the NMR study by Gagné et al.21, in sTnC helix B is only straightened when the sidechain of a glutamate (Glu 41) in loop I moves towards a Ca2+ ion as a bidentate ligand. Thus it appears that in cTnC, the sidechain interaction between Glu 40 and Ser 37 plays a role related to that of Ca2+ ligation by Glu 41 in sTnC in the opening of the N-lobe.

The two N-lobe domains of cTnC, like those of sTnC8,9, are structurally coupled through the small b-sheet. The breaking of the 38N-70O hydrogen bond in the b-sheet not only contributes to the straightening of helix B, but also releases the conformational constraints on Gly 70 and Thr 71 in the middle of Ca2+-binding loop II. Changes in the phi-psi angles of these two residues then allow helix C to move in a concerted fashion with helix B. As a result, the C-D interhelical angle in bepridil-3Ca2+-cTnC is smaller than that in 3Ca2+-cTnC.

Binding sites for bepridil

Bep1 plays the key role in the stabilization of the open N-lobe. Its phenyl group is located at the bottom of the N-lobe pocket in a small hydrophobic "cavity" (see Table 1) that seems to be the primary anchoring site for this drug molecule. Similar cavities have been found in CaM and accommodate certain bulky groups of several drugs (e.g., the trifluoromethyl group of TFP)18,19,22. The benzyl group of Bep1 protrudes towards the N-terminal linker and has extensive van der Waal’s contacts with helices B and C. The two aromatic rings of Bep1 are inserted as a "wedge" between the A-D and B-C units, sterically hindering the closing of the N-lobe (Fig. 2d). Overall, Bep1 interacts with 13 hydrophobic residues in the N-lobe (Table 1) and shields 96Å2 of N-lobe apolar surface from the solvent (calculated with Insight ’95, Molecular Simulations Inc. ).

Bep2 and Bep3 appear to mimic the binding of TnI to the C-lobe. An overlay of the current structure and that of rabbit sTnC bound to the sTnI1-47 fragment23 shows that the phenyl ring of Bep2 takes the place of the sidechain of Met 21 of sTnI (equivalent to Met 53 of cTnI) in the small hydrophobic cavity at the bottom of the C-lobe (see Table 1 and Fig. 4a). Correspondingly, the binding of Bep2 results in a bend in helix F at Ile 119, which is not observed in 3Ca2+-cTnC. Consequently, Met 120 in the current structure (equivalent to Ile 121 of rabbit sTnC, see ref. 23 for numbering scheme) occupies the position of Phe 122 of sTnC and has contacts with both the phenyl and benzyl groups of Bep2. Bep3 makes contacts with several residues on helix E (Table 1), which have also been shown to interact with TnI23. In fact, both bepridil and cTnI induce backbone conformational changes in the C-terminus of helix F and the F-G linker24. Since Bep2 and Bep3 occupy the same region of the C-lobe as TnI, it appears that these two molecules probably would not bind to cTnC in the troponin complex. By contrast, Bep1 is not likely to be displaced by TnI (see below).

The compact shape of the current structure seems to be a direct result of drug binding. The positively charged pyrrolidine groups of the bepridils bring the N- and C-lobes close together by neutralizing the negative charges at their interface. Moreover, bepridil bridges the lobes through two ionic pairs: one between Bep1 pyrrolidine and Glu 96 (Fig. 1c) and the other between Bep3 pyrrolidine and Asp99. A similar effect on CaM has been observed for the inhibitor TFP, whose positively charged piperazine ring stabilizes a compact conformation in CaM18,19. It has also been suggested that CaM cannot bind to its target proteins in this compact form18. By contrast, cTnC, when bound to the N-terminal region of cTnI (residues 33-80), adopts a more extended conformation23,25 than that of the bepridil-3Ca2+-cTnC and of CaM-TFP18,19. Correspondingly, as noted above, bepridil appears to activate rather than inhibit cTnC.

Stepwise mechanism for the opening of the N-lobe of cTnC

Previous NMR studies have shown that the 2Ca2+, 3Ca2+ and cTnI147-163-bound states of cTnC assume three distinct conformations11,12,13. A comparison of these results with the current structure and with the crystal structures of sTnC5,6,8,9 reveals the details of a stepwise mechanism for the opening of the N-lobe of cTnC. In sTnC, the interhelical angles of domains I and II, which define the "openness" of the N-lobe, depend on the conformation of three groups of "hinge residues": Phe 29-Asp 30 at the C-terminus of helix A, Ser 38-Glu 41 at the N-terminus of helix B and Val 65-Asp 66 at the C-terminus of helix C8. In the 2Ca2+ or closed state of sTnC, the hinge residues display non-helical phi-psi angles. Two critical changes lead to the opening of the N-lobe upon the binding of Ca2+ (see refs 8-9 for the detailed mechanism in sTnC): in the small b-sheet between domains I and II, the first and the last of the 4 backbone hydrogen bonds are broken; the three hinge regions then adopt a helical conformation, leading to the lengthening of helices A, B and C. Similar changes also occur in cTnC, but in two steps. As shown by NMR, in apo- or 2Ca2+-cTnC, residues Phe 27-Val 28, Ser 37-Glu 40 and Val 64-Asp 65 all display non-helical torsion angles (Fig. 2a)12, and the small b-sheet is stabilized by 4 backbone hydrogen bonds (74N-34O, 36N-72O, 72N-36O and 38N-70O)26. A comparison of the NMR structures of apo- and 3Ca2+-cTnC11,12 indicates that Asp 65 and Glu 76 in loop II move towards the incoming Ca2+ ion as carboxylate ligands during the first step of the conformational change. Consequently, helix C is extended at the C-terminus to include Val 64-Asp 65 (Fig. 2b), and the first hydrogen bond (74N-34O) is broken. The size of the N-lobe hydrophobic pocket increases slightly as the C-D interhelical angle decreases by about 10°. It should be noted that although the binding of Ca2+ to domain II does not alter the A-B interhelical angle, it does seem to "prime" domain I for further changes: helix A is lengthened by two residues (Phe 27 and Val 28, see Fig. 2b) --- either because the sidechain of Phe 27 moves further into the slightly enlarged N-lobe pocket or because the constraints imposed by the 74N-34O hydrogen bond are released. When cTnI13 or bepridil binds during the second step, the last hydrogen bond (38N-70O) in the b-sheet is broken, and helix B is extended at the N-terminus to include Ser 37-Glu 40 (Figs 2c, 2d and 3). The N-lobe then opens fully.

In summary, the lengthening of helices A, B and C is required for the full opening of the N-lobe in both sTnC7,8,9,21 and cTnC11,12,13. In both cases, however, the lengthening of helix B (which has the longest hinge region) produces the largest increase in the hydrophobic surface area of the lobe. As mentioned earlier, in sTnC the change in helix B requires the movement of a glutamate sidechain towards the Ca2+ ion in domain I21. By contrast, helix B of cTnC cannot be extended in such a fashion because domain I of this isoform does not bind Ca2+. Moreover, in cTnC the binding of a single Ca2+ ion to domain II introduces relatively minor changes11,12, which, as described above, are nonetheless essential for "priming" the opening of the N-lobe. As shown in the NMR study by Li et al.13 and in the current work, the lengthening of helix B and the subsequent opening of the N-lobe is achieved by an agent such as cTnI or bepridil, which shields the N-lobe hydrophobic pocket from the solvent. This two-stage mechanism provides an extension to the HMJ model for sTnC10, in which the N-lobe opens upon the binding of two Ca2+ ions.

Structural basis for the calcium-sensitizing effect of bepridil

It is now clear how bepridil might increase the contractile force of the cardiac myofilament17. By maintaining the open conformation of the N-lobe, bepridil may enhance its affinity for cTnI and thus increase the activity of actomyosin ATPase. In such a scheme, bepridil and cTnI would bind to the N-lobe simultaneously. Support for this mechanism comes from a study by Lin et al., which showed that a synthetic peptide covalently linked to Met 45 (in the Bep1 binding site) had little effect on regulation27. Moreover, Kleerekoper et al. have recently identified three bepridil binding sites in cTnC and have proposed that a bepridil molecule bound to the site which includes Met 45, Met 60 and Met 80 would not be displaced by cTnI15. This prediction appears to be correct. In the current structure, Bep1 is in close contact with these three methionines (Table 1). Moreover, an overlay of bepridil-3Ca2+-cTnC and 3Ca2+-cTnC-cTnI147-16313 indicates that Bep1 and cTnI147-163 bind to different regions of the hydrophobic core of the N-lobe: the former to the bottom of the N-lobe pocket, and the latter mainly to the apolar surfaces of helices A and B (Fig. 4b). Due to the predicted displacement of Bep2 and Bep3 by TnI, it appears that the Ca2+-sensitizing effect of the drug resides in Bep1 alone. Biochemical studies have also indicated that bepridil increases the Ca2+-affinity of cTnC16,17. It is apparent that bepridil hinders the closing of the N-lobe, a step that accompanies the release of Ca2+ from domain II. Increased Ca2+-affinity, by itself, may also stimulate actomyosin ATPase at intermediate Ca2+ levels17.

Implications for the future design of calcium-sensitizing agents for cTnC

Bepridil belongs to a group of pharmacological agents originally used to inhibit CaM from binding to its target proteins28. Most of these drugs have a large apolar region and a positively charged group at least 3 atoms away19,29. The apolar region seems to be essential for the binding of these agents to the hydrophobic pockets of both cTnC and CaM. By contrast, the positively charged group is probably not required for cTnC-specific drugs, since the pyrrolidine ring of bepridil does not contribute to the opening of the N-lobe in the current structure. In fact, such a group, when bound to the N-lobe, might hinder the binding of the positively charged cTnI to cTnC. Moreover, because of the structural homology between cTnC and CaM, several drugs that bind to cTnC, including bepridil and TFP, have been shown to bind to CaM as well15. As noted above, such drugs with positively charged groups may trap CaM in an inactive form. To avoid such an inhibitory effect on CaM, neutral agents resembling bepridil could be designed to modulate the activity of cTnC. Although bepridil itself, as described earlier, is not a good pharmacological agent for heart failure, the current study provides insights into the features that may be important for the design cTnC-specific calcium-sensitizers.

Methods

Crystallization and Data Collection. The A-Cys (C35S, C84S) mutant30 of chicken cTnC studied here has the same functional properties as the wild type protein. Orthorhombic (P212121) crystals with unit cell dimensions a=36.27 Å, b=62.04 Å, c=62.04 Å were grown by vapor diffusion at 4 °C from a mixture of 4 ml of protein solution (1.2 mM cTnC, 6 mM CaCl2, 2.5 mM bepridil HCl, 10mM HEPES pH 7.2) and 4 ml of reservoir buffer (30% PEG3350, 20 mM MgCl2, 50 mM HEPES pH 7.2). The partial specific volume is 1.9 Å3/Da, with one molecule per asymmetric unit. Data were collected at 95K to 2.15 Å resolution using synchrotron radiation at CHESS (beamline A1, l=0.935 Å). Data were reduced with the programs DENZO and SCALEPACK31. (See Table 2 for data collection statistics).

Structure Determination. A molecular replacement solution was obtained with the program AMoRe32, using the C-lobe of rabbit sTnC8 and the 2Ca2+-bound N-lobe of avian sTnC9 as independent search models, with data from 8.0 to 3.5 Å resolution. The final correlation coefficient was 38.8 (R factor 0.473). Positional refinement using the program XPLOR33 (R factor 0.377, with data to 2.15 Å) yielded electron density maps (2|Fo|-|Fc| and |Fo|-|Fc|) which allowed manual rebuilding of the protein model with the program O34.  Two bepridil molecules were also modeled into the electron density. Subsequent maximum likelihood refinement and simulated annealing with the program CNS_SOLVE35 aided in the modeling of ordered waters and a third bepridil. Residues 1-4, 87-91, 125-126, and 161 are disordered and were not included in the final structure. Main chain dihedral angles of all non-glycine residues lie in allowed regions of the Ramachandran plot, with 92.4% in most favored regions.

Coordinates. Coordinates have been submitted to the Protein Data Bank (accession number 1dtl).

Acknowledgments

We are grateful to J. H. Brown, Z. Grabarek, M. N. G. James and L. S. Tobacman for helpful comments on the manuscript, and to W. Liu for expression and purification of cTnC. This study was supported by grants to C. C. from the NIH and the Muscular Dystrophy Association and to J. A. P. from the NIH and the Robert Welch Foundation.

1. Farah, C. S. & Reinach, F. C. The troponin complex and regulation of muscle contraction. FASEB. J. 9, 755-767 (1995).

2. Tobacman, L. S. Thin filament-mediated regulation of cardiac contraction. Annu. Rev. Physiol. 58, 447-481 (1996).

3. Leavis, P. C. & Gergely, J. Thin filament proteins and thin filament-linked regulation of vertebrate muscle contraction. CRC. Crit. Rev. Biochem. 16, 235-305 (1984).

4. Squire, J. M., & Morris, E. P. A new look at thin filament regulation in vertebrate skeletal muscle. FASEB. J. 12, 761-771 (1998).

5. Herzberg, O. & James, M. N. G. Refined crystal structure of troponin C from turkey skeletal muscle at 2.0 Å resolution. J. Mol. Biol. 203, 761-779 (1988).

6. Satyshur, K. A., Rao, S. T., Pyzalska, D., Drendel, W., Greaser, M. & Sundaralingam, M. Refined structure of chicken skeletal muscle troponin C in the two-calcium state at 2-Å resolution. J. Biol. Chem. 263, 1628-1647 (1988).

7. Gagné, S. M., Tsuda, S., Li, M. X., Smillie, L. B. & Sykes, B. D. Structures of the troponin C regulatory domains in the apo and calcium-saturated states. Nat. Struct. Biol. 2, 784-789 (1995).

8. Houdusse, A., Love, M. L., Dominguez, R., Grabarek, Z. & Cohen C. Structures of four Ca2+-bound troponin C at 2.0 Å resolution: further insights into the Ca2+-switch in the calmodulin superfamily. Structure 5, 1695-1711 (1997).

9. Strynadka, N. C., Cherney, M., Sielecki, A., Li, M. X., Smillie, L. B. & James M. N. G. Structural details of a calcium-induced molecular switch: X-ray crystallographic analysis of the calcium-saturated N-terminal domain of troponin C at 1.75 Å resolution. J. Mol. Biol. 273, 238-255 (1997).

10. Herzberg, O., Moult, J. & James, M. N. G. A model for the Ca2+-induced conformational transition of troponin C, a trigger for muscle contraction. J. Biol. Chem. 261, 2638-2644 (1986).

11. Sia, S. K., Li, M. X., Spyracopoulos, L., Gagné, S. M., Liu, W., Putkey, J. A. & Sykes, B. D. Structure of cardiac muscle troponin C unexpectedly reveals a closed regulatory domain. J. Biol. Chem. 272, 18216-18221 (1997).

12. Spyracopoulos, L., Li, M. X., Sia, S. K., Gagné, S. M., Chandra, M., Solaro, R. J. & Sykes, B. D. Calcium-induced structural transition in the regulatory domain of human cardiac troponin C. Biochemistry 36, 12138-12146 (1997).

13. Li, M. X., Spyracopoulos, L. & Sykes, B. D. Binding of cardiac troponin-I147-163 induces a structural opening in human cardiac troponin-C. Biochemistry 38, 8289-8298 (1999).

14. Consensus recommendations for the management of chronic heart failure. On behalf of the membership of the advisory council to improve outcomes nationwide in heart failure. Am. J. Cardiol. 83(2A), 1A-38A (1999).

15. Kleerekoper, Q., Liu, W., Choi, D. & Putkey, J. A. Identification of binding sites for bepridil and trifluoperazine on cardiac troponin C. J. Biol. Chem. 273, 8153-8160 (1998).

16. MacLachlan, L. K., Reid, D. G., Mitchell, R. C., Salter, C. J. & Smith, S. J. Binding of a calcium sensitizer, bepridil, to cardiac troponin C. A fluorescence stopped-flow kinetic, circular dichroism, and proton nuclear magnetic resonance study. J. Biol. Chem. 265, 9764-9770 (1990).

17. Solaro, R. J., Bousquet, P. & Johnson J. D. Stimulation of cardiac myofilament force, ATPase activity and troponin C Ca++-binding by bepridil. J. Pharmacol. Exp. Ther. 238, 502-507 (1986).

18. Vandonselaar, M., Hickie, R. A., Quail, J. W. & Delbaere, L. T. Trifluoperazine-induced conformational change in Ca(2+)-calmodulin. Nat. Struct. Biol. 1, 795-801 (1994).

19. Cook, W. J., Walter, L. J. & Walter, M. R. Drug binding by calmodulin: crystal structure of a calmodulin-trifluoperazine complex. Biochemistry 33, 15259-15265 (1994).

20. Meador, W. E., Means, A. R. & Quiocho, F. A. Target enzyme recognition by calmodulin: 2.4 Å structure of a calmodulin-peptide complex. Science 257, 1251-1255 (1992).

21. Gagné, S. M., Li, M. X. & Sykes, B. D. Mechanism of direct coupling between binding and induced structural change in regulatory calcium binding proteins. Biochemistry 36, 4386-4392 (1997).

22. Osawa, M., Swindells, M. B., Tanikawa, J. Tanaka, T., Mase, T. & Furuya, T. Solution structure of calmodulin-W-7 complex: the basis of diversity in molecular recognition. J. Mol. Biol. 276, 165-176 (1998).

23. Vassylyev, D. G., Takeda, S., Wakatsuki, S., Maeda, K. & Maeda, Y. Crystal structure of troponin C in complex with troponin I fragment at 2.3-Å resolution. Proc. Natl. Acad. Sci. USA. 95, 4847-4852 (1998).

24. Gasmi-Seabrook, G. M. C., Howarth, J., W., Finley, N., Abusamhadneh, E., Gaponenko, V., Brito, R. M. M., Solaro, R. J. & Rosevear, P. R. Solution structures of the C-terminal domain of cardiac troponin C free and bound to the N-terminal domain of cardiac troponin I. Biochemsitry 38, 8313-8322 (1999).

25. Kleerekoper, Q., Howarth, J. W., Guo, X., Solaro, R. J. & Rosevear, P. R. Cardiac troponin I induced conformational changes in cardiac troponin C as monitored by NMR using site-directed spin and isotope labeling. Biochemistry 34, 13343-13352 (1995).

26. Spyracopoulos, L., Gagné, S. M., Li, M. X. & Sykes, B. D. Dynamics and thermodynamics of the regulatory domain of human cardiac troponin C in the apo- and calcium-saturated states. Biochemistry 37, 18032-18044 (1998).

27. Lin, X., Doston, D. G. & Putkey, J. A. Covalent binding of peptides to the N-terminal hydrophobic region of cardiac troponin C has limited effects on function. J. Biol. Chem. 271, 244-249 (1996).

28. Itoh, H., Ishikawa, T. & Hidaka, H. Effects on calmodulin of bepridil, an antianginal agent. J. Pharmacol. Exp. Ther. 230, 737-741 (1984)

29. Prozialeck, W. C. & Weiss, B. Inhibition of calmodulin by phenothiazines and related drugs: structure-activity relationships. J. Pharmacol. Exp. Ther. 222, 509-516 (1982).

30. Putkey, J. A., Dotson, D. G. & Mouawad, P. Formation of inter- and intramolecular disulfide bonds can activate cardiac troponin C. J. Biol. Chem. 268, 6827-6830 (1993).

31. Otwinowski, Z. & Minor, W. Processing of X-ray diffraction data collected in oscillation mode. In Methods in Enzymology (eds Carter, C. W. J. & Sweet, R. M.) 307-326 (Academic Press, San Diego; 1997).

32. Navaza, J. & Saludjian, P. AMoRe: an automated molecular replacement program package. In Methods in Enzymology (eds Carter, C. W. J. & Sweet, R. M.) 581-594 (Academic Press, San Diego; 1997).

33. Brünger, A. T. X-PLOR Version 3.1: A System for X-ray Crystallography and NMR. (Yale University Press, New Haven; 1993).

34. Jones, T. A., Zou, J. Y., Cowan, S. E. & Kjeldgaard, M. Improved methods for binding protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110-119 (1991).

35. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J-S, Kuszewksi, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., Warren, G. Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. D 54, 905-921 (1998).

Figure Legends

Fig. 1 The overall structure of bepridil-3Ca2+-cTnC. a, Stereo view of bepridil-3Ca2+-cTnC. cTnC comprises an N- and a C-terminal lobe. Each lobe consists of two helix-loop-helix EF-hand domains connected by a linker. The nomenclature is as follows: domain I (red) starts with helix A, followed by loop I, then helix B followed by the N-terminal linker (cyan); domain II (green) starts with helix C, followed by loop II, then helix D followed by the central linker (not included in the current structure due to disorder); domain III (yellow) starts with helix E, followed by loop III, then helix F followed by the C-terminal linker (purple and discontinuous at residues 125-126 due to disorder); domain IV (blue) starts with helix G, followed by loop IV, then helix H. There is also a helix N (black) N-terminal to domain I, which is present in TnC but not in CaM. There are two small anti-parallel b-sheets in cTnC: one between loops I and II; and the other between loops III and IV. Three Ca2+ ions, bound to domains II, III and IV, are shown as orange spheres. Three bepridils (balls and sticks) are bound to cTnC: Bep1 mainly to the N-lobe, Bep2 mainly to the C-lobe and Bep3 to both lobes. b, Ball-and-stick representation of the calcium-sensitizing agent bepridil. c, Electron density around Phe 27 (upper left), Bep1 (center) and Glu 96 (lower right) in a (2|Fo|-|Fc|) map contoured at 1.0s. The positively charged pyrrolidine ring of Bep1 forms a salt bridge with the sidechain of Glu 96.

Fig. 2 An overview of the stepwise mechanism for the opening of the N-lobe of cTnC. The N-lobes of a, Apo- or 2Ca2+-cTnC12, b, 3Ca2+-cTnC11, c, 3Ca2+-cTnC-cTnI147-16313 and d, bepridil-3Ca2+-cTnC are shown as ribbons. The A-D and B-C units are colored in grey and in black, respectively. Ca2+, bepridil and cTnI147-163 are shown as spheres, balls-and-sticks and ribbons, respectively. The "hinge" regions (residues 27-28, 37-40 and 64-65) are colored in red when in a non-helical conformation, and in yellow when in a helical conformation.

Fig. 3 A detailed view of the bepridil-induced opening of the N-lobe. a, In 3Ca2+-cTnC11, helix B displays a kink due to the non-helical torsion angles of residues 37-40 (yellow). b, The binding of bepridil results in a rearrangement of the hydrogen bonding pattern in the central b-sheet and in the N-terminal region of helix B. Consequently, helix B is extended at the N-terminus to include residues 37-40 (cyan) and its C-terminus is swung away from helix A. Note that the loss of the 38N-70O hydrogen bond also leads to a small decrease in the C-D interhelical angle. Bep1 is not shown in the figure.

Fig. 4 Comparison of the binding of bepridil and TnI to cTnC. a, Backbone overlay of the C-lobes of bepridil-3Ca2+-cTnC (red) and 2Ca2+-sTnC-sTnI1-47 (green)23 shows that Bep2 (dark red) mimics the binding of TnI (light green) to the C-lobe. Most notably, the pyrrolidine and phenyl groups of Bep2 occupy the same sites as the sidechains of Lys 18 and Met 21 of sTnI (light green balls and sticks), respectively. b, By contrast, Bep1 (dark red) and cTnI (sky blue) bind to different regions of the N-lobe pocket, as shown in the overlay of the N-lobes of bepridil-3Ca2+-cTnC (red) and 3Ca2+-cTnC-cTnI147-163 (blue)13.

_____________________________________________________________

Table 2. Residues involved in contacts (< 4.5 Å) with bepridil1

____________________________________________________________

Bep1 N-lobe F27, I36, L41, V44, M45, L48, M60, I61, V64

,V72, F77, M80, M81, S84

C-lobe E96, L97, L100

Bep2 N-lobe F20, A23, I26, F27, F77, M81

C-lobe L100, F104, I112, L117, M120, L136, F153, F156

,M157

Bep3 N-lobe M45, M50, P52, E56, M60

C -lobe E96, D99, L100, R102, M103

____________________________________________________________

1Residues whose sidechains form the hydrophobic "cavity" at the bottom of each lobe are underlined.

Residues that form salt bridges with the pyrrolidine rings of the bepridils are in bold.

____________________________________________________________________

________________________________________________________________

Table 2 Data collection and refinement statistics1

________________________________________________________________

Data collection

Resolution (Å) 2.15

Measured reflections 54,641

Unique reflections 7,998

Completeness (%) 99.7 (99.2)

Rmerge2 0.066 (0.126)

Refinement

Resolution range (Å) 30.0-2.15

No. of reflections 7,829

Sigma cutoff None

Rcryst3 0.213

Rfree4 0.276

No. of protein atoms 1185

No. of bepridil atoms 81

No. of Ca2+ atoms 3

No. of water molecules 136

r.m.s.d bond lengths (Å) 0.017

r.m.s.d angles (°) 1.48

_____________________________________________________________________

1Values for the highest resolution bin (2.23-2.15 Å) are shown in parentheses.

2Rmerge = Shkl|I - <I>|/SIhkl

3Rcryst = Shkl||Fobs| - |Fcalc||/Shkl|Fobs|, where Fcalc and Fobs are the calculated and observed structure factor amplitudes, respectively.

4Rfree is the same as Rcryst except that the summation is over 5% of the reflections, which were randomly selected and excluded from the refinement33.

________________________________________________________________________