MOLECULAR DIAGNOSTICS
 |
A laboratory technologist performs a DNA melting analysis using an unlabeled probe system in a LightScanner by Idaho Technology Inc. (Salt Lake City), which is a medium resolution instrument with high throughput capability (processing 96 samples at a time).
|
However, both approaches have limitations. Use of a labeled two-probe system (FRET probes) involves high costs and a complex system design. In addition, when labeled primers are employed, heteroduplex reassociation can influence the melt.3 Commonly used intercalating dyes as SYBRGreen I (Molecular Probes Inc.; Eugene, OR) are limited by low resolution due to dye redistribution during melting.5,6
Recently, a high-saturation intercalating dye called LCGreen I was introduced by Idaho Technology Inc. (Salt Lake City). When a molecular size ladder is melted with LCGreen I, all duplexes are clearly separated. In contrast, only the duplexes with high melting temperatures are identified when SYBRGreen I is used.7 The data suggest that LCGreen I is the better dye for DNA melting analysis of multiple products. This dye has been used in genotyping for monitoring the melting of small amplicons and unlabeled probes.7–9 The use of LCGreen I with unlabeled probes allows monitoring of both probe-amplicon duplexes and full-length double-stranded amplicon in the same melting curve. Thus, both genotyping at a specific base and mutation scanning for the full-length amplicon can be performed in the same reaction.
 |
Figure 1. (click to enlarge) A DNA melting analysis is illustrated. From a plot showing the decrease in relative fluorescence (a) of a normal sample (green line) and a heterozygous mutant (blue line) is derived (b) a plot for data presentation showing melting peaks.
|
To further evaluate the feasibility of adapting unlabeled oligonucleotide probes into routine molecular testing, unlabeled 3'-blocked probes for factor V (the Leiden mutation 1691GA) and the mutated cystic fibrosis transmembrane conductance regulator gene called CFTR (F508del) were designed and tested in three different instruments using LCGreen I. Genotyping accuracy was evaluated, as was the potential of using amplicon melting (obtained from the same melting curve) for a secondary genotyping option. The effect of SNP position on melting resolution was examined. In addition, melting resolution (data quality) was observed by comparing melting data acquired from the three instruments employed in the study: a relatively low-resolution LightCycler, a medium-resolution LightScanner made by Idaho Technology, and a high-resolution HR-1, also by Idaho Technology.
As is shown in this article, with proper asymmetric polymerase chain reaction (PCR) and probe design, unlabeled probes can successfully detect both heterozygous and homozygous mutations in all three instruments. The best data quality was obtained with the single-capillary HR-1 at a reasonable throughput of approximately 45 samples per hour.
Experimental Design
To test the design of the unlabeled probes, 10 genotyped and deidentified institutional review board–approved samples were initially used. Later, 100 previously genotyped and deidentified clinical specimens (also institutional review board approved) for each of the mutations were used for probe validation. Genomic DNA was extracted from each sample by means of the automated MagNA Pure LC instrument from Roche Diagnostics, following the manufacturer’s protocol.
 |
Table I . (click to enlarge) Unlabeled oligonucleotide probes used in the study. The underlined bases indicate the Leiden mutation and the three-base deletion in the CFTR gene.
|
Primers specific to each locus (factor V and CFTR exon 10) were designed using Primer3 software and as described in the literature.10–12 Amplicon sizes were 151 base pairs for the factor V gene and 292 base pairs for the CFTR gene. The unlabeled oligonucleotide probes specific to the factor V Leiden mutation (FP1 and FP5) and the CFTR mutation (CP1 and CP2) used in the study (see Table I) were tested for accuracy using known wild types, heterozygous mutations, and homozygous mutations.
PCR was performed using a LightCycler in a 10-µl reaction volume with LightCycler DNA master hybridization probes (Roche Diagnostics). LCGreen I in a final concentration of 1× was added before PCR, along with 100–150 ng of template DNA per reaction.
To obtain a good fluorescent signal, the researchers designed an asymmetric PCR. The final concentrations of a forward primer, a reverse primer, a specific unlabeled probe, and magnesium were 0.05 µmol, 0.5 µmol, 0.5 µmol, and 1 mmol, respectively.
PCR conditions were as follows: initial denaturation at 95°C for 5 minutes, 60 cycles at 94°C for 2 seconds (with a thermal transition rate of 20°C/sec), 56°C for 10 seconds (with a thermal transition rate of 15°C/sec), and 72°C for 20 seconds (with a thermal transition rate of 2°C/sec). The purpose of using 60 cycles was to increase the concentration of single-stranded reverse DNA for better probe binding. Following amplification, PCR products were denatured at 95°C for 30 seconds and rapidly cooled to 35°C at a thermal transition rate of 20°C/sec for heteroduplex formation.
Melting Analysis
The reaction products then were subjected to melting analysis in the three instruments.
The Instruments. The LightCycler is a rapid, medium-throughput instrument that offers 32 capillaries per run. It performs both PCR and melting-curve analysis at a standard melting resolution of about two data points per degree Celsius.
For melting, data were acquired over the transitional range of 45° to 90°C at a thermal transition rate of 0.1°C/sec. Collected data were analyzed using LightCycler software version 3.5 (Roche Diagnostics). Genotypes were identified by the melting temperatures of peaks on derivative plots.
The LightScanner is a high-throughput instrument operating at 96–384 samples per run. It performs melting analysis only, at a melting resolution of approximately 10 data points per degree Celsius. During melting, data were acquired from 45° to 90°C at a thermal transition rate of 0.1°C/sec and analyzed using Idaho Technology’s high-resolution software version 1.1. Genotypes were identified by the melting temperatures of peaks on derivative plots.
The HR-1 has a single glass capillary (GC) and therefore was the lowest-throughput melting instrument of the three used in the study. It also possessed the highest melting resolution—typically, 50 data points per degree Celsius. Data were acquired over 50–95°C at a thermal transition rate of 0.3°C/sec and analyzed using high-resolution software (version 1.1) from Idaho Technology. Derivative melting curves were plotted and genotypes identified in the same fashion as with the LightCycler and LightScanner.
Comparative Analysis. Initially, 10 genotyped samples (5 wild types, 3 heterozygous, and 2 homozygous) for each mutation (factor V Leiden 1691GA and CFTR F508del) were used to optimize unlabeled-probe genotyping in a LightCycler, a LightScanner, and an HR-1 instrument. The researchers found that the unlabeled-probe–DNA duplexes melted at a temperature below 72°C and that the amplicon duplexes would melt at temperatures higher than 75°C.
 |
Figure 2. (click to enlarge) Melting analysis (derivative plot) of the unlabeled factor V probe FP1 using (a) a LightCycler, (b) a LightScanner, and (c) an HR-1 instrument. The green, blue, and red lines indicate a normal, a heterozygous mutant, and a homozygous mutant, respectively.
|
When the FP1 unlabeled probe designed for the detection of Leiden mutation 1691GA was tested, the melting transition distinguished all genotypes (wild types, heterozygous mutation, and homozygous mutation) using any of the three instruments (see Figure 2). The best resolution was obtained from the HR-1.
 |
Table II. (click to enlarge) Melting analysis (derivative plot) of the unlabeled factor V probe FP1 using (a) a LightCycler, (b) a LightScanner, and (c) an HR-1 instrument. The green, blue, and red lines indicate a normal, a heterozygous mutant, and a homozygous mutant, respectively.
|
To address the possible effect on genotype peak separation of the mismatch position within the unlabeled probe, the researchers compared two probe configurations of similar length and GC content with mismatches at the second base (the FP1 probe) or fifth base (the FP5 probe) from the 3' end (see Table I). Results in the LightCycler showed a better melting separation for heterozygotes with the mismatch at the fifth base than for those mismatched at the second base (see Table II). Similar observations were obtained with melting in the higher-resolution LightScanner and HR-1 instruments.
Owing to similarity in melting temperatures and melting-peak patterns between wild types and homozygous mutations, melting transition data for the amplicon duplex from the LightCycler and the LightScanner could not be used for genotyping confirmation (see Figures 2a and 2b). But the amplicon melting transition taken from the HR-1 did differentiate all genotypes (see Figure 2c), with the wild-type control and homozygous mutation being distinguished by melting temperature and the heterozygous mutation being separated from the others by virtue of its unique curve shape. A high-resolution melting therefore is necessary in order for the amplicon melting transition to be used for genotyping confirmation.
For detection of CFTR F508del, the unlabeled probe CP2 was able to distinguish all genotypes in all three instruments (see Figure 3). Here again, the best melting quality was obtained with the HR-1. Melting data acquired with the CP2 probe exhibited a shoulder of unknown import associated with the mismatch peak in the LightCycler test. However, this shoulder was not observed with either the LightScanner or the HR-1. The position of the deleted bases within the probe did not strongly affect allele separation (see Table II).
Similar to the factor V Leiden experiment, in CFTR detection both the LightCycler and the LightScanner showed only minor melting-pattern differences for all genotypes using the amplicon melts alone (see Figures 3a and 3b). Genotypes therefore could not be easily distinguished using only the melting temperatures. With the high-resolution melting provided by the HR-1, heterozygous F508del gave the expected change in peak shape and separated from the wild type (see Figure 3c). However, as seen in the figure, the homozygous F508del mutation and the wild-type control were not easily distinguished. This was because the difference in their melting temperatures is very small, as predicted by the nearest-neighbor calculations.12
 |
Figure 3. (click to enlarge) Melting analysis (derivative plot) of the unlabeled CFTR probe CP2 using (a) a LightCycler, (b) a LightScanner, and (c) an HR-1 instrument. The green, blue, and red lines indicate a normal, a heterozygous mutant, and a homozygous mutant, respectively.
|
Validation. A blinded sample validation followed. For this, the unlabeled probes FP5 and CP2 were selected to test 100 deidentified clinical samples for each locus using both the LightCycler and the LightScanner. The validation procedure excluded the HR-1 because of the manual processing required.
Results from the LightCycler and LightScanner were 100% concordant with genotypes established by the FDA-approved HybProbes kit (Roche Diagnostics) for the factor V Leiden 1691GA mutation and the oligo ligation assay developed by Celera Diagnostics (Alameda, CA) for the CFTR F508del mutation. Overall, the LightScanner exhibited better melting quality (lower noise) than the Light-Cycler. One of the factor V samples containing a rare nontarget variant (1689GA) was detected by a melting-temperature shift in both instruments using the FP5 probe. This variant was originally identified by the HybProbes. The base change and its position were confirmed by bidirectional sequence analysis.
Discussion
Various technologies have been implemented in molecular diagnostics laboratories to identify such genetic alterations as SNPs and small insertions or deletions.13 Among these, DNA melting analysis represents a simple and homogeneous option.14,15
In the simple design described here, using a single unlabeled probe and a saturating DNA dye, the cost and time involved in synthesizing commonly used fluorescently labeled probes for melting analysis, such as HybProbes or Eclipse probes, are eliminated. The use of unlabeled probes and a saturating DNA dye (LCGreen I) was first reported in an engineered plasmid model.9,16 The advantages of using LCGreen I instead of the traditional SYBRGreen I have also been illustrated.17
Probes designed with at least 30 nucleotides, GC content of 40–60%, and an asymmetric PCR are key to the generation of enough single-stranded complementary amplicon for efficient hybridization and sufficient fluorescent signal. The investigators in the present study followed these guidelines. They designed all of the unlabeled probes to have a configuration of 30 nucleotides or slightly more, and around 40% GC content. The ratio of forward to reverse primers and the number of PCR cycles suitable for optimized asymmetric reactions are target dependent. In the study, the researchers used primer ratios of 1:5 and 1:10 and a range of 40–60 PCR cycles. They found that a 1:10 ratio with 60 PCR cycles was an optimal procedure for both the factor V and the CFTR unlabeled-probe assays.
Positional effects—that is, those related to the position of the SNP in the probe—were more apparent when unlabeled probes hybridized with a shorter amplicon (factor V) than when they hybridized with a longer amplicon (CFTR). This observation might be explained by the nearest-neighbor structure of a targeted DNA region affecting the thermostability of the unlabeled probe–amplicon duplexes.18,19 Or, the higher thermostability between longer amplicons may affect hybridization efficiency between unlabeled probes and amplicons. More experiments are needed to determine whether the positional effects are nearest-neighbor-structure related, amplicon size related, or due to a combination of both factors. At this point, a shorter amplicon is recommended to designers of unlabeled probes who are optimizing them to be able to use high-resolution amplicon melts for genotyping confirmation.9
Instrumentation is an important consideration for employing the unlabeled probes in a routine diagnostic laboratory. The study reported here compared three melting instruments with different melting resolutions and different throughput capabilities. In general, a single unlabeled probe performed well in all three instruments. The highest melting-data quality for probe-amplicon duplex melts was always obtained from the HR-1 high-resolution, low-throughput instrument. Next best was the medium-resolution, high-throughput LightScanner and then the LightCycler with its relatively low resolution and medium throughput. In addition, the HR-1 was the only instrument that easily distinguished all genotypes (except the homozygous F508del mutation) using amplicon melts alone.
A high-throughput clinical laboratory may find that the manual processing required by the HR-1 nullifies its advantage in melting-data quality. In addition, PCR has to be performed separately before subjecting melts to the HR-1.
The LightScanner has a lower melting resolution than the HR-1, but its throughput of 96–384 samples per run is much higher. Like the HR-1, the LightScanner performs melts only, and PCR again requires a separate thermocycler, posing a risk of cross-contamination due to opening the sample tray.
The LightCycler possesses the lowest melting resolution and just a moderate throughput of 32 samples per run; however, this instrument offers the significant advantage of a closed-tube system that minimizes cross-contamination. It performs both PCR and melts without any capillaries being opened. A modification in the unlabeled probes is available to improve data quality when a LightCycler is used.11
By incorporating DNA analogs such as locked nucleic acids, the thermostability of perfectly matched pairs can be increased while the mismatched pairs are destabilized for better melting resolution.11,20 All aspects considered, a LightScanner is the best candidate for routine use in a laboratory for high-volume molecular testing.
A scan to detect the existence of rare mutations—most likely to be heterozygotes—outside the probe detection range can be performed by monitoring amplicon melt in addition to probe melt. An HR-1 is needed for this, however, which compromises throughput. Identifying non-target-sequence variations using the amplicon duplexes melt as a secondary genotype confirmation may present challenges for clinical interpretation. Rare variants with no clinical data will have to be categorized as deleterious, benign, or uncertain.
Conclusion
The unlabeled-probe system described in this article provides several advantages for routine clinical testing for SNPs and detection of small genetic insertions and deletions. It offers high specificity and accuracy. The procedure is rapid, taking about an hour from PCR setup to data analysis. Because it requires no fluorescent labeling or sophisticated equipment, the system is inexpensive. Also, it is easy to design and perform in all three of the instruments for genotyping that were tested, without post-PCR manipulation being necessary.
 |
 |
 |
||
 |
 |
|||
Lan-Szu Chou, PhD, is research scientist, molecular genetics; Andrew Lyon is an intern, and Cindy Meadows is technical supervisor, genetics, at ARUP Laboratories (Salt Lake City). Carl Wittwer, MD, PhD, is a professor at the University of Utah school of medicine and medical director of the flow cytometry and advanced technology group at ARUP Laboratories. Elaine Lyon, PhD, is an assistant clinical professor of pathology at the University of Utah school of medicine and medical director, molecular genetics, at ARUP Laboratories. The authors can be reached at choulb@aruplab.com, meadowsc@ aruplab.com, carl.wittwer@path.utah.edu, and lyone@aruplab.com, respectively.
|
||||
1. E Lyon, “Mutation Detection Using Fluorescent Hybridization Probes and Melting Curve Analysis,” Expert Review of Molecular Diagnostics 1, no. 1 (2001): 92–101.
2. CN Gundry et al., “Rapid F508del and F508C Assay Using Fluorescent Hybridization Probes,” Genetic Testing 3, no. 4 (1999): 365–370.
3. CN Gundry et al., “Amplicon Melting Analysis with Labeled Primers: A Closed-Tube Method for Differentiating Homozygotes and Heterozygotes,” Clinical Chemistry 49, no. 3 (2003): 396–406.
4. CD Bennett et al., “The LightTyper: High-Throughput Genotyping Using Fluorescent Melting Curve Analysis,” BioTechniques 34, no. 6 (2003): 1288–1295.
5. S Kleinle, K Tabiti, and S Gallati, “Limitations of Melting Curve Analysis Using SYBR Green I-Fragment Differentiation and Mutation Detection in the CFTR-Gene,” in Rapid Cycle Real-Time PCR—Methods and Applications, ed. W Dietmaier, C Wittwer, and N Sivasubramanian (Heidelberg, Germany: Springer, 2002), 47–56.
6. KS Elenitoba-Johnson and SD Bohling, “Solution-Based Scanning for Single-Base Alterations Using a Double-Stranded DNA Binding Dye and Fluorescence-Melting Profiles,” American Journal of Pathology 159, no. 3 (2001): 845–853.
7. CT Wittwer et al., “High-Resolution Genotyping by Amplicon Melting Analysis Using LCGreen,” Clinical Chemistry 49, no. 6 (2003): 853–860.
8. M Liew et al., “Genotyping of Single Nucleotide Polymorphisms by High Resolution Melting of Small Amplicons,” Clinical Chemistry 50, no. 7 (2004): 1156–1164.
9. L Zhou et al., “Closed-Tube Genotyping with Unlabeled Oligonucleotide Probes and a Saturating DNA Dye,” Clinical Chemistry 50, no. 8 (2004): 1328–1335.
10. S Rozen and HJ Skaletsky, “Primer3 on the WWW for General Users and for Biologist Programmers,” in Bioinformatics Methods and Protocols: Methods in Molecular Biology, ed. S Krawetz and S Misener (Totowa, NJ: Humana Press, 2000), 365–386.
11. LS Chou et al., “Unlabeled Oligonucleotide Probes Modified with Locked Nucleic Acids for Improved Mismatch Discrimination in Genotyping by Melting Analysis,” BioTechniques 39, no. 5 (2005): 644–650.
12. LS Chou, E Lyon, and CT Wittwer, “A Comparison of High Resolution Melting Analysis with Denaturing High-Performance Liquid Chromatography for Mutation Scanning: Cystic Fibrosis Transmembrane Conductance Regulator Gene as a Model,” American Journal of Clinical Pathology 124, no. 3 (2005): 330–338.
13. VN Kristensen et al., “High-Throughput Methods for Detection of Genetic Variation,” BioTechniques 30, no. 2 (2001): 318–332.
14. KM Ririe, RP Rasmussen, and CT Wittwer, “Product Differentiation by Analysis of DNA Melting Curves during the Polymerase Chain Reaction,” Analytical Biochemistry 245, no. 2 (1997): 154–160.
15. RH Lipsky et al., “DNA Melting Analysis for Detection of Single Nucleotide Polymorphisms,” Clinical Chemistry 47, no. 4 (2001): 635–644.
16. WE Highsmith et al., “Use of a DNA Tool-box for the Characterization of Mutation Scanning Methods. I: Construction of the Toolbox and Evaluation of Heteroduplex Analysis,” Electrophoresis 20, no. 6 (1999): 1186–1194.
17. GH Reed and CT Wittwer, “Sensitivity and Specificity of Single-Nucleotide Polymorphism Scanning by High-Resolution Melting Analysis,” Clinical Chemistry 50, no. 10 (2004): 1748–1754.
18. J SantaLucia, HT Allawi, and PA Seneviratne, “Improved Nearest-Neighbor Parameters for Predicting DNA Duplex Stability,” Biochemistry 35, no. 11 (1996): 3555–3562.
19. HT Allawi and J SantaLucia, “Nearest Neighbor Thermodynamic Parameters for Internal GA Mismatches in DNA,” Biochemistry 37, no. 8 (1998): 2170–2179.
20. P Mouritzen et al., “Single Nucleotide Polymorphism Genotyping Using Locked Nucleic Acid (LNA),” Expert Review of Molecular Diagnostics 3, no. 1 (2003): 27–38.
