Abstract
N6-Methyladenosine (m6A) is an abundant post-transcriptional RNA modification that influences multiple aspects of gene expression. In addition to recruiting proteins, m6A can modulate RNA function by destabilizing base pairing. Here, we show that when neighbored by a 5ʹ bulge, m6A stabilizes m6A–U base pairs, and global RNA structure by ~1 kcal mol−1. The bulge most likely provides the flexibility needed to allow optimal stacking between the methyl group and 3ʹ neighbor through a conformation that is stabilized by Mg2+. A bias toward this motif can help explain the global impact of methylation on RNA structure in transcriptome-wide studies. While m6A embedded in duplex RNA is poorly recognized by the YTH domain reader protein and m6A antibodies, both readily recognize m6A in this newly identified motif. The results uncover potentially abundant and functional m6A motifs that can modulate the epitranscriptomic structure landscape with important implications for the interpretation of transcriptome-wide data.
Similar content being viewed by others
Introduction
N6-methylated adenosine (m6A) is the most abundant internal RNA modification in eukaryotic mRNAs1,2,3 and long non-coding RNAs (lncRNA)4,5 and is also found in viral RNAs6,7,8. The modification is dynamically regulated by methyl transferase9,10,11 and demethylase enzymes12,13 that are linked to human disease14,15. It frequently occurs at the 3ʹ and 5ʹ untranslated regions of mRNA where it can influence RNA splicing16,17, mRNA nuclear exportation13,18, RNA decay19, and translation initiation20,21. The modification is implicated in a growing number of processes including stem cell fate determination, stress response, DNA damage repair, microRNA biogenesis, lncRNA-mediated transcription repression, viral infection, and in the mechanisms of cancer22,23,24,25.
m6A is thought to exert its biological effects through two general mechanisms. First, it can help recruit proteins that regulate mRNA fate by influencing splicing16, export26, decay19,27, and translation initiation efficiency20,27. Many of these m6A reader proteins contain YTH domains, which specifically recognize the methyl group5. Second, m6A can exert biological effects by modulating RNA structure. In particular, studies have shown that m6A destabilizes A–U base pairing and RNA duplexes by 0.5–1.7 kcal mol−1 28,29. This destabilization has been proposed to occur due to steric contacts between the N6-methyl group and the base28 (Fig. 1a). Studies have shown that m6A can promote RNA melting and thereby enhance binding of single-stranded RNA binding proteins30,31,32. The modification can also disrupt non-canonical A–G base pairs (bps) required for protein binding33. Presence of m6A in mRNA also impedes tRNA accommodation within the ribosome and translation-elongation most likely by destabilizing A–U bps34. Interestingly, while transcriptome-wide RNA structure mapping data indicate that m6A destabilizes RNA structure in vitro and in vivo28,35, the destabilization is not observed at the m6A nucleotide per se, but rather, at the immediate 5ʹ neighbor which favors single-stranded conformations in methylated RNA28. The mechanism by which m6A destabilizes its 5ʹ-neighbor is currently unknown.
Fewer studies have examined the potential for m6A to stabilize RNA even though there is precedent for this in the literature. Prior studies have shown that when placed at dangling ends of duplexes28 or in apical loops29, m6A stabilizes RNA by 0.1–1.2 kcal mol−1 most likely due to favorable stacking interactions between the methyl group and adjacent bases. Here, by attempting to harness the destabilizing effects of m6A in the design of an RNA secondary structural switch, we discovered a bulge motif that is stabilized by m6A in a Mg2+-dependent manner. This motif is recognized by the YTH domain and m6A antibodies and can potentially help explain the global impact of m6A on RNA structure observed in transcriptome-wide studies.
Results
m6A stabilizes a junctional A–U base pair
Our initial motivation was to examine whether the destabilizing effects of m6A on A–U base pairing could be harnessed to induce a change in RNA secondary structure. We designed an RNA hairpin (B5ʹ, Fig. 1b) containing the most common4,5 (GGm6ACU) m6A consensus sequence (DRACH, where D denotes A, G, or U; R is A or G; and H is A, C or U)36,37,38. This sequence is predicted39 to fold into two iso-energetic secondary structures in which the m6A forms an A–U bp either at the junction next to a bulged guanine or deeper within the upper helix (Fig. 1b). The sequence is also predicted to form an alternative conformation in which the m6A is bulged out and in which the 5ʹ neighboring guanine forms a non-canonical G–U wobble (Fig. 1b). This alternative conformation is predicted to be destabilized by 0.6 kcal mol−1, which is within the range of m6A destabilization of A–U bps in duplex RNA28,29. Thus, we reasoned that m6A could destabilize the junctional A–U bp and promote formation of the alternative conformation containing the G–U mismatch, which can be readily detected by 1D 1H NMR40. As a control, we also carried out experiments on a corresponding RNA duplex (denoted D), in which a cytosine nucleotide was included to pair with the bulged guanine (Fig. 1b). Experiments were carried out in the presence of 3 mM Mg2+.
The m6A substitution resulted in small perturbations in NMR spectra of the control duplex D (Fig. 1c). We observed the U17-H3 imino resonance of the m6A6 partner (Fig. 1c), which also has a nuclear Overhauser effect (NOE) cross peak with the amino group of m6A6 (Supplementary Fig. 1). This indicates that m6A6 forms a m6A6-U17 Watson-Crick bp stabilized by two H-bonds, as described previously for other RNA duplexes containing m6A28. Two NOE cross peaks with comparable intensities were observed between the m6A6 methyl proton and the base m6A6-H8 and m6A6-H2 (Supplementary Fig. 1). This indicates that the amino group either deviates from a perfectly anti conformation or rapidly (on the NMR chemical shift timescale) exchanges between the anti and syn conformations (see Supplementary Note).
The NMR spectra show that unmodified B5ʹ folds into a conformation with the predicted junctional A6-U17 Watson-Crick bp, bulged G5, and with G4 either forming a weak junctional bp or a bulge (Fig. 1d and Supplementary Fig. 2a, c). The m6A6 substitution did not lead to the predicted transition to the alternative conformation, as we did not observe the upfield shifted (10–12 ppm) imino resonances characteristic of G–U wobbles (Fig. 1d and Supplementary Fig. 2a). On the contrary, the substitution sharpened the U17 imino resonance, consistent with stabilization of the junctional Watson-Crick m6A6-U17 bp (Fig. 1d and Supplementary Fig. 2a). This together with uninterrupted NOE distance-based connectivity between U17 and G16 indicate that the methylation stabilizes the helical structure (Supplementary Fig. 3b). As in D, the NMR data indicate that while m6A6 forms a m6A6-U17 Watson-Crick bp, the methyl group deviates from a perfectly anti conformation (Supplementary Fig. 1 and see Supplementary Note). Therefore, counter to our predictions, and to the behavior observed in duplexes, m6A locally stabilizes the junctional A–U bp in B5ʹ.
m6A globally stabilizes B5ʹ in a Mg2+ dependent manner
While the NMR data indicate that the m6A modification locally stabilizes the junctional A–U bp, this provides little information regarding the effect of the modification on overall RNA stability. We therefore used UV melting experiments to examine the impact of the m6A6 on RNA stability. Consistent with prior studies28,29, m6A destabilized D by 0.6 ± 0.4 kcal mol−1 at 37 °C. In stark contrast, m6A stabilized B5ʹ by 0.9 ± 0.2 kcal mol−1 at the same temperature (Fig. 1e).
To further examine the basis for m6A-dependent stabilization, we carried out UV melting and NMR experiments in the absence of Mg2+. The modification destabilized D by a slightly greater amount (~0.8 kcal mol−1) in the absence of Mg2+ (Fig. 1e). This was accompanied by broadening of the imino resonances in the absence of Mg2+ (Supplementary Fig. 3d), consistent with the destabilization observed using UV. In contrast, in the case of B5ʹ, the degree of m6A stabilization decreased significantly from 0.9 ± 0.2 kcal mol−1 to 0.3 ± 0.3 kcal mol−1 in the absence of Mg2+ (Fig. 1e). NMR spectra, including data for site-specifically labeled B5ʹ (Supplementary Fig. 2a), indicate that in the absence of Mg2+, unmodified B5ʹ folds into many distinct and slowly (on the NMR chemical shift timescale) exchanging conformations. The U17 imino resonance in methylated B5ʹ is significantly broadened, consistent with local melting of the m6A6-U17 bp in the absence of Mg2+ (Supplementary Fig. 2a and Fig. 3d). Here, the NOE data indicate that the methyl group adopts an anti conformation (Supplementary Fig. 1) and that G5 forms a G5-U17 mismatch (Supplementary Fig. 2a). Thus, the addition of Mg2+ induces a conformational switch relative to the structure without Mg2+ by stabilizing the junctional m6A6-U17 bp with a 5ʹ neighboring G5 bulge.
Structural requirements for m6A stabilization
Previous studies have shown that when placed at dangling ends, m6A stabilizes RNA duplexes by ~0.1–1.2 kcal mol−128,29. This has been attributed to favorable stacking and hydrophobic shielding of the methyl group. Interestingly, this stabilization is greater when m6A is placed at the 5ʹ (~ 1.2 kcal mol−1) versus 3ʹ (~ 0.1–0.8 kcal mol−1)28,29 end of the duplex probably due to stronger stacking with the 3ʹ neighbor. Based on the thermodynamic parameters obtained from the UV experiments (Supplementary Table 1), the observed m6A-mediated stabilization of B5ʹ is driven by more favorable enthalpy, which is consistent with formation of favorable structural interactions. In particular, the 5ʹ neighboring guanine bulge could provide m6A enough “wiggle room” to form m6A–U WC bps in which m6A optimally stacks with the 3ʹ neighbor. This may require an unusual backbone conformation at the bulge, which is stabilized by Mg2+. Such conformations with optimal stacking may not be easily accommodated within the more rigid duplex interior, resulting in a net destabilization of the duplex.
To test the structural requirements for m6A stabilization, we varied the position and structural context of m6A (Fig. 2a) using three variants of the B5ʹ RNA (B5ʹhelical, B0, and B3ʹ). We also examined a model RNA (HCV) derived from a naturally occurring m6A site identified in the 3ʹ UTR of the hepatitis C virus (HCV) genome8. In HCV, the m6A site is at the junction but on the strand opposite to the bulge. We performed NMR analysis in the presence of at 25 mM NaCl and 3 mM Mg2+ and confirmed that for all variants the methylated and unmodified constructs fold into their predicted secondary structures (Fig. 2b and Supplementary Fig. 3c).
Decreasing the conformational freedom available to m6A by moving the m6A–U bp one base pair deeper into the upper helix (B5ʹhelical, Fig. 2a) decreased the m6A dependent stabilization from 0.9 ± 0.2 kcal mol−1 to 0.4 ± 0.4 kcal mol−1 (Fig. 2c). Decreasing the conformational freedom by placing m6A opposite rather than immediately adjacent to the bulge in HCV resulted in a net destabilization by 0.4 ± 0.4 kcal mol−1(Fig. 2c). NMR spectra show that in both cases, the modification does not locally stabilize the m6A–U bp again possibly due to the lack of conformational freedom needed for optimal stacking (Fig. 2b).
Disrupting stacking with the 3ʹ neighbor through placement of a bulge 3ʹ to m6A in B3ʹ also resulted in a net destabilization by ~1.2 kcal mol−1(Fig. 2c). Interestingly, although m6A globally destabilizes B3ʹ, it did locally stabilize the m6A6–U16 bp as evidenced by sharpening of the U16 imino resonance (Fig. 2b). Finally, placement of m6A in a bulge position destabilized B0 by ~0.7 ± 0.3 kcal mol−1(Fig. 2c). Here, the NMR spectra (Supplementary Fig. 1) indicate that the methyl group adopts a syn conformation, consistent with prior studies of single-stranded RNA41, and that the modification induces the flipping out of the bulge adenine (Supplementary Fig. 4b). Loss of intra-helical stacking with the adenine bulge could explain the m6A-induced destabilization of B0. This result shows that the destabilizing effects of m6A are not limited to bps in duplexes but can extend to bulges as well.
Similar results were obtained in the absence of Mg2+ with the exception of HCV for which the destabilization was more significant in the absence (1.4 ± 0.4 kcal mol−1) versus presence (0.4 ± 0.4 kcal mol−1) of Mg2+ (Fig. 2c). Together, these studies indicate that the observed m6A dependent stabilization requires a specific motif in which an m6A–U bp is neighbored by a 5ʹ guanine bulge.
Abundance of junctional m6A base pairs transcriptome-wide
Prior transcriptome-wide nuclease mapping studies of methylated RNA from human cells revealed a clear transition in the average RNA structure at the DRACH motif upon methylation28. In the methylated RNA, the purine nucleotide immediately 5ʹ to m6A had a much higher probability of being single-stranded whereas the methylated adenine had a slightly higher probability of being double-stranded28. The two nucleotides immediately 3ʹ to m6A showed a strong tendency to be paired. Similar results were observed with in vitro and in vivo SHAPE experiments35 although the m6A site showed slightly higher reactivity, indicating a greater tendency to be unpaired. These results indicate that methylation thermodynamically biases m6A in cellular mRNAs to be located between single-stranded unpaired RNA and adjacent to helices28.
Remarkably, the m6A stabilized B5ʹ motif identified in this work captures the unique conformational signatures induced by m6A in transcriptome-wide studies. The motif partially protects m6A by positioning it at a junction, exposes the 5ʹ guanine bulge, while the 3ʹ neighbors are helical. The abundance of the 5ʹB-[m6A–X] (where B is any bulge nucleotide and X is any nucleotide) motif could increase significantly on methylation because of two combined effects. On the one hand, m6A destabilizes bps in duplexes as well as bulges by as much as ~1.7 kcal mol−1. On the other hand, it can stabilize the 5ʹB-[m6A–U] motif by ~1 kcal mol−1. The combined contribution of ~3 kcal mol−1 could significantly bias the RNA folding landscape away from structures in which m6A is embedded within duplexes and bulges toward 5ʹB-[m6A–U] and perhaps other 5ʹB-[m6A–X] motifs and to a degree that help account for the observed changes in transcriptome-wide structure-mapping data. Such a strong bias is required to explain the changes in RNA conformation induced by methylation especially since m6A occurs sub-stoichiometrically in most RNAs5,42.
To test this hypothesis, we asked what fraction of the 140,574 m6A sites that have been mapped throughout the human transcriptome43 are predicted to fold into the 5ʹB-[m6A–U] or related 5ʹB-[m6A–X] motifs as the energetically most favorable secondary structure. We then asked what additional sites are predicted to fold into the 5ʹB-[m6A–U] or 5ʹB-[m6A–X] motifs as higher energy conformations that are within the ~3 kcal mol−1 energetic threshold. These conformations could be remodeled by m6A such to adopt the 5ʹB-[m6A–U] or 5ʹB-[m6A–X] motifs as the most energetically stable conformation. In particular, we used MC-Flashfold39 to predict secondary structures for 41 nt fragments chosen such that the m6A residue was positioned in the middle. Predicted RNA structures were classified according to the secondary structural context and position of m6A (Fig. 3a). As a control, secondary structures were also predicted for 140,574 randomly selected RNA sequences from the same transcriptome that do not contain the m6A consensus sequence.
Approximately 4 and 22% of the m6A sites are predicted to fold into the 5ʹB-[m6A–U] or 5ʹB-[m6A–X] motif respectively as the most energetically favorable secondary structure (Fig. 3b). Similar abundances were obtained for the control unmodified sequences (Fig. 3b), indicating that sequences selected for methylation do not have an enhanced propensity to fold into the 5ʹB-[A–X] motif relative to random sequences. Similar results were also obtained when using a different RNA structure prediction program (RNAstructure44) and when varying the position of m6A or length of the sequences subjected to structure prediction (Fig. 3b and Supplementary Fig. 5a).
A much larger fraction of m6A sequences (~ 42 and ~ 68% for 5ʹB-[m6A–U] and 5ʹB-[m6A–X], respectively) are predicted to fold into the junctional motif with energies < 3 kcal mol−1 relative to the lowest energy structure (Fig. 3c). Similar results were obtained when predicting structures using RNAstructure44 (Supplementary Fig. 5b). These higher energy structures could become the most stable structures upon methylation. While a similar increase in abundance is observed for the control unmodified sequences (Fig. 3c), indicating that the potential to form 5ʹB-[A–X] motif is not sequence dependent, these sequences are less likely to be methylated and to experience the energetic bias. Consequently only 20% of random A sites are expected to fold into 5ʹB-[A–X] motif as the minimum free energy structure. Therefore, while further experiments are needed to assess how m6A stabilizes / destabilizes various motifs in different sequence contexts, these results indicate that m6A could strongly bias RNA folding toward 5ʹB-[m6A–X] motifs, and thus help explain the m6A induced RNA conformational changes observed in transcriptome-wide structure mapping data.
m6A is recognized in 5ʹG-[m6A–U] but not in duplexes
To examine the functional significance of the 5ʹG-[m6A–U] motif, we examined whether it can be recognized by the YTH domain m6A reader (residues 380-579) from the human YTHDF2 (NP_057342.2) protein45. Binding was measured using a fluorescence polarization (FP) assay employing fluorescein tagged RNA46. Experiments were initially carried out in the absence of Mg2+ to allow comparison with previous studies45,47,48,49,50.
As a positive control, the YTH domain binds to single-stranded RNA (SS) containing m6A and the consensus sequence (GGm6ACU) with KD = 0.18 ± 0.02 µM (Fig. 4a) in good agreement with previously reported values (0.2-2 µM)45,47,48,49. The binding affinity is diminished at least ~ 100-fold (KD > 30 µM) for unmodified SS (Fig. 4a), in good agreement with 10–100-fold weaker affinity reported previously45,47,48,49.
Surprisingly, the YTH domain binds weakly (KD > 50 µM) to both unmodified and methylated duplex RNA (D) (Fig. 4b). In both cases, the affinity is comparable to that measured for unmodified SS (Fig. 4b). This indicates that m6A sites in helical m6A–U Watson-Crick bps are not accessible for recognition by the YTH domain. This result is significant considering that helical A–U Watson-Crick bps occur frequently in the structures of RNA.
In contrast, the YTH domain does bind tightly to methylated B5ʹ (but not its unmodified counterpart; KD > 40 µM) (Fig. 4c). The binding curve deviates from a simple two-state model and the data is better fit to a model that assumes two independent binding sites (Fig. 4d). A similar behavior was previously observed for YTH domain binding to methylated ssRNA45. One of two KDs obtained from this analysis (KD1 = 0.3 ± 0.1 µM) is consistent with high affinity binding as observed for methylated SS while the other (KD2 > 50 µM) is consistent with weak binding as observed with unmodified RNA. It is possible that the YTH domain specifically binds to m6A in B5ʹ with high affinity but also weakly and non-specifically binds to other parts of the RNA (e.g. duplex or apical loop region). Alternatively, the YTH domain could bind to m6A in two different RNA conformations (e.g. helical versus more single-stranded). These results indicate that in contrast to duplexes, m6A in 5ʹB-[m6A–U] is recognized by the YTH domain.
Based on the NMR data, Mg2+ stabilizes the junctional m6A–U bp in B5ʹ potentially making m6A6 less accessible for recognition. Indeed, the binding affinity to methylated B5ʹ decreased by ~6-fold (KD1 0.4 ± 0.1 µM versus 2.2 ± 0.6 µM, Fig. 4c) in the presence of 3 mM Mg2+ but remained ~10 fold tighter than that of unmodified SS or B5ʹ and ~ 20-fold tighter than methylated D. By comparison, the binding affinity of YTH domain to methylated or unmodified SS and D as well as unmodified B5ʹ was only slightly (by ~1-fold) weakened in the presence of Mg2+ (Fig. 4a, b).
We confirmed the above results for the full-length YTHDF2 protein using in vitro RNA pulldown experiments (Fig. 4e). Here, biotinylated RNA baits were incubated with HEK293T cell lysates, and RNA:protein complexes were purified with streptavidin agarose. The bound proteins were then eluted and subjected to SDS-PAGE, followed by western blot to detect YTHDF2. In agreement with results from the FP assay, in the absence of Mg2+, modified B5ʹ enriches YTHDF2 protein by an amount comparable to modified SS. While similar enrichments were observed in the absence and presence of Mg2+ for SS, the enrichment with B5ʹ decreased 1-fold in the presence of Mg2+. In sharp contrast, YTHDF2 was barely enriched by methylated D in the absence and presence of Mg2+.
Similar results were also obtained when examining how well m6A is recognized by three m6A antibodies (Abcam polyclonal antibody, SySy polyclonal antibody and NEB monoclonal antibody) that are used in transcriptome-wide studies. As expected, all unmodified RNA constructs were not well recognized by the antibodies in dot blot assays (Fig. 5). Both methylated SS and B0 exhibited strong signals, suggesting that the antibodies can recognize unpaired m6A (Fig. 5 and Supplementary Fig. 5c). Once again, the methylated D and methylated B5ʹhelical were poorly recognized by all three antibodies. This was also the case for a 9-mer duplex lacking the stabilizing UUCG apical loop (Supplementary Fig. 5c). In contrast, the junctional m6A–U bp in B5ʹ was well recognized by all three antibodies.
Taken together, these results indicate the m6A within the 5ʹG-[m6A–U] motif is well recognized by YTH reader proteins as well as commonly used m6A antibodies, whereas m6A–U bps embedded in duplexes are not.
Discussion
A growing number of studies indicate that m6A can exert biological effects by modulating RNA structure. So far, studies have primarily focused on the destabilization of base pairing due to steric collisions with the methyl group28,29,31,33. However, m6A has also been shown to stabilize duplexes when placed at dangling ends and within apical loops28,29. Our results identify a 5ʹ bulge motif that is predicted to be abundant and stabilized by m6A in a Mg2+ dependent manner. The 5ʹ bulge most likely provides m6A the conformational freedom needed to allow the methyl group to optimally stack with the 3ʹ neighbor while maintaining Watson-Crick m6A–U pairing. Mg2+ may help to stabilize the resulting backbone conformation at the bulge by neutralizing electrostatic repulsion. In this regard, it is noteworthy that Mg2+ preferentially binds to guanine and adenine nucleotides51, thus the naturally occurring DRACH motif may enhance the propensity for the 5ʹ neighbor to bulge out. This motif with base-paired m6A stacking on its 3ʹ neighbor and with a 5ʹ-neighboring bulged nucleotide can potentially form in a variety of contexts, including bulges, apical loops, and possibly other higher order structures. Further studies are needed to assess the robustness of the observed stabilization across these different structural and sequence contexts.
While biophysical studies show that m6A destabilizes pairing of adenine nucleotides in vitro, these results were at odds with in vitro transcriptome-wide nuclease mapping28 and SHAPE35 data showing that methylation primarily acts to increase the single-stranded character of the purine nucleotide immediately 5ʹ to the m6A site, while the adenine nucleotide itself was less affected28. These data were obtained devoid of cellular proteins and the conformational differences likely originate directly from the impact of m6A on RNA structure. A bias toward the junctional m6A bps flanked by a 5ʹ bulge following methylation could help explain the increase in single-stranded character observed specifically for the purine nucleotide 5ʹ to the m6A bulge as well as the increase in the double-stranded character observed for the m6A nucleotide using transcriptome-wide nuclease mapping experiments28. This bias could be significant especially when considering that m6A selectively stabilizes the 5ʹB-[A–U] motif but destabilizes other common motifs such as duplexes and bulges. It should be noted, however, that in vivo SHAPE data35 suggest that the impact of methylation on RNA structure can be more complex, possibly due to interactions with proteins.
Our results show that m6A is effectively recognized by both YTH domain reader proteins and m6A antibodies when located in the 5ʹG-[A–U] motif. The motif could therefore also serve functional roles recruiting YTH domains and is likely to be identified in current methods used to map m6A transcriptome-wide. In contrast, our results indicate that both YTH domain reader proteins and m6A antibodies do not recognize m6A–U bps embedded within duplexes. Current methods for m6A mapping attempt to remove these structural biases through nuclease digestion and heating of the RNA prior to immunoprecipitation. However, the RNA fragments are still 30–130 nt long43 and are subjected to cooling on ice following heating, allowing for re-annealing of the RNA. Indeed, we found that m6A sites buried in A–U bps within duplexes were poorly recognized by the antibodies even when increasing the temperature and/or duration of heating with or without 8 M urea. This raises the possibility that there are m6A sites buried within duplexes that have evaded detection using current transcriptome-wide methods. This could potentially confound the interpretation of m6A induced conformational changes based on transcriptome-wide structure mapping data since motifs in which m6A is accessible such as 5ʹG-[A–U] could be over-represented compared with those in which m6A is not accessible. Additional studies are needed to examine whether these structure-specific biases in m6A recognition also occur during immunoprecipitation experiments with free-floating RNA used in the transcriptome-wide mapping of m6A sites.
Our results indicate that while m6A destabilizes duplex A–U bps and disrupts base stacking in certain motifs, the structure landscape of the epitranscriptome is also likely punctuated by motifs such as 5ʹG-[m6A–U] that are stabilized by m6A. This motif presents a readable m6A site in a structured region of RNA. While the biological significance of this motif, including whether it is recognized by specific RNA-binding proteins, requires further investigation, our findings expand the mechanisms by which m6A can modulate RNA structure and should facilitate the interpretation of transcriptome-wide data.
Methods
Sample preparation
Modified (N6-methylated adenosine, 5ʹ-biotinylated RNA and 5ʹ-fluorescein labeled RNA) and unmodified RNA oligonucleotides were synthesized using a MerMade 6 Oligo Synthesizer employing 2′-tBDSilyl protected phosphoramidite (ChemGenes) on 1 μmol standard synthesis columns (1000 Å) (BioAutomation). n-acetyl protected rC, rA, and rG phosphoramidites were used to avoid incomplete deprotection when using isobutyryl protected rG and benzoyl protected rA phosphoramidites. m6A phosphoramidite was purchased from ChemeGenes, biotin phosphoramdites, and fluorescein phosphoramdites were purchased from Glen Research. The 15N1-labeled guanosine and the 15N3-labeled uridine phosphoramidites were synthesized according to a published procedure52. ssRNA oligonucleotides were synthesized with the option to leave the final 5ʹ-protecting group (4,4ʹ-dimethoxytrityl (DMT)) on for 2ʹO deprotection and cartridge purification. UUCG-capped hairpin RNAs were synthesized with DMT group off for DMT-off 2ʹO deprotection and PAGE purification, since regular 2ʹO deprotection method might cause incomplete deprotection for UUCG-capped samples. Synthesized oligonucleotides were cleaved from the 1 μmol column using 1 mL ammonia methylamine (1:1 ratio of 30% ammonium hydroxide and 30% methylamine) followed by 2-hour incubation at room temperature to allow base deprotection. The solution was then air-dried and dissolved in 115 μL DMSO, 60 μL TEA, and 75 uL TEA-3HF for regular 2ʹO deprotection, or in 100 μL DMSO and 125 μL TEA-3HF for DMT-off 2ʹO deprotection, followed by 2.5 h incubation at 65 °C. Regular 2ʹO deprotected samples were then quenched with Glen-Pak RNA quenching buffer and loaded onto Glen-Pak RNA cartridges (Glen Research Corporation) for purification using the online protocol (http://www.glenresearch.com/). Samples were then ethanol precipitated, air dried, dissolved in water and buffer exchanged or diluted into the desired buffer (15 mM sodium phosphate, 25 mM NaCl, 0.1 mM EDTA, 10% D2O, pH 6.4 with or without 3 mM Mg2+). DMT-off 2ʹO deprotected samples were directly ethanol precipitated, and purified using 20% (w/v) denaturing PAGE, and electroeluted into 20 mM Tris buffer, pH 8, and subsequently ethanol precipitated. The samples were then dissolved in water (50 µM for hairpin and 200–500 µM for duplex), annealed by heating at 95 °C for 10 min, and cooled on ice for 30 min (for hairpin) or at room temperature for 2 h (for duplex). RNA samples were buffer exchanged at least three times using a centrifugal concentrator (EMD Millipore) into the desired buffer. The purity of methylated and unmodified oligonucleotides was verified using NMR to monitor resonances from deprotecting groups and for B5ʹ using liquid chromatography-mass spectrometry LC/MS (Novatia).
NMR spectroscopy
All NMR experiments were collected on a 600 MHz Bruker NMR spectrometer equipped with an HCN cryogenic probe. Data were processed and analyzed using NMRpipe53 and SPARKY (T.D. Goddard and D.G. Kneller, SPARKY 3, University of California, San Francisco), respectively. Resonances were assigned using 2D HSQC, HMQC, 1H-1H NOESY (mixing times of 100 ms and 150 ms), and HCN experiments in the absence of Mg2+.
UV melting
Thermal melting experiments were conducted on a PerkinElmer Lambda 25 UV/VIS spectrometer with a RTP 6 Peltier Temperature Programmer and a PCB 1500 Water Peltier System. All RNA samples were buffer exchanged at least three times with a centrifugal concentrator (EMD Millipore) to desired buffers (15 mM sodium phosphate, 25 mM NaCl, 0.1 mM EDTA, pH 6.4 with or without 3 mM Mg2+), followed by direct dilution to 3 µM with the same buffer. At least three measurements were carried out for each RNA with a sample volume of 400 µL in a Teflon-stoppered 1 cm path length quartz cell. The absorbance at 260 nm was monitored while the temperature was varied between 15 and 95 °C.
Thermodynamic parameters from UV melting experiments were fitted using nonlinear model fitting in Mathematica 10.0 (Wolfram Research) such that melting temperature (Tm) and enthalpy (ΔH) for duplex and hairpin association were obtained by the fitting to Eqs. (1) and (2), respectively54,
where T is the temperature (K), R is the gas constant (kcal mol−1), and f is the fraction of folded duplex or hairpin for Eqs. (1) and (2), respectively. ΔG and ΔS were calculated from Eqs. (3) and (4) for duplex association and hairpin folding respectively.
where CT is the total concentration of RNA. The uncertainty in Tm, ΔH, ΔG, and ΔS was obtained based on the standard deviation in triplicate measurements. The destabilization or stabilization effects of m6A was calculated using the following equations:
Secondary structure prediction
The programs MC-Flashfold package39 and RNAstructure package44 were used to predict the secondary structure of 140,574 sequences containing m6A that were identified using single-nucleotide transcriptome-wide m6A mapping of human genome build (hg19)43. Another 140,574 control sequences that do not contain the DRACH motif and that were randomly selected from the same transcriptome were also analyzed. In-house Python scripts were used to perform the analysis. The sequences were 41 nt long with m6A positioned in the middle. Analysis was repeated when varying the position and length of the m6A site (31-nt long sequences with m6A site at the eleventh or twentyfirst position). The lowest energy predicted structures were classified into different categories: duplex, unpaired, 5ʹB-[A–U], 5ʹB-[A–Y] (Y denotes A or C or G), [A–U]-B3ʹ, [A–Y]-B3ʹ and 5ʹB-[X–A]/[X–A]-B3ʹ (X denotes A or C or G or U). 5ʹ-B-[A–U] and 5ʹB-[A–Y] are motifs containing junctional A–U or A–Y bp with unpaired residues at 5ʹ side of A. Similarly, [A–U]-B3ʹ, [A–Y]-B3ʹ are motifs containing junctional A–U or A–Y bps with unpaired residues at 3ʹ side of A. 5ʹB-[X–A]/[X–A]-B3ʹ is the motif with junctional A–X bps that are on the opposite strand of unpaired residues. With the exception of G–U bps, the RNAstructure package does not predict mismatches. To calculate the population of 5ʹB-[A–Y] and [A–Y]-B3ʹ, any A–Y site that was immediately adjacent to a canonical Watson-Crick bp was counted as a mismatch. The MC-Flashfold package does predict mismatches allowing further classification of the duplex category into duplex (A–U) and duplex (A–Y). MC-Flashfold was used to predict the secondary structures setting the maximum energy difference of output structures to minimum free energy (MFE) of 0, 1, 2 or 3 kcal mol−1. For RNAstrutcure package, the maximum percent energy difference between predicted structures and MFE was set to 10, 20, or 30%. The population of 5ʹB-[A–U] or 5ʹB-[A–X] was determined to be the number of sequences predicted to form 5ʹB-[A–U] or 5ʹB-[A–X] in the output structures divided by the total number of sequences (140,574).
Protein expression and purification
A gene, codon optimized (Supplementary Table 2) for Escherichia coli expression, encoding the YTH domain (residues 380–579) of human YTHDF2 (NP_057342.2) was subcloned into the pET15b vector (the plasmid containing the codon optimized sequence was purchased from GenScript). The plasmid was transformed into E. coli strain C41(DE3). The cells were grown in LB medium containing 50 µg ml−1 ampicillin at 37 °C to an OD600 between 0.4 and 0.6. Recombinant proteins were induced by adding isopropyl-β-d-thiogalactopyranoside (IPTG) to 0.5 mM and overexpressed in LB medium at 20 °C overnight. Cells expressing YTH domain were harvested at 4 °C and stored at −80 °C or lysed immediately using a microfluidizer in the buffer containing 20 mM HEPES (pH 7.4), 200 mM NaCl, 1 mM DTT supplemented with protease inhibitors and DNase. Following centrifugation at 17,500 rpm for 3 min at 4 °C, the supernatant was loaded to Ni-NTA affinity column. The eluted protein was further purified by size exchange chromatography (SEC) using a Superdex 75 pg column. The peak fractions were collected and concentrated. The protein purity was assessed using SDS-PAGE analysis and protein concentration determined using Bradford assays.
YTH domain binding assays
Binding experiments were carried out using a previously reported46 fluorescence polarization (FP) assay at 25 °C and using a PanVera Beacon 2000 instrument (Invitrogen, Madison, WI, USA). 5ʹ-fluorescein-labeled RNAs were synthesized and purified as described in the sample preparation section. Fluorescein labeled samples were dissolved in water and then directly diluted into the desired buffer (20 mM HEPES 50 mM NaCl, 3 mM DTT, pH 8.2, with or without 3 mM Mg2+). YTH domain protein was serially diluted into 200 µL of the same binding buffer containing 2 nM 5ʹ-fluorescein-labeled RNA. Fluorescence polarization was measured at an excitation wavelength of 490 nm and an emission wavelength at 530 nm. The total fluorescent intensities did not vary significantly throughout the measurements (i.e., (Imax − Imin)/Imin ~5%, in which Imax and Imin are the maximum and minimum intensities, respectively). The binding curves were fitted to either one-site (Eq. (8)) or two-site55 (Eq. (9)) binding mode using Mathematica 10.0 (Wolfram Research).
Lt is the total concentration of protein, Rt is the concentration of RNA. A and B represent free RNA anisotropy and RNA-protein intrinsic anisotropy of the saturated protein-RNA complex. KD is the dissociation constant. In Eq. (9), KD1 and KD2 are dissociation constants of the two binding events respectively. A, B, and C represent free RNA anisotropy, RNA-Protein and RNA-[Protein]2 intrinsic anisotropy of the saturated protein-RNA complexes. By titrating excess protein against 2 nM RNA, Lfree remains in excess over concentrations of RNA-Protein and RNA-[Protein]2 complexes. Lfree is therefore well approximated by Lt and is referred to simply as L.
In vitro RNA pulldown
Biotinylated RNA samples were synthesized and purified as described in the sample preparation section. HEK293T cells were harvested at ~70–80% confluency, washed with cold PBS and lysed by dounce homogenization in lysis buffer (10 mM NaCl, 2 mM EDTA, 0.5% Triton X-100, 0.5 mM DTT, 10 mM Tris, pH 7.4), and the lysate was then brought to 150 mM KCl and 5% Glycerol (v/v). Mammalian protease inhibitor cocktail (Sigma-Aldrich) and phosphatase inhibitor cocktail (Sigma-Aldrich) were freshly added to the lysis buffer. The following steps were performed as described previously5, except that binding buffers containing different Mg2+ concentrations were used (10 mM Tris pH 7.5, 150 mM KCl, 0.5 mM DTT, 0.05% (v/v) NP-40 and 0 mM, 1.5 mM, or 3 mM MgCl2). Proteins were eluted under mild conditions with elution buffer (50 mM Tris, pH 7.5, 200 mM NaCl, 2% SDS (w/v), and 1 mM biotin) at 60 °C, 1100 rpm on a thermal block shaker. The pulldown eluent was loaded on 4–12% polyacrylamide Bis-Tris polyacrylamide gels (Thermofisher) and then transferred onto PVDF membranes (GE Healthcare, Amersham) using a wet electrophoretic transfer system (BioRad). The membrane was then blocked with 5% nonfat dry milk in 0.1% PBST (0.1% Tween-20 in 1× PBS, pH 7.4). Rabbit polyclonal YTHDF2 antibody (Aviva system biology ARP67917_P050) was diluted 1:1000 in 0.1% PBST and incubated on the membrane for 1 h at room temperature (15–25 °C) or overnight at 4 °C. Membranes were then washed in 0.1% PBST, and then incubated with HRP-conjugated goat anti-rabbit IgG (Abcam ab6721) at 1:2500 dilution in 0.1% PBST for 1 h at 25 °C. Membranes were then subsequently washed with 0.1% PBST and exposed with enhanced chemiluminescence (ECL; GE Healthcare).
Dot blot assays
RNA samples were quantified using UV spectroscopy. Methylated and unmodified RNA samples were spotted onto a nylon membrane (GE healthcare). Fluorescein labeled RNA samples (prepared as described in sample preparation section) were scanned by UVP imaging system to visualize the quantity of RNA spotted onto the membrane. The membrane was then UV-crosslinked and blocked for 1 h in 5% nonfat dry milk in 0.1% PBST (0.1% Tween-20 in 1× PBS, pH 7.4). The following steps were performed essentially as described for RNA pulldown (above). m6A antibody (SySy 202003, NEB E1610S, or Abcam ab151230) was diluted 1:1000 in 0.1% PBST and incubated on the membrane for 1 h at room temperature or overnight at 4 °C. Membranes were then washed in 0.1% PBST, incubated with HRP-conjugated goat anti-rabbit IgG (Abcam ab6721) (1:2500) in 0.1% PBST for 1 h at room temperature. The membrane was then washed again in 0.1% PBST, and developed with enhanced chemiluminescence (ECL; GE Healthcare). The SS construct used in dot blot assay is 5ʹ-GGCGGm6ACUC-3ʹ.
Code availability
The in-house Python scripts used for secondary structure prediction are available on request to the corresponding author.
Data availability
The authors declare that the data supporting the findings of this study are available within the article and its Supplementary Information files, or are available upon reasonable requests to the authors.
References
Li, X., Xiong, X. & Yi, C. Epitranscriptome sequencing technologies: decoding RNA modifications. Nat. Methods 14, 23–31 (2016).
Rottman, F., Shatkin, A. J. & Perry, R. P. Sequences containing methylated nucleotides at the 5’ termini of messenger RNAs: possible implications for processing. Cell 3, 197–199 (1974).
Desrosiers, R., Friderici, K. & Rottman, F. Identification of methylated nucleosides in messenger RNA from Novikoff hepatoma cells. Proc. Natl Acad. Sci. USA 71, 3971–3975 (1974).
Meyer, K. D. et al. Comprehensive analysis of mRNA methylation reveals enrichment in 3’ UTRs and near stop codons. Cell 149, 1635–1646 (2012).
Dominissini, D. et al. Topology of the human and mouse m6A RNA methylomes revealed by m6A-seq. Nature 485, 201–206 (2012).
Lichinchi, G. et al. Dynamics of the human and viral m6A RNA methylomes during HIV-1 infection of T cells. Nat. Microbiol. 1, 16011 (2016).
Kennedy, E. M. et al. Posttranscriptional m6A editing of HIV-1 mRNAs enhances viral gene expression. Cell Host Microbe 19, 675–685 (2016).
Gokhale, N. S. et al. N 6-methyladenosine in flaviviridae viral RNA genomes regulates infection. Cell Host Microbe 20, 654–665 (2016).
Bokar, J. A., Rath-Shambaugh, M. E., Ludwiczak, R., Narayan, P. & Rottman, F. Characterization and partial purification of mRNA N 6-adenosine methyltransferase from HeLa cell nuclei. Internal mRNA methylation requires a multisubunit complex. J. Biol. Chem. 269, 17697–17704 (1994).
Liu, J. et al. A METTL3-METTL14 complex mediates mammalian nuclear RNA N6-adenosine methylation. Nat. Chem. Biol. 10, 93–95 (2014).
Wang, Y. et al. N 6-methyladenosine modification destabilizes developmental regulators in embryonic stem cells. Nat. Cell Biol. 16, 191–198 (2014).
Jia, G. et al. N 6-methyladenosine in nuclear RNA is a major substrate of the obesity-associated FTO. Nat. Chem. Biol. 7, 885–887 (2011).
Zheng, G. et al. ALKBH5 is a mammalian RNA demethylase that impacts RNA metabolism and mouse fertility. Mol. Cell 49, 18–29 (2013).
Lin, S., Choe, J., Du, P., Triboulet, R. & Gregory, R. I. The m6A methyltransferase METTL3 promotes translation in human cancer cells. Mol. Cell 62, 335–345 (2016).
Zhang, C. et al. Hypoxia induces the breast cancer stem cell phenotype by HIF-dependent and ALKBH5-mediated m6A-demethylation of NANOG mRNA. Proc. Natl Acad. Sci. USA 113, E2047–E2056 (2016).
Xiao, W. et al. Nuclear m6A reader YTHDC1 regulates mRNA splicing. Mol. Cell 61, 507–519 (2016).
Zhao, X. et al. FTO-dependent demethylation of N 6-methyladenosine regulates mRNA splicing and is required for adipogenesis. Cell Res. 24, 1403–1419 (2014).
Fustin, J. M. et al. RNA-methylation-dependent RNA processing controls the speed of the circadian clock. Cell 155, 793–806 (2013).
Wang, X. et al. N 6-methyladenosine-dependent regulation of messenger RNA stability. Nature 505, 117–120 (2014).
Wang, X. et al. N 6-methyladenosine modulates messenger RNA translation efficiency. Cell 161, 1388–1399 (2015).
Meyer, K. D. et al. 5’ UTR m6A promotes cap-independent translation. Cell 163, 999–1010 (2015).
Zhao, B. S., Roundtree, I. A. & He, C. Post-transcriptional gene regulation by mRNA modifications. Nat. Rev. Mol. Cell Biol. 18, 31–42 (2017).
Roundtree, I. A., Evans, M. E., Pan, T. & He, C. Dynamic RNA modifications in gene expression regulation. Cell 169, 1187–1200 (2017).
Meyer, K. D. & Jaffrey, S. R. Rethinking m6A readers, writers, and erasers. Annu. Rev. Cell. Dev. Biol. 33, 319–342 (2017).
Deng, X., Su, R., Feng, X., Wei, M. & Chen, J. Role of N 6-methyladenosine modification in cancer. Curr. Opin. Genet. Dev. 48, 1–7 (2017).
Roundtree, I. A. et al. YTHDC1 mediates nuclear export of N 6-methyladenosine methylated mRNAs. eL ife 6, e31311 (2017).
Shi, H. et al. YTHDF3 facilitates translation and decay of N 6-methyladenosine-modified RNA. Cell Res. 27, 315–328 (2017).
Roost, C. et al. Structure and thermodynamics of N 6-methyladenosine in RNA: a spring-loaded base modification. J. Am. Chem. Soc. 137, 2107–2115 (2015).
Kierzek, E. & Kierzek, R. The thermodynamic stability of RNA duplexes and hairpins containing N 6-alkyladenosines and 2-methylthio-N 6-alkyladenosines. Nucleic Acids Res 31, 4472–4480 (2003).
Zhou, K. I. et al. N 6-methyladenosine modification in a long noncoding RNA hairpin predisposes its conformation to protein binding. J. Mol. Biol. 428, 822–833 (2016).
Liu, N. et al. N 6-methyladenosine-dependent RNA structural switches regulate RNA-protein interactions. Nature 518, 560–564 (2015).
Liu, N. et al. N 6-methyladenosine alters RNA structure to regulate binding of a low-complexity protein. Nucleic Acids Res. 45, 6051–6063 (2017).
Huang, L., Ashraf, S., Wang, J. & Lilley, D. M. Control of box C/D snoRNP assembly by N 6-methylation of adenine. EMBO Rep. 18, 1631–1645 (2017).
Choi, J. et al. N 6-methyladenosine in mRNA disrupts tRNA selection and translation-elongation dynamics. Nat. Struct. Mol. Biol. 23, 110–115 (2016).
Spitale, R. C. et al. Structural imprints in vivo decode RNA regulatory mechanisms. Nature 519, 486–490 (2015).
Kane, S. E. & Beemon, K. Precise localization of m6A in Rous sarcoma virus RNA reveals clustering of methylation sites: implications for RNA processing. Mol. Cell Biol. 5, 2298–2306 (1985).
Narayan, P. & Rottman, F. M. An in vitro system for accurate methylation of internal adenosine residues in messenger RNA. Science 242, 1159–1162 (1988).
Csepany, T., Lin, A., Baldick, C. J. Jr. & Beemon, K. Sequence specificity of mRNA N 6-adenosine methyltransferase. J. Biol. Chem. 265, 20117–20122 (1990).
Parisien, M. & Major, F. The MC-Fold and MC-Sym pipeline infers RNA structure from sequence data. Nature 452, 51–55 (2008).
Geerdes, H. A. & Hilbers, C. W. Ring current shifts in GU base pairs. FEBS Lett. 107, 125–128 (1979).
Engel, J. D. & von Hippel, P. H. Effects of methylation on the stability of nucleic acid conformations: studies at the monomer level. Biochemistry 13, 4143–4158 (1974).
Horowitz, S., Horowitz, A., Nilsen, T. W., Munns, T. W. & Rottman, F. M. Mapping of N 6-methyladenosine residues in bovine prolactin mRNA. Proc. Natl Acad. Sci. USA 81, 5667–5671 (1984).
Linder, B. et al. Single-nucleotide-resolution mapping of m6A and m6Am throughout the transcriptome. Nat. Methods 12, 767–772 (2015).
Reuter, J. S. & Mathews, D. H. RNAstructure: software for RNA secondary structure prediction and analysis. BMC Bioinforma. 11, 129 (2010).
Zhu, T. et al. Crystal structure of the YTH domain of YTHDF2 reveals mechanism for recognition of N 6-methyladenosine. Cell Res. 24, 1493–1496 (2014).
Rossi, A. M. & Taylor, C. W. Analysis of protein-ligand interactions by fluorescence polarization. Nat. Protoc. 6, 365–387 (2011).
Xu, C. et al. Structural basis for selective binding of m6A RNA by the YTHDC1 YTH domain. Nat. Chem. Biol. 10, 927–929 (2014).
Luo, S. & Tong, L. Molecular basis for the recognition of methylated adenines in RNA by the eukaryotic YTH domain. Proc. Natl Acad. Sci. USA 111, 13834–13839 (2014).
Li, F., Zhao, D., Wu, J. & Shi, Y. Structure of the YTH domain of human YTHDF2 in complex with an m6A mononucleotide reveals an aromatic cage for m6A recognition. Cell Res. 24, 1490–1492 (2014).
Zhang, Z. et al. The YTH domain is a novel RNA binding domain. J. Biol. Chem. 285, 14701–14710 (2010).
Zheng, H., Shabalin, I. G., Handing, K. B., Bujnicki, J. M. & Minor, W. Magnesium-binding architectures in RNA crystal structures: validation, binding preferences, classification and motif detection. Nucleic Acids Res. 43, 3789–3801 (2015).
Neuner, S., Santner, T., Kreutz, C. & Micura, R. The “speedy” synthesis of atom-specific 15N imino/amido-labeled RNA. Chemistry 21, 11634–11643 (2015).
Delaglio, F. et al. NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293 (1995).
Bloomfield, V. A., Crothers, D. M. & Tinoco, I. Nucleic Acids: Structures, Properties, and Functions (University Science Books, Sausalito, CA, 2000).
Wilson, G. M., Sun, Y., Lu, H. & Brewer, G. Assembly of AUF1 oligomers on U-rich RNA targets by sequential dimer association. J. Biol. Chem. 274, 33374–33381 (1999).
Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH image to imageJ: 25 years of image analysis. Nat. Methods 9, 671–675 (2012).
Acknowledgements
We thank Nicole Orlovsky, Atul Rangadurai, and Mary Clay, members of the Al-Hashimi laboratory, and Dr. Tony Mustoe (UNC-Chapel Hill) for assistance and critical comments on the manuscript. We thank Dr. Richard Brennan for providing the UV-Vis spectrophotometer. We thank New England Biolabs (NEB) for providing the monoclonal m6A antibody as a gift to S.M.H. We acknowledge technical support and resources from the Duke Magnetic Resonance Spectroscopy Center. This work was supported by the US National Institutes of Health (P01GM0066275 to H.M.A., R01AI125416 to S.M.H. and R00MH104712 to K.D.M.) and Burroughs Welcome Fund (to S.M.H.). The research reported in this article was performed by the Duke University faculty, research associate and was funded by US National Institute of Health contract to H.M.A.
Author information
Authors and Affiliations
Contributions
B.L., D.K.M. and H.M.A. conceived the project and experimental design. B.L. and D.K.M. prepared NMR samples and performed NMR experiments and analyzed NMR data. R.P. and C.K. prepared 15N-N1/3 labeled Guanosine and Uridine phosphoramidite. B.L. performed UV melting experiments, transcriptome-wide RNA secondary structure prediction, and dot blot assays. B.L. performed the YTH domain protein binding assays, with assistance from M.A.S. B.L. and S.H.C performed the in vitro RNA pulldown experiments. H.M.A. and B.L. wrote the manuscript with critical input from D.K.M., K.D.M., M.A.S., S.M.H., and S.H.C.
Corresponding author
Ethics declarations
Competing interests
H.M.A. is an advisor to and holds an ownership interest in Nymirum, an RNA-based drug discovery company. The remaining authors declare no competing interests.
Additional information
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Liu, B., Merriman, D.K., Choi, S.H. et al. A potentially abundant junctional RNA motif stabilized by m6A and Mg2+. Nat Commun 9, 2761 (2018). https://doi.org/10.1038/s41467-018-05243-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-018-05243-z
This article is cited by
-
Dynamic regulation and functions of mRNA m6A modification
Cancer Cell International (2022)
-
Secondary structure prediction for RNA sequences including N6-methyladenosine
Nature Communications (2022)
-
A quantitative model predicts how m6A reshapes the kinetic landscape of nucleic acid hybridization and conformational transitions
Nature Communications (2021)
-
Limits in the detection of m6A changes using MeRIP/m6A-seq
Scientific Reports (2020)