Troponin I Mutations R146G and R21C Alter Cardiac Troponin Function, Contractile Properties, and Modulation by Protein Kinase A (PKA)-mediated Phosphorylation*

Background: R146G and R21C mutations in cardiac TnI are associated with hypertrophic cardiomyopathy. Results: Both mutations blunt PKA-mediated effects on weakening cTnI-cTnC interaction and accelerating myofibril relaxation. Conclusion: Both mutations result in hypercontraction and impaired relaxation, which may contribute to increased risk to traumatic heart failure. Significance: This study increases mechanistic understanding of how single amino acid mutations result in cardiac contractile dysfunction. Two hypertrophic cardiomyopathy-associated cardiac troponin I (cTnI) mutations, R146G and R21C, are located in different regions of cTnI, the inhibitory peptide and the cardiac-specific N terminus. We recently reported that these regions may interact when Ser-23/Ser-24 are phosphorylated, weakening the interaction of cTnI with cardiac TnC. Little is known about how these mutations influence the affinity of cardiac TnC for cTnI (KC-I) or contractile kinetics during β-adrenergic stimulation. Here, we tested how cTnIR146G or cTnIR21C influences contractile activation and relaxation and their response to protein kinase A (PKA). Both mutations significantly increased Ca2+ binding affinity to cTn (KCa) and KC-I. PKA phosphorylation resulted in a similar reduction of KCa for all complexes, but KC-I was reduced only with cTnIWT. cTnIWT, cTnIR146G, and cTnIR21C were complexed into cardiac troponin and exchanged into rat ventricular myofibrils, and contraction/relaxation kinetics were measured ± PKA phosphorylation. Maximal tension (Tmax) was maintained for cTnIR146G- and cTnIR21C-exchanged myofibrils, and Ca2+ sensitivity of tension (pCa50) was increased. PKA phosphorylation decreased pCa50 for cTnIWT-exchanged myofibrils but not for either mutation. PKA phosphorylation accelerated the early slow phase relaxation for cTnIWT myofibrils, especially at Ca2+ levels that the heart operates in vivo. Importantly, this effect was blunted for cTnIR146G- and cTnIR21C-exchanged myofibrils. Molecular dynamics simulations suggest both mutations inhibit formation of intra-subunit contacts between the N terminus and the inhibitory peptide of cTnI that is normally seen with WT-cTn upon PKA phosphorylation. Together, our results suggest that cTnIR146G and cTnIR21C blunt PKA modulation of activation and relaxation kinetics by prohibiting cardiac-specific N-terminal interaction with the cTnI inhibitory peptide.

Familial hypertrophic cardiomyopathy (HCM) 2 has been identified as a major autosomal dominant disease and is highly correlated with mutations detected in myofilament contractile proteins (1). Although the majority of mutations are found in myosin and cardiac myosin-binding protein C (cMyBP-C), mutations have also been identified in thin filament regulatory proteins such as cardiac troponin I (cTnI), which is a subunit of the cardiac troponin (cTn) complex that has a critical role in the activation and relaxation of cardiac muscle (2). At the beginning of systole, with the rise of intracellular Ca 2ϩ in cardiomyocytes, Ca 2ϩ binding to cardiac troponin C (cTnC) initiates a chain of events involving dynamic and structural changes in troponin that result in the activation of the thin filament (3). In the absence of Ca 2ϩ (diastole), cTnC exists in its "closed" conformation, and cTnI binds actin tightly (and only weakly with cTnC), inhibiting actin-myosin interaction (3,4). In systole, Ca 2ϩ binding to site II of cTnC induces an "open" conformation that increases interaction between the N terminus of cTnC (NcTnC) and the cTnI switch peptide, resulting in decreased binding of the cTnI inhibitory peptide with actin (3,5). Conse-quently, this permits increased tropomyosin mobility, myosin interaction with actin to form cross-bridges, resulting in force generation (3).
␤-Adrenergic stimulation serves as an essential physiological mechanism to meet increases in circulatory demand, acting through positive inotropic-lusitropic effects (6). During ␤-adrenergic stimulation, cTnI is phosphorylated by protein kinase A (PKA) at sites Ser-23 and Ser-24 (Ser(P)-23/Ser(P)-24) that reside in the cardiac-specific N terminus of cTnI (NcTnI) (6). We (7,8) and others (6, 9 -11) have demonstrated that phosphorylation of these sites reduces the affinity of cTnC for cTnI (K C-I ), reduces Ca 2ϩ sensitivity (pCa 50 ) of tension production, increases cross-bridge cycling kinetics, and accelerates cardiac muscle relaxation. We have also reported that PKA phosphorylation of cTnI or bis-phosphomimic substitutions of cTnI (cTnI S23D/S24D ) accelerates and shortens the initial slow phase of cardiac myofibril relaxation, particularly during contraction with physiological (sub-maximal) Ca 2ϩ conditions, and thus it increases the overall speed of relaxation (7).
HCM-associated cTnI mutations were first reported by Kimura et al. in 1997 (12), including R145G/R145Q, R162W, G203S, and K206Q. Among them, the cTnI R145G mutation (cTnI R146G in rodent), which is located in the inhibitory peptide of cTnI, has received prominent attention (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25). Most previous studies investigating this mutation have focused on the Ca 2ϩ sensitivity of tension, and ATPase activity in cardiomyocytes, demembranated cardiac muscle, and transgenic mice. It is well established that cTnI-R146G mutation increases Ca 2ϩ sensitivity of myofibrillar ATPase activity and force (18 -20), reduces inhibition of actin-tropomyosin-activated myosin ATPase (14,18,19), and may have no direct effect on the crossbridge cycle (20). It has also been reported that the cTnI-R145G mutation has a significant effect on energy cost and has been associated with diastolic dysfunction (20). Another mutation, cTnI R21C , is the only identified HCM-associated mutation located at the cardiac-specific N terminus of cTnI (26 -29). In transgenic mice, the cTnI R21C mutation has been reported to prevent PKA-mediated phosphorylation in vivo (27,28). It has also been reported that isolated cardiac myocytes from R21C mice older than 12 months of age have significantly delayed Ca 2ϩ transient decay and relaxation (28). However, the mechanism for these effects and how these mutations affect the contraction and relaxation kinetics of cardiac muscle have not been studied.
Previous studies have proposed the formation of an intramolecular interaction between the N terminus and the inhibitory peptide region of cTnI upon PKA phosphorylation of Ser-23/Ser-24 of cTnI (11, 30 -32). Recently, our computational modeling results demonstrated that introduction of the S23D/ S24D substitutions (bis-phosphomimic substitutions) on cTnI (cTnI S23D/S24D ) led to the formation of an intra-subunit interaction between the N terminus and the inhibitory peptide of cTnI (8). We hypothesized that this interaction may be the structural correlate for shortening the duration and increasing the rate of the early phase of relaxation by destabilizing cTnI switch peptide interaction with NcTnC (8). Therefore, we hypothesized that introduction of an HCM mutation located in either the N terminus or the inhibitory peptide of cTnI may disrupt the formation of this intra-subunit interaction and blunt the effects of Ser-23/Ser-24 phosphorylation by PKA during ␤-adrenergic stimulation. In this work, we tested this hypothesis by studying the two HCM mutations cTnI R146G and cTnI R21C (see Fig. 1 for the location) that are located in the inhibitory peptide and the N terminus of cTnI (respectively) using combined protein biochemistry, myofibril mechanics, and computational (molecular dynamics) simulation studies. Our studies indicate that both of these cTnI mutants increase Ca 2ϩ binding of cTn (K Ca ) and K C-I in solution, increase the Ca 2ϩ sensitivity of myofibril tension development, and also prolong the early slow phase of relaxation. Importantly, both mutants blunt the ability of PKA to reduce K C-I and the Ca 2ϩ sensitivity of tension (pCa 50 ) and speed relaxation of myofibrils. Our computational modeling of cTn suggests that introduction of either mutation inhibits the formation of the intra-subunit interaction between the N terminus and the inhibitory peptide of cTnI normally seen for cTn with phosphorylation (or bisphosphomimic substitutions) of Ser-23/Ser-24. Thus, in addition to being hyper-contractile during systole, hearts with these mutations may have impaired initiation of diastole during ␤-adrenergic stimulation.
The cTnC C35S substitution was labeled with a fluorescent probe {N-[2-(iodoacetoxy)ethyl]-N-methyl}-nitrobenz-2-oxa-1,3-diazole (IANBD, M r ϭ 406.14, Life Technologies, Inc., cat-alog no. I-9) at Cys-84 (cTnC IANBD C35S ) in the dark overnight at 4°C to monitor the Ca 2ϩ -cTn (K Ca ) and cTnC-cTnI (K C-I ) binding affinities, as described previously (7, 34 -36). The labeling efficiency was determined by measuring the IANBD fluorophore to protein molar concentration ratio (7,36). The concentration of protein was determined using Bio-Rad protein assay (based on Bradford method), and the IANBD concentration in the labeled protein was determined by dividing the absorbance of the labeled protein at the maximal absorbance for the fluorophore by the extinction coefficient of IANBD (21000 M Ϫ1 cm Ϫ1 ) at a wavelength of 481 nm. The final labeling efficiency was then determined to be ϳ90%.
Purified cTnI WT , cTnI R146G , and cTnI R21C were phosphorylated using a cTnC affinity column by adding 500 units of the catalytic subunit of PKA (Sigma, catalog no. P2645). The reaction was initiate by adding 0.5 mM ATP and 6 mg/ml DTT to the column, and the column was incubated in a pre-warmed water bath at 30°C for 30 min (37). The phosphorylation profile of cTnI was determined by calculating the percentage of phosphorylated and the total amount of cTnI from Western blot (7).
Steady-state Fluorescence Measurements-All steady-state fluorescence experiments were measured using an LS50B luminescence spectrometer (PerkinElmer Life Sciences) at 15°C as described previously (7,36,40). Solution composition for this fluorescence measurement was as follows (in mM): 150 KCl, 20 FIGURE 2. A, phosphorylation (phos) profile of Ser-23/Ser-24 for purified cTnI WT , cTnI R146G , and cTnI R21C from the cTnC affinity column. Raw (B) and normalized (C) PKA phosphorylation profiles of cTnI for rat LV cardiac myofibrils exchanged with cTn-containing cTnI WT or cTnI R146G before and after PKA treatment are shown.
Ethical Approval and Tissue Preparation-Animal procedures were conducted in accordance with the National Institutes of Health Policy on Humane Care and Use of Laboratory Animals and were approved by the University of Washington Institutional Animal Care and Use Committee (IACUC). Rats were housed in the Department of Comparative Medicine, University of Washington, and cared for according to IACUC procedures. Male Sprague-Dawley rats (3 months old, 150 -250 g) were anesthetized with an intraperitoneal injection of pentobarbital (50 mg/kg) after initial exposure to isoflurane (3-5% in oxygen). When the rat had no reflexive response, its heart was rapidly excised and dissected in oxygenated physiological saline solution containing the following (in mM): 100 NaCl, 24 NaHCO 3 , 2.5 KCl, 1 MgSO 4 ⅐7H 2 O, 1 Na 2 HPO 4 , and 1 CaCl 2 (43). After this, both ventricles were cut open, and the whole heart was demembranated in skinning solution containing (in mM): 100 KCl, 9 MgCl 2 , 4 Na 2 ATP, 5 K 2 EGTA, 10 MOPS, 1% Triton X-100, pH 7.0, 50% v/v glycerol, and 1:100 dilution "protease inhibitor mixture" (Sigma, catalog no. P8340) overnight at 4°C (44,45). The heart was then washed three times in the same solution without Triton X-100 and stored at Ϫ20°C for using up to 1 week. Myofibrils from the left ventricles (LV) were used for the mechanical measurements described below.
Solutions-Composition of solution used for mechanical measurements was determined by an iterative algorithm that computes the equilibrium concentration of ions and ligands based on published affinity constants (46). Composition of relaxing solution was as follows (in mM): 80 MOPS, 1 Mg 2ϩ , 5 MgATP, 83 K ϩ , 52 Na ϩ , 15 EGTA, and 15 creatine phosphate, pH 7.0, at 15°C. The solution ionic strength was 170 mM, and the inorganic P i concentration that was determined by NMR measurement was 0.5 mM (47). All mechanical measurements were performed at a constant temperature of 15°C. The Ca 2ϩ levels (expressed as pCa ϭ Ϫlog [Ca 2ϩ ]) for activation solutions were adjusted by adding CaCl 2 . To study the effects of PKA, isolated myofibrils were incubated with 200 l of relaxing solution containing 100 units of PKA and 6 mM DTT for 45 min at 20°C.
Exchange of Recombinant cTn Complexes into Myofibrils-Muscle bundles obtained from the rat LV were rinsed twice in Rigor solution containing the following (in mM): 100 KCl, 50 Tris, 2 MgCl 2 , 1 EGTA, 1 DTT and 1:100 dilution of protease inhibitor mixture before being homogenized for 2 pulses of 30 s on ice at high speed. cTn complexes containing cTnI WT , cTnI S23D/S24D , cTnI R146G , cTnI R21C , or cTnI R21C/S23D/S24D at a final concentration of ϳ1 mg/ml were passively exchanged into isolated rat LV myofibrils in a buffer containing the following (in mM): 200 KCl, 20 MOPS, 5 MgCl 2 , 2 EGTA, 1 DTT, 4 ATP, and 1:100 dilution of protease inhibitor mixture on a slow rocker overnight at 4°C (7). Following exchange, myofibrils were washed with relaxing solution containing 1 mg/ml bovine serum albumin (BSA) twice for 30 min to remove any nonspecifically bound exogenous cTn.
Myofibril Mechanical/Kinetics Measurement-Myofibril mechanical/kinetics measurements were performed on a custom-built setup as described previously (48). Briefly, single or small bundles (ϳ2-4) of cardiac myofibrils were attached between two glass micro-tools forged from borosilicate glass capillary tubes (outer diameter 1.0 mm and inner diameter 0.5 mm, Sutter Instruments, Novato, CA), with the initial sarcomere length set as ϳ2.3 m, and perfused with solutions that can be rapidly switched. One of the needles acted as a force transducer, which deflected in a predictable manner upon application of force (48). Needle stiffness was determined by first deflecting the needle with a known amount of force using a galvanometer. Needle deflections were measured under a ϫ40 lens, and this yielded stiffness in nN m Ϫ1 . The stiffness of needles used for the experiments ranged between 5 and 11 nN m Ϫ1 . This force transducer needle was positioned over a dual diode system, which records needle displacement and correlates displacement to force development. A second straight needle was attached to the other end of the myofibril and was applied to rapidly shorten and re-stretch the myofibril through a computer interface and a Piezo-controller motor (PZT Servo controller, LVPZT amplifier, Physik Instrumente, Irvine, CA). At the end of each experiment, a calibration curve was performed in which the force transducer needle was moved in 1-m steps over the range of the diodes using micromanipulators (MP-285, Sutter Instruments, Novato, CA).
A double-barreled borosilicate glass pipette (capillary glass tubing outer diameter 2.0 mm and inner diameter 1.4 mm, SEP 0.2 mm, modified in-house to outer diameter of 0.55 mm, Warner Instruments, Hamden, CT) was used to stream low (10 Ϫ9 M, pCa 9.0) and high (10 Ϫ4 M, pCa 4.0) Ca 2ϩ -containing solutions to the mounted preparation, and stepping for solution switch over the preparation was controlled by a computerized motor (SF-77B Perfusion Fast Step, Warner Instruments). The solution change was complete in ϳ10 ms (48,49).
Activation and relaxation data were collected at 15°C and fit as described previously (48 -50). The kinetics of contractile activation (k act ; with rapid increase in Ca 2ϩ ) was obtained from a single-exponential rise to a maximum. A rapid release-restretch protocol (a sudden 20% decrease in optimal length followed by rapid stretching back to the original length after 25 ms of unloaded shortening) was applied to measure the rate of force redevelopment (k tr ). A slow phase relaxation rate (k rel, slow ) was reported as the slope of a regression line fit to the tension trace and normalized to the tension amplitude, and the slow phase duration (t rel, slow ) was measured from the onset of solution change at the myofibril to the shoulder marking the beginning of fast phase. Transition from slow to rapid phase was determined through multiple factors. An apparent change in the slope of the data or a change in the signal-to-noise ratio was often apparent at the transition. The fast phase relaxation rate (k rel, fast ) was measured from a single exponential decay fitted to the data. A t1 ⁄ 2 estimation was made in cases where the decay was not well described by a single exponential, and this was converted to a rate ϭ ln(2)/t1 ⁄ 2 . Myofibrils that contracted Ͼ10% of their optimal length were excluded from the analysis as non-isometric.
Statistics-Comparisons between groups of data were performed using paired or unpaired Student's t test as appropriate. All reported data are expressed as mean Ϯ S.E, and "n" represents the number of experimental samples in each group. Results with p Ͻ 0.05 were considered statistically significant. In this study, the R146G, R21C, WT ϩ PKA, and S23D/S24D data were compared with the WT sets; the R146G ϩ PKA data were compared with WT ϩ PKA sets, and the R21C/S23D/ S24D results were compared with S23D/S24D sets.
Computational Modeling-The initial structure of the cTn complex was built up based on the core crystal structure of Takeda et al. (51) with the addition of the N terminus of cTnI from the NMR structure provided by Howarth et al. (30). To mimic phosphorylation, a bis-phosphomimics model (cTnI S23D/S24D ) was constructed by mutating Ser-23/Ser-24 of cTnI to aspartic acid (Asp). Two systems of human cTn were prepared for simulations as follows: a cTnI R21C Ca 2ϩ -bound cTnC(1-161)-cTnI(1-172)-cTnT(236 -285) (cTnI R21C cTn model), and a cTnI R21C/S23D/S24D Ca 2ϩ -bound cTnC(1-161)-cTnI(1-172)-cTnT(236 -285) (cTnI R21C/S23D/S24D cTn model). The cTnI mutations were performed using the Mutate Residue module in VMD (52). The build-up models were immersed with TIP3P water molecules in a truncated rectangular box, which extended minimally 14 Å away from any solute atoms (53). Then, K ϩ and Cl Ϫ ions were added to neutralize the systems and brought to 150 mM ionic strength. The fully solvated systems contained 112,758 (cTnI R21C cTn model), and 112,759 (cTnI R21C/S23D/S24D cTn model) atoms, respectively. Prior to the MD simulations, we performed three steps of minimizations. Next, 150-ns MD simulations were performed under the NPT ensemble and 300 K using NAMD 2.9 (54) and the CHARMM27 force field (55). The SHAKE procedure was applied on the bonds involving hydrogen atoms, and the time step was set to 2.0 fs (56). During the sampling process, the coordinates were saved every 10 ps. The stability between site II Ca 2ϩ and its coordinating residues (Asp-65, Asp-67, Ser-69, Thr-71, Asp-73, and Glu-76) of cTnC was monitored by calculating the following distances for each 150-ns simulation as described previously (8,36). Simulations were run in triplicate.
The residue-residue contacts between cTnC and key regions of cTnI (N terminus, inhibitory peptide and switch peptide regions) were monitored over the course of the entire 450-ns simulations. Contacts between two residues were defined as described previously (36), with a carbon-carbon distance of Յ5.4 Å and a distance between any other non-carbon atoms of Յ4.6 Å being a contact. Contacts between NcTnC-switch peptide of cTnI, and cTnC-inhibitory peptide of cTnI were monitored. The intra-subunit interaction between the N terminus and the inhibitory peptide region of cTnI were also recorded. For each residue contact pair, the fraction of the simulation time that these residues were in contact was calculated for both simulation systems.

Results
Purified cTnI Phosphorylation Level from cTnC Column-Both WT, R146G, and R21C cTnI were phosphorylated by PKA using the cTnC affinity column, and the extent of cTnI phosphorylation was determined by computing the percentage of Ser-23/Ser-24 phosphorylated versus the total amount of cTnI, using Western blot analysis (7). As shown in Fig. 2A, our results suggested that our phosphorylation protocol was very efficient, with Ͼ85% phosphorylation for both WT and R146G cTnI. Consistent with previous reports (27,28), the phosphorylation of cTnI R21C was Ͻ5%, suggesting cTnI R21C may disrupt the PKA phosphorylation process at Ser-23/Ser-24 of cTnI, and thus resulted in "blunted" ␤-adrenergic stimulation effects. This may be the actual physiological/pathogenic mechanism of cTnI R21C . Thus, to determine whether it is the cTnI R21C mutation per se that is altering function or just the inability to get Ser-23/Ser-24 phosphorylated, we introduced the bis-phosphomimic substitutions S23D/S24D into cTnI R21C (cTnI R21C/S23D/S24D ) to mimic the effect of PKA phosphorylation. We (7,8,25,57) and others (58 -61) previously demonstrated that cTnI S23D/S24D can mimic the PKA phosphorylation effects on Ser-23/Ser-24 of cTnI (cTnI Ser(P)-23/Ser(P)-24 ) both structurally and functionally.
Steady-state Fluorescence Measurements of K C-I and K Ca -The effects of R146G or R21C mutation Ϯ PKA phosphorylation (or bis-phosphomimic substitutions) on cTn K Ca and the binding affinity of cTnC for cTnI (K C-I ) were determined by steady-state fluorescence measurements using a fluoroprobe IANBD, as described previously (7,36). IANBD, a sulfhydrylreactive and environment-sensitive extrinsic fluorophore, has been widely used to study the intra-molecular interactions of proteins, and labeling at Cys-84 of cTnC C35S reflects conformational and environmental changes of NcTnC that arise from Ca 2ϩ binding and/or interaction with cTnI (7,35,36,40). We first measured the conformational changes with Ca 2ϩ binding to cTn containing either cTnI R146G or cTnI R146G/Ser(P)-23/Ser(P)-24 compared with WT cTnI. As shown in the Ca 2ϩ titration curves in Fig. 3A, cTnI R146G increased (left shift) Ca 2ϩ binding affinity (K Ca ) compared with cTn containing cTnI WT , in agreement with previous studies (18 -20). The Ca 2ϩ sensitivity of the fluorescence intensity (reported as pCa 50 ) was shifted by 0.24 pCa units, from 7.07 Ϯ 0.03 (cTn with cTnI WT ) to 7.31 Ϯ 0.03 (cTn with cTnI R146G ). Consistent with our previous finding (7), phosphorylation of cTnI WT at Ser-23 and Ϫ24 (cTnI Ser(P)-23/Ser(P)-24 ) also reduced K Ca , resulting in a 0.31 pCa unit decrease (right shift). Similarly, PKA phosphorylation of cTnI R146G (cTnI R146G/Ser(P)-23/Ser(P)-24 ) reduced the Ca 2ϩ sensitivity (pCa 50 ϭ 7.03 Ϯ 0.03), resulting in a 0.28 pCa unit decrease (right shift).
In view of the "gatekeeper" role of cTnC-cTnI interaction in translating the Ca 2ϩ signal to myofilament proteins to initiate cardiac muscle contraction, we also tested how the cTnI R146G mutation Ϯ PKA phosphorylation affected K C-I . The K C-I was measured by titrating cTnI R146G or cTnI R146G/Ser(P)-23/Ser(P)-24 into cTnC IANBD C35S in the presence of 100 M Ca 2ϩ . Fig. 3B shows the IANBD fluorescence signal change as the concentration of cTnI was increased up to 0.8 M in solutions containing 0.6 M cTnC IANBD C35S . There was no further change in the fluorescence signal beyond 0.6 M cTnI, suggesting strong binding of cTnI to cTnC such that 1:1 binding was achieved. Similar to K Ca , cTnI R146G left-shifted K C-I compared with cTnI WT . As we reported previously (7), phosphorylation of cTnI WT reduced K C-I . However, this effect was completely eliminated (blunted) for the cTnI R146G mutant.

Recombinant Troponin (cTn) Complex Exchange Profiles-
The native cTn in isolated myofibrils was passively exchanged with recombinant rat cTn containing either cTnI WT , cTnI R146G , or cTnI R21C . The extent of exchange (exchange efficiency) for this procedure was periodically determined by exchanging cTn containing a cTnT-labeled at the N terminus with a c-Myc tag, to compare the c-Myc tag band versus the native cTnT band in gels and with Western blot analysis (7,57). Using this approach, we consistently see Ͼ80% endogenous cTn replacement by cTn containing the c-Myc tagged cTnT in myofibrils (7). This suggests the exchange protocol is very efficient and changes in contractile parameters should be attributed to the exchanged cTn containing either cTnI R146G or cTnI R21C .
To study the effects of PKA phosphorylation, myofibrils exchanged with cTn containing either cTnI WT or cTnI R146G were incubated with relaxing solution containing 100 units of PKA and DTT for 45 min. For cTnI R21C , PKA effects were studied by exchanging cTn containing cTnI R21C/S23D/S24D and were compared with the cTn containing cTnI S23D/S24D . The phosphorylation profile for cMyBP-C and titin (also phosphorylated by PKA incubation) was not measured; however, they should be similar for each group as paired comparisons of myofibrils containing cTnI R146G or cTnI R21C versus cTnI WT were made from each heart. The extent of cTnI phosphorylation in exchanged myofibrils (prior to PKA treatment) was inversely related to the exchange efficiency. The phosphorylation profile is plotted in Fig. 2, B and C. A very small amount of residual phosphorylated cTnI was likely present in every exchange preparation because the exchange efficiency was not 100%. It is clear that the cTnI phosphorylation level in exchanged myofibrils was quite low, as recombinant cTnI was not phosphorylated, further confirming high exchange efficiency. PKA treatment significantly increased the cTnI phosphorylation level, resulting in over 90% of cTnI phosphorylated, which is consistent with our previous observation (7).
cTnI-R146G and cTnI-R21C Mutations Effects on Myofibril Contraction-The effects of mutations on tension development and relaxation kinetics (at 15°C) were determined from isolated myofibrils from rat LV cardiac muscle exchanged with cTn containing either cTnI R146G or cTnI R21C and compared with the WT-cTn complex. Myofibrils were exposed to continually flowing solutions that were rapidly switched to provide step increases and decreases in bathing [Ca 2ϩ ], from relaxing solution (pCa 9.0) to either maximal (pCa 4.0) or submaximal (pCa 5.4, pCa 5.6, and pCa 5.8) [Ca 2ϩ ] and then back to 9.0. Representative example tension traces for cTnI WT , cTnI R146G , and cTnI R21C exchanged myofibrils during the submaximal [Ca 2ϩ ] (pCa 5.4) activation-relaxation protocol are presented in Fig. 4. A summary of tension magnitude and kinetic parameters for rat LV myofibrils exchanged with cTn containing cTnI WT , cTnI R146G , or cTnI R21C is presented in Table 1 and Figs. 5 and 6.
The rate of contractile activation (k act ) by rapid switching of solution [Ca 2ϩ ] from pCa 9.0 to 4.0 (or submaximal Ca 2ϩ levels) includes the kinetic processes of Ca 2ϩ -dependent thin filament activation, myosin cross-bridge binding, and the sub-sequent tension development. Compared with the cTnI WT -exchanged samples (3.2 Ϯ 0.3 s Ϫ1 ), k act did not differ for either the cTnI R146G -(2.7 Ϯ 0.2 s Ϫ1 ) or cTnI R21C (3.0 Ϯ 0.3 s Ϫ1 )-exchanged myofibrils at pCa 4.0 or any sub-maximal Ca 2ϩ level tested (Table 1). For all myofibrils, k act was significantly slower at sub-maximal Ca 2ϩ levels than during maximal Ca 2ϩ activations, as reported previously for rodent cardiac myofibrils (7,47), suggesting that the Ca 2ϩ sensitivity of cardiac contraction kinetics is maintained upon introduction of HCM-associated mutations. Once the activation was completed (i.e. tension was in steady state), a rapid release-restretch protocol was applied on myofibrils to measure the rate of tension redevelopment (k tr ). The k tr protocol is designed to measure the rate of myosin cross-bridge attachment and subsequent tension generation (45) when Ca 2ϩ binding to troponin is in near steady state (e.g. the thin filament is already activated). This measurement can help to differentiate the contribution of Ca 2ϩ -mediated thin filament activation versus the cross-bridge cycling kinetics to k act . At all measured conditions (cTnI R146G -or cTnI R21C -exchanged myofibrils), k tr was faster than k act at both maximal and submaximal Ca 2ϩ levels, as reported previously for cTnI WT -exchanged myofibrils (7), suggesting thin filament activation is rate-limiting for rat cardiac myofibril tension generation from rest (diastole) at 15°C. Comparing the k act /k tr ratio can give an indication of whether thin filament activation kinetics is more rate-limiting to tension development in the R146G or R21C exchanged myofibrils as compared with the WT myofibrils. Fig. 5, E and F, demonstrates that this ratio did not change upon introduction of either mutation, suggesting the activation process is maintained.
cTnI-R146G and cTnI-R21C Mutations Affect Myofibril Relaxation-Rapid deactivation of myofibrils by switching from a maximal or sub-maximal [Ca 2ϩ ] solution to relaxing solution (pCa 9.0) induced a biphasic relaxation, an initial linear tension decay followed by a more rapid (fast) exponential decay back to the baseline tension (see example trace in the inset of Fig. 4). The rate of the slow phase relaxation (k rel, slow ) is thought to be predominantly reflective of the cross-bridge detachment rate (47,50,(62)(63)(64)(65), whereas the duration of slow phase relaxation (t rel, slow ) may be influenced by the time for the troponin to move back to a "blocked" state (66). For the maximal activations, k rel, slow was unchanged for cTnI R146G -treated myofibrils (1.3 Ϯ 0.2 s Ϫ1 , Fig. 6D) compared with cTnI WT exchanged myofibrils (1.1 Ϯ 0.1 s Ϫ1 ), whereas t rel, slow of the cTnI R146G -exchanged myofibrils (105 Ϯ 7 ms, Fig. 6E) was significantly prolonged compared with those of the cTnI WTtreated myofibrils (79 Ϯ 6 ms). Similarly, at pCa 4.0, cTnI R21Cexchanged myofibrils also prolonged t rel, slow (100 Ϯ 6 ms, Fig.  6G) and did not affect k rel, slow (1.1 Ϯ 0.1 s Ϫ1 , Fig. 6F). By analyzing the contribution of slow phase on whole relaxation, we found the contributions for the cTnI WT -, cTnI R146G -, and cTnI R21C -exchanged myofibrils at pCa 4.0 were 6, 7, and 7% of the total amplitude, respectively, suggesting cross-bridge detachment was not affected by the mutations. In contrast to the slow phase of relaxation, the much larger, rapid phase of relaxation (k rel, fast ) was determined by several sarcomeric properties as well as uneven relaxation kinetics between sarcomeres in series (63-65). There was no difference in k rel, fast  NOVEMBER

FIGURE 5. Tension (A) and pCa-tension relationship (B) for cTnI WT versus cTnI R146G -exchanged myofibrils prior to and after PKA treatment. Tension (C)
and pCa-tension relationship (D) for cTnI WT versus cTnI R21C -exchanged myofibrils prior to PKA treatment and after introduction of the bis-phosphomimic mutations. The k act /k tr ratio for cTnI WT -versus cTnI R146G -(E) and cTnI WT -versus cTnI R21C -(F) exchanged myofibrils prior to and after PKA treatment (or introduction of the bis-phosphomimic mutations). *, p Ͻ 0.05; #, p Ͻ 0.01. k rel, slow or k rel, fast between cTnI WT -, cTnI R146G -, and cTnI R21Cexchanged myofibrils. However, t rel, slow was also prolonged for the cTnI R146G -exchanged myofibrils (Tables 1 and Fig. 6E). We calculated the times to reach 50% (RT 50 ) and 90% (RT 90 ) relaxation time (RT) and found that only cTnI R146G significantly prolonged the RT 50 with respect to the cTnI WT exchanged myofibrils. cTnI-R146G and cTnI-R21C Mutations Blunt the PKA Effects on Myofibril Contraction and Relaxation-We next studied the effects of PKA (or introduction of bis-phosphomimic substitutions at Ser-23/Ser-24 of cTnI) on myofibril contraction and relaxation. Consistent with previous studies (7), we found that after treating the cTnI WT -exchanged myofibrils with PKA, T max (69 Ϯ 7 mN/mm 2 ) was maintained (Fig. 5A), and pCa 50 was right-shifted 0.2 pCa units to 5.12 Ϯ 0.03, demonstrating reduced Ca 2ϩ sensitivity of tension development (Fig. 5B). Akin to the PKA-treated WT-exchanged myofibrils, T max (66 Ϯ 5 mN/mm 2 ) was also maintained (Fig. 5C), and pCa 50 was also decreased for the S23D/S24D-exchanged myofibrils (Fig. 5D), as we previously reported (7). PKA treatment of cTnI WT myofibrils (or S23D/S24D-exchanged myofibrils) also slowed k act (2.3 Ϯ 0.3 s Ϫ1 ) during maximal Ca 2ϩ activation, although the k tr (5.6 Ϯ 0.8 s Ϫ1 ) was unchanged, suggesting PKA phosphorylation affects the kinetics of thin filament activation prior to cross-bridge binding and tension development. This can be clearly seen by calculating the k act /k tr ratio (Fig. 5, E and F) for PKA-treated WT-exchanged myofibril (or S23D/S24D-exchanged myofibrils), which is significantly decreased with FIGURE 6. Slow phase relaxation at sub-maximal Ca 2ϩ level (pCa 5.4) for WT-cTn (A), cTnI R146G cTn-(B) and cTnI R21C cTn (C)-exchanged rat LV cardiac myofibrils before (black) and after (red) PKA treatment. The kinetics (k rel, slow , D) and duration (t rel, slow , E) of slow phase relaxation for cTnI WT -versus cTnI R146G -exchanged myofibrils prior to and after PKA treatment are shown. The kinetics (k rel, slow , F) and duration (t rel, slow , G) of slow phase relaxation for cTnI WT -versus cTnI R21Cexchanged myofibrils prior to PKA treatment and after introduction of the bis-phosphomimic mutations are shown. *, p Ͻ 0.05; #, p Ͻ 0.01; **, p Ͻ 0.005. NOVEMBER 13, 2015 • VOLUME 290 • NUMBER 46

JOURNAL OF BIOLOGICAL CHEMISTRY 27757
respect to WT-exchanged myofibrils prior to PKA treatment. Following PKA treatment (or introduction of bis-phosphomimic substitutions), maximal tension was also maintained for cTnI R146G -(69 Ϯ 10 mN/mm 2 ) and cTnI R21C/S23D/S24D -(70 Ϯ 5 mN/mm 2 ) exchanged myofibrils. In contrast to cTnI WT , the Ca 2ϩ sensitivity of tension development following PKA phosphorylation was not reduced (blunted) for either cTnI R146G or cTnI R21C (Fig. 5, B and D). Additionally, maximal k act (pCa 4.0) did not change for myofibrils exchanged with cTnI R146G following PKA phosphorylation or cTnI R21C/S23D/S24D , which can be clearly observed in the plots of k act /k tr ratio from Fig. 5, E and F, suggesting the slowing of thin filament activation at maximal Ca 2ϩ was also blunted.
One of the main effects of ␤-adrenergic stimulation on cardiac function is an increase in heart rate, so faster relaxation is crucial to ensure maintained or increased diastolic ventricular filling. Thus, we measured how relaxation rates were affected following PKA treatment of cTn-exchanged myofibrils. As we observed previously (7), PKA treatment of cTnI WT -exchanged myofibrils (or cTnI S23D/S24D -exchanged myofibrils) significantly increased k rel, slow (1.8 Ϯ 0.2 s Ϫ1 ) and decreased t rel, slow (66 Ϯ 3 ms) during maximal Ca 2ϩ activation, speeding up the overall relaxation. Moreover, these effects were greater at sub-maximal Ca 2ϩ levels where the heart operates. Interestingly, for all PKA-phosphorylated cTnI R146G myofibrils or cTnI R21C/S23D/S24D -exchanged myofibrils, there were no changes in either the rate (1.3 Ϯ 0.2 s Ϫ1 versus 1.2 Ϯ 0.2 s Ϫ1 ) or the duration (91 Ϯ 5 ms versus 89 Ϯ 7 ms) of slow phase relaxation, suggesting the effects of PKA to speed relaxation were blunted for both mutations. Fig. 6, A-C, is a set of example tension traces demonstrating these findings. Additionally, no changes were detected in k rel, fast with PKA phosphorylation. We also saw that upon PKA phosphorylation (or introduction of the bisphosphomimic substitutions) to cTnI WT , both RT 50 and RT 90 were significantly decreased. Importantly, these effects were blunted with the introduction of either cTnI R146G or cTnI R21C mutations.
Molecular Dynamics Simulations-We recently reported on differences in cTn dynamics between molecular dynamics models containing the WT versus R146G cTnI (8,25). Here, we studied the dynamics of WT versus R21C cTnI in our cTn model for comparison with the R146G model. Triplicate 150-ns MD simulations were compared. The root-mean-square fluctuations (RMSF) versus the protein residue numbers of each subunit were calculated, and the average (Ϯ S.D.) RMSF of the cTnC and cTnI subunits for both cTnI WT and cTnI R21C cTn systems is presented in Fig. 7, A and B. In Fig. 7, we highlight site I (pink) and site II (the Ca 2ϩ -binding loop, blue) of cTnC, and the inhibitory peptide (green) and switch peptide regions (orange) of cTnI. Similar to the cTnI R146G model (25), fluctuations were comparable for the cTnI R21C cTn model with respect to the WT model throughout most of the residues (average RMSFs of cTnC are 2.8 and 2.9 Å, respectively; and the average RMSFs of cTnI are 2.8 and 2.9 Å, respectively). Most of the regions had no changes larger than the standard deviations. The most pronounced differences to cTnC were seen in site I (cTnC residues 28 -38, 2.2 Å for cTnI WT versus 2.6 Å for cTnI R21C cTn model) and the linker loop (cTnC residues 80 -100, 2.5 Å for cTnI WT versus 2.3 Å for cTnI R21C cTn model) connecting the N-and C-terminal lobe of cTnC. The most fluctuating region detected in cTnI for both complexes was the N terminus (cTnI residues 1-41), although the helical bundle identified as the I-T arm (cTnI residues 42-137) was the most stable region in the cTnI subunits, again suggesting its structural rather than regulatory function.
We next investigated how introduction of bis-phosphomimic substitutions to Ser-23/Ser-24 of cTnI affected the dynamics of cTn containing the cTnI R21C mutation. Fig. 7, C and D, shows the average (Ϯ S.D.) RMSF of the cTnC and cTnI subunits for cTnI R21C and cTnI R21C/S23D/S24D cTn systems, respectively. Fluctuations were increased slightly in the cTnI R21C/S23D/S24D system (average RMSF 3.1 Å) with respect to the cTnI R21C system (average RMSF 2.8 Å), akin to what we previously reported for WT and cTnI S23D/S24D simulations (8), as well as the cTnI R146G and cTnI R146G/S23D/S24D simulations (25). Previously, we observed a significant change (p Ͻ 0.001) in NcTnI upon introduction of bis-phosphomimic substitutions to the WT model (8). Interestingly, introduction of bis-phosphomimic substitutions to the cTnI R21C model had very little impact on the stability of the N terminus of cTnI, similar to what we recently observed for cTnI R146G model (25). To better visualize how introduction of the R21C and/or bis-phosphomimic substitutions influences the subunit interactions among the cTn complexes, 15 snapshots taken every 10 ns during the entire MD simulations were superimposed (Fig. 7, E-G). For clarity, cTnC is shown in blue, cTnI in red, and cTnT in yellow. In contrast to the greater flexibility exhibited for the NcTnI in the cTnI S23D/S24D cTn model with respect to the WT model (8), the introduction of S23D/S24D to the cTnI R21C model had very little impact on the overall structures. As we discuss below, we speculate that this difference between the WT and R21C model upon introduction of the bis-phosphomimic substitutions may result in interference of the interaction of NcTnI with other regions of the cTn complex, and thus blunt the effects of PKA normally seen for the WT system. We next measured the time evolution of distances between the bound Ca 2ϩ ion and the six coordinating residues of site II (Asp-65, Asp-67, Ser-69, Thr-71, Asp-73, and Glu-76) over the course of each 150-ns simulation. Among these six coordinating residues, four (Asp-65, Asp-67, Asp-73, and Glu-76) exhibited no significant difference in fluctuations in any of the simulations, and these residues were always coordinated with Ca 2ϩ (results not shown). The distances for the other two coordinating residues, Ser-69 and Thr-71, fluctuated much more and varied for each run in all the four systems. Thr-71 generally did not coordinate, in agreement with the structural data from x-ray crystallography (results not shown) (51). Fig. 8 shows the distances between Ca 2ϩ and Ser-69. It is clear that Ser-69 had the most pronounced difference for Ca 2ϩ coordination, in agreement with our previous observations (8,25,27,40). The percentage of contact time for Ser-69 varies among different systems. Compared with the WT system (10%), the coordinating time of Ser-69 was increased to 23% in the cTnI R21C system. Although this was not statistically significant (p ϭ 0.3981), it suggests a stronger interaction that could provide stabilization.
This may interpret the increased Ca 2ϩ binding affinity of cTnI R21C with respect to the WT observed from the steadystate fluorescence measurements. Interestingly, the contact time was decreased to 17% upon introduction of the phospho-mimic substitutions to R21C, in agreement with the reduction of Ca 2ϩ binding affinity (K Ca ) observed from the steady-state fluorescence measurements. Interestingly, we previously (25) found that compared with the WT system, the coordinating  NOVEMBER 13, 2015 • VOLUME 290 • NUMBER 46 time of Ser-69 in the cTnI R146G system was increased to 29%. Upon introduction of the bis-phosphomimic substitutions to cTnI R146G , the contact time was decreased to 6%. Thus, the results obtained for cTnI R146G were similar to cTnI R21C (prolongation of the coordinating time). However, upon introduction of the bis-phosphomimic substitutions, cTnI R146G displayed a greater reduction of the coordinating time compared with cTnI R21C .

R146G and R21C cTnI Disrupt PKA Modulation of Contraction
Next, the residue-residue contacts of key regions were monitored over the time course of the entire 450-ns simulations. Fig.  9 shows the corresponding average contact map of residueresidue pairs (left) and the representative binding pattern (right) between the N terminus (cTnI residues 1-41, shown in red) and inhibitory peptide region (cTnI residues 138 -147, shown in blue) of cTnI for the WT (A), cTnI-S23D/S24D (B), cTnI-R21C (C) and cTnI-R21C/S23D/S24D (D) cTn models. As we reported previously (8), there were no interactions between the N terminus and the inhibitory peptide region of cTnI (Fig. 9A) for the WT model. A dramatic change was observed upon introduction of the bis-phosphomimic substitutions to Ser-23/Ser-24 of the WT complex (cTnI-S23D/S24D) (8), with residues 9 -14 interacting with residues 140 -142 of the inhibitory peptide for ϳ50% of the entire simulation (Fig.  9B). No intra-subunit interactions of cTnI were observed in the cTnI-R21C system (Fig. 9C), similar to the WT system, which is not surprising because bis-phosphomimic substitutions were not present. However, when the bis-phosphomimic residues were introduced to the cTnI-R21C system (cTnI-R21C/S23D/ S24D, Fig. 9D), the intra-subunit interaction still did not form, as we observed previously for the cTnI-R146G system (25). We have suggested that this intra-subunit interaction may destabilize contacts between the cTnI switch peptide and hydrophobic residues in the NcTnC that occur following Ca 2ϩ binding to activate contraction.
To examine this region of direct interaction between cTnC and cTnI following Ca 2ϩ binding, we examined the contact stability of the cTnI switch peptide with hydrophobic NcTnC residues. Fig. 10 shows the different contact maps of residueresidue pairs between 14 hydrophobic residues of NcTnC (from left to right: they are Phe-20, Ala-23, Phe-24, Ile-26, Phe-27, Ile-36, Leu-41, Val-44, Leu-48, Leu-57, Met-60, Phe-77, Met-80, and Met-81) and the cTnI switch peptide (cTnI residues 148 -164) for the different systems. Because Ca 2ϩ and the cTnI switch peptide were present at the start of the MD model simulations, and we did not remove Ca 2ϩ during simulations, we did not expect to see dramatic structural changes in this region. However, a change in the fluctuation of contacts can be considered as an indicator of stability of the cTnC-cTnI interaction associated with activation. As compared with the WT complex, there was little change in contact time upon introduction of the R21C mutant (Fig. 10A). A more dramatic change was seen upon introduction of the bis-phosphomimic residues to Ser-23/Ser-24 of WT complex (cTnI-S23D/S24D), suggesting decreased interaction between NcTnC hydrophobic residues and cTnI switch peptide upon phosphorylation (Fig. 10B). However, with introduction of the bis-phosphomimic to cTnI-R21C system (Fig. 10C), there was little change in fluctuation for the contacts compared with the cTnI-R21C or WT systems. Together with the contact information for the cTnI intra-subunit interaction (Fig. 9), our MD simulations suggest that phosphorylation of Ser-23/Ser-24 cTnI results in intra-subunit interaction of the cTnI N terminus with the inhibitory peptide, which reduces stability of cTnI switch peptide contacts with the cTnC hydrophobic patch, and that both the R21C and R146G cTnI mutations abrogate this action. In turn, this suggests a potential structure-based mechanism of how these mutations impair PKA regulation of contraction and relaxation.

Discussion
In this study, we tested whether HCM-associated mutations located in either the N terminus or inhibitory peptide of cTnI may disrupt the formation of an interaction between these regions that occurs with Ser-23/Ser-24 phosphorylation by PKA, thus blunting the regulatory effects on cTn that normally occur during ␤-adrenergic stimulation. We report here the effects of cTnI R146G or cTnI R21C Ϯ PKA phosphorylation (or bis-phosphomimic substitutions) on K C-I , the contractile properties of isolated rat LV cardiac myofibrils, and the whole troponin structure and dynamics changes. The most significant findings of the current study were as follows: 1) Both mutations Here, the 1st run result is shown in black; the 2nd run result is in red, and the 3rd run result is in blue.
significantly increased K Ca and K C-I compared with the cTnI WT . However, although PKA phosphorylation of cTnI resulted in a similar reduction of K Ca for WT and both mutantcontaining cTn complexes, the reduction in K C-I seen for cTnI WT was eliminated for both mutations. 2) T max was maintained for both cTnI R146G -and cTnI R21C -exchanged myofibrils, and the Ca 2ϩ sensitivity of tension (pCa 50 ) was leftshifted. However, although PKA phosphorylation (or bis-phosphomimic substitutions) decreased pCa 50 (0.2 pCa units) for WT myofibrils, this effect was blunted for both mutations. 3) PKA phosphorylation of WT myofibrils accelerated the early slow phase relaxation, especially during the sub-maximal Ca 2ϩ levels that heart operates in vivo, but most importantly, this effect was blunted for both cTnI R146G -and cTnI R21C -exchanged myofibrils. 4) MD simulations suggest the mechanism by which cTnI R146G and cTnI R21C blunt PKA-mediated reduction of K C-I , Ca 2ϩ sensitivity of tension, and the early phase of relaxation is inhibition of the formation of an intra-subunit The blue end of the spectrum (value 0) reflects no contact between residue-residue pair, and the red end of the spectrum (value 1) represents 100% contact between residue-residue pair. NOVEMBER 13, 2015 • VOLUME 290 • NUMBER 46 interaction between the N terminus and the inhibitory peptide of cTnI. This is consistently seen for WT-cTn with introduction of the bis-phosphomimic substitutions of cTnI (S23D/S24D).

R146G and R21C cTnI Disrupt PKA Modulation of Contraction
Effects of Mutations on Contractile Activation and Relaxation Kinetics-Previous studies to determine the effects of cTnI R146G on the maximal tension production and the Ca 2ϩ sensitivity of tension generation in the cardiac muscle have produced complex and sometimes contradictory results. For example, Takahashi-Yanaga et al. (19) reported that the cTnI R145G resulted in an increase in the Ca 2ϩ sensitivity of force generation and myofibrillar ATPase activity in skinned muscle fibers. In contrast, using the reconstituted actin-tropomyosinactivated myosin ATPase assay, Lang et al. (14), Takahashi-Yanaga et al. (19), and Elliot et al. (18) reported that cTnI R145G decreases the maximal ATPase in the presence of Ca 2ϩ and reduces inhibition of actomyosin ATPase activity in the absence of Ca 2ϩ . Using the human cTnI R145G -exchanged into murine myofibrils, Kruger et al. (21) reported no change in the Ca 2ϩ sensitivity of tension development; however, a slightly decreased Ca 2ϩ sensitivity was detected in the myofibrils from transgenic cTnI R146G mice. In this study, compared with WT cTn-exchanged myofibrils, T max was maintained for cTnI R146G -and cTnI R21C -exchanged myofibrils, and the Ca 2ϩ sensitivity of tension (pCa 50 ) was left-shifted by 0.13 and 0.10 pCa units, respectively. This agrees with the steady-state fluorescence measurements that showed both mutations increased K Ca . To understand the structural basis of this, we monitored the time evolution of distances between the Ca 2ϩ ion (site II) and its six coordinating residues over the course of multiple 150-ns MD simulations and found that Ser-69 coordination with Ca 2ϩ was increased with the cTnI R21C mutation (Fig. 7), in agreement with our previous report for simulations with cTn containing cTnI R146G (25). This may explain how the Ca 2ϩ sensitivity of contractile activation is left-shifted for both cTnI R146G and cTnI R21C , in accordance with previous studies on cTnC mutations where Ca 2ϩ binding at site II was stabilized (36).
To better determine how both mutations affect thin filament activation and cross-bridge kinetics, we compared the rapid release-restretch protocol (k tr ) with the Ca 2ϩ -activation protocol (k act ) during the same activation trial. Here, k tr was faster than k act for cTnI WT -, cTnI R146G -, and cTnI R21C -exchanged rat cardiac myofibrils at all Ca 2ϩ levels we tested, indicating that thin filament activation was rate-limiting for tension generation of the rat myofibrils at 15°C. This finding is consistent with our previous work in rodent hearts (7,47) but different from several other references that reported no difference between k act and k tr (62,66). As we have previously discussed (47), the presence of 0.5 mM P i in our solutions (a level of P i that is close to the physiological level of P i present in the heart) can explain this difference, and when a phosphate "mop" is used in activation solutions, the difference is eliminated. The presence of P i influences the cross-bridge tension generating isomerization specifically, without affecting thin filament activation kinetics (47). Our current results confirm those previous findings in rodents and extend them by demonstrating that a similar effect occurs at both maximal and sub-maximal Ca 2ϩ levels with introduction of either of these putative HCM-associated cTnI mutations.
Rapid, complete Ca 2ϩ removal from cardiac myofibrils (Fig.  3) by a rapid solution switching protocol induces two distinct relaxation phases (biphasic), starting with an initial early slow phase of relaxation and followed by a more rapid (fast) relaxation phase. During the slow phase, isometric conditions are maintained in sarcomeres and the force decays with a linear constant rate, indicating that k rel, slow primarily reflects the Color green (value 0) reflects no difference between the two systems; the red end of the spectrum (values above 0) reflects more contacts in the R21C, S23D/S24D, or R21C/S23D/S24D cTn system, and the blue of the spectrum (values below 0) indicates more contacts in WT, WT, or R21C model. turnover kinetics of cross-bridge cycling, dominated by the detachment rate (62,66). The duration of the initial slow phase of relaxation (t rel, slow ) depends on the Ca 2ϩ activation levels and can be influenced by the Ca 2ϩ binding and, likely, the cTnC-cTnI interaction properties of cTn. Here, compared with WT myofibrils, k rel, slow was maintained, whereas t rel, slow was prolonged upon introduction of either cTnI R146G or cTnI R21C mutations at both maximal and submaximal Ca 2ϩ levels. Consistent with previous observations (7), k rel, slow also accelerated at sub-maximal Ca 2ϩ level for all conditions. During submaximal Ca 2ϩ activations, there is less Ca 2ϩ binding to thin filaments (troponin) at any given time, so that it is easier for the thin filament to become deactivated when myosins detach. Moreover, the presence of 0.5 mM P i in our study should exacerbate this effect, as it results in a reduction of the tensionbearing cross-bridges. This may accelerate the detachment of myosin cross-bridges from the thin filament, thus contributing to an increase in the slow phase of relaxation at submaximal Ca 2ϩ levels.
Effects of PKA Phosphorylation on Contractile Activation and Relaxation Kinetics-␤-Adrenergic stimulation is a major physiological mechanism to meet the increase in circulatory demand, acting through positive inotropic and lusitropic effects (6). For cTnI, ␤-adrenergic stimulation results in the phosphorylation of Ser-23/Ser-24 of cTnI by PKA (6). Considering its key role in the heart performance, we studied how both mutations influence the PKA responsiveness. With PKA phosphorylation (or bis-phosphomimic substitutions), T max did not differ for cTnI WT -, cTnI R146G -, and cTnI R21C -exchanged myofibrils. However, although PKA phosphorylation (or with bisphosphomimic substitutions) right-shifted pCa 50 (ϳ0.2 pCa units) for WT myofibrils, this effects was blunted for both mutations (Fig. 4, B and D). In addition, during the maximal Ca 2ϩ activation, the ratio of k act /k tr was decreased from 0.67 Ϯ 0.03 to 0.48 Ϯ 0.05 for WT myofibril upon PKA treatment (or to 0.45 Ϯ 0.05 with introduction of the bis-phosphomimic substitutions), suggesting a slowed thin filament activation process with PKA phosphorylation. Interestingly, k act /k tr did not differ with PKA phosphorylation (or bis-phosphomimic substitutions) in the presence of either cTnI R146G or cTnI R21C , suggesting the slowing of thin filament activation by PKA-mediated cTnI phosphorylation during maximal Ca 2ϩ conditions is blunted. Consistent with previous work in our laboratory (7) and by others (6,9), we demonstrated that PKA treatment (or with bis-phosphomimic substitutions) increased the speed of the slow phase of relaxation for WT myofibrils, especially at the submaximal Ca 2ϩ levels that heart operates during a cardiac twitch. Most importantly, we found that this effect of PKA on slow phase relaxation was eliminated (blunted) for both mutations.
It is important to point out that the conditions for PKA phosphorylation and introduction of bis-phosphomimic substitutions (S23D/S24D) are different in vivo, because cMyBP-C and titin are also targets for PKA phosphorylation during ␤-adrenergic stimulation (67,68). So, to determine the specific effect on cTnI phosphorylation, we exchanged recombinant cTn containing cTnI S23D/S24D into cardiac myofibrils. In steady-state fluorescence studies, we found that cTnI S23D/S24D and PKA-mediated phosphorylation of cTnI (cTnI Ser(P)-23/Ser(P)-24 ) resulted in an almost identical effect, a right shift of the K C-I and K Ca curves (Fig. 3) compared with cTnI WT . The pCa 50 shift was Ϫ0.31 and Ϫ0.29 pCa units for the cTn complex containing cTnI Ser(P)-23/Ser(P)-24 versus cTnI S23D/S24D , respectively. We also saw very similar results on the modulation of thin filament activation and myofibril relaxation for PKA-phosphorylated cTnI WT -exchanged myofibrils and cTnI S23D/S24D -exchanged myofibrils. After treating the cTnI WT -exchanged myofibrils with PKA, T max was maintained and pCa 50 was right-shifted 0.2 pCa units. Similarly, T max was maintained, and pCa 50 was also decreased (0.21 pCa units) for the cTnI S23D/S24D -exchanged myofibrils. For relaxation, both PKA treatment of cTnI WT -exchanged myofibrils and cTnI S23D/S24D -exchanged myofibrils significantly increased the rate and decreased the duration of early slow phase relaxation, especially at sub-maximal Ca 2ϩ levels where the heart operates. All the above findings suggest similar functional effects of PKA phosphorylation and cTnI S23D/S24D substitutions in our systems. In this study, the PKA-phosphorylated cTnI R146G data were compared with PKA-phosphorylated cTnI WT sets, and the cTnI R21C / S23D/S24D results were compared with cTnI S23D/S24D sets. Considering the same (similar) functional effects of PKA phosphorylation and cTnI S23D/S24D substitutions, we decided it was fair to compare the PKA-phosphorylated cTnI R146G with the cTnI R21C/S23D/S24D data.
We and others (27,28) have demonstrated that cTnI R21C disrupts PKA phosphorylation of Ser-23/Ser-24 on cTnI, and this abrogates the effect of ␤-adrenergic stimulation on cTnI regulation of contraction and relaxation. This may be the actual physiological/pathogenic mechanism of cTnI R21C , where normally PKA-mediated phosphorylation of Ser-23/Ser-24 would speed relaxation of the contractile apparatus. To investigate whether "forced phosphorylation" could overcome this, we studied bis-phosphomimic substitutions S23D/S24D of cTnI R21C . Our results indicated that even with bis-phosphomimic substitutions, the phosphorylation-mediated effects on K C-I and myofibril relaxation were still blunted, suggesting that the cTnI R21C mutation per se results in the cardiac dysfunction of modulation by phosphorylation, similar to our results for the cTnI R146G mutation. Considering the similar results for cTnI R146G and cTnI R21C (with/without PKA phosphorylation or bis-phosphomimic substitutions) in both solution biochemistry and myofibril kinetics/mechanics measurements, we wanted to compare their phenotypes in humans, but unfortunately the data are rather sparse. An affected individual with the R145G mutation had ventricular hypertrophy characteristic of severe HCM (12). The missense mutation R21C has been identified in at least two families (69). In one family, a patient had apical hypertrophy after presenting with atrial fibrillation. The patient's father, three siblings, and an 18-year-old daughter all succumbed to sudden cardiac death. The clinical evaluations of three surviving mutation carriers from this family revealed that one had asymmetrical septal hypertrophy; another had isolated left atrial enlargement, and the third one had normal cardiac dimensions despite an abnormal electrocardiogram. Another family was recently also identified with the R21C mutation. This family has four members with subaortic asymmetrical R146G and R21C cTnI Disrupt PKA Modulation of Contraction NOVEMBER 13, 2015 • VOLUME 290 • NUMBER 46 hypertrophy and one mutation carrier with normal cardiac dimensions who had to be resuscitated from sudden death. Thus the sparse amount of data available suggest human phenotypes for both mutations may be similar, but more information is needed to conclude this. To understand the structural molecular level basis of how phosphorylation of cTnI residues Ser-23/Ser-24 changes the behavior of whole cTn complex and thus results in the changes in cardiac function, we performed paralleled MD simulations on WT, cTnI R146G , and cTnI R21C containing cTn and found that introduction of the bis-phosphomimic substitutions significantly altered the cTnC-cTnI interactions, particularly in the inhibitory-switch peptide regions (8). The most significant finding is that there were no intra-subunit interactions in the WT model in the absence of phosphorylation, but introduction of the bis-phosphomimic substitutions (S23D/S24D) for the WT model led to the formation of an intra-subunit interaction between the N terminus and the inhibitory peptide regions of cTnI ( Fig. 9B) (8). This intra-subunit interaction has been suggested by Solaro and co-workers (11,30,32) based on solution biochemical and spectroscopic studies (31). In addition, we also compared our simulation results of bis-phosphomimic cTnI with biochemical studies from other laboratories (70 -72). Dong et al. (70) found that bis-phosphorylation resulted in a reduction of the axial ratio of cTnI and the formation of a more compact structure upon phosphorylation using fluorescence studies. Heller et al. (71) reported that bis-phosphorylation induced a dramatic bending of the rod-like cTnI at the N-terminal extension that binds with cTnC, resulting in a significant decrease in the axial ratio of cTnI and the cTn complex overall. Reiffert et al. (72) used surface plasmon resonance to determine that the shape of cTnI changed from an asymmetrical structure to a more symmetrical one upon phosphorylation, which is consistent with the bending that results in a shorter and broader structure. Our simulations also suggest a bending at the N-terminal extension of cTnI and a more compact cTn structure upon phosphorylation that is consistent with these previous biochemical studies. We further speculate that this intra-subunit interaction may subsequently weaken interactions of the cTnI switch peptide with NcTnC, as demonstrated by increased fluctuation of contacts in MD simulations (Fig. 10B). This would allow stronger interaction between the inhibitory peptide of cTnI and actin and move the equilibrium toward thin filament deactivation. Interestingly, this intra-subunit interaction no longer formed in simulations with introduction of the cTnI R146G or cTnI R21C to the cTn complex, demonstrating that both mutations blunted the ability of cTnI phosphorylation to reposition the N-terminal extension to interact the inhibitory peptide region. These findings suggest a structural mechanism that can explain the loss of PKA-mediated modulation of thin filament activation and relaxation of myofibrils that need to occur with increasing heart rate during ␤-adrenergic stimulation and increased physical activity.
An important caution of using our simulations to explain our experimental data is that recombinant cTn subunit proteins were made from rat sequences, although the MD simulations were based on the human sequence. For cTnC, there is only one amino acid difference (Ile-119 in human and Met-119 in rat), which is a conservative substitution. For cTnT, there is also only one amino acid difference (Phe-251 in human and His-251 in rat) in the portion included in our computational model (residues 236 -285), and this is also a conservative substitution. For cTnI, there are total 13 variants in the portion used for our computational model (residues 1-172). Four of those are located in the N terminus of cTnI (Gly-4, Arg-10, Arg-13, and Ile-19 in human and Glu-4, Gly-10, Gln-13, and Val-19 in rat). One variant is located in the switch peptide of cTnI (Ala-161 in human and Thr-162 in rat). All the other variants are resided in the I-T arm region of cTnI (Leu-53, Leu-61, Ala-75, Glu-84, Ala-86, Ala-91, Ile-114, and Phe-133 in human and Met-54, Met-62, Leu-76, Val-85, Asp-87, Glu-92, Val-115,and Tyr-134 in rat). Among them, G4E, R10G, E84V, A86D, and A91E substitutions change the electrostatic charge properties and size of the amino acid; R13Q changes the charge property of the amino acid; A75V changes the size of the amino acid; and, I19V, L53M, L61M, I114V, F133Y, and A161T are conservative substitutions. There is no evidence to suggest these variants alter structure-function of the cTn complex, but this has not been studied in any detail. None of these positions has been reported to be associated with disease, which supports the idea that these sequence variants among the two species have little or no effects on the function. Additionally, most results of ATPase assays and the Ca 2ϩ -force relationship show consistent results between rodents and human. Some minor differences have been attributed to the different myosin isoforms between the two species. For these reasons, we think it is fair to compare the in vitro findings with the computational results. In the future, it would be interesting to perform the in vitro study based on the human sequence (and/or also conduct the computational modeling based on the rodent sequence) and to compare those results with our current findings to further investigate the disease-unrelated variants among species.
Author Contributions-M. R., J. A. M., and A. D. M conceived this study. Y. C. and V. R. designed the experiment; Y. C., V. R., A. T., D. W., and L. O. performed the experiment and analyzed the data; Y. C. and S. L. built up the computational model and performed the computational study; Y. C. and M. R. wrote the initial draft of the paper. All authors reviewed, edited, and approved the manuscript.