Template-directed dye-terminator incorporation (TDI) assay: a homogeneous DNA diagnostic
method based on fluorescence resonance energy transfer
Template-directed dye-terminator incorporation (TDI) assay: a homogeneous DNA diagnostic method based on fluorescence resonance energy transfer
Xiangning
Chen*
and
Pui-Yan
Kwok
Division of Dermatology, Washington University School of Medicine, 660 South
Euclid Avenue, Box 8123,
St Louis
, MO 63110,
USA
Received September 17, Revised and Accepted November 13, 1996
ABSTRACT
A new method for DNA diagnostics based on template-directed primer extension and detection by fluorescence resonance energy
transfer is described. In this method, amplified genomic DNA fragments containing polymorphic sites are incubated with a 5
'
-fluorescein-labeled primer (designed to hybridize to the DNA template adjacent
to the polymorphic site) in the presence of allelic dye-labeled dideoxyribonucleoside triphosphates and a modified Taq DNA polymerase (Klentaq1-FY). The dye-labeled primer is extended one base by the dye-terminator specific for the allele present on the
template. At the end of the genotyping reaction, the fluorescence intensities
of the two dyes in the reaction mixture are analyzed directly without
separation or purification. This homogeneous DNA diagnostic method, which we call the template-directed dye-terminator incorporation assay, is shown to be highly sensitive and
specific and is suitable for automated genotyping of large numbers of samples.
INTRODUCTION
DNA analysis is becoming increasingly important in the diagnosis of hereditary diseases, detection of infectious agents, tissue typing for
histocompatability, identification of individuals in forensic and paternity
testing and monitoring the genetic make up of plants and animals in
agricultural breeding programs (
1
). In addition, DNA analysis is crucial in large-scale genetic studies to identify susceptibility alleles associated with
common diseases such as cardiovascular diseases (
2
), autoimmune disorders (
3
) and cancer (
4
). Since each of these applications involves the analysis of a large number of
samples, simple, reliable and highly automated methods of DNA analysis are needed. Although simple tandem repeat
polymorphisms (
5
) have been used successfully in molecular genetic studies, attention is now turning to single nucleotide
polymorphisms (SNPs), the most common DNA sequence variation found in mammalian
genomes (
6
). While most SNPs do not give rise to detectable phenotypes, a significant
fraction of them are disease-causing mutations responsible for genetic diseases, including familial
breast cancer (
7
) and hemochromatosis (
8
). As the DNA sequence of the human genome is completely elucidated, large-scale DNA analysis will play a crucial role in determining the
relationship between genotype (DNA sequence) and phenotype (disease and health)
(
9
). The prospect of large-scale DNA analysis using SNPs is hampered by the cumbersome, gel-based genotyping methods currently employed in their analysis. Even
the relatively high throughput techniques, such as template-directed primer extension (
10
,
11
) and ligation assays (
12
,
13
), require immobilization of DNA followed by detection using either radioactive
reporters or the multi-step enzyme linked immunosorbent assay (ELISA). The use of high density chip arrays for allele-specific hybridization analysis (
14
,
15
) and the homogeneous 5'-nuclease allele-specific oligonucleotide hybridization assay (TaqMan ASO;
16
), both with considerable promise for high throughput, are still in the early
stages of development.
We have developed a novel detection strategy that allows the rapid analysis of
SNPs in a homogeneous assay, eliminating the need for product separation, use
of radioactivity, the multi-step ELISA or specially designed and fabricated oligonucleotide chips. Our
approach combines the specificity of enzymatic discrimination between the two alleles of an SNP in a template-directed primer extension reaction and the sensitivity of fluorescence resonance energy transfer.
Template-directed primer extension is a dideoxy chain terminating DNA sequencing protocol designed to ascertain the nature of the one base
immediately 3' of the sequencing primer, which is annealed to the target DNA
immediately 5' of the polymorphic site. In the presence of DNA polymerase and the
appropriate dideoxyribonucleoside triphosphate (ddNTP), the primer is extended
specifically by one base as dictated by the target DNA sequence at the
polymorphic site. By determining which ddNTP is incorporated, the allele(s)
present in the target DNA can be inferred. This genotyping method has been
widely used and proven to be highly sensitive and specific (
10
,
11
).
Fluorescence resonance energy transfer (FRET) is observed when two fluorescent
dyes are in close proximity and one fluorophore's emission spectrum overlaps
the other's excitation spectrum (
17
). Energy transfer is mediated by dipole-dipole interaction, where the excitation energy absorbed by one dye (donor), instead of being emmitted as fluorescence, is transferred to
the second dye (acceptor) by resonance. Spectroscopically, when the donor is
excited, its specific emission intensity decreases (quenched), while the
specific emission intensity of the acceptor increases. Changes in fluorescence
intensities of the donor and the acceptor can therefore be used as an index for
the distances between the two fluorescent dyes. The efficiency of FRET is very
sensitive to the distance between the two fluorophores. When the distance
between the two dyes increases the efficiency of energy transfer decreases
dramatically. In fact, FRET efficiency is inversely proportional to the 6th
power of the distance between the donor and acceptor dyes (
17
). For fluorophores placed on an oligonucleotide in free solution, however, the
spatial distance between the two dyes is more important than the linear
distance separating them on the oligonucleotide. This is the reason why FRET
has been observed when two dyes are placed 20 bases apart on an oligonucleotide
(
18
). The ability to detect intramolecular FRET against a background of
intermolecular FRET provides a novel and unique detection system that requires
no separation or purification of the product in DNA diagnostic assays such as
primer extension reactions. FRET can be measured either by quenching of the
donor emission, the increase in acceptor emission, or both. FRET has been
exploited successfully in automated DNA sequencing using energy transfer dye primers (
19
), in 5'-nuclease allele-specific oligonucleotide hybridization assays (TaqMan ASO;
18
), in detecting DNA hybridization (
20
) and in the study of protein-protein interaction (
21
).
In our method, named the template-directed dye-terminator incorporation (TDI) assay, the sequencing primer is 5'-labeled with the donor dye (fluorescein) and the ddNTPs
are labeled with an acceptor dye (6-carboxy-X-rhodamine, ROX). FRET occurs when the dye-labeled ddNTP is incorporated onto the sequencing primer
in the presence of DNA polymerase and target DNA. The genotype of the target
DNA molecule can be determined simply by exciting the fluorescein dye on the
sequencing primer and seeing if the acceptor dye exhibits FRET (Fig.
1
).
MATERIALS AND METHODS
Enzymes
AmpliTaq DNA polymerase was purchased from Perkin Elmer Corporation (Foster City, CA) and Klentaq1-FY DNA polymerase was purchased from the laboratories of Dr Wayne Barnes (Washington
University, St Louis, MO).
Oligonucleotides
Oligonucleotides used are listed in Table
1
. PCR and TDI primers were obtained commercially (GENSET Corporation, La Jolla, CA). The TDI primers were 5'-labeled with fluorescein and purified by reverse phase high performance liquid chromatography by
the supplier. Synthetic templates s14102-40A, s14102-40C, s14102-40G and s14102-40T were synthesized by the Genome Sequencing Center at
Washington University (St Louis, MO).
Human genomic DNA (20 ng) from 40 unrelated individuals was amplified in 40 [mu]l reaction mixtures containing 50 mM Tris-HCl, pH 9.0, 50 mM KCl, 5 mM NaCl, 1.5 mM MgCl
2
, 0.2 mM dNTP, 1 [mu]M each primer and 2 U AmpliTaq DNA polymerase. The reaction mixture was held
at 94oC for 3 min, followed by 10 cycles of 94oC for 10 s, ramping to 60oC over 90 s, held at 60oC for 30 s, followed by 30 cycles of 94oC for 10 s and 53oC for 30 s. At the end of the reaction, the
reaction mixture was cooled to 4oC to await further manipulations.
Gel purification of PCR products
The PCR products resulting from amplifying the sequence-tagged sites D18S8 (367 bp) and DXS17 (620 bp) were gel purified by running the samples on a 1% low melting point agarose gel in 1* TAE. After staining with ethidium bromide, the gel bands containing the
PCR products were excised under long wavelength (365 nm) UV transillumination.
The DNA was extracted from the gel slice using the Wizard PCR purification
system (Promega, Madison, WI) according to the manufacturer's instructions.
Dideoxyribonucleoside triphosphates
Dideoxyribonucleoside triphosphates labeled with ROX (ROX-ddA, ROX-ddC, ROX-ddG and ROX-ddU) were obtained from DuPont NEN (Boston, MA).
Unlabeled ddNTPs were purchased from Pharmacia Biotech (Piscataway, NJ).
Genotyping by the TDI assay
Genotyping reactions were performed in 20 [mu]l mixtures containing 50 mM Tris-HCl, pH 9.0, 50 mM KCl, 5 mM NaCl, 5 mM MgCl
2
, 8% glycerol, 0.1% Triton X-100, 25 nM dye-labeled TDI primer, 100 nM ROX-ddNTP and DNA template (>50 nM synthetic 40mer or PCR
products). (Note: 250 nM unlabeled ddNTPs were also added to the reaction
mixture in the reactions using synthetic oligonucleotide templates.) The
reaction mixtures were incubated at 93oC for 1 min, followed by 35 cycles of 93oC for 10 s and 50oC for 30 s. The reaction was stopped by the addition of 10 [mu]l 50 mM EDTA, pH 9.0.
Analysis of TDI reaction products by GeneScan gel electrophoresis
After the TDI reaction, 1.0 [mu]l reaction mixture was added to 5 [mu]l loading buffer (98% formamide, 10 mM EDTA). An aliquot of this mixture (2.0 [mu]l) was loaded onto a sequencing gel (6% polyacrylamide, 8 M urea, 1* TBE) for electrophoresis on an Applied Biosystems 373A automatic DNA sequencer (PE-ABD). The fluorescent species were analyzed using the
GeneScan 672 software (PE-ABD).
Analysis of TDI reaction products by fluorescence spectroscopy
The TDI reaction mixture was denatured and diluted by the addition of 125 [mu]l 0.2 N NaOH. The diluted reaction mixtures were transferred to a 96-well white microplate (PE-ABD) and the fluorescence emission determined with a Luminescence
Spectrometer LS-50B (PE-ABD). The excitation wavelength was set at 488 nm to determine the
emission intensity of fluorescein (FF, 515 nm detection wavelength) and that of
ROX with enhanced emission due to energy transfer (FR, 605 nm). The excitation
wavelength was set at 580 nm to determine the intrinsic emission intensity of
ROX (RR, 605 nm). Fluorescence enhancement (FE) is defined as FR/(RR * FF). A sample is scored as positive for a given allele when it's
fluorescence enhancement value (FE
S
) is significantly higher than that of the controls (FE
C
) at the >99% confidence level. Depending on the particular ROX-ddNTP used and the particular experiment performed, significance at this
level of confidence is achieved when FE
S
is greater than FE
C
by 15-25%.
RESULTS AND DISCUSSION
While 5'-dye-labeled primers and dye-labeled dideoxy-terminators have been used extensively in sequencing
reactions (
23
,
24
) and the sensitivity and specificity of template-directed primer extension genotyping methods are well established when single-stranded templates are used (
10
,
11
), the use of FRET as a detection method in a primer extension reaction using
double-stranded templates has not been reported prior to this work. Furthermore,
we used a member of a new class of mutant Taq DNA polymerase (
25
) that incorporates dye-labeled ddNTPs much more efficiently than the wild-type Taq DNA polymerase in our assay, instead of the Klenow fragment of
Escherichia coli
DNA polymerase I or T7 DNA polymerase commonly used in primer extension reactions (
10
,
11
). Two sets of experiments were performed to show that FRET is a simple, highly
sensitive and specific detection method in a primer extension reaction for
single base pair changes. In the first set of experiments, four synthetic
templates containing the four possible nucleotides at one particular site in
the middle of otherwise identical oligonucleotides were used to establish the
sensitivity and specificity of FRET detection of dye-terminator incorporation. In the second set of experiments, 80 PCR
products amplified from genomic DNA were used as templates in the TDI assay to
show that accurate genotyping data were produced efficiently by this assay.
DNA typing by the TDI assay using synthetic templates
DNA typing by the TDI assay using PCR products as templates
To demonstrate that double-stranded DNA samples could be used as templates in the TDI assay, genomic
DNA derived from 40 unrelated individuals was amplified and the genotype
determined by the TDI assay. Two polymorphic sequence-tagged sites, DXS17 (
26
) and D18S8 (
27
), were used to validate our assay. Gel purified PCR products amplified from
genomic DNA of 40 unrelated individuals served as templates and were placed in
two parallel TDI reactions containing the fluorescein-labeled sequencing primer and one of the allelic ROX-labeled terminators. As shown in Table
2
, all but two samples among the 40 tested for the DXS17 locus gave definitive
genotypes with the positive threshold set at FE
S
:FE
C
> 1.25. The two DNA samples that yielded no definitive genotypes were analyzed
by agarose gel electrophoresis and were shown to have very weak product bands,
indicating suboptimal PCR amplification to be the reason for the false negative
results. For the D18S8 locus, all 40 DNA samples were amplified successfully
and the PCR products were analyzed in the TDI assay as before. Table
3
shows that all 40 samples yielded definitive genotypes when the positive
threshold is set at FE
S
:FE
C
> 1.15. These results can be visualized by plotting the ratio of FF between the
two alleles against the ratio FR:RR between the two alleles (Fig.
3
). In the reaction containing the dye-terminator specific for allele 1, a sample homozygous for allele 1
displays reduced FF and RR values due to quenching and an enhanced FR value due
to FRET. In contrast, in the reaction containing the dye-terminator specific for allele 2, a sample homozygous for allele 1
displays minimal change in all three values as compared with negative controls.
A sample homozygous for allele 1 (ddC) therefore has a high (FR:RR)
1
/(FR:RR)
2
value and a low (FF)
1
/(FF)
2
value (left upper corner of the plot in Fig.
3
). Similar arguments predict that a sample homozygous for allele 2 (ddU) will
have a low (FR:RR)
1
/(FR:RR)
2
value and a high (FF)
1
/(FF)
2
value (right lower corner of the plot in Fig.
3
). Heterozygous samples have intermediate values (close to 1.0) for both ratios
and occupy the region in between the two extremes in the plot. The genotypes
for all 78 DNA samples determined by the TDI assay are in perfect concordance
with those determined by the highly accurate oligonucleotide ligation assay (
12
).
a
See Figure 2 legend for definition of abbreviations.
Based on these and other results, the false negative rate of the TDI assay is
estimated to be ~5%, with all failures attributable to suboptimal PCR amplification leading
to DNA concentrations below the detection limit. The false positive rate is ~1.5%, with contamination being the cause for such results (data not shown).
We have previously shown that Taq DNA polymerases incorporate dye-terminators at a variable rate, depending on the local sequence context,
in the presence of dNTPs (
28
,
29
). Because there are no dNTPs in the TDI assay to compete with the dye-terminators, local sequence context imposes no noticeable influence on the
assay. This is the case for the marker DXS17, where the two alleles are T and C
following the base string TT. Based on our previous work on dye-terminator sequencing peak height patterns using the Klentaq1-FY family of DNA polymerase, T incorporation following TT is inefficient while C incorporation
following TT is highly efficient in the presence of dNTPs, giving C peaks that
are two to three times the size of T peaks in the sequencing chromatogram (
29
). For the six DNA samples heterozygous for DXS17 (samples 2, 8, 14, 30, 38 and 40, Table
3
), however, the average FE
S
:FE
C
values for ROX-ddC and ROX-ddU are 2.33 and 1.98 respectively, a 15% difference that is
attributable to the difference in fluorescence properties of the dye-terminators rather than to a difference in the efficiency of dye-terminator incorporation.
Demands for genetic testing (i.e. assaying for the presence or absence of known
DNA polymorphisms or mutations) are expected to increase dramatically in the
areas of diagnostics, forensics and population studies. For example, population
studies involving thousands of individuals and hundreds of markers were performed recently to localize genes important in the development of type I diabetes mellitus (
30
). A homogeneous genotyping assay is highly suitable for large-scale genetic studies because it is not limited by a particular reaction
format and it offers the flexibility of using the best markers as they become
available for a particular application without redesigning or refabricating
high density DNA chips. Furthermore, the TDI assay is simple to set up (by
adding the standard reagent mixture to the DNA template), the results are
obtained in electronic form minutes after the allele-discriminating reaction is performed and the genotype can be assigned automatically by the use of a simple computer program.
With proof of the principle of FRET detection completed, studies are now
underway to combine PCR and TDI to achieve a one-tube reaction without the need for purification of PCR products from the
unreacted primers and dNTPs, which will interfere with the TDI reaction.
Moreover, a number of other dyes (including TAMRA, JOE and Texas Red) can act
as acceptors for fluorescein, making it possible to perform TDI for both
alleles in the same reaction vessel. Furthermore, since the principle of FRET
applies to any donor-acceptor pair, including those absorbing in the infrared region, one can
utilize near infrared donor-acceptor pairs that can be assayed in commonly used plastic ware with no concern for interfering background emissions. As DNA diagnostic tests will no doubt be performed more and more by
clinical rather than research laboratories, methods (such as the TDI assay) utilizing standard protocols that require minimal laboratory skills or manual handling will be crucial to the clinical practice
of medicine in the future.
ACKNOWLEDGEMENTS
We thank Dr Phil Buzby of DuPont NEN for generous gifts of various dye-labeled ddNTPs used in this study and Dr Daniel E.Goldberg and Perkin-Elmer Corp. for the use of the Luminescence Spectrophotometer LS-50B. We also thank Irma Bauer-Sardina, Hamideh Zakeri and Neha S. Shah for technical
assistance. This work was supported in part by a grant from the US Department of Energy (DE-FG06-94ER61909) to P.-Y.K. and a National Center for Human Genome Research NRSA
grant (1-F32-HG00156-01) to X.C.
