Binding of peptides with basic and aromatic residues to bilayer membranes: phenylalanine in the myristoylated alanine-rich C kinase substrate effector domain penetrates into the hydrophobic core of the bilayer.

Electrostatic interactions with positively charged regions of membrane-associated proteins such as myristoylated alanine-rich C kinase substrate (MARCKS) may have a role in regulating the level of free phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) in plasma membranes. Both the MARCKS protein and a peptide corresponding to the effector domain (an unstructured region that contains 13 basic residues and 5 phenylalanines), MARCKS-(151-175), laterally sequester the polyvalent lipid PI(4,5)P2 in the plane of a bilayer membrane with high affinity. We used high resolution magic angle spinning NMR to establish the location of MARCKS-(151-175) in membrane bilayers, which is necessary to understand the sequestration mechanism. Measurements of cross-relaxation rates in two-dimensional nuclear Overhauser enhancement spectroscopy NMR experiments show that the five Phe rings of MARCKS-(151-175) penetrate into the acyl chain region of phosphatidylcholine bilayers containing phosphatidylglycerol or PI(4,5)P2. Specifically, we observed strong cross-peaks between the aromatic protons of the Phe rings and the acyl chain protons of the lipids, even for very short (50 ms) mixing times. The position of the Phe rings implies that the adjacent positively charged amino acids in the peptide are close to the level of the negatively charged lipid phosphates. The deep location of the MARCKS peptide in the polar head group region should enhance its electrostatic sequestration of PI(4,5)P2 by an "image charge" mechanism. Moreover, this location has interesting implications for membrane curvature and local surface pressure effects and may be relevant to a wide variety of other proteins with basic-aromatic clusters, such as phospholipase D, GAP43, SCAMP2, and the N-methyl-d-aspartate receptor.

Many peripheral membrane proteins contain unstructured clusters of basic and aromatic residues that interact with lipid bilayers. In some cases, these clusters can bind phosphatidylinositol 4,5-bisphosphate (PI(4,5)P 2 ) 1 (Fig. 1C), a key regulator of signal transduction in cell membranes (for reviews, see Refs. [1][2][3][4][5][6][7][8]. For instance, binding of PI(4,5)P 2 to the basic-aromatic cluster on phospholipase D activates the enzyme (9 -11), whereas a similar cluster in the effector domain (residues 151-175) of the myristoylated alanine-rich C kinase substrate (MARCKS) protein may act as a reversible buffer for regulating free PI(4,5)P 2 levels in the plasma membrane (2). Peptides corresponding to basic-aromatic clusters of phospholipase D and MARCKS bind PI(4,5)P 2 with high affinity and specificity compared with monovalent acid lipids such as phosphatidylserine (11)(12)(13). To establish how these clusters can bind to the membrane, selectively sequester PI(4,5)P 2 , and possibly exert other effects on membrane structure, we need to establish their specific location within the bilayer.
This study focuses on the effector domain of the MARCKS protein, one of the major substrates for protein kinase C (for reviews, see Refs. 14 -16). MARCKS associates with membranes through an N-terminal myristoyl group and a cluster of basic-aromatic amino acids in its effector domain (16). The five Phe residues in the basic effector domain may influence membrane binding or structure in several different ways, depending on the degree to which they penetrate the membrane bilayer. For example, as we discuss in more detail below, they could (i) contribute a hydrophobic term to the membrane-binding energy, (ii) enhance the electrostatic potential due to basic residues in the cluster, or (iii) induce membrane curvature. Determining the depth of the Phe residues within the membrane is the first step in understanding why they are often associated with basic residues in peripheral membrane proteins.
An array of biophysical approaches has shed light on the conformation and location of the MARCKS-(151-175) peptide bound to lipid membranes. EPR spectroscopy (17) and circular dichroism (13) have shown that this peptide adopts an extended conformation when bound to membrane bilayers. Cafiso and co-workers (18,19) have undertaken an extensive series of EPR studies on MARCKS-(151-175) to probe its location relative to the membrane surface. They systematically substituted amino acids in the MARCKS-(151-175) sequence with cysteines derivatized with nitroxide spin labels and found that spin labels attached at the highly basic and hydrophilic N and C termini of the peptide reside on the aqueous side of the lipid phosphate head group, whereas spin labels attached to the central Phe-containing portions of the peptide are located several angstroms below the lipid head group (17).
In this study, we used a complementary technique, magic angle spinning (MAS) NMR spectroscopy, to characterize the specific membrane location of amino acids in the membranebound MARCKS-(151-175) peptide. MAS is a solid-state NMR technique for obtaining high resolution NMR spectra of membrane proteins and peptides (20 -22). Typically, protons are not observed directly in solid-state experiments because MAS cannot average strong 1 H-1 H dipolar couplings. However, because membrane lipids exhibit a high degree of rotational and segmental motion in the liquid-crystalline phase, the proton dipolar couplings are partially averaged. Oldfield et al. (23) were the first to fully recognize that high resolution 1 H spectra of lipid membranes could be obtained using MAS. More recently, several groups have extended these methods to study peptides bound to membranes (24 -27). We used two-dimensional nuclear Overhauser enhancement spectroscopy (NOESY) in combination with MAS to investigate the location of the five Phe residues of the MARCKS effector domain within the negatively charged lipid bilayer by observing the correlations between protons of the lipid groups and protons on the aromatic phenylalanine side chains.
Peptide Synthesis and Purification-Peptides corresponding to the effector domain of the MARCKS protein (residues 151-175) were synthesized by solid-phase methods at the Center for the Analysis and Synthesis of Macromolecules at Stony Brook University or by Midwest Biotech (Fishers, IN). The sequence of MARCKS-(151-175) is 151 KKKKKRFSFKKSFKLSGFSFKKNKK 175 . Synthetic peptides were purified by reverse-phase high pressure liquid chromatography (Varian Prostar) on a C 18 column with an acetonitrile:water gradient and lyophilized. The solvents contained 0.1% (w/v) trifluoroacetic acid. The purity was confirmed by matrix-assisted laser desorption ionization mass spectrometry.
Sample Preparation-The samples containing DMPC and DMPG were prepared by first co-dissolving the lipids in chloroform and removing the chloroform with a flow of argon gas. The lipids were then dissolved in cyclohexane and lyophilized to form a fluffy powder. The powder was hydrated with H 2 O or D 2 O buffer (30 mM NaCl and 20 mM HEPES, pH 7) at 30°C. For DMPC:DMPG samples containing phenylalanine methyl ester or the MARCKS peptides, the lyophilized DMPC: DMPG powder was rehydrated with buffer containing peptides. The molar ratio of DMPC to DMPG was 100:30, and the molar ratio of total lipid to peptide was ϳ100:1. After hydration, the samples were freezethawed and bath-sonicated five times and then centrifuged at 265,000 ϫ g for 1 h. The pelleted multilayers were used for NMR experiments.
The preparation of lipid membranes containing PI(4,5)P 2 and bound MARCKS peptide was carried out as follows. The PI(4,5)P 2 solution in chloroform:methanol:water was dried in a Teflon vial with a flow of argon gas; residual solvent was removed by placing the sample under vacuum for 2 h. PI(4,5)P 2 was redissolved in chloroform and mixed with other lipids by vortexing in a 100-ml glass flask. After incubation at 35°C for 20 min, the solvent was removed by rotary evaporation under vacuum. When the solvent was almost completely evaporated, the vacuum was increased to allow the remaining solvent to distill quickly. This procedure provides an even distribution of PI(4,5)P 2 in the bulk lipid on the bottom of the flask (12). Residual solvent was removed by placing the sample overnight under vacuum. The lyophilized MARCKS effector domain peptide (ϳ1 mg) was dissolved in 1 ml of buffer (30 mM NaCl and 20 mM HEPES, pH 7) and added to the lipid mixture. Multilamellar vesicles were formed by slowly swirling and vortexing with a glass bead and then centrifuged at 265,000 ϫ g for 1 h. In each sample, the molar ratio between PI(4,5)P 2 and the MARCKS effector domain was ϳ3:1. For a typical experiment, PI(4,5)P 2 was 4% of the total membrane lipid. This yielded a peptide:lipid molar ratio of 1.3:100.
Under our NMR conditions, where the lipid concentration was high, all the peptide was bound. Independent potential experiments using low concentrations of lipid showed that the MARCKS peptide bound with comparable affinity to vesicles formed with lipids having either saturated or unsaturated chains. Specifically, addition of 3 M peptide decreased the potential of 3:1 DMPC:DMPG (or POPC:1-palmitoyl-2oleoylphosphoglycerol) vesicles formed in 100 mM KCl, pH 7, at 30°C from Ϫ46 (or Ϫ43) mV to Ϫ6 (or Ϫ2) mV (data not shown). Moreover, in our NMR experiments, the peptide bound to both sides of the lipid membranes, so binding should not induce curvature in the multilamellar vesicles.
NMR Spectroscopy-For NMR experiments, the pelleted multilayers were lyophilized, rehydrated with D 2 O or H 2 O to ϳ50% (w/v), and transferred to 4-mm NMR sample rotors. High resolution MAS 1 H NMR experiments were performed at 600 or 700 MHz on Bruker Avance NMR spectrometers using 4-mm MAS rotors. The temperature was maintained at 30°C, well above the phase transition temperature of ϳ22°C for the DMPC:DMPG bilayers, for all experiments. The MAS frequency was regulated at 6000 or 6500 Ϯ 1 Hz. Two-dimensional MAS NOESY spectra were recorded in the phase-sensitive mode with 256 or 512 slices in the t 1 dimension. For all of the samples discussed below, two-dimensional NOESY spectra were obtained using a series of mixing times from 5 to 500 ms, and representative slices are typically shown for mixing times of 50 and 300 ms. Water presaturation was used on the 700-MHz spectrometer to attenuate the water signal. Experiments on the 600-MHz spectrometer were carried out in D 2 O.
Simulations of spin diffusion were carried out using a one-dimensional lattice model adapted to lipid-protein systems by Kumashiro et al. (26) and Huster et al. (27). The rate of magnetization transfer (⍀ ϭ D/r 2 ) is a function of the distance (r) between protons and the diffusion coefficient (D). For a given spin (i) in a one-dimensional lattice, the magnetization buildup is given by the following discrete diffusion equation: ⌬M i /⌬t ϭ Ϫ2⍀M i ϩ ⍀M iϩ1 ϩ ⍀M iϪ1 . The 1 H . . . 1 H distance in the simulations was 2 Å as used by Kumashiro et al. (26) and Huster et al. (27). However, the diffusion coefficient (D) was smaller, consistent with the much more fluid membranes used in our studies. We chose eight steps to model the transfer from the aromatic ring protons in phenylalanine methyl ester to the lipid acyl chain protons and varied D to obtain the best fit. The fit is not unique; the simulations presented illustrate the difference between one-step and multistep magnetization transfer. Fig. 2 presents the two-dimensional MAS 1 H NOESY spectrum of DMPC:DMPG multilayers obtained with a mixing time of 300 ms. DMPC and DMPG have comparable phase transition temperatures and are co-miscible. Mixtures of these lipids do not phase-separate (28). At the DMPC:DMPG molar ratio of 100:30 used in our experiments, the net membrane surface charge is comparable to that of the cytoplasmic leaflet of typical mammalian plasma membranes. In the two-dimensional spectrum The resonances on the diagonal correspond to the isotropic 1 H resonances that would appear in a conventional one-dimensional NMR spectrum. The most intense diagonal peak at ϳ1.3 ppm corresponds to the methylene protons of the lipid acyl chain. Cross-peaks are observed between most of the proton resonances that lie on the diagonal, and their intensities in a NOESY spectrum depend on through-space dipolar couplings. The dipolar couplings and corresponding cross-peak intensities are strong for protons that are in close proximity to one another. Fig. 3A shows 10 slices from the two-dimensional spectrum presented in Fig. 2. The slices are arranged by chemical shift from 5.28 to 0.89 ppm. The resonances are labeled 1-10 according to the labeling in Fig. 1. The diagonal peak in each slice is indicated by an asterisk. The intensities of the cross-peaks are related to the cross-relaxation rate in the NOESY experiment and the NOESY mixing time. The cross-relaxation rate is, in turn, determined by the internuclear distance and by molecular motions. The cross-relaxation rate is increased dramatically for short distances due to the r Ϫ6 distance dependence, whereas molecular motions with correlation times of Ͼ100 ns (e.g. lateral diffusion) are thought to be the dominant components for mediating relaxation (29). Fig. 3B shows the slices from the DMPC:DMPG two-dimensional 1 H NOESY spectrum collected with a mixing time of 50 ms. The drawback of using very short mixing times is that the intensity for the cross-peaks is smaller than with longer mixing times. The advantage is that these cross-peak intensities provide a more reliable indicator of protons in close proximity because contributions from spin diffusion are less important. In Fig. 3B, the individual slices are plotted such that the diagonal peaks are off-scale and the most intense cross-peak in each slice is roughly full-scale.

High Resolution MAS NMR of DMPC:DMPG-
The one-dimensional slices of the DMPC:DMPG NOESY spectra in Fig. 3 provide a qualitative guide to the location of protons in the lipid membrane because the intensities of the cross-peaks are related to the proximity of neighboring protons. For instance, in Fig. 3A (second spectrum), the resonance at 5.28 ppm corresponds to the proton on C-2 of the glycerol backbone (labeled as 1 in Fig. 1) and exhibits cross-peaks to the other glycerol protons (peaks 2 and 4) and to the choline protons (peaks 3 and 6). These are the protons that are closest in space. There is also a strong cross-peak to the acyl chain protons (peak 9). However, the acyl chain protons are, on av-erage, quite distant from the glycerol protons, and this crosspeak intensity results mainly from spin diffusion. This can be seen in the weaker intensity of the cross-peak (peak 9) at a mixing time of 50 ms (Fig. 3B, second spectrum) compared with the cross-peak at 300 ms (Fig. 3A, second spectrum). In contrast, the terminal methyl resonance at 0.89 ppm exhibits a significant cross-peak to only the acyl chain methylene resonance at 1.3 ppm at both 50 ms (Fig. 3B) and 300 ms (Fig. 3A). There is virtually no cross-relaxation between the terminal methyl protons and protons on the glycerol and choline head group. The distance between the terminal methyl and glycerol protons is considerably longer than the distance between the acyl chain and glycerol protons, and 300 ms is not long enough to allow for magnetization to diffuse along the full length of the lipid acyl chains. Gawrisch and co-workers (30,31) have shown that intrachain spin diffusion is slow and generally requires mixing times of Ͼ300 ms.
The acyl chain protons at positions 7 and 8 in Fig. 1 have resolved chemical shifts at 2.34 and 1.6 ppm, respectively. They show a dominant cross-peak to the main acyl chain resonance at 1.3 ppm and also exhibit weak cross-peaks to the protons of the glycerol backbone, lipid head group, and terminal methyl group with a mixing time of 300 ms.
High Resolution MAS NMR of Phenylalanine Methyl Ester-The one-dimensional slices in Fig. 3 provide a qualitative ruler for establishing the depth of peptides in the lipid membrane. If the peptides are buried in the hydrocarbon interior of the membrane, the peptide protons should exhibit cross-peaks predominantly to the protons of the lipid acyl chains, whereas if the peptides are bound in the polar head group region of the bilayer, they should exhibit cross-peaks predominantly to the protons of the glycerol backbone and choline head group. Peptides that bind outside of the polar head group, e.g. polylysine, should exhibit cross-peaks to only the phosphocholine head group. This approach assumes that the mixing times are sufficiently short to limit spin diffusion and that molecular motions of the peptide and lipid allow efficient cross-relaxation. We tested our approach of using peptide-lipid cross-peak intensities as a measure of proximity by obtaining a series of MAS NOESY spectra of phenylalanine methyl ester. This small molecule has a single aromatic ring and a single positive charge on the unprotected amino terminus (see inset to Fig. 5). The hydrophobic phenylalanine side chain has a favorable free energy for partitioning into the head group region, whereas the Nterminal positive charge should be attracted to, but not penetrate, the negatively charged membrane surface (32). Based on equilibrium dialysis of model tripeptides (e.g. Ala-X-Ala(O-tertbutyl)) with a charged N terminus and a protected C terminus, White and Wimley (32) estimated a free energy of Ϫ1.13 kcal/ mol for partitioning of the aromatic side chain of Phe into the head group region of POPC. These peptides localize at the membrane interface in the region of the choline head group. Positively charged amino acids, e.g. lysine and arginine, at the X position in these model tripeptides have free energies of approximately ϩ1 kcal/mol for partitioning into the head group region of POPC, suggesting that the positive charge on our phenylalanine methyl ester model compound should not penetrate the phosphocholine region of the bilayer. Fig. 4 presents the two-dimensional MAS 1 H NOESY spectrum of phenylalanine methyl ester in DMPC:DMPG multilayers. The spectrum is comparable to that of DMPC:DMPG alone with the exception of the large resonances at 7.30, 7.24, and 7.12 ppm, which are readily assigned to the aromatic protons on the phenylalanine ring. There is also a substantial peak at 4.7 ppm resulting from water protons not removed by proton presaturation.  Fig. 1 (31). The MAS frequency was 6000 Hz. The temperature was maintained at 30°C for all experiments.  Fig. 1, respectively. The resonance at 3.4 ppm marked with an asterisk results from the ␤-CH 2 group of the phenylalanine side chain. The short (50 ms) mixing time limits spin diffusion, and the spectrum indicates that the aromatic side chain is bound in the region of the choline head group, as expected (Fig. 5A). When the mixing time was increased to 300 ms (Fig. 5B), additional intense resonances occurred at 4.7 and 1.3 ppm, which are assigned to water and the lipid acyl chain CH 2 resonances, respectively. The observation of cross-peaks between the aromatic protons of phenylalanine methyl ester and protons on the lipid acyl chains suggests that spin diffusion occurs at longer mixing times. Fig. 6A presents a plot of the intensities of the acyl chain CH 2 resonance at 1.3 ppm as a function of the NOESY mixing time. Fig. 6A also presents simulations of magnetization transfer to illustrate how spin diffusion would influence the buildup curves of cross-peak intensity in the NOESY spectrum. The spin diffusion process was simulated using a one-dimensional lattice model (26, 27)  steps (solid line). The rate of magnetization transfer depends on the 1 H . . . 1 H distance for each step and the diffusion coefficient. We assumed a 2-Å distance and a diffusion constant of 0.09 Å 2 /ms (see "Experimental Procedures"). We estimated that approximately eight transfer steps are needed to relay magnetization from the choline head group to the acyl chain protons. The simulations show that, for one step, there is a rapid buildup of intensity that saturates after ϳ50 -100 ms, whereas for eight steps, magnetization builds up slowly on the last spin in the one-dimensional chain.
The NOE buildup curves for the choline methyl protons at 3.2 ppm (Fig. 6B) and for the ␣-CH 2 protons at 4.3 ppm (Fig.  6C) are roughly linear, indicating that these protons are closer in space to the aromatic protons of phenylalanine methyl ester than the acyl chain protons. The slight negative curvature seen in Fig. 6C suggests that the aromatic ring of phenylalanine methyl ester is below the choline methyl protons and near the level of the choline CH 2 protons. Together, the NOE buildup curves and simulations support the use of peptide-lipid crosspeak intensities at short mixing times as a measure of proximity.
High Resolution MAS NMR of MARCKS-(151-175) Bound to DMPC:DMPG-We performed two-dimensional MAS 1 H NOESY experiments with MARCKS-(151-175) bound to DMPC:DMPG bilayers under the same conditions used for the phenylalanine methyl ester experiments. The ratio of DMPC to DMPG was 100:30, and the ratio of peptide to total lipid was ϳ1:100. The 1 H NOESY spectra of DMPC:DMPG multilayers and phenylalanine methyl ester bound to DMPC:DMPG membranes provide points of reference for comparison. Fig. 7 (A and B) presents the aromatic slices of the twodimensional 1 H NOESY spectra of MARCKS-(151-175) obtained with mixing times of 50 and 300 ms, respectively. As for phenylalanine methyl ester, the protons from the aromatic ring of the five Phe residues in MARCKS-(151-175) are easily assigned to the intense resonance at ϳ7.3 ppm. The most intense cross-peak in both spectra corresponds to the acyl chain CH 2 resonance at 1.3 ppm. The cross-peak at 1.3 ppm is, in fact, the most intense cross-peak at mixing times from 50 to 500 ms. The NOE buildup curve for this cross-peak is shown in Fig. 6D. The rapid rise in intensity at 50 ms is consistent with direct internuclear cross-relaxation rather than spin diffusion. Comparison of the cross-peak intensities at 50 ms in Fig. 7A with those of Phe methyl ester in Fig. 5A provides information about the location of the Phe rings in MARCKS-(151-175) in the lipid bilayer. The simplest interpretation is that the Phe rings of MARCKS are buried more deeply in the bilayer. Specifically, the cross-peak at 1.3 ppm that dominates the spectrum in Fig. 7A is absent in Fig. 5A. (The resonance at 3.2 ppm in Fig. 7 is more likely to be a cross-peak to the ␤-CH 2 of the phenylalanine ring rather than to the choline methyl groups, given the intensity of the acyl chain cross-peak at 1.3 ppm.) Comparison of the relative cross-peak intensities in Fig. 7 with those in Fig. 3 indicates that, on average, the aromatic Phe protons in MARCKS-(151-175) are located closer to the lipid acyl chains than to the glycerol backbone. Specifically, there is  Fig. 3.
In contrast to phenylalanine methyl ester, MARCKS-(151-175) has five aromatic Phe groups that are not individually resolved. This raises the possibility that not all five Phe amino acids reside in the same deep location in the membrane. The rapid rise of the NOE buildup curve in Fig. 6D suggests, however, that the majority of the Phe rings are in direct contact with the lipid acyl chains.
High Resolution MAS NMR of an Alanine-substituted Version of MARCKS-(151-175) Bound to DMPC:DMPG-Alaninesubstituted versions of MARCKS-(151-175) have been used to characterize the contribution of Phe to the membrane binding affinity and position of the peptide in membranes (12,18,33). Replacing all five Phe residues with alanine both reduces the binding affinity by 10 -100-fold (12,33) and produces shallower penetration by nitroxide spin labels on the peptide (18). We tested whether the deep penetration of MARCKS-(151-175) requires all five Phe groups by synthesizing a version of the peptide with the two N-terminal Phe residues retained and the other three Phe residues replaced by Ala ( 151 KKKKKRFS-FKKSAKLSGASAKKNKK 175 ). Fig. 8 (A and B) presents the slices of the two-dimensional 1 H NOESY spectra at the level of the Phe protons obtained at 50 and 300 ms, respectively. The spectra are comparable to those shown in Fig. 7: the cross-peak at 1.3 ppm dominates the spectra, indicating that the two Phe rings are in roughly the same position as they are in the native MARCKS peptide. We obtained comparable results with an alanine version of the MARCKS peptide with only the two C-terminal Phe groups in the effector domain ( 151 KKKKKRASAKKSAKLSG-FSFKKNKK 175 ) (data not shown). Together, these data indicate that two Phe residues are sufficient to anchor these side chains in the hydrophobic core of the membrane and provide support for the idea that all five Phe residues in the MARCKS effector region have a similar position. Hence, our model (see Fig. 10) assumes that the aromatic phenylalanine rings penetrate to the level of the acyl chains, drawing the immediately adjacent basic residues into the polar head group region of the bilayer. The available evidence from EPR studies suggests that the highly charged N-terminal end of the MARCKS effector domain, which lacks any aromatic or hydrophobic residues, resides outside the envelope of the polar head group region and interacts electrostatically with acidic lipids (18).
This location is consistent with the location of pentalysine bound to membranes, as deduced from "molecular voltmeter" NMR experiments (34).
High Resolution MAS NMR of MARCKS-(151-175) Bound to Protonated POPC:PI(4,5)P 2 -The spectra presented above were obtained on DMPC:DMPG lipids, which have saturated acyl chains. We performed experiments on more biologically relevant lipid bilayers, specifically POPC membranes with 4 mol % PI(4,5)P 2 . POPC contains an unsaturated acyl chain and consequently provides a better mimic of native plasma membranes. PI(4,5)P 2 has a net negative charge of Ϫ3 or Ϫ4 at pH 7.0 (35) and was the sole negatively charged lipid in these experiments. The peptide binds with high affinity to phosphocholine:PI(4,5)P 2 bilayers, with each peptide binding ϳ3 PI(4,5)P 2 molecules electrostatically (12,36). Fig. 9 (A and B) presents the slices of the two-dimensional 1 H NOESY spectra at the level of the Phe protons obtained at 50 and 300 ms, respectively. The spectra are very similar to those in Figs. 7 and 8. The most intense cross-peak, even at very short mixing times, corresponds to the acyl chain methylene protons at 1.3 ppm. The resonances at 2.0 and 5.3 ppm have been assigned to the CH 2 -CHϭCH-CH 2 protons of the unsaturated oleoyl chain (37) and generate cross-peaks at longer mixing times. These comparisons indicate that the penetration of the MARCKS peptide into the membrane does not depend on the acyl chain unsaturation or the nature of the negatively charged lipid.

Penetration of Aromatic Phenylalanine Rings into Membrane
Bilayers-The intensities of the cross-peaks in the NOESY experiment are a function of three variables: (i) the internuclear distance between dipole-coupled spins, (ii) molecular motions that allow for cross-relaxation, and (iii) the NOESY mixing time. Cafiso and co-workers (38,39) were the first to use NOESY spectra to investigate the structure of lipid membranes. More recently, NOESY has been exploited to establish the location of small hydrophobic molecules (40,41) and peptides (24,25,42) in membrane bilayers based on the observation of peptide-lipid or small molecule-lipid cross-relaxation. The challenge in these studies has been to distinguish direct contacts from those arising from spin diffusion to correlate cross-peak intensity with internuclear distance in a quantitative fashion.
Our approach has been to use the cross-peaks from the lipid resonances as a qualitative guide to depth in the bilayer. The slices of the two-dimensional 1 H NOESY spectra shown in Fig.   FIG. 7. Slices through the  3 illustrate that the pattern of cross-peak intensities is very different as one moves from the methyl groups on the choline head group to the terminal methyl groups on the acyl chains. We propose that this pattern of relative cross-peak intensities provides a measure of depth. For instance, the protons at C-1 and C-3 of glycerol produce strong cross-peaks to the protons at C-2 of glycerol. We expect that if the Phe protons of a membrane-bound peptide reside at the level of the glycerol backbone (or phosphocholine head group), the aromatic-glycerol (or aromatic-choline) cross-peaks should dominate the spectrum. We tested our proposal by performing a series of 1 H NOESY experiments with phenylalanine methyl ester, a simple model compound. Comparing the slices of the two-dimensional spectrum of phenylalanine methyl ester with those of the lipids indicates that the aromatic Phe side chain resides at the level of the choline head group. This is the predicted position if the N-terminal charge is anchored at the membrane-solution interface, a location expected from energetic considerations and prior work with the charged form of the simple fatty acids (43)(44)(45). This surface position is further supported by a strong cross-peak between the Phe ring protons and water at long mixing times.
Comparison of the aromatic slices of the two-dimensional NOESY spectra of MARCKS-(151-175) with those of phenylalanine methyl ester (Figs. 5 and 7) indicates that the phenylalanine rings of the peptide penetrate more deeply into the bilayer. The most intense cross-peaks in the two-dimensional NOESY spectra are to the protons on the lipid acyl chains, even at the shortest mixing times used in our experiments (Fig. 7A). One difference between the phenylalanine methyl ester model compound and MARCKS-(151-175) is that the positive charges on the peptide are at the end of long lysine or arginine side chains of hydrophobic CH 2 groups that may be able to "snorkel" through the head group region to the membrane surface (46,47). Fig. 10 illustrates the location we propose for the Phe rings in MARCKS. The rings reside just below the acyl chain carbonyl groups. This represents an average position because we cannot resolve the five phenylalanine rings separately. However, the similarity between the spectra of MARCKS-(151-175) and those of alanine-substituted versions of the peptide with only two Phe rings suggests that all five Phe rings have approximately the same location.
The proposed location of the aromatic Phe side chains based on NMR measurements agrees with the location indicated by spin labeling studies, which place the backbone of the MARCKS effector domain 2-5 Å below the level of the lipid phosphate (19). In addition, the location is consistent with the observation that tryptophan in amphipathic ␣-helices that lie on the membrane surface can penetrate the acyl chain region of bilayers (48,49). Finally, if the charges on the Lys adjacent to the Phe amino acids are buried in the polar head group region and are located at the level of the lipid phosphates, then they should lead to a reorientation of the phosphocholine head group in response to the local surface charge. Such a reorientation has been observed in deuterium NMR experiments of MARCKS-(151-175) bound to POPC:1-palmitoyl-2-oleoylphosphoserine (18,19) using deuterated lipids as a molecular voltmeter (50), but not in pentalysine bound to the surface of the membrane (34).
Implications for Function of Basic-Aromatic Clusters in Membrane-associated Proteins-Clusters of basic and aromatic residues are found in a wide range of membrane-associated proteins involved in signal transduction. Phospholipase D 2 has a cluster of basic and aromatic amino acids that binds PI(4,5)P 2 (10). Moreover, peptides corresponding to the basic-aromatic regions of phospholipase D, GAP43, SCAMP2, and the N-methyl-D-aspartate receptor also bind PI(4,5)P 2 (12).
As noted in the Introduction, the aromatic residues could play several different roles in membrane-protein interactions. They could (i) contribute hydrophobic energy to the binding, (ii) enhance the electrostatic potential, and (iii) influence membrane curvature. We consider these three effects separately below.
(i) Phenylalanine has a favorable free energy for partitioning into the polar head group region of membrane bilayers (32). There is also indirect evidence that the aromatic ring of Phe has a favorable free energy for partitioning into the acyl chain region of the bilayer. For instance, Phe has a much higher occurrence in the transmembrane segments of membrane proteins of known structure than either tyrosine or tryptophan (51). In an elegant study on the preferred position of Trp and Phe in membrane bilayers, Braun and von Heijne (52) found that Phe, but not Trp, is accommodated in the hydrophobic core of the bilayer when inserted into a poly-Leu transmembrane helix. Moreover, measurements on the partitioning of benzene into bilayers show that there is no preference for benzene partitioning into either the interfacial or hydrophobic core region of the bilayer (53,54). As expected, the deep penetration of the aromatic rings of the five Phe residues increases the energy of binding of MARCKS-(151-175) to bilayer membranes. The Phe rings increase the affinity by 10 -100-fold (12,33,36).
(ii) The aromatic amino acids in basic-aromatic clusters may influence the electrostatic interactions with the membrane surface (32) by dragging the positive charge on the flanking lysine and arginine side chains deeper into the polar head group so that they can interact more effectively with the negative charges on PI(4,5)P 2 or phosphoglycerol head groups (Fig. 10). The electrostatic potential extending from the positively charged residues would be expected to increase substantially as the charge approaches the low dielectric membrane surface. Electrostatic calculations on a simple ion of charge q at an idealized membrane-water interface show that if ions are excluded from the membrane phase, then the potential in the aqueous phase extends much farther than it would for the same charge in bulk water (55). In fact, the electrostatic potential at a distance r from an ion at a low dielectric interface is twice that predicted by Debye-Hü ckel theory for an identical ion in the bulk aqueous phase. This increase in electrostatic potential would enhance the ability of the MARCKS peptide to sequester PI(4,5)P 2 . Recently, we have shown using Förster resonance energy transfer that MARCKS-(151-175) is able to electrostatically sequester PI(4,5)P 2 with high selectivity (ϳ100-fold) over monovalent acidic lipids such as phosphatidylserine (56). The high selectivity is predicted from the high valence (Ϫ4) of PI(4,5)P 2 and the Boltzmann factor associated with the interaction between MARCKS-(151-175) and PI(4,5)P 2 (2).
(iii) Finally, the peptide must strongly perturb the polar head group region to penetrate into the acyl chain region of the membrane. As the aromatic rings become buried in the hydrophobic core of the membrane, the lipid head groups must be pushed aside by the backbone and the side chains of the peptide. Monolayer studies demonstrate that the increase in the surface pressure is small (0.3 mN/m) for approximately one MARCKS-(151-175) peptide bound per 100 lipids, but the surface pressure increases markedly as the number of peptides bound per unit area increases (data not shown). Hence, binding would be expected to result in significant membrane curvature in regions of the cell membrane where MARCKS is concentrated. Membrane curvature is critical for endocytotic vesicle formation, a process that is often mediated by proteins that specifically interact with PI(4,5)P 2 (57). MARCKS is concentrated in regions of cells associated with inward curvature, such as the nascent phagosomes of macrophages (58,59).
In summary, the strong cross-peaks to acyl chain protons observed in the two-dimensional NOESY spectra indicate that the five Phe rings of MARCKS-(151-175) penetrate into the acyl chain region of the bilayer. This location suggests that the adjacent positive charges on Lys and Arg residues in the MARCKS effector domain must be pulled into the polar head group region of the lipid bilayer, where they would be more effective in sequestering PI(4,5)P 2 . More generally, the approach described above opens up a novel way to assess how clusters of basic and aromatic residues function to recruit and bind PI(4,5)P 2 and related polyphosphoinositides (e.g. PI(3,4,5)P 3 ) involved in signal transduction processes.