Mini-review
A practical approach to RT-qPCR—Publishing data that conform to the MIQE
guidelines q,qq
Sean Taylor, Michael Wakem, Greg Dijkman, Marwan Alsarraj, Marie Nguyen *
Bio-Rad Laboratories, Inc., Hercules, CA 94547, USA
a r t i c l e i n f o
Article history:
a b s t r a c t
Given the highly dynamic nature of mRNA transcription and the potential variables introduced in sample
handling and in the downstream processing steps (Garson et al. (2009) [4]), a standardized approach to
each step of the RT-qPCR workflow is critical for reliable and reproducible results. The MIQE provides this
approach with a checklist that contains 85 parameters to assure quality results that will meet the acceptance
criteria of any journal (Bustin et al. (2009) [1]).
In this paper we demonstrate how to apply the MIQE guidelines (www.rdml.org/miqe) to establish a
solid experimental approach.
Ó 2010 Published by Elsevier Inc.
1. Introduction
Real-time quantitative PCR (qPCR) has become a definitive technique
for quantitating differences in gene expression levels between
samples. Over the past 10 years, the popularity of this
method has grown exponentially, with the publication of well over
25,000 papers from diverse fields of science, including agricultural,
environmental, industrial, and medical research, making reference
to RT (reverse transcription)-qPCR data.
One of the central factors that has stimulated this impressive
growth is the increased demand from journal review panels for
the use of RT-qPCR to support phenotypic observations with quantitative,
molecular data. Furthermore, gene expression analysis is
now being used to support protein expression data from proteomics-based
assays. Since no strict guidelines have been established,
researchers have generally designed their experiments
based on information gathered from disparate sources, which has
resulted in data of variable quality.
In an effort to assist the scientific community in producing consistent,
high quality data from qPCR experiments, the minimum
information for publication of quantitative real-time PCR experiments
(MIQE) guidelines have been recently published [1]. This
has been followed by the development of an XML-based real-time
PCR data markup language (RDML) for the consistent reporting of
real-time PCR data by the RDML consortium [10]. This consortium
is active in the development of appropriate and standardized terminology,
guidelines on minimum information for biological and
biomedical investigations, and a flexible and universal data file
structure with tools to create, process, and validate RDML files.
All relevant information about the RDML project is available at
www.rdml.org.
The ultimate goal of RDML and MIQE is to establish a clear
framework within which to conduct RT-qPCR experiments and to
provide guidelines for reviewers and editors to use in the evaluation
of the technical quality of submitted manuscripts against an
established yardstick. As a consequence, investigations that use
this widely applied methodology will produce data that are more
consistent, more comparable, and ultimately more reliable.
2. Experimental design
Proper experimental design is the key to any gene expression
study. Since mRNA transcription can be sensitive to external stimuli
that are unrelated to the processes studied, it is important to
work under tightly controlled and well-defined conditions. Taking
the time to define experimental procedures, control groups, type
and number of replicates, experimental conditions, and sample
handling methods within each group is essential to minimize variability
(Table 1). Each of these parameters should be carefully recorded
prior to conducting gene expression experiments to assure
good biological reproducibility for published data.
3. RNA extraction
If samples must be collected over a period of time or in too large
a number to process immediately, they should be stored in appro-
1046-2023/$ - see front matter Ó 2010 Published by Elsevier Inc.
doi:10.1016/j.ymeth.2010.01.005
q
Information of this paper is corresponding to our currently written tech note
Bulletin No. 5859 which was published in 2009.
qq
This application note has been provided by Bio-Rad Laboratories as supplemental
educational material to this thematic special issue. This application note
was sponsored by Bio-Rad Laboratories and has not undergone a peer review
process within Elsevier.
* Corresponding author.
E-mail address: marie_nguyen@bio-rad.com (M. Nguyen).
Methods 50 (2010) S1–S5
Contents lists available at ScienceDirect
Methods
journal homepage: www.elsevier.com/locate/ymeth
priate conditions (frozen at À80 °C and/or in RNA storage solution)
until use. To minimize handling time during the RNA extraction
procedure, it is recommended that samples be processed in relatively
small batches of 10–20. The RNA extraction procedure
should include a DNase I treatment step to remove any contaminating
genomic DNA.
4. RNA quality control
Ensuring that only RNA of high purity (no contaminants) and
high integrity (not degraded) is used is one of the most critical
points in the RT-qPCR experimental workflow. Impurities in the
RNA sample may lead to inhibition of the RT and PCR can lead to
varying and incorrect quantification results [3,5]. Since sample
purity and integrity are not related, both should be assessed to
ascertain that the RNA sample meets minimal acceptance criteria
for the downstream workflow.
Purity of the sample with respect to protein contamination can
be assessed spectrophotometrically by measuring the OD260/280 ratio
in a pH neutral buffer. An OD260/280 of 1.8–2.0 indicates good
quality RNA that is devoid of protein and phenol contamination
(Fig. 1B). However, no RNA integrity information can be obtained
from a spectrophotometric reading. RNA integrity can be assessed
using several methods. The traditional method is visual inspection
after electrophoresis on a formaldehyde agarose gel in the presence
of a fluorescent dye such as ethidium bromide. Observation
of two sharp bands for the large and the small subunit ribosomal
RNAs (rRNA) with the intensity of the larger band being about
twice that of the smaller band is indicative of intact RNA. While
this method is inexpensive, the interpretation of the data is mostly
subjective and requires approximately 200 ng of total RNA, which
may be difficult if the sample is only available in limited quantities.
This method can be refined by quantifying the intensity of the
rRNA bands using an imager with densitometry scanning. While
the values of the 28S/18S ratio can vary between different tissues
or cell types, a ratio between 1 and 2 is indicative of an intact
RNA sample.
A major improvement in RNA integrity analysis came with the
introduction of microfluidics-based electrophoresis systems [9]
such as the Experion™ automated electrophoresis system (BioRad)
that combines integrity and concentration quantification of
RNA in a single step from nanogram or picogram quantities
(Fig. 1A and C). In addition to generating a virtual gel, an electropherogram,
and calculating the 28S/18S ratio, the Experion system’s
software automatically calculates and reports an RNA
quality indicator (RQI) value (Fig. 1C) that reflects the integrity of
the RNA sample based on several criteria [2] of the electropherogram.
The RQI value ranging from 1 (degraded) to 10 (intact) provides
an objective assessment of the integrity of the RNA sample
and it can be used to screen out degraded samples.
By assuring consistency in both purity and quality across all
RNA samples, variability between biological replicates can be reduced
(Fig. 1D) [9]. When a batch of RNA samples has successfully
met the standards of quality control, we recommend their immediate
use in RT-qPCR experiments, or their conversion into much
more stable cDNA by reverse transcription to preserve RNA integrity
after the quality check.
5. Reverse transcription
Given the prevalence of RNase in the environment, we recommend
performing the reverse transcription of total RNA samples
to cDNA immediately following the quality control assessment.
This will avoid the risk of RNA samples degradation from multiple
freeze/thaws before conversion to cDNA. For the RT step, the key is
to assure consistent and complete coverage of the transcribed genome
in the extracted RNA sample. Some genes are very long, but
the associated RT products for these sequences cannot be as large,
especially if the RT primers anneal only at the ends of each mRNA.
By annealing primers at the end of the mRNA plus at random
points within each mRNA sequence a good sampling of the population
of each gene is obtained. This method, which is more representative
than just annealing at the ends or at random sites within
each transcript, provides better coverage of the transcribed
genome.
A reverse transcription buffer should contain a mix of primers
that are random in sequence allowing for a better sampling of
the mRNA; RNase H, the enzyme that specifically degrades RNA
in DNA/RNA duplexes; a specific and sensitive reverse transcriptase
with broad dynamic range for RNA amounts from 1 pig to
1 pg and; a simple and fast protocol. We recommend that the same
amount of total RNA be used and that reaction time be maintained
for reverse transcription for all experimental samples to minimize
variability between biological replicates. Reverse transcribed RNA
can be stored frozen at À20 °C or À80 °C until use and diluted before
use in the downstream qPCR.
6. Primer and amplicon design
Both primer design and careful choice of target sequence are
essential to ensure specific and efficient amplification of the products.
Target sequences should be unique, 75–150 bp long with a GC
content between 50% and 60%, and should not contain secondary
structures. It is recommended that primers should have a GC content
of 50–60% and a melting temperature of 55–65 °C. Long G or C
Table 1
RT-qPCR experimental design and sample management.
Experiment design Control groups Replicates Experiment conditions Sample handling
Disease or treatment groups Time course study (i.e., t = 0) Biological (different
sample per well)
Growth conditions (media
and time or OD)
Precise time to harvest
cells or tissue
Target genes implicated Normal vs. disease (i.e.,
normal)
Technical (same sample
per well)
Days of embryonic
development
Sample extraction
method
Potential reference genes Untreated vs. drug treated
(i.e., untreated)
— Amount per mass of drug or
compound
Preservation method and
time
Number of data points to draw statistically
significant conclusions
— — Sex, phenotype Thaw and
homogenization
procedure
— — — Incubation time Total RNA extraction
procedure
This table summarizes the workflow of a typical RT-qPCR experiment from experimental design to defining control groups, replicates, and experimental conditions to the
detailed procedures for sample handling. This ultimately assures that the key steps in RT-qPCR data production lead to high quality, reproducible, and publishable data.
S2 S. Taylor et al. / Methods 50 (2010) S1–S5
stretches in the primer should be avoided, but it is recommended
to have G or C at the end of the primers.
A number of programs are available to help design primer pairs
and pick target sequences. We recommend designing oligonucleotides
using Primer-Blast (www.ncbi.nlm.nih.gov/tools/primerblast/index.cgi?LINK_LOC=BlastHomeAd),
a program developed
by NCBI that uses the algorithm Primer3 [12]. Primer sequences
are compared (blasted) to the user-selected databases to ensure
they are unique and specific for the gene of interest. The program
MFOLD (http://mfold.bioinfo.rpi.edu/cgi-bin/dna-form1.cgi) can
then be used to analyze the amplicons for potential secondary
structures that may prevent efficient amplification [14]. Ideally,
two sets of oligonucleotides should then be ordered and tested
for their performance in a qPCR.
7. qPCR validation
A validated qPCR assay is one that has been assessed for the
optimal range of primer annealing temperatures, reaction efficiency,
and specificity using a standard set of samples [1]. This will
assure that the reaction conditions, buffers, and primers have been
optimized and that the cDNA samples are not contaminated with
inhibitors of Taq polymerase. Bio-Rad has created a practical web
resource (www.bio-rad.com/genomics/pcrsupport) for qPCR design
and validation.
7.1. Determination of annealing temperature, melt curve analysis, gel
analysis of amplicon, and no template control
A critical step in a PCR is the annealing of the primers to their
target sequences. It has to be performed at the right temperature
for the primers to anneal efficiently to their targets, while preventing
nonspecific annealing and primer–dimer formation. The fastest
way to determine optimal annealing temperature is to use a therFig.
1. Analysis of RNA purity and integrity. This figure shows an example of RNA
purity and integrity analysis using spectrophotometry and Bio-Rad’s Experion
automated electrophoresis system. Mouse liver total RNA samples were subjected
to degradation by heating at 90 °C for different periods of time and analyzed for
both purity and integrity. Green boxes highlight the fact that the total RNA sample
appears by absorbance readings alone (>1.8) to be of sufficient quality, whereas RQI
measurement on the Experion system clearly shows degraded sample with low RQI
values. (A) Virtual gel image generated by the Experion software showing various
degradation levels of the RNA samples and a progressively decreasing intensity of
the 18S rRNA band; (B) the OD260/OD280 ratio measured for all samples on the
Nanodrop spectrophotometer is between 1.8 and 2.0 indicating that the samples
are devoid of protein contamination; (C) summary of RNA integrity analysis using
the Experion software. The 28S/18S rRNA ratio and the RQI indicate a decreasing
integrity of the heat treated RNA samples. A color classification allows for easy
identification of the samples that are not adequate for qPCR; (D) example of qPCR
analysis of GAPDH expression performed on degraded samples [5]. The profiles
show increasing Cq with the degraded samples. No degradation ( ); 1 h degradation
( ); 3 h degradation ( ); 5 h degradation ( ); 7 h degradation ( ).
Fig. 2. Validation of qPCR primers. (A) qPCR is performed at a range of annealing
temperatures using a thermal gradient block. Amplification profiles indicate that
the most efficient amplification occurs at the four lowest annealing temperatures
between 53 °C and 57.7 °C where the curves are at the lowest Cq; (B) a single peak
on the melt curve analysis indicates a single PCR product.
S. Taylor et al. / Methods 50 (2010) S1–S5 S3
mal cycler equipped with a temperature gradient feature. A range
of temperatures around the calculated Tm of the primers should be
tested (Fig. 2A).
It is important to check the specificity of the reaction by analyzing
the PCR product. A melt curve analysis, performed at the end of
the PCR cycles, will confirm specificity of primer annealing. The
melt curve should display a single sharp peak (Fig. 2B). In addition,
samples should be run on an agarose gel or polyacrylamide gel to
confirm that the amplicon is of the expected size. Alternatively, an
automated electrophoresis system such as the Experion automated
electrophoresis system can be used.
A duplicate, no template control (NTC) reaction should be included
in every run for each primer pair to test buffers and
solutions for DNA contamination and to assess for primer–dimers.
7.2. Establishment of a standard curve (to evaluate PCR efficiency)
The efficiency of a PCR is a measure of the rate at which the
polymerase converts the reagents (dNTPs, oligonucleotides, and
template cDNA) to amplicon. The maximum increase of amplicon
per cycle is 2-fold representing a reaction that is 100% efficient. It
is important to measure reaction efficiency as it is indicative of
problems with the qPCR that can cause artifactual results. Lowefficiency
reactions (<90%) may be caused by contaminating
Taq inhibitors, high or suboptimal annealing temperature, old
or inactive Taq, poorly designed primers, or amplicons with secondary
structures. High reaction efficiency (>110%) is generally
the result of primer–dimers or nonspecific amplicons. The most
common causes of both high and low reaction efficiencies are
poorly calibrated pipettes or poor pipetting technique.
A standard curve is generally used to determine the reaction
efficiency for any qPCR. The template for this typically is a sample
of cDNA or spiked plasmid cDNA in a sample extract. We recommend
initially producing a 10-fold dilution series over eight points
starting from the most concentrated cDNA sample, to ensure the
standard curve covers all potential template concentrations that
may be encountered during the study (i.e., broad dynamic range).
For each dilution, a standard qPCR protocol should be performed
in triplicate for all the primer pairs to be used in the experiment
and Cq values determined. The standard curve is constructed by
plotting the log of the starting quantity of the template against
the Cq values obtained. The equation of the linear regression line,
along with Pearson’s correlation coefficient (r) or the coefficient
of determination (r2
), can then be used to evaluate whether the
qPCR assay is optimized.
Ideally, the dilution series will produce amplification curves
with tight technical replicates that are evenly spaced. If perfect
doubling occurs with each amplification cycle, the spacing of the
fluorescence curves will be determined by the equation 2n
= dilution
factor, where n is the number of cycles between curves at
the fluorescence threshold (in other words, the difference between
the Cq values of the curves). For example, with a 10-fold serial dilution
of DNA, 2n
= 10. Therefore, n = 3.32, and the Cq values should
be separated by 3.32 cycles. Evenly spaced amplification curves
will produce a linear standard curve with the goal of 90–110% reaction
efficiency.
The r or r2
value of a standard curve represents how well the
experimental data fit the regression line, that is, how linear the
data are. Linearity, in turn, gives a measure of the variability across
assay replicates and whether the amplification efficiency is the
same for different starting template copy numbers. A significant
difference in observed Cq values between replicates will lower
the r or r2
value. An r with an absolute value >0.990 or an r2
value
>0.980 is desirable for RT-qPCRs. Deletion of points at both ends of
the standard curve may be required to obtain an acceptable slope
(efficiency) and r2
value. This will ultimately define the dynamic
range of cDNA concentration for each primer pair with respect to
sample dilution.
7.3. qPCR reagents, instrument, and analysis program
There are a wide variety of commercial qPCR reagent kits available.
We recommend a saturating dye that provides much better
sensitivity than SYBR
Ò
Green.
A large number of qPCR detection instruments are commercially
available. Sample volumes with a range of 10–50 ll in standard
96-well format, 0.2 ml, low-profile thermal cycler plates,
strips, or tubes from any manufacturer.
All qPCR instruments are packaged with data analysis software
and we recommend that a good software package include the following
features:
(1) Flexibility to enter plate setup information such that well
identifiers can be loaded and edited before, during, or after
the run.
(2) Ability to group wells from multiple experiments on one
plate.
(3) Built-in gene expression analysis that can normalize data to
multiple reference genes [13] and individual reaction efficiencies
[11].
(4) The ability to combine multiple plates of experimental data
into a large gene study.
8. Choice of reference genes
In RT-qPCR experiments, reference genes are used as controls to
normalize the data by correcting for differences in quantities of
cDNA used as a template [6,8,13]. A perfect reference gene is
therefore one that does not exhibit changes in expression between
samples from various experimental conditions or time points. Reference
genes must therefore be carefully selected based on experimental
data and we recommend the following protocol:
(1) Extract the total RNA from at least one or two samples from
each experimental condition or time point and confirm their
purity and quality (see Steps 2 and 3 above).
(2) Normalize the sample concentration and perform reverse
transcription PCR (see Step 4) from the same volume of each
sample.
(3) Perform the qPCR experiment using the same volume of
each cDNA sample as a template.
(4) Use the geNorm method [7] to calculate the target stability
between the different conditions (available at med-
gen.ugent.be/genorm/).
A good reference gene should have an M value below 0.5 or 1 in
homogeneous and heterogeneous samples set, respectively [13].
geNorm also helps in the selection of the optimal number of reference
genes.
Typically, between 3 and 5 good reference are required to
achieve the most accurate normalization [13].
9. Experimental reproducibility
There are two sources of variability in a gene expression experiment
that may affect the results:
(1) Biological variability which is due to inherent differences in
gene expression levels between individual organisms, tissues,
or cell culture samples.
S4 S. Taylor et al. / Methods 50 (2010) S1–S5
(2) Technical variability in the experimental process itself
which is typically associated with pipetting, poorly calibrated
pipettes, and sample quality and quantity.
To mitigate the effect of biological and technical variability, it is
generally accepted that at least three biological and two technical
replicates per biological replicate be performed for each experiment
(Fig. 3). If the experiment compares gene expression levels
between control and treated samples, the three biological replicates
should be samples that were treated in separate and independent
experiments.
10. Conclusions
RT-qPCR is the method of choice for gene expression analysis
because of its high sensitivity from samples of very low RNA concentrations.
However, in order to assure accurate and reproducible,
quantitative data, strict standard operating procedures should be
followed. All experimental details and controls should be accurately
reported when publishing gene expression experiments.
This will allow proper assessment of the data by the scientific community,
and enable informed comparison of expression data between
labs and between experiments.
In summary the key steps for most RT-qPCR experiments
include:
(1) Experimental design with appropriate number of biological
replicates and proper control samples.
(2) Sample procurement which requires adherence to strict
experimental protocols for acquisition, processing, and storage
to assure reproducibility and minimize standard deviations
between replicates.
(3) Quality control of RNA for purity and integrity.
(4) Reverse transcription to convert the total RNA to cDNA.
(5) qPCR experiments with proper primer design, choice of target
sequence and reference genes, and technical replicates.
(6) The MIQE guidelines were written to provide all the parameters
that should be met to publish acceptable results from
qPCR experiments. Here we have shown how the most
important items in the MIQE checklist can be addressed in
practice to assure high quality results.
Acknowledgment
We are grateful to Dr. Jo Vandesompele for his insightful comments
and helpful contributions to this paper.
References
[1] S.A. Bustin et al., Clin. Chem. 55 (2009) 611–622.
[2] V. Denisov et al., Development and validation of RQI: an RNA quality indicator
for the Experion automated electrophoresis system, Bio-Rad Bulletin #5761,
2008.
[3] S. Fleige, M.W. Pfaffl, Mol. Aspects Med. 27 (2006) 126–139.
[4] J.A. Garson et al., AIDS Res. Hum. Retroviruses 25 (2009) 377–378. author reply
379–381.
[5] J. Gingrich et al., Effect of RNA degradation on the data quality in quantitative
PCR and microarray experiments, Bio-Rad Bulletin #5452, 2006.
[6] L. Gutierrez et al., Plant Biotechnol. J. 6 (2008) 609–618.
[7] J. Hellemans et al., Genome Biol. 8 (2007) R19.
[8] J. Huggett et al., Genes Immun. 6 (2005) 279–284.
[9] S. Imbeaud et al., Nucleic Acids Res. 33 (2005) e56.
[10] S. Lefever et al., RDML: structured language and reporting guidelines for realtime
quantitative PCR data, Nucleic Acids Res. Advance Access Published
February 17, 2009 (accessed 21.05.09).
[11] M. Pfaffl, Nucleic Acids Res. 29 (2001) e45.
[12] S. Rozen, H.J. Skaletsky, Primer3 on the www for general users and for biologist
programmers, in: Bioinformatics Methods and Protocols: Methods in
Molecular Biology, Humana Press, Totowa, NJ, pp. 365–386.
[13] J. Vandesompele et al., Genome Biol. 3 (2002) research0034.1–
research0034.11.
[14] M. Zuker, Nucleic Acids Res. 31 (2003) 3406–3415.
Fig. 3. Experimental replicates. All experimens should be designed with a combination
of biological and technical replicates. Zhis illustrates a simple experiment
with triplicate biological samples from control and treatment/experimental conditions.
For each biological sample, three technical replicates are recommended for
the gene of interest as well as for the reference gene(s). This results in a total of at
least 36 samples plus the duplicate NTC for a total of >40 wells.
S. Taylor et al. / Methods 50 (2010) S1–S5 S5