#apaperaday: mRNA-specific readthrough of nonsense codons by antisense oligonucleotides (R-ASOs)

In today’s #apaperaday, Prof. Aartsma-Rus reads and comments on the paper titled:    mRNA-specific readthrough of nonsense codons by antisense oligonucleotides (R-ASOs)

Today’s pick from @susorov et al in @NAR_Open on antisense oligonucleotides (ASOs) to improve stop codon readthrough for nonsense mutations. DOI: 10.1093/nar/gkae624.

Nonsense mutations are mutations where the code for an amino acid is changed into a stop code by the change of 1 base pair (DNA)/nucleotide (RNA). Nonsense mutations are found in many genes and collectively account for ~10% of all pathogenic variants.

Nonsense mutations differ from frame-shifting mutations: the gene code for the protein in principle is there, if only that nonsense mutation was not used by the cell to stop protein translation. Frame-shifts mutations by contrast will lead to incorporation of wrong amino acids.

Stop codons are useful because when a protein is translated, the ribosomes (the system in the cell that translates RNA into protein), need to know the protein is complete and release it. This is done by eRF1, the protein release factor.

Nonsense mutations are stops that occur too early, before the true stop and thus a shorter protein is made, that is not functional. If however, you could suppress the use of this nonsense mutation, a full-length functional protein could be produced.

How can you suppress this? Different approaches have been used, e.g. suppressing the eRF1 protein release factor or modifying transferRNAs (t-RNA) to recognize the stop codon. There are also drugs that can suppress stop codon use, e.g. aminoglycosides like G418 and gentamicin.

Note that the dose at which these compounds suppress stop codon use are toxic. E.g. gentamicin has been tested for Duchenne to suppress nonsense mutations, but chronic use can cause deafness and kidney problems, so this is not suitable for chronic treatment.

Back to the paper: authors argue that one of the reasons these compounds are toxic, is that they do not act specifically but rather suppress all stops – the mutated ones and the true ones. Here they propose to use ASOs to achieve targeted stop codon read through.

Authors made a luciferase reporter system with a common nonsense mutation in CFTR (gene involved in cystic fibrosis). This is a UGA mutation, on position 4 (i.e. 1st code after UGA) there is a G, which makes this a strong stop. Authors made a weaker stop (UGAC) as well.

They also used a wild-type reference. Hooray for wild types :). They showed that both UGAG and UGAC reporters had 20 times less luciferase than the wild type and confirmed this also on protein level by western blot. Then they designed DNA ASOs targeting different positions.

For the weak stop, the ASOs only moderately improved translation, except for one ASO which increased luciferase 7-fold. This targeted position +8 (i.e., the 5th position after the stop codon). This ASO had no effect on the translation of the wild-type construct.

For the strong stop, this ASO increased translation only a little bit. Authors speculate that likely the eRF1 binding is too strong to be prevented. Authors then combined G-418 with the ASOs, and the readthrough efficiency was increased for both the weak and strong stop.

The weak stop achieved close to wild-type levels of luciferase activity, while the strong stop reached 25% of wild-type levels. Authors tested different stop codons, showing that UGA is the strongest one, followed by UAG and UAA—the weaker the codon, the easier the readthrough.

Authors discuss G-418 is toxic, but that combining it with ASOs allowed using lower doses to achieve readthrough. Authors also checked the location of the ASO, showing that the +8 location was crucial, with less readthrough activity for positions 9-11 and none for position >12.

Also, earlier positions were not effective. Authors used DNA ASOs in their lysate system, but for studies in cells and in vivo, you need chemical modifications or the ASO will be cleaved by nucleases or they will induce RNase-mediated cleavage.

Authors tested various chemical modifications, showing that the 5′ of the ASO had more tolerance for modifications, while in the 3′ end, LNAs to some extent were tolerated, while 2OMe and 2F modifications were not tolerated. A PS backbone reduced readthrough activity.

ASOs with mismatches also rendered ASOs ineffective. Authors then tested another nonsense mutation in their reporter system with its specific nucleotide context (it was again a strong stop). Also, there the +8 position was the best target for readthrough.

Finally, authors tested other nonsense mutations in different genes, where in each case the +8 position was most optimal. Authors also refer to a recent publication where complete restoration of protein production was reported with an ASO targeting a stop in the HBB gene.

Authors tried to reproduce this finding but were unable to. The authors of the HBB paper found that ASOs targeting +4+12 worked best. Authors of the current paper used the reported ASOs and also an ASO targeting +8 but were unable to find readthrough activity in their hands.

Note that numbering is tricky as there is no consensus on where to start counting, apparently. Authors start at the stop codon, but others start at other locations. Important to consider when comparing work (authors fortunately realized this).

Authors discuss that apparently the +8 position, so 5 after the stop codon, is important as this allows the mRNA to go into the eRF1 pathway, leading to cleavage from the ribosome and protein release. The G at position 8 is best for readthrough purposes, followed by A and pyrimidines.

Very nice work and I appreciate authors giving context so those not familiar with how stop codons work exactly can still follow the story. However, the G-418 will not be usable in humans chronically. Furthermore, authors do not discuss nonsense-mediated decay.

Nonsense-mediated decay is a process that cleaves transcripts with a nonsense mutation. This happens the first time a transcript is translated by the ribosomes. This means for ASO-mediated readthrough to happen, the ASO has to target the transcript the first time it is translated.

Otherwise, the transcript is cleaved and there is no target for the ASO. In the test system the authors used, there is no nonsense-mediated decay (I think), so they will not have had to deal with this. I would like to see how this works in more physiological settings.

So more work is needed, but this is possible because authors did the groundwork in their artificial system and optimized and tested things thoroughly with all the required references and controls.