Switch Action of Troponin on Muscle Thin Filament as Revealed by Spin Labeling and Pulsed EPR*

We have used pulsed electron-electron double resonance (PELDOR) spectroscopy to measure the distance between spin labels at Cys133 of the regulatory region of TnI (TnI133) and a native or genetically substituted cysteine of TnC (TnC44, TnC61, or TnC98). In the +Ca2+ state, the TnC44-TnI133-T distance was 42 Å, with a narrow distribution (half-width of 9 Å), suggesting that the regulatory region binds the N-lobe of TnC. Distances for TnC61-TnI133 and TnC98-TnI133 were also determined to be 38 Å (width of 12 Å) and 22 Å (width of 3.4 Å), respectively. These values were all consistent with recently published crystal structure (Vinogradova, M. V., Stone, D. B., Malanina, G. G., Karatzaferi, C., Cooke, R., Mendelson, R. A., and Fletterick, R. J. (2005) Proc. Natl Acad. Sci. U.S.A. 102, 5038–5043). Similar distances were obtained with the same spin pairs on a reconstituted thin filament in the +Ca2+ state. In the −Ca2+ state, the distances displayed broad distributions, suggesting that the regulatory region of TnI was physically released from the N-lobe of TnC and consequently fluctuated over a variety of distances on a large scale (20–80 Å). The interspin distance appeared longer on the filament than on troponin alone, consistent with the ability of the region to bind actin. These results support a concept that the regulatory region of TnI, as a molecular switch, binds to the exposed hydrophobic patch of TnC and traps the inhibitory region of TnI away from actin in Ca2+ activation of muscle.

N-lobe and C-lobe). A number of events occur when the Ca 2ϩ concentration is changed, as suggested mainly from biochemical studies. For one, Ca 2ϩ binding to the N-lobe of TnC induces the exposure of a hydrophobic pocket (5,(7)(8)(9)(10)(11). When this hydrophobic pocket is open in the ϩCa 2ϩ state, the switch segment (amino acids 116 -131) in the regulatory region (amino acids 116 -182) of TnI binds to the N-lobe (12)(13)(14)(15). Upon removal of Ca 2ϩ , the hydrophobic pocket of the N-lobe closes, and the switch segment moves out of the N-lobe of TnC. Finally, two regions (amino acids 104 -115 and 140 -182) of TnI bind actin, resulting in the inhibition of actomyosin ATPase (16 -20).
Crystal structures of the Tn core domain have been solved for human cardiac (21) and chicken skeletal troponin molecules (22). The Tn core domain has two characteristic subdomains: 1) the "regulatory head," composed of the N-lobe of TnC and the C-terminal region of TnI, and 2) the "I-T arm," composed of long coiled-coil regions from TnT2, TnI, and the C-lobe of TnC. The relative orientation of the regulatory head and the I-T arm may play a fundamental role in the regulatory mechanism of troponin. The optimal docking positions could be ascertained by fitting to a three-dimensional reconstruction from electron microscopic images of thin filaments (23,24). Alternatively, the best orientation or the best spatial relationships could be identified by fitting to polarized fluorescence data from bifunctional rhodamine probes for TnC in muscle fibers (25,26) or fluorescence resonance energy transfer (FRET) between probes attached to various Tn subunits (TnC, TnI, or TnT) on the thin filament (27).
Continuous wave electron paramagnetic resonance (CW-EPR) was used for pioneering studies on spin-labeled troponin (28 -30). Recently, we measured CW-EPR spectra from a spin label located on the Cys 133 residue next to the switch segment in the regulatory region of TnI to identify Ca 2ϩ -induced structural states (31) based on the sensitivity of spin-label mobility to flexibility and tertiary contact with a polypeptide (32). The slow spin-label mobility observed for the thin filament in the ϪCa 2ϩ state indicated tertiary contact of Cys 133 with actin because similar slow mobility was found for TnI-actin and TnI-tropomyosin-actin filaments lacking TnC, TnT, or tropomyosin. Furthermore, using nuclear magnetic resonance (NMR) and electron microscopic reconstruction, Murakami et al. (23) determined the structure of the C-terminal portion (131-182) of the regulatory region, including this cysteine, which binds to actin in the ϪCa 2ϩ state.
Global movement of Cys 133 of TnI in the complex and on the actin thin filament has been investigated by FRET and chemical cross-linking. Tao et al. (34,35) measured the average distance between TnI and TnC by FRET and found that the segment near the Cys 133 region of TnI was located near TnC in the ϩCa 2ϩ state but moved away in the ϪCa 2ϩ state. This result was supported by Ca 2ϩ -dependent cross-linking between Cys 133 and TnC (36,37). Using FRET, Miki et al. (38,39) also found that Cys 133 of TnI moved relative to Gln 41 , Lys 61 , and Cys 374 of actin on the thin filament. However, the observed changes in the average distance were relatively small (ϳ5 Å), suggesting that Cys 133 of TnI is released from the N-lobe of TnC and moves toward the actin surface. However, the N-lobe of TnC is close to the actin surface and remains near Cys 133 of TnI in the ϪCa 2ϩ state. Recently, CW-EPR and PELDOR (or DEER) spectroscopy have been used to investigate the distance between intra-and intersubunit sites in human cardiac binary TnC-I complexes and in reconstituted muscle fibers (10,11,40). These studies have verified the structures of the TnC monomer (10,11) and binary TnC-I complex (40) derived from x-ray crystallography and NMR studies. Additionally, the Ca 2ϩ -induced conformational transition of TnC in the binary complex and in fibers was determined (10,11). Here, we measured the distance distribution between Cys 133 in TnI and several residues in TnC using PELDOR to investigate the Ca 2ϩ -dependent structural changes in reconstituted thin filaments, which may be physiologically relevant. Our results suggest that the switch segment near Cys 133 of TnI is physically released from the N-lobe of TnC and fluctuates over a range of distances on a large scale (Ͼ50 Å) in the absence of actin and that this segment moves further away and presumably interacts with an actin surface in the thin filament.
Protein Preparation-Rabbit muscle protein was prepared from back and leg muscles, and chicken muscle protein was prepared from breast muscles. Actin was prepared from acetone powder of rabbit skeletal muscles, according to the method described by Spudich and Watt (41). Native troponin complexes (TnC-I-T complexes) and tropomyosin were prepared from rabbit and chicken skeletal muscle residues after extraction of myosin by methods described by Ebashi et al. (42). After dialysis with 20 mM Tris-HCl (pH 7.5), 30 mM KCl, and 0.1 mM CaCl 2 (buffer A) with 1 mM dithiothreitol (DTT), troponin was purified using a Q-Sepharose Fast Flow column (2.5 ϫ 10 cm) and eluted with a linear 30 -400 mM KCl gradient. Fractions containing TnC-I-T complexes were identified by measuring absorbance at 280 nm and by SDS-PAGE, collected, and finally dialyzed against buffer A containing 1 mM DTT.
Single cysteine mutant proteins of chicken skeletal TnC (G44C and A61C) were expressed and purified according to the method described by Nakamura (11). G44C and A61C were mutated using a Mutan-Super Express Km kit (Takara Bio Co.). The TnC gene fragments were ligated into pET15b vectors and used to transform competent Escherichia coli cells. Cells were grown at 37°C in LB medium with 0.5 mg/ml ampicillin and chloramphenicol. Cells were then induced with 1.0 mM isopropyl-␤-D-thiogalactopyranoside and incubated for 4 -5 h. Cells collected by centrifugation were sonicated in 50 mM Tris-HCl (pH 7.5) with 1 mM EDTA and 1 mM DTT. Next, ammonium sulfate (60% w/v) was added to the supernatant in 50 mM Tris-HCl (pH 7.5), 2 mM EDTA, and 1 mM DTT and centrifuged at 15,000 ϫ g for 10 min. After the supernatant was dialyzed against 10 mM Tris-HCl (pH 7.5) with 2 mM EDTA and 1 mM DTT, mutant TnC was purified using a Q-Sepharose Fast Flow column (2.5 ϫ 10 cm) and eluted with a linear 30 -900 mM KCl gradient. Fractions containing TnC were identified by measuring absorbance at 280 nm and by SDS-PAGE, collected, and finally dialyzed against buffer A containing 1 mM DTT.
Spin Labeling of the Troponin Complex and Isolation of Subunits-To remove DTT, the protein solution was dialyzed against buffer A. Troponin complexes were reacted with an equimolar amount of MSL overnight in the dark at 4°C. To remove unreacted MSL, the sample was dialyzed exhaustively against buffer A. To confirm the reaction and the removal of unreacted MSL, the dialyzed solution was measured by EPR.
MSL-labeled TnI (MSL-TnI), TnC, and TnT were isolated from MSL-Tn complexes as per Ojima and Nishita (43). MSL-Tn was dialyzed against urea buffer (6 M urea, 30 mM KCl, 1 mM EDTA, and 10 mM Tris-HCl (pH 7.6)) and applied to a CM-Toyopearl 650 M column (2.5 ϫ 10 cm). Following preequilibration with the same buffer to elute isolated TnC, MSL-TnI and TnT were eluted with a linear gradient of 30 -450 mM KCl in succession. Fractions containing TnC, MSL-TnI, and TnT were identified by measuring absorbance at 280 nm and by SDS-PAGE. TnC could be isolated almost completely from labeled TnI and TnT. Because TnT has no cysteine, the spin label cannot attach to TnT. Actually, a TnT-rich fraction showed a very weak EPR signal. Furthermore, a weak or no EPR signal was observed with TnI and TnT fractions isolated from the chicken MSL-Tn complex where TnI has Asn 133 (22). These results confirmed that Cys 133 of TnI was exclusively spin-labeled in the Tn complex from the rabbit (31,44). Therefore, we did not try further to separate TnT from labeled TnI fractions. Samples were stored in urea isolation buffer. Isolated TnC was used for spin labeling of Cys 98 .
Spin Labeling of Single Cysteines on TnCs-The mutated TnCs and TnCs isolated from native troponin complexes were dialyzed against buffer A to remove DTT. TnCs were reacted with a 10-fold molar excess of MSL overnight in the dark at 4°C. To remove unreacted MSL, the solution was dialyzed against buffer A twice. The dialyzed sample was measured by EPR to ensure that it did not show a sharp absorption line due to free mobile spin labels.
Reconstitution of the Doubly Labeled Troponin Complex-MSL-TnI and TnT isolated from rabbit MSL-troponin complexes were mixed with MSL-labeled mutant TnC and incubated at 20°C for 1 h in urea buffer (6 M urea, 1 M KCl, 0.1 mM CaCl 2 , and 20 mM Tris HCl (pH 7.6)). For reconstitution of doubly labeled troponin, this solution was gradually dialyzed to remove urea and KCl and finally dialyzed against buffer A. The reconstituted doubly labeled troponin, was purified using a Q-Sepharose Fast Flow column (2.5 ϫ 10 cm) and eluted with a linear 30 -400 mM KCl gradient. Fractions containing TnC-I-T complexes were identified by measuring absorbance at 280 nm and by SDS-PAGE and subsequently pooled. The collected solution was dialyzed against buffer A. For EPR measurement, the doubly labeled troponin solution was concentrated to ϳ100 M. This solution represented the ϩCa 2ϩ state without any further treatment or the ϪCa 2ϩ state following the addition of 5 mM EGTA.
Reconstitution of Thin Filaments-G-actin was polymerized in 20 mM imidazole-HCl (pH 7.0) with 100 mM KCl and 4 mM MgCl 2 for 30 min. Tropomyosin and MSL-Tn were added in excess to the F-actin solution (ϳ100 M, 0.3-0.5 ml) in the presence of 1 mM CaCl 2 (ϩCa 2ϩ state) or 5 mM EGTA (ϪCa 2ϩ state) and incubated for an additional hour. The molar ratio was set at 1:1:3 for tropomyosin, MSL-Tn complexes, and actin. Then, the mixture was centrifuged at 100,000 ϫ g for 30 min, and the pellet cosedimented with MSL-Tn was suspended in a small volume (20 - ATPase Assay-The inhibitory activity of troponin was measured using an actomyosin ATPase assay, as described previously (31). The ATPase reaction was monitored by NADH absorbance using a pyruvate-lactate dehydrogenase-coupled system (45). Inhibitory activity was defined as [ATPase rate (ϩCa 2ϩ ) Ϫ ATPase rate (ϪCa 2ϩ )]/ATPase rate (ϩCa 2ϩ ).
CW-EPR Measurements-EPR spectroscopy was performed using a Bruker ELEXSYS E500 spectrometer equipped with a dielectric resonator. Samples (ϳ15 l) were loaded into capillaries (inside diameter, 1.0 mm) and inserted into the resonator, and EPR spectra were acquired using 1-G field modulation amplitude at 100 kHz and 5-mW incident microwave power at 25°C.
PELDOR Measurements-We used pulsed electron-electron double resonance to measure the distance between two spin labels. PELDOR was performed with an ELEXSYS E580 FT-EPR spectrometer (Bruker Biospin) according to Nakamura et al. (11). Doubly labeled troponin complex samples (ϳ100 l) containing 30% glycerol were loaded into quartz tubes (inside diameter, 4.0 mm). The measurement temperature was set at 80 K to lengthen the spin-spin relaxation time. The time trace of electron spin echo (ESE) intensity was measured and analyzed as described previously (11).

RESULTS AND DISCUSSION
Spin Labeling of the Cys 133 Residue of TnI and TnCs-We measured dipolar interactions between spin labels located at Cys 133 in the regulatory segment of TnI and a few sites (Cys 44 , Cys 61 , and Cys 98 ) in TnC. Rabbit skeletal TnI was labeled exclusively with MSL because Cys 133 , near the switch segment of TnI, is the most reactive cysteine in the complex state (44). The labeling efficiency for rabbit troponin, estimated from double integration of the spectrum, was Ͼ0.8 mol/mol troponin com-plex. On the other hand, single cysteine-labeled TnCs (designated as TnC44, -61, and -98) yielded labeling efficiencies and actomyosin ATPase inhibitory activities of Ͼ90%. Labeled TnCs were reconstituted with labeled TnI (designated as TnI133) and TnT. Doubly labeled troponin C-I-T complexes also showed Ca 2ϩ -dependent actomyosin ATPase activity (Ͼ80%). CW-EPR spectroscopy from doubly labeled troponin complexes showed Ca 2ϩ -dependent mobility changes similar to that of singly labeled troponin complexes labeled at Cys 133 of TnI, as reported previously (31). These results indicate that doubly labeled troponin complexes function similarly to the native complex. These samples are designated as TnC44-TnI133-TnT, TnC61-TnI133-TnT, and TnC98-TnI133-TnT.
Distances between Pairs of Spin Labels Attached to TnI and TnC in the ϩCa 2ϩ State-To investigate how the N-lobe of TnC changes or modulates its interaction with the regulatory region of TnI upon Ca 2ϩ binding, we measured the distances between Cys 133 in regulatory domain of TnI and three positions, Cys 44 , Cys 61 , and Cys 98 , of the N-lobe and central helix of TnC using PELDOR. The modulation of ESE from the TnC44-TnI133-TnT (Fig. 1a) or TnC61-TnI133-TnT (Fig. 1b) complex decayed monotonically to baseline within 0.8 s. This suggests that the distance is long and the distribution is relatively  APRIL 2, 2010 • VOLUME 285 • NUMBER 14 wide because the oscillation characteristics of a well defined distance distribution are absent. The modulation depth estimated by extrapolating the baseline to time 0 was an ideal value (ϳ0.15). This indicated no aggregation because nonspecific aggregation should show very short distances with a very rapid decay of echo intensity that would not be visible and yielded a small modulation depth. To obtain the best fit, single Gaussian distance distribution equations were applied. The average distance (Å) and its half-width of half-maximum (Å) are listed in Table 1. The distances and widths were 41.7 and 9.1 Å for TnC44-TnI133-TnT and 37.8 and 11.7 Å for TnC61-TnI133-TnT, respectively. For TnC98-TnI133-TnT (Fig. 1c), we found oscillation characteristics within 0.4 s, suggesting that the distance is short and the distribution is narrow. Analysis showed that the distance and distribution were 21.8 and 3.4 Å, respectively. Inclusion of a second Gaussian distribution analysis did not improve the fit.

Switch Action of Troponin
The distances measured by PELDOR were compared with those from crystal structures and previous distance measurements. In the crystal structure (ϩCa 2ϩ state), the distances between the ␣-carbons (C␣) of TnC44-TnI133 and TnC61-TnI133 are 30.24 and 30.34 Å, respectively (Protein Data Bank code 1YTZ). These distances were considerably shorter than the distances between spin labels. This disagreement can be explained by the lengths of the side chain and spin label (ϳ9 Å). TnC44 and TnC61 exist on opposite sides of Asn 133 of TnI, bound to the N-lobe of TnC. Furthermore, these side chains (especially that of TnC44) are located outside of the N-lobe. Because the distances of C␤-C␤ for TnC44/TnI133 are larger by ϳ3 Å than those of C␣-C␣ (Table 1), the side chains of TnC44 and TnI133 must extend in opposite directions, given that the bond length of C␣-C␤ is 1.53 Å. Therefore, the interspin distances according to PELDOR must be longer than that of C␤-C␤ by at most 18 Å and equal to 48 Å at maximum. In fact, the observed distance was 42 Ϯ 9 Å, as we expected. Taken together, the observed distances are consistent with those calculated from the crystal structure, taking into account the orientation of side chains and the spacer length of the spin label.
The large observed half-widths (ϳ10 Å) seem inconsistent with published crystal structure data (Protein Data Bank code 1YTZ). TnC44 and TnC61 are not thought to be flexible because the intramolecular distances in TnC have very narrow distributions (10,11). TnI133 is exposed and moves out from the switch segment, which is clutched by the N-lobe in the crystal structure. Therefore, the side chain of Cys 133 fluctuates thermally. Additionally, the backbone of the segment including Cys 133 is thought to swing toward Cys 44 of the N-lobe. The interspin distance in TnC98-TnI133-TnT was 21.8 Å. Because the direction of the side chain of Cys 98 is nearly parallel with that of TnI133, it is reasonable to assume that the distance between these residues is nearly identical to that observed in the crystal structure (23.1 Å). A narrow distance distribution (ϳ3.4Å) suggested that the swing direction of the region including Cys 133 was vertical with respect to TnC98. In summary, the segment adjacent to Cys 133 of TnI binds to the N-lobe of TnC in the ϩCa 2ϩ state as in the crystal structure, but the Cys 133 region protrudes and swings thermally in solution (see Fig. 3a, left panel). This conclusion is supported by our previous data (31). Spin label experiments indicated that TnI133 exhibited decreased mobility but remained mobile on a nanosecond time scale in the ϩCa 2ϩ state.
These values of the reconstituted filaments are similar to or only slightly different from those of troponin alone. Therefore, our results indicate that the bond between the N-lobe of TnC and the regulatory region of TnI is not much altered by reconstitution of the troponin complex on the thin filament in the ϩCa 2ϩ state. This conclusion was supported by a previous report, which showed that the moderate mobility of a spin label  at TnI133 was identical in the TnC-I-T complex alone and in reconstituted filaments in the Ca 2ϩ state (31). Therefore, in a reconstituted thin filament, the regulatory region of TnI binds the N-lobe of TnC in a similar way to the Tn complex alone in the ϩCa 2ϩ state (Fig. 3b, left panel). However, a small difference in the distances may explain why the Ca 2ϩ affinity is severalfold weaker for the filament than for the complex alone (46,47). Our previous spin-labeling study showed that the geometry of TnC differed slightly (by Ͻ1.5 Å) in the monomer and in the TnC-I complex, to explain their nearly 10-fold affinity difference (10).

Distances between Spin-labeled TnC and -I in the
ϪCa 2ϩ State-We also tried to measure the distance between the regulatory region of TnI and the N-lobe of TnC in the ϪCa 2ϩ state using PELDOR, to examine how they move in response to Ca 2ϩ removal. PELDOR spectra of doubly labeled troponin in the ϪCa 2ϩ state indicated a gradual decay in the modulation of ESE (Fig. 1). These results suggest that a majority of spins have either weak spin-spin interactions or none at all (distance Ͼ50 Å) and broad distance distributions. However, these parameters could not be determined solely by spectral fit. We can assume that the number of spin pairs formed in the complex in the ϩCa 2ϩ state is the same as the number in the ϪCa 2ϩ state because we used the same preparations. We also used the same depth factor for the ϪCa 2ϩ state as that obtained from the fit in the ϩCa 2ϩ state. To obtain the best fit, single Gaussian distance distribution fitting was applied to the data from TnC44-TnI133-TnT (Fig. 1a) or TnC98-TnI133-TnT (Fig. 1c) interactions in the complex. The best fit showed similar broad distributions. The distances and half-widths were 29.1 and 25.2 Å for TnC44-TnI133-TnT and 41.7 and 22.7 Å for TnC98-TnI133-TnT, respectively. For TnC44-TnI133-TnT, partitioning of a second population of widely separated (Ͼ80 Å, noninteracting) spins (56%) was necessary to obtain the best fit. Inclusion of a second Gaussian function also improved the fit of TnC61-TnI133-TnT (Fig. 1b). The first population (29%) had a distance and half-width of 25.9 and 8.6 Å, whereas the second (71%) had values of 68.2 and 31.0 Å. The result obtained by two-component fit did not alter the conclusion that the distance distribution between the N-lobe of TnC and the regulatory region of TnI is very broad in the ϪCa 2ϩ state. It is therefore likely that the TnI133 region dissociates from the N-lobe of TnC and swings around it upon Ca 2ϩ removal (Fig. 3a, right panel). This agrees well with previous data indicating that spin labels of TnI133 in the Tn complex is highly mobile in the ϪCa 2ϩ state (31).
A gradual decay in ESE modulation in PELDOR spectra was also observed for reconstituted thin filaments prepared by mixing spin-labeled TnC-I-T with tropomyosin and actin (TnC44-TnI133-TnT-Tm-actin 7 and TnC61-TnI133-TnT-Tm-actin 7 ). To obtain the best fit, single or double Gaussian component fitting was applied to the data obtained from TnC44-TnI133-TnT-Tm-actin 7 (Fig. 2a) and TnC61-TnI133-TnT-Tm-actin 7 (Fig. 2b). Broad distance distributions were obtained in both cases. However, the distributions were less uniform than that of the troponin complex alone in solution. Indeed, the population with distances of 60 -80 Å or larger was dominant in the reconstituted thin filaments. This suggests that the regulatory region around TnI133 is released from TnC, fluctuates around TnC, and stays for some time at a site far away on the thin filament (Fig. 3b, right panel). However, using NMR and electron microscopic image reconstruction, Murakami et al. (23) recently showed that the C-terminal region near TnI133 binds to the surface of actin in the ϪCa 2ϩ state. It is reasonable to resolve this discrepancy by assuming that the N-lobe of TnC is mobile and tumbling relative to the C-lobe of TnC and the actin surface (26,48), whereas the regulatory region of TnI is released from TnC and fixed on the actin surface. The flexibility of the N-lobe of TnC in the complex was reported by Blumenschein et al. (49) using NMR. In the crystal structure of chicken skeletal troponin, the central helix is melted in the ϪCa 2ϩ state, whereas it is folded in the ϩCa 2ϩ state (22). On the other hand, we previously found that a part of the spin labels of TnI133 in the ϪCa 2ϩ state were immobilized completely on actin alone, as well as on the thin filament, but the remaining population were still highly mobile or moderately immobilized as in the ϩCa 2ϩ state (31), suggesting that the regulatory region of TnI is in equilibrium between being fixed on the actin surface and freely approaching the N-lobe of TnC. It is therefore likely that in a response to Ca 2ϩ the N-lobe traps a favored conformation of the regulatory region that fluctuates between the N-lobe and a site Ͼ50 Å away on the actin surface.
The Relationship to Previous Distance Measurements-From FRET data derived from the ternary complex in the Ca 2ϩ state, Luo et al. (35) showed that the distance between TnC41 or 49 and TnI133 was 40 Å, which is very close to the distance we measured between TnC44 and TnI133 (42 Å). Tao et al. (34) also showed that the distance between TnC98 and TnI133 was at least 27 Å, which was a little larger than the value our experiments yielded (22 Å). Because an attached fluorescence label is larger in size than a nitroxide spin label, it might interfere sterically with a closer approach. This close proximity was supported by disulfide bond formation between TnC98 and TnI133 (36,37). The FRET studies described above have all shown that the average distances between TnI133 and the residues of TnC in the ternary complex were only ϳ5 Å larger in the ϪCa 2ϩ state than in the ϩCa 2ϩ state. The average distance could not be directly compared because PELDOR yielded a much broader and more complex distance distribution within the Tn complex in the ϪCa 2ϩ state than in the ϩCa 2ϩ state. In agreement with FRET data, our PELDOR results indicate that the center of the Gaussian distance distribution between TnC98 and TnI133 increased by 7 Å (from 22 to 29 Å).
Miki and coworkers (38,39) measured FRET between TnI133 and residue 41 or 374 of the nucleotide binding site of actin and showed that the average distances between them decreased by 5-15 Å upon Ca 2ϩ removal. Xing et al. (50) recently showed a shift in the distance distribution by 10 -15 Å between sites in TnI and residue 374 of actin in cardiac thin filaments using time-resolved FRET. Therefore, TnI133 moves inwards on the actin filament upon Ca 2ϩ removal. The present study showed that the distances between TnI133 and various TnC sites increased, and their distributions shifted by Ͼ20 Å upon removal of Ca 2ϩ from the thin filament. Taken together, it appears that TnI133 dissociates from TnC and then moves inward on the actin filament. The differences in the amplitude of shift and the width of the distribution could be explained by increased flexibility and outward movement of the free N-lobe of TnC in the ϪCa 2ϩ state.
Conformational Changes of TnI Regulatory Region (Switch Action)-Based on the data described above, we postulated a model of the TnI regulatory region (Fig. 3b). In the ϪCa 2ϩ state, the N-lobe of TnC releases the regulatory region and begins swinging thermally in Brownian motion. The regulatory region of TnI dissociates from the N-lobe of TnC, weakly binds to a site of actin surface, and may fluctuate between TnC and actin sites. Therefore, the interspin distance distributions were broad but dominated by long distances (Ͼ50 Å). This is a stand-by state for the binding of the regulatory region to TnC (Fig. 3b, right  panel). Next, when the N-lobe binds Ca 2ϩ , it opens a hydrophobic pocket, which allows trapping a switch segment (amino acids 116 -131) of the regulatory region of TnI. The distance between TnC and TnI133 showed a relatively narrow distribution in the ϩCa 2ϩ state, suggesting that the switch segment immediately adjacent to TnI133 was fixed to the N-lobe of TnC (Fig. 3b, left panel). The release of the C-terminal actin-binding portion (amino acids 140 -182) of the regulatory region from the actin surface induces activation of muscle contraction. It is a speculative but attractive possibility that the other actin-binding site (amino acids 104 -115) as well as the C-terminal binding portion of TnI are pulled up from the actin surface by the refolding and extension of the central helix of TnC (Fig. 3b).
When the N-lobe of TnC releases Ca 2ϩ , the hydrophobic pocket closes, and the affinity of TnI for the switch segment decreases. Then, the regulatory region begins tumbling and binds again to the actin surface. Therefore, the switch segment moves outward on the filaments upon Ca 2ϩ binding as postulated earlier by Miki et al. (39).
Although the Ca 2ϩ -induced shift of tropomyosin appears to be well established by electron microscopic studies (51-54), it has not yet been satisfactorily demonstrated by distance measurements using FRET (55)(56)(57). The present PELDOR measurements clearly demonstrate the existence of large spatial rearrangements within the troponin complex. This method would be useful to test tropomyosin movement along the thin filament. The shuttling of TnI between TnC and actin sites observed here may induce changes in the flexibility of tropomyosin, which may, in turn, induce changes in the flexibility of the actin filament (33,55,57). Alternatively, the C-terminal region of TnI may anchor the protein onto actin to pull tropomyosin laterally to the outer domain of the actin filament and retain the blocked state of tropomyosin ( Fig. 3b) (53).