Near-Cognate Decoding
The first clue to what property of some peptidyl-tRNAs induces frameshifting came with the completion of the S. cerevisiae genome sequence. Two groups, one in Italy and another in France, showed that yeast encodes only 41 elongator tRNAs though the rules of decoding imply there should be 45. The codons CUG, CCG, CGA and GCG all lack a cognate isoacceptor (one that could make three strong base pairs with the codon). For example, a tRNA with anticodon CGC was expected to decode GCG, but yeast has no such tRNA. Instead, GCG must be decoded by a tRNA U*GC, where U* is a modified uridine that pairs stongly wih A but weakly with G. Significantly, each of these codons stimulates efficient +1 frameshifting. Could near-cognate decoding explain that fact?
We have shown directly that all of the 11 codons stimulate frameshifting because they are decoded by near-cognate tRNAs. Codons that allow peptidyl-tRNA slippage are recognized by tRNAs that make a weakly pair with them. Generally the tRNAs use a pyrimidine-pyrimidine wobble interaction as occurs on CUU by tRNA(Leu,UAG). The three codons that stimulate out-of-frame decoding (GCG, CGA and GUG) have wobble pairs that are either weak (U*-G) or involve a purine-purine clash (I-A or I-G). We hypothesize that the unusual structure of the codon-anticodon complex in the P site disrupts efficient recognition by cognate aminoacyl-tRNA in the A site, allowing a higher proportion of errant decoding by out-of-frame cognate tRNAs.
These data suggest an explanation for why cells encode multiple isoaccepting tRNAs. The strength of recognition by the tRNAs ensures that they will not disturb in-frame decoding of the next tRNA. This effect may have been a major force in the evolution of the genetic code and of tRNA families.