Proc. Nati. Acad. Sci. USA Vol. 76, No. 4, pp. 1775-1779, April 1979
Biophysics
Conformation of an oligopeptide in phospholipid vesicles (membrane proteins/NMR/circular dichroism/hydrophilic and hydrophobic effects/protein folding)
B. A. WALLACE AND E. R. BLOUT* Department of Biological Chemistry, Harvard Medical School, Boston, Massachusetts 02115
Contributed by Elkan R. Blout, December 28, 1978 To demonstrate a method by which the conABSTRACT formation of membrane proteins may be determined spectroscopically in model membranes, we determined the structure of a hydrophobic oligopeptide, t-butyloxycarbonylprolylleucylvalylmethyl ester, in phospholipid vesicles by nuclear magnetic resonance, circular dichroism, and infrared spectroscopy. 13C nuclear magnetic resonance and circular dichroism techniques demonstrated that the conformation of this peptide in linear hydrocarbon solutions was essentially identical to its conformation in lipid vesicles. IH nuclear magnetic resonance and infrared spectroscopy of the peptide in hydrocarbon solution then provided additional high-resolution information concerning the structure of the peptide as found in the hydrophobic portion of the lipid bilayer. The conformation of this peptide in hydrophobic media differs from its structure in hydrophilic solvents, not only in bond angles and the proportion of cis/ trans isomers about the X-proline bond, but also in its intermolecular associations.
The organization and biological activity of peptides and proteins in membranes is dependent upon the conformation they assume in a hydrophobic milieu. Clearly, the hydrocarbon chains of phospholipids present a very different structural matrix from water for the solvation of proteins. It might reasonably be expected that amino acid residues would assume different conformations in hydrophobic media. Diffraction studies of ordered two-dimensional arrays of proteins in membrane systems, such as the purple membrane of Halobacterium halobium (1), have provided some low-resolution information on the structure of membrane proteins. However; techniques that do not rely on the availability of ordered arrays of membrane proteins would be more generally useful. NMR spectroscopy has the potential for determining some bond angles between different portions of the peptide chain and the presence of inter- or intramolecular hydrogen bonds. Supplementary information at the level of gross secondary structure may be obtained from circular dichroism (CD) measurements. This work demonstrates the utility of such techniques for determining the structure of an oligopeptide in phospholipid vesicles. There are a number of problems, however, inherent in NMR studies of peptide structures in vesicles. First, the total peptide concentration must be relatively low in these samples in order to maintain the integrity of the lipid bilayer. Even with highsensitivity instruments, peptide signals are difficult to observe against the more abundant lipid background unless the compound is selectively enriched with 13C-labeled amino acids. Such enrichment may be useful for examining specific regions of large molecules; for example, studies of the orientation of gramicidin channels in lipid vesicles with '3C-enriched labels (unpublished data), but this method is, in general, not feasible for entire protein molecules. Furthermore, lipid resonances may
obscure peptide resonances, especially in proton NMR, unless 100% perdeuterated lipid is used. The major difficulty in the use of vesicles, however, is the extreme peak broadening due to immobilization of the peptide within the lipid bilayer. Even at temperatures above the lipid phase transition, 13C, resonances, especially of peptides that span the bilayer, may be too broad to be observed. Certainly, no coupling constants could be determined by proton NMR, even for small peptides. These obstacles suggest the need for a model system in which unenriched small peptides or 13C-enriched larger peptides could be examined. Such a system is described in this communication. Although the hydrophobic matrix of the lipid bilayer is not well mimicked by typical organic solvents used in NMR experiments (methanol, C2H3CN, C2HC1I, Me2SO, dioxane, etc.) because they possess the wrong geometry or polarity, it was found in this study that long-chain linear hydrocarbons do simulate the environment of the phospholipid fatty acid chains in the interior of the bilayer. Tumbling in such solvents is essentially isotropic and rapid so that sharp NMR peaks are obtained, permitting determination of coupling constants and corresponding dihedral angles by 1H NMR. Furthermore, peptide resonances are not obscured if perdeuterated hydrocarbon solvents are used. Thus, even though peptides that span the lipid bilayer are considerably immobilized in vesicles, hydrocarbon solution studies may be used to determine the structures of larger peptides and membrane proteins if suitably 13C-enriched or selectively deuterated samples were available. 13C NMR and CD spectroscopic techniques have been used in this work to demonstrate that the conformation of an oligopeptide in phospholipid vesicles is virtually identical to that in hydrocarbon solvents (albeit more immobilized within the lipid bilayer). These data, combined with results obtained from 1H NMR and infrared (IR) studies of the peptide in hydrocarbon solution, permitted determination of the structure of the peptide as found within the hydrophobic portion of the bilayer. In addition, the conformation of the peptide in hydrophobic (hydrocarbon) solution was compared to its structure in hydrophilic (aqueous) solution. The structures differ considerably, both in bond angles and in the proportion of cis/trans isomers about an X-proline bond in addition to their intermolecular association. As was expected, the overall structures differ in the extent of exposure of polar groups to solvent; these different conformations in hydrogen-bonding and non-hydrogen-bonding media have further implications for studies of protein folding and packing. Thus, we describe the structure of a peptide in lipid vesicles and propose a general method for studying the conformation of membrane proteins and peptides at high resolution. Abbreviations: CD, circular dichroism; IR, infrared; Boc, t-butyloxycarbonyl; OMe, methyl ester; LeucNH, leucine amide proton in the peptide molecule that contains a cis X-Pro bond; LeutNH, leucine amide proton in the peptide molecule that contains a trans X-Pro
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
bond. * To whom reprint requests should be addressed.
1775
1776
Proc. Natl. Acad. Sci. USA 76 (1979)
Biophysics: Wallace and Blout
MATERIALS AND METHODS Materials. t-Butyloxycarbonylprolylleucylvalylmethyl ester (BocProLeuValOMe) was prepared by Daniel Laufer as described (2). This compound migrated as a single component on analysis by thin-layer chromatography and reverse-phase high-pressure liquid chromatography. Dimyristoyl lecithin was obtained from Calbiochem. Perdeuterated dodecane and hexadecane were generously supplied by D. M. Engelman. All solvents were reagent grade. Preparation and Characterization of Vesicles. Dimyristoyl lecithin (220 ,umol) and BocProLeuValOMe (30 ,tmol) were dissolved in CHC13. The solvent was removed by evaporation and the solutes were spread as a thin layer on the walls of the evaporation vessel. 2H20 (1.2 ml) was added to the sample and the closed flask was heated on a steam bath for 20 min; the sample was flushed with nitrogen and sonicated with a microtip probe (Braun-Sonic 1510) for 0.5 hr at 370C. The vesicle sample was centrifuged briefly in a clinical centrifuge to remove any titanium particles introduced by the sonicator tip and undissolved peptide or very large liposomes. The vesicles were dialyzed against 2H20. Vesicle specimens were prepared for electron microscopy as follows: 10 gil of vesicle suspension (1 mg of lipid per ml) was placed on a Formvar-coated copper grid for 2 min; the grid was washed five times with 10 id of deionized water and stained with a 1% aqueous uranyl acetate solution for 15 sec. Grids were examined in a Zeiss electron microscope at X100,000 magnification and were photographed on Kodak SO-163 electron image film. Lipid concentrations were determined by the Fiske-Subbarow phosphate assay (3). Lipid-to-peptide ratios were analyzed by reverse-phase high-pressure liquid chromatography on an Hyp C8-23 column at 770 lb./inch2 (1 lb./inch2 = 6.9 X 103 Pa) with 75% and 100% methanol as mobile phases. Thinlayer chromatography of peptides and lipids was run on silica gel G plates in 80% MeOH/20% CHC13 (vol/vol). The chromatographs were developed with tolidine stain for peptides and molybdenum blue stain (Applied Science Laboratories, Inc., State College, PA) for lipid phosphates. Preparation of Solutions. For CD and IR samples, BocProLeuValOMe was dissolved in hexadecane, hexane, or deionized water at a concentration of 0.4 mM. In order to increase solubility of the peptide, permitting concentrations suitable for NMR studies, we used the following solvents: hydrocarbon solvent, 90% hexadecane or dodecane/10% chloroform (vol/vol) and aqueous solvent, 70% 2H20/30% methanol (vol/vol). Solutions were freshly prepared at peptide concentrations ranging from 1 to 100 mg/ml. For studies of solvent effect on -chemical shift, the hydrophobic and hydrophilic solvents varied, respectively, from 90% hexadecane/10-100% chloroform and from 70% 2H20/30-100% methanol. Instrumentation. CD spectra were recorded on a Cary 60 spectropolarimeter with a model 6001 CD attachment and a variable position detector by using a temperature-controlled sample cell of 1-mm pathlength operating over a wavelength range of 300-190 nm. Measurements were routinely made at 250 C; temperature-dependence measurements were taken in the range of 10-300C. Blank runs of solvent, or vesicles without peptide, were subtracted from the measured spectra. IR measurements used a Perkin Elmer 521 spectrophotometer operating at an ambient temperature of nt200C over a wavelength range of 4000-800 cm-1. The peptide samples were run at 1 mM solutions in 0.2-mm-pathlength CsBr cells against matched reference cells containing solvent. Fourier transform-proton NMR spectra were recorded on a Bruker 270 spectrometer. Peak positions were measured with
respect to external tetramethylsilane. Low-temperature measurements (-20° to +200C) permitted observation of NH-Ca coupling constants. 13C NMR spectra were run on Bruker 270 and Varian CFT-20 spectrometers at ambient probe temperatures of t300C. Chemical shifts are reported with respect to external CS2. T, measurements were determined by the inversion recovery method [pulse sequence (180'-r-90'-5TI)" ] (4). Vesicle measurements required the higher field (67.89 MHz), more sensitive Bruker instrument; under these conditions most peptide resonances were reasonably well separated from lipid resonances. Peptide peak assignments were established by comparison with model compounds and confirmed (for 1H NMR) by homonuclear spin decoupling. RESULTS The peptide, BocProLeuValOMe, is composed of hydrophobic amino acids commonly found in membrane proteins; Boc and methyl ester blocking groups on the NH2- and COOH-terminal residues increase its solubility in hydrophobic solvents, permitting it to partition preferentially into the hydrocarbon region of lipid bilayers. This peptide also has limited solubility in water, which permits the comparative studies of conformation in hydrophobic and hydrophilic solvents. Sonicated dimyristoyl lecithin vesicles were prepared containing phospholipid/peptide ratios of 32:1 after dialysis. These vesicles, as determined by negative-stain electron microscopy, were 250 A in diameter. They were relatively uniform in size, and the presence of peptide did not appear to perturb the vesicle structure nor to promote fusion of vesicles. High-pressure liquid chromatography and thin-layer chromatography indicated that essentially no degradation or lysolecithin formation occurred during the NMR or CD experiments. The CD spectrum of the peptide in water exhibited a molar ellipticity of 0 over the wavelength range of 300-210 nm and a decreasing ellipticity below 210 nm (Fig. 1). This result suggests an essentially random peptide conformation. In contrast, the spectrum of the peptide in hydrocarbons (either hexadecane or hexane) exhibited a relatively narrow bandwidth minimum at 227 nm with a molar ellipticity value (MO) of -3.1 X 104 and an increasing ellipticity below this wavelength (Fig. 1). Ellipticity at 227 nm in hexane was independent of tem-
1 '1 0
0
:2'0"0
220
240
260
X, nm FIG. 1. CD spectra (molar ellipticity) of BocProLeuValOMe in various solvents at 250C: A, H20; 0, hexadecane or hexane; 0,
dimyristoyl lecithin vesicles. Peptide concentration was 0.4 mM in H20, hexane, and hexadecane; 0.1 mM in vesicles.
Biophysics:
Wallace and Blout
Proc. Natl. Acad. Sci. USA 76 (1979)
perature over the range of 100-30'C. CD spectra of peptides in dimyristoyl lecithin vesicles after dialysis also exhibited a Xmin of 227 nm, with MO equal to -2.9 X 104, and a less negative ellipticity below this wavelength. The spectrum in vesicles was similar to that in hydrocarbon solution (within limits of experimental error), except that the absorption band was narrower in vesicles than in solution, perhaps indicative of the greater immobilization of the peptide in this environment. Light scattering from the vesicle samples was negligible. Before dialysis, the vesicles possessed a spectrum consisting of two components: a Xmin at 227 nm similar to the hydrocarbon solution spectrum and an increasingly negative ellipticity below 210 nm, reminiscent of the aqueous solution spectrum. After dialysis, which removed peptide present in the aqueous medium surrounding the lipid bilayers, leaving peptide that has been solubilized in the hydrocarbon region of the bilayer, only the hydrocarbon component of the spectrum remained (Fig. 1). The amount of peptide trapped in the aqueous interior volume of the vesicles must be small, as this would contribute to an "aqueous component" of the spectrum. These results strongly suggest that the peptide remaining in the vesicles is located in the interior hydrophobic portion of the bilayer, which is an environment similar to hydrocarbon solutions. Furthermore, shift reagent studies with thullium confirmed that the peptide was inaccessible to probes localized at the membrane surface, consistent with a location in the bilayer interior (unpublished data). 13C NMR spectra were obtained for the peptide in hydrocarbon and aqueous solutions and in lipid vesicles, and chemical shifts of the peptide molecule were compared (Table 1). The shifts for peptide in vesicles and in hydrocarbon solutions were essentially identical for all peaks that could be detected, while the shifts in aqueous solution were very different. Most peptide resonances were reasonably well separated from the lipid or solvent resonances. The peptide resonances were significantly broader (t4 times the peak width) in the vesicles than in the hydrocarbons, but were still narrow enough to permit detection Table 1. 13C chemical shifts Resonance
Hydrocarbon
C-O Leu C=O Pro C=O Val C=O Boc t-C Boc Ca Pro Ca Val OCH3 Ca Leu C6 Pro C# Leu Cp Val
20.33 20.50 -
-
112.11
-
-
-
134.97 140.34 140.20 145.47 152.28
134.68 -
140.21 145.46 152.30 -
-
CY Leu
164.48 167.93
CPro
-
Cb Leu
170.62 173.60 174.80
-
C. Val
20.31 20.51
35.24 112.20
C13 Pro
CH3 Boc
Solvent DML vesicles
164.51 167.88 169.35 170.01 170.61 173.65 174.82
Aqueous 17.90 18.73 19.40 33.00 111.39 132.39 134.33 140.39 140.66 145.77 152.13 161.39 162.20 164.96 168.13 169.25 170.44 171.34 174.29 175.09
Values in italics indicate peptide resonances that have chemical shifts that differ from the peptide chemical shifts in vesicles by
Biophysics
Conformation of an oligopeptide in phospholipid vesicles (membrane proteins/NMR/circular dichroism/hydrophilic and hydrophobic effects/protein folding)
B. A. WALLACE AND E. R. BLOUT* Department of Biological Chemistry, Harvard Medical School, Boston, Massachusetts 02115
Contributed by Elkan R. Blout, December 28, 1978 To demonstrate a method by which the conABSTRACT formation of membrane proteins may be determined spectroscopically in model membranes, we determined the structure of a hydrophobic oligopeptide, t-butyloxycarbonylprolylleucylvalylmethyl ester, in phospholipid vesicles by nuclear magnetic resonance, circular dichroism, and infrared spectroscopy. 13C nuclear magnetic resonance and circular dichroism techniques demonstrated that the conformation of this peptide in linear hydrocarbon solutions was essentially identical to its conformation in lipid vesicles. IH nuclear magnetic resonance and infrared spectroscopy of the peptide in hydrocarbon solution then provided additional high-resolution information concerning the structure of the peptide as found in the hydrophobic portion of the lipid bilayer. The conformation of this peptide in hydrophobic media differs from its structure in hydrophilic solvents, not only in bond angles and the proportion of cis/ trans isomers about the X-proline bond, but also in its intermolecular associations.
The organization and biological activity of peptides and proteins in membranes is dependent upon the conformation they assume in a hydrophobic milieu. Clearly, the hydrocarbon chains of phospholipids present a very different structural matrix from water for the solvation of proteins. It might reasonably be expected that amino acid residues would assume different conformations in hydrophobic media. Diffraction studies of ordered two-dimensional arrays of proteins in membrane systems, such as the purple membrane of Halobacterium halobium (1), have provided some low-resolution information on the structure of membrane proteins. However; techniques that do not rely on the availability of ordered arrays of membrane proteins would be more generally useful. NMR spectroscopy has the potential for determining some bond angles between different portions of the peptide chain and the presence of inter- or intramolecular hydrogen bonds. Supplementary information at the level of gross secondary structure may be obtained from circular dichroism (CD) measurements. This work demonstrates the utility of such techniques for determining the structure of an oligopeptide in phospholipid vesicles. There are a number of problems, however, inherent in NMR studies of peptide structures in vesicles. First, the total peptide concentration must be relatively low in these samples in order to maintain the integrity of the lipid bilayer. Even with highsensitivity instruments, peptide signals are difficult to observe against the more abundant lipid background unless the compound is selectively enriched with 13C-labeled amino acids. Such enrichment may be useful for examining specific regions of large molecules; for example, studies of the orientation of gramicidin channels in lipid vesicles with '3C-enriched labels (unpublished data), but this method is, in general, not feasible for entire protein molecules. Furthermore, lipid resonances may
obscure peptide resonances, especially in proton NMR, unless 100% perdeuterated lipid is used. The major difficulty in the use of vesicles, however, is the extreme peak broadening due to immobilization of the peptide within the lipid bilayer. Even at temperatures above the lipid phase transition, 13C, resonances, especially of peptides that span the bilayer, may be too broad to be observed. Certainly, no coupling constants could be determined by proton NMR, even for small peptides. These obstacles suggest the need for a model system in which unenriched small peptides or 13C-enriched larger peptides could be examined. Such a system is described in this communication. Although the hydrophobic matrix of the lipid bilayer is not well mimicked by typical organic solvents used in NMR experiments (methanol, C2H3CN, C2HC1I, Me2SO, dioxane, etc.) because they possess the wrong geometry or polarity, it was found in this study that long-chain linear hydrocarbons do simulate the environment of the phospholipid fatty acid chains in the interior of the bilayer. Tumbling in such solvents is essentially isotropic and rapid so that sharp NMR peaks are obtained, permitting determination of coupling constants and corresponding dihedral angles by 1H NMR. Furthermore, peptide resonances are not obscured if perdeuterated hydrocarbon solvents are used. Thus, even though peptides that span the lipid bilayer are considerably immobilized in vesicles, hydrocarbon solution studies may be used to determine the structures of larger peptides and membrane proteins if suitably 13C-enriched or selectively deuterated samples were available. 13C NMR and CD spectroscopic techniques have been used in this work to demonstrate that the conformation of an oligopeptide in phospholipid vesicles is virtually identical to that in hydrocarbon solvents (albeit more immobilized within the lipid bilayer). These data, combined with results obtained from 1H NMR and infrared (IR) studies of the peptide in hydrocarbon solution, permitted determination of the structure of the peptide as found within the hydrophobic portion of the bilayer. In addition, the conformation of the peptide in hydrophobic (hydrocarbon) solution was compared to its structure in hydrophilic (aqueous) solution. The structures differ considerably, both in bond angles and in the proportion of cis/trans isomers about an X-proline bond in addition to their intermolecular association. As was expected, the overall structures differ in the extent of exposure of polar groups to solvent; these different conformations in hydrogen-bonding and non-hydrogen-bonding media have further implications for studies of protein folding and packing. Thus, we describe the structure of a peptide in lipid vesicles and propose a general method for studying the conformation of membrane proteins and peptides at high resolution. Abbreviations: CD, circular dichroism; IR, infrared; Boc, t-butyloxycarbonyl; OMe, methyl ester; LeucNH, leucine amide proton in the peptide molecule that contains a cis X-Pro bond; LeutNH, leucine amide proton in the peptide molecule that contains a trans X-Pro
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
bond. * To whom reprint requests should be addressed.
1775
1776
Proc. Natl. Acad. Sci. USA 76 (1979)
Biophysics: Wallace and Blout
MATERIALS AND METHODS Materials. t-Butyloxycarbonylprolylleucylvalylmethyl ester (BocProLeuValOMe) was prepared by Daniel Laufer as described (2). This compound migrated as a single component on analysis by thin-layer chromatography and reverse-phase high-pressure liquid chromatography. Dimyristoyl lecithin was obtained from Calbiochem. Perdeuterated dodecane and hexadecane were generously supplied by D. M. Engelman. All solvents were reagent grade. Preparation and Characterization of Vesicles. Dimyristoyl lecithin (220 ,umol) and BocProLeuValOMe (30 ,tmol) were dissolved in CHC13. The solvent was removed by evaporation and the solutes were spread as a thin layer on the walls of the evaporation vessel. 2H20 (1.2 ml) was added to the sample and the closed flask was heated on a steam bath for 20 min; the sample was flushed with nitrogen and sonicated with a microtip probe (Braun-Sonic 1510) for 0.5 hr at 370C. The vesicle sample was centrifuged briefly in a clinical centrifuge to remove any titanium particles introduced by the sonicator tip and undissolved peptide or very large liposomes. The vesicles were dialyzed against 2H20. Vesicle specimens were prepared for electron microscopy as follows: 10 gil of vesicle suspension (1 mg of lipid per ml) was placed on a Formvar-coated copper grid for 2 min; the grid was washed five times with 10 id of deionized water and stained with a 1% aqueous uranyl acetate solution for 15 sec. Grids were examined in a Zeiss electron microscope at X100,000 magnification and were photographed on Kodak SO-163 electron image film. Lipid concentrations were determined by the Fiske-Subbarow phosphate assay (3). Lipid-to-peptide ratios were analyzed by reverse-phase high-pressure liquid chromatography on an Hyp C8-23 column at 770 lb./inch2 (1 lb./inch2 = 6.9 X 103 Pa) with 75% and 100% methanol as mobile phases. Thinlayer chromatography of peptides and lipids was run on silica gel G plates in 80% MeOH/20% CHC13 (vol/vol). The chromatographs were developed with tolidine stain for peptides and molybdenum blue stain (Applied Science Laboratories, Inc., State College, PA) for lipid phosphates. Preparation of Solutions. For CD and IR samples, BocProLeuValOMe was dissolved in hexadecane, hexane, or deionized water at a concentration of 0.4 mM. In order to increase solubility of the peptide, permitting concentrations suitable for NMR studies, we used the following solvents: hydrocarbon solvent, 90% hexadecane or dodecane/10% chloroform (vol/vol) and aqueous solvent, 70% 2H20/30% methanol (vol/vol). Solutions were freshly prepared at peptide concentrations ranging from 1 to 100 mg/ml. For studies of solvent effect on -chemical shift, the hydrophobic and hydrophilic solvents varied, respectively, from 90% hexadecane/10-100% chloroform and from 70% 2H20/30-100% methanol. Instrumentation. CD spectra were recorded on a Cary 60 spectropolarimeter with a model 6001 CD attachment and a variable position detector by using a temperature-controlled sample cell of 1-mm pathlength operating over a wavelength range of 300-190 nm. Measurements were routinely made at 250 C; temperature-dependence measurements were taken in the range of 10-300C. Blank runs of solvent, or vesicles without peptide, were subtracted from the measured spectra. IR measurements used a Perkin Elmer 521 spectrophotometer operating at an ambient temperature of nt200C over a wavelength range of 4000-800 cm-1. The peptide samples were run at 1 mM solutions in 0.2-mm-pathlength CsBr cells against matched reference cells containing solvent. Fourier transform-proton NMR spectra were recorded on a Bruker 270 spectrometer. Peak positions were measured with
respect to external tetramethylsilane. Low-temperature measurements (-20° to +200C) permitted observation of NH-Ca coupling constants. 13C NMR spectra were run on Bruker 270 and Varian CFT-20 spectrometers at ambient probe temperatures of t300C. Chemical shifts are reported with respect to external CS2. T, measurements were determined by the inversion recovery method [pulse sequence (180'-r-90'-5TI)" ] (4). Vesicle measurements required the higher field (67.89 MHz), more sensitive Bruker instrument; under these conditions most peptide resonances were reasonably well separated from lipid resonances. Peptide peak assignments were established by comparison with model compounds and confirmed (for 1H NMR) by homonuclear spin decoupling. RESULTS The peptide, BocProLeuValOMe, is composed of hydrophobic amino acids commonly found in membrane proteins; Boc and methyl ester blocking groups on the NH2- and COOH-terminal residues increase its solubility in hydrophobic solvents, permitting it to partition preferentially into the hydrocarbon region of lipid bilayers. This peptide also has limited solubility in water, which permits the comparative studies of conformation in hydrophobic and hydrophilic solvents. Sonicated dimyristoyl lecithin vesicles were prepared containing phospholipid/peptide ratios of 32:1 after dialysis. These vesicles, as determined by negative-stain electron microscopy, were 250 A in diameter. They were relatively uniform in size, and the presence of peptide did not appear to perturb the vesicle structure nor to promote fusion of vesicles. High-pressure liquid chromatography and thin-layer chromatography indicated that essentially no degradation or lysolecithin formation occurred during the NMR or CD experiments. The CD spectrum of the peptide in water exhibited a molar ellipticity of 0 over the wavelength range of 300-210 nm and a decreasing ellipticity below 210 nm (Fig. 1). This result suggests an essentially random peptide conformation. In contrast, the spectrum of the peptide in hydrocarbons (either hexadecane or hexane) exhibited a relatively narrow bandwidth minimum at 227 nm with a molar ellipticity value (MO) of -3.1 X 104 and an increasing ellipticity below this wavelength (Fig. 1). Ellipticity at 227 nm in hexane was independent of tem-
1 '1 0
0
:2'0"0
220
240
260
X, nm FIG. 1. CD spectra (molar ellipticity) of BocProLeuValOMe in various solvents at 250C: A, H20; 0, hexadecane or hexane; 0,
dimyristoyl lecithin vesicles. Peptide concentration was 0.4 mM in H20, hexane, and hexadecane; 0.1 mM in vesicles.
Biophysics:
Wallace and Blout
Proc. Natl. Acad. Sci. USA 76 (1979)
perature over the range of 100-30'C. CD spectra of peptides in dimyristoyl lecithin vesicles after dialysis also exhibited a Xmin of 227 nm, with MO equal to -2.9 X 104, and a less negative ellipticity below this wavelength. The spectrum in vesicles was similar to that in hydrocarbon solution (within limits of experimental error), except that the absorption band was narrower in vesicles than in solution, perhaps indicative of the greater immobilization of the peptide in this environment. Light scattering from the vesicle samples was negligible. Before dialysis, the vesicles possessed a spectrum consisting of two components: a Xmin at 227 nm similar to the hydrocarbon solution spectrum and an increasingly negative ellipticity below 210 nm, reminiscent of the aqueous solution spectrum. After dialysis, which removed peptide present in the aqueous medium surrounding the lipid bilayers, leaving peptide that has been solubilized in the hydrocarbon region of the bilayer, only the hydrocarbon component of the spectrum remained (Fig. 1). The amount of peptide trapped in the aqueous interior volume of the vesicles must be small, as this would contribute to an "aqueous component" of the spectrum. These results strongly suggest that the peptide remaining in the vesicles is located in the interior hydrophobic portion of the bilayer, which is an environment similar to hydrocarbon solutions. Furthermore, shift reagent studies with thullium confirmed that the peptide was inaccessible to probes localized at the membrane surface, consistent with a location in the bilayer interior (unpublished data). 13C NMR spectra were obtained for the peptide in hydrocarbon and aqueous solutions and in lipid vesicles, and chemical shifts of the peptide molecule were compared (Table 1). The shifts for peptide in vesicles and in hydrocarbon solutions were essentially identical for all peaks that could be detected, while the shifts in aqueous solution were very different. Most peptide resonances were reasonably well separated from the lipid or solvent resonances. The peptide resonances were significantly broader (t4 times the peak width) in the vesicles than in the hydrocarbons, but were still narrow enough to permit detection Table 1. 13C chemical shifts Resonance
Hydrocarbon
C-O Leu C=O Pro C=O Val C=O Boc t-C Boc Ca Pro Ca Val OCH3 Ca Leu C6 Pro C# Leu Cp Val
20.33 20.50 -
-
112.11
-
-
-
134.97 140.34 140.20 145.47 152.28
134.68 -
140.21 145.46 152.30 -
-
CY Leu
164.48 167.93
CPro
-
Cb Leu
170.62 173.60 174.80
-
C. Val
20.31 20.51
35.24 112.20
C13 Pro
CH3 Boc
Solvent DML vesicles
164.51 167.88 169.35 170.01 170.61 173.65 174.82
Aqueous 17.90 18.73 19.40 33.00 111.39 132.39 134.33 140.39 140.66 145.77 152.13 161.39 162.20 164.96 168.13 169.25 170.44 171.34 174.29 175.09
Values in italics indicate peptide resonances that have chemical shifts that differ from the peptide chemical shifts in vesicles by
