N ⁶-Hydroperoxymethyladenosine: a new intermediate of chemical oxidation of N⁶-methyladenosine mediated by bicarbonate-activated hydrogen peroxide

Abstract

N ⁶-Methyladenosine (m⁶A) represents a relatively abundant modification in eukaryotic RNA. Because m⁶A has similar properties to adenosine and a low reactivity, limited research has been focused on this nucleoside. In this study, we revealed an important intermediate in the oxidation of m⁶A through the bicarbonate-activated peroxide system. Over the course of oxidation, we found a new mechanism in which N⁶-hydroxymethyladenosine (hm⁶A), N⁶-formyladenosine (f⁶A) and N⁶-hydroperoxymethyladenosine (oxm⁶A) were intermediate products, and adenosine was the final product. In this study, oxm⁶A was isolated using HPLC and characterized by mass spectrometry, NMR and diphenyl-1-pyrenylphosphine (DPPP) fluorescence detection. This study provides a new modified nucleoside and demonstrates oxidative demethylation of m⁶A by reactive oxygen species at the nucleobase level and in RNA strands.

---

N ⁶-Methyladenosine represents the most abundant modification in the mRNA of higher eukaryotes, present at a frequency of approximately three sites on each mRNA.¹ m⁶A is also present on tRNA, rRNA and lnRNA.² This modification plays an important role in the regulation of gene expression.³ Since its discovery last century,⁴ m⁶A has been the object of relatively few studies. Recently, fat mass- and obesity-associated proteins (FTO)⁵ and AlkBH5 ⁶ were found to be m⁶A demethylases, indicating a novel regulatory mechanism in mammalian cells. Two new modifications, N⁶-hydroxymethyladenosine (hm⁶A) and N⁶-formyladenosine (f⁶A), have been found to participate in the FTO-mediated demethylation process, which may influence RNA–protein interactions and regulate gene expression.⁷ In addition, transcriptome-wide profiling of m⁶A in mRNA and lnRNA has revealed new insights into the role of RNA modification.⁸ These developments have renewed interest in the investigation of this particular, distinctive modification. Therefore, we aspire to use a chemical method to differentiate m⁶A from A.

Hydrogen peroxide is a widely used oxidant with a high content of active oxygen,⁹ but its relatively slow oxidizing rate limits its usage. Bicarbonate is present in cells and serum at high concentrations, ranging from 14.7–25 mM,¹⁰ and plays an important role in biological oxidation.¹¹ H₂O₂ and NH₄HCO₃ are environmentally friendly reagents; H₂O₂ produces only water as a by-product, and NH₄HCO₃ easily decomposes to NH₃, CO₂, and H₂O. The reaction conditions are mild at natural pH values.

Owing to its high reactivity towards secondary amines, we considered whether the oxidant could react with m⁶A. Surprisingly, instead of producing N-oxides, demethylated adenosine was produced, and the presence of several intermediates in the reaction system suggested a potential mechanism in the chemical reaction (Scheme 1). These results suggest that H₂O₂/bicarbonate can act as a reactive oxygen species (ROS) for demethylation. In this study, we determine a key intermediate in the demethylation process, and we investigate the underlying mechanism.

Scheme 1 Proposed mechanism of the demethylation process, and structures of oxidation products.

To investigate the demethylation process, we used high-performance liquid chromatography (HPLC) to monitor the reaction (UV detector at 260 nm). When a 2 mM aliquot of m⁶A was treated with 200 mM H₂O₂ and 1 M NH₄HCO₃ at 37 °C for one hour, four products were formed: A, hm⁶A, oxm⁶A and f⁶A (Fig. 1). The LC-MS data showed masses corresponding to A (267.9), hm⁶A (297.8), oxm⁶A (313.8) and f⁶A (295.9), successively in the positive-ion mode (Fig. S1†). Product A was further characterized by ¹H and ¹³C NMR (see ESI†). To confirm the occurrence of hm⁶A and f⁶A, these compounds were synthesized according to reported procedures.⁷ An equilibrium reaction between adenosine and formaldehyde produced hm⁶A (Scheme 1, Route 9). Further HPLC analysis indicated that the synthesized hm⁶A and f⁶A have the same retention times as the reported hm⁶A and f⁶A, respectively (Fig. 1b–d). We found that hm⁶A and f⁶A were unstable and could decompose to A (adenosine) during HPLC analysis (Fig. 1c and d). N⁶-Hydroperoxymethyladenosine (oxm⁶A) was found to be a new intermediate, in addition to hm⁶A and f⁶A, during the demethylation of m⁶A (Fig. 1b). When we incubated the m⁶A with bicarbonate or H₂O₂ alone, no reaction was observed (Fig. S2 and S3†).

Fig. 1 HPLC chromatograph of 2 mM m⁶A incubated with 200 mM H₂O₂ and 1 M NH₄HCO₃ at 37 °C for 0 h (a), 1 h (b) and 24 h (e). As shown in the HPLC profiles, when the reaction proceeded for 1 h, one major product (A) was produced, accompanied by three intermediates (hm⁶A, f⁶A, oxm⁶A). The synthesized hm⁶A (c) and f⁶A (d) standards have the same retention time as two of the new peaks in the reaction mixture. Because hm⁶A and f⁶A are unstable, they can coexist with A during HPLC analysis.

Diphenyl-1-pyrenylphosphine (DPPP), as a fluorescent reagent, can be used for hydroperoxide determinations.¹² When we incubated the intermediate with DPPP in the presence of butylated hydroxytoluene (BHT) at 37 °C for 1 h, the fluorescence increased, indicating the formation of a hydroperoxide intermediate (Fig. 2). Further characterization of oxm⁶A was achieved using high-resolution mass spectrometry, ¹H NMR, ¹³C NMR and TOCSY (ESI, Fig. 3, S4 and S5†), with the corresponding chemical structures shown in Scheme 1. To confirm the chemical shifts of the protons in N–H and OO–H, ¹H NMR was performed in DMSO-d₆ and in D₂O. In the DMSO-d₆ solution, the chemical shifts of the protons were 8.61 ppm (–N–H) and 11.71 ppm (–OOH) (Table 1). To confirm our hypothesis, we changed the solution to D₂O, where deuterons can be incorporated at the N–H and O–H positions because of hydrogen–deuterium (H/D) exchange behavior. As we expected, these two protons disappeared in the D₂O solution (Fig. 3). We then used total correlation spectroscopy (TOCSY) to show the H–H correlation; the TOCSY spectrum was acquired using a 600 MHz Bruker Avance II spectrometer equipped with a 5 mm triple resonance cryoprobe. The pulse sequence was DIPSI2ETGP. The relaxation delay was 1 s, with 8 acquisitions per increment, and a spectral width of 8 × 8 ppm and time domain of 2k × 176 were used. In the spectrum, the NH proton had a cross peak with CH₂ at δ (8.66, 5.26 ppm), further confirming the oxm⁶A structure. When we analysed the reaction mixture using LC-MS, we detected a relatively small mass signal of 311.8; this finding may indicate the generation of another intermediate, N⁶-carboxyladenosine, in a relatively low yield (Fig. S1c†). Meanwhile, our control experiments indicated that adenosine, uridine, cytidine and guanosine were stable in the H₂O₂/bicarbonate solution at concentrations of 200 mM H₂O₂ and 1 M NH₄HCO₃ (Fig. S6†) after one hour.

Fig. 2 Fluorescence emission spectra (λ_ex = 352 nm) of DPPP in the presence of (a) and in the absence of oxm⁶A (b) after incubation with BHT at 37 °C for 60 min.

Fig. 3 ¹H NMR spectrum (a) and TOCSY spectrum (b) of oxm⁶A in DMSO-d₆.

Table 1 ¹H chemical shifts (δ, ppm) of oxm⁶A in DMSO-d₆ at room temperature

Number	Chemical shift (δ, ppm)	Number	Chemical shift (δ, ppm)	Number	Chemical shift (δ, ppm)
1	5.33	6	5.91	11	8.66
2	3.68	7	5.23	12	5.26
3	3.96	8	5.50	13	11.72
4	4.14	9	8.46
5	4.61	10	8.31

---

Because our goal was to fully investigate the mechanism of m⁶A demethylation, we extended the reaction time to 24 hours. After 24 hours, we found that only A (primary product) and a small amount of oxm⁶A were present (Fig. 1e), whereas hm⁶A and f⁶A disappeared. This result suggested that hm⁶A and f⁶A were converted into A (Scheme 1, Routes 9 and 10).

To investigate the behaviour of oxm⁶A, it was separated from the reaction mixture, incubated in HEPES buffer (50 mM, pH 7.4) at 37 °C and then analysed by HPLC every 2 h. We found that the amount of A increased at the expense of oxm⁶A (Fig. S7 in the ESI†), and it had a half-time of approximately 8.5 h (Fig. S8†), which was more stable than hm⁶A and f⁶A (approximately 3 h).

In the H₂O₂/NH₄HCO₃ system, the hydroxyl radical was trapped by 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) to give a signal using Electron Paramagnetic Resonance (EPR) (Fig. S9†). In the reaction system, the addition of DMSO, a hydroxyl radical scavenger, dramatically decreased the chemical demethylation level of m⁶A (Fig. S10†). We speculate that the reaction underwent a hydroxyl radical mechanism. A hydroxyl radical abstracted a hydrogen atom from a methyl group to yield a carbon radical, which could then bind with O₂ to form oxm⁶A (Scheme 1, Routes 2 and 3) or bind with ˙OH to form hm⁶A (Scheme 1, Routes 4 and 8), parallel to the decomposition mechanism for 5′-hydroperoxymethyluracil and 5′-hydroperoxymethylcytosine, as proposed by Richard Wagner's group.¹³ To confirm the possibility of the ˙OH radical mechanism, we used Fenton-type reagents to react with m⁶A. The formation of hm⁶A, oxm⁶A, f⁶A and A was also observed using LC-MS analysis, confirming the reaction mechanism (Fig. S18†). Under identical experimental conditions but with the addition of a small amount of (NH₄)₂Fe(SO₄)₂ in the H₂O₂/NH₄HCO₃ reaction mixture, the reaction rate markedly increased (Fig. S11†). As the reaction is based on the hydroxyl radical mechanism, and Fe²⁺ as well as Cu²⁺ have great influences on the reaction, we therefore used Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES) to investigate the presence of iron(II) and copper(II) in the H₂O₂/bicarbonate reaction system. No signals were observed, and both the concentration of Fe²⁺ and Cu²⁺ were lower than 10 ng mL⁻¹, indicating the reaction could proceed with just a bicarbonate-activated peroxide system (the optimized operating conditions are shown in Table S1†). In the demethylation process, two pathways are shown. A hydroxyl radical attacks the methyl radical to form hm⁶A (Scheme 1, Routes 4 and 8) and O₂ attacks the methyl radical to form oxm⁶A (Scheme 1, Routes 2 and 3). The oxm⁶A and its peroxide radical can decompose to hm⁶A (Scheme 1, Routes 5, 8 and 6) and f⁶A (Scheme 1, Route 7), and we propose that the new route in the demethylation process would improve the efficiency of the demethylation reaction compared to just attacking the methyl radical by a hydroxyl radical.

Next, because m⁶A is preferentially present in the consensus sequence RRm⁶ACH (R is A/G and H is A/C/U),¹⁴ to examine whether the reaction occurs in RNA oligos, we prepared a 9-mer oligoribonucleotide (5′-CUGGm⁶ACUGG-3′) containing one m⁶A site and treated it with 10 mM H₂O₂ and 100 mM bicarbonate at 37 °C for 48 h. Because RNA may decompose in a high concentration of H₂O₂, we decreased the concentration of H₂O₂ and NH₄HCO₃. After the reaction, the oligo RNA was analysed using MALDI-TOF mass spectrometry as shown in Fig. S12.† We found a m⁶A −14 Da peak, representing a demethylation product, as well as a +14 Da peak and a +17 Da peak, which may correspond to N⁶-formyladenosine and N⁶-hydroxymethyladenosine intermediates in the demethylation pathway, respectively. At natural pH levels (pH 7.4), hm⁶A, oxm⁶A and f⁶A were relatively stable, but an alkaline phosphate digestion may accelerate their decomposition. Therefore, to verify the presence of hm⁶A, oxm⁶A and f⁶A in the oligo RNA after the reaction, we used RNase T1 followed by nuclease P1 to digest the oligo RNA,⁷ then analysed the reaction using LC-MS. In this analysis, RNase T1 can selectively digest the phosphodiester bond after G. We successfully detected the formation of A, hm⁶A, oxm⁶A, and f⁶A in the digested nucleoside, similar to our proposed mechanism for a single nucleoside (Fig. S13†).

To explore the reaction kinetics of the oxidation, two micrograms of oligo RNA were incubated with 100 μM H₂O₂ and 300 μM NH₄HCO₃ at 37 °C for 30 h in six parallel experiments, followed by digestion with nuclease P1 and alkaline phosphate. The amount of A generated from m⁶A was quantified using LC-MS every 3 hours (the calibration curve is shown in the ESI, Fig. S14†). As depicted in Fig. S15,† the A content exhibited a strong linear relationship with reaction time over a period of 30 hours. After adding Fe²⁺ to the H₂O₂/NH₄HCO₃ mixture and incubating it with oligo RNA, HPLC analysis of the enzymatically digested nucleosides in RNA showed the presence of demethylated adenosine with a decreased level of m⁶A after oxidation for 1 h (Fig. S16†).

Although FTO-mediated oxidation of m⁶A may decrease the level of m⁶A in vitro, no in vitro experiments have been reported in which a chemical reagent was used to demethylate m⁶A. We explored whether m⁶A in genomic RNA is a substrate of H₂O₂/NH₄HCO₃in vitro. Total RNA was extracted from Hela cells using the TRIzol reagent (Invitrogen) according to the manufacturer's protocol. Four micrograms of genomic RNA was incubated with 100 μM H₂O₂ and 1 mM NH₄HCO₃ at 37 °C for 12 h. After digestion with nuclease P1 and alkaline phosphatase, the solution was analysed by LC-MS. The results showed a decrease in the m⁶A/A ratio by 10% in the genomic RNA (Fig. S17†), indicating that the reagents demethylated m⁶A in vitro.

Conclusions

In conclusion, we reported a new chemical method for the oxidative demethylation of m⁶A and determined an important intermediate in the reaction system. Three intermediates, N⁶-hydroxymethyladenosine (hm⁶A), N⁶-formyladenosine (f⁶A), and N⁶-hydroperoxymethyladenosine (oxm⁶A), were characterized, and the mechanism underlying the decomposition was illustrated. We also determined that the reaction could occur in oligo RNA and genomic RNA in vitro. H₂O₂ is a reactive oxygen species that is endogenously produced during normal metabolism¹⁵ and immune responses,¹⁶ and a high concentration of bicarbonate is found in cells and serum. Thus, this route may occur in vivo and play a role in cells. ROS have been proven to directly react with genomic DNA in a chemical reaction.¹⁷ Recently, reports have shown that ROS can induce the oxidative conversion of 5mC to 5hmC in a TET dioxygenase-dependent manner,¹⁸ indicating ROS regulate the enzymatic catalytic reaction. We propose that the oxm⁶A was formed through direct oxidation by ROS in vivo, just like the nucleoside analogues formed in RNA induced by Fenton-type reagents.^17b Further study is in progress to study the presence and biological function of oxm⁶A in vivo. The discovery of the new intermediate oxm⁶A and the chemical route for the demethylation of m⁶A to A may offer new insight into the study of m⁶A.

Acknowledgements

This work was supported by the National Basic Research Program of China (973 Program) (2012CB720600, 2012CB720603), the National Science Foundation of China (nos 91213302, 21432008, 91413109) and the National Grand Program on Key Infectious Disease (2012ZX10003002-014). The oligo RNA was kindly offered by He's Group at the University of Chicago. The TOCSY spectrum was collected by Prof. Maili Liu and Dr Xiongjie Xiao at Wuhan Institute of Physics and Mathematics. We thank Prof. Bin Hu and Dr Man He for the ICP-OES service.

Notes and references

1. J. A. Bokar, The Biosynthesis and Functional Roles of Methylated Nucleosides in Eukaryotic mRNA, In Fine-tuning of RNA Functions by Modification and Editing, H. Grosjean, Topics in Current Genetics 12, Springer-Verlag, Berlin Heidelberg, 2005, pp. 141–177.
2. (a) M. Saneyoshi, F. Harada and S. Nishimura, Biochim. Biophys. Acta, 1969, 190, 264–273; (b) Y. lwanami and G. Brown, Arch. Biochem. Biophys., 1968, 126, 8–15.
3. (a) T. Pan, Trends Biochem. Sci., 2013, 11, 8–17; (b) G. Jia, Y. Fu and C. He, Trends Genet., 2013, 29, 108–115; (c) K. D. Meyer and S. R. Jaffrey, Mol. Cell. Biol., 2014, 15, 313–326; (d) Y. Fu, D. Dominissini, G. Rechavi and C. He, Nat. Rev. Genet., 2014, 15, 293–306.
4. (a) J. W. Littlefield and D. B. Dunn, Biochem. J., 1958, 70(4), 642; (b) R. H. Hall, Biochemistry, 1965, 4, 661–670; (c) P. Brookes and P. D. Lawley, J. Chem. Soc., 1962, 254, 1348–1351; (d) R. Desrosiers, K. Friderici and F. Rottman, Proc. Natl. Acad. Sci. U. S. A., 1974, 71, 3971–3975.
5. G. Jia, Y. Fu, X. Zhao, Q. Dai, G. Zheng, Y. Yang, C. Yi, T. Lindahl, T. Pan, Y. G. Yang and C. He, Nat. Chem. Biol., 2011, 7, 885–887.
6. G. Zhang, J. A. Dahl, Y. Niu, P. Fedorcsak, C. M. Huang, C. J. Li, C. B. Vagbo, Y. Shi, W. L. Wang, S. H. Song, Z. Lu, R. P. G. Bosmans, Q. Dai, Y. J. Hao, X. Yang, W. M. Zhao, W. M. Tong, X. J. Wang, F. Bogdan, K. Furu, Y. Fu, G. Jia, X. Zhao, J. Liu, H. E. Krokan, A. Klungland, Y. G. Yang and C. He, Mol. Cell, 2013, 49, 18–29.
7. Y. Fu, G. Jia, X. Pang, R. N. Wang, X. Wang, C. J. Li, S. Smemo, Q. Dai, K. A. Bailey, M. A. Nobrega, K. L. Han, Q. Cui and C. He, Nat. Commun., 2013, 4, 1798–1806.
8. (a) D. Dominissini, S. Moshitch-Moshkovitz, S. Schwartz, M. S. Divon, L. Ungar, S. Osenberg, K. Cesarkas, J. J. Hirsch, N. Amariglio, M. Kupiec, R. Sorek and G. Rechavi, Nature, 2012, 485, 201–206; (b) K. D. Meyer, Y. Saletore, P. Zumbo, O. Elemento, C. E. Mason and S. R. Jaffrey, Cell, 2012, 149, 1635–1646.
9. R. A. Sheldon, Top. Curr. Chem., 1993, 164, 21–34.
10. H. Arai, B. S. Berlett, P. B. Chock and E. R. Stadtman, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 10472–10477.
11. Y. Sun, H. Song, J. Li, M. Jiang, Y. Li, J. Zhou and Z. Guo, Biochemistry, 2012, 51, 4580–4589.
12. H. Meguro, K. Akasaka, H. Ohrui and A. D. Becke, Methods Enzymol., 1990, 186, 157–161.
13. G. S. Madugundu, J. Cadet and J. R. Wagner, Nucleic Acids Res., 2014, 42, 7450–7460.
14. (a) U. Schibler, D. E. Kelley and R. P. Perry, J. Mol. Biol., 1977, 115, 695–714; (b) C. M. Wei and B. Moss, Biochemistry, 1977, 16, 1672–1676.
15. B. Chance, H. Sies and A. Boveris, Physiol. Rev., 1979, 59, 527–605.
16. B. M. Babior and R. C. Woodman, Semin. Hematol., 1990, 27, 247–259.
17. (a) C. Bienvenu, J. R. Wagner and J. Cadet, J. Am. Chem. Soc., 1996, 118, 11406–11411; (b) H. Cao and Y. Wang, Nucleic Acids Res., 2007, 35, 4833–4844.
18. B. M. Babior and R. C. Woodman, Semin. Hematol., 1990, 27, 247–259.

---

Footnotes

† Electronic supplementary information (ESI) available: Methods, figures and spectra are included. See DOI: 10.1039/c5sc00484e

‡ J. J. Wu, H. Xiao and T. L. Wang contributed equally to this work.

---