Cellular phenotyping of secretory and nuclear prion proteins associated with inherited prion diseases.

The pathogenic mechanisms leading from mutations in the prion protein (PrP) gene to infectious disease are not understood. To investigate the possibility that cellular processing of mutant prion protein may contribute to the formation of infectious particles, a mouse PrP model system has been established using the green fluorescent protein. Three novel PrP mutants were examined employing this model system and compared with wild type as well as known mutant PrPs. Two Creutzfeldt-Jakob disease-associated PrP mutants, PrP T188K and PrP T188R, revealed a secretory pathway to the cell membrane and PrP(Sc)-like properties, i.e. enhanced proteinase K resistance and detergent insolubility similar to other mutant PrPs associated with familial prion diseases. Moreover, a recently described disease-related truncated PrP mutant, PrP Q160(Stop), showed an almost exclusive localization in the nucleus and a catabolism along the proteasomal pathway. Therefore, various distinct pathological mechanisms may cause prion diseases, and aberrant cellular processing may be included in the pathogenesis of prion diseases.

Transmissible spongiform encephalopathies in animals and man, also known as prion diseases, are rare fatal neurodegenerative disorders that can occur in sporadic, acquired, or autosomal-dominant heritable forms (1). All inherited prion diseases have been found to segregate with mutations in the coding region of the human prion protein gene (PRNP) on the short arm of chromosome 20. So far, more than 20 mutations have been detected, which are causative for prion diseases, including insertions and point and nonsense mutations. Classically, three heritable forms are distinguished according to their clinicopathological presentation, i.e. familial Creutzfeldt-Jakob disease, Gerstmann-Strä ussler-Scheinker syndrome, and fatal familial insomnia. These diseases are unique in that they can be inherited as well as transmitted to laboratory animals (via an infectious process), implying that infectious agents, called prions, are formed de novo in affected individuals. Prions consist, at least to a large extent, of a conformationally altered isoform (PrP Sc ) 1 of a host-encoded glycoprotein, PrP C . Biochemically, several differences have been described between PrP C and PrP Sc , the most prominent being the enhanced proteinase K (PK) resistance of PrP Sc and its insolubility in non-denaturing detergents. The cellular function of the prion protein is still not entirely understood, but its localization on the cell membrane makes PrP a possible candidate for a cell adhesion or receptor molecule. Recently, it has been suggested that PrP C may play a role in copper homeostasis and metabolism, since PrP binds copper via its N-terminal peptide repeat region (2)(3)(4). In addition, these data point to a function either in synaptic interaction or in the sensitivity of neurons to oxidative stress (5,6). It remains unclear whether the diseaseassociated mutants of PrP exert their effects by loss-of-function or gain-of-(dys)function of the protein and whether the resulting pathogenic mechanisms are identical to those in sporadic disease. It has, however, been demonstrated in several studies employing cell culture systems that PrP Sc from genetic as well as sporadic cases is very similar with respect to its biochemical characteristics (7,8).
To examine the fate of PrP in living cells, the green fluorescent protein (GFP) was inserted as a marker into the N terminus of the mouse prion protein (9,10), similar to a strategy recently published for bovine PrP (25). GFP chimeras with wt as well as mutant PrPs were expressed transiently under the control of a tetracycline-inducible promoter in murine neuroblastoma (N2a) cells (11). The mutant PrPs are mouse homologues of human pathogenic PrPs linked to different types of inherited prion diseases (Table I). The translocation and metabolism of the prion proteins were examined biochemically and by means of confocal laser scanning microscopy (CLSM). Thus, aberrant biochemical as well as cellular characteristics could be revealed for newly described PrP mutants, e.g. an exceptional strong nuclear localization of a shortened PrP molecule, which is rapidly degraded through the proteasomal pathway.
Assay of Protease Resistance-Equal amounts of whole cell extracts at equal volumes were PK-digested (5 g/ml) for 30 min at 37°C. All digestions were performed in 1ϫTNE buffer with 0.2% sarcosyl with no protease inhibitors included. Digestions were stopped by the addition of phenylmethylsulfonyl fluoride (2 mM), and proteins were immediately boiled in 1ϫ Laemmli buffer for 5 min and subjected to Western blot analysis.
Assay of PrP Secretion into the Medium-One day before transfection, equal amounts of cells (1 ϫ 10 6 ) were passaged to sterile 6-well plates and cultured under normal cell culture conditions. 48 h posttransfection, the cells were washed twice with Hanks' basic salt solution and cultured for 45 min in phosphate-buffered saline containing 1.0 mM Ca 2ϩ and 0.5 mM Mg 2ϩ . The cell medium was centrifuged at 660 ϫ g for 10 min at 4°C to discard cell debris, and the proteins in the supernatant were methanol-precipitated with at least 6 volumes of ice-cold methanol for several hours at Ϫ20°C. After centrifugation (20,000 ϫ g, 10 min, 4°C), the protein pellets were dissolved in TNE buffer containing 0.2% sarcosyl and subjected to Western blot analysis.
Assay of Detergent Insolubility-To determine the ratio between soluble and insoluble PrP, the cells were lysed 48 h post-transfection in ice-cold buffer containing 100 mM NaCl, 10 mM EDTA, 10 mM Tris (pH 7.5), 0.5% (v/v) Nonidet P-40, and 0.5% (w/v) sodium deoxycholate as detergents (17). Nine volumes of the same buffer (ice-cold) were added to the whole cell extracts before centrifugation (660 ϫ g, 10 min, 4°C) to remove nuclei and cell debris. The supernatants were centrifuged at 20,800 ϫ g for 100 min at 4°C to obtain detergent-soluble (supernatant) and detergent-insoluble (pellet) fractions. After methanol precipitation, both protein fractions were dissolved in lysis buffer and subjected to deglycosylation with peptide N-glycosidase F (New England Biolabs) for better PrP amount discrimination. The extracts were treated with 500 units of peptide N-glycosidase F for 4 h at 37°C according to the manufacturer's instructions and assayed by Western blot analysis.
Assay of Metabolic Effects on PrP by Inhibition of Degradation Pathways-Equal amounts of cells were passaged into 6-well plates and subjected to transfection with the DNA constructs. 48 h post-transfection, cells were incubated for 5 h in fresh medium in the presence or absence of the indicated inhibitors (30 M MG-115, 30 M lactacystin, 30 M NH 4 Cl, and 100 M chloroquine). The proteasome inhibitors MG-115 and lactacystin were from Calbiochem, and chloroquine was from Sigma. Afterward, cells were lysed in TNE buffer containing 0.2% sarcosyl, and cell extracts were subjected to Western blot analysis.
CLSM-Live cell confocal microscopy was performed on Leica TCS confocal systems (Leica TCS NT/SP2). All images were taken with 63 ϫ 0.9NA Achroplan and 40 ϫ 1.3NA Fluar objectives (Carl Zeiss). To study cellular processes, cells were imaged in Dulbecco's modified Eagle's medium that had been preequilibrated in a 10% CO 2 incubator. The cells were maintained on the microscope stage without any gas exchange using a slide chamber fitted for 12-mm round coverslips together with a heating plate. For multi-channel imaging, each fluorescent dye was imaged sequentially in the frame-interlace mode to eliminate cross talk between the channels. GFP fluorescence was excited with a 488-nm argon-ion laser line and imaged using a RSP 500 main beam splitter and a 515-nm longpass filter. Alexa 594 and ceramide TR fluorescence was excited with a 543-nm HeNe laser line and imaged using a double dichroic filter and 590-nm longpass filter settings. All image processing was performed using the Leica TCS NT software; Adobe Photoshop was used to prepare final figures for publication.

GFP-wtPrP Is Processed and Localized Like Native PrP C -
The protein chimera GFP-wtPrP was created by the insertion of the GFP-S65T DNA sequence between codons 39 and 40 of the wt mouse PrP open reading frame (Fig. 1A). To compare the biosynthesis of the GFP-wtPrP chimera with that of wt PrP C , the expression of both was first examined biochemically. Western blot analysis of transfected recombinant N2a cells (11) confirmed the synthesis of glycosylated proteins with molecular masses of about 48 -58 kDa for GFP-wtPrP and 27-37 kDa for native PrP C . Non-chimeric as well as chimeric PrP could be detected with antibodies against mouse PrP, but GFP-wtPrP could also be recognized by antibodies against GFP (Fig. 1B). To investigate whether the cellular transport of GFP-wtPrP is comparable with that of the secretory and membrane-bound protein PrP C , the localization of GFP-wtPrP in N2a cells was observed from about 20 h post-transfection by CLSM and assayed by co-localization studies. Thereby, a co-localization of GFP-wtPrP could be detected with an ER marker (Fig. 1C, upper panel) and most prominently with a Golgi marker (Fig.  1C, lower panel), but no co-localization was detectable with organelle markers for mitochondria, cytoskeleton components, or the nucleus (data not shown). Further observation of post-Golgi transport of GFP-wtPrP revealed the enrichment of fluorescent signals on the cell surface (Fig. 1D), and prolonged observations (here shown for 48 h post-transfection, but also detectable in long term, stable N2a clones expressing GFP-wtPrP) 2 confirmed a final GFP-wtPrP signal distribution in transfected N2a cells with prominent localizations in the Golgi apparatus and on the cell membrane (Fig. 1E). To exclude the possibility that even incomplete GFP-PrP chimeras show an identical cellular phenotype like GFP-wtPrP, control fusion proteins were designed carrying either the PrP sequence Cterminally (GFP-PrP⌬N) or N-terminally (GFP-PrP⌬C) in addition to GFP (Fig. 1A). Both fusion proteins could be detected in Western blot analysis as truncated and unglycosylated proteins (data not shown) and could furthermore be observed in confocal microscopy studies as homogeneously distributed signals throughout the cytoplasm and the nucleus, similar to the signal distribution of GFP alone (Fig. 1F).
Expression of Chimeric PrP Mutants-To investigate the processing and transport of PrP mutants in N2a cells by this model system, seven GFP-PrP chimeras were constructed that correspond to several distinct mutations in inherited human prion disease (Table I). Among these PrP mutants, there were three, GFP-PrP Q159 Stop , GFP-PrP T187K, and GFP-PrP T187R (corresponding to human PrP Q160 Stop , T188K, T188R), that have so far never been cellularly examined and which were assayed in this study most intensively. In addition, four cellularly well defined PrP mutants (GFP-PrP P101L, GFP-PrP W144 Stop , GFP-PrP D177N, and GFP-PrP E199K as the mouse homologues for human PrP P102L, Y145 Stop , D178N, E200K) were used as controls to evaluate the GFP-PrP model system and to extend the existing knowledge about these mutants (7,17,19,20). After expression in N2a cells, the chimeric PrP missense mutants, i.e. GFP-PrP P101L, GFP-PrP D177N, GFP-PrP T187K, GFP-PrP T187R, and GFP-PrP E199K, were fully glycosylated and synthesized to molecular masses of about 48 -58 kDa like the chimera with wt PrP ( Fig. 2A). All these glycosylated chimeras were sensitive to deglycosylation by digestion with peptide N-glycosidase F (data not shown), as is the case with native PrP. The glycosylation of the GFP-PrP chimeras is therefore neither blocked by the insertion of GFP nor by the disease-linked point mutations in mouse PrP. In contrast to the missense mutants, the chimeric PrP nonsense mutants GFP-PrP W144 Stop and GFP-PrP Q159 Stop appeared in Western blot analysis as truncated and double-banded proteins of about 40 -42 and 41-43 kDa, respectively ( Fig. 2A).
To test full-length chimeric PrP mutants with respect to their resistance against proteolysis, total cell extracts of transfected N2a cells were digested under mild conditions with PK E200K CJD a Mouse PrP contains methionine at position 128, which corresponds to codon 129 in human PrP, a polymorphic site encoding either methionine or valine.
b Mouse and human PrP differ at their corresponding positions. c EOD, familial early onset dementia. d The T188R/129V haplotype is linked to CJD. No disease-linked phenotype has been described for T188R/129M. (5 g/ml, 37°C, 30 min). Under similar conditions, enhanced PK resistance has been detected for the non-chimeric counterparts of the PrP mutants with amino acid exchanges P101L, D177N, and E199K (7,19). Transfections were carried out under identical conditions, and 72 h post-transfection lysates were prepared and assayed with matched concentrations of total protein in equal volumes (Fig. 2B). All full-length chimeras with pathogenic mutations were tested and showed an enhanced resistance against PK leading to truncated molecules. All these resistant protein signals were clearly and reproducibly visible in Western blot analysis, indicating that the chimeric PrP mutants also underwent biochemical changes that led to increased PK resistance.
To investigate the cellular localization of the PrP mutants, the expression of the chimeras was observed by CLSM. The microscopic examination revealed that the chimeric mutant proteins GFP-PrP P101L, GFP-PrP D177N, GFP-PrP T187K, GFP-PrP T187R, and GFP-PrP E199K showed no difference in localization from GFP-wtPrP (Fig. 2C). For all of these, fluorescent signals were detectable in the Golgi apparatus and on the cell membrane of N2a cells. These chimeras were trafficked along a secretory pathway like native PrP, and no differences in cellular transport and localization or the level of expression were detected. In contrast to the other chimeras, no fluorescent signals in the Golgi apparatus and on the cell membrane were detectable for the truncated chimeras GFP-PrP W144 Stop and GFP-PrP Q159 Stop (Fig. 2C). Expression of the GFP-PrP W144 Stop chimera led to a distribution throughout the cytoplasm and the nucleus, whereas the GFP-PrP Q159 Stop mutant showed an almost exclusive localization in the nucleus that was clearly distinguishable from GFP-wtPrP but also from GFP-PrP W144 Stop . All cellular phenotypes presented here could be confirmed for the chimeric and the corresponding non-chimeric PrPs by immunofluorescence analysis with anti-PrP antibodies (data not shown).
Cellular Trafficking of the Chimeric PrP Mutants-The three chimeric PrP mutants GFP-PrP T187K, GFP-PrP T187R, and GFP-PrP Q159 Stop were further analyzed with respect to their cellular transport. To investigate the intracellular translocation between ER and Golgi apparatus, the transfected cells were incubated with the fungal metabolite brefeldin A (BFA). BFA reversibly initiates the retrograde transport of Golgi vesicles into the ER, which can be stopped by a washout of the drug that leads to a subsequent reorganization of the Golgi apparatus (21). By using BFA as a cellular tool, secretory proteins could be cycled from the Golgi into the ER and back, which allows the determination and the microscopic observation of the cellular pathway(s) of the GFP-PrP chimeras. Timelapse observations of the GFP-wtPrP signal distribution in BFA-treated N2a cells revealed the complete disappearance of discrete Golgi signals within 60 min and the appearance of an "ER-typical" signal distribution (Fig. 3A). This BFA-induced phenotype was also detected in N2a cells expressing GFP-PrP T187K and GFP-PrP T187R (Fig. 3B) but was never observed for GFP-PrP Q159 Stop -expressing cells (Fig. 3A). Transfected cells showed the redistribution of GFP-PrP T187K and GFP-PrP T187R into the ER in the presence of BFA and a reorganization of discrete fluorescent signals or Golgi signals, respectively, in the absence of BFA, which clearly indicates a secretory pathway along the ER/Golgi complex for these two chimeric PrP mutants (Fig. 3B). In contrast, the chimeric PrP nonsense mutant GFP-PrP Q159 Stop showed no effect on BFA incubation of the cells. The signal distribution remained restricted to the nucleus, and no changes in the signal localization occurred even after BFA incubation for up to 20 h (data not shown), indicating that GFP-PrP Q159 Stop never entered the Golgi apparatus along its cellular pathway. To investigate the localization of the chimeric PrPs on the cell surface, the transfected cells were incubated with phosphatidylinositol-specific phospholipase C (PIPLC), which cleaves GPI-anchored proteins like native PrP C from the cell membrane. During PIPLC treatment for up to 40 min, the GFP-wtPrP-, GFP-PrP T187K-and GFP-PrP T187R-expressing cells showed an ongoing disappearance of the fluorescent signals on the cell surface, whereas no loss of fluorescence was observed for cells expressing GFP-PrP Q159 Stop (Fig. 4). To exclude that hypothetically existing membrane-bound GFP-PrP Q159 Stop was for some reason not accessible to PIPLC, we applied the protease trypsin into the culture medium and assayed the trypsin digest by Western blot analysis of whole cell extracts. However, whereas high reductions of the protein content were detected for N2a cells expressing the secretory chimeric PrPs, no effect on the protein amount was measurable for GFP-PrP Q159 Stop (data not shown). This result supported the confocal recordings that GFP-PrP Q159 Stop seemed to be exclusively localized in intracellular compartments without any trafficking along the ER/Golgi complex to the outer-membrane like wt and several mutant PrPs (Fig. 2C).
Expression and Cellular Trafficking of Non-chimeric PrP Mutants-Further examinations of PrP trafficking were carried out with the non-chimeric mouse wt and mutant PrPs. Corresponding to the analysis of the chimeric PrPs, the synthesis of the native PrPs was first examined biochemically. Western blotting revealed that the expression in N2a cells yielded glycosylated wt PrP, PrP T187K, and PrP T187R with molecular masses of ϳ27-37 kDa (Fig. 5A). Similar to the chimeric counterparts, double-banded signals could also be detected for the truncated PrP mutants, with molecular masses of ϳ14 and16 kDa for PrP W144 Stop and 15 and17 kDa for PrP Q159 Stop (Fig. 5A). Therefore, the expression pattern of the native PrPs clearly resembled that of their chimeric homologues and revealed the expected molecular weights of the prion proteins without the GFP portion.
To evaluate the amount of secreted PrP of the PrP mutants, transfected cells were cultured in fresh salt solution (physiological) for 45 min, and the proteins in the culture medium were methanol-precipitated and subjected to Western blot analysis. Under these conditions, secretion into the medium could easily be detected for wt PrP, PrP T187K, and PrP T187R but not for PrP W144 Stop and PrP Q159 Stop (Fig. 5B), support-ing our data that the PrP nonsense mutants were strictly restricted to intracellular regions and, in contrast to the other examined PrPs, not trafficked along a secretory pathway. Identical results were also obtained for the corresponding chimeric PrPs (data not shown).
Biochemical Properties of the PrP Mutants-To investigate whether PrP T187K and PrP T187R also exhibit an enhanced PK resistance like many other disease-associated full-length PrP mutants (7,8) or their chimeric PrP homologues, respectively (see Expression of Chimeric Prp Mutants in the "Results" section), whole cell extracts of N2a cells expressing PrP T187K and PrP T187R were digested with PK (5 g/ml, 37°C, 30 min) and assayed by Western blotting. Similar to their chimeric counterparts and in contrast to wt PrP, mutant PrP T187K and PrP T187R showed an enhanced PK resistance leading to truncated instead of totally digested protein and, by this, confirming the results of the GFP-PrP chimeras (Fig. 6A). In addition, the PK resistance of the shortened PrP mutants, W144 Stop and Q159 Stop , was also examined by applying identical conditions of PK incubation as used for the full-length PrP mutants. However, no specific PrP signal could be detected neither for the non-chimeric nor the chimeric shortened PrP mutants, which strongly argues for a higher sensitivity of PrP W144 Stop and PrP Q159 Stop to digestion with PK (data not shown). To test the solubility of the PrP mutants in non-denaturing detergents, whole cell extracts of transfected N2a cells were diluted with 9 volumes of ice-cold lysis-buffer and subjected to high speed centrifugation. Supernatant proteins were methanol-precipitated, and both pellet and supernatant proteins were prepared for SDS-PAGE and Western blot analysis. Under these conditions, a higher amount of wt PrP could be detected in the supernatant fraction, indicating an overrepresentation of soluble versus insoluble wt PrP (Fig. 6B). In contrast, PrP T187K and PrP T187R showed a faintly but clearly detectable overrepresentation in the pellet fraction, which indicates, in comparison to wt PrP, a higher portion of insoluble protein for these PrP mutants (Fig. 6A). Interestingly, the nonsense mutants PrP W144 Stop and PrP Q159 Stop displayed a detergentdependent discrimination of their truncated protein forms, which could be assayed by Western blot analysis. The more slowly migrating protein signal of PrP W144 Stop and the faster migrating signal of PrP Q159 Stop were represented in both the  supernatant and the pellet fraction, with a faintly overrepresentation for both mutants in the pellet fraction. However, the lower molecular weight form of PrP W144 Stop could predominantly be detected in the supernatant fraction, whereas the higher molecular weight form of PrP Q159 Stop was almost exclusively represented in the pellet fraction, indicating a higher content of insoluble protein for PrP Q159 Stop compared with PrP W144 Stop (Fig. 6B).
Degradation Pathways of the PrP Mutants-Recent studies with proteasome inhibitors have revealed that the human PrP nonsense mutant, PrP Y145 Stop , is degraded very rapidly in transfected M17 cells and that blockade of proteasomal degradation resulted in the accumulation of this PrP mutant (20). To test if application of proteasome inhibitors also affects accumulation of the other truncated PrP mutant, PrP Q159 Stop , transfected N2a cells were incubated with MG-115 (30 M), which specifically inhibits chymotrypsin-like activity of the proteasome, and with lactacystin (30 M), which acts as a covalent inhibitor of the trypsin-and chymotrypsin-like activities of the proteasome. Cell extracts were prepared after incubation of the cells with these inhibitors for 5 h, and equal amounts of total protein were subjected to Western blot analysis (Fig. 7A). Both mouse homologues, PrP W144 Stop and PrP Q159 Stop , revealed a strong increase in PrP amount in comparison to control extracts coming from the same transfected cells incubated with identical volumes of inhibitor-free solvent (Me 2 SO).
To assess the effect of acidic hydrolases on the degradation of the truncated PrP mutants, NH 4 Cl (30 M) and chloroquine (100 M) were added to N2a cells expressing PrP W144 Stop and PrP Q159 Stop . These reagents raise the pH of acidic compartments and thereby inhibit the actions of acidic hydrolases in the lysosomal and endosomal degradation pathways (22). Western blot analysis revealed that even after 24 h of incubation with NH 4 Cl and chloroquine, no significant increase was detectable for PrP W144 Stop and PrP Q159 Stop (Fig. 7A), indicating a very exclusive catabolism of the truncated PrP mutants via the proteasomal pathway. To test if these mutants are ubiquitinated in N2a cells, both proteasome inhibitor-treated and untreated cells expressing PrP W144 Stop and PrP Q159 Stop were immunoprecipitated with antibody 3B5 (16) and immunoblotted with anti-ubiquitin antibodies to detect ubiquitinated PrP forms. However, no ubiquitinated PrP W144 Stop and PrP Q159 Stop could be detected in Western blotting, regardless if cells were inhibitor-treated or untreated (data not shown).
To investigate the cellular phenotypes of the PrP nonsense mutants after blockade of proteasomal degradation, the expression of their chimeric PrP counterparts, GFP-PrP W144 Stop and GFP-PrP Q159 Stop , were assayed after incubation of cells with MG-115 (30 M) and lactacystin (30 M) for 5 h (Fig. 7B). Aberrant intracellular fluorescent deposits, i.e. small dot-like structures and broader signal accumulations, were clearly detectable in the cytoplasm and in the nucleus, confirming for both mutants the presented biochemical results (Fig. 7A) of an impaired degradation after incubation with proteasome inhibitors. All experiments involving the proteasome inhibitors were also done for wt PrP, PrP T187K, and PrP T187R, but no significant increase (or decrease) after application of MG-115 or lactacystin could be detected in a Western blot or in CLSM analysis (data not shown), indicating a non-proteasomal degradation of these full-length PrPs.
PrP Q159 Stop Is a Short-lived Protein That Shows a Directed Influx into the Nucleus-In comparison to GFP-PrP W144 Stop (and GFP-PrP⌬N and GFP-PrP⌬C, see GFP-wt Prp Is Processed and Localized Like Prp C in the "Results" section), microscopic analysis constantly revealed that GFP-PrP Q159 Stop localization was almost exclusively restricted to the nucleus of transfected N2a cells. To investigate the tendency of this protein to traffic into the nuclear compartment, bleaching experiments (23) were performed to determine the fluorescent recovery after photobleaching of the nuclear signals. The fluorescent signals in the nucleus were specifically and irreversibly bleached (10 s) with high intensities of laser light (4 -5 times higher than the scan mode), and the redistribution of new fluorescent signals were successively recorded under nonbleaching conditions. Very rapidly after complete bleaching of the nuclear fluorescence, new fluorescent signals appeared exclusively in the nucleus, and 8 min post-bleaching, the nuclear fluorescence was re-established similar to the prebleach situation (Fig. 8, a-d). This procedure was repeated several times, and a redistribution of new fluorescent signals was always restricted to the nucleus (data not shown), indicating a directed influx of this 41-43-kDa protein into this compartment instead of an unspecific homogeneous distribution throughout the cytoplasm and the nucleus.  1 and 2), PrP T187K (lane 3), and PrP T187R (lane 4) were treated (ϩ) or not treated (Ϫ) with 5 g/ml PK for 30 min at 37°C and analyzed by immunoblotting with monoclonal anti-PrP antibody 6H4. B, whole cell extracts of N2a cells expressing wt PrP (lanes 1 and 2), PrP T187K (lanes 3 and 4), PrP T187R ( lanes 5 and 6), PrP W144 Stop (W144X, lanes 7 and 8), and PrP Q159 Stop (Q159X, lanes 9 and 10) were high speed-centrifuged at 4°C for 100 min to separate detergent-soluble from detergent-insoluble molecules. Proteins in supernatant (S) fractions were methanol-precipitated, and supernatant (S) and pellet (P) fractions were subjected to Western blot analysis with antiserum Kan72. Before SDS-PAGE, samples were enzymatically deglycosylated by peptide N-glycosidase F to produce a single signal for better PrP quantification. The asterisks indicate the lower ( * ) and the higher ( ** ) molecular mass forms of PrP W144 Stop (W144X, lanes 7 and 8) and PrP Q159 Stop (Q159X, lanes 9 and 10). The bleaching of exclusively non-nuclear regions of the cells, however, revealed a continuous decrease of the nuclear fluorescence intensity that finally, after three identical bleaching events (ϳ4 min between each bleaching event, same bleaching conditions as above), led to the disappearance of any nuclear fluorescent signal within 10 min. Thus, loss of fluorescence in the nucleus after bleaching of large amounts of GFP-PrP Q159 Stop before reaching the nucleus indicates a rapid degradation of the nucleus-residing chimeras within 10 min (Fig. 8,  e-h). Noteworthy, by bleaching comparable non-nuclear regions in MG-115-treated cells, the fluorescent signal in the nucleus was, in contrast to the untreated cells, still clearly detectable after three bleaching events (ϳ4 min between each bleaching event, identical conditions as above) (Fig. 8, i-k, compare h and k) and was even easily observable after further bleaching events (data not shown), indicating a delayed degradation due to the incubation with the proteasome inhibitor. In addition, the presence of MG-115 enabled a better detection of the GFP-PrP Q159 Stop molecules before being located to the nucleus and revealed an arrangement of dot-like structures (Fig. 8, l), which could be hypothesized either as dislocation sites of the chimera from the ER membrane or ribosome-associated GFP-PrP Q159 Stop molecules (see "Discussion").

DISCUSSION
To establish a model system for the cellular observation of prion proteins by noninvasive means, mouse PrP chimeras with the GFP marker molecule have been generated that directly permit spatial and temporal recordings of PrP localization in living N2a cells. For the construction of the chimeras, we decided to insert GFP into the N terminus of PrP, between the signal sequence and the octapeptide repeat region of PrP, since the N-terminal part of the protein has been described to be, in contrast to the structured C-terminal part, highly flexible (24). In addition, we preferred the insertion of GFP into the Ninstead of the C-terminal moiety to ensure the cellular phenotyping of disease-related PrP mutants that lack the complete C terminus of PrP. Recently, the production of a GFP-PrP chimera has been described that succeeded by a similar strategy of GFP insertion in the imaging of wt and mutant bovine PrPs in mammalian cell lines (25). However, due to the large amount of data existing with the well established mouse PrP model system, we decided to create mouse PrP-GFP chimeras for a more reliable comparison with the existing results.
The chimera with wt PrP (GFP-wtPrP) behaved like native PrP in so far as it was subject to the known cellular processing of PrP C . It was fully glycosylated and, following a secretory pathway, was transported via the ER/Golgi route and tethered on the cell surface by its GPI anchor. Like PrP C , the GFP-wtPrP chimera possessed no enhanced resistance to proteinase K and was therefore clearly distinguishable from PK-resistant PrP mutants and PrP Sc .
Beyond biochemical or immunological analysis, which depend on the quality of the detection systems, our approach also provided the opportunity to directly examine the expression patterns of mutant PrPs in living cells and compare the cellular phenotypes with data from previous studies. In addition to the wt PrP chimera, seven chimeric PrP mutants were constructed and expressed in N2a cells. These mutants corresponded to known alterations in the prion protein associated with inherited prion diseases (26 -33). Biochemical as well as cellular data have been described for four of the seven disease-related PrP mutants, i.e. the three full-length PrP mutants PrP P102L, PrP D178N, PrP E200K (7,19) and the truncated molecule PrP Y145 Stop (20) but not for PrP Q160 Stop , PrP T188K, and PrP T188R (human PrPs). As described for their non-chimeric counterparts, the chimeras GFP-PrP P101L, GFP-PrP D177N, and GFP-PrP E199K (mouse PrPs) confirmed the expression pattern of glycosylated and secretory proteins, which could be detected predominantly in the Golgi apparatus and on the cell surface. Considering the biochemical characteristics of the scrapie isoform, PrP Sc , we were able to show an increased resistance against proteolysis with PK, which is in agreement with published data of human brain (17,34), transgenic mouse (35), and cell culture experiments (7) about their native PrP counterparts. Thus, even after 3 days of transient transfection and unimpeded by the insertion of GFP, the mutant PrP chimeras (but not the wild type) displayed an enhanced PK-resistance. This intriguing finding argues for a PrP Sc -like conformation of the mutant PrP chimeras.
Two newly described human pathogenic PrP mutants, PrP T188K and PrP T188R, are associated with familial forms of Creutzfeldt-Jakob disease (28,33), albeit histopathological characterizations are missing up to now. Both chimeric PrP counterparts used in this study, GFP-PrP T187K and GFP-PrP T187R, showed a cellular phenotype similar to the other disease-related full-length chimeric PrP mutants, since they were properly glycosylated and transported via the secretory pathway to the cell surface. Although a strongly reduced susceptibility of surface PrPs to PIPLC has been described in some (7,36) but not all (19) studies for disease-related PrP mutants, we were unable to detect an impaired or delayed PIPLC releasability of GFP-PrP T187K and GFP-PrP T187R from the cell membrane in comparison to GFP-wtPrP by direct microscopic observation. Thus, both mutants seem to be connected to the cell exterior exclusively by the GPI anchor. We cannot exclude that our findings might disregard differences between PrP membrane association in transient expression studies and long term stable cell clones. However, it seems unlikely that our results are due to the insertion of GFP, since unaffected PIPLC susceptibility has been shown for surface GFP chimeras with bovine wt and mutant PrP in stably transfected Chinese hamster ovary cell clones (25). It was suggested that the PIPLC resistance of several mutant PrPs is based on a physical inaccessibility of the molecules to the phospholipase, probably as a result of conformational changes that accompany conversion to the PK-resistant state (7,36,37). However, despite the unimpaired PIPLC susceptibility of chimeric PrP T187K and PrP T187R, both PrP mutants (native as well as chimeric) showed an enhanced PK resistance and increased detergent-insolubility compared with the wt controls, which clearly indicates a higher degree of aggregation characteristically for the diseaseassociated PrP isoform (7,17,25). Further investigations can now be conducted to determine whether chimeric mutant PrP aggregation is microscopically observable in long term stable expression systems. Recently, we described the construction of N2a cells, allowing inducible and quantitatively controllable expression of the mouse prion protein (11), which will enable us to perform more detailed studies of these and other chimeric PrP mutants.
A novel nonsense mutation of the PrP gene with a glutamine to stop codon change at amino acid 160 (Q160 Stop ) was found in two siblings of clinically possible or probable Alzheimer's disease (28,32). Western blot analysis of total protein extracts from peripheral blood leukocytes of the patients revealed an aberrant band pattern, suggesting a shortened prion protein.
Our results with the mouse PrP mutant homologue PrP Q159 Stop demonstrated that expression in N2a cells leads to two truncated forms of unglycosylated protein, similar to the expression pattern of the other disease-related stop mutant, human PrP Y145 Stop , in M17 human neuroblastoma cells. This Gerstmann-Strä ussler-Scheinker syndrome-linked variant with an amber mutation at codon 145 (human PrP Y145 Stop , corresponding to mouse PrP W144 Stop ) has been described to not be processed post-translationally and to be rapidly degraded through the proteasomal pathway (20). It has been shown for PrP Y145 Stop that the higher molecular mass form represents a population of the protein that still possesses its N terminus signal peptide (20). Although we did not verify the existence of the signal peptide, a higher molecular mass form of mouse mutant PrP W144 Stop and PrP Q159 Stop could clearly be detected ϳ2 kDa above the lower molecular mass forms in Western blotting, which strongly argues for a subpopulation of these mutant PrPs that conserves the signal peptide, similar to human mutant PrP Y145 Stop (20). The salient histopathological features of human PrP Y145 Stop are widespread PrP amyloid deposits in cerebral vessels and the presence of -positive neurofibrillary lesions (29). The PrP amyloid deposits have been shown to immunostain mainly with antibodies to epitopes included in the PrP sequence spanning the N-terminal 25 amino acids (38) and residues 90 -147 (29), indicating that the major PrP amyloid protein is truncated at the C terminus. No histopathology has been described for the other truncated PrP mutant, human PrP Q160 Stop (28). Microscopic observation of the expression in N2a cells revealed that chimeric PrP Q159 Stop was located to a low extent in the cytoplasm and, with much higher signal intensity, in the nucleus. Recently, the N-terminal extremity of mature PrP has been suggested to harbor a putative nuclear localization signal (nls) reminiscent of the nls of the simian virus 40 T antigen but failed to target a marker protein to the nuclear compartment (39). It therefore remains unclear whether the ϳ40-kDa GFP-PrP Q159 Stop chimera is transported into the nucleus actively or by passive diffusion. However, we were able to demonstrate that the recovery of fluorescent GFP-PrP Q159 Stop in the nucleus takes about 8 min after complete bleaching of the nucleus-residing GFP-PrP Q159 Stop , indicating a directed and rapid influx of the protein into the nucleus. In contrast to GFP-PrP Q159 Stop , the GFP-PrP W144 Stop chimera did not show a restricted nuclear local-ization but revealed, in addition, higher signal intensities in non-nuclear regions (cytoplasm, ER). Both mutants lack large regions of the C-terminal part of PrP, including the GPI anchor and the two sites of N-glycosylation, but differ in their sites of truncation. Although both mutants include the N-terminal octapeptide region and the central highly conserved hydrophobic region (40,41), the shorter mutant (mouse PrP W144 Stop or human PrP Y145 Stop , respectively) lacks almost the whole structured C-terminal region, whereas PrP Q159 Stop (or human PrP Q160 Stop ) completely harbors the first ␣-helix in PrP (42,43). Thus, this additional region might have structural consequences that finally led to the enhanced influx of chimeric PrP Q159 Stop into the nucleus. However, existing structural implications could be confirmed by detergent insolubility assays, which showed a higher content of insoluble protein for PrP Q159 Stop compared with PrP W144 Stop , indicating a higher degree of aggregation for the larger PrP mutant.
Considering the pathways of the truncated mutants with respect to their catabolism, we were able to show that both PrP W144 Stop and PrP Q159 Stop were rapidly degraded by the proteasome, confirming previous results for human PrP Y145 Stop . A half-life of ϳ10 min has been described for human PrP Y145 Stop in M17 human neuroblastoma cells (20), and we were able to demonstrate that nuclear GFP-PrP Q159 Stop was completely degraded within 10 min after bleaching of non-nuclear regions of transfected N2a cells. Thus, both truncated PrP mutants are by far the most unstable PrP forms associated with inherited prion diseases (19,20). Additionally, we were also able to demonstrate that expression of GFP-PrP W144 Stop and GFP-PrP Q159 Stop in the presence of proteasome inhibitors led to protein accumulations in cellular compartments, which are involved in proteasomal action, i.e. the cytoplasm and the nucleus (44). Fluorescent dot-like structures of GFP-PrP Q159 Stop reminiscent of ER exit sites were clearly detectable by taking enlarged confocal images of non-nuclear regions of transfected N2a cells. It has been suggested that human PrP Y145 Stop might be degraded by proteasomes on the cytosolic face of the ER, as previously reported for several secretory proteins (20, 44 -46). Our model system provided first insights into the degradation pathway of the second truncated PrP mutant associated with inherited prion disease, human PrP Q160 Stop . The protein seems to be sorted out from its secretory pathway at distinct sites of the ER membrane to be immediately degraded in the cytosol and nucleus.
The exceptionally strong nuclear localization of chimeric PrP Q159 Stop , regardless of the inhibition or non-inhibition of proteasomal function, might argue for additional properties leading to a directed influx into the nucleus. Recently, the Nterminal region of human and ovine PrP has been described to harbor nucleic acid binding and chaperoning properties similar to those of human immunodeficiency virus-1 NCp7 and retroviral nucleocapsid proteins (47). Although the putative functional domain of PrP responsible for RNA and DNA interactions is still unknown, both truncated PrP mutants were localized to cellular regions of abundant nucleic acid content, i.e. the nucleus and ER/cytoplasm including RNA-rich ribosomes. Before this study, data on the cellular processing and localization were only available for PrP Y145 Stop , which had been shown to be recognized by quality control mechanisms and degraded through the proteasomal pathway (20). We were able to demonstrate that the mouse homologue of PrP Q160 Stop is prevented from exit to the Golgi apparatus and transported to the cytosol and nucleus where it is also rapidly degraded in the proteasomes. Therefore, it seems reasonable to assume that expression of both truncated PrP mutants in the brain of an affected person results in similar pathological mechanisms that are based most probably on the impairment of proteasomal function with advanced age and/or the very slow, but continuous accumulation of truncated protein.