Vibrio cholerae hemolysin Implication of amphiphilicity and lipid-induced conformational change for its pore-forming activity Kausik Chattopadhyay 1 , Debasish Bhattacharyya 2 and Kalyan K. Banerjee 1 1 National Institute of Cholera and Enteric Diseases, Kolkata 700 010, India; 2 Indian Institute of Chemical Biology, Kolkata 700 032, India Vibrio cholerae hemolysin (HlyA), a water-soluble protein with a native monomeric relative molecular mass of 65 000, forms transmembrane pentameric channels in target bio- membranes. The HlyA binds to lipid vesicles nonspecifically and without saturation; however, self-assembly is triggered specifically by cholesterol. Here we show that the HlyA partitioned quantitatively to amphiphilic media irrespective of their compositions, indicating that the toxin had an amphiphilic surface. Asialofetuin, a b1-galactosyl-termin- ated glycoprotein, which binds specifically to the HlyA in a lectin-glycoprotein type of interaction and inhibits carbo- hydrate-independent interaction of the toxin with lipid, reduced effective amphiphilicity of the toxin significantly. Resistance of the HlyA to proteases together with the tryp- tophan fluorescence emission spectrum suggested a compact structure for the toxin. Fluorescence energy transfer from the HlyA to dansyl-phosphatidylethanolamine required the presence of cholesterol in the lipid bilayer and was synchronous with oligomerization. Phospholipid bilayer without cholesterol caused a partial unfolding of the HlyA monomer as indicated by the transfer of tryptophan residues from the nonpolar core of the protein to a more polar region. These observations suggested: (a) partitioning of the HlyA to lipid vesicles is driven by the tendency of the amphiphilic toxin to reduce energetically unfavorable contacts with water and is not affected significantly by the composition of the vesicles; and (b) partial unfolding of the HlyA at the lipid– water interface precedes and promotes cholesterol-induced oligomerization to an insertion-competent configuration. Keywords: pore-forming toxin; amphiphilicity; lipid- induced conformational change; oligomerization; protein fluorescence. Vibrio cholerae hemolysin (HlyA), a water-soluble cytolytic protein expressed by many V. cholerae El Tor O1 and non- O1 strains [1,2], belongs to a large, heterologous family of pore-forming toxins (PFT) [3,4] that are ubiquitous in prokaryotic and eukaryotic organisms. The toxin has been cloned and sequenced [5,6]. The HlyA permeabilizes a wide spectrum of eukaryotic cells including human and rabbit erythrocytes [2] and synthetic lipid vesicles [7,8] by forming transmembrane pentameric [9] diffusion channels with a diameter of approximately 1.5 nm. In addition to binding specifically to cholesterol [9], the toxin shows a lectin-like property in interacting with b1-galactosyl-terminated gly- coconjugates such as asialofetuin and asialothyroglobulin [10]. The purified toxin evokes secretion of fluid in a rabbit ligated ileal loop, suggesting its involvement in pathogenesis of cholera [11]. A consensus on the pathway of induction of membrane damage by PFTs postulates a sequence of at least three discrete biochemical events: binding of the toxin monomer to a cell surface receptor, self-assembly to an amphiphilic prepore oligomer and insertion in the lipid bilayer gener- ating a functional pore that mediates passive flux of molecules across the membrane [12–15]. Extensive studies of the interaction of V. cholerae HlyA with synthetic lipid vesicles suggest that the binding is nonspecific and nonsa- turable [8,9]. However, permeabilization of the target lipid vesicle and more precisely oligomerization of the toxin monomer to a pentameric channel shows a specific requirement for cholesterol [9,16] and is augmented dramatically by inclusion of ceramides in the lipid bilayer [9]. Recent studies indicate that sphingolipids and glycero- lipids with cone-shaped structures modify the energetic state of membrane cholesterol, which in turn promotes functionally productive interaction of the sterol with the toxin [17]. Earlier, we reported inhibition of hemolysis of rabbit erythrocytes by glycoproteins with b1-galactosyl moieties [10]. However, the sensitivity of erythrocytes to the HlyA is not correlated with the surface density of galactose. Even more intriguing was the observation that these glycoproteins inhibit the carbohydrate-independent interac- tion of the toxin with immobilized phospholipid and phospholipid-cholesterol. As there is no information on the structure of the toxin and its physicochemical charac- teristics in solution and lipid bilayer, it is difficult to speculate a molecular interpretation of the toxin–membrane interaction and subsequent events. In this communication, we show that the HlyA is a compact protein with an amphiphilic surface. Partitioning of such a molecule to a lipid bilayer seems to be driven solely Correspondence to K. K. Banerjee, National Institute of Cholera and Enteric Diseases, Kolkata-700 010. Fax: + 91 33 350 5066, Tel.: + 91 33 350 1176, E-mail: banerjeekalyan@hotmail.com Abbreviations: ANS, 8-anilino-1-naphthalene sulphonic acid; HlyA, hemolysin; PFT, pore-forming toxins (Received 10 May 2002, revised 20 June 2002, accepted 25 July 2002) Eur. J. Biochem. 269, 4351–4358 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03137.x by its tendency to avoid energetically unfavorable contact with water and is relatively insensitive to the bilayer composition. In spite of its intrinsic amphiphilicity, the toxin monomer does not possess an insertion-competent configuration. Secondly, we demonstrate that self-assembly of the toxin monomer is a spontaneous event that could occur in water as well, but at a very slow rate compared to that observed in a membrane bilayer. Finally, we provide evidence for partial unfolding of the toxin at the lipid–water interface, a conformational change that precedes and is likely to facilitate cholesterol-induced self-assembly to a transmembrane channel. EXPERIMENTAL PROCEDURES Purification of V. cholerae HlyA and assay of hemolytic activity The HlyA was purified to homogeneity as described previously [10] with some modification. V. cholerae non- O1 strain V 2 , a clinical isolate kindly supplied by R. Sakazaki, Tokyo, Japan, was grown to early stationary phase (6 h) in brain heart infusion (BHI, Becton- Dickinson) broth at 37 °C with shaking. Bacteria were removed by centrifugation at 30 000 g for 10 min at 4 °C and the culture supernatant (6 L) was concentrated approximately 100-fold by ultrafiltration through PM-10 (Amicon) membrane. Lipid-protein vesicles released by bacteria during growth were removed from the ultrafiltrate by size-exclusion chromatography on Sepharose CL-4B (Pharmacia, 50 · 2.5 cm) equilibrated with 25 m M sodium phosphate buffer containing 1 m M EDTA and 3 m M NaN 3 , pH 7.2 (Buffer A). The hemolytic activity eluted as a totally included fraction and was subjected to hydrophobic interaction chromatography on phenyl- Sepharose CL-4B (Pharmacia, 50 · 1.5 cm) equilibrated with buffer A. The HlyA bound tightly to the matrix and desorbed from the column at 46% ethylene glycol on application of a linear 0–80% ethylene glycol gradient, with the mixing chamber containing 200 mL of buffer A. Finally, the HlyA was separated from traces of proteases and low relative molecular mass contaminants by chro- matofocusing on PBE-94 (Pharmacia; 20 · 1.5 cm). The toxin bound weakly to the matrix and eluted as a slightly retarded symmetrical peak during washing of the column with buffer A. The purified toxin migrated in SDS- polyacrylamide gel electrophoresis (SDS/PAGE) [18] fol- lowing incubation in 1% SDS at 50 °C for 15 min, as a single 65 kDa polypeptide and was homogeneous in PAGE under nondenaturing condition at pH 8.3. The HlyA was stable indefinitely in 50% ethylene glycol at 4 °C. The protein concentration was determined spectro- photometrically at 280 nm based on an absorbance of 1.4 for a 1 mgÆmL )1 solution, determined by a modified Folin-Ciocalteu method [19]. The hemolytic activity was assayed by monitoring spectrophotometrically the release of hemoglobin at 541 nm or by measuring turbidity of the erythrocyte suspension at 600 nm. The purified HlyA caused 50% lysis of a 1% (v/v) suspension of rabbit erythrocytes in 10 m M sodium phosphate buffer contain- ing 150 m M NaCl, pH 7.2 (NaCl/P i ) for 1 h at 25 °Cata concentration of 4 ngÆmL )1 , corresponding to a specific activity of 60 p M . Demonstration of protein amphiphilicity Amphiphilicity was operationally defined as preference of a protein for an amphiphilic phase relative to water as measured by its equilibrium concentrations in the two phases. Because the chemical interaction of the HlyA with phase constituents might affect the distribution of the toxin in the two phases, a system consisting of water and an amphiphilic phase was constituted in two ways: (a) by using Triton X-114, a nonionic detergent that forms a single phase mixture with water at 4 °C and separates into a water-rich and detergent-rich phase at 23 °C [20], and (b) by using the interface of water and a low polarity organic solvent such as chloroform as an amphiphilic surface [21]. To monitor partitioning of the HlyA to Triton X-114, the toxin was dissolved in a 2% (v/v) solution of the detergent at 4 °C, the aqueous and detergent phases collected separately at 25 °C, precipitated with nine volumes of cold acetone and subjec- ted to SDS/PAGE for visualization of the toxin content. In the second system, the HlyA in the concentration range 25–100 lgÆmL )1 was mixed in duplicate with an equal volume of chloroform at 25 °C, vortexed for 10 min to maximize the area of water–chloroform interface and allowed to separate. The HlyA concentration in the aqueous phase was determined in triplicate by assay of the hemolytic activity as well as by the enzyme-linked immunosorbent assay (ELISA) using rabbit anti-(HlyA) IgG as described previously [10]. Preparation of liposome Phosphatidylcholine (PtdCho), phosphatidylethanolamine (PtdEtn) or a mixture (1 : 1 by weight) of PtdCho and cholesterol in chloroform/methanol (2 : 1 by volume) was evaporated to dryness, dispersed in buffer A and sonicated for 10 min at 23 kHz. Multilamellar vesicles were removed by centrifugation at 30 000 g. Large unilamellar liposomes were prepared from the pool of vesicles by size-exclusion chromatography on Sepharose CL-4B [22]. Protease treatment The HlyA was incubated with trypsin, chymotrypsin and Pronase (1 : 50, protease/toxin ratio, w/w) in NaCl/P i or NaCl/P i -2 M urea at 25 °C. The reaction was terminated by boiling in sample buffer containing 1% SDS for 5 min. Proteolytic digestion was monitored by SDS/PAGE [18]. Spectroscopic measurements Fluorescence and light scattering measurements were car- ried out with a Hitachi 4500 Spectrofluorimeter. Light scattering was recorded at a right angle to the incident beam in the wavelength range of 500–650 nm at slit widths of 2.5 nm. Because the intensity of the scattered light was relatedtothewavelength,k as 1/k 4 , the sensitivity of scattering was low in the long wavelength used; however, it eliminated interference by ultraviolet absorption of the protein. The intrinsic Trp fluorescence of the HlyA was monitored in buffer A by exciting the sample at 295 nm using band widths of 2.5 nm. Lipid-induced perturbation of the conformation of the HlyA was investigated by monitoring intrinsic Trp fluorescence of the toxin at a 4352 K. Chattopadhyay et al.(Eur. J. Biochem. 269) Ó FEBS 2002 protein : lipid ratio of 1 : 10 (w/w), a ratio at which the unbound toxin was not detected by gel filtration, using the excitation wavelength and slit widths as above. The fluorescence energy transfer from Trp in the toxin to dansyl- PtdEtn incorporated in PtdCho and PtdCho -cholesterol vesicles was followed kinetically by setting excitation and emission wavelengths at 280 and 512 nm, respectively, with 5 nm slit widths [22]. The quenching of the intrinsic Trp fluorescence by acrylamide was studied by using excitation wavelength of 295 nm, and the data were analyzed using the Stern–Volmer equation [23]. F 0 /F ¼ 1+K sv [Q] where F 0 and F refer to the fluorescence intensities in the absence and presence of acrylamide, respectively, K sv is the collisional quenching constant, and [Q] the concentration of the quencher. The binding of the dye, 8-anilino-1-naphthalene sul- phonic acid (ANS) to the HlyA monomer and oligomer was monitored spectrofluorimetrically by exciting at 390 nm a mixture of the dye and protein at concentrations of 50 l M and 90 lgÆmL )1 , respectively. The emission was recorded in the wavelength range of 410–600 nm at a slit width of 5 nm. RESULTS The HlyA is an amphiphilic protein Previously we showed that V. cholerae HlyA interacted strongly with phenyl-Sepharose CL-4B and desorbed from the hydrophobic matrix at an ethylene glycol concentration of 8.25 M [10] (see Experimental Procedure). Because surface hydrophobicity of a protein might drive it to a lipid bilayer, we explored the physicochemical nature of the HlyA surface in more details. The purified toxin was dissolved in 2% Triton X-114 at 4 °C. On separation of the solution into detergent- and water-rich phases by raising the temperature to 25 °C [20], the HlyA was found by SDS/ PAGE to be concentrated exclusively in the detergent phase (Fig. 1), a behavior that is commonly considered charac- teristic of an integral membrane protein and somewhat unusual for a water-soluble globular protein [24]. Water-soluble amphiphiles like detergents prefer the interface of water and air or a nonpolar organic solvent that enables them to attain an energetically favorable arrangement by orienting their polar and nonpolar ends to water and the less polar phase, respectively. In order to see if the partitioning of the HlyA to the detergent phase was dictated by intrinsic amphiphilicity of the protein and not by its affinity for Triton X-114 or a conformational change induced by the detergent, we monitored accumulation of the toxin at the interface of water and chloroform. The ratio of the aqueous phase concentration of the HlyA following partitioning with chloroform, c to the initial value, c o , estimated by the ELISA and also by the assay of hemolytic activity, was: 5.2 · 10 )3 (c o ¼ 100 lgÆmL )1 ), 7.6 · 10 )3 (c o ¼ 50 lgÆmL )1 )and1.2· 10 )2 (c o ¼ 25 lgÆmL )1 ) implying that approximately 99% of the toxin was actually present at the interface. For comparison, bovine serum albumin showed a c/c o value of 0.58 in an identical experiment at a protein concentration of 100 lgÆmL )1 . Interestingly, asialofetuin, a b1-galactosyl-terminated gly- coprotein [25] that binds to the HlyA with an affinity constant of 9.4 · 10 7 M )1 and inhibits its interaction with phospholipid vesicles [10] caused an approximately 10-fold increase in the bulk aqueous phase concentration of the toxin, as measured by a c/c o value of 8 · 10 )2 at HlyA and asialofetuin concentrations of 50 and 200 lgÆmL )1 ,respect- ively. These observations demonstrate: (a) the HlyA was a surface-active molecule that tended to be squeezed out of an aqueous phase; and (b) a carbohydrate ligand reduced significantly the effective amphiphilicity of the toxin, thereby rendering it energetically more compatible with an aqueous environment. Because amphiphiles form micellar aggregates to avoid exposure of the nonpolar surface to water, we examined the HlyA for self-aggregation by light scattering. The intensity of light scattered by 1 l M HlyA in water was significantly higher than that by a solution at the same concentration in 50% ethylene glycol (Fig. 2) indicating promotion of self- aggregation of the toxin by an aqueous environment. Size- exclusion chromatography of the HlyA on Biogel A-0.5 m, an agarose-based matrix with a fractionation range of 10–500 kDa led to partial exclusion of the toxin with the bulk of the material being spread out over the entire bed volume (data not shown). Use of dextran or polyacryl- amide-based matrices led to similar profiles. Such profiles Fig. 1. Distribution of the HlyA in water-Triton X-114 system. The purified toxin, dissolved in 2% (v/v) Triton X-114 in buffer A at 4 °C at a protein concentration of 100 lgÆmL )1 , partitioned into water- and detergent-rich phases at 25 °C. Aliquots of 100 lL were withdrawn from each phase, precipitated with nine volumes of cold acetone and subjectedtoSDS/PAGE. Lane 1, detergent-rich phase; lane 2, water- rich phase. Fig. 2. Dependence of the intensity of light scattered by HlyA on solvent polarity. Light scattering by the HlyA (65 lgÆmL )1 ) in buffer A (—) and 50% ethylene glycol in buffer A (– -) were recorded at slit widths of 2.5 nm. Ó FEBS 2002 Amphiphilicity and self-assembly of HlyA (Eur. J. Biochem. 269) 4353 could arise from a combination of self-aggregation of the toxin and nonspecific low affinity interaction with the matrix, both of which reflected the tendency of the protein to avoid exposure to water. The HlyA is a compact protein Having established the amphiphilic nature of the HlyA monomer, we investigated if the toxin possessed a compact solution structure characteristic of a native folded protein. The fluorescence emission spectrum of the HlyA excited at 295 nm to eliminate contribution by Tyr showed a maxi- mum at 330 nm (Fig. 3A). Exposure of the protein to 4 M guanidinium hydrochloride caused a red shift of the emission maximum to 345 nm suggesting transfer of Trp residues from a nonpolar environment in the native protein, to water in the unfolded state (data not shown). As there are 11 Trp residues at positions 31, 33, 113, 186, 210, 230, 318, 400, 402, 570 and 574 in the HlyA polypeptide [26] with possible differences in solvent exposure, we examined if the heterogeneity in microenvironments of the indolyl residues could be resolved by quenching of fluorescence by acryl- amide. Although acrylamide induced a large decrease in fluorescence intensity, there was no shift in emission maximum (Fig. 3B). The Stern–Volmer plot [23] of the relative fluorescence intensity vs. acrylamide concentration was perfectly linear over a fairly wide range of the quencher concentrations (Fig. 3C) with a K SV value of 2.2 indicating that the multiple indolyl groups behaved in essence as a single emitting center with restricted accessibility to water. These data, along with the large blue shift of the emission maximum of the native protein indicated that Trp residues were located in a nonpolar region that could be provided by the core of a compactly folded protein. Limited proteolysis of the HlyA supported a compact solution structure with restricted accessibility of peptide linkages to a water-soluble probe. There was no degradation of the toxin on incubation of the protein with trypsin (Fig. 4A) or chymotrypsin (Fig. 4B) at an enzyme : sub- strate ratio of 1 : 50. Repetition of the enzyme digestion in 2 M urea led to partial degradation of the HlyA to a 50-kDa fragment. The native toxin was, however, fairly susceptible to digestion by Pronase, a nonspecific protease (data not shown). Self-assembly of the HlyA and association with lipid bilayer Oligomerization of the HlyA, as distinct from amphiphili- city-driven nonstoichiometric self-aggregation in water described above, was monitored by stability of the pentamer in SDS at 60 °C, a characteristic it shares with several other PFTs, e.g. Staphylococcus aureus a-toxin [27]. In agreement with previous reports [9,16], self-assembly of the HlyA was essentially complete within 1 min of incubation with PtdCho-cholesterol vesicles (Fig. 5A) at 25 °Candvery slow in pure PtdCho vesicles (Fig. 5B). However, incuba- tion of the HlyA in a homogeneous dispersion of cholesterol in water at a toxin : sterol molar ratio of 1 : 60 for 20 min failed to induce detectable oligomerization (data not shown). Self-assembly of the HlyA was detected in water on storage of the toxin for several weeks but remained incomplete even after a year suggesting that oligomerization was essentially a spontaneous event that was accelerated dramatically by cholesterol in a lipid bilayer matrix. The near identity of the fluorescence emission spectra of the Fig. 3. Fluorescence emission spectra of the HlyA monomer and oligo- mers. (A) The HlyA monomer (– -), the oligomer formed during storage of the toxin in water for 30 days (— – —), and the oligomer generated in PtdCho-cholesterol vesicles and delipidated by treatment with sodium deoxycholate and size-exclusion chromatography on Sepharose CL-4B (15) (—) were dispersed in buffer A at a protein concentration of 90 lgÆmL )1 . Samples were excited at 295 nm and spectra recorded at slit widths of 2.5 nm. (B) Acrylamide-induced quenching of fluorescence of the HlyA monomer was monitored at a protein concentration of 65 lgÆmL )1 in buffer A (—) and 1 M acrylamide in buffer A (– -). Excitation wavelength and slit widths were set at 295 and 2.5 nm, respectively. (C) Stern–Volmer plot for quenching of intrinsic fluorescence of the HlyA monomer (d)and oligomer (D) by acrylamide. The protein concentration was kept constant at 65 lgÆmL )1 . Excitation and emission wavelengths were set at 295 and 330 nm, respectively, with slit widths of 2.5 nm. 4354 K. Chattopadhyay et al.(Eur. J. Biochem. 269) Ó FEBS 2002 HlyA oligomer formed in water and the one generated in PtdCho-cholesterol vesicles indicated a similarity in confor- mations of the two pentamers (Fig. 3A). Notably, oligo- merization led to no significant change in either Trp fluorescence spectrum or quenching of fluorescence by acrylamide, suggesting Trp residues sensed similar microen- vironments in the monomer and pentamer. However, the monomer and oligomer differed in binding to ANS, a fluorescent probe used widely to detect surface-exposed hydrophobic patches on proteins [28]. While there was no significant change in the ANS emission spectrum in the presence of the monomer, the oligomer induced a small twofold increase in intensity but a significant blue shift in the emission wavelength maximum by 21 nm (Fig. 6). As reported previously [10], incubation of the HlyA with PtdCho-cholesterol led to a total loss in hemolytic activity. In contrast, the HlyA bound to PtdCho remained hemo- lytically as potent as the free toxin, indicating that oligo- merization of the toxin signalled its irreversible association with the lipid bilayer. The difference in modes of association of the HlyA monomer and the oligomer with a lipid bilayer was reflected in the fluorescence energy transfer from Trp in the toxin to dansyl- PtdEtn incorporated in lipid bilayers. No increment in fluorescence intensity was observed during a 15 min incubation of pure PtdCho vesicles with the HlyA monomer (Fig. 7). In contrast, the increment in fluorescence intensity was almost immediate with PtdCho-cholesterol vesicles and essentially complete within 1 min. It appears therefore that the increment in fluorescence intensity Fig. 4. Protease-cleavage pattern of the HlyA. The toxin was incubated with the protease at an enzyme : substrate ratio of 1 : 50 (w/w) at 25 °C. (A) Lane1, digested with trypsin for 2 h; lane 2, digested with trypsin for 15 min in 2 M urea. (B) Lane 1, digested with chymotrypsin for 2 h; lane 2, digested with chymotrypsin for 15 min in 2 M urea. Fig. 5. Lipid-induced oligomerization of the HlyA detected by SDS/ PAGE. The protein : lipid weight ratio was maintained at 1 : 10 to ensure absence of the unbound toxin. Samples were treated with 1% SDS at 50 °C for 15 min, a temperature at which the oligomer does not dissociate into monomer. (A) Incubation with PtdCho-cholesterol for 1 min (lane 2) and 2 min (lane 3). Lane 1 shows HlyA in buffer A. (B) Incubation with PtdCho for 5 min (lane 2), 10 min (lane 3), 15 min (lane 4) and 20 min (lane 5). The HlyA in buffer A (lane 1) was included for comparison. Fig. 6. The binding of ANS to the HlyA monomer and oligomer. The incubation mixtures contained the monomer (– -) and oligomer (—) at a protein concentration of 90 lgÆmL )1 and ANS at 50 l M . Excitation was performed at 390 nm and slit widths were set at 5 nm (— – —) indicates ANS in buffer A. Fig. 7. Fluorescence energy transfer from the HlyA to dansyl-PtdEtn incorporated in PtdCho-cholesterol (—) and PtdCho (– -) liposomes. The toxin was incubated with liposome at a protein : lipid ratio of 2 : 1 (w/w), a ratio at which the sensitivity of the assay was found to be maximum. The incubation mixture was excited at 280 nm and fluor- escence emission was recorded at 512 nm approximately 10 s after mixing the toxin with liposome. Slit widths were 5 nm. Dansylated liposome suspensions without the HlyA served as the control. Ó FEBS 2002 Amphiphilicity and self-assembly of HlyA (Eur. J. Biochem. 269) 4355 followed exactly the time-course of self-assembly of the HlyA. Because the efficiency of the fluorescence energy transfer depends, among other things, on the proximity of the energy donor and acceptor groups [23], we interpret these observations as implying a tight association of the HlyA oligomer with the core of the PtdCho-cholesterol bilayer. Interaction with lipid induces conformational change in the HlyA monomer As incorporation of the HlyA in a lipid bilayer seemed to require cholesterol induced self-assembly of the toxin monomer, we thought it interesting to see if interaction of the toxin with the amphiphilic lipid matrix by itself induced a conformational change in the monomeric protein. To avoid complexity introduced by oligomerization of the toxin in PtdCho-cholesterol bilayer, we examined the fluorescence emission spectra of the toxin incubated with PtdCho and PtdEtn liposomes at an excitation wavelength of 295 nm (Fig. 8). Surprisingly, there was a drastic change in the spectrum, with the emission maximum showing a red shift of approximately 10 nm from 330 to 340 nm and a decrease of intensity by approximately twofold, indicating transfer of Trp residues from the nonpolar core of the protein to more polar surroundings, presumably at the lipid–water interface. These data suggest that although the HlyA monomer and oligomer had similar conformations (Fig. 3), the self- assembly involved a partially unfolded state of the protein as an intermediate that survived in the absence of cholesterol for a fairly long time without undergoing oligomerization. DISCUSSION Structure–function analysis of several PFTs, e.g. S. aureus a-toxin [13,27], aerolysin [12], pneumolysin [29], and perfringolysin [14] have established that oligomerization of the toxin is a critical cell surface event that precedes insertion of the protein in the target membrane. It is unusual for a water-soluble monomeric PFT to possess surface-exposed, uninterrupted hydrophobic stretches, and integration of such proteins with the nonpolar core of a lipid bilayer is energetically expensive. As an alternative, PFT binds to specific cell surface receptors and an increase in surface concentration by several orders of magnitude compared to that in water provokes self-assembly of the monomer to an amphiphilic oligomer. In this communication, we show that V. cholerae HlyA does not adhere to the details of this mechanistic framework of membrane permeabilization by PFTs. Despite solubility in water, the HlyA monomer is amphiphilic and the tendency of the toxin molecule to decrease its area of contact with water seems to dominate much of its functional and hydrodynamic properties. The HlyA, which was dissolved initially in 2% Triton X-114 at 4 °C, partitioned quantitatively to the detergent-rich phase on raising the temperature to 25 °C [20]. Because seques- tration of the HlyA by Triton X-114 could possibly be caused by affinity of the toxin for the detergent or by a conformational change in the native protein, we used an alternative strategy to monitor intrinsic amphiphilicity of the toxin. Amphiphilic molecules like detergents, which have high Gibbs free energy in water, adopt an energetically favorable orientation at the interface of water with air or a nonpolar organic solvent and tend to accumulate at the surface. Increase in the surface concentration of a protein, estimated indirectly by the decrease in the bulk aqueous phase concentration would therefore provide a measure of amphiphilicity. As chloroform does not affect the property of the aqueous phase due to its insolubility and lacks a functional group that can interact with a protein, we chose it as the organic solvent. The aqueous phase concentration of the HlyA was found to be 100-fold less in a water– chloroform system than in water in comparison to a twofold decrease observed with bovine serum albumin demonstra- ting that amphiphilicity was an intrinsic characteristic of the native HlyA. Amphiphilicity of the toxin was reflected in nonstoichiometric association of the protein in water, as indicated by light-scattering (Fig. 2) and gel filtration experiments. Furthermore, the complex of the HlyA with asialofetuin, a glycoprotein inhibitor of the interaction of the HlyA with synthetic lipid vesicles and biomembranes [10], was considerably more hydrophilic than the unbound toxin. On the basis of this positive correlation between effective amphiphilicity and affinity of the toxin for lipid vesicles and erythrocytes, we conclude that amphiphilicity drives the HlyA to phospholipid vesicles and is a major determinant of the interaction of the toxin with erythro- cytes. Such an interpretation is consistent with the nonspe- cific and nonsaturable interaction of the HlyA with lipid vesicles [8] and might resolve the controversy in the identity and role of the erythrocyte surface receptor in initiating the action of the toxin [10,30]. The preceding observations would suggest that the amphiphilic HlyA monomer might itself posses an inser- tion-competent configuration. We addressed the question by delinking lipid-binding from oligomerization by using PtdCho and PtdCho-cholesterol vesicles. On excitation of the HlyA incubated with such liposomes incorporating dansyl-PtdEtn as a fluorescent probe at 280 nm, transfer of Trp fluorescence energy to the dansyl moiety occurred only Fig. 8. Lipid-induced perturbation of the fluorescence emission spectrum of the HlyA. TheHlyAwasincubatedwithPtdCho(—)andPtdEtn liposomes (– -) for 10 min at a protein concentration of 50 lgÆmL )1 and a protein-lipid weight ratio of 1 : 10 and excited at 295 nm (— – —) shows the spectrum of the HlyA in buffer A at the same con- centration. Slit widths were 2.5 nm. Liposome suspensions without the HlyA served as the blank. 4356 K. Chattopadhyay et al.(Eur. J. Biochem. 269) Ó FEBS 2002 in PtdCho-cholesterol vesicles and was synchronous with the self-assembly of the toxin to the SDS-stable pentamer (Fig. 5A). In addition, the hemolytic activity of the lipid- bound HlyA correlated exactly with the quantity of the toxin present in the monomeric form, implying that the monomer and not the oligomer could exchange rapidly between the amphiphilic matrices of the synthetic lipid vesicle and erythrocyte membrane. Requirement of oligo- merization for integration of the HlyA with the nonpolar core of the lipid bilayer implies that the monomer does not adopt an insertion-competent configuration in spite of its intrinsic amphiphilicity. As a folded protein has nonpolar amino acids buried in a hydrophobic core insulated from water, it seems doubtful if the amphiphilic HlyA possesses a compact structure. Although no information is available for the HlyA struc- ture, a blue shift of 15 nm of the wavelength maximum of Trp fluorescence emission spectrum of the HlyA monomer (Fig. 3A) together with the Stern–Volmer analysis of the quenching of fluorescence by acrylamide (Fig. 3C) indicated that the multiple indolyl residues were located in a nonpolar region with limited accessibility to water. Because Trp residues are scattered along the length of the 65 kDa polypeptide chain [26], such an arrangement would imply a compact folded structure for the HlyA monomer. The compactness of the native HlyA in water was corroborated by resistance of the toxin to proteolytic degradation by trypsin and chymotrypsin (Fig. 4). Furthermore, the iden- tity in shape of the spectra of the HlyA monomer and the oligomer (Fig. 3A) and the similarity in quenching charac- teristics of the two proteins indicated that the two toxin forms shared similar global conformations. Nevertheless, the HlyA monomer and the oligomer could be distinguished by ANS. Although the increase in fluorescence intensity was relatively small in comparison to the large enhancements caused by proteins in molten globule states [31] the data suggested the presence of exposed hydrophobic patches on the oligomer that might be instrumental in conferring it the insertion-competent configuration. It may be recalled that self-assembly of monomer to insertion-competent oligomer without major changes in secondary structure and tertiary foldings have been documented for S. aureus a-toxin [13] and aerolysin [12]. Although amphiphilicity-driven partitioning of the HlyA in phospholipid vesicles devoid of cholesterol failed to trigger rapid oligomerization of the toxin, the interaction caused a profound change in the conformation of the protein (Fig. 8). A red shift of the wavelength maximum of intrinsic Trp fluorescence of the monomer by 10 nm induced by phospholipids irrespective of head groups indicated that a significant fraction of the multiple indolyl emitting centers were transferred from an apolar region in the protein core to a more polar environment, presumably at the lipid–water interface. A partial collapse of the folded HlyA, as suggested by the spatial redistribution of Trp residues, might be similar to the lipid-induced transition of the native conformation to a molten globule state observed with a number of proteins, e.g. colicin A [32], cytochrome c [33], and human apolipoprotein [34]. Cholesterol had a dramatic effect on the kinetics of self-assembly of the HlyA, an irreversible process that seemed to be guided by the higher thermodynamic stability of the oligomer; however, it exerted its effect in a lipid bilayer matrix only. Although the nature and extent of the disruption in structure of the HlyA induced by phospholipids is unclear, self-assembly of the toxin appears to involve a partially unfolded state as a fairly stable intermediate, followed by cholesterol-assisted recon- stitution to the oligomer with recovery of much of the folding patterns of the monomer. In recent years, there has been wide acceptance of chaperone-like roles of specific lipids in the correct folding of unfolded integral membrane proteins [35]. Such examples include the roles of PtdEtn and PtdCho in the folding of Escherichia coli lactose permease [36] and OmpA [37], respectively. Cholesterol might play a similar role in conversion of the HlyA monomer to the oligomer. Adoption of such a mechanism for generating functional channels has enabled the HlyA to select eukary- otic cells as targets in consonance with its survival and multiplication in animal guts and spare prokaryotic cells that lack cholesterol in their membrane. ACKNOWLEDGEMENT We thank the Council of Scientific and Industrial Research, New Delhi, India for a grant (Scheme no. 37/0955/97-EMR-II) and a senior research fellowship to K.C. We thank S. K. Bhattacharyya, Director, National Institute of Cholera and Enteric Diseases, Kolkata, for support. REFERENCES 1. Kaper, J.B., Morris, J.G. Jr & Levine, M.M. (1995) Cholera. Clin. Microbiol. Rev. 8, 48–86. 2.Yamamoto,K.,Al-Omani,M.,Honda,T.,Takeda,Y.& Miwatani, T. (1984) Non-O1 Vibrio cholerae hemolysin: purification, partial characterization and immunological related- ness to El Tor hemolysin. Infect. Immune. 45, 192–196. 3. Ojcius, D.M. & Young, J.D. (1991) Cytolytic pore-forming pro- teins and peptides: is there a common structural motif? Trends Biochem. Sci. 16, 225–228. 4. Welch, R.A. (1991) Pore-forming cytolysins of Gram-negative bacteria. Mol. Microbiol. 5, 521–528. 5. Goldberg, S.L. & Murphy, J.R. (1984) Molecular cloning of the hemolysin determinant from Vibrio cholerae El Tor. J. Bacteriol. 160, 239–244. 6. Manning, P.A., Brown, M.H. & Heuzenroeder, M.W. (1984) Cloning of the structural gene (hly) for the haemolysin of Vibrio cholerae El Tor strain O17. Gene 31, 225–231. 7. Krasilnikov, O.V., Muratkhodjaev, J.N. & Zitzer, A.O. (1992) ThemodeofactionofVibrio cholerae cytolysin. The influences on both erythrocytes and planar lipid bilayers. Biochim. Biophys. Acta 1111, 7–16. 8. Zitzer,A.,Palmer,M.,Weller,U.,Wessenaar,T.,Biermann,C., Tranum-Jensen, J. & Bhakdi, S. (1997) Mode of primary binding to target membranes and pore formation induced by Vibrio cholerae cytolysin (hemolysin). Eur. J. Biochem. 247, 209–216. 9. Zitzer, A., Zitzer, O., Bhakdi, S. & Palmer, M. (1999) Oligomer- ization of Vibrio cholerae cytolysin yields the pentameric pore and has the dual specificity for cholesterol and sphingolipids in the target membrane. J. Biol. Chem. 274, 1375–1380. 10. Saha, N. & Banerjee, K.K. (1997) Carbohydrate mediated reg- ulation of interaction of Vibrio cholerae hemolysinwithery- throcyte and phospholipid vesicles. J. Biol. Chem. 272, 162–167. 11. Ichinose, Y., Yamamoto, K., Nakasone, N., Tanabe, M.J., Takeda, T., Miwatani, T. & Iwanaga, M. (1987) Enterotoxicity of El Tor like hemolysin of non-O1 Vibrio cholerae. Infect. Immune. 55, 1090–1093. 12. Parker, M.W., Buckley, J.T., Postma, J.P.M., Tucker, A.D., Leonard, K.V., Pattus, F. & Tsernoglou, D. (1994) Structure of Ó FEBS 2002 Amphiphilicity and self-assembly of HlyA (Eur. J. Biochem. 269) 4357 the Aeromonas toxin proaerolysin in its water-soluble and mem- brane-channnel states. Nature 367, 292–295. 13. Song, L., Hobaugh, M.R., Shustak, C., Cheley, S., Bayley, H. & Gouaux, J.E. (1996) Structure of Staphylococcal a-hemolysin, a heptameric transmembrane pore. Science 274, 1859–1866. 14. Hotze, E.M., Wilson-Kubalek, E.M., Rossjohn, J., Parker, M.W., Johnson, A.E. & Tweten, R.K. (2001) Arresting pore formation of a cholesterol-dependent cytolysin by disulphide trapping syn- chronizes the insertion of the transmembrane b-sheet from a prepore-intermediate. J. Biol. Chem. 276, 8261–8268. 15. Walker, B., Braha, O., Cheley, S. & Bayley, H. (1995) An inter- mediate in the assembly of a pore-forming toxin trapped with a genetically engineered switch. Chem. Biol. 2, 99–105. 16. Zitzer, A., Harris, J.R., Kemminer, S.E., Zitzer, O., Bhakdi, S., Muething, J. & Palmer, M. (2000) Vibrio cholerae cytolysin: assembly and membrane insertion of the oligomeric pore are tightly linked and are not detectably restricted by membrane fluidity. Biochim. Biophys. Acta 1509, 264–274. 17.Zitzer,A.,Bittman,R.,Verbicky,C.A.,Erukulla,R.K., Bhakdi, S., Weis, S., Valeva, A. & Palmer, M. (2001) Coupling of cholesterol and cone-shaped lipids in bilayers augments mem- brane permeabilization by the cholesterol specific toxins strepto- lysin O and Vibrio cholerae cytolysin. J. Biol. Chem. 276, 14628–14633. 18. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227,680– 685. 19.Markwell,M.A.K.,Haas,S.M.,Beilur,L.L.&Tolbert,N.E. (1978) A modification of the Lowry procedure to simplify protein determination in membrane and lipoprotein samples. Anal. Bio- chem. 87, 206–210. 20. Bordier, C. (1981) Phase separation of integral membrane proteins in Triton X-114 solution. J. Biol. Chem. 256, 1604–1607. 21. Fersht, A. (1998) Forces between molecules and binding energies. In Structure and Mechanism in Protein Science, 1st edn, pp. 324– 348. 22. Ikigai, H. & Nakae, T. (1987) Interaction of the a-toxin of Sta- phylococcus aureus with the liposome membrane. J. Biol. Chem. 262, 2150–2165. 23. Brand, L. & Witholt, B. (1967) Flourescence measurements. Methods Enzymol. 11, 776–856. 24. Ganong, B.R. & Delmore, J.P. (1991) Phase separation tem- peratures of mixtures of Triton X-114 and Triton X-45: applica- tion to protein separation. Anal. Biochem. 193, 35–37. 25. Sharon, N. & Lis, H. (1982) Glycoproteins. In: The Proteins (Neurath, H. & Hill, R.L., eds), Vol. 5, pp. 1–144. Academic Press, New York. 26.Heidelberg,J.F.,Eisen,J.A.,Nelson,W.C.,Clayton,R.A., Gwinn, M.L., Dodson, R.J., Haft, D.H., Hickey, E.K., Peterson, J.D., Umayam, L., Gill, S.R., Nelson, K.E., Read, T.D., Tettelin, H., Richardson, D., Ermolaeva, M.D., Vamathevan, J., Bass, S., Qin,H.,Dragoi,I.,Sellers,P.,McDonald,L.,Utterback,T., Fleishmann,R.D.,Nierman,W.C.,White,O.,Salzberg,S.L., Smith,H.O.,Colwell,R.R.,Mekalanos,J.J.,Venter,J.C.& Fraser, C.M. (2000) DNA sequence of both chromosomes of the cholera pathogen Vibrio cholerae. Nature 406, 477–483. 27. Bhakdi, S. & Tranum Jensen, J. (1991) Alpha-toxin of Staphylo- coccus aureus. Microbiol. Rev. 55, 733–751. 28. Stryer, L. (1965) The interaction of a naphthalene dye with apo- myoglobin and apohemoglobin. A fluorescent probe of nonpolar binding sites. J. Mol. Biol. 13, 482–485. 29. Gilbert, R.J.C., Jimenez, J.L., Chen, S., Tickle, I.J., Rossjohn, J., Parker, M., Andrew, P.W. & Saibil, H.R. (1999) Two structural transitions in membrane pore formation by pneumolysin, the pore-forming toxin of Streptococcus pneumoniae. Cell 97, 647–655. 30. Zhang, D., Takahasi, J., Seno, T., Tani, Y. & Honda, T. (1999) Analysis of receptor for Vibrio cholerae El Tor hemolysin with a monoclonal antibody that recognizes glycophorin B of human erythrocyte membrane. Infect. Immune. 67, 5332–5337. 31. Semistonov, G.V., Rodionova, N.A., Razgulyaev, O.I., Uversky, V.N., Gripas, A.F. & Gilmanshin, R.I. (1991) Study of the Ômolten globuleÕ intermediate state in protein folding by a hydrophobic fluorescent probe. Biopolymers 31, 119–128. 32. van der Goot, F.G., Gonzales-Manas, J.M., Lakey, J.H. & Pattus, F. (1991) A Ômolten-globuleÕ membrane-insertion intermediate of the pore-forming domain of colicin A. Nature 354, 408–410. 33. Bychkova, V.E., Dujsekina, A.E., Klenin, S.I., Tiktopula, E.I., Uversky, V.N. & Ptitsyn, O.B. (1996) Molten globule-like state of cytochrome c under conditions simulating those near the mem- brane surface. Biochemistry 35, 6058–6063. 34. Clement-Collin, V., Leroy, A., Monteilhet, C. & Aggerbeck, L.P. (1999) Mimicking lipid-binding-induced conformational changes in the human apolipoprotein E N-terminal receptor binding domain: effects of low pH and propanol. Eur. J. Biochem. 264, 358–368. 35. Bogdanov, M. & Dowhan, W. (1999) Lipid-assisted protein folding. J. Biol. Chem. 274, 36827–36830. 36. Bogdanov, M., Umeda, M. & Dowhan, W. (1999) Phospholipid- assisted refolding of an integral membrane protein: minimum structural features for phophatidylethanolamine to act as a molecular chaperone. J. Biol. Chem. 274, 12339–12345. 37. Surrey, T. & Jahnig, F. (1992) Refolding and oriented insertion of a membrane protein into a lipid bilayer. Proc. Natl Acad. Sci. USA 89, 7457–7461. 4358 K. Chattopadhyay et al.(Eur. J. Biochem. 269) Ó FEBS 2002 . Vibrio cholerae hemolysin Implication of amphiphilicity and lipid-induced conformational change for its pore-forming activity Kausik Chattopadhyay 1 , Debasish Bhattacharyya 2 and Kalyan. protein and not by its affinity for Triton X-114 or a conformational change induced by the detergent, we monitored accumulation of the toxin at the interface of water and chloroform. The ratio of the. lgÆmL )1 was mixed in duplicate with an equal volume of chloroform at 25 °C, vortexed for 10 min to maximize the area of water–chloroform interface and allowed to separate. The HlyA concentration