PrP^c Does Not Mediate Internalization of PrP^Sc but Is Required at an Early Stage for De Novo Prion Infection of Rov Cells

Abstract

We have studied the interactions of exogenous prions with an epithelial cell line inducibly expressing PrP^c protein and permissive to infection by a sheep scrapie agent. We demonstrate that abnormal PrP (PrP^Sc) and prion infectivity are efficiently internalized in Rov cells, whether or not PrP^c is expressed. At odds with earlier studies implicating cellular heparan sulfates in PrP^Sc internalization, we failed to find any involvement of such molecules in Rov cells, indicating that prions can enter target cells by several routes. We further show that PrP^Sc taken up in the absence of PrP^c was unable to promote efficient prion multiplication once PrP^c expression was restored in the cells. This observation argues that interaction of PrP^Sc with PrP^c has to occur early, in a specific subcellular compartment(s), and is consistent with the view that the first prion multiplication events may occur at the cell surface.

Prion diseases are fatal neurodegenerative disorders characterized by the accumulation in the infected tissues of a detergent-insoluble and protease-resistant isoform of the cellular PrP^c protein (7, 43). The causative infectious agent, or prion, is thought to be an abnormal form of PrP^c (PrP^Sc), or a precursor of it, and prions apparently multiply by triggering the conversion of the normal PrP^c into the abnormal conformer (29). While prion multiplication does not occur in cells or tissues that do not express PrP^c (5, 9), several lines of evidence indicate that PrP^c is not sufficient to confer permissiveness. Mouse tissues expressing high levels of PrP do not necessarily accumulate prion infectivity (30), and very few PrP^c-expressing cell lines are able to propagate prions (37). This indicates that prions require an additional cellular and/or molecular factor(s) to multiply efficiently. The fact that some strains seem refractory to multiplication in cultured cells further suggests that at least some of the required factors are strain specific. The initial interactions of exogenous infecting PrP^Sc with a target cell are poorly characterized. Although a variety of cultured cell types internalize prion-infected material (13, 16, 17, 22, 24, 25, 33, 35), only a few will become infected (37), indicating that other cellular processes must occur for the establishment of cell infection. Various aspects of prion infection can be studied with Rov cells (41), including cell-to-cell prion dissemination (26) and, through a reverse genetics approach, the effects of natural polymorphisms of ovine PrP^c on prion multiplication (34). Several features make Rov cells an attractive model for the study of cell-prion interactions. First, they propagate a natural sheep scrapie strain. Second, the expression of the ovine PrP^c protein in Rov cells, and thus permissiveness to prion multiplication, can be selectively induced through the addition of doxycycline (dox) to the cell culture medium. This makes it possible to rigorously assess the influence of the PrP^c protein on the interaction of the cells with exogenous abnormal PrP and on the establishment of cell infection. In this study, we show that while the presence of the PrP^c protein is not required for efficient entry of PrP^Sc and infectivity in Rov cells, it is required at a very early stage for infection of Rov cells.

MATERIALS AND METHODS

Reagents and antibodies.

Dextran sulfate 500 (DS500) was from Sigma, and the dextran-based heparan mimetic agents HM20602 and CR36 (10) were kindly provided by D. Papy-Garcia. DS500, HM2602, and CR36 were made at 5 μg ml⁻¹ in phosphate-buffered saline (PBS) and were stored at −20°C. Aliquots diluted in cell culture medium were filter sterilized before being incubated with the cells.

4F2 monoclonal antibody (MAb) (19) was used to detect the expression of PrP^c in Rov cells. Immunoblot analysis of abnormal PrP in proteinase K (PK)-digested cell lysates was performed with ICSM18 (3) or Sha31 MAb (12).

Inoculum.

The PG127 sheep isolate and experimental sheep bovine spongiform encephalopathy (BSE) agent were serially propagated onto tg338 transgenic mice (20) that express the VRQ allele of ovine PrP^c on a mouse Prnp^0/0 background (42) to obtain the 127S and BSEov strains used in this study. Infected brains were homogenized at 10% (wt/vol) in a sterile 5% glucose solution with a Ribolyser (Hybaid, Middlesex, United Kingdom) and were sonicated for 1 to 2 min before incubation with Rov cells.

When PrP^Sc uptake was analyzed by immunofluorescence, the inoculum consisted of Rov cultures persistently infected with 127S. These cultures were incubated for 2 days in culture medium without dox, rinsed, scraped into PBS, pelleted by centrifugation, and resuspended in a 5% glucose solution. After four freeze-thaw cycles, the cell suspensions were sonicated for 1 to 2 min in a cup-horn apparatus before being diluted in cell culture medium.

Cell culture and uptake of PrP^Sc.

Rov cells (the Rov9 clone) (41) were maintained at 37°C in 6% CO₂ in alpha minimal essential medium supplemented with 10% fetal bovine serum, 100 U ml⁻¹ penicillin, and 10 μg ml⁻¹ streptomycin. Rov cells were grown to confluence in 12-well plates. To induce the expression of ovine PrP^c (the VRQ allele) in Rov cells, 1 μg ml⁻¹ dox was added to the culture medium. Infected brain homogenates diluted in 0.5 ml of cell culture medium were incubated with the Rov monolayers for the indicated periods of time. The medium was removed, and the cultures were carefully rinsed three times with PBS. The cell monolayer of each well was solubilized in 500 μl of lysis buffer (50 mM Tris-HCl, pH 7.4, 0.5% Triton X-100, 0.5% sodium deoxycholate) for 10 min at 4°C. Lysates were clarified (5,000 rpm for 1 min in a microcentrifuge), and cellular proteins were quantified by bicinchoninic acid. The proteins from each culture well were digested with PK for 2 h at 37°C (2 μg of PK for 500 μg of proteins) in the presence of 0.02% bromophenol blue to better visualize the pellets obtained after centrifugation. Pefabloc (4 mM) was added to stop the reaction, and blue pellets containing aggregated, PK-resistant PrP (PrP^res) were collected by centrifugation at 13,000 rpm in a microcentrifuge for 30 min at room temperature (27). PrP^res from each well was analyzed by immunoblotting after electrophoresis on sodium dodecyl sulfate-12% polyacrylamide gels and transfer to nitrocellulose filters. The filters were developed with an ECL reagent kit (Amersham), and chemiluminescent signals corresponding to the PrP glycoforms were quantified after acquisition with a GeneGnome digital imager (Syngene, Frederick, MD).

Detection of PrP^Sc uptake by immunofluorescence microscopy.

To specifically visualize the entry of PrP^Sc in Rov cells, an inoculum depleted of PrP^c was prepared (as described above) by using persistently infected Rov cell cultures that were incubated in cell culture medium without dox for 2 days, so as to downregulate the expression of PrP^c. Under these conditions, normal PrP was no longer detectable, while PrP^Sc was still present in large amounts. Rov cells grown on coverslips were incubated for 24 h in the presence of this inoculum. Cell monolayers were washed thoroughly with PBS, fixed with 4% paraformaldehyde-4% sucrose for 10 min, and permeabilized with 0.2% Triton X-100 in PBS for 3 min. Cells then were treated for 5 min with 3 M guanidine thiocyanate in PBS to expose PrP^Sc epitopes (39). ICSM35 MAb (3) was used to detect PrP^Sc taken up by the cells. Bound MAbs were revealed with fluorescein isothiocyanate-coupled goat anti-mouse secondary antibody. Nuclei were stained with propidium iodide. Dual-immunofluorescence confocal images were acquired by using a confocal laser-scanning microscope (CLSM 310; Carl Zeiss) equipped with a Plan-Apochromat 63× oil immersion objective (1.4 numerical aperture). Images were processed using Zeiss software for overlays of red and green channels.

Uptake of prion infectivity.

Rov cells left untreated or treated with 1 μg ml⁻¹ of dox were grown to confluence in a 75-cm² flask. The cell monolayers then were incubated with the equivalent of 2.5 μl of 10% infectious brain homogenate for 6 h. The cultures were rinsed three times with PBS, scraped into PBS, and pelleted by centrifugation. The cells were resuspended in a sterile 5% glucose solution, frozen and thawed four times, sonicated, and inoculated intracerebrally (20 μl) into tg338 mice. Mice showing neurological signs were monitored almost daily and were euthanized in extremis.

Amiloride treatment.

Rov cells were left untreated or were treated with 3 mM amiloride for 1 h prior to incubation with 2.5% of infectious brain homogenate for 30 min. We verified that this treatment prevented the hepatocyte growth factor-mediated induction of ruffles, as assessed by microscopic examination of Rov cells stained with fluorochrome-conjugated phalloidin.

Infection of Rov cells.

dox-treated confluent Rov cultures in 12-well plates were incubated overnight with the infectious brain homogenate diluted to 0.25% in the cell culture medium. In some experiments, the cells were pretreated for 24 h with sulfated glycans, and incubation of the cells with the inoculum was performed in the presence of the molecules. The monolayers then were rinsed three times with PBS, and each well was passaged into a 25-cm² flask. A few hours later, the dox-containing medium of each flask was changed and replaced with fresh medium. One week later, cell cultures were lysed in 1.5 ml of lysis buffer, and 500 μg cellular proteins was digested with PK to analyze PrP^res by immunoblotting as described above.

RESULTS

Infection requires the presence of PrP^c at the time of exposure to prions.

Although scrapie-associated fibrils represent the purest forms of prions known to date, they might be poorly relevant biologically, since such detergent-extracted, protease-treated, highly aggregated forms of abnormal PrP are unlikely to have counterparts in vivo. Thus, in this study we used homogenized brain tissue as a source of abnormal PrP and infectivity. Brain homogenate from ovine PrP tg338 mice infected with the 127S strain of sheep scrapie was diluted to 0.25% in cell culture medium and was incubated for 5 h with confluent Rov cells. The monolayers were rinsed thoroughly to remove the inoculum, and the cells were solubilized in detergent-containing lysis buffer. The cell-associated abnormal PrP was analyzed by Western blotting after PK digestion and centrifugation of the lysates. As shown in Fig. 1A, a substantial portion of exogenous PrP^Sc became associated with Rov cells. Quantitative measurements indicated that as much as 82% of the exogenously added PrP^res became associated with Rov cells after a 24-h incubation period (Fig. 1B). To determine if this process depended on the presence of PrP^c, we took advantage of the tight regulation of ovine PrP^c expression in Rov cells: cell cultures grown in the presence of dox express high levels of PrP^c, while no PrP^c is detected in untreated Rov cells (Fig. 1C) (41). Kinetic experiments using dox-treated and untreated control Rov cells showed that the association of PrP^Sc with the cells was entirely independent of the expression of PrP^c, as illustrated in Fig. 1C. The presence of PrP^Sc within individual cells could be visualized by immunostaining and confocal analysis of exposed Rov cultures (Fig. 1D). Moreover, blocking endocytosis by lowering the temperature during the exposure period strongly reduced the amount of abnormal PrP taken up by the cells. Quantification of experiments such as the one shown in Fig. 1E indicated that abnormal PrP taken up was 10- to 20-fold reduced at 4°C (n = 4 experiments). These findings confirmed that the observed phenomenon essentially involved an active internalization process. We then examined how long abnormal PrP would remain in the cells after internalization. After a 24-h exposure to infected brain homogenate, the cultures were rinsed. Some of them were lysed to quantify the uptake of abnormal PrP, while others were further incubated for up to 6 days to study its persistence in the cells. Figure 2B and C show that internalized abnormal PrP persisted for several days in the cells, implying that its degradation did not take place rapidly. We then examined the internalization of abnormal PrP from another prion strain, ovine BSE (BSEov). We chose BSEov because (i) at the terminal stage of the disease, tg338 mice expressing the VRQ allele of ovine PrP and infected with BSEov and 127S show similar amounts of PrP^res in their brains (Fig. 2A), and (ii) in contrast to 127S, BSEov does not multiply in Rov cells (data not shown). Results shown in Fig. 2B and C clearly show that abnormal PrP from BSEov was taken up less efficiently than that from 127S (35% ± 5% for BSEov and 82% ± 7% for 127S). That PrP^Sc from different strains enters the cells with distinct efficiencies indicates some selectivity of the internalization process.

FIG. 1.

Exogenous abnormal PrP is taken up by Rov cells independently of PrP^c expression. (A) Infectious brain homogenate was diluted to 0.25% in culture medium and was incubated with Rov cells (in duplicate). Five (left panel) or 8 (right panel) h later, the cultures were extensively washed and extracted, and lysates were treated with PK. Abnormal PrP taken up by the cells was analyzed by immunoblotting with anti-PrP MAb, and the level of PrP^Sc uptake was compared to that in the inoculum (input). (B) Rov cells were incubated for 5 or 24 h with infectious brain homogenate (0.25%), and abnormal PrP taken up by the cells was analyzed by immunoblotting as described in Materials and Methods. PrP^res in the cells is expressed as mean percentages ± standard deviations of PrP^res in the inoculum (input). (C) Uptake of abnormal PrP depending on PrP^c expression. The left panel shows inducible expression of ovine PrP^c in Rov cells. Equal amounts of methanol-precipitated proteins (20 μg) from Rov cultures treated with 1 μg ml⁻¹ of dox (+) or left untreated (−) were analyzed by immunoblotting with 4F2 anti-PrP MAb. The middle panel shows duplicate wells of Rov cells grown in the presence (+) or in the absence (−) of dox and incubated for 1, 8, or 28 h in the presence of infectious brain homogenate. Immunoblotting of abnormal PrP taken up by Rov cells was performed as described in Materials and Methods. The right panel shows the quantification of abnormal PrP taken up for 5 h in Rov cells expressing (+dox) or not expressing (−dox) PrP^c, expressed as mean percentages ± standard deviations of PrP^res in the inoculum (input). (D) Confocal microscopic examination of PrP^Sc internalized in Rov cells. Rov cells exposed to the inoculum (see Materials and Methods) were fixed, permeabilized, treated with guanidine thiocyanate to expose PrP^Sc epitopes, and stained with ICSM35 anti-PrP antibody. The image shows abnormal PrP staining (in green) and nuclei (in red) in an x-y plane. No PrP staining was observed when Rov cells were incubated with a mock-infected inoculum. (E) Infectious brain homogenate was incubated in duplicate for 6 h at 37°C or at 4°C with Rov cells. Immunoblotting of abnormal PrP taken up by Rov cells was performed as described in Materials and Methods. The positions of the molecular mass marker proteins are indicated (in kilodaltons).

FIG. 2.

Internalization of abnormal PrP from different strains. (A) PrP^res was isolated in duplicate from 0.6 mg of brain equivalent from terminally ill tg338 mice infected with strain 127S or BSEov; samples were analyzed on the same gel by immunoblotting with anti-PrP MAb. Note the distinct banding patterns of 127S and BSEov PrP^res. (B) Uptake and persistence in Rov cells of abnormal PrP from 127S and BSEov. 127S or BSEov brain homogenate was diluted to 0.5% in cell culture medium (input) and incubated with Rov cells. The inoculum was removed 24 h later by washing, and the cultures were extracted either immediately (uptake) or after 1 to 6 days, as indicated. Cell lysates were treated with PK, and abnormal PrP taken up by the cells was analyzed by immunoblotting with anti-PrP MAb. (C) Uptake efficiency of abnormal PrP from two different strains. 127S or BSEov brain homogenate was diluted to 0.5% in cell culture medium and incubated with Rov cells for 24 h. Abnormal PrP taken up by the cells was analyzed by immunoblotting as described in Materials and Methods, and the level of uptake was compared to that in the inoculum (input). The data represent the means ± standard errors from five independent experiments.

While providing a surrogate marker of prion multiplication in infected cells and tissues, the aggregated, PK-resistant form of PrP (PrP^res) may not be the most infectious (36). To see whether the uptake of infectivity also was independent of the presence of PrP^c, 10% infected brain homogenate was diluted 4,000-fold in culture medium and was incubated with Rov cells that either expressed or did not express ovine PrP^c. Six hours later, the cells were rinsed, collected, and inoculated into susceptible ovine PrP^c transgenic mice to quantify prion infectivity taken up by the cells. As a result, the two groups of mice (n = 7) showed similar survival times (77 ± 1.4 and 81 ± 0.7 days for untreated and dox-treated cells, respectively), establishing that Rov cells have taken up the infectivity with an efficiency that did not depend on the presence of PrP^c at the cell surface (Fig. 3). Based on the dose-dependent incubation time of the 127S agent in tg338 mice, it was calculated that about 5 to 10% of input infectivity was taken up by the cells within 6 h of exposure.

FIG. 3.

Prion infectivity is taken up by Rov cells independently of PrP^c expression. Infectious inoculum (2.5 μl of 10% brain homogenate) was incubated for 6 h with Rov cells expressing (open circles) or not expressing (filled circles) PrP^c. The cultures then were harvested and inoculated intracerebrally in tg338 mice (n = 7), and the mice were monitored until death to determine the survival time. Rov cells not exposed to the inoculum were inoculated in parallel (open triangles). The survival time of tg338 mice inoculated with 20 μl of a 1,000-fold-diluted 10% brain homogenate (filled triangles) also is shown for comparison.

The regulatable expression of PrP^c in Rov cells made it possible to investigate if abnormal PrP taken up in the absence of PrP^c still would trigger infection once PrP^c expression is restored. Control or dox-treated Rov cells were exposed overnight to infected brain homogenate to allow the uptake of abnormal PrP. The inoculum was removed, and then the cells were trypsinized and transferred to new flasks in the presence of dox and grown for 7 days to detect infection. PrP^c-expressing Rov cells exposed to the inoculum produced PrP^res (Fig. 4, lanes 1 and 2), while no PrP^res was detected in cultures that did not express PrP^c (Fig. 4, lanes 5 and 6). Remarkably, infection did not develop in cultures in which PrP^c expression was induced after the time of exposure (Fig. 4, lanes 3 and 4).

FIG. 4.

PrP^c is required at the time of inoculation for efficient cell infection. Duplicate Rov cell cultures were exposed for 18 h to infectious brain homogenate (0.25%) in the presence (lanes 1 and 2) or in the absence (lanes 3 and 4) of dox. After removal of the inoculum, the cultures were passaged and further grown for 7 days in dox-containing fresh medium. Cell lysates then were digested with PK and analyzed by immunoblotting for PrP^res. Cultures expressing no PrP^c at the moment of inoculation showed little or no PrP^res (lanes 3 and 4). These data are representative of six experiments. The signal denoted by an asterisk is due to cross-reaction of the ICSM18 MAb with PK. The positions of molecular mass marker proteins are indicated (in kilodaltons).

The above data demonstrated that Rov cells are able to efficiently internalize abnormal PrP irrespective of whether (i) the agent will multiply or (ii) PrP^c is expressed. Conversely, PrP^c expression at the time of exposure to prions appeared to be a crucial requirement for the infection to proceed.

PrP^Sc infection, but not internalization, is inhibited by heparan sulfates.

To explore if abnormal PrP has to be internalized to infect Rov cells, we studied further the uptake of PrP^Sc in these cells. A strong inhibition of the uptake of exogenous PrP^sc by sulfated glycans (1 μg/ml), such as DS500 or HM2602, has been reported in several recent studies, pointing to cellular heparan sulfates as possible receptors for abnormal PrP (16, 17, 35). To test the effects of these molecules on Rov cells, cultures were preincubated for 24 h with DS500 or the heparan mimetic agents HM2602 and CR36, and inoculum was added to the culture medium for an additional 18 h in the presence of the tested molecules. Data shown in Fig. 5A indicate that none of the sulfated glycans (at up to 25 μg/ml) inhibited the uptake of abnormal PrP in the cells (data not shown for CR36). We also looked for an effect of Na chlorate on PrP^Sc uptake. Treatment with Na chlorate, a potent inhibitor of protein sulfation (1), including cellular heparan sulfates, was found to strongly inhibit PrP^Sc uptake (17). However, we found that treatment of Rov cells with Na chlorate did not affect PrP^Sc uptake in Rov cells (Fig. 5B). Therefore, unlike the results found for CHO, GT1, and N2a cells, we failed to find an involvement of heparan sulfates in the uptake of PrP^Sc by Rov cells.

FIG. 5.

Sulfated glycans have no detectable effect on the uptake of abnormal PrP by Rov cells. Immunoblots of abnormal PrP taken up by Rov cells are shown. (A) Duplicate wells were left untreated or were treated for 24 h with the indicated concentrations of DS500 or HM2602. Infectious brain homogenate (0.25%) was added to the cells, which were further incubated for 18 h. (B) Duplicate wells were incubated for 24 h with Na chlorate (30 mM) or were left untreated (control). Infectious brain homogenate (0.25%) was added to the cells, which were further incubated for 18 h.

Macropinocytosis, a major form of endocytosis in various cell types, is involved in cell invasion by various pathogens (8, 38). To test the possibility that this internalization route might contribute to the uptake of abnormal PrP, we performed the uptake assay in the presence of amiloride, an inhibitor of macropinocytosis (45). Rov cells were pretreated with amiloride (3 mM) for 1 h (see Materials and Methods). Infected brain homogenate then was added for 30 min, and the amount of PrP^Sc taken up was quantified by immunoblotting. As a result, amiloride treatment did not affect the uptake of abnormal PrP (data not shown). Therefore, the mode of internalization of PrP^Sc in Rov cells remains to be determined.

To determine if heparan sulfates could affect transmission of infection to Rov cells, cultures were exposed to the infectious inoculum in the presence of DS500, CR36, and HM20602 (25 μg/ml). After the inoculum and the heparan sulfate molecules were washed out, the treated cultures were grown in regular cell culture medium for 1 week to detect accumulation of cell-derived abnormal PrP. Figure 6 shows that the presence of DS500, CR36, and HM20602 at the time of exposure of the cells to prions strongly inhibits infection of Rov cells. From these data it was concluded that sulfated glycans do not have a detectable effect on the uptake of PrP^sc but strongly inhibit subsequent infection of Rov cells.

FIG. 6.

Sulfated glycans inhibit Rov cell infection. Duplicate dox-treated Rov cultures were left untreated or were incubated with the indicated molecules (25 μg ml⁻¹) for 24 h. Infectious brain homogenate (0.25%) was added to the cells, which were further incubated for 18 h in the presence of the molecules. After removal of the inoculum and of the tested molecules, the cultures were passaged and further grown for 7 days in dox-containing fresh medium. The cells then were lyzed, digested with PK, and analyzed by immunoblotting for PrP^res. These data are representative of four experiments. The positions of molecular mass marker proteins are indicated (in kilodaltons).

DISCUSSION

Biochemical analyses performed in this study showed that acquisition of abnormal PrP by epithelial Rov cells is an active and a selective process, the efficiency of which can depend on the prion strain. Among the potential cellular receptors that could interact with and promote the entry of PrP^Sc, a prime candidate is the cellular form of PrP^c (6), which is constitutively endocytosed from the cell surface to intracellular compartments (15, 28). The use of the PrP^c-inducible Rov cell system allowed us to formally demonstrate that the presence of PrP^c at the cell surface is not required for internalization of abnormal PrP. Similar conclusions were recently obtained with nonpermissive CHO cells (16) and primary cultured neurons (22), therefore indicating that PrP^c is not an obligatory receptor for the entry of PrP^Sc in various cell types.

Recent studies have provided evidence that cellular heparan sulfates and/or the laminin receptor LRP/LR may be a cellular receptor for abnormal PrP. Cellular heparan sulfates are sulfated linear polysaccharides, typically linked through serine residues to proteins to form heparan sulfate proteoglycans located at the cell surface (40). Heparan sulfate proteoglycans bind numerous extracellular proteins and regulate a wide range of cellular functions (4), and they also have been shown to mediate the internalization of proteins and pathogens (21, 31, 32). The enzymatic removal of cell surface heparan sulfates (17), the inhibition of their sulfation by chlorate (17), or competition with heparin (16) and heparan mimetic agents (17, 35) prevented the binding and the uptake of detergent-extracted, aggregated forms of abnormal PrP in N2a, GT1, and CHO cells, providing evidence that cellular heparan sulfates might be involved in the entry of aggregated PrP^Sc in the cells. The laminin receptor LRP/LR also might mediate internalization of PrP^Sc, as PrP^Sc uptake can be partly inhibited in Caco-2/TC7 cells by W3 anti-LRP/LR antibody (25), while it is stimulated in LRP/LR-transfected BHK21 cells (13). LRP/LR is a heparan sulfate binding molecule (14, 18), and, interestingly, LRP/LR-mediated uptake of PrP^Sc in BHK21 cells is inhibited by sulfated glycans (13). This suggested that the LRP/LR and cellular heparan sulfates participate in PrP^Sc internalization through the same pathway. Our study, which failed to find evidence that the efficient entry of PrP^Sc in Rov cells is mediated by cellular heparan sulfates, suggests that additional mechanisms are involved in the entry of abnormal PrP in the cells. Consistent with this possibility, LRP/LR-independent uptake of PrP^Sc has been observed recently in BHK21 cells (13). One possible explanation is that internalization routes vary depending on the cell type. Alternatively, the utilization of heparan sulfates as opposed to that of other macromolecules as a receptor might depend on the biochemical state of abnormal PrP. Indeed, PrP^res obtained by detergent extraction of infected tissues in combination with proteolysis (such as that used previously [17]) results in the formation of fibrils, while membrane-associated PrP^Sc (as in the brain homogenates prepared in physiological buffers used in the present study) forms amorphous aggregates (23). The finding that large aggregates of purified PrP^res or abnormal PrP isolated by sodium phosphotungstic acid precipitation infect target SN56 cells much less efficiently than homogenized tissue (2) can be viewed as an indication that detergent-extracted, purified PrP^res and membrane-associated forms of abnormal PrP interact differently with the cells.

We show that abnormal PrP taken up in Rov cells remained undegraded for days. Using a different experimental approach, similar conclusions recently were reported by Magalhães and coworkers (22), who showed that fluorescence-labeled aggregates of PrP^res taken up by SN56 cells or primary neurons remained in a large set of intracellular vesicles, including late endosomes and lysosomes, for days. However, a salient finding in our study is that internalized PrP^Sc, while infectious for indicator mice (Fig. 3), does not necessarily initiate infection in the uptaking cells. Indeed, infection was impaired if PrP^Sc entered the cells, while PrP^c was not expressed. In a previous study, we showed that exogenous prions preferentially infect Rov cells through the apical membrane, where PrP^c is localized (27). Other studies on N2a neuroblastoma cells have shown that removal of PrP^c from the plasma membrane precludes the initiation of infection (11). Taken together, these observations indicate that PrP^c expression has to meet precise temporal and spatial requirements at the early steps for successful prion infection, and they strengthen the view that the presence of PrP at the cell surface is crucial for the establishment of infection. One possibility is that PrP^c conversion occurs primarily or at least is initiated at the plasma membrane. The finding that prions bound to physical supports, such as stainless steel wires, can transmit infection to N2a cells (44) also supports the cell surface as a possible conversion site. However, other scenarios can be envisioned to explain why intracellular PrP^Sc taken up in the absence of PrP^c is unable to infect Rov cells. The plasma membrane may be a meeting place required for the formation of complexes between PrP^Sc, PrP^c, and other putative cofactors, such as cellular heparan sulfates (17), but complexes could be driven to subcellular compartments permissive to conversion. Further work will be necessary to fully characterize the different internalization routes of PrP^Sc and to clarify their biological significance regarding the initiation of cell infection.