In most measurements of gene expression, mRNA is first reverse-transcribed into cDNA. The reverse transcription reaction is not very well understood, and it is expected to be the uncertain step in gene expression analysis. It can introduce errors produced by effects of mRNA secondary and tertiary structures, variation in priming efficiency, and properties of the reverse transcriptase (1)(2)(3)(4)(5). The aim of this work was to study the yield, reproducibility, and sensitivity of some commercially available reverse transcriptases on low to intermediate expressed genes by use of quantitative real-time PCR (QPCR).
Total RNA extraction, reverse transcription, and QPCR were performed as described in the Data Supplement that accompanies the online version of this Technical Brief at http://www.clinchem.org/content/vol50/issue9/ (6)(7). All reverse transcription reactions were run in replicates of four, using starting material from the same RNA pool prepared from bovine spleen, liver, or jejunum, which eliminated sample-to-sample variation (8). Only results for RNA from spleen are shown. Liver and jejunum gave similar results, which are provided in the online Data Supplement. To determine absolute reverse transcription yields, we added an artificial RNA MultiStandard (Roboscreen) to samples (9)(10). Eight reverse transcriptases were studied: Moloney murine leukemia virus RNase H− (MMLVH; Promega); MMLV (Promega); avian myeloblastosis virus (AMV; Promega); Improm-II (Promega); Omniscript (Qiagen); cloned AMV (cAMV; Invitrogen); ThermoScript RNase H− (Invitrogen); and SuperScript III RNase H− (Invitrogen). Reverse transcription with AMV, MMLV, and Omniscript was performed at 37 °C, whereas with cAMV, Improm-II, and MMLVH it was performed at 45 °C, and with ThermoScript and SuperScript, it was performed at 50 °C.
The cDNA synthesis yields of the intermediate to highly expressed β-actin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) genes and the low expressed genes 5-hydroxytryptamine 1a receptor (HTR1a), HTR1b, HTR2a, and HTR2b were measured by QPCR using SYBR Green I detection chemistry (7). The mean threshold cycle (Ct) and corresponding SD for all combinations of genes and reverse transcriptases are shown in Fig. 1⇓ . Because of the exponential behavior of PCR, a difference of 1 cycle in Ct between runs that differed only in the reverse transcriptase used corresponded to twofold difference in reverse transcription yield (assuming 100% PCR efficiency). For HTR1a, HTR1b, and HTR2b, the reverse transcription yields obtained with the eight reverse transcriptases were similar, whereas for GAPDH and, in particular, for HTR2a and β-actin, substantial variations were observed (Fig. 1⇓ ). For example, for HTR2a, the Ct was 32.3 cycles when SuperScript III was used, whereas it was 38.8 cycles when AMV was used. This corresponds to a 26.5 = 91-fold difference in reverse transcription yield. For HTR2b, the difference in yield with the two enzymes was only 225.4 − 25.2 = 1.14, which is 14%.
Primer hybridization relies on access to the appropriate target site in the mRNA and may vary substantially because of mRNA folding (11)(12). Reverse transcription yields could vary among the reverse transcriptases in a highly gene-dependent way as a consequence of mRNA secondary and tertiary structures. Large variation is expected for mRNAs with tight structures in which access to primer target sites is restricted. Our data suggest that this may be the case for β-actin, GAPDH, and HTR2 with our choice of primers. The reverse transcriptase that performed best for these genes was SuperScript III, which was used at 50 °C. A higher annealing temperature is often claimed to improve reverse transcription yields by reducing the degree of mRNA secondary structure, but ThermoScript, which also was used at 50 °C, did not perform particularly well. Furthermore, we found no advantage when we used reverse transcriptases without RNase activity (MMLVH, SuperScript III, and ThermoScript), which also is claimed by some vendors to improve transcription efficiency. For the six genes studied, SuperScript III gave the overall highest yield, followed by MMLVH and cAMV. AMV gave the poorest yield. The reproducibility, represented as SD of repeated experiments (Fig. 1⇓ ), was very high with all reverse transcriptases for all genes studied but HTR2a. This could be attributable to statistical variation at low copy numbers for HTR2a, which is expressed at very low yield (13).
For absolute determination of reverse transcription yields, the systems were calibrated by addition of RNA/DNA MultiStandard molecules (Roboscreen) of known concentrations. The reverse transcription yield is defined as: nmRNA is the number of mRNA molecules of a particular gene in the test sample, and ncDNA is the number of cDNA copies for that mRNA that are produced by reverse transcription. From the calibration curve of the DNA MultiStandard (10–106 DNA molecules), we obtain (14): The slope (−3.49) reflects a PCR efficiency of 93%. When we instead added RNA MultiStandard with the same sequence as the DNA standard to the test samples and reverse transcribed it before QPCR, the reverse transcription yield could be calculated as: When we added 104–106 RNA molecules, we obtained similar reverse transcription yields, whereas with 103 RNA molecules, the yields were substantially higher (Table 1⇓ ). This is probably an artifact attributable to too few RNA molecules (diluted 1:29) in the final running solution and formation of primer-dimer products detected by SYBR Green I chemistry. The latter may be avoided by use of specific probes. These data were therefore not considered when we calculated mean yields.
The detection of rare mRNA transcripts is often an issue. The yield of low-abundance mRNA has been shown to be significantly improved when carrier is used (1)(15). The reverse transcription yields for the RNA MultiStandard varied more than 100-fold. The lowest yield (0.4%) was obtained with AMV for 106 RNA molecules, and the highest yield (90%) was obtained with SuperScript III for 104 RNA molecules (Table 1⇓ ). The latter was overall the most efficient reverse transcriptase, with a mean yield of 83%. MMLV and MMLVH gave mean yields of 44% and 40%, respectively, whereas the mean yields of the other reverse transcriptases were <25%. The yield obtained with MMLVH was comparable to that reported in a previous study (15).
In conclusion, we show that reverse transcription yields vary up to 100-fold with the choice of reverse transcriptase and that the variation is gene dependent. Previously, we also reported a dependence on priming strategy (1). Hence, for quantitative gene expression measurements based on reverse transcription to be comparable among laboratories, the same enzyme, priming strategy, and experimental conditions must be used.
This work was supported by Deutsche Forschungsgemeinschaft, the Chalmers Bioscience effort, and the Crafoord Foundation.
- © 2004 The American Association for Clinical Chemistry