Protein kinase C-δ regulates the subcellular localization of Shc in H₂O₂-treated cardiomyocytes

Abstract

Protein kinase C-δ (PKCδ) exerts important cardiac actions as a lipid-regulated kinase. There is limited evidence that PKCδ also might exert an additional kinase-independent action as a regulator of the subcellular compartmentalization of binding partners such as Shc (Src homologous and collagen), a family of adapter proteins that play key roles in growth regulation and oxidative stress responses. This study shows that native PKCδ forms complexes with endogenous Shc proteins in H₂O₂-treated cardiomyocytes; H₂O₂ treatment also leads to the accumulation of PKCδ and Shc in a detergent-insoluble cytoskeletal fraction and in mitochondria. H₂O₂-dependent recruitment of Shc isoforms to cytoskeletal and mitochondrial fractions is amplified by wild-type-PKCδ overexpression, consistent with the notion that PKCδ acts as a signal-regulated scaffold to anchor Shc in specific subcellular compartments. However, overexpression studies with kinase-dead (KD)-PKCδ-K376R (an ATP-binding mutant of PKCδ that lacks catalytic activity) are less informative, since KD-PKCδ-K376R aberrantly localizes as a constitutively tyrosine-phosphorylated enzyme to detergent-insoluble and mitochondrial fractions of resting cardiomyocytes; relatively little KD-PKCδ-K376R remains in the cytosolic fraction. The aberrant localization and tyrosine phosphorylation patterns for KD-PKCδ-K376R do not phenocopy the properties of native PKCδ, even in cells chronically treated with GF109203X to inhibit PKCδ activity. Hence, while KD-PKCδ-K376R overexpression increases Shc localization to the detergent-insoluble and mitochondrial fractions, the significance of these results is uncertain. Our studies suggest that experiments using KD-PKCδ-K376R overexpression as a strategy to competitively inhibit the kinase-dependent actions of native PKCδ or to expose the kinase-independent scaffolding functions of PKCδ should be interpreted with caution.

protein kinase C-δ (PKCδ) is a member of a multigene family of related serine/threonine kinases that play vital roles in signaling events that influence cardiac contraction, ischemic preconditioning, and the pathogenesis of cardiac hypertrophy and failure (30). PKCδ is structurally characterized by a COOH-terminal catalytic domain and an NH₂-terminal regulatory domain consisting of tandem C1A-C1B motifs that bind lipid cofactors and a C2 domain that functions as a protein-protein interaction module (2). Early studies characterized PKCδ as allosterically activated enzyme, focusing on the role of lipid cofactors such as diacylglycerol or the tumor-promoting phorbol ester phorbol 12-myristate 13-acetate (PMA) to anchor the enzyme in an active conformation at membranes (in close proximity to target substrates). However, recent studies indicate that PKCδ also is phosphorylated on tyrosine residues in cells exposed to certain mitogens or proapoptotic stimuli (such as oxidative stress) and that tyrosine phosphorylation provides a mechanism to regulate PKCδ's enzymology and certain PKCδ-dependent cellular responses (29, 30). While PKCδ contains at least nine sites for independently regulated phosphorylations, most studies have focused on phosphorylation at Y³¹¹ or Y³³², residues in the V3 hinge region of the enzyme (numbering based on rodent sequence). We previously demonstrated that oxidative stress leads to a coordinate increase in PKCδ phosphorylation at Y³¹¹ and Y³³², whereas PMA selectively increases PKCδ phosphorylation at Y³¹¹ (without an associated increase in PKCδ-Y³³² phosphorylation) in cardiomyocytes (26). Progress toward identifying the functional consequences of agonist-induced changes in PKCδ tyrosine phosphorylation is recent, with evidence that Y³¹¹ phosphorylation is required for in vitro PKCδ activity toward selected target substrates (31) and certain in vivo PKCδ-dependent responses (13, 19). The functional consequences of PKCδ-Y³³² phosphorylation remain more uncertain, since mutagenesis studies failed to expose a role for Y³³² phosphorylation in the in vitro regulation of full-length PKCδ activity (31). Rather, there is limited evidence that Y³³² phosphorylation generates a docking site for the Src homology 2 (SH2) domain of Shc (Src homologous and collagen) (14), a family of three adapter proteins that link cell surface receptors to mitogenic responses [in the case of p46Shc and p52Shc (24)] or sensitize cells to oxidative stress-dependent apoptosis [in the case of p66Shc (3, 22)]. While these results suggest that tyrosine-phosphorylated PKCδ might exert an entirely different role as a signal-regulated scaffold, kinase-independent actions for PKCδ are seldom considered.

PKCδ is activated or upregulated in numerous models of cardiac ischemia and cardiac hypertrophy, and it has been implicated in the pathogenesis of reperfusion injury and adverse cardiac remodeling. However, the molecular basis for PKCδ's cardiac actions remains uncertain, since PKCδ activation has been linked to a diverse array of cellular responses, including the activation of SAPKs, activation of the PKD pathway that regulates transcription factors (including HDAC5 and cAMP response element-binding protein) and controls gene expression, phosphorylation of sarcomeric proteins that regulate cardiac contraction, cytoskeletal events that influence cardiac structure and mechanical performance, and mitochondrial events that regulate reactive oxygen species (ROS) production and the induction of apoptosis (1, 6, 12, 17, 21, 32). While Shc adapter proteins colocalize with PKCδ at surface membranes, the cytoskeleton, and in mitochondria—and there is considerable evidence that Shc and PKCδ regulate similar signaling pathways that trigger cytoskeletal remodeling, cellular growth, mitochondrial ROS generation, and ROS-dependent apoptosis—the notion that PKCδ and Shc might cooperate to regulate signaling events at specific subcellular microdomains has not previously been considered. This study examines whether Shc proteins are PKCδ-binding partners and/or PKCδ-regulated effectors in cardiomyocytes.

MATERIALS AND METHODS

Materials.

Antibodies were from the following sources: PKCδ-pY³¹¹, Shc (Cell Signaling Technology); PKCδ-pY³³², SOS1 (Santa Cruz Biotechnology); Src (Oncogene); prohibitin (Research Diagnostics); caveolin-3, β-integrin, vimentin, desmin, RACK1 (BD Biosciences); heat shock protein (HSP) 60, αB-crystallin (Stressgen); GM130 (Covance); actin (Sigma). PMA was from Sigma (St. Louis, MO). All other chemicals were reagent grade.

Cardiomyocyte culture and transfection.

All protocols and procedures used in this study were reviewed and approved by the Institutional Animal Care and Use Committee of Columbia University. Cardiomyocytes were isolated from the hearts of 2-day-old Wistar rats by a trypsin dispersion procedure using a differential attachment procedure to enrich for cardiomyocytes followed by irradiation as described previously (28). The yield of cardiomyocytes typically is 2.5–3 × 10⁶ cells per neonatal ventricle. Cells were plated on protamine sulfate-coated culture dishes at a density of 5 × 10⁶ cells/100-mm dish. Experiments were performed on cultures grown for 5 days in MEM (Gibco BRL) supplemented with 10% fetal calf serum and then serum-deprived for the subsequent 24 h. PKCδ overexpression was accomplished as described previously (28). Briefly, cardiomyocytes were exposed for 2 h to adenoviral vectors that contained nearly the full-length Ad5 genome (lacking regions E1 and E3) and a cytomegalovirus promoter driving expression of wild-type (WT) mouse PKCδ, kinase-dead (KD) mouse PKCδ (harboring a single ATP-binding site K376R substitution), or β-galactosidase (β-Gal) as control. Our previous studies showed that this method leads to overexpression of exogenous PKCδ at levels five to seven times higher than the endogenous PKCδ enzyme in cardiomyocytes (28). Protein extracts (or subcellular fractions) were prepared 48 h after transfections (following acute treatments with vehicle, PMA, or H₂O₂ as indicated).

Preparation of heavy and light membranes and detergent-insoluble cytoskeletal fractions.

Detergent-soluble and detergent-insoluble fractions were separated by washing neonatal cardiomyocyte cultures twice with cold phosphate-buffered saline (PBS) and then scraping the cells into lysis buffer containing Tris-Cl (10 mM, pH 8), NaCl (150 mM), aprotinin (50 μg/ml), leupeptin (48 μg/ml), benzamidine (50 μg/ml), PMSF (2 mM) pepstatin A (5 μM), sodium vanadate (0.1 mM), NaF (50 mM), Triton X-100 (1%), and β-octylglucoside (60 mM). The lysis buffer composition (which includes both Triton X-100 and β-octylglucoside as detergents) was designed to isolate a detergent-insoluble cytoskeletal fraction that is not contaminated with caveolae/lipid raft membranes; caveolae/lipid raft membranes are solubilized by β-octylglucoside, but not Triton X-100. Scraped cells were sonicated and centrifuged at 4°C for 15 min at maximal speed in a microcentrifuge. The supernatant was saved and used as the detergent-soluble fraction. The pellet was solubilized in SDS sample buffer, sonicated, and boiled for 5 min; the supernatant obtained following a second centrifugation in the microcentrifuge (for 5 min at maximal speed) was used as the detergent-insoluble fraction.

Subcellular fractions also were separated by differential centrifugation, essentially according to methods described by Ventura et al. (33). Briefly, neonatal cardiomyocyte cultures were rinsed in cold PBS and scraped into lysis buffer containing Tris-MOPS (10 mM pH 7.4), EGTA (1 mM), sucrose (250 mM), aprotinin (10 μg/ml), leupeptin (10 μg/ml), PMSF (250 μM), sodium vanadate (1 mM), NaF (100 mM), and sodium pyrophosphate (10 mM). Scraped cells were mechanically disrupted by 10 passages through a cylinder cell homogenizer and then centrifuged at 800 g for 5 min to pellet nuclei and unlysed cells. This initial supernatant was then centrifuged at 8,000 g for 10 min to pellet a mitochondria-enriched heavy membrane fraction, which was solubilized in SDS-sample buffer. The supernatant was removed and recentrifuged at 77,000 g for 1 h to pellet a plasma membrane fraction; the final supernatant was saved as the cytosolic fraction.

Immunoblotting studies.

Immunoblotting was performed on cell extracts or subcellular fraction according to manufacturer's instructions or methods described previously (27). In each figure, each panel represents results from a single gel (exposed for a uniform duration); in some figures, dividing lines or white spaces denote rearrangement of data from different regions of a single gel (to eliminate extraneous data that was included in the original experiment but is tangential to the issues examined in this study). Specifics are noted in individual figure legends. Detection was with enhanced chemiluminescence, and quantification was by laser-scanning densitometry. All results were replicated in at least three experiments.

Statistical analysis.

All values are expressed as means ± SE. Comparisons between groups were performed using one- or two-way analysis of variance, with a Tukey test for multiple comparisons when appropriate; significance was defined at P < 0.05.

RESULTS

PKCδ interacts with Shc in H₂O₂-treated cardiomyocytes.

A previous study concluded that PKCδ interacts with the SH2 domain of p52Shc via a Y³³² phosphorylation-dependent mechanism based on experiments that tracked complex formation by heterologously overexpressed proteins in a reductionist in vitro model (14). Studies to date have not examined whether PKCδ-Y³³² phosphorylation nucleates complexes between the native PKCδ enzyme and endogenous Shc proteins in a more physiologically relevant setting. Therefore, we used immunoprecipitation methods to examine whether stimuli that increase PKCδ tyrosine phosphorylation promote PKCδ-Shc complex formation in cardiomyocytes.

We previously reported that PKCδ displays little-to-no basal tyrosine phosphorylation in cultured neonatal cardiomyocytes (26). We showed that PKCδ phosphorylation at Y³¹¹ and Y³³² increases markedly when cardiomyocytes are treated with relatively high H₂O₂ concentrations (1–5 mM); lower H₂O₂ concentrations activate various growth regulatory signaling pathways but do not increase PKCδ tyrosine phosphorylation [even when the incubation interval is prolonged to 24 h (26)]. Figure 1A shows that PKCδ is constitutively recovered at low levels in Shc pull-downs from resting cardiomyocytes. Shc-PKCδ complex formation increases when cardiomyocytes are treated with 1–5 mM H₂O₂; lower H₂O₂ concentrations (that do not promote PKCδ tyrosine phosphorylation) do not increase PKCδ-Shc coprecipitation. Of note, control studies show that H₂O₂ does not alter PKCδ or Shc protein expression in cardiomyocytes and that PKCδ-Shc interactions are specific; PKCδ is not recovered in immune complexes prepared with an irrelevant IgG (data not shown).

Fig. 1.

Protein kinase C-δ (PKCδ) coprecipitates with Shc (Src homologous and collagen) in H₂O₂-treated cardiomyocytes. Cardiomyocytes were treated for 15 min with vehicle, H₂O₂ (5 mM, unless indicated otherwise), or PMA (300 nM). Stimulation followed a 45-min pretreatment with vehicle, GF109203X (GFX; 5 μM), or 4-amino-5-(4-methylphenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP1; 10 μM) in B and C. Extracts were immunoprecipitated with anti-Shc (A and B) or anti-PKCδ (C), and equal amounts of protein were subjected to immunoblotting for PKCδ and Shc (to verify equal protein recovery/loading and to track PKCδ-Shc complex formation) as well as other antibodies indicated in individual panels. A white space in B denotes where data from two regions of a single gel were merged for purposes of presentation. In C, NS indicates a nonspecific immnunoreactive band that migrates between p52Shc and p66Shc. All results were replicated in at least 4 independent experiments. Results for H₂O₂-dependent PKCδ-Shc complex formation are quantified in A (n = 4; *P < 0.05). SOS1, Son of Sevenless 1.

Figure 1B shows that Shc pull-downs from H₂O₂-treated cardiomyocytes contain the Y³³²-phosphorylated form of PKCδ; PKCδ in Shc pull-downs from H₂O₂-treated cardiomyocytes also is phosphorylated at Y³¹¹ (data not shown). Figure 1B also shows that PKCδ-Shc complex formation does not increase when cardiomyocytes are treated with PMA, a stimulus that increases PKCδ phosphorylation at Y³¹¹, but not Y³³² (26). While these results suggest that the PKCδ-Shc interaction might require PKCδ-Y³³² phosphorylation, additional studies with phosphatase 1 (PP1; a pharmacological inhibitor of Src kinases that prevents H₂O₂-dependent PKCδ tyrosine phosphorylation) indicate otherwise; the H₂O₂-dependent increase in Shc-PKCδ complex formation persists in cardiomyocytes treated with H₂O₂ + PP1 (Fig. 1B). PKCδ-Shc complex formation also does not require PKCδ activity, since H₂O₂-dependent Shc-PKCδ complex formation also persists in cardiomyocytes pretreated with GF109203X (an inhibitor of PKC activity; Fig. 1B).

We performed the reciprocal experiment (i.e., PKCδ immunoprecipitation followed by Shc immunoblotting) to validate the specificity of the PKCδ-Shc interaction and to determine which Shc isoforms complex with PKCδ in H₂O₂-treated cardiomyocytes. This was important, since the previous study identified a PKCδ interaction with p52Shc; PKCδ interactions with p66Shc or p46Shc have not been identified. Figure 1C shows that PKCδ binding to p46Shc, p52Shc, and p66Shc increases in cardiomyocytes treated with H₂O₂, but not PMA. This result suggests that PKCδ interacts with a domain common to all three Shc isoforms. Additional studies showed that Son of Sevenless 1 (SOS1, a guanine nucleotide exchange factor for Ras and Rac that constitutively complexes with Grb2) also is recovered in PKCδ pull-downs from cardiomyocytes treated with H₂O₂ (but not PMA or vehicle control). Since Grb2 docks through its SH2 domain to tyrosine-phosphorylated Shc, this result suggests that Shc acts as a bridge to link the Grb2-SOS1 complex to PKCδ; direct immunoblotting for Grb2 was not possible in Fig. 1, since low-molecular-weight proteins such as Grb2 migrate with the dye front under the SDS-PAGE conditions used for this experiment. Finally, Fig. 1C shows that PKCδ forms multiprotein complexes with Shc isoforms and SOS1 in cardiomyocytes pretreated with PP1 or GF109203X and then stimulated with H₂O₂ (Fig. 1C). These results indicate that the H₂O₂-induced docking interaction between PKCδ and Shc does not require PKCδ tyrosine phosphorylation or PKCδ catalytic activity.

Shc colocalizes with PKCδ in a detergent-insoluble cytoskeletal fraction in H₂O₂-treated cardiomyocytes.

In the course of studies designed to identify PKCδ-binding partners in H₂O₂-treated cardiomyocytes, we noted that a minor pool of PKCδ (amounting to ∼20% of total PKCδ immunoreactivity) accumulates as a tyrosine-phosphorylated enzyme in the detergent-insoluble fraction of cardiomyocytes treated with H₂O₂ (Fig. 2A); this pool of PKCδ, which is not solubilized in extraction buffers containing Triton X-100 and β-octylglucoside, is not recovered in the PKCδ pull-downs. Immunoblot analysis was performed to further characterize the properties of this detergent-insoluble fraction. Figure 2B shows that this detergent-insoluble fraction is highly enriched in intermediate filament (IF) proteins such as desmin and vimentin; αB-crystallin (a heat shock protein that functions as a chaperone for desmin and prevents the formation of protein aggregates) also is detected in the detergent-insoluble fraction (Fig. 2B). This fraction does not contain the mitochondria marker protein GM130, the plasma membrane marker β₁-integrin, or caveolin-3 (a marker for caveolae membranes, specialized signaling microdomains that typically contaminate detergent-insoluble cytoskeletal fractions prepared with Triton X-100, but are solubilized in extraction buffers containing β-octylglucoside).

Fig. 2.

PKCδ and Shc accumulate in a detergent-insoluble cytoskeletal fraction that is enriched in intermediate filament proteins in H₂O₂-treated cardiomyocytes. Cardiomyocytes were treated for 15 min with vehicle, the indicated concentration of H₂O₂, 1 μM norepinephrine (NE), or 100 nM EGF. Stimulation followed a 45-min pretreatment with vehicle or PP1 (10 μM) in C. Extracts were partitioned into detergent-soluble and detergent-insoluble fractions and subjected to immunoblotting with the indicated antibodies according to protocols described in materials and methods. Vehicle-treated samples were used for immunoblotting experiments in B. The white spaces in A and C show where data from two regions of a single gel were merged for purposes of presentation. Similar results were obtained in 3 separate experiments.

Figure 2A shows that the smaller p52Shc and p46Shc isoforms and trace amounts of PKCδ constitutively localize to the detergent-insoluble fraction in resting cardiomyocytes. PKCδ and all three Shc isoforms translocate to this detergent-insoluble fraction when cardiomyocytes are treated with 1–5 mM H₂O₂. Other agonists that promote PKCδ and/or Shc phosphorylation (such as norepinephrine, endothelin-1, EGF, or PMA) do not drive PKCδ or Shc to the detergent-insoluble fraction (Fig. 2A and data not shown). Additional studies show that H₂O₂-dependent PKCδ localization to the detergent-insoluble fraction does not require PKCδ tyrosine phosphorylation (since it is not inhibited by PP1; Fig. 2C) or PKCδ activity (since it is not prevented by GF109203X; data not shown). The effect of H₂O₂ to drive Shc isoforms to the detergent-insoluble fraction also persists in cardiomyocytes pretreated with PP1 or GF109203X (Fig. 2C and data not shown).

We used an adenoviral-mediated overexpression strategy to determine whether heterologously overexpressed WT-PKCδ or KD-PKCδ accumulate in—and recruit Shc to—the detergent-insoluble fraction. Figure 3A shows that WT-PKCδ partitions between the detergent-soluble and detergent-insoluble fractions, without detectable Y³¹¹ or Y³³² phosphorylation, in resting cardiomyocytes. WT-PKCδ overexpression (at levels ∼5-fold higher than the endogenous PKCδ enzyme) leads to a modest increase in p46Shc and p52Shc in the detergent-insoluble fraction (46 ± 3% and 66 ± 7%, respectively; P < 0.05; n = 8). H₂O₂ treatment increases WT-PKCδ-Y³¹¹ and Y³³² phosphorylation in both soluble and detergent-insoluble fractions, without changing WT-PKCδ partitioning between soluble and detergent-insoluble fractions. However, the H₂O₂-dependent increase in Shc localization to the detergent-insoluble fraction is amplified in cardiomyocytes that overexpress WT-PKCδ. p46Shc/p52Shc levels (combined for the analysis) are 48 ± 8% higher in the detergent-insoluble fraction of H₂O₂-treated Ad-WT-PKCδ cultures than in H₂O₂-treated Ad-β-Gal cultures (P < 0.05; n = 8). The effect of WT-PKCδ overexpression on p66Shc localization is even more striking; detergent-insoluble p66Shc levels are 5.3 ± 0.9-fold higher in H₂O₂-treated Ad-WT-PKCδ cultures than in H₂O₂-treated Ad-β-Gal cultures (P < 0.05; n = 8).

Fig. 3.

Wild-type (WT)-PKCδ and kinase-dead (KD)-PKCδ localize in, and recruit Shc to, the detergent-insoluble fraction. Adenoviral-mediated gene transfer was used to overexpress WT-PKCδ, KD-PKCδ, or β-galactosidase (β-Gal) as control (multiplicities of infection: 100 plaque-forming units/cell). A: forty-eight hours postinfection, cells were challenged for 15 min with vehicle, 300 nM PMA, or 5 mM H₂O₂ and then partitioned into detergent-soluble and detergent-insoluble fractions. WT-PKCδ and KD-PKCδ overexpression does not lead to any gross change in protein partitioning between detergent-soluble and detergent-insoluble fractions. B: immunoblotting on whole cell lysates or subcellular fractions was performed as described in materials and methods. The dashed lines in B denote where data from different regions of a single gel were merged for purposes of presentation. Similar results were obtained in 5 separate experiments.

KD-PKCδ overexpression is widely used as a strategy to identify PKCδ-specific cellular functions; this experimental approach is based on the assumption that the KD-PKCδ transgene colocalizes with (and acts as a competitive-inhibitor of) native PKCδ at its sites of action within the cell. However, Fig. 3A shows that the subcellular localization and tyrosine phosphorylation patterns of KD-PKCδ and native PKCδ (or WT-PKCδ) are quite different. KD-PKCδ partitions as a constitutively Y³¹¹- and Y³³²-phosphorylated protein to the detergent-insoluble fraction. Only ∼30% of KD-PKCδ remains (with little-to-no tyrosine phosphorylation) in soluble fraction; this pool of KD-PKCδ is inducibly phosphorylated at Y³¹¹ and Y³³² when cardiomyocytes are treated with H₂O₂. KD-PKCδ overexpression leads to an increase in basal p46Shc, p52Shc, and p66Shc partitioning to the detergent-insoluble fraction. There is no further increase in the levels of KD-PKCδ or Shc isoforms in the detergent-insoluble fraction when cardiomyocytes are treated with H₂O₂. Of note, while these experiments were performed in cells that heterologously overexpressed KD-PKCδ at levels ∼5-fold higher than endogenous PKCδ, similar results were obtained at lower levels of KD-PKCδ overexpression (data not shown).

Control experiments exposed an additional effect of WT-PKCδ or KD-PKCδ overexpression to increase in desmin expression (3.1 ± 0.6-fold for WT-PKCδ and 3.3 ± 0.4-fold for KD-PKCδ; P < 0.05, n = 4; Fig. 3B). WT-PKCδ and KD-PKCδ overexpression also increases αB-crystallin, Src, and RACK1 (receptor for activated C-kinase 1) partitioning to the detergent-insoluble fraction, without changing αB-crystallin, Src, or RACK1 protein expression [not significant (NS); n = 4; Fig. 3B]. The levels and subcellular partitioning of other cytoskeletal proteins such as actin and vimentin are not influenced by WT-PKCδ or KD-PKCδ PKCδ overexpression (Fig. 3B and data not shown).

The observation that a catalytically inactive form of PKCδ localizes as a tyrosine phosphorylated enzyme to the detergent-insoluble fraction when it is heterologously overexpressed in cardiomyocytes raises the question of whether endogenous PKCδ is tyrosine phosphorylated and recruited to the detergent-insoluble fraction when cardiomyocytes are treated with GF109203X (under conditions that inhibit PKC activity). Figure 4A shows that chronic treatment with GF109203X leads to a progressive decline in PKCδ protein abundance. However, GF109203X treatment does not increase PKCδ tyrosine phosphorylation or promote PKCδ accumulation in the detergent-insoluble fraction. Figure 4B shows that GF109203X slows PMA-dependent PKCδ downregulation, but GF109203X treatment does not lead to the accumulation of the allosterically activated form of PKCδ in the detergent-insoluble fraction.

Fig. 4.

The properties of KD-PKCδ (a constitutively tyrosine-phosphorylated enzyme that localizes to the detergent-insoluble fraction) do not phenocopy native PKCδ lacking catalytic activity. A: adenoviral-mediated gene transfer was used to overexpress WT-PKCδ or β-galactosidase as control. Cultures were treated with vehicle, GF109203X (GFX; 5 μM for 24 or 48 h), or H₂O₂ (5 mM for 15 min); stimulations were staggered, so that all stimulations were terminated 96 h postinfection. B: cultures were treated with vehicle, 300 nM PMA, or 300 nM PMA plus 5 μM GF109203X for 24 h and were then subjected to an acute 15-min challenge with vehicle or 5 mM H₂O₂. Immunoblotting was performed on whole cell lysates or subcellular fractions with antibodies that recognize PKCδ protein and PKCδ-Y³¹¹ phosphorylation according to protocols described in materials and methods. The results were replicated in 2 separate experiments.

Shc colocalizes with PKCδ in the mitochondrial fraction of H₂O₂-treated cardiomyocytes.

We examined whether mitochondria are an additional signaling compartment for native PKCδ and Shc in H₂O₂-treated cardiomyocytes. We used differential centrifugation to isolate a heavy membrane fraction that is enriched in mitochondrial marker proteins (GM130, prohibitin-1, and HSP60) and excludes plasma and caveolae membrane markers (caveolin-3 and integrin-β₁; Fig. 5); this fraction also excludes the IF protein desmin (data not shown). Figure 5 shows that PKCδ and Shc constitutively localize to this fraction. H₂O₂ treatment leads to a modest increase in PKCδ localization to the mitochondrial fraction (68 ± 12%; P < 0.05; n = 4); PKCδ is recovered as a Y³¹¹-phosphorylated enzyme in the mitochondrial fraction (Fig. 5). H₂O₂ treatment also leads to a 59 ± 8% increase in p46Shc/p52Shc and a 4.2 ± .6-fold increase in p66Shc in the heavy membrane (P < 0.05; n = 6).

Fig. 5.

PKCδ and Shc accumulate in the mitochondrial fractions of H₂O₂-treated cardiomyocytes. Cardiomyocytes were treated for 15 min with vehicle or 5 mM H₂O₂, partitioned into cytosolic, plasma membrane, and mitochondrial fractions, and then subjected to immunoblot analysis with the indicated antibodies according to protocols described in materials and methods. The white space in one panel (depicting immunoblotting for GM130) denotes merging of two regions of a single gel for presentation purposes. Similar results were obtained in 4 separate experiments.

Heterologously overexpressed WT-PKCδ enzyme partitions between the cytosol and mitochondrial fractions, without detectable Y³¹¹ phosphorylation, in resting cardiomyocytes (Fig. 6). H₂O₂ treatment leads to an increase in WT-PKCδ-Y³¹¹ phosphorylation; the Y³¹¹-phosphorylated form of WT-PKCδ is detected in both cytosolic and mitochondrial fractions (Fig. 6). WT-PKCδ overexpression increases p46Shc and p52Shc partitioning to the mitochondrial fraction in resting cardiomyocytes; WT-PKCδ overexpression also amplifies the H₂O₂-dependent translocation of all three Shc isoforms to the mitochondrial fraction.

Fig. 6.

WT-PKCδ and KD-PKCδ overexpression enhances Shc localization to the mitochondrial fraction. Adenoviral-mediated gene transfer was used to overexpress WT-PKCδ, KD-PKCδ, or β-galactosidase. Forty-eight hours postinfection, cells were challenged for 15 min with vehicle or 5 mM H₂O₂ and were then subjected to extraction and differential centrifugation to separate cytosolic and mitochondrial fractions. Immunoblotting with the indicated antibodies was performed as described in materials and methods. The data are representative of results in 4 separate experiments.

The compartmentalization pattern of KD-PKCδ is quite different. KD-PKCδ constitutively partitions as a Y³¹¹-phosphorylated enzyme to the mitochondrial fraction of resting cardiomyocytes; relatively little KD-PKCδ is recovered in the cytosolic fraction (Fig. 6). KD-PKCδ overexpression leads to a dramatic (5- to 12-fold) increase in Shc isoform partitioning to the mitochondrial fraction (n = 6). KD-PKCδ or Shc levels in the mitochondrial fraction are not increased further by H₂O₂.

DISCUSSION

PKCδ is generally viewed as a lipid-regulated kinase that phosphorylates effector substrates in diacylglycerol-enriched membranes. The notion that PKCδ might exert additional cellular actions via a kinase-independent mechanism is not generally considered, despite early evidence that a catalytically inactive form of PKCδ induces apoptosis when it is overexpressed in PMA-treated A7r5 rat aortic smooth muscle cells (8) and studies linking other PKC isoforms to kinase-independent actions (5, 18, 23, 34). This study shows that the native PKCδ enzyme interacts with Shc isoforms via a kinase-independent mechanism in H₂O₂-treated cardiomyocytes and that heterologously overexpressed PKCδ (both WT and KD transgenes) localize and recruit Shc to cytoskeletal and mitochondrial fractions. The focus on a PKCδ-dependent mechanism that regulates the subcellular compartmentation of Shc is quite reasonable and appealing, given literature linking PKCδ and Shc to many similar effector responses that influence cytoskeletal remodeling, cell migration, cell growth, and stress-induced apoptosis (15). For example, the cytoskeletal actions of PKCδ generally have been attributed to phosphorylation of actin-associated proteins such as vinculin, paxillin, and focal adhesion kinase. However, our studies suggest that PKCδ may exert an additional action to promote cytoskeletal remodeling by recruiting p46Shc and/or p52Shc (adapter protein that play important roles in the control of actin fiber assembly and actomyosin-based cell contractility). Similarly, PKCδ has been implicated in mitochondrial events that increase ROS generation and enhance susceptibility to ROS-dependent apoptosis, but the specific binding partners and/or substrates for PKCδ in mitochondria remain uncertain. We recently showed that PKCδ is required for the agonist-dependent increase in p66Shc-S³⁶ phosphorylation in cardiomyocytes (9); this modification is required for p66Shc's mitochondrial proapoptotic actions. Studies reported herein extend the analysis by showing that PKCδ also recruits p66Shc to the mitochondrial fraction, where the local actions of p66Shc to increase ROS generation and promote apoptosis would contribute to cardiovascular oxidative stress responses.

A previous study by Leitges et al. (14), showing that PKCδ-Y³³² phosphorylation generates a docking site for the SH2 domain of Shc, provided the rationale to examine whether PKCδ and Shc interact via a phospho-Y³³²-dependent mechanism in cardiomyocytes. Our immunoprecipitation studies show that native Shc isoforms form complexes with endogenous Y³³²-phosphorylated PKCδ in H₂O₂-treated cardiomyocytes. However, the observation that PKCδ-Shc complex formation is detected in cardiomyocytes treated with H₂O₂ in the presence of PP1 (which prevents PKCδ tyrosine phosphorylation) effectively excludes a Shc interaction that requires PKCδ-Y³³²-phosphorylation. The discrepancy between our findings and results obtained previously by Leitges et al. could suggest that the Y³³²-phosphorylated hinge region functions as a docking site for Shc only in certain experimental settings. In particular, the previous study showed that a glutathione S-transferase-Shc-SH2 domain fusion protein engages in an in vitro interaction with tyrosine phosphorylated WT-PKCδ, but not PKCδ-Y332F (14). These previous results, obtained in a reductionist in vitro model, do not necessarily implicate the Y³³²-phosphorylated hinge region of PKCδ as the major interaction surface for full-length Shc in vivo in a more physiological context. Our studies in cardiomyocytes suggest that other points of contact between PKCδ and full-length Shc proteins (or indirect interactions between PKCδ and Shc, due to the formation of multiprotein complexes) may assume greater functional importance in the in vivo context.

Our studies show that H₂O₂ treatment leads to PKCδ localization in the cytoskeletal fraction and that H₂O₂ drives PKCδ to the cytoskeletal fraction via a mechanism that does not require catalytic activity (since this response is not inhibited by GF109203X). Previous studies by Hahn et al. (10) showing that full-length PKCδ and the δV1 peptide fragment (a sequence corresponding to the putative δ-RACK binding domain in the regulatory domain of PKCδ) decorate similar cytoskeletal structures suggest that structural determinants in the catalytic domain also are not required. Studies reported herein also show that WT-PKCδ or KD-PKCδ overexpression leads to increased Shc localization in the detergent-insoluble cytoskeletal fraction. These results suggest that PKCδ might exert a direct (kinase-independent) role as a signal-regulated scaffold to recruit (or “hijack”) Shc to the cytoskeletal compartment. However, an alternative indirect mechanism also is possible, given the evidence that WT-PKCδ or KD-PKCδ overexpression leads to marked changes in the molecular composition of the cytoskeleton (with increases in the levels of desmin, αB-crystallin, Src, and RACK1 in the detergent-insoluble fraction). Further studies to determine whether Shc is recruited to the detergent-insoluble fraction as a result of a docking interaction with a component of the remodeled cytoskeleton (rather than a direct interaction with the PKCδ transgene) are warranted.

The observation that WT-PKCδ or KD-PKCδ overexpression results in increased desmin and αB-crystallin expression in neonatal cardiomyocyte cultures agrees with previous studies showing that desmin and αB-crystallin expression also is increased in the hearts of δV1 transgenic mice (10). However, the functional consequences of PKCδ-dependent changes in desmin and αB-crystallin expression remain uncertain. While Hahn et al. (10) previously showed that these proteins accumulate in intracellular proteinaceous aggregates that disrupt the normal subcellular architecture in δV1 mice, the further observation that mouse models of desmin and αB-crystallin overexpression display no obvious structural or functional abnormalities has been interpreted as evidence that the heart failure phenotype that develops in δV1 mice is due to chronic inhibition of PKCδ activity [rather than PKCδ-dependent changes in desmin or αB-crystallin expression (10)]. Of note, this conclusion is based on the assumption that the δV1 peptide acts exclusively as a competitive inhibitor of native full-length PKCδ. An alternative role for the δV1 peptide to regulate IF networks and cytoskeletal structures, in a manner that mimics a kinase-independent action of the regulatory domain PKCδ [analogous to the cellular actions of regulatory domain sequence of PKCε, which binds cytoskeletal proteins, regulates cytoskeletal architecture, and is required for neurite outgrowth during neuronal differentiation (23, 34)] is possible but has never been considered.

Finally, this study exposes a limitation of studies that use KD-PKCδ-K376R overexpression as a strategy to interrogate PKCδ's cardiac actions. The assumption inherent in this experimental approach is that catalytically inactive forms of PKC colocalize with (and act as a competitive inhibitors of) the cognate native PKC enzymes at their sites of action in the cell. However, even a casual review of the literature shows that many kinase-inactive PKCs aggregate and accumulate in detergent-insoluble/cytoskeletal fractions (5, 7). Studies reported herein show that KD-PKCδ-K376R aberrantly localizes as a tyrosine phosphorylated enzyme to the particulate and detergent-insoluble fractions of cardiomyocytes. Importantly, the phosphorylation and compartmentalization patterns described for the PKCδ-K376R transgene do not phenocopy the features of native PKCδ lacking catalytic activity; endogenous PKCδ does not accumulate in the detergent-insoluble/cytoskeletal fraction of cardiomyocytes chronically treated with a pharmacologic inhibitor of PKC activity. It is worth noting that our findings pertaining to the PKCδ-K376R mutant are not novel; the Pierce laboratory originally characterized the PKCδ-K376R-ATP-binding mutant as a constitutively tyrosine-phosphorylated/membrane-associated enzyme (16). These properties of the PKCδ-K376R mutant were somewhat inexplicably ignored in subsequent studies. For other PKC-ATP-binding mutants, aberrant localization to the cytoskeletal detergent-insoluble fraction is more widely recognized and generally attributed to defects in autophosphorylation at the activation loop and COOH terminus, i.e., priming autophosphorylations that stabilize the phosphatase-/protease-resistant mature conformation of the enzyme. Our previous observation that the KD-PKCδ-K376R mutant exhibits an activation loop phosphorylation defect is consistent with this formulation (25). However, the notion that autophosphorylation defects explain the abnormal maturation patterns characteristic of many catalytically inactive forms of PKC has been difficult to reconcile with literature showing that the activation loop and COOH-terminal priming sites on PKC are targets for phosphorylation in trans by other cellular enzymes. The Parker laboratory has recently resolved this dilemma by showing that ATP-binding pocket mutants lack priming phosphorylations because they adopt an open/inactive conformation; these constructs are primed (in trans, by other cellular enzymes) when cells are treated with inhibitor compounds that occupy the ATP binding pocket and stabilize the closed/active conformation (4). Additional elegant studies showing that kinase-inactive forms of PKC that retain wild-type ATP binding (for example, PKCs rendered catalytically inactive by substitutions at the catalytic aspartate that participates in the transphosphorylation reaction, but does not make any side chain contacts with ATP) are fully primed and properly folded provide further evidence that maturational phosphorylations on PKC require occupancy of the nucleotide binding pocket [and not autocatalytic activity (4)]. Recent studies from two other laboratories provide further evidence that PKCs (and related AGC kinases) use ATP as a conformational regulator, not just a phosphodonor (11, 20). Our findings, interpreted in the context of this recent literature, raise major concerns that results obtained with the KD-PKCδ-K376R mutant reflect the cellular actions of an improperly folded and aberrantly phosphorylated/localized protein, and not necessarily the kinase-independent scaffolding actions of the mature native enzyme. Future overexpression studies with catalytically inactive forms of PKCδ that are properly primed and folded should provide a more reliable strategy to examine the kinase-independent functions of PKCδ.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grants HL-77860 and HL-93343.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.