Electron Photodetachment Dissociation of DNA Polyanions in a

using the DINAMelt web server,53 http://www.bioinfo.rpi.edu/applications/hybrid/). ..... Fully annotated EPD spectra of ssB and ssD as Figures S1 ...
0 downloads 0 Views 213KB Size
Download PDF
Anal. Chem. 2006, 78, 6564-6572

Electron Photodetachment Dissociation of DNA Polyanions in a Quadrupole Ion Trap Mass Spectrometer Vale´rie Gabelica,*,† Thibault Tabarin,‡ Rodolphe Antoine,‡ Fre´de´ric Rosu,† Isabelle Compagnon,‡ Michel Broyer,‡ Edwin De Pauw,† and Philippe Dugourd*,‡

Laboratoire de Spectrome´ trie de Masse, Universite´ de Lie` ge, Institut de Chimie, Bat B6c, B-4000 Lie` ge, Belgium, and Laboratoire de Spectrome´ trie Ionique et Mole´ culaire, UMR 5579 CNRS et Universite´ Lyon 1, 43 Bd du 11 Novembre, F-69622 Villeurbanne, France

We hereby explore the effects of irradiating DNA polyanions stored in a quadrupole ion trap mass spectrometer with an optical parametric oscillator laser between 250 and 285 nm. We studied DNA 6-20-mer single strands and 12-base pair double strands. In all cases, laser irradiation causes electron detachment from the multiply charged DNA anions. Electron photodetachment efficiency directly depends on the number of guanines in the strand, and maximum efficiency is observed between 260 and 275 nm. Subsequent collision-induced dissociation (CID) of the radical anions produced by electron photodetachment results in extensive fragmentation. In addition to neutral losses, a large number of fragments from the w, d, a•, and z• ion series are obtained, contrasting with the w and (a-base) ion series observed in regular CID. The major advantage of this technique, coined electron photodetachment dissociation (EPD) is the absence of internal fragments, combined with good sequence coverage. EPD is therefore a highly promising approach for de novo sequencing of oligonucleotides. EPD of nucleic acids is also expected to give specific radical-induced strand cleavages, with conservation of other fragile bonds, including noncovalent bonds. In effect, preliminary results on a DNA hairpin and on double strands suggest that EPD could also be used to probe intra- and intermolecular interactions in nucleic acids. The analysis of biomolecules by mass spectrometric techniques has been greatly supported by the development of electrospray ionization (ESI) sources.1-3 For nucleic acids, large single strands up to the megadalton can be detected.4,5 Furthermore, higher* To whom correspondence should be addressed. E-mail: [email protected]; [email protected]. † Universite´ de Lie`ge. ‡ UMR 5579 CNRS et Universite ´ Lyon 1. (1) Yamashita, M.; Fenn, J. B. J. Phys. Chem. 1984, 88, 4451-4459. (2) Whitehouse, M.; Dreyer, R. N.; Yamashita, M.; Fenn, J. B. Anal. Chem. 1985, 57, 675-679. (3) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Science 1989, 246, 64-71. (4) Chen, R.; Cheng, X.; Mitchell, D. W.; Hofstadler, S. A.; Wu, Q.; Rockwood, A. L.; Sherman, M. G.; Smith, R. D. Anal. Chem. 1995, 67, 1159-1163. (5) Schultz, J. C.; Hack, C. A.; Benner, W. H. J. Am. Soc. Mass Spectrom. 1998, 9, 305-313.

6564 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

order nucleic acid structures such as double strands,6,7 triple helices,8 or quadruplexes8-10 can be transferred to the mass analyzer,11 and both experimental and theoretical evidence is accumulating that these structures are maintained in the gas phase.12-18 In complement to molecular mass measurements, various activation techniques have been used to fragment oligonucleotides in tandem mass spectrometry for sequence confirmation and de novo sequence determination, as well as for characterization of the intrinsic forces governing structure and function.11,19-22 The different activation techniques and the corresponding fragmentation patterns of single strands have been reviewed recently by Wu and McLuckey.23 Collision-induced dissociation (CID) is by far the most widely used fragmentation method. In CID, fragmentation is initiated by the loss of neutral or charged base. The strand subsequently fragments into complementary w and (a-base) ions, according to the commonly adopted nomenclature summarized in Scheme 1.23-25 Infrared multiple photon dissociation, which also involves vibrational excitation, leads to a frag(6) Light-Wahl, K. J.; Springer, D. L.; Winger, B. E.; Edmonds, C. G.; Camp, D. G.; Thrall, B. D.; Smith, R. D. J. Am. Chem. Soc. 1993, 115, 803-804. (7) Bayer, E.; Bauer, T.; Schmeer, K.; Bleicher, K.; Maier, M.; Gaus, H.-J. Anal. Chem. 1994, 66, 3858-3863. (8) Rosu, F.; Gabelica, V.; Houssier, C.; Colson, P.; De Pauw, E. Rapid Commun. Mass Spectrom. 2002, 16, 1729-1736. (9) Goodlett, D. R.; Camp, D. G., II; Hardin, C. C.; Corregan, M.; Smith, R. D. Biol. Mass Spectrom. 1993, 22, 181-183. (10) David, W. M.; Brodbelt, J.; Kerwin, S. M.; Thomas, P. W. Anal. Chem. 2002, 74, 2029-2033. (11) Hofstadler, S. A.; Griffey, R. H. Chem. Rev. 2001, 101, 377-390. (12) Schnier, P. D.; Klassen, J. S.; Strittmatter, E. F.; Williams, E. R. J. Am. Chem. Soc. 1998, 120, 9605-9613. (13) Gabelica, V.; De Pauw, E. J. Mass Spectrom. 2001, 36, 397-402. (14) Gabelica, V.; De Pauw, E. Int. J. Mass Spectrom. 2002, 219, 151-159. (15) Rueda, M.; Kalko, S. G.; Luque, F. J.; Orozco, M. J. Am. Chem. Soc. 2003, 125, 8007-8014. (16) Gidden, J.; Baker, E. S.; Ferzoco, A.; Bowers, M. T. Int. J. Mass Spectrom. 2004, 240, 200-210. (17) Gidden, J.; Ferzoco, A.; Shammer Baker, E.; Bowers, M. T. J. Am. Chem. Soc. 2004, 126, 15132-15140. (18) Rueda, M.; Luque, F. J.; Orozco, M. J. Am. Chem. Soc. 2006, 128, 36083619. (19) Huber, C. G.; Oberacher, H. Mass Spectrom. Rev. 2001, 20, 310-343. (20) Crain, P. F.; McCloskey, J. A. Curr. Opin. Biotechnol. 1998, 9, 25-34. (21) Nordhoff, E.; Kirpekar, F.; Roepstorff, P. Mass Spectrom. Rev. 1996, 15, 67-138. (22) Beck, J.; Colgrave, M. L.; Ralph, S. F.; Sheil, M. M. Mass Spectrom. Rev. 2001, 20, 61-87. (23) Wu, J.; McLuckey, S. A. Int. J. Mass Spectrom. 2004, 237, 197-241. 10.1021/ac060753p CCC: $33.50

© 2006 American Chemical Society Published on Web 08/22/2006

Scheme 1

mentation pattern similar to the CID case. In contrast, electron capture dissociation (ECD) produces nucleotide radical cations, and the fragmentation follows different patterns,26,27 giving rise mainly to d radical cations, a/z ions, and c/x and (c/x-base) ions (a and z, c and x ions have the same mass in palindromic sequences). However, generation of positively charged oligonucleotide is generally less efficient than the generation of the corresponding anions.27 The electron detachment dissociation (EDD) of anions, caused by bombardment with >10 eV electrons,28 is an alternative method to ECD. EDD generates radical anions and, for polynucleotides, leads to fragmentation pathways similar to those obtained by ECD.29 In the case of peptides and proteins, the fragmentation pathways accessed by radical ions in ECD allow extensive sequence coverage,30 de novo sequencing of large proteins,31 and characterization of posttranslational modifications.32-35 Several other methods for radical ion generation and fragmentation were explored. These approaches include electron-transfer dissociation by ion-ion reactions in a quadrupole ion trap,36-38 protein chemical derivatization or formation of ternary complexes to produce radicals upon CID,39-41 fast atom bombardment of ions in a quadrupole ion trap,42 or UV laser photodissociation.43-52 Recently, (24) McLuckey, S. A.; Van Berkel, G. J.; Glish, G. L. J. Am. Soc. Mass Spectrom. 1992, 3, 60-70. (25) McLuckey, S. A.; Habibi-Goudarzi, S. J. Am. Chem. Soc. 1993, 115, 1208512095. (26) Hakansson, K.; Hudgins, R. R.; Marshall, A. G.; O’Hair, R. A. J. Am. Soc. Mass Spectrom. 2003, 14, 23-41. (27) Schultz, K. N.; Hakansson, K. Int. J. Mass Spectrom. 2004, 234, 123-130. (28) Budnik, B. A.; Haselmann, K. F.; Zubarev, R. A. Chem. Phys. Lett. 2001, 342, 299-302. (29) Yang, J.; Mo, J. J.; Adamson, J. T.; Hakansson, K. Anal. Chem. 2005, 77, 1876-1882. (30) Cooper, H. J.; Hakansson, K.; Marshall, A. G. Mass Spectrom. Rev. 2005, 24, 201-222. (31) Horn, D. M.; Ge, Y.; McLafferty, F. W. Anal. Chem. 2000, 72, 4778-4784. (32) Kjeldsen, F.; Haselmann, K. F.; Budnik, B. A.; Sorensen, E. S.; Zubarev, R. A. Anal. Chem. 2004. (33) Pesavento, J. J.; Kim, Y.-B.; Taylor, G. K.; Kelleher, N. L. J. Am. Chem. Soc. 2004, 126, 3386-3387. (34) Shi, S. D.; Hemling, M. E.; Carr, S. A.; Horn, D. M.; Lindh, I.; McLafferty, F. W. Anal. Chem. 2001, 73, 19-22. (35) Migorodskaya, E.; Roepstorff, P.; Zubarev, R. A. Anal. Chem. 1999, 71, 4431-4436. (36) McLuckey, S. A.; Stephenson, J. L. Mass Spectrom. Rev. 1998, 17, 369407. (37) Syka, J. E.; Coon, J. J.; Schroeder, M. J.; Shabanowitz, J.; Hunt, D. F. Proc. Natl. Acad. Sci. U. S. A 2004, 101, 9528-9533. (38) Coon, J. J.; Shabanowitz, J.; Hunt, D. F.; Syka, J. E. J. Am. Soc. Mass Spectrom. 2005, 16, 880-882. (39) Chu, I. K.; Rodriguez, C. F.; Rodriguez, F.; Hopkinson, A. C.; Siu, K. W. M. J. Am. Soc. Mass Spectrom. 2001, 12, 1114-1119. (40) Hodyss, R.; Cox, H. A.; Beauchamp, J. L. J. Am. Chem. Soc. 2005, 127, 12436-12437. (41) Barlow, C. K.; McFadyen, W. D.; O’Hair, R. A. J. J. Am. Chem. Soc. 2005, 127, 6109-6115.

CID on radical precursor ions produced by laser-induced dissociation was investigated for peptides and phosphopeptides.51,52 These experiments show a fragmentation pattern different from direct CID in nonradical peptides. Using a similar approach, in this article, we explore the fragmentation pattern of negatively charged DNA single strands and double strands using UV light close to the DNA absorption in solution (∼260 nm). In all cases, laser irradiation causes efficient and selective electron detachment from the DNA. Subsequent CID of the produced radicals results in highly efficient fragmentation via pathways completely different from that of the CID of the corresponding even-electron species. This new technique, which we coined electron photodetachment dissociation (EPD), has promising applications both for oligonucleotide de novo sequencing and for structural studies of nucleic acid higher-order structures. EXPERIMENTAL SECTION Sample Preparation. All oligonucleotides used in this study were synthesized by Eurogentec (Liege, Belgium), with Oligold quality. Sequences A-C are self-complementary and allow the formation of double strands when stored at 4 °C for >12 h. Sequence E is capable of forming an intramolecular hairpin by base pairing between the five terminal bases (∆G° ) -4.0 kcal/ mol at 20 °C, as calculated using the DINAMelt web server,53 http://www.bioinfo.rpi.edu/applications/hybrid/). The oligonucleotide sequences are summarized in Table 1. The oligonucleotides were solubilized in doubly distilled water to obtain stock solutions with a concentration of 200 µmol/L. Aliquots of this solution were used to prepare 100 µM single-stranded DNA (i.e., 50 µM duplex concentration for A-C) in 100 mM ammonium acetate. Storing the solutions at 4 °C is sufficient to form the duplexes. Single strands and double strands from sequences A-C were analyzed by spraying the same solutions, but selecting different parent ions. The final solution injected in electrospray has a concentration of 20 µmol/L DNA (duplex concentration for sequences A-C, single-strand concentration for sequences D-F), in 50:50 (v/v) aqueous ammonium acetate 50 mM/analytical grade methanol. (42) Misharin, A. S.; Silivra, O. A.; Kjeldsen, F.; Zubarev, R. A. Rapid Commun. Mass Spectrom. 2005, 19, 2163-2171. (43) Thompson, M. S.; Cui, W. D.; Reilly, J. P. Angew. Chem., Int. Ed. 2004, 43, 4791-4794. (44) Kim, T. Y.; Thompson, M. S.; Reilly, J. P. Rapid Commun. Mass Spectrom. 2005, 19, 1657-1665. (45) Cui, W. D.; Thompson, M. S.; Reilly, J. P. J. Am. Soc. Mass Spectrom. 2005, 16, 1384-1398. (46) Williams, E. R.; Furlong, J. J. P.; McLafferty, F. W. J. Am. Soc. Mass Spectrom. 1990, 1, 288-294. (47) Guan, Z.; Kelleher, N. L.; O’Connor, P. B.; Aaserud, D. J.; Little, D. P.; McLafferty, F. W. Int. J. Mass Spectrom. Ion Processes 1996, 157-158, 357364. (48) Moon, J. H.; Yoon, S. H.; Kim, M. S. Rapid Commun. Mass Spectrom. 2005, 19, 3248-3252. (49) Oh, J. Y.; Moon, J. H.; Kim, M. S. Rapid Commun. Mass Spectrom. 2004, 18, 2706-2712. (50) Oh, J. Y.; Moon, J. H.; Lee, Y. H.; Hyung, S. W.; Lee, S. W.; Kim, M. S. Rapid Commun. Mass Spectrom. 2005, 19, 1283-1288. (51) Tabarin, T.; Antoine, R.; Broyer, M.; Dugourd, P. Rapid Commun. Mass Spectrom. 2005, 19, 2883-2892. (52) Lemoine, J.; Tabarin, T.; Antoine, R.; Broyer, M.; Dugourd, P. Rapid Commun. Mass Spectrom. 2006, 20, 507-511. (53) Markham, N. R.; Zuker, M. Nucleic Acids Res. 2005, 33, W577-W581.

Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

6565

Table 1. Summary of the Oligonucleotide Sequences notation

length

sequence

ssA ssB ssC ssD ssE ssF dsA dsB dsC

12-mer 12-mer 12-mer 15-mer 20-mer 6-mer (12-mer)2 (12-mer)2 (12-mer)2

5′-CGTAAATTTACG-3′ 5′-CGCGAATTCGCG-3′ 5′-CGCGGGCCCGCG-3′ 5′-CCAGGACGCCTCAGA-3′ 5′-CCAGGTCTGAGGCGTCCTGG-3′ a 5′-GGGTTT-3′ (5′-CGTAAATTTACG-3′)2 (5′-CGCGAATTCGCG-3′)2 (5′-CGCGGGCCCGCG-3′)2

a The underlined bases in ssE indicate the complementary bases allowing hairpin formation.

Mass Spectrometry and EPD Experiment. The experimental setup is based on the coupling of a commercial quadrupole ion trap (QIT) mass spectrometer (LCQ DUO with the MSn option, ThermoElectron, San Jose´, CA) with an optical parametric oscillator (OPO) laser. The vacuum chamber and the central ring electrode of the mass spectrometer were modified to allow the injection of UV and visible lights54 (P. Dugourd, R. Antoine, M. Broyer, and F. O. Talbot, International Patent, application number PCTFR2005003142, 2005). The laser beam enters the vacuum chamber through a quartz window flange. Then it enters and exits the trap through 3-mm apertures in the ring electrode. A sapphire window is placed on the entrance aperture and glued to the ring electrode to maintain the helium buffer pressure in the trap. A fiber optic is glued in the opposite hole. The helium pressure inside the QIT is maintained at a constant value of 1.5 mTorr, ensuring conservation of mass specifications of the QIT (resolution, accuracy, MS2, CID). The laser is a tunable Panther OPO laser with frequency doubling pumped by a 355-nm Nd3+:YAG PowerLite 8000 (both from Continuum, Santa Clara, CA) operated at a repetition rate of 20 Hz. An electromechanical shutter triggered on the rf signal of the QIT synchronizes the laser irradiation with the MS/MS events conducted in the QIT. To perform laser irradiation for a given amount of time, we add a MSn step with activation amplitude of 0%, during which the shutter is open. For all experiments, nitrogen is used as electrospray sheath gas and helium as damping and collision gas. The mass spectra are recorded in the automatic gain control mode, with a constant qz of 0.25. The electrospray source conditions were -3.9 kV on the spray needle, -10 V and 200 °C on the heated capillary, and +40 V as tube lens offset. Experiments involving duplex isolation were conducted on the uneven charge state 5-. For strands A-C, zoom scans confirmed that single-strand charge states 3- and 4- consisted of single strands only. No traces of dimers were found in the ESI full scan spectra of strands D and E. RESULTS AND DISCUSSION Electron Photodetachment and Production of the Radical Ions. We first recognized the possibility of producing and isolating radical ions while studying the three DNA duplex sequences dsA, dsB, and dsC (see Table 1), which had been studied previously by CID in a Q-TOF and in a quadrupole ion trap.55 When subjected to 2-s laser irradiation (40 laser shots) at 260 nm, the double(54) Talbot, F. O.; Tabarin, T.; Antoine, R.; Broyer, M.; Dugourd, P. J. Chem. Phys. 2005, 122.

6566 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

Figure 1. Comparison of electron photodetachment efficiency for the self-complementary 12-mer duplexes [dsA - 5H]5- (a), [dsB 5H]5- (b), and [dsC - 5H]5- (c) under 2-s irradiation at 260 nm (same injection conditions and same laser power). The asterisk indicates the parent ion.

Figure 2. Normalized electron detachment efficiency of [dsC]5- as a function of the laser wavelength.

stranded DNA undergoes charge reduction, as shown in Figure 1. No other fragments are found, contrary to what was observed with DNA irradiation at 193 nm in the pioneering work of McLafferty and co-workers.47 For the 12-mer duplexes, the ion trap resolution and mass accuracy do not allow unambiguous demonstration of electron detachment (loss one electron versus loss of one electron plus one hydrogen), but experiments with a short 6-mer single strand (see below) demonstrate that electron detachment is indeed happening. We initially chose the wavelength 260 nm because it is the maximum absorption of DNA in solution, and this wavelength is routinely used for DNA concentration measurements. We also performed experiments with varying the wavelength of the OPO laser around 260 nm in order to determine the wavelength at which electron detachment is optimal. The laser power was measured at each wavelength, and the observed electron detachment efficiency was normalized with a factor proportional to the laser fluence (Φ), which is directly proportional to the laser power and inversely proportional to the laser wavelength. The electron photodetachment cross section is calculated using eq 1. The electron detachment efficiency spectrum is subsequently normal(55) Gabelica, V.; De Pauw, E. J. Am. Soc. Mass Spectrom. 2002, 13, 91-98.

σ ) ln

I(5-) + I(4-) /Φ I(5-)

(1)

ized to unity. The results are shown in Figure 2 for the duplex [dsC]5-, which upon irradiation gives [dsC]•4-. The maximum effect is observed between 260 and 275 nm, similar to the maximum of the absorption band in solution. Note that, in this case, detachment of more than one electron is not observable. The comparison of electron photodetachment efficiency as a function of the base composition (Figure 1) shows that, the higher the G/C contents of the duplex, the more efficient the electron detachment at 260 nm. This was confirmed by experiments performed on the small doubly charged single strands dA3C3, dT3C3, dC3T3, dC3A3, and dG3T3, where only the latter gave electron photodetachment. This behavior is in agreement with the recent photoelectron experiments that showed that, due to the low ionization potential (IP) of guanine, guanine-containing oligonucleotide anions have lower electron detachment energy than non-guanine-containing species.56 In the present state of the art, little is known on the electron binding energies of polynucleotide multiply charged anions in the gas phase. Danell and Parks57 observed electron autodetachment at low rates for [7-mer]3- oligonucleotides, in a temperature-dependent and base-dependent manner. For all oligonucleotides studied here, we performed control experiments with the same trapping time and laser off, which confirm that electron detachment is dramatically enhanced upon laser irradiation around 260 nm (which corresponds to 4.77 eV) when guanines are present in the sequence. Guanine has indeed the lowest ionization potential of all four DNA bases (7.64 eV for isolated, neutral guanine).58 Furthermore, it is known that, in DNA strands, base IP lowers due to stacking (and hydrogen bonding in duplexes)59,60 and that when the strand contains guanines, the highest occupied molecular orbital can be located on the base and not on the phosphate.56,59 However, whether electron detachment comes from a thermal process or via specific electronically excited states remains to be established. As mentioned above, to check that charge reduction is caused by electron detachment, we used the small oligonucleotide 5′GGGTTT-3′ (ssF). The negative ion mode calibration is slightly perturbed by the instrument hardware modifications. The number of hydrogens is therefore not determined by accurate mass measurement, but rather by comparison of the isotopic distributions of the laser charge-reduced species with the even-electron deprotonated species observed in the full scan ESI-MS spectrum. Figure 3a shows the ESI mass spectrum obtained from a solution of ssF in the negative mode. Three charge states ([ssF - nH]n-, with n ) 1, 2, 3) are observed. In Figure 3 b, the parent ion [ssF - 2H]2- is isolated and irradiated at 260 nm during 2 s (40 laser shots). The isotopic distribution of the singly charged product (56) Yang, X.; Wang, X. B.; Vorpagel, E. R.; Wang, L. S. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 17588-17592. (57) Danell, A. S.; Parks, J. H. J. Am. Soc. Mass Spectrom. 2003, 14, 13301339. (58) Wetmore, S. D.; Boyd, R. J.; Eriksson, L. A. Chem. Phys. Lett. 2000, 322, 129-135. (59) Kim, N. S.; LeBreton, P. R. J. Am. Chem. Soc. 1996, 118, 3694-3707. (60) Starikov, E. B.; Lewis, J. P.; Sankey, O. F. Int. J. Mod. Phys. B 2005, 19, 4331-4357.

confirms that charge reduction occurs via electron detachment, leading to the radical ion [ssF - 2H]•-. Very little fragmentation is observed, the fragments being w5- and w4- and a singly charged strand, which has lost guanine (all-even-electron species). When [ssF - 3H]3- is isolated and irradiated for 2 s at 260 nm, two main peaks are obtained in the mass spectrum, as shown in Figure 3c. [ssF - 3H]•2- and [ssF - 3H]2•-, correspond to the photodetachement of one and two electrons, respectively. Similarly, very little fragmentation appears. The fragments are w52-, d52-, w4-, and w5-. Interestingly, the isotopic distribution of w5- is 1 Da lower compared to the fragment observed in Figure 3b, and the fragment is therefore noted [w5 - H]•-. In conclusion, irradiation of multiply charged oligonucleotide anions by a UV laser at 260 nm can be used to generate oligonucleotide radical anions efficiently and selectively (with minimal fragmentation). The few fragmentation pathways observed all seem to involve losses of radicals. EPD: CID on Radical Ions. The low amounts of fragments produced by electron photodetachment render electron photodetachment alone inefficient for sequencing purposes. We performed collision-induced dissociation of the radical species generated by laser irradiation (eq 2). hv

CID

[M - nH]n- 98 [M - nH]•(n-1)- 98 fragments

(2)

For the longer single strands A-E, for a given charge state n, we compared (a) the CID spectrum of the even-electron species produced by electrospray to (b) the CID spectrum of the oddelectron species produced by electron photodetachment from charge state (n + 1). Figure 4 shows such comparison for [ssA]3-, with exactly the same activation time (30 ms) and activation amplitude (11%). Sequence coverage and fragmentation efficiency are dramatically improved on the radical species. At this activation amplitude, in EPD, the parent ion has roughly the same intensity as several neutral losses, while in CID, the neutral base losses are less than 10% of the parent ion intensity. Neutral losses in EPD include base losses (guanine loss is predominant) and loss of CO and H2O/NH3 (the resolution does not allow differentiating them). The other fragment ions belong to the w and d ion series and to their complementary series a• and z•, respectively. In contrast, in CID, regular w and (a-B) ion series are observed. Similar fragmentation patterns are observed for all single strands A-F. The fragmentation pathways observed in EPD are summarized in Scheme 2. Here the radical on the parent ion is represented on the phosphate for convenience only. The low oxidation potential of guanine,58,61 the photoelectron spectra of small oligonucleotide anions,56 and the fact that the guanine content of the oligonucleotide here determines the electron photodetachment efficiency all suggest that electron detachment occurs at guanines at 260 nm. However, the efficient fragmentation all along the strand backbone indicates that the radical does not remain localized. The fragmentation of oligonucleotide ions in mass spectrometers has been reviewed recently by Wu and McLuckey.23 The fragments we observe here in 260-nm EPD are strikingly similar to the fragments reported in the literature by using various fragmentation methods involving radicals. To our knowledge, there (61) Hush, N. S.; Cheung, A. S. Chem. Phys. Lett. 1975, 34, 11-13.

Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

6567

Figure 3. (a) Electrospray mass spectrum of the 6-mer ssF and zooms on the isotopic distributions of the doubly and singly charged ion. (b) ESI-MS/MS spectrum obtained after 2-s irradiation of [ssF - 2H]2- at 260 nm and zooms on the isotopic distributions of the doubly charged (parent ion) and singly charged ion. (c) ESI-MS/MS spectrum obtained after 2-s irradiation of [ssF - 3H]3- at 260 nm and zooms on the isotopic distributions of the doubly and singly charged ion. The other fragments observed are annotated with the nomenclature of McLuckey et al.24 (Scheme 1), and comparison with normal CID was made to confirm the radical nature of the w5- ion in spectrum (c).

Figure 4. Comparison of CID and EPD of [ssA]3-. (a) CID during 30 ms at 11% activation amplitude on [ssA - 3H]3- produced by electrospray. Note the 10-fold magnification on the two fragment regions. (b) EPD: CID during 30 ms at 11% activation amplitude on [ssA - 4H]?3- produced by electron photodetachment of [ssA - 4H]4- under 2-s irradiation at 260 nm. The parent ion is abbreviated “M” in both spectra, and the fragments are annotated according to Schemes 1 and 2. Triangles indicate two electronic noise spikes.

is only one report on UV laser photodissociation of DNA strands published to date. In 1996, McLafferty’s group reported photodissociation of multiply charged dT30 under laser irradiation at 193 nm.47 Interestingly, electron photodetachment was observed and was accompanied by fragmentation into w and a ions (which have the same masses as d and z ions in palindromic sequences such as dT30). 6568 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

Methods used to date for radical ion generation from DNA anions have involved either ion-ion reactions, ion-electron reactions, or ion-neutral collisions at high energy. Radical anions generated by electron transfer upon reaction with Xe•+,62 CCl3+, 63 or O2•+ 64 produced w ions and “z” ions, but no a-B ions for polydA (62) Herron, W. J.; Goeringer, D. E.; McLuckey, S. A. J. Am. Chem. Soc. 1995, 117, 11555-11562.

Figure 5. Observed EPD fragments and their charge states for the DNA single strands A-E. Radical ion at charge state n is produced by electron photodetachment under laser irradiation of the parent ion at charge state (n + 1). n ) 3 for ssA-D and n ) 4 for ssE.

Scheme 2

and polydC.63,64 With these palindromic sequences, “z” and “a” ions have the same masses and cannot be distinguished. The two (63) McLuckey, S. A.; Stephenson, J. L.; Ohair, R. A. J. J. Am. Soc. Mass Spectrom. 1997, 8, 148-154. (64) Wu, J.; McLuckey, S. A. Int. J. Mass Spectrom. 2003, 228, 577-597.

major fragments observed by reaction with O2•+ 64 were the base loss and the wn-1 ion, which is very similar to what we observe in EPD (w52- ion in Figure 3b and c, w113- ion in Figure 4b). Liu et al. have studied radical oligonucleotides issued from electron loss upon high-energy collisions with noble gas atoms.65 With polydA, Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

6569

Figure 6. Comparison of CID and EPD of [ssE].4- (a) CID during 30 ms at 12% activation amplitude on [ssE - 4H]4- produced by electrospray. (b) EPD: CID during 30 ms at 12% activation amplitude on [ssE - 5H]?4- produced by electron photodetachment of [ssE - 5H]5- under 2-s irradiation at 260 nm. The parent ion is noted M, and the fragments are annotated according to Schemes 1 and 2.

w/d and a/z ions were observed, as well as some a-B ions. With dGCCCC, the authors could distinguish that both d and w ion series were present, together with a ions. They proposed a fragmentation mechanism giving rise to radical w ions and their complementary a ions. In our EPD spectra, although the calibration is slightly distorted due to the trap hardware modifications, by comparing with the masses of w ions produced by normal CID and taking into account the systematic calibration offset, we conclude that the radicals are located on the a and z ions. Hakansson et al. studied oligodeoxynucleotide fragmentation using ECD on positively charged DNA strands,26,27 and using EDD by electron impact on negatively charged DNA strands.29 In ECD, w, a•, d, and z• ion series were observed.26 In EDD, w/d and a•/z• series were observed for the palindromic sequences dB6 (B ) A, T, G, C). For the sequence dGCATGC, both d and w ion series and one z• ion could be distinguished.29 Another mass spectrometry fragmentation method that involves radicals is MALDI in-source decay.66-68 Interestingly, a few publications mention the observation of similar fragmentation as we observe here in EPD. Wang et al.69 reported w, d, a, and z fragment ions in in-source decay of modified oligonucleotides using the matrix mixture 3-hydroxypicolinic acid and N-(3indolylacetyl)-L-leucine (HPA/IAL) and a 337-nm N2 laser. Juhasz et al.70 reported the formation of w and d ion series in MALDI in-source decay of an 11-mer RNA using picolinic acid at 266 nm, but not using HPA at 337 nm. In summary, whatever the method used for radical ion generation, oligonucleotide fragmentation from radicals is markedly different from oligonucleotide fragmentation from closed(65) Liu, B.; Hvelplund, P.; Nielsen, S. B.; Tomita, S. Int. J. Mass Spectrom. 2003, 230, 19-24. (66) Patterson, S. D.; Katta, V. Anal. Chem. 1994, 66, 3727-3732. (67) Takayama, M. J. Am. Soc. Mass Spectrom. 2001, 12, 1044-1049. (68) Kocher, T.; Engstrom, A.; Zubarev, R. A. Anal. Chem. 2005, 77, 172-177. (69) Wang, B. H.; Hopkins, C. E.; Belenky, A. B.; Cohen, A. S. Int. J. Mass Spectrom. 1997, 169, 331-350. (70) Juhasz, P.; Roskey, M. T.; Smirnov, I. P.; Haff, L. A.; Vestal, M. L.; Martin, S. A. Anal. Chem. 1996. 68, 941-946.

6570 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

shell species and predominantly produces w, d, a•, and z• ion series instead of the w and a-B ion series that are observed in CID. However, the different techniques described above remained at the stage of proof of principle and to our knowledge were not further developed as effective oligonucleotide sequencing methods. EDD by electron impact is a recent and promising technique. EPD is an interesting alternative to EDD, as highly efficient electron detachment is obtained by increasing the photon flux. Future work will include more thorough comparison between EDD and EPD. Sequence Coverage and Potential Application of EPD for de Novo Sequencing. The method presented here, i.e., generation of radical ions by electron photodetachment by a UV laser, isolation of the radical, and fragmentation by collision-induced dissociation, can be implemented in quadrupole ion traps and in FTICR mass spectrometers. To evaluate the potential utility of the EPD method for oligonucleotide sequencing, we recorded the EPD spectra for the single strands A-E under different activation amplitudes to generate the maximum amount of fragments, summed up the scans, and inventoried the fragments. The results are summarized in Figure 5. The fully annotated EPD spectra of ssB and ssD are provided as Supporting Information Figures S1 and S2, respectively. The sequence coverage obtained by EPD is excellent. Complete sequence coverage is obtained for the 12-mers and the 15mer (the only missed C-C cleavage in ssC does not prevent sequence determination), and there are only three missed cleavages for the 20-mer ssE. Although similar sequence coverage can be obtained by Q-TOF CID on the same sequences (data not shown), the CID spectra are complicated by numerous internal fragments.71 In EPD, all major fragment ion peaks belong to the w, d, a•, and z• series, and no internal fragments are found. A few a-B ions are found as well in EPD, and those correspond to the most intense ones in CID, so they are readily identified by com(71) Rozenski, J.; McCloskey, J. A. J. Am. Soc. Mass Spectrom. 2002, 13, 200203.

Figure 7. Comparison of CID and EPD of double-stranded DNA. (a) CID during 30 ms at 10% activation amplitude on [dsB - 4H]4- produced by electrospray. (b) EPD: CID during 30 ms at 10% activation amplitude on [dsB - 5H]•4- produced by electron photodetachment of [dsB 5H]5- under 2-s irradiation at 260 nm. (c) CID during 30 ms at 9% activation amplitude on [dsC - 4H]4- produced by electrospray. (b) EPD: CID during 30 ms at 9% activation amplitude on [dsC - 5H]?4- produced by electron photodetachment of [dsC - 5H]5- under 2-s irradiation at 260 nm.

parison. The region of the neutral losses is quite rich in intense peaks, as neutral losses of CO or of guanine from the most intense fragments can also be observed, and smaller fragment peaks overlapping in that region might be hidden. This explains some of the missed cleavages for the 20-mer ssE. It is anticipated that EPD coupled to an FTICR mass spectrometer would help further increasing the sequence coverage by resolving overlapping peaks. The high sequence coverage combined with the absence of internal fragments make EPD a very promising approach for de novo sequencing of oligonucleotides. Interpretation of an EPD spectrum if the sequence is unknown (de novo sequencing) is straightforward. In a first step, thanks to the simultaneous observation, with very few exceptions, of d/a• and w/z• pairs separated by 99 Da, the d and w series can be distinguished from the a• and z• series. In a second step, the w series can be identified simply by comparison with normal CID on the even-electron oligonucleotide, where w fragments are formed, but not d fragments. An algorithm for de novo sequencing by EPD is therefore much more simple than the algorithms used for interpreting CID spectra:71 sequence tags can easily be read from the mass spectrum thanks to the absence of internal fragments, and the sequence directionality (3′ to 5′ vs 5′ to 3′) can be determined easily by a comparison with a low-energy CID spectrum. Potential of EPD for Studying Higher-Order Structures. As suggested already by Hakansson et al.,29 EDD of oligonucleotides and ECD of proteins share several characteristics: the frag-

mentation is initiated by radical ion formation (electron loss or electron gain), and the sequence coverage is large. In ECD, the fragmentation of radicals is a fast process occurring specifically at some covalent bonds, but noncovalent interactions such as hydrogen bonds retain the fragments together only the parent ion is detected. Similar processes can be at stake in EPD, as suggested also for electron-impact EDD.29 To obtain some insights into this aspect, we examined more closely the fragmentation pathways and fragmentation efficiency of ssE, which can form a hairpin structure, and of the double-stranded helices dsA-C, formed by dimerization of ssA-C. The CID spectrum and the EPD spectrum of ssE are shown in Figure 6a and b, respectively. In both fragmentation methods, the major fragment is the base loss. In EPD, one can also observe the loss of negatively charged guanine that gives [ssE-G]3- at m/z 1991.0. The major backbone fragments in CID are the w15-183ions and one a-B ion. While for ssA, at a collision energy provoking minimal fragmentation with CID, many intense fragments could already be observed by EPD (see Figure 4), this is not the case for ssE. In EPD at the same collision energy that provoked complete base loss in CID, the backbone fragments only start to appear, and the relative intensities of the w15-183- ions are not much greater than in regular CID. These results could indicate that intramolecular hydrogen bonding may retain fragments together and that EPD could be used to probe the collision energy needed to disrupt these intramolecular interactions. However, Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

6571

additional work with other oligonucleotides of same length but different structures is needed to prove this point. To test a structure for which intermolecular interactions are present with certainty, we turned back to the helices dsA-C (see Table 1). The compared CID and EPD spectra of the [ds]4species are shown in Figure 7. Only a small portion of the spectrum is shown, because no other fragments than neutral losses are detected. In quadrupole ion trap regular CID, neutral base loss is observed. The preferential base loss compared to duplex dissociation into single strands is expected because of the low charge state and of the slow activation regime in the ion trap.14,55 In EPD, in addition to base loss and CO and H2O loss, we could observe the loss of sugar-base radical from the terminal bases. This corresponds formally to loss of z1• and a1•, giving [ss + d11]4and [ss + w11]4-, respectively. As for all single strands extensive cleavage all along the sequence was observed, a possible explanation for the lack of fragmentation in the duplexes is that the strands are maintained together by weak noncovalent interactions, with maybe a partial terminal base unzipping. Altogether, the results on the hairpin and on the dsDNA suggest that EPD at moderate collision energy could be used to probe intermolecular interactions in nucleic acids. CONCLUSIONS To our knowledge, this is the first demonstration of the use of laser-induced electron photodetachment to generate radical ions of large biomolecules, which can be further fragmented by CID and give rise to interesting new fragmentation pathways. The method, coined electron photodetachment dissociation has been applied here to DNA, but the approach should be applicable to any molecule giving multiply charged anions, with an adequate wavelength for electron detachment. For nucleic acids, standard 266-nm Nd3+:YAG lasers will perfectly match the optimum electron (72) Weber, J. M.; Ioffe, I. N.; Berndt, K. M.; Loffler, D.; Friedrich, J.; Ehrler, O. T.; Danell, A. S.; Parks, J. H.; Kappes, M. M. J. Am. Chem. Soc. 2004, 126, 8585-8589. (73) Horn, D. M.; Breuker, K.; Frank, A. J.; McLafferty, F. W. J. Am. Chem. Soc. 2001, 123, 9792-9799. (74) Breuker, K.; Oh, H.; Horn, D. M.; Cerda, B. A.; McLafferty, F. W. J. Am. Chem. Soc. 2002, 124, 6407-6420.

6572

Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

photodetachment efficiency for guanine-containing oligonucleotides. However, the use of different charge states72 or of shorter wavelength is expected to extend the possibilities of the method to other analytes. The other way of producing radical ions by electron detachment is electron impact at >10 eV.28,29 In the case of nucleic acids, a major difference between electron-impact EDD and EPD is that the presence of guanines is essential for EPD to occur at 260 nm, while electron-impact EDD efficiency is less base-dependent.29 Another expected difference, yet remaining to be tested with exactly the same analytes in the two techniques, is the selectivity of the radical ion generation compared to vibrational excitation of the ion. Similarly to using ECD for protein folding studies as demonstrated by McLafferty and co-workers,73,74 EPD could be used to probe higher-order structures of nucleic acids when combined with low-energy activation methods. Conversely, when higher-energy activation methods are used for the CID of the nucleic acid single-strand radicals, intramolecular bonds are ruptured and extensive sequence coverage can be obtained. The observation of nearly complete w, d, a•, and z• series, combined with the absence of internal fragments, is opening bright perspectives for the application of EPD to nucleic acid de novo sequencing. ACKNOWLEDGMENT The GDR CNRS “Agre´gation, fragmentation et thermodynamique des syste`mes complexes isole´s” is gratefully acknowledged for financial support. V.G. is a Research Associate of the FNRS (Fonds National de la Recherche Scientifique, Belgium). SUPPORTING INFORMATION AVAILABLE Fully annotated EPD spectra of ssB and ssD as Figures S1 and S2, respectively. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review April 20, 2006. Accepted July 25, 2006. AC060753P