Unless the copy numbers are specified, all green fluorescent protein (GFP) plasmid (4700 bp) standards were constructed with a 3-fold dilution series ranging from 8100 to 100 copies. The copy numbers of the GFP plasmid and PRKR plasmid (5547 bp) are calculated based on the following formula (), in which n is the number of base pairs, m is the mass of the DNA, N_A is Avogadro's number (6.02 × 10²³ bp/mol) and M is the average molecular weight of a base pair (610 g/mol). n=(m×NA)/Mm=n×1.096×10−21g/bp

All primers were designed with Primer Express (PE Applied Biosystems, Perkin Elmer, Foster City, CA) based on target sequences obtained from the Affymetrix database (NetAffx™ Analysis Center, ) for each gene of interest. Primers were designed to target the 3′-untranslated region (3′-UTR) of transcript sequences except GFP, HsPRKR P1, HsPRKR P2 and HsPRKR P3, all of which target upstream exons. We ensured that the primer pair will specifically amplify the target sequence by searching for the nucleotide sequences that contain both primer sequences on opposing strands in the NCBI Genbank database using BLAST (). Primer pairs were discarded if the alignment of both primer sequences could result in the amplification of non-specific sequences. All new primers were tested by real-time PCR using a 3-fold serial dilution (9–0.111 ng/reaction) of gDNA and water as non-template control followed by gel electrophoresis, in order to determine amplification efficiency, specificity and the presence/absence of primer dimers. The amplification efficiency for each primer was determined from the linear slope of standard curve; only primers with a standard curve slope between −2.92 and −3.92 were used for further quantification. Inefficient and/or non-specific primers were excluded. Lastly, Verification of Amplicon Specificity using Restriction Enzymes (VASRE, see below) was performed to validate the sequence specificity of the expected amplicons. A supplementary table contains all the sequences of primers used in this paper (Online Supplementary Table).

RNA extraction was performed according to the manufacturer's manual for the RNeasy Kit (Qiagen). Both human and mouse gDNA from various tissues (human tonsil, liver, subcarinal lymph nodes and lung; mouse liver) was isolated according to the manufacturer's manual for the DNeasy Kit (Qiagen). Mouse-tail gDNA was isolated by phenol extraction after overnight proteinase K digestion, and then DNA is subsequently precipitated with ethanol and resuspended in 10 mM Tris, 1 mM EDTA, pH 7.4. For the generation of external standards, gDNA was stored as a 9 ng/μl stock at −20°C, and a fresh 3-fold serial dilution was prepared for each real-time PCR experiment (9–0.111 ng/reaction).

DNase treatment was performed on RNA samples using DNA-free (Ambion Inc.) according to the manufacturer's manual in order to remove gDNA carry-over. In a single RT reaction, 5 μg of RNA was mixed with RNase-free water (Qiagen) and 1 μl of oligo(dT₂₃) (Sigma–Aldrich Co.) to a final volume of 12 μl, then incubated at 90°C for 10 min. To this, 4 μl of 5× First-Strand buffer, 2 μl of 0.1 M DTT, 1 μl of 10 mM dNTP and 1 μl (200 U) of Superscript II (Invitrogen Co.) were added. The mixture was incubated at 42°C for 90 min followed by an extension period for 15 min at 70°C.

Real-time PCRs were performed using the ABI Prism 7900HT (PE Applied Biosystems) in 384 micro-well plates. All samples, including the external standards and non-template control, were run in triplicate. The reaction conditions had been established through a series of preliminary optimization experiments. Each 10 μl reaction contained 1× PCR buffer (Sigma–Aldrich Co.), 3 mM MgCl₂, 0.2 mM dNTP, 1 nM forward and reverse primers, 1/50 dilution of ROX reference dye (Sigma–Aldrich Co.), 3/100 000 dilution of SYBR Green I (Sigma–Aldrich Co.), 0.05 U of JumpStart Taq polymerase (Sigma–Aldrich Co.) and template DNA. Template was either a 3-fold serial dilution (9–0.111 ng/reaction) of gDNA for generation of a standard curve, water for a non-template control which was included to confirm the absence of DNA contamination in the reaction mixture, 10 ng of cDNA generated from total RNA or 1 ng of cDNA generated from mRNA. The reaction was initiated by activation of Taq polymerase at 95°C for 3 min, followed by 40 three-step amplification cycles consisting of 10 s denaturation at 95°C, 15 s annealing at 65°C and 20 s extension at 72°C. A final dissociation stage was run to generate a melting curve for verification of amplification product specificity. Real-time PCR was monitored and analyzed by the Sequence Detection System version 2.0 (PE Applied Biosystems).

Our VASRE system used a panel of six restriction enzymes (HpyCH4V, HaeIII, HinP1I, HpyCH4IV, MspI and RsaI, all from New England Biolabs) to validate the sequence specificity of the amplicons produced by real-time PCR primer pairs. The PCR products produced during real-time PCR screening of new primer pairs were digested with each enzyme in a separate reaction. Each VASRE reaction contained 0.5 U of restriction enzyme, 1× of appropriate digestion buffer, 0.8 μl of 25 mM MgCl₂, 0.3 μl of SYBR Green I 1/1000 dilution (Sigma–Aldrich Co.), 0.1 μl of ROX reference dye (Sigma–Aldrich Co.) and 1 μl of cDNA template taken from a 9 ng gDNA standard originally screened in real-time PCR to a final volume of 10 μl, and then incubated at 37°C for 1 h. The reaction mixture was returned to a clean PCR plate, and a dissociation curve was produced by measuring the peak fluorescence change as the temperature was increased. The dissociation curve after digestion with each restriction enzyme was compared to the dissociation curve generated after incubation of the PCR product in the absence of enzyme.

RT reactions for microarray were carried out as described above, except that T7 oligo(dT) primer (Affymetrix, Inc.) was used instead of oligo(dT₂₃) (Sigma–Aldrich Co.). Microarray hybridizations were performed using the Affymetrix HU133A chip (Affymetrix, Inc.) according to the user manual. The array data were analyzed with Microarray Suite ver. 5.0 (Affymetrix, Inc.).

Absolute quantification by real-time PCR requires amplification of the target sequence from a series of external standards of known quantities to generate a standard curve from which the quantity of the target sequence in the unknown sample can be calculated (18–21). Since all gene sequences for a given species are represented, gDNA could serve as a universal standard for the absolute quantification of any expressed gene. The mass of the haploid human genome (C-value) is ∼3.5 pg (), and therefore 1 ng of gDNA contains ∼286 copies of a single-copy gene. Generation of standard curves ranging up to 10 000 copies would therefore require <35 ng of gDNA. We typically generate a standard curve using a 3-fold dilution series to produce five quantities ranging from 9 ng (∼2574 copies) to 0.111 ng (∼31 copies) of gDNA isolated from tonsils as a source of normal diploid cells. A typical standard curve is shown in Figure 1 generated from primers designed against the human CCL2 gene. The C_t values range between cycles 23 and 30 (threshold = 0.20), producing a standard curve with a slope of −3.33 (R² = 0.999).

An essential consideration for using gDNA as an external standard in real-time PCR is designing target-specific primer pairs that do not span introns. Although it is common to design intron-spanning primer pairs to mitigate signals arising from contamination of RNA samples with gDNA, we routinely treat samples with DNase to avoid this non-specific amplification.

Primer pairs designed against a gene of interest were experimentally tested by generating a standard curve with tissue-derived gDNA. Acceptable primers meet the following criteria for efficiency and specificity.

High amplification efficiency: Efficiency of amplification can be determined by E%=(10(−1/slope)−1)×100%, using the slope from a standard curve plot (C_t versus log copy number). Efficiency (100%) is therefore defined by a slope of −3.32. We typically utilize primer pairs that produce standard curves with slopes between −2.92 and −3.92 (Efficiency = 100 ± 20%).

While gDNA has the advantage of acting as a source for any gene target, it also contains a large excess of non-target sequence. In contrast, plasmids or cDNA standards contain a relatively small amount of non-target sequence. We postulated that amplification efficiency may be affected by both the inherent properties of the primer pair and the complexity and nature of the DNA sample containing the target to be amplified. As gDNA is to be used as an external standard for the quantification of gene expression in cDNA, it is necessary to ensure that the differences in complexity do not confound these measurements.

Standard curves were generated by amplifying a dilution series of plasmid containing the gene for GFP by real-time PCR and plotting the C_t values against the concentration of plasmid. To examine the effect of sample complexity on amplification efficiency, increasing amounts of gDNA or cDNA were spiked into the GFP dilution series. The quantities of gDNA used were identical to quantities we typically use for generating standard curves (9–0.111 ng), and the quantities of cDNA represent typical amounts used in our experiments. The standard curves were assessed for two properties, the coefficient of correlation (R²), and the amplification efficiency (E) determined from the slope of the standard curve. As shown in Table 1, high-quality standard curves were produced in the absence or presence of background DNA (R² > 0.989). Furthermore, neither the presence of background gDNA or cDNA significantly affected amplification efficiency.

We extended this analysis by assessing GFP amplification in the presence of even higher concentrations of DNA ranging between 30.8 and 60 ng. GFP amplification was inhibited in the presence of higher gDNA concentrations (Table 2) whereas cDNA quantities over this range was not inhibitory (data not shown), suggesting that the bulk amount of background nucleic acid may not necessarily be the sole determinant in inhibiting PCR amplification and that the structural nature of the gDNA may also contribute. Mechanical shearing of gDNA effectively removed much of the inhibition of GFP amplification (Table 2).

Although the range of gDNA that we typically use in our standard curves does not impair amplification from GFP plasmid, we also wanted to ensure that amplification directly from the gDNA is not affected by the mechanical shearing that may result in loss of target sequences. To examine this, we generated standard curves using sheared or unsheared gDNA in the range of 9–0.111 ng with eight different primer pairs. Shearing the DNA affected neither the slope nor the correlation coefficient for any of the primer pairs tested (Table 3).

Plasmid DNA is often used as an external standard for absolute quantification in real-time PCR. We compared amplification from ∼14.5 fg (∼2574 copies) of plasmid containing the human PRKR gene to amplification from the same target in 9 ng (∼2574 copies) of gDNA extracted from three different tissues (HT, Human tonsil; HS, Human subcarinal lymph node; and HL, Human lung tissue). The C_t score measured in each sample (n = 12) was not significantly different between the groups (Supplementary Figure A).

HT gDNA was used to construct a standard curve from which the copy numbers of eight different target genes were determined in human gDNA from both subcarinal lymph node and lung. Each genomic sample was assessed at serial dilutions that covered the range of the standard curve. At each input level, the number of copies was not significantly different for each of the eight targets. Furthermore, the number of copies determined from each source was in agreement with each other at each dilution (Supplementary Figure B).

The three gDNA sources were compared by constructing standard curves from each and assessing gene expression in various cDNA samples. The number of copies of the PRKR target was determined in 10 ng of cDNA from U937 cell line treated with and without interferon for 4 h and in 1 ng of cDNA from peripheral CD4⁺ and CD8⁺ T cells. No significant differences were seen between the values obtained from each standard curve (Supplementary Figure C).

We have assessed copy numbers using five different primer pairs targeting different regions of the PRKR gene (Figure 3A) in 9 ng of human tonsil, subcarinal lymph node and lung gDNA. No significant differences were observed in the assessed quantity between the three sources of gDNA regardless of the target position (Figure 3B). In contrast, the copy number determined in cDNA samples obtained from the U937 cell line treated with and without interferon for 4 h or from peripheral CD4⁺ and CD8⁺ T cells is affected by the target site (Figure 3C). The assessed quantity of transcript copy decreases as the target region moves upstream from the 3′ end of the transcript. Primers closest to each other in the 3′-UTR (HsPRKR P4 and HsPRKR P5 in Figure 3A) are generally in agreement although less so when assessing expression in interferon treated cells due to the high copy numbers which exceed the standard curve. In spite of the difference in copy numbers between each primer pair, the ratios of target copy numbers between IFN treated and untreated samples and between CD8+ and CD4+ samples were comparable for each primer pair (Figure 3D).

This position-dependent effect is likely due to the efficiency of the reverse transcription reaction, which has long been appreciated, and it highlights the importance of designing primers near the 3′ end to help mitigate this effect. We examined this further by assessing the transcript copy number for eight different genes in CD4⁺ and CD8⁺ T cell cDNA using two sets of primer pairs designed against the 3′ region of each gene (Figure 4A). The assessed quantity was highly correlated (R² = 0.968) between the two sets of primers (Figure 4B).

Quantitative real-time PCR is often used to validate the results obtained from microarray experiments (24–26). We examined the relative gene expression profiles of human CD4⁺ and CD8⁺ T cells obtained by microarray analysis (M. Hyrcza, M. Ostrowski and S. D. Der, unpublished data) and validated the levels of 43 single-copy genes. Of these, 29 were not different between the two cell types (data not shown). The remaining 14 were differently expressed on the microarray and the direction of change for each was confirmed by real-time PCR although the log₂ CD8⁺/CD4⁺ ratios tended to be higher in real-time PCR than in microarray analysis (Figure 5A). Nevertheless, the quantities determined by qPCR were highly correlated (R² = 0.981) with the quantities determined by microarray analysis (Figure 5B).

Other applications that have relied upon real-time PCR are single nucleotide polymorphism detection and genotyping (4,13,14). We have applied the gDNA external standard to genotype human-tau transgenic mice using real-time PCR. Mouse-tail gDNA samples were isolated from the F2 generation of a homozygous/null cross. Expected ratios were 1/2 heterozygous, 1/4 homozygous and 1/4 null. The gDNA was amplified with two human-tau specific primer pairs as well as two primer pairs for internal reference genes (GRO1 and IL6). Human tonsil gDNA was used as an external standard to measure human-tau gene, and mouse liver gDNA was used to quantify the internal reference genes. The quantity of each gene obtained by real-time PCR was normalized with the internal reference genes to account for multiple insertions. A clear distinction is evident between the three populations with 12/20 heterozygous, 4/20 homozygous and 4/20 null (Figure 6). Multiple transgene copies of the inserted gene are also indicated, three to four copies in the haploid genome.