Regulation of PKCα activity by C1 C2 domain interactions

In this study, the role of interdomain interactions involving the C1 and C2 domains in the mechanism of activation of PKC was investigated. Using an in vitro assay containing only purified recombinant proteins and the phorbol ester, 4β-12-O-tetradecanoylphorbol-13-acetate (TPA), but lacking lipids, it was found that PKCα bound specifically, and with high affinity, to a αC1A-C1B fusion protein of the same isozyme. The αC1A-C1B domain also potently activated the isozyme in a phorbol ester- and diacylglycerol-dependent manner. The level of this activity was comparable with that resulting from membrane association induced under maximally activating conditions. Furthermore, it was found that αC1A-C1B bound to a peptide containing the C2 domain of PKCα. The αC1A-C1B domain also activated conventional PKCβI, -βII, and -γ isoforms, but not novel PKCδ or -ε. PKCδ and -ε were each activated by their own C1 domains, whereas PKCα, -βI, -βII, or -γ activities were unaffected by the C1 domain of PKCδ and only slightly activated by that of PKCε. PKCζ activity was unaffected by its own C1 domain and those of the other PKC isozymes. Based on these findings, it is proposed that the activating conformational change in PKCα results from the dissociation ofintra-molecular interactions between the αC1A-C1B domain and the C2 domain. Furthermore, it is shown that PKCα forms dimers via inter-molecular interactions between the C1 and C2 domains of two neighboring molecules. These mechanisms may also apply for the activation of the other conventional and novel PKC isozymes.

The 10 closely related isozymes that constitute the protein kinase C (PKC) 1 family of serine/threonine kinases each occupy critical nodes in the complex cellular signal transduction networks that regulate diverse cellular processes, including: secretion, proliferation, differentiation, apoptosis, permeability, migration, and hypertrophy (1)(2)(3)(4)(5)(6)(7). In common with many signaling proteins, the structure of PKC is modular, consisting of a C-terminal catalytic region containing the active site, and a regulatory region with conserved domains that mediate membrane association and activation. PKC isozymes are classified according to the structural and functional differences in these conserved domains (8,9). In the case of the "conventional" PKC␣, -␤I/␤II, and -␥ isozymes, these include the activatorbinding C1 domains, and the Ca 2ϩ -binding C2 domain. The C1 domains consist of a tandem C1A and C1B arrangement, each of which can potentially bind the endogenous activator, diacylglycerol and exogenous activators including phorbol esters. The "novel" PKC␦, -⑀, -, -, and -isozymes, contain C2 domains that lack Ca 2ϩ binding ability, while retaining functional C1A and C1B domains. The "atypical" PKC, -, and -regulatory domains also lack a functional C2 domain and contain a single C1 domain that lacks the ability to bind activators, the function of which remains obscure. Each isozyme becomes catalytically competent by undergoing multiple serine/threonine and tyrosine phosphorylations that are either autocatalytic, or catalyzed by "PKC kinases" such as the phosphoinositide-dependent kinase, PDK1 (7,10,11).
The mechanism by which the conventional PKC isozymes become membrane associated, and thus activated, involves two sequential steps. The first involves an initial Ca 2ϩ -and anionic phospholipid-dependent interaction of the C2 domain with the membrane, which is then followed by binding of diacylglycerol or phorbol esters to the C1 domains (12)(13)(14)(15). Mutagenesis studies have identified a number of critical hydrophobic residues within the C1 domains of PKC␦ and PKC␣ that appear to have distinct roles in ligand binding and membrane association (16 -18). Based on the x-ray crystallographic structure of the C1B domain of PKC␦, it has been suggested that activator binding to the C1 domain facilitates membrane-insertion by "capping" a hydrophilic groove to form a contiguous hydrophobic surface that can interact with the membrane interior (19). However, it would appear from our studies that the interaction of the phorbol ester induces a conformational change in the C1 domain, the extent of which is reflected in the activity of the enzyme, implying a more active role for phorbol ester-C1 domain interactions beyond that of presenting a hydrophobic surface to the interior of the membrane (20,21). The interaction of diacylglycerol with the C1 domains also results in an increased stereo-and regiospecificity of both membrane association and activation for PS (13,(22)(23)(24). The combined interactions of the C1 and C2 domains with the membrane provides the free energy required for structural rearrangements that lead to the dissociation of the N-terminal pseudosubstrate from the active site to allow substrate binding (25)(26)(27). This process is thought to be further facilitated by a weak interaction of the released pseudosubstrate with anionic head groups at the membrane surface (28).
There is increasing evidence supporting the notion that the existence of two C1 domains affords a complex modulatory role in the regulation of PKC activity. Phorbol esters have been shown to interact with both of the C1A and C1B domains, with distinct low and high affinities (29 -33). Furthermore, we have shown that diacylglycerol inhibits the low affinity phorbol ester interaction while enhancing high affinity phorbol ester binding, indicating that the diacylglycerol has a higher affinity for the low affinity phorbol ester-binding site than does phorbol ester itself (30,31). Additional support for the non-equivalence of the interaction of diacylglycerol, phorbol esters, and also other activators with the two C1 domains has been provided by other studies that have observed non-equivalent roles of the domains with respect to membrane association and activation (34 -37). Studies from this laboratory showed that the increased level of phorbol ester binding that results from interaction of diacylglycerol with the low affinity phorbol esterbinding site on conventional PKC isozymes corresponded to an elevated level of enzyme activity that was greater than that induced by either phorbol ester or diacylglycerol alone (30,31,38). The above, and recent data suggesting that the C1A of PKC␣ is the diacylglycerol-binding site, while the C1B domain binds phorbol ester (16,24), are compatible with the C1A and C1B domains containing the low and high affinity phorbol ester-binding sites, respectively.
Inter-domain interactions are important in PKC regulation, although detailed features of these interactions remain obscure. The conformational change that leads to the displacement of the pseudosubstrate of membrane-associated PKC isozymes involves pronounced rearrangements of the individual domains that constitute the enzyme molecule. In a recent study, it was shown that the rate of translocation of GFPtagged PKC␥ in RBL cells was markedly slower when compared with that of a truncation mutant lacking the N-terminal V1 region (14). Based on this, it was suggested that membrane association and activation resulting from binding of diacylglycerol to the C1 domains of this isozyme first requires the release of a "V1 clamp," which holds the C1 domain between the catalytic and V1 region, due to an interaction of the pseudosubstrate with the active site. In another study, the mutation of a single aspartate (Asp 55 ) residue in the C1A domain of PKC␣ was shown to result in a marked reduction in both the PS and Ca 2ϩ concentration requirements for membrane association and activation, implicating a tethering of the C1A domain to a neighboring region within the PKC␣ molecule (15,24). It was suggested that such an interaction might retain the isozyme in an inactive conformation by preventing the penetration of the C1A domain into the membrane and thus its interaction with diacylglycerol (24).
The aim of the present study was to determine the role of interdomain interactions in the mechanism of activation of PKC␣. Using in vitro activity and binding assays, it was found that PKC␣ engaged in a phorbol ester-dependent, high affinity and specific interaction with a fusion protein that contained its own C1A and C1B domains (␣C1A-C1B). The level of activity induced by interaction with the ␣C1A-C1B domain was found to be comparable with that resulting from membrane association induced under maximally activating conditions. Also, the ␣C1A-C1B domain interacted with a fusion protein containing the C2 domain of PKC␣. Taken together, these findings provide evidence for the existence of an interdomain interaction between the C1 and C2 domains, the dissociation of which is a rate-determining step in the mechanism of activation of PKC␣.
Measurements of PKC Activity-PKC isozyme activities were assayed by measuring the rate of phosphate incorporation into a peptide substrate as previously described (30). For the "conventional" PKC isoforms and the their catalytic subunits, a peptide corresponding to the phosphorylation site domain of myelin basic protein (QKRPSQRSKYL, MBP 4 -14 ) was used as the substrate, whereas assays of novel PKC and atypical PKC activity used a peptide corresponding to the pseudosubstrate region of novel PKC⑀ (⑀-peptide), in which the single alanine residue was replaced by serine (25,41,42). Briefly, the assay (75 l) consisted of 50 mM Tris-HCl (pH 7.40), 0.1 mM EGTA or CaCl 2 , 50 M MBP 4 -14 , or 50 M ⑀-peptide, TPA (500 nM or as indicated) or varying levels of DiC 8 , and fusion proteins containing the appropriate PKC domains present at a fixed concentration of 10 nM unless otherwise indicated. Where added, POPC and BPS were present at a total concentration of 150 M as large unilamellar vesicles, prepared as described previously (43). After thermal equilibration to 30°C, assays were initiated by the simultaneous addition of the required PKC isoform (0.3 nM) along with 5 mM Mg 2ϩ , 15 M ATP, 0.3 Ci of [␥-32 P]ATP (3000 Ci/mmol) and terminated after 30 min with 100 l of 175 mM phosphoric acid. Following this, 100 l was transferred to P81 filter papers, which were washed three times in 75 mM phosphoric acid. Phosphorylated peptide was determined by scintillation counting.
Measurements of PKC-C1-domain Binding Using Surface Plasmon Resonance (SPR)-Binding of C1 domain peptides to PKC isozymes was determined using SPR from measurements of the accompanying increase in refractive index as a function of time using a Biacore TM 2000 (Biacore, Inc., Piscataway, NJ). All measurements were performed at 25°C. Initially, the ␣C1A-C1B or ⑀C1A-C1B domains were captured through the respective (His) 6 tag to the nickel/NTA surface of an NTA sensor chip, prepared according to the manufacturers instructions (Biacore, Inc., Piscataway, NJ), to a level of 75 response units. Solutions containing either PKC␣ isozymes or fusion peptides at the required concentrations, in the presence or absence of 500 nM TPA were then injected over this surface and the response was measured as a function of time. The surface was regenerated after each injection by two 10-s injections of 100 mM NaOH, followed by a single 10-s injection of 10 mM HCl. After subtraction of the contribution of bulk refractive index changes and nonspecific interactions of PKC isozymes with the nickel-NTA surface, which were typically less than 1% of the total signal, the individual association (k a ) and dissociation (k d ) rate constants were obtained by global fitting of data to a 1:1 Langmuir binding model using BIAevaluation TM (Biacore, Inc.). These values were then used to calculate the dissociation constant (K D ). The values of average squared residual ( 2 ) obtained were not found to be significantly improved by fitting data to models that assumed bivalent or heterogeneous interactions between PKC isozymes and C1 domain peptides. In a separate control experiment (results not shown), it was found that the contribution of mass transport to the observed values of K D was negligible, based on the observation that these values were independent of flow rate within a range encompassing that used (10 to 50 l min Ϫ1 ).

RESULTS
To investigate the role of domain-domain intramolecular interactions in the mechanism of activation of PKC, the assumption was made that if such interactions occur within the isozyme molecule, and are involved in the activating conformational change, then peptides corresponding to the isolated domains might compete for these interactions and modulate the activity of the isozyme. To address this, various fusion proteins containing regions of the regulatory domain spanning the C1A and C1B domains were prepared, as shown in Fig. 1. Binding of these domains to PKC, and the effects on PKC isozyme activities, were determined using assay systems that contained only purified proteins along with the required peptide substrates, cofactors, and activators. By excluding membranes from the assay systems, effects could be unambiguously ascribed to direct protein-protein interactions between these domains and PKC.
PKC␣ Binds Directly to the ␣C1A-C1B Domain-Based on measurements of SPR, it was found that, in the presence of TPA, PKC␣ bound to an immobilized fusion peptide containing the isolated ␣C1A-C1B domain in a reversible, concentrationdependent manner ( Fig. 2A). The interaction was phorbol esterdependent since it was found that the level of binding in the absence of TPA was negligible. The value of K D , calculated from the ratio of the association and dissociation rate constants, indicated a high affinity interaction (Table I). Furthermore, the value of the maximal level of PKC␣ binding at equilibrium (R max ) is consistent with a PKC␣-␣C1A-C1B binding stoichiometry of 1:1, assuming maximal occupancy of ligand-binding sites. PKC␣ was also found to interact with the ⑀C1A-C1B domain (Fig. 2B), although with a markedly reduced affinity, as shown by the ϳ500-fold increase in the value of K D (Table I).
It has been suggested in a previous study that the C1A of PKC␣ may interact with residues within the C2 domain, and thereby impede the activating conformational change (24). Consistent with this, the results shown in Fig. 2C indicate that in the presence of TPA, the ␣C1A-C1B domain binds to a fusion protein containing the C2 domain of PKC␣ (␣C1A-C1B-C2).
Similar to the interaction of PKC␣ with the ␣C1A-C1B, the binding of the ␣C1A-C1B-C2 peptide to the ␣C1A-C1B domain was found to be TPA-dependent. Note that in this experiment the ␣C1A-C1B domain was initially bound to the sensor chip surface through the (His) 6 tag to a level corresponding to saturation. Under these conditions, any interactions of the ␣C1A-C1B with itself were also saturated, ruling out the possibility that the immobilized ␣C1A-C1B domain may have interacted with the C1A-C1B portion of the ␣C1A-C1B-C2 fusion protein. Importantly, the value of K D for the interaction was similar to that calculated for the interaction of the ␣C1A-C1B domain with intact PKC␣ (Table I). Furthermore, it was found that ␣C1A-C1B-C2 bound to the immobilized ⑀C1A-C1B (Fig. 2D), although with markedly reduced affinity ( Table I).
Interaction of PKC␣ C1A and C1B Domains with PKC␣ Results in Activation-PKC␣ was found to be activated by low nanomolar levels of the ␣C1A-C1B domain in the presence of a fixed concentration of TPA (500 nM) and in the absence of membrane lipids (Fig. 3A, q). This effect was phorbol ester-dependent since the ␣C1A-C1B domain negligibly affected PKC␣ activity in the absence of TPA (Fig. 3A, Ⅺ). The concentration of ␣C1A-C1B corresponding to a half-maximal increase in PKC␣ activity was ϳ1 nM, which is consistent with the value of K D for the interaction of PKC␣ with the ␣C1A-C1B domain, determined from the SPR binding data (Table I), and again indicates that the activation resulted from a high-affinity interaction. It should be noted that the maximal specific activity of PKC␣ induced by the ␣C1A-C1B domain was comparable with that determined for PKC␣ associated with membranes composed of BPS and POPC in the presence of Ca 2ϩ and TPA, each activator being at saturating levels (Fig. 3A, ૺ).
To rule out the possibility that the observed activation of PKC␣ by the ␣C1A-C1B domain may have resulted from an interaction with a site contained within the catalytic region of PKC␣, the effects of the ␣C1A-C1B domain on the activity of catalytic subunits derived from limited proteolysis of a mixed PKC␣, -␤I/II, and -␥ preparation was determined (Fig. 3A, छ). Consistent with the reported independence of the activity of the catalytic subunit from cofactor and activator requirements (44,45), the specific activity of the preparation was found to be similar to that observed for PKC␣ induced by association with BPS/POPC vesicles in the presence of Ca 2ϩ and TPA (Fig. 3A,  ૺ). Whereas PKC␣ was activated by low nanomolar levels of the ␣C1A-C1B domain, the activity of the catalytic subunit preparation was found to be unaffected by the domain within a similar concentration range (Fig. 3A, छ). This is consistent with the site of interaction of PKC␣ with the ␣C1A-C1B domain that mediates the activation being contained within the regulatory domain. However, similar to the effects on intact PKC␣ activity, the presence of the ␣C1A-C1B domain concentrations greater than 10 nM also resulted in an inhibition of catalytic subunit activity. This observation indicates that the inhibitory effects of high levels of the ␣C1A-C1B domain may involve interactions with the catalytic domain, which is consistent with the findings of a recent study that showed that GST fusion proteins containing the regulatory domains of PKC␣ and PKC⑀ each inhibited the activity of the catalytic subunit of PKC␣ (46). The possibility that the activation of PKC␣ by the ␣C1A-C1B domain may have involved GST or (His) 6 was ruled out by the finding that PKC␣ activity was unaffected by a fusion peptide containing these moieties alone (Fig. 3A, OE).
The dependence of PKC␣ activity on the concentration of TPA, determined in the presence or absence of ␣C1A-C1B, is shown in Fig. 3B. In the absence of ␣C1A-C1B, a small increase in the level of PKC␣ activity was observed within a high TPA concentration range (q). This is consistent with the results of previous studies that have indicated that PKC isozymes bind phorbol esters in the absence of membranes, although with relatively low affinity, and that this results in partial activation (47)(48)(49)(50). Similar to the effect of membrane association, the addition of the ␣C1A-C1B domain resulted in a Ͼ1000-fold  Table I. Other details are given under "Experimental Procedures."

TABLE I
Kinetic analysis of the interaction of intact PKC␣ or the ␣C1A-C1B-C2 peptide with the ␣C1A-C1B or the ⑀C1A-C1B domains using SPR The ␣C1A-C1B and ⑀C1A-C1B domains were initially immobilized on the surface of a nickel/NTA sensor chip and PKC␣, or the ␣C1A-C1B-C2 peptide, were then injected over these surfaces in the presence of 500 nM TPA. The dissociation constants (K D ) were obtained from the ratio of the association (k a ) and dissociation (k d ) rate constants, derived from global fits of response against time data to a 1:1 Langmuir binding model. Further details are given under "Experimental Procedures" and in the legend to Fig. 2 decrease in the concentration dependence for TPA induced activity, consistent with a dramatic increase in phorbol ester binding affinity (Fig. 3B, f). Thus, the calculated value of the TPA concentration corresponding to a half-maximal increase in PKC␣ activity of 5 Ϯ 2 nM is within a phorbol ester concentration range shown previously to be sufficient to elicit activation of the membrane-associated isozyme (31). Similar to TPA, the soluble diacylglycerol, DiC 8 , also activated PKC␣ in the presence of the ␣C1A-C1B domain, in a concentration-dependent manner (Fig. 3B, OE). The concentration of DiC 8 required to induce a half-maximal increase in PKC␣ activity (875 Ϯ 55 nM) was ϳ150-fold greater than that observed for TPA, which is consistent with the reduced affinity of diacylglycerols for binding to PKC compared with phorbol esters (51,52). The concentration-dependent effects of the individual C1A and C1B domains on the activity of PKC␣ induced in the presence of a saturating level of TPA, are shown in Fig. 3C. Both the ␣C1A (q) and ␣C1B (f) domains activated PKC␣ with similar concentration dependence. However, the ␣C1A and ␣C1B concentrations required to induce a half-maximal increase in activity were in each case ϳ200-fold greater than the value obtained for the full-length ␣C1A-C1B domain. Furthermore, although the ␣C1A domain induced a greater level of activity than the ␣C1B domain, this activity was ϳ3-fold lower than that induced by the ␣C1A-C1B domain.
The Effect of ␣C1A-C1B on the Activity of the PKC␣ C1A Domain Mutant, D55A-Recently, it was observed that the mutation of aspartate 55 to an alanine in the C1A domain of PKC␣ (D55A) resulted in an increase in membrane binding affinity and an increased level of Ca 2ϩ and PS independent activity. From this, it was proposed that PKC␣ might, in part, be restrained in an inactive conformation by an inhibitory intramolecular interaction involving this residue (24). Here, we further examined the intrinsic activity of D55A, first in the absence of the ␣C1A-C1B domain, membranes, Ca 2ϩ , and TPA, and showed it was higher than that of wild-type PKC␣ (Fig. 4), as reported previously (24). Consistent with the higher intrinsic activity of D55A, the addition of TPA alone resulted in significant D55A activity, relative to the small effect on wildtype PKC␣ activity. The concentration-response curves for TPA-induced activation of D55A and wild-type PKC␣, shown in Fig. 4 (inset), indicate that this corresponds to a marked decrease in the TPA concentration range required for D55A activation. In the absence of TPA, the activities of D55A and PKC␣ were both negligibly affected by the addition of the ␣C1A-C1B or ␣C1A domain. Importantly, contrasting with the TPA-induced activation of wild-type PKC␣ by the ␣C1A-C1B or ␣C1A domains, neither domain activated D55A in the presence of TPA.
Effects of C1A-C1B Domains of PKC␣, -␦, -⑀, and -on the Activities of a Panel of PKC Isoforms-To determine the isozyme specificity of the activating effect of the ␣C1A-C1B domain, the concentration-dependent effects of ␣C1A-C1B, and also the ␦C1A-C1B, ⑀C1A-C1B, and C1 domains, on the activities of PKC␣, -␤I, -␤II, -␥, -␦, -⑀, and -were each determined in the presence of TPA (Fig. 5). Consistent with the results shown in Fig. 3A, PKC␣ activity was potentiated ϳ100-fold by the ␣C1A-C1B domain in a concentration-dependent manner (Fig. 5, q). The ␣C1A-C1B domain also activated conventional PKC␤I, -␤II, and to reduced extent, PKC␥. However, the activities of novel PKC␦ or -⑀ were each unaffected. Consistent with the observation that the ⑀C1A-C1B domain bound to PKC␣,  Table I), the ⑀C1A-C1B domain was found to induce a ϳ10-fold increase in PKC␣ activity, while having negligible effects on the activities of PKC␤I and -␤II (Fig. 5, Ⅺ). This suggests that it may also share with the ␣C1A-C1B domain some limited ability to interact with PKC␣. By contrast, each of the conventional PKC␣, -␤I, -␤II, and -␥ activities were unaffected by the ␦C1A-C1B domain (Fig. 5, OE), and also by the C1 domain (Fig. 5, ƒ). Similar to PKC␣, the activities of PKC␦ and -⑀ were each potentiated by their respective C1A-C1B domains, while being unaffected by the C1 domains of the other isozymes. Furthermore, as found for PKC␣, the level of activity induced by the respective C1 domain was close to that induced by membrane association in the presence of TPA and Ca 2ϩ , each activator being present at a maximally activating concentration. By contrast to the conventional and novel PKC isozymes, the activity of atypical PKC was unaffected by its own C1 domain and those of the other PKC isozymes.
Intermolecular Interactions Involving the ␣C1A-C1B Domain Mediate in the Self-association of PKC␣-The possibility exists that once disengaged from the intramolecular interaction, the C1 and C2 domains may participate in an intermolecular interaction with their counterparts on a neighboring PKC␣ molecule, which would therefore be expected to result in the formation of dimers. To address this possibility, the activity of PKC␣ was measured as a function of its concentration, as shown in Fig. 6. The double-log plot of PKC␣ activity against concentration, obtained in the presence of a fixed level of TPA but in the absence of membrane lipids (Fig. 6, q), contained an inflection at a PKC␣ concentration of ϳ4 nM, which corresponded to a change in gradient from 0.83 Ϯ 0.5 to 2.23 Ϯ 0.3. This result is consistent with a change from 1 to 2 active sites per active PKC␣ complex and suggests the isozyme undergoes a monomer-dimer equilibrium, the dimer having a relatively higher activity. Furthermore, consistent with the notion that the dimerization of PKC␣ may be mediated by intermolecular C1-C2 domain interactions, the presence of a fixed concentration of the ␣C1A-C1B domain (Fig. 6, OE) yielded a linear relationship of slope 0.89 Ϯ 0.3 within the same PKC␣-concentration range. To address the question whether PKC␣ associated with membranes may also engage in a monomer-dimer equi-librium, the concentration dependence of PKC␣ activity was measured in the presence of BPS/POPC vesicles, Ca 2ϩ , and TPA (Fig. 6, f). The concentration of BPS (20 mol % of the total lipid concentration), Ca 2ϩ (0.1 mM), and TPA (500 nM) used each corresponded to those previously shown to induce complete membrane association of PKC␣ and a maximal level of activation of the isozyme (38). It was found that the activity of membrane-associated PKC␣ again displayed positive cooperativity with respect to PKC␣ concentration. Thus, a double-log plot of PKC␣ activity against concentration again contained an inflection corresponding to a change in the slope of the regression line from 0.78 Ϯ 0.6 to 2.02 Ϯ 0.4, which is consistent with PKC␣-PKC␣ dimerization. Furthermore, the PKC␣ concentration corresponding to the inflection point was ϳ10-fold lower than that observed for PKC␣ in the absence of membranes, indicating that in this case dimerization occurred at a lower PKC␣ concentration, consistent with an increased monomerdimer association constant.

DISCUSSION
The activation of PKC isozymes by association with membranes ultimately hinges on a conformational change induced by diacylglycerol-or phorbol ester binding to the C1 domains and PS/Ca 2ϩ binding to the C2 domain (12,15), that results in the removal of the pseudosubstrate from the active site to allow binding of a substrate (25)(26)(27). This contribution provides evidence supporting the existence of an additional intramolecular interaction between the C1 and C2 domains of PKC␣, the dissociation of which is a requirement for the activating conformational change to proceed.
The observation from SPR experiments, showing that the ␣C1A-C1B domain peptide interacted with both intact PKC␣ and with the C2 portion of the ␣C1A-C1B-C2 peptide, suggests a critical and functional interaction between the C1A, C1B, and C2 domains of the isozyme. This is further reinforced by the observation that the interaction of PKC␣ with the ␣C1A-C1B domain resulted in pronounced enzyme activity, consistent with the activating conformational change requiring the dissociation of a C1-C2 domain interaction that otherwise holds the enzyme in a "closed" inactive state. We therefore propose that the activating conformational change in PKC␣ corresponds to an equilibrium between closed and "open" states that is regulated by a C1-C2 interaction; the open active state being stabilized by interaction with a membrane surface, or with filamentous actin (53). This provides experimental evidence for an interaction involving the C2 domain that was suggested could occur in a recent study for PKC␣ (24), and also was alluded to in an earlier study for a novel PKC isozyme from Aplysia (54).
The finding that phorbol ester or diacylglycerol was absolutely required for PKC␣ activation by the ␣C1A-C1B domain (see Fig. 3A), suggests that a transient dissociation of the C1-C2 interaction may be initially required to expose the phor-bol ester-or diacylglycerol-binding sites within the C1 domains. This would be consistent with the low level of PKC activity that is induced by phorbol ester alone in the absence of membranes (see Fig. 3B and Refs. [47][48][49][50]. The observation that the interaction of PKC␣ with the ␣C1A-C1B resulted in a dramatic reduction in the TPA concentration dependence of activity (see Fig. 3B) is also consistent with the proposal that the phorbol ester-binding sites within the C1 domains are exposed by the formation of the open state.
In this model, the binding of the ␣C1A-C1B domain peptide to the C2 domain combined with phorbol ester or diacylglycerol binding to the C1 domains, would then "lock" the PKC molecule in an open state by blocking the formation of the C1-C2 interaction. This open state appears to correspond to the fully active conformation of PKC␣, based on the observation that the level of PKC␣ activity induced by interaction with TPA and the ␣C1A-C1B domain approached that resulting from membrane association induced under maximally activating conditions. The finding that the interaction of the ␣C1A-C1B domain with the ␣C1A-C1B-C2 peptide was also TPA-dependent, suggests that the properties of the ␣C1A-C1B-C2 peptide may reflect those of the intact PKC␣ molecule, in that it may also be engaged in a transient equilibrium between closed and open states; the latter being stabilized by phorbol ester binding to its C1 domains.
The finding that the dose-response curves for the activation of PKC␣ by the C1A and C1B domains were each shifted to the right by ϳ2 orders of magnitude relative to that for the fulllength ␣C1A-C1B domain (compare Fig. 3, A with C), suggests that residues in both C1A and C1B domains may participate in interdomain interactions with the C2 domain that stabilizes the closed conformation of PKC␣. The observation that the activity of D55A was still to some extent potentiated by phorbol ester (see Fig. 4), although the C1A-C2 domain interaction is absent in this mutant, also indicates that the dissociation of other interactions in addition to that between the C1A and C2 domains may be involved.
Investigations using the PKC␣ C1A domain mutant, D55A, reveal further features of the unfolding mechanism that leads to activation. The intrinsic activator-independent activity of the D55A mutant was observed to be markedly increased compared with that of wild type PKC␣ (24) (and see Fig. 4), while the TPA concentration dependence of the activity of D55A was dramatically reduced compared with that obtained for wildtype PKC␣ (see Fig. 4, inset). Furthermore, unlike the wild type PKC␣, the activity of D55A was not affected by the addition of the ␣C1A-C1B domain.
Earlier studies have suggested that the conformation of the "cytosolic" isozyme may be retained in an inactive or closed state by a "strong interaction" between the pseudosubstrate and the active site that "clamps" the C1 domains between the V1 and catalytic domain (14), although binding data and assignment of the regions of the regulatory domain involved were not determined. Nevertheless, this V1 clamp model, which was suggested to prevent the binding of diacylglycerol or phorbol ester to the C1 domains, is entirely compatible with our observations, and is extended to show that additional C1-C2 interactions are required for the unfolding of the pseudosubstrate from the active site. While in the previous study it was suggested that an electrostatic interaction between of the V1 region and the membrane surface was required to stabilize the open state of PKC␥ (14), it was found here that PKC␣ was fully activated by interaction with the ␣C1A-C1B domain in the presence of TPA, even though membranes were absent. This suggests that the reversible electrostatic interaction of the pseudosubstrate with lipid head groups favors, but may not alone, induce the formation of the open active state.
With regard to isozyme specificity, the interaction with the ␣C1A-C1B domain and the ensuing activation appears to be confined to conventional PKC␣, ␤I, ␤II, and to a lesser extent PKC␥, based on the observation that the activities of novel PKC␦, -⑀, and atypical PKC were unaffected by the domain (see Fig. 5). This apparent specificity was also suggested by the observation that the activities of PKC␣, -␤I, and -␤II were, to a lesser degree, activated by the ⑀C1A-C1B domain and were unaffected by the ␦C1A-C1B and C1 domains. The finding that the ␦C1A-C1B and ⑀C1A-C1B domains also potentiated the activities of PKC␦ and PKC⑀, respectively, suggests that intramolecular interactions involving the C1 domains of these isozymes may also mediate the activating conformational change. Although, it is possible that these interactions may be similar to those between the C1 and C2 domains of PKC␣, the organization of these domains and also the pseudosubstrate within the structures of PKC␦ and -⑀ differ markedly from that of PKC␣.
The observation that measurements of PKC␣ activity as a function of PKC␣ yielded curves that contained an inflection point corresponding to a change in gradient from ϳ1 to ϳ2 indicates a change in the reaction kinetics from being firstorder to second-order, with respect to the enzyme concentration. This is consistent with a concentration-dependent equilibrium between monomeric and dimeric forms of PKC␣. Evidence supporting the notion that PKC may be active in PKC-PKC and/or PKC-substrate complexes has also been provided elsewhere (55)(56)(57)(58)(59)(60). Furthermore, evidence that PKC may become active upon self-association in the cellular environment was presented recently based on the observation of PKC␣ dimers in lysates derived from murine B82L fibroblasts treated with calcium ionophore, phorbol ester, or epidermal growth factor (60). The self-association of PKC␣ might be mediated by the same intramolecular interactions between the C1 and C2 domains discussed above but now between two different PKC␣ molecules, since the presence of the ␣C1A-C1B peptide restored a linear relationship between enzyme concentration and activity. The observation that the interaction of the ␣C1A-C1B domain was specific for PKC␣, Ϫ␤I, Ϫ␤II, and to a lesser extent PKC␥, and that the activity of PKC␣ was relatively less affected by the C1 domains of PKC␦, -⑀, and -, suggests that PKC␣ can dimerize with itself, and potentially also with other conventional PKC isozymes. However, in the cellular environment, whether PKC isozymes form homo-or heterodimers would depend on whether these isozymes are co-localized. The roles of intermolecular PKC-PKC interactions, and the balance between inter-and intramolecular C1-C2 interactions in the mechanism of activation of PKC remain to be investigated.
Conclusion-The existence of C1-C2 domain interactions that retain the PKC␣ molecule in an inactive conformation was directly demonstrated by the observation that the domains involved bind directly to one another and that this leads to activation. The current observations support a model in which PKC␣ activation corresponds to a transition from a closed inactive to an open active state that is governed by the dissociation of the C1-C2 interaction.