Characterization of 2′-Fluoro-RNA Aptamers That Bind Preferentially to Disease-associated Conformations of Prion Protein and Inhibit Conversion*

We have isolated artificial ligands or aptamers for infectious prions in order to investigate conformational aspects of prion pathogenesis. The aptamers are 2′-fluoro-modified RNA produced by in vitro selection from a large, randomized library. One of these ligands (aptamer SAF-93) had more than 10-fold higher affinity for PrPSc than for recombinant PrPC and inhibited the accumulation of PrPres in near physiological cell-free conversion assay. To understand the molecular basis of these properties and to distinguish specific from non-specific aptamer-PrP interactions, we studied deletion mutants of bovine PrP in denatured, α-helix-rich and β-sheet-rich forms. We provide evidence that, like scrapie-associated fibrils (SAF), the β-oligomer of PrP bound to SAF-93 with at least 10-fold higher affinity than did the α-form. This differential affinity could be explained by the existence of two binding sites within the PrP molecule. Site 1 lies within residues 23–110 in the unstructured N terminus and is a nonspecific RNA binding site found in all forms of PrP. The region between residue 90 and 110 forms a hinge region that is occluded in the α-rich form of PrP but becomes exposed in the denatured form of PrP. Site 2 lies in the region C-terminal of residue 110. This site is β-sheet conformation-specific and is not recognized by control RNAs. Taken together, these data provide for the first time a specific ligand for a disease conformation-associated site in a region of PrP critical for conformational conversion. This aptamer could provide tools for the further analysis of the processes of PrP misfolding during prion disease and leads for the development of diagnostic and therapeutic approaches to TSEs.

A simple observation that emerges from the examination of the structures of all mammalian recombinant PrPs 1 that have been solved so far in their ␣-helix-rich forms is that they have a highly positively charged N-terminal tail that is flexibly disordered and a stable globular C-terminal domain (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). It is becoming clear that these structural features must have some functional significance. Indeed, there is increasing evidence that PrP may accomplish its as yet unknown physiological function by allowing both the structured C-terminal domain and the unstructured N-terminal part to interact with a variety of ligands.
The C-terminal domain seems to interact with a protein X that is thought to feature in the conversion of PrP C into PrP Sc by acting like a molecular chaperone in facilitating the unfolding of PrP C and its refolding into nascent PrP Sc (11,12). Protein X appears to bind to the side chains of residues 165-171 that form a loop and to residues on the surface of helix C that lie between the disulfide bond and the C terminus (12).
The unstructured N-terminal part of PrP can interact with a variety of molecules. Metal binding of PrP resides primarily in an octapeptide repeat motif between residues 53 and 93 (13). Although a more distal site has also been proposed (14), there seems to be a connection between metal-binding status and the conformation of PrP (15)(16)(17). Sulfated glycans interact with PrP (18) of which heparin and heparan sulfate have been shown to have three broad binding sites: residues 23-52, 53-93, and 110 -123 (19). The binding of heparin and copper comap to residues 53-93 and, interestingly, heparin binding seems to be enhanced by copper ions (19,20). Large and small nucleic acids (21)(22)(23)(24) also interact with PrP. It has been shown that human, ovine, and murine PrPs possess nucleic acid binding and chaperoning activities similar to retroviral nucleocapsid protein (25,26), and these properties are associated with the unstructured N-terminal region of the protein (5,10). Previously reported 2Ј-OH aptamers raised against recombinant hamster prion protein bound to a site between residues 23-52 within this nonspecific nucleic acid binding region (27). The physiological implications of the interactions between PrP and nucleic acid raise a number of questions that are still a matter of speculation (23)(24)(25)(26)28).
All of the published studies in which interactions between PrP and nucleic acids have been demonstrated were done with the ␣-helix-rich form of PrP. To extend these studies we decided to isolate nucleic acids that had high affinity for an infectious, prion-containing material. We used purified, non-proteinase K-treated scrapie-associated fibrils (SAF) prepared from the 263K strain as a target during in vitro selection for apatmer. This SAF preparation is expected to include all variant PrP conformations and PK sensitivities associated with infectivity in this strain of prion, and should represent an appropriate target to investigate the capacity of PrP Sc as well as other proteins that co-purify with it to retrieve high affinity aptamer ligands from a complex combinatorial RNA library.
We provide evidence that, like SAF, the ␤-oligomer of PrP bound to SAF-93 with at least 10-fold higher affinity than did the ␣-form. This differential affinity could be explained by the existence of two binding sites within the PrP molecule: Site 1 lies within residues 23-110 in the unstructured N terminus and is a nonspecific RNA binding site found in all forms of PrP. The region between residues 90 and 110 forms a hinge region that is occluded in the ␣-rich form of PrP but becomes exposed in the denatured form of PrP. Site 2 lies in the region Cterminal of residue 110. This site is ␤-sheet conformationspecific and is not recognized by control RNAs. Taken together, these data provide for the first time a specific ligand for a disease conformation-associated site in a region of PrP critical for conformational conversion. We also provide evidence that SAF-93 inhibits the accumulation of PrP res in the cell-free conversion assay under near physiological conditions.

MATERIALS AND METHODS
In Vitro Selection-All oligonucleotides used in this study were described before (29). RNAs used for in vitro selection were produced by in vitro transcription with T7 RNA polymerase in presence of 2Ј-fluoromodified pyrimidine nucleotide triphosphates (TriLink BioTechnologies, Inc., San Diego, CA), together with unmodified purine ribonucleotides in an optimized transcription buffer (30). The 2Ј-F-RNA transcripts were purified by electrophoresis on a 10% (w/v) denaturing polyacrylamide gel in TBE buffer. The pool of gel-purified 2Ј-F-RNA was heat denatured for 2 min at 95°C in deionized and filter-sterilized water, refolded for 10 min at room temperature in HMKN buffer (20 mM Hepes pH 7.2, 10 mM MgCl 2 , and 50 mM KCl, 100 mM NaCl), before being used for the selection process. The refolded 2Ј-F-RNA pool (5 nmol) was incubated with SAF purified from approximately one-half of a hamster brain, prepared as described below. Before each round of selection an aliquot of SAF was sonicated in a cup-horn probe with three pulses of 1 min each with an amplitude set at 40 and an output of 20 watts. The binding reaction was done at room temperature in HMKN buffer for 4 h. After partitioning the binding reaction was centrifuged for 1 h at 25,000 ϫ g at 10°C. The amount of unbound 2Ј-F-RNA in this first supernatant was stored at Ϫ20°C. In order to remove nonspecifically bound 2Ј-F-RNA, the pellet containing 2Ј-F-RNA-SAF complex was washed three times with 100 l of HMKN buffer. The supernatants from each wash were pooled with the first supernatant and the amount of unbound 2Ј-F-RNA determined by spectrophotometer (GeneQuant, Amersham Biosciences). The bound RNA was converted to cDNA by reverse transcription with Thermus thermophilus (Tth) DNA polymerase at 70°C for 20 min followed by 15 cycles of PCR amplification following the protocol provided by the supplier (Promega). The resulting PCR products were used as template for in vitro transcription to produce RNA for the next round of selection. The pool of RNA from round 7 was converted into DNA and cloned into pUC18 (29).
Preparation of Scrapie-associated Fibrils for Selection-SAF were prepared from the brains of hamsters that were infected with the 263K strain of scrapie (31). SAF were prepared without proteinase K treatment as described by Hope et al. (32). This preparation should include all variant PrP conformations and PK sensitivities associated with infectivity in this strain of prion. The final pellet (P285) was washed several times in water to remove traces of sarcosinate, before being resuspended in HMKN buffer pH 7.2 containing 0.02% (w/v) azide and stored at ϩ4°C.
Protein Preparation and Refolding-Truncated bovine PrPs, corresponding to residues 90 -230 and 110 -230 (human numbering) with His-tag at the N terminus were cloned, expressed, and purified as described before (33). Full-length bovine PrP corresponding to residues 23-230 without any tag was obtained by PCR amplification of genomic DNA and inserted as BglII-EcoRI restriction fragments into expression plasmid pMG939 (34) and amplified in Escherichia coli K12 1B392/ pACYRIL, which overexpresses rare arginine, isoleucine, and leucine tRNAs. Cultures were grown to saturation in Terrific Broth containing 100 g/ml ampicillin and 15 g/ml chloramphenicol at 30°C then diluted 400-fold. In late-log phase, expression of PrP was induced by raising the temperature from 30°C to 45°C for 10 min, followed by incubation at 42°C for 5 h. The cells were then harvested by centrifugation at 10 000 ϫ g for 15 min. The pellet was resuspended in ice-cold lysis buffer (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 2 mM phenylmethylsulfonyl fluoride, 10 g/ml lysozyme, 10 g/ml DNase I, 1 mg/ml sodium deoxycholate) and incubated for 30 min at 37°C. The solution was then centrifuged at 10,000 ϫ g for 30 min, and the supernatant discarded. The pellet was washed twice by resuspension in lysis buffer with centrifugation at 10,000 ϫ g for 10 min between each wash. Proteins in the pellet were dissolved by suspending it in buffer A (100 mM sodium phosphate, 10 mM Tris, pH 8.0, 8 M urea, and 10 mM 2-mercapthoethanol) and incubating for 30 min with gentle mixing. Cell debris and insoluble material were removed by centrifugation at 15,000 ϫ g for 15 min. The supernatant was loaded onto a Ni-NTA-Sepharose column (Qiagen Ltd.) pre-equilibrated with buffer A. After washing the column with the same buffer, bound proteins were eluted with buffer B (100 mM sodium phosphate, 10 mM Tris, pH 4.5, 8 M urea, and 10 mM 2-mercaptoethanol). For further purification the eluate from the column was diluted 1:2 with buffer C (50 mM Hepes pH 8.0, 8 M urea, and 10 mM 2-mercapthoethanol) and loaded onto a cation exchange chromatography column, SP-Sepharose (Amersham Biosciences). Recombinant PrP was eluted with buffer C supplemented with 1.5 M NaCl. Eluted fractions of recombinant PrP were pooled, and disulfide bonds were oxidized by stirring overnight in a 2:1 molar excess of CuCl 2 . After oxidation, the protein solution was dialyzed against 50 mM sodium acetate, pH 5.5, 1 mM EDTA, with several changes of buffer. Finally, the recombinant PrP was applied onto a size-exclusion chromatography column (Superdex 75 HR 10/30, Amersham Biosciences) equilibrated and eluted with 50 mM sodium acetate, pH 5.5.
The purified PrPs were refolded into native ␣-helical conformation or ␤-oligomeric isoform following the protocol described by Baskakov et al. (35). Size exclusion chromatography were performed at room temperature with a flow rate of 1 ml/min using either a TSK-3000 or a BIOSEP-S3000 HPLC column (300 ϫ 7.8 mm) from Phenomenex, equilibrated in 20 mM sodium acetate, pH 3.7, 0.2 M NaCl, and 1 M urea. The main peak corresponding to monomeric ␣-helical rich form or ␤-oligomer isoform of PrP was manually collected and assessed by CD. The protein concentration was determined spectrophotometrically using the molar extinction coefficient (⑀ 280 ) of 19 890 M Ϫ1 cm Ϫ1 for boPrP 90 -230 , and of 62,280 M Ϫ1 cm Ϫ1 for boPrP  . The PrP peptide spanning the region 106 -126 was synthesized, purified, and analyzed as described before (36).
Circular Dichroism-CD spectra were recorded with a Jasco J-720 CD spectrometer using the following parameters: scanning at 200 nm/ min, bandwidth of 2 nm, data spacing of 0.1 nm and using a 0.05-or 0.02-cm cuvette. Twenty individual scans were averaged, and the background spectra were subtracted.
Affinity of Monoclonal Aptamer for SAF-All RNAs used in this study were 32 P-labeled at the 5Ј-end as described before (29). Scrapieassociated fibrils from 263K strain of scrapie were prepared as described above. The final pellet from one hamster brain was resuspended in 200 l of HMKN buffer and submitted to ultrasonication. For homologous competition, 2 l of either proteinase K-treated or -untreated SAF, were incubated with 7000 cpm of 32 P-labeled monoclonal SAF-93 in HMKN buffer containing 1 g of tRNA and the mixture incubated for 1 h at room temperature. Cold homologous aptamer was added to each tube in 2-fold increasing concentrations from 0.25 to 8192 nM and incubated for 1 h. The displaced hot aptamer in each tube was separated from the remaining bound aptamer by centrifugation at 25,000 ϫ g and 10°C for 20 min. The resulting pellet was resuspended in water, and the amount of radiolabeled aptamer measured by Cerenkov counting. Each competition experiment was done in triplicate, and the data were analyzed by GraphPad PRISM.
Affinities of Monoclonal Aptamer for PrPs-Radiolabeled SAF-93 or control aptamer 19 (10,000 cpm Cerenkov) were heat-denatured in FIG. 1. Sequence alignment of isolated aptamers. The random regions from six SAF-aptamer clones were aligned using Pileup from GCG, Wisconsin.
water for 1 min at 95°C, cool to room temperature for 3 min before allowing it to refold in presence of 20 mM sodium acetate pH 5.2, 150 mM NaCl, 50 mM KCl, and 5 mM MgCl 2 for 10 min at room temperature. The refolded aptamer was then mixed with a range of increasing concentrations of PrP in various forms and lengths in presence of 1 M urea and 1 g of tRNA as nonspecific competitor in a total volume of 20 l. The binding reactions were incubated for 2 h at room temperature. In order to separate the free aptamer from PrP-aptamer complex we have used Strataclean™ resin (Stratagene) that has the property to pull-down proteins and RNA-protein complex by simple and quick centrifugation. Three l of Strataclean™ resin (Stratagene) was added to the binding reaction, and the mixture was vortexed and incubated for 1 min at room temperature before spinning in a benchtop centrifuge at top speed for 1 min. The resulting supernatant was transferred to a fresh tube, and the pellet washed with 20 l of binding buffer without tRNA. The supernatant from the washed pellet was added to the previous supernatant, this supernatant formed the unbound aptamer. The complex aptamer-PrP remaining in the pellet was resuspended in 20 l of binding buffer. The amount of radioactivity in both the supernatant, and the pellet was measured by Cerenkov counting. The binding data were analyzed by GraphPad Prism using a non-linear regression curve fitting and one site binding hyperbola equation: Slot Blots-We have used Bio-Dot® SF microfiltration apparatus and PVDF membrane (0.2-m pore size) from BioRad assembled following the manufacturer's instructions. 90 pmol of PrPs in various forms and lengths were applied by vacuum onto the PVDF membrane that was previously equilibrated in 20 mM sodium acetate buffer, pH 4.4. The PVDF membrane containing PrPs was then incubated for 1 h in 15 ml of blocking solution containing 5ϫ Denhardt's solution in HMNK buffer pH 7.2 and 20 g/ml tRNA at room temperature. The blocking solution was replaced with a fresh one and supplemented with 400,000 cpm radiolabeled and refolded SAF-93 or control aptamer 19 and allowed to incubate for 2 h or overnight at 25°C inside a hybridization oven under constant rolling. The membrane was then washed twice with 15 ml of HMNK buffer pH 7.2, wrapped inside a saran film and exposed to Storage phosphor screen for at least 2 h followed by scanning with the Storm imaging system and the autoradiographs analyzed with Image-Quant software from Molecular Dynamics.
SDS-PAGE and Western Blots-PrP proteins were separated under denaturing and reducing conditions on SDS-PAGE (37) using 18% polyacrylamide precast Novex gels (Invitrogen). After electrophoresis the gel was blotted (38) onto PVDF membrane and probed with either mAb 6H4 (Prionics) or mAb FH11 (TSE Resource center, IAH, Compton, UK) followed by a secondary antibody conjugated to alkaline phosphatase and detected with ECF substrate (Amersham Biosciences) using the chemiluminescent mode on the Storm imaging system from Molecular Dynamics. The detection with radiolabeled aptamers was performed as described under "Slot Blot." Cell-free Conversion Assay-The cell-free conversion assay without guanidinium chloride was as described by Kirby et al. (39) and based on that used by Horiuchi and Caughey (40). Briefly PrP Sc was sonicated and ϳ1 g incubated with 200 ng of [ 35 S]shaPrP in conversion buffer (50 mM citrate, pH 6.5, 50 mM KCl, 10 mM MgCl 2 , 100 mM NaCl, 0.1% (v/v) Nonidet P-40) with or without SAF-93 or control RNA, for 24 h at 37°C in a 20-l volume reaction. One-twentieth of the reaction was removed for analysis without proteinase K treatment, and the remaining was treated with 60 g/ml of proteinase K for 1 h at 37°C. Proteinase K digestion was stopped by addition of Pefabloc to 1 mM. All samples were precipitated with 20 g of bovine serum albumin and 4 volumes of ice-cold methanol at Ϫ20°C. The resulting pellet was boiled for 10 min in SDS-sample buffer and analyzed by 15% SDS-PAGE, fixed, dried, and exposed to film. Autoradiographs were quantified using Phoretix Gel Analysis software.

RESULTS
Isolation of PrP-specific 2Ј-F-RNA Aptamers-A DNA library comprising a 49 nucleotide randomized region was synthesized. In theory, this library could comprise 4 49 ϭ 8 ϫ 10 28 distinct sequences although in practice, only ϳ10 14 of these are sampled during the selection process (41). The library was transcribed to produce 2Ј-F-pyrimidine-substituted RNA and subjected to repeated cycles of in vitro selection for transcripts that bound to insoluble non-PK-treated PrP Sc fibrils isolated from the brains of 263K scrapie-infected hamsters. PrP-binding 2Ј-F-substituted nucleic acids became a significant fraction of the population by selection round 3, and the pool from round 7 was cloned as cDNA.
The sequences of the clones encoding ligands for PrP show that they probably derived from distinct members of the initial library (see Fig. 1). Some stretches of weak sequence homology between the six clones might reflect some convergence of structural features. In this study we present the data obtained from investigating the interaction between SAF-93 and PrP protein in various lengths and conformations.
Affinities of Aptamer for PrP Forms-The first step toward the characterization of the selected monoclonal 2Ј-F-SAFaptamers was to determine the strength of their binding to SAF used as a target during the in vitro selection. We have used a homologous competition binding assay in which we have measured the binding of a single concentration of 32 P-end-labeled SAF-93 to SAF preparation in presence of an increasing concentration of the same unlabeled aptamer (Fig. 2). The fitting of homlogous competition data to a one site binding model (GraphPad Prism Software, Inc.) gave an estimated K d of 23.4 Ϯ 1.2 nM without proteinase K (PK)-treatment of SAF and a K d of 26.9 Ϯ 1.1 nM against PK-treated SAF. These data indicated that the aptamer binding site lay in the PK-resistant core of disease isoform PrP, comprising residues 90 -230.
The biochemical and biophysical properties associated with the SAF material made it difficult to perform routine investigations in order to gain better understanding of the interaction between our aptamer and the disease-associated form of PrP. Therefore we decided to use purified bacterially-expressed recombinant bovine PrP of various lengths that had been refolded into ␣-helical conformation or ␤-oligomer form with an intact disulfide bridge as described by Baskakov et al. (35). The ␣-form and ␤-form (Fig. 3A) of bovine PrP eluted from the SEC with retention times very similar to those published for mouse PrP 89 -231 (35,42). The peaks that eluted at 7.19 and 9.06 min were collected and analyzed by CD (Fig. 3B) and found indeed to correspond to ␤and ␣-form of PrP 90 -230 , respectively. Similar results were obtained with the full-length PrP 23-230 (data not shown).
We used the forms of recombinant bovine PrP described above in an equilibrium binding assay with radiolabeled SAF-93 in parallel with a radiolabeled control aptamer SA-19 that was isolated against streptavidin (29) in order to determine the equilibrium binding constant of the interactions (Fig.  4). and full-length ␤-PrP 23-230 was found to be 32.2 nM (Fig. 3C) and 16.0 nM (Fig. 3D), respectively. Significantly, these affinities were close to those determined for the PK-treated and -untreated SAF (see above), indicating that SAF-93 recognized a binding site that was common between SAF and in vitro refolded ␤-form PrPs and that the aptatope was conserved between these two species.
As a complementary approach, the binding of SAF-93 to the same PrPs was assayed under native conditions by slot blotting. The results revealed that preserving the conformation of the aptamer binding site was essential for maintaining the discriminatory power between ␣and ␤-form PrPs, and that was most marked for PrP 90 -230 (Fig. 4, A and B). SAF-93 could only bind to PrP 90 -230 in its ␤-form, while the binding to ␣-form was virtually abolished (Fig. 4, A and B). Under these conditions, SAF-93 bound equally well to ␣and ␤-form PrP 23-230 (Fig. 4, A and B). Both approaches led to qualitatively similar conclusions. First, the aptamer binding site was occluded in ␣-PrP 90 -230 and only became accessible to the aptamer in ␤-PrP 90 -230 . Second, the conformation-specific binding site for SAF-93 resided downstream of residue 90. Third, the binding site for SAF-93 in the unstructured N-terminal segment upstream of residue 90 was conformationally nonspecific and could represent a region of PrP with high nucleic acid binding propensity.
Mapping the Binding Site of SAF-93 on PrP-Since the PK treatment of SAF and the truncation of recombinant bovine PrP did not affect the binding properties of the aptamer, it was clear that a binding site resided within the PK resistant core of PrP (residues 90 -230). In attempting to define the binding site more precisely, we were helped by the fortuitous appearance of a defined degradation product of PrP in Western blotting experiments (see Fig. 5).
Coomassie Blue staining of the gel (Fig. 5A) showed one major band for ␣-PrP 23-230 with the expected molecular mass of 23.6 kDa, whereas the ␣-PrP 90 -230 revealed two bands of approximately equal intensities: a band with a molecular mass of 15.9 kDa as one would expect from the bovine PrP residues 90 -230, and another band with a lower molecular mass (Fig.  5A). Mass spectrometry analysis of the protein sample in solution revealed that it contained both PrP 90 -233 and PrP 110 -230 (data not shown). In the absence of other signals we concluded with 99% confidence that the fragment 110 -230 was what caused the second band on the gel. Similar proteolysis digestion has been reported for recombinant murine prion protein (20,43). When the blot derived from this gel was probed with [ 32 P]SAF-93 (Fig. 5B), two clear bands were detected that corresponded to PrP  and PrP 90 -230 . The probing with 32 Pcontrol aptamer, SA-19, revealed a similar detection pattern, but with significantly lower intensities than those obtained with SAF-93 (see Fig. 5C). In fact, the intensity of the displayed digital picture had to be substantially increased in order to make the two bands visible. The identity of the two bands detected with SAF-93 was further confirmed by probing similar blots with mAb 6H4 (Fig. 5D) and as expected, PrP 90 -230 was not detected by mAb FH11 (Fig. 5E). Most significantly, the band that corresponded to PrP 110 -230 was not detected by SAFaptamer 93, indicating that an essential part of the aptamer binding site spanned residues 90 -110. In order to further characterize the interaction between SAF-93 and PrP 110 -230 , the latter was cloned, expressed, purified, and characterized by SDS-PAGE (Fig. 6A). We confirmed that under denaturing conditions and using North-Western blot assay SAF-93 did not bind to PrP 110 -230 (Fig. 6B) nor did control aptamer SA-19 (Fig.  6C), whereas the reactivity of this protein toward mAb 6H4 was maintained as expected (Fig. 6D). Importantly, when PrP 110 -230 was refolded into ␤-form as before, we found that the binding affinity toward SAF-93 was restored to values (Fig. 6E) very close to those calculated for ␤-PrP 23-230 and ␤-PrP 90 -230 (see Fig. 3, C and D for comparison). When PrP 110 -230 was folded into the ␣-form its interaction with SAF-93 was abolished (Fig. 6F). The control aptamer SA-19 bound only weakly to ␤-PrP 110 -230 (Fig. 6E) and not at all to ␣-PrP 110 -230 (Fig. 6F). These data provided strong evidence for the existence of a conformation-specific SAF-93 binding site downstream of residue 110. This conformation-specific aptatope is not recognized by the aptamer when the PrP 110 -230 is denatured or folded into the ␣-form.
Numerous studies have indicated that the region 106 -126, just upstream of the first ␤-sheet, is a conformationally flexible part of the protein (36,44,45) and is neurotoxic (46 -50). Because of the overlap between this "neurotoxic peptide" region and that defined as containing an SAF-93 binding site, we decided to test whether SAF-93 could bind to peptide PrP 106 -126 . By adjusting the buffer conditions, we could fold this peptide into a ␤-form, or retain it in a random coil conformation (see Fig. 7A). The affinity of SAF-93 was found to be very similar between the unstructured (Fig. 7B) and the ␤-form (Fig. 7C) of PrP 106 -126 , in the order of 400 nM, which is very close to the affinity value measured for ␣-PrP 90 -230 determined above. However, as the affinity for ␤-forms PrP had been found to be in the 20 -30-nM range (see above), this indicates that the region 106 -126 might contain only part of the aptamer binding site or that the folding of the peptide into ␤-form did not Inhibition of Cell-free Conversion of PrP by Aptamer-The preferential binding of SAF-93 to the disease-related ␤-form PrP prompted us to investigate the capacity of this artificial ligand to interfere with the conformational switch that leads to the accumulation of PrP res . We have used the cell-free conversion system under near-physiological conditions in order to give more biological relevance to the assay as a model for conversion. In this study, recombinant hamster PrP was radiolabeled during expression, purified from bacteria and refolded in vitro (39). [ 35 S]ShaPrP was incubated in the presence of unlabeled PrP Sc isolated from 263K scrapie-infected hamster brains, with or without control RNA or with increasing concentrations of SAF-93 in a guanidine-free conversion buffer at 37°C for 24 h. Following proteinase K digestion, SDS-PAGE analysis and autoradiography (Fig. 8A), the intensity of the band at ϳ17.4 kDa, representing the PK-resistant core of [ 35 S]ShaPrP res , decreased with increasing amounts of SAF-93. Even at the highest concentration, the unselected 2Ј-F-RNA library produced no significant inhibition of conversion (Fig. 8A). Similar results were obtained when the control SA-19 aptamer (29) was used instead of the unselected RNA library (data not shown). These data showed that there was a specific SAF-93-dependent inhibition of the accumulation of PrP res protein. The average IC 50 titer was estimated from the dose response graph in Fig. 8B and found to be around 40 nM for SAF-93. DISCUSSION Here we describe the isolation of novel nucleic acid ligands or aptamers for PrP, the key protein in the pathogenesis and the transmission of vCJD, BSE, and other TSEs. Because of the high propensity of recombinant PrP to interact with nucleic acid via its unstructured N-terminal domain, it was of paramount importance to discriminate between specific and nonspecific interactions. The first indication of specificity of aptamer SAF-93 came from the homologous competition assay. The fact that binding affinity was not affected by proteinase K treatment of SAF material, suggested that the aptatope lay in the PK-resistant core of PrP, comprising the 141 amino acids in the C-terminal portion of the protein, from residue 90 to 230. Moreover, the binding affinity measured for ␤-oligomer PrP spanning residues 90 -230 was very close to that of SAF mate-rial that has been treated with proteinase K. These suggest conformational similarities between SAF and ␤-oligomer PrP 90 -230 . In the light of the recently published data by Baskakov et al. (35), it is very important to stress at this point that these structural similarities might be just subtle enough to account for the existence of a common ␤-sheet rich aptatope between SAF material and ␤-oligomer PrP, even though the latter is not considered to be on the kinetic pathway for amyloid formation and cannot be regarded as a substructure of the fibrillar form (35). Because PrP Sc and PrP C both maintain the disulfide bridge (51) we believe that it was more relevant to use the non-reduced PrP to generate the ␤-oligomer form. However, we found that SAF-93 maintained the same binding affinity against ␤-form PrP with no disulfide bridge (33). 2 The simple explanation for this is that the absence of the disulfide bridge does not affect local folding into a ␤-sheet of the region downstream of residue 90 where SAF-93 aptatope occurs. In agreement with this explanation is the fact that all structures of recPrPs showed that the disulfide bridge is located between helix B and helix C, which seemed to maintain their global conformation in the plausible models for the structure of PrP Sc (52).
The molecular basis for the preferential recognition of ␤-oligomer PrP by SAF-93 is still unclear, but our data reinforce and extend previous reports that RNA interactions with proteins are generally more favorable with ␤-sheet than ␣-helix (53). We were fully aware of the fact that the analysis of the binding data with an equation that governed one binding site might not reflect the true complexity of the interaction between aptamers and PrP. However, we have analyzed the binding data between SAF-93 and ␤-PrP 23-230 with an equation that described the existence of two independent binding sites. We found an affinity value for the nonspecific site (1) of 8 M while the conformation-specific site (2) had an affinity value of 16 nM; a value that remained identical to that obtained by fitting the data to one binding site equation. This reinforces the idea that the binding of SAF-93 to ␤-form PrPs 23-230, 90 -230, and 110 -230 could be accounted for by the existence of only the conformation-specific and low affinity site (2).  The data from the mapping of the aptamer binding site onto PrP conformations provided a basis to explain the aptamer-dependent inhibition effect on the conversion of [ 35 S]ShaPrP sen to PrP res . Although the molecular mechanisms of this conversion are as yet to be elucidated, direct interaction between PrP Sc and [ 35 S]ShaPrP sen is critically important. Two non-exclusive mechanisms could explain the inhibition of the conversion by aptamer. First, the aptamer could bind directly to the PrP Sc used as a template for conversion, therefore preventing any direct and critical contact between [ 35 S]ShaPrP sen and PrP Sc . Second, and by virtue of its high affinity to the ␤-rich PrP, the aptamer might bind to any escaped de novo formed ␤-rich PrP and thereby stall the conversion.
Antibody mapping has indicated sites of interaction between PrP sen and PrP res . One or more of these areas were adjacent to the C terminus in the three-dimensional structure of PrP sen (40,54). One such surface contains residues 119 -138 and others contain residues 165-174 and 206 -223. Antibodies that bind to these areas are capable of interfering with PrP res accumulation (40). The inability of antibodies recognizing residues 90 -104 and 106 -115 to inhibit PrP res formation has been attributed to conformational rearrangements occurring during the process of conversion that might induce the antibody to dissociate (40). Indeed, many biochemical and spectroscopic studies concord to suggest that the region 90 -124 undergoes a major conformational change during the formation of PrP res (2,5,36) and (55)(56)(57). Intriguingly, the site we suggest for SAF-93 binding lies in a region defined recently by us to be involved in competing intra-chain interactions that lead to ␤-sheet-rich fibril formation in vitro (36). It is possible that SAF-93 is able to inhibit the fibril-promoting interactions between residues 104 -122 and 129 -141 that such studies with isolated peptides have indicated.
Very recently, Proske et al. (58) have isolated an aptamer to a peptide spanning residues 90 -129 of PrP, which has the ability to reduce the level of PrP Sc accumulation in persistently prion-infected neuroblastoma cells. It is intriguing that there is a potential overlap between the binding site of our SAF-93 and their aptamer DP7 (58). It should be noted, however, that region 90 -129 includes part of the N-terminal, conformation-ally non-selective nucleic acid binding region of PrP (this study) and that these aptamers do not share any similarity in either the primary or the secondary structures. It remains to be determined whether DP7 aptamer exhibits any differential affinity between ␣-helix rich and ␤-oligomer PrP, as it was selected using a short peptide of undefined structure, rather than infectious material, as employed here. Furthermore, aptamer DP7 comprises pyryimidine 2Ј-amino, 2Ј-deoxyribonucleotides, rather than the 2Ј-F modifications employed in this work. It is known that different substitutions at the 2Јposition result in changes in binding activity, probably by influencing the functional folding of RNA (29). It is significant that both aptamers were able to interfere with the conversion of PrP C into PrP Sc . Any comparison between the potency of the two aptamers in this regard must be made with caution, as the in vitro conversion assays were fundamentally different. Nevertheless, there is reason to believe that aptamer SAF-93, described here, may be the more potent inhibitor as it produced ϳ2-fold inhibition of conversion at 40 nM, compared with 700 nM, for comparable effects with DP7 (58).
Our data strengthen the existing literature about the intrinsic property of the ␣-helical form of PrP to interact with nucleic acids in general and RNA in particular, and extend the high affinity of the prion protein for RNA to the ␤-form diseaserelated PrP. These properties should be investigated further in order to gain more insight into the putative physiological role of PrP in RNA metabolism (25,26,28) and should provide ways to trap PrP in a disease-inactive conformation that might cleanse physiological fluids.