Geneticin reduces mRNA stability

Abstract

Messenger RNA (mRNA) translation can lead to higher rates of mRNA decay, suggesting the ribosome plays a role in mRNA destruction. Furthermore, mRNA features, such as codon identities, which are directly probed by the ribosome, correlate with mRNA decay rates. Many amino acids are encoded by synonymous codons, some of which are decoded by more abundant tRNAs leading to more optimal translation and increased mRNA stability. Variable translation rates for synonymous codons can lead to ribosomal collisions as ribosomes transit regions with suboptimal codons, and ribosomal collisions can promote mRNA decay. In addition to different translation rates, the presence of certain codons can also lead to higher or lower rates of amino acid misincorporation which could potentially lead to protein misfolding if a substituted amino acid fails to make critical contacts in a structure. Here, we test whether Geneticin-G418, an aminoglycoside antibiotic known to promote amino acid misincorporation-affects mRNA stability. We observe that G418 decreases firefly luciferase mRNA stability in an in vitro translation system and also reduces mRNA stability in mouse embryonic stem cells (mESCs). G418-sensitive mRNAs are enriched for certain optimal codons that contain G or C in the wobble position, arguing that G418 blunts the stabilizing effects of codon optimality.

Fig 1. G418 destabilizes mRNA in vitro.

(A) Rabbit reticulocyte lysate was used to translate an mRNA encoding firefly luciferase in the presence of G418, cycloheximide, or puromycin at increasing inhibitor concentrations (concentrations were 0.5, 5.0, and 50 ng/μL for G418 and puromycin and 0.25, 2.5 ng/μL, and 25 ng/μL for cycloheximide) to identify concentrations with modest translation inhibition. (B) At intermediate antibiotic concentrations (5.0 ng/μL G418 and puromycin, 2.5 ng/μL cycloheximide), G418 destabilizes mRNA relative to control reactions and reactions with other translation elongation inhibitors. After 30 min, in vitro translation reactions containing G418 have significantly less mRNA (*, p < 0.01) compared to control reactions, and cycloheximide and puromycin both stabilize mRNA compared to control reactions without inhibitors (*, p < 0.01, p-values from Student’s t-test). (C) Translation reactions were prepared as in (B), but firefly luciferase protein levels were measured. Very little firefly luciferase production is observed at 10 minutes, but then firefly luciferase accumulates over the remaining time course. Antibiotics consistently reduce luciferase production at all time intervals. (D) Again, translation reactions were prepared as in (B) and (C). Reactions were quenched at 15 min and loaded onto a sucrose gradient for ribosome fractionation. Fractions were collected dropwise, and nucleic acid content was quantified at 260 nm. Absorbances for sequential fractions are plotted. Regardless of treatment, a large 80S monoribosome peak was observed with minimal polyribosome peaks. Therefore, the majority of ribosomes in reticulocyte lysate exist as monoribosomes. (E) Representative fractions from (D) were probed for firefly luciferase mRNA using RT-qPCR. Consistent with the overall profile in (D), the majority of firefly luciferase mRNA can be found in fractions from the top of the gradient (Free RNA) or the 80S monoribosome peak. Minimal, but detectable firefly luciferase mRNA can be isolated from fractions in the polyribosome region of the gradient. These data suggest that most mRNA in reticulocyte lysate is translated by monoribosomes. (F) Proteins from in vitro translation reactions were analyzed by western blotting for firefly luciferase. Only full-length protein was visible (*), indicating that G418 treatment did not cause high rates of stop codon readthrough. Due to its mechanism, it is unlikely that treatment with G418 would affect the total protein level as measured here since it promotes amino acid misincorporation. Altogether, these data suggest that G418 destabilizes mRNA.

Fig 2. Amino acid misincorporation drives mRNA instability in mESCs.

(A) mESCs were treated with increasing concentrations of G418 and puromycin in the presence of azidohomoalanine (AHA). AHA incorporation was monitored by fluorescence after conjugating AlexaFluor 488 to AHA using Click chemistry. At the indicated concentrations (*), G418 and puromycin both significantly depressed new protein synthesis (p < 0.01, Student’s t-test). (B) SLAM-seq analysis was used to calculate mRNA half-lives in mESCs, comparing control cells to those grown in the presence of G418 (higher amino acid misincorporation rates) and puromycin (abortive translation elongation). Shown are violin plots for ~10,600 mRNA half-lives in the three conditions. P-values (*, p = 1.1 e-5, **, p = 3.3 e-16, and ***, p = 8.9 e-36) were calculated using the Mann-Whitney U-test. (C) Shown are codon stability scores for the fraction of codons in an mRNA correlated with mRNA half-lives. Positive correlations mean that an amino acid codon is more likely to be present in a stable mRNA (Stabilizing) and vice-versa (Destabilizing). Codons are arranged by increasing CSCs for mESCs grown under control conditions, and codon sequences are given in the graph with coloring according to wobble position nucleotide (green are AU3 codons, and purple are GC3 codons). In all cases, mESCs treated under various conditions had similar correlation coefficients, and there is a general trend with stabilizing GC3 codons. (D) mRNAs were divided into G418-sensitive (Sensitive) mRNAs and all other mRNAs (Insensitive) by calculating the ratio between mRNA half-lives in G418-treated versus puromycin-treated mESCs (see Materials and Methods). Then average codon optimality (CSC score) was calculated for each group. There is no significant difference in CSC scores between groups. (E) Similarly, ribosome density (from ref. [28]) was compared for G418-sensitive mRNAs to all other mRNAs. We again observed no difference in average ribosome density.

Fig 3. G418 preferentially destabilizes mRNAs with G/C nucleotides in the wobble position.

(A) We calculated the fraction of nucleotides at each codon position and correlated those fractions with mRNA half-lives. Consistent with published results [16], we observed destabilizing effects if the wobble position was occupied with either an A or U. Treatment with G418 or puromycin yielded results that were consistent with control mESCs. (B) We analyzed individual codons to see which were overrepresented or underrepresented in the pool of G418-sensitive mRNAs. Many of the codons with a U in the wobble position and encoding hydrophobic amino acids were underrepresented in the G418-sensitive mRNAs. Additionally, AU3 Asp and AU3 Asn codons were found. These did not consistently align with suboptimal codons. Since the Asp and Asn codons are known to have higher rates of amino acid misincorporation [38], they may act as a sensitized background to observe effects of amino acid misincorporation. Overrepresented codons contained GC3 in the wobble position. Codons were only labeled as overrepresented or underrepresented if p < 0.01 (from a Mann-Whitney U-test).