What is that about the Poly-A-Tail

“Bayer sues COVID-19 vaccine manufacturers

The pharmaceutical company Bayer has sued COVID-19 vaccine manufacturers Pfizer, BioNTech, and Moderna. The accusation: they allegedly infringed on patents with their mRNA vaccines.

The agricultural and pharmaceutical group Bayer is taking COVID-19 vaccine manufacturers Pfizer, BioNTech, and Moderna to court over allegations of patent infringement regarding their mRNA vaccines. A Bayer spokesperson confirmed that corresponding lawsuits have been filed at a federal court in the US state of Delaware.”1

The patent holders claim to have found a method (or a series of sequence modifications) that overcomes this unpredictability. By specifically designing certain sequence segments (e.g. the 3’ end region, including the poly-A signal area), one can:

• Control processing.

• Increase stability in the cytoplasm.

• Prevent unwanted “cryptic” processing (which leads to inactive mRNA).

In this patent dispute, a central point revolves around specific codon optimizations in the 3' UTR region and the poly-A tail.2 This has sparked significant interest in re-examining the patent. Consequently, this article was written, taking an unexpected turn as it revisits the issues surrounding frameshifting and structural configuration.

Let me quote two very interesting Paragraphs:

“Site directed mutations in this sequence can disrupt these functions (Conway and Wickens, 1988; Wickens et al., 1987). However, it has also been observed that sequences up to 50 to 100 bp 3′ to the putative polyA signal are also required; i.e., a gene that has a normal AATAAA but has been replaced or disrupted downstream does not get properly polyadenylated (Gil and Proudfoot, 1984; Sadofsky and Alwine, 1984; McDevitt et al., 1984). That is, the polyA signal itself is not sufficient for complete and proper processing. It is not yet known what specific downstream sequences are required in addition to the polyA signal, or if there is a specific sequence that has this function. Therefore, sequence analysis can only identify potential polyA signals.”

This means that the primary recognition signal (such as AATAAA), viewed in isolation, is insufficient for the correct and complete processing (polyadenylation) of the mRNA. Site-directed mutations in this area disrupt the function. For the process to work, additional sequences must necessarily be present, located 50 to 100 base pairs “downstream” (in the 3′ direction) of the actual signal.

If these subsequent sequences are missing, replaced, or disrupted, polyadenylation fails; even if the primary poly-A signal is completely intact. It is scientifically still unknown which exact downstream sequences are required in addition to the signal for it to function.

Because these exact contextual requirements remain unknown, merely analyzing the gene sequence can only identify potential poly-A signals, but cannot reliably predict whether they will actually be functional in practice.

“The potential causes of low steady state levels of mRNA due to the nature of the coding sequence are many. First, full length RNA synthesis might not occur at a high frequency. This could, for example, be caused by the premature termination of RNA during transcription or due to unexpected mRNA processing during transcription. Second, full length RNA could be produced but then processed (splicing, polyA addition) in the nucleus in a fashion that creates a nonfunctional mRNA. If the RNA is properly synthesized, terminated and polyadenylated, it then can move to the cytoplasm for translation. In the cytoplasm, mRNAs have distinct half lives that are determined by their sequences and by the cell type in which they are expressed. Some RNAs are very short-lived and some are much more long-lived. In addition, there is an effect, whose magnitude is uncertain, of translational efficiency on mRNA half-life. In addition, every RNA molecule folds into a particular structure, or perhaps family of structures, which is determined by its sequence. The particular structure of any RNA might lead to greater or lesser stability in the cytoplasm. Structure per se is probably also a determinant of mRNA processing in the nucleus. Unfortunately, it is impossible to predict, and nearly impossible to determine, the structure of any RNA (except for tRNA) in vitro or in vivo. However, it is likely that dramatically changing the sequence of an RNA will have a large effect on its folded structure. It is likely that structure per se or particular structural features also have a role in determining RNA stability.“

To put it more simply, the mechanisms discussed here mean: The concentration of mRNA in the cell is influenced by many factors, including inefficient synthesis, premature termination of transcription, or faulty processing in the cell nucleus. Even if an RNA molecule is fully synthesized, incorrect processing (such as during splicing or poly-A addition) can result in non-functional mRNA.

After being transported into the cytoplasm, the lifespan (half-life) of an mRNA depends significantly on its specific sequence, with translational efficiency also playing a role. Every RNA adopts a specific spatial folding determined by its base sequence. This structure is crucial for how stable the mRNA is and how it is processed. It is scientifically extremely difficult, if not impossible, to predict the structure of an RNA or to determine it precisely through experimentation.

Since the structure depends on the sequence, a drastic change in the sequence inevitably leads to a change in the RNA structure, which in turn directly affects the stability and function of the molecule.

Here we encounter a real phenomenon, known in biology as co-translational protein folding:3^,4 Even if the amino acid sequence (the blueprint) remains exactly the same, a significantly altered RNA structure, for example due to the mentioned codon optimizations, can lead to the ribosome reading the mRNA at a different rhythm. If the ribosome suddenly works faster or slower at altered positions, this can indeed influence how the amino acid chain folds into its 3D structure while it is still being produced. Theoretically, an altered RNA structure can lead to minutely different protein folds, even if the actual code is correct.

Is this really still theory? The answer should be known to everyone.5

Mulroney et al. showed that the ribosome stalls: The modification causes the ribosome to slow down or pause at certain positions (ribosomal stalling). “…occurring at ribosome slippery sequences.”

This paper is quite tricky, and I’m not the best at describing and identifying sequencing methods and the resulting isolation experiments when there are too many sequence names involved. So, I spent half the day struggling through the methodology section and arguing tooth and nail with various AIs. I’m trying to break the steps down one by one to make it bite-sized and easier to grasp what the authors actually did:

“Fluc+1FS template DNA was produced by overlap extension PCR of pUCK100Fluc [...] Fluc+1FS mRNA was produced from Fluc+1FS template DNA [...]”

The DNA template for the +1 frameshift reporter gene (Fluc+1FS) was generated via overlap extension PCR based on the firefly luciferase plasmid (pUCK100Fluc). This template is used for the in vitro transcription of the reporter RNA, which was specifically designed to make frameshifting events measurable.

The researchers artificially inserted a “slippery site” into the gene here in order to trigger and detect the error in the first place.

“UTP and CTP were substituted where required for 5-methoxyUTP, N1-methylpseudoUTP or 5-methylCTP.”

Where required, uridine triphosphates (UTP) and cytidine triphosphates (CTP) were substituted for 5-methoxy-UTP, N1-methylpseudouridine triphosphate (N1-methylpseudoUTP), or 5-methyl-CTP.

This is the methodological act in which the modification is introduced as an isolated variable into the same RNA backbone to see whether the chemistry alone triggers the error.

“The gel was stained with Coomassie dye and the region between about 60 kDa and 75 kDa [...] was excised and processed for mass spectrometry analysis [...] Peptides were assigned to their respective reading frame or junction by inspection.”

The gel was stained with Coomassie dye; the region between approx. 60 kDa and 75 kDa (corresponding to the expected size of the frameshifted protein) was excised and prepared for mass spectrometry analysis. The peptides were assigned to their respective reading frames or the transition point (junction) by manual inspection.

The protein created by the ribosome in the incorrect (+1) frame is thus physically isolated from the gel. The MS then identifies the exact sequence position where the shift from the 0-frame to the +1-frame occurred.

“A206G, U187C and U208C mRNA were transcribed from custom genes subcloned into pUC57T7 [...] U187C/U208C template DNA was produced by overlap extension PCR and reinsertion into pUC57.”

The mRNAs for the A206G, U187C, and U208C mutants were transcribed from custom gene constructs sub-cloned into the pUC57T7 plasmid. The U187C/U208C DNA template was generated via overlap extension PCR and re-inserted into pUC57.

In this step, the “slippery sequence” itself was altered. This is the point where the researchers manipulated the RNA structure to see if the ribosome slips less when the modification remains the same.”

And this is precisely where my inner skeptic kicks in: Yet, following the patent of Monsanto, the title of the paper “N1-methylpseudouridylation of mRNA causes +1 ribosomal frameshifting” by Mulrony et al. is already not entirely correct as it stands: because, this is not be attributable exclusively to the N1-methylpseudouridine substitution of uracil, but rather to the intrinsic structural alterations via codon modifications and, above all, to the modification of the 3’ UTR and the poly-A tail. And the authors demonstrate it by themself in Step 4. We know that codon modifications with N1-pseudo-U can certainly have physical consequences, such as altered melting temperatures.6 Okay. But would that alone be enough to completely overwhelm the ribosomes and force a full frameshift? As the authors' own data indicate, frameshifting events were minimized via codon modifications that altered the overall RNA structure, confirming its essential role.

One could, of course, argue: “But the T cells in vivo also responded to the +1 frameshifted peptides. Therefore, N1-methylpseudouridine must be causal.”

The researchers indeed investigated whether T cells from vaccinated individuals responded to +1 frameshift peptides. This was observed in individuals vaccinated with BNT162b2, but not in those vaccinated with ChAdOx1 (AstraZeneca).

However, this actually supports my argument: BNT162b2 already contains both codon optimization and N1-methylpseudouridine. If this complete package is administered and frameshift-derived peptides are detected, this only demonstrates that the final vaccine product can give rise to such frameshift products. It does not prove that this effect is caused exclusively by the chemical properties of N1-methylpseudouridine, because the structural variable (namely codon optimization) is an inherent and inseparable component of the BNT162b2 vaccine design.

Furthermore, the comparison with ChAdOx1 is not appropriate, as it involves an entirely different cellular and operational context that is not directly comparable. Differences in vaccine platform, antigen expression dynamics, intracellular processing, and immune presentation mechanisms introduce additional variables that make it impossible to attribute the observed differences specifically to N1-methylpseudouridine. Therefore, the absence of a comparable T-cell response after ChAdOx1 vaccination does not isolate the effect of the nucleotide modification itself, but rather reflects the broader differences between the two vaccine platforms.

However, Monsanto’s patent suggests a deeper problem within the RNA structure. And this brings us to the perennial issue that seems to have gone virtually unnoticed until today: Same codon, different letters?

Plotkin and Kudla7 show that nature exploits the redundancy of the genetic code to embed a second, superimposed layer of information into the mRNA: the physical folding code. The three-dimensional structure of the RNA is thus a direct mediator between codon selection and the actual protein quantity or quality:

”Nonetheless, in most sequenced genomes, synonymous codons are not used in equal frequencies. This phenomenon, termed codon usage bias (Figure 1), is now recognized as critical in shaping gene expression and cellular function, through its effects on diverse processes ranging from RNA processing to protein translation and protein folding.”

Bezerra et al.8 demonstrate that the “language” of the cell can not only be read but also rewritten. For the authors, RNA functions as a programming code that provides the cell with “instructions” for the synthesis of proteins with specific chemical or structural properties. They describe how codon alterations modify the entire ribosomal machinery via tRNA synthetases:

”Most codon reassignments have been linked to alterations in components of the translational machinery, namely tRNAs, aminoacyl-tRNA synthetases and the release factors that recognize stop codons [23].”

Furthermore, Liu et al.9 demonstrate that so-called “silent” mutations play a decisive functional role. Here, RNA acts not only as a mere blueprint but as a kinetic pacemaker for protein folding. The authors describe how mRNA secondary structure dynamically regulates translation: synonymous codon alterations modify the local RNA structure, which in turn serves as a physical brake or accelerator for the ribosome, thereby precisely controlling the timing of co-translational protein folding:

”It is clear that consequences of codon usage have impact at multiple steps in gene expression - including mRNP formation, mRNA structure, translational rate, protein folding, and function. It is also evident that the combination of codon usage, availability of cognate tRNAs, the ribosome unwinding mRNA, the number and length of pauses, and the overall processivity rate during translation all contribute to protein folding. Consequently, both cotranslational and posttranslational protein conformation appear to be strongly influenced by the dynamics of translation as key determinants of final folding state. At a fundamental level, these tasks are governed by mRNAs and gene-specific distribution of synonymous codons.”

The paper thereby shows that synonymous codons are key determinants of whether a protein achieves its correct functional state or undergoes misfoldings.

Therefore, as long as the entire structure of modRNA (N1-methylpseudouridine-modified versus unmodified but both sequences codon-optimized) has not been directly investigated using the same codon sequence, skepticism regarding causality is more than justified:

As proposed by Giedroc et al.10, among others11^,12 , even RNA structural changes can lead to frameshifts.

“This interaction is isosteric with the adenosine N1-2′-OH interaction found in the BWYV (Su et al., 1999), PLRV (Pallan et al., 2005) and PEMV-1 RNAs (cf. Fig. 9); however, the ScYLV and BWYV mRNA structures differ in their detailed L2–S1 hydrogen bonding and L2 stacking interactions (Fig. 10D–G).”

As Gaither et al.13 note:

“These variants modify the codon in a transcript but leave the protein unchanged, and for years were erroneously considered to be “silent.” However, the past 2 decades have seen a growing understanding that synonymous codons serve vital regulatory functions [9–12].”

Furthermore the authors summarize:

“Ribosomal collisions essentially end the RNA’s life, activating the no-go decay (NGD) pathway, and are also known to cause frame-shifts [26, 30, 92, 93]. In support of all these hypotheses, we note that the majority of our observed constraint is to preserve stability.”

Taken together, while it is plausible that N1-methylpseudouridine promotes frameshifting, it remains unclear whether this is due solely to its chemical properties or to the overall structural changes in the modRNA itself. This question has not been sufficiently resolved to classify it as an isolated causal driver.

The central argument of this article is not that N1-methylpseudouridine (N1Ψ) has no role in promoting ribosomal frameshifting. Rather, it questions whether the available experimental evidence is sufficient to classify N1Ψ as an isolated causal driver of the observed effect.

Mulroney et al. demonstrate that N1Ψ-containing mRNA can promote ⁺1 ribosomal frameshifting and provide evidence that frameshift-derived peptides are generated from the BNT162b2 mRNA platform. However, the study also shows that synonymous sequence modifications of ribosomal slippery sites substantially reduce frameshifting while the N1Ψ chemistry remains unchanged. This indicates that sequence context and RNA structure are necessary contributors to the phenomenon.

The broader literature on codon usage, RNA folding, translational kinetics, and co-translational protein folding further demonstrates that synonymous codon substitutions are not biologically neutral. They can alter RNA secondary structure, ribosome pausing, translational efficiency, and protein folding dynamics. Consequently, sequence optimization itself represents a biologically relevant variable rather than a passive background condition.

This raises an unresolved question: would comparable translational phenomena emerge from the codon-optimized sequence itself, even in the absence of N1Ψ modification? Given that synonymous codon substitutions are known to influence translation kinetics, RNA structure, and co-translational folding, the independent contribution of sequence optimization remains difficult to assess without a factorial experimental design that separates nucleotide chemistry from sequence-dependent structural effects.

The key methodological concern is that, at the level of the full-length Spike construct, the study does not employ a complete factorial design capable of independently separating nucleotide chemistry (N1Ψ versus unmodified uridine) from sequence- and structure-related variables. As a result, the relative contribution of nucleotide modification versus RNA structural context cannot be fully resolved from the available data.

Therefore, while it seems plausible that N1Ψ contributes to frameshifting, it remains unclear to what extent the observed effect reflects the chemical properties of the modification itself, its impact on RNA structural organization, or an interaction between both. The current evidence supports an interaction model involving nucleotide modification and sequence-dependent structural features, rather than conclusively establishing N1Ψ as a standalone causal factor.14

McRae et al. initially reported these findings as a Research Square preprint, which was subsequently published as a peer-reviewed study. In the context of the present discussion, the peer-reviewed version provides the most appropriate reference point, as it incorporates the additional scrutiny and validation associated with the publication process. This further supports the interpretation that the observed effects likely arise from an interplay between N1Ψ modification and sequence-dependent RNA structural features, rather than demonstrating N1Ψ as an isolated causal determinant.