Characterization of the Calcium-binding Site That Regulates Association of Protein Kinase C with Phospholipid Bilayers*

Transcription

1 "I ". THE JORNAL OF BIOLOOICAL CHEMISTRY by The Amerian Soiety for Biohemistry and Moleular Biology, In. Vol. 269, No. 19, Issue of May 13, pp , 1994 Printed in.s.a. Charaterization of the Calium-binding Site That Regulates Assoiation of Protein Kinase C with Phospholipid Bilayers* Marian MosiorS and Rihard M. Epands From the Department of Biohemistry, MMaster niversity, Hamilton, Ontario L8N 325, Canada (Reeived for publiation, Deember 27, 1993) Assoiation of alium-dependent isotypes of protein Ca2+-binding site involved the regulation of the interation of kinase C (PKC) with a phospholipid bilayer is regulated the enzyme with its substrates (5). by a single Caz+-binding site. The dependene of PKC In this study, we haraterized the Ca2+-binding site on PKC assoiation with phosphatidylserine-ontaining mem- that regulates the assoiation of the enzyme with the membranes on the onentration of Ca2+ is linear in the sub- brane. The dependene of the enzyme affinity for the memmiro- to submillimolar range. The Ca2+-regulated asso- brane on the onentration of Ca2+ revealed the funtional stoiiation of PKC with the membrane is sensitive to the hiometry of the PKCCa2+ omplex as well as the value of the fators that alter the diffuse double-layer potential pro- dissoiation onstant. dued by anioni lipids suh as phosphatidylserine (PS). We have earlier hypothesized that the inhibiting effet of This indiates that the Ca2+-binding site on the mem- Mg2+ on the affinity of PKC for the phospholipid bilayer may brane-bound enzyme senses a higher onentration of indiate that this Ca2+-binding site is loated in diret proxim- Ca2+ than is present in bulk solution. This is a onseity of the plane of the membrane (6). MLaughlin and Whitaker quene of the aumulation of Ca2+ in the layer adjaent to the plane of the membrane by the double-layer poten- (7) demonstrated that Ca2+-ontrolled exoytosis in sea urhin tial.calulationsbased on the Gouy-Chapman-Stern eggs is regulated by the magnitude of the double-layer potentheory of the diffuse double layer yielded a unique value tial produed by anioni lipids and membrane-bound ations. of the Ca2+ dissoiation onstant for the Ca2+-PKC-bi- They onluded therefore that the exoytosis is affeted by the layer omplex equal to -700 m. The soluble form of the nonspeifi aumulation of Ca2+ within the double layer. We enzyme has a 3.5 order of magnitude lower affinity for adopted their approah to pursue the same question for mem- Ca2+. The free energy of interation between the Ca2+- brane-bound PKC. The double-layer potential produed by anand PS-binding sites is large (-5 kallmol). In ontrast, ioni lipids or other lipid-binding as well as membrane-partithe interation between the diaylglyerol-binding site tioning ions is aurately desribed by the Gouy-Chapmanand either the Ca2+- or PS-binding site appears to be Stern theory (for reviews, see Refs. 8 and 9). A detailed weak. quantitative analysis of the binding of metalli ations to phospholipid bilayers omposed of PS and/or PC, published by MLaughlin et al. (7, 10, 111, provided us with all of the data Protein kinase C (PKC)' is one of the key enzymes in the required for the alulation of the Ca2+ distribution in the proxsignal transdution pathway for many ellular proesses. At imity of the PSPC bilayers, whih in turn was used to deterleast four isotypes of PKC (designated a, PI, PII, and y) require mine the loation of the Ca2+-binding site with regard to the Ca2+ for the ativation of the enzyme (1, 2). plane of the membrane. This is due to the minusule volume of The loation of the speifi binding site(s) for Ca2+ has not the double layer and the buffering apaity of EGTA. Finally, been determined, although it is generally aepted that it is we also investigated the degree of interation between PKC ontained within the seond onserved region of the regulatory binding sites for PS, diaylglyerol, and Ca". domain of PKC (1-3). The stoihiometry of the PKCCa2+ om- EXPERIMENTALPROCEDRES plex and the funtion of Ca2+-binding site(s) are not fully re- Materials-All phospholipids and 1,2-dioleoyl-sn-glyerol (DG) were solved. Membrane-bound PKC has been shown to assoiate supplied by Avanti Polar Lipids, In. (Alabaster, AL). Only DG showed with 8-10 Ca2+ ions. This assoiation has been postulated to a trae amount of ontaminant that was identified as 1,3-dioleoylglyour at the protein-lipid interfae and to ontrol the binding of erol (TLC; 19:l hloroformlaetone). Histone 111-S, protamine sulfate, Ca2+-dependent isotypes of PKC to the phospholipid bilayer (4). ATP sodium salt, EGTA, and D-sphingosine were purhased from The a (but not P) isotype seems to ontain a single high affinity Sigma. The syntheti peptide Val-Arg-Lys-Arg-Thr-Leu-Arg-Arg-Leu (VRKRTLRRL) was obtained from Bahem California (Torrane, CAI. The radiolabeled ompounds [Y-~~PIATP and [9,10-3Hldipalmitoylphos- * This work was supported by Medial Researh Counil of Canada phatidylholine were supplied by NEN (Montreal, QuBbe, Canada). Grant MT The osts of publiation of this artile were defrayed in Bovine serum albumin (fration V, fatty aid-free; Boehringer Mannpart by the payment of page harges. This artile must therefore be heim Canada, Laval, Quebe, Canada), Tris (ultrapure; Life Tehnolohereby marked "advertisement" in aordane with 18.S.C. Setion gies, In.), and all salts (Fisher) were purhased in a form that provided 1734 solely to indiate this fat. the atual lot analysis, indiating in partiular the amount of ontamii. Present address: DeDt. of Chemistrv. Indiana niversitv, Bloomine- ~~~~~ I nating alium. ton, IN Protein Kinase C Purifiation-Rat brain PKC was purified by a 8 To whom orrespondene should be addressed: Dept. of Biohemistry, MMaster nivkrsity, 1200 Main St. West, Hamilton, Ontario L8N modifiation of the proedure of Huanget al. (12) as desribed elsewhere (6). 325, Canada. Tel.: (ext ); Fax: ;E- Purified PKC displayed a single band on the silver-stained eletromail (Internet): Epand@fhs.su.MMaster.CA. phoresis gel. The speifi ativity of the enzyme for histone in a miellar The abbreviations used are: PKC, protein kinase C; PS, phosphati- assay (13) was 1-2pmolmg"min". The phospholipid-independent dylserine; PC, phosphatidylholine; DG, 1,2-dioleoyl-sn-glyerol; BSA, ativity in this assay did not exeed 2% of the total kinase ativity. bovine serum albumin; dansyl, 5-dimethylaminonaphthalene-1-sul- PKC Binding Assays-The surose-loaded vesile assay was adopted fonyl; POPS, 1-palmitoyl-2-oleoylphosphatidylserine; POPC, l-palmi- from the proedure of Rebehi et al. (14) with minor modifiations toyl-2-oleoylphosphatidylholine. desribed elsewhere (6). Membrane-bound enzyme was separated by 13798

2 Ca2+-dependent Assoiation of PKC with Membranes entrifugation at 100,000 x g for 30 min at 25 "C. Both the supernatant and the pellet were assayed under idential onditions for kinase ativity with protamine sulfate as desribed earlier (6). The vesile-assoiated kinase ativity (A,) was alulated aording to Equation 1: PA* + (P - 1) A, A, = r+p - 1 (Eq. 1) where A, and A, are the measured ativities of the bottom and top frations, respetively. The fration of sedimented vesiles (a) was alulated from the distribution of 3H-labeled PC, whih was inluded in trae amounts in all lipid mixtures. The fration of kinase ativity found in the supernatant in the absene of lipid (P) was onsistent with the lak of autosedimentation of the enzyme. All experiments were arried out in buffer omposed of 100 mm KCl, 20 mm Tris-HC1, ph 7.0, and 0.3 mg/ml BSA. Conentrations of other omponents that varied between experiments are given below. The fluoresene energy transfer measurements followed essentially the published proedure (15). A mixture of large unilamellar vesiles and PKC in a 1:5 (w/w) ratio was suspended in standard buffer at a final volume of 195 pl. The buffer also ontained 25 JIM EGTA. The emission from dansyl-labeled PE was measured at 496 nm before and after the addition of 5 pl of onentrated CaCl, solution, whih brought the onentration of free Ca2+ to the desired level. The exitation wavelength was set at 280 nm. The hange in the emission of the dansyl group, measured relative to the emission of the sample titrated with 5 pl of Ca2+-free buffer, refleted the hange in the amount of PKC assoiated with the lipid bilayer. PKC Ativity Assay-The ativity assay with histone 111-S, protamine sulfate, or the peptide VRKRTLRRL followed published proedures (16, 17). The onentration of histone and protamine sulfate was 0.2 mg/ml eah. The onentration of the peptide was 10 JIM. The reation was initiated by the addition of 40 p~ [y3'p1atp (0.2 pci/ml) and was terminated after 10 min at 30 "C by the addition of 2 ml of ie-old 25% (w/v) trihloroaeti aid or 5% (v/v) aeti aid. The samples were filtered through either Whatman GF/C or ion-exhange P-81 paper. The filters were then subsequently washed five times with 2 mlof aid solution, dried, and ounted for effiieny-orreted Cerenkov radiation. Calulation of Free Ca2+ Conentration-The total onentration of Ca2+ in stok solutions was measured by atomi absorption spetrosopy. The alulation of free Ca2+ onentration followed the established protool (181, taking into aount ontamination of all solution omponents by Ca", the effet of the ioni strength, and the interation of EGTA with Mg2' (when present). Calulation of Ca2+ Conentration in Proximity of Charged Mem- branes-the alulations, based on the equations and assumptions of the Gouy-Chapman-Stern theory of the double-layer potential (8, 9), followed the work of MLaughlin et al. (7, 10, 11). The dependene between the onentration of Ca2+ at a given distane from the membrane and the eletrostati potential at this point was desribed by the Boltzman relation. The profile of the eletrostati potential along the axis perpendiular to the plane of a membrane was desribed by the nonlinear Poisson-Boltzmann equation with the standard boundary onditions of the Gouy-Chapman-Stern theory (8, 9). Following the work of MLaughlin et al. (7, ll), we assumed that eah phospholipid, either PC or PS, oupied a surfae area of 0.7 nm2 and that the resulting eletrostati potential ould be alulated from the surfae harge density by means of the Gouy-Chapman-Stern theory (8,9). The density of bound ations was alulated using their intrinsi assoiation on- stants with either PC (7, 10) or PS (ll), taking into aount their nonspeifi aumulation by the double-layer potential in the layer immediately adjaent to the membrane surfae. The equations were solved numerially. RESLTS To quantify the assoiation of PKC with a phospholipid bilayer, we use the apparent assoiation onstant, Ka, defined as the ratio of membrane-bound to free enzyme divided by the total lipid onentration.' Two effets beome immediately apparent when K, is plotted against the onentration of free Ca". First, this dependene is linear over a wide range of Division by the total onentration of the protein-aessible lipid seemed to be more appropriate. Although approximately half of the total lipid is usually found in the inner leaflet of the large unilamellar vesiles of 0.1-pm diameter, we did not experimentally verify the atual lipid distribution. m 0 +l = nl - I k io" Cl Cl a Log fica2+i) IM1 FIG. 1. Dependene of apparent affinity of PKC for membrane with diolein on onentration of free Cas+. The apparent assoiation onstant was alulated from the binding results obtained in the surose-loaded vesile assay as desribed under "Experimental Proe- dures.'' Lipid vesiles were omposed of 15 mol % POPS, 1 mol % diolein, and POPC as a remainder. The total lipid onentration was in the range of0.01-2mm.all solutions used in the experiment ontained 100 mm KC1,0.500 mm CaCl,, and 0.3 mg/ml BSAand were buffered by 20 mm Tris-HC1 to ph 7.0. This omposition of solution was modified by the addition of 5 m~ MgC1, (O), 5 mm MnCI, (A), or 200 m~ KC1 ( +) or served as a referene without any further additions (H). The onentration of free Ca2+ was altered by the addition of EGTA and was alulated as desribed under "Experimental Proedures." When MnCl was present, no EGTA was added, and the free onentration of Ca'+ was assumed to be equal to the total onentration of Ca2+. The onentration of PKC was in the range n ~ Points. represent an average of at least two independent experiments, eah in tripliate. Bars, shown when larger than the symbol size, indiate standard deviation. Curves are graphi representations of the model of assoiation of PKC with the membrane (see the "Appendix"). The assoiation onstants were as fol- lows: Kl = 15 M-~ = 3.5 x lo6 M-~ (dotted urue); and K3 = 350 M-~ for both urves. t and K, = 7 x lo6 (dashed urue); Kl = 10 M '~ and K, onentrations of free Ca2+ (Fig. 1). Seond, the fators that derease the double-layer potential (divalent ations and high ioni strength) also derease the affinity of PKC for the membrane. The linear dependene of the apparent assoiation onstant on the free Ca2' onentration suggests that only one Ca2+binding site regulates the assoiation of PKC with the membrane. Calium ion, like other ations, is nonspeifially au- mulated in the layer immediately adjaent to the membrane by the double-layer potential produed by anioni lipids, suh as PS (8, 11). Hene, if the Ca'+-binding site that appears to regulate the assoiation of PKC with the membrane is loated within a distane from the plane of the membrane that is shorter than the Debye length (-1 nm under our experimental onditions) (81, then any fator that dereases the double-layer potential should also derease the Ca'+-dependent binding of PKC to the membrane. We used three different ways to de- rease the surfae potential of the membrane. First, we inluded in the solution one of three different divalent ations, Me, Mn2+, or Ba2+. Sine their onentration in the solution (5 mm) was 20 times lower than that of the monovalent ation K', they inreased the ioni strength of the solution only very modestly. However, these divalent ations signifiantly redued the double-layer potential through diret binding to both PS and PC (10, 11). Magnesium ion dereased the affinity of the enzyme for the membrane omposed of 15 mol % POPS, 1 mol % DG, and 84 mol % POPC by a fator of -3 (Fig. 1,O). Manga- nese ion, whih binds several times more strongly than Mg2f to both PS and PC, redued the binding of PKC to the membrane by -1.5 orders of magnitude (Fig. 1, A). Barium ion, whih

3 13800 Ca2+-dependent Assoiation - d I r: Y +J (D 51 IO6! Q, of PKC with Membranes FIG. 2. Dependene of apparent affinity of PKC for membrane without diolein on onentration of free Ca2+. Lipid vesiles were A = 496 nm upon the addition of Ca. The exitation wavelength was omposed of POPS and POPC in equimolar amounts. In one experi- 280 nm. Both slits were set at 4 nm. All experiments were onduted in ment, 20 mol % PCwas substituted by an equimolaramount of solution ontaining 100 mm KCl, 20 mm Tris-HC1, ph 7.0, 25 p~ EGTA, D-sphingosine (+). The total lipid onentration was in the range of 5 p~ lipid in the form of large unilamellar vesiles omposed of 50 mol mm.all solutions used in the experiment ontained 100 mm % POPS, 10 mol % dansyl-l-a-phosphatidylethanolamine, and POPC as KCl, mm CaCl,, and 0.3 mg/ml BSA and were buffered by 20 mm a remainder, and 8 pg of PKC. The solution was modified by the addi- Tris-HC1 to ph 7.0. This omposition of solution was modified by the tion of 5 mm MgCl, (0) or was left unhanged (). The total volume of addition of 5 mm MgCl (0) or was left unhanged ( and +). The the solution was 200 pl. The desired onentration of free Ca was onentration of free Ca2* was altered by the addition of EGTA and was ahieved by the addition of 5 pl of CaCl, stok solutions. Points and burs alulated as desribed under Experimental Proedures. The onen- represent means 2 S.D. (n = 3). tration of PKC was in the range nm. Points represent an average of at least two independent experiments, eah in tripliate. Bars, shown the Appendi~ ).~ We have tested the validity of this mass awhen larger than the symbol size, indiate standard deviation. The dashed urue is a graphi representation of the model of assoiation of tion law by making measurements for a given membrane om- PKC with the membrane (see the Appendix ). The assoiation onstants were as follows: Kl = 30 M-~, K2 = 2 x lo8 M-, and K3 = 350 M-~. binds PS with similar affinity ompared with Mn2+ (11) but has 10-fold weaker affinity for PC (lo), dereased the PKC binding to the membrane by about an order of magnitude (data not shown). Seond, the ioni strength of the solution was raised by an inrease in the onentration of KC1 from 100 to 300 mm. This moderate inrease in the ioni strength resulted in a very strong derease in the affinity of PKC for the membrane. As we shall disuss below in more detail, the effet of the ioni strength on the assoiation of PKC with the membrane appears to be more omplex than it being solely a result of the suppression of the double-layer potential. We have also investigated the effet of Ca2+ on the binding of PKC to membranes that did not ontain DG (Fig. 2). To offset the 2 orders of magnitude loss of PKC affinity for the membrane aused by the removal of diolein (61, we inreased the perentage of PS in the membrane to 50%. The addition of 20 mol % of a ationi amphiphile, sphingosine, was the third way of suppressing the magnitude of the negative potential produed by PS. The affinity of PKC for the membrane of this omposition (Fig. 2, +) was -3-fold lower than for that omposed of PS and PC only (m). The same onentration of Mg2 (5 mm) had a larger effet on the affinity of PKC for the membrane at a higher mole perent of PS. At 50 mol % PS, Mg2+ dereased the binding of PKC to the membrane by up to 20-fold (Fig. 2,O). We demonstrate below that this effet is also onsistent with the hypothesis that the Ca2+-binding site on PKC is within a distane of 1 nm from the membrane surfae. The dependene of the apparent assoiation onstant for PKC with the membrane on the onentration of free Ca2+ was linear, as in ase the of the assoiation with the membrane ontaining DG. A departure from linearity seen at higher onentrations of Ca2+ (Figs. 1 and 2) is onsistent with a derease in the double-layer potential aused by the binding of Ca to phospholipids. The binding of PKC to a membrane of defined lipid omposition an be desribed by a single assoiation onstant (see also W m rn W L.A ap 25 O t f e, +, t + +e e* + + e o 1.I I000 [Ca2+l hm1 FIG. 3. Effet of Mg2* on assoiation of PKC with membrane of large unilamellar vesiles. The relative inrease in the emission at position at total lipid onentrations that differ by up to 20-fold. The data shown in all figures represent the average of measurements made at a total lipid onentration at whih the bound fration of the enzyme was within the 1585% range. The results obtained on vesiles of the same lipid omposition but at different total lipid onentrations were idential within the limits of experimental error. The upper limit of the total lipid onentration was set at 5 m ~ The. lower ~ limit of the total lipid onentration (2 PM) was hosen so that bound PKC would not signifiantly affet the onentration of free PS in the membrane. All data shown in Figs. 1 and 2 were obtained using the binding assay employing surose-loaded large unilamellar vesiles. The surose-loaded vesile assay has been shown to be partiularly useful in studies of the assoiation of enzymes with phospholipids, yielding results onsistent with those obtained by other methods (6,141. Fig. 3 demonstrates that a derease in PKC affinity for the membrane aused by 5 mm Mg2+ was seen also when the assoiation of the enzyme with the membrane was investigated by the fluoresene resonane energy transfer from the tryptophan moieties on the protein to the dansyl flu- orophore attahed to the head group of phosphatidylethanolamine (15). Idential volumes of onentrated CaCl, solutions were added to the premixed suspensions of large unilamellar vesiles and PKC. An inrease in the fluoresene emission intensity of the dansyl group was a measure of the assoiation of the enzyme with the membrane (15). The results obtained in We have previously observed small deviations from this simple mass ationbehavior (6). Thesedisrepaniesoriginatedin a nonspeifi sedimentation of 2-7% of the total amount of PKC seen also in the absene of lipid vesiles. In the presene of 0.3 mg/ml BSA, as used throughout the experiments reported here, the autosedimentation of the enzyme fell below a detetable level. * At this onentration, the volume enlosed in spherial vesiles of 0.1-pm diameter amounts to -3% of the total volume of solution. Higher lipid onentrations require preise knowledge of the vesile volume after entrifugation to assure an aurate alulation of the fration of bound enzyme.

4 ~~ Ca2+-dependent Assoiation of PKC with Membranes I 1 lo2t al k 101' 'I l a Log (ICa2+11 FIG. 4. Dependene of apparent affinity of PKC for membrane at very low onentrations of free Ca2+. The apparent assoiation onstant was alulated from the binding results obtained in the surose-loaded vesile assay as desribed under "Experimental Proedures." Lipid vesiles were omposed of either 50 (W and 0) or 20 ( + ) mol % POPS, 1 mol % diolein, and POPC as a remainder. The total lipid onentration was in the range of m. All solutions used in the experiment ontained 100 mm KC1,0.500 mm EGTA, and 0.3 mg/ml BSA and were buffered by 20 mm Tris-HC1 to ph 7.0. This omposition of solution was modifiedby the addition of 5 MgC1, (0) or was left unhanged (W and + ). The onentration of free Ca2+ was altered by the addition of stok solutions of CaCl, and was alulated as desribed under "Experimental Proedures." The onentration of PKC was in the range nm. Points represent an average of at least two independent experiments, eah in tripliate. Bars, shown when larger than the symbol size, indiate standard deviation. Curves are graphi representations of Equation 6 from the "Appendix." The assoiation onstants were [MI as follows: Ifl = 2000 M-' and Kz = 7 x lo7 (solid urve); Kl = 800 M -~ and K, = 9 x lo6 (dotted urue); and K3 = 350 M" (dashed urue); Kl = 30 M'~ and K, = 9 x lo6 M'~ for all urves. the absene of Mg2' (Fig. 3, B) are very similar to the published data, obtained by the same method, on vesiles of idential phospholipid omposition (4). In the presene of 5 mm Mg2' (Fig. 3, 01, the onentration of Ca" required for a similar inrease in the fluoresene emission was times higher than that in the absene of Mg2'. This result remains good in quantitative agreement with that obtained by the surose-loaded vesile method on vesiles ontaining an idential fration of PS (Fig. 2, and O).5 Although the apparent assoiation onstant for PKC with the membrane depends linearly on the free Ca2+ onentration (Figs. 1 and 21, the observed effets of the double-layer potential ould result from a diret interation of the negatively harged membrane with a putative positively harged domain of the protein, instead of indiret ation through the aumulation of Ca2+ in the layer immediately adjaent to the mem- brane. The distintion between these two possibilities beomes apparent at Ca2+ onentrations below the dissoiation onstant for that ation from membrane-bound PKC. Due to the experimental onstraints on the total lipid onentration dis- ussed above, we ahieved a suffiient affinity of the enzyme for the lipid bilayer by raising the perentage of PS to 20 or 50% and inluding 1 mol % diolein. The data shown in Fig. 4 demonstrate that below a ertain onentration of free Ca2+, the affinity of PKC for the membrane depends little on the onentration of that ation. However, the threshold level of Ca2+ A negatively harged dansyl-l-a-phosphatidylethanolamine is expeted to ontribute to the nonspeifi aumulation of Ca2+ in the double layer. The relative effet of Mg2' on the magnitude of the doublelayer potential of the membrane ontaining either 50 or 60% anioni lipid is similar. We hose this partiular lipid omposition for the sake of omparison with published data. a21 Fa" K3 I PKC I SCHEME 1 seems to depend on the double-layer potential of the membrane. A derease in the perentage of PS in the membrane from 50% (Fig. 4, B) to 20% ( + inreased this threshold level of Ca2+ by about an order of magnitude. The presene of 5 mm Mg2' (Fig. 4,O) had two distint effets on the binding of PKC to the membrane ontaining 50 mol % PS and 1 mol % DG. First, the apparent assoiation onstant for the enzyme with the membrane dereased %fold in the range of Ca2+ onentrations below the apparent dissoiation onstant for this ation from the membrane-bound enzyme. This ould be attributed either to a ompetition6 between PKC and Mg2' for PS or to weaker hypothetial aumulation of the protein in the viinity of the membrane surfae by the double-layer potential. However, a substitution of 20 mol % PC by an equimolar amount of sphingosine in the membrane also ontaining 50 mol % PS and 1 mol % DG dereased the apparent assoiation onstant for PKC with the bilayer by only a small amount, from 2400 to 1900 M-'. This indiates that the enzyme itself is neither at- trated nor repelled by the negative double-layer potential produed by PS. Seond, the threshold level of Ca2+ was severalfold higher in the presene of 5 mm Mg2' (Fig. 4, 0) than in its absene (B). This resulted in -20-fold an higher affinity of PKC for the membrane in the absene of Mg2' at miromolar onentrations of Ca2+. Hene, Mg2' dereased the affinity of PKC for the membrane diretly by ompetition for PS and indiretly by diminishing the onentration of Ca2+ around its putative binding site on the enzyme. We used a simple mass ation model to desribe the data shown in Fig. 4 in a more quantitative manner. This model ontains three independent assoiation onstants that an be measured separately at different ranges of Ca2+ onentration. The assoiation onstant for PKC with the membrane of defined omposition, ICl (Sheme 11, is given diretly by the apparent assoiation onstant in the range where K, does not depend on the onentration of Ca2+ (the flat bottom part of eah urve). The assoiation onstant for membrane-bound PKC with Ca2+, K,, an be determined from the slope of the stee portion of eah urve. We have also assumed that Ca2+ an bind soluble to PKC with the assoiation onstant K3. We shall demonstrate below how this onstant an be determined. The details of the model are given in the "Appendix." The dissoiation onstant for Ca2+ The fration of free PS in the membrane was alulated from the known assoiation onstants for PS with all ations present in the solution (ll), as desribed under "Experimental Proedures." The derease in the fration of free PS aused by the assoiation of the phospholipid with Mg2' and Ca" was moderate. For example, when the membrane ontained 50 mol % PS, the presene of 5 mm Mg2' and 0.2 mm Ca" dereased the perentage of free PS from 76 to 70%.

5 13802 Ca2+-dependent Assoiation of PKC with Membranes TABLE I Conentration of ea2+ in the proximity of harged membranes The onentration of Ca2+ at a given distane from the plane of the membrane was alulated as desribed under "Experimental Proedures" assuming 1 p free Ca2+ in the bulk solution. The apparent dissoiation onstant, Kd, was obtained from a fit of the model (see the "Appendix") to the experimental data (Figs. 1, 2, and 4). The intrinsi dissoiation onstant, was alulated for the putative Ca2+-binding site loated 0.3 nm from the plane of the membrane. All numbers are rounded to 2 signifiant digits. nm 2.3 [CaZ+l at given distane from membrane [PSI EM%'] Kd Ko.3 0 nm 0.3 nm 1 nm mol % nm mm PM PM PM , from the membrane-bound enzyme obtained from the best fit of the model to the experimental data was -14 n~ for the membrane ontaining 50 mol % PS with no Mg2+ present in the solution. The presene of 5 mm Mg2' or the redution of PS ontent in the membrane to 20 mol % inreased Kz to -110 nm. These numbers were alulated with the assumption that onen- the tration of free Ca2+ around its binding site on the enzyme is idential to that in the bulk solution. However, if this binding site is within a distane from the membrane surfae shorter than 1 nm, then the Ca2+ onentration in the proximity of its putative binding site is signifiantly higher due to the nonspeifi aumulation of Ca2+ by the double-layer potential. The ombination of the Gouy-Chapman-Stern theory of the doublelayer potential with the Boltzmann equation (8,9) predits that at a distane of 0.3 nm from a membrane ontaining 50 mol % PS, the onentration of Ca2+ will be -40 PM if the onentration of this ation in the bulk solution is kept at the 1 p~ level under our experimental onditions. However, the presene of 5 m~ Mg2' redues this nonspeifi aumulation of Ca2+ to -5 p ~ At. the same distane from a membrane that ontains only 20 mol % PS, the onentration of Ca2+ will be -8 PM under these onditions (no Me). Hene, the addition of Mg2+ or a derease in the mole perent of PS from 50 to 20% redues the aumulation of Ca2+ at a distane of 0.3 nm from the membrane surfae by fators of 8 and 5, respetively. As shown in Table I, these mitigating effets of Mg2' or a lower mole perent of PS hange signifiantly over short distanes. For example, diretly at the at all onentrations of Ca2+. The onentration of Ca2+ was hanged in the range from 0.1 to 5 mm; however, the sum of the onentrations of Ca2+ and Mg2' was kept onstant at 5 mm. Sine the affinities of both ations for PS and PC are similar, the predited hanges in the double-layer potential were small. Fig. 5 shows the Ca2+ dependene of the apparent assoiation onstant for PKC with the membrane ontaining 15 mol % PS and 1 mol % DG under suh onditions. The predited saturation effet beomes apparent at millimolar onentrations of Ca2+. The dissoiation onstant for Ca2+ from soluble PKC, obtained from the best fit of the model (Fig. 5, dashed to the experimental data (W), was -3 mm. The saturation effet was also observed for membranes of other lipid ompositions at similar CaZ+ onentrations (data not shown). All the data presented above were obtained with a mixture of membrane surfae, these differenes rise to 30- and 25-fold, re- Ca2+-dependent isozymes purified from rat brain. Fig. 6 shows spetively. However, 1 nm from the surfae of the membrane, the nevertheless that at a Ca2+ onentration lower than 1 nm, the aumulation of Ca2+ will be only 2-fold smaller when Mg2' is present in the solution or when the mole perent of PS in the membrane is redued from 50 to 20%. If the Ca2+-binding site that regulates the assoiation of PKC with the phospholipid bitransloation of the enzyme was omplete when its affinity for the membrane was inreased by an inrease in the mole perent of PS in the membrane. A similar effet was ahieved when the total lipid onentration of a given omposition was approlayer ontaining PS is loated 0.3 nm from the membrane sur- priately raised (data not shown). This indiates that none of the fae, then the dissoiation onstant for Ca2+ from the membrane- Ca2+-dependent isoforms of PKC present in the mixture rebound enzyme equal to -700 nm will desribe all of the quires this ation for the assoiation with the membrane if the experimental data shown in Fig. 4 with an auray better than interation of PKC with PS and DG is strong enough. a fator of 2 (Table I). At this very low onentration ofca", the enzyme also We also applied this simple mass ation model desribing the reahed full ativity for either a syntheti peptide, VRKRTLdependene of PKC assoiation with the membrane to the data RRL (Fig. 6, O), or histone (data not shown). The dependene of obtained with membranes ontaining either 15 mol % PS and 1 the enzyme ativity for the peptide on the mole perent of PS mol % DG (Fig, 1) or 50 mol % PS (Fig. 2). The assoiation of PKC with these membranes was very weak at very low onentrations of Ca2+ (0.5 mm EGTA, no exogenous Ca2+). The assoiation onstant for PKC with the membrane, K,, was measfollowed that for the enzyme binding to the membrane. The presene of the peptide at 10 p~ in the ativity assay had little effet on the enzyme binding to the membrane (data not shown). ured on vesiles ontaining 15 mol % PS and 5 mol % DG at a DISCSSION total lipid onentration of 5 mm. Sine the affinity of PKC for Stoihiometry of Membrane-bound PKCCa" Complex-The the membrane was proportional to the mole perent of DG in linear dependene of PKC affinity for the lipid membrane on this range (6), we divided the value of Kl obtained in this way by 5 and used the result to determine the assoiation onstant for Ca2+ with membrane-bound PKC (Fig. 1). The value of 30 M-~ for Kl given for the membrane ontaining 50 mol % PS and no DG is an upper estimate. Hene, the dissoiation onstant for CaZ+ from PKC bound to the membrane, K,, of this omposition (Fig. 2, dashed urve) is no larger than 5 nm, -3-fold less than for the enzyme bound to the membrane ontaining the same mole perent of PS and, additionally, 1 mol % DG. It is apparent from Equation 6 (see the "Appendix") that if Ca2+ also binds to the soluble form of PKC, then the dependene of the apparent assoiation onstant for the enzyme, K,, on the onentration ofca2+ should level at the Ca2+ onentration lose to the dissoiation onstant for this ation from the soluble enzyme. We tested this predition under onditions in whih the double-layer potential was maintained approximately onstant If the small differenes in the nonspeifi aumulation of Ca2' OCurring under these onditions are taken into aount, the best fit yields a value of 3.3 mm.

6 ,**" Ca2+-dependent Assoiation -4.O -3.O -2.o Log [Ca2+l [MI Fr. 5. Dependene of apparent affinity of PKC for membrane at high onentrations of free CaZ+ and determination of assoiation onstant for Ca2+ with soluble enzyme. The apparent assoiation onstant was alulated from the binding results obtained in the surose-loaded vesile assay as desribed under "Experimental Proedures." Lipid vesiles were omposed of 15 mol % POPS, 1 mol % diolein, and POPC as a remainder. The total lipid onentration was in the range of mm. All solutions used in the experiment ontained 100 mmkc1 and 0.3 mg/ml BSA and were buffered by 20 mm Tris-HC1 to ph 7.0. The sum of the onentrations of Mg2' and Ca" was kept onstant at 5 mm. The onentration of PKC was in the range nm. Points represent an average of at least two independent experiments, eah in tripliate. Bars, shown when larger than the symbol size, indiate standard deviation. The solid urue is a graphi representation of Equation 6 from the "Appendix." The assoiation onstants were as follows: Kl = 15 M-', K2 = 3.5 x lo6 M-', and K3 = 350 M-' $ 1 I A 0 ' I 2 > 751 u " I o 0 Q 0 I " I I mol% PS FIG. 6. Dependene of PKC ativity and binding to phospholipid bilayer on mole perent of PS at onentrations of Ca" below dissoiation onstant of that ation from enzyme. The fration of membrane-bound enzyme (0) was determined in the suroseloaded vesile assay. Lipid vesiles were omposed of 0.2 mol % diolein, the indiated mole perent of PS, and POPC as a remainder. The total lipid onentration was 0.5 mm. The solutions used in the experiment ontained 100 mm KCl, 5 mm MgCl,, mm EGTA, and 0.3 mg/ml BSA and were buffered by 20 m Tris-HC1 to ph 7.0. The ativity of PKC for the peptide (0) was measured as desribed under "Experimental Proedures." The onentration of PKC was 200 ng/ml. The absolute value of the maximal ativity was 0.95 nmol pg" min". Points represent an average of two independent experiments, eah in tripliate. Burs, shown when larger than the symbol size, indiate standard deviation. the onentration of Ca2+ over a wide range (Figs. 1 and 2) strongly supports the hypothesis that only a single Ca2+-binding site regulates the affinity of the enzyme for the lipid bilayer ontaining either PS or PS and DG. The assoiation onstant for Ca2+ with the membrane-bound enzyme may be determined of PKC Membranes with 13ao3 from the measurements of the Caz+-dependent binding of the enzyme to the membrane (Fig. 4). An alternative hypothesis of n idential and independent Ca2+-binding sites that ould regulate the binding of the enzyme to the membrane does not apply to moleules that have multiple binding sites for the ligand(s) that remains in a phase of lower dimensionality. As it was shown theoretially and demonstrated experimentally, the binding of a multivalent maromoleule to its ligands onurrent with the transloation from three-dimensional solution to a two-dimensional surfae will result in the appearane of apparent ooperativity (19-21). Our results are onsistent with a very reent finding that the regulatory domain of PPKC ontains a single high affinity Ca2+binding site (22). However, the apparent dissoiation onstant for Ca2+ from membrane-bound PKC (700 nm) dedued from our binding experiments is several times lower than that found for the regulatory domain of PPKC. We are not able to make a definite onlusion about the identity of these Ca2+-binding sites sine the experimental methods used in these studies are very different. The existene of a single high affinity Ca2+-binding site was also demonstrated for apkc by a ompetition between Ca2+ and Gd3+ for a ommon binding site on the enzyme (5). However, this partiular Ca2+-binding site seemed to be unique for the 01 isotype. Moreover, it was impliated in the regulation of the interation of PKC with its substrates. Diret equilibrium dialysis measurements demonstrated that membrane-bound lassi PKC assoiates with 8-10 Ca2+ ions (4). A "sequential binding" model developed for the interpretation of these results suggested that Ca2+ ions bind to the protein-lipid interfae in a highly ooperative fashion and mediate the assoiation of the enzyme with the membrane (23). Our data do not reveal the total number of PKC-bound Ca2+ ions, but strongly suggest that only one of the Ca2+-binding sites regulates the assoiation of lassi PKC with the phospholipid bilayer. All Ca2+-dependent isotypes of PKC seem to share this binding site sine the Mihaelis onstants for Ca2+ are very similar for all three prinipal Ca2+-dependent isoforms of PKC expressed separately in the baulovirus system (24). Loation of Ca2+-binding Site with Regard to Membrane Surfae-Our results suggest that the Ca2+-binding site is loated within a distane of 1 nm from the membrane surfae. A value of 0.3 nm yields the best predition of the ombination of the Gouy-Chapman-Stern theory of the double-layer potential (8,9) with a simple mass ation model (see the "Appendix"). All three methods (i.e. the addition of either a ationi amphiphile to the bilayer or phospholipid-binding divalent ations to the solution as well as an inrease in the ioni strength of the solution)' that we used to derease the double-layer potential and onsequently the aumulation of Ca2+ in the proximity of the membrane surfae dereased the affinity of PKC for the membrane. Fig. 7 demonstrates that when the apparent assoiation onstant for the enzyme with the membrane is plotted against the onentration of Ca2+ at a distane of 0.3 nm from the membrane surfae, the differenes between results ob- The moderate hange in the ioni strength aused by an inrease in the onentration of KC1 from 100 to 300 mm resulted in a large 50-fold derease in the affinity of PKC for the membrane (Fig. 1, + ). Therefore, we also tested the effet of the inreased ioni strength at Ca2+ onentrations below the value of the dissoiation onstant for this ation from membrane-bound PKC (0.5 II~M EGTA, no exogenous CaZ+). nder these onditions, the affinity of PKC for the membrane ontaining 50 mol % PS and 1 mol % DG dereased 15 times when the onentration of KC1 rose from 100 to 300 mm. It is likely that the higher ioni strength of the solution dereased the affinity of the PS-binding sites on the enzyme for the phospholipid. Sine PKC ontains multiple binding sites for PS (13, 251, a small hange in the affinity of a single binding site rises geometrially with the number of binding sites.

7 13804 ea2+-dependent Assoiation of PKC with Membranes membranes omposed of PSPC mixtures with or without DG, whih we proposed earlier (6). Binding of PS, DG, and Ca2+ to their respetive binding sites on the protein provides the free +I 106- energy hange required for the hange in the protein onform +I mation and its assoiation with the membrane. The onformag 105- tional hanges in PKC ourring upon assoiation with the 0 u membrane were demonstrated diretly by the inreased sus- E 104- eptibility of the "hinge" (25, 27) and pseudosubstrate (28) do- " mains to proteolysis as well as by hanges in the CD spetra (29, 30). The free energy hange in the onformational transformation of the protein an be estimated from the ratio of the dissoiation onstants for Ca2+ from the membrane-bound and soluble enzymes, respetively. An -4 orders of magnitude differene in these onstants is equivalent to a free energy hange I of 5-6 kal/mol. It is not surprising then that either a several orders of magnitude inrease in the total lipid onentration or a substantial inrease in the mole perent of PS in the membrane provides a suffiient amount of free energy to ause the FIG. 7. Dependene of apparent affinity of enzyme for mem- assoiation and subsequent ativation of PKC at Ca" onenbrane on onentration of free CaZ+ at distane of 0.3 nm from trations below its dissoiation onstant from the enzyme, as membrane surfae. The experimental data were taken from Figs. 1 (open symbols) and 2 (losed symbols). The designation of symbols is the was demonstrated in Figs. 4 and 6. In in vitro systems, neither same as desribed for those figures. The onentration of Ca2+ at a DG (6, 15) nor Ca2+ (31) is neessary for the transloation of distane of 0.3 nm from the surfae was alulated from the bulk on- PKC to the lipid phase. However, low onentrations of these entration of free ation by means of the Guy-Chapman-Stern theory of ativators of PKC in the ell usually require their simultaneous the double-layer potential and the Boltzmann relation. The intrinsi assoiation onstants for K+, Me, Ca2+, and Mn2+ with PS were equal rise to ahieve a substantial level of PKC transloation to the to 0.16, 8, 12, and 24 M-', respetively (11). The intrinsi assoiation ell membranes and its subsequent ativation. It was reently onstants for these ations with PC were equal to 0, 2, 3, and 10 M-', demonstrated though that higher levels of Ca2+ in ells subrespetively (7, 10). All other standard assumptions of the Gouy-Chap- jeted to a pathologial shear stress ativate PKC without any man-stern theory also apply to the present ase. inrease in DG ontent in the membrane (32). tained under different experimental onditions are signifi- The addition of 1 mol % diolein to a membrane also ontainantly redued or disappear entirel~.~ Furthermore, the disso- ing 50 mol % PS had little effet on the dissoiation onstant for iation onstants for Ca2+ from membrane-bound PKC that Ca" from the membrane-bound enzyme. This would suggest seemed to depend on the mole perent of PS in the membrane little allosteri interation between diaylglyerol- and Ca2+or the presene of Mg2+ (Fig. 4) may be substituted by a single binding sites. The same mole perent of DG inreased the afvalue of -700 nm if the Ca2+-binding site is loated 0.3 nm from finity of PKC for the bilayer -100-fold at the same onentrathe membrane surfae. The effet of Mg2+ or other divalent tion of Ca2+ (Fig. 2 uersus Fig. 4, m). A fator of somewhat less ations on the affinity of PKC for the membrane ould be al- than 500 was found for membranes with 20 mol % PS (6). Thus, ternatively explained by a diret ompetition of these ations the interation between PS- and DG-binding sites also appears with Ca2+ for a ommon binding site. However, a diret ompe- to be weak. However, this interation may be stronger when tition of divalent ations for the binding site on the protein these lipids are inluded in Triton X-100-mixed mielles (33). should not depend on the mole perent of PS in the membrane. Conlusions-Our data suggest that only one Ca2+-binding As an be seen from Figs. 1 and 2, the effet of MgZ' was -5-fold site regulates the assoiation of lassi PKC with the phosphostronger when the mole perent of PS in the membrane was lipid bilayer. This binding site appears to be loated on memraised from 15 to 50%, as expeted from the differenes in the brane-bound PKC within a distane of -0.3 nm from the plane nonspeifi aumulation of Ca2+ by the double-layer potential. of the membrane. The intrinsi dissoiation onstant for Ca2' Furthermore, sphingosine ould possibly ompete with Ca2+ for for this binding site is -700 nm when orreted for the nonspea ommon binding site only if it was loated diretly at the ifi aumulation of Ca2+ in the double layer. The interation membrane surfae." In general, only the hypothesis of a Ca2+- between this binding site and the putative PS-binding sites is binding site loated within the diffuse double layer provides a strong. The onformational transformation of PKC that ours quantitative explanation to all experimental data. at the interfae requires a free energy hange of -5 kal/mol. Model of PKC Binding to Lipid Bilayer-The results pre- The interation between the diaylglyerol-binding site and sented above further support the model of PKC binding to either the PS- or Ca2+-binding site is weak. The derease in PKC affinity for the membrane aused by Mn2+ is APPENDIX higher than that expeted from the hange in the onentration of Ca". The model that we have used to desribe the dependene of It may be aused by a more pronouned ompetition of Mn" with PKC the apparent assoiation onstant for PKC with the lipid memfor PS. This ation displays 3 times higher affinity for PS than does brane, K,, on the onentration of free Ca2+ in the solution is Mg2' (11). The proximity of results obtained for membranes of different lipid ompositions (Fig. 7, losed versus open symbols) originates in the based on two assumptions. First, we assumed that thenzyme, experimental onstraints on the Ca2+ and lipid onentrations. The dis- E, assoiates with lipid, L, omposed of PSPC or PS/DG/PC in tane between the data obtained for different membranes is a funtion of their omposition, in partiular PS and DG ontent. lo The inhibitory effet of sphingosine on PKC ativity was interpreted as its ompetition with diaylglyerol (or phorbol ester) for the same binding site on PKC (26). However, as we have demonstrated here and elsewhere (61, 1 mol % DG inreased the affinity of the enzyme for PS by 2 orders of magnitude. In ontrast, 20 mol % sphingosine dereased this affinity 3 times. Hene, DG and sphingosine are not likely to ompete for the same binding site on PKC, at least when present in lipid bilayers. the form of large unilamellar vesiles. For a membrane of defined lipid omposition, this assoiation may be desribed by a mass ation equation (Equation 2) with a single assoiation onstant, Kl. Seond, we have assumed that the membranebound enzyme may bind a single Ca" ion with the assoiation onstant K, (Equation 3; or the dissoiation onstant Kd, Table I), whih is higher than the assoiation onstant for the soluble enzyme, K3 (Equation 4). It follows immediately that the asso-

8 Ca2+-dependent Assoiation of PKC with Membranes iation onstant for the enzwe-ca2+ omplex with the lipid, K4 9. MLaughlin, S. (1989) Annu. Reu. Biophys. Biophys. Chem. 18, (Equation 5), is higher than Kl. Only three of these onstants 10. MLaughlin, A,, Grathwohl, C., and MLaughlin, S. (1978) Biohim. Biophys. are independent. Ata 613, MLaughlin, S., Mulrine, N., Gresalfi, T., Vaio, G., and MLaughlin, A. (1981) Kl [El [LI = [Elb (Eq. 2) J. Gen. Physiol. 77, Huang, K.-P., Chan, K. F., Singh, T. J., Nakabayashi, H., and Huang, F. L. (1986) J. Biol. Chem. 261, Kz [E]*[Ca2+] = [ECa2+Ib (Eq. 3) 13. Hannun, Y. A,, Loomis, C. R., and Bell, R M. (1985) J. Bid. Chem. 260, K3 [E][Ca"] = [ECa"] (Eq. 4) 14. Rebehi, M., Peterson,A., and MLaughlin, S. (1992)Biohemistry 31, K4 [ECa''] [L] = [E.Ca2+Ib (E~, 5) 15. Bazzi, M.D., and Nelsestuen, G.L. (1987) Biohemistry 26, Kaibuhi, K., Takai,Y., andnishizuka,y. (1981)J. Biol. Chem. 266, Brakets denote the onentration, whereas the b 17. Chakravarthy, B. R., Bussey, A., Whitfield, J. F., Sikorska, M., Williams, R. E., and Durkin, J. P. (1991)Anal. Biohem. 196, denotes the membrane-bound enzyme. A rearrangement of 18. Fabiato, A. S., and Fabiato, F. (1979) J. Physiol. (Paris) 76, Equations 2-5 leads to the following formula (Equation 6) for 19.Reynold, J. A. (1979) Biohemistry 18, Dwyer, J. the apparent assoiation onstant for the with the D., and Bloomfield, V. A. (1981) Biopolymers Mosior, M., and MLaughlin, S. (1992) Biohim. Biophys. Ata 1106, membrane. 22. Luo, J.-H., and Weinstein, I. B. (1993) J. Biol. Chem. 268, Bazzi, M. D., and Nelsestuen, G. L. (1991) Biohemistry 30, 797CL7977 WLI(1+ Kz[CaZ+I) 24. Bums, D. J., Bloomenthal, J., Lee, M.-H., and Bell, R. M. (1990) J. Biol. Chem. Ka = (Eq. 6) 266, K3[Ca2+] 25. Newton, A. C., and Koshland, D. E., Jr. (1989) J. Bid. Chem. 264, REFERENCES 26. Hannun, Y. A,, Loomis, C. R., Memll, A. H., Jr., and Bell, R. M. (1986) J. Biol. Chern. 261, Nishizuka, Y. (1989) Caner (Phila. 63, Kishimoto, A,, Kajikawa, N., Shiota, M., and Nishizuka, Y. (1983) J. Biol. 2. Stabel, S., and Parker, P. J. (1991) Pharmaol. & Ther. 61, Chem. 268, Bums, D. J., and Bell, R. M. (1992) in Protein Kinase C: Current Conepts & 28. on; J. w., Keranen, L.M., and Newton, A. C. (1992) J. Bid. Chem. 267, Future Perspetives (Lester, D., and Epand, R. M., eds) pp. 2540, Horwood, Chihester, nited Kingdom 29. Lester, D. S., and Brumfeld, V. (1990) Int. J. Bid. Maromol. 12, Bazzi,M.D., and Nelsestuen, G.L. (1990) Biohemistry 29, Shah, J., and Shipley, G.G. (1992) Biohim. Biophys. Ata 1119, Maurer, M. C., Sando, J. J., and Grisham, C. M. (1992) Biohemistry 31, 31. Epand, R. M., Stafford, A. R., and Lester, D. S. (1992) Eur. J. Biohem. 208, Mosior, M., and Epand, R. M. (1993) Biohemistry 32, Kroll, M. H., Hellums, J. D., Guo, D., Durante, W., Razdan, K., Hrbolih, J. K., 7. MLaughlin, S., and Whitaker, M. (1988) J. Physiol. (Lond.) 396, and Shafer, A. I. (1993) J. Biol. Chem. 268, 352C~ MLaughlin, S. (1977) Curr. Top. Membr. Dansp. 9, Orr, J. W., and Newton, A. C. (1992) Biohemistry 31,