Activation of the antiviral factor RNase L triggers translation of non-coding mRNA sequences

Abstract

Ribonuclease L (RNase L) is activated as part of the innate immune response and plays an important role in the clearance of viral infections. When activated, it endonucleolytically cleaves both viral and host RNAs, leading to a global reduction in protein synthesis. However, it remains unknown how widespread RNA decay, and consequent changes in the translatome, promote the elimination of viruses. To study how this altered transcriptome is translated, we assayed the global distribution of ribosomes in RNase L activated human cells with ribosome profiling. We found that RNase L activation leads to a substantial increase in the fraction of translating ribosomes in ORFs internal to coding sequences (iORFs) and ORFs within 5′ and 3′ UTRs (uORFs and dORFs). Translation of these alternative ORFs was dependent on RNase L’s cleavage activity, suggesting that mRNA decay fragments are translated to produce short peptides that may be important for antiviral activity.

ribofootPrinter: A precision python toolbox for analysis of ribosome profiling data

“…As with other tools [8], information is saved for both 5’ and 3’ end assignment of mapped reads. As shown previously, 3’ end assignment often offers superior reading frame alignment for ~28-nt reads from ribosomes in yeast cells, while 5’ end alignment is superior for 40S subunits in yeast or 80S ribosomes in many human cell types [4, 17, 20]. The data structure can also be modified to organize read data by footprint length.…”

“…Sample data shown here were derived from WT A549 cells and previously published as AK006F and AK014F [17]. They are publicly available on NCBI GEO: GSE158044.…”

“…gene or codon) is not weighted equally in the average; highly expressed features tend to dominate. The need to consider alternatives to normalization is becoming more critical as the importance of translation outside conventional coding sequences has become clear [7, 16, 17]. In such cases, the most appropriate method of normalization may not be immediately obvious and it is therefore advantageous to consider multiple forms of normalization.…”

Compositional biases in RNA viruses: Causes, consequences and applications

“…However, when the same authors from this latter study used IAV as a model system to experimentally test their predictions, they found that codon pair bias was far more important than dinucleotide bias, using IAV as a model system (Groenke et al, 2020 ). In this latter report, the authors found that codon pair deoptimization resulted in diminished mRNA stability (Groenke et al, 2020 ); however, the codon pair deoptimization resulted in increased UpA dinucleotide frequencies, and as UpA is reported to be cleaved by RNaseL (Wreschner et al, 1981 ; Karasik et al, 2021 ), this could also explain the outcome. A bioinformatics study that used nucleotide patterns of viruses to predict host species found the two features to be discrete, but that dinucleotide bias was far more accurate than codon pair bias in identifying viral host species (Babayan et al, 2018 ).…”

“…Two possible explanations have been put forward to date to explain genomic UpA suppression. First, it was originally reported in 1981 (and subsequently verified) that UpA dinucleotides are cleaved by the cellular ISG RNaseL (Wreschner et al, 1981 ; Karasik et al, 2021 ), which could explain their deselection over evolutionary time. However, the authors further reported that RNaseL also cleaves RNA at UpU dinucleotides, and TpT/ UpU are generally not under‐represented in animal genomes or in the viruses that infect them, so the specificity and impact of RNaseL on genomic TpA/UpA content is questionable.…”