Ca2+ Differentially Regulates Conventional Protein Kinase Cs’ Membrane Interaction and Activation*

The regulation of conventional protein kinase Cs by Ca2+ was examined by determining how this cation affects the enzyme’s 1) membrane binding and catalytic function and 2) conformation. In the first part, we show that significantly lower concentrations of Ca2+ are required to effect half-maximal membrane binding than to half-maximally activate the enzyme. The disparity between binding and activation kinetics is most striking for protein kinase C βII, where the concentration of Ca2+promoting half-maximal membrane binding is approximately 40-fold higher than the apparent K m for Ca2+ for activation. In addition, the Ca2+ requirement for activation of protein kinase C βII is an order of magnitude greater than that for the alternatively spliced protein kinase C βI; these isozymes differ only in 50 amino acids at the carboxyl terminus, revealing that residues in the carboxyl terminus influence the enzyme’s Ca2+ regulation. In the second part, we use proteases as conformational probes to show that Ca2+dependent membrane binding and Ca2+-dependent activation involve two distinct sets of structural changes in protein kinase C βII. Three separate domains spanning the entire protein participate in these conformational changes, suggesting significant interdomain interactions. A highly localized hinge motion between the regulatory and catalytic halves of the protein accompanies membrane binding; release of the carboxyl terminus accompanies the low affinity membrane binding mediated by concentrations of Ca2+ too low to promote catalysis; and exposure of the amino-terminal pseudosubstrate and masking of the carboxyl terminus accompany catalysis. In summary, these data reveal that structural determinants unique to each isozyme of protein kinase C dictate the enzyme’s Ca2+-dependent affinity for acidic membranes and show that, surprisingly, some of these determinants are in the carboxyl terminus of the enzyme, distal from the Ca2+-binding site in the amino-terminal regulatory domain.

Protein kinase C is a multi-domain family of enzymes that transduces the myriad of signals resulting in phospholipid hydrolysis (1)(2)(3). The amino-terminal regulatory half of the conventional isozymes comprises three functionally distinct domains as follows: 1) an intramolecular inhibitory domain, the pseudosubstrate, that occupies the active site when the enzyme is inactive; 2) two cysteine-rich domains (C1) that are each capable of binding diacylglycerol or phorbol esters; and 3) the C2 domain that binds Ca 2ϩ and acidic phospholipids. The C1 and C2 domains function as discrete membrane-targeting motifs (4); binding of ligand to either domain alone can promote protein kinase C translocation to membranes, but both domains must be membrane-bound for the high affinity interaction resulting in maximal activation (3). The carboxyl-terminal half of the protein comprises the conserved kinase core, followed by a carboxyl-terminal region whose sequence varies among isozymes.
Understanding the mechanism by which Ca 2ϩ first recruits protein kinase C to membranes and then activates its kinase activity is essential to understanding how calcium mobilization regulates the function of protein kinase C in vivo. Although the relationship between Ca 2ϩ effects on binding and activity has not been established, much progress has been made toward describing the effects of the cation on protein kinase C activation. Since the initial reports that lipid-stimulated activation of protein kinase C requires Ca 2ϩ (5), biochemical analyses from several laboratories have established that Ca 2ϩ decreases the concentration of negatively charged phospholipids required for maximal activation (6,7) or membrane binding (8,9). It is also well established that the Ca 2ϩ -mediated increase in affinity for acidic membranes is considerably more pronounced in the presence of diacylglycerol or phorbol esters (10). The molecular mechanism for this synergy was recently found to result not from allosteric interactions between the binding sites for these two ligands but rather because each cofactor, by separate mechanisms, increases protein kinase C's affinity for phosphatidylserine (9,11,12). Thus, considerably less diacylglycerol is required to activate protein kinase C as the Ca 2ϩ concentration is raised, and vice versa, because both Ca 2ϩ and diacylglycerol cause tighter binding to phosphatidylserine (and it is this tight binding that results in release of the autoinhibitory pseudosubstrate from the active site (13)). The dissociation constant of Ca 2ϩ from protein kinase C has recently been estimated to be 3 mM in the absence of lipid and 0.7 M from the protein kinase C-lipid complex (11). Furthermore, the enzyme's membrane affinity has been shown to vary linearly with Ca 2ϩ concentration from the low micromolar to the submillimolar concentration range of this cation (11).
The stoichiometry of the protein kinase C-Ca 2ϩ complex has been reported to be 1 (11,14) to 1 or 2, depending on the isozyme (15), and to up to 8 (8). However, NMR and crystallographic data from C2 domains analyzed in the presence of Ͼ10 mM Ca 2ϩ (16) or Sm 3ϩ (17) indicate that the Ca 2ϩ -binding pocket coordinates two metal ions for the C2 domains studied; the actual affinities for each cation and whether both are bound in the presence of lipid remains to be established. Despite great advances in elucidating how Ca 2ϩ is accommodated in the protein kinase C structure, understanding how this cation affects the function of protein kinase C awaits further biochemical characterization.
This contribution examines the role of Ca 2ϩ in regulating the membrane binding and catalytic function of protein kinase C and how these functional changes correlate with structural changes. Specifically, we show that 1) membrane binding and activation of conventional protein kinase Cs display different Ca 2ϩ requirements, with activation requiring higher concentrations of the cation than membrane binding, and 2) distinct conformational changes accompany Ca 2ϩ -dependent membrane binding and Ca 2ϩ -dependent activation. In addition, we show that determinants in the carboxyl terminus modulate protein kinase C's affinity for Ca 2ϩ .
Antibodies-Polyclonal antibodies against bacterially expressed catalytic domain and bacterially expressed regulatory domain of protein kinase C ␤II were a gift of Drs. Andrew Flint and Daniel Koshland, Jr.; a polyclonal antibody against a peptide comprising residues 645-673 of protein kinase C ␤II was generously provided by Lilly, and one against residues 19 -32 of protein kinase C ␤II was from the laboratory of Dr. David S. Williams.
Protein Kinase C-Protein kinase C ␣, ␤I, or ␤II were expressed in Sf21 insect cells (Invitrogen) by infection with recombinant baculovirus (generous gifts from Robert Bell, Peter Parker, and Daniel Koshland Jr., respectively) and purified to homogeneity as described (13). The enzymes were stored at Ϫ20°C in 10 mM Tris buffer, pH 7.5 at 4°C, 150 mM KCl, 0.5 mM DTT, 0.5 mM EDTA, 0.5 mM EGTA, and 50% glycerol.
Protein Kinase C Activity Assay-Protein kinase C (typically 3 nM) activity was assayed by measuring the initial rate of [ 32 P]phosphate incorporation from [␥-32 P]ATP (50 M; 3000 Ci mmol Ϫ1 ) into saturating amounts of a protein kinase C-selective peptide (50 g ml Ϫ1 ) or protamine sulfate (0.2 mg ml Ϫ1 ) as described (13). Reactions were allowed to proceed for 6 min at 30°C. The concentrations of CaCl 2 and lipid present in the assays are indicated in the figure legends.
Centrifugation Assay for Protein Kinase C Membrane Binding-Protein kinase C (3-30 nM) was incubated with sucrose-loaded vesicles (100 M lipid; 40 mol % POPS, 5 mol % diacylglycerol, 55 mol % POPC) and various Ca 2ϩ concentrations in the presence of 20 mM HEPES buffer, 0.8 mM DTT, 0.3 mg ml Ϫ1 bovine serum albumin, 500 M EGTA, 60 mM KCl, and in the presence or absence of 5 mM MgCl 2 . Membrane binding was measured by determining the fraction of protein kinase C bound to sucrose-loaded vesicles after centrifugation for 30 min at 100,000 ϫ g, 22°C. The amount of protein kinase C in the supernatant or associated with the vesicles was measured either by assaying kinase activity toward protamine sulfate as described above or by Western blot analysis of supernatant and pellet samples.
Trypsin Sensitivity Assay for Protein Kinase C Membrane Binding-This assay takes advantage of the large increase in trypsin sensitivity of protein kinase C's hinge region that accompanies membrane binding (22). For comparison of the two binding assay methods, protein kinase C (30 nM) was incubated under conditions described above for the centrifugation assay. Samples were treated with trypsin (0.12 units ml Ϫ1 ; 10 min at 30°C), and proteolysis was quenched by addition of 0.25 volume of SDS-PAGE sample buffer. Samples were analyzed by SDS-PAGE (8% polyacrylamide) followed by electrophoretic transfer to nitrocellulose membranes and Western blotting with protein kinase C antibodies against the carboxyl terminus of ␤II. Quantitative analysis of proteolysis was carried out by densitometric scanning of the intact protein kinase C bands with a scanning densitometer (Molecular Dynamics). Percent of protein kinase C bound to membranes was correlated to the amount of proteolyzed protein as described (22).
Proteolysis-Protein kinase C (approximately 30 nM) was incubated at 30°C in the presence of HEPES, DTT (1 mM final concentration), CaCl 2 (1 nM to 5 mM final concentration), Triton X-100 (0.1% w/v final concentration) mixed micelles containing 15 mol % brain phosphatidylserine and 5 mol % diacylglycerol and proteases (concentrations and incubation times as indicated in the figure legends). Proteins were separated by SDS-PAGE on 7 or 9% polyacrylamide gels and visualized by silver staining, or in some cases, proteins were electrophoretically transferred to nitrocellulose and labeled with antibodies to protein kinase C via incubation with alkaline phosphatase-conjugated IgG and detection by the formation of the insoluble product of 5-bromo-4chloroindoyl phosphate hydrolysis.
Data Analysis-The dependence of protein kinase C binding or activity on the lipid composition of vesicles (or micelles) or the Ca 2ϩ concentration was analyzed by a nonlinear least-squares fit to a modified Hill equation as described (20). The apparent association constant, K a , for binding of protein kinase C to vesicles was calculated as the fraction of protein kinase C bound to membranes divided by the product of the fraction of protein kinase C remaining in the supernatant and the total lipid concentration (19).
Free [Ca 2ϩ ] Determinations-Concentrations of free Ca 2ϩ were calculated using a computer program kindly provided by Dr. Claude Klee (23) that takes into account pH, Ca 2ϩ , Mg 2ϩ , K ϩ , Na ϩ , EGTA, EDTA, and ATP concentrations. Binding constants used in this program were those given by Fabiato and Fabiato (24).

Ca 2ϩ Dependence for Membrane Binding Detected by Two
Independent Assays-The Ca 2ϩ dependence for the binding of protein kinase C ␤II to large unilamellar vesicles (100 M lipid) containing phosphatidylserine (40 mol %) and diacylglycerol (5 mol %) is presented in Fig. 1. Binding was detected three ways, using two independent assays, to ensure that the measured kinetics reflected properties of the protein-cofactor interaction rather than the assay system. One assay was conformationbased (E); it detects the marked increase in proteolytic sensitivity of protein kinase C's hinge region that accompanies membrane binding (approximately 100-fold increase for protein kinase C ␤II) (22). The second assay involved physical separation of membrane-bound protein kinase C from free protein kinase C by centrifugation (19). In the latter, the amount of membrane-bound protein kinase C was determined either from Western blot analysis (Ⅺ) or activity assays using the cofactorindependent substrate protamine (‚). The only compositional difference between the two assays was the presence of trypsin in the conformation-based assay. The centrifugation assay has the advantage of being a direct measure of the membrane association, whereas the conformation-based assay has the advantage of not involving separation of free enzyme from vesicles and thus avoids potential perturbation of the equilibrium between the bound and free enzyme.
Importantly, the centrifugation assay (Ⅺ, ‚) and conformation-based assay (E) yielded similar results; under the conditions of the experiment, half-maximal binding to vesicles was mediated by approximately 300 nM Ca 2ϩ . At this free Ca 2ϩ concentration, the enzyme's apparent membrane affinity was 10 4 M Ϫ1 . This affinity is comparable to that reported for the interaction of protein kinase C from rat brain with vesicles containing slightly more phosphatidylserine (50 mol %) but less diacylglycerol (1 mol %) in the presence of 1 M Ca 2ϩ (11).
Ca 2ϩ Regulation of Three Protein Kinase C Isozymes' Membrane Affinity and Activation- Fig. 2A compares the Ca 2ϩ dependence for membrane binding of three conventional protein kinase Cs as follows: ␣ (E), ␤II (Ⅺ), and the alternatively spliced ␤I (‚) which differs from ␤II in the last 50 residues at the carboxyl terminus. The Ca 2ϩ requirement for membrane binding differed modestly for these three isozymes; under the conditions of the assay, the concentrations of Ca 2ϩ mediating half-maximal binding ranged from 0.3 to 1.0 M for the three isozymes (Table I). Note that the concentration of Ca 2ϩ resulting in half-maximal membrane binding for ␤II was approximately 3-fold higher in Fig. 2A (and Table I) than that in Fig.  1; this resulted from the presence of 5 mM Mg 2ϩ in the assay in Fig. 2, which decreases the membrane surface potential and hence protein kinase C's membrane affinity (19). Importantly, the apparent Ca 2ϩ affinity of protein kinase C ␤I was 3-fold higher than the apparent Ca 2ϩ affinity of protein kinase C ␤II, despite the two having identical Ca 2ϩ -binding sites in the C2 domain. Furthermore, protein kinase C ␤I displayed a significantly higher level of Ca 2ϩ -independent membrane binding than the other two isozymes: 40% of the protein kinase C ␤I associated with vesicles in the absence of Ca 2ϩ , compared with 15 and 10% for protein kinase C ␣ and ␤II, respectively. These results reveal that the very carboxyl-terminal residues of protein kinase C influence its Ca 2ϩ -dependent membrane interaction.
The Ca 2ϩ requirement for activation differed significantly among isozymes, and, for two isozymes, this requirement was markedly different from the Ca 2ϩ requirement for membrane binding (Fig. 2B). The Ca 2ϩ dependence for both activation and binding was similar for protein kinase C ␣; however, both protein kinase Cs' ␤I and ␤II required over an order of magni-tude higher concentrations of Ca 2ϩ for half-maximal activation compared with that required for half-maximal membrane binding (Table I). Conditions for the membrane binding and activity assays were identical except that bovine serum albumin (0.3 mg ml Ϫ1 ) was included in the binding assay (its presence in activity assays did not affect results (data not shown)) and ATP and peptide substrate were present in the activity assay (their presence in binding assays does not significantly affect protein kinase Cs' interaction with vesicles 2 ). Most strikingly, the alternatively spliced ␤ isozymes displayed a difference in their Ca 2ϩ requirements for activation that was even greater than the difference in their requirements for membrane binding; the apparent K m for Ca 2ϩ (for activity) was approximately 10 ϫ higher for protein kinase C ␤II compared with protein kinase C ␤I.
The Ca 2ϩ dependence for activation was found to be sensitive to the total lipid concentration. Analysis of the data in Fig. 3 revealed that increasing the lipid concentration 10-fold resulted in a 10-fold drop in the apparent K m (from approximately 30 M in the presence of 100 M lipid to approximately 3 M in the presence of 1 mM lipid). This is consistent with binding studies by Mosior and Epand (19) showing that the interaction of protein kinase C with membranes obeys the mass action law; altering either the total lipid concentration or the total enzyme concentration affects the amount of protein kinase C that partitions with the membrane. Thus, reported K m values for Ca 2ϩ for protein kinase C are specific to particular assay conditions. Ca 2ϩ -induced Conformational Changes-The finding that protein kinase C requires different Ca 2ϩ concentrations for membrane binding and activation led us to explore whether distinct conformational changes accompany the Ca 2ϩ -dependent membrane interaction and the Ca 2ϩ -dependent activation. In a first set of experiments, we examined whether the proteolytic sensitivity of protein kinase C ␤II was altered with increasing Ca 2ϩ concentrations when in solution or when membrane-bound. Fig. 4 shows that the trypsin sensitivity of protein kinase C ␤II, in the absence of lipid cofactors, was the same in the presence of 100 nM to 5 mM Ca 2ϩ . Under the conditions of the assay, the enzyme was partially proteolyzed to generate the catalytic domain (approximately 45 kDa) and regulatory half (approximately 35 kDa; does not stain well by silver) (lanes 2-9). The similar degree of cleavage in lanes 2-9 reveals that the intrinsic activity of trypsin is not affected by altering the Ca 2ϩ concentration between 100 nM and 5 mM Ca 2ϩ (also established using BAEE as substrate, not shown).
In contrast to the cleavage in the absence of lipid, incubation of protein kinase C ␤II with phosphatidylserine:diacylglycerol mixed micelles resulted in pronounced differences in the enzyme's proteolytic sensitivity as the Ca 2ϩ concentration was raised. Two sets of changes in proteolytic rates or proteolytic fragments produced were observed. First, at Ca 2ϩ concentrations sufficient to effect membrane binding, but not activation, the enzyme was markedly more sensitive to proteolysis at the hinge region as assessed by the increased generation of the catalytic domain (compare lane 11 (100 nM Ca 2ϩ ; no binding) with lanes 12 and 13 (1 and 10 M Ca 2ϩ ; under the conditions of this assay, half-maximal binding required 1 M Ca 2ϩ )). In the presence of Ca 2ϩ concentrations too low to effect membrane binding (100 nM Ca 2ϩ ), the enzyme was proteolyzed at a similar rate as in the absence of lipid (note that the trypsin concentration in lanes 11-18 (plus lipid) was 10-fold lower than in lanes 2-9 (no lipid)). This change in proteolytic sensitivity is the well-characterized exposure of the hinge region upon mem-2 M. Mosior and A. C. Newton, unpublished results. brane binding (22). In addition, a novel 70-kDa intermediate (asterisk) became more pronounced in lanes where protein kinase C was bound to lipid but not active (e.g. lanes 12 and 13); a slightly faster migrating band (68 kDa) was barely visible below this band. A second set of changes in proteolytic sensitivity occurred as the Ca 2ϩ concentration was raised to activating levels (e.g. above 500 M Ca 2ϩ ); the 70-kDa intermediate became less apparent and a 78-kDa intermediate (arrow) became apparent. Qualitatively similar results were obtained whether protein kinase C was incubated with mixed micelles (Fig. 4) or vesicles (data not shown).

Antibodies Detect Cleavage at Carboxyl or Amino Terminus
of Protein Kinase C-To better understand the Ca 2ϩ -induced changes in protein kinase C's structure, a series of polyclonal antibodies generated against different determinants on protein kinase C were used to identify proteolytic fragments. Fig. 5A characterizes the epitope of each antibody. Protein kinase C ␤II was proteolyzed so that the native protein (80 kDa) and 78-(arrow), 70-(asterisk), and 68-kDa intermediates were trapped (the 78-and 68-kDa forms are more apparent when endoproteinase Arg-C is used; all these fragments co-migrate with the bands generated by trypsin cleavage). Samples were analyzed by SDS-PAGE and stained with silver (panel 1) or labeled with four antibodies (panels 2-5). The 78-kDa fragment (arrow) has been previously characterized and results from activation-dependent cleavage at Arg-19 in the pseudosubstrate of protein kinase C ␤II (13). This fragment was recognized by all antibodies except one generated against the bacterially expressed regulatory domain of protein kinase C: thus, the epitope for this antibody (anti-regulatory) lies in the first 18 residues of protein kinase C ␤II. The 70-kDa intermediate (asterisk) was recognized by all antibodies except an antibody generated against the carboxyl terminus of protein kinase C ␤II (residues 645-673); this revealed that the 70-kDa form was proteolyzed at the carboxyl terminus. Antibodies generated against bacterially expressed catalytic domain (anti-catalytic) and the peptide sequence from the pseudosubstrate of protein kinase C ␤II (residues 19 -32; anti-pseudosubstrate) recognized native and amino-or carboxyl-terminally cleaved protein kinase C equally well. Neither the anti-regulatory or anti-carboxyl-terminal domain antibodies labeled the 68-kDa fragment, indicating that it was truncated at both termini. Thus, one antibody recognized the first 18 residues of protein kinase C ␤II (anti-regulatory), one recognized the pseudosubstrate of protein kinase C (antipseudosubstrate), one recognized the catalytic domain (anticatalytic), and one recognized the extreme carboxyl terminus of the protein (anti-carboxyl terminus).
Ca 2ϩ Binding Induces Three Structural Changes in Protein Kinase C-The four antibodies, with epitopes spanning the length of protein kinase C, were used to characterize the tryptic fragments described in Fig. 4. Protein kinase C was partially proteolyzed by trypsin under three conditions, shown in   The proteolytic pattern resulting from inactivating and activating Ca 2ϩ concentrations was explored in more detail in the time courses of trypsinolysis in Fig. 6A. Protein kinase C was incubated with 1 nM Ca 2ϩ and no lipid (soluble conformation), with 1 M Ca 2ϩ and phosphatidylserine:diacylglycerol mixed micelles (lipid-bound but inactive), or with 200 M Ca 2ϩ and phosphatidylserine:diacylglycerol mixed micelles (lipid-bound and active) and treated with trypsin for 0 -30 min. Under the conditions of the assay, soluble protein kinase C was relatively resistant to proteolysis, whereas the lipid-bound enzyme was markedly sensitive to proteolysis at the hinge to generate the catalytic domain (apparent in Fig. 6A starting at 1 min time point in the presence of 1 and 200 M Ca 2ϩ ). Importantly, the rate of disappearance of the native (80 kDa) protein kinase C was similar in the presence of 1 and 200 M Ca 2ϩ , consistent with the hinge exposure reflecting the lipid-bound conformation of protein kinase C and being independent of the active state of the enzyme. The 70-kDa fragment resulting from carboxyl-terminal cleavage was specific to the incubation with 1 M Ca 2ϩ ; in addition, a truncated catalytic domain was apparent (this catalytic domain fragment was proteolyzed at the carboxyl terminus because it was not labeled with anti-carboxyl-terminal antibodies (data not shown)). Cleavage at the pseudosubstrate (N-nicked PKC in Fig. 6A) but not at the carboxyl terminus characterized the proteolytic pattern in the presence of activating Ca 2ϩ concentrations (lane 14). At this activating Ca 2ϩ concentration, the cleaved catalytic domain was considerably more stable than at the non-activating concentration (after 5 min of proteolysis in the presence of 1 M Ca 2ϩ , the catalytic core is almost completely proteolyzed at the  Hinge Conformational Change Is Highly Localized- Fig. 6B shows the sensitivity of protein kinase C to endoproteinase Lys-C for soluble, lipid-bound but inactive, and lipid-bound and active protein kinase C. In marked contrast to the cleavage by trypsin, Lys-C did not discriminate between the different forms of protein kinase C; proteolysis at the hinge to generate the catalytic and regulatory domains occurred at the same rate for all three conditions. To localize the site of the conformational change in the hinge, the catalytic domains generated by trypsin or Lys-C cleavage were subjected to amino-terminal sequencing. Fig. 6C shows that trypsin cleaves at two sites, Arg-307 and Lys-309, which are amino-terminal to the single Lys-C site at Lys-320. Thus, the region around residue 320 is equally accessible in all three conformations of protein kinase C (soluble, membrane-bound with or without activity), whereas a region removed by only a dozen residues is masked in the soluble conformation and undergoes a dramatic increase in exposure upon membrane binding.
Other Proteases as Conformational Probes- Table II summarizes the results of treating protein kinase C with a number of proteases. Proteases fell into three classes as follows: those such as elastase, trypsin, and to a lesser extent chymotrypsin and thermolysin that preferentially cleaved the membranebound conformation of protein kinase C at the hinge; those such as endoproteinase Glu-C and endoproteinase Lys-C that cleaved at the hinge independently of the membrane-bound conformation; and those such as endoproteinase Arg-C and endoproteinase Asp-N that did not cleave significantly at the hinge. Trypsin, elastase, and endoproteinase Arg-C were unique in being able to detect Ca 2ϩ -induced conformational changes occurring upon activation of membrane-bound protein kinase C. Curiously, Mg 2ϩ inhibited the proteolysis at the hinge catalyzed by proteases that detected the hinge conformational change attendant to membrane binding (chymotrypsin, thermolysin, and trypsin) but did not inhibit the hinge proteolysis catalyzed by enzymes that did not discriminate between soluble and membrane-bound protein kinase C (endoproteinases Glu-C and Lys-C). Mg 2ϩ did not inhibit the intrinsic activity of trypsin, assessed using BAEE as a substrate (data not shown), suggesting that Mg 2ϩ may be interacting with the protein in such a manner to alter the hinge accessibility. DISCUSSION The foregoing data establish that 1) Ca 2ϩ -dependent membrane binding and activation of conventional protein kinase Cs are differentially regulated by Ca 2ϩ , and 2) distinct conformational changes accompany these two events for the isozyme examined, protein kinase C ␤II. Kinetic studies reveal that the Ca 2ϩ requirements for membrane binding and activation of conventional protein kinase Cs differ, with activation lagging behind membrane binding. Proteolytic sensitivity experiments with protein kinase C ␤II reveal that one set of conformational changes accompanies the membrane binding mediated by Ca 2ϩ concentrations too low to activate the enzyme, and a second set of conformational changes accompanies activation resulting from higher Ca 2ϩ concentrations.
Ca 2ϩ Differentially Regulates Membrane Binding and Activation of Protein Kinase C-Kinetic analyses reveal that higher concentrations of Ca 2ϩ are required to effect activation of pro-tein kinase C compared with membrane binding. This difference is particularly striking for protein kinase C ␤II, where approximately 40-fold greater Ca 2ϩ concentrations are required to half-maximally activate the enzyme compared with the concentrations promoting half-maximal binding to membranes. This difference is only 2-fold for protein kinase C ␣ and approximately 10-fold for protein kinase C ␤I. Importantly, the Ca 2ϩ requirement for binding is similar, within a factor of 3, for protein kinase C ␣, ␤I, and ␤II; what differs most significantly between the isozymes is the Ca 2ϩ requirement for activation. Thus, all three isozymes bind to membranes with similar affinity; however, some isozymes require Ca 2ϩ concentrations that cause a much tighter binding to be catalytically competent. One explanation for the different Ca 2ϩ requirements for membrane binding and activation is that protein kinase C binds two Ca 2ϩ ions with differing affinity and different consequences on the structure (see below) and function of the protein; in this regard, the Ca 2ϩ site in the C2 domains of synaptotagmin and phospholipase C␦ have been shown to accommodate two metal ions (16,17).
Activation of protein kinase C requires removal of a basic autoinhibitory domain, the pseudosubstrate, from the substrate-binding cavity in the kinase core (3). One possibility to account for the discrepancy in Ca 2ϩ concentrations required for binding and activation is that the tighter membrane binding induced by higher Ca 2ϩ concentrations provides the energy to release the pseudosubstrate. For example, an extra kcal mol Ϫ1 in binding energy is gained by a 10-fold increase in binding affinity. Alternatively, tighter membrane binding may position protein kinase C closer to the membrane, perhaps promoting domain movements that might facilitate removal of the pseudosubstrate (for example, interaction with a basic surface in the regulatory region could neutralize the patch of acidic residues on the surface of the substrate-binding cavity that holds the pseudosubstrate in place (26)).
Modulation of Protein Kinase C's Ca 2ϩ Affinity by Carboxylterminal Interactions-An unexpected finding from this work is that the Ca 2ϩ requirements for activation and, to a lesser extent, membrane binding differ between the two alternatively spliced protein kinase C ␤ isozymes. Notably, these isozymes have identical regulatory domains and differ only in the kinase domain; the carboxyl-terminal 52 residues of ␤II are replaced by 50 residues in protein kinase C ␤I, 24 of which are identical to those in protein kinase C ␤II (25). Thus, the Ca 2ϩ -binding site in the C2 domain must be influenced by interactions with the carboxyl terminus of the kinase core. The carboxyl terminus has recently been shown to be critical to protein kinase C's regulation: phosphorylation at this region influences the enzyme's subcellular localization (27,28); this region has been reported to promote the binding of protein kinase C ␤II but not ␤I to actin (29), and deletion of this region prevents expression of mature, active protein kinase C (30). The finding that the Ca 2ϩ requirement for activation is significantly greater than that for binding in protein kinase C ␤II compared with ␤I suggests that this carboxyl-terminal interaction may be more critical in allowing catalysis than in modulating membrane binding.
A second line of evidence supports the regulation of protein kinase C's Ca 2ϩ interaction by carboxyl-terminal interactions. All protein kinase C isozymes contain two conserved phosphorylation sites on their carboxyl terminus (27); we recently showed that phosphorylation at one of these positions (Ser-660 in protein kinase C ␤II) increases protein kinase C's affinity for Ca 2ϩ by an order of magnitude (28). Whether the carboxyl terminus interacts directly with the C2 domain, perhaps providing coordination sites for Ca 2ϩ , or whether it indirectly modulates the conformation of the Ca 2ϩ -binding site remains to be established.
Distinct Conformational Changes Accompany Ca 2ϩ -dependent Membrane Binding and Activation-Using proteases as conformational probes, three regions of protein kinase C ␤II were identified that undergo structural changes representative of membrane binding and of activation: 1) the hinge, which becomes exposed whenever protein kinase C binds lipid, independently of the active conformation, 2) the carboxyl terminus, which becomes exposed only when protein kinase C is bound to lipid in its inactive conformation, and 3) the pseudosubstrate, which becomes exposed when protein kinase C is active.
The increased proteolytic sensitivity of the hinge accompanying membrane binding was originally noted by Nishizuka and co-workers (31) and forms the basis for the membrane binding assay (22) illustrated in Fig. 1. An unexpected finding from the present contribution is that the region in the hinge that becomes proteolytically labile upon membrane binding is highly localized; Arg-307 and Lys-309 are protected from proteolysis by trypsin when protein kinase C is in solution and become 2 orders of magnitude more sensitive to proteolysis upon membrane binding. In sharp contrast, Lys-320 is equally accessible to proteolysis by endoproteinase Lys-C whether protein kinase C is soluble or membrane-bound. This finding suggests a highly localized "hinge" motion, as depicted in Fig. 7. The increased proteolytic sensitivity of membrane-bound protein kinase C to elastase, chymotrypsin, and thermolysin sug-gests that the sites of cleavage of these proteases are localized near the trypsin site. An alternative explanation is that these proteases recognize interaction sites elsewhere on protein kinase C that are better exposed when protein kinase C is membrane-bound, so that proteolysis of the hinge is accelerated without sequences in this region being more exposed.
Importantly, the hinge conformational change does not result in activation of protein kinase C; membrane-bound but inactive protein kinase C has an exposed hinge, and this exposure is not influenced by the activation induced by increasing Ca 2ϩ concentrations. Consistent with this, binding of protein kinase C to non-activating acidic membranes (e.g. containing phosphatidylserine but no diacylglycerol or containing acidic phospholipids other than phosphatidylserine (9)) also is accompanied by exposure of the hinge (7). Conversely, the active conformation of protein kinase C does not require hinge exposure; we showed previously that cofactor-independent activation by protamine sulfate does not expose the hinge of protein kinase C (26). Thus, the hinge exposure results from protein kinase C's interaction with phospholipid, independently of the activation state of the kinase. One possible mechanism is that tethering of protein kinase C to acidic membranes via the C2 domain promotes the hinge conformational change; this domain, which contains the binding sites for Ca 2ϩ and acidic phospholipids, is adjacent to the hinge region (3).
The exposure of the carboxyl terminus to proteolysis is particularly interesting in that it occurs only when protein kinase C is membrane-bound but inactive; i.e. concentrations of Ca 2ϩ that promote membrane binding but not activation result in exposure of the carboxyl terminus, but higher concentrations of Ca 2ϩ that promote activation result in masking of the carboxyl terminus to proteolysis. One possibility is that activation promotes a domain movement that masks the carboxyl terminus, for example release of the pseudosubstrate from the active site might mask the carboxyl-terminal cleavage site. A second is that tight membrane binding of protein kinase C prevents access of trypsin to the carboxyl terminus. A third possibility is that the carboxyl terminus is a flexible region of the protein that transiently "opens up" as protein kinase C makes its initial low affinity membrane interaction (a conformation that can be trapped by having non-activating Ca 2ϩ concentrations present in in vitro assays) but closes up again when the enzyme binds tightly to membranes and adopts the active conforma-FIG. 7. Model for possible Ca 2؉ -dependent conformational changes that occur upon membrane binding and activation. Soluble, inactive protein kinase C with its amino-terminal pseudosubstrate occupying the active site of the catalytic domain is relatively resistant to proteolysis (left panel). Low concentrations of Ca 2ϩ promote a weak membrane interaction, but not activation, that is accompanied by two conformational changes as follows: exposure of a highly localized region in the hinge separating the kinase domain from the regulatory half of the protein, and exposure of a domain approximately 10-kDa from the carboxyl terminus; both conformational changes are detected by the increased sensitivity to trypsin (indicated T; middle panel). Higher concentrations of Ca 2ϩ result in activation, concomitant with two additional conformational changes as follows: the carboxyl terminus becomes masked again, and the amino-terminal pseudosubstrate becomes exposed (right panel). The endoproteinase Lys-C site in the hinge (indicated L) is equally exposed in all three conformations of protein kinase C. Membrane attachment is mediated by phosphatidylserine (PS) binding to the C2 domain and diacylglycerol (DG) binding to the C1 domain. The carboxyl terminus, which modulates Ca 2ϩ binding, is depicted interfacing with the C2 domain; alternatively, the carboxyl terminus could indirectly influence the conformation of the Ca 2ϩ -binding site in the C2 domain.
tion. Several lines of evidence suggest that the carboxyl terminus of protein kinase C is flexible. Modeling studies of protein kinase C's carboxyl terminus, based on the crystal structure of protein kinase A, suggest that it is a random coil which, by analogy to protein kinase A, wraps around the top of the kinase domain (26). However, the ability to autophosphorylate at residues in the carboxyl terminus suggests a flexible coil that can access the active site (27,32). The proteolytic sensitivity is also consistent with a flexible loop or hinge.
The third conformational change detected by proteases involves the activation-dependent exposure of the pseudosubstrate (13). We have previously shown that Arg-19, the first residue in the pseudosubstrate of protein kinase C ␤II, becomes sensitive to proteolysis only when protein kinase C is active and that this exposure is independent of how protein kinase C is activated; activation by its second messengers, by cofactorindependent activators such as protamine sulfate or by short chained phosphatidylcholines, is accompanied by pseudosubstrate exposure (13,26). In this contribution, we show that membrane binding alone is not sufficient for pseudosubstrate release and that activating concentrations of Ca 2ϩ are required to release the pseudosubstrate.
Conclusions- Fig. 7 presents a possible model illustrating the low affinity membrane interaction that does not result in activation, and the high affinity membrane interaction that does result in activation. Potential Ca 2ϩ -dependent conformational changes that accompany each interaction are also illustrated and explained in the legend. This model assumes that the increased proteolytic sensitivity of defined regions of protein kinase C that accompany membrane binding and activation result from conformational changes. As discussed above, protease accessibility could be affected as a result of other interactions. For example, juxtaposition of protein kinase C with the membrane could prevent trypsin from accessing the carboxyl terminus.
In summary, these data reveal that Ca 2ϩ -dependent membrane binding is not sufficient to activate protein kinase C and that the discrepancy between Ca 2ϩ requirements for membrane binding and activation varies depending on isoform, and, significantly, varies considerably between two isozymes that differ only in the carboxyl terminus. In addition, membrane binding and activation of protein kinase C ␤II appear to be accompanied by two distinct sets of conformational changes involving the amino terminus, the hinge, and the carboxyl terminus. The movement of three separate domains spanning the entire protein suggests complex interplay between the membrane-targeting domains, the kinase core, and the car-boxyl terminus. The finding that Ca 2ϩ differentially regulates protein kinase C ␤I and ␤II reveals, in particular, that the carboxyl terminus modulates Ca 2ϩ binding; the carboxyl terminus may thus provide a mechanism for differential Ca 2ϩ regulation of two very closely related isozymes that are often both present in the same cell type.