|Home | About | Journals | Submit | Contact Us | Français|
Homeostasis of cardiac function requires significant adjustments in sarcomeric protein phosphorylation. The existence of unique peptides in cardiac sarcomeres, which are substrates for a multitude of kinases, strongly supports this concept (1). We focus here on the troponin complex of the thin filaments, which contain two major proteins that participate in these phosphoryl group transfer reactions: the inhibitory protein (cardiac troponin (cTn)2 I) and the tropomyosin (Tm)-binding protein (cTnT). We describe the relatively new understanding of the molecular mechanisms of thin filament-based control of the heartbeat and how these mechanisms are altered by phosphorylation. We discuss new concepts regarding the relation between the beat of the heart and the location of thin filament proteins and their long- and short-range interactions. We also discuss elucidation of mechanisms by which these phosphorylations exacerbate or ameliorate effects of mutations in the myofilament proteins that are linked to familial cardiomyopathies.
Fig. 1 depicts an A-band region of the cardiac thin filament functional unit in the diastolic and systolic states. In diastole, force-generating reactions of cross-bridges with actin are inhibited, ATP hydrolysis is relatively low, and the sarcomere is relatively extensible (2). Properties of the giant protein titin dominate the compliance of the relaxed sarcomere (3). Interactions of thin filament regulatory proteins, the troponin heterotrimeric complex, and Tm hinder the actin-cross-bridge reaction and establish the B-state. Calcium binding to a single regulatory site on cTnC triggers a release from this inhibited state by modifications of interactions among actin, Tm, and Tn.
Evidence derived from the core crystal structure of cardiac Tn (4), from elucidation of the structures by NMR (5), from biochemical investigations of protein-protein interactions (6, 7), and from reconstructions and single-particle analysis of electron micrographs of reconstituted myofilament preparations (8, 9) provided the basis for the illustration in Fig. 1 (see Ref. 6 for a review). Apart from the lack of a thin filament lattice in the Tn core crystal structure, there was no structural information on significant regions, including the tail region of cTnT, an inhibitory peptide (Ip; which tethers cTnI to actin), the unique N-terminal peptide (~30 amino acids), and portions of the far C-terminal domain of cTnI. Thus, Fig. 1 (upper) shows binding of cTnI to actin via two regions, the highly basic Ip and a second actin-binding region. Importantly, these regions flank a switch peptide, which binds to cTnC when Ca2+ binds to the N-terminal lobe of cTnC, which houses the regulatory Ca2+-binding site, thereby participating in the mechanism by which Tn releases the thin filament from inhibition.
The C-terminal mobile domain of cTnI beyond the second actin-binding site may also participate in establishing the relaxed state. Two observations point to this possibility. The first is a report of results from reconstructions and single-particle analysis of electron micrographs indicating that that the C-terminal mobile domain of cTnI lays across azimuthally localized actins and potentially binds directly to Tm (9, 10). These structural studies on the role of the mobile domain indicated that the cTnI C terminus is an important element in driving Tm to its blocking state along the actin outer domain. The second is a report by Mudalige et al. (11) providing direct evidence for the proximity between the C-terminal mobile domain and Tm from photochemical cross-linking studies with Tm labeled at position 146 or 174. Tm-146, but not Tm-174, cross-linked to the fast skeletal TnI peptide 157–163 (DVGDWRK), which corresponds to a nearly identical C-terminal peptide in cTnI (EVGDWRK), in a Ca2+-dependent manner.
There is now strong evidence that the hypervariable N-terminal tail of cTnT is also an important element in establishing the diastolic state. A study by Tobacman et al. (12) emphasized the importance of the N-terminal tail in the relaxed state by demonstrating that the tail domain cTnT-(1–153) alone is able to induce a blocked state of the myofilaments in the complete absence of cTnI. Single-particle analysis and reconstructions of electron micrographs also demonstrated that, at low Ca2+, Tm is wedged between cTnI actin-binding peptides and the tail of cTnT from the Tn complex in register on the opposite actin strand (8, 9). Although binding of cTnT to Tm gave this Tn component its name, interactions with other components of the cTn complex are highly dependent on cTnT, which acts as a scaffold by directly binding to both cTnI and cTnC through a rigid coiled coil, the I-T arm (4). The stability of the I-T arm predicted from the core crystal structure has been verified by studies of dynamic mapping of amide hydrogens employing hydrogen/deuterium exchange, which show the I-T arm as one of a few tightly folded regions in the solution structure of cTn (13). The other region is helix 1 of cTnI, which interacts with the C-terminal lobe of cTnC.
Ca2+-triggered protein-protein interactions engage a complex process releasing thin filaments from inhibition and actively promoting force-generating interactions between myosin cross-bridges and actin. The Ca2+ sensor is cTnC, which contains EF-hands, calcium-binding sites with consensus sequences for Ca2+ and Mg2+ binding. The N-terminal lobe of cTnC contains nonfunctional site I and functional site II (regulatory site). The C-terminal lobe contains sites III and IV (14). Sites III and IV bind Ca2+ or Mg2+ with relatively high affinity and slow exchange (15, 16). Metal binding to these sites anchors cTnC to the myofilaments by its tight interaction with a near N-terminal region of cTnI-(34–71). Site II binds Ca2+ with relatively low affinity and exchange fast enough to occur within the beat of the heart (17, 18).
Calcium binding to the single regulatory site triggers contraction by opening a hydrophobic patch at the N-terminal lobe. In the absence of the rest of the thin filament proteins, there is a relatively minimum exposure of the hydrophobic patch with calcium binding (19). Full exposure requires cTnI and can be induced by the switch peptide alone (20). The observation that the calcium-dependent opening of the hydrophobic patch on the N-terminal lobe of cTnC requires the presence of the cTnI switch peptide provided evidence for a unique role of cTnI in activation of cardiac myofilaments. In contrast to the case with fast-twitch TnI, the interaction of the switch peptide of cTnI with cTnC does not appear to involve an interaction with the D/E linker. Moreover, hydrogen/deuterium exchange analysis indicated the D/E linker to be relatively dynamic (13).
An important aspect of thin filament signaling and signal transduction is that despite control by a single regulatory Ca2+-binding site on cTnC, the dependence of tension on Ca2+ is steeper than predicted by a non-cooperative binding isotherm. Hill coefficients for the steady-state relation are commonly 3–5. One possible mechanism for this steep relation is the interaction between neighboring functional units consisting of actin/Tm/Tn at a 7:1:1 ratio. Thus, the signal generated with the Ca2+ bound to Tn spreads along the thin filament or may even spread to the Tn complexes in register on adjacent actin strands (21). Detailed balance dictates that the promotion of energies of interaction in the direction from calcium-cTnC to actin-myosin also occur in the direction from actin-cross-bridge back to cTnC. It is well known that binding of rigor cross-bridges to the thin filament increases the Ca2+-binding affinity of cTnC (17). A dominant and appealing theory is that force-generating cross-bridges reacting with the thin filament also promote interactions within a functional unit and between near-neighbor functional units. However, most of the evidence for this has been developed not with force-generating but with rigor (22) or strongly bound non-cycling cross-bridges in the form of N-ethylmaleimide-modified myosin heads (2, 23, 24). Recent studies by Sun et al. (25) have emphasized the possibility that rigor or N-ethylmaleimide-modified cross-bridges may not affect the thin filament the same as strong force-generating cross-bridges (reviewed in Refs. 26 and 27). The theory advanced by Sun et al. is that cooperative spread of activation is dominated by processes at the level of the thin filaments. Their studies employed fluorescent probes to sense the on-state of specific sites of cTnC in force-generating skinned fiber bundles. These states remained the same whether or not cross-bridges were reacting in the presence MgATP with the thin filament. Data demonstrating cooperative binding isotherms of calcium binding to cTnC in reconstituted thin filaments support this concept (28, 29).
Phosphorylation of cTnI has been extensively reviewed (1, 30); we focus here on recent data indicating novel intra- and intermolecular interactions associated with these phosphorylations in thin filament signaling. PKA, as well as PKD and PKG, phosphorylates cTnI at Ser23 and Ser24 (1), whereas PKCβ and PKCδ phosphorylate these as well as other cTnI sites (31,–34). These two sites appear to be the only sites that are phosphorylated at basal physiological levels of activity in mouse and pig hearts (35, 36). Earlier studies using synthetic peptides derived from the PKA sites in the N-terminal extension of cTnI demonstrated that the phosphorylation of these residues is ordered with Ser24 as the first residue to be phosphorylated (37,–39); monophosphorylation of the peptide at Ser23 was not detected. Zhang et al. (40) also reached the same conclusion in experiments in which a cTnI mutant in which one of these Ser residues was replaced with Ala was treated with PKA. More recently, however, studies using top-down electron capture dissociation and electron transfer dissociation mass spectrometry (41) demonstrated that phospho-Ser24 was found only in the bisphosphorylated species of cTnI isolated from human tissue. Although more studies are needed to address this issue, the involvement of phosphatases at basal phosphorylation levels under physiological conditions in vivo has been considered as a possibility. Phosphatases may dephosphorylate Ser24 faster than Ser23, resulting in a faster turnover rate of the phosphorylation/dephosphorylation state of Ser24. It has been well documented that phosphorylation of these two sites results in calcium desensitization, an increase in the relaxation rate, and an increase in the cross-bridge cycling of heart muscle myofilaments (40, 42,–45), although studies of skinned myocardium from transgenic mice lacking myosin-binding protein C (MyBP-C), another substrate for PKA in myofilaments, and/or expressing non-PKA-phosphorylatable cTnI indicate that PKA phosphorylation of cTnI is responsible for calcium sensitivity and that phosphorylation of MyBP-C is responsible for enhanced cross-bridge cycling (46).
It is critical to delineate the structural information on the N-terminal extension relative to other Tn components to understand molecular mechanisms underlying these functional consequences. The N-terminal extension of cTnI interacts with the N-terminal lobe of TnC and modifies the conformational states of the N-terminal lobe of TnC (47, 48). With phosphorylation, there is a depression of the affinity of the cTnI N-terminal extension and the N-terminal lobe of cTnC (47,–50). Evidence also indicates that that the N-terminal extension of cTnI is located close not only to the N-terminal lobe of cTnC but also to the switch region of cTnI. Warren et al. (51) reported intramolecular cross-linking between cTnI at amino acid 5 or 18 and the switch region of TnI in the Tn complex. Ward et al. (52) determined which amino acid residues of the N-terminal extension participate in the interaction with cTnC. They generated a series of N-terminal deletion mutants of cTnI and measured calcium-dependent ATPase activities before and after PKA treatment. The deletion mutants up to position 18 were phosphorylated by PKA at the same rate and to the same extent as wild-type cTnI. The ΔpCa50 values of ATPase activities before and after PKA treatments were similar and substantial for all of the deletion mutants up to position 15, indicating that residues 1–15 of cTnI play only a minor role in transmitting the phosphorylation signal to other myofilament proteins. Ward et al. (53) also showed that peptide residues 1–18 of cTnI do not bind to cTnC. NMR studies (53, 54), as well as previous studies (37), demonstrated that PKA phosphorylation at Ser23 and Ser24 produced only localized structural effects in segment residues ~25–35.
A more detailed picture of the intermolecular interactions between the N-terminal extension and the N-terminal lobe of cTnC and the intramolecular interactions within TnI was proposed by Howarth et al. (5). They determined the solution structure of the bisphosphorylated N-terminal extension (residues 1–32) of cTnI and modeled the N-terminal extension into the crystal structure of the core domain of cTn (4) based on bioinformatics analysis and previous small angle scattering experiments. According to their model, the N-terminal extension interacts with cTnC near Leu29, as also demonstrated in peptide array experiments (55) with cTnI in a non-phosphorylated state. Upon bisphosphorylation, cTnI undergoes relatively drastic conformational transitions, resulting from a hinge-like movement within residues 33–42. Molecular docking experiments designed for maximum interactions of the N-terminal extension with opposite polarity showed that the acidic residues in the N-terminal end, such as Asp3, Glu4, Asp7, and Glu11, interact with the basic residues in the inhibitory region, suggesting intramolecular interactions. Photo-cross-linking experiments in the presence of Ca2+ showed that benzophenone attached to Cys5 or Cys19 cross-linked to the switch region of cTnI in the non-phosphorylated cTn complex, indicating that the N-terminal extension exists in an equilibrium between these two proposed structural states without phosphorylation (51).
Although induction by phosphorylation of cTnI at Ser23/Ser24 of a straightforward depression of Ca2+ affinity of the cTnC regulatory site has been proposed to account for increased release of Ca2+ from cTnC and enhanced relaxation, another mechanism was proposed by Baryshnikova et al. (56). As indicated above, phosphorylation of Ser23 and Ser24 modifies the conformational states of the N-terminal lobe of cTnC. They found that the presence of phosphate groups at Ser23/Ser24 does not affect calcium binding to the regulatory site of cTnC when measured in the presence of peptide residues 1–29 of the N-terminal extension of cTnI. With or without the peptide or the bisphosphopeptide, the association constant for the regulatory site was determined to be ~2 × 105 m−1. Baryshnikova et al. proposed that bisphosphorylation of peptide residues 1–29 modifies the affinity of the switch region of cTnI for the N-terminal lobe of cTnC and hence the relaxation rate. Another possibility was proposed on the basis of solution ATPase activity assays. Deng et al. (57) and Lu et al. (33) demonstrated that PKA treatment of the cTn complex resulted in a reduced maximum ATPase activity in the presence of Ca2+. These observations are supported by results of studies on the relation between thin filament length and velocity as determined by in vitro motility assay in which the apparent binding affinity of myosin for thin filaments was reduced after PKA treatment (58). Thus, it is plausible that PKA phosphorylation of cTnI modifies the affinity between cross-bridges and thin filaments.
Thr144 is located in the middle of the inhibitory region of cTnI. The equivalent position is Pro in skeletal muscle TnI. The functional consequence of phosphorylation of Thr144 by itself has not been well established. Burkart et al. (59) reported that the T144E mutation, which mimics the phosphorylation state, does not affect calcium sensitivity or maximum tension when introduced into skinned cardiac fiber bundles, whereas in the in vitro motility assay, it desensitizes to Ca2+. On the other hand, when skinned cardiomyocytes or small myocyte bundles harboring cTnI S23A/S24A to prevent phosphorylation were treated with PKCβ, Wang et al. (34) found that the isometric tension of these myocytes was sensitized to Ca2+ compared with non-PKC-treated myocytes. Because Thr144 of cTnI was phosphorylated mainly by PKC treatment in their preparations, they concluded that phosphorylation of Thr144 was responsible for the calcium sensitization. Data reported by Westfall et al. (60) indicated that the phosphorylation of Thr144 accelerates relaxation of cardiomyocytes, whereas it does not affect shortening. These divergent observations may be explained by recent findings that Thr144 is involved in length-dependent activation of tension development in skinned fiber bundles (61) and strong cross-bridge-dependent activation of the thin filaments in solution acto-S1 ATPase activity (62).
The bisphosphorylation of cTnI at its N terminus induces a lengthening of the adjacent helix TnI-(21–30) with induction of a bend in the extension, which contains acidic residues (5). On the basis of these structural modifications, we predicted that the acidic N-terminal regions of cTnI might interact intramolecularly with the basic inhibitory region of cTnI. Cross-linking studies supported this prediction (51). Functional evidence for this interaction came from the study of a phosphomimetic mutation at Ser23/Ser24 in the N-terminal extension and Thr144 in the inhibitory region (33). It was demonstrated that pseudophosphorylation at Thr144 of cTnI depressed the cooperative activation of cardiac thin filaments exclusively in the presence of the S23D/S24D mutation of cTnI. The calcium-binding properties of the thin filaments regulated by cTnI with the T144E mutation were the same as those with wild-type cTnI. The cTnI S23D/S24D mutation desensitized the thin filaments to Ca2+, as expected. Thin filaments with the S23D/S24D/T144E mutation retained the same calcium sensitivity as those with the S23D/S24D mutation, yet the Hill coefficient, indicative of cooperativity, was significantly smaller. Similar trends followed the tension development of skinned cardiac muscle fiber bundles. Moreover, studies of the cTnI R145G mutation linked to hypertrophic cardiomyopathy also indicated an interaction between the cTnI N terminus and the Ip (55). These studies demonstrated that compared with controls, calcium sensitivity is enhanced in myofilaments regulated by cTnI R145G but not when cTnI Ser23 and Ser24 are bisphosphorylated. One interpretation is that the loss of the basic residue not only depresses inhibition by the Ip but also alters electrostatic interactions with the N terminus. Transgenic mouse hearts in which residues 2–11 of cTnI were deleted showed a significant decrease in contractility and relaxation upon basal and β-adrenergic stimulation, whereas the calcium sensitivity of force development was not altered, suggesting the importance of the N-terminal part of cTnI in regulation and modification (63). These experiments provide strong indications of modulation of cTn function by an intramolecular interaction between the N-terminal extension and inhibitory region of cTnI.
Several kinases that phosphorylate cTnT have been identified, but the functional significance of the phosphorylation in thin filament signaling and integrated control of cardiac function remains unclear. Evidence for phosphorylation of cTnT by a “specific” cTnT kinase (64) and by phosphorylase kinase (65, 66) has not been followed up. Early studies (67) identified a site of phosphorylation in the N terminus at Ser1, but the significance of this site remains unknown. Sites indentified as PKC substrates in the C-terminal region have been more extensively documented. cTnT PKC-dependent phosphorylation sites are Thr197, Ser201, Thr206, and Thr287. We (68) mutated these residues to glutamate or alanine and exchanged single, double, triple, and quadruple mutants into skinned fiber bundles of mouse heart. We also exchanged the Tn complex reconstituted with cTnT phosphorylated by PKCα. Our studies of isometric tension development and actomyosin Mg-ATPase activity as a function of Ca2+ concentration identified Thr206 as a functionally critical cTnT PKC phosphorylation residue. Compared with wild-type controls, exclusive phosphorylation by PKCα or replacement by Glu induced a significant decrease in myofilament maximum tension, actomyosin Mg-ATPase activity, calcium sensitivity, and cooperativity. Tension cost was also reduced. Subsequent studies also determined that phosphorylation of cTnT depresses sliding speed in the motility assay (69). Inasmuch as PKCα activation promotes cardiac growth, engagement of this pathway coordinates hypertrophic signaling and contractile dynamics. Along these lines, Thr206 also appears to be a substrate for Raf kinase (70).
Modifications in the interaction of the N terminus of cTnI with cTnC and with the Ip region of cTnI appear to be an important mechanism in the linkage of Tn mutations to cardiomyopathies. As discussed above, studies of the cTnI R146G mutation, which is linked to hypertrophic cardiomyopathy, provide not only further indications of an interaction between the cTnI N terminus and the Ip but also an important role for phosphorylation of cTnI Ser23/Ser24 in the course of the disorder. Our studies (42) with cTnC G159D, which is linked to dilated cardiomyopathy (DCM), provide another example of an effect of phosphorylation of these cTnI residues on the functional effects of a mutant thin filament protein. In this case, the interaction appears to exacerbate the functional effects of the mutations. We (42) reported little or no effect of the mutation under base-line conditions without cTnI phosphorylation. However, with PKA-dependent phosphorylation or pseudophosphorylation of cTnI, there was a significant depression of the decrease in the calcium sensitivity of tension development compared with controls. This depression in the effect of cTnI Ser23/Ser24 phosphorylation was correlated with a reduced effect on calcium binding to the cTnC regulatory site. Another region of phosphorylation of cTnI, Ser43/Ser45, appears to modify the functional effects of a familial hypertrophic cardiomyopathy-linked mutation, Tm E180G (26). Tension developed by myofilaments controlled by Tm E180G was significantly more sensitive to Ca2+ compared with controls. In skinned fibers with wild-type thin filaments, pseudophosphorylation of cTnI Ser43/Ser45 in the anchoring domain interacting with the cTnC C-terminal lobe induced a depression in force and calcium sensitivity (59). This depression was significantly enhanced in skinned fibers controlled by Tm E180G. Interactions between the Ip and the R145G mutation linked to hypertrophic cardiomyopathy and the Ca2+-saturated C-terminal lobe of cTnC were also altered. These interactions have been determined employing NMR chemical shift mapping (71). There was a significant 14-fold decrease in the affinity with Thr144 phosphorylation and a 4-fold decrease in the presence of the R146G mutation in the peptide. Phosphorylation (Fig. 1) at cTnI residue 150, which is poorly understood, may also affect the response to the R146G mutation.
A potentially important aspect of phosphorylation of cTnT is the proximity of Thr206 to a region linked with DCM in which there is a deletion of Lys210. Comparison of the state of phosphorylation of myofilaments from hearts expressing the cTnTΔLys210 mutant with that of controls demonstrated significant net decreases in phosphorylation of cTnT, cTnI, and MyBP-C (72). However, there was an increase in site-specific phosphorylation of Thr206. Moreover, there was an increased rate of phosphorylation of Thr206 by PKCα. Evidence that cTnTΔLys210 has enhanced affinity for cTnI provides further evidence for the long-range effects of this mutation (72).
Although there has been progress in understanding the phosphatases that control phosphorylation of TnI and TnT, signaling cascades controlling these phosphatases remain poorly understood. Moreover, site-specific dephosphorylation, which is likely to be an important mechanism, has been understudied and is poorly understood. The major phosphatases controlling thin filament protein phosphorylation are PP1 and PP2A. Both of these phosphatases have been reported to have a Z-disc localization (73, 74), placing them in close proximity to the A- and I-band regions of the sarcomere and indicating that their localization may be strain-sensitive. In fact, recent studies (74) indicate that that the B56a unit of the PP2A complex localizes to the Z-disc but moves away with β-adrenergic stimulation, whereas light chain 2 phosphatase does not. Our studies (26) showing a decrease in tropomyosin phosphorylation in hearts with constitutively active p38α demonstrated co-localization of this MAPK with α-actinin at the Z-disc, as well as protein phosphatases (PP2α and PP2β). Few studies have directly compared the substrate specificities of PP2A and PP1, but there is some agreement that PP1 has preference for cTnT over cTnI (73, 75) but with no effect on dephosphorylation of Tm (73). Moreover, Jideama et al. (75) employed phosphopeptide mapping to show that PP2A induced uniform dephosphorylation of cTnI in preparations previously treated with both PKA and PKC. On the other hand, Ser23 and Ser24 were the preferred substrates for PP1. In the case of cTnT, Thr199 and an unidentified residue were the least favorable for dephosphorylation by PP1. These studies thus demonstrate specificity for dephosphorylation, a mechanism not generally taken into account when considering the integrated effects of signaling to the thin filaments. In our studies of the effects of constitutively active PKCζ (76), we observed a significant decrease in Thr phosphorylation of cTnI and cTnT notably by PKCζ T560E. To explain the apparent Thr dephosphorylation of cTnI and cTnT, we hypothesized that PKCζ exists in a complex with Pak1 and PP2A, and this was confirmed by immunoprecipitation and Western blotting. Reviewed elsewhere (77, 78) are our studies on a signaling cascade to PP2A via Gi coupling to small G-proteins (e.g. bradykinin) and potential sphingolipid signaling via p21-activated kinase.
New data advance our understanding of the molecular mechanism by which thin filament proteins control the state of the sarcomere. Understanding these molecular control mechanisms in thin filament regulation of the heartbeat has taken on new significance with the identification of many mutations linked to hypertrophic cardiomyopathy and DCM. There is also strong evidence for a role of alterations at the level of the thin filaments in acquired cardiac disorders and sudden death. The results summarized here indicate that apart from the direct effects of these mutations and modifications on function and indirect effects on stress signaling, it is important to consider the pathological process in the context of post-translational modifications of the sarcomeric proteins. The design of new diagnostics and therapies, as well as preventative measures, requires advancement of our understanding of these processes and their relative significance.
We thank colleagues for collaborations related to work cited here.
*This work was supported, in whole or in part, by National Institutes of Health Grants HL 62426, HL 64035, and HL 22231 (to R. J. S.) and HL 082923 (to T. K.). This is the sixth article in the Thematic Minireview Series on Signaling in Cardiac Sarcomeres in Health and Disease. This minireview will be reprinted in the 2011 Minireview Compendium, which will be available in January, 2012.
2The abbreviations used are: