Selection of start codon during mRNA scanning in eukaryotic translation initiation

Abstract

Accurate and high-speed scanning and subsequent selection of the correct start codon are important events in protein synthesis. Eukaryotic mRNAs have long 5' UTRs that are inspected for the presence of a start codon by the ribosomal 48S pre-initiation complex (PIC). However, the conformational state of the 48S PIC required for inspecting every codon is not clearly understood. Here, atomistic molecular dynamics (MD) simulations and energy calculations suggest that the scanning conformation of 48S PIC may reject all but 4 (GUG, CUG, UUG and ACG) of the 63 non-AUG codons, and initiation factor eIF1 is crucial for this discrimination. We provide insights into the possible role of initiation factors eIF1, eIF1A, eIF2α and eIF2β in scanning. Overall, the study highlights how the scanning conformation of ribosomal 48S PIC acts as a coarse selectivity checkpoint for start codon selection and scans long 5' UTRs in eukaryotic mRNAs with accuracy and high speed.

Fig. 1. Schematic representation of the simulation sphere.

a The ribosomal 48 S PIC in open state (PDB ID: 6GSM) is shown in surface representation. The simulation sphere around P site is encircled. b A zoomed in view of the simulation sphere. The portions of initiation factors eIF1 (indigo), eIF1A (violet), eIF2α (purple), and eIF2β (blue), the initiator tRNA (green) and mRNA (pink) are shown in cartoon representation. The AUG codon at the P site of mRNA was used for calculating the binding energy for cognate start codon:anticodon (AUG:UAC) interactions. The codon at P site was mutated to different codons to calculate the binding energy for respective mutant codon:anticodon interactions. The relative binding energies were calculated with respect to the AUG start codon.

Fig. 2. Relative binding energies of non- and near-cognate codons.

Relative binding energy profiles of (a) 3- and 2- point, and (b) 1-point mutations of the start codon. Bar chart representing the relative binding energies of 3 (turquoise), 2 (gray), and 1 (pink) point mutations calculated with respect to AUG codon. The average binding energy between codon AUG with anticodon UAC interaction is −21.4 kcal/mol, calculated using MMPBSA, over the four independent simulations. The error bars indicate the standard errors obtained from the mean of four independent simulation runs. Supplementary Data 1 contains the relevant source data.

Fig. 3. Relative binding energy profiles of near cognate (GUG and AUA) codon-anticodon interactions with respect to AUG. The simulation runs were carried out in the presence and absence of different eIFs.

For GUG and AUA, for each case, the relative binding energy is calculated with respect to AUG codon for the similar system and then the average relative binding energy is plotted as bar. The error bars indicate the standard errors obtained from the mean of relative binding energies from four independent simulation runs. AUA shows much lower relative binding energy in the absence of eIF1 alone, eIF1 and eIF1A or eIF2α. Supplementary Data 2 contains the relevant source data.

Fig. 4. Recognition of the start codon AUG by 48S PIC in open conformation.

a, b The key players involved in the codon-anticodon interactions are shown in cartoon representation in two different views rotated by 180°. The corresponding residues which may augment codon: anticodon interactions are shown in stick representations. Arg36 of eIF1 (pale blue), interacts with the first two codon bases, while Trp70 of eIF1A (blue) points towards the mRNA providing stacking interactions. eIF2α shown in purple color contains Arg54 and Arg55 residues (b), which interact with mRNA. eIF2β (cyan) interacts with both tRNA_i and eIF1 at the P site. The tRNA_i and mRNA are shown in green and pink, respectively.

Fig. 5. Root mean square deviation of tRNA_i backbone and average structures.

a Root mean square deviation (RMSD) of tRNA_i backbone from the independent simulation runs of AUG (blue), AUA (green) and GUG (red), respectively. The error bars indicate the standard error obtained from the mean of four independent simulation runs. Supplementary Data 3 contains the relevant source data. b Root mean square fluctuations (RMSF) of atoms in nucleotides of the anticodon of tRNA_i in the case of AUG (blue), GUG (red) and AUA (green) MD simulation run. The fluctuations highlight the dynamic nature of tRNA_i where the nucleotides of tRNA show more fluctuations for AUA (green) when compared to cognate AUG (blue) and near cognate GUG (red) codons. The error bars indicate the standard error obtained from the mean of four independent simulation runs. Supplementary Data 4 contains the relevant source data. c–e Snapshots of the representative structures extracted from the respective (AUG, GUG and AUA) simulation runs are shown in ribbon representation. The structures are superposed keeping a rRNA stretch from the 40S body as reference. The anticodon stem loop of tRNAi (green) is more stable for c cognate and d near-cognate codons whereas it is more dynamic for e non-cognate codon. The starting and final structures for each individual run are shown in pink and blue, respectively.

Fig. 6. Dynamics of Arg54 and Arg55 of eIF2α.

Movement of Arg54 and Arg55 of eIF2α from a AUG, b GUG and c AUA simulation runs are depicted here. The coordinates were extracted from frames of the respective trajectories at 5 ns interval. The starting structures for the runs are shown in a thicker ribbon representation. The movement of mRNA and Arg residues are more prominent for AUA. d The root mean square fluctuation (RMSF) of eIF2α atoms. High fluctuation is observed for the loop containing Arg54 and Arg55. The error bars indicate the standard error obtained from the mean of four independent simulation runs. Supplementary Data 5 contains the relevant source data.