Short Technical Reports

Low Efficiency of the of these assays. Currently, little is sion gene was a generous gift of Dr.
known about the efficiency of the re- Ching-Ping Tseng from Chang Gung
Moloney Murine verse transcriptase enzyme commonly University, Taiwan. This plasmid was
Leukemia Virus Reverse used to perform this initial conversion: used to generate mRNA runoff tran-
a modified Moloney murine leukemia scripts using the T7-MEGAshort-
Transcriptase during virus (MMLV) reverse transcriptase. script kit (Ambion, Austin, TX,
Reverse Transcription of Commercially available MMLV re- USA) as directed. Residual plasmid
verse transcriptases have been modified DNA was significantly reduced by ad-
Rare t(8;21) Fusion Gene to remove the inherent RNAse H activi- ditional DNase I (Roche Applied Sci-
Transcripts ty and thus improve the potential ence, Basel, Switzerland) digestion
cDNA yields. Commercial manufactur- steps and purification of the RNA
BioTechniques 32:768-775 (April 2002) ers suggest improved yields with this runoff transcript on RNeasy® spin
modified enzyme but fail to provide columns (Qiagen, Valencia, CA, USA).
any indication of actual conversion effi- After these additional steps, contami-
ABSTRACT ciencies. Here we describe simple, real- nating plasmid DNA could not be de-
time RT-PCR methodologies to deter- tected in 1 ng of the RNA runoff tran-
The resolving power of RT-PCR is limited mine the specific conversion efficiency scripts by the real-time PCR assay.
by the efficiency of RNA-to-cDNA conver- of MMLV reverse transcriptase. One RT of the RNA runoff transcripts
sion. Methods to determine this efficiency, assay relies on a limiting dilution step was performed in a 20-µL reaction.
using a real-time PCR assay for quantifying of a reacted RT cocktail before detec- Transcripts were mixed with 4 nM spe-
AML1-MTG 8 [t(8;21)] fusion gene tran- tion with real-time PCR and the ability cific MGT8 (7) primer (T12, 5′-AG-
scripts, are described. The efficiency is of that PCR to detect a single full- GCTGTAGGAGAATGG-3′) (MWG
shown to be directly proportional to RNA length cDNA. The other approach re- Biotech, High Point, NC, USA). RNA
template levels. The Moloney murine lates known input RNA quantities to background, when employed, consisted
leukemia virus (MMLV) reverse transcrip- detected output cDNAs. Both of these of human colon total RNA (Stratagene,
tase enzyme’s conversion efficiency was cal- two different approaches were taken, as La Jolla, CA, USA). This background
culated to be approximately 20%. The effi- the first permitted the efficiency to be RNA was repeatedly used as a negative
ciency was even lower (6%) when target accurately determined when relatively control and never demonstrated the
templates were rare (single molecules) in the small amounts of target RNA are avail- presence of the fusion gene transcripts.
RT reactions. Levels of nonspecific or back- able for conversion, whereas the second RNA (runoff transcripts and back-
ground RNA present in the RT reaction re- approach was effective for much larger ground) and primer mixtures (10 µL)
duced the reverse transcriptase’s conversion amounts of RNA template. were annealed by an initial heating to
efficiency. This background effect was partic- A leukemic fusion gene is generated 65ºC for 5 min and then cooled to room
ularly pronounced when the specific tem- by a translocation [t(8;21)] occurring temperature. A 10-µL RT cocktail [1×
plate was present in rare amounts. between chromosomes 8 [acute mye- RT buffer (Stratagene), 100 µM dNTPs
loid leukemia 1 (AML1)] and 21 [mye- (Roche Applied Science), 4 U Prime
loid leukemia gene on chromosome 8 RNase Inhibitor (Eppendorf, Ham-
INTRODUCTION (MTG8)]. This fusion gene was chosen burg, Germany), 25 U reverse tran-
because a real-time RT protocol had scriptase (Stratagene)] was then added
Reverse transcription (RT) of RNA previously been described (5) and is to the annealed RNA/primer mixtures
to cDNA combined with real-time PCR currently being utilized in house. The and reacted at 42ºC for 50 min, fol-
is widely employed for a variety of assay is highly sensitive, and previous lowed by 95ºC for 5 min.
molecular assays. Assays to detect fu- reports purport to detect three copies of Real-time PCR was used to detect
sion gene transcripts commonly found cDNA (5). full-length cDNAs (molecules that ex-
in leukemia and lymphomas are such Effects of a non-target background tend from the T7 primer past the fur-
an example (2,8). These assays permit RNA on the efficiency of the RT reac- thest PCR primer, <690 bp). Real-time
the detection of minimal residual dis- tion, and thus the assay’s ability to de- primers and TaqMan probe were as
ease in chemotherapy-treated patients tect rare RNA transcripts (AML1- described by Marcucci et al. (5). One
and can be used as biomarkers for the MTG8) within a neutral background, is correction to the AML1-MTG 8 reverse
return of the disease (1,4,6) or for the described. Whereas the data presented primer was made (5′-ATCCACAGGT-
occurrence of novel therapy-related here are specific to the t(8:21) fusion GAGTCTGGCATT-3′) that removed a
translocations (3). With few exceptions, gene RT-PCR, the results are likely in- single adenine residue after the under-
the detection of cells bearing transloca- dicative of the reverse transcriptase’s scored adenine. An initial range finding
tion-generated fusion transcripts repre- potential efficiency. RT-PCR experiment utilized four sepa-
sents the extreme limits of this technol- rate RT reactions with RNA runoff
ogy’s resolution. The initial RT step, transcript template diluted over four or-
which converts the fusion gene’s MATERIALS AND METHODS ders of magnitude. Completed 20-µL
mRNA transcript into a single strand of RT reactions were diluted with an equal
cDNA, is fundamental to the resolution A cloned AML1-MTG8 cDNA fu- amount of water, and 5 µL were then

768 BioTechniques Vol. 32, No. 4 (2002)

Short Technical Reports
 Reverse Transcriptase (Stratagene) in Converting RNA Runoff Transcript into Full-Length, PCR-Detectable cDNA
Table 1. Efficiency of Stratascript

Background Runoff RNA Positive Total Ct of Positive

RNA Molecules Reactions Reactions Reactions ± SD Efficiency

None 70 12 48 39.5 ± 0.7 19.7%

1 µg/reaction 70 4 48 40.6 ± 0.5 6.0%
Efficiency percentage was calculated using the Poisson [-ln (negative reactions/total reactions)/possible number of cDNA mol-
ecules)]. Each of the 48 reactions had the potential to contain 1.458 (70/48) cDNAs, assuming 100% conversion efficiency.
The Ct values indicate that the positive reactions contained just one detectable cDNA, thereby permitting the use of the Pois-
son calculation.

added to a 200-µL real-time PCR (100 range. Figure 1 shows the results of Non-Target Background RNA Does
µL TaqMan Master Mix reagent (2×) these real-time PCRs (10 µL) seeded Not Affect PCR Results
(Applied Biosystems, Foster City, CA, with 0, 0.5, 1, 5, or 10 plasmid mol-
USA), 6 nM both forward and reverse ecules. Real-time PCR offers quantify- There was concern that the non-tar-
primers, and 0.6 nM TaqMan probe ing abilities, as demonstrated through get background RNA used in the RT re-
(MWG Biotech). This total reaction the cycle threshold (Ct) values that vary action might have some influence on
was then distributed into 20 10-µL re- in proportion to the starting template the PCR results. To demonstrate that
actions and run on the ABI PRISM® amounts. Thus, single-molecule tem- this background RNA has no effect, a
7700 real-time PCR machine (Applied plates in this experiment yielded Ct val- single RT reaction was prepared with-
Biosystems) using the standard cycling ues of around 39.5, whereas five-mole- out background RNA, and the cDNA
conditions but with 45 rather then 40 cule template reactions had Ct values of was used as template. Real-time PCRs
cycles. Positive reactions were then 37.5. Of course, one must appreciate containing the cDNA were split into
scored. The RT reactions that yielded that it is technically impossible to place eight replicates that contained back-
less than 20 positive reactions, and thus precisely one, two, or three molecules ground RNA and eight replicates that
being within range for limiting dilution into a reaction, but an average distribu- did not. Several different amounts of
analysis, were then further analyzed. tion around those numbers was obtain- cDNA template were used, and the
Part of the remaining diluted RT reac- able. Single plasmid molecules are thus mean Ct values were calculated. Back-
tion (20 µL) was added to 950 µL real- detectable by the real-time PCR assay. ground RNA had no effect on the mean
time PCR cocktail, the mixture distrib-
uted across 96 separate PCR tubes, and
then reacted. Positive wells were
scored, and the reverse transcriptase’s
efficiency was determined.
Statistical analysis was performed
using Statistica®, a commercially avail-
able statistical software package (Stat-
soft, Tulsa, OK, USA). Differences be-
tween proportions and means were
tested using a Student’s t test. Poisson
analysis was used to determine effi-
ciencies for the limiting dilutions and is
described below. Log-linear regression
was performed to relate input RNA val-
ues to output cDNA values.

RESULTS

Single-Molecule PCR Templates

Sufficient to Generate a Real-Time
PCR Signal

To demonstrate that a single tem- Figure 1. Mean Ct values, standard deviations, and error obtained from single and multiple num-
bers of initial template molecules. Plasmid bearing the AML1-MTG8 fusion gene is approximately
plate will generate a real-time PCR sig- 4100 bp (2.5 × 106 g/mol), and 100 ag consists of 25 individual molecules. PCRs (10 µL) were seeded
nal, a SacI-linearized plasmid was seri- with 0, 0.5, 1, 5, or 10 plasmid molecules in replicates of 24, and data from positive reactions were aver-
ally diluted down to the attogram aged. Means are indicated by bars with standard errors (boxes) and deviations (whiskers).

770 BioTechniques Vol. 32, No. 4 (2002)

Short Technical Reports
Table 2. Calculation of Stratascript Reverse Transcriptase Efficiency Starting with 500 ag (1400 Molecules) Runoff RNA Transcripts
No. cDNA
molecules/ Positive Total cDNA Total cDNA molecules
Background RNA Ct range Mean Ct ± SD reactions molecules of possible 700

No Background 3/≤37.5 36.9 ± 0.5 4 12

Overall Ct 2 38.8 ± 0.6 43 86 135
39.3 ± 1.03 >37.5 to 39.5 19.3% efficiency
1/≥39.5 40.2 ± 0.4 37 37

With Background 3/≤37.5 37.5 1 3

1 µg
RNA/reaction 2 38.8 ± 0.5 26 52 89
>37.5 to 39.5 12.7% efficiency
Overall Ct 1/≥39.5 40.5 ± 0.8 34 34
39.7 ± 1.23

The Ct ranges of the one, two, and three cDNA molecule reactions were abstracted from Table 1. The efficiencies are calculated
from the total number of cDNA molecules reverse-transcribed from a possible 700 RNA molecules (as half of the original reac-
tion was limited by dilution). The Poisson calculation was not used, as the widely ranged Ct values indicated that there were more
than one cDNA molecules in the positive reaction. A simple fraction calculation was sufficient to determine the reverse tran-
scriptase efficiencies from this data. Negative reactions are not shown, but there were 10 (no background) and 35 (background).

Ct values obtained (data not shown). of rare target RNA into cDNA. The first each PCR. Thus, a positive PCR indi-
Therefore, inclusion of background approach utilized RT reactions contain- cates a single cDNA template. For this
RNA in RT reactions does not have any ing hundreds of RNA runoff transcripts approach to accurately access the RT
influence on the PCR results. templates that were reacted, and the enzyme efficiency, the dilution must be
cDNA products were then diluted over sufficient to limit just one or less cDNA
Determination of RT mRNA many PCR tubes. In this way, cDNA molecules per reaction. This can be as-
Conversion Efficiency molecules generated in the RT step are sessed by the Ct values generated. A to-
limited by dilution such that only one tal of 96 PCRs were used to screen two
Whereas manufacturers have typi- template on average will be present in separate RT reactions performed with
cally determined rates at which isotopes
are incorporated into precipitated cD-
NAs, the approaches taken here attempt
to determine an actual conversion effi-
ciency of RNA template into PCR-de-
tectable (full-length) cDNA molecules.
Determining the actual efficiency of the
reverse transcriptase’s conversion of
RNA template into cDNA is complicat-
ed by several technical difficulties. One
significant challenge is in quantifying
the number of reverse-transcribed
cDNA molecules. Real-time PCR is
ideally suited to overcome this chal-
lenge. Another technical difficulty is
determining the conversion efficiencies
for RT reactions where the RNA tem-
plate is at very low levels (1–100 RNA
molecules). Again, real-time PCR
meets this challenge with its extremely
broad dynamic range for detection.
Two different limiting dilution ap-
proaches were undertaken in an attempt
to determine accurately the efficiency of Figure 2. Efficiency of the Stratascript reverse transcriptase in converting RNA runoff transcripts
the reverse transcriptase during the RT into detectable full-length cDNA.

772 BioTechniques Vol. 32, No. 4 (2002)

Short Technical Reports
and without background RNA (1 µg). would obviously fail with higher tem-
Each RT reaction contained 50 ag plate amounts. To further explore this
runoff transcript RNA, which is approx- issue, the RT efficiency was sought us-
imately 140 copies. Half of the com- ing a bulk approach. With this ap-
pleted RT reaction was limited over 48 proach, the power of real-time PCR to
PCR tubes and assayed. Table 1 shows quantify PCR templates is fully
the results of this survey. The mean Ct brought to bear. A standard curve of Ct
values are identical to those values for values was generated for known
single templates (Figure 1); thus, the amounts of plasmid bearing the fusion
positive reactions contained just one gene. RT reactions containing known
single detectable cDNA molecule. Of amounts of runoff transcript (750–
the possible 140 runoff RNA molecules 750 000 copies) were also screened
available for cDNA conversion, just a alongside the plasmid standard curve
fraction was converted. A simple calcu- (1–10 000 copies). The sequence detec-
lation using the Poisson distribution for tor software permits unknown values to
limiting dilution analysis was then used be obtained through the use of the stan-
to determine the conversions efficien- dard curve Ct values. Each standard
cies (Table 1). The presence of back- curve point was repeated four times,
ground RNA had a considerable effect and each runoff transcript was repeated
and lowered the RT efficiencies by in three or four different RT reactions
nearly 3-fold (P = 0.04). This method is with or without the addition of back-
particularly good at determining the en- ground RNA (1 µg/reaction).
zyme efficiency when a relatively small For each RT reaction, the amount of
number of RNA templates are available input template (mRNA, the runoff tran-
for conversion, as single full-length script) was known, and the Ct values
cDNA products can be scored. obtained permitted the number of out-
The second approach taken was to put molecules (cDNA, full-length and
convert a relatively small number of amplifiable) to be determined from the
RNA template molecules and then again standard curve. There was little or no
to limit those cDNA molecules by dilu- difference between linear regressions
tion, but also to use the quantification derived from experiments with or with-
power afforded by real-time PCR. When out background RNA (likelihood ratio
an RT reaction containing 700 RNA test, P = 0.88). For this reason, the two
runoff transcripts was diluted across 96 data sets are combined. Figure 2 plots
PCR tubes, 84 positive reactions (with- the number of input mRNA molecules
out background RNA) and 61 positive against the number of detected output
reactions with background RNA were molecules (y = -0.839 + 1.037×, R2 =
detected (Table 2). However, in these ex- 0.98, P <10-6). A clearly linear rela-
periments, the range of Ct values is tionship exists between RNA input and
much greater, between 36.9 and 41.5, in- the detected cDNA output. Over the
dicating that some of the positive reac- range of relevant input concentrations
tions contained more than one cDNA (100–100 000) the log-linear regression
molecule as PCR template. In this case, implies about a 20% efficiency
the positive tubes can be subdivided into (17%–22%). With this bulk approach,
tubes that contain one or more cDNA the efficiency of the RT reaction is in
templates on the basis of the Ct value the range of the previous methodologi-
generated and the total number of cDNA cal calculations. However, the inhibito-
molecules determined. With this proce- ry effect of the background RNA ob-
dure, the RT reaction efficiency was served with the previous methods was
12.7% and 19.3% with and without not seen with this bulk approach. This
background RNA, respectively. Again, is likely because, in the bulk approach,
the effect of the background significant- higher amounts of template are con-
ly impedes the efficiency of the reverse tained within the background RNA,
transcriptase in reverse-transcribing the while the previous methods employed
RNA runoff transcripts (P < 0.002). relatively rare template in the same
While these two scoring methods amount of background. With the small-
are useful at determining the RT effi- er numbers of template, correct priming
ciency when only a relatively small of the RNA template is likely to be hin-
number of target RNA is available, they dered by background RNA.

Vol. 32, No. 4 (2002)

DISCUSSION 2.Curry, J.D., M.C. Glaser, and M.T. Smith.
2001. Real-time RT-PCR detection and quan-
tification of E2A-PBX1 fusion genes associat-
Efficiency of the MMLV reverse ed with leukaemia. Br. J. Haematol. 115:826-
transcriptase is influenced by several 831.
factors, and the resolution of the assay 3.Felix, C.A. 2001. Leukemias related to treat-
described here is clearly limited by the ment with DNA topoisomerase II inhibitors.
conversion efficiencies at the RT step. Med. Pediatr. Oncol. 36:525-535.
4.Kondo, M., K. Kudo, H. Kimura, J. Inaba,
The effects of background RNA are no- K. Kato, S. Kojima, T. Matsuyama, and K.
ticeable only when relatively small Horibe. 2000. Real-time quantitative reverse
numbers of RNA template are available transcription-polymerase chain reaction for
for conversion to cDNA. This effect is the detection of AML1-MTG8 fusion tran-
scripts in t(8;21)-positive acute myelogenous
absent when much larger amounts of leukemia. Leuk. Res. 24:951-956.
RNA template are available for RT. 5.Marcucci, G., K.J. Livak, W. Bi, M.P.
However, the overall efficiency of the Strout, C.D. Bloomfield, and M.A.
reverse transcriptase was determined to Caligiuri. 1998. Detection of minimal resid-
be only approximately 20%, which is to ual disease in patients with AML1/ETO-asso-
ciated acute myeloid leukemia using a novel
say that only 20% of a given RNA tem- quantitative reverse transcription polymerase
plate will be converted into PCR-de- chain reaction assay. Leukemia 12:1482-1489.
tectable cDNA. Conversion efficiency 6.Sugimoto, T., H. Das, S. Imoto, T. Muraya-
was considerably lower (6%) when ma, H. Gomyo, S. Chakraborty, R.
Taniguchi, T. Isobe et al. 2000. Quantitation
RNA templates in the RT reactions of minimal residual disease in t(8;21)-positive
were exceedingly rare. With efficien- acute myelogenous leukemia patients using
cies this low, one could expect that less real-time quantitative RT-PCR. Am. J. Hema-
than 1 out of 10 RT reactions would tol. 64:101-106.
yield a suitable PCR-amplifiable cDNA 7.Tobal, K. and J.A. Liu Yin. 1998. Molecular
monitoring of minimal residual disease in
for detection. At such rare levels, the acute myeloblastic leukemia with t(8;21) by
assay will be unable to confidently de- RT-PCR. Leuk. Lymphoma 31:115-120.
tect rare transcripts. The obvious solu- 8.van Dongen, J.J., E.A. Macintyre, J.A.
tion would be to increase the number of Gabert, E. Delabesse, V. Rossi, G. Saglio, E.
Gottardi, A. Rambaldi et al. 1999. Standard-
RT reactions so as to dramatically in- ized RT-PCR analysis of fusion gene tran-
crease the quantity of starting RNA in scripts from chromosome aberrations in acute
which the target template is rare. leukemia for detection of minimal residual
It should be noted that the quality of disease. Report of the BIOMED-1 Concerted
the RNA runoff transcript will directly Action: investigation of minimal residual dis-
ease in acute leukemia. Leukemia 13:1901-
affect these kinds of experiments, and 1928.
care should be taken to produce high-
quality RNA. The reverse transcriptase
This research was funded by National
efficiency should be considered when
Institutes of Health grant no. P42ES04705
employing this widely used enzyme.
from the National Institute of Environmen-
Assays to determine mRNA expression
tal Health Sciences. J.C. and C.M. are grant
will be directly affected by this limited
trainees of the University of California Tox-
enzyme efficiency, and heed should be
ic Substances Program. Dr. Alan Hubbard
taken when preparing and interpreting
is gratefully acknowledged for his statistical
standard curves used to determine RNA
advice. Address correspondence to Dr. John
expression levels. Methods to empiri-
Curry, Department of Immunology, Molec-
cally determine the efficiency of re-
ular and Cellular Biology, 439 Life Sci-
verse transcriptases have been de-
ences Addition, University of California,
scribed, and their usefulness in
Berkeley, CA 94720-3200, USA. e-mail:
determining which enzyme to employ
[email protected]
in RT-PCR assays is demonstrated.

Received 15 October 2001; accepted

REFERENCES 17 December 2001.

1.Amabile, M., B. Giannini, N. Testoni, V.

Montefusco, G. Rosti, C. Zardini, C. Ter-
John Curry, Cliona McHale,
ragna, S. Buonamici et al. 2001. Real-time and Martyn T. Smith
quantification of different types of bcr-abl University of California
transcript in chronic myeloid leukemia.
Haematologica 86:252-259. Berkeley, CA, USA

Vol. 32, No. 4 (2002) BioTechniques 775