|Home | About | Journals | Submit | Contact Us | Français|
Protein kinase C (PKC)-induced phosphorylation of troponin-I (cTnI) has been shown to regulate cardiac contraction.
Characterize functional effects of increased PKC-induced cTnI phosphorylation and identify underlying mechanisms using a transgenic mouse model (cTnIPKC-P) expressing mutant cTnI (S43E, S45E, T144E).
2D gel analysis showed 7.2 ± 0.5% replacement of endogenous cTnI with the mutant form. Experiments included: mechanical measurements (perfused isolated hearts, isolated papillary muscles, and skinned fiber preparations), biochemical and molecular biological measurements, and a mathematical model-based analysis for integrative interpretation. Compared to wild-type mice, cTnIPKC-P mice exhibited negative inotropy in isolated hearts (14% decrease in peak developed pressure), papillary muscles (53% decrease in maximum developed force), and skinned fibers (14% decrease in maximally activated force, Fmax). Additionally, cTnIPKC-P mice exhibited slowed relaxation in both isolated hearts and intact papillary muscles. The cTnIPKC-P mice showed no differences in calcium sensitivity, cooperativity, steady-state force-MgATPase relationship, calcium transient (amplitude and relaxation), or baseline phosphorylation of other myofilamental proteins. The model-based analysis revealed that experimental observations in cTnIPKC-P mice could be reproduced by two simultaneous perturbations: a decrease in the rate of crossbridge formation and an increase in calcium-independent persistence of the myofilament active state.
A modest increase in PKC-induced cTnI phosphorylation (~7%) can significantly alter cardiac muscle contraction: negative inotropy via decreased crossbridge formation and negative lusitropy via persistence of myofilament active state. Based on our data and data from the literature we speculate that effects of PKC-mediated cTnI phosphorylation are site-specific (S43/S45 vs. T144).
The trimeric protein cardiac troponin (cTn) is associated with the sarcomeric thin filament and is a key protein that regulates cardiac contraction. Calcium binds to cTn causing a conformational shift in tropomyosin and allowing actin and myosin to form a force generating cross-bridge. cTn is more than a simple on-off switch for contraction; it can exert more complex control through its phosphorylation. For example, the inhibitor subunit of cTn, cTnI, has shown significant modulatory capacity when phosphorylated by protein kinase A (PKA) and protein kinase C (PKC). Some of these aspects of control include filament sliding speed1, calcium sensitivity2, cross-bridge cycling3, and MgATPase activity4. These affect global force production and relaxation responses to afterload5, frequency5 and length6.
There are at least five phosphorylatable sites on cTnI: serines at 23, 24, 43 and 45 (S23, S24, S43, S45) and a threonine at 144 (T144). The sites nearest the N-terminal (S23, S24) are primarily phosphorylated by PKA, while the other three sites (S43, S45, T144) are primarily phosphorylated by PKC. There may be other phosphorylation sites, including S1507 and S76/T778 (in human) as well as other kinases that act on cTnI, including p21-activated kinase (PAK)7 and mammalian sterile 20-like kinase 1 (Mst1)9. In the present study, we have focused on PKC phosphorylation for two reasons: (1) PKC is upregulated during heart failure10, 11 and (2) its effects are largely dominant over those of PKA4.
We previously created a transgenic mouse with mutated PKC phosphorylation sites on cTnI: S43 and S45 were replaced with alanines to mimic unphosphorylatable sites (cTnIS43/S45-NP). Decreased phosphorylation at these sites causes a positive inotropic response12, 13, no change in relaxation13 or calcium sensitivity14, increased MgATPase activity14, and increased susceptibility to ischemic contracture15. The cTnIS43/S45-NP mouse also rescued cardiac function when it was crossed with a heart failure model (PKCε overexpression mutation)16.
To study increased PKC phosphorylation of cTnI, we created a new transgenic mouse model (cTnIPKC-P) wherein S43, S45 and T144 on cTnI were replaced with glutamic acids to simulate constitutive phosphorylation. Note this new model mutated the threonine at position 144, which the cTnIS43/S45-NP mouse did not. Authentic phosphorylation of these three sites by PKC produced responses that closely mimic the effect of pseudo-phosphorylation using glutamic acid in reconstituted fibers1. Based on previous work with the cTnIS43/S45-NP mouse, we hypothesized that the cTnIPKC-P mouse will exhibit negative inotropy and unchanged relaxation. Our goal was to test this hypothesis and determine how activator calcium, myofilament activation, and myofilament contraction contribute to the observed functional effects.
Serine 43 and 45 and threonine 144 residues on cTnI were replaced with glutamic acid to simulate constitutive phosphorylation. Steps to create the transgenic mouse were similar to the creation of the cTnIS43/S45-NP mouse and have been described elsewhere15. This mouse will be referred to as cTnIPKC-P, or the transgenic (TG) mouse.
2D-DIGE was used to determine percent replacement of endogenous cTnI with mutant cTnI in the cTnIPKC-P mice, changes in the post-translational modifications of cTnI (primarily phosphorylation) and other myofilament proteins.
The mice were anesthetized with an intraperitoneal injection of Avertin (2,2,2-tribromoethanol, 250 mg/kg body weight), and hearts quickly excised. Left ventricular (LV) pressure was measured in isolated heart preparations17; wall stress (σ) was estimated using Online Equation A2.118. Force and intracellular calcium transients [Ca]i, were measured simultaneously19 in intact posterior right ventricular (RV) papillary muscles. Isometric force and ATPase activity were measured simultaneously20 in skinned LV papillary muscle strips (sarcomere length (SL) set at 2.2 µm).
For the isolated heart and papillary muscle data, regression analysis with dummy variables was used to identify differences in various relationships between wild-type and transgenic groups. Student’s t-tests were used to compare all other data. Data are expressed as mean ± SEM.
The cTnIPKC-P mice showed no differences in body mass (WT: 30.7 ± 1.6 g, cTnIPKC-P: 31.3 ± 1.0 g), LV mass (WT: 102.4 ± 6.7 mg, cTnIPKC-P: 103.7 ± 4.2 mg), or their ratio (WT: 296 ± 9, cTnIPKC-P: 303 ± 9). There were also no overt signs of heart failure, including lethargy or differences in feeding.
Proteins from WT and cTnIPKC-P myofibrils were separated using 2D-DIGE (Figure 1A). The 2D-DIGE gels were capable of separating cTnI species with post-translational modifications (PTMs) and unmodified species (U: unmodified cTnI, Px: post-translational modified cTnI, where x = number of possible modifications). In cTnIPKC-P samples, there was an additional spot near P3, indicated by an asterisk in Figure 1A, which was not present in the WT samples. We postulated this spot represented mutant cTnI.
To confirm this spot was constitutively pseudo-phosphorylated mutant cTnI, we treated WT and cTnIPKC-P myofibrils with PP1A and PP2A1 to de-phosphorylate cTnI (Figure 1B). The WT cTnI spot profile was reduced to the U (unmodified) spot, indicating: (1) that phosphorylation was the primary post-translational modification, (2) complete dephosphorylation, and (3) that at least one phosphorylation site was associated with each Px spot. In the phosphatase-treated cTnIPKC-P samples, there were two spots: U and mutant cTnI (indicated by an asterisk in Figure 1B). The ratio of the intensity of the mutant cTnI spot to the total intensity (i.e., sum of unmodified spot (U) and mutant spot) represents the percent replacement, which was 7.2 ± 0.5% (n=9) (Online Table I).
To verify the protein identity of the spots, membrane transfers of the 2D-DIGE gels were probed with a specific pan cTnI antibody. Figure 1C illustrates multiple spots in the untreated samples whereas only one spot is visible in the phosphatase-treated sample, a pattern similar to that in the 2D-DIGE gels (Figures 1A and 1B, second panels). All six spots are identified in the Online Figure I panel B. The mutant cTnI spot was also confirmed by Western analysis (Online Figure I panel A).
There were no differences in the relative spot intensities (U, P1–P5) of cTnI in the WT and cTnIPKC-P samples, except the presence of the mutant cTnI spot (Figure 1A and Online Table I). This indicates that the basal pattern of actual phosphorylation of cTnI was unchanged in the transgenic mouse. Moreover, there were no differences in the phosphorylation levels of cardiac troponin T (cTnT), tropomyosin (Tm), myosin regulatory light chain (RLC), or myosin binding protein C (MyBP-C) (Figure 1D and Online Table I; n: WT = 4, cTnIPKC-P = 4).
Left ventricular pressure waveforms are shown for the Frank-Starling protocol for an individual WT (Figure 2A) and cTnIPKC-P (Figure 2B) animal over the same range of chamber volumes (12µl − Vmax). For a better visual comparison, the pressure waveforms from WT and cTnIPKC-P for a single chamber volume (Vmax) are shown superimposed, both as absolute values (Figure 2C, top panel) and normalized values (Figure 2C, bottom panel). Group averaged Pdev and Ped values over the entire range of chamber volumes are illustrated in Figure 2D.
The cTnIPKC-P mice exhibited reduced systolic function. The slope of the σdev – volume relationship was reduced in cTnIPKC-P mice (WT: 2.9 ± 0.1 mmHg•µl−1, n = 6, cTnIPKC-P: 2.5 ± 0.1 mmHg•µl−1, n = 6, P = 0.02), with no change in the intercept, indicating a 14% reduction in LV contractile state over all lengths (Figure 3A). Additionally, there was a decrease in the slope of the dσ/dtmax – volume relationship (WT: 0.084 ± 0.004 mmHg•ms−1•µl−1, n = 6, cTnIPKC-P: 0.074 ± 0.003 mmHg•ms−1•µl−1, n = 6, P = 0.03), with no change in the intercept, indicating a 12% decrease in the kinetic aspects of contraction (Figure 3B).
The cTnIPKC-P mice exhibited slowed relaxation as evidenced by a parallel, upward shift of the Trelax – volume relationship (intercept values: WT: 16 ± 1 ms, n = 6, cTnIPKC-P: 21 ± 3 ms, n = 6, P = 0.005), with no change in the slope, indicating a 20% increase (maximum) in the time for the LV to relax (Figure 3C). Consistent with this observation, there was a decrease in the magnitude of the slope of the dσ/dtmin – volume relationship (WT: −0.067 ± 0.003 mmHg•ms−1•µl−1, n = 6, cTnIPKC-P: −0.056 ± 0.003 mmHg•s−1•µl−1, n = 6, P = 0.03), with no change in the intercept, indicating a 16% reduction in the kinetic aspects of relaxation (Figure 3D).
The cTnIPKC-P mice showed no differences in passive properties compared to WT mice. There were no statistical differences in Ped (Figure 2D, triangles) or σed (Online Figure II) over the entire range of volumes studied.
Treatment with 1 µmol/L isoproterenol increased developed pressures to the same degree in both WT and cTnIPKC-P (Table 1). Isoproterenol also shortened rise and relaxation times to the same degree in WT and cTnIPKC-P animals (Table 1).
Figure 4A shows representative force and calcium data from one WT and one cTnIPKC-P experiment. The cTnIPKC-P mice exhibited decreased force production, indicated by a downward shift of the developed force (Fdev) – length relationship (intercept values: WT: 1.88 ± 1.22 mN•mm−2, n = 5, cTnIPKC-P: −0.51 ± 0.35 mN•mm−2, n = 6, P = 0.04, Figure 4B). There was also slowed relaxation in the cTnIPKC-P mice as evidenced by an upward shift in the dF/dtmin – length relationship (intercept values: WT: −55 ± 16 mN•mm−2•s−1, TG: 5 ± 8 mN•mm−2•s−1, P < 0.001, Figure 4C). In contrast, intracellular calcium concentration transients ([Ca]i) were unaltered in cTnIPKC-P mice, exhibiting no changes in the slope or intercept of the [Ca]i amplitude (as quantified by Rsys/Red) – length relationship (P-value: for slope = 0.53, for intercept = 0.83, Figure 4D), or the relaxation (as quantified by dR/dtmin) – length relationship (P-value: for slope = 0.23, for intercept = 0.28, Figure 4C).
There was a 14% significant decrease (n: WT = 10 fibers from 4 mice, cTnIPKC-P: 10 fibers from 3 mice) in maximally activated force (Fmax) in the cTnIPKC-P mice (Figure 5A and Table 2). While there was a similar decrease (9%) in maximal MgATPase activity (Figure 5C and Table 2), it was not statistically significant. The skinned fiber data were fit to the modified Hill equation (Equation A4.1). There were no differences in pCa50 (calcium sensitivity) or Hill coefficient (nH, cooperativity) for either the pCa-force (Figure 5B and Table 2) or pCa-MgATPase (Figure 5D) relationships. The force-MgATPase activity relationship was linear (Figure 5E) and its slope is defined as the tension cost. There were no differences in the tension cost between the WT (7.33 ± 0.13, R2 = 0.97) and cTnIPKC-P (7.41 ± 0.10, R2 = 0.98) mice (Table 2).
We created a new transgenic mouse model (cTnIPKC-P) wherein the three PKC phosphorylation sites on cTnI were mutated to glutamic acid to simulate constitutive pseudo-phosphorylation. Despite low integration of mutant protein (~ 7%), TG mice show significant functional changes, indicating high sensitivity of cardiac contraction to PKC cTnI phosphorylation. There are three main experimental findings of the present study. Compared to wild-type mice, cTnIPKC-P mice exhibited: (1) decreased active contraction and slowed relaxation, (2) a preserved response to β-adrenergic stimulation, (3) decreased maximally activated force without changes in calcium sensitivity or tension cost. We will discuss each of these observations individually, followed by an integrative interpretation that reconciles these experimental findings.
The cTnIPKC-P mouse showed ~7% replacement of endogenous cTnI with mutant cTnI. There were no differences in the basal actual phosphorylation pattern of cTnI. Therefore, there is an increase in total phosphorylation of cTnI at the PKC sites: the summation of basal actual phosphorylation (unchanged in cTnIPKC-P mice) and pseudo-phosphorylation (increased by ~7% in cTnIPKC-P mice). In spite of the relatively low level of replacement, there were significant functional effects. This is an unexpected and remarkable finding, suggesting a high sensitivity of cardiac contraction to PKC-mediated cTnI phosphorylation and potentially important physiologic and pathophysiologic roles for this post-translational regulatory process.
Recent evidence indicates that basal in vivo cTnI phosphorylation at the PKC sites is very low8, 21. Thus, the percent replacement in our TG mice, although small, may represent a physiologically relevant level of increased PKC cTnI phosphorylation. Our previous mouse model, cTnIS43/S45-NP (serine 43 and 45 mutated to alanine), showed ~50% replacement of endogenous cTnI by the mutant cTnI14. However, given the low basal phosphorylation state, a higher replacement of non-phosphorylatable sites would be required to observe functional effects.
There were also no alterations in phosphorylation of Tm, TnT, ELC or MyBP-C, indicating that the observed effects were from cTnI PKC pseudo-phosphorylation alone.
There was evidence of reduced contraction in the cTnIPKC-P mice compared to WT mice at all three levels studied: (1) isolated heart experiments showed depressed developed pressures, (2) intact papillary muscles exhibited lower developed force, and (3) skinned fibers showed significantly lower Fmax. In reconstituted fibers where the cTnI PKC phosphorylation sites were rendered constitutively pseudo-phosphorylated, a decrease in Fmax was also seen1. The cTnIS43/S45-NP mouse model showed an increase in developed pressures at high extracellular calcium levels (3.5 mmol/L [Ca])13. Together with previous data, the new transgenic model supports the hypothesis that phosphorylation of cTnI by PKC lowers the myocardium’s ability to generate active force, both under dynamic and steady-state activations.
In isolated heart and intact papillary muscle experiments, cTnIPKC-P mice also showed negative lusitropy when compared to control mice. This is consistent with the results of Pi et al, who treated wild-type mice with the PKC activator ET-1 and observed an increase in relaxation time (negative lusitropy)22. Furthermore, in their transgenic animal, in which all 5 cTnI phosphorylation sites were replaced with alanines, the effects of ET-1 were severely blunted. These results, combined with the results presented here, suggest that phosphorylation of the PKC sites on cTnI has a negative lusitropic effect. By that same logic, the cTnIS43/S45-NP mouse would be expected to exhibit positive lusitropy, but it did not13. However, in the cTnIS43/S45-NP mouse, the T144 is not mutated to an unphosphorylatable alanine. Thus, phosphorylation of T144 on cTnI may be primarily responsible for PKC-induced slowing of relaxation.
The cTnIPKC-P and WT mice responded similarly to isoproterenol (1µmol/L) infusion: increasing developed pressure and decreasing rise and relaxation times. β-adrenergic stimulation acts through PKA-activated pathways, specifically the phosphorylation of PKA sites on cTnI (S23 and S24) and cellular calcium handling proteins (e.g., sarcoplasmic reticulum ATPase pump, L-type calcium channels). These results show that the mutation we introduced in cTnI did not affect these PKA-dependent pathways
There were no differences in intracellular calcium amplitude or relaxation between WT and cTnIPKC-P mice. We have previously reported that cTnIS43/S45-NP mice do not exhibit any differences in intracellular calcium at normal extracellular calcium levels15. Others have also shown that there are no changes in intracellular calcium with PKC phosphorylation of cTnI22, 23. Normalized fluorescence values (R/Red or Rsys/Red), instead of calibrated data, were used in analyzing calcium transients. The reasons for using this approach and its validity are discussed in Appendix 3 (Online Supplement).
cTnIPKC-P mice did not exhibit differences in calcium sensitivity compared to WT mice. Burkart et al conducted experiments on detergent-extracted cardiac fibers reconstituted with three forms of mutant cTnI that were pseudo phosphorylated at the PKC sites: S43E/S45E, S43E/S45E/T144E, and T144E1. They showed that S43E/S45E and S43E/S45E/T144E fibers exhibited similar decreases in calcium sensitivity. However, there was no change in calcium sensitivity in T144E fibers, suggesting that phosphorylation at this site plays no role. This is inconsistent with data from Wang et al in which chemical phosphorylation of T144 by PKC-βII resulted in increased calcium sensitivity24.
The reconstituted fibers studied by Burkart et al showed a much higher replacement of endogenous cTnI with the mutant form (70–97%) than was present in our cTnIPKC-P mouse, though it is unclear if it is appropriate to directly compare the extent of replacement in the transgenic mouse and reconstituted system. Wang et al found that PKC-βII caused a 20–50% incorporation of radiolabeled phosphate, which was much closer to the level of phosphorylation simulated by our mutant cTnI. This suggests calcium sensitivity may be affected differently at very high, possibly non-physiologic, levels of PKC phosphorylation of cTnI.
Since our model simulates phosphorylation at all three PKC sites, it is possible there are two offsetting effects on calcium sensitivity; a decrease in calcium sensitivity due to the (pseudo)phosphorylation at S43/S45 and an increase in calcium sensitivity in response to the (pseudo)phosphorylation at T144. If PKC phosphorylates all three positions, what is the purpose of mutually offsetting effects at the different sites? One possibility is that certain PKC isoforms only phosphorylate or preferentially phosphorylate a given residue, resulting in unequal phosphorylation on the three sites. For example, PKC-βII24 and tyrosine-phosphorylated PKCδ25 preferentially phosphorylate T144.
Our goal was to identify the underlying changes in myofilamental processes that can simultaneously explain all of our results. We can group the contraction/relaxation processes into three main categories: 1) activator calcium, 2) myofilament activation, and 3) myofilament contraction.
There were no differences in intracellular calcium transients in the WT and cTnIPKC-P mice (Figure 4). This suggests there are no differences in the activator calcium.
The slope of the force–MgATPase activity relationship is the tension cost, and represents an estimate of the rate of cross-bridge detachment (g)26, 27. The cTnIPKC-P mice exhibited no change in tension cost compared to WT mice, implying unaltered g. Since the decrease in Fmax cannot be attributed to calcium activation or crossbridge detachment, it is most likely due to a decrease in crossbridge formation (f).
We used a mathematical model (Figure 6B) to interpret experimental results and determine what myofilamental processes were altered in the cTnIPKC-P mouse (see Appendix 5). This model has been used for data analysis before17, 28, 29.
Decreasing the value of f (23%) predicted the observed decrease in maximum twitch force and Fmax, but also showed decreased calcium sensitivity and no change in relaxation, contrary to our experimental observations. Therefore, there must be another altered process that slows relaxation. What parameter (or set of parameters), altered simultaneously with f, would produce decreased magnitude of contraction (both under twitching and steady-state conditions), increased relaxation time, and no change in calcium sensitivity or tension cost? We focused on parameters that would maintain the activation state (Fig. 6A): (1) increased calcium binding (increased k1), and (2) decreased crossbridge dissolution in the absence of cTnC-bound calcium (decreased d). Although the first perturbation can produce slowed relaxation, it was unable to reconcile all of the experimental observations.
The second perturbation (a decrease in f by 23% and a decrease in d by 35%) reproduced the experimental observations (compare Figure 5E to to6B,6B, ,2C2C to 6C and 6D, ,5A5A to to6E,6E, and and5B5B to to6F).6F). The magnitude of these changes closely mirrored the observations in the isolated heart (14% negative inotropy, 20% negative lusitropy) and skinned fiber (14% decrease in Fmax, less than 2% change in pCa50, nH, and tension cost). These magnitudes acted as a guide, the goal was not to identify exact parameter values, but identify a plausible solution which explained the major findings of our experiments. Our experimental data and the model-based analysis suggest that the transgenic mouse has two altered myofilament processes: decreased rate constant of crossbridge formation and a calcium-independent persistence of the active state.
The model also indicated that an isolated decrease in d was associated with an increase in calcium sensitivity and slowed relaxation. Experimental data suggests T144 phosphorylation causes an increase in calcium sensitivity24 and is primarily responsible for slowed relaxation. It is therefore tempting to attribute the calcium-independent persistence of myofilament active sate (i.e., decreased d) to T144 phosphorylation. Likewise, the model indicated that an isolated decrease in f was associated with a decrease in calcium sensitivity, no change in relaxation, and negative inotropy. Experimental data suggests that phosphorylation of S43 and S45 results in decreased calcium sensitivity1, no change in relaxation13, and negative inotropy13. Thus, it is possible that the decrease in rate constant of crossbrige formation (f) is a result of S43 and S45 phosphorylation. The precise biophysical mechanisms underlying the regulation of myofilament contractile properties by PKC-mediated phosphorylation of cTnI are not presently known.
(1) Although the 2D-DIGE data clearly indicate unchanged overall pattern of actual phosphorylation of cTnI in TG mice (Figure 1A and Online Table 1), we cannot definitely say that cTnI phosphorylation at PKA sites was unchanged. Additional data using mass spectrometry will be needed to quantify cTnI phosphorylation at individual sites. (2) Because a single TG mouse line was used in the present study (i.e., a single level of transgene expression), it is not possible to determine how the observed effects scale with the level of cTnI phosphorylation at PKC sites. Additional TG mouse lines and/or experiments using reconstituted fibers will be necessary to address this issue. (3) While it is a commonly used technique, pseudo-phosphorylation by glutamate replacement may not fully recapitulate actual phosphorylation. Furthermore, the simultaneous (pseudo)phosphorylation of all three PKC sites may or may not be a physiologically relevant pattern of cTnI phosphorylation. However, our TG mouse data, together with data from the literature, provide new insights into the effects of PKC-mediated cTnI phosphorylation.”
Our data show a small increase in cTnI phosphorylation at PKC sites produces significant functional changes, indicating high sensitivity of cardiac contraction to PKC-mediated cTnI phosphorylation. Model-based analysis predicts that these functional changes are brought about by specific changes in myofilament contractile properties: decreased rate of crossbridge formation and calcium-independent persistence of the active state. Based on our data and data from the literature, we speculate that the effects of PKC-mediated cTnI phopshorylation are site-specific (S43/S45 vs. T144).
The authors thank Kelly Clause and Dr. Partha Roy for assistance with molecular biology related issues and Dr. Kenneth Campbell for illuminating discussions regarding the mathematical model-based analysis.
Sources of Funding
This work was supported by the NIH (T32-HL76124 [JAK], R01-HL75643 [MC], PO1-HL62426 (Project 1) and RO1-HL64035 [RJS]) and the McGinnis Endowed Chair research funds [SGS].
Subject Codes:  Genetically altered mice,  Cell signaling/signal transduction,  Calcium cycling/excitation-contraction coupling