Prion Protein Aggregation Reverted by Low Temperature in Transfected Cells Carrying a Prion Protein Gene Mutation*

Prion diseases are characterized by the conversion of the normal cellular prion protein (PrPC), a glycoprotein that is anchored to the cell membrane by a glycosylphosphatidylinositol moiety, into an isoform that is protease-resistant (PrPres) and pathogenic. In inherited prion diseases, mutations in the prion protein (PrPM) engender the conversion of PrPM into PrPres. We developed a cell model of Gerstmann-Sträussler-Scheinker disease, a neurodegenerative condition characterized by PrPM-containing amyloid deposits and neuronal loss, by expressing the Gerstmann-Sträussler-Scheinker haplotype Q217R-129V in human neuroblastoma cells. By comparison to PrPC, this genotype results in the following alterations of PrPM: 1) expression of an aberrant form lacking the glycosylphosphatidylinositol anchor, 2) increased aggregation and protease resistance, and 3) impaired transport to the cell surface. Most of these alterations are temperature-sensitive, indicating that they are due to misfolding of PrPM.

All prion diseases are believed to share the same basic pathogenic mechanism that involves the conversion of a normal protein called cellular prion protein (PrP C ) into a form that is partially resistant to proteases (PrP res ) and is infectious (2). PrP C , a glycoprotein encoded by a gene (PRNP) located on human chromosome 20 (3), undergoes the non-obligatory addition of one or two N-linked oligosaccharide chains leading to the expression of three glycoforms containing none, one, or two glycans (4 -6) and the linking of a glycosylphosphatidylinositol (GPI) anchor at the C terminus (7,8). PrP C is transported to the plasma membrane through the secretory pathway and is then reinternalized and cleaved in the endocytic compartment to generate N-terminal and C-terminal fragments, the latter being returned to the cell surface (9,10). The conversion of PrP C to PrP res entails the switch from a predominantly ␣helical to a ␤-sheet conformation resulting in a form that, beside being protease-resistant and infectious, is insoluble in non-ionic detergents, has the propensity to aggregate, and can polymerize into amyloid fibrils (2).
In sporadic prion diseases, PrP res is thought to form initially as the result of a spontaneous stochastic event, and the conversion process would then be maintained autocatalytically by the endogenous PrP res . In inherited prion diseases, it has been hypothesized that the mutant PrP (PrP M ) spontaneously converts into PrP res (11).
A distinctive feature of inherited prion diseases is the phenotypic heterogeneity (1). One determinant of this heterogeneity is the PRNP codon 129, the site of a common methionine (129M)/valine (129V) polymorphism (12). Therefore, the PRNP determinant of the phenotype in inherited prion diseases is the haplotype established by the pathogenic mutation and codon 129. The Q217R-129V haplotype, which segregates with GSS, is associated with a phenotype characterized by prominent PrP deposits, often with the characteristics of amyloid, which contain PrP res forms of all sizes but especially N-terminal and C-terminal truncated forms (13).
To understand the individual steps involved in the pathogenesis of inherited prion diseases, we have generated several transfected cell models using a human neuroblastoma cell system to characterize PrP M . In this study we report on the metabolism of the PrP M expressed in cells transfected with the Q217R-129V PRNP construct. We observed complex changes in the biogenesis of PrP M which include the following: 1) expression of aberrant PrP M forms, some of which lack the GPI anchor and are retained in the endoplasmic reticulum (ER)-cis Golgi compartment; 2) presence of aggregated forms that exhibit decreased protease sensitivity in intracellular compartments and fail to reach the plasma membrane; and 3) underrepresentation of PrP M at the cell surface. Remarkably, these changes largely revert at reduced temperature, suggesting that abnormal folding of PrP M plays a major role in the pathogenesis of this GSS variant. phatidylinositol-specific phospholipase-C (PI-PLC) was purified according to Deeg et al. (14) or donated by T. Rosenberry (Case Western Reserve University).
Cell cultures were maintained at 37°C in Opti-MEM supplemented with 5% fetal calf serum and penicillin/streptomycin, in a humidified atmosphere containing 5% CO 2 . Cultures of transfected cells were supplemented with 500 g/ml hygromycin. Human neuroblastoma cells (M17) expressing normal (Q217-129V) or mutant (Q217R-129V) PrP were generated as described (15). Transfected cells were maintained in selective medium containing hygromycin. Experiments were performed at different times post-transfection on bulk-selected cells. For all experiments, cells were replated overnight and used at 90 -95% confluency. The following antibodies were used: anti-N, rabbit antiserum to synthetic peptide corresponding to human PrP residues 23-40 (B. Ghetti, Indiana University); 3F4, a monoclonal antibody that recognizes an epitope on human PrP residues 109 -112 (R. Kascsak SDS-PAGE and Western Blotting-In a typical experiment, 9 ϫ 10 6 cells were used for each condition. An equal amount of total protein was used from cells expressing either normal or mutant PrP. Protein concentrations were determined with bicinchoninic acid according to the manufacturer (Pierce), using bovine serum albumin as a standard. To detect PrP molecules in cellular extracts, cells were rinsed with PBS and lysed in a buffer containing 0.5% Nonidet P-40, 0.5% deoxycholate, and 10 mM EDTA in Tris-buffered saline (TBS, Tris 20 mM, NaCl 150 mM, pH 7.4), containing 10 g/ml each of leupeptin, antipain, pepstatin, and 1 mM phenylmethylsulfonyl fluoride (PMSF). Cell debris was cleared by centrifugation at 290 ϫ g, and the protein in the supernatant was precipitated with 5 volumes of cold methanol at Ϫ20°C. Cellular proteins were fractionated by SDS-PAGE and electrophoretically transferred to Immobilon-P (Millipore) for 2.5 h at 70 V at 4°C. Membranes containing transferred proteins were blocked in TBS containing 10% non-fat dry milk and 0.1% Tween 20 for 1 h at 37°C and probed with anti-PrP antibodies (anti-N diluted 1:4000, 3F4 diluted 1:50,000, or anti-C diluted 1:3000) dissolved in antibody dilution buffer (TBS, 1% normal goat serum and 0.05% bovine serum albumin) as described (18). Immunoreactive bands were detected with the appropriate secondary antibody conjugated to horseradish peroxidase (anti-rabbit diluted 1: 3000 and anti-mouse diluted 1:4000) and visualized on autoradiographic film by enhanced chemiluminescence (Amersham Corp.). To quantitate the relative density of immunoreactive bands, exposed autoradiographic film was scanned at 42-mm resolution with a GE10 densitometer and quantitatively analyzed using Quantity One software (PDIG20, QS30). To detect PI-PLC-cleaved PrP in the culture medium, cells were treated with PI-PLC as described below, and the released PrP was methanol-precipitated and detected as described above.
PI-PLC Digestion-Cells plated overnight in 10-cm tissue culture dishes were washed with Opti-MEM without serum and incubated with 59 ng/ml PI-PLC in fresh medium for 30 min at 37°C. The medium was collected and centrifuged at 4°C for 10 min at 290 ϫ g to pellet cell debris. PI-PLC-released proteins in the supernatant were precipitated with 5 volumes of cold methanol at Ϫ20°C for 2 h and fractionated by SDS-PAGE, and the PrP was detected by immunoblotting as described above.
Confocal Immunofluorescence Microscopy-For immunofluorescent staining of cell surface PrP, cells were grown to 50% confluency on poly-D-lysine-coated glass coverslips 18 -24 h before staining. Cells were rinsed with PBS containing 0.1% bovine serum albumin and incubated with 3F4 antibody (diluted 1:25) at 4°C for 35 min. Cells were then rinsed three times with PBS and incubated with FITC-conjugated secondary antibody for an additional 35 min at 4°C. After thorough rinsing with PBS, cells were fixed in 3% paraformaldehyde for 15 min and mounted on a glass slide in 95% glycerol containing 2% N-propyl gallate. Surface immunofluorescence/cell was quantitated by estimating the intensity of 10 different fields, divided by the number of cells. For intracellular staining, cells were washed as above and fixed in 3% paraformaldehyde for 30 min. All subsequent processing was done at room temperature. After rinsing with PBS as above, free aldehyde groups were quenched with 50 mM NH 4 Cl, and the cells were permeabilized with 0.1% Triton X-100. Nonspecific sites were blocked with PBS containing 10% goat serum, followed by 0.2% gelatin in PBS. Cells were then incubated with anti-PrP antibody (3F4) followed by FITCconjugated secondary antibody for 35-40 min each. For subcellular immunolocalization, subsequent incubations were done with anti-calnexin, anti-␣-mannosidase II, or anti-cathepsin D antibodies, respectively, followed by RITC-conjugated secondary antibodies. The cells were rinsed in PBS, mounted, and observed using a Laser Scanning Confocal Microscope (Bio-Rad).
Metabolic Labeling and Immunoprecipitation-In a typical pulsechase experiment, ϳ9.6 ϫ 10 6 cells plated overnight in 10-cm tissue culture dishes were used for each time point. Cells were washed with methionine-cysteine-free Dulbecco's modified Eagle's medium and preincubated in the same medium for 1 h at 37 or 24°C. Cellular proteins were metabolically labeled with 0.166 mCi/ml Tran 35 S-label (ICN) in labeling medium (methionine-cysteine-free Dulbecco's modified Eagle's medium with 5% dialyzed serum) for 2 or 30 min at 37 or 24°C, as indicated. The cells were washed and chased in serum-free Opti-MEM with 1 mM cold methionine and cysteine for indicated times. Where indicated, PI-PLC was added 30 min before the end of chase, and the cells were reincubated at 37°C. At the end of the chase, the medium was collected, and the cells were washed with cold PBS followed by lysis in 1% Nonidet P-40, 0.5% deoxycholate, PBS, pH 7.4, containing 10 g/ml each of leupeptin, antipain, pepstatin, and 1 mM PMSF. Clarified cell lysate and medium samples were rocked overnight with anti-PrP antibodies in the presence of 1% bovine serum albumin and 0.1% N-laurylsarcosine. Protein-antibody complexes were collected with 40 l of protein A-Sepharose (Pharmacia) and washed four times with 0.5 ml of wash buffer (150 mM NaCl, 10 mM Tris-HCl, pH 7.8, 0.1% Nlaurylsarcosine, and 0.1 mM PMSF). Bound protein was eluted by boiling in sample buffer (2% SDS, 10% glycerol, 5% ␤-mercaptoethanol) and analyzed by SDS-PAGE fluorography. PrP bands were quantitated by PhosphorImager analysis (Molecular Dynamics).
Assay of Detergent Insolubility and Proteinase K Resistance-Cells were lysed at 4°C in 1 ml of Western blot lysis buffer (TBS, Tris 20 mM, pH 7.4, NaCl 150 mM, EDTA 10 mM, Nonidet P-40 0.5% and deoxycholate 0.5%) containing protease inhibitors and clarified by centrifugation. The supernatant was ultracentrifuged at 100,000 ϫ g in a Beckman SW50 rotor for 1 h at 4°C. The high speed supernatant was collected, and the pellet fraction was redissolved in 1 ml of lysis buffer. Proteins in the supernatant and pellet fraction were precipitated with cold methanol at Ϫ20°C for 2 h, dissolved in sample buffer, and immunoblotted as above. For proteinase K treatment, cells were lysed as above in the absence of protease inhibitors, and clarified lysates were treated with 3.3 g/ml proteinase K for 5 or 10 min. The reaction was stopped by the addition of 2 mM PMSF, and the proteins were methanolprecipitated and immunoblotted.
Cell Surface Biotinylation-Subconfluent cell cultures were washed with PBS (containing 0.1 mM CaCl 2 , 1 mM MgCl 2 ) and biotinylated at 4°C with 1 mg/ml cold sulfo-NHS-biotin (dissolved in PBS) for 10 -15 min, as described (19). Excess biotin was quenched with 50 mM glycine in PBS. The cells were washed with PBS, lysed as above, and clarified by centrifugation for 10 min. PrP in the supernatant was immunoprecipitated as described above, fractionated by SDS-PAGE, and transferred to Immobilon-P. Biotinylated PrP molecules were revealed by blotting with streptavidin/horseradish peroxidase enhanced chemiluminescence (Amersham Corp.). Alternatively, biotinylated cellular proteins were retrieved by incubating the cell lysate at 4°C with streptavidin-agarose for 16 -18 h. Non-biotinylated cellular proteins in the supernatant were methanol-precipitated and dissolved in sample buffer. Biotinylated proteins were recovered after washing the beads several times with increasing stringency and eluting bound proteins by boiling in sample buffer (20). Intracellular and biotinylated samples were fractionated by SDS-PAGE, and the amount of PrP in each was revealed by immunoblotting with 3F4 or anti-C antibody.
Peptide Mapping-N and C termini of the 32-kDa form were evaluated by peptide mapping using endoproteases Asp-N and Lys-C. Cells were radiolabeled in the presence of tunicamycin to achieve a clear separation between the 27-kDa normal and 30-kDa aberrant form on SDS-PAGE. After immunoprecipitation and SDS-PAGE, the bands were visualized by autoradiography, excised, and soaked in elution buffer (Tris 20 mM, pH 8.0, SDS 1%, dithiothreitol 2 mM) for 1 h at 37°C. The gel pieces were homogenized and incubated at 60°C for 2 h, followed by treatment with 5 mM iodoacetic acid for 1 h at 4°C. The gel debris was removed by centrifugation of the homogenate in a Micropure filter (0.22 mm) inserted into a Microcon 10 unit (Amicon) at 10,000 ϫ g for 20 min. The protein sample was digested overnight at 37°C with Asp-N or Lys-C (Boehringer Mannheim) in 0.1% SDS, 50 mM sodium phosphate, pH 8.0, as described (17). The resulting peptides were separated on Tris-Tricine SDS-PAGE (21) and visualized by fluorography.

Characterization of Total Cell-associated and Plasma Membrane-anchored PrP C and PrP M at Steady State-Immunoblots
of the total cell-associated PrP C stained with the 3F4 antibody to PrP residues 109 -112 show three distinct bands corresponding to the three PrP glycoforms as follows: a diffuse band of 33-42 kDa with two highly modified glycans (H), an intermediate band of 29 -30 kDa with one complex glycan (I), and a discrete band of 27 kDa representing the unglycosylated form (U) (Fig. 1A, lane 1) (6). As expected, the H and I forms comigrate with the unglycosylated form at 27 kDa upon deglycosylation with N-glycosidase-F (PNGase-F) (Fig. 1A, lane 2). Deglycosylation also reveals an additional minor band that migrates at 20 kDa and, probably, represents the N-terminal truncated PrP fragment designated as "C2" in a previous study (Fig. 1A, lanes 2 and 4) (18). This band is more prominent in radiolabeling experiments (see below).
Immunoblots reacted with the anti-C terminus (anti-C) antibody to PrP residues 220 -231 reveal two additional minor bands migrating at ϳ22 and 18 kDa (Fig. 1B, lane 1) which are N-terminal truncated forms of the I and U isoforms, and lack the 3F4 epitope (I T , U T ; Ref. 18). These two bands are also more prominent following radiolabeling (see below). The N-terminal truncated H form is not detectable probably because it is obscured by the full-length forms. However, it is likely to comigrate at 18 kDa with the truncated I and U forms following PNGase-F treatment since the 18-kDa band markedly increases following deglycosylation (Fig. 1B, lane 2).
Immunoblots to detect PrP M with the 3F4 antibody show a prominent band of 32 kDa in addition to the three glycoforms present in PrP C (Fig. 1A, lane 3) (see below for detailed description). The pattern after deglycosylation is similar to that of PrP C , except that the deglycosylated product of 32-kDa form migrates at 30 kDa (Fig. 1A, lane 4).
Immunoblots with the anti-C antibody demonstrate that PrP M differs from PrP C in three ways as follows: 1) the I truncated form is replaced by ill defined bands of 25-22 kDa (Fig. 1B, lane 3); 2) deglycosylation reveals additional truncated forms at approximately 17 and 14 kDa (Fig. 1B, lane 4); and 3) the 32-kDa PrP M form is not detected by the anti-C antibody.
Quantitative analyses reveal that the total amounts of cellassociated PrP M and PrP C are not significantly different (total PrP M 92 Ϯ 7% of total PrP C ; n ϭ 8), but the percentage distribution of the various glycoforms differs and shows greater than 50% reduction of the PrP M U form. The H, I, and U glycoforms account for 47 Ϯ 9, 33 Ϯ 9, and 7 Ϯ 3%, respectively, of the total PrP M and are significantly different (p Ͻ 0.006) from the corresponding values of PrP C that are 64 Ϯ 5; 20 Ϯ 3, and 15 Ϯ 2%, respectively. The 32-kDa form accounts for 11 Ϯ 3% of the total PrP M .
Cell surface PrP C was released with phosphatidylinositolspecific phospholipase C (PI-PLC), an enzyme that specifically cleaves GPI anchors. Immunoblots of the released protein reacted with the 3F4 or anti-C antibodies reveal the same forms as in total cell extracts, i.e. all three full-length and truncated PrP C forms (Fig. 1C, lanes 1 and 3). The amount of cell surface PrP M released by PI-PLC is 45 Ϯ 21% (p Ͻ 0.03; n ϭ 4) less than that of PrP C , and the I and U truncated forms and the 32-kDa form are not detected (Fig. 1C, lanes 2 and 4).
The underrepresentation of PrP M at the cell surface after PI-PLC cleavage may be due to its resistance to PI-PLC because of aggregation or aberrant association with the plasma membrane (22,23). Therefore, we further examined cell surface expression of PrP C and PrP M by biotinylation and immunofluorescence.
PrP C and PrP M present at the cell surface were biotinylated and either 1) immunoprecipitated, fractionated by SDS-PAGE, and detected by blotting with streptavidin/horseradish peroxidase or 2) retrieved with streptavidin-agarose following cell lysis, fractionated, and detected by immunoblotting. The results obtained with either of these procedures are similar to those obtained with PI-PLC cleavage and show a reduction of the total surface PrP M of 48 Ϯ 23% (p Ͻ 0.006; n ϭ 4) (data not shown).
The PrP C and PrP M present on the plasma membrane were further investigated by quantitative immunofluorescence using 3F4 (Fig. 1D). Compared with PrP C , the fluorescence intensity of PrP M preparations is reduced on average by 62 Ϯ 3% (p Ͻ 0.0004; n ϭ 3), further confirming the underrepresentation of PrP M (Fig. 1D, panel 1 versus 2). Treatment with PI-PLC results in a 57 Ϯ 8% decrease of the immunofluorescence associated with PrP C and 77 Ϯ 5% of that associated with PrP M (Fig. 1D, panel 3 versus 4).
Together, these findings show that although comparable amounts of total PrP M and PrP C are present in the intracellular compartments, PrP M is underrepresented at the plasma membrane, consistent with its impaired transport to that cell locale. In addition, PrP M is processed aberrantly, producing at least two additional fragments of 14 and 17 kDa.
Therefore, metabolic studies were carried out to investigate the synthesis, processing, turnover, and transport of PrP M employing a pulse-chase paradigm.
Metabolism and Transport of PrP C and PrP M -Immunoprecipitation of newly synthesized PrP with 3F4 after a 2-min labeling with Tran 35 S-label reveals three bands that are common to both PrP C and PrP M . The two upper bands represent the precursors of the H and I forms (H P , I P ; Fig. 2A, lanes 1 and  5, upper panel) and exhibit gel mobilities different from those of the mature forms because they contain immature glycans. The lowest band comigrates with the U form ( Fig. 2A, lanes 1 and 5,  upper panel) (6,15). PrP M preparations show two additional forms migrating at 32 and 24 kDa (Fig. 2A, lane 5, upper and  lower panels, respectively). After a 40-min chase period, modification of the high mannose core glycans yields the mature H and I glycoforms that were observed in the immunoblots ( Fig.  2A, lanes 4 and 8, upper panel). An additional 20-kDa form is detected in both PrP C and PrP M preparations. Thus, the synthesis and early kinetics of the H, I, and U forms are comparable in both PrP C and PrP M cells. However, when chased for longer times, the turnovers of PrP C and PrP M are markedly different (Fig. 2B). Following a 30-min labeling and chase periods from 1 to 6 h, "intracellular" (i.e. PI-PLC-treated) PrP C and PrP M progressively decrease in quantity, although PrP M decreases more rapidly, becoming less than half the amount of PrP C after a 6-h chase (Fig. 2B, lanes 1-5 versus 6 -10). The 32-kDa form is relatively stable for 2 h and decreases thereafter (Figs. 2B, lanes 6 -10). Throughout the chase, this form accounts for 21-45% of the cell-associated PrP M , whereas the 24-kDa peptide, which is less represented, becomes undetectable after a 2-h chase (Fig. 2B, lanes 6 -10). Immunoprecipitation of the intracellular PrP M with the anti-C antibody indicates that the truncated forms are more prominent in PrP M than in the PrP C preparations after 2 or 4 h of chase (Fig. 2C,  lanes 3-5 versus 8 -10).
The PrP released from the plasma membrane by PI-PLC, and immunoprecipitated with the 3F4 antibody, peaks at the 2-h chase time point in both PrP C and PrP M cells (Fig. 2D, lanes  3 and 8). The amount of PrP M released is approximately 60% less than that of PrP C (17 Ϯ 5% versus 42 Ϯ 7% of the total cell-associated PrP M and PrP C , respectively; p Ͻ 0.002; n ϭ 3) (Fig. 2D, lanes 7-10 versus 2-5), and only the H form of PrP is detected at the surface of the mutant cells (Fig. 2D, lanes 7-10), whereas all three full-length glycoforms are well represented in the control cells (Fig. 2D, lanes 2-5). Immunoprecipitation of the PI-PLC-released PrP C with the anti-C antibody reveals a delay in the appearance of the I-and U-truncated forms with respect to the full-length forms (Fig. 2E, 3-5) consistent with the notion that these peptides result from the cleavage of full-length PrP (9,24). In contrast, in PrP M cells, the ill-defined 25-22 kDa (probably corresponding to PrP C I-truncated form) and the U-truncated forms are not transported to the surface in significant quantities (Fig. 2C, lanes 8 -10 versus Fig. 2E lanes  8 -10).
The apparent discrepancy between the immunoblot data showing comparable PrP C and PrP M pools and pulse-chase data showing that intracellular PrP M is unstable and transported only in small amounts to the cell surface, even after overnight chase times (data not shown), suggests that PrP M aggregates and is retained in an intracellular compartment. If present as an aggregate, the mutant protein may not be efficiently immunoprecipitated by our procedure. This would account for the observed decrease of PrP M in the pulse-chase experiments described above. Therefore, we studied the site of intracellular retention and determined whether the bulk of intracellular PrP M is aggregated.
Solubility and Protease Sensitivity of Intracellular and Surface PrP M -Solubility of total cell-associated and surface PrP M and PrP C was examined following treatment with non-ionic detergents and ultracentrifugation to separate aggregated from soluble proteins. Supernatant and pellet fractions were recovered and analyzed by blotting (Fig. 3, A and B). Although PrP C is almost entirely detergent-soluble, 50 Ϯ 3% (n ϭ 6) of total cell-associated PrP M , including a portion of the 32-kDa form, is detergent-insoluble (Fig. 3A, lane 3 versus 6). The insolubility is inversely related to the degree of glycosylation, the unglycosylated form being entirely insoluble. In contrast, both PI-PLC cleaved (not shown) and surface biotinylated PrP M are detergent-soluble (Fig. 3B, lane 3 versus 6). With the exception of the 32-kDa form, all detergent-insoluble PrP M is Endo-H-resistant (Fig. 3C), indicating that aggregation of the non-32-kDa fraction takes place in a compartment distal to the cis-Golgi (25).
To evaluate if aggregated PrP M could be immunoprecipitated efficiently, total cell extract, soluble, and insoluble fractions of PrP C and PrP M were subjected to immunoprecipitation, and the resulting supernatant was immunoblotted for any remaining PrP (Fig. 3D). All PrP C and almost all of the PrP M -soluble (S 2 ) fraction are immunoprecipitated by our procedure (Fig. 3D,  lanes 1-3 and 5), whereas a substantial amount of insoluble PrP M remains in the supernatant after the immunoprecipitation (Fig. 3D, lane 6). Thus, the decrease in cell-associated PrP M observed in pulse-chase experiments is primarily due to aggregation. Since the aggregated PrP M was not solubilized even by boiling in SDS (22), the kinetics of aggregation could not be determined.
The presence of proteinase K-resistant PrP M was assessed by treating PrP C and PrP M cell lysates with various concentrations of PK (0.5-10 g/ml) for different times (0 -30 min) at 37°C. Treatment with 3.3 g/ml PK for 5 min revealed that approximately 7% of the total PrP M is resistant, whereas PrP C is completely digested (Fig. 3E, lanes 2 and 3 versus 5 and 6).  8) (lower part of the gel is exposed four times longer to show the minor species of 20 and 24 kDa). B, longer pulse (30 min) and chase times (1-6 h) show that the PrP M still associated with the cell following PI-PLC treatment (intracellular PrP) turns over faster than the matching PrP C (lanes 6 -10 versus 1-5) (immunoprecipitation with 3F4). C, following immunoprecipitation with anti-C antibody, the intracellular PrP C I T and U T truncated forms are barely detectable at this exposure after 2 and 4 h of chase (lanes 3 and 4), whereas the PrP M 25-22KkDa and U T forms are present in significant amounts (lanes 8 and 9). The 20-kDa form (20) is detected with both 3F4 and anti-C antibodies and decreases progressively with increasing chase time (lanes 1-10). D, 30 min before the end of each chase time point, cells were treated with PI-PLC, and the released PrP was immunoprecipitated with 3F4. Only the H form is detected in significant amounts in PrP M preparations, in contrast to all three glycoforms (H, I, and U) of PrP C (lanes 7-10 versus 2-5). E, following immunoprecipitation of PI-PLC-released PrP with the anti-C antibody, there is a significant decrease of full-length PrP M forms as compared with PrP C (lanes 6 -10 versus 1-5), and the I T and U T forms are not detected (lanes 7-10).
All full-length PrP M forms, and possibly the 32-kDa forms, are represented in the PK-resistant fraction. All the forms show a decrease in size after PK digestion as in prion diseases (Fig. 3E,  lane 4

versus 5).
Intracellular Localization of PrP-Employing double immunofluorescence with antibodies directed to PrP and to specific cellular organelles, we observed that the majority of intracellular PrP C is co-localized with anti-␣-mannosidase II, a marker of the medial Golgi (Fig. 4A, panels 1-3). The distribution of intracellular PrP M is more widespread; part co-localizes with the Golgi apparatus, just like PrP C (Fig. 4A, panels 4 -6), and part maps with calnexin, an ER marker (Fig. 4B, panel 1 versus  2), and with cathepsin D, a marker of the endosomal-lysosomal compartment (Fig. 4B, panel 3 versus 4).
Further Characterization of the 32-kDa and the Other Aberrant PrP M Forms-The detection of the 32-kDa form, even after a 2-min pulse ( Fig. 2A), and its Endo-H sensitivity, even after a sustained chase (Fig. 3C, lanes 1 and 2; and data not shown), suggest that it remains in the ER-cis Golgi region (25). Moreover, the 32-kDa form migrates at 30 kDa on SDS gels following deglycosylation (ϳ3 kDa larger than unglycosylated PrP) and immunoreacts with the N-terminal antibody to PrP residues 23-40 (Fig. 1A, lanes 3 and 4 and data not shown) suggesting that it maintains the 22-residue C-terminal GPI signal peptide and does not acquire the GPI anchor (26). To explore this possibility, Q217R and wild type-transfected cells were labeled either with [ 35 S]methionine and -cysteine (tran 35 S-label) (Fig. 5, lanes 1 and 2) or [ 3 H]ethanolamine (Fig. 5, lanes 3  and 4), which labels the GPI anchor. In contrast to the H, I, and U forms of PrP C and PrP M , the 32-kDa form is not detected after [ 3 H]ethanolamine labeling, confirming the absence of the GPI anchor (Fig. 5, lane 2 versus 4). The persistence of the C-terminal GPI signal peptide is supported by preliminary data obtained from peptide mapping (data not shown) and may account for the lack of immunoreactivity of the 32-kDa form with the anti-C antibody.
The 24-kDa as well as the 17-and 14-kDa forms seen after deglycosylation are the other aberrant forms that are only detected in the PrP M cells ( Figs. 2A and 1B, respectively). Like the 32-kDa form, the 24-kDa form is sensitive to Endo-H (data not shown) and probably resides in the ER-cis Golgi (25). It reacts with 3F4, but not with antibodies to the N (not shown) or C termini (Fig. 2C). The 17-and the 14-kDa forms are recognized only by the anti-C antibody and are detected only after deglycosylation (Fig. 1B, lane 4).
Reversibility of Changes at 24°C-Evidence that the aggregation of the mutant protein is associated with an aberrant conformation was obtained when cells were analyzed at 24°C, a temperature known to promote folding of proteins (27,28). Pulse-chase experiments carried out at this temperature show that synthesis and turnover of intracellular PrP M is similar to that of PrP C for up to 6 h of chase (Figs. 6A, lanes 1-5 versus  6 -10) and different from that at 37°C (Fig. 2, B and D). However, the aberrant 32-and 24-kDa forms remain Endo-Hsensitive (data not shown). Moreover, the PrP M cleaved from the plasma membrane with PI-PLC also contains the I and U forms that are undetectable at 37°C (Fig. 6B, lanes 8 -10  versus Fig. 2D, lanes 8 -10). Quantitative analysis of total PrP FIG. 3. Detergent insolubility and protease resistance of PrP M . A, total cell-associated PrP C and PrP M was solubilized in detergent, and after pelleting the nuclei and cell debris (S 1 ), fractionated by high speed centrifugation to yield soluble (S 2 ), and insoluble (P 2 ) fractions. These were analyzed by SDS-PAGE and immunoblotting with 3F4. Although PrP C is completely soluble (lanes 2 and 3), PrP M is largely recovered in the detergent-insoluble fraction (lanes 5 and 6). B, biotinylated cell surface PrP M is fully soluble (lanes 5 and 6) as observed for biotinylated PrP C (lanes 2 and 3) (biotinylated surface proteins were retrieved with streptavidin-agarose and immunoblotted with 3F4. S 1 , low speed detergent-soluble lysate; S 2 and P 2 , high speed detergent-soluble (S 2 ) and -insoluble (P 2 ) fractions. C, Endo-H treatment of the insoluble fraction (P 2 ) shows that all full-length PrP M forms are Endo-H-resistant except the 32 kDa that migrates at 30 kDa after Endo-H treatment (lane 1 versus 2). D, soluble (S 1 and S 2 ) and insoluble (P 2 ) fractions of PrP C and PrP M were immunoprecipitated with 3F4, and the resultant supernatant was immunoblotted with the same antibody. In contrast to PrP C (lanes 1-3), a significant amount of PrP M (lanes 4 and 6), most of which is insoluble, fails to immunoprecipitate and is recovered in the supernatant; the soluble PrP M (S 2 ) is immunoprecipitated efficiently, except for the 32-kDa form (lane 5). E, a small amount of PrP M is resistant to PK treatment (3.3 g/ml) for 5 min (lane 5); in contrast, PrP C is completely digested (lane 2). The PK-resistant PrP M shows the expected faster mobility of the resistant C-terminal fragments and contains all the major (H res , I res , U res ) and, possibly, the 32-kDa (32 res ) forms (lanes 5 and 6).
shows that the turnover of PrP C and PrP M is comparable at 24°C, whereas at 37°C PrP M turns over more rapidly than PrP C (Fig. 6, C and D).
Immunoblots of cells exposed to 24°C show that over 60% less PrP M sediments in the detergent-insoluble form than at 37°C (Fig. 6E, lane 6; versus Fig. 3A, lane 6); the H-and I-glycosylated forms become more readily soluble than the unglycosylated U form, whereas the 32-kDa form becomes entirely soluble (Fig. 6E, lanes 5 and 6). These findings reinforce the role of glycans in protein folding and also indicate that lower temperature promotes correct folding even in the absence of glycans.
Immunofluorescence microscopy further reveals that following a 24-h exposure to 24°C, PrP M co-localizes predominantly with the Golgi apparatus as does PrP C (Fig. 6F; compare with Fig. 4A, panels 1-3). DISCUSSION The complex changes in the biogenesis of PrP M induced by the Q217R-129V genotype show that a PRNP point mutation affects the metabolism of human PrP M in significant ways as was previously shown for the D178N mutation (15). This information is relevant to the pathogenesis of human GSS. In addition, the present findings have implications regarding GPI anchor addition, folding, and storage of abnormal proteins in intracellular compartments, and the "quality control" system in the secretory pathway.
Effects of the Q217R-129V Genotype on the Biogenesis of PrP M -Although synthesis and maturation of full-length PrP M are similar to those of PrP C , aberrant PrP M forms are detected soon thereafter, and the the full-length PrP M forms undergo surprising posttranslational changes.
The aberrant PrP M forms include the 32-and 24-kDa forms and minor fragments of 17 and 14 kDa. The 32-kDa form is likely to lack the GPI anchor and to retain the 22-residue C-terminal anchoring code (8,29). Incomplete anchor addition to PrP M cannot be merely due to PrP M overexpression since the CEP4␤ expression system yields comparable synthesis of PrP C which is fully anchored ( Fig. 2A) (15). In addition, the 32-kDa form appears to be retained in the ER since it remains Endo-H-sensitive. The origin and characteristics of the other aberrant PrP M forms is unclear. On the basis of the immunoreactivity and Endo-H sensitivity, the 24-kDa peptide is either an internal or an N-terminally truncated fragment which, like the 32-kDa form, resides in the ER. The 17-and 14-kDa peptides must be generated by the aberrant cleavage of a glycosylated form.
The Q217R mutation also destabilizes the full-length PrP M forms, the majority of which aggregate, whereas a small amount may undergo degradation. Since PrP M is not transported efficiently to the plasma membrane and the tendency to aggregate is inversely related to the extent of glycosylation, the subcellular distribution of PrP M and its glycoforms are different from that of PrP C ; the relative amount of PrP M associated with the cell membrane at steady state is 45% less than the amount of PrP C and is represented largely by the highly glycosylated form, although the total intracellular pools of PrP are comparable, and all the glycoforms are represented in the PrP M pool (Fig. 1C versus A and B). The underrepresentation of the full-length I and U PrP M forms at the cell surface, in turn, may change the way truncated forms are generated.
Aggregated PrP M accumulates in intracellular compartments. Although the 32-kDa form appears to aggregate in the ER-cis Golgi, other PrP M forms aggregate in more distal intracellular compartments, judging by their resistance to Endo-H and their co-localization with markers of the endosomal-lysosomal system.
An important observation of this study is that the amount of aggregated PrP M is significantly reduced following the culture of mutant cells at 24°C. This finding suggests that aggregation of PrP M is the result of altered folding in the ER (27). A similar temperature effect has been observed on a mutant cystic fibrosis transmembrane conductance regulator with deletion of phenylalanine (CFTR⌬F508) (28). This finding has therapeutic implications since substances like glycerol have been shown to mimic the effect of low temperature on CFTR⌬F508 (30) and hence may facilitate proper folding of other mutated proteins such as PrP.
The finding that PrP M has increased resistance to protease K (PK) digestion strengthens the validity of the present cell sys- tem as a model of the human prion disease. The PK-resistant fragments generated by the exogenous protease in the Q217R mutant cells, as in the PrP Res present in prion diseases, lacks the N terminus (data not shown). Our data show that the PK-resistant PrP M contains the same forms in a similar ratio to the PrP M forms present in the aggregated fraction (Fig. 3E versus Fig. 3, A and D). This is consistent with a precursorproduct relationship between aggregated (but PK-sensitive) and PK-resistant PrP M forms as indicated (23). Thus, the present data support the notion that in inherited prion diseases, as in other prion diseases, the formation of the PK-resistant PrP M is preceded by the formation of the aggregated but proteasesensitive PrP M .
Collectively these findings indicate that the primary effect of the Q217R-129V genotype is a perturbation of the PrP M conformation. This, in turn, leads to the expression of a variety of PrP M forms, only some of which can be transported while the others aggregate and result in an altered subcellular distribution of PrP M .
The Q217R Cell Model and the Q217R Human Prion Disease-The presence of a 7-kDa internal peptide spanning from residues 81/82 to residues 145/146 has been demonstrated in FIG. 6. Reversibility of PrP M abnormalities at 24°C. A, pulse (30 min) and chase (1-6 h) at reduced temperature followed by immunoprecipitation with 3F4 shows that maturation and turnover of intracellular PrP C and PrP M are similar (lanes 1-5 versus 6 -10). B, immunoprecipitation of PI-PLC-cleaved PrP shows that PrP M is present in increased amount compared with 37°C and also contains the I and, possibly, the U forms (lanes 8 -10 versus lanes 7-10 of Fig. 2D) The surface PrP M retrieved at 6 h is less than the PrP C , consistent with a residual decrease or delay in the transport of PrP M to the plasma membrane (lanes 1-5 versus 6 -10). C and D, quantitative analysis of total PrP (intracellular ϩ PI-PLC released) by phosphorimaging confirms that the kinetics of turnover of PrP M at 24°C is similar to that of PrP C (C), whereas at 37°C PrP M turns over much more rapidly than PrP C (D).  6 with lane 6, Fig.  3A). F, immunofluorescence microscopy with the 3F4 antibody carried out after incubation at 37 or 24°C for 24 h shows that PrP M appears to co-distribute with the Golgi apparatus at 24°C, a localization similar to that of PrP C (see Fig. 4A, panels 1-3), although it is more widespread at 37°C. amyloid deposits from brains of affected subjects carrying the Q217R-129V haplotype (13). In addition, brain extracts contained the common PK-resistant PrP of 27-30 kDa, a fragment of 8 -13 kDa, and unidentified lower molecular weight fragments, the latter probably being the result of sequential proteolytic cleavage (13).
The present data suggest that during the pathogenesis of the disease, the above changes are preceded by the formation of detergent-insoluble and, in part, PK-resistant aggregates as well as by the expression of aberrant PrP M forms. The aberrant truncated forms of PrP M are indicative of an abnormal cleavage generating N-terminal fragments that contain the 106 -126 sequence that has been shown to be neurotoxic (31). However, the Q217R cell model and the human disease differ in two major ways as follows: 1) the degree of resistance to PK of the protease-resistant PrP M in the mutant cells is at most modest, and 2) the transfected cells apparently lack the above-mentioned 7-kDa internal PrP M fragment found in the brain. These differences may reflect the long incubation time and the participation of other cells in the processing of the extracellular PrP M in the inherited prion diseases (32).
The Q217R Cell Model and Other Cell Models of Prion and Nonprion Diseases-The only clear similarity between the Q217R cell model and that of FFI and CJD 178 , two inherited prion diseases genetically linked to the D178N-129M and D178N-129V haplotypes, respectively (15), is the instability of PrP M . In contrast, the FFI and CJD 178 cell models generate less protease-resistant PrP M and lack aberrant forms such as the 32-kDa form. This biochemical diversity is consistent with the phenotypic heterogeneity of inherited prion diseases (1). For example, FFI and CJD 178 are both characterized by the presence of relatively small amounts of PK-resistant PrP M and lack amyloid deposits in the brain tissue.
The present data are more difficult to compare with previous studies using mouse homologues of a variety of PRNP mutations expressed in Chinese hamster ovary cells (22,23,33). In these studies, which do not focus on the Q217R mutation, the various PrP M glycoforms, including the unglycosylated form, are well represented at the cell surface, are detergent-insoluble and protease-resistant at the plasma membrane, and when at the surface, remain attached to the plasma membrane despite cleavage of the GPI anchor by PI-PLC, suggesting that PrP M itself is tightly associated with the cell membrane. These differences may be due to the use of a heterologous system rather than the homologous system we have used or to intrinsic differences in PrP metabolism between Chinese hamster ovary and human neuroblastoma cells.
The Q217R Mutation: Implications for GPI Anchor Addition and the Quality Control System of the Secretory Pathway-Since deletion of amino acid residues in the sequence flanking the N-terminal side of the acceptor residue has not been reported to impair anchoring (34), it is remarkable that a point mutation at residue 217 interferes with the anchor linkage which normally occurs 14 amino acids downstream, at residue 231. It is unlikely that this is simply due to misfolding of the nascent PrP M to make it unfit to receive the anchor since incubation at 24°C, a temperature known to promote protein folding, does not decrease expression of the 32-kDa form or facilitate its transport past the cis Golgi compartment, as demonstrated by its persistent sensitivity to Endo-H (25); yet, the decreased temperature prevents aggregation of the 32-kDa form. Although the precise cellular locale in which the 32-kDa form accumulates is unclear, retention in the intermediate compartment has been reported for proteins that fail to receive a GPI anchor and maintain the signal peptide (35).
The ER performs a quality control function (36, 37) by which improperly folded or oligomerized glycoproteins are prevented from exiting the ER. This process is accomplished by a chaperone-mediated interplay between folding and glycosylation processes, and the misfolded proteins are eventually degraded (38 -42). Some mutant glycoproteins, however, are stopped in post-ER compartments or are allowed to reach their destination at the plasma membrane (43). The aberrant PrP M forms caused by the Q217R mutation appear to be variously affected by the quality control system. Thus, while the anchorless 32-kDa form is retained in the ER-cis Golgi compartment, anchored forms that are either unglycosylated or carry one oligosaccharide chain appear to escape the ER quality control system and aggregate in post-ER compartments. In contrast, most of the fully glycosylated H form is transported to the plasma membrane. However, even this form may have an abnormal conformation since it is likely to generate the abnormal 17-and 14-kDa peptides by an aberrant proteolytic cleavage (44 -46). Transport of an abnormal glycosylated PrP M conformer to the cell surface is consistent with the notion that N-glycosylation does not ensure proper folding but rather stabilizes conformation and prevents aggregation (42).
The Q217R neuroblastoma cells thus provide a striking model of the complex changes that a PRNP point mutation can introduce into the processing of PrP M , presumably engendering various degrees of protein misfolding. In the inherited prion disease, long incubation, aging, and cell diversity of the human brain are likely to add to this complexity.