- Email: [email protected]
Folding of amphipathic α-helices on membranes: energetics of helix formation by melittin1
Article No. jmbi.1998.2346 available online at http://www.idealibrary.com on
J. Mol. Biol. (1999) 285, 1363±1369
COMMUNICATION
Folding of Amphipathic a -Helices on Membranes: Energetics of Helix Formation by Melittin Alexey S. Ladokhin and Stephen H. White* University of California at Irvine, Department of Physiology and Biophysics Irvine, CA 92697-4560, USA
Membranes have a potent ability to promote secondary structure formation in a wide range of membrane-active peptides, believed to be due to a reduction through hydrogen bonding of the energetic cost of partitioning peptide bonds. This process is of fundamental importance for understanding the mechanism of action of toxins and antimicrobial peptides and the stability of membrane proteins. A classic example of membrane-induced folding is the bee-venom peptide melittin that is largely unstructured when free in solution, but strongly adopts an amphipathic a-helical conformation when partitioned into membranes. We have determined the energetics of melittin helix formation through measurements of the partitioning free energies and the helicities of native melittin and of a diastereomeric analog with four D-amino acids (D4,L-melittin). Because D4,L-melittin has little secondary structure in either the free or bound forms, it serves as a model for the experimentally inaccessible unfolded bound form of native melittin. The partitioning of native melittin into large unilamellar phosphocholine vesicles is 5.0(0.7) kcal molÿ1 more favorable than the partitioning of D4,L-melittin (1 cal 4.186 J). Differences in the circular dichroism spectra of the two forms of melittin indicate that bound native melittin is more helical than bound D4,L-melittin by about 12 residues. These ®ndings disclose that the free energy reduction per residue accompanying the folding of melittin in membrane interfaces is about 0.4 kcal molÿ1, consistent with the hypothesis that hydrogen bonding reduces the high cost of partitioning peptide bonds. A value of 0.6 kcal molÿ1 per residue has been observed for b-sheet formation by a hexapeptide model system. These two values provide a useful rule of thumb for estimating the energetic consequences of membrane-induced secondary structure formation. # 1999 Academic Press
*Corresponding author
Keywords: peptide-membrane interactions; membrane protein folding; free energy of transfer; folding energetics; circular dichroism
Membrane interfaces have a potent ability to induce secondary structure in a wide range of membrane-active peptides such as hormones, toxPermanent address: A. S. Ladokhin, Institute of Molecular Biology and Genetics, National Academy of Sciences of Ukraine, Kiev 252143, Ukraine. Abbreviations used: POPG, palmitoyloleoylphosphatidylglycerol; POPC, palmitoyloleoylphosphatidylcholine; OBPC, 1-oleoyl-2(9,10-dibromostearoyl)-sn-3-phosphocholine; LUV, extruded unilamellar vesicles of 100 nm diameter; ANTS, 8-aminonapthalene-1,3,6 trisulfonic acid, trisulfonic acid, 427.33 Da. E-mail address of the corresponding author: [email protected] 0022-2836/99/041363±07 $30.00/0
ins and antimicrobial peptides (Kaiser & Kezdy, 1983, 1984; Schwyzer, 1992; Maloy & Kari, 1995; White & Wimley, 1998). A classic example is the 26-residue bee-venom peptide melittin from Apis mellifera (Figure 1(a)). It is largely unstructured when free in solution, but strongly adopts an amphipathic a-helical conformation when partitioned into membranes (Vogel, 1981; Dempsey, 1990; White et al., 1998). Knowledge of the energetics of this type of folding process is important for understanding and improving the activity of antimicrobial peptides and for understanding the folding and stability of membrane proteins. Wimley & White (1996) have shown that the partitioning of small peptides into membrane interfaces # 1999 Academic Press
1364
Figure 1. The amino acid sequence of native melittin and a general thermodynamic cycle for describing the partitioning and folding peptides in membranes. (a) The amino acid sequence of native melittin. The superscripts in the sequence indicate the positions of the D-amino acid residues in D4,L-melittin. (b) General thermodynamic cycle. For peptides with strong partitioning-folding coupling, the states B and C have very low occupancy so that the effective experimental equilibrium is A $ D. See the text for details.
is dominated by the very unfavorable free energy cost of partitioning the peptide bonds (1.2 kcal molÿ1 per peptide bond for phosphocholine bilayers; 1 cal 4.186 J). They proposed that hydrogen bonding of the peptide bonds reduces this high free energy cost and thereby promotes the formation of secondary structure. Support for this
Melittin Folding on Membranes
proposal is provided by a recent study of the formation of b-sheet aggregates by the hexapeptide AcWL5COOÿ on bilayers formed from POPC, where it was shown that hydrogen bonding reduces the apparent cost of partitioning the peptide bound by about 0.6 kcal molÿ1 per residue (Wimley et al., 1998). Such a modest value for the per-residue free energy change (Gresidue) can have dramatic consequences for secondary structure formation. If ten residues were to participate in hydrogen bond formation in an a-helical peptide, for example, the folded state would be favored by 6 kcal molÿ1, meaning that the folded form would be 26,000 times more abundant than the unfolded form. An important question thus arises: how similar is the value of Gresidue for helix formation to the value observed for b-sheet formation? We examine this question using membrane-induced secondary formation by melittin as a test case. We ®nd that Gresidue for melittin is similar to the value observed for b-sheet formation by AcWL5COOÿ. The results suggest that Gresidue 0.5 kcal molÿ1 may be useful for estimating the energetic consequences of secondary structure formation on membranes. Partitioning-folding coupling The dif®culty with the determination of the energetics of helix folding on membranes is that folding is generally tightly coupled to partitioning. This can be understood from the thermodynamic cycle shown in Figure 1(b). If partitioning-folding coupling is strong, the folded B state and unfolded
Figure 2. Summaries of the estimated and experimentally determined values for partitioning and folding of melittin into POPC interfaces. (a) Estimated values for the partitioning and folding free energies for native melittin. Using the nomenclature of Figure 1(b), three thermodynamic states are shown: unfolded in aqueous phase (state A); unfolded in membrane (state C); and helix in membrane (state D). Melittin has a very low helicity in the absence of membranes. Any melittin in the folded form in water (state B) is assumed to be part of the population of conformations comprising state A. Because of partitioning-folding coupling (see the text), the occupancy of state C is very low and is a virtual state for practical experimental purposes. The range of free energy differences for the partitioning of unfolded melittin was calculated using the whole-residue interfacial hydrophobicity (H) scale of Wimley & White (1996). The range of values results from different assumptions about the interfacial location of the C-terminal polar residues (21-26 in Figure 1(a)). The range of values for the partitioning free energy of melittin in the folded form is based upon various values reported in the literature (Vogel, 1981; Kuchinka & Seelig, 1989; Schwarz & Beschiaschvili, 1989; Beschiaschvili & Baeuerle, 1991; White et al., 1998). Native melittin bound to the membrane is assumed to have 18 residues in a helical conformation (Dempsey & Butler, 1992) and is used to estimate the free energy difference per residue (Gresidue). (b) Experimentally determined values for the partitioning and folding free energies for melittin. Here, in the context of Figure 2(a), the virtual unfolded bound state of melittin is emulated by D4,L-melittin (see the text for details). The helicity difference between the membrane-bound folded and unfolded states was determined using the data of Table 1. The free energies of transfer G of D4,L-melittin and melittin were determined from molefraction partition coef®cients Kx usingG ÿRT ln Kx. The Kx were measured by an equilibrium dialysis method described in detail elsewhere (Wimley & White, 1993, 1996; Ladokhin et al., 1997; White et al., 1998). Brie¯y, a suspension of lipid vesicles was placed in one half of an equilibrium dialysis cell (Spectrum Medical Industries, Houston, TX) that was separated from the other half by a semi-permeable polycarbonate membrane. A solution of peptide was placed in the other half-cell. With constant rotation of the cells in a thermostat, melittin equilibrated across the polycarbonate membrane in less than 24 hours. We generally allowed the system to equilibrate for 48 or more hours at 25 C. Kx ([P]bil/[L])/([P]water/[W]), where [P]bil and [P]water are the bulk molar concentrations of peptide attributable to peptide in the bilayer and water phases, respectively. In the vesicle-containing half-cell, [P]total [P]bil [P]water. [L] and [W] are the molar concentrations of lipid and water (55.3 M). Due to the extremely low partitioning of D4,L-melittin, a correction of [P]water for the lipid side of the dialysis cell was necessary because the inner volume of the vesicles is not accessible to peptides that do not bind well. A correction factor to account for this phenomenon was estimated from distribution of ANTS (Molecular Probes, Eugene, OR), a dye that does not penetrate through the membrane. The correction factor was found to be 0.91 at 80 mM POPC and 0.87 at 130 mM POPC.
1365
Melittin Folding on Membranes
C state will not be signi®cantly populated in terms of experimental detectability, so that the practical thermodynamic equilibrium will between two states: the unfolded A state in solution and the folded D state on the membrane. This is certainly the case for melittin (Figure 2(a)), as demonstrated unequivocally by CD spectroscopy. The partitioning of melittin gives rise to a characteristic a-helix ellipticity at 222 nm that shows a progressive increase with lipid concentration. At the same time, there is an isodichroic point at 204 nm (Vogel, 1981; White et al., 1998), which is the expected consequence of a two-state process. The free energy of partitioning of unfolded melittin
into the bilayer interface of POPC vesicles as the folded helical form, i.e. the A $ D equilibrium (Figures 1(b) and 2(a)), has been estimated by various methods to be ÿ7 to ÿ8 kcal molÿ1 using mole-fraction partition coef®cients (Vogel, 1981; Kuchinka & Seelig, 1989; Schwarz & Beschiaschvili, 1989; Beschiaschvili & Baeuerle, 1991; White et al., 1998). The critically important membrane-bound unfolded C state cannot be observed, but must be present. It can be thought of as a virtual state (Figure 2(a)). Wimley & White (1996) have determined a whole-residue interfacial hydrophobicity scale for POPC interfaces that can be used to
Figure 2 (legend opposite)
1366 estimate the partitioning free energy of unfolded peptides. When applied to the partitioning of melittin, the free energy of transfer from the real A state to the virtual C state can be estimated to lie between ÿ2 and 4 kcal molÿ1, depending upon whether the highly polar C-terminal residues (2126, Figure 1(a)) are assumed to enter the interface. From this calculated estimate of the A $ C equilibrium and the approximate measured value of the A $ D equilibrium, the free energy difference between the unfolded and folded forms of melittin in the interface (the C $ D equilibrium) can be estimated to be between ÿ5 and ÿ12 kcal molÿ1 (Figure 2(a)). The per-residue free energy reduction driving helix formation is estimated from this broad range to be 0.3 to 0.7 kcal molÿ1, assuming a helicity of 18 residues for the folded interfacial form of melittin (Dempsey & Butler, 1992). The value for Gresidue ÿ0.6 kcal molÿ1 obtained for b-sheet formation in POPC vesicles (Wimley et al., 1998) falls comfortably within this range. This suggested that it might be an approximate universal value for estimating the energetics of secondary structure formation by peptides partitioned into bilayer interfaces. Experimental veri®cation of this prospect became possible with the availability of a diastereomeric variant of melittin (Oren & Shai, 1997). Partitioning free energy of unfolded melittin The introduction of D-amino acids into melittin should signi®cantly diminish the formation of secondary structure, while preserving the wholeresidue hydrophobicity. If such a peptide partitioned measurably into bilayers, then the free energy difference between the A state and the virtual C state (Figure 1(b)) could be determined and the energetics of the A $ C $ D thermodynamic cycle consequently estimated. For this purpose, we utilized the melittin diastereomeric analog D4,L-melittin (Oren & Shai, 1997) that has D-amino acid residues in positions 5, 8, 17 and 21 (Figure 1(a)). Their measurement demonstrated that D4,L-melittin does indeed have a greatly diminished helical content compared to melittin. We have con®rmed this observation (Table 1). The free energies of transfer of the two forms of melittin from the aqueous phase to the membrane interface of POPC LUV were measured using equilibrium dialysis (Wimley & White, 1993, 1996; Ladokhin et al., 1997; White et al., 1998) under conditions that assure that native melittin is monomeric in both the aqueous and membrane phases (Dempsey, 1990). The mole-fraction partition coef®cients Kx for D4,L-melittin and melittin were determined to be 80(30) and 4(2) 105, respectively. These values yield free energies of for D4,L-melittin and ÿ2.6(0.3) kcal molÿ1 for native melittin ÿ7.6(0.4) kcal molÿ1 (Figure 2(b)). The free energy difference of ÿ5.0(0.7) kcal molÿ1 between the ``emulated'' unfolded state of melittin and folded melittin is
Melittin Folding on Membranes Table 1. The per-residue ellipticity, y, at 222 nm and the mean helical content, fa, for native melittin and D4,L-melittin in several environments Environment Phosphate (10 mM) Methanol Bound to POPCe,f Bound to POPGe,h
Native melittina yb; fac
D4,L-Melittin
a
ÿ5.4; 0.12 ÿ23.6; 0.69 ÿ23.9; 0.70 ÿ24.0; 0.71
ÿ1.5; 0.00d ÿ8.5; 0.22 n.o.g ÿ10.5; 0.28
yb; fac
a Melittin (sequencing grade) obtained from Sigma (St. Louis, MO). D4,L-Melittin was a gift from Dr Y. Shai. b Units of 103 deg cmÿ2 dmolÿ1. CD measurements were performed using a Jasco-720 spectropolarimeter (Japan Spectroscopic Company, Tokyo). Normally, 10 to 50 scans were recorded between 185 and 260 nm using a 1 mm optical path. UV absorbance was measured with a Cary 3E spectrophotometer (Varian Analytical Instruments, Sugar Land, TX). For stock solutions in methanol, CD and absorbance were measured for the same sample in the same cuvette in order to minimize errors in determination of molar ellipticity. The errors in the determination of y are estimated to be 4 % for the methanol environment and 10 % for lipid environments. c The formula fa (y ÿ yRC)/(yH ÿ yRC) was used to estimate the fractional helical content fa of peptides in a helical conformation, where y is observed ellipticity and yRC and yH are the limiting values for a completely random coil and a completely helical conformation, respectively. Although this formula is simple and well accepted, there is a certain ambiguity in the result due to the uncertainty in prediction of what the actual values for yRC and yH should be (Green®eld & Fasman, 1969; Wu et al., 1981; Luo & Baldwin, 1997; Rohl & Baldwin, 1997; Shalongo & Stellwagen, 1997). Here, we use the following values, both at 222 nm: yRC ÿ1.5 103 deg cmÿ2 dmolÿ1 and yH ÿ33.4 103 deg cmÿ2 dmolÿ1. The yH number is calculated at 25 C for a peptide the size of melittin according to Luo & Baldwin (1997). The number for the random coil is an experimentally observed ellipticity for D4,L-melittin. Choosing the limiting pair in this way is consistent in two respects: (1) it allows one to avoid an unreasonably high helicity for membrane-bound D4,L-melittin; and (2) it brings the estimate for POPC-bound melittin into accord with the results of the amide hydrogen exchange measurements made by Dempsey & Butler (1992). d fa taken as 0 by de®nition (see above). e Lipids were obtained from Avanti Polar-Lipids (Alabaster, AL). Large unilamellar vesicles (LUV) of the lipids with an approximate diameter 0.1 mm were formed by extrusion under N2 pressure through Nucleopore polycarbonate membranes using the method of Mayer et al. (1986). The buffer solution (pH 7.0) used for partitioning experiments contained 10 mM Hepes, 50 mM KCl, 5 mM EDTA and 3 mM NaN3. To reduce the UV absorbance in CD measurements, a 10 mM potassium phosphate buffer (pH 7.0) was used. f Determined by extrapolation to complete binding (White et al., 1998). g Ellipticity not observable because of a very low partition coef®cient. h Measured at saturation (1.6 mM lipid).
consistent with the expected low population of the unfolded membrane-bound state of native melittin and with the hypothesis of that secondary structure formation is driven by a reduction in the free energy of partitioning of peptide bonds caused by hydrogen bonding (Wimley & White, 1996). Because it cannot readily form secondary structure in the interface, D4,L-melittin cannot lower its free energy of transfer into the membrane interface through the hydrogen bonding of secondary structure formation.
Melittin Folding on Membranes
Helicity of the membrane-bound forms of melittin Calculation of Gresidue for melittin helix formation requires that the average number of residues in a helical conformation be determined for the two forms of melittin partitioned into POPC membranes. The dif®culty that arises, however, is that the ellipticity of D4,L-melittin on POPC membranes cannot be measured directly because of its extremely low partition coef®cient. We therefore had to use a less direct approach. We found for native melittin that the ellipticity at 222 nm was virtually identical in methanol, when fully bound hydrophobically to POPC, and when fully bound electrostatically to the anionic lipid POPG. Assuming that a similar situation prevails for D4,Lmelittin, we used measurements of the ellipticity of the two forms of melittin in methanol and when bound to POPG to estimate their helicity difference. The results of measurements of the ellipticities (y) of both forms of melittin in the various phases (methanol, POPC and POPG) are summarized in Table 1 and Figure 3. The results for POPCbound melittin are consistent with numerous measurements on various membranes determined
Figure 3. CD spectra of melittin and its diastereomer analog, D4,L-melittin in methanol (a) and when bound to POPG LUV (b). The ellipticities and calculated helicities are summarized in Table 1. The total peptide concentrations were about 50 mM for the methanol experiments and 20 mM for the POPG experiments.
1367 by both CD and NMR spectroscopy (reviewed by Dempsey, 1990). Fractional helicities were calculated from fa (y ÿ yRC)/(yH ÿ yRC), where yRC and yH are the limiting values for completely random and completely helical conformations, respectively (see the footnotes to Table 1). The data of Table 1 indicate about 18 residues of helix for native melittin and about six residues of helix for D4,L-melittin when membrane-bound. Thus, native melittin has 12 1 more residues in an a-helical conformation than D4,L-melittin. We suggest that the only plausible region to form an a-helix in the diastereomer is T11GLPAL16, because the propensity of this region to form secondary structure is apparently stronger in membrane environments, consistent with the observation that Pro14 does not break the helix in membrane-bound native melittin (Smith et al., 1994). Energetics of melittin folding on membranes The results above, summarized in Figure 2(b), lead to a value of Gresidue of ÿ0.41(0.06) kcal molÿ1 as the per-residue free energy reduction that promotes helix formation of melittin upon partitioning into the POPC bilayer interface. However, we cannot rule out the possibility that D4,L-melittin has a conformation in which some hydrogenbonds are satis®ed internally in other than an a-helical conformation. If that is the case, then the magnitude of Gresidue may be greater than 0.41. Hirota et al. (1998) have recently reported that Gresidue ÿ0.12 kcal molÿ1 for helix formation by melittin in alcohols. Our result demonstrates in comparison the remarkable ability of the membrane interface to induce helix formation. The Gresidue that we observe for POPC membranes is well within the range of those estimated by means of the whole-residue hydrophobicity scale of Wimley & White (1996) and is satisfyingly similar to the value of ÿ0.61(0.04) kcal molÿ1 obtained for b-sheet formation on POPC membranes by AcWL5COOÿ (Wimley et al., 1998). These ®ndings provide a preliminary answer to the question posed at the beginning: the per-residue free energy reduction observed for a-helix formation on membranes is similar to the value for b-sheet formation. AcWL5COOÿ partitions into the POPC interface as a monomer and assembles into b-sheet aggregates of about 12 peptides (Wimley et al., 1998), whereas melittin partitions and folds as a single unit. The difference of 0.2 kcal molÿ1 between Gresidue for AcWL5COOÿ and melittin is not surprising, given that the amino acid sequences are strikingly different and that one must suppose that folding/assembly entropy, side-chain packing, relative exposure of side-chains to membrane and water, and the depth of membrane penetration of the secondary structure units are quite different. These structural differences make it dif®cult to attribute Gresidue solely to a reduction in the cost of partitioning a peptide bond into the interface. Differences in the hydrophobic-effect contributions to partitioning are
1368 likely to be important, for example. Nevertheless, because the cost of partitioning the peptide bond into the interface dominates whole-residue partitioning (Wimley & White, 1996), hydrogen bonding is surely dominant in secondary structure formation. Taking Gresidue ÿ0.4 to ÿ0.6 kcal molÿ1 should provide reasonable ®rst estimates of the free energy difference between the partitioning of folded and unfolded peptides. A comparison of the Gresidue values for helix formation (this work) and b-sheet formation (Wimley et al., 1998) suggests that appropriate ranges of Gresidue may be ÿ0.4 to ÿ0.5 kcal molÿ1 and ÿ0.5 to ÿ0.6 kcal molÿ1 for helix and sheet formation, respectively. The Wimley & White (1996) whole-residue hydrophobicity scale allows one to estimate the partitioning free energy of a small unfolded peptide. The partitioning free energy of the folded form can then be estimated using Gresidue. For example, the application of the whole-residue hydrophobicity scale to magainin 2 (Bevins & Zasloff, 1990) leads to an unfavorable free energy of transfer of 3.6 kcal molÿ1 for the unfolded peptide. Wieprecht et al. (1997) have estimated the helix content of bound magainin as 18 residues. If Gresidue is between ÿ0.4 and ÿ0.5 kcal molÿ1, the partitioning free energy reduction upon folding is between ÿ7.2 and ÿ9.0 kcal molÿ1. These numbers lead to an estimated partitioning free energy of ÿ3.6 to ÿ5.4 kcal molÿ1 for the folded form, which agrees satisfactorily with the experimental value of ÿ4.8 kcal molÿ1 reported by Wenk & Seelig (1998). These results are encouraging, but much remains unknown about the dependence of peptide partitioning on amino acid sequence and membrane lipid composition. The hydrophobicity-hydrophilicity pattern of the peptide sequence that de®nes the hydrophobic moment (Eisenberg et al., 1984) is undoubtedly important. Subtle differences in membrane lipid composition have already been observed to have signi®cant effects on the thermodynamics of b-sheet formation by AcWL5COOÿ. Bromination of the double-bond of the sn-2 chain of DOPC to produce OBPC leads to a decrease of Gresidue from 0.61 to 0.46 kcal molÿ1 (Wimley et al., 1998). Although many details of peptide folding on membranes remain to be discovered, the methods presented here and elsewhere provide general experimental approaches for uncovering these details in order to establish quantitative relationships between the effects of amino acid sequence and membrane composition on the partitioning and folding of peptides into membrane (White & Wimley, 1994, 1998; Wimley & White, 1996; White et al., 1998).
Acknowledgments We are grateful to Dr Yechiel Shai for his generous gift of the melittin diastereomer and to Drs William
Melittin Folding on Membranes Wimley and Kalina Hristova for stimulating discussions. This research was supported, in part, by grant GM-46823 from the National Institute of General Medical Sciences.
References Beschiaschvili, G. & Baeuerle, H.-D. (1991). Effective charge of melittin upon interaction with POPC vesicles. Biochim. Biophys. Acta, 1068, 195-200. Bevins, C. L. & Zasloff, M. (1990). Peptides from frog skin. Annu. Rev. Biochem. 59, 395-414. Dempsey, C. E. (1990). The actions of melittin on membranes. Biochim. Biophys. Acta, 1031, 143-161. Dempsey, C. E. & Butler, G. S. (1992). Helical structure and orientation of melittin in dispersed phospholipid membranes from amide exchange analysis in situ. Biochemistry, 31, 11973-11977. Eisenberg, D., Schwarz, E., Komaromy, M. & Wall, R. (1984). Analysis of membrane and surface protein sequences with the hydrophobic moment plot. J. Mol. Biol. 179, 125-142. Green®eld, N. & Fasman, G. D. (1969). Computed circular dichroism spectra for the evaluation of protein conformation. Biochemistry, 8, 4108-4116. Hirota, N., Mizuno, K. & Goto, Y. (1998). Group additive contributions to the alcohol-induced a-helix formation of melittin: implication for the mechanism of the alcohol effects on proteins. J. Mol. Biol. 275, 365-378. Kaiser, E. T. & Kezdy, F. J. (1983). Secondary structures of proteins and peptides in amphiphilic environments (a review). Proc. Natl Acad. Sci. USA, 80, 1137-1143. Kaiser, E. T. & Kezdy, F. J. (1984). Amphiphilic secondary structure: design of peptide hormones. Science, 223, 249-255. Kuchinka, E. & Seelig, J. (1989). Interaction of melittin with phosphatidylcholine membranes. Binding isotherm and lipid head-group conformation. Biochemistry, 28, 4216-4221. Ladokhin, A. S., Selsted, M. E. & White, S. H. (1997). Bilayer interactions of indolicidin, a small antimicrobial peptide rich in tryptophan, proline, and basic amino acids. Biophys. J. 72, 794-805. Luo, P. Z. & Baldwin, R. L. (1997). Mechanism of helix induction by tri¯uoroethanol: a framework for extrapolating the helix-forming properties of peptides from tri¯uoroethanol/water mixtures back to water. Biochemistry, 36, 8413-8421. Maloy, W. L. & Kari, U. P. (1995). Structure-activity studies on magainins and other host defense peptides. Biopolymers, 37, 105-122. Mayer, L. D., Hope, M. J. & Cullis, P. R. (1986). Vesicles of variable sizes produced by a rapid extrusion procedure. Biochim. Biophys. Acta, 858, 161-168. Oren, Z. & Shai, Y. (1997). Selective lysis of bacteria but not mammalian cells by diastereomers of melittin: structure-function study. Biochemistry, 36, 1826-1835. Rohl, C. A. & Baldwin, R. L. (1997). Comparison of NH exchange and circular dichroism as techniques for measuring the parameters of the helix-coil transition in peptides. Biochemistry, 36, 8435-8442. Schwarz, G. & Beschiaschvili, G. (1989). Thermodynamic and kinetic studies on the association of melittin with a phospholipid bilayer. Biochim. Biophys. Acta, 979, 82-90. Schwyzer, R. (1992). Conformations and orientations of amphiphilic peptides induced by arti®cial lipid
Melittin Folding on Membranes membranes: correlations with biological activity. Chemtracts-Biochem. Mol. Biol. 3, 347-379. Shalongo, W. & Stellwagen, E. (1997). Dichroic statistical model for prediction and analysis of peptide helicity. Proteins: Struct. Funct. Genet. 28, 467-480. Smith, R., Separovic, F., Milne, T. J., Whittaker, A., Bennett, F. M., Cornell, B. A. & Makriyannis, A. (1994). Structure and orientation of the pore-forming peptide, melittin, in lipid bilayers. J. Mol. Biol. 241, 456-466. Vogel, H. (1981). Incorporation of melittin into phosphatidylcholine bilayers: study of binding and conformational changes. FEBS Letters, 134, 37-42. Wenk, M. R. & Seelig, J. (1998). Magainin 2 amide interaction with lipid membranes: calorimetric detection of peptide binding and pore formation. Biochemistry, 37, 3909-3916. White, S. H. & Wimley, W. C. (1994). Peptides in lipid bilayers: structural and thermodynamic basis for partitioning and folding. Curr. Opin. Struct. Biol. 4, 79-86. White, S. H. & Wimley, W. C. (1998). Hydrophobic interactions of peptides with membrane interfaces. Biochim. Biophys. Acta, 1376, 339-352.
1369 White, S. H., Wimley, W. C., Ladokhin, A. S. & Hristova, K. (1998). Protein folding in membranes: determining the energetics of peptide-bilayer interactions. Methods Enzymol. 295, 62-87. Wieprecht, T., Dathe, M., Beyermann, M., Krause, E., Maloy, W. L., MacDonald, D. L. & Bienert, M. (1997). Peptide hydrophobicity controls the activity and selectivity of magainin 2 amide in interaction with membranes. Biochemistry, 36, 6124-6132. Wimley, W. C. & White, S. H. (1993). Quantitation of electrostatic and hydrophobic membrane interactions by equilibrium dialysis and reverse-phase HPLC. Anal. Biochem. 213, 213-217. Wimley, W. C. & White, S. H. (1996). Experimentally determined hydrophobicity scale for proteins at membrane interfaces. Nature Struct. Biol. 3, 842-848. Wimley, W. C., Hristova, K., Ladokhin, A. S., Silvestro, L., Axelsen, P. H. & White, S. H. (1998). Folding of b-sheet membrane proteins: a hydrophobic hexapeptide model. J. Mol. Biol. 277, 1091-1110. Wu, C.-S., Ikeda, K. & Yang, J. T. (1981). Ordered conformation of polypeptides and proteins in acidic dodecyl sulfate solution. Biochemistry, 20, 566-570.
Edited by D. Rees (Received 20 July 1998; received in revised form 9 October 1998; accepted 16 October 1998)
J. Mol. Biol. (1999) 285, 1363±1369
COMMUNICATION
Folding of Amphipathic a -Helices on Membranes: Energetics of Helix Formation by Melittin Alexey S. Ladokhin and Stephen H. White* University of California at Irvine, Department of Physiology and Biophysics Irvine, CA 92697-4560, USA
Membranes have a potent ability to promote secondary structure formation in a wide range of membrane-active peptides, believed to be due to a reduction through hydrogen bonding of the energetic cost of partitioning peptide bonds. This process is of fundamental importance for understanding the mechanism of action of toxins and antimicrobial peptides and the stability of membrane proteins. A classic example of membrane-induced folding is the bee-venom peptide melittin that is largely unstructured when free in solution, but strongly adopts an amphipathic a-helical conformation when partitioned into membranes. We have determined the energetics of melittin helix formation through measurements of the partitioning free energies and the helicities of native melittin and of a diastereomeric analog with four D-amino acids (D4,L-melittin). Because D4,L-melittin has little secondary structure in either the free or bound forms, it serves as a model for the experimentally inaccessible unfolded bound form of native melittin. The partitioning of native melittin into large unilamellar phosphocholine vesicles is 5.0(0.7) kcal molÿ1 more favorable than the partitioning of D4,L-melittin (1 cal 4.186 J). Differences in the circular dichroism spectra of the two forms of melittin indicate that bound native melittin is more helical than bound D4,L-melittin by about 12 residues. These ®ndings disclose that the free energy reduction per residue accompanying the folding of melittin in membrane interfaces is about 0.4 kcal molÿ1, consistent with the hypothesis that hydrogen bonding reduces the high cost of partitioning peptide bonds. A value of 0.6 kcal molÿ1 per residue has been observed for b-sheet formation by a hexapeptide model system. These two values provide a useful rule of thumb for estimating the energetic consequences of membrane-induced secondary structure formation. # 1999 Academic Press
*Corresponding author
Keywords: peptide-membrane interactions; membrane protein folding; free energy of transfer; folding energetics; circular dichroism
Membrane interfaces have a potent ability to induce secondary structure in a wide range of membrane-active peptides such as hormones, toxPermanent address: A. S. Ladokhin, Institute of Molecular Biology and Genetics, National Academy of Sciences of Ukraine, Kiev 252143, Ukraine. Abbreviations used: POPG, palmitoyloleoylphosphatidylglycerol; POPC, palmitoyloleoylphosphatidylcholine; OBPC, 1-oleoyl-2(9,10-dibromostearoyl)-sn-3-phosphocholine; LUV, extruded unilamellar vesicles of 100 nm diameter; ANTS, 8-aminonapthalene-1,3,6 trisulfonic acid, trisulfonic acid, 427.33 Da. E-mail address of the corresponding author: [email protected] 0022-2836/99/041363±07 $30.00/0
ins and antimicrobial peptides (Kaiser & Kezdy, 1983, 1984; Schwyzer, 1992; Maloy & Kari, 1995; White & Wimley, 1998). A classic example is the 26-residue bee-venom peptide melittin from Apis mellifera (Figure 1(a)). It is largely unstructured when free in solution, but strongly adopts an amphipathic a-helical conformation when partitioned into membranes (Vogel, 1981; Dempsey, 1990; White et al., 1998). Knowledge of the energetics of this type of folding process is important for understanding and improving the activity of antimicrobial peptides and for understanding the folding and stability of membrane proteins. Wimley & White (1996) have shown that the partitioning of small peptides into membrane interfaces # 1999 Academic Press
1364
Figure 1. The amino acid sequence of native melittin and a general thermodynamic cycle for describing the partitioning and folding peptides in membranes. (a) The amino acid sequence of native melittin. The superscripts in the sequence indicate the positions of the D-amino acid residues in D4,L-melittin. (b) General thermodynamic cycle. For peptides with strong partitioning-folding coupling, the states B and C have very low occupancy so that the effective experimental equilibrium is A $ D. See the text for details.
is dominated by the very unfavorable free energy cost of partitioning the peptide bonds (1.2 kcal molÿ1 per peptide bond for phosphocholine bilayers; 1 cal 4.186 J). They proposed that hydrogen bonding of the peptide bonds reduces this high free energy cost and thereby promotes the formation of secondary structure. Support for this
Melittin Folding on Membranes
proposal is provided by a recent study of the formation of b-sheet aggregates by the hexapeptide AcWL5COOÿ on bilayers formed from POPC, where it was shown that hydrogen bonding reduces the apparent cost of partitioning the peptide bound by about 0.6 kcal molÿ1 per residue (Wimley et al., 1998). Such a modest value for the per-residue free energy change (Gresidue) can have dramatic consequences for secondary structure formation. If ten residues were to participate in hydrogen bond formation in an a-helical peptide, for example, the folded state would be favored by 6 kcal molÿ1, meaning that the folded form would be 26,000 times more abundant than the unfolded form. An important question thus arises: how similar is the value of Gresidue for helix formation to the value observed for b-sheet formation? We examine this question using membrane-induced secondary formation by melittin as a test case. We ®nd that Gresidue for melittin is similar to the value observed for b-sheet formation by AcWL5COOÿ. The results suggest that Gresidue 0.5 kcal molÿ1 may be useful for estimating the energetic consequences of secondary structure formation on membranes. Partitioning-folding coupling The dif®culty with the determination of the energetics of helix folding on membranes is that folding is generally tightly coupled to partitioning. This can be understood from the thermodynamic cycle shown in Figure 1(b). If partitioning-folding coupling is strong, the folded B state and unfolded
Figure 2. Summaries of the estimated and experimentally determined values for partitioning and folding of melittin into POPC interfaces. (a) Estimated values for the partitioning and folding free energies for native melittin. Using the nomenclature of Figure 1(b), three thermodynamic states are shown: unfolded in aqueous phase (state A); unfolded in membrane (state C); and helix in membrane (state D). Melittin has a very low helicity in the absence of membranes. Any melittin in the folded form in water (state B) is assumed to be part of the population of conformations comprising state A. Because of partitioning-folding coupling (see the text), the occupancy of state C is very low and is a virtual state for practical experimental purposes. The range of free energy differences for the partitioning of unfolded melittin was calculated using the whole-residue interfacial hydrophobicity (H) scale of Wimley & White (1996). The range of values results from different assumptions about the interfacial location of the C-terminal polar residues (21-26 in Figure 1(a)). The range of values for the partitioning free energy of melittin in the folded form is based upon various values reported in the literature (Vogel, 1981; Kuchinka & Seelig, 1989; Schwarz & Beschiaschvili, 1989; Beschiaschvili & Baeuerle, 1991; White et al., 1998). Native melittin bound to the membrane is assumed to have 18 residues in a helical conformation (Dempsey & Butler, 1992) and is used to estimate the free energy difference per residue (Gresidue). (b) Experimentally determined values for the partitioning and folding free energies for melittin. Here, in the context of Figure 2(a), the virtual unfolded bound state of melittin is emulated by D4,L-melittin (see the text for details). The helicity difference between the membrane-bound folded and unfolded states was determined using the data of Table 1. The free energies of transfer G of D4,L-melittin and melittin were determined from molefraction partition coef®cients Kx usingG ÿRT ln Kx. The Kx were measured by an equilibrium dialysis method described in detail elsewhere (Wimley & White, 1993, 1996; Ladokhin et al., 1997; White et al., 1998). Brie¯y, a suspension of lipid vesicles was placed in one half of an equilibrium dialysis cell (Spectrum Medical Industries, Houston, TX) that was separated from the other half by a semi-permeable polycarbonate membrane. A solution of peptide was placed in the other half-cell. With constant rotation of the cells in a thermostat, melittin equilibrated across the polycarbonate membrane in less than 24 hours. We generally allowed the system to equilibrate for 48 or more hours at 25 C. Kx ([P]bil/[L])/([P]water/[W]), where [P]bil and [P]water are the bulk molar concentrations of peptide attributable to peptide in the bilayer and water phases, respectively. In the vesicle-containing half-cell, [P]total [P]bil [P]water. [L] and [W] are the molar concentrations of lipid and water (55.3 M). Due to the extremely low partitioning of D4,L-melittin, a correction of [P]water for the lipid side of the dialysis cell was necessary because the inner volume of the vesicles is not accessible to peptides that do not bind well. A correction factor to account for this phenomenon was estimated from distribution of ANTS (Molecular Probes, Eugene, OR), a dye that does not penetrate through the membrane. The correction factor was found to be 0.91 at 80 mM POPC and 0.87 at 130 mM POPC.
1365
Melittin Folding on Membranes
C state will not be signi®cantly populated in terms of experimental detectability, so that the practical thermodynamic equilibrium will between two states: the unfolded A state in solution and the folded D state on the membrane. This is certainly the case for melittin (Figure 2(a)), as demonstrated unequivocally by CD spectroscopy. The partitioning of melittin gives rise to a characteristic a-helix ellipticity at 222 nm that shows a progressive increase with lipid concentration. At the same time, there is an isodichroic point at 204 nm (Vogel, 1981; White et al., 1998), which is the expected consequence of a two-state process. The free energy of partitioning of unfolded melittin
into the bilayer interface of POPC vesicles as the folded helical form, i.e. the A $ D equilibrium (Figures 1(b) and 2(a)), has been estimated by various methods to be ÿ7 to ÿ8 kcal molÿ1 using mole-fraction partition coef®cients (Vogel, 1981; Kuchinka & Seelig, 1989; Schwarz & Beschiaschvili, 1989; Beschiaschvili & Baeuerle, 1991; White et al., 1998). The critically important membrane-bound unfolded C state cannot be observed, but must be present. It can be thought of as a virtual state (Figure 2(a)). Wimley & White (1996) have determined a whole-residue interfacial hydrophobicity scale for POPC interfaces that can be used to
Figure 2 (legend opposite)
1366 estimate the partitioning free energy of unfolded peptides. When applied to the partitioning of melittin, the free energy of transfer from the real A state to the virtual C state can be estimated to lie between ÿ2 and 4 kcal molÿ1, depending upon whether the highly polar C-terminal residues (2126, Figure 1(a)) are assumed to enter the interface. From this calculated estimate of the A $ C equilibrium and the approximate measured value of the A $ D equilibrium, the free energy difference between the unfolded and folded forms of melittin in the interface (the C $ D equilibrium) can be estimated to be between ÿ5 and ÿ12 kcal molÿ1 (Figure 2(a)). The per-residue free energy reduction driving helix formation is estimated from this broad range to be 0.3 to 0.7 kcal molÿ1, assuming a helicity of 18 residues for the folded interfacial form of melittin (Dempsey & Butler, 1992). The value for Gresidue ÿ0.6 kcal molÿ1 obtained for b-sheet formation in POPC vesicles (Wimley et al., 1998) falls comfortably within this range. This suggested that it might be an approximate universal value for estimating the energetics of secondary structure formation by peptides partitioned into bilayer interfaces. Experimental veri®cation of this prospect became possible with the availability of a diastereomeric variant of melittin (Oren & Shai, 1997). Partitioning free energy of unfolded melittin The introduction of D-amino acids into melittin should signi®cantly diminish the formation of secondary structure, while preserving the wholeresidue hydrophobicity. If such a peptide partitioned measurably into bilayers, then the free energy difference between the A state and the virtual C state (Figure 1(b)) could be determined and the energetics of the A $ C $ D thermodynamic cycle consequently estimated. For this purpose, we utilized the melittin diastereomeric analog D4,L-melittin (Oren & Shai, 1997) that has D-amino acid residues in positions 5, 8, 17 and 21 (Figure 1(a)). Their measurement demonstrated that D4,L-melittin does indeed have a greatly diminished helical content compared to melittin. We have con®rmed this observation (Table 1). The free energies of transfer of the two forms of melittin from the aqueous phase to the membrane interface of POPC LUV were measured using equilibrium dialysis (Wimley & White, 1993, 1996; Ladokhin et al., 1997; White et al., 1998) under conditions that assure that native melittin is monomeric in both the aqueous and membrane phases (Dempsey, 1990). The mole-fraction partition coef®cients Kx for D4,L-melittin and melittin were determined to be 80(30) and 4(2) 105, respectively. These values yield free energies of for D4,L-melittin and ÿ2.6(0.3) kcal molÿ1 for native melittin ÿ7.6(0.4) kcal molÿ1 (Figure 2(b)). The free energy difference of ÿ5.0(0.7) kcal molÿ1 between the ``emulated'' unfolded state of melittin and folded melittin is
Melittin Folding on Membranes Table 1. The per-residue ellipticity, y, at 222 nm and the mean helical content, fa, for native melittin and D4,L-melittin in several environments Environment Phosphate (10 mM) Methanol Bound to POPCe,f Bound to POPGe,h
Native melittina yb; fac
D4,L-Melittin
a
ÿ5.4; 0.12 ÿ23.6; 0.69 ÿ23.9; 0.70 ÿ24.0; 0.71
ÿ1.5; 0.00d ÿ8.5; 0.22 n.o.g ÿ10.5; 0.28
yb; fac
a Melittin (sequencing grade) obtained from Sigma (St. Louis, MO). D4,L-Melittin was a gift from Dr Y. Shai. b Units of 103 deg cmÿ2 dmolÿ1. CD measurements were performed using a Jasco-720 spectropolarimeter (Japan Spectroscopic Company, Tokyo). Normally, 10 to 50 scans were recorded between 185 and 260 nm using a 1 mm optical path. UV absorbance was measured with a Cary 3E spectrophotometer (Varian Analytical Instruments, Sugar Land, TX). For stock solutions in methanol, CD and absorbance were measured for the same sample in the same cuvette in order to minimize errors in determination of molar ellipticity. The errors in the determination of y are estimated to be 4 % for the methanol environment and 10 % for lipid environments. c The formula fa (y ÿ yRC)/(yH ÿ yRC) was used to estimate the fractional helical content fa of peptides in a helical conformation, where y is observed ellipticity and yRC and yH are the limiting values for a completely random coil and a completely helical conformation, respectively. Although this formula is simple and well accepted, there is a certain ambiguity in the result due to the uncertainty in prediction of what the actual values for yRC and yH should be (Green®eld & Fasman, 1969; Wu et al., 1981; Luo & Baldwin, 1997; Rohl & Baldwin, 1997; Shalongo & Stellwagen, 1997). Here, we use the following values, both at 222 nm: yRC ÿ1.5 103 deg cmÿ2 dmolÿ1 and yH ÿ33.4 103 deg cmÿ2 dmolÿ1. The yH number is calculated at 25 C for a peptide the size of melittin according to Luo & Baldwin (1997). The number for the random coil is an experimentally observed ellipticity for D4,L-melittin. Choosing the limiting pair in this way is consistent in two respects: (1) it allows one to avoid an unreasonably high helicity for membrane-bound D4,L-melittin; and (2) it brings the estimate for POPC-bound melittin into accord with the results of the amide hydrogen exchange measurements made by Dempsey & Butler (1992). d fa taken as 0 by de®nition (see above). e Lipids were obtained from Avanti Polar-Lipids (Alabaster, AL). Large unilamellar vesicles (LUV) of the lipids with an approximate diameter 0.1 mm were formed by extrusion under N2 pressure through Nucleopore polycarbonate membranes using the method of Mayer et al. (1986). The buffer solution (pH 7.0) used for partitioning experiments contained 10 mM Hepes, 50 mM KCl, 5 mM EDTA and 3 mM NaN3. To reduce the UV absorbance in CD measurements, a 10 mM potassium phosphate buffer (pH 7.0) was used. f Determined by extrapolation to complete binding (White et al., 1998). g Ellipticity not observable because of a very low partition coef®cient. h Measured at saturation (1.6 mM lipid).
consistent with the expected low population of the unfolded membrane-bound state of native melittin and with the hypothesis of that secondary structure formation is driven by a reduction in the free energy of partitioning of peptide bonds caused by hydrogen bonding (Wimley & White, 1996). Because it cannot readily form secondary structure in the interface, D4,L-melittin cannot lower its free energy of transfer into the membrane interface through the hydrogen bonding of secondary structure formation.
Melittin Folding on Membranes
Helicity of the membrane-bound forms of melittin Calculation of Gresidue for melittin helix formation requires that the average number of residues in a helical conformation be determined for the two forms of melittin partitioned into POPC membranes. The dif®culty that arises, however, is that the ellipticity of D4,L-melittin on POPC membranes cannot be measured directly because of its extremely low partition coef®cient. We therefore had to use a less direct approach. We found for native melittin that the ellipticity at 222 nm was virtually identical in methanol, when fully bound hydrophobically to POPC, and when fully bound electrostatically to the anionic lipid POPG. Assuming that a similar situation prevails for D4,Lmelittin, we used measurements of the ellipticity of the two forms of melittin in methanol and when bound to POPG to estimate their helicity difference. The results of measurements of the ellipticities (y) of both forms of melittin in the various phases (methanol, POPC and POPG) are summarized in Table 1 and Figure 3. The results for POPCbound melittin are consistent with numerous measurements on various membranes determined
Figure 3. CD spectra of melittin and its diastereomer analog, D4,L-melittin in methanol (a) and when bound to POPG LUV (b). The ellipticities and calculated helicities are summarized in Table 1. The total peptide concentrations were about 50 mM for the methanol experiments and 20 mM for the POPG experiments.
1367 by both CD and NMR spectroscopy (reviewed by Dempsey, 1990). Fractional helicities were calculated from fa (y ÿ yRC)/(yH ÿ yRC), where yRC and yH are the limiting values for completely random and completely helical conformations, respectively (see the footnotes to Table 1). The data of Table 1 indicate about 18 residues of helix for native melittin and about six residues of helix for D4,L-melittin when membrane-bound. Thus, native melittin has 12 1 more residues in an a-helical conformation than D4,L-melittin. We suggest that the only plausible region to form an a-helix in the diastereomer is T11GLPAL16, because the propensity of this region to form secondary structure is apparently stronger in membrane environments, consistent with the observation that Pro14 does not break the helix in membrane-bound native melittin (Smith et al., 1994). Energetics of melittin folding on membranes The results above, summarized in Figure 2(b), lead to a value of Gresidue of ÿ0.41(0.06) kcal molÿ1 as the per-residue free energy reduction that promotes helix formation of melittin upon partitioning into the POPC bilayer interface. However, we cannot rule out the possibility that D4,L-melittin has a conformation in which some hydrogenbonds are satis®ed internally in other than an a-helical conformation. If that is the case, then the magnitude of Gresidue may be greater than 0.41. Hirota et al. (1998) have recently reported that Gresidue ÿ0.12 kcal molÿ1 for helix formation by melittin in alcohols. Our result demonstrates in comparison the remarkable ability of the membrane interface to induce helix formation. The Gresidue that we observe for POPC membranes is well within the range of those estimated by means of the whole-residue hydrophobicity scale of Wimley & White (1996) and is satisfyingly similar to the value of ÿ0.61(0.04) kcal molÿ1 obtained for b-sheet formation on POPC membranes by AcWL5COOÿ (Wimley et al., 1998). These ®ndings provide a preliminary answer to the question posed at the beginning: the per-residue free energy reduction observed for a-helix formation on membranes is similar to the value for b-sheet formation. AcWL5COOÿ partitions into the POPC interface as a monomer and assembles into b-sheet aggregates of about 12 peptides (Wimley et al., 1998), whereas melittin partitions and folds as a single unit. The difference of 0.2 kcal molÿ1 between Gresidue for AcWL5COOÿ and melittin is not surprising, given that the amino acid sequences are strikingly different and that one must suppose that folding/assembly entropy, side-chain packing, relative exposure of side-chains to membrane and water, and the depth of membrane penetration of the secondary structure units are quite different. These structural differences make it dif®cult to attribute Gresidue solely to a reduction in the cost of partitioning a peptide bond into the interface. Differences in the hydrophobic-effect contributions to partitioning are
1368 likely to be important, for example. Nevertheless, because the cost of partitioning the peptide bond into the interface dominates whole-residue partitioning (Wimley & White, 1996), hydrogen bonding is surely dominant in secondary structure formation. Taking Gresidue ÿ0.4 to ÿ0.6 kcal molÿ1 should provide reasonable ®rst estimates of the free energy difference between the partitioning of folded and unfolded peptides. A comparison of the Gresidue values for helix formation (this work) and b-sheet formation (Wimley et al., 1998) suggests that appropriate ranges of Gresidue may be ÿ0.4 to ÿ0.5 kcal molÿ1 and ÿ0.5 to ÿ0.6 kcal molÿ1 for helix and sheet formation, respectively. The Wimley & White (1996) whole-residue hydrophobicity scale allows one to estimate the partitioning free energy of a small unfolded peptide. The partitioning free energy of the folded form can then be estimated using Gresidue. For example, the application of the whole-residue hydrophobicity scale to magainin 2 (Bevins & Zasloff, 1990) leads to an unfavorable free energy of transfer of 3.6 kcal molÿ1 for the unfolded peptide. Wieprecht et al. (1997) have estimated the helix content of bound magainin as 18 residues. If Gresidue is between ÿ0.4 and ÿ0.5 kcal molÿ1, the partitioning free energy reduction upon folding is between ÿ7.2 and ÿ9.0 kcal molÿ1. These numbers lead to an estimated partitioning free energy of ÿ3.6 to ÿ5.4 kcal molÿ1 for the folded form, which agrees satisfactorily with the experimental value of ÿ4.8 kcal molÿ1 reported by Wenk & Seelig (1998). These results are encouraging, but much remains unknown about the dependence of peptide partitioning on amino acid sequence and membrane lipid composition. The hydrophobicity-hydrophilicity pattern of the peptide sequence that de®nes the hydrophobic moment (Eisenberg et al., 1984) is undoubtedly important. Subtle differences in membrane lipid composition have already been observed to have signi®cant effects on the thermodynamics of b-sheet formation by AcWL5COOÿ. Bromination of the double-bond of the sn-2 chain of DOPC to produce OBPC leads to a decrease of Gresidue from 0.61 to 0.46 kcal molÿ1 (Wimley et al., 1998). Although many details of peptide folding on membranes remain to be discovered, the methods presented here and elsewhere provide general experimental approaches for uncovering these details in order to establish quantitative relationships between the effects of amino acid sequence and membrane composition on the partitioning and folding of peptides into membrane (White & Wimley, 1994, 1998; Wimley & White, 1996; White et al., 1998).
Acknowledgments We are grateful to Dr Yechiel Shai for his generous gift of the melittin diastereomer and to Drs William
Melittin Folding on Membranes Wimley and Kalina Hristova for stimulating discussions. This research was supported, in part, by grant GM-46823 from the National Institute of General Medical Sciences.
References Beschiaschvili, G. & Baeuerle, H.-D. (1991). Effective charge of melittin upon interaction with POPC vesicles. Biochim. Biophys. Acta, 1068, 195-200. Bevins, C. L. & Zasloff, M. (1990). Peptides from frog skin. Annu. Rev. Biochem. 59, 395-414. Dempsey, C. E. (1990). The actions of melittin on membranes. Biochim. Biophys. Acta, 1031, 143-161. Dempsey, C. E. & Butler, G. S. (1992). Helical structure and orientation of melittin in dispersed phospholipid membranes from amide exchange analysis in situ. Biochemistry, 31, 11973-11977. Eisenberg, D., Schwarz, E., Komaromy, M. & Wall, R. (1984). Analysis of membrane and surface protein sequences with the hydrophobic moment plot. J. Mol. Biol. 179, 125-142. Green®eld, N. & Fasman, G. D. (1969). Computed circular dichroism spectra for the evaluation of protein conformation. Biochemistry, 8, 4108-4116. Hirota, N., Mizuno, K. & Goto, Y. (1998). Group additive contributions to the alcohol-induced a-helix formation of melittin: implication for the mechanism of the alcohol effects on proteins. J. Mol. Biol. 275, 365-378. Kaiser, E. T. & Kezdy, F. J. (1983). Secondary structures of proteins and peptides in amphiphilic environments (a review). Proc. Natl Acad. Sci. USA, 80, 1137-1143. Kaiser, E. T. & Kezdy, F. J. (1984). Amphiphilic secondary structure: design of peptide hormones. Science, 223, 249-255. Kuchinka, E. & Seelig, J. (1989). Interaction of melittin with phosphatidylcholine membranes. Binding isotherm and lipid head-group conformation. Biochemistry, 28, 4216-4221. Ladokhin, A. S., Selsted, M. E. & White, S. H. (1997). Bilayer interactions of indolicidin, a small antimicrobial peptide rich in tryptophan, proline, and basic amino acids. Biophys. J. 72, 794-805. Luo, P. Z. & Baldwin, R. L. (1997). Mechanism of helix induction by tri¯uoroethanol: a framework for extrapolating the helix-forming properties of peptides from tri¯uoroethanol/water mixtures back to water. Biochemistry, 36, 8413-8421. Maloy, W. L. & Kari, U. P. (1995). Structure-activity studies on magainins and other host defense peptides. Biopolymers, 37, 105-122. Mayer, L. D., Hope, M. J. & Cullis, P. R. (1986). Vesicles of variable sizes produced by a rapid extrusion procedure. Biochim. Biophys. Acta, 858, 161-168. Oren, Z. & Shai, Y. (1997). Selective lysis of bacteria but not mammalian cells by diastereomers of melittin: structure-function study. Biochemistry, 36, 1826-1835. Rohl, C. A. & Baldwin, R. L. (1997). Comparison of NH exchange and circular dichroism as techniques for measuring the parameters of the helix-coil transition in peptides. Biochemistry, 36, 8435-8442. Schwarz, G. & Beschiaschvili, G. (1989). Thermodynamic and kinetic studies on the association of melittin with a phospholipid bilayer. Biochim. Biophys. Acta, 979, 82-90. Schwyzer, R. (1992). Conformations and orientations of amphiphilic peptides induced by arti®cial lipid
Melittin Folding on Membranes membranes: correlations with biological activity. Chemtracts-Biochem. Mol. Biol. 3, 347-379. Shalongo, W. & Stellwagen, E. (1997). Dichroic statistical model for prediction and analysis of peptide helicity. Proteins: Struct. Funct. Genet. 28, 467-480. Smith, R., Separovic, F., Milne, T. J., Whittaker, A., Bennett, F. M., Cornell, B. A. & Makriyannis, A. (1994). Structure and orientation of the pore-forming peptide, melittin, in lipid bilayers. J. Mol. Biol. 241, 456-466. Vogel, H. (1981). Incorporation of melittin into phosphatidylcholine bilayers: study of binding and conformational changes. FEBS Letters, 134, 37-42. Wenk, M. R. & Seelig, J. (1998). Magainin 2 amide interaction with lipid membranes: calorimetric detection of peptide binding and pore formation. Biochemistry, 37, 3909-3916. White, S. H. & Wimley, W. C. (1994). Peptides in lipid bilayers: structural and thermodynamic basis for partitioning and folding. Curr. Opin. Struct. Biol. 4, 79-86. White, S. H. & Wimley, W. C. (1998). Hydrophobic interactions of peptides with membrane interfaces. Biochim. Biophys. Acta, 1376, 339-352.
1369 White, S. H., Wimley, W. C., Ladokhin, A. S. & Hristova, K. (1998). Protein folding in membranes: determining the energetics of peptide-bilayer interactions. Methods Enzymol. 295, 62-87. Wieprecht, T., Dathe, M., Beyermann, M., Krause, E., Maloy, W. L., MacDonald, D. L. & Bienert, M. (1997). Peptide hydrophobicity controls the activity and selectivity of magainin 2 amide in interaction with membranes. Biochemistry, 36, 6124-6132. Wimley, W. C. & White, S. H. (1993). Quantitation of electrostatic and hydrophobic membrane interactions by equilibrium dialysis and reverse-phase HPLC. Anal. Biochem. 213, 213-217. Wimley, W. C. & White, S. H. (1996). Experimentally determined hydrophobicity scale for proteins at membrane interfaces. Nature Struct. Biol. 3, 842-848. Wimley, W. C., Hristova, K., Ladokhin, A. S., Silvestro, L., Axelsen, P. H. & White, S. H. (1998). Folding of b-sheet membrane proteins: a hydrophobic hexapeptide model. J. Mol. Biol. 277, 1091-1110. Wu, C.-S., Ikeda, K. & Yang, J. T. (1981). Ordered conformation of polypeptides and proteins in acidic dodecyl sulfate solution. Biochemistry, 20, 566-570.
Edited by D. Rees (Received 20 July 1998; received in revised form 9 October 1998; accepted 16 October 1998)