Failure of Prion Protein Oxidative Folding Guides the Formation of Toxic Transmembrane Forms*

Background: In vivo folding could play an essential role in prion neurodegenerations. Results: Artificial mutants causing labile PrP folds when expressed in cells originate toxic CtmPrP featured by the absence of the intramolecular disulfide bond. Conclusion: Oxidative folding impairment facilitates the formation of the toxic PrP forms. Significance: Unveiling the mechanism facilitating the formation of toxic PrP forms is crucial for the understanding and prevention of prion disorders. The mechanism by which pathogenic mutations in the globular domain of the cellular prion protein (PrPC) increase the likelihood of misfolding and predispose to diseases is not yet known. Differences in the evidences provided by structural and metabolic studies of these mutants suggest that in vivo folding could be playing an essential role in their pathogenesis. To address this role, here we use the single or combined M206S and M213S artificial mutants causing labile folds and express them in cells. We find that these mutants are highly toxic, fold as transmembrane PrP, and lack the intramolecular disulfide bond. When the mutations are placed in a chain with impeded transmembrane PrP formation, toxicity is rescued. These results suggest that oxidative folding impairment, as on aging, can be fundamental for the genesis of intracellular neurotoxic intermediates key in prion neurodegenerations.

Prion disorders are dominant gain-of-function neurodegenerations whose pathogenesis is linked to misfolded forms of the cellular prion protein (PrP C ), 3 including the prion PrP Sc and the neurotoxic CtmPrP (1)(2)(3)(4). PrP Sc is an aggregated and proteaseresistant ␤-sheet-enriched conformer of PrP C , which self-perpetuates by the templating the conversion of cell surface PrP C (1,4). In contrast, CtmPrP is an intracellular transmembrane form generated at the ER with neurotoxic properties (1,5,6). Despite that CtmPrP formation was associated with features of the ER translocation process several pathogenic mutations in the C-terminal domain such as H187R and E200K enhance its levels, suggesting a yet unexplored in vivo interplay between folding and the accumulation and action of this neurotoxic form (6 -8).
Since the enunciation of the prion hypothesis, research has focused on the mechanism by which a native PrP C structure reorganizes and acquires self-propagative features like those of PrP Sc (9 -11). The PrP C native state was assigned to the fold adopted by the chain lacking the signal sequences and containing the disulfide bond and used as reference for testing the effect of pathogenic mutations and its conversion into active prions (10,(12)(13)(14)(15)(16). However, the in vivo folding of proteins segregating into the secretory route such as PrP is a complex process participated by the ER folding machinery. This machinery coordinates processing (signal sequences removal, addition of covalent modifications, binding of cofactors, etc.), avoids undesired aggregations, and permits the acquisition of correct structure. This global process involves multiple transient protein-protein interactions with the nascent chains that can sense alterations resulting from environmental changes to the presence of mutations (17)(18)(19). Any variation in the sequence of events can impact the final product, as for the doses of secretory and transmembrane PrP forms, and its fate (6, 7, 20 -33).
Metabolic studies addressing the effect of pathogenic mutations in the C-terminal domain of PrP as disease predisposition factors have reported a wide range of alterations in processing, trafficking, aggregation, accumulation, and toxicity which varied among experimental setups, as the cell line used and the background expression of wild-type (WT) PrP C (20, 21, 23-26, 28, 29, 31, 33). These aberrancies contrast with structural reports in which the same pathogenic mutations do not impede the correct in vitro folding, but variably modify the stability, dynamics, and surface reactivity of the native state (12-16, 34, 35). Indeed, aging factors such as oxidative modifications and This article has been withdrawn by the authors. We have become aware of errors in the preparation of the following figures. There were several duplications in Fig. 2A. exhaustion of the ER folding machinery which are not considered in structural studies may play fundamental roles in the formation of pathogenic PrP.
Of the different mutations in the globular domain experimentally tested, substitutions of conserved methionines in ␣-helix 3 (hitherto PrP␣3M) provoked the largest ␣-fold destabilization (36). In particular, singly or combined M206S and M213S replacements in rHaPrP(23-231) yielded extremely labile folds with enhanced aggregation capacity (36). These mutations also mimicked the flexibility distortions impinged by sulfoxidation of such methionines found in the PrP chains in the conversion pathway (12-14, 36 -39). Despite these interesting results, the effect of these substitutions had not been addressed in living systems.
Here, we have used various cultured cells expressing PrP␣3M mutants to investigate and model the role of in vivo folding in the synthesis and accumulation of PrP forms. Unexpectedly, we found that the PrP␣3M expression is highly toxic and that such toxicity relates to the exclusive formation of Ctm-PrP due to impeded disulfide bond formation.  Table 1.

Plasmid Construction and PrP Mutant Preparation-The
Cell Culture, Transfections, and Viability Assays-The cell lines HpL 3.14 (PrP Ϫ/Ϫ , mouse hypothalamic), GpL (PrP Ϫ/Ϫ mouse glial), N2a (mouse neuroblastoma), and CHO (Chinese hamster ovary) cells were kept in Opti-MEM containing 10% fetal bovine serum and penicillin/streptomycin (40,(43)(44)(45). Transient transfections with the different PrP-coding plasmid, pEGFP-rab9 (positive control for transfection), and pcDNA3.1/ pcDNA4.1 (mock control) were performed using FuGENE 6 transfection reagent (Roche Applied Science) according to the manufacturer's instructions. Typically, 24 h after transfection the medium was exchanged and allowed for other 24 h before analysis. For viability assays, cells were harvested using trypsin/ EDTA in PBS followed by centrifugation and stained with 0.4% trypan blue for 5 min at room temperature. Total and viable cells were determined using the TC10 automated cell counter (Bio-Rad) using duplicate independent readings of experiments performed in duplicates. Displayed data are the mean Ϯ S.D. of three independent experiments. Statistical analyses were performed using the t test, with significance set to p Ͻ 0.05. For other analysis, cells were harvested by either in situ lysis in cold radioimmune precipitation assay buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10 mM EDTA, 0.5% Triton X-100, 0.5% deoxycholate) or by detachment with PBS containing 10 mM EDTA.
Expression Analysis, Detergent Solubility Assay, and Proteinase K Treatments-Cell lysates were cleared by a centrifugation of 5 min at 500 ϫ g, supplemented with 0.5 mM Pefabloc, and then precipitated with 5 volumes of methanol at Ϫ20°C. Samples were centrifuged at 10,000 ϫ g for 30 min, and the pellets were redissolved in TNE buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA). For expression studies, aliquots of the TNE-resuspended samples were diluted with Laemmli buffer and analyzed by immunoblotting. For solubility assay, the cleared cell lysates were supplemented with sarkosyl to 1% and centrifuged for 1 h at 100,000 ϫ g, 4°C, in a Beckman Optima TM Max centrifuge. Soluble fractions (supernatant) were precipitated with methanol and then together with the insoluble fractions (pellet) were analyzed by immunoblotting. For protease digestions, aliquots of cleared lysates prepared in the absence of Pefabloc were incubated for 30 min at 37°C with 20 g/ml proteinase K (PK; Promega); the proteolysis was stopped by addition of 5 mM protease inhibitor PMSF. Samples were precipitated with methanol and analyzed by immunoblotting. For protease protection assays, detached cells were homogenized in 10 mM Tris-HCl, pH 7.4, 0.1 M sodium acetate, 2 mM MgCl 2 and centrifuged at 1000 ϫ g at 4°C for 10 min. The supernatants were then centrifuged for 1 h at 100,000 ϫ g at 4°C, and the resulting pellets (microsomes) were resuspended in 50 mM Hepes, pH 7.4, 0.1 M sodium acetate, 2 mM MgCl 2 , 0.25 M sucrose. Microsome suspensions were split in three equal parts and incubated in the absence and presence of PK (5 g/ml), both with and without 0.5% Triton X-100, for 30 min at 25°C. Reactions were stopped by adding 0.5 mM PMSF and analyzed by immunoblotting.
Post-translational Covalent Modification Analysis: Glycosylation, GPI Addition, and Disulfide Bonding-Enzymatic digestions with PNGase F (New England Biolabs), Endo H (New England Biolabs), and phosphatidylinositol phospholipase C (PIPLC) (Sigma-Aldrich) were performed for 1 h at 37°C on the methanol-precipitated cell lysates, following the manufacturer's instructions. After digestion, reactions were stopped by the addition of Laemmli buffer, and PrP was analyzed by immunoblotting using 3F4 antibody. For analysis of disulfide bonds, cells were lysed in 50 mM Tris-HCl, pH 8, 1% SDS, cleared by a slow speed centrifugation, and then supplemented with or without 200 mM DTT. Both reduced and nonreduced lysates were then incubated with 100 mM Oregon Green 488 iodoacetamide (Invitrogen) for 15 min, diluted with 10 volumes of radioimmune precipitation assay buffer, and incubated with protein A-Sepharose beads (GE Healthcare) for 60 min at 4°C. After centrifugation, supernatants were incubated with mAb 3F4 for 10 h at 4°C. Protein A-Sepharose beads were then added, and after a 90-min incubation at 4°C, the protein-antibody complexes bound to protein A-agarose were sedimented by centrifugation. Pellets were washed with a buffer containing 150 mM NaCl, 10 mM Tris-HCl, pH 7.8, 0.1% sarkosyl, and 0.1% Pefabloc, and bound proteins were eluted by boiling in SDS-sample buffer. Precipitates were analyzed by SDS-PAGE and developed with goat anti-fluorescein/Oregon Green Ab (Invitrogen).
Fluorescence Microscopy Imaging-Cells were seeded on to poly-L-lysine-coated glass coverslips, transfected with the plasmids coding for HaPrP WT, HaPrP-YFP WT, and their mutants, and grown for 36 h to 60% confluence. For immunofluorescence analysis, cells were fixed with 4% paraformaldehyde in PBS containing 5% sucrose for 10 min at room temperature and washed three times with PBS. Cells were permeabilized and blocked in PBS containing 0.5% saponin, 0.1% Triton X-100, and 2% bovine serum albumin for 10 min at room temperature. Cells were incubated with anti-PrP mAb 3F4 (1:600) and with anti-␤ coatomer protein (1:600) for 1 h at room temperature. After washing with blocking buffer, samples were incubated with Alexa Fluor 488-conjugated goat antimouse IgG (1:800), Alexa Fluor 647-conjugated anti-rabbit IgG (1:800), and Hoechst 33342 (10 g/ml) in for 30 min at room temperature. Once washed, the coverslips were mounted on glass slides with ProLong Gold antifade reagent (Invitrogen). Images were captured with a confocal microscope (Leica TCS-SP-AOBS-UV) as described (40). For living cell analysis, imaging was performed as described previously (42).  (Fig. 1A). We also generated control substitutions at other conserved methionines (M134S and M154S), a substitution that preserves both the conformation and stability of the recombinant chain (M213L) and a pathogenic mutant known to alter in vivo PrP folding (A117V) (2, 3, 6, 36) (Fig. 1A). We expressed all of these constructs in PrP Ϫ/Ϫ cells (HpL and GpL) to avoid interference from endogenous WT PrP chains (Fig.  1A). Transient expression of HaPrP constructs revealed specific cytotoxicity of PrP␣3M mutants in both cell types (Fig. 1B). HaPrP WT and the control M134S, M154S, and M213L mutations all had no effect on cell viability, whereas HaPrP␣3M mutants caused approximately 40% cell death in both HpL and GpL cells upon transient transfection (Fig. 1B). Interestingly, HaPrP A117V also induced 30% cell death, suggesting possible similarities in the pathogenic process (Fig. 1B). PrP␣3M mutations generated as HaPrP-YFP fusions (42) showed the same cell loss as those on HaPrP background, arguing that YFP does not interfere in the lethal process (Fig. 1B). Next, we tested the toxicity of PrP␣3M mutants using MoPrP sequence. As with the HaPrP constructs, we found that the three PrP␣3M mutants induced approximately 40 -50% cell death (Fig. 1C). Thus, the scaffold sequence had no effect on PrP␣3M toxicity, further supporting the crucial role of these conserved residues on the in vivo PrP folding. Importantly, toxicity increases with the dose of PrP␣3M expression, as shown for variation of cell death relative to the normalized HaPrPDS signal as a function of the time after transfection (Fig. 1D).

PrP␣3M Mutants Are Highly Toxic in Cultured Cells-To
Having seen that PrP␣3M mutants are highly toxic to culture cells not expressing an endogenous PrP, we wanted to study whether expression of WT PrP affects this phenotype. We coexpressed HaPrP and MoPrP bearing the double mutant (PrPDS) with the corresponding WT PrP in a 1:1 ratio in GpL and HpL cells to determine whether this could ameliorate the toxic effect. Fig. 1E shows that PrP WT had no or little effect on PrPDS toxicity in this setup. Next, we transiently expressed both HaPrPDS and MoPrPDS in N2a cells, which express MoPrP C , and in nonneuronal CHO cells, which lack detectable PrP (Fig. 1F). Again, there was substantial toxicity in tested cells, which was apparently independent of endogenous WT PrP expression and cell type used. Altogether, these results indicate that PrP␣3M mutants exert a very high level of cell toxicity upon transient expression which is not influenced by cell type and co-expression of WT PrP.
PrP␣3M Mutants Are Processed Abnormally during Biogenesis-After determining that PrP␣3M mutants are highly toxic, we set out to characterize the molecular basis of such toxicity. Expression of WT and the control M134S and M154S mutants produced the classical 24 -38-kDa banding pattern for both HaPrP and MoPrP constructs and in both PrP Ϫ/Ϫ cell lines ( Fig. 2A). However, expression of PrP␣3M mutants yielded single bands of approximately 30 kDa (Fig. 2A). This feature has been described previously for PrP pathogenic mutants with immature glycosylation (6,20,22,25,30,32,46,47). This stimulated us to study the glycosylation of PrP␣3M mutants, focusing on HaPrPDS as a model. Indeed, whereas WT HaPrP was sensitive to PNGase F and resistant to Endo H digestions, the glycan attached to HaPrPDS was sensitive to both PNGase F and Endo H digestion (Fig. 2B). These results indicate an immature glycosylation of the PrPDS mutant. To test for the presence of a GPI anchor, the PNGase F-deglycosylated products were further digested with PIPLC. The upward shift in the bands of WT HaPrP and HaPrPDS after PIPLC treatment indicates the removal of the hydrophobic moiety and supported the presence of the GPI anchor in both chains (Fig.  2B). Also, the size similarity of WT HaPrP and HaPrPDS bands upon combined PNGase F and PIPLC digestion indicates that in both PrP chains the N-terminal signal sequence has been removed. Taken together, the post-translational processing of HaPrPDS, representative here of PrP3␣M mutants, indicates removal of the N-terminal signal sequence, immature glycosylation, and the presence of a GPI anchor.
PrP␣3M Mutants Accumulate as CtmPrP-The above biochemical features were reminiscent of previously described de novo formation of either PrP Sc -like or CtmPrP forms (6,40,41,48). To assess the conformation adopted by PrP␣3M mutants, we analyzed the detergent solubility and the resistance to proteases of HaPrPDS (1, 6). Fig. 2C shows that WT HaPrP proved completely soluble in detergents whereas HaPrPDS partitioned almost entirely into the insoluble fraction. Nonetheless, both proteins were fully digested by PK under harsh (37°C) and mild (4°C) digestion conditions, supporting the absence of prototypic PrP Sc characteristics (Fig. 2D).
Once discarded the generation of bona fide and PK-resistant PrP Sc , we next analyzed the membrane topology of HaPrP␣3M, using HaPrPA117V as control for acquiring CtmPrP topology. For this, we purified microsomes from transfected cells to carry out protease protection assays (Fig. 2E). HaPrP WT and its M134S, M154S, and M213L mutants were fully protected under mild conditions (PK digestion without detergent) but completely digested in the presence of detergents, supporting the production of the secretory PrP form. In contrast, under mild digestion conditions HaPrP A117V revealed a protease-protected fragment of approximately 21 kDa, corresponding to the ERlocalized C-terminal domain characteristic of CtmPrP (7). Digestion of microsomes from cells expressing HaPrPM206S, HaPrP213S, and HaPrPDS yielded the same 21-kDa fragment as identified for HaPrP A117V. In summary, these results indicate that HaPrP␣3M folding results in acquisition of a transmembrane CtmPrP topology.
PrP␣3M Mutants Are Retained Intracellularly and Exert ER Stress-Because CtmPrP is mainly an intracellular conformer and has been proposed to elicit an ER stress response, we then tested these properties for HaPrP␣3M (6,8,49). First, we performed indirect fluorescence microscopy studies to address the subcellular location upon transient transfection into HpL cells. As expected, WT HaPrP and its control M134S and M213L mutants were located along the secretory pathway in route to the plasma membrane (Fig. 3, A-C). In contrast, HaPrP␣3M mutants were found mostly intracellularly, with a punctuate distribution overlapping partially the Golgi (co-staining with the marker ␤COP) and ER (co-staining with the marker PDI; data not shown) membranes (Fig. 3, E-G). We also analyzed the distribution of the corresponding YFP fusion constructs. Again, WT HaPrP-YFP, HaPrPM134S-YFP, and HaPrPM213L-YFP were found mainly at the plasma membrane with some Golgi staining indicating a normal behavior (Fig. 3, J-L). However, the HaPrP HaPrP␣3M-YFP fusions, which retained toxicity (see Fig. 1B), accumulated intracellularly displaying an excess of intracellular vesicle structures containing PrP which most likely represent ER compartments (Fig. 3, N-P). Thus, the abnormal intracellular distribution of PrP␣3M mutants observed here by microscopy is in line with our biochemical analysis showing immature glycosylation, tendency to aggregate, and CtmPrP topology.
To test whether this intracellular PrP distribution was associated with ER stress, we tested the levels of key components of the unfolded protein response pathway following the expression of HaPrP mutants. Expression of WT HaPrP and of its control M134S, M154S, and M213L mutants did not affect the levels of ER chaperone BiP/Grp78 and of the apoptosis inducer transcription factor CHOP (GADD153) (Fig. 4). However, expression of all PrP␣3M mutants along with the A117V construct resulted in a significant increase of BiP/Grp78 and CHOP levels (Fig. 4). We also analyzed the levels of PDI, an ER chaperone that catalyzes thiol-disulfide exchanges and assists in disulfide bridge formation during the folding of secretory proteins. We found that the levels of PDI did not differ signifi-  OCTOBER 26, 2012 • VOLUME 287 • NUMBER 44 cantly between WT HaPrP and all of the mutants tested (Fig. 4). Taken together these results show the selective activation of the ER stress response upon accumulation of CtmPrP conformers.

Toxic CtmPrP Lacks Disulfide Bond
PrP␣3M Mutants Lack the Conventional Intramolecular Disulfide Bond-A similar toxicity as observed here for the PrP␣3M mutants has been described previously for the MoPrPC213A mutant and ascribed to a failure of the oxidative folding (30,50). Indeed, this mutant lacks the intramolecular disulfide bond which is a key determinant for the stability of the ␣2-␣3 subdomain both in vivo and in vitro (22,30,32,51,52).
To understand better the similarities between the mutant lacking the disulfide bond and the PrP␣3M mutants here, we generated the HaPrPC214A mutant, analyzed its toxicity, and tested whether it accumulates as CtmPrP. Fig. 2A shows that expression HaPrPC214A is as cytotoxic as that of HaPrP␣3M mutants, in agreement with previous observations (30). On the other hand, Fig. 2E shows that microsomes of cells expressing HaPrPC214A behaved in the protein protection assay as micro-somes from PrP␣3M and A117V mutants. These data support that the high toxicity of all these mutants is related to the adoption of CtmPrP topologies.
Then, we analyzed the state of the intramolecular disulfide bond for the PrP␣3M mutants using the differential reactivity of free and oxidized thiol groups with Oregon Green-iodoacetamide (Fig. 5A). Lysates of transfected cells which we first denatured in the absence and presence of DTT were then treated with Oregon Green-iodoacetamide for the irreversible labeling of free thiol groups. Next, PrP was immunoprecipitated with mAb 3F4 and immunoblotted with anti-Oregon Green/fluorescein antibody to detect only the fraction with free thiols. This assay clearly showed that WT HaPrP and its M134S, M154S, and M213L mutants formed stable disulfide bonds and did not interact with Oregon Green-iodoacetamide in the absence of DTT (Fig. 5). In contrast, the PrP␣3M, A117V, and C214A mutants were labeled with Oregon Green-iodoacetamide in the absence of DTT (Fig. 5B). Thus, these data suggest that a key structural feature of toxic mutants forming CtmPrP could be the presence of free thiols. To confirm this observation, we studied the effect of the DS mutation on HaPrPG123P, a mutant with described impaired capacity to generate transmembrane forms (6,20). HaPrPG123PDS shared with HaPrPDS the expression as a single band, but behaved as HaPrPG123P in topology assay (Figs. 2, F and G, 3, and 4). Importantly, HaPrPG123PDS expression produced cell viability values similar to those of WT HaPrP and HaPrPG123P. Taken together, these data support a direct relation between the lack of intramolecular disulfide bond and the formation of toxic CtmPrP.

DISCUSSION
The efficient production of secreted PrP C starts with the cotranslational translocation to the ER followed by a series of concerted processing including the cleavage of the N-terminal signal sequence, the addition of glycan chains at two facultative sites, the formation of an intramolecular disulfide bond, and a transamidation at the C terminus to add the GPI moiety (1). Of these post-translational modifications, glycosylation and disulfide bond formation depend on the cellular redox state (22,27). Impairing the ER oxidative environment or mutating the cysteines yields intracellular, diglycosylated PrP chains lacking the disulfide bond, which resembles the PrP␣3M mutants described here (22,27,30,32). In this work we demonstrated the interplay between oxidative folding and the formation of the toxic CtmPrP. Introduction of polar substitutions at the ␣3 methionines, singly or combined, precluded the formation of the intramolecular disulfide bond and dictated the stabilization of highly toxic CtmPrP topologies that killed cells. Importantly, the absence of the disulfide bond in CtmPrP and its escape from the quality control checkpoints support the impairment of the oxidative folding as key pathogenic event for the of toxic PrP forms during normal aging.
CtmPrP has been traditionally viewed as a transmembrane form whose translocation is governed by the hydrophobic region, resulting in a cytosolic N-terminal domain containing the signal sequence (2,3,49). However, recent experiments in cell cultures showed that CtmPrP lacked the N-terminal signal sequence, supporting a model of transbilayer post-translocation slippage as observed with recombinant PrP lacking disulfide bond and lipid vesicles (6,53,54). Conditions favoring the slippage by trapping the hydrophobic region at the translocon as in A117V may either kinetically delay or impose a distance constraint impeding the formation of the disulfide bond. Also, mutations that may hinder oxidoreductase recognition such as ␣3M and Y216A may favor the insertion of the N-terminal domain as shown with recombinant chains lacking a disulfide bond (30,53,54).
Our results also suggest that conditions preventing PrP oxidative folding may favor the formation of CtmPrP and, consequently, promote its deleterious effects. Indeed, both maintenance of the levels and activity of ER oxidoreductases such as Grp58 and PdIa protect against the toxicity of misfolded PrP forms (55,56). On aging, both levels and activity of ER chaperones cause a decline of oxidative folding (57). Because PrP3␣M mutants were tailored to mimic the effects of Met sulfoxidation, conditions favoring such modification on nascent chains would also facilitate the formation of CtmPrP. In this line, ER stress exhaustion concurs with an overproduction of reactive oxygen species which are the major oxidants of Met residues, as those exposed in nascent unfolded chains (58). Taken together, both decreased efficiency of the oxidative folding and increased probability of sulfoxidation of Met may then explain the agingassociated accumulation of CtmPrP forms in sporadic and inherited diseases.
The role of disulfide bonds in PrP biology has mainly focused on their contribution to the generation of prions. The analysis of highly pure preparations of PrP27-30, the protease-resistant core of PrP Sc , indicated that all Cys were forming disulfide bonds (2,59). But those samples lack CtmPrP forms due to the proteinase K digestion step used in the purification. The first evidence indirectly suggesting a pathogenic role for free thiols was provided by deletion mutants (60). Mice expressing PrP⌬177-200 and PrP⌬201-217 that are unable to forms intramolecular disulfide bonds developed signs and lesions characteristic of neuronal storage disorders. In these mice, truncated PrP was detergent-insoluble in detergent, PK-sensitive, and with migration properties resembling those of PrP␣3M mutants. Also, Cys mutants used in cellular studies resulted in PrP forms sharing key properties with PrP␣3M mutants, but their topology and toxicity were not addressed (22,27,30,32).
The finding that free thiols lead to the formation of CtmPrP has several implications. From the structural point of view, the C-terminal domain has to expand its known conformational repertoire to accommodate the absence of the ␣2-␣3 constraint and a double tether to the membrane (51,52,61). Also, the fibrillation of CtmPrP may be impeded by the diglycosylation (52,62). But the detergent insolubility of CtmPrP suggests that it may populate distinct aggregate states, adding more structural complexity. Functionally, the aggregation of PrP␣3M mutants suggests that CtmPrP may exert its toxic function through an oligomeric thiol trap. Such traps have been described in the regulation of IgM and adiponectin secretions (63)(64)(65)(66)(67)(68)(69). Importantly, given the validation of the PrP chains by the ER quality control systems the assembly of such oligomeric traps must take place upon delivery to a different environment (20). Whether other components participate in their assembly and/or stabilization and whether these factors display different affinities for the WT or mutant chains remains to be elucidated (6). Additionally, this unknown lipid-bound conformation of thiol-free PrP could indeed function as the seed for the conversion of PrP C into PrP Sc , even in traces amounts (70,71). This could provide a mechanistic explanation for the spontaneous generation of pathogenic forms of PrP in sporadic human diseases.