Anal. Chem. 2000, 72, 5092-5096
Analysis of the Degradation of Oligonucleotide Strands During the Freezing/Thawing Processes Using MALDI-MS Darryl L. Davis, Edward P. O’Brien, and Catherine M. Bentzley*
Department of Chemistry and Biochemistry, University of the Sciences in Philadelphia, 600 South 43rd Street, Philadelphia, Pennsylvania 19104
Synthetic oligonucleotide strands ranging from 5 to 25 units in length are commonly used as standards, probes, and templates in various bioanalytical applications. Until recently, their preparation, storage, and handling were regarded as unimportant, but this work provides valuable information to the contrary. The systematic degradation of oligonucleotide strands during sample preparation is investigated by repeatedly freezing/thawing short strands followed by matrix-assisted laser desorption ionization mass spectrometric (MALDI-MS) analysis. It is shown here that the longevity of an oligonucleotide strand is dependent on several factors including base composition, solution concentrations, and strand length as well as thawing conditions. Several trends in strand robustness were established. Our studies reveal that the robustness of strands is base-dependent: T-mer > A-mer > C-mer > G-mer. Likewise, an increase in the length of the strands increases the tendency of a sample to degrade. Another observation included that samples of mixed bases degrade according to structural conformations. All of these observations are attributed to the fact that the samples undergo degradation during sample/solvent isolation during freezing. The evolution of MALDI-MS has spurred on its application as a technique for the detection, analysis, and quantification of synthetic oligonucleotide strands, which are often used as standards, chemical sensors, templates, probes, and even drugs.1-5 As a result, much work has been done to improve MALDI-MS analysis including the optimization of instrumental/tuning parameters,6,7 the discovery of improved matrixes such as ATT,8-10 as * To whom correspondence should be addressed: (telephone) 215 596-8581, (fax) 215 596-8543, (e-mail)
[email protected]. (1) Chen, X.; Fei, Z.; Smith, L. M.; Bradbury, E. M.; Majidi, V. Anal. Chem. 1999, 71, 3118-3125. (2) Boncheva, M.; Scheibler, L.; Lincoln, P.; Vogel, H.; Akerman, B. Langmuir 1999, 15, 4317-4320. (3) Albagli, D.; Van Atta, R.; Cheng, P.; Huan, B.; Wood, M. J. Am. Chem. Soc. 1999, 121, 6954-6955. (4) Altman, R. K.; Schwope, I.; Sarracino, D. A.; Tetzlaff, C. N.; Bleczinski, C. F.; Richert, C. J. Comb. Chem. 1999, 1, 493-508. (5) Trabesinger, W.; Schultz, G. J.; Gruber, H. J.; Schindler, H.; Schmidt, T. Anal. Chem. 1999, 71, 279-283. (6) Saurina, J.; Hernandez-Cassou, S.; Tauler, R.; Izquierdo-Ridorsa, A. Anal. Chem. 1999, 71, 126-134.
5092 Analytical Chemistry, Vol. 72, No. 20, October 15, 2000
well as the investigation of sensitivity enhancers.11 Fortunately these fundamental advances have permitted the extension of MALDI analysis to even more complicated systems such as oligonucleotide metabolites,12 PCR products,13 and single-nucleotide polymorphisms.14 In addition the technique has been elevated to include sequencing work through enzymatic digestion15,16 or energetic dissociation.17-19 Despite all of these advances in oligonucleotide detection by MALDI, few investigative studies have been performed on the solution-handling process prior to analysis. The questions arise: What are the detrimental effects resulting from the everyday storage and handling conditions of typical strands? Are extra precautions required for different strands? Unfortunately, several researchers have already noted the degradation of these commercial samples as a result of time, detrimental storage conditions, or the repetitiveness of the freezing/thawing process.20,21 For example, Anchordoquy et al.20 noted that the fast freezing/thawing cycle was more damaging to a lipid/DNA complex than a slow freezing/thawing process. This observation of a typical drug delivery system was not only informative but also cost-effective for the drug manufacturer involved. (7) Little, D. P.; Cornish, T. J.; O′Donnell, M. J.; Braun, A.; Cotter, R. J.; Koster, H. Anal. Chem. 1997, 69, 4540-4546. (8) Juhasz, P.; Roskey, M. T.; Smirnov, I. P.; Haff, L. A.; Vestal, M. L.; Martin, S. A. Anal. Chem. 1996, 68, 941-946. (9) Tang, W.; Nelson, C. N.; Zhu, L.; Smith, L. M. J. Am. Soc. Mass Spectrom. 1997, 8, 218-224. (10) Thiede, B.; Von Janta-Lipinski, M. Rapid Commun. Mass Spectrom. 1998, 12, 1889-1894. (11) Simmons, T. A.; Limbach, P. A. Rapid Commun. Mass Spectrom. 1997, 11, 567-572. (12) Muddiman, D. C.; Cheng, X.; Udseth, H. R.; Smith, R. D. J. Am. Soc. Mass Spectrom. 1996, 7, 697-706. (13) Apffel, A.; Chakel, J. A.; Fischer, S.; Lichtenwalter, K.; Hancock, W. S. Anal. Chem. 1997, 69, 1320-1325. (14) Doktycq, J.; Hurst, G. B.; Goudarzi, S.; McLuckey, S. A.; Tang, K.; Chen, H.; Usiel, M.; Jacobson, K. B.; Woychik, R. P.; Buchanan, M. V. Anal. Biochem. 1995, 230, 205-214. (15) Haff, L. A.; Smirnov, I. P. Genome Res. 1997, 7, 378-388. (16) Smirnov, I. P.; Roskey, M. T.; Juhasz, P.; Takach, E. J.; Martin, S. A.; Haff, L.A. Anal. Biochem. 1996, 238, 19-25. (17) Bentzley, C. M.; Johnston, M. V.; Larsen, B. S.; Gutteridge, S. Anal. Chem. 1996, 68, 2141-2146. (18) Murray, K. K. J. Mass Spectrom. 1996, 31, 1203-1215. (19) Limbach, P. A. Mass Spectrom. Rev. 1996, 15, 297-336. (20) Anchordoquy, T.; Girouard, L.; Carpenter, J.; Kroll, D. J. Pharm. Sci. 1998, 87, 1046-1051. (21) Bartolini, W.; Bentzley, C. M.; Johnston, M. V.; Larsen, B. S. J. Am. Soc. Mass Spectrom. 1999, 10, 521-528. 10.1021/ac000225s CCC: $19.00
© 2000 American Chemical Society Published on Web 09/15/2000
Since it is not always possible to store oligos under ideal conditions (-20 °C, pH > 6), the repetitive freezing/thawing of an oligonucleotide strand can lead to the undesirable decomposition of a sample through the loss of a terminal phosphate group, a base, or the entire oligonucleotide unit. This observation relates to the fact that during the freezing/thawing process the oligonucleotide and buffer components isolate from the solvent system. The result is that this isolation increases the percentage of similar intermolecular attractions and bond breakages. To our knowledge, no comparative study exists which evaluates the fundamental ruggedness of various oligonucleotide strands. Therefore, this work represents a systematic analysis to distinguish the robustness of various oligonucleotide samples. Degradation results are presented as a function of oligonucleotide base composition, concentration, length, and freezing/thawing conditions. Solutions of constant base 5-mers (5′ AAA AA 3′, 5′ CCC CC 3′ etc.) as well as various strands are analyzed for their durability during several repetitive freezing/thawing cycles. The freezing/thawing rates are also studied by applying two methods: slow freezing at 0 °C or rapid freezing by placing the sample in liquid nitrogen. MALDI-MS is an ideal technique for this analysis due to its efficient and effective sample preparation. Once the method is established for oligonucleotide MALDI analysis, it only requires minutes from sample preparation to data collection. In comparison, traditional native gel electrophoresis analysis often requires hours to days for completion. The one drawback of MALDI analysis is the possibility of fragmentation during the ionization/desorption process. However, previous researchers have illustrated that the decomposition of oligo strands during the desorption/ionization event is directly proportional to molecular weight.22 Therefore, to avoid this complication, samples of relatively short length (5mers to 10-mers) were intentionally chosen to ensure the absence of degradation products from the ionization/desorption process. EXPERIMENTAL SECTION Systematic Degradation of Oligonucleotides. All oligonucleotides (Midland Certified Reagent Co., Midland, TX) were dissolved in Millipure water to yield the appropriate concentrations ranging from 1 to 50 µM. From each prepared solution 100 µL was transferred into a 1.5-mL centrifuge tube. The oligonucleotide solutions were then repeatedly frozen/thawed by one of the following techniques. During the first procedure, which tested the robustness of the sample strands under stressed situations, the solutions were frozen in liquid N2 for 1 min and then thawed in a water bath (65 °C) for 5 min. The second technique provided a comparison of typical, everyday laboratory conditions with our simulated, accelerated conditions. This method was performed under the following less harsh environment. The samples were frozen in a laboratory freezer (0 °C) for 5 h, followed by thawing on the laboratory benchtop (31 °C) for 1 h. After 10 freezing/thawing cycles, 1 µL of sample was aliquoted from the centrifuge tube and prepared for MALDI analysis. Thawing caused during shipping or from the daily freezer removal was not included as part of the cycling process since its effects are considered minimal. Each sample was repeated in triplicate to ensure reproducibility. (22) Wu, J.; Shaler, T.; Becker, C. Anal. Chem. 1994, 66, 1637-45.
Figure 1. MALDI mass spectra of a 50 µM solution of 5′ AAA AA 3′ acquired after 0, 20, and 70 cycles of freezing in liquid nitrogen followed by thawing in a 65 °C bath. ATT was used as the matrix.
Acquisition of MALDI Spectra. All spectra were acquired on the PerSeptive Biosystems Voyager-DE Biospectrometry Workstation (Framingham, MA). Each spectrum was the summation of 100 laser shots. Each sample solution was prepared/ analyzed in triplicate, and the average values of the three trials were calculated for each sample. The instrumental conditions were as follows: linear mode and negative-ion detection. In this mode, a negative 20 000 acceleration voltage is used with the delayed extraction option activated at 50 ns. After every 10 freezing/thawing cycles, 1 µL of sample was removed from the centrifuge tube and mixed with the appropriate matrix. The 10 mg/mL solution of the matrix 6-aza-2-thiothymine (Sigma Chemical Co., St. Louis, MO) was prepared daily in a 50/ 50 (v/v) solution of acetonitrile and 0.1 M ammonium citrate. ATT was chosen for this study since it is the established matrix for short oligonucleotide analysis.23 Matrix and analyte solutions were spotted on the MALDI sample stage in a 2:1 ratio (2 µL of matrix to 1 µL of analyte) and allowed to air-dry. After visual inspection, a water wash was performed on the MALDI spot to remove excess salt when necessary.24 All spectra were acquired by averaging 100 shots acquired by rastering across the sample spot. Data were analyzed using the software program Grams 386. RESULTS AND DISCUSSION Influence of the Base Composition on Decomposition. Figure 1 represents the decomposition of 5′AAA AA 3′ after undergoing 0, 20, and 70 repetitive freezing/thawing cycles, respectively. After this process, the samples exhibited decomposition peaks labeled as (1) the loss of the base (A, T, C, or G) (2) the loss of the base and phosphate group, or (3) the loss of the entire oligonucleotide unit and are labeled accordingly. For the sake of clarity, the degradation peaks of the 10-mers were labeled as (*). There was a comparison drawn among the mononucleotide, base, or phosphate loss for characteristic bases, but this will be the focus of future work. (23) Lecchi, P.; Le, H.; Pannell, L. Nucleic Acids Res. 1995, 23, 1276-1277. (24) Gardens, R.; Moroz, T.; Shippy, S.; Sweedler, J. J. Mass Spectrom. 1996, 31, 1126-1127.
Analytical Chemistry, Vol. 72, No. 20, October 15, 2000
5093
Figure 2. Plot of percent intact versus number of freezing/thawing cycles as a robustness measure for the various 5-mer strands after every 10 cycles of freezing in liquid nitrogen and freezing in a 65 °C bath.
Initially each sample was prepared and analyzed directly after receipt from the vendor and prior to the freeze/thaw cycling (cycle ) 0). This is an important step in the procedure because it determined the threshold energy of the laser irradiance and also established the absence of failure sequences as an undesirable residual product of the oligonucleotide synthesis. The comparison of the degradation of all 5-mers of different base compositions, 5′ TTT TT 3′, 5′ AAA AA 3′, 5′ GGG GG 3′, and 5′ CCC CC 3′, is illustrated in Figure 2 which depicts the plot of percentage of strand intact versus freezing/thawing cycles. The percentage of strand intact after various cycling periods was calculated according to eq 1
∑(molecular +
% intact ) (molecular ion)n cycles/
decomposition ions)n cycles × 100 (1)
This value serves as a robustness measurement for the various 5-mer strands. It is important to recognize the above ratio is a relative measurement since previous researchers have proven an influence of base composition on mass spectrometric detection as a result of ionization efficiency.24 Note that the overall trend established from these data was that the robustness of T-mer > A-mer > C-mer > G-mer. The observed trend indicates that the adenosine (A) sections of the oligo strands are more susceptible to decomposition than the thymidine (T) sections of a strand. In general, the breakdown of the A-mer is evident after only twenty freeze/thaw cycles. It is interesting to note the immediate decomposition of the A-mer during the first 20 cycles followed by very little degradation during the following 50 cycles. This observation indicates a trend for the decomposition to occur during an early phase of the freezing/ thawing cycling. In contrast, the T sections are not susceptible to decomposition, even after 70 cycles. We speculate that this observation is attributed to the interaction of each of these strands as it undergoes isolation from the aqueous solvent system. This relationship follows the tendency 5094 Analytical Chemistry, Vol. 72, No. 20, October 15, 2000
Figure 3. MALDI mass spectra of the strand 5′ AAA AA 3′. (a) 50 µM and (b) 1 µM of the strand 5′ AAA AA 3′ after undergoing 20 freezing/thawing cycles.
for the various available sites in the strand to undergo hydrogen bonding characteristic of helical duplexation, intermolecular or intramolecular bonding. More specifically, the adenine-thymidine base pair experiences hydrogen bonding at two sites, whereas the guanine-cytosine base pair undergoes bonding at three sites. The larger the number of hydrogen bonds sites a sample contains the greater the decomposition it experiences. The strands composed of G and C nucleotides have more hydrogen bonds and will therefore be more likely to lose the sugar moiety than A or T regions. It is interesting to note that this trend does not resemble the independently established trend of ionization efficiency for nucleic acid bases as they undergo mass spectrometric desorption/ ionization. Typically, the loss of the nucleic acid base is the primary fragmentation reaction that occurs during the ionization process. During this process it has been suggested that the N-glycosyl bond is weaker for the purine nucleotides than for the pyrimidine nucleotides, resulting in an overall trend of stability to be (A, G > C . T) This ionization/desorption trend is contrary to our results and allows us to conclude that our observations are strictly dependent on the solution chemistry and not a reflection of the desorption/ionization process. The ruggedness of bases under freezing/thawing conditions (T-mer > A-mer > C-mer > G-mer), however, shows some similarities to the trend of bases to withstand cleavage from the backbone in an acidic medium due to hydrolysis of the N-glycosyl linkage (T . C > G, A). Since our medium was neutral, this was not a detrimental consideration. To address the issue of which trend is more detrimental to an oligonucleotide strand (freezing/ thawing versus acidic medium), future studies could be conducted in a more acidic medium and analyzed. Influence of Solution Concentration on Decomposition of Short Strands. The following short strands of oligos 5′ TTT TT 3′, 5′ AAA AA 3′, 5′ CCC CC 3′, and 5′ GGG GG 3′ were studied to determine the concentration dependency on degradation by preparing 50, 25, and 1 µM solutions followed by analysis after 0, 20, and 50 cycles. Figure 3 shows the decomposition of the 5′ AAA AA 3′ as (a) 50 µM solution and (b) 1 µM solution after 20 harsh cycles. The general trend demonstrates that, the lower the oligomer concentration, the more susceptible a short strand is to
Figure 4. MALDI mass spectra of the strand of (a) 50 µM and (b) 1 µM solution of 5′GGG GGA AAA A 3′ analyzed after 20 cycles of harsh freezing and thawing.
decomposition. For example, after 70 cycles, the 50 µM A-mer is 46% intact, whereas the 1 µM sample is only 23% intact. (Spectra are not shown.) As already illustrated in Figure 2, the 50 µM C-mer solution exhibits a different trend from the A-mer and T-mer, with a 21% intact value after 50 cycles and a 19% intact value after 70 cycles. The lower concentrations of C-mer degraded even more quickly, with only 19% intact after only 20 cycles for the 1 µM solution. Unfortunately, the concentrations of G-mer solutions lower than 50 µM were not detectable by MALDI analysis. The T-mer was the least influenced of the four strands by concentration factors. After 70 cycles, the 50 µM solution barely degraded (98% intact) and the 1 µM solution showed 78% of the strand intact. The results from this section of the study lead to the speculation that the analyte is undergoing a separation and isolation from the remainder of the solvent system while freezing. They also support the fact that degradation is caused during this isolation process. As the analyte begins to isolate together, the more concentrated (50 µM) solution has a larger component of the oligonucleotide aggregating together than the less concentrated (1 µM) solution. Therefore, the detrimental influence of the solvent system is smaller for the 50 µM sample than for the 1 µM sample. Since the lower concentration (1 µM) has a larger solvent/analyte ratio, the harmful effects of the solvent system are more pronounced at the lower concentration. This information is valuable to researchers who store oligonucleotide strands as standards: short strands are best stored at high concentrations. Figure 4 illustrates the effect of repetitive freezing/thawing on larger mixed-base strands at various concentrations. This figure shows the sample 5′ GGG GGA AAA A 3′ as it quickly degrades to less than 4% after 20 cycles with the 50 µM sample (Figure 4a). Fifty cycles of freezing and thawing are required for similar results on the 1 µM sample. Although this tendency differs from the conclusions drawn from Figure 3, the observation is attributed to the mixed-base composition and increased length of the strand. The more complex (longer) structure causes steric hindrance among the strands themselves. These results mirror the detrimental effects of steric hindrance (high negative charge) that have been observed during the production of glass-bound oligonucleotide arrays.26,27 In our system, this prevents the efficient aggregation out of the solvent system, thereby allowing the detrimental effects of the solvent system to prevail.
Figure 5. MALDI mass spectra of (a) 5′ TTT TT 3′ and (b) 5′ TTT TTT TTT 3′ after 90 cycles of freezing/thawing conditions. Table 1. Robustness Measurements of the T-mer and A-mer as a Function of Various Strand Lengthssthe 5-mer versus the 12-mer percentage of intact oligonucleotides(%) strand length 5′ TTT TT 3′ 5′ TTT TTT TTT TTT 3′ 5′ AAA AA 3′ 5′ AAA AAA AAA AAA 3′
20 cycles 100 67 ((0.7) 90 ((0.4) 53 ((8.6)
50 cycles 98 ((0.3) 55 ((0.9) 66 ((2.5) 50 ((8.2)
70 cycles 98 ((0.2) 52 ((0.8) 56 ((5.6) 48 ((10)
Influence of Strand Length on Decomposition. The analysis for this section was performed on the strands 5′TTT TT 3′, 5′ TTT TTT TTT T 3′, 5′ AAA AA 3′, and 5′ AAA AAA AAA A 3′ at sample concentrations of 50, 25, and 1 µM after 0, 20, 50, and 70 cycles. First we noted that the earlier observed trend of lower concentration yielding more decomposition for the short strands was also observed for the longer strands in this part of the study. Our second observation stems from the spectra displayed in Figure 5. This demonstrates an influence of strand length on the decomposition of a strictly thymidine region of an oligonucleotide strand. Table 1 provides the experimental calculations of percent intact for the T-mers as well as the A-mers that were analyzed at a 50 µM concentration. The overall trend that is revealed from these studies follows the basis that as oligonucleotide length increases the likelihood of its degradation during freezing/thawing also increases. To illustrate this point, Figure 5a shows that after 90 cycles, the 5′ TTT TT 3′ experienced little degradation (only 2%) unlike the 5′ TTT TTT TTT T 3′ of the same concentration that fragmented to 40% of its original value. The results for the A-mer were not surprising since they followed the same trend as those of the T-mer. Note that for the polyadenosine sample the longer strand was more influenced by the repetitive freezing, especially during the early stages of freezing and thawing (53% intact after 20 cycles for the 5-mer versus 90% intact after 20 cycles for the 12-mer). For both samples this degradation appears to be an initial event; the predominate effects of which are detected within the first 20 cycles. (25) Bentzley, C. M.; Johnston, M. V.; Larsen, B. S. Anal. Biochem. 1998, 258, 31-37. (26) Shchepinov, M.; Case-Green, S.; Southern, E. Nucleic Acids Res. 1997, 25, 1155-1161. (27) Maskos, U.; Southern, E. Nucleic Acids Res. 1992, 20, 1675-1678.
Analytical Chemistry, Vol. 72, No. 20, October 15, 2000
5095
Figure 6. MALDI mass spectra of 50 µM of the mixed strands (a) 5′ GG GGG AAA AA 3′, (b) 5′ GG GGG CCC CC 3′ and c) 5′ TT TTT CCC CC 3′ after 50 cycles of decomposition.
All of these observation are attributed to the increased number of hydrogen-bonding interactions that occur for a longer strand and supports the conclusions drawn from Figure 4. In analyzing these results it becomes apparent that the length is equally as important as base composition in determining the robustness of a sample. Influence of Mixed Bases on Decomposition. All of the 10mers presented in Figure 6 consist of longer strands of mixed bases (50 µM) that have undergone 50 freezing/thawing cycles. From Figure 6c we notice that the TC-mer did not decomposed. The AT-mer and AC-mer also showed no signs of decomposition after 50 cycles (spectra not shown). Unexpectedly, the GA-mer degraded more than the GC-mer, which violated the trend established from the base composition studies. This observation, however, is explained by the fact that the GC-mer is selfcomplementary. Its ability to form a loop-type structural conformation identifies the GC-mer as a more stable structure and therefore makes it less inclined to undergo cycling decomposition.26 Degradation peaks from lower concentations of the strands were not discernible in the analysis. Comparison of Unusual Freezing/Thawing to Routine Handling. Most synthetic samples are not exposed to the unusual freezing/thawing conditions (liquid N2/65 °C) that our samples experienced. Therefore, a comparative study of freezing/thawing rates was performed to compare our results with average conditions (freezer/countertop) and the results of this are shown in Figure 7. In comparison, the laboratory bench/freezer approach yielded results similar to our liquid nitrogen method. The results of leaving the sample on the benchtop and returning it to the
5096 Analytical Chemistry, Vol. 72, No. 20, October 15, 2000
Figure 7. MALDI mass spectra of the 5’ AAA AA 3′ after freezing/ thawing over an 8-h period on the laboratory benchtop and in the freezer.
freezer 8 times over a 40-h period (2.5 h in the freezer; 2.5 h on the benchtop) yielded a 55% intact strand. By referring to Figure 2, a similar result would be obtained at approximately 50 cycles under the liquid nitrogen conditions. Therefore, the slower freezing and thawing performed in a typical laboratory setting is more harmful to the oligonucleotide samples than our simulated conditions. CONCLUSIONS The robustness of several oligonucleotide strands was analyzed during simulated freezing/thawing processes. The results indicated that the characteristic robustness of a strand was dependent on base composition (T-mers > A-mers > C-mers >G-mers) as well as length (5-mers were more sturdy than 10-mers for the T-mer strands). Concentration was another consideration, which was illustrated to be influential since a 50 µM concentration of the 5′ AAA AA 3′ solution illustrated less decomposition than a 1 µM solution. Experiments were also performed to compare typical laboratory conditions to our unusual freezing/thawing conditions. ACKNOWLEDGMENT This research was supported by a grant from the Office of the Vice-President for Academic Affairs at the University of the Sciences in Philadelphia. We also thank Dr. Michael Nedved and Dr. Mark Plucinsky of Johnson & Johnson for their generous assistance and instrument time. Received for review February 24, 2000. Accepted August 7, 2000. AC000225S
