Detergent-resistant membrane domains but not the proteasome are involved in the misfolding of a PrP mutant retained in the endoplasmic reticulum

Prion diseases are neurodegenerative encephalopathies of infectious, sporadic or inherited origin of humans and animals in which the transmission of the disease is based on protein conformation (Prusiner, 1998). According to the `protein only' hypothesis (Prusiner, 1998), the transmissible agent is PrP^Sc (for scrapie), a β-sheet-enriched and protease-resistant conformer of PrP^C, a cell surface glycoprotein of unknown function. The molecular basis of the conversion of PrP^C into PrP^Sc is still unknown. However, it is likely that during infectious transmission this event is catalyzed by exogenous PrP^Sc, whereas in the inherited form of the disease it occurs spontaneously as a result of a dominant mutation in the PrP gene, Prnp (Young et al., 1999).

The intracellular compartment where PrP^C/PrP^Sc conversion occurs and how this process leads to neurological dysfunction have still not been conclusively determined (Campana et al., 2005). It is also unclear whether the conversion compartment is the same for the infectious and hereditary forms of the disease. Indeed, in infected cells PrP^Sc accumulates in late endosomes and a block of endocytosis reduces scrapie production (Borchelt et al., 1992; Shyng et al., 1995; Marella et al., 2002), thus indicating that the endo-lysosomal pathway participates in scrapie formation. By contrast, it has been shown that different PrP mutants are localized principally in the ER (Ivanova et al., 2001; Drisaldi et al., 2003; Fioriti et al., 2005) and the acquisition of their scrapie-like properties appears to occur through different steps along the exocytic pathway (Harris, 2003).

Different pathological mutants have been shown to be degraded by the endoplasmic reticulum-associated degradation (ERAD) pathway and may accumulate as detergent-insoluble and partially proteinase K (PK)-resistant forms (Zanusso et al., 1999; Jin et al., 2000). Furthermore, PrPwt and some mutants have been shown to accumulate in the cytosol of neuroblastoma cells in response to proteasome inhibition and to subsequently assemble as classical aggresomes (Ma and Lindquist, 2001; Mishra et al., 2003). These data indicate that an overloading of the proteasomal system and a mislocalization of the protein in the cytosol could be involved in the scrapie-like conversion. In support of this hypothesis Lindquist and colleagues have shown that relocalization of PrP in the cytosol promoted its conversion to a self-perpetuating PrP^Sc-like conformation (Ma and Lindquist, 2002) that was selectively toxic in neuroblastoma cells, but not in non-neuronal cells and caused neurodegeneration in transgenic mice (Ma et al., 2002). However, this hypothesis has been recently questioned by the fact that neither PrPwt nor PrP mutants are a major substrate for retrotranslocation and proteasomal degradation in different transfected neuronal and non neuronal cell lines and primary neurons (Drisaldi et al., 2003), suggesting that the previous results were probably due to an increase in the synthesis of PrP induced by proteasomal inhibitors. In addition, the recent observation that some of the PrP mutants that accumulated in the cytosol upon proteasome inhibition contained an uncleaved signal peptide suggests that failure of PrP translocation into the ER rather than retrograde transport from the ER lumen could be the preferential route to the cytosol (Drisaldi et al., 2003), thus leading to the conclusion that the proteasome is not involved in the degradation of the scrapie-like misfolded mutants.

Controversy also exists regarding the presumed toxicity of cytosolic PrP (Roucou et al., 2003; Ma et al., 2002; Fioriti et al., 2005). Indeed, the fact that no increase in cytosolic PrP was found in infected cells and animals (Stewart and Harris, 2003) and that the neuronal subpopulations expressing PrP in the cytoplasm are neither necrotic nor apoptotic (Mironov et al., 2003) are in contrast to the proposed toxic and pathogenetic role of the cytosolic form.

Although there are clear indications that the ER is the key intracellular compartment in the pathogenesis of inherited prion diseases (Ivanova et al., 2001; Harris, 2003), these contrasting data indicate that mechanisms distinct from the proteasomal pathway are responsible for prion disfunction and neurotoxicity in prion diseases. Interestingly, it has been proposed that detergent-resistant membrane microdomains (DRMs) or rafts might contribute to prion conversion in infected cells (Taraboulos et al., 1995; Naslavsky et al., 1999; Baron et al., 2002). However, to date there are no studies addressing whether these microdomains have a role in scrapie-like conversion of PrP mutants. We therefore studied the metabolism and intracellular trafficking of PrPT182A, a PrP variant carrying a mutational inactivation of one of the N-glycosylation sites, which in humans is responsible for a hereditary form of CJD (Nitrini et al., 1997) and spontaneously acquires PrP^Sc-like properties (Lehmman and Harris, 1997).

As previously shown in Chinese hamster ovary (CHO) cells (Lehmann and Harris, 1997), PrPT182A partially misfolds into a PK-resistant form and is unable to reach the plasma membrane of transfected Fisher rat thyroid (FRT) cells, but is retained in the ER. Although ER retention could suggest failure to pass the ER quality-control system (Hampton, 2002) and lead to proteasomal degradation, proteasomal block induces the accumulation of an unglycosylated PrP form that fails to translocate into the ER lumen and is not cytotoxic. By contrast, the majority of the protein does not accumulate in the cytosol and does not increase its misfolding, indicating that the ERAD pathway is not responsible for the degradation of scrapie-like PrP mutant and that other mechanisms must be involved.

We found that more than 70% of PrPT182A is associated with DRMs in the ER, in a cholesterol-dependent manner. Interestingly, cholesterol depletion abolishes its raft association, promotes its accumulation and increases substantially its misfolding. These data support our previous hypothesis that cholesterol-enriched domains in the early secretory pathway are important for the correct folding of PrP (Sarnataro et al., 2004) and strongly suggest that they might also have a protective role in its pathological conversion (Campana et al., 2005).

Because of their role in the spreading of the infection from gut to central nervous system (Heppner et al., 2001) and of the similarity of their polarity with neuronal cells, epithelial cells have been shown to be a good model to study prion trafficking and infection (Villette et al., 2001; Sarnataro et al., 2002; Paquet et al., 2004; Sarnataro et al., 2004). In particular, polarized epithelial FRT cells have previously been used to characterize the exocytic pathway of mouse PrP^C (Sarnataro et al., 2002). In order to understand whether pathological PrP mutants exhibit differences in their intracellular trafficking which might have a role in the pathogenesis of the inherited diseases, we stably transfected FRT cells with a plasmid encoding PrPT182A, a pathological PrP mutant (3F4-tagged), mutated in the first consensus site for N-glycosylation (Nitrini et al., 1997). Different stably expressing FRT clones were analysed for PrPT182A expression by western blot. Compared with the wild type (wt), PrPT182A was expressed at lower levels and appeared as a sharp band in western blots, migrating at the molecular mass (∼25 kDa) of the monoglycosylated wt form (Fig. 1A, Fig. 2A).

To analyse whether PrPT182A acquires PrP^Sc-like properties in FRT cells, we examined PIPLC sensitivity and performed Triton/DOC insolubility and proteinase K (PK) resistance assays (Fig. 1), the three assays commonly used to demonstrate scrapie-like misfolding (Lehmann and Harris, 1997). In contrast to PrPwt, none of the PrP mutant was released in the aqueous phase after PIPLC treatment (Fig. 1A), indicating an abnormal attachment to the membrane or an abnormal folding of the protein that impairs access of the enzyme to the glycosylphosphatidylinositol (GPI) anchor. We also evaluated the detergent insolubility of PrPwt and the mutant by centrifuging Triton/DOC lysates at 265,000 g for 40 minutes, which should sediment PrP^Sc-like isoforms but not PrP^C. Under these conditions, ∼40-60% of PrPT182A sedimented in FRT cells (Fig. 1B, compare lanes 2 and 4). Finally, we analysed PK resistance by treating 1 mg of total protein in cellular lysates with 3.3 μg PK for 2 minutes at 37°C (Fig. 1C). Whereas PrPwt is entirely digested (Fig. 1C, lane 2), between 10 and 20% of the mutant was resistant to PK digestion, as shown by the permanence after digestion of the full-length protein (∼25 kDa) and by the appearance of a smaller isoform of ∼20 kDa (Fig. 1C, lanes 4). Taken together, these experiments indicate that in FRT cells PrPT182A displays the major biochemical hallmarks of scrapie PrP, i.e. abnormal membrane attachment, aggregation and misfolding.

In order to characterize the oligosaccharide chains of the PrP mutant, FRT cells expressing either PrPwt or PrPT182A were treated with either Endo-H or with neuraminidase (Lehmann and Harris, 1997) (Fig. 2A). As expected, PrPwt was resistant to Endo-H and sensitive to neuraminidase digestion, whereas PrPT182A was completely resistant to neuraminidase digestion and quite sensitive to Endo-H treatment (Fig. 2A), which is similar to the situation previously shown in CHO cells (Lehmann and Harris, 1997). By pulse-chase experiments we demonstrated that all the mutant remained Endo-H sensitive during the entire chase period (Fig. 2B), in contrast to the wt, which becomes Endo-H resistant after 30 minutes of chase (not shown) (Sarnataro et al., 2002). These data suggest that glycan modification enzymes have no access to the N-linked oligosaccharide chains of the PrPT182A mutant, either because of its abnormal folding and/or because of its retention in a premedial Golgi compartment where these enzymes are not present.

Therefore, in order to specifically identify the intracellular localization of the mutant protein and to understand whether a portion of the protein reaches the plasma membrane despite the incomplete sugar residues, we performed indirect immunofluorescence and confocal analysis on polarized filter-grown cells and compared the results with those of the wt protein (Fig. 3A). In contrast to the wt protein only a low plasma membrane signal of PrPT182A was found in non-permeabilized (NP) conditions, whereas there was a strong intracellular diffuse signal in permeabilized conditions (P) (Fig. 3A), indicating that PrPwt is able to reach the plasma membrane (Sarnataro et al., 2002) but the mutant is mainly retained inside the cell. In order to identify this intracellular compartment, we performed double indirect immunofluorescence experiments, labelling PrP and different intracellular compartments (Fig. 3B). Whereas the wt protein colocalized extensively with giantin, a marker of the Golgi apparatus, there was only a faint signal of PrPT182A in this organelle (Fig. 3B). In addition, we could not find any colocalization with markers of the endocytic pathway, such as EEA1 and Lysotracker (Fig. 3A) or Rab5 and Rab7 (not shown), therefore, excluding an endo-lysosomal localization for both the mutant and PrPwt (Fig. 3B). By contrast, we found an almost complete colocalization of PrPT182A with calnexin (CNX), an ER marker, that colocalized only marginally with the wt protein (Fig. 3B). Similar results were obtained with other ER markers such as binding protein (BiP), protein disulphide isomerase (PDI) and calreticulin (CLT) (not shown). Thus, these data clearly indicated that the PrPT182A mutant was retained principally in the endoplasmic reticulum.

Because ER retention of misfolded proteins might lead to degradation through the endoplasmic reticulum-associated degradation (ERAD) pathway, and because some pathological PrP mutants have been shown to be partially degraded by the proteasome (Zanusso et al., 1999; Jin et al., 2000), we investigated the role of this pathway in the degradation of PrPT182A (Fig. 4).

FRT cells expressing either PrPwt or PrPT182A were treated for 7 hours with a proteasome inhibitor, the peptide aldehyde ALLN (150 μM). In these conditions, we did not find accumulation of either the PrP mutant (Fig. 4A, lane 2) or the wt (Fig. 4A, lane 4) protein. These results were confirmed by using other proteasomal inhibitors, such as PSI 1, MG132 and lactacystin (not shown).

In order to confirm that the proteasome does not participate in the degradation of the PrPT182A mutant, we measured its degradation rate by pulse-chase labelling experiments in the presence of ALLN. Quantification of three different experiments (Fig. 4B) does not show any variation of the half-life of the mutant in the presence of ALLN treatment compared with control conditions.

Interestingly, in cells where proteasome activity was blocked, a supplementary PrP product was found that migrated in western blots between the unglycosylated and monoglycosylated PrP isoforms (Fig. 4A, arrow). To understand whether this band was the unglycosylated PrP form containing the signal peptide that was recently described in similar conditions (Drisaldi et al., 2003), we carried out a protease protection assay on isolated microsomes prepared from control and ALLN-treated cells. Although the majority of the mutant was protected by the protease treatment of intact microsomes (Fig. 4C), this band was degraded, indicating that it represented an untraslocated form associated with the cytosolic leaflet of the ER membrane (see also Fig. 5C). Similar results were also obtained for PrPwt (not shown), thus confirming previous data that the untranslocated PrP isoform is normally degraded by the proteasome both in the case of the wt and the mutant (Drisaldi et al., 2003).

We then investigated whether PrPT182A undergoes ubiquitylation when proteasomal activity was blocked (Fig. 4D,E). We immunoprecipitated PrP in denaturing conditions (necessary to access ubiquitin-linked PrP) (Yedidia et al., 2001) in control and ALLN-treated cells. Immunoprecipitates were then revealed by western blot using either an α-PrP or an α-polyubiquitin antibody. Interestingly, under ALLN treatment a smear of high molecular mass was recognised by the α-polyubiquitin antibody (Fig. 4D).

To exclude the possibility that the smear represented ubiquitylated proteins, which coimmunoprecipitated with PrP rather then polyubiquitylated forms of PrP, we separated proteins on an SDS-PAGE second dimension (Friedrichson and Kurzchalia, 1998) under reducing and denaturing conditions (Fig. 4E). In this way, we could separate PrP isoforms (which would run along a typical diagonal between low and high molecular masses) from putative PrP interactors, which are separated on the basis of their specific molecular masses and therefore run outside the diagonal of PrP. As shown in Fig. 4E the α-polyubiquitin staining that appeared after proteasomal block ran along the PrP diagonal (at high molecular mass, above the asterisk indicating the high chains of immunoglobulins). Because the only PrP form that accumulates in conditions of proteasomal block is the untranslocated form (Fig. 4A), these results suggest that the high molecular mass smear represents the polyubiquitylated untranslocated form of PrPT182A and that polyubiquitylation is required for its proteasomal degradation.

In order to directly assess whether proteasomal block induced a cytosolic accumulation of an untranslocated precursor of PrPT182A, we performed confocal immunofluorescence after ALLN treatment. In these conditions PrPT182A remained localized mainly in the ER and we could not detect any signal in the cytosol (Fig. 5A). However, we noticed an enlargement of the ER and a fragmentation of the Golgi apparatus, probably due to the general effect of ALLN on the cells (Fig. 5A). In order to exclude the possibility that we had missed a low cytosolic signal, we performed a cytosol-membrane separation assay in control conditions and after ALLN treatment (Fig. 5B). In this assay, neither PrPT182A (Fig. 5B) nor the wt protein (not shown) untranslocated precursors were found in the cytosolic soluble fraction, indicating that impairment of proteasomal activity does not promote accumulation of PrP in the cytosol, in contrast to what was previously shown (Ma and Lindquist, 2001; Ma et al., 2002). We also found that the polyubiquitylated PrPT182A was associated with the membrane fraction (Fig. 5C), suggesting that ubiquitylation of the untranslocated mutant occurs on the cytosolic leaflet of the ER, as also shown for the other ubiquitylated proteins in the lysate (Fig. 5C, right panel). In addition, after proteasomal block there was no impairment of cell viability, as assessed by TUNEL assay (not shown), suggesting that this form is not cytotoxic.

Because it has been proposed that proteasomal impairment can favour PrP scrapie-like conversion in the cytosol, we investigated whether PrPT182A and in particular its untranslocated form assumed the scrapie-like characteristics after ALLN treatment (Fig. 5D). Interestingly, the amount of PK resistant (Fig. 5D) and Triton/DOC insoluble (not shown) PrPT182A did not increase after the proteasome block, as shown by the unchanged ratio of the PK resistant forms, and the total amount of protein compared with control conditions (Fig. 5D). Thus, our data clearly indicate that the proteasome does not participate in the metabolism of the misfolded form of PrPT182A and suggest that other mechanisms must be involved.

Thus the question arises as to what is the mechanism involved in PrP mutant scrapie-like conversion. Because specialized membrane domains (DRMs or rafts) have been proposed to have a role in the pathological events leading to infectious prion diseases (Taraboulos et al., 1995; Naslavsky et al., 1999; Baron et al., 2002; Sanghera and Pinheiro, 2002; Campana et al., 2005; Kazlauskaite and Pinheiro, 2005), we studied whether PrPT182A is associated with rafts and whether this association might have a role in its scrapie-like misfolding.

To this end, we performed purification of DRMs from FRT cells expressing PrPT182A, on sucrose density gradients after lysis in TX-100 (Fig. 6A). The average of four different experiments showed that a higher proportion (70-80%) of PrPT182A, compared with the wt protein (Sarnataro et al., 2002), floated to the lighter DRM fractions of the gradients. We characterized PrPT182A association with DRMs by depleting the cells of cholesterol and sphingolipids, the two major raft components, and analysed the effect of these treatments on its flotation rate. We found that PrPT182A association to DRMs was perturbed only by depletion of cholesterol (Fig. 6A, right panel) and not by blocking the synthesis of sphingolipids (not shown), unlike PrPwt, which was affected by both treatments (Sarnataro et al., 2004). Indeed, the combined use of mevinolin and methyl-β-cyclodextrin, which depleted ∼60% of intracellular cholesterol, resulted in the complete loss of flotation of PrPT182A. Interestingly, in these conditions the ER localization of the PrPT182A mutant was not affected (Fig. S1 in supplementary material). In addition, pulse-chase labelling experiments in cholesterol-depleted cells showed that the half-life of PrPT182A was increased from 60 to 90 minutes. This led to a fivefold increase in the amount of PrP at steady state compared with control conditions (Fig. 6C, compare lanes 1 and 3).

To then understand whether cholesterol-dependent lipid microdomains had a role in the acquisition of scrapie-like properties of this PrP mutant, we analysed PK-resistance of PrPT182A after cholesterol depletion (Fig. 6C). We found that the amount of the PK-resistant form increased from 10-20% of the control cells to 80-90% in cholesterol-depleted samples, (Fig. 6C, compare lanes 2 with 4). These data therefore indicate that misfolding of the PrP mutant is facilitated by the perturbation of cholesterol-enriched membrane domains and suggest a role for these domains in preventing PrP misfolding.

The intracellular site and the mechanisms leading to PrP scrapie-like prion conversion are not clear. We have studied the intracellular trafficking and metabolism of a pathological PrP mutant, PrPT182A. We show that PrPT182A acquires scrapie-like properties in transfected FRT cells (Fig. 1) and has an altered intracellular localization compared with the wt protein, being mainly retained in the endoplasmic reticulum (Fig. 3).

Moreover, this mutant fails to acquire complex oligosaccharides, as indicated by its Endo-H sensitivity, and has a significantly shorter half-life than the wt protein (Fig. 2), which suggests that ER retention could be involved in the efficient degradation of PrP misfolded isoforms. Indeed, proteins retained in the ER that do not pass the quality control process are retrotranslocated into the cytosol and degraded by the proteasome (Hampton, 2002). Because this ERAD pathway was proposed to be involved in the metabolism and cytotoxicity of PrP (Yedidia et al., 2001; Ma and Lindquist, 2001; Ma et al., 2002) and of some PrP pathological mutants (Zanusso et al., 1999; Jin et al., 2000; Ma and Lindquist, 2001), we investigated whether it participated in the degradation of PrPT182A. In agreement with recent findings in stably transfected neuronal and non neuronal cell lines (Drisaldi et al., 2003; Fioriti et al., 2005), we found that proteasomal inhibitors led to the accumulation of an unglycosylated form of PrP that maintains the signal peptide and is unable to translocate into the ER (Fig. 4A,B). By contrast, the monoglycosylated and unglycosylated forms of both the wt and the PrP mutant do not accumulate (Fig. 4A) even after a longer proteasomal block, indicating that the proteasome is not involved in PrP degradation.

However, the half-life of PrPT182A increased following inhibition of lysosomal proteolytic activities, thus indicating that it undergoes degradation in the lysosomes (Fig. S2 in supplementary material), as previously shown for the wt protein (Taraboulos et al., 1992; Caughey et al., 1991).

Notwithstanding, a current hypothesis suggests that impairment of proteasomal degradation leads to accumulation of misfolded neurotoxic PrP molecules in the cytosol (Ma and Lindquist, 2001; Ma et al., 2002; Ma and Lindquist, 2002). Because we could not exclude the possibility that the untranslocated precursor was the PrP form able to accumulate and convert in the cytosol under proteasomal block, we analysed PrPT182A localization under these conditions. Both by immunofluorescence and subcellular fractionation we showed that PrPT182A does not accumulate in a soluble cytosolic fraction after block of the proteasome (Fig. 5A,B), but maintains a prevalent ER distribution (Fig. 5A) as in control conditions (Fig. 3B). These results indicate that, although the untraslocated precursor is exposed to the cytosolic environment (Fig. 4C), it remains associated with the external membrane of the ER and does not diffuse into the cytoplasm. Accordingly, we have shown that following proteasomal block PrPT182A is polyubiquitylated (Fig. 4D,E) and that this process occurs on the ER cytosolic leaflet (Fig. 5C), as previously described for other proteins (Zhong et al., 2004). Moreover, under proteasomal block we did not find any modification of cell viability (not shown), indicating that the untranslocated PrP form does not have any cytotoxic effects.

These results are in agreement with similar findings for other pathological PrP mutants in different cell models (Drisaldi et al., 2003; Fioriti et al., 2005), but they are in direct contrast with other data (Ma and Lindquist, 2001; Ma et al., 2002; Ma and Lindquist, 2002). These differences could be explained by speculating that only after leaving the ER membranes can PrP misfold and aggregate in the cytoplasm (Ma and Lindquist, 2001; Yedidia et al., 2001; Ma et al., 2002; Ma and Lindquist, 2002) and cause cell toxicity (Ma et al., 2002; Ma and Lindquist, 2002). We believe that in physiological conditions the untranslocated PrP precursor remains associated with the ER membranes until degraded by the proteasome. Thus, even when it accumulates after block of the proteasome, it fails to diffuse and to aggregate in the cytoplasm and therefore it is not able to promote cell death. By contrast, when the cytosolic localization of PrP is artificially forced by overexpressing a PrP construct lacking the signal peptide, as in the Lindquist model, aggregation and cytotoxicity are also induced (Ma et al., 2002).

In addition, in our hands, blocking of the proteasome does not lead to the accumulation of the misfolded PK-resistant form (Fig. 5D), indicating that the proteasome is not involved in the degradation of the PrP mutant scrapie-like toxic conformers. These results indicate that proteasomal degradation and cytosolic PrP do not have a relevant role in pathogenesis of prion diseases, contrary to what was previously postulated (Ma and Lindquist, 2001; Ma et al., 2002; Ma and Lindquist, 2002). Instead, other mechanisms that facilitate prion conversion, leading to accumulation and toxicity of PrP pathological mutants must be involved in the pathogenesis of inherited prion diseases.

It has been recently shown that PrP^Sc toxicity is associated with an increase of calcium release from the ER and with an up-regulation of ER chaperone levels (Hetz et al., 2003). This could promote ER stress, activation of an apoptotic pathway (Mattson et al., 2000) and cell death, a mechanism common to other neurodegenerative diseases (Nakagawa et al., 2000). By analogy, the pathogenic effects of PrP^Sc could be related to the presence of misfolded forms in the ER. In addition, the fact that stimulation of PrP^C retrograde transport into the ER increases PrP^Sc formation in infected cells (Beranger et al., 2002), supports a role for the ER in PrP scrapie-like conversion.

Interestingly, we have recently shown that a specific membrane environment of cholesterol-enriched microdomains in the ER could have a role in facilitating the correct folding of PrP^C (Sarnataro et al., 2004). In particular, the immature diglycosylated precursor form of PrP^C was recovered in DRMs in the ER and that it then became misfolded, as judged by the PK resistance assay, following cholesterol depletion (Sarnataro et al., 2004), indicating that DRM association in the early secretory pathway was crucial for the correct folding of PrP^C. In addition, several lines of evidence (Taraboulos et al., 1995; Naslavsky et al., 1999; Baron et al., 2002; Sanghera and Pinheiro, 2002; Kazlauskaite and Pinheiro, 2005) indicate that a raft-like lipid environment is involved in the generation of PrP^Sc in infected cells. However, the role of rafts in the scrapie-like conversion of PrP mutants has never been tested before.

To understand whether rafts, and in particular DRM association in the ER, might have a role in the scrapie-like conversion of the PrP mutant, we analysed whether PrPT182A was associated with DRMs and studied the characteristics of this association. We demonstrated that compared with the wt protein, a higher percentage of PrPT182A is associated with these microdomains in a cholesterol-dependent manner (Fig. 6A). Notwithstanding the higher DRM association of PrPT182A in the ER, this did not change the DRM partitioning of raft and non raft markers (e.g. calnexin and flotillin, not shown) neither did it impair the exocytic trafficking of endogenous proteins from the ER (not shown).

Perturbation of this association by cholesterol depletion leads to an increase of its half-life (Fig. 6B) and to its accumulation at steady state (Fig. 6C). Importantly, when PrPT182A DRM-association was perturbed the amount of the PK-resistant (scrapie-like) form was increased by approximately ninefold (Fig. 6C). Thus, our data indicate that, as for the immature PrP^C precursor (Sarnataro et al., 2004) the PrPT182A mutant also associates with DRMs in the ER in a cholesterol-dependent manner and that this association is required to prevent its pathological misfolding.

Our data on PrPT182A mutant accumulation and scrapie-like conversion in cholesterol depletion support the hypothesis that the raft environment is also necessary to stabilize the proper PrP conformation in the mutants (as shown by the fact that in normal conditions most of the protein is in rafts and is PK sensitive), therefore suggesting that prion conversion occurs outside rafts where PrP folding is destabilized and where misfolded PrP intermediates might be more prone to interact and aggregate. Our proposal of a protective role of detergent-resistant domains for PrP misfolding is also supported by the fact that binding of PrP to raft-like artificial membranes results in a stabilization of α-helical structures, while interaction with negatively charged lipid (non-raft) membranes increases β-sheet content (Critchley et al., 2004; Kazlauskaite et al., 2005).

The fact that the half-life of the mutant was increased in conditions of cholesterol depletion could be explained because the misfolded PrP formed in these conditions is more slowly degraded, or alternatively that a cholesterol- and/or raft-dependent mechanism of degradation exists and that this should contribute to misfolded PrP accumulation under cholesterol depletion. Thus, an alternative model for the role of rafts in PrP mutant scrapie-like conversion could be that they participate in the degradation of the PK-resistant mutant form indirectly, contributing to the accumulation of misconformed pathological forms when raft composition and function is affected.

Cell culture reagents were purchased from Gibco Laboratories (Grand Island, NY). The α-PrP antibodies PRI308 and SAF32 were a kind gift from J. Grassi (CEA, Saclay, France). Protein-A-Sepharose was from Pharmacia Diagnostics AB (Uppsala, Sweden). The antibodies against CNX and EEA1 were from StressGen Biotechnologies Corp. (Victoria, BC, Canada). The antibody against giantin was from BAbCO (Berkeley antibody Company, Richmond, CA). Lysotracker Red DND-99 was from Molecular Probes. The α-polyubiquitin antibody FK1 was from Affinity Research Products Ltd (Mamhead Castel, Mamhead, Exeter, UK). ALLN was from Calbiochem (La Jolla, CA). Methyl-β-cyclodextrin (βCD), mevinolin and all other reagents were obtained from Sigma Chemical Co. (St Louis, MO).

FRT cells were transfected with a cDNA encoding 3F4-tagged PrPT182A (a kind gift from Sylvain Lehmann, UPR CNRS1142, Montpellier, France), with the calcium phosphate procedure as previously described (Zurzolo et al., 1993).

FRT cells stably expressing PrPT182A were grown in F12 Coon's modified medium containing 5% FBS. Proteasomal block was performed with 150 μM ALLN for 7 hours in complete medium. Mevinolin and βCD treatments were carried out as described elsewhere (Sarnataro et al., 2002; Sarnataro et al., 2004). Cellular cholesterol levels before and after depletion were determined by a colorimetric assay (Infinity Cholesterol Reagent; Sigma Chemical Co. St Louis, MO) according to the suggested Sigma protocol number 401, as previously described (Sarnataro et al., 2004). The samples were then measured in a spectrophotometer at 550 nm.

Digestion with N-glycosidase F, endoglycosidase-H and neuraminidase (10 mU/sample) was carried out on lysates for 16 hours at 37°C. The lysates were first boiled for 3 minutes in 50 μl of 0.1 M sodium citrate, pH 5.5, containing 0.1% SDS, and then incubated with enzymes. The samples were then TCA precipitated, resuspended in SDS sample buffer, boiled and loaded onto 12% polyacrylamide gels and revealed by fluorography.

PIPLC (phosphatidyl inositol phospholipase C) treatment, Triton/DOC insolubility assay and proteinase K (PK) digestion assay were performed as previously described (Lehmann and Harris, 1997; Sarnataro et al., 2002).

FRT cells stably expressing PrP^C were grown for 4-5 days on both coverslips and transwell filters, washed with PBS, fixed in 2% paraformaldehyde, permeabilized with 0.075% saponin and processed for indirect immunofluorescence using specific antibodies. PrP^C was visualized with a TRITC-conjugated secondary antibody, CNX, giantin and EEA1 were revealed by FITC-conjugated secondary antibodies using a Zeiss laser scanning confocal microscope (LSCM 510). For lysosome staining, cells were incubated for 1 hour with Lysotracker (1:10,000) in complete medium before fixing.

Membrane topology of PrPT182A was determined as previously described (Drisaldi et al., 2003). Cells were lysed in 0.25 M sucrose and 10 mM Hepes (pH 7.4) by 10 passages through 26-gauge needles. The post-nuclear supernatant was divided into three samples: one untreated, the second digested with 250 μg /ml PK for 30 minutes at 22°C in 50 mM Tris-HCl (pH 7.5) and the third digested with PK at the same concentration in the presence of 0.5% Triton X-100 (TX-100). Samples were TCA precipitated and PrPT182A was analysed by western blotting.

Cells were lysed in 0.25 M sucrose, 10 mM Hepes (pH 7.4) by 10 passages through 26-gauge needles. After centrifugation at 2,300 g for 2 minutes, the post-nuclear supernatant was ultracentrifuged at 70,000 g for 30 minutes at 4°C in a TLA 100.2 rotor using a Beckman Optima TL ultracentrifuge. The pellet was resuspended in Triton/DOC buffer (0.5% Triton X-100, 0.5 sodium deoxycholate, 150 mM NaCl and 100 mM Tris pH 7.5) for PrP analysis, or in SDS buffer and treated as described below for polyubiquitylated protein analysis. Samples were TCA precipitated and proteins were analysed by western blotting.

Cells were lysed in 200 μl of SDS buffer (0.3% SDS, 10 mM Tris-HCl pH 7.5, 150 mM NaCl) and incubated for 10 minutes at 65°C. After addition of 2% TX-100, the samples were incubated at 65°C for 2 minutes. After addition of 800 μl TNS buffer (10 mM Tris-HCl pH 7.5, 150 mM NaCl) proteins were TCA precipitated or immunoprecipitated for the analysis on first or second dimension as follows. Immunoprecipitated (IP) samples were boiled in reducing sample buffer and resolved on 10% SDS-PAGE (first dimension). The polyacrylamide gel lane corresponding to the sample of interest was cut and boiled in reducing sample buffer for 5 minutes at 95°C and then run on 10% gels (second dimension) and analysed by western blotting (Friedrichson and Kurzchalia, 1998).

Cells grown to confluence in 150 mm dishes were harvested in cold PBS and resuspended in 1 ml lysis buffer (1% TX-100, 10 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM EDTA), left in ice for 20 minutes and Dounce homogenized (10 strokes, tight). The cell lysate was centrifuged (5 minutes, 1,300 g) to remove nuclei. Post-nuclear supernatants were mixed with an equal volume of 85% sucrose (wt/vol) in 10 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM EDTA, placed at the bottom of a discontinuous sucrose gradient (30-5%) in the same buffer and ultracentrifuged at 200,000 g for 17 hours at 4°C in an ultracentrifuge (SW41 rotor from Beckman Instruments, Fullerton, CA). Twelve fractions were harvested from the top of the gradient. A white-light-scattering band, identified in fraction 5 at the interface between 5% and 30% sucrose, contained DRM domains. Samples were TCA precipitated and proteins were analysed by western blotting.

Control, ALLN-treated or cholesterol-depleted cells plated in 60 mm dishes were pulsed for 20 minutes with 100 μCi/ml [³⁵S]methionine and chased for various times at 37°C. At the end of the chase times cells were washed with cold PBS and lysed for 20 minutes on ice in Triton/DOC or in SDS buffer. PrPT182A immunoprecipitation was performed overnight using the α-PrP SAF32 antibody coupled to protein-A-Sepharose beads. The pellets were washed twice with cold lysis buffer and three times with PBS. For Endo-H treatment the IP samples were divided and were first boiled for 3 minutes in 50 μl of 0.1 M sodium citrate pH 5.5 containing 0.1% SDS and then treated with enzymes as indicated above. The samples were then boiled with SDS sample buffer, loaded on 12% polyacrylamide gels and revealed by phosphorimager scanning.

We thank Sylvain Lehmann for the kind gift of PrPwt and PrPT182A cDNA and for helpful discussions. We also thank Jacques Grassi for the generous gift of several α-PrP antibodies and Chris Bowler for critical reading of the manuscript. V.C. received a fellowship (2003-04) from the Foreign Researchers from the French Foreign Affairs Ministry. D.S. is the recipient of a fellowship from FIRB 2001 (NE01S29H). This work was supported by Grants to C.Z. from MURST (PRIN 2004), from the Telethon Foundation (GGP0414), from the European Union (QLK-CT-2002-81628), from the Weizmann-Pasteur Foundation and from FRM (Fondation pour la Recherche Médicale).