How does N1-methylpseudouridine cause ribosomal frameshifting in mRNA vaccines?

m1Ψ drives +1 frameshifting when peptidyl-tRNA pauses in the P site; synonymous codon swaps can remove the risk.

Direct answer

N1-methylpseudouridine (m1Ψ) raises +1 frameshifting on slippery UUUC motifs because it destabilizes the codon–anticodon pair held in the ribosomal P site, letting the peptidyl-tRNA slip into a non-canonical state that re-pairs in the +1 frame. The effect depends on the motif: replacing UUUC with UUCC or UUUU removes it, and in the BNT162b2 reporters only four of six slippery motifs showed a significant increase.

1sources cited

This article was generated with WisPaper-powered search and paper analysis.

A shift of one nucleotide creates a new peptide

Ribosomal frameshifting is a decoding error in which the ribosome moves by one nucleotide instead of three, so the reading frame changes and a different peptide is produced. The paper's reporters make that error visible: a red fluorescent protein (mCherry) is translated in the normal, 0 frame, and a green one (sfGFP) is only produced if translation slips into the +1 frame past the 0-frame stop codons. The ratio of green to red signal therefore reports how often the shift happened, calibrated against a construct in which sfGFP sits in frame (100% frameshifting) and one lacking sfGFP (0%).

The concern behind the measurement is immunological. Aberrant +1-frame peptides could act as unintended antigens, so the question is not only whether m1Ψ causes frameshifting but how much, on which sequences, and whether the sequence can be redesigned to avoid it.

Only some slippery motifs respond to m1Ψ

The authors built reporters carrying the slippery UUU(U/C) motifs found in the BNT162b2 spike-coding sequence, with the vaccine's own 5′ and 3′ untranslated regions, and compared unmodified with m1Ψ-modified mRNA in HEK293T cells. Four of the six motifs showed a statistically significant increase in +1 frameshifting under m1Ψ; two did not. The same approach was applied to mRNA-1273 motifs, but low translation efficiency prevented an accurate frameshifting calculation there.

That split matters for reading the result. A slippery sequence alone does not determine the efficiency of the error, so the identity of the motif and its surroundings carry information the motif sequence does not. The measured efficiencies were also 3- to 7-fold lower than in earlier studies, which the authors attribute to differences in reporter sensitivity, and they fall in line with reports of frameshifting below 5%, the detection limit of ribosome profiling.

The same reporters were used to ask whether m1Ψ promotes stop-codon readthrough on the two consecutive UGA stop codons of BNT162b2. Readthrough stayed close to the assay's roughly 1% detection limit in both unmodified and m1Ψ-modified mRNA, so the paper concludes that m1Ψ does not drive readthrough above that limit during translation of this vaccine mRNA.

The slip does not happen while reading the codon

A natural hypothesis is that m1Ψ slows or destabilizes decoding of the slippery codon itself, giving the ribosome time to slip. The authors tested this with single-molecule FRET, using bacterial ribosomes labeled on protein L11 and Phe-tRNA labeled on the acceptor dye, so that binding, codon reading, and accommodation of the tRNA could be followed in real time on UUUC versus m1Ψm1Ψm1ΨC codons. The mix of trace classes, the dwell times for initial binding and codon reading, and the fluctuations after accommodation did not differ significantly between the two codons. A GTPase-deficient EF-Tu mutant, which makes release from the codon-reading state rate-limiting, likewise showed indistinguishable dwell times, arguing that m1Ψ does not change codon–anticodon stability in the A site.

Lowering the concentration of the tRNA that reads the slippery codon reduced 0-frame translation but did not increase +1-frame product, so slow decoding is unlikely to be the driver. The authors also slowed translocation by depleting EF-G: a known −1 frameshifting sequence responded as expected, while m1Ψm1Ψm1ΨC did not gain +1 frameshifting, arguing against stalling of the peptidyl-tRNA in the A site.

P-site destabilization is the frameshifting step

The remaining candidate is the P site, where the growing peptide is held on its tRNA. Measuring how fast peptidyl-tRNA dissociated from the P-site codon showed faster loss on m1Ψm1Ψm1ΨC than on UUUC, with a two-phase decay indicating more than one kinetic population. In the single-molecule experiments, peptidyl-tRNA normally sat in a stable P/P state; on m1Ψ-modified codons a larger fraction of traces fluctuated into a non-canonical state, 37 ± 10% versus 19 ± 6% by trace counting (30 ± 8% versus 15 ± 6% in the population analysis).

The authors' interpretation is that m1Ψ weakens 0-frame pairing in the P site and shifts the tRNA toward a conformation from which re-pairing in the +1 frame is more favorable. Frameshifting then occurs when the A site is empty and the next cognate tRNA has not yet arrived, a window the paper calls hungry P-site stalling. This mechanism was dissected in the bacterial translation system; the authors state that it remains to be shown in human translation, and they found no correlation between frameshifting and estimated tRNA availability in their reporter system, while noting that cell-type-specific tRNA pools could still matter physiologically.

Changing UUUC to UUCC or UUUU removes the effect

If the slippery motif is required for re-pairing in the +1 frame, a synonymous replacement should abolish the error. Replacing UUUC with UUCC, which would force an A–C mismatch for the phenylalanine tRNA in the +1 frame, eliminated +1-frame product in vitro. In human cells, the same substitution removed the difference between unmodified and m1Ψ-modified BNT162b2 reporters, and it also removed the difference between slippery and non-slippery m1Ψ-modified mRNA-1273 sequences in the dual-fluorescence assay.

The fourth position of the motif also matters. In the BNT162b2 screen, motifs ending in C and motifs ending in U behaved differently, which prompted a test of the downstream codon. Keeping the slippery UUU codon fixed, the authors replaced the following codon with sense codons beginning with C or U. Codons starting with C significantly increased +1 frameshifting on m1Ψ-modified mRNA, whereas codons starting with U showed no change, and the effect was specific to the first codon position. The fold change did not correlate with ribosome A-site dwell times estimated from ribosome profiling, which the authors read as evidence against tRNA availability as the explanation in this system.

Codon choice becomes a fidelity variable

The practical output is a sequence-level rule rather than a new modification. Because the frameshifting depends on a slippery UUUC motif and on a downstream codon whose first position is C, the paper's own comparison implies that synonymous recoding can reduce the error without changing the encoded protein: prioritize non-slippery synonymous motifs and avoid downstream codons beginning with C where the design allows. The authors note that synonymous replacement is not always free, since synonymous substitutions can affect protein folding, which is why the downstream-codon route is presented as an additional option.

The paper also points the rule in the opposite direction. Deliberately promoting error-prone translation could diversify the neoantigen pool for personalized cancer vaccines, particularly in tumors with low mutational burden, though this is framed as a possibility rather than a demonstrated vaccine result. The broader claim the authors make is that effects on translation fidelity should be assessed alongside immunogenicity and expression when judging a clinically relevant mRNA modification.

The rule is motif-specific, not a general m1Ψ penalty

The most useful way to read this paper is as a map of when m1Ψ-induced frameshifting appears and when it does not. It appears on UUUC motifs in the vaccine reporters, in four of six BNT162b2 sites tested, and it is amplified by a downstream codon starting with C; it disappears when the motif becomes UUCC or UUUU, and it was not detected as stop-codon readthrough above roughly 1%. The measured efficiencies were low and assay-dependent, 3- to 7-fold below earlier reports, so the absolute rate should be read with the reporter in mind rather than as a fixed property of m1Ψ.

The mechanistic dissection relied on bacterial ribosomes, and the authors explicitly leave open whether the same P-site pathway operates in human translation; the human-cell evidence is the frameshifting readout, not the single-molecule mechanism. Within those bounds, the decision-relevant comparison is between sequence variants of the same antigen: the same m1Ψ modification can be nearly silent on one slippery motif and measurable on another, so codon-level design, not the modification alone, determines whether the unintended +1-frame peptide is made.

Sources used in this answer

1

Recoding by N1-methylpseudouridine guides rational mRNA vaccine design

m1Ψ increases +1 frameshifting on slippery UUUC motifs by destabilizing P-site codon–anticodon pairing and promoting a frameshifting-prone peptidyl-tRNA state; replacing UUUC with UUCC or UUUU eliminates the effect.