Anal. Chem. 2009, 81, 3173–3179
Sequencing of Single and Double Stranded RNA Oligonucleotides by Acid Hydrolysis and MALDI Mass Spectrometry Ute Bahr,*,† Hu¨seyin Aygu¨n,‡ and Michael Karas† Cluster of Excellence “Macromolecular Complexes”, Institute of Pharmaceutical Chemistry, University of Frankfurt, 60438 Frankfurt, Germany, BioSpring GmbH, 60386Frankfurt, Germany Treatment of RNA oligonucleotides with strong acids at pH 1-2 rapidly leads to hydrolysis of the phosphodiester bonds at the 5′-position of ribose. Analysis of the resulting degradation products by MALDI coupled to an Orbitrap high resolution mass spectrometer shows almost complete mass ladders from both sides of the nucleotides without interfering fragments from base losses or internal fragments. From the mass differences between adjacent peaks of a mass ladder, the sequence can be determined. Low cleavage efficiency at the termini leads to 2mers and 3mers which can be identified by MS/MS. In this way the complete sequences of different siRNA 21mer single and double strands could be verified. This simple and fast method can be applied for controlling sequences of synthetic oligomers, as well as for de-novo sequencing. Moreover, the method is applicable for localization and identification of RNA modifications as demonstrated using the examples of an oligonucleotide with phosphorothioate backbone and of one containing 2′-methoxyribose modifications. With increasing interest in the use of RNA oligonucleotides, such as siRNAs for therapeutic purposes, the demand to develop efficient methods to characterize these molecules also increases. MALDI MS is already a well established method used for quality control of synthetic oligonucleotides. Impurities such as incomplete sequences can be directly read out of a mass spectrum. The molecular weight of the compound can provide sufficient evidence to confirm the composition of an oligonucleotide, but it is inadequate to determine the structure of an unknown. Since the early 1990s several attempts have been undertaken for sequencing DNA and RNA by mass spectrometry. A comprehensive review of different sequencing techniques is given by Limbach,1 and a short overview of mass spectrometry applied to RNA appeared recently.2 Sequence analysis based on mass spectrometric techniques includes failure analysis, enzymatic degradation, and gas-phase fragmentation (MS/MS). Failure analysis addresses the problem that automated solid-phase * To whom correspondence should be addressed. E-mail: bahr@ pharmchem.uni-frankfurt.de. † University of Frankfurt. ‡ BioSpring GmbH. (1) Limbach, P. A. Mass Spectrom. Rev. 1996, 15, 297–336. (2) Thomas, B. T.; Akoulitchev, A. V. Science 2006, 31, 173–180. 10.1021/ac900100x CCC: $40.75 2009 American Chemical Society Published on Web 03/18/2009
synthesis of oligonucleotides is not always 100%.3 Therefore, the crude synthetic mixture contains beside the intact product a distribution of “failure sequences” that differ in length by successive nucleotides. From the mass differences between adjacent failure sequence ions in the mass spectrum, the sequence can be read from the 5′-end. Exonucleases, first used by Pieles et al.,4 generate a “mass ladder” for which the mass difference between adjacent peaks in the mass spectrum matches a single nucleotide. To read the whole sequence, two exonucleases, one for digestion of the nucleotide from the 3′-end and one from the 5′-end, have to be used. However, because of the kinetics of the enzymes, multiple mass measurements have to be carried out after different time intervals. In addition to exonucleases, base-specific nucleases have also been used for DNA, RNA and modified nucleotides.5-9 Because of the limited mass resolution and mass accuracy of linear MALDI-TOF instruments, differentiation between the nucleobases cytosine and uracil has been a significant problem in the unambiguous assignment of RNA sequences. Faulstich et al.10 used fragment intensities to differentiate between C and U, Tolsen et al.8 developed a method employing a combination of exonucleases and chemical cleavage, and Hahner et al.7 used base-specific endonucleases and limited alkaline hydrolysis for obtaining complete sequence information. Gas-phase fragmentation has been extensively studied and reviewed by Wu et al.11 Different fragmentation pathways of DNA and RNA for singly charged positive ions were reported.12 Sequence determination, particularly of large oligonucleotides, forms a considerable problem for this technique and is usually (3) Keough, T.; Baker, T. R.; Dobson, R. L. M.; Lacey, M. P.; Riley, T. A.; Hasselfield, J. A.; Hesselberth, P. E. Rapid Commun. Mass Spectrom. 1993, 7, 195–200. (4) Pieles, U.; Zu ¨ rcher, W.; Scha¨r, M.; Moser, H. E. Nucleic Acid Res. 1993, 21, 3191–3196. (5) Nordhoff, E.; Karas, M.; Cramer, R.; Hahner, S.; Hillenkamp, F.; Kirpekar, F.; Lezius, A.; Muth, J.; Meier, C.; Engels, J. W. J. Mass Spectrom. 1995, 30, 99–112. (6) Bentzley, C. M.; Johnston, M. V.; Larsen, B. S.; Gutteridge, S. Anal. Chem. 1996, 68, 2141–2146. (7) Hahner, S.; Lu ¨ demann, H.-C.; Kirpekar, F.; Nordhoff, E.; Roepstorff, P.; Galla, H.-J.; Hillenkamp, F. Nucleic Acid Res. 1997, 25, 1957–1964. (8) Tolson, D. A.; Nicholson, N. H. Nucleic Acid Res. 1998, 26, 446–451. (9) Zhang, L.-K.; Gross, M. L. J. Am. Soc. Mass Spectrom. 2000, 11, 854–865. (10) Faulstich, K.; Wo ¨rner, K.; Brill, H.; Engels, J. W. Anal. Chem. 1997, 6, 4349–4353. (11) Wu, J.; McLuckey, S. A. Int. J. Mass Spectrom. 2004, 237, 197–241. (12) Andersen, T. E.; Kirpekar, F.; Haselmann, K. F. J. Am. Soc. Mass Spectrom. 2006, 17, 1353–1368.
Analytical Chemistry, Vol. 81, No. 8, April 15, 2009
3173
limited to about 15mers. Only recently, siRNA 21mers were sequenced by a dedicated CID approach applying charge reduction in an ESI-MS/MS system.13 Chemical methods for sequence analysis of oligonucleotides are used rather seldom or for special applications only. Base catalyzed hydrolysis was used to cleave methylphosphonate backbones,3,14 and alkylating agents were used to cleave phosphorothioate-linked oligonucleotides.15 Acid hydrolysis has been applied for cleavage of a 19mer DNA; however, all bases were lost before backbone cleavage occurred.5 To the best of our knowledge, this method has not successfully been applied to any kind of RNA sequence determination, although it is well-known that for RNA the cleavage of the phosphodiester bond occurs faster than the cleavage of the glycosidic bond.16 We will show here, that acid hydrolysis can be used as a simple method for generating RNA mass ladders which can then be analyzed by a high resolution instrument for unambiguous sequence determination of RNA oligonucleotides. EXPERIMENTAL SECTION All oligonucleotides listed below were obtained from BioSpring (Frankfurt, Germany): ON-1B: 5′UAU CAC UUG AUC UCG UAC AdTdT 3′ ON-1A: 5′UGU ACG AGA UCA AGU GAU AdTdT 3′ ON-2A: 5′-GGU CCG GGA UCA CGU GAU AdTdT 3′ ON-2Amut: 5′GGU CCG GGA CUA CGU GAU AdTdT 3′ ON-2B: 5′UAU CAC GUG AUC CCG GAC CdTdT 3′ ON-2AB: double strand from ON-2A and ON-2B. ON-3A: mUGmU AmCG mAGmA UmCA mAGmU GmAU mA (m ) ribose 2′Omethyl). ON-2A-SH: 5′-GsGsUs CsCsGs GsGsAs UsCsAs CsGsUs GsAsUs AsdTsdT-3′ (s ) phosphorothioate). Each sample was dissolved in pure water to a concentration of 20 µM. Four microliters of this solution were mixed with 1 µL of 3.75% trifluoracetic acid (∼pH 1). The mixture was gently shaken at 37 °C for 20-30 min. For degradation of the thioate oligonucleotide the mixture was incubated at 60 °C for 15 min. After hydrolysis, 0.5 µL of each mixture was mixed with 2 µL matrix solution directly on the MALDI target and dried in a stream of air. The matrix used was 3-hydroxypicolinic acid (45 mg/mL in water) containing 5 mg/mL diammoniumhydrogencitrate. The mass spectrum shown in Figure 2 was recorded with a MALDI-TOF mass spectrometer (Voyager DE-PRO, Applied Biosystems, Framingham, U.S.A.) in linear positive ion mode. All other experiments were obtained on a MALDI Orbitrap (XL, Thermo Scientific, San Jose, U.S.A.) in negative ion mode. Resolution at this instrument was set to 30000, and a mass range of 600-4000 was chosen. Several spectra were accumulated over a maximum of 80 laser shots, depending on the number of ions produced. MS/MS was done in the linear ion trap followed by mass analysis in the orbitrap. (13) Huang, T.; Liu, J.; Liang, X.; Hodges, B. D. M.; McLuckey, A. Anal. Chem. 2008, 80, 8501–8508. (14) Keough, T.; Shaffer, J. D.; Lacey, M. P.; Riley, T. A.; Marvin, W. B.; Scurria, M. A.; Hasselfield, J. A.; Hesselberth, E. P. Anal. Chem. 1996, 68, 3405– 3412. (15) Polo, L. M.; McCarley, T. D.; Limbach, P. A. Anal. Chem. 1997, 69, 1107– 112. (16) Pollmann, W.; Schramm, G. Z. Naturforsch. 1961, 16b, 673–678.
3174
Analytical Chemistry, Vol. 81, No. 8, April 15, 2009
Figure 1. (a) Mechanism of hydrolysis of phosphodiester bonds by strong acids. (b) Products resulting from acid hydrolysis of oligonucleotides.
RESULTS AND DISCUSSION Acid Hydrolysis. It is well known that RNA and DNA can be degraded by acids. Two hydrolysis sites are present in nucleotides, the glycosidic bond between ribose and nucleobase and the phosphodiester bonds. Already in 1961 Pollmann et al.16 measured the hydrolysis constants of phosphate and glycoside cleavages at various pH values (between 1.8-2.4). Independent of the pH, cleavage of the phosphate was about five times faster than the cleavage of the purine bases. This is in contrast to DNA where the cleavage of purine bases occurs more than 100 times faster than the backbone cleavage, and 680 times faster compared to RNA. The rate of hydrolysis for all groups was roughly proportional to the H+ concentration. The mechanism of acid hydrolysis of RNA has been intensively studied, and the results were summarized by Oivanen.17 Accordingly, acid hydrolysis follows the mechanism shown in Figure 1a. At low pH the non-bridging phosphoryl oxygen is rapidly protonated, followed by attack of the 2′-hydroxyl group on phosphorus to give a pentacoordinated species. Proton transfer to the 5′O leads to cleavage of the 5′-linked nucleotide, the resulting cyclic phosphodiester is not stable and hydrolyzes rapidly. As a result, two ladders of neutral species are formed as can be seen in Figure 1b; hydrolysis products bearing the original 5′-end (5′-fragments) have a 3′-phosphate group, and hydrolysis products with the original 3′-end (3′-fragments) have OH-groups on both ends. (17) Oivanen, M.; Kuusela, S.; Lo ¨nneberg, H. Chem. Rev. 1998, 98, 961–990.
Figure 2. MALDI mass spectrum of the 21mer ON-1B after acid hydrolysis obtained with a linear TOF instrument. Ladder 1 gives the sequence from the 3′-end, ladder 2 from the 5′-end.
For hydrolysis of RNA 21mers we tested different acids, HCl, formic acid, trifluoracetic acid, and acetic acid, with pH values between 0.5-3 and hydrolyzed the samples varying time and temperature. In agreement with the findings of Pollmann16 the hydrolysis products are independent of the pH, and only the product yield is pH dependent. To obtain a reasonable constant intensity of all fragments at given pH, temperature and hydrolysis time have to be properly selected. We used 20 min at 37 °C for hydrolysis, but to expedite the reaction, 5 min at 60 °C led to comparable results. TFA was preferred as acid, since it is available at high purity and facilitates a good crystallization with the HPA matrix. In Figure 2 the mass spectrum of the 21mer ON-1B after acid hydrolysis obtained in a linear TOF mass spectrometer is shown. The hydrolysis is not base specific, all phosphoester bonds have been cleaved with the exception of the dinucleotide on the 5′-end and the trinucleotide on the 3′-end. While the two deoxythymidines at the 3′-end can not be cleaved because of the lack of a 2′OH-group (see mechanism in Figure 1a), the cleavage of the last nucleotides at the 5′-end exhibits an efficiency too low to be detected, or the detection is limited by matrix signal background in the low mass range. Base cleavages or internal fragments resulting from multiple cleavages are not observed. Two mass ladders are obtained in one step, one from the 5′-end, the other from the 3′-end. From the mass difference between adjacent peaks in each ladder the sequence can be inferred. With respect to mass spectrometric analysis RNA sequencing is more demanding than DNA sequencing, since the nucleobases uracil and cytosine differ in mass by only 1 Da, and the mass accuracy and resolving power of linear TOFs, especially in the high mass range, is often not sufficient to determine these differences, even when internal standards are used for calibration. For this reason we used the high resolution MALDI orbitrap instrument for mass determination. The instrument delivers a resolving power of up to 100 000 (defined at mass 400) and a mass accuracy of
