The Motif of Human Cardiac Myosin-binding Protein C Is Required for Its Ca2+-dependent Interaction with Calmodulin*

Background: The motif of human cMyBP-C contains phosphorylation sites that regulate its interaction with myosin ΔS2. Results: The motif interacts with calmodulin in a calcium-dependent manner yet is independent of its phosphorylation state. Conclusion: Highly conserved residues toward the C-terminal end of the motif interact with calmodulin in an extended conformation. Significance: Calmodulin can link cMyBP-C with the calcium-signaling pathways in muscle. The N-terminal modules of cardiac myosin-binding protein C (cMyBP-C) play a regulatory role in mediating interactions between myosin and actin during heart muscle contraction. The so-called “motif,” located between the second and third immunoglobulin modules of the cardiac isoform, is believed to modulate contractility via an “on-off” phosphorylation-dependent tether to myosin ΔS2. Here we report a novel Ca2+-dependent interaction between the motif and calmodulin (CaM) based on the results of a combined fluorescence, NMR, and light and x-ray scattering study. We show that constructs of cMyBP-C containing the motif bind to Ca2+/CaM with a moderate affinity (KD ∼10 μm), which is similar to the affinity previously determined for myosin ΔS2. However, unlike the interaction with myosin ΔS2, the Ca2+/CaM interaction is unaffected by substitution with a triphosphorylated motif mimic. Further, Ca2+/CaM interacts with the highly conserved residues (Glu319–Lys341) toward the C-terminal end of the motif. Consistent with the Ca2+ dependence, the binding of CaM to the motif is mediated via the hydrophobic clefts within the N- and C-lobes that are known to become more exposed upon Ca2+ binding. Overall, Ca2+/CaM engages with the motif in an extended clamp configuration as opposed to the collapsed binding mode often observed in other CaM-protein interactions. Our results suggest that CaM may act as a structural conduit that links cMyBP-C with Ca2+ signaling pathways to help coordinate phosphorylation events and synchronize the multiple interactions between cMyBP-C, myosin, and actin during the heart muscle contraction.

The controlled modulation of Ca 2ϩ concentrations within muscle sarcomeres is the primary signaling mechanism used to regulate the sliding of myosin-based thick filaments past actinbased thin filaments to give rise to muscle contraction and relaxation. In striated muscle, Ca 2ϩ activates the thin filament by binding to troponin, thus triggering conformational changes among the thin filament accessory proteins. The result is a shift of tropomyosin on the thin filament that uncovers the binding sites where myosin heads attach to actin and impart force (1). However, there are additional layers of regulation that modulate the contractile cycle. We have previously proposed that the N-terminal modules of cardiac myosin-binding protein C can modulate the thin filament activation response to Ca 2ϩ signals, essentially buffering the Ca 2ϩ response by competing for myosin binding sites on actin while also sterically hindering tropomyosin from relaxing to its Ca 2ϩ -off position and thus holding myosin binding sites on actin open (2,3). Ca 2ϩ signals also regulate the activity of a number of kinases important in muscle action, in a number of instances via the multifunctional intracellular Ca 2ϩ receptor calmodulin (4 -6). One such calmodulin-regulated kinase is Ca 2ϩ /calmodulin-dependent protein kinase II (CaMKII) 3 that phosphorylates a critical regulatory region within cMyBP-C (7,8).
Calmodulin (CaM) is a small, dumbbell-shaped protein that is a structural homologue of the Ca 2ϩ -binding subunit of troponin (TnC), but rather than having a single known regulatory function, CaM binds to a diverse array of target proteins to modulate their functions in response to Ca 2ϩ signals (4 -6). The two globular lobes of CaM each contain a pair of Ca 2ϩbinding EF-hand (helix-loop-helix) motifs that are connected by a solvent-exposed helix of ϳ8 turns. Upon Ca 2ϩ binding, a hydrophobic cleft opens in each lobe to mediate target protein recognition and binding. In the classical Ca 2ϩ /CaM-target protein complexes, the two lobes of CaM wrap around a helical peptide sequence via flexibility in the interconnecting helix (6,9). Other modes of Ca 2ϩ /CaM binding have been observed in which CaM maintains an extended conformation (10 -14) similar to those observed for TnC interacting with its regulatory binding partner troponin I (TnI) (15).
Myosin-binding protein C (MyBP-C) was first identified as a thick filament accessory protein and has since been determined to play both structural and regulatory roles in striated muscle contraction (16 -18). MyBP-C is a modular protein made up of immunoglobulin (Ig)-type and fibronectin-like modules. An ϳ100-residue sequence, referred to as the "motif" (m), sits between two Ig modules near the N terminus (Fig. 1, A and B) and has been implicated in mediating potentially key phosphorylation-dependent interactions in the cardiac isoform (19). Interest in cardiac MyBP-C (cMyBP-C) has been stimulated following the finding that mutations in the gene encoding the cardiac isoform are a leading cause of familial hypertrophic cardiomyopathy (20 -22). cMyBP-C contains eight Ig modules and three fibronectin modules, designated C0 -C10. The C0 Ig module is cardiac specific and connects to C1 via a proline/ alanine-rich linker (Fig. 1A). In addition, there are multiple cardiac specific phosphorylation sites compared with only one in the skeletal isoform (19). The C1-m-C2 fragment (C1C2) binds to myosin ⌬S2, close to the lever arm region of the myosin head and primarily via the motif (23,24). During ␤-adrenergic stimulation, cMyBP-C undergoes phosphorylation and releases the bound myosin ⌬S2, resulting in an acceleration of the crossbridge cycling rates and force generation (23,25).
Both CaMKII and protein kinase A (PKA) can phosphorylate cMyBP-C on at least three serines within the motif (Ser 275 , Ser 284 , and Ser 304 in the human sequence, corresponding to Ser 273 , Ser 282 , and Ser 302 in mouse) (7,8,19,26) (Fig. 1C). An ordered hierarchy of phosphorylation exists among these sites (23,27,28). In the mouse protein, Ser 282 is the main target for CaMKII, and, importantly, phosphorylation at this site is necessary to induce further phosphorlyation at Ser 273 and Ser 302 by PKA. The critical role of Ser 282 phosphorylation by CaMKII is highlighted by its links to enhanced contractility after myocardial stunning (29,30), whereas the loss of cMyBP-C phosphorylation at Ser 282 results in altered response to ␤-adrenergic stimuli (31). The CaMKII-regulated phosphorylation is strictly Ca 2ϩ /CaM-dependent and can be inhibited by the Ca 2ϩ chelator EGTA or the CaM-binding peptide sequence from myosin light chain kinase (MLCK) (7,8,19,32). CaMKII inhibition reduces both cMyBP-C and TnI phosphorylation and decreases maximum force through a cross-bridge feedback mechanism (33). Together, the combined evidence suggests that Ca 2ϩ / CaM-dependent phosphorylation of cMyBP-C may be a key step for coordinating events in the contractile cycle. When directly isolated from muscle tissue, cMyBP-C is purified with endogenous CaMKII activity (7), suggesting that there is an intimate link between the protein and this Ca 2ϩ /CaM-dependent kinase.
In addition to regulating CaMKII activity, Ca 2ϩ signals regulate the activity of the Ca 2ϩ /CaM-dependent MLCK that phosphorylates yet another EF-hand relative of CaM, the myosin regulatory light chain (RLC) (34 -36). The CaM-dependent phosphorylation of cMyBP-C and RLC contribute to the contraction/relaxation cycle by modifying the local concentration of cross-bridges at the interface with actin (37). The C0 domain of cMyBP-C can directly interact with the RLC (38), whereas the cMyBP-C motif, in an unphosphorylated state, can interact with myosin ⌬S2 (24). Therefore, the N-terminal domains of cMyBP-C can interact with both components of myosin proximal to the myosin head and with actin (2,39). It thus appears that Ca 2ϩ signaling and CaM-dependent mechanisms of phosphorylation in response to these signals are integral to regulating molecular interactions between all three proteins. Because CaM is a pivotal Ca 2ϩ sensor, we asked whether CaM may serve as a structural, Ca 2ϩ -sensitive link to the regulatory role of cMyBP-C.
In this study, we have used pull-down assays, size exclusion chromatography with multiple-angle laser light spectroscopy (SEC-MALLS), tryptophan fluorescence, nuclear magnetic resonance (NMR) spectroscopy, and small angle x-ray scattering (SAXS) to identify and characterize a novel interaction between cMyBP-C and CaM. We have identified a region within the motif responsible for CaM binding and determined that the interaction is of moderate affinity, with CaM maintaining an extended conformation in the complex. We show that the interaction is Ca 2ϩ -dependent yet independent of the phos-FIGURE 1. N-terminal modules and the motif of cMyBP-C. A, schematic representation of the three N-terminal Ig modules (C0, C1, and C2) and their connectors: the proline/alanine-rich linker (P/A L ) and the motif (m). The approximate molecular weight (M w ) of each region is indicated, and the color coding for different regions of sequence/structure is conserved in panels A, B, and C. B, ribbon diagrams of (from left to right) the human C1 (PDB 2AVG) module, the structured region of mouse motif (residues 315-351) (PDB 2LHU), and the human C2 (PDB 1PD6) module with tryptophan residues (Trp 191 , Trp 196 , Trp 318 , and Trp 396 ) shown as red sticks. Note that Trp 318 in the mouse sequence is equivalent to Trp 322 in the human sequence. In the mouse motif ribbon diagram (middle), the sequence corresponding to human Cpep used in this study (incorporating Trp 318 ) is shown in yellow, and the remaining sequence is orange. C, sequence alignments of human and mouse cMyBP-C motifs and of human motif peptides (Npep and Cpep). For the human motif, the three putative phosphorylation sites are marked by asterisks above the sequence, and the only tryptophan residue, Trp 322 , is highlighted in red. Sequences corresponding to the structured region of mouse motif (amino acids (aa) 315-351) are highlighted in yellow and orange as in B. Note that for the human triphosphorylation mimic C1C2EEE, the three putative phosphorylation sites are mutated to glutamates (i.e. S275E/S284E/S304E). phorylation state of the motif. Our results provide the first structural evidence that cMyBP-C can access Ca 2ϩ signaling pathways via a physical attachment to Ca 2ϩ /CaM that could have consequences for synchronizing events between cMyBP-C and its binding partners in muscle filaments.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-The expression plasmids encoding the human cMyBP-C N-terminal constructs (C0-C1 ' C0C1, C1-m-C2 ' C1C2, C0-C1-m-C2 ' C0C2, and motif ' m) were prepared as described (40). The C1C2EEE triple glutamate mutant (S275E/S284E/S304E), which incorporates a stable triphosphomimic of the motif, was isolated from a synthetic C0C2EEE gene manufactured by Genscript Inc. employing the same protocol as described for the isolation of C0, C1, and C0C1 (3). All human cMyBP-C proteins were expressed in Escherichia coli Rosetta 2 cells and purified as described previously (41), whereas 15 N-labeled C1C2 and the triphosphomimic C1C2EEE were produced using the procedures described previously (39). The final expressed recombinant protein included an in-frame N-terminal polyhistidine tag that was not removed due to the proteolytic sensitivity of the C1C2EEE protein. CaM and 15 N-labeled CaM were expressed in E. coli Rosetta 2 cells and purified as described previously (42). Stock solutions of C1C2 and C1C2EEE were stored in a buffer containing 25 mM Tris, 350 mM NaCl, and 2 mM TCEP, pH 7.0; stock solutions of the motif were stored in buffer containing 20 mM Tris, 50 mM NaCl, 2 mM TCEP, pH 6.8; and stock solutions of CaM were stored in a buffer containing 50 mM MOPS, 150 mM NaCl, and 10 mM CaCl 2 , pH 7.4. The two synthetic peptides were purchased from Genscript Inc. Their sequences are RDSFRTPRDSKLEAPAEEDVWEILR and EDV-WEILRQAPPSEYERIAFQYG, and they are referred to as Npep and Cpep, respectively. Prior to dialysis in titration or SAXS buffers, the lyophilized Npep or Cpep was dissolved in water and 10% acetic acid, respectively, to a final concentration of ϳ4 mg/ml.
CaM-agarose Pull-down Assays-CaM-agarose pull-down assays were performed to screen which of the purified C0, C0C1, C1C2, and C0C2 cMyBP-C fragments could bind CaM. Approximately 500 l of each N-terminal fragment at 1 mg/ml in 25 mM MOPS, 50 mM KCl, 5 mM CaCl 2 , 2 mM TCEP, pH 7.0, was added to 100-l aliquots of covalently linked CaM-agarose beads (Sigma) equilibrated in the same buffer. The protein/ agarose bead slurries were left to incubate for ϳ15 min at 4°C, with gentle mixing every 2 min to distribute the bead matrix in solution, before being spun at 10,000 ϫ g for 30 s to pellet the beads. The pelleted beads were sequentially resuspended/ washed and spun four times using 500 l of the aforementioned buffer to remove any unbound protein. This process was also repeated as described using a calcium-free buffer solution, where 5 mM CaCl 2 was substituted by 10 mM EGTA throughout the protein binding and bead washing steps. To monitor cMyBP-C fragments binding to the CaM-agarose in the presence or absence of Ca 2ϩ , 20-l aliquots of washed and spun beads were added to 10 l of 25 mM Tris, pH 8.0, and 10 l of 4ϫ SDS-PAGE loading dye to form a 50% (v/v) bead/loading dye slurry. The slurry was subsequently boiled for 5 min and spun to repellet the beads, and 15 l of supernatant from each individual sample was used for SDS-PAGE analysis.
Size Exclusion Chromatography with Multiple-angle Laser Light Scattering-SEC-MALLS experiments and data analyses were carried out following the procedures described previously (42). SEC-MALLS data were collected for purified Ca 2ϩ /CaM, C1C2, and a mixture of Ca 2ϩ /CaM plus [ 15 N]C1C2 solution that contained excess Ca 2ϩ /CaM (Ca 2ϩ /CaM to [ 15 N]C1C2 molar ratio of 2:1). Typically, ϳ400 l of protein solution at ϳ100 M was analyzed in a Superdex 75 10/30 gel filtration column (GE Healthcare) coupled with the miniDAWN light scattering unit (Wyatt Technology) and a DSP refractometer (Optilab). The system was equilibrated with a standard buffer containing 25 mM MOPS, 350 mM NaCl, 5 mM CaCl 2 , 2 mM TCEP, pH 7.0. Unless otherwise stated, this standard buffer was used in all of the SEC-MALLS, tryptophan fluorescence, and NMR titration experiments reported in this study. Following similar conditions and procedures, SEC-MALLS data were also acquired for purified Ca 2ϩ /CaM, motif, and the Ca 2ϩ /CaMmotif complex with excess Ca 2ϩ /CaM (prepared by mixing Ca 2ϩ /CaM and motif at a 2:1 molar ratio).
Tryptophan Fluorescence Spectroscopy-Fluorescence spectra of cMyBP-C protein or peptide samples (including C1C2, C1C2EEE, motif, Cpep, and Npep) and their corresponding buffer solutions were obtained at 25°C using a Varian Eclipse fluorescence spectrophotometer following the same settings as described previously (42). Prior to the titration experiments, aliquots of the cMyBP-C protein or peptide and CaM stock solutions were individually dialyzed overnight against the standard buffer. To evaluate the Ca 2ϩ dependence of the binding, the 5 mM CaCl 2 in the standard buffer was replaced by 10 mM EGTA. The effect of ionic strength on the Ca 2ϩ /CaM-C1C2 interaction was evaluated by reducing the concentration of NaCl in the standard buffer to 100 mM. Similarly, the binding of motif to Ca 2ϩ /CaM was evaluated in a buffer containing 20 mM Tris, 50 mM NaCl, 5 mM CaCl 2 , 2 mM TCEP, pH 6.8.
At each titration step, aliquots of a concentrated CaM solution (Ͼ1 mM) were sequentially added to the cMyBP-C protein or peptide solution (ϳ5 M) and left to equilibrate for 10 min prior to recording their emission spectrum eight times. The eight spectra were averaged to produce the final spectrum. All averaged sample spectra were corrected for solvent effects by subtracting the emission spectra of the respective buffer across the full emission range. Dilution effects and the intrinsic fluorescence contribution from tyrosine were considered in the data analysis. Inner filter effects were checked for by sequentially titrating CaM stock into a standard tryptophan solution (ϳ40 M). No significant inner filter effects were observed for the amount of CaM used in all titration experiments (data not shown).
To determine the binding affinity (K D ) in each tryptophan fluorescence titration experiment, the relative fluorescence intensity F/F 0 values (where F is the fluorescence intensity at each titration step, and F 0 is the fluorescence intensity of Ca 2ϩfree (apo-) protein or peptide) at a chosen wavelength were plotted against the concentration of CaM and fitted to a quadratic equation based on a 1:1 binding model (e.g. see Ref. 43). For the Cpep/CaM titration data, the fit to the binding curve improved significantly when a concentration adjustment factor was included in the parameters. This adjustment factor typically ranged from 0.4 to 0.8, suggesting that the peptide concentration was lower than initially estimated from the extinction coefficient at 280 nm, possibly due to the low solubility of the peptide. For each set of titration data, the wavelength that gave the largest intensity changes between the apo and bound states was chosen for analysis. To achieve this, fluorescence spectra of the apo state and the last titration point were normalized to unity using their respective maximum intensity values. Subtraction of these normalized spectra yields the wavelength at which the largest intensity changes can be observed. Note that binding affinities can also be determined by plotting the changes in maximum emission wavelength max (i.e. the wavelength that maximal fluorescence intensity is observed) against concentrations of CaM. The K D values determined were essentially the same using F/F 0 or max .
NMR Spectroscopy-1 H-15 N HSQC spectra were recorded at 25°C using a Bruker AVIII 800-MHz NMR spectrometer equipped with a triple-resonance cryogenic probe. Stock protein solutions were dialyzed against the standard buffer overnight, and the final protein concentrations used in titration experiments were ϳ100 M for [ 15  For each addition, the solutions were mixed and equilibrated for 10 min before spectra were recorded. Relative peak heights (using the spectrum of [ 15 N]C1C2 alone as a reference) for all identifiable and nonoverlapped residues at each titration step were plotted. Residues that underwent significant intensity changes were defined as those with peak heights decreasing by more than one S.D. value from the average. Similar NMR experiments and data analyses were carried out using [ 15  used were 1:0.1, 1:0.2, 1:0.3, 1:0.4, 1:0.5, 1:0.6, 1:0.7, 1:0.85, 1:1, 1:1.5, 1:2, 1:3, and 1:4. For the Cpep titration, spectral changes presented mainly as positional rather than intensity changes; therefore, combined chemical shift perturbations (⌬␦ ppm ) were calculated based on the equation, ⌬␦ ppm ϭ ((⌬␦ HN ) 2 ϩ (0.154 ϫ ⌬␦ N ) 2 ) 1 ⁄ 2 , where ⌬␦ HN and ⌬␦ N represent the chemical shift variations in the proton and nitrogen dimensions, respectively. Residues with positions changed by more than one S.D. value from the mean were deemed to be significantly affected. Binding affinity was determined from fitting to a 1:1 binding model using ⌬␦ ppm of residues Ala 57 , Val 121 , and Ile 130 at varying concentrations of Cpep.
SAXS Data Acquisition, Analysis, and Modeling-Ca 2ϩ /CaM was freshly prepared from gel filtration in the standard buffer and immediately concentrated for SAXS measurements. For Ca 2ϩ /CaM-Cpep, the complex was prepared by simply mixing the Ca 2ϩ /CaM and Cpep solutions at a 1:3 molar ratio (note that the excess Cpep does not contribute significantly to the scattering intensity due to its small molecular volume). SAXS data were acquired at 10°C using a SAXSess instrument (Anton Paar, Graz, Austria) with 10-mm line collimation and a chargecoupled device (CCD) detector. All data reduction and analyses followed the procedures described previously (40). Low resolution ab initio shape reconstructions were performed using the program DAMMIF (45) as described previously (41). The NSD values of each set of 10 DAMMIF models were 0.621 Ϯ 0.012 for Ca 2ϩ /CaM and 0.557 Ϯ 0.015 for Ca 2ϩ /CaM-Cpep, respectively.

cMyBP-C Fragments Containing Motif-C2 Bind to Ca 2ϩ /
Calmodulin-Pull-down assays were performed by incubating CaM-agarose beads with N-terminal fragments of human cMyBP-C (C0C1, C1, C1C2, and C0C2) in the presence and absence of Ca 2ϩ (Fig. 2). Only C1C2 and C0C2, each of which includes the motif and C2, show significant binding to CaMagarose, and this binding is Ca 2ϩ -dependent. C1C2 and C0C2 appear to be pulled down to similar extents, suggesting that C0 is not involved in the interaction and that the Ca 2ϩ /CaM binding region exists within the motif and/or C2 domain.
Based on the qualitative pull-down assessments, a solution containing C1C2 and Ca 2ϩ /CaM in molar excess was subjected to SEC-MALLS. The results indicate that the first and predominant peak eluting from the column corresponds to a species with an experimental molecular mass of 48.9 Ϯ 0.2 kDa, consistent with the theoretical molecular mass of 52 kDa for a 1:1 Ca 2ϩ /CaM-C1C2 complex, whereas a second elution peak corresponds to excess Ca 2ϩ /CaM (data not shown).
Tryptophan fluorescence was then used to confirm and further characterize the binding between C1C2 and Ca 2ϩ /CaM. CaM-binding proteins commonly have a tryptophan in their CaM recognition sequences (6,9), which makes tryptophan fluorescence spectroscopy useful for probing CaM interactions, especially because CaM itself contains no tryptophans. There are four tryptophans in C1C2: Trp 191 , Trp 196 , Trp 322 , and Trp 396 . Based on the NMR and crystal structures of the C1 and C2 domains (44,46,47), there is one tryptophan located in the central hydrophobic core of each of the C1 and C2 domains (Trp 191 and Trp 396 , respectively), Trp 196 is located on the outer surface of the C1 domain, and forms part of the ␤-barrel, and Trp 322 resides within the motif (Fig. 1B).
As Ca 2ϩ /CaM was titrated into a C1C2 solution, a shift of the tryptophan emission peak was observed, with the maximum emission wavelength max changing from ϳ345 to ϳ336 nm; the blue shift is accompanied by a decrease in the maximum fluorescence intensity (Fig. 3). Tryptophan quenching is dominated by its carbonyl group and is very sensitive to changes in its local environment (48,49). The observed changes in max and intensity therefore indicate that the chemical environment of one or more of the tryptophans in C1C2 is altered upon Ca 2ϩ / CaM binding. Relative intensity changes at the wavelength giving the maximum change could be fitted to a 1:1 binding model to yield a dissociation constant (K D ) of 13.5 Ϯ 2.5 M ( Table 1). As in the pull-down assay, this interaction is Ca 2ϩ -dependent, and the binding affinity reduces markedly in the presence of EGTA ( Fig. 3 and Table 1). Further, unlike other reported CaMpeptide interactions (6,9), the affinity of the Ca 2ϩ /CaM-C1C2 interaction appears to be independent of ionic strength up to NaCl concentrations of 350 mM (Fig. 3 (Fig. 4) and shows mainly sharp and well dispersed peaks similar to the [ 15 N]C1C2 spectrum reported by Ababou et al. (44). These peaks overlay well with assignments reported for the isolated C1 and C2 domains (BMRB accession numbers 6105 and 5591) and indicate that the C1 and C2 Ig modules in C1C2 adopt the same fold as the individual domains (44,46).
Over ϳ80% of the amide resonances of C1 and C2 in [ 15 N]C1C2 could be inferred readily from the reported assignments. In contrast, the motif contributes to a substantial proportion of the unidentified peaks in the [ 15 N]C1C2 spectrum with 1 H chemical shifts between 8 and 9 ppm typical of disordered regions. Recently, Howarth et al. (50) reported NMR results showing that residues 315-351 of the 103-residue mouse motif form a small three-helical bundle (Fig. 1B), whereas the remainder of the motif is unstructured. The mouse cMyBP-C motif shares ϳ89% sequence identity with the human isoform (Fig.  1C), with only one amino acid difference in the three-helical bundle (Tyr 340 for human versus His 336 for mouse). However, despite the high degree of sequence similarity, only ϳ30% of amide assignments from the small three-helical bundle of the mouse motif (BMRB accession number 17867) could be confidently matched to corresponding peaks in our [ 15 N]C1C2 spectrum due to spectral overlap, especially in the region of the spectrum dominated by disordered structures.
Upon the addition of Ca 2ϩ /CaM to [ 15 N]C1C2, a subset of peaks in the 15 N-1 H HSQC NMR spectrum displayed a severe drop in signal intensity, with some peaks disappearing completely (Fig. 4). Specifically, ϳ66% of the amide resonance peaks mapped to the structured region of the motif showed significant intensity changes upon binding to Ca 2ϩ /CaM, including those assigned to Ala 317 , Ile 324 , Arg 326 , Ala 328 , Tyr 333 , Glu 334 , Arg 346 , Lys 350 , and Arg 351 that disappeared completely. In contrast, only a handful of peaks mapped to C1 and C2 were affected. In addition, a large proportion of peaks in the disordered region, which are likely to correspond to the unstructured residues within the motif, disappeared during the titra- Note that "C1C2 in 100 mM NaCl" data were acquired in buffer containing 100 mM NaCl, and "C1C2 in EGTA" data were acquired in buffer containing 10 mM EGTA; the rest of the titrations were performed in the standard buffer containing 350 mM NaCl (for details, see "Experimental Procedures"). The indole resonances of the four tryptophans in [ 15 N]C1C2 are among the set of residues that can be identified based on previously reported assignments (44). Interestingly, the indole peak for Trp 322 in the motif disappears at the first addition of Ca 2ϩ /CaM (Fig. 4, inset), suggesting that the environment of this tryptophan is very sensitive to the formation of the Ca 2ϩ / CaM-C1C2 complex, whereas the remaining three tryptophan indole peaks of [ 15 N]C1C2 (i.e. Trp 191 , Trp 196 , and Trp 396 from the C1 and C2 Ig modules) remain essentially unaffected throughout the course of the titrations.
The C-terminal End of the Motif Alone Is Sufficient for Binding to Ca 2ϩ /CaM-NMR titration studies suggest that the motif is predominantly responsible for the binding to Ca 2ϩ / CaM, with C1 and C2 playing only a minor role if any. Therefore, tryptophan fluorescence experiments were performed with a construct that encodes for the human motif only. Again, the characteristic blue shift and intensity decrease in the fluorescence emission spectra were observed as for C1C2 (Fig. 5), suggesting that the change in chemical environment of Trp 322 upon binding Ca 2ϩ /CaM is similar. The affinity of the interaction is also comparable between Ca 2ϩ /CaM-motif and Ca 2ϩ / CaM-C1C2 (Table 1). In addition, the binding stoichiometry remains the same at 1:1 according to SEC-MALLS analysis (data not shown).
The majority of the CaM minimal binding sequences characterized to date are 16 -30 residues in length (9, 51, 52), with two large hydrophobic residues positioned to interact with the two CaM lobes. Hence, we considered whether the minimum binding region in cMyBP-C could be substantially shorter than the 97-residue motif. Because our fluorescence and NMR results suggest that Trp 322 is likely to be critical for anchoring Ca 2ϩ /CaM, two peptides containing Trp 322 and at least one candidate for a second hydrophobic anchor were tested for binding Ca 2ϩ /CaM using tryptophan fluorescence. The sequences selected were an 18-residue peptide from the sequence mainly upstream of Trp 322 (Arg 309 -Lys 326 , termed Npep) and an alternate 23-residue peptide from the sequence mainly downstream of Trp 322 (Glu 319 -Lys 341 , termed Cpep) (Fig. 1C). Significantly, Cpep binds to Ca 2ϩ /CaM with an affinity (i.e. 4.3 Ϯ 1.3 M) comparable with that of the motif, whereas Npep displays negligible binding (Fig. 5 and Table 1). Therefore, Cpep appears to contain most or all of the amino acid sequence that is important for the Ca 2ϩ /CaM interaction.
Ca 2ϩ /CaM Binds to Cpep via the Hydrophobic Cleft in Each Ca 2ϩ -binding Lobe-In order to probe which residues in Ca 2ϩ / CaM are responsible for binding Cpep, 15 N HSQC spectra were recorded as Cpep was titrated into 15 N-labeled Ca 2ϩ /CaM. The 15 N HSQC spectrum from Ca 2ϩ /CaM alone overlays well with previously reported assignments (BMRB accession number 547), and ϳ77% of the amide resonances could be readily inferred from the reported assignments, with another 12% of peaks tentatively mapped.  As Cpep was titrated into 15 N-labeled Ca 2ϩ /CaM, many peaks in the 1 H-15 N HSQC spectrum moved, and a small subset of peaks disappeared, consistent with fast-and slow-intermediate exchange on the NMR time scale, respectively. A similar number of signals that had disappeared were observed to reappear at a molar ratio of ϳ1:4 (Ca 2ϩ /CaM to Cpep) (data not shown). The residues in CaM that undergo the largest chemical shift changes (corresponding to peaks that disappeared; Fig.  6A) include Ile 63 , Leu 69 , Met 71 , Phe 89 , Ala 103 , Thr 110 , Glu 120 , Glu 123 , Ile 125 , and Ala 128 , whereas significantly perturbed residues (defined as moving by more than one S.D. from the mean) include Phe 19 , Val 55 , Ala 57 , Thr 70 , Met 72 , Ile 85 , Val 91 , Lys 94 , Arg 106 , Met 109 , Lys 115 , Ile 130 , and Lys 148 . Interestingly, when these residues are mapped onto the crystal structure of Ca 2ϩ / CaM (53), they are all located in the hydrophobic clefts of the two Ca 2ϩ -binding lobes of Ca 2ϩ /CaM (Fig. 6B) with ϳ70% of affected peaks arising from the C-terminal lobe. In contrast, resonance peaks belonging to the interconnecting central helix (residues 73-84) show little perturbation even at the highest concentration of Cpep.
It has been reported previously that the C-lobe of Ca 2ϩ /CaM commonly acts as the initial recognition site and binds more tightly than the N-lobe in other CaM-peptide and CaM-protein interactions (6,9). To calculate the affinity of each Ca 2ϩ -binding lobe for Cpep, binding curves were constructed by following the migration of fast exchange peaks from each lobe. The peak position changes were fitted to a 1:1 binding model by nonlinear least squares regression (for examples, see Fig. 6C). In summary, all but one of the N-lobe residues examined gave dissociation constants in the range of ϳ24 -60 M, whereas all C-lobe residues gave dissociation constants of ϳ5-10 M ( Table 2), suggesting that the C-lobe does bind Cpep slightly more tightly than the N-lobe. The relatively large variations in the fitted dissociation constants and the associated errors probably reflect that the exchange regime is not strictly fast (but rather fast-intermediate), leading to errors with peak position determination. Nonetheless, the overall binding affinity for the C-terminal lobe as determined by NMR is consistent with binding affinities determined from the tryptophan fluorescence experiments (Table 1).
To confirm that the binding mode of Ca 2ϩ /CaM is conserved between its interaction with Cpep and in the context of cMyBP-C, the 15 N HSQC titration experiment was also carried out using the larger construct C1C2. Upon the addition of C1C2 to 15 N-labeled Ca 2ϩ /CaM, many peaks in the 15 N HSQC spectrum experienced a severe drop in signal intensity and disappeared (Fig. 7A). The 15 amide resonance peaks that were deemed to have experienced significant intensity changes (namely Ser 17 , Phe 19 , Thr 29 , Leu 48 , Ala 57 , Gly 59 , Ile 63 , Glu 67 , Leu 69 , Thr 70 , Met 71 , Met 72 , Ala 88 , Leu 105 , Glu 120 , Glu 123 , Ile 130 , and Met 144 ) when bound to C1C2 again mapped specifically to the hydrophobic clefts in each of the two lobes of CaM, similar to the result for Cpep (Fig. 7B). Therefore, the binding mode is likely to be conserved between Cpep and C1C2.
Ca 2ϩ /CaM Has an Extended Conformation in Its Complex with Cpep-SAXS was used to determine the solution conformation of the Ca 2ϩ /CaM-Cpep complex. The scattering profiles of Ca 2ϩ /CaM and its complex with Cpep are shown in Fig.  8A, and the associated structural parameters are summarized in Table 3. Guinier plots for Ca 2ϩ /CaM and Ca 2ϩ /CaM-Cpep (Fig. 8B, inset) are linear, consistent with the proteins in solution being predominantly monodisperse. AutoPorod (54) calculations yield volumes consistent with the expected molecular mass based on amino acid content. The probable distribution of distances between atom pairs (P(r) profiles calculated using GNOM (55) ; Fig. 8B) yields forward scattering intensity (I(0)) and radius of gyration (R g ) values in good agreement with those determined from the Guinier plots as well as the maximum linear dimension D max for the scattering particles. The structural parameters of Ca 2ϩ /CaM (R g ϭ 22.3 Å, D max ϭ 70 Å) agree reasonably with previously reported values (42,56,57) except that the R g value is ϳ1 Å (i.e. 5%) larger, probably due to the high salt concentrations in these samples. The Ca 2ϩ /CaM-Cpep complex has R g and D max values of 23.3 and 75 Å, respectively, each slightly larger than those for the unbound Ca 2ϩ / CaM. Thus, unlike the classical short helical peptides that bind CaM, there is no conformational collapse of CaM upon binding Cpep. Consistent with this, the P(r) profile of Ca 2ϩ /CaM-Cpep is similar in shape to that for CaM alone, although with a less pronounced shoulder at r ϳ45 Å (Fig. 8B) that is probably accounted for by the fact that the peptide occupies space that results in less of a "dumbbell" shape compared with CaM alone. Ab initio shape reconstructions (Fig. 8C) also show that Ca 2ϩ / CaM-Cpep is similarly extended as the unbound Ca 2ϩ /CaM with the extra mass evident that can be attributed to the bound Cpep. The scattering data are thus consistent with the NMR results indicating that Cpep interacts with both lobes of Ca 2ϩ /CaM. It is well documented for Ca 2ϩ /CaM that whereas the two hydrophobic lobes are folded into globular domains, the central For each titration data set, the data points and fitted curves were obtained using the same approaches described in Fig. 3. For clarity of representation, each titration data set has been scaled by multiplying the following constants in the (F/F 0 ) values: motif, 0.9 (•); motif in 50 mM NaCl, 0.7 (f); Cpep, 0.2 (OE); and Npep, 0.1 (ϫ). Each data point is the average of eight replicate measurements after correcting for solvent; the error bars (not shown; corresponding to Ϯ 1 S.D.) are smaller than the symbols. All titrations were performed in the standard buffer except for "motif in 50 mM NaCl" data, which used a reduced NaCl concentration (see details under "Experimental Procedures").
helix contains a flexible region (residues 77-81) that facilitates the two lobes binding peptide sequences of different lengths and composition (6,9). Kratky plots of Ca 2ϩ /CaM and Ca 2ϩ / CaM-Cpep complex are consistent with largely folded globular proteins with some flexibility (data not shown). It appears that this flexibility is retained to a similar degree in the Ca 2ϩ /CaM-Cpep complex as in Ca 2ϩ /CaM alone, consistent with the NMR titration experiments with 15 N-labeled Ca 2ϩ /CaM-Cpep showing that residues within the central helix do not experience large chemical shift changes when the two lobes bind to Cpep.
The Interaction between Ca 2ϩ /CaM and Motif Is Phosphorylation-independent-Typically C1C2 is unphosphorylated following expression in bacteria (58). To determine the specific role of the phosphorylation-induced change in charge, we generated a mutant that is a triphosphomimic, C1C2EEE, by substituting glutamate (E) for serine (S) at the CaMKII (S284E) and PKA phosphorylation sites (S275E and S304E) of human C1C2 (Fig. 1C). Interestingly, C1C2EEE binds to Ca 2ϩ /CaM in a very similar fashion and with an affinity comparable with that of the wild-type C1C2 as monitored by tryptophan fluorescence ( Fig. 3 and Table 1). The 1 H-15 N HSQC spectra of [ 15 N]C1C2EEE and [ 15 N]C1C2 overlay very well (data not shown) with few differences, which are located mainly in the region associated with disordered structure and probably corresponded to residues in close proximity to the three Ser to Glu mutations. It appears that these mutations do not induce large conformational changes in the C1 and C2 domains or in the  Table 2 for corresponding K D values).

DISCUSSION
By combining pull-down assays, SEC-MALLS, tryptophan fluorescence, and NMR titration data, we have identified a novel interaction between the N-terminal fragment C1C2 of human cMyBP-C and CaM. Previous co-sedimentation assays have failed to detect this interaction (59). The 1:1 stoichiometric interaction is Ca 2ϩ -dependent and independent of ionic strength. Furthermore, we conclude that the motif that links the C1 and C2 modules is required for the complex formation and that a 23-residue sequence (i.e. residues Glu 319 -Lys 341 ) at the C-terminal end of the motif is sufficient for maintaining this  interaction. This 23-residue peptide (Cpep) has no obvious consensus with any reported CaM-target sequences identified to date (5,6,51). Nevertheless, the formation of a Ca 2ϩ /CaM-C1C2 complex appears to be specific because NMR titration data confirm that a number of hydrophobic residues from both motif and Ca 2ϩ /CaM are involved in the complex formation. The interaction of Ca 2ϩ /CaM with the motif is via the hydrophobic clefts within the two Ca 2ϩ -binding lobes of CaM, as has been observed in all CaM-target complexes (5,6,9), and as is commonly observed the C-lobe of CaM appears to interact slightly more strongly with the motif than the N-lobe does. In addition, our results show that basic residues and methionine residues of 15 N-labeled Ca 2ϩ /CaM are among the residues that undergo large chemical shift changes during titrations with Cpep, indicating that these residues are important for the complex formation, again consistent with previous suggestions that they act to optimize the contacts between CaM and the hydrophobic interface of its target protein (6,9). It is likely that the tryptophan residue (Trp 322 ) within the motif is involved in anchoring to Ca 2ϩ /CaM, similar to what has been observed in other known CaM-target peptides (6, 9 -14). It is noteworthy that residue Trp 322 , together with most of the residues included in the 23-residue Cpep sequence, is highly conserved between isoforms and species of myosin-binding protein C (19,24), arguing that its ability to bind to Ca 2ϩ /CaM could be a crucial function of the protein. Consistent with this idea, seven mutations linked to hypertrophic cardiomyopathy have been identified in the gene encoding this 23-residue sequence (including single amino acid substitutions, such as R326Q, E334K, V342D, and L352P, and frameshift mutations at positions, such as Gln 327 , Thr 343 , and Met 348 ) (60).
In the context of CaM-binding sequences commonly being helical, it is interesting that the equivalent Cpep sequence forms two short ␣-helices within a three-helical bundle in the mouse motif structure (50) (Fig. 1B). In this configuration, residue Trp 322 lies within the hydrophobic core of the three-helical bundle, whereas for the shorter Cpep that does not include the third helix, Trp 322 is likely to be more solvent-exposed. Consistent with this idea, the tryptophan fluorescence intensity increases for Cpep as it binds to Ca 2ϩ /CaM, whereas intensity decreases are observed for all constructs containing an intact motif upon binding (Figs. 3 and 5). Unlike the classic collapsed conformations of CaM observed when binding short helical target sequences (6,9), the SAXS data show that Cpep binds to an extended Ca 2ϩ /CaM. Of note, the maximum length of the two helices of Cpep (ϳ38 Å) is sufficient to span the two Ca 2ϩbinding lobes of CaM in its extended formation. Taken together, our data suggest that CaM probably induces conformational changes within the structured three-helical bundle of the motif to expose and position Trp 322 for binding; one way this could be achieved is through hinge motions at the junctions connecting the helices.
At Ca 2ϩ concentrations below the threshold for muscle contraction, Ca 2ϩ -regulated changes in phosphorylation of cMyBP-C are related to changes in cardiac contractility (28). Activation of CaMKII requires Ca 2ϩ /CaM for autophosphorylation prior to its phosphorylating Ser 282 in the motif (19,61). Thus, the Ca 2ϩ -dependent interaction between the motif and CaM may be functionally important to the timing and sequence of phosphorylation events that regulate cMyBP-C attachment and detachment from myosin ⌬S2. Interestingly, the interaction reported here is independent of the phosphorylation state of the motif because both C1C2 and the triphosphomimic C1C2EEE bind to Ca 2ϩ /CaM with micromolar affinity. C1C2 (predominantly via the motif) has also been found to bind to myosin ⌬S2 with a K D of ϳ5 M (24), which is comparable with its binding affinity with Ca 2ϩ /CaM as identified in this study. However, the interaction between C1C2 and myosin ⌬S2 can be abolished by phosphorylation of the motif (23). Therefore, the motif appears to interact with Ca 2ϩ /CaM in a region structurally independent from the three phosphorylation sites so that phosphorylation or mutation of these sites has no impact on its interaction with Ca 2ϩ /CaM. Consistent with this idea, we have identified residues involved in direct interactions with Ca 2ϩ /CaM clustered at the structured C-terminal end of the motif.
In light of the fact that Ca 2ϩ /CaM binds to the motif independently of its apparent phosphorylation state, it is interesting to consider the idea that this interaction could provide a means to store and readily supply Ca 2ϩ /CaM. For example, CaMKII requires Ca 2ϩ /CaM to initiate the first phosphorylation event necessary for additional phosphorylation events in the motif. Alternatively, when C1C2 is bound to myosin ⌬S2, the bound Ca 2ϩ /CaM on the motif may also facilitate phosphorylation of neighboring myosin RLC via the Ca 2ϩ /CaM-dependent MLCK to promote myosin cross-bridge formation (34,62). The micro- molar range binding affinity between motif and Ca 2ϩ /CaM is weaker than affinities reported for most CaM-target proteins (K D values between 0.1 M and 0.01 nM) (6,52); therefore, these CaM-regulated proteins, in particular CaMKII and MLCK, which are responsible for modulating contractile events, could compete with the motif for Ca 2ϩ /CaM. Upon autophosphorylation of CaMKII or MLCK, CaM would be released and bind to the motif of cMyBP-C for storage.
The role of the interaction between Ca 2ϩ /CaM and the motif of cMyBP-C reported in this study has yet to be fully understood in the context of the complex molecular networks that regulate muscle contraction. Nonetheless, our observations provide the first evidence that CaM can link cMyBP-C with the Ca 2ϩ signaling pathways in muscle, which opens the possibility of the interaction playing a role in synchronizing key regulatory phosphorylation events.