Real-time PCR, Volume IX

Table of Contents

Letter from the Editor
Index of Experts
Q1: What primer design techniques do you use to
improve yield?

Q2: What protocols do you follow to prevent contamination by non-amplicons?
Q3: How do you validate if your fragment was successfully amplified? How do you quantify your results?
Q4: What validation methods do you use? How do you determine reaction efficiency?
Q5: What types of data analysis methods do you use? Do you normalize your data, and how?
List of Resources

Letter from the Editor

In this issue of GT, we bring you our ninth technical guide on real-time PCR. While it seems like not so long ago that we laid the first troubleshooting lab manual in your hands, the field has advanced, especially in the areas of quality control, experimental design, and data analysis. And while we've covered qPCR eight times before, we hope that this guide brings with it a fresh perspective on an old concept. Real-time PCR has become the gold standard in most labs for RNA and DNA amplification and quantification. It's used for everything from amplifying a target gene to verifying a hit from a microarray study, to "sequencing" genes in bacteria or viruses to detect their presence in clinical samples. We've covered using qPCR to detect microRNAs as well, and the number of applications continues to multiply.

With this in mind, though, the basics are still as tricky as ever to nail down and get just right. Luckily, our experts have some solid tips on how to optimize primer design, prevent the contaminating effects of stray DNA, and improve sample handling, among others. And while you're at it, check the back of the guide for the usual list of resources.

— Jeanene Swanson

Index of experts

Many thanks to our experts for taking the time to contribute to this technical guide, which would not be possible without them.

Reginald Beer
Lawrence Livermore National Laboratory

Mikael Kubista
TATAA Biocenter, Sweden

Michael Pfaffl
Tum, Germany

Michèl Schummer
Fred Hutchinson Cancer Research Center

Soroush Sharbati
Free University of Berlin

Q1: What primer design techniques do you use to improve yield?

Due to our extensive work in biodetection applications, we use an in-house custom-developed bioinformatics software that utilizes %CG, desired tm, desired primer length, desired amplicon size, and incorporates knowledge about conserved sequences in the targeted organisms and their near-neighbors to optimize primers and probes for the specific detection application. These primer and probe combinations are developed in silico against public and proprietary sequenced organisms. We have had very good results to date with our signatures, both in typical bench-top systems as well as our experimental microfluidic platforms.

— Reginald Beer

We use several primer design software depending on applications including beacon designer and alleleID from Premier Biosoft and e-primer from Polyclone Bioservices. We also use free design software such as Primer3. For methylation analysis by HRM we design primers using methyl primer express from Applied Biosystems. But we don't blindly rely on the primer design software. They are good, but predicting DNA thermodynamics is complex. Since primers are cheap, we usually design at least three alternative primer pairs for any new assay, order them all and then test them, if possible also in combinations. Then we move on with the pair that performed best to tune the assay conditions.

— Mikael Kubista

We are using Primer3 software to design our primers for the SYBR assays. Sequences are checked by m-fold for secondary structures of mRNA. If qRT-PCR will not work we additionally will adjust the primer design by hand. We have a first success rate of 80-90% and after adjustment nearly 100%, with a distinct single product and no primer dimers. Further we are using exclusively good primer quality from distributors with high oligo purity by HPLC purification.

— Michael Pfaffl

I have used the Web-based application Primer3 for years and I have had no reasons to switch to another application because this one does give me the flexibility I need. 80% of our primer pairs work. Another 10% work poorly and 10% don't work and need to be redesigned.

— Michèl Schummer

In our lab we use SYBR Green qPCR mainly for quantification of gene expression after reverse transcription of both mRNAs and small RnNAs such as micro-RNAs. For measuring cellular mRNA contents after reverse transcription of total RNA using random hexamers, we usually design primers (18-22 nt) applying the program Primer3. We choose primers producing an amplicon between 100 and 250 bp. To ensure high-efficiency and yield of the qPCR reaction, the primers have to possess minimal variance in GC content and melting temperature. Another criterion for choosing primers is that they have to span at least one intron to distinguish amplicons derived from cDna from those derived from genomic DNA contamination. A comfortable opportunity for choosing primers for annotated genes is also the use of primer databases such as RtPrimerDb. For quantification of miRna expression, we have developed an approach called miR-Q. miRNAs are first converted into cDna by reverse transcription using an miRna-specific oligonucleotide with 5' overhang. cDNA-molecules are quantified by a novel approach based on utilizing three DNA-oligonucleotides at different concentrations. A first oligonucleotide comprises a sequence homology to the miRNA molecule with an additional 5' overhang. The introduced 5' overhangs provide binding sequences for universal primers, which prime the final qPCR reaction utilizing the SYBR green chemistry.

— Soroush Sharbati

Q2: What protocols do you follow to prevent contamination by non-amplicons?

The droplet-based emulsion qPCR systems we have developed isolate the target DNA through the emulsion process, but we still have to follow standard protocols in the upstream sample and reagent preparation steps. This includes use of PCR hoods, gloves, gowns, alcohol, bleach solutions, and DNAse. We also employ separate clean rooms for pre-amplification chemistry verses post-amplification sample recovery and detection.

— Reginald Beer

Our laboratory in Gothenburg has taken extensive precautions to minimize/eliminate risk for contamination and control quality measures to detect it should it occur. We have detailed quality control system and for every new study we develop robust standard operating procedure, including regular calibrations and controls for all steps involving sample preparation, reverse transcription and qPCR. If genomic contamination is an issue in an RNA study, we treat the samples with DNase. If RNA contamination is an issue we use UnG (Uracil-n-glycosylase) protocols. Our pre- and post-PCR work is separated physically. PCR products are banned in spaces where samples are handled, RT is performed, and PCR prepared. Working surfaces are regularly decontaminated, and template is handled in UV cabinets, which allows us to inactivate any contaminating DNA before new work is started.

— Mikael Kubista

We have a strict decontamination program in the RNA and PCR labs. Once a week all benches will be cleaned either with 3% H2O2 bleach or 2.5% natriumhyperperchloride. Pipet sets are color codes for each lab to keep pipetting error stable and consistent. They will be decontaminated as well twice a month by UV radiation. Further we have a lab work- flow, meaning separate labs for all working steps performed: an RNA extraction lab, a qRT-PCR lab, and if necessary a separate lab where the PCR tubes will be opened and analyzed via gel electrophoresis or send for sequencing. Therefore a PCR tube will be never opened in the PCR lab, a contamination by unspecific products is "nearly impossible."

— Michael Pfaffl

Whenever possible a primer spans an exon-exon boundary or the PCR product spans an exon-exon boundary that, if genomic contamination were present, would lead to a PCR product too long to amplify in the usual 10-15 seconds. Also, the RNA prep includes a DNase step that is so efficient that we have not seen genomic contamination in over two years.

— Michèl Schummer

We follow a strict regime based on separation of templates from primers, enzyme, etc. To achieve this, the entire pipetting is performed in UV clean PCR-workstations that facilitate decontamination and avoid template spreading. Decontamination is performed by UV-irradication before and after pipetting a reaction. To ensure high quality control of qPCR reactions we use two separate UV clean PCR-workstations. The first one is only used for pipetting the mastermix where strictly no template DNA is allowed. This step avoids contamination of primers, enzyme mixes, dyes, etc., which are usually used by several operators. Once the mastermix is set, the work is continued in the second workstation by first pipetting the mastermix in wells or tubes followed by the addition of the template DNA. This rigorous standard of quality control offers efficient minimization of non-template amplicons caused by contamination. Intron spanning primers allow distinguishing between specific amplicons and those derived from genomic DNA contamination. Moreover, the use of non-reverse-transcribed controls guarantees the absence of contaminating genomic DNA.

— Soroush Sharbati

Q3: How do you validate if your fragment was successfully amplified? How do you quantify your results?

We determine successful amplification through real-time detection and post-amplification gel electrophoresis or cGE. We use suitable ladders to bracket our desired amplicon size and identify any primer/dimer products. For quantitation, we have two methods. First, we perform ct analysis to develop a standard curve, second we can look at the percentage of droplets that support amplification.

— Reginald Beer

First time we run an assay we check the PCR product in the Agilent Bioanalyzer or BioRad Experion to verify there is a single band of expected length. If the result is ambiguous we design new sets of primers. Should this not help we would consider sequencing the amplicon to verify that the template indeed has the expected sequence. Once we have established a working assay we check every PCR product by melt curve analysis. We use the new dye chromofy quite frequently. It performs as well as SYBR Green I in qPCR, but is better in melt curve analysis. This is usually sufficient to detect formation of unwanted products, and it is well suited for high throughput studies. When performing absolute quantification we determine the level of detection (LoD) of the analytical procedure. LoD should be determined for the entire process starting with sampling or sample extraction; and not only for the qPCR, since the qPCR step is never the bottleneck limiting sensitivity. LoD is calculated from replicate standard curves, and the analysis is conveniently performed with GenEx from multiD.

— Mikael Kubista

First, a new PCR amplicon will be checked by melting curve analysis, having a single peak and no primer dimers. Second, the predicted length will be checked by
high-resolution gel electrophoresis. Third, new products will be sent for sequencing.
Quantification is dependent on the question behind the problem. Normally we as animal physiologist are interested in the relative gen expression changes and therefore we are using the relative quantification approach with additional efficiency correction.

— Michael Pfaffl

Successful amplification of the actual PCR is checked on a gel (sporadically if hundreds of samples and genes are involved). Quantitation is done by SYBR green followed by melting curve. Taqman is not an option for experiments involving hundreds of genes and variants.

— Michèl Schummer

New assays and primers are generally validated by performing a gradient qPCR (e.g. with the steponePlus cycler) to find out the optimal annealing temperature. After performing the melting curve analysis, the product is separated by agarose gel electrophoresis to check for the specific product size. The amplicon is cut out and the purified Dna is sequenced to determine the specificity of primers. Generally we quantify the amplicons and use dilution series as calibration standards for quantification. The quantity of the purified amplicon is measured by spectral photometry using e.g. a nanodrop. cDNA amounts are determined by triplicate measurements for each sample using the standard curve covering the linear range of the particular assay. Expression is normalized using the geometric mean of several appropriated housekeeping gene.

— Soroush Sharbati

Q4: What validation methods do you use? How do you determine reaction efficiency?

Each signature is virtually screened against Genbank and proprietary sequence information available to the laboratory. Upon passing these criteria the signatures are screened on the bench against over 2,000 environmental and near neighbor organisms. The signatures are then further tested against all available target organisms.

— Reginald Beer

We determine PCR efficiencies by performing a standard curve. When testing assay performance the standard material can be purified PCR product, purified plasmid or cDNA. We always calculate the confidence interval of the estimated PCR efficiency using GenEx from multiD. This is critical. If too few samples are analyzed or the covered concentration range is too narrow the estimated efficiency is very uncertain, and any corrections made will rather comprise the data than improve quality. It is not advisable to run independent mini standard curves on each plate and use the independently estimated efficiencies for inter-plate calibration or any other corrections.

— Mikael Kubista

Reaction efficiency will be checked by the slope of the dilution curve in a mixture of biological samples. It is very important to perform the efficiency analysis with a biological background to test the inhibitory effects of the matrix. This result in a mean efficiency for each gene amplified. Or if working with the Rotor-Gene the single run efficiency can be determined and implemented in the REst relative quantification analysis. This results in an efficiency value for each PCR run performed.

— Michael Pfaffl

Primers are routinely checked on a variety of different tissues and with several PCR conditions, allowing assessment of amplification properties. All reactions include negative controls (no primer, no template) and positive controls (cDNA from a mix of tissues the primer pair is known to amplify. Reaction efficiency is not checked. The qPCR amplification curves give us certain clues that we subsequently investigate (e.g. one well does not reach the plateau of other wells). If too many of the tissue do not amplify the cDNA before 35 cycles, we usually use a higher concentration of the
cDNA. Routinely we use ~50 ng of cDNA in a 15 µl reaction.

— Michèl Schummer

Usually, we use qPCR as a validation method for microarray experiments. Global gene expression analysis is first done by performing two color microarray experiments. Differentially expressed genes are then measured and accurately quantified by qPCR. We determine the efficiency of qPCR reactions by analyzing our standard curves. 100% efficiency means the duplication of product per cycle. In qPcR, there is an inversely proportional connection between the log of the applied template and the ct. based on these facts, the slope m of the standard curve equal to -3.32 indicates 100% efficiency of the reaction. In addition, the coefficient of determination R2 close to 1 indicates perfect linearity between template input and Ct. The slope m, the efficiency and the coefficient of determination R2 are generally calculated and indicated by the analysis software when using a standard curve.

— Soroush Sharbati

Q5: What types of data analysis methods do you use? Do you normalize your data, and how?

For positive detections, we determine titer from the percentage of droplets amplified using Poisson statistics. No normalization or ct computation is required for this method, it is simply a binary yes or no for each droplet with the percent of droplets supporting amplification allowing a direct assessment of starting titer. We can also determine titer from the real-time monitoring of the cycle threshold of droplets that support amplification. For cycle threshold determination, we use the previously published Cepheid algorithm, which performs background averaging and subtraction until the threshold value is obtained. In our droplet publications we did not normalize our data.

— Reginald Beer

We work with all three qPCR analysis methods: absolute quantification with standard curve and reverse calibration when analyzing individual markers and partial least squares for multimarker experiments, relative quantification using the t-test or ANOVA or corresponding non-parametric methods when necessary, and multimarker expression profiling using hierarchical clustering, principal component analysis and self organizing maps for exploratory studies and artificial neural networks and support vector machines for confirmatory studies. All those methods are available in GenEx from multiD. Mode of normalization depends on the application. For absolute quantification normalization is to sample amount. For relative quantification we usually normalize to reference genes. But these must be carefully selected. They should be selected from a panel of candidate reference genes that are all independent of the studied conditions. This must be verified, since inclusion of any regulated genes in the panel introduces bias and the predicted genes will not be the best. Some people select reference genes from all the genes they study, including the regulated genes. This is completely wrong and leads unavoidably to erroneous results. Using 2-way anova, also called normfinder, in GenEx the treatment effect on the different candidate reference genes is estimated, and those genes that are affected by treatment can be removed. It is important to start with a reasonably large number of reference gene candidates; preferably more than 10. For human and mouse we offer panels of reference gene candidates that readily can be tested on representative samples to find appropriate normalizers. Software for the analysis is included. It is also important to select the appropriate number of reference genes. If equally good reference genes are available, the larger the number of reference genes, the more stable the normalization. However, if the reference genes are added in the order of decreasing stability there will be an optimum number that gives most stable result. This number can be calculated from the accumulated standard deviation. The improvements of adding extra reference genes eventually become small and may not be motivated from a cost/performance perspective. It is important to compare the confounding technical variance of the experimental procedure with the requested resolution. This can be done with the Power test in GenEx. In some cases, such as developmental studies, there may be no good reference genes at all, and higher data quality is obtained by normalizing with total RNA. Another interesting case is single-cell expression profiling. Gene expression among individual cells varies greatly and mRNA levels of unrelated genes are not correlated on the cell level. Normalization with reference genes makes no sense. On the other hand it is most intuitive to express the gene expression levels per cell.

— Mikael Kubista

Before reaction setup we normalize our sample to RNA quality and quantity. Quality will be tested by RNA integrity studies by Experion or bioanalyzer 2100, and only biological samples with acceptable RIn or RQI number will be analyzed. They should be at least in a range of 7 or higher. Any sample with RIn / RQI value lower than 7 will be re-extracted. Quantity will be checked by nanoDrop analysis, because this is the most reproducible and less variable. Normalization after PcR will be performed using internal grown references. Therefore a panel of reference genes will be tested and most stable ones will be selected by Genorm or normfinder analysis.

— Michael Pfaffl

Each 384-well S plate contains aliquots of all cDNAs used during the experiment as well as a standard made with testes cDNA in 5 dilutions (1:1, 1:3, 1:9, 1:27, 1:81) as duplicates, amplified with primers for actb (all cDNAs are sub-aliquoted and stored at -20°c for consistency). This allows us to transform the logarithmic cycle threshold (ct) values into linear values. All PCRs are normalized by the averaged expression of housekeeping genes run in triplicate. These genes are dependent on the nature of the experiment. We use at least 3 and average their median-normalized values. E.g. GaPDH is a bad housekeeping gene for certain cancer-to-normal experiments since it is elevated in cancer. Further analysis of the normalized expression data is dependent on the nature of the experiment. Unsupervised cluster analysis gives us some clues on the quality of the data (similar tissues should cluster together).

— Michèl Schummer

After generating the raw gene expression data they are normalized using the geometric mean of several stable housekeeping genes. For this purpose, we use the tool genorm, which analyzes the stability of applied housekeeping genes by calculating the stability measure m and providing a normalization factor based on the geometric mean of the housekeeping genes. In experiments studying treated samples versus untreated controls we generally calculate the log ratio of relating normalized data. A versatile tool for visualizing gene expression data for example by heat map analysis is the tm4 software package. This package was intended to be used for microarray analysis; however, e.g. calculated log ratios after qPCR analysis can be loaded as tab delimited multiple sample files into the multiExperiment viewer of tm4 to perform heatmap analysis, clustering, statistical analysis etc. This offers a convenient platform to describe and analyze comprehensive gene expression data generated by qPCR.

— Soroush Sharbati

List of Resources

Publications

Bengtsson M, Ståhlberg A, Rorsman P, Kubista m. 2005. Gene expression profiling in single cells from the pancreatic islets of langerhans reveals lognormal distribution of mRNA levels. Genome Research 15(10): 1388-92.

Bergkvist A, Forootan A, Zoric N, Strömbom L, Sjöback R, Kubista M. 2008. Choosing a normalization strategy for Rt-PCR. Genetic Engineering & Biotechnology News. 28(13).

Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, Kubista M, Mueller R, Nolan T, Pfaffl MW, Shipley GL, Vandesompele J, Wittwer CT.2009. The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clinical Chemistry 55(4): 611-22.

Hellemans J, Mortier G, De Paepe A, Speleman F, Vandesompele J. 2007. qBase relative quantification framework and software for management and automated analysis of real-time quantitative PCR data. Genome Biology 8(2): R19.

Lefever S, Hellemans J, Pattyn F, Przybylski DR, Taylor C, Geurts R, Untergasser A, Vandesompele J; rdml Consortium. 2009. Rdml: structured language and reporting guidelines for real-time quantitative PCR data. Nucleic Acids Research 37(7): 2065-9.

Lefever S, Vandesompele J, Speleman F, Pattyn F. 2009. RtPrimerdB: the portal for real-time PCR primers and probes. Nucleic Acids Research 37(Database issue): D942-5.

Murphy J, Bustin SA. 2009. Reliability of real-time reverse-transcription PCR in clinical diagnostics: gold standard or substandard? Expert Review of molecular Diagnostics 9(2): 187-97.

Tichopad A, Kitchen R, Riedmaier I, Becker C, Ståhlberg A, Kubista M. 2009. Design and optimization of reverse-transcription quantitative PCR experiments. Clinical Chemistry 55(10): 1816-23.

Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, Speleman F. 2002. Accurate normalization of real-time quantitative Rt-PCR data by geometric averaging of multiple internal control genes. Genome Biology 3(7): Research0034.

Web Tools

Genex
http://www.multid.se/genex.html

Primer3
http://primer3.sourceforge.net/
http://frodo.wi.mit.edu/primer3/input.htm
http://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi

qBase
http://medgen.ugent.be/qbase/

qBasePlus
http://www.biogazelle.com/products/

Real-time PcR data markup language (Rdml)
http://www.rdml.org/

ReSt
http://rest.gene-quantification.info/

RtPrimerdB
http://www.rtprimerdb.org/

Conferences

Advances in qPCR
http://www.selectbiosciences.com/conferences/aqpcr2009/

qPCR Symposium
www.qpcrsymposium.com
www.qpcrsymposium.eu

Quantitative PCR
http://www.healthtech.com/qpc/overview.aspx