Anal. Chem. 1997, 69, 3966-3972
Analysis of Short Tandem Repeat Polymorphisms in Human DNA by Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry Philip L. Ross* and Phillip Belgrader
Advanced DNA Technology Development Branch, Center for Medical and Molecular Genetics, Armed Forces Institute of Pathology, Rockville, Maryland 20850
The analysis of an important class of human genetic polymorphisms, short tandem repeats (STRs), using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOFMS) is described. Several model STR systems have been investigated to evaluate MALDI-TOFMS as a realistic alternative to established electrophoresis procedures, and to develop rapid and generally applicable approaches to polymerase chain reaction (PCR) product purification for MALDI-TOFMS analysis. A purification/preconcentration method for PCR product preparation based on affinity capture of biotin-labeled PCR products is demonstrated to be directly compatible with MALDI-TOFMS analysis. The entire sample preparation for MALDI-TOFMS analysis immediately following PCR amplification from human DNA extracts can be accomplished routinely in under 12 min in a single Eppendorf tube. The simplicity of this approach essentially eliminates the sample preparation bottleneck encountered with MALDI-TOFMS analysis of PCR products using existing methods. Using this method, encouraging genotyping results are demonstrated for the THO1 and TPOxx STR systems using subpicomole quantities that represent a fraction of the original dsDNA from a single PCR reaction. The technique is also demonstrated to facilitate rapid sizing of PCR fragments larger than 200 bases using MALDI-TOFMS. As described here, the analysis of DNA can be accomplished in a manner that takes advantage of the rapid and accurate analysis capabilities offered by MALDI-TOFMS. The application of the polymerase chain reaction (PCR)1 to amplification of loci containing variable number of tandem repeat (VNTR) polymorphisms has found widespread use in forensics, paternity testing, cell line identification, and characterization of disease genes.2-4 VNTRs generally consist of tandemly repeating sequence units which vary in length from 2 to 70 bases, and they vary widely between individuals in the number of sequence units. The more traditional genotyping approach, utilizing restriction
fragment length polymorphism (RFLP), is not based on PCR and, as such, requires larger quantities of intact DNA to ultimately permit accurate analysis. Coupled with the advent of PCR to generate exponential copy numbers at a given locus, the development of VNTR typing has replaced RFLP analysis, particularly in situations where minimal sample quantities are available.5 Short tandem repeat polymorphisms (STRs or microsatellites)6-16 are VNTR loci which contain repeating units of 2-7 bases that range in number from 4 to 20 (approximately) between different individuals. Genotyping using STRs has a number of practical advantages over longer VNTR loci. The shorter product size of STRs enables a greater amplification yield and, unlike longer VNTRs, permits more equal amplification of alleles with varying sizes. Additionally, the short STR amplicons are easily adaptable to development of nonoverlapping coamplification systems,11 thereby permitting genotyping of multiple loci from highly degraded or trace quantities of DNA. It has been estimated that the human genome may contain one trinucleotide or tetranucleotide repeat in every 15 kb, with up to half of the loci investigated found to be highly polymorphic.6,7,12 Dinucleotide repeats are also quite common; however, these loci are prone to amplification errors which significantly compromise accurate and reliable allele designation.11-16 Trinucleotide and tetranucleotide repeats are, therefore, more analytically informative since fewer amplification artifacts are observed. As a result, a variety of trinucleotide and tetranucleotide STR systems have become validated for routine use in forensic identity, parentage determination, and cell line authentication applications.11,13,16 Currently, analysis of STR products is accomplished using denaturing polyacrylamide gel electrophoresis with either silver staining detection or incorporation of primers with fluorescent labels for fluorescence detection. Although such methods are
* To whom correspondence should be addresed. E-mail: rossp@ email.afip.osd.mil. (1) Saiki, R. K.; Gelfand, D. H.; Stoffel, S.; Scharf, S. J.; Higuchi, R.; Horn, G. T.; Mullis, K. B. Science 1988, 239, 487-491. (2) Jeffreys, A. J.; Wilson, V.; Thein, S. L. Nature 1985, 316, 76-79. (3) Nakamura, Y.; Leppert, M.; O’Connel, P.; Wolff, R.; Holm, T.; Culver, M.; Martin, C.; Fujimoto, E.; Hoff, M.; Kumlin, E.; White, R. Science 1987, 235, 1616-1622. (4) Budowle, B.; Chakraborty, R.; Giusti, A. M.; Eisenberg, A. J.; Allen, R. C. Am. J. Hum. Genet. 1991, 48, 137-144.
(5) Reynolds, R.; Sensabaugh, G. Anal. Chem. 1991, 63, 2-15. (6) Edwards, A.; Civitello, A.; Hammond, H. A.; Caskey, C. T. Am. J. Hum. Genet. 1991, 49, 746-756. (7) Edwards, A.; Hammond, H. A.; Jin, L.; Caskey, C. T.; Chakraborty, R. Genomics 1992, 12, 241-253. (8) Polymeropoulos, M. H.; Rath, D. S.; Xiao, H.; Merril, C. R. Nucleic Acids Res. 1991, 19, 4018. (9) Polymeropoulos, M. H.; Xiao, H.; Rath, D. S.; Merril, C. R. Nucleic Acids Res. 1991, 19, 3753. (10) Bikker, H.; Baas, F.; de Vijlder, J. J. M. Hum. Mol. Genet. 1992, 1, 137. (11) Lins, A. M.; Sprecher, C. J.; Puers, C.; Schumm, J. W. Biotechniques 1996, 20, 882-889. (12) Beckman, J. S.; Weber, J. L. Genomics 1992, 12, 627-631. (13) Sprecher, C. J.; Puers, C.; Lins, A. M.; Schumm, J. W. Biotechniques 1996, 20, 266-276. (14) Schlotterer, C.; Tautz, D. Nucleic Acids Res. 1992, 20, 211-215. (15) Levinson, G.; Gutman, G. A. Mol. Biol. Evol. 1987, 4, 203-221. (16) Hammond, H. A.; Jin, L.; Zhong, Y.; Caskey, C. T.; Chakraborty, R. Am. J. Hum. Genet. 1994, 55, 175-189.
3966 Analytical Chemistry, Vol. 69, No. 19, October 1, 1997
S0003-2700(97)00312-0 CCC: $14.00
© 1997 American Chemical Society
performed with reasonable accuracy on a routine basis, the polyacrylamide gels require careful preparation and are typically run for several hours to achieve resolution greater than 4 bp. Additionally, the preparation and running of gels is a time- and labor-intensive process and consumes large quantities of reagents, some of which represent severe health hazards. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOFMS)17 has recently evolved into a technique which can routinely obtain accurate molecular weight information on purified and/or synthetic oligonucleotides up to 50 bases in length.18-21 MALDI-TOFMS offers the advantages of short analysis times and the ability to obtain unambiguous molecular weight information without the need for allelic ladders as size standards. Finally, current capabilities exist for highthroughput MALDI-TOFMS analysis since a large number of small dried spots of material can be placed on a single sample probe. Therefore, MALDI-TOFMS has the potential to replace gel-based technology for analysis of length polymorphisms such as STRs. In practice, however, the advantageous features of MALDITOFMS have not been demonstrated as a consequence of fragmentation and purification issues pertaining to succesful desorption and detection of intact DNA strands. The development of picolinic acid derivatives22,23 for MALDI matrix preparations has alleviated oligomer fragmentation problems.24,25 However, purification adequate for succesful detection of PCR products has required tedious and lengthy procedures.26-32 Such approaches in general have relied on ethanol precipitation26-30 and/or commercial purification kits15,31,32 to purify and concentrate DNA fragments. Since these procedures involve multiple centrifugation and lyophilization/evaporation steps requiring large initial quantities of PCR product, they do not legitimately represent rapid or efficient modes of analysis. Ethanol precipitation in particular can promote considerable sample losses since it relies on solubility equilibria to isolate and wash the DNA. Furthermore, previous reports of detection of PCR products by MALDI-TOFMS have not demonstrated resolution sufficient to routinely resolve fragments (17) Hillenkamp, F.; Karas, M.; Beavis, R. C.; Chait, B. T. Anal. Chem. 1991, 63, 1193A-1203A. (18) Pieles, U.; Zu ¨ rcher, W.; Scha¨r, M.; Moser, H. E. Nucleic Acids Res. 1993, 21, 3191-3196. (19) Jurinke, C.; van den Boom, D.; Jacob, A.; Tang, K.; Wo¨rl, R.; Ko ¨ster, H. Anal. Biochem. 1996, 237, 174-181. (20) Juhasz, P.; Roskey, M. T.; Smirnov, I. P.; Haff, L. A.; Vestal, M. L.; Martin, S. A. Anal. Chem. 1996, 68, 941-946. (21) Bentzley, C. M.; Johnston, M. V.; Larsen, B. S.; Gutteridge, S. Anal. Chem. 1996, 68, 2141-2146. (22) Shaler, T. A.; Yan, Y.; Wickham, J. N.; Wu, K. J.; Becker, C. H. Rapid Comm Mass Spectrom. 1995, 9, 942. (23) Wu, K. J.; Shaler, T. A.; Becker, C. H. Anal. Chem. 1994, 66, 1637-1645. (24) Tang, W.; Zhu, L.; Smith, L. M. Anal. Chem. 1997, 69, 302-312. (25) Zhu, L.; Parr, G. P.; Fitzgerald, M. C.; Nelson, C. M.; Smith, L. M. J. Am. Chem. Soc. 1995, 117, 6048-6056. (26) Cha´ng, L.-Y.; Tang, K.; Schell, M.; Ringelberg, C.; Matteson, K. J.; Allman, S. L.; Chen, C. H. Rapid Commun. Mass Spectrom. 1995, 9, 772-774. (27) Hurst, G. B.; Doktycz, M. J.; Vass, A. A.; Buchanan, M. V. Rapid Commun. Mass Spectrom. 1996, 10, 377-382. (28) Tang, K.; Allman, A. L.; Chen, C. H.; Cha´ng, L. Y.; Schell, M. Rapid Commun. Mass Spectrom. 1994, 8, 183-186. (29) Tang, K.; Taranenko, N. I.; Allman, S. L.; Cha´ng, L. Y.; Chen, C. H. Rapid Commun. Mass Spectrom. 1994, 8, 727-730. (30) Doktycz, M. J.; Hurst, G. B.; Habibi-Goudarzi, S.; McLuckey, S. A.; Tang, K.; Chen, C. H.; Uziel, M.; Jacobson, K. B.; Woychik, R. P.; Buchanan, M. V. Anal. Biochem. 1995, 230, 205-214. (31) Liu, Y.-H.; Bai, J.; Liang, X.; Lubman, D. M.; Venta, P. J. Anal. Chem. 1995, 67, 3482-3490. (32) Liu, Y. H.; Bai, J.; Zhu, Y.; Liang, X.; Semeniak, D.; Venta, P. J.; Lubman, D. M. Rapid Commun. Mass Spectrom. 1995, 9, 735-743.
4 bases apart in the >60 bp size range, as is required for analysis of STR loci. These considerations have thus rendered MALDITOFMS unsuitable for DNA size analysis on a routine basis. Clearly, for MALDI-TOFMS to become acceptable as a practical tool for rapid genetic analysis, new methodology enabling direct PCR product sizing must be explored. Such a method must permit sample purification following amplification of real DNA extracts, preferably in a single-tube format. In this report, we present initial findings from application of MALDI-TOFMS to STR analysis using representative loci that are currently employed extensively in forensic and human identity applications.11,13,16 During this work, a truly rapid, single-tube approach to purification of PCR products involving immobilization on magnetic beads is utilized. The use of related immobilization strategies for sample preparation prior to MALDI-TOFMS analysis of peptides33 and synthetic oligonucleotide sequence ladders has been reported.34-36 DNA fragment immobilization in this fashion permits extensive washing of the captured molecules. However, solid phase immobilization has not been demonstrated to facilitate direct mass spectrometric analysis of PCR products, particularly in an entirely self-contained format. We demonstrate that solidphase immobilization can be readily implemented for direct PCR product sizing of real DNA extracts using MALDI-TOFMS. The method described here is an adaptation of approaches used in solid phase sequencing of DNA,37,38 whereby sense and antisense strands are separated after immobilization. The approach makes use of biotin-labeled PCR primers designed to immediately flank the polymorphic region of interest. After a single stage of amplification, the double-stranded, biotinylated PCR product is then captured onto streptavidin-coated magnetic particles, which facilitate extensive washing of the DNA. The product is then denatured, and an aliquot containing free ssDNA is directly analyzed by MALDI-TOFMS. It is demonstrated that the entire procedure can be performed directly in the PCR reaction tube in under 12 min and is well suited to rapid STR analysis. Accurate genotyping information can be routinely obtained from several hundred femtomoles of single-stranded DNA. We also show that the technique can be of general utility for rapidly obtaining size information of PCR products larger than 250 bases through MALDI-TOFMS analysis. EXPERIMENTAL SECTION PCR Amplification of STR Loci. Primers were synthesized for several STR loci, as summarized in Table 1, on an Applied Biosytems 392 DNA synthesizer (Perkin-Elmer, Applied Biosystems Division, Foster City, CA). The primers were designed to flank the repeat region as closely as possible while still maintaining reasonable annealing/melting characteristics. In all cases, the primers were used without further purification. Initial experiments focused on amplification of the trinucleotide repeat locus COL1A, which is an ACT repeat located inside intron (33) Girault, S.; Chassaing, G.; Blais, J. C.; Brunot, A.; Bolbach, G. Anal. Chem. 1996, 68, 2122-2126. (34) Tang, K.; Fu, D.; Kotter, S.; Cotter, R. J.; Cantor, C. R.; Ko ¨ster, H. Nucleic Acids Res. 1995, 23, 3126-3131. (35) Chou, C.-W.; Bingham, S. E.; Williams, P. Rapid Commun. Mass Spectrom. 1996, 10, 1410-1414. (36) Ko ¨ster, H.; Tang, K.; Fu, D.-J.; Braun, A.; van den Boom, D.; Smith, C. L.; Cotter, R. J.; Cantor, C. R. Nature Biotechnol. 1996, 14, 1123-1128. (37) Hultman, T.; Stahl, S.; Hornes, E.; Uhlen, M. Nucleic Acids Res. 1989, 17, 4937-4946. (38) Tong, X.; Smith, L. M. DNA Sequence 1993, 4, 151-162.
Analytical Chemistry, Vol. 69, No. 19, October 1, 1997
3967
Table 1. Sequence and Molecular Weight Data for STR Loci Observed in the Present Study locus
allele sizesa
COL1A
6, 7, 8, 9, 10 (76-92 bp)
THO1
6, 7, 8, 9, 9.3, 10 (67-83 bp)
TPOxx
8, 9, 10, 11 (80-92 bp)
a
strand
primer sequenceb
repeat (MW)
MW range
forward reverse forward reverse forward reverse
5′-ACATGTGCCATAGTATTAAATCCC-3′ 5′-CTATATTCTTATCCTGAGTAAAAACCA-3′ 5′-CCTGTTCCTCCCTTATTTCCC-3′ 5′-B-GGGAACACAGACTCCATGGTG-3′ 5′-GGCACTTAGGGAACCCTCAC-3′ 5′-B-TCCTTGTCAGCGTTTATTTGC-3′
ACT (906.6) TGA (946.6) CATT (1210.8)
22331-25051 22642-25482 20215-25059
AATG (1259.8)
24914-28693
Listed as number of repeat units and product sizes observed in the current study only. b B indicates biotin label.
12 of the human collagen R-2 type I gene, one of two genes implicated in dominant osteogenesis imperfecta.39 The expected molecular weights of single strands of the corresponding PCR products are included in Table 1. Since molecular ions of single strands are observed exclusively in MALDI-TOFMS, the expected molecular weights of both strands are given. A biotinylated forward primer was also synthesized to enable direct comparison between conventional and capture-based purification. Amplification of COL1A samples was performed using 1X PCR buffer II (Perkin Elmer, Norwalk, CT) 2.5 mM MgCl2, 0.2 mM deoxynucleotide triphosphates, 1.0 µM each primer, 1.25 units of Taq DNA Polymerase (Perkin Elmer), and 20-50 ng of template DNA, in 50 µL reaction volumes. DNA was isolated from whole blood using the Puregene DNA isolation kit (Gentra Systems, Minneapolis, MN). PCR was performed using a Perkin-Elmer 9600 thermal cycler under the following conditions: 95 °C denature for 30 s, 32 cycles of 95 °C for 30 s, 63 °C for 30 s, 72 °C for 30 s, followed by a 10 min soak at 73 °C. For development of the biotin-streptavidin capture technique, the tetranucleotide repeats located in the human tyrosine hydroxylase gene (THO1, AATG repeat)9 and thyroid peroxidase gene (TPOxx, TGAA repeat)10 were used. These STR loci are among several used widely for identity testing and other genetic profiling applications.11,13,16 In the present experiment, primers were designed to immediately flank the repeat region (Table 1), whereas commercial THO1 and TPOxx systems have allele sizes of 179-203 and 232-248 bp, respectively using optimized primer sets. Biotin label on the 3′ end was incorporated using BiotinON phosphoramidite (Clontech Labs, Palo Alto, CA). The yield of biotinylation was estimated by MALDI-TOFMS analysis to be ≈60%, assuming approximately equal sensitivity of labeled and unlabeled oligonucleotide. Amplification of THO1 and TPOxx products was performed in some cases using the reagent conditions as described above. In initial experiments, the PCR buffer used with Taq polymerase was STR 10× buffer (Promega Corp., Madison, WI). Alternatively, Pfu DNA polymerase and Pfu 10× buffer (Strategene, La Jolla, CA) were used according to the manufacturer’s instructions. Thermal cycling conditions were as follows: 96 °C denature for 1 min, followed by 10 cycles of 94 °C for 30 s, 60 °C for 30 s, 72 °C for 45 s, followed by 20 cycles of 90 °C for 30 s, 60 °C for 30 s, 72 °C for 30 s, followed by a 7 min soak at 72 °C. Purification of PCR Products. The various conventional strategies, namely ethanol precipitation, size-exclusion microfiltration, and membrane dialysis, used for preconcentration and/ or purification of the STR amplicons have been described in some detail previously.26-32 Ethanol precipitation in the presence of (39) Pepe, G. Hum. Mutat. 1993, 2, 300-305.
3968 Analytical Chemistry, Vol. 69, No. 19, October 1, 1997
Figure 1. Schematic outline of sample preparation protocol followed for purification and MALDI-TOFMS sizing of biotin-labeled, PCRamplified STR products.
concentrated ammonium acetate was performed on selected reamplified COL1A samples. Size-exclusion cartridges (Microcon 30, Microcon 50, Amicon Inc., Beverly MA) were used according to the manufacturer’s instructions, with the addition of one or more further wash steps to remove PCR components prior to spinelution of the DNA. Membrane dialysis40 was performed by floating a 0.025 µm membrane filter (Millipore Inc., Bedford, MA) onto a small volume of water, onto which was placed 5 µL of PCR product and 0.5 µL of acetonitrile. These approaches yielded very poor data quality and required nested PCR in order to observe any mass spectral response; therefore, these were ultimately abandoned in favor of PCR product purification using streptavidincoated magnetic particles. The procedure involved in capturing PCR products onto streptavidin-coated magnetic particles is outlined schematically in Figure 1. Magnetic particles (100 µL of suspension as supplied, PerSeptive Biosystems, Framingham, MA) were washed twice and resuspended with 0.5 M NaCl/0.005 M EDTA solution. Equal volumes of PCR product and magnetic particle suspension were then combined and incubated at room temperature. In all cases, PCR product from single 50 µL reactions following a single-stage PCR reaction were used for affinity capture-based approaches. The supernatant was removed following magnetic separation, and the captured PCR product was washed with distilled deionized water (40) Go ¨risch, H. Anal. Biochem. 1988, 173, 393.
and/or 0.1 M ammonium acetate. The double-stranded DNA was then denatured either by treatment with 0.1 N NaOH or by incubation at 90 °C. Following NaOH denaturation, the resultant solution containing free strands was desalted by membrane dialysis as described above, and an aliquot was taken for analysis. In the case of heat denaturation, which was used for all of the affinity capture results shown here, the resulting free strands were directly analyzed by MALDI-TOFMS. Typically, the DNA-bead complex was resuspended in 6 µL of H2O and was then incubated at 90 °C for periods ranging from 30 to 120 s. After magnetic separation, a 1.2 µL aliquot was then taken for MALDI-TOFMS analysis. Heat denaturation offered the advantage of simplicity and substantial time savings with no apparent change in data quality and was, therefore, the method of choice for further application to STR typing. No useful signals were observed from direct MALDI analysis of magnetic particles containing immobilized double-stranded PCR products. Succesful PCR product purification was observed for all enzyme/buffer combinations used. MALDI-TOF Mass Spectrometry. Mass spectrometry was performed on a Voyager RP (PerSeptive Biosystems) workstation operating in linear, static acceleration mode utilizing a nitrogen laser (337 nm). Negative ion mode was used for all results shown here and, in general, was observed to give slightly superior performance to positive ion mode. The matrix solution was prepared as a 4:1 molar ratio of 3-hydroxypicolinic acid/picolinic acid in a solution containing 0.07 M ammonium citrate and 30% acetonitrile. Spectra were typically acquired by averaging 30-70 laser shots. The molecular ion (M - H)- and dimer ion (M2 H)- of horse cytochrome c (Sigma Chemical Co., St. Louis, MO) were used for calibration. RESULTS AND DISCUSSION Conventional Purification Approaches. As described in the previous section, ethanol precipitation, size-exclusion purification, and membrane desalting were explored in preliminary investigations. In general, it was observed that a second stage of amplification (nested PCR) was required in order to achieve reasonable spectral quality and reproducibility using such approaches. It was also observed that size-exclusion purification incurred significant loss of amplified DNA, as indicated by the presence of detectable quantities of DNA following several elution steps. Size-exclusion purification also was not effective in complete removal of primers and primer dimers from PCR reaction mixtures. Apart from these constraints, the detection of double-stranded DNA also limits the utility of such approaches. Since only ions corresponding to single DNA strands are produced in MALDI, the resultant molecular weight profiles are composites of the forward and reverse strands of each individual STR allele. There is a substantial molecular weight variation between the strands, as illustrated in Table 1, thereby introducing an element of uncertainty into allele designation. Comparison of representative double-stranded and single-stranded product from the COL1A locus (homozygous 10, 10) is shown in Figure 2. With the doublestranded product, a peak molecular weight value is obtained which actually does not correspond to the molecular weight average of individual forward and reverse strands. Differing desorption efficiency between the two strands may arise from differences in base composition, the result being a molecular weight envelope with a peak value closer to that of the more intense component.
Figure 2. Typical mass spectra obtained for a homozygous 10, 10 individual at COL1A locus. (a) Purification of DNA fragment following reamplification using Microcon 30 spin filtration device. PCR is performed using Taq polymerase with unlabeled primers. Molecular weight assignment is based on peak value. (b) Purification of DNA following single amplification using affinity capture with heat denaturation. PCR amplification is performed with Taq polymerase and a biotinylated reverse primer. Molecular weight is assigned by data analysis software.
Reliable molecular weight assignment is difficult to perform with the poor data quality represented by double-stranded PCR product analysis. The degree of uncertainty surrounding information gathered from dsDNA measurements, therefore, ultimately limits the applicability of this method for genuine unknown samples. Of particular importance is the fact that such measurements may not accurately determine the addition or deletion of a single base. Affinity Capture Purification. The use of streptavidin-coated magnetic particles has been investigated in various capacities as a means of preparing biotinylated peptides and synthetic oligonucleotides for mass spectrometric analysis.33-36 The streptavidin-biotin interaction is essentially irreversible, thereby permitting various manipulations of the captured molecule. The approach has, nevertheless, remained unexplored as a means of performing rapid and direct sample preparation of amplified DNA (from real DNA extracts) for mass spectrometric analysis. The aim here is to utilize solid phase immobilization as a single purification step, such that additional sample purification steps and/or nested PCR are avoided. Figure 2, which compares analysis of COL1A PCR product using size-exclusion purification (a) and affinity capture purification (b) helps to illustrate the advantages brought about by solid phase immobilization of PCR products. Apart from eliminating the need for nested PCR, affinity capture/denaturation yields single-stranded DNA, which circumvents the molecular weight ambiguity of dsDNA analysis by MALDI-TOFMS. The result is a distinct molecular weight readout rather than an average of two strands with markedly differing molecular weights. Analysis of single-stranded rather than double-stranded DNA, in addition to complete removal of all other PCR-derived contaminants, contributes to the improved sensitivity observed with the affinity capture method. Analytical Chemistry, Vol. 69, No. 19, October 1, 1997
3969
Figure 4. Comparison of MALDI-TOFMS analysis of THO1 product purified using a 30 min protocol and a 12 min protocol. Product from the same PCR amplification (Pfu polymerase) was used for each.
Figure 3. MALDI-TOFMS analysis of THO1 products from several individuals. Sample in (a) is homozygous (9.3, 9.3) K562 cell line DNA used as positive control DNA in commercial STR systems (Promega, Madison WI). All samples were amplified using Pfu DNA polymerase and were purified by affinity capture, followed by 90 °C denaturation for 90 s.
Rapid, Single-Tube PCR Product Purification. Initial application of the affinity capture approach utilized chemical denaturation (0.1 N NaOH) followed by membrane dialysis to isolate ssDNA for analysis. Although analysis was performed succesfully in this manner, the approach requires an undesirable transfer of material to a dialysis filter. By the simple implementation of a brief heat denaturation stage rather than NaOH denaturing, it was found that reliable MALDI-TOFMS analysis was possible with no sacrifice in resolution or sensitivity. Heat denaturation, however, eliminates the longer incubation period and dialysis step necessitated by chemical denaturation. In this way, the entire purification was accomplished in under 30 min and can be performed directly in the PCR reaction tube. This time is inclusive of all steps, including prewashing of the magnetic particles and final deposition of an aliquot of the sample, with matrix, onto the sample probe. Figure 3 shows genotyping results obtained for several individuals following a single-stage PCR amplification at the THO1 locus. The figure illustrates the high degree of variability encountered at this locus from randomly selected individuals among a normal population. The K562 sample was tested as an additional confirmation of correct genotyping results.41 For the samples shown in Figure 3, the DNA was heat denatured in water for 90 s following capture and washing. An aliquot of the single-stranded THO1 DNA was then combined with matrix and directly analyzed by MALDI-TOFMS. Absorbance measurements (260 nm) indicate single-stranded product con(41) STR-THO1 typing of individuals in this study was performed using Promega (Madison, WI) Geneprint STR systems.
3970 Analytical Chemistry, Vol. 69, No. 19, October 1, 1997
centrations of 2.5-3.0 pmol/µL following denaturing. For the heterozygous samples, this concentration corresponds to ∼600 fmol of ssDNA/spot/allele, depending on allele sizes, assuming a 1:1 matrix/analyte dilution ratio (1.5 µL each) divided between two replicate 1 µL spots. With the affinity capture technique as demonstrated here, the implementation of MALDI-TOFMS for rapid PCR-based DNA size analysis can be realized. Since the peaks are clearly defined, absolute molecular weight assignment and separation between neighboring alleles become more definitive. The need to prepare and run size standards for each analysis is, therefore, eliminated. One important feature of the THO1 locus is the occasional occurence of the 9.3 allele (Figure 3a), which corresponds to deletion of a single base from one of the tetranucleotide repeat units. Based on the absolute molecular weight information obtained, clear mass differences of approximately 300 Da between the 9.3 and 10 alleles are found. It is, therefore, a straightforward matter to distinguish between 9.3 and 10 alleles. Optimization of STR Analysis via Affinity Capture. Efforts aimed at further optimization of the purification protocol, with the eventual goal of minimizing overall time and number of operations, have been very encouraging. By reduction of the incubation period involved with streptavidin-biotin capture, the overall purification procedure can be simplified considerably, as demonstrated in Figure 4. Purification of the heterozygous THO19,10 sample was accomplished by using a 2 min capture incubation period, followed by two 100 µL water washes and a 0.1 M (100 µL) ammonium acetate wash. The overall preparation time, from preliminary washing of the magnetic particles to final deposition of sample with matrix onto the probe surface, was 12 min. These improvements do not appear to have significantly compromised genotyping accuracy, measurement sensitivity, or resolution between neighboring alleles. For certain individuals, low-intensity peaks from analysis of THO1 products were observed which correspond to products 4 bases shorter or larger than the main alleles. These features
Figure 6. MALDI-TOFMS analysis of biotinylated 268 base PCR product following single amplification from a hypervariable region of the human mitochondrial genome. Product was purified by affinity capture with 90 °C heat denaturation for 90 s. Inset: expansion of molecular ion region showing peak molecular weight assignment.
Figure 5. Comparison of amplification/purification of TPOxx STR locus in K562 DNA using Taq DNA polymerase (a) and Pfu DNA polymerase (b and c). Identical voltage conditions were used between (a) and (b). Between (b) and (c), the potential on the first acceleration grid was reduced from 21 250 to 17 500 V, expressed as a percentage of sample stage (backing plate) voltage (25 000 V) in acquisition software. Improved resolution and slightly reduced signal intensity are observed at the lower voltage setting.
correspond to repeat slippage products,42 which is a common occurrence in amplification of dinucleotide repeat loci15,16 and is occasionally observed for some tetranucleotide repeat loci. The addition of an extra nucleotide base was also observed for some DNA samples, particularly at the TPOxx locus, resulting in an increase of ∼300 Da from expected values for some alleles. To address and minimize these complications, amplification using Pfu DNA polymerase was investigated. This enzyme has a demonstrated lower error rate than that of Taq polymerase, and the lower processivity of Pfu promotes higher fidelity, blunt-ended DNA synthesis from templates with susceptibility to form secondary structures. Use of this enzyme appears to alleviate production of lower and higher molecular weight products corresponding to repeat slippage and nontemplated base addition, respectively, as shown in Figure 5. In the case of amplification of TPOxx loci using Taq polymerase (Figure 5a), a shift in the peak maxima of over 300 Da from the expected molecular weight for both alleles is observed. Amplification using Pfu polymerase succesfully eliminated the apparent nontemplated base addition, as well as alleviated problems with formation of repeat slippage products. Complete elimination of this artifact was not achieved for all DNA samples, as is observed in Figure 3b. An apparent increase in sensitivity is also observed on going from Taq to Pfu amplification, which is not necessarily expected, since the proofreading ability of Pfu typically occurs at the expense (42) Walsh, P. S.; Fildes, N. J.; Reynolds, R. Nucleic Acids Res. 1996, 24, 28072812.
of superior amplification yield. With Pfu polymerase, sufficient resolution and sensitivity are observed to facilitate the somewhat challenging task of reliable instrument optimization (Figure 5b,c) for resolution. Upon reducing the first acceleration grid voltage from 21 250 to 17 500 V (expressed as a percentage of the sample stage voltage of 25 000 V), substantially improved resolution is observed. The lowering of the first grid potential improves ion focusing at the sacrifice of sensitivity. Thus, fairly robust and consistent signal at a particular sample spot is necessary to succesfully balance resolution and sensitivity. The ability to perform such an optimization demonstrates that MALDI analysis of PCR products can advance from mere detection of a signal. As described here, the affinity capture approach, combined with MALDI-TOFMS, is simple and rapid enough to rival simple agarose gel electrophoresis to check PCR reaction yields and verify correct amplicon size. As a quick test of this, a sample of amplified, biotin-labeled mitochondrial DNA was analyzed. The primers amplify a highly polymorphic 268 bp product in the control region of the human mitochondrial genome. The mass spectrum in Figure 6 reveals correct size of the amplification product (expected molecular weight ) 81355) within 83 mass units (∼0.1%) using external calibration (BSA, negative ion mode). The spectrum was obtained with a preparation time of 15 min, which is equal to or less than the time involved in preparation and running of agarose gels. It should be noted that the sample spot was fairly homogeneous in this instance, in that it was possible to obtain similar spectra at a number of locations along the sample. Further efforts to investigate the effective upper molecular weight limit attainable were not pursued. The experimental molecular weight range suggested by this result is quite encouraging. Although limited mass accuracy accompanies the poor resolution in this mass range, sufficient accuracy is observed to distinguish between fragments varying in length by 4 bases. This measurement indicates that the potential exists to develop multiplex amplification systems with nonoverlapping tetranucleotide repeat loci extending well beyond 100 bases in length. Efforts to optimize the amplification of STR loci, including development of multiplex Analytical Chemistry, Vol. 69, No. 19, October 1, 1997
3971
analysis capabilities, are currently underway, as are modifications to further simplify and improve upon the sample preparation approach. Apart from use directly for mass spectrometric genotyping, this approach holds promise for further applications. Of particular interest is implementation of the procedure described here for automated sample preparation through the use of 96 well magnetic stations (Dynal Inc., Oslo, Norway). In addition, both the immobilized and remaining free single-stranded DNA can be utilized for confirmatory analysis, for sequencing (for homozygous cases) or hybridization-based experiments. The precision with which molecular weight information can be obtained on the STR products may approach a level whereby issues such as sequence heterogeneity can be assessed.42,43 It has been observed for certain alleles of STR loci that repeat units do not uniformly comprise the same base composition. Such sequence differences impart a change of 9 Da or more. Sufficient molecular weight accuracy to address this heterogeneity is not achieved in the present experiments. However, it has been demonstrated for mixed-base synthetic oligonucleotides that a single base sequence variant (∆m ) 15 Da) can be easily identified using MALDITOFMS with delayed extraction.20 Thus, one future goal in this work, which will require the implementation of delayed extraction, is to achieve sufficient resolution and mass accuracy such that this heterogeneity can be identified as it occurs. For the substantially larger single-stranded PCR products examined here, such a task would additionally require multiple molecular weight measurements from individual samples in order to reach an appropriate confidence level. (43) Puers, C.; Hammond, H. A.; Jin, L.; Caskey, T. C.; Schumm, J. W. Am. J. Hum. Genet. 53, 1993, 953-958.
3972
Analytical Chemistry, Vol. 69, No. 19, October 1, 1997
CONCLUSION We have demonstrated that the analysis of PCR-amplified STR loci is easily performed using MALDI-TOFMS. Evaluation of several conventional methods for sample purification prior to mass spectral analysis reveals severe limitations with respect to time, sample quantity, and data quality. Implementation of a rapid, affinity-based sample purification strategy is shown to be an effective means of obtaining precise molecular weight information in a short period of time. The procedure, as demonstrated here, is unique in that no ancillary reamplification, purification, preconcentration, or desalting steps of any kind are required for succesful analysis. Thus, the entire amplification and preparation can be performed in a single Eppendorf tube. Furthermore, since only simple pipetting, magnetic separation, and heating operations are involved, the procedure is amenable to conventional laboratory automation. Encouraging genotyping results have been obtained using the affinity capture of biotinylated STR products in a procedure that is routinely accomplished in under 12 min. Accurate genotyping results were obtained from approximately 600 fmol of purified product. The ability to quickly obtain molecular weight information on a 268 base PCR product is also demonstrated. ACKNOWLEDGMENT The authors thank Ms. Kristal Allen-Weaver for technical assistance. The views presented here are strictly the opinion of the authors and in no way reflect the position of the U.S. Army, the U.S. Air Force, or the U.S. Department of Defense. Received for review March 20, 1997. Accepted July 2, 1997.X AC970312T X
Abstract published in Advance ACS Abstracts, August 15, 1997.
