Real-time closed tube single nucleotide polymorphism (SNP) quantification in pooled samples by quencher extension (QEXT)

Abstract

Quencher extension (QEXT) is a novel single step closed tube real-time method to quantify SNPs using reporters and quenchers in combination with primer extension. A probe with a 5′-reporter dye is single base extended with a dideoxy nucleotide containing a quencher dye if the target SNP allele is present. The extension is recorded from the quenching (reduced fluorescence) of the reporter dye. This avoids the influence of the unincorporated dye-labeled nucleotides, resulting in high accuracy and a high signal-to-noise ratio. The relative amount of a specific SNP allele is determined from the nucleotide incorporation rate in a thermo-cycling reaction. We tested the QEXT assay using five SNPs in the Listeria monocytogenes inlA gene as a model system. The presence of the target SNP alleles was determined with high statistical confidence (P < 0.0005). The quantitative detection limits were between 0 and 5% for the targeted SNP alleles on a background of other SNP alleles (P < 0.05). The QEXT method is directly adaptable to current real-time PCR equipment and is thus suited for high throughput and a wide application range.

INTRODUCTION

A wide range of different single nucleotide polymorphism (SNP) detection methods are currently available (reviewed in 1,2). Commonly, SNP analyses involve PCR amplification and SNP detection. The SNP detection approaches with the highest signal-to-noise ratios are generally based on chain termination with dideoxy nucleotides (3). In addition, the combined use of reporter and quencher dyes is a robust and widely employed way of measuring reactions in real time in closed tube systems. We have combined the benefits of these two principles in the quencher extension (QEXT) assay. QEXT is a novel single step closed tube real-time SNP detection method. The principle of QEXT is that a probe with a reporter dye is single base extended with a dideoxy nucleotide (4) containing a quencher if the target SNP allele is present. The extension is recorded from the quenching (reduced fluorescence) of the reporter dye (Fig. 1A–C).

Figure 1.

Schematic representation of the QEXT method. The DNA is shown as black bars, the DNA polymerase as green ovals (P), the reporter dye as blue circles (R) and the dideoxy cytosine labeled with the quencher as ddC with red circles (Q). (A) Before incorporation of the quencher, the reporter emits fluorescent light after excitation. (B) A DNA polymerase incorporates the quencher containing dideoxy cytosine if the target SNP allele is present. (C) After sequence-specific incorporation of the quencher, the emitted fluorescence from the reporter is quenched. The measured QEXT signal is the reduced fluorescence from the reporter. (D) Normalized QEXT data from the Applied Biosystems 7700 fluorescence detector (see Materials and Methods for details). Each point represents a fluorescence measurement. Results from three samples containing 0, 25 and 100% of a target SNP allele are shown to illustrate the method. The thin lines represent the regression curves, while the bold lines represent the initial tangent (derivative) of the regression curve. (E) The negative of the slope of the tangents for the three samples shown in (d) were plotted against the respective percentages of the target SNPs. Similar standard curves were used for quantification.

In QEXT the amount of target SNP allele is determined from the initial nucleotide incorporation rate in a thermo-cycling reaction (Fig. 1D and E). We obtained a very high signal-to-noise ratio (>20-fold) with the QEXT assay. The quantitative detection limits were between 0 and 5% for the target SNP alleles on a background of corresponding non-target SNP alleles. The QEXT method is directly adaptable to current real-time PCR equipment and is thus suited for high throughput and a wide application range.

There is a diversity of applications with a need for high throughput methods to quantify SNPs in pooled or mixed DNA samples (5). One example is the use of SNPs in epidemiological studies of pathogens. Analyzing SNP frequencies in large sets of pooled samples may help us to better understand the population dynamics of pathogens, while direct analyses of mixed natural samples could be used to study pathogens in the environment. We used Listeria monocytogenes in our model experiments. This bacterium causes severe problems in modern food production and the fatality rate of listeriosis can be as high as 30% (6). Listeria monocytogenes is abundant in the environment, but only a few of the subtypes are actually human pathogens (7). We have recently developed a new DNA array approach for qualitative SNP detection in L.monocytogenes (8) strains. The DNA array approach, in contrast to QEXT, however, is relatively labor intensive and not suited for high throughput SNP quantification in pooled or mixed samples.

MATERIALS AND METHODS

The strains analyzed in this work are listed in Table 1. The DNA was purified as previously described (8). The primers inlA-F (5′-GGAGCTAACCAAATAAGTAACATCAGT-3′) and inlA-R (5′-TATCCGTACTGAAATTCCATTTAGTT-3′) were used to amplify the inlA gene. The reactions included 10 pmol primers, 200 µM each deoxynucleotide triphosphate, 10 mM Tris–HCl (pH 8.8), 1.5 mM MgCl₂, 50 mM KCl, 0.1% Triton X-100 (w/v), 1 U DynaZyme DNA polymerase (Finnzymes Oy, Espoo, Finland) and purified DNA in a final volume of 50 µl. Between 30 and 35 cycles were used in the amplification reactions, with denaturation at 95°C for 30 s, annealing at 55°C for 30 s and extension at 72°C for 30 s. All reactions were initiated with 4 min denaturation at 94°C and ended with a 7 min extension at 72°C. The samples were treated with 4 U shrimp alkaline phosphatase and 20 U exonuclease I at 37°C for 30 min to inactivate the nucleotides and to degrade residual PCR primers. Finally, the shrimp alkaline phosphatase and the exonuclease I were heat inactivated at 95°C for 10 min. The treated products were used as templates in the QEXT reaction.

Table 1. Listeria monocytogenes strains used in this study.

Straina	Originb	Serotype
SS 12067	Human epidemic	4b
SS 9618	Unknown	4
SS 8819	Human epidemic	4b
SS 7751	Unknown	4
SS 5001	Human sporadic	1/2a
SS 7785	Unknown	1
SS 65500	Unknown	1
SS 5223	Unknown	4
MF 762	Fish product	1
MF 3591	Fish	1
MF 1348	Fish processing plant	1
MF 3477	Fish	1
MF 3140	Fish product	1
MF 701	Environment	4
MF 502	Meat	1
MF 3066	Fish	1
MF 3035	Fish processing plant	1
MF 921	Fish processing plant	1
MF 2778	Fish processing plant	1
MF 1037	Environment	1

^aSS, Statens Seruminstitut, Copenhagen, Denmark; MF, MATFORSK Norwegian Food Research Institute, Ås, Norway.

^bThe SS strains are described by Rasmussen et al. (15), while the MF strains are described by Rudi et al. (8).

The 5′ 6-FAM-labeled probes inlA1 (5′-CGCAGCCACT TAAGGCAATTTTTAATG-3′), inlA2 (5′-TCGGCTGGGC ATAACCAAATTAGCGA-3′), inlA3 (5′-GGCTTATATCA CTTATATTATTAAAGTACAA-3′), inlA4 (5′-GGAAAAG GAACGACAACATTTAGTGGAAC-3′) and inlA5 (5′-AAA TGCACCAGTAAACTACAAAGCAAA-3′) were used in the QEXT reaction. The 25 µl QEXT reaction contained 5 µl PCR product (treated as described above), 1× Thermosequenase reaction buffer, 1 pmol each probe, 10 pmol quencher Tamra ddCTP (Perkin Elmer, Boston, MA) and 8 U Thermosequenase (Amersham Biosciences, Little Chalfont, UK). The cyclic labeling was done with denaturation at 95°C for 30 s and labeling at 55°C for 1 min. The labeling was repeated for 40 cycles. The fluorescent spectra from 500 to 660 nm were recorded every 7 s with an Applied Biosystems 7700 fluorescence detector (Applied Biosystems, Foster City, CA) during the thermocycling. The signals were measured in relative fluorescence units (RFU). The QEXT signals were calculated for each cycle during the denaturation phase by subtracting the Tamra signals at 576 nm from the reporter signals (6-FAM) at 541 nm. The Tamra signals were thus used as internal references for normalization of the reporters.

Regression curves were obtained using polynomial regression employing the QEXT signals (Microsoft Excel® 2000; Microsoft Corp., Redmond, WA). The initial slopes of the regression curves were used for quantification. Standard curves were constructed from samples containing known amounts of the given targets. Finally, the amount of SNP alleles in unknown samples was determined from the standard curves. Statistical tools provided in Minitab release 13.30 (Minitab Inc., State College, PA) were used to determine confidence levels.

RESULTS AND DISCUSSION

Five SNPs identified in the L.monocytogenes inlA gene (8) were analyzed by QEXT. This gene was chosen because it is a pathogen determinant and has a high evolutionary rate (9). inlA is important for the invasion of eukaryote cells. The inlA genes were amplified by PCR from the strains shown in Table 1. The amplifications were confirmed by agarose gel electrophoresis (results not shown). The presence or absence of target SNP alleles was determined from the initial quencher nucleotide incorporation rates during thermocycling (examples are shown in Fig. 1D and E). All five SNPs were detected with high signal-to-noise ratios (Fig. 2). The SNPs were confirmed independently by DNA sequencing (not shown). The two-sample t-test showed that for all the probes target and non-target SNP alleles could be detected with a confidence limit of P < 0.0005. The slight signal reduction for some of the non-target SNP alleles may be due to misincorporation by the DNA polymerase.

Figure 2.

SNP scoring for pure isolates. Twenty different strains with known sequences for the inlA gene were analyzed. The bases at the labeling positions are shown in each panel. The mean and standard deviations for the samples containing the target (open bars) and non-target SNPs (shaded bars) are also shown.

The possibility of quantifying SNPs in mixed samples by QEXT was investigated by diluting PCR products containing the targeted SNP alleles in a background of PCR products with non-target SNP alleles. The amounts of templates were kept constant while the ratios of the SNP alleles were varied (Fig. 3). The detection limit was between 0 and 5% for all the SNPs tested. The signal reduction rates for these samples were significantly different (P < 0.05) as determined by a two-sample t-test. Although there was a difference between the 0 and 1% samples for all five SNPs tested, these could not be separated at the P < 0.05 level. In theory, the incorporation rate should be proportional to the amount of target. Linearity was obtained for all the SNPs in the range 0–50%. The squared regression coefficients (R²) were 0.99 for inlA, 0.96 for inlA2, 0.98 for inlA3, 0.99 for inlA4 and 1.0 for inlA5. The incorporation rate was lower than expected for the 75 and 100% samples for inlA3 (Fig. 3). The apparent underestimation for this probe could be due to the high incorporation rate. The initial signal reduction is probably influenced by label saturation, i.e. the probes labeled in previous cycles reduce the amount of probes that can be labeled in the following cycles. The QEXT assay was optimized for sensitivity in our example. We used the maximum amount of PCR products tolerated in the QEXT assay to obtain a high sensitivity. Reducing the total amount of PCR product would increase the accuracy in the cases with high concentrations of target SNPs (>50%), but would give a lower sensitivity.

Figure 3.

Quantification of SNPs in mixed samples. The initial signal reduction is plotted against the percentage of the target SNP allele for probes inlA1–inlA5. The dilution series of PCR products containing the target SNP alleles on a background of PCR products containing the corresponding non-target SNP alleles are shown (squares): inlA1, MF 3140 in MF 701; inlA2, MF 701 in MF 3140; inlA3, MF 3140 in MF 701; inlA4, MF 3140 in MF 3035; inlA5, MF 701 in MF 3140. PCR products from all 20 strains used in this study were also pooled and the signal reduction for the respective probes determined (triangles). Error bars represent standard deviations from three replicates.

The frequency of SNP alleles was determined in pooled samples of PCR products from the 20 strains used in this work. Equal amounts of the PCR products were pooled after amplification (Fig. 3). The expected frequency for inlA1 was 50%, while the determined frequency was 41%. The corresponding values were 55 and 59% for inlA2, 5 and 5% for inlA3, 60 and 63% for inlA4 and 45 and 46% for inlA5, respectively. This shows that the QEXT assay is suitable for quantifying SNPs in pooled samples. The average error was 6.3%. This error includes both the errors in the method and the uncertainties due to variation in DNA amounts when mixing the samples. We did not analyze samples where the bacteria were pooled prior to DNA purification or samples where purified DNAs were pooled prior to PCR amplification. The reason is that we wanted to test the accuracy of the QEXT method and not introduce additional uncertainties (5). Pooling prior to amplification could lead to distortions due to different amplification efficiencies of the SNP alleles. Amplification prior to pooling can in many cases be beneficial in order to standardize the amount of input template.

We have demonstrated that the QEXT assay has a high signal-to-noise ratio and provides a means to accurately quantify SNPs in pooled samples using readily available real-time PCR equipment. 5′-Nuclease PCR is the most widely used alternative technology for analyzing SNPs using real-time PCR equipment. 5′-Nuclease PCR is an integrated PCR amplification and SNP detection system, while these steps are separated in QEXT. The basic difference for SNP detection between 5′-nuclease PCR and QEXT is that the specificity in 5′-nuclease PCR is hybridization dependent, while the discrimination in QEXT is done by the DNA polymerase. It is generally possible to obtain a higher specificity using enzymatic discrimination than using physical discrimination by hybridization. FQ-TDI is a recently developed method where the direct quenching of dyes when incorporated into DNA is measured (10). The quenching in FQ-TDI depends on both the primary and secondary structure of the DNA and, hence, may vary considerably from primer to primer. In the QEXT assay the use of both a reporter and a quencher dye consistently ensures a high signal-to-noise ratio for all primers. In addition, the fact that we record the reporter signal at a different wavelength from that of the quencher-labeled nucleotides also contributes to the high signal-to-noise ratios. QEXT requires dye-labeled primers, which is an additional cost compared to FQ-TDI. The total cost of a SNP detection system, however, is a combination of probe, design and optimization costs, in addition to success rate and quality of the results.

The QEXT method represents a robust, sensitive and simple alternative to existing real-time technologies such as fluorescence polarization (11), real-time PCR (12,13), FQ-TDI (10) and the invader assay (14). Fluorescent polarization detection is sensitive to temperature and viscosity and is not adaptable to current real-time PCR equipment, while the invader assay requires sophisticated probe design. Taken together, a high signal-to-noise ratio, a low detection limit, high accuracy and a closed tube SNP detection system make QEXT especially suitable for high throughput quantitative screenings.