Almost from the time that the genetic code was first deciphered, molecular biologists have tried to determine how the 30S subunit decodes mRNA with such high fidelity. The free energy for a codon- anticodon interaction is only about 2 to 3 kcal mol^-1 more favorable for a codon to interact with its cognate anticodon than for that same codon to interact with a near cognate anticodon with just a single nucleotide mismatch. This free energy difference predicts an error rate about 10 - to 100-fold greater than that actually observed. It seemed likely that the ribosome plays some role in stabilizing codon-anticodon interactions. This hypothesis received support from biochemical experiments with bacterial ribosomes that showed N1 methylation of highly conserved adenines at positions 1492 (A1492) and 1493 (A1493) of the 16S rRNA impaired A-site tRNA binding. Similar impairment also was observed when these adenines were changed to guanine or cytosine. Although these experiments indicated that A1492 and A1493 help the ribosome to recognize the shape of the codon-anticodon helix at the A-site, they did not show how they do so.
    Ramakrishnan and coworkers turned to x-ray crystallography to solve the problem. They began by soaking an oligonucleotide containing the tRNA^Phe anticodon stem-loop and a U6 hexanucleotide into crystals of the T. thermopbilus 30S ribosomal subunit. The x-ray diffraction data showed that a correct codon-anticodon match causes A 1492 and A1493 to flip out of the loop in which they are normally located and the highly conserved guanine at position 530 (G530) to switch from a syn-to anti conformation. In their new conformations, A1493 and 1492 interact with the first and second base pairs of the codon-anticodon helix, respectively, while G530 interacts with both the second position of the anticodon and the third position of the codon. These conformational changes allow the ribosome to closely inspect the first two base pairs of the codon-anticodon helix so that it can discriminate between Watson-Crick base pairs and mismatches. The environment of the third or “wobble” position appears to be able to accommodate a noncanonical base pair such as the GU wobble.