The Effector Domain of Myristoylated Alanine-rich C Kinase Substrate Binds Strongly to Phosphatidylinositol 4,5-Bisphosphate*

Both the myristoylated alanine-rich protein kinase C substrate protein (MARCKS) and a peptide corresponding to its basic effector domain, MARCKS-(151−175), inhibit phosphoinositide-specific phospholipase C (PLC)-catalyzed hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP2) in vesicles (Glaser, M., Wanaski, S., Buser, C. A., Boguslavsky, V., Rashidzada, W., Morris, A., Rebecchi, M., Scarlata, S. F., Runnels, L. W., Prestwich, G. D., Chen, J., Aderem, A., Ahn, J., and McLaughlin, S. (1996) J. Biol. Chem. 271, 26187–26193). We report here that adding 10−100 nmMARCKS-(151−175) to a subphase containing either PLC-δ or -β inhibits hydrolysis of PIP2 in a monolayer and that this inhibition is due to the strong binding of the peptide to PIP2. Two direct binding measurements, based on centrifugation and fluorescence, show that ≈10 nmPIP2, in the form of vesicles containing 0.01%, 0.1%, or 1% PIP2, binds 50% of MARCKS-(151−175). Both electrophoretic mobility measurements and competition experiments suggest that MARCKS-(151−175) forms an electroneutral complex with ≈4 PIP2. MARCKS-(151−175) binds equally well to PI(4,5)P2 and PI(3,4)P2. Local electrostatic interactions of PIP2 with MARCKS-(151−175) contribute to the binding energy because increasing the salt concentration from 100 to 500 mm decreases the binding 100-fold. We hypothesize that the effector domain of MARCKS can bind a significant fraction of the PIP2 in the plasma membrane, and release the bound PIP2 upon interaction with Ca2+/calmodulin or phosphorylation by protein kinase C.

typical mammalian cell, it plays many important roles in signal transduction and cell biology (reviewed in Ref. 1). For example, PIP 2 is the source of three second messengers: diacylglycerol, inositol 1,4,5-trisphosphate (IP 3 ), and phosphatidylinositol 3,4,5-trisphosphate (reviewed in Refs. 2Ϫ4). PIP 2 also recruits proteins containing pleckstrin homology (PH) (reviewed in Ref. 5) and other domains (reviewed in Ref. 6) to the plasma membrane and is a cofactor for the activation of phospholipase D (reviewed in Refs. 7,8). Finally, it is required for exocytosis (reviewed in Ref. 9) and activates several different ion channels (reviewed in Ref. 10). How can PIP 2 play so many different roles? The pool of PIP 2 hydrolyzed by phosphoinositide-specific phospholipase C (PLC) may be sequestered in cholesterol-enriched lipid "rafts" (11), or the enzymes that synthesize PIP 2 may be concentrated in specific regions of the plasma membrane (4). Alternatively, proteins may reversibly bind much of the PIP 2 in the plasma membrane, blunting changes in the level of free PIP 2 that would otherwise occur. We explore here the hypothesis that myristoylated alanine-rich protein kinase C substrate (MARCKS) not only binds a significant fraction of the PIP 2 in the plasma membrane, but also releases it upon binding by Ca 2ϩ /calmodulin or phosphorylation by protein kinase C (PKC).
MARCKS (reviewed in Refs. 12 and 13) is a ubiquitous PKC (14) substrate, present at high concentration in many cell types; for example, its concentration in brain is Ϸ10 M (13,15), comparable to the concentration of PIP 2 . It has two conserved regions required for membrane binding: a myristoylated N terminus and a basic effector domain (residues 151Ϫ175). The mechanism by which MARCKS binds to the plasma membrane is well understood (reviewed in Refs. 16 and 17); the N-terminal myristate inserts into the bilayer and the cluster of basic residues in the effector domain interacts electrostatically with acidic lipids. When the effector domain of MARCKS binds to Ca 2ϩ /calmodulin or is phosphorylated by PKC, its interaction with acidic lipids is reversed (reviewed in Refs. 12, 13, 16, and 17).
Previous work suggested that the effector domain of MARCKS interacts with PIP 2 ; both MARCKS and its effector domain peptide, MARCKS-(151Ϫ175), inhibit the PLC-catalyzed hydrolysis of PIP 2 in vesicles, and this inhibition is reversed by Ca 2ϩ /calmodulin binding or PKC phosphorylation of MARCKS-(151Ϫ175) (18). Although the simplest interpretation of these results is that the effector domain binds strongly to PIP 2 , it is difficult to rule out potential artifacts because MARCKS-(151Ϫ175) also aggregates the vesicles. For example, the putative large lateral domains formed in membranes by MARCKS-(151Ϫ175) (18,19) are probably an aggregation-related artifact (20).
The objective of this study was to test our hypothesis more rigorously. First, we investigated whether MARCKS-(151Ϫ175) inhibits the PLC-catalyzed hydrolysis of PIP 2 in phospholipid monolayers, a system that eliminates the possibility of artifacts due to vesicle aggregation. Second, we measured directly the binding of MARCKS-(151Ϫ175) to bilayers containing PIP 2 using two independent techniques. Next, we conducted competition and electrophoretic mobility experiments to investigate the stoichiometry of the complex formed between MARCKS-(151Ϫ175) and PIP 2 . Finally, we examined the importance of nonspecific electrostatic interactions by comparing the specificity of MARCKS-(151Ϫ175) for PI(4,5)P 2 versus PI(3,4)P 2 and measuring the effect of salt on the binding. The results all support the hypothesis that the effector domain of MARCKS binds with high affinity to PIP 2 in membranes; under "Discussion," we consider the biological implications of this strong interaction.
Monolayer Measurements-We measured PLC-catalyzed hydrolysis of PIP 2 (reviewed in Ref. 27) in a monolayer as described previously (28,29). In a typical experiment, 60 l of 55 M PC/PS/[ 3 H]PIP 2 (66.5:33:0.5) in chloroform was carefully deposited onto the surface of a solution containing 100 mM KCl, 25 mM HEPES, 0.1 mM EGTA, pH 7.5, 15 ml in a 5-cm-diameter Teflon trough. Once the chloroform had evaporated (10 min), we measured the surface pressure of the monolayer (typically 25 mN/m) using a square piece of filter paper and a balance as described previously (30). MARCKS-(151Ϫ175) was then added to the subphase; the magnetic stir bar at the bottom of the trough mixed MARCKS-(151Ϫ175) uniformly in the subphase within Ϸ1 min. After 3 min, we added PLC-␦ 1 or PLC-␤ 1 (final concentration Ϸ 0.1 nM), mixed the subphase solution for another 3 min, then added 100 M CaCl 2 to the subphase to produce a free concentration of Ca 2ϩ Ϸ 2 M, as measured using a Ca 2ϩ electrode in separate experiments. A free Ca 2ϩ concentration of 2 M produces significant activation of both PLC-␦ and -␤ on monolayers and other systems (29,31,32). We collected 200-l aliquots of the subphase at different times after the addition of Ca 2ϩ and measured the radioactivity due to [ 3 H]IP 3 . Unless specified, all measurements were done at 25°C with a monolayer consisting of PC/PS/ [ 3 H]PIP 2 (66.5:33:0.5) at a surface pressure of Ϸ 25 mN/m. We did several control experiments. First, we determined that the PLC-catalyzed hydrolysis of PIP 2 in the absence of Ca 2ϩ was negligible (data not shown), as expected for the Ca 2ϩ -dependent PLC-␦ and -␤ enzymes. Second, we showed that the initial hydrolysis rate (the initial slope of the hydrolysis curves, such as those shown in Fig. 1) was proportional to PLC concentration (data not shown). Third, we confirmed that increasing the surface pressure, , decreased the hydrolysis rate significantly (28,29), presumably because a hydrophobic region in the catalytic domain of PLC inserts into the monolayer (33). Fourth, we determined that the percentage of inhibition of PIP 2 hydrolysis was independent of PLC concentration; increasing the PLC-␦ 1 concentration 5-fold did not change the percentage of inhibition produced by 150 nM MARCKS-(151Ϫ175) (data not shown).
Other studies showed the initial rate of hydrolysis catalyzed by both PLC-␦ 1 and PLC-␤ 1 depends on the mole fraction of PIP 2 (28,31,32,34,35). Thus, if MARCKS-(151Ϫ175) binds to PIP 2 , decreasing the free concentration of PIP 2 in the monolayer, we expect the initial rate of hydrolysis to decrease.
Vesicle Preparations-We used multilamellar vesicles (MLVs), large unilamellar vesicles (LUVs), and sucrose-loaded LUVs, prepared as described in detail elsewhere (20,36). MLVs for the electrophoretic mobility measurements were prepared by drying the lipid mixture on a rotary evaporator, adding a solution containing 100 mM KCl, 1 mM MOPS, pH 7.0, and gently swirling the solution for several min. LUVs for the fluorescence binding measurements were prepared by drying the lipid mixture on a rotary evaporator, hydrating the lipids in a solution containing 100 mM KCl, 1 mM MOPS, pH 7.0, then taking the MLVs through five cycles of freezing and thawing followed by 10 extrusion cycles through a stack of two polycarbonate filters (0.1-m-diameter pore size) using a Lipex Biomembranes Extruder (Vancouver, British Columbia, Canada) (37). Sucrose-loaded LUVs for the centrifugation binding measurements were prepared in a similar manner, except that the lipid film was hydrated in a solution containing 176 mM sucrose, 1 mM MOPS, pH 7.0. After extrusion, the solution outside the sucroseloaded LUVs was exchanged for 100 mM KCl, 1 mM MOPS, pH 7.0.
Centrifugation Binding Measurements-We measured the binding of [ 3 H]NEM-MARCKS-(151Ϫ175) to sucrose-loaded PC/PIP 2 LUVs using the centrifugation technique described previously (36). Sucrose-loaded PC/PIP 2 LUVs were mixed with 2 nM [ 3 H]NEM-MARCKS-(151Ϫ175) and centrifuged at 100,000 ϫ g for 1 h. We measured the radioactivity of the supernatant and the pellet to calculate the percentage of [ 3 H]NEM-MARCKS-(151Ϫ175) bound. We minimized loss of the peptide to the tube in these experiments (and the fluorescence experiments described below) by pre-treating the tube with sonicated PC vesicles, which coats the tube surface with PC; MARCKS-(151Ϫ175) binds only weakly to PC vesicles (Fig. 2), but strongly to plastic.
Binding measurements were also done at several different salt concentrations (100Ϫ500 mM KCl). Because the iso-osmotic sucrose solution corresponding to [KCl] Ͼ 200 mM is viscous and difficult to extrude, we used a mixture of sucrose and KCl as an internal solution for the vesicles.
To describe the binding of the peptide to lipid vesicles without making assumptions about the absorption mechanism, we use a molar partition coefficient, K, as described previously (38,39) , where [P] m is the molar concentration of peptide bound to the membrane, [L] is the molar concentration of lipid accessible to the peptide (Ϸ one-half of the total lipid concentration for these LUVs because the peptide interacts only with the outer leaflet of the bilayer; the vesicles are not permeable to the peptide, which is added to a solution of preformed vesicles), and [P] is the molar concentration of free peptide in the bulk aqueous phase. For most of our binding measurements (e.g., Fig. 2 Fluorescence Binding Measurements-We determined the molar partition coefficient, K, for the binding of acrylodan-MARCKS-(151Ϫ175) to PC/PIP 2 LUVs using a fluorescence technique described in detail elsewhere (20). Briefly, acrylodan is a polarity-sensitive fluorescent dye with an excitation peak at Ϸ370 nm and an emission peak at Ϸ520 nm in water (40). When the acrylodan-labeled peptide binds to a membrane, the fluorophore inserts into the bilayer, shifting the emission peak to Ϸ455 nm and increasing the fluorescence intensity. We added acrylodan-MARCKS-(151Ϫ175) to PC/PIP 2 LUVs and measured the fluorescence of the mixture at 455 nm, corrected as described elsewhere (20), to calculate the percentage of bound peptide at a given lipid concentration.
Stoichiometry Measurements-We obtained information about the stoichiometry of the peptide-PIP 2 complex by examining how the fraction of peptide bound decreases when increasing concentrations of peptide are added to a solution of sucrose-loaded PC/PIP 2 LUVs. We used the centrifugation technique with a constant lipid concentration (10,000 nM accessible lipid in the form of sucrose-loaded PC/PIP 2 (99.7:0.3) LUVs or 30 nM accessible PIP 2 ) to obtain the results shown in Fig. 3. About 80% of MARCKS-(151Ϫ175) is bound when the peptide is present at low concentration (e.g. 2 nM). As the peptide concentration increases, it binds a significant fraction of the accessible PIP 2 and the fraction of bound peptide decreases. We performed the experiments with both MARCKS-(151Ϫ175) binds only weakly to PC vesicles ( Fig. 2), so the binding sites in PC/PIP 2 LUVs should be PIP 2 . If we assume that one MARCKS-(151Ϫ175) binds to one PIP 2 , we get Equation 2, where K a is the apparent association constant for 1:1 binding, [P-PIP 2 ] is the molar concentration of the complex of peptide and PIP 2 , [P] is the molar concentration of free peptide in the bulk aqueous phase, [PIP 2 ] is the molar concentration of free PIP 2 , [P] tot is the sum of bound and free peptide molar concentrations, and [PIP 2 ] tot is the sum of bound and free PIP 2 molar concentrations.
These three equations were combined into a single quadratic expression that predicts the 1:1 binding curve (shown in Fig. 3 with K a ϭ 2 ϫ 10 8 M

Ϫ1
, as derived from the 1% PIP 2 binding data in Fig. 2A). Electrophoretic Mobility Measurements-The electrophoretic mobility (velocity/field) of MLVs was measured as described previously (41) and used to calculate the zeta potential, the electrostatic potential at the shear plane, which is located about 0.2 nm from the surface (42). A mixture containing MARCKS-(151Ϫ175) and a low concentration of MLVs was loaded into a glass electrophoresis tube that had been pre-washed with MARCKS-(151Ϫ175) to minimize the loss of the peptide. An electrical field was applied, and the electrophoretic mobility of the vesicles, u, was measured directly using a stopwatch and a microscope. The zeta potential was calculated using the Helmholtz-Smoluchowski equation (42,43).
is the zeta potential of a vesicle, u is the velocity of the vesicle in a unit electric field, is the viscosity of the aqueous solution, r is the dielectric constant of the aqueous solution, and 0 is the permittivity of free space. As discussed previously (42), the zeta potential is proportional to the surface charge density and thus to the number of charged peptides that absorb to the vesicles.
A control experiment showed that MARCKS-(151Ϫ175) has a similar effect on the zeta potential of 98:2 PC/dipalmitoyl PI(4,5)P 2 and 98:2 PC/native bovine brain PI(4,5)P 2 MLVs (data not shown). Fig. 1 illustrates the effects of MARCKS-(151Ϫ175) on the PLC-catalyzed hydrolysis of PIP 2 . We deposited a mixture of lipids containing [ 3 H]PIP 2 at the air-water surface; added MARCKS-(151Ϫ175), PLC, and CaCl 2 sequentially to the subphase; collected samples of the subphase at the indicated times; and measured the [ 3 H]IP 3 to determine the percentage of [ 3 H]PIP 2 hydrolyzed as a function of time ( Fig. 1).

MARCKS-(151Ϫ175) Inhibits the PLC-catalyzed Hydrolysis of PIP 2 in a Monolayer-
In the absence of MARCKS-(151Ϫ175), PLC-␦ 1 produces rapid hydrolysis of PIP 2 following addition of CaCl 2 to the subphase (free [Ca 2ϩ ] Ϸ 2 M); Ϸ25% of the PIP 2 in the monolayer is hydrolyzed in 5 min (open circles in Fig. 1A). Similar results were reported previously (28,29,35). Addition of low concentrations of MARCKS-(151Ϫ175) inhibits this hydrolysis; 50 nM MARCKS-(151Ϫ175) produces Ϸ50% inhibition, and 150 nM produces almost complete inhibition (decrease in initial rate of hydrolysis or slope of line in Fig. 1A). (The measurements were done at a surface pressure of 25 mN/m to minimize the amount of PLC required. Control experiments at ϭ 30Ϫ35 mN/m, at which the area of a lipid in a monolayer corresponds to that of a lipid in a bilayer (see Footnote 2 in Ref. 28), produced qualitatively similar results (data not shown) to the data shown in Fig. 1A.) Fig. 1B shows that MARCKS-(151Ϫ175) also inhibits hydrolysis of PIP 2 catalyzed by PLC-␤ 1 , a different isoform of PLC; 10 nM MARCKS-(151Ϫ175) produces Ϸ50% inhibition, and 50 nM produces almost complete inhibition.
The monolayer results ( Fig. 1) are consistent with those reported previously using phospholipid vesicles if one notes that higher concentrations of peptide were required to produce inhibition in the vesicle experiments because most of the peptide was bound to the vesicles (18). Specifically, Ϸ10 M MARCKS-(151Ϫ175) produced 90% inhibition of PLC-catalyzed hydrolysis in vesicles containing Ϸ10 M PIP 2 . As expected, reducing the concentration of lipid decreased the concentration of peptide required for inhibition in the vesicle experiments (18). In the monolayer experiments shown in Fig.  1, there is much less PIP 2 than peptide, so most of the peptide is free in the subphase. (If the PIP 2 in a monolayer containing  Table I. 0.5% of this lipid were dispersed uniformly through the subphase, it would be present at a concentration of Ϸ1 nM).
The simplest interpretation of the results illustrated in Fig.  1 is that MARCKS-(151Ϫ175) decreases the rate of hydrolysis by binding to PIP 2 in the monolayer and competing with the catalytic domain of PLC for PIP 2 . The PLC assay system, however, is complicated. We considered three other possible explanations for our results. First, the inhibition could be due to the interaction of MARCKS-(151Ϫ175) with PLC. However, we could obtain no evidence for such an interaction using fluorescently labeled MARCKS and PLC (data not shown). Second, the inhibition could be due to the neutralization of the monovalent acidic lipids in the monolayer by MARCKS-(151Ϫ175); the rate of the PLC-catalyzed hydrolysis of PIP 2 decreases as the mole fraction of monovalent acidic lipids decreases (28), and addition of 50 nM MARCKS-(151Ϫ175) almost neutralizes the 2:1 PC/PS surface (44). However, we obtained similar results for both PLC-␦ 1 and PLC-␤ 1 using PC/PIP 2 (99.5:0.5) monolayers (Table I); these monolayers are essentially electrically neutral prior to the addition of MARCKS-(151Ϫ175). Thus, the inhibition is not due to a decrease in the electrostatic surface potential of the monolayer. (We also observed similar effects using PS/PIP 2 (99.5:0.5) monolayers (Table I), so the inhibition does not depend strongly on the mole fraction of monovalent acidic lipids in the monolayer.) Third, the inhibition could be due to irreversible denaturation/complexation of PLC by a trace contaminant in the peptide solution. This is unlikely because addition of calmodulin reverses the inhibition induced by MARCKS-(151Ϫ175) in the monolayer system (data not shown), as was observed previously in the vesicle system (18).
Experiments measuring the effects of other molecules on the PLC-catalyzed hydrolysis of PIP 2 support our interpretation that MARCKS-(151Ϫ175) inhibits the reaction by binding to PIP 2 . Neomycin, a molecule that forms a 1:1 complex with PIP 2 with an equilibrium dissociation constant in the 1Ϫ10 M range (45), also inhibits the PLC-catalyzed hydrolysis of PIP 2 ; 2 M neomycin produces a significant inhibition, and 20 M neomycin produces Ϸ50% inhibition of the PLC-␦ 1 -catalyzed hydrolysis of PIP 2 (data not shown). Conversely, simple polybasic peptides such as pentalysine and heptalysine, which do not bind with high affinity to PIP 2 (46), do not inhibit the PLC-catalyzed hydrolysis of PIP 2 . For example, 10 M heptalysine, which binds sufficiently strongly to a 2:1 PC/PS membrane to reduce the zeta potential Ϸ50% from ϷϪ50 to Ϫ30 mV (20), has little effect on the hydrolysis of PIP 2 under conditions similar to those described in Fig. 1 (data not shown).
MARCKS-(151Ϫ175) Binds Strongly to Vesicles Containing PIP 2 -To test more directly our hypothesis that MARCKS-(151Ϫ175) binds to PIP 2 , we used both a centrifugation tech-nique ( Fig. 2A) and a fluorescence technique (Fig. 2B) to measure the binding of MARCKS-(151Ϫ175) to vesicles containing PIP 2 . Fig. 2A Fig. 2B, we use a fluorescence technique to measure the binding of acrylodan-MARCKS-(151Ϫ175) to PC/PIP 2 vesicles. The acrylodan probe is polarity-sensitive; when it binds to a lipid bilayer, its fluorescence signal increases and is blue-shifted. Fig. 2B shows that acrylodan-MARCKS-(151Ϫ175) binds to PC/PIP 2 vesicles containing 0.1% and 1% PIP 2 with K Ϸ 10 5 and 10 6 M Ϫ1 , respectively, results comparable to those shown in panel A. Thus, measurements using both the centrifugation and the fluorescence techniques show that MARCKS-(151Ϫ175) binds strongly to PC/PIP 2 vesicles.
MARCKS-(151Ϫ175) Binds to More than One PIP 2 - Fig. 3 illustrates that, as the concentration of MARCKS-(151Ϫ175) increases, the percentage of MARCKS-(151Ϫ175) bound to PC/ PIP 2 vesicles decreases (open circles). The curve illustrates the predicted effect assuming the peptide forms a 1:1 complex with PIP 2 ; for example, increasing the concentration of MARCKS-(151Ϫ175) from 2 nM to 10 nM should have little effect on the fraction of peptide bound because the accessible concentration of PIP 2 ϭ 30 nM. If only 1:1 complexes were formed, Ϸ50 nM MARCKS-(151Ϫ175) should be required to decrease the percentage of peptide bound from Ϸ80% to 50%, but Fig. 3 shows that only 10 nM peptide produces this reduction. These data strongly suggest that one MARCKS-(151Ϫ175) binds to n (n Ͼ 1) PIP 2 . It is difficult, however, to deduce with any confidence the stoichiometry of the peptide/lipid interaction from these data because the theoretical expressions are complicated and model-dependent (47).  2 , and the percentage of PIP 2 hydrolyzed was measured as in Fig. 1. [Pep] indicates the concentration of MARCKS-(151-175) added to the subphase of the monolayer. % inhibition ϭ 100% ϫ (initial rate of hydrolysis without peptide Ϫ rate with peptide)/rate without peptide. This was determined from experiments similar to those shown in Fig. 1. The data are representative results from at least two sets of experiments. Although both electrostatic repulsion between the negatively charged PIP 2 molecules and entropy effects make it unlikely that n-mers of PIP 2 exist in the absence of peptide, we did two control experiments to investigate this possibility. First, addition of 0.1 mM EDTA to the bathing solution did not change the binding of [ 3 H]NEM-MARCKS-(151Ϫ175) to PC/PIP 2 (99.9:0.1) LUVs (data, similar to Fig. 2A, not shown), which suggests that trace concentrations of divalent ions are not producing n-mers of PIP 2 . Second, replacing 20% of PC by 1-stearoyl-2-arachidonoyl-sn-glycero-3-phosphatidylcholine did not change the binding of [ 3 H]NEM-MARCKS-(151Ϫ175) to PC/PIP 2 (99:1) LUVs; similarly, changing the lipid composition from PC/PIP 2 (99.9: 0.1) to PC/PE/cholesterol/PIP 2 (39.9:30:30:0.1) did not affect the binding (data, similar to Fig. 2A, not shown). Thus, the arachidonic acid chain at the 2-position on PIP 2 is unlikely to produce n-mers.
The simplest interpretation of the results shown in Fig. 3 is that one peptide binds sequentially to n PIP 2 monomers to form a complex. Experiments with PC/NBD-PIP 2 (99.9:0.1) vesicles show addition of MARCKS-(151Ϫ175) quenches the fluorescence, consistent with the peptide binding to several NBD-PIP 2 and inducing a local demixing (data not shown).
Valence of PIP 2 -A comparison of panels A and B in Fig. 4 illustrates that, in the absence of MARCKS-(151Ϫ175), MLVs containing 3% or 6% PS have similar zeta potentials to MLVs containing 1% or 2% PIP 2 , respectively. These results suggest that, although the maximum valence of PIP 2 is Ϫ5, the effective valence of PIP 2 is Ϫ3 at pH 7.0, in agreement with previous results (48). NMR experiments show about 1 proton is bound to PIP 2 at pH 7 (48,49). The results in Fig. 4 suggest that a K ϩ is also bound to PIP 2 in 100 mM KCl to yield an effective valence of Ϫ3. However, we do not know the valence of PIP 2 when it is bound to a basic peptide such as MARCKS-(151Ϫ175); it could lose a bound proton or potassium ion.
Stoichiometry of Complex from Zeta Potential Data-When  Fig. 4A, however, show that the zeta potentials of 99:1 PC/PIP 2 or 98:2 PC/PIP 2 MLVs remain negative after the addition of 10 Ϫ8 Ϫ10 Ϫ5 M MARCKS-(151Ϫ175), providing additional evidence that one MARCKS-(151Ϫ175) binds to several PIP 2 to form an electroneutral complex (i.e. one ϩ13 valent MARCKS-(151Ϫ175) combines with four ϷϪ3 valent or three ϷϪ4 valent PIP 2 ). We stress that the results in Figs. 3 and 4 provide only indirect evidence about the stoichiometry of the complex; the question needs to be addressed using more direct experimental approaches.

MARCKS-(151Ϫ175) Binds Only Weakly to Vesicles Containing Small Fractions of Other Acidic
Lipids-Addition of MARCKS-(151Ϫ175) to 6% PS or 6% PI vesicles has little effect on the zeta potential of the vesicles (Fig. 4B). The peptide exerts an intermediate effect on the zeta potential of PC/PI(4)P vesicles; 10 Ϫ7 M MARCKS-(151Ϫ175) changes the zeta potential of 4% PI(4)P vesicles from Ϫ18 mV to Ϫ15 mV (data not shown).
Although MARCKS-(151Ϫ175) binds strongly to vesicles containing high mole fraction of monovalent acidic lipids (e.g. PS and PG) due to nonspecific diffuse double layer effects, it binds only weakly to vesicles containing a low mole fraction (Ͻ10%) of monovalent acidic lipids (Fig. 4B and Ref. 25). Thus, the strong binding of MARCKS-(151Ϫ175) to PC/PIP 2 vesicles apparent from the data in Figs. 2 and 4A is not due to the average electrostatic potential (zeta potential) of the vesicles.
Local Electrostatic Effects Are Important-We measured the effect of ionic strength on the binding of MARCKS-(151Ϫ175) to PIP 2 . Fig. 5 shows that increasing [KCl] from 100 mM to 500 mM decreases the peptide's molar partition coefficient Ϸ100fold for both 99:1 PC/PIP 2 and 99.9:0.1 PC/PIP 2 vesicles; the average electrostatic potential at the surface of these vesicles is very small or negligible and cannot drive the strong binding of MARCKS-(151Ϫ175) to PIP 2 . Specifically, the zeta potential (average potential 0.2 nm from surface) is only Ϫ8 mV (Fig. 4) or Ϫ1 mV for 1% PIP 2 or 0.1% PIP 2 vesicles, respectively. Note that the molar partition coefficient does not change when the peptide concentration is increased from 2 nM (Fig. 5, filled symbols) to 5 nM (Fig. 5, open symbols). This observation that the molar partition coefficient is independent of peptide concentration when the [lipid] Ͼ Ͼ [peptide] (see Equation 1), provides an important control against several potential artifacts (e.g. the binding sites are not saturated).
Three observations suggest that the binding of MARCKS-(151Ϫ175) to PIP 2 depends on a local electrostatic interaction. First, peptide binding decreases as the salt concentration increases (Fig. 5). Second, MARCKS-(151Ϫ175) binds with similar affinity to PI(4,5)P 2 and PI(3,4)P 2 (Fig. 4A). Third, MARCKS-(151Ϫ175) binding to the phosphoinositides (PI, PI(4)P, and PIP 2 ) increases with lipid charge (Fig. 4). The local positive electrostatic potential adjacent to a MARCKS-(151Ϫ175) peptide adsorbed to a bilayer containing 30% monovalent acidic lipid is illustrated in Fig. 3  bound to PC/PIP 2 vesicles (data not shown). Thus, the picture (17) of an extended MARCKS-(151Ϫ175) with its aromatics penetrating the polar head group region derived previously from CD, EPR (50), and monolayer measurements (25) is probably also valid for the interaction of MARCKS-(151Ϫ175) with PC/PIP 2 vesicles. DISCUSSION Previous work showed that both the MARCKS protein and a peptide corresponding to its effector domain, MARCKS-(151Ϫ175), inhibit the PLC-catalyzed hydrolysis of PIP 2 in phospholipid vesicles containing physiological concentrations (33%) of the monovalent acidic lipid PS; interaction of the peptide with Ca 2ϩ /calmodulin or phosphorylation by PKC reverse the inhibition (18). The results we report here show that this reversible inhibition is not an artifact related to the peptide-induced aggregation of the vesicles (20); MARCKS-(151Ϫ175) inhibits the PLC-catalyzed hydrolysis of PIP 2 in monolayers comprising a mixture of PC/PS/PIP 2 (Fig. 1). The simplest interpretation of these results is that the effector domain of MARCKS binds to PIP 2 with high affinity and competes successfully with the catalytic domain of PLC for this lipid. Our observation that Ϸ100 nM MARCKS-(151Ϫ175) produces 90% inhibition of PLC-catalyzed hydrolysis of PIP 2 ( Fig.  1) thus provides information about the affinity of the peptide for PIP 2 presented in a physiologically relevant PC/PS/PIP 2 surface. This information is difficult to obtain by conventional binding techniques because the peptide also binds strongly to surfaces containing physiological concentrations of monovalent acidic lipids like PS (25).
The binding of MARCKS-(151Ϫ175) to vesicles containing monovalent acidic lipids such as PS involves three forces: a long range Coulomb attraction, a short range Born/dehydration repulsion, and a short range hydrophobic attraction of the aromatic residues for the polar head group region of the bilayer (25). Although these three forces also must be important in mediating the interaction of the peptide with membranes containing PIP 2 , this binding is not yet well understood at a molecular level. Additional experimental information is required to understand the role of the 5 Phe and other specific residues in the binding of MARCKS-(151Ϫ175) to PIP 2. Ongoing NMR experiments, EPR measurements with spin-labeled PIP 2 , 2 and our quenching measurements with fluorescent PIP 2 should provide useful information. The direct binding measurements we report here, however, are sufficient to show that the effector domain of MARCKS binds to PIP 2 with high affinity (Figs. 2Ϫ5) and sequesters the PIP 2 away from the catalytic domain of PLC (Fig. 1). Extrapolating from our observations on the MARCKS protein and effector domain peptide in model systems, we hypothesize that the effector domain of MARCKS reversibly sequesters a significant fraction of the PIP 2 in the plasma membrane of cells.
Biological Implications of MARCKS Effector Domain Binding to PIP 2 - Fig. 6 illustrates what is known about the binding of MARCKS to membranes. Work from many different laboratories has shown the MARCKS protein binds to phospholipid vesicles-and by implication to plasma membranes in cellsthrough hydrophobic interaction of its N-terminal myristate with the interior of the bilayer and electrostatic interaction of its basic effector domain (residues 151Ϫ175) with acidic lipids (Refs. 54Ϫ58; reviewed in Refs. 16 and 17). Fig. 6 illustrates the 13 positively charged residues (blue plus signs) and 5 aromatic residues (green ovals) in the effector domain; much of what we know about the interaction of the effector domain with membranes comes from experiments with the MARCKS-(151Ϫ175) peptide, which appears to bind in a similar manner to the effector region in the intact protein (reviewed in Ref. 17). The regions of MARCKS flanking the effector domain contain negatively charged residues (red minus signs) and should be repelled from the negatively charged surface. What is new is the recognition that the effector region can bind PIP 2 (lipids with red circles in Fig. 6) with high affinity; in a PC/PIP 2 membrane, Ϸ4 PIP 2 /peptide are bound. Biological membranes such as the plasma membrane also contain monovalent acidic lipids that may affect the binding stoichiometry, so the exact number of PIP 2 bound is not known. It must be significant, however, because both the peptide and protein inhibit PLC-catalyzed hydrolysis of PIP 2 on membranes that contain 33% monovalent 2 D. Cafiso, personal communication. acidic lipids (see Fig. 1 and Ref. 18).
There are several corollaries to our hypothesis that MARCKS binds a significant fraction of PIP 2 in the plasma membrane and releases it upon interaction with Ca 2ϩ /calmodulin or phosphorylation by PKC. First, if MARCKS is to bind a significant fraction of the PIP 2 , it must be present in cells at concentrations comparable to PIP 2 ; it is in many cell types (e.g. 10 M in brain, Refs. 13 and 15). Second, if reversible binding of PIP 2 by MARCKS is important for phosphoinositide function, overexpression of MARCKS should stimulate increased production of PIP 2 to maintain a constant free concentration of PIP 2 in the membrane; this has been observed in some cells (59). Third, in cells where MARCKS is distributed nonuniformly in the plasma membrane (presumably because of protein-protein interactions, possibly with actin), PIP 2 should be colocalized with MARCKS; in fibroblasts, both MARCKS and PIP 2 are concentrated in membrane ruffles (60Ϫ62). Fourth, the mechanism by which MARCKS is targeted selectively to the plasma membrane is unknown; the interaction of the effector domain with PIP 2 documented here could contribute to this targeting, as could the less specific electrostatic targeting mechanism that Silvius and co-workers (63) recently proposed with respect to the targeting of Ki-Ras4B to the plasma membrane. Fifth, and perhaps most importantly, local activation of PKC or calmodulin should produce a local increase in the free concentration of PIP 2 in the plasma membrane; although this has been observed in model systems (see Fig. 3