The amino acid-trinucleotide relationships of the genetic code have long prompted interest in finding a historical rationale for why particular nucleotide triplets correspond to specific amino acids. Efforts to find stereochemical complementarity between a particular amino acid side chain and a specific triplet have been largely unsuccessful, as have been attempts to demonstrate specific amino acidtrinucleotide complexes in solution. Because the connection between specific amino acids and trinucleotides is made through the aminoacylation of transfer RNAs (tRNAs), the catalysts of aminoacylations and their interactions with tRNAs have been intensively studied (Gieg~ et al., 1993; Saks et al., 1994; Martinis and Schimmel, 1995; McClain, 1995). These investigations already have shown that some aminoacyl-tRNA synthetase catalysts do not interact with the anticodon trinucleotide, establishing that for these examples the relationship between a specific trinucleotide and a specific amino acid is indirect. More typically, anticodonsynthetase interactions occur and contribute significantly to aminoacylation efficiency and specificity (Schulman, 1991; Saks et al., 1994). However, specific aminoacylations still take place when anticodons are deleted from tRNA structures (Frugier et al., 1994; Martinis and Schimmel, 1995). Collectively, these observations have established that nonanticodon-containing RNA sequence and structure per se contain information that is interpreted as an amino acid. Since the discovery of.RNA catalysis, tRNAs have increasingly been viewed as molecules that need to be understood in terms of how the theater of proteins emerged from an early RNA world. Based on recent experimental observations, RNA-catalyzed aminoacylation of tRNA-like molecules is plausible (lllangasekare et al., 1995). Because the aminoacyl ester linkage is higher in energy than the amide bond, two or more aminoacyI-RNAs could in principle congregate to yield peptides spontaneously. Whether early peptides led directly to aminoacyl-tRNA synthetases is not known, but many believe that these are ancient enzymes that were among the first proteins to emerge from the RNA world. This realization has added motivation to understanding the design and evolution of tRNA synthetases and the mechanisms by which they interpret RNA sequences and structures in terms of specific amino acids. tRNA Structure Can Be Viewed as Two Domains with Distinct Functions Although tRNAs vary in size from 75 to 93 nucleotides, they typically are comprised of 76 nucleotides that can be arranged as a cloverleaf structure with self-complementary bases making four distinct helical segments and three loops. The 3' ends of all tRNAs end in the single-stranded sequence N73CCAo., where N73 is any of the four nucleotides and the free 2'and 3'-hydroxyl groups on the terminal adenosine contain the amino acid attachment site. The parts of the cloverleaf structure are the acceptor stem, the dihydrouridine (D) stem-loop, the anticodon stem-loop, and the T~C stem-loop. This structure is folded in three dimensions into an L-shaped molecule that consists of two domains (Figure 1). One domain is formed by coaxial stacking of the T~'C stem onto the acceptor helix, while the other results from stacking the anticodon stem onto the D stem. In this structure, the trinucfeotide anticodon of the genetic code and the amino acid attachment site are segregated between the different domains, where they are 76/~ apart. These two structural domains are also distinct functional units. The anticodon-containing domain serves as a reading head that interacts with the mRNA template. The acceptor -T~C minihel ix that terminates in the NCCA tetranucleotide is a substrate for sequence-specific aminoacylation with many different amino acids. Aminoacylation specificity and efficiency generally depend on the nature of the N 73 "discriminator" base and on one or two base pairs among the first four in the acceptor stem. Thus, the sequences and structures of tRNA acceptor stems can be thought of as an operational "RNA code" that is devoid of the anticodon trinucleotides of the genetic code (Schimmel et al., 1993; Frugier et al., 1994). Two Domains of tRNAs Interact with Separate Domains of tRNA Synthetases All tRNA synthetases catalyze the same basic reaction: activation of an amino acid by reaction with ATP to form a bound aminoacyl adenylate and reaction of the bound adenylate with the cognate tRNA to form am'inoacyl-tRNA.