Determinants of the in vivo folding of the prion protein. A bipartite function of helix 1 in folding and aggregation.

Misfolding of the mammalian prion protein (PrP) is implicated in the pathogenesis of prion diseases. We analyzed wild type PrP in comparison with different PrP mutants and identified determinants of the in vivo folding pathway of PrP. The complete N terminus of PrP including the putative transmembrane domain and the first beta-strand could be deleted without interfering with PrP maturation. Helix 1, however, turned out to be a major determinant of PrP folding. Disruption of helix 1 prevented attachment of the glycosylphosphatidylinositol (GPI) anchor and the formation of complex N-linked glycans; instead, a high mannose PrP glycoform was secreted into the cell culture supernatant. In the absence of a C-terminal membrane anchor, however, helix 1 induced the formation of unglycosylated and partially protease-resistant PrP aggregates. Moreover, we could show that the C-terminal GPI anchor signal sequence, independent of its role in GPI anchor attachment, mediates core glycosylation of nascent PrP. Interestingly, conversion of high mannose glycans to complex type glycans only occurred when PrP was membrane-anchored. Our study indicates a bipartite function of helix 1 in the maturation and aggregation of PrP and emphasizes a critical role of a membrane anchor in the formation of complex glycosylated PrP.

ent at the plasma membrane of neuronal and lymphatic cells. During transit through the secretory pathway, the C-terminal domain of PrP C is modified by the attachment of two N-linked complex carbohydrate moieties (Asn 180 and Asn 196 ) (1-4) and a glycosylphosphatidylinositol (GPI) anchor at serine 230 (5).
Structural studies revealed that the mammalian prion protein is composed of a flexibly disordered N terminus and a structured C-terminal domain. This autonomously folding domain contains three ␣-helical regions and a short, two-stranded ␤-sheet (6 -8). Indeed, the C-terminal region seems to confer pathogenicity; after limited proteolysis of PrP Sc , a C-terminal fragment, denoted PrP 27-30 (aa ϳ90 -230), is formed which transmits the disease. Furthermore, studies in transgenic mice indicated that N-terminally truncated PrP C molecules (⌬23-89) support PrP Sc propagation (9,10).
Molecular dynamics simulations of human PrP suggest that the structured part of the protein is stabilized by the N-linked glycans. Moreover, the two partially sialylated oligosaccharide moieties generate a negative electrostatic field that covers the whole surface of helix 2 and helix 3 (11). Regarding the possible physiological or pathological relevance, PrP glycosylation has been shown to have an impact on the conformational transition of PrP C into PrP Sc , to influence the selective neuronal targeting of PrP Sc , and to contribute to the phenomenon of strain diversity (12)(13)(14)(15)(16)(17)(18)(19)(20).
The co-and posttranslational modifications of PrP C are initiated with the cleavage of the N-terminal signal peptide (aa  and the transfer of core glycans while 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, a series of enzymatic reactions in the endoplasmic reticulum (ER) and Golgi compartment. Every asparagine present in the consensus sequence Asn-X-Ser and Asn-X-Thr can serve as an acceptor site for N-linked glycosylation; however, not all such sites present in a polypeptide are actually modified. Similarly, it is unknown why some of the core glycans, like those present in PrP C , are terminally processed, whereas others remain as high mannose structures. So far, it is assumed that the structure of the polypeptide is an important determinant for both modifications.
In this study, we analyzed the in vivo folding and maturation of PrP C and showed that helix 1 as well as membrane attachment play a crucial role in the postranslational modifications of PrP C . express PrP-L108M/V111M for immunostaining with the monoclonal antibody 3F4 (21). Amino acid deletions and substitutions were introduced by PCR cloning techniques. The deletions include the following amino acids: ⌬GPI, aa 230 -254; ⌬TM, aa 113-133; ⌬H1, aa 141-171; ⌬H1*, aa 144 -156; ⌬N, aa 27-89. In PrPmtGPI, the serines at positions 230 -232 have been replaced by threonines. For the generation of PrP-CD4, the CD4 transmembrane sequence and cytoplasmic tail (aa 395-457) were fused to the C terminus of PrP⌬GPI. CD4 sequences were amplified by PCR from pSPOX-CD4PrP (22). All amino acid numbers refer to the mouse PrP sequence (GenBank TM accession number M18070).
Cell Lines, Transfections, Endo H Digestion, ␣-Glucosidase, and ␣-Mannosidase Inhibitor Treatment-N2a and ScN2a cells were cultivated as described (24). SH-SY5Y is a human neuroblastoma cell line (DSMZ number ACC 209). SH-SY5Y cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. Cells were transfected by a liposome-mediated method using LipofectAMINE Plus reagent according to the manufacturer's instructions (Invitrogen). 1-Deoxymannojirimycin (DMJ) or castanospermine (Calbiochem) was dissolved in phosphate-buffered saline (PBS) and added to the cell culture medium (10 g/ml). For endo H digestion, protein lysates were adjusted to 0.5% SDS, boiled for 10 min, and then digested with endoglycosidase H (New England Biolabs) for 1 h at 37°C as specified by the manufacturer.
Detergent Solubility Assay and Proteolysis Experiments-As described previously (25), cells were washed twice with cold PBS, scraped off of the plate, pelleted by centrifugation, and lysed in cold buffer A (0.5% Triton X-100 and 0.5% sodium deoxycholate in PBS). The lysate was centrifuged at 15,000 ϫ g for 20 min at 4°C. After boiling in Laemmli sample buffer, supernatants and pellets were examined by immunoblotting. For proteolysis experiments, cells were lysed as described above and incubated with proteinase K (PK) (Roche Molecular Biochemicals) at the concentrations indicated in Fig. 6 at 4°C for 60 min. The reaction was terminated by the addition of Pefabloc SC (Roche Molecular Biochemicals) and boiling in Laemmli sample buffer. Residual PrP was detected by Western blotting.
Northern Blot Analysis-RNA was extracted and analyzed by Northern blotting as described earlier (26). RNA bound to the membrane was hybridized with a 32 P-labeled DNA probe specific for mouse BiP. As a positive control, mock-transfected cells were incubated with tunicamycin (10 g/ml) (Sigma) for 1 h prior to RNA extraction.
Western Blot Analysis and Indirect Immunofluorescence-Immunoblot analysis and immunofluorescence were performed as described previously (26). A Leica TCSNT/DMIRB confocal system (Heerbrugg, Switzerland) was used for confocal laser-scanning microscopy. Quantification was performed using AIDA 1.0 image analysis software (Raytest).
Metabolic Labeling and Immunoprecipitation-Cells were starved for 30 min in methionine-free modified Eagle's medium (Invitrogen) and subsequently labeled for 30 or 60 min with 300 Ci/ml Pro-mix L-[ 35 S] in vitro cell label mix (Amersham Biosciences; Ͼ37 TBq/mmol) in methionine-free modified Eagle's medium medium. For the chase, the labeling medium was removed, and the cells were washed twice and then incubated in complete medium for the time indicated. Immunoprecipitation of PrP was carried out as described previously (24). Immunoprecipitation products were analyzed by SDS-PAGE. Gels were impregnated with Amplify (PerkinElmer Life Sciences) and exposed to film. Phospholipase C and Trypsin Treatment-Cells were washed twice with ice-cold PBS and then maintained on ice. Phosphatidylinositol phospholipase C (PIPLC) (Roche Molecular Biosciences) in PBS was added to the cells for 2 h at 4°C (0.5 units/ml). The cells were washed extensively with PBS and then lysed with cold buffer A. The distribution of PrP was analyzed by the detergent solubility assay (see above). For PIPLC treatment of extracts, cells were lysed in Triton buffer (0.1% Triton X-100 in PBS), and PIPLC was added to the lysate. After incubation on ice for 1 h, extracts were fractionated by centrifugation and analyzed by Western blotting. Trypsin treatment (0.25%, w/v) of the intact cells was carried out in the cell culture dish on ice. The digest was terminated by the addition of soy bean trypsin inhibitor (Invitrogen). The cells were collected by a brief centrifugation, washed with trypsin inhibitor, and then lysed in cold buffer A. PrP present in the cell culture supernatant was precipitated by trichloroacetic acid and then boiled in Laemmli sample buffer.
Expression and Characterization of Recombinant PrP-Wild type PrP and mutant PrP sequences were amplified by PCR and cloned into the vector pPROEX-HTa (Amersham Biosciences) via the EheI and HindIII restriction sites to add a N-terminal extension of His 6 . Transformed Escherichia coli DH5␣ cells were grown at 37°C to an OD ϭ 0.6, and protein expression was induced for 3 h by isopropyl-1-thio-␤-Dgalactopyranoside (0.5 mM). PrP was purified under denaturing conditions (8 M urea, 50 mM Tris, pH 8) by Ni 2ϩ -nitrilotriacetic acid chromatography. Proteins were eluted (8 M urea, 100 mM sodium acetate, pH 4.6), and fractions containing PrP were pooled. For the formation of the disulfide bonds, proteins were diluted (0.1 mg/ml) and air-oxidized overnight using CuSO 4 (1 M) as a catalyst (27). To concentrate PrP, proteins were adjusted to pH 8 with NaOH, rebound to an Ni 2ϩ -nitrilotriacetic acid column, and eluted (8 M urea, 200 mM imidazole, 100 mM sodium phosphate, pH 7). Refolding was induced by 50-fold dilution into 100 mM sodium phosphate, pH 7. For the limited proteolytic digestion, proteinase K (1 g/ml; Roche Molecular Biochemicals) was added, and the samples were incubated at room temperature for the times indicated in Fig. 5. The reactions were stopped by adding 5 mM phenylmethylsulfonyl fluoride (Roche Molecular Biochemicals) and boiling in Laemmli sample buffer. Samples were analyzed using SDS-PAGE and Coomassie Blue staining.

Helix 1 Directs Folding and
Maturation of PrP C in the Secretory Pathway-To study PrP C folding in the secretory path-way of mammalian cells, we expressed different PrP mutants with deletions proximal to helices 2 and 3 in mouse neuroblastoma (N2a) cells (Fig. 1A). The highly structured helix 2-loophelix 3 motif was left untouched, because it contains the two complex N-linked glycans and the disulfide bridge, modifications known to be important for proper folding (1, 2, 7). Initially, we used the glycosylation status of the PrP mutants to monitor the efficiency of folding. After transient expression in is not significantly expressed on the cell surface. Cells were incubated on ice with trypsin (Trypsin ϩ) to remove cell surface proteins. After cell lysis, residual PrP was detected by Western blotting. As a control, mock-treated cells were analyzed in parallel (Trypsin Ϫ). B and C, PrP⌬H1 is not GPI-anchored. B, live cells were incubated at 4°C with PIPLC to release GPI-anchored proteins from the cell surface or mock-treated (PIPLC Ϫ) (lanes 1-4). In parallel plates (lanes 5 and 6), cells were incubated in fresh medium for 3 h at 37°C. PrP present in the cell lysate (Lysate) or cell culture supernatant (Medium) was detected by Western blotting. C, cell lysates were prepared in Triton buffer (0.1% Triton X-100 in PBS) and incubated on ice with PIPLC or mock-treated (PIPLC ϩ/Ϫ). After centrifugation, PrP present in the detergent-soluble (Sup) or -insoluble (Pellet) fraction was detected by Western blotting.
N2a cells, the PrP mutants were analyzed by endo H digestion and Western blotting (Fig. 1B). Endo H only cleaves high mannose and hybrid structures; consequently, complex type glycans of wild type PrP (wtPrP) could not be liberated by this enzyme (Fig. 1B, wt). Similarly, PrP mutants devoid of the complete N terminus (⌬N, aa 27-89 deleted) or the putative transmembrane domain (⌬TM, aa 113-133 deleted) resisted endo H cleavage, revealing the presence of complex N-linked glycans (Fig. 1B, ⌬N and ⌬TM). However, after endo H digestion, the electrophoretic mobility of PrP⌬H1 (aa 141-171 deleted) and of the double mutant PrP⌬N⌬H1 (aa 27-89 and 141-171 deleted) increased, indicating that mutants lacking helix 1 were not complex-glycosylated (Fig. 1B, ⌬H1 and ⌬N⌬H1). Since the ⌬H1 deletion includes helix 1 (aa 144 -156) together with the second ␤-strand (aa 160 -162) of the prion protein, we also tested a construct in which only helix 1 has been deleted (PrP⌬H1*). PrP⌬H1* expressed in N2a cells showed exactly the same phenotype as PrP⌬H1, indicating that the absence of helix 1 was responsible for the observed effect. Of note, a faster migrating PrP species present in cell extracts from PrP⌬TM-expressing cells was also sensitive to endo H digestion (Fig. 1B, ⌬TM).
To investigate the apparent lack of complex glycosylation of PrP⌬H1 in more detail, we performed metabolic labeling experiments and analyzed PrP by immunoprecipitation. In addition, transfected cells were treated with DMJ, which inhibits ␣-mannosidase I, thereby interfering with terminal processing of N-linked glycans (28). After the pulse, two major PrP species were detected in wtPrP-and PrP⌬TM-expressing cells ( Fig. 2A,  lanes 1 and 3). The faster migrating bands represented PrP containing high mannose N-linked glycans, which could be released by endo H (data not shown). The upper diffuse band was composed of PrP glycoforms that had undergone terminal processing (i.e. contained complex type glycans). Notably, these complex glycosylated PrP species of wtPrP and PrP⌬TM were missing in cells that had been treated with DMJ ( Fig. 2A, lanes 2 and 4). In contrast, PrP⌬H1 was exclusively present as a faster migrating PrP species, identical to PrP⌬H1 generated in the presence of DMJ ( Fig. 2A, lanes 5 and 6). Pulse/chase experiments revealed that the majority of high mannose glycoforms of wtPrP and PrP⌬TM were converted to complex structures within 30 min (Fig. 2B, lanes 2 and 4). In contrast, the N-linked glycans of PrP⌬H1 were not processed into complex structures (Fig. 2B, lane 6).
These initial experiments revealed that deletion of the complete N terminus or of the internal putative transmembrane domain does not interfere with the complex glycosylation of PrP. Deletion of helix 1, however, generates a high mannose glycoform of PrP that is not terminally processed.
PrP⌬H1 Is Not Membrane-anchored and Is Secreted into the Cell Culture Medium-To analyze the cellular trafficking of PrP⌬H1, intact cells were incubated with trypsin to remove all cell surface proteins (Fig. 3A) or with PIPLC to liberate GPIanchored proteins specifically (Fig. 3B). After trypsin treatment, lysates prepared from wtPrP-or PrP⌬TM-expressing cells had lost most of their PrP, indicating that it was localized at the cell surface (Fig. 3A, lanes 2 and 4). Similarly, after PIPLC treatment, wtPrP and PrP⌬TM were found almost quantitatively in the cell culture supernatant (Fig. 3B, wt, ⌬TM, lane 4). In contrast, the relative amount of PrP⌬H1 present in extracts prepared from trypsin-or PIPLC-treated cells was not reduced when compared with the mock-treated control, nor was the cell culture supernatant of PIPLC-treated PrP⌬H1-expressing cells enriched in PrP (Fig. 3, A, lane 6, and  B, ⌬H1, lane 4). Because the trypsin and PIPLC treatments were performed at 4°C, possible secretion of PrP⌬H1 would not have been detected in these assays. Therefore, transfected cells were cultivated in serum-free medium for 3 h at 37°C, and then the cell culture supernatant was analyzed for the presence of PrP. In contrast to wtPrP, PrP⌬H1 was efficiently secreted under these conditions. Interestingly, the faster migrating, endo H-sensitive glycoform of PrP⌬TM was also secreted into the cell culture supernatant at 37°C, indicating once more that the deletion of the putative transmembrane domain generated a heterogeneous PrP population, with the major fraction being folded correctly (Fig. 3B, lanes 5 and 6).
The results of these studies indicated that PrP⌬H1 was not GPI-anchored; instead, it was secreted into the cell culture medium. To demonstrate further that PrP⌬H1 failed to receive a GPI anchor, we performed an in vitro PIPLC digestion. After cell lysis in cold Triton X-100 buffer (0.1% in PBS) and subsequent centrifugation, wtPrP, like other GPI-anchored proteins, fractionated into the detergent-insoluble phase (Fig. 3C, Pellet, lane 1). Incubation of the cell lysates with PIPLC prior to centrifugation rendered the majority of wtPrP and PrP⌬TM Triton X-100 soluble, indicative of their release from the GPI anchor (Fig. 3C, Sup, lanes 2 and 4). PrP⌬H1, however, displayed a different solubility profile. Interestingly, a substantial fraction of PrP⌬H1 was readily solubilized in Triton X-100 buffer at 4°C (Fig. 3C, Sup, lane 5). Moreover, after PIPLC digestion of the extracts, the relative amount of PrP⌬H1 found in the supernatant fraction was not increased (Fig. 3C, Sup, lane 6). Consistent with the results described above (Fig. 3, A  and B), the endo H-sensitive glycoform of PrP⌬TM was also partially soluble in 0.1% Triton X-100 (Fig. 3C, Sup, lane 3).
In the next step, confocal laser-scanning microscopy was performed to corroborate the findings from the biochemical analysis presented above. As expected, wtPrP, PrP⌬TM, and PrP⌬N were located predominantly on the surface of N2a cells. In cells expressing PrP⌬H1 or PrP⌬N⌬H1, however, PrP immunoreactivity was found exclusively intracellularly, following a distinct pattern indicative of the Golgi compartment (Fig. 4,  N2a). Since PrP⌬H1 is efficiently secreted, the intense staining of the post-ER compartment is probably not due to an intracellular retention but rather illustrates the highest local concentration of PrP⌬H1 and PrP⌬N⌬H1 on their way to the cell surface. To exclude the possibility that either the described effects were specific for N2a cells or endogenous wtPrP might influence the processing of transfected PrP mutants, we included SH-SY5Y cells in our analysis, an established neuro-blastoma cell line of human origin that does not express detectable amounts of endogenous PrP (data not shown). Indeed, the analysis of SH-SY5Y cells shown in Fig. 4 corroborated our findings in N2a cells; in SH-SY5Y cells staining of PrP⌬H1 and PrP⌬N⌬H1 was also most intense in the Golgi compartment, whereas wtPrP, PrP⌬TM, and PrP⌬N were again targeted to the plasma membrane (Fig. 4, SH-SY5Y).
Taken together, our data indicated that deletion of helix 1 has a profound effect on posttranslational modifications and cellular trafficking of PrP; GPI anchor attachment is prevented, and a high mannose glycoform of PrP⌬H1 is secreted into the cell culture medium.
A Highly Conserved Hydrophobic Residue in Helix 3 Is Required for Correct Folding of PrP C -Our data presented above revealed that the presence of helix 1 is a prerequisite for correct folding and trafficking of PrP C . We then thought of a way to interfere with the formation of helix 1 without having to delete this domain. Based on the NMR structure of mouse PrP C , a hydrophobic side chain at residue 204 of helix 3 interacts with phenylalanine 140, glutamic acid 145, tyrosine 148, and tyrosine 149 and thereby provides an essential stabilization of helix 1 structure (Fig. 5A). Of note, this residue, either methionine (human, mouse) or isoleucine (hamster), is extremely well conserved between species (29). To disrupt helix 1 packing, we substituted serine or arginine for methionine 204 to create PrP-M204S or PrP-M204R (Fig. 5B). The Western blot analysis presented in Fig. 5C indicated that both PrP-M204S and PrP-M204R did not receive complex glycans and were secreted into the cell culture supernatant, similarly to PrP⌬H1. Support for the assumption that wtPrP and PrP-M204S had different conformations was provided by a biochemical characterization of the recombinant proteins. Wild type PrP and PrP-M204S were expressed in E. coli, purified, and oxidized as described under "Experimental Procedures" (Fig. 5D). After dilution into an aqueous buffer, the proteins were subjected to a limited proteolysis (Fig. 5E). In the case of r-wtPrP, a fragment of about 15 kDa was generated, most likely comprising the C-terminal globular domain. Western blot analysis with an antibody specific for the C terminus of PrP supported this idea (data not shown). In contrast, r-PrP-M204S was digested without a detectable intermediate (Fig. 5E).
Thus, PrP C maturation seems to be dependent on an invariant hydrophobic amino acid located in helix 3, most likely due to a stabilization of helix 1 packing. A limited proteolytic digestion of recombinant wtPrP and PrP-M204S indicated conformational differences between these proteins.
Disruption of Helix 1 Restores Core Glycosylation of PrP⌬GPI-Based on previous studies, deletion of the GPI anchor results in the formation of detergent-insoluble PrP molecules that are not glycosylated (30 -33). Notably, this misfolded PrP⌬GPI molecule is a preferred substrate for the formation of a PK-resistant PrP isoform in an in vitro-based conversion model. The phenotype of full-length PrP lacking the C-terminal GPI anchor signal sequence is illustrated in Fig. 6A; PrP⌬GPI expressed in N2a cells was mainly unglycosylated and secreted into the cell culture medium. Surprisingly, when we deleted helix 1 from PrP⌬GPI, we found that the double mutant was core-glycosylated (Fig. 6A, ⌬H1⌬GPI). Encouraged by our previous results with the amino acid substitution at position 204, we introduced the same mutation into the GPI anchor-deficient PrP construct. Similarly to the deletion of helix 1, the M204S substitution prevented the formation of unglycosylated PrP⌬ GPI; PrP-M204S⌬GPI was present as a high mannose glycoform sensitive to endo H digestion (Fig. 6A, M204S⌬GPI). Support for the assumption that PrP⌬GPI and PrP-M204S⌬GPI had different conformations was provided by a limited proteolytic digestion. Whereas PrP-M204S⌬GPI was PK-sensitive, a significant resistance against PK digestion was observed for PrP⌬GPI (Fig. 6B).
These results revealed that deletion or destabilization of helix 1 restored core glycosylation of PrP⌬GPI in vivo. In contrast to PrP⌬GPI, PrP-M204S⌬GPI and PrP⌬H1⌬GPI were glycosylated and sensitive to a limited proteolytic digestion.
The C-terminal GPI Anchor Signal Sequence Mediates Core Glycosylation of wtPrP-From previous studies, it is known that the location of the glycosylation consensus site with respect to its distance from the C terminus influences the efficiency of N-linked glycosylation (34). In case of the prion protein, both acceptor sites for the core glycans are very close to the C terminus (Asn 180 and Asn 196 ). To analyze the role of the C-terminal GPI anchor signal sequence in the core glycosylation of PrP C , we mutated the cleavage/attachment site ( site) for the GPI anchor. In PrPmtGPI, the serines at positions 230 -232 have been replaced by threonines, which was expected to impair the efficiency of GPI anchoring (35). In this way, the original length of the nascent PrP polypeptide chain (aa 1-254) was preserved without allowing the GPI anchor attachment to occur. Interestingly, PrPmtGPI was glycosylated; however, the N-linked glycans of PrPmtGPI were sensitive to endo H (i.e. were not converted into complex structures (Fig. 6C, mtGPI,  Endo H ϩ)). To further investigate the role of a membrane anchor in the formation of complex glycosylated PrP C , we replaced the C-terminal GPI anchor sequence with the heterologous CD4 transmembrane domain. As expected from previous studies (22,36), PrP-CD4 was complex glycosylated and present at the plasma membrane (Fig. 6D).
By comparing the absence of the GPI anchor signal sequence with a nonfunctional GPI anchor signal sequence, we found that the presence of the GPI anchor signal sequence, whether functional or not, is necessary to allow the attachment of the FIG. 6. Glycosylation of PrP is mediated by the C-terminal signal sequence, PrP folding, and membrane anchoring. A, deletion or destabilization of helix 1 promotes core glycosylation of PrP⌬GPI. N2a cells were transiently transfected with PrP mutants lacking the C-terminal GPI anchor attachment signal in the wild type (⌬GPI) or disrupted helix 1 (⌬H1⌬GPI and M204S⌬GPI) background. Secretion of PrP at 37°C into the cell culture medium (M) and the glycosylation status of PrP present in the cell lysate (L) were analyzed as described in the legend to Fig. 5. B, helix 1 directs the formation of PK-resistant aggregates. N2a cells expressing PrP⌬GPI or PrP-M204⌬GPI were lysed in detergent buffer and subjected to a limited digestion with proteinase K (1:100, 4°C for 60 min). Remaining PrP was analyzed by Western blotting. C, a nonfunctional GPI anchor sequence leads to core glycosylated PrP. A PrP mutant with a mutated -site of the C-terminal signal sequence (mtGPI) was expressed in N2a cells and analyzed as described for A. D, an artificial transmembrane anchor restores maturation of PrP C . N2a cells were transiently transfected with PrP-CD4. The supernatant cell culture medium (M) and the cell lysate (L) were analyzed as described for A. To monitor the glycosylation status of PrP-CD4, cell lysates were digested with endo H (EndoH ϩ). Localization of PrP-CD4 was visualized by indirect immunofluorescence of permeabilized cells using confocal laser-scanning microscopy as described in the legend to Fig. 4. core glycans. However, interfering with the native folding of PrP C by disrupting helix 1 relieves the need for the extra stretch of C-terminal amino acids. For the conversion of high mannose glycans to complex type glycans, membrane anchoring of PrP C is required, be it via the authentic GPI anchor or a heterologous transmembrane anchor.
Misfolded PrP Does Not Induce the Unfolded Protein Response and Can Be Reinternalized by Heterologous Cells-Removal of the three terminal glucose residues from the core glycan is generally regarded as an indication of a successful folding of the nascent chain (37,38). To monitor this modification, we inhibited the ER glucosidases by castanospermine (39) and analyzed different PrP constructs by immunoprecipitation. As expected, formation of complex glycosylated wtPrP or PrP⌬TM was inhibited by castanospermine. Instead, immature PrP species appeared, which were clearly distinguishable from high mannose PrP generated in the presence of DMJ (Fig. 7A,  wtPrP, ⌬TM). The difference in electrophoretic mobility can be attributed to the three additional glucose residues present in each of the two core glycans. Similarly, PrP⌬H1 present in castanospermine-treated cells was clearly larger than in the untreated control, indicating that the initial processing of the core glycan (i.e. the removal of the terminal glucose residues) was not impaired in PrP⌬H1 (Fig. 7A, ⌬H1). Similar results were obtained for PrP-M204S and PrPmtGPI (data not shown).
To detect possible adverse effects of the expression of mutant PrP, we analyzed the unfolded protein response pathway. Induction of the unfolded protein response is indicative of the accumulation of immature or misfolded protein in the ER and leads to the up-regulation of a variety of proteins, such as the ER chaperone BiP (40,41). However, none of the PrP mutants induced increased transcription of BiP in a detectable manner (Fig. 7B). As a positive control, strong induction was found after tunicamycin treatment of N2a cells. To analyze the fate of secreted PrP molecules further, we collected the supernatant cell culture medium, removed any detached cells by filtration, and applied the medium to untransfected N2a cells. 24 h later, the N2a cells were analyzed for the uptake of heterologous PrP, which could be distinguished from endogenous PrP by the presence of the 3F4 epitope. Interestingly, a significant amount of heterologous PrP was present in the lysates of N2a cells which had been incubated with the cell culture medium containing secreted PrP-M204S, PrPmtGPI, or PrP⌬GPI, whereas no 3F4-positive PrP was present in cells incubated with medium derived from cells transfected with wtPrP. It should be noted that heterologous PrP was exclusively present in the detergent-insoluble fraction of the N2a cells (Fig. 7C, P).
In summary, these experiments indicated that the high mannose glycans of the analyzed PrP mutants are initially processed in the ER by glucosidases. Expression of these PrP mutants does not induce the unfolded protein response; the immature PrP glycoforms lacking a GPI anchor are rather secreted and can be reinternalized by heterologous cells. DISCUSSION A conformational transition of PrP C into PrP Sc is implicated in the pathogenesis of prion diseases. Our study provides experimental evidence that deletion of the N-terminal domain of PrP C including the putative transmembrane region and the first ␤ strand have no impact on the formation of complex glycosylated and GPI-anchored PrP C . Helix 1, however, plays a crucial role in the folding pathway of PrP C ; it seems to be essential for GPI anchoring and complex glycosylation. Whereas the GPI anchor signal sequence allows core glycosylation to occur, whether or not attachment of the GPI anchor FIG. 7. The immature PrP molecules undergo initial trimming of the glucose residues and do not induce an unfolded protein response. A, cells transfected with wtPrP, PrP⌬TM, or PrP⌬H1 were cultivated in the presence of castanospermine (CSP) or DMJ (10 g/ ml) and metabolically labeled with [ 35 S]methionine for 60 min. PrP was analyzed by immunoprecipitation and compared with untreated controls (Ϫ). Different glycoforms are indicated. complex: N-linked glycans with complex structure; G 3 Man 9 GNA 2 , core glycans composed of 2 N-acetylglucosamine (GNA), 9 mannose (Man), and 3 terminal glucose (G) residues; Man 9 GNA 2 , core glycans without the terminal glucose residues. B, Northern blot analysis of RNA prepared from untransfected (mock) or PrP-transfected cells using a BiP-specific probe. As a positive control, mock-transfected cells were incubated with tunicamycin (10 g/ml) for 1 h prior to RNA extraction. C, secreted PrP can interact with heterologous cells. Cell culture medium collected from N2a cells transiently transfected with wt-PrP, PrP-M204S, PrPmtGPI, or PrP⌬GPI was filtered and then added to untransfected N2a cells. These cells were lysed 24 h later, and PrP present in the detergent insoluble (P) and soluble fraction (S) was analyzed by immunoblotting with the anti-PrP antibody 3F4. takes place, membrane anchorage of PrP C is necessary for the terminal processing of glycans (Fig. 8).
Core Glycosylation of PrP C Is Dependent on the C-terminal Signal Sequence-The addition of the core glycans occurs while the emerging polypeptide chain is still associated with the translocon. Constraints are imposed by the location of the glycosylation consensus site with respect to its distance from the C terminus (34), and in PrP both glycosylation sites are close to the C terminus (Asn 180 and Asn 196 ). Consequently, PrP⌬GPI lacking the C-terminal amino acids of the GPI anchor signal sequence was glycosylated very inefficiently, indicating that the C-terminal signal sequence in addition to mediating the GPI anchor attachment is also required for efficient core glycosylation. Indeed, the expression of PrPmtGPI with a mutated, nonfunctional GPI anchor signal sequence supported this hypothesis; in this case, core glycosylation was restored, most likely by providing an extra stretch of amino acids distal to the glycosylation sites. Interestingly, we found that the C-terminal amino acids of the GPI anchor signal sequence were only required to ensure glycosylation of wtPrP. Interfering with the formation of helix 1, exemplified by PrP⌬H1 or PrP-M204S, relieved the need for a C-terminal extension to allow core glycosylation to occur; in contrast to PrP⌬GPI, PrP⌬H1⌬GPI and PrP-M204S⌬GPI were efficiently core-glycosylated similarly to PrPmtGPI. Unglycosylated PrP⌬GPI has also been observed in earlier studies (30,31,33). In one study, the authors discussed the possibility that PrP glycosylation and membrane anchorage are cooperative processes (30). However, the fact that PrP⌬H1⌬GPI, PrP-M204S⌬GPI, and PrPmtGPI can be core-glycosylated in the absence of a membrane anchor points to a different scenario. In this context, it is important to note that folding of PrP is one of the most rapid folding reactions measured to date (42). It therefore is conceivable that PrP⌬GPI, due to the C-terminal truncation, is released from the translocon more rapidly than wtPrP or PrPmtGPI. Deletion or destabilization of helix 1 in PrP⌬H1⌬GPI or PrP-M204S⌬GPI, on the other hand, might delay cotranslational protein folding and thereby prolong the association of these mutants with the translocon. Support for the assumption that destabilization of helix 1 might be the cause for the observed differences in folding was obtained by the biochemical characterization of recombinant PrP. The appearance of a proteolytic fragment after limited PK digestion of recombinant wtPrP but not PrP-M204S indicated variant protein conformations.
GPI Anchor Attachment Is Dependent on PrP Folding and Is Required for Complex Glycosylation-A variety of proteins with diverse physiological functions, such as signal transduction, neuronal guidance, or cell-cell interaction, are linked to the plasma membrane through a GPI anchor (43)(44)(45). GPI modification occurs in the ER rapidly after synthesis, with the cleavage of the C-terminal signal peptide followed by the transfer of the GPI anchor from a dolichol carrier to the acceptor amino acid, called the site (46). Like the N-terminal signal peptides, the C-terminal GPI anchor signal sequences do not have conserved sequences but rather characteristic features; only a few amino acids with small side chains are allowed at the , the ϩ 1, and the ϩ 2 positions, and a hydrophobic domain is required about 10 -12 amino acids distal to the site (35,45,(47)(48)(49). It is known that a single amino acid substitution in the GPI anchor signal sequence can significantly interfere with Similarly to wtPrP, mutants lacking the unstructured N terminus (PrP⌬N) or the putative transmembrane domain (PrP⌬TM) were complex-glycosylated and GPI-anchored. The GPI anchor could be replaced by a heterologous transmembrane domain (PrP-CD4) without interfering with postranslational modifications. A second group of mutants was characterized by a disrupted helix 1. In this case, core glycans were attached, and terminal glucose residues were trimmed; however, GPI anchor attachment and conversion of the high mannose glycans into complex structures failed. Interestingly, a nonfunctional C-terminal GPI anchor signal sequence generated a similar PrP glycoform (PrPmtGPI). A third phenotype was observed by deleting the GPI anchor signal sequence from wtPrP. PrP⌬GPI was mainly unglycosylated and partially resistant to proteolytic digestion. Of note, core glycosylation of PrP⌬GPI could be restored by disrupting helix 1 (PrP⌬H1⌬GPI or PrP-M204S⌬GPI).
GPI anchor attachment (50); however, much less is known about the role of the protein conformation in this process.
Our study presents evidence that in addition to the amino acid sequence of the GPI anchor signal sequence, protein folding has a major impact on the efficiency of GPI anchor attachment. Despite the presence of a correct signal sequence, PrP⌬H1 and PrP-M204S did not receive a GPI anchor. To our knowledge, this is the first example to show that protein folding can modulate the transfer of a GPI anchor. The analysis of other PrP mutants supported this hypothesis. Deletion of the complete N terminus, which lacks any known structure, did not interfere with the GPI anchoring of PrP⌬N. Interestingly, the majority of PrP⌬TM, with a deletion of the putative transmembrane domain, was GPI-anchored, suggesting that this domain has only a minor impact on the native folding of PrP. Of note, an absence of the GPI anchor was not only seen for the mutant that lacks helix 1 (PrP⌬H1) but also for PrP-M204S and PrP-M204R, containing only a single amino acid substitution in helix 3. Based on the NMR structure of wtPrP, the most likely explanation for this effect is that helix 1 is destabilized by preventing hydrophobic interactions with helix 3. Although the analysis of recombinant PrP implies conformational differences between wtPrP and PrP-M204S, other mechanisms, such as a direct effect on the GPI anchor transfer reaction, can not be excluded at the moment. A detailed analysis of recombinant PrP mutants with refined biophysical methods, preferentially NMR, will help to answer this question. Impairment of the GPI anchor attachment also had a major impact on further posttranslational modifications, the conversion of the core glycans into complex structures. None of the analyzed PrP mutants without a GPI anchor was complexglycosylated. However, complex glycosylation was not specific for GPI-anchored PrP; PrP-CD4 fixed to the membrane via the heterologous CD4 transmembrane domain was complex-glycosylated as well. The absence of terminal glycosylation could be due either to an altered conformation of the PrP mutants, such as PrP⌬H1 or PrP-M204S, or to the lack of a membrane anchor. In the later scenario, membrane anchoring could provide a specialized environment (i.e. the membrane surface, necessary for native folding of PrP C ). The membrane components might directly promote PrP C folding or indirectly prevent intermolecular PrP interactions of hydrophobic domains. Support for the notion that the membrane anchor might also influence the maturation of wtPrP was provided by PrPmtGPI, which is characterized by a mutated GPI anchor cleavage/attachment site. It can be assumed that this mutant has a structure similar to wtPrP; nevertheless, it was not complex-glycosylated. Interestingly, an influence of the membrane environment on the structure of recombinant PrP was recently shown by two in vitro studies (51,52).
Implications for the Physiological Role of PrP C and the formation of PrP Sc -Incomplete postranslational modifications are often used as a diagnostic marker for proteins that have failed to acquire their native conformation. Such immature proteins are mainly degraded through a mechanism known as ER-associated protein degradation (53)(54)(55)(56). Previous studies showed that the degradation of wild type and mutant PrP via the ER-associated protein degradation pathway is possible (57)(58)(59)(60). However, in our hands, neither unglycosylated PrP⌬GPI nor the high mannose PrP mutants were efficiently degraded via this pathway (data not shown). Instead, the majority of these mutants were transported through the secretory pathway and finally secreted, consistent with earlier reports (30,31,33). This might suggest that the PrP mutants were not sensed as misfolded by the cellular quality control systems. Two observations support such a hypothesis. First, the terminal glu-cose residues were removed from the core glycans, and second, even the overexpression of PrP mutants did not result in the activation of the unfolded protein response. Interestingly, the secreted PrP mutants had a long lifetime and could interact with heterologous cells. Recently, several studies shed light on possible physiological functions of PrP C , including a role in signal transduction pathways, in protein-protein interactions or in copper homeostasis (61)(62)(63)(64)(65)(66). Therefore, it will be important to determine whether immature PrP species are physiologically active or might have a trans-dominant activity. Indeed, an earlier study with transgenic mice indicates a such a dominant activity. Expression of a mutant PrP lacking a part or the complete putative transmembrane domain caused ataxia and neuronal cell death (67). Similar to our studies in cell culture, this mutation did not abolish the formation of complex glycosylated PrP C , but it might well be that the secreted high mannose glycoform of PrP⌬TM we observed was also present in these transgenic mice. It would be interesting to analyze whether expression of PrP⌬H1 or PrPM204S, generating exclusively the secreted high mannose glycoform, had a similar or even aggravated phenotype in transgenic mice.