WO2023186932A1 - Methods of rna stabilisation - Google Patents

Abstract

This invention relates to the production of a messenger RNA (mRNA) molecule with increased stability that comprise attaching a poly(A) tail comprising one or more N6-methyladenosine (m6A) residues to an RNA molecule comprising a coding sequence to produce an mRNA molecule. mRNA molecules and methods for their production and use are provided.

Description

Methods of RNA Stabilisation

Field

The present invention relates to methods and compounds for increasing the stability of RNA in vivo.

Background

Trypanosoma brucei (T. brucei) is a protozoan unicellular parasite that causes lethal diseases in sub- Saharan Africa: sleeping sickness in humans and² nagana in cattle⁴. The infection can last several months or years mostly because T. brucei escapes the immune system by periodically changing its variant surface glycoprotein (VSG)². The T. brucei genome contains around 2000 antigenically distinct VSG genes⁵, but only one VSG gene is actively transcribed at a given time. Transcriptionally silent VSG genes are switched on by homologous recombination into the Bloodstream Expression Site (BES) or by transcriptional activation of a new BES2, resulting in parasites covered by ~10 million identical copies of the VSG protein⁶.

VSG is essential for the survival of bloodstream form parasites. VSG is not only one of the most abundant proteins in T. brucei, but it is also the most abundant messenger RNA (mRNA) in bloodstream forms (4-11 % of total mRNA)⁷⁸. VSG mRNA abundance is a consequence of its unusual transcription by RNA polymerase I and its prolonged stability⁹. The half-life of VSG mRNA has been estimated to range from 90-270 min, contrasting with the 12 min, on average, for other transcripts¹⁰. The basis for its unusually high stability is not known. It is thought to derive from the VSG 3’untranslated region (UTR), which contains two conserved motifs, a 9-mer and a 16-mer motif (usually called 16-mer, but the first and last position are less conserved, the conserved core is a 14-mer), found immediately upstream of the poly(A) tail⁵’¹¹. Mutational studies have shown that the 16-mer conserved motif is essential for VSG mRNA high abundance and stability¹², even though its underlying mechanism is unknown.

VSG expression is highly regulated when the bloodstream form parasites undergo differentiation to the procyclic forms that proliferate in the insect vector¹³. The BES becomes transcriptionally silenced and VSG mRNA becomes unstable¹⁴, which results in rapid loss of VSG mRNA and replacement of the VSG coat protein by other surface proteins (reviewed in¹⁵). The mechanism by which VSG mRNA becomes unstable during differentiation remain unknown. The surface changes are accompanied by additional metabolic and morphological adaptations, which allow procyclic forms to survive in a different environment in the insect host¹⁵.

RNA modifications have been recently identified as important means of regulating gene expression. The most abundant internal modified nucleotide in eukaryotic mRNA is N6-methyladenosine (m⁶A)¹⁶'¹⁷, which is widespread across the human and mouse transcriptomes and is often found near stop codons and the 3’UTR of the mRNA encoded by multiple genes^{18 19}. In these organisms, m⁶A is synthesized by a methyltransferase complex whose catalytic subunit, METTL3, methylates adenosine in a specific consensus motif. Demethylases responsible for removing m⁶A from mRNA have also been identified²⁰’²¹. m^sA affects several aspects of RNA biology, for instance contributing to mRNA stability, mRNA translation, or affecting alternative polyadenylation site selection (reviewed in¹). Summary

The present inventors have found that the presence of N^B-methyladenosine (m⁶A) residues in the poly(A) tail increases the stability of mRNA molecules and extends their half-life in vivo. This may be useful in increasing protein expression in vivo, for example for improving the efficacy of RNA therapeutics.

A first aspect of the invention provides a method of producing a messenger RNA (mRNA) molecule comprising providing an RNA molecule comprising a coding sequence, attaching a poly(A) tail comprising one or more N⁶-methyladenosine (m⁶A) residues to the RNA molecule to produce an mRNA molecule.

A second aspect of the invention provides a method of stabilising a messenger RNA (mRNA) molecule in a mammalian cell comprising attaching a poly(A) tail comprising one or more m⁶A residues to the mRNA molecule, wherein the m^sA residues in the poly(A) tail stabilises the mRNA molecule

A third aspect of the invention provides an mRNA molecule comprising a poly(A) tail comprising one or more m⁶A residues.

A fourth aspect of the invention provides a pharmaceutical composition comprising an mRNA molecule of the third aspect.

A fifth aspect of the invention provides a kit for use in a method of the first or second aspect comprising a poly(A) tail comprising one or more N⁶-methyladenosine (m⁶A) residues and an RNA ligase.

A sixth aspect of the invention provides a kit for use in a method of the first or second aspect comprising adenosine, N⁶-methyladenosine (m⁶A) and a poly(A) polymerase.

Other aspects and embodiments of the invention are described in more detail below.

Brief Description of the Figures

Figure 1 shows that m⁶A is present in the poly(A) tail of VSG mRNA and other transcripts, a, Overlap chromatogram of nucleoside modifications detected in mRNA mammalian BSF by LC-MS/MS. Data are ratios between peak areas, b, Enrichment of nucleoside modifications in mRNA relative to total RNA. Two- way ANOVA with sidak correction for multiple test (”** m6A, m6,6A ,m7G and ml A P<0.0001 ). N = 5 biological samples, c, m^sA levels quantified using standard curve. Bar represents mean, n = 3 or 4 biological replicates. Unpaired two tailed t-test: mammalian bloodstream or insect total RNA vs mRNA P<0.0001 ; mammalian bloodstream mRNA vs Insect procyclic mRNA P= 0.4162. d, Scatter plot of m6A enrichment relative to average transcript expression, expressed as Iog2 counts per million reads mapped (GPM). Transcripts enriched or depleted in m⁶A IP sample relative to Input sample are indicated in red or blue, respectively. Moderated t-test adjusted with Benjamin Hochberg false discover rate. Triangles represent VSGs. N = 3 independent IPs. e, Gene set enrichment analysis. Line indicates the enrichment score distribution across VSG genes, ranked according to the Iog2 fold change between m⁶A-IP and input samples, f, Schematics of oligonucleotides used in RNase H digestion of VSG mRNA and expected digestion products (g). SL: spliced leader; dT: poly deoxi-thymidines, g, m⁶A immunoblotting of mammalian bloodstream forms total RNA digested with RNase H after pre-incubation with indicated oligonucleotides. Methylene Blue stains rRNA. Tub: 3-Tubulin. n = 2 independent experiments, h, Mass-spectrometry analysis of total RNA digested independently with enzymes RNase T1 and RNase A. Total RNA was extracted from Trypanosoma brucei (BSF, n = 3; PCF, n = 2), Trypanosoma congolense (n = 2), Trypanosoma cruzi (n = 1 ), Leishmania infantum (n = 2) and human cells (HEK293T, n = 1 ).

Figure 2 shows that m⁶A is removed from VSG mRNA prior to its degradation, a, Schematics of VSG mRNA transcript and analyses described in this figure, b, VSG transcript levels (RT-qPCR, pink), m⁶A levels (immunoblotting, light blue) and length of poly(A) tail (PAT assay, dark blue) after transcription halt by actinomycinD (ActD). Data are mean ± s.d. Two-way ANOVA with sidak correction for multiple test. Black asterisks denote significance between mRNA and m^sA. Grey asterisks denote significance between poly(A) tail and m⁶A (****P<0.0001 , *P=0.0104 in mRNA vs m6A in 15 min, *P=0.0224 in mRNA vs m⁶A in 30 min, *P=0.0169 in poly(A) vs m6A in 30 min), n = 3 transcription inhibition experiments, c, Northern blotting of VSG decay from parasites treated with ActD. Total RNA was incubated with an oligonucleotide located 368 nt upstream of VSG poly(A) tail and digested with RNaseH. Probe hybridizes with conserved 16-mer motif. AO is the VSG 3’end fragment in which the poly(A) tail was removed by oligo dT-RNase H digestion. Methylene Blue stains rRNA. n = 3 transcription inhibition experiments, d, m⁶A immunoblotting of bloodstream form total RNA extracted from parasites treated with ActD (c). Methylene Blue stains rRNA. e, VSG transcript levels and m⁶A levels during parasite differentiation from bloodstream to procyclic forms. Total RNA was extracted at different time points after inducing differentiation with cis-aconitate. Data are mean ± s.d. Two-way ANOVA with sidak correction for multiple test (****P<0.0001 ). n = 3 parasite differentiation experiments, f, m6A immunoblotting of parasites differentiating to procyclic forms (e). Methylene blue stains rRNA.

Figure 3 shows that the inclusion of m6A in the VSG poly(A) tail depends of de novo transcription, a, Parasites were treated with cis-aconitate (CA), and after washing away compound, parasites were placed in culture in 3 different conditions. Labels 1-5 indicate the conditions at which parasites were collected for immunoblotting analysis (Panel b). b, m⁶A immunoblot at each of the 5 conditions (Panel a), n = 3 independent experiments, c, Quantification of immunoblotting in (a). Two-way ANOVA with sidak correction for multiple test. (****P<0.0001 . Data are mean ± s.d. Black asterisks refer to condition 3, grey asterisks to condition 5. d, VSG mRNA levels measured by RT-qPCR. Two-way ANOVA with sidak correction for multiple test. (****P<0.0001 ). Data are mean ± s.d. n = 3 independent experiments, e, m^sA immunofluorescence analysis. Parasites were treated with Nuclease P1 (NP1 ) or ActD. Nuclei were stained with Hoechst. Arrows points to weak m⁶A signal, f, Proportion of m6A signal in nucleus and cytoplasm, n = 4 experiments with 125 parasites in each. Data are mean ± s.e.m. g, m6A levels expressed as mean fluorescence intensity (MFI). Unpaired two tailed t-test (****P<0.0001 , *** P= 0.001 ). Data are mean ± s.d. n = 5 independent experiments, h, RNA-FISH analysis of VSG2 transcripts. Three representative cells are shown, i, Proportion of VSG mRNA signal in nucleus and cytoplasm (h). Data are mean ± s.e.m. n = 5 independent experiments with 34 parasites in each j, m⁶A Dot-Blot of subcellular fractions. Quantity of spotted RNA is indicated, n = 3 fractionation experiments, k, Quantification of dot-blot m⁶A signal (j). Unpaired two tailed t-test P= 0.8753. Data are median. Scale bars, 4pm; DIC, differential interference contrast. Figure 4 shows that the conserved VSG 16-mer motif is required for inclusion of m⁶A in adjacent poly(A) tail, a, Schematics of VSG double-expressor (DE) cell-lines. VSG117 was inserted in the active bloodstream expression site, which contains VSG2 at the telomeric end. In DE1 , VSG117 contains its endogenous 3'UTR with the conserved 16-mer motif (sequence in blue). In DE2, the 16-mer motif of VSG117 was scrambled (sequence in orange), b, Transcript levels of VSG1 17 and VSG2 transcripts (RT-qPCR), normalized to transcript levels in cell-lines expressing only VSG2. One-way ANOVA with sidak correction for multiple test. (P=0.7612 for VSG2 wt vs VSG117 wt in DE1 , ****P<0.0001 for VSG2 wt vs VSG117 mut in DE2). n = 3 independent clones, c, m6A immunoblot of mRNA from DE1 and DE2 cell-lines. RNase H digestion of VSG2 mRNA was used to resolve VSG2 and VSG117 transcripts. 50ng and 12.5ng of DE1 were loaded in two separate lanes, n = 3 independent clones, d, m6A index calculated as the ratio of m6A intensity and mRNA levels, measured in Panels c and b, respectively, und., undetectable. # intensities measured in lane 3 of Panel c. e, m6A enrichment in VSG genes. m6A-RIP sequencing data was used to calculate, for each VSG gene, the ratio between the number aligned reads in IP versus Input samples. Only VSG transcripts detected in IP sample were used for this analysis. Blue and orange indicate the presence or absence, respectively, of the 16-mer motif in the 3’UTR. Unpaired two-sided Mann-Whitney test (P<0,0001 ,****). f, Scatter plot of m⁶A-RIP enrichment relative to transcript levels of detectable VSG transcripts. Colour code identical to panel e. Dashed lines represents log2FC=1 and log2FC=-1 . Spearman correlation between the data was R=-0.35. n = 3 for input samples and m6A immunoprecipitated samples.

Figure 5 shows that VSG 16-mer motif inhibits CAF1 and poly(A) tail deadenylation, a, The length of the VSG poly(A) tail was measured using Poly(A) tailing (PAT) assay after transcription halt by ActD. WT and Mut-16-mer cell-lines were compared, b, VSG117 transcript levels (measured by RT-qPCR, pink) and length of poly(A) tail after transcription halt by ActD. Values were normalized to 0 hour. Two-way ANOVA with sidak correction for multiple test. Black asterisks refer to mRNA, grey asterisks refer to poly(A) tail ****P<0.0001 . *** P=0.0002 in VSG117 wt poly(A) tail vs VSG117 mut poly(A) tail in 15 min). Data are mean ± s.d. n = 3 transcription inhibition experiments, c, Length of VSG poly(A) tail upon CAF1 downregulation and after transcription halt by ActD. Poly(A) length was measured by PAT assay. Two-way ANOVA with sidak correction for multiple test. (****P<0.0001 ). Data are mean ± s.d. n = 3 transcription inhibition experiments, d, VSG transcript levels upon CAF1 downregulation and after transcription halt by ActD. Significance was measured by two-way ANOVA with sidak correction for multiple test. (*P=0.0191 ). Data are mean ± s.d. n = 3 transcription inhibition experiments, e, RNA-FISH analysis of VSG8 of 4 indicated conditions. DIC, differential interference contrast. Scale represents 4um. f, VSG8 transcript levels expressed as mean fluorescence intensity (MFI) levels of FISH signal. The proportion of nuclear and cytoplasmic staining was calculated as described in Fig. 3. Data are mean ± s.d. Unpaired two-sided t-test (p-value <0,0001 ,”**). n = 3 biological replicates, 100 cells per replicate.

Figure 6 shows the experimental setup used to test in HeLa cells the effect of m⁶A in the poly(A) tail of in vitro transcribed RNAs.

Figure 7 shows the four steps required to generate in vitro methylated mRNAs. Figure 8 shows a quantification of reporter mRNA levels in HeLa cells. Transcript levels were quantified by qPCR and normalized to beta-Tubulin gene, 24 hours post-transfection. Different amounts of GFP mRNA were transfected into HeLa cells (0, 100, 500 and 2500ng). One sample received 100 ng of mRNA without the transfection reagent, preventing mRNA from being internalized by the cell, allowing us to confirm that no external RNAs remained after washing the cells. In the final sample, 500ng of RNA was not treated with Antarctic phosphatase, which results in cellular stress due to residual NTPs and thus reduced eGFP mRNA levels.

Figure 9 shows a quantification of decay of reporter mRNA in HeLa cells. T ranscript levels were quantified by qPCR at multiple times after transfection and normalized to beta-Tubulin gene and time point of Ohr (moment of transfection). Data for eGFP mRNA with methylated poly(A) tail is displayed in blue and nonmethylated in orange.

Detailed Description

This invention relates to methods for increasing the stability of a messenger RNA (mRNA) molecule in a mammalian cell by attaching to the mRNA molecule a poly(A) tail comprising one or more N⁶- methyladenosine (m⁶A) residues.

The presence of one or more m⁶A residues in the poly(A) tail of an mRNA molecule is shown herein to increase the stability of the mRNA molecule. For example, the presence of one or more m⁶A residues in the poly(A) tail may increase the in vivo half-life of the mRNA molecule and reduce the in vivo deadenylation and degradation of the mRNA molecule. An mRNA produced by a method described herein with one or more m⁶A residues in its poly(A) tail may have improved stability relative to a control mRNA without m⁶A residues in the poly(A) tail. An mRNA produced by a method described herein with one or more m⁶A residues in its poly(A) tail may exhibit a biphasic decay curve (for example, indicating that mRNA degradation occurs in two independent steps).

A messenger RNA (mRNA) molecule is a single-stranded RNA that comprises a nucleotide sequence that encodes a protein (a coding sequence or CDS). The coding sequence of the mRNA is translated in the cytoplasm of a cell by ribosomes to produce the encoded protein. An mRNA molecule may further comprise a 5’ cap and a 5’ untranslated region (UTR) located upstream of the coding sequence and a 3’UTR and a poly(A) tail located downstream of the coding sequence.

An RNA molecule as described herein may be a precursor to an mRNA molecule and may be converted into a mature mRNA molecule by a method described herein. For example, an RNA molecule may lack a poly(A) tail but may comprise one or more other elements of an mRNA molecule, such as a 5’ cap, a 5’UTR, coding sequence and 3’UTR. In some embodiments, the RNA molecule may comprise a 5’ cap, 5’UTR, coding sequence and 3'UTR. An mRNA molecule may be produced by adding a poly(A) tail to the 3' end of the RNA molecule. In other embodiments, the RNA molecule may comprise a 5’UTR, coding sequence and 3'UTR.

An mRNA molecule may be produced by adding a 5’ cap to the 5’ end of the RNA molecule and a poly(A) tail to the 3’ end of the RNA molecule. The 5’ cap and the poly(A) tail may be added simultaneously or sequentially in any order. A poly(A) tail is a sequence consisting of adenosine monophosphate bases that is located at the 3’ end of an mRNA molecule. The poly(A) tail contributes to the nuclear export, translation and stability of mRNA in eukaryotic cells.

A poly(A) tail may be 20 to 300 bases in length, for example, 75 to 125 bases in length.

1 or more, 2 or more, 3 or more, 4 or more, 5 or more, 10 or more, 20 or more, 30 or more, 40 or more, 50 or more, 60 or more 70 or more, 80 or more, 90 or more 100 or more, 150 or more, or 250 or more of the adenosine bases in the poly(A) tail may be methylated. For example, they may be N⁶-methyladenosine (m6A) bases.

1% or more, 2% or more, 3% or more, 4% or more 5% or more, 10% or more, 20% or more, 30% or more, 40% or more or 50% or more of the adenosine bases in the poly(A) tail may be methylated. For example, they may be N⁶-methyladenosine (m6A) bases.

The m6A residues may be distributed randomly or non-randomly in the poly(A) tail.

The coding sequence of the mRNA molecule may encode a peptide or protein. The peptide or protein may be transiently expressed in the cytoplasm of a cell, following transfection of the cell with the mRNA molecule. Suitable coding sequences for any peptide or protein of interest are well-known in the art. For example, the coding sequence of an mRNA may encode an antigen, such as a cancer or pathogenic antigen, an antibody molecule or a therapeutic protein, such as a cytokine, angiogenic factor or antibody molecule. The coding sequence of an mRNA may encode a metabolic protein (for example, insulin or glucagon-like peptide-1 ) or a peptide toxin (for example, Botulinum toxin).

In some embodiments, the coding sequence of the mRNA molecule may encode multiple peptides or proteins. The sequences encoding each peptide or protein may be separated by a self-cleaving peptide coding sequence, such as a 2A peptide coding sequence. The self-cleaving peptide causes cleavage of the nascent peptide chain during translation and separates each individual peptide or protein. Suitable 2A peptides may include T2A, P2A, E2A and F2A peptides (Poddar et al (2018) supra; Kim et al (2011 ) PLoS ONE 6, e18556). In some embodiments, multiple proteins may be expressed from a single open reading frame (ORF).

A 5’ cap is a modified base that is located at the 5’ end of the mRNA. The 5’ cap promotes nuclear export and translation and inhibits degradation by exonucleases. The 5’ cap may be N7-Methyl-3'-O- Methylguanosine, N7-Methylguanosine or an analogue thereof, such as anti-reverse cap analogue (ARCA), 3'-O-Me-m7G(5')ppp(5')G, (m7G(5')ppp(5')G) or (m7G(5')ppp(5')Gm). The 5’ untranslated region (UTR) is the region of the mRNA molecule upstream (5’) of the translation initiation codon of the coding sequence. Suitable 5’UTR sequences are well known in the art. The 3’ untranslated region (UTR) is the region of the mRNA molecule downstream (3’) of the translation termination codon of the coding sequence. Suitable 3’UTR sequences are well known in the art. In some embodiments, the UTR is a housekeeping gene UTR (for example, to facilitate the targetting of multiple different cell types). Alternatively, the UTR may be selected in order to target specific cell types. In some embodiments, 3’ UTR may be a 3’UTR with a Trypanosome 16-mer motif.

In some embodiments, one or more naturally occurring ribonucleotide bases of an RNA molecule or mRNA molecule described herein may be replaced by a modified ribonucleotide bases. Modified ribonucleotide bases include pseudouridine (4^J), N1-methylpseudouridine (ml ^P), 5-methylcytidine (m5C), 5- hydroxymethylcytosine (5hmC), 5-methyluridine (m5U) and 2-thiouridine (s2U). An RNA molecule or mRNA molecule may comprise one or more modified ribonucleotide bases. Further modified ribonucleotide bases are well known in the art. Suitable techniques for the production of RNA molecules containing modified ribonucleotide bases are well known in the art. In some embodiments, a poly(A) tail may comprise uracil (U) and/or guanine (G) ribonucleotide bases.

A method of producing a messenger RNA (mRNA) molecule as described herein may comprise attaching a poly(A) tail comprising one or more N⁶-methyladenosine (m6A) residues to an RNA molecule to produce a polyadenylated mRNA molecule.

The RNA molecule may be produced by in vitro transcription of a DNA template molecule. A DNA template molecule comprising a coding sequence operably linked to an RNA polymerase promoter may be contacted with an RNA polymerase, such that the DNA template molecule is transcribed by the RNA polymerase to produce an RNA comprising the coding sequence

The DNA template molecule may for example be a plasmid or a PCR product. In some embodiments the DNA template molecule may be a synthetic DNA molecule. The DNA template molecule may be a DNA primer. The DNA template molecule may comprise a transcriptional promoter.

The coding sequence of the DNA molecule may be codon optimised for human expression. The DNA template molecule may be contacted with the RNA polymerase under conditions suitable for in vitro transcription. The RNA polymerase may be T7 RNA polymerase. The RNA polymerase may be aT3 RNA polymerase. The RNA polymerase may be an SP6 RNA polymerase. In some embodiments, the RNA polymerase may be a polymerase as described by Dousis, A et al. Nat Biotechnol (2022).

Suitable methods and reagents for in vitro transcription are well established in the art. For example, in vitro transcription may be performed using a megaMEGAscript T7 kit (AM1333); a Vaccinia Capping System (NEB M2080S); a TrasnLink (CleanCap® Reagent AG) system; an E. coli Poly(A) Polymerase (NEB M0276S) and/or using CutSmart® Buffer (NEB).

Following transcription, the transcribed RNA molecule may be isolated and/or purified. For example, the RNA molecule may be isolated by gel extraction, cellulose purification, Tangential Flow Filtration (TFF), using RNase III or by HPLC.

Following transcription, the 5’ end of the RNA molecule may be capped. For example, a modified ribonucleotide, such as N7-Methyl-3'-O-Methylguanosine or N7-Methylguanosine may be added to the 5’ end of the RNA molecule. Suitable methods and reagents for the 5’ capping of RNA molecules are well established in the art.

The 5’ end of the RNA molecule may be capped before, simultaneously or after the attachment of the poly(A) tail.

Following capping, the capped RNA molecule may be isolated and/or purified. For example, the capped RNA molecule may be isolated by column purification or ethanol precipitation. Suitable methods for isolation and/or purification are well known in the art.

Following transcription, a poly(A) tail may be attached to the 3’ end of the RNA molecule. The poly(A) tail may be attached before, simultaneously or after the capping of the 5’ end of the RNA molecule.

In some embodiments, the poly(A) tail may be attached to the RNA molecule using a poly(A) polymerase. For example, the RNA molecule may be contacted with a poly(A) polymerase in the presence of m⁶ATP. Polymerisation by the polymerase may generate a poly(A) tail at the 3’ end of the RNA molecule that contains one or more m⁶A bases.

Suitable poly(A) polymerases and methods for their use in generating poly(A) tails are well known in the art. For example, suitable poly(A) polymerase include an E. coli poly(A) polymerase (NEB M0276S).

The RNA molecule may be contacted with the poly(A) polymerase in the presence of a mixture of A and m⁶A. The concentration of m⁶A in the mixture relative to A ([m⁶A]/[A]) may be 0.1 to 10, for example, 0.1 , 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1 , 2, 3, 4, 5, 6, 7, 8, 9 or 10. For example, the proportion of m⁶A in the mixture relative to A may be 5: 95, 10:90, 20:80, 30: 70. 40:60, 50: 50, 60: 40, 70: 30, 80: 20, 90: 10 or 95: 5. Polymerisation by the polymerase in the presence of the mixture may generate a poly(A) tail at the 3’ end of the RNA molecule that contains A bases and one or more m⁶A bases. Suitable reaction reagents and conditions are well known in the art.

In other embodiments, the poly(A) tail may be attached to the RNA molecule using an RNA ligase. For example, the RNA molecule may be contacted with poly(A) molecule comprising one or more m⁶A bases in the presence of the RNA ligase. Ligation of the poly(A) molecule to the RNA molecule by the ligase may generate a poly(A) tail at the 3' end of the RNA molecule that contains one or more m⁶A bases i.e. the poly(A) molecule forms a poly(A) tail when ligated to the RNA molecule by the ligase.

A poly(A) molecule comprising one or more m⁶A bases may be synthesised by conventional techniques.

Suitable RNA ligases and methods for their use are well known in the art. For example, suitable RNA ligases may include T4 RNA ligase.

Following attachment of the poly(A) tail, the mRNA molecule may be isolated and/or purified. For example, the mRNA molecule may be isolated by column purification, gel extraction, or HPLC. The mRNA molecule may be isolated by cellulose purification (for example, as described by Baiersddrfer M, et al.. Mol Ther Nucleic Acids. 2019 Apr 15;15:26-35).

In some embodiments, the mRNA molecule may further be purified by treatment with a phosphatase (for example, an alkaline phosphatase) to degrade residual NTPs. Suitable phosphatases are well known in the art and include Antarctic phosphatase.

The invention also provides an mRNA molecule with a poly(A) tail comprising one or more m^sA bases. A suitable mRNA may be produced by a method described above.

In some embodiments, the mRNA molecule may be an RNA vaccine. For example, the mRNA molecule may comprise a coding sequence that encodes a pathogenic antigen. Expression of the pathogenic antigen in a cell of a host may elicit an immune response to the antigen in a host

In other embodiments, the mRNA molecule may be an RNA therapeutic. For example, the mRNA molecule may comprise a coding sequence that encodes a therapeutic protein. Expression of the therapeutic protein in a cell of a host may elicit a therapeutic effect, for example it may replaces or compensates for a defective protein in a host. Suitable therapeutic proteins may include methylmalonyl-CoA mutase, CFTR, SERPINA1 (a-1 -antitrypsin), cytokines, such as IL-12. IL-15, IL-23, IL-36, IFNa, and GM-CSF, angiogenic factors, such as VEGF, immunomodulatory proteins such as 0X40 Ligand (OX40L), and antibody molecules, such as bispecific antibodies (e.g. riboMABs™)

A mRNA molecule or lipid nanoparticle described herein may be used to transfect a mammalian cell in vitro, ex vivo or in vivo. Following transfection, the coding sequence of the mRNA molecule may be expressed in the cell.

Following production and/or purification, an mRNA molecule may be formulated into a suitable delivery vehicle. For example, the mRNA may be encapsulated into a lipid nanoparticle (LNP) or a polymer -based nanoparticle. Suitable delivery vehicles include polymers. Suitable techniques for encapsulating mRNA in LNPs are well known in the art. For example, as described by Paunovska, K., et al. Nat Rev Genet 23, 265- 280 (2022); Jayaraman M et al. Angew Chem Int Ed Engl. 2012 Aug 20;51 (34):8529-33; Maier MA, et al.. Mol Ther. 2013 Aug;21(8):1570-8 and Pardi N, et al. J Control Release. 2015 Nov 10;217:345-51.

While it is possible for a delivery vehicle, such as a lipid nanoparticle (LNP), or a mRNA molecule described herein to be used (e.g., administered) alone, it is often preferable to present it in the form of a pharmaceutical composition, which may comprise at least one component in addition to the delivery vehicle or mRNA molecule. Thus pharmaceutical compositions may comprise, in addition to the delivery vehicle or mRNA molecule, a pharmaceutically acceptable excipient, carrier, buffer, stabilizer or other materials well known to those skilled in the art.

The term “pharmaceutically acceptable,” as used herein, pertains to compounds, ingredients, materials, compositions, dosage forms, etc., which are, within the scope of sound medical judgment, suitable for use in contact with the tissues of the subject in question (e.g., human) without excessive toxicity, irritation, allergic response, or other problem or complication, commensurate with a reasonable benefit/risk ratio. Each carrier, diluent, excipient, etc. must also be “acceptable” in the sense of being compatible with the other ingredients of the formulation. Suitable carriers, excipients, etc. can be found in standard pharmaceutical texts, for example, Remington: The Science and Practice of Pharmacy , 23rd edition, Academic Press.

Pharmaceutical compositions and formulations may conveniently be presented in unit dosage form and may be prepared by any methods well known in the art of pharmacy. Such methods include the step of bringing into association the delivery vehicle or mRNA molecule with the carrier which constitutes one or more accessory ingredients. In general, the compositions are prepared by uniformly and intimately bringing into association the active compound with liquid carriers.

A pharmaceutical composition comprising a delivery vehicle or mRNA molecule may be administered alone or in combination with other treatments, either simultaneously or sequentially dependent upon the condition to be treated. For example, a pharmaceutical composition comprising an lipid nanoparticle (LNP) or mRNA molecule may be administered in combination with an adjuvant or other therapeutic agent.

A mRNA molecule or delivery vehicle may be useful in a method of treatment or therapy. For example, a method of treatment may comprise administering a mRNA molecule, lipid nanoparticle or cell to an individual in need thereof. The purpose of the method of treatment depends on the peptides or proteins encoded by the coding sequence of mRNA. Methods may for example be for the treatment of pathogenic infection, metabolic diseases, heart disease, and cancer.

Other aspects and embodiments of the invention provide the aspects and embodiments described above with the term “comprising” replaced by the term “consisting of’ and the aspects and embodiments described above with the term “comprising” replaced by the term ’’consisting essentially of”.

It is to be understood that the application discloses all combinations of any of the above aspects and embodiments described above with each other, unless the context demands otherwise. Similarly, the application discloses all combinations of the preferred and/or optional features either singly or together with any of the other aspects, unless the context demands otherwise.

Modifications of the above embodiments, further embodiments and modifications thereof will be apparent to the skilled person on reading this disclosure, and as such, these are within the scope of the present invention.

All documents and sequence database entries mentioned in this specification are incorporated herein by reference in their entirety for all purposes.

“and/or” where used herein is to be taken as specific disclosure of each of the two specified features or components with or without the other. For example “A and/or B” is to be taken as specific disclosure of each of (i) A, (ii) B and (iii) A and B, just as if each is set out individually herein.

Experimental Here we show that N6-methyladenosine is an RNA modification enriched in T. brucei mRNA. Importantly, this study revealed that m⁶A is present in mRNA poly(A) tails, and half of m⁶A is located in only one transcript (VSG mRNA). We identified a cis-acting element required for inclusion of m⁶A at the VSG poly(A) tail and, by genetically manipulating this motif, we showed that m6A blocks poly(A) deadenylation, hence promoting VSG mRNA stability. We provide the first evidence that poly(A) tails of mRNA can be methylated in eukaryotes, playing a regulatory role in the control of gene expression. m⁶A is enriched in the VSG polv(A) tail

To investigate if T. brucei RNA harbours modified nucleosides that could play a role in gene regulation, we used liquid chromatography-tandem mass spectrometry (LC-MS/MS) to detect possible modifications of RNA nucleosides in poly(A)-enriched RNA (mRNA) and total RNA (mainly composed of rRNA, tRNA and other non-coding RNAs). 34 modified nucleosides were detected: 15 were detected in mRNA and 19 were only detected in total RNA (Fig. 1a). Some of these modifications had been previously detected in T. brucei RNA including Am, which is found in the mRNA cap structure, and m3C, m5C and Gm in tRNA and rRNA22-24. Comparison of the intensities of each RNA nucleoside in mRNA vs total RNA revealed that N6- methyladenosine (m6A) is 10-fold more abundant in mRNA. 10 other modifications are equally distributed in mRNA and total RNA and 4 modifications are 30-60-fold enriched in total RNA than mRNA (Fig. 1 b). Given the importance of m6A for RNA metabolism in other eukaryotes, we focused on this specific modification in T. brucei.

We used an isotope-labelled m6A nucleoside standard to quantify m6A in poly(A)-enriched, poly(A)-depleted and total RNA fractions from two stages of the parasite life cycle, i.e., the mammalian bloodstream form (BSF) and the insect procyclic form (PCF). The chromatograms of the poly(A)-enriched fraction (mRNA) revealed a peak corresponding 282->150 mass transition and an elution time of 10 min, identical to the elution time observed in the m6A standard. The m6A peak was barely detectable in the total and poly(A)- depleted RNA, indicating that most (if not all) m6A is present in mRNA and absent from rRNA and tRNAs. Similar results were obtained in RNA fractions from the procyclic form insect stage. For both stages of the life cycle, the chromatograms of total RNA and poly(A)-depleted samples contained a peak with an identical mass transition, but an earlier elution time (6.5 min), which likely reflects N1-methyladenosine (ml A), a modification commonly found in rRNAs and tRNAs²⁵-²⁶.

The m6A standard allowed us to quantify the abundance of m6A in mRNA fractions of bloodstream and procyclic forms: m6A represents 0.06-0.14% of total adenines in mRNA (Fig. 1c). In other words, in 10,000 adenosines, 6-14 are methylated to form m6A. This proportion is lower than in mammalian cells (0.1- 0.4%¹⁶’¹⁷).

To identify the transcripts enriched in m6A, we performed m6A RNA immunoprecipitation (m6A-RIP) in nonfragmented RNA, followed by sequencing of both input and immunoprecipitated (IP) transcripts. Similar strategies have been described before to study methylation at the gene level²⁷-²⁸. Differential expression analysis showed that the Variant Surface Glycoprotein 2 (VSG2) is the most represented transcript in the immunoprecipitate sample. To calculate the enrichment of m6A per transcript, we normalized the number of reads in the IP sample by the number of reads in the Input sample. Fig. 1d shows that VSG2 is the transcript with highest average expression (Log2 CPM = 15.1 , in the x-axis) and one of the most enriched in m6A (Log2 Fold Change = 2.5, in the Y axis). 7 of the top 10 genes are VSGs or VSG-related (VR) genes. Gene Set Enrichment Analysis showed that VSG transcripts are indeed enriched in mBA (Fig. 1 e). Non- VSG transcripts can also be enriched in m6A, such as cyclophilin A and ubiquitin carrier protein (UCP) (Fig. 1 d). To confirm that VSG is the transcript harbouring most m6A and to map m6A within the VSG transcript, we performed immunoblotting with an antibody that specifically recognizes m6A (Extended Data Fig. 3a) and in which we site-selectively cleaved the VSG transcript with ribonuclease H (RNase H). Poly(A) RNA was incubated with DNA oligonucleotides that annealed at different sites along the length of the VSG transcript. RNase H digestion of the RNA:DNA hybrids result in fragments of predicted sizes (Fig. 1f). If the VSG transcript is the band with the intense m6A signal, the band detected by immunoblotting would “shift” to one or more fragments of smaller size.

In the control condition, in which RNA was pre-incubated without any oligonucleotide or with a control oligonucleotide that annealed with a-tubulin (another abundant transcript in T. brucei), we observed an m6A- positive smear, confirming that m6A is present in multiple mRNA molecules. Strikingly, however, we also observed an intense band of around 1 .8 kb, which coincides with the size of actively transcribed VSG (Fig. 1g). This prominent band corresponds to 50% of m6A signal, it is not detected in mRNA from insect stage of the parasite life cycle (in which VSG is not expressed), nor in mouse liver RNA. A band of similar mobility was also observed when we used an oligonucleotide that hybridized to the spliced leader (SL) sequence, a 39nt sequence that contains the mRNA cap and that is added to every mRNA by a trans-splicing reaction29. This indicates that mBA is neither present in the spliced leader sequence, nor in the mRNA cap structure. In contrast, when we used oligonucleotides VSG-A, VSG-B and VSG-C, which hybridized to three different unique sites in the VSG sequence, we observed that the major m6A band shifted, and in all three conditions, the 3'end fragment contained the entire m6A signal (Fig. 1g). Importantly, VSG-C oligonucleotide is adjacent to the beginning of the poly(A) tail. Thus, the 3’ fragment released upon RNase H digestion with VSG-C corresponds to the poly(A) tail of VSG mRNA. This fragment, which contains the entire mBA signal from the VSG transcript, is heterogeneous in length and shorter than 200 nt (Fig. 1g).

To further confirm that the 3’fragment released after incubation with VSG-C and RNase H corresponded to the VSG poly(A) tail, we performed RNase H digestion in RNA pre-incubated with a poly(T) oligonucleotide. Consistent with the results using VSG-C, the major band detected by mBA-antibody completely disappears, further supporting the idea that in bloodstream forms most m6A is present in the poly(A) tail of VSG mRNA (Fig. 1f-g). Interestingly, digestion of RNA hybridized with poly(T), also abolished the smear detected by m6A-antibody (Fig. 1g), indicating that most m6A present in non-VSG transcripts appears to be located in their poly(A) tails. Notably, a similar approach to digest poly(A) tails does not affect m6A levels in mammalian mRNA of HeLa cells¹⁸.

Most m6A in mRNA from human cells is preceded by guanosine³⁰³¹. Since m6A in T. brucei appears to be localized in the poly(A) tail, we wanted to understand if m6A is in a similar sequence context in trypanosome mRNA. To test this, we performed RNA digestion with selective nucleases, similar to assays originally performed to establish the m6A sequence context in mammals30. In these experiments, we digested trypanosome RNA with RNase T 1 , which cleaves RNA after G, or RNase A, which cleaves RNA after pyrimidines (C/U). These enzymes leave the subsequent nucleotide with a 5’hydroxyl. Thus, if m6A follows G or C/U, it would be released with a 5’-hydroxyl, after RNase T1 or RNase A digestion, respectively. Next, the RNA is digested with nuclease P1 , which leaves any other m6A with a 5’-phosphate. m6A can then be quantified using isotope-labelled m6A as a standard. m6A that is preceded by an A can be extrapolated by subtracting total m6A levels from m6A preceded by G and C/U. Using this approach, m6A in mammalian mRNA is primarily preceded by G (Fig. 1 h), as described previously30. However, m6A in T. brucei bloodstream form mRNA was primarily preceded by A (53%), and only 27% of m6A is preceded by G (Fig.

1 h). This is consistent with a strong enrichment of m6A in the poly(A) tails. We also tested m6A in related species: Trypanosoma congolense, Trypanosoma cruzi and Leishmania infantum. Interestingly, T. congolense and T. cruzi showed an even higher enrichment of m6A after an A (69% and 68%, respectively) (Fig. 1 h). In contrast, L. infantum showed a unique digestion pattern, whereby 53% m6A is located after a C or a U (Fig. 1 h). With these results, we predict that the three more closely related species, T. brucei. T. congolense and T. cruzi, harbour a large fraction of m6A in the poly(A) tail of transcripts, while L. infantum contains m6A in an internal region of the transcripts and in a consensus motif different from mammalian cells.

Our results show that trypanosomatids harbour a large fraction of m6A in a sequence context different from mammalian cells. In T. brucei, although transcripts from >300 genes harbour m6A, the VSG gene family is the most represented. Importantly, we also show that around 50% of m6A in a cell is present in the poly(A) tail of the actively transcribed VSG mRNA. Based on the m6A frequency in the T. brucei transcriptome and the enrichment in VSG, we estimate that there are nearly four m6A molecules per VSG mRNA.

Timina of m6A removal

The half-life of VSG transcript is 90-270 minutes, while the median mRNA half-life in trypanosomes is minutes¹⁰. Given that removal of the poly(A) tail often precedes RNA degradation, we hypothesized that the presence of m6A in the poly(A) tail could contribute to this exceptional VSG mRNA stability.

We tracked m6A levels in VSG mRNA as it undergoes degradation in three independent conditions. We first inhibited transcription in bloodstream form parasites with actinomycin D (ActD) and, for the next 6 hours, we quantified the amount of VSG mRNA that remains by RT-qPCR. We also measured the levels of m6A in VSG (by immunoblotting) and the length of the VSG poly(A) tail using Northern blotting and the Poly(A) Tail- Length Assay (PAT), which involves ligation of adaptors to the 3’end of poly(A) tails and two consecutive PCRs using VSG-specific forward primers (Fig. 2a). The amplified fragments contain part of the open reading frame (ORF), the 3’UTR of VSG transcript and the downstream poly(A) tail, whose size is variable between different transcript molecules.

VSG mRNA has previously been shown to exhibit biphasic decay: in the first hour after transcription blocking VSG mRNA levels remain high and, only in a second phase, do VSG mRNA levels decay exponentially¹⁴’³². Consistent with these earlier findings, we detected no major changes in mRNA abundance during the first hour after actinomycin D treatment (lag phase, or first phase); however, afterwards VSG exhibited exponential decay (second phase) (Fig. 2b). Northern blotting and PAT assay revealed that during the one- hour lag phase, the length of the VSG poly(A) tail was stable, but then it rapidly shortened during the second phase (Fig. 2c). At 120 and 240min, the intensity of the VSG poly(A) tail drops rapidly, indicating that the transcript was also rapidly degraded. This indicates that there is a specific time-dependent step that triggers the rapid shortening of the VSG poly(A) tail and the subsequent degradation of the VSG transcript. Immunoblotting revealed that the m6A levels also decreased, but strikingly the loss of m6A preceded the shortening of the poly(A) tail and subsequent mRNA decay (Fig. 2b, Fig. 2d). In fact, m6A levels decrease exponentially during the first hour after actinomycin D, taking around 35 min for total mRNA m6A levels to drop 50%, while VSG mRNA only reached half of the steady-state levels around 2hours (Fig. 2b). These results indicate that m6A is removed from VSG mRNA prior to the deadenylation of the poly(A) tail, which is quickly and immediately followed by degradation of the transcript.

When bloodstream form parasites undergo cellular differentiation to procyclic forms, VSG is downregulated as a consequence of decreased transcription and decreased mRNA stabilty¹⁴. To test whether m6A is also rapidly removed from VSG mRNA prior to its developmentally programmed degradation, we induced differentiation in vitro by adding cis-aconitate to the medium and changing the temperature to 27°C. Parasites were collected and total RNA was extracted in different time points. Quantitative RT-PCR showed that the levels of VSG mRNA stayed stable for around one hour, which was followed by an exponential decay (Fig. 2e). Importantly, immunoblotting analysis showed that during parasite differentiation, m6A intensity in the VSG mRNA dropped faster than the VSG-mRNA levels, reaching half of the steady-state levels in 23 min (Fig. 2e-f). Thus, during parasite differentiation, we observed again that the removal of m6A precedes the loss of VSG mRNA levels.

Overall, these results show that in two independent conditions, m6A is removed from the VSG transcript earlier than the VSG transcript is deadenylated and degraded, suggesting that m6A may need to be removed from the VSG transcript before it can be degraded.

VSG mRNA is methylated in the nucleus

In most organisms, m6A is generated by methylation of adenosine residues within a specific consensus sequence by the METTL3 methyltransferase or its orthologsl . Given that in T. brucei, m6A is present in the poly(A) tail, a different mechanism is likely used. Indeed, trypanosomes lack a METTL3 ortholog³³, indicating that a different pathway would be required to acquire m6A in the poly(A) tail. To understand how m6A accumulates in the VSG mRNA, we used parasite differentiation as a natural inducible system of VSG downregulation. This process is reversible in the first two hours34. Parasite differentiation was induced by adding cis-aconitate for 30 min (as described above, Fig. 2e), and then was washed away (Fig. 3a). Immunoblotting revealed that m6A was reduced after 30 min of cis-aconitate treatment (Fig. 3b-c), while mRNA levels remained unchanged (Fig. 3d). When cells were allowed to recover or 1 hour in the absence of cis-aconitate (Flask 4, Fig. 3a), we observed that the intensity of the m6A signal returned to normal levels (Fig. 3c), while mRNA levels continued to remain constant (Fig. 3d). These results indicate that the levels of m6A can be recovered without a net increase in VSG mRNA levels. The net levels of VSG mRNA result from a balance between de novo transcription and degradation. To test if the recovery of m6A levels after cisaconitate removal was due to de novo transcription, parasites were cultured in the presence of actinomycin D (Flask 5, Fig. 3a). We observed that the intensity of m6A in the VSG transcript was not recovered. Instead, the m6A levels decreased to ~20% (Fig. 3c). Overall, these results indicate that de novo transcription is required to re-establish m6A levels in VSG mRNA. If m6A is incorporated into VSG mRNA soon after transcription, and if it remains in the poly(A) tail until it gets degraded, we should be able to detect m6A in the nucleus and in the cytoplasm of the parasites.

Immunofluorescence analysis with an antibody against m6A showed a punctate pattern both in the nucleus (20%) and cytoplasm (80%) (Fig. 3e, 3f). To confirm that this m6A signal originated from RNA, we incubated the fixed cells with nuclease P1 prior to the antibody staining, which specifically cleaves single-stranded nucleic acids without any sequence-specific requirement. This treatment caused a marked reduction in the intensity of the m6A signal, indicating that the immunoreactivity of the m6A antibody derives from RNA, and not from non-specific interactions with cellular proteins (Fig. 3e, 3g). As an additional control, we treated with actinomycin D for 2 hours prior to immunofluorescence analysis. This treatment is expected to result in reduced cellular mRNA levels. Immunostaining with the m6A antibody showed a drop in the intensity of the m6A signal by around 40% (Fig. 3e, 3g), which is similar to the trend observed by immunoblotting (Fig. 2c). Overall, these data show that the m6A immunostaining likely reflects m6A in mRNA and validates the immunoblotting results.

To understand how m6A is incorporated into VSG mRNA, we performed RNA-FISH analysis of VSG2 (Fig. 3h) and compared it with m6A localisation (Fig. 3e). Quantification of the FISH signal reveals that -19% of the signal is in the nucleus and 81% in the cytoplasm (Fig. 3i). Given that this transcript distribution is very similar to the subcellular distribution of m6A, the data suggests that the concentration of m6A per transcript is relatively similar in the two compartments. To confirm this hypothesis, we biochemically fractionated parasites into nuclear and cytoplasmic fractions and quantified m6A in each fraction. Fractionation was confirmed by DAPI staining of the nuclei and Western blotting. Equal masses of RNA from nuclear and cytoplasmic fractions were spotted on a nylon membrane and hybridized with anti-m6A antibody (Fig. 3j). Quantification of m6A signal showed that the m6A intensity was similar in the two fractions (Fig. 3k), revealing that the concentration of m6A per transcript is similar in the two cell compartments. Taken together, our data indicate that methylation of VSG mRNA poly(A) takes place in the nucleus, soon after transcription.

A VSG motif is required for methylation m6A is added to the VSG mRNA poly(A) tail soon after transcription, probably still in the nucleus. The m6A- RIP analysis showed that m6A is particularly enriched in VSG transcripts (Fig. 2d). We therefore asked how the VSG poly(A) tails are selected for preferential enrichment of m6A. It has been previously shown that each VSG gene contains a conserved 16-mer motif (5’-TGATATATTTTAACAC-3’) in the 3’UTR adjacent to the poly(A) tail that is necessary for VSG mRNA stability¹². It has been recently shown that an RNA-binding complex binds this motif and stabilizes the transcript by a yet unknown mechanism¹²'³⁵. Here, we hypothesized that this 16-mer motif may act in cis to promote inclusion of m6A of the adjacent poly(A) tail. VSGs are essential proteins that are transcribed monoallelically from a telomeric location called the Bloodstream Expression Site (BES). If we mutagenized the 16-mer motif from the monoallelically transcribed VSG2 gene, this would reduce the levels of VSG2 protein, which is lethal for the parasites¹². To solve this problem, a VSG2-expressing parasite line was genetically modified to introduce a reporter VSG gene (VSG117) in the same BES by homologous recombination. The resulting cell-lines were called VSG double- expressors (DE), because they simultaneously express the endogenous VSG2 and the reporter VSG117 (Fig. 4a). In the DE1 cell line, the VSG117 gene contained a wild-type 16-mer motif. In the DE2 cell-line, VSG117 contained a 16-mer motif in which the sequence was scrambled (5'-GTTATACAAAACTTTT-3') (Fig. 4a). As has been previously reported, the transcript levels of VSG2 and VSG117 are dependent on each other and are dependent on the presence of the 16-mer motif12. RT-qPCR analysis showed that the two VSGs have roughly the same levels in DE1 cell-line. However, in DE2, VSG117 transcript is about 7-fold less abundant than VSG2 (Fig. 4b), confirming that the 16-mer motif is important for the abundance of VSG transcripts.

To test whether the 16-mer motif is required for inclusion of m6A in VSG poly(A) tails, we performed m6A immunoblotting of cellular RNA obtained from the two double-expressor cell lines. Given that VSG2 and VSG117 transcripts have similar sizes (~1 .8 kb), we used RNase H to selectively cleave VSG2 before resolving the RNA on gel. VSG2 cleavage was performed by incubating the total RNA sample with an oligonucleotide that hybridizes to the VSG2 ORF followed by incubation with RNase H (as described in Fig. 1 f-g). As expected, the VSG2-m6A-containing fragment is smaller and runs faster on an agarose gel (Fig. 4c). An “m6A index”was calculated by dividing the relative intensity of m6A in each VSG band (Fig. 4c) by the corresponding relative transcript levels measured by RT-qPCR (Fig. 4b). A low m6A index indicates a given transcript has fewer modified nucleotides (Fig. 4d).

Whenever the 3’UTR of VSG transcripts contain a 16-mer WT (VSGs with a blue box in Fig. 4a), VSG m6A bands are detectable by immunoblot and the m6A index varies between 20-140 arbitrary units. In contrast, when the 16-mer motif is mutagenized (VSG117 with orange box in Fig. 4a), the VSG m6A is undetectable (Fig. 4c), and the m6A index therefore is too low to calculate. These results indicate that the motif is required for inclusion of m6A in the VSG transcript. If the 16-mer motif played no role in m6A inclusion, the VSG117 m6A index would be identical in both cell-lines (DE1 and DE2), i.e. around 20. Given that the RT-qPCR quantifications showed the relative intensity of VSG117 16-merMUTis -0.10 (Fig. 4b), the predicted intensity of the VSG117 16-merMUTm6A band would have been 20 x 0.10= 2.0 arbitrary units. To be sure that a band with this level of m6A would be detected on an immunoblot, we ran a more diluted DE1 RNA sample in lane 3 (Fig. 4c). The intensity of the VSG117 16-merWTband is 2.1 arbitrary units (Fig. 4c and 4d), and it is readily detected in the immunoblot. Given that we could not detect any band corresponding to a putative methylated VSG117 16-merMUTin DE2 (even after over exposure of the immunoblot, Extended Data Fig. 6a), we conclude that the VSG conserved 16-mer motif is necessary for inclusion of m6A in the VSG poly(A) tail. A similar immunoblotting analysis was performed in an independent pair of DE cell-lines (DE3 and DE4) that express a different reporter VSG (VSG8). Consistently, m6A was not detectable when the 16-mer motif of VSG8 was mutagenized, further supporting the conclusion that this motif is required for detectable methylation.

To determine if the 16-mer motif is associated with the presence of m6A in the VSG transcript, we used the m6A-RIP data to compare the enrichment of m6A in VSG transcripts with and without a 16-mer motif (Fig. 4e). m6A enrichment is calculated as the ratio between number of normalized reads in immunoprecipitated versus input samples. Among the 20 VSG transcripts detected after immunoprecipitation (Fig. 1 d), we found that the 11 VSG transcripts with a 16-mer motif (blue bars) are on average, 5-fold more enriched in m6A than the nine VSG transcripts lacking a 16-mer motif (orange bars) (log2FC=2.1 , P< 0.0001 , Mann-Whitney test) (Fig. 4e). The 11 transcripts with a 16-mer motif encode fully functional VSG proteins or one pseudogene and all genes are located in the specialized subtelomeric loci from where VSG can be transcribed (BES). In contrast the nine VSG transcripts that lack the 16-mer motif are located in non-BES sites and most of them (seven) are pseudogenes.

To confirm that the m6A-enrichment detected in VSGs containing a 16-mer motif was not simply a reflection of higher expression of those VSG genes, the transcript levels measured from the m6A-RIP input sample were plotted against m6A-enrichment for each of the 20 detected VSG genes (Fig. 4f). As expected, VSG2 is the active gene and with the largest CRM. Importantly, we found no correlation between m6A enrichment and transcript levels, indicating that the yield of m6A enrichment in the immunoprecipitation experiment is not dependent on the abundance of the transcript. This data indicates that the observation that 16-mer motif containing transcripts are more enriched in m6A is independent of VSG transcript levels and reflects the functional link between the 3’UTR motif and the presence of m6A in the adjacent poly(A) tail. m6A is required for VSG mRNA stability

The unusual localization of m6A in the poly(A) tail suggests that the underlying biochemistry of m6A formation in trypanosomes is different from what has been described in other eukaryotes. Consistently, sequence searches using hidden Markov models (hmmer.org) did not find a METTL3 methyltransferase, nor Alkbh5 demethylase orthologs in kinetoplastida³³. Given that, at this stage the mechanism of m6A formation in the poly(A) tail is unknown and therefore cannot be directly blocked, we used the genetic mutants of the 16-mer conserved motif to enquire about the function of m6A in VSG mRNA.

To test the role of the 16-mer motif on poly(A) length in mRNA stability, we measured VSG mRNA stability in 16-merWTand 16-merMUTcell-lines. VSG mRNA half-life was measured by blocking transcription for 1 hour with actinomycin D (the duration of the lag phase during VSG mRNA decay) and the levels of VSG mRNA were followed by RT-qPCR. PAT assay clearly shows that, when the VSG117 transcript contains the 16-mer motif (16-merWTcell-line), the length of the VSG117 poly(A) tail is stable for 1 hour (Fig. 5a-b). In contrast, VSG117 transcripts containing a scrambled 16-mer motif exhibited very rapid shortening of the poly(A) tail. In this case, there was no detectable lag phase — instead, the length of the poly(A) tail was reduced to 25% of its original length after just 15 minutes and was undetectable after 1 hour (Fig. 5a-b). Consistent with the fast kinetics of poly(A) deadenylation, in the absence of an intact 16-mer motif, VSG117 transcript levels decayed very rapidly with a half-life of ~20 minutes (Fig. 5b). These experiments show that when the VSG conserved 16-mer motif is mutated and m6A is lost, the VSG transcript is no longer stable and exhibits rapid poly(A) deadenylation and a marked reduction of mRNA stability.

CAF1 is the deadenylase responsible for the deadenylation of most transcripts in T. brucei³⁶. Here, we asked if CAF1 also deadenylates VSG mRNA and whether the presence of m6A affected this process. To test this, we downregulated CAF1 by tetracycline-inducible RNA interference. We followed the decay of the VSG2 transcript and its poly(A) tail length after actinomycin D treatment (Fig. 5c-d). In WT conditions (-Tet condition), the shortening of the poly(A) tail and decrease in VSG transcript levels starts one hour post actinomycin D treatment and the poly(A) tail is entirely deadenylated in 4 hours (Fig. 5c-d). In contrast, when CAF1 is downregulated, in the first two hours post-actinomycin D treatment, both the length of the poly(A) tail and VSG transcript levels remained unchanged. Only after 4 hours, we see a slightly shorter poly(A) tail and a decrease in VSG transcript levels. The fact that the phenotype of CAF1 depletion is only detected after 1 hour post-actinomycin D treatment demonstrates that CAF1 does not have strong activity on the VSG poly(A) tail during the first hour of treatment. Given our previous finding that it takes about 1 hour to remove m6A from VSG mRNA (Fig. 2b), CAF1 may be partially inhibited while VSG poly(A) tail is methylated, but once m6A has been removed from the poly(A) tail, CAF1 may then be able to rapidly deadenylate the VSG transcript.

To further understand the mechanism by which m6A inhibits VSG deadenylation, we performed RNA-FISH to determine the subcellular localization of VSG. Given that CAF1 is localized predominantly in the cytoplasm³⁷, we predicted that a VSG transcript with a mutagenized 16-mer motif (and hence lower levels of m6A) would be unaffected in the nucleus but would be rapidly degraded by CAF1 in the cytoplasm. To test this, we genetically modified the CAF1-inducible RNAi cell-line in order to have a reporter VSG (VSG8) where the 16-mer motif was either wild-type (wt) or mutagenized (Fig. 5e-f).

When the VSG8 has a WT 16-mer motif, depletion of CAF1 leads to a small increase in VSG8 transcript abundance and VSG8 remains distributed -20% in the nucleus and 80% in cytoplasm relative to the condition when CAF1 is present (Fig. 5e-f). In contrast, when the 16-mer motif of VSG8 is mutagenized (i.e. when m6A levels are undetectable) and CAF1 is present, the levels of VSG8 in the cytoplasm show a sharp decrease, while the nuclear signal is less affected (Fig. 5e-f). Finally, when the 16-mer motif is mutagenized and CAF1 is depleted, we observe a significant recovery of VSG8 RNA-FISH signal, especially in the cytoplasm, indicating CAF1 is responsible for deadenylating most mutagenized VSG transcript and this process takes place in the cytoplasm (Fig. 5e-f).

In the two conditions in which VSG8 have a mutagenized 16-mer motif, the nuclear levels of VSG8 are significantly lower (MFI is around 100) than in the conditions where VSG8 has a WT 16-mer motif (MFI is around 200) (Fig. 5e-f). These results suggest that when m6A levels are reduced in the poly(A) tail, the stability of the VSG mRNA is also partially reduced in the nucleus and this appears to be independent of CAF1. Together this data is consistent with a model in which m6A inhibits CAF1 activity in the cytoplasm, reducing poly(A) deadenylation and thus contributing to VSG mRNA stability.

Effect of m6A in the poly(A) tail of in vitro transcribed RNAs

An enhanced Green florescent protein reporter mRNA with m⁶A in the poly(a) tail was generated by in vitro transcription (IVT). A control version was the same transcript without the m⁶A in the poly(A) tail. The eGFP mRNAs were transfected to cells (in this case HeLa cells) and their abundance and decay kinetics (stability) measured (Fig 6).

Generation of in vitro poly(A) methylated mRNAs mRNA was produced by in vitro transcription (IVT) (Figure 7). The first step in this process was to have a DNA template with our gene of interest, which can be any gene that will be useful according to the desired application and should include not only a codon-optimized coding sequence but also appropriate regulatory sequences. The DNA template started with a T7 promoter, to start the in vitro transcription (the IVT kit is based on a T7 RNA polymerase, that will produce the mRNA from the sequence of the DNA template). The DNA template can be a plasmid (amplified in bacteria) or an amplification product (generated by PCR, polymerase chain reaction). The second step was the in vitro transcription reaction, in which the T7 RNA polymerase (megaMEGAscript T7 Kit, AM1333) transcribed the mRNAs from the sequence in the DNA template. The RNA was produced and the DNA removed by DNase I treatment followed by ethanol precipitation (to remove the reaction components and concentrate the RNA). This produced the in vitro synthesised RNAs. As the T7 RNA polymerase had several rounds of abortive transcription (stops before the end, producing shorter RNAs) in addition to the desired RNA and could produce double-stranded RNAs (dsRNAs), these unwanted RNAs were removed. This can be done by a number of methods and the choice of method depends on the application. Suitable methods could be HPLC (high-pressure liquid chromatography), cellulose purification or gel extraction. The third step was the addition of the 5'cap to the mRNA using the Vaccinia Capping System (NEB M2080S). This generated a capped mRNA (the cap is necessary to improve stability and translation in vivo and is part of the normal mature mRNAs). This capping was also done co-transcriptionally with specific kits (CleanCap® Reagent AG, TriLink). After the capping reaction, the capped mRNA was purified by column purification (RNeasy, QIAGEN) to remove all the capping reagents. The fourth step was the addition of a methylated poly(A) tail at the 3’end of the in vitro transcribed mRNA. One approach was the ligation of a commercially synthesized methylated RNA poly(A) oligo (with the number and positions of m6A defined). The ligation was based on T4 RNA Ligase enzyme. At the end, the desired ligated mRNA was purified from the non-ligated mRNAs. A second approach to polyadenylate the mRNA was via the enzymatic reaction of a poly(A) polymerase. In this method, the poly(A) tail was synthesized by extension from the end of the mRNA by the poly(A) polymerase (E. coll Poly(A) Polymerase, NEB M0276S) with the ATP as adenosine donor. To add the m6A, a pre-methylated ATP (m6ATP) was added to the reaction. In this way, m6A was incorporated in the poly(A) tail mixture interspersed with adenosines. In our protocol, we used a 50% ATP : 50% m6ATP mixture. We used the normal E. coll Poly(A) Polymerase (NEB M0276S) protocol with one alteration that allowed the incorporation of the m6A: the normal poly(A) polymerase buffer was replaced by CutSmart Buffer (NEB). Finally, polyadenylated mRNA was column purified (RNeasy, QIAGEN) to remove the polyadenylation reagents. Additionally, the mRNA was treated with Antarctic Phosphatase neb (NEB M0289S) to remove residual NTPs from the previous reactions, that could be activate cellular immune responses, followed by another column purification. The Antarctic Phosphatase treatment is optional but may improve the response of cells if NTPs are contaminating the purified sample.

Delivery of mRNA into a cell-line in vitro

The delivery of the methylated RNAs into mammalian cells depends on the cells to be transfected, the goal of the transfection, the application, etc. We transfected cells in culture (Hela cells) using the lipofectamine reagent (Lipofectamine™ 2000 T ransfection Reagent 11668019). Figure 8 shows the effect of transfecting different amounts of eGFP mRNA (0, 100, 500 and 2500 ng). 24 hours after transfection the cells were washed three times with PBS, the mRNA levels were measured by quantitative PCR. We observed that the eGFP mRNA detected in the RNA was proportional to the amount of RNA used for transfection. The sample without RNA (0 ng) showed the background levels and was the same as the sample that contained the RNA but no reagent for transfection, 100 ng (no transfection). This means that if there was no transfection, the RNA outside the cells was removed by the PBS washes. Therefore, the RNA measured in the transfected samples is mainly RNA that was transfected (intracellular RNAs) and not RNA that was outside the cells (non-transfected RNAs). In addition, the RNA not treated with Antarctic phosphatase (500 ng no Antarctic) showed less intracellular RNA than the 500ng transfection. This indicated that the removal of residual NTPs improves the amount of cellular RNA in the cells Measuring the stability of methylated mRNAs

Upon transfection, as the RNAs enter the cells and are used and degraded by the cellular degradation machinery, we expected the mRNA levels to be degraded at a certain rate. If the presence of m⁶A in the poly(A) tail leads to increased stability, the rate of degradation should be slower than that of non-methylated RNAs. HeLa cells were transfected with eGFP mRNA and mRNA levels were measured 2, 4, 6, 8 and 24 hours after transfection (Figure 9). We observed that the levels of both methylated and non-methylated RNA increased in the first 6 hours after transfection. After 6 hours of transfection, we observed that the non- methylated mRNA levels decreased, while the methylated ones remained stable (with a small increase until 24 hours) (Fig 9). This demonstrates that m⁶A protects the mRNA from degradation.

The classic function of a poly(A) tail is to suppress mRNA degradation and to promote translation. Poly(A)- binding proteins (PABPs) bind to the poly(A) tail and stimulate mRNA translation by interaction with translation initiation factors³⁸. Removal of the poly(A) tail by deadenylase complexes is a prerequisite for mRNAs to enter into 5'-> 3’or 3'-> 5'degradation pathways³⁹⁴⁰. In this report, we identified a novel mechanism by which a poly(A) tail contributes to mRNA stability. We found that the presence of m6A in the poly(A) tail of VSG transcripts inhibits RNA degradation, most likely by impeding CAF1-mediated deadenylation.

The presence of m6A in the poly(A) tail is so far unique to trypanosomes. In other eukaryotes, m6A has been mainly detected by m6A mapping approaches around the stop codon and 3’UTR, where it plays a role in mRNA stability and translation¹. A mapping study was recently published in T. brucei in which m6A was mapped in internal regions of transcripts³. m6A was not reported to be in the poly(A) tail in this previous study. However, m6A mapping relies on aligning m6A-containing RNA fragments to genomic sequence. Since the poly(A) sequence is not encoded in the genome, any m6A-containing poly(A) tail would not be mappable and therefore not detected in this or any other previous m6A mapping study. It remains unclear how m6A gets into the poly(A) tail. The presence of m6A in the poly(A) tail suggests that an unusual RNA- methyltransferase will directly or indirectly bind to the 16-mer motif and methylate adenosines that are either adjacent to the 16-mer motif or become more proximal via a loop-like conformation of the poly(A) tail. This would explain why orthologs of the canonical METTL3 enzyme do not exist in the trypanosome genome³³. A recent study has shown that a RNA stabilizing complex, MKT1 complex, binds to the 16-mer motif³⁵, but this complex does not contain any homologues of m6A readers, writers or erasers. Deadenylation is the first step in the main mRNA decay pathway in eukaryotes⁴¹. T. brucei is not an exception29. In this study we showed that m6A seems to protect the poly(A) tail from deadenylation by CAF1. The molecular mechanism behind this stabilizing effect is unknown. It is possible that the CAF1 deadenylase is inefficient on a methylated poly(A) tail. There is structural and biochemical evidence that poly(A) tails adopt a tertiary structure that facilitates the recognition by some mammalian deadenylases (CAF1 and Pan2)⁴². When a poly(A) tail contains m6A, the tertiary structure may be not properly formed and deadenylase activity is inhibited, as has been shown with guanosine residues¹² within an oligo-A oligonucleotide⁴². In this model, a putative demethylase may be required to remove the methyl group, which could then allow the VSG poly(A) tail to be efficiently deadenylated by CAF1 . Alternatively, the stabilizing effect of m6A could result from recruitment of a specific RNA-binding protein, that prevents the poly(A) tail from being deadenylated. T. brucei has around 2000 VSG genes, but only one is actively transcribed at a given time². It has been proposed that the maximal amount of VSG mRNA per cell is dependent on a post-transcriptional limiting factor dependent on the presence of the 16-mer motif ¹²- The inclusion of m6A in poly(A) tails may be this factor. When the 16-mer motif is present in both VSG genes, both get partially methylated and their abundance is reduced to about half of a single-VSG expressor; however, when the 16-mer motif is absent from one of the VSGs, the second VSG is more methylated and the transcripts become more abundant. Our work is the first report of an RNA modification in poly(A) tails. We show that m6A is present in the poly(A) tail of T. brucei mRNAs, it is enriched in the most abundant transcript (VSG), and that m6A acts as a protecting factor stabilizing VSG transcripts from CAF1 deadenylation activity. It will be important for future studies to identify the enzymes and proteins involved in adding, reading, or removing m6A. Given the importance of VSG regulation for chronic infection and parasite transmission, drugs that interfere with m6A incorporation in poly(A) tails are expected to block parasite virulence. Understanding these regulatory epitranscriptomic processes may open up possibilities for developing therapeutic strategies to treat sleeping sickness.

References

1. Zaccara, S et al. Nat Rev Mol Cell Biol 20, 608-624 (2019).

2. Horn, D. Mol Biochem Parasitol 195, 123-129 (2014).

3. Liu, L. et al. Biochem Biophys Res Commun 508, 1286-1290 (2019).

4. Franco, J. R et al Parasitology 141 , 748-760 (2014).

5. Cross, G. A., et al. Mol Biochem Parasitol 195, 59-73 (2014).

6. Cross, G. A. Parasitology 71 , 393-417 (1975).

7. Nilsson, D. et al. PLoS Pathog 6, (2010).

8. Kraus, A. J et al. Sci Rep 9, 12281 (2019).

9. Gunzl, A. et al. Eukaryot Cell 2, 542-551 (2003).

10. Fadda, A. et al.. Mol Microbiol 94, 307-326 (2014).

11. Berberof, M. et al. Embo J 14, 2925-2934 (1995).

12. Ridewood, S. et al.. Mol Microbiol 106, 614-634 (2017).

13. Roditi, I. et al.. J Cell Biol 108, 737-746 (1989).

14. Ehlers, B et al. Mol Cell Biol 7, 1242-1249 (1987).

15. Matthews, K. R. J Cell Sci 118, 283-290 (2005).

16. Wei, C. M. et al. Cell 4, 379-386 (1975).

17. Perry, R. P. et al. Cell 4, 387-394 (1975).

18. Meyer, K. D. et al. Cell 149, 1635-1646 (2012).

19. Dominissini, D. et al. Nature 485, 201-206 (2012).

20. Jia, G. et al.. Nat Chem Biol 7, 885-887 (2011 ).

21. Zheng, G. et al. Mol Cell 49, 18-29 (2013).

22. Freistadt, M. S et al. J Biol Chem 263, 15071-15075 (1988).

23. Bangs, J. D. et al. J Biol Chem 267, 9805-9815 (1992).

24. Perry, K. L. et al Proc Natl Acad Sci U S A 84, 8190-8194 (1987).

25. Hauenschild, R. et al. Nucleic Acids Res 43, 9950-9964 (2015).

26. Boccaletto, P. et al.. Nucleic Acids Res 46, D303-D307 (2018).

27. Molinie, B. et al. Nat Methods 13, 692-698 (2016).

28. Xiang, J. F. et al. Mol Cell 69, 126-135. e6 (2018).

29. Clayton, C. E. Open Biology 9, 190072 (2019).

30. Schibler, U. et al J Mol Biol 115, 695-714 (1977).

31. Linder, B. et al. Nat Methods 12, 767-772 (2015).

32. Hoek, M., et a/ Mol. Biochem. Parasitol. 120, 269-283 (2002).

33. Iyer, L. M., et al. Bioessays 38, 27-40 (2016).

34. Domingo-Sananes, M. R., et al J Cell Biol 211 , 455-468 (2015).

35. Melo do Nascimento, L. et al.. Elife 10, (2021 ).

36. Fadda, A. et a/ RNA 19, 937-947 (2013).

37. Schwede, A. et al. Nucleic Acids Res 36, 3374-3388 (2008).

38. Gallie, D. R. Translation (Austin) 2, e959378 (2014).

39. Garneau, N. L. et al Nat Rev Mol Cell Biol 8, 113-126 (2007).

40. Chen, C. Y. & Shyu, A. B. Wiley Interdiscip Rev RNA 2, 167-183 (2011 ).

41. Decker, C. J. & Parker, R. Genes Dev 7, 1632-1643 (1993).16

42. Tang, T. T. L. et al. Nat Struct Mol Biol 26, 433-442 (2019).

Claims

Claims

1 . A method of producing a messenger RNA (mRNA) molecule comprising providing an RNA molecule comprising a coding sequence; and attaching a poly(A) tail comprising one or more N⁶-methyladenosine (m⁶A) bases to 3' end of the RNA molecule to produce an mRNA molecule.

2. A method of stabilising a messenger RNA (mRNA) molecule in a mammalian cell comprising attaching a poly(A) tail comprising one or more m⁶A bases to the 3’ end of the mRNA molecule, wherein the m⁶A residues in the poly(A) tail stabilise the mRNA molecule.

3. A method according to claim 1 or claim 2 wherein the presence of one or more m⁶A residues in the poly(A) tail increases the in vivo half-life of the mRNA molecule

4 A method according to any one of the preceding claims wherein the poly(A) tail is 20 to 300 bases in length.

5. A method according to any one of the preceding claims wherein 10 or more of the bases in the poly(A) tail are m⁶A.

6. A method according to any one of the preceding claims wherein 5% or more of the bases in the poly(A) tail are m⁶A.

7. A method according to any one of the preceding claims wherein the mRNA comprises a coding sequence for an antigen or a therapeutic protein.

8. A method according to any one of the preceding claims wherein the poly(A) tail is attached by a method comprising contacting the RNA molecule or mRNA molecule with a poly(A) polymerase in the presence of m⁶A bases, such that the poly(A) polymerase generates a poly(A) tail comprising one or more m⁶A bases at the 3’ end of the mRNA molecule or RNA molecule.

9. A method according to any one of claims 1 to 7 wherein the poly(A) tail is attached by a method comprising contacting the RNA molecule or mRNA molecule with a single stranded poly(A) molecule comprising one or more m⁶A bases in the presence of an RNA ligase, such that the single stranded poly(A) molecule comprising one or more m^sA bases is ligated to the 3’ end of the mRNA molecule or RNA molecule to form a poly(A) tail.

10. A method according to any one of the preceding claims further comprising isolated or purifying the mRNA molecule.

11. A method according to any one of the preceding claims further comprising encapsulating the mRNA molecule in a lipid nanoparticle (LNP).

12. A method according to any one of the preceding claims further comprising formulating the mRNA molecule or lipid nanoparticle into a pharmaceutical formulation with a pharmaceutically acceptable excipient.

13. An isolated mRNA molecule comprising a poly(A) tail comprising one or more m⁶A residues.

14. An isolated mRNA molecule according to claim 13, wherein the mRNA molecule is produced by a method of any one of claims 1 to 12.

15. An isolated mRNA molecule according to claim 13 or 14, wherein the mRNA molecule comprises a coding sequence for an antigen.

16. An isolated mRNA molecule according to any one of claims 13 to 15, wherein the mRNA molecule comprises a coding sequence for a therapeutic protein.

17. A lipid nanoparticle comprising an mRNA molecule according to any one of claims 13 to 16.

18. A pharmaceutical composition comprising an mRNA molecule according to any one of claims 13 to 16 or a lipid nanoparticle according to claim 17.

19. A kit for use a method according to any one of claims 1 to 12 comprising a single stranded poly(A) molecule comprising one or more N⁶-methyladenosine (m^sA) residues and an RNA ligase.

20. A kit for use a method according to any one of claims 1 to 12 comprising adenosine, N⁶- methyladenosine (m⁶A) and a poly(A) polymerase.