Mutagenesis, Vol. 17, No. 5, 365-374,
September 2002
© 2002 UK Environmental Mutagen Society/Oxford University Press
Ligation of a primer at a mutation: a method to detect low level mutations in DNA
Department of Radiation Oncology, Dana Farber Cancer Institute, Harvard Medical School, Boston, MA, USA and 1 School of Dentistry, University of California Los Angeles, Los Angeles, CA, USA
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Detection of low frequency mutations following exposure to mutagens or during the early stages of cancer development is instrumental for risk assessment and molecular diagnosis. We present a sensitive new method to detect trace levels of DNA mutations induced within a large excess of wild-type sequences. The method is based on mutation-induced generation of new restriction enzyme recognition sites. A DNA sequence is amplified from genomic DNA or cDNA using a high fidelity polymerase. The purified PCR product is digested with a restriction enzyme that recognizes the newly generated restriction site, partially dephosphorylated and ligated with an oligonucleotide at the position of the mutation. The ligated oligonucleotide is then utilized in two rounds of PCR to amplify the mutated DNA but not the wild-type allele that contains no restriction site. An A
T polymorphism in mRNA (tenascin gene, A2366T, AsnIle) and a GA polymorphism in genomic DNA (Ku gene, G74582A, ValIle), both of which generate a restriction site for the enzyme SAU3A1, demonstrate the application. Eleven patient samples pre-characterized for the G74582A polymorphism in the repair gene Ku are used to demonstrate the reliability of this approach. This technique quantitatively detects the Ku GA polymorphism at a mutant frequency of 1.6x10-6 relative to the wild-type allele. Mutations in p53 that are frequently induced by mutagens can readily be detected using the present method. As an example, using a second enzyme BbvI, a mutation frequently encountered in human cancers (G14154A mutation, p53 codon 245, ArgGln) was detected in patient samples. The process does not require radioactivity, utilizes established procedures and overcomes several factors known to produce false positives in RFLP-based assays. The present amplification via primer ligation at the mutation (APRIL-ATM) has potential applications in the detection of mutagen-generated genetic alterations, early detection of tumor marker mutations in bodily discharges and the diagnosis of minimal residual disease. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Detection of mutated cells carrying altered DNA sequences within a large excess (>104-fold) of unaltered, wild-type cells is useful in the fields of molecular mutagenesis and carcinogenesis (Steingrimsdottir et al., 1996
; van Zeeland and Vrieling, 1999), early tumor detection (Dong et al., 2001) and molecular diagnosis (Parsons and Heflich, 1997; Monteiro et al., 2000; van Houten et al., 2000). Assays for measuring mutations in vivo, for example, are important for the assessment of mutagenic hazards to humans and the mutation load of specific individuals, i.e. the level of mutations acquired following exposure to chemical or environmental agents during their lifetime or following specific incidents. Such mutations are initially present at a very low frequency (for example <10-4; Steingrimsdottir et al., 1996; van Zeeland and Vrieling, 1999) relative to the wild-type allele in the sequences of the exposed tissue and their detection presents a major technical challenge. Yet, the development of sensitive mutagenicity tests is essential if genotoxic agents are to be correctly identified and potential human exposures minimized, where possible (Jenkins et al., 1998). Similarly, for molecular diagnosis early detection of low frequency K-ras mutations in stools correlate well with colon cancer development (Dong et al., 2001), and detection of minimal residual disease (MRD) in distant sites from head and neck tumors might aid in early detection of recurrences that are frequent in this type of cancer (van Houten et al., 2000).
A widely used approach for detection of known mutations is PCR–RFLP (Friedman et al., 1990; Eiken et al., 1991; Bos and Van Mansfeld, 1992; van Mansfeld and Bos, 1992; Parsons and Heflich, 1997; Bazrafshani et al., 2000; Plendl et al., 2001). Following PCR amplification of a sequence from genomic DNA, digestion with a restriction endonuclease generates an agarose gel-resolved extra fragment either only on the wild-type allele or, alternatively, only on the mutant allele (Parsons and Heflich, 1997). This approach has proven to be generally reliable and versatile. The sensitivity of PCR–RFLP, however, is low, as it can only detect mutations when these comprise >1–10% of the amplified sequences, i.e. it is mainly useful for detecting clonally expanded mutations in tumors or frequently occurring polymorphisms (Parsons and Heflich, 1997). Methods that overcome this limitation and can identify mutations in a large excess of wild-type sequences include the restriction site mutation assay (RSM; Parry et al., 1990; Jenkins et al., 1998), the radiolabeled probe assay (Haliassos et al., 1989), MutEx/ACB PCR (Parsons and Heflich, 1998), PCR/LDR (Barany, 1991) and others (reviewed by Parsons and Heflich, 1997; van Houten et al., 2000). Among the most promising approaches are the `enriched PCR' methods, the PCR–RFLP (Felley-Bosco et al., 1991; Pourzand and Cerutti, 1993) and RSM (Parry et al., 1990; Jenkins et al., 1998) assays. These methods rely on the presence of a restriction endonuclease recognition site in the wild-type sequence whose digestion prevents successful PCR amplification of a DNA segment encompassing the site. If, however, a mutation is present in the recognition sequence, PCR is enabled and a measurable product is generated. Using this principle, detection of low frequency mutations in the region 10-4–10-8 have been reported (Parsons and Heflich, 1997; Jenkins et al., 1998). One disadvantage of these approaches is that unless wild-type sequences are 100% digested, false positives can be produced (Parsons and Heflich, 1997; Jenkins et al., 1998).
A new primer ligation PCR-based approach is presented here which combines the reliability of classical PCR–RFLP and the high sensitivity of RSM/PCR–RFLP. The method follows the initial steps of classical PCR–RFLP, i.e. PCR amplification of a sequence from genomic DNA or cDNA is followed by digestion with a restriction enzyme which only cuts when a mutation generates its recognition sequence (Figure 1). Following that, a PCR primer is ligated uniquely at the position of the digested mutation. This approach allows ligation-mediated PCR amplification of the mutated, but not the wild-type, sequences (Figure 1). We have identified conditions that allow this procedure to be performed with minimal or no interference from non-specific ligation of the primer and to achieve quantitative detection of mutations following PCR amplification. It is shown that the method can detect a mutation down to a frequency of 1.6x10-6 relative to the wild-type alleles, for the sequences examined, in a reproducible fashion and without generation of false positives/negatives. We demonstrate detection of polymorphisms and a frequently encountered p53 mutation from patient samples. This method should hold promise for the detection of low frequency mutations, such as those that are induced by mutagens and carcinogens or that accumulate in the early stages of tumor induction.
|
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Preparation of genomic DNA and cDNA
Tissues were obtained from patients operated on for lung adenocarcinomas from the Massachussetts General Hospital Tumor Bank and from the Cooperative Human Tissue Network. Specimens were removed immediately after surgery and frozen in liquid nitrogen. Genomic DNA from frozen lung samples was then extracted using a commercial kit (QIAamp DNA mini kit; Qiagen Inc.). Typically, 30 mg lung tissue yielded 20–30 µg purified genomic DNA. To prepare double-stranded cDNA, mRNA from cultured osteosarcoma SAOS-2 cells was extracted using oligo(dT)-coated magnetic beads (Dynabeads mRNA Direct kit; Dynal, Lake Success, NY) and was used to synthesize cDNA (Universal Riboclone Synthesis; Promega, Madison, WI).
PCR amplification of the Ku gene, the p53 gene and tenascin gene regions
Regions of the Ku gene in genomic DNA and tenascin gene mRNA that contain polymorphisms which generate a new restriction site for SAU3A1 were examined. To amplify a 334 bp Ku region from genomic DNA, a 15 cycle PCR reaction with a high fidelity polymerase (Advantage HF-2; Clontech, Palo Alto, CA) was used. The primers used for PCR amplification were 5'-GGG GAG CAC AAT TTC CCT TC-3' (forward) and 5'-GGA ACT GGA ACT CAA GGC AAG-3' (reverse). PCR thermocycling conditions were: 94°C for 30 s; 10 cycles of 94°C for 20 s, 65°C for 20 s and 68°C for 20 s, with the annealing temperature decreasing by 1°C/cycle (touchdown PCR); 15 cycles of 94°C for 10 s, 55°C for 20 s and 68°C for 20 s; 68°C for 6 min; 4°C; hold. Following amplification, the PCR product was gel purified to remove the small amount of genomic DNA (QIAquick gel extraction kit; Qiagen Inc.). Gel-purified DNA was quantitated with picogreen (Molecular Probes, OR). A similar protocol was also used to amplify a 377 bp tenascin gene fragment from double-stranded cDNA. The primers used were 5'-ACC AGC ACC ATC TTC-3' (forward) and 5'-GAC ACT GCC GCT GTG GTA-3' (reverse). Finally, a 266 bp long p53 region containing exon 7 was amplified from genomic DNA, from a wild-type and from a patient sample, using the same PCR conditions as for the Ku gene. The patient sample contained a GA mutation that generates a BbvI restriction site (5'-GCAGC-3') in codon 245. Primers used were 5'-AAG GCG CAC TGG CCT CAT CTT-3' (forward) and 5'-GAG GTG GAT GGG TAG TAG-3' (reverse). All PCR products were sequenced at the Dana Farber Cancer Institute Molecular Biology Laboratory.
SAU3A1/BbvI digestion
Purified PCR products (300 ng) were digested with 1 U of the 4 bp-cutter enzyme SAU3A1 or the 5 bp-cutter BbvI (New England Biolabs, Beverly, MA). Incubation was carried out for 3 h at 37°C in the buffers supplied by the company. The samples were then purified using a commercial kit (QIAquick columns; Qiagen Inc.).
Dephosphorylation and ligation
To partially remove 5'-phosphate groups from DNA ends, samples were treated with calf intestinal phosphatase (CIP; New England Biolabs) either prior to or, alternatively, following SAU3A1/BbvI digestion. Dephosphorylation was performed in CIP reaction buffer (100 mM NaCl, 50 mM Tris–HCl, 10 mM MgCl2, 1 mM DTT) for 1 h at 37°C. The samples were then purified using QIAquick columns. For ligation of SAU3A1-digested, dephosphorylated samples, DNA was resuspended in 6 µl of ligation buffer (50 mM Tris–HCl, 10 mM MgCl2, 10 mM DTT, 1 mM ATP, 25 µg BSA, pH 7.5) for ligation of asymmetric linkers corresponding to SAU3A1 restriction sites. The linkers 5'-AAC TGT GCT ATC CGA GGG AAT GGA ACC TGA GTC CTC TCA CCG CA-3' (4 µl from a 3.6 mg/ml stock) and 5'-GAT CTG CGG TGA-3' (4 µl from a 1 mg/ml stock) were mixed with the DNA sample, annealed at 50°C and then slowly cooled to 10°C. To this was added 3 µl of T4 DNA ligase (2000 U/µl; New England Biolabs), followed by incubation at room temperature for 15–30 min. A similar protocol was also followed for ligation of BbvI-digested samples except that, because the enzyme cuts outside its recognition site and does not yield a palindromic sequence, two sequence-specific linkers are required for ligation. Each linker is for use with either the forward or the reverse gene-specific primer in the subsequent PCR reactions. For use with the forward primer, the linkers 5'-AAC TGT GCT ATC CGA GGG AAT GGA ACC TGA GTC CTC TCA CCG CA-3' and 5'-CCT CTG CGG TGA-3' were ligated. For use with the reverse primer, the linkers 5'-AAC TGT GCT ATC CGA GGG AAT GGA ACC TGA GTC CTC TCA CCG CA-3' and 5'-GAG GTG CGG TGA-3' were ligated.
The first and second rounds of PCR
For samples containing SAU3A1-specific linkers, the first PCR amplification was carried out in a Perkin Elmer Gene-Amp PCR 9600 system (PE Biosystems, Foster City, CA) using a primer (5'-AAC TGT GCT ATC CGA GGG AA-3') corresponding to the first 20 bases of the 44mer oligonucleotide that was ligated and either the forward or the reverse primer utilized in the original high fidelity amplification from genomic DNA or cDNA. The cycling utilized in conjunction with Titanium Taq polymerase (Clontech) was: 94°C for 30 s; 25 cycles of 94°C for 30 s and 72°C for 60 s; 72°C for 5 min; 4°C; hold. Negative controls (no template DNA) were run each time. For the second round of PCR, 2 µl of the first PCR product were re-amplified, without further purification, using a restriction site-specific primer 5'-CTG AGT CCT CTC ACC GCA GAT C-3' corresponding to the end of the ligated 44mer oligonucleotide plus a 5'-GATC-3' extension, i.e. the SAU3A1-recognition sequence. The same PCR protocol as in the first round PCR was retained. Finally, for samples containing BbvI-specific linkers, similar PCR conditions were applied except that the restriction site-specific primer contained the BbvI-generated sequences. PCR products were analyzed on an agarose gel stained with ethidium bromide.
Dilution of heterozygous into homozygous samples
PCR products that were determined via sequencing to be heterozygous at the examined Ku gene position were diluted into homozygous samples in order to produce mixed samples with decreasing mutation frequency (i.e. decreasing ratio of mutant to wild-type alleles) down to a mutation frequency of 1x10-6. Alternatively, the whole genomic DNA from heterozygous samples was mixed with genomic DNA from homozygous samples so that mutation frequencies down to 5x10-6 were obtained. To prepare the 5x10-6 mutation frequency, 0.1 ng heterozygous genomic DNA (sim;15 cells equivalent) was added to 10 µg homozygous DNA. The mixed sample was then split into 10 PCR reactions of 1 µg each, after which the PCR products were re-mixed into one sample. All experiments were reproduced at least three times.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Mutation detection in cDNA (tenascin gene) and in genomic DNA (p53 gene codon 245)
cDNA reverse transcribed from cultured SAOS-2 cell mRNA was sequenced and found to contain an A2366
T polymorphism that converts a 5'-GAAC-3' to a 5'-GATC-3' site on one allele of the 3123 bp tenascin cDNA (heterozygous, Figure 2). The generation of this new SAU3A1 restriction site in the tenascin gene (GenBank accession no. XM_004201) was therefore used to validate the present technique. Amplification using the designated primers from cDNA results in a 357 bp sequence (tenascin gene positions 2323–2680) that contains the new SAU3A1 restriction site 14 bp from the end. Digestion with SAU3A1 resulted in a fragment that could be resolved on an ethidium gel (Figure 2A, classical PCR–RFLP). Alternatively, the digested PCR product was processed using the present APRIL-ATM procedure, as outlined in Figure 1
. Figure 2B depicts the product obtained following the first PCR amplification, using either the reverse primer plus the ligated primer (lane 1) or the forward primer plus the ligated primer (lane 2). Using this combination of primers, the expected band is resolved in lane 1 but not in lane 2, while additional products (smear) may also be observed in the background. In Figure 2C the product of the second PCR is observed, using the restriction site-specific primer together with the reverse (lane 1) or the forward (lane 2) primer. Single bands with no background are now resolved in both lanes. If the length of the second PCR primer (20mer) is taken into account, the sizes of the bands observed in Figure 2C correspond to the expected products following ligation at the SAU3A1 restriction site at position 2366. Therefore, the current method can be applied in two alternative formats, i.e. using either the forward or the reverse primers in conjunction with the ligated primer. In addition, in contrast to classical PCR–RFLP, which shows two bands following digestion of the heterozygous sample (i.e. the original and the digested fragments), the present procedure generates single bands, because only the mutated fragments are amplified.
|
Detection of the G14154
A mutation in codon 245 of the p53 gene (GenBank accession no. X54156) was performed by applying the method to genomic DNA from a normal individual and from a patient with a GA mutation (Figure 2E). Strong bands are demonstrated for the patient sample (lanes 3 and 4), which are absent from the normal sample (lanes 1 and 2). As with tenascin, detection of the mutation is possible using two alternative formats, i.e. either using the forward (lanes 1 and 3) or the reverse primers (lanes 2 and 4) in conjunction with the ligated primer. Extraction of the assay-generated band from the agarose gel and sequencing confirmed that the ligated oligonucleotide had ligated at the mutation position and produced correct identification of the mutation (Figure 2F, sequencing directly from genomic DNA; Figure 2E, sequencing of the APRIL-ATM-isolated gel band).
Mutation detection in genomic DNA (Ku gene)
Genomic DNA extracted from discarded surgical lung tissue was used to amplify regions of the Ku gene (GenBank accession no. AC008123) with a G74582A polymorphism that generates a new SAU3A1 restriction site. The designated primers were used to amplify Ku region 74355–74689 using a high fidelity polymerase. Samples from six patients were sequenced in the resulting 334 bp region and three were found to contain a heterozygous polymorphism that converts the `wild-type' 5'-GGTC-3' to 5'-GATC-3', while the remaining were found to be wild-type on both alleles. The newly generated SAU3A1 restriction site in the heterozygous patients lies at position 227 of the 334 bp long Ku PCR fragment. Of these six samples, one wild-type (homozygous) and one heterozygous PCR fragment were then further analyzed via classical PCR–RFLP and via the present method. Figure 3A demonstrates that, in agreement with the electropherograms, following digestion of the PCR products with SAU3A1 an sim;230 bp fragment appears for the heterozygous sample (lane 2) but not for the wild-type (homozygous, lane 1). The SAU3A1-generated fragment in lane 2 is incomplete, since <50% of the heterozygous sample is digested, despite purification of the original PCR product and the lengthy enzymatic digestion applied (3 h). The sources of this incomplete digestion, which is common in PCR–RFLP, are currently not well understood. However, it is clear that they reduce the already limited ability of PCR–RFLP to distinguish heterozygous samples in the presence of excess wild-type sample.
|
Next, the two SAU3A1-digested samples were ligated with a linker primer and processed by the present protocol (Figure 3B
). Figure 3B1 depicts the product obtained following the first PCR amplification, using the forward and ligated primers for the wild-type (lanes 1 and 2) and heterozygous (lanes 3 and 4) samples. For the samples in lanes 1 and 3 the dephosphorylation step was omitted. The heterozygous samples depict strong bands around 360 bp, which are absent from the wild-type. Additional bands can be seen in all samples, depending also on whether the dephosphorylation step after SAU3A1 digestion was included or omitted. Figure 3B2 depicts the same samples following the second PCR step, using the forward and restriction site-specific primers. The heterozygous samples (lanes 3 and 4) depicted single, clear, strong bands in this case. These bands were completely absent from the wild-type samples (lanes 1 and 2). The effect of dephosphorylating on the two samples prior to the SAU3A1 enzymatic digestion can be seen in Figure 3B3. In the first two lanes wild-type (lane 1) and heterozygous (lane 2) samples were subjected to the first PCR of the present protocol. Unclear results were obtained when only a single PCR round was applied, since the band that corresponds to the mutation is very faint and another band is also evident. Following the second PCR of the wild-type and heterozygous samples the restriction fragment is clearly present for the heterozygous sample (lane 4). However, a high background appears in the wild-type sample (lane 3). Thus it appears that the partial dephosphorylation step improves the signal-to-noise ratio. This empirical observation is currently not fully understood. It may be that DNA ends that undergo `illegitimate ligation' (`background') are more rapidly dephosphorylated than DNA ends that generate the `signal'. The dephosphorylation, under the conditions applied, is estimated to be at least 95% efficient, as evidenced by control experiments performed in the early stages of assay development (data not shown). Despite that, the remaining 5% phosphorylated DNA ends are sufficient to generate a strong PCR signal with a very low background. Taken together, the data in Figure 3 indicate that the most clear, background-free identification of the mutation is obtained by following the entire APRIL-ATM protocol in the sequence outlined in Figure 1. To further verify the detection, position and nature of the mutation, the ethidium gel band isolated from the heterozygous sample (Figure 3B2, lane 4) was extracted and sequenced. The electropherogam (Figure 3B) demonstrates clearly that the method selects only the mutant (GA) fragment from the heterozygous Ku sequence. Therefore, APRIL-ATM allows further characterization of the mutated fragment. To further examine the reproducibility and reliability of the procedure in correctly identifying the G74582A polymorphism in Ku when samples from different patients were used, a set of four genomic DNA samples of unknown status for the Ku SNP were prospectively examined via APRIL-ATM and then sequenced. Figure 4A demonstrates that the method correctly identified the presence or absence of the polymorphism, i.e. two samples were confirmed as heterozygous and two as wild-type (homozygous), in agreement with the sequencing-derived electropherograms. A further 11 patient samples whose status had already been verified by sequencing were also retrospectively examined via APRIL-ATM (Figure 4B). The assay verified the Ku G74582A polymorphism status correctly in all cases.
|
Detection limit
To establish the detection threshold of the present method we performed a series of dilutions of the heterozygous PCR products into wild-type PCR products, following their amplification from genomic DNA (Figure 5A
). Figure 5A1 demonstrates detection of the mutants for mutant:wild-type ratios (i.e.`mutation frequencies') of 0.5, 0, 5x10-3, 5x10-4 and 5x10-5 (lanes 1–5, respectively). Further dilutions of the mutants are demonstrated in Figure 5A2, and their correct detection for mutation frequencies of 0.5, 5x10-5, 5x 10-6, 1.6x10-6 and 0 (lanes 1–5, respectively) is depicted. Densitometric evaluation of the ethidium gel bands in Figure 5A2 demonstrates that the amount of PCR product generated by the assay is approximately proportional to the initial mutation frequency (lanes 2–5; lane 1 is `saturated'). Therefore, quantitation of mutation frequency is possible following the present protocol. Experiments for elucidation of the detection limit of APRIL-ATM were reproduced with the Ku gene three times to verify the 1.6x10-6 mutant-to-wild type limit (not shown). At even lower mutation frequencies (<1.6x10-6) background bands became evident in the application of APRIL-ATM. This limit is possibly due to the onset of PCR errors in the initial amplification step. The polymerase used (Advantage HF-2) is a mixture of modified Taq plus Deep Vent polymerases (Clontech Inc., personal communication). Future substitution of Advantage HF-2 with Pfu polymerase can potentially increase the detection limit of APRIL-ATM, since Pfu has a higher fidelity than Deep Vent. It should be noted, however, that the dependence of the selectivity of the assay on polymerase errors is also a function of the polymerase fidelity in the sequence context of the target region at the target base. Therefore sequence-to-sequence variation of APRIL-ATM selectivity may also occur.
|
The dilution experiment was also repeated by mixing genomic DNA from a heterozygous sample into genomic DNA from a wild-type sample. The resulting mutation frequencies were 5x10-5 and 5x10-6. To reproducibly achieve such low mutation frequencies following the mixing of whole genomes we added 100 pg mutant genomic DNA to 10 µg wild-type genomic DNA (i.e. roughly equivalent to mixing 15 heterozygous cells with 1.6x106 homozygous cells). Using lower amounts of mutant or wild-type DNA could be expected to result in statistical fluctuations, since, for example, 10 pg mutant genomic DNA would correspond to just 1.5 cells. The process used was found to reproducibly and clearly detect traces of heterozygous sample added to homozygous DNA (Figure 5B
, lanes 1–5). Following extraction and purification from the ethidium gel of the fragment in lane 3 a sequencing reaction was performed (Figure 5, electropherogram), which showed only the mutated 5'-GATC-3' sequence and no evidence of the wild-type 5'-GGTC-3' sequence. This demonstrates clearly that, at a mutant frequency of 5x10-6, the technique isolates almost exclusively mutant GA fragments and no false positives are present. |
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The current method versus classical PCR–RFLP
The mutation-dependent disappearance or appearance of an agarose gel-resolved restriction fragment has for a long time served as a simple, reliable method to aid research and clinical diagnosis. Such a `classical' PCR–RFLP is extensively used to detect crucial mutations in a wide variety of genes, including ras (Bos and Van Mansfeld, 1992
; van Mansfeld and Bos, 1992), cystic fibrosis (Friedman et al., 1990), PKU (Eiken et al., 1991), p53 (Behn et al., 1998), SPINK1 (Plendl et al., 2001) and others (Parsons and Heflich, 1997). The existence of an enzyme that differentiates between mutant and wild-type sequences at the desired position is a prerequisite for using the method. However, the availability of numerous restriction enzymes allows substantial versatility to such RFLP-based approaches. In several cases even if a recognition sequence does not exist a priori, it can be artificially introduced by specially designed PCR primers (Haliassos et al., 1989; Kumar and Dunn, 1989; Day et al., 1999a,b), thus increasing the range of sequences that can be examined via PCR–RFLP. For example, in a random DNA sequence >20% of bases are contained within a pre-existing 4 base restriction site and >60% of bases are within a 4 base subsequence that can be converted into a restriction site by using the `mismatched primer approach' (Day et al., 1999b). On the other hand, the drawback of classical PCR–RFLP is its sensitivity, as it cannot resolve mutations present in <1–10% of the sample (Chen and Viola, 1991; Kahn et al., 1995) and, therefore, very low abundance mutations cannot be detected. Occasionally, even for the heterozygous situation, the restriction fragment is not easily resolved on an ethidium gel due to inadequate performance of the enzyme or the presence of inhibiting impurities (see Figure 3A, lane 2). Radiolabeled probes have been used to increase PCR–RFLP sensitivity to the level of 10-4–10-6 (Haliassos et al., 1989; Nakazawa et al., 1990, 1992), however, the practical disadvantages of handling radioactive materials and the lengthy protocols involved render this approach cumbersome.
The present approach, APRIL-ATM, retains the advantages of reliability and versatility of classical PCR–RFLP while increasing the sensitivity to the level of detecting mutant frequencies down to 1.6x10-6 without the use of radioactivity. The approach is easy to adopt, since the steps involved (digestion, ligation, dephosphorylation and amplification) are common practice in most laboratories. Following ligation of an oligonucleotide at a mutation-generated restriction site, a series of two modest cycle PCR amplifications is employed to generate enough mutant DNA to be resolved on a regular ethidium gel. The second PCR is nested to the first one and contains the SAU3A1 enzyme recognition site (5'-GATC-3'). This second, restriction site-specific PCR is necessitated by the fact that ligation of the linker oligonucleotide is not a very specific process (Zirvi et al., 1999) and linker molecules are also added (inefficiently) to the ends of the undigested sequences. Even traces of such `illegitimate' ligation result in unwanted bands and a `smear' following the first PCR amplification step (Figure 3B1 and B3). The second round of PCR largely eliminates the unwanted fragments (Figure 3B2). The overall improvement afforded by APRIL-ATM over classical PCR–RFLP can be assessed by comparison of Figure 3A with 3B2. Furthermore, unlike radioactivity-based PCR–RFLP methods (Nakazawa et al., 1990), APRIL-ATM produces enough DNA at a mutation frequency of 5x10-6 to allow further verification of the mutation via sequencing. The characteristics of the polymorphisms and mutations examined in the present work are summarized in Table I.
|
Further advantages and disadvantages of APRIL-ATM
An advantage of the present approach is that its design drastically minimizes known sources of false negatives or false positives. RFLP-based methodologies that rely on the destruction of the wild-type sequence to enrich for mutated alleles may generate false positives resulting from incomplete enzymatic digestion (Steingrimsdottir et al., 1996
; van Houten et al., 2000). RSM, for example, may produce restriction fragments that, upon sequencing, are shown to contain 50–70% false positives (Jenkins et al., 1998). As a result, wild-type destruction assays are restricted to enzymes that can achieve near complete digestion (Jenkins et al., 1999), thereby limiting the sites that can be tested for mutation. To overcome this problem, an inverse restriction site mutation assay was developed where mutation converts a restriction site for a first enzyme to a restriction site for an alternative, second enzyme (Jenkins et al., 1999). True positives are more reliably assessed by the inverse RSM assay, however, its application is only possible for the very few sequences that satisfy this double condition. In contrast, APRIL-ATM produces no signal if enzymatic digestion at the target sequence is incomplete. Thereby previously identified problems, such as heteroduplex formation during PCR (Jenkins et al., 1998; Watzinger et al., 2001), DNA damage (Jenkins et al., 1998) or methylation sensitivity (Jenkins et al., 1998), all of which cause incomplete digestion, do not result in false positives using the present approach, since incomplete digestion generates no bands. For false negatives to appear with APRIL-ATM, on the other hand, the enzyme should not be able to digest its recognition sequence at all. At a mutation frequency of 5x10-6, DNA isolated by APRIL-ATM was sequenced and found to contain no evidence of the wild-type sequence (Figure 5, electropherogram). The starting amount of tissue required to generate the 10 µg total genomic DNA utilized in this experiment was sim;15 mg, well within the amount of tissue extracted in a small biopsy. Therefore, appropriately chosen mutations at a frequency of 5x10-6 in a biopsy sample can be expected to be reliably diagnosed using the present approach. A further consequence of the reliability of APRIL-ATM is that it is not restricted to selected enzymes but any of the existing restriction endonucleases can be used, whether efficient or inefficient `cutters'. For example, using the enzyme BbvI APRIL-ATM detected a p53 codon 245 mutation (Figure 2D) which is frequently (>2%) encountered in stomach, soft tissue and colon cancers (Hernandez-Boussard et al., 1999; McKinzie et al., 2001) and can be used as a tumor marker (Dong et al., 2001).
A disadvantage of APRIL-ATM over approaches such as MutEx/ACB PCR (Parsons and Heflich, 1998) is that it cannot detect all mutations but only those that generate a new restriction site, which amount to sim;20–30% of all target sequences (Zirvi et al., 1999). An improvement will result if the assay is combined with the `mismatched primer approach' (Haliassos et al., 1989; Day et al., 1999b), in which case >60% of all mutations would be detectable (work in progress). Furthermore, the application of two consecutive PCR steps in APRIL-ATM increases the sensitivity of the assay, but may also increase the risk of carryover contamination. Finally, unless the gene-specific primers are carefully chosen, the nested PCR step may result in PCR artifacts. Taken together, these considerations suggest that APRIL-ATM will be used as a complement to existing RFLP-based approaches that rely on the destruction of wild-type sequences, because sequences that do not form a restriction site in the wild-type may form a restriction site in the mutant, and vice versa. Therefore, APRIL-ATM reinforces and substantially enriches the currently available methodologies for RFLP-based mutation detection.
In summary, APRIL-ATM is a major improvement over classical PCR–RFLP and a sensitive new approach for the quantitative detection of low abundance mutations or polymorphisms in a variety of DNA sequences, down to a mutation frequency of 1.6x10-6, without using radioactivity. The design of the assay minimizes known sources of false positives and false negatives and also allows further characterization of the mutation via sequencing. The reliability of the assay should allow it to aid the measurement of mutagen exposures, molecular clinical diagnosis for detection of minimal residual disease or early detection of tumor markers in bodily discharges.
|
Acknowledgments |
|---|
We are grateful to Dr Edward Fox, Director of the Dana Farber Molecular Diagnostics Laboratory, for helpful discussions during the design and execution of this work. The assistance of Mohamet Miri and Frank Haluska MD in obtaining tissue specimens from the Massachussetts General Hospital Tumor Bank and of the Cooperative Human Tumor Network staff is also acknowledged. This research was supported in part by NIH grants CA 83234 and CA/HG 90422.
|
Notes |
|---|
2 To whom correspondence should be addressed at: Longwood Radiation Oncology Center, Brigham–Dana Farber Children's Hospitals, Level L2, Radiation Therapy, 75 Francis Street, Boston, MA 02115, USA. Tel: +1 617 6326905; Fax: +1 617 6326900; Email: mmakrigiorgos{at}partners.org
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
-
Barany,F. (1991) Genetic disease detection and DNA amplification using cloned thermostable ligase. Proc. Natl Acad. Sci. USA, 88, 189–193.
Bazrafshani,M.R., Ollier,W.E. et al. (2000) A novel PCR-RFLP assay for the detection of the single nucleotide polymorphism at position –1082 in the human IL-10 gene promoter. Eur. J. Immunogenet., 27, 119–120.[ISI][Medline]
Behn,M., Qun,S. et al. (1998) Frequent detection of ras and p53 mutations in brush cytology samples from lung cancer patients by a restriction fragment length polymorphism-based `enriched PCR' technique. Clin. Cancer Res., 4, 361–371.
Bos,J.L. and Van Mansfeld,A.D. (1992) Detection of point mutations in p21ras genes. Ann. Med., 24, 207–209.[ISI][Medline]
Chen,J. and Viola,M.V. (1991) A method to detect ras point mutations in small subpopulations of cells. Anal. Biochem., 195, 51–56.[ISI][Medline]
Day,J.P., Bergstrom,D. et al. (1999a) Nucleotide analogs facilitate base conversion with 3' mismatch primers. Nucleic Acids Res., 27, 1810–1818.
Day,J.P., Hammer,R.P. et al. (1999b) Nucleotide analogs and new buffers improve a generalized method to enrich for low abundance mutations. Nucleic Acids Res., 27, 1819–1827.
Dong,S.M., Traverso,G. et al. (2001) Detecting colorectal cancer in stool with the use of multiple genetic targets. J. Natl Cancer Inst., 93, 858–865.
Eiken,H.G., Odland,E. et al. (1991) Application of natural and amplification created restriction sites for the diagnosis of PKU mutations. Nucleic Acids Res., 19, 1427–1430.
Felley-Bosco,E., Pourzand,C. et al. (1991) A genotypic mutation system measuring mutations in restriction recognition sequences. Nucleic Acids Res., 19, 2913–2919.
Friedman,K.J., Highsmith,W.E.Jr et al. (1990) Cystic fibrosis deletion mutation detected by PCR-mediated site-directed mutagenesis. Clin. Chem., 36, 695–696.
Haliassos,A., Chomel,J.C. et al. (1989a) Detection of minority point mutations by modified PCR technique: a new approach for a sensitive diagnosis of tumor-progression markers. Nucleic Acids Res., 17, 8093–8099.
Haliassos,A., Chomel,J.C. et al. (1989b) Modification of enzymatically amplified DNA for the detection of point mutations. Nucleic Acids Res., 17, 3606.
Hernandez-Boussard,T., Rodriguez-Tome,P. et al. (1999) IARC p53 mutation database: a relational database to compile and analyze p53 mutations in human tumors and cell lines. International Agency for Research on Cancer. Hum. Mutat., 14, 1–8.[ISI][Medline]
Jenkins,G.J., Chaleshtori,M.H. et al. (1998) Mutation analysis using the restriction site mutation (RSM) assay. Mutat. Res., 405, 209–220.[ISI][Medline]
Jenkins,G.J., Takahashi,N. et al. (1999) Inverse restriction site mutation (iRSM) analysis. Mutation detection involving the formation of restriction enzyme sites in target genes. Mutagenesis, 14, 37–42.
Kahn,S.M., Jiang,W. et al. (1995) Diagnostic detection of mutant ras genes in minor cell populations. Methods Enzymol., 255, 452–464.[ISI][Medline]
Kumar,R. and Dunn,L.L. (1989) Designed diagnostic restriction fragment length polymorphisms for the detection of point mutations in ras oncogenes. Oncogene Res., 4, 235–241.[ISI][Medline]
McKinzie,P.B., Delongchamp,R.R. et al. (2001) Prospects for applying genotypic selection of somatic oncomutation to chemical risk assessment. Mutat. Res., 489, 47–78.[ISI][Medline]
Monteiro,C., Marcelino,L.A. et al. (2000) Molecular methods for the detection of mutations. Teratog. Carcinog. Mutagen., 20, 357–386.[ISI][Medline]
Nakazawa,H., Aguelon,A.M. et al. (1990) Relationship between chemically induced Ha-ras mutation and transformation of BALB/c 3T3 cells: evidence for chemical-specific activation and cell type-specific recruitment of oncogene in transformation. Mol. Carcinog., 3, 202–209.[ISI][Medline]
Nakazawa,H., Aguelon,A.M. et al. (1992) Identification and quantification of a carcinogen-induced molecular initiation event in cell transformation. Oncogene, 7, 2295–2301.[ISI][Medline]
Parry,J.M., Shamsher,M. et al. (1990) Restriction site mutation analysis, a proposed methodology for the detection and study of DNA base changes following mutagen exposure. Mutagenesis, 5, 209–212.
Parsons,B.L. and Heflich,R.H. (1997) Genotypic selection methods for the direct analysis of point mutations. Mutat. Res., 387, 97–121.[ISI][Medline]
Parsons,B.L. and Heflich,R.H. (1998) Detection of basepair substitution mutation at a frequency of 1x10(–7) by combining two genotypic selection methods, MutEx enrichment and allele-specific competitive blocker PCR. Environ. Mol. Mutagen., 32, 200–211.[ISI][Medline]
Plendl,H., Siebert,R. et al. (2001) High frequency of the N34S mutation in the SPINK1 gene in chronic pancreatitis detected by a new PCR-RFLP assay. Am. J. Med. Genet., 100, 252–253.[ISI][Medline]
Pourzand,C. and Cerutti,P. (1993) Genotypic mutation analysis by RFLP/PCR. Mutat. Res., 288, 113–121.[ISI][Medline]
Steingrimsdottir,H., Beare,D. et al. (1996) Development of new molecular procedures for the detection of genetic alterations in man. Mutat. Res., 353, 109–121.[ISI][Medline]
van Houten,V.M., Tabor,M.P. et al. (2000) Molecular assays for the diagnosis of minimal residual head-and-neck cancer: methods, reliability, pitfalls and solutions. Clin. Cancer Res., 6, 3803–3816.
van Mansfeld,A.D. and Bos,J.L. (1992) PCR-based approaches for detection of mutated ras genes. PCR Methods Appl., 1, 211–216.[Medline]
van Zeeland,A.A. and Vrieling,H. (1999) Mutation in mammalian cells in vitro and in vivo as indicators of exposure to carcinogens. IARC Sci. Publ., 146, 355–365.[Medline]
Watzinger,F., Horth,E. et al. (2001) Quantification of mRNA expression by competitive PCR using non-homologous competitors containing a shifted restriction site. Nucleic Acids Res., 29, E52–2.[Medline]
Zirvi,M., Bergstrom,D.E. et al. (1999) Improved fidelity of thermostable ligases for detection of microsatellite repeat sequences using nucleoside analogs. Nucleic Acids Res., 27, e41.
Received on February 19, 2002; revised on April 19, 2002; accepted on April 22, 2002.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Li, R. Berbeco, R. J. Distel, P. A. Janne, L. Wang, and G. M. Makrigiorgos s-RT-MELT for rapid mutation scanning using enzymatic selection and real time DNA-melting: new potential for multiplex genetic analysis Nucleic Acids Res., June 9, 2007; 35(12): e84 - e84. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W.-H. Liu, M. Kaur, G. Wang, P. Zhu, Y. Zhang, and G. M. Makrigiorgos Inverse PCR-Based RFLP Scanning Identifies Low-Level Mutation Signatures in Colon Cells and Tumors Cancer Res., April 1, 2004; 64(7): 2544 - 2551. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. V. Harden, D. C. Thomas, N. Benoit, K. Minhas, W. H. Westra, J. A. Califano, W. Koch, and D. Sidransky Real-Time Gap Ligase Chain Reaction: A Rapid Semiquantitative Assay for Detecting p53 Mutation at Low Levels in Surgical Margins and Lymph Nodes from Resected Lung and Head and Neck Tumors Clin. Cancer Res., April 1, 2004; 10(7): 2379 - 2385. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Kaur and G. M. Makrigiorgos Novel amplification of DNA in a hairpin structure: towards a radical elimination of PCR errors from amplified DNA Nucleic Acids Res., March 15, 2003; 31(6): e26 - e26. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||