Cancer research would greatly benefit from technologies that allow simultaneous screening of several unknown gene mutations. Lack of such methods currently hampers the large-scale detection of genetic alterations in complex DNA samples. We present a novel mismatch-capture methodology for the highly efficient isolation and amplification of mutation-containing DNA from diverse nucleic acid fragments of unknown sequence. To demonstrate the potential of this method, heteroduplexes with a single A/G mismatch are formed via cross-hybridization of mutant(T→G) and wild-type DNA-fragment populations. Aldehydes are uniquely introduced at the position of mismatched adenines via the Escherichia coli glycosylase, MutY. Subsequent treatment with a biotinylated hydroxylamine results in highly specific and covalent biotinylation of the site of mismatch. For PCR amplification,synthetic linkers are then ligated to the DNA fragments. Biotinylated DNA is then isolated and PCR amplified. Mutation-containing DNA fragments can subsequently be sequenced to identify type and position of mutation. This method correctly detects a single T→G transversion introduced into a 7-kb plasmid containing full-length cDNA from the p53 gene. In the presence of a high excess wild-type DNA(1:1000 mutant:normal plasmids) or in the presence of diverse DNA fragment sizes, the DNA fragments containing the mutation are readily detectable and can be isolated and amplified. The present Aldehyde-Linker-Based Ultrasensitive Mismatch Scanning has a current limit of detection of one base substitution in 7 Mb of DNA and increases the limit for unknown mutation scanning by two to three orders of magnitude. Homozygous and heterozygous p53 regions (G→T,exon 4) from genomic DNA are also examined, and correct identification of mutations is demonstrated. This method should allow large-scale detection of genetic alterations in cancer samples without any assumption as to the genes of interest.
Screening human tissues for mutations and polymorphisms is of major interest in biomedical research because a significant portion of inherited and acquired human diseases is attributed to such genetic alterations. Generation and accumulation of mutations is accepted as a necessary event in malignant transformation and carcinogenesis(1–5). Base substitution mutations in particular result commonly from exposure to mutagens (5) and are believed to play a major role in the formation of several cancers. p53 mutations have been found in 50% of all cancers, and 75% of these are base substitutions (6). Inherited polymorphisms may also play a major role in cancer formation (4, 6). Despite accumulated knowledge, the generation of transformed cells and their progression to malignancies are likely to involve multiple mutations (4). Thus, it remains unclear whether the limited number of genetic changes identified thus far are the driving events in the formation of the genetically unstable, invading phenotype or whether crucial,unidentified mutations generated upstream are also involved(7).
Understanding the dependence of carcinogenesis on specific mutations is contingent on the availability of technologies that can sensitively and effectively screen cancer samples for unknown mutations in several fragments (genes) simultaneously. The ability to detect mutant alleles in the presence of an excess normal alleles, which often“contaminate” cancer samples, is another requirement that must be satisfied for screening mutations in cancer (8). Although powerful methods are available to screen short sequences for known mutations (9), most current methods for identifying unknown mutations remain limited to DNA fragments of a single size,typically <1 kb (10), and require a high proportion of mutant DNA in the sample for adequate sensitivity.
We present a novel approach for mutation detection (Fig. 1) that addresses several of the requirements described above. The method combines: (a) the specificity of mismatch repair glycosylases for faithful mismatch detection; (b) the high selectivity of the biotin-avidin interaction for isolation of biotinylated mutant DNA fragments; and (c) the sensitivity of PCR to enable identification of the isolated mutant DNA. After formation of heteroduplexes using “control” and “mutant” DNA samples, mismatched adenines are excised via MutY glycosylase, in a process that generates an unsaturated open chain aldehyde on the sugar,at the position of the apyrimidinic site. A biotinylated hydroxylamine(ARP3) is then covalently reacted with the aldehyde, and the biotinylated DNA is linker ligated and separated from nonbiotinylated DNA using streptavidin-coated microspheres. The microspheres are then directly used in a PCR reaction, which proceeds using the intact DNA strand that is complementary to the biotinylated strand held on the microspheres. Using this novel approach, isolation of mutated fragments does not require prior knowledge of the sequence, can be performed in the presence of 1000-fold excess normal alleles, and can be carried out for large (≥7 kb) and non-uniform DNA fragment sizes simultaneously.
MATERIALS AND METHODS
DNA Test Systems and Formation of Heteroduplexes.
Plasmids (7091-bp long) containing wild-type and mutant (1336T→G transversion, a serine→alanine change) full-length human p53 sequence(1843 bp) was produced via site-specific mutagenesis, via a modification of the procedure described previously (11). Briefly, human wild-type p53 cDNA was inserted into the BamHI site of the expression vector pcDNA3.1(+) (Invitrogen,CA). To generate the S378A mutation, p53 was subcloned into the BamHI site of pAlter-1 and mutagenesis carried out using the Altered Sites II mutagenesis system (Promega Corp., Madison, WI). The sequence of the mutagenesis oligonucleotide was: GGG TCA GTC TAC CGC CGC CCA TAA A. Mutations were conformed by sequencing, and the p53 cDNA was then subcloned into pcDNA3.1(+).
Prior to heteroduplex formation, mixtures of mutant plasmid (mutp53)and wild-type plasmid (wtp53) were formed at desired ratios in Sau3A1 buffer (10 mm NaCl, 1 mm Bis Tris Propane-HCl,1 mm MgCl2, and 0.1 mmDTT, pH 7.0), and samples were digested with Sau3A1 enzyme (New England Biolabs) for 2 h at 37°C. Digestion lead to formation of 31 double-stranded fragments with a wide distribution of sizes up to∼1000 bp, with AGCT ends. After digestion, the mixture was heated at 94°C for 2 min, gradually cooled down, and incubated for 1 h at 65°C and then allowed to cool slowly to room temperature. The sample solutions were then treated with 30 mm freshly prepared hydroxylamine (pH 7.0, for 1.5 h at room temperature) to remove traces of spontaneously forming aldehydes in DNA (12, 13). After incubation, the samples were phenol:chloroform extracted, ethanol precipitated, and resuspended in the buffer of choice.
As a test system for chemiluminescence studies, 5′ fluoresceinated double-stranded oligonucleotides with or without a centrally located A:T→ C:G base substitution were synthesized (Oligos Etc; (+)strand sequence: fluorescein-5′-GTC TCC CAT CCA AGT ACT AAC CAG GCC CGA CCC TGC TTG GCT TCC GAT T-3′). For cross-hybridization of mutant and wild-type sequences, equimolar amounts (∼0.3 μg) of each oligonucleotide were mixed in 40 mm Tris-HCl (pH 7.5), 20 mmMgCl2, and 50 mm NaCl and then heated at 94°C for 2 min and gradually cooled down and incubated for 1 h at 65°C to form heteroduplexes. The oligomers were then treated with hydroxylamine as described above.
Genomic DNA from one patient known to harbor a heterozygous mutation (first sample, G→T transversion at codon 110, exon 4 of p53)or from a second patient with no mutation (second sample), was used to apply ALBUMS in screening genomic DNA. The presence or absence of the mutations in exon 4 had been demonstrated via sequencing at the Dana-Farber Cancer Institute Molecular Diagnostics Laboratory. To amplify and purify exon 4 for ALBUMS-screening, PCR with a high-fidelity polymerase (Advantage HF-2; Clontech) was used. The 20-mer primers used for PCR amplification of the 399-bp region were:5′-CAA CGT TCT GGT AAG GAC AA-3′ (forward) and 5′-GCC AGG CAT TGA AGT CTC AT-3′ (reverse). PCR thermocycling conditions were: 94°, 30 s; (94°, 20 s/65°, 20 s/68°, 20 s) × 10 cycles,with annealing temperature decreasing 1°/cycle (touchdown PCR);(94°, 10 s/55°, 20 s/68°, 20 s) × 25 cycles;68°, 6 min; 4°; Hold. After amplification, the PCR product was gel-purified to remove the small amount of genomic DNA (QIAquik gel extraction kit; Qiagen, Inc.). Heteroduplex formation was performed as described above for the plasmid system.
Verification of Biotinylated Aldehyde-reactive Probe Reactivity with MutY-treated Heteroduplexes.
Heteroduplexes were resuspended in MutY buffer (10 mmHEPES-KOH, 0.1 mm KCl, and 10 mm EDTA, pH 7.0)prior to incubation with MutY (Trevigen) using 1 unit of enzyme/μg DNA, at 37°C for 1 h. After MutY treatment, the sample was incubated with 5 mm biotinylated ARP (14),a biotinylated hydroxylamine, from Molecular Probes, for 30 min at room temperature. Unreacted ARP was removed by microbiospin-6 column(Bio-Rad) filtration, and the samples were fluoresceinated by a 1-h exposure to a commercially available fluoresceination reagent(Fluorescein Label IT reagent; 1 μl reagent/μg DNA in 4-morpholinepropanesulfonic acid buffer, pH 7.5, at 37°C). Excess reagent was then removed by microbiospin filtration. In experiments using the synthetic prefluoresceinated heteroduplexes, the fluoresceination step was omitted. The doubly-labeled (ARP and fluorescein) DNA fragments were then immobilized onto neutravidin-coated microplates (Pierce) in the presence of 5 nm anti-fluorescein-Fab-alkaline phosphatase (Boehringer),and chemiluminescence was performed using an Intensified Charge Couple Device camera (Princeton Instruments) as described (13). Experiments were repeated three to five times.
Generation of Random Aldehyde-containing DNA Fragments.
To test the ability of the method to isolate and amplify DNA fragments that contain aldehydes, aldehyde-containing apurinic/apyrimidinic sites were artificially introduced to DNA prior to reaction with 5 mm ARP. For this purpose,Sau3A1-digested wild-type plasmids were treated with hydroxylamine,extracted in phenol-chloroform, and precipitated in ethanol. The samples were then depurinated via exposure to sodium citrate (pH 3.5)at 38°C for different time periods (15). The depurination was stopped by raising the pH to 7.0. The reaction mixture was then purified via microbiospin-6. Experiments were repeated three times.
Isolation, PCR Amplification, and Sequencing of Mutation-containing DNA.
After MutY treatment or acid depurination and exposure to ARP, the DNA fragments were extracted in phenol-chloroform and precipitated in ethanol. Samples were then resuspended in 60 μl of ligation buffer for ligation of asymmetric linkers corresponding to Sau3A1-restriction sites, following the protocol of Hubank and Schatz (16). The linkers used were a 24-mer, 5′-AgCACTCTCCAgCCTCTCACCgCA-3′ and a 12-mer, 5′-gATC TgCggTgA-3′. The DNA sample was mixed with the linkers,annealed to 50°C, and then cooled down to 10°C. Three μl of T4 DNA ligase (400 units/μl) were then added and incubated overnight at 15°C. To isolate biotinylated from nonbiotinylated DNA fragments, the ligase-treated samples were mixed with streptavidin-coated magnetic beads (Dynal) and gently shaken for 3 h at room temperature. After thorough washing, the microspheres were suspended in 20 μl of 60 mm Tris-HCl, 15 mm ammonium sulfate, and 3.5 mm MgCl2 (pH 8.5) and used directly in a 50-μl PCR reaction, using the 24-mer oligonucleotide as a primer. Amplification was then carried out in a Perkin-Elmer Gene-Amp PCR 9600 system, as described (16): 3′ incubation at 72°C, followed by addition of Taq polymerase and incubation 5′ at 72°C. Twenty cycles of 1 min at 95°C and 3 min at 72°C were then performed. After a final extension for 10 min at 72°C, the products were held at 4°C overnight. The PCR products (70% of total product)were analyzed in a 1.5% horizontal agarose gel (3 h at 100 V) and stained with 1 μg/ml ethidium bromide. Experiments were repeated three to five times. Sequencing of PCR products was carried out by excising DNA from the agarose gel (Geneclean II kit; Bio101) and processing them on an automated sequencer.
To examine the presence of mutations in a specific genetic region, the linker-ligation step in the above protocol can be omitted. After isolation of ARP-biotinylated DNA molecules with streptavidin-microspheres, PCR primers for the desired genetic region were added, followed by amplification. To apply this approach for the plasmid system, PCR for a 236-bp region encompassing the known T→G mutation in the plasmid was used after binding to streptavidin-microspheres (5′-ACT CAA GGA TGC CCA GGC TG-3′ forward primer, and 5′CCT ATT GCA AGC AAG GGT TC-3′ reverse primer; PCR cycling: 95°C, 1 min and then (95°C 30 s; 68°C, 1 min) for 25 cycles using Advantage-2 polymerase (Clontech); finally, 72°C 5 min, and hold at 4°C). The presence of mutations in the isolated p53-exon 4 from genomic DNA was similarly detected using the PCR conditions described above. Thus, after incubation with microspheres,PCR was applied using the same primers used for exon 4 during amplification from genomic DNA. These experiments were repeated twice.
Verification of ARP Reactivity with MutY-treated Heteroduplexes.
For a verification of the ARP binding to MutY-recognized A/G mismatches, a chemiluminescence methodology was implemented. When MutY acts on synthetic heteroduplexes formed as a result of cross-hybridization of A:T→C:G mutant and wild-type alleles, strong chemiluminescence signals were obtained (Fig. 2,A, Lane 3). When MutY or ARP were omitted,almost background signals were generated (Fig. 2,A, Lanes 1and 2). Similar to a fluoresceinated aldehyde reactive probe on which we reported earlier (13), ARP also appears to react with the MutY-generated aldehydes at positions of A/G mismatches that are presumed to form. Strong chemiluminescence signals were also obtained when heteroduplexes derived from cross-hybridization of plasmid digests were MutY treated (Fig. 2,A). Therefore,chemiluminescence can provide signals relating to the overall T→G mutations in a sample consisting of diverse DNA fragments. The chemiluminescence signals are proportional to the amount of DNA applied on the microplates (Fig. 2 B, inset).
Isolation and PCR Amplification of Aldehyde-containing DNA Fragments.
To validate the procedure (Fig. 1) for the quantitative isolation and PCR amplification of ARP-reacted DNA fragments, wtp53 plasmid was first digested with Sau3A1 into 31 DNA fragments of varying sizes, ligated to synthetic linkers, and PCR amplified. The PCR products (Fig. 3,B, Lanes 1 and 2) demonstrated similar bands as the original digest (Fig. 3,A, Lane 2). In the absence of primers or DNA, no PCR product was obtained (not shown). Next, after Sau3A1 digestion the DNA was exposed for short periods to mild acidic conditions. This exposure introduces stoichiometric amounts of aldehyde-containing apurinic/apyrimidinic sites in DNA (13, 15), which are known to couple specifically to ARP (14, 17). ARP-containing fragments were then trapped on streptavidin-coated microspheres and PCR amplified. Fig. 3,C demonstrates a gradual increase in the PCR products as a function of depurination. When ARP was omitted, no PCR products were obtained (Fig. 3,C, Lane 5). In addition,the PCR products contain bands similar to the original plasmid digest(Fig. 3 A, Lane 2), which is consistent with the expected (5, 12, 15) random generation of aldehydes during depurination.
Isolation and PCR Amplification of T→G Transversion-containing DNA Fragments.
To demonstrate isolation and amplification of T→G transversions in diverse DNA fragments of unknown sequence, mutant (T→G) and wild-type(T-containing) plasmids were Sau3A1-digested and cross-hybridized to form A/G and T/C-containing heteroduplexes. The sample was then MutY-treated and biotinylated with ARP. Fig. 4,A depicts the PCR products obtained when heteroduplexes(mutant + wild type) or homoduplexes (wild type + wild type) were used. A strong band is evident for the heteroduplexes,whereas no such band is present for the homoduplexes. The PCR product profile is strikingly different from the one in Fig. 3, where all DNA fragments were present in the PCR product. Examination of the p53 sequence revealed two Sau3A1 digestion sites flanking the mutated thymidine. Therefore, including the added linkers, the mutation-carrying fragment should be ∼602 bp (554 bp plasmid + 48 bp linkers), which agrees with the data in Fig. 4,A. To further determine the identity of the resulting fragment, the DNA was recovered from the gel and sequenced in parallel with the original wild-type and mutant sequences. Sequencing clearly demonstrated (Fig. 4 B) the recovery of the mutation-containing sequence out of a mixture of diverse DNA fragment sizes, via ALBUMS.
Next, the selectivity of the method was tested by repeating the procedure in the presence of decreasing mutant:wild-type plasmid ratios(1:1, 5:95, 1:99, and 1:1000). These plasmid mixtures were Sau3A1 digested to generate complex DNA fragment populations,cross-hybridized, and processed for isolation of mutants. Fig. 5 demonstrates that ALBUMS can isolate and amplify the mutation-containing fragment when one mutant allele is present in up to 1000-fold excess of wild-type alleles in the original plasmid mixtures.
Detection of the Presence of a Mutation in a Specific Sequence.
To demonstrate correct identification of a mutation in a specific region, a 236-bp fragment encompassing the known T→G mutation in the plasmid was amplified out of the ALBUMS-selected DNA fragments. Fig. 6,A demonstrates the isolation of a 236-bp fragment when heteroduplex DNA populations (1:1, mutant:wild type) are formed(Lane 1) but not when homoduplex DNA (wild-type alone) is used (Lane 2). Next, a 399-bp region (p53 exon 4, 449 bp including the primers) amplified from patient genomic DNA samples were similarly examined. Fig. 6,B, Lanes 1 and 2, demonstrates the PCR product amplified from genomic DNA from the two patient samples. These two genomic DNA samples had been precharacterized via sequencing at the DFCI Molecular Diagnostics Laboratory and were found to be heterozygous at codon 110 (CGT→CTT,first sample) and homozygous (second sample), respectively. Fig. 6 C demonstrates the application of ALBUMS on these two PCR products. It can be seen that, after ALBUMS, a PCR product is obtained for the heterozygous (Lane 1) but not for the homozygous(Lane 2) sample.
Improvement in Unknown Mutation Detection.
The ability to simultaneously screen several DNA fragment sizes currently eludes most methods for unknown mutation detection (8, 9). Sequencing via hybridization on DNA microarrays (DNA chips;Ref. 18) allows the single-step sequencing of up to 15–30 kb arbitrarily chosen DNA; however, the method cannot currently handle the presence of an excess of normal alleles, which is often found in tumor samples. In addition, the sequences to be examined for mutation have to be known in advance and tiled on the microarray, so that a special chip must be built for each set of interrogated sequences.“Capture” methods, such as those that use the MutS enzyme for immobilization of mismatches (19), in principle can bind mismatches in numerous DNA fragments simultaneously without prior knowledge of the sequence. However, as reported earlier (9, 19) and from our own experience,4nonspecific binding of DNA poses limits to the selectivity of this approach.
The present work demonstrates a major improvement in the central problem of unknown mutation detection, i.e., the ability to scan unknown and diverse DNA sequences for isolation and amplification of a rare mismatch (“finding the needle in a haystack”). In this novel approach, after the formation of heteroduplexes, the high specificity of MutY for mismatched adenine is used to uniquely introduce aldehydes at positions of mutations. Previously, radioactive,biotinylated, or fluoresceinated hydroxylamines had been shown to efficiently trap aldehyde-containing abasic sites (12, 17, 20, 21). Furthermore, our group has demonstrated that a fluoresceinated aldehyde-reactive probe resulted in sensitive chemiluminescent detection of MutY-generated total aldehydes in a DNA sample (13). However, identification of the mutant gene is of paramount importance; hence, methods to isolate, amplify, and identify the mutation are urgently needed. To this end, in this work a procedure was devised to isolate, PCR amplify, and characterize the mutations. A biotinylated ARP (14, 17) was used to introduce biotin at the aldehydes formed after MutY-mediated excision of adenine from mismatched adenine:guanine pairs. After immobilization of biotinylated DNA on streptavidin support, the intact (complementary)strand serves as the template for PCR. The MutY-generated aldehydes are the only reactive sites for ARP (traces of spontaneously occurring aldehydes are removed via hydroxylamine pretreatment). Unlike electrophoresis or chromatography-based methods, ALBUMS relies on the avidin-biotin interaction for mutation isolation and detection and not on DNA fragment size discrimination. Consequently, a problem that traditionally restricted the use of enzymes for mutation detection,nonspecific strand cutting by the enzyme used or by contaminating nucleases (10), is of no consequence to the present approach. Fig. 5 demonstrates that a single base substitution introduced into a 7-kb plasmid is still detectable when the mutant:wild type ratio is 0.1%. By increasing the PCR cycles and by overexposing the gels, the PCR product in the heteroduplexes becomes stronger (not shown); however, nonspecific bands start appearing in the homoduplexes, as well. Because under the conditions adopted in this paper the 0.1% mutant dilution is only just detectable, in practice a 0.1–1% mutant:wild type ratio should be taken as the current limit for ALBUMS. Conventional methods rarely can detect 1 mutant allele in the presence of 10 wild-type alleles(22); therefore ALBUMS extends the detection limit by a factor 10–100. If the comparison to other methods is performed in terms of the lowest frequency of mutated bases detectable in the presence of nonmutated DNA, the result in Fig. 5 indicates that ALBUMS has a detection limit of 1 mutated base in 7 Mb of DNA. Conventional methods for detecting unknown mutations typically detect up to 1 mutated base in 3 kb of DNA (22), whereas DNA chips can detect up to ∼1 mutated base in 15–30 kb of DNA (18). Accordingly, ALBUMS improves the lowest limit for unknown mutation scanning by two to three orders of magnitude. Base substitution mutations that can be detected with such sensitivity are the four transversions (T→G/G→T or A→C/C→A), because A/G mismatches will result whenever DNA heteroduplexes are formed. In addition, MutY recognizes A/C mismatches (23); therefore, (G→A/A→G,C→T/T→C) are also potentially detectable. A number of alternative DNA base changes should also be detectable using the present approach in conjunction with other glycosylases, e.g., the thymidine glycosylase from Thermophilus aquaticus (mainly G/T mismatches but also G/G; Ref. 24) and from HeLa cells(mainly G/T mismatches but also A/A; Ref. 25). The method will currently not detect small deletions or frameshifts because these are not recognized by glycosylases. In addition, homozygous mutations would only be detectable upon addition of a wild-type DNA to serve as a template for heteroduplex formation. In its present format, ALBUMS requires a starting total (mutant plus wild type) DNA of 1 μg. This translates to a minimum requirement for ∼1 ng of mutant DNA in the sample, because 0.1% mutant DNA can be detected. ALBUMS is applicable to any form of DNA (genomic, cDNA, or plasmid), is nonisotopic, and can be adapted to a high throughput format.
The present development allows highly selective scanning for mutations and/or single nucleotide polymorphisms without prejudice as to which are the DNA sequences (genes) of interest. The method relies on the formation of heteroduplexes from a wild-type sample (e.g.,genomic DNA or cDNA, from noncancerous tissue) and a mutated sample(e.g., DNA from cancerous tissue). Unlike other mutation detection methods that use heteroduplex formation (e.g.,denaturing gradient gel electrophoresis, chemical cleavage of mismatch,and denaturing high-precision liquid chromatography; Ref.10), the heteroduplexes in ALBUMS are formed in a complex hybridization of several DNA fragments simultaneously. Once mutated DNA fragments are isolated and amplified via ALBUMS, identification of the corresponding genes is feasible with established methods(e.g., sequencing, cloning, and Southern blot). Large-scale hybridization approaches, such as application on appropriate DNA arrays, may also be envisioned. In this manner, point mutations in several oncogenes, for example, could be screened in a single experiment. Therefore, ALBUMS should enable multiplex screening of cancer samples for mutation-containing genes or for identification of point mutations that promote malignant transformation of cells exposed to mutagens. G→T transversions in particular are the major mutation generated by the tobacco carcinogen benzo(a)pyrene and are found in lung cancers of smokers (26). Because traces of DNA containing G→T can be detected with the present method, G→T transversion scanning in several DNA fragments simultaneously could help identify key mutated genes leading to lung cancer or could develop into an early detection method for such malignancies. Further envisioned ALBUMS applications include genotyping and polymorphism studies and the role of mutations in diseases other than cancer. Practical applications may include screening plasmids in mutagenesis experiments and quantification of PCR errors.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in part by USPHS Grants K04 CA69296 and 1R21/R33 CA83234–01 (to G. M. M.) by the NIH and by a grant from the Starr Foundation.
The abbreviations used are: ARP, aldehyde reactive probe; ALBUMS, aldehyde-linker-based ultrasensitive mismatch scanning.
The assistance of Dr. Judy E. Garber in obtaining the patient samples is gratefully acknowledged.