Proc. Natl. Acad. Sci. USA

Vol. 76, No. 8, pp. 3852-3856, August 1979 Biophysics

Pore formation in lipid membranes by alamethicin (peptide antibiotic/peptide-lipid interaction/pore structure/excitable membranes/infrared attenuated total reflection spectroscopy)

URS PETER FRINGELI AND MARIANNA FRINGELI Laboratory for Physical Chemistry, Swiss Federal Institute of Technology, CH-8092 Zurich, Switzerland

Communicated by V. Prelog, April 23,1979

ABSTRACT The conformation of the linear ge tide antibiotic alamethicin in dipalmitoyl phosphatidylcholine multilayers was investigated in the absence of an electric field by means of infrared attenuated total reflection spectroscopy. Alamethicin was found to be incorporated into the lipid membrane not only in the dry state but also in an aqueous environment. Its molecular conformation, however, changed from a helix when dry to an extended chain when aqueous. The extended chain aggregated to di- and multimers spanning the lipid bilayer. The equilibrium concentration of alamethicin in the surrounding water was 90 nM, which is in the range of concentrations used in black film experiments. The corresponding molar ratio of lipid to peptide was 80:1. Concerning the molecular mechanism of electric field-induced pore formation, one has to conclude that the dipole model proposed by several authors is very unlikely because it is based on the assumption that the major part of alamethicin is adsorbed on the membrane surface, from which small amounts flip into the membrane under the influence of an electric field. An alternative mechanism is proposed, based on a field-induced conformational change of the peptide from the extended state to a helix. This transition is favored by the resulting dipole moment of the alamethicin helix.

Alamethicin is a peptide antibiotic isolated from the fungus Trichoderma viride. Under proper conditions it can produce action potentials similar to those of nerve axons (1). After these early investigations in 1968 there was considerable confusion in the literature with respect to the primary structure, the interaction with lipid bilayers, and the mechanism of pore formation; e.g., alamethicin was believed to be a ring peptide with 18 amino acids until Martin and Williams (2) showed in 1976 that the molecule is linear and consists of 19 amino acids. The following primary structure of the alamethicin RF30 (ALA) component is taken from Gisin et al. (3): 1

5

area determined by pressure/area measurements of spread ALA (330 A2), Gordon and Haydon also concluded that there is surface adsorption-i.e., predominantly hydrophilic interaction. The most interesting question, namely that of the molecular mechanism of pore formation under the influence of a properly applied electric field, is thus generally explained by insertion of a small fraction of ALA from the surface into the membrane, leading to conducting pore aggregates (5-15). Contradictory opinions exist only with respect to the origin of different conduction levels of ALA-doped black membranes-i.e., whether they arise (i) from conformational or positional changes within preformed oligomers (5, 11) or (ii) from variations in the number of molecules involved in the oligomer (8-10, 12-15). Most of these conclusions are based on electrical measurements at black lipid membranes-i.e., on information on a macroscopic level, which generally does not directly relate to molecular structure and mechanism. In this paper we report polarized infrared (IR) attenuated total reflection (ATR) spectra of ALA-doped DPPC multibilayer membranes in the absence of an electric field. Experiments have been carried out on materials in the dry state and in an aqueous environment. In the latter case the conditions were similar to those used in black film experiments. From our data, which reveal direct information on molecular structure and orientation, we conclude that at equilibrium ALA must be incorporated into the lipid bilayer. This finding is in contradiction to the assumption of surface adsorption (5-15) and implies that the current models for pore formation have to be revised. METHODS For general information on ATR technique the reader is referred to ref. 16. More details concerning its application to membrane research are given in ref.. 17. ALA was kindly provided by G. B. Whitfield of Upjohn, and L-a-dipalmitoyl phosphatidylcholine was obtained from R. Berchtold (Biochemical Laboratory, Bern). Both components were mixed in chloroform at a molar ratio of 50:1. Some drops of the solution were spread on both sides of a zinc selenide ATR plate. Multibilayers were formed by evaporating the solvent (17). Incorporation of ALA into the lipid bilayer required further exposure of the membrane to liquid water (18-19). The time to reach equilibrium distribution of ALA between the membrane and the surrounding water phase was several hours to a few days at 250C, depending on the thickness of the multilayer membrane (19). In this experiment an average thickness of 28 bilayers was used, resulting in an equilibration time of about 80 hr. The final molar ratio DPPC/ALA was 80:1.

10

Ac-Aibu-Pro-Aibu-Ala-Aibu-Ala-Gln-Aibu-Val-Aibu-Gly15

19

Leu-Aibu-Pro-Val-Aibu-Aibu-Glu-Gln-Phol, in which Aibu is a-aminoisobutyric acid and Phol is phenylalaninol. A distinct hydrophobicity of the peptide is expected from the amino acid composition. However, on the basis of NMR data, Lau and Chan (4) have postulated a predominant hydrophilic interaction between ALA and dipalmitoyl phosphatidylcholine (DPPC). Gordon and Haydon (5) have studied the adsorption of ALA at the interface between glyceryl monooleate in n-decane and sodium chloride solution. The area per molecule was found to be 530 A2. From comparison with the 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.

Abbreviations: ATR, attenuated total reflection; IR, infrared; DPPC, dipalmitoyl phosphatidylcholine; ALA, alamethicin RF3O. 3852

Biophysics: Fringeli and Fringeli RESULTS Spectroscopic considerations Molecular structure determination of ALA is mainly based on the analysis of position, shape, and polarization of typical peptide absorption bands such as amide I [nL80% v(C=O), ;10% v(C-N), z10% b(N-H)], amide II [t60% b(N-H), t40% v(C-N)] (20), and v(N-H). It is well established that the frequencies and shapes of amide group vibrations critically depend on peptide secondary structure (20-22). The IRspectroscopic behavior of a-helices and /3-structures of peptides with finite chain length (n < 18), which is of particular interest in our case, has been investigated by Chirgadze and Nevskaya (23, 24). Polarized light is a powerful tool in our application. However, a prerequisite for its use in molecular structure determination is the knowledge of the direction of the corresponding transition dipole moment. For the N-H stretching mode the direction is expected to be parallel to the N-H bond. Concerning the amide I and II vibrations, the transition moments lie approximately parallel to the C==O double bond and perpendicular to the N-H bond, respectively. However, if the peptide assumes a regular structure, such as an a-helix or A-pleated sheet, the corresponding bands split into two (infinite chain) or more (finite chain) IR-active components. In the case of the a-helix, the main components of the transition moments are directed parallel (amide I) and perpendicular (amide II) to the helix axis, whereas for parallel /-pleated sheet structure the amide I and II vibrations are polarized predominantly perpendicular and parallel to the chain direction, respectively (23, 24). A more detailed discussion of this subject will be published elsewhere. IR-ATR technique also enables efficient measurements in aqueous environment-i.e., under biological conditions. Nevertheless, it is not easy to compensate the strong water absorption bands near 1640 cm-' and 3400 cm-' by means of an ATR reference cell with an accuracy required for a significant analysis of difference spectra (amide I and N-H stretching, respectively) with respect to band position, shape, and polarization. For that reason amide I band analysis was performed in liquid 2H20. Incorporation of ALA into DPPC multibilayer

membranes Special attention must be paid to the preparation of the model membrane. Because it turned out that contact with liquid water is a prerequisite for ALA incorporation into the DPPC bilayers (see above), a given volume of pure water was pumped in a closed cycle through the ATR cell until the ALA amount in the membrane remained constant. The ALA concentration in the surrounding aqueous phase could then be determined from the reduction of the ALA content in the membrane. In a typical experiment it was found that an aqueous concentration of 90 + 20 nM is in equilibrium with a molar ratio lipid/peptide of 80:1 in the membrane. It should be noted that this finding is in good agreement with results published by Gordon and Haydon (5), a fact that needs further discussion because of different interpretations on the molecular level. The striking spectroscopic effects observed upon incorporation of ALA into the DPPC membrane are demonstrated by the IR spectra in the amide I/II region (Fig. 1) and by Fig. 2A. For comparison the sum spectrum is presented in Fig. 1 a. The spectrum of the model membrane as prepared from chloroform solution (molar ratio DPPC/ALA 50:1) by evaporation of the solvent is presented in Fig. 1 b. Treatment of this sample by liquid water for

half an hour followed by careful drying (see below) resulted in the spectrum shown in Fig. 1 c, whereas a different sample that was equilibrated with liquid water for 77.5 hr resulted in the

Proc. Natl. Acad. Sci. USA 76 (1979)

3853

Frequency, cm' FIG. 1. Structural changes of ALA in DPPC multilayer membranes. Changes depending upon sample preparation and environment are reflected by amide I and amide II IR ATR absorption bands (absorbance plot, a = -ln T, in which T is transmittance). 1 and I, Parallel and perpendicular polarized incident light, respectively. a, Sum spectrum of pure DPPC and ALA, dry N2, 250C. b, Model membrane prepared from CHCl3 solution (10 mM DPPC, 0.2 mM ALA), dry N2, 250C. c, Sample b after 30-min contact with liquid H20, dry N2, 50'C after adequate drying. d, Model membrane prepared as for b, 77.5 hr in contact with liquid H20, dry N2, 500C after adequate drying, molar ratio 80:1. e, Sample d in liquid H20 before drying, 250C, molar ratio 80:1. f, Sample d after addition of liquid 2H20 containing 90 nM ALA (equilibrium concentration), 250C, molar ratio 80:1. g, Sample f after adequate drying, kept in N2 (gas) with 30,000 ppm 2H20 (-90% relative humidity), 250C, molar ratio 80:1. h, Sample f after sudden drying with dry N2 (gas), 500C, molar ratio 80:1.

spectrum of Fig. 1 d. All samples (Fig. 1 a-d) have been investigated in a dry nitrogen flow. The temperature was 500C because under these conditions the interference of the OH bending vibration of water bound at the polar part of the membrane with the amide I vibration of ALA (z1650 cm-')

is negligible. The most significant changes in polarization of amide I and II bands may be explained as follows: (i) The amide I and II bands in Fig. 1 a do not show significant polarization. Therefore one may assume that there is no preferred orientation of the amide groups with respect to the ATR plate (17). (ii) Significant change of the amide I band with respect to polarization and shape is observed in Fig. 1 b, thus indicating some molecular ordering of ALA by the lipid bilayer matrix

prepared from chloroform solution. (iii) Considerable polarization of both amide I and amide II bands is achieved by bringing the membrane into contact with liquid water for 30 min, indicating that after this treatment the

3854

Proc. Natl. Acad. Sci. USA 76 (1979)

Biophysics: Fringeli and Fringeli

A.

layer. However, there are many arguments against this inter-

loc

I

pretation:

...

(i) The wavenumbers observed for amide I vibrations are not consistent with a 3-structure (20-24). (ii) ALA chains expanded on top of the DPPC polar heads

B Ar

LIL

t

.1

DPPC

i

\

- ';

a

x1o

J

--

P ~~X2

'--'----''-''--''

x1 -f i

_

A ....rI t\-n-

--

i

'''' ''''1

DPPC/

ALA

....' ''1

0-t L fJW _ apt

x1

3500

3000

2500 Frequency, cm-'

1300

FIG. 2. (A) OH-, N1H-, and N2H-stretching region of sample g from Fig. 1, after drying with N2 (gas), 50'C, molar ratio 80:1. Approximately three amide deuterons nearest to the COOH-terminus have been exchanged by protons resulting from ;150 ppm residual H20 in the N2 (gas) flow; cf. Table 2. (Absorbance plot, a = -ln T.) (B) Difference spectrum between pure DPPC and DPPC/ALA in the CH2-wagging region. Aqueous environment, 250C. (Absorbance plot, = - In T, difference 2X expanded). 1 and I, Parallel and perpendicular polarized incident light, respectively. a

amide I transition dipole moments are oriented predominantly parallel to the hydrocarbon chains of DPPC, whereas the transition moments of amide II are directed perpendicular to the hydrocarbon chains (Fig. 1 c). Polarization is additionally enhanced after the fully equilibrated membrane is dried (Fig. 1 d). The most probable explanation of the observed spectroscopic facts is that under the influence of liquid water ALA penetrated into the membrane. Upon drying the ALA folded to a helix, which, however, remained fully incorporated. As concluded from the amide I dichroic ratio (17), which is >5, and from the results in Table 1, the helix axis must be parallel to the hydrocarbon chains-i.e., approximately perpendicular to the ATR plate. This explanation is fully supported by the polarization of the N-2H stretching vibration [v(N-2H)] at 2500 cm-1 (Fig. 2 A) because the transition dipole moment of this vibration is directed along the N-2H bond-i.e., parallel to the helix axis. Other explanations seem unlikely; e.g., one could imagine that ALA lies unfolded on the top of the polar head groups of DPPC. In order to achieve polarizations comparable to those of Fig. 1 c and d, the peptide chains must assume a (-sheet structure, with a sheet orientation perpendicular to the plane of the biTable 1. Influence of ALA on hydrocarbon chain ordering and orientation in dry DPPC bilayers at 250C 4* al a1 it Remarks Sample 1.1 From CHCl3 0.135 270 0.391 DPPC/ALA 50:1 2.3 Equilibrated 0.147 190 0.758 DPPC/ALA 80:1 DPPC 310 1.0 Equilibrated 0.165 0.364 * Angle between hydrocarbon chains and normal to bilayers; cf. ref.

17. t Increase of all-trans hydrocarbon chains with respect to pure DPPC.

must be bound to them by hydrogen bonds oriented perpendicularly to the bilayer plane. On the other hand, hydration of the membrane enhances the amount of water in the hydrophilic region between two adjacent lipid double layers about 10-fold to ;25 wt %, resulting in a thickness of the water layer of -20

A(25). Because it is very unlikely that the hydrogen bonds mentioned above are sterically protected against access of water molecules, two easily detectable effects should be expected: first, rapid hydrogen-deuterium exchange of all amide protons in a humid atmosphere (30,000 ppm 2H20), and second, a strong reduction of the amide I polarization. Both effects require dissociation of the crosslinking and orienting hydrogen bonds between lipid and peptide. This process should be favored by the considerable swelling of the water layer as mentioned above, resulting in a competition between lipid/peptide and lipid/H20 hydrogen bonding, respectively. However, neither of the two effects could be observed (cf. Table 2 and Fig. 1 g). (iii) In the case of dry ALA/L-1-palmitoyl-2-palmitoleoyl phosphatidylcholine there is a significant interaction of ALA with partially oxidized double bonds, indicating that ALA must penetrate at least as far as the middle of a lipid monolayer. On the other hand, if the penetration depth of ALA were only half a monolayer one would expect a significant lowering of hydrocarbon chain ordering. However, it is found that hydrocarbon chain ordering in DPPC is significantly enhanced (Table 1) by interaction with ALA in dehydrated membranes. This interesting fact will be discussed in detail elsewhere. Because helical ALA assumes about the same length as DPPC, one should assume that the helical rods are fully incorporated into one monolayer of the bilayer membrane. (iv) The tilt angle between the hydrocarbon chains of DPPC and the normal to the bilayer is reduced from 270 to 190 upon interaction of ALA (cf. Table 1). It is unlikely that such a reorientation of the whole membrane can be induced by ALA adsorbed only to the polar head groups. (v) Comparing the N-H stretching bands [v(N-H)] of a nonequilibrated and an equilibrated DPPC/ALA membrane, one observes a shift from 3330 cm-1 to 3350 cm-1 accompanied by narrowing of the shape (26). [For comparison, v(N-H) of pure crystalline ALA is observed at 3322 cm-'. ] This fact points again to an enhanced hydrophobic rather than hydrophilic environment when ALA is transferred from the nonequilibrated state into the equilibrated state of the membrane. This observation is paralleled by a corresponding shift of the amide I band (Fig. 1 a-d), which is also expected as a result of enTable 2. 1H-2H exchange of amide protons at 250C Estimated flexible Accessible amide sequence, Time to ALA equiligroups, Humidity, % residues brate ppm System 1 day 15 ± 8 17-19 DPPC/ALA 150 41 ± 3 -1 day 12-19 80:1 30,000 100 1-19 Minutes Liquid water ± ALA 30,000 -1 day 33 3 14-19 100 1-19 Minutes Liquid water At 250C, 30,000 ppm H20 90% relative humidity. Note that residue 19 is the COOH-terminal amino acid of ALA.

Biophysics: Fringeli and Fringeli hanced interaction of some amide groups near the COOH terminus with the hydrocarbon chains of the lipid. (vi) Finally, one should also consider that ALA consists of 16 hydrophobic and only 3 hydrophilic residues, a fact that should favor incorporation into the lipid phase. Supported by the arguments mentioned above, one is led to the conclusion that for equilibrated dry DPPC/ALA model membranes ALA must be completely incorporated into the hydrophobic phase and assume a helical structure. The COOH-terminus is in contact with the polar headgroups of DPPC. Further information on the interaction of the COOH terminus of ALA with the polar headgroups of phosphatidylcholine is given elsewhere.

Unfolding of ALA helix in an aqueous environment So far only structural features of dry DPPC/ALA membranes have been discussed. In order to get relevant biological information it must be checked whether the molecular structures of DPPC and ALA remain unaltered in an aqueous environment. This, however, is not the case at all, especially for ALA. In order to keep the amide I region free from H20 (liquid) absorption, the membrane was investigated in liquid 2H20 containing the corresponding equilibrium concentration of ALA. Two striking effects were observed: (i) The high degree of amide I polarization observed in equilibrated dry membranes (Fig. 1 d) vanishes completely, paralleled by a shift of the amide I band to lower wavenumbers (Fig. 1 f). (ii) The amide II band (-60% 6 (N-H) vanishes after a few minutes; i.e., the amide hydrogen atoms are replaced by deuterium.

Fact ii implies that the ALA helix becomes unstable or is even unfolded upon addition of liquid water, because otherwise the 'H-2H exchange could not be as rapid as was-observed. The significant frequency shift of the amide I band between deuterated helical ALA (Fig. 1 g) and the conformation existing in liquid 2H20 (Fig. 1 f) indicates that in the latter case a helical conformation is unlikely. Moreover, the drastic change in polarization must result from a rearrangement of amide I transition moments from parallel to predominantly perpendicular orientation with respect to the hydrocarbon chains. A reorientation of the helix axis parallel to the plane of the bilayers would explain the change in polarization but not the observed frequency shift. There is, however, a conformational change of ALA that is in agreement with experimental facts i and ii-namely, unfolding of the helix and aggregation to a parallel 3-structure with peptide chains directed parallel to the hydrocarbon chains. Concerning the frequencies of amide I band components, it should be noted that the high-frequency component at 1653 cm-1 is consistent with a parallel arrangement of two extended ALA chains (see above and ref. 24), whereas the low frequency component at 1635 cm-1 is consistent with a hexameric parallel association of ALA chains (24). These aggregates are long enough to span the lipid bilayer membrane and may be considered as preaggregates for pore formation. Several additional experiments have been performed to check the interpretation above. First, liquid H20 was used instead of 2H20. No use of the amide I band was made because of strong overlapping with the OH bending mode of H20. However, liquid water absorption can be compensated without problems in the amide II region by means of an adequate ATR cell in the reference beam. As demonstrated by Fig. 1 e, this band exhibits a distinct z-polarization (dichroic ratio RATR 2.6)-i.e., perpendicular to the plane of the bilayer (17)-which indicates that the mean orientation of the N-H bonds of the amide groups is approximately parallel to this plane, a finding

Proc. Natl. Acad. Sci. USA 76 (1979)

3855

that is consistent with the unpolarized amide I band (Fig. 1 f), thus supporting the interpretation of extended ALA chains spanning the membrane. At this point it should be mentioned that for random orientation of ALA in a membrane with an aqueous environment the dichroic ratio should be 1.55 for both amide I and amide II bands (cf. ref. 17). Second, the attempt was made to "freeze" ALA in the extended state by sudden drying of the membrane in a nitrogen flow. Fig. 1 h shows that a considerable fraction of ALA chains or parts of them could be kept in the extended state, as concluded from polarization and band position. In dry membranes, however, this state is thermodynamically unstable; therefore helix formation should spontaneously-occur. The rate can be drastically enhanced when the membrane is exposed to a wet atmosphere (-30,000 ppm, t90% relative humidity at 250C) for about 30 min at 250C followed by a temperature rise to 50'C, where the sample was kept for further 30 min before being dried in a nitrogen flow and cooled down to 25"C again. By this procedure the strong amide I polarization commonly observed in dry equilibrated DPPC/ALA membranes is restored. It should be noted that the same procedure, when applied to nonequilibrated membranes (Fig. 1 b and c), is not able to enhance the polarization of the amide I band, a fact that also supports the interpretation that ALA chains expanded by liquid water span the membrane-i.e., are incorporated into the hydrophobic phase. Third, information about the interaction of ALA with saturated hydrocarbon chains can be obtained from the wagging sequence ['y,(CH2)] of the methylene groups. This sequence is produced only by hydrocarbon chains assuming an all-trans conformation (17). Dry DPPC/ALA membranes exhibit a significantly higher amount of all-trans hydrocarbon chains than pure DPPC (Table 1); however, if the same membrane is exposed to liquid water, hydrocarbon chain ordering becomes smaller than in the corresponding pure DPPC membrane, as demonstrated by the difference spectrum in Fig. 2B. This is good evidence that under these conditions ALA must assume a conformation less regular than the helical rods existing in the dry system, thus reducing hydrocarbon chain ordering of the lipid. Finally, one should remember that contact with liquid water is a prerequisite for incorporation of ALA into the DPPC membrane. This fact again supports the conclusion that in an aqueous environment ALA must be located in the membrane and not as an adsorbate on the polar head groups. Molecular flexibility of ALA It has been demonstrated above that the molecular flexibility of alamethicin is drastically enhanced by interaction with liquid water. The question now arises whether helix destabilization can be achieved also by hydration from the gas phase. Hydrogen-deuterium exchange experiments are an excellent tool to obtain this information. These experiments will be described in a separate paper (26), but the main results are summarized in Table 2. Two interesting facts should be mentioned. First, the flexibility near the COOH-terminus is slightly enhanced upon incorporation of ALA into the lipid membrane, a fact that is also reflected by the appearance of a high-frequency component of the amide I band (Fig. 1 d). Second, the helix of amino acids 1-11 has turned out to be stable enough to prevent any water access from the gas phase at 250C during a period of observation of several weeks. From Table 2 one has to conclude that ALA exhibits at least three different degrees of flexibility along its chain. Furthermore, the fluctuation frequency for helix unfolding depends critically on the degree of hydration. Evidently in dry membrane the helical conformation

3856

Proc. Natl. Acad. Sci. USA 76 (1979)

Biophysics: Fringeli and Fringeli

is predominant, whereas in an aqueous environment ALA is predominantly unfolded and spans the membrane.

DISCUSSION In the sections above it has been demonstrated that in the absence of an electric field ALA is without doubt located in the membrane and not adsorbed to its polar surface as generally assumed. As a consequence, molecular mechanisms proposed for pore formation must be revised if the assumption of surface adsorption was essentially involved (8-10, 12-14). Finally, it must be shown that our investigations in aqueous environment have been performed under conditions corresponding to those used in black film experiments. The typical equilibrium concentration of ALA in the aqueous phase was determined to be 90 i 20 nM, which is in the range of antibiotic concentrations used for single pore measurements (27). The molar ratio DPPC/ALA in the membrane corresponding to the concentration mentioned above was 80:1, resulting in 3.2 1010 ALA molecules per mm2 of a monolayer. It should be noted that this value is in good agreement with the determination of surface concentration by interfacial tension measurements that resulted in zt10l0 ALA molecules per mm2 in a glyceryl monooleate monolayer; this is equivalent to an area per ALA molecule of. 530 A2 at saturation (5). In the view of our results this area cannot be related to the area of the expanded molecule (330 A2), because also in this case ALA would be expected to be incorporated into the glyceryl monooleate membrane-i.e., have a much smaller cross-section. A final remark concerning the molecular mechanism of pore formation should be made. At present it is not possible to give a decisive answer because experiments resulting in information on a molecular level are not available. Application of the IRATR technique with electric field stimulation should be informative. Nevertheless, it should be noted that our results presented here strongly support a mechanism already proposed by Eisenberg et al. (7), Gordon and Haydon (5), and Hall (28): ALA aggregates open the pore by field-induced conformational changes. Our findings indicate that these aggregates are not located at the membrane water interface but span the membrane, forming a kind of dimeric to multimeric parallel 3-structures (j3). Because ALA can also assume helical conformation (a) depending on the environment, we propose a conformational equilibrium a r± 13 that in an aqueous environment is predominantly on the right-hand side. Jung et al. (29) have found evidence favoring this proposal. According to our results, the chains in the :-structure have the same direction as the helix axis of a. Thus when an electric field E is applied across the membrane with the positive pole on the side of the COOHterminus the equilibrium constant K = [1A]/[aLis altered in favor of the helix according to K = K0-exp[-(-,E)/RT]. Here , denotes the dipole moment per mole of helices, R is the gas constant, and T is the absolute temperature. Assuming 3.5 debyes as dipole moment per amide group (30) (charge separation equal to one-half elementary charge), one may expect a maximum dipole moment of about 70 debyes for completely helical ALA. In the case of El = 200 kV/cm one obtains K = K0-0.31. The amount of helix increase p depends on the magnitude of K°, the equilibrium constant at rest. For an arbitrary value of K° = 102, p will be about 2% of the initial amount of fl-structure according to 1 - exp[-(-,E)/RT] 1 + KO -

exp[-(A4,E-)/RT]

100%.

Of course this treatment is much too simple for adequate description of the complex process. Nevertheless it helps to estimate the fraction of field-induced conformational changes.

Helix flexibility is smallest near the NH2 terminus; i.e., local KO is minimum. This fact should favor field-induced initiation of helix formation. Furthermore, it seems very probable that ALA does not respond to the electric field as an isolated molecule but rather cooperatively with adjacent elongated chains of the corresponding aggregate resulting in a more efficient field effect. Because, in this case the number of molecules per aggregate should play a role, one could imagine that the probability for pore opening at a given electric field strength will depend on the size of the aggregate. These reflections demonstrate that electric field-induced helix formation could be a mechanism for pore opening due to local disturbance of the membrane. However, further experimental work is required for substantiation. We thank Professor H. H. Gfinthard for continuous interest and support. Financial support by the Swiss National Science Foundation (Projects no. 3.521.0,75 and no. 3.192.0.77) and by the Emil Barell Foundation (Hoffmann-La Roche, Basel, Switzerland) is gratefully

acknowledged. 1. Mueller, P. & Rudin, D. 0. (1968) J. Theor. Biol. 18,222-258. 2. Martin, D. R. & Williams, R. J. P. (1976) Biochem. J. 153, 181-190. 3. Gisin, B. F., Kobayashi, S. & Hall, J. E. (1977) Proc. Nati. Acad. Sci. USA 74,115-119. 4. Lau, A. L. Y. & Chan, S. I. (1975) Proc. Nati. Acad. Sci. USA 72, 2170-2174. 5. Gordon, L. G. M. & Haydon, D. A. (1975) Philos. Trans. R. Soc. London 270,433-447. 6. Cherry, R. J., Chapman, D. & Graham, D. E. (1972) J. Membr. Biol. 7, 325-344. 7. Eisenberg, M., Hall, J. E. & Mead, C. A. (1973) J. Membr. Biol. 14, 143-176. 8. Baumann, G. & Mueller, P. (1974) J. Supramol. Struct. 2, 538-557. 9. Boheim, G. (1974) J. Membr. Biol. 19,277-303. 10. Mueller, P. (1976) Ann. N. Y. Acad. Scd. 264,247-265. 11. Gordon, L. G. M. & Haydon, D. A. (1976) Biochim. Biophys. Acta 436,541-556. 12. Boheim, G., Kolb, H.-A., Bamberg, E., Apell, H.-J., Alpes, H. & Lauger, P. (1977) in Electrical Phenomena at the Bilogical Membrane Level, ed. Roux, E. (Elsevier, Amsterdam), pp. 289-310. 13. Boheim, G. & Benz, R. (1978) Biochim. Biophys. Acta 507, 262-270. 14. Boheim, G. & Kolb, H.-A. (1978) J. Membr. Biol. 38,99-150. 15. Kolb, H. A. & Boheim, G. (1978) J. Membr. Biol. 38, 151-191. 16. Harrick, N. J. (1967) Internal Reflection Spectroscopy (Wiley Interscience, New York). 17. Fringeli, U. P. (1977) Z. Naturforsch. 32c, 20-45. 18. Fringeli, U. P., Fringeli, M. & Guinthard, Hs. H. (1978) Ber. Bunsenges. Phys. Chem. 82,922. 19. Fringeli, U. P. (1979) J. Membr. Biol., in press. 20. Miyazawa, T. (1967) in Biological Macromolecules, ed. Fasman, G. D. (Dekker, New York), Series 1, pp. 69-103. 21. Koenig, J. L. (1972) J. Polymer Sci. D-6, 59-177. 22. Frushor, B. G. & Koenig, J. L. (1975) in Advances in Infrared and Raman Spectroscopy, eds. Clark, R. J. H. & Hexter, R. E. (Haydon & Son, London), Vol. 1, pp. 35-97. 23. Nevskaya, N. A. & Chirgadze, Yu. N. (1976) Biopolymers 15, 637-648. 24. Chirgadze, Yu. N. & Nevskaya, N. A. (1976) Biopolymers 15, 627-636. 25. Gottlieb, M. H. & Eanes, E. D. (1974) Biophys. J. 14, 335342. 26. Fringeli, U. P. (1979) Z. Naturforsch. Tedl C, in press. 27. Boheim, G., Irmscher, G. & Jung, G. (1978) Biochim. Biophys. Acta 507, 485-506. 28. Hall, J. (1975) Biophys. J. 15, 934-939. 29. Jung, G., Dubischar, N. & Leibfritz, D. (1975) Eur. J. Biochem. 54,395-409. 30. Hol. W. G. J., van Duijnen, P. T. & Berendsen, H. J. C. (1978) Nature (London) 273, 443-446.