Anal. Chem. 1996, 68, 576-579
Effect of Impurities on the Matrix-Assisted Laser Desorption Mass Spectra of Single-Stranded Oligodeoxynucleotides Thomas A. Shaler, Juanita N. Wickham, Kristin A. Sannes, Kuang Jen Wu, and Christopher H. Becker*
Molecular Physics Laboratory, SRI International, Menlo Park, California 94025
The effect of impurities on the analysis of single-stranded DNA oligomers by the technique of matrix-assisted laser desorption/ionization with time-of-flight mass spectrometry has been studied using the matrix 3-hydroxypicolinic acid and 355-nm pulsed light. By mixing the DNA oligomers with different concentrations of impurities and recording mass spectra, limits are set on the tolerable level of a given impurity in a sample. The tolerance limits for sodium chloride, potassium chloride, sodium acetate, sodium fluoride, sodium dodecyl sulfate (SDS), and manganese(II) chloride were found to be approximately 10-2 M. It was found that magnesium salts degraded the mass spectrum at much lower levels of 10-4 M. The organic compounds tris(hydroxymethyl)aminomethane (Tris), urea, dithiothreitol (DTT), glycerol, and ethylenediaminetetraacetic acid (EDTA), when present as its ammonium salt, were tolerable at concentrations into the range of 0.25-0.5 M, while the organic polyamine compound spermine substantially degraded the mass spectrum at concentrations above 10-2 M. When comparing these results for DNA analysis with previously reported limits for protein analysis, large differences are seen for some of the impurities tested. The technique of matrix-assisted laser desorption/ionization (MALDI) for the mass analysis of biomolecules is known to be relatively insensitive to the presence of small molecule and ionic impurities. Impressive examples have been reported showing good quality mass spectra of proteins and peptides from crude biological samples.1 The analysis of DNA by MALDI had lagged behind that of proteins until the discovery of the UV-sensitive matrix 3-hydroxypicolinic acid (3HPA).2,3 Nucleic acid samples, such as those isolated from biological specimens or those that are the products of enzymatic reactions, usually contain various contaminating species. The types of contaminants typically encountered include buffer salts, various small organic compounds, and proteins. For example, in the Sanger dideoxy DNA sequencing reactions, the product mixtures contain Tris, EDTA, NaCl, MgCl2, DTT, glycerol, DNA polymerase enzyme, and bovine serum albumin as well as unreacted nucleoside triphosphates. For such reaction products to be analyzed by MALDI, it is necessary to determine whether the impurities (1) Beavis, R. C.; Chait, B. T. Proc. Natl. Acad. Sci. U.S.A., 1990, 87, 68736877. (2) Wu, K. J.; Steding, A.; Becker, C. H. Rapid Commun. Mass Spectrom. 1993, 7, 142-146. (3) Wu, K. J.; Shaler, T. A.; Becker, C. H. Anal. Chem. 1994, 66, 1637-1645.
576 Analytical Chemistry, Vol. 68, No. 3, February 1, 1996
that are present will interfere with the analysis, necessitating a purification step. It is worth noting that the current method of analyzing DNA sequencing reactions by polyacrylamide gel electrophoresis requires no purification prior to loading samples onto the gel. It is also important to consider that almost no purification method is 100% effective at removing all impurities from a sample, and for many biomolecule purification methods such as dialysis, the level of purification is directly related to the amount of time allowed for the purification. Since time is frequently an important factor for an analytical procedure, one would certainly like to perform the minimum amount of purification prior to analyzing a sample. Impurity tolerances for the MALDI analysis of proteins and peptides have been reported. Cottrell and co-workers have reported quantitative tolerance limits for various impurities that are frequently encountered in the analysis of proteins and peptides.4 In the analysis of nucleic acids by IR-MALDI with succinic acid as a matrix, Hillenkamp and co-workes have reported the effect of addition of various metal ion salts at concentrations of 20 mM.5 The extensive adduct ion formation that was observed for added salt conditions could be reduced through the use of ammonium-containing cation exchange resin, or by the direct addition of ammonium salts such as ammonium acetate.5 The addition of diammonium citrate has also been shown to be effective at reducing the level of adduct ions caused by background levels of alkali metal salts in UV-MALDI analysis of DNA.3,6 We have investigated the effects of impurity concentration on UVMALDI measurements on nucleic acids for several different impurities that might be encountered in such samples and have determined the tolerance limits of these impurities for singlestranded DNA (ssDNA) samples. The results of these studies are presented here. EXPERIMENTAL DETAILS Purified synthetic single-stranded DNA was obtained through the custom synthesis services of either National Biosciences, Inc. (Plymouth, MN) or Keystone Laboratories (Menlo Park, CA). Mixed-base 21-mer and 41-mer ss-DNA were used, with the sequences 5′-ACTCGATGGTACACGTACACGTACGTG-3′ and 5′(4) Mock, K. K.; Sutton, C. W.; Keane, A.; Cottrell, J. S. In Techniques in Protein Chemistry IV; Angeletti, R. H., Ed.; Academic Press, Inc.: San Diego, CA, 1993. (5) Nordhoff, E.; Cramer, R.; Karas, M.; Hillenkamp, F.; Kirpekar, F.; Kristiansen, K.; Roepstorff, P. Nucleic Acids Res. 1993, 21, 3347. (6) Pieles, U.; Zurcher, W.; Schar, M.; Moser, H. E. Nucleic Acids Res. 1993, 21, 3191. 0003-2700/96/0368-0576$12.00/0
© 1996 American Chemical Society
GTACATGCACATGTACATCGATGACTGCTGACTGCTAGATC3′. Tris(hydroxymethyl)aminomethane (Tris), anhydrous magnesium sulfate, sodium acetate, potassium chloride, sodium dodecyl sulfate (SDS), glycerol, magnesium chloride, manganese(II) chloride, and urea were obtained from Mallinckrodt (Paris, KY), while sodium chloride and sodium fluoride were from J. T. Baker Co. (Phillipsburg, NJ). Ethylenediaminetetraacetic acid (EDTA) and 1,2-diaminocyclohexane-N,N,N′,N′-tetraacetic acid (CDTA) were purchased from Aldrich (Milwaukee, WI). Spermine and dithiothreitol (DTT) were purchased from Sigma (St. Louis, MO). All chemicals were obtained as high-purity materials and were used without further purification. All solutions were prepared at room temperature in water that was deionized in a Millipore Milli-Q system (18 MW cm). The ss DNA was dissolved to yield 100 mM stock solutions, as determined by UV absorbance measurements at 260 nm. Stock solutions of EDTA and CDTA were prepared by suspending the free acid in a small amount of water and adding ammonium hydroxide until the material dissolved and a pH of 7 was achieved. All other solutions were prepared by standard gravimetric and volumetric techniques. Saturated solutions of the matrix 3-hydroxypicolinic acid2 (∼0.5 M) were prepared in 50 mM diammonium citrate containing 25% (v/v) acetonitrile. The additive diammonium citrate has been shown to be effective at reducing the levels of alkali-metal adduct ion formation6 as well as providing some pH buffering.3 Samples were prepared for analysis by mixing 1 mL of DNA stock solution with 1 mL of impurity solution, or with 1 mL of water for controls. A 2-mL aliquot of matrix solution was added to the mixture, and a 1-mL drop of the resulting solution was applied to the sample stage and allowed to dry in the air at room temperature. Mass spectra were recorded on a home-built 2-m time-of-flight apparatus previously described in detail.3 All spectra were obtained with sample bias voltages of 28 kV and two-stage ion extraction/acceleration (13 kV potential across the first stage). A laser wavelength of 355 nm in 5-ns pulses (Spectra Physics Model DCR-1) was used throughout the study, with energy densities slightly above the observed threshold, which is typically about 10-15 mJ/cm2 for the control samples with no impurities. Higher fluences were sometimes necessary to obtain signals above the tolerance limit, while at very high impurity concentrations, no discernible signals were obtained over wide ranges of laser intensities. Since we have previously reported that yields for positive and negative ions of ssDNA with this matrix are approximately the same,3 most of the impurities have been tested at only one polarity (positive). For some of the impurities tested, spectra were recorded in both positive and negative ion modes to compare the relative sensitivities. Tolerable levels of the impurities were set by first changing the amounts of the added impurities by factors of 5-10 in order to bracket the onset of signal degradation. Subsequent measurements were then made by varying impurity concentration by a factor of 2 or less in order to obtain as accurate a value as possible. RESULTS AND DISCUSSION The effect of impurity concentration on the positive ion MALDI TOF mass spectra for a single-stranded DNA 21-mer has been measured. We have found that the impurities investigated fell into two groups, which we label type I and type II impurities. Type I impurities were characterized by extensive, concentrationdependent adduct formation over a wide range of concentrations.
Table 1. Tolerance Limits below Which Adequate Mass Spectra Could Be Obtained for a DNA 21-mer in the Positive Ion Mode by MALDI-TOF Analysis impurity
type
tolerance limit (mM)
NaCl NaOAc SDS NaF KCl NH4OAc NH4CDTA NH4EDTA MgCl2 MgSO4 MnCl2 Tris glycerol urea dithiothreitol spermine EDTA + MgCl2
I I I I I II II II I I I II II II II I I
25 25 10 5 10 100 500 250 0.1 0.2 20 500 2% v/v (∼250 mM) 500 500 10 0.2
Type II impurities tended not to form intense adduct ions and exhibited their ill effects mainly through a decrease in signal stability and/or by preventing proper crystallization of the sample. We have assigned tolerance limits for a given impurity as being roughly the point at which adduct ions become more intense than [M + H]+ ([M - H]- for negative ions), and/or the signal-tonoise ratio decreases by more than about a factor of 4, and/or the width of the peak increases by more than about a factor of 3. The signal is noticeably degraded at this point. The measured tolerance limit for each of the impurities studied is listed in Table 1. Given the significant degree of sample-to-sample variation, we estimate that these values are accurate to within about a factor of 2. Impurity concentrations below the tolerance limits caused no or only minor adverse effects. At impurity concentrations somewhat above the tolerance limit, it was found to be necessary to increase the laser power in order to obtain a signal. Eventually, a concentration of impurity was reached which completely inhibited the signal, making it impossible to obtain a mass spectrum. Five different alkali-metal salts have been studied: sodium chloride, sodium fluoride, sodium acetate, SDS, and potassium chloride. All five were found to have tolerance limits of the same magnitude, though a slightly lower limit was found for sodium fluoride. This lower tolerance for fluoride may be due to the fact that significant amounts of HF will be formed when the sample is mixed with the acidic matrix solution, and the HF, capable of forming strong hydrogen bonds, may have a disruptive effect on the matrix-DNA interaction. Overall, the tolerance for alkalimetal salts is on the order of 10-2 M. The effect of adding different amounts of NaCl on the mass spectrum of the 21-mer is demonstrated in Figure 1. The spectrum depicted in Figure 1c was obtained from a sample containing 25 mM NaCl; this is the tolerance limit for NaCl. At this level of NaCl impurity, the signal is broader than the control by a factor of 3, and the peak maximum has shifted to correspond to that of the Na+ adduct. At a 10-fold higher concentration of NaCl (Figure 1d), the signal has degraded to an unresolved, broad distribution of Na+ adducts whose maximum corresponds to binding of about 10 sodiums to the DNA. The spectrum in Figure 1d could only be obtained by the use of substantially higher laser power. Analytical Chemistry, Vol. 68, No. 3, February 1, 1996
577
Figure 1. Positive ion MALDI TOF mass spectrum of 21-mer ssDNA showing the effect of added NaCl: (a) control sample with no added NaCl, (b) 5 mM, (c) 25 mM, and (d) 250 mM. Spectrum c corresponds to the tolerance limit for NaCl. Each spectrum is the average of 100 laser shots.
The tolerance limit for magnesium salts was found to be 2 orders of magnitude lower than that for the alkali salts, being in the 10-4 M range. The effect of magnesium salts, like the alkali salts, was also found to be independent of the identity of the counterion. In another set of experiments, the efficacy of adding the chelating agents EDTA and CDTA to magnesium-containing DNA solutions was investigated. Both EDTA and CDTA were tested alone and were found to have high tolerance limits of 0.5 M. Different levels of EDTA were added to a 21-mer sample that contained 5 × 10-3 M Mg2+, and the mass spectrum was recorded. It was found that only slight improvements in the signal could be observed by using concentrations of EDTA above 100 mM, though the spectra were still a great deal worse than those of control samples. Results depicting representative spectra from these experiments are shown in Figure 2. CDTA was found to be even less effective than EDTA. Another divalent metal salt, manganese(II) chloride, was tolerable at levels similar to those found for the alkali-metal salts. This result suggests that the 2+ charge state is not the sole cause for the greater intolerance for magnesium, even when taking into consideration the difference in effective nuclear charge between magnesium and manganese. The ionic radius, Lewis acidity, and relative hardness of the metal ions probably also play a role in their relative affinities for DNA. Most of the organic impurities that were tested were found to be well tolerated at concentrations into the range of 0.25-0.5 M. Neither Tris nor urea nor DTT had any noticeable effects in terms of the visible appearance of the sample spots, though they did make the signal less stable at concentrations near the tolerance limit. Glycerol, however, at the higher concentrations tested, prevented crystallization of the matrix and formed glassy samples. Spermine was the one organic impurity tested that formed intense adducts at low millimolar concentrations and had a tolerance limit similar to that of the alkali metal salts. The impurities Tris, EDTA, NaCl, and MgSO4 were further examined for their effect on negative ion mass spectra. Additionally, the effect of the size of the analyte was investigated for those same four impurities by comparing results for the 21-mer (MW 578 Analytical Chemistry, Vol. 68, No. 3, February 1, 1996
Figure 2. Positive ion spectra depicting the effect of adding EDTA to a solution of 21-mer ssDNA containing 5 mM MgCl2. Spectrum a is a DNA-only control which has, in addition to the [M + H]+ peak, small matrix adduct peaks. Spectrum b was obtained from a sample that contained DNA and MgCl2 only. Spectrum c was obtained from a sample that had 2 mM EDTA added, while the sample that gave spectrum d contained 200 mM EDTA. Each spectrum is the sum of 100 laser shots. Table 2. Polarity Dependence and Size Dependence of the Impurity Tolerance Limits (mM) for Selected Impurities 21-mer
41-mer
impurity
+ ion
- ion
+ ion
- ion
Tris NaCl MgSO4 EDTA
500 20 0.2 250
400 10 0.1 250
400 5 0.02 100
200 5 0.02 100
Figure 3. Positive ion MALDI TOF mass spectrum of 41-mer ssDNA showing the effect of added NaCl at concentrations of (b) 1 mM, (c) 5 mM, and (d) 25 mM. The control spectrum (a) shows a small peak on the low-mass side of the peak corresponding to a failure product 40-mer produced in the synthesis. Each spectrum is the sum of 100 laser shots.
≈ 6300) with those for a 41-mer (MW ≈ 12 300). These results are presented in Table 2, and the effect of added NaCl on the mass spectra of the 41-mer is demonstrated in Figure 3. A comparison of the tolerance limits measured for positive and negative ions reveals that, while the limits are of the same
magnitude, those for negative ions may be slightly lower when Na+ and Mg2+ are considered. Whether these minor differences are significant or not awaits a more thorough study. More significant are the differences found in the tolerance limits between the 21-mer and the 41-mer, especially for Na+ and Mg2+. The 41-mer samples were found to have lower tolerance limits for both Na+ and Mg2+compared to the 21-mer samples. This size effect is likely a result of the increased number of sites for adduct formation in the larger DNA strand. There are two potential mechanisms by which an impurity can have a deleterious effect on a MALDI mass spectrum. The first is through adduct formation. In principle, any species present in a sample that has an affinity for the analyte can attach itself to the analyte in the sample preparation or at some point during the desorption/ionization process. Analytes that have ionizable (in the Brnønsted acid-base sense) functional groups would be expected to have an affinity for, and thus be sensitive to, ionic impurities. For proteins and peptides, which have both acidic and basic groups, adduct ions to both cationic and anionic impurities can be observed. Since DNA is a polyanion in neutral solution, it interacts strongly with positively charged counterions, and likewise in the mass spectrum, adducts to metal ions are observed. The metal ion adducts occur in both positive and negative ion mass spectra of DNA to a similar degree. The degree to which cationic additives associate with DNA in solution has recently been measured by Ma and Bloomfield.7 Based on electrophoretic mobility measurements, they determined that the extent of charge neutralization of the DNA increased as the charge of the counterion increased. Our results show that solution-phase association is not an accurate predictor of the degree to which an impurity causes ill effects in the MALDI experiment. This can be seen in the comparison of Mn2+ with Mg2+. In addition, we have found under solution conditions in which Mg2+ should be completely chelated by EDTA that adduct formation was still prevalent. Nonionic organic impurities, which associate with DNA through hydrogen bonding and hydrophobic interactions, tend not to form strong adduct ion peaks in the mass spectrum except at concentrations of the impurity approaching that of the matrix. The tolerance limit for alkali-metal salt impurities in the analysis of proteins and peptides was reported by Cotrell and coworkers4 to be 1 M. The greatly reduced tolerance of alkalimetal salts by DNA (10-2 M) can be explained by the fact that peptides and proteins tend to have relatively few ionizable groups per molecule, whereas a DNA n-mer has n - 1 acidic groups per molecule. It is also worthwhile to note the dramatic difference between proteins and DNA in the tolerance for the detergent SDS. The reported tolerance limit for SDS in protein MALDI was 0.01%, which, calculated on a molar basis, is a concentration of 3 × 10-4 M. In the analysis of DNA, we find that SDS is tolerated at levels that are essentially the same as for any other alkali-metal salt. This difference might be explained by the fact that proteins have a highly folded structure in solution and can be unfolded by the detergent, while ssDNA generally exists as a random coil in solution and is less affected by the presence of detergent. DNA (7) (8) (9) (10)
Ma, C.; Bloomfield, V. A. Biopolymers 1995, 35, 211-216. Fan, X.; Beavis, R. C. Org. Mass Spectrom. 1993, 28, 1424-1429. Beavis, R. C.; Bridson, J. N. J. Phys. D: Appl. Phys. 1993, 26, 442. Jensen, O. N.; Barofsky, D. F.; Young, M. C.; von Hippel, P. H.; Swenson, S.; Seifried, S. E. Rapid Commun. Mass Spectrom. 1993, 7, 496-501. (11) Kitaigorodskii, A. I. Molecular Crystals and Molecules; Academic Press: New York, NY, 1973; pp 95-121.
is also less likely to associate with the negatively charged dodecyl sulfate ions than is a protein. The second way that an impurity can cause problems is by disrupting the crystallization of the matrix. Impurities such as glycerol, for example, which is capable of extensive hydrogen bonding, will cause drying to form a glassy rather than a crystalline sample above a certain concentration. We have found that glassy samples require the use of significantly higher laser power densities than are needed for crystalline samples and that they produce broad, poor quality spectra. Beavis and co-workers have suggested that nonvolatile liquid impurities might also inhibit ion formation by coating the crystals of the sample.8,9 The tolerable level of glycerol that we find for DNA is the same as that which has been reported for proteins.4,8 Deleterious effects of glycerol at about the same level have also recently been noted in the MALDI analysis of a cross-linked protein-nucleic acid complex.10 However, since the matrix 3-HPA has a relatively high water solubility even in cold water, we have not been able to successfully employ the technique of water-washing the dried samples to remove excess glycerol as has been reported for studies with proteins.1,8,10 The work of Beavis and co-workers has shown that peptide and protein analytes are incorporated into the crystal lattice of the matrix.9 It is expected that this incorporation holds for DNA also. While it is not possible for large analyte molecules to be incorporated substitutionally into the small matrix molecule lattice, it might be possible for the analyte to exhibit interblock solubility11 or to be incorporated in large defect sites in the crystal. Excessive impurity levels might change the quality of the crystal defects, leading to poor incorporation of the analyte and resulting in fewer good spots on the sample and a corresponding decreased signal stability. CONCLUSIONS We have determined the sensitivity to ionic and organic impurities in the MALDI analysis of ss-DNA and found that the types and levels of tolerable impurities are vastly different from previously reported values for protein analysis. The greater degree of sensitivity to ionic impurities for DNA has important implications for the MALDI analysis of DNA. Many of the reactions of the DNA-processing enzymes are carried out in saltcontaining buffers. The results are especially relevant to the analysis of the products of DNA polymerase reactions, in which magnesium is present as an enzyme cofactor, and indicate the necessity for a preliminary purification step or alternative approaches. We have observed that a simple in situ method of removing magnesium by the addition of a chelating agent (EDTA) is only partially successful at reducing the level of magnesium adduct ion formation in the mass spectrum. ACKNOWLEDGMENT Financial support from the National Institutes of Health, National Center for Human Genome Research (Grant No. HG00174), is gratefully acknowledged. J.N.W. thanks the National Science Foundation for support through the Research Experience for Undergraduates Program. Received for review March 15, 1995. Accepted October 25, 1995.X AC9502662 X
Abstract published in Advance ACS Abstracts, December 1, 1995.
Analytical Chemistry, Vol. 68, No. 3, February 1, 1996
579
