Enhanced Detection of Oligonucleotides in UV MALDI MS Using the

Anal. Chem. 1999, 71, 2866-2870

Enhanced Detection of Oligonucleotides in UV MALDI MS Using the Tetraamine Spermine as a Matrix Additive John M. Asara and John Allison*

Department of Chemistry, Michigan State University, East Lansing, Michigan 48824

Matrix-assisted laser desorption/ionization mass spectrometry (MALDI MS) has been used to analyze oligonucleotides. However, success has been limited by cation adduction and high detection limits. Both of these problems are due to the high net negative charge that oligonucleotides carry on the phosphodiester backbone. Comatrixes such as ammonium salts with UV absorbers such as 3-hydroxypicolinic acid, 2,4,6-trihydroxyacetophenone, and 6-aza-2-thiothymine have been used to improve the spectral quality for oligonucleotides in MALDI MS. Organic bases have also been used as co-matrixes; however, the most popular matrix, 3-hydroxypicolinic acid, is not compatible with these additives. We have found that the tetraamine spermine as a matrix additive can successfully eliminate cation adduction and lower the detection limits for DNA in the MALDI experiment, without having to resort to desalting steps. The results suggest that multiply protonated spermine molecules function better than ammonium ions in neutralizing oligonucleotides and displacing alkali cations. Protonated spermine is chemically similar to ammonium ions since it binds to the phosphate backbone and releases protons to the phosphate groups. Spermine can be used successfully with the matrixes 6-aza-2-thiothymine and 80% anthranilic acid/ 20% nicotinic acid but not with 3-hydroxypicolinic acid. The additive also works well for the analysis of metalated DNA. Matrix-assisted laser desorption/ionization mass spectrometry (MALDI MS) has been used to analyze oligonucleotides.1-3 Applications of MALDI in this area can be limited by alkali cation adduction and high detection limits.1,2,4 In MALDI MS, spectra for oligonucleotides are “typically acquired from 1 to 10 pmol of a nucleic acid sample”.1 In contrast, low femtomole levels of DNA have been analyzed using nanoelectrospray with FT-MS.5 The high * To whom correspondence should be addressed. E-mail: allison@ cem.msu.edu. (1) Nordhoff, E.; Kirpekar, F.; Roepstorff, P. Mass Spectrom. Rev. 1996, 15, 67-138. (2) Murray, K. K. J. Mass Spectrom. 1996, 31, 1203-1215. (3) Limbach, P. A. Mass Spectrom. Rev. 1996, 15, 297-336. (4) Roskey, M. T.; Juhasz, P.; Smirnov, I. P.; Takach, E. J.; Martin, S. A.; Haff, L. A. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 4724-4729. (5) McLafferty, F. W.; Kelleher, N. L.; Begley, T. P.; Fridriksson, E. K.; Zubarev, R. A.; Horn, D. M. Curr. Opin. Chem. Biol. 1998, 5, 571-578.

2866 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

net negative charge that oligonucleotides can accumulate, due to the -1 charge on each phosphodiester group along the DNA backbone, complicates the MALDI analysis. There are two aspects to the charge state of the analyte that are relevant in MALDI analysis. If the analyte is trapped in the matrix in ionic form, desorption may or may not be possible depending on the charge. Sufficient energy appears to be unavailable for the desorption of multiply charged species in MALDI, which predominantly forms singly charged species.6 This results in either high detection limits or no signal at all. If the multiply charged anionic analyte lowers its charge in either the precipitation process or the desorption process, it frequently does so by extracting protons from the matrix and/or forming adducts with alkali metal ions such as Na+ and K+.1,2 In this way, the charge state of the molecule is lowered, allowing for desorption; however the signal may be distributed over many m/z values, since a variety of ionic species of the type [(analyteq-)(H+)x(Na+)y(K+)z]- are formed. For high-mass species that cannot be isotopically resolved, the presence of these peaks results in poor resolution and peak asymmetry, making accurate m/z assignments difficult.1 This problem also raises detection limits and frequently requires cleanup steps such as RP-HPLC, microdialysis, or cation exchange in the analysis to remove salts prior to MALDI target formation.1 These complications have limited the ability of MALDI MS “to routinely sequence oligonucleotides 15-20 bases long”.7 It has been shown in fast atom bombardment mass spectrometry (FAB MS), another desorption/ ionization (D/I) technique that produces mainly singly charged ions, that matrix additives can lower the charge on polyionic analytes and allow for more efficient D/I of highly charged analytes. Additives developed at Michigan State University for specific applications include the use of triflic acid for the analysis of positively charged organometallic complexes8 and the use of octylnicotinium bromide for the enhanced detection of negatively charged phosphate-containing biomolecules.9 Highly charged oligonucleotides and phosphopeptides have been analyzed with improved D/I efficiency through the use of (6) Wu, Z.; Biemann, K. Int. J. Mass Spectrom. Ion Processes 1997, 165-166, 349-361. (7) Henry, C. Anal. Chem. 1997, 2433A-246A. (8) Asara, J. M.; Uzelmeier, C. E.; Dunbar, K. R.; Allison, J. Inorg. Chem. 1998, 37, 1833-1840. (9) Huang, Z.-H.; Shyong, B.-J.; Gage, D. A.; Noon, K. R.; Allison, J. J. Am. Soc. Mass Spectrom. 1994, 5, 935-948. 10.1021/ac981406l CCC: $18.00

© 1999 American Chemical Society Published on Web 05/27/1999

matrix additives.1,2,10 Matrix additives can reduce adduct formation with alkali cations and provide protons to lower the net negative charge on these complexes. Ammonium acetate was first used to displace alkali metal ions with NH4+ successfully in the MALDI experiment by Currie et al.11 Since then, many applications using ammonium salts have appeared in the literature12,13 with the most successful being the use of diammonium hydrogen citrate (DAHC) by Pieles et al.14 Ammonium salts are very useful additives since complexed ammonium ions lose ammonia upon desorption/ionization, leaving protons behind. DAHC as a matrix additive in MALDI has been used in combination with many matrixes such as 2,4,6-trihydroxyacetophenone (THAP), 6-aza-2thiothymine (6-ATT), 80% anthranilic acid/20% nicotinic acid (80/ 20), and 3-hydroxypicolinic acid (3-HPA).1 3-HPA has become the most popular matrix for the analysis of DNA.1,2 However, ammonium ions do not completely displace alkali metal cations, and they have less of an effect for higher mass analytes.1 Basic neutral amines such as triethylamine and imidazole have also been shown to reduce the formation of alkali metal cation adducts and improve spectral quality for DNA in MALDI MS as well.15 The development of new additives for DNA analysis by MALDI MS is important, offering advantages such as improved resolution and enhanced sensitivity. Here we demonstrate the use of the tetraamine spermine as a useful matrix additive for the direct MALDI analysis of oligonucleotides, without incorporating a desalting step. Spermine, NH2(CH2)3NH(CH2)4NH(CH2)3NH2, is found in all organisms.16,17 Polyamines such as putrescine, spermidine, and spermine are essential to all living cells; their depletion has been shown to inhibit cell growth and affect gene expression.17 Spermine, in its tetraprotonated form at cell pH,18 stabilizes DNA by shielding the negative charges from the phosphate backbone, thus reducing negative charge repulsion within and between DNA strands.17,19,20 With complexed spermine, DNA condenses into a more compact structure.17,19-21 Divalent cations such as Mg2+ have also been shown to serve this function.20-22 Polyamines are therefore regulators of secondary and tertiary structures of nucleic acids.20 The X-ray structure of spermine crystallized from a phosphatebuffered solution shows that its amino groups form direct hydrogen bonds to phosphate oxygens.20 In DNA, spermine molecules are almost always found in areas of high phosphate density.17 However, X-ray crystallographic data reported in recent (10) Asara, J. M.; Allison, J. J. Am. Soc. Mass Spectrom. 1999, 10, 35-44. (11) Currie, G. J.; Yates, J. R.,III. J. Am. Soc. Mass Spectrom. 1993, 4, 955-963. (12) Zhu, Y. F.; Taranenko, N. I.; Allman, S. L.; Martin, S. A.; Haff, L.; Chen, C. H. Rapid Commun. Mass Spectrom. 1996, 10, 1591-1596. (13) Li, Y. C. L.; Cheng, S.-W.; Chan, T.-W. D. Rapid Commun. Mass Spectrom. 1998, 12, 993-998. (14) Pieles, U.; Zu ¨ rcher, W.; Scha¨r, M.; Moser, H. E. Nucleic Acids Res. 1993, 21, 3191-3196. (15) Simmons, T. A.; Limbach, P. A. Rapid Commun. Mass Spectrom. 1997, 11, 567-572. (16) Esposito, D.; Del Vecchio, P.; Barone, G. J. Am. Chem. Soc. 1997, 119, 2606-2613. (17) Tippin, D. B.; Sundaralingam, M. J. Mol. Biol. 1997, 267, 1171-1185. (18) Nakai, C.; Glinsman, W. Biochemistry 1977, 16, 5636-5641. (19) Lodish, H.; Baltimore, D.; Berk, A.; Zipursky, S. L.; Matsudaira, P.; Darnell, J. Molecular Cell Biology; Scientific American: New York, 1995; pp 344346. (20) Feuerstein, B. G.; Williams, L. D.; Basu, H. S.; Marton, L. J. J. Cell. Biochem. 1991, 46, 37-47. (21) Bloomfield, V. A. Biopolymers 1997, 44, 269-282. (22) Cohen, S. S. Nature 1978, 274, 209-210.

years suggest that spermine may also bind to DNA bases to a small extent.17,20 While many spermine molecules could complex with a single DNA strand, crystallized forms of DNA with only a small number of spermine molecules attached have been characterized.17,20,22,23 This raises the question of whether there are extensive or limited interactions of spermine with DNA in solution. The role of spermine in reducing the net charge of DNA in solution suggests that it should be evaluated as an additive, to participate in a similar process in the MALDI experiment. It should also be noted that spermine is frequently added to solutions from which oligonucleotides are crystallized. In some cases, its presence is required for crystal growth,17,20,23 although the polyamine is not incorporated into the crystal that is formed. Since crystal formation is part of the MALDI process, the role of spermine in this experiment may provide information on the extent of spermine-DNA interactions in the solution from which the MALDI target is formed. Spermine has been cited in the past as a contaminant in the MALDI analysis of oligonucleotides,24 although this is not surprising since 3-HPA was the matrix used in that study; it has been found that 3-HPA as a matrix is incompatible with organic bases.15 When spermine was considered as an impurity, it was reported that, for a particular 21-mer, the intensity of a spermine adduct of the analyte matched that of the protonated analyte in the MALDI spectrum, at a spermine concentration of 10 mM. In the spectra shown here, using other matrixes, spermine adducts are not observed. It will be shown that spermine, when used as a matrix additive in MALDI for the analysis of oligonucleotides, can completely eliminate cation adduction and lower detection limits. The spermine molecule functions much like ammonium ions since it appears to bind to the phosphate backbone and release protons to the phosphate groups, either in the crystal growth process or in the subsequent desorption/ ionization step of MALDI. This additive can be used with those most commonly used MALDI matrixes for oligonucleotide analysis, notably 6-ATT and 80/20. It will also be shown that the MALDI spectrum of metalated DNA can be enhanced using spermine as well. EXPERIMENTAL DETAILS The 12-mer oligonucleotide d(CCTCTGGTCTCC) was purchased from the Yale University Oligonucleotide Synthesis Facility (New Haven, CT). The 17-mer sequencing primer d(GTAAAACGACGGCCAGT) was purchased from Boehringer Mannheim (Indianapolis, IN). The platinated 12-mer d(CCTCT{GGPt2+}TCTCC) was a gift from Dr. Kim R. Dunbar at Michigan State University. All oligonucleotides were desalted via size exclusion chromatography using a Sephadex G-25 Fine (Pharmacia Biotech, Piscataway, NJ) column. The oligonucleotides were prepared in water at a stock concentration of 5 pmol/µL and then further diluted to 1 pmol/µL and 75 fmol/µL. An aliquot of d(CCTCTGGTCTCC) was also prepared at a concentration of 1 pmol/µL in10 mM NaCl. An aliquot of d(CCTCT{GGPt2+}TCTCC) was prepared in 20 mM sodium acetate at a concentration of 1 pmol/µL. The d(CCTCTGGTCTCC) strand was digested in water with ∼50 units of exonuclease III, a 3′f 5′ exonuclease isolated from E. coli. The (23) Egli, M.; Williams, L. D.; Gao, Q.; Rich, A. Biochemistry 1991, 30, 1138811402. (24) Shaler, T. A.; Wickham, J. N.; Sannes, K. A.; Wu, K. J.; Becker, C. H. Anal. Chem. 1996, 68, 576-579.

Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

2867

Figure 1. MALDI mass spectra of d(GTAAAACGACGGCCAGT) in the presence of 10 mM NaCl in (a) 3-hydroxypicolinic acid with diammonium hydrogen citrate, (b) 6-aza-2-thiothymine with diammonium hydrogen citrate, and (c) 6-aza-2-thiothymine with spermine.

digestion was monitored every 30 min for 6 h and then at 12 h and 24 h. The matrixes 3-hydroxypicolinic acid, 6-aza-2-thiothymine, anthranilic acid, and nicotinic acid were purchased from Sigma Chemical (St. Louis, MO). When spermine (Fluka, Milwaukee, WI) was used as an additive, it was prepared at a concentration of 12.5 mM in saturated matrix solution (1/1 acetonitrile/water). When diammonium hydrogen citrate (J. T. Baker, Phillipsburg, NJ) was used, it was prepared at a concentration of 25 mM in 1/1 acetonitrile/water in saturated matrix solution. Samples were prepared by mixing 1 µL of analyte solution and 1 µL of co-matrix solution and allowed to air-dry on a gold sample plate. Linear MALDI mass spectra were recorded on a Perseptive Biosystems (Framingham, MA) Voyager Elite delayed extraction time-of-flight reflectron mass spectrometer equipped with a nitrogen laser (337 nm, 3 ns pulse). The accelerating voltage was 23 kV, the delay time selected was 50 ns, the grid voltage was set to 93.0% of the accelerating voltage, and the guide wire voltage was set to a magnitude of 0.25% of the accelerating voltage. Typically, 64-128 laser shots were averaged for each spectrum. All spectra shown were taken in the negative-ion mode. RESULTS AND DISCUSSION We show here that the addition of spermine to the MALDI matrix solution results in improved spectra for single-stranded oligonucleotides and that it eliminates the need for a desalting step when such analytes are under study. Figure 1a shows the MALDI spectrum of 1 pmol of a 17-mer deoxyribonucleotide using 3-HPA/DAHC as the co-matrix. This is a commonly used matrix for such analytes. To demonstrate the complications that arise when salts are present, the sample was introduced in a solution that was also 10 mM in NaCl. Note that the ammonium salt cannot 2868 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

successfully displace all of the sodium ions for the high concentration of NaCl used and the large number of potential sites at which Na+ could form a complex. For this discussion, M will be used to indicate the neutral form of the analyte, in which all of the negative charges have been neutralized by protons (oligon- + nH+). Thus, if the analyte forms a negatively charged species in the MALDI experiment which contains one Na+, it would lose two protons, hence the designation of [M + Na - 2H]-. Figure 1b shows the MALDI spectrum obtained for the same 17-mer, with another commonly used matrix combination, 6-ATT/DAHC. While the range of sodium ion adducts may be smaller than that shown in Figure 1a, the resolution is lower. This results in a very broad signal that spans nearly 100 Da due to cation adduction. Figure 1c shows the spectrum when 6-ATT/spermine is used as the matrix. Even at the large salt concentration, spermine is effective, much more so than ammonium ions, in displacing Na+ ions and providing protons to negatively charged phosphate groups for neutralizationsyielding a single peak for the deprotonated form of the neutral molecule at m/z 5223. Without alkali metal ion adducts, the highest resolution of the instrument can be realized, and most meaningful m/z values can be determined. The capability to simply eliminate cationization is very important, especially when mixtures of oligonucleotides with similar molecular weights are under study; an extensive series of alkali metal ion adducts can mask the presence of a second component. The result of spermine addition is also an increase in the signal-to-noise (S/N) ratio of the experiment. Spermine is not as effective with 3-HPA; therefore, other matrixes such as 6-ATT must be used. For 6-ATT, spermine clearly is more effective than ammonium salts. The interaction of metal ions such as platinum with DNA is important to be able to study, since the anticancer activity of cisplatin has been related to strong DNA-Pt interactions.25,26 Of course, for MALDI to be useful in studying such interactions, the complex must remain intact during all of the processes associated with the analysis. That is, the metal should not be displaced from the oligonucleotide by either matrix molecules or additives. Figure 2a shows the spectrum of a platinated DNA strand using 3-HPA/ DAHC as the co-matrixes. The spectrum shows that many sodium adducts are formed. In this case, the sample was prepared/ supplied for analysis in a sodium acetate-buffered solution. While the spacing of the peaks clearly indicates that sodium ion adducts are being formed, even in the presence of ammonium ions, one cannot assume that the lowest m/z peak in the cluster represents the [M - H]- ions. Again, the alkali metal ion adduct formation makes the detection of additional species of similar mass difficult. Figure 2b shows the MALDI spectrum of the same sample using the 80/20/DAHC matrix. Apparently, with this matrix, ammonium ions are more effective in displacing sodium ions. It has been shown previously that the 80/20 matrix is a good choice when one attempts to detect platinated DNA.27 Figure 2c shows the MALDI spectrum of the same platinated DNA compound using 80/20 and spermine. The sodium adducts have completely been eliminated by spermine, giving a sharp peak representing [M + Pt2+(NH3)2 - 3H]- at m/z 3773. A peak for the free DNA is also seen as the [M - H]- ion at m/z 3546. We believe that this is (25) Sherman, S. E.; Lippard, S. J. Chem. Rev. 1987, 87, 1153-1181. (26) Reedijk, J. Inorg. Chim. Acta 1992, 198-200, 873-881. (27) Gonnet, F.; Kocher, F.; Blais, J. C.; Bolbach, G.; Tabet, J. C.; Chottard, J. C. J. Mass Spectrom. 1996, 31, 802-809.

d(CCTCT{GGPt2+}TCTCC)

Figure 2. MALDI mass spectra of in (a) 3-hydroxypicolinic acid with diammonium hydrogen citrate, (b) 80% anthranilic acid/20% nicotinic acid with diammonium hydrogen citrate, and (c) 80% anthranilic acid/20% nicotinic acid with spermine.

indicative of an incomplete reaction, rather than an indication of an equilibrium that exists between the platinated and platinumfree forms, since other spectra have been obtained for the same metal complex, more highly purified by HPLC, in which the peak at m/z 3546 is not present. In working with this Pt-DNA complex, we have found that the 80/20 matrix is superior to 6-ATT, in terms of allowing the metalated complex to remain intact. This is consistent with data reported by Gonnet et al.27 A comment should be made that, using other matrixes and excitation wavelengths, Pt(NH3)2+-DNA complexes have been shown to exhibit monoand diammine losses.28 This is not observed for the systems studied here. While the data shown are typical for platinum complexes, using the two matrixes and two additives shown, complexes involving other metals are much more sensitive to the chemical nature of the matrix.29 The data shown in Figure 2 show that spermine as a co-matrix eliminates the need for desalting and purification procedures. Spermine, particularly with the 6-ATT matrix, also allows for detection of oligonucleotides at lower levels than with DAHC. Without spermine, even with purified samples that are Na+ free, detection limits are higher. Figure 3a shows a portion of the MALDI spectrum of 75 fmol of a desalted 12-mer deoxyribonucleotide using the most common matrix combination of 3-HPA and diammonium hydrogen citrate. Using our MALDI-TOF instrument, no signal is obtained at this low femtomole concentration. Figure 3b shows the MALDI spectrum of the same 12-mer using (28) Costello, C. E.; Nordhoff, E.; Hillenkamp, F. Int. J. Mass Spectrom. Ion Processes 1994, 132, 239-249. (29) Asara, J. M.; Dunbar, K. R.; Allison, J. Unpublished results. (30) Guo, L.-H.; Wu, R. In Methods in Enzymology; Colowick, S. P., Kaplan, N. O., Eds.; Academic: New York, 1983; Vol. 100, p 60.

Figure 3. MALDI mass spectra of 75 fmol of d(CCTCTGGTCTCC) in (a) 3-hydroxypicolinic acid with diammonium hydrogen citrate and (b) 6-aza-2-thiothymine with spermine.

the 6-ATT/spermine combination. There is a substantial increase in desorption/ionization efficiency observed. It is also noteworthy that, for this particular sample, the 75 fmol of analyte is deposited with 12.5 nmol of spermine. Even when the spermine/analyte ratio is ∼105, this additive interacts with multiple sites and provides protons to lower the charge state, but the interactions are sufficiently weak that all interacting spermine molecules are lost at some point in the MALDI process. That is, there is no peak representing the adducts of one or more spermine molecules with the analyte (each spermine would add a mass of 202 Da). We have found the 6-ATT/spermine combination to be useful in the analysis of mixtures of oligonucleotides as well. We had been doing an experiment to study the utility of exonuclease III. This enzyme is typically used to digest double-stranded DNA starting at the 3′ end.30 Figure 4 shows the results of an exonuclease III digestion of the same 12-mer used in Figure 3, from one point 12 h into a digestion. In this experiment, 1 pmol of desalted DNA was being digested and one-fourth of the sample was mixed with the common 3-HPA/DHAC matrix mixture. Thus, with approximately 50 fmol of each digestion product present, two or three components could be detected, as shown in Figure 4a. If the same sample amount is analyzed using the 6-ATT/spermine combination, more components of the mixture are detected, and at higher S/N ratios. Much of the sequence can be determined using the peaks shown. It is also noteworthy that the relative response apparently is different for the two matrix mixtures as well. Also, notice the peak at m/z 1840 is 98 Da higher than the peak labeled G, representing H3PO4. This peak may be due to an incomplete digestion product. Finally, a comment should be made on the utility of spermine as an additive. Since spermine is not compatible with 3-HPA, the most effective matrix for oligonucleotide applications, we suggest that the real strength of using spermine is in the analysis of Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

2869

Figure 4. MALDI mass spectra of ∼ 50 fmol of each component of a 3′ exonuclease digestion mixture (total amount initially ) 250 fmol) of d(CCTCTGGTCTCC) in 3-hydroxypicolinic acid with diammonium hydrogen citrate and in 6-aza-2-thiothymine with spermine.

samples at 1 pmol or below, without a desalting step. When larger quantities of DNA (>1 pmol) are characterized, both 3-HPA/NH4+ and 6-ATT/spermine may be reasonable matrix choices, depending on the salt content and mass of the oligonucleotide. CONCLUSION In the past two decades, the role that polyamines play in living cells at the molecular level has been an important topic of research. Since spermine-assisted crystallization leads to the neutral (protonated) form of DNA, it follows that proton transfer from protonated spermine nitrogens to negatively charged phos-

2870 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

phate groups takes place in the crystallization step of the MALDI experiment, not the desorption event. If spermine molecules remain attached after crystallization is complete, they are lost in the desorption step. It is not surprising that spermine is more effective than ammonium ions in neutralizing oligonucleotides. In the dynamics of complexation of positively charged protonated nitrogen centers with negatively charged phosphates, multidentate ligands are expected to yield the more stable products and to have higher formation constantssanalogous to simpler systems involving chelating agents. The fact that the mass spectral results show such complete formation of the deprotonated form of the DNA analyte suggests that spermine-oligonucleotide interactions are extensive and complete. Spermine appears to sample all phosphate groups and provide protons to these sites. Such data may be useful in modeling polyamine-nucleic acid complexes, certainly suggesting more than a passive counterion role for protonated polyamines. ACKNOWLEDGMENT The authors thank Dr. James H. Geiger for suggesting the use of polyamines in MALDI analysis and Dr. Kim R. Dunbar and Elizabeth Lozada for desalting the oligonucleotides and for providing the cis-platin oligonucleotide complex. Mass spectral data were acquired at the MSU Mass Spectrometry Facility, which is partially supported by Grant RR-00480 from the Biotechnology Research Technology Program of the National Center for Research Resources of the NIH. Received for review December 18, 1998. Accepted April 9, 1999. AC981406L