Properties of Scrapie Prion Protein Liposomes*

Purified scrapie prions contain one identifiable mac- romolecule, PrP 27-30, which polymerizes into rod-shaped amyloids. The rods can be dissociated with retention of scrapie infectivity upon incorporation of PrP 27-30 into detergent-lipid-protein complexes (DLPC) as well as liposomes. As measured by end-point titration, scrapie infectivity was increased >lOO-fold upon dissociating the rods into liposomes. The incor- poration of PrP 27-30 into liposomes was demonstrated by immunoelectron microscopy using colloidal gold. Detergent extraction of prion liposomes followed by chloroform/methanol extraction resulted in the reappearance of rods, indicating that this process is reversible. Scrapie prion infectivity in rods and liposomes was equally resistant to inactivation by irradi- ation at 254 nm and was unaltered by exposure to nucleases. A variety of lipids used for producing DLPC and liposomes did not alter infectivity. Fluorescently labeled PrP 27-30 in liposomes was used to study its entry into cultured cells. Unlike the rods which re- mained as large fluorescent extracellular masses, the PrP 27-30 in liposomes rapidly entered the cells and was seen widely distributed within the interior of the cell. PrP 27-30 is derived by limited proteolysis from a larger protein designated PrPs’ which is membrane bound. PrPsc in membrane

Purified scrapie prions contain one identifiable macromolecule, PrP 27-30, which polymerizes into rodshaped amyloids. The rods can be dissociated with retention of scrapie infectivity upon incorporation of PrP 27-30 into detergent-lipid-protein complexes (DLPC) as well as liposomes. As measured by end-point titration, scrapie infectivity was increased >lOO-fold upon dissociating the rods into liposomes. The incorporation of PrP 27-30 into liposomes was demonstrated by immunoelectron microscopy using colloidal gold. Detergent extraction of prion liposomes followed by chloroform/methanol extraction resulted in the reappearance of rods, indicating that this process is reversible. Scrapie prion infectivity in rods and liposomes was equally resistant to inactivation by irradiation at 254 nm and was unaltered by exposure to nucleases. A variety of lipids used for producing DLPC and liposomes did not alter infectivity. Fluorescently labeled PrP 27-30 in liposomes was used to study its entry into cultured cells. Unlike the rods which remained as large fluorescent extracellular masses, the PrP 27-30 in liposomes rapidly entered the cells and was seen widely distributed within the interior of the cell. PrP 27-30 is derived by limited proteolysis from a larger protein designated PrPs' which is membrane bound. PrPsc in membrane fractions was solubilized by incorporation in DLPC, thus preventing its aggregation into amyloid rods. The functional solubilization of scrapie prion proteins in DLPC and liposomes offers new approaches to the study of prion structure and the mechanism by which they cause brain degeneration.
Scrapie is a transmissible, degenerative neurological disease of sheep and goats. The clinical signs of neurologic dysfunction become manifest after a prolonged incubation period (Eklund et al., 1967;Gajdusek, 1977;Prusiner, 1982b). The scrapie agent, in contrast to viruses, resists inactivation by procedures that modify or hydrolyze nucleic acids (Alper et *This work was supported by Research Grants, AGO2132 and NS14069 from the National Institutes of Health and Senator Jacob Javits Center of Excellence in Neuroscience NS22786 Grant as well as by gifts from Sherman Fairchild Foundation, FUR-Nabisco, Inc., Armstrong McDonald Foundation, Inc., and Bernard Osher Philanthropic Fund. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  Diener et al., 1982;Prusiner, 1982); however, it is sensitive to reagents which modify proteins (Prusiner et al., 1981;McKinley et al., 1983;Prusiner et al., 1983). The requirement of a protein for infectivity and the unusual properties of the scrapie agent prompted the introduction of the term "prion" to denote this novel class of infectious pathogens (Prusiner, 1982(Prusiner, , 1987Diener, 1987). The only identifiable macromolecule of purified preparations of scrapie prions is a protein, designated PrP 27-30 (Bolton et al., 1982;Prusiner et al., 1982;McKinley et al., 1983). This protein is derived by proteinase K digestion from a larger molecule of apparent molecular weight 33,000-35,000, denoted PrPS' (Oesch et al., 1985;Barry et al., 1986;Meyer et al., 1986). Both PrP 27-30 and PrPSc are membrane bound, aggregate into rod-shaped polymers upon detergent extraction, and can be purified as aggregates Meyer et al., 1986).
Recently, we described the functional solubilization of purified PrP 27-30 from the rod-shaped polymers into detergent-lipid-protein complexes (DLPC)' and liposomes (Gabizon et al., 1987). PrP 27-30 in rods was mixed with phosphatidylcholine and 2% sodium cholate, vortexed, and sonicated to form DLPC. The 100,000 x g supernatant of this mixture contained most of the PrP 27-30 demonstrating that the protein had been solubilized. After removing the detergent by dialysis, PrP 27-30 was found incorporated into liposomes. Both the DLPC and liposomes exhibited scrapie prion infectivity.
In this report, we extend our initial observations on prion DLPC and liposomes. The properties of these novel forms of the prion are described and compared to those previously reported for the rods which possess the properties of amyloid McKinley et al., 1986). In contrast to liposomes which are closed spherical vesicles, DLPC are small fragments of these phospholipid vesicles. Because of their small size, the DLPC were used for all studies except those where detergents interfered with the experimental protocols such as immunoelectron microscopy and cell fusion. In these studies, liposomes were formed upon removal of the detergent by dialysis. The formation of rods, DLPC, and liposomes is a reversible process with all forms being infectious. These properties of prions are unlike those of viruses and are most compatible with a one-component system.

MATERIALS AND METHODS'
The abbreviations used are: DLPC, detergent-lipid-protein complexes; PC, phosphatidylcholine: SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; FITC, fluorescein isothiocyanate; Sarkosyl, sodium dodecyl sarcosinate; PMSF, phenylmethylsulfonyl fluoride. * Portions of this paper (including "Materials and Methods," part of "Results," Figs. 2-8, and Tables I and 11) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.

RESULTS
The studies described in a previous communication (Gabizon et al., 1987) and the Miniprint section of this paper have used PrP 27-30 which is purified as insoluble rod-shaped polymers. Subsequent PrP 27-30 disaggregation without denaturation was accomplished by a combination of detergent and phospholipids which led to the formation of DLPC. PrP 27-30 is derived from a larger molecule, PrPk, during purification which utilizes limited proteinase K digestion. This digestion hydrolyzes the N-terminal 67-amino acid residues (Prusiner et al., 1984a;Basler et al., 1986). PrP+ is a membrane-bound protein which polymerizes upon detergent extraction (Meyer et al., 1986).
The detergent solubilization of PrPC bound to membranes  Laemmli (1970). Prior to electrophoresis, selected fractions were digested (+) with proteinase K (PK) in order to distinguish the scrapie PrP isoform from the cellular one. After electrotransfer from the electrophoretic gel to nitrocellulose, the blot was stained with rabbit antisera raised against purified hamster brain PrP 27-30. Molecular weight markers are given in kilodaltons. from uninfected hamster brains was included as a control (supernatant fluid, lanes 5 and 6; pellet, lanes 11 and 12).

Digestion of PrPC by proteinase K is shown in lane 5.
The formation of DLPC by addition of Sarkosyl and phosphatidylcholine to scrapie-infected microsomal membranes did not alter the scrapie infectivity titer. The microsomal membrane fraction had a titer of lo8.' ID, units/ml; after detergent extraction in the presence of phospholipid followed by ultracentrifugation, the titer of the supernatant fraction was lo".* ID, units/ml.

DISCUSSION
For many years, investigators reported the membranebound nature of the scrapie agent but were unable to solubilize the infectious entity (Hunter, 1972(Hunter, , 1979Hunter et al., 1974). In fact, the intimate association of scrapie infectivity with membranes gave rise to the "membrane hypothesis" (Hunter et al., 1967) and to the notion that the scrapie agent was inseparable from cellular membranes, and thus, could not be purified (Hunter et dl., 1971). Subsequently, purification schemes, based upon the detergent-induced polymerization of scrapie prions into rod-shaped amyloids, were developed by us (Prusiner et al., 1980b(Prusiner et al., , 1981 and modified by others (Diringer et al., 1983a;Hilmert and Diringer, 1984;Hope et al., 1986).
Although the rod-shaped scrapie prion polymers could be dissociated, this required denaturation of the component protein, PrP 27-30 or PrP%, until recently. Denaturation of PrP 27-30 was registered by conversion from a protease-resistant to a sensitive form which is accompanied by a loss of scrapie infectivity; this diminution of infectivity greatly impeded characterization of the prion particle (Bolton et al., 1984). Dissociation of the prion rods by a combination of detergent and phospholipid was a significant advance since infectivity was retained under these conditions (Gabizon et al., 1987).
We now report the extraction of PrP-and scrapie infectivity from membranes directly into DLPC without aggregation into rods as an intermediate step (Fig. 1). The development of a protocol for solubilization of scrapie prions should facilitate the development of new methods for purification and characterization of these novel infectious pathogens.
The number of hypotheses which can be seriously considered to explain the molecular structure of the infectious scrapie agent have been significantly constrained by recent studies (Diener et al., 1982;Oesch et al., 1985;Carlson et al., 1986;Gabizon et al., 1987;Prusiner, 1987;Westaway et al., 1987). The studies reported here and elsewhere have made the possibility that scrapie is caused by a virus seem remote and have forged a strong case for PrP% being an integral and necessary component of the infectious particle (Diener et al., 1982;McKinley et al., 1983;Oesch et al., 1985;Gabizon et al., 1987;Diener, 1987).
Our studies show that morphological transformation of the scrapie PrP isoform from a rod-shaped polymer to a liposome is accompanied by a significant increase in infectivity. This increase in biological activity is generally >10-fold as assessed by incubation time bioassay and -100-fold as reported here as measured by end-point titration (Fig. 3). That PrP 27-30 is actually transferred into the liposomes during disruption of the rods is well documented by immunoelectron microscopy. Fig. 2 shows that the liposomes acquire antigens which bind PrP 27-30 antisera as registered by the attachment of colloidal gold second antibodies. The dispersion of PrP 27-30 was indirectly demonstrated by fusion of the prion liposomes to cultured cells (Fig. 8). Fluorescein isothiocyanate (F1TC)labeled rods were seen as clumps adjacent to cells while prion

Properties of Scrapie Prion
Protein Liposomes liposomes fused to cells resulted in the transfer of fluorescent PrP 27-30 to the interior of the cells. It will be of interest to learn whether cultured cells can be more efficiently infected with prions using liposomes rather than rods. In our initial study of prion phospholipid vesicles, we reported the resistance of scrapie infectivity in DLPC and liposomes to inactivation by nucleases and Zn++ (Gabizon et al., 1987) in accord with other studies using purified prion rods (Bellinger-Kawahara et al., 198713). Those results have been extended by adding nucleases to the rods and then dissociating the rods into DLPC in the presence of the nucleases (Table I). Again, no change in scrapie infectivity was found. Irradiation at 254 nm of the DLPC produced an inactivation curve virtually identical to that observed for the rods in the experiment reported here and earlier studies (Fig.  4) (Bellinger-Kawahara et al., 1987a). The resistance of scrapie infectivity to inactivation by irradiation at 254 nm in preparations of purified prion rods suggests that if prions contain a nucleic acid, it will be -5 bases in length for a single-stranded molecule or 30-45 base pairs for a doublestranded molecule (Bellinger-Kawahara et al., 1987a). The validity of these earlier studies is strengthened by the observations reported here; the same results were obtained after prions were transformed from rods into DLPC. The DS7 value for prion DLPC is in good agreement with values reported two decades earlier for murine scrapie agent in brain homogenates (Alper et al., 1967).
In contrast to resistance of prion rods and DLPC to procedures that modify nucleic acids, prolonged exposure of the rods and DLPC to proteolytic digestion resulted in a significant decrease in scrapie infectivity (Fig. 5). Inactivation of prion rods and DLPC by proteinase K was both a function of the time of digestion and the concentration of protease as reported earlier for both partially purified (Prusiner et al., 1981) and more extensively purified samples . Our experiments clearly demonstrate that the resistance of PrP 27-30 to enzyme-catalyzed proteolysis is an intrinsic property of the scrapie PrP isoform and not a consequence of its polymerization into amyloid rods.
We do not understand the molecular basis for solubilization of PrP& by a combination of detergent and phospholipid. No detergent alone or in combination with other detergents has been identified which solubilizes the majority of either PrPS' or scrapie infectivity (Millson and Manning, 1979;Prusiner et al., l978,1980aPrusiner et al., l978, , 1980bDiringer et al., 1983b). Recent studies on the aspartate receptor show a similar phenomenon where a combination of detergent and phospholipid solubilized this protein in a biologically active form while detergents alone did not (Bogonez and Koshland, 1985). Many transport proteins must be solubilized in the presence of lipids in order to isolate and reconstitute them in a functionally active state (Maron et al., 1979;Barzilai et al., 1984;Braiman et al., 1987;Moriyama and Nelson, 1987).
The studies presented here and elsewhere (Meyer et al., 1986;Gabizon et al., 1987) suggest that a complex equilibrium may exist between the membrane-bound and rod polymer forms of PrPS'. This putative equilibrium appears to be influenced by the relative concentrations of detergent and lipid. Detergent extraction of membranes from scrapie-infected brains results in the formation of rods. The rods can be dissociated by phospholipids in the presence of detergent; from the resulting DLPC, the rods can be reformed by raising the detergent level (Fig. 6). In its most simple representation, the system behaves as if it is an equilibrium between the membrane-bound and rod forms of PrPSc that can be de-scribed by the following equation:

r P s ] . [lipid] [detergent]
where PrP%,b, is membrane-bound PrPSc, K is the equilibrium constant, and PrPzd is the amyloid rod form of PrPSc. A process similar to that described by the above equation appears to occur i n vivo where aggregated PrPS" found in scrapie-infected brains in the form of amyloid filaments is probably a consequence of cell death which releases PrPS" into the extracellular space where it polymerizes into filaments (Bendheim et al., 1984;DeArmond et al., 1985). Even in advanced disease, most of the PrPSc appears to be membrane bound (Meyer et al., 1986, suggesting that the membrane form of PrP" may be important with respect to transmitting disease from cell to cell (Hay et al., 1987b;Prusiner and DeArmond, 1987).
While PrPC has been shown to be localized almost exclusively to the external surface of cells where it is anchored by a phosphatidylinositol glycolipid (Stahl et al., 1987), the topology of PrPS" is under investigation. Studies on PrP 27-30 have shown that, like PrPC, it contains a phosphatidylinositol glycolipid. Interestingly, PrP contains a hydrophobic domain of sufficient length to span the membrane (Bazan et al., 1987). This domain was found buried within the lipid bilayer in cellfree translation studies (Hay et al., 1987a). A second form of PrP presumed to be secretory was found within the interior of microsomal vesicles used in cell-free translation studies (Hay et al., 1987b). Although some portion of PrP 27-30 is on the surface of the liposomes (Fig. 2), the topology of PrP 27-30 or PrPSc within liposome membranes remains to be established.