Troponin Regulatory Function and Dynamics Revealed by H/D Exchange-Mass Spectrometry*

Muscle contraction is tightly regulated by Ca2+ binding to the thin filament protein troponin. The mechanism of this regulation was investigated by detailed mapping of the dynamic properties of cardiac troponin using amide hydrogen exchange-mass spectrometry. Results were obtained in the presence of either saturation or non-saturation of the regulatory Ca2+ binding site in the NH2 domain of subunit TnC. Troponin was found to be highly dynamic, with 60% of amides exchanging H for D within seconds of exposure to D2O. In contrast, portions of the TnT-TnI coiled-coil exhibited high protection from exchange, despite 6 h in D2O. The data indicate that the most stable portion of the trimeric troponin complex is the coiled-coil. Regulatory site Ca2+ binding altered dynamic properties (i.e. H/D exchange protection) locally, near the binding site and in the TnI switch helix that attaches to the Ca2+-saturated TnC NH2 domain. More notably, Ca2+ also altered the dynamic properties of other parts of troponin: the TnI inhibitory peptide region that binds to actin, the TnT-TnI coiled-coil, and the TnC COOH domain that contains the regulatory Ca2+ sites in many invertebrate as opposed to vertebrate troponins. Mapping of these affected regions onto the troponin highly extended structure suggests that cardiac troponin switches between alternative sets of intramolecular interactions, similar to previous intermediate resolution x-ray data of skeletal muscle troponin.

Muscle contraction is tightly regulated by Ca 2؉ binding to the thin filament protein troponin. The mechanism of this regulation was investigated by detailed mapping of the dynamic properties of cardiac troponin using amide hydrogen exchange-mass spectrometry. Results were obtained in the presence of either saturation or non-saturation of the regulatory Ca 2؉ binding site in the NH 2 domain of subunit TnC. Troponin was found to be highly dynamic, with 60% of amides exchanging H for D within seconds of exposure to D 2 O. In contrast, portions of the TnT-TnI coiled-coil exhibited high protection from exchange, despite 6 h in D 2 O. The data indicate that the most stable portion of the trimeric troponin complex is the coiled-coil. Regulatory site Ca 2؉ binding altered dynamic properties (i.e. H/D exchange protection) locally, near the binding site and in the TnI switch helix that attaches to the Ca 2؉ -saturated TnC NH 2 domain. More notably, Ca 2؉ also altered the dynamic properties of other parts of troponin: the TnI inhibitory peptide region that binds to actin, the TnT-TnI coiled-coil, and the TnC COOH domain that contains the regulatory Ca 2؉ sites in many invertebrate as opposed to vertebrate troponins. Mapping of these affected regions onto the troponin highly extended structure suggests that cardiac troponin switches between alternative sets of intramolecular interactions, similar to previous intermediate resolution x-ray data of skeletal muscle troponin.
Striated muscle contraction is reversibly activated and tightly controlled by Ca 2ϩ binding to the key regulatory protein of the contractile apparatus, troponin (Tn 2 , reviewed in Ref. 1,2). In vertebrate troponins, this regulation involves a Ca 2ϩ -dependent attachment of subunit TnC NH 2 domain to a ϳ10-residue, amphipathic, "switch" helix of subunit TnI (3)(4)(5). The switch helix is within a 9-kDa portion of TnI that appears to interact with actin and tropomyosin so as to shut off muscle contraction (6,7), specifically when Ca 2ϩ (and the switch helix) dissociate from the TnC regulatory domain.
Ca 2ϩ -saturated cardiac and skeletal muscle troponin structures each have been solved by x-ray crystallography at high resolution (4,5). Results agree with each other in many details, and the findings are important advances after many years when only much smaller portions of troponin had been solved at atomic resolution. On the other hand, the position of the TnC regulatory NH 2 domain differs between these two structures. In addition, the connecting residues between this domain and the remainder of troponin are not seen in the cardiac data. The mechanistic implications of these differences are unclear. In the presence of Mg 2ϩ /EGTA, an atomic model of skeletal muscle troponin has been hypothesized from 9 Å x-ray data (5). High resolution data have not been reported for either isoform in the presence of low Ca 2ϩ .
Several other aspects of the regulatory mechanism remain unknown. The inhibitory effects of TnI on the thin filament are not understood at atomic resolution, and are functionally complex (8 -12). Also, there is little understanding of the effects of regulatory Ca 2ϩ dissociation on parts of troponin outside the TnC NH 2 region. Finally, the regulatory action of many invertebrate troponins cannot be comprised by the mechanism stated above. These troponins have Ca 2ϩ binding sites only within the COOH domain of TnC (13,14), and none in the TnC NH 2 domain. Thus, there are major gaps in the current understanding of how troponins control muscle contraction.
The present work represents a new approach to this subject: amide hydrogen exchange (15)(16)(17)(18)(19) of troponin. Solvent-exposed amide hydrogens in peptides rapidly exchange with solution H (or D in D 2 O), with rates of ϳ10 s Ϫ1 (20). In contrast, exchange is blocked by hydrogen bonding, such as that of most backbone amide hydrogens in folded proteins. Exchange rates in proteins do not correspond to unfolding or refolding rates. Rather, they are very much slower. This is because, under usual conditions, local refolding rates greatly exceed the H/D exchange rate of the unfolded region. Exchange at the slowest exchanging amides tends to be governed, i.e. protected from the solvent-exposed rate, by global protein folding stability. Other amide hydrogens exhibit faster exchange, intermediate between those of solvent-exposed hydrogens and core hydrogens, reflecting wide variation in local flexibility or folding stability across different regions of an overall folded protein. The degree of H/D exchange protection is a measure of local folding stability within the context of a globally folded protein (15)(16)(17)(18)(19). By characterizing exchange rates at multiple sites, either by NMR or by mass spectrometry, the dynamic behavior of the protein can be mapped. Results loosely correlate with crystallographic B-factors, but reflect different protein properties and provide different information (21). Finally, by measuring the effects of physiological perturbations, such as Ca 2ϩ binding to troponin, on H/D exchange rates, one may detail intramolecular signal transduction.
In the present study, mass spectrometry is used to map amide hydrogen exchange in the human cardiac troponin core domain (TnC-TnI-TnT-(183-288)). Data for two conditions are reported: in the presence of either full saturation or subsaturation of the regulatory Ca 2ϩ binding site (site II), which is located in the TnC NH 2 domain. Troponin is a highly extended protein, ϳ20 nm in length (1), without a large globular region that might be highly protected from H/D exchange. Even the troponin "core domain," which includes TnC, most of TnI, and the COOH terminus of TnT, lacks a globular truly core-like region (4). Correspondingly, in the current study much of troponin is found to undergo H/D exchange rapidly. Exchange is completed fully within seconds or a minute of exposure to D 2 O. The most prominent exception to this pattern is within the 46-residue TnT-TnI coiled-coil portion of troponin. Parts of the coiled-coil are protected from H/D exchange for hours and evidently comprise the most tightly folded portion of the ternary troponin complex.
Current results also show that Ca 2ϩ saturation of site II affects H/D exchange rates across many parts of troponin. Effects are not confined to the TnC NH 2 domain and the TnI switch helix. Rather, Ca 2ϩ binding has long range allosteric effects. Based on these detailed results, we suggest that the intra-troponin switch regulating muscle contraction involves a conversion between two states of the troponin core domain. The states are stabilized by two different sets of intra-and intersubunit interactions, and involve the coiled-coil and TnC COOH domain as well as the region of troponin where Ca 2ϩ binds. This proposal is discussed in terms of previous structural studies of cardiac and skeletal muscle troponins. It gives added weight to a hypothesized atomic structure for skeletal muscle troponin in the presence of low Ca 2ϩ concentration, derived from 9 Å x-ray data (5).

EXPERIMENTAL PROCEDURES
Protein Preparation, Protein Fragment Generation, and Protein Fragment Identification-Human cardiac troponin was prepared by reconstitution of bacterially expressed (22) TnI, TnC, and COOH-terminal TnT-(183-288) (expression plasmid obtained by PCR of full-length TnT cDNA.). After 1:1:1 mixture and serial dialysis to remove initial denaturant and to decrease 1 M NaCl to 0.1 M NaCl, the complex was isolated via ion exchange chromatography using a 6-ml Resource S column connected to an FPLC (AKTA, Amersham Biosciences), and stored at Ϫ80°C.
To identify troponin fragments that could be examined for H/D exchange, 5 g of troponin in 50 l of 10 mM NaHPO 4 (pH 7.0), 0.1 M NaCl, was mixed with 50 l of 0.1 M NaHPO 4 (pH 2.5), followed by the addition of 5 g of porcine pepsin dissolved in 10 mM HCOONa (pH3.75). The digested sample (5 min 0°C) was injected into a micropeptide trap connected to C18 HPLC column (5 cm ϫ 1 mm, Vydac) attached to a Finnigan FT-ICRMS (Thermoelectron). Digested peptides were eluted at 90 l/min for tandem mass spectrometry. Peptic fragments were identified using the search algorithm Bioworks 3.2 (Thermoelectron) and manual confirmation.

H/D Exchange of Troponin and Subsequent Isotope Analysis by HPLC-Electrospray Ionization FT-ICR MS-
To initiate exchange, performed at 25°C, 5 g of troponin complex was transferred by Zeba Desalt Column (Pierce) into H 2 O buffer (10 mM NaHPO 4 (pH 7.1), 0.1 M NaCl), and then diluted to 5.9 M troponin by adding 9 parts (v/v) D 2 O buffer: 10 mM NaDPO 4 (pD read 7.1), 0.1 M NaCl, 0 or 1 mM CaCl 2 . Exchange was quenched after different time intervals by adding an equal volume of iced 100 mM NaDPO 4 (pD 2.5), 0.1 M NaCl, and quickly flash-freezing the sample in liquid nitrogen. Samples were stored at Ϫ80°C.
For analysis, samples were quickly thawed and digested with 5 g of pepsin (Sigma) on ice for 5 min followed by immediate injection into the micropeptide trap (Michrom Biosources, Inc) attached to C18 HPLC column connected to FT-MS. Peptides were eluted within 10 -12 min using a gradient of 5-35% acetonitrile at a flow rate of 90 l/min. Individual peptide envelopes were easily recognizable by correspondence to the possible post-exchange m/z of pre-H/D-identified peptides; 1/z Da increase per exchanged hydrogen. The centroid of each peptide was determined using the software package MagTran (23). To allow correction for backward D/H exchange during pepsin digest and HPLC-MS, fully deuterated control was prepared (and digested, and peptides assessed by MS) by incubation of troponin complex for 2 h at 40°C in 2 M deuterated urea, 100 mM NaDPO 4 (pD 2.5). Corrections for back exchange (D to H) during processing were determined for each peptide. The average back exchange was 23%.
Curve Fitting of Exchange Kinetic Data-Non-linear least squares curve fitting of peptide mass increases over time was performed with Scientist (Micromath). Peptides, subpeptides, and/or overlapping peptides were grouped for global curve fitting. This improved error estimates for some peptides, relative to the alternative approach of subtraction that is sometimes effective (24). (In this alternative, exchange in a subpeptide is subtracted from exchange in a larger, parent peptide, and the kinetics of the subtracted result are fit, with larger errors in some cases.) The globally fit groups of peptides were: TnC [25-56, 28 - Primary Structure Assignments of H/D Exchange Protection-For peptides from most of troponin (TnT, TnI, and the TnC COOH-terminal domain), the sizes of the observed H/D exchange transitions matched the peptide lengths or length overlaps. This allowed straightforward assignment of the transitions to specific residues, as shown under "Results and Discussion." In most of the remaining cases, a peptide (or nonoverlapping subpeptide region) contained both a detectable transition and other hydrogens that exchanged before the first 5-s time point. Assignment was again straightforward for these peptides, because the observed transition, which indicated detectable protection from exchange, could be attributed to the subset of residues that were folded sufficiently to be identifiable by x-ray crystallography. The primary exception to this method of transition assignment was in the TnC NH 2 domain, where there was no unambiguous basis for assignment of specific residue NHs to many of the kinetically identified transitions. For illustration of approximate mapping, the different rates were provisionally assigned to conform with previously published work, such as the troponin high resolution structural data and NMR studies of TnC.
The assigned exchange rates (k ex ) were converted to estimated protection factors (K cl ϭ k chem /k ex ), where k chem was the geometric average (range 1.8 -8.2) of the unprotected H/D exchange rates expected (20) for each residue in the peptide. These protection factors can be considered to represent local folding stability, measured in the context of the globally folded troponin molecule. In the figures, protection factors are mapped on the troponin primary and tertiary structures. However, these maps are altered very little if direct exchange rate values are substituted and mapped instead.
Ca 2ϩ Concentrations and Binding Saturation-Cardiac troponin sites III and IV bind Ca 2ϩ with affinity ϳ10 8 M Ϫ1 , the regulatory site II binds Ca 2ϩ with affinity ϳ10 6 M Ϫ1 , and site I is fully defunct (1,25). To perform H/D exchange with sites II-IV saturated was straightforward: 1 mM CaCl 2 was included. To specifically examine effects of site II Ca 2ϩ , we sought a condition in which site II was not saturated, but sites III and IV remained fully saturated. In principle, H/D exchange of some portions of troponin might be accelerated considerably by even minor subsaturation of sites III and IV. Therefore, to ensure that effects of lower Ca 2ϩ concentrations were due to changes in site II saturation only, our lower Ca 2ϩ condition remained in the M range. Adding mM Mg 2ϩ to an EGTA-chelated sample was an unused alternative, with the limitation that the Mg 2ϩ affinity of sites III and IV is not high enough to assure saturation and also that Mg 2ϩ is not Ca 2ϩ .
Experiments using a Ca 2ϩ -sensitive fluorescence probe showed that site II was mostly but not completely empty under our low Ca 2ϩ conditions. FuraFF (Invitrogen) spectra and Ca 2ϩ affinities (measured with fluorescence titration data fit to simple binding isotherms) were characterized and used to examine solution Ca 2ϩ concentrations. When troponin was transferred by spin column into the H 2 O buffer and then added in 1:10 part to the D 2 O buffer, the free Ca 2ϩ rose by 4 M, implying that 4 M Ca 2ϩ dissociated from 5.9 M troponin (the final protein concentration). Accordingly, troponin site II was at minimum 2/3 dissociated and at most 1/3 bound to Ca 2ϩ under these conditions, which replicated those used for H/D exchange measurements. This conclusion of substantial unsaturation of site II was confirmed by the observed differences in H/D exchange rates in the presence of low as opposed to high Ca 2ϩ . However, because site II was not completely empty under the low Ca 2ϩ conditions examined, the true effects on exchange rates may be larger than those detected in the present report.

RESULTS
Kinetic Results-Extensive (5 s to 6 h) H/D exchange kinetic data were obtained for 43 peptides derived by digestion of troponin exposed to D 2 O in the presence of saturating, 1 mM Ca 2ϩ . In separate experiments, kinetic H/D exchange data were obtained from 37 of these same 43 peptides, plus one other, derived by digestion of troponin exposed to D 2 O containing 4 M Ca 2ϩ . (This Ca 2ϩ concentration was subsaturating for regulatory Ca 2ϩ binding site II, under these experimental conditions.) The peptides are shown in Fig. 1. The high Ca 2ϩ peptides provided data concerning 89% of the amide NH groups in the experimental troponin, and the lower Ca 2ϩ peptides included 80% of the amide NH groups. The average peptide contained 15.5 exchangeable amide bonds, and overlapped in 60% of its length with one or more other peptides.
The exchange data (peptide mass increase versus duration of D 2 O exposure) spanned a wide time range. Results, therefore, could report transitions occurring over multiple timescales, from seconds to hours. This was in fact observed (Figs. 2 and 3, panels described in detail below). H/D exchange rates ranged across almost four orders of magnitude, from unprotected amide hydrogens that exchanged before the first, 5-s time point, to amide hydrogens that were unexchanged after 6 h in D 2 O. Often, there was more than one transition evident within an individual peptide. Note that all peptides reported in Figs. 2 and 3 begin at zero mass increase at zero time.
The availability of so much data, and over such a broad time range, raised the possibility of achieving a near-comprehensive, local mapping of troponin dynamics. Accordingly, the kinetic analysis was conducted to provide assessments that were as full and as precise as possible. Curve fitting was employed to determine transition rate constants and transition magnitudes. Because of the  wide time range examined, it was necessary when measuring rate constants to treat the data as the sum of exponentials, whenever the results so implied. Also, to optimize measurement of transitions, global curve fitting proved essential. Data from peptides and their subpeptides or overlapping peptides were combined and fit simultaneously. In the global fitting procedure, exchange rates of those NHs present in more than one peptide were required to be equal within all such peptides. Curve fitting was broadly successful for transition rate estimation, as shown in Figs. 2 and 3, and in Tables 1 and 2.
Fast Hydrogen Exchange Occurs in Large Portions of Troponin-Before considering the detailed data across troponin, it useful to note the broad pattern of the results. Fig. 4 shows colorcoded mapping of the degree of protection from H/D exchange. Results are delineated according to the troponin complex primary structure, and presented for both experimental conditions, high Ca 2ϩ and low Ca 2ϩ . The largest single kinetic category was unprotected amide hydrogens that exchanged before the first, 5-s time point, with protection factor ϭ log (k chem / k ex ) Ͻ 1.2. Fully 53% of the assessed amides (colored reddishbrown in the figure) exchanged before the first 5-s time point when troponin was placed in D 2 O containing low Ca 2ϩ . An almost equal fraction of amide hydrogens (47%) exchanged before 5 s in the presence of high Ca 2ϩ . Some of these fast exchanging hydrogens are in regions of troponin that are, as far as is known, predominantly disordered in the absence of binding partners actin and tropomyosin. However, most of these fast-exchanging hydrogens are in regions with structure identifiable by x-ray and/or NMR. Therefore, results suggest these regions are loosely folded rather than disordered. They appear to have too little protection from exchange to be detected in the present study. For example, amide hydrogens with Ͻ 20-fold protection would experience nearly complete hydrogen-deuterium exchange within 5 s. Summing the extent of these reddish brown-colored regions with the extent of the red-colored regions exhibiting fast but measurable exchange rates (1.3 Ͻ log(k chem /k ex ) Ͻ 1.7)), the H/D exchange rates of 66% of the troponin complex suggest loose or absent local folding in the presence of saturating Ca 2ϩ concentration. Similarly, 58% of the examined portion of the protein has these properties (i.e. either red or reddish brown in the figure) in the presence of low Ca 2ϩ . The overall picture is of a highly dynamic molecule, regardless of whether site II is saturated with Ca 2ϩ . Fig. 4 shows that some regions of the primary structure contain hydrogens that are very highly protected from exchange. The violet-colored regions do not exchange in seconds. Very much to the contrary, they have protection from amide hydrogen exchange beyond the last 6-h time point. The largest structural region showing such protection is in the heart of the TnT-TnI coiled-coil, involving corresponding regions of both subunits. The direct kinetic data showing this is in Fig. 2A. This panel shows H/D exchange kinetic data for three nested TnT peptides. All exchange slowly, exemplifying the slow H/D exchange in much of the coiled-coil.

H/D Exchange Rates within Troponin Regions Distant from the Regulatory Ca 2ϩ Binding Site-
More generally, Fig. 2A can be used to illustrate how the kinetic exchange data were analyzed. Solid lines are best fit curves, providing transition sizes and rate constants used elsewhere in the present report. The three dashed lines in Fig. 2A, similar to all panels in Figs. 2 and 3, indicate the expected   (Fig. 2B) and in the presence of lower Ca 2ϩ (Fig. 2C), reaching in each case only 50% maximal possible exchange (dashed lines). In contrast to many other regions of troponin that are highly dynamic according to the H/D exchange data, this region of the coiled-coil likely serves as a key component of the overall stability of the ternary troponin complex.
Close examination of Fig. 2C shows that an H/D exchange transition involving about six amides occurred during the first minute in both peptides, i.e. within amides 98 -109 shared by the two peptides. H/D exchange rates for these and other hydrogens are listed in Tables 1 and 2. Remarkably, this Fig. 2C first minute transition was absent in the presence of mM Ca 2ϩ (Fig. 2B). Similarly, much more exchange occurred in the first minute within TnI-(117-124) in the presence of lower Ca 2ϩ than in the presence of higher Ca 2ϩ (both curves shown in Fig.  2D). Thus two TnI peptides within the coiled-coil undergo significant allosteric effects detected by changes in H/D exchange rates, despite being located far from the Ca 2ϩ binding site in the TnC NH 2 domain and outside the portions of troponin generally implicated as part of regulatory Ca 2ϩ signaling. The current findings suggest that regulatory signal transduction includes broad regions of troponin, beyond the TnC regulatory domain.
Similar effects were observed in the TnC COOH-terminal peptide 154 -161 (Fig. 2G). Filled or open symbols indicate conditions of TnC site II saturation or unsaturation, in this as well as in the other panels. The TnC COOH domain includes high affinity Ca 2ϩ sites III and IV, which were saturated under both conditions examined in the present study (1). Nevertheless, the figure shows a difference between the conditions. Specifically, 3 or 4 of the TnC-(154 -161) amide hydrogens exchanged with a Ca 2ϩ concentration-sensitive rate. They exchanged at ϳ1,000 s when the Ca 2ϩ concentration was M and at ϳ10,000 s when the Ca 2ϩ was mM. (The remaining 3 or 4 hydrogens exchanged within the first 5 s.) This implies that the TnC H-helix, the end of which is within this peptide, is affected by Ca 2ϩ binding to the TnC NH 2 domain.
Of course, many of the mapped amides in troponin exhibited exchange rates that were indistinguishable, regardless of Ca 2ϩ .  (Fig. 4) of the locations of 4 timescales for H/D exchange: before 5 s (unprotected); within the first minute; at ϳ1000 s (panel E only); and later than 6 h (panel E only). The overall pattern indicated no effect of Ca 2ϩ and a dynamic helix, loosely folded except for high protection from exchange where the helix is in direct contact with the COOH domain of TnC (see below).

H/D Exchange Rates within Troponin Regions Proximal to the Regulatory Ca 2ϩ
Binding Site-One of the most diverse regions for exchange rates was the NH 2 domain of TnC, which contains Ca 2ϩ site II. Many of the amide hydrogens exchanged quickly (data exemplified by Fig. 3A). Others exchanged over a range of time scales, as shown by the variegated coloring of the TnC 2-57 region of Fig. 4, and as exemplified by Fig. 3B. Saturating Ca 2ϩ resulted in fewer hydrogens that were completely unprotected from exchange in the TnC NH 2 domain (Fig. 4). High Ca 2ϩ appeared to increase the protection of some amide hydrogens and decrease the protection of others. Current results qualitatively agree (but are insufficient for quantitative comparison because of long peptides) with findings on isolated TnC or the TnC NH 2 domain by x-ray crystallography (5,26), NMR (25,27), and H/D exchange (28), all of which indicate changes in this domain on Ca 2ϩ binding. Fig. 3C shows H/D exchange kinetics in the presence of mM Ca 2ϩ for TnI peptide 156 -169 and subpeptide 162-169. The longer of these two includes part of the TnI switch helix that binds to the TnC NH 2 domain in a Ca 2ϩ -dependent manner. A  (20) for the unfolded state, and are color coded on a logarithmic scale. Both high Ca 2ϩ and low Ca 2ϩ findings are shown. Violet regions are the least dynamic, most protected from exchange (generally not exchanging after 6 h). Reddish-brown regions are the most dynamic, evidencing no protection (exchange completed by 5 s). This indicates either weak or absent folding within native troponin. In the TnC NH 2 domain, the relative location of exchange transitions could not be determined within residues 14 -24 and also within residues 26 -57; the colors may be scrambled.
transition over the first 10 s was observed only in the larger peptide (uppermost curve), defining the location of the transition as the switch helix. This is as expected: the switch helix is folded and has detectable protection from exchange under high Ca 2ϩ conditions, when it is attached to TnC.
In contrast, data for the same two peptides (Fig. 3D) indicated no protection of the switch helix from H/D exchange in the presence of low Ca 2ϩ , i.e. unsaturated site II. The key within the panel to this inference is that by 5 s the larger peptide already incorporated at least 6 more deuteriums than did the smaller peptide. Thus, the additional six amide segment present in the larger peptide, a segment that includes the switch helix amides, underwent full exchange by 5 s. The data in Fig.  3D for TnI-(162-169) are of mediocre quality, but are of use when globally assessed with the higher quality TnI-(156 -169) data. For example, because the longer peptide undergoes complete exchange by 60 s, the shorter peptide must do so as well.
Similarly, data for TnI peptide 125-152 and its subpeptide 125-134, shown in Fig. 3, E and F, allow assessment of exchange in another important region: TnI-(137-148). Early work on troponin (reviewed in Ref. 1) showed that the corresponding peptide from skeletal muscle troponin binds to the thin filament with inhibitory effect as an isolated peptide. This segment is included only in the larger of the two peptides shown in panels E and F. In the presence of mM Ca 2ϩ (panel E), a large kinetic transition occurred at ϳ5 s in the larger peptide, with magnitude such that the inhibitory segment must be included. This is consistent with uniform dynamics detected by EPR (29). A smaller transition occurred at ϳ400 s, and is assignable to the end residues of the coiled-coil because it occurred in both peptides of the panel. (The shorter peptide comprises the end of the coiled-coil.) Data from the same two TnI peptides indicated that release of Ca 2ϩ from TnC site II, i.e. lower Ca 2ϩ concentration conditions, altered exchange kinetics substantially (Fig. 3, F versus E). The longer peptide incorporated more deuterium within the first 5 s in the presence of lower Ca 2ϩ than it did in the presence of higher Ca 2ϩ . However, this was not the only qualitative change for this longer peptide; there was a converse effect as well. With lower Ca 2ϩ conditions, incorporation flattened by 5 s, so that by 15 s there was much more exchange into the longer peptide when the Ca 2ϩ concentration was higher than when it was lower. These results imply large effects of Ca 2ϩ binding to TnC site II on the dynamic properties of the TnI inhibitory region. Some residues exhibited faster exchange, and some exhibited slower exchange. The data at 30 s and beyond were relatively noisy for the larger peptide, so the global fit at later times was dominated primarily by the data for the shorter peptide. Neither peptide reached maximal H/D exchange by 6 h, another contrast from panel E.

DISCUSSION
Thin filament activation involves troponin movement, and troponin dynamics have proven a productive subject for understanding regulation (e.g., Ref. 12,30)). The present report adds to this literature by employing amide hydrogen exchange, which provides particularly rich dynamic information. No exogenous probes need be attached, and insight is obtained for all of the protein backbone, including both mobile and immobile regions. Furthermore, when exchange rate constants can be determined as in the present report, a wealth of quantitative dynamic insight is gained. Fig. 5 illustrates concisely many of the findings in the present report. It shows the degree of protection from H/D exchange, as measured in the presence of mM Ca 2ϩ . The data are mapped onto the high resolution structure of the Ca 2ϩ -saturated cardiac troponin core domain, using a log scale color coding. Enormous variation was seen across this highly asymmetrical, extended protein. The most protected regions, in violet or dark blue, identify the components that are key to formation and overall stability of the ternary complex. These are the TnT-TnI coiled-coil, particularly the most highly protected portion in both subunits, and the attachment of TnI helix 1 to TnC. The H/D exchange results demonstrate these aspects both qualitatively and quantitatively. Furthermore, these findings can be seen in the figure to be in striking contrast to the overall dynamic aspects of troponin, most of which are shown in red or reddish brown, indicating weak local folding. Many features are notable from this mapping. The TnT-TnI coiled-coil was highly protected, but neither uniformly along its length, nor uniformly between TnT and TnI. The TnT was more protected, suggesting that when the coiled-coil strands transiently dissociate, the TnT strand is more stable. High protection began at the base of the coiled-coil, was most profound near the center, and greatly diminished near the COOH termini of both TnT and TnI components of the coiled-coil. In support of the validity of this mapping, there is an atomic resolution explanation for the most highly protected region that exists across both TnT and TnI strands: the region closely coincides with charged side chain networks that are important for troponin function, that extend across the coiled-coil, and that involve TnT residues E244 and K247 and TnI residues K106, D110, R79, and R103 (31).
The three-turn TnI switch helix in Fig. 5A is noteworthy. It overlies the TnC NH 2 domain that contains regulatory site II, and is colored red to indicate weak but measurable protection from exchange (gray-colored portions were not measured.) H/D exchange within the TnI switch helix remained relatively fast, despite full saturation of site II via 1 mM Ca 2ϩ . Protection from exchange was a modest ϳ50-fold. This suggests weak, energetically facile dissociation of the switch helix from Ca 2ϩ -saturated TnC, as may be required for regulatory switch function.
Similarly, the data indicate that the TnI "inhibitory" region (boxed residues including TnI 140 in Fig. 4) is folded sufficiently under Ca 2ϩ -saturating conditions to be ϳ30-fold protected from H/D exchange (i.e. the data show that k chem /k ex Ϸ 30). The inhibitory region was not identified in the high resolution x-ray structure of Ca 2ϩ -saturated cardiac troponin (4), but was ordered and identified in the high resolution structure of Ca 2ϩsaturated skeletal muscle troponin (5). Our cardiac troponin data indicate that, in solution, the Ca 2ϩ -saturated cardiac isoform involves a folded rather than disordered state for the TnI inhibitory region, corresponding to the findings for skeletal muscle troponin. Thus, it is part of the intra-troponin Ca 2ϩ switch, in addition to any role it may have in effecting inhibition of muscle contraction.
In contrast to the coiled-coil, TnT helix 1 (at far left of Fig.  5A) and most of the TnI helix 1 were highly dynamic. The exception to this within TnI helix 1 is high protection from exchange where it comes in contact with the globular, COOH domain of TnC. This suggests a tight interaction important for troponin complex formation.
The effects of Ca 2ϩ on troponin dynamics are illustrated in Fig. 5B. Blue regions were more protected (inferred to be less flexible and more tightly folded) when Ca 2ϩ was saturating, and red regions were less protected when Ca 2ϩ was saturating. Regions with statistically insignificant alteration in exchange kinetics are colored silver (with a light blue tint). For green regions the effects of Ca 2ϩ could not be assessed. Ca 2ϩ increased the protection from exchange for some regions and decreased protection for others, with more of the former. The figure illustrates clearly that changes were not confined to the TnC NH 2 domain and overlying TnI switch helix. Rather, allosteric effects were evident in the TnC COOH domain, in the TnI strand of the coiled-coil and at the COOH terminus of both strands of the coiled-coil. Actin and tropomyosin, of course are not present in these experiments, and cannot explain these effects. These Ca 2ϩ -mediated effects on many parts of this extended molecule raise two questions: how are the effects transmitted, and what relation do these effects have to the mechanism of thin filament regulatory function?
The answers to these questions, ultimately, should reference troponin and thin filament structures, which are known in significant part. In particular, the current results should be considered in relation to the proposed atomic structure of skeletal muscle troponin in low Ca 2ϩ , hypothesized from 9 Å x-ray data obtained in the presence of Mg 2ϩ /EGTA (5). In Fig. 5C, the low Ca 2ϩ cardiac troponin H/D exchange data are mapped onto the homologous residues of this skeletal muscle troponin model. The orientation is different from that shown in Fig. 5, A and B. Also, regardless of orientation, the TnC NH 2 domain has a different position within this structure. Importantly, just beyond the far end of the coiled-coil, the inhibitory region of TnI folds back sharply in this model to form tight contacts not observed in high Ca 2ϩ conditions. This includes contacts with the TnC NH 2 domain (bottom) and the end of the TnI strand of the coiled-coil.
To explain the current H/D exchange findings, we propose that a similar conformation occurs in cardiac troponin, as part of an intra-troponin switch (Fig. 6). This conformation alters the H/D exchange properties of the inhibitory region and the end of the TnI strand of the coiled-coil, as well as the exchange properties of both the corresponding portion of the TnT strand, and the TnC COOH domain to which the TnT attaches. In this way, we suggest that the effect of Ca 2ϩ on troponin is to promote a switch involving several components. In one conformational state the TnI switch helix attaches to the TnC regulatory domain, and the inhibitory region attaches to the D/E helix. In the second conformation, the inhibitory region of TnI attaches to the end of the coiled-coil. For vertebrate troponins, Ca 2ϩ binding to the TnC NH 2 domain influences this switching. We speculate that for many invertebrate troponins, switching instead is affected by Ca 2ϩ binding to the TnC COOH domain, which interacts with the end of the coiled-coil.
Weaker binding between the two coiled-coil strands, when the Ca 2ϩ concentration is low, is suggested by the faster H/D exchange properties extending through most of the TnI strand of the coiled coil in the presence of subsaturating Ca 2ϩ . This mechanism may explain how site II Ca 2ϩ binding can have such a remote effect on H/D exchange. Furthermore, this may have functional significance (32). Increased dynamics of the TnI portion of the coiled-coil supports the possibility that it interacts with actin differently, depending upon Ca 2ϩ conditions. At present there is little structural data regarding the interaction of the TnT-TnI coiled-coil with other thin filament proteins (33).
In summary, the troponin role in the regulation of muscle contraction consists of a Ca 2ϩ -triggered change in quaternary structure, a switching between alternative sets of intra-troponin interactions. The current results are consistent with models of regulation in which COOH-terminal portions of TnI interact with actin and tropomyosin to inhibit muscle contraction under low Ca 2ϩ , relaxing conditions. The present work implies that intra-troponin interactions, evident in the H/D exchange data, and involving the end of the TnT-TnI coiled-coil, stabilize a troponin conformation that positions the TnI COOH terminus where it can affect actin and the position of tropomyosin on actin (7,34), regulating muscle contraction. FIGURE 6. Illustration of the proposed effects of Ca 2؉ on troponin. Top, selected, indicated regions from the high resolution Ca 2ϩ -saturated skeletal muscle troponin structure. Bottom, same regions are shown within the proposed structure of Mg 2ϩ /EGTA skeletal muscle troponin, which was derived from intermediate resolution data in the same study (5). Cardiac TnI regions with Ca 2ϩ -sensitive H/D exchange rates are shown in hot pink, including the inhibitory region (stick representation), the switch helix, and much of the coiled-coil. A glycine at the coiled-coil terminus is shown in spherical representation, to indicate where the Ca 2ϩ -sensitive changes in the TnI inhibitory region commences in the proposed model. In high Ca 2ϩ , the cardiac TnI inhibitory region exhibits H/D exchange protection, as would be predicted from this skeletal muscle troponin structure, but not from the cardiac troponin x-ray structure (Fig. 5). Several changes occur when Ca 2ϩ dissociates from the TnC NH 2 domain. The TnI switch helix dissociates and becomes disordered. The inhibitory region turns to tightly interact with the end of the coiled coil and changes its position greatly. Interactions with TnC helix H are altered. The stabilizing effects of the inhibitory region on the D/E helix are lost.