Characterization of Byproducts from Chemical Syntheses of Oligonucleotides Containing 1-Methyladenine and 3-Methylcytosine.

Oligonucleotides serve as important tools for biological, chemical, and medical research. The preparation of oligonucleotides through automated solid-phase synthesis is well-established. However, identification of byproducts generated from DNA synthesis, especially from oligonucleotides containing site-specific modifications, is sometimes challenging. Typical high-performance liquid chromatography (HPLC), mass spectrometry (MS), and gel electrophoresis methods alone are not sufficient for characterizing unexpected byproducts, especially for those having identical or very similar molecular weight (MW) to the products. We used a rigorous quality control procedure to characterize byproducts generated during oligonucleotide syntheses: (1) purify oligonucleotides by different HPLC systems; (2) determine exact MW by high-resolution MS; (3) locate modification position by MS/MS or exonuclease digestion with matrix-assisted laser desorption ionization-time of flight analysis; and (4) conduct, where applicable, enzymatic assays. We applied these steps to characterize byproducts in the syntheses of oligonucleotides containing biologically important methyl DNA adducts 1-methyladenine (m1A) and 3-methylcytosine (m3C). In m1A synthesis, we differentiated a regioisomeric byproduct 6-methyladenine, which possesses a MW identical to uncharged m1A. As for m3C, we identified a deamination byproduct 3-methyluracil, which is only 1 Da greater than uncharged m3C in the ∼4900 Da context. The detection of these byproducts would be very challenging if the abovementioned procedure was not adopted.

Introduction

n class="Chemical">Oligonucleotides synthesized chemically are widely used as drugs and research tools in biology, chemistry, and medicine. Solid-phase synthesis of DNA and RNA oligonucleotides has been well-developed, and the automated phosphoramidite-based chemical process has become highly efficient.[1−3] Phosphoramidites of regular and some modified nucleotides are commercially available, and certain oligonucleotide strands could be readily obtained from commercial sources.[4−7] Besides the great development in the synthesis of oligonucleotides, the differentiation of byproducts from the product oligonucleotides, especially on those containing site-specifically modified structures, is sometimes ignored by the end users. If those byproducts or small impurities were not identified and removed, it could have devastating consequences for the corresponding biological assays and medical treatments.[8,9] For certain instances, it is challenging to identify some byproducts generated in the synthesis and deprotection steps, especially the byproduct that has a molecular weight (MW) identical or very similar to the desired product oligonucleotide.

We have synthesized various n class="Chemical">oligonucleotides containing modified structures in the past, focusing on alkyl or aryl DNA adducts, by using solid- and solution-phase DNA synthesis.[10−14] In the syntheses of oligonucleotides containing 1-methyladenine (m1A) and 3-methylcytosine (m3C), we observed byproducts in a neutral context that are either regioisomer (identical MW) to the product or have 1 Da in MW greater than the product in a ∼4900 Da context (Figure a). m1A and m3C are formed by exogenous and endogenous alkylating agents mainly in single-stranded DNA and have been detected both in vitro[15−21] and in vivo.[15,20,22−29] Both adducts are cytotoxic and block DNA replication and are the best substrates for the AlkB family DNA repair enzymes (Figure b).[30−34] m3C has also been proposed as an epigenetic biomarker for cancer.[35]

Figure 1

Oligonucleotide products and byproducts studied in this work. (a) Structures of adducts and byproducts from chemical syntheses and (b) the alkyl adduct m1A, as an example, is repaired in the presence of the AlkB enzyme and necessary cofactors.

pan class="Chemical">Oligonucleotide products and byproducts studied in this work. (a) Structures of adducts and byproducts from chemical syntheses and (b) the alkyl adduct pan class="Chemical">m1A, as an example, is repaired in the presence of the AlkB enzyme and necessary cofactors.

Most of our DNA syntheses successfully provided target n class="Chemical">oligonucleotides. However, some oligonucleotides contained side reaction contaminants generated during the synthesis or deprotection steps, requiring further purification and identification. To that end, we applied a rigorous quality control process which entails the following steps: (i) synthesize an adduct-containing oligonucleotide from the corresponding phosphoramidite; (ii) purify the product oligonucleotide by both reverse-phase and anion-exchange high-performance liquid chromatography (HPLC); (iii) measure the exact MW of the oligonucleotide by high-resolution mass spectrometry (HR-MS), certain impurities can also be detected by HR-MS; (iv) determine the modification position by either MS/MS or exonuclease digestion with matrix assisted laser desorption ionization-time of flight (MALDI-TOF) analysis; and (v) test the biological activity of the adduct-containing oligonucleotide by appropriate enzymatic assays. We are assuming that the modified nucleotide has been extensively investigated for its stability under solid-phase chemical synthesis and deprotection beforehand; however, the strategy outlined above should confirm this. Nuclear magnetic resonance (NMR; 1H, 13C, and 31P) is a powerful tool for structural integrity studies of the phosphoramidite on the milligram scale.[36−38] However, NMR becomes impractical for characterizing minor impurities within the modified structures in synthetic oligonucleotides because of increasing spectral complexity, as well as the final amount of material available (e.g. on a micro-mole or less scale). Below, we report the detailed characterization of byproducts in the syntheses of oligonucleotides containing m1A and m3C. The characterization and separation of those byproducts provide confidence in the quality of oligonucleotides used in further biological experiments.

Results

Identification of the 6-Methyladenine Byproduct from m1A Synthesis

We were trying to incorporate n class="Chemical">m1A into an oligonucleotide as a substrate for the AlkB repair reaction.[34,39] A 16mer oligonucleotide containing m1A was prepared by using the commercially available phosphoramidite of m1A (Figure a). The final product was deprotected under standard conditions by treating the crude oligonucleotide with ammonium hydroxide at 80 °C for 3 h. The oligonucleotide was then tested by both reverse-phase and anion-exchange HPLC (see the Experimental Section for detailed conditions). The resulting chromatograms showed a single peak under both conditions (retention time 10.2 min in Figure a and 4.0 min in Figure S22). HR electrospray ionization-time of flight (ESI-TOF) MS analysis of the sample exhibited m/z at 1224.715 at its −4 charge state, which is in good agreement with the theoretical m/z 1224.711 expected of the product oligonucleotide (Figures a and S1 and Table ).

Figure 2

Reverse-phase HPLC chromatograms of 16mer DNA alkyl products and byproducts. The retention time of a modification-containing oligonucleotide is labeled on top of the corresponding chromatogram. The modifications are as follows: (a) m6A; (b) m1A; (c) A; (d) mixture of m3C + m3U; (e) m3C; and (f) m3U.

Figure 3

HR ESI-TOF MS analyses of 16mer DNA oligonucleotides containing target modifications and byproducts. Data represent the −4 charge envelopes, and the monoisotopic peak (all 12C, 14N, etc.) values are labeled above the first peak in each peak envelope. (a) The oligonucleotide (containing m6A) generated from initial m1A synthesis; (b) m1A; (c) m3C; and (d) m3U. The predicted and empirical isotopic pattern using related oligonucleotides can also be a useful tool in purity assessment.

Table 1

Calculated and Observed Monoisotopic MW and m/z Value of Modified Oligonucleotidesa

lesion or base	MW (calculated) of neutral species	m/z (calculated) –4 charge peak	m/z (observed) –4 charge peak
16mer m6A	4902.877	1224.711	1224.715
16mer m1A	4902.877	1224.711	1224.715
16mer m3C	4878.866	1218.709	1218.703
16mer m3U	4879.850	1218.955	1218.957

For m1A and m3C syntheses, the sequence of the 16mer was 5′-GAAGACCTXGGCGTCC-3′, where X indicates the position of the modified bases.

Reverse-phase HPLC chromatograms of 16mer DNA alkyl products and byproducts. The retention time of a modification-containing n class="Chemical">oligonucleotide is labeled on top of the corresponding chromatogram. The modifications are as follows: (a) m6A; (b) m1A; (c) A; (d) mixture of m3C + m3U; (e) m3C; and (f) m3U.

n class="Disease">HR ESI-TOF MS analyses of 16mer DNA oligonucleotides containing target modifications and byproducts. Data represent the −4 charge envelopes, and the monoisotopic peak (all 12C, 14N, etc.) values are labeled above the first peak in each peak envelope. (a) The oligonucleotide (containing m6A) generated from initial m1A synthesis; (b) m1A; (c) m3C; and (d) m3U. The predicted and empirical isotopic pattern using related oligonucleotides can also be a useful tool in purity assessment.

For pan class="Chemical">m1A and n>n class="Chemical">m3C syntheses, the sequence of the 16mer was 5′-GAAGACCTXGGCGTCC-3′, where X indicates the position of the modified bases.

Collision-induced dissociation (CID) MS/MS analysis was used to determine the location of the alkyl adduct in the synthesis of n class="Chemical">m1A-containing 16mer oligonucleotide (Figure ). The MS/MS results, presented in detail in the Supporting Information (Figure S5 and Table S1), confirmed that modification occurred at the eighth position from the 3′ end of the oligonucleotide, and the modified base had the same MW as m1A. Analysis of 3′ end fragmentations indicated that w15 to w8 fragments all contain an extra methyl modification (Figure , Table S1, and Figure S10). On the other hand, w7 to w1 ions showed only standard DNA sequences without methyl modification (Table S1). These results suggested that the modification is located between w8 and w7, which is the eighth position from the 3′ end of the 16mer oligonucleotide. From the 5′ fragmentations, we observed the methyl modification in ions from (a15-C) to (a10-G), but no such modification in ions from (a9-X) to (a2-A) (Figures and S9 and Table S1). These results suggested that the modification is located between (a10-G) and (a9-X), which is the ninth position from the 5′ end (Figure ). The fragmentation patterns from both the 5′ and 3′ ends are consistent with the proposed m1A-containing DNA sequence (Figure ).

Figure 4

Predicted fragmentation pattern from CID of the 16mer oligonucleotide products from the m1A and m3C syntheses. X denotes the modified nucleotides.

Predicted fragmentation pattern from CID of the n class="Chemical">16mer oligonucleotide products from the m1A and m3C syntheses. X denotes the modified nucleotides.

To further confirm the location of the modified base, we performed enzyme digestion coupled with MALDI-TOF MS analysis.[11−13,40,41] The procedure has been widely used to confirm the location of lesion positions by determining the mass changes after partial digestion from the 3′ end by the exonuclease snake venom phosphodiesterase (SVP).[42−44]Figure shows the positive ion MALDI-TOF MS spectra of the 3′ to 5′ SVP exonuclease digestion of the product pan class="Chemical">oligonucleotide at different time intervals. The m/z of 4904.2 at 0 min (before digestion, as a control) indicated that the 16mer DNA sequence contained a methylated DNA base ([M+H]+ theoretical m/z = 4903.9 Da; herein theoretical m/z values are shown in parentheses after the observed m/z values). After 1 min SVP digestion, while the product ion disappeared, three new lower masses appeared at m/z = 4615.7 (4614.8), 4326.7 (4325.8), and 4022.7 (4021.7), which correspond to the 15, 14, and 13mer fragments generated from 3′ cleavage, respectively (Figure ). The signal intensity of three new peaks were significantly increased with the disappearance of the original product pan class="Chemical">oligonucleotide ion at m/z = 4904.2.

Figure 5

Time-course MALDI-TOF analyses of SVP digestion products of the 16mer product oligonucleotide (containing m6A) generated from m1A synthesis. Theoretical masses are listed in the inset. The theoretical monoisotopic mass at 2745.5 is highlighted in red because it is the smallest digestion product containing the modification.

Time-course MALDI-TOF analyses of SVP digestion products of the 16mer product n class="Chemical">oligonucleotide (containing m6A) generated from m1A synthesis. Theoretical masses are listed in the inset. The theoretical monoisotopic mass at 2745.5 is highlighted in red because it is the smallest digestion product containing the modification.

Further digestion resulted in a number of smaller fragments. In the 3 min digestion (Figures and S17), the signal of m/z = 3075.1 (3074.6) matches the sequence for the fragment that was cleaved one base before the lesion. The m/z = 2746.4 (2745.5, containing a methyl modification) represents the digestion occurring on the 3′ side of the modified n class="Chemical">adenine position (Figure ), which also persisted in the 5 min digestion spectrum. The signal at 2419.1 (2418.5, no methyl modification) in the 3 and 5 min digestions represents the fragment generated after liberation of deoxymethyladenosine 5′ monophosphate. Digestions practically stopped at m/z = 1223.6/8 (1222.3), which represents the m/z of a 4mer nucleotide of the 5′ end. Taken together, these results confirm that the modified structure is a methylated adenine and it is at the ninth position of the proposed oligonucleotide from the 5′to 3′ direction.

The chromatographic and mass spectral evidence presented above indicates that the product n class="Chemical">oligonucleotide could have the MW expected for a 16mer containing the m1A adduct in the correct position. We then tested the repair of this product by the AlkB protein, which is an enzyme that repairs various alkylated DNA adducts (Figure b).[32,34] The result showed the repair efficiency of this oligonucleotide by AlkB was very low, even when 2.5 μM of AlkB was incubated with the 5.0 μM product oligonucleotide (Figures b and S24). This observation was in contrast to the report that m1A is a good substrate of AlkB (e.g. 0.2 μM of AlkB is able to repair 5.0 μM m1A completely in 1 h).[10] We suspected that the structure in the product oligonucleotide may not be m1A, even though it had a MW identical to m1A. The reverse-phase HPLC results also showed an anomalous phenomenon. For example, the positively charged m1A nucleobase[34] should be more polar than the natural base adenine and thus have a shorter retention time on a C18 column. However, the opposite was observed: adenine has a shorter retention time than the product (Figure a,c).

Figure 6

HPLC profiles of the AlkB repair reactions on different alkyl substrates. The three chromatograms within each panel represent one set of repair reaction including the oligonucleotides of starting material, reaction mixture, and product. For example, in the panel containing chromatograms a–c, (a) represents the starting material m6A, (b) represents the repair reaction of m6A by AlkB, and (c) represents the pure product adenine synthesized separately. Chromatograms shown in a–c relevant to m6A repair and in j–l relevant to m3U repair were analyzed under reverse-phase conditions. Chromatograms shown in d–f relevant to m1A repair and in g–i relevant to m3C repair were analyzed under anion-exchange conditions (see the Experimental Section for detailed information). (a) m6A; (b) m6A + AlkB; (c) A; (d) m1A; (e) m1A + AlkB; (f) A; (g) m3C; (h) m3C + AlkB; (i) C; (j) m3U; (k) m3U + AlkB; and (l) U.

HPLC profiles of the AlkB repair reactions on different alkyl substrates. The three chromatograms within each panel represent one set of repair reaction including the n class="Chemical">oligonucleotides of starting material, reaction mixture, and product. For example, in the panel containing chromatograms a–c, (a) represents the starting material m6A, (b) represents the repair reaction of m6A by AlkB, and (c) represents the pure product adenine synthesized separately. Chromatograms shown in a–c relevant to m6A repair and in j–l relevant to m3U repair were analyzed under reverse-phase conditions. Chromatograms shown in d–f relevant to m1A repair and in g–i relevant to m3C repair were analyzed under anion-exchange conditions (see the Experimental Section for detailed information). (a) m6A; (b) m6A + AlkB; (c) A; (d) m1A; (e) m1A + AlkB; (f) A; (g) m3C; (h) m3C + AlkB; (i) C; (j) m3U; (k) m3U + AlkB; and (l) U.

It has been reported that n class="Chemical">6-methyladenine (m6A) could be generated from a Dimroth rearrangement of m1A under basic conditions (Figure a and Table ).[34,45] We therefore conducted a new batch synthesis of m1A under a milder deprotection condition (25 °C for 16 h). The product oligonucleotide was eluted as a single peak in both HPLC columns (Figure b for C18 and Figure d for anion-exchange). The HPLC retention time in reverse-phase HPLC exhibited a shorter retention time (8.5 min, Figure b) than the unmodified control adenine (9.0 min, Figure c) or the previous product (10.2 min, m6A, Figure a), which is in good agreement with the theory that m1A should be more polar than both A and m6A.

The MW of the new product was characterized by HR-MS and had the correct m/z of n class="Chemical">m1A at 1224.715 (1224.711) (Figures b and S2 and Table ). The MS/MS results (e.g., the w7, w8, a9-X, and a10-G ions, Figures , S6, S11, and S12, and Table S2) demonstrated that the modification is at the ninth position from the 5′ end and it is a methylated adenine. The exonuclease digestion with MALDI-TOF analyses (Figures S18 and S21a) also confirmed the location and identity of the adduct. For example, the 2746.1 (before the adduct) and 2418.8 (after the adduct) peaks at 3 min digestion in Figure S18 were present. Finally, incubation with AlkB showed an excellent repair efficiency against the new product oligonucleotide: 5.0 μM adduct was mostly repaired in 1 h with the 0.2 μM enzyme (Figure e).[10] Taken together, these results confirmed the newly synthesized product as a 16mer oligonucleotide containing m1A at the ninth position from the 5′ end (Figure a).

Identification of the m3U Byproduct from m3C Synthesis

We also wanted to synthesize an n class="Chemical">oligonucleotide containing m3C as the substrate for the AlkB family DNA repair reactions within the same sequence context as for the m1A oligonucleotide (Figure a).[34,39] After automated synthesis and standard deprotection (80 °C for 3 h), the HPLC chromatogram from a C18 column showed two peaks with similar intensities (9.5 and 11.6 min, Figure d), which were readily separated (Figure e,f). HR-MS analysis showed that the early eluting peak (9.5 min, Figure e) had an m/z value of 1218.703 at its −4 charge state (Figures c and S3), which is in agreement with that of the m/z of the target m3C-containing oligonucleotide (theoretical m/z 1218.709, Table ). The second species at 11.6 min (Figure f) showed an m/z of 1218.957 at its −4 charge state (Figures d and S4). Compared to the 9.5 min peak, the m/z value of the second species was 0.254 unit higher at the −4 charge state, indicating that the MW (4879.850 Da) of the second species is 1.016 Da higher than the first one. This is consistent with the m/z value of an oligonucleotide containing 3-methyluricil (m3U) at its −4 charge state (1218.955, Table ). m3U could potentially be generated from deamination of m3C under basic conditions (Figure a).[46] The MS/MS results of the two oligonucleotides (e.g., the w7, w8, a9-X, and a10-G ions, Figures , S7, S8, and S13 to S16, Tables S3, and S4) demonstrated that the modifications are at the ninth position from the 5′ end. The exonuclease digestion with MALDI-TOF analyses (Figures S19, S20, and S21b) also confirmed the location and identity of the adducts. At 7 min digestion, signature peaks were 2722.2 (before the adduct) and 2418.8 (after the adduct) for m3C (Figure S19); and 2723.3 (before the adduct) and 2418.7 (after the adduct) for m3U (Figure S20). The C18 HPLC retention time of the two species showed an m3C-containing oligonucleotide had a shorter retention time (9.5 min, Figure e) than m3U (11.6 min, Figure f). This result indicated that m3C is more polar than m3U, which supports that m3C is positively charged at pH 7.0 (Figure a).[34] The AlkB reaction (0.2 μM) on the m3C-containing oligonucleotide showed a good repair efficiency: 5.0 μM m3C adduct was mostly repaired to cytosine after 1 h (Figure h). This was not the case for m3U-containing oligonucleotides under a similar condition, which showed almost no repair to m3U (Figure k). These observations were consistent with the observations that m3C is a strong substrate but m3U is a weak substrate for AlkB under both in vitro and in vivo conditions.[34,47]

To avoid the formation of n class="Chemical">m3U, we carried out a second batch synthesis of m3C-containing oligonucleotides under a milder deprotection condition (25 °C for 16 h). This time we observed only the m3C-containing species in the HPLC chromatogram (similar to Figure e). The identity of the newly synthesized m3C was confirmed by the aforementioned spectroscopic and enzymatic tests. These results indicate that the formation of the byproduct m3U was from deamination of m3C catalyzed by harsh deprotection conditions (high temperature and alkaline).[46,47] Serendipitously, the formation of m3U/m6A in the first batch of synthesis (80 °C for 3 h) provides a new method for chemically preparing m3U/m6A-containing oligonucleotides.

Discussion

For the synthesis of n class="Chemical">m1A, we observed a byproduct m6A, which exhibited identical MW and location to m1A. It is difficult to differentiate m1A and m6A without conducting bioassays and careful HPLC retention analysis. As for m3C, it was the m3U byproduct that showed a 1 Da difference in the ∼4900 Da context (Table ). The two oligonucleotides could be eluted as a single species in HPLC under certain conditions; thus, it is imperative to perform HR-MS analysis and enzymatic reaction to distinguish the two products. For HPLC analysis of modified oligonucleotides, we used a combination of two different separation systems, such as reverse-phase (e.g. C18) and anion-exchange chromatography. For example, it is not easy to find a suitable condition for fully separating 16mer-containing m1A and A with C18 columns (Figure b,c). However, it is relatively easy to separate them with an anion-exchange column because m1A is positively charged at neutral pH but A is not (Figures a, 6d and 6f).[34] On the other hand, it is not easy to distinguish m6A and A under anion-exchange conditions (Figures S22 and S23) because of their neutrality at pH 7.0, but it is possible to separate them on the C18 column (Figure a,c).

The MW of the n class="Chemical">oligonucleotide can be determined by HR-MS, and the location of the modification can be identified by MS/MS or exonuclease digestion with MALDI-TOF analysis. To further confirm the identity of the oligonucleotide from a biological activity perspective, we recommend running enzymatic reaction on the product. In this work, we used SVP exonuclease to digest all oligonucleotides formed in the syntheses. A specific enzyme for certain types of modification could greatly help elucidate the structure of modification. Therefore, we tested the repair efficiency of m1A and m3C and their byproducts by the AlkB repair enzyme.[32,34] The AlkB protein in Escherichia coli is an α-ketoglutarate/Fe(II)-dependent dioxygenase that repairs various alkyl DNA adducts, including m1A and m3C (Figure ).[30−32,39,48−50] We found that m1A and m3C are good substrates but m6A and m3U are weak substrates of AlkB.[34,47,48] For other modified structures, it is highly recommended that a biological or enzymatic assay should be adopted for identifying the product. The reason for adopting an enzymatic test is because a byproduct (e.g. m6A) may have an MW identical to the target product (e.g. m1A), which may be hard to differentiate by LC and MS analyses, including the MS/MS and exonuclease digestion with MALDI-TOF analysis.

Conclusion

Chemical synthesis of n class="Chemical">oligonucleotides is important for conducting various biological, chemical, and medical research including oligonucleotide drug development. The field of oligonucleotide synthesis has progressed such that DNA and RNA containing standard bases can be ordered from commercial sources. Most of these oligonucleotides are in high quality, but it is a good practice to perform quality control to confirm that the products have the correct sequences in high purity. A simple HPLC/MS analysis should be sufficient for most applications. However, a more rigorous and stringent quality control procedure[51−53] should be adopted for site-specifically modified oligonucleotides, such as epigenetic marks, DNA adducts, and drug candidates. In some cases, byproducts are generated from side chemical reactions during standard automated synthesis and deprotection steps. For this reason, specific deprotection conditions may be required for preparation of oligonucleotide-containing modifications, such as epigenetic biomarkers 5-hydroxymethyl-dC, 5-formyl-dC, and 5-carboxy-dC.[54,55] After synthesis and deprotection, it is even more important to carry out a thorough purification and characterization procedure to ensure (1) complete removal of protecting groups, (2) product having high purity, and (3) modification having the correct position and identity. In this study, we used a reliable and robust procedure to characterize byproducts from m1A and m3C syntheses. The protocol used here could be helpful for identifying byproducts generated from other oligonucleotide syntheses.

Experimental Section

Synthesis of Oligonucleotides Containing m1A and m3C

n class="Chemical">Oligonucleotides (16mer) with the sequence 5′-GAAGACCTXGGCGTCC-3′ containing the lesions at the X position were made by using solid-phase phosphoramidite chemistry on a MerMade-4 Oligonucleotide synthesizer.[10,34,56] The phosphoramidites were purchased from ChemGenes. N1-Methyl deoxyadenosine (n-fmoc) phosphoramidite was used for m1A synthesis. N3-Methyl deoxycytidine (n-bz) phosphoramidite was used for m3C synthesis. The final cleavage and deprotection step was usually carried out by treating the oligonucleotide with concentrated aqueous ammonium hydroxide (28%) at 80 °C for 3 h. The modified cleavage and deprotection step for minimizing the byproduct formation was at 25 °C for 16 h. The concentration of an oligonucleotide was determined by measuring UV absorbance at 260 nm. The extinction coefficient (ε) of a certain adduct is calculated as its unmodified counterpart because of the negligible variation between the values in the context of a 16mer DNA.

Purity Test of Oligonucleotides Containing m1A and m3C and Related Byproducts by HPLC

The purity of n class="Chemical">oligonucleotides was tested by both reverse-phase (C18) and anion-exchange HPLC methods. For the reverse-phase test, liquid chromatographic separation was performed by using a Phenomenex Luna Semi-Preparative C18 column (9 × 250 mm, 5 μm) at a flow rate of 3 mL/min. Solvent A was 100 mM triethylammonium acetate in water, and solvent B was 100% acetonitrile. A solvent gradient was carried out under the following conditions: 2.0% of B for 0.2 min, 2.0 to 9.0% of B over 0.3 min, 9.0 to 9.4% of B over 16 min, 9.4 to 70.0% of B over 0.5 min, 70.0% of B for 3 min, 70.0 to 2.0% of B over 0.5 min, and 2.0% B over 4.5 min. The column oven was set at 40 °C. The UV signal at 260 nm was used to detect the oligonucleotide absorbance. For the anion-exchange LC analysis, oligonucleotides were purified and tested by using a Thermo DNAPac PA-100 anion-exchange column (4 × 250 mm, 13 μm) with solvent A as water and solvent B as 1.5 M ammonium acetate in water. A solvent gradient was carried out under the following conditions: 50.0% of B for 1 min, 50.0 to 52.0% of B over 2 min, 52.0 to 75.0% over 1 min, 75.0% of B for 2 min, 75.0 to 50.0% of B over 0.5 min, and 50.0% of B for 4.5 min. The flow rate was at 4.0 mL/min.

HPLC/MS Analysis

HR-MS analyses of n class="Chemical">oligonucleotides were performed on AB Sciex triple quadrupole-TOF mass spectrometers (ABI4600 and ABI5600). ESI was conducted by using a needle voltage of 4.0 kV in a negative ion mode. A heated capillary was set at 600 °C. The nebulizer gas pressure was 40 psi; the heater gas pressure was 40 psi; the curtain gas pressure was 25 psi; the declustering potential was −220 V; and the collision energy was −10 V. Liquid chromatographic separation was performed using a Phenomenex Luna C18 column (4.6 × 100 mm; 5 μm) at a flow rate of 0.4 mL/min. Solvent A was 10 mM ammonium acetate in water, and solvent B was 100% acetonitrile. A solvent gradient was carried out under the following conditions: 2.0% of B for 0.5 min, 2.0 to 17.4% of B over 11 min, 17.4 to 60.0% of B over 0.1 min, 60.0% of B for 2 min, 60.0 to 2.0% of B over 0.1 min, and 2.0% B over 3.3 min.

MS/MS Analysis

n class="Chemical">Oligonucleotide fragmentation analyses were performed by manually injecting the oligonucleotide samples of 100 pmol/μL into AB Sciex triple quadrupole-TOF mass spectrometers (ABI4600 and ABI5600). The syringe flow rate was set at 10 μL/min. ESI was conducted by using a needle voltage of 4.5 kV in a negative ion mode. A heated capillary was set at 400 °C. The nebulizer gas pressure was 75 psi; the heater gas pressure was 25 psi; the curtain gas pressure was 25 psi; the declustering potential was −100 V; and the collision energy was −15 V. The parent ion m/z for m6A, m1A, m3C, and m3U were selected as 816.14 (−6 charge), 816.14 (−6 charge), 974.77 (−5 charge), and 974.96 (−5 charge), respectively. Data analyses were performed with the AB Sciex Analyst TF software 1.7.

Exonuclease Digestion with MALDI-TOF Analysis

The modified n class="Chemical">oligonucleotides were characterized by 3′–5′exonuclease digestion followed by MALDI-TOF analysis. In general, 1.0 μL of sample containing 200–250 pmol of a modified oligonucleotide was used for digestion. SVP (0.2 unit) was added together with 6.0 μL of ammonium citrate (100 mM, pH 9.4) and 6.0 μL of water for the 3′ to 5′ digestion.[11] For the MALDI-TOF analyses, 1.0 μL of the mixture was used at a certain time point until the digestion was almost finished; the digestion was quenched by mixing with 1.0 μL of the MALDI analysis matrix (3-hydroxypicolinic acid and diammonium hydrogen citrate in a 1:1 ratio). Samples were analyzed by a Shimadzu Axima Performance MALDI-TOF mass spectrometer using a 50 Hz laser with a power setting of 90 in a positive ion reflection mode with 500 shots collected.[13]

Enzymatic Reaction with the AlkB Protein and HPLC Analysis

All reactions were performed at 37 °C for 1 h in a reaction buffer containing 70.0 μM Fe(NH4)2(SO4)2·6H2O, 0.93 mM α-ketoglutarate, 1.86 mM n class="Chemical">ascorbic acid, and 46.5 mM HEPES (pH 8.0).[10] The reactions were quenched by adding 10.0 mM EDTA followed by heating at 95 °C for 5 min. Typically, 0.25 μM of the purified AlkB protein was incubated with 5.0 μM m6A and m1A in the presence of all cofactors in a 20 μL reaction volume. For m3C and m3U, 0.18 μM of AlkB was incubated with 5.0 μM of oligonucleotides. Samples were then analyzed under an HPLC condition that was able to separate the substrate and the product.

To separate the starting material and the product of the enzymatic reactions for the four lesions, different HPLC conditions were optimized by using either C18 or anion-exchange columns. The UV detection was set at 260 nm. Specifically, (1) n class="Chemical">m6A and its repaired product A were analyzed by using a Phenomenex Luna Semi-Preparative C18 column (10 × 150 mm, 5 μm) under the following conditions: solvent A was 100 mM n>n class="Chemical">triethylammonium acetate in water, and solvent B was 100% acetonitrile. A solvent gradient was carried out under the following conditions: 2.0% of B for 0.2 min, 2.0 to 9.0% of B over 0.3 min, 9.0 to 9.4% of B over 16 min, 9.4 to 70.0% of B over 0.5 min, 70.0% of B for 3 min, 70.0 to 2.0% of B over 0.5 min, and 2.0% B for 4.5 min. The flow rate was at 3.0 mL/min. The column oven was set at 40 °C. This protocol was also used to analyze m3U and its repair product U. (2) m1A and its repaired product A were analyzed by using a Thermo DNAPac PA-100 anion-exchange column (4 × 250 mm, 13 μm) with solvent A as water and solvent B as 1.5 M ammonium acetate in water. A solvent gradient was carried out under the following conditions: 50.0% of B for 1 min, 50.0 to 52.0% of B over 2 min, 52.0 to 75.0% over 1 min, 75.0% of B for 2 min, 75.0 to 50.0% of B over 0.5 min, and 50.0% of B for 4.5 min. The flow rate was at 4.0 mL/min. (3) m3C and its repaired product C were analyzed with a similar protocol to m1A/A analysis, except that the solvent gradient was changed to the following conditions: 50.0% of B for 1 min, 50.0 to 57.0% of B over 1 min, 57.0% of B for 2 min, 57.0 to 50.0% of B for 0.5 min, and then 50.0% of B for 4.5 min.