Structural Determinants of Conformationally Selective, Prion-binding Aptamers
2004; Elsevier BV; Volume: 279; Issue: 13 Linguagem: Inglês
10.1074/jbc.m310928200
ISSN1083-351X
AutoresNatalie M. Sayer, Matthew Cubin, Alexandre Rhie, Marc D. Bullock, Abdessamad Tahiri‐Alaoui, William James,
Tópico(s)RNA regulation and disease
ResumoWe have recently described the isolation of 2′-fluoropyrimidine-substituted RNA aptamers that bind selectively to disease-associated β-sheet-rich forms of the prion protein, PrP, from a number of mammalian species. These aptamers inhibit the accumulation of protease-resistant forms of PrP in a prion-seeded, in vitro conversion assay. Here we identify the minimal portions of two of these aptamers that retain binding specificity. We determine their secondary structures by a combination of modeling and solution probing. Finally, we identify an internal site for biotinylation of a minimized, synthetic aptamer and use the resultant reagent in the detection of abnormal forms of PrP in vitro. We have recently described the isolation of 2′-fluoropyrimidine-substituted RNA aptamers that bind selectively to disease-associated β-sheet-rich forms of the prion protein, PrP, from a number of mammalian species. These aptamers inhibit the accumulation of protease-resistant forms of PrP in a prion-seeded, in vitro conversion assay. Here we identify the minimal portions of two of these aptamers that retain binding specificity. We determine their secondary structures by a combination of modeling and solution probing. Finally, we identify an internal site for biotinylation of a minimized, synthetic aptamer and use the resultant reagent in the detection of abnormal forms of PrP in vitro. Transmissible spongiform encephalopathies (TSEs) 1The abbreviations used are: TSE, transmissible spongiform encephalopathy; SA, streptavidin. involve spongiform degeneration and astrocyte gliosis in the central nervous system leading to dementia and death (1Hur K. Kim J.I. Choi S.I. Choi E.K. Carp R.I. Kim Y.S. Mech. Ageing Dev. 2002; 123: 1637-1647Crossref PubMed Scopus (45) Google Scholar). They can have genetic, sporadic, or infectious etiologies, exemplified by fatal familial insomnia, sporadic Creutzfeldt-Jakob disease, and variant Creutzfeldt-Jakob disease in humans, respectively. They all involve the modification of prion precursor protein PrPc, which is constitutively expressed in mammalian cells, particularly neurons, as a GPI-anchored plasma membrane glycoprotein. During many TSE diseases, infectivity is found to be associated with amyloid deposits of PrP, in a form referred to as PrPSc (2Prusiner S.B. McKinley M.P. Bowman K.A. Bolton D.C. Bendheim P.E. Groth D.F. Glenner G.G. Cell. 1983; 35: 349-358Abstract Full Text PDF PubMed Scopus (833) Google Scholar), and lack of a gene for PrP renders animals insusceptible to prion diseases (3Prusiner S.B. Groth D. Serban A. Koehler R. Foster D. Torchia M. Burton D. Yang S.L. DeArmond S.J. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 10608-10612Crossref PubMed Scopus (419) Google Scholar, 4Bueler H. Aguzzi A. Sailer A. Greiner R.A. Autenried P. Aguet M. Weissmann C. Cell. 1993; 73: 1339-1347Abstract Full Text PDF PubMed Scopus (1814) Google Scholar). These and related observations support the prion theory of TSE etiology, in which self-templated conformational changes in PrP are necessary and sufficient for disease progression and transmission (5Prusiner S.B. Science. 1982; 216: 136-144Crossref PubMed Scopus (4123) Google Scholar). The encipherment of the pathological properties of TSEs within the conformation of PrP is not fully understood, but whereas the membrane-proximal domain of PrPc is mainly α-helical in structure, PrPsc is dominated by β-sheets (6Pan K.M. Baldwin M. Nguyen J. Gasset M. Serban A. Groth D. Mehlhorn I. Huang Z. Fletterick R.J. Cohen F.E. Prusiner S.B. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 10962-10966Crossref PubMed Scopus (2083) Google Scholar). In contrast to PrPC, PrPsc is insoluble in nondenaturing conditions and is partially resistant to proteases (7Prusiner S.B. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 13363-13383Crossref PubMed Scopus (5166) Google Scholar). Diagnostic tests for the in vivo identification of prions are not yet ideal. Disease cannot be confirmed until clinical signs appear or even until a post mortem of brain tissue shows spongiform changes and prion accumulation using immunohistochemical methods. Only recently has it been possible to isolate antibodies that are able to differentiate between PrPc and PrPsc (8Paramithiotis E. Pinard M. Lawton T. LaBoissiere S. Leathers V.L. Zou W.Q. Estey L.A. Lamontagne J. Lehto M.T. Kondejewski L.H. Francoeur G.P. Papadopoulos M. Haghighat A. Spatz S.J. Head M. Will R. Ironside J. O'Rourke K. Tonelli Q. Ledebur H.C. Chakrabartty A. Cashman N.R. Nat. Med. 2003; 9: 893-899Crossref PubMed Scopus (242) Google Scholar), so current methods rely on a pretreatment with protease, which may obscure the presence of relevant, protease-sensitive conformations. The most promising of diagnostic tests have been using fluorescence, such as multi-spectral ultraviolet fluorescence looking at specific fluorescence of proteins when excited by UV radiation (9Rubenstein R. Gray P.C. Wehlburg C.M. Wagner J.S. Tisone G.C. Biochem. Biophys. Res. Commun. 1998; 246: 100-106Crossref PubMed Scopus (38) Google Scholar) and fluorescence correlation spectroscopy, which recognizes single fluorescent molecules in solution as they pass between the exciting laser beam and the objective of a confocal microscope (10Bieschke J. Giese A. Schulz-Schaeffer W. Zerr I. Poser S. Eigen M. Kretzschmar H. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 5468-5473Crossref PubMed Scopus (217) Google Scholar). Another interesting approach by Safar et al. (11Safar J.G. Scott M. Monaghan J. Deering C. Didorenko S. Vergara J. Ball H. Legname G. Leclerc E. Solforosi L. Serban H. Groth D. Burton D.R. Prusiner S.B. Williamson R.A. Nat. Biotechnol. 2002; 20: 1147-1150Crossref PubMed Scopus (202) Google Scholar) is that of a conformation-dependent immunoassay; this uniquely quantifies PrP isoforms by the binding pattern of different antibodies to both denatured and native forms. There has been some evidence that PrPsc can be detected in the blood of infected rodents (12Brown P. Cervenakova L. McShane L.M. Barber P. Rubenstein R. Drohan W.N. Transfusion. 1999; 39: 1169-1178Crossref PubMed Scopus (270) Google Scholar, 13Brown P. Rohwer R.G. Dunstan B.C. MacAuley C. Gajdusek D.C. Drohan W.N. Transfusion. 1998; 38: 810-816Crossref PubMed Scopus (288) Google Scholar), as well as in the urine of scrapie-infected hamsters (14Shaked G.M. Shaked Y. Kariv-Inbal Z. Halimi M. Avraham I. Gabizon R. J. Biol. Chem. 2001; 276: 31479-31482Abstract Full Text Full Text PDF PubMed Scopus (164) Google Scholar). The amount of PrPsc outside the central nervous system is very small compared with that in the brain, so it might be necessary to amplify the small amount there using, for example, the cyclic PrPsc amplification process (15Saborio G.P. Permanne B. Soto C. Nature. 2001; 411: 810-813Crossref PubMed Scopus (1022) Google Scholar). For a review of these and other diagnostic assays see Ref. 16Ingrosso L. Vetrugno V. Cardone F. Pocchiari M. Trends Mol. Med. 2002; 8: 273-280Abstract Full Text Full Text PDF PubMed Scopus (40) Google Scholar. Given the limitations of antibodies, aptamers provide an alternative approach to the ligand-based detection of PrPsc. Aptamers are nucleic acids isolated in vitro for their ability to bind a molecule of interest by the SELEX procedure (17Tuerk C. Gold L. Science. 1990; 249: 505-510Crossref PubMed Scopus (8166) Google Scholar). This involves cycles of binding and amplification starting from about 1014 different nucleic acid molecules, which eventually results in the isolation of a few sequences that have good affinity for targets such as CD4 (18Kraus E. James W. Barclay A.N. J. Immunol. 1998; 160: 5209-5212PubMed Google Scholar), streptavidin (19Tahiri-Alaoui A. Frigotto L. Manville N. Ibrahim J. Romby P. James W. Nucleic Acids Res. 2002; 30: 1-9Crossref PubMed Scopus (50) Google Scholar), and HIV-1 gp120 (20Sayer N. Ibrahim J. Turner K. Tahiri-Alaoui A. James W. Biochem. Biophys. Res. Commun. 2002; 293: 924-931Crossref PubMed Scopus (54) Google Scholar, 21Khati M. Schumann M. Ibrahim J. Sattentau Q. Gordon S. James W. J. Virol. 2003; 77: 12692-12698Crossref PubMed Scopus (151) Google Scholar). Protein binding by aptamers results mainly from hydrogen bonds and electrostatic forces, which means that the best targets are often basic, positively charged molecules. Aptamers can have a subnanomolar affinity and a high specificity (for a review see Ref. 22James W. Curr. Opin. Pharmacol. 2001; 1: 540-546Crossref PubMed Scopus (43) Google Scholar), Because of the number of different potential ligands that can be screened by the SELEX process, aptamers have often been raised against molecules when other approaches have failed. They have many advantages over antibodies in that they are nonimmunogenic, smaller, and usually of a better affinity and can be chemically synthesized. This, together with their contrasting physico-chemical properties to antibodies, makes aptamers potentially useful reagents for the detection of previously overlooked conformations of PrP. We have recently described the isolation of 2′-fluoropyrimidine-substituted RNA aptamers that were selected from randomized libraries using infectious prions as the target material (23Rhie A. Kirby L. Sayer N. Wellesley R. Disterer P. Sylvester I. Gill A. Hope J. James W. Tahiri-Alaoui A. J. Biol. Chem. 2003; 278: 39697-39705Abstract Full Text Full Text PDF PubMed Scopus (180) Google Scholar). One of these aptamers, SAF-93 was shown to bind to two regions of PrP: (i) a conformationally insensitive, diffuse region lying between residues 23 and 110 that binds nonspecifically to nucleic acids and (ii) a site lying between residues 110 and 230. The latter site was occluded in the normal α-helix-rich conformation of the protein but exposed in the disease-associated β-sheet-rich forms and was not bound by control nucleic acids. SAF-93 also inhibited the conversion of α helix-rich to β sheet-rich forms in a prion-nucleated, in vitro conversion assay (23Rhie A. Kirby L. Sayer N. Wellesley R. Disterer P. Sylvester I. Gill A. Hope J. James W. Tahiri-Alaoui A. J. Biol. Chem. 2003; 278: 39697-39705Abstract Full Text Full Text PDF PubMed Scopus (180) Google Scholar). SAF-93 is therefore an interesting ligand for PrP and may reveal conformations of PrP not yet identified. Before the properties of this and another aptamer, SAF-76 (also described in this paper), could be exploited, more information was required about the structural features essential for their binding to prions. In this paper, we present the results of experiments designed to facilitate the use of prion-binding aptamers as molecular reagents for the investigation of prion diseases. Firstly we determined the solution structure of the aptamers and mapped the binding motifs onto the determined structure. We then characterized the minimal structural features that were responsible for specific binding and designed a novel method for determining a residue within the essential structure of the best aptamer that could be modified without affecting its affinity for PrP. With this information, we went on to synthesize modified aptamers for a range of analytical and diagnostic applications. Protein Preparation and Refolding—Full-length bovine PrP, corresponding to residues 23-230 (human numbering) was cloned, expressed, and purified as described before (23Rhie A. Kirby L. Sayer N. Wellesley R. Disterer P. Sylvester I. Gill A. Hope J. James W. Tahiri-Alaoui A. J. Biol. Chem. 2003; 278: 39697-39705Abstract Full Text Full Text PDF PubMed Scopus (180) Google Scholar). Briefly, the protein was expressed in Escherichia coli and purified from inclusion bodies by nickel-nitrilotriacetic acid affinity chromatography and reversed phase high pressure liquid chromatography. The purified PrP was refolded into native α-helical conformation or β-oligomeric isoform following the protocol described in Ref. 24Baskakov I.V. Legname G. Baldwin M.A. Prusiner S.B. Cohen F.E. J. Biol. Chem. 2002; 277: 21140-21148Abstract Full Text Full Text PDF PubMed Scopus (390) Google Scholar. Finally, the monomeric α-helical rich form and β-oligomer isoform of PrP were purified by high performance size exclusion chromatography (23Rhie A. Kirby L. Sayer N. Wellesley R. Disterer P. Sylvester I. Gill A. Hope J. James W. Tahiri-Alaoui A. J. Biol. Chem. 2003; 278: 39697-39705Abstract Full Text Full Text PDF PubMed Scopus (180) Google Scholar). The protein structures were checked using CD, as previously described (23Rhie A. Kirby L. Sayer N. Wellesley R. Disterer P. Sylvester I. Gill A. Hope J. James W. Tahiri-Alaoui A. J. Biol. Chem. 2003; 278: 39697-39705Abstract Full Text Full Text PDF PubMed Scopus (180) Google Scholar). In Vitro Transcription—Transcription of 2′-fluoropyrimidine-containing RNA was carried out largely as described in Ref. 25Heidenreich O. Kang S.H. Brown D.A. Xu X. Swiderski P. Rossi J.J. Eckstein F. Nerenberg M. Nucleic Acids Res. 1995; 23: 2223-2228Crossref PubMed Scopus (40) Google Scholar. Briefly, transcription by T7 RNA polymerase (New England Biolabs) was incubated at 37 °C for at least 6 h in 40 mm Tris, pH 8.1, 6 mm magnesium chloride, 1 mm spermidine, 5 mm dithiothreitol, and 1 mm NTPs. The products were then treated for 30 min at 37 °C with 50 units of DNase 1 (Roche Applied Science) in 150 mm sodium acetate, pH 5.2, 10 mm magnesium chloride. The reaction was stopped by phenol extraction, and the transcript was recovered by ethanol precipitation. 32P End Labeling of RNA—For 5′ labeling, the terminal 5′ phosphate was removed using bacterial alkaline phosphatase (New England Bio-labs) and replaced with [32P]phosphate from [γ-32P]ATP using T4 polynucleotide kinase (Roche Applied Science). Labeling at the 3′ end was achieved by ligating [32P]pCp to the 3′ end of the aptamer using T4 RNA ligase (New England Biolabs). Ligation was carried out for 2 h at 37 °C. The labeled RNAs were electrophoresed on 12% polyacrylamide, 8 m urea gels, visualized by autoradiography, and recovered by passive elution from gel slices. Identification of the Core Binding Sequence Using Boundary Experiments—A library of nested deletion fragments was generated from either 3′ or 5′ end-labeled aptamer by mild alkaline hydrolysis (26Aurup H. Williams D.M. Eckstein F. Biochemistry. 1992; 31: 9636-9641Crossref PubMed Scopus (128) Google Scholar) and partitioned into PrP-binding and nonbinding fractions using Strata-Clean™ (Stratagene), as described before (20Sayer N. Ibrahim J. Turner K. Tahiri-Alaoui A. James W. Biochem. Biophys. Res. Commun. 2002; 293: 924-931Crossref PubMed Scopus (54) Google Scholar). The products were analyzed by denaturing 18% PAGE. Enzymatic Probing and Footprinting—32P end-labeled RNA was digested in 1× HMNK, pH 7.2, buffer (20 mm HEPES, pH 7.2, 100 mm sodium chloride, 50 mm potassium chloride, 5 mm magnesium chloride, 0.04% Nonidet P-40) containing 1 μg of tRNA at 20 °C for 5 min with either nuclease V1 (Pierce; 5 × 10-3 unit) or S1 (Amersham Biosciences; 0.05 unit). Footprinting was achieved by incubating an identical aptamer preparation with prion protein for 1 h at 25 °C and then subjecting the mixture to appropriate nuclease digestion. The digestions were terminated by phenol extraction. The RNA was precipitated using ethanol, then dissolved in formamide buffer, and analyzed by denaturing PAGE (see below). Urea-denaturing PAGE of 32P-Labeled RNA—All of the samples for truncation and footprinting analysis were heated to 95 °C for 3 min in formamide loading buffer and run on 40-cm-long 18% or 20% polyacrylamide, 8 m urea gels at 25 watts for 3 h. A partial alkaline hydrolysate of the RNA along with a G residue ladder was run alongside analysis samples to aid sequence identification. The G residue ladder was generated by digestion of 50,000 cpm (Cerenkov) denatured RNA at 55 °C in 10 μl of 20 mm sodium citrate, 1 mm EDTA, 7 m urea, pH 4.6, with RNase T1 (Amersham Biosciences; 5 × 10-3 units). Binding Affinity of Aptamers—End-labeled aptamer (5000 cpm; Cerenkov) was heat-denatured at 95 °C for 2 min in water, cooled to room temperature, and refolded in 1× AMNK buffer (20 mm sodium acetate, pH 5.2, 100 mm sodium chloride, 50 mm potassium chloride, 5 mm magnesium chloride, 0.04% Nonidet P-40) for 10 min. After refolding, the aptamer was incubated with different concentrations of prion protein in the presence of 1 μg of tRNA in 30 μl for 1 h at 25 °C. The bound RNA was then separated from unbound using 0.1 volume of StrataClean™ resin and centrifuged at 13,000 rpm for 3 min, and the supernatant (containing unbound RNA) was removed. The resin was washed with 30 μl of 1× AMNK buffer and again centrifuged (the wash was pooled with the unbound RNA). The resin (complexed with PrP-bound RNA) was resuspended in 60 μl of 1× AMNK buffer, and the amounts of radioactivity corresponding to PrP-bound and unbound RNA were measured for each concentration of protein. The data were fitted using nonlinear curve fitting to a hyperbolic equation by Graph Pad Prism. Identification of Sites at Which Incorporation of Biotin Does Not Interfere with PrP Binding—Transcription of aptamer was done as described above, with the addition of 2 μm biotin-16-dUTP (Roche Applied Science). After completion, the transcription was dialyzed against water for 3 h instead of phenol extraction to remove nonincorporated 5-allyl biotinyl UTP. To isolate molecules that could bind a single streptavidin tetramer, the dephosphorylated, biotinylated RNA was incubated for 30 min with 30 μl of streptavidin (2 mg/ml). The complex was then run on an 8% Tris-borate native polyacrylamide gel, and the band corresponding to aptamer bound to one tetramer of streptavidin was cut out, passively eluted from the gel, and ethanol-precipitated. The gel-purified and biotinylated RNA was then 5′ end radiolabeled as described above. The double-labeled RNA was ethanol-precipitated and complexed with streptavidin as before. The complex was then incubated with 40 μl of β-form PrP90-230 (1 mg/ml) for 30 min at 25 °C. The ternary complex (PrP-aptamer-streptavidin) was then partitioned from the binary complex (aptamer-streptavidin) using StrataClean™ resin (to which streptavidin does not bind). RNA was recovered from both fractions, partially hydrolyzed using alkali, and analyzed by urea-denaturing PAGE, as described above, along with a transcript without biotin incorporated as a control. Slot Blots Using Chemically Synthesized Biotinylated Aptamer—30 pmol of full-length PrP was bound to a polyvinylidene difluoride membrane (equilibrated in 0.8% (v/v) acetic acid, 350 mm β alanine) by applying a vacuum. The membrane with protein attached was then washed with Tris-buffered saline-Tween and 10 mm magnesium chloride and then prehybridized in the same buffer with the addition of 5% milk powder and 20 μg/ml tRNA for 1 h. The membrane was then incubated overnight in 20 pmol of SAF-93(1-34,35bU,36-60), (chemically synthesized and biotin derivitized at position 35) that had been refolded in 1× HMNK and added to 20 ml of Tris-buffered saline-Tween, 10 mm magnesium chloride, 5% milk powder, 20 μg/ml tRNA. The membrane was blocked again for 1 h, washed, incubated in presence of streptavidin alkaline phosphatase (0.25 μg/ml) (Sigma) for 30 min, washed, and then developed using enzyme chemifluorescence substrate as described before. All of the incubations were carried out at 25 °C. Primary and Secondary Structure Determination of SAF-76 and SAF93—To elucidate their secondary structures, SAF-76 and SAF-93 were transcribed in vitro, purified, labeled at the 5′ end with 32P, and refolded. They were then subjected to limiting digestion using the structure-sensitive nucleases S1 (preference for single-stranded regions) and V1 (preference for double-stranded regions) (27Ehresmann C. Baudin F. Mougel M. Romby P. Ebel J.P. Ehres-mann B. Nucleic Acids Res. 1987; 15: 9109-9128Crossref PubMed Scopus (661) Google Scholar) and analyzed using denaturing PAGE, as described (see "Experimental Procedures"). The results (see supplemental figures) were used to constrain the MFOLD helix prediction algorithm (28Zuker M. Nucleic Acids Res. 2003; 31: 3406-3415Crossref PubMed Scopus (10406) Google Scholar) and to eliminate the suboptimal structures that were inconsistent with experimentally determined structural features. The most empirically compliant structures are illustrated in Fig. 1 (A and B). It must be noted that RNA structures are often dynamic (29Nguyen D.H. DeFina S.C. Fink W.H. Dieckmann T. J. Am. Chem. Soc. 2002; 124: 15081-15084Crossref PubMed Scopus (41) Google Scholar). Aptamers, which have been selected for their ability to bind to a target protein with good affinity, may not always be fully structured in the absence of protein. For this reason, even though multiple repetitions of these experiments gave entirely consistent results, we cannot be categorical that these putative secondary structures are as well determined as, say, those of tRNAs. With these caveats, the aptamers appear to fold into two and three helical domains, respectively, separated by extended, nonhelical regions. That is, SAF-76 appears to comprise two domains consisting of helices 1, 2, and 3 and of helices 4, 5, and 6, respectively, and SAF-93 appears to comprise three domains consisting of helix 1, of helices 2, 3, and 4, and of helix 5, respectively. It seemed very possible that only one of these domains in each case was actually necessary for interaction with PrP, particularly because the aptamers are relatively long (117 and 116 nucleotides, respectively). Accordingly, we determined the 3′ boundary of the PrP-binding regions of each by separating a partial hydrolysate of 5′-labeled full-length aptamer into PrP-binding and non-PrP-binding fractions and analyzing each by urea-denaturing PAGE (Fig. 1, C and D). In each case, an approximate 3′ boundary could be identified as the point at which the intensity of fragments in the lane of nonbinding RNA became greater than that in the lane of PrP-binding RNA, as indicated in Fig. 1. Again, it should be noted that even very short fragments retain some binding to PrP, probably through the nonspecific interaction with the N-terminal tail of the protein. This results in a slight blurring of the boundary between the essential and inessential regions of aptamer. Nevertheless, in both cases, it seemed possible to eliminate a substantial 3′ portion of the aptamer without loss of PrP binding. When the 3′ boundaries were placed on the secondary structure previously determined, a paradox immediately presented itself in both cases. In SAF-76 helix 1 was split by the 3′ boundary, and in SAF-93 helix 2 was split by the 3′ boundary. That is, in each case, only one-half of a predicted helix remained in the 3′-truncated, PrP-binding aptamer. Three explanations seemed possible. First, the actual secondary structures might be very different from those presented. This could result from a large number of noncanonical, tertiary interactions stabilizing much weaker canonical helices than had been considered by the MFOLD algorithm. Second, the truncated aptamer might still bind to PrP but with much reduced affinity, as a result of the loss of one helix. Third, the truncated aptamer might fold into an altogether different structure from its full-length parent. To resolve these possibilities, we used the primary and secondary structure information to generate defined, truncated aptamers and analyzed both their affinity for PrP and their secondary structures. Structure and Affinity of Truncated Aptamers—The double-stranded DNA template used to transcribe the parental aptamers in vitro was amplified using the standard 5′ SELEX primer together with a 3′ primer corresponding to the 3′ boundary determined above. The parental aptamers and their truncated derivatives are hereinafter named to indicate the nucleotides retained from the parent. These truncated derivatives were then analyzed by enzymatic probing as described previously. Again the experiments were repeated a number of times to ensure that the cleavages shown were consistent. SAF-76(1-38) showed anomalous nuclease sensitivity when compared with its parent, SAF-76(1-117) (Fig. 2A). Nucleotides that had formerly appeared to be single-stranded were now sensitive to nuclease V1, whereas those previously showing signs of double-strandedness were now sensitive to nuclease S1, making the parent-like structure shown in Fig. 2A untenable. When we constrained the modeling of SAF-76(1-38) by MFOLD with the new data, the structure shown in Fig. 2B emerged. Intriguingly, this contains secondary structural features found within the 5′ region of SAF-93, with which it shows some sequence homology. When we compared the affinity of the truncated aptamer with its parent (Fig. 2C), it appeared that removal of the 3′ portion reduced the affinity of SAF-76 for α-form PrP but maintained its affinity for β-form PrP. This is consistent with the idea that binding to the α-form is dominated by nonspecific nucleic acid interactions in the unstructured N terminus of PrP, which would lessen as the length of the RNA decreased. Although the conformational specificity of SAF-76(1-38) to β-oligomeric form PrP was improved compared with the full-length aptamer, its affinity was still weaker than that of aptamer SAF-93. We constructed three deletion mutants of SAF-93. The first, SAF-93(1-60), was based on the 3′ boundary determined above. The second, SAF-93(3G,28-60), was, in addition, truncated at the 5′, in accordance with indications from a 5′ boundary experiment (data not shown), thereby removing helix 1. The third, SAF-93(1-33,43-45,48-51), was designed to preserve helix 1, shorten helix 3, and remove sequences corresponding to the "half-helix" 2. Nuclease sensitivity experiments, together with computer modeling, indicated that the secondary structure of each of these truncated aptamers was as anticipated (Fig. 2D). Interestingly, nucleotides 31 and 32 of SAF-93(1-60), which had appeared to be in helix 2 of the full-length SAF-93(1-116), were still sensitive to nuclease V1 in the truncated version, in which helix 2 could not form. This is strong evidence for stable, noncanonical interactions between this region and some other portion within the 5′ half of SAF-93. When we measured the affinity of the truncated versions of SAF-93 for both conformations of PrP, we found that removal of the 3′ half to generate SAF-93(1-60) had little effect on the binding to the α-form while slightly improving the affinity for the β-oligomeric form (Fig. 2E). The shortening of helix 3 and removal of region 2 in SAF-93(1-33,43-45,48-51), reduced the binding affinity for the β- oligomeric form of PrP. However, destruction of helix 1 in SAF-93(3G,28-60) and disruption of region 2, comprising presumptively tertiary structure, effectively destroyed the specificity of binding to β-form PrP. These results correlate with the footprinting studies on SAF-93(1-60), which showed that much of the sequence was protected by binding to β-form PrP (see supplemental figures). Taken together, the results show that SAF-93(1-60) defines a minimal, conformationally selective aptamer for PrP and comprises two helices bounding a region of tertiary structure. Generation of an Internally Derivitized Aptamer for PrP Detection—One consequence of the minimization procedure above was that relatively little of SAF-93(1-60) was free to interact with bulky molecules such as streptavidin once it was bound to β-form PrP, as indicated by the footprinting results (see supplemental figures). This was disadvantageous in many potential applications of aptamer technology. Indeed, introduction of a biotin at the 3′ end, the most convenient position for site-specific, post-transcriptional incorporation, resulted in a ligand that, when complexed with streptavidin, had very poor PrP binding characteristics (data not shown). Consequently, we undertook a systematic search for internal uracils that could be so derivitized without detriment. We exploited the fact that all of the uracils in SAF-93(1-60) were part of the nonhydrolyzable nucleotide, 2′-fluoro-2′-deoxyuridine. This means that, in the unmodified SAF-93(1-60), a partial alkaline hydrolysate of 5′-labeled aptamer gives a purine ladder, and bands corresponding to U are absent. On the other hand, biotinylated uracil was associated with 2′ OH 5-allyl-biotinyl uridine, which is readily cleaved by alkali, producing an additional band at any U position into which it is incorporated. We spiked transcription of SAF-93(1-60) with 2 μm 5-allyl-biotinyl UTP to produce an average incorporation rate of one biotinylated uracil/molecule, presumably at random, at the nine available positions. We then complexed the partially biotinylated RNA with streptavidin (SA) and purified the RNA that was bound to one SA tetramer from that which bound none, two, or more molecules of SA. We then incubated the SA-aptamer complex with PrP and partitioned the PrP-binding complexes using the resin StrataClean™, which fortuitously binds PrP but not SA. We subjected the PrP-binding and nonbinding complexes to partial alkaline hydrolysis and analyzed them on denaturing PAGE (Fig. 3, A and B). The results show that the most 5′-proximal uracil that can be derivitized with biotin-streptavidin while retaining its PrP binding properties is U35. In the 5′ region of U35, the hydrolysis ladder is identical in both the biotin-SA-PrP-binding sample and the control sample. However, in the 3′ end of U35, there is an additional cleavage event in the biotin-SA-PrP-binding sample, the bands migrate more slowly (by approximately one nucleotide equivalent) and are less distinct than in the control sample. This is consistent with the incorporation of an additional 487 Da (corresponding to allyl-biotin) and the replacement of 2′-fluoro with 2′OH at position 35. Because of the resolution of the gel, it is not possible to exclude the possibility that uracils 3′ to U35 could also be derivitized in this way without loss of function. To test whether our interpretation of this experiment was correct, we chemically synthesized an aptamer corresponding to the sequence identified above, namely SAF-93(1-34,35bioU,36-60) and determined its affinity for the two isoforms of PrP when complexed with streptavidin. The results show that, as anticipated, the affinity of the aptamer for β-form PrP was the same in the absence as in the presence of streptavidin: Kd = 8.4 ± 1.4 and 6.8 ± 2.9 nm, respectively (Fig. 4A). This aptamer was then used to probe slot blots of PrP in either the β-oligomeric or α-monomeric conformation (Fig. 4B). The aptamer detected the β-form much more effectively than the α-form and did not bind to an equivalent amount of human serum albumin. We have recently described aptamers that bind a region of PrP that is intimately involved in the conformational shifts that occur in the generation of prions and can inhibit conversion in cell-free assays systems (23Rhie A. Kirby L. Sayer N. Wellesley R. Disterer P. Sylvester I. Gill A. Hope J. James W. Tahiri-Alaoui A. J. Biol. Chem. 2003; 278: 39697-39705Abstract Full Text Full Text PDF PubMed Scopus (180) Google Scholar). Although the original aptamers had many promising features, for example, their resistance to degradation conferred by 2′-fluoro substitution, they were not ideally suited to analytical or therapeutic applications because of their length and complexity. Accordingly, we have used a range of experimental and computational tools to define the essential functional structures of two prion-binding aptamers. By analyzing the functional primary and secondary structure of the original 116-nucleotide-long aptamer using a range of techniques, we have been able to preserve its conformational specificity while reducing it to a size suitable for chemical synthesis. Moreover, using a new method for screening internal modifications, we have been able to derivitize it with biotin to generate a ligand that can readily be applied in multiple detection formats. The synthetic aptamer described here (SAF-93(1-34,35bioU,36-60)) is a novel and promising tool for the investigation of PrP isoforms in vivo. For example, it could be used as a detection reagent in slot or dot blotting applications, as shown here. Specifically, one could envisage applying multiple samples of blood, cerebro-spinal fluid, brain homogenate, or other biological materials in parallel to a polyvinylidene difluoride or similar support membrane. SAF-93(1-34,35bioU,36-60) could then be used to develop the membrane, followed by detection using enzyme-conjugated streptavidin and chemoluminescent, chemofluorescent, or chromogenic imaging. Perhaps more conveniently, high throughput plate format assays could be developed in which aptamer-derivitized assay plates are used to capture abnormal PrP from biological samples. Detection of the captured PrP could again by through the use of SAF-93(1-34,35bioU,36-60) and any of the proprietary streptavidin-based systems. More biologically informative uses include the replacement of anti-PrP antibodies with SAF-93(1-34,35bioU,36-60) in histochemical analysis of biopsy and post mortem samples of experimentally infected model animals or suspect cases. Certain investigations that are not possible using currently available antibodies might become possible using the aptamer. For example, the specificity of SAF-93(1-34,35bioU,36-60) for abnormally folded PrP does not depend on the resistance of the protein to proteinase K. This means that putative forms of PrP that are abnormally folded and either infectious or pathogenic but not fully resistant to PK could be sought directly. Moreover, because the selective reactivity of abnormally folded PrP for aptamer is expressed under native conditions and does not require the presence of denaturants, as is commonly the case with antibody-based systems, the aptamer approach opens the way for a simpler exploration of the appearance of infectious material in vivo under conditions in which classic PrPSc is not presently detectable. We thank James Hope for advice and constructive discussion at early stages of this work and Andrew Gill for help with protein characterization. Download .pdf (.08 MB) Help with pdf files
Referência(s)