- Email: [email protected]
Mass-spectrometry DNA sequencing
Mutation Research 573 (2005) 3–12
Review
Mass-spectrometry DNA sequencing John R. Edwards a, b , Hameer Ruparel a, b , Jingyue Ju a, b, ∗ a
Columbia Genome Center, Columbia University College of Physicians and Surgeons, Room 405A, Russ Berrie Medical Science Pavilion, New York, NY 10032, USA b Department of Chemical Engineering, Columbia University, New York, NY 10027, USA Received 1 July 2004; accepted 15 July 2004 Available online 11 February 2005
Abstract Matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) has been explored widely for DNA sequencing. Compared to gel electrophoresis based sequencing systems, mass spectrometry produces very high resolution of sequencing fragments, rapid separation on microsecond time scales, and completely eliminates compressions associated with gel-based systems. While most of the research efforts have focused on using mass spectrometers to analyze the DNA products from Sanger sequencing or enzymatic digestion reactions, the read lengths attainable are currently insufficient for large-scale de novo sequencing. The advantage of mass-spectrometry sequencing is that one can unambiguously identify frameshift mutations and heterozygous mutations making it an ideal choice for resequencing projects. In these applications, DNA sequencing fragments that are the same length but with different base compositions are generated, which are challenging to consistently distinguish in gel-based sequencing systems. In contrast, MALDI-TOF MS produces mass spectra of these DNA sequencing fragments with nearly digital resolution, allowing accurate determination of the mixed bases. For these reasons mass spectrometry based sequencing has mainly been focused on the detection of frameshift mutations and single nucleotide polymorphisms (SNPs). More recently, assays have been developed to indirectly sequence DNA by first converting it into RNA. These assays take advantage of the increased resolution and detection ability of MALDI-TOF MS for RNA. © 2005 Elsevier B.V. All rights reserved. Keywords: MALDI-TOF mass spectrometry; Single nucleotide polymorphism; DNA sequencing
Contents 1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MALDI-TOF MS analysis of nucleic acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DNA sequencing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
∗ Corresponding author. Tel.: +1 212 851 5172; fax: +1 212 851 5176. E-mail address: [email protected] (J. Ju).
0027-5107/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.mrfmmm.2004.07.021
4 4 6
4
4. 5. 6.
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
Detection of frameshift mutations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 RNA sequencing assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1. Introduction The ability to sequence DNA accurately and rapidly is revolutionizing biology and medicine. This rapid technology development involving engineering, chemistry, biology, and computer science makes it possible to move from studying single genes at a time to analyzing and comparing entire genomes. With the completion of the human genome sequence map [1,2], resequencing polymorphic areas of the genome linked to disease development will contribute greatly to the understanding of the molecular basis of disease and developing new therapeutics. High accuracy is required for these genetic variation studies. The current state-ofthe-art technology for high throughput DNA sequencing is the capillary array DNA sequencer using laser induced fluorescence detection [3–7]. Although this technology addresses the throughput and read length requirements of large scale DNA sequencing projects, the accuracy required for detecting genetic variations needs to be improved for biological research and clinical diagnostics. For example, electrophoresis based sequencing methods have difficulty detecting heterozygotes unambiguously and are not 100% accurate for a given base due to compressions in GC rich regions [8,9]. In addition, the first few bases after the priming site are often masked by the high fluorescence signal from the excess dye-labeled primers or dye-labeled terminators, and are difficult to identify. While other techniques such as pyrosequencing have been widely used for detecting SNPs and DNA variations [10], unambiguous detection of homopolymeric regions in DNA poses a challenge for pyrosequencing. Conventional electrophoresis based DNA sequencing methods have difficulty in characterizing frameshift mutations, because a deletion or insertion in any one allele will cause the sequencing reaction between the two alleles to be out of phase, leading to difficulties in interpreting the sequencing data. The primary limitations of current electrophoresis based DNA sequencers are the long time it takes for
samples to travel across a gel electrophoretically and the resolving capability of the gel itself. In this regard, mass spectrometry has emerged as a potential alternative tool for the separation and detection of nucleic acids. Traditional mass spectrometers that have been previously used to analyze small molecules cannot be used for this purpose since the relatively delicate nature of the phosphate linkage between bases in DNA makes efficient intact volatilization of DNA molecules nearly impossible without fragmentation. Two “soft” ionization techniques, developed for the analysis of biological and synthetic polymers, have been employed in the analysis of nucleic acids, namely electrospray ionization (ESI) and matrix assisted laser desorption ionization (MALDI) [11,12]. Due to its low sampling speed, ESI mass spectrometry is mainly used in SNP detection methods in pooled DNA samples [13–15]. MALDI time-of-flight (MALDI-TOF) mass spectrometry (MS) has been widely explored for nucleic acids detection due to the relatively large mass ranges easily detectable, the generation of mainly singly charged molecular ions and ease of automation for sample preparation and data collection. MALDI-TOF MS assays have been developed that range from short sequencing (up to 100 bases) to genotyping and insertion/deletion detection assays.
2. MALDI-TOF MS analysis of nucleic acids In MALDI-TOF MS sample analysis, a matrix solution is added to the DNA fragments to be analyzed (Fig. 1). The matrix–analyte mixture is then spotted on to a target plate and allowed to crystallize. The resulting crystal is hit with a laser to ionize the analyte and introduce it into the flight tube. The common mechanism attributed to the ionization process is that when the laser hits the crystal, a plume of matrix and DNA fragments is released from the surface. Initially, the matrix molecules are primarily ionized [16]. Subsequent collisions of the matrix and DNA molecules generate DNA ions. These ions are then made to pass through
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
5
Fig. 1. Schematic of MALDI-TOF mass spectrometry for DNA analysis. Nucleic acid samples that have been co-crystallized with a matrix are ionized with a nitrogen laser. An electric field is applied to accelerate the DNA ions towards the detector. Lighter ions hit the detector first, while the slower traveling heavier ions hit the detector later.
an electric field, which causes them to fly through the flight tube to the detector. In its simplest form, conservation of energy can be used to relate the electric potential energy of the DNA fragment to its kinetic energy. Thus, lighter ions (smaller DNA fragments) travel faster than heavier ions (larger DNA fragments) leading to a separation of fragments in the flight tube based on their mass difference. The time of flight is measured and a calibration factor is used to convert this value into the mass-to-charge ratio. Several MALDI-TOF MS based techniques have been developed for DNA sequencing and SNP detection. Due to the highly accurate nature of mass spectrometry measurements, these techniques are unparalleled in terms of their ability to consistently and accurately detect nucleic acid identity. The mass resolu-
tion in theory can be as good as 1 Da. A major limitation of all these techniques is the maximum detectable mass range by MALDI-TOF MS of nucleic acids with sufficient resolution to determine the nucleotide identity without fragmentation. In order to measure DNA masses accurately with MALDI-TOF MS over a large mass range, these limitations need to be overcome. For example, despite the use of soft ionization techniques such as MALDI, large DNA molecules have been shown to easily fragment in the mass spectrometer. Furthermore, alkaline and alkaline earth salts have been shown to easily adhere to the negatively charged DNA backbone complicating the data analysis. The discovery of 3-hydroxypicolinic acid (3-HPA) as a matrix for MALDI has significantly facilitated efficient DNA ionization and improved sample resolution and sen-
6
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
sitivity [17]. While much effort has been focused on discovering alternative matrices to enhance the ionization ability of DNA, the newly discovered matrices still need further development [18–20]. Currently samples can be desalted through a variety of methods including ion exchange columns and solid phase capture techniques. Another major limitation in the practical implementation of MALDI-TOF MS for DNA analysis is that the DNA analytes do not co-crystallize with the matrix homogenously. Often during analysis the laser must be slowly moved along the crystal to find spots with a large concentration of DNA in order to obtain high signal-to-noise ratios. Several methods have been used to overcome this limitation, including the use of piezoelectric spotters to spot down MALDI samples that have a diameter less than the impact area of the laser and hydrophobic surfaces that aid in the uniform dispersion of DNA fragments during the crystallization process [21,22]. For example, Guo et al. reported the use of a procedure for preparing MALDI samples on a hydrophobic surface for detecting low-concentration of oligonucleotides [22]. Collectively, these developments have had a huge impact on the ability to automate MALDI-TOF MS analysis and tremendously increase throughput.
3. DNA sequencing Despite the completion of the human genome project, new technologies that can sequence DNA faster and cheaper are still in great demand. Such technologies are needed for resequencing areas in the human genome to determine genetic variations for disease gene discovery, and for de novo sequencing of other organisms to determine what makes different species unique. By eliminating the need for gel electrophoresis, MALDI-TOF MS analysis of DNA fragments has the potential to greatly enhance the accuracy of these sequencing efforts. While numerous enzymatic digestion methods have been used, the best techniques for DNA sequencing so far have relied upon the Sanger dideoxy procedure [23] to generate DNA sequencing fragments. Several groups using a variety of sample purification procedures have reported DNA sequencing results based on this method [22,24–29]. Using cleavable primers, Monforte and
Becker demonstrated read lengths up to 100 bp [29]. In their published procedures, Monforte and Becker purified the DNA sequencing sample using a cleavable biotinylated primer, so that the extension fragments from the primer are captured by streptavidin coated magnetic beads at the 5 end of the extension fragments, while the other components in the sequencing reaction are washed away. Fu et al. reported the sequencing of exons 3 and 5 of the p53 gene using MALDI-TOF MS with an average read length of 35 bp [21]. They processed the sequencing samples using immobilized DNA templates on a solid phase for one cycle extension. The extended DNA fragments are hybridized on the immobilized templates, while the other components in the sequencing reaction are eliminated. These efforts established the feasibility of using MALDI-TOF MS for DNA sequencing. However, in both methods, falsely stopped DNA sequencing fragments (fragments terminated at dNTPs instead of ddNTPs) were not eliminated and were introduced into the mass spectrometer. In addition, four separate reactions were used, one for each dideoxynucleotide terminator analogous to the approach used in dye-labeled primer sequencing. It has been shown that false stops and dimerized primers that are formed in the mass spectrometry analysis process produce extra peaks in the mass spectra, preventing accurate base identification [25]. Since primers are the key components in a Sanger sequencing reaction, they are typically used in excess to drive the reaction forward, generating a high yield of DNA sequencing fragments. A high number of unreacted primers in the mixture being analyzed can however often create dimer peaks (2M + H) in the mass spectrum, which can directly interfere with interpretation of the MS data [25]. False stops are caused during a DNA extension reaction when the enzyme prematurely falls off the DNA template thereby terminating a DNA fragment with a dNTP instead of a ddNTP. False stops generate extension fragments with a mass difference of 16 Da less than the correctly terminated fragment. If such an event occurs, it can lead to difficulties with the subsequent data analysis, since the resulting peak can be at a mass position confusing the identification of the expected peak. Falsely stopped DNA fragments vary in size and cannot be eliminated by chromatographic methods. Ideally, for DNA sequencing with MALDI-TOF MS, one would like to establish an approach that allows sequencing reactions to be performed in one tube
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
to simplify sample preparation, to use cycle sequencing for increasing the yield of the DNA sequencing fragments, and to isolate pure DNA sequencing fragments free from false stops and other sequencing reaction components. Edwards et al. have explored the use of solid phase capturable dideoxynucleotides (SPCddNTPs) to develop a high fidelity method to sequence DNA by mass spectrometry [30]. While this method is similar to previously developed methods that rely on the analysis of Sanger sequencing reaction product fragments, it eliminates falsely stopped DNA fragments, excess primers and salts to generate a high yield of pure sequencing products for efficient data analysis. The SPC-sequencing assay takes advantage of the high specificity of the interaction between biotin and streptavidin by using biotinylated dideoxynucleotide terminators in the sequencing reaction to ensure the efficient removal of salts, excess primers and false stops from the DNA sequencing products. Briefly, in the SPC-sequencing approach, a DNA template, dNTPs (A, C, G, T), biotinylated dideoxynucleotides [biotin-ddNTPs], a primer and DNA polymerase are combined in one tube (Fig. 2). After polymerase extension and termination reactions, a series
7
of DNA fragments with different lengths are generated. The sequencing reaction mixture is then incubated with a streptavidin-coated solid phase such as magnetic beads. Only DNA sequencing fragments that are terminated with biotinylated dideoxynucleotides at the 3 end are captured on the solid phase. Excess primers, falsely terminated DNA fragments, enzymes and all other components from the sequencing reaction are washed away. The biotinylated DNA sequencing fragments are then cleaved off the solid phase by disrupting the interaction between biotin and streptavidin using ammonium hydroxide or formamide to obtain a pure set of DNA sequencing fragments. These fragments are then mixed with matrix (3-HPA) and loaded onto a mass spectrometer to produce accurate mass spectra of the DNA sequencing fragments. Since each of the four nucleotides (A, C, G, T) has a unique molecular mass, the mass difference between adjacent peaks of the mass spectra gives the sequence identity of the nucleotides. An example of a mass spectrum from SPCsequencing is shown in Fig. 3. The first peak in the spectrum corresponds to the extension of the primer by the first nucleotide that is complementary to the corresponding nucleotide in the DNA template. The differ-
Fig. 2. Schematic representation of SPC-sequencing. Sanger sequencing reactions are carried out with biotinylated dideoxynucleotides. After solid phase capture of the correctly terminated DNA sequencing fragments, excess primer, incorrectly terminated DNA fragments, and other impurities are removed. The pure DNA products are then released from the solid phase for analysis by mass spectrometry.
8
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
sulting in a diminished resolving capacity of the mass spectrometer for larger DNA fragments [29]. Fei et al. have shown that using dye labeled ddNTP paired with a conventional dNTP to space out the mass difference can increase the detection resolution in a SNP assay [31]. Similarly, the use of biotinylated terminators with increased mass differences to generate DNA sequencing products significantly improves sequence identification by MS. This increase in mass resolution is very useful for heterozygote detection. While the read lengths of MALDI-TOF MS based assays are still insufficient for de novo sequencing, these assays may prove ideal for resequencing projects that target a particular set of mutations. A variety of single base extension assays have been designed for this purpose, however sometimes these assays are not ideal since they require previous knowledge of the preceding sequence for primer design and synthesis. In highly variable regions of a particular gene, these methods may not suffice, as sampling only a few bases at a time could prove very inefficient. DNA sequencing on the other hand allows primer design flexibility and accurate sampling across a short range of nucleotides along the sequence. Fig. 3. DNA sequencing mass spectrum generated from Biotin-16ddUTP and Biotin-11-dd(A,G,C)TPs. The spectrum shows the sequence of a PCR template amplified from genomic DNA that is wild type at the BRCA15382insC locus. The number assigned to each peak corresponds to the difference in mass between that peak and it’s preceding peak, which is used to identify the base that the peak represents.
ence in mass between each successive peak can be measured to determine the identity of the nucleotide that corresponds to that peak. No primer peak is seen in the mass spectrum, since the primers are not biotinylated and are removed after solid phase capture, eliminating the possibility of false peaks caused by primer dimers. There are no peaks due to false stops in the spectrum either. When biotin-ddNTPs with different molecular weights are used, the smallest mass difference between any two sequencing fragments is 16 Da; with conventional ddNTP terminators the smallest mass difference is only 9 Da (the difference between A and T). Thus, the A/T heterozygous peaks are not well resolved in the MALDI-TOF mass spectrum using conventional ddNTPs [25]. It has been shown that as DNA fragment size increases so does the mass spectral peak width re-
4. Detection of frameshift mutations For genetic screening of the deletion and insertion mutations, such as the 185delAG and 5382insC mutations in the BRCA1 gene commonly occurring in members of the Ashkenazi population with a strong family history of breast and ovarian cancer [32], approximately 5–20 bps of clean DNA sequences around the mutation site are all that are needed to accurately determine the nature of the mutation. SPC-sequencing can be used to accurately detect and characterize frameshift mutations in PCR-amplified templates from genomic DNA while avoiding many of the limitations faced by other mutation-detection methods [33]. As schematically shown in Fig. 4, the principal advantage of SPCsequencing for mutation detection is the highly accurate identification of different bases that might coexist at a single locus along the DNA as a direct consequence of substitution or frameshift mutations. MALDI-TOF MS analysis displays the sequence data in the form of distinct mass peaks, facilitating the accurate characterization of frameshift mutations as compared to
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
9
Fig. 4. Schematic representation of SPC-sequencing used for frameshift mutation detection. Wild-type DNA has the same sequence for both alleles leading to the production of a single peak at every position in the mass spectrum that will correspond to the nucleotide existing at that position in the sequence (left panel). However, mutant DNA with an insertion or deletion (insertion shown here as an example) in one allele can have two different bases coexisting at the same sequence position. This leads to the formation of two distinct mass peaks at each position in the spectrum (right panel). By calculating the corresponding mass differences, the sequences of both alleles can be simultaneously read and the mutation site can be accurately characterized.
other sequencing methods that measure fluorescent or radioactive signals emitted from the labeled DNA. Sequencing results from a capillary electrophoresis sequencer and MALDI-TOF MS for a region around the mutation site 5382insC in exon 20 of the BRCA1 gene were compared in Fig. 5. Sequencing data produced by capillary electrophoresis does not allow determination of the allele with a high degree of confidence. The sequencing results using MALDI-TOF MS and biotin-ddNTPs are accurate and allow correct determination of the sequence of both alleles in the DNA containing a frame shift mutation in one copy. Likewise the same analysis can be performed for deletion mutations, as exemplified by the analysis of a 185delAG mutant of the BRCA1 gene (Fig. 6). There is a single peak corresponding to one base at each position before the mutation site. Beyond the mutation site, however, there are two distinct peaks at every position, clearly demonstrating the heterozygosity, and facilitating an efficient analysis of the sequence before and after the mutation site. In the capillary electrophoresis sequencing data (Figs. 5A and 6A), the insertion (or deletion) mutation
causes a frameshift in the sequence with the mutation. Consequently, DNA fragments are produced with the same length, but terminating with different nucleotides. It might be possible to confirm the presence of a mutation by recognizing this distinct pattern in the sequencing data for a mutant in a combination with sequencing data from the reverse direction, but an accurate characterization of the mutation from these data might still prove difficult and tedious. While highly experienced researchers can sometimes determine these types of mutations by directly looking at sequencing data, in general this method is highly error prone and requires manual analysis. In the mass spectrum (Figs. 5B and 6B), the initial single peaks represent the wild type portion of the sequence. Since the frameshift mutation has not yet occurred, the sequences of both alleles are in phase giving rise to a single peak. However, there is an occurrence of doublet peaks after the first peak followed by doublet peaks at all subsequent positions. The mutation site can be identified as the position where the doublet peaks are first observed. MALDI-TOF MS analysis using biotinylated ddNTPs provides a fast reliable
10
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
Fig. 5. (A) Fluorescence electropherogram for a 5382insC mutant. Peaks are clear before the mutation site and ambiguous beyond the mutation site (after “C” at position 118). (B) Mass spectrum showing sequencing data for a 5382insC mutant. The appearance of a succession of clear double peaks at and after the third position confirms the presence of the insertion of a “C”. The wild-type DNA sequence is 5 -. . .CCAGGA. . .-3 , while the mutant DNA sequence is 5 -. . .CCCAGG. . .-3 .
method to make this determination, which can be easily automated for high-throughput data analysis.
5. RNA sequencing assays Since the advent of MALDI-TOF MS as a method to analyze nucleic acids, it has been shown that RNA has
Fig. 6. (A) Sequence electropherogam for the 185delAG mutant. Peaks are clear before the mutation site and ambiguous beyond the mutation site (after “T” at position 217). (B) Mass spectrum showing the sequencing data for a 185delAG mutant. The first single peak corresponds to a “T” in both alleles showing there is no mutation up to that point. The deletion of a C and T in one of the two alleles (A and G in the forward sequence) in the subsequent two positions causes the appearance of a series of double peaks starting from the second position. The reverse wild-type DNA sequence is 5 -. . .TCTAAGA. . .-3 , while the mutant DNA sequence is 5 -. . .TAAGATT. . .-3 .
several distinct advantages over DNA for direct measurement with this technique. RNA has been shown to be less prone to fragmentation during the analysis process. In addition, salts are less likely to stick to RNA in comparison to DNA, easing sample preparation [34]. To take advantage of the higher sensitivity offered by
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
the use of RNA due to these factors, several assays have been developed including using DNA as a template and synthesizing RNA ladders using a modified form of Sanger sequencing with 3 -deoxynucleotide triphosphates (3 -dNTPs) and phage SP6 or T7 RNA polymerase. Using four reactions (one reaction for each terminator), Kwon et al. have shown that read lengths of over 50 nucleotides can be determined [34]. This same group also demonstrated that a single-base extension assay using RNA polymerase could be used for genotyping. In this assay, to circumvent the limitation caused by the 1 Da mass difference between U and C, dUTP was substituted with ␣-thio-dUTP. Other methods have explored direct sequencing of RNA using enzymatic digests. The mass difference between adjacent fragments is measured to determine the identity of each base. This technique is also complicated by the 1 Da mass difference between U and C. In the sequencing results U and C were differentiated by their distinct peak intensities that differ due to the discrepancy in cleavage efficiency of the RNase used during the enzymatic reaction. Sequencing results from short synthetic templates using this method were reported [35]. Another technique that has been explored involves the use of base-specific RNases to cleave RNA products generated by transcription from a PCR product amplified using a promoter-tailed primer. In this method, all possible cleavage products for the fragment to be analyzed are pre-computed to form distinct fingerprint patterns, which are then used to determine the sequence at cleavage points. With this technique it is possible to amplify larger gene products and analyze their reaction products to determine de novo mutation sites [36].
6. Future directions The primary limitation in using MALDI-TOF MS based techniques for DNA sequencing is the relatively short read lengths that can be accurately determined. The difficulty involves the amount of ion current available to simultaneously measure different DNA fragments and the effects this has on the sensitivity and resolution of larger DNA fragments. In this regard, there have been reports of the detection of large DNA molecules of 1000 nucleotides in length. However, the peak width in the data generated can range over sev-
11
eral 100 Da and the mass accuracy is quite poor. To improve detector sensitivity and begin to overcome some of these limitations, cryogenic detectors for mass spectrometry have been explored [37]. While these detectors have been shown to be able to detect single molecules they have yet to be fully implemented in the detection of DNA. Currently, MALDI-TOF MS techniques provide rapid and efficient methods for detecting SNPs and mutations using single base extension assays and short sequencing assays. Sequencing methods are able to detect insertion/deletion mutations with great accuracy. Automation has been or can easily be implemented for most assays allowing unparalleled levels of throughput for SNP detection and mutation screening projects that will have a profound impact on the determination of the effects of SNPs and mutations in the function of the human genome.
References [1] E.S. Lander, L.M. Linton, B. Birren, C. Nusbaum, M.C. Zody, J. Baldwin, Initial sequencing and analysis of the human genome, Nature 409 (2001) 860–921. [2] J.C. Venter, M.D. Adams, E.W. Myers, P.W. Li, R.J. Mural, G.G. Sutton, The sequence of the human genome, Science 291 (2001) 1304–1351. [3] L.M. Smith, J.Z. Sanders, R.J. Kaiser, P. Hughes, C. Dodd, C.R. Connell, C. Heiner, S.B.H. Kent, L.E. Hood, Fluorescence detection in automated DNA sequencing analysis, Nature 321 (1986) 674–679. [4] J. Ju, C. Ruan, C.W. Fuller, A.N. Glazer, R.A. Mathies, Fluorescence energy transfer dye-labeled primers for DNA sequencing and analysis, Proc. Natl. Acad. Sci. U.S.A. 92 (1995) 4347–4351. [5] J. Ju, A.N. Glazer, R.A. Mathies, Cassette labeling for facile construction of energy transfer fluorescent primers, Nucleic Acids Res. 24 (1996) 1144–1148. [6] O. Salas-Solano, E. Carrilho, L. Kotler, A.W. Miller, W. Goetzinger, Z. Sosic, B.L. Karger, Routine DNA sequencing of 1000 bases in less than one hour by capillary electrophoresis with replaceable linear polyacrylamide solutions, Anal. Chem. 70 (1998) 3996–4003. [7] I. Kheterpal, J. Scherer, S.M. Clark, A. Radhakrishnan, J. Ju, C.L. Ginther, G.F. Sensabaugh, R.A. Mathies, DNA sequencing using a four-color confocal fluorescence capillary array scanner, Electrophoresis 17 (1996) 1852–1859. [8] J.M. Bowling, K.L. Bruner, J.L. Cmarik, C. Tibbetts, Neighboring nucleotide interactions during DNA sequencing gel electrophoresis, Nucleic Acids Res. 19 (1991) 3089–3097. [9] H. Yamakawa, O. Ohara, A DNA cycle sequencing reaction that minimizes compressions on automated fluorescent sequencers, Nucleic Acids Res. 25 (1997) 1311–1312.
12
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
[10] M. Ronaghi, M. Uhlen, P. Nyren, A sequencing method based on real-time pyrophosphate, Science 281 (1998) 363. [11] J.B. Fenn, M. Mann, C.K. Meng, S.F. Wong, C.M. Whitehouse, Electrospray ionization for mass spectrometry of large biomolecules, Science 246 (1989) 64–71. [12] R.C. Beavis, B.T. Chait, Matrix-assisted laser-desorption mass spectrometry using 355 nm radiation., Rapid Commun. Mass Spectrom. 3 (1989) 436–439. [13] H. Oberacher, P.J. Oefner, G. Holzl, A. Premstaller, K. Davis, C.C. Huber, Re-sequencing of multiple single nucleotide polymorphisms by liquid chromatography-electrospray ionization mass spectrometry, Nucleic Acids Res. 30 (2002) e67. [14] S. Zhang, C.K. Van Pelt, X. Huang, G. Schultz, Detection of single nucleotide polymorphisms using electrospray ionization mass spectrometry: validation of a one-well assay and quantitative pooling studies, J. Mass Spectrom. 37 (2002) 1039–1050. [15] J.J. Walters, W. Muhammad, K.F. Fox, A. Fox, D. Xie, K.E. Creek, L. Pirisi, Genotyping single nucleotide polymorphisms using intact polymerase chain reaction products by electrospray quadrupole mass spectrometry, Rapid Commun. Mass Spectrom. 15 (2001) 1752–1759. [16] C.M. Land, G.R. Kinsel, Investigation of the mechanism of intracluster proton transfer from sinapinic acid to biomolecular analytes, J. Am. Soc. Mass Spectrom. 9 (1998) 1060–1067. [17] K.J. Wu, A. Steding, C.H. Becker, Matrix-assisted laser desorption time-of-flight mass spectrometry of oligonucleotides using 3-hydroxypicolinic acid as an ultraviolet-sensitive matrix, Rapid Commun. Mass Spectrom. 7 (1993) 142–146. [18] Y.F. Zhu, C.N. Chung, N.I. Taranenko, S.L. Allman, S.A. Martin, L. Haff, C.H. Chen, The study of 2,3,4trihydroxyacetophenone and 2,4,6-trihydroxyacetophenone as matrices for DNA detection in matrix-assisted laser desorption/ionization time-of-flight mass spectrometry, Rapid Commun. Mass Spectrom. 10 (1996) 383–388. [19] E. Nordhoff, A. Ingendoh, R. Cramer, A. Overberg, B. Stahl, M. Karas, F. Hillenkamp, P.F. Crain, Matrix-assisted laser desorption/ionization mass spectrometry of nucleic acids with wavelengths in the ultraviolet and infrared, Rapid Commun. Mass Spectrom. 6 (1992) 771–776. [20] P. Lecchi, H.M. Le, L.K. Pannell, 6-Aza-2-thiothymine: a matrix for MALDI spectra of oligonucleotides, Nucleic Acids Res. 23 (1995) 1276–1277. [21] D.J. Fu, K. Tang, A. Braun, D. Reuter, B. Darnhofer-Demar, D.P. Little, M.J. O’Donnell, C.R. Cantor, H. Koster, Sequencing exons 5–8 of the p53 gene by MALDI-TOF mass spectrometry, Nat. Biotechnol. 16 (1998) 381–384. [22] Y. Liu, X. Sun, B. Guo, Matrix-assisted laser desorption/ionization time-of-flight analysis of low-concentration oligonucleotides and mini-sequencing products, Rapid Commun. Mass Spectrom. 17 (2003) 2354–2360.
[23] F. Sanger, S. Nicklen, A.R. Coulson, DNA sequencing with chain-terminating inhibitors, Proc. Natl. Acad. Sci. U.S.A. 74 (1977) 5463–5467. [24] M.C. Fitzgerald, L. Zhu, L.M. Smith, The analysis of mock DNA sequencing reactions using matrix-assisted laser desorption/ionization mass spectrometry, Rapid Commun. Mass Spectrom. 7 (1993) 895–897. [25] M.T. Roskey, P. Juhasz, I.P. Smirnov, E.J. Takach, S.A. Martin, L. Haff, DNA sequencing by delayed extraction-matrix-assisted laser desorption/ionization time of flight mass spectrometry, Proc. Natl. Acad. Sci. U.S.A. 93 (1996) 4724–4729. [26] N.I. Taranenko, S.L. Allman, V.V. Golovlev, N.V. Taranenko, N.R. Isola, C.H. Chen, Sequencing DNA using mass spectrometry for ladder detection, Nucleic Acids Res. 26 (1998) 2488–2490. [27] F. Kirpekar, E. Nordhoff, L.K. Larsen, K. Kristiansen, P. Roepstorff, F. Hillenkamp, DNA sequence analysis by MALDI mass spectrometry, Nucleic Acid Res. 26 (1998) 2554–2559. [28] E. Nordhoff, C. Luebbert, G. Thiele, V. Heiser, H. Lehrach, Rapid determination of short DNA sequences by the use of MALDI-MS, Nucleic Acid Res. 28 (2000) e86. [29] J. Monforte, C. Becker, High-throughput DNA analysis by timeof-flight mass spectrometry, Nat. Med. 3 (1997) 360–362. [30] J.R. Edwards, Y. Itagaki, J. Ju, DNA sequencing using biotinylated dideoxynucleotides and mass spectrometry, Nucleic Acids Res. 29 (2001) e104. [31] Z. Fei, T. Ono, L.M. Smith, MALDI-TOF mass spectrometric typing of single nucleotide polymorphisms with mass-tagged ddNTPs, Nucleic Acids Res. 26 (1998) 2827–2828. [32] L.S. Friedman, E.A. Ostermeyer, C.I. Szabo, P. Dowd, E.D. Lynch, S.E. Rowell, M.C. King, Confirmation of BRCA1 by analysis of germline mutations linked to breast and ovarian cancer in ten families, Nat. Genet. 8 (1994) 399–404. [33] H. Ruparel, M.E. Ulz, S. Kim, J. Ju, Digital detection of genetic mutations using SPC-sequencing, Genome Res. 14 (2004) 296–300. [34] Y. Kwon, K. Tang, C. Cantor, H. Koster, C. Kang, DNA sequencing and genotyping by transcriptional synthesis of chainterminated RNA ladders and MALDI-TOF mass spectrometry, Nucleic Acids Res. 29 (2001) e11. [35] K. Faulstich, K. Worner, H. Brill, J.W. Engels, A sequencing method for RNA oligonucleotides based on mass spectrometry, Anal. Chem. 69 (1997) 4349–4353. [36] S. Krebs, I. Medugorac, D. Seichter, M. Forster, RNaseCut: a MALDI mass spectrometry-based method for SNP discovery, Nucleic Acids Res. 31 (2003) e37. [37] G.C. Hilton, J.M. Martinis, D.A. Wollman, K.D. Irwin, L.L. Dulcie, D. Gerber, P.M. Gillevet, D. Twerenbold, Impact energy measurement in time-of-flight mass spectrometry with cryogenic microcalorimeters, Nature 391 (1998) 672–675.
Review
Mass-spectrometry DNA sequencing John R. Edwards a, b , Hameer Ruparel a, b , Jingyue Ju a, b, ∗ a
Columbia Genome Center, Columbia University College of Physicians and Surgeons, Room 405A, Russ Berrie Medical Science Pavilion, New York, NY 10032, USA b Department of Chemical Engineering, Columbia University, New York, NY 10027, USA Received 1 July 2004; accepted 15 July 2004 Available online 11 February 2005
Abstract Matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) has been explored widely for DNA sequencing. Compared to gel electrophoresis based sequencing systems, mass spectrometry produces very high resolution of sequencing fragments, rapid separation on microsecond time scales, and completely eliminates compressions associated with gel-based systems. While most of the research efforts have focused on using mass spectrometers to analyze the DNA products from Sanger sequencing or enzymatic digestion reactions, the read lengths attainable are currently insufficient for large-scale de novo sequencing. The advantage of mass-spectrometry sequencing is that one can unambiguously identify frameshift mutations and heterozygous mutations making it an ideal choice for resequencing projects. In these applications, DNA sequencing fragments that are the same length but with different base compositions are generated, which are challenging to consistently distinguish in gel-based sequencing systems. In contrast, MALDI-TOF MS produces mass spectra of these DNA sequencing fragments with nearly digital resolution, allowing accurate determination of the mixed bases. For these reasons mass spectrometry based sequencing has mainly been focused on the detection of frameshift mutations and single nucleotide polymorphisms (SNPs). More recently, assays have been developed to indirectly sequence DNA by first converting it into RNA. These assays take advantage of the increased resolution and detection ability of MALDI-TOF MS for RNA. © 2005 Elsevier B.V. All rights reserved. Keywords: MALDI-TOF mass spectrometry; Single nucleotide polymorphism; DNA sequencing
Contents 1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MALDI-TOF MS analysis of nucleic acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DNA sequencing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
∗ Corresponding author. Tel.: +1 212 851 5172; fax: +1 212 851 5176. E-mail address: [email protected] (J. Ju).
0027-5107/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.mrfmmm.2004.07.021
4 4 6
4
4. 5. 6.
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
Detection of frameshift mutations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 RNA sequencing assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1. Introduction The ability to sequence DNA accurately and rapidly is revolutionizing biology and medicine. This rapid technology development involving engineering, chemistry, biology, and computer science makes it possible to move from studying single genes at a time to analyzing and comparing entire genomes. With the completion of the human genome sequence map [1,2], resequencing polymorphic areas of the genome linked to disease development will contribute greatly to the understanding of the molecular basis of disease and developing new therapeutics. High accuracy is required for these genetic variation studies. The current state-ofthe-art technology for high throughput DNA sequencing is the capillary array DNA sequencer using laser induced fluorescence detection [3–7]. Although this technology addresses the throughput and read length requirements of large scale DNA sequencing projects, the accuracy required for detecting genetic variations needs to be improved for biological research and clinical diagnostics. For example, electrophoresis based sequencing methods have difficulty detecting heterozygotes unambiguously and are not 100% accurate for a given base due to compressions in GC rich regions [8,9]. In addition, the first few bases after the priming site are often masked by the high fluorescence signal from the excess dye-labeled primers or dye-labeled terminators, and are difficult to identify. While other techniques such as pyrosequencing have been widely used for detecting SNPs and DNA variations [10], unambiguous detection of homopolymeric regions in DNA poses a challenge for pyrosequencing. Conventional electrophoresis based DNA sequencing methods have difficulty in characterizing frameshift mutations, because a deletion or insertion in any one allele will cause the sequencing reaction between the two alleles to be out of phase, leading to difficulties in interpreting the sequencing data. The primary limitations of current electrophoresis based DNA sequencers are the long time it takes for
samples to travel across a gel electrophoretically and the resolving capability of the gel itself. In this regard, mass spectrometry has emerged as a potential alternative tool for the separation and detection of nucleic acids. Traditional mass spectrometers that have been previously used to analyze small molecules cannot be used for this purpose since the relatively delicate nature of the phosphate linkage between bases in DNA makes efficient intact volatilization of DNA molecules nearly impossible without fragmentation. Two “soft” ionization techniques, developed for the analysis of biological and synthetic polymers, have been employed in the analysis of nucleic acids, namely electrospray ionization (ESI) and matrix assisted laser desorption ionization (MALDI) [11,12]. Due to its low sampling speed, ESI mass spectrometry is mainly used in SNP detection methods in pooled DNA samples [13–15]. MALDI time-of-flight (MALDI-TOF) mass spectrometry (MS) has been widely explored for nucleic acids detection due to the relatively large mass ranges easily detectable, the generation of mainly singly charged molecular ions and ease of automation for sample preparation and data collection. MALDI-TOF MS assays have been developed that range from short sequencing (up to 100 bases) to genotyping and insertion/deletion detection assays.
2. MALDI-TOF MS analysis of nucleic acids In MALDI-TOF MS sample analysis, a matrix solution is added to the DNA fragments to be analyzed (Fig. 1). The matrix–analyte mixture is then spotted on to a target plate and allowed to crystallize. The resulting crystal is hit with a laser to ionize the analyte and introduce it into the flight tube. The common mechanism attributed to the ionization process is that when the laser hits the crystal, a plume of matrix and DNA fragments is released from the surface. Initially, the matrix molecules are primarily ionized [16]. Subsequent collisions of the matrix and DNA molecules generate DNA ions. These ions are then made to pass through
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
5
Fig. 1. Schematic of MALDI-TOF mass spectrometry for DNA analysis. Nucleic acid samples that have been co-crystallized with a matrix are ionized with a nitrogen laser. An electric field is applied to accelerate the DNA ions towards the detector. Lighter ions hit the detector first, while the slower traveling heavier ions hit the detector later.
an electric field, which causes them to fly through the flight tube to the detector. In its simplest form, conservation of energy can be used to relate the electric potential energy of the DNA fragment to its kinetic energy. Thus, lighter ions (smaller DNA fragments) travel faster than heavier ions (larger DNA fragments) leading to a separation of fragments in the flight tube based on their mass difference. The time of flight is measured and a calibration factor is used to convert this value into the mass-to-charge ratio. Several MALDI-TOF MS based techniques have been developed for DNA sequencing and SNP detection. Due to the highly accurate nature of mass spectrometry measurements, these techniques are unparalleled in terms of their ability to consistently and accurately detect nucleic acid identity. The mass resolu-
tion in theory can be as good as 1 Da. A major limitation of all these techniques is the maximum detectable mass range by MALDI-TOF MS of nucleic acids with sufficient resolution to determine the nucleotide identity without fragmentation. In order to measure DNA masses accurately with MALDI-TOF MS over a large mass range, these limitations need to be overcome. For example, despite the use of soft ionization techniques such as MALDI, large DNA molecules have been shown to easily fragment in the mass spectrometer. Furthermore, alkaline and alkaline earth salts have been shown to easily adhere to the negatively charged DNA backbone complicating the data analysis. The discovery of 3-hydroxypicolinic acid (3-HPA) as a matrix for MALDI has significantly facilitated efficient DNA ionization and improved sample resolution and sen-
6
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
sitivity [17]. While much effort has been focused on discovering alternative matrices to enhance the ionization ability of DNA, the newly discovered matrices still need further development [18–20]. Currently samples can be desalted through a variety of methods including ion exchange columns and solid phase capture techniques. Another major limitation in the practical implementation of MALDI-TOF MS for DNA analysis is that the DNA analytes do not co-crystallize with the matrix homogenously. Often during analysis the laser must be slowly moved along the crystal to find spots with a large concentration of DNA in order to obtain high signal-to-noise ratios. Several methods have been used to overcome this limitation, including the use of piezoelectric spotters to spot down MALDI samples that have a diameter less than the impact area of the laser and hydrophobic surfaces that aid in the uniform dispersion of DNA fragments during the crystallization process [21,22]. For example, Guo et al. reported the use of a procedure for preparing MALDI samples on a hydrophobic surface for detecting low-concentration of oligonucleotides [22]. Collectively, these developments have had a huge impact on the ability to automate MALDI-TOF MS analysis and tremendously increase throughput.
3. DNA sequencing Despite the completion of the human genome project, new technologies that can sequence DNA faster and cheaper are still in great demand. Such technologies are needed for resequencing areas in the human genome to determine genetic variations for disease gene discovery, and for de novo sequencing of other organisms to determine what makes different species unique. By eliminating the need for gel electrophoresis, MALDI-TOF MS analysis of DNA fragments has the potential to greatly enhance the accuracy of these sequencing efforts. While numerous enzymatic digestion methods have been used, the best techniques for DNA sequencing so far have relied upon the Sanger dideoxy procedure [23] to generate DNA sequencing fragments. Several groups using a variety of sample purification procedures have reported DNA sequencing results based on this method [22,24–29]. Using cleavable primers, Monforte and
Becker demonstrated read lengths up to 100 bp [29]. In their published procedures, Monforte and Becker purified the DNA sequencing sample using a cleavable biotinylated primer, so that the extension fragments from the primer are captured by streptavidin coated magnetic beads at the 5 end of the extension fragments, while the other components in the sequencing reaction are washed away. Fu et al. reported the sequencing of exons 3 and 5 of the p53 gene using MALDI-TOF MS with an average read length of 35 bp [21]. They processed the sequencing samples using immobilized DNA templates on a solid phase for one cycle extension. The extended DNA fragments are hybridized on the immobilized templates, while the other components in the sequencing reaction are eliminated. These efforts established the feasibility of using MALDI-TOF MS for DNA sequencing. However, in both methods, falsely stopped DNA sequencing fragments (fragments terminated at dNTPs instead of ddNTPs) were not eliminated and were introduced into the mass spectrometer. In addition, four separate reactions were used, one for each dideoxynucleotide terminator analogous to the approach used in dye-labeled primer sequencing. It has been shown that false stops and dimerized primers that are formed in the mass spectrometry analysis process produce extra peaks in the mass spectra, preventing accurate base identification [25]. Since primers are the key components in a Sanger sequencing reaction, they are typically used in excess to drive the reaction forward, generating a high yield of DNA sequencing fragments. A high number of unreacted primers in the mixture being analyzed can however often create dimer peaks (2M + H) in the mass spectrum, which can directly interfere with interpretation of the MS data [25]. False stops are caused during a DNA extension reaction when the enzyme prematurely falls off the DNA template thereby terminating a DNA fragment with a dNTP instead of a ddNTP. False stops generate extension fragments with a mass difference of 16 Da less than the correctly terminated fragment. If such an event occurs, it can lead to difficulties with the subsequent data analysis, since the resulting peak can be at a mass position confusing the identification of the expected peak. Falsely stopped DNA fragments vary in size and cannot be eliminated by chromatographic methods. Ideally, for DNA sequencing with MALDI-TOF MS, one would like to establish an approach that allows sequencing reactions to be performed in one tube
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
to simplify sample preparation, to use cycle sequencing for increasing the yield of the DNA sequencing fragments, and to isolate pure DNA sequencing fragments free from false stops and other sequencing reaction components. Edwards et al. have explored the use of solid phase capturable dideoxynucleotides (SPCddNTPs) to develop a high fidelity method to sequence DNA by mass spectrometry [30]. While this method is similar to previously developed methods that rely on the analysis of Sanger sequencing reaction product fragments, it eliminates falsely stopped DNA fragments, excess primers and salts to generate a high yield of pure sequencing products for efficient data analysis. The SPC-sequencing assay takes advantage of the high specificity of the interaction between biotin and streptavidin by using biotinylated dideoxynucleotide terminators in the sequencing reaction to ensure the efficient removal of salts, excess primers and false stops from the DNA sequencing products. Briefly, in the SPC-sequencing approach, a DNA template, dNTPs (A, C, G, T), biotinylated dideoxynucleotides [biotin-ddNTPs], a primer and DNA polymerase are combined in one tube (Fig. 2). After polymerase extension and termination reactions, a series
7
of DNA fragments with different lengths are generated. The sequencing reaction mixture is then incubated with a streptavidin-coated solid phase such as magnetic beads. Only DNA sequencing fragments that are terminated with biotinylated dideoxynucleotides at the 3 end are captured on the solid phase. Excess primers, falsely terminated DNA fragments, enzymes and all other components from the sequencing reaction are washed away. The biotinylated DNA sequencing fragments are then cleaved off the solid phase by disrupting the interaction between biotin and streptavidin using ammonium hydroxide or formamide to obtain a pure set of DNA sequencing fragments. These fragments are then mixed with matrix (3-HPA) and loaded onto a mass spectrometer to produce accurate mass spectra of the DNA sequencing fragments. Since each of the four nucleotides (A, C, G, T) has a unique molecular mass, the mass difference between adjacent peaks of the mass spectra gives the sequence identity of the nucleotides. An example of a mass spectrum from SPCsequencing is shown in Fig. 3. The first peak in the spectrum corresponds to the extension of the primer by the first nucleotide that is complementary to the corresponding nucleotide in the DNA template. The differ-
Fig. 2. Schematic representation of SPC-sequencing. Sanger sequencing reactions are carried out with biotinylated dideoxynucleotides. After solid phase capture of the correctly terminated DNA sequencing fragments, excess primer, incorrectly terminated DNA fragments, and other impurities are removed. The pure DNA products are then released from the solid phase for analysis by mass spectrometry.
8
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
sulting in a diminished resolving capacity of the mass spectrometer for larger DNA fragments [29]. Fei et al. have shown that using dye labeled ddNTP paired with a conventional dNTP to space out the mass difference can increase the detection resolution in a SNP assay [31]. Similarly, the use of biotinylated terminators with increased mass differences to generate DNA sequencing products significantly improves sequence identification by MS. This increase in mass resolution is very useful for heterozygote detection. While the read lengths of MALDI-TOF MS based assays are still insufficient for de novo sequencing, these assays may prove ideal for resequencing projects that target a particular set of mutations. A variety of single base extension assays have been designed for this purpose, however sometimes these assays are not ideal since they require previous knowledge of the preceding sequence for primer design and synthesis. In highly variable regions of a particular gene, these methods may not suffice, as sampling only a few bases at a time could prove very inefficient. DNA sequencing on the other hand allows primer design flexibility and accurate sampling across a short range of nucleotides along the sequence. Fig. 3. DNA sequencing mass spectrum generated from Biotin-16ddUTP and Biotin-11-dd(A,G,C)TPs. The spectrum shows the sequence of a PCR template amplified from genomic DNA that is wild type at the BRCA15382insC locus. The number assigned to each peak corresponds to the difference in mass between that peak and it’s preceding peak, which is used to identify the base that the peak represents.
ence in mass between each successive peak can be measured to determine the identity of the nucleotide that corresponds to that peak. No primer peak is seen in the mass spectrum, since the primers are not biotinylated and are removed after solid phase capture, eliminating the possibility of false peaks caused by primer dimers. There are no peaks due to false stops in the spectrum either. When biotin-ddNTPs with different molecular weights are used, the smallest mass difference between any two sequencing fragments is 16 Da; with conventional ddNTP terminators the smallest mass difference is only 9 Da (the difference between A and T). Thus, the A/T heterozygous peaks are not well resolved in the MALDI-TOF mass spectrum using conventional ddNTPs [25]. It has been shown that as DNA fragment size increases so does the mass spectral peak width re-
4. Detection of frameshift mutations For genetic screening of the deletion and insertion mutations, such as the 185delAG and 5382insC mutations in the BRCA1 gene commonly occurring in members of the Ashkenazi population with a strong family history of breast and ovarian cancer [32], approximately 5–20 bps of clean DNA sequences around the mutation site are all that are needed to accurately determine the nature of the mutation. SPC-sequencing can be used to accurately detect and characterize frameshift mutations in PCR-amplified templates from genomic DNA while avoiding many of the limitations faced by other mutation-detection methods [33]. As schematically shown in Fig. 4, the principal advantage of SPCsequencing for mutation detection is the highly accurate identification of different bases that might coexist at a single locus along the DNA as a direct consequence of substitution or frameshift mutations. MALDI-TOF MS analysis displays the sequence data in the form of distinct mass peaks, facilitating the accurate characterization of frameshift mutations as compared to
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
9
Fig. 4. Schematic representation of SPC-sequencing used for frameshift mutation detection. Wild-type DNA has the same sequence for both alleles leading to the production of a single peak at every position in the mass spectrum that will correspond to the nucleotide existing at that position in the sequence (left panel). However, mutant DNA with an insertion or deletion (insertion shown here as an example) in one allele can have two different bases coexisting at the same sequence position. This leads to the formation of two distinct mass peaks at each position in the spectrum (right panel). By calculating the corresponding mass differences, the sequences of both alleles can be simultaneously read and the mutation site can be accurately characterized.
other sequencing methods that measure fluorescent or radioactive signals emitted from the labeled DNA. Sequencing results from a capillary electrophoresis sequencer and MALDI-TOF MS for a region around the mutation site 5382insC in exon 20 of the BRCA1 gene were compared in Fig. 5. Sequencing data produced by capillary electrophoresis does not allow determination of the allele with a high degree of confidence. The sequencing results using MALDI-TOF MS and biotin-ddNTPs are accurate and allow correct determination of the sequence of both alleles in the DNA containing a frame shift mutation in one copy. Likewise the same analysis can be performed for deletion mutations, as exemplified by the analysis of a 185delAG mutant of the BRCA1 gene (Fig. 6). There is a single peak corresponding to one base at each position before the mutation site. Beyond the mutation site, however, there are two distinct peaks at every position, clearly demonstrating the heterozygosity, and facilitating an efficient analysis of the sequence before and after the mutation site. In the capillary electrophoresis sequencing data (Figs. 5A and 6A), the insertion (or deletion) mutation
causes a frameshift in the sequence with the mutation. Consequently, DNA fragments are produced with the same length, but terminating with different nucleotides. It might be possible to confirm the presence of a mutation by recognizing this distinct pattern in the sequencing data for a mutant in a combination with sequencing data from the reverse direction, but an accurate characterization of the mutation from these data might still prove difficult and tedious. While highly experienced researchers can sometimes determine these types of mutations by directly looking at sequencing data, in general this method is highly error prone and requires manual analysis. In the mass spectrum (Figs. 5B and 6B), the initial single peaks represent the wild type portion of the sequence. Since the frameshift mutation has not yet occurred, the sequences of both alleles are in phase giving rise to a single peak. However, there is an occurrence of doublet peaks after the first peak followed by doublet peaks at all subsequent positions. The mutation site can be identified as the position where the doublet peaks are first observed. MALDI-TOF MS analysis using biotinylated ddNTPs provides a fast reliable
10
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
Fig. 5. (A) Fluorescence electropherogram for a 5382insC mutant. Peaks are clear before the mutation site and ambiguous beyond the mutation site (after “C” at position 118). (B) Mass spectrum showing sequencing data for a 5382insC mutant. The appearance of a succession of clear double peaks at and after the third position confirms the presence of the insertion of a “C”. The wild-type DNA sequence is 5 -. . .CCAGGA. . .-3 , while the mutant DNA sequence is 5 -. . .CCCAGG. . .-3 .
method to make this determination, which can be easily automated for high-throughput data analysis.
5. RNA sequencing assays Since the advent of MALDI-TOF MS as a method to analyze nucleic acids, it has been shown that RNA has
Fig. 6. (A) Sequence electropherogam for the 185delAG mutant. Peaks are clear before the mutation site and ambiguous beyond the mutation site (after “T” at position 217). (B) Mass spectrum showing the sequencing data for a 185delAG mutant. The first single peak corresponds to a “T” in both alleles showing there is no mutation up to that point. The deletion of a C and T in one of the two alleles (A and G in the forward sequence) in the subsequent two positions causes the appearance of a series of double peaks starting from the second position. The reverse wild-type DNA sequence is 5 -. . .TCTAAGA. . .-3 , while the mutant DNA sequence is 5 -. . .TAAGATT. . .-3 .
several distinct advantages over DNA for direct measurement with this technique. RNA has been shown to be less prone to fragmentation during the analysis process. In addition, salts are less likely to stick to RNA in comparison to DNA, easing sample preparation [34]. To take advantage of the higher sensitivity offered by
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
the use of RNA due to these factors, several assays have been developed including using DNA as a template and synthesizing RNA ladders using a modified form of Sanger sequencing with 3 -deoxynucleotide triphosphates (3 -dNTPs) and phage SP6 or T7 RNA polymerase. Using four reactions (one reaction for each terminator), Kwon et al. have shown that read lengths of over 50 nucleotides can be determined [34]. This same group also demonstrated that a single-base extension assay using RNA polymerase could be used for genotyping. In this assay, to circumvent the limitation caused by the 1 Da mass difference between U and C, dUTP was substituted with ␣-thio-dUTP. Other methods have explored direct sequencing of RNA using enzymatic digests. The mass difference between adjacent fragments is measured to determine the identity of each base. This technique is also complicated by the 1 Da mass difference between U and C. In the sequencing results U and C were differentiated by their distinct peak intensities that differ due to the discrepancy in cleavage efficiency of the RNase used during the enzymatic reaction. Sequencing results from short synthetic templates using this method were reported [35]. Another technique that has been explored involves the use of base-specific RNases to cleave RNA products generated by transcription from a PCR product amplified using a promoter-tailed primer. In this method, all possible cleavage products for the fragment to be analyzed are pre-computed to form distinct fingerprint patterns, which are then used to determine the sequence at cleavage points. With this technique it is possible to amplify larger gene products and analyze their reaction products to determine de novo mutation sites [36].
6. Future directions The primary limitation in using MALDI-TOF MS based techniques for DNA sequencing is the relatively short read lengths that can be accurately determined. The difficulty involves the amount of ion current available to simultaneously measure different DNA fragments and the effects this has on the sensitivity and resolution of larger DNA fragments. In this regard, there have been reports of the detection of large DNA molecules of 1000 nucleotides in length. However, the peak width in the data generated can range over sev-
11
eral 100 Da and the mass accuracy is quite poor. To improve detector sensitivity and begin to overcome some of these limitations, cryogenic detectors for mass spectrometry have been explored [37]. While these detectors have been shown to be able to detect single molecules they have yet to be fully implemented in the detection of DNA. Currently, MALDI-TOF MS techniques provide rapid and efficient methods for detecting SNPs and mutations using single base extension assays and short sequencing assays. Sequencing methods are able to detect insertion/deletion mutations with great accuracy. Automation has been or can easily be implemented for most assays allowing unparalleled levels of throughput for SNP detection and mutation screening projects that will have a profound impact on the determination of the effects of SNPs and mutations in the function of the human genome.
References [1] E.S. Lander, L.M. Linton, B. Birren, C. Nusbaum, M.C. Zody, J. Baldwin, Initial sequencing and analysis of the human genome, Nature 409 (2001) 860–921. [2] J.C. Venter, M.D. Adams, E.W. Myers, P.W. Li, R.J. Mural, G.G. Sutton, The sequence of the human genome, Science 291 (2001) 1304–1351. [3] L.M. Smith, J.Z. Sanders, R.J. Kaiser, P. Hughes, C. Dodd, C.R. Connell, C. Heiner, S.B.H. Kent, L.E. Hood, Fluorescence detection in automated DNA sequencing analysis, Nature 321 (1986) 674–679. [4] J. Ju, C. Ruan, C.W. Fuller, A.N. Glazer, R.A. Mathies, Fluorescence energy transfer dye-labeled primers for DNA sequencing and analysis, Proc. Natl. Acad. Sci. U.S.A. 92 (1995) 4347–4351. [5] J. Ju, A.N. Glazer, R.A. Mathies, Cassette labeling for facile construction of energy transfer fluorescent primers, Nucleic Acids Res. 24 (1996) 1144–1148. [6] O. Salas-Solano, E. Carrilho, L. Kotler, A.W. Miller, W. Goetzinger, Z. Sosic, B.L. Karger, Routine DNA sequencing of 1000 bases in less than one hour by capillary electrophoresis with replaceable linear polyacrylamide solutions, Anal. Chem. 70 (1998) 3996–4003. [7] I. Kheterpal, J. Scherer, S.M. Clark, A. Radhakrishnan, J. Ju, C.L. Ginther, G.F. Sensabaugh, R.A. Mathies, DNA sequencing using a four-color confocal fluorescence capillary array scanner, Electrophoresis 17 (1996) 1852–1859. [8] J.M. Bowling, K.L. Bruner, J.L. Cmarik, C. Tibbetts, Neighboring nucleotide interactions during DNA sequencing gel electrophoresis, Nucleic Acids Res. 19 (1991) 3089–3097. [9] H. Yamakawa, O. Ohara, A DNA cycle sequencing reaction that minimizes compressions on automated fluorescent sequencers, Nucleic Acids Res. 25 (1997) 1311–1312.
12
J.R. Edwards et al. / Mutation Research 573 (2005) 3–12
[10] M. Ronaghi, M. Uhlen, P. Nyren, A sequencing method based on real-time pyrophosphate, Science 281 (1998) 363. [11] J.B. Fenn, M. Mann, C.K. Meng, S.F. Wong, C.M. Whitehouse, Electrospray ionization for mass spectrometry of large biomolecules, Science 246 (1989) 64–71. [12] R.C. Beavis, B.T. Chait, Matrix-assisted laser-desorption mass spectrometry using 355 nm radiation., Rapid Commun. Mass Spectrom. 3 (1989) 436–439. [13] H. Oberacher, P.J. Oefner, G. Holzl, A. Premstaller, K. Davis, C.C. Huber, Re-sequencing of multiple single nucleotide polymorphisms by liquid chromatography-electrospray ionization mass spectrometry, Nucleic Acids Res. 30 (2002) e67. [14] S. Zhang, C.K. Van Pelt, X. Huang, G. Schultz, Detection of single nucleotide polymorphisms using electrospray ionization mass spectrometry: validation of a one-well assay and quantitative pooling studies, J. Mass Spectrom. 37 (2002) 1039–1050. [15] J.J. Walters, W. Muhammad, K.F. Fox, A. Fox, D. Xie, K.E. Creek, L. Pirisi, Genotyping single nucleotide polymorphisms using intact polymerase chain reaction products by electrospray quadrupole mass spectrometry, Rapid Commun. Mass Spectrom. 15 (2001) 1752–1759. [16] C.M. Land, G.R. Kinsel, Investigation of the mechanism of intracluster proton transfer from sinapinic acid to biomolecular analytes, J. Am. Soc. Mass Spectrom. 9 (1998) 1060–1067. [17] K.J. Wu, A. Steding, C.H. Becker, Matrix-assisted laser desorption time-of-flight mass spectrometry of oligonucleotides using 3-hydroxypicolinic acid as an ultraviolet-sensitive matrix, Rapid Commun. Mass Spectrom. 7 (1993) 142–146. [18] Y.F. Zhu, C.N. Chung, N.I. Taranenko, S.L. Allman, S.A. Martin, L. Haff, C.H. Chen, The study of 2,3,4trihydroxyacetophenone and 2,4,6-trihydroxyacetophenone as matrices for DNA detection in matrix-assisted laser desorption/ionization time-of-flight mass spectrometry, Rapid Commun. Mass Spectrom. 10 (1996) 383–388. [19] E. Nordhoff, A. Ingendoh, R. Cramer, A. Overberg, B. Stahl, M. Karas, F. Hillenkamp, P.F. Crain, Matrix-assisted laser desorption/ionization mass spectrometry of nucleic acids with wavelengths in the ultraviolet and infrared, Rapid Commun. Mass Spectrom. 6 (1992) 771–776. [20] P. Lecchi, H.M. Le, L.K. Pannell, 6-Aza-2-thiothymine: a matrix for MALDI spectra of oligonucleotides, Nucleic Acids Res. 23 (1995) 1276–1277. [21] D.J. Fu, K. Tang, A. Braun, D. Reuter, B. Darnhofer-Demar, D.P. Little, M.J. O’Donnell, C.R. Cantor, H. Koster, Sequencing exons 5–8 of the p53 gene by MALDI-TOF mass spectrometry, Nat. Biotechnol. 16 (1998) 381–384. [22] Y. Liu, X. Sun, B. Guo, Matrix-assisted laser desorption/ionization time-of-flight analysis of low-concentration oligonucleotides and mini-sequencing products, Rapid Commun. Mass Spectrom. 17 (2003) 2354–2360.
[23] F. Sanger, S. Nicklen, A.R. Coulson, DNA sequencing with chain-terminating inhibitors, Proc. Natl. Acad. Sci. U.S.A. 74 (1977) 5463–5467. [24] M.C. Fitzgerald, L. Zhu, L.M. Smith, The analysis of mock DNA sequencing reactions using matrix-assisted laser desorption/ionization mass spectrometry, Rapid Commun. Mass Spectrom. 7 (1993) 895–897. [25] M.T. Roskey, P. Juhasz, I.P. Smirnov, E.J. Takach, S.A. Martin, L. Haff, DNA sequencing by delayed extraction-matrix-assisted laser desorption/ionization time of flight mass spectrometry, Proc. Natl. Acad. Sci. U.S.A. 93 (1996) 4724–4729. [26] N.I. Taranenko, S.L. Allman, V.V. Golovlev, N.V. Taranenko, N.R. Isola, C.H. Chen, Sequencing DNA using mass spectrometry for ladder detection, Nucleic Acids Res. 26 (1998) 2488–2490. [27] F. Kirpekar, E. Nordhoff, L.K. Larsen, K. Kristiansen, P. Roepstorff, F. Hillenkamp, DNA sequence analysis by MALDI mass spectrometry, Nucleic Acid Res. 26 (1998) 2554–2559. [28] E. Nordhoff, C. Luebbert, G. Thiele, V. Heiser, H. Lehrach, Rapid determination of short DNA sequences by the use of MALDI-MS, Nucleic Acid Res. 28 (2000) e86. [29] J. Monforte, C. Becker, High-throughput DNA analysis by timeof-flight mass spectrometry, Nat. Med. 3 (1997) 360–362. [30] J.R. Edwards, Y. Itagaki, J. Ju, DNA sequencing using biotinylated dideoxynucleotides and mass spectrometry, Nucleic Acids Res. 29 (2001) e104. [31] Z. Fei, T. Ono, L.M. Smith, MALDI-TOF mass spectrometric typing of single nucleotide polymorphisms with mass-tagged ddNTPs, Nucleic Acids Res. 26 (1998) 2827–2828. [32] L.S. Friedman, E.A. Ostermeyer, C.I. Szabo, P. Dowd, E.D. Lynch, S.E. Rowell, M.C. King, Confirmation of BRCA1 by analysis of germline mutations linked to breast and ovarian cancer in ten families, Nat. Genet. 8 (1994) 399–404. [33] H. Ruparel, M.E. Ulz, S. Kim, J. Ju, Digital detection of genetic mutations using SPC-sequencing, Genome Res. 14 (2004) 296–300. [34] Y. Kwon, K. Tang, C. Cantor, H. Koster, C. Kang, DNA sequencing and genotyping by transcriptional synthesis of chainterminated RNA ladders and MALDI-TOF mass spectrometry, Nucleic Acids Res. 29 (2001) e11. [35] K. Faulstich, K. Worner, H. Brill, J.W. Engels, A sequencing method for RNA oligonucleotides based on mass spectrometry, Anal. Chem. 69 (1997) 4349–4353. [36] S. Krebs, I. Medugorac, D. Seichter, M. Forster, RNaseCut: a MALDI mass spectrometry-based method for SNP discovery, Nucleic Acids Res. 31 (2003) e37. [37] G.C. Hilton, J.M. Martinis, D.A. Wollman, K.D. Irwin, L.L. Dulcie, D. Gerber, P.M. Gillevet, D. Twerenbold, Impact energy measurement in time-of-flight mass spectrometry with cryogenic microcalorimeters, Nature 391 (1998) 672–675.