The power and utility of RNAi to silence gene expression has made it the tool of choice for reverse genetics in eukaryotic systems. The most frequently used applications for RNAi in mammalian systems include: testing hypotheses for gene function, functionally screening and identifying target genes, validating genes as potential drug targets, and examining the use of short interfering RNAs (siRNA) as therapeutic agents.
This article describes how to determine the percent remaining gene expression and percent knockdown of gene expression when assessing the effects of siRNAs with real-time RT-PCR. It also discusses how the change is affected by these calculations.
This article describes how to determine the percent remaining gene expression and percent knockdown of gene expression when assessing the effects of siRNAs with real-time RT-PCR. It also discusses how the change is affected by these calculations.
Calculating % Remaining Gene Expression and % Knockdown
The comparative C
T method (ΔΔC
T) for relative quantitation is a commonly used method for measuring siRNA-induced silencing or knockdown of a particular gene when using TaqMan® Gene Expression Assays. In this method, data are normalized using a control transcript (e.g., 18S rRNA), and the normalized expression value (ΔC
T) for the gene of interest in the experimental sample is compared to the equivalent ΔC
T for the negative control siRNA-treated sample (NC).
- The experimental sample (e.g., RNA from siRNA-transfected cells) is assessed with two TaqMan assays: one that measures the RNA levels of the gene of interest (e.g., GAPDH) and one that measures an endogenous calibrator RNA (e.g., 18S rRNA). The normalized expression value ΔCT-Sample is calculated as follows: ΔCT-Sample = (CT-Sample with GAPDH primer/probe – CT-Sample with 18S rRNA primer/probe)
- A nontargeting negative control siRNA-transfected RNA sample is also required. The same TaqMan assays used to assess the experimental samples are used to assess the negative control samples. The ΔCT-NC is calculated as follows:
- ΔCT-NC = (CT-NC with GAPDH primer/probe – CT-NC with 18S rRNA primer/probe)
- The normalized expression values of the target gene are compared in the experimental sample and negative control by calculating ΔΔCT:
- ΔΔCT = ΔCT-Sample – ΔCT-NC
- The ΔΔCT is used to calculate the percent remaining gene expression and the percent knockdown
Understanding the Variability of the Data
The formula for converting ΔΔC
T to gene expression levels (percent remaining gene expression or percent knockdown) distorts the appearance of the variability of samples. When the ΔΔC
T is small, there is a larger incremental change in percent remaining gene expression or percent knockdown compared to when the ΔΔC
T is large, because one C
T value represents a 2-fold change. Samples can have similar precision at the original C
T levels, the ΔC
T-NC, ΔC
T-Sample, or the final ΔΔC
T level, yet display wide variability in the percent remaining gene expression or percent knockdown (Figure 2). See sidebar on page 10 for a brief discussion of precision and accuracy.
Figure 2. Data Variability Translates to Large Differences in the Ability to Identify Targets. The variability disappears as the percent remaining gene expression decreases and the percent knockdown increases.
The comparison of variability in ΔΔC T, percent remaining gene expression, and percent knockdown is shown graphically in Figures 2B and 2C.
The precision of the original real-time assay is the same at low or high ΔΔC T values. However, when ΔΔC T values are low, even small variability results in large changes in percent knockdown, essentially making target genes harder to identify. As ΔΔC T values increase, the variability appears diminished by the calculation, making the target genes easier to identify. This is important when viewing graphs and error bars for percent remaining gene expression or percent knockdown of different samples. The error bars of the samples with higher knockdown will appear smaller, even when the raw data used in the knockdown calculation was as variable as other samples with low knockdown and error bars that appear larger.
Figure 3. Examples of Accuracy and Precision for Normally Distributed Data. The light brown line indicates the “true” answer (zero) for a test. The black line represents an accurate and precise result (0 ± 1). The blue curve shows data that is less precise (has increased variability) but still accurately defines the average (0 ± 10). The green curve (10 ± 10) represents data that are both more variable and less accurate than data graphed by the black or blue curves, but it is correct a small percentage of the time. The orange curve represents very precise data (10 ± 1); however, it is inaccurate, because it does not encompass the correct answer.
Figure 2. Data Variability Translates to Large Differences in the Ability to Identify Targets. The variability disappears as the percent remaining gene expression decreases and the percent knockdown increases.
- Rows 1–9 (Figure 2A) show how an incremental change of 0.25 in ΔΔCT alters the percent remaining gene expression and percent knockdown from 0 to 2 ΔΔCT. The remaining gene expression ranges from 0 to 25%, which corresponds to a knockdown of 0–75%.
- Rows 10–14 show an incremental change of 1.0 in ΔΔCT across a range of 3 to 7 ΔΔCT. The remaining gene expression changes 12.5–0.78% (slightly more than 11%). The corresponding 11% change in knockdown ranges from 87.5% to 99.2%.
- Rows 15–22 show an incremental change of 1.0 in ΔΔCT across a range of 8 to 15 ΔΔCT. Despite this large range of change at the ΔΔCT level, the remaining gene expression changes by less than 0.4%. Over the 7 point spread in ΔΔCT (8 to 15 ΔΔCT), the knockdown varied only from 99.6% to 100%.
The comparison of variability in ΔΔC T, percent remaining gene expression, and percent knockdown is shown graphically in Figures 2B and 2C.
The precision of the original real-time assay is the same at low or high ΔΔC T values. However, when ΔΔC T values are low, even small variability results in large changes in percent knockdown, essentially making target genes harder to identify. As ΔΔC T values increase, the variability appears diminished by the calculation, making the target genes easier to identify. This is important when viewing graphs and error bars for percent remaining gene expression or percent knockdown of different samples. The error bars of the samples with higher knockdown will appear smaller, even when the raw data used in the knockdown calculation was as variable as other samples with low knockdown and error bars that appear larger.
Figure 3. Examples of Accuracy and Precision for Normally Distributed Data. The light brown line indicates the “true” answer (zero) for a test. The black line represents an accurate and precise result (0 ± 1). The blue curve shows data that is less precise (has increased variability) but still accurately defines the average (0 ± 10). The green curve (10 ± 10) represents data that are both more variable and less accurate than data graphed by the black or blue curves, but it is correct a small percentage of the time. The orange curve represents very precise data (10 ± 1); however, it is inaccurate, because it does not encompass the correct answer.