Identification of Novel Proteinase K-resistant C-terminal Fragments of PrP in Creutzfeldt-Jakob Disease*

The central event in the pathogenesis of prion diseases, a group of fatal, transmissible neurodegenerative disorders including Creutzfeldt-Jakob disease (CJD) in humans, is the conversion of the normal or cellular prion protein (PrPC) into the abnormal or scrapie isoform (PrPSc). The basis of the PrPC to PrPSc conversion is thought to involve the diminution of α-helical domains accompanied by the increase of β structures within the PrP molecule. Consequently, treatment of PrPSc with proteinase K (PK) generates a large PK-resistant C-terminal core fragment termed PrP27-30 that in human prion diseases has a gel mobility of ∼19-21 kDa for the unglycosylated form, and a ragged N terminus between residues 78 and 103. PrP27-30 is considered the pathogenic and infectious core of PrPSc. Here we report the identification of two novel PK-resistant, but much smaller C-terminal fragments of PrP (PrP-CTF 12/13) in brains of subjects with sporadic CJD. PrP-CTF 12/13, like PrP27-30, derive from both glycosylated as well as unglycosylated forms. The unglycosylated PrPCTF 12/13 migrate at 12 and 13 kDa and have the N terminus at residues 162/167 and 154/156, respectively. Therefore, PrP-CTF12/13 are 64-76 amino acids N-terminally shorter than PrP27-30 and are about half of the size of PrP27-30. PrP-CTF12/13 are likely to originate from a subpopulation of PrPSc distinct from that which generates PrP27-30. The finding of PrP-CTF12/13 in CJD brains widens the heterogeneity of the PK-resistant PrP fragments associated with prion diseases and may provide useful insights toward the understanding of the PrPSc structure and its formation.

Prion diseases consist of a group of fatal and transmissible neurodegenerative disorders including scrapie and bovine spongiform encephalopathy in animals, and Creutzfeldt-Jakob disease (CJD), 1 fatal familial insomnia (FFI), and Gerstmann-Strä ussler-Scheinker disease (GSS) in humans (1,2). Human prion diseases can be sporadic, familial, or acquired by infection. Most of them are characterized by the deposition of an abnormal prion protein, PrP Sc , in brain. PrP Sc derives from its host-encoded normal cellular isoform PrP C that is predominantly expressed in brain but also at lower levels in many other tissues (3)(4)(5). Mature human PrP C contains 209 amino acids encompassing residues 23-231, a disulfide bridge between residues 179 and 214, two consensus sites for N-linked glycans at residues 181 and 197, and it is attached to cell membranes via a C-terminal glycosylphosphatidylinositol (GPI) anchor (6 -12). Although PrP C and PrP Sc have an identical primary structure, they have distinct physicochemical properties. PrP C exists as a detergent-soluble monomer and is readily degraded by proteinase K (PK), whereas PrP Sc forms detergent-insoluble aggregates and shows high resistance to PK digestion (6,(13)(14)(15)(16)(17). Following treatment with PK, PrP Sc generates a protease-resistant core, referred to as PrP27-30 that is N-terminally truncated at the N terminus between residues 78 and 103 (6,18). It is thought that the PrP C to PrP Sc conversion involves the decrease of ␣-helical domains accompanied by the increase of ␤ structures in the midportion of the PrP molecule (19 -21). However, the precise location and extent of these structural changes within the PrP molecule and therefore, the tertiary and quaternary structures of PrP Sc , are largely a matter of speculation.
Full-length PrP Sc and PrP27-30 are the only known components of the naturally occurring infectious agent causing prion diseases (1) and are thought to be the primary cause of the histological changes in brains of subjects with prion diseases (22). However, the role of PrP Sc in the pathogenesis of these changes is poorly understood. This issue is further compounded by the presence of other derivatives of PrP Sc in human and animal prion diseases, particularly the 7-8-kDa-truncated PrP fragments found in GSS (23)(24)(25)(26) and the 16-and 7-kDa Cterminal fragments in scrapie-infected hamsters (27,28).
We have now characterized two novel C-terminal fragments of PrP in brains of patients with sporadic CJD (sCJD). These PrP fragments migrate at about 12 and 13 kDa on Tris-Tricine gradient gel, and have the N terminus that begins at residues 162/167 and 154/156, respectively, as determined by automated Edman degradation. They are resistant to PK digestion and derive from both glycosylated and unglycosylated forms. We propose that these fragments are distinct from PrP27-30 and derive from a subpopulation of full-length as well as Nterminally truncated PrP Sc carrying a different conformation.
Brain Tissues-Human brain tissues were collected at autopsy or biopsy and were kept frozen at Ϫ80°C until use. Brains from subjects with sporadic CJD were confirmed by histological examination, and by immunohistochemistry and immunoblotting to show the presence of PrP Sc as described (29,32). The control human brains were obtained from individuals unaffected by prion disease.
Molecular Genetics-Genomic DNA was extracted from frozen brain tissue. The methionine/valine polymorphism at codon 129 was determined as we described previously (33).
PK Digestion and Deglycosylation of PrP-Brain homogenates (10%, w/v) were prepared on ice in lysis buffer (100 mM NaCl, 10 mM EDTA, 0.5% Nonidet P 40, 0.5% sodium deoxycholate, 10 mM Tris-HCl, pH 7.5), and samples were centrifuged at 1,000 ϫ g for 10 min to remove cellular debris. To prepare the detergent-insoluble fraction (P2), brain homogenates were centrifuged at 14,000 ϫ g for 25 min at 4°C, and supernatants were further centrifuged at 100,000 ϫ g for 1 h at 4°C. The pellets were resuspended in lysis buffer. For PK digestion of PrP, samples were incubated with PK at 100 g/ml for 1 h at 37°C, and the digestion was then terminated by the addition of 3 mM phenylmethylsulfonyl fluoride. For deglycoyslation of the protein, samples were denatured and incubated in the presence of recombinant peptide N-glycosidase F (PNGase F) according to the manufacturer's protocol (Roche Applied Science).
Two-dimensional gel electrophoresis was performed as described by the supplier with minor modifications using PROTEIN IEF cell (Bio-Rad). For the first dimension isoelectric focusing, P2 fraction boiled in gel loading buffer was precipitated by 9-fold volume of chilled acetone at Ϫ20°C for 1.5 h, followed by centrifugation at 16,000 ϫ g for 15 min at 4°C. The pellets were resuspended in 10% trichloroacetic acid for 2 h at room temperature and were centrifuged at 16,000 ϫ g for 15 min. The pellets were resuspended in 200 l of rehydration buffer (7 M urea, 2 M thiourea, 2% CHAPS, 0.5% ampholine pH 3-10, and trace amounts of bromphenol blue), and were loaded onto the immobilized pH gradient (IPG) strips for rehydration at 50 V for 14 h. The rehydrated IPG strips were focused for about 40 kVh. For the second dimension SDS-PAGE, the focused IPG strips were equilibrated for 15 min each in equilibration buffer 1 (6 M urea, 2% SDS, 20% glycerol, 2% DTT, and 0.375 M Tris-HCl, pH 8.8), and then in equilibration buffer 2 (6 M urea, 2% SDS, 20% glycerol, 2.5% iodoacetamide and 0.375 M Tris-HCl, pH 8.8). The equilibrated strips were loaded onto 16.5% Tris-Tricine SDS-PAGE gels (Bio-Rad Criterion gels). Electrophoresis was conducted at 120 V.
Proteins on the SDS-PAGE gels were electrotransferred onto PVDF membranes at 70 V for 2 h at 4°C. The membranes were blocked with 5% nonfat milk in TBST (150 mM NaCl, 0.05% Tween 20, 10 mM Tris-HCl, pH 7.6) overnight at 4°C prior to incubation with antibodies. Membrane-bound proteins were probed with either anti-C antibody at 1:3,000 or with 3F4 antibody at 1:50,000. Following washes with TBST, the PVDF membranes were incubated with appropriate secondary antibodies conjugated with horseradish peroxidase (Amersham Biosciences). The PrP bands were visualized on Kodak X-Omat films by enhanced chemiluminescence (ECL Plus kit, Amersham Biosciences).
Purification of PK-resistant PrP-CTF 12/13-Brain tissues (5-10 g) were used for purification of PK-resistant PrP Sc fragments according to the published method (34), as modified (35,36). After the final sedi-mentation of PK-resistant PrP Sc fragments by ultracentrifugation, the pellets were denatured and treated with PNGase F. PrP-CTF 12/13 were purified further by micropreparative continuous elution SDS-PAGE (16% gel, Mini Prep Cell apparatus, Bio-Rad). Proteins were eluted at a flow rate of 70 l/min and collected into 400 l fractions. The fractions containing PrP-CTF 12/13 were pooled and lyophilized. Immunoblotting with anti-C antibody was used to monitor the purification throughout the procedures.
N-terminal Protein Sequencing-Purified proteins were separated by 10 -20% mini Tris-Tricine SDS-PAGE gradient gels (Novex pre-cast gel, Invitrogen), transferred onto Problott membranes (Applied Biosystems, Foster City, CA), and visualized by Coomassie Blue staining. N-terminal protein sequencing by automated Edman degradation was performed as described previously (18), at the ProSeq Microsequencing Facility (Boxford, MA) with an Applied Biosystems 477A Protein Sequencer. N-terminal sequencing typically proceeded for 10 cycles. Multiple N-terminal sequences were obtained by alignment of the experimentally determined amino acids at each cycle with the translated human PrP sequence (Swiss-Prot accession number: P04156).
Triton X-114 Phase Separation-Phase separation of PrP in Triton X-114 was performed as described (37) with minor modification. Brain tissues were homogenized in 1% Triton X-114 in PBS. 200 l of brain homogenate was centrifuged at 14,000 rpm for 30 s at 4°C. The supernatant (100 l) was incubated at 37°C for 15 min and centrifuged at 14,000 rpm for 1 min. A solution of 1% Triton X-114/PBS (600 l) was added into the detergent phase and samples were split into two parts. The samples were incubated with (ϩ) or without (Ϫ) phosphatidylinositol-specific phospholipase C (PIPLC) (Sigma) at 4°C overnight, and the phase separation repeated.
Sandwich Blotting of GPI-anchored PrP with Proaerolysin-GPIanchored proteins were detected with proaerolysin-mediated sandwich blotting according to the published procedure (38). Proteins transferred onto the PVDF membrane were incubated with proaerolysin (Protox Biotech, Victoria, British Columbia, Canada) at 0.5 g/ml in PBS for 1 h at room temperature. The blot was incubated with anti-proaerolysin monoclonal antibody (Protox Biotech, Victoria, British Columbia, Canada) at 1:8,000 for 1 h. After wash with TBST and incubation with HRP-conjugated IgG (sheep anti-mouse second antibody), the proaerolysin-bound GPI anchored proteins were visualized by enhanced chemiluminescence (ECL Plus kit, Amersham Biosciences).
Immunoprecipitation-Immunoprecipitation of PrP from P2 fraction was performed as described (39) with minor modifications. In brief, anti-PrP monoclonal antibodies (8B4) at 100 g were coupled to 7 ϫ 10 8 tosyl activated magnetic Dynabeads in 1 ml of phosphate-buffered saline, pH 7.5 (PBS) at 37°C for 20 h, and the beads were then washed twice with 0.1% bovine serum albumin (BSA) in PBS. The antibodyconjugated beads were incubated with 0.1% BSA, 0.2 M Tris-HCl, pH 8.5 at 37°C for 4 h to block nonspecific binding sites, followed by two washes with 0.1% BSA in PBS. For immunoprecipitation of PrP, 100 l of antibody-conjugated beads was incubated with 870 l of lysis buffer in the presence of 30 l of P2 fraction at room temperature for 3 h. The immune complex-containing beads were washed three times with washing buffer (2% Nonidet P 40 and 2% Tween-20 in PBS, pH 7.5). PK digestion and deglycosylation of PrP were performed as described above. Samples were mixed with 2ϫ gel loading buffer, heated at 95°C for 5 min, and subjected to SDS-PAGE and immunoblotting.

PK-resistant C-terminal Fragments of PrP Are Present in
sCJD Brains-We prepared the detergent-insoluble fraction from brains of 29 subjects affected by various subtypes of sCJD, which included five subtypes identified as MM1, VV1, MM2, MV2, and VV2 sCJD, according to the classification of Parchi et al. (40). These preparations were treated with PK and analyzed by Tris-Tricine gradient gels and immunoblotting using anti-C, an antibody to the C-terminal domain of PrP. Following this procedure, we observed two sets of PK-resistant PrP fragments (Fig. 1). The first set consisted of the well-known PrP27-30 (16,29,32). It migrated as three bands of 30 -35, 27-28, and 19 -21 kDa that correspond to N-terminally truncated forms of diglycosylated, monoglycosylated, and unglycosylated PrP Sc , respectively. Unlike the PrP27-30, the second set of PK-resistant fragments migrated at ϳ12 and ϳ13 kDa, and were designated PrP C-terminal fragments 12/13 kDa (PrP-CTF 12/13) (Fig. 1). They varied considerably in amount and ratio among different subtypes of sCJD (Table I). In 26 of the 29 sCJD brains examined, PrP-CTF 12/13 accounted for up to 24% of all PK-resistant PrP fragments as measured by densitometry. The analysis of the ratio, which was examined in 13 cases, often showed a better representation of PrP-CTF13 than PrP-CTF12 but the opposite was occasionally true (Table I). The presence, amount and ratio varied from case to case and showed no apparent correlation with the amount of PrP27-30 and the sCJD subtype ( Fig. 1 and Table I). In three cases, PrP-CTF 12/13 were present only in trace amounts as their detection required longer exposure of the immunoblots. The study of the brain distribution, carried out in 14 cases showed that PrP-CTF 12/13 generally were better represented in the neocortex than in the hippocampal formation and subcortical structures such as putamen, thalamus, and cerebellum.
PrP-CTF12/13 Are Truncated Glycoforms of PrP Sc -Because PrP contains two consensus sites for N-linked glycans at residues 181 and 197 that contributes to its heterogeneity (6,11,12), deglycosylation by PNGase F is often used to simplify the gel migration pattern and to reveal the size of the PrP backbone. As shown in Fig. 2, PrP glycoforms were readily detectable in control brains, which upon deglycosylation, shifted mainly to a 27 kDa band corresponding to the full-length PrP and an 18 kDa band corresponding to an N-terminally truncated PrP fragment with an N terminus starting at residue 111/112, as characterized previously (29). They were invariably sensitive to PK digestion as expected for normal PrP C . PrP-CTF12/13 in the molecular mass range of 12 and 13 kDa were not detected in control brains with or without deglycosylation. PrP-CTF12/13 were barely detectable in untreated brain homogenates from sCJD subjects. However, PK treatment and removal of glycans with PNGase F independently resulted in a substantial increase in amount of PrP-CTF12/13 with the same gel mobility of 12 and 13 kDa as that observed following treatment with both PK and PNGase F (Fig. 2). These findings indicate that PrP-CTF12/13, like PrP27-30 (16,29,32), are primarily generated by PK treatment of both glycosylated and unglycosylated PrP Sc forms. PrP-CTF12/13 are unlikely to be generated during post mortem autolysis because they are present with the same characteristics in sCJD brain tissue obtained at biopsy, in which autolysis is not expected to play a role (data not shown).
PrP-CTF12/13 Are Part of the C-terminal Domain-PrP-CTF12/13 immunoreacted only with anti-C antibody (29) against the C terminus of PrP (residues 220 -231), but not with 3F4 antibody (30) recognizing an epitope located in the mid-PrP region (residues 109 -112) (Fig. 3). This finding suggests that PrP-CTF12/13 are generated by truncation of PrP at sites C-terminal to residue 112. Amino acid sequencing by automated Edman degradation showed that the N terminus of the PrP-CTF12 mainly started at residue 162 and occasionally at residue 167, while PrP-CTF13 preferentially began at residue 154 and infrequently at residue 156 (Table II). Therefore, PrP-CTF12/13 are made of the C-terminal domain of PrP spanning from residues 162/167 to 231 and residues 154/156 to 231, respectively.
Two-dimensional Immunoblotting of PrP-CTF12/13-We further examined PK-treated, detergent-insoluble fraction (P2) from sCJD brains using two-dimensional gel electrophoresis which separates proteins not only on the basis of the relative mass but also of net electrical charges (Fig. 4). A broad set of spots with molecular mass of 20 -30 kDa and pH 5.2-8.0 that  (32,40). The relative abundance between the 13 and 12 kDa PrP-CTF is qualitatively indicated by ϩ (less abundant) or ϩϩ (more abundant) as estimated by visual inspection of immunoblots probed with anti-C antibody. Due to the closeness of the PrP-CTF12/13 bands, accurate quantitation became difficult and therefore was not attempted. In some cases, only trace amounts of PrP-CTF12/13 were detected following long exposure of immunoblots.
Effect of PK and PNGase F on PrP-CTF12/13. Brain homogenates from control (left panel) and sCJD (right panel) subjects were untreated or treated with PK and PNGase F (either separately or in combination). PrP-CTF12/13 were present in sCJD samples but not in controls. In sCJD brains, PrP-CTF12/13 were barely detectable in untreated preparations, but PK treatment and removal of glycans with PNGase F independently increased the amount of PrP-CTF12/13 without changing their gel mobility. The immunoblots were probed with anti-C antibody. Molecular mass markers are shown in kDa.
FIG. 1. Immunoblot of PrP-CTF12/13 and PrP27-30 from sCJD. P2 fractions from two cases of sCJD were incubated with PK at 100 g/ml for 1 h at 37°C. The samples were loaded onto 10 -17% Tris-Tricine gradient gel and the proteins were transferred onto PVDF membrane. PK-resistant PrP fragments were detected on immunoblots using the anti-C antibody to the PrP C terminus. Two groups of fragments were observed. PrP-CTF12/13 migrated as two bands at 12 and 13 kDa, while PrP27-30 migrated as three bands corresponding to the diglycosylated form as the upper band at ϳ30 kDa, the monoglycosylated form as the middle band at ϳ24 kDa, and the unglycosylated form as the lower band at ϳ19 -21 kDa. Immunoblots from two representative cases of sCJD are shown (lane 1, sCJD MV2; lane 2, sCJD VV1). Molecular mass markers are indicated in kDa.
immunoreacted with both anti-C antibody (Fig. 4A) and 3F4 (Fig. 4C), corresponded to glycoforms of PrP27-30 (29,41,42). Two additional sets of spots were detected by anti-C antibody (Fig. 4A) but not by the 3F4 antibody (Fig. 4C). A distinct set of 4 -6 spots with a pI within pH 5.0 -6.5 and a molecular mass of 12-13 kDa is likely to represent the unglycosylated form of PrP-CTF12/13 readily detectable on SDS-PAGE gels. Another set of spots with more acidic pI values within pH 4.5-6.0 and a mass of ϳ20 -21 kDa is likely to represent the glycosylated species of the PrP-CTF12/13 (Fig. 4A). To strengthen this conclusion, the PK-treated P2 fraction of sCJD brains was deglycosylated by PNGase F. After this treatment, the heterogeneity of PrP spots was greatly reduced. A group of 4 -5 PrP spots with a mobility of 19 -20 kDa and pI between pH of 6.5-8.5 were detected by both anti-C (Fig. 4B) and 3F4 (Fig. 4D). These spots matched the deglycosylated PrP27-30 reported in previous studies (41,42). Another group of 4 -6 spots displaying a gel mobility of 12-13 kDa and a pI within pH of 5-6.5 was detected only by anti-C (Fig. 4B), and corresponded to the pI of the unglycosylated PrP-CTF12/13. Therefore, the latter group is likely to derive from both the original unglycosylated PrP-CTF12/13 species and those generated following deglycosylation. The multiple spots present even after deglycosylation in both PrP27-30 and PrPCTF12/13 are likely due to the heterogeneity of the GPI moiety covalently linked to the C terminus (29,41,42). No major differences were observed between the two-dimensional gel profiles of PrPCTF12/13 from CJD cases with PrP Sc types 1 and 2. However, detailed studies are needed to exclude minor variations.
We computed the expected molecular weight and pI of unglycosylated PrP-CTF12/13 using the ProtParam program of the SWISS-PROT (ca.expasy.org/tools/protparam.html). The FIG. 3. Antibody mapping of PrP-CTF12/13. Brain homogenates from sCJD subjects were either untreated or treated with PK and PNGase F. PrP bands were detected on immunoblots using either anti-C antibody (upper panel) or 3F4 antibody (lower panel). In all preparations, PrP-CTF12/13 were detected only by anti-C but not by 3F4. Therefore, PrP-CTF12/13 are N-terminally truncated PrP fragments. Molecular mass markers are shown in kDa.

TABLE II N-terminal sequencing of PrP-CTF12/13
Protein sequencing of the PK-resistant PrP-CTF12/13 fragments was performed using automated Edman degradation as described under "Experimental Procedures." N-terminal amino acid sequence was derived from amino acids released in each consecutive cycle of Edman degradation using the translated human PrP sequence as alignment reference. The initial yields of all samples ranged from 2 to 10 pmol, and the average repetitive yields were about 90%.  4. Two-dimensional immunoblots of PrP-CTF12/13 and PrP27-30. PK-treated P2 fraction from a sCJD case was either untreated or deglycosylated with PNGase F. Proteins were separated by two-dimensional gel electrophoresis. PrP was detected on immunoblots using either anti-C antibody or 3F4 antibody. A and B, two-dimensional immunoblots probed with anti-C antibody. Without deglycosylation (A), the spots corresponding to glycoforms of PrP27-30 migrated in a broad region from molecular mass 30 to 19 kDa and with pI between 5.2 and 8.0. The spots corresponding to PrP-CTF12/13 displayed pI between 5.0 and 6.5, while additional spots with pI within pH 4.5-6.0 and a mass of ϳ20 -21 kDa are likely to represent the glycosylated form of PrP-CTF12/13. After deglycosylation by PNGase F (B), two groups of PrP spots were detected, representing unglycosylated PrP27-30 and PrP-CTF12/13 species, respectively. C and D, two-dimensional immunoblots probed with 3F4 antibody. In both untreated (C) and PNGase F-treated (D) samples, spots corresponding to glycosylated and unglycosylated PrP-CTF12/13 were not detected by 3F4 antibody. However, spots corresponding to glycoforms of PrP27-30 (C) and those representing deglycosylated PrP27-30 (D) were readily detected by 3F4 antibody. The positions of pI gradient and molecular mass markers (in kDa) are indicated on the bottom and on the left, respectively.
PrP sequences containing C-terminal 78 and 70 amino acids matching those of PrPCTF12/13 were assigned molecular masses of 9.4 and 8.4 kDa, which, adding ϳ4 kDa for the GPI anchor, become 13.4 and 12.4 kDa, comparable to the 13 and 12 kDa determined by the gel migration. Theoretical pIs of PrP-CTF12/13 are 5.2-5.8, not very different from the pIs indicated by two-dimensional electrophoresis. The subtle difference in pI (ϳ0.5) could derive from the heterogeneity of the GPI anchors. These data confirm that the PrP-CTFs are C-terminal fragments of PrP Sc ; i.e. they result from PK cleavages located much more toward the C terminus than the cleavages that generate PrP27-30.
PrP-CTF12/13 Likely Contains the GPI Anchor-To further determine if the PrP-CTF12/13 fragments contain the GPI anchor, we subjected brain homogenates to phase partitioning in the detergent Triton X-114 that keeps hydrophobic proteins in the detergent phase and soluble proteins in the aqueous phase. Following treatment with PIPLC, GPI-anchored proteins released from the membrane are supposed to shift from the detergent phase into the aqueous phase. As shown in Fig.  5A, PrP species from both normal and CJD brains including PrP-CTF12/13 were mostly recovered in the detergent phase without PIPLC treatment. After PIPLC treatment, most of the PrP from the control shifted from the detergent phase to the aqueous phase, as it is also observed for other GPI-anchored proteins due to the loss of hydrophobic diacylglycerol portion of the anchor (37). In contrast, PrP species from CJD including PrP-CTF12/13 remained in the detergent phase after PIPLC treatment, suggesting that the GPI anchor is either lacking or is resistant to PIPLC treatment as shown previously (43).
To exclude the possibility that PrP-CTF12/13 is anchorless, we carried out sandwich blotting using proaerolysin, a toxin binding specifically and with high affinity to GPI-anchored proteins, which was then detected with a specific antibody (38). The PrP fragments detected by the anti-C antibody in a PKand PNGase F-treated P2 fraction enriched in PrP-CTF12/13 co-migrated with the fragments detected by the monoclonal antibody to proaerolysin (Fig. 5B). This suggests that PrP-CTF12/13 fragments contain the C-terminal GPI anchor.

PrP-CTF12/13 Can Be Derived from PK Digestion and PN-Gase F Deglycosylation of Both Full-length and Truncated PrP
Forms in Vitro-Previous studies from other and our laboratories have demonstrated that the untreated detergent-insoluble fraction (P2) prepared from CJD brains contains not only fulllength PrP but also forms that are truncated at different sites of the N terminus (29,41,44). To determine if PrP-CTF12/13 derive from either full-length or truncated PrP, the full-length PrP molecules were separated from truncated forms by immunoprecipitation with 8B4, an antibody to the N terminus of PrP (31). The separated full-length and truncated PrP molecules in the 8B4 immunoprecipitate and the supernatant following immunoprecipitation, respectively, were then subjected to PK digestion and PNGase F deglycosylation. As shown in Fig. 6, full-length, but not truncated PrP was observed in immunoprecipitate while truncated PrP forms including PrP-CTF12/13 were present in supernatant only. After treatments with PK and PNGase F, PrP27-30 was found in both immunoprecipitate and supernatant in similar amount, suggesting that PrP27-30 derive from both full-length and truncated PrP. However, PrP-CTF12/13 appeared to be more abundant in supernatant than in immunoprecipitate, suggesting that while PrP27-30 derives from both full-length PrP in similar amounts, PrP-CTF12/13 derive preferentially from truncated PrP. DISCUSSION We have identified two novel C-terminal fragments of PrP, PrP-CTF12/13, in human brains affected by different subtypes of sporadic CJD. The PrP-CTF12/13 are present along with the FIG. 5. Characterization of GPI anchor in PrP-CTF 12/13. A, immunoblot of PIPLC-treated PrP-CTF12/13. Brain homogenates (10% w/v) were prepared in 1% Triton X-114 in PBS at 4°C. Following the removal of cellular debris, phase separation was initiated at 37°C after samples were incubated at 4°C overnight in the presence (ϩ) or absence (Ϫ) of PIPLC. PrP in the detergent (D) and aqueous (A) phases was determined by SDS-PAGE and immunoblotting with anti-C antibody. B, sandwich blotting of PrP-CTF12/13 with proaerolysin. PrP enriched in P2 fraction from CJD brains was digested with PK and was then deglycosylated by PNGase F. Proteins were separated on Tris-Tricine SDS-PAGE (10 -17% gradient gel). The blot was incubated with proaerolysin at 0.5 g/ml in PBS for 1 h at room temperature. Proaerolysin bound to GPI-anchored proteins were detected with immunoblotting with an anti-proaerolysin monoclonal antibody. The position of PrP-CTF12/13 is indicated by an asterisk (*).
FIG. 6. Generation of PrP-CTF12/13 from full-length and truncated PrP in vitro. P2 fraction from sCJD subjects was processed for immunoprecipitation of full-length PrP using beads conjugated to 8B4, a monoclonal antibody to the N terminus of PrP (31). The resulting immunoprecipitate and supernatant are expected to contain full-length and truncated PrP forms, respectively. Immunoprecipitate and supernatant were either untreated or treated with both PK and PNGase F. The samples were then subjected to SDS-PAGE, followed by immunoblotting with anti-C antibody. Following treatment with PK and PN-Gase F, PrP27-30 was detected in both immunoprecipitate and supernatant in similar amounts. In contrast, PrP-CTF12/13 appeared to be more represented in the truncated PrP fraction (supernatant) than in the full-length PrP fraction (immunoprecipitate).
well-known PK-resistant PrP fragments commonly referred to as PrP27-30 (16). Like PrP27-30, PrP-CTF12/13 are primarily generated by in vitro limited proteolysis and represent a PKresistant C-terminal core of PrP Sc that includes glycosylated and unglycosylated forms. The glycosylated forms of PrP-CTF12/13 have a gel mobility of ϳ20 -21 kDa, whereas the two unglycosylated forms migrate to ϳ12 kDa and ϳ13 kDa and are generated by cleavages at residues 154/156 and 162/167, respectively. The identification of PrP-CTF12/13 widens the heterogeneity of PK-resistant fragments of PrP associated with human prion diseases. It also raises questions concerning 1) the nature and mode of formation of these fragments, 2) how these fragments can be accommodated in current models of PrP C to PrP Sc conversion, and 3) the role that they may play in the pathogenesis of CJD.
PK-resistant Fragments in Human Prion Diseases-The discovery of PrP-CTF12/13 brings to a total of three groups of PrP fragments resistant to PK that are associated with human prion diseases (Fig. 7). The most common and best known is PrP27-30 (16). It is the only PK-resistant, disease-associated species that has been proven to retain infectivity. It includes a large number of fragments of different size due to the highly ragged N terminus that spans from residues 78 -103 (18). PrP27-30 appears to have an intact C terminus identical to that of PrP C including the presence of the disulfide bridge (6 -12). The site of cleavage of the full-length PrP Sc by PK and, therefore, the size of the PK-resistant PrP27-30, is influenced by the PrP genotype at codon 129 of the PrP gene, the location of a common methionine/valine (M/V) polymorphism (18). Our previous studies have shown that there are two major types of PrP27-30 in human prion diseases. In PrP27-30 type 1, the N terminus commonly starts at residue 82 but this type also includes several secondary species with the N terminus starting between residues 78 and 97. In contrast, the N terminus of PrP27-30 type 2 commonly starts at residue 97 with secondary sites between residues 92 and 103 (18). Furthermore, PrP27-30 types 1 and 2 co-distribute with distinct disease phenotypes (18,32) and are conserved upon transmission to receptive animals (45).
The second family of PK-resistant PrP fragments includes much smaller species with relative molecular mass of 7-8 kDa (PrP7-8) (Fig. 7). PrP7-8 has only been observed in subjects associated with mutations in the PrP gene linked to GSS, a group of familial prion diseases characterized by the presence of PrP amyloid deposits. PrP7-8 is an internal fragment that varies in size according to the PrP gene mutation (23)(24)(25)(26), but overall spans from residues 74 -90 to residues 146 -153 (25). PrP7-8 is present in PK-untreated GSS brain preparations but increases in quantity following PK-treatment. Furthermore, it is expressed exclusively by the mutated allele (23)(24)(25)(26). Therefore, it is likely that in vivo PrP7-8 represents a PK-resistant region of PrP, possibly generated by an abnormal PrP processing pathway linked to GSS-causing mutations. The infectivity of PrP7-8 has not been carefully assessed. However, a PrP synthetic peptide that spanned residues 89 -143 and contained a GSS mutation accelerated or possibly triggered a GSS-like condition following inoculation to transgenic mice carrying the same GSS mutation but not to wild-type mice (46). Furthermore, before inoculation the synthetic PrP peptide had to be refolded in a ␤-structure-rich conformation, therefore mimicking the naturally occurring PrP7-8. These data suggest that PrP7-8 can propagate the prion disease only when an appropriate PrP gene mutation is present in both donor and recipient, arguing that while PrP27-30 can recruit wild type PrP, PrP7-8 requires the presence of a GSS mutation.
PrP-CTF12/13, the third group of PK-resistant PrP fragments as reported in the present study, are different from PrP27-30 not only in size but also because they do not show any effect of the PrP genotype at codon 129 on their PK cleavage site and correlation with the disease phenotype. They are FIG. 7. Diagram of human PrP C and PK-resistant PrP fragments observed in human prion diseases. The NMR-derived but linearized human PrP C structure comprises an unstructured N-terminal domain and a globular C-terminal domain that contains three ␣-helices (␣1-3) and two short ␤-strands (␤1-2) (47). Post-translational modifications include a disulfide bridge (S-S) between residues 179 and 214, two consensus sites for N-linked glycans at residues 181 and 197 and a GPI anchor at the C terminus. PrP27-30 is a well characterized PK-resistant PrP fragment present in most human prion diseases. It is generated by PK cleavage between residues 74 and 103 of human PrP Sc in sCJD producing a highly ragged N terminus (18). PrP7-8 has to date only been observed in subjects with GSS and comprises a group of PK-resistant fragments spanning residues 74/90 to 146/153 (23)(24)(25)(26). PrP-CTF12/13, identified in this study, are PK-resistant and contain both glycosylated and unglycosylated forms, but it remains to be determined whether they retain the disulfide bridge and the ␣-helix 2 and 3 structures. also quite different from PrP7-8 as for the size, the region of the PrP they embody and the lack of obligatory relation with PrP gene mutations (Fig. 7). Pertain to the above discussion on human prion diseases, it is of interest to note that a 16 kDa C-terminal domain spanning from PrP residues ϳ115 to ϳ217 (27) and a 7 kDa fragment derived from the extreme PrP C terminus (28) have been found in PK-treated preparations of scrapie-infected hamsters.
Possible Modes of Formation of PrP-CTF12/13-An interesting feature of PrP-CTF12/13 is that their N terminus matches the C terminus of PrP7-8 and the combination of these two fragments corresponds to PrP27-30 (Fig. 7). This finding raises the question as to whether upon PK treatment PrP-CTF12/13 are generated by two separate cleavages of the full-length PrP Sc that would generate both PrP7-8 and PrP-CTF12/13. Alternatively, they may originate from another species of abnormal PrP that carries a PK-resistant core located closer to the C terminus than that of PrP27-30. Our failure to consistently detect PrP7-8 in preparations containing PrP-CTF12/13 makes the former possibility unlikely. Therefore, in vivo PrP-CTF12/13 may represent the C-terminal region of a subpopulation of PrP Sc in which the PK-resistant core is displaced 64 -76 residues toward the C terminus when compared with that of the PrP Sc that generate PrP27-30. Since the detergentinsoluble fraction enriched in PK-resistant PrP contains both full-length and N-terminally truncated PrP forms (29), we reasoned that PrP-CTF12/13 might be the PK-resistant core of these truncated PrP forms while PrP27-30 is the PK-resistant core for the full-length forms. However, PrP-CTF12/13 could be recovered from both full-length and truncated PrP forms following PK-and PNGase treatments, although more appeared to be recovered from the truncated than from full-length forms. Therefore, generation of the PrP-CTF12/13 is not strictly related to the size of the PrP Sc . If indeed PrP-CTF12/13 represent the C-terminal region of a subpopulation of PrP Sc in which the PK-resistant core is displaced toward the C terminus, this "PrP Sc " subpopulation must have a tertiary structure different from that of the PrP Sc generating PrP27-30. This conclusion raises the question of how PrP-CTF12/13 fit in the current models of PrP27-30 formation. Structural NMR studies (47) have shown that the N-terminal region (residues 23-125) of human PrP C is flexible and unstructured, while the C-terminal domain contains 3 ␣-helices (␣-helix 1, residues 144 -154; ␣-helix 2, residues 173-194; and ␣-helix 3, residues 200 -228) and 2 short ␤-strands (␤-strand 1, residues 128 -131 and ␤-strand 2, residues 161-164) (Fig. 7). Although the tertiary structure of PrP Sc is still unknown, a common model holds that the PKresistant core of PrP Sc corresponding to PrP27-30, is formed by conversion of the ␣-helix 1 to ␤ structure, while the secondary and tertiary structures of the C terminus are preserved (48). The C-terminal end of the ␤ structure domain is unknown but it is tentatively placed in the vicinity of residue 170, i.e. at the beginning of ␣-helix 2 (49,50). The formation of PrP-CTF12/13 does not fit in this model. If the formation of a ␤ structure followed by oligomerization, which protects the C terminus, is the mechanism that also provides PK resistance to PrP-CTF12/ 13, the newly formed ␤ structure must start between residues 154 -167, i.e. near the N terminus of the ␣-helix 2. This suggests that other PrP regions closer to the C terminus than that of PrP27-30 may trigger the formation of ␤ structure. The limited available space between the N terminus of the putative ␤ structure domain in PrP-CTF12/13 and the ␣-helix 2 raises the question of whether the PrP-CTF12/13 C-terminal region maintains the original structure or is also converted to ␤ structure. Based on the evidence that the disulfide bridge joining ␣-helices 2 and 3 is present and is required for infectivity, the prevailing hypothesis is that the original globular structure of the PrP173-228 C-terminal domain is maintained in PrP27-30. However, a model in which the tertiary structure of PrP27-30 differs from that of PrP C has also been proposed (49). Finally, PrP-CTF12/13 might derive from PrP through the activation of an alternative metabolic pathway. Increase in PK-resistant PrP has been recently demonstrated in cell culture when the proteasome activity is impaired (51).
Role of PrP-CTF12/13 in the Pathogenesis of CJD-It would certainly be of interest to know whether PrP-CTF12/13 are capable of propagating prion diseases and whether they are neurotoxic. One can speculate that if in vivo PrP-CTF12/13 are part of a subpopulation of PrP Sc in which the ␤ structure conformation is located in region immediately C-terminal of residues 154 -167, this "PrP Sc " variant might be infectious. This possibility is supported by two sets of data obtained with PrP carrying various deletions, referred to as miniprions. One set showed that a redacted PrP (PrP106) lacking the 141-176 region can propagate the prion disease suggesting that this region is not required for infectivity (52). This finding is relevant because according to current models the PrP141-176 region is believed to have at least part of the ␤ structure, which seems also to be largely absent in PrP-CTF12/13. The second set of experiments argues that miniprions lacking most of, or all, the C-terminal PrP region, which makes up the bulk of the PrP-CTF12/13, cannot transmit the prion disease, although they are highly toxic (53). Whether PrP-CTF12/13 are infectious and/or toxic in PK-treated PrP preparations as PrP27-30 remains to be determined. If they were infectious, they would further reduce the size of the PrP Sc region required for infectivity.
In conclusion, the finding of the PrP-CTF12/13 present in sCJD-affected brains paves the way to further studies of nature, formation, and pathogenic role of prions. The data reported here raise the possibility that a novel subpopulation of PrP Sc carrying a conformation different from that of the classical PrP Sc is present in prion diseases.