Principles of melting-curve based analysis of genetic variation
Post-PCR melting curve analysis (MCA) has become a robust and well-established method to characterize gene fragments, e.g., for microbiological identification or the detection of mutations and single nucleotide polymorphisms (SNPs). In the case of SNP studies, the goal can either be to detect the presence of different allelic variants of given SNPs in a sample set, or to identify polymorphic positions at which previously unknown SNPs do occur without determining the exact allele in the first place. In the latter case, melting curve analysis is used as a screening technique to reduce the number of sequencing reactions required to detect new variants.
Known SNPs present in PCR amplicons have been investigated for many years with sequence-specific, labeled probes that bind with different strength to the different alleles or allele combinations in the SNP-containing region, depending on whether they fully match the target sequence or contain mismatches (probe melting). Another detection principle makes use of special DNA-binding dyes whose binding characteristics allow them to be used at high concentrations without inhibiting PCR. Due to its saturating, homogeneous staining of PCR products, such dyes give sharp, unique melting profiles that allow the differentiation between homo – and heterozygote samples, and, less frequently, even between homozygous wild-types and mutants (whole amplicon melting).
Geno typing of known sequence variants
For both probe-based and whole-amplicon melting, raw data is generally represented by plotting fluorescence over °C , with the relevant temperature range and acquisition rate (data resolution) being higher with generic high resolution dyes than with labeled probes (Fig.1. Ia, IIa). However, it can be analyzed more conveniently by looking at the negative first derivatives (-dF/dT), revealing melting temperatures as peaks (Fig. 1 Ib, IIb). A comparison of data derived from homozygous wild-types, homozygous mutant and heterozygous samples can reveal differences in peak number, peak position or both. If high-resolution dyes are used instead of probes, the differences obtained between melting peaks are often detectable, but sometimes not big enough to allow a clear differentiation of homozygotes (Tmdye, Fig. 1 Ib) The use of sequence-specific labeled probes provides more reliable results and clearer separation (Tmprobe, Fig. 1 IIb), with heterozygotes showing two peaks, and homozygotes only one peak at different positions for wild-type and mutant. Software algorithms have been developed that make use of this latter principle to identify SNP alleles. Advanced algorithms are furthermore able to carry out an automated calling of the corresponding sample genotypes, by grouping samples with similar melting curve shape together, either automatically or (optional) based on standards included in the experiment (not shown).
High-resolution melting detects novel sequence variants
By definition, the scanning of amplicons for unknown variations cannot be done with sequence-specific probes since the location of the sequence around the potential polymorphic site(s) is not known. Designing a 30 bp probe for every potential polymorphic site on a given gene is just not a realistic approach. However, generic DNA-binding dyes with saturating binding characteristics can be used instead in combination with instruments allowing melting signals to be analyzed at high resolution (Ref. 2, 4). In such an analysis, unknown sequence variations in diploids like humans become apparent in heterozygous samples during post-PCR melting curve analysis due to the presence of heteroduplex DNA. When amplified and melted, these samples show melting curves with significantly different profiles (shapes) than those derived from homozygous (wild-type or mutant) samples. Software algorithms are available to analyze these differences by first normalizing the data and then temperature-shifting the curves such that differences of Tm between homozygotes with very similar curve shape disappear and heterozygotes stand out more clearly due to their differently shaped curves. In cases where Tm differences between homozygotes are big enough, these can be more obviously displayed by omitting the temperature shifting step (not shown). By finally plotting the difference in fluorescence between each sample, the homozygotes and heterozygote samples can be easily identified.
A Choice of SNP Discovery and Analysis Methods
Melting curve analysis has advantages over other mutation detection methods because it is based on a robust, post-PCR biophysical measurement. Less sequence information is needed to design a genotyping assay, probes can be designed with greater flexibility and the same dye or probe can be utilized for all known or unknown alleles present and investigated. Melting curve analysis reveals a maximum of information. Nowadays rather than only analyzing the position of a melting peak, more sophisticated algorithms now take the shape of melting curves into account for automatic grouping.
For pre-sequencing, a close-tube gene scanning approach like melting curve analysis with a high-resolution dye offers greater convenience and throughput compared with traditional methods (e.g., dHPLC).
In summary, the methods and principles presented for SNP analysis here have shown to be highly robust and informative, enabling their use in diagnostic testings (e.g., Factor II and Factor V Leiden variant test kits provided by Roche for the LightCycler(R) 2.0 System). For life science research, they have enabled large-scale research studies to discover new sequence variants and confirm the presence of known ones, as shown in the following examples:
Examples of melting curve analysis used in clinical research
In a study by Evrard et al. samples of individuals with cancer were screened for sequence variations in the cytidine deaminase (CDA) gene, which plays a key role in the response to chemotherapy with antimetabolite drugs (Ref. 5). In the search for a relationship between toxic side effects and genetic variants of the CDA gene, High Resolution Melting revealed sequence variants due to distinct patterns in melting curve shape. This allowed limiting sequencing to those samples that actually contain anomalous sequence information.
In this study, the LightCycler(R) 480 High Resolution Melting Mastermix, which contains a saturating fluorescent dye, was used on the LightCycler(R) 480 System from Roche Applied Science. PCR was carried out using a touchdown protocol, with annealing temperatures ranging from 70 to 60 °C. High Resolution Melting curve data were obtained at a rate of 25 acquisitions per °C.
High Resolution Melting analysis separated the DNA samples clearly into different groups. Twenty-two out of 46 samples were obviously heterozygous for a uniform sequence variation in the 173 bp fragment derived from exon 1 (Fig. 2, red curves). Sequencing linked this variation to a known SNP A79C resulting in a Lys27Gln change. All other amplicons had the wild-type sequence. These findings were confirmed by sequencing, showing 100% accuracy of the performed analysis. When a longer 223 bp amplicon was investigated, HRM revealed the presence of another variation in two cases (Fig. 3, green curves). Sequencing disclosed that these two samples were heterozygous for SNP A79C and in addition for IVS1+37 G>A.
In another study, Bauer et al. used High Resolution Melting curves to screen for CFTR mutations. In order to assess the 19 most prevalent CFTR alleles in Germany, exons 4, 7, 10, 11, 14b, 19, 20, and 21 were investigated (Ref. 1).
Heterozygous CFTR mutations (R347P, F508del, I507del, G542X, R553X, G551D, 1717-1G>A, 2789+5G>A, W1282X, N1303K) could easily be detected using a standard PCR protocol
The two studies described here show that sensitivity and specificity of High Resolution Melting-based gene scanning were comparable to those obtained with previously used, more conventional scanning methods (e.g., dHPLC), with the advantage that results were obtained much faster and with much less effort.
The LightCycler(R) 480 System is a versatile platform for the discovery and analysis of genetic variation (instrument available for 96- and / or 384-well plates) In particular, it is currently the only available plate-based real-time PCR system allowing mutation scanning by high-resolution melting (Ref.2 and 4). The optionally available LightCycler(R) 480 Genotyping and Gene Scanning Softwares can be combined with optimized ready-to-use PCR master mixes for both genotyping and gene scanning. The system thus provides highly accurate results based on the analysis of melting curve profiles.
LIGHTCYCLER is a trademark of Roche. Patent and license disclaimer information is available online (http://www.lightcycler.com).
1. Bauer, P. et al. LightCycler(R) 480 High Resolution Melting for CFTR Mutation Screening (2007). European Society of Human Genetics Meeting, Nice, poster presentation.
2. Dujols V, Kusukawa N, McKinney JT, Dobrowolsky SF, Wittwer CT. High-resolution melting analysis for scanning and genotyping., in Real Time PCR. Tevfik D, ed., Taylor and Francis, Abingdon, 2006.
3. Evrard et al (2007). Mutation scanning of the cytidine deaminase (CDA) and dihydropyrimidine dehydrogenase (DPYD) genes by high-resolution melting curve analysis on the LightCycler(R) 480 System in cancer patients treated with antimetabolites drugs. Human Genome Meeting 2007, Montréal, CA. Poster presentation, available online on www.lightcycler480.com.
4. Herrmann, M. G., J. D. Durtschi, et al. (2007). “Expanded Instrument Comparison of Amplicon DNA Melting Analysis for Mutation Scanning and Genotyping.” Clin Chem.
5. Mercier, C., Raynald, C. et al. (2007). “Toxic death case in a patient undergoing gemcitabine-based chemotherapy in relation with cytidine deaminase downregulation.” Pharmacogenetics and Genomics (in press).
6. Montgomery, J., C. T. Wittwer, et al. (2007). “Simultaneous mutation scanning and genotyping by high-resolution DNA melting analysis.” Nat Protoc 2(1): 59-66.
1 Roche Applied Science, Nonnenwald 2, 82372 Penzberg, Germany.
2 Roche Applied Science and Molecular Diagnostics, Laval, Quebec, Canada Correspondence should be addressed to Roche Applied Science and Molecular Diagnostics Canada.