A real-time-PCR read-out is given as the number of PCR cycles ("cycle threshold" Ct) necessary to achieve a given level of fluorescence. For this study, the Ct was fixed in the exponential phase of the PCR (Figure 2A, linear part of the fluorescence curve). During the initial PCR cycles, the fluorescence signal emitted by SYBR-Green I bound to PCR product was usually too weak to register above background. Indeed, a difference could not be defined until after about 15 PCR cycles. During the exponential phase of the PCR the fluorescence doubled at each cycle. After 35 cycles, the intensity of fluorescent signal usually began to plateau, indicating that the PCR had reached a saturation status. As a Ct is proportional to the logarithm of initial amount of target in a sample, the relative concentration of one target with respect to another is reflected in the difference in cycle number (ΔCt) necessary to achieve the same level of fluorescence (Figure 2A).

Optimal design of the PCR primers is essential for accurate and specific quantification using real-time PCR. For the assay described herein, detection is based on the binding of the SYBR-Green I dye into double stranded PCR products, which is a sequence independent process. While this assay is cheaper than the more commonly used 5' nuclease assay, it loses the additional level of specificity introduced by the hybridisation of the PCR product to a specific fluorescent TaqMan probe. The sensitivity of detection with SYBR-Green I may therefore be compromised by the formation of primer-dimers, lack of specificity of the primers, primer concentration (which can be limiting) and the formation of secondary structures in the PCR product. All of these factors could lead to the creation of unexpected double-stranded DNA product, which would incorporate SYBR-Green I and register a fluorescent signal. The 5'-nuclease assay using TaqMan probes would also be compromised by primer dimers. Although they are not detected, they do alter the amplification efficiency of the specific product.

The optimisation of the primer concentration is essential. Each set of primers work best under different concentration conditions. Primer design is also very restrictive: the annealing temperature is restricted to 58–60°C, which corresponds to the optimal working conditions for the AmpliTaq Gold DNA polymerase enzyme (Perkin Elmer), and the length of the PCR product is set between 80 and 150 bp. The use of software dedicated to RT-PCR primer design, such as Primer Express™ (Perkin Elmer), is highly recommended. For all experiments described in this report, each pair of primers was designed following the manufacturer's instructions to amplify products from the genes of interest ranging from 81 to 122 bp (Primer Express™, Perkin Elmer). Primer concentration is usually determined to be optimal when specific amplification relative to primer-dimers was maximal in a positive versus negative control experiment. However, the choice of the most appropriate conditions for our assay corresponds to the combination of primer concentration allowing a ratio of target gene to reference gene of 1 in a DNA sample with no genetic abnormality. This ratio is calculated using the ΔCt methods (ΔCt = Ct of the target – Ct of the reference). A delta Ct of 0 indicates a ratio of 1 between the target and the reference (2^-(ΔCt) = 2⁰ = 1). A number of factors can influence this assay, and the optimisation consists in compensating these factors to obtain a ratio of 1 in a DNA sample with no genetic abnormality. The inclusion of such a normal DNA sample on each experiment also provides a reference sample for a Δ-ΔCt method of calculation, however, as the normalisation will relate to a value close to 0 for the second ΔCt, both methods essentially give the same result.

Complete absence of primer dimer was rarely achieved in a PCR negative control (Figure 2B and 2C), however, in an experimental negative including a template but no target, dimers were not observed. Furthermore, the inclusion of a heat dissociation experiment at the end of the PCR suggests that a single product is present (even if an electrophoresis is used once at the end of the first optimisation plate to confirm it). The degree of an eventual primer-dimer contribution (Figure 2C) to the overall fluorescent signal of the PCR negative control can also be detected in such a dissociation experiment. As an example, 3 sets of primers were used to amplify 3 targets (OPA1, GLI and MYC-C) in an optimisation assay where the only variable was the concentration of primers in the reaction (Figure 2D). For these examples, the concentration of primer for GAPDH was 100 nM for GLI and MYC-C and 50 nM for OPA1. However, for other target genes it was necessary to either increase or decrease the primer concentration for GAPDH (see Table 1). For this demonstration, we used normal PBMC derived DNA template and normalised the expression of these 3 target genes to GAPDH. The optimisation experiment consists in finding the conditions necessary to obtain a ratio of for target gene to reference. The first set of primers (black bars, MYC-C) showed a bell shaped optimisation pattern where the intermediate concentrations were optimum and giving ratio of target to reference close to 1. The second set of primers (grey bars, OPA1) showed a plateau type of curve where most of the higher concentrations were of similar efficiency. The last primer set (open bars, GLI) showed poor amplification at both extremes of the concentration range, and a single, efficient concentration of primer. In these optimisation experiments, primer dimers were not observed. Higher primer concentration had a limiting effect on the PCR reaction and would be detrimental to the accuracy of the quantification.

For an accurate quantification it is also necessary to verify that PCR efficiency is independent of the initial amount of target DNA. Each pair of primers was therefore tested across several logs dilution series of a positive control DNA sample. Figure 3A shows such an experiment for TREC quantification (coding and signal). It was necessary to use cloned template for this purpose, as a TREC rearrangement sequence is not available in genomic DNA. The two plots obtained for the signal and coding TREC (and the internal reference gene, GAPDH), were linear, demonstrating equal efficiency of the individual PCR assays over a large range of initial template concentrations. The titration plots were also super-imposed, demonstrating that the three PCR assays were equally efficient despite the differences in PCR product sizes and base pair constitution. Similar experiments were performed for all the other genes, using genomic DNA however (data not shown, and Table 1).

One of our major aims was to determine the limit of sensitivity of our assay in allowing accurate quantification of gene amplification, deletion or rearrangement from DNA extracted from relatively few cells. For this purpose we estimated the number of cells present in a given sample by real-time PCR quantification of the copy number of the housekeeping gene GAPDH, each cell having two allelic copies of this gene. In principle, any gene would be suitable, however, as we were expecting to look at deletions, amplifications and rearrangements, we chose a gene that was unlikely to be involved in any of these processes. We have assayed the linearity of the relationship between real-time Ct for GAPDH reading with cloned GAPDH target. The use of cloned template also provided a read-out that reflected the number of molecules present in the reaction, which could then be used for an absolute quantification (see Figure 4A). Therefore, Figure 3A also shows a standard curve correlating the GAPDH-Ct reading to the number of cloned GAPDH molecules, allowing us to correlate the GAPDH-Ct of a sample with a number of GAPDH molecules. This number, when divided by 2, would thereby correspond to the number of cells from which the DNA was initially extracted. We assayed the relationship between the amount of DNA in cell extracts and GAPDH readings. The relationship between GAPDH-Cts and the spectrophotometer-measured concentrations of forty genomic DNA samples extracted from PBMCs under the same conditions but at different times and by different research workers is linear as demonstrated in Figure 3B. The relationship between the initial number of cells in a sample and the DNA concentration of the same sample was also linear (data not shown), indicating a similar efficiency of the DNA extraction method without regard for the initial number of cells in the samples. Together these results validate the use of GAPDH-Cts as an internal measure of the amount of DNA in each sample, reflecting the number of cells present in that sample.

The original PCR conditions described in the reports by Doueck and colleagues 14 were reproduced to clone two PCR products (380 bp each) for both signal and coding TRECs. Real-time PCR primers were designed flanking the site of the recombination, to amplify PCR products of 99 and 122 bp, specific for signal and coding TRECs respectively (see Figure 1). Optimisation and standardisation were performed (see Figure 2 and 3). Two quantification methods were used to compare a set of DNA samples extracted from CD4+T-cells in ten healthy individuals. The first method determined the number of TREC molecules per μg of DNA, relying on DNA concentration obtained by optical density readings. Our results were in line with published values 1415 for a similar age group (Figure 4A). The second method estimated the number of cells in the reaction by quantifying GAPDH copy number as described above, giving the number of cells containing a TREC as a percentage of total cells (Figure 4A). Results obtained with these methods were comparable and could be super-imposed. The second method has the advantage of providing a more understandable scale. This method was validated on a sorted cell population of less than fifty cells (data not shown), demonstrating that it could be adapted to samples with a very small yield of DNA 16. The reproducibility of the PCR assay was determined by including a standardisation sample in 17 consecutive PCR plates, run by 2 independent manipulators (Figure 4C). In this particular sample, the mean TREC content was 9.08 %, with a standard deviation of 0.611 %.

We next assessed the general applicability of our experimental procedure to determine levels of relative gene amplification. We used our assay to quantify GLI copy number in genomic DNA extracted from cell lines known to exhibit GLI gene amplification at various levels 17. We proceeded through optimisation as in Figure 2. Analysis using our assay indicated slightly higher levels of amplification (black bars) than previously reported (open bars) 18 (Figure 5A). Similar results were observed when multiplex-fluorescent PCR was used to analyse GLI copy number in the DNA extracted from these cell lines (grey bars). We also looked for evidence of MYC-C gene amplification in two cell lines HT29 and colo320 19 and in normal PBMCs using our assay. After performing optimisation, results (black bars) were similar to published amplification levels (open bars) for both cell lines but were normal in healthy PBMCs (Figure 5A).

We also applied the SYBR-Green I assay to tumour samples for the detection of the MYC-N gene amplification. These samples had been tested previously by standard procedures for the diagnosis of neuroblastoma using fluorescence in-situ hybridisation (FISH) 10. Our results (Figure 5B, black bars) compared well with the established amplification factor recorded for these tumours (open bars) and confirmed the presence of MYC-N amplification in 6 tumours. Using the Δ-ΔCt methods of calculation with one of the sample becoming the reference (either sample 4777, 100 fold amplification or sample 4334, 50 fold amplification) does not change the results significantly.

Point mutations and small insertions or deletions in the OPA1 gene have been shown to underlie more than half of all cases of dominant optic atrophy. In addition two larger deletions in OPA1 have been detected, each in a single family. One of these deletes exon 20 and the surrounding intronic DNA 13, while the other deletes the entire gene and over half a megabase of surrounding DNA 12. Traditional PCR-based mutation detection methods such as single strand contamination polymorphism (SSCP) analysis would miss these mutations since only the normal allele would amplify. We therefore designed a SYBR-Green I assay to quantify the number of copies of OPA1 exon 20 present in any given sample, which should therefore detect both of these deletions. We tested this on genomic DNA from patients and healthy members of the family (Figure 6A) with the deletion of the entire gene, and were able to confirm a reduction in relative copy number in these samples as compared to healthy members of the family (figure 6B and 6C). In this particular assay, exon 20 of the OPA1 gene is the target sequence. It is 165 bp long and limited the design of exon specific primers. The ratio of target gene to reference obtained in normal PBMCs during optimisation was 0.96 (Figure 2C and Table 1). However, the ΔCts in affected individual was on average more than 1 cycle compared to unaffected individuals indicating a reduction of at least 50%. We also applied this assay to the quantification of gene copy number in a transgenic mouse model. The gene Hop was 5' FLAG-tagged. Endogenous Hop sequences were also used as reference gene for the quantification. We identified founders carrying either 1 or 2 copies of the transgene (data not shown).

PBMCs were recovered using a step gradient sucrose separation procedure (Lymphoprep, Ficomed, Oslo, Norway) from 5.0 ml of heparinized blood. DNA was extracted using a standard proteinase K (Roche, Mannhein, Germany) digestion (over night at 55°C) followed by phenol/choloroform extraction (ready red, Appligen-Oncor, Ilkirch France) and DNA precipitation using ethanol. DNA was recovered by centrifugation, dissolved in TE (TRIS 10 mM, EDTA 1 mM, pH8), and stored at -20°C. For the TREC assay, a further CD4+ T-cell separation was performed using magnetic depletion (CD4+ T-cell negative selection kit, Metachem, Meylan, France).

Real-time PCR was performed using an ABI 7700 Sequence Detection System (PE Applied-Biosystems) in the presence of SYBR-green. The optimisation of the real-time PCR reaction was performed according to the manufacturer's instructions (PE Applied-Biosystems, User Bulletin 2 applied to the SYBR-Green I core reagent protocol) but scaled down to 25 μl per reaction. The PCR conditions were standard (SYBR-Green I core reagent protocol) and all reagents were provided in the SYBR-Green I core reagent kit, including AmpliTaq-GOLD polymerase (PE Applied-Biosystems). After optimisation (see Results section), nucleotide primers were used at various concentrations for the detection and quantification of GAPDH, signal and coding TREC, GLI, MYC-C, MYC-N, OPA1, FLAG and Hop.

Gene dosage of GLI was analysed by Fluorescent Differential PCR employing FAM labelled oligonucleotide primers 18. Two reference loci were also used: exon 19 of the cystic fibrosis transmembrane conductance regulator (CFTR) at 7q31 and ubiquitin-conjugating enzyme pseudogene (UBE2L2) at 12q12. PCR amplification was carried out in 30 μl 10 mM Tris-HCl buffer, pH 9.0, containing 50 mM KCl, 0.1% (v/v) Triton X-100, 1.5 mM MgCl₂, 0.2 mM dNTPs, 5 μM of each primer for UBE2L2 and CFTR, 100 ng of genomic DNA, 7.5 μM of the GLI primers and 2 units Taq DNA polymerase. PCR conditions consisted of an initial denaturation step of 95°C for 5 minutes, followed by 24 cycles of 95°C for 1 minute, 58°C for 1 minute and 72°C for 1 minute, with a final extension of 72°C for 10 minutes. A PCR using human genomic DNA as a control was routinely run in parallel. Moreover, experiments were performed in triplicate to ensure reproducibility of the technique. The FAM labelled products were separated through a denaturing polyacrylamide gel on an ABI 377 fragment analyser (PE Applied Biosystems). Quantitative analysis of the PCR products was performed using Genescan 672 software^© (PE Applied Biosystems). GLI gene dosage was calculated by comparing the ratios of GLI to reference gene peak areas generated by control DNA with those generated by the cell line DNAs, as described 18. Primer sequences were as follows: GLI-F-d TGA TGC AGT TCC TTT ATT ATC AGG; GLI-R-(FAM)-d AGA GTA GGG AAT CTC ATC CAT CA, giving a product of 200 bp. UB-F(FAM)-d CGA AGA GCA CAC TTA AAG ATC TG; Ub-R-d GGT CAG CCT-GAA GTG GAT GCT CA generating a 173 bp product. CFTR19-F-dCCT ACC AAG TCA ACC AAA CC; CFTR19-R(FAM)-dACA TTG CTT CAG GCT ACT GG; generating a product of 268 bp.