Spectroscopic characterization of the alkylated α-sarcin cytotoxin: analysis of the structural requirements for the protein-lipid bilayer hydrophobic interaction

ELSEVIER

Biochimica et Biophysica Acta 1252 (1995) 43-52

Biochl"PmPic~a et BiophysicaA~ta

Spectroscopic characterization of the alkylated a-sarcin cytotoxin: analysis of tlhe structural requirements for the protein-lipid bilayer hydrophobic interaction Maria Gasset, Jos6 M. Manchefio, Javier Lacadena, Alvaro Martinez del Pozo, Mercedes Ofiaderra, Jos~ G. Gavilanes * Departmento de Bioqufmica y Biologfa Molecular, Facultad de Qufmica, Universidad Complutense, 28040 Madrid, Spain Received 24 October 1994; revised 21 April 1995; accepted 9 May 1995

Abstract

a-Sarcin is a ribosome-inactivating protein that translocates across lipid bilayers, these two abilities explaining its cytotoxic character. This protein is composed of a single polypeptide chain with two disulfide bridges. Reduction and carboxyamidomethylation of a-sarcin results in protein unfolding, based on the results of the spectroscopic characterization of the chemically modified protein. The absorption and fluorescence emission bands of the tryptophan residues of the modified protein appear blue- and red-shifted, respectively. Far-UV circular dichroism analysis reveals the presence of residual secondary structure (/~-strands and turns) in the alkylated protein. This retains its ability to interact with lipid bilayers. It promotes vesicle aggregation, lipid-mixing between bilayers and leakage of the intravesicular aqueous contents. The modified protein tends to abolish the phase transition of acid phospholipids as detected by differential scanning calorimetry and depolarization measurements of fluorescence-labelled vesicles. The protein gain access to vesicle-entrapped trypsin. The fluorescence emission of the tryptophan residues is blue-shifted upon interaction of the protein with the bilayers, and anthracene incorporated into the hydrophobic core of the membranes quenches the tryptophan fluorescence emission of the protein. The secondary structure of the alkylated protein interacting with lipid vesicles has been studied by infrared spectroscopy. An increase in the a-helix and turn contents and a concomilant decrease in the /3-structure content are observed upon interaction with the bilayers. The results obtained are discussed in terms of the structural requirements for the interaction of c~-sarcin with lipid membranes. Keywords: Antitumor protein; Cytotoxin; Protein unfolding; Spectroscopy of protein; Protein-lipid interaction; ot-Sarcin

1. Introduction

a-Sarcin is the best known proteins produced by several hibit more than 90% amino-acid was early described as a potent several tumours induced in mice

member of a family of that exsequence similarity [1 ]. It antitumour agent against [2,3]. Recent studies have

Aspergillusstrains,

Abbreviations: ANTS, 8-aminonaphthalene-l,3,6-trisulfonic acid; CD, circular dichroism; DMPC, dimyristoylphosphatidylcholine; DMPG, dimyristoylphosphatidylglycerol; DMPS, dimyristoylphosphatidylserine; DPH, 1,6-diphenyl-l,3,5-hexatriene; DPX, p-xylene bis(pyridinium bromide); DSC, differential scanning calorimetry; FTIR, Fourier transform infrared; NBD-PE, N-(7-nitro-2-1,3-benzoxadiazol-4-yl)dimyristoylphosphatidylethanolamine; PG, phosphatidylglycerol; RET, resonance energy transfer; Rh-PE, N-(lissamine rhodamine B sulfonyl)diacylphosphatidylethanolamine; o~SRC, reduced and carboxyamidomethylated o~-sarcin. * Corresponding author. Fax: +34 1 3944159. 0167-4838/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 01 67 - 4 8 3 8 ( 9 5 ) 0 0 1 06-9

revealed that a-sarcin is cytotoxic for many human tumour cell lines [4], and it is being studied as anfitumour agent under the form of either free protein or conjugated to monocional antibodies (immunotoxin) [5]. This cytotoxin displays a very specific ribonuclease activity that inactivates the ribosomes (see [6] for a review) and inhibits the cellular protein biosynthesis. Therefore, the cytotoxicity of a-sarcin would involve two general steps, i.e., entering the cell membrane and inactivation of the ribosomes. No protein membrane receptors have been so far described for a-sarcin. However, this protein strongly interacts with phospholipid vesicles promoting their aggregation and fusion [7-9]. The protein translocates across the membrane of asolectin vesicles in the absence of any permeabilizing agent [10], in a way which may be related to its ability in passing through cell membranes. The interaction of c~-sarcin with lipid vesicles shows an electrostatic component. This may be explained by the net

44

M. Gasset et al. /Biochimica et Biophysica Acta 1252 (1995) 43-52

positive charge of the protein [11], although the three-dimensional distribution of charge inherent to the native protein conformation could be in addition required. The effect of a-sarcin on bilayers also shows a hydrophobic component [10,12], which is difficult to explain by simply considering the primary structure of the protein. This is a highly polar molecule without hydrophobic stretches of more than three residues [11]. Thus, an unfolded a-sarcin form might give valuable information about the conformational requirements of the protein-vesicle interaction. The study of unfolded proteins has recently gained interest because the growing evidence on the involvement of these protein forms in cellular processes. Thus, although protein translocation across membranes is a not well understood biochemical event, partial protein unfolding plays a major role in the processes occurring in the membrane insertion and translocation of bacterial protein toxins [ 13-15]. Also, imported proteins need to have a translocation competent conformation (partially folded) [16,17] for translocation across membranes to occur, ce-Sarcin is a protein composed of a single polypeptide chain of 150 amino acids containing four cysteine residues, which form two disulfide bridges [12]. In addition, one of the two disulfides bridges in ct-sarcin would apparently be a quite important structural constraint, since it is built between cysteine 6 and cysteine 148 bringing together both the NH 2- and the -COOH terminal ends of the molecule. In the experiments described below, we have prepared alkylated a-sarcin in order to produce an unfolded protein form. The study of the interaction of reduced and carboxyamidomethylated ce-sarcin with lipid bilayers has been performed to analyze the structural requirements that c~-sarcin fits on its interaction with membranes. 2. Materials and methods

a-Sarcin was purified from cultures of Aspergillus giganteus MDH 18894. The isolation procedure was based on that described [2], with minor modifications. The protein concentration was determined by absorbance measurements at 280nm based on an E(1 mg/ml, 280nm, l cm) value of 1.34 [18]. Reduction of disulfide bonds of c~-sarcin and carboxyamidomethylation of the resulting thiol groups was performed as follows: c~-sarcin ( 1 0 m g / m l protein concentration) was maintained for 1 hour at 37°C in 1.0 M Tris-HC1 buffer (pH 8.0) containing 2 mM EDTA and 6 M guanidinum hydrochloride; reduction of disulfide bridges was achieved by addition of 25 mM dithiothreitol final concentration (10-fold molar excess over the total cysteine content) and incubation at 37°C for 90 min; iodoacetamide was then added to a final concentration of 250mM, and the mixture was incubated in the dark for another 60 min at 37°C. The alkylated protein was purified by gel filtration on a Biogel P-2 column equilibrated in 50 mM acetic acid and lyophilized. The alkylated protein does not produce the specific 400-nucleotide ce-fragment from the 28S rRNA

of eukaryotic ribosomes, under the assay conditions previously described for native ce-sarcin [19,20]. Protein samples were hydrolysed in evacuated and sealed tubes for 24h at 105°C in 5.7N HC1, containing 0.1% ( w / v ) phenol. Amino-acid analyses (three determinations for each protein sample) were performed on a Beckman Model 6300 automatic analyzer equipped with an IBM-AT based System Gold enhancement. Circular dichroism (CD) spectra were obtained on a Jobin Yvon Mark III dichrograph fitted with a 250W xenon lamp. The spectra were recorded at 0.2 n m / s scanning speed. Samples were analyzed in either 0.05 or 0.01 cm optical path cells. CD results were expressed in units of deg. cm 2 dmo1-1 of amino-acid residue. These values were calculated on the basis of 113 as the mean residue weight for the amino acids in c~-sarcin [11]. Fluorescence measurements were performed on a SLM Aminco 8000 spectrofluorimeter. The contribution of the solvent to the spectra has been subtracted for all samples. Lightscattering effect of the protein-lipid complex was reduced by exciting the sample with vertically polarized light and measuring the horizontally polarized emitted light. Spectra were recorded in 0.2 cm optical-path thermostated cells. The slit widths were 4 nm for both excitation and emission beams, c~-Sarcin contains 8 tyrosine and 2 tryptophan residues [11]. The relative contribution of both types of fluorophores to the intrinsic protein fluorescence emission spectra was evaluated from the emission spectra recorded for excitation at both 275 and 295 nm. Normalization of both emission spectra was performed by considering that the protein fluorescence emission above 380 nm only arises from the tryptophan residues. Synthetic phospholipids, dimyristoylphosphatidylglycerol (DMPG), dimyristoylphosphatidylcholine (DMPC), dimyristoylphosphatidylserine (DMPS), were purchased from Avanti Polar Lipids; egg yolk phosphatidylglycerol (PG) was from Sigma-Aldrich. Soybean asolectin (Sigma) was purified by dry acetone treatment. The different lipid vesicles were prepared at 1 m g / m l phospholipid concentration in 50mM Mops buffer (pH 7.0) containing 0.1M NaC1 and 1 mM EDTA. Sonication for 30 min in a water-bath sonifier at 5°C above the phase transition of the corresponding phospholipid a n d / o r freeze-thawing and extrusion through 0.1/zm pore polycarbonate membranes (five cycles) were performed for the vesicle preparation. The average size of the vesicle population was 100nm as determined by Coulter counting. All the different protein-lipid complexes were prepared by adding the protein to freshly prepared vesicles at the required lipid/a-sarcin molar ratio [7]. The lipid concentration was determined as described [21 ]. Trypsin-containing asolectin vesicles were prepared as described above, except that the lipid film was rehydrated (at 1 0 - 2 0 m g / m l phospholipid concentration) in 50mM Mops buffer (pH 7.0) containing 0.1 M NaCl and 8 1 0 m g / m l Tos-Phe-CHzC1 ('TPCK')-treated trypsin

M. Gasset et al. / Biochimica et Biophysica Acta 1252 (1995) 43-52

(Merck). The vesicles were separated from the non-encapsulated material in Sephadex G-75 columns packed in a 5 ml disposable syringe; w~sicles were eluted by centrifugation of the column at 500 g for 2 min. This procedure was repeated twice to remove all untrapped trypsin [10]. Vesicles containing coencapsulated 8-aminonaphthalene-l,3,6-trisulfonic acid (ANTS) and its quencher DPX (p-xylylene bis(pyridinium bromide)) were used for the vesicle leakage experiments. Vesicles were prepared by five cycles of freeze-thawing and further extrusion through polycarbonate membranes (0.1/.~m pore diameter), in 10mM Tris-HC1 buffer (pH 7.5) containing 20mM NaC1, 12.5 mM ANTS and 45 mM DPX. Vesicles were separated from the non-encapsulated material by gel filtration on a Sephadex G-75 column ,equilibrated in 10mM Tris-HC1 buffer (pH 7.5) containing 100mM NaCI and 1 mM EDTA [22]. The emission of ANTS in the leakage experiments was measured at 510nm for excitation at 386nm [22]. The fluorescence emission of the vesicles containing both ANTS and DPX is set to zero (%) leakage, while the fluorescence emission of the lysed vesicles (by adding 1% Triton X-100) is set to 100% leakage. Quenching of the fluorescence emission of protein tryptophans by anthracene incorporated in phospholipid vesicles has been also analyzed. Anthracene was added to the chloroform lipid solution, at the required molar ratio, prior to the preparation of the dry film of phospholipids. Other details of the vesicle preparation were as above described. Emission spectra for excitation at 295 nm were recorded after addition of the protein to the vesicle suspension. The apparent (%) energy transfer from tryptophans to anthracene is defined by the equation: (%) = [1 - ( F / F o ) ] × 100, when F and F 0 are the fluorescence emission intensities at 340nm in the presence and in the absence of anthracene, respectively. Lipid-mixing of membranes was monitored as described [23]. 1% N-(7-nitro-2-1,3-benzoxadiazol-4-yl)dimyristoylphosphatidylethanolamine (NBD-PE) and 0.6% N-(lissamine-rhodamine B sulfonyl)diacylphosphatidylethanolamine (Rh-PE) were incorporated into the lipid vesicles. These fluorescence labelled vesicles were mixed with unlabelled vesicles. (1:9 proportion). Lipid-mixing resulted in an increase of the NBD-PE fluorescence due to the decrease in the fluorescence energy transfer between NBD-PE (donor) and Rh-PE (acceptor) by dilution of the fluorescence probes due to the mixing of the membrane lipids. The efficiency of the energy transfer (%RET) is defined as above, the fluorescence emission intensities at 530 nm for excitation at .450 nm in the presence and in the absence of Rh-PE being F and F 0, respectively. Normalization of the measurements was performed by independent experiments in the presence of 1% Triton X-100 [23]. The kinetic analysis of the lipid-mixing was performed by continuously recording the NBD-PE fluorescence emission variation in an assay consisting of (1:1) labelled/unlabelled vesicles.

45

Fluorescence depolarization measurements were performed on the SLM Aminco 8000 spectrofluorimeter equipped with 10mm Glan-Thompson polarizers. Labelling of the vesicles with 1,6-diphenyl-l,3,5-hexatriene (DPH) was performed as previously described [24]. Protein-vesicles mixtures were incubated for 1 h above the transition temperature of the corresponding phospholipid, and later cooled down. The fluorescence emission was measured at 425 nm for excitation at 365 nm, after equilibration of the samples at the required temperature. This parameter was measured by using an inner temperature probe. Aggregation of the lipid vesicles induced by the protein was measured by recording the absorbance variation produced by addition of the protein to a lipid vesicle preparation. Absorbance at 360nm was continuously monitored on a Beckman DU-8 spectrophotometer equipped with a thermostated cell holder. Controls without protein were performed for each experiment. Differential scanning calorimetry (DSC) analysis of the protein-lipid complexes was performed on a Microcal MC 2 calorimeter at 30°C/h scanning rate, as previously described [12]. The calorimetric unit was interfaced to an IBM PC microcomputer for automatic data collection and analysis. The analyses were performed with lipid-protein complexes prepared at 4°C in the instrument cell. A heating/cooling cycle was first completed. No differences on the further recorded scans were observed after repeated heating/cooling cycles. Fourier transform infrared (FTIR) spectroscopy measurements were performed on a Nicolet 520 instrument equipped with a DTGS detector. Samples were placed into a thermostated demountable cell (Harrick) with CaF 2 windows. Samples were prepared on the window by mixing 10/.tl of deuterated-buffer (d-buffer; 50mM Mops buffer (pH 7.0) containing 0.1 M NaC1, prepared in 2H20) or the required amounts of phospholipid in d-buffer. Phospholipids were maintained for 1 h above the phase transition temperature in the d-buffer. Protein samples were maintained in d-buffer for at least 2 h prior any measurement to allow hydrogen-deuterium (H × D) exchange. The path length used was 50/xm and the sample chamber was continuously purged with dry air. A minimum of 216 scans per sample were taken, averaged, apodized with a HappGenzel function and Fourier-transformed to give a nominal resolution of 2cm -~. Amide I' band was analyzed by means of Fourier self-deconvolution and curve fitting according to [25]. 3. Results 3.1. Spectroscopic characterization o f the reduced and carboxyamidomethylated ol-sarcin

Amino-acid analysis of the reduced and carboxyamidomethylated a-sarcin ( a S R C ) reveals the pres-

46

M. Gasset et al, / Biochimica et Biophysica Acta 1252 (1995) 43-52

ence of four carboxymethylcysteines, indicating the completeness of the reaction. This chemical modification promotes protein unfolding. In fact, the spectroscopic characterization of o~SRC shows typical features of an unfolded protein (Fig. 1). The UV-absorbance spectrum of a S R C is blue-shifted in comparison to that of the native protein, as expected for the protein chromophores in a more polar environment after unfolding. The absorbance maximum of native o~-sarcin appears at 278 nm, whereas it is located at 275nm in a S R C (Fig. 1A). The CD spectrum of a S R C in the peptide bond region (far-UV) (Fig. 1B) is consistent with a high contribution of random conformation (the minimum absorption band, 7r-~-* transition, is below 200nm) and a negligible a-helix content (the negative ellipticity at 220 nm, n-Tr * transition, is very low). Analysis of this CD spectrum reveals the presence of some residual secondary structure in c~SRC, turns and /3-strands (Fig. 1B). The observed loss of the characteristic protein bands in the near-UV CD spectrum [26] upon protein modification is in agreement with the disruption of the native tertiary structure (data not shown). The fluorescence emission spectrum for excitation at 275nm shows two maxima centered at about 305 nm and 345 (Fig. I C). The tryptophan contribution (two residues in the protein) to this spectrum has been deduced by obtaining the emission spectrum for excitation at 295 nm, a wavelength where the absorption of tyrosine is less than 10% of that measured at 275 nm. The tyrosine contribution (8 residues) is deduced from the comparison of the emission spectra for excitation at 275nm (Tyr + Trp; spectrum 1 in Fig. 1C) and 295nm after normalization (Trp; spectrum 2 in Fig. 1C). The tyrosine emission band is centered at 305 nm, and a single emission band of tryptophan appears at 345 nm. The fluorescence emission of both fluorophores in ceSRC is greatly

210

220

modified in comparison with that in native oz-sarcin (Fig. I C, spectra 4 and 5). The chemical modification produces an about 3-fold increase in the fluorescence emission of tyrosine with no shift in the position of the emission maximum. No tyrosine to tryptophan energy transfer was observed in native a-sarcin [26]. Therefore, the increase in the tyrosine fluorescence emission of o~SRC cannot be attributed to an interruption of such a transfer. This tyrosine fluorescence emission increase in a SRC should be related to a decreased quenching by adjacent groups resulting from a conformational change arising from the chemical modification. A similar increase in the tyrosine fluorescence emission was also observed for the acid pH-induced unfolding in a-sarcin [26]. Regarding the tryptophan emission, two maxima at 325 and 335nm are observed in native oz-sarcin (Fig. IC, spectrum 4), which were explained in terms of different environments for the two Trp residues of c~-sarcin [26]. A red-shift in the fluorescence emission of protein fluorophores is expected to occur upon protein unfolding. The shielding of the Trp residues from water by the protein matrix produces a blue-shift of the fluorescence emission in comparison to that of the free amino acids, and this is lost upon unfolding. A single and red-shifted emission band of tryptophan is observed in otSRC. This indicates that the two Trp residues are more exposed in o~SRC, in agreement with the protein unfolding. The tryptophan emission in o~SRC is about 2-3-fold increased in comparison with that in native o~-sarcin. This may indicate that the Trp emission was quenched by some protein group in the native conformation, and the quenching decreases upon the conformational change promoted by the chemical modification. An increase of the tryptophan fluorescence emission of about the same magnitude was also observed in a conformational transition in native

230

240 nm

A

0.4

2.0

S

0.3

!

1

/

-3

9

-6

~0.2

1.0

x

,

I I

-9

o

I

0.1

I

-12

270

290

310 nm

310

330

350

370

390 nm

Fig. 1. Spectroscopic characterization of aSRC. (A) Absorbance (A, recorded in 1 cm optical-path ceils) spectra of native a-sarcin ( - - ) and a S R C (---). (B) Far-UV circular dichroism spectra of native a-sarcin ( - - ) , and a S R C (---). Ellipticity values are given in units of deg. cm 2 dmol-] of amino-acid residue. ( . ) Theoretical values corresponding to the secondary structure percentages (oz-sarcin, 21% a-helix, 33% antiparallel /3-sheet, 18% /3-turn, 28% random; aSRC, 21% antiparallel /g-sheet, 17% /3-turn, 62% random) calculated from the CD spectra according to [41]. (C) Fluorescence emission (F, expressed in arbitrary units) spectra of aSRC: spectrum 1, for excitation at 275 nm; spectrum 2, for excitation at 295 nm and normalized to the former spectrum at wavelengths above 380nm; spectrum 3, calculated difference spectrum, (1) minus (2); spectra 4 and 5, fluorescence emission spectra corresponding to the tryptophan and tyrosine contributions, respectively, in native o~-sarcin at the same protein concentration than a SRC. All the samples were dissolved in 50 mM Mops buffer (pH 7.0) containing 0.1 M NaCI.

M. Gasset et al. / Biochimica et Biophysica Acta 1252 (1995) 43-52

0.4

A

03 o~ 02 < <1 0.1

I

t

I

1

2

3

I 4

.~-D#''IO 2

60 II¢I

"40 a~

20 I

I

i

I

1

2

3

4

p / L xlO 2

Fig. 2. Aggregation and lipid-mixing of vesicles promoted by otSRC. (A) Increase in the absorbance values at 360nm (AA360) produced by interaction of DMPG vesicles (30nmol) and a S R C at different ( P / L ) protein/lipid molar ratios. Results represent the average of three different experiments. (B) Effect of a S R C on energy transfer efficiency (%RET) between NBD-PE (donor) and Ria-PE (acceptor) incorporated into DMPG vesicles (1.0:0.6:98.4, molar ratto; 9:1, unlabelled/labelled vesicles). A constant total phospholipid concentration of 150/zM and different a S R C / l i p i d ( P / L ) molar ratios were employed in these measurements. The (%RET) values (average of three determinations) were measured as ( % R E T ) = [ I - ( F / F o ) ] × I O 0 , where F and F 0 are the fluorescence emission intensities at 530nm in the presence and in the absence of Rh-PE, respectively. Experiments were performed in 50mM Mops buffer (pH 7.0) containing 0.1 M NaC1 and 1 mM EDTA, at 37°C.

o~-sarcin, centered at pH 8.0 and attributed to the deprotonation of the oz-NH 2 group of the protein [26]. Finally, these spectroscopic characteristics of a SRC are not modified upon incubation of the protein at either 4°C or 37°C for at least one week, what indicates that spontaneous refolding does not occur.

3.2. Interaction of reduced and carboxyamidomethylated a-sarcin with acidic phospholipid vesicles Addition of o~SRC to a preparation of acid phospholipid vesicles produces a large increase in the turbidity of the lipid dispersion, due to the significant light-scattering of the resulting complexes. The time-course of this process was studied by measuring the absorbance variation at 360 nm (AA360). A time-dependent increase in the (AA360) value is observed when a SRC is added to DMPG vesicles (data not shown). About 30min are required for completion of the process at the lowest protein concentration studied. A plot of the maximum absorbance variation

47

versus protein concentration is given in Fig. 2A. The analyzed effect becomes saturated at about 0.04 protein/phospholipid molar ratio. Native c~-sarcin produces a similar effect, and under identical experimental conditions exhibits a 0.02:1 saturating protein/phospholipid molar ratio [7]. At the saturating ratio, the extent of the absorbance variation produced by o~SRC is about 1.5-fold higher than that promoted by native o~-sarcin. No absorbance variation is observed when phosphatidylcholine vesicles are used, as also described for the native protein. This effect of c~SRC on acidic vesicles must involve electrostatic interactions inherent to the vesicle aggregation revealed by this analysis. But a SRC also destabilizes the bilayers. In fact, c~SRC promotes lipid-mixing between membranes of DMPG, as detected by a classical resonance energy transfer (RET) assay [23] (Fig. 2B). A decreased RET efficiency between the donor NBD-PE and the acceptor Rh-PE (from 77% to about 25% RET) is observed depending on the amount of a S R C present. The variation is saturated at about 0.02:1 protein/phospholipid molar ratio. The maximum extent of the decrease in RET efficiency and the saturating protein/lipid molar ratio are almost coincident to those obtained for native a-sarcin [8]. When Fig. 2A and 2B are compared in terms of protein/phospholipid saturating molar ratio, it can be concluded that aggregation of mixed bilayers occurs, since AA360 still increases at protein concentration higher than that corresponding to the maximum lipid-mixing observed. c~SRC also produces leakage of the aqueous contents in egg-phosphatidylglycerol vesicles. Egg-PG vesicles were used instead of DMPG due to their higher encapsulation efficiency [8]. ANTS has been found to be a reliable and convenient fluorophore to continuously monitor leakage from the vesicle interior [27]. DPX is highly water-soluble and efficiently quenches the ANTS fluorescence when coencapsulated with this fluorophore. Dilution of ANTS and DPX in the external medium, as consequence of the vesicle leakage, prevents the quenching of the ANTS fluorescence. When both, ANTS and DPX are coencapsulated in lipid vesicles, the rate of dequenching is only dependent on the rate of dilution of ANTS and DPX into the external medium [27]. The kinetics of the leakage process is analyzed by measuring the ANTS fluorescence variation. The %leakage is dependent on the c~SRC concentration and the process takes about 10min for completion at the lowest ct SRC concentration studied. The extent of leakage increases and reaches a plateau as the c~SRC concentration increases; the maximum %leakage produced is about 85%, in comparison to that produced by 1% Triton X-100 considered as 100% (lysed vesicles). The initial rate of the process increases almost linearly in the protein concentration range considered (Fig. 3A). The leakage of the aqueous contents can be concomitant with or following mixing of bilayer lipids, depending on the resulting structures. Therefore, it is of interest to

48

M. Gasset et al. / Biochimica et Biophysica Acta 1252 (1995) 43-52

%L

%L/s

A

80

O.8

80

G6

40

0.4

20

O.2 t

0.75 %

I

I

I

1 . 5 0 2.25

3D0

rl-'~l'xl02

B

80

a

60

b

1.4~

40

1.oO

20 I

20

I

40

20

60

I

I

60

80

s

aa~

$

Fig. 3. Leakage of aqueous content of lipid vesicles induced by a SRC. (A), Leakage of the aqueous contents of egg PG vesicles induced by o~SRC, monitored as fluorescence emission dequenching of ANTS coencapsulated with DPX. ( • ) Extent of the leakage, expressed as percent leakage [(%)L], and ( • ) initial rate of leakage, expressed as [(%)L]/s, vs. protein/lipid molar ratio (P/L). The [(%)L] has been calculated from the fluorescence variation at 510 nm at 10 min after addition of o~SRC; values are referred to the fluorescence variation of vesicles disrupted with (1% final concentration) Triton X-100, considered as 100%. The excitation wavelength was 386nm. Vesicles (0.1 m g / m l ) were dissolved in 10mM Tris-HC1 buffer (pH 7.5) containing 0.1M NaC1 and l m M EDTA. (B) Kinetics of the (a) lipid-mixing and (b) leakage of aqueous content of egg PG vesicles (0.1 m g / m l ) induced by a S R C (5nmol/ml). Values are expressed as percentages of the maximum fluorescence variation in either assay, RET (for lipid-mixing) and dequenching of ANTS (for leakage) (see Section 2). The inset shows a plot of the chart tracings in (B) according to the equation I n [ 1 0 0 / ( 1 0 0 - % ) ] vs. time (s), (%) being the values from (B).

examine the correspondence between lipid-mixing and leakage as a function of the time. A delay between both processes is expected when non-leaky structures are initially formed upon the protein-vesicle interaction. No delay is observed when both lipid-mixing and leakage of the aqueous content induced by a SRC are kinetically compared (Fig. 3B). When these chart tracings are analyzed by plotting I n [ 1 0 0 / ( 1 0 0 - % ) ] vs. time (Fig. 3B, inset), the lipid-mixing could be ascribed to a first-order process (Fig. 3B, inset plot a) as it would be expected if the aggregation of vesicles is not the rate-limiting step.

This assay has been performed at a higher protein/lipid molar ratio than that required for saturation of the lipidmixing, and therefore aggregation of vesicles is not expected to be rate-limiting. A similar analysis has been performed for the leakage kinetics (Fig. 3B, inset plot b). The slope of the plot decreases what would indicate that the rate constant of the process decreases as the process proceeds (Fig. 3B, inset). According to this, it seems that the lipid-mixing promoted by ceSRC in these vesicles is a leaky process in the first stages, but once the lipid-mixing affects to many bilayers the leakage occurs through a slower process than the lipid-mixing (Fig. 3B, inset). The analysis of the effect of a protein on the thermotropic behaviour of vesicles renders information about the potential nature of the protein-vesicle interaction. The involvement of a hydrophobic component in the interaction can be deduced from these studies, a SRC modifies the thermotropic behaviour of acid phospholipid vesicles but does not produce any effect on phosphatidylcholine vesicles. This has been studied by measuring the fluorescence anisotropy of DPH-labelled vesicles. DPH is a fluorescence probe for the hydrophobic core of lipid bilayers. The anisotropy of the DPH fluorescence emission decreases upon the gel to liquid-crystalline transition of phospholipid bilayers due to the concomitant increase of the membrane fluidity. Therefore, the phospholipid phase transition can be monitored by measuring the DPH anisotropy. Both transition temperature and amplitude can be thus analyzed, a SRC promotes a decrease of the amplitude of the phospholipid phase transition (Fig. 4A). The protein tends to abolish the phase transition, but the transition temperature is not greatly affected, o~SRC concentrations higher than that corresponding to 10:1 phospholipid to protein molar ratio does not modify this latter anisotropy profile. These results suggest that ceSRC induces a disorganization of the acyl chains and changes in their mobility, which is in agreement with a penetration of the polypeptide into the bilayers. The effect of a SRC on the thermotropic behaviour of phospholipid vesicles was also analysed by DSC. The results obtained for DMPG vesicles and a SRC mixtures at different lipid to protein molar ratios are given in Fig. 4B. The protein causes a decrease in the enthalpy change (area under the heat capacity curve) for the phase transition of the phospholipid. A broadening in the transition is also observed as the protein proportion increases, in a similar way as observed from the measurements performed with DPH-labelled vesicles. These results suggest an important disorganization of the phospholipid bilayer upon interaction with c~SRC, many phospholipid molecules being removed from participating in the gel-to-liquid lipid phase transition. An analysis of these data in terms of the equation A H / A H o = 1 - Na(P/L), where AH 0 is the enthalpy change in the absence of protein and ( P / L ) is the proteinto-lipid molar ratio, would give a value of Na = 60 for the mean number of phospholipid molecules affected by

M. Gasset et al. / Biochimica et Biophysica Acta 1252 (1995) 43-52

r

15

20

25

3O

I

I

I

I

35 "C I

A

0.35 0.30

tion, a SRC is degraded based on the results of the electrophoretic analysis of the reaction mixture (data not shown)• This indicates that a SRC gain access to the vesicle-entrapped trypsin. This was also observed for the native o~-sarcin [10].

3.3. Changes on the enuironment of tryptophan residues of aSRC upon interaction with phospholipid vesicles

0.25 0.20

It is well known that the fluorescence emission of the indole moiety of tryptophan is very sensitive to the dielectric constant of its environment. In this regard, the fluorescence emission spectrum of a SRC is greatly modified upon interaction with phospholipid vesicles. Fig. 5 shows a summary of the results obtained for DMPG vesicles. The interaction of c~SRC with the vesicles results in a blue-shift (Fig. 5, inset) in the position of the emission maximum wavelength (from 345nm to about 330nm) with a concomitant increase in the fluorescence emission intensity (about 3-fold) (Fig. 5). These changes would become saturated at about 125:1 lipid to protein molar ratio. The increased quantum yield would imply a greater shielding from the polarizable groups that are usually responsible for tryptophan quenching. These observations are interpreted in terms of a transfer of the indole group of tryptophan to a more hydrophobic environment upon interaction with the lipid vesicles. This is confirmed by the results obtained

o.15 O.lO 0.05

49

1

I

I

I

1

I

O1-1

AC,

I

20

I

25

i

30

"C

Fig. 4. Effect of a S R C on the phase transition of phospholipid vesicles. (A) Steady-state anisotropy ( r ) variation of DPH-labelled DMPG vesicles upon interaction with o~SRC at: (4), 10:1; (3), 50:1; (2), 100:1 lipid/protein molar ratios. (1) DMPG vesicles in the absence of protein. These studies were performed in 50 mM Mops buffer (pH 7.0) containing 0.1 M NaCI and 1 mM EDTA. The phospholipid concentration employed was 80/xg/ml, containing h l000 DPH/lipid weigth ratio. Anisotropy values were determined after equilibration of the samples for 10min at each temperature. Values are expressed as the average of three different determinations. (B) Effect of a S R C on the DSC heating profiles of DMPG vesicles. Thermograms correspond to: (a) DMPG; (b to e) 400:1, 200:1, 100:1 and 50:1 lipid/c~SRC, respectively. Samples were prepared in 50mM Mops buffer (pH 7.0) containing 0.1M NaCI and I mM EDTA. The lipid concentration was 0 . 3 m g / m l . Samples were subjected to a previous heating/cooling cycle, and the thermograms shown were obtained under heating conditions at 3 0 ° C / h scanning speed. No differences were observed after repeated heating/cooling cycles.

otSRC. Similar results have been also obtained for DMPS vesicles (data not shown). o~SRC has been incubated with trypsin-containing vesicles in the presence of a large excess of external hen egg-white trypsin inhibitor, enough to completely inhibit the total amount of proteinase present in the assay. This experiment has been previously performed for native o~sarcin under identical conditions [10]. After 4 h of incuba-

i 4.61

•

•

4.0 3.5

Rift

340 ~~.~.

3.0

2.0

•0

2

o.s

1.0

1.5

2.0

2.s

L/P xl()2

3.0

Fig. 5. Effect of DMPG vesicles on the fluorescence emission properties of otSRC. ( • ) Fluorescence emission intensity at 330nm, for excitation at 275nm, expressed in arbitrary units (but the same than in Fig. 1C); ( • ) wavelength of the emission maximum (nm). Values are given for different ( L / P ) lipid to protein molar ratios (3/xM constant protein concentration; 50/xg/ml), and are the average of three different determinations. The inset shows the fluorescence emission spectra of (a) otSRC and (b) DMPG/otSRC vesicles at 150:1 lipid to protein molar ratio. Fluorescence emission is expressed in the same arbitrary units as above. Samples were dissolved in 50mM Mops buffer (pH 7.0) containing 0.1 M NaC1 and 1 mM EDTA. Experiments were performed by maintaining a constant otSRC concentration. Spectra were obtained by recording horizontally polarized emission intensities for vertically polarized excitation beam to avoid light-scattering interferences.

50

M. Gasset et al. / Biochimica et Biophysica Acta 1252 (1995) 43-52

A N T / D M P G xlO 2 2 3 4

1 A

'

I

I

I

%

D M P G vesicles [31], and the fluorescence of tryptophan residues was also quenched by anthracene [32] in a similar extent to that observed in o~SRC.

3.4. Infrared spectroscopy of aSRC interacting with lipid bilayers 20 10

% 30 O

20 10 •

I

2

I

I

I

4 6 8 ANT/Protein

Fig. 6. Quenching of the a SRC tryptophan fluorescence by anthracene incorporated into DMPG vesicles. The effect of (ANT) anthracene is evaluated by considering an apparent energy transfer efficiency (%), defined by the equation (%)= [1- (F/F0)] × 100, where F and F0 are the fluorescence emission intensities at 340 nm on excitation at 295 nm in the presence and in the absence of anthracene respectively. (A): o~SRC was added to the anthracene-labelled vesicles at different ANT/DMPG molar ratios, and incubated for I h at 37°C prior to the measurements. The experiments have been performed at a constant aSRC concentration of 1.47. l0 6 M, and two lipid/protein molar ratios, ( * ) 188:1 and ( . ) 70:1, have been considered. (B) The data in (A) are plotted as (%) apparent energy transfer efficiency vs. anthracene to aSRC molar ratio.

from measurements of quenching of the protein fluorescence by anthracene incorporated in the lipid vesicles. Anthracene may safely be expected to be located in the acyl chain region of the bilayer, based on its hydrophobicity [28]. The fluorescence emission spectrum of the indole group of tryptophan overlapped the absorption spectrum of the anthryl group, thus allowing energy transfer measurements [28]. Anthracene incorporated in D M P G vesicles quenches the tryptophan emission at 340nm of c~SRC, and the intensity of the anthracene fluorescence emission peaks around 4 0 0 n m is concomitantly increased. The resuits obtained are shown in Fig. 6. An increase in the apparent energy transfer is observed upon increasing the acceptor concentration, the process being saturated depending on both D M P G / p r o t e i n and anthracene/protein concentration ratios (Fig. 6A and B). The extent of the energy transfer to anthracene is similar to that observed for cardiotoxin II [29] and mellitin [30] or the blocked dipeptide Boc-Trp-Phe-OET [28], when these polypeptides are partitioned into anthracene-labelled liposomes. Native c~-sarcin also shows an increased fluorescence quantum yield (about 2-fold) of both Tyr and Trp residues upon interaction with

The effect of the lipid vesicles on the secondary structure of a S R C has been examined by using infrared spectroscopy, since sample light-scattering hampers the circular dichroism spectra. The usefulness of infrared spectroscopy to probe secondary structure of proteins in membranes is well recognized [33]. Conformational sensitive amide r band region of the infrared spectra after Fourier self-deconvolution of ceSRC in the presence and in the absence of bilayers are given in Fig. 7. Upon vesicle interaction, a SRC undergoes a conformational change according to the amide I' band modifications observed. Percentages of each type of secondary structure can be determined from the Gaussian peaks composing the FTIR spectrum of a protein. However, even the deconvolved spectrum of ceSRC is sufficiently featureless that curve fitting cannot be done reliably. Increasing resolution enhancement factor and decreasing the Lorentzian half-width band only results in spectrum asymmetrization but not in band differentiation. This has been also reported for denatured ribo-

I

I

I

I

I

I

I

I

I

B

I

I 1000

!

t 1860

I

I 1040

I

I

I

wivlnumber crn 1

Fig. 7. Secondary structure of a S R C in the presence of lipid vesicles. (A) Absorbance (A) FTIR spectra in the amide I' region, at 37°C, of a S R C both (1) in the absence and (2) in the presence of DMPG vesicles (150:1 lipid-to-protein molar ratio). Part B shows the difference spectra (absorbance difference AA; i.e., spectrum 2 minus spectrum 1). Fourier self-deconvolution was performed with a resolution enhancement factor (K value) of 2 and a Lorentzian half-width band of 30cm -] . Samples were prepared on the CaF 2 window by mixing 10/xl of a SRC (5.0 m g / m l in 50mM Mops buffer (pH 7.0) containing 0.1 M NaCI prepared in D20) and d-buffer or the required amount of DMPG in d-buffer. Corrections for the isotopic effect on the pH were not considered. The frequency limits for the different secondary structure elements were: 16621645cm - I , a-helix; 1689-1682cm -I and 1637-1613cm I, fl-sheet; 1644.5-1637cm - t , random; 1682-1662.5cm-t, /3-turns [25]. The absorbance scale is the same in both parts of the figure.

M. Gasset et al. / Biochimica et Biophysica Acta 1252 (1995) 43-52

nuclease A [34]. Therefore, we have calculated the difference spectrum by choo'.fing a S R C in the absence of vesicles as reference. The difference spectrum is also reported in Fig. 7. This allows a qualitative estimation of the conformational changes produced. Negative bands in this difference spectrum accounts for structure elements lost while positive bands correspond to gained elements. It is clear that the lipid vesicles promotes conformational changes at level of the secondary structure in a SRC. The protein gains a-helix and turns while some fl-structure is lost. This loss of fl-structure corroborates the presence of this arrangement in the residual secondary structure of a S R C as deduced fron~ the CD measurements. Native a-sarcin also increased its a-helix content upon interaction with the lipid vesicles, also according to FTIR measurements [31 ].

4. Discussion

The previous studies about the effect of a-sarcin on bilayers have revealed the involvement of both electrostatic and non-ionic inl:eractions. Neutralization of the charged head groups of the acid phospholipids by the positively charged protein is the most plausible explanation for part of the effects promoted by a-sarcin in bilayers. But the interaction of the protein at the level of the hydrophobic core of the bilayers is an intriguing question in a protein like a-sarcin, since it contains 24 basic and 17 acid residues and about 60% of the remaining residues have polar side chains, and only very short hydrophobic amino-acid stretches are present. The effects promoted by a protein must be related to its either native or ligand-induced structure. Thus, the regions of a-sarcin involved in hydrophobic interactions would be either present in the native conformation or result from the bilayer-induced conformation. The observation of a conformational change in a-sarcin promoted b'.¢ acid bilayers [31] does not exclude any of the abow~ two possibilities. An unfolded protein form would render valuable information regarding to this. Thermally denatured a-sarcin partially refolds [26] and consequently is not a reliable model for these purposes. The use of denaturing agents must be discarded because they can also affect the bilayers. Simple reduction of the disulfide bridges of a-sarcin was also discarded to avoid any refolding of the polypeptide due to potential reoxidation reactions. Thus, we have prepared reduced and carboxyamidomethylated a-sarcin. Iodoacetamide was selected as a sulfhydryl blocking group instead of iodoacetate, since this reagent introduces four negative charges in the modified protein. The spectroscopic characterization of a SRC shows shifts of the tryptopharL absorption and emission bands, which are typical features of an unfolded polypeptide. This is corroborated by the near-UV and far-UV circular dichro-

51

ism measurements. However, a SRC retains some elements of residual secondary structure, turns and /3-strands according to the CD analysis. This is not surprising since denatured states of proteins frequently exhibit significant amounts of persistent secondary structure and a significant fraction is suggested to be native-like [35,36]. c~SRC interacts with acidic phospholipid vesicles, as it occurs with native a-sarcin. Thus, a SRC promotes vesicle aggregation of acidic vesicles. This could be expected because vesicle aggregation is mainly related to electrostatic interactions and the net charge of the modifed protein is not sensibly different to that of native a-sarcin. This is an important observation which indicates that the native conformation of a-sarcin is not required for the electrostatic interaction of the protein with bilayers. In addition, a SRC and native a-sarcin do not interact with PC vesicles. Therefore, although the spectroscopic characterization of a SRC shows a more polar environment for the aromatic chromophores of the protein, which might be related to the presence of solvent-exposed hydrophobic residues in a SRC, these are not the driving force for the interaction with bilayers. Electrostatic interactions are first required for the interaction of a SRC with membranes, thus explaining the absence of interaction with PC vesicles. But, a SRC also destabilizes the bilayers, i.e. lipid-mixing, leakage of intravesicular aqueous contents, and a large reduction of the amplitude of the phospholipid phase transition (detected by fluorescence anisotropy and differential scanning calorimetry measurements), and parts of the molecule are transferred to the hydrophobic core of the bilayer as the energy transfer of tryptophan to anthracene demonstrates. In addition, c~SRC gain access to vesicle-entrapped trypsin as described for the native protein. Thus, a SRC and native a-sarcin hydrophobically interact with bilayers although they have different conformation. In this regard, a-sarcin exhibit some amino-acid sequence similarity with ribonucleases of the RNase T1 subfamily [37] whose three-dimensional structure is well known. Based on this, we have proposed a model for the structure of a-sarcin [37] in which a /3-sheet composed of five strands is the central part of the molecule. This /3-sheet is hydrophobic in a-sarcin and has been tentatively proposed to be involved in hydrophobic interactions with bilayers [37]. Although unfolded, a SRC retains residual secondary structure, 20% r-structure is deduced from the far-UV CD spectrum and the FTIR analysis corroborates the presence of this peptide bond secondary arrangement. The residual secondary structure of a SRC may be related to the proposed /3-sheet hydrophobic locus of a-sarcin, thus explaining the hydrophobic interaction of both protein forms, native and a SRC, with bilayers. In opposition to this reasoning, the interaction of a SRC with bilayers may be considered as a non-specific effect of a denatured protein. However, a SRC is highly water-soluble, as the native protein, does not tend to aggregate and, moreover, a hydrophobic domain in a SRC is so difficult to find as in

52

M. Gasset et al. / Biochimica et Biophysica Acta 1252 (1995) 43-52

native o~-sarcin. In this sense, it has been shown [38-40] that several water-soluble proteins, unfolded at low pH, expose a hydrophobic domain, which enable them to destabilize membranes but they avidly aggregate in solution. In summary, the obtained results indicate that the native conformation of ~-sarcin is not required for the electrostatic interaction of the protein with bilayers. The secondary structure elements presents in o~SRC (turns and /?-strands) are enough to promote destabilization of the hydrophobic core of the bilayer. The conformational changes produced by the lipids affect to these structural elements since the gained o~-helix is accompanied by a decreased /?-structure. This would be in agreement with the involvement of the potential /?-structure of ce-sarcin in the hydrophobic interactions of this cytotoxin with membranes.

Acknowledgements We thank Prof. Jos~ M. Gonzfilez-Ros (Department of Neurochemistry and Institute of Neurosciences, University of Alicante, Spain) for providing access to the infrared spectrometer. We are also indebted to Prof. Erik Goormaghtigh (Universit6 Libre de Bruxelles, Belgium) for providing the software package of the FTIR analysis and critical reading of the manuscript and valuable suggestions. This work has been supported by Grants PB90/0007 and PB93/0090 from the DGICYT (Spain).

References [1] Arruda, L.K., Mann, B.J. and Chapman, M.D. (1992) J. lmmunol. 149, 3354-3359. [2] Olson, B.H., Jennings, J.C., Roga, V., Junek, A.J. and Schuurmans, D.M. (1965)Appl. Microbiol. 13, 322-326. [3] Wool, I.G. (1984) Trends Biocbem. Sci. 9, 14-17. [4] Turnay, J., Olmo, N., Jim6nez, A., Lizarbe, M.A. and Gavilanes, J.G. (1993) Mol. Cell. Biochem. 122, 39-47. [5] Wawrzynczak, E.J., Henry, R.V., Cumber, A.J., Parnell, G.D., Dernyshire, E.J. and Ulbrich, N. (1991) Eur. J. Biochem. 196, 203-209. [6] Wool, I,G., Endo, Y., Chang, Y.L. and Gliick, A. (1990) in The Ribosome: Structure, Function and Evolution (Dahler, A., Garret, R.A., Moore, P.B., Schlessinger, D. and Warner, J.R., eds.), pp. 203-214, American Society of Microbiology, Washington, DC. [7] Gasset, M., Martfnez del Pozo, A., Ofiaderra, M. and Gavilanes, J.G. (1989) Biochem. J. 258, 569-575. [8] Gasset, M., Ofiaderra, M., Thomas, P.G. and Gavilanes, J.G. (1990) Biochem. J. 265, 815-822. [9] Manchefio, J.M., Gasset, M., Lacadena, J., Ram6n, F., Marffnez del Pozo, A, Ofiaderra, M. and Gavilanes, J.G. (1994) Biophys. J. 67,

1117-1125. [10] Ofiaderra, M., Manchefio, J.M., Gasset, M., Lacadena, J., Schiavo, G., Martlnez del Pozo, A. and Gavilanes, J.G. (1993) Biochem. J. 295, 221-225.

[11] Sacco, G., Drickamer, K. and Wool, I.G. (1983)J. Biol. Chem. 258, 5811-5818. [12] Gasset, M., Ofiaderra, M., Marffnez del Pozo, A., Schiavo, G., Laynez, J., Usobiaga, P. and Gavilanes, J.G. (1991) Biochim. Biophys. Acta 1068, 9-16. [13] Montecucco, C., Papini, E. and Schiavo, G. (1991) in Sourcebook of Bacterial Protein Toxins (Alouf, J.E. and Freer, J.H., eds.), pp. 45-56, Academic Press, London. [14] Parker, M.W. and Pattus, F. (1993) Trends Biochem. Sci. 18, 391-395. [15] London, E. (1992) Biochim. Biophys. Acta 1113, 25-51. [16] Chen, W.J. and Douglas, M.G. (1987) J. Biol. Chem. 262, 1560515609. [17] Kumamoto, C.A. (1991) Mol. Microbiol. 5, 19-22. [18] Gavilanes, J.G., Vfizquez, D,, Soriano, F. and M~ndez, E. (1983) J. Protein Chem. 2, 251-261. [19] Lamy, B. and Davies, J. (1991) Nucleic Acid Res. 19, 1001-1006. [20] Lacadena, J., Marffnez del Pozo, A., Barbero, J.L., Manchefio, J.M., Gasset, M., Ofiaderra, M., L6pez-Otln, C., Ortega, S., Garc~a, J. and Gavilanes, J.G. (1994) Gene 142, 147-151. [21] Bartlett, G.R. (1959) J. Biol. Chem. 234, 466-468. [22] Ellens, M., Bentz, J. and Szoka, F.C. (1984) Biochemistry 23, 1532-1538. [23] Struck, D., Hoekstra, D. and Pagano, R.G. (1981) Biochemistry 20, 4093-4099. [24] Gavilanes, J.G., Lizarbe, M.A., Municio, A.M. and Ofiaderra, M. (1985) Int. J. Pept. Protein Res. 26, 187-194. [25] Goormaghtigh, E., Cabiaux, V. and Ruysschaert, J.M. (1990) Eur. J. Biochem. 193, 409-420. [26] Marffnez del Pozo, A., Gasset, M., Ofiaderra, M. and Gavilanes, J.G. (1988) Biochim. Biophys. Acta 953, 280-288. [27] Smolarsky, M., Tietelbaum, D., Sela, M. and Gitler, C. (1977) J. Immunol. Methods 15, 255-265. [28] Uemura, A., Kimura, S. and Imanishi, Y. (1983) Biochim. Biophys. Acta 729, 28-34. [29] Batenburg, A.M., Boagis, P.E., Rochat, H., Verkleij, A.J. and De Kruijff, B. (1985) Biochemistry 24, 7101-7110. [30] Batenburg, A.M., Hibbeln, J.C.L., Verkleij, A.J. and De Kruijff, B. (1987) Biochim. Biophys. Acta 903, 155-165. [31] Gasset, M., Ofiaderra, M., Goormaghtigh, E. and Gavilanes, J.G. (1991) Bioehim. Biophys. Acta 1080, 51-58. [32] Manchefio, J.M., Gasset, M., Lacadena, J., Mart~nez del Pozo, A., Ofiaderra, M. and Gavilanes, J.G. (1994) in Biological Membranes: Structure, Biogenesis and Dynamics (Op den Kamp, A.F., ed.), pp. 269-276, Springer, Berlin. [33] Goormaghtigh, E. and Ruysschaert, J.M. (1991) in Molecular Description of Biological Components (Brasseur, R., ed.), Vol. 1, CRC Press, Boca Raton, FL. [34] Sosnick. T.R. and Trewhella, J. (1992) Biochemistry 31, 8329-8335. [35] Montelione, G.T. and Scheraga, H.A. (1989) Acc. Chem. Res. 22, 70-76. [36] Shortle, D. (1993) Curr. Opin. Struct. Biol. 3, 66-74. [37] Manchefio, J.M., Gasset, M., Lacadena, J., Mart~nez del Pozo, A., Ofiaderra, M. and Gavilanes, J.G. (1995) J. Theor. Biol. 172, 259-267. [38] Hong, K., Yoshimura, T. and Papahadjopoulos, D. (1985) FEBS Lett. 191, 17-23. [39] Kim, J. and Kim, H. (1986) Biochemistry 25, 7867-7874. [40] Harter, C., Biichi, T., Semenza, G. and Brunner, J. (1988) Biochemistry 27, 1856-1864. [41] Perczel, A., HollSsi, M., Tusnfidy, G. and Fasman, G.D. (1991) Protein Engng. 4, 669-679.