Aromatic Residue Position on the Nonpolar Face of Class A Amphipathic Helical Peptides Determines Biological Activity*

The apolipoprotein A-I mimetic peptide 4F (Ac-DWFKAFYDKVAEKFKEAF-NH2), with four Phe residues on the nonpolar face of the amphipathic α-helix, is strongly anti-inflammatory, whereas two 3F analogs (3F3 and 3F14) are not. To understand how changes in helix nonpolar face structure affect function, two additional 3F analogs, Ac-DKLKAFYDKVFEWAKEAF-NH2 (3F-1) and Ac-DKWKAVYDKFAEAFKEFL-NH2 (3F-2), were designed using the same amino acid composition as 3F3 and 3F14. The aromatic residues in 3F-1 and 3F-2 are near the polar-nonpolar interface and at the center of the nonpolar face of the helix, respectively. Like 4F, but in contrast to 3F3 and 3F14, these peptides effectively inhibited lytic peptide-induced hemolysis, oxidized phospholipid-induced monocyte chemotaxis, and scavenged lipid hydroperoxides from low density lipoprotein. High pressure liquid chromatography retention times and monolayer exclusion pressures indicated that there is no direct correlation of peptide function with lipid affinity. Fluorescence studies suggested that, although the peptides bind phospholipids similarly, the Trp residue in 4F, 3F-1, and 3F-2 is less motionally restricted than in 3F3 and 3F14. Based on these results and molecular modeling studies, we propose that the arrangement of aromatic residues in class A amphipathic helical molecules regulates entry of reactive oxygen species into peptide-phospholipid complexes, thereby reducing the extent of monocyte chemotaxis, an important step in atherosclerosis.

Human apolipoprotein A-I (apoA-I), 1 the major protein component of high density lipoproteins (HDL), has been the subject of intense investigation in many laboratories because of its strong antiatherogenic properties (1)(2)(3)(4)(5)(6)(7). Two major theories on the role of HDL in the development of atherosclerosis are as follows: (i) reverse cholesterol transport (HDL transfers cholesterol from peripheral tissues to the liver for excretion) (5), and (ii) the lipid oxidation theory (it has been shown that HDL traps oxidized lipids that are responsible for the production of cytokines) (7). Several enzymes (notably platelet-activating factor acetylhydrolase and paraoxonase) that are present on the surface of HDL are responsible for destroying the biological activity of oxidized lipids (6). Different strategies have been used by several laboratories to support both theories (8,9). However, recently, it has been suggested that the two mechanisms may be two sides of the same coin (10).
We have used apoA-I mimetic class A amphipathic helical peptides to understand the functions of apoA-I that may be playing a primary role in inhibiting atherosclerosis (10,11). With this approach, we have shown that either intraperitonial administration of the class A amphipathic helical peptide 5F or the oral administration of peptide 4F synthesized from all-Damino acids inhibit atherosclerosis in dyslipidemic mouse models without altering plasma cholesterol levels (12,13). Furthermore, investigations using apoA-I peptides support the concept that their ability to inhibit atherosclerosis is related to their ability to remove "seeding molecules" (reactive lipid hydroperoxides formed by the oxidation of phospholipids containing arachidonic acid) from the low density lipoprotein (LDL) surface, thereby inhibiting free radical oxidation chain reactions and the subsequent propagation and release of cytokines (14). Preliminary studies suggest that oral administration of D-4F to LDL receptor null and apoE null mice causes the rapid formation and clearance of small HDL-like particles containing peptide, cholesterol, and apoA-I and paraoxonase, capable of converting proinflammatory HDL into anti-inflammatory HDL (15). These results, coupled with the previous observations that class A amphipathic helical peptide associates with HDL (16), suggest that the peptide is active in a lipid-associated form.
However, not all class A amphipathic helical peptides are anti-inflammatory to the same extent (17,18). Using a series of peptides in which the hydrophobicity of the nonpolar face was increased by substituting Ala, Leu, or Val residues with Phe residues, we have shown that greater hydrophobicity per se is not sufficient to increase the anti-inflammatory properties of a class A amphipathic helical peptide (17). The addition of one more Phe to the nonpolar face of Ac-18A-NH 2 (2F) either at position 3 (3F 3 ) or at 14 (3F 14 ) (Fig. 1) resulted in peptides that were no longer able to inhibit LDL-induced monocyte chemotactic activity (17). However, a peptide analog containing four Phe residues on the nonpolar face (4F) was the most active in this series in inhibiting LDL-induced monocyte chemotactic activity, whereas the extensively studied 2F analog was not as effective. Furthermore, 2F did not inhibit atherosclerosis in a diet-induced mouse model of atherosclerosis (19), whereas 4F was highly effective (18). The further addition of Phe residues (5F, 6F, and 7F) did not enhance this property. We therefore hypothesized that a particular arrangement of aromatic residues on the nonpolar face is more important in producing a peptide with maximum anti-inflammatory properties than is the overall hydrophobicity of the peptide.
To investigate this, we have synthesized two additional an-alogs of 3F. In one analog, with the amino acid sequence Ac-DKLKAFYDKVFEWAKEAF-NH 2 (referred to as 3F-1), the aliphatic residues are at the center of the nonpolar face, and aromatic residues are near the interfacial Lys residues, whereas in another 3F analog with the amino acid sequence Ac-DKWKAVYDKFAEAFKEFL-NH 2 (referred to as 3F-2), the -electron-containing aromatic residues are at the center of the nonpolar face (see Fig. 1A). The primary sequence and wheel projection of these peptides along with those of 2F, 3F 3 , 3F 14 , and 4F for comparison are shown in Fig. 1A. The space-filling models of these peptides with the Lys residues snorkeled (20,21) are shown in Fig. 1B. We have studied the physical-chemical and anti-inflammatory properties of these two peptides and 4F in comparison with 3F 3 and 3F 14 . The results show that although all of the peptides have similar abilities to form an amphipathic helix, the two peptides 3F-1 and 3F-2 are biologically active, similar to 4F. This effect appears to be related to FIG. 1. A, helical wheel representation of the parent molecule 2F, the antiatherogenic 4F, and the four 3F analogs. The wheel is projected along the axis of the helix from N to C terminus with the hydrophobic side facing downward. The primary structure is given above each wheel diagram. The amino acid composition of each of these peptides is the same, but the sequences are different. The plus and minus signs denote the charge on the amino acids at neutral pH. Red denotes an acidic residue, blue denotes a basic residue, and boldface black denotes aromatic residues. The line shows the interface between the hydrophobic face (lower part) and the hydrophilic face (upper half) as given by us earlier (20). Although the hydrophobic faces appear to be similar in each peptide, their abilities to allow water to penetrate into the lipid are different. B, space-filling models of the parent peptide, 2F, the antiatherogenic peptide, 4F, and the four 3F peptides. These models were created using silicon graphics; the lysine residues were snorkeled (22), and the models were energy-minimized. The view down the long axis of the helix shows that 2F, 3F 3 , and 3F 14 are wedge shaped, whereas 4F, 3F-1, and 3F-2 are more cylindrical in shape. The blue color in the space-filling model denotes a nitrogen atom, whereas the red color denotes an oxygen atom. the position of aromatic residues with respect to the surface of lipoprotein particles, which affect molecular arrangements near the interface and alter the lateral interactions within the particle. Changes in these interactions could promote the transfer of oxidized lipids (found on the surface of LDL particles) to HDL-like particles, thus making LDL less effective in inducing monocyte chemotaxis, an important step in the initiation of atherogenesis.
Hemolysis Assay-The hemolysis assay was carried out as described by Tytler et al. (22). Briefly, red blood cells (RBCs) were collected from EDTA-treated human blood by centrifugation. The cell pellet was washed three times with phosphate-buffered saline to remove plasma and the buffy coat. A suspension of 1% erythrocytes in phosphatebuffered saline with or without peptide was incubated at 37°C for 10 min. The suspension was centrifuged at 16,000 ϫ g for 3 min. Hemolysis was measured as hemoglobin content (absorbance at 540 nm) of the supernatant. Base-line hemolysis was the hemolysis of RBCs incubated with phosphate-buffered saline. Hemoglobin released by 0.1% Triton X-100 was taken to be 100% lysis. Inhibition of lysis by peptides was measured by incubating RBCs with a 10 M concentration of peptides and adding 2 M 18L to the cell suspension. Hemolysis was expressed as a percentage of the Triton X-100 lysis.
Monocyte Chemotaxis Assay-The procedure detailed by Navab et al. (6) was followed. Briefly, human aortic endothelial cells and smooth muscle cells were isolated. Human aortic smooth muscle cells were grown in microtiter well plates until confluence, and the human aortic endothelial cells were added. Cocultures were treated with 20 g of 1-palmitoyl-2-arachidonyl-sn-glycero-3-phosphocholine (PAPC) and 1 g/ml of hydroperoxyeicosatetraenoic acid (HPODE) in the absence or presence of HDL (50 g of cholesterol) or peptides (66 M) for 8 h. The supernatants were collected and assayed for chemotactic activity.
Determination of Lipid Hydroperoxide Content-Lipid hydroperoxides were measured colorimetrically by the FOX2 assay (23). The ability of the peptides to remove lipid hydroperoxides from lipoproteins was assessed by incubating each of them with plasma from Watanabe rabbits. Plasma from these rabbits is known to possess lipid hydroperoxides (24,25). 1 ml of plasma was incubated with peptide at a final concentration of 44 M for 1 h at 37°C. The lipoproteins were then separated on two Superose 6 (Amersham Biosciences) columns in tandem using the Bio-Rad BioLogic Duoflow system. 0.5-ml fractions were collected, and absorbance was monitored at 280 nm. The peaks were analyzed for hydroperoxide content. Briefly, 900 l of the FOX reagent containing 300 M FeSO 4 7H 2 O, 120 M xylenol orange, 25 mM H 2 SO 4 , and 4.4 mM butylated hydroxytoluene in methanol was added to 100 l of the sample and incubated in the dark for 30 min, and absorbance was measured at 560 nm. The values were compared with a standard of cumene hydroperoxide to quantitate the hydroperoxide content.
Lipoproteins-Low density lipoprotein (d ϭ 1.019 g/ml) and high density lipoproteins (d ϭ 1.063-1.21 g/ml) were isolated using the procedure by Havel et al. (26) from normal volunteers under the protocol approved by Human Research Subject Protection committees of the University of Alabama at Birmingham and UCLA. In some cases, butylated hydroxytoluene (20 mM in ethanol) was added to freshly isolated plasma at a concentration of 20 M, and the lipoprotein was separated by fast protein liquid chromatography using the methods previously described (27). The LDL, HDL, and lipoprotein-deficient serum had endotoxin levels below 20 pg/ml, which is well below the threshold needed for induction of monocyte adhesion or chemotactic activity. The concentrations of lipoproteins reported in this study are based on protein content.
Peptide Synthesis-Peptides were synthesized by the solid phase method with a Protein Technologies PS-3 automatic peptide synthesizer using the procedures described previously (12,17). Peptides were purified using a preparative HPLC system (Beckman Gold), and the purity of the peptides was determined by mass spectral analysis and analytical HPLC. Circular Dichroism-CD spectra were recorded on an AVIV model 215 spectropolarimeter using a quartz cell with a 0.1-cm path length at 25°C. Peptide solutions in phosphate-buffered saline were used at concentrations ranging from 10 to 400 M. The effect of lipid binding on the secondary structure of the peptide was studied using peptide-lipid complexes (1:10 mol/mol) with 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) and 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC). Since the results were similar, we have reported results with POPC in order to be able to compare with the other studies. These complexes were prepared by adding the appropriate volume of peptide solution to POPC multilamellar vesicles (MLVs). The MLVs were prepared by dissolving a known amount of lipid in chloroform and slowly removing the solvent by evaporation under a thin stream of nitrogen. Residual solvent was removed by storing the lipid film under vacuum overnight. An appropriate volume of buffer was added to hydrate the thin lipid film, which was then vortexed. The lipid-peptide complexes were prepared by adding the required volume of peptide solutions to give the desired lipid to peptide molar ratio. CD measurements were also done in the presence of 50% trifluoroethanol. The program CD Spectra Deconvolution (CDNN), version 2.1, was used to calculate the secondary structure content (28).
Right Angle Light Scattering Measurements-Association of these peptides with POPC was determined by following the dissolution of POPC MLVs by right angle light scattering using an SLM 8000C photon counting spectrofluorometer as described (29). POPC MLVs were prepared by evaporating a solution of POPC (Avanti Polar) under nitrogen and hydrating the lipid film with phosphate-buffered saline (pH 7.4). The sample containing 105 M POPC and an equimolar amount of peptide was maintained at 25°C and continuously stirred. Turbidity clarification was monitored at 400 nm for 30 min. Complete dissolution of POPC vesicles was achieved by the addition of Triton X-100 to a final concentration of 1 mM.
Surface Pressure Measurements-Monolayer exclusion pressure measurements give the affinity of the peptides for a lipid-water interface. The procedure of Phillips and Krebs (30, 31) was followed using egg phosphatidylcholine (EPC) as per their procedure. An insoluble monolayer of EPC was spread at the air-water interface in a Teflon dish at room temperature to give an initial surface pressure ( i ) in the range of 5-45 micronewtons/m. A solution of peptides in phosphate-buffered saline containing 1.5 M guanidinium chloride was carefully injected in to the subphase to give a final concentration of 50 g/dl. The guanidinium chloride was diluted in the subphase to a final concentration of Յ1 mM to allow the peptides to renature. The subphase was stirred continuously, and the increase in EPC monolayer surface pressure (⌬) was recorded until a steady state value was obtained. The value of the initial surface pressure ( i ) at which the peptides no longer penetrate the EPC monolayer (i.e. the exclusion pressure ( e )) was calculated by extrapolating the i versus ⌬ linear regression fit to ⌬ ϭ 0 micronewtons/m.
Intrinsic Tryptophan Fluorescence-Fluorescence spectra of both peptide and peptide-lipid complexes were recorded at room temperature with excitation at 295 nm using a SPF-500 spectrofluorometer. These studies were done using DMPC as the lipid of choice in order to compare them with earlier published results of other peptides (17). The relative fluorescence intensity was measured in ratio mode that corrects for time dependent fluctuations in lamp intensity. Quenching of tryptophan fluorescence by potassium iodide and acrylamide was determined by adding aliquots of stock solutions of potassium iodide (4 M) and acrylamide (4 M) to a 2.7-ml solution of either peptide or peptide-lipid complexes (32). Stock solutions of potassium iodide contained 1 mM sodium thiosulfate (Na 2 S 2 O 3 ) to prevent the formation of I 3 Ϫ . Inner filter corrections were applied for acrylamide quenching (32). Emission spectra were taken after each addition, and emission intensity at max was determined. Peak emission intensity was used for all calculations. The quenching data were analyzed according to the Stern-Volmer equation, where F 0 and F represent the fluorescence in the absence and presence of the quencher, and [Q] is the concentration of the quencher. K SV is the Stern-Volmer constant.
Large Unilamellar Vesicles (LUVs)-Lipid films were made by dissolving appropriate amounts of lipid in a mixture of chloroform/methanol (2:1 (v/v)) and dried in a test tube under nitrogen to deposit the lipid as a thin film on the wall of the tube. Final traces of solvent were removed in a vacuum chamber attached to a liquid nitrogen trap for 2-3 h. Dried films were kept under argon gas at Ϫ30°C if not used immediately. Films were hydrated with buffer, vortexed extensively at room temperature, and then subjected to five cycles of freezing and thawing. The homogeneous lipid suspensions were then further processed by 10 passes through two stacked 0.1-m polycarbonate filters (Nucleopore Filtration Products, Pleasanton, CA) in a barrel extruder (Lipex Biomembranes, Vancouver, BC) at room temperature. LUVs were kept on ice and used within a few hours of preparation. Lipid phosphorus was determined by the method of Ames (33).
Fluorescence Spectroscopy Studies with DPH-PC-Peptide solutions were prepared in buffer containing 10 mM Hepes, 0.14 M NaCl, 0.1 mM EDTA, pH 7.4. The peptide concentrations were determined by absorbance at 280 nm against buffer as reference. Films containing POPC/ DPH-PC (400:1) were prepared from a solution in chloroform/methanol (2:1). The film was hydrated in buffer. 100-nm diameter LUVs were then made by extrusion as described previously. Fluorescence was measured in quartz cuvettes at 37°C, in an SLM Aminco Bowman Series II spectrofluorometer, with an excitation of 355 nm and slits set to 4-nm bandwidth in both excitation and emission. The fluorescence emission of 2 ml of 50 M LUVs in the same buffer was recorded. Then small aliquots of a peptide solution, also in the corresponding buffer, were added successively, and the fluorescence emission was recorded after each addition. The emission maximum was found at 431 nm.
Red Edge Excitation Shift (REES)-The emission spectra of tryptophan were measured at 25°C in an SLM Aminco Bowman Series II luminescence spectrofluorometer. The excitation wavelength was varied every 5 nm between 280 and 310 nm, and the emission spectra were recorded over 310 -350 nm three times and averaged. Measurements were done in quartz cuvettes containing 2 ml of buffer (10 mM NaH 2 PO 4 , 0.14 M NaF, 1 mM EDTA, pH 7.4). Aliquots of solutions of peptides in buffer were added to the cuvette to a final concentration of 10 M. Measurements were repeated with the addition of 100 M EPC LUVs. To reduce scattered light intensity, excitation polarization was set to 90°and emission to 0°, with 4-nm bandwidth. Spectra were corrected for instrumental factors, and controls were subtracted. Peak emission wavelengths were recorded.
Statistical Analysis-Statistical analysis was performed as previously described (18) using one-way analysis of variance, and significance was defined as p Ͻ 0.05. Post hoc Tukey tests were performed. Peptides were compared with base line and with each other.

Removal of Hydroperoxides from LDL and Inhibition of PAPC ϩ HPODE-induced Monocyte
Chemotaxis-We previously reported that the anti-inflammatory properties of these peptides correlate well with their ability to scavenge seeding molecules from the LDL surface, inhibiting LDL-induced monocyte chemotaxis (6,7). To determine their ability to scavenge hydroperoxides from LDL, the peptides were incubated with plasma from Watanabe rabbits that are known to have oxidized LDL (24,25). We used 4F as a positive control peptide to compare the present results with the previously published results (17). The lipoproteins were separated on Superose 6 columns, and the fractions were tested for hydroperoxide content. As shown in Fig. 2A, 3F-2 LDL-induced monocyte chemotaxis is due to the oxidation of LDL phospholipids containing arachidonic acid. This generates specific biologically active phospholipids, which in turn cause the artery wall cells to produce the potent monocyte chemoattractant MCP-1, which in turn induces monocyte chemotaxis (6,7). We compared the ability of these peptides to inhibit the monocyte chemotactic activity resulting from the incubation of human artery wall cells with the arachidonic acid containing phospholipid, PAPC, and the fatty acid hydroperoxide, HPODE. The effect of these peptides on PAPC-HPODE-induced monocyte chemotaxis (Fig. 2B) showed that, whereas both 3F-2 and 3F-1 effectively inhibited PAPC-HPODE-induced monocyte chemotaxis, 3F 3 and 3F 14 showed only small effects. This parallels their ability to scavenge lipid hydroperoxide molecules from LDL. There was no statistical difference between the activities of 3F-2 and 3F-1. However, the peptide 4F was the most effective and significantly different from 3F-1 and 3F-2.
Inhibition of 18L-induced Red Cell Lysis by 3F Peptide Analogs-Our earlier results showed that apoA-I and apoA-I mimetic peptides are not themselves lytic at low peptide/lipid ratios, and furthermore they stabilize membranes from lysis induced by the lytic peptide 18L (22). We tested the 3F peptides for lytic activity using RBCs and compared them to 2F and 4F (control peptides). The lysis caused by 0.1% Triton X-100 was FIG. 2. A, ability of peptides to scavenge lipid hydroperoxides from LDL. The effect of the 3F peptides and 4F on lipid hydroperoxide levels in LDL was determined. 1 ml of plasma from Watanabe rabbits (used as a source of oxidized LDL) was incubated with peptides at a final concentration of 44 M, and the lipoproteins were separated on Superose 6 columns using fast protein liquid chromatography. The hydroperoxide concentration in the LDL fraction was determined using cumene hydroperoxide as a standard. The data was analyzed using one-way analysis of variance, and the significance value is marked. Compared with the control samples, 4F, 3F-2, 3F-1, and 3F 3 reduced the concentration of lipid hydroperoxide significantly. The reduction seen by 4F, 3F-2, and 3F-1 appeared to be statistically more significant (**, p Ͻ 0.001) than that seen by 3F 3 ( †, p ϭ 0.001). However, 3F 14 is not significantly different from the control. B, inhibition of oxidized lipid-induced monocyte chemotaxis. The effect of the 3F peptides on PAPC ϩ HPODE-induced monocyte chemotaxis was determined at 66 M concentration of peptide. 20 g of PAPC was added together with 1 g/ml of HPODE to cocultures of human artery wall cells as described previously (7,8), and human HDL at 50 g/ml cholesterol or no addition was made to the cocultures (No Addition), or the peptides were also added at 66 M. After 8 h of incubation, supernatants were collected and assayed for monocyte chemotactic activity using standard neuroprobe chambers. The data are mean Ϯ S.D. of the number of migrated monocytes in nine fields for triplicate samples. *, p Ͻ 0.05 compared with PAPC ϩ HPODE alone.
considered to be 100%. The 3F peptides and 4F by themselves did not cause any lysis, even at a concentration of 10 M, whereas 2 M 18L was able to almost completely lyse the cells (98% of the level of Triton X-100). We also tested these peptides for their ability to inhibit 18L-mediated lysis (Fig. 3A). There is a range of potencies exhibited by these peptides in inhibiting 18L-mediated hemolysis. The results indicate that 3F 14 was not as effective in inhibiting 18L-mediated red cell lysis, whereas 2F and 3F 3 caused about 40% inhibition of 18L lysis. The peptides that were most effective in inhibiting 18L-induced cell lysis were 3F-1, 3F-2, and 4F. Both 3F-1 and 4F inhibited lysis by 65%, whereas 3F-2 was the most potent among the four peptides in inhibiting 18L-induced hemolysis (90% inhibition). A plot of reduction of lipid hydroperoxides by peptides ( Fig. 2A) versus 18L-induced percent RBC lysis (Fig. 3A) showed a remarkable correlation (Fig. 3B).
HPLC Retention Times and Surface Pressure- Table I lists the HPLC retention time on a C-18 column and exclusion pressure from an EPC monolayer. Among the four 3F analogs, peptide 3F-2 has the shortest retention time on a C-18 reversed phase HPLC column, and 3F 3 has the highest retention time. Exclusion pressure from an EPC monolayer was also the least for the peptide 3F-2, and highest for 3F 3 ( Table I). The results suggest that among the four 3F analogs, the presence of aromatic amino acids at the center of the nonpolar face decreases the lipid-associating ability (binding to reversed phase HPLC column containing C-18 acyl chain) and EPC monolayer exclusion pressure. The addition of another Phe to produce 4F marginally increases the retention time and exclusion pressure compared with 3F 3 and 3F 14 .
Tryptophan Fluorescence and Quenching Studies-In the peptide 3F-2, Trp is at the center of the ␣-helix nonpolar face, whereas it is at the polar-nonpolar interface in the other three peptide analogs. All of the peptides show a significant shift in fluorescence maximum ( max ) with DMPC, indicating that all of the four peptides associate similarly with lipid (Table II). The shift in max from solution state to lipid-associated state for 3F 3 and 3F 14 are 14 and 13 nm, respectively, whereas the other two peptides shifted less (11 nm). Peptide 3F-2 shows the largest increase in fluorescence intensity on binding to DMPC. This appears to be due to the position of Trp at the center of the nonpolar face. From aqueous phase to lipid-associated form, this Trp residue undergoes the greatest change in environment. Quenching studies by both iodide and acrylamide (Table  II) indicate that the Trp residues become less exposed to the quenchers (both KI and acrylamide) in the presence of lipid. The K SV values suggest that the Trp environment is similar in all the peptides in aqueous and lipid-bound states.
Right Angled Light Scattering Studies-Turbidity clarification (Fig. 4) studies indicate that all of the peptides clear multilamellar vesicles of POPC. The peptide 3F-2 appears to clear turbidity due to POPC multilamellar vesicles as rapidly as the 3F 3 and 3F 14 peptides. However, the peptide 3F-1 appears to clarify the least, suggesting that this peptide forms larger complexes than the other peptides. Calculation of t 50 , the time taken for reducing the light scattering intensity by 50%, from Fig. 4 indicates that 3F-1 and 3F-2 have the fastest rates of clarification (Table I), in agreement with these two peptides having increased surface pressure and increased HPLC retention times.
Circular Dichroism Studies-Ellipticity at 222 nm was dependent on the concentration of the peptide for all of the 3F analogs. Fig. 5 shows that there is an increase in ellipticity due to self-association, up to 300 M for all four peptides, beyond which it appears to plateau. The concentration for attaining 50% of maximal helicity varies for these peptides and indicates the differences in self-association of peptides. However, at 400 M, all four peptides seem to have similar helicities (40 -60%). The arrangement of hydrophobic residues on the nonpolar face FIG. 3. A, effect of the peptides on 18L-induced hemolysis. The effect of the peptides on 18L-induced RBC lysis was studied by coincubating RBCs with 18L and the control peptide, 4F, and the four 3F analogs. Hemolysis was measured as hemoglobin content in the lysate (absorbance of the lysate at 540 nm). Hemolysis by 0.1% Triton X-100 was taken as 100% lysis. The data are expressed as a percentage of Triton X-100 lysis. The data were analyzed by one-way analysis of variance. All of the peptides protect the RBCs from 18L-induced lysis. However, the protection by 3F 14 is not statistically significant. All of the others were significantly protective. **, p Ͻ 0.001. B, correlation between RBC lysis and lipid hydroperoxide scavenging by peptides. The percentage of lysis obtained with the different peptides (Fig. 3A) and the lipid hydroperoxide content in the LDL fraction of Watanabe plasma after incubation with the peptides (Fig. 2A) are plotted for each peptide. 3F-1 (Ⅺ), 3F-2 (E), 3F 3 (ƒ), 3F 14 (‚), 4F (छ). The regression coefficient was not calculated, since data from separate experiments are combined. appears to determine helicity in solution. All of the peptides possess similar helicity values (50 Ϯ 10%) at two different concentrations (100 and 400 M) in the presence of POPC or 50% trifluoroethanol (results not shown). These results indicate that the potential to form an amphipathic helical structure for all of these sequences is similar, and the secondary structure in the presence of lipid is also similar. Fluorescence Studies with DPH-PC-DPH-PC has been previously used to study the penetration of water into phospholipids bilayers (34). This lipid, when incorporated into POPC vesicles, produces increased fluorescence due to the hydropho-bic environment. However, if the DPH-PC in the PC bilayer is exposed to water, the fluorescence is quenched. This lipid can therefore be used to determine the extent of water penetration into the membrane. We determined the extent to which the addition of the peptides increased the quenching of the DPH-PC probe embedded in LUVs of POPC. We compared the properties of the four 3F peptides with that of 4F. Peptide 4F caused the greatest quenching of the fluorescence of DPH-PC (Fig. 6). At low peptide concentrations, the peptides 3F-2 and 3F-1 were the next most potent. The peptides 3F 3 and 3F 14 were not able to efficiently quench the DPH fluorescence. The presence of the aromatic groups in clusters could change the packing of the lipid acyl chains to allow for a greater penetration of water, which would cause quenching of the probe. Although one would not expect many water molecules to penetrate into the membrane, water molecules may gain access transiently, or the access into the membrane could occur as a result of water binding to the peptides, as has been observed in other systems (35). Alternatively, the peptides themselves may cause the quenching of the DPH-PC.
The results described in the legend to Fig. 6 were obtained by adding peptide to LUVs, maintaining a lipid/peptide ratio of 5 or greater. We have also measured the relative fluorescence intensity of DPH-PC in lipid particles that were solubilized by peptide (results not shown). This was done by incubating MLVs of POPC/DPH-PC (400:1) with a solution of one of the peptides at a lipid/peptide molar ratio of either 1:1 or 2:1. These mixtures were maintained at 25°C for up to 10 h. The resulting fluorescence of the solubilized lipid was compared with that of an identical mixture without peptide.
Unlike the case described above, of peptide added to LUVs, these solubilized lipid micelles exhibited fluorescence intensities that were similar to but slightly greater than that of the lipid alone. There was little difference in the fluorescence emission from the mixtures containing the different peptides. These results suggest that in the case of the LUVs, the lipid was not converted into a micellar form in the time and peptide concentrations used. We interpret the small effect of the peptides on the DPH-PC fluorescence in the solubilized micelles as being a consequence of the peptide interacting with only a small fraction of the total lipid that is present at the edge of the discoidal micelles. The biological effects presented in the present work, of inhibition of hemolysis (Fig. 2) and of monocyte chemotaxis (Fig. 3), do not involve the use of complexes in the form of discoidal micelles.
We have also analyzed the quenching of DPH-PC in LUVs at low peptide concentrations, where solubilization of the lipid would be minimal. We have calculated the dependence of quenching on peptide concentration for each of the peptides in the range between 0 and 3.5 M peptide (Fig. 6). The slopes of these plots are given in Table III. The peptide 4F is clearly the most potent in promoting quenching of the DPH-PC fluorescence, followed by 3F-2. 3F-1 has a somewhat higher slope than 3F 3 or 3F 14 , but the difference is not statistically significant because of the markedly nonlinear dependence of quenching on  the concentration of 3F-1, resulting in an increase in the error of the estimation of the slope at this range of very low peptide concentrations. REES-REES is the shift in the wavelength of maximum fluorescence emission toward higher wavelengths, caused by a change in the excitation wavelength toward the red edge of the absorption band. This phenomenon occurs with fluorophores in a motionally restricted environment in which the solvent dipole reorientation is slower than the lifetime of the excited state. REES has been used as a measure of the rigidity of the environment of Trp in proteins/peptides (36). There can also be other factors such as proton exchange in the excited state or other electrostatic effects that influence REES (37). However, for a series of peptides with similar amino acid compositions, conformation, and Trp emission wavelength, as with most proteins (36), the magnitude of REES is dependent on local motion. Despite the fact that the Trp environment is similar in all of the peptides studied here, there is a clear difference between the more biologically active analogs 3F-2, 3F-1, or 4F that do not exhibit REES either in lipid or buffer (Fig. 7, A and C) and the less active analogs 3F 3 and 3F 14 that exhibit a significant effect. In the presence of LUVs of egg PC, the change in emission max at ex is about 20 nm for 3F 14 and 14 nm for 3F 3 . These results suggest that the Trp is motionally restricted in 3F 3 and 3F 14 , whereas it is not in 4F, 3F-1, and 3F-2.

DISCUSSION
The series of peptides studied here have many structural similarities. All of the 3F peptides have identical amino acid compositions. In the presence of lipid, all of these peptides have similar secondary structures, and they all have similar hydrophobicities. They insert in a similar fashion into monolayers of EPC as shown with surface pressure measurements and into bilayers as evidenced by the fluorescence emission from Trp.
Despite all of these similarities, there is a significant difference in the biological potency of these peptides with respect to inhibition of 18L-induced hemolysis (Fig. 3A) and anti-inflammatory action (Fig. 2, A and B). These facts suggest that some relatively subtle differences among these peptides in their interaction with membranes account for the larger differences in the observed anti-inflammatory properties.
It is clear that the arrangement of aliphatic and aromatic amino acids on the hydrophobic face of the helix has significant effects on biological properties of Class A amphipathic helical peptides. In the present study, we introduce two new 3F analogs, 3F-1 and 3F-2, with the aromatic groups being either at the hydrophobic/hydrophilic interface of the amphipathic helix in 3F-1 or at the center of the hydrophobic face in 3F-2. This arrangement of aromatic residues changes the angle of the nonpolar face. In both 3F-1 and 3F-2, the angle is 180°, whereas it is 160°in the other peptides (Fig. 1A). Thus, the surface area of the nonpolar face is much greater in these two new analogs. CD results in 50% trifluoroethanol and in the presence of POPC indicate that all of the sequences have similar potential to form amphipathic helical structures (results not shown). Concentration-dependent CD studies in aqueous solution indicate that all of the peptides self-associate, although at different concentrations (Fig. 5).
ApoA-I has been shown to inhibit LDL-induced monocyte The experiment was repeated with independently prepared liposomes and peptide solutions and exhibited the same relative order among the different peptides but with somewhat greater error between experiments than the precision within an experiment. An analysis is given in Table III.  chemotaxis (6,7). The seeding molecules in LDL, hydroperoxyeicosatetraenoic acid and HPODE, induce the formation of specific oxidized phospholipids, which cause monocyte chemotaxis. Removal of these lipid hydroperoxides by HDL and apoA-I inhibits this process (6,7). In these studies, we have used HPODE-PAPC, which directly generates the oxidized phospholipids that induce monocyte chemotaxis. In this report, we show that HPODE-PAPC-induced monocyte chemotaxis is inhibited by both 3F-2 and 3F-1. These two peptides and 4F were better inhibitors than HDL, whereas 3F 3 was relatively ineffective and 3F 14 had no significant effect (Fig. 2B). This suggests that the arrangement of aromatic amino acids on the nonpolar face plays an important role in the complex process of inhibiting oxidized phospholipid-induced monocyte chemotaxis. It is interesting to note that although 3F 14 clarifies lipid similarly to other peptides and possesses a relatively higher HPLC retention time and a higher monolayer exclusion pressure than 3F-1 and 3F-2, it is not as effective in inhibiting monocyte chemotaxis. The peptides that were effective in inhibiting oxidized phospholipid-induced monocyte chemotaxis also effectively removed lipid hydroperoxides from LDL ( Fig.  2A). The peptides 3F 3 and 3F 14 were relatively ineffective in removing the LDL-lipid hydroperoxides from WHHL rabbit plasma and in inhibiting oxidized phospholipid-induced monocyte chemotaxis. Preliminary studies suggest that oral administration of D-4F to LDL receptor null and apoE null mice causes the rapid formation of small HDL-like particles containing peptide, cholesterol, apoA-I, and paraoxonase (15). These results coupled with the previous observations that class A amphipathic helical peptide associates with HDL (16) suggest that the peptides exhibit anti-inflammatory properties in lipidassociated form.
We have earlier used inhibition of 18L-induced red cell hemolysis (22) to screen the properties of class A peptides. The inhibition of superoxide production and neutrophil degranulation by class A peptides reported by us earlier (38) and 18Linduced lysis presented here suggest that the mechanism of inhibition of hemolysis and inhibition of HPODE-induced monocyte chemotaxis may all be related to stabilization of membrane by class A peptide analogs. We had shown in a dose-response study that 50 g of 18A was able to inhibit 18L-induced hemolysis by 50% (22), a result similar to the present observations with 2F. In the present group of peptides, we have compared the relative potency of class A peptides to inhibit 18L-induced hemolysis. Analogs 4F, 3F-2, and 3F-1 were significantly effective in inhibiting 18L-induced hemolysis, whereas 2F, 3F 3 , and 3F 14 were not as effective (Fig. 3A). It is interesting to note that there is a close relationship between the ability of the model peptides to inhibit 18L-induced RBC lysis and their ability to lower LDL lipid hydroperoxides (Fig.  3B). Molecular modeling studies (Fig. 1B) show that whereas the three less effective peptides show a similar wedge-shaped cross-section, the active peptides possess hydrophobic residues that would give a cylindrical shape to the peptides.
Tryptophan and tyrosine residues, when in close proximity to a lipid interface, are known to engage in electron transfer reactions that enable them to function as free radical scavengers capable of quenching hydroperoxide-initiated lipid peroxidation (39 -41). Since these peptides possess Tyr and Trp residues, they may play a role as scavengers of lipid hydroperoxides. The presence of multiple phenylalanine residues in these peptides may further enhance the stability of the tyrosyl radical (39). In the active 3F-2 analog, Tyr and Trp are one turn apart on the nonpolar face. However, Trp and Tyr are not in close proximity either in the helical wheel representation or in the linear sequence of 4F, yet this peptide is effective in scavenging lipid hydroperoxide from LDL. In addition, we have seen that an amphipathic helical peptide from apoJ sequence, which does not possess either Tyr or Trp, is able to inhibit lipid hydroperoxide-induced monocyte chemotaxis. 2 This suggests that in the peptides studied here, other factors such as the environment of the nonpolar face may play a major role, although the ability of the Trp and Tyr residues to act as free radical scavengers cannot be ignored presently.
There are two physical properties that vary in the same way as the biological activity. These are the REES (Fig. 7) and the quenching of DPH-PC (Fig. 6). The REES suggests that the more active peptides 3F-1, 3F-2, and 4F are in a less conformationally restricted environment, and the increased quenching of DPH-PC indicates that these active peptides, particularly 4F (Fig. 6), allow a greater penetration of water molecules into the hydrophobic milieu of the membrane. We are currently working on further defining the factors among these peptides that account for the differences in both these physical properties as well as their different biological effects. However, one feature that strikes us is the difference in "cross-sectional shape" of these peptides when modeled as ␣-helices with the Lys residues in a snorkel position (Fig. 1B). The peptide helices would lie near the lipoprotein (or peptide-lipid) interface with the hydrophobic face of the helix facing toward the lipid (bottom of the molecular models shown in Fig. 1, A and B). The molecular models show that the less active peptides (2F, 3F 3 , and 3F 14 ) have a relatively small cross-sectional area of the helix nonpolar face so that they are wedge-shaped. In contrast, peptides 3F-1, 3F-2, and 4F have a more cylindrical shape, because the nonpolar face of the helix is larger (Fig. 1B). We suggest that these peptides affect the lateral pressure profile and molecular packing of the lipoprotein particle surface in different ways.
The concept of lateral pressure profile was developed by Cantor (42) and used initially to explain the action of general anesthetics but has been shown to have applicability to may aspects of membrane structure and function (43). In this description, the lateral interactions in the phospholipid-water interface vary with the depth of penetration of a molecule such as a peptide into the phospholipid milieu. Presently, the relative depths of penetration of wedge-and cylinder-shaped peptides are not known. However, relative to the wedge-shaped peptides, the cylinder-shaped peptides seem to perturb the lipid acyl chain packing more because of their larger helix nonpolar face. This concept of looser acyl chain packing with the active peptides is supported by the results of red edge excitation shift experiments (Fig. 7) and fluorescence studies with DPH-PC, which show the quenching of DPH-PC fluorescence by the presence of water molecules in the hydrophobic milieu (Fig. 6). In contrast, the effect of wedge-shaped peptide molecules appears to be mainly on the interfacial region, since they have a small nonpolar face (Fig. 1B), thus causing minimal effects on the packing of the lipid acyl chains.
In summary, our studies show that factors other than overall hydrophobicity are important for the anti-inflammatory properties of class A peptides. Furthermore, for the first time, these studies differentiate the role of aromatic hydrophobic residues and aliphatic residues in altering the biological activity of a class A amphipathic helical peptide. The effects of interaction of class A amphipathic helical peptides on lipid packing depend upon the nature of hydrophobic residues on the nonpolar face. The presence of water in the milieu of the hydrophobic region of the peptide-lipid complex could determine the extent of anti-atherogenicity of class A amphipathic helical peptides. The presence of water molecules in turn could allow for the transfer of oxidized lipids from the LDL surface to peptide-containing particles, thereby rendering LDL less effective in inducing monocyte chemotaxis, an important step in the initiation of atherogenesis.