Integrity of Helix 2-Helix 3 Domain of the PrP Protein Is Not Mandatory for Prion Replication*

Background: Prions are self-propagating β-sheet-rich conformers of PrP protein, however, domains involved in conformational modification remain undetermined. Results: Peptide insertions in the H2-H3 inter-helix segment or C-terminal part of H2 do not prevent prion replication. Conclusion: The center of the H2-H3 domain of PrP is not critical for prion conversion. Significance: These data improve current knowledge on structural features of prions. The process of prion conversion is not yet well understood at the molecular level. The regions critical for the conformational change of PrP remain mostly debated and the extent of sequence change acceptable for prion conversion is poorly documented. To achieve progress on these issues, we applied a reverse genetic approach using the Rov cell system. This allowed us to test the susceptibility of a number of insertion mutants to conversion into prion in the absence of wild-type PrP molecules. We were able to propagate several prions with 8 to 16 extra amino acids, including a polyglycine stretch and His or FLAG tags, inserted in the middle of the protease-resistant fragment. These results demonstrate the possibility to increase the length of the loop between helices H2 and H3 up to 4-fold, without preventing prion replication. They also indicate that this loop probably remains unstructured in PrPSc. We also showed that bona fide prions can be produced following insertion of octapeptides in the two C-terminal turns of H2. These insertions do not interfere with the overall fold of the H2-H3 domain indicating that the highly conserved sequence of the terminal part of H2 is not critical for the conversion. Altogether these data showed that the amplitude of modifications acceptable for prion conversion in the core of the globular domain of PrP is much greater than one might have assumed. These observations should help to refine structural models of PrPSc and elucidate the conformational changes underlying prions generation.

The process of prion conversion is not yet well understood at the molecular level. The regions critical for the conformational change of PrP remain mostly debated and the extent of sequence change acceptable for prion conversion is poorly documented. To achieve progress on these issues, we applied a reverse genetic approach using the Rov cell system. This allowed us to test the susceptibility of a number of insertion mutants to conversion into prion in the absence of wild-type PrP molecules. We were able to propagate several prions with 8 to 16 extra amino acids, including a polyglycine stretch and His or FLAG tags, inserted in the middle of the protease-resistant fragment. These results demonstrate the possibility to increase the length of the loop between helices H2 and H3 up to 4-fold, without preventing prion replication. They also indicate that this loop probably remains unstructured in PrP Sc . We also showed that bona fide prions can be produced following insertion of octapeptides in the two C-terminal turns of H2. These insertions do not interfere with the overall fold of the H2-H3 domain indicating that the highly conserved sequence of the terminal part of H2 is not critical for the conversion. Altogether these data showed that the amplitude of modifications acceptable for prion conversion in the core of the globular domain of PrP is much greater than one might have assumed. These observations should help to refine structural models of PrP Sc and elucidate the conformational changes underlying prions generation.
Prions cause fatal neurodegenerative diseases such as Creutzfeldt-Jakob disease, Gerstmann-Straussler-Scheinker syndrome, and fatal familial insomnia in humans, scrapie in sheep and goats, bovine spongiform encephalopathy in cattle, and chronic wasting disease in cervids. These disorders are associated with the conformational conversion of a monomeric cellular prion protein, PrP C , 3 to an aggregated, pathological form, PrP Sc (1,2). Although the structure of PrP C is well characterized (3)(4)(5), that of the abnormally folded conformer PrP Sc remains a critical challenge. PrP C is a conserved cell surface, GPI-anchored glycoprotein that comprises an unstructured NH 2 -proximal half followed by a globular part arranged in three ␣-helices and a short antiparallel ␤-pleated sheet. A single disulfide bond keeps helices 2 and 3 closely arranged. Although PrP C and PrP Sc appear to have identical primary structures, they differ profoundly in secondary structure and biophysical properties. PrP C is largely helical and sensitive to proteinase K (PK) digestion, whereas PrP Sc is enriched in ␤-structure, insoluble, and partially resistant to PK (6 -11). The conformational dynamic changes underlying PrP Sc formation remain enigmatic, as well as the structural features conferring PK resistance to the infectivity associated, prion protein core, usually starting at residue ϳ82-97 and extending up to the C terminus (12)(13)(14)(15). There is evidential consensus that the N-terminal moiety, removed by PK treatment in vitro or lysosomal proteases in cells (16,17), is not essential for the replication because prions were generated from the N-terminal-deleted PrP (either ⌬23-88 or ⌬32-93) in infected transgenic mice (18,19). Decisive experimental data about which region of the PrP PK-resistant domain carries the information for structural transition are still lacking, yet a number of theoretical models have been proposed. Some models consider either the N- (20,21) or C-terminal part (22)(23)(24)(25) of the PK-resistant domain as the minimum essential region that supports structural modifications, whereas others involve both parts of the PK-resistant domain as being essential for prion conversion (11,26). It is worth noting that the punctual missense mutations associated with inherited forms of prion diseases are clustered between residues 97 and 148 and between 178 and 238 (27). Reverse genetic approaches to express modified PrP sequences have provided some clues but are constrained by the fact that mini-mal amino acid substitutions in the PrP protein can hamper prion replication and/or may have a dramatic effect on the susceptibility to infection (28 -31). Deletions within the proteaseresistant domain, whereas inducing neuropathology (32)(33)(34)(35), generally do not allow prion replication, except for PrP106, i.e. PrP with two deletions (⌬23-88, ⌬141-176), the so called miniprion (18). However, subsequent studies led these authors to propose the participation of the 141-176 region in the structural changes (20). Recently Myc or tetracysteine tags were introduced by insertion or sequence modification at one each end of the PK-resistant domain (36 -38), and shown to be compatible with prion conversion. However, introduction of such tags farther within this domain had so far prevented prion replication (36,38). Here we report the conversion into prion of several PrP mutants with an insertion of 8 -16 amino acids in the middle of the H2-H3 region, hence demonstrating an unexpected tolerance to sequence modification in a central part of the PK-resistant core.

EXPERIMENTAL PROCEDURES
Ethic Statement-All animal experiments have been performed in accordance with the European Community Council directive 86/609/EEC.
Generation of PrP Insertion Mutants-Insertions and substitutions were introduced by site-directed mutagenesis (QuikChange II mutagenesis kit, Stratagene) into sheep PrP (allotype Val 136 -Arg 154 -Gln 171 ) cloned into pTRE plasmid (Clontech) using mutagenic primers (supplemental Table S1). The plasmid encoding PrP-ins203, i.e. with an in-frame insertion of 24 bp, was isolated during screening of an experiment originally intended to modify the second site of glycosylation (31). The sequences of all the mutant constructs were verified by sequencing. Numbers indicated for each insertion refer to the position of the first amino acid of the insert according to the sheep PrP numbering.
Cell Culture and Isolation of Rov Cells-Rov cells are epithelial RK13 cells that stably express either wild-type (WT) or mutant ovine PrP using a tetracycline-inducible system (31,39). They were obtained by transfection and puromycin selection and were grown in Opti-MEM medium (Invitrogen) supplemented with 10% fetal calf serum (FCS) and antibiotics, and split at 1:4 after trypsin dissociation once a week. To assess PrP C expression, cells were cultivated in the continuous presence of 1 g/ml of doxycycline (Sigma).
Prion Infection of Cell Cultures-Rov cells were infected with the cloned 127S scrapie strain as previously described (31) using 1% (w/v) brain pool homogenates of terminally ill tg338 mice (44). Two days postexposure cells were washed, incubated for 2 additional days, trypsinized, and split at 1:10 dilution. Cells were further incubated for 1 week before analysis (passage 1) and split at 1:4 dilution at each following passage. To test the infectivity of cultures propagating mutant PrPr Sc cells were harvested at passage 6 -8 postexposure, pelleted, frozen and thawed, then sonicated, and used as inoculums.
Cell Lysis, Proteinase K Digestion, and PNGase Treatment-Cells were washed twice with cold phosphate-buffered saline and whole cell lysates were prepared at 4°C in TL1 buffer (50 mM Tris-HCl, pH 7.4, 0.5% sodium deoxycholate, 0.5% Triton X-100). Lysates were clarified by centrifugation for 2 min at 800 ϫ g and protein concentrations were determined by microBCA assay (Pierce). For PrP res , lysates were incubated with 4 g of PK per 1 mg of protein, for 2 h at 37°C, then centrifuged for 30 min at 22,000 ϫ g. Pellets were dissolved in Laemmli sample buffer and boiled for 5 min at 100°C. When needed, 500 units of PNGase F (New England Biolabs, MA) and 1% Nonidet P-40 were added to denatured proteins that were further incubated at 37°C (17).
Transgenic Mice, in Vivo Infections, and Tissue Homogenate Preparation-To test for the infectivity of PrP Sc -ins203, 5 ϫ 10 6 Rov-ins203 or RK13 cells were harvested by scraping at passage 8 post-infection. Cell pellets were resuspended in 100 l of medium containing 5% glucose and antibiotics, then frozen and thawed three times before sonication for 1 min. A 20-l aliquot of these cellular lysates was injected intracranially to tg338 mice overexpressing the VRQ allele of ovine PrP (44). Brains and spleens of infected animals were harvested at the terminal stage of the disease and analyzed for PrP res content and distribution by immunoblots and histoblots, respectively, as described previously (45).
Immunoblotting Detection of PrP C and PrP res -Either 12 or 4 -12% NuPage BisTris polyacrylamide gels (Invitrogen) were used for SDS-PAGE. For PrP C analysis, 50 g of protein were deposited on the gel. For PrP res , samples corresponding to the PK-resistant PrP present in 50 g of protein of the cellular lysate were loaded on the gel, except for PrP res -ins(Gly 6 ) and PrP res -ins(FLAG), which an equivalent of 250 g of protein were loaded as indicated in the figure legends (Figs. 4D and 5D). The transfer of proteins, their detection and revelation were described previously (17).
Conformational Stability Assay-Cellular lysates were brought to the desired concentration of guanidine HCl, incubated for 1 h at 20°C under agitation, then diluted to reach the same volume and guanidine HCl concentration (800 l, 0.5 M). 200 l of 5ϫ digestion buffer (750 mM Tris-HCl, pH 7.4, 10% Sarkosyl and 20 g/ml of PK) was added and samples were incubated for 1 h at 37°C under agitation, before methanol precipitation and analysis by immunoblotting.
Cu 2ϩ -Immobilized Metal Affinity Chromatography (IMAC) Hitrap Chromatography-The AKTA Purifier100 FPLC chromatographic system was used (GE Healthcare) as previously described (17,46). One milliliter of cell lysates (1 mg of protein) made in TNT buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 1% Triton X-100) supplemented with protease inhibitors (Roche Applied Science) was injected into the column. Columns were washed and a linear gradient of 3-200 mM imidazole in TNT buffer was applied to elute column-bound proteins at a flow rate of 1 ml/min for 10 min; 0.5-ml fractions were collected. To solubilize aggregates of PrP Sc -ins(His 6 ), all buffers were complemented with 4 M urea. Eluted fractions were either treated with 5 g/ml of PK and precipitated with cold methanol after addition of 20 g of BSA as a protein carrier or methanol precipitated without PK treatment for total PrP. Individual columns were dedicated to each type of lysate.
Immunofluorescence-Cells were fixed with 4% paraformaldehyde and PrP C was detected as previously described (31,47). For PrP Sc immunodetection, fixed and permeabilized cells were treated for 5 min with 3.5 M guanidium thiocyanate and washed three times before exposure to ICSM33 mAb as previously described (42,47), or to both ICSM33 and M2 anti-FLAG mAbs and then to fluorescent anti-IgG2b and anti-IgG1. Images were acquired with an Axio observer Z1 microscope (Zeiss) and a coolsnap HQ 2 camera (Photometrics) driven by the Axiovision imaging system software.
NMR Measurements-Recombinant proteins (WT PrP, PrP-ins191, and PrP-ins193) were produced and purified from Escherichia coli as published previously (25). HSQC experiments were run at 25°C on 0.5-0.7 mM 15 N-labeled samples of wildtype PrP and mutants using a VARIAN INOVA 600 MHz. The samples were in 5 mM MOPS buffer, pH 7.0. Water suppression was achieved with a standard Watergate pulse sequence.
Circular Dichroism-CD measurements were carried out on a Jasco-715 spectropolarimeter equipped with a thermostated cell holder controlled by a Jasco Peltier element. Far-UV CD spectra were recorded from 260 to 195 nm at 25°C in a 1-mm path-length quartz cuvette at a protein concentration of 12 M in 5 mM MOPS buffer, pH 7.0. Each CD spectrum was obtained by averaging 15 scans collected at a scan rate of 200 nm/min. Baseline spectra obtained with buffer were subtracted for all spectra. The thermal denaturation scans were obtained by heating a 12 M sample from 20 to 97°C at 2°C/min and by monitoring the ellipticity at 220 nm with a 4-s averaging time and a 0.2°C data pitch. The thermogram was analyzed assuming a two-state model.

PrP C with Extended H2-H3 Loop Can Convert into Prion-In
a previous study aimed at producing PrP glycosylation mutants (31), we identified a mutant that contained an insertion of 8 amino acids downstream from the second glycosylation site, within the unstructured loop joining helices H2 and H3 of sheep PrP (Fig. 1, A and B). This mutant, denoted PrP-ins203, arose fortuitously probably due to a misalignment of the mutagenic primers used. To examine the effects of this mutation on PrP glycosylation and permissiveness to prion infection, PrP-ins203 was stably expressed in RK13 cells. Immunoblotting analysis revealed the presence of mono-and diglycosylated PrP-ins203 but the apparent lack of the unglycosylated form (Fig. 1C). By quantification on overexposed blots, this latter form was found to be 15-20-fold less abundant than WT PrP (Fig. 1D). PrP-ins203 polypeptide showed a slightly reduced mobility compared with WT PrP, consistent with the presence of additional residues (Fig. 1E). The anti-PrP mAb V14, which recognizes a conformational epitope (amino acids 191-199) encompassing a region located nearby upstream of the site of insertion of the octapeptide ins-203 (5,41), was unable to recognize PrP-ins203 ( Fig. 1E and supplemental Fig. S1). This suggested that the presence of additional amino acids just downstream of the V14 epitope somehow modifies the local conformation of the PrP. Yet the insertion mutant was correctly routed to the cell surface, as ascertained by PrP immunofluorescence labeling of intact cells (Fig. 1F) and presence in lipid rafts (supplemental Fig. S2).
Cells expressing either WT or mutant PrP were infected with the 127S strain of ovine prion and analyzed for their PrP res content at subsequent passages of cultures. The immunoblot in Fig. 2 shows that PrP-ins203 was converted into a self-propagating, protease-resistant species with amounts of PrP res comparable with those in cells expressing WT PrP (Fig. 2). After PNGase treatment, the PrP res fragment of the insertion mutant was shown slightly bigger in size than WT PrP, thus indicating that the inserted amino acid stretch was protected from protease digestion in PrP Sc . To further characterize the PrP-ins203 mutant we subjected it to a conformational stability assay using guanidine hydrochloride as a denaturing agent (48). The insertion did not appreciably modify the resistance to denaturation (Fig. 3A), because half-denaturation of the mutant and WT PrP Sc occurred at 1.4 Ϯ 0.2 and 1.3 Ϯ 0.4 M, respectively.
PrP-ins203 conferred cells similar susceptibility to prion infection as WT PrP in cell cultures exposed to serial dilutions of 127S-infected mouse brain homogenate. Both Rov-wt and Rov-ins203 produced large amounts of PrP res after exposure to inoculum at the 10 Ϫ3 dilution, whereas no PrP res was detected beyond the 10 Ϫ4 dilution (Fig. 3B). To confirm that PrP-ins203 led to the generation of infectious prion, the cell homogenate from 8 passages on Rov-ins203 cultures was inoculated to naive Rov-ins203 or Rov-wt cells, and to tg338 mice. Both homotypic and allotypic cell cultures were successfully infected based on PrP res accumulation (Fig. 3C). Mice inoculated with the PrP-ins203-derived homogenate died at 61 Ϯ 2 days with typical TSE signs (Fig. 3D) and showed the presence of PrP res with a glycoform pattern and regional deposition in the brain typical of the 127S strain (supplemental Fig. S3). Altogether these results demonstrate that a mutant PrP with an 8-amino acid insertion in the PK-resistant core is efficiently convertible into bona fide prion.
Insertion of Glycine Residues, or Extension of the H2-H3 Loop Up to 16 Amino Acids Do Not Prevent Prion Replication-We modified the original insertion already containing 3 glycine residues to create a segment containing 6 glycines (Fig. 4A). Because a feature of the polyglycine stretch is to be incompatible with the formation of secondary, either ␣-helix or ␤-sheet structures, it was assumed that the insert ins203(Gly 6 ) would neither participate to elongation of H3 nor convert into a ␤-sheet segment if PrP Sc were to be produced. When expressed in cells, PrP-ins203(Gly 6 ) showed a glycoform pattern with a low level of the unglycosylated form as for the original PrP-ins203, and some decrease in the proportion of the diglycosylated form (Fig. 4B). PrP-ins203(Gly 6 ) was efficiently routed to the cell surface and found to be associated with lipid raft fractions (supplemental Fig. S2) and was converted into PK-resistant species upon infection with 127S prion (Fig. 4, C and D). PrP Sc -ins203(Gly 6 ) was readily propagated along with five subpassages, although the amount of PrP res was reduced by at least FIGURE 2. 127S-infected Rov-ins203 cells produce self-propagating PrP res . Infected cell cultures were split once a weak and samples of PK-digested cellular lysates were analyzed by immunoblotting, the passage number is indicated at the top of the upper panel. Note that the unglycosylated form of PrP Sc -ins203 was not detected to this level of exposure and that, as expected, mock-infected cells (m.i.) did not produce protease-resistant PrP. In the lower two panels, PrP res was treated with PNGase F before immunoblotting. Sha31 anti-PrP antibody recognizes an epitope located in the helix H1, whereas V14 antibody recognizes a conformational epitope previously localized nearby and upstream of the insert. 10-fold compared with that in WT PrP Rov cells (Fig. 4D). Thus, extension of the loop between H2 and H3 by a polyglycine stretch remained compatible with PrP Sc formation, suggesting that this part of the protein is unstructured in misfolded conformers.
We also introduced an 8-amino acid FLAG sequence into the original ins203 octapeptide, enabling us to recognize the presence of the insert using a relevant antibody (Fig. 5A). As for other ins203 insertions, the unglycosylated forms of PrP-ins-(FLAG) were produced in low amounts in RK13 cells (Fig. 5B), and FLAG-PrP was efficiently expressed at the cell surface (Fig.  5C). Rov-ins203(FLAG) were susceptible to prion infection and generated self-propagating PrP Sc , however, the PrP res levels were markedly reduced compared with WT PrP (Fig. 5D). Cells in which PrP Sc had accumulated efficiently exhibited numerous aggregates labeled, after guanidium thiocyanate denaturation, by anti-PrP and anti-FLAG antibodies (Fig. 5E). We further showed that PrP Sc -ins(FLAG) autopropagated in cell culture was also infectious for Rov-wt cells (Fig. 5F). Altogether these results established that a PrP C with a 4-fold increase, i.e. from 5 to 21 amino acids, of the size of the sequence between H2 and H3 can be converted into bona fide prions.
Production of His-tagged PrP Sc -As a means to facilitate PrP purification, the first 6 amino acids of ins203 were substituted by histidine residues and the resulting PrP-ins203(His 6 ) was expressed in RK13 cells (Fig. 6A). The functionality of the tag was tested by loading cellular lysates onto Cu 2ϩ -IMAC columns. Because PrP already contains an N-terminal octarepeat sequence, a strong binding site for divalent ions, both WT and the His-tag PrP C were retained on such columns (Fig. 6B). However, PrP-ins(His 6 ) was eluted at a much higher concentration of the Cu 2ϩ competitor imidazole (150 mM) than WT PrP (90 mM), indicating that the His tag actually increased the avidity for immobilized copper ions of PrP (Fig. 6C), including diglycosylated molecules where complex N-glycan chains are attached close to the tag. Interestingly, the fractions in which His tag PrP was eluted contained at least 10-fold less contaminant proteins than the fractions containing WT PrP (not shown).  To determine whether the H2-H3 segment was accessible to interaction into PrP Sc molecules as well, Rov-ins(His 6 ) cells were subjected to prion infection. Infected cells produced PrP res from the first passage on, the amount of which increased gradually on subpassaging to reach a level comparable with that produced by Rov-wt after 7-8 passages. His-tag PrP Sc was able to infect cultures expressing homotypic as well as WT PrP sequence (Fig. 6, C and D). Anti-His antibodies recognized the PK-resistant fragment (Fig. 6C, right panel). The half-denaturation value of PrP Sc -ins(His 6 ) in conformational stability assay was 1.3 Ϯ 0.25 M, similar to WT and ins203 PrP res (see above).
We also asked whether a His tag inserted in the PK-resistant core would still bind to copper-immobilized ions. Although we did show binding of the His tag PrP Sc to IMAC Cu 2ϩ columns (Fig. 7), the bulk of PrP Sc did not bind and was either excluded in the flow-through (Fig. 7, top panel) or retained on the column even after application of imidazole at high concentrations (not shown). To improve the purification of His-tagged PrP Sc , cellular lysates were incubated with 4 M urea prior to application on the columns. Proteins recovered by elution with increasing imidazole concentrations were PK-digested and analyzed by immunoblotting. WT PrP Sc , present in Rov cellular lysates essentially as N-terminal truncated species (17), bound to Cu 2ϩ columns with low avidity and started to elute at 40 mM imidazole (Fig. 7, lower panel). His-tagged PrP res was eluted at higher concentrations of imidazole, starting at 88 mM (Fig. 7,  middle panel), indicating that the tag could be functional for the PrP res species under adequate conditions of solubility.

Integrity of C-terminal Part of Helix-2 Is Not Essential for
Prion Conversion-To determine how variable the exact location of the 8-amino acid insertion could be, we moved it upstream and downstream of the original position (Fig. 8A). The insert was introduced at the other end of the inter helix loop (position 198), or in helix H2 (195 to 187) or helix 3 (207 and 210). These mutants were expressed and efficiently reached the cell surface (supplemental Fig. S4A), except PrP-ins210, which was not detectably expressed at the membrane but showed intracellular accumulation, essentially at a juxtanuclear position (supplemental Fig. S4B). PrP-ins198 did not confer Rov cells susceptibility to prion infection, indicating that even in the unstructured segment considered, the site of insertion might be critical. Cells expressing H3 mutants, Rov-ins207 and Rov-ins210, were not permissive to the infection, yet the convertibility of PrP-ins210 could not be adequately tested due to impaired trafficking.
In contrast, PrP with insertion into H2, 3 amino acids before the end of the helix (position 195) and then further upstream with 1 amino acid increment up to 193 conferred permissiveness to the cells and led to reproducible and robust infection as attested by the formation of self-propagating PrP res (Fig. 8B). Demonstration of the infectivity of each of the three mutant PrP Sc was ascertained by using cellular homogenates to challenge naive Rov-wt and homotypic cells (supplemental Fig. S5). Insertion further upstream markedly affected (PrP-ins192) or even hampered (PrP-ins191 and PrP-ins187) prion replication (Fig. 8B). These results showed that insertion of extra amino FIGURE 6. Insertion of six histidines increased the copper affinity of PrP C and PrP-ins(His 6 ) was converted into a His-tagged prion. A, the glycoform pattern of PrP-ins(His 6 ) shows a reduction of the unglycosylated form similar to the original ins203 PrP (4F2 mAb). B, WT PrP was eluted from the Cu 2ϩ IMAC column at a lower concentration of imidazole than PrP-ins(His 6 ). Note that the monoglycosylated forms were eluted after the diglycosylated ones for both WT and ins203(His 6 ) PrP. Immunoblots were revealed with the 4F2 mAb. C, immunoblot showing that PK-resistant PrP was generated upon infection. The rate of PrP res increased notably after five to six passages of the cell culture (left panel, Sha31 mAb). The presence of the His tag on the PK-resistant fragment was ascertained using an anti-His 6 mAb (right panel). D, Rov cells expressing WT PrP and Rov-ins(His 6 ) were susceptible to de novo infection by cellular homogenate of 127S-infected Rov-ins(His 6 ) harvested at passage six (PK-treated samples and Sha31 mAb).
acids in the part of the protein corresponding to last two turns of H2 (Fig. 9) did not prevent prion formation, whereas introduction of the same insertion further upstream in the helix impaired conversion.
Insertions in H2 C Terminus Do Not Interfere with PrP-fold-To understand whether the inserted sequences affect the fold of PrP, we recorded NMR 15 N HSQC experiments, which are sensitive to protein tertiary fold. As insertions in H2 could be more destabilizing for mutants than insertion in the loop, we specifically analyzed PrP-ins193 and PrP-ins191. The former is competent for prion conversion but not the latter, although they are both expressed at the cell surface and found associated with FIGURE 7. Retention of PrP-ins(His 6 ) prion by chromatography affinity. A, part of the PK-resistant PrP-ins(His 6 ) produced by infected cells was retained with high affinity to the column. B, PrP Sc -ins203(His 6 ) solubilized with 4 M urea was more efficiently retained on IMAC copper columns than the WT PrP res . At this particular concentration urea allowed complete PK digestion of PrP C but preserved the protease resistance of both the His tag and WT PrP Sc . Eluted samples were treated by PK and analyzed by immunoblotting (Sha31 antibody). The insert is in red and amino acids that belong to, or are originally from helix-2 are in green. Note that the bottom sequence contains two inserts, each at a different position. The name of the mutant is on the left and its capacity to convert into prion is indicated on the right. B, assay for PrP res content of the insertion mutant PrP modifying the loop, the helices, or both. The same amount of protein digested by PK was used throughout. The passage number is indicated above each panel and the asterisk indicates that a longer exposure is shown.
lipid rafts (supplemental Figs. S2 and S4). Moreover the insertion is moved only 2 amino acids toward the center of H2 in the PrP-ins191 with respect to the PrP-ins193. The spectra of the wild-type and mutant proteins were similar (Fig. 10A). They all showed a good resonance spreading that is typical of folded proteins indicating that the insertions do not significantly affect the tertiary structure, which is thus retained. Minor differences can reasonably be attributed to a different packing of the secondary structural elements in the three proteins that leads to different local chemical environments. These results are consistent with previous observations that had proven that the presence of the disulfide bridge in the middle of H2 and H3 is essential for retaining the fold of the core element H2-H3 (25): we do not observe a major destabilization because the insertions introduced do not affect the disulfide bridge.
To understanding further the role of the insertions, which is whether they introduce additional secondary structures by elongating the helices or merely substitute the native structure, we recorded far-UV CD spectra of PrP-ins191 and PrP-ins193 and compared them with that of the wild-type protein (Fig.  10B). They were identical within experimental error suggesting that the additional residues do not promote formation of an additional secondary structure. Finally, we questioned whether the insertions would have an effect on the fold stability. We recorded temperature scans followed by CD. We observed that, whereas having very similar highly cooperative behavior, the wildtype seemed to reproducibly be 2-3°C more stable than the two mutants suggesting that, if at all, the insertions have a mild destabilizing effect (Fig. 10B). Taken together these results tell us that these insertions do not perturb the overall fold of PrP C .
Conversion into Prion of Mutant with Two Insertions-We have also constructed a double mutant by inserting the octapeptide concomitantly at two places previously found to be tolerant to insertion, i.e. at position 195, into H2, and at position 203, in the loop between H2 and H3. Upon infection, PrP-ins(195 ϩ 203) was convertible into a prion that was readily and reproducibly (3 times out of 3 trials) propagated in cell culture (Fig. 8). Thus again, mutant PrP with 16 additional amino acids in a central part of the PK-resistant core proved to be apt to form prion.

DISCUSSION
In this study, we expressed a number of sheep PrP mutants displaying a peptide insertion at various places and we show that nine mutants with exogenous sequences introduced in the central part of the H2-H3 region remain convertible into prions, thus questioning how critical is the integrity of this region for the conversion into PrP Sc . This issue is important particularly in view of several recent reports proposing the crucial involvement of the H2-H3 region in misfolding events leading to prion formation (11,22,24,26,49,50).
PrP Proteins with Extended H2-H3 Interhelix Segment Remain Convertible into Prions-Four PrP mutants were built with insertions at position 203 that increased about 2-4-fold the size of the loop joining H2 and H3. The resulting proteins were glycosylated and properly routed to the cell surface, whereas sharing a notable reduction of unglycosylated PrP species. The latter feature might result from an increased glycosylation efficiency (51) or a reduced stability of this form, or both, as observed for the F198S mutation in human (52)(53)(54) and for its equivalent in the sheep sequence, F201S (31), both of which are close to the insertion site. Upon prion infection ins203 mutants produced self-propagating PrP Sc albeit at different levels. The original PrP-ins203 generated as much PrP res as WT PrP, whereas cells expressing the His-tagged PrP had to undergo several passages before reaching similar levels, suggesting an adaptation of the prion strain to this particular substrate or selection of most performing components (55). The other two PrP mutants produced lower quantities of PrP res , which may reflect reduced replication efficiency, or a lower stability of the misfolded protein in the cells. Nevertheless, these insertion mutants of PrP Sc were de novo infectious for naive cells or for transgenic mice expressing the sheep PrP. Altogether these results proved that bona fide prions could be generated despite a substantial increase in size of the segment separating H2 and H3, the heart of structured PrP Sc . The magnitude of acceptable changes in this region was impressive because not only was the size of the H2-H3 connecting segment increased, but the physicochemical characteristics of the loop were also drastically altered. In PrP-ins(His 6 ) and PrP-ins-(FLAG), in particular, the calculated isoelectric point and charge at pH 7 of the loop segment were markedly modified.
PrP C and PrP Sc Conformations of Insertion Mutants in Position 203-The loss of the conformational V14 epitope in PrP-ins203 might suggest some local dynamic change of the conformation of the region immediately upstream of the insert (56), as the epitope was reported to cover the domain between the two glycosylation sites, from the middle of H2 to the beginning of the H2-H3 loop (5,41). However, another possibility is that the size of the segment encompassing the epitope would have been previously underestimated. PrP Sc -ins203 had a similar resistance to guanidium denaturation as WT PrP Sc , suggesting no major structural variation between the two misfolded proteins, despite dramatic under-representation of aglycosylated species in PrP Sc -ins203. Moreover, the strain features in tg338 mice of PrP Sc -ins203 were typical of the 127S strain propagated on WT PrP cultures. Importantly, the introduction of six consecutive glycines in the insert did not preclude prion conversion. This FIGURE 9. Schematic structure of WT PrP with the indication of regions tolerant to insertion. ␣-Helices are shown in red and the two short ␤-sheets in yellow. H2 and H3 are linked by a disulfide bond that maintains them close to each other, and the unstructured inter-helix segment forms a loop. The position where the four different insertions (ins203, ins203(Gly 6 ), ins203(His 6 ), and ins203(FLAG)) were introduced, just before H3, is indicated. The C-terminal part of H2 in which octapeptides can be introduced without preventing prion conversion is colored in dark blue.
finding strongly suggests that the corresponding segment was not changed into the ␤-sheet in PrP Sc , and comes in support of the structurally based prediction that the loop between H2 and H3 remains unstructured in PrP Sc (20,22).
Prions Can Be Tagged inside the PK-resistant Core-The insertion of tags might be useful to study the biology of prion infection. Previous observations tended to indicate that the ends of the PK-resistant fragment could be modified, whereas the inside part could not (36 -38). In fact, as shown here, tags can be introduced within the protease-resistant segment without blocking prion replication. Despite their location between two heavy N-glycan chains, the His tag and FLAG tag were easily accessible in PrP C and proved to be useful for purification and immunodetection experiments. Yet, purification or detection of His-tagged PrP Sc could not be achieved without partial solubilization of aggregates by urea. Also, PrP Sc -ins(FLAG) was not recognized by anti-FLAG antibodies without prior guanidine treatment of the cells. Both results are consistent with previous observations on other tagged PrP Sc and the inaccessibility of all these tags to proteolytic enzymes in misfolded PrP (36 -38).

Integrity of the Last Two Turns of H2 Is Not Mandatory for
Prion Replication-Insertion of the octapeptide (AQQ-GGGFT) either at the beginning of the H2-H3 loop (position 198) and in the N-terminal part of H3 (position 207) impaired the formation of prion. Nevertheless, the integrity of the extremity of H3 or of the first residue of the H2-H3 loop is unlikely to be essential for prion conversion because insertion of another sequence at close proximity appeared to be compatible with PrP res formation (57). In contrast, introduction of the octapeptide in the C-terminal part of H2 allowed the generation of four PrP mutants that all remained prion-compatible as attested by self-propagation of PrP res and de novo infectivity of the mutant PrP Sc . Of note, the levels of newly formed PrP res decreased while moving the insertion from position 195 to 192, with no conversion occurring beyond this position. As the thread of helix comprises a little more than 3 amino acids, it means that insertion in the last two C-terminal turns of H2 allowed the formation of bona fide prions.
Furthermore, a double mutant with an insert in the last turn of H2 and another in the H2-H3 loop was competent for prion FIGURE 10. Structural analyses of PrP with insertions in H2 conferring permissiveness to 127S or not. A, comparison of the NMR HSQC spectra at 25°C of ins191 (top), ins193 (middle), and wild-type PrP (bottom) recorded at 600 MHz. B, comparison of the far-UV CD (top) and the temperature scans followed at 220 nm (bottom) of ins191 (red), ins193 (blue), and wild-type (black) PrP recorded at 25°C. The spectra were normalized for the concentration.
conversion and propagation, showing that the simultaneous modification of both regions was tolerated. Like the PrP-ins-(FLAG) mutant mentioned above, PrP-ins(195 ϩ 203) harbors 16 additional amino acids, strengthening the view that neither the size of the middle part of the H2-H3 region nor the amino acid sequence are critical for 127S prion replication. Altogether these findings were unexpected as the H2-H3 region is proposed to be essential for prion conversion according to several, recently published models (22,26,49), in which these two helices are proposed to undergo drastic conformational change and to adopt a ␤-sheet conformation in the prion state (11,23,50). Other models, however, consider that it is mainly the N-terminal moiety of the PK-resistant fragment that supports structural modification (20,21,58). Also, a recent paper has reported productive infection in cell cultures expressing a PrP molecule with two point mutations substituting original amino acids by cysteines, one in the H2 end and the other in the loop (59).
Insertions in H2 End Preserve Overall Fold of the H2-H3 Domain-We wonder whether the insertions could have destabilizing effects on the structure of PrP C and particularly on the H2-H3 domain. This question concerned both the PrP construct competent for prion conversion and those that are not. We thus turned to structural analysis of the recombinant insertion mutant PrP assuming that their structures would reflect characteristics of the corresponding glycosylated PrP anchored at the membrane of mammalian cells. The conclusions from these works that insertions ins193 are compatible with prion conversion and ins191, are not compatible, do not induce significant alteration of the overall fold of PrP.
As a whole, if we assume the increasingly supported hypothesis of a conformational change of the H2-H3 domain in prions (22)(23)(24)(25)(26), the comparison of infection data with structural analysis of PrP-ins193 and PrP-ins191 may suggest that when approaching the disulfide bridge, which could induce specific constraints to surrounding areas, modifications of H2 are less important for the overall fold of PrP C than for its conversion into PrP Sc .
In conclusion we showed that it is possible to generate bona fide prions with 8 -16 additional amino acids inserted in the middle of the PK-resistant domain, revealing an unexpected tolerance to PrP sequence change in the prion core. More specifically we showed that octapeptide insertions in either the H2-H3 loop or the last two C-proximal turns of H2 were not critical for prion conversion. These findings indicate that the integrity and sequence specificity of the loop and the two adjacent turns of H2 are not determinant for the structure of PrP Sc and will help to validate and refine structural models.