N6-Adenosine Methylation of a Single Site in SARS-CoV-2 5′-UTR Promotes Translation

Reviewer #1 (Public review):

Summary:

A prevailing view is that translation of 5' capped mRNAs, i.e. mRNAs that are translated via ribosome scanning, is inhibited by highly structured 5' untranslated regions (5' UTRs). Despite having a common, structured 5' UTR, the mRNAs produced by the SARS-CoV-2 virus are efficiently translated. In this study, the authors identified a DRACH motif in stem-loop 3 (SL3), suggesting a potential site of m6A methylation of A74 by the enzyme METTL3. Given that such m6A modifications are known to disrupt RNA structure formation, the authors tested the hypothesis that this may be the basis underlying the efficient translation of these mRNAs. Mutational approaches complemented by METTL3 siRNA knockdown were employed to support this hypothesis. Additional experiments showed that this is required for efficient association of a reporter mRNA with polysomes (indicative of active translation), and suggest that the 5' UTR is more highly structured when methylation is abrogated.

Strengths:

The data clearly indicate that N6 methylation of A74 is required for efficient translation of SARS-CoV-2 mRNAs.

Weaknesses:

While the evidence supports the authors' central hypothesis, there are two issues that should be addressed. The first is that all of the approaches are indirect. All of the evidence for the presence of mRNA structural elements is based on computational and genetic analyses. We now know that there is something there, but we still do not know what it is. The authors need to use a biochemical approach to actually map the structural elements of the 5' UTR and determine how such structure(s) are changed by loss of methylation. The second hinges on the assumption that these mRNAs are translated via canonical ribosome scanning. RNA viruses are well-known to use a variety of other mechanisms, e.g. internal ribosome entry signals and ribosome tethering, to promote efficient translation. Alternatives to ribosome scanning should be considered.

Reviewer #2 (Public review):

The study addresses the conundrum of how the mRNAs of SARS-CoV-2 are efficiently translated since the 5' leader, which has common elements for all the viral genes, is highly structured. The authors test the hypothesis that m6A modification at position 74 is key to this translation. First, the authors show convincingly that this site is modified. Then, with extensive transfected reporter experiments using luciferase assays as well as sucrose gradient sedimentation, this modification is shown to be key for efficient translation. While the mechanism of this effect is not entirely clear (see comments/suggestions below), the authors show that it is independent of YTH "reader" proteins and likely involves altered interactions between SL3 (which contains A74) and downstream elements in the UTR. These results are important because they both offer insight into the function of m6A in gene expression and suggest how they may be important for the translation of viral mRNA in particular.

While the data on their own make the overall case that the m6A modification in the 5'UTR of the viral genes is important to their expression, there are several things worth considering that could refine the model and make it more convincing.

It is not clear whether putative uORF translation, particularly translation of the uORF that begins with a CUG codon at position 59 in the 5'UTR (as shown in Finkel et al., Nature 2020), would be impacted by this modification (as it includes the putative m6A site at position 74). It is also worth considering whether SL3 melting by translation of this uORF would alter the proposed mechanism.

It is a bit unclear why the A74T mutant was put in the longer construct while the C75G mutant was put in a shorter construct. While not essential, the mechanistic arguments would be stronger if the same construct had been used to compare the mutations.

While the authors show that "global depletion of m6A modification does not grossly alter translation efficiency" in a general sense (page 11), it would be of interest to know whether any host mRNAs with 5'UTR m6A (i.e., ACTA2 and COX8A, mentioned in this study) are affected by the mechanism here (i.e. run them in the luciferase assay).

The authors show that the YTH "reader" proteins have a very small inhibitory effect (1.6-fold) on the translation of the viral mRNA with 5'UTR m6A. However, it remains unclear how important this is or whether it is generally true for host mRNAs with this modification.

It is reassuring to see controls for changes in RNA levels in the supplemental material. The RNAs were generally stable under the experimental parameters explored, which would rule out RNA-decay-based mechanisms of m6A regulation. However, it should be noted that mRNA level experiments appear to have been done at 24 h while luciferase measurements were done at 48 h (as noted on p. 23, gene expression vs luciferase activity). It is not clear whether any RNA decay phenotypes would be apparent at 24 h.

The authors use the term "ribosome profiling" (for example, on page 10), but it would appear the experiment performed is actually "polysome profiling" or "sucrose gradient sedimentation" since it did not involve ribosome footprinting.

Reviewer #3 (Public review):

Aly et al investigate the potential for a single N6-methyladenosine RNA modification in the context of the 5' UTR sequence of SARS-CoV-2 to regulate translation of a downstream luciferase reporter transfected into cells. They show using meRIP (m6A RNA IP) that this site is methylated in the plasmid-driven transcript, and convincingly show it mediates reporter translational efficiency using knockdown of the m6A methyltransferase METTL3 and mutation of the modified UTR site together with analysis of the transcript's association with polyribosomes. They suggest that the benefit to translation conferred by the modification is through its effect on the secondary structure of the 5' UTR, based on an RT-PCR-based assay in control and METTL3 knockdown cells linking RT processivity to translation (luciferase) output. They also extend their conclusions to two cellular mRNA 5' UTRs, also reported to contain a single m6A modification, and show METTL3-dependent changes in RNA structure stability, hinting at a broader significance of this mechanism of m6A control of gene expression.

The conclusions of the paper are mostly well supported by the data presented, though validation of knockdown of METTL3 (and reader proteins) is absent.

A major limitation of the work is the exclusive use of the reductionist artificial reporter system in uninfected cells. Though the 5' UTR site they identify is methylated in the context of a transcript generated in the nucleus (where the m6A installing complex is mainly localized, and believed to act exclusively in uninfected cells), how frequently this site is modified, if at all, on viral RNAs generated within cytoplasmic membrane-bound replication organelles. Similarly, whether the translation regulation by a single m6A modification identified here occurs within the context of an infected cell, in which there are many changes to the RNA and translational regulatory landscape, also remains to be tested.

How this work can be reconciled with others that have concluded either little potential for translational regulation by 5' UTR modification (Guca et al 2024; PMID: 38244546) or that an eIF3-mediated mechanism is responsible (Meyer et al, 2015 PMID: 26593424) is not addressed in the discussion.