In vivo conversion of cellular prion protein to pathogenic isoforms, as monitored by conformation-specific antibodies.

The central event in prion disease is thought to be conformational conversion of the cellular isoform of prion protein (PrP(C)) to the insoluble isoform PrP(Sc). We generated polyclonal and monoclonal antibodies by immunizing PrP(C)-null mice with native PrP(C). All seven monoclonal antibodies (mAbs) immunoprecipitated PrP(C), but they immunoprecipitated PrP(Sc) weakly or not at all, thereby indicating preferential reactivities to PrP(C) in solution. Immunoprecipitation using these mAbs revealed a marked loss of PrP(C) in brains at the terminal stage of illness. Histoblot analyses using these polyclonal antibodies in combination of pretreatment of blots dissociated PrP(C) and PrP(Sc) in situ and consistently demonstrated the decrease of PrP(C) at regions where PrP(Sc) accumulated. Interestingly, same mAbs showed immunohistochemical reactivities to abnormal isoforms. One group of mAbs showed reactivity to materials that accumulated in astrocytes, while the other group did so to amorphous plaques in neuropil. Epitope mapping indicated that single mAbs have reactivities to multiple epitopes, thus implying dual specificities. This suggests the importance of octarepeats as a part of PrP(C)-specific conformation. Our observations support the notion that loss of function of PrP(C) may partly underlie the pathogenesis of prion diseases. The conversion of PrP(C) to PrP(Sc) may involve multiple steps at different sites.

Prion diseases, such as scrapie in sheep and goats and Creutzfeldt-Jakob disease in humans, are transmissible neurodegenerative disorders (spongiform encephalopathy). A major component of the infectious agent responsible for these diseases is thought to be a post-translationally modified form of a host-encoded membrane glycoprotein PrP C , 1 termed PrP Sc (1). Conformational differences are observed between PrP C and PrP Sc . PrP Sc has a large number of ␤-sheets and a diminished ␣-helical content compared with PrP C (2)(3)(4); hence, it is relatively resistant to protease digestion. The protease resistance of PrP Sc has been widely accepted as the physico-chemical basis to distinguish between PrP C and PrP Sc . A recent study demonstrated that plasminogen has selective affinity to PrP Sc but not to PrP C (5).
It is also ideal to use antibodies distinguishing conformations of PrP C and PrP Sc . However, since PrP C is phylogenetically highly conserved and is expressed in various tissues, including immune systems (6,7), immune tolerance means restriction in generating antibodies. To overcome this problem, other workers have used mice devoid of PrP C to generate monoclonal antibodies (mAbs) to prion protein (PrP) (8 -13). Almost all mAbs so far obtained recognize both PrP Sc and PrP C . Exceptionally, one of the mAbs (clone 15B3) was reported to be specific for PrP Sc (11). To gain insight into conversion processes, it is important to characterize epitopes reflecting conformational differences between PrP C and PrP Sc .
PrP C , expressed on the cell surface with a C-terminal glycosylphosphatidylinositol anchor, is expressed in most tissues of uninfected animals. The function of PrP C has remained obscure. However, it has been suggested that PrP C has a role for normal synaptic function (14,15). Furthermore, ablation of the prion protein gene (Prnp) caused cell death in some circumstances in vivo and in vitro (16,17). Although recent studies suggested that the ectopic expression of the Prnd gene, encoding a homolog of PrP C , was involved in cell death (32), expression of PrP C antagonized cell death signaling in both cases (17,18). These results raised the possibility that functional loss of PrP C might partly underlie the pathogenesis of prion diseases. However, the expression level of the mRNA of the Prnp gene is unaltered in cases of scrapie infection (19). Thus, the equilibrium of PrP C and PrP Sc in the pathogenesis remained to be examined.
We immunized Prnp Ϫ/Ϫ mice with native PrP C from wild type mice (Prnp ϩ/ϩ ) and established mAbs to PrP. These mAbs specifically recognized PrP C in solution with a marked decrease of PrP C at the terminal stage of illness. Epitope mapping of these mAbs and polyclonal antibodies from mice immunized with native PrP C suggested the importance of octarepeats as part of PrP C specific conformation.

EXPERIMENTAL PROCEDURES
Generation of Prnp Ϫ/Ϫ Mice-Prnp Ϫ/Ϫ mice were generated by homologous recombination, as described (20). The entire Prnp gene open reading frame as well as 3Ј-end of the intron 2 was replaced with a pgk-neo gene cassette. The resulting allele is almost identical to that generated by another group (16). The homologous recombinant ES clones were injected into blastocysts of C57BL/6 mice to give rise to chimeras. Mice heterozygous for the mutation (Prnp ϩ/Ϫ ) were obtained by crossing the chimeras with C57/BL6J mice. The heterozygotes (Prnp ϩ/Ϫ ) were further intercrossed to obtain mutation homozygotes (Prnp Ϫ/Ϫ ). The mouse line has been maintained by backcrossing to C57BL/6. Genotypes of the mice were determined by Southern blot analysis or by polymerase chain reaction analysis of DNAs prepared from tails of the mice. The primers used were 5Ј-GTACAGTAGAC-CAGTTGCTC-3Ј and 5Ј-CAGAGTCAAGATCTCCTAGT-3Ј for the wildtype allele and 5Ј-CTCGTGCTTTACGGTATCGC-3Ј and 5Ј-CAGAGT-CAAGATCTCCTAGT-3Ј for the mutated allele.
Antibodies-Fluorescein isothiocyanate-conjugated anti-mouse IgG and IgM, horseradish peroxidase (HRP)-conjugated anti-mouse IgM, fluorescein isothiocyanate-conjugated anti-rabbit IgG, and HRP-conjugated anti-rabbit IgG antibodies were purchased from Jackson Immunoresearch Laboratories. Biotinylated anti-mouse IgM antibody was purchased from Vector Laboratories. Anti-PrP peptide (position of the peptide on mouse PrP was amino acid residues 213-226) rabbit serum (Ab.Mo-VI) (21) was used as a reference antibody.
Immunization and Fusion Protocols-For immunization, 10% brain homogenates or 2 ϫ 10 7 thymocytes of Prnp ϩ/ϩ mice were given intraperitoneally to Prnp Ϫ/Ϫ mice four times at 2-3-week intervals. The brains were homogenized with RPMI medium and then centrifuged at 3,000 rpm for 30 min. The supernatant fraction was emulsified with Freund's complete and incomplete adjuvants for first and subsequent immunizations, respectively. Thymocytes were given intraperitoneally without adjuvants. The spleen cells of the immunized mice were fused to P3U1 mouse myeloma cells (P3 ϫ 63Ag8U.1) and cultured, as described (22). Subclasses of the mAbs were determined using a mouse mAb isotyping kit (Amersham Pharmacia Biotech).
PrP recombinant Baculovirus-Autographa californica nuclear polyhedrosis virus (AcNPV) and recombinant virus stocks were grown and assayed in monolayers of Spodoptera frugiperda ovary cells IPLB-SF-21AE (SF21AE) (23) in TC100 medium containing 10% (v/v) fetal bovine serum. The Prnp open reading frame fragment was amplified by polymerase chain reaction and subcloned into the BamHI site of the transfer vector pAcYM1S. Primers used were 5Ј-GGGATCCAGTCAT-CATGGCGAACCT-3Ј and 5Ј-GGGATCCACGAGAAATGCGAAGGAA-3Ј. SF21AE cells were cotransfected with AcNPV DNA and transfer vector DNA by Lipofectin (Life Technologies, Inc.), and then the recombinant baculovirus PrP-AcNPV was selected, as described (24). The expression of recombinant PrP was confirmed by indirect fluorescent assay (IFA) and WB by using Ab.Mo-VI (21). PrP-AcNPV-infected cells were washed with phosphate-buffered saline and then fixed with acetone for 5 min at room temperature. These fixed cells were used for screening of hybridomas, using an IFA test (see details below). The antigenicity of the recombinant PrP was evaluated by immunizing 7-8-week-old Prnp Ϫ/Ϫ , and Prnp ϩ/ϩ mice with PrP-AcNPV-infected cells.
Scrapie Prions and Animals-The Obihiro strain of scrapie prion (PrP Sc ), which had been passaged in ICR/Slc mice more than 10 times, was prepared from infected brains homogenized in phosphate-buffered saline and intracerebrally inoculated into 3-week-old C57BL/6 (Prnp ϩ/ϩ ) and Prnp Ϫ/Ϫ mice, as described (21). At appropriate time points, the brains were collected and used for PrP Sc extraction and immunoprecipitation. Noninfected brains of Prnp Ϫ/Ϫ and Prnp ϩ/ϩ mice served as controls. For histoblot analysis, we used the Sc237 strain of scrapie prion, which had been passaged through Syrian Golden hamsters more than 10 times. Three-week-old hamsters were intracerebrally inoculated with Sc237 prion, as described (21). PrP Sc extraction from infected animal brains and an enzyme-linked immunosorbent assay (ELISA) were done, as described (21).
Immunohistochemical Analysis-Mouse brains were frozen in liquid N 2 and then cut into 5-m-thick cryosections. The sections were fixed with acetone for 10 min and then reacted with primary antibodies for 30 min. After washing three times in phosphate-buffered saline, the sections were incubated with fluorescein isothiocyanate-conjugated antimouse IgG or IgM. Recombinant baculovirus-infected cells were also examined using the same procedure.
Immunoprecipitation (IP)-The brains were homogenized (10%, w/v) in 0.5% Nonidet P-40, 0.5% sodium deoxycholate, 100 mM NaCl, 10 mM EDTA, in 10 mM Tris-HCl (pH 7.5) and then precleared by centrifugation at 11,000 ϫ g for 30 min at 4°C. The samples were incubated with antibodies at 4°C for 1 h. The antigen-antibody complexes were collected on beads. Immuno Assist MG-PP (Kanto Chemical) and protein G beads (Amersham Pharmacia Biotech) were used for mAbs and Ab.Tg, respectively. After washing three times, the beads were mixed with SDS-sample buffer and boiled for 5 min. The precipitated PrP was detected by WB, using biotinylated Ab.Tg and streptavidin-HRP (Life Technologies, Inc.).
Histoblot Analysis-Histoblot analysis was done as described (25). Briefly, mouse and hamster brains were frozen in liquid N 2 , and 8-mthick cryosections were prepared and placed on glass slides. The glass slides carrying the sections were immediately pressed onto Immobilon-P membranes for 1 min. The membranes were used with one of the following pretreatments: 100 g/ml proteinase K for 1 h at 37°C or hydrated autoclaving pretreatment for 10 min at 115°C. To detect PrP Sc , hydrated autoclaving pretreatment was done following proteinase digestion. The pretreated membranes were incubated with Ab.Tg. Bound antibodies were detected using HRP-conjugated anti-mouse IgG and chemiluminescence, as described for WB.
Epitope Analysis-To determine the epitopes, antibodies were incubated with a gridded array of peptides comprising 122 polypeptides of 13 amino acids, shifted by 2 amino acids and covering the entire mouse sequence. The peptides were covalently attached at COOH termini to a cellulose support, as individual spots (Jerini Biotools, Berlin). Bound antibodies were detected using HRP-conjugated anti-mouse IgG or IgM and chemiluminescence, as described for WB.

RESULTS
Generation of Mice Devoid of PrP C -We generated mice devoid of PrP C by gene targeting. As shown in Fig. 1, the entire open reading frame and 3Ј-end of the second intron were replaced with the pgk-neo gene cassette. The genotype of mice was determined by either Southern blotting or polymerase chain reaction analysis of DNA prepared from tails of the mice. An example of Southern blot analysis of tail DNAs from crosses between heterozygotes is shown in Fig. 1B. Immunoblot analysis of lysates from cerebra showed the absence of PrP C in Prnp Ϫ/Ϫ mice and a reduced level in Prnp ϩ/Ϫ mice compared with findings in Prnp ϩ/ϩ mice (Fig. 1C). Homozygous mutant mice showed no obvious behavioral changes and at a young age appeared to have normal motor activity. However, the aged mice showed tremor and ataxia, as reported (16).
Generation of Polyclonal and Monoclonal Antibodies to PrP-Prnp Ϫ/Ϫ mice were immunized with brain homogenates and thymocytes of Prnp ϩ/ϩ mice. The sera from immunized  (Table I). In contrast, the sera from Prnp ϩ/ϩ mice inoculated with the same antigens showed no reactivities in these assays. Antisera (Ab.Tg) served as anti-pan-PrP polyclonal antibodies throughout this study. We then attempted to generate mAbs, using a conventional B cell hybridoma technique. In initial studies, we wanted to screen the hybridomas by IFA against thymocytes from Prnp ϩ/ϩ mice and by ELISA, using brain homogenates of Prnp ϩ/ϩ mice, but we did not detect candidate clones secreting mAbs reactive to PrP. This failure was attributed to limited concentrations of antigens. Thus, we decided to use recombinant proteins expressed by baculoviruses. The recombinant baculovirus (PrP-AcNPV) carrying murine Prnp open reading frame, under the promoter of the polyhedrin gene, was plaquepurified. SF21AE cells infected with PrP-AcNPV showed expression of recombinant PrP revealed by IFA, using Ab.Tg ( Fig.  2A). WB analysis using Ab.Tg detected an immunoreactive band of the expected size in homogenates of infected SF-21 cells (Fig. 2B). Furthermore, Prnp Ϫ/Ϫ mice immunized with lysates of PrP-AcNPV-infected SF21 cells generated antibodies reactive to PrP, in various forms (Table I). Thus, PrP-AcNPV infected SF-21 cells could serve as an antigen for screening mAbs. Splenocytes from mice immunized with brain homogenates were fused with P3U1 mouse myeloma cells, and the resulting hybridomas were screened by IFA against PrP-AcNPV-infected SF21 cells. Noninfected SF21 cells were served as controls. Seven clones of hybridomas were established.
Characterization of mAbs-All of these mAbs reacted to neither PrP C nor PrP Sc in WB and showed only a weak reaction to PrP Sc fractions, using ELISA. These mAbs were IgMs. Despite evidence that the mAbs did not react with PrP in WB, the immunoreactivity of mAbs was confirmed by immunoprecipitation. Immunoprecipitates from brain lysates with mAbs were size-fractionated, blotted onto a membrane, and probed with Ab.Tg. All mAbs efficiently immunoprecipitated PrP from brain samples of wild-type healthy mice. Two mAbs, 4A3 and 11H1, representing two distinct groups that showed different properties in IFA, as described later, were extensively characterized. Representative data from mAbs 4A3 and 11H1 are shown in Fig. 3A (lane 2). These mAbs reacted weakly to brain homogenates from scrapie-affected mice (Fig. 3A, lanes 3). No signals were detected with Prnp Ϫ/Ϫ brain homogenates (Fig.  3A, lanes 1), thus indicating the PrP specificity of the immunoreactive bands. On the other hand, Ab.Tg detected larger amounts of PrP from the scrapie-infected brains (Fig. 3A, lane  3). These results suggested that PrP C was selectively identified by IP, using these mAbs. However, none of the mAbs showed cell surface staining of wild-type thymocytes (data not shown), but Ab.Tg did show staining. Sequential IP with mAbs and Ab.Tg indicated that the PrP C immunoprecipitated with the mAbs is a subfraction of PrP C immunoprecipitated with Ab.Tg (Fig. 4). On the other hand, sequential immunoprecipitation of 4A3 and 11H1 suggested that the fraction precipitated by these mAbs is probably identical (Fig. 4). We observed similar results with all the mAbs used. These results suggest that these mAbs Prnp ϩ/ϩ thymocyte 64,000 ϩ ϩ Prnp ϩ/ϩ brain homogenate 128,000 ϩ ϩ PrP-AcNPV-infected SF21 cell 64,000 ϩ ϩ a Purified PrP Sc was used as an antigen for ELISA. b Immunoreactivity of antisera to mouse PrP Sc in WB. c Immunoreactivities to thymocytes of Prnp ϩ/ϩ mice and PrP-AcNPVinfected SF21 cells. Tg of PrP from the scrapie-affected mouse brain. Scrapie-affected mice were killed at 0, 9, 12, 15, and 20 weeks postinfection. Brain homogenates were examined with (ϩ) or without (Ϫ) proteinase K (PK) treatment. PK(Ϫ) samples represent the total amounts of PrP C and PrP Sc , while PK(ϩ) samples represent the amounts of PrP Sc . Note the increase of proteinase K-resistant PrP at 20 weeks postinoculation. C, immunoprecipitates from samples same as in B. D, brain homogenates used for B and C were immunoprecipitated with 11H1 mAb. In C, same homogenates were immunoprecipitated with anti-pan-PrP (Ab.Tg) antibody. 11H1-reactive PrP was markedly decreased at 20 weeks postinoculation, while Ab.Tg detected a larger amount of PrP (PrP C and PrP Sc ) even in the same sample. are specific to a subfraction of PrP C .
Dynamics of PrP C to PrP Sc Conversion in the Brain-The findings that the mAbs immunoprecipitated small amounts of PrP from infected brains (Fig. 3A) suggested the exhaustion of PrP C caused by the exponential conversion of PrP C to PrP Sc . Thus, we measured the equilibrium of PrP C and PrP Sc during the pathogenesis. Homogenates from mouse brains at 0, 9, 12, 15, and 20 weeks postinfection, immunoprecipitated with mAbs and Ab.Tg, were probed using the biotinylated Ab.Tg (Fig. 3, C and D). The immunoprecipitates with Ab.Tg might represent the total amount of PrP (Fig. 3C). The total amount of PrP was also assessed by WB without proteinase K pretreatment (Fig.  3B). Proteinase K resistant products (PrP Sc ) accumulated in the brains at terminal stages of illness (Fig. 3B). The PrP Sc was first detected at 15 weeks postinfection and dramatically increased by 20 weeks postinfection (Fig. 3B). Consistent with the WB analysis, the total PrP immunoprecipitated with Ab.Tg increased within 20 weeks postinfection (Fig. 3C). In contrast, the mAbs-related immunoprecipitates were reduced dramatically by 20 weeks postinfection (Fig. 3D). The results were obtained with mAb 11H1 and similar results were seen with mAb 4A3 (data not shown). These results suggested that PrP C is exhausted by conversion to PrP Sc , at the terminal stage of illness.
Spatial Distribution of PrP C and PrP Sc -To gain insight into the regional distribution of PrP C and PrP Sc , we attempted to define conditions to discriminate PrP C and PrP Sc , using histoblot analysis. The pan-PrP-specific Ab.Tg was used as a probe. Without any pretreatment of the membrane, diffuse signals were observed in both scrapie-affected and mock-affected wildtype mice (Fig. 5A, first column from the left). These signals disappeared by proteinase K treatment (Fig. 5A, second column). With autoclave pretreatment, the signal from scrapieaffected mice was enhanced (Fig. 5A, third column), while the signal from mock affected mice was slightly reduced. When the autoclave pretreatment was followed by proteinase K digestion (Fig. 5A, fourth column), the signal from scrapie affected mice was further enhanced. The same treatment completely eliminated the signal from the mock affected mice. No signals were detected in case of Prnp Ϫ/Ϫ mice, regardless of pretreatment conditions. These results indicate that signals on the membrane (without pretreatment) preferentially represent PrP C , while PrP Sc was specifically detected on the membrane exposed to a combination of hydrated autoclaving and proteinase K digestion (Fig. 5A, fourth column). It should be noted that the PrP C signal was reduced in the scrapie-affected sample, particularly in the cerebral cortex (Fig. 5A, first column). To better understand temporal and spatial relationships between amounts of PrP C and PrP Sc , we inoculated scrapie prion Sc237 into the brains of hamsters, which were killed at 38 and 80 days after inoculation, respectively. Mock-infected hamsters were also examined, as controls. As shown in Fig. 5B, PrP C but not PrP Sc was present in normal hamster brains. PrP Sc deposition was first detected in the thalamus at 38 days postinoculation, and the PrP C signal decreased specifically in that region. At the terminal stage of illness, PrP Sc had spread to all brain regions, and the PrP C signal was decreased in those areas (Fig. 5B). The distribution of PrP Sc revealed by the histoblot was consistent with the distribution of PrP Sc plaque deposits detected by immunohistochemistry, using the reference antibody (Ab.Mo-VI) (data not shown). The results from histoblot and IP analyses consistently indicated the loss of PrP C at the terminal stage of illness.
Immunohistochemical Dissociation of Multiple Isoforms-Ab.Tg recognized PrP C in Prnp ϩ/ϩ mice (Fig. 6, panel 2). Homogeneous signals representing PrP C were distributed throughout the entire brain areas except for nuclei. In the scrapie-affected mouse brain, granulous and/or amorphous signals appeared (Fig. 6, panel 1). Seven mAbs were classified into two groups in terms of reactivities to distinct materials in IFA. All mAbs showed homogeneous and weak signals in the normal unaffected mouse brain (Fig. 6, panels 5 and 8, and Table II) in agreement with their reactivities to PrP C by IP. However, for the scrapie-affected samples, four (clones 1H8, 4A3, 6B5, and 9A8) showed stellar structures (Fig. 6, panel 4, and Table II), while the others (7H8, 8G6, and 11H1) showed plaque structures in neuropil (Fig. 6, panel 7, and Table II). Double staining of the former group together with the anti-GFAP antibody indicated that stellar materials were PrPs that had accumulated in astrocytes (data not shown). There were no signals in Prnp Ϫ/Ϫ brain samples (Fig. 6, panels 3, 6, and 9). These results suggested that the mAbs recognized not only PrP C but also abnormal isoforms in the acetone-fixed sections.
Epitope Mapping-A single mAb recognizes PrP C , under a certain condition (in solution) but recognizes abnormal isoforms under another condition (acetone-fixed tissue section). To FIG. 4. Sequential immunoprecipitation of PrP C . Brain homogenates were sequentially immunoprecipitated with 11H1 and 4A3 mAbs and polyclonal Ab.Tg in the order indicated. Ab.Tg precipitated a large amount of PrP C from the samples that had been immunoprecipitated with mAbs. On the other hand, mAbs precipitated a limited amount of PrPC from those samples.
FIG. 5. Regional distribution of PrP C and PrP Sc . A, discrimination of PrP C and PrP Sc by histoblot analysis, using Ab.Tg. The blots of scrapie-affected Prnp ϩ/ϩ , mock-affected Prnp ϩ/ϩ , and Prnp Ϫ/Ϫ mouse brain samples were reacted with Ab.Tg., with or without pretreatment of hydrated autoclaving and proteinase K. PrP C was detected on nonpretreated membranes (far left). PrP Sc was specifically detected by the combination of autoclaving and proteinase K treatments (right-hand column). B, temporal examination of PrP C and PrP Sc in scrapie-affected hamsters, using histoblots. Hamsters were inoculated intracerebrally with scrapie prion Sc237. At postinfection day 38 and 80, brain samples were taken. Mock-infected brain was also included (day 0). Note the decrease of PrP C at sites where PrP Sc was accumulated. elucidate the molecular basis for these observations, we determined the epitopes recognized by mAbs and Ab.Tg, using a gridded array of synthetic peptides consisting of 122 13-residue peptides, sequentially shifted in steps of 2 amino acids and covering the whole mouse PrP sequence (Fig. 7A). Ab.Tg recognized distinct clusters of polypeptides, probably reflecting the conformation of native PrP C used as an immunogen. Four discontinuous regions appeared to constitute dominant epitopes. Interestingly, all mAbs examined recognized multiple discontinuous peptides, as summarized in Fig. 7B. Two mAbs (4A3 and 11H1) as well as Ab.Tg recognized sequences around octarepeat sequences. In the 4A3-1 segment, a pair of tripeptides ((T/G/S/)WG; positions 55-57, 63-65, 71-73, 79 -81, and 87-89) seemed to constitute an epitope. Two perfect repeats (GQPHGGGWG; positions 57-65 and 81-89), but not repeats containing a substituted residue (GQPHGGSWG; positions 65-73 and 73-81), were preferential epitopes recognized by Ab.Tg. mAb 4A3 also recognized a different region (4A3-2, GNDWEDR; positions 141-147), the sequence of which completely overlapped with the epitope of mAb 15B3-1, which is specific for PrP Sc (11). mAbs 11H1 and 8G6 also recognized  169 -179)). In addition to the octarepeat region, Ab.Tg also recognized Ab.Tg-2 (QWNKP; positions 97-101) and Ab.Tg-3 (DWEDRYYRE; 143-151). The position of Ab.Tg-3 was close to that of 15B3-1. To examine the reproducibility of the experimental system, we analyzed sera from four independent immune mice. Ab.Tg-1 and -2 were detected with all sera, while Ab.Tg-3 was detected with three of four immune sera samples. These highly reproducible results indicate the reliability of our experimental system. DISCUSSION To gain insight into epitopes representing PrP C conformation and in vivo conversion processes of PrP C to PrP Sc , we immunized Prnp Ϫ/Ϫ mice with native PrP C , prepared as brain FIG. 6. Indirect fluorescent assay with Ab.Tg and mAbs 4A3 and 11H1 to scrapie-affected Prnp ؉/؉ mice. Scrapie-affected and mock-affected Prnp ϩ/ϩ mouse brains were rapidly frozen and then cryosectioned. Prnp Ϫ/Ϫ mice served as controls. Brain samples were examined using Ab.Tg (panels 1-3), mAb 4A3 (panels 4 -6), and 11H1 (panels 7-9) as primary antibodies. Ab.Tg detected the PrP C throughout the entire brain areas as indicated by smear signals (panel 2). In addition to similar signals, granulous and amorphous signals representing PrP Sc were observed in the scrapie-affected brain (panel 1). mAbs 4A3 and 11H1 also weakly revealed PrP C in the mock-affected brain (panels 5 and 8). These mAbs detected stellar structures and plaques, respectively (panels 4 and 7). No signal was observed in Prnp Ϫ/Ϫ mice, using these antibodies (panels 3, 6, and 9).

TABLE II Characteristics of monoclonal and polyclonal antibodies
Antibody Ab.Tg IgG ϩ ϩ ϩ ϩ ϩ ϩ ϩ ϩ a IP analysis with PrP C and PrP Sc prepared from normal (C) and scrapie-affected (Sc) mouse brains, respectively. b WB analysis with PrP C and PrP Sc prepared from normal (C) and scrapie-affected (Sc) mouse brains, respectively. c For ELISA we used partially purified PrP Sc . d IFA with acetone-fixed cryosection from normal (C) and scrapie-affected (Sc) brains, and PrP-AcNPV-infected SF21AE cells. ϩA, stellar signals in astrocytes; ϩN, amorphous signals in neurophils; ϩ, positive; ϩ/Ϫ, weakly positive; Ϫ, negative.
homogenates or intact thymocytes from wild-type mice. With this attempt, we obtained polyclonal antibody, named Ab.Tg, and seven mAbs. The Ab.Tg was highly reactive to both PrP C and PrP Sc (Table II); therefore, it was used as a pan-PrPspecific antibody. Unlike the Ab.Tg, none of the mAbs detected any isoforms of PrP by WB analysis; they showed only weak reactivities to partially purified PrP Sc samples in ELISA (Table  II). However, all of these mAbs efficiently immunoprecipitated PrP in brain lysates from noninfected mice but not from scrapie-infected mice, indicating preferential affinity to PrP C in solution. We observed that the immunoprecipitable PrP C was markedly decreased in the infected brain by one-fifth to onesixth of the level in noninfected brain, depending on PrP Sc accumulation. At this time point, we did not observe cell loss to this magnitude. Thus, the loss of PrP C probably reflects a reduction of PrP C per cell rather than a loss of cells in diseased brains. Consistent results were also obtained by histoblot analyses. The Ab.Tg detected diffusely distributed PrP C on the histoblot, without any pretreatment (Fig. 5A). In contrast, the same antibody specifically detected PrP Sc on the blot exposed to hydrated autoclaving and partial digestion with proteinase K (Fig. 5A). The time course study revealed that the site of the PrP C -decrease correlated with the site at PrP Sc depositing (Fig.  5B). Although the expression level of mRNA of the Prnp gene is unchanged with scrapie infection (19), our results suggest that the amount of PrP C in infected brains is exhausted by conversion of PrP C to PrP Sc , thus providing a molecular basis for the notion that the loss of function of PrP C may partly account for pathogenesis of the central nervous system (14 -18).
Although all of the mAbs we used showed preferential affinities for PrP C in solution, they were divided into two groups in terms of reactivities to tissue sections. In immunohistochemistry for acetone-fixed scrapie brain, some mAbs (1H8, 4A3, 6B5, and 9A8) recognized a stellar structure (Fig. 6, panel 4), while other mAbs (7H8, 8G6, and 11H1) reacted with plaque structures in neuropil (Fig. 6, panel 7). In mock samples, a weak and diffuse signal was evident. Stellar and plaque structures were never observed (Fig. 6, panels 5 and 6). Thus, these stellar and plaque structures might represent disease-specific isoforms of PrP. mAb 4A3 recognized an isoform that accumulates in as-trocytes (Fig. 6, panel 4). The PrP may be an intermediate form in converting to the fully pathogenic PrP Sc . It has been reported that PrP Sc first accumulates in astrocytes prior to development of neuropathological changes (26) and that the astrocyte-specific induction of PrP in Prnp Ϫ/Ϫ mice restores susceptibility to scrapie (27). These data suggest that astrocytes play an important role in scrapie pathogenesis. The mAb 4A3 may be useful to study the roles of astrocytes in scrapie disease. The mAb 11H1 may recognize the isoform closely related to the fully pathogenic PrP Sc . The immunohistochemical staining pattern of the mAb 11H1 resembled results obtained with rabbit polyclonal antibodies to synthetic peptide or scrapie-associated fibrils in formalin-fixed samples treated with either hydrated autoclaving (21) or guanidine (28). Either guanidine or hydrated autoclaving pretreatment has been found essential for immunofluorescent detection of PrP Sc plaques. The mAbs and Ab.Tg facilitated detecting abnormal isoforms of PrP, with solely acetone fixation. The distinct immunohistochemical staining patterns of two groups of mAbs suggests that conversion from PrP C to PrP Sc involves multiple steps at different sites in vivo.
The mAbs were preferentially reactive to PrP C in IP but were immunohistochemically reactive to abnormal isoforms of PrP. One of the intriguing observations is that all mAbs recognized multiple discontinuous linear peptides that show no apparent similarities (Fig. 7). A similar unusual nature was noted for a PrP Sc -specific clone 15B3 (11). Clone 15B3 recognized three distinct linear polypeptides. At least one of each of the epitopes recognized by three mAbs examined in this study overlapped with one of the 15B3 epitopes. Especially, the segments of 4A3-2 and 11H1-3 completely overlapped with epitopes 15B3-1 and 15B3-2, respectively. Epitope 8G6-2 is closely related to the 15B3-2 epitope. We suggest that these epitopes largely contributed to reactivities of these mAbs to abnormal isoforms in acetone-fixed tissue sections. Epitopes 15B3-1/4A3-2 and 15B3-2/11H1-3/8G6-2 may represent isoforms accumulating in astrocytes and in the neuropil, respectively.
Regarding epitopes specific to PrP C , the octarepeat region is notable. Two of three mAbs, 4A3 and 11H1, reacted to the region of octarepeats. Furthermore, results from polyclonal antibody Ab.Tg suggested that this region is likely to be one of the major epitopes on PrP C . Because native PrP C was used as an immunogen, the epitopes recognized by Ab.Tg may reflect conformation of PrP C . A pair of tripeptides ((T/G/S/)WG; positions 55-57, 63-65, 71-73, 79 -81, and 87-89) seemed to constitute an epitope of mAb 4A3. Ab.Tg recognized two perfect repeats (GQPHGGGWG; positions 57-65 and 81-89) but not other repeats that carried a substitution (GQPHGGSWG; positions 65-73 and 73-81). The substitutions at 71 and 79 residues (Gly to Ser; underlined) affect the immunoreactivity of Ab.Tg. A model suggested that the octarepeat region is constrained by four Cu(II)-coordinating histidines into a compact structure (29). The tripeptides involved in 4A3 reactivity are located at centers of loops generated by copper binding (29). The epitopes for some mAbs generated by DNA-mediated immunization of Prnp Ϫ/Ϫ mice were also located at an octarepeat region (8). These results suggest that the octarepeats are important for antigenicity of PrP C , and copper may contribute to related characteristics. Consistent with this notion, it has been reported that copper can enhance restoration of proteinase K resistance and infectivity after the denaturation of PrP Sc with guanidine hydrochloride (30).
It is noteworthy that the polyclonal antibodies from immunized mutant mice revealed limited regions with regard to being effective epitopes, although the Prnp Ϫ/Ϫ mice were immunologically intolerant of PrP. We examined sera from four independent immune mice and observed reproducible results among the sera against the PrP C immunogen. Epitopes 1 and 2 of Ab.Tg (Ab.Tg-1 and Ab.Tg-2) were always detected, and the third one (Ab.Tg-3) was detected in three of four immune sera, indicating their dominance as epitopes. The NH 2 -terminal sequences between positions 23 and 40 seldom functioned as an epitope, and therefore were unstable. These results suggested that the amino-terminal half of PrP C has a characteristic conformation, in a good agreement with findings with the mAbs we characterized. We also observed distinct epitope patterns on PrP Sc fractions. 2 This experimental strategy may aid in understanding species barrier mechanisms and differences in infectious prion strains replicated in a given animal species.
Data from experiments using these mAbs and polyclonal antibodies we developed revealed a marked decrease of PrP C in brains of animals at the terminal stage of illness, thereby providing a molecular basis for the hypothesis that loss of function of PrP C may have some role in the pathogenesis of prion diseases. Results from epitope mapping and our immunohistochemical studies suggest that conversion of PrP C to PrP Sc may involve multiple steps at different sites.