We proposed a new real-time RT–PCR scheme for miRNA quantification (). It included two steps: RT and real-time PCR. First, the stem–loop RT primer is hybridized to a miRNA molecule and then reverse transcribed with a MultiScribe reverse transcriptase. Next, the RT products are quantified using conventional TaqMan PCR.
Figure 1 Schematic description of TaqMan miRNA assays, TaqMan-based real-time quantification of miRNAs includes two steps, stem–loop RT and real-time PCR. Stem–loop RT primers bind to at the 3′ portion of miRNA molecules and are reverse (more ...)
The dynamic range and sensitivity of the miRNA quantification scheme were first evaluated using a synthetic cel-lin-4 target. Synthetic RNA was quantified based on the A260 value and diluted over seven orders of magnitude. The cel-lin-4 TaqMan miRNA assay showed excellent linearity between the log of target input and CT value, demonstrating that the assay has a dynamic range of at least 7 logs and is capable of detecting as few as seven copies in the PCR reaction ().
Figure 2 Dynamic range and sensitivity of the TaqMan lin-4 miRNA assay. (A) Amplification plot of synthetic lin-4 miRNA over seven orders of magnitude. Synthetic RNA input ranged from 1.3 × 10−3 fM (equivalent to 7 copies per reaction) to 1.3 × (more ...)
Eight additional miRNA assays were also validated using mouse lung total RNA. The RNA input ranged from 0.025 to 250 ng (). The CT values correlated to the RNA input (R2 > 0.994) over four orders of magnitude. A negative control assay, cel-miR-2, did not give a detectable signal, even in reactions with 250 ng mouse total RNA.
Correlation of total RNA input to the threshold of cycle (CT) values for eight miRNA assays. Mouse lung total RNA input ranged from 0.025 to 250 ng per RT reaction. A Caenorhabditis elegans miRNA (miR-2) was included as a negative control assay.
The expression profile of five miRNAs was determined in seven different mouse tissues to create a miRNA expression map. The copy number per cell was calculated based on the input total RNA (assuming 15 pg/cell) and the standard curve of synthetic lin-4 target. Several interesting observations were made from this expression map. First, miRNAs are very abundant, averaging 2390 copies per cell in these tissues. The level of expression ranged from less than 10 to 32
090 copies per cell. Of the 12 miRNAs, miR-16 and miR-323 were the most and least abundant miRNAs, respectively, across all tissues. In addition, each tissue had a distinctive signature of miRNA expression. The overall level of miRNA expression was highest in mouse lung and lowest in embryos. Finally, the dynamic range of miRNA expression varied greatly from less than 5-fold (let-7a) to more than 2000-fold (miR-323) among these seven tissues ().
Expression profiles of five miRNAs across seven mouse tissues
To assess the need for RNA isolation, we added cell lysates directly to miRNA assays. The equivalent of 2.5–2500 cells were added directly to 7.5 µl RT reactions. When detected, the CT values correlated (R2 > 0.998) to the number of cells in the RT reactions over at least three orders of magnitude ().
Dynamic range of eight TaqMan miRNA assays using OP9 cell lysates. The number of cell input ranged from 3 to 2500 cells per RT. A Caenorhabditis elegans miRNA (miR-2) was used as a negative control.
The effect of non-specific genomic DNA on TaqMan miRNA assays was also tested for 12 assays. Results showed no difference in CT values in the presence or absence of 5 ng of human genomic DNA added to the RT reactions, suggesting that the assays are highly specific for RNA targets (data not shown). Based on this observation, we added heat-treated cells directly to miRNA quantification assays. illustrates the comparison of miRNA quantification using purified total RNA, cell lysates and heat-treated cells derived from an equal number of HepG2 cells. Adding heat-treated cells directly to the miRNA assays produced the lowest CT values, and good concordance was observed among all three different sample preparation methods.
Comparison of heat-treated cells, cell lysate and total RNA for real-time quantitation of 10 miRNAs. The level of miRNA expression is measured in the threshold cycles (CT). Approximately 400 HepG2 cells were analyzed per PCR.
The reproducibility of TaqMan miRNA assays was examined by performing12 miRNA assays with 16 replicates performed by two independent operators (data not shown). The standard deviation of the CTs averaged 0.1, demonstrating the high precision of the assays.
Solution hybridization-based miRNA northern analysis was used as an independent technology to compare with TaqMan miRNA assays (). We observed that hybridization-based miRNA analyses were less reproducible and that concordance with TaqMan assays varied from target to target. There was a general concordance between the two methods (R2 = 0.916) for miR-16 across five mouse tissue samples. However, correlations were relatively low for less abundant miRNAs, such as miR-30 (R2 = 0.751).
Comparison of TaqMan miRNA miR-16 assay to solution-based northern hybridization analysis. Total RNAs from mouse kidney, liver, lung, spleen and testicle tissues were used.
Hybridization methods can lack specificity for the mature miRNAs. We investigated the ability of the TaqMan miRNA assays to differentiate between the mature miRNAs and their longer precursors, using synthetic targets for pri-miRNA precursors, pri-miR-26b and pri-let-7a and pre-miRNA precursor pre-miR-30a (). TaqMan assays designed to detect either precursors or mature miRNAs were tested with synthetic targets averaging 1.5 × 108 copies per RT reaction (1.3 × 107 copies per PCR). TaqMan miRNA analyses with only pri-miRNA precursor molecules produced CT values at least 11 cycles higher than analyses with mature miRNA ones. This result implies that if mature miRNA and precursor were at an equal concentration, the latter would contribute <0.05% background signal to the assay of mature target. For pre-miR-30a where the mature miRNA miR-30a-3p is located at the 3′ end of the pre-miR-30a sequence, a difference of 8.4 CT was observed. The results showed that TaqMan miRNA assays are specific to mature miRNAs. However, the assay specificity is better if the miRNA is located at the 5′ strand of the pre-miRNA precursor. Experiments analyzing total RNA instead of synthetic targets indicated that the precursors are at least two orders of magnitude less abundant than mature miRNAs, based on CT differences of 7 or more for miR-26b-1 and let-7a-2 precursors. Considered together, these results suggest that the TaqMan miRNA assays are highly specific for the mature miRNAs.
Discrimination between mature miRNAs and their pri- or pre-miRNA precursors
The ability of the TaqMan miRNA assays to discriminate miRNAs that differ by as little as a single nucleotide was tested with the five synthetic miRNAs of let-7a, let-7b, let-7c, let-7d and let-7e (). Each miRNA assay was examined against each miRNA. Relative detection efficiency was calculated from CT differences between perfectly matched and mismatched targets, assuming 100% efficiency for the perfect match. Very low levels of non-specific signal were observed, ranging from zero to 0.3% for miRNAs with 2–3 mismatched bases and only 0.1–3.7% for the miRNAs that differed by a single nucleotide. Most cross-reactions resulted from G–T mismatches during the RT reaction (let-7a assay versus let-7c target etc.). Only the targeted miRNA was detected if more than three mismatched bases between any two miRNAs were present.
Figure 7 Discrimination power of let-7 miRNA assays. Relative detection (%) calculated based on CT difference between perfectly matched and mismatched targets. A total of 1.5 × 108 copies of synthetic RNA was added to RT reaction. The concentration was (more ...)
We compared the discrimination ability of the TaqMan miRNA assays to that of solution-based hybridization analysis (). In our hands, the hybridization method discriminated well between let-7a and let-7b. However, poor or no discrimination was observed among let-7a, let-7c and let-7d, which differ by 1–3 nt.
Poor discrimination of miRNAs with solution hybridization-based (northern) analysis.
We speculated that stem–loop primers might provide better RT efficiency and specificity than linear ones. Base stacking of the stem might enhance the thermal stability of the RNA–DNA heteroduplex. Furthermore, spatial constraint of the stem–loop would likely improve the assay specificity in comparison to conventional linear RT primers. We compared the sensitivity and specificity of the stem–loop and linear RT primers using synthetic miRNAs for let-7a (). We observed several advantages for the stem–loop RT. First, in the presence of the synthetic let-7a target, the CT values between linear and stem–loop RT methods differed by 7, indicating that the efficiency of stem–loop RT was at least 100 times higher. Secondly, stem–loop RT discriminated better between miRNAs that differ by two bases based on ΔCT values. Finally, the stem–loop RT was at least 100 times better able to discriminate between the mature miRNA and its precursor, based on the ΔCT (precursor versus mature) of 7.
Figure 9 Specificity of TaqMan miRNA assays between stem–loop and linear RT primers. Mature let-7a-specific assay was tested against let-7a, let-7e and pri-miR precursor let-7a-3. ΔCT represents the CT difference between two targets or methods. (more ...)