Proc. Natl. Acad. Sci. USA Vol. 93, pp. 7475-7480, July 1996 Biochemistry

Anti- aptamers recognize amino acid sequence and bind a (in vitro selection/SELEX/combining site) WEI XU AND ANDREW D. ELLINGTON* Department of Chemistry, Indiana University, Bloomington, IN 47405 Communicated by Norman R Pace, Indiana University, Bloomington, IN, April 3, 1996 (received for review December 1, 1995)

ABSTRACT In vitro selection of binding been shown to interact with the anion exosite of the molecule species (aptamers) is superficially similar to the immune (5), whereas selected to bind to the Tax protein of response. Both processes produce biopolymers that can rec- human T-lymphotropic virus type I can recognize disparate ognize targets with high affinity and specificity. While anti- faces of the protein and differentially disrupt protein-protein bodies are known to recognize the sequence and conformation interactions with cellular transcription factors (6). of protein surface features (), very little is known We have employed the human virus type about the precise interactions between aptamers and their 1 (HIV-1) Rev protein as a model system to further explore epitopes. Therefore, aptamers that could recognize a partic- how peptide and protein epitopes are recognized by aptamers. ular epitope, a peptide fragment of human immunodeficiency Rev binds to and facilitates the transport of mRNAs contain- virus type 1 Rev, were selected from a random sequence RNA ing a 234 residue Rev responsive element (RRE; ref. 7). pool. Several of the selected RNAs could bind the free peptide Interactions between the Rev protein and its primary binding more tightly than a natural RNA , the Rev-binding site, the 30 residue Rev-binding element (RBE), have previ- element. In accord with the hypothesis that protein and ously been shown to be mediated by an arginine-rich motif nucleic acid binding cusps are functionally similar, interac- (ARM) that spans residues 34-50 (8, 9). In the current study, tions between aptamers and the peptide target could be RNA aptamers were selected to bind to the isolated Rev34_50 disrupted by sequence substitutions. Moreover, the aptamers peptide. Like , the selected RNAs could specifically appeared to be able to bind with different solution recognize the sequence of the peptide. Like antibodies, the conformations, implying an induced fit mechanism for bind- anti-peptide aptamers could bind the corresponding native ing. Just as anti-peptide antibodies can sometimes recognize epitope on the Rev protein, albeit with lower affinity. These the corresponding epitope when presented in a protein, the results suggest that artificially selected binding cusps may have anti-peptide aptamers were found to specifically bind to Rev. much in common with those found on natural molecules and have important implications for the development of novel In vitro selection techniques have been used to isolate nucleic diagnostic reagents and strategies for intracellular immuniza- acid molecules that can bind tightly and specifically to a wide tion. range of target molecules (1). The sequence and functional diversity generated by in vitro genetic experiments is reminis- cent of the molecular diversity generated by an immune MATERIALS AND METHODS response. When the binding abilities of naturally selected Peptides, , and RNAs. Peptides were a gift of protein antibodies and artificially selected nucleic acid aptam- Amgen Biologicals. Peptides were made by standard auto- ers are compared, they are found to be strikingly similar. Like mated procedures. The purity and concentration of each antibodies, aptamers can recognize molecules as small as peptide was confirmed by HPLC, mass spectrometry, and theophylline (Mr = 120) and as large as 30S ribosomal particles amino acid analysis. The sequence of the 16-mer peptide (Mr > 106). Like Fab fragments, aptamers can form corresponding to the ARM of Rex (positions 1-16) was monovalent complexes with their targets that have dissociation MPKTRRRPRRSQRKRP. The sequence of the 15-mer pep- constants in the nanomolar range. The specificity of aptamers tide corresponding to the ARM of Tat (positions 46-60) was for their targets can be similar to that of antibodies for their SYGRKKRRQRRRPPQ; this peptide was kindly provided by : RNAs selected to bind to anti-peptide antibodies can Maria Zapp (University of Massachusetts Medical Center). compete with the cognate peptide ligand (2). RNAs selected Rev and a glutathione S-transferase (GST)-Rex fusion protein to bind to one isozyme of protein kinase C can discriminate were overexpressed in and were generous gifts against isozymes that are as much as 96% homologous (3), of Maria Zapp. Recombinant Tat protein was overexpressed in while RNAs selected to bind to avian myeloblastosis virus E. coli and was obtained from Agmed (Bedford, MA). reverse transcriptase can discriminate against the Moloney A minimal RBE that binds Rev as well as the 94-nt stem IID murine leukemia virus reverse transcriptase and vice of the RRE was used for competition experiments. The versa (4). sequence of the minimal element was 5' GGGAACUCGAU- The hypothesis that the two types of biopolymers are GAAGCGAGCUCUUGGGCGCAGCCUCAAU- functionally similar can be further assessed by determining to GAGGCUGACGGUACAAGUACUGACUUCG- what extent the properties of antibodies predict the properties GAUCCCUGC; underlined positions correspond to residues of aptamers. Such predictions have already been borne out in 45-54 and 64-75 of the HIV-1 RRE [numbering according to several instances. For example, antibodies have long been Malim et al. (10)]. The minimal RBE was transcribed from a known to recognize particular surface features (epitopes) on double-stranded DNA oligomer by using an Ampliscribe kit antigens. Aptamers now also appear to recognize epitopes on target molecules. selected to bind to have Abbreviations: RRE, Rev responsive element; RBE, Rev binding element; ARM, arginine-rich motif; sRev, selection Rev; uRev, un- The publication costs of this article were defrayed in part by page charge modified Rev. payment. This article must therefore be hereby marked "advertisement" in *To whom reprint requests should be addressed. e-mail: accordance with 18 U.S.C. §1734 solely to indicate this fact. [email protected]. 7475 Downloaded by guest on October 2, 2021 7476 Biochemistry: Xu and Ellington Proc. Natl. Acad. Sci. USA 93 (1996) (Epicentre Technologies, Madison, WI) and was purified on a Competition Assays with the RBE. Aptamers were assayed denaturing polyacrylamide gel as described (11). for their ability to compete with the RBE for binding to Rev. Derivatization of Resin. A peptide corresponding to resi- Either radiolabeled N71 pool or selected clones (2.4 ,g; 0.6 dues 34-50 of the Rev protein of HIV-1 (sRev; see Fig. 4) was ,LM) were heated to 75°C for 3 min in 1 x binding buffer (50 used for selections. A milligram of peptide was coupled to 0.5 p,l) and allowed to cool to ambient temperature over S min. ml bed volume of Affi-gel 10 (Bio-Rad) according to the Radiolabeled RBE (1.8 ,ug; 0.6 ,uM; 50 ,l) was thermally manufacturer's instructions. The amount of peptide present in equilibrated in a separate tube. The two RNA samples were solution before and after the derivatization was determined by mixed and Rev protein was added to a final concentration of a Coomassie blue assay (12). The concentration of the peptide 0.3 ,M. The mixture was incubated at ambient temperature for on the resin was estimated to be 0.5 mg/ml (0.2 mM). A 60 min. A small portion (5 ,ul) of the reaction mixture was set "negative selection" resin was generated by derivatizing Affi- aside, and the remainder of the binding reaction was filtered gel 10 with ethanolamine. over HAWP modified cellulose filters (Millipore). The filters Pool Design and Construction. A RNA pool containing a were washed twice with 500 ,ll of lx binding buffer. Bound 71-nt completely random sequence tract was used as a starting RNAs were eluted by boiling three times in 100 ,ul water for point for selections (13). To avoid amplification artefacts, the 3 min, and the eluates were pooled (300 ,lI) and precipitated. primer sequences flanking the randomized region corre- Unfiltered and filtered samples were analyzed on a 10% sponded to those used by Crameri and Stemmer (14), with the denaturing polyacrylamide gel. The number of counts in exception that one of the primers (42.71), was appended to individual bands were determined by using a Phosphorlmager include T7 RNA polymerase promoter sequences. The se- (Molecular Dynamics). quence of 42.71 was 5' GGTAATACGACTCACTATAGG- GAGATACCAGCTTATTCAAT. A double-stranded DNA RESULTS pool was generated from the original synthetic oligomer via a large-scale PCR (13), and an RNA pool was transcribed from In Vitro Selection of Aptamers That Bind a Peptide from a portion (1 ,g; 1 x 1013 sequences) of this pool. HIV-1 Rev. The RNA recognition domain of HIV-1 Rev is an In Vitro Selection of Aptamers That Could Bind an Immo- ARM that spans residues 34-50. A 17-mer peptide (sRev) bilized Peptide. In round 1, RNA (5 gg, 5 library equivalents) corresponding to this domain was used as a target for in vitro in 1 x binding buffer (500 ,ul of 10 mM Hepes, pH 7.4/100 mM selection experiments. A RNA pool containing 71 random NaCl) was mixed with negative selection resin (50 ,ul) and sequence positions was incubated with an affinity resin con- 10 The binding reaction was poured taining sRev. Nonbinding species were removed by washing, incubated at 7°C for min. and binding species were eluted with high salt. The stringency into a small plastic tube, briefly centrifuged, and the super- of the selection was progressively increased by washing the natant removed. The supernatant was mixed with affinity resin resin with higher concentrations of monovalent cations before (50 ptl) and incubated at 7°C for 20 min. The supernatant was elution. removed as before, and the affinity resin was washed twice After 10 cycles of selection and amplification, the amount of with 500 ,ul of 1 x binding buffer and once with 500 ,ul of 1 x RNA that could be retained on the resin had increased (6.2% HS buffer (10 mM Hepes, pH 7.4/1 M NaCl). Finally, selected bound) compared with the starting population (0.2% bound), RNAs were eluted by washing the affinity resin twice with 100 and the selected RNAs could bind to the resin better than the ,ul of lx elution buffer (10 mM Hepes, pH 7.4/1 M NaCl/ wild-type RBE (1.2% bound). The selected pool was cloned 0.1% Triton X-100). The eluted species were generally a small and sequenced. Multiple, different sequence families were fraction of the total population, varying from 0.3% to 5% of present in the selected population (Fig. 1). In general, variants the total applied species in the early cycles. Eluates were within sequence families appeared to be related by mutation, pooled and counted. Selected RNAs were amplified as de- but different sequence families showed no relationship to one scribed (11, 13) and used for additional cycles of selection. another, the RBE, or aptamers previously selected to bind Subsequent rounds were similar to the first, except in rounds Rev. The C8 and C18 aptamer families were independently 6-8 1 x binding buffer contained 250 mM NaCl and in rounds derived, yet showed extensive similarity to one another (Fig. 1). 9 and 10 1 x binding buffer contained 400 mM NaCl. In rounds Although the anti-peptide aptamers did not share continu- 6-10 the wash with lx HS buffer was omitted. ous primary sequence motifs, several of the aptamers are Mobility Shift Assays. The ability of RNAs to bind to free predicted to contain a structural motif similar to the arginine- peptide was assayed by gel mobility shift assays. Radiolabeled binding pocket of HIV-1 TAR (Fig. 2a; ref. 15). Aptamers C17 RNA samples were thermally equilibrated in 50 mM Hepes and C8 are predicted to fold to form a UA bulge loop 5' to the (pH 7.4) and 100 mM NaCl by heating to 75°C for 3 min and paired stem GAG:CUC (Fig. 2b). Although this bulge loop then cooling to ambient temperature over 5 min. RNAs (0.5 contains only two residues, as opposed to three in TAR, ,ug; 1.2 ,uM) were mixed with peptides (25 ng; 1.0 ,uM) in 10 mutational analyses have revealed that it should still specifi- ,ul total volume of 1 x shift buffer (10 mM Hepes, pH 7.4/100 cally interact with arginine (18). To determine if the TAR-like mM KCl/1 mM dithiothreitol/1 mM MgCl2/1 mM EDTA/50 secondary structural motif played a role in binding the sRev ,ug of tRNA per ml/5% glycerol). The binding reaction was peptide, the 5' and 3' boundaries of the C17 aptamer were incubated on ice for 30 min. RNA-peptide complexes were mapped by deletion analysis. The minimal C17 aptamer is separated from free RNA by electrophoresis on a 10% poly- shown in Fig. 2b and still contains the TAR-like motif. acrylamide gel (29:1, acrylamide/bisacrylamide; lx TBE) at 8 Anti-Peptide Aptamers Can Interact Tightly and Specifi- W for 3 hr. The temperature of the gel remained at 4-7°C cally with Free Peptide. While the selection favored species throughout the run. Dissociation constants for RNA-peptide that recognized the immobilized ligand, aptamer families that complexes were determined by quantitating mobility shifts as could bind to free sRev peptide were identified by using a a function of peptide concentration. RNA concentration was mobility shift assay. Two ofthe major families, C8 and C17, and 0.5 nM, and the peptide concentration varied from 5 to 400 several of the less populated families (e.g., C24, C33) could be nM. readily shifted on a native gel by free peptide, but the other Mobility shifts with proteins were the same as those carried major family (C2) could not (Fig. 1). Interestingly, sequence out with peptides, except that the final concentration of RNA variants within a family showed different propensities for was 50 nM, the final concentration of protein was 1.5 ,uM, and binding to free peptide. the binding reaction was incubated at ambient temperature for The most prevalent aptamers from four of the families (C8, 30 min and then assayed at 4°C. C17, C24, and C33) that interacted with free peptide were Downloaded by guest on October 2, 2021 Biochemistry: Xu and Ellington Proc. Natl. Acad. Sci. USA 93 (1996) 7477

Shift GAAAUCAAUA (17) - 2:. ly CAGUCCUCGGGUGCUAACCAUGkCACCCMAGGCCUUAGUGGUAUGUGGUUCUUG am ------A------A------_SA------______------C------G--G------C------C------G------G------C-C------A--- C8: G-- GCGAAUACGUCAUAUUGGUAGUAUG---UAGAG.CGUGGUGCAUCUCCAAACUGCUG1) C18: GCU _G&ACGGUUGG AAUCGU CUUGCCAUGAUUUACGUAUCC GUGAUG (2) - family C17: GUAUUCUGGUGGUUUA A GG GUCCUUUGGUUGGACUACAGUGGAGGUUCUCUUA (3) + family ------4------G:-G------U--G 4------U------U------U--G :-:------^-----A------> G------G------z-----U A ----- C-GU---- AAUGGAAAUGCGCAGCGACCAUGAAACCUCGCAUGGUUCAUCGAUUGUUUGGAUAGUGUCUGUGUG (2) - R:.ly~ami1 ______------&------~ ~ ------_------______A Cl: GCAGUUAACCAAGCCUGCAUACUGGAUAGACGGCUUAUCCGACUGAAUGCCUCCCGAAAGGUGCAGUU C9: GCGCAAACCCGAAGAAUGCCCAAAUUGAUCCAGAGCAAGUGGGAAUGAUAUAAAGUACCUGGUCCUGG C15: UCCAAACCCCGUUGAGAGUUGAUCCGGUCUAGGGAAUGGGAAAGAAGUAGGUAUCGAAGAGAAUGUACCCU + C24: AGGACUGAAAUAUUCACGUUGACGUUGUCUUGGAGUGCUGAUGGAAACCAAUAUGAWUAAUGGGUCCUG C33: ACGCAGCGACUGUGGUGGUGAGCGGUU GCGUAACUUGAUUUAAGCAAGUACUCAUGGCCGAACCUCUA C34: UCUAGUCAAGUUGCAAUCUCCGGUGGGGUGGUAACCGAGGAACACGUUUCGGGUGUAUAGGCUAGCG C37: UCUACCAGAGCGAGUGUGCUGAACGUUCUAAGGACGGGAUUGAAUCGAGAUGCGUAUACUAGGACCUUACG + C52: GCUUGGUACCGAGCUCGGAUCCACGUAGUAACGGGCCGCCAGUGUGCUGGAAUUCGGGUCGUUCUUG FIG. 1. Families of anti-peptide aptamers. RNA molecules from round 10 were cloned and sequenced. There were several families that contained highly related variants. The number of times each variant was isolated is indicated in parentheses. The C8 and C18 families contain related motifs (underlined). Different aptamers were assayed for their ability to be shifted by free peptide (RNA/peptide = 1:1). Results are recorded on the far right of the figure; -, no observable interaction; +, an observed interaction (shifted band >10% of input RNA). examined in greater detail. While both the anti-peptide aptam- and perhaps the sequence of the peptide, as opposed to merely ers and the RBE could shift the sRev peptide, the dissociation binding to any positively charged ligand. constants for the anti-peptide aptamers ranged from 19 to 36 Anti-Peptide Aptamers Recognize Peptide Sequence. A nM, while the dissociation constant for the RBE was 44 nM. panel of peptides in which the core Rev sequence was replaced The fact that selected RNAs showed slightly higher affinity for with runs of alanines was used to assess how well the RBE and the free peptide than did the RBE was consistent with the anti-peptide aptamers could recognize the amino acid se- elution data for the selected pool. The specificities of the quence (Fig. 4). Because the anti-peptide aptamers appeared selected RNAs were probed by determining whether they to recognize an unmodified Rev peptide (uRev) at least as well could bind to other arginine-rich sequences. While a peptide as the original sRev target, unmodified variants were used in corresponding to the Rev ARM (sRev) could efficiently shift this comparison. anti-peptide aptamers, peptides corresponding to ARMs from The anti-peptide aptamers could tolerate some amino acid HTLV-I Rex and HIV-1 Tat could not (Fig. 3). This result substitutions, but were extremely sensitive to others. Different indicated that the aptamers were recognizing the overall shape aptamers demonstrated differential sensitivities to amino acid substitutions, and presumably recognized the original peptide a. TAR target in different ways. When the four amino acids at the amino terminus of uRev were substituted by alanines J all of the anti-peptide aptamers tested retained 51 c A G c (uA4Rev), GGGU UCUCUG GUUAGCCAGA c G I I 11 II 1*1 1 1 1 1 1 1 1 1 1 1 1 1lIl GI binding activity. However, when substitutions were extended 3,CCCAAGGGAU CAAUCGGUCU CU GAG by an additional three alanines (uA7Rev), the C17 aptamer could no longer bind. All of the aptamers were found to be b. Aptamers particularly sensitive to substitutions at R41 and/or R44 [compare results with uRevA4(41-44) and uRevA2(42-43)], 017 \ GA%dCMnimaI C17 CAAUUGUA UUAA AGU *Au 10 10 0 G 06 UU 0 0D a C UC- U AGUJ GGUC a.i > x & 0 A ,I bUAU \)A GA% @ @ _ G 0X > X U*~GAG uuAibAu G o 0 0 0^, 0. X C I CIt UZi-cc cc( CC 0 d:~~~~r01cc u 4uc UGCU AAAGAG .0 ,. k "

A GA A A 08CGA uAAG UGA G CG CU W CGUGA AUGI.I. GAGUAGUAUG b4614 &i 4. .,A C17 C24 C8 C33 AGA FIG. 3. Specificity of anti-peptide aptamers. The most prevalent members of the C8, C17, C24, and C33 families were challenged with FIG. 2. Structures of arginine-binding RNAs. Each of these RNAs peptides corresponding to the ARMs of several different proteins. The contains a particular bulge loop structure adjacent to a stem: 5'-UN1_2- sequences of the Rev, Rex, and Tat peptides are described in detail in GAG ... CUC-3'. These bulge loop structures are boxed. (a) The Materials and Methods. These peptides have similar numbers of secondary structure of the TAR element of HIV-1 (16). (b) The arginine residues and similar overall charges. Anti-peptide aptamers secondary structures of the C17 and C8 aptamers were predicted by were readily shifted by the cognate peptide from Rev (RNA/peptide = using the program MULFOLD (17). The minimal C17 aptamer was 1:1), while no complex formation was observed with peptides derived identified by deletion analysis. from Rex and Tat. Downloaded by guest on October 2, 2021 7478 Biochemistry: Xu and Ellington Proc. Natl. Acad. Sci. USA 93 (1996)

pept ide FBE C8 C17 C24 C33 T RQARRNRRRRWRERQR sPev Succ- ++ (1.0, 1.0) ++ ++ ++ ++ uRev_ - (0I,) ++ ++ ++ ++ uA4 Pev AAAA - 0.14 0.12) + + ++ ++ uA7Pev AAAAAAA - ~ ++ - ++ ++ uRevA4(41-44) AAAA u RevA2 (42-43) A t+(0.14,0.12) ++ + ++ + URevA6 AAAAA URevA3 AAA ++ ++ + ++ sRevn Succ- -OONH2 ++ ++ ++ ++ ++ sRevA(39)n Succ-=A -ONH2 - (0,0) ++ ++ ++ ++ sRevA(44)n Succ- -CONH2 - (0, 0) ++ + ++ sRevA6n Succ- AAAAAA-CONH2 - (0 , 0) 99evA3n Succ- A A-CONH2 ++(0.94, 0.90) + + ++ uRevN(44) _~~N ++ ++ - ++ uRevK(44) _K~~~~~- ++ ++ - ~++ FIG. 4. Anti-peptide aptamers recognize the amino acid sequence. Interactions with anti-peptide aptamers were probed with a panel of peptide variants by using a gel mobility shift assay. The sequence of the Rev peptide is indicated at the top of the figure; identities with the wild type are indicated by filled boxes, while substitutions are explicitly written. "Succ-" is an amino-terminal succinylation, while "CONH2" is a carboxy-terminal amidation. The names of peptides are on the left. Binding efficiencies are reported relative to interactions with the original sRev peptide target, and are broken down into three classes: -, 0-10% of the sRev control; +, 11-80% of sRev; + +, >80% of sRev. The numbers in parentheses are binding efficiencies relative to the sRev peptide for two separate determinations with the RBE. and to substitutions between positions 45 and 47 (compare with proteins from which arginine-rich peptides were derived: uRevA3 and uRevA6). In particular, the C24 aptamer was Rev, Rex, and Tat. In accord with earlier results with peptides, found to require a single arginine residue at position 44 for the anti-peptide aptamers were shifted by Rev, but not by Rex binding. or Tat (Fig. 5). Because the RBE recognizes the a-helical shape of the Rev The fact that anti-peptide aptamers could bind to peptides ARM as well as its sequence (9), it was possible that the with different a-helical propensities while the RBE could not patterns of binding observed with anti-peptide aptamers were suggested that the aptamers recognized and induced the the result of differential recognition of a peptide conformer. formation of a particular conformer, likely non-a helical. If so, To evaluate this possibility, the binding efficiencies of the RBE the anti-peptide aptamers might be expected to pay an ener- and anti-peptide aptamers were assessed with a panel of getic price for disrupting the structure of the Rev a helix. To peptide variants that contained end caps known to promote a test this hypothesis, aptamers were allowed to compete with helicity. As expected, while the wild-type RBE could not the RBE for binding to limiting amounts of Rev. While the recognize the peptide uRev, the anti-peptide aptamers could unselected pool was incapable of competing with the RBE for readily recognize both unmodified and end-capped peptides. binding to Rev, the anti-peptide aptamers could compete. Nonetheless, the same patterns of amino acid sequence sub- These results confirmed that the same epitope was recognized stitutions interfered with binding: substitution of the six car- on both the peptide and the protein. In accord with predic- boxy-terminal amino acid residues with alanines (sRevA6n) tions, while the anti-peptide aptamers bound to an unstruc- blocked binding to all variants, while substitution of only three tured peptide (uRev) better than the RBE, they bound the carboxy-terminal amino acid residues (sRevA3n) restored a-helical Rev ARM from 3- to 9-fold less well than the RBE. binding to all but the C8 aptamer. Anti-Peptide Aptamers also Bind the Rev Protein. Because anti-peptide aptamers appeared to recognize sequence more DISCUSSION strongly than shape, we attempted to determine if the Rev Antibodies interact with discrete regions of protein surfaces ARM could be recognized within its natural context: the Rev known as epitopes. Just as particular antibodies that bind to protein of HIV-1. Mobility shift assays were again carried out antigens are selected during immunization, particular RNAs that bind to protein targets can be selected in vitro. It was therefore germane to inquire whether the functional charac- c ~~~cc e e~xc teristics that have been observed for antibody recognition of protein epitopes also applied to the aptamer recognition of 0- - ...... protein targets. II;*sS Ii.. .: i $+, .-.r To generate aptamers that recognized a continuous epitope, a peptide corresponding to amino acids 34-50 of the Rev protein of HIV-1 was immobilized and used as a target for selection. This peptide must assume an a-helical conformation to interact with the RBE (9), but it was unknown what C17 C24 C8 C33 conformation it assumed on the column. By succinylating the peptide at the amino terminus and leaving the carboxy termi- FIG. 5. Anti-peptide aptamers can interact with the Rev protein. nus free, it was thought that the immobilized peptide would be Each of the aptamer families was assayed for its ability to shift various ARM proteins. The protein/RNA ratios in these experiments were able to assume a variety of conformations, and thus that 30:1, as opposed to 1:1 for assays with peptides. As was the case with aptamers that differed in sequence and structure from the ARM peptides (Fig. 3), each anti-peptide aptamer could shift the RBE and from previously selected anti-Rev aptamers would be cognate Rev protein, but not the noncognate Rex or Tat proteins. identified. Downloaded by guest on October 2, 2021 Biochemistry: Xu and Ellington Proc. Natl. Acad. Sci. USA 93 (1996) 7479 Aptamers selected from a random sequence population Nieuwlandt et al. (19) have speculated that aptamers to could indeed bind tightly and specifically to the free peptide substance P bind to a particular peptide conformation, since target. Our results compare favorably to a selection carried out the reverse orientation of the peptide sequence does not against the 11 amino acid substance P peptide (19). Substance interact with the aptamer. However, these results can also be P contained a single arginine and formed complexes with explained by assuming that the anti-substance P aptamers do anti-peptide RNAs that had a Kd of 190 nM, while the Rev not interact with the reverse orientation because they are peptide (sRev) contained multiple arginines and formed com- simultaneously recognizing the side chains and backbone plexes with anti-peptide RNAs that had Kd values of 19-36 nM. moieties of the peptide. Nonetheless, it is likely that the The increase in affinity seen in these selections relative to aptamer-substance P complex is conformationally con- substance P is likely due to more extensive interactions be- strained. Because substance P assumes a random conforma- tween the positively charged peptide and the negatively tion in solution, the anti-substance P aptamers may be another charged RNA molecules. The fact that positively charged example of an induced fit mechanism for target recognition. targets can often successfully elicit high-affinity RNA aptam- The ability of antibodies to recognize epitopes divorced ers has previously been observed and commented on (20). from their original structural context (27, 28) implied that As anticipated, the aptamers showed little sequence or peptide could be used to generate anti-protein structural similarity to the wild-type RBE. However, several of antibodies (29-31). Anti-peptide aptamers can also bind to a the aptamers are predicted to contain a structural motif similar protein containing the sequence corresponding to the peptide. to the arginine-binding pocket of HIV-1 TAR (Fig. 2a; ref. 15). Although it is difficult to generalize from a single peptide that The importance of the TAR-like motif for binding to free has high affinity for a natural RNA, our results suggest that it peptide can be inferred from sequence comparisons. Three may also be possible to use protein subsegments as targets for minor variants of the C17 family contain a CUC to UUC generating aptamers. If so, it should greatly simplify the substitution (Fig. 1). These three variants cannot bind to free generation of anti-protein nucleic acid probes for use in sRev peptide, while the major C17 family member and one diagnostic assays. Populations of anti-peptide aptamers might other minor variant that retain the CUC motif can bind to free someday even prove useful for intracellular immunization peptide. Similarly, the arginine-binding motif is found in the against viruses (32). If the aptamer does recognize a non-a- C8 family but not in the similar C18 family; the C8 family can helical conformer, then it is possible that it acts as a "RNA bind to free sRev peptide while the C18 family cannot. chaperone" and denatures a portion of the Rev protein during Interestingly, the same TAR-like motif has also been found in its interaction. Anti-peptide antibodies that denature their other RNAs that bind to ARMs, including aptamers selected corresponding epitopes have been shown to be neutralizing to bind Rev (11) and the natural Rex-binding element (21). (33); anti-peptide RNAs may prove to be similarly efficacious. The repetition of the secondary structural motif in all of these RNAs suggests that it may play a role in arginine-binding, just This research was supported by National Institutes of Health Grant as it does in TAR. PHS R01 AI 36083, a Cottrell Scholar Award (A.D.E.) from Research The selected aptamers appear to share a number of func- Corporation, a Pew Scholar Award (A.D.E.), an Office of Naval tional characteristics with antibodies. For example, antibodies Research Young Investigator Award (A.D.E.), and a National Science can recognize the amino acid sequence of an epitope. Our Foundation Young Investigator Award (A.D.E.). that can also recognize the results demonstrate aptamers 1. Gold, L., Polisky, B., Uhlenbeck, 0. C. & Yarus, M. (1995)Annu. amino acid sequence of an epitope. Several alanine substitu- Rev. Biochem. 64, 763-797. tion variants of the Rev peptide sharply reduced aptamer 2. Tsai, D. E., Kenan, D. J. & Keene, D. J. (1992) Proc. Natl. Acad. binding. The C24 aptamer appeared to be able to specifically Sci. USA 89, 8864-8868. recognize R44 in the context of the Rev peptide. In general, 3. Conrad, R., Keranen, L. M., Ellington, A. D. & Newton, A. C. however, the aptamers seemed to be less sensitive to amino (1994) J. Biol. Chem. 269, 32051-32054. acid substitutions than the RBE. Alanine scan experiments 4. Chen, H. & Gold, L. (1994) Biochemistry 33, 8746-8756. have revealed that the RBE recognizes multiple amino acids 5. Paborsky, L. R., McCurdy, S. N., Griffin, L. C., Toole, J. J. & throughout the Rev34_50 peptide (9). It is unclear whether the Leung, L. L. K. (1993) J. Biol. Chem. 268, 20808-20811. aptamers recognize a more delimited region of 6. Tian, Y., Adya, N., Wagner, S., Giam, C.-Z., Green, M. R. & anti-peptide Ellington, A. D. (1995) RNA 1, 317-326. the peptide than the RBE, or if the slightly higher affinities of 7. Cullen, B. R. & Malim, M. H. (1991) Trends Biochem. Sci. 16, aptamers for the Rev peptide can compensate for the loss of 346-350. some amino acid contacts. 8. Kjems, J., Calnan, B. J., Frankel, A. D. & Sharp, P. A. (1992) Antibodies frequently interact with peptide or protein li- EMBO J. 11, 1119-1129. gands by induced fit mechanisms (22-24). Natural RNAs have 9. Tan, R., Chen, L., Buettner, J. A., Hudson, D. & Frankel, A. D. also been found to bind their targets using induced fit. In (1993) Cell 73, 1031-1040.- particular, the RBE and the Rev34_50 peptide have been shown 10. Malim, M. H., Hauber, J., Le, S. Y., Maizel, J. V. & Cullen, B. R. to undergo conformational changes when they interact with (1989) Nature (London) 338, 254-257. one another (16, 25). The RBE induces an a-helical confor- 11. Giver, L., Bartel, D., Zapp, M., Pawul, A., Green, M. & Ellington, A. D. (1993) Nucleic Acids Res. 21, 5509-5516. mation in the Rev peptide on binding and cannot productively 12. Scopes, R. K., ed. (1987) Protein Purification: Principles and interact with non-a-helical peptides (9, 25). While the Rev Practice (Springer, New York). ARM requires a helical peptide configuration, other ARMs, 13. Conrad, R. C., Baskerville, S. & Ellington, A. D. (1995) Mol. such as the Tat ARM, can bind to RNA independent of the Diversity 1, 69-78. peptide conformation (26). Similarly, the anti-peptide aptam- 14. Crameri, A. & Stemmer, W. P. (1993) NucleicAcids Res. 21,4410. ers can bind to peptides irrespective of their a-helical propen- 15. Puglisi, J. D., Ruoying, T., Calnan, B. J., Frankel, A. D. & sity. The anti-peptide aptamers may therefore recognize an Williamson, J. R. (1992) Science 257, 76-80. extended rather than an a-helical conformation of the peptide, 16. Battiste, J. L., Tan, R., Frankel, A. D. & Williamson, J. R. (1994) or a portion of the peptide that does not readily form a helix. Biochemistry 33, 2741-2747. 17. Jaeger, J. A., Turner, D. H. & Zuker, M. (1989) Methods Enzy- In support ofthis hypothesis, although the aptamers could bind mol. 183, 281-306. to a relatively unstructured peptide ligand (uRev) better than 18. Weeks, K. M. & Crothers, D. M. (1991) Cell 66, 577-588. the wild-type RBE, they could not bind as well as the RBE to 19. Nieuwlandt, D., Wecker, M. & Gold, L. (1995) Biochemistry 34, the same a-helical peptide sequence in the context of the Rev 5651-5659. protein. Taken together, these results imply that aptamers may 20. Lato, S. M., Boles, A. R. & Ellington, A. D. (1995) Chem. Biol. recognize their targets by induced fit. 2, 291-303. Downloaded by guest on October 2, 2021 7480 Biochemistry: Xu and Ellington Proc. Natl. Acad. Sci. USA 93 (1996)

21. Baskerville, S., Zapp, M. & Ellington, A. D. (1995) J. Virol. 69, 27. Geysen, H. M., Meloen, R. H. & Barteling, S. J. (1984) Proc. 7559-7569. Natl. Acad. Sci. USA 81, 3998-4002. 22. Stanfield, R. L., Fieser, T. M., Lerner, R. A. & Wilson, I. A. 28. Rodda, S. J., Geysen, H. M., Mason, T. J. & Schoofs, P. G. (1986) (1990) Science 248, 712-719. Mol. Immunol. 23, 603-610. 23. Rini, J. M., Schulze-Gahmen, U. & Wilson, I. A. (1992) Science 29. Lerner, R. A. (1984) Adv. Immunol. 36, 1-44. 255, 959-965. 30. Rowlands, D. J. (1992) FEMS Lett. 100, 479-482. 24. Wilson, I. A. & Stanfield, R. L. (1995) Curr. Opin. Struct. Biol. 3, 31. Van Regenmortel, M. H. V. '(1992) FEMS Lett. 100, 483-488. 113-118. 32. Baltimore, D. (1988) Nature (London) 335, 395-396. 25. Tan, R. & Frankel, A. D. (1994) Biochemistry 33, 14579-14585. 33. Wien, M. W., Filman, D. J., Stura, E. A., Guillot, S., Delpey- 26. Tan, R. & Frankel, A. D. (1995) Proc. Natl. Acad. Sci. USA 92, roux, F., Crainic, R. & Hogle, J. M. (1995) Nat. Struct. Biol. 2, 5282-5286. 232-243. Downloaded by guest on October 2, 2021