Translation initiation by the Kozak mRNA sequence is based on a conformational readout on the ribosome

The recognition mechanism of Kozak mRNA, typically comprising purines in the -3 and +4 positions flanking the AUG start codon, has remained enigmatic for decades. To address this fundamental function in eukaryotes during translation initiation, we analysed the cryo-EM structures of human 48S preinitiation complexes with mRNA point mutations. They reveal a fan-like intercalation of the pre-codon triplet into the 18S ribosomal RNA (rRNA), while the -3 purine as opposed to the pyrimidine in the non-Kozak context favours stabilization of the ternary complex between initiation factor eIF2 and initiator tRNA. Specificity towards the +4 purine is achieved beyond a single residue recognition by mutual conformational adaptations of eIF1A, mRNA and rRNA and involves the insertion of a reading head in which decoding residue A1825 (rRNA) stacks with the A-site codon to stabilize the fully accommodated state. Hence, instead of relying on base pairing as in bacteria, the specific recognition of the Kozak sequence on eukaryotic ribosomes is based on an induced-fit mechanism that triggers a conformational readout of the mRNA.