Site-specific phosphorylation of myosin binding protein-C coordinates thin and thick filament activation in cardiac muscle.

The heart's response to varying demands of the body is regulated by signaling pathways that activate protein kinases which phosphorylate sarcomeric proteins. Although phosphorylation of cardiac myosin binding protein-C (cMyBP-C) has been recognized as a key regulator of myocardial contractility, little is known about its mechanism of action. Here, we used protein kinase A (PKA) and Cε (PKCε), as well as ribosomal S6 kinase II (RSK2), which have different specificities for cMyBP-C's multiple phosphorylation sites, to show that individual sites are not independent, and that phosphorylation of cMyBP-C is controlled by positive and negative regulatory coupling between those sites. PKA phosphorylation of cMyBP-C's N terminus on 3 conserved serine residues is hierarchical and antagonizes phosphorylation by PKCε, and vice versa. In contrast, RSK2 phosphorylation of cMyBP-C accelerates PKA phosphorylation. We used cMyBP-C's regulatory N-terminal domains in defined phosphorylation states for protein-protein interaction studies with isolated cardiac native thin filaments and the S2 domain of cardiac myosin to show that site-specific phosphorylation of this region of cMyBP-C controls its interaction with both the actin-containing thin and myosin-containing thick filaments. We also used fluorescence probes on the myosin-associated regulatory light chain in the thick filaments and on troponin C in the thin filaments to monitor structural changes in the myofilaments of intact heart muscle cells associated with activation of myocardial contraction by the N-terminal region of cMyBP-C in its different phosphorylation states. Our results suggest that cMyBP-C acts as a sarcomeric integrator of multiple signaling pathways that determines downstream physiological function.

Contraction of cardiac muscle is initiated by activation of the actin-containing thin filaments, but is modulated by structural changes in the myosin-containing thick filaments. Calcium binding to troponin induces an azimuthal movement of tropomyosin on the surface of the thin filaments which allows myosin head domains from the neighboring thick filaments to strongly attach to actin (1). Subsequently, small conformational changes in the actin-attached myosin catalytic domain are amplified by the essential and regulatory light chain-containing myosin light chain domain or “lever arm” associated with the release of Adenosine 5′-triphosphate hydrolysis products (2, 3). This “working stroke” produces piconewton-scale force and nanometer-scale displacement of the thin filaments toward the center of the sarcomere.

Heart muscle contractility is also regulated by posttranslational modifications of sarcomeric proteins, including phosphorylation of the regulatory components of the thick filaments (4). Phosphorylation of these components has been widely implicated in the regulation of cardiac output, and altered phosphorylation levels have been frequently associated with heart failure (5), further underlining their functional significance. In the current study, we focused on phosphorylation of cardiac myosin binding protein-C (cMyBP-C), a thick filament-associated protein with important regulatory functions in both healthy and diseased states of the heart. The functional significance of cMyBP-C phosphorylation is highlighted by the fact that ablation of either cMyBP-C or its phosphorylation leads to pathological hypertrophy in animal models, suggesting that cMyBP-C phosphorylation is essential for normal heart function (6, 7).

cMyBP-C is localized to 9 transverse stripes in the central region of each half-thick filament, called the C-zone, via interactions of its C-terminal anchoring region with the myosin tails and titin (Fig. 1), closely matching the ∼43-nm periodicity of the myosin head domains. In contrast, interactions of its regulatory N-terminal domains are less well defined, and binding sites for both myosin and actin have been identified in vitro (8). Myosin interactions of cMyBP-C’s N terminus are generally associated with an inhibitory effect on contractility, and both structural and functional studies suggest that cMyBP-C stabilizes the thick filament OFF state by tethering myosin head domains to the surface of the thick filament backbone (9, 10). In contrast, cMyBP-C’s N-terminal domains have also been shown to bind actin and activate the thin filament presumably by moving tropomyosin away from its blocked position, which increases the calcium sensitivity of its regulatory units (11, 12).

Fig. 1.

Hierarchical phosphorylation of C1mC2 by PKA. (A, Top) Diagram of the sarcomere. The thick filament D-, C-, and P-zones and M-line are labeled accordingly. (A, Bottom) Domain organization and known protein interactions of cMyBP-C. The proline/alanine-rich linker (P/A) between domains C0 and C1 and the m-motif (m) between domains C1 and C2 are labeled accordingly. Sequence alignment of the m-motif residues encompassing the conserved phosphorylation sites (red) is shown below, with canonical PKA consensus sequences (RRXS) underlined. Serines are numbered according to the rat cMyBP-C sequence. (B) Time (t)-dependent in vitro phosphorylation of C1mC2 by PKA analyzed by Phos-tag–SDS/PAGE. (C) Chromatogram demonstrating the separation of phosphorylated C1mC2 constructs (0P, 1P, 2P, and 3P) by IEC on the Resource S column. Blue, green, and red traces correspond to phosphorylation reactions stopped at different time points. (D) MS analysis of phosphorylated residues in IEC fractions shown in C. Example MALDI-MS/MS spectra for tris-phosphorylated C1mC2 (C1mC2-3P, blue) and bis-phosphorylated C1mC2 (C1mC2-2P, green) are shown with peaks labeled with their corresponding phosphorylated peptide sequences. Note that the peak at the 2,236.126 mass-to-charge ratio (m/z) corresponding to S313-phosphorylated peptide found in C1mC2-3P is missing in the C1mC2-2P spectrum (indicated by a red box). The peak at 1,256.671 m/z is not shown in the C1mC2-2P spectrum for clarity. Interpretation of the MS data is summarized in the box on the right. Details are provided in text. In vitro kinase assays with S-A–substituted C1mC2 analyzed by Phos-tag–SDS/PAGE (E) and deconvolution into phosphorylation of individual serine residues (F) are shown. Mean ± SEM (n = 6). Statistical significance of differences between values was assessed with a one-way ANOVA followed by Tukey’s post hoc test: *P < 0.05, ***P < 0.001.

Both the inhibitory and activating interactions of cMyBP-C are believed to be controlled by its phosphorylation state. The cardiac specific “m-motif” between domains C1 and C2 contains a series of conserved serine residues that are phosphorylated by several protein kinases in vivo and in vitro (13) (Fig. 1). Protein kinase A (PKA) has been identified as the primary kinase acting upon cMyBP-C (14), mediating some of the effects of β-adrenergic stimulation on myocardial function, such as increased cross-bridge kinetics, decreased calcium sensitivity, and accelerated relaxation (15, 16). More recently, other protein kinases have been shown to phosphorylate cMyBP-C with distinct specificities for m-motif serines, suggesting that each phosphorylation site might have distinct regulatory effects (13).

However, the detailed function and mechanism underlying cMyBP-C phosphorylation at individual sites remain poorly understood, likely due to the complexity of its interaction with other sarcomeric components, which is further compounded by the multiplicity of signaling pathways acting upon cMyBP-C and its associated phosphorylation states (17). Current mechanistic hypotheses of the effects of cMyBP-C phosphorylation are largely based on in vivo experiments in animal models (6, 18) or tissues derived from those animals (19) in which phosphorylation levels and associated regulatory mechanisms cannot be controlled at the molecular level. Moreover, serine-to-aspartate substitutions have been frequently used as a convenient model for phosphorylation, although recent studies showed that these substitutions only partially recapitulate the effects of phosphorylation (20). Conversely, in vitro experiments with fully phosphorylated proteins or proteins containing serine-to-aspartate substitutions do not recapitulate the in vivo complexity of cMyBP-C’s interactions and its dynamic phosphorylation (21–23).

The aim of the present work was to understand the molecular mechanism of cMyBP-C phosphorylation and its structural and functional effects on both thin and thick filament-based regulation in heart muscle cells. We investigated the relationship of individual cMyBP-C phosphorylation sites and potential cross-talk using several protein kinases known to phosphorylate cMyBP-C. We utilized cMyBP-C fragments with well-characterized phosphorylation states to show that site-specific phosphorylation has distinct effects on its interaction with and regulation of the thin and thick filaments, combining in vitro biochemical binding assays with in situ structural measurements in intact heart muscle cells. The results show that cMyBP-C acts as a sarcomeric integrator of different signaling pathways to determine downstream physiological effects, and that the functional effects of cMyBP-C phosphorylation can only be understood by combining thin and thick filament-based mechanisms into an integrated model of contractile regulation.

Results

Hierarchical Phosphorylation of cMyBP-C by PKA.

To elucidate the regulatory function of cMyBP-C phosphorylation, we phosphorylated a recombinant fragment of rat cMyBP-C containing domains C1, the phosphorylatable m-motif, and C2 (C1mC2; Fig. 1) with the catalytic subunit of PKA in vitro (Fig. 1). The C1mC2 fragment is a convenient model for studying the function of full-length cMyBP-C, particularly with respect to its effects on both thin and thick filament structure (11, 12, 20) and its regulation by phosphorylation (20, 24, 25). Although rodent and human C1mC2 have a high sequence identity (>92%), suggesting a conserved molecular function, species-specific effects cannot be excluded (26).

Incubation of C1mC2 for 30 to 60 min with PKA resulted in a mixture of unphosphorylated, monophosphorylated, bis-phosphorylated, and tris-phosphorylated protein, which could be clearly separated by both Phos-tag and sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS/PAGE) (Fig. 1). Individual C1mC2 phospho-species were isolated by stopping the kinase reaction at different time points and separating the phosphorylated proteins by ion-exchange chromatography (IEC) (Fig. 1). The phosphorylation level and homogeneity (>95%) of each IEC fraction (e.g., 1P, 2P, 3P) were confirmed by electrospray ionization (ESI) mass spectrometry (MS) (). Tetrakis-phosphorylated C1mC2 was only observed after prolonged incubation with PKA at 30 °C, suggesting that the fourth site is a poor substrate for PKA.

Phosphorylated amino acid residues in each IEC fraction were identified by proteolytic digestion followed by phospho-peptide enrichment and matrix-assisted laser desorption ionization (MALDI) MS and ESI-MS () (details are provided in ). Analysis of the IEC fraction corresponding to monophosphorylated C1mC2 (C1mC2-1P) revealed specific phosphorylation on a single serine residue, S288, in agreement with previous studies suggesting that S288 in the cardiac-specific insertion is the initial PKA phosphorylation site (14). Bis-phosphorylated C1mC2 (C1mC2-2P) contained phosphorylated serine residues only in positions 288 and 279, and tris-phosphorylated C1mC2 (C1mC2-3P) contained phosphorylated serines only in positions 288, 279, and 313. As an example, the MALDI-MS spectrum of C1mC2-3P is shown in Fig. 1, with peaks corresponding to identified phospho-peptides labeled accordingly. The peak at a 2,236.126 mass-to-charge ratio corresponding to the S313-phosphorylated peptide is missing in the MALDI-MS spectrum of the bis-phosphorylated C1mC2 (C1mC2-2P) (Fig. 1 , Inset, Top Left, red box), suggesting that S313 is phosphorylated after S288 and S279. In contrast, phosphorylation of S308 by PKA was only observed after prolonged incubation, suggesting that this serine is a poor substrate for PKA in vitro. Thus, our results suggest that PKA phosphorylation of the cardiac-specific m-motif follows a concerted hierarchical mechanism in the sequence of S288, followed by S279, followed by S313 (Fig. 1 , Inset, Top Right).

To further test this conclusion, we prepared serine-to-alanine (S-A) substitutions of each individual PKA site in C1mC2, and analyzed their phosphorylation profiles by Phos-tag–SDS/PAGE (Fig. 1). Total phosphate incorporation after 60 min of incubation was strongly reduced in S288A-substituted C1mC2 (1.25 ± 0.08 mol of inorganic phosphate [Pi]/mol [mean ± SEM]; n = 6; ) compared with wild type (2.94 ± 0.18 mol of Pi/mol [mean ± SEM]; n = 6). S279A- and S313A-substituted C1mC2 showed intermediate levels of phosphorylation, although S279A had a stronger inhibitory effect than S313A (1.6 ± 0.11 and 2.17 ± 0.12 mol of Pi/mol, respectively), in agreement with the hierarchical phosphorylation sequence proposed above.

Next, we analyzed the effects of S-A substitutions on phosphate incorporation by PKA at each individual phosphorylation site (Fig. 1). Consistent with the hierarchical model, phosphorylation of S288 was not significantly affected by substitution of either S279 or S313 by alanine. Similarly, S279 phosphorylation was not inhibited by substitution of either S288 or S313 by alanine, although in the native fragment, this residue is phosphorylated after S288. This suggests that unphosphorylated S288 may be involved in intramolecular interactions that prevent access to S279, and that substitution of S288 by alanine abolishes this interaction. If so, substitution of S288 by alanine might therefore partially mimic phosphorylation of this site. In contrast, S313 phosphorylation was inhibited by substitution of either S288 or S279 to alanine, further supporting the proposal that this residue is phosphorylated downstream of S288 and S279. Contrary to the effects of S-A substitutions described above, PKA phosphorylation of S308 was greatly increased in C1mC2-S313A (∼50%) compared with wild-type control (∼10%), suggesting that phosphorylation of S313 per se has an inhibitory effect on phosphorylation of S308.

cMyBP-C Phosphorylation by Non-PKA Kinases Reveals Regulatory Coupling of Phosphorylation Sites.

Although PKA is considered to be the primary kinase acting upon cMyBP-C in vivo, several other kinases have been shown to be able to phosphorylate specific residues within the cardiac-specific m-motif (13). Moreover, these kinases have been frequently shown to be activated during heart disease, underlining their functional significance.

Ribosomal S6 kinase II (RSK2) has been suggested to specifically phosphorylate serine residues in the cardiac-specific m-motif insertion in response to activation of the mitogen-activated protein kinase/extracellular signal-regulated kinase pathway (27), and we used site-specific S-A substitution and MS to confirm these results (Fig. 2 and ). RSK2 primarily phosphorylated rat C1mC2 at S288 in vitro, although we observed phosphorylation of other residues at higher enzyme-to-substrate ratios or after prolonged incubation. In contrast, α-adrenergic stimulation leads to an increase in cMyBP-C phosphorylation primarily via stimulation of protein kinase Cε (PKCε) (28, 29) [and protein kinase D (PKD) (30)] activity. We used site-directed mutagenesis, MS, and Western blot analysis to confirm that PKCε specifically phosphorylates S308 in C1mC2 in vitro (Fig. 2 and ). Phosphorylation of S288 was not a prerequisite for phosphorylation of S308 by PKCε as shown previously (31).

Fig. 2.

C1mC2 phosphorylation by RSK2 and PKCε. (A) RSK2-mediated phosphorylation of S288 in rat C1mC2 was confirmed by in vitro kinase (IVK) assays using wild-type and S288A- substituted C1mC2 as well as MALDI-MS phosphorylation site profiling. m/z, mass-to-charge ratio. (B) PKCε phosphorylation of S308 in rat C1mC2 was confirmed by both IVK assays using phosphoablated C1mC2-S308D and ESI-MS. (C) Western blot analysis of RSK2- and PKCε-phosphorylated C1mC2 using PKA site (RRXpS)-specific and S288 (pS288)-specific antibodies. Note that PKCε-phosphorylated C1mC2 is not recognized by either of the antibodies, confirming S308 as the main phosphorylation site. (D) PKA IVK assays with unphosphorylated, S288-phosphorylated, or S308-phosphorylated C1mC2 analyzed by Phos-tag–SDS/PAGE. t, time. Analysis of individual site phosphorylation (E) and total level of phosphate incorporation by PKA (F) are shown. Mean ± SEM (n = 3 to 9). Statistical significance of differences between values was assessed with a one-way ANOVA followed by Tukey’s post hoc test: *P < 0.05, **P < 0.01.

Next, we investigated potential cross-talk between cMyBP-C–mediated phosphorylation by RSK2, PKCε, and PKA using purified C1mC2 fragments site-specifically phosphorylated by RSK2 and PKCε as substrates in PKA kinase assays. As expected from the sequential phosphorylation of C1mC2 described above, S288 phosphorylation by RSK2 significantly accelerated phosphorylation of S279 by PKA, but had no effect on phosphorylation of residues downstream of S279 (i.e., S313) (Fig. 2 ). In contrast, phosphorylation of S308 by PKCε decreased the rate of PKA phosphorylation of S313, suggesting that S308P is a negative regulator of cMyBP-C PKA phosphorylation of S313, consistent with previous results in isolated cardiomyocytes (32). Moreover, both S288 by RSK2 and S308 by PKCε phosphorylation reduced overall phosphate incorporation into C1mC2 by PKA by ∼1 mol of Pi/mol of C1mC2 (Fig. 2). Although this is expected for phosphorylation of S288 as a main PKA site, S308 is primarily phosphorylated by PKCε and PKD, further confirming the antagonistic effects of PKA- and PKCε/PKD-mediated phosphorylation of cMyBP-C.

In summary, the multiple cMyBP-C phosphorylation sites are not independent, and exhibit positive and negative regulatory coupling with partially overlapping specificity for multiple kinases.

Site-Specific Phosphorylation Controls Thick and Thin Filament Binding of C1mC2.

To characterize the effects of site-specific cMyBP-C phosphorylation on its proposed interactions with both the myosin-containing thick and actin-containing thin filaments (12, 20), we determined the binding affinities of C1mC2 in its different phosphorylation states for isolated myosin S2 and native thin filaments (NTFs) in vitro. The affinity of C1mC2 for its binding site on myosin, the first 126 amino acids of myosin S2 (S2Δ), was characterized by microscale thermophoresis (MST). As previously shown, unphosphorylated C1mC2 binds S2Δ in a biphasic manner, corresponding to high- and low-affinity binding sites with dissociation constants (Kds) of 20.7 ± 3.5 μmol/L (mean ± SEM; n = 6) and >400 μmol/L, respectively (20). Moreover, the Kd for myosin S2Δ of λ-protein phosphatase–treated native full-length cMyBP-C isolated from rat ventricular tissue was 24.2 ± 3.9 μmol/L (mean ± SEM; n = 4) (), in good agreement with the results obtained for recombinant rat C1mC2.

Monophosphorylation of either S308 or S288 slightly reduced C1mC2’s affinity for myosin S2Δ as indicated by an increase in Kd to ∼30 μmol/L (Fig. 3 and ). In contrast, both PKA bis-phosphorylation (S288 and S279) and tris-phosphorylation (S288, S279, and S313) completely abolished S2Δ binding, indicating that either bis-phosphorylation or phosphorylation of S279 per se controls the cMyBP-C–myosin S2 interaction. We addressed this question by phosphorylating C1mC2-S288A with PKA and isolating the monophosphorylated species (C1mC2-S288A-1P) using IEC. Both unphosphorylated and monophosphorylated C1mC2-S288A bind myosin S2Δ with a Kd similar to that measured for the wild-type protein (Kd of ∼20 μmol/L), suggesting that bis-phosphorylation is necessary and sufficient to abolish cMyBP-C–myosin S2 interaction (Fig. 3 and ). To further test this conclusion, we sequentially phosphorylated C1mC2 with RSK2 and PKCε, and measured the affinity of the bis-phosphorylated C1mC2 (S288 and S308) for myosin S2Δ (). The RSK2/PKCε bis-phosphorylated C1mC2 did not bind to myosin S2Δ, further supporting the hypothesis that bis-phosphorylation per se, independent of the phosphorylation site combination, abolishes cMyBP-C–myosin S2 interaction. These results are consistent with the largely ionic nature of cMyBP-C’s main interaction site in the m-motif and myosin S2 (20).

Fig. 3.

Effects of site-specific phosphorylation on binding of C1mC2 to myosin S2Δ and NTFs. (A) Normalized MST curves for C1mC2 and its different phosphorylation states titrated against myosin S2Δ. (B) Normalized MST binding curves for C1mC2-S288A (brown) and C1mC2-S288A-1P (black) titrated against myosin S2Δ. NTF cosedimentation data for C1mC2 and its different phosphorylation states in the absence (C, pCa 9) and presence (D, pCa 4.5) of 32 μmol/L CaCl2 are shown. Calculated Kd (E) and Bmax (F) values for the NTF binding assays in the absence (filled circles) and presence (open circles) of CaCl2 are shown. Mean ± SEM (n = 4 to 10). Statistical significance of difference was assessed with a one-way ANOVA followed by Tukey’s post hoc test: *P < 0.05, **P < 0.01.

We measured the affinity of C1mC2 in its different phosphorylation states for isolated bovine cardiac NTFs using high-velocity cosedimentation. Unphosphorylated C1mC2 (0P) binds NTFs in the absence of calcium (pCa 9) in a saturable manner with a Kd of ∼20 μmol/L and a maximal binding capacity (Bmax) of ∼1, indicating stoichiometric binding of C1mC2 to actin (Fig. 3 and ). Addition of calcium (pCa 4.5) had no effect on C1mC2 binding to NTFs as indicated by identical Kd and Bmax values (Fig. 3 and ). Strikingly, both PKA monophosphorylation (S288) and bis-phosphorylation (S288 and S279), as well as monophosphorylation of S308 by PKCε, showed no change in the affinity of C1mC2 for NTFs (Kd ∼ 20 μmol/L), but decreased Bmax to ∼0.5, suggesting a lower binding capacity of C1mC2 in the partially phosphorylated states. As before, full calcium activation of NTFs had no additional effect on either Kd or Bmax. However, PKA tris-phosphorylation (3P) significantly decreased C1mC2’s affinity for NTF with an estimated Kd of ∼60 μmol/L independent of [Ca2+]. Addition of calcium increased Bmax from ∼0.5 to ∼1 for C1mC2-3P, suggesting an interplay between tris-phosphorylation of C1mC2, thin filament binding, and calcium activation (23).

These results suggest that cMyBP-C phosphorylation regulates contractility partly via differential modulation of its affinity for myosin and actin binding sites, so that phosphorylation leads to a redistribution of cMyBP-C’s N-terminal domains from their myosin binding sites in the thick filaments to their actin binding sites in the thin filaments. Monophosphorylation weakens and bis-phosphorylation abolishes thick filament binding, and only tris-phosphorylation affects C1mC2 binding to regulated thin filaments.

Site-Specific Phosphorylation of C1mC2 Controls Its Effect on Thin and Thick Filament Structure.

Next, we used a bifunctional rhodamine probe attached to the E-helix of cardiac troponin C (cTnC-E) to monitor structural changes in the thin filaments of demembranated ventricular muscle cells associated with the activating effect of C1mC2 described previously (12). Unphosphorylated C1mC2 activates the force and thin filament structure of ventricular trabeculae in the absence of Ca2+ (pCa 9) with a half-maximal effective concentration (EC50) of ∼20 μmol/L (Fig. 4, filled red circles). Maximum isometric force at [C1mC2] = 40 μmol/L is only ∼60% of that measured during Ca2+ activation in the absence of C1mC2 (Fig. 4), although the level of thin filament activation as reported by the cTnC probe is significantly higher, suggesting that exogenous C1mC2 has both activating and inhibitory effects on contractility.

Fig. 4.

Effects of site-specific phosphorylation of C1mC2 on active force and thin and thick filament structure in ventricular trabeculae. (A) Concentration-dependent effect of C1mC2 on force generation (black squares) and thin (red filled circles) and thick (red open circles) filament structure in ventricular trabeculae. TnC and RLC probe orientations are expressed as the order parameter

, which is +1 for a probe dipole orientation parallel to the filament axis and −0.5 for a perpendicular orientation.

for full calcium activation is shown in orange on the left axis. (B) Isometric force of ventricular trabeculae in the presence of 40 μmol/L C1mC2 in its different phosphorylation states. ACT, activating solution (pCa 4.5); REL, relaxing solution (pCa 9). TnC (C) and RLC (D) probe orientations of ventricular trabeculae in the presence of 40 μmol/L C1mC2 in its different phosphorylation states are shown. Mean ± SEM, with the number of trabeculae (n), is indicated in each panel. Statistical significance of differences between groups was assessed with a one-way ANOVA followed by Tukey’s post hoc test: *P < 0.05, **P < 0.01, ***P < 0.001. ns, not significant.

Effects of site-specific phosphorylation of C1mC2 on active force and thin and thick filament structure in ventricular trabeculae. (A) Concentration-dependent effect of C1mC2 on force generation (black squares) and thin (red filled circles) and thick (red open circles) filament structure in ventricular trabeculae. TnC and RLC probe orientations are expressed as the order parameter