Flexizymes: their evolutionary history and the origin of catalytic function.

Transfer RNA (tRNA) is an essential component of the cell's translation apparatus. These RNA strands contain the anticodon for a given amino acid, and when "charged" with that amino acid are termed aminoacyl-tRNA. Aminoacylation, which occurs exclusively at one of the 3'-terminal hydroxyl groups of tRNA, is catalyzed by a family of enzymes called aminoacyl-tRNA synthetases (ARSs). In a primitive translation system, before the advent of sophisticated protein-based enzymes, this chemical event could conceivably have been catalyzed solely by RNA enzymes. Given the evolutionary implications, our group attempted in vitro selection of artificial ARS-like ribozymes, successfully uncovering a functional ribozyme (r24) from an RNA pool of random sequences attached to the 5'-leader region of tRNA. This ribozyme preferentially charges aromatic amino acids (such as phenylalanine) activated with cyanomethyl ester (CME) onto specific kinds of tRNA. During the course of our studies, we became interested in developing a versatile, rather than a specific, aminoacylation catalyst. Such a ribozyme could facilitate the preparation of intentionally misacylated tRNAs and thus serve a convenient tool for manipulating the genetic code. On the basis of biochemical studies of r24, we constructed a truncated version of r24 (r24mini) that was 57 nucleotides long. This r24mini was then further shortened to 45 nucleotides. This ribozyme could charge various tRNAs through very simple three-base-pair interactions between the ribozyme's 3'-end and the tRNA's 3'-end. We termed this ribozyme a "flexizyme" (Fx3 for this particular construct) owing to its flexibility in addressing tRNAs. To devise an even more flexible tool for tRNA acylation, we attempted to eliminate the amino acid specificity from Fx3. This attempt yielded an Fx3 variant, termed dFx, which accepts amino acid substrates having 3,5-dinitrobenzyl ester instead of CME as a leaving group. Similar selection attempts with the original phenylalanine-CME and a substrate activated by (2-aminoethyl)amidocarboxybenzyl thioester yielded the variants eFx and aFx (e and a denote enhanced and amino, respectively). In this Account, we describe the history and development of these flexizymes and their appropriate substrates, which provide a versatile and easy-to-use tRNA acylation system. Their use permits the synthesis of a wide array of acyl-tRNAs charged with artificial amino and hydroxy acids. In parallel to these efforts, we initiated a crystallization study of Fx3 covalently conjugated to a microhelix RNA, which is an analogue of tRNA. The X-ray crystal structure, solved as a co-complex with phenylalanine ethyl ester and U1A-binding protein, revealed the structural basis of this enzyme. Most importantly, many biochemical observations were consistent with the crystal structure. Along with the predicted three regular-helix regions, however, the flexizyme has a unique irregular helix that was unexpected. This irregular helix constitutes a recognition pocket for the aromatic ring of the amino acid side chain and precisely brings the carbonyl group to the 3'-hydroxyl group of the tRNA 3'-end. This study has clearly defined the molecular interactions between Fx3, tRNA, and the amino acid substrate, revealing the fundamental basis of this unique catalytic system.