Mutant prion protein acquires resistance to protease in mouse neuroblastoma cells

Prion diseases (transmissible spongiform encephalopathies) are fatal neurodegenerative diseases that include CreutzfeldtJakob disease, kuru, GerstmannSträusslerScheinker syndrome and fatal familial insomnia in humans, scrapie in sheep and bovine spongiform encephalopathy in cattle (Prusiner et al., 1996 ). Prion diseases can be acquired or inherited or are of idiopathic origin. Inherited forms show an autosomal dominant trait and are linked to point or insertional mutations in the PrP gene (PRNP in humans, Prnp in mice) encoding the prion protein (PrP) (Prusiner & DeArmond, 1994 ). Prion diseases are characterized by the accumulation of a proteinase K (PK)-resistant isoform (PrP^Sc) of the cellular protease-sensitive prion protein (PrP^C).

PrP^C, a neuronal glycoprotein, is anchored to the plasma membrane by a glycosylphosphatidylinositol (GPI) moiety. The function of PrP^C, which is a copper-binding protein (Brown et al., 1997a ), is still not clearly understood. Recent data point to a function in synaptic interaction (Herms et al., 1999 ), resistance of neurons to oxidative stress (Brown et al., 1997b ) or signal transduction (Mouillet-Richard et al., 2000 ).

Studies investigating PrP^C metabolism have demonstrated that PrP^C turns over with a half-life of 36 h in mouse neuroblastoma cells (Borchelt et al., 1990 ; Caughey et al., 1989 ). Chicken PrP^C has been shown to recycle between the plasma membrane and an endosomal compartment (Shyng et al., 1993 ). During its normal processing, PrP^C is cleaved at the central hydrophobic region, yielding a C-terminal, GPI-anchored fragment that can be identified in brain tissue and in cultured cells (Chen et al., 1995 ; Harris et al., 1993 ; Jiménez-Huete et al., 1998 ; Pan et al., 1992 ).

PrP^Sc is derived from PrP^C by a post-translational process that seems to involve species-specific molecular interactions between the two isoforms. PrP^C is necessary for infectivity and pathology, as demonstrated by experiments showing complete resistance of Prnp-knockout mice to infection with prions (Brandner et al., 1996 ; Brown et al., 1996 ; Büeler et al., 1993 ). PrP^C is rich in α-helices, while PrP^Sc exhibits a greater content of β-sheets (Caughey et al., 1991 ; Pan et al., 1993 ; Safar et al., 1993 ). The nuclear magnetic resonance structure of full-length recombinant PrP^C revealed that the region comprising amino acids 121231 contains a high level of secondary structure, including three α-helices and a two-stranded antiparallel β-sheet, whereas the N terminus is highly flexible (Hornemann et al., 1997 ; Riek et al., 1996 , 1997 ). Copper binding seems to confer a distinct structure on the N terminus (Viles et al., 1999 ). Synthesized peptides that are homologous to residues 109122 and 106126 form β-sheets spontaneously (Forloni et al., 1993 ; Gasset et al., 1992 ), with the internal sequence AGAAAAGA (residues 112119) displaying the highest tendency to form amyloid (Gasset et al., 1992 ). Peptides containing the AGAAAAGA motif are toxic to neurons in culture, whereby the sequence AGAAAAGA was found to be necessary but not sufficient for the neurotoxic effect (Brown, 2000a ; Forloni et al., 1993 ). A recent study using in vitro conversion assays of PrP^C to PK-resistant PrP^Sc indicates that synthetic peptides derived from the central part of PrP^C (amino acids 106141) have an inhibitory effect on this conversion (Chabry et al., 1998 ). The presence of residues 119 and 120 (the two last residues within the motif AGAAAAGA) seems to be crucial for this inhibitory effect, arguing for their involvement in the intermolecular interaction that leads to PK-resistant PrP. Mutant PrP molecules carrying deletions of amino acids 108121 or 114121 are not convertible to PrP^Sc when overexpressed in scrapie-infected neuroblastoma cells (Hölscher et al., 1998 ; Muramoto et al., 1996 ). The central hydrophobic region, spanning all or most of the sequence AGAAAAGA, therefore plays an important role in the conversion process.

For further insight into the function of this region, we decided to establish a cell culture model. To develop such a system, we expressed PrP molecules carrying mutations at and around this hydrophobic sequence in murine neuroblastoma cells and analysed the mutant proteins for proteolytic processing and resistance to protease. We report here that one of these mutant PrP molecules acquired protease resistance in comparison with wild-type PrP.

Reagents and antibodies.
DNA-modifying enzymes were purchased from Promega and dideoxy-sequencing reagents from Amersham. Cell-culture reagents were obtained from Biochrom KG. N-Glycosidase F (PNGase F) was purchased from New England BioLabs. All other reagents were from Sigma. The following PrP-specific antibodies were used: mouse mAb 3B5, directed against amino acids 6884 of human PrP (Krasemann et al., 1996 ), mAb 6H4, directed against amino acids 155163 of bovine PrP (Korth et al., 1997 ), mAb 13A5, directed against an epitope of Syrian hamster PrP (Rogers et al., 1991 ), and polyclonal antibody Kan72, directed against amino acids 89103 of murine PrP (moPrP) (Hölscher et al., 1998 ). Alkaline phosphatase (AP)-conjugated goat anti-mouse antibody was from Dianova and the AP-conjugated goat anti-rabbit antibody from Dako. TRITC-coupled goat anti-mouse antibody was purchased by Dianova.

Brain tissues and N2a cells.
Tissues were obtained from brain hemispheres of tg81 mice (Scott et al., 1989 ), hamster and hamster infected with the scrapie strain 263K. The murine neuroblastoma cell line N(euro-)2a (Klebe & Ruddle, 1969 ) was obtained from the ATCC. N2a cells were grown in DMEM containing 10% foetal calf serum and penicillin/streptomycin in an atmosphere of 10% CO₂.

Mutant Prnp genes.
The generation of the plasmid pCI-Prnp has been described previously (Windl et al., 1999 ). The hamster-specific 13A5 epitope was inserted into the wild-type Prnp^a allele via overlap-extension (OE)-PCR (Higuchi et al., 1988 ) by using the outer primers T3 (5' AATTAACCCTCACTAAAGGG 3') and T7neo (5' TAATACGACTCACTATAGG 3') and the overlapping primers 13A5up (5' CCAAAATGCATCATGGGCC 3') and 13A5do (5' CCCATGATGCATTTTGGCA 3'). After cleavage with XbaI and XhoI, the PCR product was cloned into the corresponding restriction sites of pCI-neo (Promega) to give pCI-Prnp13A5. Based on pCI-Prnp13A5, plasmid pCI-PrnpΔSmaI, containing a Prnp gene without its original SmaI site, was established. The following primers were used in an OE-PCR: ΔSmaIup (5' CCCTGCCCAGGATACCGGC 3') and ΔSmaIdo (5' CGGTATCCTGGGCAGGGAAG 3').

Plasmid pCI-PrnpΔSmaI was the starting point for pCI-PrnpΔ105125, pCI-PrnpA112119 and pCI-PrnpG112119. Plasmid pCI-PrnpΔ105125 was generated via OE-PCR by using the following overlapping primers: Δ105125up (5' CCAGCATGTACCCGGGTTTGCTGGGCTTGT 3') and Δ105125do (5' CAGCAAACCCGGGTACATGCTGGGGAGCG 3'). In addition to the deletion (codons 105125), this construct carried a new SmaI restriction site by changing codons 104 and 126 silently from CCA to CCC and from GGC to GGG. Constructs pCI-PrnpA112119 and pCI-PrnpG112119 were generated based on pCI-PrnpΔ105125. The primers A112119up (5'ACCAAGGCCCCCCACTACTGCCGCAGCTGCCG CAGCCGCTGCCACATGCTTGAGGTTGGTTTT3') and A112119do (5' AAAACCAACCTCAAGCATGTGGCAGCGGCTGCGGCA GCTGCGGCAGTAGTGGGGGGCCTTGGT3') or G112119up (5' ACCAAGGCCCCCCACTACTCCCCCACCTCCCCCACC CCCTCCCACATGCTTGAGGTTGGTTTT3') and G112119do (5' AAAACCAACCTCAAGCATGTGGGAGGGGGTGGGGGAGGTGGGGGAGTAGTGGGGGGCCT-TGGT 3') were hybridized with each other and the resulting fragments were cloned into the SmaI site of pCI-PrnpΔ105125 to give pCI-PrnpA112119 and pCI-PrnpG112119, respectively. The sequences of all modified Prnp genes were verified by DNA sequence analysis.

The plasmid pHA58 carries the hygromycin-resistance gene under the control of the murine pgk-1 promoter. The construction of the vector phCMV*-Prnp has been described previously (Windl et al., 1999 ). This plasmid was originally derived from the plasmid pUHD 10-3 and carries the wild-type Prnp^a allele under the control of the transactivator-responsive promoter (PhCMV*-1). The plasmids pCI-Prnp13A5 and pCI-PrnpG112119 were cleaved with XhoI and the plasmid pUHD 10-3 was cleaved with EcoRI. After blunting the 5' protruding ends of the three constructs with the Klenow fragment of DNA polymerase, these constructs were cleaved with XbaI. The inserts excised from pCI-Prnp13A5 and pCI-PrnpG112119 were cloned between the Klenow-treated EcoRI site and the XbaI site of pUHD 10-3 to produce phCMV*-Prnp 13A5 and phCMV*-PrnpG112119.

Construction of recombinant cell lines.
N2a cells were stably transfected with the different pCI-Prnp constructs using Effectene or Superfect (both from Qiagen) as transfection reagents according to the manufacturers instructions. After transfection, antibiotic-resistant cells were selected in 400 µg/ml geniticin (G418) and polyclonal cell lines were obtained after 4 weeks. The level of PrP overexpression in the different cell lines was assayed by Western blot analysis (see below).

The use of the tetracycline-inducible expression system required stably double-transfected N2a cell lines carrying the reverse transactivator as a first component and wild-type and G112119 moPrP genes under the control of the transactivator-responsive promoter as a second component. N2a clones expressing the first component, the reverse tetracycline-controlled transactivator (rtTA), have been established previously (Windl et al., 1999 ). One of the resulting N2artTA clones (clone 1) was subjected to a second round of transfection, a co-transfection with either phCMV*-Prnp13A5 or phCMV*-PrnpG112119 combined with pHA58 (Prnp plasmid and hygromycin-resistance plasmid in a 19:1 mixture) using Effectene as the transfection reagent. Cells were selected in the presence of hygromycin (200 µg/ml) and geniticin (200 µg/ml).

Immunocytochemistry.
Cells were grown in 24-well plates and were fixed with 10% paraformaldehyde in PBS for 10 min or with methanol for 30 min at 4 °C. After rinsing with PBS, the cells were incubated with mouse mAb 3B5 at a dilution of 1:5 for 1 h at room temperature. Cells were rinsed and then incubated with TRITC-conjugated secondary antibody at a dilution of 1:200 for 30 min at room temperature. After rinsing, the cells were embedded in PBS and examined with a fluorescence microscope (Olympus).

Western blot analysis.
Cells grown in 75 cm² tissue culture flasks were lysed in 500 µl ice-cold lysis buffer containing 50 mM TrisHCl, pH 7·5, 150 mM sodium chloride, 0·5% NP-40 and 0·5% sodium deoxycholate supplemented with 1 mM PMSF. Brain tissue was homogenized in 9 vols ice-cold lysis buffer. Where indicated, lysates or brain homogenates were treated with PNGase F according to the manufacturers instructions. Proteins were separated by SDSPAGE (12·5 or 15% acrylamide), transferred to a PVDF membrane and processed further as described elsewhere (LeGendre, 1990 ). For detection of PrP and its fragments, the membrane was incubated with one of various primary antibodies at a dilution of either 1:2000 (13A5, 6H4 and Kan72) or 1:50 (3B5). This was followed by incubation with AP-conjugated goat anti-mouse or goat anti-rabbit antibody (diluted 1:1000). Immunopositive signals were detected by using the chemiluminescent substrate CDP-Star (Tropix) according to the manufacturers instructions. The light signals were monitored, documented and quantified with a CCD camera system (raytest) and the accompanying software (DIANA, TINA; raytest).

Assay of protease resistance.
Cell lysates were treated with PK at different concentrations (3·3, 10, 20 or 50 µg/ml) for 10 min at 37 °C or they were digested with 3·3 µg/ml PK at 37 °C for 0, 10, 20, 40 or 60 min, as described previously (Lehmann & Harris, 1996 ). Proteolytic digestions were terminated by adding 2 mM PMSF and analysed directly by Western blot. The protease resistance of PrP in brain tissue from scrapie-infected hamster was analysed by digesting the homogenized brain with PK at a concentration of 20 µg/ml for 1 h at 37 °C. Digestion was terminated by the addition of 2 mM PMSF. Proteins in the brain homogenate were methanol-precipitated, denatured in 7 M guanidine hydrochloride and methanol-precipitated again. The resulting pellet was resuspended in lysis buffer.

Overexpression of wild-type and mutant moPrPs in N2a cells
The wild-type moPrP contains the highly amyloidogenic sequence AGAAAAGA (positions 112119) in the central region comprising residues 105125 (Fig. 1). We generated several mutations in this sequence motif of PrP to study the fate of these mutant prion proteins at the cellular level. Two mutant moPrPs carried exclusively either glycine or alanine residues instead of the AGAAAAGA sequence (designated G112119 or A112119). A third mutant form of moPrP (designated Δ105125) had a deletion in the central region including AGAAAAGA. Wild-type and mutant moPrPs were epitopically tagged by substituting a methionine residue for an isoleucine at position 138 (Fig. 1). The introduction of this epitope into moPrP allowed the detection of the protein by the mAb 13A5 (Rogers et al., 1991 ). The wild-type and mutant moPrP genes were positioned under the control of the strong hCMV promoter and stably transfected into N2a cells.

(22K): Fig. 1. Scheme of wild-type and mutant moPrPs. The coding regions of the normal (moPrP) and mutant Prnp genes [WT (13A5-tagged), A112119, G112119 and Δ105125] are depicted by bars. Numbers indicate amino acid positions, CHO indicates N-glycosylation sites and GPI indicates the attachment site of the glycolipid anchor. The signal sequence and the octapeptide repeat region are highlighted by shaded bars. The replacement of an isoleucine (I) by a methionine (M) at position 138 creates an epitope for the hamster-PrP-specific mAb 13A5 (Rogers et al., 1991 ). The hatched box depicts the highly amyloidogenic region AGAAAAGA, and the amino acid changes in A112119 and G112119 are recorded.

In order to determine whether the mutant PrP molecules were expressed and delivered to the cell surface, we performed immunofluorescence microscopy with stably transfected polyclonal lines of N2a cells. Only overexpressed exogenous PrP (Fig. 2ad), and not endogenous PrP of untransfected N2a cells (Fig. 2e), was detectable. Overexpressed wild-type, Δ105125, G112119 and A112119 moPrPs were all detected on the surface of intact cells. The staining pattern around the cell periphery was similar before and after permeabilization (data not shown).

(36K): Fig. 2. Overexpression of wild-type, Δ105125, G112119 and A112119 moPrPs. (a)(e) N2a cell lines overexpressing wild-type (WT) (a), A112119 (b), G112119 (c) and Δ105125 moPrP (d) as well as normal N2a cells (e) were labelled with mouse mAb 3B5 and stained with a fluorescein-coupled secondary antibody. Cells were observed by immunofluorescence microscopy. The exposure time was chosen to detect only overexpressed exogenous PrP. (f) Lysates of stably transfected N2a cell lines overexpressing wild-type (lane 1) and mutant (lanes 24) moPrPs were analysed by immunoblotting with mouse mAb 3B5. Equal amounts (100 µg) of protein were loaded in all lanes of the polyacrylamide gel. Positions of marker proteins are indicated.

The size and glycosylation pattern of the mutant proteins were examined by Western blot (Fig. 2f). All four recombinant PrP molecules migrated as a series of three bands, which represent the un-, mono- and diglycosylated forms of the proteins. Wild-type, G112119 and A112119 moPrPs (Fig. 2f, lanes 13) were the same size as PrP^C from other sources, whereas the Δ105125 protein bands were shifted to a lower molecular mass (Fig. 2f, lane 4; also see Fig. 3). The levels of expression of A112119, G112119, Δ105125 and wild-type moPrPs in stably transfected N2a cell lines were comparable, but were much higher than the expression of endogenous moPrP in normal N2a cells (see Fig. 3b).

(34K): Fig. 3. Characterization of proteolytic fragments of wild-type (WT), Δ105125, G112119 and A112119 moPrPs. (a)(b) Brain homogenates of normal hamster and tg81 mice (a, lanes 1 and 2) and lysates of N2a cell lines overexpressing recombinant moPrP (a, lanes 36; b, lanes 14) as well as normal N2a cells (a, lane 7; b, lane 5) were treated with PNGase F for 3 h. Equal amounts (100 µg) of deglycosylated proteins were subjected to immunoblotting with the monoclonal antibodies 13A5 (a) and 6H4 (b). The positions of full-length PrP and the C-terminal fragment (C1) are indicated to the left and positions of marker proteins are indicated to the right. (c) Quantification of the immunopositive signals of the C-terminal fragment (C1) compared with the amount of full-length PrP in the different mutants following staining with antibody 6H4. The ratio between the C1 fragment and full-length PrP was determined for each lane and means were calculated. Bars indicate standard deviations based on three experiments.

Examination of PrP-specific fragments following deglycosylation
The sugar moieties of wild-type PrP^C are N-linked and can be cleaved from their respective asparagines by PNGase F, giving rise to a protein of 27 kDa (Haraguchi et al., 1989 ). In order to study the PrP-specific fragments of the mutant moPrPs following deglycosylation, we performed Western blot analysis with different antibodies. Mouse mAb 13A5 immunoreacted with the 13A5-tagged wild-type, Δ105125, G112119 and A112119 moPrPs (Fig. 3a, lanes 36) as well as with hamster PrP in normal hamster or tg81 mice (Fig. 3a, lanes 1 and 2), but could not detect untagged endogenous PrP of normal N2a cells (Fig. 3a, lane 7). mAb 6H4, which reacted with normal moPrP, also failed to detect endogenous PrP (Fig. 3b, lane 5) due to the low level of expression.

Treatment of cell lysates with PNGase F reduced the heterogeneity of the PrP-specific bands in 13A5-tagged wild-type, G112119 and A112119 moPrPs to a fragment of 27 kDa [indicated by PrP in Fig. 3a (lanes 35) and b (lanes 13)], which is consistent with the molecular mass of full-length deglycosylated PrP containing the GPI anchor. Δ105125 moPrP showed the expected difference in molecular mass of about 2·3 kDa in comparison with wild-type moPrP (Fig. 3a, lane 6; b, lane 4). A prominent band of 18 kDa (designated C1) was recognized by mAbs 13A5 and 6H4 in blots of wild-type and A112119 moPrPs [Fig. 3a (13A5), lanes 3 and 4; b (6H4), lanes 1 and 2]. It corresponded in size with the major C-terminal fragment of PrP^C in hamster and tg81 mice (Fig. 3a, lanes 1 and 2). The PrP-specific fragment of 18 kDa was not detectable or was only weakly detectable with both antibodies in blots of Δ105125 and G112119 moPrPs [Fig. 3a (lanes 5 and 6), b (lanes 3 and 4) and c]. This finding was highly significant and reproducible. These experiments showed that, while all mutants carried N-linked glycosylation, the amount of the potentially proteolytic fragment C1 was dependent on the mutation introduced.

G112119 moPrP is resistant to protease
In order to test the mutant moPrPs for protease resistance, the lysates were digested for 10 min with 3·3 µg/ml PK and treated with PNGase F to simplify the evaluation. Wild-type, A112119 and endogenous moPrPs, as well as PrP of tg81 mice, were completely degraded under these conditions (Fig. 4a, lanes 2, 8, 10 and 12), but the mutant G112119 moPrP gave a clear protease-resistant signal that migrated in the 22 kDa range (Fig. 4a, lane 6).

(63K): Fig. 4. Protease resistance of mutant moPrPs. (a) Protein lysates of N2a cell lines overexpressing recombinant moPrP (lanes 18), lysate from normal N2a cells (lanes 9 and 10) or lysate from brain homogenate of tg81 mice (lanes 11 and 12) were treated with PK (3·3 µg/ml) at 37 °C for 10 min (+) or left undigested (-). All lysates were incubated with PNGase F for 3 h and equal amounts of protein (100 µg) were analysed by Western blot using the rabbit polyclonal antiserum Kan72. (b) Equal amounts (100 µg) of lysates of N2a cells overexpressing wild-type moPrP (lanes 15) or G112119 moPrP (lanes 610) were digested at 37 °C for 0, 10, 20, 40 or 60 min with PK (3·3 µg/ml). An equivalent amount of brain homogenate from hamster scrapie was left undigested (lane 11) or treated with PK (20 µg/ml) at 37 °C for 60 min (lane 12). Positions of marker proteins are indicated.

To assess the extent of protease-resistance of G112119, we subjected lysates of G112119 and wild-type moPrPs to digestion with 3·3 µg/ml PK for 10, 20, 40 and 60 min at 37 °C. In contrast to the complete degradation of wild-type moPrP (Fig. 4b, lanes 25), significant amounts of immunoreactive G112119 moPrP remained after 10 and 20 min (Fig. 4b, lanes 7 and 8). Comparison with the protease-resistant core of hamster PrP^Sc (Fig. 4b, lane 12) showed that the fragments of both hamster PrP^Sc and G112119 moPrP remaining after proteinase digestion migrated at a similar size range and faster than respective untreated samples (Fig. 4b; compare lanes 6 and 7 as well as lanes 11 and 12).

Doxycycline-induced expression of wild-type and G112119 moPrPs
The protease resistance of G112119 moPrP was examined further with the reverse tetracycline-inducible expression system. The two stably transfected cell lines rtTAwild-type and rtTAG112119 displayed inducibility (Fig. 5a; compare lane 2 with lane 3 and lane 4 with lane 5) when tested for background expression before induction and overexpression after induction with doxycycline. The level of expression increased about 2- to 3-fold after induction with 2 µg/ml doxycycline in the medium for 20 h. Comparison of the amount of wild-type or G112119 moPrP in the uninduced state with the level of endogenous PrP in untransfected N2a cells shows a background of wild-type and G112119 moPrPs (Fig. 5a, lanes 2 and 4).

(37K): Fig. 5. Doxycycline-induced overexpression of wild-type and G112119 moPrPs. (a) Stably double-transfected N2a cell lines were tested before (-) and after (+) induction with doxycycline (2 µg/ml medium) for 20 h. Expression of wild-type (lanes 2 and 3) and G112119 (lanes 4 and 5) moPrPs was analysed by Western blot with the mAb 3B5. A cell lysate from normal N2a cells was added as a control (lane 1). Equal amounts (100 µg) of protein were loaded in all lanes. (b) Protein lysates of stably double-transfected N2a cell lines expressing wild-type (lanes 710) or G112119 (lanes 14) and protein lysates of normal N2a cells (endogenous; lanes 5 and 6) were treated with PK (3·3 µg/ml) at 37 °C for 10 min or left untreated as indicated before (lanes 1, 2, 7 and 8) or after (lanes 3, 4, 9 and 10) induction with doxycycline (Dox; 2 µg/ml medium) for 20 h. Equal amounts (100 µg) of protein were then subjected to immunoblotting using Kan72 antiserum. (c) Proteins in lysates of cells expressing wild-type (lanes 68) or G112119 (lanes 15) moPrPs after induction with doxycycline (2 µg/ml medium) were treated with PK at different concentrations (3·3, 10, 20, 50 µg/ml) at 37 °C for 10 min (lanes 25, 7 and 8) or left undigested (lanes 1 and 6). Equal amounts (100 µg) of protein were analysed by Western blot and the rabbit polyclonal antiserum Kan72. Positions of marker proteins are indicated.

In order to examine whether G112119 became protease-resistant in the tetracycline-inducible system, we performed a digestion with 3·3 µg/ml PK for 10 min with lysates of cells before and after induction for 20 h with 2 µg/ml doxycycline. In contrast with the complete degradation of wild-type (Fig. 5b, lanes 8 and 10) and endogenous (Fig. 5b, lane 6) moPrP, we found protease-resistant G112119 moPrP in the induced as well as in the uninduced state (Fig. 5b, lanes 2 and 4). The amount of protease-resistant G112119 moPrP correlated with the amount of total G112119 moPrP before induction (Fig. 5b, lanes 1 and 2) as well as after induction (Fig. 5b, lanes 3 and 4) and was about 50% of the total G112119 moPrP in both expression states. The clear increase in the amount of protease-resistant PrP was therefore acquired within 20 h after induction. In the induced state, G112119 moPrP showed a clear shift of the diglycosylated, monoglycosylated and unglycosylated bands after digestion with PK (Fig. 5b, lane 4) compared with the three full-length bands before digestion (Fig. 5b, lane 3). Treatment of G112119 moPrP in the induced state with 3·3 µg/ml PK for different lengths of time yielded protease-resistant G112119 moPrP after digestion for 10 and 20 min (data not shown), which corresponded to our results with the permanently expressed mutant moPrPs under the control of the hCMV promoter.

The high level of expression of wild-type and G112119 moPrPs in the induced state gave us reason to test the extent of resistance to PK at higher concentrations. After induction for 20 h with 2 µg/ml doxycycline, we treated lysates of the N2a cell lines rtTAG112119 and rtTAwild-type with PK at concentrations of 3·3, 10, 20 and 50 µg/ml for 10 min (Fig. 5c). Wild-type moPrP was degraded completely (Fig. 5c, lanes 7 and 8), whereas immunoreactive G112119 moPrP still remained after digestion with 10 µg/ml PK (Fig. 5c, lane 3). The amount was still about 12% of full-length G112119 moPrP (Fig. 5c, compare lane 3 with lane 1).

By employing both a permanent and an inducible expression system, we could show that the acquisition of protease resistance is an intrinsic property of G112119 moPrP in N2a cells.

Three different mutants of PrP were overexpressed in N2a cells and analysed alongside normal N2a cells and N2a cells overexpressing wild-type PrP. The three mutations were positioned in the central, evolutionarily highly conserved domain of the protein in order to reveal any consequences of changing or deleting the highly amyloidogenic palindrome AGAAAAGA (residues 112119 in moPrP) (Gasset et al., 1992 ). By immunofluorescence analysis, we showed that all mutant prion proteins were transported to a large extent to the cell surface, as was the wild-type protein. Most mutant PrPs that have so far been stably overexpressed in cells reach the cell surface, with the notable exception of proteins mutated at the first N-glycosylation site (Lehmann & Harris, 1997 ). In all mutant proteins presented here, both glycosylation sites were used. Therefore, we did not find any indication of a defect in the post-translational transport of the mutant PrPs to the cell surface, whereas a few pathogenic human mutations, such as P102L, D178N and F197, as well as an insertional mutation with nine additional octapeptides, were shown to have an effect on the delivery of the molecules to the cell surface (Ivanova et al., 2001 ).

A certain amount of normal PrP^C is cleaved physiologically, which is identifiable by the presence of a C-terminal fragment (Chen et al., 1995 ; Harris et al., 1993 ; Jiménez-Huete et al., 1998 ; Pan et al., 1992 ). It is the amount of this C-terminal fragment relative to the full-length PrP that is altered substantially in two of the mutant PrPs presented here. It is easily conceivable why this might be a consequence of the extensive deletion in mutant Δ105125 moPrP, in which all residues N- and C-terminal to the putative cleavage site (residue 110 or 111) have been removed (Chen et al., 1995 ). The strongly inhibitory effect of the mutant G112119 on the efficacy of cellular cleavage is probably determined directly by a mutation-induced alteration in protein conformation, but there are several possible explanations for how proteolytic processing is influenced by the altered structure of G112119 moPrP: (i) the mutated residues might themselves be a poor substrate for the putative cleaving enzyme; (ii) the structure of the whole protein might be changed and the cleavage site might be less accessible; or (iii) due to the mutation introduced, the protein might be distributed on the cell surface in a different manner and therefore be in infrequent contact with the processing enzyme.

Analysis of the protease resistance of the various PrP mutants showed that G112119 moPrP displayed resistance against PK digestion at a concentration of 3·3 µg/ml for 10 min at 37 °C. These conditions are much milder than those used in the detection of PrP^Sc in brain homogenates of diseased animals and humans (e.g. 20 µg/ml for 30 min at 37 °C; Büeler et al., 1994 ), but were used by Lehmann & Harris (1997) in the analysis of Prnp genes carrying human-pathogenic mutations. In the latter study, the authors found that, under such conditions, the introduction of the human-pathogenic mutations P102L, D178N, T183A and E200K and an insertion of six additional octapeptides in the ORF of Prnp at the respective locations (P101L, D177N, T183A and E199K) rendered the mutated proteins resistant to PK at the above concentration. The PK resistance was always accompanied by enhanced detergent-insolubility of the mutated proteins (Daude et al., 1997 ; Lehmann & Harris, 1997 ). Similar PK resistance was reported for pathogenic mutations by other groups (Priola & Chesebro, 1998 ; Singh et al., 1997 ) but, by and large, the low concentration of PK, the dependence of these results on the cell type used (Petersen et al., 1996 ; Priola & Chesebro, 1998 ) and the lack of demonstrable infectivity resulting from this type of protease resistance leave it open whether the generated protease-resistant PrP is low-concentration PrP^Sc or unrelated to infectivity. We could show that the extent of protease resistance of the mutant G112119 moPrP was slightly higher than under the standard conditions of 3·3 µg/ml for 10 min at 37 °C, but still much lower than the resistance of PrP^Sc in infected animals or in infected tissue culture (Hölscher et al., 1998 ). Yet, protease resistance is an intrinsic property of G112119 moPrP in N2a cells and was not a chance event in a particular N2a cell clone, as we confirmed the protease resistance of this particular mutant by expressing it in a different expression system, the doxycycline-inducible expression system. So far, this system has only been used to express wild-type PrP genes in transgenic animals and N2a cells (Tremblay et al., 1998 ; Windl et al., 1999 ). The mutant G112119 moPrP showed a concurrent increase in the amount of protease-resistant PrP after induction of expression by doxycycline, indicating that the acquisition of protease resistance occurs soon after the biosynthesis of this mutant protein.

These findings have two general meanings for the conversion of certain mutant PrP molecules to protease-resistant isoforms. (i) The acquisition of a mild form of protease resistance of mutant prion proteins in tissue culture is not restricted to mutations known to cause inherited prion diseases in humans. This might point to the fact that there are many more (and probably more devastating) pathogenic mutations possible in PrP that might not exist in living animals because they interfere with the evolutionary success of the respective carrier. A recent study using peptides randomized in this hydrophobic domain supports this argument, as the author showed that these peptides are more toxic than the neurotoxic fragment 106126 (Brown, 2000a ). (ii) Changes in the central hydrophobic region have immediate consequences for the structure or the processing of PrP. This is in line with studies on other mutants in this domain that have used different experimental set-ups (Hegde et al., 1998 ; Hölscher et al., 1998 ; Muramoto et al., 1996 ). Recently, an alternative topological variant of PrP called ^CtmPrP has been proposed as a key factor in the development of neurodegeneration in prion diseases (Hegde et al., 1998 , 1999 ). The fragment size generated from ^CtmPrP by mild PK digestion was found to be smaller than PK-resistant PrP^Sc. The general importance of ^CtmPrP for the pathogenesis of prion diseases remains controversial (Stewart & Harris, 2001 ).

Structural data on full-length recombinant PrP define the N-terminal half of the protein (residues 23120 in moPrP and residues 23124 in hamster PrP) as flexibly disordered or random-coil (Donne et al., 1997 ; Riek et al., 1997 ). Copper binding seems to confer a distinctive structure on the N-terminal octapeptide repeats (Viles et al., 1999 ), but the motif AGAAAAGA is positioned at the hydrophobic C terminus of this unstructured part of PrP, and a recent refinement of the structure by Liu et al. (1999) has confirmed that it does not adopt regular secondary structure. Yet, the latter authors found indications of multiple discrete conformations in this hydrophobic domain, which might imply that this region is metastable and could therefore play an essential role in the formation of PrP^Sc. It was suggested that PrP^Sc interacts with recombinant PrP^C at amino acid residues 112119 of the mouse sequence (Brown, 2000b ), which might indicate that the sequence encompassing residues 112119 is of special importance to the conversion of PrP^C to PrP^Sc. Copper ions reversibly induce PrP^C to become protease-resistant (Quaglio et al., 2001 ), but it remains to be seen in what way the binding of copper to the N-terminal repeats and the concomitant adoption of structure influences the structural transitions that potentially take place in this more C-terminal hydrophobic domain of the N terminus. Our data demonstrate that a certain mutation in the hydrophobic region induces the ready formation of PK-resistant PrP. Therefore, it can be hypothesized that this mutation stabilizes a conformation that shares features with the infectious form of PrP. Whether prion infectivity was generated de novo in tissue culture will be tested in future by using infectivity assays and transgenic animals.

We thank Stanley B. Prusiner for antibody 13A5 and the tg81 mice, Bruno Oesch for antibody 6H4 and Alexander Buerkle for antibody Kan72. We also thank Martin Groschup for providing brain from hamsters and hamsters infected with scrapie strain 263K, B. Zevnik for his gift of plasmid pHA58 and H. Bujard for his generous gifts of plasmids pUHG 172-1 and pUHD 10-3. We are indebted to Connex GmbH (Martinsried, Germany) for the use of their S2 facility. This work was supported by grants from the European Commission (grants no. BMH4-CT98-6040 and ERBFMH4 CT98 6040).

Footnotes

^b Present address: Department of Neuropathology, Oxford, UK.