Phosphorylation specificities of protein kinase C isozymes for bovine cardiac troponin I and troponin T and sites within these proteins and regulation of myofilament properties.

Protein kinase C (PKC) isozymes alpha, delta, epsilon, and zeta, shown to be expressed in adult rat cardiomyocytes, displayed distinct substrate specificities in phosphorylating troponin I and troponin T subunits in the bovine cardiac troponin complex. Thus, because they have different substrate affinities, PKC-alpha, -delta, and -epsilon phosphorylated troponin I more than troponin T, but PKC-zeta conversely phosphorylated the latter more than the former. Furthermore, PKC isozymes exhibited discrete specificities in phosphorylating distinct sites in these proteins as free subunits or in the troponin complex. Unlike other isozymes, PKC-delta was uniquely able to phosphorylate Ser-23/Ser-24 in troponin I, the bona fide phosphorylation sites for protein kinase A (PKA); and consequently, like PKA, it reduced Ca2+ sensitivity of Ca2+-stimulated MgATPase of reconstituted actomyosin S-1. In addition, PKC-delta, like PKC-alpha, readily phosphorylated Ser-43/Ser-45 (sites common for all PKC isozymes) and reduced maximal activity of MgATPase. In this respect, PKC-delta functioned as a hybrid of PKC-alpha and PKA. In contrast to PKC-alpha, -delta, and -epsilon, PKC-zeta exclusively phosphorylated two previously unknown sites in troponin T. Phosphorylation of troponin T by PKC-alpha resulted in decreases in both Ca2+ sensitivity and maximal activity, whereas phosphorylation by PKC-zeta resulted in a slight increase of the Ca2+ sensitivity without affecting the maximal activity of MgATPase. Most of the in vitro phosphorylation sites in troponin I and troponin T were confirmed in situ in adult rat cardiomyocytes. The present study has demonstrated for the first time distinct specificities of PKC isozymes for phosphorylation of two physiological substrates in the myocardium, with functional consequences.

Actomyosin MgATPase, the molecular motor of cardiac muscle contraction, is regulated by Tn 1 and Tm. The Tn complex consists of three subunits, i.e. the Ca 2ϩ -binding TnC, the ATPase-inhibiting TnI, and the Tm-binding TnT. The current understanding of Ca 2ϩ -stimulated cardiac contraction at the level of contractile apparatus, based largely on evidence obtained from skeletal muscle studies, can be summarized as follows. In the relaxed state (in the absence of Ca 2ϩ ), TnI and Tm prevent cross-bridge cycling between actin and myosin heads through steric blocking of myosin-binding sites on actin (1)(2)(3) and possibly by inhibiting a kinetic step in the actomyosin ATP hydrolysis cycle, such as release of P i (4 -6) or ADP (7) from the actin-myosin complex. The contractile apparatus becomes activated in the presence of Ca 2ϩ , which is a process beginning with a conformational change in TnC upon binding of Ca 2ϩ (8), resulting in an increased affinity of TnC for the ATPase-inhibiting region of TnI (9) and eventually functional detachment of TnI from actin (10). This allows Tm to roll toward the center of the actin helix groove and unmask myosinbinding sites on actin and ultimately leads to MgATPase activation and muscle contraction (11,12).
In addition to direct activation by Ca 2ϩ through its binding to TnC, actomyosin MgATPase can be further modulated by phosphorylation of contractile proteins (13). It has been reported that phosphorylation of TnI by PKA resulted in decreased affinity of Ca 2ϩ for TnC without affecting the maximal activity of MgATPase (13)(14)(15)(16). We reported earlier that TnI was also effectively and stoichiometrically phosphorylated by PKC in vitro (16 -19) and in situ in adult rat cardiomyocytes (20) at multiple and similar sites, resulting in decreased maximal activity of the enzyme (16, 19 -22) that was accompanied by altered interactions of phosphorylated TnI with other contractile protein components (21,22). More recently, with the use of TnI mutants in which several identified phosphorylation sites (18) were substituted by Ala or deleted, we have identified that phosphorylation of Ser-43/Ser-45 by PKC was largely responsible for the reduced ATPase activity (16). We further observed that Ser-23/Ser-24, the bona fide PKA phosphorylation sites in TnI (13), could be cross-phosphorylated by PKC when TnI was exhaustively phosphorylated, leading to decreases in both Ca 2ϩ sensitivity and maximal activity of the enzyme (16). Like TnI, TnT has been shown to be phosphorylated stoichiometrically and at multiple sites by PKC in vitro (17)(18)(19), leading to reduced Ca 2ϩ -stimulated actomyosin MgATPase activity (19,21). This effect was characterized by reduced affinity of the phosphorylated TnT toward Tm-actin and accordingly reduced affinity of the thin filament (regulated actin) toward myosin (21). Although TnT was also phosphorylated by PKC in adult rat cardiomyocytes (20,23), its in situ phosphorylation sites have not yet been determined.
PKC is the product of a gene superfamily, and 13 subspecies have been identified (24). Expression of PKC isozymes in cardiomyocytes is developmentally regulated, i.e. higher numbers and levels of the isozymes have been found in fetal/neonatal than in adult cells (25)(26)(27)(28). The current consensus is that in adult cardiomyocytes PKC-⑀ is abundantly expressed (25)(26)(27)(28)(29); PKC-␦ and -are readily detectable (25)(26)(27)(28)(29); PKC-␣ is either absent (25)(26)(27) or present (28,29); and PKC-␤ and -␥ are not expressed (25)(26)(27)(28). In agreement with Puceat et al. (28) and Ventura et al. (29), we could readily detect PKC-␣ by Western blots and the PS/Ca 2ϩ /diacylglycerol-dependent protein kinase activity (most likely due to PKC-␣) in the homogenates of adult rat cardiomyocytes. 2 It has been hypothesized that the individual PKC isozymes may have specific cellular functions as a consequence of, in part, phosphorylation of specific substrate proteins (24,30). Because brain pan PKC preparations (mixtures of isozymes) were used in all of our earlier phosphorylation and functional studies of contractile proteins (16 -22), involvements of the individual isozymes remain unclear. In the present studies, we have investigated these crucial issues. We found that PKC isozymes exhibited distinct phosphorylation specificities with certain functional consequences.

EXPERIMENTAL PROCEDURES
Preparations of Cardiac Contractile Proteins, Synthetic Peptides, Recombinant PKC Isozymes, and PKA-Bovine heart ventricles were used as the source of contractile proteins. TnC, TnI, and TnT were purified according to the method of Potter (31) and stored at Ϫ70°C in 50 mM Tris-HCl (pH 8.0) containing 6 M urea, 1 mM EDTA, and 15 mM 2-mercaptoethanol. Tn complex was reconstituted from the three individual subunits (31). Tm (32), F-actin (33), and myosin S-1 (33) were prepared as described by others. In order to prevent oxidation of TnI and Tm, 1 mM DTT or 15 mM 2-mercaptoethanol was added to all solutions throughout the preparation and reconstitution procedures. TnI and TnT peptides containing phosphorylation sites were synthesized at the Emory University Microchemical Facility. Recombinant PKC isozymes (␣, ␦, ⑀, and ) were purified from baculovirus-infected Sf9 insect cells (34 -36). PKA was purified from bovine heart extracts (37).
Phosphorylation of Contractile Proteins, TnI and TnT Peptides, and Phosphopeptide Analysis-Prior to phosphorylation, TnI and TnT were dialyzed against 10 mM Tris-HCl (pH 7.5) containing 1 mM DTT with sequential KCl concentrations of 1, 0.7, and 0.3 M (16). The conditions of phosphorylation by PKC and PKA were essentially as described elsewhere (16,19,21,22) except that PKC-␣ was activated by PS (20 g/ml)/CaCl 2 (100 M)/diolein (5 g/ml); PKC-␦ and PKC-⑀ were activated by PS/diolein but omitting CaCl 2 ; and PKC-was activated by PS only. KCl (0. Control incubations were conducted in parallel experiments using heat-inactivated PKC and PKA. Two-dimensional tryptic peptide mappings of phosphorylated TnI and TnT were carried out as described previously (16, 18 -20). Alternatively, the Tn complex was first reconstituted from the TnC, TnI, and TnT subunits (31), and the resulting complex was phosphorylated. TnI and TnT were then separated by SDS-polyacrylamide gel electrophoresis and excised from the gels for phosphopeptide analysis.
In Situ Phosphorylation in Cardiomyocytes-Isolated adult rat cardiomyocytes were prelabeled with 32 P i for 2 h and treated with or without 100 nM TPA for 10 min, and the 32 P-labeled myofibrillar proteins were separated by SDS-polyacrylamide gel electrophoresis as described earlier (20,23). TnI and TnT were then recovered for phos-phopeptide analysis (20, 23) as mentioned above.

RESULTS
Distinct substrate specificities of the four PKC isozymes (␣, ␦, ⑀, and ) shown to be expressed in adult rat cardiomyocytes (25)(26)(27)(28)(29) for phosphorylation of TnI and TnT in the bovine cardiac Tn complex (0.5-20 M) were observed (Fig. 1). Thus, because they display different substrate affinities (indicated by apparent K m values), PKC-␣ phosphorylated TnI considerably more than TnT; PKC-␦ and PKC-⑀ phosphorylated TnI much more than TnT, and PKCphosphorylated TnT much more than TnI. Furthermore, differential substrate inhibitions of the isozymes were also noted at high Tn concentrations. Notably, at 20 M Tn, phosphorylation of TnI by PKC-␣ was markedly reduced, that by PKC-␦ was moderately attenuated, and that by PKCwas nearly completely diminished; phosphorylation of TnT by PKC-␦ or PKC-was also nearly completely inhibited. These findings indicated that PKC isozymes displayed discrete specificities for the two proteins present in the Tn complex with respect to phosphorylation extents and substrate affinities or inhibitions. The above findings were confirmed in a parallel, time-dependent (5-120 min) experiment where 2 M Tn complex was phosphorylated by the isozymes (Fig. 2). The 2 R. L. Raynor and J. F. Kuo, unpublished data. same conclusion was reached in separate experiments where adult rat cardiac Tn, instead of bovine Tn, was used as substrate. 2 As previously reported (13,15,20,39), PKA phosphorylated only TnI in the bovine or rat cardiac Tn complex (data not shown).
It was suspected that PKC isozymes might exhibit further defined substrate specificities in that they would differentially phosphorylate multiple sites in TnI. We found that this was indeed the case (Fig. 3). At about the midpoint (0.2-1.0 mol of P/mol) of maximal phosphorylation of TnI by the individual isozymes and PKA, the relative extents of 32 P incorporation into the individual tryptic phosphopeptide spots (in decreasing order) were 2Ͼ3Aϳ1ϾϾ3BϾ4ϳ5ϳ6 for PKC-␣, 1ϳ3Aϳ5Ͼ2Ͼ3Bϳ6Ͼ4 for PKC-␦, 2Ͼ4Ͼ3AϾ1ϾϾ3Bϳ5ϳ6 for PKC-⑀, 4Ͼ2ϾϾ others for PKC-, and 5ϾϾ3BϾ2Ͼ others for PKA. The most noteworthy finding was that 32 P incorporation into spot 5 was uniquely and preferentially catalyzed by PKC-␦, mimicking the action of PKA. Minor labeling of spot 5 was observed only when TnI was exhaustively phosphorylated by other PKC isozymes, such as that shown for PKC-␣ (with an extent of phosphorylation of 2.5 mol of P/mol). It was also observed that 32 P labeling of spot 4 was selectively catalyzed by PKC-⑀ and -and that labeling of spots 1 and 3A by PKC-⑀ was the lowest. We have determined earlier (16,18) that phosphopeptide spot 1 contained phosphorylation site Ser-78, spot 2 contained sites Ser-43/Ser-45, spot 3A (formerly 3) contained site Thr-144, and spot 5 contained sites Ser-23/Ser-24. The phosphorylation sites in spots 3B, 4 and 6 are still unknown. The selective phosphorylation of Ser-23/Ser-24 by PKC-␦ and PKA has been previously confirmed by the absence of spot 5 in two TnI mutants, whereby these phosphorylation sites were either substituted by Ala residues (S23A/S24A) or the N-terminal sequence containing these sites was deleted (N32) when the mutants were phosphorylated by pan PKC (16) or by the individual PKC isozymes. 3 The somewhat elongated spot 5 from the sample phosphorylated by PKC-␦, compared with the relatively focused one from that phosphorylated by PKA, might reflect a difference in relative extents of phosphorylation of Ser-23 and Ser-24; Zhang et al. (39) recently reported that Ser-23 is phosphorylated more rapidly than Ser-22 by PKA and that phosphorylation of both residues is functionally relevant.
Distinct specificities for PKC isozymes in phosphorylating multiple sites in bovine TnT were also noted ( Fig. 4). At about the midpoint (0.4 -0.7 mol of phosphate incorporated per mol of TnT) of maximal phosphorylation, tryptic peptide mappings of phosphorylated TnT revealed that 32 P incorporation into phosphopeptide spots (in decreasing order) was 4AϾ5Ͼ2ϳ6Ͼ7ϾϾ1ϳ3ϳ4Bϳ8ϳ9 for PKC-␣, 2Ͼ6Ͼ4AϾ7 Ͼ5ϾϾothers for PKC-␦, 4AϾ6Ͼ5Ͼ1ϳ2ϳ3Ͼ7 for PKC-⑀, and 8Ͼ9ϾϾ others for PKC-. At higher levels of phosphorylation of TnT (1.2-2.1 mol/mol), phosphorylation of spots 1 and 3 by PKC-␣, -␦, and -⑀ became evident but that of spots 1, 2, 3, and 4A by PKCremained insignificant or undetectable (data not shown). We have previously determined that spot 1 contained phosphorylation site Thr-190, spot 4A (formerly 4) contained site Thr-199, and spot 3 or 5 contained site Thr-280 (18). More recently, spot 6 was determined to contain phosphorylation site Ser-194 in TnT that was phosphorylated by brain pan PKC, 3 a site previously identified by others (40). It was intriguing that PKCexclusively catalyzed 32 P incorporation into spots 8 and 9, previously unrecognized major phosphopetides containing unknown phosphorylation sites. The phosphorylation sites in spots 2 and 7 are also unknown at present. In order to establish the physiological relevance of sites phosphorylated in the free TnI and TnT subunits, we examined whether the same sites in the respective subunits would also be phosphorylated in the Tn complex or in cardiomyocytes. Phosphopeptide maps, essentially the same as those shown above for the free subunits (Figs. 3 and 4), were observed for phosphorylation (approximately 1 mol/mol) of the Tn complex by the respective PKC isozymes (data not shown). A minor difference, however, was noted for the latter. PKC-␦ and -⑀ (compared with -␣) preferentially phosphorylated spot 3A (compared with spots 1 and 2) in TnI. Certain in vitro phosphorylation sites in the free TnI subunit shown in Fig. 3 were also identified in cardiomyocytes incubated under the basal condition, and phosphorylation of these sites was stimulated by TPA (Fig. 5). These findings, while confirming an earlier study (20), also showed that phosphorylation in spot 5 (Ser-23/Ser-24) that was enhanced by TPA was likely due to activation of PKC-␦. It was surprising that the site in spot 8 in TnT was principally phosphorylated in cardiomyocytes and that this phosphorylation was not stimulated by TPA (Fig. 5). Because this site was specific for PKC- (Fig. 4) and this atypical isozyme is not activated by diacylglycerol (or TPA) and Ca 2ϩ (24) that PKC-is likely to be the major PKC subspecies that phosphorylates TnT in vivo in cardiomyocytes. Other in vitro phosphorylation sites in TnT seen in Fig. 4, however, were slightly stimulated by TPA (Fig. 5).
The functional effects of TnI and TnT phosphorylation by PKC isozymes or PKA on Ca 2ϩ -stimulated MgATPase of reconstituted actomyosin S-1 were examined next (Figs. 8 and 9), and the results are summarized (Table I). It was found that TnI phosphorylation by PKC-␣ primarily caused a decrease in the maximal activity, whereas that by PKA exclusively caused a reduction in the Ca 2ϩ sensitivity of the enzyme (Fig. 8). Phosphorylation of TnI by PKC-␦, on the other hand, had dual actions in that it affected both parameters, functioning as a hybrid of PKC-␣ and PKA (Fig. 8, Table I). The dual effects of PKC-␦ were consistent with its unique ability to cross-phosphorylate the typical PKA phosphorylation sites Ser-23/Ser-24 in spot 5 in addition to the typical PKC phosphorylation sites, including Ser-43/Ser-45 in spot 2 (Fig. 3). It has been reported, with the use of mouse and rat TnI mutants, that phosphorylation of Ser-43/Ser-45 by pan PKC decreased the maximal activity (16) and that phosphorylation of Ser-23/Ser-24 by PKA (15,16) or pan PKC (16) reduced the Ca 2ϩ sensitivity of Ca 2ϩstimulated actomyosin MgATPase. We recently reported that when recombinant mouse wild-type TnI was exhaustively phosphorylated, brain pan PKC also significantly phosphorylated Ser-23/Ser-24 and reduced Ca 2ϩ sensitivity (16). The findings from the present study indicated that PKC-␦, at least in part, present in the pan PKC preparation might be responsible for those observations. Although the functional consequence of TnI phosphorylation by PKC-⑀ was not examined here, we suspected that it would function like PKC-␣ because both isozymes caused little or no phosphorylation of Ser-23/Ser-24 (Fig. 3). The effect of PKCwas also not studied because of its low ability to phosphorylate TnI (Figs. 1, 2, and 6). Next, we examined the effects of TnT phosphorylation by PKC isozymes. Because phosphopeptide maps (Fig. 4) indicated that PKC-␣, -␦, and -⑀ phosphorylated similar sites (therefore likely having similar effects) that were distinct from those phosphorylated by PKC-, we chose only PKC-␣ and -for the functional study ( Fig. 9). Because phosphorylation of TnT by PKC-␣ yielded marked decreases in both Ca 2ϩ sensitivity and activity of MgATPase, in line with earlier studies with pan PKC (19,20,21), it was somewhat surprising that phosphorylation by PKCat distinct, unknown sites resulted in a slightly increased Ca 2ϩ sensitivity without affecting the activity of MgATPase (Fig. 9, Table I). It was also noted that phosphorylation of TnT by PKC-␣ produced effects on MgATPase that were more pronounced than the effects of phosphorylation of TnI by PKC-␣ or PKA. DISCUSSION We have presented evidence in the present study showing that PKC isozymes displayed distinct specificities in phosphorylating two physiological substrates in the myocardium (i.e. TnI and TnT), various sites within these proteins, and synthetic peptides containing the respective phosphorylation sites. It seems particularly worth noting that (a) PKC-␦ was unique among all isozymes in its ability to mimic PKA in phosphorylating Ser-23/Ser-24 (spot 5) in TnI, resulting in a decreased Ca 2ϩ sensitivity of actomyosin S-1 MgATPase, and (b) PKCwas unique in its ability to selectively phosphorylate unknown sites in spots 8 and 9 in TnT, leading to a slight increase in the Ca 2ϩ sensitivity without affecting the activity of MgATPase.
The crucial issues needed to be addressed are whether the in vitro phosphorylation of TnI and TnT is physiologically relevant and which PKC isozymes are functionally involved in cardiomyocytes. As reported earlier (20), the in vitro TnI phosphorylation sites for the isozymes have been collectively con- The data shown are the means Ϯ S.E. of three experiments, and the curves drawn are the "best fits" of the data to Hill's equation using nonlinear regression. The maximal activity obtained for the unphosphorylated (Control) TnI was taken as 100%. The findings were confirmed in two other sets of experiments. firmed in the in situ phosphorylation in cardiomyocytes in the present study (Fig. 5). It was surprising, however, that only spots 8 and 9 in TnT containing the unknown sites specific for PKCwere predominantly phosphorylated in myocytes, with other in vitro sites for other isozymes being only minimally phosphorylated. It appears that, at least under the present experimental conditions, all PKC isozymes (and PKA) are likely to be involved in phosphorylation of TnI, whereas PKCis likely to be the major subspecies responsible for phosphorylation of TnT in cardiomyocytes in situ and perhaps in vivo. PKC isozymes ␣, ␦, and ⑀ phosphorylated TnT in vitro at Thr-190 (spot 1), Ser-194 (spot 6), Thr-199 (spot 4A), and Thr-280 (spot 3 or 5) to various extents (Fig. 4), and phosphorylation by PKC-␣ ( Fig. 9 and Table I) and PKC-⑀ 3 led to decreased Ca 2ϩ sensitivity and activity of MgATPase. The above sites thus far identified are all located in the C-terminal half of TnT where binding to Tm and TnC occurs (11,12) and possibly where TnT may interact with actin (41,42). It is conceivable, therefore, that the unknown sites (spots 8 and 9) phosphorylated by PKCmight be located distal to the region containing other phosphorylation sites in order to produce different effects. It may be added that we have shown earlier that all in vitro phosphorylation sites for brain pan PKC in two other cardiac myofilament proteins, i.e. C-protein (20) and myosin light chain-2 (23), in addition to those in TnI mentioned earlier (20) (Fig. 5) were also confirmed in experiments using adult rat cardiomyocytes treated with or without TPA, isoproterenol, or phenylephrine. In this respect, it is of interest to identify certain pharmacological and pathophysiological conditions under which sites other than those in spots 8 and 9 (specific for PKC-) would be phosphorylated in TnT in cardiomyocytes through activation of other isozymes. Analogous to the earlier study on TnI mutants (16), the functional consequences of phosphorylation of specific sites in TnT by PKC isozymes could be further clarified by site-directed mutagenesis of its phosphorylation sites. Finally, pharmacological approaches could be used to explore the role of PKC isozymes in cardiomyocytes. For example, Puceat et al. (28) reported that phenylephrine induced translocation of PKC-␦ and -⑀, but not that of PKC-␣ and -, and that TPA caused down-regulation of PKC-␣ to be the fastest and that of PKC-⑀ to be the slowest. Martiny-Baron et al. (36) reported that indolocarbazole Gö6976 potently inhibited (IC 50 of 2 nM) the classical PKC isozymes (such as ␣) without affecting the novel (such as ␦ and ⑀) and atypical (such as ) isozymes at a concentration as high as 3 M. Studies on these issues are currently in progress.
In conclusion, the present study has demonstrated distinct substrate preferences for PKC isozymes. More importantly, dramatic differences emerged upon examination of phosphorylation sites in TnI and TnT, which appear to correlate in some instances with distinct functional differences. This biochemical approach promises to be very useful in determining the functional specificities of PKC isozymes in cardiac myofibrils.