Anal. Chem. 2005, 77, 2373-2380
A Deoxynucleotide Derivatization Methodology for Improving LC-ESI-MS Detection Jimmy Flarakos,† Wennan Xiong,‡ James Glick,‡ and Paul Vouros*,‡
Barnett Institute and Department of Chemistry and Chemical Biology, Northeastern University, Boston, Massachusetts 02115, and Deparment of Biological Engineering, Massachusetts Institute of Technology, 56-738, Cambridge, Massachusetts 02129
We have developed a novel LC-UV-MS derivatization method for the analysis of deoxyguanosine monophosphate adducts that demonstrates enhanced signal intensities relative to underivatized analytes in positive ion mode electrospray ionization MS. Detection of DNA nucleotide adducts is normally conducted in negative ion mode, which requires basic mobile phases that make chromatographic separations difficult and reduce MS sensitivity. Utilizing coupling reagents typically employed in peptide synthesis, several different deoxyguanosine nucleotide phosphoramidates and phosphomonoesters were synthesized in high conversion yield and under mild reaction conditions. The derivatives were characterized by MS/MS and reaction conversion yields determined from the DADUV traces. The derivatives were evaluated for ionization efficiencies, fragmentation patterns, and reversed-phase chromatographic properties by LC/ESI-MS/MS. Overall, the hydrophobic derivatives showed increases in ionization efficiency and improved peak shape. Rank ordering of the derivatizing agents was initially established using the dGp-modified derivatives. The best derivatizing agent, hexamethyleneimine, showed a 3-4-fold signal enhancement compared to underivatized dGp and was selected for additional evaluation. A model system using the carcinogen, N-acetoxy-2-acetylaminofluorene (AAAF), was used to synthesize a N-acetyl-(2-aminofluorenyl)-guanosine 5′-monophosphate (dGpAAF) adduct, which was subsequently derivatized with hexamethyleneimine. Detection limits for dGphex and dGpAAFhex, purified by HPLC, were 10- and 3-fold higher (S/N) than their respective underivatized analogues. Practical applicability, with similar improvements in sensitivity, was established by derivatizing adducts isolated from calf thymus DNA exposed to AAAF. Our results demonstrate the utility of simple reactions for the enhanced detection of a mononucleotide in positive ion mode ESI MS and the application of this technique for the detection of dGpDNA adducts at the low-femtomole level. The study of DNA adducts as a critical indicator for the onset of tissue carcinogenicity has been known for several decades.1-3 * Corresponding author. Phone: 617-373-2840, fax: 617-373-2693, E-mail:
[email protected]. † Northeastern University. ‡ Massachusetts Institute of Technology. 10.1021/ac0483724 CCC: $30.25 Published on Web 03/12/2005
© 2005 American Chemical Society
DNA adducts are covalent complexes formed from biotransformation reactions of carcinogenic chemicals yielding reactive metabolite intermediates, which attach to selective bases of cellular DNA. Although cellular enzymes correct any damaged genetic material resulting from chemical exposure, small amounts of damaged DNA, if not repaired, may lead to the initiation of carcinogenesis. Genotoxic carcinogen adduction is believed to be initiated through the electrophilic addition of a carcinogenic agent to DNA. The subsequent DNA adduct may interfere with sequence translation leading to deleterious modifications to gene expression.4 Many studies have identified an increased cancer risk related to chemical exposure and the presence of DNA adducts.1,3,5,6 Thus, DNA adducts can serve as good biomarkers of chemical exposure and are used to determine the extent of damage to genetic material. Therefore, accurate quantification of DNA adducts is important to generate dependable epidemiological databases.7 Currently, the methods used for the analysis of DNA adducts include the following: immunoassays,8-10 radioactivity and scintillation counting,11,12 liquid chromatography coupled to electrochemical detection,13-15 accelerator mass spectrometry,16-19 liquid (1) Brookes, P.; Lawley, P. D. Nature 1964, 202, 781-784. (2) Kriek, E.; Miller, J. A.; Juhl, U.; Miller, E. C. Biochemistry 1967, 6, 177182. (3) Miller, J. A.; Miller, E. C. Environ. Health Perspect. 1983, 49, 3-12. (4) Farmer, P. B. Toxicol. Lett. 2004, 149, 3-9. (5) Colin Garner, R. Mutat. Res./Fundam. Mol. Mech. Mutagen. 1998, 402, 67-75. (6) Baird, W. M.; Mahadevan, B. Mutat. Res./Fundam. Mol. Mech. Mutagen. 2004, 547, 1-4. (7) Farmer, P. B.; Sweetman, G. M. A. J. Mass Spectrom. 1995, 30, 13691379. (8) Gyorffy, E.; Anna, L.; Gyori, Z.; Segesdi, J.; Minarovits, J.; Soltesz, I.; Kostic, S.; Csekeo, A.; Poirier, M. C.; Schoket, B. Carcinogenesis 2004, 25, 12011209. (9) Wang, H.; Lu, M.; Weinfeld, M.; Le Chris, X. Anal. Chem. 2003, 75, 247254. (10) Divi, R. L.; Beland, F. A.; Fu, P. P.; Von Tungeln, L. S.; Schoket, B.; Camara, J. E.; Ghei, M.; Rothman, N.; Sinha, R.; Poirier, M. C. Carcinogenesis 2002, 23, 2043-2049. (11) Booth, E. D.; Kilgour, J. D.; Robinson, S. A.; Watson, W. P. Chem. Biol. Interact. 2004, 147, 195-211. (12) Petruzzelli, S.; Tavanti, L. M.; Celi, A.; Giuntini, C. Am. J. Respir. Cell Mol. Biol. 1996, 15, 216-223. (13) Zhou, L.; Yang, J.; Estavillo, C.; Stuart, J. D.; Schenkman, J. B.; Rusling, J. F. J. Am. Chem. Soc. 2003, 125, 1431-1436. (14) Park, J. W.; Cundy, K. C.; Ames, B. N. Carcinogenesis 1989, 10, 827-832. (15) Germadnik, D.; Pilger, A.; Rudiger, H. W. J. Chromatogr., B 1997, 689, 399-403. (16) Dingley, K. H.; Roberts, M. L.; Velsko, C. A.; Turteltaub, K. W. Chem. Res. Toxicol. 1998, 11, 1217-1222.
Analytical Chemistry, Vol. 77, No. 8, April 15, 2005 2373
chromatography or capillary electrophoresis coupled to mass spectrometry,20-26 and32P-postlabeling.27-30 32P-Postlabeling is a mature technique that can be used to detect modifications as low as one adduct per 109-1011 normal nucleotides.5 However, while the technique is very sensitive, it lacks the capacity to supply any structural information. This shortcoming of 32P-postlabeling and the ability of LC/MS to provide structural information has permitted MS-based techniques to dramatically gain in popularity, becoming essential in the characterization of DNA adducts. Although there are numerous reports on the use of LC/MS methods for the analysis of DNA adducts in their nucleoside form,20,21,26,31 there are few examples of the use of LC/MS to detect nucleotide adducts.32-34 The relative lack of appropriate methods for DNA adducts in the nucleotide form highlights the need for a suitable method capable of detecting the wide array of all DNA adducts. Phosphate adducts that form on the 5′ end of DNA represent an important class of DNA adducts that requires suitable methods for analysis. Using the classical methods of chemical and enzymatic digestion techniques for the digestion of DNA, the loss of the 5′-phosphate group can result in the loss of important carcinogen adducted phosphates. DNA phosphate adducts experience little or no repair and persist in vivo, providing an alternative means for quantifying DNA damage.35-38 Methyl and ethyl 5′-phosphate adducts were first studied in 1978 by Swenson and Lawley with 32P post-labeling techniques.39 However, the lack of structural information made the characterization of the DNA phosphate adduct incomplete. Recently, Haglund and co-workers used LC/MS/MS to structurally characterize N-ethyl-N-nitroso(17) Martin, E. A.; Brown, K.; Gaskell, M.; Al-Azzawi, F.; Garner, R. C.; Boocock, D. J.; Mattock, E.; Pring, D. W.; Dingley, K.; Turteltaub, K. W.; Smith, L. L.; White, I. N. Cancer Res. 2003, 63, 8461-8465. (18) Lightfoot, T. J.; Coxhead, J. M.; Cupid, B. C.; Nicholson, S.; Garner, R. C. Mutat. Res. 2000, 472, 119-127. (19) Goldman, R.; Day, B. W.; Carver, T. A.; Mauthe, R. J.; Turteltaub, K. W.; Shields, P. G. Chem. Biol. Interact. 2000, 126, 171-183. (20) Gangl, E. T.; Turesky, R. J.; Vouros, P. Anal. Chem. 2001, 73, 2397-2404. (21) Ding, J.; Vouros, P. J. Chromatogr., A 2000, 887, 103-113. (22) Gangl, E. T.; Turesky, R. J.; Vouros, P. Chem. Res. Toxicol. 1999, 12, 10191027. (23) Barry, J. P.; Norwood, C.; Vouros, P. Anal. Chem. 1996, 68, 1432-1438. (24) Szeliga, J.; Page, J. E.; Hilton, B. D.; Kiselyov, A. S.; Harvey, R. G.; Dunayevskiy, Y. M.; Vouros, P.; Dipple, A. Chem. Res. Toxicol. 1995, 8, 1014-1019. (25) Wolf, S. M.; Vouros, P. Anal. Chem. 1995, 67, 891-900. (26) Wolf, S. M.; Vouros, P. Chem. Res. Toxicol. 1994, 7, 82-88. (27) Snyderwine, E. G.; Davis, C. D.; Nouso, K.; Roller, P. P.; Schut, H. A. Carcinogenesis 1993, 14, 1389-1395. (28) Randerath, K.; Reddy, M. V.; Gupta, R. C. Proc. Natl. Acad. Sci. U.S.A. 1981, 78, 6126-6129. (29) Moller, L.; Zeisig, M.; Vodicka, P. Carcinogenesis 1993, 14, 1343-1348. (30) Gillardeaux, O.; Perin-Rousel, O.; Perin, F. IARC Sci. Publ. 1993, 133139. (31) Aussenac, J.; Chassagn, D.; Claparols, C.; Charpentier, M.; Duteurtre, B.; Feuillat, M.; Charpentier, C. J Chromatogr., A 2001, 907, 155-164. (32) Gennaro, L. A.; Vadhanam, M.; Gupta, R. C.; Vouros, P. Rapid Commun. Mass Spectrom. 2004, 18, 1541-1547. (33) Cai, Z. Anal. Sci. 2001, 17, a199-a201. (34) Doerge, D. R.; Yi, P.; Churchwell, M. I.; Preece, S. W.; Langridge, J.; Fu, P. P. Rapid Commun. Mass Spectrom.: RCM 1998, 12, 1665-1672. (35) Den Engelse, L.; De Graaf, A.; De Brij, R. J.; Menkveld, G. J. Carcinogenesis 1987, 8, 751-757. (36) Den Engelse, L.; Menkveld, G. J.; De Brij, R. J.; Tates, A. D. Carcinogenesis 1986, 7, 393-403. (37) Haglund, J.; Henderson, A. P.; Golding, B. T.; Tornqvist, M. Chem Res. Toxicol. 2002, 15, 773-779. (38) Haglund, J.; Ehrenberg, L.; Tornqvist, M. Chem. Biol. Interact. 1997, 108, 119-133. (39) Swenson, D. H.; Lawley, P. D. Biochem. J. 1978, 171, 575-587.
2374
Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
urea alkylated phosphate adducts. Although the alkylating agent N-ethyl-N-nitrosourea is not listed as a carcinogen, its reactivity is similar to N-methyl-N-nitrosourea (a listed carcinogen) and may be used to study its DNA alkylating characteristics.40 Therefore, losing the phosphate adducts during the digestion procedure potentially distorts the representation of adduction activities of carcinogens on native DNA. In addition, cleavage of the phosphate group adds another digestion step that increases the loss of total nucleosides during sample preparation. The development of a sensitive LC/MS methodology for the analysis of DNA nucleotide adducts has been a challenge in bioanalytical chemistry, complicated by the presence of a native phosphate group. Therefore, nucleotides have been analyzed in negative ion mode, which required basic mobile phases that made reversed-phase chromatographic separations tedious and may often result in low MS sensitivity. Detection in positive mode would allow the use of acidic mobile phases resulting in synergistic effects of increased k′ (capacity factor), better chromatographic resolution, and increased MS sensitivity. Several techniques have been implemented to address the lack of chromatographic retention. For example, ion pairing chromatography with triethylammonium bicarbonate41 or N,N-dimetlylhexylamine42 has been used with some success to address these issues. However, these elaborate methods lack robustness, simplicity, and reproducibility when compared to those used for the analysis of nucleoside adducts. There have been several reports involving derivatization of nucleotides.34-38 Hashizume synthesized TMS derivatives of nucleotides for GC/MS analysis by displacing acidic hydrogen atoms on hydroxyl groups.43 Similarly, permethylated derivatives have been prepared44,45 for GC/MS analysis. Recently, Nordstrom reported use of propionyl or benzoyl derivatives to increase hydrophobicity and ESI response.46 Although, specific derivatization of the phosphate group has been accomplished by Yerkevich using phenylboronic acids, the relatively low conversion yield does not make this method amenable to quantitative analysis.47 It is essential in DNA nucleotide adduct analysis that derivatization reactions be specific and not add to the complexity of structural determination. To demonstrate feasibility of our derivatization methodology, we needed to fulfill certain criteria: the derivatization needed to be simple, efficient, amenable to small sample sizes and reagent volumes, and rapid (less than 24 h). Furthermore, the reaction conversion yield needed to be relatively high, i.e., greater than 90%. It was also desirable to have a method that was not labor intensive with a minimum of sample transfer steps to reduce sample losses. More importantly, the selected derivative needed to actually increase k′ and also improve ionization efficiency (40) Haglund, J.; Van Dongen, W.; Lemiere, F.; Esmans, E. L. J. Am. Soc. Mass Spectrom. 2004, 15, 593-606. (41) Huber, C. G.; Krajete, A. Anal. Chem. 1999, 71, 3730-3739. (42) Tuytten, R.; Lemiere, F.; Dongen, W. V.; Esmans, E. L.; Slegers, H. Rapid Commun. Mass Spectrom. 2002, 16, 1205-1215. (43) Hashizume, T.; Sasaki, Y. Anal. Biochem. 1966, 15, 199-203. (44) Baker, K. M. Adv. Mass Spectrom. Biochem. Med. 1976, 1, 231-237. (45) Pettit, G. R.; Einck, J. J.; Brown, P. Biomed. Mass Spectrom. 1978, 5, 153160. (46) Nordstrom, A.; Tarkowski, P.; Tarkowska, D.; Dolezal, K.; Astot, C.; Sandberg, G.; Moritz, T. Anal. Chem. 2004, 76, 2869-2877. (47) Yurkevich, A. M.; Kolodkina, II; Varshavskaya, L. S.; Borodulina-Shvetz, V. I.; Rudakova, I. P.; Preobrazhenski, N. A. Tetrahedron 1969, 25, 477-484.
Figure 1. Column switching setup in the back-flush mode.
relative to an underivatized nucleotide. It was also essential that the derivatization worked at sufficiently low analyte levels for in vitro or in vivo applications. Notwithstanding the important aspects listed above, the most significant factor in a suitable MS method is that the derivative retain the structural features of the adduct for characterization and trace level analysis by tandem MS techniques. To satisfy these criteria, coupling reagents typically employed in peptide synthesis were used to prepare several different deoxyguanosine nucleotide phosphoramides and phosphomonoesters in high conversion yield and under mild conditions. The different derivatives were initially evaluated using deoxyguanosine monophosphate (dGp) as the model analyte. Once a suitable derivative was identified, the dGp adduct of N-acetoxy-2-acetylaminofluorene (AAAF) was then used as a model system for assessment of the applicability of the derivatization protocol to carcinogen adducted DNA.25,26,39 EXPERIMENTAL SECTION Chemicals. Deionized water was generated by a Milli-Q-Plus water system from Millipore (Waltham, MA). HPLC grade methanol and acetonitrile were obtained from Fisher Scientific (Pittsburgh, PA) while ammonium acetate and all other reagents were purchased from Sigma Chemical Co. (St. Louis, MO) unless otherwise specified. Samples were centrifuged in 1.5-mL siliconized microcentrifuge tubes with the Marathon 21000 (R) refrigerated multipurpose centrifuge. Both were purchased from Fisher (Pittsburgh PA). The reactions were carried out at room temperature, overnight in DMSO. The derivatives were characterized by MS/MS and reaction conversion yields determined from their UV traces. Several phosphoramidate and phosphomonoester reactions were used to prepare a reference sample of the N-acetyl-(2-aminofluorenyl)guanosine 5′- monophosphate (dGpAAF) adduct, which was previously prepared in our laboratory.26 The dGpAAF derivatives were analyzed under the same conditions as the deoxyguanosine 5′- monophosphate (dGp) derivatives. HPLC. All HPLC separations was performed on a HewlettPackard 1090 liquid chromatography system equipped with a DR5 ternary solvent delivery system and diode-array UV (Wilmington,
DE). All conversion yield determinations, on crude mixtures, were conducted on a Waters (Milford, MA) Atlantis dC18 50 × 3 mm, 3-µm reversed-phase column, with a flow rate of 0.5 mL/min and an injection volume of 20 µL. UV absorbance was monitored at 254 and 280 nm with a 16-nm slit width and the reference wavelength of 480 nm. Isolation of HPLC fractions and monitoring of in vitro reaction completion were conducted with a second Rheodyne valve installed on the HP 1090 to automate the fraction collection process (Figure 1). The loading pump for this column switching configuration was a Waters 616 multidelivery system (Milford, MA). Synthesis of dGp Derivatives. A 3-mg sample of TBTU was added to a DMSO solution containing 1 mg of dGp, derivatizing reagent, and 5 µL of TEA. The reaction mixture was stirred at room temperature for 12 h and was followed by storage at -20 °C. To avoid degradation, sample mixtures were diluted just prior to LC/MS analysis. In Vitro DNA Reaction. A 3-mg sample of calf thymus DNA (type I) in 3 mL of 10 mM citrate buffer, pH 6.0, was reacted with 1 mg of N-acetoxy-N-acetyl-2-aminofluorene (dissolved in 1 mL of acetonitrile) at room temperature for 24 h. Excess acetonitrile was evaporated in vacuo. After extraction of the reaction mixture with ethyl acetate (3 × 3 mL), the DNA was precipitated with 60 µL of 5 M NaCl and 2 mL of ice-cold ethanol. The sample was then centrifuged at -10 °C for 20 min at 10 000 rpm, after which the excess liquid was siphoned from the DNA pellet. DNA was reconstituted in 3 mL of water, and enzymatic hydrolysis was accomplished with 3 mL of Dnase I (1000 units/mL, 10 mM TrisHCl, pH 7.5, 10 mM MgCl2, 10 mM CaCl2) and 3 mL of reaction buffer (200 mM Tris-HCl pH 8.3, 10 mM MgCl2). After 3 h, 7 µL of snake venom phosphodiesterase I (20 units/mg in 110 mM Tris-HCl, pH 8.9, containing 110 mM NaCl, 15 mM MgCl2) was added and the resultant mixture incubated at room temperature for 6 h. The reaction was monitored for completion with column switching (details discussed later in this section), and the digestion was terminated by the addition of 3 volumes of ice-cold ethanol followed by vortexing. Insoluble material was pelleted by centrifugation at -10 °C for 20 min at 10 000 rpm. The resulting Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
2375
supernatant was transferred and evaporated to a residual volume of 100 µL. A Waters Oasis 30-mg HLB (Milford, MA) solid-phase extraction cartridge was employed for adduct isolation from the DNA digest. Unmodified nucleotides were eluted with 10% (v/v) methanol, and AAF-modified nucleotides were eluted with 50% (v/v) methanol. All fractions were evaporated to dryness and reconstituted in 100 µL of deionized water. Comparison of dGp and dGp Derivatives. Mobile-phase solvents A and B, respectively, were 10 mM ammonium acetate in water and 10 mM ammonium acetate in methanol. Three methods were used to identify suitable MS/MS parameters for the derivatized mononucleotides: method 1 was 5/95% (B/A) (v/v) for derivatives 1-5 (Table 1), method 2 employing 20/80% (B/A) (v/v) was used for derivatives 6-8. Conversion yield determinations for dGp and dGp derivatives were performed with method 3 generic gradient of 5% B for 3 min, 5-30% B 3-7 min, and held at 30% for 7-15 min, with a total run time of 20 min. Comparison of dGpAAF and dGpAAF Derivatives. Three methods were used to identify suitable MS/MS parameters for the underivatized dGpAAF and dGpAAF derivatives: method 1 (dGpAAF, dGpAAFmethyl, and dGpAAFDMAP), method 2 (dGpAAFpiperidine), and method 3 (dGpAAFhexamethyleneimine) composed of 50/50, 60/40, and 65/35% (B/A) (v/v). respectively. Isolation and Purification of dGpAAF and dGpAAFhexamethyleneimine Derivative. The column switching system employed a binary column and pump configuration. The first pump, possessing the weaker mobile phase, is used to load the sample onto a loading column. After 2 min, the valve is switched so the loading column is in-line with the analytical column. This allows the second pump, having the stronger mobile phase, to back-flush the analytes from the loading column onto the analytical column for further separation. Back-flushing the analytes gave a more compressed peak and better signal-to-noise ratio relative to forward elution onto the analytical column. The loading and analytical columns were a Zorbax XDB C-18 4.6 × 12.5 mm 5 µm (Wilmington, DE). The loading mobile phase consisted of 60/40 or 50/50% (dGpAAF or dGpAAFhex) 25 mM ammonium acetate in water and methanol, at a flow rate of 2.0 mL/min for 1.5 min. The analytical mobile phase consisted of 60/40 25 mM ammonium acetate in water and methanol at a flow rate of 0.5 mL/min. Fractions were isolated and evaporated to dryness in preparation for the analysis. This system was also employed to monitor the in vitro reactions. Comparison of Limits of Detection (LOD) of Purified dGpAAF and dGpAAFhex. The purified samples were reconstituted in 1 mL of methanol, vortexed, and placed in a clean 1-cm path-length quartz cuvette equipped with a Teflon cap. Solvent blanks were recorded under identical conditions, and the baseline was subtracted from the experimental spectrum using a Varian Bio 100 spectrometer (Walnut Creek, CA). Molar extinction coefficients for dGp and AAF were determined separately by taking the UV spectrum of 8 µg/mL AAF and 10 µg/mL dGp. Using Beer’s law (A ) bc), the extinction coefficient () was calculated from the empirical absorbance (A), concentration (c), and path length (b). The sum of the extinction coefficients for dGp and AAF were used as the extinction coefficients for dGpAAF and dGpAAFhex. 2376
Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
Lowest limit of detection (LLOD) analyses were performed on a New Objective (Woburn, MA) Aquasil C18 50 × 0.075 mm, 3 µm, at a flow rate of 100 nL/min. A 0.5-µL sample of solution containing 500 pg/mL dGpAAF or dGpAAFhex was injected in triplicate preceded by a blank. From these purified solutions, the dGpAAF and dGpAAFhex were diluted to a volume yielding an arbitrarily determined signal-to-noise ratio of approximately 3.3 to 1. Mass Spectral Analysis. Ion Trap Mass Spectral Analysis. All ion trap mass spectral analyses were performed on a Finnigan LCQ classic quadrupole ion trap mass spectrometer (San Jose, CA) operating in the positive ion mode. ESI parameters were as follows: capillary temperature 175 °C; spray voltage 4.5 kV; capillary voltage 35.0 V; sheath and auxiliary gas flow rate (nitrogen) 40 and 3, respectively. All MS data were collected in full scan m/z range of 150-750 Da. Triple Quadrupole Spectral Analysis. All triple quadrupole analyses were performed using a Finnigan TSQ 700 triple quadrupole mass spectrometer (San Jose, CA). The heated capillary temperature was set at 175 °C, the collision cell pressure was set to 0.9 mTorr, the second quadrupole offset voltage was set to -30 V, the conversion dynode was set to -15 kV and the electron multiplier to 1500 V. Electrospraying of the capillary column effluent was initiated by positioning the µESI tip ∼3 mm from the entrance of the TSQ 700 heated capillary and applying a potential of 1.5 kV to the stainless steel liquid junction housing. Following injection, an instrument control language procedure was initiated that set the instrument to acquire positive ions in a selected reaction monitoring (SRM) mode. Specifically, the transitions were m/z 569 > 373 (dGpAAF) and 650 > 373 (dGpAAFhex) with a dwell time of 0.4 s, respectively, and an interchannel delay of 0.4 Da. The total run time was 10 min. RESULTS AND DISCUSSION An accurate evaluation of derivatized compounds could only be established by clearly setting specific criteria to define our model for enhancing detection of DNA adducts. In the process, we needed to develop a method to reduce LC/MS detection limits for nucleotide analysis. Focusing on this global purpose, we concentrated on three parameters: improved retention, peak shape, and higher peak area ratios. Intuitively, reducing the polarity of nucleotides via derivatization to form more hydrophobic entities would be expected to positively affect these parameters to varying degrees. Criteria for Selection of Derivatizing Agents. Selection of derivatizing agents was based on two major criteria: hydrophobicity and basicity. The increased hydrophobicity would serve the dual purpose of improving chromatographic performance and sensitivity. The phosphate group in dGp has two pKas, 2 and 7. At neutral pH, the phosphate moiety possesses a net negative charge. This gives the phosphate group its characteristic hydrophilicity and thus makes it difficult to analyze this class of compounds with reversed-phase chromatography. The derivatizing agents were selected based on their ability to displace the acidic hydrogen or hydroxyl group on the phosphonium moiety. We selected compounds that had at least one nitrogen to provide us with a basic group and that also possessed a relatively hydrophobic scaffold structure. Since adducts are difficult to synthesize in high
Figure 2. General reaction scheme for synthesis of dGp derivatives. Table 1. Validation Parameters for the LC/MS/MS Analysis of DGp Derivativesa
Figure 3. Representative extracted ion chromatograms of (a) dGphex and (b) dGpAAFhex derivatives.
a Solvent A, 10 mM ammonium acetate; solvent B, methanol. Methods: *1, 5/95% B/A (dGp Rt)3.00); **2, 20/80% B/A; ***3, 5%B for 3 min. Conditions: 5-30% B over 5 min, hold for 7 min at 30%. Total run time 20 min.
conversion yield, we initially tested all the derivatizing agents on pure dGp according to the general scheme shown in Figure 2. To optimize reaction conversion yields and reduce reaction time, it was essential to perform the reactions in the absence of water since water is one of the products of the reaction. This optimization permitted the reactions to go to completion and resulted in nearly quantitative conversions of greater than 90%. Selection of Optimized dGp Derivative. Table 1 summarizes the results from the synthesis of eight dGp derivatives. All of the derivatizing agents (except for the methylating agent) possessed at least one nitrogen heteroatom to augment the modified dGp’s ability to stabilize a positive charge and, thereby, facilitate the formation of a protonated ion during the electrospray ionization process. The modified mononucleotides exhibit a significant difference in chromatographic retention on the Atlantis column. The hexamethyleneimine derivative 8 was the most retention followed by N-methylpiperazine (6). These derivatives required higher organic content in their mobile-phase compositions to elute within 15 min. The piperidine derivative 7 was analyzed with method 2 and was fourth most retentive. The remaining derivatives 1-5 were run with the same method, and their corresponding retention times are listed in the far right column. A second significant feature was the product ion of each derivative. This is extremely important for quantifying and structurally identifying nucleobase adducts. When subjected to
MS/MS, dGphex exhibits a characteristic cleavage of the glycosidic bond that yields a fragment of 152 Da containing the nucleobase (Figure 3a). Similarly, derivatized dGp adducts such as dGpAAFhex yield a 373-Da fragment that also contains the nucleobase (Figure 3b). This selective and highly efficient transition is important for the sensitive detection and identification of DNA adducts by tandem MS. For example, the loss of the deoxyribosemonophosphate group facilitates recognition of all deoxynucleotide adducts in a complex mixture using constant neutral loss scanning techniques in LC/MS/MS. Furthermore, retention of the charge on the nucleobase moiety provides for the use of MS3 experiments to obtain additional structural information. Based on the above considerations, the reagents used for derivatives 1, 7, and 8 appear to fulfill the criteria for the facile formation of the aglycon ion (m/z 152 in dGp) and provide the best conversion yield for the formation of the dGp derivative. These derivatives were then selected for further evaluation by LC/ MS/MS analysis of the model DNA adduct dGpAAF. In addition, although it did not fit our general criteria of high conversion yield and characteristic product ion, the DMAP derivative 4 was included to investigate how a native cationic derivatizing agent would affect the MS/MS fragmentation of a DMAP derivatized adduct. We anticipated a quaternary amine compound such as the DMAP derivative would significantly enhance the ionization in positive ion mode while reducing background noise by the elimination of acid from the LC/MS mobile phase. However, when the acid was removed from the mobile phase, the DMAP derivative showed less ionization efficiency than when acid was present (data not shown). We suspect that the positive charge on the quaternary amine was partially neutralized by the anion on the phosphate oxygen that is not protonated above pH 2.5. This neutralization resulted in an overall lower ionization efficiency for the DMAP derivative without including acid in the mobile phase. Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
2377
Table 2. Comparison of Signal-to-Noise Ratiosand Peak Area Intensities for Modified dGp Adducts compound dGpAAF dGpAAF dGpAAF dGpAAF-CH3 dGpAAF-CH3 dGpAAF-CH3 dGpAAF-DMAP dGpAAF-DMAP dGpAAF-DMAP dGpAAF-pip dGpAAF-pip dGpAAF-pip dGpAAF-hex dGpAAF-hex dGpAAF-hex
normalized S/N 1 1 1 5 5 5 2 2 2 6 6 6 9 9 9
Rt 3.59 3.57 3.53 4.62 4.61 4.62 5.13 5.1 5.13 3.69 3.82 3.74 3.17 3.14 3.24
MS/MS2
area MS
area MS2
total
569/373 569/373 569/373 582/373 582/373 582/373 673/373 673/373 673/373 636/373 636/373 636/373 650/373 650/373 650/373
2.77 × 2.27 × 107 2.18 × 107 3.92 × 107 4.22 × 107 3.93 × 107 1.95 × 107 1.95 × 107 1.90 × 107 6.85 × 107 6.14 × 107 7.24 × 107 8.86 × 107 9.15 × 107 1.48 × 108
9.19 × 1.09 × 107 1.12 × 107 2.39 × 107 2.24 × 107 2.39 × 107 2.11 × 108 1.84 × 108 1.56 × 108 3.42 × 107 3.74 × 107 4.16 × 107 1.91 × 107 3.46 × 107 3.17 × 107
3.69 × 3.36 × 107 3.30 × 107 6.31 × 107 6.47 × 107 6.31 × 107 2.16 × 107 2.13 × 107 2.06 × 107 1.03 × 108 9.89 × 107 1.14 × 108 1.08 × 108 1.26 × 108 1.80 × 108
Selection of Derivatives for Modified DNA Adduct. N-Acetyl(2-aminofluorenyl)guanosine 5′-monophosphate was reacted to form the methyl, DMAP, piperidine, and hexamethyleneimine derivatives, and the resulting adducts were successfully analyzed in positive ion mode. In a fashion similar to the dGp, the derivatized adducts showed improved ionization efficiencies, had diagnostically important fragmentation patterns, and showed increased retention when compared to their underivatized counterparts (Table 2).To effectively compare the peak area intensities, it was necessary to sample the signals in a consistent manner. Since some derivatives were susceptible to in-source fragmentation, we summed the precursor and product ion and used the average areas to determine the peak area signal enhancement. We also set a retention time window of 3.2-5.2 min and adjusted the isocratic binary composition of each modified adduct to permit them to elute within this window. As expected, dGpAAFpip and dGpAAFhex derivatives required a higher percentage of organic in the mobile phase in order to elute within the preset retention time window. Since the dGpAAFhex was more hydrophilic than dGpAAFpip, the higher percentage of organic in the mobile phase resulted in a sharper peak and yielded better signal-to-noise ratio relative to the unmodified adduct (dGpAAF). Intuitively, the higher organic composition contributes to the increased volatility of the mobile phase. The higher volatility permits more efficient droplet formation during the electrospray process producing higher peak area ratios (Table 2). Although it is not clear whether the improved chromatography or improved ionization efficiency of the derivatives was the more influential factor, it is clear that there is a definite overall increase in signalto-noise ratio. The results in Table 2 suggest a 9- and 4-fold enhancement, respectively, in signal-to-noise ratio and peak area for dGpAAFhex relative to underivatized dGpAAF. However, since the modified and unmodified adducts fragment in-source to different degrees, it was necessary to use SRM to accurately and consistently quantify the signal-to-noise ratio and peak area enhancement. LLOD Determination. A pure sample of dGpAAF and dGpAAFhex was used to determine the limit of detection and the subsequent comparison of the unmodified and modified adducts. Both compounds were compared by injecting 50 pg of each adduct at a mobile-phase composition resulting in a retention time of 5.70 2378
Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
107
108
107
average 3.45 × 107 6.36 × 107 2.12 × 107 1.05 × 108 1.38 × 108
and 5.50 min for dGpAAFhex and dGpAAF, respectively. A minimum of 3 times signal-to-noise ratio as calculated from the spectral software was defined for the LLOD. A blank was injected before each sample and showed no interference at the retention time of each analyte. Figure 4 clearly shows a 2-fold increase in area, which can be attributed to improved ionization efficiency. However, the striking feature is the signal-to-noise enhancement of the dGpAAAFhex relative to dGpAAF. There is a minimum of 10-fold increase in signal-to-noise ratio most notably due to the lower background noise. Both signals are ∼1 × 106 counts; however, the dGpAAFhex is run at higher organic composition leading to lower background noise and a sharper peak. The result is an LLOD of 77 and 88 fmol for dGpAAFhex and dGpAAF, respectively. In fact, with the relatively large signal-to-noise ratio of the modified adduct, it should be possible to comfortably decrease the LLOD of the dGpAAFhex by an additional factor of 3. Calf Thymus in Vitro Results. To simulate derivatization conditions in an in vitro system, calf thymus DNA was reacted with AAAF, and after sample cleanup, the dG-C8-AAF adduct was derivatized with the hexamethyleneimine derivatizing agent. The modified and unmodified adducts were then compared under isocratic conditions and eluted within the same retention time range. Blanks were extracted with the calf thymus DNA to establish specificity. The results are illustrated in Figure 5 where equimolar mixtures of dGpAAF (left pane) and dGpAAFhex (right pane) are compared. The results show that the derivatized adduct has approximately twice the peak area and signal-to-noise ratio relative to its underivatized counterpart and supports the practical value of the derivatization technique. CONCLUSION We have demonstrated the utility of derivatizing deoxynucleotides for LC/MS analysis in the positive ion mode in order to enhance sensitivity and facilitate chromatographic separations under commonly used reversed-phase chromatography conditions. Previously, analysis of nucleotides was limited to negative ion mode in basic mobile phase or in positive ion mode using complex ion-pairing schemes. Retention of anionic nucleotides is minimal at best requiring high aqueous composition mobile phases. This reduces the ionization efficiency during the electrospray process
Figure 4. Evaluation of LLODs of dGpAAF and dGpAAFhex in SRM mode.
Figure 5. Analysis of calf thymus DNA adducts of (a) dGpAAF blank, (b) dGpAAF, (c) dGpAAFhex blank, and (d) dGpAAFhex.
by primarily reducing the volatility of the mobile phase. Using acidic mobile phases improves reversed-phase chromatographic peak shape by protonating exposed silanol groups. Although new generations of columns have less silanol activity, secondary interactions are still prevalent even in premium silica-based columns. Thus, separations in acidic mobile phases are the chromatographer’s preferred mode of operation. This limitation was circumvented with the described derivatation technique permitting analysis with a wide array of acidic mobile phases on conventional C18 columns.
The results clearly indicate that the hexamethyleneimine modified derivatives of dGp and dGpAAF yield higher sensitivity than their unmodified counterparts by increasing the hydrophobicity and ionization efficiency of the nucleotide adducts, which enhances their detection in positive ion electrospray mode. Our data showed a minimal 2-3-fold increase in detectability of nucleotide adducts both in aqueous solutions and when reacted with calf thymus DNA. Another important feature of this method is its specificity toward nonadducted phosphate groups. In preliminary experiments, the hexamethyleneimine derivatizing Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
2379
reagent did not displace any preformed phosphate adducts (data not shown). This unique property of hexamethyleneimine reagent will permit the simultaneous determination of nucleobase and phosphate adducts, in a single analysis, and allow for the analysis of backbone DNA phosphate adducts without the loss of information associated with enzymatic digestion. ACKNOWLEDGMENT We acknowledge the support of Eriks Rozners for the use of the ultraviolet spectrophotometer. Special thanks to the Analytical Sciences group at MDS Pharma Services for their donation of the column supplies and equipment. In addition, the National
2380
Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
Institutes of Health, Grant RO1CA6930-06 and CA77114, is thanked for their generous support. This is publication number 852 from the Barnett Institute. Note Added after ASAP Publication. The article was originally posted on 3/12/05. An error in Figure 2 was corrected. The article was reposted on 3/22/05.
Received for review November 3, 2004. Accepted January 31, 2005. AC0483724
