Molecular Analysis of the Interactions between Protein Kinase C-ε and Filamentous Actin*

Protein kinase C-epsilon (PKC-ε) contains a putative actin binding motif that is unique to this individual member of the PKC gene family. We have used deletion mutagenesis to determine whether this hexapeptide motif is required for the physical association of PKC-ε and actin. Full-length recombinant PKC-ε, but not PKC-βII, -δ, -η, or -ζ, bound to filamentous actin in a phorbol ester-dependent manner. Deletion of PKC-ε amino acids 222–230, encompassing a putative actin binding motif, completely abrogated this binding activity. When NIH 3T3 cells overexpressing either PKC-ε or the deletion mutant of this isozyme were treated with phorbol ester only wild-type PKC-ε colocalized with actin in zones of cell adhesion. In binary reactions, it was possible to demonstrate that purified filamentous actin is capable of directly stimulating PKC-ε phosphotransferase activity. These and other findings support the hypothesis that a conformationally hidden actin binding motif in the PKC-ε sequence becomes exposed upon activation of this isozyme and functions as a dominant localization signal in NIH 3T3 fibroblasts. This protein-protein interaction is sufficient to maintain PKC-ε in a catalytically active conformation.

PKC-⑀ is a typical multidomain protein in which the overall structural organization has been conserved in orthologous genes from yeast to mammals (2,22). However, in mammals PKC-⑀ has acquired short sequence motifs in the regulatory N-terminal region that are not evident in invertebrates (AplII of Aplysia and PKC d98F of Drosophila; Ref. 15) and are postulated to function as localization signals in the subcellular targeting of this protein kinase. These putative targeting signals include peptide motifs reported to be capable of anchoring PKC-⑀ within the cytoskeletal matrix (10,11,13) and domains which appear to support interactions with either the Golgi or plasma membranes of mammalian cells (16,17). This diversification may potentially recruit PKC-⑀ into distinct, and spatially segregated, multimeric signaling complexes in a speciesand cell-specific manner and contribute to the striking functional versatility of this isozyme.
Mammalian PKC-is most closely related to PKC-⑀ in overall sequence identity and similarity (66 and 82% in human Swiss-Prot numbers P24723 and Q02156, respectively) and yet each displays a distinct pattern of distribution within the same cell, with PKC-being concentrated in the Golgi apparatus and PKC-⑀ accumulating at areas of cell-cell contact in NIH 3T3 fibroblasts (18). For this reason, relatively small sequence motifs or signal patches may be the primary determinants of the disparity in their topogenic fate. Alignment of the amino acid sequences for human PKC-⑀ and PKC-reveals that the principal difference between the two polypeptides can be attributed to an insertion of 56 additional residues within the hinge region of PKC-⑀. The PKC-⑀ hinge region contains two consensus peptide motifs, a PEST sequence (19) and a sequence that is quite similar to the "destruction box" of mitotic cyclins (RLGL-DEFNF, residues 402-410 in human PKC-⑀), that are known to target cytosolic proteins for ubiquitin-dependent destruction by the 26 S proteasome (19,21). In addition, in vivo sorting assays conducted using truncated fragments of PKC-⑀ suggest that upon activation the hinge region may be involved in targeting the protein to the plasma membrane (16). The N-and C-terminal variable extensions of PKC-⑀ and -show a significant degree of sequence identity (53 and 57%, respectively) and only the N terminus contains stretches of three or more residues in which nonconservative substitutions are encountered (residues 28 -36 and 139 -154 in human PKC-⑀). The only remaining stretch of amino acids in which sequence divergence between PKC-⑀ and -is apparent can be found within the conserved C1 domain, where a putative actin-binding motif that is unique to PKC-⑀ has recently been identified (13). We * This work was supported by Grants DK45718 and ES8397 from the National Institutes of Health (to M. K. W. and D. M. T.), respectively. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  have now investigated the functional importance of this putative actin-binding site by deletion mutagenesis and expression of the mutagenized PKC-⑀ cDNA in the Spodoptera frugiperda (Sf9) insect cell line and cultured NIH 3T3 fibroblasts.
Binary interactions between purified actin and a variety of recombinant PKCs were performed under in vitro conditions designed to provide direct tests of our hypotheses that filamentous actin (A F ): 1) represents a novel and specific PKC-⑀-binding protein that, 2) recognizes and physically interacts with the previously identified actin-binding motif in PKC-⑀, and 3) is capable of maintaining this kinase in a catalytically active conformation (13). These analyses revealed that interactions between purified A F and full-length recombinant PKC-⑀ were quite specific, because PKC-␤ II , -␦, -, and -did not bind A F at a physiological ionic strength. Deletion of amino acid residues 222-230 in mouse PKC-⑀ completely abrogated this binding activity without altering the affinity of the deletion mutant (⌬PKC-⑀ 222-230 ) for 4␤-phorbol 12,13-dibutyrate (4␤-PDBu). Moreover, immunofluorescence microscopy of PKC overexpressing NIH 3T3 cells affirmed that PKC-⑀, but not ⌬PKC-⑀ 222-230 , colocalized with actin. Finally, assays of in vitro phosphotransferase activity indicated that A F may also function as an agonist of PKC-⑀, in the absence of lipid metabolites or membrane phospholipids.
Actin Preparations-Rabbit skeletal muscle ␣-actin was purified as described (22) and polymerized to generate A F by dialysis against 4 mM imidazole, pH 7.5, containing 2 mM MgCl 2 and 0.5 mM ATP. Monomeric actin (A G ) was generated by dialyzing A F for 48 h against G-buffer (5 mM Tris-HCl, pH 7.6, containing 0.2 mM CaCl 2 , 0.2 mM ATP, and 0.5 mM ␤-mercaptoethanol). Cys 374 of actin was labeled with the dye AEDANS as described in Ref. 23 and a stable and soluble derivative of A G was generated using the heterobifunctional reagent 3-maleimidobenzoic acid N-hydroxysuccinimide ester (MBS) according to Bettache et al. (24). The two C-terminal residues of actin (Cys 374 and Phe 375 ) were selectively removed by digestion of AEDANS-labeled A G with bovine pancreatic type I trypsin according to the procedure described in Ref. 25.
Subcellular Fractionation and Generation of Tryptic PKC-⑀ Fragments-Cytosolic fractions were obtained from rat cerebral cortex (13). This cytosolic fraction contained an average of 7 ng of PKC-⑀/g of protein.
Cytosolic fractions containing tryptic PKC-⑀ fragments were generated by predigesting cytosol with trypsin at a protein to enzyme mass ratio of 40:1 for 5 min at room temperature before adding soybean trypsin inhibitor at four times the trypsin concentration.
Generation of Overexpressing Cell Lines-NIH 3T3 fibroblasts were purchased from the American Type Culture Collection (Rockville, MD) and infected with pLXSN recombinant retrovirus or pLXSN harboring the genes for p3/PKC-⑀ or p5/⌬PKC-⑀ 222-230 . Plasmid pLXSN constructs were transfected into the amphotrophic packaging line PA317 using Lipofectin and selecting with 400 g/ml G418. The titer of the resulting retrovirus was amplified by sequential passage between the 2 and PA317 packaging cell lines before infection of NIH 3T3 cells as described in Ref. 27. Stably expressing cell lines were selected and subcloned by limiting dilution (28) in 500 g/ml G418 to yield 16 subclones derived from single cells. The PKC-⑀-and ⌬PKC-⑀ 222-230 overexpressing NIH 3T3 cells were screened for PKC protein expression by Western blot analyses and representative clones were used for immunofluorescence studies.
Immunofluorescence Microscopy-Indirect immunofluorescence localization was performed as described in Ref. 29. Where specified, vector controls and PKC-⑀ overexpressing NIH 3T3 clones were treated with 100 nM phorbol 12-myristate 13-acetate (PMA) for 10 min. Cells were rinsed twice with phosphate-buffered saline, fixed with 4% paraformaldehyde in phosphate-buffered saline for 30 min, washed twice with phosphate-buffered saline containing 0.1 M glycine, and permeabilized with blocking buffer (phosphate-buffered saline containing 1% bovine serum albumin, and 2% normal goat serum, and 0.4% saponin) for 5 min. Cells were incubated with either anti-PKC-⑀ (diluted 1:500 for vector transfected cells or 1:2,000 for PKC-overexpressors) or anti-mannosidase II (diluted 1:500) in blocking buffer for 1 h, washed five times, and then incubated with an appropriate secondary antibody (fluorescein isothiocyanate-conjugated anti-rabbit IgG when costained with Texas red-conjugated phalloidin or Texas red-conjugated antimouse IgG when costained with anti-mannosidase II monoclonal antibody) in blocking buffer for 2 h.
Actin Binding-Actin binding assays were performed in A F buffer (4 mM Tris-HCl, pH 7.5, containing 10 M CaCl 2 , 30 mM KCl, 120 mM potassium proprionate, 1 mM MgCl 2 , 150 M ATP, and 100 mM ZnCl 2 ) as described previously (13). Standard curves for PKC immunostaining were obtained for each gel and used to estimate the percentage of A F -dependent cosedimentation ((A F ϩ PKC)-(PKC)). Where indicated MBS-actin was added to A F at the specified mass ratios. Samples were incubated for 15 min at room temperature in the presence of either 500 nM 4␤-PDBu or an equimolar concentration of the inactive phorbol ester, 4␣-PDBu, unless otherwise specified. Phospholipids were not included in any of the assay mixtures in order to rule out the possibility of phospholipid-mediated protein-protein interactions. A F pellets were analyzed by SDS-PAGE and immunoblotting. Densitometric analysis of the autoradiograms was performed using ImageQuant 3.3 (Molecular Dynamics, Inc.).
[ 3 H]PDBu Binding-Binding assays were performed using polyethylene glycol precipitation of recombinant 6xHis-tagged PKC-⑀ or ⌬PKC-⑀ 222-230 (25 ng) and increasing concentrations of [ 3 H]PDBu (0.1-25 nM) in the absence or presence of 100 g/ml phosphatidylserine, as described (30). Bovine ␥-globulin (4 mg/ml) was included in the assay mixture as a carrier protein. Specific binding represents the difference between the precipitated [ 3 H]PDBu in the absence (total binding) and presence (nonspecific binding) of 10 M unlabeled PDBu and the concentration of free [ 3 H]PDBu was measured for each tube.
Trypsin Sensitivity-Binding assays were performed using a modification of the method described previously (31). Equal amounts of cytosolic and actin-bound PKC-⑀ were resuspended in 15 l of A F buffer with 500 nM 4␣-PDBu or 4␤-PDBu. Samples were treated with a specified concentration of trypsin for 5 min at room temperature and proteolysis was quenched by addition of soybean trypsin inhibitor followed by SDS-PAGE sample buffer (10 mM Tris-HCl, pH 6.8, containing 9% SDS, 15% glycerol, 2% ␤-mercaptoethanol, 0.05% bromphenol blue). Western blot analysis and densitometric scanning of the intact PKC-⑀ bands was used to determine the extent of degradation.
Phosphotransferase Activity-Purified recombinant PKC-␤ II or -⑀ (20 ng) was added to 20 mM HEPES buffer, pH 7.4, containing 150 mM KCl, 0.1 mM EGTA, 10 mM MgCl 2 , and 200 g/ml of the specified substrate. Samples were preincubated for 10 min in the presence of 4␣-PDBu or 4␤-PDBu before the addition of 100 M ATP containing 1 Ci of [␥-32 P]ATP (10 Ci/mmol) and incubation for an additional 5 min at 30°C. Samples were blotted onto pieces of P81 ion-exchange paper, washed three times in 100 ml of 30% (v/v) acetic acid, and counted using a liquid scintillation spectrometer.
Data Analysis-Values shown are representative of at least three or more experiments, unless otherwise specified, and treatment effects were evaluated using a two-sided Student's t test. Errors are standard errors of the mean (S.E.) of averaged results and values of p Ͻ 0.05 were taken as a significant difference between means.

PKC-⑀ Preferentially Binds A F through Reversible Ionic Interactions-Recombinant
PKC-⑀ cosediments with purified A F under physiological ionic strength conditions, this interaction is both saturable and significantly enhanced by 4␤-PDBu (Ref. 13; Fig. 3C). Addition of an A G derivative, MBS-A G , that maintains its monomeric form under physiological conditions (24) did not inhibit this binary interaction at a unitary MBS-A G :A F mass ratio (Fig. 1A). However, soluble MBS-A G did significantly decrease the in vitro cosedimentation of PKC-⑀ with A F at the nonphysiologically high mass ratio of 3:1 (Fig. 1A), possibly due to relatively low affinity interactions between this enzyme and MBS-A G . Similar competition studies, performed using synthetic peptides and cytosolic PKC-⑀, indicate that hydrophilic forces may contribute to the high affinity interactions between PKC-⑀ and A F (13). To more systematically examine the nature of this protein-protein interaction, A F cosedimentation assays were performed using purified PKC-⑀ in the presence of either Triton X-100 or salt (KCl). While increasing concentrations of Triton X-100 (0.01-1.0%, v/v) had no effect on the amount of PKC-⑀ that cosedimented with purified A F (data not shown), high concentrations of salt completely disrupted this protein-protein interaction (IC 50 ϭ 375 mM, Fig. 1B).
The C-terminal Half of PKC-⑀ (Residues 320 -785) Does Not Cosediment with A F -Limited trypsinization of PKC-⑀ results in the formation of three proteolytic fragments that can be separated by SDS-PAGE (Fig. 2A, lane 1, of left and right  panels). These tryptic cleavage sites have been mapped to the hinge region of PKC-⑀ and shown to yield one N-terminal (1-319) and two C-terminal (320 -785 and 364 -785) fragments of the holoenzyme (32). When these tryptic fragments of PKC-⑀ were coincubated with A F in the presence of 500 nM 4␤-PDBu, only undigested PKC-⑀ and its N-terminal fragment (PKC-⑀ 1-319 ) cosedimented with A F (Fig. 2A, lane 2 of left panel). It was also apparent that both PKC-⑀ and PKC-⑀ 1-319 had become more concentrated in the A F pellets (compare lanes 1 and 2 of Fig. 2A, left panel) and that this interaction was effectively reversed by the addition of a synthetic peptide that is identical to the PKC-⑀ putative actin-binding motif (LKKQET; Fig. 2A, lane 3 of left panel). Neither C-terminal peptide (PKC-⑀ 320 -785 or PKC-⑀ 364 -785 ) could be detected using anti-C-terminal antibodies in the same actin pellets (Fig. 2A,  lane 2 of right panel). Additional A F cosedimentation assays indicated that the synthetic hexapeptide LKKQET competes with intact, cytosolic, PKC-⑀ for A F binding in a relatively specific manner. In these assays, 4␤-PDBu (500 nM) increased the binding of PKC-⑀ to A F by more than 3-fold (Fig. 2B, inset) and increasing concentrations of the synthetic peptide LKKQET completely inhibited this protein-protein interaction (Fig. 2B). Scrambling the amino acid sequence of this synthetic peptide to KQLKTE resulted in a loss of inhibition and a synthetic peptide reported to inhibit the binding of PKC-⑀ to the PKC anchoring protein RACK1 (DIINALCF; see Ref. 33) also failed to interfere with the cosedimentation of PKC-⑀ and A F (Fig. 2B). Yet another synthetic peptide that proved to be ineffective in these competition assays (EAVSLKPT; data not shown) was identical to PKC-⑀ 14 -21 and reportedly inhibits the association of activated PKC-⑀ with cross-striated structures in saponin-permeabilized cardiomyocytes (10).
Identification of an Actin Binding Motif in PKC-⑀-We expressed in Sf9 cells the full-length PKC-␦ and PKC-⑀, as well as a deletion mutant of PKC-⑀ (⌬PKC-⑀ 222-230 ) that lacks a sequence motif previously implicated in the in vitro binding of PKC-⑀ to A F (13). These kinases and A F were purified from Sf9 cell lysates and rabbit skeletal muscle, respectively, to greater than 85% homogeneity (Fig. 3A). Although the nonapeptide deleted from ⌬PKC-⑀ 222-230 is located between the C1A and C1B phorbol ester-binding domains of this protein (34), this mutation did not significantly alter the specific binding of [ 3 H]PDBu to PKC-⑀ in the absence or presence of phosphatidylserine (Fig. 3B). Intact and mutated PKC-⑀ bound to [ 3 H]P- DBu with a K d of 2.3 Ϯ 0.6 nM (n ϭ 3) in the presence of phosphatidylserine (Fig. 3B, left saturation curves). Phosphatidylserine was required for the high affinity binding of [ 3 H]P-DBu (10 nM) to both PKC-⑀ and ⌬PKC-⑀ 222-230 , as the omission of this phospholipid decreased specific [ 3 H]PDBu binding by about 90% (Fig. 3B). The ability of ⌬PKC-⑀ 222-230 to bind A F was measured using cosedimentation assays and compared with that of the full-length recombinant PKC-⑀. Recombinant PKC-␦ was also used as a negative control, since it has high sequence homology with PKC-⑀ (66% similarity and 45% identity) but does not contain the consensus actin-binding motif (13,35). PKC-␦ did not cosediment with A F in the present experiments, in the absence or presence of 4␤-PDBu (Fig. 3C). In contrast, treatment with 4␤-PDBu resulted in more than a 7-fold enhancement of PKC-⑀ binding to A F (Fig. 3C). Interestingly, the PKC inhibitor GF109203X (1 M) had no affect on this protein-protein interaction (data not shown). While ⌬PKC-⑀ 222-230 consistently displayed a weak association with A F (0.063 Ϯ 0.008 PKC bound /PKC total ), this interaction was not enhanced by increasing concentrations of 4␤-PDBu (Fig. 3C). Fig. 3D shows a more extensive comparison of the A F binding properties of PKC-⑀ to those members of the PKC family having the highest degree of overall sequence identity to PKC-⑀ (PKC-␦ and -) or that have previously been reported to associate with A F (PKC-␤ II and -). Only recombinant PKC-⑀ and ⌬PKC-⑀ 222-230 reliably cosedimented with A F under the in vitro conditions used in these experiments (Fig. 3D).
Colocalization of Activated PKC-⑀ and Actin in PKC-⑀ Overexpressing NIH 3T3 Cell Lines-Immunofluorescence microscopy was used to examine the redistribution of activated PKC-⑀ in G418-resistant NIH 3T3 clonal derivatives overexpressing either the intact enzyme or ⌬PKC-⑀ 222-230 . Wild-type NIH 3T3 cells and cells transfected with an empty vector did not reliably stain for PKC-⑀ (not shown), due to a low level of expression for the endogenous isozyme (18). Staining of nonstimulated and PMA-treated PKC-⑀ overexpressing NIH 3T3 cells is shown in Fig. 4, A and C, respectively. The cellular morphology and distribution of PKC-⑀ staining was reminiscent of that reported previously (18). Staining for PKC-⑀ was diffuse throughout the cytoplasm of unstimulated cells, where it showed a punctate pattern that extended from around the nucleus toward the cell periphery and into long cytoplasmic extensions (Fig. 4A, arrow). In these resting cells, there was a convincing superimposition of PKC-⑀ and A F , costained using phalloidin-Texas Red (Fig. 4, compare A and B), within the long cellular processes that were frequently observed. A small portion of PKC-⑀ also appeared to colocalize with a mannosidase II-rich component of the Golgi (data not shown). These observations support the conclusion that PKC-⑀ is capable of binding to both skeletal A F in vitro and nonmuscle A F in cultured fibroblasts. After 10 min in 100 nM PMA, the long cytoplasmic extensions of PKC-⑀ overexpressing cells retracted as the cells flattened out and displayed prominent ruffling of the plasma membrane. We observed a dramatic redistribution of activated PKC-⑀ and A F to the cell margins, typically in areas of cell-cell contact, and dissolution of actin stress fibers (Fig. 4, C and D). In contrast, clones overexpressing ⌬PKC-⑀ 222-230 showed a predominantly perinuclear staining pattern in the absence of PMA that was not associated with A F (Fig. 4, compare E and F). Upon PMA treatment, ⌬PKC-⑀ 222-230 became more diffusely distributed throughout the cell, where it showed a coarse punctate texture that was not associated with A F (Fig. 4, compare G and H). Moreover, membrane ruffling and the dissolution of actin stress fibers was no longer apparent in ⌬PKC-⑀ 222-230 overexpressing fibroblasts, although cortical A F staining did appear to become more intense in the presence of PMA (Fig. 4H).
A F Binding Increases the Proteolytic Sensitivity of PKC-⑀ -We have hypothesized that hydrophilic interactions between A F and the C1 region of PKC-⑀ may anchor this kinase in an active conformation within microfilamentous structures (13). To investigate this possibility, we probed the topology of PKC-⑀ by performing protease sensitivity assays. The hinge region of PKCs becomes markedly more susceptible to trypsinization through a conformational change that accompanies activation (31). Indeed, the susceptibility of cytosolic PKC-⑀ to tryptic cleavage was increased by roughly an order of magnitude in the presence of 500 nM 4␤-PDBu (Fig. 5B). Digestion with increasing concentrations of trypsin resulted in three proteolytic fragments of PKC-⑀ of the predicted molecular masses ( Figs. 2A  and 5A). Here we report that the pattern of PKC-⑀ proteolysis observed in the presence of purified A F was essentially equivalent to that observed in the presence of 4␤-PDBu (Fig. 5B), consistent with an A F -induced exposure of the PKC-⑀ hinge region.
Purified A F Is Sufficient to Stimulate Recombinant PKC-⑀ Phosphotransferase Activity-Although provocative, the equivocal nature of protease accessibility assays makes the use of independent assays obligatory. For this reason, we performed phosphotransferase assays to directly examine the binary interactions between purified A F and recombinant PKC-⑀, in the complete absence of any classical PKC activators. Titration of A F , in the presence of PKC-⑀ (20 ng) and the substrate peptide myelin basic protein fragment 4 -14 (MBP 4 -14 ), showed that A F physically interacts with this isozyme to produce a 4-fold increase in kinase activity (Fig. 6A). Maximal in vitro activity , or 120 g of purified A F (triangle) before adding trypsin at the concentrations indicated (nanograms of trypsin/g of cytosol). Samples were analyzed by 10 -15% SDS-PAGE followed by immunoblotting with anti-PKC-⑀ antiserum and digitization of the Ͼ90 kDa band corresponding to the undigested isozyme. Data are the averages of two independent experiments. occurred at an A F concentration of approximately 20 g/ml (Fig. 6A, EC 50 ϭ 4.7 g/ml). Importantly, the addition of 500 nM 4␤-PDBu to this protein mixture significantly increased the level of maximal kinase activity produced by A F alone (Fig. 6A). This additive effect of 4␤-PDBu on the V max of recombinant PKC-⑀ may reflect the existence of two independent mechanisms for activation and this possibility merits further investigation. In parallel assays, it was confirmed that purified A F was not an effective agonist of recombinant PKC-␤ II when measured using MBP 4 -14 as a substrate in the presence (Fig.  6A) or absence of 4␤-PDBu and Ca 2ϩ (Ref. 36 and data not shown).
Competition assays performed using synthetic peptides identical to PKC-⑀ 223-228 and rVI-RACK1 234 -241 indicated that A F interacts with the former hexapeptide motif to stimulate PKC-⑀ phosphotransferase activity (Fig. 6B). As previously mentioned (10), an octapeptide derived from PKC-⑀ 14 -21 has no effect on A F -induced PKC-⑀ activity and it is shown here that equimolar concentrations of rVI-RACK1 234 -241 (DIINALCF) was also an ineffective antagonist.
The effect of A F -binding on the substrate specificity of PKC-⑀ was examined because chimeric studies have shown that the N-terminal regulatory domain of PKC-⑀ restricts the substrate specificity of this isozyme (37) and A F has been reported to disrupt the ability of PKC␤ II to phosphorylate a variety of substrates (36). In the present study, we measured PKC-⑀ phosphotransferase activity in the absence and presence of A F using two synthetic peptides (␦ peptide and MBP 4 -14 ) and three purified proteins (histone IIIS, calponin, and protamine HCl) as substrates. While A F significantly enhanced the phosphorylation of these substrates by PKC-⑀, it had no effect on the substrate preference of this kinase: ␦ peptide Ͼ Ͼ protamine Ͼ MBP 4 -14 Ͼ histone IIIS Ͼ calponin (Fig. 6C).
Interaction of PKC-⑀ with the C Terminus of Actin-The flexible C terminus of ␣-actin contains the four cleavage sites most accessible to proteolysis (25). Here we have used limited trypsinization to selectively remove the C terminus dipeptide of A G (Cys 374 -Phe 375 ) to examine the importance of the C terminus in interactions with PKC-⑀. The minimal time required for tryptic cleavage of the Lys 373 -Cys 374 bond was established by monitoring the disappearance of AEDANS, a fluorescent stain known to specifically label Cys 374 of A G (Fig. 7A). Equivalent amounts of PKC-⑀ cosedimented with intact and truncated A F (Fig. 7B). In contrast, the C-terminal of A F proved to be essential for the A F -induced stimulation of PKC-⑀ phosphotransferase activity (Fig. 7C). Finally, titrating the ratio of truncated to intact A F gradually led to a complete reversal of PKC-⑀ stimulation by intact A F (Fig. 7D). DISCUSSION A recurring theme in topological biochemistry is that cytosolic protein kinases may be recruited to perform distinct functions based on the localization signals they have acquired and their microenvironment at the time of activation. Thus, spatiotemporal gradients of required cofactors may activate multiple PKC isozymes within a common subcellular domain and further select for the isozyme(s) capable of anchoring to a cognant binding partner. It is presently understood that the resulting segregation of PKC isozymes directs the assembly of a coherent signaling apparatus in which the kinase(s) becomes colocalized with its intended substrates. Here we report two substantive findings that should contribute to continuing advances in our understanding of such isozyme-specific PKC signaling cascades. First, A F recognizes and binds directly to a hexapeptide motif that is unique to the regulatory C1 domain of PKC-⑀. Second, physical interactions between A F and PKC-⑀ are dynamic and capable of stimulating PKC-⑀ kinase activity in the absence of lipid metabolites.
It has previously been demonstrated that the activation of PKC-⑀ exposes an actin binding motif that binds A F and that a hexapeptide derived from the C1 domain of PKC-⑀ competitively inhibits this protein-protein interaction (13). Because several actin-binding proteins have conserved homologous peptide motifs (35), we hypothesized that this cluster of highly charged amino acids in PKC-⑀ may function as an actin-binding site. While putative PKC localization signals have commonly been deduced from such peptide competition studies (10,36,40) or binding assays using fragments of the full-length isozyme (16,17,41), the limitations inherent to each experimental approach argue in favor of circumspection. Here we have used PCR-based mutagenesis to delete the putative actin binding motif in PKC-⑀ in order to more directly examine its functional importance for the in vitro interactions between PKC-⑀ and A F . Our data suggest that the interface between PKC-⑀ and A F was generally electrostatic in nature and that the polymeric form of actin presented the optimal surface for PKC-⑀ binding. A F cosedimentation assays with recombinant PKC-⑀ and ⌬PKC-⑀ 222-230 demonstrated the C1 domain of PKC-⑀ contains an authentic actin binding motif. This A F localization signal is positioned in a flexible stretch of 22 amino acids that separate two compact ␣/␤ structural motifs (C1A and C1B) that function as cellular receptors for diacylglycerol and phorbol esters. Thus, deletion of the actin-binding site may have altered the local topology of this activator binding domain and even long range conformational effects cannot be ruled out at this time. However, we note that the affinity of [ 3 H]PDBu binding to PKC-⑀ was not altered as a result of this deletion and that the remaining 13 amino acids separating the PKC-⑀ C1A and C1B domains was comparable in length to the 15 amino acids naturally found in all conventional PKCs (␣, ␤, and ␥).
While only vertebrate homologues of PKC-⑀ contain the consensus actin binding motif LKX 2 EX, both PKC-␤ II and -have previously been reported to bind actin (36,42). In this respect, the disparity in the amounts of recombinant PKC-␤ II , -⑀, andthat cosedimented with A F , in a physiologically relevant buffering system, was unexpected and striking. It seems likely that the evolution of PKC actin binding motifs may have emerged, in both vertebrates and invertebrates, through divergent pathways and that distinct cofactors are required to support these protein-protein interactions. In the case of PKC-⑀, low affinity interactions with A F were detected in the absence of phorbol esters using the deletion mutant ⌬PKC-⑀ 222-230 . In contrast, the physical association of A F with the intact enzyme was of a higher affinity (13) and significantly enhanced in the presence of 4␤-PDBu. These findings suggest that interactions between A F and its cognate binding site in PKC-⑀ may be positively regulated by allosteric interactions between diacylglycerol and the C1 homology domains of PKC-⑀. The finding that arachidonic acid synergistically interacts with diacylglycerol to promote A F binding to PKC-⑀ (13) is consistent with this model.
Multiple localization signals have been tentatively identified in the PKC-⑀ sequence (10,11,13,16,17). Two of these signal motifs have been mapped to the C1 domain of PKC-⑀, the actin binding motif discussed here and a Golgi localization signal that reportedly dominates all other localization signals when the enzyme is in an inactive conformation (16). Importantly, deletion of the actin binding motif did not alter the Golgi localization of inactive PKC-⑀ that has been observed in this and previous studies. Thus, comparisons of NIH 3T3 cells stably overexpressing PKC-⑀ and ⌬PKC-⑀ 222-230 may make it possible to study the functional importance of A F -PKC-⑀ interactions without interfering with its regulation of other cellular functions. In the present study, activated PKC-⑀ accumulated with A F in what appeared to be adhesion zones when NIH 3T3 fibroblasts were induced to attach and spread by the introduction of PMA. The failure of ⌬PKC-⑀ 222-230 overexpressing NIH 3T3 cells to establish such actin-based structures when treated with PMA indicates that the actin binding motif becomes a dominant localization signal when PKC-⑀ is activated and that A F -PKC-⑀ interactions may play an important role in regulating normal cell-cell as well as cell-substrate cohesion. A potential role of PKC-⑀ in regulating cell adhesion has recently been highlighted by the finding that this isozyme becomes selectively activated in HeLa cells during spreading (7). PKC activation has also been shown to be required to initiate fibroblast spreading on a fibronectin substrate (38) and to precede actin polymerization in the adhesion of HeLa cells to a gelatin substratum (39).
The finding that A F binding makes PKC-⑀ substantially more vulnerable to proteolytic degradation raised the possibility that this protein-protein interaction may lead to relief of the inhibition exerted by the pseudosubstrate domain on the kinase domain. On a structural level, it remains to be convincingly demonstrated that A F is actually capable of disrupting this interdomain interaction. However, it has now been established that A F is sufficient to stimulate PKC-⑀ phosphotransferase activity without drastically altering its substrate specificity. In contrast, A F apparently inhibits the phosphorylation of PKC-␤ II substrates by promoting the autophosphorylation of one or more Ser/Thr residues in this conventional isozyme (36). Preliminary evidence indicates that the C terminus of actin may play an important role in the A F -induced stimulation of PKC-⑀ activity, although this remains to be proven since removal of the C-terminal dipeptide of actin can cause significant changes in the overall topology of actin filaments (20). The present results, together with our previous studies (13), indicate that A F may be a bifunctional anchoring protein that maintains PKC-⑀ in an active conformation within cytoskeletal structures and appear to be assembled during cell adhesion. Further studies involving a comparison of PKC-⑀ and ⌬PKC-⑀ 222-230 overexpressing NIH 3T3 cell lines should reveal whether A F -PKC-⑀ interactions participate in the regulation of cell adhesion or the oncogenic cascade that is induced by overexpressing this gene in fibroblasts (5).