Effect of the D178N mutation and the codon 129 polymorphism on the metabolism of the prion protein.

Prion diseases are thought to be caused by the conversion of the normal, or cellular, prion protein (PrP) into an abnormal protease-resistant conformation (PrP). There are three familial forms of human prion disease, Creutzfeldt-Jakob disease (CJD), Gerstmann-Straussler-Scheinker syndrome, and fatal familial insomnia (FFI) which are all expressed at advanced age despite the congenital presence of the mutant prion protein (PrP). The cellular mechanisms that result in the age-dependent conversion of PrP into PrP and the unique phenotypes associated with each PrP are unknown. FFI and a familial type of Creutzfeldt-Jakob disease (CJD), share the D178N mutation in the PrP gene but have distinct phenotypes linked to codon 129, the site of a methionine/valine polymorphism (129M/V). We analyzed PrP processing in cells transfected with constructs reproducing the FFI and CJD genotypes. The D178N mutation results in instability of the mutant PrP which is partially corrected by N-glycosylation. Hence, only the glycosylated forms of PrP reach the cell surface whereas the unglycosylated form is degraded. The unglycosylated PrP is also under-represented in the brain of FFI patients validating the cell model. These results offer new insight into the effect of the D178N mutation on the metabolism of the prion protein.

Mutations underlying an increasing number of inherited diseases are being discovered. The task now is to define the individual steps through which the mutated protein causes a specific disease, in some cases by becoming a pathogen, often after a symptom-free interval of several decades (Hamilton et al., 1992).
The prion protein (designated PrP C ) has been implicated in a variety of human and animal diseases referred to as prion diseases, spongiform encephalopathies, or transmissible amyloidoses (Prusiner and DeArmond, 1994). Prion diseases can occur either sporadically or by infectious transmission in both humans and animals. In addition, 18 mutations in the PrP C gene, identified as PRNP, have been reported to be associated with inherited forms of prion diseases in humans .
The inherited prion diseases associated with PRNP mutations fall into three major groups: Creutzfeldt-Jakob disease (CJD), 1 Gerstmann-Strä ussler-Scheinker syndrome, and the recently discovered fatal familial insomnia (FFI) . Despite their phenotypic differences, FFI and one familial type of CJD (CJD 178 ) are both linked to a single mutation of PRNP at codon 178 resulting in the substitution of asparagine for aspartic acid (D178N) (Goldfarb et al., 1992). They differ, however, at the PRNP codon 129, a natural polymorphic site encoding either methionine or valine (Goldfarb, Petersen et al., 1992). Codon 129 of the mutant allele specifies the disease phenotype associated with the D178N mutation: the 129M,D178N haplotype is linked to FFI, the 129V,D178N haplotype to CJD 178 (Goldfarb, Petersen et al., 1992).
A variety of data clearly demonstrate the importance of the mutant PrP (PrP M ) in the pathogenesis of FFI and CJD 178 and other inherited prion diseases. They provide no information, however, on the precise metabolic events that lead to the disease. In this study we used transfected human neuroblastoma cells to examine the effect of the PRNP D178N mutation associated with either the 129M or 129V codon on the synthesis and metabolism of the PrP M .

EXPERIMENTAL PROCEDURES
Cloning and Production of Cell Lines-M-17 human neuroblastoma cells (Ross et al., 1983; provided by Dr. June Biedler) were transfected with the episomal vector CEP4␤ (Hambor et al., 1988) containing prion protein coding sequence (27 g/10-cm plate) using the cationic lipid DOTAP (Boehringer Mannheim, 46 g/10-cm plate) in serum-free Opti-MEM (Life Technologies, Inc.). The inserted prion protein coding sequence, under the control of the cytomegalovirus promoter, was either normal (178D) or mutant (D178N) with either a Met or Val at codon 129 (129M or 129V), a natural polymorphic site in the PRNP (Fig. 1). The prion protein coding sequence was obtained by amplifying human genomic DNA with primers that flank the entire coding sequence: DG2, 5Ј-GTACTGAGAATTCGCAGTCATTAT-GGCGAACCTTGG-3Ј; and DG1, 5Ј-GTACTGAGGATCCTCCTCATC-CCACTATCAGGAAGA-3Ј, the underlined sequences correspond to the prion gene sequences, nucleotides 41 to 70 and 792 to 814, respectively, according to the sequence of Kretschmar (1986). Using the EcoRI and BamHI sites (bold face) in the primers, the coding sequence was cloned into the bacterial plasmid pVZ1 (Henikoff and Eghtedarzadeh, 1987). We created the codon 129V alleles corresponding to normal or D178N coding sequence using oligonucleotide-directed mutagenesis to convert the codon 129M alleles (Bio-Rad Muta-Gene phagemid in vitro mutagenesis kit). Transfected cells were grown as bulk selected hygromycin-resistant cultures in Opti-MEM with 5% calf serum supplemented, iron-enriched (Life Technologies, Inc.), and 500 g/ml hygromycin B (Calbiochem). For each experiment, cells were removed from the flask with trypsin, counted, and identical numbers of cells seeded onto plates and grown overnight to ϳ95% confluence. All experiments were repeated with cells from independent transfections and at different times post-transfection to avoid selection bias.
Source of Tissue-Samples of gray matter were obtained from the occipital cortex of an FFI subject with the D178N mutation and one * This work was supported by National Institutes of Health Grants AG-08992 and AG-08155 and the Britton Fund. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Dept. of Pathology, Institute of Pathology, 2085 Adelbert Rd., Case Western Reserve University,Cleveland,; E-mail: rbp@po.cwru.edu. octapeptide repeat deletion in the mutant allele and from an age matched control.
Cells were tested for the presence of proteinase K-resistant PrP in the following manner. Whole cell homogenates were prepared from 10 7 cells, after washing the plates with cold PBS, by lysing with 9 volumes of 100 mM NaCl, 10 mM Tris, 10 mM EDTA, 0.5% Nonidet P-40, and 0.5% sodium deoxycholate, pH 7.4. The postnuclear supernatant was digested with proteinase K in either of 2 ways: 2 ϫ 10 5 cell equivalents were digested in a 20-l reaction for 1 h with 0.5, 1, 5, or 10 g/ml proteinase K (Boehringer Mannheim) or 2 ϫ 10 5 cell equivalents were digested with 5 g/ml for 5, 10, 15, or 20 min. The proteinase K reaction was terminated by addition of PMSF to 3 mM. The samples were precipitated with 4 volumes of methanol, resuspended in sample buffer, resolved on 14% SDS-PAGE, blotted onto Immobilon P, and detected with the 3F4 antibody.
Brain tissue was homogenized in 9 volumes of 0.32 M sucrose, 20 mM Tris, pH 7.5, containing 2 g/ml PMSF, 1 g/ml leupeptin, 1 g/ml pepstatin. The supernatant was clarified by centrifugation at 1000 ϫ g for 10 min, and recentrifuged at 100,000 ϫ g for 1 h to obtain a membrane fraction. The total homogenate and the membrane fraction, before and after deglycosylation, were resolved on 12% SDS-polyacrylamide gels, transferred to Immobilon P, and probed with an anti-aminoterminal antibody (a polyclonal rabbit serum to a synthetic peptide, human PrP 23-40; Chen et al. (1995)) provided by B. Ghetti, Indiana University, Indianapolis, IN. Quantitative analyses of the immunoblots was performed with a computer-assisted laser scanner (LKB Ultrascan XL).
Biotinylation of Cell Surface PrP-Cells were plated at 5 ϫ 10 6 cells per 10-cm plate or 1.8 ϫ 10 6 cells per 35-mm well. After incubation overnight at 37°C the cells were held on ice for 5 min with or without pretreatment with PI-PLC at 37°C for 30 min. The cells were washed three times with PBS, pH 7.4, containing 0.1 mM CaCl 2 and 1 mM MgCl 2 . The cells were biotinylated by adding cell impermeable NHS-Biotin (Pierce) at 250 g/ml for 10 min at 4°C. The reaction was quenched by adding 50 mM glycine in PBS, pH 7.4, at 4°C followed by 2 washes with the same. Cells were lysed in PBS, pH 7.4, containing 0.5% Nonidet P-40 and 1 mM PMSF. The lysates were precleared with 20 l of Protein A-agarose (Boehringer Mannheim) for 1 h at room temperature, centrifuged at 2,100 ϫ g for 10 min, and then the supernatant was immunoprecipitated with the monoclonal antibody 3F4 or the polyclonal antibody anti-C described by . The samples were run on 14% SDS-PAGE, blotted onto Immobilon P, and detected with streptavidin-horseradish peroxidase (1:1000) followed by ECL (Amersham).
Pulse-Chase PI-PLC-Cells were grown to 95% confluence on 10-cm plates. Cells were washed once with methionine-deficient MEM (deficient MEM, ICN) and then incubated in deficient MEM for 30 min (37°C). A 30-min pulse with 0.5 mCi of Trans 35 S-label (ICN) in deficient MEM was followed by a 0, 30-min, or 2-h chase in Opti-MEM. Following the chase, cells were washed once with Opti-MEM and then incubated in Opti-MEM Ϯ PI-PLC (59 ng/ml) for 2 h at 4°C. Media was then collected and cells were lysed for immunoprecipitation.
Pulse-Chase PI-PLC in the Presence of Tunicamycin-Cells were grown to 95% confluence on 10-cm plates. Cells were washed once with deficient MEM and then incubated for 1 h in deficient MEM Ϯ 2 g/ml tunicamycin (Boehringer Mannheim). 2 g/ml was empirically determined to be the lowest concentration that would effectively block gly-cosylation while not affecting protein synthesis. 2 A 1-h pulse with 0.5 mCi of Trans 35 S-label in deficient MEM Ϯ tunicamycin (2 g/ml) was followed by a 0 or 4-h chase in Opti-MEM lacking tunicamycin. Following the chase, cells were washed once with Opti-MEM and then incubated in Opti-MEM Ϯ PI-PLC (59 ng/ml) for 2 h at 4°C. Media was then collected and cells were lysed for immunoprecipitation.
Pulse-Chase in the Presence of Brefeldin A (BFA)-Cells were grown to 90% confluence on 60-mm plates. Cells were washed once with deficient MEM and then incubated in deficient MEM for 30 min (37°C). A 30-min pulse with 0.175 mCi of Trans 35 S-label in deficient MEM with 1 g/ml BFA (Epicentre Tech) was followed by a 0, 30-min, 60-min, 90-min, 1-h, or 2-h chase in Opti-MEM with 1 g/ml BFA. 1 g/ml was empirically determined to be the lowest concentration that would inhibit transport beyond the Golgi. 3 Cells were washed in phosphatebuffered saline and lysed for immunoprecipitation.
Cell Lysis and Immunoprecipitation-Media was removed and centrifuged at 290 ϫ g for 10 min (4°C) to eliminate any cells and 1 ml was used for immunoprecipitation. Cells were washed once with phosphatebuffered saline and then lysed in 1 ml of 0.5% Nonidet P-40/phosphatebuffered saline supplemented with 1 mM PMSF. After incubating on ice for 10 min, lysates were centrifuged at 2100 ϫ g for 10 min (4°C) to remove nuclei and large debris. 250 l of the cell lysates were immunoprecipitated in a total volume of 1 ml. The prion protein was immunoprecipitated in 1% bovine serum albumin, 0.1% N-laurylsarcosine (with 1 M PMSF added) by rocking at 4°C for 16 h. Immunoprecipitation was carried out with one of the following antibodies: 3F4, anti-C, an antiserum to the carboxyl-terminal residues 220 -231 , or anti-N, an antiserum to the amino-terminal residues 23-40. Protein-antibody complexes were bound to Protein A-agarose beads (Boehringer Mannheim) by rocking at 4°C for 3 h. To wash, beads were pulse spun at 690 ϫ g (4°C), the supernatant was aspirated, and the beads were resuspended in 0.5 ml of ice-cold wash buffer (150 mM NaCl, 10 mM Tris, pH 7.8, 0.1% N-laurylsarcosine with 0.1 mM PMSF). Beads were washed this way six times and then resuspended in sample buffer (2% SDS, 10% glycerol, 5% ␤-mercaptoethanol) and boiled to release the bound proteins.
PNGase-F and Endoglycosidase H Treatment-Immunoprecipitated protein was removed from the beads and precipitated in 4 volumes of methanol to remove sample buffer. Protein pellets were resuspended in denaturing buffer (0.5% SDS, 1% ␤-mercaptoethanol) and boiled for 10 min. Samples were deglycosylated with PNGase-F (1000 units, New England Biolabs, in 1% Nonidet P-40, 50 M sodium citrate, pH 7.5) or endoglycosidase H (1000 units, New England Biolabs, in 50 M sodium citrate, pH 5.5) for 60 min at 37°C. Proteins were again precipitated with methanol and the resulting pellet was boiled in sample buffer and separated on 16% SDS acrylamide gels.
SDS-PAGE, Fluorography, and PhosphorImage Analysis-Proteins were separated on 16% acrylamide gels (37.5:1 acrylamide:bis-acrylamide). Gels were fixed in methanol:acetic acid:H 2 O (40:10:50) for 15 min, dehydrated in dimethyl sulfoxide for 1 h and radioactivity was enhanced by rocking the gels in PPO/dimethyl sulfoxide (22%) for 90 min. Gels were then dried and exposed to film (Kodak XAR-5). Densitometric analysis of the fluorograms was performed using a PDI 420oe scanner. Some gels were also analyzed using a Molecular Dynamics Phos-phorImager. Microsoft Excel 5.0 was used for statistical analysis.

Expression and Localization of the PrP-Human
PrP C is a 209-amino acid glycoprotein anchored to the cell membrane by a glycosylphosphatidylinositol (GPI) moiety (Prusiner and DeArmond, 1994). As is the case for other membrane glycoproteins, most of the PrP C molecules receive N-linked oligosaccharide chains, a disulfide bond, and the GPI anchor while being translocated into the lumen of the endoplasmic reticulum (ER) (Vidugiriene and Menon, 1994). During this processing PrP C is also folded. The oligosaccharide chains are trimmed in the ER and additional modifications take place in the Golgi apparatus. Human PrP C has two sites for N-glycosylation corresponding to residues 181 and 197 (Locht et al., 1986;Robakis et al., 1986). Three major forms of mature PrP C are found which differ in their degree of glycosylation and are referred to as glycoforms. They include one unglycosylated form and two glycosylated forms, presumably mono-and diglycosylated (Caughey et al., 1989;Monari et al., 1994).
We used human neuroblastoma cells transfected with DNA constructs expressing the human prion protein sequence to determine the effect of the D178N mutation on the metabolism of PrP (Fig. 1). The constructs are based on an episomal vector, so hygromycin-resistant bulk cultures, rather than cloned cell lines, were used. The advantage of this system is that the expression level is not influenced by the site of integration facilitating direct comparison in metabolic studies of the PrP expressed by different constructs. Initially, we compared the cell surface forms of PrP C and PrP M expressed in transfected cells following cleavage of the GPI anchor with the enzyme PI-PLC ( Fig. 2A).
The PrP C released from transfected control cells shows three distinct glycoforms which are similar to those previously described in non-transfected mouse neuroblastoma cells ( Fig. 2A; Caughey et al., 1988;Scott et al., 1988). Since our cells overexpress PrP C we compared, in a separate experiment, the ratio of PI-PLC released glycoforms with those of a non-transfected human cell, M17 BE2C a clonal line from the same tumor, to ensure that we were not observing a processing bias toward less glycosylated forms due to an overload of the glycosylation system. The non-transfected control cells express the three PrP glycoforms in a ratio similar to that of our transfected cells expressing normal PrP. This indicates that the processing observed in the transfected cells is an accurate reflection of the normal pathway (data not shown). PrP M released from the cells transfected with the D178N mutant coding sequence and either the 129M codon (129M,D178N; FFI haplotype) or the 129V codon (129V,D178N; CJD haplotype) show two differences when compared with the control cells ( Fig. 2A): 1) an overall decrease in the PI-PLC-released PrP M (the loading of the samples from the mutant cells was adjusted by a factor of 3 so that they would be comparable to the control cells) and 2) a selective decrease in the forms that migrate most rapidly in gels. Densitometric analyses demonstrated that only one-third the amount of PrP was released from the surface of the cells expressing PrP M compared to cells expressing PrP C (1:0.34 Ϯ 0.18; n ϭ 5; p Ͻ 0.0002) and the most rapidly migrating form, which accounts for 8.3 Ϯ 3.6% (n ϭ 5) of the PI-PLC released PrP in the control cells, is virtually undetectable in the mutant cells. The cells transfected with the vector alone show no immunoreactivity, indicating that the PrP expressed in the transfected cells is coded entirely by the inserted constructs.
After treatment of the PI-PLC released PrP with PNGase F, which cleaves N-linked oligosaccharides, the PrP migrates as a single band (Fig. 2B) which has the same mobility as the most rapidly migrating form in the untreated preparations. This finding confirms that the three PrP forms differ in N-glycosylation and that the most under-represented PrP M glycoform at the cell surface is the unglycosylated form.
To confirm that the unglycosylated PrP M was missing from the cell surface, and not merely resistant to PI-PLC, we surface labeled cells with biotin that were untreated or previously treated with PI-PLC. As shown in Fig. 3, cells expressing PrP M showed a marked reduction in surface labeling after treatment with PI-PLC indicating that mutant PrP M is not resistant to PI-PLC. Thus, the under-representation of unglycosylated PrP M released from the surface of the cells by PI-PLC (Fig. 2) is not a result of an inability to cleave the GPI anchor of PrP M .
Metabolism of the PrP-The under-representation of PrP M on the surface of the mutant cells could be the result of rapid turnover or unsuccessful transport. In addition, the underrepresentation of the unglycosylated PrP M may result from its selective degradation or increased glycosylation. We investigated synthesis and glycosylation of PrP M using a pulse-chase paradigm. After the chase, the cells were treated with PI-PLC to remove the GPI-anchored protein from the cell surface. The cell lysate (intracellular PrP) and the medium (cell surface PrP) were immunoprecipitated and analyzed by SDS-PAGE.
At the end of the labeling period (time 0; Fig. 4A), all cell lysates exhibit the same immunoprecipitable bands in comparable amounts including three distinct bands and a slowly migrating smear. The three discrete bands are a high mobility band corresponding to the unglycosylated PrP and two bands with slightly lower gel mobility. The latter two bands are sensitive to digestion with endoglycosidase H (endo H), indicating that they correspond to the PrP forms glycosylated with the high mannose core and that they are still located in the ER or the Golgi complex (data not shown; Caughey et al. (1989)). After a 30-min chase, as a result of modifications to the sugar chains, these two forms become endo H resistant (data not shown). The slowly migrating smear is the PrP on which the carbohydrates have been highly modified (Caughey et al., 1989). At 2 h, only mature forms of PrP are detected (Fig. 4A) types accounts for 1.9 Ϯ 2.3% of the total PrP M while the unglycosylated PrP C accounts for 8.2 Ϯ 2.6% in control cells (n ϭ 6; p Ͻ 0.003; Fig. 4C). Although there is also an apparent under-representation of the intermediate form, as noted above, the species migrating at this position change with time making quantitation difficult. Similar results are obtained using antibodies directed against the amino (anti-N) and carboxyl (anti-C) termini, except that anti-N recognizes an additional band at 20 kDa which is an unglycosylated, carboxyl-terminal truncated form of PrP and anti-C detects a truncated form of the unglycosylated PrP, not recognized by 3F4, that migrates at approximately 18 kDa (data not shown; Chen et al. (1995)). Truncated forms lacking the amino terminus have been described (Pan et al., 1992;Harris et al., 1993;Chen et al., 1995).
The unglycosylated PrP M is a significant fraction of the total PrP M at the zero time point indicating that this form is normally produced by all cell types (18.0 Ϯ 2.2% versus 23.3 Ϯ 2.2% in control cells; n ϭ 8). At the zero time point, the labeled PrP is still in the ER-Golgi complex region as shown by the barely detectable quantity of PrP that is PI-PLC-released from the cell surface (Fig. 4B). The decrease of the intracellular unglycosylated PrP M occurs at the 30-min and 2-h time points when the glycosylated PrP M forms are still intracellular or are being inserted into the plasma membrane. When analyzed separately, the unglycosylated cell surface PrP M is undetectable after 2 h in the 129M mutant cells (n ϭ 4), but accounts for 0.6 Ϯ 0.4% of total PrP in 129V mutant cells (n ϭ 4; Fig. 4D). In contrast, the unglycosylated form accounts for 3.2 Ϯ 1% and 5.3 Ϯ 0.6% in 129M and 129V control cells (n ϭ 4; p Ͻ 0.008 (129 M) and p Ͻ 0.0008 (129V) two tailed t test; Fig. 4D). These findings, and the previous finding that PrP M is under-represented at the cell surface in the steady state, indicates that PrP M is inefficiently transported through the secretory pathway. This is especially evident for the less modified glycoforms. The absence of the truncated, unglycosylated 20-kDa form on the surface of the mutant cells also supports this conclusion (Fig. 4B).
In addition to the surface and intracellular forms of PrP detailed above, we detected a small amount of PrP (Ͻ5% of the total) in the chase medium without addition of PI-PLC (data not shown). Epitope mapping indicates that the secreted PrP corresponds to the previously described PrP form lacking the last four carboxyl-terminal residues and the GPI anchor (Stahl et al., 1993). The secreted PrP is under-represented in the media from the mutant cells relative to the normal cells, indicating that the low amount of PrP M at the surface of the mutant cells is not the result of increased secretion.
Transport and Stability of the PrP-To better characterize the effect of glycosylation on the transport of the PrP M , we labeled cells in the presence of the antibiotic tunicamycin which blocks the synthesis of the high mannose core preventing glycosylation of newly synthesized proteins in the ER (Elbien, 1987). We analyzed the cell lysate and media from PI-PLC-treated cells as in the pulse-chase experiment described above, with or without addition of tunicamycin. In contrast with the untreated controls, the cell lysates from normal and mutant tunicamycin-treated cells show only one readily detectable PrP band at the zero time point (Fig. 5A) which, as expected, migrates with the unglycosylated form of the PrP. After a 4-h

FIG. 4. Unglycosylated PrP M is not retained in cells (A) nor does it efficiently reach the cell surface (B). A, cells were labeled
for 30 min at 37°C with Trans 35 S-label, the media was removed, and cells were chased for the times indicated above the lane. PI-PLC was added 30 min before the end of the chase after the chase media was removed. The cell homogenate was immunoprecipitated with monoclonal antibody 3F4. The precipitated PrP was resolved on 16% SDSpolyacrylamide gels and visualized by fluorography. M, mature; I, intermediate; U, unglycosylated. B, the media from the PI-PLC-treated cells was immunoprecipitated and analyzed in the same manner as the cell homogenate. M, mature; I, intermediate; U, unglycosylated; 20kd, minor proteolytic product. C and D, graphical representation of the glycoforms of PrP present in the cell or on the cell surface as a percent of total PrP (panel C: *, p Ͻ 0.003; panel D: *, p Ͻ 0.008, p Ͻ 0.0008). chase period, the intracellular PrP, although decreased by over 90% (93 Ϯ 2.3%; n ϭ 6), is still present in the tunicamycintreated cells expressing PrP C (Fig. 5A). In contrast, after the 4-h chase, the PrP M is barely detectable in the tunicamycintreated mutant cells although the amount of PrP M in the tunicamycin treated control and mutant cells is comparable immediately after labeling (Fig. 5A). At the cell surface, PrP C is readily detectable in the tunicamycin-treated control cells at the zero time point and increases 3.5-fold during the 4-h chase (Fig. 5, B and C). In contrast, PrP M is present at one-half the amount of PrP C at the surface of the tunicamycin-treated mutant cells at time 0 and is essentially unchanged over the 4-h chase (Fig. 5C).
The finding that when glycosylation is prevented PrP M barely reaches the cell surface and is undetectable inside the cell shortly after its synthesis lends additional support to the notion that the unglycosylated form of the D178N PrP M is degraded in an intracellular compartment and that PrP M , but not PrP C , requires glycosylation to facilitate transport to the cell surface.
Cellular Compartment of PrP Degradation-We used brefeldin A (BFA), which blocks the transport of glycosylated proteins from the ER through the Golgi complex (Lippincott-Schwartz et al., 1991), to determine whether PrP M is degraded when it is retained in the combined ER-Golgi compartment that forms following BFA treatment. When BFA is added to the cells expressing PrP C , all three PrP C forms synthesized during the labeling period are easily detectable for the entire 2-h chase period (Fig. 6). During this time the two glycosylated forms become insensitive to endo H, indicating that the N-linked carbohydrates have been modified, although to a lesser extent than in untreated cells (data not shown). Over a 2-h chase, all three forms of PrP M are also readily evident. The unglycosylated form from the 129M,D178N cells exhibits a modest decrease as a percent of total PrP over the 2-h chase indicating degradation or conversion to a glycosylated form. When compared with the unglycosylated PrP C , the percent of unglycosylated 129M PrP M decreases by 48.4 Ϯ 2.4% from the 0 time FIG. 5. Tunicamycin impairs transport to the cell surface and alters stability of PrP M . A, cells were preincubated in media lacking methionine and cysteine with or without 2 g/ml tunicamycin. The media was removed and the cells were labeled for 60 min in media containing Trans 35 S-label and 2 g/ml tunicamycin. 30 min into the labeling period PI-PLC was added to one plate. After the labeling period the media was removed, a cell homogenate was immediately prepared from the plate treated with PI-PLC which was immunoprecipitated with the monoclonal antibody 3F4. Media lacking the 35 S label was added to the second plate, after 3.5 h at 37°C, PI-PLC was added and following an additional 30-min incubation at 37°C the media was removed and a cell homogenate was prepared and immunoprecipitated with 3F4. The immunoprecipitated samples were analyzed on SDSpolyacrylamide gels and visualized by fluorography. M, mature; I, intermediate; U, unglycosylated. B, the media from the PI-PLC-treated cells was immunoprecipitated with monoclonal antibody 3F4, analyzed on SDS-polyacrylamide gels, and visualized by fluorography. M, mature; I, intermediate; U, unglycosylated; 20kd, minor proteolytic product. C, graphical representation of the amount of PrP present on the surface of mutant and control cells treated with tunicamycin expressed as a percent of the total labeled PrP at time 0 or after a 4-h chase. *, p Ͻ 0.0045; two-tailed t test).

FIG. 6. Unglycosylated 129M PrP M is less stable than 129V PrP M in cells treated with brefeldin A.
A, cells were preincubated in media lacking methionine and cysteine and labeled for 30 min in media containing Trans 35 S-label and 1 g/ml brefeldin A. Cells were chased for the times indicated above the lanes, cell homogenates were prepared and immunoprecipitated with monoclonal antibody 3F4. The immunoprecipitated samples were resolved on SDS-polyacrylamide gels and visualized by fluorography. The fastest migrating band is the unglycosylated PrP; the two upper bands are glycosylated with the high mannose core which is modified to endoglycosidase H resistance over the course of the 2-h chase. G, glycosylated; U, unglycosylated. B, graphical representation of the decrease in the unglycosylated PrP as a percent of total PrP after a 2-h chase in brefeldin A-treated cells (*, p Ͻ 0.003). point to the 2-h time point, whereas the percent decrease of PrP C is 27.9 Ϯ 6.5% over the 2-h chase (n ϭ 3, p Ͻ 0.05; two-tailed t test). In the cells expressing 129V,D178N PrP M , the unglycosylated form is more stable than in the cells expressing 129M,D178N as it decreases by only 17.5 Ϯ 1.1% during the 2-h chase. The decrease in the unglycosylated PrP M observed in the two mutant cells is significantly different (n ϭ 3, p Ͻ 0.003). This result is consistent with the observation that more unglycosylated PrP M reaches the cell surface in the 129V,D178N cells than in the 129M,D178N cells. Although there is modest turnover of the unglycosylated PrP M in the presence of BFA, which may reflect conversion to glycosylated forms, the unglycosylated PrP in either the mutant or normal cells is relatively stable in the presence of BFA keeping in mind that in untreated cells expressing PrP M the unglycosylated form is barely detectable at the end of a 2-h chase. Thus, the degradation of the unglycosylated PrP M is likely to occur in a cellular compartment beyond the Golgi, probably the endosomal-lysosomal system. D178N Mutant Cells Lack PrP res -Cell homogenates from normal and mutant cells were tested for the presence of proteinase K-resistant PrP. Enzyme titrations from 0.5 to 10 g/ml in 1-h incubations at 37°C as well as a time course of protease digestion using 5 g/ml proteinase K for 5, 10, 15, and 20 min failed to demonstrate any proteinase K-resistant PrP. Concentrations of proteinase K 10 -200 times greater than this are typically used to demonstrate PrP res in brain homogenates from affected individuals or cell extracts from scrapie-infected cells (Monari et al., 1994;Neary et al., 1991).
PrP M Is Under-represented in the Brain-To determine whether the unglycosylated PrP M is decreased in the brain of subjects with FFI, the ratio of the PrP glycoforms was examined on Western blots of brain microsomal fractions obtained from a normal brain and a brain of an FFI affected subject. The FFI subject had, in addition to the D178N mutation, a deletion of one octapeptide repeat within codon 76 and 91 of the mutant allele. The presence of the deletion does not alter the characteristics of the FFI disease phenotype or of the PrP res when compared to FFI subjects without deletion . Histologically normal brain regions lacking PrP res were used to avoid the potential effect of the lesions and of the conversion into PrP res on the amount of PrP M . Immunoblots of normal human gray matter tissue stained with an anti-N serum show three major bands corresponding to three different glycoforms which migrate as a single band after deglycosylation (Fig. 7). Densitometric analyses revealed that the unglycosylated PrP M is present at one-third of the amount of the unglycosylated PrP C expressed by the normal allele (1:0.33 Ϯ 0.02) (Fig. 7,  lanes 1 and 2). However, after deglycosylation, the product of the two alleles is comparable (1:0.86 Ϯ 0.09), indicating that the decrease of the unglycosylated form is selective (Fig. 7,  lanes 3 and 4). DISCUSSION The central event in the pathogenesis of the prion diseases is thought to be a change in the conformation of PrP C (Prusiner and DeArmond, 1994) that renders the PrP C protease resistant (PrP res ) (Hope et al., 1986;Caughey and Raymond, 1991;Pan et al., 1993;Safar et al., 1993). The abnormally conformed PrP res is believed to act as a template for the conversion of newly synthesized PrP C into PrP res . This mechanism is thought to be shared by all forms of prion diseases: sporadic, transmitted, and inherited. In the sporadic form, the change in conformation of PrP C would be the consequence of either a somatic mutation or of a stochastic event involving the direct conformational modification of a PrP C molecule. In the transmitted forms, the conversion would be triggered by the exogenous PrP res . In the inherited prion diseases the pathogenic mutation presumably predisposes the PrP M to spontaneous conversion into PrP res , however, this conversion occurs as a function of age even though mutant protein is produced throughout the life of the individual.
The study of the metabolism of the PrP M in cells transfected with constructs homologous to the two PRNP haplotypes linked to FFI and CJD 178 was undertaken to assess the metabolic differences, if any, between the two forms of PrP M and PrP C . Studying cell models of FFI and CJD 178 is of special interest since FFI and CJD 178 share the same D178N mutation in PRNP, but have two different phenotypes providing a striking example of phenotypic heterogeneity (Goldfarb, Petersen et al., 1992). Since the only heterogeneity in the PRNP coding sequence between these two diseases is at codon 129 of the mutant allele, the phenotypic differences are likely to be due to the amino acid present at position 129 of PrP M which in all FFI subjects examined to date is methionine and in all CJD 178 subjects is valine (Goldfarb, Petersen et al., 1992;. These differences extend to the PrP res associated with the two diseases (Monari et al., 1994). The ratio of the three glycoforms and the size of the PrP res fragment generated by proteinase K digestion are different. In FFI the unglycosylated form is under-represented and the PrP res fragment generated by proteinase K treatment is smaller than in CJD 178 (Monari et al., 1994). The difference in size of the PrP res fragments, which is due to different sites of cleavage by the proteinase K in the two PrP res forms, is consistent with the hypothesis that the PrP res present in FFI and CJD 178 have distinct conformations. The simplest explanation for these findings, based on a large body of experimental data, is that the presence of methionine or valine at position 129 of the PrP M results in PrP res that differ in the ratios of the glycoforms and in conformation. In turn, these differences determine two distinct disease phenotypes (Monari et al., 1994;. The metabolism of PrP C we observed in the human cell line expressing the normal PRNP is in accord with the data previously reported for the mouse neuroblastoma N2A cells (Caughey et al., 1989;Taraboulos et al., 1992;Harris et al., 1993). In our transfected neuroblastoma cells, the ratio of the three glycoforms is essentially the same as the ratio found in non-transfected cells (data not shown). We also observed an ϳ20-kDa band in the cells transfected with the normal construct which corresponds to one of the N-terminally truncated form described previously Chen et al., FIG. 7. Unglycosylated PrP M is under-represented in the brain of a subject affected by FFI. Membrane fractions obtained from the occipital cortex of a control subjects (lane 1) and one FFI subject heterozygous for a deletion of one of the octarepeats within PRNP codons 76 and 91 (lane 2). The samples were immunoblotted and stained with an antibody that recognizes the amino-terminal region of PrP. As expected the unglycosylated form migrates as two uneven bands in the FFI subjects because of the deletion, whereas in the control the unglycosylated form migrates as a single band. The samples were treated with the enzyme PNGase F to remove the N-linked sugars and analyzed as above (control,lane 3;FFI,lane 4). In the FFI subject the two bands, which now contain all the original glycoforms, are comparable indicating that the unglycosylated form is selectively decreased. (U, unglycosylated; M, mature). 1995). This band was more pronounced following treatment of the cells with tunicamycin or digestion of the samples with PNGase F, indicating that most of this truncated form is glycosylated and consequently co-migrates with the higher molecular weight forms as demonstrated by . In addition to this fragment, we also observed an additional ϳ20-kDa form that was immunoprecipitated by the anti-N serum to the amino-terminal region of PrP C . 3 This fragment is currently being characterized.
The cell lines transfected with the D178N mutant coding sequence and either the methionine or valine codon at position 129 synthesize three high mannose PrP M glycoforms in amounts comparable to those of control cells. The mutant transfectants also express three mature PrP forms, one unglycosylated and the other two glycosylated. However, in contrast to controls, these three forms are all decreased in amount at or after they have passed the trans Golgi compartment and have been transported to the cell surface. Moreover, the decrease preferentially affects the unglycosylated form which is vastly under-represented at the cell surface. This change is more severe in the cells bearing the FFI 129M,D178N coding sequence than in the cells with the CJD 178 129V,D178N sequence. The most reasonable explanation is that the decreased amount of all forms of D178N PrP M and the selective underrepresentation of the unglycosylated form are due to the decreased stability and/or impaired transport of the PrP M .
The experiments with tunicamycin and BFA provide insight into the stability and transport characteristics of the two D178N PrP M that we examined. Following treatment with tunicamycin, which abolishes N-glycosylation, cells expressing PrP C show a single PrP C unglycosylated form (Elbien, 1987). Although, as previously reported, total PrP C is decreased in the tunicamycin-treated control cells (Caughey et al., 1989), the time required for transport to the surface of the unglycosylated form is comparable in treated and untreated cells, indicating that transport of PrP C to the plasma membrane is relatively unaffected by tunicamycin. In tunicamycin-treated cells expressing either the 129M,D178N or the 129V,D178N coding sequence, intracellular PrP M is readily detectable at the zero time point in both cell types, but becomes essentially undetectable at 4 h. These findings demonstrate that the lack of glycans destabilizes PrP M and are consistent with the conclusion derived from the previous experiments that transport of the unglycosylated PrP M form to the cell surface is preferentially impaired in both D178N cells.
The BFA experiment provides information concerning the cellular locale in which the PrP M is digested. In the presence of BFA, N-linked glycoproteins are retained in the BFA-generated ER-Golgi compartment (Sampath et al., 1992). However, in this hybrid compartment the oligosaccharide chains acquire endo H resistance (Kornfield and Kornfield, 1985). As expected, all three PrP C glycoforms are present in significant amounts inside the cell, presumably in the ER-Golgi compartment. They are relatively stable over a 2-h period and are not transported to the cell surface. BFA has a slight but significantly different effect on the unglycosylated form of PrP M expressed in the cells transfected with the 129M,D178N and 129V,D178 constructs. While the unglycosylated form of PrP M is degraded or glycosylated in the 129M,D178N cells at a slightly greater rate than the unglycosylated form PrP C in control cells, it is relatively constant in the 129V,D178N cells. Thus, some of the unglycosylated "FFI-like" PrP M may be digested in the ER-Golgi (Fra and Sitia, 1993). The finding that the unglycosylated form of the PrP M is relatively stable in this compartment suggests that during normal processing in untreated cells the unglycosylated PrP M is digested in the endosomal-lysosomal system. Our data also support the hypothesis that N-linked oligosaccharides aid in the folding process and provide structural stability to proteins (Helenius, 1994;Jethmalani et al., 1994).
Taken together these data indicate that the D178N PrP M is unstable and this instability is partially corrected by N-glycosylation. Thus, while the glycosylated forms are synthesized and transported to the cell surface in a fairly normal fashion, although in reduced amounts, the unglycosylated form remains unstable and is broken down before it reaches the cell surface. In contrast to the recent reports by Harris (1995, 1996) on the metabolism of mouse homologues of several pathogenic human mutations in Chinese hamster ovary cells, the PrP M expressed in our cells was released from the cell surface by PI-PLC suggesting a normal association with the cell membrane. In addition, the PrP M produced in our cells was fully digested by proteinase K. Whether the discrepancies between these results are the consequence of homologous versus heterologous systems or the sequence differences between the human and mouse prion protein remains to be determined.
We examined the PrP M from the brain of a FFI subject with the D178N mutation and one octapeptide repeat deletion in the mutant allele in an area of the brain lacking PrP res (Bosque et al., 1992;. Because of the 8-residue difference, the PrP M and PrP C can be separated by gel electrophoresis. The unglycosylated PrP M was present at one-third of the amount of the unglycosylated PrP C . These findings are consistent with the conclusion that the unglycosylated form of the 129M,D178N PrP M is also unstable in the brain cells of FFI patients and that the neuroblastoma cells transfected with the 129M,D178N construct recapitulate the early metabolic events of the PrP M occurring in brain cells of the FFI patients. The reduced stability of the D178N PrP M may be based on the spatial proximity of residues 178 and 129. Secondary structure predictions generated by the alignment of PrP C sequences place these two residues in close proximity of each other in the hydrophobic core (Huang et al., 1994;Nguyen et al., 1995). If this model of PrP secondary structure is correct, methionine and valine at position 129 would be strategically located to influence the conformation of the D178N PrP M (Huang et al., 1994). The brefeldin A experiment suggests that a subtle but significant difference exists in the processing of PrP M in 129M and 129V cells. The minor differences do not immediately shed light on the different phenotypes observed in FFI and CJD 178 , however, this may reflect the complexity of the disease process. Within the more complex cellular environment provided by the intact brain, and the longer time frame needed to develop the disease, the subtle differences we observe between the 129M and 129V PrP M proteins may be sufficient to result in two distinct diseases.
Finally, although the present study provides no direct data as to the mechanism by which PrP M is converted into PrP res in FFI and CJD 178 , it suggests that in the FFI brain the ratios of PrP res and PrP M glycoforms are comparable. As expected from the studies by Caughey et al. (1990), the prion protein must reach the cell surface for it to be converted to PrP res explaining the observation that in FFI very little unglycosylated PrP res is detected (Monari et al., 1994). Therefore, the under-representation of the unglycosylated PrP res in FFI appears to result solely from the degradation of the unglycosylated form before it reaches the cell surface. inspiration and critical reading of the manuscript. Sandra Bowen provided assistance in preparation of the manuscript.