In this article we will discuss about the isolation and sequencing of heteroduplex fragments from DHPLC analysis of somatic, low-frequency mutations.
Introduction to Isolation and Sequencing of Mutations by DHPLC:
Identification of somatic mutations in samples from solid tumors represent some additional challenges compared to the identification of heterozygous germline mutations. Biopsies from solid tumors often contain various amounts of normal cells and the tumor itself can be heterogeneous with respect to the mutation. Thus, the frequency of any one mutant allele is usually unknown.
If the tumor is homogenous for a mutation in one allele of a gene and has lost the remaining wild-type copy of the gene, and the normal tissue component has been minimized through dissection of the sample, the frequency of the mutant allele can be very high.
Otherwise, a heterogeneous tumor sample containing a large amount of normal tissue will have a very low frequency of the mutant allele. In either case, the heteroduplex fraction will be small and the DHPLC conditions must be optimized in order to reduce the amount of polymerase induced errors and, if possible, to sharpen the wild- type peak.
When a sample has been identified with a heteroduplex peak pattern, the mutation has to be characterized. If the frequency of the mutant allele is high, the sample can easily be sequenced directly. If the frequency is low, the heteroduplex peak(s) has/have to be isolated, re-amplified, and sequenced.
DNA fragments can be isolated with the WAVE Nucleic Acid Fragment Analysis System. Previously described applications include subcloning of PCR products and purification of template DNA for sequencing. Here, a procedure for isolation, re-amplification and sequencing of low-frequency mutations is described.
As examples two mutations in exon 8 of the p53 gene (TP53) are analyzed. This particular exon is selected based on its characteristic melting domain profile with a central ‘high-melting’ domain (Figure 9- 1). The two mutations were detected using denaturing gradient gel electrophoresis (DGGE) and are selected based on their localization in different domains.
Mutation 1 is a G>T located in the 5′ ‘low-melting’ domain and is therefore expected to be easily identified by DHPLC. Mutation 2 is an A>T in the central ‘high-melting’ domain and is expected to be more difficult to recognize.
Results and Discussion:
i. DHPLC Conditions:
Genomic DNA was isolated from tumor samples as described previously. Exon 8 of TP53 was amplified using a nested PCR approach. In the first round, approximately 10 ng of genomic DNA was amplified in a 25-µl reaction using 1.25 U of Optimase DNA polymerase for 25 cycles (annealing temperature: 56°C). The primers used were 5′- AAATGGGCAGGTAGGACCTGAT (sense) and 5′-GTGCTAGGAAAGAGGCAAGGAA (anti- sense).
The product of the latter reaction was used either as a template for the second round of PCR for the DHPLC analysis or for the re-amplification prior to the sequencing reaction as described below. The second round of PCR amplification was performed in a 25-µl reaction using 1.25 U of Optimase polymerase, but for 30 cycles (annealing temperature: 62°C). The primers used were 5′-ACTGCCTCTTGCTTCTCTTTTCC (sense) and 5′-AATCTGAGGCATAACTGCACCC (anti- sense). The anti-sense primer was labeled with 6-FAM (ABI) at the 5′-end.
DHPLC was performed on a WAVE Nucleic Acid Fragment Analysis System from Transgenomic, which was equipped with a fluorescence detector (PMT voltage: high; response time: 0.1 sec) and controlled by the HSM software version 3.0. The buffers used were: buffer A: 0.1 M triethylammonium acetate (TEAA); buffer B: 0.1 M TEAA/25% acetonitrile (ACN); buffer D: 75% ACN. Gradient conditions were: 0 min: Buffer A = (105 – B)%, Buffer B = (B – 5)%; 0.1 min: Buffer A = (100 – B)%, Buffer B = B%; 6.1 min: Buffer A = (88 – B)%, Buffer B = (B + 12)%; 6.2-6.7 min: Buffer D = 75% ACN; 6.8-9.0 min: Buffer A = (105 – B)%, Buffer B = (B – 5)%. At 59°C B = 49, at 64°C B = 45, and at 65°C B = 44. Flow rate was 0.9ml/min. Each PCR reaction was diluted 1:1 with water in order to get the signal at 59°C below the detection limit of the detector. Six-µl aliquots from the same PCR reaction were analyzed at each of the three temperatures.
As predicted by the melting curve analysis in Figure 9-1, the optimal temperature for detection of mutation 1 is 59°C where the two heteroduplex peaks are clearly separated and the homoduplex mutant peak containing a T at position 38 can be seen as a shoulder in front of the wild-type peak containing a G at this position (Figure 9-2A).
The heteroduplex peaks from mutation 2 can also be recognized at 59°C, but only as a narrow shoulder (Figure 9-2B). Optimal separation of mutation 2 is seen at 64°C (Figure 9-3A). At 65°C mutation 2 can still be recognized but the homozygous wild-type/mutant peak (A versus T at position 110) is now fairly broad (Figure 9-3B).
Although the two heteroduplex peaks of mutant 2 elute as one peak, they are still easily recognized at both 64 and 65°C. In order to test the sensitivity of the fragment collection and re-sequencing approach, genomic DNA from mutant 2 was diluted with increasing amounts of wild-type genomic DNA and analyzed together with the original mutant 2. One of these dilutions is included in Figures 9-3A and 9-3B.
ii. Fragment Collection:
Fragments were collected with a WAVE Fragment Collector Model FCW 200 controlled by the Fragment Collector software version 188.8.131.52 (Transgenomic). For elution profiles like mutation 1 (Figure 9-2A) it is possible to use ‘Peak’ detection where the fraction is automatically collected around a given peak.
For mutation 1, an approximately 100-µl fraction (corresponding to 0.11 min collection time) is collected around the initial heteroduplex peak. For profiles like mutation 2, where the fragment of interest is not a peak, an alternative strategy is to use a ‘Whole Window’ approach where approximately 50-µl fractions are collected every 0.1 min independent of the signal. In the example shown in Figure 9-3B, fractions were collected from 5.40 – 5.46, 5.50 – 5.56, and so on until 6.20 – 6.26 minutes.
Depending on the design of the fragment collector, a ‘dead volume’ may be present in the instrument. With the FCW 200 a small volume from the valve to the tip of the dispenser will be carried on to the next collection during multiple collections from a single sample and from the last collection of one sample to the first collection of the next.
To avoid any contamination from one sample to another, an initial fraction is collected from every sample before any attempts to isolate the heteroduplex peak(s). In the examples shown in Figures 9-2A, 9-3A, and 9-3B, this first fraction was collected from 2.00 – 2.11 minutes (approximately 100µl) where the signal is at baseline.
iii. Re-Sequencing of Isolated Fragments:
The elution buffer in collected fractions does not affect any subsequent PCR reactions even when present at 20% by volume. For the re-amplification of the collected heteroduplex fractions in Figures 9-2A, 9-3A, and 9-3B, 1 µl of the collected material was re-amplified in a 25-µl reaction using 1.25 U Optimase polymerase for 30 cycles.
PCR products were purified from agarose gels using the QIAquick Gel Extraction Kit prior to sequencing with ABI PRISM BigDye. Terminators v1.0 on an ABI PRISM 310 Genetic Analyzer (Applied Biosystems).
In order to obtain uniform sequence data from various PCR products, all re-amplification reactions were performed with tailed PCR primers.
Thus, for TP53 exon 8 the sense primer is 5′-ctcctgttccgaccctgccACTGCCTCTTGCTTCTCTTTTCC and the anti-sense primer is 5′-cggaacaggagagcgcacgAATCTGAGGCATAACTGCACCC.
The nucleotides in lower case represent the two universal sequencing tails/primers that are used for bi-directional sequencing. The exon specific part of each primer, shown in upper case, are identical to the primers used to generate the PCR products for the WAVE System analysis.
Initial attempts to use -21 M13 and M13 REV as universal sequencing primers did not provide satisfactory results. Instead, the ABI PRISM Primer Express software version 1.0 (Applied Biosystems) was used to identify a new set of primers that could also be used with the pGEM-3Zf(+) plasmid which is provided as a positive control template with the BigDye(tm) Terminators kits.
Figure 9-4 shows sequencing results for mutation 1. The wild-type and mutation 1 ‘direct’ sequences are derived from re-amplifications of the same initial first round of PCR that was used as a template for the PCR products analyzed on the WAVE System.
Although it is present at a fairly low level, the G>T mutation in mutation 1 ‘direct’ is clearly seen in both directions and would probably have been identified without the need for an isolation and re-amplification of the heteroduplex peak. However, the sequence of the mutation 1 ‘purified’ illustrates how the intensity of the mutant signal can be significantly improved by the isolation/re-amplification approach.
Figure 9-5 shows sequencing results for mutation 2. In this case, the very faint signal in the mutant 2 diluted ‘direct’ sequencing result would not have been recognized as more than random noise.
However, even though the fraction collected at 64°C includes a significant proportion of wild-type molecules (the signal of the true wild-type is elevated from the base-line, see Figure 9- 3A), the mutant fraction is enriched to a point where it can be identified in the sequence analysis. At 65°C (Figure 9-3B), the proportion of wild-type molecules is even higher and the mutant signal in the sequence is only slightly improved.
DHPLC is a sensitive method for the detection of the presence of low-frequency mutations that may not be reliably recognized by direct sequencing. Combined with a fragment collector, it is also possible to perform automated isolation of small heteroduplex peaks allowing for the characterization of these mutations by re-amplification and sequencing.
The sensitivity of DHPLC for detection of low-frequency mutations and the quality of the sequence data obtained from small heteroduplex peaks depend on the quality of the PCR product and the melting profile. Any errors introduced by the polymerase during the PCR reaction will accumulate as a broad peak in front of the wild-type peak and lower the ratio of the true mutation in the sequence analysis.
As demonstrated here, Optimase polymerase can provide PCR products virtually free of polymerase-induced errors even after a nested approach with 25 and 30 cycles, respectively.
The nested approach was selected mainly in order to describe a general strategy for genomic DNA templates of variable quality. Thus, the conditions described here are also suitable for DNA isolated from paraffin-embedded tumor tissue.
As DNA from paraffin-embedded sections is often highly degraded, PCR amplicons should be designed as short as possible. However, if the primers get to close to the intron/exon boundaries, it can be difficult to obtain high quality sequence data from both directions.
With the universal sequence primers described here, consistent sequence data can be obtained from very near the end of the amplicon-specific primers. Furthermore, as the same two primers can be used for different amplicons more uniform sequence data can be obtained.