On-Line Cation Exchange for Suppression of Adduct Formation in

Jingjing Xu , Jing Gao , Chengli Yu , Han He , Yiming Yang , Daniel Figeys , and ... Tawnya G. Flick , Catherine A. Cassou , Terrence M. Chang , and E...
0 downloads 0 Views
Download PDF
Anal. Chem. 1998, 70, 5288-5295

On-Line Cation Exchange for Suppression of Adduct Formation in Negative-Ion Electrospray Mass Spectrometry of Nucleic Acids Christian G. Huber* and Michael R. Buchmeiser

Institute of Analytical Chemistry and Radiochemistry, Leopold-Franzens-University, Innrain 52a, A-6020 Innsbruck, Austria

One major difficulty in the analysis of nucleic acids by electrospray mass spectrometry is represented by the affinity of the polyanionic sugar-phosphate backbone for nonvolatile cations, especially ubiquitous sodium and potassium ions. A simple on-line sample preparation system comprising a microflow pumping system and 45× 0.8-mm-i.d. microcolumns packed with weak or strong cation-exchange resins is described for the efficient removal of cations from nucleic acid samples. Samples were analyzed by flow injection analysis at a 3-5 µL/min flow of 10 mM triethylamine in 50% water-50% acetonitrile. After on-line desalting, mass spectra of oligonucleotides revealed no significant sodium adduct peaks. Moreover, signal-to-noise ratios were greatly enhanced compared to direct injection of the samples. Electrospray mass spectrometry with on-line sample preparation allowed accurate molecular mass determinations of picomole amounts of crude oligonucleotide preparations ranging in size from 8 to 80 nucleotides within a few minutes. The good linearity of the calibration plot (R2 ) 0.9988) over at least 2 orders of magnitude and a relative standard deviation in peak areas of less than 9% permitted the sensitive quantitative measurement of oligonucleotides in a concentration range of 0.2-20 µM with selected-ion monitoring. Finally, the on-line sample preparation system was evaluated for the mass spectrometric analysis of complex oligonucleotide mixtures. The ability to characterize biological macromolecules such as proteins or nucleic acids by mass spectrometry (MS) has substantially benefited from the development of new soft ionization techniques, namely electrospray ionization (ESI)1 and matrixassisted laser desorption/ionization (MALDI)2. The applicability of ESI-MS and MALDI-MS to the mass spectrometric analysis of nucleic acids has been greatly extended by recent advances in sample preparation methods, ion-source technology, solution- and gas-phase chemistry, and mass analyzers.3,4 While MALDI seems * Corresponding author. Tel.: +43 512 507 5176. Fax: +43 512 507 2767. E-mail: [email protected]. (1) Witehouse, C. M.; Dreyer, R. N.; Yamashita, M.; Fenn, J. B. Anal. Chem. 1985, 57, 675-679. (2) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301. (3) Nordhoff, E.; Kirpekar, F.; Roepstorff, P. Mass Spectrom. Rev. 1996, 15, 76-138. (4) Limbach, P. A. Mass Spectrom. Rev. 1996, 15, 297-336.

5288 Analytical Chemistry, Vol. 70, No. 24, December 15, 1998

to be restricted to nucleic acid molecules of less than ∼700 kDa,5-7 Fuerstenau and Benner8 and Cheng et al.9 recorded ESI mass spectra of nucleic acids with molecular weights as high as several MDa. It is, however, quite difficult to perform mass measurements on nucleic acids by either ESI-MS or MALDI-MS, especially if the molecular mass surpasses 10 kDa.10 The main difficulties arise due to the formation of adducts between nucleic acids and ubiquitous cations such as sodium, potassium, or magnesium ions which persist even after ionization and desolvation during the ESI or MALDI process.11,12 The presence of adducts is a problem, particularly in the case of large nucleic acids, since the probability of forming adducts with counterions other than protons increases with increasing chain length of the nucleic acid molecule. Because of cation adduction, the molecular ions are dispersed among several different species of different m/z ratios, resulting in highly complex spectra and decreased sensitivity. Moreover, accurate mass measurements are hampered because the resolution of the generally employed quadrupole or ion-trap mass analyzers is insufficient to resolve the different adduct peaks of charge states higher than ∼20-. Effective cation removal is, therefore, a prime prerequisite to obtain more readily interpretable mass spectra, high mass accuracy, and satisfactory sensitivity. Reduction of cation adduction has been attempted by several different approaches. Stults and Marsters were able to improve the quality of ESI mass spectra of oligonucleotides up to 77-mers through replacement of metal cations by treatment with ammonium acetate, followed by ethanol-2-propanol precipitation.13 Nevertheless, this method is time-consuming and is not quantitative, relatively large amounts of sample are required, and multiple (5) Bai, J.; Liu, Y. H.; Lubman, D. M.; Siemieniak, D. Rapid Commun. Mass Spectrom. 1995, 8, 687. (6) Berkenkamp, S.; Kirpekar, F.; Hillenkamp, F. Science 1998, 281, 260262. (7) Liu, Y. H.; Bai, J.; Lubman, D. M.; Venta, P. J. Anal. Chem. 1995, 67, 34823490. (8) Fuerstenau, S. D.; Benner, W. H. Rapid Commun. Mass Spectrom. 1995, 9, 1528-1538. (9) Cheng, X.; Camp, D. G.; Wu, Q.; Bakhtiar, R.; Springer, D. L.; Morris, B. J.; Bruce, J. E.; Anderson, G. A.; Edmonds, C. G.; Smith, R. D. Nucleic Acids Res. 1996, 24, 2183-2189. (10) Portier, N.; Van Dorsselaer, A.; Cordier, Y.; Roch, O.; Bischoff, R. Nucleic Acids Res. 1994, 22, 3895-3903. (11) Smith, R. D.; Loo, J. A.; Edmonds, C. G.; Barinaga, C. J.; Udseth, H. R. Anal. Chem. 1990, 62, 882-889. (12) Bleicher, K.; Bayer, E. Biol. Mass Spectrom. 1994, 23, 320-322. (13) Stults, J. T.; Marsters, J. C. Rapid Commun. Mass Spectrom. 1991, 5, 359363. 10.1021/ac980791b CCC: $15.00

© 1998 American Chemical Society Published on Web 11/06/1998

Table 1. Sequences and Molecular Properties of Oligonucleotides Used in This Study

dT8 dT15 dT16 pdA12 pdA13 pdA14 pdA15 pdA16 pdA17 pdA18 sequence 1 sequence 2 sequence 3 sequence 4 sequence 5 sequence 6 sequence 7

sequence

length (nt)

Mr (Da)

T8 T15 T16 5′-phosphorylated A12 5′-phosphorylated A13 5′-phosphorylated A14 5′-phosphorylated A15 5′-phosphorylated A16 5′-phosphorylated A17 5′-phosphorylated A18 CCG TAG CCG ATG TGA CTC GTG CCG TAG CCC GAT GTG ACT CGT G CCG TAG TAC CCG ATG TGA CTC GTG CCG TAG TGC CCG ATG TGA CTC GTG AAT TAG GAC CCG ATG TGA CTC GTG TGA TGA TGA TGC GTG AAG ACA GTA GTT CCC TGA CTC TGA reverse complementary sequence in pBr322 at position 3419-3489

8 15 16 12 13 14 15 16 17 18 21 22 24 24 24 39 80

2371.59 4500.96 4805.15 3776.59 4089.81 4403.02 4716.24 5029.45 5342.67 5655.88 6438.27 6727.45 7344.86 7360.86 7392.92 12 061.99 24788.22

precipitations are necessary to eliminate cations in long-chain oligonucleotides.10,14,15 Greig and Griffey reported on the use of organic bases (e.g., triethylamine, imidazole, piperidine) as additives to eliminate cation adduction.16 Because complete removal of metal ions is usually not achieved, this approach is limited to small oligonucleotides of less than 10 kDa.10 Polymeric cationexchange beads, mixed with the sample on the tip of the FAB probe or the MALDI target, have been used to remove cations in FAB-MS 17 and MALDI-MS.18 A combination of ultrafiltration and treatment with an ammonium-loaded cation-exchanger has been applied for off-line sample preparation by Cheng et al.19 before ESI-MS analysis of oligonucleotides. Reversed-phase solid-phase extraction20 and reversed-phase HPLC have been utilized for offline21,22 or on-line23-25 desalting of DNA and RNA. Lui et al. electrosprayed oligonucleotides from solutions containing as much as 250 mM sodium chloride by passing the sample solution through an efficient on-line microdialysis cleanup system comprising a 20-cm-long cellulose hollow fiber of 200 µm i.d. and a molecular weight cutoff of 13 000.26 Although several methods have already been described for the desalting of nucleic acids samples before ESI-MS, there is still a (14) Naito, Y.; Ishikawa, K.; Koga, Y.; Tsuneyoshi, T.; Terunuma, H.; Arakawa, R. Rapid Commun. Mass Spectrom. 1995, 9, 1484-1486. (15) Tsuneyoshi, T.; Ishikawa, K.; Koga, Y.; Naito, Y.; Baba, S.; Terunuma, H.; Arakawa, R.; Prockop, D. J. Rapid Commun. Mass Spectrom. 1997, 11, 719722. (16) Greig, M. J.; Griffey, R. H. Rapid Commun. Mass Spectrom. 1995, 9, 97102. (17) Vollmer, D. L.; Gross, M. L. J. Mass Spectrom. 1995, 30, 113-118. (18) Nordhoff, E.; Ingendoh, A.; Cramer, R.; Hillenkamp, F.; Karas, M.; Stahl, B.; Overberg, A. Rapid Commun. in Mass Spectrom. 1992, 6, 771-776. (19) Cheng, X.; Gale, D.; Udseth, H. R.; Smith, R. D. Anal. Chem. 1995, 67, 586-593. (20) Deroussent, A.; Le Caer, J.-P.; Gouyette, A. Rapid Commun. Mass Spectrom. 1995, 9, 1-4. (21) Little, D. P.; Thannhauser, T. W.; McLafferty, F. W. Proc. Natl. Acad. Sci. U.S.A. 1995, 95, 2318-2322. (22) Little, D. P.; Chorush, R. A.; Speir, J. P.; Senko, M. W.; Kelleher, N. L.; McLafferty, F. W. J. Am. Chem. Soc. 1994, 116, 4893-4897. (23) Emmett, M. R.; Caprioli, R. M. J. Am. Soc. Mass Spectrom. 1994, 5, 605613. (24) Bleicher, K.; Bayer, E. Chromatographia 1994, 39, 405-408. (25) Apffel, A.; Chakel, J. A.; Fischer, S.; Lichtenwalter, K.; Hancock, W. S. Anal. Chem. 1997, 69, 1320-1325. (26) Liu, C.; Wu, Q.; Harms, A. C.; Smith, R. D. Anal. Chem. 1996, 68, 32953299.

need for new methods that are inexpensive, fast, quantitative, efficient, and easy to setup and handle, that minimize dilution of the sample, and that allow facile regeneration. We have, therefore, developed an on-line cation-exchange sample preparation system for the removal of cations. It is shown that this fast and simple method is applicable not only to the analysis of short- and longchain synthetic oligonucleotides but also to the quantitative determination of oligonucleotides and mixture analysis. EXPERIMENTAL SECTION Chemicals and Oligonucleotides. Sodium hydroxide (p.a.), hydrochloric acid (37%, p.a.), glacial acetic acid (p.a.), methanol (p.a.), and HPLC gradient-grade acetonitrile were obtained from Merck (Darmstadt, Germany). Triethylamine (TEA) was purchased from Fluka (Buchs, Switzerland). A 1 M stock solution of triethylammonium acetate, pH 7.0, was prepared by mixing equimolar amounts of triethylamine and glacial acetic acid. For preparation of all solutions, high-purity water (Epure, Barnstead Co., Newton, MA) was used. The standards of phosphorylated and nonphosphorylated oligonucleotides (pdA12-18, dT16) were purchased as sodium salts from Pharmacia (Uppsala, Sweden). Synthesized oligonucleotides, obtained from Microsynth (Balgach, Switzerland), were used without further purification (Table 1). The sequence of the synthesized 80-mer was the same as the reverse complementary sequence in the pBr322 plasmid from position 3410-3489 relative the Eco RI recognition site. Preparation of Cation-Exchange Microcolumns. The characteristics of the cation-exchange resins used in this study are summarized in Table 2. The Amberlite CG-120-II and Amberlite CG-50-II cation exchangers were obtained from Fluka, and the Dowex 50WX8 ion-exchange resin was from Serva (Heidelberg, Germany). Small particles present in the Amberlite CG-50-II resin were removed by 5-fold sedimentation. The preparation of the carboxylic acid-functionalized particles (ROMP-(COOH)2; ROMP ) ring-opening metathesis polymerization) from 1,4,4a,5,8,8ahexahydro-1,4,5,8-exo,endo-dimethanonaphthalene and endo,endo[2.2.1]bicyclohept-2-ene-5,6-dicarboxylic anhydride by ROMP has been described elsewhere.27 The ROMP-(COOH)2 particles were (27) Buchmeiser, M. R.; Atzl, N.; Bonn, G. K. J. Am. Chem. Soc. 1997, 119, 9166-9174.

Analytical Chemistry, Vol. 70, No. 24, December 15, 1998

5289

Table 2. Characteristics of Cation-Exchange Resins Used in this Study resin ROMP-(COOH)2 Amberlite CG-50-II Amberlite CG-120-II Dowex 50WX8 a

chemical composition

functional group

particle size (µm)

capacity (mequiv/g)a

poly(norborn-2-enedicarboxylic acidco-hexahydrodimethanonaphthalene) polymethacrylate/DVB PS/DVB PS/DVB

carboxylic acid

30-50

1.9

carboxylic acid sulfonic acid sulfonic acid

38-75 38-75 38-75

8.8 3.0 4.8

Determined by titration with 0.05 M NaOH.

extensively washed with 1 M sodium hydroxide to remove oligomeric carboxylic acids. All cation exchangers were converted to the hydrogen form by treatment of 500 mg of resin with hydrochloric acid (1 M, 50 mL) on a nutsch followed by washing with deionized water (150 mL). Subsequently, the resins were treated with a mixture of water, acetonitrile, and triethylamine (1:1:1, v/v, 30 mL). The resins were washed with deionized water until the eluate had a pH between 8 and 9 (∼1000 mL). Finally, the particles were suspended in 10 mM triethylamine in 50% water-50% acetonitrile (v/v, 50 mL). A 0.5-mL portion of this suspension was used to pack the resins into a 70-mm-long piece of Teflon tubining of 1.6 mm (1/16 in.) o.d. and 0.8 mm i.d. (Upchurch Scientific, Oak Harbor, WA) with the help of a 1-mL plastic syringe fitted with a 0.9-mm-o.d. stainless steel needle. The resin was retained by two filter disks cut from a sheet of Whatman GF/C glassfiber filter (Whatman, Kent, England) with the help of a 1/16-in. PEEK fingertight fitting (Upchurch). The two filter disks were held in place by a 1/16 in-1/16 in. stainless steel union (Swagelock, Solon, OH). After packing, the cation-exchange microcolumn was trimmed to a length of 45-50 mm with a razor blade. Electrospray Mass Spectrometry with On-Line CationExchange Sample Preparation. ESI-MS was performed on a Finnigan MAT TSQ 7000 triple-quadrupole mass spectrometer (Finnigan MAT, San Jose, CA) equipped with the electrospray ion source. For analysis with pneumatically assisted ESI, an electrospray voltage of 3.5-3.8 kV and a sheath gas pressure of 410 kPa were employed. The temperature of the heated capillary was set at 200 °C. Mass spectra were recorded by scanning the third quadrupole Q3; scan range and scan time are given in the figure captions. Total ion chromatograms and mass spectra were recorded on a DEC-Alpha 3000 workstation with the ICIS software version 7.01 (Finnigan). Mass calibration was performed in the positive ionization mode by direct infusion of a solution of horse heart myoglobin (Sigma, St. Louis, MO) and methionyl-arginylphenylalanyl-alanine (Finnigan). A syringe pump (model 980-532, Harvard Apparatus, South Natick, MA) equipped with a 250-µL glass syringe (Unimetrics, Shorewood, IL) was used for continuous infusion experiments. For flow injection analysis, samples were injected by means of a Rheodyne 7725 injector with a 5-µL sample loop (Cotati, CA) into a 3-5 µL/min flow of 10 mM triethylamine in 50% water-50% acetonitrile (v/v) provided by a micropump (model Rheos 4000, Karlskoga, Sweden). The inlet of the cation-exchange microcolumn was connected to the injector, and the column outlet was connected to a 300-mm piece of fused silica tubing (0.15 mm o.d. and 0.05 mm i.d.) which served as the electrospray capillary. 5290 Analytical Chemistry, Vol. 70, No. 24, December 15, 1998

Ion-Pair Reversed-Phase High-Performance Liquid Chromatography. The HPLC system consisted of a low-pressure gradient pump (model 480 GT, Gynkotek, Germering, Germany), a degasser (Knauer, Berlin, Germany), a column oven (Gynkotek) set at 50 °C, a diode array UV detector (model UVD-320, Gynkotek) set at 254 nm, a sample injection valve (model 8125, Rheodyne Inc., Cotati, CA) with a 20-µL sample loop, and a PCbased data system (GynkoSoft, Version 5.32, Gynkotek). Columns (60 × 4 mm i.d.) packed with alkylated poly-styrene/divinylbenzene beads were prepared according to the previously published protocol28 and are commercially available as DNASep columns from Transgenomic Inc. (Santa Clara, CA). Oligonucleotides were eluted with a 30-min linear gradient of 5-15% acetonitrile in 0.1 M triethylammonium acetate at a flow rate of 550 µL/min and a temperature of 50 °C. RESULTS AND DISCUSSION Characterization of the On-Line Ion-Exchange Sample Preparation System. The strategy applied for reducing the amount of adducts in nucleic acid samples embodies the on-line exchange of cations with triethylammonium ions in a triethylammonium-loaded cation-exchange microcolumn. Column dimensions need to be as small as possible to minimize sample dilution, yet large enough to provide sufficient ion-exchange capacity. Microcolumns of 0.4 mm i.d. could be fabricated with some experience, but they were prone to clogging during the packing procedure. Columns of 0.8 mm i.d.were, therefore, preferred because they were easy to pack with the polymer particles. In initial experiments, cation-exchange columns of 20 mm length were connected to the needle of a 100-µL syringe filled with the sample solution, and spectra were recorded by continuous infusion with a syringe pump. The minimum volume of sample solution required to obtain a stable signal during continuous infusion was 50-100 µL. Flow injection analysis, on the other hand, allowed the analysis of much lower sample volumes (0.5-5 µL) at the cost of a ∼3-fold dilution of the sample due to band broadening in the cation-exchange microcolumn. For flow injection analysis, a column length of 45 mm was necessary to allow connection of the column to the injector and electrospray capillary with the help of standard fingertight fittings. The constant flow of triethylamine solution was provided by either a syringe pump or an HPLC micropump, the latter offering the possibility of monitoring the column backpressure (typically 2-6 bar) and eliminating the need to repeatedly refill the syringe with triethylamine solution. (28) Huber, C. G.; Oefner, P. J.; Bonn, G. K. Anal. Biochem. 1993, 212, 351358.

Figure 1. Mass spectra of dT15 (a) without and (b) with on-line cation-exchange sample preparation. Inset, expanded view of the 3charge state. Cation-exchange microcolumn, ROMP-(COOH)2, 45 × 0.8 mm; scan, 500-2500 amu in 5 s; flow injection into 10 mM TEA in 50% water-50% acetonitrile (v/v), 3 µL/min; injection volume, 5 µL; sample, 400 pmol of dT15Na15.

The performance of the on-line sample preparation system was tested by flow injection analysis of 400 pmol of dT15-sodium salt without (Figure 1a) and with cation-exchange microcolumn (Figure 1b). Both mass spectra showed series of multiply charged ions with charge states from 2- to 7-. The extensive adduction of sodium cations, especially in the 2- and 3- charge states, is evident in the upper mass spectrum (Figure 1a), obtained without cation-exchange microcolumn. The presence of at least 13 sodium adduct signals in the 3- charge state of the pentadecamer (see inset in Figure 1a) indicates that sodium adduction occurs not only at the sugar-phosphate backbone but also at the nucleobases. This observation is consistent with the slight acidity of the hydrogen linked to N-3 in thymine (pKa ) 9.9) due to the two neighboring keto functions.29 Upon on-line cation exchange, cations are replaced with triethylammonium ions, which dissociate and evaporate during the ESI process, leaving the nucleic acid molecules in the hydrogen form. The absolute abundance of the most intense signal increased by a factor of 10, and the signalto-noise ratio changed by a factor 2.9 from 15:1 to 41:1 (Figure 1b). The almost complete elimination of sodium adduction allowed a precise molecular mass determination with a measured (29) Clauwaert, J.; Stockx, J. Z. Naturforsch. B 1968, 23, 25-30.

mass of 4502.0 ( 0.3 Da, which compares well to the theoretical mass of 4500.98 Da. The three small peaks appearing at higher m/z in the 3- charge state (see inset in Figure 1b) correspond to absolute mass differences of 14, 36, and 53 and are, therefore, not due to cation adducts. For the evaluation of different weak and strong cation-exchange resins, the presence of sodium adducts in a synthetic oligonucleotide was studied. Before packing, the dry resins were swollen in 10 mM triethylamine in 50% water-50% acetonitrile (v/v). To avoid swelling or shrinking of the polymer particles during operation of the microcolumns, it is mandatory to keep the concentration of acetonitrile in the eluents and washing solutions always constant. Cation-exchangers of 30-80-µm particle size were chosen because the resultant backpressure (2-6 bar) was small enough to allow packing into 45- × 0.8-mm i.d. Teflon tubes without the need for a high-pressure pump. The performance of different cation-exchangers was investigated by comparing the molecular mass region of the deconvoluted mass spectra of a dT16 raw product (Microsynth) recorded without cation-exchanger and with microcolumns packed with ROMP-(COOH)2, Dowex 50XW8, Amberlite CG-50-II, and Amberlite CG-120-II, respectively. The ratio of signal intensity of monosodium adduct to protonated oligonucleotide was reduced from 1:8 without cation exchange to practically background noise level with on-line cation exchange (ratios between 1:50 and 1:30). Among the different strong and weak cation-exchange resins, no significant difference in terms of effectiveness of cation removal, signal intensity, or charge state distribution was observed. Moreover, the ion-exchange capacity of the individual resins (Table 2) is high enough in all cases to guarantee efficient desalting. The minimum amount of oligonucleotide that can be loaded onto a 45- × 0.8-mm-i.d. ROMP-(COOH)2 microcolumn without loss in desalting efficiency was characterized by continuous infusion of 200 µL of a 0.2 µg/µL solution of dT16-Na16 (Pharmacia) at a flow rate of 5 µL/min. Without the cation-exchange column, Na1-Na7 adducts were observed, and the monosodium adduct was the most abundant species in the deconvoluted mass spectrum (ratio of 1:0.82 for the intensities of the 4828 Da:4805 Da signals). With on-line cation exchange, sodium adduction was efficiently reduced, giving a ratio of 1:40 for the intensities of the 4828 Da: 4805 Da signals. This ratio did not significantly change even after infusion of 10 nmol of oligonucleotide, demonstrating that at least 10 nmol of oligonucleotide, corresponding to 200 flow injection analyses of 50-pmol sample size, can be analyzed without a decrease in the effectiveness of cation removal. After extensive use, the cation-exchange microcolumns were reconverted to the triethylammonium form by flushing the columns with 50% water50% acetonitrile (v/v, 250 µL), nitric acid (1 M in 50% water-50% acetonitrile v/v, 150 µL), and 50% water-50% acetonitrile (v/v, 250 µL) and subsequently reequilibrated with triethylamine solution (10 mM, 1000 µL) in 50% water-50% acetonitrile (v/v). Nevertheless, since the regeneration procedure required significantly more time than the packing of a new microcolumn and since the amount of resin required for one microcolumn was very low (∼5 mg), replacement of used cation-exchange microcolumns was usually preferred. Analysis of Synthetic Oligonucleotides. Several oligonucleotide preparations of varying concentration (5-50 µM), salt Analytical Chemistry, Vol. 70, No. 24, December 15, 1998

5291

Table 3. Molecular Masses of Oligonucleotides oligonucleotidea

measured mass ( sda (Da)

relative deviationc (%)

sequence 1 sequence 2 sequence 3 sequence 4 sequence 5 sequence 6 sequence 7

6439.7 ( 0.9 6728.5 ( 0.6 7345.5 ( 1.0 7361.4 ( 0.9 7394.4 ( 0.9 12063.0 ( 0.9 24790.4 ( 2.6

0.022 0.015 0.012 0.0073 0.020 0.0084 0.0088

a Sequences in Table 1. b sd, standard deviation calculated by the deconvolution program from the series of multiply charged ions. c Deviation from the mass calculated from the sequence (Table 1).

Figure 3. Mass spectrometric analysis of a synthetic 80-mer oligonucleotide raw product (a) without and (b) with on-line cation removal. Inset, deconvoluted mass spectra. Cation-exchange microcolumn, ROMP-(COOH)2, 45 × 0.8 mm; scan, 500-1500 amu in 5 s; flow injection into 10 mM TEA in 50% water-50% acetonitrile (v/ v), 5 µL/min; injection volume, 2.5 µL; sample, 250 pmol of sequence 7 (Table 1).

Figure 2. Negative ESI-MS of a synthetic 39-mer oligonucleotide: (a) original mass spectrum, (b) deconvoluted mass spectrum. Cationexchange microcolumn, ROMP-(COOH)2 45 × 0.8 mm; scan, 5001500 amu in 5 s; flow injection into 10 mM TEA in 50% water-50% acetonitrile (v/v), 3 µL/min; injection volume, 5 µL; sample, sequence 6 (Table 1).

content, and purity (45-98%) ranging in size from 21 to 80 nucleotides (nt) were analyzed by ESI-MS using the on-line desalting procedure. The measured molecular masses are summarized in Table 3. It can be seen that all masses were obtained with accuracy between 0.007 and 0.02% irrespective of length or purity of the analyzed oligonucleotide. The values compare well to those reported in the literature for the mass determination of oligonucleotides by ESI-MS with a triple-quadrupole mass spectrometer.10 In most oligonucleotide samples, n - 1 and n - 2 5292 Analytical Chemistry, Vol. 70, No. 24, December 15, 1998

failure sequences were readily detected. Figure 2 shows as an example the electrospray mass spectrum of a 39-mer (measured mass 12 063.1 Da) with a series of multiply charged ions from 12- to 22-. Failure sequences are detected at 11 757.9 and 11 427.9 Da. The mass differences of 305.2 and 330.0 Da closely correspond to the masses of deoxythymidine monophosphate (304.2 Da) and deoxyguanosine monophosphate (329.2 Da), respectively, the first two nucleotides at the 5′-end of sequence 6, which clearly demonstrates the utility of ESI-MS for the characterization of synthetic oligonucleotides. The longer the nucleic acid molecule, the more compelling becomes the adequate removal of cations before ESI-MS analysis. As reported by Portier et al.,10 a 72-mer had to be desalted by 3-fold treatment with 10 M ammonium acetate followed by ethanol precipitation in order to get high-quality mass spectra. We used a synthetic 80-mer to examine the applicability of the on-line cationexchange desalting procedure for sample preparation of large oligonucleotides (>20 000 Da). Figure 3 compares the mass spectra of the unpurified 80-mer (theoretical mass 24 788.22 Da) recorded without (Figure 3a) and with (Figure 3b) the on-line cation-exchange system. The deconvoluted mass spectrum of the nondesalted sample revealed a broad peak with a molecular mass

Figure 5. Peak area reproducibility test by repetitive injection of 2.0 µL of dT8. Cation-exchange microcolumn, ROMP-(COOH)2, 45 × 0.8 mm; selected ion monitoring at 592.9, 789.5, and 473.5 amu; scan time, 5 s; flow injection into 10 mM TEA in 50% water-50% acetonitrile (v/v), 5 µL/min; injection volume, 2.0 µL; sample, 46 pmol of dT8.

Figure 4. Analysis of a synthetic 80-mer oligonucleotide after preparative fractionation by ion-pair reversed-phase HPLC (a) without and (b) with on-line cation removal. Inset, deconvoluted mass spectra. Cation-exchange microcolumn, ROMP-(COOH)2, 45 × 0.8 mm; scan, 500-1500 amu in 5 s; flow injection into 10 mM TEA in 50% water50% acetonitrile (v/v), 5 µL/min; injection volume, 2.5 µL; sample, sequence 7 (Table 1).

of 24 906 Da, which is 115 Da higher than the theoretical value (see inset in Figure 3a). The peak broadening and shift to higher mass is due to sodium adduction and the presence of another unresolved signal from a byproduct. With the desalted sample, the correct molecular mass of the target product (24 790.4 Da) is obtained. The observed byproduct of 24 861.0 Da is likely the result of incomplete deprotection of the oligonucleotide after solidphase synthesis. The mass difference of 70.6 Da possibly corresponds to an isobutyryl protecting group that has been used to protect the amino group in deoxyguanosine-phosphoramidite during synthesis. Nevertheless, at this point it cannot be rationalized why only one protecting group remains unhydrolyzed in a molecule with a total of 28 G nucleotides. The high chemical background in the spectrum of the unpurified 80-mer did not allow the identification of failure sequences, although the multitude of noise between the major signals indicated the presence of truncated sequences. To investigate the influence of sample purity on spectrum quality without and with on-line cation-exchange, 10 µg of crude 80-mer was fractionated by ion-pair reversed-phase HPLC using a column packed with a micropellicular poly-styrene/divinylben-

zene stationary phase.28,30 In the chromatogram, two major peaks and a number of small peaks before the main product indicating the presence of failure sequences were observed. Without online desalting, a molecular mass of 24 822.0 Da was measured with the fraction containing the main product, which is still 32 Da higher than the theoretical mass because of sodium adduction (Figure 4a). With on-line desalting, however, the correct mass of 24 790.1 Da is obtained. A comparison of Figures 4b and 3b demonstrates that there is no difference in measured mass between the raw product and the product purified by micropreparative ion-pair reversed-phase HPLC. However, spectrum quality is greatly enhanced upon chromatographic purification of the raw product because the chemical background in the spectrum is substantially reduced. This allowed even the detection of failure sequences in the synthetic 80-mer (see inset in Figure 4). Quantitative Analysis of Oligonucleotides. So far, much of the attention in mass spectrometric analysis of nucleic acids focused on the accurate determination of molecular mass and qualitative detection of contaminating sequences. A relative quantification of contaminating oligonucleotides by ESI-MS has been accomplished by plotting the intensity ratios of 27- and 28mer oligonucleotides as a function of the concentration ratios and a fairly good correlation was observed.10 Nevertheless, absolute quantitative information about the sample composition is of utmost importance in many experiments dealing with nucleic acids. Therefore, we evaluated the performance of flow injection analysis with on-line cation-exchange sample preparation for quantitative analysis of nucleic acids. The reproducibility of peak areas was checked by repeated injections of dT8 using the mass spectrometer in the selected-ion monitoring mode (Figure 5). Since the (30) Huber, C. G.; Stimpfl, E.; Oefner, P. J.; Bonn, G. K. LC-GC 1996, 14, 114127.

Analytical Chemistry, Vol. 70, No. 24, December 15, 1998

5293

Figure 7. Analysis of a mixture of five heterooligonucleotides: (a) original mass spectrum, (b) deconvoluted mass spectrum. Cationexchange microcolumn, Amberlite CG-120-II, 45 × 0.8 mm; scan, 500-1500 amu in 5 s; flow injection into 10 mM TEA in 50% water50% acetonitrile (v/v), 5 µL/min; injection volume, 5.0 µL; sample, sequences 1-5 (Table 1), ∼20 pmol of each.

Figure 6. Mass spectra of a mixture of a homologous series of phosphorylated oligodeoxyadenylic acids: (a) original mass spectrum, (b) deconvoluted mass spectrum. Cation-exchange microcolumn, Dowex 50XW8, 45 × 0.8 mm; scan, 250-1500 amu in 5 s; flow injection into 10 mM TEA in 50% water-50% acetonitrile (v/v), 5 µL/ min; injection volume, 5.0 µL; sample, pdA12-18, ∼75 pmol each. Peak assignments correspond to charge state and number of adenosine units in the oligonucleotide, respectively.

efficiency of the electrospray process strongly depends on the composition of the electrosprayed solution, it is obligatory to dissolve the oligonucleotide sample in the same solvent as the electrospray solution. To keep the composition of the sample solution the same as the solvent for flow injection analysis, aqueous oligonucleotide solutions were diluted with an equal volume of 20 mM TEA in acetonitrile. With these precautions, the peak areas of six injections of 46 pmol dT8 were reproducible with a relative standard deviation of 8.8%. In the selected-ion5294 Analytical Chemistry, Vol. 70, No. 24, December 15, 1998

monitoring mode, a 16-mer was determined with a lower mass detection limit of 1 ng (400 fmol) at a signal-to-noise ratio of 3:1. The plot of peak areas versus concentration showed good linearity (R2 ) 0.9988) over a range of at least 2 orders of magnitude (0.220 µM dT16) with the equation of the linear regression line being Y ) 296.2X - 22.72 (Y, peak area [arbitrary units]; X, amount of injected oligonucleotide [pmol]; N ) 13). Analysis of Oligonucleotide Mixtures. We finally investigated the performance of ESI-MS with on-line cation removal for the characterization of nucleic acids differing by a single nucleotide unit and for nucleic acids of the same chain length but different base composition. Complex mass spectra consisting of various series of multiply charged ions make data analysis more difficult with increasing size and number of mixture components.19 Since the presence of cation adducts would further increase the complexity of the mass spectra, removal of cations is mandatory, particularly in mixture analysis. Figure 6a depicts the mass spectrum of a series of homologous phosphorylated oligodeoxyadenylic acids, 12-18 nt in length. Sixteen signals were observed in the original mass spectrum for the seven components present in the mixture. Charge states and molecular masses were unambiguously assigned by means of the deconvolution software (Figure 6b). For all seven oligonucleotides, at least three different charge states were detected in the m/z range of 500-1500 amu. Some of the charge states overlap, i.e., the 6- charge state of pdA12, 7- of pdA14, 8- of pdA16, and 9- of pdA18 in the signal at

m/z 628 as well as the 5- charge state of pdA15 and 4- of pdA12 in the signal at m/z 942. In another experiment, a mixture of five different heterooligonucleotides (22-24-mers, 20 pmol each) was analyzed (Figure 7a). Despite the high complexity of the spectrum and the relatively low signal-to-noise ratio, the correct masses of all five oligonucleotides were obtained after deconvolution (Figure 7b). Sequence 3 (measured mass 7345.5 Da) differs from sequence 5 (measured mass 7361.4 Da) by the mutation of AfG, and the measured mass difference of 15.9 Da is in good agreement with the theoretical mass difference of 16.0 Da. Since the smallest possible mass difference for a point mutation in deoxynucleic acids is 9.0 Da (TfA) and the two peaks of 15.9-Da mass difference are resolved to baseline, it can be concluded that all four possible point mutations are detectable by ESI-MS with on-line cationexchange sample preparation. In sequence 5 (measured mass 7394.4 Da), the CCG triplet at the 5′-end and T at position 7 of sequence 4 have been substituted with AAT and G, respectively (theoretical mass difference 48.1, measured mass difference 48.9 Da). Sequence 1 (measured mass 6438.7 Da) and sequence 2 (measured mass 6728.5 Da) are internal deletion sequences of sequence 4, where TA (theoretical mass difference 617.4 Da, measured mass difference 617.0) and TAC (theoretical mass difference 906.6 Da, measured mass difference 905.8 Da), respectively, are missing. CONCLUSIONS Adequate sample preparation is of prime importance for the acquisition of high-quality mass spectra using ESI-MS. The

present work demonstrates that on-line sample preparation using microcolumns packed with weak or strong cation-exchange resins is an efficient method for removal of cations in nucleic acids samples prior to analysis by ESI-MS. One laboratory-made and several commercially available cation-exchange resins of 30-80µm particle size are shown to be useful as packing materials, and no difference in effectiveness of desalting among the various resins is observed. The sample preparation system is fast and easy to set up without the need for specialized or expensive equipment. The applicability of the system for the confirmation of the identity and the determination of the purity of nucleic acids by means of ESI-MS is illustrated by the analysis of oligonucleotides ranging in size from 8 to 80 nt. The high tolerance for impurities such as failure sequences and salts, the low cost of consumables, the short time required for a single measurement, and the possibility to obtain quantitative information about mixtures make the method very attractive for routine mass spectrometric analysis of picomole amounts of nucleic acids. ACKNOWLEDGMENT This work was supported by grants from the Austrian Science Fund (P-11740 GEN and P12963 GEN).

Received for review July 16, 1998. Accepted September 29, 1998. AC980791B

Analytical Chemistry, Vol. 70, No. 24, December 15, 1998

5295