|Home | About | Journals | Submit | Contact Us | Français|
Mutation scanning using high-resolution melting curve analysis (HR-melt) is an effective and sensitive method to detect sequence variations. However, the presence of a common SNP within a mutation scanning amplicon may considerably complicate the interpretation of results and increase the number of samples flagged for sequencing by interfering with the clustering of samples according to melting profiles. A protocol describing simultaneous high-resolution gene scanning and genotyping has been reported. Here, we show that it can improve the sensitivity and the efficiency of large-scale case/control mutation screening. Two exons of ATM, both containing a SNP interfering with standard mutation scanning, were selected for screening of 1356 subjects from an international breast cancer genetics study. Asymmetric PCR was performed in the presence of a SNP-specific unlabeled probe. Stratification of the samples according to their probe-target melting was aided by customized HR-melt software. This approach improved identification of rare known and unknown variants, while dramatically reducing sequencing effort. It even allowed genotyping of tandem SNPs using a single probe. Hence, HR-melt is a rapid, efficient and cost-effective tool that can be used for high-throughput mutation screening for research, as well as for molecular diagnostic and clinical purposes.
A key step in the search for potentially pathogenic genetic variants in disease susceptibility genes is mutation screening of coding exons and proximal intronic splice consensus sequences of the entire gene in large subject series. Mutation scanning using high-resolution melting curve analysis (HR-melt) prior to sequencing has been described as an effective, sensitive and economical method to detect genetic variations and to reduce sequencing efforts (De Leeneer, et al., 2008; Reed and Wittwer, 2004; Takano, et al., 2008). HR-melt analysis using unlabeled probes can also be used for genotyping (Liew, et al., 2004; Seipp, et al., 2007; Zhou, et al., 2004). HR-melt relies on the use of LCGreen® Plus double-stranded DNA fluorescent dye and specifically designed instruments for data collection, such as the LightScanner® (Idaho Technology), which can be used for high-throughput analyses. HR-Melt offers several obvious advantages as compared to traditional mutation scanning methods. Not only efficient, this method is secure due to its closed-tube nature, and is amenable to automation for high-throughput mutation discovery. It provides simultaneous acquisition of up to 384 fluorescent melting signals in about 5 min and also fits seamlessly into a resequencing workflow because of its non-destructive nature. A particularly demanding application of HR-melt is analysis of candidate intermediate-risk susceptibility genes by case-control mutation screening, which will often require complete mutation screening of >1000 cases and >1000 controls to achieve reasonable statistical power. However, the presence of a common SNP within a mutation scanning amplicon may considerably complicate the interpretation of results and increase the number of samples flagged for sequencing by interfering with the clustering of melt curve groups according to melting profiles.
Recently, a protocol for simultaneous mutation scanning and genotyping using HR-melt analysis has been described (Montgomery, et al., 2007; Zhou, et al., 2005). The method combines both LCGreen® Plus dye and unlabeled oligonucleotide probes in an asymmetric PCR, leading to simultaneous production of probe-target and whole amplicon double-stranded DNA duplexes that can be analyzed from the same HR-melt run.
In this study, we aimed to apply the method to improve sensitivity and efficiency of large-scale case/control mutation scanning of the ATM gene (GenBank reference sequence NM_000051; MIM# 607585) in some specific situations (Figure 1). We have actually screened the 62 coding exons of ATM in > 1000 subjects, and did not meet any technical difficulties with the vast majority of exons/amplicons when using standard HR-melt mutation scanning (Tavtigian, et al., manuscript in preparation). However, screening of some ATM amplicons illustrates challenges that may be encountered and cost-effective solutions to these challenges. We chose the 36th and 59th coding exons of ATM, both containing a common SNP that interferes with standard HR-melt mutation scanning (Table 1). Mutation screening of the 36th coding exon was challenging because the amplicon contains a common missense SNP (c.5557G>A, allele frequency of 23% in the Caucasian population) adjacent to a less common SNP (c.5558A>T, 1% in the Caucasian population), as well as 2 other rare SNPs (c.5497-15G>C and c.5497-8T>C, 0.5% and 1 % in the Caucasian population respectively). The 59th coding exon amplicon contains an A>C substitution (c.8786+8, 1.7% in the Caucasian population), located downstream of the splice donor site, disturbing standard HR-melt analysis.
For each amplicon, an unlabeled probe was designed to anneal to the region surrounding the common SNP of interest. Stratification of the samples according to their probe-target melting profiles was facilitated by customized LightScanner® software (Idaho Technology). The conceptual idea is that common SNP genotypes are called from the probe-target melting data. Analysis of the whole amplicon melt curve data (e.g., mutation scanning) is then performed separately on heterozygous and homozygous sample subsets to distinguish curve shape differences due to presence of other unknown variants. This approach is particularly valuable for large-scale mutation screening studies, where systematic resequencing of the whole gene in all samples is too laborious and expensive.
Mutation screening was performed on 697 early onset and/or familial breast cancer cases and 659 controls enrolled in an international breast cancer genetics study. These include subjects collected through the Northern California, Australian, and Ontario sites of the Breast Cancer Family Registry (BCFR), subjects collected through the Kathleen Cunningham Foundation for Research of familial Breast Cancer (kConFab, Australia), and subjects enrolled in a Thai case-control study. The mutation screening included in this project had approval by the IARC IRB and the local IRBs of each of the centers from which we received samples. All DNAs were extracted from lymphocyte samples or lymphoblastoid cell lines using standard procedures, then normalized at a concentration of 15 ng/ul and arrayed in 384-well plates. Each plate included negative controls (with no DNA), and a DNA sample from Chimpanzee was added. This sample is used as quality control to assess the efficiency of HR-melt assay for rare sequence variants detection, as the Chimpanzee genome is evolutionarily close enough to the Human genome that almost all amplicons work, but different enough from human that most amplicons will harbor a few sequence variations.
Nested primer pairs were designed to amplify specifically the 36th and 59th coding exons of ATM (NM_000051), including intron-exon junctions, and are available on request. Unlabeled probes were designed to anneal to the SNP of interest, following Idaho Technology’s recommendations. All probes were blocked at their 3’ end by C3-blocker to prevent extension during PCR.
For the 36th coding exon, the unlabeled probe (5’-CCAAGATACAAATGAATCAT-3’) was designed to target the major allele of the common SNP c.5557G>A. This probe also annealed to the major allele of the adjacent SNP c.5558A>T (Figure 2.A). For the 59th codon exon, the unlabeled probe (5’-CAGAAGGTAAGTGATATGAAGTAAAGGAGG-3’) was designed to target the major allele of SNP c.8786+8A>C.
Primary PCR (PCR1) was performed in 8 μl reaction volume containing 30 ng of template DNA, 1.5 mM MgCl2, 200 μM dNTP, 200 nM forward and reverse primers, 0.04 U/μl of Platinum® Taq Polymerase, and 1X PCR buffer supplied by the manufacturer (Invitrogen, Paisley, Scotland). The amplification protocol consisted of 25 cycles with amplification steps at 94 °C, 60 °C, and 72 °C for 30 s each.
Asymmetric nested PCR (PCR2) was then performed in 6 μl reaction volume containing 2 μl of 1:100 diluted PCR1 product, 1.5 mM MgCl2, 132 μM dNTP, 100 nM limiting primer, 500 nM excess primer (primer asymmetry ratio of 1:5), 500 nM unlabeled probe, 0.48X LCGreen® Plus (Idaho Technology, Salt Lake City, Utah, USA), 0.04 U/μl of Platinum® Taq Polymerase, and 1X Buffer (Invitrogen). The amplification protocol consisted of 55 cycles with amplification steps of 94 °C for 30 s, 60 °C for 30 s, and 72 °C for 40 s each. For an optimal efficiency of HR-melt, PCR2 primers were designed to amplify amplicons with a maximum length of 350 bp.
Prior to LightScanner® analysis, PCR2 products were heated to 90°C, then slowly cooled to 20°C to promote heteroduplex formation and detection. Melting was monitored from 35°C to 90°C on a LightScanner® instrument (Idaho Technology). PCR amplification led to simultaneous production of probe-target and whole amplicon duplexes that were analyzed from the same HR-melt run. Since probe-target duplexes are shorter than whole amplicon double stranded DNA duplexes, they melt at a lower temperature. Short probe-target duplexes and larger whole amplicon double-stranded DNA duplexes can therefore be analyzed in two distinct temperature windows (Figure 1).
Genotyping and mutation scanning analyses were carried out using the LightScanner® software (Idaho Technology). The region of the probe melting was analyzed using the “Genotyping” mode and the region of DNA melting was analyzed using the “Scanning” mode (Montgomery, et al., 2007). Stratification of the samples according to their probe-target melting profile was facilitated by a customization of the commercial Lightscanner® software. This new version provides the option to export probe-target melting groups as subsets, which are subsequently used for independent scanning of the amplicons according to probe-target melting profile results.
PCR2 products showing different melting curves from the reference group were sequenced using the BigDye Terminator, version1.1 (Applied Biosystems, Foster City, California) and run on a 96-capillary Spectrumedix Sequencer (Transgenomics, Glasgow, UK) according to the manufacturer’s recommendations.
Many research groups have published studies on mutation screening of the ATM gene and have reported many different rare sequence variants detected with various methodologies. Some have used SSCP with a relatively low sensitivity, others have used DHPLC with a relatively higher sensitivity. Recently, we have completed a large-scale mutation screening of ATM using a relatively new procedure, HR-melt. Although the object of this report is to describe a methodological improvement to the basic HR-melt protocol that enhances the effectiveness of mutation screening, we would like to emphasize that basic HR-melt mutation screening strategy provides good sensitivity for detection of sequence variants. For instance, we assessed the results of 13 studies where different mutation scanning methodologies had been used for mutation detection in ATM (Angele, et al., 2003; Atencio, et al., 2001; Broeks, et al., 2007; Brunet, et al., 2008; Buchholz, et al., 2004; Dork, et al., 2001; Gonzalez-Hormazabal, et al., 2008; Izatt, et al., 1999; Livingston, et al., 2004; Maillet, et al., 2002; Renwick, et al., 2006; Sommer, et al., 2003; Teraoka, et al., 2001; Thorstenson, et al., 2001). When considering only Caucasian subjects for this pooled analysis, 142 carriers of a rare missense variant (i.e. with carrier frequency <1%) were observed among 2661 subjects (mutation detection rate: 142/(2661*3056)=0.000017 variants*subjects-1*codon-1). We compared those results to our data. We identified 101 carriers of a rare missense variant when performing the analysis by HR-melt on 1356 subjects of non-African origin (mutation detection rate: 101/(1356*3056)=0.000024 variants*subjects-1*codon-1). Hence, the rate ratio between our results and the pooled analysis is 1.4 (p-value = 0.008) confirming the higher sensitivity of the HR-melt approach. Nonetheless, we observed that for the two amplicons containing the 36th and 59th coding exons, classic mutation scanning by HR-melt analysis failed to provide clear variant clustering due to the presence of a frequent SNP within the amplicons studied (Figure 2.B and Figure 3.A).
Analysis of the 36th coding exon of ATM was hampered by the presence of several SNPs reported in this amplicon: a common SNP immediately adjacent to a rare SNP (c.5557G>A and c.5558A>T, respectively) and 2 other rare SNPs (c.5497-8C>T and c.5497-15G>C) (Figure 2.A). We assessed an unlabeled oligonucleotide probe designed to anneal to the common G allele at the polymorphic site c.5557G>A and to the common A allele at the adjacent rarer SNP position c.5558A>T. The LightScanner software identified 3 different groups in the “Genotyping” mode (Figure 2.C), which were analyzed independently in the “Scanning” mode. Subset a was scanned for mutations and the automatic call identified 2 melting profiles. Sequencing analysis revealed that all samples from this subset carried the common G allele for c.5557 and some were also c.5497-15 GC heterozygotes (Figure 2.D). Mutation scanning analysis of subset b identified heterozygous samples that were either c.5557GA or c.5558AT. A third group emerging from subset b analysis corresponded to double heterozygotes c.[5557GA(+)5497CT] (Figure 2.E). Subset c revealed 4 different groups in mutation scanning: c.5557AA, c.[5557AA(+)5497-8CT], c.[5557AA(+)5497-8CC], and lastly, c.[5557GA(+)5558AT] (Figure 2.F). Analysis of the whole sample set (1356 subjects) identified one additional rare missense variant (c.5633C>T).
Another example illustrating the difficulty of interpreting HR-melt using the standard mutation scanning mode in the presence of a common SNP is provided by the exon 59 amplicon (Figure 3.A). Within the reference group, sequencing of a few samples with melt curves near the edge of the ‘normal‘ melt curve distribution revealed that some were AC heterozygous for the SNP c.8786+8. Since their melting pattern was hardly distinguishable from the one produced by the “wild-type” group, other samples carrying the same SNP could have been missed, even though our standard practice is to sequence a small fraction of samples from the edge of the HR-melt normal grouping. Moreover, other variants in the vicinity of this SNP might also produce a melting pattern similar to that of the reference group. Thus, to improve the detection of SNP c.8786+8A>C and the detection of new rare nearby variants, an unlabeled probe was designed to hybridize to the common A allele of the SNP. As expected, genotyping allowed distinction of homozygous c.8786+8AA samples from heterozygous c.8786+8AC samples. Analysis of the probe-target melting region in the same 384 samples also identified 2 samples presenting a third distinct profile. One of the samples was c.8786+8CC. The second sample, from the Chimpanzee DNA used as quality control, was homozygous CC for a new variant (c.8786+11) located downstream SNP c.8786+8, and therefore in the probe-target region (Figure 3.B). Both had been missed in the standard mutation scanning analysis initially performed. Groups corresponding to the 3 probe-target melting profiles were further analyzed as individual subsets using the mutation scanning mode. No novel variant was identified in the c.8786+8AA subset (Figure 3.C) nor in the c.8786+8AC subset (Figure 3.D). However, analysis of the whole sample set (1356 subjects) using the simultaneous genotyping and mutation scanning approach succeeded in identifying a total of 4 rare variants (c.8672-43T>C, c.8672-22T>G, c.8741T>C and the chimpanzee variant c.8786+11T>C), which had not been detected or were hardly distinguishable using the standard mutation scanning mode (data not shown). Thus, simultaneous genotyping and mutation scanning substantially improved the characterization of the samples where standard mutation scanning provided ambiguous and not fully reliable results.
Here, we discuss the usefulness of simultaneous genotyping and mutation scanning in the context of large-scale mutation screening projects. The search for new potentially deleterious genetic variants in candidate susceptibility genes requires the screening of coding sequences and splice junctions of entire genes in large sets of cases and controls. HR-melt analysis has repeatedly been reported as an efficient method for mutation scanning. Although it has been reported that different heterozygotes within the same amplicon could be distinguished from each other based on their curve shape differences (Graham, et al., 2005) (Garritano et al. personal communication), mutation screening performed on a large number of samples renders the analysis more complex. Moreover, screening of entire genes often requires the screening of genomic regions containing common SNPs that can interfere with the mutation scan and complicate the interpretation of the results (De Leeneer, et al., 2008). Systematic resequencing of all variant samples is the most common approach to this issue. However, when applied on large series, this latter approach is expensive, laborious and time-consuming (Sevilla, et al., 2002).
Simultaneous genotyping and mutation scanning by HR-melt analysis represents an attractive alternative for high-throughput analysis. The genotyping method was chosen because it could be easily integrated in our existing mutation scanning workflow. Other genotyping methods could have been chosen, but they would have added extra steps to our mutation screening protocol, and would also have required the use of another laboratory instrument. By performing genotyping and mutation scanning simultaneously using HR-melt, we avoided multiple manipulations, and waste of biological material and reagents. Lab contaminations issues were also reduced. For amplicons that contain a common SNP, we postulated that stratification of HR-melt data by common SNP genotype prior to mutation scanning analysis would increase the detection sensitivity for those rare variants, whose melting patterns may be either 1) essentially the same as, and consequently masked by, the melt curve of a common SNP heterozygote; 2) masked by the extra noise present in a large scale melt curve analysis that contains 2 common genotypes; or 3) buried within the melt curve data of an amplicon whose data complexity overcomes the standard software’s ability to group.
Here, we showed that simultaneous genotyping and mutation scanning is suitable to easily distinguish up to nine different genotype combinations, in the case of the 36th coding exon of ATM. Automatic clustering by the analysis software showed complete concordance with sequencing results. In addition, this approach offers the advantage of directly queuing asymmetric PCR products for sequencing. We validated on a series of 90 samples that sequencing reactions from asymmetric PCR products and standard sequencing reactions performed equally.
Study of the 59th coding exon pointed out that the position of a variant within the amplicon and/or the nature of the sequence surrounding the variant are likely to play a critical role on the accuracy of mutation detection by standard HR-melt analysis. Our study provides evidence that in some sequence contexts, some sequence variants may be missed by the classical HR-melt approach, especially when mutation scanning is performed in a 384-well format. This issue has been discussed in a technical assessment of the HR-melt protocol by the UK National Genetics Reference Laboratory (http://www.ngrl.org.uk/Wessex/downloads.htm), and the authors concluded to the existence of sequence variations “intrinsically difficult” to detect by HR-melt.
Simultaneous genotyping and mutation scanning represents therefore a valuable asset since it can easily be integrated in large-scale, high-throughput mutation scanning workflows. Although the risk of missing a rare variant might still remain, this method showed better sensitivity for the identification of novel rare variants, and better accuracy for distinguishing different genotype groups, than the standard HR-melt mutation scanning method. Having validated this approach to screen the 36th and 59th coding exons of ATM efficiently, 8 additional probes were designed to improve the mutation screening of the whole gene in our sample sets. Our general experimental strategy was to design an unlabeled probe for each variant reported to have a frequency >1% in dbSNP database in the regions of interest, in our sample series.
Thus, 10 out 66 ATM amplicons could have been predicted beforehand to require simultaneous genotyping and mutation scanning (15%). We also applied this approach to two additional amplicons during the course of the study to facilitate their mutation screening. The first one contained a SNP found to be common in our sample series (rs3092910:T>C), and the second one contained a novel SNP 54 bp upstream of the 43th coding exon of ATM, that we initially identified using the standard mutation scanning approach (Table 2). For all studied amplicons, cycling conditions (annealing temperature and number of cycles) were optimized in presence of LC Green for mutation screening. Our experience showed that after PCR optimization, none of the ATM amplicons failed to amplify in the presence of LCGreen. For amplicons requiring simultaneous genotyping and mutation scanning, we re-optimized the PCR conditions in presence of the probe. We also verified that the 1:5 primer concentrations ratios would not impair the HR-melt analysis. Initial protocols had to be modified in some cases, especially by adjusting MgCl2 concentration.
Using our strategy, a higher level of confidence in mutation scanning results can be reached when simultaneously proceeding to genotyping using unlabeled probes. We have shown that this approach can dramatically reduce the amount of sequencing required, compared to sequencing all variants that have a melt curve indicative of the presence of a sequence variant, and recommend the method whenever one of three criteria is met 1) the cost of excess sequencing due to the presence of a known common variant in an amplicon will exceed the ~$50-$75 setup cost of the unlabeled probe assay; 2) there is great concern that the presence of a known common SNP will mask the presence of an unknown rare SNP; or 3) it is important, within the mutation screening context, to detect all of the minor allele homozygotes of a common SNP located with an amplicon of interest.
The potential of HR-melt for cost-effective and sensitive high-throughput genotyping and mutation scanning has been reported in numerous studies. For example, Takano et al. and De Leeneer et al. described HR-melt as an economical screening method to detect mutations in BRCA1 and BRCA2 (De Leeneer, et al., 2008; Takano, et al., 2008). In their work, the authors emphasized the advantages, both in time and cost, offered by the use of HR-melt. Cost-effective and rapid methods for screening are indeed highly needed for mutation screening and testing, particularly for molecular diagnostic purposes in medium and low-resources countries. For mutation discovery studies, this technique would also be beneficial since it enables large-scale case-control or population studies at low-cost, but with a sensitivity and an accuracy higher than the current mutation scanning gold-standard, DHPLC (Chou, et al., 2005).
In conclusion, simultaneous genotyping and mutation scanning is another methodology that confirms that HR-melt is a rapid, efficient and cost-effective tool that can be used for high-throughput mutation screening for research, as well as for molecular diagnostic and clinical purposes.
TN-D is recipient of a fellowship from the Fondation de France and SG is recipient of a fellowship from University of Pisa and a Special Training Award from the International Agency for Research on Cancer. This work was supported by the National Cancer Institute, National Institutes of Health under RFA-CA-06-503 and R01-CA121245-01A2, and through cooperative agreements with members of the Breast Cancer Family Registry and P.I.s, including Cancer Care Ontario (U01 CA69467), Northern California Cancer Center (U01 CA69417), and University of Melbourne (U01 CA69638). The content of this manuscript does not necessarily reflect the views or policies of the National Cancer Institute or any of the collaborating centers in the CFR, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government or the CFR. We would like to thank the many families who contribute to the BCFR. We also wish to thank Heather Thorne, Eveline Niedermayr, all the kConFab research nurses and staff, the heads and staff of the Family Cancer Clinics, and the Clinical Follow Up Study (funded by NHMRC grants 145684, 288704 and 454508) for their contributions to this resource, and the many families who contribute to kConFab. kConFab is supported by grants from the National Breast Cancer Foundation, the National Health and Medical Research Council (NHMRC) and by the Queensland Cancer Fund, the Cancer Councils of New South Wales, Victoria, Tasmania and South Australia, and the Cancer Foundation of Western Australia.