Figures
Abstract
Genotyping of thiopurine S-methyltransferase (TPMT) is recommended for predicting the adverse drug response of thiopurines. In the current study, a novel version of allele-specific PCR (AS-PCR), termed competitive real-time fluorescent AS-PCR (CRAS-PCR) was developed to analyze the TPMT*2 genotype in ethnic Chinese. This technique simultaneously uses wild-type and mutant allele-specific scorpion primers in a single reaction. To determine the optimal conditions for both traditional AS-PCR and CRAS-PCR, we used the Taguchi method, an engineering optimization process that balances the concentrations of all components using an orthogonal array rather than a factorial array. Instead of running up to 264 experiments with the conventional factorial method, the Taguchi method achieved the same optimization using only 16 experiments. The optimized CRAS-PCR system completely avoided non-specific amplification occurring in traditional AS-PCR and could be performed at much more relaxed reaction conditions at 1% sensitivity, similar to traditional AS-PCR. TPMT*2 genotyping of 240 clinical samples was consistent with published data. In conclusion, CRAS-PCR is a novel and robust genotyping method, and the Taguchi method is an effective tool for the optimization of molecular analysis techniques.
Citation: Chen D, Yang Z, Xia H, Huang J-F, Zhang Y, Jiang T-N, et al. (2014) Enhanced Specificity of TPMT*2 Genotyping Using Unidirectional Wild-Type and Mutant Allele-Specific Scorpion Primers in a Single Tube. PLoS ONE 9(4): e91824. https://doi.org/10.1371/journal.pone.0091824
Editor: Chiyu Zhang, Institut Pasteur of Shanghai, Chinese Academy of Sciences, China
Received: September 22, 2013; Accepted: February 16, 2014; Published: April 4, 2014
Copyright: © 2014 Chen et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported in part by the Transformation Foundation on Scientific and Technological Achievements of The Third Military Medical University [2012XZH03]; The Nature Scientific Foundation of China [31370853]; The National High Technology Research and Development Program of China [2011AA02A121 and 2013AA020204]; Preferential Foundations of Science; and Technology Activities for Students Study Abroad of Chongqing [YuLiuZhu201201]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: A patent describing the use of TPMT*2 CRAS-PCR system for clinical diagnostic purposes is currently under review (State intellectual property office of The P.R.C, Application ID: 20140028809.5). This does not alter our adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors.
Introduction
Since their invention [1], thiopurine drugs, including azathioprine (AZA), 6-mercaptopurine (6MP), and 6-thioguanine (6TG), have significantly advanced the treatment of hematologic malignancies, organ transplantation, inflammatory bowel disease (IBD) and some autoimmune diseases (e.g., rheumatoid arthritis, autoimmune hepatitis) [2]. Thiopurine drugs are metabolized by thiopurine methyltransferase (TPMT) [2], which may have single nucleotide polymorphisms (SNPs), resulting in null or decreased enzyme activity [1], [2], [3], [4], [5]. Since the initial identification of SNPs in TPMT, more than 30 mutant alleles of the gene have been identified, of which TPMT*2, *3A, *3B, *3C, and *4 are the most common [1], [2], [3], [4], [5]. About 3–14% of patients have a heterozygous TPMT genotype with one variant TMPT allele [1], [2], [3], [4], [5]. Independent and professional groups have recommended analyzing the TPMT genotype status to determine thiopurine drug dosing [5], [6], [7], [8], [9]. Accordingly, the Clinical Pharmacogenetics Implementation Consortium (CPIC) recommends that the initial dosage of AZA or 6-MP be reduced by 30–70% for patients who possess heterozygous genotypes of TPMT*2, *3A, *3C, or *4 [6], [7]. Therefore, TPMT genotyping may be applicable for personalized therapy to predict the adverse response to thiopurine drugs.
Among the methods of TPMT genotyping, restriction-fragment length polymorphism (RFLP)-southern blot assay, RFLP-PCR, oligonucleotide ligation assay, and allele-specific PCR (AS-PCR) or the amplification-refractory mutation system (ARMS) have been used. AS-PCR is one of the most common methods and uses (AS) primers or probes labeled with fluorescent reporter dyes (usually hydrolysis probes or dual hybridization probes) to identify the genotypes of the targeted SNPs [10], [11], [12], [13], [14], [15]. Because only a single nucleotide difference exists between the defined and opposite genotypes, allelic determination is often hampered by cross-hybridization between either defined genotypic probes and opposite amplicons, or defined genotypic AS primers and opposite templates [16], [17], [18], [19], [20]. Consequently, stringent reaction conditions (e.g., high annealing temperatures, low AS primer concentrations) are needed to avoid non-specific genotypic determination. However, the optimization of traditional AS-PCR is costly and time-consuming, and optimization cannot always completely eliminate non-specific amplification (7). Therefore, novel strategies to avoid the non-specific genotypic determination of AS-PCR are needed.
Generally, molecular biologists optimize assays using factorial methods, which involve testing all the levels of all factors against one another, requiring numerous experiments [21], [22], [23]. As an alternative approach to reduce time and effort, the Taguchi method [21], [22], [23], was first applied to molecular biological methodologies for the optimization of traditional end point PCR in 1994 [24]. The Taguchi method has also been used for DNA amplification fingerprinting [25], end-point multiplex PCR [26], [27], [28], real-time quantitative PCR using SYBR Green chemistry [29] and fluorescence resonance energy transfer (FRET) probes [30]. Although this method has been effective for several molecular assays, it has yet to be demonstrated for AS-PCR [31].
The Taguchi method has been widely used in automotive and electronics industrial design, principally in development trials, for the establishment of optimal conditions for a particular process using a minimal number of experiments [28], [32], [33], [34], [35]. This method is applied towards experimental objectives that are “nominal-the-best”, “smaller-the-better”, “larger-the-better”, or “zero-point proportional” [21], [22], [23]. Quality loss functions (signal-to-noise ratios; S/N ratios) penalize deviations from prediction values, and consequently, orthogonal arrays can be designed to examine multiple factors in few experimental trials [21], [22]. Thus, a trial investigating the effects and interaction of five PCR reaction components (control factors) each at four levels would require an experiment with 1024 (i.e., 45) separate experiments if tested by traditional array design. Using the Taguchi method, an estimate of the effect of each component could be arranged into an L16(45) orthogonal array that only requires 16 experiments.
In the present study, a novel method of AS-PCR, termed competitive real-time fluorescent AS-PCR (CRAS-PCR), was developed to analyze the genotypes of the TPMT*2 variant (dbSNP: rs1800462G>C; p.A80P; c.238G>C; g.11421G>C). This method uses wild-type allele-specific forward (WT-ASF) and mutant allele-specific forward (MT-ASF) scorpion primers in a single reaction mixture. The Taguchi method was used to optimize reaction components and evaluate reaction characteristics of both traditional AS-PCR and CRAS-PCR. Compared with traditional AS-PCR, non-specific amplification can completely be eliminated in CRAS-PCR following optimization by the Taguchi method. Preliminary application of CRAS-PCR in the analysis of 240 clinical samples further confirmed CRAS-PCR as a routine, reliable and simple method to analyze SNP genotypes, including the TPMT*2 variant.
Materials and Methods
Materials and equipment
The traditional end-point PCR and real-time PCR instruments used in this study were the GeneAmp PCR System 2700 (Applied Biosystems, Foster City, CA, USA) and Exicycler™ 96 Real-time Quantitative Thermal Block (Bioneer, Daejeon, Korea). Real-time PCR data were collected and analyzed using Exicycler™ 96 Software (Bioneer). Ex Taq Hot Start Version DNA Polymerase (TaKaRa, Dianian, China) and a MutanBEST Kit (TaKaRa) were used to construct wild-type quality control (WT-QC) and mutant-type quality control (MT-QC) plasmids. Real-time PCR was performed with 2× master mix for Premix Ex Taq (Perfect Real Time; TaKaRa). QC plasmids were sequenced using a BigDye Terminator V3.1 Cycler Sequencing Kit (Applied Biosystems, Foster City, CA, USA) and the ABI Prism 3500 Genetic Analyzer (Applied Biosystems). Oligonucleotide sequences were synthesized at Sangon Biotech Co., Ltd. (Shanghai, China; Table 1). Genomic DNA of human whole blood was extracted using the QIAamp DNA Blood Mini Kit (QIAGEN GmbH, Hilden, Germany). GelRed (10,000×; Biotium Inc., Hayward, CA, USA) was diluted to 100× concentrations in 6× Loading Buffer (TaKaRa, Dalian, China) for nucleic acid staining following agarose gel electrophoresis [36]. Images of stained gels were captured with a UV transilluminator (Vilber Lourmat, Marne la Valled, France).
Samples and genomic DNA extraction
This study was approved by the institutional review board and ethics committee of Southwest Hospital (Chongqing, China). Human blood samples were collected from ethnic Han Chinese volunteers living in the Chongqing territory. Written informed consent was obtained prior to collection. Coagulation of the blood samples was prevented with dipotassium ethylenediamine tetraacetic acid (K2-EDTA, 1.5 mg/ml blood). Genomic DNA was extracted using the QIAamp DNA Blood Mini Kit (Qiagen GmbH) according to the manufacturer's instructions. The genomic DNA concentration of each clinical sample was determined using the Quant-iT™ PicoGreen dsDNA Assay Kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions.
Construction of quality control (QC) plasmid
The PCR reaction mixture (20 μl) used to amplify whole exon 4 fragments containing TPMT*2 included 1× Ex Taq Buffer (Mg2+ plus; TaKaRa), 250 μM of each dNTP (TaKaRa), 0.5 units of Ex Taq HS (TaKaRa), 200 nM each of forward and reverse primer (HQ-589 and HQ-590 in Table 1) and 20 ng of human genomic DNA. Reactions were performed as follows: denaturation at 94 °C for 5 min; 35 cycles of 94 °C for 30 sec, 55 to 65 °C for 30 sec and 72 °C for 30 sec; and final extension at 72 °C for 5 min. DNA sequencing of the PCR products was performed using the BigDye Terminator V3.1 Cycler Sequencing Kit (Applied Biosystems). All sequencing analyses were performed at least twice in duplicate, and both forward and reverse sequencing were performed. After the wild-type allele of TPMT*2 was confirmed, the PCR products were cloned to construct the WT-QC plasmid. The WT-QC plasmid was subsequently used to prepare the MT-QC plasmid using the MutanBEST Kit (TaKaRa) according to the manufacturer's instructions.
Real-time PCR
Three types of primers were designed: WT-ASF and MT-ASF scorpion primers targeting WT and MT TPMT*2 alleles, and a common reverse (CO-R) primer (Table 1). For each Assay Type, 1×104 copies were used of plasmids WT-QC, MT-QC, or mixed QC (MIX-QC; an equal ratio of WT- and MT-QC).
The real-time PCR mixture (20 μl) contained 1× Premix Ex Taq master mix (Perfect Real Time; TaKaRa), one of the three QC plasmid types, and variable concentrations of primers (Table S1 to S3). Reactions were performed with the following cycling conditions: initial denaturation of 95 °C for 30 sec; followed by 40 cycles of 95 °C for 30 sec and the desired annealing temperature (Table S1) for 30 sec (with single fluorescence acquisition). All temperature transition ramping rates were at 2.5°C/sec. The quantification cycle (Cq) values were automatically determined by Exicycler™ 96 Software (Bioneer). All reactions were performed at least in duplicate with no-template control to monitor for contamination and non-specific products.
The Taguchi method
The cycle number at which the fluorescence significantly differs from the background noise (the Cq) is inversely proportional to the log of the initial quantity of DNA [16], [37], [38]. For each of the three Assay Types (Table 2), the Cq was selected as the response variable, except for Assay No. 3-3, for which the difference in Cq values (ΔCq) between WT and MT was selected as the response variable. Four factors were optimized at four levels (Table S1); therefore, according to the Taguchi method, an L16(45) orthogonal array was designed (Table S2, S3). All calculations of the Taguchi method were done by inputting the formulae described below (Eq. 1–8) and corresponding values into a Microsoft Excel spreadsheet.
An optimized assay should have the lowest possible Cq (or the lowest possible ΔCq for Assay No. 3-3); therefore, the “smaller-the-better” equation (Eq. 1) was employed to calculate the S/N ratio (η) using an experimentally determined Cq, except for Assays No. 1-2 and 2-1. For these assays, the “larger-the-better” equation (Eq. 2) was used to evaluate and avoid the non-specific amplification of WT-ASF and MT-ASF primers on opposite templates. In the absence of an amplification signal (i.e., no Cq values determined by Exicycler™ 96 Software), a manual Cq value of 40 was assigned. The “smaller-the-better” and “larger-the-better” S/N ratio were calculated as follows:(1)
(2)where η = the S/N ratio, r = the number of repeated runs, y = the response (Cq or ΔCq).
The percent contribution (PC) of each factor to the total variations observed in each experiment was calculated as follows [21] (Eq. 3 to 5):(3)
(4)
(5)where SST = the total sum of squares, n = the number of experiments in the orthogonal array, ηi = the S/N ratio of each experiment, and
= the average S/N ratio of all experiments; SSX = the sum of squares of the signal-to-noise ratio of factor X, in which f = the number of factors, RXi is the number of runs conducted at level i of factor X;
= the average S/N ratio for each level of each factor; VE = the variance of error; dofX = the degree of freedom of factor X. The F-ratio and p-value of each factor were further analyzed by Analysis of Variance (ANOVA).
To predict the S/N ratio of a reaction carried out with optimal conditions, the following equation was used [21] (Eq. 6):(6)where ηm = the overall mean of S/N ratios at the optimal level for all factor combined,
= the mean S/N ratio at optimal levels of each factor i.
The 95% confidence interval at optimal conditions was calculated as follows [21] (Eq. 7):(7)where F(95%,1,dofE) = the first F value with the degree of freedom equal to 1 and the degree of freedom of error (dofE) as the second degree of freedom at 95% confidence; VE = the variance of error; neff = the effective sample size determined by N/(1+dofTF), where N = the total number of experiments, and dofTF = the degree of freedom of all factors combined.
As confirmation, the S/N ratio should fall within the 95% confidence interval, given that a prediction model is suitable. The confidence interval was calculated using the following equation [21] (Eq. 8):(8)where nconf = the number of confirmatory tests conducted.
Preliminary application of the CRAS-PCR system to clinical samples
WT-, MT- and MIX-QC plasmids at 10-fold serial dilutions (i.e., 1×108 to 1×101 copies) were further used to evaluate optimized CRAS-PCR conditions obtained from the Taguchi method. The optimized CRAS-PCR conditions were then used to analyze the TPMT*2 genotype of 240 clinical samples in which 50 to 150 ng of genomic DNA were used in reaction mixtures. Using a conversion factor of 6.6 pg per diploid human cell [39], [40], this concentration equals 1.5×104 to 4.5×104 copies of the corresponding genotypic allele. The Cq values of each sample were assessed within CY5 and 6-FAM fluorescent channels, indicating wild-type and mutant alleles, respectively. To confirm the results of CRAS-PCR, an exon 4 fragment containing TPMT*2 for each sample was amplified and both forward and reverse sequencing were carried out (Table 1) as described in previous sections.
Results
The principles of CRAS-PCR
The CRAS-PCR system (Figure 1) is a novel version of AS-PCR in which unidirectional WT- and MT-AS primers (WT-ASF and MT-ASF in the present study) are used in a single tube to eliminate the non-specific amplification commonly occurring in traditional AS-PCR. By nature, the primer pairs in PCR reactions attempt to trigger amplification of input DNA under the pressure of the thermodynamic driving force of thermophilic DNA polymerase. In the AS-PCR system, a defined AS primer has only a single base at its 3′-end that distinguishes the genotype of the input DNA [10], [12], [13], [19], [20]. Accurate annealing between the AS primer and the corresponding complementary input DNA triggers a genotype-specific amplification reaction [19], [20]. However, if only a single genotypic AS primer (e.g., MT genotype) is included in the reaction, under the thermodynamic driving force of DNA polymerase, the single base terminal mismatch between primers and template could easily trigger the non-specific amplification of an input DNA having opposite genotype (e.g., WT genotype) [41]. Moreover, weak-destabilization effects of terminal mismatches could further promote non-specific amplification [42]. Although stringent reaction conditions can be used to dramatically reduce or eliminate non-specific amplification, optimization is time-consuming, and sometimes unsuccessful.
The WT and MT oligonucleotides refer to WT-ASF and MT-ASF primers, respectively. The WT-ASF, MT-ASF and CO-R primers depicted in the left panel were included in a single reaction. The middle panel shows the predicted reaction response for WT homozygotes, MT homozygotes, and WT/MT heterozygotes. Their corresponding amplification curves are indicated in the right panel, for which the red (CY5 fluorescent channel) and green (6-FAM fluorescent channel) lines indicate the amplification signals of WT and MT alleles, respectively.
For any given SNP variant, at least one WT- or MT-genotypic allele is present in the AS-PCR system. Therefore, if both primers are present, at least one genotypic AS primer will perfectly match the input DNA with the corresponding genotype and trigger specific amplification. Such amplification would perfectly fulfill the thermodynamic driving force of thermophilic DNA polymerase [41] and thus dramatically reduce or eliminate the weakly destabilizing effects of terminal mismatch [42]. Therefore, for CRAS-PCR, amplification triggered by perfectly matched AS primers always occurs. The availability of specific amplification can dramatically reduce or eliminate the non-specific amplification in traditional AS-PCR systems [10], [12], [19], [20].
Because the amplicon size of different genotypic input DNAs may be the same or indistinguishable by agarose gel electrophoresis, unidirectional competitive AS primers in CRAS-PCR systems are designed as scorpion primers [43], [44]. A unique fluorescent reporter and corresponding quencher is labeled at the 5′- and 3′-end of the stem-loop structure of scorpion primers (Figure 1). Therefore, a defined fluorescent signal can be used to directly identify the genotype of the input DNA (e.g., CY5 fluorescent channel for the WT genotype and 6-FAM fluorescent channel for the MT genotype in the present study; Table 1) [43], [44]. To further distinguish the genotype, additional mismatches can be introduced penultimate from the 3′-end for both WT-ASF and MT-ASF primers according to the Web-based Allele-Specific PCR Primer (WASP) principle (Table 1) [42]. Moreover, because distinguishable alleles are present, the fluorescent signal produced by one AS primer could serve as an internal positive control for another. For example, the negative fluorescent signal of the MT-ASF primer might be a true negative if a positive fluorescent signal is provided from the WT-ASF primer (e.g., for WT homozygotes), but might be a false negative if a negative fluorescent signal is provided from WT-ASF primers (e.g., if input DNA was accidently omitted).
Construction of WT-QC and MT-QC plasmids
To develop a system for optimization of CRAS-PCR, the PCR products of the TPMT exon 4 locus (Figure 2A) were cloned to prepare the WT-QC plasmid for TPMT*2. This plasmid was further used to prepare a corresponding MT-QC by site-directed mutagenesis. Both plasmids were sequenced to confirm the TPMT*2 genotype (Figure 2B; Figure S1–3). A 1∶1 mixture of above two plasmids was used as MIX-QC plasmids.
Panel A shows the PCR products of TPMT exon 4 containing the TPMT*2 locus at a 55 to 65°C gradient annealing temperature (lane 1 to 8). Panels B and C show the sequencing chromatogram images of WT- and MT-QC plasmid, in which the arrows indicate the location of TPMT*2, and the underlined bases show the codon containing TPMT*2.
Comparison of traditional AS-PCR and CRAS-PCR by the Taguchi method
According to the principles of the Taguchi method, a modified L16 (45) orthogonal array was designed with 16 experiments using four factors with four levels each (Table S1 and S2). Based on the AS primers, the conditions were optimized and characteristics of traditional AS-PCR (Assay Type 1 and 2) and CRAS-PCR (Assay Type 3; Table 2, Table S1 and S2) were evaluated. Moreover, each Assay Type was further subgrouped into three assays based on the genotype of the input DNA (WT-, MT-, or MIX-QC plasmid; Table 2). CY5 and 6-FAM fluorescence signaled the positive amplification of the TPMT*2 WT- or MT-genotype, respectively (Table 1 and 2; Figure 1).
Assay Types 1 and 2 (Table 2) represent traditional AS-PCR for which only WT- or MT-AS primers were used, and primer extension theoretically only occurs when the 3′-end is perfectly complementary to the allele present in the input sample [19], [20]. However, the results of the Taguchi method shows that non-specific amplification occurred almost uniformly for the MT-QC plasmid in Assay No. 1-2 and for WT-QC plasmid in Assay No. 2-1 (Figure 3A and 3B; Table S4 and S5). This further confirms that the thermodynamic driving force of thermophilic DNA polymerase can promote the non-specific amplification of traditional AS-PCR, despite the additional mismatch in the AS primers [10], [12], [13], [19], [20], [42].
Amplification curves of traditional AS-PCR (Panels A and B) and CRAS-PCR (Panels C and D) at 53 °C annealing temperature. Non-specific amplification by traditional AS-PCR was completely eliminated for CRAS-PCR. For Panel A (Assays No. 1-1 and 1-2) and Panel B (Assays No. 2-1 and 2-2), the open circles indicate the amplification curves of corresponding labeled QC plasmids. For Panels C and D (Assay No. 3-1 and 3-2), the corresponding QC plasmids are labeled on the upper right corners.
Compared with traditional AS-PCR, all experiments in the CRAS-PCR system (Assay Type 3) showed specific amplification regardless of the genotype of the input DNA (Figure 3C and D, Table S6 and S7). This result further confirms that the thermodynamic driving force of thermophilic DNA polymerase can be perfectly fulfilled in the CRAS-PCR system, thus eliminating the non-specific amplification commonly occurring in traditional AS-PCR.
Optimization and evaluation of traditional AS-PCR and CRAS-PCR by the Taguchi method
To determine the optimal conditions for traditional AS-PCR and CRAS-PCR, S/N ratios (η) were calculated (Table 3; Table S4 to S7). The mean η for each level of each factor was used to determine the optimal level (highest S/N ratio) and to calculate the PC values of each factor (Table 3). The PC values reflect the amount of variation of each factor. For traditional PCR (Assay Types 1 and 2), the annealing temperature had the highest PC values and significantly contributed to both the amplification efficiency and specificity, for which lower annealing temperatures yielded higher amplification efficiency (Assay No. 1-1 and 2-2), and higher annealing temperatures yielded higher amplification specificity (Assay No. 1-2 and 2-1). Moreover, the AS primer concentration had the highest PC values and affected the amplification efficiency and specificity: higher concentrations (Assay No. 1-1 and 2-2) yielded higher efficiency and lower concentrations (Assay No. 1-2 and 2-1) yielded higher specificities. These characteristics are consistent with the common thermodynamic features of traditional AS-PCR. Therefore, the results further confirm that the Taguchi method is useful for understanding the features of molecular analysis methods such as AS-PCR in the current study.
Compared with traditional AS-PCR, in CRAS-PCR the annealing temperature had the lowest PC values regardless of the genotype of input DNA, which indicates that the CRAS-PCR maintained specific amplification under relaxed annealing temperature (Table 3). Moreover, for a defined homozygous genotypic input DNA (Assay No. 3-1 and 3-2), the corresponding AS primer had the highest PC values and significantly contributed to the amplification efficiency. However, for heterozygous input DNA (Assay No. 3-3), only MT-ASF had significantly higher PC values.
Further detailed analysis of Assay Type 3 showed that the optimal CRAS-PCR conditions were dependent on the genotype of the input DNA (Table 3). According to the principle of the Taguchi method and parameters used to calculate S/N ratios, the optimal levels of each factor reflect the optimal components used for WT and MT homozygotes (Assay No. 3-1 and 3-2, respectively) and heterozygotes (Assay No. 3-3) (Table 2 and 3).
Evaluation of CRAS-PCR performance following optimization
Both the WT- and MT-alleles of TPMT*2 provide important information for clinical applications. Therefore, an optimal CRAS-PCR method should equivalently amplify each genotypic DNA, either in homozygous or heterozygous state. To further evaluate and confirm the characteristics of the optimized CRAS-PCR system, WT-, MT- and MIX-QC plasmids were used as input DNA for CRAS-PCR performed under the conditions optimized for Assays No. 3-1 to 3-3 (Table 3). The results further confirmed specific amplification for all three assays (Table 4, Table S8). Comparative analysis showed that the optimal level of Assays No. 3-1 and 3-2 had high amplification efficacy for the corresponding genotypic DNA, but not for the opposite genotypic DNA. For Assay No. 3-3, the amplification of heterozygous DNA (MIX-QC plasmid) had a low ΔCq value (ΔCq = 0.44; SD = 0.24), indicating equivalent amplification of each genotype. Triplicate experiments at optimal levels were performed, and S/N ratios and mean response were compared with predicted values. All assays had observed S/N ratios and mean response within the 95%confidence interval (Table 4), indicating that each model was accurate.
Based on above results, the reaction conditions of Assay No. 3-3 (0.2 μM each of WT- and MT-ASF primers, 0.4 μM of CO-R primers, and annealing temperature at 55 °C) were confirmed as the optimal parameters to analyze the TPMT*2 genotype. Ten-fold serially diluted WT-, MT- and MIX-QC plasmid (i.e., 1×108 to 1×101 copies of each type of plasmid) was used to further evaluate features of the above optimized CRAS-PCR system. Using a conversion factor of 6.6 pg per diploid human cell [39], [40], this is equal to 330 μg to 33 pg of human genomic DNA for WT- and MT-QC plasmid, and μg to 66 pg for MIX-QC plasmid in μl of reaction mixture, which replicates the human genomic DNA concentration used in real-time PCR. For homozygous genotypic DNA (i.e., WT- or MT-QC plasmid) at 10-fold serial dilution concentrations (1×108 to 1×101 copies), specific amplification consistently occurred (Figure 4A, B), which further confirms the specificity and selectivity of the optimized CRAS-PCR system. For the amplification efficiencies of homozygous (i.e., WT- or MT-QC plasmid) or heterozygous (i.e., MIX-QC plasmid) genotypic DNA, the WT- and MT-alleles had equivalent amplification efficiencies and Cq values regardless of the genotype of input DNA (Figure 4C, D). Furthermore, the CRAS-PCR system specifically determined the genotype of input homozygous DNA at a limitation of 10 copies, demonstrating a high level of sensitivity. Consistent with the data from Table 4, the average ΔCq value (0.42) for the serially diluted MIX-QC plasmid fell within the predicted 95% CI, further verifying the accuracy of the model.
Amplification curves of 10-fold serial dilutions (1×108 to 1×101 copies) of WT- (red lines) and MT-QC (green lines) are shown in Panels A and B, respectively. All assays showed specific amplification of targeted genotypic alleles. Calibration plots of homozygous (i.e., WT- or MT-QC plasmid) and heterozygous (i.e., MIX-QC plasmid) genotype alleles are shown in Panels C and D, respectively. The amplification efficiencies of homozygous WT- and MT-alleles (in Panel C) were 96.59% (slope: −3.406; R2 = 0.999) and 99.11% (slope: −3.343; R2 = 0.998), respectively. The amplification efficiencies of WT- and MT-alleles in heterozygous samples (in Panel D) were 95.07% (slope: −3.446; R2 = 0.998) and 98.27% (slope: −3.364; R2 = 0.999), respectively. The average ΔCq values of WT- and MT-alleles in heterozygous samples (Panel D) were 0.42 (SD: 0.19).
To further evaluate the ability to detect low-frequency or rare mutant alleles, various concentrations of MT-QC plasmid (1×106 to 1×101 copies) were mixed with serially diluted WT-QC plasmid (1×106 to 1×101 copies). The results show that the current CRAS-PCR system has ability to distinguish allelic differences at sensitivities of 1% regardless of the concentration of input heterozygous DNA (Figure 5). For example, when the input WT-QC plasmid is at 1×106 copies, the current CRAS-PCR system has the ability to detect 1×104 copies of MT-QC plasmid although the current CRAS-PCR system has the ability to detect as low as 10 copies of mutant allele without the interference of high-abundant WT-alleles (Figure 4). The ability to distinguish alleles (i.e., 1%) is similar to traditional AS-PCR) [45], [46]. This result indicates that the higher abundance of certain genotypic DNA would suppress or eliminate low-frequency or rare opposite genotypic DNA.
The concentration of WT-QC plasmid (red lines) was 1×106 copies in Panel A to C and 1×104 copies in Panel D to F. The mutation percentage of MT-QC plasmid (green lines) in each assay is indicated as labeled, in which 0% mutations indicates homozygous input DNA (i.e., WT-QC plasmid); 50% mutations indicates 1×106 copies in Panel B and 1×104 copies in Panel E; and 1% mutations indicates 1×104 copies in Panel C and 1×102 copies in Panel F. The results of other combinations of WT- and MT-QC plasmid showed equivalent results, i.e., 1% sensitivity of mutation analysis (data not shown).
TPMT*2 frequencies in Chinese analyzed by CRAS-PCR
To verify the specificity for clinical application of the optimized CRAS-PCR system, the optimal levels of the four factors in Assay Types No. 3-3 were further used to analyze the TPMT*2 genotype of 240 clinical blood samples from ethnic Chinese. This population is known to be almost exclusively WT [11], [47], and therefore, a predominant lack of positivity for the MT allele would rule out the possibility of non-specific amplification by the MT-ASF primer. Each sample was analyzed in duplicate, and WT-, MT- and MIX-QC plasmid were tested in parallel to monitor the amplification quality. The results showed that, as expected, all genotypes were WT, with no MT or heterozygous genotype samples (Figure 6). These results were further demonstrated by DNA sequencing of exon 4 fragments containing TPMT*2. Thus, this assay demonstrates the effective application of CRAS-PCR optimized by the Taguchi method for genotype determination of clinical DNA samples.
Examples of amplification curves of clinical samples (Panel A) and corresponding QC plasmids (Panel B) in which the CY5 (red lines) and 6-FAM (green lines) fluorescent channel indicate WT- and MT-genotype signals, respectively. In Panel B, the amplification curves for WT- (labeled with 1), MT- (labeled with 2) and MIX- (labeled with 3 and 4) QC plasmid are shown as a control.
Discussion
We have provided a comprehensive analysis of the differences between AS-PCR and CRAS-PCR and have also described efficient methodology for evaluation and optimization of each approach. In traditional AS-PCR, AS primers or AS probes are used to analyze the genotype of input DNA [10], [11], [12], [13], [14], [15], [19], [20]. At least one of the primer pairs targeting variant sequences is chosen from a polymorphic area with mutations located at or near its 3′-end. However, successful genotype determination is dependent on stringent conditions (annealing temperature, and primer and Mg2+ concentrations) [10], [11], [19], [20], [42], which could affect the sensitivity; additionally, the specificity is often hampered by cross-annealing, even when stringent reaction conditions are used [17], [18], [42]. Moreover, because the amplicon fragment size is often the same, two separate reactions are used to analyze the WT and MT genotype of input DNA [19], [20], [43], [44]. By nature, if primer pairs and corresponding templates are present, the PCR system tries to produce amplicons due to the thermodynamic driving force of thermophilic DNA polymerase [41], often leading to non-specific amplification in AS-PCR [10], [11], [19], [20], [42]. However, because CRAS-PCR uses both primers in a single reaction, for any defined input DNA a specific amplification reaction will occur, satisfying the thermodynamic driving force and reducing nonspecific amplification (Figure 1). The design parameters for CRAS-PCR outlined in this study demonstrate the advantage of the CRAS-PCR system for the avoidance of nonspecific amplification (Figure 3).
Regardless of the methodology for PCR, the amplification of input DNA by thermophilic DNA polymerase depends on reaction components (e.g., primer or Mg2+ concentrations) and thermal cycling parameters (e.g., annealing temperature and duration); therefore, reproducibility and accuracy is contingent on adequate optimization [41]. Optimization demands an investigation of the interaction between multiple variables and involves unusually large experiments [31], [41]. Traditionally, complete optimization can be achieved by testing each variable component independently by the factorial method, which is costly and time-consuming [31]. However, in the current study, the Taguchi method, a Design of Experiment (DOE) [21], [22], [23] was used to optimize conditions and evaluate characteristics of AS-PCR and CRAS-PCR. The Taguchi method was selected because it provides a systematic approach towards designing experiments [21], [22], [23], and consequently has been the most widely used technique in automotive and electronics industrial design for the past two decades [21], [22], [23], [28], [32], [33], [34], [35]. One of the important steps in Taguchi's technique is the selection of orthogonal arrays, which helps to determine the optimum level for each parameter and establishes the relative importance of individual parameters using a small set of all possibilities in a minimal number of experiments [21], [22], [23]. In the present study, according to the principle of the Taguchi method, a modified L16(45) orthogonal array was selected to optimize conditions and evaluate characteristics of both traditional AS-PCR and CRAS-PCR. Compared with factorial methods requiring 64 (43) experiments for traditional AS-PCR (Assay Types 1 and 2) and 256 (44) experiments for CRAS-PCR (Assay Type 3), only 16 experiments were need for each above assay, which dramatically increased the efficiency of optimization and evaluation.
The effectiveness of the Taguchi method was demonstrated using a variety of statistical methods. In all assays of the Taguchi method, the PC values of errors were less than 15% (Table 3), which indicated the acceptable reliability of orthogonal array selection, and significant factors were analyzed in each assay [21], [22], [23]. In assays of traditional AS-PCR (Assay Type 1 and 2), both “smaller-the-better” and “larger-the-better” strategies were selected in which the Cq values served as response variables. The annealing temperature had the highest PC values and its p values were less than 0.05. This result indicates that the annealing temperature had significant effect on both amplification efficiency and specificity for traditional AS-PCR. We aimed to determine the contribution of higher and lower concentrations of AS primers for efficiency (Assay No. 1-1 and 2-2 in Table 3) and specificity (Assay No. 1-2 and 2-1 in Table 3). The results indicated that both higher annealing temperature and lower AS primer concentrations contribute to the specificity of AS-PCR. Among the two factors, the annealing temperature was the most important parameter because it had significant higher PC values (Table 3).
For CRAS-PCR, the “smaller-the-better” strategy was used in which the Cqs or ΔCqs were used as response variables based on the genotype of the input DNA. Compared with the results of traditional AS-PCR, the annealing temperature almost always had lower PC values, and did not significantly affect the efficiency or specificity of the CRAS-PCR system. Moreover, for input DNA having homozygous genotype (Assay No. 3-1 and 3-1 in Table 3), the concentration of the corresponding AS primers had the highest PC values. However, using the ΔCq as a response variable for input DNA having heterozygous genotype (Assay No. 3-3 in Table 3), only MT-ASF primers had higher PC values, which indicates that the concentration of MT-ASF primers had more effect on the uniform amplification of both WT- and MT-genotypic input DNA (Table S6). Notably, compared with traditional PCR (Assay Type 1 and 2) in which non-specific amplification is present in almost all experiments, there was no non-specific amplification in any experiments for the CRAS-PCR system (S4 to S6). Therefore, the non-specific amplification of the traditional AS-PCR system can be eliminated or dramatically decreased if both genotypic primers are used in a single tube, such as in the CRAS-PCR system developed in the current study.
Among various single nucleotide mismatches, GA, CT, TT and CC mismatches almost always have strong destabilization strength, AA and GG mismatches have medium destabilization strength, and CA and GT mismatches have weak destabilization strength [42], [48]. In many cases, a single nucleotide mismatch at the 3′-end of the AS primer is not enough to specifically distinguish mutant alleles from wild-type ones. As a general rule, an additional mismatch can be introduced into the 3′-terminal position to enhance the destabilization strength of the AS primers to provide greater ability to distinguish alleles in AS-PCR systems. For example, if the 3′-terminal mismatch is a weak one (e.g., GT mismatch), a strong penultimate mismatch could be engineered, and vice versa [42], [48]. In our preliminary experiments, two strategies were used to design WT- and MT-ASF primers: one was fully complementary to the corresponding genotype alleles; another manually introduced a mismatched nucleotide at the penultimate position according to the WASP principle [42]. Compared with fully complementary AS primers, the currently designed AS primers (i.e., WT- and MT-ASF) consistently showed higher specificities in traditional AS-PCR systems (data not shown). The results of the Taguchi method indicate that the CRAS-PCR system using AS primers designed according to the WASP principle almost had annealing temperature-independent specificity and selectivity, which suggests that the design strategies of AS primers are crucial to such a feature of the CRAS-PCR system. However, the current study just evaluates the allele-distinguishing abilities of the CRAS-PCR system to specifically identify polymorphisms having strong (i.e., CC mismatch) to medium (i.e., GG mismatch) destabilization effects. Another thorough study should be further performed to analyze the ability to identify AS primers having only weak destabilization effects, such as CA and GT mismatches, in the CRAS-PCR system.
Based on the results of the Taguchi method and the final objective of TPMT*2 genotyping, optimal levels of four factors were used to evaluate the genotype of 240 clinical samples. The frequency of TPMT*2 genotype was essentially similar to previous publications, which indicated that TPMT*3C was found in 2.3% Chinese, but TPMT*2, *3A and *3B were not detected [11], [47]. Based on the frequency of TPMT*2, *3A, *3B and *3C in ethnic Chinese, a CRAS-PCR system targeting TPMT*3C is being developed in our laboratory according to the principles described in the current study.
In summary, a novel type of AS-PCR, CRAS-PCR, was developed in the present study. The Taguchi method was used to optimize conditions and evaluate characteristics of both traditional AS-PCR and CRAS-PCR. Compared with factorial methods, the Taguchi method provided a powerful statistic approach to determine optimal factor levels and understand characteristics. Compared with traditional AS-PCR, CRAS-PCR completely eliminated non-specific amplification at more relaxed conditions, which suggests that CRAS-PCR could serve as a powerful and robust routine genotyping tool for SNPs analysis, including TPMT*2, in a single tube.
Supporting Information
Figure S1.
Sequencing Chromatograph of WT-QC Plasmid.
https://doi.org/10.1371/journal.pone.0091824.s001
(PDF)
Figure S2.
Sequencing chromatograph of MT-QC plasmid.
https://doi.org/10.1371/journal.pone.0091824.s002
(PDF)
Figure S3.
Sequencing alignment between the reference genomic sequence and QC plasmids.
https://doi.org/10.1371/journal.pone.0091824.s003
(PDF)
Table S1.
Four factors and four levels for corresponding assays.
https://doi.org/10.1371/journal.pone.0091824.s004
(PDF)
Table S2.
The modified L16(45) orthogonal array used to optimize traditional AS-PCR (Assay Types 1 and 2).
https://doi.org/10.1371/journal.pone.0091824.s005
(PDF)
Table S3.
The modified L16(45) orthogonal array used to optimize CRAS-PCR (Assay Type 3).
https://doi.org/10.1371/journal.pone.0091824.s006
(PDF)
Table S4.
Quantification cycles of duplicate runs (Cq1 and Cq2) for all experiments and their corresponding S/N ratio (η) in Assay Type 1.
https://doi.org/10.1371/journal.pone.0091824.s007
(PDF)
Table S5.
Quantification cycles of duplicate runs (Cq1 and Cq2) for all experiments and their corresponding S/N ratio (η) in Assay Type 2.
https://doi.org/10.1371/journal.pone.0091824.s008
(PDF)
Table S6.
Quantification cycles of duplicate runs (Cq1 and Cq2) for all experiments and their corresponding S/N ratio (η) in Assay No. 3-1 and 3-2.
https://doi.org/10.1371/journal.pone.0091824.s009
(PDF)
Table S7.
Quantification cycles of quadruplet runs (Cq1 to Cq4) for all experiments and their corresponding S/N ratio (η) of each sample ΔCq in Assay No. 3-3.
https://doi.org/10.1371/journal.pone.0091824.s010
(PDF)
Table S8.
Quantification cycles of triplicate runs (Cq1 to Cq3) for confirmatory experiments and their corresponding S/N ratio (η) in Assay Type 3 (CRAS-PCR).
https://doi.org/10.1371/journal.pone.0091824.s011
(PDF)
Author Contributions
Conceived and designed the experiments: QH WLF. Performed the experiments: DC ZY HX YZ. Analyzed the data: QH DC ZY JFH. Contributed reagents/materials/analysis tools: QH GYW ZRC TNJ HX JFH. Wrote the paper: QH.
References
- 1. Elion GB (1989) The purine path to chemotherapy. Science 244: 41–47.
- 2. Ford LT, Berg JD (2010) Thiopurine S-methyltransferase (TPMT) assessment prior to starting thiopurine drug treatment; a pharmacogenomic test whose time has come. J Clin Pathol 63: 288–295.
- 3. Appell ML, Berg J, Duley J, Evans WE, Kennedy MA, et al. (2013) Nomenclature for alleles of the thiopurine methyltransferase gene. Pharmacogenet Genomics 23: 242–248.
- 4. Duley JA, Somogyi AA, Martin JH (2012) The future of thiopurine pharmacogenomics. Pharmacogenomics 13: 1549–1552.
- 5. Nguyen CM, Mendes MA, Ma JD (2011) Thiopurine methyltransferase (TPMT) genotyping to predict myelosuppression risk. PLoS Curr 3: N1236.
- 6. Relling MV, Gardner EE, Sandborn WJ, Schmiegelow K, Pui CH, et al. (2013) Clinical pharmacogenetics implementation consortium guidelines for thiopurine methyltransferase genotype and thiopurine dosing: 2013 update. Clin Pharmacol Ther 93: 324–325.
- 7. Relling MV, Gardner EE, Sandborn WJ, Schmiegelow K, Pui CH, et al. (2011) Clinical Pharmacogenetics Implementation Consortium guidelines for thiopurine methyltransferase genotype and thiopurine dosing. Clin Pharmacol Ther 89: 387–391.
- 8. Swen JJ, Nijenhuis M, de Boer A, Grandia L, Maitland-van DZA, et al. (2011) Pharmacogenetics: from bench to byte—an update of guidelines. Clin Pharmacol Ther 89: 662–673.
- 9. Carter MJ, Lobo AJ, Travis SP (2004) Guidelines for the management of inflammatory bowel disease in adults. Gut 53 Suppl 5V1–V16.
- 10. Yates CR, Krynetski EY, Loennechen T, Fessing MY, Tai HL, et al. (1997) Molecular diagnosis of thiopurine S-methyltransferase deficiency: genetic basis for azathioprine and mercaptopurine intolerance. Ann Intern Med 126: 608–614.
- 11. Collie-Duguid ES, Pritchard SC, Powrie RH, Sludden J, Collier DA, et al. (1999) The frequency and distribution of thiopurine methyltransferase alleles in Caucasian and Asian populations. Pharmacogenetics 9: 37–42.
- 12. Niens M, Spijker GT, Diepstra A, Te MG (2005) A factorial experiment for optimizing the PCR conditions in routine genotyping. Biotechnol Appl Biochem 42: 157–162.
- 13. Ameyaw MM, Collie-Duguid ES, Powrie RH, Ofori-Adjei D, McLeod HL (1999) Thiopurine methyltransferase alleles in British and Ghanaian populations. Hum Mol Genet 8: 367–370.
- 14. Ford L, Kampanis P, Berg J (2009) Thiopurine S-methyltransferase genotype-phenotype concordance: used as a quality assurance tool to help control the phenotype assay. Ann Clin Biochem 46: 152–154.
- 15. Schutz E, von Ahsen N, Oellerich M (2000) Genotyping of eight thiopurine methyltransferase mutations: three-color multiplexing, “two-color/shared” anchor, and fluorescence-quenching hybridization probe assays based on thermodynamic nearest-neighbor probe design. Clin Chem 46: 1728–1737.
- 16. Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, et al. (2009) The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clin Chem 55: 611–622.
- 17. Oliver DH, Thompson RE, Griffin CA, Eshleman JR (2000) Use of single nucleotide polymorphisms (SNP) and real-time polymerase chain reaction for bone marrow engraftment analysis. J Mol Diagn 2: 202–208.
- 18. Shi C, Eshleman SH, Jones D, Fukushima N, Hua L, et al. (2004) LigAmp for sensitive detection of single-nucleotide differences. Nat Methods 1: 141–147.
- 19. Gibson NJ (2006) The use of real-time PCR methods in DNA sequence variation analysis. Clin Chim Acta 363: 32–47.
- 20. Lyon E, Wittwer CT (2009) LightCycler technology in molecular diagnostics. J Mol Diagn 11: 93–101.
- 21.
Ross PJ (1988) Taguchi techniques for quality engineering: loss function, orthogonal experiments parameter and tolerance design. New York: McGraw-Hill.
- 22.
Taguchi G (1986) Introduction to quality engineering: designing quality into products and processes (trans: Organization AP). The Organization AP, Tokyo: Productivity Press Inc.
- 23.
Morgan E (2005) Chemometrics: Experimental Design. New York: John Wiley and Sons.
- 24. Cobb BD, Clarkson JM (1994) A simple procedure for optimising the polymerase chain reaction (PCR) using modified Taguchi methods. Nucleic Acids Res 22: 3801–3805.
- 25.
Caetano-Anolles G (1998) DAF optimization using Taguchi methods and the effect of thermal cycling parameters on DNA amplification. Biotechniques25: : 472–476, 478–480.
- 26. Dobay O, Juhasz E, Ungvari A, Jeney C, Amyes SG, et al. (2009) Taguchi optimisation of a multiplex pneumococcal serotyping PCR and description of 11 novel serotyping primers. Acta Microbiol Immunol Hung 56: 327–338.
- 27. Ramakrishna US, Kingston JJ, Harishchandra SM, Batra HV (2013) Taguchi optimization of duplex PCR for simultaneous identification of Staphylococcus aureus and Clostridium perfringens alpha toxins. FEMS Microbiol Lett 340: 93–100.
- 28. Ballantyne K, van Oorschot R, Mitchell R (2008) Reduce optimisation time and effort: Taguchi experimental design methods. Forensic Sci Int Genet Suppl Ser 1: 7–8.
- 29. Thanakiatkrai P, Welch L (2012) Using the Taguchi method for rapid quantitative PCR optimization with SYBR Green I. Int J Legal Med 126: 161–165.
- 30. Celani DSH, Moyses CB, Pontes FJ, Duarte RN, Sanches DSC, et al. (2011) Molecular assay optimized by Taguchi experimental design method for venous thromboembolism investigation. Mol Cell Probes 25: 231–237.
- 31. Rao RS, Kumar CG, Prakasham RS, Hobbs PJ (2008) The Taguchi methodology as a statistical tool for biotechnological applications: a critical appraisal. Biotechnol J 3: 510–523.
- 32. Kaladhar M, Subbaiah KV, Rao CS, Rao KN (2011) Application of Taguchi approach and Utility Concept in solving the Multi-objective Problem when turning AISI 202 Austenitic Stainless Steel. Journal of Engineering Science and Technology Review 1: 55–61.
- 33.
Kamaruddin S, Khan ZA, Wan KS (2004) The use of the Taguchi method in determining the optimum plastic injection moulding parameters for the production of a consumer product. Journal Mekanikal: 98–110.
- 34. Rama RS, Padmanabhan G (2012) Application of Taguchi methods and ANOVA in optimization of process parameters for metal removal rate in electrochemical machining of Al/5%SiC composites. IJERA 3: 192–197.
- 35. Bayrak OU, Hattagoglu F, Hinislioglu S (2010) Determination of Modulus of Rupture of Pavement Concrete with Silica Fume and Fly Ash using Taguchi Technique. International Journal of Civil and Structural Enginnering 3: 518–533.
- 36. Huang Q, Baum L, Fu WL (2010) Simple and practical staining of DNA with GelRed in agarose gel electrophoresis. Clin Lab 56: 149–152.
- 37. Huggett JF, Foy CA, Benes V, Emslie K, Garson JA, et al. (2013) The digital MIQE guidelines: Minimum Information for Publication of Quantitative Digital PCR Experiments. Clin Chem 59: 892–902.
- 38. Higuchi R, Fockler C, Dollinger G, Watson R (1993) Kinetic PCR analysis: real-time monitoring of DNA amplification reactions. Biotechnology (N Y) 11: 1026–1030.
- 39. Saiki RK, Gelfand DH, Stoffel S, Scharf SJ, Higuchi R, et al. (1988) Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239: 487–491.
- 40. Lo YM, Zhang J, Leung TN, Lau TK, Chang AM, et al. (1999) Rapid clearance of fetal DNA from maternal plasma. Am J Hum Genet 64: 218–224.
- 41.
Yuryev A (2007) PCR Primer Design. New Jersey: Humana Press.
- 42. Wangkumhang P, Chaichoompu K, Ngamphiw C, Ruangrit U, Chanprasert J, et al. (2007) WASP: a Web-based Allele-Specific PCR assay designing tool for detecting SNPs and mutations. BMC Genomics 8: 275.
- 43. Mackay J, Landt O (2007) Real-time PCR fluorescent chemistries. Methods Mol Biol 353: 237–261.
- 44. Kubista M, Andrade JM, Bengtsson M, Forootan A, Jonak J, et al. (2006) The real-time polymerase chain reaction. Mol Aspects Med 27: 95–125.
- 45. Jarry A, Masson D, Cassagnau E, Parois S, Laboisse C, et al. (2004) Real-time allele-specific amplification for sensitive detection of the BRAF mutation V600E. Mol Cell Probes 18: 349–352.
- 46. Anderson SM (2011) Laboratory methods for KRAS mutation analysis. Expert Rev Mol Diagn 11: 635–642.
- 47. Kubota T, Chiba K (2001) Frequencies of thiopurine S-methyltransferase mutant alleles (TPMT*2, *3A, *3B and *3C) in 151 healthy Japanese subjects and the inheritance of TPMT*3C in the family of a propositus. Br J Clin Pharmacol 51: 475–477.
- 48. Bui M, Liu Z (2009) Simple allele-discriminating PCR for cost-effective and rapid genotyping and mapping. Plant Methods 5: 1.