Differentiating between near- and non-cognate codons in Saccharomyces cerevisiae

Abstract

Background: Decoding of mRNAs is performed by aminoacyl tRNAs (aa-tRNAs). This process is highly accurate, however, at low frequencies (10(-3) - 10(-4)) the wrong aa-tRNA can be selected, leading to incorporation of aberrant amino acids. Although our understanding of what constitutes the correct or cognate aa-tRNA:mRNA interaction is well defined, a functional distinction between near-cognate or single mismatched, and unpaired or non-cognate interactions is lacking.

Methodology/principal findings: Misreading of several synonymous codon substitutions at the catalytic site of firefly luciferase was assayed in Saccharomyces cerevisiae. Analysis of the results in the context of current kinetic and biophysical models of aa-tRNA selection suggests that the defining feature of near-cognate aa-tRNAs is their potential to form mini-helical structures with A-site codons, enabling stimulation of GTPase activity of eukaryotic Elongation Factor 1A (eEF1A). Paromomycin specifically stimulated misreading of near-cognate but not of non-cognate aa-tRNAs, providing a functional probe to distinguish between these two classes. Deletion of the accessory elongation factor eEF1Bgamma promoted increased misreading of near-cognate, but hyperaccurate reading of non-cognate codons, suggesting that this factor also has a role in tRNA discrimination. A mutant of eEF1Balpha, the nucleotide exchange factor for eEF1A, promoted a general increase in fidelity, suggesting that the decreased rates of elongation may provide more time for discrimination between aa-tRNAs. A mutant form of ribosomal protein L5 promoted hyperaccurate decoding of both types of codons, even though it is topologically distant from the decoding center.

Conclusions/significance: It is important to distinguish between near-cognate and non-cognate mRNA:tRNA interactions, because such a definition may be important for informing therapeutic strategies for suppressing these two different categories of mutations underlying many human diseases. This study suggests that the defining feature of near-cognate aa-tRNAs is their potential to form mini-helical structures with A-site codons in the ribosomal decoding center. An aminoglycoside and a ribosomal factor can be used to distinguish between near-cognate and non-cognate interactions.

Figure 1. The decoding center and dual-luciferase reporters for determining rates of translational misreading in yeast.

Panel A. The codon:anticodon mini-helix in the decoding center is stabilized by base-pairing at all three positions of the mini-helix favoring A-minor interactions with flipped out bases G520, A1492 and A1493. PyMol (Delano Scientific, LLC) was used to generate this figure based on the coordinates 1IBM in the RSCN Protein Data Bank . Panel B. In all missense reporters, transcription is initiated from the yeast ADH1 promoter, and terminated at a sequence from the CYC1 3′ UTR. The luciferase genes from Renilla and firefly are cloned in frame to produce a fusion of the two proteins. The sense reporter has the AGA codon encoding arginine at amino acid residue 218 in the catalytic site of firefly luciferase. Missense reporters contain the indicated mutations at this position, which encode the indicated amino acids. Efficiencies of missense suppression were calculated by dividing the ratio of firefly/Renilla luciferase generated from cells harboring the missense test vectors by the ratio of firefly to Renilla luciferase generated from cells harboring the sense control plasmid.

Figure 2. Proposed basis for near-cognate codon-anticodon interactions.

Top. Base pairing between tRNA^Arg2 anticodon and UGU codon (left), and between tRNA^Arg3 anticodon and the AGC and AGU codons. Bottom. I•U, U•C, and U•U base pairing.

Figure 3. Effects of the eEF1 complex mutants on mis-reading of near- and non-cognate codons.

Misreading of the non-cognate UCU and near-cognate AGC codons by mutant forms of eEF1A (Panel A), or by isogenic strains with tef3Δ, tef4Δ, or tef3Δ tef4Δ double null mutants or tef5Δ strains expressing the K120R S121Δ I122Δ allele compared to isogenic wild-type strain (Panel B). Panel C. Misreading of all seven missense codons by cells lacking both forms of eEF1Bγ (tef3Δ tef4Δ). Effects of the indicated mutants are depicted as fold of isogenic wild-type cells. ** indicates p values of <0.01; * indicates p values of <0.05.

Figure 4. Characterization of alleles of RPL5.

Panel A. Effects of the K27E rpl5 mutanton misreading of seven missense codons. Effects on the indicated missense reporters are depicted as fold of isogenic wild-type cells. ** indicates p values of <0.01. Panel B. Paromomycin dilution spot assays. Ten-fold dilutions (10⁶→10¹ CFU) of logarithmically growing cells were arrayed onto H-leu medium containing paromomycin (1 mg/ml) or no drug control plates. Cells were grown at 25°C for 3 days.

Figure 5. Modeling of mutations in eEF1A and eEF1Bα that influence influence misincorporation of missense aa-tRNAs.

The nucleotide exchange factor eEF1Bα and the fitted aa-tRNA present clashes indicating that they do not interact with eEF1A simultaneously. PyMOL (Delano Scientific, LLC) was used with the coordinates 1G7C of yeast eEF1A:eEF1Bα (amino acids 114–206) in complex with GDPNP . The ribbon structure of eEF1A is shown in blue, eEF1A mutated bases are shown in cyan, and GDPNP is indicated in magenta. Panel A. tRNA (yellow) was fitted into the structure based on coordinates obtained from the crystal structure of the EF-Tu:Phe-tRNA^Phe:GCPNP complex (1TTT in the RSCN Protein Data Bank, . Panel B. Ribbon structure of eEF1Bα from 1G7C is shown in red, and the KSI residues in the mutant form used in this study are indicated in salmon.