RNA aptamers to the adenosine moiety of S-adenosyl methionine: structural inferences from variations on a theme and the reproducibility of SELEX
RNA aptamers to the adenosine moiety of S-adenosyl methionine: structural inferences from variations on a theme and the reproducibility of SELEXDonald H. Burke1 and Larry Gold1,2,*
1Department of Molecular, Cellular and Development Biology, University of Colorado, Boulder, CO 80309-0347, USA and 2NeXstar Pharmaceuticals, Inc., 2860 Wilderness Place, Boulder, CO 80301, USA
Received November 12, 1996;Revised and Accepted March 12, 1997
ABSTRACT
We used in vitro selection (SELEX) to isolate RNA `aptamers' to S-adenosyl methionine (SAM). Individual aptamer sequences conform to the structural element noted previously for adenosine binding in selections for aptamers to ATP and NAD+. When we compare the patterns of sequence conservation among 65 adenosine-binding sequences to the published structure of the adenosine aptamer, we find that the most highly conserved nucleotides contact the bound adenosine directly, and that one conserved nucleotide outside the binding pocket is in position to stabilize nucleotides within the binding pocket. The aptamer's ability to bind diverse adenosine-containing cofactors is easily understood in terms of its mode of binding, which leaves the 5' position exposed to solvent. We propose that aptamers that bind their targets away from the reactive moiety may be particularly well suited for catalysis. Finally, we estimate that one sequence in 1011 may be able to form this structural motif, and that there may be many other adenosine-binding motifs that have escaped detection because of their lower representation in the starting random pools.
INTRODUCTION
Many groups have used in vitro selections (SELEX; 1 ,2 ) to isolate RNA molecules (aptamers) that bind small molecules, including nucleotide cofactors (3 -5 ). RNA aptamers originally selected to bind ATP were used as the starting material to evolve self-kinasing RNA (6 ). Similar two-stage strategies may work for selecting a number of cofactor-dependent RNA enzymes (ribozymes).
S-Adenosyl methionine (SAM) is the primary methyl donor for methyl transfer reactions, and it is involved in generating radicals for ribonucleotide reduction in at least one class of ribonucleotide reductase enzymes (7 ). The adenosyl moiety of SAM and other nucleotide cofactors is often considered an evolutionary leftover from the RNA World (8 ,9 ), and by extension, as evidence that these cofactors were among the first used by primitive organisms. Given the importance of both methyltransfer and radical reactions, SAM is a likely candidate to have participated in RNA World metabolism. As a first step toward isolating methyltransferase ribozymes, we used SELEX to isolate SAM-binding RNA `aptamers.' The results of these efforts are described here.
MATERIALS AND METHODS
The starting RNA population (10 ) contained ~6 * 1014 (1 nmol) sequences of the form 5'-gggcauaagguauuuaauuccaua(N80)uugauucggaugcuccgguagcucaacucg-3', where `N' is any nucleotide and the sequences in lower case were used as primer binding sites during reverse transcription and amplification. Approximately 27 nmol of RNA in 10 mM MgCl2, 200 mM NaCl, 50 mM Bis-Tris-HCl, pH 6.4, was applied to a 4% beaded agarose column derivatized with 2.7 mM adenosine monophosphate (Sigma). Weakly bound RNA was removed with 10 column vol of buffer. Specifically bound RNA was eluted with buffer that contained 5.0 mM SAM using a rigorous elution regime described elsewhere (D.H.B. and L.G., submitted), then it was precipitated with glycogen, reverse transcribed, amplified by PCR, and transcribed for the next round of selection. cDNA made from the product of the final round was amplified with primers that contained restriction sites so that they could be cloned for sequencing.
RESULTS AND DISCUSSION
During the first four rounds of selection, <1% of the input RNA eluted specifically upon addition of free SAM, climbing to 2.3% in the fifth, 38.2% in the sixth and 50.2% in the seventh round. Not surprisingly, the final population could also be readily eluted from the AMP-derivatized resin with free 5' AMP. Thirty-five different sequences were obtained from 53 seventh-round isolates (Fig. 2 A-D). Thirty-four of these converged on a structural motif identified previously in selections of RNA aptamers for ATP and NAD+ (3 ,4 ) and known to recognize the adenine base and some features of the sugar (4 ). Consistent with the previous studies, we find two stems of variable sequence separated by an asymmetric internal bulge-loop with GGAAGAAACUG (highly conserved positions underlined) along one strand and a single bulged G in the other (Fig. 2 E). Twelve of our sequences included base pairs with the 3' fixed sequence uucgGaugc and used the uppercase G as the bulged G (Fig. 2 C). G:C was the most common base pair 5' to the bulged G throughout the population, whether or not the fixed sequences were part of the structure. A G:C pair 5' to the bulged G improves affinity ~10-fold relative to an A:U pair at the same position (4 ), thereby accounting for its predominance in our selection and previously.
Assuming that the informational content at each position (i.e., ability to tolerate nucleotide substitutions) among adenosine aptamers with equivalent affinities is reflected in the statistics of Table 1 , then the 14 nt of the conserved element (including a G29 to pair with C18), are expected to appear in one out of every 414 * Pf = 6.3 * 107 sequences, where Pf is the product of each fractional representation (shown as `percent consensus' in Table 1 ). Furthermore, one sequence in ~1500 will be able to form seven additional base pairs (total of four on each side including the C18:G29 pair, and allowing G-U pairs). As a result, at least one sequence in 1011 will contain a functional adenosine-binding element using some variation of the motif described here. Therefore, its repeated isolation in three labs from starting populations with >1014 individual members may derive largely from its overrepresentation in the initial populations, rather than its being intrinsically the `best' RNA aptamer for adenosine. By extension, there may be structurally distinct aptamers in these pools with equivalent or better affinity that have escaped detection because of their lower representation in the starting random pools. Ribozymes built from such aptamers may or may not share the features of ribozymes built from the adenosine aptamers described here.
ACKNOWLEDGEMENTS
Funding for this work was provided by grant GM19963 from NIH to L.G., by postdoctoral fellowship CHE-9302453 from NSF to D.H.B., and by additional support from NeXstar Pharmaceuticals, Inc. We thank Drs Dom Zichi and Tom Shields for assistance with the molecular graphics, and Drs Julie Feigon and Feng Jiang for helpful discussion of their aptamer structures and for sharing coordinates prior to release.
