The biphasic assay mechanism
TSP employs a biphasic PCR mechanism. Assays are performed using two sets of primers: a pair of primers to enrich the target sequence (locus-specific primers) and a second nested primer pair to amplify only the enriched target DNA containing the interrogated allele(s) (nested locus-specific primers). A large difference in annealing temperature between the two sets of primers separates their participation into the different stages of the reaction, thereby allowing both sets of primers to be present in the same reaction vessel. Depending on the platform used to detect the genotyping products, the benefits of this four-primer, biphasic PCR mechanism range from increased sensitivity and specificity for allele-specific primer extension chemistry to size-controlled PCR products for high resolution melt detection. These benefits result from initial enrichment of the target locus, prior to interrogation of the SNP by the nested locus-specific primers.
The use of a temperature switch during PCR enables a sequence of two amplifications to take place in a single-step, closed-tube reaction: amplification of the target sequence at a high annealing temperature, followed by detection of the harbored SNP at a low annealing temperature. This is achieved by manipulating the design of the locus-specific (LS) and nested locus-specific (NLS) primers. The LS primers are designed with a high melting temperature and to produce a PCR product greater than 400 bp that flanks the SNP allele. The NLS primers are designed with two distinct regions: a core region and a tail region and to amplify a product of less than 300 bp. The core region is complementary to sequence flanking the SNP and has a low melting temperature. The 5'-tail region is non-complementary to the DNA template and increases the melting temperature of the primer once the tail is incorporated into the PCR product. During the first phase of PCR amplification, the use of a high annealing temperature ensures that only the LS primers participate in the reaction, thereby enriching the target sequence harboring the SNP. In the second PCR phase, the annealing temperature is decreased to allow the core region of the NLS primers to hybridize to the enriched target sequence. After several PCR cycles of low-temperature annealing to allow the NLS primers to become incorporated into the DNA template, the annealing temperature is again increased. The NLS primers are present in the reaction at a sufficiently high concentration so that, in combination with their now increased melting temperature, they can compete effectively with the LS primers to result in the accumulation of the NLS primer product(s). These TSP genotyping products can be detected by a variety of single-marker methods.
TSP genotyping by size polymorphism
The detection of TSP genotyping products by size polymorphism is achieved using allele-specific primer extension chemistry. In the TSP assay, allele-specificity is conferred by the terminal 3' nucleotide of the NLS forward primer and the SNP genotype is determined from the size of the TSP genotyping products. The target allele is detected by the presence of the smallest PCR product, which is amplified by the NLS forward and reverse primer pair (the allele-specific PCR product). The absence of the target allele is observed by the presence of the largest PCR fragment, which is the product of the LS forward primer and NLS reverse primer (the alternate allele PCR product). The alternate allele PCR product also acts as a positive control against a failed PCR assay. In heterozygote samples, both PCR products are amplified, thereby providing co-dominant allele detection (Figure 1).
TSP genotyping by high resolution melt analysis
The allelic discrimination of TSP genotyping products by high resolution melt (HRM) analysis is achieved by detecting the difference in melting temperature between the PCR fragments amplified for the different alleles. The TSP assay is performed using a pair of LS primers to enrich the target sequence from the genomic template, and a pair of NLS primers to capture the harbored SNP. The NLS primers are designed without a 3' nucleotide complementary to one of the SNP alleles, and hence the assay produces a single PCR fragment of constant size regardless of the captured SNP allele(s), which is the result of the NLS primer pair. The NLS primers are designed to position the SNP towards the centre of the TSP genotyping product and to produce a PCR fragment with an optimal length for HRM analysis, which is typically between 100 and 300 bp. The advantage of this assay configuration is that the size of the TSP genotyping product and the position of the SNP within the PCR fragments can be readily adjusted to maximize allele discrimination sensitivity (Figure 2).
The strategy for developing the biphasic PCR mechanism
During the development of TSP, the most critical steps to achieve were: (1) complete separation of the amplification step for enrichment of the target sequence from the step for interrogating the harbored SNP, and (2) efficient transition in the second phase of the reaction from continued enrichment of the target locus to the amplification of the SNP genotyping products. In developing the biphasic PCR mechanism for TSP, experiments were focused on allele-specific PCR assays designed for endpoint SNP genotyping by the detection of size polymorphism on agarose gel. Robust allele-specific PCR assays are challenging to develop due to the difficulty associated with achieving absolute allele-specificity [4, 5, 14, 15]. Hence, allele-specific PCR assays provided a useful model for developing the biphasic PCR mechanism, since it was expected that these assays would provide the greatest challenge to achieving an efficient transition in the second phase of the reaction from enrichment of the target locus to amplification of the expected SNP genotyping products. In addition, allele-specific PCR allowed the accumulation of PCR product to be visualized on agarose gel at each stage of the reaction, which assisted the development of the optimal parameters for TSP primer design and cycling conditions.
Defining parameters for locus-specific primer design
The first parameters investigated for the development of TSP were the melting temperature and cycle number for the LS primers. A melting temperature range of 60-65°C was chosen for designing the LS primers. This enabled the first phase of the PCR amplification to be performed at a high annealing temperature, allowing hybridization of only the LS primers, while the second phase of the PCR could be performed at a lower annealing temperature for stringent annealing of the NLS primers. Experiments were performed using 12 primer pairs that generated PCR products ranging from 500 to 1300 bp in length and a subset of 16 DNA samples. The primer concentration was maintained at 0.1 μM to ensure the locus specific primers became limiting in the second phase of amplification and to minimize primer-dimer formation.
Performing a standard thermal cycling reaction at 58°C annealing with 35 cycles of amplification produced detectable levels of PCR product of the expected size for all of the LS primer sets tested, confirming the specificity of the amplifications. To determine the number of PCR cycles required for enrichment of a target locus, the reactions were terminated at 5-cycle intervals and the products were separated on an agarose gel. Fifteen cycles was selected for enriching the target locus in the first phase of amplification, as this was the cycle threshold before which detectable levels of locus-specific product were produced when between 20 and 50 ng of genomic DNA was used as starting material. These assay parameters ensured that the locus-specific product was never amplified to a detectable level, thereby simplifying the assay output for endpoint SNP genotyping by minimising the total number of bands that would be detected in the final TSP genotyping assay.
Defining parameters for nested locus-specific primer design
Several parameters were essential to consider when developing the NLS primers to ensure that they did not engage in the first phase of amplification, were able to compete with the LS primers once they had engaged, and did not amplify non-specifically or result in mega-priming [16].
To ensure that the NLS primers would not engage in the first phase of PCR amplification, the optimal core melting temperature of the NLS primers was investigated using 12 SNPs on a subset of 16 DNA samples. This was done empirically by designing primer cores with melting temperatures ranging from 35 to 55°C, increasing in 5°C increments. Standard thermal cycling reactions were performed with an annealing temperature of 58°C with 35 cycles of amplification using only the NLS forward and reverse primer pair to determine which primer sets did not produce an NLS product directly from genomic template. Separation of the amplification products on agarose gel revealed that NLS primers with a melting temperature of 50°C or less produced no visible product, while those with a melting temperature of 55°C produced an amplification product, irrespective of the genotype of the genomic template, implying non-specific amplification.
Pairs of NLS primer cores with melting temperatures of 50°C or less were tested under TSP cycling conditions (see Materials and Methods) in the presence of the LS forward and reverse primers using DNA samples with known genotypes. In these reactions, it was observed on agarose gel that the NLS primers were unable to effectively compete with the LS primers (Figure 3). For example, when an NLS primer pair with a core melting temperature of 45°C was tested in the TSP configuration for allele-specific PCR, the presence-absence of the 147 bp allele-specific PCR product corresponded to the correct genotype. However, each reaction also produced the 271 bp alternate allele PCR product and so incorrectly indicated a heterozygous state (Figure 3B). This suggested the NLS primers could not sufficiently outcompete amplification with the LS primer pair in the presence of the SNP allele to allow an efficient transition from enrichment to interrogation of the target allele.
To improve the ability of the NLS primers to compete with the LS primers, the addition of 5' non-complementary tails to NLS primers with a core melting temperature between 35 and 50°C was investigated. These tails increased the overall melting temperature of the NLS primers from 5 to 20°C above their core melting temperature. A tail that increased the overall melting temperature to 55°C or above resulted in the NLS primers efficiently engaging in the first phase of amplification, which resulted in non-specific allele amplification regardless of the genotype of the genomic template. Similarly, a tail that did not increase the overall melting temperature above 50°C did not allow the NLS primers to compete with the LS primers once the annealing temperature was increased in the second phase of amplification (Figure 4). A tail that increased the overall melting temperature to 53°C when added to NLS primers with a core melting temperature of 45°C was observed to not bind in the first phase of amplification and was able to out-compete the LS primers in the homozygous wildtype state. However, these primer parameters were still not sufficient to provide equal amplification for heterozygous alleles.
To improve the efficiency of the NLS primer amplification in the heterozygous state, the primer concentration was initially increased ten-fold over the LS primer pair. This greatly improved the amplification efficiency of the NLS primers. However, in allele-specific TSP assays it also resulted in the accumulation of only allele-specific PCR product, even in heterozygous samples. Decreasing the primer ratio to 5:1 balanced the accumulation of the allele-specific and alternate allele PCR products in allele-specific PCR assays during the second phase of amplification.
Overall, the final parameters selected for TSP assay design were found to be a pair of LS primers with a melting temperature of 63°C (with a range of 60-65°C) and a pair of NLS primers with a core melting temperature of 45°C (range 43-47°C) with a short 5' non-complementary tail that increased the overall melting temperature to 53°C (range 52-55°C) once incorporated into PCR product. The optimal ratio of NLS to LS primer was 5:1.
Validation of a distinct two-stage amplification mechanism
To validate that TSP amplification was partitioned into two distinct stages, the assay was performed with a variety of primer combinations configured for allele-specific PCR or HRM detection. The accumulation of TSP genotyping products was monitored in real-time using SYBR Green and confirmed by endpoint detection on agarose gel.
In real-time PCR, assays performed under standard TSP cycling conditions with only the LS forward and reverse primers showed the rapid accumulation of product by 35 cycles (Figure 5a). This was confirmed by agarose gel electrophoresis to correspond to the amplification of the full-length target region. Without NLS primers in the reaction, there was no switch to amplification of the final TSP genotyping products (Figure 5c). In contrast, assays performed with only the NLS forward and reverse primers resulted in the delayed accumulation of PCR product until after the annealing temperature was lowered following the first 15 cycles (Figure 5b). In the final phase of TSP, an additional 10 cycles of amplification was needed to produce sufficient PCR product during the real-time analysis to detect the binding of the NLS primers to the genomic template. Endpoint detection of these products on agarose gel revealed the non-specific amplification of the allele-specific TSP genotyping product in the absence of the target allele in assays configured for allele-specific PCR. This resulted from the indiscriminate hybridization of the NLS primers directly to the genomic template in the second phase of amplification due to the absence of the LS primers to first enrich the target region. These results demonstrate an efficient partitioning of the participation of the LS and NLS primers into the first and second stages of PCR, respectively. TSP cycling performed using both LS and NLS primers showed an amplification profile similar to reactions containing only LS primer, but consistently produced less fluorescence at each cycle (Figure 5c). This reduced fluorescence corresponds to the transition from the amplification of LS product to the NLS product, and is observed because the NLS product is significantly shorter than the LS product. SYBR Green dye binds only to double-stranded DNA, producing an increase in fluorescence that is influenced by the length(s) of the PCR product(s). Endpoint agarose gel detection showed the expected TSP genotyping products in assays configured for both allele-specific PCR (Figure 1b) and HRM detection (Figure 2b). The real-time analysis therefore also demonstrates an efficient transition from the amplification of LS product to the accumulation of NLS product in the second phase of the reaction.
To further confirm the importance of the biphasic PCR mechanism for the amplification of the expected TSP genotyping products, the same set of reactions was repeated, however this time with constant cycling conditions (i.e. 35 cycles at 58°C annealing), which replicate the conditions of a published allele-specific PCR genotyping method [15]. These conditions did not produce the expected genotyping products that were seen with TSP thermal cycling (Figure 6), implying the importance of the distinct temperature switch in the TSP assay for controlling the sequential amplification of the genotyping products.
Evaluation of TSP for endpoint SNP genotyping
To evaluate the use of TSP for endpoint SNP genotyping, assays were developed for the detection of SNP genotyping products by size polymorphism, and for detection by HRM analysis. The genotyping specificity of the assays was initially tested on a subset of DNA samples with known zygosity.
All TSP assays developed for HRM detection amplified only the size-defined region harboring the SNP deliminated by the NLS primers and produced the expected genotypes for the DNA samples tested (Figure 2), but seven of 87 TSP assays developed for allele-specific PCR resulted in the "leaky" amplification of alleles. "Leaky" amplification was defined by the presence of the NLS primer product (allele-specific PCR product) in DNA samples known to be homozygous for the alternate allele. To overcome the problem of non-specific and leaky amplification, the addition of a secondary mismatch in the NLS forward primer was investigated. The addition of a deliberate secondary mismatch near the 3' terminus of an NLS primer has been demonstrated to improve allele-specificity by destabilising the binding of the primer and improving primer extension stringency [5]. The nature and position of the secondary mismatch was calculated using WebSNAPER and TM MISMATCH functions [17, 18]. Assays performed with a secondary mismatch at the n-2 or n-3 position were unable to outcompete amplification of the LS product, resulting in the amplification of both allele-specific and alternate allele PCR products for homozygous samples. Rather, redesigning the TSP markers to interrogate the alternate SNP allele proved to be the most effective method for stabilizing allele-specificity. Whilst some of the redesigned assays still produced a small amount of non-specific allele amplification, the signal-to-noise ratio was always sufficiently high to ensure unambiguous genotype assignment.
The genotyping accuracy for the 87 allele-specific PCR markers was determined in a blinded study using subsets of 376 different DNA samples, which generated a total of 11,232 data points. The range of DNA samples tested included doubled haploid, F1 and F3 plants. Doubled haploid plants were expected to be homozygous at all SNP loci, while the F1 and F3 plants were expected to contain both homozygous and heterozygous SNP loci. About 25% of the 11,232 TSP genotypes inferred heterozygosity at the SNP locus. Comparing the SNP datasets generated using the TSP markers with those produced using an independent genotyping method revealed 100% concordance in the genotype assignments. The independent genotyping data was generated using allele-specific primer extension assays performed on the BioPlex™ microsphere suspension array platform (BioRad), as described by [19] using the same DNA samples, and for the same SNP loci.
The robustness of the TSP assay was further demonstrated by amplifying DNA samples extracted using a variety of common methods (see Materials and Methods). Although the quality of the DNA templates was variable and often crude, the TSP assays produced the correct genotypes from the range of DNA samples, demonstrating that DNA quality was not a limiting factor. The concentration of the DNA template was the greatest variable to the success of marker amplification and accurate genotyping. The TSP assay was optimal with a starting DNA input of 100 ng or less.