N⁶-methyladenosine in poly(A) tails stabilize VSG transcripts

Summary

RNA modifications are important regulators of gene expression¹. In Trypanosoma brucei, transcription is polycistronic and thus most regulation happens post-transcriptionally². N⁶-methyladenosine (m⁶A) has been detected in this parasite, but its function remains unknown³. Using RNA immunoprecipitation, we found that m⁶A is enriched in 342 transcripts, with an enrichment in Variant Surface Glycoproteins (VSGs). Strikingly, ~50% of the m⁶A is located in the poly(A) tail of the actively expressed Variant Surface Glycoprotein (VSG) transcript. m⁶A residues are removed from the VSG poly(A) tail prior to deadenylation and mRNA degradation. Computational analysis revealed an association between m⁶A in the poly(A) tail and a 16-mer motif in the 3’UTR of VSG. Using genetic tools, we showed that the 16-mer motif acts as a cis-acting motif required for inclusion of m⁶A in the poly(A) tail. Removal of this motif from the VSG 3’ UTR results in poly(A) tails lacking m⁶A, rapid deadenylation and mRNA degradation. To our knowledge this is the first identification of an RNA modification in the poly(A) tail of any eukaryote, uncovering a novel post-transcriptional mechanism of gene regulation.

Introduction

Trypanosoma brucei (T. brucei) is a protozoan unicellular parasite that causes lethal diseases in sub-Saharan Africa: sleeping sickness in humans and 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)^7,8. 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) tail^5,11. 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. 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 identified^20,21. m⁶A affects several aspects of RNA biology, for instance contributing to mRNA stability, mRNA translation, or affecting alternative polyadenylation site selection (reviewed in¹).

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.

Results

m⁶A is enriched in the VSG poly(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, Extended Data Fig. 1 and Extended Data Table 1). 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 rRNA^22–24. 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. 1b). Given the importance of m⁶A for RNA metabolism in other eukaryotes, we focused on this specific modification in T. brucei.

Figure 1. 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, m^6,6A, m⁷G and m¹A P<0.0001). N = 5 biological samples. c, m⁶A levels quantified using standard curve in Extended Data Fig. 2d. 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 log2 counts per million reads mapped (CPM). 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. P values: Supplementary Table 1. 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 log2 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: β-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).

(see also Supplementary Figure S1 and Source Data Figure 1)

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 (Extended Data Fig. 2a), 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 (Extended Data Fig. 2b). 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 (m¹A) (Extended Data Fig. 2c), a modification commonly found in rRNAs and tRNAs^25,26.

The m⁶A standard allowed us to quantify the abundance of m⁶A in mRNA fractions of bloodstream and procyclic forms (Extended Data Fig. 2d): m⁶A represents 0.06–0.14% of total adenines in mRNA (Fig. 1c; Extended Data Fig. 2e). 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 non-fragmented RNA, followed by sequencing of both input and immunoprecipitated (IP) transcripts. Similar strategies have been described before to study methylation at the gene level^27,28. 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 m⁶A (Fig. 1e). Non-VSG transcripts can also be enriched in m⁶A, such as cyclophilin A and ubiquitin carrier protein (UCP) (Fig. 1d).

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 α-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, Extended Data Fig. 3b). 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 (Extended Data Fig. 3c–d). 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 m⁶A 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 m⁶A 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 m⁶A-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 guanosine^30,31. 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 T1, 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. 1h), 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. 1h). 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. 1h). In contrast, L. infantum showed a unique digestion pattern, whereby 53% m⁶A is located after a C or a U (Fig. 1h). 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.

Timing of m⁶A removal

The half-life of VSG transcript is 90–270 minutes, while the median mRNA half-life in trypanosomes is 13 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.

Figure 2. 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 m⁶A. 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. A0 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. Quantification is in Extended Data Figure 4. d, m⁶A immunoblotting of bloodstream form total RNA extracted from parasites treated with ActD (c). Methylene Blue stains rRNA. Quantification is in Extended Data Figure 4. 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. Same colour code as in Panel b. 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.

(see also Supplementary Figure S1 and Source Data Figure 2)

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 exponentially^14,32. 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, Extended Data Fig. 4). 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 and Extended Data Fig. 4). 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 orthologs¹. 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 for 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 cis-aconitate 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.

Figure 3. 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). Drawings were obtained from smart.servier.com. 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, m⁶A 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, 4μm; DIC, differential interference contrast.

(see also Supplementary Figure S1 and Source Data Figure 3)

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; Extended Data Fig. 5a). 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; Extended Data Fig. 5b). 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; Extended Data Fig. 5b), 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, Extended Data Fig. 5c). 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 (Extended Data Fig. 5d, 5e). 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 mechanism^12,35. 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.

Figure 4. 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 VSG117 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 immunoprecipited samples.

(see also Supplementary Figure S1 and Source Data Figure 4)

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. 1f–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-mer^MUT is ~0.10 (Fig. 4b), the predicted intensity of the VSG117 16-mer^MUT m⁶A band would have been 20 × 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-mer^WT band 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-mer^MUT in 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) (Extended Data Fig. 6b–c). 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. 1d), 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 CPM. 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-mer^WT and 16-mer^MUT cell-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-mer^WT cell-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.

Figure 5. 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.

(see also Supplementary Figure S1 and Source Data Figure 5)

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 (Extended Data Fig. 7). 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, Extended data Fig. 6b).

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.

Discussion

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 pathways^39,40. 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 residues 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. The Rudenko lab has 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.

Extended Data

Extended Data Fig. 1. Chemical structures of RNA modifications found in T. brucei.

a, 15 modifications were detected in poly(A)-enriched RNA (mRNA). b, 19 modifications were not detected in poly(A)-depleted RNA. All modifications were detected in total RNA. Structures were obtained from the database Modomics (http://genesilico.pl/modomics/).

Extended Data Fig. 2. Detection of m⁶A in T. brucei by mass-spectrometry.

a-b, Chromatograms obtained by LC-MS/MS analysis of a N6-methyladenosine standard and three RNA samples of T. brucei bloodstream form (BSF, Panel a) or insect procyclic stage (PCF, Panel b): total RNA, RNA enriched with poly(T)-beads (i.e., poly(A)-enriched RNA) and RNA that did not bind to polyT-beads (i.e., poly(A)-depleted RNA). c, Chromatogram obtained by LC-MS/MS analysis of a N1-methyladenosine standard with the 282->150 mass transition. The m¹A peak is detected at 6.5 minutes. d, Standard curve of m⁶A. Increasing quantities of commercially synthesized m⁶A were loaded on the HPLC column and the area under the chromatogram peak was measured. e, Quantification of the m⁶A/A(%) in the mRNA in a second independent experiment. n = 5 mRNA samples.

(see also Source Data of Extended Figures)

Extended Data Fig. 3. m⁶A detection in T. brucei by immunoblotting.

a, Specificity of anti-m⁶A antibody. Oligonucleotides containing either m⁶A (positive control), unmodified adenosine or m¹A (negative controls) were manually spotted in the membrane, which was incubated with anti-m⁶A antibody. The antibody specifically recognized the oligos with m⁶A, while exhibiting low cross-reactivity to the oligos with only unmodified adenosine or containing m¹A. b, m⁶A signal intensity in the immunoblot, measured by Image J, in the whole lane containing the poly(A)-enriched RNA of bloodstream forms. c, The intensity of the ~1.8 kb band was divided by the signal intensity of the entire lane. n = 5 biological replicates. d, m⁶A immunoblotting of RNA samples from two stages of T. brucei life cycle. Samples (from left to right): total RNA (Total), Poly(A)-enriched (A+) RNA and Poly(A)-depleted (A−) RNA from mammalian BSF and insect PCF. The last lane contains total mouse liver RNA (Mouse). 2 μg of total RNA, 2 μg of poly(A)-depleted RNA and 100 ng of poly(A)-enriched RNA was loaded per lane. rRNA was detected by staining RNA with methylene blue to confirm equal loading between total and poly(A)-depleted fractions. As expected rRNA is undetectable in the poly(A)-enriched fraction.

(see also Supplementary Figure S1 and Source Data of Extended Figures)

Extended Data Fig. 4. Poly(A) tail length, m⁶A and mRNA levels during VSG turnover.

a, Levels of m⁶A (immunoblot), length of VSG poly(A) tail (RNase H – northern blot) and levels of VSG mRNA (northern blot) after transcription halt by ActD. Signals were normalized to time point 0hr. The pattern observed is consistent with Figure 2b. Two-way ANOVA with sidak correction for multiple test. ( ****P<0.0001,*P=0.0190 in mRNA vs m6A in 15 min, ***P=0.0004 in poly(A) vs m6A in 15 min, ***P=0.0001 in mRNA vs m6A in 120 min,*P=0.0136 in poly(A) vs m6A in 240 min). n = 4 biological samples for mRNA and m⁶A levels, n = 3 biological samples for poly(A) tail length. Data are mean ± s.d.

(see also Source Data of Extended Figures)

Extended Data Fig. 5. Subcellular distribution of m⁶A in bloodstream form parasites.

a, Proportion of m⁶A signal in nucleus and cytoplasm. Data are mean ± s.d. n = 4 independent experiments. b, Quantification of mean fluorescence intensity (MFI) levels of m⁶A in five independent replicates in three different conditions: untreated BSF, nuclease P1 (NP1)-treated BSF, and actinomycin D (ActD)-treated BSF. Raw MFIs were obtained, the average of the untreated BSF equalled to 100%, and all other values normalized to 100%. Data are mean ± s.e.m. c, Distribution of VSG2 mRNA in the nucleus and cytoplasm in single marker (SM) cell line (single VSG expression) and in the clones that express a second reporter VSG (6 clones of DE1 express VSG117 containing a WT 16-mer motif, 7 clones of DE2 express VSG117 containing a mutagenized 16-mer motif). VSG mRNA was quantified by FISH. Nucleus was delimited by Hoechst staining. Total signal was set as 100% and the nucleus and cytoplasm represented as percentage of total signal. Error bars represent s.d. d, Microscopic observation of nuclei purified after fractionation protocol. Nuclei were stained with Hoechst. The nuclear purification was compared with initial lysates and with the cytoplasmic fraction. Scale bars, 10μm; N = 1 independent experiment. e, Western blot of subcellular fractions (total lysate, nuclear and cytoplasmic fractions) using antibodies against a nuclear protein (histone H2A; custom rabbit polyclonal 1:5000) and a cytoplasmic protein (β-tubulin; mouse monoclonal KMX-1 1:1000). For each sample, we loaded a protein equivalent to the same amount of cells. n = 3 independent experiments.

(see also Supplementary Figure S1 and Source Data of Extended Figures)

Extended Data Fig. 6. VSG double-expressor (DE) cell lines immunoblotting.

a, Overexposure of full immunoblot shown in Figure 4c. Three independent 16-mer^WT clones and three independent 16-mer^MUT are shown (C1-C6). Note that with this exposure, most intense bands are saturated. The purpose of this high exposure is to observe the region of blot corresponding to the VSG117 transcript. No VSG117 band is observed in the 16-mer^MUT clones. It is also possible to observe a weak VSG2 band in the VSG2 single expressor lane and in the 16-mer^MUT clones, which correspond to incomplete RNase H digestion of VSG2 transcript. N = 3 independent clones for each genotype (C1-C6). b, Schematics of VSG double-expressor (DE) cell-lines DE3 and DE4. VSG8 was inserted in the active bloodstream expression site, which naturally contains VSG2 at the telomeric end. In DE3, VSG8 contains its endogenous 3’UTR with the conserved 16-mer motif (sequence in blue). In DE4, the 16-mer motif of VSG8 was scrambled (sequence in orange). c, m⁶A immunoblot of mRNA from DE3 and DE4 cell-lines, in which CAF1 was further depleted by RNAi by adding Tetracycline (Tet). RNase H digestion of VSG2 mRNA was used to resolve VSG2 and VSG8 transcripts. Two independent DE3 clones and two independent DE4 clones are shown (C1-C4), each with (+) or without (−) CAF1 downregulation.

(see also Supplementary Figure S1)

Extended Data Fig. 7. CAF1 depletion.

CAF1 transcript levels measured by RT-qPCR in CAF1 RNAi cell-line used in Figure 5, Panels c-d and e-f. CAF1 downregulation was induced by adding tetracycline (Tet) to the medium. Unpaired two tailed t-test (**** P>0.0001). Data are mean ± s.d. n = 3 independent clones.

Extended Data Table 1.

Mass-spectrometry features of 34 nucleoside modifications found in T. brucei.

Base	Modification	Mass transition	Retention time (min)
Adenosine	m6A	282>150	9.90
	m1A	282>150	6.66
	Am	282>136	9.38
	g6A	369>237	8.47
	ct6A	395>263	6.90
	m6,6A	296>164	11.80
	hm6A	298>166	9.93
	ms2m6A	328>196	6.0
	i6A	336>204	16.63
	ms2t6A	459>327	13.94
	t6A	413>281	12.76
	hn6A or m6t6A	427>295	13.89
Guanosine	m1G	298>166	8.46
	m7G	298>166	7.45
	m2,2,7G	328>196	5.36
	m2G	298>166	9.58
	Gm	298>152	9.80
	m2,2G	312>180	10.70
	m2,7G	312>180	8.58
Cytidine	m4,4C	272>140	7.94
	m3C	258>126	6.42
	m4C	258>126	7.96
	m5C	258>126	6.76
	Cm	258>112	7.81
Uridine	se2U	306>174	8.45
	ncm5U	302>170	7.96
	mcm5U	317>185	7.95
	ncm5s2U or nchm5U	318>186	8.47
	nm5s2U	290>158	5.36
	Y	245>125	4.5
	acp3U	346>214	6.55
	cmnm5se2U	394>262	8.47
Other	OHyW	525>393	5.37
	o2yW	541>409	5.39

Extended Data Table 2.

List of oligonucleotides.

PURPOSE	SEQUENCE
AnTaT VSG qPCR	ACAACCACGGAAAGTGACG CACTTTTTGTCGCCATAAGC
VSG2qPCR	AGCAGCCAAGAGGTAACAGC CAACTGCAGCTTGCAAGGAA
VSG117 qPCR	AAGCGACAACAGATAAATGC CTTTGCAAGCATTATTTTCC
18S qPCR	ACGGAAT GGCACCACAAGAC GTCCGTTGACGGAATCAACC
16-mer mutagenesis	TTTGTTATACAAAACTTTTCAAAACCAGCCGAGATTTTGTG TTTGAAAAGTTTTGTATAACAAAAGTTTTCAAGTAGCAAGG
PAT adaptor	5’-Ph-CCAGTGAGCAGAGTGACGAGGACTCGAGCTCAAGC- 3ddA-3’
PAT Rev 1	GCTTGAGCTCGAGTCCTCG
PAT Rev 2	CGTCACTCTGCTCACTGG
PAT rev A0	GAGGACTCGAGCTCAAGCGCGTGTTAAAATATATCAG
PAT AnTaT 1	AATCCCCGAATTGTAAATGG
PAT AnTaT 2	TTTCTGCCGCATTTGTGG
PAT VSG117 1	AAGCGACAACAGATAAATGC
PAT VSG117 2	ATTCGCCCTCAGTGCTGC
AnTaT VSG RNase H A	TACTCGTCGTTGGCTGCTTG
AnTaT VSG RNase H B	TATTTTACTGCATAGGGCGT
AnTaT VSG RNase H C	GCGTGTTAAAATATATCAGA
Spliced leader RNase H	CAATATAGTACAGAAACTGT
β-tubulin RNase H	TACGGAGTCCATTGTACCTG
Oligo d(T)	TTTTTTTTTTTTTTTTTT
VSG2 RNase H	TCCGGCTGTTTCGTTTCT
Dot blot oligo A	ACTAGCTTAACTACGACCTCCTGAG
Dot blot oligo m6A	ACTAGCTTAACT-m6A-CGACCTCCTGAG
Dot blot oligo m1A	ACTAGCTTAACT-m1A-CGACCTCCTGAG GGAGAAAGAATAGTAACCCTTTCATCAAAGAAAATAGTCGAA
VSG8 cloning	GCTTATGCGAACAGCGAGCACAACC GAGGGGGGAAATTTGAGGGGGGAAAGGGCTGCAGGAATTC TTAAAAAAGCAAGGCCACAAATGC
VSG8qPCR	ATGGAAAGGAACCAACAACG CCCTCCTTTGTCTTATCTTTGC GGAGTTTGGTATAATCCGTTCG
CAF1qPCR	GCGTGGATTTGAAGTTACCG
PAT VSG2 1	AAGGTAGCAGATGAGACTGC
PAT VSG2 2	TTAGCAAGACCCCTCTTTGG
VSG RNase H northern	ACGCCCTATGCAGTAAAATA
VSG motif probe	GCGTGTTAAAATATATC

Extended Data Table 3.

Statistical parameters of time course experiments.

Figure	Measure	Experimental approach	Curve fit	Lag phase (XO) (min)	Decay constant (K) (min−1)	Half-life (min)	R2
2b	mRNA	Transcription inhibition (ActD)	Biphasic	60.0	0.019	36.9	0.91
2b	poly(A)	Transcription inhibition (ActD)	Biphasic	42.9	0.015	45.4	0.95
2b	m6A	Transcription inhibition (ActD)	Exponential	NA	0.019	37.4	0.93
2d	mRNA	Differentiation	Biphasic	53.5	0.012	58.7	0.88
2d	poly(A)	Differentiation	Exponential	NA	0.031	22.7	0.93
5d	mRNA	Transcription inhibition (ActD)	Exponential	NA	0.019	36.8	0.95
5b	poly(A)	Transcription inhibition (ActD)	Exponential	NA	0.074	9.3	0.98
S4	mRNA	Transcription inhibition (ActD)	Biphasic	32.7	0.009	76.3	0.85
S4	poly(A)	Transcription inhibition (ActD)	Biphasic	27.8	0.006	114.5	0.91
S4	m6A	Transcription inhibition (ActD)	Exponential	NA	0.026	26.2	0.97

Curves were fitted to the decay of VSG mRNA (pink), length of poly(A)-tail (dark blue) and m⁶A levels (light blue). Curves in which the measured variable decayed from T=0hr were called “One phase”. Those in which the measured variable decayed only after an initial constant period were called “Biphasic”.

Data Availability Statement

Datasets of RNA-seq of m⁶A-RIP experiments are deposited in SRA accession code: PRJNA786734. The following panels have associated raw figures: 1b, 1c, 1e, 1h, 2b, 2c, 3c, 3d, 3f, 3g, 3i, 3k, 4b, 4e, 4f, 5b, 5c, 5d, 5f, ext2d, ext2e, ext3c, ext4, ext5a, ext5b, ext5c, ext7. The following public databases were used; Trypanosome genome database (TryTrypDB): https://tritrypdb.org/tritrypdb/app and RNA modifications database (MODOMICS): http://genesilico.pl/modomics/. All data is available upon request.