Molecular fossils of an ancestral code revealed by large-scale analysis of tRNAs.

Modern tRNAs are thought to have evolved from simpler RNA minihelices resembling acceptor stems, some of which could be specifically aminoacylated. A central and testable prediction of minihelix-based models is that vestiges of ancestral coding relationships should remain embedded within the acceptor stems of contemporary tRNAs and should persist after correction for structural constraints. Here, we ask whether cognate anticodon-like triplets are statistically enriched in nonanticodon regions of modern tRNAs beyond what would be expected from structural and compositional constraints alone. Using a curated dataset of 77,335 canonical tRNA sequences and a structure-aware statistical framework, we find that for a subset of amino acids cognate anticodon-like triplets are significantly enriched in acceptor stems and, to a lesser extent, in other structural domains. Notably, these amino acids include those previously shown to support specific minihelix aminoacylation [Y. M. Hou, P. Schimmel, Nature 333, 140-145 (1988); C. Francklyn, P. Schimmel, Nature 337, 478-481 (1989)]. The strongest acceptor-stem signals survive structure-preserving null models, and cluster nonrandomly against proposed chronologies of amino acid incorporation into the genetic code. Together, these findings identify molecular fossils within modern tRNAs consistent with the persistence of an ancestral operational code, and provide refined statistical support for the notion that extant codon-amino acid relationships evolved from early aminoacylation systems involving RNA minihelices.