Michael W. Pfaffl   Graham W. Horgan & Leo Dempfle
Nucleic Acids Research 2002 May 1; 30(9): E36

=>      download  new  REST ©  versions     <=

Summary

Introduction

Reverse transcription (RT) followed by polymerase chain reaction (PCR) is a powerful tool for the detection and quantification of mRNA. Nowadays real-time RT-PCR is widely and increasingly used, because of its high sensitivity, good reproducibility, and wide quantification range (1, 2). It is the most sensitive method for the detection and quantification of gene expression levels, in particular for low abundant mRNA (1, 2), in tissues with low concentrations of mRNA (e.g. bone marrow, fatty tissues), from limited tissue sample(e.g. biopsies, single cells) (3, 4) and to elucidate small changes in mRNA expression levels (1, 2, 5). However, it is a very complex technique with various substantial problems associated with its true sensitivity, reproducibility and specificity and, as a fully quantitative methodology, it suffers from the problems inherent in real-time RT-PCR. Generally two quantification strategies can be performed: An absolute and a relative quantification. In absolute quantification the absolute mRNA copy number per vial is determined by comparison to appropriate external calibration curves (2). The relative expression is based on the expression levels of a target gene versus a reference gene and is adequate for most purposes to investigate physiological changes in gene expression levels. Some mathematical models have already been developed to calculate the relative expression ratios of single samples (6-8), with or without efficiency correction. Equation 1 shows the most convenient mathematical model, which includes an efficiency correction for real-time PCR efficiency of the individual transcripts (6). 

Animal experiment,  total RNA extraction and reverse transcription

Total RNA extraction was performed from rat liver as described previously (12). Adult rats were either fed with physiological zinc concentrations (control group, 58 ppm Zn, n = 7) or suffered 22 to 29 days under zinc depletion (sample group, 2 ppm Zn, n = 6) (13). Isolated total RNA integrity was electrophoretically verified by ethidium bromide staining and by an average optical density (OD) OD260/OD280 nm absorption ratio of 1.97 (range 1.78 – 2.09). Either 330 ng, 1000 ng or 3000 ng total RNA were reverse transcribed with 100 U of Super script II Plus RNase H- Reverse Trancriptase (Gibco Life technologies, Gaithersburg, USA) in a volume of 40 µl, using 100 µM random hexamer primers (Pharmacia Biotech, Uppsala, Sweden) according to the manufacturers instructions. Therefore concentrations of 8.25 ng, 25 ng or 75 ng cDNA (= reverse transcribed total RNA) per µl were achieved. 

Optimisation of RT-PCR

Highly purified salt-free (HPSF) primer for target gene Metallothionein (MT) (forward primer: CTC CTG CAA GAA GAG CTG CT; reverse primer: TCA GGC GCA GCA GCT GCA CTT) and for reference gene GAPDH (Forward primer: GTC TTC ACT ACC ATG GAG AAG G; Reverse primer: TCA TGG ATG ACC TTG GCC AG) were generated commercially (MWG Biotech, Ebersberg, Germany). The MT primer set is able to amplify the transcripts of MT isoform 1 and MT isoform 2 mRNA. Conditions for real-time PCRs were optimised in a gradient cycler (Mastercycler Gradient, Eppendorf, Germany) with regard to Taq DNA polymerase (Roche Molecular Biochemicals, Basel, Switzerland), forward and reverse primers, MgCl2 concentrations (Roche Molecular Biochemicals), and various annealing temperatures (54°C–66°C). RT-PCR amplification products were separated on a 4% high resolution NuSieve agarose (FMC Bio Products, Rockland, USA) gel electrophoresis and analysed with the Image Master system (Pharmacia Biotech). Optimised conditions were transferred to the following LightCycler real-time PCR protocol. 

LightCycler real-time PCR

For determination of test and software variations all applications of different total cDNA input were performed in triplets (MT 1-3 and GAPDH 1-3). Real-time PCR mastermix was prepared as follows (to the indicated end-concentration): 6.4 µl water, 1.2 µl MgCl2 (4 mM), 0.2 µl forward primer (0.4 µM), 0.2 µl reverse primer (0.4 µM) and 1 µl LightCycler - Fast Start DNA Master SYBR Green I (Roche Molecular Biochemicals). 9 µl of master-mix was filled in the glass capillaries and 1 µl volume cDNA (either 8.25 ng, 25 ng or 75 ng) was added as PCR template. Capillaries were closed, centrifuged and placed into a cycling rotor. A four step experimental run protocol was used: (1) Denaturation program (10 min @ 95°C), (2) amplification and quantification program repeated 40 times (15 s @ 95°C;10 s @ 60°C for MT or 10 s @ 58°C for GAPDH; 20 s @ 72°C; 5 s @ 86°C for MT or 5 s @ 84°C for GAPDH with a single fluorescence measurement), (3) melting curve program (60°C to 99°C with a heating rate of 0.1°C per second and a continuous fluorescence measurement)   (4) cooling program down to 40°C. To improve SYBR Green I quantification a high temperature fluorescence measurement point at the end of the fourth segment was performed (14). It melts the unspecific PCR products below the chosen temperature, e.g. primer dimers, eliminates the non-specific fluorescence signal and ensures accurate quantification of the desired GAPDH and MT real-time RT-PCR product, respectively. For the described mathematical model it is necessary to determine the crossing points (CP) for each transcript. CP is defined as the point at which the fluorescence rises appreciably above the background fluorescence. In this study “Second derivate maximum Method” was performed for CP determination, using LightCycler Software 3.5 (15).

Statistics

Figure 1: Page 1 - Introduction

On the introduction page the basic settings are made for the REST application. Up to four genes and one reference gene can be labelled. Different background colours in the spreadsheets and the print command are shown and described. Pink cells indicate cells for data input, blue cells indicate data output, grey cells are used for calculation purposes and output of the CP variation, the red box will start the Randomisation Test itself and the printer icon indicates “print this page”. Further, the relative expression equation is given with direct links to the data input section on page CP input + Randomisation Test. 

Figure 2: Page 2 - PCR efficiency

The PCR efficiency calculation is facultative and not obligatory for the user. To generate the data basis for the determination of PCR efficiency of each transcript, it is recommended to use various dilutions in triplets of a pool of all available cDNAs. This ensures the best estimation of the PCR efficiency. If the user wants to determine the real-time PCR efficiencies, an import via copy and paste of cDNA starting concentrations in dilution row and the corresponding CP values measured by the real-time PCR machine is possible. Depending on the real-time PCR platform used, CP values can be determined either by “Threshold Cycles” = “Fit Point Method” (all platforms) or “Second derivate maximum Method” (only LightCycler). Up to three CPs can be inserted in the table (run 1 - 3) per cDNA starting concentration and REST determines the slope with a logarithmic algorithm, as published earlier (1, 6, 18), as well as an indication of the linearity of this logarithmic alignment using Pearson correlation coefficient. The real-time PCR efficiencies were calculated from the slope, according to the established equation E = 10 [–1/slope] (1, 18). E is in the range from 1 (minimum value) to 2 (theoretical maximum and optimum). If no real-time PCR efficiencies are calculated here, REST assume an optimal efficiency of E = 2.0 on the following pages and further procedures.

Figure 3: Page 3 - CP input + Randomisation Test

Figure 4: Page 4 - Ratio + Variation output

 
 Statistical model:  Pair Wise Fixed Reallocation Randomisation Test ©

Differences in expression between control and treated samples were assessed in group means (figure 1) for statistical significance by randomisation tests (19, 20). Permutation or randomisation tests are a useful alternative to more standard parametric tests for analysing experimental data. They have the advantage of making no distributional assumptions (such as Normality) about the data, while remaining as powerful as more standard tests (20). They calculate p-values by obtaining the proportion of random allocations of the observed data to the control and treated sample groups that would give greater indications of a treatment effect than that observed. If this is small, then there is evidence that the observed treatment effect is not simply the result of random allocation. The test thus makes no assumptions concerning the distribution of measured gene expression in any hypothesised population – it assumes only the random allocation of treatment. In practice, it is impractical to examine all possible allocations of data to treatment groups, and a random sample is drawn. If 2000 or more samples are taken, a good estimate of p-value (SE < 0.005 at p = 0.05) is obtained. In the applied Pair Wise Fixed Reallocation Randomisation Test © for each sample, the CP values for reference and target genes are jointly reallocated to control and sample groups (= pair wise fixed reallocation), and the expression ratios are calculated as described above. They are deemed to give greater indications of a treatment effect than that actually observed if   | log R | > | log R0 |   where R0 is the true expression ratio and R the result of reallocation. In Pair Wise Fixed Reallocation Randomisation Test © a two sided test was performed. The randomisation tests were carried out using a Microsoft Excel ® macro (17) attached to a purpose-built spreadsheetand running in the background of REST.

Confirmation of primer specificity

Specificity of RT-PCR products was documented with high resolution gel electrophoresis and resulted in a single product with the desired length (MT: 106 bp, GAPDH: 197 bp). In addition a LightCycler melting curve analysis was performed which resulted insingle product specific melting temperatures: 87.4°C (GAPDH) and 89.7°C (MT). No primer primer-dimers formation were generated during the applied 40 real-time PCR amplification cycles.

Real-time PCR amplification efficiencies and variation

Real-time PCR efficiencies were calculated from the slopes given in LightCycler software (15). The corresponding real-time PCR efficiency (E) of one cycle in the exponential phase was calculated according to the equation: E = 10 [–1/slope], as described earlier (1, 6, 18). Investigated transcripts showed real-time PCR efficiency rates for MT (EMT = 1.67) and GAPDH (EGAPDH = 1.88) in theinvestigated range from 120 pg to 75 ng cDNA input, repeated 6 times, with high linearity (Pearson correlation coefficient  r > 0.989). To mimic different reverse transcription efficiencies and to confirm precision and reproducibility of real-time PCR, as well as for REST, three replicates of real-time RT-PCR at each of various cDNA input concentrations (3-times more and 3-times less concentrated) were performed and real-time RT-PCR and REST variations (CV) were determined. As shown in table 1 variations of investigated transcripts are based on the CP variation and remained stable between 2.43% and 10.03% for MT and 1.59% and 12.89% for GAPDH; the later showing a dependence on the cDNA input in real-time PCR. CP itself decreased with increasing cDNA input in both factors and groups. 

Table 1: ANOVA of CV of inter-assay variation of MT and GAPDH crossing points (CP) determined in rat liver by real-time RT-PCR started either with 8.25 ng (1-3), 25 ng (4-6) or 75 ng (7-9) cDNA per capillary. Given are the mean CP and the coefficient of variation (CV), each one based on n=7 for control and n=6 for sample, respectively.

Variation and reproducibility of REST

On the basis of the previously published mathematical model (6) REST calculates the relative expression ratios on the basis of group means for target gene MT versus reference gene GAPDH and tests the group ratio results for significance. Normalised and not normalised expression results were compared: Normalised by GAPDH expression: As presented in table 2, the down-regulation factor (reciprocal value of ratio) of MT mRNA in the case of zinc deficiency was calculated by REST starting from different cDNA concentrations. Further different runs (MT 1-3 and GAPDH 1-3 => n = 3 x 9) were compared to calculate all possible combinationsbetween individual real-time runs. Derived variations and the influence of deviating cDNA starting amounts on the REST calculated relative expression ratio, and the significance of the performed randomisation test are presented in table 2. Over all investigated combinations (n = 27) a mean factor of down-regulation of 44.505 (CV = 26.83%) was observed. No significant differences between cDNA starting concentration on expression ratio could be found. 

No normalisation by GAPDH

In table 3 the factor of down-regulation of MT mRNA in the case of zinc deficiency was calculated by REST without normalisation by reference gene. For MT a mean down-regulation factor of 28.081 (CV = 10.22%) and for GAPDH of 0.677 (CV = 29.79%) were observed. No significant differences between cDNA starting concentration on expression ratio could be found either for MT or GAPDH. 

Crossing point determination

For the determination of CP in general two methods can be chosen: “Fit Point Method” or adequate methodologies like “Threshold Cycle” (22, 23) where CP will be measured at constant fluorescence level and “Second derivative maximum Method” where CP will be measured at the maximum increase or acceleration of fluorescence, even if the fluorescence levels between curves are different (18). Besides the LightCycler, the“Fit Point Method” or “Threshold Cycle” are used in TaqMan ® (PE Applied Biosystems,  Foster City, CA, USA), RotoGene ® (Corbett Research, Sydney, NSW, Australia), iCycler ® Thermal Cycler (Bio-Rad, Hercules, CA, USA) and Multiplex Quantitative PCR System ® (Stratagene, La Jolla, CA, USA). “Second derivate maximum Method” is an algorithm exclusively used in LightCycler software (15). 

Normalisation

The normalisation of the target gene with an endogenous standard is recommended. REST allows for a normalisation of the target genes with a reference gene. On  both mathematical models the Pair Wise Fixed Reallocation Randomisation Test © is performed and the results are presented in the appropriate output windows. Researchers can decide if they want to correct the data or not. The basis of data normalisation is the expression result of an endogenous desirable unregulated reference gene transcript to compensate inter PCR variations (sample to samplevariations) between the runs. If the CP deviation of the chosen reference gene has the same mean in the control as in sample group mean ( DCPref  (MEAN contol – MEAN sample) = 0 ) then stable and constant reference gene mRNA level is given. Real-time RT-PCR specific errors in the quantification of mRNA transcripts are easily compounded with any variation in the amount of starting material between the samples. This is especially relevant when the samples have been obtained from different individuals, and will result in the misinterpretation of the derived expression profile of the target genes (1). Here some questions arise: What is the appropriate reference gene for an experimental treatment and investigated tissue? (11, 24). Commonly used housekeeping genes (9) are suitable for reference genes, since they are present in all nucleated cell types and necessary for basic cell survival. The mRNA synthesis of housekeeping genes is considered to be stable in various tissues, even under experimental treatments (9-11). But numerous treatments and studies have already shown that the mentioned housekeeping genes are regulated and vary under experimental conditions (25-28). It remains with the investigator to decide which reference gene is best for a reliable normalisation. In addition, the endogenous control should be expressed at roughly the same CP level as the target gene (1). At the same CP level, reference and target underwent already the same cycle condition, real-time RT-PCR kinetics, in respect to polymerase activation (heat activation of polymerase) or inactivation and reaction end product inhibition by the generated RT-PCR product (29). The software can give you the essential hints if a normalisation via the chosen reference is useful (by the factor of regulation and p-value-nn of the randomisation test of the reference), or if the reference is not suitable, because it is significantly regulated. 

Relative quantification software

Up to now only one relative quantification software for real-time PCR is available and distributed by Roche Molecular Biochemicals “LightCycler Relative Quantification Software” (30). The mathematical algorithm on which the Roche Molecular Biochemicals software is based is unpublished, and might be the one discussed earlier (6, 8). 

Pair Wise Fixed Reallocation Randomisation Test ©

 Conclusion

Acknowledgements