Post-translational import of the prion protein into the endoplasmic reticulum interferes with cell viability: a critical role for the putative transmembrane domain.

Aberrant folding of the mammalian prion protein (PrP) is linked to prion diseases in humans and animals. We show that during post-translational targeting of PrP to the endoplasmic reticulum (ER) the putative transmembrane domain induces misfolding of PrP in the cytosol and interferes with its import into the ER. Unglycosylated and misfolded PrP with an uncleaved N-terminal signal sequence associates with ER membranes, and, moreover, decreases cell viability. PrP expressed in the cytosol, lacking the N-terminal ER targeting sequence, also adopts a misfolded conformation; however, this has no adverse effect on cell growth. PrP processing, productive ER import, and cellular viability can be restored either by deleting the putative transmembrane domain or by using a N-terminal signal sequence specific for co-translational ER import. Our study reveals that the putative transmembrane domain features in the formation of misfolded PrP conformers and indicates that post-translational targeting of PrP to the ER can decrease cell viability.

Prion diseases in humans and animals are characterized by the accumulation of PrP Sc , a partially protease-resistant isoform of the cellular prion protein PrP C . PrP Sc is generated through a conformational transformation of PrP C and represents the major component of infectious prions (reviewed in Refs. [1][2][3][4]. PrP 1 is post-translationally modified by the attachment of two N-linked complex carbohydrate moieties (Asn 180 and Asn 196 ) (5-7) and a glycosylphosphatidylinositol (GPI) anchor at serine 231 (8) as well as by the formation of a disulfide bond between Cys 178 and Cys 213 . Studies with recombinant PrP purified from bacteria revealed that the formation of the disulfide bond is essential for the native folding of PrP (9).
The co-and post-translational modifications of PrP C are initiated with the cleavage of the N-terminal signal peptide (aa  and the transfer of core glycans, whereas the nascent chain is still associated with the translocon. Shortly after the protein is fully translocated, the GPI anchor is attached to the acceptor amino acid close to the C terminus. The final maturation of PrP C includes the processing of the core glycans into complex-type glycans by a series of enzymatic reactions in the endoplasmic reticulum (ER) and Golgi compartment. Posttranslational modifications, like N-linked glycosylation and GPI anchor attachment, are often used as diagnostic markers to monitor efficient import into the ER. In the case of PrP, however, we and others have shown that PrP devoid of a GPI anchor remains mainly unglycosylated but is imported efficiently into the ER and transported through the secretory pathway (10 -14). It has been found that the only specific marker for ER import of PrP is a cleaved N-terminal signal sequence (10). 2 Misfolding of PrP C in the cytosol or in the ER can induce neurodegeneration in the absence of PrP Sc . Neurotoxic properties of cytosolic PrP aggregates were observed after proteasomal inhibition in cultured cells or after the forced expression of cytosolic PrP in transgenic mice (15). Other studies revealed that a minor fraction of PrP can be synthesized as an integral membrane protein with two different topologies, termed either Ntm PrP, with the N terminus facing the ER lumen, or Ctm PrP, with the C terminus facing the ER lumen. Amino acids 112-135 of PrP were identified as a putative transmembrane domain (TM) (16), and mutations in the TM domain were found to alter the relative amount of Ctm PrP and Ntm PrP (17). Remarkably, the increased synthesis of Ctm PrP has been shown to coincide with progressive neurodegeneration both in Gerstmann-Strä ussler-Scheinker syndrome patients with an A117V mutation and in transgenic mice carrying a triple mutation within the putative TM domain (AV3) (17). Other studies on the putative TM domain indicated that the hydrophobic stretch of amino acids 106 to 126 has the propensity to form fibrils (18) and can induce cell death in cultured cells (19 -22).
In mammalian cells, secretory proteins are usually translocated into the ER via the co-translational pathway, which requires the binding of the signal recognition particle (SRP) to the nascent protein. The SRP directs the targeting of the whole nascent chain-ribosome complex to its receptor (SRPR) on the ER membrane, resulting in the transfer of the growing protein to the translocation pore and a direct release into the ER lumen as soon as the synthesis has finished. An alternative posttranslational import pathway has been described in mammalian cells for proteins of less than 75 amino acids in length; in this case the completely synthesized protein is targeted to the * This work was supported by grants from the Deutsche Forschungsgemeinschaft (TA 167/2 and SFB 596), the Bayerischen Staatsminister fü r Wissenschaft, Forschung und Kunst (for Prion, MPI3) and the Bundesmisterium fü r Bildung und Forschung (01KO0110). 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) X53424 (yeast GAS1), M33556 (yeast KRE5), and M18070 (mouse PRNP).
In this study we analyzed the biogenesis of mouse PrP and showed that the putative transmembrane domain induces the misfolding of PrP during post-translational targeting to the ER. As a consequence, unprocessed and misfolded PrP associates with ER membranes and interferes with cell viability.
Protein Preparation of Yeast Extracts and Endo H Treatment-Yeast cells were grown to midlog phase. 50 OD units were poisoned with 10 mM NaF, and proteins were prepared as described previously (39,40) with the following modifications. Cells were washed in lysis buffer (phosphate-buffered saline containing 0.5% yeast protease inhibitor mixture; Sigma), resuspended in lysis buffer, and broken using glass beads in the Mini-BeadBeater-8 (Biospec Products). A postnuclear supernatant (cell lysate) was prepared by centrifugation at 1000 ϫ g for 5 min at 4°C. To assess the glycosylation status of proteins, the lysate was digested with endoglycosidase H (Endo H, NEB) following the manufacturer's instructions.
Endo H Digestion and Secretion Analysis in N2a Cells-For Endo H digestion, cell lysates were prepared in 0.5% SDS, boiled for 10 min, and digested with Endo H (NEB) for 1 h at 37°C (10). After boiling in Laemmli sample buffer, the lysates were examined by immunoblotting. To analyze the secretion of PrP, cells were cultivated in fresh medium for 3 h at 37°C. PrP present in the supernatant cell culture medium was precipitated by trichloroacetic acid and analyzed by Western blotting.
Western Blotting-Following SDS-PAGE, proteins were transferred onto a nitrocellulose membrane and analyzed as described previously (32).
Colony Assay-Yeast cells were grown to stationary phase. A serial dilution was prepared (5-fold in sterile H 2 O) and spotted onto selective or rich media plates. Growth was monitored after 2 days of incubation at 30°C.
Membrane Fractionation-Localization of recombinant proteins was analyzed by Renografin density gradient centrifugation (39,40). Cells were washed three times in sorbitol buffer (0.8 M sorbitol, 50 mM Tris, 10 mM NaF, pH 7.6) and in TE (50 mM Tris, 1 mM EDTA, pH 7.6). Lysates were prepared in TE as described above, and 0.5 ml were mixed with 0.5 ml of 76% Renografin, successively overlaid with 1 ml at 34, 30, 26, and 22% Renografin and centrifuged at 150,000 ϫ g for 20 h at 4°C. 14 fractions were removed from the top of the gradient. Membranes were diluted 10-fold with TE, pelleted at 100,000 ϫ g for 1 h at 4°C, and resuspended in Laemmli sample buffer.
Cell-free Translation and PK Protection Assay-For in vitro translation the TNT Quick transcription/translation kit (Promega) was used according to the manufacturer's instructions. Briefly, 20 l of TNT, 1 g of plasmid-DNA, and 1 l of [ 35 S]methionine (15 mCi/ml, Amersham Biosciences) were adjusted to 25 l and incubated for 45 min at 30°C. For co-translational import 2 l of canine microsomal membranes (Promega) were included in the reaction. All translation reactions were stopped by the addition of 100 M emetine (Sigma). For post-translational import the membranes were added after stopping the translation. All samples were incubated for a further 45 min at 30°C and were then divided into two aliquots. For the PK protection assay, proteinase K (0.5 mg/ml, Roche Applied Science) was added to one aliquot and incubated for 1 h on ice. The protease was then inactivated by the addition of 5 mM phenylmethylsulfonyl fluoride (Sigma) for 5 min on ice. Samples were boiled in Laemmli sample buffer and separated by SDS-PAGE. Gels were fixed for 30 min, soaked in Amplify (Amersham Biosciences) for another 30 min, and dried. Signals were visualized using a phosphorimaging device (FLA-2000, Fuji; AIDA 3.26 software, Raytest, Staubenhardt, Germany).

PrP Targeted to the ER Is Unprocessed and Interferes with
Yeast Growth-Our aim was to use the yeast model to specifically analyze the import of PrP into the ER. In yeast the hydrophobicity index of the N-terminal signal sequence determines whether a co-or post-translational translocation pathway is used (25). Therefore, we expressed PrP with its own signal peptide (pPrP) and also with the ER-targeting peptide of the endogenous GPI-anchored protein Gas1p (gPrP) (41,42). Both signal sequences have a hydrophobicity index (Kyte-Doolittle) lower than 2, which suggests a post-translational targeting of gPrP and pPrP to the ER (25). gPrP contains the C-terminal Gas1p GPI anchor signal sequence, and further PrP constructs were generated lacking either the GPI anchor attachment signal (gPrP⌬GPI, pPrP⌬GPI) or lacking both the ER targeting and the GPI anchor attachment signals (cyto-PrP) (Fig. 1A). It should be noted that PrP⌬GPI expressed in mammalian cells is completely imported into the ER (10 -14).
Expression of these PrP constructs was analyzed by Western blotting using the monoclonal anti-PrP antibody 3F4 (Fig. 1B). To monitor N-linked glycosylation of PrP, yeast extracts were treated with endo-␤-N-acetylglucosaminidase H, which removes all asparagine-linked sugar moieties from yeast proteins (43). It appeared that neither the Gas1p nor the PrP signal sequence was able to mediate the import of PrP into the yeast ER. All PrP constructs remained unglycosylated. Moreover, the majority of gPrP (ϳ95%), pPrP⌬GPI (100%), and gPrP⌬GPI (ϳ50%) contained uncleaved N-terminal signal sequences. In this context it is important to note that in mammalian cells PrP⌬GPI is mainly unglycosylated, but in contrast to gPrP⌬GPI and pPrP⌬GPI expressed in yeast, it is N-terminally processed. In the same study we showed that PrP containing a nonfunctional C-terminal GPI anchor signal sequence is both N-terminally processed and core-glycosylated (10). Further biochemical analysis revealed that gPrP⌬GPI (Fig. 1D), gPrP, pPrP⌬GPI, and cyto-PrP (data not shown) were highly insoluble in detergent buffer and partially resistant to proteolytic digestion, corroborating earlier findings (44).
In the course of these experiments we noticed a retarded growth of yeast cells expressing PrP with an uncleaved Nterminal signal sequence, be it gPrP, gPrP⌬GPI, or pPrP⌬GPI. To analyze the growth of PrP-expressing cells in more detail, yeast cultures were serially diluted and spotted onto selective agar plates (Fig. 1C, SCD-ura). Both gPrP and gPrP⌬GPI showed a reduced growth compared with the vector control. Interestingly, decreased growth was specifically linked to the expression of PrP targeted to the ER and was also observed for pPrP⌬GPI (data not shown). PrP expressed without a N-terminal signal peptide (cyto-PrP) had no adverse effects on cell growth.
These experiments revealed that expression of mouse PrP with a signal peptide specific for post-translational targeting to the ER generates N-terminally unprocessed misfolded PrP and interferes with yeast growth. Cytosolically expressed PrP lacking the signal peptide is misfolded as well; however, this has no adverse effects on yeast growth.
Deletion of the Putative Transmembrane Domain Restores Processing of PrP and Yeast Growth-Previous studies revealed that the ER import of PrP is modulated by an internal stretch of hydrophobic amino acids (aa112-135). Instead of being synthesized as a secreted protein, this domain can direct the formation of a transmembrane topology (45). Different studies indicate that this stretch of hydrophobic amino acids may have additional properties; short peptides comprising residues 106 -126 formed fibrils in vitro (18) and induced cell death in cultured cells (19 -22).
To address the role of the putative TM domain in PrP folding and ER import, several deletion mutants lacking the complete TM domain and also PrP-AV3 and PrP-G122P were included in our analysis. The triple A to V and the G to P substitutions are located within the putative transmembrane domain ( Fig. 2A).
Although the AV3 mutation enhances the formation of Ctm PrP, the G122P mutation abolishes the formation of any transmembrane topology, be it Ctm PrP or Ntm PrP (17,46). Western blot analysis revealed that all PrP mutants lacking the putative TM domain were glycosylated, indicated by an increased electrophoretic mobility after Endo H digestion (Fig.  2B, Endo H ϩ). In addition, these glycosylated mutants were N-terminally processed. The mutants with amino acid substitutions in the putative TM domain, gPrP-G122P and gPrP-AV3, remained unprocessed and unglycosylated, similar to gPrP (data not shown), and interfered with yeast growth (Fig.  2C). However, cells expressing PrP mutants with a deleted putative TM domain grew at wild-type rates (Fig. 2C).
Our experiments revealed that PrP mutants lacking the putative TM domain were N-terminally processed and glycosylated, although they were targeted to the ER via the Gas1p signal peptide. In addition, these mutants did not interfere with yeast growth.
Unprocessed PrP Shows Prolonged Association with ER Membranes-PrP targeted to the ER via the PrP or Gas1p signal peptide (pPrP, gPrP) had biochemical properties similar to cyto-PrP and also seemed to remain in the cytosol. However, cyto-PrP did not interfere with yeast growth. To analyze the cellular localization of the different PrP mutants in more detail, we performed a Renografin density gradient centrifuga- tion (39,40). This analysis revealed a specific feature of PrP mutants that interfered with yeast growth.
As expected, cytosolic PrP remained in the bottom fractions together with the subunit Gim2p (34) of the cytosolic GimC complex (Fig. 3A). The glycosylated and N-terminally processed PrP mutant with a deleted TM domain (gPrP⌬28 -156) was found in the same fractions as Vhp1p, the 100-kDa subunit of the vacuolar V-ATPase (47). Further analysis revealed that the other N-terminally processed and glycosylated PrP mutants analyzed, gPrP⌬TM and gPrP⌬28 -140, were also exported from the ER and transported to the vacuole (data not shown). In contrast, the N-terminally unprocessed PrP mutant gPrP⌬GPI, as well as gPrP, gPrP-AV3, and gPrP-G122P (data not shown), was membrane-associated and co-localized with the ER and plasma membrane fraction. Because gPrP⌬GPI has no GPI anchor and was not secreted into the culture medium (data not shown), we assumed that it was associated with ER vesicles.
In summary, these data indicate that the PrP mutants characterized by an uncleaved N-terminal signal peptide have a unique feature; they show prolonged association with ER membranes.

In Mammalian Cells PrP⌬GPI Is N-terminally Processed and Transported through the Secretory Pathway-Our analysis
in yeast revealed that PrP⌬GPI expression significantly interfered with cell growth. The question arose whether a similar phenotype could be observed in mammalian cells. PrP⌬GPI was previously characterized by us and other groups, and adverse effects on the growth of mammalian cells have not been described (10 -14). The phenotype of PrP⌬GPI expressed in mouse neuroblastoma (N2a) cells is illustrated in Fig. 3B. Similar to gPrP or to pPrP⌬GPI expressed in yeast, this PrP mutant is mainly unglycosylated. In contrast to the situation in yeast, however, in mammalian cells PrP⌬GPI is N-terminally processed and transported through the secretory pathway ( Fig.  3B and Ref. 10). To test whether the putative transmembrane domain also had an effect on the glycosylation of PrP⌬GPI in mammalian cells we analyzed PrP⌬TM⌬GPI in N2a cells, and core glycosylation was shown to be restored (Fig. 3B).
Thus, in mammalian cells, PrP⌬GPI, even though mainly unglycosylated, is transported through the secretory pathway and does not affect cell growth. These data suggest that the specific phenotype of gPrP⌬GPI in yeast, such as an adverse effect on cell growth and a prolonged association with ER membranes, is linked to the uncleaved signal peptide.
Directing PrP to a Co-translational Import Pathway in Yeast Cells Restores Processing of PrP and Cell Growth-In contrast to mammalian cells, in yeast many proteins destined for the secretory pathway can be imported post-translationally, i.e. independently of the signal recognition particle SRP and its receptor (SRPR). However, the import of a subset of yeast proteins is strictly dependent on SRP/SRPR. It has been shown previously that these proteins contain a hydrophobic signal sequence with a hydrophobicity index (Kyte-Doolittle) higher than 2 (25).
To address the possibility that the effect of PrP expression in yeast is due to a post-translational import pathway, we analyzed the biogenesis of PrP containing a N-terminal ER signal sequence specific for co-translational import. To do so, we replaced the moderately hydrophobic Gas1p ER signal of gPrP and gPrP⌬GPI with the extremely hydrophobic signal sequence of Kre5p (hydrophobicity index Ͼ 3) (48), generating kPrP and kPrP⌬GPI, respectively (Fig. 4A). The Western blot analysis revealed that kPrP was N-glycosylated and that the N-terminal signal sequence of kPrP⌬GPI was cleaved (Fig. 4B). Remarkably, the expression of kPrP⌬GPI did not interfere with yeast growth (Fig. 4C). We then compared the subcellular localization of kPrP⌬GPI with that of gPrP⌬GPI. Indeed, PrP expressed with the Kre5p signal sequence was exported from the ER and found in the vacuole fraction (Fig. 4D). Interestingly, kPrP⌬GPI was highly insoluble in detergent and partially resistant to proteolytic digestion (data not shown), similar to gPrP and cyto-PrP expressed in yeast (Fig. 1) and PrP⌬GPI expressed in N2a cells (Fig. 6B in Ref. 10).
Thus, PrP expressed with a signal sequence specific for cotranslational translocation is imported into the yeast ER. Nterminally processed PrP molecules, whether or not they are glycosylated, have no adverse effects on cell growth.
Post-translational Import of PrP⌬TM in a Mammalian System-Recent in vitro studies showed that PrP is not imported post-translationally into mammalian microsomal vesicles (49). Our experiments described above revealed that in yeast the deletion of the putative TM domain allowed the import and processing of PrP even when it was targeted to the ER via the post-translational pathway by the Gas1p signal (gPrP⌬TM). Consequently, we asked whether the deletion of the putative TM domain would also enable the post-translational import of PrP⌬TM in a mammalian system. To address this question we performed in vitro import experiments in rabbit reticulocyte lysate supplemented with dog pancreas microsomes. To specifically analyze post-translational import the translation was stopped with emetine (50) prior to the addition of microsomal membranes. The import of PrP was monitored by the appearance of core-glycosylated protein (Fig. 5, g) and its protection from exogenously added proteinase K (Fig. 5, PKϩ). Corroborating earlier findings, full-length PrP was exclusively imported co-translationally into microsomes (Fig. 5, PrP, co). PrP with a deleted TM domain, however, was efficiently imported post-translationally (Fig. 5, PrP⌬TM, post). Similarly, a PrP mutant with a larger N-terminal deletion including the TM domain was imported post-translationally (Fig. 5, PrP⌬28 -156, post). In the context of a previous study (51) that described ER targeting of PrP via the C terminus, it is important to note that none of our PrP constructs analyzed for post-translational ER import contained the C-terminal GPI anchor signal sequence.

DISCUSSION
Aberrant conformations of the prion protein are a hallmark of prion diseases. However, the pathogenic mechanisms leading to neuronal cell death still remain enigmatic. By using a yeast model system we were able to characterize the features of misfolded PrP that specifically interfere with cell viability: an uncleaved N-terminal signal sequence and a prolonged association with ER membranes. This unprocessed PrP was generated only in the presence of the putative transmembrane domain and only during post-translational targeting of PrP to the ER.
The Putative Transmembrane Domain Interferes with Posttranslational Import of PrP into the ER-In initial experiments we expressed PrP with its own N-terminal signal peptide and also with the signal peptide of the endogenous yeast GPIanchored protein Gas1p. Under both conditions full-length PrP and PrP⌬GPI accumulated with uncleaved signal peptides and FIG. 3. Unprocessed PrP associates with ER membranes. A, PrP containing an uncleaved N-terminal signal sequence associates with ER membranes and is not transported anterogradely. Subcellular localization of proteins indicated on the left was analyzed by the Renografin density gradient centrifugation of yeast cell lysates. Fractions were removed from the top of the gradient and analyzed by Western blotting; the antibodies are indicated on the right. B, in mammalian cells, PrP⌬GPI is secreted into the cell culture medium. N2a cells were transiently transfected with PrP⌬GPI and PrP⌬TM⌬GPI. After 24 h the medium was changed, and the cells were cultivated for 3 h at 37°C. The medium was collected; cells were washed twice, lysed in detergent buffer, and fractionated by centrifugation. PrP present in the whole lysate (L) or in the cell culture medium (M) was analyzed by Western blotting. To detect N-linked high mannose glycans, cell lysates were incubated with Endo H prior to Western blot analysis (EndoH ϩ).
were not imported into the ER (Figs. 1 and 6). In our first approach to analyze this phenomenon mechanistically, we deleted the internal domains and could show that all PrP mutants lacking the putative transmembrane domain were N-terminally processed, received N-linked glycans, and were imported into the ER (Figs. 2 and 6). A different way to promote import of full-length PrP, or PrP⌬GPI, into the yeast ER was to use the highly hydrophobic signal sequence of Kre5p (hydro-FIG. 4. An extremely hydrophobic signal sequence restores processing and anterograde transport of PrP in yeast. A, the signal sequences of mouse PrP (ss-PrP), yeast Gas1p (ss-Gas1), and Kre5p (ss-Kre5) and their respective hydrophobicities. The hydrophobicities were calculated and plotted using Dnasis 1.2. B, the Kre5p signal sequence restores processing of PrP in yeast cells. PrP was expressed with the Kre5p signal (kPrP, kPrP⌬GPI) and compared with PrP expressed with the N-terminal signal sequence of Gas1p signal (gPrP⌬GPI). Proteins were analyzed as described for Fig. 1B. Note that kPrP receives N-linked glycans and that the N-terminal signal sequence is cleaved off from kPrP and kPrP⌬GPI. C, expression of kPrP⌬GPI does not interfere with yeast growth. A colony assay of yeast cells expressing kPrP⌬GPI or gPrP⌬GPI was performed as described for Fig. 1C. D, anterograde export of kPrP⌬GPI from the yeast ER. Subcellular localization of kPrP⌬GPI and gPrP⌬GPI was analyzed as described for Fig. 3A. As a marker for the vacuolar fraction Vph1p was used (vector). phobicity index Ͼ 3), which is specific for co-translational import (Figs. 4 and 6). Based on these findings it seems most likely that the putative transmembrane domain specifically interferes with the post-translational import of PrP into the ER.
This model was supported by a mammalian in vitro ER import model system. Full-length PrP was not imported posttranslationally into dog pancreas microsomes (Fig. 5), corroborating earlier results (49). Efficient post-translational import was observed, however, for PrP constructs lacking the TM domain (Figs. 5 and 6).
How could the TM domain interfere with post-translational ER import? One possibility is that the TM domain induces a conformation in which the N-terminal signal peptide is buried. Thereby, the signal sequence would be functionally inactivated and post-translational targeting to the ER would be prevented. However, the fact that PrP expressed with a signal peptide was found in association with ER membranes, in contrast to cytosolic PrP, which does not have a signal peptide, points to a different scenario. Interference of the TM domain with posttranslational translocation itself seems more plausible, suggesting that not the targeting to the ER membrane but the translocation process might be impaired. Two models, which are not mutually exclusive, could explain this phenomenon. First, during post-translational targeting the C-terminal domain of PrP might adopt a tight conformation incompatible with a subsequent translocation into the ER, analogous to the impaired post-translational import of dihydrofolate reductase into mitochondria when stabilized by methotrexate (52,53). In   FIG. 6. Biogenesis of PrP mutants in yeast and mammalian cells. The scheme displays the structural features of the PrP mutants analyzed in this study and their degree of post-translational modifications. Please note that a decreased cell growth is linked specifically to PrP mutants, which associate with ER membranes and contain an uncleaved signal peptide. These mutants include the putative TM domain and were generated during post-translational targeting to the ER.

FIG. 5.
Post-translational import into the ER of PrP mutants lacking the putative transmembrane domain. The indicated PrP constructs were translated in vitro using the TNT Quick system. Microsomal membranes (membranes ϩ) were included immediately to analyze co-translational import (co) or after stopping the translation reaction with emetine to study post-translational import (post). Imported core-glycosylated proteins (g) were identified by protease protection assay using 0.1 mg/ml proteinase K for 1 h on ice (PK ϩ). As a control, precursor proteins (p) were translated in the absence of microsomal membranes (membranes Ϫ). Radioactively labeled proteins were separated on SDS-PAGE and visualized by phosphorimaging. this context, it is important to note that after a limited proteolytic digestion the N terminus of misfolded PrP is rapidly degraded, whereas the C terminus adopts a partially PK-resistant core. Alternatively, the TM domain could interact with components present in the cytosol or in the ER membrane. Such an interaction, obviously not present during co-translational import, would directly interfere with the translocation process.
N-terminally Unprocessed PrP Associates with ER Membranes and Interferes with Cell Viability-Under physiological conditions in mammalian cells, PrP is post-translationally modified by the cleavage of the N-terminal signal peptide, the addition of two N-linked glycans, and a C-terminal GPI anchor. When we concentrated on the analysis of different PrP⌬GPI constructs, we observed three different degrees of post-translational modifications. 1) pPrP⌬GPI and gPrP⌬GPI were neither N-terminally processed nor glycosylated. 2) kPrP⌬GPI was also not glycosylated; however, the N-terminal signal peptide was cleaved. 3) gPrP⌬GPI mutants devoid of the putative TM domain (gPrP⌬TM, gPrP⌬28 -140, gPrP⌬28 -156) were both N-terminally processed and glycosylated. In a previous study our group analyzed the folding of PrP in the secretory pathway of mammalian cells and identified different PrP mutants, which were compromised in the post-translational modifications characteristic of PrP C . None of these mutants, however, contained an uncleaved signal peptide (10). In a different study we were also able to show that complex glycosylation of PrP C , a modification that does not occur in yeast cells, is dispensable for correct trafficking through the secretory pathway (54).
The retarded growth of yeast cells was specifically linked to the accumulation of PrP molecules falling in class 1, i.e. containing an uncleaved signal peptide (Fig. 6). Another feature, which was observed exclusively for these PrP constructs, was a prolonged association with ER membranes (Figs. 3 and 4).
How might the prolonged association of gPrP interfere with yeast growth? Previous studies proposed that impaired integrity of cellular membranes, because of a pore forming activity of the putative TM domain, could contribute to the pathological changes in prion diseases (55)(56)(57). In our yeast system, decreased cell growth was dependent on both the N-terminal signal peptide and the TM domain. Thus, it is conceivable that the signal peptide targets misfolded PrP to the ER membrane, where PrP might interact with a component of the ER membrane or the translocon, thereby impairing its function. A straightforward approach to test this hypothesis would be to perform co-immunoprecipitation experiments, but unfortunately the rapid aggregation of PrP interferes with such an approach. Alternatively, the yeast model offers the possibility to perform a suppressor screen to identify cellular components involved in the decreased growth of gPrP-expressing yeast cells.
PrP Misfolding in the Cytosol and Neurodegeneration-Our present study emphasizes the idea that aberrant PrP conformers in the cytosol and/or at the ER membrane have cytotoxic properties. In the mammalian system two PrP species, Ctm PrP and cytosolic PrP aggregates, have previously been shown to induce neurodegeneration independent of PrP Sc . The increased synthesis of Ctm PrP, as a result of mutations in the putative TM domain, is linked to inherited prion diseases in transgenic animals and humans (17,46). Neurotoxicity and neurodegeneration linked to misfolded PrP in the cytosol was observed in cultured cells after treatment with proteasomal inhibitors and in transgenic mice after forced cytosolic PrP expression (15). In both models no infectious PrP Sc was found, similar to our PrP aggregates formed in yeast cells (data not shown). Indeed, it was suggested that the neurotoxic and the transmissible agent in prion disease might be distinct species (46).
Is there a possible relationship between the different toxic PrP species described thus far? In cell culture models Ctm PrP has not been shown to cause cell death, even though a particular PrP mutant was exclusively present as Ctm PrP (58). This might reflect a specific vulnerability of post-mitotic neurons and/or a prolonged pathogenesis in animals/humans. It is interesting, however, that both gPrP expressed in yeast and Ctm PrP contain uncleaved N-terminal signal peptides (49,58).
In the cell culture model described by Ma et al. (15), cytosolic accumulation of PrP and cell death of N2a cells was induced by a proteasomal inhibitor. A recent report revealed that the cytosolic PrP molecules found after such a treatment contain an uncleaved signal peptide (59). Thus, the cytotoxic species in the N2a cell culture model is very similar to gPrP, which interferes with yeast growth.
In contrast to the animal model described (15), we did not see any adverse effects of PrP expressed in the cytosol and lacking a N-terminal signal peptide. This might be explained by the fact that in transgenic mice PrP was expressed in the background of endogenous PrP C . This leaves the possibility that cytosolic PrP mediates its toxic effect by interfering with the import of endogenous PrP C and/or the recruitment of the latter into cytosolic aggregates. Alternatively, the adverse effects of cytosolic PrP might be too moderate to be detected in fast diving cells.
Our study is the first demonstration that PrP misfolding can interfere with the viability of non-mammalian cells. The use of the yeast system could help to define cellular pathways that are sensitive to the accumulation of misfolded PrP and might thereby lead to a better understanding of mammalian prion diseases.