High-Resolution DNA Melting Analysis
Luming Zhou,1 Lesi Wang,1 Robert Palais,2 Robert Pryor,1 Carl T. Wittwer1, 3*
1
2
3
Department of Pathology, University of Utah, Salt Lake City, UT 84132, USA.
Department of Mathematics, University of Utah, Salt Lake City, UT, 84109, USA
Associated Regional and University Pathologists (ARUP), Salt Lake City, UT 84109,
USA
*To whom correspondence should be addressed. E-mail: carl.wittwer@path.utah.edu
One sentence summary: High-Resolution DNA melting analysis with unlabeled probes
and saturating DNA dyes provides both mutation scanning of entire PCR products and
specific genotyping in the same closed-tube reaction.
Abstract
Simple DNA melting, when performed rapidly at high-resolution, can
simultaneously scan for mutations and specifically genotype sequence alterations in
DNA. For example, if one or more unlabeled oligonucleotide probes and a saturating
fluorescent DNA dye are included in asymmetric PCR, subsequent melting analysis
without additional processing reveals both PCR product and probe melting transitions.
Unbiased hierarchal clustering of SNPs using either unlabeled probe or PCR product
portions of the melting curve reveal identical genotypes. Multiple unlabeled probes in
one reaction can genotype many sequence variants with simultaneous scanning of the
entire PCR product, vastly reducing the need for re-sequencing.
Melting of the DNA double helix to separate random coils is conventionally
monitored during slow heating with UV absorbance. The process may take hours to
perform, requires large amounts of DNA, and the number of samples that can be
observed is limited. In contrast, techniques that monitor DNA melting with fluorescence
have become popular along with real-time PCR (1). Only a few minutes are required,
PCR produces enough DNA for fluorescence analysis, and many samples can be
analyzed. Dyes that stain double-stranded DNA are commonly used to identify PCR
products by their melting temperature (2). Similarly, different hybridization probe
designs have been developed for genotyping (3-6). Rather than using additional probes
for each allele analyzed, discrimination is by melting temperature (Tm) instead of
fluorescent color (7). Unlike a static “dot blot” at a single temperature, an entire melting
profile is obtained across the region of temperature transition. When performed in the
same tube as amplification, melting analysis provides a homogeneous, closed-tube
system for product identification or genotyping that requires no processing or separation
steps.
The power of DNA melting analysis depends on its resolution. Fluorescent
melting analysis is currently limited in resolution by instrumentation, software and the
fluorescent labels used. Although melting transitions can be observed with covalentlylabeled probes or incorporation of fluorescently-labeled primers (8), dyes that generically
stain double-stranded DNA are attractive for simplicity and cost (9). Recently, a series of
saturating DNA dyes that maximize detection of mismatched duplexes (heteroduplexes)
and high-resolution fluorescent melting instruments have been developed (10).
High-resolution melting analysis for gene scanning analyzes the shape of the PCR
product melting transition. Screening for heterozygous SNPs within products up to 1,000
bp has a sensitivity and specificity of 97% and 99%, respectively (11). In many cases,
the PCR product melting transition also allows genotyping without probes (9, 12).
Greater specificity can be obtained by including unlabeled hybridization probes (6).
Specific genotypes are inferred by correlating sequence alterations under the probes to
changes in the probe melting temperature. We now report simultaneous PCR product and
unlabeled probe melting analysis in the same reaction. When applied to genotyping,
analysis of both PCR product and probe melting regions allows redundant verification of
genotype. When integrated into mutation scanning, common polymorphisms and
mutations can be genotyped, in addition to screening for unexpected, rare sequence
variants. Unbiased, hierarchal clustering is used to group the melting curves into
genotypes.
One, two, or even more unlabeled probes can be used in a single PCR. With one
probe, the genotype of an SNP can be established by both PCR product and probe
melting. With two probes of different melting temperatures, two regions of sequence
variation can be genotyped and the rest of the PCR product scanned for rare sequence
variants. Multiple probes can be used to identify the location and even the sequence of
new unexpected variants. In each case, asymmetric PCR produces both full length
duplex product and excess single stranded product complementary to the probes. When
the mixture is heated, melting transitions for both full length product and product/probe
duplexes occur at different temperatures. Information on materials and methods is
available on Science Online (13).
With a single unlabeled probe, both PCR product melting and unlabeled probe
melting can independently be used to identify SNP genotype (Fig. 1). Human DNA
samples of different genotypes (Factor V Leiden, G>A) were patterned onto a 384-well
plate, amplified by asymmetric PCR in the presence of an unlabeled probe and a
saturating DNA dye, and melted at 0.1°C/s with high-resolution acquisition of
fluorescence and temperature. The entire melting profile, showing both the melting
region of the unlabeled probe and the melting region of the PCR product is shown in Fig.
1A. Two transitions are apparent. Depending on genotype, the probe melts at 58-68°C,
while the PCR product melts at 78-82°C. When only the probe melting region is
considered as a derivative plot (Fig. 1B), wild type samples matched to the probe are
most stable, homozygous mutant samples are destabilized by about 6°C, and
heterozygous samples show both transitions. When only the PCR product region is
considered, the wild type samples are more stable than the homozygous mutants by about
2°C (Fig. 1A). The heterozygous samples are best distinguished by the skewed shape of
the curve and the broad overall transition. The low temperature shoulder results from
heteroduplexes and enables PCR product scanning as previously described (11).
Heterozygotes are easily identified on temperature-overlayed difference plots (14) shown
in Fig. 1C. When unbiased hierarchal clustering is applied to either the PCR product or
the unlabeled probe melting data, identical assignments result (Fig. 1D).
Product melting analysis identifies sequence variants anywhere between two
primers, while probe melting analysis identifies variants under a probe. In Fig. 1, both
analyses give the same result because the sequence variant is both under the probe and
between the primers. If product melting analysis indicates a variant but the probe does
not, then the variation occurs between the primers but not under the probe. Probes can be
placed at sites of common sequence variation so that in most cases, if product scanning is
positive, the sequence variant will be identified by the probes, greatly reducing the need
for sequencing. Multiple probes can be used if they differ in melting temperature and if
each allele presents a unique pattern of probe and/or product melting. For example, many
different mutations in the cystic fibrosis transconductance regulator (CFTR) gene cause
cystic fibrosis, and many additional sequence variations in this gene are polymorphisms
that do not cause disease.
Three SNPs in two regions of exon 11 of the CFTR gene were analyzed with two
unlabeled probes (Fig. 2). Two of the mutations were only six bases apart, allowing one
of the probes to cover two mutations (Fig. 2A). The normalized melting curves (Fig. 2B)
show regions of probe melting (56-74°C) and PCR product melting (80-83°C). On
casual observation, it is not clear from the normalized melting curve what information
can be extracted. However, when the probe region is displayed as a derivative plot (Fig.
2C), the melting transitions of all common alleles under both probes are apparent. Both
unlabeled probes were matched to the wide type sequence, but one of the probes was
made shorter and contained dU instead of dT to decrease its melting temperature (15, 16).
The more stable probe covered a single SNP, resulting in two alleles being separated by
Tm, both being more stable than all alleles of the less stable probe. The less stable probe
covered two SNPs, resulting in three peaks for common genotypes (17). The specific
probe mismatch and its position within the probe affect duplex stability, allowing probe
design that distinguishes multiple alleles. A difference plot of the PCR product melting
transition is shown in Fig 2D. The heterozygous, wild type, and homozygous mutant
samples are clearly different. However, it is difficult to distinguish between the
heterozygotes by PCR product melting alone. Unbiased hierarchal clustering grouped all
heterozygotes together (data not shown). The three heterozygotes are all in the same
SNP class (Liew), resulting in the same heteroduplex mismatches (C:A and T:G) and
homoduplex matches (C:G and A:T). Although predicted stabilities using nearest
neighbor thermodynamics are not identical (18, 19), definitive genotyping required the
use of probes. Product melting easily identifies the presence of heterozygotes, unlabeled
probes may be necessary for genotyping.
Three SNPs and two deletions within exon 10 of the CFTR gene were also
analyzed with two unlabeled probes (Fig. 3). The higher melting probe covered a single
SNP, while the lower melting probe covered two SNPs and two deletions (Fig. 3A). The
normalized melting curves (Fig. 3B) show regions of low temperature probe (56-67°C),
high temperature probe (67-75°C), and PCR product (80-83°C) melting. When the probe
regions are displayed as a derivative plot (Fig. 3C), all five heterozygous genotypes
follow unique paths that allows genotyping and distinguishes them from wild type. Four
of the heterozygotes show resolved peaks, while one is identified by a broad peak
resulting from a relatively stable mismatch (an A:G mismatch near one end of the probe
in an AT-rich region). Allele discrimination does not require a unique Tm for each allele,
only that the curves are different in some region of the melting transition. A difference
plot of the PCR product melting transition is shown in Fig. 3D. The double heterozygote
shows the greatest deviation from wild type because two mismatches are present within
the PCR product. The four single heterozygotes are all easily distinguishable from wild
type. In contrast to exon 11, all heterozygotes could be genotyped by either PCR product
or probe melting. Consideration of both regions provides independent confirmatory
evidence of genotype. As expected, a blinded study of 48 samples with the genotypes
shown in Figs. 2 and 3 were all correctly identified.
Although common sequence variants can usually be genotyped with one or two
unlabeled probes in the same reaction, more than two probes and/or sequential reactions
can also be used. For example, multiple overlapped probes can locate unexpected rare
variants to within the region covered by one probe (20). Once located to a probe,
additional probes can be designed to identify the exact position and sequence of the
variation. It is true that sequencing is a more direct approach for identifying new,
previously unknown variations or if the amplified region is highly variable. However, in
the vast majority of genetic analysis, the amplified wild type sequence is known and
potential common variants are limited. In these cases, specific genotyping can be
obtained by simple DNA melting. No fluorescent oligonucleotides or separations are
required, and both amplification (15 min) and analysis (1-2 min) are rapid. Current realtime instruments are neither required nor desirable because of limited resolution. The
time and cost reduction for genetic analysis is liberating.
References and Notes
1. C. T. Wittwer, N. Kusukawa, in Diagnostic Molecular Microbiology; Principles and
Applications, D. Persing et al., Eds. (ASM, Washington, 2003), pp. 71-84.
2. K. M. Ririe, R. P. Rasmussen, C. T. Wittwer, Anal Biochem. 245, 154 (1997).
3. M. J. Lay, C. T. Wittwer, Clin Chem. 43, 2262 (1997).
4. P. S. Bernard, R. S. Ajioka, J. P. Kushner, C. T. Wittwer, Am J Pathol. 153, 1055
(1998).
5. A. O. Crockett, C. T. Wittwer, Anal Biochem. 290, 89 (2001).
6. L. Zhou, A. N. Myers, J. G. Vandersteen, L. Wang, C. T. Wittwer, Clin Chem. 50,
1328 (2004).
7. C. T. Wittwer, M. G. Hermann, C. N. Gundry, K. S. Elenitoba-Johnson, Methods 25,
430 (2001).
8. C. N. Gundry et al., Clin Chem. 49, 396 (2003).
9. C. T. Wittwer, G. H. Reed, C. N. Gundry, J. G. Vandersteen, R. J. Pryor. Clin Chem.
49, 853 (2003).
10. C. T. Wittwer, V. E. Dujols, G. Reed, L. Zhou, PCT WO 2004/038038 A2
(published patent application).
11. G. H. Reed, C. T. Wittwer. Clin Chem. 50, 1748 (2004).
12. Liew M, et al., Clin Chem. 50, 1156 (2004).
13. Materials and methods are available as supporting material on Science Online.
14. Temperature-overlayed difference plots are obtained from normalized melting curves
as previously described (8, 9). Briefly, a fluorescence range of the normalized melting
curves is selected (usually low fluorescence/high temperature) and each curve is shifted
to overlay a standard sample within this range. Then, the fluorescence difference of each
curve from the average of all wild type curves is plotted at all temperature points.
15. P. D. Ross, F. B. Howard. Biopolymers 68, 210 (2003).
16. M. W. Carmody, C. P. H. Vary. BioTechniques 15, 692 (1993).
17. M. G. Herrmann, S. F. Dobrowolski, C. T. Wittwer. Clin. Chem. 46, 425 (2000).
18. K. J. Breslauer, R. Frank, H. Blocker, L. A. Marky. Proc Natl Acad Sci U S A. 83,
3746 (1986).
19. N. Peyret, P. A. Seneviratne, H. T. Allawi, J. SantaLucia Jr. Biochemistry 38, 3468
(1999).
20. H. Millward, W. Samowitz, C. T. Wittwer, P. S. Bernard. Clin Chem. 48, 1321
(2002).
21. We thank N. Kusukawa for reviewing the manuscript. This work was supported by
grants from the State of Utah Centers of Excellence program, the NIH (GM072419 and
GM073396) and Idaho Technology.
Supporting Online Material
www.sciencemag.org
Materials and Methods
Tables S1, S2
Figure Legends
Fig. 1. High-resolution melting analysis showing both PCR product and unlabeled probe
melting. Three genotypes of the Factor V Leiden SNP (wild type, R506QHet and
R506QHom) were patterned onto a 384-well plate. Asymmetric PCR was performed in
the presence of the saturating DNA dye, LCGreen® PLUS, and an unlabeled probe
covering the SNP (13). Without any processing after PCR, the plate was melted in a
prototype LightScanner™ at 0.1°C/s and 25 readings acquired every °C. The notemplate controls samples (two wells at bottom right) did not show a melting transition.
After baseline normalization, both PCR product and probe melting transitions are
apparent (A). Derivative plots of just the probe region (B) visually cluster by genotype,
as do difference plots of the PCR product melting transition (C). Visual clustering was
confirmed by unbiased hierarchal clustering showing the expected “SNP” pattern on the
plate (D). The same pattern was obtained after analyzing just the probe region (B), the
product region (C), or both (D).
Fig. 2. High-resolution melting analysis of exon 11 of the CFTR gene using two
unlabeled probes. On probe covers two SNPs (G551D and R553X) and the other covers
G542X (A). Melting transitions were observed with the saturating DNA dye, LCGreen I,
on the HR-1 instrument at 0.3°C/s with 65 readings every °C (13). Normalized
composite melting curves (B), probe derivative plots (C), and product difference plots
(D) are shown. Five independent amplifications and melting curves are shown for each
genotype.
Fig. 3. High-resolution melting analysis of exon 10 of the CFTR gene using two
unlabeled probes. On probe covers two SNPs (I506V and F508C) and two deletions
(I507del and F508del) while the other covers Q493X (A). Melting conditions are
described in Fig. 2. Normalized composite melting curves (B), probe derivative plots
(C), and product difference plots (D) are shown. Five independent amplifications and
melting curves are shown for each genotype.
Supporting Online Material
Materials and Methods.
DNA Samples. Human genomic DNA of known Factor V Leiden genotype (1)
was kindly de-identified and provided by ARUP. Heterozygous genomic DNA samples
with selected cystic fibrosis mutations (Table S1) were from the Coriell Institute for
Medical Research. In addition, a G542X homozygote was obtained from the same
source.
Primers and probes. Primer and probe sequences and concentrations are listed
in Table S2 and were synthesized by the University of Utah core synthesis facility.
Predicted probe Tms (2, 3) were lower than observed Tms, perhaps because of dye
stabilization and the non-equilibrium conditions/rates used for melting (0.1-0.3°C/s).
In general, probes that melted lower than PCR extension temperatures work well. The
melting temperature of different probe/allele duplexes can be adjusted by probe length,
mismatch position, and probe dU vs dT content (4, 5).
Asymmetric PCR. PCR for Factor V was performed in 384-well format and 5 µl
volumes, and included 20 ng of genomic DNA in 50 mM Tris, pH 8.3 with 3 mM MgCl2,
0.2 mM each dNTP, 500 µg/ml BSA, 1X LCGreen® PLUS (Idaho Technology), 0.2 U
KlenTaq1™ (DNA Polymerase Technology), and 70 ng TaqStart™ antibody (Clontech).
PCR was performed in a 9700 thermal cycler (ABI) with an initial denaturation at 94°C
for 10 s, followed by 50 cycles of 94°C for 5 s, 57°C for 2 s, and 72°C for 2 s. After
PCR, the samples were heated to 94°C for 1 s and then cooled to 10°C before melting.
PCR for amplification of CFTR exons 10 and 11 was performed in 10 µl volumes
and included 50 ng of genomic DNA in 50 mM Tris, pH 8.3 with 2 mM MgCl2, 0.2 mM
each dNTP, 500 µg/ml BSA, 1X LCGreen I (Idaho Technology) and 0.4 U Taq
polymerase (Roche). The PCR was performed in capillaries on a LightCycler (Roche)
with an initial denaturation of 95ºC for 10 seconds followed by 45 cycles of 95ºC for 1 s,
54ºC for 0 s, and 72ºC for 10 s. After amplification, the samples were heated to 95ºC for
0 s and rapidly cooled to 40ºC before melting.
Melting Acquisition. When samples were amplified on 384-well plates, melting
acquisition occurred on a prototype version of the LightScanner™ (Idaho Technology).
This instrument is similar to the LightTyper® (6) modified for high-resolution melting of
LCGreen dyes. The standard 470 nm light-emitting diodes were replaced with 450 nm
light-emitting diodes (Bright-LED Optoelectronics). In addition, the optical filters were
changed to 425-475 nm excitation and 485 nm long-pass emission filters (Omega
Optical). Temperature homogeneity across the plate, temperature precision and control,
as well as fluorescence precision and frequency of acquisition were all increased over the
LightTyper. The plate was heated from 55 to 88°C at 0.1°C/s with a 300 ms frame
interval, 15 ms exposure and 100% LED power, resulting in about 25 points/°C.
Melting of CFTR exons was performed on the HR-1 high-resolution melting
instrument (Idaho Technology). After PCR, each capillary was transferred to the HR-1
and melted from 50ºC to 90ºC with a slope of 0.3ºC/s, resulting in 65 points/°C.
Melting Analysis. Melting curves were analyzed on custom software written in
LabView (National Instruments). Normalization and background subtraction was first
performed by fitting an exponential to the background before and after the melting
transitions of interest. Derivative plots of probe melting transitions were obtained by
Salvitsky-Golay polynomial estimation as previously described (7). Melting curves of
PCR products were compared on difference plots of temperature-overlayed, normalized
melting curves (8, 9). These analytical methods have been previously applied to
mutation scanning (10-12) and HLA matching (13).
Agglomerative, unbiased hierarchical clustering of melting curve data was
performed by standard methods (14), custom programmed in LabView. The distance
between curves was taken as the average absolute value of the fluorescence difference
between curves at each temperature acquisition. The number of groups was
automatically identified by selecting the largest ratio of distances between adjacent group
transitions.
Table S1. Heterozygous CFTR sequence alterations studied.
Name
Nucleotide position
Amino acid change
exon
Q493X
C to T at 1609
Gln to Stop at 493
10
I506V
A to G at 1648
Ile to Val at 506
10
ΔI507
3 bp deletion between 1648 and 1653
deletion of Ile at 507
10
ΔF508
3 bp deletion between 1652 and 1655
deletion of Phe at 508
10
F508C
T to G at 1655
Phe to Cys at 508
10
G542X
G to T at 1756
Gly to Stop at 542
11
G551D
G to A at 1784
Gly to Asp at 551
11
R553X
C to T at 1789
Arg to Stop at 553
11
Table S2. Primer and probe sequences and concentrations.
Name
Sequence1
[PCR] (µM)
Factor V
100
FV for
CTGAAAGGTTACTTCAAGGAC
0.1
FV rev
GACATCGCCTCTGGG
0.5
FV probe
TGGACAGGCGAGGAATACAGGTT-P
0.4
CFTRx10
201
CFx10 for
ACTTCTAATGATGATTATGGG
0.05
CFx10 rev
ACATAGTTTCTTACCTCTTC
0.5
CFx10 probe 1
GTTCTCAGTTTTCCTGGATTATGCCTGGCAC-P
0.5
CFx10 probe 2
AAUAUCAUCUUUGGUGUUUCCUAUGAUGAAUAU-P
0.5
CFTRx11
1
Size (bp)
198
CFx11 for
TGTGCCTTTCAAATTCAGATTG
0.05
CFx11 rev
CAGCAAATGCTTGCTAGACC
0.5
CFx11 probe 1
CCAAGTTTGCAGAGAAAGACAATATAGTTCTTGGAGAA-P
0.5
CFx11 probe 2
AGGUCAACGAGCAAGAAUUUCUUUAGGU-P
0.5
Positions of sequence variation are shown in bold. A 3’ terminal phosphate is indicated by, “-P”.
References and Notes.
1. M. J. Lay, C. T. Wittwer. Clin Chem. 43, 2262 (1997).
2. K. J. Breslauer, R. Frank, H. Blocker, L. A. Marky. Proc Natl Acad Sci U S A. 83,
3746 (1986).
3. N. Peyret, P. A. Seneviratne, H. T. Allawi, J. SantaLucia Jr. Biochemistry 38, 3468
(1999).
4. P. D. Ross, F. B. Howard. Biopolymers 68, 210 (2003).
5. M. W. Carmody, C. P. H. Vary. BioTechniques 15, 692 (1993).
6. C. D. Bennett, et al. Biotechniques 34, 1288 (2003).
7. C. T. Wittwer, N. Kusukawa, in Diagnostic Molecular Microbiology; Principles and
Applications, D. Persing et al., Eds. (ASM, Washington, 2003), pp. 71-84.
8. C. N. Gundry et al., Clin Chem. 49, 396 (2003).
9. C. T. Wittwer, G. H. Reed, C. N. Gundry, J. G. Vandersteen, R. J. Pryor. Clin Chem.
49, 853 (2003).
10. J. T. McKinney et al. Mol. Genet. Metab. 82, 112 (2004).
11. C. Willmore et al. Am. J. Clin. Pathol. 122, 206 (2004).
12. S. F. Dobrowolski et al. Hum Mutat. 25, 306 (2005).
13. L. Zhou et al. Tissue Antigens 64, 156 (2004).
14. R. O Duda, P. E. Hart, D. G. Stork. Pattern Classification (Wiley, New York, 2nd
ed., 2000).