Tyrosine Phosphorylation Modifies Protein Kinase C δ-dependent Phosphorylation of Cardiac Troponin I*

Our study identifies tyrosine phosphorylation as a novel protein kinase Cδ (PKCδ) activation mechanism that modifies PKCδ-dependent phosphorylation of cardiac troponin I (cTnI), a myofilament regulatory protein. PKCδ phosphorylates cTnI at Ser23/Ser24 when activated by lipid cofactors; Src phosphorylates PKCδ at Tyr311 and Tyr332 leading to enhanced PKCδ autophosphorylation at Thr505 (its activation loop) and PKCδ-dependent cTnI phosphorylation at both Ser23/Ser24 and Thr144. The Src-dependent acquisition of cTnI-Thr144 kinase activity is abrogated by Y311F or T505A substitutions. Treatment of detergent-extracted single cardiomyocytes with lipid-activated PKCδ induces depressed tension at submaximum but not maximum [Ca2+] as expected for cTnI-Ser23/Ser24 phosphorylation. Treatment of myocytes with Src-activated PKCδ leads to depressed maximum tension and cross-bridge kinetics, attributable to a dominant effect of cTnI-Thr144 phosphorylation. Our data implicate PKCδ-Tyr311/Thr505 phosphorylation as dynamically regulated modifications that alter PKCδ enzymology and allow for stimulus-specific control of cardiac mechanics during growth factor stimulation and oxidative stress.

Protein kinase C␦ (PKC␦) 2 is a ubiquitous serine/threonine kinase implicated in a wide range of cellular responses (1,2). PKC␦ is conventionally viewed as a lipid-dependent enzyme that is anchored to membranes in close proximity to target substrates through interactions with lipid cofactors. However, there is recent evidence that PKC␦ also is dynamically regulated through activation loop (Thr 505 ) phosphorylation (3,4). For other PKC isoforms, activation loop phosphorylation is a stable "priming" phosphorylation completed during de novo enzyme synthesis (5). In the case of cPKCs, activation loop phosphorylations are mediated by phosphoinositide-dependent kinase-1 and are essential to generate a catalytically competent enzyme. Although newly synthesized PKC␦ also undergoes maturational phosphoinositide-dependent kinase-1dependent Thr 505 phosphorylation, PKC␦ differs from other PKC isoforms in that 1) PKC␦ is a catalytically active enzyme even without Thr 505 phosphorylation and 2) PKC␦-Thr 505 phosphorylation is dynamically regulated through an autocatalytic mechanism (4). Although there are hints that Thr 505 phosphorylation might contribute to the control of PKC␦ enzymology, a PKC␦-Thr 505 autophosphorylation mechanism that regulates the actions of PKC␦ toward a physiologically relevant substrate in a differentiated cell has never been reported.
PKC␦ also is regulated through tyrosine phosphorylation. However, the consequences of PKC␦ tyrosine phosphorylation remain disputed, because PKC␦ tyrosine phosphorylation is variably linked to increased, decreased, or unchanged PKC␦ activity (1). Inconsistencies in the literature have been attributed to the presence of multiple tyrosine residues throughout the structure of PKC␦ (including Tyr 52 , Tyr 64 , Tyr 155 , Tyr 187 , Tyr 311 , Tyr 332 , Tyr 512 , and Tyr 523 , numbering based upon rodent sequence) that are targets for independently regulated phosphorylations by Src family kinases, c-Abl, and potentially other tyrosine kinases. The general consensus in the literature is that PKC␦ tyrosine phosphorylation patterns vary according to the nature of the inciting stimulus and dictate the functional properties of the enzyme in cells. Our recent studies show Tyr 311 and Tyr 332 (residues in the hinge region of PKC␦ that links the regulatory and kinase domains) are phosphorylated in vivo in H 2 O 2 -treated cardiomyocytes and in vitro in kinase assays performed with Src (4,6). We also obtained unanticipated evidence that Src and PKC␦ tyrosine phosphorylation promote PKC␦-Thr 505 autophosphorylation in cells. These results run counter to the prevailing assumption that PKC␦ tyrosine and Thr 505 phosphorylations are independently regulated mechanisms and suggest that at least some of the dynamically regulated changes in PKC␦-Thr 505 phosphorylation in cells can be attributed to tyrosine phosphorylation (i.e. PKC␦-Thr 505 autophosphorylation constitutes a final common mechanism to control PKC␦ activity, including during oxidative stress).
PKCs induce structural and functional ventricular remodelling both by activating signaling pathways that alter gene expression and by directly phosphorylating myofilament proteins that control cardiac contraction (7). cTnI, the "inhibitory" subunit of the troponin complex, is an important PKC substrate that plays an indispensable role in Ca 2ϩ -dependent regulation of myofilament function (8). At low intracellular [Ca 2ϩ ], cTnI is anchored to the actin filament via peptides (an inhibitory peptide 138 GKFKRPTLRRVR 149 ) and a second actin-binding site (residues 161-188) flanking a so-called switch peptide (9). This tethering of cTnI acts together with other regulatory proteins to prevent strong, force-generating interactions of myosin cross-bridges with thin filaments. At high intracellular [Ca 2ϩ ], the switch peptide binds strongly to the cTnC regulatory domain and induces a movement of cTnI away from actin thereby releasing inhibition (for recent reviews see Refs. 10 and 11 and references within). Myofilament function is further regulated and fine-tuned to hemodynamic load, by cTnI phosphorylation. cTnI contains three phosphorylation clusters (Ser 23 / Ser 24 , Ser 43 /Ser 45 , and Thr 144 ) that exert distinct effects on cardiac function. Ser 23 /Ser 24 phosphorylation desensitizes myofilaments to Ca 2ϩ and leads to an enhanced rate of relaxation, augmented cross-bridge cycling, and accelerated unloaded shortening velocity. In contrast, Ser 43 /Ser 45 phosphorylation leads to a decrease in maximal actomyosin Mg-ATPase, Ca 2ϩ -activated force, and cross-bridge cycling rate. Despite the fact that Thr 144 is critically positioned in the inhibitory peptide region, the functional impact of Thr 144 phosphorylation has been less intensively studied. Recent studies suggest that Thr 144 phosphorylation may influence myofilament Ca 2ϩ sensitivity (12)(13)(14).
PKC-dependent cTnI phosphorylation is believed to be particularly important in the development of hypertrophy/failure syndromes, where coordinate increases in PKC isoform expression and cTnI phosphorylation have been linked to reduced actin-myosin interactions and depressed contractile function (7,15,16). However, conventional models of PKC␦ activation, which attribute PKC isoform specificity entirely to translocation events that localize the enzyme to membranes, do not adequately explain PKC-dependent cTnI phosphorylation in the sarcomere. A role for tyrosine-phosphorylated PKC␦ that is released from membranes and recovered as a constitutively active, lipid-independent enzyme in the soluble fraction of H 2 O 2 -treated cardiomyocytes would be more plausible (6). This study examines the role of PKC␦ as a cTnI kinase, with a particular focus on the role of Src (and PKC␦ tyrosine phosphorylation) to regulate PKC␦-dependent phosphorylation of cTnI and transduce physiologically relevant changes in contractile function.
cTnI Phosphorylation in Adult Ventricular Cardiomyocytes-Left ventricular myocytes enzymatically isolated from male Sprague-Dawley rats according to a modification of published methods were plated for 2 h on laminin-coated dishes (10 g/ml; Invitrogen) at a density of 5 ϫ 10 4 rod-shaped cells/ 35-mm dish (27). The cells were washed with fresh media, pretreated with vehicle (Me 2 SO), GF109203X (5 M), or PP1 (10 M) and then stimulated with H 2 O 2 (5 mM) or PMA (100 nM) for 15 min. Cell lysis was in ice-cold radioimmune precipitation assay buffer (with protease/phosphatase inhibitor mixture) and was followed by centrifugation (16,000 ϫ g, 4°C, 15 min) to pellet myofilament proteins, which were then resuspended in a urea/thiourea sample buffer and separated on 15% acrylamide gels. The gels were stained with Pro-Q Diamond (to detect phosphorylated proteins) followed by Sypro Ruby (to detect total protein), and analysis was by imaging on a Typhoon 9410 molecular imager (GE Healthcare) using ImageQuant TL software.
Cardiomyocyte Culture and PKC␦ Immunoprecipitation for Kinase Assays-Cardiomyocytes were isolated from hearts of 2-day-old Wistar rats by a trypsin dispersion procedure that uses a differential attachment procedure followed by irradiation to enrich for cardiomyocytes (6). The cells were plated on protamine sulfate-coated culture dishes at a density of 5 ϫ 10 6 cells/100-mm dish and grown in minimum essential medium (Invitrogen) supplemented with 10% fetal calf serum for 4 days and then serum-deprived for 24 h prior to experiments. The cells were treated with vehicle, 5 mM H 2 O 2 , or 100 nM PMA for 15 min and then lysed in homogenization buffer, and cell extracts were subjected to immunoprecipitation with anti-PKC␦ (Santa-Cruz) and used in immunocomplex kinase assays modified only to include cTn complex as substrate according to methods described previously (4,6,12).
In Incubations were for 30 min at 30°C in the absence or presence of Src or c-Abl (0.66 units) and lipid cofactors; PS-PMA was included in assays to allosterically activate PKC␦ and render it a better substrate for Src-dependent tyrosine phosphorylation (4).
Peptide Mapping Studies-For peptide mapping studies, kinase reactions (see above) were stopped by adding 27 l of 4ϫ SDS-PAGE sample buffer. The proteins were separated by SDS-PAGE and transferred to nitrocellulose, and the band corresponding to PKC␦ was excised from the membrane, cut into small pieces, and treated for 30 min at 37°C with polyvinylpyrrolidone (0.5%, w/v) in acetic acid (100 mM), followed by five water washes (to remove the acid) and a 10-min incubation at room temperature in the dark with 100 mM iodoacetate to carboxymethylate PKC␦. The membrane pieces were then washed three times with water and twice with 50 mM ammonium bicarbonate and incubated overnight at 37°C in 60 l of a buffer containing 42 mM ammonium bicarbonate, 17 M HCl, and 10 g of sequencing grade trypsin. Digested peptides were eluted from the membrane by sonication and lyophilized, and the residue was reconstituted in 0.1% trifluoroacetic acid and fractionated by reverse phase-HPLC on a Vydac semimicro C 18 column (2.1 ϫ 250 mm). The peptides were eluted with a linear gradient from 0.1% trifluoroacetic acid in water to 0.1% trifluoroacetic acid in acetonitrile over 140 min at a flow rate of 1 ml/min. The eluant was monitored at 220 nm, and fractions were collected every 30 s for Cherenkov counting. Radioactive peptides of interest were submitted to the Howard Hughes Medical Institute/Columbia University Protein Chemistry Core Facility for sequencing by MALDITOF mass spectrometry.
Mechanical Measurements in Skinned Rat Ventricular Myocytes-Rat ventricular myocytes were mechanically isolated and mounted in the experimental set-up as described previously (20). Isometric force production and the rate constant of force redevelopment (k tr ) were assessed in single, skinned myocytes, as described previously (20), before and after PKC␦ treatment alone (as described above) or in combination with Src kinase. All of the PKC␦ treatments were in the presence of lipid cofactors (PS/PMA). Lipid cofactors and Src alone had no effect on mechanical parameters.
Statistical Analysis-All of the values are presented as the means Ϯ S.E., and the values of p Ͻ 0.05 were the criteria of statistical significance. Statistical evaluation was by Student t test (KaleidaGraph, Synergy Software).

RESULTS
cTnI Phosphorylation via a PKC␦/Src-dependent Mechanism in H 2 O 2 -treated Cardiomyocytes-Initial studies examined the mechanisms that control cTnI phosphorylation in situ in the myofilament lattice of cultured adult rat cardiomyocytes subjected to oxidative stress by H 2 O 2 treatment. Fig. 1 shows that PMA (which directly activates PKCs) and H 2 O 2 induce similar increases in cTnI phosphorylation. The H 2 O 2 -dependent increment in cTnI phosphorylation is completely abrogated by GF109203X (a relatively nonselective inhibitor of PKC isoforms), and it is markedly blunted by PP1 (an inhibitor of Src family kinases). The observation that H 2 O 2 increases cTnI phosphorylation through a PKC-dependent mechanism that requires tyrosine kinase activity provided the rationale to consider a role for tyrosine-phosphorylated PKC␦.
PKC␦ was immunopurified from vehicle-, PMA-, and H 2 O 2treated cardiomyocyte cultures and used in assays with cTn ternary complexes (consisting of equimolar cTnI, cTnC, and cTnT) as substrate to resolve the cellular actions of PKC␦ from the actions of other PKC isoforms. Native PKC␦ is recovered from quiescent cardiomyocyte cultures with only very low lipid-independent cTnI kinase activity; cTnI phosphorylation is markedly increased when lipid activators, PS/PMA, are added to the in vitro kinase assay ( Fig. 2A). PMA treatment of cardiomyocyte cultures, prior to PKC␦ immunoprecipitation, leads to a minor increase in basal and lipid-dependent cTnI kinase activity, in association with a trace increase in overall PKC␦ tyrosine phosphorylation. Oxidative stress induced by H 2 O 2 treatment results in a marked increase in PKC␦ tyrosine phosphorylation and a dramatic increase in PKC␦-dependent phosphorylation of cTnI, including in assays without lipid cofactors (i.e. PKC␦ is recovered from cardiomyocytes subjected to oxidative stress as a lipid-independent cTnI kinase). Although H 2 O 2 treatment has been linked to the activation of an array of serine/threonine kinases that may potentially associate with PKC␦ during immunoprecipitation, control experi- Adult ventricular myocytes were treated with vehicle (dimethyl sulfoxide (DMSO)), GF109203X (5 M), or PP1 (10 M) followed by H 2 O 2 (5 mM) or PMA (100 nM) for 15 min. The gels were stained with Pro-Q Diamond (phosphoproteins stain) followed by Sypro Ruby (total protein stain), and the phospho/ total protein ratios were determined (n ϭ 4, means Ϯ S.E.; *, p Ͻ 0.05). The results are normalized by setting the mean value for vehicle-treated samples to 100.
ments identify similar pharmacologic profiles for the cTnI kinase activity recovered from resting and H 2 O 2 -treated cultures; in both cases, cTnI phosphorylation is fully inhibited by 5 M GF109203X and 5 M Gö6983 (broad spectrum PKC inhibitors) but not by 10 M Gö6976 (a conventional PKC isoformselective inhibitor that also directly inhibits the catalytic activity of PKD, another potential H 2 O 2 -activated cTnI kinase; Fig. 2 and data not shown). These results implicate PKC␦ as a lipidindependent cTnI kinase in cardiomyocytes exposed to oxidative stress.
PKC␦ kinase activity may be activated by H 2 O 2 indirectly as a result of Src-dependent tyrosine phosphorylation or directly through cysteine oxidation (17). To assess the importance of tyrosine phosphorylation in H 2 O 2 -dependent changes in PKC␦ activity, we compared cTnI phosphorylation by PKC␦ recovered from myocytes treated with H 2 O 2 alone or H 2 O 2 ϩPP1 (to inhibit PKC␦ tyrosine phosphorylation). Fig. 2B shows that H 2 O 2 increases overall PKC␦ tyrosine phosphorylation (tracked with a general anti-phosphotyrosine antibody) and PKC␦-Tyr 311 and -Tyr 332 phosphorylation (tracked by phospho site-specific antibodies (PSSAs)); H 2 O 2 -dependent PKC␦ tyrosine phosphorylation is blocked by PP1, under conditions that also block H 2 O 2 -dependent activation of Src (tracked by immunoblotting for Src activation loop Tyr 416 phosphorylation). PP1 attenuates (by 28.4 Ϯ 4%, n ϭ 4, p Ͻ 0.05) but does not entirely block the H 2 O 2 -dependent increase in cTnI phosphorylation by PKC␦. These results indicate that H 2 O 2 increases PKC␦ phosphorylation of cTnI through at least two mechanisms that differ in their requirements for Src-depend-ent tyrosine phosphorylation. The Src and tyrosine phosphorylationdependent mechanisms that regulate PKC␦ activity are the focus of the remainder of the studies in this manuscript.
Src Phosphorylates PKC␦ at Tyr 311 and Tyr 332 , Leading to Enhanced PKC␦ Phosphorylation of cTnI-We recently used PSSAs to identify Tyr 311 and Tyr 332 as sites for Src-dependent PKC␦ phosphorylation (4). Recombinant PKC␦ (human sequence) was subjected to in vitro kinase assays without and with active Src in buffers containing [␥-32 P]ATP and PS/PMA, and radiolabeled PKC␦ was subjected to peptide mapping analysis as an alternative method to map sites for Src-dependent PKC␦ phosphorylation sites (without relying on available PSSAs). Fig. 3A shows that Src increases 32 P incorporation into PKC␦, leading to the appearance of a PKC␦ species that migrates more slowly in SDS-PAGE and displays increased PKC␦-Tyr(P) 311 and PKC␦-Tyr(P) 332 immunoreactivity. Src also increases PKC␦ autophosphorylation at Thr 505 , consistent with our recent observation that the activation loop of PKC␦ is the target of an autophosphorylation reaction that is augmented by Src (4).
PKC␦ purified by SDS-PAGE and blotted to nitrocellulose was excised from the membrane and digested with trypsin, and peptide fragments were separated by reverse phase-HPLC (Fig.  3B). Radioactive peptide fragments detected in assays with PKC␦ ϩ Src (but not in assays with PKC␦ alone) were sequenced by MALDI-TOF mass spectrometry to determine sites for Src-dependent PKC␦ phosphorylation. Fig. 3B shows that three distinct radioactive peaks were detected in reverse phase-HPLC chromatograms of phospho-peptide fragments derived from in vitro kinase assays with PKC␦, Src, and [ 32 P]ATP (and not in vitro kinase assays with PKC␦ alone). Peaks 1 and 2 were identified by mass spectrometry analysis as the oxidized and reduced forms of a Tyr(P) 332 -containing fragment (note that differences in oxidation status arise during sample preparation and should not be construed as evidence that this modification occurs in cells). Peak 3 contained a Tyr(P) 311 fragment, as well as nonphosphorylated forms of peptides containing Tyr 64 and Tyr 52 (other sites on PKC␦ that have been reported to be phosphorylated under certain stimulatory conditions). Collectively, these results identify Tyr 311 and Tyr 332 as the major sites for in vitro Src-dependent PKC␦ tyrosine phosphorylation.
Having identified Tyr 311 and Tyr 332 as major targets for Srcdependent PKC␦ phosphorylation in vitro and H 2 O 2 -dependent PKC␦ phosphorylation in vivo, we performed in vitro kinase assays with PKC␦, active Src, a kinase buffer containing ; cell extracts were subjected to immunoblotting for PKC␦-Tyr(P) 311 and Src-Tyr(P) 416 (with Src protein as a loading control, top right), and PKC␦ pull-downs were immunoblotted for PKC␦-Tyr(P) 332 and Tyr(P) (with PKC␦ as a loading control; the anti-PKC␦-Tyr(P) 332 antibody displays too much nonspecific immunoreactivity to be used directly on cell extracts, middle right). PKC␦ in immune complexes also was used in kinase assays with cTn complex as substrate; a representative autoradiogram of cTnI phosphorylation is illustrated (bottom right). IP, immunoprecipitation.
[ 32 P]ATP and cTn complex (consisting of wild type (WT) cTnT, WT-cTnC, and WT-cTnI) to determine whether Srcdependent PKC␦ tyrosine phosphorylation leads to a change in PKC␦-dependent phosphorylation of cTnI. PKC␦ phosphorylates cTnI in a lipid-dependent fashion in vitro, with cTnI phosphorylation markedly increased in assays performed in the presence of active Src (under conditions that lead to PKC␦ tyrosine phosphorylation; Fig. 3C). The Src-dependent increment in cTnI phosphorylation cannot be attributed to direct myofibrillar protein phosphorylation by Src, because 1) cTnI phosphorylation is completely blocked by GF109203X and 2) anti-Tyr(P) antibodies detect tyrosine-phosphorylated forms of PKC␦ and Src, but no anti-Tyr(P) immunoreactivity comigrating with cTnI was detected. These results indicate that the Srcdependent increment in cTnI phosphorylation results from serine/threonine phosphorylation by PKC␦ rather than direct cTnI tyrosine phosphorylation by Src.
Conclusions derived from PhosphorImager studies were validated by immunoblot analysis (Fig. 4B, bottom panel). An anti-cTnI-Ser(P) 23 /Ser(P) 24 PSSA that specifically recognizes the Ser 23 /Ser 24 -phosphorylated form of cTnI detects similar PKC␦dependent phosphorylation of WT-cTnI and cTnI mutants harboring single residue substitutions at Ser 43 /Ser 45 and Thr 144 (alone and in combination); this PSSA does not detect phosphorylation of the cTnI-S23D/S24D mutant. Importantly, radiolabeling studies show that Src increases overall WT-cTnI phosphorylation by PKC␦, but direct immunoblotting studies show that Src does not increase PKC␦-dependent cTnI-Ser 23 / Ser 24 phosphorylation (compare Fig. 4B, top and bottom panels). We performed additional studies with an anti-pTXR phospho-motif antibody (Cell Signaling Technology; that recognizes phospho-threonines flanked by an arginine in the ϩ2 position, which is predicted to recognize cTnI phosphorylation at KRPT 144 LR, but not RS 23 S 24 AN or KIS 43 AS 45 RK) to localize the Src-dependent increment in PKC␦-dependent cTnI phosphorylation. The anti-pTXR PSSA does not recognize cTnI phosphorylation in assays with PKC␦ alone (without Src), where PKC␦ acts primarily as a cTnI-Ser 23 /Ser 24 kinase. Rather, this antibody selectively recognizes phosphorylation of WT-cTnI, cTnI-S23D/S24D, and cTnI-S43E/S45E (i.e. forms of cTnI that can be phosphorylated at Thr 144 ) in assays performed with PKC␦ ϩ Src (Fig. 4B, bottom panel); the cTnI-S43E/S45E/T144E and cTnI-T144E mutants are not detected. These immunoblotting studies validate the use of the anti-pTXR PSSA to selectively track cTnI-Thr 144 phosphorylation and validate conclusions derived from radiolabeling experiments tracking overall cTnI phosphorylation (by Phosphor-Imager); both techniques identify an effect of Src to selectively increase PKC␦ cTnI-Thr 144 kinase activity. Additional studies show that PKC␦ functions as a cTnI-Thr 144 kinase in assays with Src having either cTnI-S23D/S24D or cTnI-S23/24A as substrate (Fig. 4C). These results argue that Thr 144 phosphorylation does not require prior phosphorylation (or negative charge) at Ser 23 /Ser 24 (i.e. cTnI phosphorylation does not appear to be a hierarchical process). Collectively, these results provide surprising evidence that PKC␦ acquires a new activity, directed toward a different site on cTnI (namely Thr 144 ), when it is tyrosine-phosphorylated by Src; Src does not increase PKC␦dependent cTnI-Ser 23 /Ser 24 phosphorylation.
In vitro kinase assays were performed in buffers containing PS/PMA, which is included both to activate PKC␦ (in assays without Src) and to induce a conformational change required for Src-dependent PKC␦ tyrosine phosphorylation (in assays with Src); our previous studies established that Src phosphorylates PKC␦ only in assays performed with PS/PMA (4). Given recent evidence that the PKC␦ C1 domain is comprised of C1A and C1B domains with markedly different affinities for PMA and the natural lipid cofactor DAG (i.e. PMA binds with high affinity to the PKC␦-C1B domain, whereas DAG anchors full-length PKC␦ to membranes via an interaction with the PKC␦-C1A domain (18)), it was important to determine whether the Src-dependent changes in PKC␦dependent cTnI phosphorylation also are detected in assays with DAG. Fig. 4D shows that DAG effectively substitutes for PMA in these experiments. PS/DAG supports PKC␦-dependent cTnI-Ser 23 / Ser 24 phosphorylation, in assays with PKC␦ alone and mimics the effect of PMA to render PKC␦ a substrate for Src, leading to PKC␦-Tyr 311 /Tyr 332 phosphorylation and PKC␦-dependent cTnI-Thr 144 phosphorylation in assays with Src.
cTnI is a substrate for several physiologically and pathophysiologically relevant serine/threonine kinases that control contractile function, including other PKC isoforms, PKA, and PKD. We performed in vitro kinase assays with PKA, PKD, PKC␣, PKC␤II, PKC␦, and PKC⑀ (under conditions calibrated to achieve similar levels of autophosphorylation) and used immunoblotting with anti-cTnI-Ser 23 /Ser 24 and anti-TXR (cTnI-Thr 144 ) PSSAs to determine whether Src regulates the cTnI kinase activity of these other enzymes. All of these enzymes increase 32 P incorporation into cTnI, tracked by PhosphorImager analysis; immunoblotting studies show that PKC␦, PKC␣, PKC␤II, PKA, and PKD are all effective cTnI-Ser 23 /Ser 24 kinases (Fig. 5). In contrast, PKC⑀ promotes cTnI phosphorylation (detected by PhosphorImager), but PKC⑀ is a relatively poor cTnI-Ser 23 /Ser 24 kinase. Rather, PKC⑀ and PKC␤ are weak cTnI-Thr 144 kinases; PKC␣, PKD, PKA, and allosterically activated PKC␦ (in assays without Src) do not phosphorylate cTnI at Thr 144 . Of note, PKC⑀-and PKC␤II-dependent increases in cTnI-Thr 144 phosphorylation are modest in magnitude compared with the robust increase in cTnI-Thr 144 phosphorylation induced by Src-phosphorylated PKC␦. Moreover, the effect of Src to convert PKC␦ into a cTnI-Thr 144 kinase is quite specific. Although Src also tyrosine phosphorylates PKC⑀ and PKD (but not PKC␣ or PKC␤II), Src has no effect on (or actually decreases, in the case of PKC␣) the cTnI kinase activities of these other enzymes (measured by Phosphor-Imager analysis or by immunoblotting with the anti-cTnI-pSer 23 /Ser 24 PSSA). Src does not convert any of these other enzymes into effective cTnI-Thr 144 kinases. Finally, because PKD also is a known substrate for c-Abl (19), c-Abl-dependent modulation of PKD activity also was considered. Fig. 5B shows that PKD undergoes an autophosphorylation reaction; 32 P incorporation into PKD is increased slightly by Src (which pro-motes PKD tyrosine phosphorylation) and to an even greater extent by c-Abl (although the very high nonspecific anti-Tyr(P) immunoreactivity in assays with c-Abl alone thwarted efforts to directly track c-Abl-dependent PKD tyrosine phosphorylation). PKD does not acquire cTnI-Thr 144 kinase activity when tyrosine-phosphorylated by Src or c-Abl.
Src-dependent Changes in PKC␦ Substrate Specificity Require Tyr 311 and Thr 505 -We used a mutagenesis approach to explore the structural basis for Src-dependent changes in PKC␦ substrate specificity. Fig. 6 shows that Tyr 3 Phe substitutions at Tyr 311 or Tyr 332 (that selectively abrogate Src-dependent PKC␦ phosphorylation at the cognate tyrosine residues) have no effect on PKC␦-dependent cTnI-Ser 23 /Ser 24 phosphorylation. However, the Tyr 3 Phe substitution at Tyr 311 completely abrogates the Srcdependent increment in cTnI phosphorylation by PKC␦ (detected by PhosphorImager) and cTnI-Thr 144 phosphorylation (detected by immunoblot analysis). In contrast, a Tyr 3 Phe substitution at Tyr 332 does not interfere with these activities.
We used a similar mutagenesis approach to examine the role of activation loop phosphorylation. Although recombinant PKC␦ (which displays at most trace Thr 505 phosphorylation prior to in vitro kinase assays) undergoes pronounced Thr 505 autophosphorylation during in vitro kinase assays (particularly in the presence of Src), WT-PKC␦, PKC␦-Y311F, and PKC␦-Y332F enzymes recovered from COS-7 cells exhibit similar levels of basal Thr 505 phosphorylation that do not increase during in vitro kinase assays (with or without Src; compare Figs. 3A and 6). Basal PKC␦-Thr 505 phosphorylation (for WT and Tyr 3 Phe substituted enzymes heterologously overexpressed in COS-7 cells) is presumed to be attributable to a phosphoinositide-dependent kinase-1dependent mechanism that mediates activation loop phosphorylation in trans during de novo enzyme synthesis (and is not  COS cells were transfected with plasmids that drive expression of WT and Y311F, Y332F, T505A, and T505E substituted forms of PKC␦ fused to GFP. PKC␦ was immunoprecipitated with anti-GFP, subjected to immunoblotting with anti-GFP to validate equal protein recovery (C), and subjected to immunocomplex kinase assays without and with lipid cofactor, Src, or c-Abl as indicated. cTnI phosphorylation was quantified by Phosphor-Imager (and is expressed relative to the basal level of phosphorylation for each PKC␦ construct (A). PKC␦-Tyr 311 , -Tyr 332 , and -Thr 505 phosphorylation and cTnI-Ser 23 /Ser 24 and Thr 144 phosphorylation were detected by immunoblot analysis according to the legend for B. The results were replicated in two separate experiments. D, a schematic that marks the various phosphorylation sites of PKC␦ and the main kinases implicated in phosphorylating each site.
prevented by Y311F or Y332F mutations). T505A or T505E substitutions prevent activation loop phosphorylation, but they do not interfere with Src-dependent PKC␦-Tyr 311 (or Tyr 332 ) phosphorylation; T505A or T505E substitutions also do not alter overall cTnI phosphorylation (detected by 32 P incorporation into cTnI) or cTnI-Ser 23 /Ser 24 phosphorylation (detected by immunoblot analysis with the anti-cTnI-pSer 23 /Ser 24 PSSA) in assays without Src. However, T505A and T505E substitutions abrogate the Src-dependent increment in PKC␦-depend-ent overall cTnI phosphorylation (detected by Phosphor-Imager) and the Src-dependent acquisition of cTnI-Thr 144 kinase activity (detected with the anti-TXR PSSA). These results indicate that PKC␦ is catalytically active without Thr 505 phosphorylation, but that Thr 505 phosphorylation is necessary (and that a phospho-mimetic substitution is not sufficient) for Src-dependent regulation of PKC␦ substrate specificity. Collectively, these results indicate that Tyr 311 and Thr 505 cooperate in the Src-activated mechanism that alters the substrate specificity of PKC␦. The conclusion that the substrate specificity of PKC␦ is influenced by a phosphorylation event at Tyr 311 , but not Tyr 332 , is supported further by studies with c-Abl, which selectively increases PKC␦ phosphorylation at Tyr 311 (not Tyr 332 ) and supports PKC␦dependent cTnI-Thr 144 phosphorylation (Fig. 6).
PKC␦ Regulates Cardiac Contractile Function in a Src-dependent Manner-We performed mechanical studies on skinned (detergent extracted) ventricular myocytes (20) to determine whether Src (and tyrosine phosphorylation) alters PKC␦-dependent regulation of contraction. PKC␦ alone reduces submaximal Ca 2ϩ -activated isometric tension (force/ cross-sectional area of the myocyte at pCa 5.5), without changing maximal Ca 2ϩ -activated tension (at pCa 4.5) in skinned myocytes (Table 1 and Fig. 7). However, PKC␦ does not alter cross-bridge cycling kinetics as reflected in the rate constant of force redevelopment (k tr ) following a rapid release-restretch maneuver at either submaximum or maximum Ca 2ϩ concentration. Control studies show that lipids (PS/PMA) and Src alone do not significantly alter mechanical function. These results are expected from previous data characterizing cTnI-Ser 23 /Ser 24 phosphorylation as a modification that reduces submaximal tension but does not alter maximal Ca 2ϩ -activated tension or k tr (21,22). The functional response to treatment with Src-phosphorylated PKC␦ is quite different. PKC␦ ϩ Src lowers tension and the rate of force redevelopment at maximal but not submaximal Ca 2ϩ -activated tension in skinned myocytes (Table 1 and Fig. 7). These results are in agreement with previous studies linking cTnI-Thr 144 phosphorylation to a decrease in maximum Ca 2ϩ -activated tension (12,23). The observation that Src-phosphorylated PKC␦ does not mimic the effect of PKC␦ alone to lower submaximal Ca 2ϩactivated isometric tension suggests that a Ser 23 /Ser 24 phosphorylation- The traces show the recovery of force following a mechanical detachment of cross-bridges, to drop force to near-zero values. A, top panel, representative records used for the determination of k tr rates before (black) and after (red) PKC␦ treatment demonstrate no change in rate of force recovery. Histograms (inset) summarize lack of effect of PKC␦ on maximum Ca 2ϩ -activated parameters. A, bottom panel, force tracings demonstrate that Src (tyrosine-phosphorylated) PKC␦ (blue) decreases the rate of recovery of force (i.e. return to equilibrium more slowly) and therefore k tr . Histograms of maximum Ca 2ϩactivated tension and k tr values demonstrate significant decreases in these parameters in myocytes treated with Src (tyrosine-phosphorylated) PKC␦. B, top panel, tracings show no difference between control (black) and PKC␦ (red) on the rate of force recovery at submaximal Ca 2ϩ , although submaximal tension production was significantly low (histogram). B, bottom panel, force tracings reflect a minor (although nearly significant p ϭ 0.07) shift in force recovery with PKC␦ pretreated with Src kinase (blue). In contrast to PKC␦ alone, no significant change in submaximal tension production was observed (histogram, bottom). All of the measurements were performed in the presence of lipid activators PS/PMA. The results are reported as the means Ϯ S.E. (* indicates p Ͻ 0.05).

TABLE 1 Summary of mechanical properties of detergent extracted (skinned) rat ventricular myocytes before and after kinase treatments
The data are presented as the means Ϯ S.E. Maximum tension and the rate constant of force redevelopment (k tr ) were measured at pCa 4.5 in myocytes before and after treatment with PKC␦ alone (n ϭ 4) or with PKC␦ in the presence of Src kinase (n ϭ 5). Submaximal tension and k tr were measured at pCa 5.5. induced effect is attenuated when cTnI is simultaneously phosphorylated at Ser 23 /Ser 24 and Thr 144 (by Src-phosphorylated PKC␦). These mechanical studies provide a functional correlate to the in vitro biochemical studies and support the conclusion that when allosterically activated by PS/PMA PKC␦ selectively phosphorylates cTnI at Ser 23 /Ser 24 and leads to a fall in submaximal tension (without altering contractile function at maximal Ca 2ϩ ) and that when it is prephosphorylated by Src PKC␦ acquires an additional activity to phosphorylate cTnI at Thr 144 (which prevents the Ser 23 /Ser 24 phosphorylation-dependent decrease in submaximal Ca 2ϩactivated tension and results in a decrease in force and rate of force redevelopment at maximal Ca 2ϩ ).
Other laboratories, conducting similar mechanical studies, considered analogous changes in tension and k tr (as reported in Table 1) significant. Patel et al. (22) reported that PKA accelerates the rate of force development in murine skinned myocardium expressing ␣or ␤-tropomyosin. They reported small but significant changes in the submaximal rate of force redevelopment after PKA treatments of non-transgenic myocardium with 100% ␣-tropomysin (increased by 15%) and trangenic myocardium with 60% ␤-tropomysin (increased by 18%) myocardium. Korte et al. (28) showed that loaded shortening, power output, and rate of force development (WT ϭ 5.8 Ϯ 0.9 s Ϫ1 versus MyBP-C Ϫ/Ϫ ϭ 7.7 Ϯ 1.7 s Ϫ1 ) are increased with knock-out of cardiac myosin binding protein-C. Moreover similar shifts (to those presented here) in Ca 2ϩ sensitivity of force are closely related to events common in fatal human cardiomyopathies.

DISCUSSION
This study provides novel evidence that cTnI phosphorylation by PKC␦ is controlled by Src and tyrosine phosphorylation. These observations are highly significant for several reasons. First, previous studies have implicated tyrosine phosphorylation as a mechanism that alters the activity of PKC␦, but there is still no consensus as to the precise nature of this regulatory control. Tyrosine phosphorylation has variably been implicated as a mechanism that increases or decreases PKC␦ activity. Although there have been isolated reports that link tyrosine phosphorylation to changes in PKC␦ activity toward only selected substrates, studies of substrate specificity have been confined to assays with peptides or proteins with little-to-no physiologic significance. The results reported herein link a tyrosine phosphorylation-dependent change in PKC␦ activity to a physiologically important event in the cell, namely the phosphorylation of cTnI. Second, this study is the first to map tyrosine phosphorylation sites that control PKC␦ activity. Our experiments implicate Tyr 311 (which is a major target for Srcdependent phosphorylation during both stimulation by PMA and oxidative stress) and Thr 505 (a site that is phosphorylated via an autocatalytic reaction that is increased by Src) in this form of regulatory control. In contrast, Tyr 332 (which also is phosphorylated by Src during oxidative stress but is not a substrate for c-Abl and is not phosphorylated during stimulation with PMA) does not mediate Src-dependent changes in PKC␦ substrate specificity. Third, an effect of PKC␦-Tyr 311 /Thr 505 phosphorylation to regulate the phosphorylation of only selected sites on a single substrate is unprecedented in the literature. Post-translational modifications typically induce more general changes in enzyme activity measured as a change in maximal catalytic activity (K cat ) or affinity for substrate or ATP (K m ). Although a recent study using a proteomic approach provided intriguing hints that activation loop phosphorylation can influence PKC␦ substrate specificity (24), a post-translational modification that selectively regulates the phosphorylation of selected sites on a single substrate has not previously been identified.
The role of PKA-dependent cTnI phosphorylation at Ser 23 / Ser 24 in the control of cardiac dynamics is well documented (10). Although PKC phosphorylation of cTnI was originally mapped to Ser 43 /Ser 45 and Thr 144 , recent studies also implicate Ser 23 /Ser 24 as an alternate (and perhaps more important) phosphorylation site for certain PKC isoforms (and PKC-activated effectors). Our results also show that PKC␣, PKC␤, PKC␦, and PKD act in a similar manner to phosphorylate cTnI at Ser 23 / Ser 24 ; only PKC⑀ is uniquely identified as a poor cTnI-Ser 23 / Ser 24 kinase. We also link PKC␦-dependent cTnI-Ser 23 /Ser 24 phosphorylation to a decrease in force development at submaximal Ca 2ϩ , but no change in maximal Ca 2ϩ -activated tension or cross-bridge cycling kinetics in skinned cardiomyocytes (the predicted response based upon studies examining the regulatory actions of PKA, the prototypical cTnI-Ser 23 /Ser 24 kinase).
The data summarized in Table 1 show that at submaximal Ca 2ϩ concentrations Ser 23 /Ser 24 phosphorylation leads to an 83% fall in tension (from 19.7 Ϯ 1.2 to 3.3 Ϯ 0.3 kN/m 2 ). This significant change in tension is in agreement with previous studies detailing the effect of PKA-dependent phosphorylation of cTnI-Ser 23 /Ser 24 on cardiac muscle. The Ser 23 /Ser 24 phosphorylation effect on tension (3.3 Ϯ 0.3 kN/m 2 ) is largely reversed under oxidative stress conditions through additional phosphorylation of Thr 144 (10.6 Ϯ 3.1 kN/m 2 ). This 68% increase in tension indicates improved sarcomeric Ca 2ϩ responsiveness.
Allosterically activated PKC␦ is not a cTnI-Thr 144 kinase. In fact, whereas cTnI-Thr 144 phosphorylation originally was attributed to PKCs, the physiologically relevant cTnI-Thr 144 kinase (and the functional consequences of cTnI-Thr 144 phosphorylation) remains uncertain. cTnI-Thr 144 phosphorylation mechanisms have largely been inferred from studies with T144A or T144E-substituted cTnI mutants, because a PSSA that directly tracks cTnI-Thr 144 phosphorylation (similar to the cTnI-pSer 23 /Ser 24 PSSA) is not available. The identification of the anti-pTXR phospho-motif antibody as a reagent that can be used to track cTnI-Thr 144 phosphorylation in immunoblotting experiments allowed a comparison of cTnI-Thr 144 phosphorylation by individual PKC isoforms that was not previously possible. Our studies identify a low level of cTnI-Thr 144 phosphorylation by PKC␤ and PKC⑀ (consistent with recent studies that attribute cTnI-Thr 144 phosphorylation to PKC-␤II (14)) but no cTnI-Thr 144 phosphorylation by related kinases such as PKC␣, PKA, PKD, and allosterically activated PKC␦ (i.e. the form of PKC␦ that is anchored to membranes by lipid cofactors). In contrast, tyrosine-phosphorylated PKC␦ is a robust cTnI-Thr 144 kinase. These results suggest that tyrosine-phosphoryl-ated PKC␦ is the most physiologically relevant cTnI-Thr 144 kinase and that cTnI-Thr 144 phosphorylation would be most prominent in vivo in conditions such as oxidative stress (where activation of Src leads to PKC␦ tyrosine phosphorylation). Current concepts regarding the functional consequences of cTnI-Thr 144 phosphorylation are based largely on studies that use mutagenesis approaches or genetic models, with evidence that cTnI-Thr 144 phosphorylation accelerates myofilament relaxation in adult rat myocytes, decreases Ca 2ϩ sensitivity in sliding filament assays, or sensitizes myofilaments to Ca 2ϩ (without changing maximum tension) in skinned myocytes from transgenic mice that express the cTnI-Ser 23 /Ser 24 A mutant (12)(13)(14). Importantly, the studies reported herein examine the physiologic consequences of a coordinate increase in cTnI phosphorylation at Ser 23 /Ser 24 and Thr 144 induced by Src-phosphorylated PKC␦. Here, the Ca 2ϩ -desensitizing effect of PKC␦ alone is no longer detected. Rather, PKC␦ ϩ Src lowers tension and the rate of force redevelopment at maximal Ca 2ϩ , an effect that is predicted to contribute to oxidative stress-induced myocardial dysfunction. On the other hand, at submaximal Ca 2ϩ concentrations, where the myocardium typically operates, oxidative stress events that activate Src-PKC␦ signaling pathways could enhance cardiac function. This effect on one hand might be beneficial in maintaining systolic contractile function of the failing myocardium. On the other hand this might be detrimental during ␤-adrenergic stimulation hindering the expected increase of relaxation rate. These data underscore the importance of understanding events that modulate troponin I phosphorylation and ultimately fine-tune sarcomeric function.
Recent NMR data suggest that cTnI phosphorylation at Ser 23 /Ser 24 weakens the interaction of the cTnI N-terminal residues 1-30 with cTnC, induces a lever-like bending around residues 33-42, and facilitates its repositioning to bind the inhibitory region of cTnI (9). This association between N-terminal and inhibitory regions of cTnI is stabilized by favorable electrostatic interactions between an acidic patch consisting of residues Asp 3 , Glu 4 , Asp 7 , Glu 11 , and a basic patch Arg 142 , Arg 146 , and Arg 149 , surrounding Thr 144 . Interestingly, the R146G mutation (linked to familial hypertrophic cardiomyopathy in humans) leads to myofilaments with increased Ca 2ϩ sensitivity and lack of responsiveness to phosphorylation at Ser 23 /Ser 24 (25,26). In the case of the R146G mutant, replacement of an Arg with Gly presumably induces its effect by disrupting the electrostatic interactions between the acidic N terminus and the inhibitory region of cTnI. In our case, the addition of a bulky, negatively charged phosphate group to the basic patch, through Thr 144 phosphorylation, could induce a similar effect. Our mechanical data are consistent with this model, because Thr 144 phosphorylation alleviates the Ca 2ϩ -desensitizing effect of Ser 23 /Ser 24 phosphorylation. Our data further suggest that phosphorylation at Thr 144 (on a Ser 23 /Ser 24 phosphorylation background) prevents the cTnI inhibitory region from properly interacting with actin-tropomyosin, leading to a decrease in maximum force.
Collectively, our studies identify distinct signaling modes for PKC␦, a single PKC isoform whose substrate specificity can be dynamically regulated through tyrosine phosphorylation. This study focuses on a dual role for PKC␦ as both an allosterically activated and a tyrosine-phosphorylated enzyme leading to the phosphorylation of functionally distinct sites on cTnI, potentially underlying differences in PKC-dependent regulation of myofilament contraction in the normal heart and in the context of diseases associated with oxidative stress. This novel tyrosine phosphorylation-dependent change in PKC␦ substrate specificity is likely to represent a general regulatory mechanism that also influences PKC␦ phosphorylation of other cellular substrates and might be targeted in the future for therapeutic advantage.