Reconstitution of Prion Infectivity from Solubilized Protease-resistant PrP and Nonprotein Components of Prion Rods*

The scrapie isoform of the prion protein, PrPSc, is the only identified component of the infectious prion, an agent causing neurodegenerative diseases such as Creutzfeldt-Jakob disease and bovine spongiform encephalopathy. Following proteolysis, PrPSc is trimmed to a fragment designated PrP 27–30. Both PrPSc and PrP 27–30 molecules tend to aggregate and precipitate as amyloid rods when membranes from prion-infected brain are extracted with detergents. Although prion rods were also shown to contain lipids and sugar polymers, no physiological role has yet been attributed to these molecules. In this work, we show that prion infectivity can be reconstituted by combining Me2SO-solubilized PrP 27–30, which at best contained low prion infectivity, with nonprotein components of prion rods (heavy fraction after deproteination, originating from a scrapie-infected hamster brain), which did not present any infectivity. Whereas heparanase digestion of the heavy fraction after deproteination (originating from a scrapie-infected hamster brain), before its combination with solubilized PrP 27–30, considerably reduced the reconstitution of infectivity, preliminary results suggest that infectivity can be greatly increased by combining nonaggregated protease-resistant PrP with heparan sulfate, a known component of amyloid plaques in the brain. We submit that whereas PrP 27–30 is probably the obligatory template for the conversion of PrPCto PrPSc, sulfated sugar polymers may play an important role in the pathogenesis of prion diseases.

PrP Sc , the abnormal isoform of PrP C , is the only known component of the prion, an agent causing fatal neurodegenerative disorders such as bovine spongiform encephalopathy and Creutzfeldt-Jakob disease (1). It has been postulated that prion diseases propagate by the conversion of PrP C molecules into proteaseresistant and insoluble PrP Sc molecules by a mechanism in which PrP Sc serves as a template (2). Whereas some PrP Sc may be insoluble in vivo (3), it is well documented that most PrP Sc , as well as its protease-resistant core denominated PrP 27-30, precipitate into insoluble aggregates (also known as prion rods) when membranes from scrapie-infected brains are extracted with detergents such as sarkosyl (4). In addition to PrP Sc , prion aggregates were shown to contain nonprotein components, which include sphingolipids as well as polysaccharides (5)(6)(7). The traces of nucleic acids present in prion rods are believed to be too small to function as coding tools (8).
No physiological role has ever been attributed to any nonprotein components of prion rods.
Disruption of prion rods into detergent protein lipid complexes resulted in the retention of their protease resistance property concomitantly with an increase in their prion infectivity, suggesting that solubilized PrP Sc is more infectious than the aggregated prion protein (9). Contrarily, disruption of prion rods by sonication and SDS resulted in a protease-sensitive PrP with complete loss of infectivity (10).
As opposed to methods to disrupt prion aggregates, we have recently introduced a new experimental procedure that results in the production of nonaggregated PrP Sc or PrP 27-30 molecules by inhibition of the primary detergent-induced aggregation used for rod formation (11). When membranes from brains of hamsters terminally ill with scrapie were incubated in the presence of Me 2 SO in addition to sarkosyl and subsequently applied to a sucrose density gradient, the protease-resistant PrP molecules (PrP [27][28][29][30] were divided between the light fractions, containing soluble or poorly aggregated PrP 27-30 molecules, and the heaviest fractions, containing insoluble and heavily aggregated PrP 27-30 molecules (11). Interestingly, when light and heavy fractions of such gradients, containing similar concentrations of protease-resistant PrP, were inoculated into hamsters, the infectivity of the light fractions was lower by more than 2 logs than the infectivity of the heavy fractions. Light fractions produced in parallel in the absence of Me 2 SO, which did not contain any detectable PrP, presented the same low infectivity as the Me 2 SO light fractions, suggesting that this residual infectivity was not due to the presence of Me 2 SO-solubilized PrP 27-30 molecules. We attribute the low infectivity present in both light fractions to the fact that the brain extracts were applied to the sucrose gradient from the top, and therefore some small prion aggregates, containing undetectable PrP 27-30, may not have sedimented.
In this work, we investigated whether molecules other than protease-resistant PrP might have a physiological role in prion infectivity. To this effect, we combined the low infectious protease-resistant Me 2 SO-solubilized PrP described above with nonprotein components that remain in prion aggregates subsequent to denaturation and harsh protease digestion (NPH Sc ). 1 In some experiments we substituted the NPH Sc frac-* 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.
¶ Supported by a grant from the Israel Center for the Study of Emerging Diseases.
ʈ Supported by grants from the European Community and by the Agnes Ginges Center for Human Neurogenetics. To whom correspondence should be addressed: Dept. of Neurology, Hadassah University Hospital, Jerusalem, Israel 91120. Fax: 972-2-6429441; E-mail: gabizonr@hadassah.org.il. 1 The abbreviations used are: NPH, nonprotein heavy; NPH Sc , NPH fraction originating from a scrapie-infected hamster brain; PK, proteinase K; L, light; H, heavy; GndSCN, guanidium thiocyanate; H Sc , heavy fraction originating from a scrapie-infected hamster brain; NPH N , heavy fraction after deproteination, originating from a normal hamster brain; L Sc , light fraction from a scrapie-infected hamster brain; L Sc Me2SO , light fraction from a scrapie-infected hamster brain solubilized with Me 2 SO. tion for heparan sulfate, a known component of prion rods. Our results show that the addition of the deproteinized sedimented fraction (NPH Sc ) to low infectious solubilized protease-resistant PrP restores the prion infectivity to its original values. We therefore propose that in addition to PrP Sc , prion infectivity may depend upon, or at least be largely facilitated by, the presence of other components of prion rods.

EXPERIMENTAL PROCEDURES
Sucrose Gradients-Three hundred l of 10, 15, 20, 25, 30, and 60% sucrose in phosphate-buffered saline were loaded into centrifuge tubes adapted for the TLS-55 rotor of the TL 100 ultracentrifuge (Beckman Instruments) to form a zonal gradient. Normal or scrapie brain membranes (25 l containing 15 g/ml protein) were diluted with STE buffer (100 mM NaCl, 10 mM Tris, pH 7.4, 1 mM EDTA) containing 2% sarkosyl to a final volume of 240 l. When appropriate, Me 2 SO (10%) was added to the brain extract and incubated for 16 h at 4°C before the extract was loaded on top of the gradient and centrifuged in a TLS-55 tube ultracentrifuge (Beckman) at 55,000 rpm (g av ϭ 100,000) for 1 h at 20°C. Before immunoblotting with anti-PrP monoclonal antibody 3F4 (12), gradient samples of equal volume were collected and digested with proteinase K (PK, 40 g/ml) for 60 min at 37°C.
Sample Preparation-The three top fractions of each gradient were pooled and denominated fraction L (light). The three pooled bottom fractions (H, heavy) were tested either directly or after a deproteination treatment that included the following: 1) denaturation with 4.5 M guanidium thiocyanate (GndSCN) (final concentration) for 15 min, 2) precipitation with methanol to wash out the GndSCN, 3) resuspension in 2% sarkosyl STE buffer before digestion with PK (100 g/ml for 60 min at 37°C) to form NPH (non protein heavy) fractions. To form the combined fractions, original samples (L, NPH, or HS (heparan sulfate)) were mixed at equal volumes and incubated for 16 h at 4°C. Original samples as well as combinations (detailed in Fig. 2) were assayed for the presence of PrP and infectivity. All volumes of original samples were adjusted before inoculation to contain the same concentration of L or NPH samples present in the mixtures.
Heparan Sulfate-400 l of the L Sc Me2SO sample were incubated with bovine kidney heparan sulfate (Sigma) for 16 h (4 mg/ml) at 25°C. The resulting fraction was denominated HS/L Sc Me2SO . Heparanase Digestion-400 l of the NPH Sc sample were precipitated in methanol and resuspended in STE buffer before the addition of 25 units/ml of heparinase III (Sigma) for 16 h at 37°C. The resulting sample was denominated NPH*.
In Vivo Infectivity Experiments-Five male Syrian hamsters, 4 weeks old, were inoculated intracerebrally with 50 l of each of the samples to be tested for prion infectivity. Animals were tested daily. Prion titers were measured by monitoring the incubation period until the appearance of symptoms (13). Before inoculation into hamsters, samples containing only L or H fractions were supplemented with 2% sarkosyl to retain similar concentrations of the components in the infectivity assay.
Immunoblotting of Brain PrP Sc -10% (w/v) of brain tissue from FIG. 1. Solubilization of PrP 27-30 by Me 2 SO. a, membranes from scrapie-infected brains extracted with 2% sarkosyl in the presence or absence of 10% Me 2 SO or from normal brain extracted with sarkosyl only were subjected to a sucrose density gradient (10 -60%) as described (11). Fractions collected from the gradients were digested in the presence and absence of PK (40 g/ml for 60 min at 37°C) and immunoblotted with monoclonal antibody 3F4. The three pooled bottom fractions (H) were tested either directly or after a deproteination treatment that included the following: 1, denaturation with 4.5 M GndSCN (final concentration) for 15 min; 2, precipitation with methanol to wash out the GndSCN; 3, resuspension in 2% sarkosyl STE buffer before digestion with PK (100 g/ml for 60 min at 37°C) to form NPH fractions. When applicable, NPH samples were digested by heparanase (25 units/ml of heparinase III (Sigma) for 16 h at 37°C). To form the combined fractions, original samples (L, NPH, or H) were mixed at equal volumes and incubated for 16 h at 4°C. The HS concentration was 4 mg/ml. Original samples as well as combinations (detailed in Fig. 1) were assayed for the presence of PrP, infectivity, and structure by electron microscopy. NPH Sc * , NPH Sc sample digested with heparanase; L N , light fraction from a normal hamster brain. ; 8, NPH Sc /L Sc ; 9, NPH Sc * / L Sc Me2SO . All samples were digested prior to immunoblotting with 40 g/ml for 60 min at 37°C. scrapie-infected hamsters (frozen at Ϫ80°c following flash freezing in liquid nitrogen) was homogenized in cold sucrose buffer (10 mM Tris, 0.3 M sucrose in phosphate-buffered saline). 2% sarkosyl was added to 50-l samples before digestion with 40 g/ml PK for 60 min at 37°C.
Statistical Analysis-To compare the study groups, Anova and the nonparametric Kruskal-Wallis test were applied. In addition, multiple pairwise comparisons were performed using the Dunnett and Scheffe methods. The tests were performed using the SPSS for Windows computer program.

RESULTS
Brain membranes from scrapie-infected and uninfected hamsters were extracted with sarkosyl in the presence and absence of Me 2 SO and, following an overnight incubation, applied to a 10 -60% sucrose gradient as described (11). Gradient fractions were digested with PK and immunoblotted with ␣PrP monoclonal antibody 3F4 (Fig. 1a). As can be seen in the figure, the light fraction prepared in the presence of Me 2 SO contained a considerable fraction of the total PrP 27-30. To assure that the concentration of PrP 27-30 present in the light fractions obtained with Me 2 SO was significantly higher than that present in the light fraction without Me 2 SO, we immunoblotted with an anti-PrP antibody several 10-fold dilutions of the Me 2 SO scrapie light fraction. As can be seen in Fig. 1b, the concentration of PrP 27-30 in the Me 2 SO light fractions was at least 1000 times larger than its concentration in the light fraction produced without Me 2 SO. Fig. 2 presents the organization of the reconstitution experiments and the samples used for infectivity assays. The three first fractions (of 12), as well as the three last fractions of each sucrose gradient, were pooled and denominated L and H fractions, respectively. All gradient fractions were digested with 40 g/ml PK at 37°C for 60 min. Part of each H fraction was totally denatured with 4.5 M GndSCN and, after methanol precipitation, digested again with PK to produce the NPH fraction. Subsequently, L and NPH fractions from different sources (normal and scrapie-infected with and without Me 2 SO) were used to create the combined samples specified in Fig. 2.
All samples to be evaluated for infectivity were precipitated by methanol (to remove traces of Me 2 SO) and resuspended into inoculation buffer (1% bovine serum albumin in phosphatebuffered saline) to contain PrP 27-30 at comparable concentrations (Fig. 3a). No protease-resistant PrP was detected in the NPH Sc samples following the denaturation/protease digestion treatment (Fig. 3a, lane 2). PrP 27-30 was also absent from the light samples produced in the absence of Me 2 SO, as well as in the NPH Sc /L Sc sample (Fig. 3a, lanes 4 and 8). When the bioassays were completed, we also tested the brains of the animals inoculated with each sample for the concentration and electrophoretic pattern of PrP after PK digestion (Fig. 3b). Although incubation times for the different samples varied widely (see Table I), the concentration of PrP 27-30, as well as the banding pattern of the protein, was the same regardless of the inocula administered to the hamsters. Histoblot analysis of all brains were also identical (data not shown). This suggests that the manipulations performed in this work did not produce a new prion strain. Although it was repeatedly shown that different strains of prions can be characterized by these parameters (14 -18), end point titration analyses are required to prove this point conclusively.
All samples described in Fig. 2 (original and combined) were bioassayed for prion infectivity (Fig. 4 and Tables I and II).   Table II presents the significance of the variability among the treatment groups (p values). Although p Ͻ 0.05 (*) is considered significant enough in this kind of test, we also noted the extremely significant comparisons where p was smaller than 0.001 (**). p values of 1 or close to 1 suggest similarity between samples. The disease incubation times for animals inoculated with similarly prepared samples in the different experiments were pooled in the general calculations because no statistically significant difference was found between them. We also calculated the titers (log ID 50 ) from the median of disease incubation time as described (13). However, because the accuracy of titers calculated from disease incubation times (in days) is a debatable issue, we based all the statistical analyses directly on the disease incubation times. A graphic representation of the results can be seen in Fig. 4. Whereas very high infectivity was present in the H Sc fraction, no infectivity whatsoever was observed when this fraction was first denatured with 4.5 M GndSCN and then digested with PK, resulting in the NPH Sc fraction. These results indicate that the NPH Sc fraction cannot convert in vivo PrP C into PrP Sc , because even after a long incubation time (more than 300 days), no animals inoculated with these samples present any disease symptoms. Moreover, no traces of PrP 27-30 were observed in their brains even after 300 days (Fig. 3a), suggesting that no subclinical infection was established in these animals. Preliminary experiments (data not shown) also suggest that NPH Sc cannot convert PrP C to PrP Sc in vitro, because the brain inoculation of combined fractions containing NPH Sc and light fractions from normal hamsters without PK digestion (containing large quantities of PrP C ) did not result in any disease symptoms or PrP Sc accumulation after more than 300 days.
As shown here and in our previous work (11), whereas samples H Sc and L Sc Me2SO contained similar concentrations of PrP 27-30 (Fig. 3a), only the aggregated fraction, H Sc , presented high prion infectivity. This can be seen by comparing the titer, incubation time, and p values for both samples (Tables I and  II). However, when the noninfectious NPH Sc was incubated with the L Sc Me2SO fraction to form the NPH Sc /L Sc Me2SO , infectivity was restored to the levels observed in the H Sc samples. This was not the case for the NPH Sc /L Sc combination, suggesting that the presence of nonaggregated PrP 27-30 in the L Sc Me2SO was essential for the restoration of prion infectivity.
No increase in infectivity was detected when L Sc Me2SO was combined with NPH N . This may imply that a putative second prion component is not present in normal brain. Although it may be so, it is more probable that such a second component needs to combine with protease-resistant PrP in a specific fashion to sediment into the heavy fractions of the sucrose gradient in significant quantities.
One of the candidate molecules for a prion second component is HS. This sugar polymer was found in brain amyloid deposits of Alzheimer's disease as well as of prion diseases (19). In addition, sulfated sugars seem to have an important role in the metabolism of PrP Sc (20). Molecules such as pentosan sulfate and low molecular weight heparin have been shown repeatedly to inhibit the production of PrP Sc in scrapie-infected neuroblastoma cells (ScN2a cells). HS itself has been shown to either increase or reduce PrP Sc accumulation in these cells, depending on the experimental setup (21)(22)(23). These molecules have also been shown to inhibit prion disease pathogenesis in vivo (24 -26). This suggests that sulfated sugars of specific size and properties may either help form prions or disrupt prion forma-tion, probably depending on some kind of competition mechanism.
To test whether one of the NPH Sc components is an HS-like molecule, we digested the NPH Sc fraction with heparanase in two of the reconstitution experiments, prior to its combination with the L Sc Me2SO fraction. In the third experiment, we substituted HS for NPH Sc . As can be seen in Fig. 4, the combined results of these three experiments suggest a role for HS in prion infectivity. Whereas the infectivity of the reconstituted sample containing heparanase-digested NPH Sc (NPH Sc * / L Sc

Me2SO
) was higher than that of the light fractions presenting low infectivity, such as L Sc or L Sc Me2SO , it was significantly lower FIG. 4. Bioassay of inoculated samples. Male Syrian hamsters, 4 weeks old, were inoculated intracerebrally with 50 l of each of the samples to be tested for prion infectivity. Animals were tested daily. Prion titers were measured by monitoring the incubation period until the appearance of symptoms (13). a, graphic representation of disease incubation time as described in Table I; b, infectivity titers as described in Table I. Titers were calculated from incubation times as described (13). than the infectivity of both the H Sc and the NPH Sc /L Sc Me2SO fractions that present high infectivity (see Table II for p values). In addition, the HS/L Sc Me2SO combined sample showed a considerably higher infectivity than the L Sc Me2SO sample alone. In addition, the clinical features of the disease as well as the neuropathology of the animals infected with HS/L Sc Me2SO were identical to those of the classical 263 strain in Syrian hamsters (data not shown and Ref. 27), suggesting that the HS/L Sc Me2SO sample did not contain a new prion strain.
Extensive experiments are required to establish whether the effect of HS is unique or whether other sugar polymers, sulfated or not, may serve as the backbone of prion rods. DISCUSSION The results presented here indicate that production of prion infectivity requires the presence both of protease-resistant PrP and of nonprotein components of prion rods and suggest that these components may well be sulfated sugar polymers. It remains to be established whether sulfated sugar polymers are indeed a fundamental component of prions or whether their function, although not essential, greatly facilitates prion propagation and the establishment of prion infection.
Unsuccessful attempts to dissociate and reconstitute prion infectivity were performed years ago, even before PrP Sc was identified as a necessary components for infectivity (28). Prion infectivity could not yet be associated with protease-resistant PrP molecules produced by an array of in vitro conversion protocols (29 -31) and has even been suggested to exist in the absence of detectable protease-resistant PrP in the inocula (32). In view of our results presented here, which suggest both prion components are required, we suggest that prion infectivity can be transmitted by a few (and therefore undetected) molecules of PrP Sc , if associated with the appropriate nonprotein components. Our results also open the way to a new line of in vitro conversion experiments, which may hopefully result in full in vitro production of prion infectivity.
Whereas the function of PrP Sc as a template in the PrP C to PrP Sc conversion stands on solid grounds (33), the pathological role of a putative second component is unclear. A possible role for any functional molecule present in the NPH Sc fractions may be to anchor the PrP Sc molecules associated with it to the appropriate target in the host. Without a polymer such as HS, most PrP Sc molecules may be cleared from the brain before the PrP C to PrP Sc conversion reaction has been established in enough cells to establish a process of infection (34). The kind of sugar polymer used as rod backbone and anchor may play a role in modifying parameters of prion infectivity.
The fact that sulfated sugar polymers such as HS may have a crucial function in prion structure and propagation suggests several plausible explanations of the fact that small sulfated sugar polymers such as pentosan sulfate were shown to inhibit the production of PrP Sc in cells (21,22). These molecules may compete with the sugar polymer functioning as prion component for the right cell targets to which the template PrP Sc molecules should be docked. Otherwise, these molecules may compete with the prion sugar component for the binding of newly formed PrP Sc molecules. Interestingly, polyamines were shown recently to inhibit PrP Sc accumulation from ScN2a cells (35). We suggest that these highly positively charged polymers may bind to the highly negatively charged sulfated sugars and thereby facilitate the clearance of newly formed and still nonaggregated protease-resistant PrP. The results presented here therefore provide an explanation of the fact that such disparate molecules as small sulfated sugar polymers and positively charged polymers may both prove effective in the treatment of prion diseases.