The 118–135 peptide of the human prion protein forms amyloid fibrils and induces liposome fusion1

J. Mol. Biol. (1997) 274, 381±393

The 118 ±135 Peptide of the Human Prion Protein forms Amyloid Fibrils and Induces Liposome Fusion Thierry Pillot1, Laurence Lins2, Marc Goethals3, Berlinda Vanloo1 Johan Baert4, Joel Vandekerckhove3, Maryvonne Rosseneu1* and Robert Brasseur2 1

Laboratory for Lipoprotein Chemistry, Department of Biochemistry, Universiteit Gent, Belgium 2

Centre de Biophysique MoleÂculaire NumeÂrique, Faculte des Sciences Agronomiques de Gembloux, Gembloux, Belgium 3

Flanders Interuniversity Institute for Biotechnology Universiteit Gent, B-9000 Gent Belgium 4

Interdisciplinary Research Center, University of Leuven Campus Kortrijk, Belgium

The prion protein (PrPC) is a glycoprotein of unknown function normally found at the surface of neurons and of glial cells. It is involved in diseases such as bovine spongiform encephalopathy, and Creutzfeldt-Jakob disease in the human, where PrPC is converted into an altered form (termed PrPSc). PrPSc is highly resistant towards proteolytic degradation and accumulates in the central nervous system of affected individuals. By analogy with the pathological events occuring during the development of Alzheimer's disease, controverses still exist regarding the relationship between amyloidogenesis, prion aggregation and neuronal loss. To unravel the mechanism of PrP neurotoxicity and understand the interaction of PrP with cellular membranes, a series of natural and variant peptides spanning residues 118 to 135 of PrP was synthesized. The potential of these peptides to induce fusion of unilamellar lipid vesicles was investigated. According to computer modeling calculations, the 120 to 133 domain of PrP is predicted to be a tilted lipid-associating peptide, and to insert in a oblique way into a lipid bilayer through its N-terminal end. In addition to amyloidogenic properties exhibited in vitro by these peptides, peptide-induced vesicle fusion was demonstrated by several techniques, including lipid- and core-mixing assays. Elongation of the 120 to 133 peptide towards the N- and C-terminal ends of the PrP sequence showed that the 118 to 135 PrP peptide has maximal fusogenic properties, while the variant peptides had no effect. Due to their high hydrophobicity, all peptides tested were able to interact with liposomes to induce leakage of encapsulated calcein. We demonstrate also that the propensity of the peptides to fold as an a-helix increases their fusogenic activity, thus accounting for the maximal fusogenic activity of the most stable helix at residues 118 to 135. These data suggest that, by analogy with the C-terminal domain of the b-amyloid peptide, the fusogenic properties exhibited by the prion peptides might contribute to the neurotoxicity of these peptides by destabilizing cellular membranes. # 1997 Academic Press Limited

*Corresponding author

Keywords: prion protein; synthetic peptide; a-helix; membrane fusion; amyloid ®brils

Introduction The cellular prion protein (PrPC) is a 33 to 35 kDa sialoglycoprotein expressed mainly in neurons and, to a lesser extent, in other cells (for Abbreviations used: SUV, LUV, small and large unilamelar visicles, respectively; PC, L-a-phosphatidylethanolamine; PS, L-a-phosphatidyl-L-serine; PC, L-a-phosphatidylcholine; SM, bovine sphingomyelin; TFE, tri¯uoroethanol; HFP, hexa¯uoro-2-propanol; E/M, excimer/monomer ratio. 0022±2836/97/480381±13 $25.00/0/mb971382

review, see Prusiner, 1994). The protein is either bound to the cell surface by a glycosyl phosphatidylinositol anchor or secreted as a soluble derivative (Chen et al., 1995; Perini et al., 1996). Prion diseases are unusual fatal neurodegenerative disorders occurring in human and in animals, that are characterized by dementia and motor dysfunction, and neuropathologically by cerebral spongiosis and amyloidosis. Mutations in the PRNP gene cosegregate with hereditary human disorders such as familial Creutzfeldt-Jakob disease, Gerstmann# 1997 Academic Press Limited

382 StraÈussler-Scheinker disease, and fatal familial insomnia (Prusiner, 1994). The central molecular event underlying these different forms of prion diseases is likely to be a posttranslational alteration of the PrPC molecule that yields an abnormal protease-resistant PrP isoform designated PrPc (Oesch et al., 1994). This isoform, which probably accounts for the entire infectious prion particles, accumulates abnormally in the brain. The structural details of this transition remain uncertain, but they might involve the conversion of a-helices into b-sheets in critical regions of the polypeptide chain. Molecular modeling studies suggested that PrPC might fold into a four ahelix bundle protein, with helices at residues 109 to 122, 129 to 141, 178 to 191 and 202 to 218 (Huang et al., 1995). One or more of the putative a-helices might refold into b-sheets upon conversion of PrPC to PrPSc. Results of solid-state NMR studies have recently suggested that residues 112 to 121, predicted to belong to the ®rst a-helix of the bundle might play a key role in the conformational transition involved in the development of prion diseases (Heller et al., 1996). In contrast with the model prediction of the helical structure of the PrPC, Riek et al. (1996) demonstrated, by NMR studies of the 121 to 231 domain of the mouse prion protein, the presence of two antiparallel b-sheets at residues 128 to 131 and 161 to 164 and of three a-helices (Riek et al., 1996). The presence of a b-sheet at position 128 to 131 might be critical for the initiation of the transition from PrPC to PrPSc. The formation of nascent PrPSc appears to involve interactions between PrPC and PrPSc, and can be modeled by PrP peptides. It has been demonstrated, using PrP knockout mice, that normal PrPC is necessary to PrPSc-induced infectivity and neurotoxicity (Brown et al., 1996). A common feature between prion-associated diseases and Alzheimer's disease is the deposition of extracellular protein aggregates in the form of amyloid. It has been proposed that PrPSc might be an aggregate of PrPC, but the relationship between aggregation, amyloidogenesis, and infectivity remains unclear. In vitro amyloid formation by either PrP of PrP-related peptides has been extensively described (Come et al., 1993). Altogether, these results suggest that the residues 106 to 147 of the prion protein are capable to spontaneously form amyloid-like ®brils, and that the PrP sequence at residues 118 to 133 represents the amyloidogenic domain of the PrP (Jarrett & Lansbury, 1993). These amyloidogenic properties might be related to the conformational instability of this domain of the prion protein. It has further been proposed that perturbation of the packing environment of the highly conserved residues, belonging to the two ®rst modelized a-helices of PrP, is a possible mechanism for triggering the conversion of PrPc to PrPSc, where a-helices appear to be converted into b-sheets as a primary event in amyloid formation. The mechanism of induced-prion protein neurotoxicity is still poorly documented. It has been suggested that a

Fusogenic Properties of the 118 ±135 Prion Peptide

prion protein fragment (e.g. residues 106 to 126) can induce neurodegenerescence, which occurs by apoptosis (Forloni et al., 1993). This peptide, which is toxic for neurons, is, however, unable to destroy neurons which do not express PrPC (Brown et al., 1996). These results indicate that aggregation of PrPSc is not a necessary requirement for neurotoxicity and infectivity and that PrPSc degradation products may play a role in the nerve cell degeneration that occurs in prion-related encephalopathies. A portion of the PrP sequence, including amino acids 118 to 133, is highly conserved across species and shares homology with the C-terminal domain of the Alzheimer's b-amyloid peptide. In analogy with the molecular events occuring in Alzheimer's disease, this prion peptide has been used to analyze the kinetics of amyloid formation in the prionrelated disease (Come et al., 1993). We have recently demonstrated that the C-terminal domain of the Alzheimer's b-amyloid peptide (e.g. amino acids 29 to 42) has in vitro fusogenic properties on liposomes, due to a tilted insertion of the peptide into the lipid bilayer (Pillot et al., 1996, 1997). Moreover, we proposed that this tilted lipid-associating peptide, which is critical for b-amyloid formation, could directly mediate the interaction of the b-amyloid peptide with cell membrane and thereby the peptide toxicity by causing further damages to the cell. Membrane destabilization, due to the tilted penetration of fusion peptides into cellular membranes is induced by a variety of peptides (Brasseur, 1991; Brasseur et al., 1997). These hydrophobic peptides penetrate the lipid bilayer at an angle of 45 to 50 , due to the hydrophobicity gradient along their sequence (Zimmerberg et al., 1993). Computer modeling of the prion protein showed that a prion-peptide encompassing residues 120 to 133 has properties similar to those of the fusion Alzheimer's peptide (Pillot et al., 1996). We therefore investigated the fusogenic properties of a series of synthetic prion peptides and report in this paper, that together with its amyloidogenic capacity, the 118 to 135 peptide of the prion protein is able to induce fusion of unilamellar lipid vesicles with concommitant perturbation of membrane permeability as monitored by calcein leakage. The a-helical structure of the peptides seems critical for their interaction with a lipid bilayer and for their fusogenic activity. Our results propose an alternative mechanism for the prion cytotoxicity through a direct perturbation of the cellular plasma membrane similar to that induced by viral fusion peptides and by the Alzheimer's b-amyloid peptide. The penetration of the cell membrane by the peptide can cause membrane perturbations and further cell damages.

Results Computer modeling of the prion peptides The sequence and properties of the synthetic prion peptides are summarized in Table 1. In ana-

383

Fusogenic Properties of the 118±135 Prion Peptide Table 1. Properties of the synthetic prion peptides Peptide Prion (120±133) Prion (120±133, 0 ) Prion (120±133, 85 ) Prion (120±135) Prion (118±135) Prion (118±135, 0 ) Ab(29±42)

Peptide sequence AVVGGLGGYMLGSA ASVGGLMGYLGGVA MVVGGLGGYALGSA AVVGGLGGYMLGSAMS AGAVVGGLGGYMLGSAMS AGGVVGGLGGYMLASAMS GAIIGLMVGGVVIA

Predicted orientation

Angle of insertion

Mean hydrophobicity

Hydrophobic moment

Oblique Parallel Perpendicular Oblique Oblique Parallel Oblique

45 0 85 45 25 0 50

0.63 0.63 0.63 0.58 0.57 0.57 0.89

0.077 0.145 0.074 0.075 0.100 0.098 0.039

Mean hydrophobicity and hydrophobic moment were computed using the Eisenberg consensus scale (Brasseur et al., 1992). Underlined amino acids correspond to position changes compared to the prion (120±133) and (188±135) wild-type peptides. The angle of insertion in a lipid bilayer was calculated according to Brasseur (1991).

logy with the fusogenic Ab(29 to 42) peptide, the prion peptides have high hydrophobicity and are predicted to insert in an oblique way in a lipid bilayer at angles between 25 for the prion (118 to 135) peptide to 45 for the shorter prion (120 to 133) peptide. In addition, variant peptides with the same amino acid composition, named prion (120 to 133, 0 ), prion (120 to 133, 85 ), and prion (118 to 135, 0 ), were designed in order to take either a parallel or a perpendicular orientation at a lipid/ water interface (Table 1). As carried out previously for the viral fusion peptides (Horth et al., 1991) and for the Alzheimer b-amyloid peptide (Pillot et al., 1996), the conformation of the synthetic prion peptides at a lipid/water interface was calculated by

energy minimization (Brasseur, 1991). The most probable conformations of the peptides are illustrated in Figure 1, showing the tilted orientation of the prion (120 to 133), (120 to 135) and (118 to 135) peptides, together with the parallel orientation of the mutant peptides. The oblique orientation of the wild-type peptides compared to the lipid bilayer is due to the concentration of the most hydrophobic residues at the N-terminal extremity of the peptide, thereby creating a C-N hydrophobicity gradient along the sequence. As previously described for the Alzheimer b-amyloid peptide (Pillot et al., 1996; Brasseur et al., 1997), the prion peptides penetrate the phospholipid bilayer or insert into the lipid phase through their more hydrophobic N-terminal

Figure 1. Computer modeling of the mode of insertion of the prion peptides within a lipid matrix. For simplicity reasons, lipids were not drawn. The horizontal line represents the interface between the hydrophobic (upper) and the hydrophilic (bottom) phases.

384 end and the eight to nine N-terminal residues insert into the lipid to destabilize the regular packing of the phospholipid acyl chains (Figure 1). Lipid-mixing induced by the tilted prion peptides The induction of intervesicular lipid-mixing by peptides, as a measure of their fusogenic activity, was tested with PC/PE/PS/SM/cholesterol SUVs utilizing a probe dilution assay (Morris et al., 1988; Pillot et al., 1996). SUVs labeled with pyrenelecithin were mixed with unlabeled vesicles and the ratio of the excimer to monomer intensity of the pyrene probe was measured as a function of time. A decrease of the excimer intensity and an increase of the monomer intensity due to the dilution of the probe into the fused vesicles is a measure of the fusion. As shown in Figure 2, among the six peptides tested, only the prion (120 to 135) and (118 to 135) peptides, and to lesser extent the prion (120 to 133) peptide, were able to decrease the E/M ratio of mixed unlabeled and

Figure 2. Time course of lipid-mixing of PC/PE/PS/ SM/cholesterol SUVs induced by 13.8 mM of different peptides. Peptide aliquots were added to a mixture of labeled SUVs (3 mg phospholipid) containing 2.5 mol% of Pyr-PC and of unlabeled SUVs (12 mg phospholipid) in a 10 mM Tris-HCl, pH 8.0 buffer, containing 150 mM NaCl, 1 mM NaN3, and 0.1 g/l Na-EDTA. The Pyr-PC excimer/monomer ratio was monitored at room temperature, and is plotted, as a percentage of the initial value, versus time. Experimental values represent mean of three separate experiments. The different peptides used in this experiment were: PrP (120 to 133) (^), PrP (120 to 133, 0 ) (~), PrP (120 to 133, 85 ) (^), PrP (120 to 135) (&), PrP (118 to 135) ( & ), and PrP (118 to 135, 0 ) (~).

Fusogenic Properties of the 118 ±135 Prion Peptide

labeled SUVs. The mutant prion peptides had no fusogenic activity, as predicted from computer modeling and from the calculation of the peptide obliquity (Table 1). The longer prion (118 to 135) peptide had the strongest effect inducing to a E/M ratio decrease of 43%, whereas the prion (120 to 133) and (120 to 135) peptides decreased the E/M ratio by 32 and 18%, respectively. In a control experiment, a prion (120 to 133) peptide, with a randomized sequence: AVGVLGGYGMLGSA, showed no fusogenic activity, thus con®rming the validity of the experimental approach (data not shown). Similar results were obtained using labeled and unlabeled LUV preparations to monitor lipid-mixing induced by the prion peptides (data not shown). The dependency of both the extent and kinetics of the lipid-mixing process on the peptide concentration, was examined by adding increasing amounts of prion (120 to 133), (120 to 135) and (118 to 135) peptides to a ®xed amount of SUVs. The time course of the E/M ratio decrease induced by addition of various amounts of the three peptides showed that all peptides are able to induce lipid-mixing, in a concentration-dependent manner. In order to compare the relative fusogenic activity of the prion peptides, the extent of the E/ M ratio decrease was measured ten minutes after addition of increasing amounts of peptides (Figure 3). The prion (118 to 135) peptide was found to be more active than the shorter peptides, as it causes fusion more rapidly and to a higher

Figure 3. In¯uence of the peptide concentration on the extent of lipid-mixing of PC/PE/PS/SM/cholesterol SUVs induced by the PrP (118 to 135) ( & ), the PrP (120 to 135) (&) and the PrP (120 to 133) (^) peptides. Percentage of E/M ratio decrease was plotted versus peptide concentration, after ten minutes of SUVs and peptide incubation.

385

Fusogenic Properties of the 118±135 Prion Peptide

extent than the prion (120 to 133) and (120 to 135) peptides. At a 46 mM concentration, the prion (118 to 135) peptide caused a 48% of E/M ratio decrease after ten minutes incubation with the vesicles, whereas the E/M ratio decreased by 34 and 19%, respectively, when the prion (120 to 135) and (120 to 133) peptides were mixed with the vesicles at the same peptide concentration. Similar results were obtained using PC/PE/cholesterol SUVs (data not shown). In control experiments, incubation of mixed labeled and unlabeled vesicles in the absence of peptide or incubation of labeled vesicles alone with the prion peptides, had no effect on the E/M ratio (data not shown). These results strongly suggest that the observed lipid-mixing induced by the prion peptides is the result of intervesicular lipid-mixing, as previously described (Morris et al., 1988). Core-mixing induced by the synthetic prion peptides In order to further document the fusogenic properties of the prion peptides previously monitored by the lipid-mixing assay, experiments were carried out to demonstrate content-mixing of two vesicle populations labeled separately (Kendall & McDonald, 1982). When the prion peptides were added to a mixture of calcein, Co2‡- and EDTA-containing PC/PE/PS/SM/cholesterol SUVs, an increase of the ¯uorescence was observed (data not shown). The observed increase in ¯uorescence, due to an increased concentration of free calcein, could be a consequence of EDTA-Co2‡ complex formation occuring within the inner compartment of the fused vesicles. On the other hand, if leakage of the vesicle contents occurs, dilution and subsequent dissociation of calcein-Co2‡ complex might also take place. This would increase the emission intensity of the free calcein. Fluorescence due to leakage of vesicle contents can be eliminated by adding 0.4 mM Co2‡ (chelated in a 1:1 molar ratio with citrate) in the outer phase. Figure 4 illustrates the effect of increasing concentrations of the different prion peptides on the core-mixing of calcein, Co2‡- and of EDTA-containing SUVs after ten minutes of incubation in the presence of Co2‡ in the external phase. These results show that only the prion (118 to 135) and (120 to 135) peptides could induce a signi®cant content-mixing of the two populations of vesicles at the concentration range used in the lipid-mixing experiments. In agreement with the lipid-mixing assay, the prion (118 to 135) peptide is most fusogenic as it induces a 21% increase of the calcein emission intensity at 46 mM, compared to 16% for the (120 to 135) peptide. On the contrary, the shorter (120 to 133) peptide was not able to induce signi®cant mixing of the core content of the lipid vesicles, since the ¯uorescence increase was only 7% at the highest peptide concentration used. In control experiments, we demonstrated that neither the prion (118 to 135, 0 ), (120 to 133, 0 ), and (120 to 133, 85 ) mutant peptides were able to cause signi®-

Figure 4. Mixing of liposomal contents induced by the PrP peptides. Peptides were dissolved at 1 mg/ml in 50% TFE and peptide aliquots were added to a mixture of calcein, Co2‡- and of EDTA-containing SUVs (1:1 ratio, 15 mg phospholipids) in a 10 mM Tris-HCl, pH 8.0 buffer, containing 150 mM NaCl, 1 mM NaN3 and 0.4 mM Co2‡ chelated with 0.4 mM citrate. Core-mixing of the two populations of vesicles was followed as described in Materials and Methods and the percentage of the maximum calcein ¯uorescence after ten minutes incubation was plotted as a function of the peptide concentration. PrP (118 to 135) ( & ), PrP (120 to 135) (&), and PrP (120 to 133) (^). 100% of the calcein ¯uorescence was established by lysing the vesicles with Triton X-100 (0.5%, w/v) in the presence of 10 mM EDTA.

cant content-mixing of the two population of vesicles (data not shown). Prion peptides-induced leakage of liposomal contents The results of the core-mixing assay suggest that, concomitantly with membrane fusion and vesicle core-mixing, release of vesicular content occurs in the presence of the prion peptides. To assess the leakage of vesicular content due to the interaction with the synthetic prion peptides, we performed leakage measurement of free calcein encapsulated in SUVs. The addition of the prion peptides to SUVs (15 mg of phospholipids) produced membrane destabilization as measured by the release of calcein from the vesicles. Figure 5 illustrates the concentration-dependent release of calcein from SUVs, induced by the addition of the prion (120 to 133), (120 to 135) and (118 to 135) peptides. Within the concentration range, the rate of calcein leakage was signi®cant for the two longer peptides, and to a lesser extent for the prion (120 to 133) peptide. As

386

Figure 5. Leakage of liposomal contents induced by the PrP peptides. Peptides were dissolved at 1 mg/ml in 50% TFE and peptide aliquots were added to calcein SUVs (15 mg phospholipids) in a 10 mM Tris-HCl, pH 8.0 buffer, containing 150 mM NaCl, 1 mM NaN3, and 0.01 g/l Na-EDTA. Calcein release from the vesicles was followed as described in Materials and Methods and percentage of calcein release after ten minutes of incubation with increasing concentrations of prion peptides was plotted as a function of the peptide concentration. PrP (118 to 135) ( & ), PrP (120 to 135) (&), and PrP (120 to 133) (^). 100% of leakage was established by lysing the vesicles with Triton X-100 (0.5%, w/v).

shown in Figure 5, after ten minutes incubation, the prion (118 to 135) peptide was more active than the other peptides, in agreement with the results from the lipid- and core-mixing assays (Figures 3 and 4). Under the same experimental conditions, the parallel and perpendicular variant prion peptides at a concentration of 46 mM induced only 5 to 12% of calcein release (data not shown). Although we cannot rule out an aspeci®c interaction of these peptides with the SUVs due to their high hydrophobicity, these results suggest, however, that the insertion of the wild-type peptides into the lipid phase enhances the release of the encapsulated calcein. Altogether, these results suggest that the fusion of SUVs induced by the prion peptides is correlated with lipid bilayer destabilization via a direct interaction of the peptides with the lipid phase and that this process is more ef®cient for the prion (118 to 135) peptide. Lipid-binding properties of the prion peptides The results from the lipid- and core-mixing experiments, together with the calcein leakage from the liposomes, suggest that the prion peptides

Fusogenic Properties of the 118 ±135 Prion Peptide

interact directly with the lipid bilayers to perturb the packing of the lipid acyl chains. In order to monitor the lipid-binding properties of the different prion peptides used in this study, we separated free and lipid-associated peptides by size-exclusion chromatography. Mixtures of lipid vesicles and peptides, as well as vesicles and peptides alone, were applied to a Sephadex G-25 column equilibrated in a 10 mM Tris-HCl buffer containing 150 mM NaCl (Figure 6). Peptides were detected by measurement of the Tyr-¯uorescence at 305 nm, while the lipid vesicles were monitored by measurement of the pyrene monomer and excimer ¯uorescence. As shown in Figure 6A, when mixed with SUVs under the same conditions as for lipidmixing assays, the prion peptides eluted in the void volume of the column together with the liposomes. On the contrary, peptides alone were recovered within the volume of the column at an elution volume of 4 to 5 ml. When the ¯uorescence intensity of pyrene-PC-monomers was used to followed pyrene-labeled SUVs, these appeared in the void volume of the column. Interestingly, the ¯uorescence emission intensity of the pyrene-monomer was signi®cantly increased when the SUVs were incubated with the fusogenic prion (120 to 135) and (118 to 135) peptides (Figure 6B), whereas the (120 to 133) peptide induced only a slight increase in ¯uorescence emission intensity. These results showed that the three prion peptides are able to bind to lipid vesicles while inducing vesicle aggregation and fusion. These data were con®rmed by the decrease of the pyrene excimer ¯uorescence intensity when SUVs were mixed with the prion (118 to 135) and (120 to 135) fusion peptides (Figure 6C). In separate experiments, we demonstrated that the variant prion peptides had lipidbinding properties, but we did not observe any modi®cation of the pyrene ¯uorescence emission intensity (data not shown). Due to their high hydrophobicity, the mutant peptides are probably able to associate with liposomes, but cannot induce vesicle fusion, in agreement with their predicted orientation (Table 1). Conformational analysis of the synthetic prion peptides The secondary structure of the prion peptides was evaluated from the measurement of their circular dichroism spectra at increasing concentrations of TFE. In a 10 mM sodium phosphate (pH 7.4) solution, all the peptides exhibited mainly random coil structures with a characteristic negative band at 198 nm (data not shown). When dissolved in the presence of increasing concentrations of TFE, a well known a-helix promoting solvent, a-helical structures appeared, with characteristic negative CD bands at 208 and 222 nm and a positive band at 192 nm. As shown in Figure 7, at 20% TFE, the prion (118 to 135) peptide was mainly ahelical, as deconvolution of the CD spectra indicated that the percentage of a-helical structure

Fusogenic Properties of the 118±135 Prion Peptide

387

Figure 7. Circular dichroism spectra of the PrP (118 to 135) peptide at 23 C in the presence of increasing concentrations of TFE. Peptide was dissolved, at 0.1 mg/ml, 20 minutes before CD measurements in a 10 mM phosphate (pH 7.4) buffer containing 20% (1), 15% (2), 10% (3), 5% (4), or 0% (5) of TFE.

Figure 6. Monitoring of the lipid-binding properties of the synthetic prion peptides. Binding of the prion peptides to SUVs was monitored by size exclusion chromatography as described in Materials and Methods. A, 15 mg of the PrP (118 to 135) ( & ), the PrP (120 to 135) (&), and the PrP (120 to 133) (^) peptide in the absence of lipid: 15 mg of the PrP (118 to 135) (^), the PrP (120 to 135) (~), and PrP (120 to 133) (&) peptides incubated with SUVs (15 mg lipids) were loaded on a Sephadex G-25 column and the peptides were followed by Tyr-¯uorescence as a function of the elution volume.

reached 40 to 45%. Decreasing the TFE concentration resulted in a markedly decrease of the a-helical structure of the peptide, which had no resolved structure in an aqueous buffer (Figure 7). Under the same experimental conditions, the prion (120 to 135) and (120 to 133) peptides were 40 and 25%, respectively, a-helical (data not shown). These results suggest that the prion peptides require different concentrations of TFE to fold as an a-helix and that the longer peptides (e.g. 120 to 135 and 118 to 135) have a higher tendency to fold as an a-helix than the shorter (120 to 133) peptide. These results also suggest that these structural differences observed between the three prion peptides might account for the differences of fusogenic properties of these different peptides. When solubilized in different detergents, the prion (118 to 135) peptide had different a-helixforming propensities. As shown in Figure 8, the prion (118 to 135) peptide could fold as a stable a-helix in SDS, but not in cholate. Moreover, incubation with liposomes increased the a-helical content of this peptide, but to a lesser extent than in SDS. These data suggest that the hydrophobic environment within an SDS micelle or a lipid

Monitoring of the Pyr-PC monomer (B) and excimer (C) ¯uorescence in the absence of peptide (~) or in the presence of the PrP (118 to 135) ( & ), the PrP (120 to 135) (&), and the PrP (120 to 133) (^) peptides.

388

Fusogenic Properties of the 118 ±135 Prion Peptide

decreased fusogenic activity. As shown in Table 2, the lesser extent of lipid-mixing induced by the (120 to 133) peptide, compared to the longer peptides, might be related to the limited ability of this 14-residue peptide to fold as a stable a-helix. These results suggest that the secondary structure of the peptides is critical for interaction with the lipid bilayer and, therefore, for vesicle fusion. Although the computer modeling calculations predicted that the (120 to 133) prion peptide should have fusogenic properties, the low ef®ciency of the prion (120 to 133) peptide to induce core-mixing of the vesicles might be due to the poor propensity of this peptide to fold as a stable a-helix. The prion peptides form amyloid fibrils in vitro

Figure 8. The circular dichroism spectra of the PrP (118 to 135) peptide dissolved at 0.1 mg/ml in a 10 mM phosphate buffer at pH 7.4 in the presence of 0.1% (w/v) SDS (1), of PC/PE SUVs (peptide/lipid molar ratio 1:100) (2), and of 0.5% (w/v) sodium cholate (3).

bilayer might contribute to stabilize the a-helical structure of the peptide. Similar results were obtained with the prion (120 to 133) and (120 to 135) peptides (data not shown). Influence of the peptide secondary structure on their fusogenic properties In order to assess the in¯uence of the secondary structure on the fusogenic activity, the peptides were dissolved in decreasing concentrations of TFE and added to SUVs preparations. Compared to the peptides dissolved in 50% TFE described above, solubilizing the peptides in 20% TFE did not affect the extent of the intervesicular lipid-mixing induced by the peptides, whose secondary structure remained unchanged (Table 2). A further decrease of the percentage of TFE in the peptide solubilizing buffer, decreased the a-helical content of the peptide, and this was associated with a

The prion peptides used in this study were sparingly soluble in aqueous solution and readily formed ®brils in vitro. In order to demonstrate nucleated growth with prion-related peptides, it was necessary to remove pre-existing aggregates from the sample solution. According to recent observations (Come et al., 1993; Evans et al., Pillot et al., 1997; Wood et al., 1996), we prepared stock solutions of prion peptides in HFP at a concentration of 5 mg/ml and removed the solvent under vacuum prior to use. Figure 9 shows the time course and the extent of aggregate formation at 25 C as assessed by thio¯avin-T ¯uorescence measurements (LeVine, 1993), illustrating the different kinetics of aggregate formation for the three prion peptides. Aggregate formation begins within 10 to 20 hours for all peptides, and reaches completion at about 40 to 70 hours. Interestingly, the longer peptides (e.g. 120 to 135 and 118 to 135) start to form aggregates more rapidly than the (120 to 133) prion peptide, which form amyloid-like ®brils only after one day incubation. Furthermore, in all experiments conducted under these experimental conditions, the ®brils from the (120 to 133) prion peptide were signi®cantly less abundant at equal concentration of peptide than for the longer prionrelated peptides. Negative staining electron micrographs (Figure 10) show that aggregates of the (118 to 135) prion peptide exhibited the standard features of amyloid-like ®brils. These are approxiÊ in width and of indeterminate length, mately 100 A with some evidence of a twist along the longitudinal axis (Figure 10A). Some of these aggregates exhibit substructural features in the electron micro-

Table 2. In¯uence of the peptides secondary structure on their lipid-mixing properties Prion (120± 133) TFE (%) 10 15 20 50

Prion (120±135)

Prion (118±135)

a-Helix (%)

E/M ratio decrease (%)

a-Helix (%)

E/M ratio decrease (%)

a-Helix (%)

E/M ratio decrease (%)

5 13 25 28

ND 4 18 19

12 23 40 45

15 21 33 34

11 29 42 49

19 26 45 47

Percentages of a-helical contents were calculated according to Johnson (1990).

Fusogenic Properties of the 118±135 Prion Peptide

389

Figure 9. Time course of the PrP (118 to 135) ( & ), the PrP (120 to 135) (&), and the PrP (120 to 133) (^) peptide aggregation. Amyloid formation was measured by thio¯avin-T ¯uorescence.

scope, being apparently composed of parallel alignments of unbranched ®laments (Figure 10B).

Discussion As previously reported (Come et al., 1993), there is a strong homology between the 29 to 42 C-terminal domain of the Alzheimer's b-amyloid peptide and the 118 to 135 domain of the prion protein. We recently showed that the C-terminal domain of the b-amyloid peptide (e.g. residues 29 to 40 and 29 to 42) has fusogenic properties (Pillot et al., 1996, 1997), and proposed that, in analogy with the amyloid peptide, the (118 to 133) prion peptide belongs to class A of the oblique-orientated peptides (Brasseur et al., 1997). In the present paper, we showed that the 118 to 135 peptide of the human prion protein destabilizes lipid vesicles and induces liposomal fusion. The rationale for this study was to propose a mechanism through which the prion protein, or its degradation products, might be toxic to neuronal membranes, independently of the formation of PrPSc and its accumulation and aggregation in the brain. The destabilizing properties of the (118 to 135) prion protein peptide were investigated in a model system consisting of phospholipid vesicles, representing an hydrophobic ± hydrophilic interface such as present in biological membranes at the lipid ± water interface or in proteins in the core/surface regions. The (118 to 135) prion peptide induces vesicular fusion as demonstrated by mixing of both lipid membranes and inner content of ves-

Figure 10. Electron microscopy of the ®brils generated in vitro by the PrP (118 to 135) peptide (bars represent 100 nm).

icles, whereas parallel and perpendicular variant peptides had no activity. These results are in agreement with those obtained with the b-amyloid peptide (Pillot et al., 1996) and with the C-terminal peptide of apolipoprotein A-II (Lambert et al., 1997). We observed increased fusogenic activity for the prion (120 to 135) and (118 to 135) peptides, compared to the (120 to 133) peptide, suggesting that the number of hydrophobic residues penetrating the lipid phase together with the propensity of the peptides to fold as a stable a-helix are critical parameters for liposome fusion (Pillot et al., 1996). Peptide binding to the vesicles was accompanied by leakage of the encapsulated calcein, in agreement with recent observations showing that a neurotoxic fragment of the prion protein (e.g. 106 to 126 residues) forms ion permeable channels in pla-

390 nar lipid bilayer membranes (Lin et al., 1997). The 118 to 135 prion peptide penetrates the lipid phase through its more hydrophobic N-terminal end, which corresponds to the hydrophobic domain of the cytotoxic 106 to 126 prion peptide. As previously described for the b-amyloid peptide (Arispe et al., 1996), channel or pore formation might be the cytotoxic mechanism of action of amyloidogenic peptides found in prion-related diseases or other amyloidoses, possibly triggering apoptosis (Forloni et al., 1993). Recent data show that the 106 to 126 related-prion peptide might induce neuronal cell injury through a mechanism involving oxidative stress (Rizzardini et al., 1997). We thus propose that, based upon our experimental results, a direct interaction of PrPSc and/or its degradation products with neuronal membranes might play a role in the degeneration of nerve cells that occurs in prion-related encephalopathies. It has been recently suggested that PrPSc mutants or infectious PrPSc remain tightly associated with the plasma membrane after enzymatic cleavage of their glycosylphosphatidylinositol anchor, by an unknown mechanism (Lehmann & Harris, 1996). The mode of interaction of the prion protein or of prion-related peptides with the neuronal membranes has not been elucidated yet and it remains unclear whether it enhances the membrane succeptibility to further injury. Membrane fusion induced by fusogenic peptides is a cooperative mechanism and it might involve di-, tri- or tetramerization of peptides (Rapaport & Shai, 1994). The fusion peptide of in¯uenza virus hemagglutinin has an amphiphilic a-helical structure and it induces liposomal fusion through a cooperative mechanism. There is a structural homology between potential amyloidogenic peptides such as the prion peptide (Come et al., 1993) and other fusion peptides (Pillot et al., 1996). Moreover, the presence of glycine residues in the apparent consensus sequence XGXXXG is critical for fusogenic activity (Gray et al., 1996). Substitution of single glycine residues in the fusion peptide sequence induces profond structural changes and this further suggests that tertiary and quaternary structural interactions might be important for the binding of the fusion peptides to lipid bilayers. Such sequences were detected both in the prion protein (Come et al., 1993) and in the Alzheimer b-amyloid peptide (Pillot et al., 1996). According to the ion channel formation properties of the 106 to 126 prion peptide, we cannot rule out the oligomerization of the 118 to 135 prion peptide during the fusion process. However, we observed no fusogenic activity of an aggregated solution of the 118 to 135 fusion peptide. The synthetic prion peptides were further demonstrated to generate amyloid-like ®brils in vitro, as monitored by thio¯avin-T ¯uorescence measurements and by negative-staining electron microscopy. In agreement with a previous report (Come et al., 1993), we showed that the 118 to 135 domain in the human prion protein is highly amyloido-

Fusogenic Properties of the 118 ±135 Prion Peptide

genic. Interestingly, the (120 to 133) prion peptide is less amyloidogenic than the longer peptide, suggesting that the two residues at the N-terminal end are critical for amyloidogenesis. The oblique-orientated peptide described here is likely to be exposed at the surface of the prion protein according to both calculated and NMR-derived structures (Huang et al., 1995; Riek et al., 1996). Moreover, in both structures, it should consist of mixed conformational elements, either an helix-turn-helix in the theoretical model, or a b-sheet and a a-helix in the NMR structure. This peptide is therefore likely to possess high conformational mobility and to be structurally instable. A conformational change from b-sheet to a-helix might therefore occur either under lipidbinding or through its interaction with another domain of the prion protein. We therefore hypothesize that this particular property of the fusion prion peptide might be important in the conversion of PrPC to PrPSc. This particular domain of the prion protein has an intrinsic tendency to self-associate and is able to destabilize an hydrophobic ±hydrophilic interface. The combination of these two properties might induce the exposure of the hydrophobic core of the protein to the extracellular milieu and perturb the folding of the protein. Whether this fragment can destabilize the threedimensional structure of the prion protein remains to be established. Tilted or oblique-orientated peptides can represent a simple and ef®cient way of inducing lipid bilayer destabilization (Brasseur, 1991; Brasseur et al., 1997). The fusogenic properties of the 118 to 135 prion-protein peptide described in this study might thus account for the cytotoxicity of the prion protein, or its degradation products, through a direct interaction with cellular membranes. Moreover, the presence of such a tilted peptide within the predicted neurotoxic domain of the prion protein might account for several biochemical properties of the infectious PrPSc. It remains to be established whether this fusogenic prion peptide, for which we demonstrated in vitro amyloidogenic properties, could be involved in the conformational transition of the PrPC to the PrPSc by facilitating the destabilization of the prion protein core.

Materials and Methods Materials L-a-Phosphatidylethanolamine (PE) L-a-phosphatidyl-L-serine (PS) from L-a-phosphatidylcholine (PC) from egg

from egg yolk, bovine brain, yolk, cholesterol and bovine serum albumin were purchased from Sigma. Bovine sphingomyelin (SM) was from Matreya Inc. (USA). 1-Palmitoyl-2-pyrene(14)-phosphatidylcholine (Pyr-PC) was a kind gift from Somerharju (Helsinki, Finland). All reagents for peptide synthesis and sequencing were purchased from Applied Biosystems. The tri¯uoroethanol (TFE) and hexa¯uoro-2-propanol (HFP) used for sample preparation were of the highest grade from Sigma. 20 ,70 -{bis(carboxymethyl)aminomethyl} ¯uorescein (calcein) was from Molecular Probes.

391

Fusogenic Properties of the 118±135 Prion Peptide Synthesis and purification of peptides Peptides were synthesized by the standard F-moc solid-phase method, on an Applied Biosystems model 431A peptide synthesizer. The peptide-resin conjugate was cleaved with tri¯uoroacetic acid, and the peptide was precipitated with tributylmethyl ether and recovered by centrifugation at 2000 g. The residue was dried for two hours in Speedvac concentrator (Savant Instr., Farmingdale NY). The peptides (20 to 40 mg) were resuspended in acetonitrile/water (8:10, v/v) to get rid of all scavenger molecules and the insoluble peptide was recovered by centrifugation. The washing step with water was repeated ten times. The purity and correct sequences of all peptides were veri®ed by electronspray ionisation mass spectrometry using a Fisons/VG Platform (Manchester, UK) mass. Molecular modeling of peptides by energy minimization Modelization of all peptides was carried out as previously described (Brasseur, 1991) and the method used is that applied to the study of polypeptide conformation (Brasseur et al., 1992). The method used to predict conformational structure of the peptides accounted for the contribution of the lipid ± water interface, the concomitant variation of the dielectric constant, and the transfer energy of atoms from a hydrophobic to hydrophilic environment (Brasseur et al., 1993). The structure, mode of insertion, and orientation of the peptides were studied in a dipalmitoylphosphatidylcholine molecular layer. In this model, the interaction energy (sum of contributions from van der Waals energy interactions, torsional potential energy, electrostatic interactions, and transfer energy between peptide and dipalmitoylphosphatidylcholine in the monolayer) was calculated and minimized until the lowest energy state of the entire aggregate was reached. All calculations were performed on an Olivetti CP586, by using PC-TAMMO‡ (Theoretical Analysis of Molecular Membrane Organization) and PC-PROT‡ (Protein Plus Analysis) software. Graphs were drawn with the PC-MGM‡ (Molecular Graphics Manipulation) program. Lipid-mixing experiments Small unilamelar vesicles (SUVs) were prepared from a mixture of PC/PE/PS/SM/cholesterol (10:5:7.5:7.5:16, w/w) when necessary in the presence of 2.5 mol% of Pyr-PC, as previously described (Pillot et al., 1996). All lipids were dissolved in chloroform at 10 mg/ml, and dried under a stream of nitrogen. Pyr-PC was dissolved in chloroform/ethanol at a concentration of 2 mM. Dried mixed lipids were hydrated in a 10 mM Tris-HCl buffer, pH 8.0, containing 150 mM NaCl, 0.1 g/l Na-EDTA and 1 mM NaN3. The lipid suspension was sonicated at 23 C, using a Branson soni®er, under nitrogen at 32 watts, four times for 15 minutes. After sonication, the labeled and unlabeled vesicles were applied to a Sepharose CL 4B column. SUVs were separated from larger particles eluting in the void volume of the column. In order to obtain an homogenous preparation, only the top fractions of the SUV elution peak were collected and pooled. Phospholipid concentration was determined by an enzymatic assay (BioMeÂrieux, France). Large unilamellar vesicles (LUVs) were prepared from phospholipids as follows. Dry lipids were hydrated in buffer and dispersed by vortexing to produce large multimellar vesicles. The lipid suspension was freeze-thawed ®ve times

and then extruded ten times through two stacked 100 nm pore size polycarbonate ®lters in a pressure extruder (Lipex Biomembranes, Inc., Vancouver, Canada). The LUV and SUV preparations could be stored at 4 C for at least one week, without signi®cant size change of the vesicles during storage. Fusion of pyrene-labeled SUVs together with unlabeled vesicles, at a 1/4 (w/w) ratio, was measured using a ¯uorescence probe dilution assay. The pyrene excimer/monomer intensity ratio was measured as a function of time after addition of increasing quantities of the peptides dissolved at a concentration of 1 mg/ml in either 50 or 20% (v/v) TFE. Fusion resulted in a decrease of the excimer intensity, and a slight increase of the monomer ¯uorescence. Emission spectra were obtained on an Aminco SF 500 spectro¯uorimeter at 25 C. The pyrene excimer/monomer (E/ M) ratio was calculated from the excimer and monomer ¯uorescence intensity at 475 and 398 nm, respectively, with an excitation wavelength of 346 nm. The excitation and emission band widths were set at 2 and 5 nm, respectively. Under these conditions, unloaded vesicles in the presence of the peptides gave no signi®cant signal due to light scattering. All experiments were performed in a 10 mM Tris-HCl buffer, pH 8.0, containing 150 mM NaCl, and 0.1 g/l Na-EDTA for a ®nal volume of 500 ml. The ®nal TFE concentration in the reaction mixture was less than 5% and a correction was made for the effect of the solvent on the excimer/monomer ratio. The ®nal peptide concentrations were in the range of 6.9 to 46 mM. Core-mixing experiments Vesicle fusion was also investigated by a core-mixing assay developed by Kendall & McDonald, 1982). Lipid vesicles were prepared as described above except that the phospholipid mixture was prepared by hydrating 2 mg of phospholipids in 1.5 ml of a 10 mM Tris-HCl buffer, pH 8.0, containing 150 mM NaCl, and 1 mM NaN3 and containing calcein at 8.0 mM plus CoCl2 at 1.0 mM or EDTA at 20 mM. Vesicles were then vigorously vortexed for two minutes and sonicated as previously described. Untrapped solutes were removed by two successive elutions on a Sephadex G-25 column with 10 mM Tris-HCl buffer, pH 8.0, containing 150 mM NaCl, and 1 mM NaN3. The lipid concentrations of the liposome suspensions were determined by phosphorus analysis. In a standard experiment, calcein, Co2‡- and EDTA-containing vesicles were mixed at a 1:1 molar ratio in a 10 mM Tris-HCl buffer, pH 8.0, containing 150 mM NaCl and 1 mM NaN3. Peptides from concentrated stock solutions were added at different concentrations and the calcein ¯uorescence was followed using an Aminco SPF 500 spectro¯uorimeter, with excitation and emission wavelengths of 490 and 520 nm, respectively, and with excitation and emission slits of 5 nm. The maximum ¯uorescence yield was determined in the presence of 0.5% Triton X-100 and 10 mM EDTA. In some experiments, CoCl2 at 0.4 mM chelated with 0.4 mM citrate were present in the reaction mixture to assess the leakage of encapsulated components. All experiments were performed under constant stirring at 25 C in the ¯uorimeter cuvette. Leakage of liposome contents The leakage of liposome contents to the external medium was kinetically monitored by measuring the release of calcein trapped inside the vesicles (Defrise-Quertain

392 et al., 1989). The SUVs were prepared as previously described except that dried lipids were rehydrated in a 10 mM Tris-HCl buffer, pH 8.0, containing 150 mM NaCl, 0.1 g/l Na-EDTA, 1 mM NaN3 and 40 mM calcein. At this concentration, ¯uorescence self-quenching occurs. Nontrapped calcein was removed by gel chromatography on Sephadex G-25. Vesicles were eluted with a 10 mM TrisHCl buffer, pH 8.0, containing 150 mM NaCl, 0.1 g/l NaEDTA, 1 mM NaN3. Calcein entrapped at a self-quenching concentration in SUVs, increases in ¯uorescence when it leaks from inside the liposomes. Upon addition of the peptides from a 50% (v/v) TFE stock solution, the increase in calcein ¯uorescence was followed, using an Aminco SPF 500 spectro¯uorimeter, with excitation and emission wavelengths of 490 and 520 nm, respectively, and with excitation and emission slits of 5 nm 100% leakage was established by lysis of the vesicles with 0.5% (w/ v) Triton X-100 (Defrise-Quertain et al., 1989). The percentage of calcein release is de®ned as: % F(t) ˆ (It ÿ I0/ Itot ÿ I0)  100, where I0ˆ initial ¯uorescence, Itotˆ total ¯uorescence observed after addition of Triton X-100, and It- ¯uorescence at time t. In our experimental conditions, in the absence of peptide, the spontaneous leakage rate was less than 1%/ten minutes. The volume of the reaction mixture was ®xed at 500 ml. All experiments were with constant stirring at 25 C. Lipid-binding experiments The ability of the prion peptides to associate with lipids, was assessed by gel ®ltration. Pyrene-labeled SUVs (3 mg phospholipid) were mixed with unlabeled vesicles (12 mg phospholipids) in a total volume of 500 ml and the lipid mixture was applied on a Sephadex G-25 column. The vesicles were eluted in a 10 mM Tris-HCl buffer, pH 8.0, containing 150 mM NaCl, 0.1 g/l NaEDTA, 1 mM NaN3 and 500 ml fractions were collected. For each fraction, the pyrene excimer/monomer (E/M) ratio was calculated from the excimer and monomer ¯uorescence intensity as described above. In separate experiments, 15 mg peptides dissolved in 50% TFE (v/v) were applied to the column and peptides were detected in the eluted fraction by monitoring the optical density at 280 nm and by measuring the tyrosine-¯uorescence emission using excitation and emission wavelengths of 280 and 305 nm, respectively. For lipid-binding experiments, SUVs and peptides were mixed together, incubated for 15 minutes at room temperature and the mixture was applied to the gel ®ltration column. Peptides and lipids were monitored as described above. Under our experimental conditions, SUVs had no effect on the absorbance of the peptides at 280 nm and the solvent used to prepare the peptide stock solutions had no in¯uence on the pyrene E/M intensity ratio. Aggregation conditions and thioflavin-T analysis All peptides were solubilized at 5 mg/ml in HFP and stored at ÿ80 C. For aggregation experiments, peptides were concentrated to dryness and resolubilized in a 10 mM Tris-HCl buffer, pH 8.0, containing 150 mM NaCl to a ®nal concentration of 250 mM. The peptide solutions were then ®ltered through 0.22 mM ®lter, and peptide concentrations were measured by amino acid analysis. Samples were brie¯y agitated prior to each measurement. Amyloid formation was quanti®ed using the thio¯avin-T ¯uorescence method (LeVine, 1993). Brie¯y, 10 ml of the aggregation reaction was added to 990 ml of

Fusogenic Properties of the 118 ±135 Prion Peptide 3 mM thio¯avin-T in 50 mM sodium phosphate buffer, pH 6.0. Immediately thereafter, ¯uorescence was read with excitation and emission wavelengths of 450 and 482 nm, respectively, and with excitation and emission slits of 5 and 10 nm, respectively. Electron microscopy Samples for electron microscopy (EM) were prepared by dissolving the peptides in 30% (v/v) formic acid and by stirring at room temperature for one hour. The peptide solutions were further dialyzed overnight at 4 C against distilled water. EM samples (10 ml) were applied to Formvar-carbon-coated grids, negatively stained with 2% (w/v) uranyl acetate, and examined by using a Zeiss 10C transmission electron microscope operating at 60 kV. Circular dichroism measurements To overcome problems of peptide solubility at high concentration, fresh peptide stock solutions (5 mg/ml) were prepared in HFP and peptides were further diluted in TFE. Circular dichroism spectra were obtained at 23 C in a Jasco 710 spectropolarimeter calibrated with 0.1% (w/v) d-10-camphorsulfonic acid solution (Brasseur et al., 1992). Portions of peptide stock solutions were removed and diluted in a 10 mM sodium phosphate buffer, pH 7.4, containing different percentages of TFE, SDS, sodium cholate or lipids. Peptide concentrations were measured by amino acid analysis. Nine spectra were recorded 20 minutes after dissolving the peptides at a ®nal concentration of 0.1 mg/ml. The percentages of secondary structure were estimated by curve-®tting on the entire ellipticity curve between 184 and 260 nm according to the variable selection procedure developed by Johnson (1990).

Acknowledgements We wish to thank H. Caster and J. Taveirne for excellent technical assistance. This work was supported by grant G006196 from the Belgian National Funds for Scienti®c Research and by BioMed2 Program CT 898.96.

References Arispe, N., Pollard, H. B. & Rojas, E. (1996). Zn2‡ interaction with Alzheimer amyloid b protein calcium channels. Proc. Natl Acad. Sci. USA, 93, 1710± 1715. Brasseur, R. (1991). Differentiation of lipid-associating helices by use of three-dimensional molecular hydrophobicity potential calculation. J. Biol. Chem. 266, 16120± 16127. Brasseur, R., De Meutter, J., Vanloo, B., Goormaghtigh, E., Ruysschaert, J. M. & Rosseneu, M. (1992). Mode of assembly of amphipathic helical segments in model high density lipoproteins. Biochim. Biophys. Acta, 1043, 245±252. Brasseur, R., Vanloo, B., Deleys, B., Lins, L., Labeur, C., Taveirne, J., Ruysschaert, J. M. & Rosseneu, M. (1993). Design and properties of synthetic model peptides for the amphipathic helices of plasma apolipoproteins. Biochim. Biophys. Acta, 1170, 1 ± 7. Brasseur, R., Pillot, T., Lins, L., Vandekerckhove, J. & Rosseneu, M. (1997). Peptides in membrane: tipping

Fusogenic Properties of the 118±135 Prion Peptide the balance of membrane stability. Trends Biochem. Sci. 22, 167± 171. Brown, D. R., Schmidt, B. & Kretzschmar, H. A. (1996). Role of microglia and host prion protein in neurotoxicity of a prion protein fragment. Nature, 380, 345± 347. Chen, S. G., Teplow, D. B., Parchi, P., Teller, J. K., Gambetti, P. & Autilio-Gambetti, L. (1995). Truncated forms of the human prion protein in normal brain and in prion diseases. J. Biol. Chem. 270, 19173± 19180. Come, J. H., Fraser, P. E. & Lansbury, P. T. (1993). A kinetic model for amyloid formation in the prion diseases: importance of seeding. Proc. Natl Acad. Sci. USA, 90, 5959± 5963. Defrise-Quertain, F., Cabiaux, V., Vandenbranden, M., Wattiez, R., Falmagne, P. & Ruysschaert, J. M. (1989). pH-dependent bilayer destabilization and fusion of phospholipidic large unilamellar vesicles induced by diphteria toxin and its fragments A and B. Biochemistry, 28, 3406± 3413. Evans, K. C., Berger, E. P., Cho, C. G., Weisgraber, K. H. & Lansbury, P. T. (1995). Apolipoprotein E is a kinetic but not a thermodynamic inhibitor of amyloid formation: implications for the pathogenesis and treatment of Alzheimer's disease. Proc. Natl Acad. Sci. USA, 92, 763± 767. Forloni, G., Angeretti, N., Chiesa, R., Monzani, E., Salmona, M., Bugiani, O. & Tagliavini, F. (1993). Neurotoxicity of a prion protein fragment. Nature, 362, 543 ± 546. Gray, C., Tatulian, S. S., Wharton, S. A. & Tamm, L. K. (1996). Effect of the N-terminal glycine on the secondary structure, orientation of the in¯uenza hemagglutinin fusion peptide with lipid bilayers. Biophys J. 70, 2275± 2286. Heller, J., Kolbert, A. C., Larser, R., Ernst, M., Bekker, T., Baldwin, M. A., Prusiner, S. B., Pines, A. & Wemmer, D. E. (1996). Solid-state NMR studies of the prion protein H1 fragment. Protein Sci. 5, 1655± 1661. Horth, M., Lambrecht, B., Chuah Lay Khim, M., Bex, M., Thiriart, C., Ruysschaert, J.-M., Burny, A. & Brasseur, R. (1991). Theoretical and functional analysis of the SIV fusion peptide. EMBO J. 10, 2747± 2755. Huang, Z., Prusiner, S. B. & Cohen, F. E. (1995). Scrapie prions: a three-dimensional model of an infectious fragment. Folding Design, 1, 13 ± 19. Jarrett, J. T. & Lansbury, P. T. (1993). Seeding `onedimensional crystallization' of amyloid: a pathogenic mechanism in Alzheimer's disease and scrapie?. Cell, 73, 1055± 1058. Johnson, W. C., Jr (1990). Protein secondary structure and circular dichroism: a practical guide. Proteins: Struct. Funct. Genet. 7, 205± 214. Kendall, D. A. & McDonald, R. C. (1982). A ¯uorescence assay to monitor vesicle fusion and lysis. J. Biol. Chem. 257, 13892± 13895. Lambert, G., Decout, A., Vanloo, B., Rouy, D., Duverger, N., Kalopissis, A., Vandekerckhove, J., Chambaz, J., Brasseur, R. & Rosseneu, M. (1997).

393 The C-terminal helix of human apolipoprotein A-II promotes the fusion of unilamellar liposomes and displaces apolipoprotein A-I from high density lipoproteins. Biochemistry. In the press. Lehman, S. & Harris, D. A. (1996). Mutant and infectious prion proteins display common biochemical properties in cultured cells. J. Biol. Chem. 271, 1633± 1637. LeVine, H. (1993). Thio¯avine T interaction with synthetic Alzheimer's disease b-amyloid peptides: detection of amyloid aggregation in solution. Protein Sci. 2, 404 ±410. Lin, M. X., Mirzabekov, T. & Kagan, B. L. (1997). Channel formation by a neurtoxic prion protein fragment. J. Biol. Chem. 272, 44± 47. Morris, S. J., Bradley, D., Gibson, C. C., Smith, P. D. & Blumenthal, R. (1988). Use of membrane-associated ¯uorescence probes to monitor fusion of bilayer vesicles: application to rapid kinetics using pyrene excimer/monomer ¯uorescence. In Spectroscopic Membrane Probes (Loem, L. M., ed.), pp. 161± 191, CRC Press, Boca Raton, USA. Oesch, B., Jensen, M., Nilsson, P. & Fogh, J. (1994). Properties of the scrapie prion protein: quantative analysis of protease resistance. Biochemistry, 33, 5926± 5931. Perini, F., Vidal, R., Ghetti, B., Tagliavini, F., Frangione, B. & Prelli, F. (1996). PrP27-30 is a normal soluble prion protein fragment released by human platelets. Biochem. Biophys. Res. Commun. 223, 572± 577. Pillot, T., Goethals, M., Vanloo, B., Talussot, C., Brasseur, R., Rosseneu, M., Vandekerckhove, J. & Lins, L. (1996). Fusogenic properties of C-terminal domains of the Alzheimer beta-amyloid peptide. J. Biol. Chem. 271, 28757± 28765. Pillot, T., Brasseur, R., Vandekerchove, J. & Rosseneu, M. (1997). Apolipoprotein E inhibits liposome fusion induced by the Alzheimer's beta-amyloid peptide. Nutr. Metab. Cadiovasc. Dis. 7, 276± 280. Prusiner, S. B. (1994). Inherited prion diseases. Proc. Natl Acad. Sci. USA, 91, 4611± 4614. Rapaport, D. & Shai, Y. (1994). Interaction of ¯uorescently labeled analogues of the amino-terminal fusion peptide of Sendai Virus with phospholipid membranes. J. Biol. Chem. 269, 15124± 15131. Riek, R., Hornemann, S., Wider, G., Billeter, M., Glockshuber, R. & WuÈthrich, K. (1996). NMR structure of the mouse prion protein domain PrP (121231). Nature, 382, 180± 182. Rizzardini, M., Chiesa, R., Angeretti, N., Lucca, E., Salmona, M., Forloni, G. & Cantoni, L. (1997). Prion protein fragment 106-126 differentially induces heme oxygenase I mRNA in cultured neurons and astroglial cells. J. Neurochem. 68, 715± 720. Wood, S. J., Maleeff, B., Hart, T. & Wetzel, R. (1996). Physical, morphological and functional differences between pH 5.8 and 7.4 aggregates of the Alzheimer's amyloid peptide Ab. J. Mol. Biol. 256, 870 ± 877. Zimmerberg, J., Vogel, S. S. & Chernomordik, L. V. (1993). Mechanism of membrane fusion. Annu. Rev. Biophys. Biomol. Struct. 22, 433± 466.

Edited by I. B. Holland (Received 27 May 1997; received in revised form 28 August 1997; accepted 29 August 1997)