Published online March 22, 2004 Nucleic Acids Research, 2004, Vol. 32, No. 5 e53 DOI: 10.1093/nar/gnh051
Sensitive and speci®c detection of K-ras mutations in colon tumors by short oligonucleotide mass analysis Matilde E. Lleonart*, Santiago RamoÂn y Cajal1, John D. Groopman2 and Marlin D. Friesen Unit of Gene±Environment Interactions, International Agency for Research on Cancer, 69372 Lyon, France, 1 Department of Pathology, Hospital Vall d'Hebron, Paseo Vall d'HebroÂn 119±129, 08035 Barcelona, Spain and 2 Department of Environmental Health Sciences, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD 21205, USA Received December 4, 2003; Revised February 4, 2004; Accepted March 3, 2004
ABSTRACT Short oligonucleotide mass analysis (SOMA) is a technique by which small sequences of mutated and wild-type DNA, produced by PCR ampli®cation and restriction digestion, are characterized by HPLC-electrospray ionization tandem mass spectrometry. We have adapted the method to speci®cally detect two common point mutations at codon 12 of the c-K-ras gene. Mutations in DNA from 121 colon tumor samples were identi®ed by SOMA and validated by comparison with sequencing. SOMA correctly identi®ed 26 samples containing the 12GAT mutation and four samples containing the 12AGT mutation. Sequencing did not reveal mutant DNA in three samples out of the 26 samples shown by SOMA to contain the 12GAT mutation. In these three samples, the presence of mutant DNA was con®rmed by SOMA analysis after selective PCR ampli®cation in the presence of BstN1 restriction enzyme. Additional mutations in codons 12 and 13 were revealed by sequencing in 24 additional samples, and their presence did not interfere with the correct identi®cation of G to A or G to T mutations in codon 12. These results provide the basis for a sensitive and speci®c method to detect c-K-ras codon 12mutated DNA at levels below 10±12% of wild-type DNA. INTRODUCTION A single base change in DNA can lead to altered cellular behavior, inducing oncogenesis, if crucial genes are modi®ed. Mutations in the c-K-ras gene contribute to the development of colon cancer, and such mutations seem to appear early in the colon tumorigenesis pathway (1,2). Detection of such mutations at low levels would be bene®cial for early diagnosis,
prognosis and the evaluation of therapeutic outcome in cancer treatments (3). The possibility of low-frequency mutation detection is of particular interest for early detection of malignant diseases, for evaluation of surgical margins and for monitoring relapse. The c-K-ras gene is commonly mutated in human cancers, especially in lung and colon tumors (4,5). c-K-ras point mutations have been described at codons 12, 13, 59 and 61 (6,7) and, depending on the location and the kind of mutation, histopathological differences have been described in animal models (8) and humans in different kinds of cancer (6,9). Most K-ras mutations are localized at codon 12 and they have been associated with tumor progression and shortened patient survival in colon (10) and non-small cell lung cancer (NSCLC) (11), although, in some cases, such association has not been detected. At the moment, a number of methods for detecting single base mutations in DNA sequences have been described. Those methods include the ampli®cation refractory mutation system (12), single-strand conformation polymorphism (SSCP) and allele-speci®c ampli®cation (13), mismatch ampli®cation mutation assay (14), oligonucleotide ligation assay (15), ligation chain reaction (16), enriched PCR (17), PCR primerintroduced restriction analysis with enrichment of mutant alleles (18), PCR-based assays (19), point mutation detection using exonuclease ampli®cation-coupled capture technique (20), restriction fragment length polymorphism (RFLP) (7,21) and denaturing gradient gel electrophoresis (DGGE) (22). These techniques have several disadvantages for automating DNA diagnosis, which in some cases include the requirement for various hybridization conditions, the use of two PCR ampli®cations and, ®nally, the need for electrophoresis. No ideal method for diagnosis, in which minimal steps are required to give maximal sensitivity, is in universal use for cases of known mutations in a single codon. Previous studies have shown that mass spectrometric analysis of PCR products is feasible for detecting point mutations in DNA (23). Several mass spectrometric approaches for analyzing short oligonucleotides have been used: electrospray ionization mass spectrometry (ESI-MS) (24) and
*To whom correspondence should be addressed at present address: Department of Pathology, Hospital Vall d'HebroÂn, Paseo Vall d'HebroÂn 119±129, 08035 Barcelona, Spain. Tel: + 34 934894169; Fax: +34 932746818; Email: [email protected]
Nucleic Acids Research, Vol. 32 No. 5 ã Oxford University Press 2004; all rights reserved
e53 Nucleic Acids Research, 2004, Vol. 32, No. 5 matrix-assisted laser desorption ionization time of ¯ight mass spectrometry (MALDI-TOF) (25±27) by using puri®ed or intact PCR products (28). Short oligonucleotide mass analysis, previously described as SOMA (23), an approach which uses mass spectrometric analysis of PCR-ampli®ed oligonucleotides, has previously been successfully used to selectively and sensitively detect variant sequences involving a single base change at a codon, for example codon 1307 of the APC gene (26), codon 249 of the p53 gene (29±31) and certain polymorphisms with a prognostic value (27). This technique exploits the utility of mismatched bases within primer sequences described for the PCR approach. PCR ampli®cation and subsequent restriction digestion generate small (7- to 15mer) DNA sequences of wild-type and/or mutated genomic DNA in suf®cient quantities for sensitive and speci®c sequence analysis by ESI-MS/MS. The validity of the mutant DNA presence given by ESI-MS/ MS was con®rmed by a previously described PCR strategy using BstN1 (32±34). This strategy, known as restriction endonuclease-mediated selective PCR, exploits the thermostable properties of BstN1 and Taq polymerase, allowing enhanced ampli®cation of mutant K-ras, when compared with a PCR where no mutant enhanced ampli®cation has been performed. We used SOMA to selectively detect two c-K-ras codon 12 mutant sequences in DNA from 121 colon adenocarcinomas. In this study, we provide detailed information on how the SOMA method can also be used to detect speci®c mutations even when multiple mutations are possible in a codon, and we compared these results with sequencing. However, looking solely at the two most common mutations in a single codon, as we did, does not rule out other mutations at the same codon or other mutations at other codons, such as 13, 59 or 61 in the case of K-ras (6±8). We raise the importance of the SOMA method as a technique that can be used for routine diagnosis in clinical practice.
MATERIALS AND METHODS Samples Samples of tumor biopsy DNA from patients diagnosed with colon adenocarcinoma had been collected at the Department of Pathology at Clinica Puerta de Hierro (Madrid) in relation to a previous study involving other genetic alterations (35). SOMA PCR PCR primers which amplify a 93 bp DNA sequence around codon 12 are shown in Figure 1. Both the 43mer forward (F1) and 42mer reverse (R1) primers introduce the six-base recognition sequence CTGGAG for the type II restriction enzyme BpmI (New England Biolabs, Beverly, MA) 17±22 and 15±20 bases, respectively, from the 3¢ end of each primer. PCR was performed with Taq-Platinum (Invitrogen, Cergy Pointoise, France) in a 25 ml total volume using 150 ng of genomic DNA. Thermal cycling conditions for all primer sets were 95°C for 2 min, followed by 35 cycles of 95°C for 30 s, 55°C for 30 s and 72°C for 30 s. (Perkin Elmer Thermal Cycler).
PAGE 2 OF 8
Restriction digestion of this ampli®ed DNA with BpmI produces an 8mer DNA sequence (Fig. 1, inset) with the codon 12 sequence located at the 5¢ end of the sense oligonucleotide. Ampli®cation and restriction digestion are independent of the eight-base sequence so a mixture of wild-type and mutated DNA sequences is produced. The restriction enzyme BstN1 is commonly used to enhance K-ras codon 12 mutation (32±34). Forward primer (F2), a modi®cation of primer F1 involving replacement of a G with a C three bases from its 3¢ end (Fig. 1), produces ampli®ed DNA containing a CCÃWGG recognition sequence for the restriction enzyme BstNI (New England Biolabs). As it is heat stable, BstN1 can be included in the PCR mix (2 U per reaction) to destroy wild-type DNA during each ampli®cation cycle, enhancing ampli®cation of mutated over wild-type DNA. Thermal cycling conditions using BstN1 were the same as those described above except that the elongation step was done at 60°C. Preparation and puri®cation of SOMA oligonucleotides Following PCR ampli®cation, DNA was cleaned up by phenol/chloroform extraction and ethanol precipitation by adding one-third volume of 7.5 M ammonium acetate (to reduce the level of sodium adduct ions) and 6 vol of ethanol, and holding the solution at ±80°C for 1 h. A 2 ml aliquot of See-DNA (Amersham, Orsay, France) was added to the PCR products to enhance precipitation. DNA was precipitated after centrifugation and the resulting pellet was digested with 8 U of BpmI for 2 h at 37°C in a ®nal reaction volume of 50 ml. This procedure produces an 8mer DNA sequence (Fig. 1). The 8mer DNA sequence was then puri®ed using 1 vol of phenol/ chloroform/isoamyl alcohol (25:24:1 by vol) (Sigma, St Louis, MO) and precipitated again in the presence of 2 ml of SeeDNA, one-third volume of 7.5 M ammonium acetate and 6 vol of ethanol. After washing the colored pellet with 70% ethanol, samples were allowed to air dry. Before HPLC-MS analysis, samples were resuspended in 6 ml of a solution of aqueous 0.4 M 1,1,1,3,3,3-hexa¯uoro-2-propanol (HFIP) (Sigma) and methanol (85:15, v/v). HPLC conditions Before introduction into the mass spectrometer, oligonucleotides were further puri®ed by HPLC, carried out at ~20 ml/ min on a 15 cm 3 800 mm i.d. Vydac C-18 reversed phase Ê pore size) (LC Packings, Amsterdam, column (5 mm, 300 A The Netherlands). To obtain this ¯ow rate, an HP1100 pump (Agilent, Palo Alto, CA) operating at 0.32 ml/min was connected to an LC Packings Accurate micro¯ow splitter. HPLC solvents were prepared from a stock solution of aqueous 0.8 M HFIP, adjusted to pH 7.0 with triethylamine, then diluted to 0.4 M (with water for solvent A and methanol for solvent B) as described previously (24). An initial mobile phase concentration of 35% B was held for 2 min, then programmed to 65% B in 9 min. The entire 5 ml sample was injected onto the HPLC under these initial conditions, as the oligonucleotides are concentrated at the head of the column and low molecular weight salts and other impurities are washed away.
PAGE 3 OF 8
Nucleic Acids Research, 2004, Vol. 32, No. 5 e53
Figure 1. Forward (F1 and F2) and reverse (R1) SOMA PCR primers that were used. Primers amplify a 93 bp DNA fragment in the K-ras gene. The inset shows the two wild-type 8mer SOMA oligonucleotides which are produced by restriction digestion of the ampli®ed DNA with the enzyme BpmI. p indicates a 5¢-phosphate group. Bold letters indicate the mutation introduced in the primer to amplify the DNA.
Mass spectrometric analysis of SOMA oligonucleotides Mass chromatograms were obtained on a LC Quattro mass spectrometer (Micromass, Manchester, UK) equipped with an electrospray ionization source operated in the negative ionization mode. The electrospray capillary was held at ±5 kV and the cone potential was typically 77 V. Ion source temperature was 100°C and the electrospray desolvation temperature was 215°C. Argon pressure in the collision cell was ~5 3 10±3 mBar. As the 8mer DNA elutes into the mass spectrometer, it dissociate into a pair of sense and antisense 8mer oligonucleotides for each wild-type (12GGT) or mutated sequence selected. Electrospray ionization of such an 8mer oligonucleotide produces a series of multiply charged ions. For example, the negative electrospray mass spectrum for the wild-type sense (WT-s) K-ras oligonucleotide (Fig. 2a) shows a major peak for the [M-2H]2± ion at m/z 1279.6. A smaller signal is present for the triply charged [M-3H]3± ion at m/z 852.8; the singly charged [M-H]± molecular ion at m/z 2560.6 is absent. To maximize the sensitivity of the method, HFIP was added to the HPLC mobile phase to increase the relative intensity of doubly charged ions and to minimize sodium adduction (29). To obtain sequence information, the ®rst sector of the tandem mass spectrometer is set to selectively pass ions with a mass-to-charge ratio corresponding to the [M-2H]2± ion of a chosen oligonucletide. The beam of ions is then collided, at an energy of 37 V, with a curtain of argon gas to produce a spectrum of sequence-speci®c CID fragment ions (Fig. 2b) which are analyzed by the second sector of the tandem mass spectrometer. An ax±Bx series of fragment ions is formed from the 5¢ end of the oligonucleotide and a wx series of ions from the 3¢ end (Fig. 2c). Fragment ions speci®c for wild-type and mutated sense and antisense oligonucleotides are monitored as the compounds elute into the mass spectrometer. For speci®c detection of aspartate and serine mutation (Fig. 3), the mass spectrometer was programmed to acquire data in the multiple reaction monitoring mode monitoring ®ve sequence-speci®c [M-2H]2± ion fragments for the sense (s) and antisense (as) SOMA oligonucleotides: 12GGT(WT)-s, 1279.6®1139.2 or 650.4
(both fragment ions are sequence speci®c); 12AGT-s/12GAT-s, 1271.8®1123.7 or 650.4 (either sequence can produce this fragment ion); 12AGT-as, 1236.3®963.6; 12GAT-as, 1236.3®659.4; and 12GGT(WT)-as, 1228.8®659.4. Sequencing Primers used for PCR ampli®cation of a region of exon 1 of the K-ras gene which includes codon 12 produced a 240 bp DNA fragment (sense, 5¢-CCTTATGTGTGACATGTTC-3¢; and antisense, 5¢-CTCATGAAAATGGTCAGAG-3¢). The reaction was carried out in a ®nal volume of 20 ml with a primer concentration of 0.25 mM under the following PCR conditions: 96°C for 5 min, followed by 35 cycles of 96°C for 30 s, 55°C for 20 s and 60°C for 4 min. Samples from the PCR were puri®ed with QIAquick gel extraction columns (Qiagen, Courtabeuf, France). The sequencing reaction was based on the rhodamine terminator using ¯uorescent ddNTPs. Samples were run on a 4.8% polyacrylamide gel (250 ml of APS 10% + 35 ml of TEMED) during 3.5 h at 3000 V and 51°C and read using a Prism 377 DNA sequencer (Applied Biosystems, Foster City, CA). Comparison of SOMA with sequencing Sequencing and SOMA require approximately the same time to perform. Both methods involve: DNA extraction, PCR ampli®cation, puri®cation and injection into the mass spectrometer (10 min per sample) or sequencer (30±60 min per sample). SOMA requires an additional restriction digestion after the PCR ampli®cation, which increases the analysis time by 1±2 h. The cost of consumable items for SOMA and sequencing is comparable; the cost of the tandem mass spectrometer used in this study is about twice that of a capillary electrophoresis sequencer. RESULTS AND DISCUSSION Comparison of SOMA with sequencing The major objective of this work was to show that the SOMA technique can detect speci®c point mutations of well-known genes, such as codon 12 of the K-ras gene, even when present at levels below 10±12% of wild-type DNA (Fig. 3). Detection
e53 Nucleic Acids Research, 2004, Vol. 32, No. 5
PAGE 4 OF 8
Figure 2. (a) Negative electrospray mass spectrum of the wild-type-s oligonucleotide shown in Figure 1. The mass of this peak depends on the number of G, C, A and T bases present in the oligonucleotide. It gives no information about sequence. Under the solvent conditions described herein, the [M-2H]2± ion peak at m/z 1279.6 is the most intense ion, and sodium and other metal-adduct ions (see inset) are kept to a minimum. (b) CID mass spectrum of the [M-2H]2± oligonucleotide ion (m/z 1279.6) shown in Figure 1. Fragment ion peaks in the mass spectrum give oligonucleotide sequence information. (c) The [an±Bn]± series of fragment ions gives sequence information about the 5¢ end of the molecule; the wx± series of fragment ions gives sequence information about the 3¢ end.
of a speci®c mutation by SOMA does not interfere with the detection of other mutations when the latter occur in the same codon. Tumor biopsy DNA samples from 121 patients diagnosed with colon adenocarcinoma were analyzed by SOMA at codon 12 of the c-K-ras gene, and controlled by sequencing in separate PCRs. We focused on differentiating the aspartate from the serine mutation versus other variants occurring at codon 12 of the K-ras gene (Fig. 3). In other variant groups, we considered wild-type and/or other mutants less frequent than aspartic or
serine at codon 12. We performed the study in such a manner because we focused on the predominant mutational pattern as described in the literature for K-ras (36,37). Under these conditions, by SOMA we detected four samples with the 12AGT mutation, 26 samples with the 12GAT mutation, versus 91 samples that were considered as other variants (wild-type and/or other mutations). After sequencing, we detected four samples containing the 12AGT mutation and 23 samples containing the 12GAT mutation instead of the 26 previously detected by SOMA.
PAGE 5 OF 8
Nucleic Acids Research, 2004, Vol. 32, No. 5 e53
Figure 3. SOMA mass chromatograms illustrating the speci®c detection of 12AGT and 12GAT variants in the DNA from eight tumor samples shown by sequencing to contain (a) wild-type DNA, (b) wild-type and 12AGT DNA and (c) wild-type and 12GAT DNA. Five CID fragment ions were monitored for each DNA sample.
Moreover, sequencing revealed 16 samples containing other mutations at codon 12, and ®nally eight samples contained a G®A mutation at codon 13 (13GAC) versus 70 wild-type samples. Detection of low levels of mutated DNA by SOMA An excess amount of wild-type DNA may hamper mutant DNA detection. Different approaches involving PCR, as mentioned above, have been considered to avoid wild-type DNA interference for an accurate detection. For three tumor DNA samples out of 121, 12GAT mutant DNA was detected by SOMA but was not con®rmed by sequencing (Fig. 4). Experiments were done in triplicate. In the SOMA method, F1 and R1 primers were used (Fig. 1). We hypothesized that when the 12GAT mutation present in the DNA extracted from tumor tissue is