![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
|
Genome Res. 13:1754-1764, 2003 ©2003 by Cold Spring Harbor Laboratory Press; ISSN 1088-9051/03 $5.00 Methods Allele-Specific Holliday Junction Formation: A New Mechanism of Allelic Discrimination for SNP Scoring1 FreshGene, Inc., Concord, California 94520, USA 2 Department of Pathology, Molecular Pathology, UC Davis Medical Center, Sacramento, California 95817, USA
We report here a new mechanism for allelic discrimination—allele-specific Holliday Junction formation. The Holliday Junction (HJ) is a unique DNA structure that can be formed in a sequence-nonspecific manner by routine PCR. To cause the PCR-based HJ formation to occur in an allele-specific manner, the PCR primers are manipulated such that an extra mismatch next to a SNP of interest is introduced between a target and a reference amplicon and a GC-clamp is added. Based on this new mechanism, novel SNP genotyping methods were developed, including a homogeneous fluorescence polarization (FP) competition assay that requires neither labeled primers/probes nor expensive enzymes/substrates. Using this novel genotyping technology, we were able to convert >95% of SNP sequences into genotyping assays that work well under a universal set of assay conditions and achieved 100% accuracy in clinical samples.
SNPs can be detected in either a sequence-specific or sequence-nonspecific way (Kowk 2001  ).
Sequence-specific detection relies on four general mechanisms for allelic
discrimination: allele-specific hybridization, allele-specific nucleotide
incorporation, allele-specific oligonucleotide ligation, and allele-specific
invasive cleavage (Kowk 2001).
We describe here a fifth mechanism for allelic discrimination: allele-specific
Holliday Junction (HJ) formation.
It has been shown by Panyutin and Hsieh
(1993
The Holliday Junction has physical and biochemical properties that are very different from those of duplex DNA or single-stranded DNA. These differences can be exploited in various ways for the detection of Holliday Junctions and the development of novel SNP-scoring methods. We demonstrate here two ways of detecting Holliday Junctions (HJ): a homogeneous, FP (fluorescence polarization) competition assay and a heterogeneous, gel electrophoresis assay. Through the successful development of genotyping assays for 108 SNPs randomly selected from the NCBI SNP database and 11 clinically relevant SNPs, we show that Holliday Junction-based genotyping provides a robust, very inexpensive technology for the scoring of a diverse set of SNPs or mutations with a high degree of accuracy.
Mismatch-Induced Holliday Junction Formation of PCR Amplicons Can Be Detected by Gel Electrophoresis When a forward primer and two tailed reverse primers are used to amplify DNA samples,  50% of the amplicons will be partial duplexes
(Fig. 1B). These partial
duplexes will form Holliday Junctions and undergo spontaneous branch migration
(Fig. 1B). When such PCR
amplicons of one genotype are mixed with the amplicons of a different genotype
at 1:1 ratio, it can be calculated that 6% of the total amplicons should
form stable Holliday Junctions. On the other hand, when the amplicons of one
genotype are mixed with the amplicons of the same genotype, no stable Holliday
Junctions will form. In theory, the presence or absence of Holliday Junctions
should be detectable on gels because Holliday Junctions have much lower
mobility than duplex amplicons. However, such PCR-generated Holliday Junction
bands were never detected on gels before when ethidium bromide was used to
visualize the DNA bands. When more sensitive DNA dyes such as SybreGold and
SybreGreen were used, we were able to detect them directly on gels
(Fig. 2).
PCR-generated Holliday Junction bands were first detected in our early studies when PCR amplicons were relatively long(>120 bp). Deriving PCR amplicons from a heterozygote is essentially the same as mixing the PCR amplicons of one genotype with the amplicons of a different genotype at 1:1 ratio. For each SNP tested, we detected a DNA band that was only present in PCR assays from heterozygotes. Figure 2A shows that this heterozygote-specific band moved more slowly than its respective duplex amplicons and only this band was up-shifted by RuvA and RuvC, two proteins that specifically bind to Holliday Junctions. These data strongly imply that the heterozygote-specific bands are, indeed, Holliday Junction bands. This was confirmed later by using synthetic Holliday Junctions that mimic the PCR-generated HJ in very short amplicons (Fig. 2B). Figure 2B shows that a DNA band of much lower mobility was present when the target amplicons from a C/C homozygote of SNP rs#1551570 were mixed with the T/T reference amplicons (r-T) but absent when mixed with the C/C reference amplicons (r-C; Fig. 2B, lanes 2–9). Moreover, this band was the only band up-shifted by RuvA (Fig. 2B, lane 10), and it had the same mobility as the synthetic HJ on gel (Fig. 2B, lane 11).
Allele-Specific Holliday Junction Formation Requires an Introduced
Extra Mismatch and in Most Cases, the Addition of a GC-Clamp To find out if all types of SNPs (A/T, A/G, C/T, G/T, A/C, and G/C) can induce HJ formation in short amplicons, we generated all possible mismatches by using forward primers (Table 1A, SEQ ID NO 3–7) that contain different "wobbles" at the SNP ss#4129 (NCBI SNP database) and a 6-bp overlap with the reverse-tailed primers (Table 1B, T1-2 and T2-2, SEQ ID NO 31 and 32.) As a result, the wobbles overwrite the naturally occurring C in the genomic DNA template used (a C/C homozygote) and thus generate a series of short amplicons containing all possible SNPs at the same distance from both amplicon ends and in the same sequence environment. Each wobble-containing forward primer was used in combination with the (T1-2/T2-2, 1:1) reverse-tailed primer mix to generate a 67-bp amplicon. The extent to which various mismatches impede branch migration in these amplicons was judged by the intensity of respective HJ bands on 6% PAGE gels (Fig. 3A).
Figure 3A shows that for the 67-bp amplicons, a single A/T, A/G, or A/C mismatch did not produce any detectable HJ at all—judging by the lack of HJ bands on gels and, later, by the lack of inhibition in FP competition assays (data not shown). A noticeable but weak HJ band was produced only by a single G/C mismatch, and the HJ band produced by a single T/G or C/T mismatch is barely visible.
Therefore, a single-base mismatch, A/T, A/G, or A/C, cannot induce
detectable HJ formation in these short amplicons. One possible way to solve
this problem is to introduce an extra single-base mismatch (e.g., A/T, A/G, or
A/C) next to the SNP of interest. This extra mismatch would be, by itself, too
weak to block branch migration but, in combination with the SNP, might be
strong enough to block branch migration and cause detectable HJ formation. To
test this, four target PCR amplicons, each with a predetermined genotype (A,
T, C, or G) at SNP ss#4129, were produced by using the four forward primers
(SEQ ID 13–16) that have a 4-bp overlap with the tailed reverse primers
(T1-2 and T2-2, SEQ ID NO 31 and 32.) At the same time, four reference PCR
amplicons, each with a predetermined genotype (A, T, C, or G) at SNP ss#4129
and an extra T
To test whether the introduction of an extra mismatch can be used widely to
achieve allele-specific HJ formation, we randomly selected Figure 4 illustrates the development of an HJ-based genotyping assay for SNP ss#4130, an SNP with an A/T-rich sequence context. When target amplicons were amplified from genomic DNA sample #1 of the M08PDR panel using primer sets 4130F/(4130T1 + 4130T2) (Table 1A,B, SEQ ID 19, 37, and 38) and then mixed with each of the two reference amplicons amplified using primer sets 4130ref-C/(4130T1 + 4130T2) (Table 1A,B, SEQ ID 22, 37, and 38) and 4130ref-T/(4130T1 + 4130T2) (Table 1A,B, SEQ ID 23, 37, and 38), the introduced extra mismatch alone was not enough to achieve allele-specific Holliday Junction formation (Fig. 5, lanes 1,2) However, when an 8-bp or 7-bp GC-clamp was added to the two redesigned target forward primers 4130F2 and 4130F3, respectively, allele-specific Holliday Junction formation was achieved (Fig. 5, lanes 3–6).
We succeeded in developing gel-based genotyping assays for all the SNPs selected, including108 SNPs randomly picked from the NCBI SNP database and 11 clinically relevant SNPs. Figure 6A shows the data of a typical PAGE gel-based genotyping assay, whereas Figure 6B shows that of a typical agarose gel-based assay. The genotype of the genomic DNA sample was determined by detecting the formation of HJ between the target amplicons and the two reference amplicons. If the genomic DNA was a homozygote of one allele, it only formed Holliday Junctions with the reference amplicon of the other allele (Fig. 6A, samples 1, 3, 6, 7; Fig. 6B, samples 76–80). In contrast, if the genomic DNA was a heterozygote, it formed Holliday Junctions with both reference amplicons, and the HJ bands were about half of the intensity of those formed in homozygotes (Fig. 6A, samples 2, 4, 5; Fig. 6B, samples 70–75).
Holliday Junctions Can Be Detected Homogeneously by an FP Competition
Assay
Figure 7B shows the data of
a typical FP genotyping assay: an HJ-based genotyping assay developed for a
G/A SNP ss#4188 was tested against the 90 genomic DNA samples of the Corriel
Human Diversity Panel. For each sample, the number of Holliday Junctions
formed between the target amplicons and each of the two reference amplicons
was determined by the two FP values of the tracer: FP(+refA) and FP(+refG).
The difference between the two FP values was then calculated: RuvA binds to all Holliday Junctions, including those that formed in an allele-specific manner or those that formed nonspecifically. One source for nonspecific Holliday Junction formation is the genomic DNA template used in PCR. The second source is truncated primers that exist in low-quality (salt-free) tailed reverse primers. These nonspecific Holliday Junctions do not interfere with the detection of the specific HJ band on gel, probably because they are a heterogeneous population with randomly distributed lengths. However, these nonspecific Holliday Junctions do bind to RuvA and interfere with the FP-based HJ detection. Therefore, the homogeneous FP competition assay has stricter requirements than the gel-based heterogeneous assay: First, the genomic DNA concentration used in the PCR should be 0.2 ng/µL or less. Second, the two tailed reverse primers usually need to be PAGE-purified, although desalted primers are always satisfactory for the three forward primers. The only labeled oligonucleotide for the FP homogeneous assay is the 18-bp fluorescein-labeled oligonucleotide used to make the tracer. Because the tracer used in the assay is universal for all SNPs and the working concentration for the tracer is only 300–500 pM, the cost for the tracer is negligible. RuvA protein can be easily and inexpensively manufactured in large scale and can be stored in a refrigerator for more than 6 mo without detectable loss of activity.
We succeeded in developing genotyping assays for 108 SNPs randomly selected
from the NCBI SNP database. These SNPs cover all possible mismatches (G/C,
A/T, A/G, A/C, C/T, G/T) in various sequence contents (from extremely AT-rich
to extremely GC-rich). Among the 108 SNP assays, 95% of them use the universal
PCR/assay conditions outlined in Methods and To determine the sensitivity of the technology and its potential application to diagnostics, we developed genotyping assays for 11 clinically relevant mutations. All 11 SNP assays performed well under a set of universal PCR/assay conditions. Moreover, the genotyping assays for Factor V Leiden, the prothrombin G20210 [GenBank] mutation, and the two MTHFR polymorphisms were each tested blindly on 50 clinical samples (the samples were made anonymous and received an exemption from the Institutional Review Board) obtained from UC Davis Medical Center that had previously been tested using standard genotyping technologies. The DNA concentrations of these 150 clinical samples were unknown, and a 1000x dilution was applied universally for all of them for PCR for the FP-based assays. The allele-specific Holliday structure formation genotyping was found to have 100% accuracy in these samples as compared with the clinical results.
We describe here a new allelic discrimination mechanism— allele-specific Holliday Junction formation. Based on this mechanism, two SNP genotyping methods were developed: a gel-based heterogeneous assay and an FP-based homogeneous assay. Although the two methods rely on different detection modalities, the assay development for them proved to be interchangeable. In addition to the two detection methods illustrated here, it is foreseeable that a variety of other methods and instruments such as capillary electrophoresis, liquid chromatography, or FRET-based assays can also be used for the same purpose.
Like other SNP genotyping methods, the key to successful assay design for
Holliday Junction-based SNP genotyping is the design of PCR primers. On
average, by following the primer design rules outlined in Methods and
illustrated in Figure 4,
successful genotyping assays performed under a universal set of PCR/assay
conditions can be developed for There are already many SNP genotyping methods available today, each equipped with its unique set of desirable attributes. For an example, some heterogeneous assays (such as RFLP) do not require expensive labeled primers but are labor-intensive and time-consuming. In contrast, homogeneous assays (such as the TaqMan or the READIT assay) are fast and can be easily automated but all require expensive labeled primers/probes or expensive enzymes/substrate. We developed here an FP-based, homogeneous genotyping assay that is fast, easily automatable, and, at the same time, does not require expensive labeled primers/probes or enzymes/substrates. In fact, it is the only homogeneous assay available today that uses universal assay conditions and requires neither labeled primers/probes nor expensive enzymes/substrates. However, our FP-based assay is not yet a close-tube assay like the Taqman assay. We are working on the development of a thermostable tracer and RuvA to make our assay a close-tube assay. Not only can Holliday Junction-based genotyping assays be used for scoring SNPs on individual DNA samples, they can also be used to detect allele frequency directly in pooled DNA samples. As shown in Figure 6, the number of HJs formed in heterozygotes is about half of those formed in homozygotes. This is compatible with the fact that, in heterozygotes, only 50% of the target DNA amplicons can form HJs with each of the two reference amplicons. Therefore, the number of HJs formed between the target DNA amplicons and the reference DNA amplicons seems to reflect the allele frequency in the target DNA sample. Using DNA samples with known allele frequency at 10% increments (0%, 10%, 20%, 30%, 40%, and 50%), we were able to show a clear correlation between allele frequency and the relative amount of HJ formation (Fig. 2C). Besides various single-plex genotyping assays that have already been or can be developed based on this new allelic discrimination mechanism, allele-specific Holliday Junction formation also has the potential to be used for the development of genotyping technologies capable of performing multiplex assays in parallel. For example, because the length of the amplicons and thus the size of the HJs formed between them can be arbitrarily determined through primer design, multiple SNPs can be genotyped in a single lane on PAGE gels or capillary electrophoresis if the Holliday Junctions formed for different SNPs are of different size and therefore have different mobility. A more drastic approach, however, will involve performing the allelic discrimination for multiple SNPs in parallel directly on genomic DNA followed by the isolation/purification of the allelic discrimination products—Holliday Junctions using RuvA. The identities of individual Holliday Junctions purified can then be elucidated, for an example, on a micro-array after amplification by PCR. Certainly, many technical obstacles need to be overcome before this potential can be applied to practice. Finally, our FP-based competition assay also has the potential to be used as a homogeneous yet inexpensive method for the detection of gene expression or the detection of pathogens. As long as two partial duplexes have tails that are complementary to each other, a Holliday Junction will form between them regardless of the sequence homology between the duplex portions of the two partial duplexes (Fig. 1A). In fact, the less homology shared between the sequences of the duplex parts of the two partial duplexes, the more stable the Holliday Junction. When two tailed reverse primers are used to amplify a target DNA, part of the target amplicons will be partial duplexes (Fig. 1B). The partial duplex portion of the target amplicons can be easily transformed into Holliday Junctions by annealing with partial duplexes that have complementary tails but heterologous sequence at their duplex part. The resulting HJs can then be detected homogeneously using our FP-based competition assay. The target amplicons, in these cases, can derive from host RNA/genomic DNA or from the RNA/DNA of various pathogens such as HIV, HBV, or HCV.
SNP Assay Design We selected 91 SNPs from the batch of SHGC/AffymetrixChip/Seq-Genomic (from ss#4117 to ss#4210 except ss#4131, ss#4178, and ss#4208) and 17 SNPs from the batch of SHGC/Affymetrix Chip-Genomic (from ss#3985 to ss#4003 except ss#3989) from the NCBI SNP database. Among the four SNPs that were omitted for assay development, ss#4131 does not have the sequence information about its allele position, ss#4178 and ss#4208 have longrepetitive sequences, and ss#3989 was skipped because it had been extensively used before for PCR in our lab and might cause some contamination problem. Of the total SNPs, 11 were clinically relevant: 7  -thalassemia mutations (-28, cd17, cd70/72, IVS-2 nt 654,
cd27/28, cd41/42 and +40/+43), the Factor V Leiden mutation, the Prothrombin
gene G20210
[GenBank]
A mutation, and two common MTHFR polymorphisms (MTHFR677 and
MTHFR1298). For each SNP, a total of five primers were designed for the development of a genotyping assay: three short (18–36 bp) forward primers and two long(40–50 bp) tailed reverse primers (Fig. 4.) For the three forward primers, one was designed to produce the target amplicons with the pair of the tailed reverse primers, whereas each of the other two forward primers was designed to produce reference amplicons with the same pair of tailed reverse primers. The two reference amplicons have known genotypes at the SNP of interest as well as an introduced extra mismatch next to the SNP of interest (Fig. 4.) In particular, each reference primer has at least a 4-bp-long, perfectly matched overlap with the tailed reverse primers so that the reference amplicons have a predetermined sequence (Fig. 4). Depending on whether the extra mismatch is introduced at the base immediately before or after the SNP of interest, the tailed reverse primers begin at (or in other words, their 3' ends terminate at) the base that is 1 base or 2 bases after the SNP of interest, respectively (Fig. 4). This rule for designing tailed reverse primers holds true unless it causes 3'-dimer or hairpin formation. The hairpin and 3'-dimer problems can be solved by the introduction of permutations of the primer at appropriate positions or by moving the 3' ends of tailed primers farther away from the SNP of interest. For the SNPs that are in a very GC-rich sequence environment, no GC-clamp is needed (Fig. 4A). For the SNPs that are in an AT-rich or average GC content environment, a 2–8-bp GC-clamp must be added to both the target and the reference amplicons to achieve allele-specific Holliday Junction formation. The GC-clamp can be added to the target and the reference amplicons by attaching it to all of the three forward primers (Fig. 4B.) In both cases, the target forward primers used to amplify the target DNA ends just before the SNP of interest (Fig. 4A,B). Alternatively, the GC-clamps are added only to the target forward primer during primer design to avoid long reference forward primers (Fig. 4C.) In these cases, the GC-clamp is added to the reference amplicons in a second round PCR using the target forward primer as the forward primer and the amplicons produced by the reference forward primers as the template.
All primers were designed using the software Oligo 6 purchased from
Molecular Biology Insights, Inc. The general rules for primer design are as
follows: (1) the priming part of each primer—the part that actually
hybridizes with the template— should have a Tm above
66°C. (2) The final Tm for each target forward primer
after the addition of a 0–8-bp GC-clamp should be above 78°C. (3) If
the base next to the SNP is G or C, a G A or C A mutation should be introduced
there in the reference forward primers to introduce an A/G or A/C mismatch
between the target and the reference amplicons. If the base next to the SNP is
T, a T
Oligonucleotides
PCR Amplification The target amplicons and each of the reference amplicons were either amplified separately or amplified together in a single PCR reaction. When amplified together, the forward primer of the PCR reaction was a mixture of the target forward primer and one of the reference forward primers at a ratio of 4:1.
Branch Migration
RuvA and RuvC Proteins
Band-Shift Analysis
Tracer
FP Competition Assay Using Analyst AD When LJL HE-96B plates were used, 10 µL of PCR/branch migration products was mixed, in each well, with 10 µL of 0.67 nM tracer before adding 4 µL of a 1:400 dilution of the 50 µM RuvA stock solution. After incubation at room temperature for 10 min, the fluorescence polarization of the samples was measured using an Analyst AD fluorescence plate reader (Molecular Devices Inc.). When LJL HE-384B plates were used, 5 µL of PCR/branch migration products was mixed, in each well, with 3 µL of 1 nM tracer before adding 2 µL of a 1:400 dilution of the 50 µM RuvA stock solution. After incubation at room temperature for 10 min, the fluorescence polarization of the samples was measured using an Analyst AD fluorescence plate reader (Molecular Devices Inc.).
PAGE Gel Analysis
Agarose Gel Analysis
Synthetic Holliday Junctions
The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC section 1734 solely to indicate this fact.
Article and publication are at http://www.genome.org/cgi/doi/10.1101/gr.997703.
3 Corresponding author.
Kowk, P.-Y. 2001. Methods for genotyping single nucleotide polymorphisms. Annu. Rev. Genomics Hum. Genet. 2: 235-258.[CrossRef][Medline]
Lishanski, A. 2000. Screening for single-nucleotide
polymorphisms using branch migration inhibition in PCR-amplified DNA.
Clin. Chem. 46:
1464-1470. Lishanski, A., Kurn, N., and Ullman, E.F. 2000. Branch migration inhibition in PCR-amplified DNA: Homogeneous mutation detection. Nucleic Acids Res. 28: E42. Panyutin, I.G. and Hsieh, P. 1993. Formation of a single base mismatch impedes spontaneous branch migration. J. Mol. Biol. 230: 413-424.[CrossRef][Medline] West, S.C. 1997. Processing of recombination intermediates by the RuvABC proteins. Annu. Rev. Genet. 31: 213-244.[CrossRef][Medline]
Received November 15, 2002;
accepted in revised format April 18, 2003.
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
ABSTRACT
RESULTS
A mutation (mm1) introduced at the base immediately
before the SNP, were produced by using the four forward primers (SEQ ID
17–20) that all have a 4-bp overlap with the tailed reverse primers
(T1-2 and T2-2, SEQ ID NO 31 and 32.) Each of the 4 target amplicons was then
mixed at a 1:1 ratio with each of the four reference amplicons. All 16
possible mixtures were subjected to denaturation and branch migration. The
formation of HJ in these 16 mixtures was then visualized on 6% PAGE gels
(
FP
= FP(+refA) - FP(+refG). For an A/A homozygote such as sample #76, FP(+refA)
is much higher than FP(+refG) and 
