Direct Evidence of Generation and Accumulation of β-Sheet-rich Prion Protein in Scrapie-infected Neuroblastoma Cells with Human IgG1 Antibody Specific for β-Form Prion Protein*

Background: PrPSc is believed to have a β-sheet-rich structure. Results: Established human IgG1 stained β-form PrP in prion-infected cells but had no inhibitory activity against the propagation of PrPres. Conclusion: β-Sheet-rich PrP was generated and accumulated in cells. Significance: β-Form PrP aggregates play roles in cytotoxicity, whereas prion or PrPSc is responsible for prion propagation. We prepared β-sheet-rich recombinant full-length prion protein (β-form PrP) (Jackson, G. S., Hosszu, L. L., Power, A., Hill, A. F., Kenney, J., Saibil, H., Craven, C. J., Waltho, J. P., Clarke, A. R., and Collinge, J. (1999) Science 283, 1935–1937). Using this β-form PrP and a human single chain Fv-displaying phage library, we have established a human IgG1 antibody specific to β-form but not α-form PrP, PRB7 IgG. When prion-infected ScN2a cells were cultured with PRB7 IgG, they generated and accumulated PRB7-binding granules in the cytoplasm with time, consequently becoming apoptotic cells bearing very large PRB7-bound aggregates. The SAF32 antibody recognizing the N-terminal octarepeat region of full-length PrP stained distinct granules in these cells as determined by confocal laser microscopy observation. When the accumulation of proteinase K-resistant PrP was examined in prion-infected ScN2a cells cultured in the presence of PRB7 IgG or SAF32, it was strongly inhibited by SAF32 but not at all by PRB7 IgG. Thus, we demonstrated direct evidence of the generation and accumulation of β-sheet-rich PrP in ScN2a cells de novo. These results suggest first that PRB7-bound PrP is not responsible for the accumulation of β-form PrP aggregates, which are rather an end product resulting in the triggering of apoptotic cell death, and second that SAF32-bound PrP lacking the PRB7-recognizing β-form may represent so-called PrPSc with prion propagation activity. PRB7 is the first human antibody specific to β-form PrP and has become a powerful tool for the characterization of the biochemical nature of prion and its pathology.

Prion diseases are fatal and transmissible neurodegenerative disorders that include Creutzfeldt-Jakob disease in humans and bovine spongiform encephalopathy in cattle. It has been proposed that a misfolded form of prion protein is responsible for the infectivity of prion disease, and the pathogenesis of prion disease involves a conformational change of prion protein (PrP) 2 from PrP C to PrP Sc (1). PrP C is a monomeric isoform, rich in ␣-helical structure, and sensitive to digestion by proteinase K (PK). In contrast, multimers are PrP Sc characterized as PrP res by enhanced resistance toward PK digestion. Consequently, it is believed first that PrP Sc might have a ␤-pleated sheet structure because all amyloids studied have been found to have this structure and second that this ␤-sheet-rich PrP Sc binds specifically to PrP C , propagating its altered conformation via a templating mechanism, which is a prion activity. In other words, it was demonstrated that PrP Sc , defined as PrP27-30, was generated and accumulated in prion-infected cells or brain. However, there is no direct evidence of whether prion or PrP Sc has a ␤-sheet-rich structure, although there is suggestive evidence of FTIR using centrifugation-purified aggregates of prion-infected brain extracts (2,3).
Antibodies to these proteins are powerful tools to clarify these questions. However, 1) the purification procedure of PrP Sc was based on the centrifugation precipitation of molecular aggregates using PEG or sodium phosphotungstic acid, which does not necessarily guarantee the specificity to a prion (2, 4 -7). 2) The specificity of an anti-PrP Sc antibody was defined by the existence of PrP27-30 resulting from PK-treated prion-infected brain or cell extracts that are immunoprecipitated with tested antibodies. These experiments showed that PrP27-30 was resolved from PK-resistant PrP but do not directly indicate that PrP27-30 is generated from ␤-sheet-rich PrP; i.e. they rather suggest a generation of ␤-sheet-rich PrP by experimental procedures including the use of PK, denaturing agents, or detergents (8). 3) All antibodies established so far are either anti-PrP Sc /PrP C or anti-PrP C antibodies (9). More recently, mouse monoclonal IgG W261, which reacts exclusively with PrP Sc but not PrP C , has been reported by cell fusion technology using spleen cells immunized with sodium phosphotungstic acid-precipitated PrP Sc derived from prion-infected brain extracts (10). This study did not address the question of whether PrP Sc was the ␤-form PrP or not.
In this study, using the conformation-defined recombinant PrPs and a human single chain Fv-displaying phage library, we have established two human IgG, PRB7 and PRB30, which are specific to the ␤-form but not the ␣-form of recombinant PrP of human, bovine, sheep, and mouse. Epitope mapping analysis showed that PRB7 IgG recognized residues 128 -132 of the fulllength prion protein.
When prion-infected ScN2a cells were cultured in the presence of PRB7, apoptotic cells with numerous PRB7 binding signals including large aggregates were gradually generated during 4 days of culture. This finding is the first direct evidence of the generation and accumulation of ␤-sheet-rich prion protein in ScN2a cells. Interestingly, in these apoptotic cells, SAF32-staining granules were distinct from PRB7-binding aggregates, suggesting that SAF32-binding PrP does not have a PRB7-recognizing ␤-sheet structure, whereas PRB7-binding PrP may not have the N-terminal octarepeat region of PrP. After ScN2a cells were cultured in the presence of PRB7 or SAF32 for 3 days, PK-treated cell lysate was immunoblotted by 6D11 to examine the inhibitory effects of PRB7 IgG on the generation/accumulation of PrP Sc . Surprisingly, PRB7 IgG had no influence, whereas SAF32 strongly inhibited the generation/accumulation of PrP res .
Thus, this study reports the first establishment of a human IgG antibody recognizing ␤-form PrP but not ␣-form PrP and the use of this antibody to provide direct evidence of the de novo generation and conversion of ␤-sheet-rich PrP in prion-infected cells. PRB7 IgG can be a powerful tool to purify the ␤-form PrP generated de novo and demonstrate its biochemical basis and significance to elucidate structural evidence of prion infectivity and neurotoxicity.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-The pQE30 vector and Escherichia coli (M15) were obtained from Qiagen. The 6D11 antibody was purchased from Signet. A silver stain II kit was from Wako. Phenylmethylsulfonyl fluoride (PMSF) and anti-␤-tubulin (I) antibody were purchased from Sigma. Recombinant anti-PrP Fab HuM-P, HuM-D18, HuM-D13, HuM-R72, and HuM-R1 were purchased from Inpro Biotechnology. GAHu/ Fab/Bio was purchased from Nordic Immunology Inc. Horseradish peroxidase (HRP)-conjugated anti-goat mouse IgG, alkaline phosphatase-conjugated goat anti-mouse IgG, and goat anti-human IgG Fc-HRP were purchased from Jackson ImmunoResearch Laboratories. HisTrap TM HP, the anti-His tag antibody, and HiTrap TM Protein A HP were from GE Healthcare. The mouse anti-E tag monoclonal antibody, HRP-conjugated anti-E tag mAb, the anti-E tag antibody-Sepharose column, and the anti-M13 monoclonal antibody were purchased from Amersham Biosciences. The 3,3Ј,5,5Јtetramethylbenzidine solution was from Calbiochem. SAF32 was purchased from SPI Bio (Ann Arbor, MI). Control human IgG/ was purchased from Bethyl Laboratories. Alkaline phosphatase-conjugated streptavidin and HRPconjugated streptavidin were purchased from Vector Laboratories. The Dye Terminator Cycle Sequencing FS Ready Reaction kit was from Applied Biosystems. NheI, BstApI, ApaI, BsaWI, BbsI, and HindIII were obtained from New England Biolabs (Ipswich, MA). T4 DNA ligase was obtained from Takara Bio Inc. Opti-MEM, the SuperScript TM First-Strand Synthesis System, pcDNA3.1 TM (Ϫ) mycHisA, pcDNA3.1(Ϫ) Zeo, the FreeStyle TM MAX reagent, Annexin V-Alexa Fluor 568, Alexa Fluor 488-labeled anti-human IgG, Alexa Fluor 546-labeled anti-mouse IgG, and Alexa Fluor 488succinimidyl ester were obtained from Invitrogen. Anti-␤-actin antibody was purchased from Abcam. The labeling of antibody with Alexa Fluor 488 was performed according to the manufacturer's instructions. DAPI was purchased from Cambrex Bio Science.
Expression Vector-PrP expression vectors were constructed as described (11). Briefly, DNA fragments corresponding to human PrP residues 23-231 were independently amplified by polymerase chain reaction (PCR) using forward primer (5Ј-gcggatccaagaagcgcccgaagcctgga-3Ј) and back primer (5Ј-ccaagcttctatcagctcgatcctctctggtaata-3Ј) from genomic DNA of HEK293T cells. The fragments were digested with BamHI and HindIII and inserted into a pQE30 vector bearing the His tag sequence. Bovine PrP , mouse PrP(23-231), or sheep PrP(25-234) expression vectors were constructed in a similar way.
Protein Expression and Purification-Constructed histidinetagged recombinant full-length prion protein (rPrP; amino acids 23-231) was expressed in E. coli (M15) and purified by HisTrap HP as described (8). Briefly, fractions containing human rPrP were diluted to a final concentration of 1 mg/ml using a 0.1 M Tris buffer (pH 8.0) containing 9 M urea and 100 mM ␤-mercaptoethanol. The dialyzed solution was diluted 3-fold with 0.1% (v/v) trifluoroacetic acid in water, loaded on a 10 ϫ 150-mm WP300 C4 HPLC column (GL Sciences, Tokyo, Japan), and eluted using a gradient of 0.1% (v/v) trifluoroacetic acid in acetonitrile. Fractions containing oxidized PrP were lyophilized and stored at Ϫ20°C as described (8).
Refolding-Refolding of rPrP was performed according to the methods of Jackson et al. (12). Briefly, to prepare ␣-form PrP, a stock solution of rPrP (1 mg/ml in 6 M GdnHCl) was dialyzed into a 10 mM sodium acetate, 10 mM Tris acetate buffer (pH 8.0). In the case of ␤-form PrP, rPrP was dissolved in a 100 mM dithiothreitol (DTT), 6 M GdnHCl, 10 mM Tris acetate (pH 8.0) buffer for 16 h and then dialyzed with a 10 mM sodium acetate, 1 mM DTT (pH 4.0) buffer. A half-volume of the dialysis buffer was replaced with fresh buffer, and this procedure was repeated every 12 h for a total of 10 times. Refolded proteins were stored at 4°C. The PrP concentration was determined using a spectrophotometer, Ultrospec 2000 (Amersham Biosciences), using a molar extinction coefficient E 280 of 56,650 M Ϫ1 cm Ϫ1 .
Circular Dichroism Spectroscopy-Circular dichroism (CD) spectra were recorded in a 0.1-cm cuvette with a J-820 spectrometer (Jasco) scanning at 20 nm/min with a bandwidth of 1 nm and data spacing of 0.5 nm. Each spectrum represents the average of five individual scans after subtracting the background spectra (8).
Atomic Force Microscopy-Atomic force microscopy images were recorded on an SPI-3800 (Seiko Instruments, Chiba, Japan) immediately after dilution of the 5.4 mg/ml solution of the ␤-form PrP and displaying on a fresh mica surface.
PK Digestion-Recombinant PrP (100 g/ml) was treated with the indicated concentration of PK for 1 h at 37°C. ScN2a cell lysates were treated with PK (final concentration, 20 g/ml) for 30 min at 37°C, then PMSF at a final concentration of 2 mM was added, and the mixture was centrifuged at 15,000 rpm for 45 min at 4°C. The pellet was dissolved in a 1ϫ SDS-PAGE sample buffer and boiled for 10 min for immunoblotting analysis.
Immunoblotting-SDS-PAGE was performed as described (13,14). Purified PRB7 IgG was separated by 10% SDS-PAGE, and the gel was stained with the silver stain II kit.
Immunoblotting was performed as described (13,15). Briefly, samples were added to a 2ϫ sample buffer, boiled for 10 min, separated on a 16.5% Tricine-polyacrylamide gel, and electrotransferred onto a PVDF membrane at 150 V for 90 min using a semidry electroblotter (Sartorius, Tokyo, Japan). The membrane was blocked with 5% skim milk in PBS overnight. Specific binding of the antibody to proteins was determined by incubating the membrane for at least 1 h with the 6D11 antibody (1:4000). HRP-conjugated goat anti-mouse IgG served as a secondary antibody (1:5000). Protein signals were visualized using ECL on a LAS-1000 image analyzer (Fuji Film, Tokyo, Japan). The Tricine gel was stained with the silver stain II kit.
Phage Libraries-The human single chain Fv (scFv)-displaying M13 phage library constructed using the pCANTAB 5E (4.5-kb) phagemid vector was used in this study as described (13,15). The peptide-displaying phage libraries (Ph.D.-12) were purchased from New England Biolabs. The Ph.D.-12 library contains linear peptides composed of 12 random amino acids.
Biopanning-Biopanning was performed as described (13)(14)(15). To isolate the antibody phage specific to ␤-form PrP, the scFv phage library (1 ϫ 10 12 pfu/50 l) was preabsorbed with ␣-form PrP (5 g) coating an immunotube under the conditions of a folding buffer (pH 8.0). One hour later, unbound phages were collected and added to a ␤-form PrP (5 g)-coated immunotube under the conditions of a folding buffer (pH 4.0). After 1 h of incubation at room temperature, the unbound phages were discarded by washing the immunotubes seven times with folding buffer (pH 4) containing 0.1% Tween 20. The bound phages were then eluted with 0.1 M glycine-HCl (pH 2.2), immediately neutralized with 1 M Tris-HCl (pH 9.1), and amplified by infection with log phase E. coli strain TG1 cells. The phages and soluble scFv were prepared as described (13)(14)(15). To isolate peptide phage to PRB7 IgG, unrelated control human IgG/ and PRB7 IgG were separately immobilized on polystyrene 96-well microplates (300 ng/50 l/well). The Ph.D.-12 phage library (4 ϫ 10 11 transforming units) was preabsorbed on a plate immobilized by unrelated control human IgG/. The phages were incubated at room temperature for 1 h with immobilized PRB7 IgG, and the binding phages were eluted with 0.1 M glycine-HCl (pH 2.2). The eluate was immediately neutralized with 1.0 M Tris-HCl (pH 9.1) and amplified by infecting ER2738.
To perform the epitope mapping of PRB7 IgG, a microtiter plate was coated with 50 ng/50 l/well PRB7 IgG or control IgG/ for 1 h at 4°C. After blocking the plates with 0.25% BSA, PBS, Ph.D.-12 phage clones (50 l/well of 1 ϫ 10 11 virions/ml) were added to each well followed by incubation with a biotinylated anti-M13 monoclonal antibody (1:2000) in combination with HRP-conjugated streptavidin (1:2000).
For the construction of the PRB7 IGHC1 expression vector, the Vhls fragment and the Vh fragment were digested with BstApI. These two fragments were ligated with T4 DNA ligase. The ligated sample was amplified by PCR using Vhls forward primer 2 (5Ј-AAAAATCTAGAGCTAGCGATGGCATG-3Ј) and the Vh back primer. This amplified fragment and the IGHC1 fragment were digested with ApaI, ligated, and amplified by PCR using Vhls forward primer 2 and the IGHC1 back primer. This amplified fragment was digested with NheI and HindIII and inserted into a pcDNA3.1(Ϫ) mycHisA vector.
For the construction of the PRB7 IGHC1 expression vector, the Vlls fragment and the Vl fragment were digested with BsaWI, ligated with T4 DNA ligase, and amplified by PCR using the Vlls forward primer and the Vl back primer. This amplified fragment and the IGKC fragment were digested with BbsI, ligated, and amplified by PCR using the Vlls forward primer and the IGKC back primer. The amplified fragment was digested with NheI and HindIII and inserted into a pcDNA3.1(Ϫ) Zeo vector.
Protein Expression and Purification of PRB7 IgG-FreeStyle 293F cells (Invitrogen; 1 ϫ 10 6 cells/ml) were co-transfected with an equimolar mixture (final total DNA concentration, 1 g/ml) of a heavy-chain expression vector and a light-chain expression vector with the FreeStyle MAX reagent (final concentration, 1 l/ml) according to the manufacturer's instructions (23). Transfections were performed in 500-l cultures as described above. One week later, the supernatant was harvested from the culture by centrifugation. PRB7 IgG was purified using HiTrap Protein A HP according to the manufacturer's instructions (13).
Computer Simulation of PrPs-The computer simulation was performed on the basis of the NMR structure of the human PrP (amino acids 124 -227; Protein Data Bank code 1QM0) (24) using PyMOL software. Human IgG was simulated by use of Molecular Operating Environment TM (MOE, Version 2009.10) (supplemental Fig. S3).
Cell culture-ScN2a cells, N2a58 cells overexpressing mouse PrP (a genotype with codons 108L and 189T), FF32 cells, or N2aL1 cells were used. FF32 cells were cloned from F3 cells (an N2a58 cell line infected with a prion Fukuoka-1 strain derived from mouse brain infected with a brain of a Gerstmann-Straüssler-Scheinker syndrome patient) or N2aL1 cells (an N2a58 cell line infected with the 22L strain derived from mouse brain infected with scrapie) as described previously (25,26). All cells were cultured in Opti-MEM containing 10% fetal bovine serum and penicillin-streptomycin at 37°C in 5% CO 2 . Cells (5 ϫ 10 4 cells/well in a 24-well plate) were cultured in the ␤-Sheet-rich Prion Protein Aggregates in ScN2a Cells absence or presence of varying concentrations of PRB7 IgG, SAF32, control human IgG/ antibodies (azide-free), Alexa Fluor 488-PRB7 IgG, or Alexa Fluor 488-control human IgG/ antibodies (azide-free) at 37°C on 24-well microplates (Iwaki Glass Co.) (27).
Immunostaining-Immunostaining was performed as described (28 -30). Briefly, cells were cultured in an Advanced TC TM glass bottom cell culture dish (Greiner Bio-One, Germany). Three days later, cells were fixed with 4% paraformaldehyde for 20 min at room temperature and then washed with PBS. In the case of staining of ␤-actin or ␤-tubulin, cells were fixed with cold methanol for 30 min at Ϫ20°C. Fixed cells were permeated with 0.2% saponin, PBS for 20 min; washed with PBS; and blocked with 3% BSA, PBS for 30 min at room temperature. These cells were incubated with SAF32 (5 g/ml) for 1 h at room temperature in 1.5% BSA, 0. Confocal Microscopy Analysis-Fluorescence images were acquired using a Zeiss LSM 700 confocal laser microscope (Carl Zeiss, Tokyo, Japan). Orthogonal projections were generated from z-stacks, and profile projections were generated using Zeiss LSM software (Carl Zeiss).

Characterization of Recombinant Human Full-length PrP-
PrP was prepared using recombinant prion protein as described (8). CD analysis showed the typical characteristics of the ␣-form of PrP with a structure rich in ␣-helical content and of the ␤-form with a ␤-sheet rich structure, respectively (Fig. 1A) (31). Atomic force microscopy of the ␤-form of PrP displayed a round shaped particle, showing the formation of globular structures with an apparent diameter (horizontal) of ϳ47 nm and an apparent height of ϳ1.6 nm (Fig. 1B). ␣-Form PrP was digested completely with 2 g/ml PK, whereas ␤-form PrP showed resistant fragments at 8 g/ml PK treatment (Fig. 1C). D18 antibody recognizes the epitope 133-157 of ␣-form PrP but not ␤-form PrP (17)(18)(19). In accordance with this finding, D18 did not bind to this ␤-form PrP. D13 (epitope 96 -106 of PrP), R72 (epitope 152-163 of PrP), or the P antibody (epitope 96 -105 of PrP) did not distinguish the structural difference between the ␣-form and the ␤-form of PrP (Fig. 1D) (17)(18)(19).
Biopanning against ␤-Form of PrP-The immunotube for biopanning was prepared by coating the tube with ␤-form PrP (5 g) dissolved in pH 4.0 solution. The phage solution (10 12 pfu/50 l) preabsorbed to ␣-form PrP was added to the ␤-form PrP-coated immunotube under the condition of a folding buffer (pH 4.0). After two rounds of biopanning selection, ELISA was performed under the pH 8.0 condition to determine the binding activity to ␣-form PrP because ␣-form PrP was folded under the pH 8.0 condition. On the other hand, the binding activity to ␤-form PrP was examined under the pH 4.0 condition. A total of 384 clones were isolated and tested for binding activity to antigen by ELISA. Two scFv phage clones, designated as PRB7 and PRB30, were selected to be specific for the ␤-form of PrP without binding activity to the ␣-form of PrP ( Fig. 2A).
PBR7 and PBR30 scFv were then purified by affinity column and tested for their binding specificity using a pH 7.3 buffer (Fig. 2B). Both scFv clearly showed marked binding activity to the ␤-form but not the ␣-form of PrP or an unrelated negative control protein, His-tagged human TIM-3.
It is possible that the pH of the experimental condition influences the conformation of PrPs or scFv, resulting in gain or loss of the binding activity. Therefore, to examine more carefully the effects of the pH condition on PrP/scFv binding, after plates had been coated with folded PrP at the corresponding pH, ELISA was performed for 1 h under either pH 4.0 or 8.0. The almost identical results indicated that the binding of PRB7 and PRB30 was specific to ␤-form but not ␣-form PrP (supplemental Fig. S4). In concert with the conformation specificity of these clones, they showed no binding activity to the SDS-denatured form of full-length PrP, indicating that both scFv are specific to the conformational structures of ␤-form PrP (data not shown). It was confirmed that PRB7 reacted with the ␤-form PrP in native PAGE (supplemental Fig. S5).
Gene Usage of PRB7 and PRB30-The amino acid sequence of scFv clones was deduced from the DNA sequence. The complementarity-determining regions (CDR1-CDR3) and the frame regions (FR1-FR4) were assigned by searching the IMGT/V-QUEST database. PRB7 and PRB30 have unique sequences in both Vh and Vl genes, respectively (Fig. 3C).
PRB7 and PRB30 Recognize Conformation of ␤-Form PrP Monomer and Oligomer but Not That of ␣-Form PrP-PrP was fractionated by size exclusion chromatography. The ␣-form of PrP was eluted as a single peak (␣-form PrP fraction 1) at an elution time of 11.8 min in a running buffer at pH 5.0 (20 mM sodium acetate, 200 mM sodium chloride, 1 M urea, 0.02% azide). For the ␤-form of PrP, the sample was resolved as two peaks at an elution time of 7.5 (␤-form PrP fraction 1) and 12.6 min (␤-form PrP fraction 2) in a running buffer at pH 4.0 (20 mM sodium acetate, 200 mM sodium chloride, 1 M urea, 0.02% azide) (Fig. 4A). ␣-Form PrP fraction 1 exhibited a typical ␣-helical spectrum, whereas ␤-form PrP fraction 1 and ␤-form PrP fraction 2 exhibited a typical ␤-sheet structure in CD analysis (Fig. 4B).
To test the binding activity of these fractions to PRB7 and PRB30, each fraction was immobilized on microtiter plates. As shown in Fig. 4C, PRB7 or PRB30 bound to ␤-form PrP fraction 1 as well as ␤-form PrP fraction 2. Both scFv showed no binding to ␣-form PrP, amyloid ␤42 monomer peptide, IL-18, or BSA. In contrast, the authentic mouse anti-PrP monoclonal antibody SAF32 bound to every PrP fraction irrespectively of their conformations. Its binding epitope is located at the N-terminal octarepeat region of the ␣-form of PrP (27). Unrelated scFv (RE51) gave no binding reactions to any coated proteins.
PRB7 IgG and Its Binding Specificity-IgG form of PBR7 scFv was constructed to endow PRB7 scFv with molecular stability. PRB7 IgG purified with a Protein A column was analyzed by APRIL 20, 2012 • VOLUME 287 • NUMBER 17

JOURNAL OF BIOLOGICAL CHEMISTRY 14027
SDS-PAGE, which indicated that the IgG form associated with a ␥and -chain with a disulfide bond (Fig. 5A). The PRB7 IgG showed an identical binding specificity to PRB7 scFv with specific binding to ␤-form PrP but not ␣-form PrP, whereas SAF32 bound equally to ␤-form PrP and ␣-form PrP (Fig. 5B). Furthermore, PRB7 IgG showed no binding activity to A␤42 soluble form, prefibril oligomers, or fibril forms (Fig. 5B).
The amino acid sequence homology search using ClustalW version 3.1 indicated that the pepPRB7-14 peptide motif had weak homology to 124 -136, particularly 128 -132 of the human fulllength prion protein, suggesting that PRB7 IgG recognizes this epitope (Fig. 6B). The 124 -136 region is depicted in red in the PyMOL display of the NMR structure of the ␣-form of PrP (24).  (12). The CD spectrum of each PrP preparation is shown in Fig. 7A. When the binding activity of PRB7 IgG to these PrPs was examined by ELISA, PRB7 IgG exclusively bound to the ␤-form of PrP but not the ␣-form of PrP (Fig. 7B, left  panel). This reactivity is shared by human, bovine, sheep, and mouse PrP. SAF32 strongly recognized both forms of every species without any discrimination (Fig. 7B, right panel).

␤-Form PrPs Are Generated and Accumulate de Novo in
Prion-infected ScN2a Cells-PRB7 recognizes the fine structure of the conformation of PrP and does not stain ScN2a cells fixed with paraformaldehyde. Furthermore, PRB7 IgG hardly stained

␤-Sheet-rich Prion Protein Aggregates in ScN2a Cells
cell surfaces of non-fixed ScN2a cells, suggesting scarce existence of ␤-form PrP on the cell surface (supplemental Fig. S6). Therefore, to examine the generation of ␤-form PrP in prioninfected cells, prion-infected ScN2a cells were cultured in the presence of PRB7 IgG or SAF32. Three days later, cells were fixed with paraformaldehyde. After permeation with saponin, cells were stained with Alexa Fluor 488-anti-human IgG to detect the complex of PRB7 IgG-PrP or SAF32-PrP (Fig. 8A). Control cells, N2a58 non-infected cells, or ScN2a cells in the presence of normal IgG were not stained with this protocol. As the binding of normal IgG to N2a58 cells or ScN2a cells was not observed by confocal microscopy or flow cytometry analysis, N2a58 cells or ScN2a cells were found to be Fc␥ receptor-negative. In contrast, PRB7 IgG or SAF32 stained the cytoplasm of ScN2a but not N2a58 in a distinct image. Three staining patterns of cytoplasm were observed in ScN2a incubated with PRB7 IgG, i.e. cells with varying numbers of small granules, cells with very large aggregates in the cytoplasm, and cells with no staining. SAF32 stained cell surface PrP of ScN2a and N2a58 cells. When detection sensitivity of confocal laser microscopy was increased, SAF32 was detected by Alexa Fluor 546-labeled anti-mouse IgG in the cytoplasm region below the cell surface FIGURE 5. Characterization of PRB7 IgG antibody. A, SDS-PAGE of PRB7 IgG. Protein A column-purified PRB7 IgG (300 ng) was separated by 10% SDS-PAGE under reducing or non-reducing condition and detected by silver staining. B, PrP conformers (50 ng/50 l/well) were detected by PRB7 IgG or SAF32 (0.5 g/ml). A␤42 conformers (50 ng/50 l/well) were tested for binding reactivity to PRB7 IgG or 6E10 (mouse anti-A␤42 antibody) as described under "Experimental Procedures." 2ME, 2-mercaptoethanol. Data represent the mean values ϮS.D. membrane of ScN2a. When these cells were fixed with methanol and stained with anti-␤-actin or anti-␤-tubulin antibody, the PRB7 IgG-bound aggregates were evident in ScN2a cells cultured with PRB7 IgG (Fig. 8B). The details of the PRB7staining features of ScN2a cells cultured with PRB7 IgG are investigated in Fig. 9. About 4000 cells were classified on the basis of the number of stained granules. Cells without granules amounted to 71%, and those with one granule amounted to 16%. Cells with more than 10 granules amounted to 0.6%. Among these cells, 0.4% of cells with very large aggregates in the cytoplasm were particularly conspicuous. Although ScN2a cells with enormous aggregates in the entire cytoplasm were stained with anti-␤-actin (Fig. 8B), these cells were analyzed for the existence of nuclei over the perpendicular axis using serial images taken by confocal microscopy, and traces of nuclei were detected by DAPI (Fig. 10A, white arrow). Annexin V weakly

␤-Sheet-rich Prion Protein Aggregates in ScN2a Cells
stained these cells with enormous aggregates, indicating that these cells had undergone apoptosis, whereas Annexin V did not stain cells that were strongly DAPI-positive (Fig. 10A). When cells with enormous aggregates were stained with SAF32 in addition to PRB7, however, this staining image observed from the perpendicular axis was not completely merged in PRB7-and SAF32-staining PrP (Fig. 10B, white arrow). A magnified image of the B region of Fig. 10B indicated that PRB7 IgG stained very large tangled fibrillary aggregates, whereas SAF32 appeared to independently stain distinctive tiny granules dispersed in the entire cytoplasm (Fig. 10C, right panel). Similarly, in the case of ScN2a cells with clear DAPI-positive nuclei and numerous small PRB7 IgG-bound granules, SAF32 stained another part of the cytoplasm distinct from PRB7 IgG (Fig. 10C,  left panel). These cells with a number of tiny PRB7-positive granules were clearly stained with Annexin V, indicating that they had undergone apoptosis (Fig. 10D). These results suggested that PRB7 IgG-bound granules were antibody complexes with N-terminal region-deleted PrP. To confirm whether PRB7 IgG-bound complexes were composed of PrP, ScN2a cells cultured with Alexa Fluor 488-labeled PRB7 IgG were denatured with 6 M GdnHCl and stained with an another anti-PrP antibody, 6D11, which recognizes the epitope located at 93-109 of PrP (32). PrP is sandwiched with PRB7 IgG and 6D11 in ELISA (supplemental Fig. S7). As shown in Fig. 10E, SAF32 or 6D11 did not visualize tiny aggregates in the cytoplasm of non-denatured ScN2a cells, whereas PRB7 IgG stained large aggregates in the same cells (Fig. 10E, panels 1 and 2). In the case of GdnHCl-denatured ScN2a cells, SAF32 or 6D11 stained numerous tiny aggregates in the cytoplasm (Fig. 10E,   FIGURE 7. Binding specificity of PRB7 IgG by ELISA analysis. A, CD spectra of ␣-form PrP and ␤-form PrP. Recombinant PrP was refolded as described under "Experimental Procedures." B, ELISA of PRB7 IgG antibodies binding to PrP. Left panel, PRB7 IgG (ranging from 0 to 1 g/ml) was incubated with PrP-coated plates (100 ng/50 l/well). Normal human IgG was used as a control. The secondary antibody was HRP-conjugated anti-human IgG (1:5000 dilution). Right panel, SAF32 (0.1 g/ml) equally binds to both ␣-form PrP and ␤-form PrP (100 ng/50 l/well). The background was determined using HRP-conjugated anti-human IgG alone. Data represent the mean values ϮS.D. deg, degrees. APRIL 20, 2012 • VOLUME 287 • NUMBER 17 panels 3 and 5). PRB7 IgG showed a distinct image in the same cells (Fig. 10E, panel 4) where two kinds of staining images were observed, i.e. non-merged and merged aggregates with PRB7 IgG and 6D11 (Fig. 10E, panels 6 and 8). Fig. 10E, panel 7, shows a three-dimensional image of Fig. 10E, panel 6. These results suggested that there were not only PRB7 IgG-bound aggregates that had lost SAF32 and 6D11 epitopes but also 6D11 epitoperetaining aggregates bound with PRB7 IgG.

␤-Sheet-rich Prion Protein Aggregates in ScN2a Cells
PRB7 IgG Showed No Inhibitory Activity on Accumulation of PrP res in ScN2a Cells-ScN2a cells were cultured in the absence or presence of PRB7 IgG or SAF32. Four days later, cell extracts were prepared in the presence of varying concentrations of PK. The resultant PrP res was evaluated by immunoblotting with 6D11. As shown in Fig. 11, SAF32 strongly inhibited the accumulation of PrP res , indicating the internalization of SAF32 antibody in the cytoplasm after binding to cell surface PrP (Fig. 8).
In contrast to the strong inhibitory activity of SAF32, PRB7 IgG showed no effects on the accumulation of PrP res , similar to the effects of control IgG. These results indicated that SAF32 but not PRB7 IgG blocked prion or PrP Sc , preventing the generation of ␤-form PrP.

DISCUSSION
A number of studies showed that prions generate and accumulate PrP Sc , which shows resistance to PK degradation. It is postulated that PrP Sc converts PrP C to PrP Sc by its templating activity. PrP Sc was purified as a multimer or aggregates. The protease-resistant core of PrP Sc , designated PrP27-30, polymerizes into an amyloid. Many purified amyloids have been found to have a ␤-pleated sheet structure. From these findings, it is believed that prion or PrP Sc may have a ␤-sheet-rich structure (3,5).
Is PrP Sc a ␤-form PrP? Antibodies should be a powerful tool to solve this question. Despite the insufficient biochemical characterization of PrP Sc , many attempts have been made to establish antibodies with fine specificity using mice immunized FIGURE 8. ␤-Form PrPs are generated and accumulate de novo in prion-infected ScN2a cells. A, ScN2a or N2a58 cells were cultured in the absence or presence of PRB7 IgG, SAF32, control human IgG/, or control mouse IgG (5 g/ml) in 10% FBS, Opti-MEM. Three days later, cells were fixed with paraformaldehyde and permeated for staining with Alexa Fluor 488-labeled anti-human IgG (green), Alexa Fluor 546-labeled anti-mouse IgG (red), or DAPI (blue) as described under "Experimental Procedures." Each box indicates a representative staining pattern. Box 1, Box 2, and Box 3 show Large/many granules, a few tiny granules, or no granules stained with PRB7 IgG, respectively. B, ScN2a cells cultured in the presence of PRB7 IgG or control human IgG/ (hIgG) (green) as described in A were fixed with methanol and stained with anti-␤-actin antibody (red), anti-␤-tubulin antibody (red), or DAPI (blue) as described under "Experimental Procedures."
In this study, using conformation-defined recombinant ␤-form PrP and an antibody-displaying phage library, we established a ␤-form PrP-specific human IgG1 antibody, PRB7 IgG. This antibody does not recognize generic oligomer or aggregate forms of unrelated proteins in accordance with the binding activity of PRB7 or PRB30 scFv. PRB7 IgG is the first human IgG antibody specifically binding to ␤-form PrP monomer and oligomers but not ␣-form PrP. It is suggested that PRB7 IgG recog-nizes the epitope 128 -132 of ␤-form human PrP. PRB7 IgG is equally cross-reactive to the ␤-form of bovine, sheep, and mouse PrP.
As PRB7 IgG does not recognize paraformaldehyde-fixed or denatured PrP molecules, we simply observed the ScN2a cells cultured with PRB7 IgG. We have demonstrated here that prion-infected ScN2a cells specifically internalized PRB7 IgG and accumulated the PrP granules bound with PRB7 IgG in the cytoplasm (Fig. 8). This is the first direct demonstration that ␤-form PrP was generated and accumulated in prion-infected cells in the physiological condition. Because neither ScN2a nor N2a58 cells were stained with normal IgG as assessed by confocal microscopy or flow cytometry, these cells were Fc␥ receptor-negative. Nevertheless, ScN2a cells internalized PRB7 IgG, suggesting that complexes of ␤-form PrP targeted by PRB7 IgG FIGURE 9. Distribution pattern of PRB7 IgG-positive granules in ScN2a cells cultured with PRB7 IgG. Cells were stained with PRB7 IgG or DAPI as described in Fig. 8. A total of 4203 cells were classified by the number of granules in the cytoplasm (positive cells/total cells observed). DIC, differential interference contrast.
localizing on the cell surface were endocytosed into the cytoplasm (29). If this ␤-form PrP corresponds to PrP Sc , this observation is in accordance with the finding that PrP Sc was generated on a lipid raft on the cell surface (43). When ScN2a cells were cultured with PRB7 IgG, they gradually generated an increasing number of tiny PRB7-binding granules in the cytoplasm. Among them, about 0.4% of cells had enormous aggregates in the cytoplasm. Vertically severed images of these cells show traces of nuclei in cells that were weakly stained with Annexin V, indicating that they had undergone apoptosis (Fig.  10A). These findings suggested that an enormous aggregate resulted from the fusion of many tiny ␤-form granules primarily endocytosed from the cell surface but was not the result of ␤-form PrP propagation in the cytoplasm.
SAF32 bound tiny granules dispersed in the cytoplasm that were distinct from large aggregates stained with PRB7 IgG (Fig.  10C). These results suggested first that PRB7-positive ␤-form PrPs lost or hid the N-terminal octarepeat region to which SAF32 bound and second that granules stained with SAF32 were not generated from ␤-form PrP.
It was suggested that PrP C -PrP Sc conversion, which is physiologically prevented by an energy barrier, may be a spontaneous stochastic event favored by mutations in the PRNP gene or acquired by infection with exogenous PrP Sc . Accordingly, it is conceivable that GdnHCl treatment of cells additively induces the conformational conversion of PrP, leading to the formation of aggregates in experiments that immunohistologically attempted to show granules of PrP Sc in the cytoplasm (8). It is noted that when ScN2a cells were stained with SAF32 or 6D11 with a denaturing pretreatment by 6 M GdnHCl numerous tiny PrP granules became visible in the cytoplasm, and their images were quite distinct from those of PRB7-staining aggregates. Our results indicated that apoptotic cells showed two distinct types of PrP granules in the cytoplasm: one was SAF32-negative and PRB7 IgG-positive granules (N-terminal region-deleted; ␤-form PrP), and the other was PRB7 IgG-negative and SAF32positive granules (full-length PrP lacking a ␤-sheet-rich conformation) (Fig. 10E). Related studies have reported that PK-sensitive PrP Sc was contained in prion aggregates (44) and more recently that there is heterogeneity in PrP Sc conformers including small soluble PrP aggregates (6). Regarding the requirement of full-length PrP for prion activity, the answer is not consistent: i.e. Prusiner et al. (50) reported that the purified PK-digested PrP27-30 was not infectious, whereas Anaya et al. (45) have recently reported that small infectious PrP res aggregates were recovered in the absence of strong in vitro denaturing treatments from prion-infected cultured cells. On the other hand, using an anti-aggregated PrP IgM antibody, 15B3, Biasini et al. (6,47) reported the successful purification of pathological fulllength prion aggregates, and Cronier et al. (46) found that PKsensitive PrP Sc was involved in prion aggregates. Our findings in Fig. 10 suggest that PrP composed of tiny aggregates visualized with SAF32 or 6D11 after 6 M GdnHCl treatment of cells may be a full-length PrP with ␣-conformation.
To examine the influence of PRB7 IgG on the templating activity of ␤-form PrP to accumulate PrP Sc , ScN2a cells were cultured in the presence of PRB7 IgG or SAF32, and their PKtreated cell lysates were immunoblotted with 6D11 to evaluate the amount of PrP res . Surprisingly, PRB7 IgG had no effect, FIGURE 10. ScN2a cells exhibit PRB7 IgG-positive aggregates under apoptosis. Cells were stained with PRB7 IgG (green), SAF32 (red), or DAPI (blue) as described in Fig. 8. A, ScN2a cells with very large aggregates in the cytoplasm underwent apoptosis. ScN2a cells cultured with PRB7 IgG were incubated with Annexin V (red) followed by fixing, permeation, and staining with anti-human IgG or DAPI as described under "Experimental Procedures." Orthogonal projections of serial confocal sections are shown by histograms alongside one z-section taken from the middle of the cell (as indicated by the red guidelines). White bar A or B indicates the range of the guideline. The white arrow in the B range indicates the pattern of a huge aggregate-bearing ScN2a cell. Scale bar, 20 m. B, very large aggregates in the cytoplasm are PrP. ScN2a cells cultured with PRB7 IgG were fixed, permeated, and stained with SAF32 followed by secondary antibody staining with Alexa Fluor 488-labeled anti-human IgG or Alexa Fluor 546-labeled anti-mouse IgG. Orthogonal projections of serial confocal sections are shown by histograms alongside one z-section taken from the middle of the cell (as indicated by the red guidelines). White bar A or B indicates the range of the guideline. The white arrow in the B range indicates the pattern of a huge aggregate-bearing ScN2a cell. C, PRB7 IgG-positive images do not merge with SAF32-positive images. Left image, prion-infected ScN2a cells containing a number of small PRB7 IgG-bound granules. SAF32 stained the entire region of cytoplasm except PRB7 IgG-bound granules. Right image, prion-infected ScN2a cells containing large tangled fibrillary aggregates stained with PRB7 IgG. A number of tiny granules were stained with SAF32 independently of PRB7 IgG-positive aggregates. D, ScN2a cells with a number of PRB7 IgG-positive granules around nuclei underwent apoptosis. ScN2a cells cultured with PRB7 IgG were incubated with Annexin V followed by fixing, permeation, and staining with anti-human IgG or DAPI as described in A. E, ScN2a cells cultured with Alexa Fluor 488-PRB7 IgG were fixed, permeated, treated with (ϩ) or without (Ϫ) 6 M GdnHCl, and stained with 6D11 (epitope 93-109 of PrP (32)). Panel 1 is an image merged with DAPI and SAF32 staining of non-treated ScN2a cells. Panel 2 is an image merged with DAPI, PRB7 IgG, and 6D11 staining of non-treated ScN2a cells. Panels 3-8 show the images of 6 M GdnHCl-treated ScN2a cells. Panels 4 -7 show images of an identical cell. Panel 3 is an image merged with DAPI and SAF32 staining. Panel 4 is an image stained with Alexa Fluor 488-PRB7 IgG. Panel 5 is an image stained with 6D11. Panel 6 is an image merged with PRB7 IgG and 6D11 staining. Panel 7 is a three-dimensional image of panel 6. Panel 8 shows a cell containing heterogeneous PRB7 IgG-bound aggregates merged with (white arrow) or without (white arrowhead) 6D11 staining in the presence of numerous tiny aggregates stained with 6D11 in the cytoplasm. Staining was performed as described under "Experimental Procedures." DIC, differential interference contrast. FIGURE 11. Effect of PRB7 IgG on accumulation of PrP res . ScN2a cells were cultured with PRB7 IgG, SAF32, or control human IgG/ antibodies ranging from 0 to 5 g/ml for 4 days at 37°C. In the case of "Ϫ", ScN2a was cultured without antibody for 4 days at 37°C. Cell lysates were treated with PK (final concentration (conc.), 20 g/ml) as described under "Experimental Procedures." Lysates were separated on a 16.5% Tricine-polyacrylamide gel and immunoblotted with 6D11 antibody followed by HRP-conjugated anti-goat mouse IgG as described under "Experimental Procedures." whereas SAF32 strongly inhibited the generation/accumulation of PrP Sc . Similar results have been reported in which a PrP Sc -specific antibody had no prion-clearing effect, whereas the PrP C -specific antibody showed marked clearing activity (10). Because the mimotope of PRB7 is located at 128 -132, if this ␤-pleated sheet forms an interface for prion aggregation PRB7 IgG may block the assembly of PrP aggregates because the binding IgG is much larger than the target molecule (PrP). We examined this activity using FF32 cells or N2aL1 cells and obtained the same results as for ScN2a cells (data not shown). In this respect, our result suggested that, in the prion conversion reaction, PRB7 IgG-recognizing ␤-form PrP did not play a role in driving the conversion of the PrP conformation and appears rather to be an end product that loses prion activity and accumulates in the cytoplasm. This end product may trigger cell cytotoxicity via the mitochondrial machinery (42). This is in concert with the result that ␤-form recombinant PrP does not work as a template for prion amplification in protein misfolding cyclic amplification (41,48). It is conceivable that PRB7-negative, SAF32-positive granules as a full-length, non-␤-form PrP might be responsible for prion propagation (Fig. 10, C and E). Our results may be in agreement with the finding that prion propagation and toxicity in vivo occur in two distinct mechanistic phases (42). Thus, we have not obtained evidence to determine whether PRB7-recognizing ␤-form PrP represents the so-called PrP Sc that has both conversion activity and cell cytotoxicity. Further investigation is needed to determine whether ␤-form PrP affinity-purified with PRB7 shows prion activity such as the infectious and conversion/accumulation activity of the prion.
Although a number of studies illuminated the accumulation of PrP res in prion-infected cells or brain extracts, there is little literature giving much attention to the phenomenon in which most prion-infected ScN2a cells vigorously proliferate to survive in culture (49). The low frequency of PRB7 IgG-positive granules in ScN2a cells suggests the stochastic nature of the conversion of ␣-form to ␤-form PrP. The spongelike degeneration of prion-infected brain may reflect this feature.
In conclusion, we report the establishment of PRB7 IgG, which is the first human antibody discriminating the ␤-form from the ␣-form of PrP in the physiological condition. Using this antibody, we have demonstrated direct evidence of the generation and accumulation of ␤-form PrP in prion-infected ScN2a cells. It can be used to purify ␤-form PrP molecules without the need for protease digestion and may be useful to demonstrate the biochemical basis of the relationship of ␤-form PrP to PrP Sc and to elucidate the structure-based evidence for prion infectivity and neurotoxicity.